76 
Fishery Bulletin 95 ( I ), 1997 
Table 6 
Numbers of tags returned and estimates of numbers of tags recovered from various unloading locations. Median reporting rates, 
median numbers of tags recovered, and their respective 95% confidence intervals, were determined from 1,000 bootstrap replica- 
tions based on random sampling from the specified beta (5) or uniform ( U ) distributions. Parameters for the beta distributions 
B(x,y,a,b) are x=r, y=N-r+l, where N is the number of tags seeded, r is the number of seeded tags returned, a is the minimum 
possible reporting rate (based on the local tag-return rate), and b is the maximum possible reporting rate (1). The estimations for 
Korea and Taiwan could not be carried out in the usual way because of zero or very small numbers of seeded or RTTP (or both) tag 
returns. Estimations for these locations are described fully in the text. 
Unloading 
location 
Number of 
tags returned 
Reporting rate 
Number of tags recovered 
Probability 
distribution 
Median 
95% confidence 
interval 
Median 
95% confidence 
interval 
American Samoa 
2,070 
5(254,71,0.016,1) 
0.784 
0.739-0.826 
2,639 
2,505-2,802 
Japan 
1,969 
5(39,42,0.015,1) 
0.492 
0.386-0.595 
4,000 
3,307-5,104 
Korea 
4 
0.002 
0.002-0.003 
1,798 
1,412-2,386 
Philippines 
6,671 
5(4,2,0.223,1) 
0.764 
0.476-0.961 
8,727 
6,940-14,003 
Puerto Rico 
297 
5(9,8,0.002,1) 
0.525 
0.301-0.753 
565 
395-988 
Solomon Islands 
2,226 
5(5,1,0.126,1) 
0.877 
0.526-0.994 
2,540 
2,239-4,232 
Taiwan 
0 
0.000 
0 . 000 - 0.000 
327 
257-434 
Thailand 
2,218 
5(31,56,0.017,1) 
0.366 
0.276-0.466 
6,061 
4,761-8,043 
Tagging vessel 
243 
1.000 
1 . 000 - 1.000 
243 
243-243 
Other 
2,568 
[7(0.5, 1.0) 
0.746 
0.518-0.987 
3,442 
2,601-4,956 
Total recoveries 
18,266 
0.586 
0.490-0.671 
31,166 
27,208-37,264 
port tags, in a large-scale tuna tagging experiment 
in the western tropical Pacific Ocean. 
Tag-shedding rates were estimated by fitting a tag- 
shedding model to double-tagging data. The appli- 
cation of a double-tagging experiment to the estima- 
tion of the rate at which tags are shed from single- 
tagged fish requires several assumptions that are 
discussed in detail by Beverton and Holt (1957). In 
this study, there are four assumptions that warrant 
discussion. First, it must be assumed that the shed- 
ding rates of tags applied in the double-tagging ex- 
periment are the same as those for single-tagged fish. 
This assumption might fail if, for example, less care 
was taken with double tagging than with single tag- 
ging because of the need to return fish to the water 
within certain time limits. In the RTTP tagging ex- 
periment, taggers were instructed to take as much 
care in implanting each tag in double-tagged tuna 
as they would for single-tagged tuna. Although it is 
not possible to test this assumption with the limited 
amount of double-tagging data, the similarity in re- 
turn rates of double- and single-tagged tuna (South 
Pacific Commission 2 ) suggests that there had not 
2 South Pacific Commission. 1994. Oceanic Fisheries Programme 
work programme review 1993-94 and work plan 1994-95. Sev- 
enth standing committee on tuna and billfish; 5-8 August 1994, 
Koror, Palau, South Pacific Commission, Noumea, New 
Caledonia. Working paper 5, 66 p. 
been a gross violation. If the assumption did fail, as 
described above, the shedding rates as applied to 
single-tagged tuna would be overestimated. 
Second, it is necessary to assume for double-tagged 
fish that the events potentially resulting in shedding 
of tags are random and independent with respect to 
the two tags. If this assumption fails, there will be 
fewer observations of fish retaining one tag, and con- 
sequently shedding rates will be underestimated. 
This assumption is difficult to test unless it is pos- 
sible to identify fish that have shed both tags, which 
of course will not normally be the case under field 
conditions. The techniques adopted in this experi- 
ment (individual tag placement on opposite sides of 
the fish) were designed to facilitate compliance with 
this assumption, but the actual extent of compliance 
remains unknown. 
Third, it must be assumed that the first (primary) 
and second (companion) tags applied to fish in a 
double-tagging experiment have the same probabili- 
ties of shedding. This assumption might fail if, for 
example, the companion tag is less securely im- 
planted because tagging on the opposite side of the 
fish is an unfamiliar task. This assumption can be 
tested by using the frequencies of primary and com- 
panion tag retention in fish that were recaptured 
bearing one tag. In this double-tagging experiment, 
there were 68 returns that consisted of one tag. Of 
