ZOBODAT - www.zobodat.at

Zoologisch-Botanische Datenbank/Zoological-Botanical Database

Digitale Literatur/Digital Literature

Zeitschrift/Journal: Carolinea - Beiträge zur naturkundlichen Forschung in Südwestdeutschland

Jahr/Year: 2006

Band/Volume: 63

Autor(en)/Author(s): Buchy Marie-Céline

Artikel/Article: An Elasmosaur (Reptilia: Sauropterygia) from the Turanian (Upper Cretaceous) of Morocco 5-28 ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at carolinea, 63 (2005): 5-28, 11 Abb.; Karlsruhe, 15.03.2006 5

An Elasmosaur (Reptilia: Sauropterygia) from the Turanian (Upper Cretaceous) of Morocco

M a r ie -C éline Buchy

Abstract qui pourraient indiquer une pathologie liée à un âge The subcomplete skeleton of an elasmosaurid plesio­ avancé. Certains aspects de l’anatomie des élasmo­ saur recently acquired by the Staatliches Museum für saures suggèrent qu’ils s’alimentaient partiellement Naturkunde Karlsruhe (State Museum of Natural His­ aux dépends des invertébrés du fond marin. tory in Karlsruhe) is described. The specimen comes from the early Turanian of the Goulmima region, Mo­ Autor rocco, and represents one of the most complete elas- Marie -Céline Buchy , Universität Karlsruhe, Geolo­ mosaurs known in the world. Owing to the present gisches Institut, Postfach 6980, D-76128 Karlsruhe, uncertainties about elasmosaurs systematics, it is Germany. Current address: Geowissenschaftliche referred mainly on stratigraphical grounds to a new Abteilung, Staatliches Museum für Naturkunde, Erb­ species of the genus Libonectes. Some of the bones prinzenstrasse 13, D-76133 Karlsruhe, Germany; of the holotype (e.g. cervical vertebrae, ulna) show an [email protected] unusual type of preservation that might indicate an old-age related pathology. Some aspects of elasmo­ Introduction saurs anatomy suggest non obligatory feeding upon Elasmosaurs form a family of plesiosaurs cha­ sea floor invertebrates. racterised by an elongate neck bearing a small, brevirostrine head. Elongation of the neck is Kurzfassung Ein Elasmosaurier (Reptilia: Sauropterygia) aus achieved by both elongation of the vertebrae and dem Turon (Oberkreide) von Marocco increase in vertebral count. They appear in the Das fast vollständige Skelett eines elasmosauriden fossil record during the Early Jurassic of Europe Plesiosauriers, welches kürzlich vom Staatlichen Mu­ and are known world-wide by the Late Cretace­ seum für Naturkunde Karlsruhe erworben wurde, wird ous (Persson 1963, Ba r d e t 1995, Ba r d e t et al. beschrieben. Das Stück stammt aus dem frühen Turon 1999, Lazo & C ichow o lski 2003), though they der Goulmima-Region Marokkos und ist eines der voll­ are most abundant in the Upper Cretaceous se­ ständigsten Elamosauridenskelette der Welt. Wegen diments of North America (e.g. W illiston 1890, der gegenwärtigen Unklarheiten in der Elasmosaurier- systematik wird das Stück hauptsächlich aus stratigra­ 1906, W elles 1962, C arpenter 1999, Sto rrs fischen Gründen einer neuen Art der Gattung Libonec­ 1999, G asparini et al. 2003, Ever har t in press). tes zugeordnet. Einigen Knochen des Holotyps (z.B. The record of elasmosaurs from Africa is limited Halswirbel, Ulna) zeigen eine ungewöhnliche Art der to fragmentary remains (for a review see W erner Erhaltung, welche auf alterspathologische Ursachen & B a r d e t 1996) and some unspecified remains hindeutet. Einige Aspekte der Elasmosaurieranatomie from the area of Goulmima, Morocco (Ba r d e t et legen die Möglichkeit nahe, dass die Tiere gelegentlich al. 2000, Ba r d e t pers. comm.). Meeresbodeninvertebraten gefressen haben. Goulmina has yielded a rich fish assemblage Résumé (C avin 1995, 1997, 1999, 2001, C avin et al. Un Élasmosaure (Reptilia: Sauropterygia) du 2001) and marine reptiles (Ba r d e t et al. 2003a, Turonien (Crétacé supérieur) du Marocco b). Most described specimens were acquired via Le squelette subcomplet d’un plésiosaure élasmosau- fossil dealers, their exact origin being unknown. ride récemment acquis par le Staatliches Museum für Several workers undertook field work in order to Naturkunde Karlsruhe (Muséum d’Histoire Naturelle define the provenance of the vertebrate-bearing de Karlsruhe) est décrit. Le spécimen provient du Tu­ nodules (C avin 1999, C avin et al. 2001, Ba r d e t et ronien inférieur de la région de Goulmima, Maroc, et al. 2003a) and concluded that the nodules occur représente l’un des élasmosaures les plus complets at the top of the Cenomanian-Turonian section, au monde. En raison des incertitudes entourant à ce jour la systématique des élasmosaures, il est attribué, in the Unit 4 defined by F errandini et al. (1985). surtout sur des arguments stratigraphiques, à une This unit is dated by the ammonite Mammites to nouvelle espèce du genre Libonectes. Certains des os the early Turanian and was deposited under open de I holotype (par exemple des vertèbres cervicales, platform conditions at the maximum of the Ceno­ I ulna) présentent un type de préservation inhabituel manian-Turonian transgression (F errandini et ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 6 carolinea, 63 (2005)

al. 1985, C avin 1999, Ba r d e t et al. 2003a). Ma­ age of the fossil (Chancellor et al. 1994, C. If- rine reptiles from Goulmima include mosasaurs RiM pers. comm.). and plesiosaurs (Ba r d e t et al. 2000, 2003a, b, The specimen is currently on exhibition at the Buchy et al. 2005, Ba r d e t pers. comm.). The SMNK, resting on its ventral side (Plate 1). Most specimen herein described is the first diagnostic elements of the girdles are therefore inaccessible elasmosaur of the assemblage, and one of the for study. The length of the skeleton as preserved most complete and best preserved elasmosaurs is about 7.2 m. Mechanical preparation was com­ known in the world. pleted prior to purchase by the SMNK. For the non-preserved right limb elements distal to the Institutional abbreviations propodials, casts of the left ones were substituted ANSP: Academy of Natural Sciences, Philadel­ (Plate 1).The left forelimb is mostly genuine (Fig. phia, USA; AMNH: American Museum of Natural 8), although the distal-most portion probably be­ History, New York, USA; CIT: California Institute longs to another limb (see Description). The fifth of Technology, Pasadena, USA; DMNH: Denver metatarsal (e.g. W elles 1943, C aldw ell 1997, Museum of Natural History, Denver, USA; FMNH: Sato 2003) situated cranially indicates that the Field Museum for Natural History, Chicago, USA; hindlimb placed on the left side of the specimen KUVP: Kansas University, Vertebrate Paleontolo­ is either visible in ventral aspect or is actually the gy, Museum of Natural History, Lawrence, USA; right one (Fig. 9). The proximal two thirds of this MSGB: Museo del Servicio Geológico, Bogotá, hindlimb are genuine; its distal third is a cast of Colombia; NZGS: New Zealand Geological Sru- the distal extremity of the left forelimb (see De­ vey, Lower Hutt, New Zealand; RSM: Royal Sas­ scription). Except for a few aesthetic patches of katchewan Museum, Regina, Canada; SMNK: coloured resin, mainly in the skull and mandible Staatliches Museum für Naturkunde Karlsruhe; (Figs 1, 2, 4), the rest of the skeleton appears SMNS: Staatliches Museum für Naturkunde genuine. Stuttgart; SMUSMP: Southern Methodist Univer­ All preserved neural arches are fused, indicating sity, Shuler Museum of Natural History, Dallas, an adult individual sensu Brown (1981). USA; UCMP: University of California Museum of Numerous cervical vertebrae of the caudal three- Paleontology, Berkeley, USA; UNSM: University fourths of the neck, some ribs and long bones of Nebraska State Museum, Lincoln, USA. (Figs 7-9, Table 1) are incompletely preserved. When the affected element is not distorted, a Material and Preservation brownish area that contrasts with the surroun­ The specimen (Plate 1, Figs 1-9) was obtained ding light beige matrix replaces the missing bone via a fossil dealer in the autumn of 2002, together substance as if it were a natural cast. The darker with the polycotylid plesiosaur SMNK-PAL 3861 brown edge of the damaged bone, where ex­ (Buchy et al. 2005); it is housed in the SMNK posed, has a finished, smooth aspect. Other ele­ under accession number SMNK-PAL 3978. It ments are distorted as if they were soft in places consists in a subcomplete, articulated, 3-dimen- when diagenesis occurred, e.g. cervicals wedged sionally preserved skeleton lacking parts of the between adjacent cervicals, without broken bone left limbs and the terminal caudal vertebrae; both fragments (Fig. 7). This uncommon preservation right limbs are missing except for the humerus is interpreted as due to a probable pathology that and femur. The locality data associated with might indicate an old adult (see Discussion). SMNK-PAL 3978 are the same as for SMNK- PAL 3861 (Buchy et al. 2005): both come from Systematic Palaeontology ‘some kilometres around Goulmima’, Morocco. Sauropterygia O w e n , 1860 The calcareous matrix surrounding SMNK-PAL Plesiosauria de Bla in v ille , 1835 3978 is similar to that surrounding SMNK-PAL Plesiosauroidea W e lle s , 1943 3861, which agrees with the vertebrate-bearing Elasmosauridae C o p e , 1869 nodules known from the early Turonian of the Libonectes C a r p e n t e r , 1997 Goulmima area (Cavin 1995, 1997, 1999, 2001, Libonectes atlasense n. sp. Cavin et al. 2001, Ba r d e t et al. 2003a, b, Buchy et al. 2005, Ba r d e t pers. comm.). The presence Holotype - SMNK-PAL 3978, skeleton lacking of the ammonite Fagesia superstes (Ko ssm at , both right limbs distal to the propodials, parts of 1897) preserved with SMNK-PAL 3978 confirms the left limbs and the terminal caudal vertebrae the late early Turonian to early middle Turonian (Plate 1, Figs 1-9). ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turonian of Marocco 7

