This article considers the statistical issues relevant to the comparative method in evolutionary biology. A generalized linear model (GLM) is presented for the analysis of comparative data, which can be used to address questions regarding the relationship between traits or between traits and environments, the rate of phenotypic evolution, the degree of phylogenetic effect, and the ancestral state of a character. Our approach thus emphasizes the similarity among evolutionary questions asked in comparative studies. We then discuss ways of specifying the sources of error involved in a comparative study (e.g., measurement error, error due to evolution along a phylogeny, error due to misspecification of a phylogeny) and show how the impact of these sources of error can be taken into account in a comparative analysis. In contrast to most existing phylogenetic comparative methods, our procedure offers substantial flexibility in the choice of microevolutionary assumptions underlying the statistical analysis, allowing researchers to choose assumptions that are most appropriate for their particular set of data and evolutionary question. In developing the approach, we also propose novel ways of incorporating within-species variation and/or measurement error into phylogenetic analyses, of estimating ancestral states, and of considering both continuous (quantitative) and categorical (qualitative or 'state') characters in the same analysis.
ASJC Scopus subject areas
- Ecology, Evolution, Behavior and Systematics