Xenoposeidon is the earliest known rebbachisaurid sauropod dinosaur

Introduction

The fossil record of sauropod dinosaurs extends through most of the Mesozoic, from the Late Triassic (Lallensack et al., 2017) to the very end of the Cretaceous (Riera et al., 2009; Sellés et al., 2016). However, their record in the earliest Cretaceous, as for most dinosaurs, is much less rich (Tennant et al., 2017). In fact, almost the entire record of sauropodomorphs in the first three ages of the Cretaceous rests on fossils from Europe (Tennant, Chiarenza & Baron, 2018: fig. 11: parts C–F)—only Leinkupal from the Berriasian/Valanginian of Argentina (Gallina et al., 2014), and perhaps Euhelopus (based on a revised age for the Mengyin Formation, Borinder, Poropat & Kear, 2016) have been named in this period from outside of Europe. In this context, sauropods from earliest Cretaceous formations in Europe are particularly important for our understanding of the evolution of this group.

Xenoposeidon proneneukos is a neosauropod sauropod dinosaur from the Berriasian–Valanginian (earliest Cretaceous) Ashdown Formation of the Wealden Supergroup of southern England (Taylor & Naish, 2007). It is represented by a single partial mid-to-posterior dorsal vertebra, NHMUK PV R2095 (Fig. 1; BMNH R2095 at the time of the original description by Taylor & Naish, 2007; NHMUK R2095 at the time of Taylor, 2017). This element consists of the centrum and the base of a tall neural arch, broken off below the transverse processes and zygapophyses. Despite its fragmentary nature, it is recognisably different from all other sauropods, and Taylor & Naish (2007) diagnosed it on the basis of six characters that they considered unique among sauropods. Here, I will present a revised diagnosis.

Taylor & Naish (2007: 1554–1557) compared the Xenoposeidon vertebra to those of the main neosauropod groups—Diplodocoidea, Camarasauridae, Brachiosauridae and Titanosauria—and concluded that it could not be convincingly referred to any of these groups (see Fig. 2). Their phylogenetic analysis (pp. 1157–1558 and fig. 6) corroborated this by recovering Xenoposeidon as a neosauropod in all most parsimonious trees, but in a polytomy with all other neosauropods, wholly unresolved save that the clade Flagellicaudata was preserved in all MPTs.

In light of Wilson & Allain’s (2015) redescription of the rebbachisaurid diplodocoid Rebbachisaurus garasbae from the mid-Cretaceous of Morocco, and the availability of more photographs and models of rebbachisaurid material, it has now become possible to reinterpret the idiosyncratic system of laminae found in Xenoposeidon, and to refer it confidently to an existing family-level clade.

Validity of Xenoposeidon

Upchurch, Mannion & Barrett (2011: 497–498), in a review of Wealden sauropods, reassessed Xenoposeidon, accepting its validity and concurring with Taylor & Naish (2007) that it was difficult to place within any recognised sauropod clade. However, they tentatively proposed a basal somphospondylan identity for it. Similarly, Mannion et al. (2013: 151) tentatively considered it most likely a basal macronarian.

D’Emic (2012: 651) asserted that ‘the absence of diagnostic features renders Xenoposeidon a nomen dubium’. However, his assessment was mistaken in several respects. For example, the extension of the base of the neural arch to the posterior extremity of the centrum is clearly not, as he asserted, due to damage. D’Emic claimed that dorsal vertebrae illustrated by Osborn & Mook (1921: plates LXIX and LXXII) have forward-sloping neural arches resembling those of Xenoposeidon: in reality, only one posterior dorsal vertebrae out of four complete dorsal columns illustrated in that monograph shows a forward slope, and it differs so much from its fellows that this can only be interpreted as the result of crushing. D’Emic further claimed that the lamina patterns observed in Xenoposeidon can be recognised in other sauropods, but I have been unable to find morphology resembling them in the descriptions he suggests: Osborn and Mook 1921 for Camarasaurus, Riggs, 1903 for Brachiosaurus (probably a typo for Riggs, 1904, which also does not depict similar patterns), Carballido et al. (2011) for Tehuelchesaurus. However, a similar pattern does appear in Rebbachisaurus, as will be discussed below. D’Emic (2012: 651) is probably correct that the ‘asymmetric neural canal’ described by Taylor & Naish (2007: 1553–1554) is a misreading of the tall anterior fossa as being the anterior portion of the neural canal: as Taylor and Naish pointed out, ‘The vacuity is filled with matrix, so the extent of its penetration posteriorly into the neural arch cannot be assessed’. Nevertheless, the shape and size of the fossa is unique among sauropods, and it is bounded by laminae which do not seem to be medial CPRLs—see below. In summary, as will be shown in more detail below, X. proneneukos is a valid, diagnosable taxon, contra D’Emic (2012).

