Introduction

Baffin Bay, located between Canada and Greenland, was flanked by three large continental ice sheets during the Last Glacial Maximum (LGM; here considered lasting from ~25 to 18 ka BP)1,2,3,4. The Laurentide (LIS), Innuitian (IIS), and Greenland (GIS) ice sheets advanced across the continental shelves surrounding Baffin Bay and are postulated to have reached the shelf edge in many places (Fig. 1)3,4,5,6,7. In northern Baffin Bay, the LIS and GIS merged with the IIS, forming a continuous ice sheet with an ice shelf8. With the deglacial climate warming, the ice sheets retreated landward from their glacial marine-terminating ice margins to their present-day close-to-minimum extent, mainly being confined to onshore settings9,10. Around Baffin Bay today, the GIS margin is mostly land-based, while the LIS and IIS have disintegrated into smaller ice masses (e.g., Agassiz, Barnes, and Penny ice caps) and glaciers over Baffin Island and Canadian Arctic Archipelago (Fig. 1c).

Fig. 1: Overview maps showing the geographic location of Baffin Bay, its main oceanographic currents and glacial history, and core locations.
figure 1

a Baffin Bay is bordered to the north by the Canadian Arctic Archipelago (CAA), Baffin Island to the west, and Greenland to the east. The Baffin Island Current (BIC) transports colder and fresher waters originating from the Arctic Ocean southwards through Davis Strait, merging with the Labrador Current (LC) flowing into the North Atlantic Ocean. The West Greenland Current (WGC), an admixture of the Arctic-sourced East Greenland Current (EGC) and Atlantic-sourced Irminger Current, carries warmer and more saline waters into Baffin Bay along the west Greenland coast. b Reconstructed maximum extent of the Laurentide (LIS), Innuitian (IIS), and Greenland (GIS) ice sheets during the LGM (~25 ka) after Batchelor et al.3 (solid line) and Dalton et al.7 (dotted line, truncated around NW Greenland). c The location of the 79 radiocarbon-dated sediment cores (filled circles) compiled in this study (gray circles indicate cores with basal till, and red circles without) distributed into six depocenters (differentiated by black and gray dashed lines): (i) Central Baffin Bay (CBB), (ii) Baffin Bay Slope (BBS), (iii) Outer, (iv) Mid, (v) Inner Baffin Island Shelf (BIS) and West Greenland Shelf (WGS), and (vi) Northern Baffin Bay (NBB). Note that all cores taken from an Outer Shelf depocenter are from the WGS and none from the BIS. The locations of cross-shelf troughs on the WGS (e.g., Melville Bay, Upernavik, Uummannaq, Disko Bay) are indicated. The light gray cover represents the present-day ice margin (after Harrison et al.70). Bathymetric information in (b) and (c) is based on the International Bathymetric Chart of the Arctic Ocean Version 3.071.

Full size image

Ice-sheet dynamics and the associated changes in the rate and pattern of erosion, and also the mode of sediment transport, had a strong control on sediment deposition in Baffin Bay since the last glacial period11,12. However, the reconstruction of the LGM to modern sedimentation history in Baffin Bay is limited by the scarcity of directly dated sediment cores due to the paucity of datable material in the sediments11,12,13,14,15,16. Because recent developments in accelerator mass spectrometry (AMS) dating techniques have reduced the amount of material needed for radiocarbon-based age determinations17, data availability is increasing rapidly18,19,20. This has led to an increase in field studies providing local information on sedimentation rates in Baffin Bay, including refinements of deglacial ice-margin dynamics (timing/pattern) across the basin19,20,21,22,23. Previous reconstructions of paleo-ice sheet extent in Baffin Bay were based largely on undated glacial geomorphological evidence3,7,9,10. Despite this progress, no single study to date focuses on the overall development of sedimentation patterns in Baffin Bay since the maximum ice extent during the LGM.

