early BA temperatures were nearly as warm as
those of the earliest Holocene when Northern
Hemisphere summer insolation peaked (Fig.
4). SST likewise rose by 4°C off the British
Columbian coast (15) between 15.5 and 14 ka.
Temperature increases of this magnitude pro-
vided the mechanism to rapidly melt the CIS and
place its equilibrium line potentially above all
but the highest peaks in the region (19, 20).
Substantial mass loss during this period may
also explain depletion of d18O in planktic fo-
raminifera (25) from the Gulf of Alaska via
ocean surface freshening (Fig. 4). GIA results
(8, 21) show that between 14.5 and 14.0 ka, the
CIS alone contributed 2.5 to 3.0 m of sea-level
rise, representing about 17.5 to 21% of melt-
water pulse 1A (Fig. 4). Those models also re-
veal that the CIS lost nearly one-half of its Last
Glacial Maximum mass in as little as 500 years
(Fig. 4) and changed from a continental-scale ice
sheet to a complex network of alpine glaciers, ice
fields, and ice caps. Our data also confirm that
the CIS had disappeared by the end of the YD
(Fig. 1), likely in response to abrupt warmth that
signaled the end of the Pleistocene (1).
Ironically, strong loss of mass from the CIS
and the adjacent Laurentide Ice Sheet during the
early BA may have helped trigger climate dete-
rioration that slowed its demise and allowed
alpine and cirque glaciers to advance (Fig. 4).
As shown in the TraCE-21ka experiment (22)
and in more recent experiments of a combined
general circulation model and hydrologic model
that simulates Laurentide and Cordilleran Ice
Sheet saddle collapse at ~14.5 ka (20), introduc-
tion of meltwater into the Arctic and Atlantic
oceans can weaken the Atlantic meridional over-
turning circulation (AMOC). This weakening, in
turn, can lead to widespread hemispheric cooling
that could cause alpine glaciers and margins of
the CIS to advance into newly deglaciated terrain.
Meltwater-induced changes in AMOC are, like-
wise, hypothesized to have initiated the YD (26).
The greatly reduced volume of the CIS before the
inception of the YD (Fig. 4) argues against CIS
meltwater playing any substantial role as a trig-
ger for the YD.
Our study reveals that the CIS and associated alpine glaciers responded to abrupt climate
change (Fig. 4) and challenges the traditional
view that the CIS covered large regions of westernmost Canada as late as 12.5 ka. Our results
also support GIA models that place major mass
loss of the CIS in line with other Northern Hemisphere ice sheets at ~14.5 ka, in response to
abrupt, hemisphere-wide climate amelioration
(3, 4). Alpine areas in the region had clearly
emerged from the ice sheet before the BA, and
alpine glaciers advanced soon after the onset of
the BA and during the YD (Fig. 4).
The current understanding of the peopling of
the Americas requires humans to have migrated
to the south of the ice sheets after the end of
the Last Glacial Maximum (~18 ka) but before
14.6 ka (18). Existing geologic evidence (17) and
ice sheet models (9) rule out migration between
the Cordilleran and Laurentide ice sheets before
13.4 ka. Although some intermediate and high
elevation sites became ice free during the BA,
lower elevations in the interior of British Columbia
did not become ice free until the end of the YD
(Fig. 1), making the route for human migration
across the Cordillera unlikely during the latest
The complexity of CIS decay can explain the
age equivalence of moraines (Fig. 2) constructed
by glaciers of different lengths (see SM). The
thinning CIS exposed rugged mountainous areas
that were transformed into a labyrinth of valley
glaciers. In such a scenario, cirque glaciers might
reform and construct small moraines during
climate reversals such as the BA and YD. Simultaneously, valley glaciers left from the decaying ice sheet might reinvigorate and advance
to positions many kilometers down-valley from
cirque headwalls during climate reversals. Although marginal retreat was common during
the demise of the CIS (6), especially over lower
relief topography, our hypothesis for substantial
ice loss at high elevations accords with both
conceptual (27) and GIA-based models (8, 21).
Numerical ice flow models (5, 7, 19) support widespread mass loss through marginal retreat and
thinning of the CIS during the period 14 to 10 ka,
and moderate resolution (5 km) models (5)
100° W 60° W
[14.5 - 14 ka]
[14.5 - 14 ka]
Fig. 3. Rapid ice sheet thinning and climate anomalies within the Bølling-Allerød and
Younger Dryas. (A) GIA-modeled (8, 21) surface ice elevation change for the Cordilleran Ice
Sheet and western sector of the Laurentide Ice Sheet between 14.5 and 14.0 ka. (B) (Left)
TraCE-simulated (22) anomalies in surface air temperature between 14.0 to 13.8 ka and
14.4 to 14.2 ka. Dark gray shading is the extent of ice cover at 14 ka. (Right) Thicker and dashed
contours denote, respectively, zero and negative anomalies in surface air temperature from
12.2 to 12.0 ka and 12.6 to 12.5 ka. Dark gray shading (8) is the extent of ice cover at 12 ka.
(C) Same as (B) but for precipitation anomalies.