A62 A major likelihood-based approach gives problematic estimates of diversification dynamics and rates

Abstract The diversity of life is shaped by rates of speciation and extinction, and so estimating these rates correctly is crucial for understanding diversity patterns among clades, regions, and habitats. In 2011, Morlon and collaborators developed a promising likelihood-based approach to estimate speciation and extinction and to infer the model describing how these rates change over time based on AICc. This approach is now implemented in an R package (RPANDA). Here, we test the accuracy of this approach under simulated conditions, to evaluate its ability to correctly estimate rates of speciation, extinction, and diversification (speciation—extinction) and to choose the correct underlying model of diversification (e.g. constant or changing rates of speciation and extinction over time). We found that this likelihood-based approach frequently picked the incorrect model. For example, with changing speciation rates over time, the correct model was chosen in only ∼10 per cent of replicates. There were significant relationships between true and estimated speciation rates using this approach, but relationships were weak when speciation rates were constant within clades. Relationships were consistently weak between true and estimated rates of extinction and of diversification. Overall, we suggest that results from this approach should be interpreted with considerable caution.