Conference on Applied Statistics in Agriculture

BOOTSTRAP ESTIMATION AND COMPARISON OF AN INDEX OF PHYLOGENETIC CORRELATION

A common objective of bioinformatic analyses is to assess the similarity of species, given a biological trait or characteristic. Phylogenetic correlation is one means to achieve this objective. Such measures provide a means to evaluate evolutionary models and history as well as having potential application to ecological relationships including host preference selection. Typically, these measurements are based on the deviation of an observed phylogeny from a Brownian evolutionary model. Statistical inference for this difference is assessed through likelihood ratio tests. These tests, in turn, rely on the assumption of a Normal likelihood within the phylogenetic trait. In addition, statistical comparison of estimated phylogenetic correlations between competing phylogenies or traits has not been addressed. In this paper, a bootstrap resampling methodology is proposed for two common phylogenetic correlation metrics, Pagel’s λ and Blomberg’s K. The underlying bootstrap distribution of the estimates will be utilized as a means of computing confidence limits as well as carrying out hypothesis testing. The method will be demonstrated using phylogenetic and metabolomic data related to the host specificity of an insect, Ceutorhynchus cardariae Korotyaev, on a wide range of Brassicaceae species.