Sea-ice dynamics in an Arctic coastal polynya during the past 6500 years


The production of high-salinity brines during sea-ice freezing in circum-arctic coastal polynyas is thought to be part of northern deep water formation as it supplies additional dense waters to the Atlantic meridional overturning circulation system. To better predict the effect of possible future summer ice-free conditions in the Arctic Ocean on global climate, it is important to improve our understanding of how climate change has affected sea-ice and brine formation, and thus finally dense water formation during the past. Here, we show temporal coherence between sea-ice conditions in a key Arctic polynya (Storfjorden, Svalbard) and patterns of deep water convection in the neighbouring Nordic Seas over the last 6500 years. A period of frequent sea-ice melting and freezing between 6.5 and 2.8 ka BP coincided with enhanced deep water renewal in the Nordic Seas. Near-permanent sea-ice cover and low brine rejection after 2.8 ka BP likely reduced the overflow of high-salinity shelf waters, concomitant with a gradual slow down of deep water convection in the Nordic Seas, which occurred along with a regional expansion in sea-ice and surface water freshening. The Storfjorden polynya sea-ice factory restarted at ~0.5 ka BP, coincident with renewed deep water penetration to the Arctic and climate amelioration over Svalbard. The identified synergy between Arctic polynya sea-ice conditions and deep water convection during the present interglacial is an indication of the potential consequences for ocean ventilation during states with permanent sea-ice cover or future Arctic ice-free conditions.


The sinking of dense waters on Arctic and Antarctic shelves through recurrent cooling and rejection of salt during sea-ice growth is a key contributor to global ocean circulation [25] with 10% of contemporary deep waters formed in the Arctic Ocean and the Barents Sea derived from these brine-enriched shelf waters [33]. High sea-ice production in Arctic coastal polynyas facilitates dense water production and ocean stratification, thus inhibiting the upward mixing of warm Atlantic water and sea-ice melt [1]. Coastal polynyas are persistent and recurrent areas of open water that occur within locations of otherwise consolidated and thicker ice cover. Amongst these, the Storfjorden coastal polynya in southern Spitsbergen (Fig. 1) is known to be an important sea-ice factory [19] and a significant source of brine rejection [33]. Dense brine-enriched waters from Storfjorden cascade downslope before flowing north [43], where they descend to depths of more than 2000 m [24] and account for up to 15% of the total dense water generated in the entire Arctic [13, 45]. However, this has likely changed in the past either as a contributor to, or because of climate change at high latitudes. Indeed, millennial-scale reconstruction of past brine formation in the Storfjorden polynya based on the sedimentary distribution of calcareous and agglutinated benthic foraminifera has revealed a systematic pattern of high (low) intensities during cold (warm) climate periods over the last 15,000 years [36, 37]. In contrast, large annual variability in brine formation has also been observed during the most recent warm periods during the last century. Thus, reduced brine formation, and hence, strongly reduced export of dense water to the Arctic Ocean occurred during periods with exceptionally warm Atlantic water advection and reduced sea-ice coverage in the Barents Sea, while intense brine formation was re-established during periods of recurrent cooling [60]. Accordingly, since the process of brine rejection is largely dependent on the seasonal formation of sea-ice, past reconstruction of sea-ice coverage coupled with environmental inferences from benthic foraminifera assemblages in the Storfjorden polynya [36, 37] provides a more direct indication of past brine formation, and thus potentially, a new measure for evaluating the significance of Arctic coastal polynyas with respect to dense water formation on a glacial-interglacial timescale. This approach provides an alternative to the still disputed use of benthic foraminiferal stable isotope records as a measure of the influence of brine-enriched shelf waters on deep water production [15, 29, 35]. In this study, we combine down-core records of organic geochemical biomarkers of sea-ice variability (IP25) [8] and open water phytoplankton (brassicasterol) with source-specific, sea-ice derived terrigenous sediments, supplemented by published agglutinated foraminifera data (% of total benthic foraminifera) [36, 37]. We hereby present evidence that changes in sea-ice coverage and inferred brine formation in the Storfjorden polynya over the past 6500 years coincide with past variability in deep water renewal in the Nordic Seas. As such, we highlight the importance of Arctic coastal polynyas as one significant driver of deep water renewal processes during the present interglacial.

Fig. 1

Study area and investigated marine sediment surface, floodplain and core samples. Major oceanographic features and the maximum sea-ice extent are indicated. Inset shows outline of Storfjorden Polynya (grey shaded) superimposed on the geology of Svalbard. Arsenic concentrations (ppm) in both floodplain sediments onshore and marine surface sediments are shown

Regional setting

Storfjorden, in southeastern Spitsbergen, is a ca. 200-km-long inlet, separated from the open ocean by a shallow sill (~120 m). Surface waters are seasonally stratified, with sea-ice and brine formation taking place each winter in the inner fjord [43]. Strong northeasterly winds blow sea-ice away from the eastern shelf, producing a large latent heat polynya, where high sea-ice production and continuous freezing generates cold (<−1.9 °C) and salty (34.8 to >35.8 psu) water [19, 45], which sinks and fills the central basin, finally overflowing the sill. Depending on its salinity (34.3–35.3 psu) [45], the brine may continue downslope reaching 2000 m into the deep-intermediate water of the Greenland Sea [24, 33]. Surface currents in Storfjorden are controlled by southwestward flowing, ice-covered polar waters from the Arctic Ocean. The East Spitsbergen Current (ESC) balances the bottom currents that transport the dense water out of Storfjorden towards the deep ocean. Sediments deposited in Storfjorden are enriched in organic carbon (up to 2.4 wt%) and largely dominated by terrigenous derived organic matter [58]. Terrigenous sediments are largely supplied by local (fast) ice entrainment processes and episodic freezing/melting processes in the polynya. Alternatively, terrigenous sediments transported by polar surface waters (ESC) to Storfjorden are released during frequent melting episodes and are deposited in Storfjorden [58].

