High-Resolution Record of Late Glacial and Deglacial Sea Ice Changes In

Earth and Planetary Science Letters 403 (2014) 446–455

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

High-resolution record of late glacial and deglacial sea ice changes in Fram Strait corroborates ice–ocean interactions during abrupt climate shifts ∗ Juliane Müller , Ruediger Stein

Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Am Alten Hafen 26, 27568 Bremerhaven, Germany a r t i c l e i n f o a b s t r a c t

Article history: The transition from last glacial to deglacial and subsequently to modern interglacial climate conditions Received 18 December 2013 was accompanied by abrupt shifts in the palaeoceanographic setting in the subpolar North Atlantic. Received in revised form 7 July 2014 Knowledge about the role that sea ice coverage played during these rapid climate reversals is limited Accepted 14 July 2014 since most marine sediment cores from the higher latitudes provide only a coarse temporal resolution Available online 21 August 2014 and often poorly preserved microfossils. Here we present a highly resolved reconstruction of sea ice Editor: J. Lynch-Stieglitz conditions that characterised the eastern Fram Strait – a key area for water mass exchange between Keywords: the Arctic Ocean and the North Atlantic – for the past 30 ka BP. This reconstruction is based on the sea ice distribution of the sea ice biomarker IP25 and phytoplankton derived biomarkers in a sediment core subpolar North Atlantic from the continental slope of western Svalbard. During the late glacial (30 ka to 19 ka BP), recurrent LGM advances and retreats of sea ice characterised the study area and point to a hitherto less considered Heinrich Event 1 oceanic (and/or atmospheric) variability. Along-lasting perennial sea ice coverage in eastern Fram Strait Younger Dryas persisted only at the very end of the Last Glacial Maximum (i.e. from 19.2 to 17.6 ka BP) and was AMOC abruptly reduced at the onset of Heinrich Event 1 – coincident with or possibly even inducing the collapse of the Atlantic Meridional Overturning Circulation (AMOC). Further maximum sea ice conditions prevailed during the Younger Dryas cooling event and support the assumption of an AMOC reduction due to increased formation and export of Arctic sea ice through Fram Strait. Asigniﬁcant retreat of sea ice and sea surface warming are observed for the Early Holocene. © 2014 Elsevier B.V. All rights reserved.

1. Introduction MARGO Project, 2009). Further, the rate of North Atlantic Deepwa- ter (NADW) formation in the Greenland–Iceland–Norwegian (GIN) There exists general consensus that the subpolar North At- Seas and the strength of the Atlantic Meridional Overturning Cir- lantic up to the Fram Strait – the only deepwater connection to culation (AMOC) are research targets for understanding the often the Arctic Ocean and the main outlet for sea ice (Aagaard, 1982; abrupt climate shifts that characterised the transition from last Aagaard and Carmack, 1989; Rudels and Quadfasel, 1991)– ex- glacial to interglacial conditions (Rahmstorf, 2002; McManus et perienced significant fluctuations in the oceanic circulation during al., 2004). Most of these reconstructions of AMOC strength are the late glacial and deglacial and that these fluctuations are as- based upon stable isotope data obtained from foraminifera and sociated with global climate changes. Various proxy and model provide valuable information on the ventilation or stratification of (time-slice) reconstructions of the Last Glacial Maximum (LGM; the water column (e.g. Bauch et al., 2001; Benway et al., 2010; commonly defined as the period between 23 ka and 19 ka BP) Waelbroeck et al., 2011). led to the meanwhile widely approved notion of a seasonally ice- Only very limited information is available about the variable free eastern corridor in the northern North Atlantic permitting sea ice extent and thus its considerable effect on the water column moisture release required for the last glacial build-up of Northern structure in the subpolar North Atlantic during the late glacial and Hemisphere ice-sheets (Hebbeln et al., 1994; Sarnthein et al., 2003; deglacial remains unclear. Microfossil-based sea ice reconstruc- Schulz and Paul, 2004; Meland et al., 2005; de Vernal et al., 2006; tions mainly build upon assemblage analyses of diatoms (Koç et al., 1993; Ran et al., 2006), foraminifera (Sarnthein et al., 2003; * Corresponding author. Tel.: +49 331 288 2224. Slubowska-Woldengen et al., 2008)or dinoflagellate cysts (de Ver- E-mail address: [email protected] (J. Müller). nal et al., 2013) and provide important palaeoenvironmental http://dx.doi.org/10.1016/j.epsl.2014.07.016 0012-821X/© 2014 Elsevier B.V. All rights reserved. J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455 447

Fig. 1. Study area with core sites MSM5/5-712-2 (yellow circle) and PS2837-5 (orange circle). The major ocean currents (NC = Norwegian Current, WSC = West Spitsbergen Current; EGC = East Greenland Current) and the modern average summer sea ice extent (white dotted line) are indicated. Also included is the modelled LGM sea ice distribution by Stärz et al. (2012) with MSM5/5-712-2 and PS2837-5 core sites. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) information. The often lowered preservation potential of sea ice assessment of palaeo sea ice conditions. Given that other factors associated diatoms and foraminifera (due to silica or carbonate than sea ice might control phytoplankton productivity, and this dissolution) in the Arctic Ocean and subpolar North Atlantic (e.g. has been recently shown for the Labrador Sea (Weckström et al., Scott and Vilks, 1991; Schlüter and Sauter, 2000; Zamelczyk et al., 2013), caution should be exercised when interpreting PIP25 values. 2014), however, often hamper the reconstruction of the spatial and The Arctic-wide applicability of the PIP25 index and the accuracy temporal distribution of sea ice in the past. The sea ice biomarker of the quantitative sea ice estimates derived from this approach proxy IP25, ahighly branched isoprenoid (HBI) monoene with 25 thus require further evaluation. Amore recent derivative is the carbon atoms (C25) derived from sea ice diatoms (Belt et al., 2007; so-called DIP25 index (Cabedo-Sanz et al., 2013), which builds Brown et al., 2014), has been found stable within Arctic Ocean upon the observation by Rowland et al. (2001) that the degree sediments for at least the entire Quaternary (Stein and Fahl, of unsaturation of C25-HBI molecules produced by a certain di- 2013) and thus enables the assessment of sea ice changes in atom species decreases with decreasing algal growth temperature. the High Northern Latitudes even during the distant past. De- Fahl and Stein (2012) and Stein et al. (2012) thus suggested and spite the novelty of this proxy and a certain lack of knowledge demonstrated that the ratio of sedimentary C25-HBI diene to IP25 about how environmental factors (salinity, nutrient and light avail- contents could serve as an indicator for palaeo sea surface temper- ability) might affect the productivity of IP25 synthesising sea ice atures (SSTs) in the Arctic realm. This approach has been supported diatoms, IP25 has been applied frequently for reliable sea ice re- by palaeoenvironmental reconstructions (Stein and Fahl, 2012; constructions covering different areas of the Arctic Ocean (for Cabedo-Sanz et al., 2013) and surface sediment analyses (Xiao et recent reviews see Stein et al., 2012; Belt and Müller, 2013). With al., 2013). Further investigations into the reliability of the DIP25 as the PIP25 index, a recent approach combining IP25 and phyto- well as the PIP25 index, however, are required to fully exploit the plankton biomarker data to distinguish between different sea ice capacity of these approaches (we refer to Belt and Müller, 2013 for conditions, a promising step towards semi-quantitative sea ice re- an overview of potential limitations of both indices). constructions, in particular for the Fram Strait area, has been made One of the first IP25-based sea ice reconstructions has been (Müller et al., 2011). The consideration of phytoplankton-derived conducted on a sediment core PS2837-5 from the Yermak Plateau biomarkers (e.g. brassicasterol or dinosterol; Volkman, 2006) – (Fig. 1), where the authors found significant variations in the sea indicative of open-water conditions – alongside IP25 proves very ice cover in northern Fram Strait during the past 30 ka BP (Müller helpful to avoid a misleading interpretation of the absence or et al., 2009). Unfortunately, this record does not provide a suitably minimum concentration of IP25, which may result from either high temporal resolution to document the transitions from cold to a lack of sea ice or, on the contrary, a permanent and thick warm (and vice versa) climate intervals in much detail. sea ice cover reducing the light penetration through the ice and Herein we present a new high-resolution sea ice record based thus inhibiting ice algae growth (Horner and Schrader, 1982; on biomarker data of a sediment core from the eastern Fram Müller et al., 2009). The PIP25 index (the ratio of IP25 against the Strait. By considering both IP25 and the phytoplankton derived sum of IP25 and a phytoplankton biomarker) basically constitutes biomarker lipids brassicasterol and dinosterol (Kanazawa et al., a proxy for the intensity of a sea ice cover during spring and sum- 1971; Volkman, 2006)we gain valuable insight into the palaeo- mer with high PIP25 values (> 0.7) reflecting an extended and ceanographic and environmental development in this climate sen- low PIP25 values (< 0.5) reflecting a reduced ice cover (Müller sitive area between 30 ka and 9ka BP. The core site at the western et al., 2011). Müller et al. (2011, 2012), however, also advise continental margin of Svalbard (Fig. 1) is ideally suited to record that rather than merely calculating the PIP25 index, the individ- even short-term shifts in the North Atlantic oceanic and/or at- ual IP25 and phytoplankton biomarker concentrations and – if mospheric circulation system and we find evidence for hitherto available – further proxy data should be considered for a proper unknown rapid sea ice changes during the late glacial. Finally, 448 J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455

