DUALCOR: a phylogenetic comparative method to evaluate phylogenetic correlation between a character and a non-evolving external variable

Abstract

A new phylogenetic comparative method called DUALCOR is presented to evaluate the evolutionary response of a character to non-evolving external factors, such as environmental variables. The method treats the character as a typical evolving feature of an organism that is reconstructed on a given tree, whereas the external factor is treated as unrelated to the phylogeny. DUALCOR first calculates the correlation/regression between the observed values of the character and the external factor; then it maps the character onto the phylogeny, shuffles the changes among branches, and re-evolves the character to yield new terminal values uncorrelated with the observed values of the external factor, allowing users to examine whether the observed degree of correlation can be attained at random. This is repeated n (say, 999) times, thereby using the dual nature of characters to construct a permutation test that is shown to satisfy requirements of Generalized Monte Carlo procedures. In addition, we provide an empirical example with a reverse test where the external variable (features determined largely by non-heritable factors) is the dependent variable.