Scoles et al.: Global p.hylogeography of Scomber 
839 
et al., 1996). Bluefish sampled from Brazil and the 
eastern U.S. coast were even more divergent 
(8=1.48%) (Goodbred and Graves, 1996). But in the 
eastern Atlantic, no significant differentiation was 
seen among S. japonicus samples from the eastern 
Mediterranean Sea (J-ISR), Ivory Coast (J-IVC), and 
South Africa (J-SAF). Additionally, although no 
haplotypes were shared between bluefish samples 
from the Mediterranean Sea and South Africa 
(8=0.35%), haploty pes in the two samples were closely 
related (Goodbred and Graves, 1996). In contrast, the 
less mobile gray mullet, Mugil cephalis, sampled from 
the Mediterranean Sea, and South Africa were highly 
divergent (8=2.92) (Crosetti et al., 1994). These data 
suggest that fewer barriers to gene flow occur be- 
tween northern and southern regions of the eastern 
Atlantic versus those of the western Atlantic. The 
tropical water of the east Atlantic occupies the small- 
est area of all the tropical regions, and temperate 
species are suspected of crossing this region by mov- 
ing under the warm water mass, thereby achieving 
an antitropical distribution (Briggs, 1974). 
It is possible that S. japonicus larval or adult ex- 
change occurs between Gulf of Guinea and the north- 
ernmost extension of the southwest Atlantic S. 
japonicus population assisted by trans-Atlantic equa- 
torial currents. These locations are separated by the 
most narrow region of the Atlantic Ocean. Haplotype 
frequencies were significantly different among all 
samples across the Atlantic Ocean except in one 
pairwise comparison of J-ARG and J-IVC. Genetic 
similarity between these samples may reflect recent 
isolation or contemporary gene flow. Eastern and 
western Atlantic populations of S. japonicus are, 
however, morphologically differentiated on the ba- 
sis of nonoverlapping gill-raker counts (Starks, 1921; 
Matsui, 1967). 
The sample of S. australasicus from the Red Sea 
(A-RED) had the highest estimated haplotypic diver- 
sity. No haplotypes were shared between the samples 
of S. japonicus from the Atlantic Ocean and the Red 
Sea, suggesting no gene flow through the Suez Ca- 
nal. Fifty-two fish species are known to have made 
the passage, all but one from the Red Sea to the Medi- 
terranean Sea (Golani, 1993). Rastrelliger kanagurta , 
a close relative of Scomber, is among those known to 
have invaded the Mediterranean Sea from the Red 
Sea (Collette, 1970). The samples of S. japonicus from 
the eastern Mediterranean Sea (J-ISR) and S. 
australasicus from the Red Sea were divergent by 
8=0.72%, one fixed restriction site difference, and an 
additional site that was nearly fixed. These findings 
do not exclude the possibility of movement through 
the Suez Canal, which may be revealed by larger 
sample sizes. 
Fishery data indicate that S. japonicus in the re- 
gion of Japan and Taiwan represent separate popu- 
lations with unique spawning grounds and larval 
retention areas (Sato, 1990). Although haplotypes in 
the sample from the eastern Pacific (J-CAL) were 
not found in other S. japonicus samples, haplotypes 
in S. japonicus samples from the northwest region 
of the Pacific Ocean ( J-JPN and J-TWN ) were shared 
at similar frequencies. The data of our study are con- 
sistent with exchange among the two western Pa- 
cific populations. Whether exchange is the result of 
larval drift or adult movement is indeterminable; 
however, movement of adult S. australasicus is 
known to occur between these regions (Chang and 
Wu, 1977). Greater trans-Pacific divergence than that 
observed in S. japonicus was seen in gray mullet 
sampled from Hawaii and the Galapagos Islands 
(8=1.4%) (Crosetti et al., 1994), suggesting these 
population were isolated for a considerably longer 
period than mackerel. 
The level of nucleotide sequence divergence be- 
tween Atlantic and Pacific samples of S. japonicus 
was lower than expected. It has been suggested that 
in mammals, mitochondrial DNA evolves at rate of 
2% nucleotide sequence divergence per million years 
(Brown et al., 1979). When this rate of divergence is 
applied to the 1.4% net nucleotide sequence diver- 
gence observed between Atlantic and Pacific S. 
japonicus, it appears that the two populations were 
isolated about 700,000 years. Such estimates of di- 
vergence are approximate at best, but if this rate of 
divergence is valid in Scomber japonicus, the ob- 
served nucleotide sequence divergence is lower than 
would have resulted if these populations were isolated 
by the uplift of the Isthmus of Panama, 3. 1-3.6 million 
years ago ( Keigwin, 1978 ). Gene flow may have occurred 
between the two regions since the uplift of Panama, 
during times of warmer climates (Avise, 1992). 
Samples of S. australasicus between New Zealand 
and Australia, regions separated by nearly 2000 km 
of deep ocean, were genetically homogeneous. Per- 
haps there is a high potential for genetic exchange 
in S. australasicus, which was also indicated by com- 
parison of S. australasicus samples from the north- 
ern Pacific. Although a test of homogeneity between 
A-JPN and A-MEX was marginally significant 
(P=0.013), the mean nucleotide sequence divergence 
was only 0.02%, which contrasted with the 0.30% 
divergence and fixed restriction site difference be- 
tween samples of S. japonicus from the eastern and 
western Pacific Ocean. 
The lowest genotypic and nucleotide sequence di- 
versity in the study occurred in a sample from the 
Revillagigedo Islands (A-MEX) and may indicate the 
origin of this population by means of a founder event 