Table 1 Equations for the calculation of IP25-based proxies to estimate sea ice coverage (PBIP25, PDIP25) and relative sea surface temperatures (DIP25). Proxy Equation References

PBIP25 (sea ice cover) PBIP25 = IP25/(IP25 + (brassicasterol × c)); Müller et al., 2011, 2012 c = mean IP25/mean brassicasterol; c = 0.0127 PDIP25 (sea ice cover) PDIP25 = IP25/(IP25 + (dinosterol × c)); Müller et al., 2011, 2012 c = mean IP25/mean dinosterol; c = 0.0322 DIP25 (relative SST) DIP25 = C25-HBI diene/IP25 Fahl and Stein, 2012; Cabedo-Sanz et al., 2013 through the integration of proxy and model data on deglacial concentrations have been normalized to the TOC contents of the AMOC and SST variability we are able to put our findings in a more samples. The PIP25 index, in the following also termed PBIP25 and comprehensive palaeoceanographic context. PDIP25 index referring to the use of the phytoplankton biomarkers brassicasterol (B) and dinosterol (D), respectively, has been cal- 2. Regional setting culated according to Müller et al. (2011, 2012). We note that in ◦ ◦ a few sediment samples from the late glacial interval concentra- The core site MSM5/5-712-2 (78 54.94 N, 6 46.03 E; 1487 m tions of brassicasterol and dinosterol are close to the detection water depth) at the western continental slope of Svalbard is limit (here < 4μg/gOC). The identification of the trace compound located close to the current summer sea ice margin (Fig. 1). IP25 is not possible in these samples. Regarding previous observa- This eastern part of Fram Strait remains predominantly ice-free tions of contemporaneous minimum phytoplankton biomarker and throughout the year due to the advection of warm Atlantic water minimum IP25 contents in samples from areas (or intervals) char- via the West Spitsbergen Current (WSC), which is the northward acterised by a permanent sea ice coverage (Stein and Fahl, 2013; extension of the Gulf Stream and the Norwegian Current (NC), Xiao et al., 2013) and to avoid ambiguous PIP25 values of zero respectively (Aagaard, 1982). In contrast, the western Fram Strait reflecting ice-free conditions, we appointed a PIP25 index of 1 to experiences a continuous discharge of polar water and sea ice via these samples. The calculation of the DIP25 index (i.e. the ratio of the East Greenland Current (EGC; Aagaard and Coachman, 1968; the C25-HBI diene to IP25) is not possible, where IP25 is absent. Ac- Martin and Wadhams, 1999). This export of sea ice is mainly con- cordingly, DIP25 values of sediment sections where both C25-HBI trolled by Arctic Ocean and North Atlantic atmospheric high and isomers are absent or just at the detection limit have been set 0 low pressure systems governing wind strength and thus ice-drift thus reflecting coolest sea surface conditions. The equations used patterns (Jung and Hilmer, 2001; Kwok et al., 2004). Further, varia- for the calculation of each index are provided in Table 1. Concen- tions in the oceanic heat transport towards the Arctic Ocean affect trations of dinosterol and the C25-HBI diene are provided in the the SSTs and thus sea ice conditions in Fram Strait (Zhang et al., supplement (S3). 1998; Spielhagen et al., 2011). The successive cooling and densi- All data is available via the PANGAEA online data repository fication of surface waters carried by the NC and the WSC drives (http://dx.doi.org/10.1594/PANGAEA.833668). the thermohaline convection and NADW formation in the GIN Seas and hence impacts on the AMOC (Aagaard et al., 1985; Rudels 3.1. Core chronology and Quadfasel, 1991; Rudels, 1996). In this regard, the distribution of sea ice insulating the ocean surface from the atmosphere The age model of the lower section (covering the deglacial and (and also by releasing either freshwater during melt or dense late glacial time interval) of core MSM5/5-712-2 is based upon brines during freezing), plays an important role as it directly af- 8 AMS 14C ages (Table 2) obtained from tests of the planktic fects the temperature and density of underlying water masses and foraminifer Neogloboquadrina pachyderma (s) and magnetic suscep- thus the thermohaline convection in the study area (Rudels, 1996; tibility and TOC correlations with adjacent sediment cores (see Mauritzen and Häkkinen, 1997). Supplement S1). Following Zamelczyk et al. (2014), we correlated the magnetic susceptibility values of core MSM5/5-712-2 with the 3. Material and methods Western Svalbard slope MS stack by Jessen et al. (2010) and the susceptibility values of core PS2837-5 from the Yermak Plateau Sediment core MSM5/5-712-2 was obtained during a Maria S. (Niessen et al., 1997; see S1). Further, by means of correlating Merian cruise in 2007 (Budéus, 2007) and has been kept frozen the TOC contents of core MSM5/5-712-2 and core PS2837-5 from − ◦ ( 30 C) until further treatment. Samples for organic geochemi- the Yermak Plateau (S1; Birgel and Hass, 2004) we identified sev- cal analyses have been freeze-dried, homogenised and were stored eral tie points and selected three AMS 14C ages of core PS2837-5 in brown glass vials. After the removal of carbonates by adding (Nørgaard-Pedersen et al., 2003; Table S2) that we adopted to hydrochloric acid, total organic carbon (TOC) contents were deter- improve the herein presented age model of core MSM5/5-712-2 mined using a CS-125 (Leco) carbon–sulfur analyser. For biomarker (Fig. 2). analyses, ca. 3–4 g of sediment were extracted by means of an A marine reservoir correction of 440 yr (Mangerud and Gullik- Accelerated Solvent Extractor (DIONEX ASE 200) using a mix- sen, 1975) has been applied during the conversion of radiocarbon ture of dichloromethane and methanol (2:1 v/v). To permit quan- ages into calendar years before present (cal. years BP) using the tification of biomarkers, 7-hexylnonadecane and cholesterol-d6 Fairbanks radiocarbon calibration (Fairbanks et al., 2005). The core (cholest-5-en-3β-ol-D6) were added as internal standards prior chronology builds on a linear interpolation between calibrated ages to this step. Following Müller et al. (2011, 2012), hydrocarbons (Fig. 2). and sterols were separated by means of open-column (SiO2) chromatography using n-hexane (5 ml) and methylacetate:n-hexane 4. Results and discussion (20:80 v/v; 6 ml), respectively. Prior to gas chromatography and gas chromatography–mass spectrometry (GC–MS) analyses, sterol 4.1. The late glacial and Last Glacial Maximum (LGM) ◦ fractions were silylated with 300 μl BSTFA (60 C, 2h). Com- pound analyses of both fractions and the further identification The oldest interval of the record (30–26 ka BP) is characterised and quantification of individual biomarker lipids were performed by a significant drop from previously high to minimum concen- as described in detail by Müller et al. (2012). Absolute biomarker trations of both the sea ice proxy IP25 and the phytoplankton J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455 449