Taking advantage of this data availability, our study presents a new spatio-temporal analysis of radiocarbon-based sedimentation rate patterns across Baffin Bay since the LGM, based on an extensive collection of both previously published (n = 68) and new radiocarbon-dated sediment cores (n = 11). Besides documenting the successive activation of various depocenters, it provides detailed information on the pattern of ice margin retreat and new sedimentation rate-based estimates of subglacial erosion rates (for the GIS).

Sedimentation in Baffin Bay over the last 25,000 years

Baffin Bay is a >2000 m deep semi-enclosed oceanic basin between the Canadian Arctic Archipelago and Greenland, connecting the Arctic Ocean to the North Atlantic Ocean (Fig. 1). Its continental shelves (up to 500 m deep) are dissected by U-shaped over-deepened cross-shelf troughs (up to 100 km wide and 950 m deep, Fig. 1), that were carved and filled by fast-flowing ice streams associated with the repeated advance of the LIS, IIS, and GIS during previous Quaternary glaciations4,10,24,25. During full-glacial conditions, the advance of these paleo-ice streams through the cross-shelf troughs enabled the delivery of large volumes of eroded terrigenous and shelf material beyond the trough mouth, forming huge sediment aprons on the adjacent continental slopes, termed trough mouth fans5,26,27,28,29.

The new compilation of 79 14C-dated sediment cores from Baffin Bay (Fig. 1c and Supplementary Table 1) comprises primarily previously published data complemented by eleven new cores. Only cores with a minimum of two 14C ages, all based on marine micro- and macro-fossil remains, were included in this compilation (see the “Methods” section and Supplementary Table 2). Sedimentation rate (SR) in the majority (86%) of cores in this compilation is constrained by at least three 14C ages, with an overall range of 2–26 14C ages per core (see the “Methods” section for a rationale of using 2 ages as a minimum and Supplementary Fig. 1). Regionally, the 79 sediment cores are interpreted to represent six individual depocenters, based on a combination of their geographical location, water depth, and relative distance from the present-day coastline (Fig. 1c and Supplementary Table 1): (1) the deep basin (7 cores), which largely comprises water depths >1500 m, termed as Central Baffin Bay (CBB); (2) the slope (15 cores) occupying the water depth interval of ~1500 m until the shelf break (~500 m), referred here as Baffin Bay Slope (BBS); (3) the Outer Shelf (3 cores), with all three cores from the West Greenland Shelf (WGS); (4) the Mid WGS and Baffin Island Shelf (BIS) (8 cores); (5) the WGS and BIS Inner Shelves (30 cores, incl. 4 fjord sites); and (6) the Northern Baffin Bay (NBB) (16 cores), which is treated here as a separate depocenter due to its location at the confluence of the LIS, IIS, and GIS.

The age-depth relationships for these 79 cores were constructed using the rapid Bayesian approach provided by the open-source software UNDATABLE in MATLAB30. All 14C ages were re-calibrated applying the Marine20 calibration curve31 and region-specific marine reservoir age corrections (Supplementary Table 2). The resulting median SRs from UNDATABLE were then binned into 1 ka time-slices (Supplementary Fig. 2 and Supplementary Table 3) and combined in a boxplot for all cores in each depocenter (Fig. 2). In addition, mean SRs were calculated for each sediment core for the four time-intervals considered here (see below and Fig. 3).

Fig. 2: Sedimentation rates in Baffin Bay and Greenland Ice Sheet and climate dynamics.
figure 2

1-kyr-binned sedimentation rates of all sediment cores within a specific depocenter in Baffin Bay are compared with model and proxy reconstructions of Greenland Ice Sheet (GIS) areal extent and air temperature evolution since the Last Glacial Maximum (LGM). a The LGM, early deglaciation, late deglaciation, and postglacial intervals are determined based on observed changes in GIS modeled areal extent37 and reconstructed annual air temperature from proxy data of the GISP2 ice core record72. b–g Sedimentation rates are separated into the 6 depocenters delineated in Baffin Bay: (i) Central Baffin Bay (CBB), (ii) Baffin Bay Slope (BBS), (iii) Outer Shelf, (iv) Mid Shelf, (v) Inner Shelf, and (vi) Northern Baffin Bay (NBB), with the number of cores in each depocenter is given in brackets. Note that the three shelf depocenters outlined combine cores from the West Greenland Shelf as well as from the Baffin Island Shelf (except for the outer shelf from which no cores exist on the Baffin Island side). The light blue bars mark the sedimentation rate interval of 10–50 cm ka−1. The box plots show the median, minimum, maximum, lower (1st) and upper (3rd) interquartile range, and outliers of sedimentation rates.

