Publication Abstract

CIE Review of BSAI and GOA Pacific Cod Stock Assessment Models.

Executive Summary

This CIE review covers the stock assessment models for the stocks of Pacific cod in the Eastern Bering Sea and Gulf of Alaska. This reviewer was impressed by the amount and quality of work carried out on the assessment models for these stocks, and by the high standard of data collection and analyses that underpin these models. The best scientific information available appears to have been used throughout in the stock assessment models. Recommendations cover a number of areas, as stipulated in the Terms of Reference. These include ongoing support for the age-determination work and any work that would advance the Nichol temperature data, if practicable, to help better model annual variation in survey catchability. The current use of the Jensen-based estimates of natural mortality, and external estimation of both between-individual variability in size-at-age and input sample sizes for size and age composition data are supported, as are the use of the Nichol

 

et al. study to tune survey catchability, the practice of fixing at least one fishery selectivity to be asymptotic, and the partitioning of data to allow for data features and limitations in SS functionality. Survey data are a key input to the assessment models, and should remain so, with fishery CPUE included (but not fitted) as a useful and independent comparison. Areas that could be improved include modelling time-varying selectivity with a constrained random walk over time instead of in time blocks, using a more flexible growth curve than the current version of the Richards curve (e.g. for EBS cod), and incorporating bi-modal selectivity as an option (e.g. for GOA cod). Furthermore, a coarser bin structure for the larger sizes that does not lead to a significant loss of information could be explored, as could the possibility of fixing σR to a sensible value instead of pursuing a time-consuming iterative procedure. Areas of concern include the statistical treatment in the overall likelihood of non-independent data, the internal estimation of the extent of ageing bias (is this possible?), and the “perfect” model fit to the GOA sub-27 survey data, indicating over-parameterisation. There are also concerns about whether the year-to-year changes in model structure, which should reduce somewhat as models mature, remain justified.

 

 

http://www.fakr.noaa.gov/npfmc/PDFdocuments/catch_shares/Pcod/PcodModelsRev411Oliveria.pdf

J.A.A. De Oliveira*

Individual CIE Report, Alaska Fisheries Science Center, Seattle, 14-18 March 2011. Prepared for the Center of Independent Experts. 26pp