Kurmademys kallamedensis, GAFFNEY & MOODY & WALKER, 2001

GAFFNEY, EUGENE S., MOODY, RICHARD T. J. & WALKER, CYRIL A., 2001, Kurmademys, a New Side-Necked Turtle (Pelomedusoides: Bothremydidae) from the Late Cretaceous of India, American Museum Novitates 3320, pp. 1-16 : 4-15

Kurmademys kallamedensis , new species

TYPE SPECIMEN: ISI R152 , a nearly complete skull lacking the dorsal part of the prefrontals, the posterior part of the crista supraoccipitalis, and part of the left quadratojugal .

TYPE LOCALITY: Near the village of Kallamedu , Tamil Nadu, southern India. Map of locality is in Sastry et al. ( 1972) .

HORIZON: Kallamedu Formation of the described from other exposures of the Kal­ Ariyalur Group. Formation named and de­ lamedu (Matley, 1929; Yadagiri and Ayyasscribed by Sastry et al. ( 1972), who corre­ ami, 1987). The Kurmademys locality is a lated it with the uppermost Maastrichtian; the small pocket of fine­grained sandstone and Cretaceous­Tertiary boundary is its upper clay, about 6 in. thick. It also contained croclimit (Sastry et al.: 6). Dinosaurs have been odiles, gar scales, and freshwater gastropods

and bivalves and is interpreted as a freshwater pond deposit.

DIAGNOSIS: As for genus.

ETYMOLOGY: For the Kallamedu Formation.

REFERRED MATERIAL: ISI R 155A, partial skull; ISI R 155B, partial skull; ISI R 155C, partial skull; ISI R 158, partial skull; ISI R 159, partial skull.

DESCRIPTION

The Kurmademys kallamedensis type skull, ISI R152 has a premaxilla­condyle median length of 47.2 mm, a maximum width of 49.4 mm, and a height from condylus mandibularis of the quadrate to the top of the skull roof of 28.2 mm .

The order of elements and the order of topics within elements follows Gaffney ( 1979). More general features of the pleurodire skull and explanations of terminology can also be found in Gaffney ( 1979) and are not repeated here.

Although at present six skulls of Kurmademys are known, the type specimen, ISI R152 , forms the basis of this description, with a few contributions from the other specimens. The type skull is the best preserved and nearly complete. The other five skulls are not as well preserved and are not yet fully prepared .

PREFRONTAL

Almost all of both prefrontals are missing in the type skull, ISI R152 . The only fragments of prefrontal remaining are the ventralmost tips of right and left prefrontals lying on the medial side of the maxillae in the anterior orbital walls. There is more of the right prefrontal than the left. The dorsal area of the prefrontals has been restored arbitrarily on ISI R152 . However, ISI R158 has prefrontals preserved .

The shape and extent of prefrontal in Kurmademys is consistent with that seen in most other pelomedusoids, such as FR 4922. They meet for their entire length on the midline. There are no nasals.

FRONTAL

Both frontals are nearly complete in ISI

R 152, but their anteriormost margins are breaks rather than sutures, so the contact with the prefrontals is not determinable. However, the extent preserved is comparable to the frontal as seen in Pelomedusa and Pelusios , so it is not likely that much frontal is missing.

As in other pelomedusoids, the frontal of Kurmademys contacts the postorbital laterally and the parietal posteriorly. Its contacts and relative size are very similar to Pelusios and Pelomedusa . On the ventral surface the frontal of Kurmademys has a parasagittal ridge separating fossa orbitalis from the sul­ otic as in most other pelomedusoids. Postecus olfactorius. The sulcus is slightly narrow­ riorly the parietal contacts the supraoccipital er anteriorly than in Pelusios , but widens and prootic as in other pleurodires. The paposteriorly as in Pelusios and most pelome­ rietal of Kurmademys also has a short ventral dusoids. The processus inferior parietalis process below its lateral margin that contacts meets the posterior edge of the frontal ridge the dorsal process of the palatine in the latdorsally as in FR 4922, without a ventral eral wall of the sulcus palatinopterygoideus process as in Pelusios . (Antunes and Broin, 1988).