Type horizon and locality - Lower Turonian (Up­ chin extends until level with the rostral end of the per Cretaceous) Unit 4 of the Cenomanian-Turo- maxilla. 52-53 cervical and 5-7 pectoral verte­ nian limestone bar, north of Goulmima, Errachi- brae. The axis neural arch is 1.5 time higher than dia Province, Morocco (Ferrandini et al. 1985; its centrum; the axis postzygapophysis entirely Cavin 1999: fig. A; Bardet et al. 2003a: fig. 1). overhangs the centrum of the third cervical. Etymology - Latin, from the north African Atlas Referred specimen - SMNS 81783, poorly pre­ (Mountain Range). served skull and mandible (Fig. 10), from the Diagnosis - An elasmosaur differing from the type same locality and age as the holotype. specimen of Libonectes (Carpenter 1997,1999) in the following features: The ventral margin of Description the upper cheek is concave between premaxillary Axial skeleton tooth 3 and maxillary tooth 3 on the right side, Skull (Figs 1, 2, 4, 5) and 5 on the left. The naris is situated level with The skull is typically elasmosaurian, comprising maxillary teeth 4-5, which are the largest teeth a preorbital section (about 38% of skull length) in the upper tooth row. The mandibular rostrum slightly shorter than the postorbital section, comprises 3 teeth; a ventrocaudally oriented, flat which is distinguished by a high parietal crest.

Table 1. Bones of SMNK-PAL 3978 probably affected by a pathological decrease in bone density; location of the lesions. Most of the girdles and dorsal vertebrae being obscured by matrix or currently inaccessible, the list is not exhaustive. Lesions preceded by an asterisk are not clearly visible but suspected because of important distortion of the bone that cannot be explained by diagenesis (see Preservation and e.g. Fig. 7).

Bone Place of lesion left humerus *linear, at the distal extremity of the shaft, transverse to the shaft left ulna linear, in the middle of the shaft, transverse to the shaft left femur ventral half proximal to at least mid shaft right femur cranioproximal portion until at least mid shaft /right' tibia *linear, in the middle of the shaft, transverse to the shaft ,riqht‘ fibula ‘ linear, in the middle of the shaft, transverse to the shaft C15 left dorsocaudal corner of centrum and base of neural spine C16 caudal articular margin of centrum C17 ‘ oblique line ventrocaudally directed through centrum C23 ‘ ventrocaudal corner of centrum C24 ‘ caudal articular margin of centrum C26 ‘ middle of centrum C28 caudal articular margin of centrum and left pedicel C29 dorsocaudal corner of centrum and neural arch in continuity (except caudal margin of the arch) C30 ‘ cranioventral margin of left rib C32 caudal articular margin of centrum C33 cranial articular margin of centrum C34 dorsal portion of caudal articular margin of centrum C35 ‘ rectangular lesion in left dorsocaudal corner of centrum C36 ‘ centrum C38 ‘ centrum C39 cranial articular margin of centrum C40 cranial articular margin of centrum and cranial portion of neural arch C41 ‘ right dorsocaudal corner of centrum C42 ‘ right side of centrum and neural arch C46 middle of caudal left articular margin of centrum C47 cranial articular margin of centrum C48 caudal articular margin of centrum C51 cranial articular margin of centrum C57 base of left neural arch right rib 63 in the broken area (see text) right rib 64 transverse area 20 mm proximal to articular head ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 8 carolinea, 63 (2005)

