Formulating quantitative methods to evaluate fishery-management systems: what fishery processes should be modelled and what trade-offs should be made?

• Published in: ICES journal of marine science 1999-12, Vol.56 (6), p.900-916
• Main Authors: Mcallister, M. K., Starr, P. J., Restrepo, V. R., Kirkwood, G. P.
• Format: Article
• Language: English
• Subjects: Bayesian; decision analysis; fishery management procedure; fishery- management system; Marine; Monte Carlo simulation; operating model; stock assessment; uncertainty
• ISSN: 1054-3139; 1095-9289
• DOI: 10.1006/jmsc.1999.0547

A wide variety of modelling approaches have been developed over the past few decades to provide quantitative advice to fishery managers about the potential consequences of proposed alternative management actions and to improve the design of fishery- management systems. This article surveys these approaches to identify the main differences among them in their application of decision theory, particularly in the methods used to account for uncertainty, and to identify model structures that need to be considered. Six interlinked model structures are identified: population dynamics, data collection, data analysis and stock assessment, setting harvest controls, the harvest decision process, and imperfect implementation. Their inclusion and design will depend on the institutional arrangements for fishery management. For example, some evaluations have included the last two structures to account for the flexibility of fishery managers in their making of annual harvest decisions and limitations on their ability to implement management policy precisely. Other evaluations have excluded them because they have evaluated “management procedures” whereby all participants would be bound to strictly follow a preset protocol for data collection, data analysis, and a harvest decision rule. In addition, this article identifies some trade-offs that are often made when modelling methods are formulated that can strongly affect model results and the advice given. Partly as a result of these trade-offs, analysts have recently been combining some of the originally divergent aspects of different modelling approaches. Examples of this include combining the use of Bayesian statistical methods with the use of alternative operating models (rather than only one) to account for both parameter and model uncertainty.