When and how to incorporate biotic interactions into models of species niches and distributions: Getting Grinnell, Elton, and Hutchinson all to agree
Great interest exists in understanding the role of biotic interactions in influencing community composition and species distributions (ranges), both currently and under climate change. The literature now indicates that biotic interactions can hold strong effects on species distributions, but no guidelines exist for integrating them into correlative ecological niche models that identify suitable areas. Fortunately, recent distinctions between what have been termed “Grinnellian” and “Eltonian” niches provide helpful insight. Under this framework, Grinnellian niche models are defined by density-independent factors not affected by the focal species (scenopoetic variables, as proposed by Hutchinson), which typically are relevant at relatively coarse grains and vary over large geographic extents. Currently, scenopoetic variables are generally considered to be abiotic, with biotic interactors instead associated with Eltonian niche models that include density-dependent feedbacks between the two species (and that are usually relevant at local scales). However, by considering population-level effects of the interacting species (rather than the effects on individuals), I here ask which classes of biotic interactions could be scenopoetic.
Some classes of biotic interactors can be assigned as scenopoetic or non-scenopoetic, but others require additional information regarding the particular species involved. Facilitators and harming ammensals constitute scenopoetic variables (and are relevant to the focal species). Conversely, commensals and harmed ammensals do not (but are irrelevant anyway). In contrast, determinations for remaining classes (competitors, mutualists, predators/prey, consumers/resources; all of which may be important) require information regarding the particular species involved. Scenopoetic biotic interactors should be included as predictors in Grinnellian models. Relevant non-scenopoetic interactors must be taken into account in other ways. They can be considered in post-processing of a Grinnellian model if the interaction affects the species’ ranges in ways that depend on the scenopoetic (e.g., abiotic) conditions present. In contrast, when the effects of the interaction are contingent upon the relative local population levels, interactors must be studied via Eltonian models (often including the outputs of Grinnellian models as inputs). This revised theoretical framework should promote more effective linking across scales as well as facilitate estimations of dispersal and demography under future climate change. Progress is needed regarding how to consider strings of interactors, as well as interactions whose effects vary across space or time.