Expected future performance of salmon abundance forecast models with varying complexity

Abstract

We evaluated the scope for improving abundance forecasts for fishery management using Sacramento River fall Chinook salmon (Oncorhynchus tshawytscha) as a case study. A range of forecast models that related the Sacramento Index (SI; an index of adult ocean abundance) to jack (estimated age 2) spawning escapement the previous year were considered. Alternative models incorporated effects of density dependence, local environmental conditions, the abundance of the previous cohort, and trends or autocorrelation in the jack-to-SI relationship. Forecast performance was assessed in terms of bias, accuracy, ability to track trends in the SI, and management objectives. Several models achieved higher accuracy than the model used for management, but no single model performed best across all criteria, and substantial forecast error remained across all approaches considered. Environmental models generally performed better than the management model, but there were differences in the relative importance of individual environmental variables over time and among model formulations. Accounting for model selection uncertainty in environmental models decreased their forecast performance. Simpler models often had similar or better performance than environmental models. In particular, the model incorporating temporally autocorrelated errors demonstrated potential for modest forecast improvement with relatively little additional model complexity.Résumé : Nous avons évalué le potentiel d'amélioration des prévisions d'abondance pour la gestion des pêches en utilisant le saumon quinnat (Oncorhynchus tshawytscha) d'automne du fleuve Sacramento comme étude de cas. Divers modèles de prévision reliant l'indice Sacramento (SI; un indice de l'abondance océanique des adultes) à l'échappée de géniteurs unibermarins (âge estimé de 2 ans) l'année précédente ont été examinés. Les modèles incorporaient, selon le cas, les effets de la dépendance sur la densité, des conditions ambiantes locales, de l'abondance de la cohorte précédente et des tendances ou de l'autocorrélation du lien unibermarins-SI. La performance de prévision a été évaluée en termes des biais, de l'exactitude, de la capacité de faire ressortir des tendances du SI et des objectifs de gestion. Plusieurs modèles ont produit une plus grande exactitude que le modèle utilisé pour la gestion, mais aucun modèle n'a surclassé les autres pour tous les critères, et des erreurs de prévision importantes subsistaient pour toutes les approches examinées. Les modèles environnementaux donnaient généralement de meilleurs résul-tats que le modèle de gestion, mais il y avait des différences en ce qui concerne l'importance relative de variables environnementales précises dans le temps et entre les formulations de modèle. La prise en compte de l'incertitude associée au choix du modèle dans les modèles environnementaux réduisait leur performance de prévision. Dans bien des cas, les modèles plus simples performaient aussi bien ou mieux que les modèles environnementaux. Plus particulièrement, le modèle incorpo...

Beyond Biomass: Valuing Genetic Diversity in Natural Resource Management

“…Welfare in the model stems from harvest profits. To specify the harvest exploitation rule, we start with the Pacific Fishery Management Council's plan for Sacramento River Fall Chinook (SRFC) as summarized by Winship et al (). We make two adjustments to this rule in our model.…”

“…Second, we smooth the harvest rate function. The rule specified in Winship et al () is based on a constant escapement approach but includes modifications that introduce several discontinuities. To avoid potentially erratic features in the value and policy functions, we use a smooth approximation to the rule.…”

Role of maturation and mortality in portfolio effects and climate resilience

“…To simulate forecast error, we obtained records of recent forecasts versus postseason abundance estimates from fishery management documents (PFMC 2021b) and reconstructed what the current forecast approach would have predicted for earlier years (back to 1995) with data available at the time (Winship et al 2015). The logged ratio between the forecast and the postseason abundance estimate was well described by a lognormal distribution (Satterthwaite and Shelton 2022) with log-scale mean 0.132 and log-scale standard deviation 0.486 (Fig.…”

Implementing Ecosystem-Based Management Principles in the Design of a Salmon Ocean Ecology Program

“…For example, a conceptual model of the system has been developed for Chinook salmon in Central California (Wells et al, 2016; Figure 3). Winship et al (2015) explored the utility of incorporating metrics of environmental conditions identified in that framework into annual abundance forecasts made for Sacramento River Fall Chinook, typically the predominant contributor to ocean salmon fisheries off California and much of Oregon. Many of the indicators highlighted in that conceptual model are currently reported to fisheries managers in annual reports to the regional management council (i.e., PFMC); this is done to improve decisionmaker and stakeholder understanding of the status and trends of key physical and biological indicators throughout the CCE (Levin et al, 2009;Pacific Fisheries Management Council [PFMC], 2013; California Current Integrated Ecosystem Assessment [CCIEA], 2020).…”