For permission to copy, contact editing@geosociety.org
q 2004 Geological Society of America 729
GSA Bulletin; May/June 2004; v. 116; no. 5/6; p. 729?742; doi: 10.1130/B25316.1; 5 figures; 1 table; Data Repository item 2004079.
Glacial Lake Agassiz: A 5000 yr history of change and its relationship
to the d18O record of Greenland
James T. Teller?
Department of Geological Sciences, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada
David W. Leverington?
Center for Earth and Planetary Studies, National Air and Space Museum, Smithsonian Institution, Washington, D.C.
20560-0315, USA
ABSTRACT
Lake Agassiz was the largest lake in
North America during the last period of de-
glaciation; the lake extended over a total of
1.5 3 106 km2 before it drained at ca. 7.7
14C ka (8.4 cal. [calendar] ka). New com-
puter reconstructions?controlled by
beaches, isostatic rebound data, the margin
of the Laurentide Ice Sheet, outlet eleva-
tions, and a digital elevation model (DEM)
of modern topographic data?show how
variable the size and depth of this lake were
during its 4000 14C yr (5000 cal. yr) history.
Abrupt reductions in lake level, ranging
from 8 to 110 m, occurred on at least 18
occasions when new outlets were opened,
reducing the extent of the lake and sending
large outbursts of water to the oceans.
Three of the largest outbursts correlate
closely in time with the start of large d18O
excursions in the isotopic records of the
Greenland ice cap, suggesting that those
freshwaters may have had an impact on
thermohaline circulation and, in turn, on
climate.
Keywords: Lake Agassiz, history, out-
bursts, Greenland isotopic record.
INTRODUCTION
During the last period of global deglacia-
tion, large lakes formed along the margins of
continental ice sheets in North America, Eu-
rope, and Asia. The complex history of North
American proglacial lakes has been summa-
rized by Prest (1970), Teller (1987, 2004),
Dyke and Prest (1987), Klassen (1989),
?E-mail: tellerjt@ms.umanitoba.ca.
?E-mail: leveringtond@nasm.si.edu.
Dredge and Cowan (1989), Karrow and Oc-
chietti (1989), Lewis et al. (1994), and others.
Several special volumes of papers (e.g., Teller
and Clayton, 1983; Karrow and Calkin, 1985;
Teller and Kehew, 1994) and hundreds of
journal articles over the past century have pro-
vided important insight into this extensive
lake system. Closely linked to the history of
proglacial lakes is the chronology of the rout-
ing of North American glacial runoff, which
has been discussed by many for various re-
gions, and synthesized by several, including
Teller (1987, 1990a, 1990b, 1995) and Lic-
ciardi et al. (1999), and tied to past global
ocean circulation by Broecker et al. (1989),
Clark et al. (2001), Teller et al. (2002), and
others.
The largest of all proglacial lakes was Lake
Agassiz, which developed and expanded
northward along the margin of the Laurentide
Ice Sheet as it retreated downslope into the
Hudson Bay and Arctic Ocean basins in Sas-
katchewan, Manitoba, Ontario, Quebec, and
Northwest Territories. During the history of
Lake Agassiz, overflow was carried from the
lake to the oceans by way of four different
routes (Fig. 1): (A) southward via the Min-
nesota and Mississippi River Valleys to the
Gulf of Mexico, (B) northwestward through
the Clearwater-Athabasca-Mackenzie River
Valleys to the Arctic Ocean, (C) eastward
through a series of channels that led to the
Great Lakes (or to glacial Lake Ojibway) and
then to the St. Lawrence Valley and North At-
lantic Ocean, and, finally, when Lake Agassiz
completely drained, (D) northward and east-
ward through Hudson Bay and Hudson Strait
to the North Atlantic Ocean, between the Kee-
watin and Labrador ice centers.
The extent, configuration, and depth of
Lake Agassiz, as with all proglacial lakes, was
a result of the interaction among (1) location
of the ice margin, (2) topography of the newly
deglaciated surface, (3) elevation of the active
outlet, and (4) differential isostatic rebound
(Teller, 1987, 2001). This interaction was
complex, partly because it involved glacier
melting and ice-dam failure, as well as ice
readvances and surges into the lake basin.
Thus, overflow outlets were opened and oc-
casionally closed again by a fluctuating re-
treat. Outlet erosion also played a role. The
interrelationship between differential isostatic
rebound and the geographic location of the
outlet was especially important in dictating the
extent and depth of Lake Agassiz. As de-
scribed by Teller (2001), whenever the outlet
carrying overflow was at the southern end of
the basin, lake levels receded everywhere to
the north of the isobase (i.e., the contour of
equal isostatic rebound) through the outlet
(Fig. 2A). Whenever the outlet was not at the
southern end of the basin, transgression oc-
curred to the south of the outlet, while regres-
sion occurred everywhere to the north of it
(Figs. 2B and 2C). Isobases across the Agassiz-
Ojibway basin are shown in Figure 3, along
with the locations of the main outlets.
As a result of the abrupt opening and clos-
ing of outlets, plus the interaction of differ-
ential isostatic rebound and the use of many
different outlet channels, the level of Lake
Agassiz was constantly changing, and its his-
tory is very complex. Overall, its history is
one of northward expansion, punctuated by
abrupt drops in lake level when lower outlet
channels were deglaciated; each decline was
then followed by transgressive deepening of
the lake in the basin south of the isobase
through the outlet and regression north of that
isobase due to differential rebound.
In this paper, a series of maps are presented
730 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Figure 1. Routing of runoff to the oceans from Lake Agassiz (Teller et al., 2002). Ice
margin at ca. 9 14C ka shown by dashed line. Outflow paths A?D described in text.
that outline the extent of Lake Agassiz at 18
different transgressive maximums and 15 in-
tervening minimums. These span the history
of the lake from ca. 10.9 14C ka (thousand ra-
diocarbon years B.P.; 12.94 thousand calendar
years B.P.), until its final stage at ca. 7.7 14C
ka (8.45 cal. ka), after it had amalgamated
with glacial Lake Ojibway. In addition, we
discuss some of the relationships of these out-
bursts to the d18O curves in the Greenland
GISP2 and GRIP ice cores.
LAKE RECONSTRUCTION
Maps depicting various stages in the history
of Lake Agassiz have been published for more
than a century, beginning with those in U.S.
Geological Survey Monograph 25 by Upham
(1895). Johnston?s (1946) field work on the
beaches provided important new data on
shorelines in the Canadian part of the lake,
north of where Upham had worked. Modern
reconstructions have been guided by the
mapped and correlated strandlines of Upham
(1895) and Johnston (1946) and have been
supplemented by new topographic map data,
field work, and aerial photograph interpreta-
tion (see Elson, 1983). More recent recon-
structions of Lake Agassiz have extended the
area of the lake farther north (Teller et al.,
1983; Elson, 1983; cf. Smith and Fisher,
1993), and, because glacial Lake Ojibway
amalgamated with Lake Agassiz during the
last several hundred years of its history, this
eastern extension, as described by Vincent and
Hardy (1979) and Veillette (1994), is included
in our reconstructions. Elson (1967), Prest
(1970), and Dyke and Prest (1987) provided a
number of snapshots of the lake through its
history, and some, such as Clayton and Moran
(1982), Teller (1985), Klassen (1989), Dredge
and Cowan (1989), and Thorleifson (1996),
presented maps and integrated its history with
deglaciation across central North America.
Lake reconstructions in various regions and at
various times were also published by research-
ers in the volume Glacial Lake Agassiz (Teller
and Clayton, 1983), among them Brophy and
Bluemle (1983), Dredge (1983), Klassen
(1983a, 1983b), Fenton et al. (1983), and
Schreiner (1983). Smith and Fisher (1993) and
Fisher and Souch (1998) expanded the lake
northwestward to the Clearwater-Mackenzie
outlet.
Reconstructions in this paper expand on
those in Leverington et al. (2000, 2002a),
which used, and projected northward, the cor-
related beach elevations in Thorleifson (1996),
which were derived from Johnston (1946) and
Teller and Thorleifson (1983) and were inte-
grated with isobase strandline data to the east
from Vincent and Hardy (1979) (Fig. 3A). In
contrast to Figure 3A, which shows the trend
of isobases at 100 km spacing, Figure 3B pre-
sents the trend of isobases at a vertical spacing
of 25 m on the Upper Campbell water plane.
