Canary Rockfish (Sebastes pinniger) stock assessment for British Columbia in 2022

Canary Rockfish (Sebastes pinniger, CAR) ranges from the Gulf of Alaska southward to northern Baja California. In British Columbia (BC), the apparent area of highest concentration occurs on the west coast of Vancouver Island and at the heads of the three gullies in Queen Charlotte Sound. This species occurs along the west coast of Graham Island and in the western sections of Dixon Entrance, but the apparent abundance is lower. In 2007, the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) assessed the coastal population of CAR in British Columbia as ‘Threatened’, based on an analysis of survey indices and the threat from commercial fishing. As a result, the species was considered for legal listing under the Species at Risk Act (SARA). A 2007 stock assessment (also acting as a recovery potential assessment) by Stanley et al. (2009) estimated that CAR was in the ‘Cautious Zone’ in the DFO Sustainable Fisheries Framework (DFO 2009a) but an update to that assessment conducted in 2009 concluded that CAR was in the ‘Healthy Zone’ when using a credible steepness value. In 2011, a decision was made not to list Canary Rockfish under Schedule 1 of the SARA. In 2019, Bill-C-68 was enacted to amend the Fisheries Act with the Fish Stocks provisions, prompting a national review of the approximately 180 stocks using Sustainability Surveys with the aim to include the majority of those stocks in regulation over the next five years. Canary Rockfish is one of 18 groundfish stocks in the Pacific Region being considered for inclusion. The purpose of this CAR stock assessment is to evaluate the current stock status and provide advice suitable for input to a sustainable fisheries management plan. This stock assessment evaluated a BC coastwide population harvested by two fisheries: a combined bottom and midwater trawl fishery accounting for over 95% of the catch and an ‘other’ fishery which combined a range of capture methods but was mostly longline. Midwater trawl catches of CAR were combined with bottom trawl for the purposes of this stock assessment. Analyses of biology and distribution did not support separate regional stocks for CAR. A single coastwide stock was also assumed by Stanley et al. (2009) and the subsequent update. The assessment used an annual catch-at-age model tuned to six fishery-independent trawl survey series, a bottom trawl catch per unit effort (CPUE) series, annual estimates of commercial catch since 1935, and age composition data from survey series (23 years of data from three surveys) and the commercial fishery (37 years of data). The model started from an assumed equilibrium state in 1935, the survey data covered the period 1967 to 2021 (although not all years were represented) and the CPUE series provided an annual index from 1996 to 2021. A two-sex model, which estimated M for each sex and the stock-recruitment steepness parameter, 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 1950–2012, and selectivity for the three surveys with age frequency (AF) data and the commercial trawl fleet. The survey and CPUE scaling coefficients (q) were determined analytically. Fourteen sensitivity analyses evaluated with MCMC were conducted relative to the base run to test the effect of alternative model assumptions. A further three runs were made with an environment Pacific decadal oscillation (PDO) index series to evaluate the effect of this series on the estimated recruitment trajectory. These models were also evaluated with MCMC. The base run estimated the CAR spawning population biomass at the end of 2022 (median with 0.05 and 0.95 quantiles) to be 0.78 (0.57, 1.0) relative to B0 and to be 3.0 (1.9, 4.9) relative to BMSY. This latter result suggested that the 2023 CAR spawning population lay well in the Healthy zone (with a probability >0.99). Projections predicted that the stock will remain in the Healthy zone up to the end of 2032 at all evaluated catch levels up to 2,000 t/y. Older female CAR were absent from the AF data (females older than age 40 were rare) while the male CAR AF data extended to above age 60. The previous CAR stock assessment assumed a fixed M = 0.06 for all males and for females up to age 13; females age 14 and older had M = 0.12. This stock assessment approached this problem in three ways: 1) estimating a separate M for males and females to get the best fit to the AF data; 2) estimating separate M values for males and females up to age 13 and then estimating new M values for both sexes from age 14 and higher; 3) while estimating single M values for each age, allowing the female selectivity to the commercial fishery and for five of the six surveys to decline with older ages, creating a cryptic population of female spawners. This assessment found that all three models could fit the data credibly, with the first option being the most parsimonious. This run was selected as the base run. The other two options were more optimistic relative to B0 and BMSY than was the base run. The median estimates by the 14 sensitivity runs for B2023/B0 ranged from 0.62 to 0.97 and for B2023/BMSY ranged from 2.40 to 3.22, indicating that all 14 sensitivity runs lay well in the Healthy zone. These analyses included, higher and lower pre-1996 catch histories, higher and lower recruitment standard deviation (σR) assumptions, adding the two hard bottom long line (HBLL) survey series, dropping the CPUE series, substituting an alternative CPUE series, omitting ageing error, adding AF data from the Hecate Strait (HS) and west coast Haida Gwaii (WCHG) synoptic surveys, and using two alternative ageing error functions in addition to the three hypotheses for female natural mortality described above. Incorporating the environmental PDO index series into the stock assessment resulted in an unsatisfactory conclusion: the degree to which the index series was able to influence the recruitment pattern was dependent on the weight given to the index series. Choosing the weight was arbitrary and higher weights resulted in a deterioration of the fit to the fishery data. This procedure is effectively a correlation analysis because there is no functional link between the index series with the population dynamics.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.