Encounter risk is critical to many aspects of ecology and conservation, spanning topics such as predator-prey interactions, mate finding, dispersal, and disease dynamics. However, encounter risk is rarely explicitly linked with conservation management actions. For some marine mammals, collisions with watercraft cause a significant portion of overall mortality, and conservation efforts would benefit greatly from the ability to quantify the effects of management on lethal encounter risk. The establishment of protection zones is viewed as a primary management objective to reduce mortality risk, yet implementation of these zones can be contentious. This work applies recent advances in encounter rate theory, nested in a Bayesian Belief Network, to quantify the relative risk of lethal collisions between Florida manatees (Trichechus manatus latirostris) and watercraft in a real landscape. Abundance of manatees and boats are two critical state variables that must first be estimated and modeled in space and time to apply this framework in a landscape. We do this by combining multiple sources of information (including counts from aerial surveys) with advances in Bayesian statistical estimation. We also apply decision theory, using two different optimization approaches (simulated annealing and linear integer programming (LIP), to determine optimal configuration of protection zones under various costs and constraints scenarios.
Expected manatee abundance and distribution varied significantly among seasons, therefore so did the mortality risk and the optimal configuration of protection zones. The most important habitat covariate for predicting abundance shifted from distance to seagrass in the warm season (late spring - early fall), to distance to development in the cool season (late fall and early spring) and cold season (winter). This likely reflects a change in behavior time budgets from foraging to thermoregulation with cold weather. We derived estimates of total mortality risk in the landscape under current management and no protection scenarios, to quantify the predicted effectiveness of current management. Surprisingly, the predicted optimal configuration of zones was very similar between optimization approaches, even though we prioritized connectivity with simulated annealing and did not with LIP. We developed cost proxies for burdens to managers and waterway users that vary in space, time, and zone. Cost, cost weighting, and specific management constraints, along with the assumed relationships of speed with probability of avoidance and mortality risk, strongly influence what is considered optimal. This work provides a useful tool for exploring the effectiveness of existing zones, and the implementation of optimal management protection for marine mammals.