70 
Fishery Bulletin 95( 1 ), 1997 
Tagger 
100,000 ■ 
90.000 - 
80.000 ■ 
70.000 - 
60.000 ■ 
50.000 - 
40.000 - 
30.000 ■ 
20.000 - 
10,000 - 
0 ■ 
□ Single tagging 
□ Double tagging 
Skipjack 
Yellowfin 
Species 
Bigeye 
3.000 
2,500 
2.000 
• 1,500 
- 1,000 
- 500 
0 
C 
10,000 
Double tagging 
T l i — i — T — i — i — r — i — i — i — i — i — i 
15 25 35 45 55 65 75 85 95 105 115 125 135 
Length (cm) 
Figure 1 
Distributions of single-tag and double-tag releases by (A) tagger, 
(B) species and (C) size. 
£^(t 2 , t x I p,L) = -y,ln 
i= 1 
^ 2 ^ 2 ; ) 
1- P 0 (^ 2; ) 
X'" 
7=1 L 
1 — Pq (ty ) 
(4) 
where the terms in square brackets represent the 
probabilities of two tags and one tag being observed 
for each recapture, given that at least one tag is ob- 
served. Maximum-likelihood estimates of p andL can 
therefore be obtained by minimizing with 
respect to the parameters. 
The model was fit to pooled recapture data, 
to data classified by species, and to data classi- 
fied by tagger. As an approximate indication of 
the overall losses due to tag shedding for each 
data set, the proportion of tags retained after 
two years (99% of RTTP tag returns were re- 
captured within two years of release), Q 2yr , was 
calculated from Equation 2 by using the esti- 
mated parameters. Approximate 95% confi- 
dence intervals for Q 2 were obtained by the 
percentile method (Efron, 1982) applied to dis- 
tributions of Q,, v; . generated from 1,000 paramet- 
ric bootstrap (or Monte Carlo) replicates of each 
data set. The replicates were constructed by 
using the observed distributions of times at lib- 
erty, and the numbers of tags observed for each 
pseudo-return were determined randomly with 
the conditional probabilities of a recaptured tuna 
bearing two tags or one tag, i.e. and 
respectively, given the estimated parameters. 
The statistical significance of improvements 
in fit of models that included species-specific 
and tagger-specific shedding parameters was 
determined by using likelihood-ratio tests 
(Kendall and Stuart, 1961). 
Tag-seeding experiments 
Rationale Tag seeding was carried out by ob- 
servers placed on board purse-seine vessels as 
part of regional and national observer pro- 
grams. The purse-seine fleet was targeted for 
tag-seeding experiments for several reasons. 
First, purse seiners account for most of the tuna 
catch in the western Pacific (and also recovered 
most tags); the estimation of reporting rates for 
this gear type in particular was therefore of 
critical importance. Second, the large, modern 
purse seiners typical of the western Pacific fleet 
handle large quantities of tuna very rapidly, 
with little opportunity for onboard inspection 
of individual fish for tags. As a result, tagged 
tuna recaptured by purse seiners were mostly 
detected during unloading (when individual fish 
are handled) or during the initial stages of pro- 
cessing in canneries. The efficacy of tag detec- 
tion during these periods was unknown prior 
to the commencement of the tagging experi- 
ment; it was feared that delayed detection of 
tags might result in significant losses which, if 
ignored, would compromise the objectives of the 
tagging experiment. Third, the very fact that 
most tagged tuna recaptured by purse seiners 
