Skull

Frontal (UP-VIL-10-9; Fig. 2A–B)

The frontal represents the only mosasauroid skull bone found up to now in Villeveyrac-L’Olivet. With the exception of its anterior part, it is almost complete and of relatively large size (preserved length = 80 mm). Both posterolateral corners are broken off but the left one (Fig. 2A–B) permits us to reconstruct their shape and lateral expansion with a relative degree of certainty. The posterior right corner exhibits an oval puncture (about 5 × 10 mm) with sharp and elevated edges (especially the medial one), suggesting a bite mark. As a whole the bone is long and narrow (width above orbits = 29 mm). Posterior to the orbits, although incomplete, the posterolateral corners of the frontal appear to have been sharp and laterally expanded (estimated width at the parietal contact = 50 mm). The margins are almost straight above the orbits, and then curve medially anteriorly (Fig. 2A). Anteriorly, the frontal is incomplete, and its contribution to the narial emargination cannot be assessed. Posteriorly, the suture for the parietal is straight, vertical and with low interlocking ridges (cf. Bell, 1997). The dorsal surface of the frontal bears discrete ornamentations. The area between the posterior corners is slightly concave and lined by transverse longitudinal ridges posteriorly, as well as with discontinuous small ridges converging towards the middle of the bone anteriorly. Anterior to this area that ends around the middle of the orbits, the frontal bears some slight longitudinal ridges and a slight median elevation that in most mosasaurs extends anteriorly onto the internarial bar. In ventral view, the median groove for the olfactory canal is only partly embraced ventrally by a descending process. It is largely open anteriorly, and then narrows at the orbital level. Posteriorly, there is a large surface that cannot be strictly considered a tabular boss (sensu Bell, 1997), inclined at about 45° from the horizontal for the cerebral hemispheres. This surface bears rough striations and two slender notches located posterolaterally to the median groove (Fig. 2B). The sutural surface for the prefrontal is much larger than that for the postorbitofrontal and they are not in contact, indicating that the frontal takes part in the dorsal margin of the orbit. Both represent large horizontal overlapping zones.

Following Bell (1997), the following combination of characters indicates that this frontal belongs to a plesiomorphic mosasauroid: lateral sides sinusoidal (char 9-0), no median dorsal keel (char 12-0), sharp posterior ala (char 13-0), no contact of prefrontal and postorbital sutures ventrally (char 30-0), no tabular boss anterior to the frontal-parietal suture (char 16-0) is straight and simple, frontal suture with parietal opposing and with low interlocking ridges (char 17-0), no frontal invasion of the parietal and parietal foramen considered as located far from the frontal suture (char 24-0). On the contrary, the following character is apomorphic for the studied frontal: frontal-prefrontal contact consisting in a broad horizontal ventral surface (char 15-1). The narial emargination (char 11) cannot be checked as this region is broken. However, contrary to Bell (1997), the olfactory canal partly embraced ventrally by descending processes of the frontal (Bell, 1997, char 14-1) and the long and narrow morphology of the frontal (Bell, 1997, char 10-1) are here considered as plesiomorphic for mosasauroids (when compared with Varanus). Especially, the long and narrow morphology of the frontal in general looks like that of plesiomorphic mosasauroid forms, such as Aigialosaurus, Russellosaurus and Tethysaurus (N.B., pers. obs.). This morphology strongly differs from that of derived mosasauroids, which usually possess large triangular frontal tables (Bell, 1997). It differs from that of Aigialosaurus from the Cenomanian-Turonian of Croatia (see Dutchak and Caldwell, 2006) in having more sinusoidal lateral margins. It differs from that of Russellosaurus from the Turonian of Texas (Polcyn and Bell, 2005), which exhibits more convex lateral borders and a well-defined triangular surface for the cerebral hemispheres. It roughly looks like that of Tethysaurus from the Turonian of Morocco (Bardet et al., 2003) in its general morphology, but in Tethysaurus (like in Russellosaurus) the triangular area for the cerebral hemispheres is much sharper. No frontal has been described for Pannoniasaurus, from the Santonian of Hungary (Makádi et al., 2012), the only mosasauroid genus currently known from continental deposits. However, an unpublished specimen from the same outcrop (courtesy of L. Makádi, Budapest) looks similar to that of UP-VIL-10-9 (N.B., pers. obs.). Pending both the description of the Hungarian material and the discovery of additional material in Villeveyrac-L’Olivet, the frontal here described is referred to an indeterminate plesiomorphic mosasauroid, most probably belonging to the Tethysaurine clade sensu Bardet et al. (2005). The size of the frontal, as compared with the Tethysaurus holotype, indicates an animal probably about 3 meters long (N.B., pers. obs.).

