658 



Fishery Bulletin 91(4). 1993 



an albacore of a specific initial size over time. This is 

 equivalent to using the L . and K estimates in a Fabens- 

 type length increment model to make an approximate 

 comparison of predicted growth trends independent of 

 assumed or estimated t„. Despite substantial differ- 

 ences in the L„ and K estimates from the two analy- 

 ses, there was little difference in the predicted growth 

 of a 40-cm albacore over time within the range of the 

 data (Fig. 4). 



Growth rates of tagged albacore 



The 28 tag release-recapture records with com- 

 plete growth data included a wide variety of sizes 

 (61.0-96.5cm), times-at-large (54-1,790 days) and 

 release-recapture locations well distributed 

 throughout the principle fishing areas. Estimated 

 average growth rates during the periods at large 

 were 0.17-1.13cmmonth '. For albacore at lib- 

 erty <500 days, growth rate showed the expected 

 decline with increasing mid-size (half way be- 

 tween release and recapture sizes); but this de- 

 cline was negligible for albacore at liberty >500 

 days (Fig. 5). Given the relatively small number 

 of records available, no attempt was made to es- 

 timate the VB growth parameters based on tag 

 release-recapture data. 



Comparison of tag return growth 

 increments with the fitted models 



To compare the consistency of the VB models de- 

 rived earlier with the tag-return data, we first 

 calculated sets of predicted recapture lengths for the 

 28 tag return records by using the two sets of L„ and 

 K estimates, derived from length-frequency and verte- 

 bral-ring-count data, in Equation 2. We also made a 

 third set of predictions using the L . estimate and twice 

 the K estimate obtained from the length-frequency 

 analysis (L„-97.1cm, K=0A78 yr~M. These parameters 

 are consistent with an assumption that the observed 



