Markov chain models predict the consequences of experimental extinctions

Predicting the effects of species extinction is challenging because of ecosystem complexity. Using novel long-term experimental deletions of two mussel species, I found that a multi-species Markov chain model, which was parameterized under non-experimental conditions, accurately predicted the effects of local extinction in an intertidal community. Systematically deleting species from the model predicted declining ecosystem biomass with reduced biodiversity, but relatively constant ecosystem resilience following disturbance. The effects of reduced biodiversity on these variables ranged widely, illustrating the need for species-specific predictions of extinction effects. The results show that Markov chain models can be useful tools for predicting consequences of local species extinction without extensive experimentation.