Are spatially informed food web models better than non-spatial ones?

Background/Question/Methods

A central question in ecology is to understand how networks of interacting species such as food webs shape biodiversity and species persistence in nature. It was shown that the stability of food webs decays with diversity, a long-standing problem known as the complexity-stability dilemma. In the last decade, multiple stabilizing factors have been described. For example, it was shown that some features of food web topology like, a relatively low degree of nestedness and high modularity were factors that could stabilize food web dynamics. Moreover, it was shown that the spatial structure of interacting populations can also have stabilizing effects. Nevertheless, we know little about whether there exists a relationship between the spatial distribution of the species embedded in a food web, and its topology. The challenge now is to understand how food web structure is constrained by the spatial distribution of species to address how specific spatial distributions will affect food web stability. As a first approach to address that issue, I generated theoretical food webs using well–known probabilistic models (e.g. cascade model) parameterized with data from empirically compiled food webs (Interaction Web DataBase). Second, I generated theoretical spatial distributions such as those found in empirical systems (i.e. nested, mosaic and homogeneous distributions). Third, I used those spatial distributions to trim forbidden interactions from the theoretically generated food webs based on non co-occurrence between a given pair of species. The latter procedure yields what I call spatially informed food webs. Finally, I compared the fit of the non-spatial theoretically generated food webs for each food web model to the empirical food web to the fit of the spatially informed food webs. 

Results/Conclusions

My preliminary results show that for the modular (or mosaic) spatial distribution, the spatially informed food webs are a better fit to the empirical food web than the non-spatial ones. Also, for the nested spatial distribution, the spatially informed food webs show consistently worse fits than the non-spatial theoretical food webs. On the other hand, spatially informed food webs with random distributions show sometimes better fits than other kinds of distributions. Overall, these results suggest that the spatial distribution of interacting species may be relevant to understand food web topology and stability, while they underline the necessity for further study with respect to which particular kinds of species distributions constrain topology and how.