Food web simulations: stochastic variability and systems-based conservation

Individual-Based Modeling in Conservation Biology A major challenge for food-web research is studying diversity and variability more explicitly. This means a focus on individual-level variability in populations (Bolnick et al., 2011) that hopefully might help to better understand how structural properties predict dynamical behavior (Dunne, 2006). One reason why linking structure to dynamics is still a hard challenge can be that intrapopulation variability is relatively poorly considered in most models. Yet defining developmental stages as graph nodes is a step toward managing this challenge: for example, in many food-web models certain species are represented by separate graph nodes that include juveniles and adults. Trophic status and network dynamics can be quite sensitive to this kind of demographic aggregation (or resolution) of the web and are expected to be influenced by individual-level differences in terms of behavior and feeding habits. Individual-level variability includes genetic, demographic, and stochastic factors and to date it is not easy to incorporate all these in most modeling frameworks. Yet developing the methodological background of individual-based modeling seems to be very useful for future research and applications. Since individual-level differences are more important in smaller populations (Lande, 1988), studying their effects explicitly is relevant for conservation efforts. Using network metrics as proxies or predictors of food-web dynamics is an old but still open issue. For an example, network hubs are supposed to be species of key importance and hubbish networks are thought to be safe against errors but vulnerable against attacks (Montoya and Solé, 2002). We make structural predictions routinely but we are very poor in testing these on either real-time series or simulation models. One way to make our knowledge more robust here is by adopting a comparative approach: studying spatio-temporal food-web gradients can inform about possible re