Considerable uncertainty surrounds the extent and timing of the advance and retreat of the Greenland Ice Sheet (GIS) on the continental shelf bordering Baffin Bay during the last glacial cycle. Here we use marine geophysical and geological data to show that fast-flowing ice sheet outlets, including the ancestral Jakobshavn Isbræ, expanded several hundred kilometers to the shelf edge during the last glaciation ca. 20 ka. Retreat of these outlets was asynchronous. Initial retreat from the shelf edge was underway by 14,880 calibrated (cal) yr B.P. in Uummannaq trough. Radiocarbon dates from the adjacent Disko trough and adjoining trough-mouth fan imply later deglaciation of Jakobshavn Isbræ, and, significantly, an extensive readvance and rapid retreat of this outlet during the Younger Dryas stadial (YD). This is notable because it is the first evidence of a major advance of the GIS during the YD on the West Greenland shelf, although the short duration suggests that it may have been out of phase with YD temperatures.
The mass balance and stability of polar ice sheets is related strongly to the fast-flowing ice streams and outlet glaciers that deliver the majority of ice to the surrounding ocean (Bamber et al., 2007). In the case of the Greenland Ice Sheet (GIS), a major focus of current research is short-term changes to the dynamics of its marine-terminating outlets (e.g., Moon and Joughin, 2008). Despite their importance to GIS mass balance, however, our understanding of their longer-term, centennial- to millennial-scale behavior remains poor (Alley et al., 2010). This is particularly the case in central West Greenland where fast-flowing outlets, including Jakobshavn Isbræ, drain large ice-sheet basins (105 km2) into fjords that open onto the inner continental shelf from where bathymetric troughs extend to the shelf edge (Fig. 1). Submarine fans occur offshore of the mouths of these troughs (Fig. 1). To date, the record of GIS history preserved in these troughs and fans has been little investigated.
This paper focuses on two troughs: Disko trough, which extends 370 km from the shelf edge to the mouth of Jakobshavn Isfjord, and Uummannaq trough, which extends 300 km from the shelf edge into Uummannaq Fjord (Fig. 1). Multibeam swath bathymetry, subbottom profiles, and sediment cores from these troughs provide information on seafloor morphology, acoustic stratigraphy, sedimentology, and chronology related to the former expansion of GIS outlets. In this paper, we describe landforms and sediments recording the advance and retreat of these outlets across the shelf; we discuss the associated chronology and highlight the wider paleoglaciological significance for GIS extent during the Last Glacial Maximum (LGM) and Younger Dryas (YD; 12.8–11.7 ka; Steffensen et al., 2008).
EVIDENCE FOR GROUNDED ICE FLOW WITHIN CROSS-SHELF TROUGHS
In Uummannaq Fjord and the adjoining shelf trough, streamlined subglacial bedforms up to 16 km long, 0.6 km wide, and with elongation ratios of 1:40 are present on the seafloor. In dimensions, the bedforms are consistent with megascale glacial lineations, although some appear to initiate from bedrock outcrops and resemble crags and tails (Fig. 2A). Such elongate subglacial bedforms are recognized as a product of formation beneath fast-flowing ice and their presence is diagnostic of former ice streaming (cf. King et al., 2009). The bedforms can be traced westward along the trough to the shelf edge. At the shelf edge, subbottom profiler records imaged a sediment ridge over 10 m high (Figs. 2B–2C). Core VC45 from the ridge (Fig. 3) bottomed out in a massive, stiff diamicton consistent with subglacial till. The ridge is interpreted as a moraine formed at the terminus of a fast-flowing outlet that extended along the trough to the shelf edge. A radiocarbon date of 14,880 calibrated (cal) yr B.P. (15,519–14,243 cal yr B.P.; laboratory code AA-89913) (Fig. 3; Table DR2 in the GSA Data Repository1) on benthic foraminifers from glaciomarine mud 5 cm above the till constrains the timing of retreat of the Uummannaq outlet from the shelf edge. This interpretation is supported by a date of 14,060 cal yr B.P. from glaciomarine mud above glaciogenic debris flows in a core from the Uummannaq fan, which provides a minimum age for retreat from the shelf edge (Ó Cofaigh et al., 2012).
In Disko Bugt, drumlins, crags and tails, and crudely streamlined forms up to 4 km long and 0.8 km wide characterize the seafloor and record westerly ice flow across the bay from Jakobshavn Isfjord (Fig. 2D). Toward the western edge of the bay, bedform orientation switches to southwest (reflecting bedrock structure), recording flow via a narrow, 1000-m-deep trough onto the mid-shelf where, in turn, subglacial lineations 6 km long and with elongation ratios of at least 1:21 record flow onto the outer shelf (Fig. 2E).
In the outer trough, subbottom profiles imaged a sediment drape up to 4 m thick (Fig. DR1 in the Data Repository). Sediment cores show that the drape comprises laminated and massive glaciomarine muds that are underlain by stiff diamicton. Core VC20 was collected in the outer-shelf trough, 65 km from the shelf edge (Figs. 1 and 3). The lowest unit, from 539 to 520 cm, comprises diffusely laminated sand. This is overlain by 95 cm of stiff, massive diamicton with occasional low-angle planar discontinuities. The sand is separated from the diamicton by 7 cm of contorted sand and diamicton. Thin sections from the diamicton show parallel fractures, plasma streaks, fractured grains, turbate structures, and lineaments (see the Data Repository), consistent with subglacial shear (Hiemstra and Rijsdijk, 2003). A subglacial interpretation is also supported by poor sorting, high shear strength, and massive structure.
Both the till and underlying sand contain rare whole and fragmented shells. A valve of Nuculana pernula in core VC20 from 524–525 cm in the deformed zone was dated to 12,240 cal yr B.P. (12,522–11,952 cal yr B.P.; lab code Beta-265217) while a shell fragment from 477.5–480 cm in the till was dated to 12,050 cal yr B.P. (12,346–11,757 cal yr B.P.; Beta-265216) (Fig. 3). Because the shells are not in situ, accelerator mass spectrometry (AMS) radiocarbon dates on them provide a maximum age for the till and associated ice advance. Retreat of this outlet is constrained by dates on marine shells in cores from the outer-mid shelf. A shell fragment from 408 cm in ice-proximal glaciomarine sediments above the till in core VC20 was dated to 12,240 cal yr B.P. (12,523–11,958 cal yr B.P.; AA91731) (Fig. 3). The fragmented nature of the sample, and its similarity in age to the shells within and beneath the till, indicates that it is a reworked sample, incorporated into ice-proximal glaciomarine sediments during retreat. It provides a maximum age for deglaciation of the outer-shelf trough. Further constraint on the timing of retreat is provided by a date of 12,050 cal yr B.P. (12,346–11,757 cal yr B.P.; McCarthy, 2011) on in situ shells from glaciomarine sediments in mid-shelf core MSM-343340 (from RV Maria S. Merian cruise MSM05–3) (Fig. DR5; Table DR2). The age range of this date overlaps with the dates on reworked shells in till in core VC20, suggesting a short-lived readvance onto the outer shelf followed by rapid retreat to the mid-shelf during the YD. The retreat occurred over at least 110 km, implying a retreat rate of 22–275 m a−1, consistent with recent retreat rates of marine-terminating GIS outlets (e.g., Moon and Joughin, 2008; Bjørk et al., 2012)
Cores from the Disko fan provide additional support for a YD readvance onto the outer shelf. Normally graded, massive and laminated sands (Fig. 3; Fig. DR4) are consistent with deposition by mass flow when the ice sheet was at the shelf edge (Ó Cofaigh et al., 2003). Dates on reworked shells from these mass-flow units give a maximum age for sediment delivery to the slope. The youngest dates range from 13.8 to 12.2 ka (Fig. 3; Table DR2). Hence, shells from organisms living on the shelf and slope as recently as 12.2 ka (fan and shelf core VC20) to 12.0 ka (VC20) were incorporated into till and mass flows. In core VC35, an increase in ice-rafted debris (IRD) above 160 cm is constrained by a date of 12,270 cal yr B.P. (Beta-272271) at 175 cm (Fig. 3). Increased IRD input to the fan after 12.2 ka is consistent with evidence from core VC20, supporting a YD readvance and retreat on the outer shelf.
