Body mass – biomass scaling modulates species keystone-ness to press perturbations

In this study, we systematically explored (1) the relationships between species’ topological and energetic traits and their keystone status, (2) how these relationships depend on the community biomass structure, and (3) how keystone-ness identified by linear approximations corresponds to the potentially nonlinear effects that species can have on their communities. To address these, we used empirical, mass-balanced Ecopath food webs as well as dynamic food web simulations generated using the Allometric Trophic Network (ATN) framework. These varied not only in complexity—such as species richness and connectance—but also in biomass structure, trending either bottom-heavy or top-heavy, i.e., scaling exponents of species’ body mass to biomass towards negative or positive. We defined keystone species as those that, when subjected to press perturbations, exert disproportionately large effects on the biomasses of all other species within the community. To quantify species’ keystone-ness, we employed two approaches. First, we derived the normalized inverse community matrix for both empirical and model-generated food webs at equilibrium, quantifying species’ keystone-ness as their per capita net effect, i.e., the net effect of a unit perturbation-induced change in the biomass of the perturbed species on the equilibrium biomasses of all other species. Second, for model food webs, we simulated species-specific press perturbations, either by reducing production gain or increasing metabolic loss, and measured the per capita community biomass change. We then examined the relationships between species’ topological and energetic traits and their keystone-ness, as well as how these relationships depend on the food webs’ biomass structure. We thereby evaluated the consistency between species’ (linearly approximated) per capita net effect and (potentially nonlinear) per capita community biomass change regarding conclusions on species’ keystone status.