Table 2 AMS 14C ages for Maria S. Merian core MSM5/5-712-2 and additional AMS 14C ages adopted from Polarstern core PS2837-5 (bold) and from correlation tie points (TP 7, TP 8, TP 9; italic) from the Western Svalbard Slope magnetic susceptibility stack published by Jessen et al. (2010). Amarine reservoir age of 440 yr has been assumed according to Mangerud and Gulliksen (1975). Core depth AMS 14C age Calibrated age BP (1σ ) Lab. reference Literature reference 214.5 cm 7920 ± 50 8733 ± 113 Poz-30723 Aagaard-Sørensen et al., 2014 280.5 cm 8780 ± 50 9784 ± 119 KIA 37423 Aagaard-Sørensen et al., 2014 322.5 cm 9140 ± 50 10 269 ± 52 Poz-30725 Aagaard-Sørensen et al., 2014 350 cm 10 500 ± 50 12 450 ± 78 KIA 7571 Nørgaard-Pedersen et al., 2003 428.5 cm 11 920 ± 60 13 749 ± 62 Poz-30726 Aagaard-Sørensen et al., 2014 480 cm 12 215 ± 60 14 002 ± 84 KIA7572 Nørgaard-Pedersen et al., 2003 660 cm 12 500 ± 70 14 480 ± 161 KIA10864 Nørgaard-Pedersen et al., 2003 687.5 cm 14 210 ± 80 16 663 ± 157 Poz-30727 Zamelczyk et al., 2014 716–716.5 cm 16 760 ± 120 19 907 ± 158 Poz-38427 Zamelczyk et al., 2014 762.25 cm 18 860 ± 150 22 445 ± 151 Poz-30728 Zamelczyk et al., 2014 801 cm 19 710 ± 130 23 550 ± 184 – Jessen et al., 2010 842 cm 20 140 ± 130 24 047 ± 154 – Jessen et al., 2010 876 cm 22 900 ± 200 27 494 ± 265 – Jessen et al., 2010 882.5–883 cm 24 036 ± 186 28 774 ± 245 Poz-30729 Zamelczyk et al., 2014

may relate to different ecological requirements (e.g. temperature, salinity, nutrient availability, different blooming periods) of brassicasterol and dinosterol producing algae. Recently, apotential sea ice origin of brassicasterol (but not dinosterol) has been discussed by Belt et al. (2013) and needs to be considered. Given the over- 2 all similarity between PBIP25 and PDIP25 values (R = 0.8; Figs. 3 and S4) and that no correlation exists between IP25 and brassicasterol concentrations (R2 = 0.01; S4), we suggest that most of the brassicasterol identified in this core has its source in open- water phytoplankton. However, the moderate abundance of IP25 and high brassicasterol (and dinosterol) contents between 26 ka and 23 ka BP, lead us to conclude that a retreat of the previously extended sea ice cover towards the north (or west) per- mitted spring sea ice algal and summer phytoplankton blooms at the western continental slope of Svalbard. Similarly, Müller et al. (2009) observed a reduced sea ice cover between 27 ka and 24 ka BP further to the north at the western Yermak Plateau. The peak brassicasterol contents at about 24 ka BP (Fig. 3) concur with the maximum abundance of the subpolar foraminifer T. quinqueloba in this core (Zamelczyk et al., 2014), which also points to a temporary retreat of sea ice. Coincident with this reduction in sea ice, a short but intensive discharge of glacial meltwater is recorded at various sites along the western continental slope of Svalbard, which caused a high accumulation of (terrigenous) organic matter (S1) Fig. 2. Age model and linear sedimentation rates of core MSM5/5-712-2. Circles refer and ice rafted detritus (Knies and Stein, 1998; Vogt et al., 2001; to reservoir corrected and calibrated 14C ages with respective error bars. Yellow Jessen et al., 2010). This melting event and the seasonally ice- circles are 14C ages adopted from sediment core PS2837-5; red circles are 14C ages adopted from the Western Svalbard slope MS stack by Jessen et al. (2010). (For free sea surface conditions – a basic requirement for late glacial interpretation of the references to color in this figure legend, the reader is referred continental ice-sheet growth – possibly resulted from a stronger to the web version of this article.) advection of warm Atlantic water along the eastern corridor of the Nordic Seas up to Fram Strait as previously proposed by Hebbeln biomarker brassicasterol (Fig. 3) indicative of unfavourable living et al. (1994), Dokken and Hald (1996) or Andersen et al. (1996). conditions for sea ice diatoms as well as for phytoplankton prefer- Significant short-term fluctuations in the concentrations of both ring ice-free water. Previously, such simultaneous minima in both IP25 and brassicasterol point to rapid changes in the palaeo- biomarker contents have been interpreted to reflect a permanent ceanographic setting in eastern Fram Strait between 23.2 ka and and very thick sea ice cover that inhibits any primary (sea ice al- 19.2kaBP (i.e. during the LGM; Fig. 3). Peak IP25 and high bras- gae and phytoplankton) productivity due to an insufficient light sicasterol concentrations at about 23.2 ka, 22.6 ka, 21.5 ka and at availability (Müller et al., 2009, 2011; Stein and Fahl, 2013). Fur- about 20 ka BP are punctuated by intervals when both biomarker ther, no or only little organic matter (i.e. IP25) is released to the contents drop to minimum or zero values (Fig. 3) as seen before water column and thus towards the seafloor when the ice is not between 28.3 ka and 26 ka BP. These fluctuations possibly refer to sufficiently melted during summer. The PBIP25 (and PDIP25) val- rapid advances (restraining algal growth) and retreats (promoting ues accordingly reach maximum values of 1 (Fig. 3) suggesting a algal growth) of the sea ice cover at the core site and we sug- permanent sea ice cover between 28.3 ka and 26 ka BP, whereas gest that these shifts between a perennial and a reduced sea DIP25 values of 0 indicate lowest SSTs. A breakup of this peren- ice cover result from short-term perturbations in the advection nial sea ice cover at 26 ka BP is inferred from rapidly increasing of warm Atlantic water towards the north and/or rapidly chang- IP25 and brassicasterol concentrations (Fig. 3). This amelioration of ing atmospheric circulation patterns controlling Arctic Ocean sea sea surface conditions is also mirrored by a significant decrease ice export through Fram Strait. Recurrent sea ice advances (and in PIP25 (∼ 0.4–0.7) and increasing DIP25 values (3–6). The slight SST cooling) and retreats (and SST warming) are also depicted by offset between PBIP25 and somewhat higher PDIP25 values (Fig. 3) oscillating PBIP25, PDIP25 and DIP25 values (Fig. 3). Interestingly, 450 J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455

Fig. 3. Biomarker concentrations, PIP25 and DIP25 indices of sediment core MSM5/5-712-2 and insolation values (Laskar et al., 2004). Light blue vertical shadings indicate 14 cooling events with extended sea ice cover. Red triangles refer to AMS C datings. Red circles in the PIP25 (and DIP25) record indicate where a value of 1 (0) has been appointed due to zero or minimum biomarker concentrations. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) these fluctuations occur during an interval that is characterised The following period between 19.2 ka and 17.6 ka BP, finally by the temporally confined occurrence of the infaunal foraminifer seems to reflect the latest phase of the LGM (post LGM) with a species Siphotextularia rolshauseni in a sediment core PS1243 from permanent sea ice cover inhibiting sea ice algae (IP25 = 0) and the Nordic Seas (Bauch et al., 2001). This species seems to be in- phytoplankton (brassicasterol ≈ 0) productivity at the core site dicative of well-oxygenated water conditions (Gupta et al., 2006), for more than 1000 yr (Fig. 3). A perennial ice cover and lowest which corroborates the assumption of a modified oceanic circula- SSTs are also displayed by maximum PBIP25 and PDIP25 values and tion in the northern North Atlantic during the late glacial period. minimum DIP25 values, respectively (Fig. 3). Further, Zamelczyk A significant though highly variable Atlantic water inflow to the et al. (2014) observed a drop in the total foraminifer flux and southern part of the GIN Seas, however, is also suggested by e.g. a high shell fragmentation in the respective sediments of core Rasmussen and Thomsen (2008), Rørvik et al. (2013) and Weinelt MSM5/5-712-2 at about 19 ka BP which they attribute to a cooling of the sea surface. This onset of a perennial sea ice cover in et al. (2003) from foraminifer data. A model simulation by Stärz eastern Fram Strait seems to coincide with the final expansion (ac- et al. (2012) using GLAMAP SST data (see Fig. 1 for the mod- cording to Andersen et al., 1996) or retreat (see Jessen et al., 2010 elled summer sea ice extent in Fram Strait between 23 ka and for further references and discussion) of the SBIS at the western 19 ka BP) supports the assumption that the core site was located in shelf edge of Svalbard. In any case, the close vicinity of the SBIS close vicinity to the maximum summer sea ice margin and hence could have supported the formation of perennial (fast) sea ice in suited to record these short-term sea ice fluctuations during the the study area. In addition, melting of Northern Hemisphere ice- late glacial. These model results also point towards a perennial sheets since at least 19 ka BP (Clark et al., 2004) and the associated sea ice cover at the PS2837-5 core site in northern Fram Strait freshwater discharge to the Arctic Ocean would have promoted (Fig. 1), which is in well agreement with the biomarker data of this an overall increase in sea ice formation. In addition, the perma- core (Müller et al., 2009). Regarding the overall late glacial palaeo- nent post LGM sea ice coverage at the core site was likely related ceanographic situation in the study area, the recurrent formation to a final cooling of Atlantic surface waters, which is also indi- of highly productive polynyas in front of the Svalbard–Barents Sea cated by minimum Mg/Ca SSTs recorded at ODP site 980 in the Ice Sheet (SBIS) due to katabatic winds (for further discussion see eastern subpolar North Atlantic (Fig. 4; Benway et al., 2010). Con- e.g. Knies et al., 1999) may be considered as a further explanation cerning the strength of the AMOC, it has been demonstrated in var- for the repeated maxima in both biomarker records. ious proxy and model studies that the oceanic overturn remained J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455 451