In other pelomedusoids, such as Baurue­

mys, FR 4922, and Pelusios , the frontal is JUGAL

thickened lateral to the sulcus olfactorius

The jugal in ISI R 152 is preserved on both ridge and forms a wall for the fossa orbitalis

sides, but surrounding contacts are seen only anteriorly and a dorsal margin of the ptery­

on the right side.

go­palatine channel posteriorly. This wall is

The jugal is relatively small in Kurmadeusually continuous with the posterior orbital

mys, similar to Pelusios and Pelomedusa , wall laterally and ventrally. However, in Kur­

and much smaller than in podocnemidids. mademys the frontal is relatively thin and

The jugal is exposed in the posterior margin there is no distinct connection between the

of the orbit, but its exposure is reduced by a sulcus olfactorius ridge and the posterior or­

posterodorsal process of the maxilla. Venbital wall. The pterygo­palatine channel is

trally the jugal contacts the maxilla, and posrelatively open dorsally, at the anterior end,

teriorly the quadratojugal. The jugal is not in contrast to most other pelomedusoids,

exposed on the dorsal edge of the cheek which have a more restricted ridge of bone

emargination due to a contact of the quadhere.

ratojugal and maxilla. This is unusual be­

cause Kurmademys has a cheek emargination PARIETAL at least as extensive as in Pelomedusa and

Pelusios , and they have a large exposure of Both right and left parietals are complete

the jugal along this emargination.

in ISI R 152, except for small cracks and

The medial process of the jugal in ISI breaks.

R 152 is preserved and visible on both sides. The dorsal plate of the parietal is not ex­

It is basically similar to that area described tensive in Kurmademys . Anteriorly there is a

by Gaffney (1979: figs. 53, 130) for Pelusios . transverse contact with the frontal, and lat­

The medial process of Kurmademys contacts erally a contact with the postorbital. There is

the palatine medially and the maxilla anteno contact between the parietal and quadra­

riorly. The jugal forms part of the anterior tojugal; the postorbital is widely exposed

wall of the adductor muscle chamber as in along the edge of the temporal emargination.

most pelomedusoids. The jugal contacts the The temporal part of the parietal is about as

postorbital medially and the pterygoid venextensive as in the living Pelomedusa , which

tromedially. Ventrally the jugal contacts the is slightly more emarginate than most Pelusios .

maxilla. All of these contacts are quite sim­ The ventral wall of the parietal, the pro­

ilar to Pelusios .

cessus inferior parietalis, can be seen in both

sides of ISI R 152. The wall is similar to that

Q in other pelomedusoids. The anterior margin UADRATOJUGAL

of the wall is formed completely by the pa­ The quadratojugal in ISI R 152 is nearly rietal and curves ventrally without the pos­ complete on the right side. Some small parts terior indentation seen in some Pelusios . of the dorsal and ventral margin are probably Ventrally the parietal contacts the pterygoid, missing, but based on the tapering of the which is low until it reaches the area anterior bone and on surrounding bones, very little of to the foramen nervi trigemini, where the the quadratojugal is missing. On the left side, pterygoid/parietal contact is more dorsal. The however, only the anterior part and a small foramen nervi trigemini in Kurmademys is fragment of the posterodorsal contact with formed by the parietal, pterygoid, and pro­ the quadrate are preserved. The quadratojugal of Kurmademys is unusually small for pelomedusoids, because the temporal and cheek emarginations are both extensive. Kurmademys is unique among bothremydids in having such a small quadratojugal and narrow temporal arch. The quadratojugal contacts the quadrate posteriorly, the postorbital anterodorsally, the jugal anteriorly, and the maxilla anteroventrally. The dorsal margin of the quadratojugal forms the lateral edge of the temporal emargination and the ventral margin forms the dorsal edge of the cheek emargination.

SQUAMOSAL

Both right and left squamosals are present in ISI R 152 and both have some damage to their posterolateral margins, but are otherwise complete.

As in most turtles, the squamosal of Kurmademys is cone shaped, forming the posterolateral portion of the antrum postoticum. As in other turtles, the squamosal of Kurmademys fits onto the circular posterior end of the quadrate and contacts the opisthotic medially. Kurmademys has an extensive temporal emargination and the squamosal has no contact with parietal or postorbital. The right side of ISI R 152 also shows that, although the squamosal has a narrow anterior process lying on the quadrate, the process does not reach the quadratojugal. In Pelusios and Pelomedusa there is usually a contact between the quadratojugal and squamosal, although there is some individual variation with some specimens having a very slight contact or no contact. However, none of the specimens of Pelusios or Pelomedusa available to us show the degree of reduction of quadratojugal and squamosal nor the extent of dorsal exposure of the quadrate seen in Kurmademys . This feature is unique to Kurmademys . The dorsolateral surface of the squamosal in Kurmademys is rounded, with no parasagittal ridge or wall seen in other Pelomedusoides. This ridge is a function of the degree of temporal emargination and is present to a varying extent in all other Pelomedusoides.

