The impact of tip age distribution on reconstructing trait evolution using phylogenetic comparative methods

Collecting data for use in constructing phylogenies is a valuable but time- and resource-consuming pursuit. As a result, indicators of the potential value of including certain species in a phylogeny a priori could prove useful when planning this stage of research. Here, we used a simulation approach to investigate whether there are trends in the ability for phylogenetic comparative methods to recover the correct model of trait evolution based on certain characteristics of the phylogeny. First, we used multiple diversification rates to simulate phylogenies containing varying proportions of fossil and extant tips. We then simulated the evolution of a single trait across each phylogeny using multiple continuous trait evolution models. We then compared the fit of the correct and incorrect models to the simulated traits. This quantitative evaluation allows us to discern whether there are certain tip characteristics associated with identifying the correct trait evolution models. Our results indicate that the inclusion of fossils can be highly beneficial to reconstructing certain trait histories (e.g., Ornstein-Uhlenbeck and ACDC) but not to others (e.g., Brownian motion). In fact, in many cases, increasing the proportion of fossils in a phylogenetic dataset is far more beneficial, and perhaps more time- and resource-efficient, than increasing the number of extant taxa in the dataset. Our results corroborate previous findings that the inclusion of fossil tips can vastly improve the reconstruction of trait histories, but also show that this effect is often stronger for older fossils.