Food web conservation vs. strategic threats: A security game approach.

• Strategic interactions over the food web significantly affect conservation prioritization. • Security game approach and system reliability theory are used for ecological modeling. • Model prioritizes food web species to allocate defensive resources vs. a strategic threat. • Model quantifies the variation in food web reliability as the agents' strategies change. • The model provides an overarching intervention strategy to minimize web unreliability. Multi-species conservation is of critical concern in ecosystem management science. In this context, modeling the effect of strategic threats on decision-making is a challenging problem that has not been sufficiently addressed. Using a security game approach, this paper investigates the optimal conservation of a food web against a strategic threat. The model builds upon the non-cooperative Stackelberg game, wherein conservator (defender) and adversary (attacker) play as leader and follower, respectively. The objective of the defender is to preventively maximize the entire web reliability, under financial and ecological constraints. The defender optimally manipulates the populations of an optimal subset of species to achieve this. In contrast, the attacker attempts to maximize web unreliability by decreasing the population of selected species, using limited resources. A meta-heuristic algorithm is developed to compute the equilibrium strategy, and the model is validated through numerical examples. Additionally, in a scenario-based approach, it is examined how the defense and attack strategies, as well as food web reliability, change as the population of keystone species change. The results also show that the combinational use of mathematical optimization and food web-specific conservation prioritization indices yields a practical tool for food web conservation prioritization. The results specifically yield theoretical insights into how to optimally control trophic cascade effects due to changing keystone species populations. A step-wise methodology is proposed to implement the model. [ABSTRACT FROM AUTHOR]