Results 1 to 1 of 1
11-08-2010, 03:12 PM #1
- Join Date
- May 2009
- Read 0 Reviews
Escapement goals for sockeye salmon with informative prior probabilities based on hab
By Harold J. Geiger, Jeffery P. Koenings
Fisheries Research, Volume 11, Issues 3-4, Fish Population Dynamics: Solving Fishery Management Problems, August 1991, Pages 239-256, ISSN 0165-7836, DOI: 10.1016/0165-7836(91)90004-Y.
Managers of sockeye salmon (Oncorhynchus nerka) fisheries have difficulty setting escapement goals without an extensive history of spawner-recruit observations. Part of this difficulty comes from a failure to incorporate information other than spawner-recruit observations into the process. We show how Bayesian decision theory can be used to combine habitat and subjective information with the spawner-recruit history. We offer suggestions that will improve the process of setting escapement goals by: (1) simplifying the process so that a number of reasonable models can be examined simultaneously; (2) exposing the subjective element of the analysis so that it too can be observed and criticized in a somewhat standard and fair manner; (3) promoting learning about the sockeye systems by establishing norms and searching in a standardized fashion for systems that do not behave as expected; and (4) providing a means to keep the search for optimum escapements from looking in places that experience has shown are not fruitful, or even harmful. Without reference to spawner-recruit observations, we make initial guesses at a system's Ricker parameters by studying euphotic volume and other measures, including smolt population characteristics. These guesses are translated into Bayesian prior probabilities. The prior probabilities and system-specific spawner-recruit histories are used to construct posterior probabilities for each model. The consequences of choosing an escapement goal are modeled with a loss function, and the expected losses are calculated with the posterior probabilities. The escapement goal with the lowest expected loss is recommended, regardless of which model seems to be the one most likely to be correct.