Value of data in stock assessment models with misspecified initial abundance and fishery selectivity

Abstract

The age-structured assessment model available in the MSEtool R package assesses stock status and exploitation for varying data availability, from limited to rich datasets. We investigated model accuracy in relation to data availability, population exploitation levels, initial population assumption and fishery selectivity misspecification. Estimates were accurate in all conditions when data were available for a stock in an unfished state. However, for estimates to be accurate without complete exploitation data, total catch and abundance index data needed to span more than two stock generations. When the data time series was shorter than two generations, fishery mean lengths spanning one generation improved relative estimates (e.g. depletion), but precise estimates of unfished recruitment required fishery age- or length-structured data.