Climatic control on the retreat of the Laurentide Ice Sheet margin in easternmost Québec–Labrador (Canada) revealed by cosmogenic nuclide exposure dating

The Laurentide Ice Sheet (LIS) was the largest ice sheet in the Northern Hemisphere during the last glacial cycle. The effects of its demise on global climate and sea‐level changes during the subsequent deglaciation are unequivocal. Understanding the interplay between ice sheets and long‐term or short‐term (e.g. abrupt) climatic events is therefore crucial for predicting future rates of ice sheet melting and their potential contribution to sea‐level changes. Here, we present 37 new 10Be surface exposure ages from easternmost Québec–Labrador that allow us to identify close ties between regional deglaciation history and climate. These results reveal that the LIS was disconnected from the Newfoundland Ice Cap by ~14.1 ka. Samples collected from moraine boulders indicate that this event was followed by five major stillstands and/or readvance stages of the LIS margin. Integrating our new moraine ages with those of earlier studies allows us to depict a temporal framework including events at ~12.9, ~11.5, ~10.4, ~9.3 and ~8.4–8.2 ka. These moraine ages highlight a strong sensitivity of the LIS to temperature changes in the Northern Hemisphere, as the documented continental ice margin stabilizations coincide with abrupt cooling events recorded in Greenland ice cores. These observations support the idea of a negative feedback mechanism induced by meltwater forcings into the North Atlantic Ocean which, in turn, provoked repeated cold reversals during the Younger Dryas and early Holocene.


Introduction
The ice sheets that once covered the Northern Hemisphere played an undeniable role in modulating global changes since the Last Glacial Maximum (LGM), as their growth and decay were closely interconnected with atmospheric/oceanic circulation and sea-level changes (Alley et al., 1997;Clark et al., 2000;Carlson et al., 2008;Denton et al., 2010;Briner et al., 2020;Lowell et al., 2021).Warming conditions during the deglaciation that followed the LGM provoked enhanced freshwater inputs into the North Atlantic Ocean, slowing the Atlantic Meridional Overturning Circulation (AMOC) and ultimately reducing air temperature over the Northern Hemisphere (Barber et al., 1999;Fisher et al., 2002;McManus et al., 2004;Rohling & Pälike, 2005;Carlson & Clark, 2012;Jennings et al., 2015;Süfke et al., 2022).However, the interplay between such feedbacks and ice mass evolution is still poorly constrained, as Holocene climate evolution and its sensitivity to known forcings remain in some cases elusive (Axford et al., 2009;Jennings et al., 2015;He & Clark, 2022).Documenting the response of former ice sheets -such as the Laurentide Ice Sheet (LIS)to external forcing, and the timing of this response, is critical for assessing the contribution of melting ice masses to past and future sea-level rise.
The study of stillstands and readvances of the LIS margin during its overall retreat provide key information on late Wisconsinan and early Holocene climate fluctuations, which is essential for (i) improving our knowledge on the long-term glacial and deglaciation history of former ice sheets, (ii) determining their sensitivity to climate change and (iii) identifying trigger mechanisms of ice margin behaviour (Lesnek & Briner, 2018;Briner et al., 2020;Young et al., 2020).Identification of causes, such as drainage of large glacial lakes, contributes to assessing their potential role in, and response to, millennial-scale abrupt climatic events (e.g.Fisher et al., 2002;Lajeunesse & St-Onge, 2008;Yu et al., 2010;Dubé-Loubert et al., 2018;Leydet et al., 2018;Brouard et al., 2021;Süfke et al., 2022).Dating and reconstructing the evolution of former ice sheet margins is therefore crucial for understanding the long-term interconnections between ice sheets and the global climate system and identifying the interfering feedback mechanisms that influence deglaciation (Briner et al., 2020;Lowell et al., 2021).
The development of cosmogenic exposure dating has contributed substantially to improving reconstructions of the deglacial history in North America by allowing direct dating of landforms deposited during stagnation or retreat stages of ice margins (Balco, 2020).Over the last few decades, an ever growing number of studies have focused on identifying and dating major stillstands and readvances of the LIS margin during its retreat following the LGM (i.e.Marsella et al., 2000;Clark et al., 2003;Briner et al., 2005Briner et al., , 2007Briner et al., , 2009;;Balco et al., 2009;Young et al., 2012Young et al., , 2020aYoung et al., , 2021;;Bromley et al., 2015;Davis et al., 2015;Ullman et al., 2016;Margreth et al., 2017;Crump et al., 2020;Lowell et al., 2021).However, as these studies focused mostly on specific sectors of the LIS, large geographical gaps in available absolute ages from glacial landforms remain, especially at its eastern fringe.Despite the relatively large number of radiocarbon ages available in southern Labrador and adjacent easternmost Québec that provide minimum-limiting ages for deglaciation (i.e.King, 1985;Dyke et al., 2003;Dalton et al., 2020), the chronology of most of the major moraine systems marking stabilizations/readvances of the retreating LIS margin in the region is still poorly constrained.
In this paper, we report on the deglaciation history of the eastern sector of the LIS by using terrestrial cosmogenic 10 Be exposure dating of boulders from major moraine systems.This chronological dataset coupled with previously published and new radiocarbon ages allow us to (1) constrain the position of the ice margin during its retreat, and (2) assess retreat rates of the ice margin in order to identify time intervals of important changes.These results provide a revised deglacial chronological framework across five former ice-marginal stillstand/ readvance positions located along a 500-km-long transect from the coast to the Labrador hinterland.These results allow us to discuss the potential causes for deposition of moraines in easternmost Québec-Labrador and their implications for regional climate.

