A Comparison of Estimates of First Nations Catches of Fraser River Sockeye Salmon from 1996 to 1999 by Scale-based Discriminant Function Models and Run Reconstruction Models

Historically, the Pacific Salmon Commission has used run reconstruction models to assign stock of origin to sockeye salmon (Onchorhynchus nerka) catches in the Fraser River First Nations fishery. Since 1987, scale samples have been collected from sockeye caught in the First Nations fishery to enable the Pacific Salmon Commission to make direct scale-based estimates of the catch by stock group. Scale parameters measured from the sockeye scales were used in linear discriminant function analyses (DFA) and the results of the analyses were applied to fishery catches to estimate the catch by stock group in the First Nations fishery. This report provides a comparison of catch by stock group in the Fraser River First Nations fishery from run reconstructions and DFA models for the years 1996 to 1999. Estimates from the scale-based DFA and the run reconstruction models were generally similar. However, in several cases catch estimates by stock group within years differed substantially (sometimes > 50%) between the two estimators. These differences in catch estimates between DFA and run reconstruction were likely attributable to: i) adverse sockeye migration conditions in the Fraser River in some years (i.e. high discharge and high water temperatures) that affected the relative vulnerability of individual sockeye stocks to First Nations fisheries,and ii) the inability of run reconstruction models to adjust catch estimates for these effects. The scale samples collected for the DFA analyses provided direct measures of the stock groups present in in-river fishery catches, consequently, this technique was sensitive to differences in stock vulnerability to fisheries attributable to the unusual migratory conditions. Scale-based DFA models should provide more accurate estimates of catch by stock group in the Fraser River First Nations fishery because of their ability to assess the variability in sockeye timing within and between years. Consequently, this technique is considered preferable to run reconstruction. Recommendations for improving the scale-based DFA approach include more thorough (increasing sample sizes and numbers of collections at designated sites) and consistent (better spatial and temporal coverage across years) sampling of First Nations catches of sockeye salmon in the Fraser River. Additionally, the number of sampling locations in the Fraser River should likely be increased to more accurately estimate catches by stock group as a result of the generally larger catch allocations of Fraser River sockeye to First Nations fisheries.