Inference of secondary extinctions in marine communities using food-web modelling across the Permian–Triassic mass extinction

The Permian–Triassic mass extinction (PTME) is characterized by the highest extinction rates in marine communities in the entire Phaneozoic. As the event coincides with extreme environmental perturbations such as rapid global warming, most palaeobiological studies assume all extinctions during the PTME occurred as primary loss of taxa directly caused by physical environmental stress. However, secondary extinction cascades triggered by changes in ecological interactions can be as, if not more, important drivers of species loss. Herein, we modelled food webs of marine communities across the PTME using the most comprehensive occurrence dataset from the western Tethyan tropical deposits at Meishan, China, at a very high stratigraphical resolution to infer secondary extinctions. Using robustly inferred predator-prey networks for each time interval, we classified extinctions as secondary under the following conditions: i) a predator became extinct when prey diversity declined, ii) a primary consumer (e.g., filter feeders) became extinct following a decrease in primary productivity, iii) a species within the trophic network became extinct following either increased predation pressure (increased normalized out-degree) or a loss of resources (declined normalized in-degree). Based on these classifications, the majority of the extinctions can be attributed to secondary extinction due to perturbations in primary productivity or changes in trophic interactions. Modelling trophic interactions in these extinct communities reveals the importance of considering evidence for secondary extinctions to better understand the processes giving rise to biodiversity patterns in deep time.