Materials and methods

We studied inorganic elements and organic biomarkers in sediment surface samples (0–1 cm) taken with multicorer equipment from the western Barents Sea (Fig. 1; Table 1) and a gravity core JM10-10GC (77.41°N, 20.10°E, 123 m water depth, hereafter referred to as JM10), taken within the Storfjorden polynya where brines form today (Fig. 1). The surface samples were sliced onboard, frozen, and subsequently freeze-dried prior to analysis.

Table 1 Inorganic geochemical data from Barents Sea surface sediments

Inorganic geochemistry

All surface samples were analysed for major and trace elements using a Philips PW 1480 WD XRF instrument equipped with an Rh X-ray tube. For XRF major elements about 2 g of finely ground sample was preheated over a gas burner to remove any organic material before pre-ignition at 1000 ± 50 °C for at least 1 h. 4.200 ± 0.005 g Li2B4O7 (Claisse, Quebec, Canada) is mixed with 0.600 ± 0.005 g pre-ignited sample and fused to glass beads in Pt—5% Au-crucible. The method for determination of trace element with XRF is based on pressed pellets. 1.2 ± 0.005 g Hoechst wax was mixed with 5.4 ± 0.005 g dried and fine-ground sample material in a Spex Mixer/Mill for at least 1 min. The mixture was pressed to a pellet in a Herzog pelletizing press, with an applied force around 20 kN for 20 s. Methods accuracy for arsenic (As) and aluminium (Al) was tested with several certified reference materials (CRM), as shown in Tables 2 and 3. Relative percent difference between the duplicate samples was within ±10%. Al-normalisation was applied for As data in core JM10 to avoid dilution due to variable sedimentation rates (19–104 cm/ka) in the record [36]. A correlation coefficient R 2 = 0.88 between As/Al ratio and As concentrations (ppm) in core JM10 indicate no dilution effects on the As concentrations in the sediments.

Table 2 Methods accuracy for measuring As by XRF
Table 3 Methods accuracy for measuring Al by XRF

Concentrations of leachable elements in the same sample set was measured by ICP-AES with the instrument PerkinElmer 4300 DV. Nitric acid extraction was used to estimate the amounts of As and Al present in the nonsilicate fraction of the sediment in all surface sediments and core JM10. 1.000 ± 0.001 g of freeze-dried sediment was digested with 20 ml 7 M HNO3 for 30 min at 120 ± 4 °C in autoclave (CertoClav Sterilizer, CV-EL 18LGS), following the procedure described in the Norwegian Standard NS 4770 from 1994. After cooling, the sample was filtered through Whatman grade 597 and further diluted. The analysed solution contains 10 ppm Rh as internal standard and about 10% HNO3 (v/v). Method quantification limits, respectively, 20 mg/kg Al and 2 mg/kg As, is based on ten times the standard deviation for ten replicates of method blanks. Relative percent difference between the duplicate samples was within ±10%. Certified reference material Mess-3 (marine sediment for trace elements and other constituents, NRC-CNRC Canada) was routinely analysed to test methods analytical performance. The correlation coefficients between XRF and ICP-AES based arsenic and aluminium concentrations of 73 surface samples is r 2 = 0.95 and 0.75, respectively. Arsenic concentrations in the remaining text are based on the ICP-AES extraction method to allow comparison with published As concentration in floodplain and overbank deposits from Spitsbergen [32] (Fig. 1).


The biomarkers IP25 [8] and brassicasterol were quantified following addition of internal standards (9-octylheptadec-8-ene, 10 µL; 10 µg mL−1; 5α-androstan-3β-ol, 10 µL; 10 µg mL−1, respectively), extraction (DCM/Methanol; 3 × 3 mL, 2:1 v/v) and purification of extracts using silica column chromatography (IP25: hexane, 6 mL; brassicasterol: 20:80 methylacetate/hexane, 6 mL). Further purification of the IP25 containing fraction was achieved by Ag-ion chromatography (Supelco discovery Ag-Ion; 0.1 g) with saturated hydrocarbons (hexane; 1 mL) and unsaturated hydrocarbons (including IP25: acetone; 2 mL) eluted as two single fractions. All partially purified fractions were analysed using gas chromatography–mass spectrometry (GC–MS) according to established methods [5]. Brassicasterol was derivatized (BSTFA; 50 µL, 70 °C, 1 h) prior to analysis by GC–MS.


The chronology of the upper 325 cm of JM10 is based on 7 AMS 14C radiocarbon dates obtained on bivalves and monospecific samples of the benthic foraminiferal species N. labradorica (Table 4)  (see details in [37]). All AMS 14C dates were calibrated to calendar ages by applying the Calib7.02 program [47] and the Marine13 calibration curve [40]. The applied age model is consistent with the published model of Rasmussen and Thomsen [37]. The sedimentation rates vary between 19 and 104 cm/ka, with highest values (104 cm/ka) in the upper part of the sediment core (~1.0–0.5 ka BP), and lowest values (19 cm/ka) between ~2.8 and ~1.0 ka BP. Moreover, sedimentation rates in the lowermost part of the record (2.8 to ~6.5 ka BP) vary between 49 and 76 cm/ka. The quality of the dated material was checked by measuring bivalves and N. labradorica in two different samples within the same depth interval (324–326 cm). The dates are identical within error (Table 4), excluding the possibility of re-deposition of the bivalves in this environmental setting. However, we caution the reader that the observed changes in sedimentation rates between 2.8 and 1.0 ka BP are based on dating results from bivalves only, due to the lack of sufficient planktic or benthic foraminifera in this interval.

