Events and Seminars : 2014 Seminars

THE USE OF BAYESIAN METHODS IN FISHERIES POPULATION DYNAMICS MODELING

ELIZABETH A. BABCOCK, PhD
Assistant Professor
Department of Marine Biology and Ecology
Rosenstiel School of Marine and Atmospheric Science
University of Miami
TUESDAY, APRIL 28, 2015
11:00 a.m.–12:00 p.m., CRB 692

Bayesian methods are commonly used in fisheries science because Bayesian methods allow the use of prior information from biological studies to inform the parameters estimated using fisheries data. For example, the natural mortality rate (M) is a parameter that must be estimated in many models used in fisheries stock assessment, but fisheries data are often uninformative about this parameter. Natural mortality rates can be estimated from tagging studies on the species of interest, or inferred from a meta-analysis across populations or species. A Bayesian depletion model including in-season recruitment was applied to the standardized catch per unit effort (cpue) of Caribbean spiny lobster (Panulirus argus) to estimate the abundance and fishing mortality of lobsters at two fished sites in Belize. The model used uninformative priors for all parameters except M, which was given an informative prior based on studies of other populations of the same species. At Glover’s Reef, a typical model estimated abundance of 66–79000 lobsters. The depletion model applied to data from the 2012 season at Port Honduras found an abundance of ~12000 lobsters. The models estimated a harvest fraction of 70% at both sites. Frequentist depletion models generally use a fixed value of M. The Bayesian prior allows the parameter to be informed by the prior while allowing for some uncertainty in the estimate. The Bayesian framework also facilitates the estimation of probabilities, such as the probability that the fishing mortality rate is above a target level.