Abstract
Understanding the patterns underlying phenotypic diversification across the tree of life has long been a fundamental aim in evolutionary and comparative biology. Classic and recent work has demonstrated both the wide variability in evolutionary rate throughout time and across lineages and the importance of characterizing these patterns in explaining the evolutionary proceses that generate biological diversity. A less extensive literature has shown that this variability extends to different aspects of phenotype, with separate suites, or modules, of traits within organisms showing different, “mosaic” patterns in rate and disparity across species. A merging of these two perspectives would identify modules of traits that display similar mosaic patterns in evolutionary tempo and mode. However, tools to do so have been limited. In this study, I introduce a new method for the identification of suites, or modules, of continuous traits that display shared patterns in evolutionary disparity across lineages. The approach defines a separate model of evolutionary disparification for each module defined by a phylogeny with branch lengths proportional to disparity. Module memberships and the number of modules are inferred using a greedy hill climbing approach that combines several different strategies to the unsupervised learning of classification and mixture models.