Vertebrae

General. Ten vertebrae are available for study. Their original positions along the vertebral column were estimated after comparisons with extant lizards and fossil pythonomorphs. The vertebral material consists of two cervical vertebrae (UP-VIL-10-11 being more posterior than UP-VIL-cast 1588; Fig. 2C–H), six dorsal vertebrae (USTL-OLV 3CV, USTL-OLV R 05, USTL-OLV R05 CV, UP-VIL-10-10, UP-VIL-10-12 and UP-VIL-10-13; Fig. 2I–L), one caudal (UP-VIL-10-03; Fig. 2M–N), and one indeterminate vertebra (UP-VIL-10-04). There are important size differences between the specimens. The two cervical vertebrae are of similar size (CL = 44.7 mm in UP-VIL-10-11, CL = 45.7 in 1588), the two dorsal vertebrae USTL-OLV R 05 and UP-VIL-10-10 as well (CL = 61.9 and 60.4, respectively), whereas the others are of various sizes (CL = 50.2 in USTL-OLV 3CV, CL = 41.7 in USTL-OLV R05 CV, CL = 33.0 in UP-VIL-10 12, CL = 20.3 in UP-VIL-10-03, CL = 10.2 in UP-VIL 10 13). Despite size differences, suggesting different individuals (the largest vertebrae are comparable in size to those of Tethysaurus), these vertebrae show, beyond natural intracolumnar variability, similar anatomical features suggesting that they belong to the same taxon.

All vertebral centra are procoelous, triangular in ventral view (Fig. 2E, H, I), and clearly lack pachyostosis. They are widened anteriorly because of the lateral extension of the paradiapophyses but are rather straight thereafter. There is no precondylar constriction. In dorsal view, vertebrae are wider anteriorly than posteriorly, especially because of the lateral extension of the paradiapophyses (Fig. 2F, J). Indeed, prezygapophyses are not strongly laterally extended (at less than 45°), although the width between the prezygapophyses is larger than across the postzygapophyses. Prezygapophyses rather strongly project anteriorly, extending well beyond the anterior border of the cotyle (Fig. 2G, L). Prezygapophyses are much inclined on the horizontal and prezygapophyseal facets are elongated (Fig. 2J). The narrowest part of the interzygapophyseal constriction is positioned posteriorly, at about two thirds of the neural arch length (Fig. 2F, J). Zygapophyses lie clearly above the floor of the neural canal and above paradiapophyses (Fig. 2C, K). The epidiapophyseal ridge is sharp posterior to the dorsal border of the paradiapophysis and even seems to protrude posterolaterally, forming a convexity at this place. Only UP-VIL-10-11 displays a well-preserved zygantrum (Fig. 2D), although apparently no zygantral facet is present. Zygosphenes are not preserved although a V-shaped poorly preserved structure in the two cervical vertebrae (Fig. 2C, F) may represent rests of rudimentary notched zygosphenes. The neural canal is subtriangular (with a trilobate aspect) and much smaller than the cotyle, both in height and width (Fig. 2C, K). The cotyle is ovoid (Fig. 2K). The posterior borders of the neural arch are concave. In posterior view, the neural spine is thick and rather short. Only sufficiently well-preserved on vertebrae UP-VIL-10-11, it seems to rise behind the zygosphene, at the level of the posterior limit of the prezygapophyses, in a very gentle slope that increases progressively posteriorly (Fig. 2G). It also markedly widens posteriorly and ends in a large knob. Posteriorly, it does not seem to extend beyond the condyle. Paracotylar, subcentral, lateral, and/or zygantral foramina are present, although their occurrence and number vary. The gently convex ventral surface of the cervical vertebra UP-VIL-cast 1588 suggests that a hypapophysis was present, although the posterior portion of the centrum is incomplete and probably bore only a small peduncle (Fig. 2G). The ventral surface of UP-VIL-10-11 is nearly flat, and probably bore only a small peduncle (like in the varanoid squamate Pachyvaranus; Houssaye et al., 2011). All dorsal vertebrae display a shallow sagittal furrow that originates directly posterior to the cotyle, where it is the deepest, and becomes shallower posteriorly to a flat end.

Cervical vertebrae (UP-VIL-10-11, UP-VIL-cast 1588; Fig. 2C–H)

In the cervical vertebrae, the axis of the cotyle-condyle system appears straight. Paradiapophyses extend well below the ventral border of the cotyle (Fig. 2C, G). Observation of the most anterior cervical vertebra (UP-VIL-cast 1588) suggests that the hypapophyseal peduncle is co-ossified with the centrum (Fig. 2H). However, it cannot be described, as the posterior part of the centrum is eroded ventrally (Fig. 2H).

Paradiapophyseal articular facets appear rather posteriorly located in the anterior most (preserved) cervical vertebra (although still in the centrum first half), as compared to the other vertebrae; they gradually move anteriorly, from posterior cervical to posterior dorsal vertebrae. They are located much posteriorly to the posterior limit of the prezygapophyses in cervicals (Fig. 2G), which is not the case in dorsal vertebrae (Fig. 2L). In the posterior cervical, paradiapophyses are dorsoventrally well expanded and their dorsal part bends posteriorly (Fig. 2G).

Dorsal vertebrae (USTL-OLV 3CV, USTL-OLV R 05, USTL-OLV R05 CV, UP-VIL-10-10, UP-VIL-10-12 and UP-VIL-10-13; Fig. 2I–L)

Dorsal vertebrae lack any trace of swelling on the ventral face of the centrum (Fig. 2I, L). The dorsal vertebrae preserved are considered rather posterior ones based on the fact that the ventral border of the paradiapophyses lies well above the ventral base of the cotyle (Fig. 2K). Paradiapophyses are located anteriorly; their dorsal part curves posteriorly (Fig. 2L), as in the cervical vertebrae. However, they are positioned much more dorsally than in cervicals and the paradiapophyseal facet is shorter dorsoventrally. The axes of the cotyles and condyles are inclined, the cotyle facing anteroventrally and the condyle being oriented posterodorsally (Fig. 2L).

Anterior caudal vertebra (UP-VIL-10-03; Fig. 2M–N)

The poorly preserved fragmentary vertebra UP-VIL-10-03 bears both transverse processes and peduncles for articulation with haemal arches; it is thus an anterior caudal. The transverse processes are broken off; their bases show that they are located anteriorly, originate from the dorsal part of the centrum clearly above the level of the cotylar ventral rim. The centrum is much longer than wide and high. The cotyle is ovoid (Fig. 2M). The centrum also displays a sagittal groove in ventral view (Fig. 2N). Two ventral peduncles for articulation with chevron bones are located posteriorly (Fig. 2N).

Discussion. All vertebrae are procoelous, with dorsoventrally depressed cotyles and condyles. The axis of the cotyle-condyle system is oblique in (at least) dorsal vertebrae; the hypapophysis originates from the centrum; there seem to be articulated caudal chevrons; the narrowest part of the interzygapophyseal constriction, as seen in dorsal view, occupies a posterior position. These vertebrae are thus characteristic of the Varanoidea sensu Lee (1997), i.e. the Varanidae, Helodermatidae and Pythonomorpha sensu Lee (1997) following Lee (2009). Intracolumnar variation in the axis of the cotyle-condyle system has already been described in varanoid lizards (e.g. Arcanosaurus; Houssaye et al., 2013). The presence of a zygosphene-zygantrum system within these taxa defines the Pythonomorpha (Lee, 1997). The occurrence of a well-developed zygantrum (although articular facets could not be observed) and the possible zygosphene indicate that such a system may have been present. Moreover, the vertebrae differ from those of Varanus in e.g., the absence of precondylar constriction, the size and trilobate shape of the neural canal, from Helodermatidae and Lanthanotidae in the triangular shape of the centrum in ventral view (A.H. pers. obs.; Rieppel, 1980).

