Testing species' deviation from allometric predictions using the phylogenetic regression

Phylogenetic generalized least squares (PGLS) has become one of the most commonly used phylogenetic comparative methods. Despite its common use, descriptions, and applications of methods to test for species' deviations from allometric predictions using phylogenetic regression have been piecemeal. We simplify previous computational descriptions of PGLS standard errors in a manner that can be easily generalized toward more complex general linear models. We focus on the implementation of phylogenetic analysis of covariance, which provides a direct test for the equality of intercepts and slopes. Our computational descriptions allow testing whether individual species, or a group of species, deviate significantly from allometric predictions. The use of PGLS confidence and prediction intervals and phylogenetic analysis of covariance is exemplified in an analysis of brain structure volumes in primates.