Optimality in reserve selection algorithms: When does it matter and how much?

Abstract

This paper responds to recent criticisms in Biological Conservation of heuristic reserve selection algorithms. These criticisms primarily concern the fact that heuristic algorithms cannot guarantee an optimal solution to the problem of representing a group of targeted natural features in a subset of the sites in a region. We discuss optimality in the context of a range of needs for conservation planning. We point out that classical integer linear programming methods that guarantee an optimal solution, like branch and bound algorithms, are currently intractable for many realistic problems. We also show that heuristics have practical advantages over classical methods and that suboptimality is not necessarily a disadvantage for many real-world applications. Further work on alternative reserve selection algorithms is certainly needed, but the necessary criteria for assessing their utility must be broader than mathematical optimality.