Introduction: Understanding how species are distributed in space and how they interact with each other is central for scientific and conservation purposes. Species' distributions and interactions result from a complex interplay of local trophic dynamics, dispersal processes, resource availability, and abiotic factors governed by the landscape matrix, which also determines the spatial connectivity for organisms' dispersal and resource fluxes. River networks not only exhibit universal spatial structures, but their dendritic landscape structure is tightly linked to species and metacommunity processes therein.
Methods: Here, using a mechanistic model of spatially connected food webs integrating both essential biological and hydrological aspects, we investigate how food-web properties vary in space, and how these patterns are influenced by key model parameters. We then contrast our predictions with a suite of null models, where different aspects (such as spatial structure or trophic interactions) of the spatial food-web model are alternatively relaxed.
Results: We find that species richness is highest in areas where local nutrient load is maximal (lowland headwaters, according to our default assumption). Overall, species richness is positively associated with link density, modularity and omnivory, and negatively related to connectance, nestedness, and niche overlap. However, for metrics such as connectance and omnivory, stochasticity of trophic interactions is a much stronger predictor than spatial variables such as distance to outlet and drainage area. Remarkably, relationships between species richness and food-web metrics do not generally hold in null models, and are hence the outcome of coupled biological and physical (i.e., hydrological) processes characteristic to river networks.
Discussion: Our model generates realistic patterns of species richness and food-web properties, shows that no universal food-web patterns emerge as a result of the riverine landscape structure, and paves the way for future applications aimed at disentangling metacommunity dynamics in river networks.