Centennial-scale variability of sea-ice cover in the Chukchi Sea since AD 1850 based on biomarker reconstruction

Figure 3 shows the downcore profile of TOC (%), biogenic silica (%), TN (%), C/N molar ratio, δ¹³C (‰) and δ¹⁵N (‰) in the C07 core since AD 1850. TOC values range from 1.46% to 2.93% at the core top, with a mean value of 1.98%, and show a gradual increase towards present except for lower values from the late 1970s and early 1980s (figure 3(a) and table S2). Biogenic silica shows broadly similar trends as TOC except for a sharp decay in the uppermost layer after a maximum of 17.3% at AD 1990 (figure 3(b)). TN content increases in a narrow range (0.15%–0.34%) with lower values in the 1970s and at the end of the 19th century, similar to TOC. The C/N ratio show more variability between 7.8 and 11.5 with a mean value of 9.6. The δ¹³C values of organic matter span from −22.3‰ to −23.6‰ at the core-top with a mean value of −22.7‰. The δ¹⁵N values vary between 7.7‰ and 8.1‰ at the core-top with little changes downcore (figures 3(c)–(f)).

Zoom In Zoom Out Reset image size

Generally, all biomarker concentrations show a broad increase towards present (figure 4 and table S2). IP₂₅ is generally low until AD 1930, except for a brief increase around AD 1870 (1.4 μg g⁻¹ TOC) coincident with decreasing TOC and biogenic silica values, followed by a gradual increase accelerating in the 1960s to reach highest values in the 1990s (4.2 μg g⁻¹ TOC) (figure 4(a)). The downcore profile and range of concentrations of HBI II and IP₂₅ are very similar, while HBI III is an order of magnitude lower, ranging from 0.01 to 0.47 μg g⁻¹ TOC (figures 4(b) and (c)). Brassicasterol abundances are low until the abrupt increase in the 1990s to a maximum of 27.9 μg g⁻¹ TOC (figure 4(d)). The dinosterol profile is quite different and shows strong decadal-scale variability that share resemblance with terrigenous biomarkers (campesterol and β-sitosterol) (figures 4(e) and (f)).

Zoom In Zoom Out Reset image size

The high TOC content of the core top (2.93%) falls in the upper range of TOC values reported in surface sediments across the Chukchi Sea shelf (1.25%–2.56%) by Goñi et al (2013) (table S2). TN values range from 0.15 to 0.34% (table S2) and show significant positive correlation with TOC (r ² = 0.68, p< 0.01, figure 5) suggesting similar sources and fate of C and N. The C/N ratio has commonly been used to distinguish between algal and terrigenous organic matter (OM) (Hedges et al 1986). Fresh marine algae typically exhibit C/N ratios around 4–6 whereas in vascular plants these values are above 20 (Meyers 1994). C/N ratios >10 typically indicate significant terrestrial influence (Grebmeier et al 1989). Downcore C/N values at C07 vary between 7.8 and 11.5 (mean value of 9.5, table S2) indicating notable and increasing contribution of land-derived inputs since the late 1970s likely due to enhanced river run-off resulting thawing permafrost and ground ice melting induced by Arctic warming. The SCC flowing eastward towards the Chukchi shelf where it is diverted offshore by PW could occasionally contribute to the transport of terrestrial material to the core site (green arrows in figure 1). In addition, the central current BSW (blue arrow in figure 1) is likely to carry fine sediment particles from the Bering Strait and shallow southern Chukchi shelf to C07 site (Weingartner et al 2005, Darby et al 2009, Yamamoto et al 2017). A substantial amount of terrestrial organic matter to marine sediments may have be delivered via the melting of sea ice (Eicken et al 2005). Lalande et al (2007) also underlined commonly found high concentrations of lithogenic material in the lower layers of sea ice implying substantial contribution to vertical fluxes.

Zoom In Zoom Out Reset image size

The δ¹³C provides complementary information on organic carbon sources. Previous evaluation of isotopic end-members in the northern Bering and Chukchi Seas indicates distinctly δ¹³C values for marine phytoplankton (−21.2‰) and terrigenous C₃ plant material (−27.0‰) (Naidu et al 2000). However, it is important to note that the δ¹³C values of OM from sea ice algae can vary between −15‰ to −18‰ (Schubert and Calvert 2001). Belt et al (2008) further reported that the mean δ¹³C value for IP₂₅ (−19.3‰) lies at the heavily ¹³C-enriched marine end-member for Arctic OM. Therefore, δ¹³C downcore values of −22.3‰ to −23.6‰ are likely indicative of marine sources with some contribution of isotopically lighter terrestrial organic material. More depleted δ¹³C values towards present are also indicative of enhanced inputs from the land over the last decades as also inferred from C/N ratios (figure 3(d)) and higher amounts of campesterol and β-sitosterol, in particular since the 1960s (figure 4(f)).

The downcore profile of δ¹⁵N values, spanning from 7.7‰ to 8.1‰, indicate a long-term increase. The δ¹⁵N values of OM produced by plankton range between 7‰ and 10‰, whereas in C3 higher plants they are comprised between 0‰ and 2‰ (Meyers 1997). The δ¹⁵N trend may feature enhanced nutrient utilization by phytoplankton due limited nutrient availability under stratified conditions triggered by sea ice melting as postulated by Lee et al (2012) based on N and C uptake rates. Increasing riverine freshwater inputs also contribute to ocean stratification and freshening of the Arctic Ocean (Li et al 2009). A recent modeling study has shown that riverine water discharge would also impact significantly on the heat budget of the Arctic Ocean and consequently contribute to the sea-ice decline (Park et al 2020).

As can be seen from figure 4, sympagic HBIs (IP₂₅ and HBI II) and pelagic biomarkers (HBI III, brassicasterol and dinosterol) display similar trends but at different concentration levels. The presence of IP₂₅ and pelagic biomarkers throughout the core indicates seasonal sea-ice conditions at C07 since AD 1850 (Bai et al 2019, Hu et al 2020). All HBIs occur at extremely low levels in the older part of the record pointing out extensive sea-ice cover, except for a short interval of higher values between AD 1865 and AD 1875 (figures 4(a)–(c)). These results are consistent with historical sea ice data from whaling ship logbooks throughout the northern Bering Sea reporting more severe sea-ice conditions in the 1850s to 1860s, comparable to the 1970s taken as a reference for recent cold conditions (Mahoney et al 2011). Indeed, while winter sea-ice extent was similar during both periods, summer sea-ice extended further South in the Bering Strait in the 1850s to 1860s due to thicker sea ice and slower melting (Mahoney et al 2011). The logbooks also describe a period of sea-ice retreat to the northern Bering Strait, which corresponds to sympagic HBIs and HBI III peaks found at AD 1865–1875 in our core (Mahoney et al 2011; figure 4). Subsequent decades of the early 1900s are characterized by lower sympagic HBIs caused by a temporary return to colder conditions until a first stepwise increase at AD 1930 and second one in the 1990s (figures 4(a) and (b)). Stable HBI values in the 1940–1960s are consistent with temporarily cold conditions reported by Mahoney et al (2011).

The IP₂₅ profile of the neighboring core ARA01B-03MUC (73.52 °N, 168.94 °W, 72.5 m water depth; figure 1) located northwest of C07 shows similar features (Kim et al 2019) (figure 6(a)). However, HBI III and brassicasterol decrease earlier in the upper ARA01B-03MUC, while IP₂₅ declines synchronously in both cores. High brassicasterol together with high IP₂₅ in ARA01B-03MUC since AD 1950 are indicative of sea-ice edge conditions. This is further supported by the co-eval lower and decreasing abundance of HBI III attesting the retreat of sea ice (0.15–0.07 µg g⁻¹ TOC for ARA01B-03MUC versus 0.2–0.5 µg g⁻¹ TOC for C07) (figure 6(b)).

