Stikine Sockeye Salmon Management Model; Improving Management Uncertainty

Salmon fisheries that harvest Stikine River sockeye salmon (Oncorhynchus nerka) in marine waters in Southeast Alaska and in inriver Canadian fisheries are managed under the Pacific Salmon Treaty (PST). The historical Stikine Management Model (SMM) is the agreed to inseason forecast model. Using the SMM as the base model, six new forecast models were developed to estimate inseason the terminal run size of the Stikine River sockeye salmon stocks. In the first part, initial model selection, regression models utilizing various combinations of historical harvest and catch per unit (CPUE) data were developed. The dependent variable, terminal run size by stock, was based on postseason run reconstructions. Different models were evaluated based on the criteria of the lowest Akaike Information Criterion value. In the second part, inseason modeling, along with the historical SMM model, six forecast models were evaluated using weekly inseason data. Four of the models were second order polynomial regression models and two models were historical run timing models based on the inriver run size. Model predictions were averaged for an early and late season forecast. The inseason performance of the models was evaluated based on the percent error between the inseason estimate and the ‘true’ postseason estimate.