Volume 48, Issue 8 e2020GL091602
Research Letter
Free Access

Seasonally Resolved Holocene Sea Ice Variability Inferred From South Pole Ice Core Chemistry

Dominic A. Winski

Corresponding Author

Dominic A. Winski

Climate Change Institute and School of Earth and Climate Sciences, University of Maine, Orono, ME, USA

Correspondence to:

D. A. Winski,

[email protected]

Search for more papers by this author
Erich C. Osterberg

Erich C. Osterberg

Department of Earth Sciences, Dartmouth College, Hanover, NH, USA

Search for more papers by this author
Karl J. Kreutz

Karl J. Kreutz

Climate Change Institute and School of Earth and Climate Sciences, University of Maine, Orono, ME, USA

Search for more papers by this author
David G. Ferris

David G. Ferris

Department of Earth Sciences, Dartmouth College, Hanover, NH, USA

Search for more papers by this author
Jihong Cole-Dai

Jihong Cole-Dai

Department of Chemistry and Biochemistry, South Dakota State University, Brookings, SD, USA

Search for more papers by this author
Zayta Thundercloud

Zayta Thundercloud

Department of Earth Sciences, Dartmouth College, Hanover, NH, USA

Search for more papers by this author
Jiayue Huang

Jiayue Huang

Department of Atmospheric Sciences, University of Washington, Seattle, WA, USA

Search for more papers by this author
Becky Alexander

Becky Alexander

Department of Atmospheric Sciences, University of Washington, Seattle, WA, USA

Search for more papers by this author
Lyatt Jaeglé

Lyatt Jaeglé

Department of Atmospheric Sciences, University of Washington, Seattle, WA, USA

Search for more papers by this author
Joshua A. Kennedy

Joshua A. Kennedy

Department of Chemistry and Biochemistry, South Dakota State University, Brookings, SD, USA

Search for more papers by this author
Carleigh Larrick

Carleigh Larrick

Department of Chemistry and Biochemistry, South Dakota State University, Brookings, SD, USA

Search for more papers by this author
Emma C. Kahle

Emma C. Kahle

Department of Earth and Space Sciences, University of Washington, Seattle, WA, USA

Search for more papers by this author
Eric J. Steig

Eric J. Steig

Department of Earth and Space Sciences, University of Washington, Seattle, WA, USA

Search for more papers by this author
Tyler R. Jones

Tyler R. Jones

Institute of Arctic and Alpine Research, University of Colorado, Boulder, CO, USA

Search for more papers by this author
First published: 06 April 2021
Citations: 8

Abstract

Variability in sea ice is a critical climate feedback, yet the seasonal behavior of Southern Hemisphere sea ice and climate across multiple timescales remains unclear. Here, we develop a seasonally resolved Holocene sea salt record using major ion measurements of the South Pole Ice Core (SPC14). We combine the SPC14 data with the GEOS-Chem chemical transport model to demonstrate that the primary sea salt source switches seasonally from open water (summer) to sea ice (winter), with wintertime variations disproportionately responsible for the centennial to millennial scale structure in the record. We interpret increasing SPC14 and circum-Antarctic Holocene sea salt concentrations, particularly between 8 and 10 ka, as reflecting a period of winter sea ice expansion. Between 5 and 6 ka, an anomalous drop in South Atlantic sector sea salt indicates a temporary sea ice reduction that may be coupled with Northern Hemisphere cooling and associated ocean circulation changes.

Key Points

  • Millennial-scale Holocene variations in sea salt sodium at the South Pole primarily originate from changes in winter sea ice extent

  • Antarctic Holocene sea salt values increased, especially from 8,000 to 10,000 years ago, reflecting a zonally symmetric change in sea ice

  • We infer reduced Atlantic sector winter sea ice from 5,000 to 6,000 years ago, possibly related to North-South Atlantic Ocean heat redistribution

Plain Language Summary

Sea ice variability has a dramatic effect on regional and global climate. Because sea ice extent has such a large summer to winter difference, seasonally specific records of past sea ice conditions are necessary to properly interpret sea ice/climate relationships. Here, we present a sea salt record from the South Pole Ice Core, which represents Southern Hemisphere sea ice changes during the last 11,400 years. We use an atmospheric chemistry model to show that wintertime sea salt in the South Pole Ice Core comes mostly from salty snow originating from sea ice. Wintertime sea ice variations are responsible for most of the long-term variability in the South Pole sea salt record. Ice core data across Antarctica show increasing sea salt concentrations since 11,400 years ago, representing cooling and sea ice expansion, particularly between 8,000 and 10,000 years ago. Between 5,000 and 6,000 years ago, a drop in sea salt indicates an abrupt reduction in sea ice cover in the South Atlantic. Interestingly, paleoclimate data suggest that sea ice was more extensive in the North Atlantic at this time, indicating a linked and opposing sea ice signal in the North and South Atlantic most likely due to changing ocean circulation.

1 Introduction

Sea ice in the Southern Ocean exhibits some of the most pronounced seasonality in the global climate system. During late winter, Antarctica is surrounded by an average of 18.5 million km2 of sea ice, diminishing to 3.1 million km2 during summers (Parkinson, 2014; Shepherd et al., 2018). Despite rising global temperatures, Southern Ocean sea ice had remained remarkably stable until 2016, despite model projections predicting declining Antarctic sea ice (Turner & Comiso, 2017; Turner et al., 2015). Since 2016, the Southern Ocean has exhibited abrupt reductions in sea ice extent (Parkinson, 2019). However, consistent observations of Antarctic sea ice are restricted to the short satellite era (since 1979), which hinders our ability to disentangle anthropogenic changes from natural variability, to understand multi-decadal variability, or to investigate feedbacks with climate more broadly.

The recently drilled South Pole ice core (SPC14; Casey et al., 2014) provides a new opportunity to advance our understanding of Holocene sea ice variability in the Southern Ocean. Recent studies have conclusively demonstrated connections between South Pole and low latitude climate (Clem et al., 2020; Jones et al., 2016; Turner et al., 2020). The breadth and duration of meteorological and aerosol data from the South Pole (e.g., Clem et al., 2020; Lazzara et al., 2012; Sheridan et al., 2016) is unique among ice coring locations. SPC14 has the only ice core chronology on the East Antarctic plateau that allows seasonal partitioning of chemistry records during the Holocene (Winski et al., 2019). Here we use this seasonal Holocene chronology and major ion chemistry data to provide new insights into winter sea ice variability and Southern Hemisphere climate during the Holocene.

2 Study Site

The South Pole has a relatively high modern snow accumulation rate (8 cm water equivalent yr−1; Lilien et al., 2018; Mosley-Thompson et al., 1999; Winski et al., 2019), providing the unique opportunity to develop a seasonally resolved ice core chemistry record from East Antarctica. Surface winds at the South Pole originate primarily from slightly (grid) east of the prime meridian but with contributions from both the Ross and Weddell sectors of Antarctica (Clem et al., 2020; Lazzara et al., 2012; Sheridan et al., 2016). Previous studies have thoroughly characterized the South Pole aerosol climatology in relation to meteorological conditions, finding that sources of marine and temperate air to the South Pole are varied, but aerosols primarily arrive within the mid-to-upper troposphere between grid northwest and southwest, between the Ross and Weddell Seas (Arimoto et al., 2008; Harris, 1992). For both terrestrial and marine aerosols, delivery to the South Pole requires cyclonic activity to penetrate the polar cell, bringing relatively warm, moist and aerosol-laden air to the South Pole in what have been referred to as ‘salt storms' (Bodhaine et al., 1986; Hogan et al., 1997; Parungo et al., 1981; Shaw et al., 1988). In these cases, typically occurring during winter months, a strong influx of lower-latitude air is needed to overcome the strong subsidence over the South Pole and prevailing katabatic winds. Whereas the flux of particulate matter to the South Pole is highest during summer months, sea salt aerosols exhibit winter peaks in both the atmosphere (Bodhaine et al. 19861987; Parungo et al., 1981; Sheridan et al., 2016; Tuncel et al., 1989) and snowpack (Bergin et al., 1998; Cole-Dai & Mosley-Thompson, 1999; Ferris et al., 2011; Legrand & Delmas, 1984; Legrand & Kirchner, 1988; Whitlow et al., 1992). At the South Pole, both wet and dry deposition of ionic impurities occur in roughly equal measure (Legrand & Delmas, 19871988; Legrand & Kirchner, 1988), which is supported by the lack of correlation between accumulation rate and chemical concentrations from previous South Pole ice core studies (e.g., Ferris et al., 2011).

3 Methods

SPC14 was collected between 2014 and 2016 and analyzed using the Dartmouth melter system and ion chromatography (Winski, et al., 2019; supporting information), resulting in timeseries of the following major ions: Cl, NO3, SO42−, Na+, Mg2+, and Ca2+. The SP19 chronology was based on the very consistent seasonal variations in Na+ and Mg2+ ions, constrained with volcanic tie points (Figures 2 and 3 of Winski et al., 2019). Each summer, these ions reached their annual minimum, demarcating years within the SP19 timescale. The definition of annual horizons in the Holocene SP19 timescale based on this signal, as well as its consistency, allow us to extract information about the seasonality of Na+ and Mg2+ chemistry by calculating the minimum and maximum ionic concentrations for each year. The annual minimum Na+ and Mg2+ concentrations occur, by definition, on January 1st during the austral summer, while the annual maximum concentrations occur during winter or early spring (Bergin et al., 1998; Legrand & Kirchner, 1988; Whitlow et al., 1992). We use multiple well-established methods to infer sea salt and non-sea salt (crustal, volcanic, biogenic, etc.) contributions of each ion at South Pole (Figures S1–S3), all of which yield consistent results. These methods also allow us to define Na+ originating from marine sources (ss-Na+), which we use as our primary sea salt metric.

We use output from GEOS-Chem model (Bey et al., 2001) simulations of sea salt aerosol mobilization and transport to constrain the relative source contributions of ss-Na+. Critical to our study, the GEOS-Chem model allows for the partitioning of Na+ on a seasonal basis at the South Pole into two origins: open water, and sublimation of blowing snow over sea ice. The assumptions and description of the model runs used here are discussed by Jaeglé et al. (2011), Huang and Jaeglé (2017), and Huang et al. (2020).

4 Results

4.1 The Holocene SPC14 Sodium Record

The Holocene SPC14 major ion records are presented in Figure 1. Consistent with other studies (e.g., Bertler et al., 2005; Fischer et al., 2007; Wolff et al., 20062010), principal component analysis and ionic ratios (see supporting information) indicate Na+, Mg2+ and Cl are dominated by marine sea salt contributions, Ca2+ is a mix of crustal and sea salt sources, and SO42− likely receives contributions from volcanic emissions, marine biogenic emissions and sea salt. Consistent with prior work (Hogan et al., 1997; Shaw et al., 1988), we interpret a substantial contribution of sea salt during snow events based on the lack of a significant negative relationship between Na+ and accumulation rate (Table S2), as would be expected if South Pole were a purely dry depositional environment (Kreutz et al., 2000).

Details are in the caption following the image

Concentrations of major ions in SPC14 during the Holocene. From top to bottom are plotted Na+, Cl, SO42−, Mg2+, Ca2+, annual accumulation rate (water equivalent) and δ18O. ss-Na+ and nss-Ca2+ are included as dashed lines. All timeseries have undergone low-pass filtering at 200 years for clarity.

Temporally, Na+ and Cl records are predictably similar and exhibit a rising trend, most pronounced between 8 and 10 ka, but interrupted by a temporary drop from 5 and 6 ka. The early Holocene sea salt rise is most clearly evident in ss-Na+, which increases 32% between 8-10 ka. The magnitude of the 5–6 ka event is roughly 2 ppb in Na+ and 8 ppb in Cl, approximately a 20%–25% anomaly for each. This drop is not visible in Ca2+ (or nss-Ca2+), which shows broadly elevated Ca2+ values during the mid-Holocene. Instead, the major feature within the Ca2+ and nss-Ca2+ record is a 4-fold rise in concentration during the last millennium. The 5–6 ka sea salt drop is not present in the SPC14 accumulation or δ18O timeseries, both of which show relatively stable values near 5–6 ka BP (Figure 1). In contrast, the prominent rise in Na+ and Cl during the early Holocene is associated with a concurrent drop in δ18O. High frequency (sub-centennial) variability is high throughout all ionic records, although the white component of the signal does not contain consistently strong frequencies.

4.2 Seasonality of the Sea Salt Signal

The GEOS-Chem model outputs are displayed in Figure 2a, where both sea salt from the open ocean and blowing snow from sea ice exhibit a seasonal cycle with high concentrations during the late winter, consistent with observations (e.g., Sheridan et al., 2016, Figure 2a) and previous modeling studies (Levine et al., 2014). Seasonality is much higher in snow blowing from sea ice than in open ocean sources, such that the primary source of sea salt to the South Pole switches seasonally between open water and sea ice. During the austral summer, when sea salt aerosol concentrations are low, open water constitutes up to 93% of the sea salt delivered to the South Pole (Figure 2). During winters, when concentrations are highest, snow from sea ice contributes up to 55% of the South Pole sea aerosol. Multiplying seasonal SPC14 ss-Na+ values by the GEOS-Chem percent contributions shows a seasonally consistent contribution from open ocean sources with a seasonally variable contribution from sea ice (Figure 2b). This pattern holds across all major ice core sites in Antarctica with source seasonality increasing with proximity to the coast (Figure S4). These results clarify previous work interpreting the significance of sea salt records in Antarctic ice cores by demonstrating that the mechanistic source of ss-Na+ to interior Antarctica changes distinctly by season (Fischer et al., 2007; Iizuka et al., 2008; Wolff et al., 2006).

Details are in the caption following the image

Mean modeled sea salt aerosol surface concentration at the South Pole as a function of month and source (a). Sources are delineated into open ocean (blue) and blowing snow from sea ice (brown). The range in monthly values among the 9 years of output is shown with vertical bars. Observational data of median monthly scattering, interpreted as sea salt aerosol (e.g., Sheridan et al., 2016), is shown for comparison (black dashed line). Seasonal concentrations of SPC14 ss-Na+ from open ocean (purple) and sea ice (cyan) sources are shown in (b). During summer months, lower ss-Na+ concentrations originate almost entirely from open water, whereas during winters, sea ice sources contribute up to 55% of ss-Na+ to the South Pole during months when concentrations are highest.

Because of the very consistent Na+ seasonality over the 11,400-year record, and because this consistency was the basis for assigning annual positions in the SP19 timescale, we calculated maximum (winter) and minimum (summer) ss-Na+ throughout the Holocene for each year (Figure 3). Both seasonal signals display similar patterns including a rise from 8 to 10 ka, a subsequent drop to a local minimum at roughly 5.5 ka and rising concentrations from 2-5 ka. However, millennial-scale features are more apparent in the wintertime Na+ data. At decadal and lower frequencies, correlations between maximum (winter) and mean annual Na+ are high (r = 0.91), exceeding parallel correlations between minimum (summer) and mean annual values (r = 0.77) (Figure S5). The greater relative contributions, higher variance and closer correlation with mean annual values together indicate that wintertime sodium concentration is more representative of the overall annual variability. During the 5–6 ka event, maximum (winter) sodium decreases sharply by nearly a third, while minimum (summer) sodium remains comparatively stable, barely emerging from the overriding centennial scale variability. Given the GEOS-Chem results showing the wintertime importance of sea ice as a Na+ source (Figure 2), a winter Na+ reduction at 5–6 ka (Figure 3) suggests that winter and early spring sea ice conditions may be associated with the millennial scale structure in the SPC14 sea salt record.

Details are in the caption following the image

The Holocene sea salt sodium record in SPC14. Shading indicates a running histogram of mean annual ss-Na+ concentrations. Units indicate the percentage of years within a 200-year window within each 1 ppb size bin. 200-year low pass filtered timeseries of annual mean (black), annual minimum (red) and annual maximum (blue) ss-Na+ are overlaid, indicating mean, summer and winter ss-Na+ concentrations.

5 Discussion

5.1 Two Antarctic Holocene Sea-Salt Signals

In Figure 4b, we show Holocene sea salt records from SPC14, Dronning Maud Land (DML; Fischer et al., 2007), Dome C (Rothlisberger et al., 2002; Wolff et al., 2010), Taylor Dome (Mayewski et al., 1996; Steig et al., 2000), Talos Dome (Mezgec et al., 2017), WAIS-Divide (Markle et al., 2018) and Siple Dome (Kreutz et al., 1997). A principal component analysis reveals two dominant signals (PC1 and PC2) in Antarctic sea-salt variability over the Holocene. PC1 represents over 50% of the centennial-scale variability, showing a prevailing increase in Na+ during the Holocene in all records except Taylor Dome (where Na+ begins increasing at ∼7 ka). This increase is particularly pronounced between 8-10 ka BP in all records but Talos Dome and Taylor Dome. Most of the Antarctic records have high correlations with PC1, especially Dome C (r = 0.83), Siple Dome (r = 0.87), SPC14 (r = 0.82), and WAIS-Divide (r = 0.89). This consistency on the broadest scale suggests an Antarctic-wide phenomenon occurring throughout the Holocene.

Details are in the caption following the image

Comparison between Holocene Antarctic sodium records. Panels A and B show the locations of global (a) and Antarctic (b) sites. PCs 1 and 2 are from ice core sodium at Taylor Dome (Steig et al., 2000), Talos Dome (Mezgec et al., 2017), Siple Dome (Kreutz et al., 1997), Dome C (Rothlisberger et al., 2002), DML (Fischer et al., 2007), WAIS-Divide (Markle et al., 2018) and SPC14 (red). The first and second principal components of these seven sodium records are shown in the table (eigenvector components >0.20 in bold) and in Panel C (black). PC1 reflects increasing sea ice, especially during the early Holocene (gray shading) as supported by increasing sea ice presence (cyan, Divine et al., 2010), sea surface temperature (maroon, Shevenell et al., 2011), and negative ss-Na+− δ18O correlations (purple). PC2 primarily reflects the shared signal between DML and SPC14 (blue and red). Between 5 and 6 ka BP (gray shading), ss-Na+ is decoupled from δ18O (purple) but is consistent with GISP2 Na+ (orange, Mayewski et al., 1994), North Atlantic IRD (gray, Bond et al., 2001) and Bermuda Rise 231 Pa/230Th (green, McManus et al., 2004).

In contrast, PC2 is primarily weighted toward DML and SPC14. PC2 exhibits a drop between 5 and 6 ka, as seen exclusively in the DML (Fischer et al., 2007) and SPC14 records. At DML, ss-Na+ drops approximately 50% at 6 ka, compared with a concomitant ∼20% drop in ss-Na+ at SPC14. DML represents an overwhelmingly Atlantic source region with ∼80% of 4-days trajectories originating in the Atlantic during winters (Reijmer et al., 2002). In contrast, only ∼40% of July airmasses arrive at the South Pole from the Atlantic sector (Harris et al., 1992), with the majority of the remaining trajectories passing over W. Antarctica. Thus, we interpret the 5–6 ka anomaly as representing a regional phenomenon specific to the Atlantic sector sea salt source region common to DML and SPC14, with the more muted SPC14 signal representing dilution of the signal by contributions from the Pacific sector. Supporting this hypothesis are correlations between South Pole aerosol data (σsp – 550 nm; Bodhaine et al., 19861987) and ERA5 (Hersbach et al., 2020) sea ice reanalysis fields showing that South Pole atmospheric sea salt is positively correlated with sea ice concentration within the southeast Atlantic Ocean (r = 0.56, p < 0.05; Figure S6).

5.2 Interpretation of the Holocene Secular ss-Na+ Rise

Based on the combination of our seasonal SPC14 ss-Na+ analysis and the GEOS-Chem results, we hypothesize that wintertime sea ice variability is the primary cause of the Antarctic-wide secular sodium rise during the Holocene (PC1), as well as the 5–6 ka sodium anomaly restricted to the Atlantic sector (PC2). This is consistent with the interpretation of the DML and Dome Fuji records as primarily sensitive to sea ice extent (Iizuka et al., 2008; Stenni et al., 2010). The large increase in SPC14 ss-Na+ between 6-10 ka is consistent with sea ice presence records (Xiao et al., 2016), particularly from TN057-13 in the southeast Atlantic (Bianchi & Gersonde 2004; Divine et al., 2010, Figures 4a and 4c). Holocene cooling is also apparent in reconstructed sea surface temperature TN057-13 (Anderson et al., 2009) as well as from JPC-10 and ODP-1098, located west of the Antarctic Peninsula (Etourneau et al., 2013; Shevenell et al., 2011; Figures 4a and 4c). A long-term increase in Holocene sea ice indicates cooler late-Holocene conditions attributed to decreasing insolation amplified by internal mechanisms, as has been inferred by Antarctic-wide water isotope comparisons (Masson et al., 2000) and sea ice records (Divine et al., 2010; Etourneau et al., 2013; Hodell et al., 2001). A larger-than-modern Ross Ice Shelf during the early Holocene (Yokoyama et al., 2016) may have covered either some of the sea ice sodium sources or the polynyas that generate much of the sea ice in the Ross Sea (Comiso et al., 2011). This may account for the absence of steeply rising ssNa+ between 8 and 10 ka at Taylor and Talos Domes.

In addition to changes in sea ice, ss-Na+ variability on millennial timescales has been attributed to the distillation (e.g., rainout) of impurities under varying temperature (and therefore saturation vapor pressure) conditions (Markle et al., 2018; Yung et al., 1996). Analysis of the SPC14 δ18O record (Steig et al., 2021) provides insight to this mechanism, as a major shift in atmospheric distillation would be apparent (Markle et al., 2018). We calculate running 41-year correlations between ss-Na+ and δ18O throughout the Holocene, finding a variable but typically significant relationship with r-values averaging −0.45 (p < 0.01), and with the strongest relationship between 8 and 10 ka (r = −0.79; Figure 4c). The magnitude of this relationship is approximately 1.7 ppb of ss-Na+ for each permil δ18O. At WAIS-Divide the ss-Na+ - δ18O relationship developed for the last 68 ka is expectedly steeper (∼4 ppb ss-Na+ for each permil δ18O) and more exponential given the larger glacial-interglacial changes (Markle et al., 2018). In this interpretive framework, the secular Holocene increase in ss-Na+ could indicate cooling Holocene conditions with less ss-Na+ rainout enroute to the ice core sites. Thus, this is consistent with enhanced sea ice, and both increases in sea ice and decreases in rainout may have contributed to the secular Holocene sodium rise.

5.3 Interpretation of the Abrupt 5–6 ka ss-Na+ Decline

Our interpretation of the 5–6 ka ss-Na+ decline (PC2) as an abrupt reduction in Atlantic-sector wintertime sea ice extent is consistent with previous studies (Fischer et al., 2007; Stenni et al., 2010) interpreting the DML sodium record as reflecting sea ice changes. Records from marine cores TPC286 (62° S) and TPC078 (56° S) are near the inferred sodium source region for the South Pole during the instrumental period (Figure S6) and show reduced winter sea ice during the mid-Holocene relative to the early and late Holocene (Collins et al., 20122013). A mid-Holocene sea ice reduction is not observed in marine records further to the north (Bianchi & Gersonde 2004; Divine et al., 2010; Nielsen et al., 2004; Xiao et al., 2016), from the southern Indian Ocean (Mashiotta et al., 1999; Orme et al., 2020), nor from the Antarctic Peninsula (Domack et al., 2001), further emphasizing the geographic specificity of the mid-Holocene ss-Na+ anomaly. In contrast, a progressive sea ice increase and/or sea surface cooling is widely observed (Divine et al., 2010; Etourneau et al., 2013; Hodell et al., 2001; Xiao et al., 2016), consistent with the secular Antarctic-wide Na+ increase (PC1).

