Iron cycling in Arctic methane seeps

Anoxic marine sediments contribute a significant amount of dissolved iron (Fe2+) to the ocean which is crucial for the global carbon cycle. Here, we investigate iron cycling in four Arctic cold seeps where sediments are anoxic and sulfidic due to the high rates of methane-fueled sulfate reduction. We estimated Fe2+ diffusive fluxes towards the oxic sediment layer to be in the range of 0.8 to 138.7 μmole/m2/day and Fe2+ fluxes across the sediment-water interface to be in the range of 0.3 to 102.2 μmole/m2/day. Such variable fluxes cannot be explained by Fe2+ production from organic matter–coupled dissimilatory reduction alone. We propose that the reduction of dissolved and complexed Fe3+ as well as the rapid formation of iron sulfide minerals are the most important reactions regulating the fluxes of Fe2+ in these cold seeps. By comparing seafloor visual observations with subsurface pore fluid composition, we demonstrate how the joint cycling of iron and sulfur determines the distribution of chemosynthesis-based biota.


Introduction
Iron is a critical micro-nutrient for marine phytoplankton and photosynthesis in the surface ocean (e.g., Boyd et al. 2000). As one of the most bio-available iron species, inorganic aqueous iron is supplied to the ocean by rivers, aeolian transport, glacial meltwater, icebergs, and hydrothermal vents and from anoxic marine sediments (Raiswell and Canfield 2012;Tagliabue et al. 2017). In the absence of organic ligands, Fe(OH) 3 0 is the primary inorganic aqueous iron species at pH 8 whereas Fe 2+ dominates the aqueous iron pool under anoxic conditions (Raiswell and Canfield 2012). In anoxic marine sediments, Fe 2+ is mostly produced through the reductive dissolution of iron (oxyhydr)oxide driven by the decomposition of particulate organic matter (POC) (or dissimilatory iron reduction (DIR), hereafter) (Lovley and Phillips 1988). The rates of Fe 2+ production through this process depend on factors such as the quantity/type of organic matter, bottom seawater dissolved oxygen concentration (Lyons and Severmann 2006;Dale et al. 2015), the reactivity of different iron minerals (Raiswell and Canfield 1998;Larsen and Postma 2001), and solution pH (Straub et al. 2001). It has been well documented that large amounts of sulfide produced during sulfate reduction can effectively scavenge Fe 2+ to form authigenic iron sulfide minerals, a process that has been commonly documented along global continental margins (Schulz et al. 1994;Reimers et al. 1996;Niewöhner et al. 1998;Riedinger et al. 2004;Raiswell and Anderson 2005;Riedinger et al. 2005;Lim et al. 2011;Fischer et al. 2012;Riedinger et al. 2017). The Fe 2+ that is not consumed through the precipitation of iron sulfide minerals diffuses towards the oxic sediment layer where it is partially oxidized to form iron (oxyhydr)oxide precipitates. The small fraction of Fe 2+ that eventually Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00367-020-00649-5) contains supplementary material, which is available to authorized users. escapes the sediments can enhance the precipitation of ferrihydrite (Fe 4 HO 8 •4H 2 O) in the bottom water and can be transported by bottom currents (Lyons and Severmann 2006;Raiswell and Canfield 2012;Lenstra et al. 2019). Seafloor macrofauna is known to either stimulate Fe 2+ release to the bottom sediment or enhance Fe 2+ consumption through oxidation in the surficial sediments, depending on the process involved. Bioturbation and bioirrigation increase the burial of labile organic matter and iron oxides which consequentially accelerates Fe 2+ production through DIR (Canfield et al. 1993;Aller 1994). The active exchange between pore fluid and the overlying seawater as a result of faunal activities facilitates the escape of excess Fe 2+ (Severmann et al. 2010). On the other hand, the introduction of oxygen into the sediment, as a result of pumping seawater by seafloor animals, also enhances the oxidation of Fe 2+ and thus consumption (Aller 1980;Aller 1982). Dale et al. (2015) have shown that Fe 2+ fluxes from continental margin sediments were previously underestimated and more data are needed to refine global iron budgets. Cold seep sediments, where large quantities of methane fuel high rates of sulfate reduction, are an ideal environment to study the cycling of Fe 2+ under persistent anoxic and sulfidic conditions, a case that resembles conditions of fast sulfate reduction due to high organic matter turnover rate along productive continental margins. The chemosynthesis-based macrofauna, whose survival depends on the release of reduced compounds from the cold seep sediments, may also play a significant role in determining the fate of Fe 2+ in surficial sediments. Here, we investigate iron cycling from four Arctic methane seeps along the northern Norwegian margin in water depths ranging from 220 to 380 m ( Fig. 1), where geochemical data are scarce to constrain the fate of iron in the sediments. We present porewater and sediment data to discuss the reactions and processes that affect Fe 2+ cycling in shallow methane-rich anoxic sediments. In addition, we show how iron and sulfide fluxes impact the distribution of seafloor chemosynthesis-based biota, and vice versa.

