Tendaguria

Mannion, Philip D., Upchurch, Paul, Schwarz, Daniela & Wings, Oliver, 2019, Taxonomic affinities of the putative titanosaurs from the Late Jurassic Tendaguru Formation of Tanzania: phylogenetic and biogeographic implications for eusauropod dinosaur evolution, Zoological Journal of the Linnean Society 185, pp. 784-909 : 817-822

CF. TENDAGURIA

Material: MB.R.2091.31 (G45) – middle–posterior cervical vertebra from Quarry G, located approximately 600 m south of Tendaguru Hill, Lindi District, south-eastern Tanzania ( Fig. 1); Upper Dinosaur Member (Upper Saurian Bed), Tendaguru Formation; Tithonian, Late Jurassic ( Janensch, 1929a; Bonaparte et al., 2000; Aberhan et al., 2002; Remes, 2007; Bussert et al., 2009).

DESCRIPTION AND COMPARISONS OF MB.R.2091.31

Bonaparte et al. (2000) tentatively referred a cervical vertebra (MB.R.2091.31) to Tendaguria tanzaniensis , primarily on the basis of its low, undivided neural spine. MB.R.2091.31 is a vertebra from the middle to posterior region of the cervical series (see Table 7 for measurements). Overall it is fairly complete ( Fig. 22), although it is missing some material from the right side of the neural arch, and has undergone some further damage in places since its description by Bonaparte et al. (2000) (see also: Janensch, 1929a: fig. 9).

The centrum is anteroposteriorly short, with an average Elongation Index [aEI = centrum length (excluding condyle) divided by the mean average value of the posterior mediolateral width and dorsoventral height] of 1.5. The ventral surface of the opisthocoelous centrum is extremely thin dorsoventrally. This surface is strongly concave mediolaterally along its anterior third (excluding the condyle), with this concavity decreasing in prominence posteriorly, such that the posteriormost portion of the ventral surface is flat. A sharp, anteroposteriorly oriented midline ridge runs along the deepest portion of this ventral concavity. The possession of a midline keel or ridge along the ventral surface of cervical centra is the plesiomorphic sauropod condition ( Upchurch, 1998), but this is lost in many derived eusauropods ( Poropat et al., 2016), with the notable exception of dicraeosaurids and rebbachisaurids ( Whitlock, 2011a).

Although both parapophyses are incomplete, it is clear that their dorsal surfaces are unexcavated, as is also the case in non-neosauropods and many derived somphospondylans ( Upchurch, 1998; Mannion et al., 2013). The parapophyses are restricted to the anterior third of the non-condylar centrum and project mainly laterally, contrasting with the strongly ventrolaterally projecting middle–posterior cervical parapophyses of several somphospondylans [the mid-Cretaceous East Asian taxa Daxiatitan , Erketu and Euhelopus ( Wilson & Upchurch, 2009; D’Emic, 2012), and the Late Cretaceous Argentinean titanosaur Overosaurus ( Coria et al., 2013) ], as well as the diplodocoids Apatosaurus and Nigersaurus (Mannion et al., 2013) . The lateral surface of the centrum is heavily and deeply excavated, with the lateral pneumatic foramen divided into numerous separate chambers by subvertical laminae. This lateral foramen occupies most of the centrum length. Either a breakage or a possible further excavation within the foramen reveals additional pneumatic space further within the centrum.

The diapophysis is supported by a prominent PCDL that projects anterolaterally, as well as very slightly dorsally. There is no evidence for an anterior centrodiapophyseal lamina (ACDL), but its absence might be a preservational artefact. At the anterolateral corner of the dorsal surface of the centrum, a stout ridge bifurcates into a prominent CPRL medially and a second lamina laterally that runs mainly vertically, joining the medial-most portion of the PRDL. As this lamina does not meet the prezygapophysis, it is not an instance of a true bifid CPRL, a feature that appears to be restricted to diplodocoid cervical vertebrae ( Whitlock, 2011a, b). A deep subtriangular fossa is formed in between these two laminae. The PRDL is a dorsoventrally thin, plate-like structure that projects posterolaterally and slightly ventrally. The dorsal surface of the PRDL is excavated by a shallow, though extensive fossa. The prezygapophysis extends only a very short distance beyond the anterior condyle of the centrum. Its dorsomedially facing articular surface is mildly convex mediolaterally, whereas the articular surface of the postzygapophysis is mildly concave and faces ventrolaterally. The dorsal surface of the postzygapophysis is rugose and forms an epipophysis; the latter does not extend beyond the posterior margin of the postzygapophysis. There is no ‘pre-epipophysis’ or epipophyseal–prezygapophyseal lamina (EPRL). A prominent, dorsoventrally thin, posteroventrally oriented PODL is present.

Large portions of the outer surface of the neural arch have been broken away, presumably because the arch is highly pneumatic (see below). In dorsal view, the neural spine has a trapezoidal outline, with mildly concave anterior and posterior margins, and a mediolaterally oriented long-axis. In lateral view, the anterior margin of the neural spine slopes at an angle of approximately 60°, such that it faces anterodorsally, whereas the posterior margin is mainly vertically oriented. The anterior surface of the neural spine is rugose. There is evidence for a SPRL, a deep postspinal fossa and a SPOL, although the latter is coated in plaster. The unbifurcated neural spine is extremely low, projecting only a short distance above the level of the postzygapophyses. Although low middle–posterior cervical neural spines are known in other sauropod taxa, e.g. Australodocus ( Remes, 2007) and Euhelopus ( Wilson & Upchurch, 2009) , these tend to be bifid structures; as such, MB.R.2091.31 shares this potentially unusual combination of a low, single neural spine with the holotypic anterior dorsal vertebrae of Tendaguria (as well as Moabosaurus ).

INTERNAL TISSUE STRUCTURE OF MB.R.2091.31

The vertebra is pneumatized by several large and small subcircular camerae ( Wedel, 2003) that are regularly distributed throughout the bone ( Fig. 23A–P). Pneumatic cavities are closely packed and internally separated only by thin walls of bone, with the external bone walls slightly thicker. Distinctly smaller camerae occur only in a few places, in particular in the condyle and lateral to the neural canal, whereas the predominant pneumatic camerae of the vertebral body are considerably larger.

The centrum bears no distinct midline strut of bone, but only a thin median wall that is replaced posteriorly by a dorsoventrally elongate camera with very thin bone walls ( Fig. 23K). The condyle is hollowed out by large, rounded camerae that are separated by thin bone walls, and are surrounded by smaller camerae ( Fig. 23B, C). Ventrally, these smaller camerae hollow out the parapophyses ( Fig. 23D), and posteriorly they increase in size. The pneumatic camerae are connected to the external surface and extend into the lateral pneumatic fossa of the centrum and into the CDF. At the diapophyseal level, the vertebral body becomes transversely narrow, consisting only of a few pneumatic camerae that are separated by very thin bone walls ( Fig. 23E–H). The walls of the neural canal apparently have collapsed in places, so that separation to the surrounding camerae is incomplete. Generally, pneumatic camerae of smaller and larger size throughout the vertebra, surround and are connected to the neural canal ( Fig. 23M).

The prezygapophyses are massive around their articular surface, and become hollowed out at the same level at which pneumatic camera start to develop within the condyle ( Fig. 4b). The pneumatic camera of each prezygapophysis opens laterally into the PRCDF and posteriorly excavates the SPRL ( Fig. 23C). The postzygapophysis is pneumatized by large camerae at its base, but its posterior half (including the articular surface) is apneumatic ( Fig. 23K, L).The lateral neural spine is hollowed out by a right and a left pneumatic camera, which close just anterior to the postspinal fossa ( Fig. 23G).

Only rounded camerae are present, which are regularly distributed throughout most of the vertebra, and are of varying dimensions. This polycamerate pattern of pneumatization strongly resembles that observed in the diplodocid Apatosaurus ( Wedel, 2003) , but differs from other diplodocids ( Wedel, 2003; Schwarz et al., 2007). The absence of camellae distinguishes MB.R.2091.31 from titanosauriforms, such as Giraffatitan ( Schwarz & Fritsch, 2006) , Sauroposeidon ( Wedel et al., 2000; Wedel, 2003), and derived titanosaurs (e.g. Powell, 1992; Wedel, 2003; Woodward & Lehman, 2009; Cerda et al., 2012).

SAUROPODA MARSH, 1878

EUSAUROPODA UPCHURCH, 1995

MAMENCHISAURIDAE YOUNG & ZHAO, 1972