New evidence for preservation of contemporary marine organic carbon by iron in Arctic shelf sediments

The Barents Sea is located between 70 and 81° N off the northern Norwegian coast and is the largest of the six pan-Arctic shelf seas. It covers an area of 1.6 million square kilometres with an average water depth of 230 m (Carmack et al 2006). For a detailed descriptions about the modern climate setting and ecosystem of the Barents Sea, we refer to extensive overviews and reviews published during the past two decades (Loeng et al 1997, Wassmann et al 2006, Jakobsen and Ozhigin 2011, Smedsrud et al 2013, Dalpadado et al 2014, Jørgensen et al 2015). In brief, the oceanic circulation pattern of the western Barents Sea is dominated by the relatively warm northward-flowing North Atlantic Current (temperature 2 °C–8 °C, salinity >35‰) and cold Arctic currents (Spitsbergen and Persey; temperature <0 °C, salinity <35‰) entering the Barents Sea from the northeast. The relatively sharp boundary between these water masses forms the oceanographic polar front (figure 1) (Harris et al 1998) which is mainly determined by the shelf bathymetry and is, therefore, relatively stable from year to year (Drinkwater 2011). The northern Barents Sea is seasonally ice-covered with maximum and minimum ice coverage in March–April and August–September, respectively. The heat content of the Atlantic water keeps the southern Barents Sea permanently ice-free. River runoff into the Barents Sea is very limited. Only one larger river, the Petchora River, enters directly into the south-eastern Barents Sea in Russia. Rivers on the Kola Peninsula, on Svalbard and in Norway are small. Thus, sediment discharge through river inflow is low (e.g. Politova et al 2020) and the main processes responsible for Barents Sea surface sediment distribution are re-deposition by winnowing from shallow banks into troughs and depressions, and deposition from sea ice. Hence, sedimentation rates are generally low, 0.04–2.1 mm y⁻¹ since the last glacial period (Faust et al 2020), but can be much higher proximal to glacier outlets, for instance in places close to Svalbard. The present ecological setting, as in all Arctic shelf seas, is characterized by very pronounced seasonal fluctuations in insolation and primary production. Despite the relatively short duration of the growing season in the Arctic, the Barents Sea is a high productivity shelf area where 40% of the total primary production of the Arctic Ocean takes place (Sakshaug 2004). Throughout the western Barents Sea, oxygen penetration was repeatedly analysed by direct measurements and indirect indicators (e.g. pore water profiles), and was found to be between ∼2 and 6 cm below the sediment–water interface (Vandieken et al 2006, Nickel et al 2008, Freitas et al 2020, Stevenson et al 2020, Faust et al 2021). In addition, ²¹⁰Pb profiles and direct measurements of benthic faunal activity show a distinct sediment mixed layer in the Barents Sea that extends to a sediment depth of approximately 2 cm (Carroll et al 2008, Solan et al 2020).

Four sediment cores were collected by using a multi-corer along a south-north gradient in the western Barents Sea during the Changing Arctic Ocean Seafloor cruise (JR16006) in summer 2017 (figure 1). Sediment cores from station B13, B14, B15, B16 (supplementary table S1) were sliced in 0.5 cm intervals from 0 to 2 cm and in 1 cm intervals thereafter. Samples were stored in plastic bags at −20 °C immediately after recovery on-board the Royal Research Ship James Clark Ross. Prior to any chemical sediment analysis, all samples were freeze-dried and homogenized by gentle grinding using an agate mortar and pestle. Between two and four downcore samples from each sediment core were selected for radioactive carbon-14 content (Δ¹⁴C) and stable isotopic δ¹³C signature analysis (supplementary tables S2 and S3).

