Scoles et al.: Global phylogeography of Scomber 
831 
Table 3 
Probabilities of significance from Roff and Bentzen’s (1989) chi- 
square analysis with 1000 randomizations of the data of samples 
of Atlantic Scomber japonicus. Pacific S. japonicus , S. 
australasicus, and S. scombrus that shared haplotypes. 
Comparison 
X 2 
Number of 
simulations 
exceeding yj 
Probability 
Atlantic Scomber japonicus 
J-ISR J-IVC J-SAF 
J-FLA J-ARG 
191.36 
0 
<0.001** 
J-ISR J-IVC J-SAF 
48.70 
451 
0.451 ns 
J-ARG J-IVC J-SAF 
60.84 
28 
0.028* 
J-ARG J-IVC 
22.21 
68 
0.070 NS 
J-ARG J-SAF 
26.89 
0 
0.001** 
J-FLA J-ARG 
28.37 
1 
0.001** 
J-FLA J-ISR J-IVC 
58.40 
16 
0.016* 
J-FLA J-ISR 
32.00 
1 
0.001*’ 
J-FLA J-IVC 
24.33 
14 
0.014* 
Pacific Scomber japonicus 
J-JPN J-TWN 
10.67 
658 
0.658 NS 
Scomber australasicus 
A-JPN A-MEX A-AUS 
A-NZL 
128.94 
0 
<0.001** 
A-JPN A-MEX 
17.75 
13 
0.013* 
A-AUS A-NZL 
11.67 
718 
0.718 ns 
Scomber scombrus 
S-ENG S-MAS 
20.17 
0 
<0.001** 
Significantly different at a=0.05. 
’’Significantly heterogeneous at a=0.01. 
NS Not significant. 
Scomber japonicus Of the three species, 
samples of S. japonicus spanned the greatest 
geographical area, and mtDNA restriction site 
analysis of this species revealed the largest 
range of divergences. In the North Pacific, 
samples from Japan (J-JPN) and Taiwan (J- 
TWN) shared four haplotypes, three of which 
occurred in more than one individual in each 
sample (nos. 1-3, Table 2), and no significant 
heterogeneity was observed between the 
samples (P=0.658, a=0.05, 5=-0.010%). In con- 
trast, a comparison of the pooled data of J-JPN 
and J-TWN with S. japonicus from the eastern 
North Pacific ( J-CAL) revealed one fixed restric- 
tion site difference, a net nucleotide sequence 
divergence of 5=0.30%, and genotype distribu- 
tions were significantly different (PcO.001, 
cc=0.05). 
Many closely related haplotypes were ob- 
served in samples of S. japonicus from the At- 
lantic and Mediterranean Sea, and several were 
shared among two or more sample locations 
(Table 2). No significant differences were ob- 
served in the distribution of haplotypes among 
J-ISR, J-IVC, and J-SAF, from the eastern At- 
lantic (P=0.451, a=0.05, 5=-0.0Q3%-0.015%, 
Table 4), or between J-IVC and J-ARG across 
the Atlantic (P=0.07, a=0.05, 5=0.025%). All 
other tests of haplotype frequencies among At- 
lantic samples were significant (P< 0.05). 
sample. Although the net nucleotide sequence diver- 
gence between A-JPN and A-MEX was small 
(8=0.021%), the distribution of haplotypes between 
the samples was significantly heterogeneous 
(P=0.013, a=0.05). Together, the Australian and New 
Zealand samples of S. australasicus from the South 
Pacific were genetically distinct from the combined 
North Pacific samples (A-JPN and A-MEX). 
The mean corrected nucleotide sequence divergence 
between the pooled groups was high (8=0.54%), 
and only one haplotype was shared between the 
pooled South and North Pacific samples (no. 79), oc- 
curring at very different frequencies (0.05 and 0.51, 
respectively). 
The Red Sea sample of S. australasicus was dis- 
tinct from all other collections of this species. The 
mean net nucleotide sequence divergence between 
A-RED and the pooled data of all other S. aus- 
tralasicus was 8=0.86%. The Red Sea sample was less 
divergent from the S. australasicus samples from 
Australia and New Zealand (8=0.51%) than were the 
S. australasicus samples from Japan and Mexico 
( 8 = 1 . 2 %) 
Phylogeographic patterns 
MtDNA restriction site analysis and neighbor-join- 
ing of S. japonicus and S. australasicus mtDNA 
haplotypes revealed five major clusters: S. japonicus 
from the Pacific Ocean, S. japonicus from the Atlan- 
tic Ocean, S. australasicus from the Red Sea, a “ubiq- 
uitous” cluster of haplotypes of S. australasicus from 
all other samples of this species, and a “unique” clus- 
ter of only New Zealand and Australia S. austral- 
asicus haplotypes (Fig. 3). 
Parsimony analyses were used to explore cladistic 
relationships among the haplotypes of S. japonicus 
and S. australasicus identified by mtDNA restriction 
site analysis. Parsimony analysis among all Scomber 
haplotypes could not be conducted because S. 
scombrus mtDNA sequences were so divergent that 
it was not possible to determine homologous restric- 
tion site characters among the three species. Mainly 
interested in patterns occurring in S. japonicus, we 
chose a root in S. australasicus (haplotype no. 92) on 
the basis of results of parsimony analyses of cyto- 
chrome b sequences that included S', scombrus and a 
