A method for testing the assumption of phylogenetic independence in comparative data

Abstract

When making a comparison between two or more traits across taxa, the assumption of phylogenetic independence must be tested. Empirically determining the validity of this assumption before and after applying a phylogenetically based comparative method (PCM) can provide researchers with a rigorous methodological approach for analysing comparative data. This approach can help resolve current debates regarding whether it is always appropriate to apply a PCM, and whether PCMs are in fact successful in accounting for all of the historical non-independence in comparative data. To use this methodological approach, however, a generally applicable statistical diagnostic that can test the assumption of phylogenetic independence in comparative data is required. In this context, I present the application of a statistical diagnostic called the ‘test for serial independence’. This diagnostic can be applied to almost all currently employed PCMs, including Felsenstein’s (1985) independent contrasts. Furthermore, I demonstrate the application of the test for serial independence by analysing three data sets from the literature. The results of these analyses show that this diagnostic can successfully detect different degrees of phylogenetic autocorrelation in small and large phylogenies as well as in different types of phenotypic characters. The challenges and difficulties associated with applying the proposed methodological approach and the test for serial independence are discussed.