contemporary range of distribution. Fish from a
Missourian refugium would have recolonized the
headwaters of the James Bay and Hudson Bay areas
of Ontario and Quebec through the connection be-
tween Lake Agassiz and Lake Ojibway-Barlow
(about 9500 years ago) (Crossman & McAllister
1986) resulting in a northerly distribution of de- Exploitation
scendants. As mentioned previously, Mississippian
fish would have used the Warren, Brule, and Chica Since the existence of aboriginal culture in North
go dispersal routes to colonize the southern part of America, lake sturgeon have been a key food
the range (Mandrak & Crossman 1992). The large source, especially during spring ceremonial festiv-
number of haplotype 2 fish sampled from northern ities at lake sturgeon spawning sites. During the
sites (Nelson River, Manitoba; Moose River basin, early 1800’s, lake sturgeon was also sought after as a
Ontario; Waswanipi River, East Megiscane region, trade item since the isinglass (a formofgelatin ob-
Quebec) coupled with the very high predominance tained from the inner lining of the swimbladder)
of haplotype 1 fish in the south is consistent with the could be used as a clarifying agent in wine, beer and
two refugium hypothesis. The observation that all jelly making. Records of its harvest were first kept
21 fish sampled from the Flambeau River, a tribu- by the Hudson Bay Company which provided abo-
tary of the Mississippi River were haplotype 1 lends riginal people access to markets (Holzshamm &
further credence to the idea that populations de- McCarthy 1988). There is no evidence that the liar-
rived from a Mississippian refugium were haplo- vest levels of lake sturgeon prior to the 1860’s had
type 1. According to the two refuge hypothesis, the any influence on population levels.
three haplotype 2 fish collected by Guenette et al. Early settlers to North America did not value
(1993) in the Ottawa River and Lake of Two Mountain lake sturgeon as a food source. This changed about
tains could be the result of secondary contact 1855 when a market for caviar developed at Sand-
among the two refugial groups; Lake Ojibway-Bar- usky, Ohio, on Lake Erie, followed by the sale of
low was connected with the Ottawa River up until smoked flesh in 1860. Spurred by these market de-
about 8000 years ago (Crossman & McAllister mands, local markets and fisheries spread to lakes
1986). According to the single refugium model, lake Huron. Ontario, Superior, Nipissing, and Nipigon
sturgeon recolonized Canada from a single Missis- and Lake of the Woods. In every case, after an ini-
sippian refugium (Guenette et al. 1993) a scenario tial high yield. the fisheries displayed a rapid and
compatible with the paleogeographic evidence permanent decline to very low levels (Harkness &
(Crossman & McAllister 1986, Mandrak & Cross- Diamond 1961). For example, the Lake Erie catch
man 1992). Mississippian refugium fish would have fell from over 2500 tonnes per year to less than 500
contained both haplotypes, one of which was large- tonnes between 1885–1895.
ly eliminated in the southern part of the range via The development of commercial fisheries for
stochastic lineage extinctions (Avise et al. 1987). lake sturgeon in Manitoba were delayed relative to
The few haplotype 2 fish observed in the Lake of those in Ontario. The harvest from Lake Winnipeg
Two Mountains and the Ottawa River would be the and its tributaries, the Red and Assiniboine rivers,
remnants of a historically more common lineage. peaked at 445 tonnes in 1900 and crashed to 13
We favor the two refuge hypothesis given the dis- tonnes by 1910 when the fishery was closed (Hous-
tribution of haplotypes over the broader geograph- ton 1987). The fishery reopened after 6 years with
ic scale illustrated in Figure 1. Stochastic lineage ex- fluctuating landings until 1928 when it was again
tinction is expected to be random with respect to closed. Declining catches in Lake Winnipeg
the extinction of haplotypes in specific populations. spurred interest in more northern locations such as
It is difficult to envisage that all the southerly pop- the Nelson and Churchill rivers where commercial
ulations sampled by us and Guenette et al. (1993) fishing began in 1907 and 1924, respectively. Succes-
from throughout the Great Lakes and Mississippi
regions would have had the same haplotype elim-
inated by chance, and it is difficult to imagine differ-
ential selection of mtDNA haplotypes.