Reassessing the affinities of vertebral remains from Permo-Triassic beds of Gondwana

The Sanga do Cabral Formation of southern Brazil has a rich fossil tetrapod assemblage and is suggested to have an Early Triassic age mainly based on the presence of the parareptile Procolophon trigoniceps. However, a Permo-Triassic age can be also suggested for this unit taking into account the presence of putative Permian taxa and some previous stratigraphic assessments. We describe here several large vertebrae from the Sanga do Cabral Formation that display a distinctive morphology that includes the presence of a transverse distance across postzygapophyses more than twice the transverse width of the centrum, and accessory articulation structures in the neural arch that remind the hyposphene and hypantrum present in some basal parareptiles and diadectomorphs. Vertebrae with a similar large size and morphology had been previously reported from the same locality as belonging to the genus Procolophon based on their parareptile appearance (mainly the presence of a swollen neural arch) and the fact that the vertebrae were collected at the same locality where a large fragmentary skull assigned to this taxon was found. However, these vertebrae lack a comparable consistent morphology with those of Procolophon and basic statistical analyses demonstrate that these vertebrae are significantly larger than those expected in the largest known Procolophon skulls of South Africa. The morphology of these vertebra is consistent with that present in seymouriamorphs, pareiasaurs and diadectomorphs, but the absence of exclusive diagnostic characters precludes an assure assignation to either of these taxa. According to their current stratigraphic range, seymouriamorphs are the most plausible postulation, as their younger representatives are known from Late Permian deposits of Russia, but the other candidates cannot be excluded. The presence of any representative of those groups in the Sanga do Cabral Formation would be important because: (1) they would represent the first and only known record of seymouriamorphs or diadectomorphs in Gondwana, regarding the characters that the described vertebrae share with these groups; (2) they would suggest a survivorship for pareisaurs up to the Latest Permian or through the Permian-Triassic boundary, according to geochronological data currently available for this unit; (3) they might also suggest a Late Permian age for at least part of the Sanga do Cabral Formation if the intraformational conglomerates that yielded the vertebrae resulted from the rejuvenation of older levels of the same unit, and they do not include reworking of stratigraphically older strata.