Full size image
Fig. 3: Spatio-temporal distribution of sedimentation rates and ice-margin positions in Baffin Bay.
figure 3

The figure depicts the mean sedimentation rates for all individual sediment cores in Baffin Bay across the four-time intervals (a–d) considered here and the hypothesized contemporary ice-margin positions. The ~25 ka ice margins (in a) are according to Batchelor et al.3 (solid black line) and Dalton et al.7 (dotted black line). The ~18 ka BP (in a), ~12 ka (in b), and ~6 ka BP (in c) ice margins are according to Dalton et al.7 (dotted lines) and from Dalton et al.10 (solid lines) who adopted the previous Greenland ice-margin reconstruction by Dyke9. The red dashed lines highlight new ice margin positions for ~18 (in a) and ~12 ka BP (in b), respectively, as suggested in this study. The postulated Younger Dryas ice margin (orange lines in b) is based on observed grounding zone wedges on the mid-shelf interpreted as periods of ice stillstand or short-term re-advances20,28,35,43,46,48. The present-day ice margin (in d) is after Harrison et al.70.

Full size image

As only 8 of the 79 sediment cores from the shelves penetrated the deglacial sediments and reached underlying tills (marked as gray circles in Fig. 1c; see the “Methods” section and Supplementary Fig. 3), we note that the oldest ages retrieved from the cores only present minimum ages of the onset of sedimentation following ice retreat. Cores that did recover tills beneath dated sediments are of exceptional importance, as they permit temporal constraints of ice retreat and the onset of marine sedimentation at the respective site. This is especially useful for the West Greenland Shelf, where previous reconstructions of the timing and pattern of ice-sheet retreat are based largely on near-shore/terrestrial geochronological controls3,9,23.

From the LGM to ~15 ka BP, sedimentation is only recorded in cores from the CBB and BBS (Fig. 2 and Supplementary Fig. 2). During the LGM (~25 to ~18 ka BP), the median SRs (note that the median SRs is always based on the 1-kyr bins of the median SRs generated by the UNDATABLE age model) were generally low in the CBB, averaging slightly below 10 cm ka−1. In contrast, the BBS cores show higher rates of ~20 cm ka−1 (Fig. 2). While rates mainly remained low in the CBB (~10 cm ka−1) during the early deglaciation (~18 to ~12 ka BP), the BBS sites reveal faster accumulation between 30 and 50 cm ka−1 (Fig. 2). The following late deglaciation (~12 to ~6 ka BP) is marked by a sharp decrease in the median SRs to <5 and ~10 cm ka−1 in the CBB and BBS sites (Fig. 2), respectively, that remained rather constant through the following postglacial period (<6 ka BP).

The oldest outer shelf record from the WGS has a mean SR of ~200 cm ka−1 and dates back to the end (12–13 ka BP) of the early deglaciation (Fig. 2 and Supplementary Fig. 2). During the late deglaciation, median SRs in the outer shelf region drastically declined to ~10 cm ka−1 to remain at this level throughout the postglacial. The earliest documented mid- and inner-shelf sedimentation on the BIS and WGS is documented for the late deglaciation (Fig. 2 and Supplementary Fig. 2). On the mid-shelf, median SRs decreased from ~80 cm ka−1 to a rather constant ~15 cm ka−1 after ~8 ka BP. The inner-shelf median SRs revealed a similar decreasing trend through the late deglaciation, however, on a much higher level, decreasing from ~300 to ~80 cm ka−1 (Fig. 2 and Supplementary Fig. 2). Comparably high median SRs of ~50 cm ka−1 also persisted throughout the postglacial.