Table 4 AMS14C dates and calibrated dates for core JM10-10GC as published by Rasmussen and Thomsen [36, 37]

Results and discussion

Proxies for sea-ice dynamics

To interpret our down-core record, we first provide the background to our combined proxy data by presenting measurements obtained from surface sediments that reflect the modern physico-geography of the region. Arsenic (As) concentration in near-shore unpolluted marine sediments is normally between 5 and 10 ppm [54]. Sedimentary arsenic is principally associated with sesquioxide material (mostly hydrous iron oxides) as shown by a positive correlation between As and Fe (r 2 = 0.65). Arsenic concentration in our Barents Sea surface sediments varies between 2 and 105 ppm, with a clear geographical boundary along the Marginal Ice Zone (MIZ) (Fig. 2). South of the MIZ, the mean As concentration (7 ppm) resembles values in uncontaminated soils from northern Scandinavia [39], while for sites north of the MIZ, a mean concentration of 27 ppm is significantly higher than the global average for coastal marine sediments (5–10 ppm; [54]). The enrichment in the northern sediments is, however, probably not related to diagenetic redox-cycling processes seen in other shelf environments [48] since As anomalies are not correlated with other redox-sensitive elements such as Mn (r 2 < 0.2). Instead, it is more likely that natural sources of As-rich deposits and dissolved As in the water column are the causes of the sedimentary enhancements. As-rich sediments are most likely transported by sea-ice and released along the MIZ [21], while dissolved As can be taken up by phytoplankton blooms in the MIZ, and thus incorporated into the sedimentary cycle [10]. Indeed, local As anomalies are reported from Paleogene sequences, SW Spitsbergen (Fig. 1) [32] and As concentrations as high as 225 ppm have been recorded in coal seams interbedded with marine and lacustrine siltstones and shales [22]. Arsenic anomalies (>50 ppm) also occur in nearby floodplain sediments sourced from Carboniferous-Cretaceous organic-rich deposits along the coastline adjacent to Storfjorden (Fig. 1) [32]. Coastal freezing processes along the shoreline or within coastal polynyas [16] allow entrainment of As-enriched sediments in sea-ice with subsequent release during melt within the MIZ. Other As anomalies in sediments are reported from the Laptev Sea and Kara Sea shelves [21, 27, 28], where incorporation of As-enriched particles in newly formed sea-ice and transportation within the Transpolar Drift and East Spitsbergen Current may have caused the As anomalies identified below the MIZ in the northwestern Barents Sea (Fig. 2). Hence, we use the As anomalies in the sedimentary record as evidence for newly formed sea-ice that allowed incorporation of terrigenous (As-rich) particles in coastal areas, and subsequent sea-ice melting and release of As-rich ice-rafted sediments within the MIZ. To complement the As data, we also measured the distribution of the organic geochemical sea-ice proxy IP25 in the same surface sediments. IP25 is a highly specific lipid biosynthesized by certain diatoms residing in the underside of seasonal Arctic sea-ice [11] and whose presence and abundance in sediments is strongly associated with overlying sea-ice cover [6, 8], including the Barents Sea [7, 31]. In general, higher or increasing sedimentary abundances of IP25 are positively associated with seasonal sea-ice occurrence (or change) as shown through various surface and down-core records from across the Arctic [6]. However, lower IP25 abundances have been found in sediments from regions of much higher or near-permanent sea-ice cover including East Greenland [3] and the High Arctic (>80°N) [52, 59]. In such settings, the abundances of phytoplankton biomarkers, including brassicasterol, are also low; both observations being consistent with light-inhibited and, therefore, low biological productivity.

Fig. 2

Proxy data for modern sea-ice variability in the Barents Sea. Left Arsenic concentration (As in ppm) in Barents Sea surface samples. Right Sea-ice biomarker IP25 concentration in Barents Sea surface samples [31]. Storfjorden Polynya (stippled polygon), studied core position JM10 (green square), maximum of marginal ice zone (MIZ) (black line) and the Barents Sea polar front (stippled line) are indicated

Sea-ice dynamics in an Arctic coastal polynya

The accumulation of IP25 in the MIZ sediments [31] closely resembles the spatial distribution of As (Fig. 2) consistent with recurrent freezing and melting of sea-ice in the region. Furthermore, the release of sea-ice debris is known to stimulate phytoplankton blooms during spring, resulting in high export production rates during peak-bloom stages within the MIZ [38]. Through particle scavenging, this provides an additional mechanism that leads to enhanced sedimentary As. Down-core analyses of these sea-ice (IP25, As) and phytoplankton (brassicasterol) proxies (Fig. 3), therefore, provide a temporal measure of variable sea-ice coverage in the Storfjorden polynya, and by inference, changes in high-salinity brine rejection due to variable polynyal activity resulting from freezing/melting processes. The results are discussed for three different time intervals (6.5–2.8, 2.8–0.5, <0.5 ka BP), with the boundary at 2.8 ka based on the gradual decline of the IP25 concentration between 3.0 and 2.5 ka and the abrupt increase in percentages of agglutinated forams at this time (Fig. 4). Notched box-whisker plots for the distributions of As/Al, IP25, and brassicasterol in these time intervals (Fig. 3) confirm that, for all parameters, the median in the interval 2.8–0.5 ka is largely different from the median in the time intervals 6.5–2.8 and 0.5–0 ka on 5% level. However, on a 5% level, notched regions of As/Al distribution in intervals 2.8–0.5 and 0.5–0 ka do overlap (Fig. 3), implying that paleoenvironmental conditions for sedimentary As deposition during these time intervals were not significantly different compared to the interval 6.5–2.8 ka (see discussion below).