Each Lake Agassiz beach defines its own re-
bound curve, because isostatic rebound varied
through time; older beaches have slightly
steeper curves, whereas younger ones have
gentler curves (see Teller and Thorleifson,
1983, Fig. 2). The trend of the isobases over
the life of the lake are assumed to be the same,
although we recognize that changes in relative
thickness of the Laurentide Ice Sheet during
deglaciation are likely to have led to some
changes in isobase orientation and
configuration.
The isostatically deformed rebound curves
were used to generate a paleo?water surface
of Lake Agassiz associated with each beach
(Mann et al., 1999; Leverington et al., 2002b).
Each rebound surface was computationally
projected to intersect modern topography (de-
fined by a digital elevation model [DEM] with
individual grid dimensions of ;1 3 1 km),
generating an outline of the lake south of the
Laurentide Ice Sheet (Leverington et al., 2000,
2002a, 2002b). The GLOBE (Globe Task
Team, 1999) and ETOPO 5 (NGDC, 1988) da-
tabases were used as the sources for the mod-
ern DEM. Because control points for isostatic
rebound are nearly all at the margin of the
lake, the simple curvilinear isobases repre-
senting this rebound (Fig. 3) may be more
complex, as suggested by Dredge and Cowan
(1989) and Rayburn and Teller (1999). How-
ever, lake bathymetry and total lake volume
would have been influenced only slightly by
such isostatic variability, and the coincidence
of actual beaches and wave-trimmed cliffs
with the topographically modeled shorelines
indicates that the outlines of various lake stag-
es are accurately depicted, except perhaps in
far northern regions where rebound curves
were projected and are not controlled by
strandline data.
The ice margin across the Lake Agassiz ba-
sin through time is controlled in only a few
places, mainly by scattered and dated end mo-
raines and by some dated lithostratigraphic re-
lationships. Additional ??local?? control for
placement of the ice margin is based on the
linkage of lake levels (beach elevations) to
specific outlets carrying overflow at a given
time, which were controlled by the ice margin.
Because of this uncertainty, and because the
Laurentide Ice Sheet margin was very dynam-
ic during retreat?repeatedly surging into the
lake and rapidly calving back (e.g., Clayton et
al., 1985; Dredge and Cowan, 1989)?we
have kept the ice margins shown in Figure 4
relatively constant through time across that
part of the basin where there is no control; we
acknowledge that the ice margin fluctuated,
extending beyond our ??average?? margin at
times and retreating north of it at other times,
Geological Society of America Bulletin, May/June 2004 731
GLACIAL LAKE AGASSIZ
Figure 2. Cartoons showing the influence of differential isostatic rebound through time (T1, T2, T3, T4) on lake level, resulting from
overflow through (A) an outlet in the southern end of the basin, (B) an outlet between the northern and southern ends of the basin,
and (C) an outlet in the northern end of the basin (after Larsen, 1987; Teller, 2001). Through time, older lake levels (beaches) become
elevated or submerged with respect to the contemporaneous (horizontal) level.
but large, deep lake basins are not good places
for the evidence of short-lived marginal po-
sitions to be preserved.
The level of Lake Agassiz fluctuated con-
siderably throughout its history, with numer-
ous transgressive maximums identified by
mapped (and named) beaches and intervening
low-level stages controlled by the elevation of
newly opened outlets. Subsequent uplift of
those outlets led to deepening of the lake be-
fore the next, lower outlet was deglaciated;
these intervening minimum levels did not
form beaches and their outlines are controlled
by the elevation of the spillover points in our
paleotopographic database. The mapped min-
imums were implicit in previous volumetric
calculations (Mann et al., 1999; Leverington
et al., 2000, 2002a; Teller et al., 2002), but
have never been published.
Although substantial volumes of water were
released between the transgressive maximums
and subsequent drawdown levels (see draw-
down values in caption to Fig. 4), the lake did
not always change substantially in areal ex-
tent. In Figure 4, we show a selection of
paired maximums and minimums (e.g., A1-
A2, B1-B2); six additional pairs showing only
relatively small change in area between pre-
ceding and subsequent lake outlines are also
available.1 None of the minimum lake outlines
in Figure 4 has been published before; six of
the 12 maps in Figure 4 that show the trans-
gressive maximums have been published pre-
viously as bathymetric maps by Leverington
et al. (2000, 2002a), although two have been
substantially modified.
1GSA Data Repository item 2004079, Figure
DR1, is available on the Web at http://www.
geosociety.org/pubs/ft2004.htm. Requests may also
be sent to editing@geosociety.org.
HISTORY OF THE RISES AND FALLS
OF LAKE AGASSIZ
Dating of Events
There are many beaches and wave-cut
scarps in the Lake Agassiz basin. Some are
large, distinct, and continuous for many kilo-
meters; others are not (Upham, 1895). The
ages of beaches and subsequent outbursts are
controlled by a limited number of radiocarbon
dates. Organic matter is rare in the beaches of
Lake Agassiz, so their ages are commonly
based on dated glacial events in the Agassiz
basin or correlation with events outside of the
basin, such as events recorded in the Superior,
Huron, and Michigan basins (e.g., Drexler et
al., 1983; Teller and Mahnic, 1988; Lewis et
al., 1994; Colman et al., 1994) and in outlet
channels that carried overflow (Smith and
Fisher, 1993; Fisher, 2003). There are several
732 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Figure 3. (A) Isobases across the Lake Agassiz-Ojibway basin (Teller and Thorleifson, 1983: after Johnston [1946], Walcott [1972], and
Vincent and Hardy [1979]). Lines of equal isostatic rebound (isobases 1?12) are spaced at 100 km intervals. General locations of main
outlets are shown (S, NW, E1, E2, and O). (Caption continued on p. 733.)
TABLE 1. CONVERSION OF INTERPRETED RADIOCARBON AGES OF LAKE AGASSIZ BEACHES TO
THE RANGE OF CALENDAR YEARS (DEFINED BY 1s) AND MEAN AGES IN YEARS BEFORE PRESENT
OF THAT RANGE
Map Beach name 14C age Calendar age range Mean of cal. range
A Herman 10,900 13,019?12,860 12,940
B Norcross 10,100 11,697?11,564 11,630
C Tintah 9900 11,257?11,230 11,240
D U. Campbell 9400 10,671?10,578 10,620
E L. Campbell 9300 10,550?10,429 10,490
F McCauleyville 9200 10,398?10,264 10,330
G Blanchard 9000 10,208?10,185 10,200
H Hillsboro 8900 10,149?9935 10,040
I Emerado 8800 9908?9775 9840
J Ojata 8700 9684?9600 9640
K Gladstone 8600 9548?9540 9540
L Burnside 8500 9527?9494 9510
M Ossawa 8400 9472?9428 9450
N Stonewall 8200 9249?9089 9170
O The Pas 8000 8996?8788 8890
P Gimli 7900 8701?8636 8670
Q Grand Rapids 7800 8592?8543 8570
R Kinoje?vis 7700 8474?8421 8450
Note: Calendar dates estimated using CALIB 4.3 program of Stuiver and Reimer (1993) and Stuiver et al.
(1998). Subsequent drawdowns of the lake (the outbursts) are nearly the same age as the beaches. ??Map?? refers
to Figure 4 and Figure DR1 (see footnote 1 in text) lake reconstructions.
accepted temporal pins in the Lake Agassiz
beach and outburst chronology that are based
on radiocarbon dates in the basin; these are
related to formation of specific beaches and
lie within a generally accepted small age
range: (1) the Herman beach (11.0?10.8 14C
ka; Fig. 4A1), (2) Upper Campbell beach
(9.4?9.3 14C ka; Fig. 4D1), and (3) the final
drainage of the lake (7.7 14C ka). The two
beaches below the Upper Campbell beach (the
Lower Campbell and McCauleyville) (Figs.