Study areas
The investigated cold seeps are located along the glaciogenic northern Norwegian continental margin (Fig. 1). As one of the largest trough fans south of Svalbard (Lucchi et al. 2012), the major sediment types from Storfjordrenna are glacigenic diamictites and texturally heterogeneous marine sediments (Lucchi et al. 2012) as the result of repeated growth and retreat of grounding glaciers shaping the seafloor (Patton et al. 2015). The release of methane, shallow subsurface occurrence of gas hydrate, and chemosynthesis-based siboglinid frenulate polychaetes were documented from Storfjordrenna in recent papers (Hong et al. 2017;Sen et al. 2018a, b). Densely distributed craters and elevated methane concentrations (20-60 nM) in the bottom water have recently been reported from the Bjørnøyrenna area in the north central Barents Sea (Long et al. 1998;Andreassen et al. 2017), where ice sheets have carved the seafloor during the last glaciation and exposed the Middle Triassic bedrock (Long et al. 1998;Andreassen et al. 2017). Offshore of the Vesterålen Islands of northern Norway, the Hola trough is a cross-shelf feature with a width of 12 km and water depth around 200 m (Sauer et al. 2016). Active methane seepage was documented by recent geophysical (Chand et al. 2008) and geochemical investigations (Sauer et al. 2015(Sauer et al. , 2016. Ullsfjorden, a 70-km-long fjord in northern Norway, has a maximum water depth of ca. 285 m and numerous pockmarks that may be related to gas escape (Plassen and Vorren 2003;Sauer et al. 2016). The sediments in Ullsfjorden are composed of mostly glaciomarine trough fill (Plassen and Vorren 2003).

Sampling and analyses
The 13 sediment cores examined here were collected during three cruises from 2013 to 2016. Sediment cores were recovered by various techniques: box corer (BC), multicorer (MC), and gravity corer (GC) as well as push corer (PC) and blade corer (BLC) using a remotely operated vehicle (ROV) onboard R/V Helmer Hanssen. The sediment cores from Storfjordrenna and Bjørnøyrenna were collected with the assistance of either a towed camera or a ROV (see Supplementary material for more information). Porewater was sampled by either 10-cm or 5-cm HCl-washed (ca. 10%) rhizon samplers (Seeberg-Elverfeldt et al. 2005) with a pore size of 0.15 μm. Immediately after core recovery, MilliQ-rinsed rhizons were inserted through pre-drilled holes in the liners at 1-to 3-cm intervals, depending on the core length and expected redox zones. Titration of total alkalinity (TA) with HCl (0.012 M) and analyses of dissolved Fe 2+ spectrophotometrically with a ferrospectral complex as the color reagent (Traister and Schilt 1976) were performed onboard shortly after the pore fluids were collected. Samples for total sulfide (ΣHS), δ 13 C of dissolved inorganic carbon (DIC), cations, and anions were preserved onboard and analyzed on shore. For quantification of sulfide species in the sediments (i.e., acid-volatile sulfide (AVS) and chromium-reducible sulfide (CRS)) and their sulfur isotopic composition, chemical extraction of the sediments from cores S-904MC and S-1521GC was performed (Canfield et al. 1986;Fossing and Jørgensen 1989). Descriptions of the analytical protocols are provided in the Supplementary material. All pore fluid and solid-phase data are reported in Tab. S1 and Tab. S2, respectively, of the Supplementary material.

