Designing Regulatory Systems That Are Robust to Uncertainties Caused by Changes in Production and Observation Parameters

Decision rules for harvest management need to account for possible persistent changes in stock-recruitment parameters.  Changes in recruitment carrying capacity do not imply changes in optimum fishing mortality rates, but changes in density-independent mortality rates of juvenile fish (strength of recruitment compensation) do imply need for fishing rate adjustments.  Stochastic dynamic programming shows that the best way to respond to such changes depends on how quickly the changes occur.  In particular, relatively slow and/or persistent changes result in the optimum policy being to choose that fishing mortaltiy rate and/or spawning stock goal that is optimum with respect to current productivity, even when it its known that productivity will continue to change but with unknown future direction.  Rapid changes on time scales of 1-5 years should generally be treated the same as random, uncorrelated recruitment noise, through use of a stationary decision rule of either fixed escapement or fixed fishing mortality rate form.  A critical need in monitoring programs to better respond to productivity changes is direct monitoring of current fishing mortality rates, though methods such as tagging and spatial analysis of area swept by fishing relative to fish distributions.