The marbled murrelet (Brachyramphus marmoratus) is an endangered seabird that nests in coastal forests from Alaska to California. The value of these forests for human use, coupled with the difficulty of determining whether a forest stand is occupied by nesting marbled murrelets, poses a dilemma for land managers. Should they implement a costly survey to gather information on whether a potential nest site is occupied, or should they allow human use, effectively assuming the site is unoccupied? This article demonstrates the application of the partially observable Markov decision process (POMDP) as a framework for addressing this question. The analysis yields a policy in which the optimal action is a function of the decision-maker's subjective probability that a potential nest site is occupied by marbled murrelets. By incorporating stochastic state dynamics and the choice of whether to invest in learning, the POMDP provides a formal representation of adaptive management when active learning is possible.

JEL Classification Codes: Q22, Q23