Detecting Patterns of Trait Evolution

Abstract

This chapter examines a variety of methods for detecting the patterns of evolution of a certain trait. It first considers common metrics for evaluating phylogenetic signal and compares fit of alternative evolutionary models. When trait data deviates significantly from assumptions of Brownian motion (BM), the phylogenetic distances separating taxa on a time-calibrated tree might not accurately capture phenotypic distance between species. If we are able to identify the correct model of trait evolution, we can transform the branch lengths on the phylogenetic tree to match. Nevertheless, we might still favor using the untransformed tree because complex evolutionary models might not generalize across traits. The chapter also reviews alternative models of trait evolution, including a model of constrained evolution, multirate and multioptima models, speciational models, and white noise model. Finally, it looks at some of the common models for reconstructing ancestral states for discrete and continuous data.