Combining species distribution models and value of information analysis for spatial allocation of conservation resources

摘要

1. Managers often have incomplete information to make decisions about threatened species management, and lack the time or funding needed to obtain complete information. Value of information (VOI) analysis can assist managers in deciding whether to manage using current information or monitor to reduce uncertainty before managing. However, VOI analysis has not yet been applied to spatial allocation of monitoring resources across a landscape. 2. Here, we demonstrate how to make the best use of data from species distribution models (SDMs) and VOI analysis to assess the value of land protection decisions for single and multiple-species objectives across a heterogeneous landscape. Our method determines the situations where one should monitor before protecting the land, and those where one should act based on current incomplete information. Further, we prioritize land planning units based on cost-effectiveness (expected number of occurrences protected per dollar spent) and identify properties to target for monitoring or immediate conservation. 3. In a single species case study, we found that the optimal decision was to act based on current information when the prior probability of detecting an occurrence in a survey was low. When probability of detection was high, it was most effective to monitor the majority of units. In a multi-species case study, monitoring was only optimal in 50% of cases, due to high inferred probability of at least one occurrence of a threatened species in many units. When compared to a simulation where units were monitored by default, using VOI to determine which units were monitored or prioritized for immediate conservation led to an increase in the expected number of occurrences protected. 4. Synthesis and applications. Using a combination of species distribution models and value of information analysis can assist managers in efficiently distributing limited resources for protected area allocation. Our results suggest that if managers can use valu