Body size correlates with discrete-character morphological proxies

Principal coordinates analysis (PCoA) is a statistical ordination technique commonly applied to morphology-based cladistic matrices to study macroevolutionary patterns, morphospace occupation, and disparity. However, PCoA-based morphospaces are dissociated from the original data; therefore, whether such morphospaces accurately reflect body-plan disparity or extrinsic factors, such as body size, remains uncertain. We collated nine character–taxon matrices of dinosaurs together with body-mass estimates for all taxa and tested for relationships between body size and both the principal axis of variation (i.e., PCo1) and the entire set of PCo scores. The possible effects of body size on macroevolutionary hypotheses derived from ordinated matrices were tested by reevaluating evidence for the accelerated accumulation of avian-type traits indicated by a strong directional shift in PCo1 scores in hypothetical ancestors of modern birds. Body mass significantly accounted for, on average, approximately 50% and 16% of the phylogenetically corrected variance in PCo1 and all PCo scores, respectively. Along the avian stem lineage, approximately 30% of the morphological variation is attributed to the reconstructed body masses of each ancestor. When the effects of body size are adjusted, the period of accelerated trait accumulation is replaced by a more gradual, additive process. Our results indicate that even at low proportions of variance, body size can noticeably affect macroevolutionary hypotheses generated from ordinated morphospaces. Future studies should thoroughly explore the nature of their character data in association with PCoA-based morphospaces and use a residual/covariate approach to account for potential correlations with body size.