74 
Fishery Bulletin 95(1 ), 1997 
Table 4 
Numbers of seeded-tag releases and returns, by unloading location. The 95% confidence intervals were calculated assuming a 
binomial distribution. 
Unloading location 
Number seeded 
Number returned 
Return rate 
95% confidence interval 
American Samoa 
324 
254 
0.784 
0.735-0.828 
Japan 
80 
39 
0.487 
0.374-0.602 
Korea 
16 
0 
0.000 
0.000-0.206 
Philippines 
5 
4 
0.800 
0.284-0.995 
Puerto Rico 
16 
9 
0.562 
0.299-0.802 
Solomon Islands 
5 
5 
1.000 
0.478-1.000 
Thailand 
86 
31 
0.360 
0.260-0.471 
Total 
532 
342 
0.643 
0.600-0.684 
Table 5 
Numbers of seeded tag releases and returns from American Samoa, by year. The 95% confidence intervals were calculated assum- 
ing a binomial distribution. 
Year 
Number seeded 
Number returned 
Return rate 
95% confidence interval 
1990 
23 
3 
0.130 
0.028-0.336 
1991 
116 
95 
0.819 
0.737-0.884 
1992 
50 
48 
0.960 
0.863-0.995 
1993 
101 
83 
0.822 
0.733-0.891 
1994 
34 
25 
0.735 
0.556-0.871 
Total 
324 
254 
0.784 
0.735-0.828 
Estimation of overall reporting rate for the 
RTTP The variation in return rates of seeded tags 
by unloading location (and possibly over time for 
some locations) means that the simple, pooled return 
rate of seeded tags may provide a biased estimate of 
the overall tag-reporting rate for the RTTP. There- 
fore, the tag-seeding data were stratified by unload- 
ing location, and an overall average reporting rate 
weighted by the estimated numbers of RTTP tags 
recovered at those locations was determined. I did 
not attempt to take into account the possible varia- 
tion in reporting rates by time because of insufficient 
information for most locations. 
The estimates of the numbers of RTTP tags recov- 
ered at various locations and in total are shown in 
Table 6. Median reporting rates, numbers of tags 
recovered, and their 95% confidence intervals are 
based on bootstrap sampling from the reporting-rate 
probability distributions indicated in Table 6. The 
relationships between the bootstrap distributions 
and the sample means p ■ from tag seeding are shown 
in Figure 2. For some locations, notably Philippines 
and Solomon Islands, p f overestimates the median 
of the probability distribution of p-. This is due to 
the small numbers of seeded tags in these locations 
and the resulting effect on the shape of the assumed 
underlying probability distributions. 
The estimation of tag recoveries and reporting 
rates for Korea and Taiwan had to be treated differ- 
ently because no RTTP tags were returned from Tai- 
wan and only four tags were returned from Korea 
(which were given to a SPC staff member during a 
brief visit). Additionally, it was not possible to seed 
tagged fish into shipments bound for Taiwan. There- 
fore, there was no basis for estimating tag recover- 
ies and reporting rates in Korea and Taiwan directly 
from tag-seeding and RTTP tag-return data. 
However, other information was available to de- 
rive estimates for these locations. During the period 
of the RTTP, approximately 100,000 t of tuna was 
processed annually by canneries in Korea, all of 
which was supplied by Korean purse seiners (Lewis 1 ). 
1 Lewis, A.D. 1993. Product flows of tuna in the western Pa- 
cific, 1991 with likely trends during 1992. Sixth standing com- 
mittee on tuna and billfish; 16—18 June 1993, Pohnpei, Feder- 
ated States of Micronesia, South Pacific Commission, Noumea, 
New Caledonia. Information paper 2, 7 p. 
