Noll et a!.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscho 
135 
North Pacific coast line was severely altered, reducing 
the sizes of the enclosed basins as well as the circulation 
among them. 5 18 0 records (e.g. Bartlein et al., 1991; Roh- 
ling, 1998), and limited data from Beringia (Hopkins, 1982) 
suggest that during most of the mid-Wisconsin glaciation, 
sea level was 50 m lower than at present and that at the 
LGM, the level was 120 to 130 m lower in areas not af- 
fected by ice loading or tectonic activity (Fairbanks, 1989). 
Sakhalin Island was connected to the mainland and Hok- 
kaido Island, blocking the northern outflow of the Sea of 
Japan; and the connection of a smaller Sea of Okhotsk with 
the Pacific Ocean through the Kurile islands was restrict- 
ed. In the Bering Sea, the extensive continental shelf was 
exposed, blocking circulation through the Bering Strait. To 
the south, the Aleutian Islands and the Alaska Peninsula 
were joined to about 170°W, and many of the islands to the 
west were also connected, thus limiting the connections be- 
tween the Gulf of Alaska and the Bering Sea. In Cook In- 
let, glaciers may have remained advanced throughout the 
middle Wisconsin glaciation (Reger and Updike, 1983). 
Pink salmon spawning habitat along the Gulf of Alaska 
was nearly completely eliminated at the LGM and much 
of the Beringian coastline was probably ice bound much of 
the year, minimizing available freshwater habitat. Coastal 
areas of the Gulf of Alaska, including most of the continen- 
tal shelf, were extensively glaciated during the Quaterna- 
ry Period, although a few isolated areas of the outer coast 
may have been ice free (Hamilton, 1986); and the ice cover 
of the eastern Aleutians coalesced with the Alaska Penin- 
sula and ice caps covered all the major islands (Thorson 
and Hamilton, 1986). In these areas there may have been 
ephemeral streams from melted snow or ice at the south- 
ern margin, near the present shelf break. Although the 
rivers draining the Yukon-Kuskokwim region flowed over 
the exposed shelf and probably served as a refugium, the 
Bering Sea appears to have had sea ice cover much of 
the year (Sancetta, 1983; Sancetta and Robinson, 1983); 
and seasonal sea ice may have persisted as far south as 
54°N for 6-8 months a year during the LGM (de Vernal 
and Pedersen, 1997). On fhe Asian side, glaciation was 
much less extensive and included some alpine glaciation; 
but few glaciers extended to tidewater, except possibly on 
the southeast side of the Kamchatka Peninsula (Bespalyy, 
1984; Velichko, 1984; Anderson, 1981). Freshwater habi- 
tat was probably not reduced to the extent of the Gulf of 
Alaska coast. Nevertheless, the Sea of Okhotsk, like the 
Bering Sea, probably had sea ice cover much of the year 
(Sancetta, 1983; Sancetta and Robinson, 1983). 
Harsh conditions greatly reduced marine surface water 
productivity over the entire region (Morley et al., 1991; 
Keigwin et al., 1992; deVernal and Pederson, 1997). Micro- 
fossils indicate that the subarctic Pacific Ocean was sim- 
ilar to the present day Sea of Okhotsk, with cold fresh 
surface water and a highly stratified water column. Sea 
surface temperatures estimated by CLIMAP were 2° to 
4°C colder than present throughout the year over most 
of the area (Moore et al., 1980). Off Japan, temperatures 
were even colder (>6°C) at the LGM, indicating that the 
cold Oyashio Current penetrated farther south than it 
does at present (Moore et al., 1980). 
Most of the description above considered the LGM, but 
5 18 0 records indicate that periodically other major glacial 
advances occurred at about 135,000 BP ( 18 0 stage 6), 
about 225,000 BP ( 18 0 stage 8), and so forth (Bartlein et 
al., 1991). Another less extensive advance may have oc- 
curred in the North Pacific region about 75 ka BP (Hop- 
kins, 1982). It is likely that those events also influenced 
the distribution and demographics of salmon species in- 
cluding pink salmon. 
Many of the streams populated by pink salmon are short 
coastal streams that are transient. Consequently, in maxi- 
mizing productivity opportunities, pink salmon may exhib- 
it a higher level of gene flow (straying) than other Pacific 
salmon (Quinn, 1984). Geologic evidence suggests that the 
LGM did not affect Asian streams as broadly or severely as 
Alaskan streams. However, the freshwater environments 
were less favorable than at present, and it would be expect- 
ed that the harsh marine environment severely reduced 
productivity, probably creating a situation in which many 
local populations repeatedly went extinct. The ability to 
exploit spawning habitat rapidly would have been an ad- 
vantage during the LGM and many local extinctions were 
probably followed by recolonization. As a consequence, it is 
likely that a few systems provided the source for stock colo- 
nization following deglaciation. In the eastern range where 
habitat was ice-covered, re-establishment of pink salmon 
probably depended on colonization from the Bering or more 
southerly refugia, or possibly from the “off-year” broodline, 
if the rigid two year life cycle relaxes in marginal environ- 
ments, such as appears to have happened in the Lauren- 
tian Great Lakes (e.g., Kwain and Chappel, 1978). 
Overall, the geological events should have accentuated 
geographic boundaries and increased the importance of 
random drift. Divergence between Asian and North Amer- 
ican populations suggests colonization from different re- 
fugia, and significant, but lesser, regional divergence sup- 
ports homing in pink salmon, at least regionally. However, 
the low overall divergence observed among both Asian and 
North American populations (G sr -0 .023) suggests that 
the populations studied are either recently diverged and 
derived from a single population or from genetically simi- 
lar ancestral populations, or that there is sufficient gene 
flow to arrest divergence. One of the advantages of study- 
ing pink salmon is that there are two broodlines occupy- 
ing much of the same range. It will be interesting to ex- 
amine the genetic structure of odd-broodline pink salmon 
in the same range, which will represent a second natural 
experiment with which to examine the influences of geog- 
raphy, oceanography, and geologic history on the genetic 
structure of pink salmon populations. 
Acknowledgments 
We are indebted to many individuals who assisted in pro- 
viding samples and grateful to Hanhvan Thi Nguyen and 
Bichhang Thi Nguyen for their valuable assistance in 
the laboratory. R. Wilmot provided helpful comments and 
help with federal bureaucratic processes. Two anonymous 
reviewers provided constructive criticism. 
