Enhanced methane cycling across the Laptev Sea signaled by time-integrated biomarkers of aerobic methane oxidation

Elevated methane concentrations in seawater have been reported over extensive areas of the East Siberian Arctic Seas, overlying thawing subsea permafrost. However, observed methane concentrations of the ephemeral seawater are highly variable across both space and time, compromised by both the timing of rare measurements and storm-driven exchanges to the atmosphere. Here, we applied time-integrated signals of the δ¹³C-composition of specific C₃₀ hopanoids (diploptene, hop-17(21)-ene, neohop-13(18)-ene and diplopterol) in surface sediments to trace aerobic methane oxidation as a proxy for enhanced methane cycling. Interpretations of hopanoids and possible sources were further assessed by 16S-rRNA analyses in the surface sediments. The consistently low δ¹³C-C₃₀ hopenes signals, ranging between −57.5 ‰ to −37.1 ‰ (n=23) across the Laptev Sea shelf indicated aerobic methane oxidation. This suggests ubiquitous methane cycling with the most pronounced intensities in the outer shelf region, broadly consistent with the observed methane concentrations. Notably, depleted δ¹³C-C₃₀ hopenes were also found in the mid-shelf region of the Laptev Sea, earlier thought to be an area of comparatively low methane cycling. High methane concentrations were also observed in the vicinity of the Lena River delta, yet the isotopically heavier δ¹³C-C₃₀ hopenes may here reflect a combination of lower aerobic methane oxidation, a greater relative abundance of type II methanotrophs (lower isotope fractionation during hopanoid production) and isotope dilution from non-methanotrophic sources. While this complicates the biomarker interpretation in the unique setting near the Lena River delta, the δ¹³C-C₃₀ hopenes were still much lower than δ¹³C-organic carbon, indicating aerobic methane oxidation and a clear methane cycling signal also in this regime. Taken together, the results unravel the wider cross-shelf patterns of enhanced methane cycling in the Laptev Sea through probing of methane fossilised in membrane lipids of aerobic methanotrophs, with the molecular-isotopic pattern being preserved in the sedimentary archive.

1 Introduction

Large uncertainties remain regarding methane (CH₄) emissions from both anthropogenic and natural systems (Saunois et al., 2025; Schaeffer et al., 2025). Terrestrial permafrost stores massive amounts of carbon, but climate change and Arctic amplification have enhanced its thawing, causing net CH₄ emissions to the atmosphere (Ramage et al., 2024; Hugelius et al., 2024). Subsea permafrost, formed from the inundation of the northern tundra during the last deglaciation, is often overlooked. Partly due to its inaccessibility, limiting the number of observations. However, observations display highly elevated CH₄ concentrations in the shallow seawater column overlying subsea permafrost (e.g., Semiletov, 1999; Shakhova et al., 2010; Steinbach et al., 2021), which is currently thawing about ten times faster than nearby terrestrial permafrost in Northeastern Siberia (Shakhova et al., 2017; Wild et al., 2022).

The East Siberian Arctic Shelf (ESAS), the world's largest continental shelf sea, is very shallow (on average, ca. 50 m) and estimated to host the majority of the world's subsea permafrost (Romanovskii and Hubberten, 2001; Romanovskii et al., 2005; Overduin et al., 2019). The concentration of CH₄ in the surface waters of the ESAS is, particularly in the Laptev Sea, much higher than what would be expected from CH₄ levels in the atmospheric boundary layer, suggesting that the thawing subsea permafrost system may serve as a conduit for CH₄ release to the atmosphere (Shakhova et al., 2010, 2014, 2019). A key challenge to constrain the geospatial pattern of CH₄ releases is the short-term variability of water column CH₄ concentrations, together with the sparseness of observations over time in this remote region. While subsea permafrost thaw and related CH₄ emissions are governed by environmental changes on decadal time-scales (Shakhova et al., 2017, 2019), the sparse and infrequent measurements of water column CH₄ concentrations might capture neither temporal nor geospatial trends of CH₄ emissions from these systems. Water sampling has revealed the general pattern of elevated CH₄ concentrations and discovered a few major hotspots, but this pattern is strongly affected by short-term variability governed by storms and the shallow water column of the Laptev Sea (e.g., Shakhova et al., 2014). Furthermore, a recent modeling study concluded that current atmospheric observation networks are not able to detect abrupt CH₄ releases from this region (Wittig et al., 2024). To better grasp the CH₄ dynamics over a longer timespan than what is captured in snapshot seawater samples, there is a need for an integrated metric to observe patterns of long-term CH₄ release from the subsea permafrost system over both space and time scales.