The antrum postoticum is preserved on both sides of ISI R 152, and its internal extent is visible. The antrum is larger in Kurmademys than in any other Pelomedusoides. Pelusios , Pelomedusa , and FR 4992 have a large antrum postoticum, but the antrum of Kurmademys is even larger. The size in FR 4992, Pelusios , and Pelomedusa is interpret­ ed as the primitive condition for Pelomedusoides, because this is the condition in chelids. The Kurmademys condition is tentatively interpreted as a unique autapomorphy of this genus, because our current analyses show Kurmademys deep within the Bothremydidae .

POSTORBITAL

The postorbital is preserved on both sides of ISI R 152. The posterior edge of the left postorbital is missing bone when compared with the more complete right postorbital.

The size and relations of the postorbital in Kurmademys are very similar to the postorbital in Pelusios and Pelomedusa . The postorbital of Kurmademys lies between the orbit anteriorly and the temporal margin posteriorly and forms part of the margins of those openings. Medially the postorbital contacts the frontal anteriorly and the parietal posteriorly. Laterally the postorbital contacts the jugal anteriorly and the quadratojugal posteriorly. All of these contacts are as in Pelusios and Pelomedusa . FR 4922 differs in having a broad contact of the parietal and quadratojugal posterior to the postorbital.

The ventral process of the postorbital is also similar to that in Pelusios and Pelomedusa . As exposed in the anterior wall of the temporal fossa, the postorbital contacts the parietal medially, the pterygoid ventrally, and the jugal laterally. On the right side, which is better preserved than the left side, there is a small contact between the parietal and pterygoid, preventing exposure of the postorbital in the pterygo­palatine channel at this point. In the posterior wall of the fossa orbitalis, the ventral process of the postorbital contacts the dorsal process of the palatine medially in a sloping suture. Laterally the postorbital contacts the jugal. The medial surface of the ventral process of the postorbital forms the lateral wall and the lateral part of the roof of a relatively short pterygo­palatine channel.

PREMAXILLA

Both premaxillae are present in ISI R 152 and are nearly complete.

Laterally the premaxilla contacts the max­ soids. The apertura is not produced anteriorly illa in a parasagittal suture, and it contacts as in Pelusios and Podocnemis . The dorsal the other premaxilla medially. The posterior process of the maxilla in Kurmademys is margin of the premaxilla forms at least part thinner than in Pelusios and Pelomedusa and of the apertura narium interna, but has a bro­ is similar to FR 4922. The ventral margin of ken edge medially. The broken edge does not the orbit, is formed by the maxilla, which has show a sutural surface anywhere and there is a dorsal process along the posterior margin no fragment of a vomer, but it is possible that of the orbit reducing the contribution made one was present. The premaxilla in FR 4922 by the jugal to the orbit. Forms such as FR has a process of the maxilla lying behind it, 4922 and Podocnemis have a posterodorsal but this is absent in ISI R 152. The dorsal process, but it is separated from the orbit by surface of the premaxilla forms part of the the jugal. Pelusios and Pelomedusa do not floor of the fossa nasalis. In Kurmademys the have this process, but its expression is varipremaxillae curve dorsally to form a sharply able in pelomedusoids. The posterior edge of rising median ridge in the fossa. This median the maxilla forms the margin of the cheek ridge is present in other pelomedusoids, but emargination. Between the cheek emarginait is lower and smaller than in Kurmademys . tion and the jugal, the maxilla contacts the The ventral surface of the premaxilla bears quadratojugal. the continuation of the labial ridge. The la­ The horizontal plate of the maxilla is exbial ridge of Kurmademys is narrower than posed in the orbital floor where the maxilla in Pelusios and lacks the anterior projection contacts the palatine medially and forms a of the margin of the apertura narium externa small part of the border of the large foramen in that form. This area between the apertura orbito­nasale. Posteriorly the maxilla connarium externa and labial ridge is very thin tacts the jugal. in Kurmademys , similar to FR 4922. Kur­ The triturating surfaces of Kurmademys mademys has a shallow median notch similar are narrow anteriorly and widely expanded to Pelusios and wider than in FR 4922. In posteriorly. The maxilla itself, however, tacontrast to Pelusios and Pelomedusa , Kur­ pers posteriorly, so that the palatine forms mademys has a posterior extension to the pre­ the posterior and medial portion of the tritmaxilla that bears a distinct, ventrally facing urating surface. The triturating morphology concavity on the midline that is the ventral is similar to that in Foxemys , which is also surface of the dorsal ridge in the fossa na­ narrow anteriorly and expanded posteriorly, salis. Lower jaws of Kurmademys show a with a significant contribution from the palmarked symphyseal hook. The flat part of the atine. In both skulls the narrow, anterior part triturating surface narrows considerably from has a raised medial edge along the lingual the maxilla to form a narrow shelf between ridge. They differ in that Foxemys has a the concavity and the labial ridge. This mor­ slight pinching of the snout anteriorly, as in phology is also seen in Bothremys and Ro­ Rosasia ; this is absent in Kurmademys , sasia as well as Neochelys and some gener­ where the snout is straight. Foxemys has two alized cryptodires like baenids. There is no accessory ridges, absent in Kurmademys . sign of a foramen praepalatinum. The triturating surface in Kurmademys is