Flux and Gibbs free energy calculations
The fluxes of Fe 2+ towards the oxic sediment layer (F ox hereafter) were determined by applying Fick's law: where φ is the porosity (0.7; Sen et al. 2018a), D is the diffusion coefficient at the bottom water temperatures of the coring locations, and dC dX is the measured concentration gradient of Fe 2+ . For the diffusion coefficient of Fe 2+ , we considered both the molecular diffusion and bioturbation following where D′ Fe is the D in Eq.
(1) and D Fe is the tortuosity-corrected molecular diffusion coefficient (which ranges from 1.98E−6 to 2.01E−6 cm 2 /s) at different bottom water temperatures, and D b is the coefficient for bioturbation assuming the process resembles diffusion (Boudreau 1997). We estimated D b with the empirical relationship by Middelburg et al. (1997).
To estimate the Fe 2+ flux towards the sediment-water interface (i.e., F eff ), we used the equation proposed by Boudreau and Scott (1978): where C Fe.L is the pore water concentration of Fe 2+ at the bottom of the oxic layer in the sediments (mole/cm 3 ), and kl is the first-order rate constant of Fe 2+ oxidation (1/s), and L is the thickness of the oxygenated layer in centimeters. Fig. 1 Locations of the four Arctic cold seeps and the corresponding sediment cores investigated. Cores with available sedimentary sulfur data and S/Al ratios from XRF core scanning were labeled in yellow. Site labels of cores with apparent changes in downcore sulfate concentrations were underlined Determining the values of L and C Fe.L is challenging as we have no measurement for the dissolved O 2 concentration in the pore fluid. We chose a value of 0.1 cm for L based on the observed rapid reduction in nitrate concentration within the uppermost centimeter below the seafloor ( Fig. 2) which indicates a likely thin oxic sediment layer. A similar oxygen penetration depth has been detected from in situ microprofiler measurements (Boetius and Wenzhofer 2013) at seep environments. Dale et al. (2015) assigned depths of oxygen penetration to be less than 0.1 cm for severely hypoxic to anoxic scenarios. We assume these environments have similar redox conditions as our study areas based on the high methane- fueled sulfate reduction rate. For the values of C Fe.L , we estimated the Fe 2+ concentration at 0.1 cmbsf by interpolating the measured concentrations from 0 and 1 cmbsf. The resulting concentrations varied from 0.03 to 11.25 μM. The highest C Fe.L appears at S-1063MC and S-1064MC . The kl value was calculated following Millero et al. (1987): where k is the rate constant describing the overall hydrolysis equilibria of reduced iron species (Millero et al. 1987), I is the ionic strength for which we used the value 0.686 as calculated with the bottom seawater concentrations, and T is the temperature for bottom water at each site. We adopted the O 2 concentration of 320 μM based on the values reported by Anderson et al. (1988). By assuming a pH of 8.1 (Ofstad et al. 2020) and pK w values at the corresponding bottom water temperature at our sites (Millero 2001), we can derive [OH − ] in the equation. The calculated fluxes and the parameters required for the calculation are reported in Tab. S3 in the Supplementary material. Similar calculations were done previously for the sediments from Svalbard fjords (Wehrmann et al. 2014). The fluxes obtained from Svalbard fjords are compared with the fluxes derived from our studied sites. We also calculated molar Gibbs free energy (ΔG 0 r ) with measured pore fluid compositions to determine whether a set of reactions is thermodynamically plausible. Detailed descriptions for these calculations can be found in the Supplementary material. The downcore ΔG r changes are shown in Fig. S2 and reported in Tab. S6.
Compared with the cores with nearly constant downcore sulfate concentration, more apparent decreases in sulfate concentrations can be observed from cores S-904MC, S-1029PC, B-1123BLC, H-21PC, and U-26PC. The reduction of sulfate coupled to anaerobic oxidation of methane is mostly responsible for the decreasing sulfate in these cores as suggested by the elevated TA and low δ 13 C values of dissolved inorganic carbon ( Fig. S1 from Supplementary material). Distinct increases in Fe 2+ with the highest concentrations ranging from 40 to 100 μM were observed in the top 5-10 cm of these cores. Within the depth ranges where Fe 2+ is detected, ΣHS is usually below the detection limit (~20 μM) or barely measurable. Greater amounts of ΣHS are detected only below the depths where Fe 2+ can no longer be detected. Gravity core S-1521GC has a similar sulfate concentration gradient as S-904MC. The low sampling resolution for the top 40 cm of sediments however prohibits us from documenting the pore fluid profiles in detail. Cores S-1029PC and B-1123BLC show sudden declines in sulfate concentration within intervals of a few centimeters. Less than 10 μM of Fe 2+ can be detected in the top centimeters of these two cores with ΣHS concentrations as high as 12 mM.
We also report the concentrations of pore fluid nitrate from the Storfjordrenna and Bjørnøyrenna cores (Fig. 2). In general, up to 20 μM of nitrate can be detected from the depths where abundant Fe 2+ is detected. The measured high nitrate concentration cannot be explained by contamination of nitric acid during the cleaning of rhizons and sample vials. We used only hydrochloric acid during our cleaning procedure. Also, the sample vials for nutrient samples were not acid washed before use. Complete consumption of nitrate is only observed within the deeper sediments of cores S-904MC, S-1029PC, and B-1123BLC where greater than 4 mM of ΣHS is detected (Fig. 2).

Iron flux calculation
We estimated F ox ranging from 0.8 to 138.7 μmol/m 2 /day and F eff varying from 0.3 to 102.2 μmol/m 2 /day ( Fig. 3 and Tab. S3). In general, high F eff corresponds to high F ox (Fig. 3). The fractions of Fe 2+ escaping from the sediment columns, as defined by the ratios between F eff and F ox (R in Fig. 3), are higher at our sites compared with those from Svalbard fjords (Fig. 3), despite the generally lower concentrations of Fe 2+ detected in our cores.

Sediment sulfide speciation
AVS includes different nano-particles of iron monosulfides, such as mackinawite and greigite, and even a small fraction of pyrite (Rickard and Luther 2007). CRS, on the other hand, has been shown to be composed of mostly pyrite (Canfield et al. 1986). Our data show variable amounts of CRS and AVS in the sediments from cores S-904MC and S-1521GC (Fig. 4). CRS abundance increases abruptly from ca. 175 μmol S/g (μmole sulfur in gram of dry sediment) in the uppermost sediments to more than 1500 μmol S/g at greater depths in both cores. The increase of CRS in core S-904MC corresponds to the depth where high Fe 2+ concentrations were detected (Fig.  4). Abundant AVS (up to 154 μmol S/g) was detected in the first 30 cm of core S-1521GC. AVS is barely detectable at sediment depths approximately below 30 cmbsf at S-1521GC.
The δ

Discussion
We focus on the following reactions/processes that contribute to the cycling of iron in the sediments of the investigated cold seeps: (1) the release of Fe 2+ from DIR, (2) the release of Fe 2+ from other reactions such as the reduction of dissolved Fe 3+ , (3) precipitation of iron sulfide minerals, and (4) oxidation of Fe 2+ in the sediments as a result of biological disturbance. We show that Fe 3+ reduction and iron sulfide precipitation are likely the most important reactions governing the fate of Fe 2+ at the investigated sites. There are other possible reactions, such as oxidation of Fe 2+ by nitrate and oxygen as well as the precipitation of vivianite and incorporation of Fe 2+ into carbonates that can also contribute to the consumption of Fe 2+ in pore fluid (Egger et al. 2016). The pore fluid profiles however suggest only minor contributions of these reactions. We will thus not discuss these reactions in detail.

Production of Fe 2+ in the Arctic cold seep sediments
Glacial erosion of iron-rich terrestrial rocks around Svalbard supplies iron (oxyhydr)oxide to marine sediments. Release of Fe 2+ to the pore fluid through DIR explains the high Fe 2+ concentrations observed in Svalbard fjord sediments (Wehrmann et al. 2014). The similar maximum concentrations of ammonium and phosphate (varying by factors of two and four, respectively) detected from the 11 cores at our three investigated areas (except for U-26PC and S-1521GC; Fig. S1 from the Supplementary material) suggest comparable rates of POC turnover and DIR, which, however, cannot explain the variable iron fluxes that could be two orders of magnitude different (Fig. 3). Different sets of reactions are required to explain the calculated wide range of F ox .
The observation of high Fe 2+ concentrations within the intervals where nitrate is still detectable from Storfjordrenna and Bjørnøyrenna (Fig. 2) points to a Fe 2+ producing reaction that is thermodynamically more favorable than nitrate reduction. Following the classic early diagenesis sequence (Froelich et al. 1979), high Fe 2+ concentrations can only be detected at depths where nitrate is completely reduced if Fe 2+ is primarily supplied by DIR (c.f., Canfield et al. 1993). As a result, we suspect that a significant fraction of the Fe 2+ observed in our pore fluids is produced by the reduction of dissolved and complexed Fe 3+ (e.g., Fe(OH) 4 − and Fe(OH) 2 + ) in the oxic and suboxic sediments. Reduction of complexed Fe 3+ is energetically similar to oxic POC degradation and nitrate Fig. 3 Fe 2+ fluxes towards the oxic sediment layer (F ox ) and efflux (F eff ). The ratios between F eff and F ox (R) represent the fraction of Fe 2+ escaping from the sediment columns. In general, we observed both higher F eff and R from the investigated areas as compared with Svalbard fjords (Wehrmann et al. 2014), despite the lower Fe 2+ concentrations detected in our pore fluid reduction based on our Gibbs free energy calculation of various redox pairs (Fig. S2). A similar conclusion that chelated Fe 3+ may increase the redox potential of iron reduction was drawn by Lovley and Phillips (1988) and Thamdrup (2000). The concentration of Fe 3+ is generally low in the marine environment due to the low solubility of Fe(III) oxide (e.g., 10 −8 M with ferrihydrite; Stumm and Morgan (2012)). However, it has been shown that, in the presence of chelators, Fe 3+ can be solubilized from the surface of oxide minerals (Lovley 1997). Furthermore, Fe 3+ can bind to organic ligands and maintain solubility in seawater (Luther III et al. 1992;Bau et al. 2013). Our calculations suggest that such a process is thermodynamically more favorable if Fe 3+ is at the nanomolar level, a reasonable value considering the concentration of organic ligands in seawater (Millero 1998). Future work determining the concentrations of chelators (e.g., organic ligands) is required to test our hypothesis for the co-occurring high Fe 2+ and nitrate concentrations in the shallow sediments.

Consumption of Fe 2+ through authigenic iron sulfide formation
Much of the pore fluid Fe 2+ is consumed through the precipitation of iron sulfide minerals, a process that is inferred from the abundance of solid-phase sulfur in S-904MC and S-1521GC (Fig. 4). We detected an order of magnitude more abundant AVS and CRS as compared with previous studies (Lim et al. 2011;Wehrmann et al. 2014). At S-1521GC, AVS Fig. 4 Concentrations of sulfate, Fe 2+ , and ΣHS and the amounts of CRS and AVS as well as their corresponding δ 34 S values from the two selected cores from Storfjordrenna is most abundant at 3 cmbsf (0.9 wt% AVS-Fe or 154 μmole AVS-S/g sediments) which is 3 to 10 times higher than the content reported by Wehrmann et al. (2014). The highest CRS-Fe abundance from Wehrmann et al. (2014) is ca. 0.25 wt% whereas, from both S-904MC and S-1521GC, we measured more than 5 wt% of CRS-Fe. When comparing the abundance of CRS with porewater profiles from S-904MC, we noted that the rapid increase of CRS abundance at the ca. 10 cm corresponds to the interface between porewater Fe 2+ and ΣHS (Fig. 4). This correlation may be explained by the rapid formation of pyrite in shallow sediment depths (< 10 cmbsf). We did not observe any significant changes in CRS abundance or in the δ 34 S-CRS across the modern sulfate-methane-transition (SMT) from S-1521GC, though the variable CRS abundance throughout the core makes it difficult to detect potentially small increases in CRS at the SMT. The parallel trend between δ 34 S-AVS and ΣHS concentration suggests that AVS uses sulfide and is in equilibrium with the modern pore fluid. On the other hand, the lack of correlation between δ 34 S-CRS and ΣHS concentration implies that CRS likely integrates the long-term pore fluid sulfide signals produced at different times and rates of sulfate reduction.

Fe 2+ fluxes in the surficial sediments of Arctic cold seeps
Comparison of our flux estimates with those from Svalbard fjords (Wehrmann et al. 2014) and other regions (Dale et al. 2015 and the references therein) shows lower F ox but comparable F eff from the investigated cold seeps. Fe 2+ can be oxidized in the oxic and suboxic surface sediments by dissolved oxygen and nitrate, respectively, and form nanoparticulate ferrihydrite aggregations that are later scavenged by suspended sediments (Raiswell and Anderson 2005). This re-oxidation process can be quantified by the ratios between F ox and F eff ( Fig. 3 and Tab. S3). With the exception of anomalous values from cores S-932MC and S-1064MC, on average, 40% of the Fe 2+ that diffuses towards the surface sediments is oxidized. In other words,~60% of the Fe 2+ produced through iron reduction in the sediment columns can escape the oxidation process in surficial sediments and enter the bottom water (Fig. 3). Such a high proportion may be due to a thin oxic surface layer of sediments facilitating effective escape, though future measurements of dissolved oxygen concentration in the pore fluid are required to verify this. We are aware that our choice of oxygen penetration depth (L in Eq. (3)) has a profound effect on F eff . Nonetheless, even if we assigned double the value for L (0.2 cm), which reduces F eff by 53%, our conclusion that more Fe 2+ is able to escape from the sediments as compared with those from the Svalbard fjord sites still holds true. Homogenization of surface sediment by benthic fauna, which has been accounted for by the D b coefficients in our flux calculation (Tab. S3), may enhance oxidation and lower the fraction of Fe 2+ leaving the sediment column. However, our calculation suggests that such bioturbation can only account for 2-3% of the Fe 2+ oxidation in surface sediments.

Interaction between benthic macrofauna and subsurface geochemistry
The survival of chemosynthesis-based animals at cold seeps is highly dependent on the cycling and flux of sulfur species (Freytag et al. 2001;Dubilier et al. 2008;Sen et al. 2018a), which in turn also affects the cycling of Fe 2+ and vice versa. From the area where core 938MC was recovered, the seafloor is devoid of any chemosynthesis-based megafauna but colonized by non-seep specialist fauna such as Thenea sponges and anemones (Fig. 5a). There is no detectable ΣHS in the top 10 cm of the sediments as most of the ΣHS has been removed by iron sulfide precipitates. Even though no CRS/ AVS abundance data is available at this site, the high S/Al ratio within this sediment interval (Fig. S3 in the Supplementary material) supports this conclusion. The high Fe 2+ concentration therefore serves as a "geochemical cap" preventing ΣHS from leaking to the bottom ocean, consistent with the low sulfide flux. This in turn allows for colonization by non-seep specialist fauna that does not have adaptations for dealing with sulfide toxicity. In the area where S-904MC was recovered, a dense distribution of Oligobrachia worms containing symbiotic sulfur-oxidizing bacteria in their trunks (Sen et al. 2018b) can be observed (Fig. 5b). These worms likely rely on dissolved sulfide as their primary energy (and ultimately nutrition) source (Sen et al. 2018b) and are able to harvest sulfide deep in the sediments across their half-meterlong tubes. Indeed, ΣHS concentration is below the detection limit (~20 μM) in the uppermost centimeter from these sediments with Fe 2+ concentrations slightly over 100 μM (Fig.  5b). Rapid iron reduction and high Fe 2+ concentration prevent most of the sulfide from leaking to the bottom ocean at this location as well. Core B-1123BLC represents an extreme case with high ΣHS supply and greater than 12 mM ΣHS detected at 15 cmbsf (Fig. 5c). Not only were dense distributions of Oligobrachia worms observed in seafloor images of where this core was recovered from, but additionally, filamentous bacteria (likely sulfur-oxidizing bacteria Lösekann et al. 2008) colonizing the anterior ends of Oligobrachia tubes that extend a few centimeters into the water column were also observed. This colonization results in a fuzzy appearance of the worm tubes as visible in Fig. 5c. The high concentration of ΣHS close to the sediment surface and potential leakage to the bottom water accounts for these worm tubes being covered in bacteria, that can use this leaking sulfide as an energy source. These bacteria require oxygen in addition to sulfide for carbon fixation and therefore need to locate themselves at the redox boundary, which is often restricted to the sediment-water interface. Leakage of sulfide into the bottom water extends the niche of these bacteria, and the worm tubes provide a settlement surface from which to exploit it.
It is also noteworthy that, immediately adjacent to the bacteria-covered worms, microbial mats were present but not the Oligobrachia worms themselves (the area where imprints of the blade corer are visible in Fig. 5c). We speculate that the sulfide flux in this field of bacterial mats is even higher and that appreciable quantities of hydrogen sulfide may have leaked to the bottom water. Even though Oligobrachia worms are equipped to take up oxygen in hypoxic conditions and further have adaptations against sulfide poisoning, the amount of sulfide reaching the bottom water in this location might be beyond their limits. This could explain their absence from this seafloor patch. Such an explanation is, however, complicated by the fact that the bacteria making up the mat are also aerobic and require access to oxygen. Future studies targeting the two habitats are required to compare the concentrations of hydrogen sulfide in the bottom water.
We also suspect a complicated interaction between nitrogen, sulfur, and iron species occuring in the sediments showing fast sulfide/Fe 2+ turnover (e.g., the sediments from Fig. 5 b  and c). Apparent downcore decrease to almost exhaustion in nitrate concentration is only observed from cores S-904MC, S-1029PC, and B-1123BLC with ΣHS concentrations higher than 4 mM. We propose that the high ΣHS concentrations promote the reduction of nitrate in the deeper sediments (> 10 cmbsf) which is supported by our Gibbs free energy calculation that coupling of denitrification with sulfide oxidation can be equally energetic as the POC-induced nitrification (Fig. S2). It has been shown that the cable bacteria Desulfobulbaceae are able to remotely couple these two redox pairs in marine sediments (Marzocchi et al. 2014). The competition for sulfide between iron sulfide formation and nitrate reduction may also determine the fate of Fe 2+ in such sulfiderich sediments. Future work is needed to examine these competing pathways.

