Rebbachisauridae, Bonaparte, 1997

Canudo, José I., Carballido, José L., Garrido, Alberto & Salgado, Leonardo, 2018, A new rebbachisaurid sauropod from the Aptian-Albian, Lower Cretaceous Rayoso Formation, Neuquén, Argentina, Acta Palaeontologica Polonica 63 (4), pp. 679-691 : 686-689

Rebbachisauridae

The amount of information recovered from the skull of Lavocatisaurus and its position amongst the basal rebbachisaurids provides the first opportunity to investigate the evolution of certain characters in one of the least-known lineages of Neosauropoda. Below, we list and briefly discuss the evolution of certain characters upon which the new taxon described here sheds further light.

Preantorbital fenestra.—The presence of a preantorbital fenestra was recognized as a derived character of Jobaria and more derived sauropods ( Wilson and Sereno 1998; Wilson 2002). Amongst Neosauropoda it is a widespread character but with varying degrees of development. This fenestra is poorly developed in basal macronarians (e.g., Camarasaurus , Giraffatitan ; Madsen et al. 1995; Janensch 1935), being similar in shape and size to the posterior maxillary foramen of basal sauropodomorphs (e.g., Plateosaurus ; Sereno 2007). It differs from the posterior maxillary foramen in its anteroventral direction and in its connection to the antorbital fenestra ( Wilson and Sereno 1998). Its presence in derived titanosaurs and diplodocids is much clearer, as the preantorbital fenestra is usually a well-developed opening in the maxilla (e.g., Diplodocus , Tapuiasaurus ; Marsh 1884; Zaher et al. 2011). The maxillae here described possess a large opening that can be identified as the preantorbital fenestra. Although it resembles that of diplodocids and derived titanosaurs, in Lavocatisaurus the preantorbital fenestra is extremely developed, occupying a great area of the maxilla. As noted above, a similarly placed and apparently equally large fenestra was first identified as the antorbital fenestra in Nigersaurus ( Sereno et al. 1999: fig. 2D; Sereno and Wilson 2005: fig. 5.5). To judge by the published images ( Sereno et al. 1999; Sereno and Wilson 2005) and personal observation of the preserved Nigersaurus maxilla (MNN GDF512), the latter fenestra corresponds more probably to the preantorbital fenestra. Indeed, the anterior margin of the preantorbital fenestra is at the height of the subnarial foramen in Nigersaurus ( Sereno et al. 1999: fig. 2D), whereas the antorbital fenestra, as reconstructed ( Sereno and Wilson 2005; Sereno et al. 2007), is well above the foramen. Therefore, Lavocatisaurus and Nigersaurus share the presence of a large preantorbital fenestra, which occupies around half of the maxillary body. The preantorbital fenestra of both rebbachisaurids is not inset in any kind of fossa, differing from most flagellicaudatans known (e.g., Dicraeosaurus , Apatosaurus ; see Tschopp et al. 2015).

Infratemporal fenestra. In basal sauropodomorphs, the infratemporal fenestra is posteriorly placed with respect to the orbit, being anteriorly delimited by the postorbital anterodorsally) and jugal (anteroventrally), and posteriorly by the squamosal (posterodorsally) and the quadratojugal posteroventrally) (e.g., Plateosaurus ). A similar position and configuration of the infratemporal fenestra is observed in Camarasaurus and Europasaurus ( Madsen et al. 1995; Marpmann et al. 2015), although in these taxa the fenestra is placed posteroventrally to the orbit, with the anterior margin close to the anterior margin of the orbit. In contrast, in most sauropods the squamosal–quadratojugal contact is missing, with the quadrate forming a small posterior edge of the infratemporal fenestra (see Wilson and Sereno 1998). This is observed in most basal sauropods and neosauropods (e.g., Giraffatitan , Diplodocus ). In diplodocimorphs, the infratemporal fenestra projects anteriorly, its anterior margin going beyond that of the orbit (e.g., Diplodocus , Apatosaurus ; Berman and McIntosh 1978). This anterior displacement of the infratemporal fenestra seems to be greater in rebbachisaurids, with that of Nigersaurus placed completely anteriorly to the orbit ( Sereno et al. 2007). The anterior displacement of the infratemporal fenestra in rebbachisaurids seems to have had its origin early in their evolution, as most of the infratemporal fenestra is positioned anteriorly to the orbit in Lavocatisaurus , with its posterior margin slightly posterior to the anterior margin of the orbit Fig. 3). Amongst rebbachisaurids, this is not the only morphological change in the configuration of the infratemporal fenestra. In Lavocatisaurus the jugal articulates with the squamosal, a novel articulation not previously described for any sauropodomorph, but that can also be detected in Limaysaurus and Nigersaurus (see description). Therefore, the infratemporal fenestra of rebbachisaurids is not delimited by the postorbital. In Lavocatisaurus this fenestra is certainly bounded by the jugal (anteriorly and posterodorsally) and the squamosal (posteriorly). The wide ventral end of the squamosal shows no signs of articulation with the quadratojugal, making it possible that the squamosal fails to contact the quadratojugal, as in Flagellicaudata ( Tschopp et al. 2015). Nevertheless, a badly preserved articulation for the quadratojugal (ventrally) and the jugal (dorsally) can be detected in the preserved quadrate, making it impossible to discern whether the squamosal–quadratojugal contact was present or not. Based on the restoration of Nigersaurus , the ventral margin of the infratemporal fenestra should be bounded by the quadratojugal (not preserved either in Lavocatisaurus or in Nigersaurus ).

