The uniqueness of the phylogenetic regression

Abstract

The phylogenetic regression provides the hypothesis testing facilities of general linear models for comparative data with incompletely known phylogenies. It applies Ridley's radiation principle. It takes a thorough-going regression approach and so applies a Brownian motion model not to all the variables, but just to the error in the regression model. The phylogenetic regression is effectively unique as a regression method for comparative data—any substantially different method must be statistically unsound. The phylogenetic regression has been justified by analytical work and by computer simulations. Methods lacking explicit justification should be treated with suspicion. The statistical way forward for comparative biology lies in well-defined, well-justified methods.