A spatially disaggregated, length-based, age-structured population model of yellowfin tuna (Thunnus albacares) in the western and central Pacific Ocean

A spatially disaggregated, length-based, age-structured model for yellowfin tuna (Thunnus albacares) in the western and central Pacific Ocean is described. Catch, effort, length-frequency and tagging data stratified by quarter (for the period 1962−99), seven model regions and 16 fisheries are used in the analysis. The model structure includes quarterly recruitment in each region, 20 quarterly age classes, independent growth patterns for juveniles and adults, structural time-series variation in catchability for all non-longline fisheries, age-specific natural mortality, and age-specific movement among the model regions. Acceptable fits to each component data set comprising the log-likelihood function were obtained. The model results suggest that declines in recruitment, and as a consequence, population biomass, have occurred in recent years. Although not obviously related to overexploitation, the recruitment decline suggests that the productivity of the yellowfin tuna stock may currently be lower than it has been previously. Recent catch levels appear to have been maintained by increases in fishing mortality, possibly related to increased use of fish aggregation devices in the purse-seine fishery. A yield analysis indicates that average catches over the past three years may have slightly exceeded the maximum sustainable yield. The model results also reveal strong regional differences in the impact of fishing. Such heterogeneity in the fisheries and the impacts on them will need to be considered when future management measures are designed. Additional keywords: length-based model, statistical age-structured model, spatial model, stock assessment Yel ow fi n tuna popul ati on model John H am pon and D vdA. Fournier MF01049 Introduction Yellowfin tuna (Thunnus albacares) is an important component of the tuna fishery in the western and central Pacific Ocean (WCPO), with annual catches over the past decade averaging almost 400000 t (Lawson 2000). Catches are distributed over a wide area, from the Philippines and Indonesia in the west to Hawaii, Kiribati and French Polynesia in the east (Fig. 1). The majority of the catch is taken by industrial purse seiners (~60%) and longliners (~17%), and by a variety of small-scale fishing gears (gillnet, troll, handline and others) in the Philippines and eastern Indonesia (~20%). Yellowfin tuna is believed to constitute a discrete genetic stock in the WCPO (Ward et al. 1994). Previous assessments of yellowfin tuna in the WCPO have relied mainly on analyses of standardized longline catch-per-unit-effort (CPUE) and tagging data (Hampton et al. 1999). However, with formal international management arrangements for tuna fisheries in the WCPO currently being negotiated (Anon. 2000), there is a need for a more sophisticated assessment approach that can indicate the current status of the resource relative to defined biological reference points (Schnute and Richards 1998) and evaluate alternative harvest strategies. A model that sufficiently captures the complexity of the population dynamics and fisheries is required as a first step. In particular, because yellowfin tuna is captured at different ages by different gear types, such a model will need to incorporate age structure to be able to predict the stock response to changes in harvest levels by the different fisheries. Also, the wide distribution of the stock and fisheries coupled with spatial heterogeneity in yellowfin tuna fishery and population parameters (including movement patterns), requires treatment of the spatial distribution of the fisheries, the fish and fish movement. The MULTIFAN-CL model (Fournier et al. 1998) provides a good basis for a spatially disaggregated, agestructured model for yellowfin tuna. MULTIFAN-CL is age structured, with the age composition of catches being estimated from length-frequency samples. The model is spatially disaggregated, with the population and fisheries stratified into a number of regions within the overall stock range. Other features of the model, such as the treatment of process error in the fishing effort – fishing mortality 938 John Hampton and David A. Fournier relationship, also make it a suitable choice as a starting point for a yellowfin tuna stock assessment model. This paper describes the adaptation and extension of the MULTIFAN-CL model for the stock assessment of yellowfin tuna in the WCPO. The major extension of the model involves the incorporation of tagging data into the model to facilitate the estimation of age-dependent movement and natural mortality rates. Various adaptations of the model to deal with specific features of yellowfin tuna biology (e.g. non-von Bertalanffy growth of juveniles) and fisheries are described as a base-case example. A yield analysis is integrated into the model as an example of how the model results might be interpreted in a fishery management context.