Revised diagnosis

Xenoposeidon differs from all other known sauropods in five respects. Compare the following characters with the state in mid-posterior dorsal vertebra of other sauropods as shown in Fig. 2.

1. Neural arch covers dorsal surface of centrum. The posterior margin is continuous with that of the centrum, such that in lateral view the posterior margin of the vertebra forms a single smooth curve (Fig. 3: 1a, 1b). In most sauropod dorsals, the base of the neural arch is some way forward of the posterior margin of the centrum. Even in R. garasbae, where the posterior margin of the neural arch approaches that of the centrum, there is a distinct kink in lateral view between the posteroventral slope of the ventral part of the arch’s posterior border, and the vertical margin of the centrum (Fig. 4B).

2. Neural arch slopes anteriorly 30°–35° relative to the vertical, as determined by the orientation of the posterior articular surface of the centrum (Fig. 3: 2). In fully lateral view, vertical orientation of the posterior articular surface is difficult to determine because the bone extends slightly further posteriorly at centrum mid-height than more dorsally or ventrally, but it is easy to see in a slightly posterolateral view, as can determined from the 3D model (File S1).

3. Sharp oblique lamina above lateral fossa forms ventral border of a broad, flat area of featureless bone. The fossa beneath this ridge-like lamina (Fig. 3: 3a) contains nested within it a deeper lateral foramen; and above it, below the ‘M’-shaped complex of laminae that are discussed in detail below, the bone is quite flat and smooth (Fig. 3: 3b).

4. Very large, teardrop-shaped anterior fossa, nearly as tall as the posterior articular facet of the centrum and half as transversely broad as it is tall (Fig. 3: 4). In R. garasbae, the fossa is proportionally nearly as tall, but much narrower (Wilson & Allain, 2015: fig. 3E).

5. Arched laminae form vaulted boundary of anterior fossa. These laminae (Fig. 3: 5a) cannot be interpreted as the medial CPRLs (Fig. 3: 5b), as those arise separately from the neural arch pedicels. These laminae arising from the pedicels cannot instead be regarded lateral CPRLs, as those laminae are located on the lateral face of the neural arch, as will be discussed below. Furthermore, the point where the arched supporting laminae meet at the top of their arch is located some way ventral to the location inferred for the prezygapophyses based on the trajectory of the preserved portions of the medial CPRLs (Fig. 5).

Reinterpretation of Xenoposeidon

Taylor & Naish’s (2007) history, geography, geology and description of the Xenoposeidon specimen require no revision, and should continue to be considered definitive: this paper does not supersede the earlier description, but should be read in conjunction with it.

The illustrations of the specimen in the original paper, however, were in monochrome and omitted the dorsal and ventral views. The present paper supplements these illustrations with a colour depiction from all six cardinal directions (Fig. 1), an oblique view (Fig. 5) and a high-resolution 3D model of the specimen (File S1).

More importantly, Taylor & Naish’s (2007) interpretation of some features of the vertebra, particularly the ‘M’-shaped complex of laminae on the lateral faces of the neural arch, was mistaken. Although the neural spine and dorsal part of the neural arch are missing, including the pre- and post- zygapophyses and lateral processes, they wrote that ‘sufficient laminae remain to allow the positions of the processes to be inferred with some certainty’. But their inferences were incorrect. Taylor & Naish (2007: 1553) interpreted the cross-shaped structure on the anterodorsal part of the left lateral face of the neural arch as the site of the parapophysis, despite the lack of any articular facet in that location. This influenced their interpretation of the four laminae that met at that point as the ACPL below, the PPDL above, the PRPL anteriorly and an unnamed accessory infraparapophyseal lamina posteroventrally, which they interpreted as homologous with a PCPL (Fig. 6A). Similarly, they did not attempt to identify either the long lamina running up the posterior edge of the lateral face of the neural arch (designating it only ‘posterior lamina’) or the lamina forming a shallow ‘V’ with the ‘accessory infraparapophyseal lamina’, simply calling it an ‘accessory postzygapophyseal lamina’ (Fig. 6A).

