Language selection

Search

Research Document 2021/004

Walleye Pollock (Theragra chalcogramma) stock assessment for British Columbia in 2017

By Starr, P.J. and Haigh, R.

Abstract

A new stock assessment is presented for two British Columbia (BC) stocks of Walleye Pollock (Theragra chalcogramma, WAP), with the BC North stock encompassing the three most northerly Pacific Marine Fisheries Commission (PMFC) major areas (5C, 5D, 5E) and the BC South stock including the remaining four outside PMFC major areas (3C, 3D, 5A, 5B plus minor areas 12 & 20). These stock definitions were selected on the basis of a difference in observed mean weights, with BC North mean weights estimated near 1.0 kg/fish while the equivalent BC South mean weights averaged near 0.5 kg/fish. A delay-difference production model was used to assess each stock, using data from fishery-independent surveys, a CPUE series derived from commercial bottom trawl catch rates, and an annual mean weight series derived from unsorted commercial catch samples. Because there are no useable BC ageing data, we used survey age samples from the Gulf of Alaska (GoA) to specify growth for the BC North stock. The BC South stock proved more problematic, with the GoA growth model unable to fit the BC South observed mean weights, eventually requiring us to use a published WAP growth model from the Asian Sea of Okhotsk. Each stock assessment explored a range of plausible natural mortality values as well as a range of ages for the knife-edge selectivity assumption because the biomass indices and the mean weight data used in the delay-difference model were not informative for these parameters. The stock assessment was conducted in a Bayesian framework, where the best fit to the data was used as the starting point for a search across the joint posterior parameter distributions using the Monte Carlo Markov Chain (MCMC) procedure. Twelve runs were made for the BC North stock and 11 for the BC South stock, with each run consisting of 60 million MCMC iterations, sampling every 50,000th iteration, discarding the first 200 draws for burn-in, leaving 1,000 draws to comprise the posterior. Composite reference (model averaged) scenarios were used to represent each stock, with the model average for both stocks consisting of eight model runs selected on the basis of a subjective evaluation of the quality of the MCMC posterior. Each composite reference scenario included three values for instantaneous natural mortality (M=0.25, 0.30, 0.35) and covered two or three ages at which knife-edge recruitment (k) to the fishery occurred (k=3, 4 in BC North and including k=5 in BC South). The MCMC posteriors for the two composite scenarios were constructed by pooling the 1000 MCMC samples from each of the selected runs to give a posterior of 3,000 samples for BC North and 6,000 samples for BC South, thus giving equal weight to each run. The composite reference scenario was evaluated against historical reference points (HRPs) based on the reconstructed spawning biomass trajectory due to concerns about the stability of estimating B0 and B2017. The HRP Bavg, the average spawning biomass from 1967-2016, was used as a proxy for BMSY, and Bmin, the minimum spawning biomass from which it subsequently recovered to Bavg, was used in place of 0.4BMSY. Twice Bmin was used in place of 0.8BMSY. The average exploitation rate over the period 1967-2016 (uavg) was used in place of uMSY. The biomass at the beginning of 2017 for the model average BC North stock was evaluated as being primarily above the USR while the 2017 beginning year biomass for the BC South stock was evaluated as being entirely above the USR. For each stock, the assessment provides a decision table which evaluates the probability of the model average case staying above five reference points across a wide range of 22 constant catches. However, the paper warns that the probabilities in these decision tables should be viewed cautiously as the delay-difference model used in this stock assessment is not capable of making reliable multi-year projections because it has no latent age structure to inform predictions and the stock-recruitment function is poorly determined.

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: