Shedding light on the ‘dark side’ of phylogenetic comparative methods

Abstract

Phylogenetic comparative methods are becoming increasingly popular for investigating evolutionary patterns and processes. However, these methods are not infallible – they suffer from biases and make assumptions like all other statistical methods. Unfortunately, although these limitations are generally well known in the phylogenetic comparative methods community, they are often inadequately assessed in empirical studies leading to misinterpreted results and poor model fits. Here, we explore reasons for the communication gap dividing those developing new methods and those using them. We suggest that some important pieces of information are missing from the literature and that others are difficult to extract from long, technical papers. We also highlight problems with users jumping straight into software implementations of methods (e.g. in r) that may lack documentation on biases and assumptions that are mentioned in the original papers. To help solve these problems, we make a number of suggestions including providing blog posts or videos to explain new methods in less technical terms, encouraging reproducibility and code sharing, making wiki-style pages summarising the literature on popular methods, more careful consideration and testing of whether a method is appropriate for a given question/data set, increased collaboration, and a shift from publishing purely novel methods to publishing improvements to existing methods and ways of detecting biases or testing model fit. Many of these points are applicable across methods in ecology and evolution, not just phylogenetic comparative methods.