Holocene paleoceanography of the Northeast Greenland shelf

. The Northeast Greenland shelf is highly sensitive to climate and ocean variability because it is swept by the East Greenland Current, which, through the western Fram Strait, forms the main pathway of export of sea ice and cold water masses from the Arctic Ocean into the North Atlantic Ocean. We carried out benthic foraminiferal assemblage, stable isotope- and sedimentological analyses of a marine sediment core retrieved from the Northeast Greenland shelf (core DA17-NG-ST7-73), which 10 provided a multiproxy reconstruction of Holocene paleoceanographic conditions. The results reveal significant variations in the water masses and thus, in the strength of the East Greenland Current over the last ca. 9.4 ka BP. Between 9.4 and 8.2 ka BP the water column off Northeast Greenland was highly stratified, with cold, sea ice-loaded surface waters and strong influx of warm Atlantic Water in the subsurface. At ~8.4 ka BP a short-lived peak in terrestrial elements may be linked to influx of ice-berg transported sediments and thus, to the so-called 8.2 ka event. Holocene Thermal Maximum like conditions prevailed from 8.2 to 15 6.2 ka BP, with a strong influence of the Return Atlantic Current and a weakened transport of Polar Water in the upper East Greenland Current. After 6.2 ka BP we recorded a return to a more stratified water column with sea-ice loaded surface waters and still Atlantic-sourced subsurface waters. After 4.2 ka BP increased Polar Water at the surface of the East Greenland Current and reduction of the Return Atlantic Water at subsurface levels led to freshening and reduced stratification of the water column and (near) perennial sea-ice cover. The Neoglaciation started at 3.2 ka BP at our location, characterized by a strengthened East 20 Greenland Current. Cold subsurface water conditions with possible sea-ice cover and minimum surface water productivity persisted here throughout the last ~3 kyr. shelf (Syring et core That study reveals a cold interval with increased sea-ice cover between 9.3 and 7.9 ka found in relatively low numbers, although there is some increase in species linked to chilled Atlantic Water (Figs. 6 and Continuously increasing benthic δ 13 C values indicate strong ventilation of the water column (Fig. 7). The results point to cold and unstable conditions with 435 minimum surface water productivity and increasing sea-ice cover at our location. During this period, the EGC was likely strengthened compared to previously, with a thick layer of cold and fresh PW on the surface and recirculated AAW inflow from the Arctic at subsurface levels, as seen in the chilled Atlantic Water group. Telesinski in particular at locations 515 influenced by the EGC. This gradual cooling trend in the northern North Atlantic is associated therefore with a continuous transition from a positive towards a negative NAO phase (e.g., Andersen et al., 2004; Orme et al., 2018). Positive NAO phases are


Introduction
The acceleration of climatic changes in the Arctic and Subarctic regions is particularly marked by the drastic reduction of summer sea-ice cover. According to model simulations, the Arctic Ocean may become seasonally ice-free as early as 2040-2050 (Stroeve 25 et al., 2012). However, despite the extreme societal and environmental relevance of this sea-ice reduction, Holocene marine environments and corresponding natural sea-ice states around Greenland are still not well understood. The Northeast (NE) Greenland continental shelf is a particularly important region for studying and understanding the mechanisms that control sea-ice formation, melt and drift. It is the broadest shelf along the Greenland margin, extending more than 300 km from the coastline. To the west it reaches the East Greenland coastline with its major marine-terminating outlets of the Greenland Ice Sheet, while to the 30 east it is bounded by the Fram Strait, which is the only deep passage between the Arctic Ocean and the rest of the world oceans.
This shelf region is highly sensitive to climate and ocean variability because it underlies the East Greenland Current (EGC), which, through the western Fram Strait, is the main pathway of export of sea ice and cold water masses from the Arctic Ocean into the North Atlantic Ocean (Rudels and Quadfasel, 1991). The outflow of melt water/sea ice is a key factor that determines the North Atlantic freshwater budget, stratification and influences the deep-water formation. Moreover, waters on the NE Greenland 35 https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. 100 The Polar Front in the Greenland Sea represents the eastern limit of perennial sea-ice cover, and the location of the sea-ice edge in the summer depends on the extent of the sea-ice export from the Arctic Ocean (Vinje, 1977). The sea-ice margin and the location of the Polar Front lies today east of our study site (Danish Meteorological Institute (DMI) and NSIDC, 2012), however, the Northeast Water (NEW) Polynya is in fairly close proximity to our core location. This is a seasonally ice-free or only loosely icecovered area south of Nordøstrundingen (Fig. 1) on the eastern Greenland coast. Two ice barriers support the seasonal formation 105 of the NEW Polynya, the Ob Ice Barrier by pushing drift ice eastward, while the Norske Øer Ice Barrier (NØIB) blocks the northward flowing sea ice which is entrained in the NE Greenland coastal current (Schneider and Budéus, 1994). It starts to open around May/June, then gradually increases its size and closes in September (Pedersen et al., 1993). The polynya's maximum extent to the north can reach high latitudes up to 83°N, beyond the NE Greenland shelf, and to the east the polynya can occupy the entire NE Greenland shelf with only little sea-ice left in the area (Schneider and Budéus, 1997). 110

Material and methods
Gravity core DA17-NG-ST7-73G (79.068° N 11.903° W) was collected during the NorthGreen Expedition in September 2017 at 120 385 m water depth in an isolated bathymetrical depression on the NE Greenland shelf (Fig. 1). In addition, we collected Rumohr core DA17-NG-ST7-72R (79.072° N 11.888° W, 384.6 m water depth) with an intact sediment-water interface from the same station, which provided information for the age model of the gravity core. Temperature and salinity of the water column were measured in situ by a Seabird CTD (Conductivity-Temperature-Depth) profiler before the deployment of the gravity corer, with the core site being identified based on Innomar® subbottom profiler data obtained prior to coring. 125 The 410 cm long gravity core (73G) and the 88 cm long Rumohr core (72R) were stored at ca 3 °C before they were split lengthwise.
The archive halves were scanned with an Itrax X-ray fluorescence core scanner (Cox Analytical) at the Department of Geoscience, Aarhus University (AU), Denmark. The scan was conducted with a Molybdenum tube (other settings: stepsize: 200 µm; exposure time: 10 s; voltage: 30 kV, current: 30 mA). The core scanning provided a line scan image, a 2-cm-wide radiographic image of the 130 https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. center of the core (voltage: 60 kV; current: 45 mA; exposure time: 2000 ms) and high-resolution analysis of bulk geochemical composition. From the results of the X-ray fluorescence scanning, we show the elements Si, K, Ti and the Ca/Fe ratio on Fig. 3.
The age-depth model of core 73G is based on six Accelerator Mass Spectrometry radiocarbon (AMS 14 C) dates, of which five are of mixed benthic foraminifera and one is of a worm tube lining. AMS 14 C dating was performed at the Laboratory of Ion Beam 135 Physics, ETH Zürich, Switzerland. Radiocarbon ages were calibrated using the Marine20 calibration dataset (Heaton et al., 2020) and a local marine reservoir age correction (ΔR) of 0 ± 50 years (Reimer and Reimer, 2001). The age-depth model was made with the OxCal v4.4 software (Ramsey, 2008), using a P-sequence depositional model. Beyond the lowest radiocarbon date, the model is extrapolated toward the bottom of the core. The age of the upper part of the core is constrained by correlation to Rumohr core 72R by selected XRF elements (appendices, Fig. A1), to determine the amount of top sediment lost during gravity coring. 140 Samples for foraminiferal, ice-rafted debris (IRD) and stable isotope analyses were taken at 5 cm intervals at the Department of Geoscience, AU. The samples were weighed, wet-sieved with distilled water through sieves of 1000, 100 and 63 µm, dried at 40 °C for 24 hours and weighed again. For the foraminiferal assemblage analyses both planktic and benthic (calcareous and agglutinated) species were counted and identified to species level under a stereomicroscope for the size fractions of 63-100 µm 145 and >100 µm; counts were subsequently combined prior to percentage calculations. When the samples contained excessive material, prior to counting they were split into equal parts containing >300 specimens, using a microsplitter. In these cases, foraminiferal tests of one part were counted and identified, and the total number of foraminifera was calculated. A minimum of 300 benthic specimens were identified for each sample, except for four samples (10-11 cm, 15-16 cm, 30-31 cm, 195-196 cm).
These four samples contained only between 242-296 specimens. These data were, therefore, treated with some caution, but they 150 were still included in the calculations for relative and absolute abundances. Absolute foraminiferal concentrations were calculated as individuals/g wet sediment. Relative abundances of both benthic agglutinated and calcareous species were calculated as a percentage of the total benthic (calcareous and agglutinated) foraminiferal fauna. However, diagrams of the separate percentage calculations of the two benthic foraminiferal groups (agglutinated species of total agglutinated assemblage and calcareous species of total calcareous assemblage) is shown in the appendices (Figs. A2 and A3). 155 For stable isotope analysis 5-200 specimens of the benthic foraminiferal species Elphidium clavatum were picked from every sample. The number of picked tests was restricted by the number of available, clearly identifiable tests. The specimens belonged mainly to the size 100-1000 µm fraction, supplemented with few specimens from the 63-100 µm fraction, when necessary. The oxygen and carbon isotope analysis of foraminiferal calcite was performed at the Leibniz Laboratory for Radiometric Dating and 160 Stable Isotope Research at the Christian-Albrechts-University of Kiel using a MAT253 (Thermo Scientific) mass spectrometer system and a Kiel IV carbonate preparation device, with an analytical accuracy of <0.08 ‰ for δ 18 O and <0.05 ‰ for δ 13 C. Results are expressed in δ notation referring to the PDB (Pee Dee Belmnite) standard, while using NBS-19 and IAEA-603 carbonate standards.

165
The content of IRD was calculated at 5 cm sample intervals using the dry weight of the size fraction >1000 µm divided by the wet weight of the whole sample.

Core description
Core DA17-NG-ST7-73G consist entirely of olive grey (5Y 4/1) marine silt with some clay, except the lowermost part of the core.
The sediment of the lowest 40 cm of the core is much coarser, containing up to 51 % sand; it is also darker in colour (Fig. 3). In the rest of the core the sand fraction is on average 1.5 %. 175

Chronology
The age model was developed using six AMS 14 C measurements on mixed benthic foraminifera and on a worm tube lining from 180 core 73G (Table 1). In order to evaluate potential loss of core top sediment during the coring process, we compared the XRF data of gravity core 73G and Rumohr core 72R (Fig. A1 in appendices), which indicated that 12 cm sediment was missing from the top of core 73G; this information was used when constructing the age model of core 73G. This age model suggest that the 410 cm long sediment core covers the last ~11.2 ka BP (Fig. 4). However, the results of the grain size, IRD and X-ray fluorescence analysis indicate that beyond the lowest radiocarbon date (at 345 cm depth, 9.4 ka BP) the sediments change significantly (Fig. 4). In the 185 bottom 65 cm of the core preliminary analysis of foraminiferal content revealed the presence of species indicating a Pliocene or early Pleistocene age (Feyling-Hanssen, 1976;Feyling-Hanssen, 1980;Feyling-Hanssen et al., 1983), in addition to the normal in situ foraminiferal assemblage. We suggest that these Plio-/Pleistocene foraminifera are reworked into a glaciomarine setting. Due to this very different environment and the lack of 14 C dates in this interval, the age model of the bottom 65 cm of core 73G is currently uncertain. Therefore, apart from a short description of the foraminiferal content in the results, we have omitted the bottom 190 65 cm of the core (between 345 and 410 cm) from further analyses and discussion, focussing on the last 9.4 ka. https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License.

X-ray fluorescence
Throughout the whole core, Si, K and Ti follow the same pattern. All three records show a relatively stable trend from the base of the core until ~318 cm depth (ca. 8.5 ka BP). At 316 cm (ca. 8.4 ka BP) all three elements displays peak counts, followed by a decrease until 161 cm (ca. 4.2 ka BP). There is a short-term drop in all three element counts at 161 cm depth (ca. 4.2 ka BP), which 205 is followed by a steady increase until the top of the core (Fig. 3). The Ca/Fe ratio follows a steadily decreasing trend throughout the core, with slightly stronger fluctuations from the end of the core until ~270 cm (ca. 7 ka BP) and in the top 55 cm (last ~1.5 ka BP). The record shows a pronounced peak between 335 and 338 cm (ca. 9.1 ka BP), and two minor peaks at ~213 cm (ca. 5.5 ka BP) and at ~34 cm (ca. 0.9 ka BP) (Fig. 3).

410-345 cm core depth
The bottom 65 cm of the core are characterised by a mixture of two assemblages. An assemblage of relatively small, well-preserved specimens of Arctic species include Cassidulina reniforme, Elphidium clavatum, Islandiella norcrossi and Melonis barleeanum.

345-0 cm core depth
We counted in total 70 samples, and identified 65 benthic (44 calcareous and 21 agglutinated) and three planktic foraminiferal 220 species (Table A1 in the appendices) in the top 345 cm of core 73G. Both benthic and planktic specimens were well-preserved and showed minor or no signs of post-mortem dissolution of the tests.
The concentration of planktic foraminifera is extremely low throughout the core, it lies between 0 and 13 ind. g -1 sed. Planktic 225 foraminifera are represented primarily by the polar species Neogloboquadrina pachyderma, occasionally accompanied in the interval between 340-130 cm (ca. 9.4-3.4 ka BP) by specimens of the subpolar species Turboratalita quinqueloba and Neogloboquadrina incompta. The overall occurrence of planktic foraminifera is averaging 2 % of the total foraminiferal fauna.
On average, benthic agglutinated foraminifera account for 65 % of the total foraminiferal assemblage. From the end of the record until 210 cm (ca. 5.4 ka BP) the calcareous species periodically outnumber the agglutinated, however, from this point on and until 230 the top of the record, agglutinated foraminifera continuously dominate the benthic assemblage ( abundances of E. clavatum shows the least variability throughout the record, however, it follows a steadily decreasing trend towards the top of the core (Fig. 5). In the following sections, we describe foraminiferal assemblage zones (ecozones) that were defined by visual interpretation of the species abundances. Boundaries were placed where major changes occurred in the relative abundances of the most abundant benthic calcareous and agglutinated species, indicating changes in the environment (Figs. 5 and 6).

Ecozone I. (345-310 cm; ca. 9.4-8.2 ka BP) 250
This ecozone is characterized by the highest content of total foraminifera (171 ind. g -1 sed.), including the highest concentrations of planktic (13 ind. g -1 sed.) and benthic calcareous (128 ind. g -1 sed.) foraminifera throughout the core. These three categories show two peaks in this interval: the first and bigger at ca. 9 ka BP and the second, smaller one at ca. 8.7 ka BP. We find in this interval the lowest agglutinated/calcareous ratio, as the calcareous specimens almost continuously outnumber the agglutinated ones. The average concentration of agglutinated foraminifera is 29 ind. g -1 sed. in this zone, and it does not vary significantly 255 throughout the other intervals either (27-29 ind. g -1 sed.). The benthic calcareous assemblage is dominated by C. reniforme (relative abundance on average 14 %), followed by S. horvathi (on average 12 %), E. clavatum (on average 11 %) and C. neoteretis (on average 9 %). In this ecozone, we can recognize a distinctive decrease in planktic and benthic calcareous foraminiferal concentrations, and as a 260 consequence, a higher benthic agglutinated/calcareous species ratio. The relative abundance of C. neoteretis promptly increases, and it becomes the dominating species of the benthic calcareous fauna in this interval (highest relative abundance 27 %, on average 13 %), followed by C. reniforme (on average 10 %). Pullenia bulloides suddenly appears in the record and, at the same time, the relative abundance S. horvathi drastically decreases and it remains around 1 % during the whole period. Even though the relative abundances of E. clavatum decrease compared to the previous interval, this species shows its highest relative abundance throughout 265 the core in this ecozone at 255 cm (17 %; ca. 6.5 ka BP).The relative abundance of agglutinated species Adercotryma glomerata significantly increases from 310 cm on and remains on the same level throughout this zone.
agglutinated species A. glomerata shows a drastic peak at the beginning of this ecozone, and the relative abundances of T. torquata, Textularia cf. earlandi and Spiroplectammina biformis increase significantly compared to the previous interval. The concentrations of planktic foraminifera and the relative abundances of C. neoteretis further decrease and reach their lowest 285 level throughout the core, while C. reniforme shows an increase compared to ecozone IV, just like S. feylingi. The relative abundances of agglutinated species T. torquata, T. cf. earlandi and S. biformis continue to increase, while Saccamina difflugiformis shows a steep rise unique to this interval.

Stable isotopes
From the 70 analysed samples the results of three samples were not accepted: one (180 cm) due to technical problems, and two 290 gave too low signal due to small sample size (0 cm and 130 cm). The sample at 140 cm gave a comparatively very low δ 18 O (2.25 ‰) and a very high δ 13 C value (-1.07 ‰), but these outliers were nonetheless accepted.
Except for the one outlier, the variation of δ 18 O values is rather small throughout the core (between 2.85 and 3.69 ‰, with a mean of 3.29 ‰), and the values show a stable trend, albeit with a slight increase towards the top of the core. The δ 13 C values show a constant increase from the bottom to the top of the core, varying between -1.07 and -2.61 ‰ (Fig. 7). 295

Environmental significance of foraminiferal assemblages
In order to be able to describe the changes in water masses over time on the NE Greenland shelf, we place selected benthic calcareous foraminifera species into groupings that are based on environmental preferences of the species (Table A2 in appendices).
The Atlantic Water group includes C. neoteretis and P. bulloides. These species indicate warm and saline AW inflow underneath cold and low salinity surface waters (e.g., Mackensen and Hald, 1988;Seidenkrantz 1995;Rytter et al., 2002;Jennings et al., 2004;300 Jennings et al., 2011;Cage et al., 2021). The chilled Atlantic Water group includes I. norcrossi and M. barleeanum. These species have been previously linked to cool AW (e.g., Slubowska-Woldengen et al., 2007;Perner et al., 2011;Perner et al., 2015;Cage et al., 2021), and we use them to represent the relative contribution of chilled AW recirculated in the Arctic Ocean to the EGC. The Arctic Water group includes S. horvathi and E. arctica. Both species live at present beneath perennial sea ice, in the deeper part of the Arctic Ocean, thus indicating cold, Arctic originated deep-waters (e.g., Green, 1960;Lagoe, 1979;Wollenburg and Mackensen, 305 1998;Jennings et al, 2020). We consider S. feylingi as a sea-ice edge indicator species that tolerates unstable conditions (Knudsen and Seidenkrantz, 1994;Seidenkrantz, 2013); its increase may refer to the location of a sea-ice margin at the study site. Moreover, we use in the interpretation the abundances of the agglutinated species A. glomerata, T. earlandi, T. torquata, S. biformis and S. difflugiformis. A. glomerata is often correlated with Atlantic sourced waters in high latitudes (e.g., Korsun, 1997, Lloyd 2006;Perner et al., 2012). Many studies find T. earlandi, T. torquata and S. biformis in areas influenced by cold, low-salinity 310 Arctic water, often in glaciomarine environments (e.g., Jennings and Helgadottir, 1994;Korsun and Hald, 2000;Perner et al, 2012;Perner et al., 2015, Wangner et al., 2018. S. difflugiformis is described as closely associated with Polar Water on the East Greenland shelf and fjords (Jennings and Helgadottir, 1994).   Although the study site is located on the central flow route of the EGC, the path of the influx of the Atlantic-sourced water, which is clearly identified as the bottom layer through the CTD (Fig. 2), is less clear. The temperature profile with bottom water temperatures not exceeding 1 °C suggest that the Atlantic-sourced water is closer in character to that found in Westwind Trough 325 https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. than the Atlantic-sourced water presently seen in Norske Trough. This may indicate a primarily AAW source (Rudels et al., 2005;Schaffer et al., 2017). The Atlantic-sourced water may thus reach the site via the Westwind Trough, but may also be due to winddriven upwelling. This process leads to considerable modification of local water masses and shelfbreak exchanges (e.g., Estrade et al., 2008). In the Arctic it is often associated with the onshore transport of warm, and relatively salty AW to the continental shelves (e.g., Kirillov et al., 2016). 330

Bottom 65 cm of the sediment core
The combined presence of an assemblage of well-preserved small tests of Arctic foraminiferal species and many extinct Plio-Pleistocene specimens suggest that a Plio-Pleistocene assemblage was reworked from older deposits such as Kap København (Feyling-Hanssen, 1976;Feyling-Hanssen, 1980;Feyling-Hanssen et al., 1983;Funder et al., 2001), into a glaciomarine 335 environment with in situ Arctic foraminifera. Potentially the breakup and significant retreat of a nearby glacier caused reworking of older sediments as also previously suggested for other sites (e.g., Seidenkrantz et al., 2019), but further studies are required to resolve this issue.

Early Holocene (ca. 9.4-8.2 ka BP; ecozone I.)
This cold interval was characterized by high percentages of S. horvathi, which has previously been considered linked to high sea-340 ice concentrations (e.g., Green, 1960;Lagoe, 1979;Wollenburg and Mackensen, 1988;Jennings et al., 2020) combined with the presence of Atlantic Water indicator species. This combination of Atlantic-sourced bottom water and sea ice points to a highly stratified water column in the early Holocene (Fig. 5). Cold, heavily sea ice-loaded surface waters entrained in PW of the upper EGC would have characterised the upper water column, while at the same time the influx of warm Atlantic-sourced water from either the RAC or the AAW (or a combination of both) was also strong. When compared to the data from the top of the core, the 345 water column stratification between the cold and warmer waters was likely more pronounced than today. Relatively high counts of the terrestrially-derived elements Si, K and Ti, together with relatively low δ 18 O values indicate increased meltwater influence from the Greenland Ice Sheet (Figs. 3 and 7).
The interval was also characterised by high surface water productivity, as shown by high concentrations of planktic and benthic calcareous foraminifera, the presence of E. arctica, which thrives in high-productivity environments (Wollenburg and Kuhnt, 2000) 350 and a pronounced peak in Ca/Fe ratio, which is considered as a general proxy for paleoproductivity (e.g., Vare et al., 2009). The planktic foraminifera may have been transported to the site in the subsurface water layer from a nearby more open-ocean site, but they still suggest increased surface water-bioproductivity in the general region. Low and slowly rising benthic δ 13 C values can also be related to high level of bioproductivity, due to decomposition of isotopically light organic carbon in the lower part of the water column. We propose that high productivity combined with sea-ice indicates that our site was located west of the Polar Front, but 355 that the Polar Front was situated in close proximity to our location in this time interval. It should be noted that in the XRF data, we recognize a short-lived, but clear peak in terrestrially-derived sediments around 8.5-8.4 ka BP (Fig. 3). This may be potentially linked to increased iceberg release during the so-called 8.2. ka event (i.a., Alley et al., 1997;Barber et al., 1999), however, this signal seems to be only reflected in the results of the sedimentological analyses, and not in the foraminiferal assemblage changes/stable isotope results, although the lower temporal resolution of the latter may also not allow us to identify any changes. 360 Our suggestions of cold water and sea ice at the surface is in accordance with a recent sea-ice reconstruction from the NE Greenland shelf (Syring et al., 2020a; core PS93/025). That study reveals a cold interval with increased sea-ice cover between 9.3 and 7.9 ka https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. BP, shown by lower concentrations of marginal sea-ice proxy HBI III and IP25, interpreted as low sea-ice algal productivity caused by an expansion of the sea-ice cover (Fig. 7). Foraminiferal records from the Southeast Greenland shelf (Jennings et al., 2011;365 cores MD99-2322 and MD992317) suggests similarly cold and unstable sea-surface conditions between 9.4 and 8.1 ka BP.
Moreover, cool deep-water conditions in the eastern Fram Strait during this period (Consolaro et al., 2018; core JM10-330) point to a weakened RAC, suggesting that the strong water column stratification on the NE Greenland shelf was due to increased AAW influx.

Holocene Thermal Maximum (ca. 8.2-6.2 ka BP; ecozone II.) 370
The interval from 8.2 to 6.2 ka BP was characterised by the warmest bottom-water conditions of the Holocene on the NE Greenland shelf, as well as reduced sea-ice cover, as indicated by the almost complete absence of sea-ice indicator species and maximum abundances of Atlantic Water species and other species indicating stable bottom waters. C. neoteretis and A. glomerata had their highest relative abundances during this interval (except for an extreme peak of A. glomerata at ~4 ka BP), and P. bulloides appeared in the record after a long absence around 8 ka BP (Figs. 5 and 6), suggesting highly stable bottom waters (Rytter et al., 2002). A 375 concurrent significant decrease in planktic and benthic calcareous foraminiferal concentrations indicates a transition towards lower surface and subsurface water productivity, also shown in decreasing Ca/Fe element ratios.
We suggest that this may be explained by a strengthening of the RAC, transporting RAW to our site and causing increased entrainment of warm Atlantic-sourced water into the EGC. As a consequence of likely vertical expansion of the Atlantic-sourced water layer, surface waters likely also warmed, and the Polar Front moved from our location to further north or further inland 380 during this period, which represents the warmest of our record, and which we here refer to as the Holocene Thermal Maximum  (Fig. 7), which would have resulted in a stronger RAC. In line with our findings, Müller et al. (2012) (core PS2641-4) and Jennings et al. (2011) (cores MD99-2322and MD99-2317 reconstructed relatively warm conditions during the first part of the mid Holocene on the ME and SE Greenland shelf in surface and subsurface waters, respectively. Also further south in the Nordic Seas, thermal optimum-like conditions with warm sea surface temperatures prevailed until ca. 6 ka BP (Bauch et al., 2001;core PS1243). 390

Cooling after the HTM (ca. 6.2-4.2 ka BP; ecozone III.)
From 6.2-4.2 ka BP benthic foraminiferal assemblages again start to resemble those of the period prior to 8.2 ka BP, with increased percentage of S. horvarthi and a decrease in the Atlantic water indicator C. neoteretis (Fig. 5). This suggests return to a more stratified water column with sea-ice loaded surface water and Atlantic-sourced subsurface waters. However, in contrast to Ecozone I, neither the benthic and planktic foraminiferal concentrations nor stable isotope data suggest increased bioproductivity. 395 Along West Spitzbergen the inflow of subsurface AW, and thus the RAC weakened after 7 ka BP (Consolaro et al., 2018;core JM10-330;), and a gradual cooling of surface waters started around 7-6.6 ka BP (Müller et al., 2012;core MSM5/5-712), accompanied by increased sea-ice cover (Slubowska-Woldengen et al., 2007;core JM02-440). A sediment core retrieved north from our location (Syring et al., 2020a;Zehnich et al., 2020; core PS93/025) shows seasonally more extended sea-ice cover (derived 400 from decreasing HBI III and increasing IP25 concentrations; Fig. 7) and decreasing primary production after 5.5 ka BP. https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. Accordingly, at the same time, Perner et al. (2015) (core PS641-4) recorded the southward relocation of Polar Front to close to their site in the Foster Bugt area. Giraudeau et al. (2004) (core MD99-2269) and Ran et al. (2006) (core MD99-2275) recorded increased influence of cool, low salinity polar waters of the EGC on the North Iceland Shelf between 6.5-3.5 ka BP and 6.8-5.6 ka BP, respectively. In the Iceland Basin a long-term cooling trend started at between 6.8 and 6.1 ka BP (Orme et al., 2018;Van 405 Nieuwenhove et al., 2018; core DA12-11/2).

Further cooling in the mid Holocene (ca. 4.2-3.2 ka BP; ecozone IV.)
Our record shows clear evidence for significant changes occurring at 4.2 ka BP. The period starts at our location with a sudden rise in the relative abundances of calcareous sea-ice indicator species S. feylingi and in the relative abundances of agglutinated species T. earlandi, T. torquata and S. biformis (Figs. 5 and 6), which are often connected to cold, low-salinity PW (Jennings and 410 Helgadottir, 1994;Korsun and Hald, 2000;Perner et al., 2012;Perner et al., 2015;Wangner et al., 2018). Furthermore, the interval is marked by lowest abundances of warm Atlantic Water species and very low concentrations (under 1 ind./g) of planktic foraminifera. The benthic agglutinated/calcareous foraminiferal ratio increases and the terrestrially derived elements Si, K and Ti show minimal values (Fig. 3), indicating very low meltwater influence in the EGC. Slightly higher benthic δ 18 O values and increasing δ 13 C values after 4.2 ka BP (Fig. 7) point also to general cooling and weak influence of AW. Accordingly, the light δ 18 O 415 spike found in the middle of this interval cannot be attributed to sudden bottom water warming on the basis of the foraminiferal fauna. It is more readily explained by brine rejection during sea ice formation, which can carry the light isotopic signal from the surface to the ocean bottom. It should here be noted that apart from a single sample with increased relative frequencies of the warmer-water agglutinated species A. glomerata (Figs. 6 and A3), there is no evidence of a shorter-lived climate excursion that might be linked to the so-called 4.2ka BP event (Weiss, 2017), only a general change in environment at this time. 420 We suggest that increased PW at the surface of the EGC and reduction of the RAC at subsurface levels led to freshening and reduced stratification of the water column at our site; it may also have experienced (near) perennial sea-ice cover. The reduced strength of the RAC was likely caused by an overall decrease in AW transport from the south with the North Atlantic and West Spitsbergen currents to the eastern Fram Strait, accompanied by surface-water cooling and increased sea-ice coverage after 5.5 ka 425 BP (Hald et al., 2007;core MSM5/5-712;Werner et al., 2013;core PS1878;Telesinski et al., 2014b;core PS1878;Consolaro et al., 2018;core JM10-330;Fig. 7). Harsh conditions and strengthened EGC with permanent sea-ice cover dominated the central Greenland shelf as well (Perner et al., 2015;Kolling et al., 2017;core PS641-4). Further south on the shelf Jennings et al. (2002) and Perner et al. (2016) (core JM96-1206) described increased PW influence in the EGC, glacier advance and iceberg rafting during this period. 430

Neoglaciation (ca. 3.2-0.3 ka BP; ecozone V.)
The period from ca. 3.2 -0.3 ka BP is characterised by a further increase in the relative abundances of sea-ice edge indicator species S. feylingi and the agglutinated species S. difflugiformis. Atlantic-water indicator C. neoteretis is only found in relatively low numbers, although there is some increase in species linked to chilled Atlantic Water (Figs. 5, 6 and 7). Continuously increasing benthic δ 13 C values indicate strong ventilation of the water column (Fig. 7). The results point to cold and unstable conditions with 435 minimum surface water productivity and increasing sea-ice cover at our location. During this period, the EGC was likely strengthened compared to previously, with a thick layer of cold and fresh PW on the surface and recirculated AAW inflow from the Arctic at subsurface levels, as seen in the chilled Atlantic Water group. https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License.

Development of Holocene oceanographic conditions on the NE Greenland shelf
Early Holocene warming of the circum-Arctic region (Kaufmann et al., 2004) led to enhanced Greenland Ice Sheet and glacier melting around the Arctic. During the Last Glacial Maximum (~26-19 ka BP) the Greenland Ice Sheet extended onto the shelf offshore NE Greenland, perhaps even reaching the shelf break (Bennike and Weidick, 2001;Arndt et al., 2017). However, as a 450 consequence of rising air temperatures due to the orbitally-forced Northern Hemisphere summer insolation maximum (reported at approximately 10 ka BP; Andersen et al., 2004;Jansen et al., 2008), the deglaciation of the outer coast and retreat of the ice margin to its present location occurred already between 11.7 and 9.3 ka BP (Larsen et al., 2018); the ice shelf of 79NG rapidly retreated through the fjord between 9.6 and 7.5 ka BP (Syring et al., 2020b). The drastic ice recession of the early Holocene produced an extended meltwater surface layer in the Greenland region prior to 8.6 ka BP . The extensive melting of 455 the Greenland Ice Sheet was strong enough to act as negative feedback to the early Holocene warming, and delayed the HTM with 2 kyr at our location compared to the eastern parts of the Nordic Seas (Blaschek and Renssen, 2013). Thus, between 9.4 and 8.2 ka BP cold, heavily sea ice-loaded surface waters and strongly stratified water column characterized the NE Greenland shelf. The strong freshwater discharge might have also have weakened the recording of the 8.2 ka cooling event at our location, although a short-term increase in terrestrial deposits suggest transport of sediments from land via meltwater or icebergs during this event. This 460 cold event was described in a number of paleoceanographic archives from Greenland and the northern North Atlantic (e.g., Risebrobakken et al., 2003;Ellison et al., 2006;Rasmussen et al., 2007), and it is believed to be connected to the collapse of the Laurentide Ice Sheet and drainage of Lake Agassiz (e.g., Alley et al., 1997;Barber et al., 1999;Hillaire-Marcel et al., 2007;Hoffman et al., 2012). However, in many other records from areas that are mainly influenced by the EGC, WGC or the Baffin-Labrador Current system (Keigwin et al, 2005;Sachs 2007;Seidenkrantz et al., 2013), the 8.2 ka cooling event cannot be 465 recognised; these sites were all characterised by the permanent presence of a low-salinity, cold surface layer at that time, which may also explain the weak signal of this event at our site.
After ca. 8.2 ka BP, our data suggest that increased entrainment of warm Atlantic-sourced water within the RAC changed the water column properties on the NE Greenland shelf. During the subsequent Holocene Thermal Maximum, both subsurface and surface 470 water temperatures increased, and consequently, open water conditions prevailed in the next ca. 2 kyr close to our location. Due to the warming, the floating ice margin of 79NG decreased (Bennike and Weidick, 2001), but on the other hand, the fresh, less dense basal melt water of the glacier isolated the landfast sea-ice cover from the warm subsurface waters, and thus, stabilized the NØIB (Mayer et al., 2000;Syring et al., 2001b). Further south, parallel to the warming of the EGC (Jennings et al., 2011;Müller et al., 2012), the Irminger Current increased its strength, as recorded on the North Icelandic Shelf (Ran et al., 2006;Cabedo-Sanz 475 et al., 2016). After the Thermal Maximum, cooling started on the NE Greenland shelf between 6.2 and 4.2 ka BP (e.g., this study; Andrews et al., 1997;Jennings et al., 2002;Zehnich et al., 2020), with decreased, but still persistent inflow of subsurface AW from the RAC.
The EGC became stronger, with sea-ice loaded surface waters and relatively warm Atlantic-sourced subsurface waters. Coincident 480 with the expansion of the EGC, several studies from the Nordic Seas (e.g., Bauch et al., 2001;Hall et al., 2004;Hald et al., 2007) infer a weakening of the AMOC, increased water column stratification and less ventilated subsurface during this period. In line with a decreased flux of recirculating AW onto the NE Greenland shelf, the Northeast Greenland Ice Margin started to advance from its mid Holocene minimum around 6 ka BP (Larsen et al., 2018). Reduced amounts of warm AW on the inner continental shelf most probably reduced the basal melting within the 79NG fjord and may have contributed to the re-advance of the ice shelf 485 of 79NG seen from around 4.5 ka BP on (Bennike and Weidick, 2001;Syring et al., 2020b).
The Neoglacial cold interval started on the East Greenland shelf approx. 3.5-3.2 ka BP, with increased freshwater forcing from the Arctic Ocean and advance of the Greenland Ice Sheet (this study; Andersen et al., 2004;Jennings et al., 2011). According to model simulations of Renssen et al. (2006), the expansion of sea ice may be associated with a cooling triggered by a negative solar 490 irradiance anomaly, which was amplified through a positive oceanic feedback mechanism. The cooling caused temporary relocation of deep-water formation sites in the Nordic Seas, which was accompanied by a distinct reduction in AMOC strength (Hall et al., 2004). The increase in sea-ice extent stratified the water column and hampered the deep-water formation, leading to additional cooling and more sea ice (Renssen et al., 2006).

Variations of AW and PW entrainment in the EGC during the Holocene and their implications for the general climatic 495 trends in the Nordic Seas
Distinct variability in subsurface water mass properties on the NE Greenland shelf through the Holocene point to broad scale changes in the proportion of RAW/AAW and PW in the EGC, and thus, to a strong covariance with the northward heat transport in the Nordic Seas. The increased entrainment of warm Atlantic-sourced water into the EGC that we document during the HTM, coincides with increased advection of warm AW in the West Spitzbergen Current, driven by increased wind force and/or by 500 stronger thermohaline circulation (Sarthein et al., 2003;Slubowska-Woldengen et al., 2007;Knudsen et al., 2011).
During the mid Holocene, the summer insolation declined at the northern latitudes due to the changes in the Earth´s orbital parameter (Berger, 1978). From 7 ka BP, the summer sea surface temperatures of the Nordic Seas started to decrease (Koç et al., 1993;Andersen et al., 2004). This cooled AW progressed northward carrying less and less heat to the North Atlantic and West 505 Spitsbergen currents (Sarthein et al., 2003;Slubowska-Woldengen et al., 2007;Risebrobakken et al., 2011;Consolaro et al., 2018;Hald et al., 2007), also due to a weakened AMOC (e.g., Bauch et al., 2001;Hall et al., 2004;Hald et al., 2007;Knudsen et al., 2011). In response to the reduced warm water transport through the RAC, the proportion of cooled AW and PW in the EGC increased, which ended the ameliorated conditions on the East Greenland shelf.

