276 



Fishery Bulletin 9 1 (2), 1993 



10 20 30 40 50 60 

 AGE (YR) 



10 20 30 40 50 60 70 

 AGE (YR) 



Figure 4 



von BertalanfFy growth curves for male (top) and 

 female (bottom) sablefish in Gulf of Alaska, cal- 

 culated from length-at-age data. 



For west coast sablefish, 52,743 fish were 

 released, with 4220 recoveries between 1971 

 and 1991. Of these recoveries, 1083 had us- 

 able growth-increment information (Figs. 

 5,6). Pot gear was the dominant release gear, 

 but both trawl and pot were important re- 

 covery gears (Table 2). 



Estimated von Bertalanffy parameters for 

 west coast sablefish based on tagging data 

 can be compared with von Bertalanffy pa- 

 rameters estimated directly from length- 

 at-age data i Table 4) from specimens collected 

 in trawl and pot surveys in 1983 and 1989, 

 and aged during 1991 and 1992 (Fig. 7). For 

 this dataset, only James' estimates compared 

 well with estimates based on direct ages. If 

 James' estimates were not available, we could 



not conclude that estimates of growth parameters based on tag- 

 ging were consistent with length-at-age data generated by age 

 readers. Unexpectedly for the west coast dataset, James' esti- 

 mates of SE(L„) were smaller than for the unweighted Fabens' 

 estimates. 



Another possible problem when analyzing growth-increment 

 data is that the observed increments may be dependent on the 

 tag or recovery gears that are used. Our west coast sablefish 

 tag-recapture dataset provides an opportunity to examine 

 whether observed growth increments may depend on the fishing 

 gears used for tagging and recapture. From this dataset, we 

 selected only fish with pot gear releases and trawl or pot gear 

 recoveries. We then regressed growth increment against sex, 

 recovery gear, size-at-release, and time at liberty. The analysis 

 showed that fish recovered by pot gear could be expected to have 

 growth increments 3.7 cm larger (significant a=0.0001) than fish 

 recovered by trawl gear after the other independent variables 

 were taken into account. Thus, release and recovery gears can 

 be expected to affect the growth curves estimated from tag- 

 recapture data. 



Estimates from Pacific cod tag-recapture data 



Pacific cod has also proved to be a very difficult species to di- 

 rectly age (Kimura & Lyons 1990). For eastern Bering Sea 

 (Fig. 1) Pacific cod, 12,396 tagged fish were released, with 375 

 recoveries between 1982 and 1989. Of these recoveries, 284 had 

 usable growth-increment information (Fig. 8). Bottom trawl was 

 the only release gear, but both trawl and longline were impor- 

 tant recovery gears (Table 2). Because of the small tag- 

 recapture sample size available for Pacific cod, we combined 

 male and female data into a single fit. 



As with sablefish, Pacific cod tag recaptures can show nega- 

 tive growth increments (Fig. 8). For the complete tag-recapture 

 dataset, unweighted and weighted Fabens' estimates were bi- 

 ased when compared with estimates based on length-at-age data 

 (Table 5). James' estimates did not exist within a reasonable 

 range for K,, (0.01<K o <1.0). Using recoveries with only positive 

 growth increments allowed James' estimates to be close to those 

 based on length-at-age data (Table 5). After deleting only the 

 larger negative growth increments (i.e., larger in absolute value), 

 data with increments greater than -10 cm gave a standard error 

 for L using James' method of 53.4 cm. These results indicate 

 that for James' estimates, the sample size for Pacific cod was 

 insufficient to overcome variability in the data. 



Discussion 



Our simulation study confirmed that James' robust estimators 

 for estimating von Bertalanffy parameters from tag-recapture 

 data show little bias when variability can be described by simple 

 observational error and/or variability in h r . In contrast, the 

 unweighted and weighted Fabens' estimators sometimes had 