The general vertebral morphology is clearly distinct from that observed in snakes (e.g. wider across the prezygapophyses than across the postzygapophyses, absence of prezygapophyseal processes and absence of sagittal keel on precloacal vertebrae). The clear absence of pachyostosis distinguishes the present taxon from most stem-ophidiomorphs (sensu Palci and Caldwell, 2007), the pythonomorph Carentonosaurus Rage and Néraudeau (2004) and the unnamed “pythonomorph from Touraine” (Houssaye, 2010) (see Houssaye, 2013). The shape of the centrum in ventral view is distinct from the Y-shape (marked by a strong width reduction just behind the paradiapophyses) observed in the non-pachyostotic stem-ophidiomorphs Dolichosaurus and Coniasaurus (Caldwell, 1999 and Caldwell, 2000). In addition, the articulation of the zygapophyses is nearly horizontal in these two taxa (A.H., pers. obs.). Although diagnostic characters for mosasauroid vertebrae are almost lacking, the shape of the centrum, lateral projection and anterior position of the paradiapophyses, the relative size of the neural canal, in accordance with the general size of the vertebrae, strongly suggest a mosasauroid affinity. Prezygapophyses project much more anteriorly and are more strongly inclined on the horizontal than in the plesiopedal and plesiopelvic mosasauroids (sensu Bell and Polcyn, 2005 and Caldwell and Palci, 2007; A.H., pers. obs.).

The Villeveyrac-L’Olivet material is distinct from most hydropedal mosasauroids, notably in the relatively long centrum, the obliquity of the cotyle-condyle system in (at least) dorsal vertebrae, and the lack of reduction or vertical orientation of the zygapophyses in the dorsal region. The shape of the centrum in ventral view differs from Dallasaurus, where it is more elongated, cylindrical and convex in ventral view (Bell and Polcyn, 2005). The material cannot be distinguished from Tethysaurus, based on the original description (Bardet et al., 2003), but also based on the combination of the following observations: prezygapophyses strongly projecting anteriorly and not so much laterally, and much inclined on the horizontal; prezygapophyseal facets elongated; sharp epidiapophyseal ridge posterior to the dorsal border of the paradiapophysis, protruding posterolaterally; dorsal centra flat with a sagittal shallow furrow; and caudal vertebra centrum (at least in anterior caudals) much longer than wide and high. However, the impossibility to determine if a true zygosphene-zygantrum system occurs prevents us from attributing this material to Tethysaurus. However, it differs from Pannoniasaurus (Tethysaurus's sister taxon) in the occurrence of paracotylar foramina (Makádi et al., 2012).

Limbs

Metacarpal (UP-VIL-10-06; Fig. 2O–P)

The first right metacarpal is a rather large bone (length = 37 mm; proximal width = 22 mm; distal width = 18 mm). It is long and not twisted. The proximal end is expanded into a shallow transverse oval cup for articulation with the distal carpal row. This cup is bordered laterally, medially and posteriorly by large crests. The distal end is more slender that the proximal one. It has a convex hemi-circular rolling surface for articulation with the proximal phalanx. This shape permits great extension and flexion movements but restricted lateral ones. Because of the development of both proximal and distal extremities, both lateral and medial surfaces of the shaft are concave in dorsal and ventral views. Both dorsal and ventral surfaces are almost smooth.

As a whole, this first metacarpal is comparable in shape to that of the extant Varanus, though being more robust in general proportions, and of basal mosasauroids for which this bone is known (N.B., pers. obs.). In general, the first metacarpal is more than one third of the tibia length in mosasauroids whereas it is less than 1/3 in Varanus (N.B. pers. obs.). The respective length of the first metacarpal and of the tibia (see description below) from Villeveyrac-L’Olivet (MC1: 37 mm; Tibia: 108 mm) indicates that these two bones probably belong to the same taxon and individual (a hypothesis under which the ratio was more than 1/3), suggesting that the MC1 probably belonged to a plesiopedal mosasauroid rather than to a varanoid (sensu Bell and Polcyn, 2005).

