INFLUENCE OF AGE-SELECTIVE SURVEYS ON THE RELIABILITY OF STOCK SYNTHESIS ASSESSMENTS

We used Monte Carlo simulations to examine the reliability of assessments following a method of separable catch-at-age analysis (Methot's stock synthesis approach). Our simulations were based on the widow rockfish (Sebastes entomelas) fishery off the west coast of the continental United States. We designed the study to examine how the performance of the method depended upon the age-specific selectivity pattern of a survey of relative abundance. When survey selectivity increased asymptotically with age, stock synthesis estimates of final year biomass (and most other parameters) tended to be closest to the correct value. When the survey concentrated on young fish the estimates were least reliable. When the survey selectivity was ''dome-shaped,'' concentrating on intermediate-aged fish, estimates were somewhat less precise (and less accurate) than for the asymptotic pattern but much more precise than for the survey concentrating on young fish. Predictably, estimates of recruitment and biomass were better at a higher fishing mortality rate and when the magnitude of survey error was lower. Generally, mean errors in estimated biomass and recruitment were positive, while median errors were close to zero. This pattern is due to transformation bias associated with a lognormal error structure.