RELATIVE IMPACT OF CHARACTER CODING DIFFERENCES AND STRATIGRAPHIC INFORMATION ON THE SUPPORT FOR ALTERNATIVE EARLY DINOSAUR PHYLOGENIES

Since 2017, multiple studies have cast doubt on the traditional view of large-scale dinosaur phylogeny, according to which the long-necked, herbivorous sauropodomorphs and the ancestrally carnivorous theropods form a monophyletic group (Saurischia) to the exclusion of the “bird-hipped” ornithischians. Emerging alternatives to this traditional arrangement include a sister-group relationship between Ornithischia and Sauropodomorpha (Ornithischiformes); a topology in which the ornithischians are sister to, or even nested within, Theropoda (Ornithoscelida); and the inclusion of the putatively non-dinosaurian Silesauridae within Ornithischia. Some (though not all) of these hypotheses have received support from formal phylogenetic analyses, a fact that has been often attributed to character coding differences among the underlying datasets. However, following a common practice within the parsimony phylogenetic framework, such analyses usually each produced a single point estimate of dinosaurian phylogeny, without determining whether it explained observed character data significantly better than the competing hypotheses. Moreover, despite the fact that improved congruence between phylogeny and stratigraphy has been put forward as a key piece of evidence in favor of some of the recently proposed non-standard hypotheses, none of the analyses performed so far has allowed stratigraphic ages to inform phylogenetic inference. Here, I address both of these shortcomings by taking advantage of the tools recently made available in the RevBayes phylogenetic software package, and conduct extensive Bayes factor model comparisons among 20 alternative phylogenetic topologies in both time-free and tip-dated settings. By repeating the comparisons on three recently published datasets that have produced discordant parsimony point estimates, I also show how this approach can be extended to evaluate the relative contributions of character scoring changes and stratigraphic information to the preference for competing early dinosaur topologies. The results obtained here highlight the importance of explicit hypothesis testing in modern morphological phylogenetics, and of integrating character data and stratigraphic evidence in a single coherent statistical framework.