raised anteriorly along the margin of the ap­ MAXILLA ertura narium interna. The posterior expand­

ed area is slightly concave. There are no ac­ Both maxillae of ISI R 152 are complete cessory ridges on the triturating surface, except for the distal ends of the dorsal pro­ which ends posteriorly in a V­shaped margin cesses, which are missing. completely formed by the palatine. The vertical or alveolar plate of the maxilla is deep and fairly massive, not narrow as VOMER in FR 4922, but similar to Podocnemis . The maxilla forms the apertura narium externa There is no vomer present nor are there anteriorly and it is wider at its base than in sutural surfaces remaining for a vomer. How­ FR 4922, similar to most other pelomedu­ ever, the bone edges in this area are not entirely complete, and the morphology surrounding the apertura narium interna is very close to that in Foxemys , which has a well­developed vomer. Thus it is quite possible that one was present in Kurmademys .

PALATINE

Both palatine bones are present in ISI R 152, but are missing some of the anterior edges that form the margin of the apertura narium interna and the possible vomerine contact.

The anterolateral part of the palatine in Kurmademys contacts the maxilla and forms the posteromedial part of the triturating surface. The triturating surface is a low platform that ends in a V­shaped margin completely formed by the palatine. Foxemys is similar to Kurmademys in this area. The palatine forms the posterior margin of the apertura narium interna, but most of this margin is missing in ISI R 152. The choanal grooves are barely discernible in Kurmademys ; they are better defined in Foxemys . The foramen palatinum posterius is formed in the palatine­pterygoid suture by both bones ( fig. 4), as in Foxemys and in contrast to FR 4922, Pelusios , Pelomedusa , and Podocnemis , in which most of the foramen is in the palatine. As in other pleurodires, there is a median contact with the other palatine and a transverse, posterior contact with the pterygoid. On the dorsal surface both right and left palatines are visible and free of matrix. The palatine forms the medial part of the orbital floor and the lateral margin of the large foramen orbito­nasale.

Posteriorly the palatine of Kurmademys has a large dorsal process forming the lateral wall of the pterygo­palatine channel. This process contacts the jugal laterally, the postorbital dorsolaterally, and the parietal dorsomedially. The process tapers dorsally, so that its medial edge is higher than its lateral edge. The medial edge forms that lateral margin of the adductor channel. This dorsal process of the palatine reaches the parietal in the postorbital wall, which is quite unusual and has not been found so far in other pelomedusoids. The posterior wall of the orbit is complex in pelomedusoids and particularly so in Kurmademys . Behind the dorsal process of the palatine is the ventromedial process of the postorbital and a short ventral process of the parietal, all visible on the posterior surface of the postorbital wall.

QUADRATE

Most of both quadrates are complete and free of matrix. Part of the medial area of the left quadrate is broken and partially restored with something awful.

The squamosal lies at the posterolateral corner of the quadrate and its relations and contacts with the quadrate in Kurmademys are similar to those in Pelusios and Pelomedusa . The quadrate exposure along the lateral edge of the temporal emargination prevents contact of squamosal and postorbital. Anteriorly the quadrate contacts the quadratojugal, but the contact is relatively small because of the extensive temporal and cheek emargination. The quadratojugal contact is smaller in Kurmademys than in Pelusios , Pelomedusa , or FR 4922.

