Gharrett et al.: Phylogeographic analysis of mitochondrial DNA variation in Oncorhynchus kisutch 
537 
Figure 5 
Comparison of the effective number of haplotypes (n c ) and the aver- 
age nucleotide divergence between haplotypes within a species for 
25 species of fish estimated from haplotype diversity {h ). The number 
of effective haplotypes is monotonically related to N c ]i (p is the muta- 
tion rate); larger average nucleotide divergences require longer times 
and reduced gene flow to develop. The following species were plotted: 
A = red grouper ( Richardson and Gold, 1993, 1997); B = white marlin 
(Graves and McDowell, 1995); C = hardhead catfish (Avise et al., 
1987, 1989); D = American eel (Avise et al., 1986, 1989); E = oyster 
toadfish (Avise et al., 1987, 1989); F = American shad (Bentzen et al., 
1989); G = redear sunfish (Avise et al., 1987, 1988, 1989, 1992); H = 
Sicyopterus stimpsoni (Zink et al., 1996); I = lake whitefish (Bernat- 
chez and Dodson, 1991); J = red snapper (Gold et al., 1994, 1997); 
K = coho salmon (Carney et al., 1997); L = black drum (Richardson 
and Gold, 1993; Gold et al., 1994); M = spotted sunfish (Avise et al., 
1987, 1989, 1992; Avise, 1992); N = coho salmon (this study); O = NW 
Atlantic capelin (Dodson et al., 1991); P = bowfin (Avise et al., 1987, 
1989, 1992; Avise, 1992); Q = striped marlin (Graves and McDowell, 
1994; Graves and McDowell, 1995); R = Stenogobius hawaiiensis 
(Zink et al., 1996); S = Lentipes concolor (Zink et al., 1996); T = 
greater amberjack (Richardson and Gold, 1993); U = Awaous gua- 
mensis (Zink et al., 1996); V = Atlantic herring (Kornfield and Bog- 
danowicz, 1987; Richardson and Gold, 1993); W = largemouth bass 
(Nedbal and Phillipp, 1994); X = red drum (Gold and Richardson, 
1991; Gold et al., 1993); Y = warmouth (Avise et al., 1987, 1989, 1992; 
Avise, 1992); Z = NE Atlantic capelin (Dodson et al., 1991). Estimates 
for coho salmon in this study (N) and our previous study (K) are 
circled. 
estimates of nucleotide divergence were inflated 
more than threefold. However, the data remain 
useful for examining intraspecific structure, and 
strong heterogeneity among collections indicated 
structure among populations or at higher levels 
of population hierarchy. 
The unrooted tree (Fig. 4) depicting relation- 
ships among collections shows that the Bering 
Sea and the Karluk River collections cluster sep- 
arately from the other collections. The cluster- 
based structure of the “gene tree” accentuates 
the divergence between geographic regions be- 
cause the highest abundance of E-cluster haplo- 
types occurs in the Bering Sea (and Karluk) col- 
lections. In fact, the Delta Clearwater collection is 
fixed for E-cluster haplotypes. The AMOVA anal- 
yses revealed a population structure that appears 
stronger within regions than among regions. Both 
estimates of gene flow (one to two females per 
generation) are low in relation to allozyme-based 
estimates of other Pacific salmon species (i.e. an 
exchange of 6.6 to 16.4 individuals per genera- 
tion for pink salmon [McGregor, 1983; Beacham 
et al., 1985; Noll et al., 2001]; 6.3 to 9.0 for chum 
salmon [Kondzela et al., 1994; Phelps et al., 1994; 
Wilmot et al., 1994; Winans et al.. 1994]; 1.3 to 7.1 
for sockeye salmon [Wood et al., 1994; Varnavs- 
kaya et al., 1994]; and 1.2 to 4.0 for chinook salm- 
on [Gharrett et al., 1987; Utter et al., 1989; Bart- 
ley and Gall, 1990] ) but consistent with estimates 
from coho salmon allozyme data (Wehrhan and 
Powell, 1987; Bartley et al., 1992). 
It is essential to keep in mind that such gene- 
flow estimates assume an equilibrium between 
gene flow and random drift. The distribution of 
coho salmon haplotypes suggests that a broad 
equilibrium may not exist. For example, the Delta 
Clearwater River population was nearly fixed for 
haplotype H, an E-cluster haplotype found only in 
the Bering Sea populations. Its haplotype compo- 
sition reflects its geographic isolation from other 
coho populations. Also, the haplotype distribution 
from the Kamchatka River population was sur- 
prising because it is so similar to Southeast Alas- 
kan populations, but differed from the geographi- 
cally closer Bering Sea populations. This makes 
sense if the genetic structure of extant coho salm- 
on populations is strongly influenced by historic 
events, such as historic random drift, colonization 
from eastern populations, or survival in an differ- 
ent glacial refugia, and that an equilibrium between gene 
flow and random drift has not yet been reached. 
Synthesis 
The data obtained from restriction site analysis provide a 
present day “snapshot” of coho salmon mtDNA variation 
and have two levels of resolution. One level is the geo- 
graphic distribution of haplotypes and the haplotype com- 
positions of the populations sampled. The other level is 
the pattern and extent of divergence among the hap- 
lotypes. The mtDNA haplotype compositions of popula- 
tions indicate that coho salmon, at least in their Alaskan 
range, generally have lower gene flow than other Pacific 
salmon species, although the comparisons assume equi- 
libria between gene flow and random drift that may not 
yet have been reached. Coho salmon exhibit divergence 
among populations within regions, but generally not fixed 
differences. Many of the haplotypes were found in popu- 
lations throughout the range examined, which suggests 