Fig. 3

Notched box-whisker plots for all down-core measurements of JM10-10GC for the parameters As/Al, IP25, and brassicasterol in the time intervals 0–500 a BP, 500–2800 a BP, and 2800–6500 a BP. White lines mark the estimated positions, and notched intervals the 95%-confidence limits for the medians of the distributions

Fig. 4

Proxy data for sea-ice variability in the Storfjorden polynya (core JM10-10GC) over the past 6500 years BP. Bottom to top: IP25 concentration (µg/gSed and µg/gTOC), As/Al ratio (×1000), agglutinated foraminifera (% of total benthic assemblages), and brassicasterol concentration (µg/gSed). Stippled lines indicate the mean values for each proxy in the three intervals discussed in the main text. Note that IP25 concentrations normalized to µg/g Sediment and µg/g TOC indicate no dilution effect on biomarker records due to variable sedimentation rates

Consistent with the surface sediment data, As/Al and IP25 co-vary in the 6500-year record (core JM10), with highest values between 2.8 and 6.5 ka, a decreasing trend towards 0.5 ka, and an increase towards the core-top (Fig. 4). The occurrence of IP25 at the core-top is consistent with modern observations of annual sea-ice formation in the polynya [19], while its presence throughout the record demonstrates persistent (but variable) seasonal sea-ice occurrence. Highest IP25 concentrations and As/Al ratios between 6.5 and 2.8 ka are accompanied by enhanced brassicasterol concentrations and lower relative abundances of agglutinated foraminifera (Fig. 4), implying a variable sea-ice margin and recurrent melting/freezing periods with associated phytoplankton blooms. These modern-like conditions, with seasonal sea-ice formation and increased polynyal activity, are in accordance with environmental inferences from calcareous and agglutinated foraminiferal assemblages in the fjord during this time interval [37]. Our proxy data are also consistent with simulations of increased sea-ice production (+15%) and extent (+14%) in the circum-Arctic [9], likely as a consequence of the flooding of the Arctic Siberian shelf [4] and potentially positive ocean-sea ice-atmosphere feedbacks in the Barents Sea [44], and further evidenced by reduced sea surface temperatures off western Svalbard around 5 ka [56] (Fig. 5). Elsewhere, a gradual southward expansion of the MIZ has been reconstructed for the Canadian Archipelago [52] and the Fram Strait [30]. Werner et al. [56] hypothesized that the occurrence of heavy winter sea-ice off the western Svalbard coast after 5.2 ka BP is due to established modern sea-ice production in the Arctic Ocean after the Holocene transgression. The distinct cooling trend in the Nordic Sea connected to the sea-ice expansion as a consequence of the flooding [9] and declining insolation [26] (Fig. 5) provides the prerequisite for the advection and persistent presence of seasonal sea-ice in the Storfjorden polynya.

Fig. 5

Sea-ice reconstruction, brine formation and deep water penetration to the Arctic over the past 6500 years. Bottom to top, IP25 concentration (µg/gSed) in Storfjorden, seawater-derived Nd isotope variations expressed as ɛNd in the eastern Fram Strait [55], planktic foraminifera δ13C from western Svalbard/Barents Sea [42, 56], benthic foraminiferal δ13C from Greenland Sea [51], June/July/August (JJA) air temperatures over Svalbard [14], sea surface temperatures (SST) off western Svalbard [46, 56, 57], Greenland ice core data from DYE-3, GRIP and NGRIP on the GICC05 timescale [53] and solar irradiance [26]

A distinct change in sea-ice coverage in the Storfjorden polynya occurred after 2.8 ka BP. While a seasonally fluctuating MIZ similar to its present (winter) location prevailed along western Spitsbergen [30], the reduced sea-ice and phytoplankton biomarkers together with higher mean proportions of agglutinated foraminifera (Fig. 4) demonstrated a clear change in sea-ice conditions in the Storfjorden polynya between 2.8 and 0.5 ka, with low entrainment/freezing of terrestrial sediments, diminished surface water productivity and dense/packed sea-ice coverage. At the same time, on the western Svalbard/Barents Sea margin, decreasing values in planktic δ13C records and a downward migration of the planktic foraminifera Neogloboquadrina pachyderma sin., also point to surface water freshening and saltier, warmer sub-surface waters (Fig. 5) [42, 57], thus preconditioning the setting for extensive sea-ice formation. The dominance of the calcareous benthic foraminifera species Elphidium excavatum in the Storfjorden sediments provides further evidence for more extensive seasonal ice cover [37]. A permanent sea-ice cover in Storfjorden is also in agreement with observations from western coastal Svalbard, where enhanced formation of shore-fast sea-ice and/or dense sea-ice coverage has been suggested [17]. On the other hand, pulses of advected Atlantic water along the Barents and Svalbard margin during this period [42, 55] did not influence the persistent sea-ice coverage in Storfjorden. However, confirmation of the latter requires a higher resolution IP25 record, as intervals with more variable sea-ice conditions inferred from highly fluctuating proportions of agglutinated foraminifera, which are not covered with the current IP25 dataset (Fig. 4). In the meantime, the high-resolution As/Al record of constantly low values (<0.5) throughout this interval implies dense sea-ice coverage, suggesting that increased proportions of agglutinated foraminifera in some intervals may reflect variable preservational conditions under the dominant influence of Arctic waters rather than strong polynyal activity, and thus brine formation. However, the latter needs to be explored further with additional records from the Storfjorden area and adjacent trough.

Rapidly increasing phytoplankton production, and enhanced IP25 concentrations demonstrate that the sea-ice factory restarted abruptly ~0.5 ka BP, at which time, sediment entrainment/release processes also recovered, with higher As/Al ratios towards the core-top (Fig. 4). The establishment of a highly fluctuating sea-ice boundary would have finally led to formation of a coastal polynya with seasonally variable sea-ice conditions. Enhanced IP25 and brassicasterol concentrations are largely consistent (except one interval centered on 0.3 ka) with increased proportions of agglutinated foraminifera (Fig. 6), supporting inferences by Rasmussen and Thomsen [36] of an intensified but variable polynyal activity. Thus, constantly high sea-ice production throughout the last ~500 years is likely the result of inferred mild summer temperatures on Spitsbergen including the Little Ice Age (Fig. 5) [14]. These modern-like conditions in Storfjorden, with variable sea-ice coverage over the last 500 years, contrast the more dense/packed sea-ice conditions in the preceding interval (~2.8–0.5 ka BP), but corroborate a recent biomarker-based sea-ice reconstruction for western Svalbard, which showed a gradual decline in spring sea-ice concentration over the past 400 years [12].

