Print PageTheoretical models of optimal prey choice predict that intraspecific diet polymorphisms can occur as a result of phenotypic variation in realized prey profitability. Competing models correspond to alternative scenarios of how individuals vary in their ranking of preferred and sub-optimal prey, and are associated with distinct predictions about how individual variation in diet will be affected by consumer density. Testing of these models with empirical data sets is mostly lacking, but this is changing with the emergence of network analysis as a powerful tool for examining patterns of individual resource use. We summarize and elaborate on existing network analytical techniques for linking data to theory, and provide a case study using dietary data from sea otters. We analyzed 63,701 feeding dives recorded from 74 tagged otters at 3 sites in California that spanned the full range of possible sea otter densities. We used weighted indexes to describe patterns of diet nestedness and modularity in individual-resource networks for each site. We disentangled the contribution of key prey types and more sporadically-consumed prey types to the patterns of nestedness and modularity using fractional diet composition analysis (FDCA). We examined the distribution of pairwise correlations in rank prey preferences at each site.
Results/Conclusions
We found that the degree of individual dietary specialization was greatest at the high-density site, and in contrast there was a high degree of overlap among individual diets at the low-density site. Nestedness in individual diets was significant at all three sites, and the high-density and medium-density sites also showed significant levels of modularity in individual resource use patterns (as compared to null models), but this was not the case for the low-density site. Using FDCA to focus in on those prey types forming the core components of individual diets caused nestedness to disappear from all 3 populations, and modularity to disappear in the low-density population (because all individuals shared the same highest-ranked prey type as their core diet component) but increase in the medium and high-density populations. Prey preference rankings were highly correlated among individuals from within the same diet modules, but largely un-correlated among individuals from different diet modules. Our results suggest that patterns of diet specialization in sea otters can be viewed as supportive of multiple prey choice models. We propose that sporadically-consumed prey and core prey may serve different functions, and their utilization could be driven by different selective forces.
