Decision theory provides a quantitative and transparent mechanism for planning sequential decisions (policies) which maximize some objective in the face of uncertainty. Conservation planning and natural resource management frequently rely on the tools of decision theory to determine strategies, such as where to place marine protected areas, how to set fishing quotas, how to balance control of invasive species with protection of endangered species, or how to predict the response of individual human and institutional actors in response to polices or regulation. A defining feature of the optimal control algorithms typically employed in these problems has been an underlying assumption of stationarity. However, many ecological systems can experience extremely long transient periods, making this assumption dubious.
Results/Conclusions
I demonstrate how non-stationarity due to long transients can lead typical optimal control methods to make bad policy decisions. I then explore the potential to avoid these outcomes using adaptive learning techniques that iteratively correct for this error, and contrast this approach with simple precautionary strategies. I illustrate these results using examples setting harvest quotas in marine fisheries.