Regional setting
Easternmost Québec-Labrador (Canada) hosts a series of moraines marking the overall retreat of the LIS during the Lateglacial and early Holocene (Fig. 1).The region lies within the Grenville geological province of the Canadian Shield and is underlain mainly by Proterozoic quartzofeldspathic gneisses, granites and anorthosites (Greene, 1974;Hynes & Rivers, 2010).Regional physiography consists of a hilly 'peneplain' ranging from 300 to 500 m in elevation that is deeply incised by structural valleys.The study area also includes the Lake Melville depression, an estuary that stretches 200 km inland from the Atlantic coast, and the Mealy Mountain Massif, a prominent plateau-topped highland reaching over 1000 m.Except for the area located to the northwest of Lake Melville, the study area is characterized by a generally thin ice-contact cover of deposits and valleys partially filled with glaciomarine to glaciofluvial deposits (Fulton & Hodgson, 1979;King, 1985).Five extensive morainic systems (>100 km long) previously identified in easternmost Québec-Labrador represent major successive positions of the retreating LIS margin following the late Wisconsinan glacial cycle (i.e. from SE to NW: Brador, Belles Amours, Paradise, Little Drunken and Sebaskachu moraine systems: Fulton & Hodgson, 1979;King, 1985;Occhietti et al., 2011).Although these moraines undeniably represent mappable, regional-scale stillstands and/or potential readvances of the LIS margin, their age remains poorly constrained and their correlations with other sectors of the LIS to the north (e.g.Baffin Island) or to the west (e.g.St. Lawrence estuary) are still debated (e.g.King, 1985;Grant, 1992;Occhietti et al., 2011).
Early workers proposed that easternmost Québec-Labrador was not covered by the LIS during the late Wisconsinan glaciation (Coleman, 1921).The idea of a restricted LIS in the region was later invoked: some authors positioned the LGM margin of the LIS at the Brador Moraine near the present coastline (Ives, 1978;Vilks & Mudie, 1978) and others suggested that it was located even more inland at the Paradise Moraine based on the scarcity of geomorphological evidence further east (Fig. 1; Fulton & Hodgson, 1979).In more recent investigations, severals authors revised this restricted position by demonstrating that the LIS completely covered easternmost Québec-Labrador at the LGM, reaching as far as the shelf edge in the Labrador Sea (Josenhans et al., 1986;Piper, 1991;Grant, 1992;Roger et al., 2013).After its separation from the Newfoundland Ice Cap, the LIS margin stabilized at a position located at the modern coast of eastern Québec-Labrador, depositing the Brador and the Belles Amours moraine systems (Fig. 1; Grant, 1992).The age of the Brador and the Belles Amours moraine systems were estimated at 12.5 14 C ka BP (~15 cal ka BP) and 11.0 14 C ka BP (~13 cal ka BP), respectively (King, 1985).Grant (1992) proposed that the LIS margin stabilized when it grounded at the marine limit to build the Brador Moraine at 12.6 14 C ka BP (~15 cal ka BP) and deposited the Belles Amours Moraine shortly after (<200 years) in the frame of a regional readvance.The ice margin then retreated westward and deposited successively the Paradise and Little Drunken moraine systems during two stillstands and/or readvances of the LIS margin (King, 1985).The Paradise Moraine was proposed to correspond to the Younger Dryas (YD) event, therefore correlating to the St-Narcisse Moraine of southern Québec (Dyke & Prest, 1987;King, 1985;Grant, 1992;Occhietti, 2007), while the Little Drunken Moraine was speculated to correspond to the Québec North Shore Moraine deposited at ~10.8 cal ka BP (Dubois & Dionne, 1985;Dietrich et al., 2019).However, the abandonment of the Paradise Moraine and Québec North Shore Moraine have been recently dated at 10.3 ± 0.6 and 9.2 ± 0.5 ka, respectively, using cosmogenic exposure dating (Ullman et al., 2016).After this stage of deglaciation, the LIS margin stabilized again at the west end of Lake Melville and deposited the Sebaskachu Moraine.The Little Drunken Moraine has been tentatively correlated to the Sebaskachu Moraine by several authors (i.e.Blake, 1956;Fulton & Hodgson, 1979;Occhietti et al., 2011), although no formal geomorphological connection has been observed between the two systems.The ice margin is estimated to have retreated from the Sebaskachu Moraine position at 8.0 14 C ka BP (~8.8 cal ka BP) and reached central Labrador by 7.0 14 C ka BP (~7.8 cal ka BP; King, 1985).Similarly, cosmogenic exposure dating on erratic boulders yielded ages of 8.6 ± 0.6 ka west of Lake Melville and 7.5 ± 0.4 ka for central Labrador (Ullman et al., 2016).Ice retreat then took place rapidly with the final ablation of the LIS at ~5.5 14 C ka BP (~6.0 cal ka BP) over the central Québec-Labrador Peninsula (Richard et al., 1982;Clark et al., 2000;Jansson, 2003;Occhietti et al., 2011;Dubé-Loubert et al., 2018), leaving series of small recessional moraines across the landscape.

Field mapping and sampling
Positions of the former margins (i.e.moraines) of the LIS were mapped to spatially constrain its successive extents during deglaciation of the easternmost Québec-Labrador sector (Fig. 1).Our mapping builds upon previous investigations on moraines across the study area that was, however, sporadic and not sufficiently constrained in some sectors (Dubois & Dionne, 1985;Grant, 1992;Klassen et al., 1992).Interpretation of Landsat satellite imagery allowed us to refine and extend previous maps of moraine deposits.All mapped moraine systems were then visited in the field and sampled.Additionally, the chronology ( 10 Be and 14 C) was used to interpolate ice margin positions in areas lacking wellpreserved moraine deposits.
Surface exposure ages on moraine boulders, erratics and bedrock outcrops were obtained using cosmogenic beryllium-10 dating (hereafter referred to as 10 Be; Fig. 2).About 1 kg of rock was collected from the upper 2 cm of boulder surfaces using a handheld rocksaw, a hammer and a chisel.The flat top of stable boulders embedded in the moraine matrix was targeted to minimize overturning possibility, post-deposition exhumation and extreme weathering.Where available, quartz veins or quartz-rich material were extracted from the boulders.Samples were precisely documented in the field, including site description, coordinates and elevation obtained from a handheld GPS.

Sample preparation and analysis
Samples were processed at the cosmogenic nuclide laboratory at the University of Strasbourg following a wellestablished protocol modified from Kohl and Nishiizumi (1992) and Bierman et al. (2002).All samples were crushed and sieved to isolate the 250-1000 µm fraction, which was subsequently treated by leaching in an HCl solution to eliminate oxides and organic material.Up to nine ultrasonic leaching cycles with diluted HF and HNO 3 (~1% each) solution were needed to purify the quartz.A commercial 9 Be carrier (~0.25-0.5 mg) was added to the samples before dissolving the pure quartz in concentrated HF (48%) and HNO 3 (68%).Beryllium was isolated through anion and cation exchange columns and then precipitated as hydroxide.Beryllium hydroxide was dried and calcinated to BeO at 750°C. 10 Be/ 9 Be ratios were measured at ASTER, the French Accelerator Mass Spectrometry (AMS) at CEREGE (Aix-en-Provence).The measurements of samples were normalized against an in-house CEREGE standard, STD11, with an assumed 10 Be/ 9 Be ratio of (1.191 ± 0.013) × 10 −11 (Braucher et al., 2015) and a 10 Be half-life of 1.387 ± 0.012 Ma (Chmeleff et al., 2010;Korschinek et al., 2010).A procedural blank was processed with each batch of samples to evaluate 10 Be contamination during laboratory procedures.Blank 10 Be/ 9 Be ratios range from 5.21 ± 0.42 × 10 −15 to 6.12 ± 0.45 × 10 −15 ; average blank 10 Be/ 9 Be ratios were 5.74 ± 0.59 × 10 −15 (n = 8) (Table 1).10 Be age calculations 10 Be ages were calculated using the online exposure age calculator formerly known as the CRONUS-Earth online exposure age calculator version 3.0 (Balco et al., 2008; https://hess.ess.washington.edu/).Calibration was done using the northeastern North America (NENA) 10 Be production rate of 4.04 ± 0.27 atoms g −1 a −1 (Balco et al., 2009) and the nuclide-and time-dependent LSDn scaling scheme (Lifton et al., 2014).The NENA production rate is statistically identical to other high-latitude production-rate calibration datasets such as the Baffin Bay (Young et al., 2013) and the Rannoch Moor (Putnam et al., 2019) production rates.For comparison, the global production rate (Borchers et al., 2016) would result in ages that are ~5% younger.By comparison, using the St or Lm scaling schemes (Lal, 1991;Stone, 2000) would result in exposure ages that are ~5% older than the LSDn-derived 10 Be ages.The effects of postglacial erosion on 10 Be ages were neglected as sampled bedrock and boulder surfaces commonly displayed glacial striations, suggesting little surface erosion since deglaciation.Topographic shielding was not taken into account for boulders and sampled surfaces as it was estimated to be negligible (<1%).Glacio-isostatic uplift corrections were not accounted for in the final results (Supporting Information Table S1), as its effects on cosmogenic nuclide production are not well constrained (Balco et al., 2009;Young et al., 2020b).Snow cover corrections were also omitted in the calculation of the ages as we sampled boulders located on top of moraine crests, which are often wind-swept and thus probably prevented any significant snow accumulation (Supporting Information Table S2).Individual 10 Be exposure ages are presented with 1σ uncertainty (Table 1).Stated uncertainties are analytical only.Outliers were identified using the CRONUS-Earth online calculator and defined as 10 Be ages that are >2σ older or younger from the mean of the remaining 10 Be ages.Probability distribution function (PDF) plots were generated using the iceTEA (Tools for Exposure Ages from ice margins; http://ice-tea.org/en/)'Import and plot ages' tool (Jones et al., 2019), in which the moraine ages were defined as the error-weighted mean of the sample population and the error-weighted internal uncertainty, excluding outliers.
We include the production rate uncertainty in quadrature when comparing and discussing landform ages with independent climate records and radiocarbon ages.Additionally, 10 Be ages from Ullman et al. (2016) were recalculated using the same parameters as above to allow direct comparison with our new 10 Be ages (Supporting Information Table S3).

Radiocarbon dating
Accelerator mass spectrometry (AMS) radiocarbon dating was carried out on marine shells collected in 2019.The four obtained AMS 14 C ages were calibrated within the age-depth modelling process and converted to calendar The 10 Be ages were calculated with the 'LSDn' scaling scheme (Lifton et al., 2014), using the CRONUS-Earth online calculator version 3.0 (Balco et al., 2008; http://hess.ess.washington.edu/)and the production rate calculated with the NENA calibration data set of Balco et al. (2009).The 10 Be AMS standard applied was the ASTER in-house STD-11 with a 10 Be/ 9 Be ratio of 1.19 × 10 −11 (Braucher et al., 2015) and a 10 Be half-life of 1.387 ± 0.012 Ma (Chmeleff et al., 2010;Korschinek et al., 2010).A constant thickness of 2 cm and a rock density of 2.65 g cm −3 was applied for all samples.No erosion and no topographic shielding were accounted for in our calculations.Snow cover corrections were not applied (Supporting Information).Asterisks (*) mark the samples that are considered as outliers and were not included in the mean age calculations (in bold).Numbers in parentheses includes the production rate uncertainty.
years using the online software Calib 8.2 (Stuiver & Reimer, 1993; http://calib.org/)with the Marine20 radiocarbon age calibration curve (Heaton et al., 2020).A local reservoir correction (ΔR) of −2 ± 69 was used to account for the regional offset of the world ocean 14 C age (McNeely et al., 2006).These four ages complement a dataset of 41 previously published 14 C ages that have also been recalibrated.Marine samples were calibrated using the aforementioned parameters, whereas terrestrial samples were calibrated using the IntCal20 radiocarbon age calibration curve (Reimer et al., 2020).All individual radiocarbon ages are presented as the mean of the calibrated age range with 1σ uncertainty (Table 2).

Retreat rates
The moraine map was used to reconstruct retreat rates of the LIS in easternmost Québec-Labrador.When ice-contact deposits were absent (i.e.west of Lake Melville), retreat isochrons of former ice margin positions (i.e.7.9, 7.6 and 7.5 ka) were tentatively drawn perpendicular to the ice-flow direction where minimum limiting ages are available.Ice retreat rates were then calculated along time-distance transects perpendicular to the retreating ice margin.The transects were hand-drawn radially from the ~7.5-ka isochron of the LIS to the coast, where they are at intervals of ~50 km.This approach allowed each moraine system to be crossed at least three times, since no transect intersects all five mapped ice margin positions.The mean linear retreat rate was then calculated between every ice margin position using the ages for each moraine system presented in this paper for estimating the land-based ice margin retreat.These retreat rates represent minimum values as they do not take into account possible readvances of the LIS margin and their respective durations.The AMS 14 C ages were calibrated within the age-depth modelling process, using the online software Calib 8.2 with the Marine20 (Heaton et al., 2020) and IntCal20 (Reimer et al., 2020) radiocarbon age calibration curves.A local reservoir correction (ΔR) of −2 ± 69 was used to account for the regional offset of the world ocean 14 C age, as determined by McNeely et al. (2006).

Results
10 Be ages and moraine systems Moraine systems along the SE-NW transect of eastern Québec-Labrador are here briefly described.The resulting 10 Be ages from the sampled boulders and surfaces are provided for each of these moraine systems.Three erratic boulders located above marine limit were sampled in the easternmost sector of the study area, where morainic landforms are sparse.These erratic boulders constrain a series of closely spaced recessional moraines that have not yet been linked to any major moraine system.Since the moraine ridges were mostly located below marine limit, three boulders -located on both sides-were selected to assess the timing of their formation; two 'coastal' boulders (LBD19-30 and LBD19-31) were located within 2 km of the coast and one 'inland' boulder (LBD19-28) was sampled ~40 km from the coast.The coastal erratic boulders provided 10 Be ages of 14.1 ± 0.6 and 12.5 ± 0.5 ka (Table 1), whereas the erratic boulder collected 40 km inland provided a 10 Be age of 12.0 ± 0.5 ka (Table 1).
The Brador Moraine consists of small-amplitude moraine ridges coinciding with the marine limit (~140 m) (Fig. 3A) and outwash deposits occupying lower areas in narrow valleys.Some moraine ridges show traces of local glaciotectonic deformation, indicating readvance of the ice margin before stabilization.Five boulders were sampled from three different segments of the Brador Moraine over a distance of ~75 km.These boulders yielded consistent 10 Be ages ranging between 12.6 ± 0.7 and 13.4 ± 0.5 ka and a weighted average of 12.9 ± 0.2 ka (Table 1; Fig. 4).
The Belles Amours Moraine consists of series of small sinuous ridges, each a few metres high and less than 20 m wide (Fig. 3B).This moraine system has a SW-NE orientation and is cross-cutting the Brador Moraine west of Blanc Sablon (Fig. 1).Five boulders were sampled from two of the main crests of the Belles Amours Moraine.They yielded 10 Be ages ranging from 11.2 ± 0.5 to 12.0 ± 0.4 ka (Table 1), resulting in a weighted average of 11.5 ± 0.3 ka (Fig. 4).Additionally, a bedrock surface sampled 4 km behind the Belles Amours Moraine and at 75 m above sea level (m.a.s.l.) -located below marine limit (~130 m.a.s.l.; Grant, 1992)-yielded an age of 10.8 ± 0.4 ka (LBD19-46; Table 1), indicating the emergence from marine waters and hence representing a minimumlimiting age constraint.
The Paradise Moraine consists of a 20-km-wide complex of glaciofluvial deposits pitted with kettles, as well as fields of hummocky and ribbed moraines (Fig. 3C).East of the Mealy Mountains, the moraine overprints glacial lineations at an odd angle, testifying to an ice flow reorganization that suggests an ice margin readvance.Five samples were collected on the Paradise Moraine and yielded a large range of ages from 11.4 ± 0.4 to 37.1 ± 1.2 ka (Table 1).Two samples yielded (too-) old 10 Be ages of 20.8 ± 0.7 and 37.1 ± 1.2 ka, while the three remaining samples have exposure ages ranging from 11.4 ± 0.4 to 12.7 ± 0.5 ka and a resulting weighted average of 12.1 ± 0.6 ka (Fig. 4).
The Little Drunken Moraine is located ~250 km inland from the coast.It has a lobate geometry and is defined by series of 5-to 40-m-high till ridges (Fig. 3D) associated with extensive outwash-plain deposits in proglacial valleys.Five samples were collected from the main crest of the Little Drunken Moraine: three samples yielded 10 Be ages ranging from 9.8 ± 0.4 to 11.3 ± 0.4 ka, with two ages at 13.5 ± 0.4 and 14.7 ± 0.5 ka considered as outliers (Table 1).The weighted average of the three remaining ages is 10.6 ± 0.7 ka (Fig. 4).
Two bedrock surfaces were sampled from the top of Mokami Hill (488 m.a.s.l.; Fig. 1) in order to assess ice thinning at the head of Lake Melville (LBD19-74 and LBD19-75).Although located <50 m apart, these sampled yielded disparate 10 Be ages of 10.0 ± 0.4 and 13.8 ± 0.5 ka (Table 1).Dating surfaces from the mountain top west of Lake Melville therefore remains inconclusive.The large span of ages (>3 ka) suggests that (i) at least one sample is unreliable, or (ii) they have endured differential erosion.Although it is likely that the older age is unreliable based on its location and nearby ages (i.e. the Sebaskachu Moraine), it is impossible to assess the actual exposure age of the mountain surface at this point and more samples should be collected to document the thinning of the LIS in the region during the early Holocene.
The Sebaskachu Moraine, located west of Lake Melville, consists of small linear moraines with well-defined narrow crests ~10 m high on the plateau (Fig. 3E); in valleys it consists of large (>50 m) subaqueous ice-contact depositional systems.Five boulders sampled from the Sebaskachu Moraine on the plateau yielded 10 Be ages ranging between 7.9 ± 0.3 and 9.1 ± 0.3 ka (Table 1).The weighted average of these five samples is 8.4 ± 0.4 ka (Fig. 4).Additionally, a boulder was sampled from the top of a valley moraine near Happy Valley-Goose Bay at 65 m.a.s.l. and yielded an age of 7.8 ± 0.4 ka (LBD19-84; Table 1).This site -located below marine limit (~135 m.a.s.l.; Fitzhugh, 1973)-provides the timing of the emergence rather than ice retreat.The regional relative sealevel curve from Fitzhugh (1973) suggests that the sample was shielded by the water column for up to 600 years.It therefore only provides a minimum estimate of the deglaciation, although adding these 600 years would closely match the weighted age calculated for the Sebaskachu Moraine.
Five samples were also collected 40 km west of Happy Valley-Goose Bay on minor moraine ridges (<3 m) in the Peter's River valley, above marine and lake limits.The samples yielded three 10 Be ages ranging from 7.5 ± 0.3 to 9.0 ± 0.4 ka, with two outliers at 6.8 ± 0.4 and 12.4 ± 0.5 ka (Table 1).The three remaining samples are, however, slightly disparate with a 10 Be weighted average age of 8.1 ± 0.5 ka (Fig. 4).

Radiocarbon ages
Three shell samples were collected in the southern extension of the Sebaskachu moraine, near Happy Valley-Goose Bay.These shells were embedded within the till composing the subaqueous ice-contact system, therefore possibly pre-dating readvance of the ice margin.These samples yielded consistent ages of 8.5 ± 0.1 cal ka BP (LBD19-62), 8.5 ± 0.1 cal ka BP (LBD19-61) and 8.4 ± 0.1 cal ka BP (LBD19-81).A shell collected in glaciomarine muds of the Churchill River bank 50 km west of the Sebaskachu Moraine yielded an age of 8.2 ± 0.2 cal ka BP (LBD19-79).Together, these results provide maximum and minimum limiting ages for the deposition of the Sebaskachu Moraine, which are in agreement with (i) former 14 C dating (Fig. 5; Table 2), and (ii) the cosmogenic exposure age of the Sebaskachu Moraine (8.4 ± 0.4 ka).

Discussion
Ice margin stabilizations in easternmost Québec-Labrador and regional correlations The new cosmogenic exposure ages allow us to define the timing of deposition of the moraine systems of easternmost Québec-Labrador that record major readvance and/or stabilization stages of the LIS margin during its overall northwestward retreat.These results are consistent and in chronological order in most cases, except for the Paradise Moraine that appears older than the Belles Amours Moraine deposited ~100-120 km to the east.The timing and correlations of the moraine systems are discussed below with respect to previously published data.
The age of the Brador Moraine system of 12.9 ± 0.7 ka (including the production rate uncertainty propagated through in quadrature) is supported by a robust 10 Be chronology and corresponds to the beginning of the YD chronozone.This age is further supported by a series of minimum-limiting radiocarbon ages ranging from 11.4 ± 0.1 to 12.2 ± 0.2 cal ka BP and maximum-limiting ages ranging from 12.7 ± 0.2 to 13.0 ± 0.2 cal ka BP (Fig. 5; Table 2).Additionally, the inland erratic boulder dated at 12.0 ± 0.5 ka, supported by a nearby age of 12.4 ± 0.1 cal ka BP (Engstrom & Hansen, 1985;Fig. 5; Table 2), suggests that the recessional moraine system located between the two erratic boulder sites ('coastal' and 'inland') may correspond to or slightly succeed the Brador Moraine.There is no indication, however, for its extension north of this sector, but it is highly probable that it is located beyond the coast of southeastern Labrador as foraminifera samples offshore yielded ages ranging from 11.1 ± 0.2 to 12.3 ± 0.3 cal ka BP (Table 2).Our results therefore suggest that the Brador Moraine is time-equivalent to the St-Narcisse Moraine in southern Québec (Occhietti, 2007) and possibly to one of the icecontact grounding-zone wedge systems observed offshore the Québec North Shore region (Lajeunesse et al., 2019).It would also be contemporaneous with the Saglek Moraine located in the Torngats Mountains of northern Labrador (Clark et al., 2003).Consequently, the coastal erratic boulders suggest deglaciation of the coast of easternmost Québec-Labrador between 14.1 ± 0.6 and 12.5 ± 0.5 ka.Opening of the Belle Isle Strait  1).The black and dashed lines represent the weighted average and the 1σ interval.[Color figure can be viewed at wileyonlinelibrary.com] and isolation of the Newfoundland Ice Cap from the LIS therefore occurred prior to that timing, presumably during the Bølling-Allerød warm period (14.7-12.9ka BP).This interpretation supports what has previously been proposed by Shaw et al. (2006) who argued that disconnection between the two ice masses occurred between 14.8 and 14.0 cal ka BP.In this scheme, the time frame proposed by Grant (1992) appears to be erroneous, as they argued that the opening of the Strait of Belle Isle and separation of the two ice masses occurred prior to 15 cal ka BP.
The age of 11.5 ± 0.7 ka for the deposition of the Belles Amours moraine ridges indicates a stabilization of the LIS margin at the beginning of the Holocene or near the end of the YD.This timing for the deposition of the Belles Amours moraine system contradicts the interpretation of Grant (1992) who argued that their deposition may have taken place only a few centuries after deposition of the Brador Moraine on the basis of only a minor relative sea-level fall between the two events.However, a relatively stagnant ice margin during the YD and a significant readvance -as evidenced by the crosscutting relationship of the two moraines (Fig. 1)-may have significantly limited the rate of glacio-isostatic rebound in the region, similar to observations in Greenland during the Neoglacial ice expansion (Long et al., 2009).Although the mapped extent of this moraine is relatively limited (~100 km), the 12.0 ± 0.5 ka age of the inland erratic suggests that the ice margin was probably located west of this site at the time of deposition of the Belles Amours Moraine.Its northern extent can, however, only be tentatively mapped.Additionally, a basal lacustrine sample dated at 11.5 ± 0.2 cal ka BP (Lamb, 1980;Fig. 5; Table 2)-similar to the mean cosmogenic exposure age of the Belles Amours Moraine samples-may provide an indication for its approximate location near that site.The timing of deposition of the Belles Amours Moraine suggests a correlation with the Mars-Batiscan Moraine in southern Québec (Occhietti, 2007), although the physical connection between the two systems remains ambiguous.It is possible that the Belles Amours Moraine corresponds offshore to one of the youngest ice-contact grounding-zone wedges (Lajeunesse et al., 2019) and, onshore, to the Baie Trinité Moraine on the Québec North Shore (Fig. 1B; Occhietti et al., 2011).However, further dating would be needed to confirm the correlation as the timing of deposition of the Baie-Trinité Moraine is still not well constrained.
Direct dating of the Paradise Moraine remains inconclusive.Two of the samples yielded ages >20 ka, while the three remaining samples yielded an average age of 12.1 ± 0.9 ka.This age is stratigraphically problematic as it is older than the Belles Amours Moraine.Dated boulders located west of the Paradise Moraine system yielded a mean age of 10.4 ± 0.6 ka (Ullman et al., 2016) or 9.4 ± 0.8 ka when recalculated using the same parameters as for our 10 Be ages (Supporting Information Table S3; Fig. 5).Interestingly, four samples from Ullman et al. (2016) also yielded erroneous and largely too old ages (site CL1B in the original publication).The abundance of old ages implies that boulder recycling was important during deposition of the Paradise Moraine.However, it is uncertain where these boulders came from and when they were first  2).For ages from this study, the reader is referred to Fig. 1.All ages are reported in thousand years (ka).Full lines represent mapped moraines and dotted lines represent potential extension of these systems.Extension of the moraine systems and retreat isochrons of former ice margin positions (i.e.7.9, 7.6 and 7.5 ka) were tentatively drawn based on available minimum-and maximum-limiting ages.[Color figure can be viewed at wileyonlinelibrary.com] exhumed.Similarly, the dating of lake sediments, from about 50 km east of the moraine, yielded 14 C ages ranging up to >30 cal ka BP (Lamb, 1978) and were interpreted as the result of contamination by old carbon (King, 1985).These different occurrences of ages >20 ka suggest that the boulders sampled from the Paradise Moraine were affected by isotopic inheritance yielding artificially old and erroneous exposure ages.This isotopic inheritance points toward generally low glacial erosion rates and/or transport of supraglacial material on the plateau southwest of the Mealy Mountains.However, an age of 9.4 ± 0.1 cal ka BP (King, 1985;Fig. 5; Table 2) collected from basal lake sediments 20 km west of the moraine provides an indication for a later deglaciation.A moraine system located east of the Mealy Mountains correlates with the Paradise Moraine and corresponds to a lobe emanating from Lake Melville, leaving the summits of the Mealy Mountains deglaciated at that time.Radiocarbon dating of marine shells along a riverbank east of this moraine system yielded ages ranging from 9.2 ± 0.2 to 9.6 ± 0.3 cal ka BP (Hodgson & Fulton, 1972;Fig. 5; Table 2), which is in line with the exposure age retained for the abandonment of the Paradise Moraine at 9.4 ± 0.8 ka (recalculated from Ullman et al., 2016).Correlations with other sectors of the Québec-Labrador Ice Dome are challenging at this stage as the extension of the Paradise Moraine beyond the study area remains elusive.
The Little Drunken Moraine cosmogenic exposure ages also show a wide distribution with two outliers.These results should be interpreted with caution since they were also collected from a site on the plateau southwest of the Mealy Mountains; these samples were therefore potentially subject to isotopic inheritance similarly to the Paradise Moraine samples.Nonetheless, the resulting weighted mean of 10.6 ± 1.0 ka matches the 14 C ages for the Québec North Shore Moraine, which has been previously mapped as the southwestern extension of the Little Drunken Moraine (Dubois & Dionne, 1985;Dietrich et al., 2019).In turn, recent cosmogenic dating of the Québec North Shore Moraine suggested a 10 Be age of 9.2 ± 0.5 ka for its abandonment (Ullman et al., 2016).Recalculation of these ages again gives an even younger moraine age of 8.5 ± 0.7 ka.Regardless of the calculation method, ages from Ullman et al. (2016) are consistent with a radiocarbon age of 9.3 ± 0.1 cal ka BP (King, 1985;Fig. 5; Table 2) collected from the base of a lacustrine record located less than 10 km northwest of the Little Drunken Moraine.The extension of the moraine is arguably represented across Lake Melville either on seismic data (Fig. 1C-M1; Syvitski & Lee, 1997) or on swath bathymetry imagery (Fig. 1C-M2; Gebhardt et al., 2020).Which of these two moraines represents the Little Drunken Moraine is unclear as the physical connection between these systems is missing.The connection with an unnamed moraine system located north of Lake Melville (Fig. 1C-M3) -interpreted to be equivalent to the Little Drunken Moraine by Batterson et al. (1987)-is also absent.However, a minimum-limiting age of 8.8 ± 0.2 cal ka BP (Vilks et al., 1987;Fig. 5; Table 2) on shells in Lake Melville indicates that the ice margin probably extended across the lake at the time of the deposition of the Little Drunken Moraine.To the north, a moraine system of similar age (Tasiuyak Moraine-9.3cal ka BP) reported in the Nain-Okak region could represent the northern extension of the ice margin at the time of deposition of the Little Drunken Moraine (Andrews, 1963;Recq et al., 2020).
Cosmogenic exposure dating of the Sebaskachu Moraine yielded an age of 8.4 ± 0.6 ka, which is in strong agreement with the radiocarbon 14 C ages (<8.4 ± 0.1 and >8.2 ± 0.2 cal ka BP; Table 2) bracketing the moraine formation age.A shell collected in deglacial deposits on top of the moraine near Happy Valley-Goose Bay further supports abandonment of the Sebaskachu Moraine by 8.3 ± 0.2 cal ka BP (Lowdon & Blake, 1980; Table 2).Additionally, shells collected from three ice-contact deltas north of our sample site yield ages of 8.4 ± 0.1, 8.2 ± 0.2 and 8.2 ± 0.1 cal ka BP (Fig. 5; Table 2).As these deltas are all located downstream from lakes, it is unlikely that their deposition correspond to stabilization of the ice margin further inland.These ages confirm the northern extension of the Sebaskachu Moraine up to the Adlatok Valley, where a ~20-km gap remained in the moraine mapping (Fig. 1 C).The southern extension of the moraine remains undefined, but minimum-limiting radiocarbon ages of 8.4 ± 0.1 and 8.3 ± 0.1 cal ka BP (King, 1985;Fig. 5; Table 2) collected in lake sediments suggest that the ice margin at that time was probably located ~50 km west of the Little Drunken Moraine.The age of the Sebaskachu Moraine corresponds to the timing of deposition of the extensive Sakami Moraine in western Québec (Hillaire-Marcel et al., 1981;Hardy, 1982;Lajeunesse & Allard, 2003;Lajeunesse & St-Onge, 2008;Ullman et al., 2016) and the Cockburn moraines on Baffin Island (Bryson et al., 1969;Andrews & Ives, 1978;Dyke, 1979;Miller, 1980;Briner et al., 2007;Young et al., 2012).
The Peter's Moraine exposure age of 8.1 ± 0.7 ka is comparable to a nearby site from Ullman et al. (2016), which gave an age of 7.7 ± 0.4 ka.The moraine age is similar to the age of 8.2 ± 0.2 cal ka BP on a shell collected in glaciomarine deposits along the Churchill River (Fig. 1; Table 2).It is also coeval with another shell collected in a nearby outcrop that yielded an age of 8.2 ± 0.4 cal ka BP (Lowdon & Blake, 1975;Fig. 5; Table 2).As a minor moraine system partly controlled by topographic features, it is unlikely that correlations can be made with other systems of the Québec-Labrador Ice Dome.It probably represents subordinate recessional moraines from the Sebaskachu Moraine located ~30 km to the east, as suggested by the short time interval (<0.3 ka) separating both landforms.
To summarize, while the Brador, Belles Amours and Sebaskachu moraines are particularly well dated at ~12.9, ~11.5 and ~8.4-8.2 ka respectively, the age of the Paradise and Little Drunken moraines remains uncertain.In addition, published cosmogenic exposure ages (Ullman et al., 2016) and recalibration of legacy radiocarbon ages (King, 1985) for the deglaciation west of Lake Melville allow us to refine later positions of the LIS margin for the interior of Québec-Labrador up to Churchill Falls and allow us to estimate the associated retreat rates for that period.A radiocarbon age indicates that the ice margin was located at the head of the Churchill River valley by 8.0 ka (King, 1985;Fig. 5; Table 2).Cosmogenic exposure ages (Ullman et al., 2016) and radiocarbon ages (King, 1985) indicate that the ice margin retreated rapidly to Lake Winokapau by ~7.6 ka and to Churchill Falls by ~7.5 ka (Fig. 5; Table 2).

Ice-sheet response to abrupt climatic forcings
Controls on the formation of the Brador, Belles Amours and Sebaskachu moraines, which yield robust ages (Fig. 4), are first investigated in this discussion.The significance of the Paradise and Little Drunken moraines, less well dated, need further discussion as their timing remain uncertain.
The occurrence of the Brador Moraine (12.9 ± 0.7 ka) roughly outlining the marine limit (~140 m.a.s.l.) along the coast of easternmost Québec-Labrador was speculated as representing a mass balance readjustment (re-equilibration) of the LIS when the calving ice margin reached the Brador highlands following disconnection from the Newfoundland Ice Cap (Grant, 1992).The role of the topography in stabilizing marine-based ice margins has been extensively discussed in recent literature (e.g.Jamieson et al., 2012;Batchelor et al., 2019;Brouard & Lajeunesse, 2019).However, this role appears to be secondary in this particular situation as the moraine is also observed on the plateau between valleys.Although it is difficult to deny the role played by the transition from marine-to land-based in stabilizing the ice margin, it probably reflects an equilibrium state sufficient for stabilization following a readvance -evidenced by glaciotectonic deformation-triggered by the sudden decrease in temperature by 2°C (Fig. 6; Rasmussen et al., 2014) at the beginning of the YD (12.9-11.7 ka BP).
The age of the Belles Amours Moraine (11.5 ± 0.7 ka) suggests that it corresponds to the Preboreal Oscillation (11.5-11.3ka BP), which is outlined by a sharp fall in temperatures in Greenland ice cores at the beginning of the Holocene (Fig. 6; Kobashi et al., 2017).Freshwater input into the North Atlantic by abrupt glacial lake drainages is postulated to have provoked alteration of the AMOC that resulted in several prominent early Holocene abrupt cooling events, including the Preboreal Oscillation, the 9.3-ka event and the 8.2-ka event (Barber et al., 1999;Fisher et al., 2002;Alley & Ágústsdóttir, 2005;Fleitmann et al., 2008).These freshwater inputs are expressed in the Labrador Sea as detrital carbonate peaks (DCPs) associated with a high concentration of ice-rafted debris (Jennings et al., 2015).In the case of the Preboreal oscillation, it is considered to have been triggered by drainage of the glacial Lake Agassiz via the Mackenzie River (Fisher et al., 2002;Süfke et al., 2022) and corresponds in time with DCP 1 (Fig. 6; Jennings et al., 2015).
The Sebaskachu (8.4 ± 0.6 ka) and Peter's (8.1 ± 0.7 ka) moraine ages correspond closely to the well-known 8.2-ka cooling event recorded in various sediment cores in the North Atlantic region (Barber et al., 1999;von Grafenstein et al., 1998;Jennings et al., 2015;Kleiven et al., 2008) and ice cores of the Greenland Ice Sheet (Rasmussen et al., 2014).This cooling event is widely considered to be caused by the drainage of glacial Lake Agassiz-Ojibway into the North Atlantic via the Hudson Strait (e.g.Barber et al., 1999;Clarke et al., 2004;Lajeunesse & St-Onge, 2008;Brouard et al., 2021).Moraine crests in the vicinity of Peter's Moraine are minor and may as well be topographically controlled, and hence climatically insignificant, recessional features.Nonetheless, the close occurrence of both moraines has a clear temporal coincidence with the abrupt cooling event related to the collapse of the LIS over Hudson Bay at 8.2 ka and recorded in the Labrador Sea as DCP 7 (Fig. 6; Jennings et al., 2015).However, the timing of the Sebaskachu Moraine at the start of this event could also support the hypothesis of a broader cooling perturbation beginning at ~8.6 ka BP (Rohling & Pälike, 2005;Morrill et al., 2013) as a response to the early opening of the Tyrell Sea (Fig. 6; Jennings et al., 2015).
In view of the tightly clustered individual (n = 5) ages for the Belles Amours Moraine centred at 11.5 ± 0.3 ka, the 10 Be ages for the Paradise (12.1 ± 0.6 ka) and Little Drunken (10.6 ± 0.7 ka) moraines reported here are unlikely as they are unquestionably too old.Based on their stratigraphic position, the Belles Amours Moraine is expected to be older than the Paradise Moraine and, consequently, significantly older than the Little Drunken Moraine.To cope with these discrepancies, published cosmogenic exposure ages (Ullman et al., 2016) were incorporated in our age model.Their recalculations show ages of 9.4 ± 0.8 and 8.5 ± 0.7 ka for the Paradise Moraine and Québec North Shore-Little Drunken Moraine, respectively (see Supporting Information Table S3).While there are discrepancies between the ages from the original publication and our recalculations, the uncertainties still overlap between datasets.However, two key parameters could partly explain the younger ages of the samples in regard to the actual targeted landforms: (i) Ullman et al. (2016) sampled boulders sometimes located well behind morainic crests (up to 25 km), therefore systematically minimizing the moraine ages; and (ii) some samples were collected below the reported marine limit of 150 m (Carlson, 2020) for the site CL1, consequently delaying their exposure ages by a few centuries.These ages, representing minimum values for the abandonment of the moraines, are nonetheless coherent with minimum-limiting radiocarbon ages in the region and, consequently, complete our dataset without any chronological overlap between moraine segments.Combining both datasets could therefore allow the correlation of the Paradise Moraine to the 10.3-ka event and the Little Drunken Moraine to the 9.3-ka event, as suggested by Ullman et al. (2016).While the cause of the 10.3-ka cooling event remains elusive, its consequences, recorded by (i) the presence of DCP 2 (Fig. 6; Jennings et al., 2015) and (ii) the widespread stabilization of the ice margins in western Greenland and eastern Canada (Young et al., 2020a(Young et al., , 2021) ) are unequivocal and a correlation with the major Paradise Moraine is most likely.The 9.3-ka event, which is correlated to DCP 3 or 4 (Fig. 6; Jennings et al., 2015), was speculated to have been triggered either by the Noble Inlet readvance across Hudson Strait (Jennings et al., 2015) or a meltwater outburst flow from Lake Superior (Yu et al., 2010).This abrupt cooling event has been recorded around Baffin Bay as a widespread ice margin stabilization event (e.g.Lesnek & Briner, 2018;Crump et al., 2020;Young et al., 2020a) and may therefore correlate with the timing of deposition of the Little Drunken Moraine.The occurrence of two stabilizations in Lake Melville (M1, M2-Fig.1C) could in fact correspond to the two DCP peaks observed around that event, if the moraine system identified on seismic data (M1) could be the result of a drawdown of the ice as speculated by several authors (i.e.Fulton & Hodgson, 1979;Vilks et al., 1987).This drawdown into Lake Melville may have provoked only a localized stabilization, although its occurrence elsewhere remains to be determined.
Finally, the compilation of cosmogenic exposure ages presented here, together with previously published cosmogenic and radiocarbon ages, identifies margin stabilizations and/or readvances of the eastern fringe of the LIS at ~12.9, ~11.5, ~10.4,~9.3 and ~8.4-8.2 ka.This comprehensive 10 Be-based chronology clearly demonstrates that the timing of moraine deposition in easternmost Québec-Labrador was coeval with cold climatic events recorded in Greenland ice cores (Fig. 6), indicating that the LIS margin was interconnected with climate fluctuations at least regionally.The synchronicity of moraine deposition along the eastern fringe of the LIS from Baffin Island to southern Québec -that connect both continental-and marine-based segments of the ice margin-suggests strongly that readvances of the ice margin were triggered by major, short-term regional climatic deteriorations rather than by a mechanical adjustment or re-equilibration of the ice sheet (Hillaire-Marcel et al., 1981;Dubois & Dionne, 1985;Clark et al., 2000).This deglaciation chronology supports results from around Baffin Bay suggesting that early Holocene ice margin readvances and/or stabilizations of the LIS -and Greenland Ice Sheet-were caused by meltwater pulses into the Labrador Sea (Young et al., 2020a(Young et al., , 2021)).Similar to their interpretation, detrital carbonate peaks from the Cartwright Saddle on the Labrador Shelf (Jennings et al., 2015) preceding the stabilizations/readvances of the ice margin in easternmost Québec-Labrador possibly indicate that abrupt cold reversals of the early Holocene were strongly influenced by meltwater pulses weakening the AMOC.While early Holocene cooling events appear to be triggered by freshwater input -often caused itself by abrupt lake drainage-into the Atlantic Ocean and recorded by detrital carbonate events in the Labrador Sea, further investigation should focus on identifying trigger mechanisms for such freshwater outbursts.The 1000-to 1200-year cyclicity behind those cooling events may suggest an external   (2021).(E) Average retreat rates in southern Labrador reconstructed from this study (black line) with the standard error of the mean (shading).Dashed dark grey line is from Dalton et al. (2020).Note that retreat rates are represented in a logarithmic scale.(F) Greenland mean-annual temperatures reconstructed using gas-phase δ 15 N-N 2 measurements (purple -±1σ; Buizert et al., 2014).(G) Greenland mean-annual temperatures reconstructed using gas-phase δAr-N 2 measurements (red -±2σ; Kobashi et al., 2017).(H) δ 18 O record from NGRIP project (orange; Rasmussen et al., 2014) forcing that influenced the rapid retreat of the ice margin prior to these events, thus favouring meltwater outburst.First proposed by Denton & Karlén (1973) -and later advocated by Bond et al. (1997)-the cyclicity of Holocene cooling events has been for decades very puzzling to researchers.Bond-like events, speculated by some to be influenced by enhanced solar activity (e.g.Bond et al., 2001), may certainly still be relevant to identify a source for such cycles, although their occurrence may as well simply be the result of several trigger mechanisms (Young et al., 2021).

Retreat rates reflect long-term climatic trends
Retreat rates across the study area probably reflect long-term (centennial-to millennial-scale) climatic trends, with minimum recession during colder periods and increasing rates corresponding to increasing temperatures.Minimum recession rates of the LIS margin in easternmost Québec-Labrador were observed during the YD (~25 m a −1 ) and again between 9.3 and 8.2 ka (~50 m a −1 ).The YD was marked by significantly lower temperatures over a relatively long period of ~1200 years (Fig. 6; Buizert et al., 2014), therefore contributing to a significant slowdown in ice retreat.A similar assessment could be made for the period between 9.3 and 8.2 ka, when another significant reversal in the warming temperature trend is observed (Fig. 6; Kobashi et al., 2017).Increasing retreat rates across the study area from 11.5 to 9.3 ka (75-90 m a −1 ) could be linked to the warming climatic trend recorded in Greenland ice cores (Fig. 6; Kobashi et al., 2017).Collapse of the LIS can be recorded after 8.2 ka -regardless of climatic trend-as retreat rates started increasing drastically, reaching up to 1,000 m a −1 by 7.5 ka (Fig. 6).
Several lines of explanation can be invoked to explain the collapse of the LIS in central Québec-Labrador: (i) rapidly increasing temperatures in the Arctic and subarctic regions -similar to present-day observations-directly affecting its accumulation zone and resulting in negative mass balance (Carlson et al., 2009); or (ii) the Québec-Labrador Ice Dome reached a tipping point from which it was no longer in balance with climate and sustainable under early Holocene climatic conditions.Recurrent climatic cooling phases provoked by meltwater discharge events were suggested to have helped 'artificially' sustain the Québec-Labrador Ice Dome during the early Holocene (Ullman et al., 2016).In this perspective, it is possible that the absence of such events after 8.2 ka favoured the collapse of the LIS as a lagged response to the overall Holocene warming trend (Shakun et al., 2012;Buizert et al., 2014).With a constant negative mass-balance at the time, the Québec-Labrador Ice Dome eventually disappeared by 6 ka (Richard et al., 1982;Clark et al., 2000;Jansson, 2003;Occhietti et al., 2011;Dubé-Loubert et al., 2018).
However, it can be stressed that calculating retreat rates between stabilizations of the ice margin is only tentative in this situation, smoothing inappropriately the signal of ice-sheet readvances.Therefore, using an approach integrating estimates of hiatus duration (e.g.Lowell et al., 2021) would be more appropriate in order to exclude uncertainties related to the length of stabilizations and magnitude of potentially preceding readvances.However, this method would require further dating to better constrain the timing of the ice margin readvances and duration of the stabilizations.

