Noll et al.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 
125 
Figure 1 
Map of location sites where fish were collected for tissue samples. Unless otherwise indicated, one collection of 
fish from the lower river was made. Japan: 1 = Yurappu River, Yacumo Hatchery (22 Sep 1990), 2 = Kushiro 
Hatchery (29 Sep 1990), 3 = Tokushibetsu Hatchery (17 Sep 1990). Sakhalin Island, Russia: 4 = Lutoga River, 
lower river (13 and 17 Jul 1990) and hatchery (29 Jul 1990 and 8 Nov 1990), 5 = Dolinka River (6 and 16 Aug 
1990), 6 = Ochepukha River, lower river (8 and 22 Aug, and 3 Sep 1990) and upper river (6 and 9, and 11 Sep 
1990) and Znamenka River, a major tributary of Ochepukha River ( 13, and 26 Aug 1990, 3, 6, , 9, 11, 13, 17, and 
19 Sep 1990), 7 = Monetka River ( 18 and 19 Jul 1990). Magadan region, Russia: 8 = Tauy River (26 Jul 1990), 9 
= Uglekanka River (3 Aug 1990). Western Kamchatka Peninsula, Russia: 10 = Vorovskaya River (31 Jul 1990), 
11 = Kol River (30 Jul 1990), 12 = Pymta River (29 Jul 1990), 13 = Utka River (29 Jul 1990), 14 = Bistraya 
River (3 Aug 1990). Eastern Kamchatka Peninsula, Russia: 15 = Ossora River (28 Jul 1990), 16 = Karaga River 
(2 Aug 1990). Alaska: 17 = Prince William Sound, Annin F. Koernig Hatchery, Evans Island, Sawmill Bay, five 
collections in 1990. 
locus were tested for departure from expected Hardy- 
Weinberg equilibrium frequencies by using a Pearson x 2 - 
goodness-of-fit test. Alleles were pooled to eliminate class- 
es with expected values of less than four. Isoloci could not 
be tested for Hardy-Weinberg equilibrium because single- 
locus data were not available. 
Homogeneity of allele frequencies was examined by us- 
ing log-likelihood ratio analysis (G-test, Sokal and Rohlf, 
1995); tests were performed among collections within a 
river, among rivers within a region, and among regions. 
The regions were Japan (Hokkaido Island), Sakhalin Is- 
land, western Kamchatka, eastern Kamchatka, and Alas- 
ka (Prince William Sound). Significance of tests that indi- 
cated low probability (P<0.G5) based on y 2 distributions 
and that had small expected numbers in rare classes ( <4 ) 
were confirmed by using a Monte-Carlo procedure analo- 
gous to the one described by Roff and Bentzen ( 1989). Sig- 
nificance of multiple tests was corrected according to Coo- 
per (1968). Heterogeneity within and among regions was 
compared by using an approximate F-statistic: 
^ 'df among, df within ~ ^ among^f among^''^ within^f within ^ 
Neighbor-joining trees (Saitou and Nei, 1987) were con- 
structed by using chord distances (Cavalli-Sforza and Ed- 
wards, 1967; Wright, 1978). Genetic variability within and 
among streams, regions, and continents was partitioned 
hierarchically by gene diversity analysis (Chakraborty 
and Leimar, 1987) and analysis of variance (Weir and 
Cockerham, 1984; Weir, 1996). Average unbiased heterozy- 
gosities and their standard errors were calculated accord- 
ing to Nei ( 1978). 
The five collections from Prince William Sound were, 
in a strict sense, temporal collections from the mixed fish- 
ery in the Sound and were not intended to represent col- 
lections from discrete drainages. Nevertheless, for conve- 
nience of analysis, they were treated as unique collections 
from a single system in the genetic distance and gene di- 
versity analyses. 
Results 
A total of 54 genetic loci were scored. The 17 loci with fre- 
quencies of the common allele less than 0.95 in at least 
