700 
Fishery Bulletin 99(4) 
samples deviated considerably from zero (F sr =0.018 to 
0.070, PcO.OOl) for all loci. Significant departures of F ST 
from zero were observed for three loci ( Ttho-1 *, Ttho-4*, 
and Ttho-6*) among all samples except the NE Atlantic 
sample ( F ST =0.013 to 0.085, P<0.005) and among the 
three Pacific samples (F S7 —0.018 to 0.074, PcO.OOl). There 
were also significant differences in allele frequencies be- 
tween some pairwise comparisons (Table 2). At Ttho-1 *, 
the SW Pacific sample showed significant difference from 
all other samples (all PcO.OOl), and there was also a sig- 
nificant difference between the SE Pacific and SW At- 
lantic samples (P=0.003). The NE Atlantic sample was 
significantly heterogeneous in comparison with all other 
samples in Ttho-4* (all PcO.OOl). Difference between the 
NW and SE Pacific samples was also significant in this 
locus (P=0.001). In Ttho-6 * the NE Atlantic sample was 
again significantly heterogeneous in comparison with all 
other samples (all PcO.OOl). Furthermore, there were also 
significant differences between the NW and SE Pacific, be- 
tween the SW and SE Pacific, and between the SW Pacific 
and SW Atlantic samples in this locus (all PcO.OOl). In 
Ttlio-7 * the NE Atlantic sample showed significant differ- 
ences from all three Pacific samples (PcO.OOl for the NW 
and SW Pacific, P=0.003 for the SE Pacific) but not from 
the SW Atlantic sample (P=0.017). 
Discussion 
Because the number of alleles observed in microsatellite 
loci is usually large and the frequency of each allele may 
be low, a large sample size is necessary for satisfying sub- 
sequent statistic analyses. Ruzzante (1997) showed that a 
sample size of 50 <n< 100 individuals is generally satis- 
factory, although this size depends on allele number and 
frequency. Size of our samples ranged from 32 to 48, close 
to the lower margin of this threshold. Nevertheless, the 
distinct status of the Northeast Atlantic albacore sample 
from others is obvious. Although our study shared the 
same sample lots with Chow and Ushiama (1995), their 
mtDNA analysis revealed much less genetic differentia- 
tion between samples from the North and South Atlantic. 
MtDNA analysis is thought to be a more effective indi- 
cator for population subdivision than nuclear DNA (Nei 
and Li, 1979). But for albacore in the Atlantic, this is 
obviously not the case. No selection toward microsatellite 
alleles is obvious; therefore, differences in evolutionary 
rate between mtDNA and nDNA may explain differences 
in allele frequency distribution in the Atlantic albacore. 
Because mtDNA variation observed by Chow and Ush- 
iama (1995) was much lower than the microsatellite DNA 
variation observed in our present study, there might have 
been insufficient elapsed time for unique mitochondrial 
genotypes to have arisen within the existing stocks. Thus, 
the rapidly evolving microsatellites appear to reflect alba- 
core population subdivision. 
MtDNA analyses have also failed to detect genetic differ- 
ence between samples from northern and southern hemi- 
spheres within the Pacific (Chow and Ushiama, 1995). In 
contrast, present microsatellite analysis detected signifi- 
Table 2 
Pairwise comparison of F ST between five albacore samples. 
NW 
Pacific 
SW 
Pacific 
SE 
Pacific 
SW 
Atlantic 
Ttho-1* 
SW Pacific 
0.1401 
SE Pacific 
0.005 
0.099 7 
SW Atlantic 
0.018 
0.249' 
0.0461 
NE Atlantic 
-0.004 
0.154 7 
-0.003 
0.023 
Ttho-4* 
SW Pacific 
0.003 
SE Pacific 
0.032 7 
0.012 
SW Atlantic 
0.006 
-0.001 
0.017 
NE Atlantic 
0.15U 
0.131 7 
0.094 7 
0.131 7 
Ttho-6* 
SW Pacific 
0.020 
SE Pacific 
0.038 7 
0 . 100 7 
SW Atlantic 
0.006 
0.052 7 
0.010 
NE Atlantic 
0.135 7 
0.208 7 
0.0431 
0.086 7 
Ttho-7* 
SW Pacific 
0.012 
SE Pacific 
0.006 
0.003 
SW Atlantic 
0.020 
0.006 
0.001 
NE Atlantic 
0.592 7 
0.042 7 
0.040 7 
0.017 
1 Significant difference (P<0.005). 
cant differences among three Pacific samples, clearly indi- 
cating that microsatellites appear to be more sensitive and 
powerful in detecting more subtle signals of genetic differ- 
entiation in albacore samples than mtDNA analysis. These 
results support several ecological and morphometric stud- 
ies (Nakamura, 1969; Lewis, 1990) which assumed negligi- 
ble migration of albacore across the equator in the Pacific. 
Separate North and South Pacific albacore stocks is a rea- 
sonable assumption because two major spawning grounds 
confined in the western to mid tropical Pacific are spatio- 
temporarily separated (Nishikawa et al., 1985). However, it 
is difficult to explain genetic differentiation between south- 
west and southeast Pacific albacore samples because no 
major spawning ground of albacore has been determined 
in the southeast Pacific. Microsatellite analysis of a sam- 
ple from a different year class and a larger sample size is 
necessary to better define the observed genetic differences 
among the Pacific samples and the South Atlantic sample. 
Acknowledgments 
We gratefully acknowledge R. D. Ward, Commonwealth 
Scientific and Industrial Research Organization, Austra- 
lia, for reading the manuscript. This study was partially 
supported by the Fishery Agency of Japan. 
