Stewart: Defining migration rates of Parophrys vetulus 



481 



that might serve to disrupt movement along the coast. 

 Although conceptually appealing, these extensions may 

 cause technical problems for MCMC because of the low 

 information content of the aggregated historical tag- 

 ging data. 



Movement rates are notoriously difficult to estimate 

 (Xiao, 1996). However, the general approach to inte- 

 grated tag analysis based on maximum likelihood has 

 been found to be reliable through simulation testing 

 (Maunder, 2001). In this application, the error structure, 

 although intended to accommodate clustered recoveries 

 and the inclusion of zero recoveries in many space-time 

 combinations, may be inadequate for the observed level 

 of variability in recoveries. Specifically, there were many 

 cases of observed recoveries in areas where no recoveries 

 could have been predicted under simple models, given 

 the structure imposed by the population dynamics that 

 were assumed. Future extension and simulation testing 

 to evaluate other error structures, such as zero-inflated 

 Poisson or negative binomial models, should be undertak- 

 en. Further, there may be interactive effects of the prior 

 distributions used, despite lack of observed response to 

 one-dimensional sensitivity testing. Some important ele- 

 ments of uncertainty may not have been included in the 



present analysis; however, uncertainty is a problem faced 

 by most modeling applications in fisheries. 



It is unlikely that future stock assessments will be 

 structured around areas as small as PMFC areas. Con- 

 version of the movement rates reported here for use 

 as priors in stock assessments will require assump- 

 tions regarding the distribution of biomass within areas 

 modeled and the correspondence of the areas modeled 

 to PMFC areas. Research survey data may provide a 

 fishery-independent source on which to base these as- 

 sumptions. The estimation of fishing mortality in this 

 analysis should also be considered if it is to be applied 

 within a stock assessment framework. These issues 

 would appear to be no more daunting than currently 

 accepted assumptions of thorough mixing across the 

 coast or of completely isolated stock groups. 



Tagging programs for groundfish species off the U.S. 

 west coast have generally decreased over the last 75 

 years, and there are no plans for large-scale tagging 

 programs in the future. An analysis of all available tag- 

 ging data may therefore provide the only quantitative 

 source of guidance and an important avenue to allow 

 for uncertainty in movement rates without new data 

 with which to estimate these rates. This type of method 



