Putting a lake together: Integrating synthetic data and field observations to build a better food web

Abstract

Food webs provide context to understand how ecological communities will respond to environmental change, but revealing their structure typically relies upon time-intensive sampling and analysis of species' diets. As a result, all food web models require some unavoidable simplifications because of limited data availability, whether temporally, spatially, or taxonomically. Large databases of published trophic interactions have made this process somewhat easier, but knowledge gaps persist. We combine the use of databases with extensive field surveys, including gut-content analysis, to generate a food web for Lake George, NY. Including aquatic plants, phytoplankton, zooplankton, macroinvertebrates, and fish, our analysis identified 279 genera in the lake involved in 1910 interactions. After removing genera with no identified interactions or improbable interactions and grouping some genera into higher categories, the food web included 49 nodes with 484 interactions among them. The network structure of the inferred Lake George food web exhibits several common patterns such as relatively few trophic levels and the prevalence of tritrophic chains. Our results suggest that constructing food webs from databases provides a useful first step to determine topology. However, in situ sampling allowed us to account for additional interactions, as only 50 of the 106 directly observed interactions between fish and their prey were also found in published databases. Finally, we highlight the need to focus on developing a better understanding of herbivory in lakes, as species interactions among the diverse plankton and macroinvertebrate populations are not well known.