METHODS FOR THE ANALYSIS OF COMPARATIVE DATA IN EVOLUTIONARY BIOLOGY.

Abstract

Inferences regarding phylogenetic patterns and constraints on the evolution of characters often can be derived only from comparisons of extant species. If the phylogeny of these species is known, then the mean phenotypes of taxa can be partitioned into heritable phylogenetic effects and nonheritable residual components. Methods are presented for the estimation of phylogenywide means of characters, the variance-covariance structure of the components of taxon-specific means, and the mean phenotypes of ancestral taxa. These methods, which are largely an extension of maximum-likelihood techniques used in quantitative genetics, make an efficient use of the data, are unbiased by phylogenetically uninformative contributions to mean phenotypes, and take into account fully the nonindependence of data resulting from evolutionary relationships. Statistical tests are introduced for evaluating the significance of phylogenetic heritability and of correlations between traits, and expressions are given for the standard errors of ancestral mean phenotype estimates. It is argued that the covariance structure of phylogenetic effects provides a description of a macroevolutionary pattern, whereas that for the residual effects, when corrected for sampling error, is more closely related to a microevolutionary pattern.