Among the various unusual features of the Xenoposeidon vertebra, the ‘M’-shaped set of laminae is immediately apparent in lateral view (Fig. 4A): a line can be traced from the anterior margin of the neural arch’s lateral face up the ACPL to the cross that was interpreted as the parapophysis, then posteroventrally down the ‘accessory infraparapophyseal lamina’, then posterodorsally up the ‘accessory postzygapophyseal lamina’ and finally down the posterior margin of the neural arch’s lateral face, along the ‘posterior lamina’. Photographs of other specimens that were available to us at this time did not apparently manifest similar features.

But subsequent work on R. garasbae (Wilson, 2012: 100, fig. 9; Wilson & Allain, 2015)—and an associated video of the rotating vertebra (see Acknowledgements)—show that Rebbachisaurus has a similar complex of laminae (Fig. 4B), which are described by Wilson & Allain (2015: 6) as the second of the eight autapomorphies that they listed for the species: ‘infrazygapophyseal laminae (lat. CPRL, CPOL) that intersect and pass through neighbouring costal laminae (ACPL, PCDL) to form an ‘M’ shape’.

Because the illustrated dorsal vertebra of Rebbachisaurus—MNHN MRS 1958—is substantially complete, it is possible to follow the trajectories of the laminae that participate in the ‘M’ to their apophyses, and so determine their true identities (Fig. 4). The two vertically oriented laminae—the outer pillars of the ‘M’—continue up past the top of the ‘M’. The anterior one supports the parapophysis, and the posterior supports the diapophysis. Also, the two laminae that form the valley in the middle of the ‘M’ support the prezygapophysis and postzygapophysis: in both cases, as noted by Wilson and Allain, they intersect the vertical lamina before continuing to meet their respective zygapophyses. The four laminae that make up the ‘M’, from anterior to posterior, are therefore the ACPL, posterior part of the lateral CPRL, anterior part of the CPOL, and PCDL. Of these, the intersection between the ACPL and lateral CPRL is clearly visible in left lateral view of MNHN MRS 1958 (Fig. 4B). The intersection between the CPOL and PCDL is less apparent in this view, though clear in three dimensions. Both laminae continue dorsally beyond this intersection, but their paths are somewhat changed at the point of contact, with the dorsal portion of the PCDL inclining more anteriorly, and the rod-like CPOL apparently passing through the sheet of bone formed by the PCDL to meet the postzygapophysis.

The referred R. garasbae specimen NMC 50844 described and illustrated by Russell (1996: 388–390 and fig. 30) is also broadly consistent with this morphology. It is not possible to be definite about the laminar intersection based only on line drawings of the specimen from the four cardinal directions, but, as illustrated in Russell’s fig. 30c, the lateral CPRL does appear to pass through the ACPL. The CPOL seems in this specimen to originate posterior to the PCDL, not intersecting with it. However, this difference from the holotype dorsal may be serial variation since, as Russell notes, the relatively longer centrum of his specimen indicates a more anterior serial position than for the holotype’s dorsal vertebra; and this interpretation is corroborated by the observation that, based on lamina trajectories, the anteroposterior distance between the parapophysis and diapophysis was less in NMC 50844 than in the holotype.

In light of these Rebbachisaurus specimens, the mysterious laminae of Xenoposeidon are more readily explained. It is now apparent that the cross on the side of the Xenoposeidon vertebra is not the site of the parapophysis, as Taylor & Naish (2007: 1553) proposed, but merely the intersection of two laminae that pass right through each other: the ACPL, running dorsolaterally, and the lateral CPRL, extending anterodorsally to the (missing) prezygapophysis (Fig. 6B). Similarly, the ‘posterior lamina’ is the PCDL, and it intersects with the CPOL, though the intersection is lost in NHMUK PV R2095 (Fig. 6B). Both the parapophysis and diapophysis of the Xenoposeidon vertebrae would likely have been located some distance above the preserved portion, the former anterior to the latter.