Most of the quadrate is involved in the formation of the cavum tympani and its two spaces, the antrum postoticum and the precollumellar fossa. The antrum postoticum of Kurmademys is unusually large for pelomedusoids; it is as large as the antrum in Emydura , the presumed primitive condition for pelomedusoids. The antrum of Kurmademys is swollen to completely fill the area inside the space formed by squamosal and quadrate. The precollumellar fossa is also deep and very large in Kurmademys , and also comparable in size to primitive chelids. However, other pelomedusoids, such as Pelusios and Pelomedusa , also have a large precollumellar fossa. In many features, the cavum tympani of Kurmademys is more primitive than in other bothremydids.

The other feature of interest in the cavum tympani is the incisura columellae auris, which still has the stapes present in the right quadrate of ISI R 152. The incisura is reduced to a completely closed, small foramen containing only the stapes in Kurmademys , in contrast to the open incisura of Foxemys . Arenila , Zolhafah, Bothremys , and Taphrosphys have the closed incisura as in Kurmademys but in Araripemys, FR 4922, and chelids, it is open. The combination of a completely closed incisura columellae auris with a gigantic antrum postoticum is a combina­ the exoccipital. The basioccipital contact of tion unknown so far in pleurodires. The the quadrate characterizes the Bothremydiquadrate of Kurmademys has a kidney­ dae plus Podocnemididae , but the exoccipital shaped cavum tympani as in Bothremys . Al­ contact is more restricted, found so far only though the incisura columellae auris is com­ in the Bothremydidae . pletely closed by bone, behind it is a groove The medial contacts of the quadrate informed in the quadrate for the eustachian clude the opisthotic and prootic as in other tube. Dorsally the groove slopes up to a hor­ turtles. The roof of the fenestra postotica (adizontal, straight­edged ridge that separates itus canalis stapedio­temporalis) has a low the eustachian tube surface from the opening parasagittal ridge showing the passage of the for the stapedial artery in the fenestra pos­ more lateral stapedial artery from the more totica. medial (and ventral) lateral head vein. This

The medial contact of the quadrate is with ridge is largely formed by the quadrate, with the prootic anteriorly. As in nearly all other the opisthotic contributing medially. The feturtles, the prootic and quadrate form the fo­ nestra postotica of Kurmademys is subdividramen stapedio­temporale. The foramen sta­ ed by bony partitions, so that the lateral head pedio­ temporale is certainly placed more an­ vein and stapedial artery are separated from teriorly in Kurmademys than in Emydura , but the more medial parts of the fenestra postonot any more anteriorly than in FR 4922 or tica. The opisthotic forms this wall dorsally Pelusios and Pelomedusa . The canalis sta­ and the quadrate forms it ventrally. On both pedio­temporalis can be followed posteriorly sides of ISI R 152 the contact of these two on the right side to the aditus canalis stape­ bones is broken, and a small amount of comdio­temporalis, which is partially divided pression is visible on the left side. The quadfrom the rest of the fenestra postotica, show­ rate also forms the ventral margin of a small ing the entry of the stapedial artery into the opening medial to the one just described, but skull. A separate canal for the stapes itself is lateral to the foramen jugulare posterius, present from the incisura columellae auris to which seems to be a remnant of a more open the aditus canalis stapedio­temporalis as in fenestra postotica. In the Bothremydidae in Bothremys and Taphrosphys . general the fenestra postotica is strongly sub­

Behind the prootic there is a contact with divided and separated by bony partitions. In the supraoccipital in ISI R 152 ( fig. 3A), that Kurmademys the bony partitions are thinner intervenes between the usual opisthotic­pro­ and some are probably represented by cartiotic contact. This contact also occurs in all lage or thin bone, allowing the collapse and other Bothremys Group taxa. The opisthotic breakage of foramen edges during fossilizacontact appears to be reduced in favor of ex­ tion. There is a well­developed quadrate­expansion of the supraoccipital contact when occipital contact medially as in Bothremys ; compared with a more generalized pleurodire this is considerably more extensive than the like Pelusios . The quadrate­opisthotic con­ small contact in FR 4922. tact is anteromedial to posterolateral in Kur­ The presumed foramen chorda tympani inmademys, between the supraoccipital and ferius is present on the posterior surface of squamosal. the processus articularis, roughly similar in

On the ventral surface the quadrate in Kur­ position to Podocnemis . Because of glue and mademys contacts the pterygoid anterome­ crud on the right quadrate, this is only visible dially from the base of the condylus mandi­ on the left. bularis along the anterolateral edge of the processus articularis, as in most pleurodires, PTERYGOID such as Emydura and Pelusios . Kurmademys has a medial process of the quadrate as in Both pterygoids are preserved in Kurmaother pleurodires, that contacts a narrowly demys and both are nearly complete. Most of exposed prootic, and broadly contacts the ba­ the thin pterygoid flange extending ventrally sisphenoid. Behind the basisphenoid, the from the quadrate process is missing from quadrate has a broad contact with the basi­ both pterygoids; the right one is more preoccipital. Dorsal to that the quadrate contacts served than the left.

On the ventral surface the pterygoid contacts the palatine in a roughly transverse suture that trends slightly anterolaterally. The foramen palatinum posterius is formed in the palatine­pterygoid suture as in nearly all bothremydids. Medially the pterygoids meet on the midline for a bit less than half their length. They are separated posteriorly by the triangular basisphenoid.

As in all pleurodires, there is a laterally projecting processus trochlearis pterygoidei. In Kurmademys the processus does not extend at a sharp right angle as in the Santana bothremydids (Gaffney et al., in press), but is only slightly less than a right angle, much as in Foxemys . It is not as acute as in chelids and Araripemys . The flange or web that extends ventrally from the base of the processus trochlearis pterygoidei along the quadrate process in all pleurodires is mostly missing in Kurmademys . The portion preserved is consistent with that seen in other bothremydids.

The posterolaterally extended quadrate processus in Kurmademys is narrower and longer than in FR 4922, Araripemys , pelomedusids, and chelids. In these groups the process is relatively flat and more horizontal, while in Kurmademys and some other bothremydids, such as Foxemys , it is narrower and more vertical. This condition seems to be related to the presence of a ventrally concave depression in the posterolateral part of the pterygoid in these forms ( fig. 4). This depression is presumably the pterygoideus muscle attachment site. In Kurmademys the depression is shallower than in Nigeremys and Foxemys , but it covers a larger area. Its margins are not as well defined in Kurmademys as it is in those taxa. There is no development of an overhang of this depression by the pterygoid as in the Podocnemididae .

The foramen posterius canalis carotici interni in Kurmademys lies entirely within the basisphenoid ( fig. 3B); the pterygoid does not participate in its formation as it does in many other bothremydids. The Kurmademys condition is unique within pelomedusoids. The posterior margin of the pterygoid contacts the narrowly exposed prootic between the basisphenoid and quadrate contacts.

Most of the dorsal surface of the pterygoid is visible in Kurmademys , although the region inside the cavum cranii is variably obscured by pieces of matrix. The crista pterygoidea is relatively low. The pterygoid forms the ventral margin of the foramen nervi trigemini as in other bothremydids, but the foramen is not placed very close to the foramen stapedio­temporale as it is in many other bothremydids, such as Bothremys and Foxemys . The pterygoid forms the floor of the sulcus palatinopterygoideus, which lies between the side wall of the cavum cranii and the processus trochlearis pterygoidei.

The anterior contacts of the pterygoid at the base of the processus trochlearis pterygoidei are visible on both sides of ISI R 152. The pterygoid plus the palatine and jugal form the postorbital wall, as exposed posteriorly in the adductor muscle chamber. The pterygoid has a very narrow contact with the parietal medially, broader contacts with the postorbital more laterally, and with the jugal most laterally.

SUPRAOCCIPITAL

The supraoccipital in ISI R 152 is complete ventrally and anteriorly, but is missing the posterior part of the crista supraoccipitalis.

The supraoccipital in turtles is Y­shaped, with paired lateral projections forming the medial part of the cavum labyrinthicum on each side. In cryptodires and most pleurodires the supraoccipital has a tripartite suture, with the prootic and opisthotic visible on the dorsal surface of the otic chamber. It is unusual to find that in a group of bothremydids—the Bothremys Group of Lapparent de Broin and Werner ( 1998)—the supraoccipital contacts the quadrate and separates the prootic from the opisthotic. In Kurmademys this unusual condition is present. The supraoccipital on the right side has a broad contact with the quadrate laterally and separates the prootic from the opisthotic. On the left side the supraoccipital is complete and separates the prootic and opisthotic, but the quadrate is damaged. This degree of quadrate contact by the supraoccipital is similar in Kurmademys, Bothremys , Foxemys , and Rosasia . The contact is absent in Taphrosphys and indeterminate in Zolhafah , Nigeremys , and Arenila .