The skull is ogival in dorsal aspect, and triangu­ PAL 3978 are treated here. Measurements on lar in lateral aspect (Figs 1,2). W illisto n (1890) the skull and mandible are given in Table 2. and numerous others since (e.g. W elles 1952, The left premaxilla houses 4 teeth, the right one C arpenter 1999, Sato 2003) complained of the 5. The premaxillomaxillary suture commences poor preservation of sutures and osteological at the rostral-most point of the nans; it smoothly details in most elasmosaurian skulls. SMNK-PAL curves laterally and ventrally (Figs 1,2). On both 3978 is no exception. Although the skull is only sides, its ventral-most extent is obscured by the slightly sheared rightward (Fig. 4), it is impossible third visible dentary tooth and adjacent matrix. to determine the course of many of the sutures Two ridges parallel the interpremaxillary suture due to numerous breaks. This is especially the (the dorsal midline) on the surface of the left pre­ case in the interorbital area, where no individual maxilla for about half the rostrum length. The pre­ bone can be distinguished with certainty, and no maxilla forms the medial margin of the naris, but suture distinguished from breaks (Figs 2, 4). The the extent of the bone in the interorbital area can­ lateral shear caused breakage to the dorsolateral not be determined. The ventral margin of the right portions of the squamosal in a symmetrical man­ upper jaw is concave from premaxillary tooth 3 to ner (Figs 1, 2); this breakage pattern, known in maxillary tooth 3. The ventral concavity of the left several specimens, led some authors to suggest upper jaw extends from premaxillary tooth 3 to the presence of a quadratojugal (W illiston 1890, maxillary tooth 5. On both sides, the ventral mar­ W elles 1943, 1949, 1952, 1962). Flowever, no gin of the upper jaw is obscured by matrix caudal quadratojugal is nowadays considered existing in to the rostral margin of the orbit, that is, caudal to plesiosaurs (e.g. W illiston 1925, Ro m er 1956, maxillary tooth 9 (Fig. 5). Brown 1981, Sto rrs 1997, C a r penter 1997). The right naris is oval in dorsal aspect, its long Due to the same distortion, the suspensorium of axis oriented rostrolaterally; the original outline SMNK-PAL 3978 is inclined rostroventrally (Fig. of the left one cannot be determined. Both are 1); given the general pattern of deformation, that situated level with maxillary teeth 4-5. On the left affected the skull in a dorsoventral and rightward side, the 7 rostral-most maxillary teeth alterna­ direction (see Fig. 4), it was probably subvertical te with dentary teeth; another maxillary tooth is in vivo. visible ventral to the orbit; all other left maxillary Owing to these uncertainties about the location teeth are obscured by matrix and dentary teeth. of the sutures, the present description considers Similarly, on the right, 6 maxillary teeth alternate the skull as a whole instead of focusing on indi­ with dentary teeth, and a further maxillary tooth vidual bones. Moreover, the skull appears very is visible ventral to the orbit. The maxilla appears conservative in the group (Car p e n te r 1999), to form most of the lateral margin of the naris; the and was described in detail by W illiston (1890), caudal margin of the naris is formed by the pre­ W elles (1943) and C a r penter (1997) among frontal. Though, the suture between the maxilla others; therefore only features specific to SMNK- and prefrontal is only visible on the left, where the outline of the naris is unclear. The maxillop- refrontal suture runs to the rostral margin of the Table 2. Measurements in mm of the skull and mandible orbit, whose rostrolateral corner is formed by the of SMNK-PAL 3978. Definition of the measurements was maxilla. The jugal appears to form most of the discussed by Buchy et al. (2005: Appendix); here, owing ventral margin of the orbit; the suture of this bone to the lateral distortion of the skull, the skull length is with the maxilla runs caudoventrally but is soon measured between the rostral extremity of the skull and obscured by teeth and matrix. The rostromedial the caudal-most margin of the quadrate condyle. margin of the orbit is formed by the prefrontal. skull length 400 The medial margin of this bone runs from the temporal fenestra length 165 caudal-most point of the naris to the centre of the orbit maximal diameter 57 medial margin of the orbit. preorbital length 155 A break affecting the centre of the right postorbi­ tal bar, and a similar, symmetrical structure on the skull width 185 left side probably occurred at the medial margins interorbital minimal width 60 of the postorbitals, which, I conclude, form the mandible length 450 caudal margins of the orbits. The lateral margin mandibular symphysis length 40 of the postorbital (its suture with the jugal) runs maximal visible length of the dentary tooth row 270 caudally from the caudoventral corner of the left ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turonian of Marocco 9 orbit in lateral aspect, sigmoidal; it is preserved broken. Otherwise the teeth are well preserved, for about 30 mm. The temporal bar is broken in but only visible in labial view. Basal diameter and its middle on both sides; it is unclear whether this height of the teeth when determinable are given break can be interpreted as the caudal extremity in Table 3. The tooth count, occlusion pattern and of the jugal. location of the naris and orbit compared to maxil­ The parietal crest is parallel-sided in dorsal lary and dentary teeth are shown in Figure 5. aspect, and partly reconstructed; no parietal fo­ The rostral-most premaxillary tooth pair is stron­ ramen is visible, although it might not be identifi­ gly procumbent. The second premaxillary tooth able in the caudal portion of the poorly preserved pair is curved and points rostroventrally; suc­ interorbital area. The caudal extent of the parie­ cessive upper teeth become progressively more tal and its suture with the squamosal cannot be vertical. The dentary teeth are more uniform in determined. The contact between quadrate and squamosal is also invisible. As preserved, the caudal part of the skull is boomerang-shaped in dorsal aspect. Table 3. Basal diameter and height in mm of the tooth crowns of SMNK-PAL 3978. Italicised values are esti­ Mandible (Figs 1, 3-5) mate; crowns whose height is in brackets are recently The mandibular rostrum houses 3 tooth positions. broken, while the sign V following crown height indica­ tes a naturally worn apex. The mandibular symphysis commences level with left premaxillary tooth 4. In ventral aspect, it is vi­ Left Basal Height Right Basal Height sible running straight rostrally for about 20 mm. It tooth no diameter tooth no diameter cannot be traced further rostrally. From the rostral pmx 1 5 17 pmx1 6 21 extremity of the visible portion of the symphysis a 2 10 30+ 2 6 (17) ridge diverges rostrolaterally on either side. The 7 triangular area between these ridges is flat. In la­ 3 9 25+ 3 (13) teral aspect the rostral margin of the mandible 4 9 33+ 4 9 (21) is straight, caudoventrally oriented, and persists 5 8 (14) so until level with the rostral-most extent of the mx 1 7 15+ mx 1 5 (17) maxilla. The ventral margin of the mandible cau­ 2 7 24+ 2 4 15+ dal to this point is straight, horizontal. The dorsal 3 7 (20) 3 7 (10) margin of the lower jaw is slightly concave until 4 11 41 + 4 7 13+ about the level of the rostral margin of the supra- 5 10 39+ 5 8 23+ temporal fenestra; it is horizontal further caudally. 6 6 27+ 6 5 (16) Seventeen dentary teeth are visible on the left, 7 4 (10) 7 5 13+ 16 on the right plus one empty alveolus (Table 8 / (10) 8 3 (10) 3). The dentary is the only visible bone in lateral d 1 8 27+ d 1 7 (24) aspect until level with the middle of the tempo­ ral bar. There, a vertical break is continuous with 2 10 33 + 2 6 25+ the break affecting the temporal bar; the caudal 3 10 34+ 3 6 30+ extent of the dentary cannot be determined. The 4 3 (10) 4 5 21 + angulosurangular suture emerges caudal to this 5 9 (27) 5 6 (31) break at mid-height on the mandible and runs 6 2 (7) 6 8 30+ caudally; the dorsal margin of the angular then 7 7 17+ 7 6 20+ outlines the glenoid fossa. The dorsal margin of 8 / / 8 6 (24) the angular cannot be traced further caudally on 9 8 30+ 9 6 28 the lateral surface of the retroarticular process. 10 7 28+ 10 7 28 The rostral margin of the articular (its suture with 11 7 29 11 7 22+ the surangular) is not identifiable.The retroarticu­ 12 7 23 12 7 27+ lar process extends caudally until level with the centre of cervical vertebra 3. 13 6 14 13 / / 14 8 22+ 14 8 30+ Dentition (Figs 1-5) 15 6 13 15 7 22 The enamel is damaged on several teeth. Most 16 7 20+ 16 8 30+ of the largest teeth of the right row were recently 17 3 6 17 7 / ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 10 carolinea, 63 (2005) orientation; all point vertically and are slightly verse process cannot be defined. The length of curved. the cervicals and pectorals, where determinable, The enamel exhibits faint striae toward the base is given in Table 4. On the right side, a ridge is of the crowns on the labial face. All tooth crowns preserved at mid-height on the centra until C39; are subcircular to weakly oval in basal cross- on the left until C48 (the ridge is less expressed section; no carina is visible. Right premaxillary from C42 to C48). The cross-sectional shape of tooth crowns 1 and 5, maxillary crown 5, denta­ the vertebrae is unclear, as most breaks pass ry crowns 10 and 12, and possibly left maxillary through vertebrae rather than between them. crown 5 exhibit a basal band of wrinkled enamel, Where visible, the centra are weakly dorsoven- 1-2 mm in thickness, presumably all around the trally compressed in cross-section (Fig. 7). A crown base. This might mark a growth abnorma­ centrimetric intervertebral space is preserved lity, maybe due to a deficiency; further prepara­ until between C39 and C40, and probably bet­ tion could clarify whether it actually occurs mostly ween C40 and C41; further caudally, the articular on the right side and help establish a diagnosis. surfaces of subsequent vertebrae as preserved Most apices were either broken recently or worn appear separated by only a few millimetres. From in vivo (Table 3); the intact ones are rounded in­ C4, half the length of the postzygapohysis of each stead of sharp and taper abruptly. vertebra projects over the cranial margin of the succeeding centrum. Until C16 the neural spines Vertebrae and ribs (Figs 6, 7) are inclined caudally at an angle of 45° to the ho­ The atlas/axis complex is preserved close to its rizontal. Further caudally, the neural spines seem original position, but due to incomplete prepara­ to point vertically, but are only completely preser­ tion only its caudal portion is visible. On the right ved between C39 and C41. The neural spines of side, the base of both neural arches, part of the C41 to C43 are fan-shaped distally; whether this odontoid, and most of the axis and its cervical represents their original shape or is due e.g. to rib are exposed, the bone surface being poorly pathology or diagenesis is unknown. preserved. On the left, the axis neural arch and All cervical ribs are fused to the centra; the suture centrum and the proximal-most portion of the line is visible as a swollen line on C5 to C7 The cervical rib are visible (Fig. 6). The distal-most third cervical rib is subtriangular in lateral aspect. portion of the atlas neural arch appears to con­ Its distal extremity tapers caudally and reaches a tact the axis neural arch dorsocaudal to the zy- level 10 mm caudal to the cranial margin of the gapophyseal contact; adhering matrix, though, renders interpretation problematic. The suture of the atlas neural arch with the odontoid as visible Table 4. Length in mm of the cervical and pectoral cor­ on the right is almost straight and runs slightly pora of SMNK-PAL 3978. Italicised values are estimate. cranioventrally, forming a ridge; the suture bet­ No. Length No. Length No. Length ween the axis neural arch and its centrum ex­ (mm) (mm) (mm) tends caudally from the caudal-most point of the suture of the atlas neural arch with the odontoid. atax / C21 80 C40 90 C3 34 C22 69 C41 The axis neural arch extends caudally over the 100 C4 35 C23 75 C42 95 cranial half of cervical centrum 3; the postzyga- C5 34 C24 66 C43 85 pophysis projects entirely caudal to the axis cen­ C6 32 C25 65 C44 90 trum. The axis neural arch is 1.5 time higher than C7 40 C26 / C45 95 its centrum. A conspicuous, rounded ridge runs C8 38 C27 / C46 95 parallel to the dorsal margin of the neural arch C9 40 C28 80 C47 97 at mid-height. The right axis rib smoothly curves C10 42 C29 72 C48 80 distocaudally; it tapers distally and its distal ter­ C11 45 C30 79 C49 87 minus is rounded. C12 55 C31 75 C50 82 C32 Together with the atlas and axis, the neck com­ C13 55 80 C51 83 C14 60 C33 80 C52 65 prises at least 52 cervical vertebrae; vertebra 53 C15 69 C34 78 C53 79 is too poorly preserved to determine whether it is C16 60 C35 72 P1 69 a cervical or a pectoral. Vertebra 54 is a pectoral C17 70 C36 80 P2 68 and so are the following 5 vertebrae. The 60th C18 66 C37 80 P3 73 vertebra of the series is either the last pectoral C19 65 C38 78 P4 80 or first dorsal, but the insertion area of the trans­ C20 70 C39 90 P5 80 ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 11 centrum of C4. A break runs through C4, obscu­ The caudal-most median elements of the gastral ring the shape of the associated ribs. Between basket are preserved in articulation between the C5 and C46, the shaft of each rib is nearly ver­ vertebral column and left hindlimb. The gastralia tical, and its distal extremity is expanded crani- exhibit the typical, spindle shape, with a shallow, ally and caudally, being thus hammer-shaped. longitudinal groove on their exposed surface. Further caudally, the cervical ribs taper distally. Each rib is longer than the height of the respecti­ ve centrum on C4 to C10; the ribs are as long as Appendicular skeleton the centra are high on C11 to C18, and shorter Pectoral girdle than the height of the centra between C19 and As the specimen is exposed, only isolated por­ C50. The ribs then become progressively longer tions of the pectoral girdle are visible, so that toward the thorax. Except maybe the third pair, all no reliable interpretation can be given. The right ribs are laterally convex. They insert on the vent­ pectoral fenestra is insufficiently exposed to de­ rolateral portion of the centra, slightly caudally, at termine whether a pectoral bar is present. least until C48. The pectoral ribs are only partly preserved; they were not fused to the transverse Pelvic girdle (Fig. 9) processes and, like the thoracic ribs, collapsed The left pubis partly overlies the left femur. The and now lie on their cranial face. The proximal cranial-most portion of its medial margin is in turn portion of the shaft of the pectoral ribs exhibits a overlain by what is probably the same portion of well-defined, caudally directed crest. the right pubis. Sacral ribs obscure the rest of the The dorsal vertebrae are hardly visible, due to medial margin of the left pubis. The medial and matrix and ribs. In the cranial portion of the tho­ cranial margins of the pubis appear to form an rax, 10 are identified by their transverse proces­ obtuse angle. The cranial margin runs straight ses. Toward the pelvic girdle, the column is dis­ craniolaterally, at least in its preserved, medi­ articulated. Vertebrae 70 to 77 are lying on their al-most portion. The lateral margin is smoothly right side, and their neural arches are separated concave. The left pubis is still in natural contact from the respective centra; vertebra 75 partial­ with the laterocranial-most, acetabular portion of ly overlaps vertebra 76, and a further vertebra the ischium. The proximal portion of the ilium is might be present in the area. The first clear sacral probably represented by an enlarged acetabu­ vertebra, bearing a large, kidney-shaped rib facet lar head, smoothly curving into a parallel-sided in the centre of the lateral face of the centrum, shaft, preserved over about 70 mm. is the 76th (or 77th) in the series. The following vertebra is also a sacral, while the 78th (or 79th) Forelimb (Fig. 8) is too poorly preserved to determine whether it The shaft of the left humerus is set off from its is a sacral or the first caudal. Where visible, the distal portion, both parts being separated by a dorsal and sacral centra are cylindrical, slight­ break. It thus appears that the same disease as ly higher than long; they appear to be in close several vertebrae (see Material and Discussion; contact with one another. The neural spines are Table 1) and ulna (see below) affected the hume­ missing on all dorsal vertebrae, broken at their rus at the distal end of its shaft; the humerus may base. On the right side, seven of the cranial-most have been compressed proximodistally when the thoracic ribs are largely preserved, collapsed like carcass touched the sea floor, breakage occur­ the pectoral ribs; like them, they were not fused ring in the affected area. Comparison with the to the transverse processes. The right ribs asso­ preserved portion of the right humerus appears ciated with vertebrae 60 to 63 had been broken to confirm that the left humerus was about one at the level of the ventral curvature of the shaft, fourth longer than it is as preserved. The cranial and calluses grew at the fractured area. The left margin of the humerus is only slightly concave ribs are missing, or are represented by disarticu­ in its proximal portion. In its distal third the cra­ lated fragments. nial margin forms a distinct shoulder. The caudal Twelve caudals are partly preserved, although margin of the humerus is smoothly concave. The much of these is reconstructed, at least super­ distocaudal-most corner of the bone is missing. ficially. The centra are wider than long. All neural The articular facets for the radius and ulna are spines are missing. The poorly preserved trans­ shallowly concave in dorsal aspect and subequal verse processes are horizontal, with parallel cra­ in length. The interarticular area is situated level nial and terminal margins. with the centre of the shaft. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 12 carolinea, 63 (2005)