normally (i.e. weak but not halted) operating AMOC during the latest phase of the LGM.

4.2. The last deglaciation

Abruptly rising concentrations of IP25 and, in particular, of brassicasterol point to a return of ice-free summers at the western continental margin of Svalbard at about 17.6 ka BP (Fig. 3). Adrop in PBIP25 and PDIP25 values from maximum (1) to minimum values (< 0.2) within ca. 300 yr and increasing DIP25 values at the core site coincide with rising Mg/Ca temperatures (Benway et al., 2010) and decreasing planktic δ18Ovalues observed in the North Atlantic (Dokken and Jansen, 1999; Grousset et al., 2001) which suggest a significant sea surface warming and freshening (Fig. 4). A slightly later retreat of sea ice is also traceable in the biomarker record of core PS2837-5 (Müller et al., 2009). This breakup of a previously perennial ice cover in Fram Strait, coincident with the onset of Heinrich Event 1 at 17.5 ka BP, apparently accompanied (or amplified) a major shift in the oceanic setting in the North At- lantic. The sudden reduction in sea ice coverage correlates well with a decrease in δ18O values observed in several marine sediment cores from the northeast Atlantic and the GIN Seas (see e.g. Fig. 4) that reflects a rapid freshening of the sea surface (and thus stratification) due to the onset of deglacial meltwater discharge (Bauch et al., 2001; Weinelt et al., 2003; Benway et al., 2010; Telesinski´ et al., 2014). In their study of a sediment core SU90-09 from the western side of the mid-Atlantic Ridge, Grousset et al. (2001) identified European ice sheet derived IRD and concluded that a destabilisation of these ice sheets (just prior to the major Laurentide surge) might have served as a trigger for Heinrich Event 1. This supports our assumption that a general collapse of Arctic Ocean (landfast) sea ice caused a large-scale destabilisation of tidewater glacier fronts which resulted in increased iceberg calving and a rapid export of sea ice and icebergs into the North Atlantic (Fig. 5). This sudden and concentrated freshwater forcing consequentially led to a significant weakening of the AMOC, which is inferred from e.g. abruptly decreasing δ13Cvalues at ODP site 980 (Fig. 4; Benway et al., 2010) or an increase in the sedimen- 231 230 Fig. 4. Compilation of North Atlantic proxy and model data documenting oceanic tary Pa/ Th values (a proxy for the meridional overturning changes during the last deglaciation. 231Pa/230Th values, δ13Cvalues of benthic circulation) of a core from the subtropical North Atlantic (Fig. 4; foraminifera (McManus et al., 1999, 2004) and the modelled AMOC anomaly (green McManus et al., 2004; see Burke et al. (2011) for a thorough dis- curve; Ritz et al., 2013) display changes in the oceanic overturning. Planktic δ18O cussion of the 231Pa/230Th approach). In addition, Ritz et al. (2013) values refer to the sea surface temperature and/or freshening at the Vøring Plateau (Dokken and Jansen, 1999). Mg/Ca temperatures (Benway et al., 2010)and the DIP25 provided model data estimating a 14 Sv reduction in the oceanic index of core MSM5/5-712-2 indicate (relative) sea surface temperature variability. overturn during Heinrich Event 1 (Fig. 4). 18 Changes in the sea ice coverage at eastern Fram Strait are reflected in the PBIP25 The abrupt minimum in foraminifer δ Ovalues and peak record of core MSM5/5-712-2. (For interpretation of the references to color in this foraminifer flux rates observed in core MSM5/5-712-2 between figure legend, the reader is referred to the web version of this article.) 17 ka and 16.5 ka BP (Zamelczyk et al., 2014) coincide with elevated brassicasterol concentrations (Fig. 3), which may point to relatively undisturbed until ca. 17.5 ka BP (Fig. 4; for further dis- an increase in primary productivity at the core site in response to cussion see e.g. McManus et al., 2004; Waelbroeck et al., 2011; a freshwater and thus elevated nutrient discharge. During the later Ritz et al., 2013). Heavy δ18Ovalues documented by Dokken and part of Heinrich Event 1, rising PIP25 values seem to reflect a suc- Jansen (1999) in a core from the Norwegian Sea (Fig. 4) thus seem cessive re-expansion of sea ice until 14.5 ka BP, when a minor low in DIP values (suggestive of lowered SSTs), an increase in IP to be associated with the continued (subsurface) advection of high 25 25 and a decrease in brassicasterol concentrations and hence rising salinity Atlantic water (Fig. 5). The sea ice coverage in Fram Strait P IP values (Fig. 3) point to a more distinct short-term increase apparently served as an effective insulator limiting the oceanic B 25 in sea ice cover, which might be attributed to meltwater pulse 1A heat loss to the atmosphere and hence promoted the northward (MWP-1A). This 14.5 ka BP melting event, which has been associ- subsurface penetration of Atlantic Water. We suggest that the ated with a significant sea level rise during the Bølling warm pe- screening effect of a persistent perennial sea ice coverage in Fram riod (Deschamps et al., 2012), is also traceable in various sediment Strait and probably also in the adjacent Arctic Ocean (Fig. 5) could cores from Fram Strait and adjacent areas (Birgel and Hass, 2004; have contributed to an unusual heat accumulation in glacial inter- Jessen et al., 2010). At the core site, MWP-1A caused exceptional mediate Arctic Ocean water depths as was reconstructed by Cronin high sedimentation rates (Fig. 2) and the accumulation of high et al. (2012). Further, permanent and, in particular, landfast sea ice amounts of, probably terrigenous, meltwater derived organic car- would have hampered the calving of icebergs (in terms of glacier bon (S1). buttressing) and the export of icebergs and thus restrained the The onset of the Younger Dryas – a millennial-duration cold freshwater release to the GIN Seas. This could have sustained a spell at the termination of the last deglaciation – is marked by a 452 J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455

Fig. 5. Permanent and immobile post LGM (19.2–17.6 ka BP) summer sea ice in the Arctic Ocean and Fram Strait (left); break-up of permanent ice and export of sea ice and icebergs to the North Atlantic during Heinrich Event 1 (centre); increased sea ice formation and discharge through Fram Strait during the Younger Dryas (right). Extent of continental ice sheets (green shadings) during the late LGM after Dyke et al. (2002) and Mangerud et al. (2004), and during the Younger Dryas after Bradley and England (2008). Dark grey shaded areas refer to the LGM sea level (after Stärz et al., 2012). Melt-/freshwater supply from major Arctic river systems and ice drift patterns (Transpolar Drift and Beaufort Gyre) are indicated for the Younger Dryas. Circles refer to core sites that experienced sea ice coverage (see Table S5; yellow circle = MSM5/5-712-2). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) minor peak in IP25 and increasing brassicasterol concentrations at Though we observe a rather similar (i.e. extended) sea ice cov- ca. 13.2 ka BP (Fig. 3). Thereafter, between 12.8 ka and 12.6 ka BP, erage in our study area during the Younger Dryas and post LGM, a very severe and extended (lasting until summer) sea ice coverage the overall boundary conditions of these two climate intervals are is noticeable from maximum IP25 concentrations (up to 1.6μg/g fundamentally different. A weak (yet slowly rising) insolation and OC) and a remarkable drop in brassicasterol concentrations to ca. thus low summer melt rates, a significantly lowered sea level, and 5–6 μg/g OC, which are only slightly higher values than during shallow shelves exposed or covered by the Eurasian and North the late LGM. We suggest that during the Younger Dryas, in con- American ice sheets characterised the Arctic Ocean during the trast to the LGM, astronger insolation (Fig. 3) and a hence higher late LGM. We suggest that these conditions patronised a perma- light penetration through the ice promoted the productivity of sea nent, thick, and immobile sea ice coverage in Fram Strait between ice diatoms. Peak PBIP25 and PDIP25values (> 0.9) and minimum 19.2 and 17.6 ka BP (Fig. 5). During the Younger Dryas, in con- DIP25 values (< 3) indicate that an almost perennial sea ice cover- trast, an almost maximum insolation promoted the disintegration age and consequentially lowered SSTs persisted for at least 100 yr of remaining continental ice sheets and a hence strong freshwater (12.8–12.7 ka BP; Figs. 3 and 4). discharge to the Arctic Ocean. The rising sea level caused a flood- It is assumed that the Younger Dryas cold event resulted from ing of the shallow shelves of the Kara and Laptev Seas thus putting a deterioration of deep-water formation in the North Atlantic (and the “sea ice factories” of the Arctic Ocean into operation (Fig. 5). thus a weakening of the AMOC) due to an enormous discharge of Furthermore, high summer melt rates and changes in the oceanic freshwater from North American proglacial lakes into the North At- and atmospheric circulation (e.g. a strengthened Transpolar Drift) lantic (Broecker et al., 1989; Clark et al., 2001; Teller et al., 2002; would have forced sea ice export via Fram Strait into the GIN Seas Katz et al., 2011; Rayburn et al., 2011). Recently, Murton et al. thus decelerating the AMOC during the Younger Dryas (Fig. 5). (2010) identified a northwestern outburst flood path from Lake Agassiz via the Mackenzie Delta, which probably accounted for a 4.3. The early Holocene meltwater routing across the Arctic Ocean that led to a 30% weakening of the AMOC (Condron and Winsor, 2012). In this sense, Fahl The onset of the Holocene warm period is marked by a sig- and Stein (2012) related a significant increase in sea ice coverage nificant retreat of the sea ice cover at the western continental at the southern Lomonosov Ridge to a preceding freshwater event margin of Svalbard as is inferred from highest brassicasterol con- that affected the Arctic Ocean at the onset of the Younger Dryas centrations (20–40 μg/gOC) and minimum IP25 concentrations (Spielhagen et al., 2005). A pan-Arctic routing of freshwater and a (0.1–0.2μg/gOC) between 12 ka and 9ka BP (Fig. 3). PBIP25 and hence elevated formation of sea ice is also inferred from sedimen- PDIP25 values shift to lowest and DIP25 values to maximum values, tological and geochemical data of sediment cores from the cen- pointing to a minimum sea ice coverage and substantially warmer tral Arctic Ocean (Not and Hillaire-Marcel, 2012) and central Fram SSTs, respectively (Fig. 3). The productivity of brassicasterol synthe- Strait (Maccali et al., 2013). This observation highlights the enor- sizing phytoplankton was apparently stimulated by the maximum mous effect of a freshening of the Arctic Ocean for the formation insolation (Fig. 3), which, in turn, probably also limited the for- of sea ice during a period of significantly increased insolation. Even mation (and persistence) of sea ice in eastern Fram Strait. Further, sediment cores from sites located much further to the south carry the general recovery of the AMOC (since ca. 11.7 ka BP; Fig. 4) information about the prevalence of sea ice during the Younger supported an effective heat transfer from the tropical towards the Dryas (Fig. 5; Knies, 2005; Slubowska-Woldengen et al., 2008; subpolar North Atlantic (McManus et al., 2004). This is also sug- Bakke et al., 2009; Cabedo-Sanz et al., 2013; see Table S5). We gested by Risebrobakken et al. (2011), who argued that the early assume that this southward propagation of Arctic sea ice (and Holocene warming recorded at various sites in the Nordic Seas freshwater) accounted substantially for the repeated weakening of reflects an intensified heat advection due a reorganisation of the the AMOC between 12.8 ka and 11.8 ka BP (Figs. 4 and 5). oceanic circulation following the last deglaciation. J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455 453

5. Implications of late glacial sea ice fluctuations for LGM climate to a restrained freshwater export from the Arctic to the North At- reconstructions lantic, which, in turn, would have promoted a normally operating AMOC. The sudden break-up of the ice cover and the concen- This is the first documentation of rapidly changing sea ice con- trated release of high amounts of sea ice and icebergs trapped ditions in the Fram Strait during the late glacial and LGM that may therein at 17.6 ka BP then could have acted as a trigger for the indicate that the heat flow to the Arctic Ocean via this nar- AMOC break-down during Heinrich Event 1. Similarly, an almost row passage was rather unstable. The alternating phases of ice perennial sea ice cover in Fram Strait coincided with a weakened advances and ice retreats between 26 and 19 ka BP reveal a AMOC during the Younger Dryas. The latter expansion of sea ice significant (though often unconsidered) short-term variability in likely resulted from the overall increase in sea ice formation in the the oceanic (and/or atmospheric) circulation in the northernmost Arctic Ocean due to a significant deglacial freshwater release and part of the subpolar North Atlantic throughout a critical period sea level rise. The high insolation and a probably strengthened or of high latitude ice-sheet growth. We argue that this may be of reorganised oceanic and atmospheric circulation within the Arc- general importance for palaeoceanographic proxy reconstructions tic Ocean promoted the export of sea ice through Fram Strait to and model simulations focusing on the LGM, which, at least for the North Atlantic where it impacted on thermohaline processes these studies, is commonly defined as the time period between and contributed to the Younger Dryas AMOC weakening. Highest 23 ka and 19 ka BP (e.g. by the MARGO Project, 2009). Most phytoplankton and lowest ice algae productivity characterised the proxy and numerically modelled LGM time-slice reconstructions early Holocene when a maximum insolation led to a significant sea display only a mean state of climate and sea surface conditions, surface warming and sea ice retreat in the study area. respectively (e.g. Pflaumann et al., 2003; Meland et al., 2005; Acknowledgements MARGO Project Members, 2009; Otto-Bliesner and Brady, 2010; Stärz et al., 2012) and thus the temporal development through- The Geosciences Department of the Alfred Wegener Institute out these 4ka remains unresolved. The eventuality of a discon- is acknowledged for funding of J. Müller. Kirsten Fahl and Wal- tinuous northward heat transport towards the GIN Seas, however, ter Luttmer are kindly thanked for the laboratory management sheds new light on the interpretation of proxy data derived from in Bremerhaven. We wish to thank Heather Benway for provid- sedimentary archives. Uncertainties about reconstructed LGM sea ing foraminiferal stable isotope and Mg/Ca data of ODP core 980. surface conditions thus may not only be related to an often prob- Katarzyna Zamelczyk is thanked for magnetic susceptibility data of lematic age control or the different analytical approaches used to core MSM5/5-712-2. Special thanks go to Michael Schreck for sub- estimate SSTs (for discussion see e.g. Rosell-Melé and Comes, 1999; stantial logistical support in Bremerhaven. Christian Hass is greatly Meland et al., 2005; de Vernal et al., 2006) but they may also re- acknowledged for invaluable discussions and comments on an ear- late to a time-variable heat flux at the different core sites. With lier version of this manuscript. We wish to acknowledge the three regard to the sea ice history in eastern Fram Strait, we suggest anonymous reviewers for their constructive comments. This arti- that a long-term maximum (i.e. perennial) sea ice coverage pre- cle benefited, in particular, from the very helpful comments of vailed only between 19.2 ka and 17.6 ka BP, which is just be- reviewer #3 concerning the stratigraphy and the age model of core yond the conventional LGM time window (23–19 ka BP). This MSM5/5-712-2. may give reason to deliberate whether the LGM – in terms of palaeoceanographic reconstructions – should be defined as afixed Appendix A. Supplementary material time interval or as the period characterised by most severe (max- Supplementary material related to this article can be found on- imum) last glacial climate conditions in a specific area. Concern- line at http://dx.doi.org/10.1016/j.epsl.2014.07.016. ing large-scale (i.e. global) climate reconstructions, of course, only the former would prove suitable or at least feasible. However, lo- References cal (time-transgressive) phenomena, such as the permanent sea ice coverage in Fram Strait and the associated palaeoceanographic Aagaard, K., 1982. Inflow from the Atlantic Ocean to the Polar Basin. In: Rey., L. (Ed.), The Arctic Ocean. Comité Arctique International, Monaco, pp. 69–82. impact, would remain disregarded. Transient palaeoclimate model Aagaard, K., Carmack, E.C., 1989. The role of sea ice and other fresh water in the experiments as well as comprehensive palaeoceanographic recon- Arctic circulation. J. Geophys. Res. 94 (C10), 14485–14498. structions based on continuous sedimentary records (ideally with Aagaard, K., Coachman, L.K., 1968. The East Greenland Current north of Denmark a sufficiently high temporal resolution) could provide valuable in- Strait. Arctic 21 (3), 181–200. Aagaard, K., Swift, J.H., Carmack, E.C., 1985. Thermohaline circulation in the Arctic formation about regionally and temporally different responses to Mediterranean Seas. J. Geophys. Res. 90 (C3), 4833–4846. late glacial climate change. Aagaard-Sørensen, S., Husum, K., Hald, M., Marchitto, T., Godtliebsen, F., 2014. Sub sea surface temperatures in the Polar North Atlantic during the Holocene: 6. Conclusions planktic foraminiferal Mg/Ca temperature reconstructions. Holocene 24 (1), 93–103. Andersen, E.S., Dokken, T.M., Elverhøi, A., Solheim, A., Fossen, I., 1996. Late qua- The reconstruction of sea ice conditions in eastern Fram Strait ternary sedimentation and glacial history of the western Svalbard continental by means of the sea ice proxy IP25 and phytoplankton derived margin. Mar. Geol. 133 (3–4), 123–156. biomarkers gives new insight into the highly variable palaeoceano- Bakke, J., Lie, O., Heegaard, E., Dokken, T., Haug, G.H., Birks, H.H., Dulski, P., Nilsen, graphic evolution in the subpolar North Atlantic throughout the T., 2009. Rapid oceanic and atmospheric changes during the Younger Dryas cold period. Nat. Geosci. 2 (3), 202–205. late glacial, deglacial and the early Holocene. Recurrent phases of Bauch, H.A., Erlenkeuser, H., Spielhagen, R.F., Struck, U., Matthiessen, J., Thiede, J., sea ice advances and retreats during the late glacial indicate that Heinemeier, J., 2001. A multiproxy reconstruction of the evolution of deep and a rather discontinuous (or pulse-like) heat flow along the eastern surface waters in the subarctic Nordic seas over the last 30,000 yr. Quat. Sci. Nordic Seas governed sea ice conditions in the study area. Only Rev. 20 (4), 659–678. Belt, S.T., Brown, T.A., Ringrose, A.E., Cabedo-Sanz, P., Mundy, C.J., Gosselin, M., during the latest phase of the LGM (between 19.2 and 17.6 ka BP), Poulin, M., 2013. Quantitative measurement of the sea ice diatom biomarker a long-lasting permanent and thick sea ice cover persisted at the IP25 and sterols in Arctic sea ice and underlying sediments: further considera- core site. This accumulation of ice masses in Fram Strait (and prob- tions for palaeo sea ice reconstruction. Org. Geochem. 62, 33–45. ably also in the adjacent Arctic Ocean) at the very end of the LGM Belt, S.T., Massé, G., Rowland, S.J., Poulin, M., Michel, C., LeBlanc, B., 2007. A novel chemical fossil of palaeo sea ice: IP . Org. Geochem. 38 (1), 16–27. could have been a crucial pre-requisite for the AMOC collapse dur- 25 Belt, S.T., Müller, J., 2013. The Arctic sea ice biomarker IP25: areview of current ing the following Heinrich Event 1. We suggest that the permanent understanding, recommendations for future research and applications in palaeo ice cover could have hampered the calving of icebergs and thus led sea ice reconstructions. Quat. Sci. Rev. 79 (1), 9–25. 454 J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455