Fig. 6

Down-core variability of sea-ice (IP25), phytoplankton (brassicasterol), and agglutinated foraminifera indicators in Storfjorden polynya over the past ca. 500 years. Orange bars indicate the correspondence of high sea-ice variability, phytoplankton production and strong polynyal activity as inferred from higher proportions of agglutinated foraminifera [36]. Blue bar shows no response

Relationship between sea-ice, brines and deep water production

In modern times, it is well known that dynamic sea-ice production and brine rejection within the wind-driven polynyas in the circum-Arctic are important contributors for deep water convection in the Nordic Seas and Arctic Ocean [2, 43, 45]. Further, Bauch et al. [4] suggested that for the Last Glacial Maximum, enhanced sea-ice production and dense bottom water formation could be attributed to the formation of katabatic wind-driven polynyas in front of the western Svalbard-Barents Sea ice sheet. Similarly, based on calcareous and agglutinated foraminifera, Rasmussen and Thomsen [36, 37] showed that the strength of brine formation in the Storfjorden polynya over the last 15 ka BP was largely related to climatic conditions, with enhancements during cold periods (and vice versa). However, such studies were based on rather unselective proxies for sea-ice reconstruction (i.e. stable isotopes and assemblages of benthic and planktic foraminifera), which potentially limits their value in terms of confirming the significance of brine rejection on deep water formation in palaeo records (Fig. 5)

Table 5 Down-core variability of IP25 concentrations in JM10-10GC

In this study, we demonstrate temporal coherence between our more direct proxy-based sea-ice reconstruction (and inferred brine intensity changes) and changes to deep water convection obtained from local and other regional records from the Nordic Seas (Fig. 5). Thus, the recurrent freezing/melting of sea-ice in the Storfjorden polynya and associated strong brine formation between 6.5 and 2.8 ka BP coincides with less radiogenic ɛ Nd values (−9.4 to −10.6) from western Spitsbergen, as seen for present-day deep water penetration to the Arctic Ocean [55] (Fig. 5). During the same interval, high convection rates in most areas of the Nordic Seas is evident from high carbon isotope values in both planktic and benthic foraminifera [4, 42], together with a period of maximum ventilation in the Greenland Sea (Fig. 5) [50, 51] and AMOC strengthening [20]. In contrast, more permanent sea-ice cover and probably subdued brine formation in Storfjorden polynya after 2.8 ka, is accompanied by a prominent shift to more radiogenic ɛNd along the western Spitsbergen continental margin (Fig. 5) [55]. At the same time, freshening of surface waters and intensification (thickening) of sea-ice in the Fram Strait has been deduced from carbon isotope data of planktic foraminifera [56] (Fig. 5) and elevated IP25 abundances [30], while increased sea-ice production in the Arctic and export through Fram Strait also coincides with a proposed reduction of deep convection in the Greenland Sea [51] (Fig. 5). Modelling results also suggest that negative anomalies in total solar irradiance ~2.7 ka may have been responsible for local shutdown of deep water formation in the Nordic Seas at this time [41], which when superimposed on decreasing insolation (Fig. 5) may have stimulated positive oceanic feedbacks, such as enhanced stratification, expansion of sea-ice and less deep water formation leading to additional cooling and more sea-ice (e.g. [50, 51]. Regardless of the ultimate trigger for the abrupt changes in sea-ice coverage in Storfjorden polynya at ~2.8 ka, the timing of such solar-forced cooling events demonstrates that the most severe climatic conditions in the Nordic Seas and circum-Arctic reduced the contribution of Arctic sea-ice factories (i.e. polynyas) to deep water production.

The enhancement of the sea-ice factory and phytoplankton production in Storfjorden at ~0.5 ka BP, when recurrent freezing/melting of sea-ice in the polynya coincides largely with the increased admixture of deep waters from the Nordic Sea (less radiogenic ɛNd) (Fig. 5) and increased proportions of agglutinated foraminifera (Fig. 6), supports the notion of enhanced brine formation during stronger polynyal activity. The transition to more intense polynyal activity ~0.5 ka BP, coupled to higher sea-ice variability thereafter, also aligns with observations from western Svalbard, where spring sea-ice concentration has steadily declined over the past 400 years [12] and heat transport into the Arctic via the West Spitsbergen Current has increased [14, 46].

Implications and conclusions

The Arctic Ocean halocline is maintained by the contribution of cold and brine-enriched deep waters [1, 13], which are formed because of high sea-ice production in coastal polynyas over the continental shelves (Fig. 7) [49]. Tamura and Ohshima [49] showed that the current polar amplification of global warming will lead to negative trends in sea-ice production in most of the Arctic polynyas and with future projections of a summer ice-free Arctic Ocean (IPCC 2013), sea-ice factories in Arctic coastal polynyas may lose their significance entirely (Fig. 7). A likely cause for this trend could be delayed sea-ice freezing and increased Arctic air temperatures [49]. The last time a similar scenario occurred was during the Holocene Thermal Maximum when Arctic Ocean sea-ice cover was believed to be less than half of the minimum summer extent in 2007 [18]. Indeed, Årthun et al. [60] showed that during periods of maximum warming in the central Barents Sea, formation of brine-enriched shelf waters, and thus export of deep waters to the Nordic Seas and Arctic Ocean, was strongly reduced. Whether this reduced export contributed to the slow down in AMOC in the twentieth century [34] remains speculative. However, from this study we conclude that sea-ice production in Arctic coastal polynyas is highly sensitive to variable, externally forced climate or ocean feedback mechanisms. The correspondence between high (low) polynyal activity and variable sea-ice conditions in one important Arctic sea-ice factory, and observations of stronger (weaker) deep water renewal processes in the Nordic Seas during the present interglacial highlight the potential consequences for ocean ventilation during states with permanent sea-ice cover or future Arctic ice-free conditions.

Fig. 7

Location of coastal polynyas in the Arctic with variable sea-ice dynamics. a Modern sea-ice distribution and strong deep convections (crosses) with vigorous sea-ice factories (orange polygons) for brine-enriched shelf water formation. Blue arrows cold, ice-covered surface currents. Red arrows: warm, saline Atlantic-derived water masses. b Less sea-ice in the Arctic with shut down of sea-ice factories (open polygons) and slow down of deep convection (minus). NWP North Water polynya, NEWP Northeast Water polynya, NB Nordbukta, WBP Whaler’s Bay polynya, SP Storfjorden Polynya, KSP Kara Sea polynya, LSP Laptev Sea polynya


  1. 1.

    Aagaard K, Coachman L, Carmack E (1981) On the halocline of the Arctic Ocean. Deep Sea Res Part a Oceanogr Res Pap 28:529–545. doi:10.1016/0198-0149(81)90115-1

    Article  Google Scholar 

  2. 2.

    Aagaard K, Swift J, Carmack E (1985) Thermohaline circulation in the Arctic Mediterranean Seas. J Geophys Res 90:4833–4846

    Article  Google Scholar 

  3. 3.

    Alonso-Garcia M, Andrews JT, Belt S, Cabedo-Sanz P, Darby D, Jaeger J (2013) A comparison between multi-proxy and historical data (AD 1990––1840) of drift-ice conditions on the East Greenland shelf (~66°N). The Holocene 23:1872–1883

    Article  Google Scholar 

  4. 4.

    Bauch HA et al (2001) Chronology of the Holocene transgression at the North Siberian margin. Glob Planet Change 31:125–139

    Article  Google Scholar 

  5. 5.

    Belt S, Brown TA, Navarro-Rodriguez A, Cabedo-Sanz P, Tonkin A, Ingle R (2012) A reproducible method for the extraction, identification and quantification of the Arctic sea ice proxy IP25 from marine sediments. Anal Methods 4:705–713

    Article  Google Scholar 

  6. 6.

    Belt S, Müller J (2013) The Arctic sea ice biomarker IP25: a review of current understanding, recommendations for future research and applications in palaeo sea ice reconstructions. Quat Sci Rev 79:9–25. doi:10.1016/j.quascirev.2012.12.001

    Article  Google Scholar 

  7. 7.

    Belt ST, Cabedo-Sanz P, Smik L, Navarro-Rodriguez A, Berben SMP, Knies J, Husum K (2015) Identification of paleo Arctic winter sea ice limits and the marginal ice zone: optimised biomarker-based reconstructions of late Quaternary Arctic sea ice. Earth Planet Sci Lett 431:127–139. doi:10.1016/j.epsl.2015.09.020

    Article  Google Scholar 

  8. 8.

    Belt ST, Massé G, Rowland SJ, Poulin M, Michel C, LeBlanc B (2007) A novel chemical fossil of palaeo sea ice: IP25. Org Geochem 38:16–27. doi:10.1016/j.orggeochem.2006.09.013

    Article  Google Scholar 

  9. 9.

    Blaschek M, Renssen H (2013) The impact of early Holocene Arctic shelf flooding on climate in an atmosphere-ocean-sea-ice model. Clim Past 9:2651–2667. doi:10.5194/cp-9-2651-2013

    Article  Google Scholar 

  10. 10.

    Broecker WS, Peng T-H (1982) Tracers in the sea. Lamont-Doherty Geological Observatory Columbia University, New York

    Google Scholar 

  11. 11.

    Brown TA, Belt ST, Tatarek A, Mundy CJ (2014) Source identification of the Arctic sea ice proxy IP25. Nat Commun. doi:10.1038/ncomms5197

    Google Scholar 

  12. 12.

    Cabedo-Sanz P, Belt S (2016) Seasonal sea ice variability in eastern Fram Strait over the last 2000 years. Arktos 2:22. doi:10.1007/s41063-41016-40023-41062

    Article  Google Scholar 

  13. 13.

    Cavalieri DJ, Martin S (1994) The contribution of Alaskan, Siberian, and Canadian coastal polynyas to the cold halocline layer of the Arctic Ocean. J Geophys Res Oceans 99:18343–18362. doi:10.1029/94jc01169

    Article  Google Scholar 

  14. 14.

    D’Andrea WJ, Vaillencourt DA, Balascio NL, Werner A, Roof SR, Retelle M, Bradley RS (2012) Mild Little Ice Age and unprecedented recent warmth in an 1800 year lake sediment record from Svalbard. Geology 40:1007–1010. doi:10.1130/g33365.1

    Article  Google Scholar 

  15. 15.

    Dokken TM, Jansen E (1999) Rapid changes in the mechanism of ocean convection during the last glacial period. Nature 401:458–461. doi:10.1038/46753

    Article  Google Scholar 

  16. 16.

    Eicken H, Reimnitz E, Alexandrov V, Martin T, Kassens H, Viehoff T (1997) Sea-ice processes in the Laptev Sea and their importance for sediment export. Cont Shelf Res 2:205–233

    Article  Google Scholar 

  17. 17.

    Forwick M, Vorren T (2009) Late Weichselian and Holocene sedimentary environments and ice rafting in Isfjorden, Spitsbergen. Palaeogeogr Palaeoclimatol Palaeoecol. doi:10.1016/j.palaeo.2009.06.026

    Google Scholar 

  18. 18.

    Funder S et al. (2011) A 10,000-year record of arctic ocean sea-ice variability-view from the beach. Science 333:747–750. doi:10.1126/science.1202760

  19. 19.

    Haarpaintner J, Gascard J-C, Haugan PM (2001) Ice production and brine formation in Storfjorden, Svalbard. J Geophys Res 106:14001–14013

    Article  Google Scholar 

  20. 20.

    Hall I, Bianchi G, Evans J (2004) Centennial to millennial scale Holocene climate–deep water linkage in the North Atlantic. Quat Sci Rev 23:1529–1536

    Article  Google Scholar 

  21. 21.