Importantly, we see no evidence in the SPC14 δ18O for a change in rainout that could account for the 5–6 ka event. A plot of running correlations between ss-Na+ and δ18O reveals that this association disappears during the 5–6 ka event (Figure 4c). This is also apparent in Figure 1 wherein clear drops in Na+ and Cl are visible while no major changes occur in δ18O or accumulation rate. Given the transport of ss-Na+ to the South Pole through the meridional flux of ‘salt storms' (e.g., Shaw et al., 1988), and the prevalent association between ss-Na+ and δ18O (Figure 4c), it is difficult to explain the mid-Holocene ss-Na+ drop through a rainout mechanism. Furthermore, if a reduction in meridional transport were responsible, it is probable that SPC14, being the more distal site, would see a larger ss-Na+ drop compared with EDML; however, the opposite is observed.

5.4 A Global 5–6 ka Signal?

Sodium concentrations from the GISP2 and GRIP ice cores in central Greenland are elevated from 5-6 ka (Mayewski et al., 19941997, Figures 4a and 4c), interpreted as reflecting very cold conditions in high northern latitudes (De Angelis et al., 1997; O'Brien et al., 1995). In Renland, δ18O amplitudes were elevated at this time, reflecting variability in Atlantic Meridional Overturning Circulation (AMOC) and associated sea ice cover (Hughes et al., 2020). The Greenland and Antarctic records are consistent with the hypothesis of reduced North Atlantic Deep Water formation leading to a reduction in AMOC (Bond et al., 1997). This hypothesis invokes a shift in heat content from the North Atlantic to the South Atlantic near 6 ka, analogous to the bipolar see-saw observed in ice core records spanning the last glacial period (Blunier & Brook, 2001; Broecker, 1998). Marine studies (Bianchi & McCave 1999; Hansen & Østerhus, 2000; Thornalley et al., 2013) show more vigorous AMOC strength near Iceland and Scotland between 7 and 8 ka and then decreasing strength from 5 to 7 ka. At the Bermuda Rise, McManus et al. (2004) find that the only significant anomaly in Holocene AMOC strength was a weakening between approximately 5 and 6 ka. A hypothetical AMOC weakening between 5 and 6 ka may explain both the widespread cold anomalies in the Northern Hemisphere (Mayewski et al., 2004; Wanner et al., 20112015) as well the inferred reduction in Weddell Sea ice inferred from ss-Na+ in SPC14 and DML.

6 Conclusions

Results from the GEOS-Chem model (Huang & Jaeglé, 2017) indicate that the SPC14 sea salt record is primarily sensitive to changes in wintertime sea ice extent, consistent with the strong seasonality of the ice core chemistry and prior interpretations. Sea salt levels at the South Pole, and at most Antarctic ice core sites, increased throughout most of the Holocene and particularly between 8-10 ka (Figure 4), interpreted as an Antarctic-wide wintertime sea ice increase. Between 5 and 6 ka, an anomalous drop in sea salt is recorded at SPC14 and DML, but is not present in the δ18O record (Steig et al., 2021) or other Antarctic sea salt records. We hypothesize that this drop reflects an abrupt decline in wintertime sea ice restricted to the South Atlantic/Weddell region. Evidence from Southern (e.g., Collins et al., 2012) and Northern Hemisphere (e.g., Bond et al., 2001; Mayewski et al., 1994; McManus et al., 2004) proxy records supports our hypothesis and indicates a possible bipolar influence of weakened AMOC at that time (Bond et al., 1997).

Acknowledgments

This research was funded by US National Science Foundation grants 1443336 (Osterberg), 1443397 (Kreutz) 1443663 (Cole-Dai), 1702266 (Alexander), 1443105, 1141839 (Steig) and NASA grant NNH18ZDA001N (Jaeglé). We thank Xiahong Feng, Sarah Das, Robert Hawley, and Eric Posmentier for their helpful suggestions; Mark Twickler, Joseph Souney and the SPICEcore Science Coordination Office for administering the project; the U.S. Ice Drilling Program for support activities through NSF Cooperative Agreement 1836328; the 109th New York Air National Guard for airlift in Antarctica; the field team who helped collect the core; the members of South Pole station who facilitated the field operations; the National Science Foundation Ice Core Facility for ice core processing; and the many student researchers who produced the data underlying this research.

    Data Availability Statement

    The SPC14 major ion record is accessible at https://www.usap-dc.org/view/dataset/601399.