An Investigation of Ageing Requirements to Support the British Columbia Sablefish (Anoplopoma fimbria) Operating Model

Abstract

Since the inception of the Management Strategy Evaluation (MSE) process in 2011 (Cox et al. 2011), the Sablefsh operating model (OM) has informed the British Columbia (BC) fshery’s harvest strategy approach by assessing the stock against reference points and simulating future stock dynamics and fshery responses. Originally, the MSE process was initiated to address the decline of the Sablefsh spawning stock after the 1980s. Despite implementing management changes in 2011, the resulting recovery was slow, and it wasn’t until high recruitments began in 2016-2018 that the spawning stock recovery was assured. These 1 - 3 record high year classes moved the stock above optimal biomass levels in a short period, and by 2022 the spawning stock was estimated to be 1.32 times the female spawning biomass at maximum sustainable yield, BMSY (DFO 2023a). Current scientifc challenges to the Sablefsh MSE process pivot on ensuring the OM can continue to provide reasonable estimates of stock status in the presence of record high recruitment from the 2016-2018 year classes (DFO 2023a). Accurately estimating abnormally large and small recruitments is critical and carries asymmetric risks because overestimation or underestimation can lead to long-term stock status risks or economic setbacks, respectively. OM estimates of recruitment strength rely on age composition data from research surveys and commercial fsheries, as does the estimation of feet-specifc gear selectivity parameters that characterize the age and size at which fsh are caught. However, recent reductions in sample sizes of age composition data, especially from commercial sources, have increased reliance on less direct data via Bayesian priors on selectivity parameters developed from a mark-recapture tagging program (Cox et al. 2023). It is therefore essential to evaluate how sensitive the OM is to the quantity of age data and to strategically allocate ageing resources among survey and fshery sample sources. Proper age data usage is key to managing the effect of recruitment uncertainty in stock assessments, thereby ensuring that operating models are able to (a) provide timely estimates of stock status that are relatively robust to multiple sources of uncertainty, and (b) simulate realistic stock and fshery monitoring data for testing management procedure performance. This Science Response process was initiated by DFO Science in the Pacifc Region to increase understanding of how age composition sample size and the allocation of ageing effort among age data sources affects Sablefsh operating model performance with respect to bias and precision of management parameters. This evaluation is undertaken in the context of the full Sablefsh management system by using closed-loop feedback projections that apply the current Sablefsh management procedure (MP) to simulated data each year to determine annual catch levels. Establishing a scientifcally defensible rationale for both the number of ages and allocation of ages to data sources will inform the annual Sablefsh ageing request. While these analyses are specifc to Sablefsh, the framework developed is relevant to stock assessment programs for multiple species groups that face similar challenges due to limited ageing capacity. Overall, results show that, as expected, increased age sample sizes produce lower parameter uncertainty in selectivity estimates, especially for age compositions derived from otolith sampling during the annual Sablefsh research survey. However, under a scenario where the stock collapses due to a persistent recruitment failure, the sample size and source of Sablefsh age composition data does not appear to have a strong infuence on detecting spawning stock biomass or status with respect to the limit reference point (LRP). We discuss options for improving the response of the Sablefsh management system to persistent low recruitment in the Conclusions section. Inability to detect persistent recruitment failures notwithstanding, results under average recruitment conditions suggest that sample sizes in the range of 1200 - 1600 ages / year are suffcient to support the current Sablefsh operating model. Bias and precision for a subset of key model parameters have little improvement above 1600 total samples, and deteriorate at a faster rate when total sample size drops below 1200 aged fsh. The identifed lower limit of 1200 samples / year aligns with recent samples sizes, while the upper value of 1600 samples / year is similar to sample sizes from the years 2010-2016. This Science Response Report results from the regional peer review of March 25, 2024 on the Effects of Age Composition Data on Operating Model Performance for Sablefsh (Anoplopoma fmbira) in British Columbia.