Abstract
Methods for reserve selection and conservation planning often ignore uncertainty. For example, presence‐absence observations and predictions of habitat models are used as inputs but commonly assumed to be without error. We applied information‐gap decision theory to develop uncertainty analysis methods for reserve selection. Our proposed method seeks a solution that is robust in achieving a given conservation target, despite uncertainty in the data. We maximized robustness in reserve selection through a novel method, “distribution discounting,” in which the site‐ and species‐specific measure of conservation value (related to species‐specific occupancy probabilities) was penalized by an error measure (in our study, related to accuracy of statistical prediction). Because distribution discounting can be implemented as a modification of input files, it is a computationally efficient solution for implementing uncertainty analysis into reserve selection. Thus, the method is particularly useful for high‐dimensional decision problems characteristic of regional conservation assessment. We implemented distribution discounting in the zonation reserve‐selection algorithm that produces a hierarchy of conservation priorities throughout the landscape. We applied it to reserve selection for seven priority fauna in a landscape in New South Wales, Australia. The distribution discounting method can be easily adapted for use with different kinds of data (e.g., probability of occurrence or abundance) and different landscape descriptions (grid or patch based) and incorporated into other reserve‐selection algorithms and software.