572 
Fishery Bulletin 95(3), I 997 
A 
Seychelles 
Gulf of 
Mexico 
California 
Mexico 
California 
Philippines 
Philippines 
- Coral Sea 
Hawaii 
Seychelles 
Hawaii 
Kiribati 
Gulf of 
Mexico 
Sri Lanka 
Sri Lanka 
Mexico 
Kiribati 
J 1 
Coral Sea 
L 
1 
_l 
0.050 
0.025 
0.000 0.014 
0.007 0.000 
Genetic distance 
Percentage divergence 
Figure 3 
UPGMA phenograms for the mtDNA data with (A) unbiased genetic dis- 
tance (Nei 1978), and (B) percentage nucleotide divergence (Nei, 1987). 
was significant following Bonferroni 
correction to a levels (west-central 
Pacific versus east Pacific, P=0.456, 
a=0.05; Indian versus east Pacific, 
P=0.316, a=0. 025; Atlantic versus east 
Pacific, P=0. 119, a=0.017; Atlantic ver- 
sus Indian, P=0. 047, a=0. 0125; Atlan- 
tic versus west-central Pacific, P=0.032, 
a=0.010; Indian versus west-central 
Pacific, P=0.0135, a=0.008), the three 
pairwise comparisons of the Atlantic 
Ocean, Indian Ocean, and west-central 
Pacific all showed P-values less than 
0.05. 
Clearly, the mtDNA data do not dif- 
ferentiate west-central Pacific Ocean 
collections from east Pacific Ocean col- 
lections but, considering the inter- 
ocean analyses alone, do provide some 
support for the delineation of Atlantic 
Ocean, Indian Ocean, and Pacific 
Ocean stocks. 
otide divergence, pairwise nucleotide divergence 
ranged from 0.040% to -0.025% (mean 0.004%). 
There was little correspondence between these two 
mtDNA dendrograms, and this lack of correspon- 
dence, together with the low distances observed, sug- 
gests that the tree topologies are unreliable. 
Because there is no significant mtDNA differen- 
tiation between the six Pacific Ocean collections 
(Table 5; and Ward et al., 1994) nor between the two 
Indian Ocean collections (Table 5), the collections 
within each ocean were pooled to test for interoce- 
anic differences. A comparison of the three oceans 
yielded a chi-square analysis that was significant 
(P=0.009, a=0.05) and a genetic diversity analysis 
bordering on significance (observed G ST =0.010, 
G sT.nuii =0 005 ±0.003, P=0.059). A pairwise compari- 
son of the oceans showed that all pairs were signifi- 
cant (Indian versus Atlantic, P=0.047, a=0.05; Pa- 
cific versus Atlantic, P=0. 017; Pacific versus Indian, 
P=0.009). 
Finally, the mtDNA data were analyzed to see 
whether they offered any support to the conclusion 
from the GPI-A* data that there are (at least) four 
yellowfln tuna stocks. The four putative stocks con- 
sisted of the following units: Atlantic (Gulf of Mexico), 
Indian (Seychelles and Sri Lanka), west-central Pa- 
cific (Coral Sea, Kiribati, Philippines, Hawaii), and 
east Pacific (California and Mexico). Chi-square 
analysis of mtDNA data from these four regions in- 
dicated limited but significant (P=0.Q24) heteroge- 
neity. Although none of the six pairwise comparisons 
Discussion 
Samples of yellowfin tuna from the Pacific, Indian, 
and Atlantic oceans were compared with respect to 
four polymorphic allozyme loci and with respect to 
mtDNA variants. 
No significant allele frequency differences were 
observed for three of the allozyme loci, but the fourth 
locus, GPI-A*, showed considerable differentiation. 
Across all collections, the “true” G gT indicated that 
about 11% of the variation at this locus was attrib- 
utable to differences between collections. Two geneti- 
cally distinguishable groups were apparent. One con- 
sists of eastern Pacific Ocean and Indian Ocean fish, 
with a high frequency of the GPI-A*lb allele, the 
other of Atlantic Ocean and west-central Pacific 
Ocean fish, with a high frequency of the GPI-A* 100 
allele. Because there are no migration routes between 
the eastern Pacific Ocean (California and Mexico) and 
the Indian Ocean that avoid the west-central Pacific 
Ocean, and between the Atlantic Ocean and west- 
central Pacific Ocean that avoid the Indian Ocean, 
there is reason to believe that there are at least four 
stocks of yellowfin tuna: Atlantic Ocean, Indian 
Ocean, west-central Pacific Ocean, and eastern Pa- 
cific Ocean. 
Sharp ( 1978) also examined GPI-A* allele frequen- 
cies in western and eastern Pacific populations. His 
GPI-A* allele frequencies for collections from Ecua- 
dor and Mexico were very similar to our California 
and Mexico frequencies, and his GPI-A* frequencies 
