| Summary: | Dissertation (Ph.D.) University of Alaska Fairbanks, 1999 Recruitment estimates for Pacific herring, Clupea pallasi, populations in the Bering Sea and Northeast Pacific Ocean are highly variable, difficult to forecast, and crucial for determining optimum harvest levels. Age-structured population models for annual stock assessments of the sac-roe fisheries rely on fishery and survey age composition data tuned to an auxiliary survey of total biomass. In Chapter 1, the first age-structured model for Norton Sound herring was developed similarly to existing models. Estimates of variability from age-structured stock assessment models for Pacific herring are often not calculated. In Chapter 2, a parametric bootstrap procedure using a fit of the Dirichlet distribution to observed age composition data was developed as a quick and easy method for computing error estimates of model estimates. This bootstrap technique was able to capture variability beyond that of the multinomial distribution. This technique can provide estimates of variability for existing population models with age composition data requiring little change to the original model structure. Recruitment time series from Pacific herring stock assessment models for 14 populations in the Bering Sea and Northeast Pacific Ocean were analyzed for links to the environment. For some populations, recruitment series were extended backward in time using cohort analysis. In chapter 3, correlation and multivariate cluster analyses were applied to determine herring population associations. There appear to be four major herring groups: Bering Sea, outer Gulf of Alaska, coastal SE Alaska, and British Columbia. These associations were combined with an exploratory correlation analysis of environmental data in chapter 4. Appropriate time periods for environmental variables were determined for use in Ricker type environmentally dependent spawner-recruit forecasting models. Global and local scale environmental variables were examined in forecasting models, resulting in ...
|
|---|