Author: David R. BernardPublisher:
ISBN:
Category : Chinook salmon
Languages : en
Pages : 31
Book Description
An alternative to the current method of forecasting size of the annual terminal run of large (age 1.3-1.5, approximately >659 mm from mid eye to tail fork) Chinook salmon Oncorhynchus tshawytscha to the Taku River is demonstrated. Both the current method and the alternative are based on sibling relationships within brood years; however, the alternative is based on median forecasts from power functions with lognormal error. Bayesian analysis using the program WinBUGS was used to avoid bias in forecasted run size from measurement error, and to quantify conditional uncertainty in forecasts. Because of greater imprecision in estimates of annual run size in earlier years, only data from 1995 to 2013 were used to exercise both current and alternative methods. From jackknifed hindcasting the mean percent error for the current method was +13% and +7% for the alternative; the former had two negative predictions. A Bayesian forecast with the alternative method produced the posterior probability distribution for the estimated terminal run size in 2014 with median 24,440 and mean 25,980 large salmon. Management targets under the Pacific Salmon Treaty for Taku fisheries were described, and the forecast posterior probability distribution was used to calculate probabilities of meeting those targets in 2014. Relevance of forecasting with a truncated data series of recent years was discussed. Instructions were given on how to expand the WinBUGS code to produce a forecast for 2015 and for years beyond.