Overview

The comparative method is one of the oldest and most widely used approaches to studying evolution. The rationale is that a group of species contains more variation than can be created in an experiment or using observations on a single species, and comparisons across species can be used to test broad questions in evolutionary theory. One of the key issues in comparative analysis is the problem of phylogeny. Phylogeny can create problems through generating non-independence of data, which compromises statistical tests, but it also generates opportunity by allowing the evolution of traits to be mapped. The modern comparative method is based on modelling the evolutionary process, using models of trait evolution to generate statistical models that can fitted to trait data.

We review a range of the current models and techniques used in comparative analysis. We begin by looking at techniques for modelling continuous traits, concentrating on methods for measuring variation in the rate of evolution through time, speciational modes of evolution, constraints on traits and variable levels of phylogenetic dependence. We then look at issues of uncertainty in data and how this may be incorporated, including uncertainty resulting from phylogenetic error, measurement error and other forms of non-independence. Developments on the analysis of discrete traits are described, including the use of modern Bayesian model averaging and selection methods. Finally we describe how the links between macroevolution (speciation and extinction) and trait evolution can be uncovered.