Phylogenetic Analyses of Adaptation and Constraint in Lacertid Lizards

Increased phylogenetic information (often based on analyses of DNA sequences) has revolutionized comparative biology. Phylogenies are used to guide the choice of species for comparisons and to identify traits that are homologous versus analogous. They are also used to perform “phylogenetically correct” statistical analyses that attempt to account for the fact that related species tend to resemble each other (exhibit “phylogenetic signal”). The first fully phylogenetic method for continuous-valued traits was “independent contrasts” (Felsenstein, 1985), based on a simple Brownian motion model of character evolution. This algorithm is now viewed as a special case of (phylogenetic) generalized least-squares approaches (Grafen, 1989; Martins and Hansen, 1997; Garland and Ives, 2000; Rohlf, 2001). Recent methods to test for and quantify phylogenetic signal show that it is ubiquitous, but stronger in some types of traits (e.g., body size, other morphometric traits) than in others (especially behavior). All of the foregoing approaches may employ Monte Carlo simulations along specified phylogenies (Martins and Garland, 1991; Garland et al., 1993) or randomization tests for hypothesis testing. Current work aims to develop more realistic models of character evolution, sometimes by transformation of branch lengths, to avoid over-correction for possible phylogenetic effects, and to incorporate within-species variation (e.g., measurement error) into phylogenetic comparative analyses. (Methods for categorical traits [e.g., see Ridley, Maddison, Grafen, Pagel] will not be covered.)