The crista supraoccipitalis is usually relatively short in bothremydids. In Kurmademys it is broken posteriorly and its length is in­ occipital. A shallow, median concavity lies determinate. between the paired tuberculae and is formed

almost entirely by the basioccipital.

EXOCCIPITAL Both exoccipitals are preserved and only PROOTIC

lack the condylus occipitalis, although there

Both prootics in ISI R 152 are preserved;

is some breakage around the posterior foram­

the right one is nearly complete and the left

ina.

one has a partially eroded dorsal surface.

The exoccipital contacts the supraoccipital

The prootic is exposed on the dorsal and

dorsally, the opisthotic dorsolaterally, the

anterior surface of the otic chamber with the

quadrate ventrolaterally, and the basioccipital

following contacts: the parietal medially, the

ventrally. The quadrate­exoccipital contact

quadrate laterally, the supraoccipital posteri­

occurs in all Bothremydidae and is absent in

orly, and the pterygoid ventrally. There is no

other pleurodires.

prootic­opisthotic contact in Kurmademys .

Dorsomedially the exoccipital forms the

The foramen stapedio­temporale is formed in

lateral and ventral margin of the foramen

the prootic­quadrate suture. In contrast to all

magnum. Ventromedially the exoccipital pre­

other bothremydids, the foramen opens an­

sumably participates in the formation of the

terodorsally rather than anteriorly. It is visi­

condylus occipitalis, but this structure is bro­

ble in dorsal view in Kurmademys , but in

ken off at its base in ISI R 152, so the pres­

other bothremydids it is barely or not visible

ence or absence of the basioccipital in the

in dorsal view. The position of the foramen

condylus is not determinable. The foramen

stapedio­temporale in Kurmademys is very

jugulare posterius in Kurmademys is formed

similar to that in pelomedusids and chelids.

entirely by the exoccipital. The bone sur­

The foramen nervi trigemini is formed by the

rounds most of the foramen, but on each side

prootic dorsolaterally, the parietal dorsome­

the foramen is open laterally due to the pres­

dially, and the pterygoid ventrally. The fo­

ence of a narrow fissure. This fissure is dif­

ramen is best preserved on the right side; the

ferent in shape on both sides and may be due

left foramen nervi trigemini is larger and has

to breakage, in which case the original con­

broken edges.

dition of the foramen would be closed. Kur­

On the ventral surface of ISI R 152, the

mademys has two foramina nervi hypoglossi

prootic is exposed where the three bones (the

as in all other pelomedusoids; they lie near

pterygoid, basisphenoid, and quadrate) meet

the base of the condylus occipitalis, ventro­

( fig. 4). This is in the deepest part of the

lateral to the foramen magnum. The more

concavity formed by these bones. The con­

dorsal foramen is formed entirely within the

cavity is presumed to be for the pterygoideus

exoccipital, but the more ventral one is

muscle attachment. The prootic is exposed

formed in the exoccipital­basioccipital su­

and has a very similar shape on both sides,

ture.

so this is not interpreted as a consequence of

preservation or an artifact. The prootic has a

BASIOCCIPITAL

distinct foramen, here interpreted as the fo­

The basioccipital in ISI R 152 is nearly ramen nervi facialis (VII), for the facial complete. A small amount of breakage is vis­ nerve, always closely associated with the ible on each tuberculum basioccipitale and prootic ossification. The form of the prootic the broken condylus occipitalis does not exposure in Kurmademys is not like that in clearly show basioccipital sutures. any other pleurodire. The primitive condition The basioccipital of Kurmademys has a of the ventral prootic exposure occurs in broad, transverse contact with the basisphe­ chelids, pelomedusids, and Araripemys , all noid anteriorly. Posterolaterally the basioc­ of which have a large prootic exposure with cipital contacts the quadrate, and posterodor­ the foramen posterius canalis carotici interni sally there is a broad contact with the exoc­ in the prootic. In all bothremydids, podoccipitals. The tuberculum basioccipitale is nemidids, and FR 4922 the prootic does not formed about equally by the quadrate and ex­ contain the internal carotid as it does in chelids and pelomedusids. In FR 4922 the prootic is partially exposed in a narrow space between the basisphenoid and quadrate, similar to the primitive position found in chelids and pelomedusids, but is distinctly posterior to the prootic, as exposed in Kurmademys . It is likely that the prootic exposure in Kurmademys is not a retention of a primitive state, but is an autapomorphy of this genus, probably related to the development of the pterygoideus concavity that removed covering elements. Small areas of the prootic are variably exposed in other bothremydid skulls due to erosion or the development of the concavity.