The radius is subquadrangular with a convex cra­ ?second phalanx; the third phalanx is genuine. nial margin and a concave caudal margin. The whole digit V is missing, except for the dis­ Only the distal portion of the ulna is preserved. A tal-most portion of what appears to be the fourth natural cast (see Material and Discussion) see­ phalanx. mingly replaced the middle portion of its shaft; it appears that more of the bone was preserved Hindlimb (Fig. 9) proximally as its proximal edge consists of freshly At least proximal to mid-shaft, the ventral half of broken bone. The cranial margin of the ulna as the left femur is entirely missing and was pro­ preserved is concave; its distal margin is divided bably preserved as a cast (see Material and into a cranial facet for the intermedium or cen­ Discussion). The caudal margin of the left femur trale (see Caldw ell 1997, Sato 2003), a cen­ is smoothly concave. Due to the overlying pu­ tral, longer facet for the ulnare, and a third facet, bis, only the straight articular facet for the fibula slightly longer than the facet for the intermedium is visible. Taking into account the shortening of but incompletely preserved, for supernumerary the left humerus, both propodials have about the bone(s). same proximodistal dimension. The right femur The radiale is subrectangular with rounded an­ preserves no further anatomical details, its visib­ gles and convex cranial and caudal margins. Its le, ?dorsal aspect being weathered and having proximocranial corner forms a proximally direc­ been partly preserved as a cast. ted hook. The intermedium is hexagonal. The As noted earlier (see Material), the portion of the ulnare is subquadrangular with rounded angles, hindlimb placed on the left side of the mounted a convex cranial margin and a concave caudal specimen distal to the femur is either the right margin. one or visible in ventral view. For the present de­ Identification of the elements of the distal carpal scription, the latter hypothesis is adopted; it ap­ and tarsal rows in plesiosaurs was last discussed pears to agree better with the loss, probably due by Sato (2003); it appears reasonable to consi­ to weathering, of the distal portions of both right der the second element as fused distal carpals limbs. (tarsals) 2 and 3, as they articulate with metacar- The middle portion of the tibia is reconstructed; pals (-tarsals) II and III. Distal carpals 1 and 2+3 it was probably preserved as a cast. The cranial are subquadrangular, the former one-third larger margin of the tibia is convex, its caudal margin than the latter. Distal carpal 4 is subrectangular, concave; its cranial corners are rounded while its craniocaudally elongate. The caudal-most ele­ caudal corners are pointed. Its proximal margin is ment of the carpal row is metacarpal V, which is convex cranially and concave caudally; its distal typically hour-glass-shaped and proximodistally margin is weakly convex. elongate; it is incompletely exposed proximally. Less than the distal half of the fibula is genuine. Metacarpals l-lll are entirely preserved; they are The preserved portions of its cranial and caudal hour-glass-shaped as are the phalanges. Only margins are parallel, craniodistally directed. Its the proximal-most portion of metacarpal IV is distal margin is convex, poorly divided into two genuine. articular facets, for the calcaneum and astraga­ In the stylopod, a break runs distal to phalanges lus (traditionally interpreted respectively as fibu- I6, II5, III4, IV3 and ?V4 (see further). Proximal to lare and intermedium and recently reassessed this break, the phalanges of the first digit regu­ by Caldw ell [1997] and Sato [2003]). larly decrease in size distally. Distal to the break, Thecentrale (traditionally tibiale, C aldw ell 1997, the phalanges forming the cranial-most digit are Sato 2003) is subrectangular with rounded crani- larger than the last phalanges of the proximal oproximal and caudoproximal corners and a con­ portion of digit I, thus casting a serious doubt cave cranial margin. The proximal margin of the about the origin of this genuine, distal fragment. astragalus is obscured by matrix; both its cranial It most likely does not belong where it is located, and caudal margins are straight; its distal mar­ and digit I was probably the shortest digit of the gin is concave cranially and convex caudally. The forelimb, as is the case for all plesiosaurs known proximal margin of the calcaneum is obscured (e.g. C aldw ell 1997). It is unknown, where the by matrix as well. The bone is enlarged distally, distal fragment comes from. Therefore, 6 phalan­ where its otherwise straight cranial and caudal ges are surely preserved in digit I, 5 in digit II, margins diverge. Its distal margin is convex. and 4 in digit III. Of digit IV, bones are reconstruc­ The first distal tarsal is trapezoidal, its longer ted between the metacarpal and the centre of the margin being the proximal one. Its cranioproxi- ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 13 mal corner is overlain by the hooked-shaped, considered geographically or stratigraphically cranially directed cranioproximal corner of dis­ defined groups of taxa. As a result, most elas- tal tarsal 2+3 (see above). Apart from this hook, mosaurian genera currently considered valid are distal tarsal 2+3 Is subquadrangular. Distal tar­ monospecific and often monotypic, many being sal 4 is subquadrangular, its proximal and distal defined on poorly diagnostic features or material margins slightly convex, Its cranial and caudal (Table 5). A profound lack of understanding of margins slightly concave. Metatarsal V is hour­ the anatomy of elasmosaurs in an evolutionary glass-shaped and follows the distal tarsal row perspective results, which is exemplified by the caudally. number and nature of characteristics considered A break filled with resin runs from the caudodistal reliable for diagnosing elasmosaurian genera corner of metatarsal I through the shaft of meta­ (Carpenter 1999): number of cervical verte­ tarsals II—IV and the proximal margin of phalanx brae, structure of the atlas/axis complex, position V1 .The metatarsals otherwise are genuine, hour­ of the naris compared to maxillary teeth, and, for glass-shaped, and of similar size. some taxa, absence of a pectoral bar in adults. The stylopod is broken and reconstructed further Hydralmosaurus W e lle s , 1943 appears unique distally, between the phalanx I2 and the distal in the extension of the caudodistal margin of its margin of V1. The distal third of the hindlimb is humerus. Of these characteristics, the import of a cast of the forelimb (see above). Thus only the the number of cervical vertebrae on taxonomy following bones are surely genuine: phalanx 11, should be thoroughly assessed in a group that the caudal half of phalanxes I3 and I4 and the is defined by the elongation of its neck, not only proximocaudal corner of the phalanx I5; half of by elongation of individual vertebrae but also by phalanxes 111 and M2, phalanx M3 and half of increase in their number. Individual variation in phalanxes II4; half of phalanx 1111, phalanxes the number of vertebrae, including cervicals, oc­ III2 and III3; most of phalanxes IV1, IV2 and IV3; casionally occur in mammals (e.g. Ro u vièr e & phalanxes V1, V2 and half of phalanx V3. D elm as 1985, N ickel et al. 2004); large-scale studies in Recent reptiles are missing that would Referred specimen (Fig. 10) help estimate individual variations and their mag­ SMNS 81783 is a weathered skull. It exhibits all nitude. Additionally, few elasmosaurs are known features observed in SMNK-PAL 3978 listed in with complete cervical column and material al­ the diagnosis; moreover, it confirms that the sus- lowing independent taxonomical identification. pensorium is subvertical. Given the uncertainty concerning the tooth count when skull and mandibles are preserved in oc­ Discussion clusion (e.g. C arpenter 1997), the extent of the Comparative palaeontology nares and their position relative to the maxillary Although elasmosaurids are among the most teeth is usually difficult to determine precisely, famous reptiles of the Mesozoic (e.g. Shuler their margins being broken and pressed inward. 1950, T aylor 1997) and form an anatomically These features are frequently used as a means homogenous, well-defined group among Ple- to quantify a subjective impression concerning siosauria, there are few adequately described the caudal retreat of the naris and skull propor­ skulls (Table 5). Referral of partial and some­ tions, which depends much upon the amount of times juvenile specimens, based on post-cranial distortion the skull underwent. features of poorly proved taxonomical import, SMNK-PAL 3978 undoubtedly represents an has often added new portions of the skeleton or elasmosaurid plesiosaur as defined, for instance, even skull to taxa established upon incomplete by Brown (1993) and Ba r d e t et al. (1999). With holotypes, potentially creating chimeras in great 58 to 60 cervicals and pectorals (‘cervicals’ sen­ need of reassessment. In consequence, elasmo­ su Ca r penter 1999), SMNK-PAL 3978 is similar saur systematics is currently still mostly based to most Late Cretaceous elasmosaurs for which on postcranial features, as they were established this figure is known with certainty (Table 5). It in the latest reviews of American taxa (W elles probably only clearly differs from Elasmosaurus, 1943, 1952, 1962); taxonomy is in great need which has as many as 72 cervicals (C o pe 1868, of a comprehensive revision taking into account C a r penter 1999). ontogeny and individual variation. Long-awaited The humerus of SMNK-PAL 3978 is incompletely recent clarifications (Sto rrs 1999, C a r p e n ­ preserved caudally; though, the central location ter 1997 1999, G rossm ann in prep.) mainly of the transitional area between the facets for Table 5. Late Cretaceous elasmosaurs of import for comparison with SMNK-PAL 3978. The first cited specimen is the holotype of the taxon. The number of 14 cervicals includes pectorals (for comparison with Carpenter 1999) when not otherwise stated and is given for the holotype if not otherwise stated.

