Carl, Peter, and I’s paper on the power of comparative methods just appeared in Evolution. The paper is ArXived here with code here, Carl has slides and a blog post (here and here). As you can see from the number of links Carl is a big proponent of open science, in fact he kept an open lab notes throughout this work.

Comparative methods remain one of the most useful ways for us to understand the adaptive and ecological significance of particular phenotypes. Applications of this methods rely on phylogenies in order to distinguish between convergence (giving us natural experimental replication) and common descent. These methods also given us many useful insights into the tempo and mode of phenotypic change.

To understand rates of phenotypic evolution on phylogenies and provide the tools for the comparative method a range of increasing complicated models of trait evolution have been developed incorporating varying rates of evolution etc. However, the amount of information present in these data is limited as the traits at the tips of a tree are potentially highly correlated due to the underlying phylogeny. This means that we should be careful to understand the limitations of these data to distinguish between complex models, and also acknowledge that many of the assumptions that we often use in model fitting and hypothesis choice are not valid (e.g. our asymptotic assumptions). In this paper we produce a general framework for exploring these issues through simulations. A lot of this work was driven forward by Carl’s desire to really understand the validity of the assumptions underlying the application of these methods, so we hope that this will get a lot of use in the future in guiding the design and analysis of these types of data.