To quantify the amount of organic carbon bound to iron (oxyhydr)oxides in our samples, we applied a method described in detail by Lalonde et al (2012) and Salvadó et al (2015). Briefly, 0.25 g of sediment was transferred into 30 ml centrifuge tubes. Fifteen millilitre of a solution containing 0.27 M trisodium citrate (Na₃C₆H₅O₇·H₂O) and 0.11 M sodium bicarbonate (NaHCO₃) was added, mixed and heated to 80°C in a water bath; 0.1 M sodium dithionite (Na₂S₂O₄; 0.25 g) was added to the mixture, temperature was maintained at 80°C, and the tube was shaken every 5 min. After 15 min, the mixture was centrifuged for 10 min at 4000 rpm, the supernatant was decanted, and 200 μl of 12 N HCl were added to prevent Fe(III) precipitation. The remaining sediment samples were rinsed three times with artificial seawater and then freeze-dried. To quantify potential organic carbon loss unrelated to metal oxide dissolution, a control experiment was conducted: A 0.25 g aliquot of each sample was treated the same way as the reduction experiment, but the complexing and reducing agents (sodium citrate and sodium dithionite) were replaced with sodium chloride to reach a solution of the same ionic strength. All samples were weighed after the experiment to account for mass loss.

Organic carbon content of the bulk sediment before and after the reduction and control experiments was analysed on decarbonated samples using 10% (vol.) HCl, rinsed three times and dried overnight at 50°C. Organic carbon content was determined with a LECO SC-144DR combustion analyser at the University of Leeds, UK (Faust et al 2021). The certified reference material LECO 502-062 and blanks were included in every batch, and results are given in weight percentage. The relative error of the organic carbon analysis was ±1.7%. To account for the mass loss during the extraction experiment we applied the mass balance calculation of Salvadó et al (2015).

The δ¹³C and Δ¹⁴C of organic carbon associated with reactive iron were determined by difference from the control and the reductive leach residue and compared to bulk untreated sediments (supplementary tables S2 and S3). In more detail, bulk sediment samples before and after the reduction and control experiments were moistened with a small amount of deionised water, covered by glass fibre filter papers and placed into a sealed glass desiccator vessel together with a beaker of concentrated hydrochloric acid to hydrolyse any carbonate in the sample over three days via acid fumigation at the Natural Environment Research Council Radiocarbon Facility. Decarbonated samples were freeze-dried before combustion to CO₂ in a sealed quartz tube. After combustion, sample CO₂ was cryogenically recovered on a vacuum line. An aliquot of recovered sample CO₂ was used to measure δ¹³C on a dual inlet stable isotope mass spectrometer (Thermo Fisher Delta V). This value was used for normalisation of the measured sample ¹⁴C/¹³C ratio. A second aliquot of sample CO₂ was prepared to graphite and the sample ¹⁴C/¹³C ratio was measured by accelerator mass spectrometry at the Scottish Universities Environmental Research Centre AMS laboratory. The ¹⁴C enrichment of organic carbon in each sample was calculated as Δ¹⁴C_org based on the relative difference between the isotope ratio of the absolute international standard (relative to the year of measurement, 2021) and the age-corrected sample isotope ratio, where the latter was first normalised to-25% δ¹³C Vienna Pee Dee Belemnit (VPDB), using the measured δ¹³C_org value described above (Stuiver and Polach 1977). The fraction modern (F¹⁴C) is also reported where F¹⁴C = (Δ¹⁴C_org + 1000)/991.448 based on the correction factor for measurement in 2021. To quantify potential ¹⁴C contamination from the organic carbon extraction process, three process standards, modern (TIRI Barley Mash; TBM), background (Anthracite) and carbon free silica sand, were separately processed alongside the natural sediment samples in both the control and the reaction experiment. The aliquots of Anthracite standard show slightly elevated values above the laboratory blank (i.e. ¹⁴C added during sample treatment), indicating an addition of a small amount of modern carbon during the extraction procedure. Accordingly, we applied a background correction to samples and TBM standards of a value of F¹⁴C = 0.006 ± 0.001. This value was obtained via the average of measured values for Anthracite processing blanks. This addition of a small quantity of carbon during processing is consistent with results for silica blanks, which showed a carbon content of 0.01%–0.02%.

δ¹³C and Δ¹⁴C of the organic carbon bonded to reactive iron phases (δ¹³C-Fe_R; Δ¹⁴C-Fe_R; supplementary tables S2 and S3) were calculated assuming that the isotopic composition of the control sample (Δ¹⁴C-Fe_R control) is comprised of a mixture of organic carbon from reactive iron (the unknown Δ¹⁴C-Fe_R) and carbon forming the residue after the reductive removal of reactive iron (measured Δ¹⁴C-Fe_R reaction), such that mass balance leads to equations (1) and (2):

$$\begin{align}{\Delta ^{14}}{\text{C-F}}{{\text{e}}_{\text{R}}} = 1/{\text{fOC-F}}{{\text{e}}_{\text{R}}} *( {{\Delta ^{14}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{control}} - {\mkern 1mu} {\Delta ^{14}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{reaction}}} ) + {\mkern 1mu} {\Delta ^{14}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{reaction}}\end{align} \tag{ 1 }$$

$$\begin{align}{\delta ^{13}}{\text{C-F}}{{\text{e}}_{\text{R}}} = 1/{\text{fOC-F}}{{\text{e}}_{\text{R}}} *\left( {{\delta ^{13}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{control}} - {\mkern 1mu} {\delta ^{13}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{reaction}}} \right) + {\mkern 1mu} {\delta ^{13}}{\text{C-F}}{{\text{e}}_{\text{R}}}{\mkern 1mu} {\text{reaction}}\end{align} \tag{ 2 }$$

where fOC-Fe_R is the fraction of total organic carbon bond to reactive iron from Faust et al (2021).

To determine the ¹⁴C carbonate contents of benthic foraminifera, providing constraint on the ¹⁴C activity of marine dissolved inorganic carbon (DIC) at time of deposition, aliquots of bulk sediment samples were washed through a 63 μm sieve before being dried at 50°C, and microfossils were picked using a Leica MZ12 binocular microscope. All benthic foraminifera used for radiocarbon analyses were a mixture epifaunal or shallow infaunal species, dominated by rotaliids (>90%). Pyrgo and other miliolid species, which have previously been associated with anomalous radiocarbon measurements in the Arctic (Ezat et al 2017), were not analysed. The radiocarbon activities of the carbonate microfossils were obtained via the Mini Carbon Dating System at the Bristol Radiocarbon Accelerator Mass Spectrometry facility. Carbonate specimens were acidified to release CO₂ (<100 μgC), which was measured directly without graphitisation (Tuna et al 2018).

The stable carbon isotope signature of organic matter (δ¹³C_org) in marine sediments reflects the isotopic composition of the carbon source and the fractionation between ¹²C and ¹³C during photosynthesis (Hayes 1993). As the contribution of C₄ plant types is insignificant in the Arctic region (Collins and Jones 1986, Still et al 2003), δ¹³C_org can be a reliable proxy to identify marine versus terrigenous organic matter in Barents Sea sediments (e.g. Schubert and Calvert 2001). Marine organic carbon is isotopically enriched in ¹³C compared to terrestrial C₃ plant material (Arthur et al 1985) and typical endmember values are −20.1‰ and −26.1‰ for marine and terrigenous organic matter, respectively, in the Barents Sea region (Knies and Martinez 2009, Pathirana et al 2014). Our results show that δ¹³C_org signatures of all bulk untreated surface sediment samples vary between −22.1‰ and −23.7‰ (average −22.5‰) suggesting a bulk organic matter pool more dominated by marine organic matter (Martens et al 2021 and ref. therein). The δ¹³C_org values show a slight decrease with sediment depth (figure 2 and supplementary figure 1). The decline in δ¹³C_org is correlated with a Δ¹⁴C_org decrease (r = 0.8; supplementary figure S2) which occurs as total organic carbon contents and related biomarker records documenting relatively steady first-order organic matter decay with depth (Stevenson et al 2020, Faust et al 2021). This indicates no significant changes in the type of organic matter deposited at the seafloor, hence, we attribute these changes to a preferential degradation of labile marine organic matter, and not to a gradual change in organic carbon sources over time.

