How does the RMP work? Effectively, it is a two phase procedure. The first phase, which has been extensively tested and was reviewed in detail by the NMFS peer panel, consists of a series of simulation trials of the proposed management rule _ the Catch Limit Algorithm (CLA). This rule produces a catch quota from three pieces of information; the catch history of the whale stock, and the estimated mean and variance in abundance of the stock from all surveys conducted on it. Uncertainty is the keyword for the CLA. The estimate of uncertainty is used to set catch limits. It does this in an extremely risk-averse fashion, especially when it comes to stock protection. The CLA uses the requisite data to revise its range of uncertainty about the productive potential of the whale stock and the estimation accuracy (bias) of its abundance estimate. It uses a set of parameters, including the protection level and equilibrium target level (both defined as a fraction of pristine stock levels) to set harvest quotas. The CLA is extremely conservative in that:
The first phase of the RMP was tested with an exhaustive series of 100 year simulation trials, using a more detailed population dynamics model than that on which the CLA is based. A series of "robustness" trials varied parameters to simulate a wide range of plausible alternative scenarios to the base cases (e.g., reduced stocks due to global warming, inaccurate reporting of historical catches).
Three measures of performance evaluation -- relating to stock protection, stablity of harvest and maintenance of harvest yield -- were considered, though, in my opinion, stock protection was weighted most highly. The model used for validation is a generic model. It represents not a specific, but a typical whale stock. Numbers are reckoned relative to pristine stock levels, not actual abundance values. This is a new approach to resource management. We try the shoe on by simulating its attributes and checking out how it fits and, more importantly, how it wears. The NMFS peer panel was mostly impressed with the approach and the robustness trials done. But where is the science?
The science comes into the second phase of the RMP, the so-called implementation phase. Here push comes to shove, with respect to a specific whale stock. Rather than a generic whale stock the model in this phase is based on the actual stocks in a region. The spatial interrelationship of the stocks translates into how the CLA quota is to be regionally allocated. Actual harvest and survey data are to be used, but it is the same CLA as in the first phase. In the examples given in the IWC reports (concerning Antarctic and North Atlantic minke whale stocks), initial estimates of the spatial distribution of abundance were given. Presumably these estimates will be refined if and when actual stock quotas are to be set. If data on multispecies or multistock interactions or interactions of the stocks with environmental variables are available, these should be included in the model _ as recommended by the NMFS peer review panel. Additional robustness trials, some on combinations of parameters changes over plausible ranges for the stocks in question, should be done as part of the implementation phase, to make it the best scientific assessment of stock biology and abundance assessment possible. For example, the NMFS peer panel suggested development of some promising methods, such as genetic techniques for stock identification to best separate existing stocks. The NMFS peer panel did not consider the moral or political aspects of whaling. We had not collaborated with each other or with members of the RMP team. The fact that our consensus agreement supported use of the current CLAs, with reassessment after 20 years, providing that additional robustness trials are run at the implementation phase, speaks to the fine work done by the RMP team. I believe we should move on from here to implement the current CLA procedure, using the best scientific evidence on those stocks where the situation merits it and not continue testing it further using a generic model. If a better CLA procedure is proposed in the future, replace the current procedure with the new one. This is in the spirit of how the algorithm was developed in the first place - as the percieved best of 5 independently developed candidate algorithms. If the CLA doesn't work with a particular species (e.g., allows harvest when the population should be protected) with the best scientificially most defensible model (based on available data and scientific evidence to support it) in the stock implementation phase then we should reassess using the algorithm for that species.
What of the application of the CLA and its simulation testing to fisheries? There are striking differences between fish and baleen whales, including variable year to year recruitment for fish, and large differences between fish (even marine fish) in life history strategies and population dynamics. These differences preclude the use of a generic model for fish stocks. Different algorithms will have to be developed for different regions and must consider such factors as the bycatch of some either protected or undesired species by fishing targeted on another species. For this reason the CLA may have to have different rules for different situations, the rules representing the objectives and constraints for management and the fishing industry as well as the biology of the fish stocks involved. However, the strength of such a procedure is that forming a CLA will involve setting a series of quota setting rules that both management and industry can buy into. In conjunction with policies that vest fishers in the future of the stocks they harvest, such a Individual Transferable Quotas (ITQ) the CLA approach can foster long term harvest and marketing strategies which will lend greater stability and sustainability to the stocks harvested.