Eveson et al.: Incorporating fishery observer data into an integrated catch-at-age and multiyear tagging model 



501 



DitL. 



th^ 



0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 



/W, 



M, 



M, 



LtL 





D-CL 



Al 



J: 



Q:^ 



c 

 (U 



M4 



0.10 0.15 0.20 0.25 

 F2 



Jl 



0.10 0.15 (1.20 



^3 



Tnje 



Median 



Mean 



Li 



0.6 0.8 1.0 I.: 

 Pi 



1.4 1.6 



0.5 0.6 



0.7 



0.8 0.9 1.0 



Figure 2 



Histograms of the 1000 estimates obtained for each parameter under scenario 1 (see 

 Table ll. The true parameter value, median, and mean of the estimates are indicated 

 by vertical lines. M,= natural mortality rate for age i fish; F, = fishing mortality rate 

 for age / fish; P^ = population size of tagged cohort at age 1 (in 100,000s); A= tag 

 reporting rate for the unobserved component of the fishery (assumed to be constant 

 for scenario 1). 



In regard to precision, we estimated fishing mortality 

 rates, cohort size, and reporting rates with much 

 greater precision (CVs generally in the range of 0.10 

 to 0.20) than that for the natural mortality estimates 

 (CVs often exceeding 0.50) across all scenarios (Table 

 3). Only when natural mortality was constrained to 

 be constant across ages (scenario 8) was reasonable 

 precision achieved for this parameter (CV of 0.22). Of 

 the fishing mortality parameters, the estimates for the 

 oldest age of recapture (i.e., F^ in scenario 11, F^ in all 

 other scenarios) always had the highest CV, and usually 

 notably so. 



Comparing the CVs for a specific scenario with those 

 for scenario 1, we found that the results were generally 

 predictable, at least in terms of direction (Table 3). 

 For example, increasing the value used for the fishing 

 mortality rate (scenario 3) or for the reporting rate 

 (scenario 5) resulted in greater precision (i.e., lower 



CVs) for all parameters, because these changes lead 

 to more tag returns. The results for scenario 2 were 

 not instantly as intuitive. We would expect increasing 

 natural mortality to give higher CVs because more 

 fish would die naturally, leaving fewer tagged fish to 

 be caught. Although small increases were observed 

 in the CVs for the other parameters, large decreases 

 were observed for the natural mortality estimates. 

 This serves as a reminder that the CV is calculated 

 in relation to the true parameter value, and therefore 

 direct comparisons for parameters whose true values 

 have been changed are more complicated. When the 

 SDs of the natural mortality estimates from scenario 

 2 were compared with those from scenario 1 instead of 

 the CVs, they did in fact increase (although this may 

 in part be due to the fact that fewer estimates are 

 truncated at their lower bound of zero when the value 

 used for natural mortality is higher). 



