Why is vertebral pneumaticity in sauropod dinosaurs so variable?

The vertebrae of sauropod dinosaurs have distinctive and complex pneumatic features including fossae and foramina in the sides of their centra. These vary between individuals, serially within individuals, and even between the left and right sides of single vertebrae. This presents a conundrum because bone is usually the least variable material in the vertebrate body. Blood vessels, however, are much more labile, as can be seen in the varied occurrence of vascular foramina in the vertebrae not only of sauropods, but also of birds, crocodilians and mammals. Vascular variation arises in part from the ontogeny of vertebrae, which in their embryonic state are vascularised from within the neural canal: the hand-off from these vessels to others which penetrate from outside is not always completed. In birds, pneumatizing diverticula enter the vertebrae alongside blood vessels, in the vascular foramina that they form, before excavating the surrounding bone into larger pneumatic foramina. We propose that the same was true in sauropods, and that variation of vascularization directly caused variation of pneumatization. In most vertebrae, a single vascular foramen carries both artery and vein, but occasionally these vessels separate and each forms a separate foramen. This explains why in rare cases individual sauropod vertebrae have two pneumatic cavities on a single side: each arises from the vascular foramen formed by either artery or vein. Qeios, CC-BY 4.0 · Article, February 25, 2021 Qeios ID: 1G6J3Q.2 · https://doi.org/10.32388/1G6J3Q.2 1/11


Introduction
The vertebrae of sauropod dinosaurs are distinctive not only because of their size but also because they have complex pneumatic features. These include fossae and foramina, in both the centrum and neural arch; and laminae connecting landmarks such as the zygapophyses, diapophyses and parapophyses (Wedel 2003). For this reason, sauropod vertebrae are unusually diagnostic and are frequently used in species determination (McIntosh 1990).
Bone is generally the material that varies the least between individuals of a species, with muscle, nerves and especially blood vessels being more prone to variation (Berger 1956:435-439, Moore et al. 2010. However, while pneumatic features of sauropod vertebrae can be characteristic of a species, genus or clade, they are also highly variable: not only between individuals, but also along the column of an individual (e.g. Diplodocus carnegii, Hatcher 1901:plates 3 and 7), and even sometimes between the sides of a single vertebra. Examples of the latter include the single vertebra that is the Xenoposeidon proneneukos holotype: Taylor andNaish (2007:1552; Figure 1); and the sequence of vertebrae in the tail of Giraffatitan brancai MB.R.5000 (Wedel and Taylor 2013:5-7 and figures 4 and 5). In contrast, the vertebrae of mammals, non-dinosaurian reptiles and even other dinosaurs are much more uniform, exhibiting less individual, serial and bilateral variation. Why are sauropod vertebrae so much more variable?  Analysis It has been generally assumed that variation in pneumatic features is essentially random: as Witmer (1997:64) wrote of the antorbital paranasal sinus in archosaurs, "pneumatic diverticula are viewed simply as opportunistic pneumatizing machines, resorbing as much bone as possible within the constraints imposed by local biomechanical loading regimes".
However, here we will develop another explanation. Bremer (1940:200) demonstrated that in extant birds, developing diverticula follow blood vessels as they radiate through the body: "Into this loose tissue, along the vein, the air sac [i.e. diverticulum in modern usage] finally grows in the form of a long tube … The actual entrance of the air sac into the main marrow cavity is effected at first at the internal opening of the vein". O'Connor (2006O'Connor ( :1208O'Connor ( -1209 confirmed that "vascular injection studies on birds with pneumatic postcrania reveal that nutrient vessels share (i.e., co-occupy) foramina with pneumatic diverticula to gain access to the medullary space". It is parsimonious to assume the same was true in sauropods.
But vascularization of vertebrae is itself highly variable, and it is common for the pattern of vertebrae with and without external vascular foramina to be random. For example, in a juvenile specimen of the crocodilian Tomistoma, only about half of the first 13 caudal vertebrae have vascular foramina on each side, and they are not the same vertebrae on each side ( Figure 2).  they are absent or too small to make out in vertebrae 1-4, 8 and 12-13. In left lateral view, vascular foramina are apparent in the centra of caudal vertebrae 4-7 and 9; they are absent or too small to make out in vertebrae 1-3, 8, and 10-13. Caudal centra 5-7 and 9 are therefore vascularised from both sides; 4 and 10-11 from one side only; and 1-3, 8 and 12-13 not at all.
In sauropods, too, vascularization is variable along the vertebral sequence ( Figure 3) and between sides of an individual vertebra.  like most of the sequence, has a single vascular foramen on the right side of its centrum, but caudal 8 has two; others, including caudal 1, have none.
Why is vascularization so variable?
The ontogenetic development of vertebral vasculature is a complex process. Early in embryonic development, the spinal cord is much larger than the vertebrae. Arteries serve the cord first; then as the notochord segments and is replaced by the cartilaginous anlagen of the vertebrae, branches of the medullary arteries tunnel into the cartilage and support the growth of the vertebrae (Amato 1959). All the blood supply to developing vertebrae therefore comes from inside the neural canal.
Branches of the segmental arteries subsequently penetrate the vertebrae from the outside, and anastomotic connections develop inside the vertebra, connecting the internal and external systems (see example in Cramer 2014:figure 2.4).
As the growth of the vertebrae outpaces that of the cord, there is a handoff of arterial supply from the original medullary arteries that serve the cord to the secondary, external arteries: foramina inside the neural canal shrink with age, while those on the external surface of the vertebra enlarge (Smuts 1975:35). However, this handoff is not always completed, and asymmetric arterial supply is common (Smuts 1975). Consequently centra frequently lack an external vascular foramen on one or both sides. This is not a problem as the medullary arteries can provide the necessary blood supply, but in sauropods the absence of such external foramina means there is no point of entry for a diverticulum that otherwise would subsequently produce pneumatic cavities in the side of the bone.
In summary, since external pneumatic cavities follow vascularization of the outer wall, and the latter is variable, it follows that pneumatization is also variable, reflecting the variation in the soft tissues that guide its development.
The The morphogenetic rules governing cervical and caudal pneumatization are less clear than in the torso. In the cervical column, the arteries that supply the spinal cord branch from the paired vertebral arteries, which lie alongside the vertebral centra, and this may explain the "centrum first" pattern of pneumatization in the cervical vertebrae of non-avian theropods (Benson et al. 2012). The caudal vertebrae of juvenile diplodocids are less pneumatic than those of adults (Melstrom et al. 2016, Hanik et al. 2017, and in Giraffatitan extensive caudal pneumaticity in present only in large individuals (Wedel and Taylor 2013). These observations suggest that caudal pneumatization in sauropods continued for several years, after the vascular handoff from neural canal arteries to arteries on the external surface of the centrum, possibly explaining the mix of "centrum first" and "arch first" pneumatization observed in sauropod caudal vertebrae.
It is also notable that paired pneumatic fossae or foramina occur lateral or dorsolateral to the neural canal in every archosaurian clade with postcranial pneumaticity (Figure 4). These fossae and foramina occur in taxa with and without lateral cavities in the centra, and with and without laminated neural arches, so they are probably the most consistent osteological correlates of pneumaticity across non-avian ornithodirans. The consistent appearance of vertebral pneumaticity in areas adjacent to the neural canal corroborates the hypothesis that segmental spinal arteries were crucial in "piloting" pneumatic diverticula as they developed. Hatcher 1901: plate 6).

