Species interactions cause non-additive effects of multiple environmental stressors on communities

Abstract

Cumulative ecological impacts of chronic, extreme, and often novel, anthropogenic environmental changes (i.e., stressors) often differ from the sum of their individual effects. Uncertainty over the causes of such non-additivity among multiple stressors confounds forecasts of their net ecological impact. Although stressors can interact directly within the environment to mediate their combined effects on communities, species interactions likely also play key roles. Here, we use a simulation model to explore how species interactions cause community responses (changes in species richness and total biomass) to paired stressors to differ from what we would expect based on the individual effects of each stressor (the additive effect). We demonstrate how interspecific interactions cause communities to respond non-additively to stressors, and how this depends on whether these interactions are negative or positive and whether the stressors have positive or negative impacts on the community property of interest. When pairwise species interactions involve at least one negative interaction (i.e., competition or predation), stressors combine to have greater than expected negative impacts (e.g., species or biomass loss) and less than expected positive impacts (e.g., biomass increases). In contrast, reciprocally positive interactions between species (i.e., facilitation) generally cause stressors to have additive, or slightly less than additive, net effects on species richness and community biomass. While species interactions determine the nature of the combined impact of multiple stressors (i.e., greater than or less than expected), species co-tolerance and stressor timing (i.e., sequential vs simultaneous application) only modify the magnitude of this effect. These findings highlight how interactions among species can contribute to non-additive responses by communities to environmental change, in addition to those caused by interactions among stressors themselves.