52 
Fishery Bulletin 107(1 ) 
Figure 4 
Dissimilarity matrix of 0 values for genetic distance data derived from Chinook salmon ( Oncorhynclius tshawytscha) populations 
from the west coast of Vancouver Island. Values of 6 where 8= 0 are shown as white, 8 < 0.02 as light gray, 8 < 0.04 as dark 
gray, and 8 > 0.04 as black. Populations that show genetic affiliation but are outside the geographic region are denoted by < >. 
ber of groups occurs at k = 4, corresponding to four 
geographic locations: Quatsino Sound, Nootka Sound, 
Clayoquot+Barkley sounds, and southwest Vancouver 
Island (Fig. 6C). At this point, additional groups do 
not cause the observed data to continue to drop sub- 
stantially below the reference distribution. When k=9, 
a second peak in the optimum groupings oc- 
curred which corresponded to the partitioning 
of Barkley Sound and Clayoquot Sound popula- 
tions, similar groupings were derived from the 
UPGMA tree; the vertical line in Figure 3B 
indicates the corresponding number of groups 
and group membership as determined by bi- 
PORGS. The four regional groups, Quatsino 
Sound, Nootka Sound, Clayoquot + Barkley 
sounds, and southwest Vancouver Island, each 
formed a cluster on the UPGMA tree; however, 
the optimum number of clusters is not obvious. 
The dendrogram appears to show greater genet- 
ic distance between the populations of Marble 
River and Colonial River than between popula- 
tions of Nootka Sound and Clayoquot+Barkley 
sounds. Overall, the partition and agglomera- 
tive methods produced similar results. 
Discussion 
This article provides a new method for clustering 
genetic distance data by partitioning optimally 
with RGS, where acceptable partitions reduce 
intracluster distance. For this analysis, we used 
1.E+13 -| 
1.E+12 - 
1.E+11 ■ 
$ 1.E+10- 
c 1.E+09- 
/ 
g 1.E+08- 
T -♦ \ 
/ ^ \ 
o 1.E+07- 
-g 1.E+06- 
1 '* \ \ 
a3 1.E+05 - 
T v 4 * \ 
E 1.E+04- 
/ AV '■ \ 
Z 1.E+03 - 
If* V \ 
1.E+02 - 
/ / 'y \ \ 
1.E+01 - 
Jf 4 \_ ♦ \ 
1 . t+UU 1 
1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 
Number of groups (k) 
Figure 5 
Number of occurrences by k groups for the simulated data set 
( A ) and for the Chinook salmon (Oncorhynclius tshawyts- 
cha) data set (n = 19) with random search and 5.0 x 10 8 iterations 
and the total number of all set partition occurrences 
required for 19 populations (Stirling numbers of second kind 
( ■ ). 
