Table 5. — Estimates of immediate (1 - p) and annual instan- 

 taneous it) tag shedding rates for northwestern Atlantic bluefin 

 tuna double-tagging study for all years combined (1971-77) 

 based on a 3-yr return period using unweighted and weighted 

 linear regression models. (The weights used in the weighted 

 model were equated to the ratio of the number of returns of 

 double- tagged fish during each return period to the total number 

 of returns of double-tagged fish during all periods.) 



Model and tag type 



1 - P 



L (annual) 



Linear regression: 

 Plastic dart 

 Metal dart 



Combined 



Weighted linear regression: 

 Plastic dart 

 Metal dart 

 Combined 



0.027 

 0.049 



0.040 



0.033 

 0.049 



0.042 



22886 

 0.19201 

 0.20452 



0.19200 

 0.18213 

 0.18596 



indicate the magnitude of the variances of the 

 estimates. 



Our estimate of (1 - p) is slightly greater than 

 the overall estimate of 0.027 given for bluefin tuna 

 from the northwest Atlantic by Lenarz et al. 

 (1973). The difference is small relative to the pre- 

 cision of the estimates. Our estimate of (1 - p) for 

 northwest Atlantic bluefin tuna is less than the 

 value of 0.10 reported for Pacific yellowfin tuna by 

 Bayliff and Mobrand (1972) and the value of 0.12 

 reported for North Pacific albacore by Laurs et al. 

 (1976). 



Our estimate of L is less than the overall esti- 

 mate of 0.31 reported by Lenarz et al. (1973) for 

 bluefin tuna and the L estimate of 0.278 reported 

 for yellowfin tuna by Bayliff and Mobrand (1972). 

 Our L estimate is greater than the estimates of 

 between 0.086 and 0.098 reported for albacore by 

 Laurs et al. (1976). 



As previously noted, there was no significant 

 difference in return rates found for the two types of 

 dart tags for 1971-77. However, from examination 

 of the data presented in Table 1, there appeared to 

 be changes occurring in the shedding rates of each 

 type of tag and a difference between the 1971-73 

 and 1974-77 time intervals. Therefore, we calcu- 

 lated (1 - p) andL for each time interval and con- 

 ducted chi-square tests (df = 6) for differences in 

 returns over three recapture periods ik = 3) be- 

 tween time intervals and between tag types (Table 

 6). We found significant differences between time 

 intervals for each of the tag types and significant 

 differences between tag types for each of the time 

 intervals. The plastic dart tags became less 

 efficient, i.e., L increased over the time intervals, 

 and the metal dart tags improved, i.e., L decreased 

 over the time intervals. 



The model of Chapman et al. (1965), which was 



Table 6. — Estimates of immediate (1 - p) and einnual instan- 

 taneous (L) tag shedding rates for northwestern Atlantic bluefin 

 tuna double-tagging study for time intervals 1971-73 and 1974- 

 77 based on a * = 3-yr return period. (A contingency table, 7x2, 

 was constructed containing the number of double and single 

 returns for each of the three recapture periods plus the number of 

 nonreturns for each tag type and each time interval.) Results of 

 chi-square tests (df = 6) for differences in double and single tag 

 returns and total nonreturns between time intervals and tag 

 types over a 3-yr recapture period are given. 



Tag type and 

 time interval 



L (annual) 



Chi-square value 



Plastic dart: 

 1971-73 

 1 974-77 



Metal dart: 

 1971-73 

 1974-77 



1971-73: 

 Plastic dart 

 Metal dart 



1974-77: 

 Plastic dart 

 Metal dart 



029 

 0.023 



0.140 

 0.007 



029 



0.140 



0.023 

 0.007 



0.14838 

 0.28455 



0.37163 

 0.17242 



0.14838 

 0.37163 



0.28455 

 0.17242 



64.286" 

 33.489" 

 18.924" 

 18.135" 



"P«0.01. 



modified by Bayliff and Mobrand (1972), assumes 

 constant L over recapture periods. We decided to 

 examine values of L over the two pairs of recap- 

 ture periods k = (1, 2) and k = (2, 3) to determine 

 how well our data fit the model. Since only two 

 recapture periods were used, L and ( 1 - p) were 

 estimated by solving two simultaneous equations. 



For the tag types and time intervals examined, 

 there is an indication that L is not constant (Table 

 7). In fact, L increased in three out of four cases. 

 The sequence of events could have happened due 

 to chance alone, for if the changes in L came from a 

 binomial distribution with P =0.5, then the prob- 

 ability of L decreasing in three of the four cases or 

 L increasing in three of the four cases is ^0.25. 

 However, L during the second time period is more 

 than 60% >L in the first time period in three cases 

 and only 16% <L in the first time period in the 

 other case. While the data do not provide conclu- 

 sive evidence that L is not constant, it would be 

 dangerous to extrapolate beyond the time period 

 used for analysis. 



We previously noted that the 1974 releases were 



Table 7. — Estimates of annual instantaneous (L) tag shedding 

 rates for northwestern Atlantic bluefin tima double-tagging 

 study for 1971-73 and 1974-77 based on return periods of A = (1, 



2) and k = (2, 3). 



183 



