FITTING SURPLUS-PRODUCTION MODELS WITH MISSING CATCH DATA USING ASPIC: EVALUATION WITH SIMULATED DATA ON ATLANTIC BLUE MARLIN

SUMMARY Assessments of sailfish and marlins usually rely on application of surplus–production models, because size and age composition of catches are not known. However, even annual catches of these species are unreported in some Atlantic fisheries where they are taken primarily as bycatch. Past ICCAT billfish assessments have omitted the missing data, an ad hoc approach that reduces credibility of an assessment. In theory, if recent catch and effort data are available for fisheries lacking historical catch data, it should be possible to fit a surplus–production model by estimating historical catches from corresponding data on fishing effort. Enhancements to ASPIC (the computer program used for non-equilibrium surplus-production modeling in previous ICCAT assessments) for this task were evaluated on simulated fisheries data generated by a computer program constructed around life history characteristics of Atlantic blue marlin (Makaira nigricans). The resulting simulated population included sex, size, and age structure on a monthly basis; growth was sexually dimorphic, with females attaining larger asymptotic mean sizes; and size at age was variable. Annual recruitment was determined from spawning biomass with a Beverton–Holt stock-recruitment function, as modified by density-indep endent stochastic survival. The simulation model was used to generate sample time series of simulated catches and population abundance histories using several alternative assumptions about natural and fishing mortalities, the slope of the stock-recruitment relationship, and measurement error. The resulting time series of simulated catches and abundances were fitted with and without missing catches using ASPIC. Estimates of maximum sustainable yield (MSY) and the ratios of final-year stock biomass to biomass at MSY (B./B MSY ) and fishing mortality to fishing mortality at MSY (F./F MSY ) were compared to known values from the simulations; results with and without missing catches were contrasted. Results characterize biases most likely arising from dissimilarity of the strongly age-structured simulation model and the age-aggregated surplus–production model. Nonetheless, fitted values for B./B MSY and F./F MSY averaged very near the true values. ASPIC results obtained with missing catches were very similar to those based on complete catch data, but not surprisingly were somewhat more variable. RESUME