Specific biomarkers in surface sediments offer an opportunity to study large-scale variabilities in CH₄ and its dynamics integrated over time scales of years to decades. Bacteria in oxygenated sediments and in the water column can utilize CH₄ as an energy and carbon source through aerobic CH₄ oxidation (AeOM), converting it to carbon dioxide (CO₂; Hanson and Hanson, 1996), or assimilating it into biomass (e.g., lipid synthesis). Hopanoid lipids such as diploptene and diplopterol are important constituents of the cell membranes of Methane Oxidizing Bacteria (MOB) carrying out these processes (Summons et al., 1994), but can also derive from bacteriohopanepolyols (BHPs) upon diagenesis (Mackenzie et al., 1981). Hopanoids and BHPs are not only produced by MOB, but can stem from a variety of aerobic and anaerobic bacteria utilizing CH₄ and/or inorganic/organic carbon (Rohmer et al., 1980; Summons et al., 1999; Sinninghe Damsté et al., 2004; Belin et al., 2018). Thus, hopanoid analyses are frequently combined with compound-specific isotope analysis of stable carbon isotopes (δ¹³C) to relate these biomarkers to CH₄ cycling. The isotopically depleted stable carbon isotopes of CH₄, δ¹³C-CH₄ (as low as −90 ‰; Milkov and Etiope, 2018), is used to differentiate CH₄-derived hopanoids from other sources. Thereby, C₃₀-hopanoids (hopanoids with 30 carbon atoms) have been used as a proxy to infer enhanced CH₄ cycling in past climates (e.g., Hinrichs, 2001; Hinrichs et al., 2003; Birgel and Peckmann, 2008; Sun et al., 2022; Blumenberg et al., 2024; Yan et al., 2025) and contemporary systems (Davies et al., 2016; Inglis et al., 2019). Generally, enhanced CH₄ cycling has been linked to decreasing δ¹³C values of hopanoids (e.g., Inglis et al., 2019; van Winden et al., 2020; Yan et al., 2025). Therefore, δ¹³C-hopanoids may be used as a time-integrated signal of enhanced CH₄ cycling. In particular, hopanoid carbon chain lengths ≤ C₃₀ are often more depleted in ¹³C compared to C₃₁-hopanoids and therefore closely associated with MOB production of hopanoids (Inglis et al., 2019). While the proxy has been widely used to study CH₄ dynamics in past climates, contemporary system calibrations of ≤ C₃₀ hopanoids only exist in peatlands (≤ C₃₀ hopanoids between −21 ‰ and −45 ‰; Inglis et al., 2019) and lacustrine systems (diploptene ranging from −38.8 ‰ to −68.8 ‰; Davies et al., 2016). The lack of a large-scale comparison of CH₄ and hopanoids in marine systems leaves uncertainties when interpreting δ¹³C-hopanoids as tracers of CH₄ cycling in geological records. In summary, δ¹³C-hopanoids in oxygenated surface sediments can be an informative tool to constrain a time-integrated signal of enhanced CH₄ cycling on a years-decade scale, complementing observations of CH₄ concentrations in seawater which is highly variable over much shorter time-scales.

Figure 1 The Laptev Sea sampling locations and station numbers, including estimates of the current Arctic land and subsea permafrost covers. Permafrost datasets were generated from Obu et al. (2019) and Overduin et al. (2019). The bathymetry was plotted using IBCAO version 5.0 (Jakobsson et al., 2024).

Observations across the Laptev Sea suggest the presence of AeOM, but its efficiency is questionable and spatially variable. The Laptev Sea water and surface sediments are well oxygenated (Stepanova et al., 2017; Brüchert et al., 2018; Maciute et al., 2025), with the water column oxygen saturation ranging from 70 %–100 % during expeditions between 2015 and 2020 (Xie et al., 2023). This combined with high CH₄ concentrations, indicates the general possibility for AeOM. Furthermore, microbial studies have confirmed the presence of MOB in Laptev Sea sediments and water column (Samylina et al., 2021; Savvichev et al., 2023). Recent machine-learning estimations deemed the global AeOM to be generally efficient in coastal waters (Mao et al., 2022), but AeOM observations in the vicinity of the Lena River delta display a rather weak methanotrophic sink, with a CH₄ turnover time of 193 d in mixed sea/river water (Bussmann et al., 2017). In contrast, incubation experiments in the outer Laptev Sea cold seep region suggest more efficient AeOM in those surface sediments (Tikhonova et al., 2021), but with very low AeOM rates in the water column (300–1000 d turnover time, Shakhova et al., 2015). Taken together, AeOM is active in the Laptev Sea, but observations are diverging on the scale and efficiency of AeOM across the Laptev Sea. Such discrepancies highlight the importance of constraining the geospatial distribution of AeOM to unravel areas of more intense CH₄ cycling.