¹⁴C is commonly utilized for age determination of the carbon-bearing component. The ¹⁴C signature of marine organic matter (Δ¹⁴C_org) is a marker for modern, pre-aged or fossil carbon and, therefore, provides some information about the organic matter source, especially if combined with δ¹³C_org signatures. Marine surface sediments from circum-Arctic shelf regions reveal a substantial range of published Δ¹⁴C_org values, outlining large-scale differences in organic carbon sources to the present seafloor (Martens et al 2021). The Laptev and East Siberian Seas receive substantial carbon contributions from remobilization of thawing permafrost or other older deposits. In contrast, terrigenous organic carbon input in the Kara Sea is mainly contemporary (i.e. recent plant cover), and Barents and Chukchi Sea sediments are dominated by modern carbon derived from marine primary producers (Martens et al 2021).

The organic carbon Δ¹⁴C_org signatures of our bulk untreated Barents Sea sediment samples vary between −260‰ and −760‰ and decrease with sediment depth due to the time-dependant decay of ¹⁴C (figure 2, supplementary figure S1). Bulk Δ¹⁴C_org values in the first centimetre of each core are in accordance with published Δ¹⁴C_org data of bulk surface sediments from the western Barents Sea, which range from −245‰ to −504‰ (n = 11) (Martens et al 2021). Sediment cores from stations B13, B14 and B16 show very similar Δ¹⁴C_org values (−260‰, −292‰, −291‰) in the first centimetre. At station B15, Δ¹⁴C_org is lower (−558‰) as is δ¹³C_org. Due to a very shallow mean bioturbation depth (<2 cm) at all investigated stations (Carroll et al 2008, Solan et al 2020), the presence of old bulk organic matter at the seafloor can be explained by a combination of generally low sedimentation rates in the Barents Sea (figure 2), leading to long residence times of material at the seafloor (Griffith et al 2010), and the lateral transport of material across the shelf, which could include pre-aged organic matter from terrestrial or marine sources (e.g. Vonk et al 2014).

Dissimilatory iron reduction is an important process in the anaerobic degradation of organic matter in marine sediments. Iron (oxyhydr)oxides are reduced, and Fe²⁺ is released into the pore waters (Burdige 1993). Following upward diffusion out of the iron reduction zone, this Fe²⁺ is oxidised mainly by molecular O₂, but also by NO₃ ⁻ and solid Mn (oxyhydr)oxides (Burdige 1993). This oxidation usually occurs in the upper centimetres of a shelf sediment profile below the oxygenated, strongly bioturbated surface sediment layer, and leads to an authigenic enrichment of sedimentary Fe_R (Froelich et al 1978) below the sediment-water interface. Organic carbon has a strong affinity to such freshly precipitating Fe(III) phases (e.g. ferrihydrite), and it has therefore been proposed that coprecipitation of organic carbon with or adsorption to, this authigenic Fe_R within the sediment is the main OC-Fe_R coupling process that promotes the stabilization of sedimentary organic matter (Lalonde et al 2012, Riedel et al 2013, Chen et al 2014, Barber et al 2017, Chen and Sparks 2018). However, several recent investigations indicate that the fraction of the total organic carbon content bound to Fe_R (fOC-Fe_R) is not generally controlled by Fe_R availability (Sirois et al 2018, Faust et al 2020, 2021). Downcore fOC-Fe_R profiles in combination with pore water composition and sedimentary Fe_R contents reveal that iron redox cycling and associated authigenic Fe_R formation within the sediment are less important for the coupling of Fe_R to organic carbon than assumed (Faust et al 2021). Indeed, substantial amounts (>10%) of the total organic carbon content is bound to Fe_R at the sediment-water interface above the zone of authigenic Fe_R precipitation. This raises the question of how much of the OC-Fe_R is allochthonous, i.e. formed in the overlying water column, in sea ice, or on land; and how much is autochthonous, i.e. formed by biogeochemical processes within the sediments.

