We demonstrated, using an individual-based evolutionary model that cannibalism can also account for the evolution of density-dependent behavioral ‘phase change’ in locusts. A key (and differentiating) aspect of our modeling is that we considered a framework in which behavioral interactions and plasticity are not pre-determined, but rather they evolve. At low densities locusts’ evolutionary stable strategy is to actively avoid contact with others, resulting in solitarious-type behavior. As local density increases above a critical value, however, the favored strategy is for individuals to move away from those who approach while being attracted from those who move away. Thus, our model suggests that a low density avoidance behavior followed by a transition to gregarious behavior beyond the critical density, resulting in high density bands, are both individual-level adaptations to minimize cannibalism. our cannibalism hypothesis also accounts for the observed hysteresis in locust phase change behavior. Thus phase change minimizes the costs of cannibalism, and also may allow individuals to effectively migrate out of nutrient poor environments, providing a novel and parsimonious explanation for the evolution of phenotypic plasticity in locusts.