Language selection

Search

Research Document - 2006/064

Catch-age models for Pacific herring: Evaluation of alternative assumptions about fishery and stock dynamics and alternative error distributions

By Haist, V., and J. Schweigert

Abstract

The herring stock assessment catch-age model (EASM) was designed for Pacific herring and includes assumptions unique to those stocks. A new catch-age model (NASM) was recently developed for a B.C. herring objective based fisheries management evaluation. The two models differ in some of their assumptions about the fisheries and population dynamics, and it is not clear which should be used in the annual stock assessment process.

We present another catch-age model, HCAM, which incorporates the structure and assumptions of both the EASM and the NASM, with the purpose of determining which assumptions of the EASM and NASM models result in better performance. To the extent possible, objective criteria are used to assess model performance.

Both the EASM and NASM models over-weight the data (that is, residuals tended to be over-dispersed relative to the error assumed for individual data components) and both models have strong age-related patterns in the residuals from fitting the age composition data. The EASM analyses have strong retrospective bias in abundance estimates; for the NASM-like analyses the retrospective patterns are not as large and relatively unbiased.

We evaluated numerous implementations of the HCAM, attempting to minimize the areas of concern with the EASM and the NASM implementations. An implementation of this model was developed that we believe has better performance than either the EASM or the NASM, although it incorporates aspects of both those models. Diagnostics from the HCAM implementation that we believe indicate better performance are; a reduction in the magnitude of the retrospective pattern; a reduction in the magnitude of the age-related pattern in age-composition residuals, and better coherence between the assumed and empirical estimates of the lack of model fit to the data.

Accessibility Notice

This document is available in PDF format. If the document is not accessible to you, please contact the Secretariat to obtain another appropriate format, such as regular print, large print, Braille or audio version.

Date modified: