Ecological management of stochastic systems with long transients

Long transients may be common in ecological dynamics, but the implications such dynamics have for ecological management have not been fully explored. Long transient periods can easily be mistaken for stable state dynamics, but may require dramatically different management policies. Here, I explore the optimal management of stochastic ecological systems that may contain either a tipping point or a ghost attractor: an important mechanism which can give rise to long transients. I consider three approaches of increasing sophistication: (1) dynamic management under a fixed model, (2) management that accounts for uncertainty over possible models, and (3) adaptive management that actively learns the correct model over the management process. This analysis confirms the prediction that long transients can create considerable uncertainty and give rise to very different optimal management policies, and also illustrates that dynamic management that can either plan for this uncertainty or actively learn to decrease the uncertainty can promote successful management of long transients.