Age sample sizes and their effects on growth estimation and stock assessment outputs: Three case studies from US West Coast fisheries

Age estimates are often used as conditional age-at-length (CAAL) data in stock assessment models to internally estimate growth and other life history parameters of fish populations. Obtaining sufficient numbers of age estimates is not a trivial task, as it requires considerable sampling effort, the development of aging criteria, the validation of age observations, and, finally, the production aging of sufficient numbers of age structures to inform the assessment models. In this study, we evaluated the influence of CAAL data in several assessment models of West Coast groundfish, by both randomly selecting only portions of available CAAL data and by selecting intermittent numbers of years of CAAL data for inclusion in models. These sensitivity tests were conducted on recent stock assessment models for Blackgill Rockfish (Sebastes melanostomus), Bocaccio (Sebastes paucispinis), and Pacific Sanddab (Citharichthys sordidus), as these three species represent a fairly wide range of life history types. The results showed that for the Bocaccio model, the assessment outputs from model runs with reduced numbers of CAAL data were very comparable with the model run with all available data. However, when CAAL data were reduced, the Blackgill and Sanddab models estimated different spawning outputs, stock depletions, and growth functions. Reducing CAAL data in these two models also resulted in larger estimation uncertainties in the assessment outputs. Overall, our results suggest that it is important to obtain more age data from long-lived species over as long a time span as possible, and aging samples from intermittent years may be considered if there are constraints in time or effort to age all available samples. Published by Elsevier B.V.