Detecting Trait-Dependent Diversification Under Diversification Slowdowns

Testing whether a certain biological trait significantly affects clade diversification is central to macroevolutionary research. To this end, many scientists use constant-rate estimators (CR estimators) of diversification. However, it has never been examined whether these estimators report meaningful relationships between traits and diversification even when the diversification itself decelerates over time. In this study, I simulate trait-driven diversification concurrently with diversification slowdowns. Then, I test whether CR estimators manage to uncover the simulated relationships. Results suggest that CR estimators are robust against violation of rate constancy and successfully detect trait-dependent diversification in spite of diversification declines. Interestingly, correct results were recovered whether clade age correlated with clade diversity or not. Further comparison of CR estimators with QuaSSE suggested that QuaSSE performs better under constant diversification, but tends to report spuriously significant outcomes when diversification decelerates (=elevated Type I error). Given that diversification slowdowns have been recently reported for a wide range of taxa, these findings may be of particular relevance for future diversification studies.