When undertaking model-based fisheries stock assessments, the approaches that are adopted must be applicable to the type(s) and quality of data that are available. Non-equilibrium models (e.g. biomass-dynamic and statistical catch-at-age models) are recommended whenever sufficient, reliable data exist. For many fisheries with limited data, however, the only models suitable for assessment are those assuming the population is at equilibrium with respect to processes such as mortality and recruitment (e.g. equilibrium surplus production models, catch curve and per-recruit analyses). For fisheries that lack both a reliable time series of (total) catches and abundance indices, model-based assessments typically apply catch curve analyses, and often combined with simple equilibrium population models (e.g. per recruit analysis). That is, catch curve analyses are typically applied to age and/or length frequency data obtain an estimate of total mortality, which is then related directly to natural mortality or used in further analyses to provide a relative measure of yield or biomass.
Traditional forms of catch curve analysis (e.g. linear catch curve analysis or the Chapman & Robson method) make the above strong equilibrium assumptions, and other “simplifying” assumptions, such as knife-edge selectivity, all of which are unlikely to be true. It is possible, however, to avoid several of these assumptions, but at the cost of increased model complexity (i.e. requiring more parameters and potentially more data). To what extent can the considerable limitations of these analyses be overcome, and is the increased complexity associated with the required modifications always necessary, or even advisable? What else might be done to improve catch curve-based assessments for such data limited fisheries? With these questions in mind and using data for demersal finfish species in Western Australia, this presentation will describe the approaches we have used when attempting to address these issues.