Small et al.: Population and stock indentification of Oncorhynchus kisutch 
855 
Table 6 
Percentage of classifications of individual fish to regions for 34 populations of coho salmon from British Columbia. In this jack- 
knife analysis the fish tested was not included in the discriminant analysis sample, which included the rest of the fish. The 
percentage of fish correctly classified to region is in the “correct” column and the percentage of fish from that region misclassified 
to other regions is read across the row. The percentage of fish from other regions misclassified to a particular region is read down 
the region column, n is the number of fish from the region. 
Region 
Region 
n 
Correct 
Pallant 
C. Coast 
L. Fraser 
Thompson 
WCVI 
ECVI 
NCVI 
Nass 
Skeena 
Total 
Pallant 
92 
48.91 
6.52 
8.70 
5.43 
6.52 
7.61 
13.04 
1.09 
2.17 
100 
C. Coast 
219 
27.85 
10.05 
5.94 
6.85 
4.11 
17.81 
11.42 
7.31 
8.68 
100 
L. Fraser 
1018 
53.73 
4.72 
5.89 
2.85 
3.05 
8.74 
12.87 
5.11 
3.05 
100 
Thompson 
798 
84.96 
1.13 
2.51 
2.63 
0.25 
3.26 
3.01 
1.00 
1.25 
100 
WCVI 
115 
60.00 
6.09 
6.09 
11.30 
2.61 
5.22 
3.48 
2.61 
2.61 
100 
ECVI 
338 
44.08 
5.03 
10.36 
13.61 
2.37 
5.03 
10.06 
6.80 
2.66 
100 
NCVI 
133 
43.61 
4.51 
5.26 
18.05 
6.02 
1.50 
10.53 
6.77 
3.76 
100 
Nass 
133 
20.30 
0.75 
13.53 
14.29 
6.02 
6.77 
8.27 
9.02 
21.05 
100 
Skeena 
412 
52.67 
3.40 
5.58 
6.55 
6.07 
1.94 
6.80 
6.31 
10.68 
100 
sea, or traversed the shifting freshwater waterways 
of northern B.C., to recolonize the upper reaches of 
major watersheds as far south as the Nass and 
Skeena rivers (Lindsey and McPhail, 1986). Dispersal 
of coho salmon from a central B.C. coastal refuge(ia) 
located on the mainland or unglaciated portions of 
Vancouver or the Queen Charlotte Islands (or both) 
(Warner et al., 1982) likely established the hetero- 
geneous mainland-coastal population group, of which 
lower Fraser coho salmon populations may be a dis- 
tinctive offshoot. 
Of the four regional components of biodiversity in 
B.C. coho salmon, the Thompson and upper Fraser 
is the most distinctive. Little introgression has ap- 
parently occurred between the lower Fraser and 
Thompson River coho salmon populations despite 
their common passage through the lower Fraser 
River on their return to spawning grounds for over 
3,000 generations. Coho salmon populations are few 
and small in the Fraser River drainage above its 
confluence with the Thompson River, and our data 
show no evidence of introgression between the coho 
salmon of the Thompson-Fraser Rivers and the up- 
per Skeena watersheds such as that postulated for 
sockeye salmon (Wood et al., 1994). However, our 
sampling of the Skeena watershed is limited to date, 
and the current analysis (in which larger, relatively 
infrequent, alleles have been binned) has limited 
power for the detection of historical gene exchange 
through an analysis of rare alleles. 
The Skeena River watershed has been identified 
as the southern limit for freshwater fish and sock- 
eye salmon that dispersed from a northern glacial 
refuge (Lindsey and McPhail, 1986; Wood et al., 1994; 
Bickham et al., 1995). Thus, it seems likely that the 
coho salmon populations of the upper Skeena and 
Nass watersheds are derived from Beringia. The ge- 
netic intermediacy of lower Skeena and Nass popu- 
lations between upper Skeena-Nass and neighboring 
coastal populations suggests a hybrid nature for the 
lower river populations. The extent to which the com- 
posite nature of these populations reflects historical 
or current gene flow (or both) between the two found- 
ing groups, and the adaptive consequences of such 
introgression, has yet to be determined. 
The coho populations of the lower Fraser drainage 
basin formed a cohesive genetic group in the NJ 
analysis of genetic distance in this study. This was 
in sharp contrast to the heterogeneity observed 
among other coastal mainland and island populations 
likely derived from one or more coastal refugia. The 
heterogeneity of central coast populations may re- 
flect the existence of several coastal refugia, intro- 
gression from northern coho populations originating 
from Beringia that have yet to be well characterized, 
or simply founder effects in the establishment of in- 
dividual coastal populations from a single refuge, as 
postulated for sockeye salmon (Wood et al., 1994). If 
the heterogeneity does result from an admixture of 
several founding groups among coastal coho popula- 
tions, the lower Fraser River populations may bet- 
ter represent the origin al genetic profile of fish from 
a single, possibly southern, coastal refuge. Interest- 
ingly, Vancouver Island coho salmon were more fre- 
quently misclassified as lower Fraser than as cen- 
tral coast fish in the mixed-stock fishery simulations 
of our study, in spite of their apparently greater ge- 
netic similarity to central coast populations. If the 
