Assessing the impacts of environmental and ecological variables on the performance of Fraser sockeye salmon forecast

Abstract

The Canadian Fraser River sockeye salmon ( Oncorhynchus nerka) is one of the largest stock complexes in North America, supporting major commercial, recreational, and First Nations fisheries. Sockeye fisheries management relies on an annual pre-season forecast of adult returns. In this study, we developed a comprehensive framework for visualization with a Taylor diagram to evaluate pre-season forecast models annually and identify external drivers important for forecasting sockeye returns. Specifically, we incorporated five new covariates, including sea surface temperature in the Gulf of Alaska and the abundance of salmon species, into the existing forecast models. Results revealed better performances by both the Ricker and Power models when coupled with the newly included covariates. Moreover, models selected over a decade ago underperformed compared to those selected based on our recent retrospective analysis from 2009 to 2020. We advocate for continuous evaluation of forecast models in the face of environmental change, emphasizing the necessity of developing models that incorporate non-stationary processes and assess the impacts of environmental and ecological factors on Fraser sockeye dynamics.