Benway, H.M., McManus, J.F., Oppo, D.W., Cullen, J.L., 2010. Hydrographic changes Knies, J., 2005. Climate-induced changes in sedimentary regimes for organic mat- in the eastern subpolar North Atlantic during the last deglaciation. Quat. Sci. ter supply on the continental shelf off northern Norway. Geochim. Cosmochim. Rev. 29 (23–24), 3336–3345. Acta 69 (19), 4631–4647. Birgel, D., Hass, H.C., 2004. Oceanic and atmospheric variations during the last Knies, J., Stein, R., 1998. New aspects of organic carbon deposition and its pale- deglaciation in the Fram Strait (Arctic Ocean): a coupled high-resolution oceanographic implications along the northern Barents Sea margin during the organic-geochemical and sedimentological study. Quat. Sci. Rev. 23 (1–2), 29–47. last 30,000 years. Paleoceanography 13 (4), 384–394. Bradley, R.S., England, J.H., 2008. The Younger Dryas and the sea of ancient ice. Quat. Knies, J., Vogt, C., Stein, R., 1999. Late Quaternary growth and decay of the Sval- Res. 70 (1), 1–10. bard/Barents Sea ice sheet and paleoceanographic evolution in the adjacent Broecker, W.S., Kennett, J.P., Flower, B.P., Teller, J.T., Trumbore, S., Bonani, G., Wolfli, Arctic Ocean. Geo Mar. Lett. 18 (3), 195–202. W., 1989. Routing of meltwater from the Laurentide Ice Sheet during the Koç, N., Jansen, E., Haflidason, H., 1993. Paleoceanographic reconstructions of surface Younger Dryas cold episode. Nature 341 (6240), 318–321. ocean conditions in the Greenland, Iceland and Norwegian seas through the last Brown, T.A., Belt, S.T., Tatarek, A., Mundy, C.J., 2014. Source identification of the 14 ka based on diatoms. Quat. Sci. Rev. 12 (2), 115–140. Kwok, R., Cunningham, G.F., Pang, S.S., 2004. Fram Strait sea ice outflow. J. Geophys. Arctic sea ice proxy IP25. Nat. Commun. 5, 4197. http://dx.doi.org/10.1038/ ncomms5197. Res. 109 (C1), C01009. Budéus, G., 2007. Short Cruise Report RV Maria S. Merian Cruise MSM05/5. U. o. Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A.C.M., Levrard, B., 2004. Hamburg Institute of Oceanography. Along-term numerical solution for the insolation quantities of the Earth. As- tron. Astrophys. 428 (1), 261–285. Burke, A., Marchal, O., Bradtmiller, L.I., McManus, J.F., François, R., 2011. Application Maccali, J., Hillaire-Marcel, C., Carignan, J., Reisberg, L.C., 2013. Geochemical signa- of an inverse method to interpret 231Pa/230Th observations from marine sedi- tures of sediments documenting Arctic sea-ice and water mass export through ments. Paleoceanography 26 (1), PA1212. Fram Strait since the Last Glacial Maximum. Quat. Sci. Rev. 64, 136–151. Cabedo-Sanz, P., Belt, S.T., Knies, J., Husum, K., 2013. Identification of contrasting Mangerud, J., Gulliksen, S., 1975. Apparent radiocarbon ages of recent marine shells seasonal sea ice conditions during the Younger Dryas. Quat. Sci. Rev. 79, 74–86. from Norway, Spitsbergen, and Arctic Canada. Quat. Res. 5 (2), 263–273. Clark, P.U., Marshall, S.J., Clarke, G.K.C., Hostetler, S.W., Licciardi, J.M., Teller, J.T., Mangerud, J., Jakobsson, M., Alexanderson, H., Astakhov, V., Clarke, G.K.C., Henrik- 2001. Freshwater forcing of abrupt climate change during the last glaciation. sen, M., Hjort, C., Krinner, G., Lunkka, J.-P., Möller, P., Murray, A., Nikolskaya, Science 293 (5528), 283–287. O., Saarnisto, M., Svendsen, J.I., 2004. Ice-dammed lakes and rerouting of the Clark, P.U., McCabe, A.M., Mix, A.C., Weaver, A.J., 2004. Rapid rise of sea level 19,000 drainage of northern Eurasia during the Last Glaciation. Quat. Sci. Rev. 23 years ago and its global implications. Science 304 (5674), 1141–1144. (11–13), 1313–1332. Condron, A., Winsor, P., 2012. Meltwater routing and the Younger Dryas. Proc. Natl. MARGO Project Members, 2009. Constraints on the magnitude and patterns of ocean Acad. Sci. http://dx.doi.org/10.1073/pnas.1207381109. cooling at the Last Glacial Maximum. Nat. Geosci. 2 (2), 127–132. Cronin, T.M., Dwyer, G.S., Farmer, J., Bauch, H.A., Spielhagen, R.F., Jakobsson, M., Nils- Martin, T., Wadhams, P., 1999. Sea-ice flux in the East Greenland Current. Deep Sea son, J., Briggs, W.M., Stepanova, A., 2012. Deep Arctic Ocean warming during the Research Part II. Top. Stud. Oceanogr. 46 (6–7), 1063–1082. last glacial cycle. Nat. Geosci. 5 (9), 631–634. Mauritzen, C., Häkkinen, S., 1997. Influence of sea ice on the thermohaline circula- de Vernal, A., Hillaire-Marcel, C., Rochon, A., Fréchette, B., Henry, M., Solignac, S., tion in the Arctic–North Atlantic Ocean. Geophys. Res. Lett. 24 (24), 3257–3260. Bonnet, S., 2013. Dinocyst-based reconstructions of sea ice cover concentration McManus, J.F., Francois, R., Gherardi, J.M., Keigwin, L.D., Brown-Leger, S., 2004. Col- during the Holocene in the Arctic Ocean, the northern North Atlantic Ocean and lapse and rapid resumption of Atlantic meridional circulation linked to deglacial its adjacent seas. Quat. Sci. Rev. 79, 111–121. climate changes. Nature 428 (6985), 834–837. de Vernal, A., Rosell-Melè, A., Kucera, M., Hillaire-Marcel, C., Eynaud, F., Weinelt, McManus, J.F., Oppo, D.W., Cullen, J.L., 1999. A 0.5-million-year record of millennial- M., Dokken, T., Kageyama, M., 2006. Comparing proxies for the reconstruction scale climate variability in the North Atlantic. Science 283 (5404), 971–975. of LGM sea-surface conditions in the northern North Atlantic. Quat. Sci. Rev. 25 Meland, M.Y., Jansen, E., Elderfield, H., 2005. Constraints on SST estimates for the (21–22), 2820–2834. northern North Atlantic/Nordic Seas during the LGM. Quat. Sci. Rev. 24 (7–9), Deschamps, P., Durand, N., Bard, E., Hamelin, B., Camoin, G., Thomas, A.L., Hender- 835–852. son, G.M., Okuno, J.I., Yokoyama, Y., 2012. Ice-sheet collapse and sea-level rise Müller, J., Massé, G., Stein, R., Belt, S.T., 2009. Variability of sea-ice conditions in the at the Bolling warming 14,600 years ago. Nature 483 (7391), 559–564. Fram Strait over the past 30,000 years. Nat. Geosci. 2 (11), 772–776. Dokken, T.M., Hald, M., 1996. Rapid climatic shifts during isotope stages 2–4 in the Müller, J., Wagner, A., Fahl, K., Stein, R., Prange, M., Lohmann, G., 2011. Towards Polar North Atlantic. Geology 24 (7), 599–602. quantitative sea ice reconstructions in the northern North Atlantic: a combined Dokken, T.M., Jansen, E., 1999. Rapid changes in the mechanism of ocean convection biomarker and numerical modelling approach. Earth Planet. Sci. Lett. 306 (3–4), during the last glacial period. Nature 401 (6752), 458–461. 137–148. Dyke, A.S., Andrews, J.T., Clark, P.U., England, J.H., Miller, G.H., Shaw, J., Veillette, J.J., Müller, J., Werner, K., Stein, R., Fahl, K., Moros, M., Jansen, E., 2012. Holocene cooling 2002. The Laurentide and Innuitian ice sheets during the Last Glacial Maximum. culminates in sea ice oscillations in Fram Strait. Quat. Sci. Rev. 47, 1–14. Quat. Sci. Rev. 21 (1–3), 9–31. Murton, J.B., Bateman, M.D., Dallimore, S.R., Teller, J.T., Yang, Z., 2010. Identification Fahl, K., Stein, R., 2012. Modern seasonal variability and deglacial/Holocene change of Younger Dryas outburst flood path from Lake Agassiz to the Arctic Ocean. of central Arctic Ocean sea-ice cover: new insights from biomarker proxy Nature 464 (7289), 740–743. records. Earth Planet. Sci. Lett. 351–352, 123–133. Niessen, F., Kleiber, H., Nehrke, G., 1997. Physical properties of sediment core PS2837-5. In: Stein, R., Fahl, K. (Eds.), Scientific Cruise Report of the Arctic Ex- Fairbanks, R.G., Mortlock, R.A., Chiu, T.-C., Cao, L., Kaplan, A., Guilderson, T.P., Fair- pedition ARK-XIII/2 of RV Polarstern in 1997, vol. 255. Alfred Wegener Institute banks, T.W., Bloom, A.L., Grootes, P.M., Nadeau, M.-J., 2005. Radiocarbon calibra- for Polar Marine Research, Bremerhaven, pp. 115–120. tion curve spanning 0 to 50,000 years BP based on paired 230Th/234U/238Uand 14 Nørgaard-Pedersen, N., Spielhagen, R.F., Erlenkeuser, H., Grootes, P.M., Heinemeier, Cdates on pristine corals. Quat. Sci. Rev. 24 (16–17), 1781–1796. J., Knies, J., 2003. Arctic Ocean during the Last Glacial Maximum: Atlantic and Grousset, F.E., Cortijo, E., Huon, S., Hervé, L., Richter, T., Burdloff, D., Duprat, J., We- polar domains of surface water mass distribution and ice cover. Paleoceanogra- ber, O., 2001. Zooming in on Heinrich layers. Paleoceanography 16 (3), 240–259. phy 18 (3), 1063–1082. Gupta, A.K., Sarkar, S., Mukherjee, B., 2006. Paleoceanographic changes during the Not, C., Hillaire-Marcel, C., 2012. Enhanced sea-ice export from the Arctic during the past 1.9 Myr at DSDP Site 238, Central Indian Ocean Basin: benthic foraminiferal Younger Dryas. Nat. Commun. 3, 647. proxies. Mar. Micropaleontol. 60 (2), 157–166. Otto-Bliesner, B.L., Brady, E.C., 2010. The sensitivity of the climate response to the Hebbeln, D., Dokken, T., Andersen, E.S., Hald, M., Elverhøi, A., 1994. Moisture supply magnitude and location of freshwater forcing: last glacial maximum experi- for northern ice-sheet growth during the Last Glacial Maximum. Nature 370, ments. Quat. Sci. Rev. 29 (1–2), 56–73. 357–360. Pflaumann, U., Sarnthein, M., Chapman, M., d’Abreu, L., Funnell, B., Huels, M., Kiefer, Horner, R., Schrader, G.C., 1982. Relative contributions of ice algae, phytoplankton, T., Maslin, M., Schulz, H., Swallow, J., van Kreveld, S., Vautravers, M., Vogel- and benthic microalgae to primary production in nearshore regions of the Beau- sang, E., Weinelt, M., 2003. Glacial North Atlantic: sea-surface conditions re- fort Sea. Arctic 35 (4), 485–503. constructed by GLAMAP 2000. Paleoceanography 18, 1065. Jessen, S.P., Rasmussen, T.L., Nielsen, T., Solheim, A., 2010. A new Late We- Rahmstorf, S., 2002. Ocean circulation and climate during the past 120,000 years. ichselian and Holocene marine chronology for the western Svalbard slope Nature 419 (6903), 207–214. 30,000–0 cal years BP. Quat. Sci. Rev. 29 (9–10), 1301–1312. Ran, L., Jiang, H., Knudsen, K.L., Eiríksson, J., Gu, Z., 2006. Diatom response to the Jung, T., Hilmer, M., 2001. The link between the North Atlantic Oscillation and Arctic Holocene climatic optimum on the North Icelandic shelf. Mar. Micropaleon- sea ice export through Fram Strait. J. Climate 14 (19), 3932–3943. tol. 60 (3), 226–241. Kanazawa, A., Yoshioka, M., Teshima, S.-I., 1971. The occurrence of brassicasterol in Rasmussen, T.L., Thomsen, E., 2008. Warm Atlantic surface water inflow to the the diatoms, Cyclotella nana and Nitzschia closterium. Bull. Jpn. Soc. Sci. Fish. 37, Nordic seas 34–10 calibrated ka B.P. Paleoceanography 23 (1), PA1201. 889–903. Rayburn, J.A., Cronin, T.M., Franzi, D.A., Knuepfer, P.L.K., Willard, D.A., 2011. Timing Katz, B., Najjar, R.G., Cronin, T., Rayburn, J., Mann, M.E., 2011. Constraints on Lake and duration of North American glacial lake discharges and the Younger Dryas Agassiz discharge through the late-glacial Champlain Sea (St. Lawrence Low- climate reversal. Quat. Res. 75 (3), 541–551. lands, Canada) using salinity proxies and an estuarine circulation model. Quat. Risebrobakken, B., Dokken, T., Smedsrud, L.H., Andersson, C., Jansen, E., Moros, M., Sci. Rev. 30 (23–24), 3248–3257. Ivanova, E.V., 2011. Early Holocene temperature variability in the Nordic Seas: J. Müller, R. Stein / Earth and Planetary Science Letters 403 (2014) 446–455 455

