Abstract

Sabertooth craniodental adaptations have evolved numerous times amongst carnivorous mammals. Some of the most extreme sabertooth adaptations are found within the carnivoran subfamily Barbourofelinae. However, the evolutionary origins of this group have been uncertain for more than 170 years, with variable placement as an independent case of sabertooth acquisition, as a clade within the Nimravidae (Eocene to Oligocene ‘false sabertooth cats’), or as a member of the Machairodontinae (true sabertooth cats such as Smilodon). Here we present a novel approach to assessing the validity of three independent sabertooth clades within Carnivora. We performed a total-evidence Bayesian analysis in Beast2 across all major carnivoran families, using the fossilized birth-death (FBD) model and incorporating 223 morphological characters, nuclear and mitochondrial gene sequences, and stratigraphic occurrence data. Our results place barbourofelines as terminal members of the Nimravidae, sister to the Nimravini (0.91 posterior probability), a relationship not found in prior cladistic studies. Ancestral area estimation performed in the R package BioGeoBEARS best supports a primarily European paleobiogeographic center for the barbourofelines with multiple dispersal events to other continents, a finding in direct opposition to past hypotheses for this group. Furthermore, new patterns in convergence between nimravids and machairodontines were revealed via Bayesian ancestral state estimation in BayesTraits. Results support a hypothesis of cats copying nimravids, and nimravids cats in certain aspects of sabertooth morphology, and not total evolutionary independence of these features as typically envisioned.