Understanding the impact of species extinctions on the functioning of complex food webs is a challenging task. Some recent theoretical work has shown that the impacts of single species lost can be predicted with only a few variables, such as the biomass and the trophic rank of the extinct species. In nature, however, extinctions are rarely isolated events and extinction cascades are very likely to happen. Our objective in this study was to develop a general method to predict the impacts of multiple species extinctions in complex food webs without a detailed knowledge of all species interactions. We set up a biodiversity-ecosystem functioning experiment with mesocosms of marine and freshwater communities. The food webs were made of 9 and 10 species and had a complex structure including intra-guild predation, omnivory and non-trophic interactions. We assembled mesocosms with 0, 1, 3, 6, 9 (marine) and 0, 1, 2, 4, 10 (freshwater) extinct species, and performed 10 random extinctions sequences. We measured 12 and 7 ecosystem functions for the marine and freshwater systems respectively to estimate primary and secondary production. Ecosystem functioning was analyzed with a nested linear model that partition the variance between species richness (additive and non-additive), species identity and community composition.
Results/Conclusions
We found that in the freshwater system, species richness is a poor predictor of ecosystem functioning, while in the marine system the relation with species richness was linear and more predictable. The independent contribution of each species was the best predictor of ecosystem functioning for both systems, with explained variance ranging from 10% to 35%. The full model including all effects (species richness, identity and composition) was slightly better with explained variance ranging between 31% and 57%. We also found that the contribution of each species to ecosystem functioning could be predicted by their body size and diet. Our results suggest that the contribution of diversity per se to ecosystem functioning varies across systems and that in all cases it is not a good predictor of change, as it explains only a small fraction of the variance. Our results also suggest that, even when interactions are expected to be important, a good knowledge of each species trait is sufficient to explain to a great extent the impact of multiple extinctions. This finding is of great importance since the prediction of the impact of biodiversity losses in complex, more realistic food webs, may be simpler than the current theory would predict.