FISHERY BULLETIN: VOL. 79, NO. 2 



100 200 300 400 500 600 700 800 900 1000 MOO 1200 



DAYS OUT (A) 



Figure 4. — Residuals from the fit of Model 1 to the Group A albacore data, as a function of days out. 



tagging and then recover to normal within the 

 first year of liberty. While this basic response to 

 tagging stress appears to be a reasonable hy- 

 pothesis, the estimates of a and /3 w^ere very 

 unstable and little confidence could be placed on 

 projected growth rate recovery patterns. Further, 

 although the extended model fits the data better 

 than the ordinary von Bertalanffy in the sense of 

 reducing residual variation, as expected, this 

 improvement is only slight; considering the high 

 variance displayed in the data we do not reject 

 the simpler von Bertalanffy model. 



DISCUSSION AND CONCLUSIONS 



The results of our growth analysis are strength- 

 ened by their consistency with other findings, but 

 the assumptions of our analysis need to be tested. 

 In particular, we assumed that our estimate ofL^^, 

 125 cm, was the same for all groups of North 

 Pacific albacore. If L3c(B -I- C)>L:c(A), our con- 

 clusions concerning differences in growth rate are 

 reinforced. But if the South fish. Group A, actually 

 tend toward a larger asymptote in fork length 

 than the North fish of Groups B and C (there is no 

 evidence of this), then the differences between 

 estimates of K might not be significant. For 



example, if we assume Lx(A) = 130 cm and 

 Ly^iB + C) = 120 cm, the differences vanish. More 

 to the point, unless the Ly's are the same, the 

 comparison of growth rates between groups is no 

 longer conveniently reduced to a comparison of K 

 estimates. When the assumption of equal Lx is 

 valid, the actual value ofL^ assumed is relatively 

 unimportant as far as the covariance analysis is 

 concerned; our conclusions were the same when 

 Ly was fixed at 120, 130, and 135 cm. 



In Model 1, the standard von Bertalanffy model, 

 we also assumed that the growth rate was un- 

 altered by the presence of the tag or by the stress 

 imposed in its application. In our analysis of 

 Model 2 we explored the question of whether 

 tagging might have affected growth rate in a 

 specified way, and Model 2 fits our data only 

 slightly better than Model 1. However, effects of 

 the sort we hypothesized might easily be masked 

 by high variance in the data. Nevertheless, if the 

 effect of tagging were simply to reduce the normal 

 growth rate, K, suddenly and permanently to a 

 lower level, K', it would go undetected by our 

 analysis. To determine the validity of the tag- 

 effect assumption, we need to compare the growth 

 rates of tagged fish with those of untagged, "con- 

 trol" fish. 



300 