4E1 and 4F1) have radiocarbon dates associ-
ated with them that indicate that they were
deposited shortly after 9400 yr B.P. (Teller et
al., 2000). The age of The Pas beach (Fig.
4O1), which dates the end of the routing of
Agassiz water into the Superior basin, is 8.2?
8.0 14C ka (Teller and Mahnic, 1988; Thor-
leifson, 1996; Teller et al., 2002), probably
closer to 8.0 14C ka, on the basis of the paleo-
magnetically dated change in sediment style
in the Superior basin described by Mothersill
(1988).
The ages of two of the older beaches, the
Norcross and Tintah, are controversial. Early
researchers attributed them to sequential
??stair-step?? drops in lake level after the Her-
man beach was formed when lower routes
were opened into the Superior basin (e.g.,
Johnston, 1946; Elson, 1967; Fenton et al.,
1983). Dating of two cores in the southern
outlet of Lake Agassiz led Fisher (2003) to
conclude that this interpretation was correct,
and he considered that they formed between
ca. 10.9 and 10.8 14C ka. In contrast, others
have concluded that the Norcross and Tintah
beaches formed later, after centuries of lower
lake level (the Moorhead phase) (e.g., Thor-
leifson, 1996; Teller et al., 2000; Teller, 2001;
Leverington et al., 2000; Fisher and Smith,
1994). These researchers argued that differ-
ential isostatic rebound forced Lake Agassiz
to rise briefly to the Norcross level sometime
Geological Society of America Bulletin, May/June 2004 733
GLACIAL LAKE AGASSIZ
Figure 3. (Caption continued from p. 732.) (B) Isobases across the Lake Agassiz basin at vertical intervals of 25 m on the rebounded
surface associated with the Upper Campbell beach. Maps on older (or younger) beaches, which has undergone more (or less) differential
rebound, would show steeper (or gentler) gradients.
between 10.4 and 10.1 14C ka after the Moor-
head low-water phase. Following a drop in
lake level, glacial readvance at ca. 10.0 14C ka
closed the newly opened northwestern outlet,
forcing the lake to rise to the Tintah level. We
adopt the Teller (2001) interpretation in this
paper, but we acknowledge that the absence of
radiocarbon dates from any of the older beach-
es and the small number from Agassiz outlets
do not allow exact dating of these beaches.
Between 9.4 14C ka and the final drainage
of Lake Agassiz, more than 15 recognized
beaches developed in the basin (Johnston,
1946; Elson, 1967; Teller and Thorleifson,
1983); most are linked to a topographically
distinct overflow route from the lake (Teller
and Thorleifson, 1983; Leverington and Teller,
2003). The ages associated with these lake
stages (beaches), and with the outbursts that
ended those stages, all lie between 9.4 and 7.7
14C ka, and it is possible that the time needed
to form each beach and the time represented
by the intervening transgression were about
the same. Thus, 1700 14C yr (5 2170 cal. yr)
divided by 15 outbursts equals an average of
113 radiocarbon years (145 cal. yr) between
each beach and each outburst. Complicating
the age assignment are the so-called radiocar-
bon plateau at ca. 10,000 14C yr B.P. (10.6?
10.0 and 9.6 14C ka) and several other such
plateaus during the time of Lake Agassiz over-
flow (e.g., 8.75 14C ka and 8.25 14C ka) (e.g.,
Bradley, 1999, p. 68) as well as the overall
nonequivalence of radiocarbon and sidereal
years due to long-term changes in atmospheric
14C content (e.g., Stuiver and Reimer, 1993).
Table 1 shows our interpretation of the radio-
carbon ages of Lake Agassiz beaches and their
conversions to calendar years using the
CALIB 4.3 program of Stuiver and Reimer
(1993) and Stuiver et al. (1995).
Snapshots of Lake Agassiz Through Time
Although Lake Agassiz began to develop in
the southern end of the basin before its first
large beach formed at ca. 11.7 14C ka (13.4
cal. ka) (Fenton et al., 1983), our reconstruc-
tions begin with the oldest extensive beaches
in the Lake Agassiz basin, the Herman beach-
es, formed while overflow was through the
southern outlet (e.g., Upham, 1895; Elson,
1967; Fenton et al., 1983); the lowest (youn-
gest) of this closely spaced group of beaches
is shown in Figure 4A1. The age of the Her-
man beaches and of the subsequent 110 m
drop in lake level is between 11.0 and 10.8
14C ka (see Licciardi et al., 1999; Fisher,
2003), probably ca. 10.9 14C ka (12.9 cal. ka).
A total of 9500 km3 of Lake Agassiz waters
were released abruptly into the Lake Superior
basin near Thunder Bay, Ontario (outlet E1 of
Fig. 3A) at this time, and the size of the lake
decreased from ;134,000 km2 to 37,000 km2.
This was the largest outburst until the final
drainage of the lake (Teller et al., 2002). On
the basis of the interpretation of beach ages
by Fisher (2003), as discussed above, this 110
m drop in lake level would have occurred in
several steps over about a century, and the
Norcross and Tintah beaches formed as Lake
734 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Figure 4. Snapshots of the history of Lake Agassiz showing the extent of the lake (black area) at various times; maps of the 12 lake
stages in this sequence that are not shown here (map pairs E1-E2, F1-F2, H1-H2, K1-K2, L1-L2, and M1-M2; in square brackets in
this caption) are available in Figure DR1 (see footnote 1 in text). Each of the map pairs (A1-A2 to O1-O2) shows the transgressive
maximums (which are based on mapped beaches) and the subsequent postoutburst minimums; each minimum is followed by a new
transgressive phase (south of the outlet) caused by differential isostatic rebound. The last three lake stages (P, Q, and R) are depicted
only by transgressive maximums. Values given below for lake-level decline are from Teller et al. (2002) and Leverington and Teller
(2003) and relate to change in water level estimated in the outlet area at that time. Arrows show routing of outburst and baseline
overflow at time of each lake stage. See comments in text about position of ice margins, Table 1 for age of each lake stage, and Figure
5 for routing of overflow and relationship to Greenland isotopic record. (A1) Herman beach stage (lower level) and (A2) lake following
110 m drawdown. (B1) Norcross beach stage and (B2) lake following 52 m drawdown. (C1) Tintah beach stage and (C2) lake following
30 m drawdown. (D1) Upper Campbell beach stage and (D2) lake following 30 m drawdown. [(E1) Lower Campbell beach stage and
(E2) lake following 11 m drawdown.] [(F1) McCauleyville beach stage and (F2) lake following 16 m drawdown.] (G1) Blanchard beach
stage and (G2) lake following 10 m drawdown. [(H1) Hillsboro beach stage and (H2) lake following 8 m drawdown.] (I1) Emerado
beach stage and (I2) lake following 18 m drawdown. (J1) Ojata beach stage and (J2) lake following 20 m drawdown. [(K1) Gladstone
beach stage and (K2) following 9 m drawdown.] [(L1) Burnside beach stage and (L2) lake following 13 m drawdown.] [(M1) Ossawa
beach stage and (M2) lake following 15 m drawdown.] (N1) Stonewall beach stage and (N2) lake following 58 m drawdown. (O1) The
Pas beach stage and (O2) lake following 18 m drawdown. (P) Gimli beach stage. (Q) Grand Rapids beach stage. (R) Kinoje?vis beach
stage; just before Lake Agassiz-Ojibway drained into the Tyrrell Sea.
M
Agassiz overflow eroded the southern outlet
below the Herman beach level.
It is important to note that the routing of
overflow during this early period remains un-
certain. New field work has raised the ques-
tion of whether overflow from Lake Agassiz
was directed east through the Great Lakes
(outlet E1 in Fig. 3A) for centuries after the
Herman beach was abandoned, as has been the
interpretation by all researchers, or whether
overflow may have been through the Clear-
water outlet to the Athabasca-Mackenzie val-
ley system (outlet NW in Fig. 3A) between
10.8 and 9.4 14C ka (12.8?10.6 cal. ka). The
latter requires that the northwestern outlet was
deglaciated shortly after 11 14C ka, which is
much earlier than most evidence indicates. Re-
gardless of the routing of overflow at this
time, the extent of Lake Agassiz is not likely
to have been much different except along its
western margin. The computer-generated ex-
tent of the lake immediately after this draw-
down is shown in Figure 4A2; as previously
discussed, this lake is not represented by a
beach because subsequent rising waters re-
worked shoreline sediment upslope.
During the next 700?800 yr, waters deep-
ened to the south of the isobase through the
eastern outlet (isobase 6 in Fig. 3A), and the
lake margin transgressed over the dry lake
bed; waters shallowed in the region between
that isobase and the Laurentide Ice Sheet mar-
gin (see Fig. 2B). There are many radiocarbon
dates on vegetation that was buried by sedi-
ments deposited during this transgression (see
summary of dates in Appendix B of Licciardi
et al., 1999). The Norcross beach represents
the maximum extent of this transgression just
before 10.1 14C ka (Teller, 2001) (Fig. 4B1);
it also dates the start of the next drop in lake
level. Figure 4B2 shows the extent of Lake
Agassiz following the opening of the north-
western outlet to the Arctic Ocean (outlet NW
in Fig. 3A), based on the work of Fisher and
Smith (1994). Differential isostatic rebound,
combined with the redamming of the north-
western outlet (Thorleifson, 1996), raised wa-
ter levels to the Tintah beach level by ca. 9.9
14C ka (Fig. 4C1), prompting renewed over-
flow through the southern outlet for a short
time (Teller, 2001). When ice retreated again
from the northwestern outlet, lake level
abruptly dropped, and the extent of the lake
decreased again (Fig. 4C2). The difference be-
tween our interpretation of beach formation
during this period of lake history and that of
Fisher (2003) is that he has related formation
of the Norcross and Tintah beaches to rapid
erosion of the southern outlet immediately fol-
lowing formation of the Herman beach.
Over the next 400 yr, between ca. 9.8 and
9.4 14C ka, greater isostatic rebound of the
northwestern outlet (on isobase 7, Fig. 3A)
compared to that of the southern outlet (iso-
base 1) caused Lake Agassiz to transgress
southward until it reached the channel that had
previously carried overflow at the southern
end of the basin; this process resulted in the
stranding of the largest and most extensive
beach in the basin, the Upper Campbell beach,
at ca. 9.4 14C ka (10.6 cal. ka), as shown by
the outline of the lake in Figure 4D1. Within
a few years, a lower eastern outlet channel
from Lake Agassiz was deglaciated, and wa-
ters were again routed into the Great Lakes,
this time via the Lake Nipigon basin (outlet
E2, Fig. 3A), and the level of the lake abruptly
dropped (Fig. 4D2).
Subsequent abrupt declines in lake level oc-
curred when new, lower, eastern outlet chan-
nels were deglaciated; following each drop
there was a new deepening of waters (trans-
gression) south of the isobase through the out-
let, as there had been between previous draw-
downs. Although there are a few other Lake
Agassiz beaches, which lie between these
transgressive maximums, most are not well
developed and may be storm beaches or off-
shore bars (see Teller, 2001). Figure 4 shows
lake stages related to five transgressive max-
imums and subsequent minimums after over-
flow was rerouted eastward (G1-G2, I1-I2, J1-
J2, N1-N2, O1-O2), and three late-stage
maximums routed through the Ottawa River
Valley after ca. 8.0 14C ka (8.9 cal. ka) (Fig.
4P, 4Q, 4R). Six map pairs showing maximum
and minimum stages during this period are not
shown in Figure 4, but are available in Figure
DR1 (see footnote 1; i.e., map pairs E1-E2,
Lower Campbell; F1-F2, McCauleyville: H1-
H2, Hillsboro; K1-K2, Gladstone; L1-L2,
Burnside; and M1-M2, Ossawa).
At ca. 8.0 14C ka, Lake Agassiz amalgam-
ated with glacial Lake Ojibway, which had
been expanding independently in eastern On-
tario and adjacent Quebec (Fig. 3A); it is pos-
sible that the drawdown of Lake Agassiz fol-
lowing formation of The Pas beach (Fig. 4O2)
may have resulted from this amalgamation.
Overflow was subsequently routed out
through the Ottawa River Valley; the Kinoje?v-
is outlet carried overflow just before the final
drainage of the lake (e.g., Vincent and Hardy,
1979) at ca. 7.7 14C ka (8.45 cal. ka). Figure
4R shows the outline of the lake at this time,
which extended over an area of 841,000 km2
and is correlated to the Ponton beach in the
Geological Society of America Bulletin, May/June 2004 735
GLACIAL LAKE AGASSIZ
736 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Figure 4. (Continued.)
Geological Society of America Bulletin, May/June 2004 737
GLACIAL LAKE AGASSIZ
Figure 4. (Continued.)
738 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Figure 4. (Continued.)
Agassiz part of the basin (Leverington et al.,
2002a). An alternative two-step scenario for
the final drainage of Lake Agassiz was dis-
cussed by Leverington et al. (2002a), where
the eastern (Ojibway) part of the lake com-
pletely drained but only part of the western
region did, leaving a residual lake of
;408,000 km2 in Manitoba and adjacent
northern Ontario (Leverington et al., 2002a,
Fig. 2F). Following the final drainage of Lake
Agassiz-Ojibway, waters of the Tyrrell Sea
transgressed south over lacustrine sediments
in the Hudson Bay Lowland (??marine limit??
of Fig. 3A), and modern drainage was estab-
lished across the old lake bed.
RELATIONSHIP OF LAKE AGASSIZ
OUTBURSTS TO THE GREENLAND
ISOTOPIC RECORD
Influxes of fresh water to the North Atlantic
Ocean from North America have been corre-
lated with changes in thermohaline circulation
(THC) and climate (e.g., Broecker et al., 1985,
1989; Rahmstorf, 1995; Manabe and Stouffer,
1995, 1997; Alley et al., 1997; Clark et al.,
2001). This relationship has been linked to the
main cooling episodes that interrupted late
glacial warming, namely, Heinrich 1, the
Younger Dryas, the Preboreal Oscillation, and
the 8.2 cal. ka cooling. Changes in the site of
meltwater delivery to the oceans during late-
glacial time (associated with changes in con-
tinental-scale glacial drainage basins) and
short high-flux injections of water related to
catastrophic outbursts from Lake Agassiz have
been interpreted as possible forcing mecha-
nisms (Clark et al., 2001; Teller et al., 2002).
Some outbursts from Lake Agassiz may have
triggered changes in THC, and these changes
may have been sustained when there was an
associated re-rerouting of overflow to a dif-
ferent ocean.
Some evidence for episodes of climatic
cooling comes from the oceans, but the best
record of temperature fluctuation during the
5000-cal.-yr-long period when Lake Agassiz
waters may have had an impact on THC and
climate is found in the isotopic and ionic rec-
ords of the GRIP and GISP2 ice cores in
Greenland (e.g., O?Brien et al., 1995; Grootes
and Stuiver, 1997; Johnsen et al., 2001). Giv-
en the potential impact on climate of fresh-
water injections to the North Atlantic Ocean,
it is relevant to compare the chronology of
Lake Agassiz outbursts to the isotopic record
of Greenland.
Figure 5 shows the fluctuations of d18O val-
ues in the GISP2 and GRIP ice cores of
Greenland, which serve as proxies for tem-
Geological Society of America Bulletin, May/June 2004 739
GLACIAL LAKE AGASSIZ
Figure 5. Record of d18O values in Greenland ice of GRIP (Johnsen et al., 2001) and GISP2 (Stuiver et al., 1995) cores plotted with the
record of catastrophic outbursts from Lake Agassiz (A?R); GRIP data are 55 cm averages, and GISP2 data are bidecadal (isotopic data
from Cross, 2002). Negative isotopic excursions related to the Younger Dryas, Preboreal Oscillation (PBO), and 8.2 ka event are
identified. Ages are in calendar years B.P.; selected ages are shown in radiocarbon years. Each Lake Agassiz outburst has been inter-
preted as occurring in ;1 yr and is represented by a bar whose height relates to the total volume of the outburst (Teller et al., 2002);
the final outburst was 163,000 km3 and is indicated by an arrow. Note that if each lake drawdown occurred in 1 yr or less, as suggested
by Teller et al. (2002), a 10,000 km3 outburst would result in a flux of 0.32 Sv (i.e., 320,000 m3?s21 for 1 yr). Dashed line represents
baseline overflow from Lake Agassiz. The ??total runoff?? includes all flow through routes A, B, C, and D of Figure 1 (Licciardi et al.,
1999, Appendix A) but excludes the outbursts; note how little flow changed through time.
740 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
perature variation (Johnsen et al., 2001). Also
shown are the times (and magnitudes) of Lake
Agassiz outbursts; the radiocarbon dates of
these outbursts have been converted to cal-
endar years as shown in Table 1.
Negative excursions in d18O are correlated
with decreases in atmospheric temperature
(e.g., Dansgaard, 1961; Bradley, 1999) and, in
turn, with climate cooling in the North Atlan-
tic region; some have related these isotopic
excursions to the influx of freshwaters that in-
hibited THC and delivery of warm waters into
high latitudes (e.g., Broecker et al., 1989;
Bjo?rck et al., 1996; Barber et al., 1999; Clark
et al., 2001; Renssen et al., 2001). Because the
dating of isotopic excursions in the Greenland
ice cap is controlled by the measurement of
annual snow layers, the excursions? ages in
calendar years have generally been accepted,
although there are variable differences of up
to nearly 200 yr in interpreted age between the
GISP2 and GRIP ice cores; events in the
GRIP core generally show younger ages than
those in the GISP2 core (Southon, 2002), as
can be seen in Figure 5. These age differences
are very important when trying to establish a
cause-and-effect relationship between climate
change and Agassiz flood outbursts that are
spaced at 100?200 yr, especially because the
impact of freshwater on THC and, in turn, on
climate and, in turn, on the isotopic record of
Greenland is not known with any certainty.
Models show changes in THC to be variable
in response time, in duration, and in the mag-
nitude of change in flux and temperature, as a
result of variation in duration, magnitude, and
site of influx to the ocean. Furthermore, the
time of the freshwater flux into the ocean may
greatly influence the THC response, because
oceans can be pushed over critical thresholds
when combined with other contemporary forc-
ing events (cf. Alley et al., 2001) and when
oceans are in circulation modes that make
them more vulnerable to change (e.g., Fanning
and Weaver, 1997). In short, both the ocean
response to freshwater forcing and the fresh-
water forcing itself are complex, and changes
are nonlinear (see Rahmstorf, 2000; Ganopol-
ski and Rahmstorf, 2001).
Lake Agassiz Outbursts and the
Greenland d18O Isotopic Record
Several d18O isotopic excursions in the
Greenland record have been correlated with
outbursts from Lake Agassiz or with the re-
direction of Lake Agassiz overflow (Fig. 5),
which may have reduced the flux of warm wa-
ters into the North Atlantic:
1. The Younger Dryas cooling began at ca.
12.9 cal. ka and is clearly recorded in the
Greenland isotopic record (Fig. 5). Broecker
et al. (1989), Fanning and Weaver (1997),
Clark et al. (2001), Teller et al. (2002), and
others related the start of this well-known
cooling to either a Lake Agassiz outburst and/
or the redirection of Agassiz overflow into the
North Atlantic at 12.9 cal. yr ka. This is the
time of the first (Herman) drawdown of Lake
Agassiz (Figs. 4A1 to 4A2), which initiated a
period of overflow from Agassiz to the North
Atlantic Ocean that lasted for more than 1000
cal. yr. Contrary to the interpretation that large
fluxes of Agassiz water led to a reduction in
THC, it is interesting to note that the 11.6 cal.
ka outburst occurred at the end of the negative
isotopic excursion associated with the Youn-
ger Dryas and close to the time when overflow
from Lake Agassiz shifted from being routed
through the Great Lakes to being routed into
the Arctic Ocean (Fig. 5). An analysis of dated
lacustrine records, tree rings, and the GRIP ice
core by Bjo?rck et al. (1996) places the end of
the Younger Dryas at 11.45?11.39 cal. ka,
which is 100?200 yr later than is indicated in
the isotopic records of the GISP2 ice core
(Fig. 5).
2. The Preboreal Oscillation (PBO) began
with a slow cooling almost immediately after
the end of the Younger Dryas and ended with
an abrupt short cold episode between 11.4 and
11.3 cal. ka, as recorded in the GISP2 core
(Fig. 5). In the GRIP ice core, the PBO cool-
ing is dated ;200 yr later at 11.2?11.05 cal.
ka, which immediately followed the Lake Ag-
assiz outburst at 11.2 cal. ka (Fig. 5) as noted
by Fisher et al. (2002). In contrast, in the
GISP2 core, the PBO precedes the 11.2 cal.
ka outburst but follows the 11.6 cal. ka out-
burst by several centuries. Complicating the
attempt to link Agassiz outbursts with the
PBO is the fact that there was a comparable
outburst of water into the North Atlantic
Ocean from the Baltic Ice Lake of Europe just
before the PBO, which probably had an im-
pact on THC and contributed to the PBO cool-
ing event (Bjo?rck et al., 1996). We think that
the chronological association of the PBO cold
event?recorded in both Greenland and north-
western Europe 200?300 years after the end
of the Younger Dryas (Bjo?rck et al., 1996)?
with large post?Younger Dryas outbursts from
Lake Agassiz and the Baltic Ice Lake suggests
a cause-and-effect relationship.
3. The 8.25?8.1 cal. ka negative d18O ex-
cursion in the GISP2 core, and shortly after
that in the GRIP core (Fig. 5), followed the
final drainage (outburst) from Lake Agassiz,
dated at ca. 8.4 cal. ka. A number of research-
ers have linked this increased freshwater flux
with the 8.2 ka isotopic event (e.g., Alley et
al., 1997; Barber et al., 1999; Clark et al.,
2001; Renssen et al., 2001; McDermott et al.,
2001; Teller et al., 2002; Clarke et al., 2003,
2004). Other cooling events dated between 8.4
and 8.0 cal. ka have been recorded in Europe,
North America, and elsewhere (e.g., von Gra-
fenstein et al., 1998; McDermott et al., 2001;
Hughen et al., 1996). Dean et al. (2002) as-
sociate the abrupt change in proportion of
land, water, and ice at the time of the drainage
of Lake Agassiz with a fundamental change
in atmospheric circulation.
It is interesting to note that the 7.7 14C ka
(8.45 cal. ka) mean calibrated age of the first
postglacial marine shells in Hudson Bay, de-
termined by Barber et al. (1999) and correlat-
ed with the final drainage of Lake Agassiz,
precedes the start of the Greenland isotopic
excursion by ;150 yr in the GISP2 core and
;250 yr in the GRIP core (Fig. 5). This dif-
ference in age between the freshwater influx
and isotopic response could be reduced by in-
creasing the ??local deviation?? (DR) for
??global-mean surface reservoir age for car-
bonate?? used by Barber et al. (1999) to values
closer to those found in Hudson Bay itself.
Alternatively, we suggest that this seeming lag
in response can be accounted for by the two-
step model for the final drainage of Lake Ag-
assiz that was proposed by Leverington et al.
(2002a), in which the second step of the final
drawdown provided the critical mass of fresh-
water that triggered a change in THC in an
ocean that had reached a relatively stable in-
terglacial mode. In fact, a two-stage cooling
around the time of the 8.2 ka event has been
identified in speleothems of Ireland (Baldini
et al., 2002) and lacustrine records in Norway
(Nesje and Dahl, 2001), and a two-step release
of Lake Agassiz waters has been modeled by
Clarke et al. (2004).
It is possible that the small isotopic excur-
sions in the GISP2 core at ca. 10.4?10.3,
10.1?10.0, and 9.8?9.9 cal. ka are related to
small Agassiz outbursts (Fig. 5). In an analy-
sis of D14C ka records of German oak and
pine, Bjo?rck et al. (1996) noted a distinct
anomaly at 10.1 cal. ka that corresponds to a
1.5?2.0 8C temperature drop in the GRIP core,
as well as other D14C anomalies at ca. 11.0
and 9.4 cal. ka. However, correlation of these
small isotopic excursions with relatively small
Agassiz outbursts is very tenuous, partly be-
cause there is uncertainty in precisely dating
the outbursts and partly because there are dif-
ferences in chronologies between the GISP2
and GRIP cores.
Geological Society of America Bulletin, May/June 2004 741
GLACIAL LAKE AGASSIZ
SUMMARY AND CONCLUSIONS
Lake Agassiz had a long history of frequent
and abrupt changes in size and depth. As the
Laurentide Ice Sheet retreated, progressively
lower outlets from Lake Agassiz carried over-
flow to the oceans through four main routes
(Fig. 5). These outflow changes, in combina-
tion with differential isostatic rebound, result-
ed in a lake that repeatedly expanded and
abruptly contracted (Fig. 4). Beaches and
wave-cut strandlines record the transgressive
maximums of at least 18 stages between 10.9
and 7.7 14C ka (12.9 and 8.45 cal. ka). Be-
tween these maximums were low-level stages
that developed when new and lower outlets
opened and brought about an abrupt draw-
down of the lake. These outbursts were fol-
lowed by a slow deepening in the southern
part of the basin, as waters south of the outlet
transgressed because of differential isostatic
rebound.
The largest Lake Agassiz outbursts were
those related to abrupt drawdowns in lake lev-
el from the Herman, Norcross, Tintah, Upper
Campbell, Stonewall, and Kinoje?vis lake stag-
es (Figs. 4A, 4B, 4C, 4E, 4N, and 4R). Three
of these outbursts occurred near the start of
large d18O isotopic excursions in the GISP2
and GRIP ice cores from Greenland (Fig. 5),
namely, those related to the Younger Dryas,
Preboreal Oscillation, and 8.2 cal. ka cooling.
These large Agassiz outbursts may have trig-
gered those cool episodes by affecting the
THC. Smaller outbursts have no clear rela-
tionship to the isotopic excursions in the ice
cores. We suggest that the difference between
the estimated 8.45 cal. ka age of the final
drainage of Lake Agassiz and the 8.2 cal. ka
cooling event may be accounted for by a two-
step drainage of Lake Agassiz or by increasing
the ??local deviation?? for the marine carbonate
reservoir effect used by Barber et al. (1999)
to date the final drainage.
ACKNOWLEDGMENTS
Funding for this research was provided by the
Natural Sciences and Engineering Research Council
(NSERC) of Canada (to Teller) and by a Smithson-
ian Institution Fellowship (to Leverington). Green-
land isotope data provided by the National Snow
and Ice Data Center, University of Colorado at
Boulder, and the World Data Center?A for Paleo-
climatology, National Geophysical Data Center,
Boulder, Colorado. Reviews by Tim Fisher and
Dave Liverman helped improve an early version of
this paper, and discussions with Wally Broecker,
George Denton, Tom Lowell, and Gary Comer
about outlets have helped advance our thinking.
REFERENCES CITED
Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Tay-
lor, K.C., and Clark, P.U., 1997, Holocene climatic
instability: A prominent widespread event 8200 years
ago: Geology, v. 25, p. 483?486.
Alley, R.B., Andandakrishnan, S., Jung, P., 2001, Stochastic
resonance in the North Atlantic: Paleoceanography,
v. 16, p. 190?198.
Baldini, J.U., McDermott, F., and Fairchild, I.J., 2002,
Structure of the 8200-year cold event revealed by a
speleothem trace element record: Science, v. 296,
p. 2203?2206.
Barber, D.C., Dyke, A., Hillaire-Marcel, C., Jennings, A.E.,
Andrews, J.T., Kerwin, M.W., Bilodeau, G., McNeely,
R., Southon, J., Morehead, M.D., and Gagnon, J.-M.,
1999, Forcing of the cold event of 8200 years ago by
catastrophic drainage of Laurentide lakes: Nature,
v. 400, p. 344?348.
Bjo?rck, S., Kromer, B., Johnsen, S., Bennike, O., Ham-
marlund, D., Lemdahl, G., Possnert, G., Rasmussen,
T.L., Wohlfarth, Hammer, C.U., and Spurk, M., 1996,
Synchronized terrestrial-atmospheric deglacial records
around the North Atlantic: Science, v. 274,
p. 1155?1160.
Bradley, R.S., 1999, Paleoclimatology: Reconstructing cli-
mates of the Quaternary (2nd edition): Boston, Aca-
demic Press, 610 p.
Broecker, W., Peteet, D., and Tind, D., 1985, Does the
ocean-atmosphere have more than one stable mode of
operation?: Nature, v. 315, p. 21?26.
Broecker, W.S., Kennett, J., Flower, B., Teller, J.T., Trum-
bore, S., Bonani, G., and Wolfli, W., 1989, Routing of
meltwater from the Laurentide Ice Sheet during the
Younger Dryas cold episode: Nature, v. 341,
p. 318?321.
Brophy, J.A., and Bluemle, J.P., 1983, The Sheyenne River:
Its geological history and effects on Lake Agassiz, in
Teller, J.T., and Clayton, L., eds., Glacial Lake Ag-
assiz: Geological Association of Canada Special Paper
26, p. 174?186.
Clark, P.U., Marshall, S.J., Clarke, G.K.C., Hostetler, S.W.,
Licciardi, J.M., and Teller, J.T., 2001, Freshwater forc-
ing of abrupt climate change during the last glaciation:
Science, v. 293, p. 283?287.
Clarke, G.K.C., Leverington, D.W., Teller, J.T., and Dyke,
A.S., 2003, Superlakes, megafloods, and abrupt cli-
mate change: Science, v. 301, p. 922?923.
Clarke, G.K.C., Leverington, D.W., Teller, J.T., and Dyke,
A.S., 2004, Paleohydraulics of the last outburst flood
from glacial Lake Agassiz and the 8200 BP cold
event: Quaternary Science Reviews, v. 23,
p. 389?407.
Clayton, L., and Moran, S.R., 1982, Chronology of late
Wisconsinan glaciation in middle North America:
Quaternary Science Reviews, v. 1, p. 55?82.
Clayton, L., Teller, J.T., and Attig, J., 1985, Surging of the
southwestern part of the Laurentide Ice Sheet: Boreas,
v. 14, p. 235?241.
Colman, S.M., Clark, J.A., Clayton, L., Hansel, A.K., and
Larsen, C.E., 1994, Deglaciation, lake levels, and
meltwater discharge in the Lake Michigan basin: Qua-
ternary Science Reviews, v. 13, p. 879?890.
Cross, M., 2002, Greenland summit ice cores, 1997, Boul-
der, Colorado: National Snow and Ice Data Center in
association with the Wold Data Center for Paleocli-
matology at NOAA-NGDC, and the Institute of Arctic
and Alpine Research, CD-ROM.
Dansgaard, W., 1961, The isotopic composition of natural
waters with special reference to the Greenland ice cap:
Meddelelser om Gr?nland, v. 165, p. 1?120.
Dean, W.E., Forester, R.M., and Bradbury, J.P., 2002, Early
Holocene change in atmospheric circulation in the
northern Great Plains: An upstream view of the 8.2
ka cold event: Quaternary Science Reviews, v. 21,
p. 1763?1775.
Dredge, L.A., 1983, Character and development of northern
Lake Agassiz and its relation to Keewatin and Hud-
sonian ice regimes, in Teller, J.T., and Clayton, L.,
eds., Glacial Lake Agassiz: Geological Association of
Canada Special Paper 26, p. 117?131.
Dredge, L.A., and Cowan, W.R., 1989, Quaternary geology
of the southwestern Canadian Shield, in Fulton, R.J.,
ed., Quaternary geology of Canada and Greenland:
Boulder, Colorado, Geological Society of America,
Geology of North America, v. K-1, p. 214?249.
Drexler, C.W., Farrand, W.R., and Hughes, J.D., 1983, Cor-
relation of glacial lakes in the Superior basin with
eastward discharge events from Lake Agassiz, in Tell-
er, J.T., and Clayton, L., eds., Glacial Lake Agassiz:
Geological Association of Canada Special Paper 26,
p. 309?329.
Dyke, A.S., and Prest, V.K., 1987, Late Wisconsinan and
Holocene history of the Laurentide ice sheet: Ge?ogra-
phie Physique et Quaternaire, v. 41, p. 237?263.
Elson, J.A., 1967, Geology of glacial Lake Agassiz, in
Mayer-Oakes, W.J., ed., Life, land, and water: Win-
nipeg, University Manitoba Press, p. 36?95.
Elson, J.A., 1983, Glacial Lake Agassiz?Discovery and a
century of research, in Teller, J.T., and Clayton, L.,
eds., Glacial Lake Agassiz: Geological Association of
Canada Special Paper 26, p. 21?41.
Fanning, A.F., and Weaver, A.J., 1997, Temporal-
geographical meltwater influences on the North Atlan-
tic Conveyor: Implications for the Younger Dryas: Pa-
leoceanography, v. 12, p. 307?320.
Fenton, M.M., Moran, S.R., Teller, J.T., and Clayton, L.,
1983, Quaternary stratigraphy and history in the south-
ern part of the Lake Agassiz basin, in Teller, J.T., and
Clayton, L., eds., Glacial Lake Agassiz: Geological As-
sociation of Canada Special Paper 26, p. 49?74.
Fisher, T.G., 2003, Chronology of glacial Lake Agassiz
meltwater routed to the Gulf of Mexico: Quaternary
Research, v. 59, p. 271?276.
Fisher, T.G., and Smith, D.G., 1994, Glacial Lake Agassiz:
Its northwest maximum extent and outlet in Saskatch-
ewan (Emerson phase): Quaternary Science Reviews,
v. 13, p. 845?858.
Fisher, T.G., and Souch, C., 1998, Northwest outlet chan-
nels of Lake Agassiz, isostatic tilting and the migrat-
ing continental drainage divide, Saskatchewan, Can-
ada: Geomorphology, v. 25, p. 57?73.
Fisher, T.G., Smith, D.G., and Andrews, J.T., 2002, Pre-
boreal oscillation caused by a glacial Lake Agassiz
flood: Quaternary Science Reviews, v. 21,
p. 873?878.
Ganopolski, A., and Rahmstorf, S., 2001, Rapid changes of
glacial climate simulated in a coupled climate model:
Nature, v. 409, p. 153?158.
GLOBE Task Team, 1999, The Global One-kilometer Base
Elevation (GLOBE) elevation model. Version 1.0:
Boulder, Colorado, National Oceanic and Atmospheric
Administration, National Geophysical Data Center,
http://www.ngdc.noaa.gov/mgg/global/.
Grootes, P.M., and Stuiver, M., 1997, Oxygen 18/16 vari-
ability in Greenland snow and ice with 1023 to 1025-
year time resolution: Journal of Geophysical Research,
v. 102, p. 26,455?26,470.
Hughen, K.A., Overpeck, J.T., Trumbore, S., and Peterson,
L.C., 1996, Rapid climate changes in the tropical re-
gion during the last deglaciation: Nature, v. 380,
p. 51?54.
Johnsen, S.J., Dahl-Jensen, D., Gundestrip, N., Steffensen,
J., Clausen, H., Miller, H., Masson-Delmotte, V., Sve-
inbjo?rnsdottir, A., and White, J., 2001, Oxygen isotope
and paleotemperature records from six Greenland ice-
core stations: Camp Century, Dye-3, GRIP, GISP2,
Renland and NorthGRIP: Journal of Quaternary Sci-
ence, v. 16, p. 299?307.
Johnston, W.A., 1946, Glacial Lake Agassiz, with special
reference to the mode of deformation of the beaches:
Geological Survey of Canada Bulletin 7, 20 p.
Karrow, P.R., and Calkin, P.E., eds., 1985, Quaternary evo-
lution of the Great Lakes: Geological Association
Canada Special Paper 30, 258 p.
Karrow, P.F., and Occhietti, S., 1989, Quaternary geology
of the St. Lawrence Lowlands of Canada, in Fulton,
R.J., ed., Quaternary geology of Canada and Green-
land: Boulder, Colorado, Geological Society of Amer-
ica, Geology of North America, v. K-1, p. 321?389.
Klassen, R.W., 1983a, Assiniboine delta and the
Assiniboine-Qu?Appelle valley system?Implications
concerning the history of Lake Agassiz in southwest-
ern Manitoba, in Teller, J.T., and Clayton, L., eds.,
742 Geological Society of America Bulletin, May/June 2004
TELLER and LEVERINGTON
Glacial Lake Agassiz: Geological Association of Can-
ada Special Paper 26, p. 211?229.
Klassen, R.W., 1983b, Lake Agassiz and the late glacial
history of northern Manitoba, in Teller, J.T., and Clay-
ton, L., eds., Glacial Lake Agassiz: Geological As-
sociation of Canada Special Paper 26, p. 97?115.
Klassen, R.W., 1989, Quaternary geology of the southern
Canadian interior plains, in Fulton, R.J., ed., Quater-
nary geology of Canada and Greenland: Boulder, Col-
orado, Geological Society of America, Geology of
North America, v. K-1, p. 138?173.
Larsen, C.E., 1987, Geological history of glacial Lake Al-
gonquin and the upper Great Lakes: U.S. Geological
Survey Bulletin B1801, 35 p.
Leverington, D.W., and Teller, J.T., 2003, Paleotopographic
reconstructions of the eastern outlets of glacial Lake
Agassiz: Canadian Journal of Earth Sciences, v. 40,
p. 1259?1278.
Leverington, D.W., Mann, J.D., and Teller, J.T., 2000,
Changes in the bathymetry and volume of glacial
Lake Agassiz between 11,000 and 9300 14C yr B.P.:
Quaternary Research, v. 54, p. 174?181.
Leverington, D.W., Mann, J.D., and Teller, J.T., 2002a,
Changes in the bathymetry and volume of glacial
Lake Agassiz between 9200 and 7700 14C yr B.P.:
Quaternary Research, v. 57, p. 244?252.
Leverington, D.W., Teller, J.T., and Mann, J.D., 2002b, A
GIS model for reconstruction of late Quaternary land-
scapes from isobase data and modern topography:
Computers & Geosciences, v. 28, p. 631?639.
Lewis, C.F.M., Moore, T.C., Jr., Rea, D.K., Dettman, D.L.,
Smith, A.J., and Mayer, L.A., 1994, Lakes of the Hu-
ron basin: Their record of runoff from the Laurentide
Ice Sheet: Quaternary Science Reviews, v. 13,
p. 891?922.
Licciardi, J.M., Teller, J.T., and Clark, P.U., 1999, Fresh-
water routing by the Laurentide Ice Sheet during the
last deglaciation, in Clark, P.U., Webb, R.S., and Keig-
win, L.D., eds., Mechanisms of global climate change
at millennial time scales: American Geophysical
Union Geophysical Monograph 112, p. 171?202.
Manabe, S., and Stouffer, R.J., 1995, Simulation of abrupt
climate change induced by freshwater input to the
North Atlantic Ocean: Nature, v. 378, p. 165?167.
Manabe, S., and Stouffer, R.J., 1997, Coupled ocean-
atmosphere model response to freshwater input: Com-
parison to Younger Dryas event: Paleoceanography,
v. 12, p. 321?336.
Mann, J.D., Leverington, D.W., Rayburn, J., and Teller, J.T.,
1999, The volume and paleobathymetry of glacial
Lake Agassiz: Journal of Paleolimnology, v. 22,
p. 71?80.
McDermott, F., Mattey, D.P., and Hawkesworth, C., 2001,
Centennial-scale Holocene climate variability revealed
by a high-resolution speleothem d18 record from SW
Ireland: Science, v. 294, p. 1328?1131.
Mothersill, J.S., 1988, Paleomagnetic dating of late glacial
and postglacial sediments in Lake Superior: Canadian
Journal of Earth Sciences, v. 25, p. 1791?1799.
Nesje, A., and Dahl, S.O., 2001, The Greenland 8200 cal
yr BP event detected in loss-on-ignition profiles in
Norwegian lacustrine sediment sequences: Journal of
Quaternary Science, v. 16, p. 155?166.
