Sizing up phylogenetic testing in geometric morphometrics: a case study of allometry

Accounting for phylogenetic relatedness in the analysis of shape is generally deemed necessary to acknowledge the non-independence between species due to common ancestry. However, some response variation can be both attributable to a predictor variable and phylogenetically structured, an effect termed conservative trait correlation (CTC). For CTC s, the effects of predictors versus phylogenetic structure cannot easily be disentangled. This is problematic because standard phylogenetic tests wholly attribute CTC to phylogenetic structuring when adjusting error distributions for relatedness. By contrast, non-phylogenetic methods retain CTC as wholly predictor-related, neglecting to estimate the effects of phylogenetic structuring altogether. Characterising how CTC contributes to a shape response is therefore crucial to support the interpretation of shape analyses. We here demonstrate a strong CTC effect in the geometric morphometric shape variation among 370 crania from the speciose Australian rock-wallabies (genus Petrogale). Specifically, significant allometry supporting mechanistically well-explained allometric patterns such as scaling of the braincase (Haller’s rule) and snout length (craniofacial evolutionary allometry) is reported by ordinary least squares (OLS) regression, but not phylogenetic generalised least squares (PGLS) analysis. We then apply two methods capable of characterising the missing variation: variation partitioning (VARPART), which estimates the proportion of variation shared between the predictor and phylogeny, and multi-response phylogenetic mixed models (MR-PMM), which identify the strength of correlation within the phylogenetic component of trait variance. Both methods show that CTC dominates the allometric shape variation in our sample. Our results therefore highlight CTC estimation as an important additional tool to clarify the relationship between morphological evolution and its predictors.