Artifacts or attributes? Effects of resolution on the Little Rock Lake food web

A detailed and relatively evenly resolved food web of Little Rock Lake, Wisconsin, was constructed to evaluate the sensitivity of food-web patterns to the level of detail (degree of resolution) in food-web data. This study presents definitions (e.g., ecosystem food webs) and methods for constructing and reducing the resolution of food webs to provide relatively pragmatic and rigorous touchstones for consistency in future food-web studies. This analysis suggests that food-web patterns such as the scale-invariant links-per-species ratio, short chain lengths, and limited number of trophic levels are constrained by the resolution of food-web data rather than by ecological factors. Patterns less sensitive to changes in resolution such as directed connectance (the proportion of observed directed links to all possible directed links) may be robust food-web attributes. The food web of Little Rock Lake appears to be the first highly and evenly resolved food web of a large natural ecosystem originally documented for the purpose of examining quantitative food-web patterns. This ecosystem food web contains roughly twice as many species as the largest web to date. It also may provide the most credible portrait available of the detailed trophic structure of a whole ecosystem. The 93-trophic-species web of Little Rock Lake differs from previously published trophic-species webs by having more links per species (L/S = 11), longer chain lengths (average: ≥10, maximum: ≥16), species at higher trophic levels (maximum: = 12), higher fractions of intermediate species, and smaller fractions of top species and links to top species. The sensitivity of quantitative food-web patterns to changes in resolution was examined in several series of trophically aggregated Little Rock Lake webs. Each of the series starts with a highly and relatively evenly resolved web with 182 consumer, producer, and decomposer taxa and ends with low-resolution webs with 9 aggregates of taxa. Taxa were aggregated based on the proportion of predators and prey shared by the taxa. Different series of webs were generated using different criteria for linking aggregates to evaluate the sensitivity of food-web patterns to linkage criteria. The sensitivity analysis revealed that several, but not all, quantitative food-web patterns are very sensitive to systematic aggregation of the web. Sensitive patterns include number of links per species, linkage complexity, the distributions of chain lengths and species among trophic levels, and the proportions of top species and links to top species. Less-sensitive patterns include connectance, the ratio of predators to prey, the proportions of intermediate and basal species, and the proportions of links that are between intermediate and basal species. Directed connectance is the only pattern examined that is both very robust to trophic aggregation and generally comparable to other community webs. Quantitative food-web patterns in published community webs are generally similar to highly aggregated Little Rock Lake webs (versions with 9-40 aggregates). These findings suggest that previously described community food webs are severely aggregated versions of more elaborate webs similar to that of Little Rock Lake.