Hampton: Estimates of tag-reporting and tag-shedding rates for tuna in the tropical Pacific Ocean 
69 
scale tuna tagging program, the Regional Tuna Tag- 
ging Project (RTTP), in the western tropical Pacific. 
From 1989 to 1992, 146,581 tagged skipjack tuna, 
Katsuwonus pelamis, yellowfin tuna, Thunnus 
albacares, and bigeye tuna, Thunnus obesus, were 
released throughout the western tropical Pacific from 
the Philippines and eastern Indonesia to approxi- 
mately 170°W. This area is fished by purse-seine, 
pole-and-line, longline, handline, and troll vessels, 
which have collectively harvested more than one 
million metric tons of tuna per year since 1989 
(Lawson, 1994). As of 31 May 1995, 18,266 tagged 
fish had been recaptured and the tags and accompa- 
nying recapture information returned to SPC. Tagged 
tuna were recaptured by all of the fishing methods 
of the western Pacific fishery. Most tag returns (76%) 
originated from purse seiners, consistent with the pro- 
portion of total catch attributed to that gear (67% for 
1990-1993). Few additional tag recoveries are expected. 
One of the major objectives of the tagging program 
was to estimate the rates of fishing-induced and natu- 
ral mortality by using models similar to Equation 1, 
so that the impacts of the fishery on the stocks could 
be assessed. It was therefore necessary to obtain es- 
timates of type-1 and type-2 tag losses. In this pa- 
per, I focus on the estimation of tag-shedding rates 
and tag-reporting rates. Tag-shedding rates were 
estimated from double-tagging experiments carried 
out in conjunction with the tag-release program. Dif- 
ferences in shedding rates among species and differ- 
ences among individual taggers were evaluated. Tag- 
reporting rates were estimated from tag-seeding ex- 
periments in which tuna caught by purse seiners 
were surreptitiously tagged by fisheries observers 
prior to the fish being placed in the fish wells. Dif- 
ferences in the rates of reporting seeded tags by spe- 
cies, time, and port of unloading were investigated. 
An estimate of the overall reporting rate of recov- 
ered RTTP tags and its variability, which takes into 
account the variability in tag reporting among un- 
loading ports, was obtained. 
Materials and methods 
Double-tagging experiments 
Field operations Tagging was carried out on a pole- 
and-line vessel from which tuna were captured with 
standard commercial gear. Only uninjured fish that 
were cleanly hooked in the jaw were selected for tag- 
ging. Fish with excessive mouth damage, eye dam- 
age, or gill damage were not tagged. Selected fish 
were placed in a vinyl tagging cradle and their fork 
lengths measured to the nearest centimeter. For 
single-tagged fish, a Hallprint '' 13 cm dart tag was 
inserted by using a sharpened stainless steel appli- 
cator, into the musculature at an angle of about 45°, 
1-2 cm below the posterior insertion of the second 
dorsal fin. Smaller ( 10-cm) tags were used for tuna less 
than 35 cm FL. Ideally, the tag barb was anchored be- 
hind the pterygiophores of the second dorsal fin. 
Throughout the three-year tag release program, a 
small sample (approximately 3%) of the tagged tuna 
were double tagged. Double tagging occurred on par- 
ticular days chosen in advance by the cruise leader 
and on such days, most fish were double tagged. The 
objectives were for each principle tagger to double 
tag at least 400 tuna, and for the double-tag releases 
to be as representative as possible of the species and 
size composition of the single-tag releases. These 
objectives were largely accomplished (Fig. 1). 
The technique for double tagging was identical to 
that of single tagging, with the exception that a sec- 
ond tag was inserted on the opposite side of the fish, 
1-2 cm anterior to the first tag to avoid damaging it 
with the applicator. For single and double tagging, 
fish were generally out of the water for less than ten 
seconds. 
Data analysis Observations of the numbers of tags 
retained by double-tagged tuna at recapture can be used 
to estimate tag-shedding rates. I used a simple tag- 
shedding model (Beverton and Holt, 1957; Hampton 
and Kirkwood, 1990), which defines the probability, 
Q(t), of a tag being retained at time t after release as 
Q(t) = ( 1- p)exp(-Lt), (2) 
where p is the immediate type-1 shedding rate and 
L is the continuous type-2 shedding rate. These pa- 
rameters can be estimated from a double-tagging 
experiment under the assumption that all tags not 
immediately shed have identical shedding probabili- 
ties that are independent of the status of the com- 
panion tag. Given this assumption, the probabilities 
of two, one, and no tags being retained at time t af- 
ter release are, respectively, 
P 2 (t) = Q(t) 2 , 
P 1 (t) = 2Q(t)[l-Q(t)] (3) 
P 0 (t) = [l-Q(t)f. 
Consider a double-tagging experiment resulting in 
m recaptures of fish bearing two tags at times t 2i ( i = 
1, ..., m) and in n recaptures bearing one tag at times 
ty (J = 1, ..., n). The negative log likelihood of the 
data (t.,, tj) given the model parameters p and L is 