    Hölemann JA, Schirmacher M, Kassens H, Prange A (1999) Geochemistry of surficial and ice-rafted sediments from the Laptev Sea (Siberia) Estuarine. Coast Shelf Sci 49:45–59. doi:10.1006/ecss.1999.0485

    Article  Google Scholar 

  22. 22.

    Jensen H (2000) Resultater av kjemiske analyser av prøver av Svalbard kull og tilgrensende bergarter over, under og mellom kull fløtsene. NGU, Trondheim

    Google Scholar 

  23. 23.

    Jochum KP, Willbold M, Raczek I, Stoll B, Herwig K (2005) Chemical characterization of the USGS reference glasses GSA-1G, GSC-1G, GSD-1G, GSE-1G, BCR-2G, BHVO-2G and BIR-1G using EPMA, ID-TIMS, ID-ICP-MS and LA-ICP-MS. Geostand Geoanalytical Res 29(3):285–302. doi:10.1111/j.1751-908X.2005.tb00901.x

    Article  Google Scholar 

  24. 24.

    Jungclaus JH, Backhaus JO, Fohrmann H (1995) Outflow of dense water from the Storfjord in Svalbard: a numerical model study. J Geophys Res Oceans 100:24719–24728. doi:10.1029/95jc02357

    Article  Google Scholar 

  25. 25.

    Killworth PD (1983) Deep convection in the World Ocean. Rev Geophys 21:1–26. doi:10.1029/RG021i001p00001

    Article  Google Scholar 

  26. 26.

    Laskar J, Robutel P, Joutel F, Gastineau M, Correia ACM, Levrard B (2004) A long-term numerical solution for the insolation quantities of the Earth. Astron Astrophys 428:261–285. doi:10.1051/0004-6361:20041335

    Article  Google Scholar 

  27. 27.

    Loring DH, Dahle S, Naes K, Dos Santos J, Skei JM, Matishov GG (1998) Arsenic and other trace metals in sediments from the Kara Sea and the Ob and Yenisey Estuaries. Russia Aquat Geochem 4:233–252. doi:10.1023/A:1009691314353

    Article  Google Scholar 

  28. 28.

    Loring DH, Næs K, Dahle S, Matishov GG, Illin G (1995) Arsenic, trace metals, and organic micro contaminants in sediments from the Pechora Sea, Russia. Mar Geol 128:153–167. doi:10.1016/0025-3227(95)00091-C

    Article  Google Scholar 

  29. 29.

    Mackensen A, Schmiedl G (2016) Brine formation recorded by stable isotopes of Recent benthic foraminifera in Storfjorden: palaeoceanographical implications. Boreas 45:552–566. doi:10.1111/bor.12174

    Article  Google Scholar 

  30. 30.

    Müller J, Werner K, Stein R, Fahl K, Moros M, Jansen E (2012) Holocene cooling culminates in sea ice oscillations in Fram Strait. Quat Sci Rev 47:1–14. doi:10.1016/j.quascirev.2012.04.024

    Article  Google Scholar 

  31. 31.

    Navarro-Rodriguez A, Belt ST, Knies J, Brown TA (2013) Mapping recent sea ice conditions in the Barents Sea using the proxy biomarker IP25: implications for palaeo sea ice reconstructions. Quat Sci Rev. doi:10.1016/j.quascirev.2012.11.025

    Google Scholar 

  32. 32.

    Ottesen RT et al (2010) Geochemical atlas of Norway, Part 2: Geochemical atlas of Spitsbergen. Chemical composition of overbank sediments. Norges geologiske undersøkelse/Norges vassdrags- og energidirektorat, Trondheim

    Google Scholar 

  33. 33.

    Quadfasel D, Rudels B, Kurz K (1988) Outflow of dense water from a Svalbard fjord into the Fram Strait. Deep Sea Res Part A Oceanogr Res Pap 35:1143–1150. doi:10.1016/0198-0149(88)90006-4

    Article  Google Scholar 

  34. 34.

    Rahmstorf S, Box JE, Feulner G, Mann ME, Robinson A, Rutherford S, Schaffernicht EJ (2015) Exceptional twentieth-century slowdown in Atlantic Ocean overturning circulation. Nat clim change. doi:10.1038/NCLIMATE2554

    Google Scholar 

  35. 35.

    Rasmussen TL, Thomsen E (2009) Stable isotope signals from brines in the Barents Sea: implications for brine formation during the last glaciation. Geology 37:903–906. doi:10.1130/g25543a.1

    Article  Google Scholar 

  36. 36.

    Rasmussen TL, Thomsen E (2014) Brine formation in relation to climate changes and ice retreat during the last 15,000 years in Storfjorden, Svalbard, 76–78°N. Paleoceanography 29:911–929. doi:10.1002/2014pa002643

    Article  Google Scholar 

  37. 37.

    Rasmussen TL, Thomsen E (2015) Palaeoceanographic development in Storfjorden, Svalbard, during the deglaciation and Holocene: evidence from benthic foraminiferal records. Boreas 44:24–44. doi:10.1111/bor.12098

    Article  Google Scholar 

  38. 38.

    Reigstad M, Carroll J, Slagstad D, Ellingsen I, Wassmann P (2011) Intra-regional comparison of productivity, carbon flux and ecosystem composition within the northern Barents Sea. Prog Oceanogr 90:33–46. doi:10.1016/j.pocean.2011.02.005

    Article  Google Scholar 

  39. 39.

    Reimann C, Matschullat J, Birke M, Salminen R (2009) Arsenic distribution in the environment: the effects of scale. Appl Geochem 24:1147–1167. doi:10.1016/j.apgeochem.2009.03.013

    Article  Google Scholar 

  40. 40.

    Reimer PJ et al (2013) IntCal13 and Marine13 radiocarbon age calibration curves, 0–50,000 years cal BP. Radiocarbon 55:1869–1887

    Article  Google Scholar 

  41. 41.

    Renssen H, Goosse H, Muscheler R (2006) Coupled climate model simulation of Holocene cooling events: oceanic feedback amplifies solar forcing. Clim Past 2:79–90

    Article  Google Scholar 

  42. 42.

