干旱区生态环境知识资源中心

A food web describes the feeding links between species in a community. The species in many food webs are organized into groups of highly linked species that are weakly linked to species in other groups. A Bayesian approach to identifying such groups in an observed food web is described. This approach extends a previous non-Bayesian one that does not exploit information about the relatively high density of links within groups and relatively low density between groups. Under the new approach, this information is encoded through prior distributions for within- and between-group link densities. The approach is shown to work well on simulated food webs. Results are presented of the application of the method to the Coachella Valley desert food web.