To further validate and characterise a potential allochthonous OC-Fe_R source in Arctic marine sediments, we compare our bulk Δ¹⁴C_org and δ¹³C_org signatures with those of Δ¹⁴C-Fe_R and δ¹³C-Fe_R as well as Δ¹⁴C content of benthic foraminifera. To identify the type and amount of organic carbon bound to Fe_R we conducted an iron oxide extraction (reaction experiment) based on the method originally developed by Mehra and Jackson (1958), modified for marine sediments by Lalonde et al (2012). To account for organic carbon released during the reaction experiment that was not related to iron phases, we conducted a sodium chloride extraction (control experiment) without the complexing and reducing agents trisodium citrate and sodium dithionate. The comparison of the δ¹³C_org and Δ¹⁴C_org values between the reaction experiment and control experiment reveal slightly more depleted values in the control experiment at station B13 (11.5 cm), B14 (22.5 cm) and B16 (2.5 cm, 6.5 cm; figure 2). Such differences have been reported before (e.g. Salvadó et al 2015) and they indicate that in some cases the sodium chloride treatment, washed out labile unbound organic matter. This could be an effect of different organic matter sources, degradation state or liberation of very labile OC-Fe_R during sodium chloride treatment (Fisher et al 2020). Nevertheless, following extraction treatment of our sediment samples, the isotopic signatures of Δ¹⁴C_org and δ¹³C_org in the solid residues of the control versus reaction experiments show clear differences from the bulk sediment samples at all stations (figure 2). The reaction and control experiment residues have mostly lower (more negative) Δ¹⁴C_org and δ¹³C_org values. It follows that the treatment liberated organic carbon that was relatively enriched in ¹³C and ¹⁴C from the sediments and selectively liberated organic matter with a more marine and younger isotopic signature. When we calculate the δ¹³C and Δ¹⁴C signatures of the organic carbon bound to reactive iron phases (δ¹³C-Fe_R; Δ¹⁴C-Fe_R, equations (1) and (2)), we find they are considerably higher than the respective values in most of the studied bulk samples (figure 2). To further investigate these isotopic disequilibria between organic carbon fractions of the bulk, control and reaction residues, we calculate the respective 'fraction modern' (F¹⁴C) offsets (see Soulet et al 2016 for details) of the contemporaneous carbon reservoirs of F¹⁴C-Fe_R, bulk organic carbon and carbonate (benthic foraminifera; figure 3).

Zoom In Zoom Out Reset image size

Benthic calcareous fossils (e.g. foraminifera) are commonly used to derive age-depth relationships for marine sediments over the last ∼50 ka. In comparison to bulk organic carbon, which represents a mixture of organic carbon from various sources, of various ages and different degradation states, the calcareous material of benthic organisms reflects bottom-water radiocarbon activity (i.e. a single carbon source) that is not significantly influenced by processes within the sediment. Thus, radiocarbon ages based on Δ¹⁴C content of benthic calcareous fossils are assumed to reveal the most accurate ages of sediment deposition (e.g. Skinner and Bard 2022). The ¹⁴C offsets of F¹⁴C-Fe_R and benthic foraminifera (carbonate) in relation to bulk organic carbon show that over time (downcore), the offsets between carbonate and bulk organic carbon show an increasing trend (figure 3). As bioturbation activity is low at all investigated stations (Solan et al 2020, Faust et al 2021), we assume that this is a function of decomposition and preferential degradation of fresh and young marine organic matter in the bulk sediment. However, we cannot completely exclude the possibility that the observed offset change is caused by a variation in local ¹⁴C reservoir effect i.e. bottom water dissolved inorganic carbon age changes e.g. due to oceanographic variability. The F¹⁴C-Fe_R offset is (apart from sample B16; 6.5 cm) always >1 and is relatively stable, ranging between 1.23 and 1.64. Thus, the organic carbon bound to iron is considerably younger than the bulk organic carbon. Moreover, the F¹⁴C-Fe_R offset relative to biogenic carbonate shows that F¹⁴C-Fe_R at the top of the sediment cores is of similar age, or even younger, than carbonate. The downcore decrease in the Fe_R-carbonate offset could be related to the loss of some fresh and labile organic carbon bound to iron. Alternatively, the type/source of the organic carbon bound to Fe_R could have changed in the past. But irrespective of the reason, our F¹⁴C-Fe_R data show that the organic carbon bound to iron is young in all samples, indicating rapid and lasting sequestration of CO₂ and a direct link to contemporaneous terrestrial and/or marine primary productivity.

