CONSTRAINING THE LATE TRIASSIC MASS EXTINCTION USING THE FOSSILIZED BIRTH-DEATH RANGE MODEL

The Late Triassic extinction event led to a substantial loss of biodiversity but the underlying drivers of this event appear complex, and it has not been established whether this was a sudden or gradual event. We address this question using the fossilized birth-death (FBD) range model and global stratigraphic range data for 14 marine invertebrate classes. The FBD process is a phylogenetic model that incorporates the species diversification and fossil sampling processes explicitly. The model can be applied to the analysis of stratigraphic ranges and used to estimate speciation and extinction rates during different geological intervals. In this model, fossils are distributed over time according to a uniform Poisson sampling process, and rates may vary across intervals in a piecewise manner. We introduce an extension of the model that allows for the possibility that we may only know whether a given taxon was sampled or not during each geological interval, but not the frequency or at which precise time points the taxon was sampled (referred to as presence/absence sampling). We assess the performance of our modelling framework using simulations and demonstrate that although precision decreases given presence/absence data, we still obtain reliable estimates of diversification and sampling parameters. We apply the new model to the analysis of the Triassic-Jurassic extinction interval, for which presence/absence data only is available at the sub-stage level, and we re-examine the diversification dynamics across this important and turbulent episode in earth’s history. Together, our findings demonstrate the importance of explicitly accommodating the diversification and sampling processes in combination. Our new model should be of broad application in paleodiversity studies, given that many fossil datasets are comprised of presence/absence data only, making it challenging to analyze such datasets using existing phylogenetic models. Future development of phylogenetic models in palaeobiology should continue to consider seriously the nature of the incompleteness of fossil occurrence databases.