It appears from Dalla Vecchia (1999: fig. 47, left part) that in the holotype and only vertebra of Histriasaurus boscarollii, ‘WN-V6’, the CPOL on the right side of the vertebra intersects with the PCDL in the same way as in Rebbachisaurus, though it is not possible to determine whether the lateral CPRL similarly intersects the ACPL. Dorsal vertebrae of other rebbachisaurid sauropods, however, do not appear to feature the distinctive ‘M’ and intersecting laminae of Rebbachisaurus and Xenoposeidon:

• The 3D model of a dorsal vertebra of Nigersaurus (Sereno et al., 2007) shows that the lateral CPRLs originate anterior to the ACPLs and the CPOLs posterior to the PCDLs, so that there is no intersection. A subtle ‘V’ shape does appear high up on the lateral faces of the neural arch, between the ACPL and the PCDL, but it seems unrelated to the lateral CPRL and CPOL.

• Unpublished 3D models of an anterior dorsal neural arch and a more posterior dorsal vertebra of Katepensaurus (L. M. Ibiricu, 2017, personal communication) as illustrated in figs. 3A and 5A of Ibiricu et al. (2017) show that in both vertebrae, the lateral CPRLs originate anterior to the ACPLs, and the CPOLs seem to originate posterior to the PCDLs—though damage to the posterior portion makes the latter uncertain.

• The laminae do not appear to intersect in the illustrated dorsal vertebra of Demandasaurus (Torcida Fernández-Baldor et al., 2011: fig. 9).

• The sole known vertebra of Nopcsaspondylus seems to have an entirely different pattern of lamination (Mannion, 2010: fig. 5) with no lamina intersections like those of MNHN MRS 1958.

No determination can be made for other rebbachisaurids as they are insufficiently preserved or illustrated (e.g. Limaysaurus, Amazonsaurus, Cathartesaura), or simply lack posterior dorsal vertebral material (e.g. Rayososaurus, Tataouinea, Comahuesaurus, Zapalasaurus).

However, one cannot rule out the possibility that complete and well-preserved posterior dorsal vertebrae of most or all rebbachisaurids have Rebbachisaurus-like intersecting laminae: even in those species for which a well-preserved vertebra lacks them, this could be due to serial variation, with these features only fully developing in the more posterior dorsals.

Xenoposeidon, then, resembles Rebbachisaurus in the possession of a distinctive ‘M’ on the lateral face of the neural arch, in the intersecting lateral CPRL and ACPL, and in the elevation of the parapophysis above the level of the prezygapophysis, as inferred from the trajectories of the lateral CPRL and ACPL—a complex of related features. Although at first glance they do not closely resemble each other, Xenoposeidon and Rebbachisaurus, while geometrically different, are topologically similar.

A superficially similar ‘M’-shaped complex of laminae is also found in dorsal vertebrae of the saltasaurine titanosaur Neuquensaurus (Salgado, Apesteguía & Heredia, 2005: figs. 3 and 4). However, this is not homologous to the situation in Rebbachisaurus and Xenoposeidon, as different laminae are involved: Salgado, Apesteguía & Heredia (2005: 626) identify the inner arms of the ‘M’ as the PCPL and a novel accessory PCDL which they term the APCDL. (This APCDL, together with the ventral portion of the PCDL proper, constitute the ‘ventrally forked infradiapophyseal lamina’ of Salgado, Coria & Calvo, 1997.) It is apparent from the illustrations of Salgado, Apesteguía & Heredia (2005: figs. 3C and especially 4A–4B) that the APCDL of Neuquensaurus is not contiguous with, and cannot be considered a part of, the CPOL.

Regarding the significance of the elevated parapophysis, since no complete or nearly complete rebbachisaurid dorsal column has been described, comparisons with other, better represented sauropods are warranted. In the probable basal diplodocoid Haplocanthosaurus, the dorsal margin of the parapophyseal facet reaches the level of, and is coincident with, the prezygapophyseal facet around dorsal vertebra 7 or 8, but never rises any higher than this in more posterior vertebrae (Hatcher, 1903: plate I). In the more distantly related diplodocid diplodocoids Apatosaurus and Diplodocus, the parapophysis never migrates far enough dorsally to reach a position level with the prezygapophyses, even in the most posterior dorsals (Gilmore, 1936: plate XXV; Hatcher, 1901: plates VII, VIII).