Taxon Additional Material Occurrence Number of cervical Skull length

reference(s) vertebrae (mm) ©Staatl. Mus.f.NaturkdeKarlsruhe&Naturwiss.Ver.e.V.;downloadunterwww.zobodat.at ,'O O o o O JO b ~ § o g | _C 2 CD co cu2 CO o LO .0 ^ cp CO c co §.sajg 2 O O > > . Q CD D- £ £ -DQ-S q I > ^ _0 o coco I CO HI Q) .Q co ¿j a p T3 CO ■E 3 T "p ^ O ^ i co J2 CO ^ ¡—] ¥ 8 3 0 P co^ CO > co co co c co .52 c 0 Q.U{ co" 8 ? ^ ^ W 9 S nO) S? ) CO) , n ffl — LU - Q T— 5 > 13 o>=5 > § Cfl E 1 ¥ co o — ffl CM -S C -o r ¡ CD" -Q co E ® t 'p w CT). JD WI ^ > "O 0 S> c — - CD v* CO -= n o

o- 0 - Q.*-CO S "O CD _ _ CM0

g d 0 ^ O O C!—"O Q_7T"co' ■5 “2 .Q LU COCT) CD o i S - 3 0 < 5 3 0 CL « « L- L- 3 0 O c co 13 CD ° O 5 3 ® co o> 03 - T m o 0 co C o 0 F- O r5 -S O ^ 0 rS0 r5-S ^ O Uj l 3 _l 3 0 3 0 0 -b0 0 s 03 Q3 0 CO is 3 0 . Q CO v Oco cu _ O O CO 03 _3 - - C l Q oo - i i l l i O Z 3 Z O -3£ -= o) .o co P- > > 8 CDO o o cc o F 0 io 0 0 03

B 1 03 t 03 5 cl 0 "co 0 -p o -

m 0 CL 0 0 0 o " m"o £ §■ o O 2 CO | 0 ^ w 3 3 0 co-p 03 CO CO o 2 . r ® 03 0 O CO 0

'c 3 CO < 8 o 0 2 d ^ 1 . 0 —0o 0 |— _ 03 03 _ > -5= _ *0 C\J C —- —- C C\J co 0 O 0 0 03 03 0 co rf cl 0 Tf0 0 0 0 0 0 8 0 c 0 CO 0 0 - 5 E ™ tc P .9 c H ~ ~ H ~ < o

8 > 0 ^ E Q. co COO CO o a 3 co 3 o a o 0 ^ _c JO.2 o co o- O co “3 ccO F < < F z p 52 co 9 CO9 S co0 1 ¡ 0 0 IU to LL <-U i— CL O ) = 8 CO < 0 o 9 > o "50 0 ^ c - 03 o W CT . £ " -a .2 sz =1 E o CO 0 -QE 0 I 0

a

o 0

M . Q Q CM 0 " _ T 5 CO a O H H Q- 0 p 8 1 1 CM - h o ,c ,§c § ° 0 c « 0 ^ "CT 0 0 i- CL 5 L C CL 0 8 E __

, c 0 1 cl < o 0 O 0 carolinea, 63 (2005)63 carolinea, CO

j C j i ®co o 2 h~-C 5 ■§ J 3 _J m -o m C W O w 3C D ^ 03 l2cM ooS-S LL 3 _0 O ■> ffl ^ O ■> ffl^ 0 in 0 in o S m S S D)0 0 - c E 0 c 8 ° 3 E L= c = CL OCO < ^ ® z ^ ® z

l m -2 Q- c 0 0

i -2 <■ lo uj 0

t - 0 t .0 CD > Q Q Z o .=L o Z ni 1- ( 3 °6 0 = : O O CO ~ o CO CO - 0 w CM m - h o 3 0 ^ 0 0 ^ 5 2 2 0 s i E Q. C 0 CO 3

