Differential effects along a stressor gradient cause nonlinear B-EF relationships
Stressors can change both diversity (B) and ecosystem functioning (EF), as such leading to B-EF relationships that predict the functioning of stressed ecosystems for a given amount of stress-induced diversity loss. Using a well-known community model, we theoretically explore how toxicity-induced changes in fi (per-capita contributions of species i to EF) and Ki (carrying capacity of species i) map to changes in B and EF. Next, we test how applying different concentration ranges of the toxicant would result in different B-EF relationships.
We simulated a community of 20 ecologically equivalent species, each with an initial density of 100, assuming intraspecific competition was higher than interspecific competition, a prerequisite for co-existence. We considered the special case where species sensitivity for contaminant-induced reductions in Ki is proportional to species sensitivity for contaminant-induced reductions in fi, and that fi is a more sensitive endpoint than Ki, i.e. that changes in per-capita functional rates occur sooner (i.e. at lower concentrations) than changes in the maximal density attained.
At stress levels only affected the median of the community-mean per-capita contributions (fi) to ecosystem functioning (EF), no significant linear B-EF relationship (with B = Evenness) could be found. Stress levels affecting only fi and Ki (and thus density) changed Shannon evenness by <1%, resulting in steep B-EF slopes. Stress levels affecting Shannon evenness, next to fi and Ki, led to significant B-EF slopes that were less steep. Stress levels affecting fi, Ki (and thus density), Shannon evenness, and richness, produced the flattest B-EF slopes, yet resulted in the most profound changes in EF. Overall, the model suggests that concentration-range-specific mechanisms driving EF loss cause nonlinear B-EF relationships across the complete concentration range.
Our results were sensitive to changes in the degree of interspecific competition. Additional simulations run for higher degrees of interspecific competition (but still less than interspecific competition) linearized overall B-EF slopes, mostly because evenness became more sensitive to stress than density. This is sensible since higher competition among species will more rapidly lead to uneven communities for the same stressor effects on density. Nevertheless, the overall B-EF relation remains nonlinear, as found for the initial parameter setting.