Identifying the mechanisms which allow large, complex food webs to persist is a central goal of ecology. Seminal theoretical work of May and others identified the challenges that complexity imposes on the coexistence of species in large communities, yet complexity, including trophic omnivory, is common in real food webs. Many structural properties have been shown to stabilize large food webs, while dispersal has been shown to play a significant role in species coexistence. However, it is not well understood how the mechanisms for maintaining coexistence in trophic communities that are isolated and those which are interconnected by dispersal interact. Here, we report the results of models of complex spatial food webs. Simulated food webs were constructed using a size-based “niche model” paired with consumer-resource models where vital rates were determined by empirical allometric relationships. Species within the food web were allowed to disperse among habitat patches.
Our models produced a wide array of dynamics, including stability, dramatic fluctuations, and extinctions. Our results clearly showed a general increase in the ability of species within food webs to coexist as a result of the effects of dispersal and spatial asynchrony. Surprisingly, we observed an inverse relationship between the ability of food webs to persist when isolated and their tendency to be asynchronous when spatial, limiting the ability of food webs that are persistent when isolated to benefit from persistence generated by spatial asynchrony. Individual food webs showed a wide range of idiosyncratic sensitivities to the influence of spatial effects, with the presence of omnivory having little notable impact. Rather, the overall persistence of spatial food webs appeared to be determined by a complex interplay between the size-structure of a given food web, which influences the stability of the food web in isolation, and the action of spatial coexistence mechanisms. Our findings suggest that a food web’s persistence when isolated cannot readily predict its persistence in a spatial environment.