0 I °8 h - • 0 d !

— ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 15 radius and ulna clearly differs from the situation hing) and a pointed, triangular coronoid process characteristic of Hydralmosaurus (Welles 1943, (which is invisible due to poor preservation in Carpenter 1999). most other elasmosaurs known from otherwise The neural arch of the axis of SMNK-PAL 3978 is adequate material). SMNK-PAL 3978 differs from caudally directed, as in Libonectes, Styxosaurus the holotype and only known specimen of the ta­ and Elasmosaurus, rather than rising dorsally xon in its limb anatomy and most likely in lacking (Carpenter 1999); the postzygapophysis of the the craniolateral process of the pubis that is so axis extends further caudally than the caudal ar­ conspicuous in Terminonatator ( Sato 2003). ticular face of the axis, overhanging the centrum Tuarangisaurus Wiffen & M oisley , 1986 pos­ of C3, a condition also known in Libonectes but sesses a brevirostrine skull (beak index 34, ver­ not in Styxosaurus, Thalassomedon or Elasmo­ sus about 40 in SMNK-PAL 3978), with a wider saurus (Carpenter 1997, 1999). The atlas/axis maxillary rostrum than SMNK-PAL 3978; the of Hydrotherosaurus appears to be similar in pro­ nares are situated level with maxillary teeth 2-3 portions to that of Elasmosaurus (Welles 1943, (Wiffen & M oisley 1986: figs 1-3, Carpenter Carpenter 1999). In contrast to Libonectes mor- 1999: fig. 14). The mandibular rostrum extends gani, whose axis neural arch is about half the until between the second and third dentary teeth height of its axis centrum, the axis neural arch (Wiffen & M oisley 1986, Long 1998). Only of SMNK-PAL 3978 is about 1.5 time higher than about half the length of the axis postzygapophy­ its axis centrum. The nares of SMNK-PAL 3978 sis projects caudally over centrum 3 (Wiffen & are situated dorsal to the largest maxillary teeth Moisley 1986: fig. 4). (numbers 4-5), similar to Libonectes morgani It appears therefore that SMNK-PAL 3978 shares (where the largest teeth are number 3-4), while the diagnostic features of none of the Late Cre­ in Styxosaurus the nares are situated caudal taceous elasmosaurid genera currently conside­ to the largest maxillary teeth (Welles & B ump red as valid (Welles 1962, Wiffen & M oisley 1949, Carpenter 1999). In Thalassomedon the 1986, Bardet et al. 1999, Carpenter 1999, nares are circular in outline and situated above Storrs 1999, Gasparini et al. 2003, Sato 2003). maxillary teeth 3-5; the rostral-most teeth appear However, I consider it likely that the impossibility not to be procumbent (Carpenter 1999). of referring this specimen to any hitherto descri­ Callawayasaurus colombiensis (Welles , 1962), bed genus is most likely rooted in the inadequa­ from the Aptian of Colombia, has 56 preserved cy of many holotypes or referred specimens and cervicals (Carpenter 1999). Its nares are situ­ their description and illustration, and, as a con­ ated above maxillary teeth 3-5; its skull is more sequence, of elasmosaur taxonomy in general. I brevirostrine than that of SMNK-PAL 3978 (Wel ­ therefore do not elect to create a new genus that les 1962: figs 3, 4). The atlas/axis complex of would mainly be diagnosed by features unknown Callawayasaurus is unknown. in other taxa. Based on similar stratigraphical oc­ Aphrosaurus furlongi Welles , 1943 possesses currence, SMNK-PAL 3978 is referred to the Tura­ 57 cervicals, plus 3 pectorals based on a refer­ nian genus Libonectes Carpenter , 1997 and the red juvenile specimen (Welles 1943; the same features unique to SMNK-PAL 3978 are arbitra­ author gives 59 cervicals plus 3 pectorals in his rily given specific import to acknowledge the dif­ 1952 diagnosis). The antebrachial space is redu­ ferences observed between the type of Libonec­ ced in Aphrosaurus furlongi W elles , 1943, as tes, L. morgani (Welles , 1949), and SMNK-PAL the radius has an almost straight caudal margin 3978. These are listed in the diagnosis. and the ulna an almost straight cranial one. The carpals of this taxon are said to be very thick, and Bone lesions the humerus appears narrow distally (Welles The affected bones are not adjacent (Table 1) 1943). This taxon is poorly defined; in any case, and in the clearest instances they are neither SMNK-PAL 3978 shares none of these features distorted (the original outline being preserved; with it. see Material, Fig. 7) nor displaced. It is highly The (adult) holotype of Terminonatator Sato , unlikely that diagenesis caused such preserva­ 2003 comprises a poorly preserved skull and tion in these bones. In contrast, several other partial postcranium. Its diagnosis is somewhat bones clearly show signs of diagenetic distortion puzzling in emphasising nine premaxillary teeth in the presumably affected areas (e.g., the right (including a central one that could result from humerus and caudal-most cervical vertebrae: abnormal growth, if not merely from lateral crus­ Fig. 8 and 7, respectively). The lesions are here ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 16 carol inea, 63 (2005)

interpreted as resulting from a pathology indu­ 2003), represent one of the few groups of tetra- cing a decrease in bone density. Provided this pods whose caudal-most tooth crowns are not pathology affected the mineral frame of the bone lower than the rostral ones. Lowering of the cau­ (osteoid), the natural cast would then represent dal tooth crowns is a well-justified phenomenon remains of the non-mineral fraction of the bone. on a biomechanical ground (e.g. G reaves 1983, Due to reduced mineral frame, the affected areas 1995). The caudal part of the tooth row encoun­ of the bones would have been ‘soft’ in vivo ; col­ ters the highest forces during occlusion; more lapse of the humerus, for instance, at the level massive teeth in this part of the row better resist of the affected area is explained by the resulting the stress. Being located closer to the pivot, the softening of the bone (R o thsc h ild & M artin path traversed by their apex during the occluding 1993). Pathologies inducing such a multilocular phase is shorter, hence the caudal-most teeth decrease in bone density without abnormal bone are lower. M a ssar e (1987) attributed the form growth include osteoporosis and osteomalacia, of elasmosaur teeth a piercing (of fish) func­ and metabolic or hormonal diseases; disuse or tion. If that is arguable for the rostral-most teeth, immobilisation also result in reduction of bone it is certainly not the case for the caudal ones. density but appear unlikely in the case and locati­ Interaction of the caudal-most dentary teeth of on of the lesions observed in SMNK-PAL 3978 (cf. SMNK-PAL 3978 (and most elasmosaurs) with Rothschild & M artin 1993). Avascular necrosis, even low caudal maxillary teeth in order to pierce maybe resulting from decompression syndrome, a prey item would require such a gape (see e.g. could also be invoked, although the lesions in Fig. 1, Table 3) that it appears impossible: if dur­ SMNK-PAL 3978 do not appear localised in the ing the movement (provided it is allowed by the areas usually affected (subarticular portions of occipital joint) the skull is considered as the mo­ femur and humerus, Ro thsc h ild & M artin 1987, bile element, the squamosal would be brought 1993, Ro thschild & S to rrs 2003). Rothschild into contact with the axis neural arch. This would & S to rrs (2003) did diagnose decompression mean squeezing the supraspinsal ligament and syndrome in plesiosaurs, but their systematic local epaxial and depressor musculature (e.g. examination yielded no affected vertebrae while Frey 1988). If the skull is considered immobile SMNK-PAL 3978 exhibits numerous affected during occlusion, and only the mandible moves, cervical vertebrae (Table 1). According to these the movement (if allowed by the mandibular ar­ authors, the vertebrae were possibly highly vas- ticulation) would bring the retoarticular process cularised as an evolutionary response to the risk to collide with the suspensorium. The caudal of decompression syndrome. Proper diagnosis of portion of the tooth row cannot be considered the pathology in SMNK-PAL 3978 requires syste­ a prey-piercing device, and still large caudal matic examination of the internal structure of the teeth are retained throughout the whole elas­ affected areas, at least with X-rays (R othschild mosaur history. If the function of these teeth was & M artin 1993), which is not currently possible, not related to a vertical movement of the jaws, as the specimen is on exhibition. then one has to consider a function aligned with A different type of lesion is visible on some bones a horizontal movement - of the water flow: it is of the forelimb (Fig. 8). On the dorsal surface of here preliminarily suggested that the elasmosaur the left humerus, a scratch mark terminated by tooth apparatus would constitute better a sieving a subcircular depression surrounded by a wider device. C o llin & J anis (1997) argued that sus­ area where the bone appears pressed-in is pre­ pension feeding is most likely absent in fossil ma­ sent in the middle portion of the shaft. A larger, rine reptiles. Although the head of elasmosaurs subcircular, depressed area filled with resin at the is not large compared to their body (as would be broken edge of the shaft possibly also represents required for suspension feeding: C o llin & J anis a bite mark; the ulna shows a similar mark in its 1997), the mandible is consistently ogival in ven­ middle. The dorsal surface of the radius exhibits tral aspect and tall in lateral aspect, which leaves at least five possible circular, smaller bite marks. space and attachment areas for a well-developed hyomandibular musculature and gular pouch. Elasmosaur dentition, preliminary remarks The absence of a bony secondary palate in ple­ Elasmosaurids, together with the other long­ siosaurs stands in contradiction to the hypothesis necked, short-headed plesiosaurs of debated af­ of suspension feeding; however C o llin & J anis finities, cryptoclidids and plesiosaurids (O’Keefe (1997) recognise that a soft tissue equivalent 2001, G asparini et al. 2003, O’Keefe & W ahl can develop. The existence of such a device was ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 17

