Ward et a I.: Population structure of Thunnus albacares 
569 
slightly less than 1% of the observed diversity arose 
from differences between collections and could be 
attributed to sampling error alone (G STnu[l ). For GPI- 
A*, the observed value of G ST , at 12%, was much 
larger than the value attributable to sampling error 
(about 1%). The “true” G ST e stimate of GPI-A* — the 
difference between G sr and G STnull — is thus around 
11%, indicating that about 11% of the observed di- 
versity at the GPI-A* locus comes from differences 
between collections. 
The GPI-A* heterogeneity (Fig. 1) was further ex- 
plored by comparing all collections pairwise (Fig. 2; 
Table 4). This comparison essentially revealed two 
groups of collections: 1) the west-central Pacific 
Ocean and the Atlantic Ocean (Gulf of Mexico) col- 
lections; and 2) the Indian Ocean (Sri Lankan) and 
eastern Pacific Ocean (Californian and Mexican) col- 
lections. Within each of these two groups there was 
no significant differentiation, but between them dif- 
ferentiation was marked. This conclusion holds af- 
ter Bonferroni corrections of a levels for multiple 
tests. The GPI-A* 100 allele was the most frequent 
allele in the west-central Pacific Ocean and the At- 
lantic Ocean group, whereas the GPI-A * 75 allele was 
the more frequent allele in the Indian Ocean and the 
eastern Pacific Ocean group. The genetic differen- 
tiation of the Atlantic Ocean collection from the In- 
dian Ocean collection suggests that fish from these 
areas constitute separate stocks; the separation of 
the Indian Ocean collection from the west-central 
Pacific Ocean collections suggests that fish from these 
areas constitute separate stocks; the separation of 
the west-central Pacific Ocean collections from the 
eastern Pacific Ocean collections suggests that fish 
from these areas constitute separate stocks; and the 
separation of the eastern Pacific Ocean collections 
Table 2 
Mitochondrial DNA haplotype frequencies (Bel I and Eco RI haplotypes respectively), sample sizes (n), haplotype diversities (h) 
and percent nucleotide diversities (% n.d.). Abbreviations are defined in Table 1. Seych. = Seychelles. 
Locus Haplotype 
Atlantic 
Indian 
Pacific 
GOM 
Seych. 
S. Lan. 
Philipp. 
Cl. Sea 
Kiribati 
Hawaii 
Calif. 
Mexico 
mtDNA AA 
0.266 
0.319 
0.381 
0.286 
0.340 
0.443 
0.276 
0.294 
0.325 
AB 
0.543 
0.407 
0.333 
0.509 
0.402 
0.364 
0.537 
0.463 
0.425 
AC 
— 
— 
— 
— 
— 
0.011 
0.015 
— 
— 
AF 
— 
0.011 
— 
— 
— 
— 
0.007 
0.049 
0.025 
AG 
— 
— 
— 
0.006 
— 
— 
0.007 
— 
— 
BA 
0.011 
0.011 
— 
0.025 
0.010 
— 
0.007 
0.049 
— 
BB 
0.064 
0.066 
— 
0.031 
0.052 
0.068 
0.060 
0.073 
0.025 
CA 
0.032 
0.011 
— 
0.031 
0.062 
0.011 
0.015 
0.024 
0.050 
CB 
0.021 
0.088 
0.286 
0.068 
0.103 
0.057 
0.045 
0.024 
0.125 
CO 
0.021 
— 
— 
— 
— 
— 
— 
— 
— 
DB 
— 
— 
— 
— 
0.010 
— 
— 
— 
— 
EB 
— 
— 
— 
— 
0.021 
— 
— 
— 
— 
LB 
— 
— 
— 
— 
— 
0.011 
— 
— 
— 
MB 
— 
— 
— 
— 
— 
0.011 
— 
— 
— 
NB 
— 
— 
— 
0.019 
— 
— 
0.007 
— 
— 
PB 
— 
— 
— 
— 
— 
— 
0.015 
0.024 
— 
OA 
— 
— 
— 
0.006 
— 
— 
— 
— 
0.025 
OB 
— 
— 
— 
0.006 
— 
0.023 
— 
— 
— 
QA 
— 
0.011 
— 
— 
— 
— 
— 
— 
— 
QB 
0.011 
0.011 
— 
— 
— 
— 
0.007 
— 
— 
WA 
0.032 
0.033 
— 
— 
— 
— 
— 
— 
— 
ZB 
— 
0.022 
— 
0.006 
— 
— 
— 
— 
— 
A2B 
— 
— 
— 
0.006 
— 
— 
— 
— 
— 
Q2B 
— 
0.011 
— 
— 
— 
— 
— 
— 
— 
n 
94 
91 
21 
161 
97 
88 
134 
41 
40 
h 
0.634 
0.727 
0.695 
0.655 
0.712 
0.670 
0.633 
0.705 
0.712 
% n.d. 
0.998 
1.263 
1.017 
1.017 
1.174 
1.027 
0.901 
1.099 
1.047 