OPISTHOTIC

Both opisthotics are preserved in ISI R 152; the right one is nearly complete and the left one is missing a small part anteriorly.

In dorsal view, the opisthotic has these contacts: supraoccipital anteromedially, the squamosal laterally, and the exoccipital posteromedially. There is no prootic­opisthotic contact. The opisthotic in Kurmademys ends posteriorly at about the same level as the squamosal; it does not extend posteriorly beyond the squamosal as in pelomedusids, Araripemys , and FR 4922. Ventrally the opisthotic forms the roof and some of the subdivisions of the fenestra postotica. In ISI R 152 the opisthotic divides the fenestra postotica into two portions: a more lateral one that seems to have contained the stapedial artery and lateral head vein, and a more medial one with unknown contents. The medial foramen has a finished dorsal margin, although as preserved, the ventral margin is broken. Presumably the more medial foramen just held cartilage as in living turtles with an open fenestra postotica (Gaffney, 1979).

BASISPHENOID

The basisphenoid is complete and well preserved in ISI R 152. The cavum cranii is largely free of matrix and some of the dorsal surface of the basisphenoid is visible.

The basisphenoid in Kurmademys is not strongly triangular as in other bothremydids but is more pentagonal. It is a relatively large element, wider than long. The anterior contact with the pterygoids trends posterolaterally and anteromedially, and the angle this suture makes with the midline is similar to that in Foxemys and Zolhafah . At the anterolateral corner of the basisphenoid, between the pterygoid and quadrate contacts, is a short contact with the prootic. In nearly all bothremydids the prootic is covered, so this contact is unusual. The lateral margin of the basisphenoid is a long, parasagittal contact with the quadrate. This contact in Kurmademys is longer than in any other bothremydid; Foxemys and Polysternon most closely approach it. Posteriorly the basisphenoid has a transverse contact with the basioccipital.

In contrast to all other bothremydids, in Kurmademys the foramen posterius canalis carotici interni is formed completely by the basisphenoid, without participation of the pterygoid. However, the foramen is very close to the pterygoid suture, particularly on the right side. The foramen posterius canalis carotici interni in Kurmademys is also placed farther anteromedially than in any other bothremydid. This could be explained morphologically by a reduced ossification of the canalis caroticus internus posteriorly. The canalis in all bothremydids travels anteromedially and slightly dorsally to enter the sella turcica. If the canalis in a form like Bothremys (which has the foramen posterius canalis carotici interni placed far posterolaterally) were to be exposed by the removal of bone ventrally, the foramen would appear to migrate anteromedially along the path of the canalis caroticus internus. It is possible that this condition could result from the development of a deep pterygoideus muscle concavity, which is formed directly ventral to the canalis caroticus internus. Although Kurmademys has a distinct pterygoideus concavity, it is relatively shallow compared to such forms as Foxemys , and Foxemys does not have the foramen posterius canalis carotici interni placed anteromedially.

The dorsal surface of the basisphenoid in Kurmademys has the dorsum sellae and sella turcica visible. The dorsum sellae overhangs the sella turcica and has the foramen anterius canalis carotici interni also hidden in dorsal view and lying at the posterolateral corner of the sella turcica. There is a small processus clinoideus on the left side; the right one is broken. The shape and general proportions of the basisphenoid. Foxemys , Polysternon, Rothe dorsum sellae and sella turcica are similar sasia, Zolhafah , and Bothremys can be united to those in Pelusios . The degree of overhang by having a foramen stapedio­temporale very of the dorsum sellae, however, is greater in close to the foramen nervi trigemini and a Kurmademys than it is in Pelusios . The ros­ broad preorbital part of the skull, features abtrum basisphenoidale is fused into a single sent in Kurmademys .

structure, but its anterior end shows the two Kurmademys has characters in common ossified trabeculae rather than the single ros­ with the Bothremys Group, but the state of trum seen in Pelusios . There is no sign of a the current analysis shows only a few steps foramen nervi vidiani in the left sulcus cav­ from Kurmademys as the sister group to the ernosus. The right side still has some matrix. Nigeremys Group plus Bothremys Group. The addition of as yet undescribed taxa will RELATIONSHIPS help to resolve its phylogenetic position.