838 
Fishery Bulletin 96(4), 1998 
vided no evidence for the hybrid origin of individu- 
als of intermediate coloration (Kijima et al., 1986). 
The systematic status of geographically distant S. 
japonicus populations has been problematic because 
this species is characterized by considerable morpho- 
logical variability. Several names have been used for 
the groups occurring in different areas: S. japonicus, 
western North Pacific (Houttuyn, 1782); S. colias, 
eastern Atlantic (Gmelin, 1789); S. grex, western 
North Atlantic (Mitchell, 1815); S. diego, eastern 
North Pacific (Ayres, 1857); S. peruanus, eastern 
South Pacific (Jordan and Hubbs, 1925); and 
Pneumatophorus japonicus marplatensis, western 
South Atlantic (Lopez, 1955). Differences in pigmen- 
tation, tooth crenulation, scale size, and gill-raker 
counts on the lower first arch distinguish fish from 
the Pacific, western Atlantic, and eastern Atlantic. 
These groups were synonymized into the single spe- 
cies S. japonicus by Matsui (1967) because “[morpho- 
logical] differences are smaller [between populations] 
than those which separate sympatric species of 
Scomber and Rastrelliger , and it thus seems correct 
to regard S. japonicus as a single polytypic species.” 
The mtDNA data did not refute the specific status 
of S. japonicus and S. australasicus reported by 
Matsui ( 1967). In fact, a greater divergence was found 
among two mtDNA lineages within S. australasicus 
than between any other pairs of samples of S. 
japonicus and S. australasicus. The two divergent 
S. australasicus mtDNA lineages, which differed by 
an average of 8=1.84% according to the restriction 
site data, were equally represented in both South 
Pacific collections. A cursory allozyme analysis of 10 
loci from 10 individuals of each lineage revealed no 
significant allelic differences in nuclear encoded 
genes (Scoles, 1994). The phylogenetic pattern ob- 
served in S', australasicus appears to be the result of 
postdivergence introgressive hybridization. 
A similar phylogenetic pattern has been observed 
in other pelagic marine fishes. In blue marlin 
( Makaira nigricans), “ubiquitous” haplotypes occur- 
ring in both the Atlantic and Pacific oceans were 
highly divergent from other Atlantic haplotypes, 
forming a “unique” clade revealed by mtDNA analy- 
sis restriction fragment analysis (n=114, 8=0.15%, 
Graves and McDowell, 1995) and by sequencing 612 
bp of the mtDNA cytochrome b gene (rc= 26, 8=1.6%, 
Finnerty and Block, 1992). Similarly, sailfish 
( Istiophorus platypterus) had Pacific and Atlantic 
clades that differed by 8=0.27% (Graves and 
McDowell, 1995). The patterns in these fishes and 
in S. australasicus demonstrate that considerable 
mtDNA divergence can occur among haplotypes 
within a single sample, and suggest the limitations 
of basing taxonomic status only on mtDNA data. 
The mtDNA data indicated the possibility of mul- 
tiple taxonomic lineages within S. japonicus. The two 
major groups of S', japonicus (Atlantic and Pacific) 
are differentiated by nearly as much as are the two 
different lineages within S. australasicus. The nucle- 
otide sequence divergence between some S. japonicus 
groups is nearly as great as between groups of S. 
japonicus and S. australasicus (Table 4). Unlike the 
two divergent S. australasicus lineages that were 
observed in single samples, the S. japonicus groups 
are geographically partitioned. The paraphyletic re- 
lationship (Figs. 3-5), geographic isolation, high level 
of genetic divergence (Table 4), and morphological 
differences among the S. japonicus lineages support 
recognition of separate species: Scomber japonicus 
Houttuyn, 1782, in the Pacific; and Scomber colias 
Gmelin, 1789, the Atlantic. This conclusion assumes 
that there is no significant sampling error (see Moore, 
1995) and should be tested by further morphological 
and nuclear DNA data analyses. The Red Sea popu- 
lation of Scomber, previously considered as S. 
japonicus (Matsui, 1967), was re-examined by Baker 
and Collette ( 1998), who have assigned the Red Sea- 
northern Indian Ocean population to Scomber 
a ustralasicus on the basis of morphological characters. 
Phylogeographic patterns and their origins 
The distribution of haplotypes between eastern and 
western Atlantic samples of S. scombrus indicated 
the populations do not share a common gene pool. 
The close relationship among haplotypes in these 
samples (Fig. 2) contrasts with the finding of two 
divergent mtDNA matrilines (8=3.7%) in samples of 
capelin, Mallotus villosus, from eastern and west- 
ern regions of the North Atlantic (Dodson et al., 1991). 
An intermediate level of divergence to that of <S. 
scombrus and of capelin was observed between 
samples of bluefish ( Pomatomus saltatrix, 8=0.26%, 
1 fixed difference) from the eastern and western At- 
lantic Ocean (Goodbred and Graves, 1996). The lower 
divergence in S. scombrus may be the result of more 
recent isolation or greater vagility. 
Patterns of divergence in S. japonicus and other 
species in the Atlantic Ocean appear to be highly 
influenced by species’ vagility and warm water in the 
tropics. Haplotype distributions were significantly 
different between S. japonicus samples from the 
western Atlantic Ocean (J-FLA and J-ARG), and 
unique haplotypes occurred within each at relatively 
high frequencies (10-20%). Similary, a population- 
genetic study of the shortfin mako shark demon- 
strated significantly different mtDNA haplotype fre- 
quencies between samples from the western North 
and South Atlantic Ocean (8=0.13%,P<0.001) (Heist 