Taylor & Naish (2007: 1554) argued that Xenoposeidon could not at that time be convincingly referred to Rebbachisauridae because Rebbachisaurus differs from NHMUK PV R2095 in five ways: ‘possession of a very prominent PCDL (mistranslated as PCPL in the original), large and laterally diverging prezygapophyses, depressions at the base of the neural arch (Bonaparte, 1999: 173), lateral foramina not set within fossae, and a strongly arched ventral border to the centrum’. Of these features, the first is now recognised as occurring in Xenoposeidon; the second appears to be an outright error, as the prezygapophyses of Rebbachisaurus meet on the midline, and in any case the situation in Xenoposeidon is not known. ‘Depressions at the base of the neural arch’ seems to be a mistranslation of Bonaparte’s original Spanish, ‘profundas depresiones en la base de la espina neural’, which refers not to the neural arch but the neural spine, and since this portion is not preserved in Xenoposeidon, it is not informative for our purposes. The 3D model of the Rebbachisaurus dorsal suggests that its lateral foramina are set in shallow depressions, but these are far less pronounced than those of Xenoposeidon. This leaves the stronger arching of the ventral border of the centrum in Rebbachisaurus, but this difference is not convincing given that the ventral margin of the NHMUK PV R2095 posterior cotyle is incomplete and the anterior end of the centrum is missing: the ventral border was likely rather more arched when the vertebra was complete.

In conclusion, the weight of morphological evidence, including the camerate internal tissue structure of the centrum that is exposed in anterior view (Fig. 1B), supports including Xenoposeidon within Rebbachisauridae. This is compatible with the observation of Taylor & Naish (2007: 1557), in whose phylogenetic analysis ‘various most-parsimonious trees also recover Xenoposeidon in many other positions, including as a … rebbachisaurid’.

Revised reconstruction

In light of the reassignment of Xenoposeidon to Rebbachisauridae, and the reinterpretation of its laminae, I present a new reconstruction of how the vertebra NHMUK PV R2095 might have looked when complete (Fig. 7). As in MNHN MRS 1958, the parapophysis and diapophysis are both elevated above the zygapophyses. The lateral CPRL and ACPL meet at a point where they project laterally about the same distance from the vertebra, as is apparent from the preserved portion of the vertebra; but the CPOL is assumed to pass through a sheet-like PCDL as in Rebbachisaurus, because it is clear from breakage in NHMUK PV R2095 that the PCDL extended further laterally from the body of the neural arch than the preserved portion indicates. The neural spine, composed as in Rebbachisaurus of pre- and post-spinal laminae together with the left and right SDLs, is shown fading out at the top, as there is no way to determine its height. The condyle that is the centrum’s anterior articular surface is reconstructed as only slightly convex, as in Rebbachisaurus. It is shown almost immediately anterior to the preserved portion of the centrum, because the camerae in the dorsal part of the anteriormost preserved portion reach their point of dorsalmost excavation a short distance behind the front part, indicating that the cortex at this point was curving down over the camerae to form the condyle.

It is instructive to compare this with the original reconstruction of the vertebrae (Taylor and Naish, 2007: fig. 5). The new reconstruction has a taller neural arch, a far more elevated parapophysis, a more posteriorly located diapophysis (no longer dorsal to the parapophysis) and a shallower condyle, as that of the original reconstruction was drawn with those of brachiosaurs in mind.

Systematic Palaeontology

• Dinosauria Owen, 1842

• Saurischia Seeley, 1888

• Sauropodomorpha Huene, 1932

• Sauropoda Marsh, 1878

• Neosauropoda Bonaparte, 1986

• Diplodocoidea Marsh, 1884

• Rebbachisauridae Sereno et al., 1999

• Xenoposeidon Taylor and Naish, 2007

• Xenoposeidon proneneukos Taylor and Naish, 2007

Holotype. NHMUK PV R2095, the Natural History Museum, London. A mid-to-posterior dorsal vertebra consisting of partial centrum and neural arch.

Revised diagnosis: Differs from all other sauropods in the following characters:

1. Neural arch covers dorsal surface of centrum, with its posterior margin continuous with that of the centrum.

2. Neural arch slopes anteriorly 30°–35° relative to the vertical.

3. Sharp oblique lamina above lateral fossa forms ventral border of a broad, flat area of featureless bone.

4. Very large, teardrop-shaped anterior fossa.

5. Arched laminae form vaulted boundary of anterior fossa, enclosed within the medial CPRLs.

Discussion

Supplemental Information