Reply: Scientific inquiry and fish stock assessment in the Canadian Department of Fisheries and OceansandReply: The interplay of policy, politics, and science

Abstract

The collapse of Atlantic cod was the unifying theme of Memorial University’s EcoResearch project “Sustainability in a cold ocean coastal environment.” The project entailed a broad-scale examination of the history, context, and consequences of that collapse. One objective was to analyze the role of science in fishery management decisions. The research question that became the title of our paper (Hutchings et al. 1997) was developed within that framework. We concluded that an environment in which information can be controlled, for whatever reason, is not one in which the scientific process can exercise its full potential. Our attention was drawn to the institutional structure that results in such an environment. After identifying what we perceived to be undesirable consequences of having science fully integrated within government, we posed the question, how should the structure be changed? Doubleday et al.’s (1997) comment is a welcome contribution to this examination. They make the valid point that the present stock assessment review process is more open to nonDFO employees than in the past. However, they charge that by misinterpreting and selectively quoting from Department of Fisheries and Oceans (DFO) documents, Hutchings et al. (1997) “give a false impression of stock assessments of Northern cod in the 1980s and 1990s.” Because of our comparatively minor treatment of northern cod stock assessments, Doubleday et al.’s point of departure is surprising and draws attention away from the main points of our perspective. We examined neither the structure of stock assessment models nor their ability to estimate stock size and fishing mortality, rendering superfluous the discussion of the Alverson and Harris reports by Doubleday et al. Doubleday et al. (1997) describe the stock assessment review process presently used by the DFO. We agree that the recently modified system has many desirable aspects. For example, the presence of fishers and non-DFO scientists at assessment meetings has been viewed in a positive light by all parties. However, the integration of outsiders in the assessment process is not nearly as complete as Doubleday et al. imply. First, assessment documents are not written in a form readily accessible to non-scientists. Second, in Atlantic Canada, involvement of non-DFO scientists can range from a full review of assessment documents received prior to assessment meetings (a good practice) to a simple response to general open invitations to attend meetings and provide comment (a considerably less effective practice). However, involvement by nonDFO employees in actual stock assessment data analysis, i.e., running and tuning Virtual Population Analyses (VPAs), is rare, if it occurs at all. Thus, in practice, the production of, and the ability to comprehensively review, stock assessment documents remains a DFO “in house” affair in many important respects. Doubleday et al. (1997) appear to have misunderstood our discussion of scientific concerns regarding northern cod in 1986. We focused on the convergence of conclusions among DFO and non-DFO scientists regarding both the overestimation of northern cod biomass and the underestimation of fishing mortality on this stock. We cited Winters (1986) because of the many uncertainties that he identified. Although scientific uncertainty with respect to stock size and fishing mortality prompted several recommendations by CAFSAC at the time, it was not until 1989 that these uncertainties were reflected by somewhat reduced catch quotas. Doubleday et al. (1997) do not explain why these significant recommendations were not enacted within the time frame specified by CAFSAC. Regarding the projected increases in northern cod spawning stock biomass between 1992 and 1994 (Doubleday et al. 1997), these were based on numbers-at-age estimates from VPA. Uncertainties in these numerical estimates were not quantified. In addition, the projections did not account for the well-known “retrospective” problem associated with VPA numerical estimates (Smith and Gavaris 1993; Myers and Cadigan