The goal of the Herring Research and Monitoring is to improve the ability to predict herring populations through research and monitoring. The ability to predict population levels require a model of some sort. The model can vary from a simple statistical relationship to one that utilizes many inputs to predict survival and recruitment. Currently the Alaska Department of Fish and Game uses an age-structure-analysis (ASA) model to predict herring biomass levels in Prince William Sound.
The existing ASA model minimizes the sum of squares differences between model estimates and various data sources. One aspect of this project is to move that ASA model into a Bayesian framework that allows the model to output not only the best fit value, but also provides a probability distribution so one can determine the likely range of possible biomass levels. The new model is to be used to evaluate the importance of each input through a process of leaving out variables and examining how much the model population estimates change. The final component of the project is to examine other herring populations to determine how often do populations crash and how long does it take for them to recover.
What we will learn
The new model will make it easier to see the range of likely solutions to the estimated herring population level. We will determine which inputs have the greatest value to the model and the frequency of herring collapse and recovery.