Phenotypic plasticity is a key factor for the success of organisms in heterogeneous environments. Although many forms of phenotypic plasticity can be induced and retracted repeatedly, few extant models have analyzed conditions for the evolution of reversible plasticity. We present a general model of reversible plasticity to examine how plastic shifts in the mode and breadth of environmental tolerance functions (that determine relative fitness) depend on time lags in response to environmental change, the pattern of individual exposure to inducing and noninducing environments, and the quality of available information about the environment. We couched the model in terms of prey-induced responses to variable predation regimes. With longer response lags relative to the rate of environmental change, the modes of tolerance functions in both the presence or absence of predators converge on a generalist strategy that lies intermediate between the optimal functions for the two environments in the absence of response lags. Incomplete information about the level of predation risk in inducing environments causes prey to have broader tolerance functions even at the cost of reduced maximal fitness. We give a detailed analysis of how these factors and interactions among them select for joint patterns of mode and breadth plasticity.

To begin identifying what behavioral details might be needed to characterize community dynamics and stability, we examined the effect of prey behavioral responses to predation risk on community dynamics and stability. We considered the case of prey altering their foraging effort to trade off energy gain and predation risk. We used state-dependent dynamic optimization to calculate the optimal trade-off for four models of prey behaviorally responding to predation risk. We consider a fixed behavior model in which prey use constant levels of foraging effort and three flexible behavior models in which prey change their foraging effort according to their physiological state and their perceived level of predation risk. Flexible behavior was destabilizing at the community level as evidenced by higher predator-prey oscillations and lower community persistence times. The mechanisms by which prey estimated predation risk also affected community stability. We found that community dynamics resulting from prey with flexible behavior and fixed perception of risk approximated community dynamics resulting from prey with flexible behavior and perfect information about predation risk, however neither approximated the community dynamics resulting from prey with flexible behavior and flexible perception of risk. Thus, whether it might be possible to abstract complex behavior with simpler rules when modeling community dynamics depends on the prey's behavioral mechanisms, which are empirically poorly known.