IMPLICATIONS FOR GIS HISTORY DURING THE LGM AND YD
This is the first study of GIS extent, flow dynamics, and retreat from the central West Greenland shelf. Previous reconstructions of the GIS in West Greenland have been based largely on terrestrial and nearshore data (e.g., Roberts and Long, 2005; Funder et al., 2011). Subglacial landforms, combined with till in dated cores, provide direct evidence for an extensive grounded GIS on the West Greenland shelf during the last glaciation. Major ice-sheet outlets emanated from Disko Bugt and Uummannaq Fjord and drained to the shelf edge.
Our data further imply that retreat of adjacent GIS outlets was asynchronous (Fig. 4). Retreat in Uummannaq trough was underway by 14.8 ka. In Disko trough, the available chronology indicates that retreat did not commence until 13.8 ka, but the margin of Jakobshavn Isbræ subsequently underwent a major fluctuation during the YD (Fig. 4). Currently we cannot determine if the Uummannaq outlet also underwent a YD readvance. Regionally, however, two sets of moraines have been mapped on the Southwest Greenland shelf, and it has been suggested that the outer most “Hellefisk moraines” are LGM in age, and the inner “Fiskebanke moraines” date to the YD (Weidick et al., 2004; Funder et al., 2011). The moraines are undated but the YD advance of Jakobshavn Isbræ onto the shelf documented here suggests that it is at least plausible that the Fiskebanke moraines are of YD age. The short duration of this readvance, combined with the fact that retreat also took place during the stadial, suggests that the dynamic response of Jakobshavn Isbræ may have been out of phase with YD temperatures. This is consistent with published records that indicate spatial variability in Greenland glacier and ice-sheet margin response in the YD (Jennings et al., 2006; Kelly et al., 2008; Carlson et al., 2008; Funder et al., 2011).
Considerable differences between West and East Greenland with respect to the timing of GIS retreat from the continental shelf are also implied (Fig. 4). Retreat in East Greenland was under way by 18–17 ka (Evans et al., 2002; Jennings et al., 2006; Håkansson et al., 2007), leading retreat in West Greenland by 2000–3000 yr (Fig. 4). The influence of warm ocean water impinging on the shelf has been proposed as the driver for retreat in the East (Jennings et al., 2006; Knutz et al., 2011), but the controls on initial retreat in the West are less clear. There is evidence for the warm West Greenland Current (WGC) influencing subsurface waters in northern Baffin Bay by 10.9 ka (Knudsen et al., 2008) and coastal waters around Thule from 10.2 ka (Funder, 1990). However, the arrival of the WGC on the mid-shelf in Disko trough, and thence into Disko Bugt, did not occur until after ca. 9 ka (Lloyd et al., 2005; McCarthy, 2011), later than the YD retreat of that outlet.
An alternative explanation for the short-lived YD advance and retreat in Disko trough is that it represents a surge, controlled, at least in part, by subglacial topography and ice thickness. The overall sinuous nature of the trough, and the islands at the mouth of Disko Bugt, could have acted to stabilize grounded ice during advance. This is analogous to the role of topographic pinning points on retreating fjord glaciers. However, surging would have lowered the ice surface profile via dynamic thinning, and Carlson et al. (2008) proposed that southern GIS outlets were subject to surface ablation during the YD. A thin outlet would have been more susceptible to collapse and rapid retreat. This underscores the importance of understanding ice-stream processes, dynamics, and bed topography at the scale of individual ice-sheet drainage basins if we are to accurately predict how such ice masses might respond to future climate change and sea-level rise.
This work was funded by the UK Natural Environment Research Council and the U.S. National Science Foundation. We thank I. Leighton and A. Ratcliffe for assistance with the microfabric data. Insightful reviews by S. Funder and an anonymous reviewer improved the manuscript. This paper is a contribution to the PAST Gateways (Palaeo-Arctic Spatial and Temporal Gateways) program.