In the NBB, sedimentation is documented since the early deglaciation interval at ~14. 5 ka BP, much earlier than the shelf records from BIS and WGS (Fig. 2 and Supplementary Fig. 2). Median SRs of ~40–70 cm ka−1 lasted until about 10 ka BP when during the late deglaciation, the rates decreased to ~30–50 cm ka−1, eventually becoming rather stable around 35 cm ka−1 during the postglacial.

Discussion

Spatial sedimentation patterns in Baffin Bay have varied substantially (Figs. 2 and 3) since the LGM, when continental ice (LIS, IIS, GIS) expanded onto the shelves of Baffin Bay (Fig. 1). When the continental ice sheets advanced to or near the shelf breaks during the LGM3,7,9,10, the CBB and BBS probably were the only accommodation space available as documented by the available sediment cores; therefore, they were most likely the only active depocenters in Baffin Bay (Figs. 2 and 3a). However, sedimentation rates of <20 cm ka−1 suggest low terrigenous sediment input into CBB (Figs. 2 and 3). The low amounts of ice-rafted detritus observed in some of the deep basin records32,33 point to reduced iceberg rafting and imply sedimentation mainly from turbid meltwater plumes11,33, which would be consistent with a perennial ice-covered (sea ice and ice-shelf) Baffin Bay during the LGM8,34. At the same time, the BBS received more sediment, particularly in the southeastern Baffin Bay off West Greenland (up to 10-fold more, Fig. 3a). There, common glacigenic debris flows and turbidity currents provided the trough mouth fans with high amounts of sediment released from ice streams grounded at the mouths of cross-shelf troughs5,12, accompanied by sediments delivered by turbid meltwater plumes34,35.

The higher median SRs in the BBS records observed for the early deglaciation (Fig. 2) indicate faster-accumulating sediments during initial ice sheet recession (Fig. 3). These higher rates reflect the influence of ice-rafted detritus, glacigenic debris flows, and meltwater-induced deposition34,35. In contrast, SRs in the more distal CBB only show a slight increase between 13 and 16 ka BP (Fig. 2). During the LGM and the early deglaciation, the fastest deposition occurred at the Disko trough mouth fan in southern Baffin Bay (Fig. 3a, b), suggesting that the feeding glacier, Jakobshavn Isbrae, was most efficient in delivering sediments. This result aligns with present-day observations that indicate that this ice stream is by far the fastest-moving glacier in Greenland36.

The exclusivity of sedimentation in the CBB and BBS from the LGM to ~15 ka BP (Fig. 2) suggests prolonged ice sheet stability on the Baffin Bay shelves, most likely extending to near the shelf edges, at least within most of the cross-shelf troughs. This observation is consistent with empirical and model reconstructions of the GIS margin denoting close to the full-glacial maximum areal extent during this interval23,37. However, local changes in sedimentation style at the slope might indicate an initial small-scale retreat from the very shelf edge off West Greenland, possibly occurring as early as 17 ka BP34. Oldest marine sediments overlying a subglacial till were deposited on the upper slope at ~14.8 ka BP35, followed by outer shelf sedimentation around 13 ka BP35,38 (Fig. 2). In the NBB, the oldest dated sediments on till have ages of ~14.5 to 15.3 ka BP19,21, coeval with the collapse of the up to 500 m-thick ice shelf covering northern Baffin Bay8. Ice sheet destabilization and the retreat from the shelf edges thus occurred during the relatively warm Bølling–Allerød interstadial39 and was probably forced by rising northern latitude summer insolation40,41 and the strengthened advection of warmer waters into Baffin Bay triggered by the reinvigorated meridional heat transported by the Atlantic Meridional Overturning Circulation (AMOC)34,42,43. These data from West Greenland allow us to re-assess previously hypothesized ice margin position on the West Greenland shelf at 18 ka BP9. The presence of tills and the absence of early-deglaciation sediments on the shelf suggest that at 18 ka BP, the ice still extended to near the edge of the shelf (Fig. 3a).