510
The northward flow of AW with the North Atlantic and West Spitsbergen currents to the eastern Fram Strait, and consequently the RAW reaching the EGC, arrived to a minimum during the late Holocene, after 4.5 ka BP (Slubowska-Woldengen et al., 2007).
Consequently, the EGC strengthened, carrying a thick layer of fresh PW southward. A pronounced colder period after 4 ka BP is recognized in some (but not all) reconstructions north and south of Iceland, in the Greenland Sea and to the west of Svalbard (Moossen et al., 2015;Orme et al., 2018;Telesinski et al., 2014b;Werner et al., 2016;respectively); in particular at locations 515 influenced by the EGC. This gradual cooling trend in the northern North Atlantic is associated therefore with a continuous transition from a positive towards a negative NAO phase (e.g., Andersen et al., 2004;Orme et al., 2018). Positive NAO phases are https://doi.org/10.5194/cp-2021-59 Preprint. Discussion started: 2 June 2021 c Author(s) 2021. CC BY 4.0 License. accompanied by stronger westerlies carrying moist air over Europe and Siberia, an increased AW inflow through Fram Strait, and warmer temperatures in the Arctic, which lead to a reduction in sea-ice formation. During intervals of a negative NAO these phenomena are reversed (Kwok, 2000;Hurrell and Deser, 2010), and high pressure over Greenland and consequently strengthened 520 northerly winds over East Greenland can lead to strengthened EGC and increased PW transport (e.g., Ionita et al, 2016). The late-Holocene trend towards a stronger and fresher EGC is likely to have increasingly impacted the Subpolar Gyre dynamics. Increased freshwater input to the Labrador Sea may have weakened the Subpolar Gyre circulation by preventing deep convection (Hillarie-Marcel et al., 2001), and consequently reduced the amount of water entrained into the North Atlantic Current .
This, on the other hand, may have forced enhanced northward heat transport via the North Atlantic Current, and may have helped 525 to restart (Thornalley et al., 2009) the previously weakened AMOC (e.g., Cronin et al., 2003;Oppo et al., 2003).

Conclusions
The presented multiproxy study, based on benthic foraminiferal assemblage-, geochemical-and sedimentological analyses of sediment core DA17-NG-ST7-73G allowed us to reconstruct changes in sea surface productivity, subsurface water temperatures, sea-ice conditions, Greenland Ice Sheet melting and thus, in the strength of the EGC over the last ca. 9.4 ka BP on the NE Greenland 530 shelf.
-Between 9.4 and 8.2 ka BP the water column was highly stratified, with cold, heavily sea ice-loaded surface waters and strong influx of warm waters in the subsurface. High surface water productivity suggests that the polar front was close to our location.
-A short-lived peak in terrestrially-derived elements suggesting transport of sediments from land via meltwater or icebergs 535 may be linked to the so-called 8.2 ka BP event.
-The interval from 8.2 to 6.2 ka BP was characterised by the warmest bottom-water conditions of the Holocene on the NE Greenland shelf, with low surface water productivity and strong AW influence (RAC) in the weakened EGC.
-After the HTM the water column started the resemble that of the period prior to 8.2 ka BP. The EGC became stronger, with sea-ice loaded surface waters and relatively warm Atlantic-sourced subsurface waters. The subsurface inflow of 540 warm AW from the RAC decreased, but remained persistent.
-After 4.2 ka BP increased PW at the surface of the EGC and reduction of the RAC at subsurface levels led to freshening and reduced stratification of the water column and (near) perennial sea-ice cover.

-
The period from ca. 3.2 until 0.3 ka BP is characterised by cold and unstable conditions, with minimum surface water productivity, and possible sea-ice cover at our location. During this period, the EGC was likely strengthened, with a thick 545 layer of cold and fresh PW on the surface and strong recirculated AAW inflow from the Arctic at subsurface levels.
-The proportion of PW and AW in the EGC shows a strong covariance with the northward heat transport to the northern North Atlantic. Thus, the cooling trend that characterizes the Holocene after the HTM led to strengthened EGC and increased PW transport, probably due to transition to a negative NAO phase. This stronger and fresher EGC in the late-Holocene is likely to have impacted the Subpolar Gyre circulation and helped to strengthen the AMOC 550  (Knudsen and Seidenkrantz, 1994) Stainforthia fusiformis (Williamson, 1858) Stetsonia horvathi (Green, 1959) Trifarina fluens (Todd in Cushman and McCulloch, 1948) Triloculina tricarinata (d´Orbigny, 1846) Triloculina trihedra (Loebich and Tappan, 1953) Planktic Neogloboquadrina incompta (Cifelli, 1961) Neogloboquadrina pachyderma (Ehrenberg, 1861) Turborotalita quinqueloba (Natland, 1938)

Data availability
Data presented in this manuscript have been uploaded to PANGEA (under review).

Author contribution
Teodora Pados-Dibattista and Marit-Solveig Seidenkrantz developed the research idea. Teodora Pados-Dibattista carried out the 575 sampling, data collection and the data analysis, with the help of Christof Pearce, Henrieka Detlef, Jørgen Brendtsen and Marit-Solvieg Seidenkrantz. Christof Pearce performed the age modelling of the core. Teodora Pados-Dibattista prepared the manuscript with contributions from all co-authors.

Competing interests
Author Marit-Solveig Seidenkrantz is co-editor-in-chief of the journal. 580