Teeth and keratinized sheaths.—The well-preserved premaxillae, maxillae, dentaries and teeth of Lavocatisaurus yield an improved understanding of many aspects of the feeding adaptations in Rebbachisauridae . A great number of teeth were described for Nigersaurus , which has 58 upper teeth (4 premaxillary and 25 maxillary per side) and 68 lower teeth (34 in each dentary). In contrast, only 7 alveoli are present in the single dentary recovered for Demandasaurus (Torcida-Baldor et al. 2011) . In Lavocatisaurus we found 32 teeth in the upper series (4 in the premaxilla and 12 in the maxilla per side), and 44 teeth are present in the lower series (22 per dentary). The number of teeth in Lavocatisaurus is similar to that of non-neosauropod sauropods (e.g., 20 dentary teeth are present in Jobaria ; Sereno et al. 1999), but differs from the number of lower tooth positions in other, non-rebbachisaurid diplodocimorphs and camarasauromorphs ( Wilson and Sereno 1998). Basal rebbachisaurids thus seem to keep the plesiomorphic state (20 teeth or more per dentary), with a marked increment in tooth positions in Nigersaurus and a decrease in Demandasaurus .

All the preserved teeth in Lavocatisaurus (both upper and lower) have highly asymmetrical enamel, as in derived rebbachisaurids ( Nigersaurus , Limaysaurus , Demandasaurus ) and differing from the symmetrical enamel of the more basal rebbachisaurid Comahuesaurus ( Salgado et al. 2004) . This indicates that, whereas symmetrical enamel was present in basal forms, asymmetrical enamel evolved prior to the diversification of the clade into its two major lineages (Rebbachisaurinae and Limaysaurinae). As noted above, the upper teeth of Lavocatisaurus are much larger (both longer and wider) than those of the lower series. Although a similar pattern was mentioned for Nigersaurus ( Sereno et al. 2007) , in the case of Lavocatisaurus this difference is of an order such that the largest maxillary teeth are almost three times bigger than the dentary teeth ( Fig. 3). The upper teeth present a low-angle wear facet on their lingual surface, but none of them has the second, labial, wear facet as described for Nigersaurus , although slightly polished surfaces are recognized on its tips. The lingual low-angle wear facet in Nigersaurus has been interpreted as being a result of toothtooth occlusion. The enormous differences in tooth size in Lavocatisaurus , especially close to the mid-point of the tooth series, make it less likely that the smooth and planar low-angle wear facets were the product of tooth-to-tooth occlusion. This is due to the fact that a single lower tooth cannot produce a single wear facet wider than its root. Therefore, the tooth-to-tooth abrasion producing this low-angle wear facet should be revised, at least in Lavocatisaurus , pending further studies on its micro-wear (see below). The second, labial high-angle facet present in Nigersaurus is also present in diplodocids and dicraeosaurids, and has been regarded as being a result of tooth-to-plant abrasion during groundlevel browsing ( Sereno et al. 2007). The absence of a labial, high-angle wear facet in Lavocatisaurus could indicate a different browsing height in basal rebbachisaurids or different movements when cutting plants, although another possibility is that it results from different types of food in diplodocoids. Whether or not the dentary teeth present similar facets to those of the upper teeth remains an open question calling for further and more complete discoveries.

It is widely accepted that sauropods lacked any masticatory adaptation, such as cheeks ( Upchurch et al. 2007) or beaks (Sereno 2007). Nevertheless, the presence of a beak was suggested for Bonitasaura and other related taxa, based on the horizontal edge posterior to the tooth-bearing region ( Apesteguía 2004; but see Wilson 2005). The presence of a keratinized sheath was suggested in Nigersaurus ( Sereno et al. 2007) on the basis of the presence of numerous neurovascular grooves in the maxilla and dentary. The presence of a keratinous beak was recently proposed by Wiersma and Sander (2016) in Camarasaurus and other sauropods, based on the presence of exposed roots and extensive vascular foramina. For these authors, the presence of such a structure would explain the numerous ITRs reported for several sauropods; it would have served as extra protection from abrasive plants and provided a continuous cutting surface. The presence of ITRs and dentigerous bones in Lavocatisaurus is evidence against the consumption of bones by insects ( Britt et al. 2008), reinforcing the taphonomic separation of the teeth as suggested by Wiersma and Sander (2016). As in Nigersaurus , the premaxilla and maxilla of Lavocatisaurus are laterally covered with numerous irregular grooves that are more numerous in the lower part of these bones. A similar pattern, though much more developed, can be observed in the dentaries. The dorsal margin of the dentaries has a depressed area that runs posteriorly from the whole lateral margin of the dentary ramus and ascends dorsally at the anterolateral margin. A similar, but less developed, morphology was described in Demandasaurus and suggested to be equivalent to the vascular canal shown in Nigersaurus by Sereno et al. (2007; see also Torcida-Baldor et al. 2011). In Lavocatisaurus the laterodorsal surface of the dentary, above this marked step, is full of irregular, middle-to-largesized pores and neurovascular foramina. Such morphology is frequently cited as evidence of keratin sheaths (see Sereno 2007). This region of the dentary coincides with the horizontally placed, anterior lateral edge of the maxilla ( Fig. 3), which is sharp and covered with irregular grooves, both medially and laterally. Given this morphology, we do not rule out the presence of keratinous sheaths in the anteriormost edentulous region of the maxillae and dentaries. The presence of a keratinous sheath covering the anteriormost region of the snout could explain the ITRs of Lavocatisaurus (see Wiersma and Sander 2016), as well as the polished, single, low-angle wear facet of the upper teeth (see above). If this is the case, the lingual surface of the teeth, with thinner enamel, will have been subject to wear against a keratinous beak. Nevertheless, further analyses are needed in order to test this hypothesis further.