Previous work on OC-Fe_R coupling in marine sediments further afield, has focused primarily on bulk carbon content, and/or δ¹³C_org (δ¹³C_org-Fe_R) in surface sediments (Lalonde et al 2012, Barber et al 2014, Ma et al 2018, Zhao et al 2018, Wang et al 2019). In accordance with our bulk δ¹³C_org and δ¹³C_org-Fe_R findings, a global data set of sediment from various depositional environments, including freshwaters, estuaries, river deltas, shelf sediments and the deep sea, shows that, in most cases, δ¹³C_org-Fe_R is enriched in ¹³C relative to the bulk organic matter (Lalonde et al 2012). The authors attributed the isotopic shift towards higher values to selective association of certain types of organic matter rich in proteins and carbohydrates, as these are more likely to establish protective inner-sphere complexes with reactive iron phases. Follow-up studies on surface sediments from the shelf areas of China indicated more complex δ¹³C_org-Fe_R signatures with values ranging from −49‰ to −4‰ (mean −23‰) (Zhao et al 2018, Wang et al 2019). Their finding of larger regions with substantially enriched and/or depleted δ¹³C_org-Fe_R values relative to bulk organic matter, and a trend of more enriched values towards the continental margin, indicated a preference for terrigenous organic carbon binding with iron. Presuming that the organic carbon–Fe_R bounding occurs mainly in the sediment, the authors suggested that the observed δ¹³C_org-Fe_R variability was caused by selective sequestration and release of ¹³C-depleted organic carbon either during the OC-Fe_R binding process or by its reduction under anoxic conditions (Wang et al 2019). Our downcore δ¹³C_org-Fe_R signatures from the Barents Sea do not support the assumption of preferential binding of terrigenous organic carbon in the oxic part of the core, although the slight downcore decrease (¹³C depletion) might indeed be related to selective release of ¹³C-enriched organic carbon. But the driving force(s) for the OC-Fe_R binding is still not known and it could be physical, chemical and/or biological mechanisms that initiate an isotopic fractionation. This hinders the assignment of the δ¹³C_org as organic carbon source indicator.

A more robust attempt to identify the type and origin of the organic carbon bound to Fe_R in marine sediment is a dual-carbon isotope approach similar to ours. To the best of our knowledge, only one study used both δ¹³C_org and Δ¹⁴C_org to evaluate the origin of the OC-Fe_R in marine sediments (Salvadó et al 2015). They showed that along the Eurasian Arctic shelf, δ¹³C_org-Fe_R and Δ¹⁴C_org-Fe_R signatures point towards an older and more terrestrial source in the Laptev Sea, probably related to coastal erosion and thawing permafrost. However, in areas where marine phytoplankton is an important sedimentary organic carbon source, e.g. in the East Siberian Sea and towards the outer shelf areas, their data imply a younger and marine plankton-dominated source of the organic carbon bound to Fe_R. Thus, the δ¹³C_org-Fe_R and Δ¹⁴C_org-Fe_R spatial variability in Chinese and Eurasian shelf sediments implies that the origin of the organic carbon varies distinctly with proximity to land and is related to the dominant organic matter source (marine versus terrigenous), of the bulk sedimentary composition.

Remarkable though is that, disregarding the spatial variability of the δ¹³C_org-Fe_R and Δ¹⁴C_org-Fe_R signatures in Eurasian Arctic shelf surface sediments, a re-examination of these data reveals that the δ¹³C_org and Δ¹⁴C_org offsets between bulk and Fe_R-bound organic carbon show F¹⁴C offsets >1 and enriched δ¹³C_org values compared to the bulk sediment composition the East- and West-East Siberian Sea (figure 4). In a similar way, the four sediment cores from the Barents Sea have organic carbon bound to Fe_R that is enriched in ¹³C and ¹⁴C compared to the bulk organic carbon content, irrespective of sediment depth/age. These findings indicate a rapid and preferential binding of fresh and marine organic matter with Fe_R. Thus, Fe_R not only protects organic matter from degradation in marine sediments over millennial time scales (Faust et al 2021), Fe_R also sequestered contemporary carbon across the Arctic Shelf which further highlights the potential efficiency of this 'rusty carbon sink'.

Zoom In Zoom Out Reset image size