LAUSR

Morphological data sets are misleading in salamander (Caudata) phylogeny due to the relative homoplasy of the dermal skull observed in paedomorphic forms, leading to the trend of excluding morphology when exploring questions of salamander phylogeny. Investigations in caecilians (Gymnophiona) have demonstrated that the inclusion of braincase morphology can rescue morphological phylogenetic analyses and produce topologies congruent with molecular data sets. We scanned 28 species (25 genera) of salamander, representing all 10 families, with high‐resolution micro‐computed tomography to investigate braincase variation. We describe the morphology of the braincase for all 10 families and distinguish between paedomorphic and metamorphic morphologies. Our results demonstrate a general uniformity amongst metamorphic species with variation largely restricted to the occipito‐otic region. A greater range of variation is observed within paedomorphic forms than would be expected when considering the homoplasy of the dermal skull. Obligate paedomorphic forms demonstrate considerably more variation in the anterior braincase than do facultative paedomorphs, which we suggest is evidence of a greater complexity in the evolution and development of these forms than neoteny alone would produce. This raises the question of character independence within morphological data sets and warrants further investigation into the correlation of other characters before morphological data are omitted.