Pacific Herring Operating Model Update

A purpose-built statistical catch-at-age model was presented with the following features: flexibility in modelling natural mortality (M), a method for integrating data from surface and dive surveys in estimation of the survey index, representation of timing of all fisheries throughout the year, inclusion of spawn-on-kelp (SOK) fisheries, and an age-composition likelihood function that captures correlation among ages. The new model is a spatially integrated statistical catch-at-age herring (SISCAH) model, however the spatial capabilities of the model were not used here. SISCAH was compared to the previous assessment model with a transition analysis, using data through 2022. Suitability of SISCAH as an assessment model for the five major Pacific Herring stocks was evaluated using common statistical metrics including goodness of fit, simulation self-tests, retrospective analyses, and sensitivity tests. Trends in biomass, depletion, and recruitment were evaluated for each major stock. Utility of SISCAH as an operating model was evaluated by: i) conditioning on the historical time series (1951-2022), ii) projecting the population over a 15 year time frame, iii) applying example precautionary approach (PA) compliant management procedures to the projected population, and iv) evaluating performance against existing conservation and biomass objectives using closed loop simulations. Density-dependent natural mortality was added to represent potential ecosystem impacts (e.g., depensatory predation) on Pacific Herring stocks. Evidence for depensation in natural mortality varied among regions, and was strongest in Haida Gwaii (HG), Central Coast (CC), and West Coast Vancouver Island (WCVI), and weak in Prince Rupert District (PRD) and Strait of Georgia (SOG). Model fit was improved by introducing a weighting method for combining surface and dive survey indices within a single year, rather than the previous practice of treating post-1987 surface observations as coming from the dive survey only. SISCAH and the previous model show similar time series trends in biomass and recruitment. SISCAH natural mortality estimates match the previous model for HG and WCVI but appears to be constrained at lower levels of natural mortality for PRD, CC and SOG. SOK removals were included in the new model, represented as using closed ponding methods although representation using open ponding methods is also possible in future versions. Including these removals had little effect on biomass trends since ponding mortality is generally low, but yield curves were sensitive to allocation of quota among fisheries. As an operating model SISCAH reproduced historical population trends and simulated future trends and observational data consistent with the historical observations. Example management procedure evaluations were presented using perfect information simulations. Density-dependent natural mortality has stock-specific effects, affecting estimates of long term average unfished biomass and thus reference points, and consequently the perception of stock status over time. For example, stocks which show stronger evidence of depensation, such as HG, have lower estimates of long-term average unfished biomass which corresponds to lower limit reference points estimates. Equilibrium yield curves were produced using 200-year simulations and were the basis for maximum sustainable yield (MSY)-based reference point calculations. These curves were highly sensitive to the allocation of catch between fisheries; an allocation based on the recent-most 10-years historical average was used in this analysis (excluding years without fishing). Estimated harvest rate at MSY may be used to guide for tuning management procedure evaluations, for compliance with the PA policy reporting requirements, and for evaluation of proposed best practices for forage species. A minimum 3 year cycle for management strategy evaluation (MSE) updates is recommended, unless new evidence reveals exceptional circumstances. A process for implementing the new assessment and operating model, updates to the MSE, and identification of exceptional circumstances should be developed in a phased approach in consultation with managers, First Nations and stakeholders. Environmental variability was modelled implicitly via natural mortality dynamics (i.e., implicit predation), recruitment variability, and inter-annual variability in natural mortality around the depensatory relationship to biomass; however, specific advice on the impacts of climate change and changes to ocean productivity were not addressed in the analysis. Future work should include examining alternative parametrizations of density-dependent natural mortality, survey catchabilities, and spatial structure.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.