Food webs are complex networks of predation interactions. Most of the models proposed to explain these interactions are not directly related to species features like body size or phylogeny. In 2000, Williams and Martinez introduced their famous niche model of food web structure. In their model, the authors suppose that each species is characterized by a niche value, a number between 0 and 1 drawn at random according to some rules. Their aim is to reproduce qualitatively the structure of a food web. Here we propose a statistical version of this model that allows a quantitative estimation of the niche values for the species of observed food webs. These estimation are made using Bayesian statistics coupled with MCMC techniques.

Results/Conclusions

With these estimations, we first can fit food webs and compare the predicted webs with the observed ones, and in this way evaluate the accuracy of the niche model. Secondly, we relate the estimated niche values with external biological information such as body size or phylogeny.