Computed tomography reveals multiple origins of extreme caudal vertebral pneumaticity in sauropod dinosaurs.

Postcranial skeletal pneumaticity (PSP) is common in the presacral vertebrae of sauropod dinosaurs, but seemingly rare in their caudal vertebrae. Where identified, evidence for caudal vertebral PSP in sauropods is primarily based on the presence of external features, such as lateral fossae. However, such fossae can only be regarded as unequivocally pneumatic if communication between them and internal pneumatic bone texture can be confirmed. Based upon evidence from internal imaging, caudal vertebral PSP was previously known only in a rebbachisaurid diplodocoid (anterior caudal centrum and neural arch), some somphospondylan titanosauriforms (anterior caudal neural arches only) and saltasaurine titanosaurs (anterior-posterior caudal centra and neural arches). Here, we present novel CT scan data of caudal vertebrae of six Middle-Late Jurassic sauropods, representing several eusauropod lineages. We synthesise these new data with a comprehensive critical appraisal of purported external and internal evidence for caudal vertebral PSP in Sauropodomorpha. Newly sampled specimens of the non-neosauropod eusauropods Cetiosaurus sp. (anterior caudal centrum), 'Cetiosaurus glymptonensis' (middle caudal centrum) and the mamenchisaurid Wamweracaudia keranjei (anterior-middle caudal vertebrae), as well as the dicraeosaurine diplodocoid Dicraeosaurus sattleri (anterior caudal vertebra), are apneumatic. By contrast, the anterior-posterior caudal centra and neural arches of the diplodocine diplodocoid Tornieria africana possess deeply invasive external fossae that communicate with internal pneumatic chambers. Shallow external fossae on the centra and neural arches of at least the anteriormost 24 caudal vertebrae of the brachiosaurid titanosauriform Giraffatitan brancai communicate with internal pneumatic chambers. We observe a repeated pattern of PSP invading the anterior caudal vertebrae, with at least five independent acquisitions and/or reversals within Neosauropoda. Furthermore, for the first time, we demonstrate that extreme caudal vertebral PSP, in which pneumaticity extends into the middle-posterior region of the tail, is not restricted to saltasaurines, with this having also evolved independently in diplodocines and brachiosaurids. Finally, we find that both small- and large-bodied sauropods, including those with relatively short and long tails, evolved (and lost) caudal vertebral PSP. Therefore, the development of caudal vertebral PSP in sauropods does not appear to correspond with changes in body shape or mass. Instead, it might result from the opportunistic nature of pneumatic diverticula. However, given the high degree of inter- and intraspecific plasticity in its phylogenetic and serial distribution, we recognise that the evolution of PSP into the tail of sauropods might have been driven by a set of as-yet unknown, complex selective pressures.