Tibia (UP-VIL-10-08; Fig. 2Q–R)

The preserved right tibia is slightly crushed, especially its proximal end. It is a long and narrow bone, its length being approximately three times greater than its proximal width (see Table 2). Though slightly crushed, the shaft was probably almost flat and not bowed in life. Both anterior and posterior margins are slightly concave, the anterior more than the posterior one. The dorsal (extensor) surface of the tibia is flat and bears on its proximo-anterior part the cnemial crest; though broken, it appears slightly developed. The ventral (flexor) surface is also almost flat but, because of the larger condyle expansion it bears a proximo-medial domed relief. The proximal articulation for the femur is larger than the distal one. It is composed of two flat condyles, the posterior being larger than the anterior one, both separated by a slight oblique groove. The distal articulation is saddle-shaped for articulation with the astragalus. Both epiphyses are clearly ossified.

In hydropedal mosasauroids (sensu Bell and Polcyn, 2005), the tibia is generally a short bone, being slightly longer than wide (Russell, 1967). The rather long and narrow morphology of UP-VIL-10-08, in which the proximal extremity represents about 1/3 of the total length of the bone, looks like that of basal mosasauroids (Caldwell, 1996). These proportions differ from those of limbed terrestrial squamates, such as Varanus, in which the tibia is a much more elongate and gracile bone, its proximal extremity representing about 1/4 of the total length (see Table 2). More especially, the Villeveyrac-L’Olivet tibia resembles that of plesiopedal/plesiopelvic mosasauroids in which this bone is preserved (Aigialosaurus, Carsosaurus, Komensaurus and Haasiasaurus; see references in Table 2). However, UP-VIL-10-08 differs from all these mosasauroids by several characters that could indicate as a whole that it belongs to a more derived mosasauroid, possibly of the plesiopedal/hydropelvic grade. They are: •

clearly ossified epiphyses;

The respective size of the bones and the fact that they have been found in relative close association indicate that at least some of them could belong to the same individual. As well, the anatomical characters of the bones indicate that probably only one squamate taxon is represented in the Villeveyrac-L’Olivet continental assemblage. The combination of characters found on both the frontal, the vertebrae and possibly the tibia, identifies the taxon as a basal mosasauroid probably of the plesiopedal/hydropelvic grade and belonging to the Tethysaurinae clade (sensu Makádi et al., 2012). On the basis of the size estimations made on the frontal, the tibia and the largest vertebrae, the Villeveyrac-L’Olivet mosasauroid was probably a large animal about 3 meters long, with limbs poorly adapted to an aquatic life.

The occurrence of such a basal mosasauroid in the continental outcrop of the Villeveyrac-Mèze Basin, though appearing at first sight surprising, is not so unexpected and, on the contrary, rather reinforces the idea of the occurrence of freshwater mosasauroids, a group until recently considered as strictly marine. Indeed, several reports of mosasauroid remains in comparable Late Cretaceous continental outcrops of Europe have been made in Hungary (Makádi et al., 2012), Portugal (Antunes and de Lapparent de Broin, 1988, Bardet et al., 2008 and Sauvage, 1897–1898), and southern France (E. Buffetaut and L. Makádi, pers. comm; see Csiki-Sava et al., 2015). It is not impossible that some large varanoid vertebrae mentioned by Buffetaut et al. (1999) and Laurent et al. (2002) could belong to the same, or at least a closely related mosasauroid.

The ecological implications of the occurrence of such an aquatic taxon in these continental palaeoenvironments remain unresolved, though several hypotheses can be proposed. As noted above, the Villeveyrac-L’Olivet mosasauroid was a rather large animal, about 3 m long. Pannoniasaurus of Hungary was larger, probably up to 6 m long (see Makádi et al., 2012) and some vertebrae found in southern France are also relatively large (E. Buffetaut and L. Makádi, pers. comm.). At least in Villeveyrac-L’Olivet, the faunal list (see Table 1) indicates that large aquatic predators are lacking as the crocodyliforms are both poorly represented and of moderate size. The mosasauroid of Villeveyrac-L’Olivet, because of its relatively large size, could represent one of the largest known predators in the aquatic zones of this continental environment, as noted previously for Pannoniasaurus (Makádi et al., 2012). Finally, though mosasauroids have acquired ovoviviparity very early in their evolution, indicating that they probably were freed from the terrestrial realm (Caldwell and Lee, 2001) and gave birth in open oceanic habitats (Field et al., 2015), it cannot be ruled out that these relatively protected freshwater continental environments could have been used by more basal mosasauroids as nurseries.