In spite of the generally acknowledged importance of interspecific interactions for the dynamics of natural populations, most decisions in resource management and conservation are still made on the basis of single-species models. Many researchers have called for more comprehensive approaches, and there are a growing number of attempts to include the community context in management decisions. Dynamic food web models describing the community's consumer-resource interactions are a common means of providing that context, but there are continuing arguments about what elements are essential for a predictive food web model. Most previous models leave out classes of interaction such as mutualism and facilitation, dynamic traits such as behavior and life history, spatial structure, and externally imposed temporal variability, among other factors. How would greater inclusion of such phenomena improve our understanding of food web function? How would that aid us in predicting and managing changes in real food webs that may occur with disturbances such as habitat loss, species invasions, or climate change? This session brings together a range of people who have introduced more biological details to dynamic food web models or who have applied such models towards conservation and management questions. This session incorporates a variety of approaches to making food web models more realistic. Topics addressed by talks will include the incorporation of nontrophic interactions such as mutualism, effects arising from behaviors such as dispersal, inducible defenses, and foraging decisions, spatial and temporal structure, ecosystem functioning, and the importance of understanding how these and other factors determine the success of species in communities. The talks are united by the common theme of our practical need to expand dynamic models beyond simple pairwise, trophic, and mean-field perspectives. The range of topics allows comparisons of the promises and challenges presented by different approaches.