Conclusion
This paper reports on 37 new cosmogenic 10 Be and four new radiocarbon ages in easternmost Québec-Labrador that allow us to establish a chronological framework of major moraine systems deposited during stabilizations/readvances stages of the LIS margin.These results improve the deglaciation history drawn for the region by earlier workers and document periods of regional ice margin stabilization at ~12.9, ~11.5, ~10.4,~9.3 and ~8.4-8.2 ka.The combined ages outline a close relationship between climate and glaciodynamics of the LIS margin during the late Wisconsinan-early Holocene transition and are in line with previous work on Baffin Island and western Greenland suggesting that glacial dynamics were controlled by abrupt climatic events and possibly synchronous along the margin of entire ice sheets.The physical connection between the easternmost Québec-Labrador moraines and other systems of the eastern fringe of the LIS remains, however, ambiguous.Although correlations can be made with major moraine systems of southern Québec, large geographical gaps and chronological uncertainties persist.The transition between a land-to marinebased ice margin makes it difficult to accurately connect undated moraines from one region to another without a complete swathe of bathymetric coverage offshore.
Future work focusing on connecting the easternmost Québec-Labrador moraine systems with those of the Québec North Shore and northern Labrador is needed to provide a larger-scale palaeogeographical perspective of the major stages of the LIS margin in a period of rapid climatic fluctuations across a wide range of latitudes, as well as marine to continental contexts.Further direct dating is needed on moraine systems across the Québec-Labrador region to assess more precisely the extent of the LIS and the final stages of deglaciation at the YD-early Holocene transition.A better chronological constraint would not only be essential for future empirical-based numerical modelling assessing the contribution of the LIS to global sea-level changes, but it would also greatly increase our understanding of abrupt climatic cooling events induced by ice sheet-derived meltwater discharge into the North Atlantic.Although an almost consistent cyclical pattern of ~1000-1200 years is observed between the major ice margin stabilizations reported here during the early Holocene, the origin of these events as negative feedback triggered by meltwater inputs into the North Atlantic has yet to be explored as an analogue to potential forthcoming climate variation.

Figure 1 .
Figure 1.Location of the study area with mapped moraine systems.(A) Extent of the ice cover (light grey) in North America during the Last Glacial Maximum, modified from Dalton et al. (2020).(B) Location of mapped moraine systems of the LIS discussed in this study, modified from Occhietti et al. (2011).NL: Newfoundland.(C) Easternmost Québec-Labrador and major moraine systems presented in this study.Sample locations of our new cosmogenic

Figure 4 .
Figure 4. Probability distribution function (PDF) plots of 10 Be ages for moraine systems in easternmost Québec-Labrador, excluding outliers (see Results and Table 1).The black and dashed lines represent the weighted average and the 1σ interval.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 5 .
Figure 5. Previously published ages in easternmost Québec-Labrador.Cosmogenic 10 Be ages are recalculated from Ullman et al. (2016).Radiocarbon 14 C ages are compiled from various sources (Table2).For ages from this study, the reader is referred to Fig.1.All ages are reported in thousand years (ka).Full lines represent mapped moraines and dotted lines represent potential extension of these systems.Extension of the moraine systems and retreat isochrons of former ice margin positions (i.e.7.9, 7.6 and 7.5 ka) were tentatively drawn based on available minimum-and maximum-limiting ages.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 6 .
Figure 6.(A) Exposure ages of moraine systems of easternmost Québec-Labrador from this study.(B) Exposure ages of moraine systems of Labrador and Québec from Ullman et al. (2016).(C) Exposure ages of moraine systems of LIS outlet glaciers on Baffin Island from Young et al. (2020a).Additionally, we used one site from Briner et al. (2007) and recalculated in Young et al. (2013).(D) Composite record of Baffin Bay moraine deposition, from Young et al.(2021).(E) Average retreat rates in southern Labrador reconstructed from this study (black line) with the standard error of the mean (shading).Dashed dark grey line is fromDalton et al. (2020).Note that retreat rates are represented in a logarithmic scale.(F) Greenland mean-annual temperatures reconstructed using gas-phase δ 15 N-N 2 measurements (purple -±1σ;Buizert et al., 2014).(G) Greenland mean-annual temperatures reconstructed using gas-phase δAr-N 2 measurements (red -±2σ;Kobashi et al., 2017).(H) δ 18 O record from NGRIP project (orange;Rasmussen et al., 2014).(I) Total carbonate weight chronology in core MD99-2236 collected on the Labrador Shelf (blue; Jennings et al., 2015; Fig. 1).Vertical bars represent cold intervals discussed in the text.YD: Younger Dryas; PB: Preboreal.[Color figure can be viewed at wileyonlinelibrary.com] Figure 6.(A) Exposure ages of moraine systems of easternmost Québec-Labrador from this study.(B) Exposure ages of moraine systems of Labrador and Québec from Ullman et al. (2016).(C) Exposure ages of moraine systems of LIS outlet glaciers on Baffin Island from Young et al. (2020a).Additionally, we used one site from Briner et al. (2007) and recalculated in Young et al. (2013).(D) Composite record of Baffin Bay moraine deposition, from Young et al.(2021).(E) Average retreat rates in southern Labrador reconstructed from this study (black line) with the standard error of the mean (shading).Dashed dark grey line is fromDalton et al. (2020).Note that retreat rates are represented in a logarithmic scale.(F) Greenland mean-annual temperatures reconstructed using gas-phase δ 15 N-N 2 measurements (purple -±1σ;Buizert et al., 2014).(G) Greenland mean-annual temperatures reconstructed using gas-phase δAr-N 2 measurements (red -±2σ;Kobashi et al., 2017).(H) δ 18 O record from NGRIP project (orange;Rasmussen et al., 2014).(I) Total carbonate weight chronology in core MD99-2236 collected on the Labrador Shelf (blue; Jennings et al., 2015; Fig. 1).Vertical bars represent cold intervals discussed in the text.YD: Younger Dryas; PB: Preboreal.[Color figure can be viewed at wileyonlinelibrary.com]

Table 1 .
Sample characteristics, AMS measurement results and 10 Be ages.

Table 2 .
Radiocarbon and calibrated radiocarbon ages from material collected in easternmost Québec-Labrador.