Beach mice are federally threatened species that inhabit coastal dunes of Alabama and Florida. These species are affected by multiple stressors, including habitat loss, predation from cats, light pollution, and other factors associated with human use of beaches. Given the small range of each subspecies of beach mouse and the high desirability of their habitat by humans, conservation opportunity is both limited and associated with a high cost. Consequently, conservation efforts must be well-targeted to achieve recovery objectives. In this study, we assessed habitat objectives for 3 subspecies of beach mice in the panhandle of Florida (Perdido Key, Choctawhatchee, and St. Andrew beach mouse). To accomplish this, we first created a spatially explicit Bayesian network model that predicted the probability of occurrence for each 30-m pixel across each subspecies’ range in Florida. Occurrence was predicted from the suitability of a pixel and its neighborhood, which reflected numerous factors including burrow suitability, food availability, and predator avoidance. Using the recovery criteria outlined in the species recovery plans, we derived a range of habitat objectives for each subspecies under different management scenarios, demonstrating empirically the importance of targeted conservation efforts and influencing current management strategies for these species.
Results/Conclusions
When coupled with established population objectives, the models provided insight into how much habitat is available, how much more is needed, and where conservation or restoration efforts can most efficiently achieve established objectives. Because there are a multitude of reasons and opportunities for habitat restoration, we provided managers a range of hectares needed to reach subspecies population objectives. Recovery plans state that either 50% (Perdido Key or Choctawhatchee) or 87% (St. Andrew) of critical habitat must be both protected and occupied. To calculate this area, we identified existing critical and protected areas and assumed a pixel was occupied if its predicted probability was >= 70%. The models suggest that habitat objectives are met for two subspecies while the St. Andrew subspecies requires 347 +/- 0 to 1081 +/- 17 additional hectares. Managers can use the model to run scenarios on a focal area (e.g. will implementing a cat capture program increase the probability of beach presence) and use the maps to identify contiguous areas that could be connected by habitat corridors. We will present these and other results regarding the processes that control the distribution and abundance of these species, and discuss how our approach informs restoration and monitoring efforts.