Zoom In Zoom Out Reset image size

The downcore profile of dinosterol is different from that of sympagic HBIs and brassicasterol (figures 4(d) and (e)). Low brassicasterol and HBI III from the late 19th century until AD 1920–1930 are coherent with extensive sea ice in the Chukchi Sea (figure 4). However, high amounts of dinosterol (i.e. >10 μg g⁻¹ TOC) are surprising and could reflect other source organisms such as sea-ice microalgae as reported by Nichols et al (1990) in sea ice samples from Antarctica. Note that the downcore profile of brassicasterol and dinosterol in ARA01B-03MUC show similar differences (Kim et al 2019). Another common feature in C07 and ARA01B-03MUC is the recent decline of IP₂₅ and brassicasterol, supporting accelerating summer sea ice retreat in the Chukchi Sea driven by climate change in recent decades (figures 6(a) and (c)).

Increased terrigenous sterols (campesterol and β-sitosterol) are consistent with C/N and δ¹³C data and further evidence high and increasing river run-off (figure 4(f)). With the decline of sea ice, airborne continentally-derived material from adjacent continents could also contribute more to Arctic sediments. Indeed, in addition to fluvial deposits, terrestrial constituents transported over long distances from their emission sources as aerosols accumulate in sea ice and get released during the melting season. This is evidenced by terrigenous n-alkanes at station DM (figure 1) showing high export fluxes as sea ice breaks-up in summer (Bai et al 2019). Ren et al (2020) also reported similar results with high terrigenous campesterol and β-sitosterol fluxes in summer contrasting with no detectable export in winter.

Semi-quantitative estimates of sea ice can be calculated from the PIP₂₅ index by combining sympagic IP₂₅ with pelagic phytoplankton biomarkers using the following expression:

$$\begin{equation*}{\text{PI}}{{\text{P}}_{25}} = \frac{{\left[ {{\text{I}}{{\text{P}}_{25}}} \right]}}{{\left[ {{\text{I}}{{\text{P}}_{25}}} \right] + \left[ {{\text{phytoplankton biomarker}}} \right] \times c}}\end{equation*}$$

$$\begin{align} & {\text{with}} \nonumber \\ & c = \frac{{{\text{mean I}}{{\text{P}}_{{\text{25}}}}{\text{ concentration}}}}{{{\text{mean phytoplankton biomarker concentration}}}}\end{align}$$

where c is a balance factor accounting for the different magnitude concentrations between IP₂₅ and phytoplankton biomarkers, such as brassicasterol and dinosterol that were used as the phytoplankton end-member to derive P_BIP₂₅, P_DIP₂₅. More recently, HBI III has also been proposed as an alternative pelagic phytoplankton end-member to calculate P_IIIIP₂₅ index.

Indeed, surface sediments from the Barents and Chukchi seas have shown elevated abundances of HBI III in the vicinity of the sea-ice edge or in the MIZ (Belt et al 2015, Smik et al 2016, Ribeiro et al 2017, Belt 2018, Bai et al 2019). Yet, different relationships of this compound to sea ice have been postulated. For example, maximum HBI III in surface sediments from the western Barents Sea has been associated with winter sea-ice extent and/or the MIZ (Belt et al 2015) while Ribeiro et al (2017) correlated HBI III in East Greenland fjord sediments to mid-July MIZ conditions. Uncertainties still remain as to the HBI III producers and habitat mainly because of the lack of modern ocean observations. Knowledge on HBI III has been essentially inferred from correlations between concentrations in surface sediments and SIC at different seasons. The only sediment trap data on HBI III in the Arctic Ocean are from Bai et al (2019) at station DM (figure 1) that shows high fluxes at moderate sea-ice cover (<50%) and their decrease towards ice-free conditions that are consistent with production by sea-ice edge thriving diatoms. Furthermore, cross-plots of HBI III against IP₂₅ downcore concentrations exhibit a stronger positive relationship than brassicasterol against IP₂₅ (figure S2) in agreement with HBI III production being closely associated with the sea-ice edge. This interpretation is also supported by co-eval higher fluxes of the IP₂₅ and HBI III from end of July to early August at the station DM (Bai et al 2019). Synchronous changes of IP₂₅ and HBI III have been further interpreted by Belt (2018) as reflecting rapidly changing sea ice (advancing and retreating) such as for example the two-step concomitant rise of HBI III and IP₂₅ in the1970s and early 1990s in our core (figures 6(a), (b) and 7; Smik et al 2016, Belt 2018, Bai et al 2019).

