Pattern formation in a model of competing populations with nonlocal interactions

We analyze and compute an extension of a previously developed population model based on the well-known diffusive logistic equation with nonlocal interaction, to a system involving competing species. Our model involves a system of nonlinear integro-differential equations, with the nonlocal interaction characterized by convolution integrals of the population densities against specified kernel functions. The extent of the nonlocal coupling is characterized by a parameter delta so that when delta -> 0 the problem becomes local. We consider critical points of the model, i.e., spatially homogeneous equilibrium solutions. There is generally one critical point in the first quadrant (i.e., both population densities positive), denoting coexistence of the two species. We show that this solution can be destabilized by the nonlocal coupling and obtain general conditions for stability of this critical point as a function of delta, the specific kernel function and parameters of the model. We study the nonlinear behavior of the model and show that the populations can evolve to localized cells, or islands. We find that the stability transition is supercritical. Near the stability boundary solutions are small amplitude, nearly sinusoidal oscillations, however, when delta increases large amplitude, nonlinear states are found. We find a multiplicity of stable, steady state patterns. We further show that with a stepfunction kernel function the structure of these islands, a highly nonlinear phenomenon, can be described analytically. Finally, we analyze the role of the kernel function and show that for some choices of kernel function the resulting population islands can exhibit tip-splitting behavior and island amplitude modulation. (C) 2013 Elsevier B.V. All rights reserved.