Conclusions
In this study, we present pore fluid and sediment geochemical data to constrain the fate of Fe 2+ in four cold seeps along the northern Norwegian margin. We show that DIR alone cannot explain the wide range of iron concentrations and fluxes observed at the investigated sites. The co-appearance of Fe 2+ and nitrate in pore fluid in cores from Storfjordrenna and Bjørnøyrenna as well as our thermodynamic calculations leads us to propose that aqueous Fe 3+ reduction plays a significant role in producing Fe 2+ . A significant fraction of the Fe 2+ is precipitated as iron sulfide minerals, as supported by the high abundance of AVS and CRS. We demonstrate that, Fig. 5 Comparison of seafloor observations (a-c) as well as pore fluid Fe 2+ and ΣHS profiles (d-f). The joint cycling of iron and sulfur determines the distribution of chemosynthesis-based animals despite the lower Fe 2+ fluxes towards the oxic sediment layer into the bottom water as compared with those estimated in Svalbard fjord sediments, a larger proportion of Fe 2+ is able to escape the sediments from our study areas. This can be attributed to a potentially thin oxic layer in sediments of our sites. The comparison between seafloor macrofauna assemblage and subsurface geochemical profiles reveals the interaction and geochemical control on the distribution of chemosynthesis-based macrofauna. MISO (Multidisciplinary Instrumentation in Support of Oceanography) (Dr. Daniel Fornari and Gregory Kurras) operating the TowCam deep-sea imaging system and the ROV team (Prof. Dr. Martin Ludvigsen, Frode Volden, Stein Nornes, and Pedro de la Torre) from the Norwegian Centre for Autonomous Marine Operations and Systems (AMOS) for their assistance in sample collection. We would also like to thank Mrs. Haoyi Yao for her assistance in porewater sampling and analyses. The constructive and insightful comments from Dr. Susann Henkel and two anonymous reviewers greatly improve the quality of this manuscript. The handling of the paper by the Editor-in-Chief of Geo-Marine Letters Dr. Gabriele Uenzelmann-Neben is also much appreciated.
Funding information Open Access funding provided by UiT The Arctic University of Norway. This work was supported by the Research Council of Norway through its Centres of Excellence funding scheme (project number 223259) and NORCRUST (project number 255150).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.