Many of the insights offered by analyzing the structure of food webs (e.g. by identifying compartments or groups of trophically similar species) rely on aggregation of predator-prey interactions to the species level. However, predator-prey interactions are fundamentally individual-level processes. In fact, there is often large variability in diet among the individuals of a species. In marine systems especially, there is often substantial intraspecific variation in diet driven largely by variation in body size. Given this variability, we can view the process of assembling a species-level food web as already applying a grouping of the individuals in the community, implicitly and a priori. However, this grouping may obscure the actual individual-level traits and states that structure feeding relationships. In this work, we explore the potential importance and structure of individual-level variability in diet through the use of an extensive database of predator gut contents collected by NOAA's Alaska Fisheries Science Center from the Gulf of Alaska and Bering Sea. We analyze the effect of size on variability in diet both within and between species and explore novel extensions of existing group models that explicitly incorporate this variability and identify structure that would otherwise be hidden at the species level.
Results/Conclusions
Our analyses of the gut contents data show a substantial effect of predator size on both the size and type of prey consumed. As expected, larger predators tend to consume larger prey, but diet composition also shifts as predators (e.g. cod) become large enough to exploit certain prey (e.g. crabs). This suggests that size is likely to account for a large portion of the variability in predator diet as individuals undergo ontogenetic niche shifts as they grow and highlights the mechanistic detail that can be lost by aggregating food webs to the species level. Moreover, these analyses suggest that size may provide a more effective way to organize predators than species identity in this system. At the same time, the assembly of a species-level food web nonetheless provides valuable insights into the structure of the system, as the application of a Bayesian group model identifies several clusters of predators with similar diets. These two results highlight the potential for merging size- and trait-based perspectives with species-based perspectives. Future work will develop methods that can explicitly account for hierarchical organization (i.e. individuals nested in species nested in trophic groups) to better understand the mechanisms structuring feeding relationships.