This working paper includes pre-season 1999 stock size forecasts for nine sockeye, five pink, and five chum salmon stocks or stock groupings in northern British Columbia, statistical areas 1-10. The recommended forecasts are based on simple models that have been evaluated in a previous working paper.
The recent 5-yr mean model is a simple time-series approach that effectively accommodates gradual changes (autocorrelated anomalies) in productivity. For northern populations of sockeye salmon, this model has performed as well, or better than other models because variations in the independent variables used by other models have been small, and their effects have been obscured by other factors. However, for Skeena River sockeye, the 5-yr mean model should be rejected in favour of the “smolt” or “sibling age-class” models that include the effect of measured, record low smolt production in two consecutive years. Although the sibling model has not performed especially well in retrospective analyses under typical conditions, it’s use is recommended for 1999 to capture the combined effects of poor freshwater production for the 1994 and 1995 brood years as well as apparent poor marine survival for the 1994 brood year. This recommendation is also consistent with the precautionary approach to fisheries management because the sibling model predicts the lowest stock size in 1999.