Abstract: Ecological communities comprise many species interacting in various ways. Because of these interactions, species' per-capita growth rates often depend in nonlinear ways on their own densities, and on the densities of other species. Predicting and explaining community dynamics requires modeling those nonlinearities with sufficient accuracy. Historically, one particular class of nonlinearities has been regarded as both particularly challenging and particularly important to identify: higher order interactions (HOIs). A HOI is indicated when a model accurately describing the dynamics of isolated species and pairs of species fails to accurately predict the dynamics of the full multispecies community. HOIs are challenging to identify because, by definition, they cannot be identified in experiments that study single species or pairs of species in isolation from the rest of the community. However, HOIs arise from the same underlying ecological mechanisms that give rise to other nonlinearities. Those other nonlinearities can be just as challenging to identify as HOIs, and are just as important to identify in order to predict and explain community dynamics. Here, I show that both HOIs and other nonlinearities occur in extremely simple laboratory communities, and that they are sensitive to environmental conditions. I grew replicate populations of each of three species of bacterivorous ciliate protists in all possible 1-, 2- and 3-species combinations, in each of two environments (low or high resource enrichment). The experiment lasted dozens of generations, providing long-term data from which to parameterize models of population dynamics. I parameterized models from the 1- and 2-species communities, and used the parameterized models to predict the dynamics of the 3-species communities. At low enrichment, the three species exhibited a HOI: a Lotka-Volterra competition model fit the 1- and 2-species dynamics, but failed to predict the 3-species dynamics. At high enrichment, the three species did not exhibit a HOI. Instead, one species exhibited nonlinear intraspecific competition it did not exhibit at low enrichment. Thus, the same three species could either exhibit an HOI, or some other nonlinearity, depending on the environmental conditions. This confirms recent theory predicting that the occurrence and strength of both HOIs and other nonlinearities is likely to be highly idiosyncratic. I discuss the implications of these results for future research on HOIs as a distinct class of nonlinearity. I argue that distinguishing HOIs from other nonlinearities is an artificial distinction, and that future research should focus more broadly on all the nonlinearities that drive community dynamics.
