Applications of Bayesian Statistical Methods to Fisheries

Moderators: Steve Fleischman, Milo Adkison, Hal Geiger, and Noble Hendrix
Emails: steve_fleischman@fishgame.state.ak.us, Milo.Adkison@uaf.edu, Hal_Geiger@fishgame.state.ak.us and nhendrix@r2usa.com.

Date: Thursday, September 15, 2005
Time: 8:00 am to 5:40 pm
Location: Egan 9-10

Bayesian statistics provides a unified framework for quantifying uncertainty, in which new data are considered in the context of pre-existing knowledge. Until recently, Bayesian methods were of limited utility because they required the solution of often-intractable analytical problems. However new advances in hardware and software have now made it possible to draw samples from posterior distributions with Markov-Chain Monte Carlo and other techniques. One result has been rapid growth in the use of Bayesian methods for complex fisheries models. A key feature of Bayesian statistics is its ability to seamlessly incorporate prior and auxiliary information. In particular, Bayesian hierarchical models borrow strength from an ensemble of estimates to lend coherence and precision to inference about individual stocks. The ability of Bayesian methods to synthesize information from diverse sources in this way fits perfectly with the AFS 2005 theme of creating a fisheries mosaic. The ability of Bayesian methods to make more efficient use of information will become especially helpful if funding for research continues to decline. In this symposium we plan to highlight innovative techniques for using Bayesian statistics in fisheries science applications. To our knowledge, a symposium of this sort has not previously been held. It has generated considerable interest and we have confirmed presentations from first-rate researchers as far away as Nova Scotia, New Zealand and South Africa.

Link to list of presentations in this symposium

Sponsor:

Alaska Department of Fish and Game