Effects of sampling effort on characterization of food-web structure

A critical and poorly understood aspect of food-web theory concerns the possibility that variable observation effort, such as widely different sampling intensities among investigators, confounds structural food-web patterns. We evaluated this possibility by simulating the effects of variable observation effort on the structure of a food web including 77 insect species found inside the stems of 10 species of grasses. A highly detailed description of the trophic structure of this community was provided by an exhaustive sampling program involving dissection of 164 215 grass stems over 12 yr. Most significantly, the data describe the frequency at which each of the consumers and their 126 different trophic links were observed. During the simulated increase in sampling, consistent trends were observed among trophic-species webs as the species richness of these webs increased to a maximum of 73 trophic species. Connectance remained surprisingly constant, while the fractions of top and basal species decreased, and the fraction of intermediate species increased. These trends were much less consistent among taxonomic-species webs. This suggests that comparative analyses of connectance among trophic-species webs constructed with varying degrees of moderate observation effort are generally robust. In contrast, the fractions of species appear to require corrections for variable sample sizes for such analyses. These corrections may be easier to develop for trophic-species webs because of their relatively simple changes with sampling effort.