Search

Aspects of the potential application of multispecies/ecosystem modelling to advise the management of South African fisheries are discussed. In general, reliable predictive ability from such models is likely to be achieved sooner for top predators, because relatively fewer links need to be modelled. Accordingly, discussion concentrates on the deliberations of scientific workshops on modelling marine mammal – fisheries interactions held by two international marine mammal commissions during 2002. Five questions are posed, and some responses suggested, relating to the development of a framework for multispecies/ecosystem modelling to contribute to South African fisheries management: (1) should such models be used for testing or making decisions; (2) do they appreciably reduce uncertainties associated with single-species models; (3) are whole ecosystem or minimum realistic models more appropriate; (4) what computer software is best suited to implement such approaches; (5) what are the overall cost implications? Caution is expressed that general scientific acceptance of predictive reliability for such models (as required for their use for management) is unlikely in the short term, and will probably require considerable data collection and complex analysis at not insubstantial cost.Afr. J. mar. Sci. 26: 53–61

Ecosystem-orientated thinking is increasingly incorporated into fishery management. Given the complexity of ecosystem processes, there is a need to evaluate the tools used to steer this thinking critically. ECOPATH with ECOSIM (EwE), an aggregate system-modelling package, is currently the most widely employed approach to assess the ecosystem effects of fishing. The basic equations and assumptions, strengths and weaknesses, and the potential of this approach to contribute to practical fisheries management advice are reviewed. Strengths include the structured parameterization framework, the inclusion of a well-balanced level of conceptual realism, a novel representation of predator-prey interaction terms, and the inclusion of a Bayes-like approach (ECORANGER) to take account of the uncertainty associated with values for model inputs. Weaknesses in model structure include the constraining nature of the mass-balance assumption (of ECOPATH) for initiating projections, the questionable handling of life history responses such as compensatory changes in the natural mortality rates of marine mammals, overcompensatory stock-recruit relationships that result from default parameter settings, possible problems in extrapolating from the microscale to the macroscale, as well as some (not too far-reaching) mathematical inconsistencies in the underlying equations. There is a paucity of systematic and stepwise investigations into model behaviour and properties, and users are cautioned against non-critical use of the default settings. An important limitation related to the predominant use of EwE as a “black-box” modelling tool is that some users fail to consider a range of alternative interaction representations. As with all multispecies approaches, the major limitation in applying the EwE approach lies in the quality and quantity of available data. Current EwE applications generally do not adequately address uncertainty in data inputs and model structure. Prudent EwE applications that utilize good data and are based upon rigorous statistical analyses can complement the quantitative predictions of traditional single-species models. They could be particularly useful in some contexts if output in the form of probability distributions encompassing a range of likely ecosystem responses were to be coupled with attempts to extend Operational Management Procedure (OMP) approaches to fisheries management beyond the singlespecies level. In particular, such applications could serve as the operating models of the underlying dynamics that are used for computer simulation testing of OMPs.