    Sarnthein M, Van Kreveld S, Erlenkeuser H, Grootes PM, Kucera M, Pflaumann U, Schulz M (2003) Centennial-to-millennial-scale periodicities of Holocene climate and sediment injections off the western Barents shelf, 75 degrees N. Boreas 32:447–461. doi:10.1080/03009480310003351

    Article  Google Scholar 

  43. 43.

    Schauer U (1995) The release of brine-enriched shelf water from Storfjord into the Norwegian Sea. J Geophys Res Oceans 100:16015–16028. doi:10.1029/95jc01184

    Article  Google Scholar 

  44. 44.

    Semenov VA, Park W, Latif M (2009) Barents Sea inflow shutdown: a new mechanism for rapid climate changes. Geophys Res Lett. doi:1029/2009gl038911

  45. 45.

    Skogseth R, Haugan PM, Haarpaintner J (2004) Ice and brine production in Storfjorden from four winters of satellite and in situ observations and modeling. J Geophys Res Oceans. doi:10.1029/2004jc002384

    Google Scholar 

  46. 46.

    Spielhagen RF et al (2011) Enhanced modern heat transfer to the Arctic by warm Atlantic water. Science 331:450–453. doi:10.1126/science.1197397

    Article  Google Scholar 

  47. 47.

    Stuiver M, Reimer PJ (1993) Extended C-14 data-base and revised CALIB 3.0 C-14 AGE calibration program. Radiocarbon 35:215–230

    Article  Google Scholar 

  48. 48.

    Sullivan KA, Aller RC (1996) Diagenetic cycling of arsenic in Amazon shelf sediments. Geochim Cosmochim Acta 60:1465–1477. doi:10.1016/0016-7037(96)00040-3

    Article  Google Scholar 

  49. 49.

    Tamura T, Ohshima KI (2011) Mapping of sea ice production in the Arctic coastal polynyas. J Geophys Res. doi:10.1029/2010jc006586

    Google Scholar 

  50. 50.

    Telesiński MM, Bauch HA, Spielhagen RF, Kandiano ES (2015) Evolution of the central Nordic Seas over the last 20 thousand years. Quat Sci Rev 121:98–109

    Article  Google Scholar 

  51. 51.

    Telesiński MM, Spielhagen RF, Bauch HA (2014) Water mass evolution of the Greenland Sea since late glacial times. Clim Past 10:123–136. doi:10.5194/cp-10-123-2014

    Article  Google Scholar 

  52. 52.

    Vare L, Massé G, Gregory T, Smart C, Belt S (2009) Sea ice variations in the central Canadian Arctic Archipelago during the Holocene. Quat Sci Rev 28:1354–1366. doi:10.1016/j.quascirev.2009.01.013

    Article  Google Scholar 

  53. 53.

    Vinther BM et al (2006) A synchronized dating of three Greenland ice cores throughout the Holocene. J Geophys Res. doi:10.1029/2005jd006921

    Google Scholar 

  54. 54.

    Wedepohl KJ (1991) The composition of the upper earth’s crust and the natural cycles of selected metals. Metals in natural raw materials. Natural resources. In: Merian E (ed) Metals and their compounds in the environment. VCH, Weinheim, pp 3–17

    Google Scholar 

  55. 55.

    Werner K, Frank M, Teschner C, Mueller J, Spielhagen RF (2014) Neoglacial change in deep water exchange and increase of sea-ice transport through eastern Fram Strait: evidence from radiogenic isotopes. Quat Sci Rev 92:190–207. doi:10.1016/j.quascirev.2013.06.015

    Article  Google Scholar 

  56. 56.

    Werner K, Spielhagen RF, Bauch D, Hass HC, Kandiano E (2013) Atlantic water advection versus sea-ice advances in the eastern Fram Strait during the last 9 ka: multiproxy evidence for a two-phase Holocene. Paleoceanography 28:283–295. doi:10.1002/palo.20028

    Article  Google Scholar 

  57. 57.

    Werner K, Spielhagen RF, Bauch D, Hass HC, Kandiano E, Zamelczyk K (2011) Atlantic Water advection to the eastern Fram Strait—multiproxy evidence for late Holocene variability. Palaeogeogr Palaeoclimatol Palaeoecol 308:264–276. doi:10.1016/j.palaeo.2011.05.030

    Article  Google Scholar 

  58. 58.

    Winkelmann D, Knies J (2005) Recent distribution and accumulation of organic carbon on the continental margin west off Spitsbergen. Geochem Geophys Geosyst. doi:10.1029/2005gc000916

    Google Scholar 

  59. 59.

    Xiao X, Fahl K, Müller J, Stein R (2015) Sea-ice distribution in the modern Arctic Ocean: biomarker records from trans-Arctic Ocean surface sediments. Geochim Cosmochim Acta 155:16–29

    Article  Google Scholar 

  60. 60.

    Årthun M, Ingvaldsen RB, Smedsrud LH, Schrum C (2011) Dense water formation and circulation in the Barents Sea. Deep Sea Res Part I 58:801–817. doi:10.1016/j.dsr.2011.06.001

    Article  Google Scholar 

Download references


This work is a contribution to the CASE Initial Training Network funded by the European Community’s 7th Framework Programme FP7 2007/2013, Marie-Curie Actions, under Grant Agreement No. 238111. The research is part of the Centre for Arctic Gas Hydrate, Environment and Climate and was supported by the Research Council of Norway through its Centres of Excellence funding scheme Grant No. 223259. We thank the reviewers for their help improving the manuscript significantly.

Author information



Corresponding author

Correspondence to Jochen Knies.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Knies, J., Pathirana, I., Cabedo-Sanz, P. et al. Sea-ice dynamics in an Arctic coastal polynya during the past 6500 years. Arktos 3, 1 (2017).

Download citation


  • Arctic
  • Storfjorden
  • Polynya
  • Holocene
  • Sea ice