The phylogeny of varanoid lizards and the affinities of snakes

Abstract

Evidence that platynotan squamates (living varanoid lizards, snakes and their fossil relatives) are monophyletic is presented. Evolutionary relationships within this group are then ascertained through a cladistic analysis of 144 osteological characters. Mosasauroids (aigialosaurs and mosasaurs), a group of large marine lizards, arc identified as the nearest relatives of snakes, thus resolving the long-standing problem of snake affinities. The mosasauroid-snake clade (Pythonomorpha) is corroborated by 40 derived characters, including recumbent replacement teeth, thecodonty, four or fewer premaxillary teeth, supratemporal-prootic contact, free mandibular tips, crista circumfenestralis, straight vertical splenio-angular joint, loss of posterior ramus of the coronoid, reduced basipterygoid processes, reduced interpterygoid vacuity, zygosphene-zygantral articulations, and absence of epiphyses on the axial skeleton and skull. After mosasauroids, the next closest relatives of snakes are varanids (Varanus, Saniwa and Saniwides) and lanthanotids (Lanthanotus and Cherminotus). Derived features uniting varanids and lanthanotids include nine cervical vertebrae and three or fewer pairs of sternal ribs. The varanid-lanthanotid-pythonomorph clade, here termed Thecoglossa, is supported by features such as the of platynotan relationships should not cause problems regarding homoplasy and instability in the resultant tree. Snakes would have been expected to have slotted neatly into the final cladogram, with little homoplasy, and their position would have been robust and stable. This is indeed what has happened in this analysis. Snakes possess not just the (admittedly few) diagnostic characters of platynotans as a whole, but also a particular set of derived features that allow them to be placed in various nested subgroups of platynotans (e.g. varanoids, thecoglossans, pythonomorphs). The initial tentative decision to include snakes in the ingroup (Platynota) was strongly supported by the results of the subsequent cladistic analysis, which showed that snakes also possess the synapomorphics of various nested subgroups of platynotans and are thus embedded deeply within Platynota. This reasoning might sound dangerously circular, but it is not. If snakes were not platynotans, inclusion of them (mistakenly) into a cladistic analysis of platynotans would not have yielded such clear results. In this scenario, snakes would not be expected to have many of the synapomorphies uniting various subgroups of platynotans. Furthermore, any apparent synapomorphics that snakes did share with some platynotans would have arisen by chance (convergently), since snakes really lie outside platynotans. Thus, the combination of derived characters found in snakes would not conform to the nested pattern of synapomorphics identified within platynotans. It is unlikely, for instance, that snakes would convcrgcntly evolve all the synapomorphics of pythonomorphs, but none of the synapomorphies of the Varanus-Lanthanotus clade (the sister group of pythonomorphs). Rather, snakes would be expected to have randomly evolved a few but not all of the characters of both these groups. This would mean that they cannot be placed with confidence in either group. Thus, if snakes were not platynotans, forcing them into a cladistic analysis of platynotans would have resulted in their position within the group being poorly resolved and highly unstable. © 1997 The Royal Society..