Merging theory and experiments to predict and understand coextinctions

In an era of mass extinction, predicting the consequences of species loss has become a priority for ecologists. Extinction of one species can trigger the loss of dependent species, sometimes leading to cascades of extinctions. Simulations predict that cascading extinctions should be commonplace, but empirical observations of extinction cascades rarely match those predicted by simulation. By contrast, species-removal field experiments have yielded surprises, such as novel interactions following removals. Thus, given this mismatch, the true predictive value of extinction simulation studies is unknown. We explore the value of validating extinction simulations with observational and experimental studies. We propose a new framework that unites both approaches to studying extinction cascades, and which reveals new opportunities to couple theory and data.