Research Document 2025/004
Pacific Ocean Perch (Sebastes alutus) Stock Assessment for British Columbia in 2023
By Starr, P.J., and Haigh, R.
Abstract
Pacific Ocean Perch (Sebastes alutus, POP) ranges from Honshu in Japan to Baja California in Mexico. In British Columbia (BC), POP occurrence is almost continuous, with apparent breaks in upper Hecate Strait (probably too shallow) and off the SW coast of Haida Gwaii (steep terrain unsuitable for trawling). Hotspots (high catch per unit effort [CPUE] density) occur in Moresby Gully, around Anthony Island, off Rennell Sound, off the NW coast of Haida Gwaii, and in Dixon Entrance near Langara Island.
This stock assessment evaluated a BC coastwide population with three subareas, each with separate fisheries in 5ABC (Queen Charlotte Sound), 3CD (west coast Vancouver Island), and 5DE (west coast Haida Gwaii). The fisheries were dominated by trawl gear (>99%) with minor removals by other gear types (e.g., longline). Midwater trawl catches of POP were most prevalent in 3CD, but only after 2007. Midwater trawl activity in 5ABC was moderate to low, and in 5DE was minimal. For this stock assessment, bottom and midwater trawl records were combined.
The assessment used an annual catch-at-age model tuned to six fishery-independent trawl survey series, annual estimates of commercial catch since 1935, and age composition data from survey series (29 years of data from five surveys) and the commercial fishery (43 years of data from three fisheries). The model started from an assumed equilibrium state in 1935; the survey index data covered the period 1967 to 2022 (although not all years were represented).
A two-sex model, which estimated M for each sex and the stock-recruitment steepness parameter h, was implemented in a Bayesian framework using the Markov Chain Monte Carlo (MCMC) ‘No U-Turn Sampling’ (NUTS) procedure. In addition to natural mortality and steepness, the parameters estimated by this model included average recruitment over the period 1935–2014, recruitment distribution parameters to allocate coastwide recruitment, and selectivity for the 5ABC commercial fleet (shared with 3CD and 5DE) and for the five surveys using age frequency (AF) data. The survey scaling coefficients (q) were determined analytically. Single-area models assuming three independent stocks were also run to compare with the subarea results from the multi-area base run model. Ten primary sensitivity analyses, evaluated with ‘Markov Chain Monte Carlo’, were conducted relative to the base run to test the effect of alternative model assumptions. A further three sensitivity runs were made only to the level of the ‘mode of the posterior distribution’ (MPD) because the results differed little from the base run.
The base run estimated the POP spawning population biomass at the start of 2024 (median with 0.05 and 0.95 quantiles) to be 0.58 (0.42, 0.81) relative to B0 and 2.3 (1.4, 3.9) relative to BMSY. This latter result suggested that the 2024 POP spawning population was positioned well in the Healthy zone coastwide and by subarea.
The median MCMC estimates by the 10 primary sensitivity runs for B2024/B0 ranged from 0.54 to 0.64 and for B2023/BMSY ranged from 2.08 to 2.53, indicating that all 10 sensitivity runs lay well in the Healthy zone. These analyses included: parameterising the weighting of AFs, higher and lower pre-1996 catch histories, higher and lower recruitment standard deviation (σR) assumptions, omitting ageing error, and using two alternative ageing error vectors.
The greatest uncertainty in this stock assessment was the relative size of the three subareas, an issue that was demonstrated by varying the choice of the recruitment reference area in two of the sensitivity runs compared to the base run. This uncertainty centred on the size of the 3CD subarea, which varied between 14% and 22% of the total B0 biomass over the two sensitivity runs and the base run. The larger 5ABC subarea varied between 52% and 60% of the total B0 biomass for the same three models while the 5DE biomass was relatively constant near 20% of the total B0 biomass over all three runs. The total B0 biomass was similar for the two sensitivity runs and the base run. This result implies that, while the overall yield from this population is reasonably well understood, the 3CD subarea should be managed with caution in conjunction with the other two subareas.
The impact of environmental covariates was not modelled in this stock assessment because of inconclusive results obtained in previous stock assessments (POP in 2017, Canary Rockfish in 2022). Instead, a projection run was made after arbitrarily reducing the mean recruitment by 50% to represent a “worst case” scenario in terms of recruitment over the next 10 years.
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: