Neural Spine Bifurcation in Sauropod Dinosaurs of the Morrison Formation: Ontogenetic and Phylogenetic Implications

It has recently been argued that neural spine bifurcation increases through ontogeny in several Morrison Formation sauropods, that recognition of ontogenetic transformation in this ‘key character’ will have sweeping implications for sauropod phylogeny, and that Suuwassea and Haplocanthosaurus in particular are likely to be juveniles of known diplodocids. However, we find that serial variation in sauropod vertebrae can mimic ontogenetic change and is therefore a powerful confounding factor, especially when dealing with isolated elements whose serial position cannot be determined. When serial position is taken into account, there is no evidence that neural spine bifurcation increased over ontogeny in Morrison Formation diplodocids. Through phylogenetic analysis we show that neural spine bifurcation is not a key character in sauropod phylogeny and that Suuwassea and Haplocanthosaurus are almost certainly not juveniles of known diplodocids. Skeletochronology based on the sequence of skeletal fusions during ontogeny can provide relative ontogenetic ages for some sauropods. Although such data are sparsely available to date and often inconsistent among sauropod genera they provide another line of evidence for testing hypotheses of ontogenetic synonymy. Data from skeletal fusions suggest that Suuwassea and Haplocanthosaurus are both valid taxa and that neither is an ontogenetic morph of a known diplodocid. Wedel & Taylor, Neural Spine Bifurcation in Sauropods PalArch’s Journal of Vertebrate Palaeontology, 10(1) (2013) © PalArch Foundation 2 Introduction Among tetrapods, sauropod dinosaurs are unusual in that many taxa have deeply bifid neural spines in their presacral vertebrae. Many mammals have shallowly bifid spines in their cervical vertebrae, but usually only the neurapophysis is divided, whereas in sauropods the division is more extensive. In the most extreme cases the midline cleft extends to the roof of the neural canal, completely dividing the neural spine into bilaterally paired metapophyses (figure 1). Bifid presacral neural spines evolved several times independently in sauropods, and are present in some mamenchisaurids, all known diplodocids and dicraeosaurids, the basal macronarian Camarasaurus, the basal somphospondyls Euhelopus, Erketu, and Qiaowanlong, and the derived titanosaur Opisthocoelicaudia (Wilson & Sereno, 1998; Ksepka & Norell, 2006; You & Li, 2009; figure 2). In addition, the tips of the proximal caudal neural spines are often weakly bifid in diplodocids (e.g. Diplodocus carnegii CM 84/94, Hatcher, 1901: plate 9). In contrast, non-pathological bifid neural spines are uncommon in extant tetrapods, and are limited to the cervical vertebrae in certain large-bodied, longnecked birds (Rhea, Tsuihiji, 2004: figure 2b; Casuarius, Schwarz et al., 2007: figure 5b; Dromaius, Osborn 1898: figure 1; Theristicus, Tambussi et al., 2012: 7; also in the recently extinct Dromornithidae, Gastornithidae, and Phorusracidae, Tambussi et al. 2012: 7), the thoracic vertebrae in some bovids (e.g. zebu Bos indicus, Mason & Maule, 1960: 20), and the lumbar vertebrae of sirenians (Kaiser, 1974). Cervical neural spines in humans and many other mammals have paired tubercles at their tips (Kapandji, 2008: 190191; Cartmill et al., 1987: figure 2-3a; figure 3). They are therefore sometimes described as being bifid (e.g. White & Folkens, 2000: 145). The appearance of bifurcation is caused by the outgrowth of bone at the spine tip to anchor the large transversospinalis muscles. This is a different phenomenon from the non-union of the endochondral portions of the vertebral spine, which occurs pathologically in humans (and presumably all other vertebrates) as spina bifida cystica and spina bifida occulta (Barnes, 1994: 46-50 and figures 3.5 and 3.6). The developmental underpinnings of bifid neural spines in sauropods are not well underFigure 1. A cervical vertebra of Apatosaurus ajax YPM 1860 showing complete bifurcation of the neural spine into paired metapophyses. In dorsal (top), anterior (left), left lateral (middle), and posterior (right) views. Wedel & Taylor, Neural Spine Bifurcation in Sauropods PalArch’s Journal of Vertebrate Palaeontology, 10(1) (2013) © PalArch Foundation 3 Figure 2. Consensus phylogeny of sauropods based on the strict consensus trees of Taylor (2009), Ksepka & Norell (2010) and Whitlock (2011). The first of these provides the skeleton of the tree including outgroups, basal sauropods and macronarians; the second gives the positions of Erketu and Qiaowanlong; the last provides a detailed phylogeny of Diplodocoidea. Taxa with bifid neural spines are highlighted in blue. Haplocanthosaurus and Suuwassea, whose positions are disputed by Woodruff & Fowler (2012) are shown in bold. Wedel & Taylor, Neural Spine Bifurcation in Sauropods PalArch’s Journal of Vertebrate Palaeontology, 10(1) (2013) © PalArch Foundation 4 stood. It is possible that in some vertebrae the paired embryonic neural arch elements never fused except to form a roof over the neural canal. In contrast, in the genus Camarasaurus it is possible that many of the presacral neural spines were not bifid in young animals, and that the degree of bifurcation increased over the course of ontogeny (see below). In a recently-published paper, Woodruff & Fowler (2012) argued that the degree of bifurcation of sauropod neural spines was ontogenetically controlled, with the simple, undivided spines of juveniles gradually separating into paired metapophyses over the course of posthatching ontogeny. Based on this inferred ontogenetic trajectory, Woodruff & Fowler (2012) further argued that currently recognized sauropod taxa are oversplit, and that when ontogenetic transformations were taken into account, it would be necessary to synonymize several taxa. In particular, they argued that the Morrison Formation diplodocoid Suuwassea was a juvenile of a known diplodocid (Ibidem: 6-8), that Haplocanthosaurus and Barosaurus were likewise suspect (Ibidem: 9), and that rebbachisaurids were possibly paedomorphic dicraeosaurids (Ibidem: 8-9). Our goals in this paper are, first, to re-examine the evidence for an ontogenetic increase in neural spine bifurcation in sauropods, and then to evaluate the synonymies proposed by Woodruff & Fowler (2012). Although bifid neural spines also occur in other sauropods, as noted above, the hypotheses of Woodruff & Fowler (2012) depend on ontogenetic inferences drawn from Morrison Formation sauropod taxa, and therefore we are confining our discussion to those taxa (e.g. Camarasaurus, Haplocanthosaurus, and the Morrison diplodocoids).