indeed suggested for pliosaurs (McHenry pers. logy, Karlsruhe (Germany), 27 June - 01 July 2000, comm.). Some fossil evidence hint at an inflexible Abstract Book. neck in plesiosaurs, at least in its cranial-most Bardet , N., Pereda Suberbiola , X. & J alil , N.-E. (2003a): A new polycotylid plesiosaur from the Late part (Buchy et al. 2005), but still the length of the Cretaceous (Turanian) of Morocco. - C. R. Palevol, neck stands in contradiction to biomechanical re­ 2:307-315. quirements of a suspension feeder as analysed Bardet , N., Pereda Suberbiola , X. & J alil , N.-E. by Collin & J anis (1997). (2003b): A new mosasauroid (Squamata) from the One hypothesis never explored to account for Late Cretaceous (Turanian) of Morocco. - C. R. elasmosaurs’ peculiar anatomy is sediment siev­ Palevol, 2: 607-616. ing for epi- or endobionts. Elasmosaurs evolved Blainville , H. D. de (1835): Description de quelques during highest transgression phases, when large, espèces de reptiles de la Californie, précédée de shallow, soft substrate areas became available 1‘analyse d‘un système général d'Erpetologie et d'Amphibiologie. - Nouv. Ann. Mus. Nat. Hist, nat., for marine reptiles (e.g. Smith et al. 1994; Bardet Paris, 3 (4): 233-296. 1995; Noe et al. in prep.). Under this hypothesis Brown, D. S. (1981): The English Upper Jurassic Plesi- the chin of SMNK-PAL 3978 forms a convenient osauroidea (Reptilia) and a review of the phylogeny plough. The intensive pitting of the premaxil­ and classification of the Plesiosauria. - Bull. Br. Mus. lary surface of elasmosaurs (e.g. Welles 1962, (Nat. Hist.), Geol., 35 (4): 253-347. Wiffen & M oisley 1986, Bardet et al. 1999) Brown, D. S. (1993): A taxonomic reappraisal of the could have housed exits for sensory terminations families Elasmosauridae and Cryptoclididae (Repti­ of the fifth cranial nerve, functionally analogous lia: Plesiosauroidea). - Revue de Paléobiologie, Vol. to Recent walruses’ sensory bristles (Ridgway & spéc., 7: 9-16. uchy M.-C., étayer F. & rey E. (2005): Osteolo­ Harrison 1987). Although absent in SMNK-PAL B , M , F , 3978—at least in its present condition—gastro- gy of Manemergus anguirostris n. gen. et sp., a new plesiosaur (Reptilia, Sauropterygia) from the Upper liths are regularly encountered within elasmo­ Cretaceous of Morocco. - Palaeontographica, A saurs’ abdominal area (for a review see Cicimur- 272:97-120. ri & E verhart 2001) and would be essential to Caldwell , M. W. (1997): Limb osteology and ossifica­ access digestible parts of invertebrates. Finally, tion patterns in Cryptoclidus (Reptilia: Plesiosauroi­ this does not contradict occasional capture of dea) with a review of Sauropterygian limbs. - J. Vert. fish when available, but it would better explain Pal., 17 (2): 295-307 the rarity of such identifiable gastric contents Carpenter , K. (1997): Comparative cranial anatomy (Cicimurri & E verhart 2001). of two North American Cretaceous plesiosaurs. - In: Callaway J. M. & Nicholls E. L. (Eds): Ancient Mari­ ne Reptiles: 191-216; San Diego (Academic Press). Carpenter , K. (1999): Revision of North American Acknowledgements elasmosaurs from the Cretaceous of the western in­ Thanks are due to E. ‘D.’ Frey (SMNK) and R. Schoch terior. - Paludicola, 2 (2): 148-173. (SMNS) for access to specimens in their care; and to Cavin , L. (1995): Goulmimichthys arambourgi n.g., E. ‘D.’ Frey also for comments upon the manuscript; to n.sp., un Pachyrhizodontidae (Actinoptrygii, Teleos- C. Ifrim (University of Karlsruhe) for ammonite deter­ tei) d’une nouvelle localité à nodules fossilifères du mination; to N. Bardet (Paris), M. Everhart (Hays), L. Turonien inférieur marocain. - C. R. Acad. Sci. Paris, Noe (Cambridge) for discussion and literature; to K.T. 321, Il a: 1049-1054. Smith (Yale University) who improved the present text Cavin , L. (1997): Nouveaux Teleostei du gisement du by correcting the language and making valuable com­ Turonien inférieur de Goulmima (Maroc). - C. R. ments. Acad. Sci. Paris, Sci. Terre et planètes, 325: 719- 724. Cavin , L. (1999): Occurrence of a juvenile teleost, References Enchodus sp., in a fish gut content from the Upper Bardet , N. (1995): Évolution et extinction des reptiles Cretaceous of Goulmima, Morocco. - Palaeontology marins au cours du Mésozoïque. - Palaeovertebra- Sp. Pap., 60: 57-72. ta, 24 (3-4): 177-283. Cavin , L. (2001): Osteology and phylogenetic relation­ Bardet , N., Godefroit, P. & Sciau , J. (1999): A new ships of the teleost Goulmimichthys arambourgi Ca­ elasmosaurid plesiosaur from the Lower Jurassic of vin, 1995, from the Upper Cretaceous of Goulmima, Southern France. - Palaeontology, 42: 927-952. Morocco. - Eclog. geol. Helvetica, 94: 509-535. Bardet , n ., Pereda Suberbiola , X., Jalil , N.-E. & Ia - Cavin , L., Boudad , L., Duffaud , S., Kabiri , L., Le ROChène S. M. (2000): Marine reptiles from the early Lœuff, J., Rouget , I. &T ong, H. (2001): L’évolution ate Cietaceous of the Goulmima region, Morocco. paléoenvironnementale des faunes de poissons du 5 European Workshop on Vertebrate Palaeonto­ Crétacé supérieur du bassin du Tafilalt et des régi- ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 18 carolinea, 63 (2005)

