Evaluation of virgin recruitment profiling as a diagnostic for selectivity curve structure in integrated stock assessment models

摘要

Virgin recruitment (R-0), the equilibrium recruitment in the absence of fishing, is an often used parameter in fisheries stock assessment for scaling population size. We describe and evaluate the use of the R-0 likelihood component profile to diagnose selectivity misspecification, using simulation analysis for bigeye tuna in the eastern Pacific Ocean. The profile is evaluated under two types of selectivity misspecification: (1) misspecified shape and (2) misspecified temporal variation. The results indicate that length-composition data can provide substantial information on R-0 estimation when the model is correctly specified, but can substantially bias estimates of absolute abundance when selectivity is misspecified. Although contradictory profiles for length-composition and abundance index data result from selectivity misspecification, they may not be useful in determining which survey or fishery selectivity is misspecified. The R-0 profile selectivity diagnostic is based on the influence of composition data on absolute abundance. However, perhaps a more problematic and difficult to detect issue is the impact of length-composition data on biomass trends. The age-structured production model diagnostic could be applied to identify bias in both absolute biomass and biomass trend caused by age- or length-composition data in the presence of model misspecification.