The disadvantage of derivation: conserved systematic flaws in primary data have repeatedly biased the phylogenetic inference of Temnospondyli (Tetrapoda, Amphibia)

Abstract

ABSTRACTPhylogenetic analyses and their resultant tree topologies underlie paleobiological studies. Regardless of the type of study, the relationships of focal taxa are foundational, whether implemented in a qualitative or a quantitative framework. This reliance places a premium on the continued refinement of both phylogenetic methods and inference. Temnospondyls are a diverse clade of non-amniote (‘amphibian’) tetrapods whose phylogenetic relationships have been extensively explored due to their speciose nature, widespread occurrence in Paleozoic and Mesozoic paleoenvironments, and putative relationship to extant amphibians. Despite being studied by a diversity of workers, there is only one dataset that is widely employed to test the broad-scale relationships of Temnospondyli, that of Schoch (2013). This dataset has been reused in several high-profile studies testing the question of lissamphibian origins, and the original resultant topology has been widely adopted by taxonomic specialists and non-specialists alike. However, close examination of this matrix reveals discernible patterns of problematic codes related to non-homology, dependency, and unsubstantiated data (e.g., codes for postcranial characters for taxa with no known postcrania). These patterns, in conjunction with their prevalence, warrant a thorough survey of the entire matrix and subsequent reanalysis of its various forms to test whether previously published findings regarding the relationships of temnospondyls and the origins of lissamphibians are substantiated. A thorough reassessment of this matrix and several of its high-profile derivates revealed that the phylogeny of temnospondyls is more poorly known than depicted by the literature and that certain hypotheses of lissamphibian origins within Temnospondyli lack phylogenetic support.