104 
Fishery Bulletin 99(1) 
59% 
6 1 °< 
J87% 
\ 54% 
70% 
Quesnel 1 2 MF 
Stuart 13 MF 
Nechako 14 MF 
_ Chilko 11 MF 
Lower Shuswap 8 TR 
70% L___' Middle Shuswap 9 TR 
Eagle 10 TR 
59% 
Coldwater 6 TR 
Nicola 7 TR 
Bridge 5 MF 
■fr 
67% 
Tete Jaime 18 LTF 
Red Chilliwack 3 UF 
Slim 20 UF 
Indianpoint 19 UF 
Red Chehalis 4 UF 
— ■— — Cottonwood 16 MF 
Holmes 17 UF 
■ Bowron 15 UF 
Hanison 1 LF 
Wlrite Chilhwack 2 LF 
Genetic distance 0.03 
i 
0.02 
—I — 
0.01 
—I — 
Figure 3 
Neighbor-joining dendrogram of 20 chinook salmon populations 
from the Fraser River watershed. The dendrogram shows branch 
lengths, and the bootstrap values at the nodes were the percent- 
age of dendrograms in which the populations beyond the node were 
grouped together out of 1000 dendrograms. The numbers follow- 
ing the place names correspond to map numbers in Figure 1. “MF”, 
“TR, “UF” and “LF” indicate middle Fraser, Thompson River, upper 
Fraser and lower Fraser River respectively. 
all loci combined. All values were significantly 
greater than zero (P<0.005), indicating that pop- 
ulations tended towards disequilibrium at each 
locus. Gametic disequilibrium tests indicated that 
loci were unlinked. 
Because of a suspected null allele at Ots 102, 
most populations showed a deficiency of heterozy- 
gotes at this locus; this locus also had a high and 
significant F is value. When the dendrogram analy- 
sis was redone with the corrected data set, the re- 
gional structure was improved, in that all the mid- 
dle Fraser River populations grouped together and 
bootstrap values were higher for regional nodes. 
Thus the major conclusion of this study was un- 
affected by the suspected null allele. However be- 
cause the null allele was not observed in the pedi- 
gree analysis, its presence remains speculative and 
we have kept the original data set for the analysis 
presented here. Observed homozygote excesses are 
likely not due to our inability to resolve alleles be- 
cause we observed a large number of individuals 
that amplified at the other two loci of the study but 
failed to amplify at Ots 102, suggesting the pres- 
ence of a null homozygote. Observed heterozygote 
deficiency could also be due to a partial disequi- 
librium; however no single population showed dis- 
equilibrium at all three loci. 
Temporal stability of allele frequencies, and 
population and regional variability 
Comparisons of samples from different years for the 
same population suggested that there is temporal 
stability in allele frequencies within populations. 
With the exception of a single comparison within 
the Harrison River samples (the 1988 samples dif- 
fered from the 1992 samples [P nondifferentia- 
tion=0.0003]), sample sets within populations were 
indistinguishable. Analysis of molecular variance showed 
that 97.57% (P<0.01, 5719 df) was within populations and 
that 1.80% (P<0.01, 16 df) of the total genetic variance was 
among populations. A small but significant 0.63%, (P<0.01, 
3 df) of the genetic variance was apportioned between 
regions. Because population year classes were not signif- 
icantly different from each other and variability among 
populations exceeded variability among year classes, year 
classes were combined into single populations for the rest 
of the analysis. All regions were significantly different from 
each other in pairwise tests (P<0.008). 
Dendrogram 
To determine if there was a pattern to allele frequency 
differences, we constructed a NJ dendrogram (Fig. 3) of 
pairwise genetic distances. This analysis suggested that 
geography at least partially underlies genetic relation- 
ships among chinook salmon populations in the Fraser 
River drainage. The NJ dendrogram consisted of four ma- 
jor branches, and are congruent with regional groupings 
(Fig. 3). The lower Fraser River branch includes Harrison 
River and Chilliwack and is supported in 96% of the den- 
drograms. The Thompson River branch is supported at 
70% and includes Bridge River (from the middle Fraser 
River) in 59% of the dendrograms. The middle Fraser River 
branch is supported in 59% of the dendrograms. The larg- 
est branch, supported in 32% of the dendrograms, includes 
all the upper Fraser River populations, as well as the Chill- 
iwack-red and Chehalis-red hatchery populations, and Cot- 
tonwood River from the middle Fraser River region. 
This graphic analysis of genetic relationships shows re- 
gional groupings consisting of lower, middle, and upper 
Fraser River population groups, and a well-defined Thomp- 
son River group. These groupings are modestly supported 
because most of the genetic variance is within populations 
and only a small amount of the variance is among regions. 
Population and regional differences 
Wright’s F statistics (Wright, 1951) were calculated to 
determine the degree of structuring between and within 
the regional population groupings. When all the popula- 
tions were kept separate, the F st values indicated diver- 