At the end of the early deglaciation (13–11 ka BP), the main depocenters shifted very quickly from the BBS via the outer shelf to the inner shelf (Fig. 2) (corresponding to 200–300 km on the WGS), indicating an extremely rapid retreat of the ice sheets from the shelf edge towards the coast (see ref. 23). Especially, if also considering that during the Younger Dryas (YD)44 the ice sheet retreat was halted (or even reversed) as indicated by grounding zone wedges on the Baffin Island and West Greenland shelves20,28,43,45,46,47,48. During this YD mid-shelf stillstand, SRs were highest at the outer shelf (Figs. 2 and 3b), whereas inner shelf deposition is only documented after the YD cold spell at ~11.5 ka BP (Fig. 2). Although no direct evidence exists, the west Greenland mid-shelf grounding zone wedges and the here proposed later onset of inner-shelf sedimentation suggest that the margin of the GIS persisted on the mid-shelf until 12 ka BP (Fig. 3b, c). Thus, west Greenland inner shelf deglaciation primarily occurred only after the YD cooling in response to early-Holocene climate and ocean warming49,50.

The fast retreat of the LIS, IIS, and GIS between 13 and 11 ka BP was characterized by massive calving of icebergs leading to enhanced deposition of iceberg-rafted detritus in Baffin Bay, including the widely documented Baffin Bay Detrital Carbonate Layer 08,22,51. Within the cross-shelf troughs, the retreating ice sheets opened new accommodation spaces and, therefore, facilitated shelf deposition that eventually caused sediment starvation in the CBB and BBS depocenters, beginning with the late deglaciation (Fig. 2), resulting in low median SRs for CBB (<5 cm ka−1) and BBS (~10 cm ka−1).

All shelf sections show initially very high SRs that continuously decreased until ~6 ka BP, with the dominant depocenter (i.e., fastest deposition) on the inner shelves (Fig. 2). The high inner shelf SRs are facilitated by the huge accommodation space provided by the over-deepened inner shelf troughs46, located most proximal to the remaining ice masses. The same pattern of decreasing sedimentation rates during this period is also observed in the NBB (Fig. 2)19,21. As a consequence of the shelf deglaciation, sedimentation in Baffin Bay was dominated by hemipelagic sediments over the last 10–8kyr, with only minor contributions from ice-rafted detritus until the beginning of Neoglacial19,20,45.

During the subsequent postglacial period (<6 ka BP), sedimentation on the shelves leveled off in all areas (Fig. 2). At this time, the LIS and IIS had disintegrated into smaller ice caps over Baffin Island and the Canadian Arctic Archipelago, whereas the GIS attained its Holocene minimum areal extent10,23,37. Thus, comparably low median SRs after ~6 ka BP were most likely triggered by reduced sediment input from land to the Baffin Bay shelves, probably due to decreasing sub-ice sheet erosion rates in the postglacial period and from the trapping of meltwater sediment in the inner fjords52,53. Still, the highest SRs (>50 cm ka-1) occurred at the most ice-proximal sites on the inner shelves (Figs. 2 and 3d). During the last ~3 ka BP, slightly enhanced median SRs are observed on the inner- and mid-shelves (Fig. 2), probably related to enhanced sediment input caused by Neoglacial re-advance of local glaciers as indicated by renewed ice-rafted detritus sedimentation19,20,45,54.

Taking advantage of the much better database for the WGS (32 cores; Fig. 1c), we use this comprehensive overview of sedimentation rates to reconstruct erosion rates of the part of GIS draining into the Baffin Bay for the late deglaciation (late deglacial) and postglacial periods. Based on the overall mean of the binned median SR for all cores for these intervals from the four major cross-shelf troughs (Figs. 1c, 3), we calculated mean sediment accumulation rates for each trough (see the “Methods” section). Using the areal extents of the individual troughs and their paleo-ice stream drainages25, estimated sediment mass fluxes were converted to approximate subglacial erosion rates (Supplementary Table 4; see also refs. 47,48).

Based on the calculated sediment accumulation rates, our estimated late deglacial (12–6 ka BP) bedrock erosion rates for West Greenland range from 0.08 mm yr−1 (Uummannaq Trough) to 0.36 mm yr−1 (Melville Bay), whereas postglacial rates range from 0.04 to 0.06 mm yr−1, except for Disko Trough with 0.12 mm yr-1 (Supplementary Table 4). Although at the lower end, these results largely agree with estimates resulting from geophysical assessments of deglacial sediment volumes in Greenland shelf troughs (0.29–0.52 mm yr−1, these are high estimates due to smaller effective drainage areas considered55), from cosmogenic nuclide (10Be) analyses for centennial-scale (0.3–0.8 mm yr−1) and orbital-scale (0.1–0.3 mm yr−1) erosion rates56,57, and from a 10-year sediment load analysis in a southwest Greenland river (0.5 mm yr−1)58. The consistency of these estimates based on various methods underlines the high erosion capacity of the GIS, as average subglacial erosion rates of 0.01–0.1 mm yr-1 have been given for polar regions59.

Mean erosion rates derived from all four West Greenland shelf troughs decrease from 0.17 mm yr−1 for the late deglacial to 0.08 mm yr−1 for the postglacial, reflecting a roughly 55% reduction in subglacial erosion of West Greenland, a trend also observed off Petermann glacier entering Nares Strait55. This trend towards decreasing erosion rates probably reflects the transition from active deglacial (ice) retreat to relatively stable postglacial conditions.

Methods

Distribution of 14C ages in cores used in this study

The primary age-control points used in the computation of the SRs for the 79 cores from Baffin Bay, shown in Fig. 2 and Supplementary Fig. 2, range from a minimum of 2 to a maximum of 26 14C ages per core (Supplementary Fig. 1; our census date is June 30, 2023). Generally, the number and distribution of age-control points in the individual sediment records do have a considerable effect on the information content of the derived SRs, especially on the temporal resolution of the SR record. We decided to use two ages as a lower boundary, as these two ages present the minimum to calculate a reliable SR within the time interval constrained by these two ages, plus an estimate of the SR for the youngest core section assuming a recent age for the core top (see below). Of course, these only give mean values for two sections of the core, potentially masking temporal SR variabilities. Nevertheless, these mean values provide valuable information to assess at least average local SRs. In all the selected depocenters, cores with only a few as well as with a considerable number of ages co-occur (see Supplementary Fig. 1), always revealing a consistent pattern of regional SR development, which even counts for the deep basin cores having the lowest overall numbers of 14C ages (see also Fig. 3). Thus, in sum, higher number of ages per core provide a higher resolved SR record, but even records with only two ages provide valuable information on mean SRs. To limit the related uncertainty, in most cases, we refrain from interpreting SR data from single records but focus on the overall trend in sedimentation patterns, and we advise readers to note this. Despite the limitations mentioned above, these data represent currently our best tool for addressing the glacial to Holocene development of the Baffin Bay sedimentary systems.

Age calibration and sedimentation rates

All 14C ages from each core were re-calibrated individually within the age-depth modeling process of UNDATABLE30 (settings: nsim = 105, bootpc = 30, xfactor = 0.1). For the age calibration, region-specific marine reservoir age correction (ΔR) values60 were applied to Holocene and deglacial 14C ages (<15,000 years) and a higher ΔR value (250 ± 100 years)22 for 14C ages > 15,000 years from CBB and BBS (see Supplementary Table 2). As the 14C ages obtained from bulk sediment organic matter and Portlandia arctica, due to the “Portlandia Effect,” often yield considerably older ages compared to other marine fossils (see refs. 14 and54,55,56,61,62,63,64), they were excluded from the age model constructions. In addition, the top of each core is assumed to be of recent age (i.e., 0 ka BP ± 50 years, with recent age defined as the year 1950 AD). We note that the assumption of a modern age of the core top could potentially lead to underestimation of the SR, especially in cores interpreted to lack younger Holocene sediments12,21,45,65. Finally, the topmost (youngest) age for each core was excluded from the bootpc (bootstrapping process) to force the algorithm to include this age in the construction of the age-depth relationship.

Note that the here presented SRs for the Baffin Bay cores were not corrected for any mass-transport deposits as these are usually only documented for those few cores analyzed by computer tomography18,19 (but see also ref. 12), while for nearby cores with a similar stratigraphy such mass-transport deposits are not documented21,54,66. Thus, to treat all records the same way, even the few documented cases12,19,20 have not been considered. As datings reveal that the sediment in such mass-transport deposits is of rather similar age as in the overlying deposits18, these mass-transport deposits are also considered to represent time-related sediment input associated with the retreat of ice sheets.

Besides computing the age-depth relationship, UNDATABLE calculates each sediment core’s median SRs (in cm ka−1). We binned these rates into one-kiloyear time slices and wrote the mean values to the central position of each slice. The binned SRs are presented in logarithmic scale (log10) in Fig. 2 to emphasize variability in areas (cores) with largely varying values. Additionally, step-plots of the binned SRs for all cores are displayed in Supplementary Fig. 2.

Cores with basal till

Sediment cores that have been interpreted to contain basal till deposits, indicative of paleo-ice extent on Baffin Bay, are indicated (gray-filled circles) in Fig. 1c, including gravity core GeoB22357-3 from the Clyde Trough. The thick diamicton bed (LF2) identified at the bottom of core GeoB22357-367 was previously interpreted as a glacigenic debris flows45. In light of the newly acquired computed tomographic data (see ref. 19) revealing a rather massive and over-compacted basal diamicton and the absence of any internal structures consistent with glacigenic debris flows (Supplementary Fig. 3; see ref. 67), this basal unit is re-interpreted here as ice-contact sediments. This presence of subglacial till, together with the presented 14C ages, further substantiates evidence that the mid-shelf area of the Clyde shelf trough was glaciated (stillstand or LIS re-advance) by the Clyde Ice Stream during the YD stadial, as postulated by Couette et al.45.

Estimating subglacial erosion rates

Based on the available late deglacial and postglacial sedimentation rates, assuming an average dry bulk density of 0.8 g cm−3 for Greenland fjord and shelf sediments61 and considering the area covered by the four WGS troughs (Melville, Uummannaq, Upernavik, and Disko)25, we calculated the total mass accumulation for these four troughs for the two time slices considered here (Supplementary Table 4). As biogenic contributions (total organic carbon and biogenic carbonate) to the sediment accumulation are usually <5%18,68,69, these have been neglected. In a second step, the calculated mass accumulations were applied to the respective drainage areas25 to calculate average denudation/erosion rates, thereby considering a density for the predominantly crystalline rocks of 2.7 g cm−3 (refs. 47,48). Mean erosion rates for West Greenland have been calculated by averaging the erosion rates of the individual drainage areas of the respective shelf troughs, thereby considering the size of the respective drainage basins.

Of course, there are several limitations to this approach: (1) the limited SR-data coverage for the various troughs on the WGS, as, for example, considering only proximal inner shelf records (e.g., Melville Bay) will lead to an overestimation of the sediment mass accumulation in a trough (and vice versa), (2) the late deglacial rates have to be considered as minimum estimates as most sediment cores considered here did not penetrate the entire deglacial sediment column, thus missing the fast accumulating ice proximal sediments deposited directly after the local deglaciation, and (3) neither sediment bypassing of the troughs, e.g., via iceberg rafting, nor input from the Canadian side is considered. Nevertheless, the consistency of the data from the various investigated cross-shelf troughs, as well as the alignment with estimates based on other methods, underlines the applicability of this approach (see ref. 61).