When studying extinct organisms, which phylogenetic methods are the most useful to determine patterns of evolutionary relationship? How well do current classifications reflect the patterns discovered? Using Athyridida (Upper Ordovician–Lower Jurassic) as a case study, we utilize parsimony, Bayesian Mk, asymmetrical rates, and fossilized birth–death process models, with and without character partitions, to compare results from different methods of inference, to test previous phylogenetic hypotheses and examine morphological character evolution in this long-lived group of extinct brachiopods. Because different phylogenetic methods utilize different models of evolution involving different sets of assumptions, they can result in different patterns of relationship, making it necessary to test multiple methods and then evaluate thoughtfully the various results obtained.

We discovered that the four main athyridide higher taxa we focus on largely maintain their coherence as clades in most of the analyses, but relationships among them vary substantially, with implications for the evolution of characters important in their classification. We were able to characterize in detail the athyridide external valve characters that are more variable than internal characters, quantifying the commonly held impression that internal features are more likely to be homologues and thus more reliable in identifying relationships than external characters. Because taxa in classifications are still frequently used as clade proxies in macroevolutionary studies, it is necessary to obtain and compare the most robust hypotheses of relationship among named taxa in order to evaluate both character homology and homoplasy and taxonomic fidelity to hypotheses of evolution.