the role of oceanic heat advection versus changes in orbital forcing. Paleo- Stärz, M., Gong, X., Stein, R., Darby, D.A., Kauker, F., Lohmann, G., 2012. Glacial short- ceanography 26 (4), PA4206. cut of Arctic sea-ice transport. Earth Planet. Sci. Lett. 357–358, 257–267. Ritz, S.P., Stocker, T.F., Grimalt, J.O., Menviel, L., Timmermann, A., 2013. Estimated Stein, R., Fahl, K., 2012. A first southern Lomonosov Ridge (Arctic Ocean) 60 ka IP25 strength of the Atlantic overturning circulation during the last deglaciation. Nat. sea-ice record. Polarforschung 82 (1), 83–86. Geosci. 6 (3), 208–212. Stein, R., Fahl, K., 2013. Biomarker proxy shows potential for studying the entire Rørvik, K.-L., Rasmussen, T.L., Hald, M., Husum, K., 2013. Intermediate water Quaternary Arctic sea ice history. Org. Geochem. 55, 98–102. ventilation in the Nordic seas during MIS 2. Geophys. Res. Lett. 40 (9), Stein, R., Fahl, K., Müller, J., 2012. Proxy reconstruction of Cenozoic Arctic Ocean 1805–1810. sea-ice history: from IRD to IP25. Polarforschung 82 (1), 37–71. Rosell-Melé, A., Comes, P., 1999. Evidence for a warm Last Glacial Maximum in the Telesinski,´ M.M., Spielhagen, R.F., Bauch, H.A., 2014. Water mass evolution of the K K Greenland Sea since late glacial times. Clim. Past 10 (1), 123–136. Nordic seas or an example of shortcomings in U 37 and U 37 to estimate low sea surface temperature? Paleoceanography 14 (6), 770–776. Teller, J.T., Leverington, D.W., Mann, J.D., 2002. Freshwater outbursts to the oceans Rowland, S.J., Belt, S.T., Wraige, E.J., Masse, G., Roussakis, C., Robert, J.M., 2001. Ef- from glacial Lake Agassiz and their role in climate change during the last fects of temperature on polyunsaturation in cytostatic lipids of Haslea ostrearia. deglaciation. Quat. Sci. Rev. 21 (8–9), 879–887. Phytochemistry 56 (6), 597–602. Vogt, C., Knies, J., Spielhagen, R.F., Stein, R., 2001. Detailed mineralogical evidence for two nearly identical glacial/deglacial cycles and Atlantic water advection to Rudels, B., 1996. The thermohaline circulation of the Arctic Ocean and the Green- the Arctic Ocean during the last 90,000 years. Glob. Planet. Change 31 (1–4), land Sea. In: Wadhams, P., Dowdeswell, J.A., Schofield, A.N. (Eds.), The Arctic and 23–44. Environmental Change. Gordon and Breach, Amsterdam, pp. 87–99. Volkman, J.K., 2006. Lipid markers for marine organic matter. In: Volkman, J.K. (Ed.), Rudels, B., Quadfasel, D., 1991. Convection and deep water formation in the Arctic Handbook of Environmental Chemistry, vol. 2. Springer-Verlag, Berlin, Heidel- Ocean–Greenland Sea system. J. Mar. Syst. 2 (3–4), 435–450. berg, pp. 27–70. Sarnthein, M., Pflaumann, U., Weinelt, M., 2003. Past extent of sea ice in the north- Waelbroeck, C., Skinner, L.C., Labeyrie, L., Duplessy, J.C., Michel, E., Vazquez Riveiros, ern North Atlantic inferred from foraminiferal paleotemperature estimates. Pa- N., Gherardi, J.M., Dewilde, F., 2011. The timing of deglacial circulation changes leoceanography 18 (2), 1047–1058. in the Atlantic. Paleoceanography 26 (3), PA3213. Schlüter, M., Sauter, E., 2000. Biogenic silica cycle in surface sediments of the Green- Weckström, K., Massé, G., Collins, L.G., Hanhijärvi, S., Bouloubassi, I., Sicre, M.-A., land Sea. J. Mar. Syst. 23 (4), 333–342. Seidenkrantz, M.-S., Schmidt, S., Andersen, T.J., Andersen, M.L., Hill, B., Kuijpers, Schulz, M., Paul, A., 2004. Sensitivity of the ocean–atmosphere carbon cycle to A., 2013. Evaluation of the sea ice proxy IP25 against observational and diatom ice-covered and ice-free conditions in the Nordic Seas during the Last Glacial proxy data in the SW Labrador Sea. Quat. Sci. Rev. 79 (1), 53–62. Maximum. Palaeogeogr. Palaeoclimatol. Palaeoecol. 207 (1–2), 127–141. Weinelt, M., Vogelsang, E., Kucera, M., Pflaumann, U., Sarnthein, M., Voelker, A., Er- Scott, D.B., Vilks, G., 1991. Benthic foraminifera in the surface sediments of the lenkeuser, H., Malmgren, B.A., 2003. Variability of North Atlantic heat transfer deep-sea Arctic Ocean. J. Foraminiferal Res. 21 (1), 20–38. during MIS 2. Paleoceanography 18 (3). Slubowska-Woldengen, M., Koç, N., Rasmussen, T.L., Klitgaard-Kristensen, D., Hald, Xiao, X., Fahl, K., Stein, R., 2013. Biomarker distributions in surface sediments from M., Jennings, A.E., 2008. Time-slice reconstructions of ocean circulation changes the Kara and Laptev seas (Arctic Ocean): indicators for organic-carbon sources on the continental shelf in the Nordic and Barents Seas during the last and sea-ice coverage. Quat. Sci. Rev. 79 (1), 40–52. 16,000 cal yr B.P. Quat. Sci. Rev. 27 (15–16), 1476–1492. Zamelczyk, K., Rasmussen, T.L., Husum, K., Godtliebsen, F., Hald, M., 2014. Sur- Spielhagen, R.F., Erlenkeuser, H., Siegert, C., 2005. History of freshwater runoff across face water conditions and calcium carbonate preservation in the Fram Strait the Laptev Sea (Arctic) during the last deglaciation. Glob. Planet. Change 48 during marine isotope stage (MIS) 2, 28.8–15.4 kyr. Paleoceanography 29 (1), (1–3), 187–207. 1–12. Spielhagen, R.F., Werner, K., Sørensen, S.A., Zamelczyk, K., Kandiano, E., Budeus, G., Zhang, J., Rothrock, D.A., Steele, M., 1998. Warming of the Arctic Ocean by a Husum, K., Marchitto, T.M., Hald, M., 2011. Enhanced modern heat transfer to strengthened Atlantic Inflow: model results. Geophys. Res. Lett. 25 (10), the Arctic by warm Atlantic water. Science 331 (6016), 450–453. 1745–1748.