Zoom In Zoom Out Reset image size

HBI III, as well as brassicasterol and dinosterol were thus used to calculate PIP₂₅ indexes and derive semi-quantitively sea ice changes since AD 1850. On average, concentrations of brassicasterol were about five times higher, dinosterol about twelvefold higher and HBI III about 12 times lower than IP₂₅, resulting in balance factor c of 0.20, 0.08 and 11.74, respectively. Downcore P_BIP₂₅ values vary between 0.19 and 0.79, which according to Müller et al (2011), correspond to low sea ice to MIZ conditions, with only one value above the permanent sea-ice cover threshold (i.e. P_BIP₂₅ >0.75) (figure 6(d)). The sharp drop in the 1920s −1930s suggests a transient period of sea-ice retreat (figure 7(a)). Since AD 1930, P_BIP₂₅ values increase from 0.54 to 0.74, indicating proximal MIZ conditions except for the last decade of accelerated decline (figure 6(d) and table S2). Indeed, P_BIP₂₅ shows a decreasing trend towards less/variable sea-ice cover since the late 1990s as expected from the recent unprecedented decline of sea ice in the northern Chukchi Sea (Walsh et al 2016). P_IIIIP₂₅ values show different trends and variability that share more resemblance with the sea ice extent in the Chukchi Sea of Walsh et al (2016) with relatively high and stable values from the 1880s to the 1920s followed by a continuing decrease reflecting the progressive and long-term retreat of seasonal sea ice extent (figure 6(d) and table S2). Yet, the drastic reduction of the last decades is not reproduced by the P_IIIIP₂₅ index. This result can be explained by HBI III production tending to zero as surface waters become ice-free during summer as opposed to brassicasterol (Bai et al 2019). The P_DIP₂₅ values show similar features as P_BIP_25, despite discrepancies between dinosterol and brassicasterol profiles, except since the 1980s when P_DIP₂₅ indicates increased sea ice contrary to observations (Walsh et al 2016) (figures 4(d), (e) and 6(d)). Diverging spatial distribution of these two sterols has also been underlined by Su et al (2022) in the Eastern Siberian Sea. It has been speculated that dinosterol in the Arctic Ocean might be produced by other organisms than dinoflagellates (Nichols et al 1990). Due to remaining source uncertainties of this sterol in the Arctic Ocean and inconsistent trends at moderate to low sea ice, P_DIP₂₅, was not further considered for sea-ice reconstruction.

Finally, the tri-unsaturated HBI ratio (HBI TR₂₅) was also calculated using the following equation of Belt et al (2019) (HBI TR₂₅= [Z-triene]/([Z-triene] + [E-triene])) to explore this new biomarker as a potential proxy of spring phytoplankton bloom in the Arctic and sub-Arctic. The downcore values of HBI TR₂₅ are all below the spring phytoplankton bloom threshold of 0.62 from which we can conclude that primary production did not reach high levels as suggested by the δ¹⁵N values pointing out limited nutrient availability of surface waters (figure 6(e)).

4.4.1. SIC reconstruction

4.4.2. Key physical drivers

4.4.3. Links with the AO

4.4.4. Links with the PDO

4.4.5. Links with the AMO and AMOC