Discussion
As noted by O'Connor (2006O'Connor ( :1208, "Whereas arteries and veins often utilize a single nutrient foramen within a given vertebra, occasionally there are separate foramina for each". Similarly, Travan et al. (2015) show that in the cervical vertebrae of humans the transverse foramen, which the vertebral artery and vein pass through, is sometimes double, with the two vessels each passing through its own opening rather than the usual shared opening. (In rare cases, a triple transverse foramen occurs, with the sympathetic nerve plexus passing through a third opening rather than sharing the opening used by one or both blood vessels.) A similar phenomenon can be observed in the tail of the Brontosaurus excelsus holotype YPM 1980, in which the right side of the centrum of caudal 7 has the usual single vascular foramen but that of caudal 8 has two (Figure 3). If our hypothesis that pneumatization follows vascularization is correct, then then this could explain why there are sometimes two pneumatic fossae on one side of a centrum, for example the left side of caudal 25 of the Giraffatitan brancai tail MB.R.5000 ( Figure 5): the two vascular foramina carrying artery and vein were each followed by a pneumatic diverticulum and each developed into a pneumatic fossa. Vascular foramina are rarely if ever seen in sauropod vertebrae that feature pneumatic fossae or foramina. Understandably they do not appear alongside these features, as the cavities were excavated around the blood vessels; but why do vascular foramina not appear within pneumatic cavities?
When a blood vessel enters a bone through a vascular foramen it is still detectable in CT scans as a tunnel through the trabeculae :figures 3 and 11), but the vessels usually arborize into arterioles and capillaries quickly once they're inside. So before a bone becomes pneumatized by a fossa, the artery has already branched into many small vessels. When the diverticulum subsequently enlarges the vascular foramen into a pneumatic fossa, pneumatization likely excavates the bone around the already-branched arterial tree that existed inside the bone. There should therefore be multiple vascular foramina inside the fossa, representing the multiple branches of the artery -as can be observed in a least some vertebrae of ducks ( Figure 6). However, these foramina will be much smaller than those that remain at the surface of apneumatic vertebrae. They may not be well preserved by fossilization, and even when preserved they will be difficult to spot during fossil preparationespecially as pneumatic cavities in large, delicate bones are already difficult to prepare. We recommend that sauropod workers carefully check pneumatic fossae and foramina for evidence of contained vascular foramina.

Conclusion
In general, bones are the least variable part of a body, followed by muscles, nerves, and finally blood vessels, which are very variable in all vertebrates. Pneumatic fossae and foramina are skeletal features, so they might be expected to fall at the least variable end of the spectrum. But since diverticula follow blood vessels as they develop, the variability of pneumatic features in bones is not a coincidence: the variability of blood vessels causes the variability of diverticula, and of their skeletal traces.