ons avoisinantes (Sud-Est du Maroc): implications wegungsapparat. - 8th Edition, 626 pp.; Stuttgart paléobiogéographiques. — C. R. Acad. Sei. Paris, (Parey Verlag). Sei. Terre et planètes, 333: 677-683. O'Keefe , F. R. (2001): A cladistic analysis and taxono­ Chancellor , G. R., Kennedy , W. J. & Hancock, J. M. mic revision of the Plesiosauria (Reptilia: Sauropte­ (1994): Turonian ammonite faunas from central Tuni­ rygia).-A cta Zool. Fennica, 213: 1-63. sia. - Palaeontology Sp. Pap., 50:1-118. O'Keefe , F. R. & Wahl , W. (2003): Current taxonomic Cicimurri, D. J. & Everhart , M. J. (2001): An elasmo- status of the plesiosaur Pantosaurus striatus from saur with stomach contents and gastroliths from the the Upper Jurassic Sundance Formation, Wyoming. Pierre Shale (Late Cretaceous) of Kansas. - Trans. - Paludicola, 4(2): 37-47 Kansas Acad. Sei., 104 (3/4): 129-143. Owen , R. (1860): On the orders of fossil and recent Collin, R. & Janis , C. M. (1997): Morphological cons­ Reptilia, and their distribution in time. - Rep. Brit. traints on tetrapod feeding mechanisms: why where Ass. Adv. Sei., 1859, 29: 153-166. there no suspension-feeding marine reptiles? - In: Persson , P. O. (1963): A revision of the classification Callaway J. M. & Nicholls E. L. (Eds): Ancient Marine of the Plesiosauria, with a synopsis of the strati- Reptiles: 451-466; San Diego (Academic Press). graphical and geographical distribution of the group. Cope , E. D. (1868): On a new large enaliosaur. - Proc. - Lunds Universitets Arsskrift, N.F. Avd. 2, 59: 1-59. Acad. Nat. Sei. Philadelphia, 20: 92-93. Ridgway , S. H. & Harrison , R. J. (1987): Handbook of Cope , E. D. (1869): Synopsis of the extinct Batrachia, marine mammals. 5 volumes; San Diego (Academic Reptilia and Aves of North America. - Trans. Amer. Press). Phil. Soc., n. s., 14:1-252. Römer , A. S. (1956): Osteology of the Reptiles. 772 pp.; Everhart , M. J. (in press): Bite marks on an elasmo- Chicago and London (University of Chicago Press). saur (Sauropterygia; Plesiosauria) paddle from the Rothschild , B. M. & Martin , L. D. (1987): Avascular Niobrara Chalk (Upper Cretaceous) as probable necrosis: occurrence in diving Cretaceous mosa- evidence of feeding by the lamniform shark, Cretox- saurs. - Science, 236: 75-77 yrhina mantelli. - Pal Arch. Rothschild, B. M. & Martin , L. D. (1993): Paleopatho­ Ferrandini , M., Philip , J., Babinot , J.-E, Ferrandini , logy: disease in the fossil record. - 386 pp.; Boca J. &T ronchetti, G. (1985): La plate-forme carbona- Raton (CRC Press). tée du Cénomanien-Turonien de la région d'Erfoud- Rothschild , B. M. & Storrs , G. W. (2003): Decom­ Errachidia (Sud-Est marocain): stratigraphie et pa­ pression syndrome in plesiosaurs (Sauropterygia: léoenvironnements. - Bull. Soc. géol. France, sér; 8, Reptilia). - J. Vert. Pal., 23(2): 324-328. I: 559-564. Rouvière , H. & Delmas , A. (1985): Anatomie humaine. Frey , E. (1988): Anatomie des Körperstammes von - 12th Edition, t. 2, Masson, Paris. Alligator mississipiensis Daudin. - Stuttgarter Beitr. Sato , T. (2003): Terminonatator ponteixensis, a new Naturk., Ser. A, 42:1-106. elasmosaur (Reptilia; Sauropterygia) from the Up­ Gasparini , Z., Bardet , N., Martin , J. E. & Fernandez , per Cretaceous of Saskatchewan. - J. Vert. Pal., 23 M. (2003):The elasmosaurid plesiosaur Aristonectes (1): 89-103. Cabrera from the latest Cretaceous of South Ameri­ Shuler, E. W. (1950): A new elasmosaur from the Ea­ ca and Antarctica. - J. Vert. Pal., 23(1 ): 89-103. gle Ford Shale of Texas. - Southern Methodist Univ., Greaves , W. S. (1983): A functional analysis of carnas- Fondren Science Series, 1 (2): 1-30. sial biting. - Biol. J. Linn. Soc., 20: 353-363. Smith, A. G., Smith, D. G. & Funnel, B. M. (1994): At­ Greaves , W. S. (1995): Functional predictions from las of Mesozoic and Cenozoic coastlines. - 99 pp.; theoretical models of the skull and jaws in reptiles Cambridge (Cambridge University Press). and mammals. - In: Thomason, J. J. (Ed.): Functio­ Storrs , G. W. (1997): Morphological and taxonomic nal morphology in vertebrate paleontology: 99-115; clarification of the genus Plesiosaurus. - In: Calla ­ Cambridge Cambridge (University Press). way , J. M. & Nicholls , E. L. (Eds.): Ancient Marine Kossmat , F. (1897): Untersuchungen über die Südindi­ Reptiles: 145-190; San Diego (Academic Press). sche Kreideformation. Zweiter Teil. - Beitr. Paläont. Storrs, G. W. (1999): An examination of Plesiosauria Österreich-Ungarns u. Orient, 11: 1-46 (108-153), (Diapsida: Sauropterygia) from the Niobrara Chalk plates 1-8 (12-19). (Upper Cretaceous) of central North America. - Univ. Lazo , D. G. & Chichowolski, M. (2003): First plesiosaur Kansas Paleont. Contr., 11:1-15. remains from the lower Cretaceous of the Neuquen Taylor , M. (1997): Before the dinosaur: the historical Basin, Argentina.-J. Paleont., 77 (4): 784-789. significance of the fossil marine reptiles. - In: Cal ­ Long, J. A. (1998): Dinosaurs of Australia and New Zea­ laway J. M. & Nicholls E. L. (Eds): Ancient Marine land and other animals of the Mesozoic Era. - 188 Reptiles: XIX-XLVI; San Diego (Academic Press). pp.; Sydney (University of New South Wales Press). W elles , S. P. (1943): Elasmosaurid plesiosaurs with a Massare , J. A. (1987): Tooth morphology and prey pre­ description of new material from California and Co­ ference of Mesozoic marine reptiles. - J. Vert. Pal., 7 lorado. - Mem. Univ. Calif., 13(3): 125-215. (2): 121-137. Welles , S. P. (1949): A new elasmosaur from the Nickel, R., Schummer , A. & Seiferle , E. (2004): Eagle Ford Shale of Texas. - Fondren Sei. Ser., Lehrbuch der Anatomie der Haustiere, Bd. 1 Be­ Southern Methodist Univ., 1: 1-28. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 19

Welles , S. P. (1952): A review of the North American Wiffen , J. & Moisley , W. L. (1986): Late Cretaceous Cretaceous elasmosaurs. - Univ. Calif. Pubis geol. reptiles (Families Elasmosauridae and Pliosau- Sci., 29(3): 47-144. ridae) from the Mangahouanga Stream, North Is­ Welles , S. P. (1962): A new species of Elasmosaur land, New Zealand. - N. Z. J. Geol. Geophys., 29: from the Aptian of Colombia and a review of the 205-252. Cretaceous plesiosaurs. - Univ. Calif. Pubis Geol. Williston , S. W. (1890): A new plesiosaur from the Sci., 44(1): 1-96. Niobrara Cretaceous of Kansas. - Trans. Kansas Welles , S. P. & Bump , J. (1949): Alzadasaurus pem- Acad. Sci., 1 2 : 174-178. bertoni, a new elasmosaur from the Upper Cretace­ Williston, S. W. (1906): North American plesiosaurs: ous of South Dakota. - J. Paleont., 23: 521-535. Elasmosaurus, Cimoliasaurus, and Polycotylus. Werner , C. & B ardet , N. (1996): New record of elas­ -A m . J. Sci., 21:221-236. mosaurs (Reptilia, Plesiosauria) in the Maastichtian Williston , S. W. (1925). Osteology of the reptiles. of the Western Desert of Egypt. - Berliner geowiss. - 300 pp.; Cambridge, USA (Harvard University Abh., E18: 335-341. Press).

Plate 1: SMNK-PAL 3978, Libonectes atlasense n. sp., holotyope; general view of the mounted specimen. Total length about 7.2 m. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 20 carolinea, 63 (2005)

Figure 1: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; (a) skull in left lateral view; ( b) interpretative line drawing. Abbreviations: a: angular; art: articular; d: dentary; en: naris; j: jugal; mx: maxilla; o: orbit; par: parietal; pmx: premaxilla; po: postorbital; pf: prefrontal; Q: quadrate; rp: retroarticular process; sa: surangular; sq: squa­ mosal. Matrix is hatched. Shaded areas are reconstructed. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turonian of Marocco 21

Figure 2: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; (a) skull in dorsal view; ( b) interpretative line drawing. Abbreviations as in Figure 1, plus: tf: temporal fenestra. Matrix is hatched. Shaded areas are reconstruc­ ted. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 22 carolinea, 63 (2005)

Figure 3: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; (a) skull in ventral view; (b) interpretative line drawing. Abbreviations as in Figure 1. Matrix is hatched. Shaded areas are reconstructed. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turonian of Marocco 23

Figure 4: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; (a) skull in rostral view; ( b) Interpretative line drawing. Abbreviations as in Figure 1. Matrix is hatched. Shaded areas are reconstructed. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 24 carolinea, 63 (2005)

Figure 6: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; interpretative drawings of the atlas/axis complex (a) in left lateral view; ( b) in right lateral view. Abbreviations: ax: axis; C3: third cervical vertebra; nax: axis neural arch; nat: atlas neural arch; od: odontoid; pz: atlas postzygapophysis; r: cervical rib. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at Buchy : Elasmosaur from the Turanian of Marocco 25

Figure 7■ SMNK-PAL 3978, Libonectes atiasense n. sp., ho'otyp6', cervicals37^4^ 9 between C37 and C39; arrow points at the cranial portion of C40 preserved as a natural cast, prooaoiy pathology (see text). Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 26 carol inea, 63 (2005)

Figure 8: SMNK-PAL 3978, Libonectes atlasense n. sp., holotype; (a) left forelimb in dorsal view; ( b) interpretative drawing; (c) right humerus in dorsal view; (d) interpretative drawing. Abbreviations: c1-4: distal carpals 1-4 (see text); H: humerus; i: intermedium or centrale (see text); R: radius; Re: radiale; U: ulna; Ue: ulnare; l-V: metacarpals l-V. Arrows point cranially. Dashed lines mark breaks or reconstructed margins. Dotted lines mark possible bite and scratch marks. Scale bar 50 mm. ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at

Buchy : Elasmosaur from the Turanian of Marocco 27

Figure 9: SMNK-PAL 3978, Libonectes atlasense n. sp„ holotype; (a) left¡1) d S i ’vtew or'fhe fefU>ne in ventral view are visible in dorsal view while the rest of the limb is either the righ on ce; centra|e; il: ilium; is: ischium; (see text); (b) interpretative drawing. Abbreviations: as: astragalus, c . • tatarsa|s \.\j Arrows for F lemur; t'.ibula; Pu: pubis; sr; sacral rib; 11-4; distal tarsals M bar 50 mm. orientation ot the limb elements point cranially. Dashed lines mark brea ©Staatl. Mus. f. Naturkde Karlsruhe & Naturwiss. Ver. Karlsruhe e.V.; download unter www.zobodat.at 28 carol inea, 63 (2005)

Figure 10: SMNS 81783, Libonectes atlasense n. sp., referred specimen in left dorsolateral view. Scale bar 50 mm.