Scoles et al.: Global phylogeography of Scomber 
825 
1970). Scomber japonicus and S. scombrus have simi- 
lar larval stage durations: eggs of S. scombrus and 
S. japonicus hatch in less than 6 d, and schooling 
behavior begins when larvae metamorphose (about 
15 mm) which occurs at 24-9 d and 22-9 d, respec- 
tively (Hunter and Kimbrell, 1980; Ware and Lam- 
bert, 1985). Thus the duration of passive planktonic 
transport of the early life history stages is usually 
29 d or less, but probably extends into the early ju- 
venile stages as well. Little spawning or larval ecol- 
ogy data are available for S. australasicus, but it is 
expected that it has a similar early life history. 
Exchange between regional populations of S. 
scombrus in the North Atlantic appears sufficient to 
maintain genetic homogeneity. Two spawning groups 
in the northwest Atlantic, the “northern contingent” 
in the southern Gulf of St. Lawerence, and the “south- 
ern contingent” between Cape Cod and Cape 
Hatteras, were identified by Sette ( 1950) by size com- 
position and tagging data. However, these groups 
were not discriminated by meristic or growth char- 
acter analysis (MacKay and Garside, 1969; Simard 
et al., 1992), or by allozyme analysis (Maguire et al., 
1987). In the northeast Atlantic, two spawning 
groups, the “western stock” south and west of the 
British Isles, and the “North Sea stock,” were iden- 
tified by tagging (Hamre, 1980), but these groups 
were not distinguished by allozymic differences 
(Jamieson and Smith, 1987). 
Matsui (1967) showed greater phenotypic varia- 
tion in S. japonicus than in the other two species of 
Scomber, possibly because of its wider distribution. 
Populations of S. japonicus from the eastern and 
western Atlantic had nonoverlapping distributions 
of gill-raker counts, and other variable morphologi- 
cal characters of Atlantic populations were similar, 
including belly spots, strongly crenulated teeth, and 
large scales. In contrast, Pacific S. japonicus exhib- 
ited lightly crenulated teeth, no belly spots, and 
smaller scales (Matsui, 1967). Analysis of four poly- 
morphic allozyme loci revealed no divergence among 
samples of S. japonicus from the southeast Atlantic 
off Namibia (Zenkin and Lobov, 1989), but heteroge- 
neity in immunological reactivity suggested popula- 
tion structure off the northwest African coast (Weiss, 
1980). A study of 14 polymorphic allozyme loci showed 
significant differences between samples from the 
north- and southeastern Pacific Ocean, which sug- 
gested reduced gene flow across the tropics in com- 
parison with other similarly distributed pelagic fishes 
(Stepien and Rosenblatt, 1996). Differences in growth 
rates and morphological characters among samples 
within the southwest Atlantic led Perrotta (1993) to 
conclude that S', japonicus populations within the 
region had diverged to the level of subspecies. 
To evaluate the genetic relationships among the 
three species of Scomber, and the discontinuous popu- 
lations within them, we examined mitochondrial (mt) 
DNA restriction sites, and cytochrome b sequences. 
Analysis of mtBNA has proven useful in revealing 
phylogeographic structure in a variety of marine and 
freshwater fish species (Avise, 1992). Because 
mtDNA is clonally inherited, information regarding 
historical phylogenetic relationships is retained, 
hence analysis of mtDNA can render considerable 
information on historical relationships among popu- 
lations and the mtDNA lineages they possess. 
Materials and methods 
Specimen collection 
Samples of 15 to 21 individuals each of S. scombrus, 
S. australasicus, and S. japonicus were obtained from 
15 locations (Table 1, Fig. 1). Specimens of Ras- 
trelliger kanagurta, a species of the sister group to 
Scomber (Collette et al., 1984), were obtained from 
Sri Lanka. Whole fish were frozen after collection 
and shipped to the laboratory on dry ice. 
MtDNA preparation and analysis 
MtDNA-enriched genomic DNA was isolated from 
3 g of lateral red muscle, or whole young-of-year fish 
with digestive tracts removed (A-NZL only) accord- 
ing to Chapman and Powers (1984) method, modi- 
fied by the omission of sucrose step gradients and 
the use of 1.5% sodium dodecyl sulfate for mitochon- 
drial lysis. DNAs were digested with 10 hexameric 
(Apal, Bgll, Bsu36l, Oral, Pvull, Seal, Stul, Sspl, 
Hpal, Spe I) and 2 multi-hexameric (Aval, Hae II) re- 
striction endonucleases following manufacturers’ 
(Stratagene and Gibco-BRL) instructions. DNA frag- 
ments were separated by electrophoresis in 1.0 or 
1.5% agarose gels, transferred to nylon membranes 
by Southern transfer, and hybridized to a biotin-la- 
beled probe as described previously (Scoles and 
Graves, 1993). The probe was made by separately 
cloning four yellowfin tuna, Thunnus albacares, 
mtDNA fragments that cover the entire mitochon- 
drial genome in the Pstl site of pBluescript SK- 
( Stratagene). MtDNA fragments were visualized by 
using the BluGene Non-Radioactive Nucleic Acid 
Detection Kit (Gibco-BRL). 
A 12-letter composite mtDNA haplotype, indicat- 
ing the fragment pattern for each enzyme, was de- 
veloped for each individual. Letters were assigned 
to restriction fragment patterns as they were encoun- 
tered, beginning with A’ for the closely related 