NGDC (National Geophysical Data Center), 1988, ETOPO
5 Data Announcement 88-MGG-02, in Digital relief
of the surface of the Earth: Boulder, Colorado, NOAA
(National Oceanic and Atmospheric Administration),
National Geophysical Data Center, http://
www.ngdc.noaa.gov/mgg/global/.
O?Brien, S.R., Mayewski, P.A., Meeker, L.D., Meese, D.A.,
Twickler, M.S., and Whitlow, S.I., 1995, Complexity
of Holocene climate as reconstructed from a Green-
land ice core: Science, v. 270, p. 1962?1964.
Prest, V.R., 1970, Quaternary geology of Canada, in Doug-
las, R.J.W., ed., Geology and economic minerals of
Canada: Geological Survey of Canada, Economic Ge-
ology Series, no. 1, p. 675?764.
Rahmstorf, S., 1995, Bifurcations of the Atlantic thermo-
haline circulation in response to changes in the hy-
drological cycle: Nature, v. 378, p. 145?149.
Rahmstorf, S., 2000, The thermohaline ocean circulation:
System with dangerous thresholds?: Climate Change,
v. 46, p. 247?256.
Rayburn, J.A., and Teller, J.T., 1999, Northwestern drainage
of glacial Lake Agassiz: New information and a re-
vised interpretation: Calgary, CANQUA-CGRC Meet-
ing, Program and Abstracts, p. 60?61.
Renssen, H., Goose, H., Fichefet, T., and Campin, J.-M.,
2001, The 8.2 kyr BP event simulated by a global
atmospheric-sea-ice-ocean model: Geophysical Re-
search Letters, v. 28, p. 1567?1570.
Schreiner, B.T., 1983, Lake Agassiz in Saskatchewan,
inTeller, J.T., and Clayton, L., eds., Glacial Lake Ag-
assiz: Geological Association of Canada Special Paper
26, p. 75?96.
Smith, D.G., and Fisher, T.G., 1993, Glacial Lake Agassiz:
The northwestern outlet and paleoflood: Geology,
v. 21, p. 9?12.
Southon, J., 2002, A first step to reconciling the GRIP and
GISP2 ice-core chronologies, 0?14,500 yr B.P.: Qua-
ternary Research, v. 57, p. 32?37.
Stuiver, M., and Reimer, P.J., 1993, Extended 14C data base
and revised CALIB 3.0 14C age calibration program:
Radiocarbon, v. 35, p. 215?230.
Stuiver, M., Grootes, P.M., and Braziunas, T.F., 1995, The
GISP2 18O climate record of the past 16,500 years and
the role of the Sun, ocean, and volcanoes: Quaternary
Research, v. 44, p. 341?354.
Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S.,
Hughen, K.A., Kromer, B., McCormac, F.G., van der
Plicht, H., and Spurk, M., 1998, INTCAL98 radiocar-
bon age calibration 24,000?0 cal BP: Radiocarbon,
v. 40, p. 1041?1083.
Teller, J.T., 1985, Glacial Lake Agassiz and its influence on
the Great Lakes, in Karrow, P.F., and Calkin, P.E., eds.,
Quaternary evolution of the Great Lakes: Geological
Association of Canada Special Paper 30, p. 1?16.
Teller, J.T., 1987, Proglacial lakes and the southern margin
of the Laurentide Ice Sheet, in Ruddiman, W.F., and
Wright, H.E., eds., North America and adjacent
oceans during the last deglaciation: Boulder, Colorado,
Geological Society of America, Geology of North
America, v. K-3, p. 39?69.
Teller, J.T., 1990a, Volume and routing of late glacial runoff
from the southern Laurentide Ice Sheet: Quaternary
Research, v. 34, p. 12?23.
Teller, J.T., 1990b, Meltwater and precipitation runoff to the
North Atlantic, Arctic, and Gulf of Mexico from the
Laurentide Ice Sheet and adjacent regions during the
Younger Dryas: Paleoceanography, v. 5, p. 897?905.
Teller, J.T., 1995, The impact of large ice sheets on conti-
nental paleohydrology, in Gregory, K., Baker, V., and
Starkel, L., eds., Global continental paleohydrology:
New York, John Wiley and Sons, p. 109?129.
Teller, J.T., 2001, Formation of large beaches in an area of
rapid differential isostatic rebound: The three-outlet
control of Lake Agassiz: Quaternary Science Reviews,
v. 20, p. 1649?1659.
Teller, J.T., 2004, Controls, history, outbursts, and impact
of large late-Quaternary proglacial lakes in North
America, in Gilespie, A., Porter, S., and Atwater, B.,
eds., The Quaternary Period in the United States: De-
velopments in Quaternary Science, v. 1: Amsterdam,
Elsevier, p. 45?61.
Teller, J.T., and Clayton, L., eds., 1983, Glacial Lake Ag-
assiz: Geological Association of Canada Special Paper
26, 451 p.
Teller, J.T., and Kehew, A.E., eds., 1994, Late glacial his-
tory of proglacial lakes and meltwater runoff along
the Laurentide Ice Sheet: Quaternary Science Re-
views, v. 13, p. 795?981.
Teller, J.T., and Mahnic, P., 1988, History of sedimentation
in the northwestern Lake Superior basin and its rela-
tion to Lake Agassiz overflow: Canadian Journal of
Earth Sciences, v. 25, p. 1660?1673.
Teller, J.T., and Thorleifson, L.H., 1983, The Lake Agassiz?
Lake Superior connection, inTeller, J.T., and Clayton,
L., eds., Glacial Lake Agassiz: Geological Association
of Canada Special Paper 26, p. 261?290.
Teller, J.T., Thorleifson, L.H., Dredge, L.A., Hobbs, H.C.,
and Schreiner, B.T., 1983, Maximum extent and major
features of Lake Agassiz, in Teller, J.T., and Clayton,
L., eds., Glacial Lake Agassiz: Geological Association
of Canada Special Paper 25, p. 43?45.
Teller, J.T., Risberg, J., Matile, G., and Zoltai, S., 2000,
Postglacial history and paleoecology of Wampum,
Manitoba, a former lagoon in the Lake Agassiz basin:
Geological Society of America Bulletin, v. 112,
p. 943?958.
Teller, J.T., Leverington, D.W., and Mann, J.D., 2002,
Freshwater outbursts to the oceans from glacial Lake
Agassiz and their role in climate change during the
last deglaciation: Quaternary Science Reviews, v. 21,
p. 879?887.
Thorleifson, L.H., 1996, Review of Lake Agassiz history,
in Teller, J.T., Thorleifson, L.H., Matile, G., and Bris-
bin, W.C., eds., Sedimentology, geomorphology and
history of the central Lake Agassiz basin: Winnipeg,
Manitoba, Field Trip Guidebook B2, Geological As-
sociation of Canada/Mineralogical Association of
Canada Annual Meeting, May 27?29, p. 55?84.
Upham, W., 1895, The glacial Lake Agassiz: U.S. Geolog-
ical Survey Monograph 25, 658 p.
Veillette, J.J., 1994, Evolution and paleohydrology of gla-
cial Lakes Barlow and Ojibway: Quaternary Science
Reviews, v. 13, p. 945?971.
Vincent, J.-S., and Hardy, L., 1979, The evolution of glacial
Lakes Barlow and Ojibway, Quebec and Ontario:
Geological Survey of Canada Bulletin 316, 18 p.
von Grafenstein, U., Erlenkeuser, H., Mu?ller, J., Jouzel, J.,
and Johnsen, S., 1998, The cold event 8200 years ago
documented in oxygen isotope records of precipitation
in Europe and Greenland: Climate Dynamics, v. 14,
p. 73?81.
Walcott, R.I., 1972, Late Quaternary vertical movements in
eastern North America: Quantitative evidence of
glacio-isostatic rebound: Reviews of Geophysics and
Space Physics, v. 10, p. 849?884.
MANUSCRIPT RECEIVED BY THE SOCIETY 4 JANUARY 2003
REVISED MANUSCRIPT RECEIVED 8 SEPTEMBER 2003
MANUSCRIPT ACCEPTED 23 SEPTEMBER 2003
Printed in the USA