Combining field experiments and individual-based modeling to identify the dynamically relevant organizational scale in a field system

Community ecologists continually strive to build analytical models that realistically describe long-term dynamics of the systems they study. A key step in this process is identifying which details are relevant for predicting dynamics. Currently, this remains a limiting step in development of analytical theory because experimental field ecology, which provides the key empirical insight, and theoretical ecology, which translates empirical knowledge into analytical theory, remain weakly linked. I illustrate how an individual-based computational model of species interactions is a useful way to bridge the gulf between empirical research and theory development. I built a computational model that reproduced key natural history and biological detail of an old-field interaction web composed of a predator species, a herbivore species and two plant groups that had been the subject of extensive previous field research. I examined, using simulation experiments, how individual behavior of herbivores in response to changing resource and predator abundance scaled to long-term population-level and community-level dynamics. The simulation experiments revealed that the long-term community dynamics could be highly predictable because of two counterintuitive reasons. First, seasonality was a strong forcing variable on the system that removed the possibility of serial dependence in population abundance over time. Second, because of seasonality, short-term behavioral responses of herbivores played a much stronger role in shaping community structure than longer-term processes such as density responses. So, simply knowing the short-term responses of herbivores at the evolutionary ecological level was sufficient to forecast the long-term outcome of experimental manipulations. This study shows that an individual-based model, once it is calibrated to the real-world field system, can provide key insight into the biological detail that analytical models should include to predict long-term dynamics.