Non-trophic interactions are widely recognized as critically important in ecological systems, yet most models of large communities consider only trophic interactions (i.e. food webs). Non-trophic interactions may be incorporated into large food web models using interaction modifications (IMs), whereby the magnitude of a direct interaction between two species may be affected by the density of a third. However, non-trophic interactions have diverse causes, which often imply IMs which are structured in specific ways. Modeling IMs in sufficient detail to reflect this diversity could make adding IMs to large food web models much less tractable, but how important is doing so for accurately capturing the system-wide effects of IMs? Put another way, how much does each structural element of an adaptive behavior contribute to that behavior's community-wide effects? We compare assembled food webs containing IMs which correspond to adaptive foraging in the face of a tradeoff between foraging rewards and predation risk, to webs with patterns of IMs that retain various combinations of some, but not all, of the nonrandom structural constraints implied by that adaptive behavior. We examine how different the effects of various 'partially -randomized' IMs are, compared to IMs corresponding to this particular adaptive behavior.
Results/Conclusions
Depending on the parameter values chosen, webs generated by our 'full' optimal foraging model differed from those generated by a similar model without IMs in the means of several food web descriptors, and in the temporal variability of these metrics through the assembly process. In general, webs generated by our 'full' model were the most different from the no-IMs model webs. Webs generated by models containing more of the nonrandom features of the full model were usually more similar to the full model's webs than those generated by models with fewer of the nonrandom features, such that a given model's behavior was brought closer to that of the full model by each nonrandom structural element of the full model that was incorporated. However, one or more models also often had effects that were in the opposite direction of those of the full model. In some cases, predictions of the full model were sufficiently different from all others to suggest that the adaptive nature of that model's IMs (i.e. that they maximized an adaptive forager's growth rate) played a critical role in total IM effects. Mechanistic explanations for IM effects in each model will also be discussed.