In this study, we explore AeOM biomarkers as time-integrated proxies of CH₄ cycling in Arctic Ocean water and sediments, focusing on the Laptev Sea. We hypothesize that known CH₄ ebullition hotspots in the Outer Laptev Sea and Inner Laptev Sea (OLS and ILS), where > 1000 nM dissolved CH₄ concentrations have been observed (Shakhova et al., 2010, 2014; Steinbach et al., 2021), have high concentrations of AeOM-tracing C₃₀ hopanoids (diploptene, hop-17(21)-ene, neohop-13(18)-ene and diplopterol) with low δ¹³C values. In contrast, we hypothesize that the mid-shelf region without any discovered CH₄ hotspots, yet with dissolved CH₄ concentrations in the range 10–60 nM (Fig. 2) display lower concentrations of C₃₀ hopanoids more enriched in ¹³C. Observations of AeOM biomarkers are also compared to water CH₄ concentrations and 16S-rRNA data on abundance of MOB- and non-methanotrophic hopanoid-producing bacteria. We finally use this dataset to explore the geospatial distribution of CH₄ cycling across the 600 km wide Laptev Sea shelf. Insights from this study also allows to establish some general considerations for interpreting hopanoids in geological records.

2 Methods

2.1 Study area and the ISSS-2020 expedition

During the last deglaciation, sea level rise caused the northern regions of East Siberian permafrost to be submerged, thus forming the present coverage of Laptev Sea subsea permafrost (Romanovskii and Hubberten, 2001; Romanovskii et al., 2005; Nicolsky and Shakhova, 2010; Shakhova et al., 2017; Lindgren et al., 2018; Overduin et al., 2019). The Laptev Sea water column is shallow (average water depth ∼ 50 m), with well oxygenated top mm to cm of surface sediments (Brüchert et al., 2018; Maciute et al., 2025). Ebullition appears to be a key transport vector in this shallow region, transferring large amounts of CH₄ to the water column (Shakhova et al., 2010, 2014; Steinbach et al., 2021). The Inner (ILS; latitudes < 73.9° N and longitudes ∼ 130° E) and Outer Laptev Sea hotspot regions (OLS; latitudes ≥ 76.1° N and longitudes 125-131° E) are established as areas of extensive CH₄ cycling in the water column, with their shallowness causing frequent ventilation of the water column CH₄ to the overlying atmosphere (e.g., Shakhova et al., 2010, 2014; Thornton et al., 2016, 2020; Steinbach et al., 2021; Chernykh et al., 2023). Results from campaigns of subsea permafrost drilling have indicated rapid thawing of subsea permafrost in the ILS region (e.g., Shakhova et al., 2017). Thawing subsea permafrost is considered to facilitate CH₄ release from the very shallow ILS (Shakhova et al., 2014; Sapart et al., 2017; Wild et al., 2022). In the deeper, yet still shallow OLS (46–72 m), a thermogenic gas reservoir has been constrained as the main CH₄ source using triple isotope fingerprinting (Steinbach et al., 2021). The Laptev Sea is also strongly influenced by the Lena River plume (Shakhova and Semiletov, 2007; Shakhova et al., 2010). While the ILS receives substantial amounts of organic carbon from the Lena River (Vonk et al., 2012; Holmes et al., 2012; McClelland et al., 2016; Bröder et al., 2018; Martens et al., 2022), the riverine export of CH₄ is negligible in comparison to the regional CH₄ emissions (Shakhova and Semiletov, 2007; Shakhova et al., 2010). The mid-outer Laptev Sea transect (MLS; latitudes between 73.9–76.1° N and longitudes ∼ 130° E) is, in contrast, an area of lower CH₄ concentrations than the ILS and OLS, yet the still elevated CH₄ levels over extensive MLS scales (Shakhova et al., 2010) call for investigations on potential CH₄ sources also for this region. Samples in the current study were collected from the Laptev Sea during the International Siberian Shelf Study in 2020 (ISSS-2020; September–November 2020) aboard the research vessel Akademik Keldysh (Fig. 1).

2.2 Sediment and Water Sampling

2.2.1 Sediment sampling

Sediment samples were retrieved using an Oktopus Multicorer System that sampled eight cores per station with an undisturbed sediment-water interface. The penetration depth of the liners were maximum of 30 cm. Overlying water was drained off the surface sediment in the cores and the sediment was subsequently sectioned in 1 cm intervals onboard using an Oktopus Core Extruder. All sediment subsamples for organic geochemical analyses were immediately frozen in plastic seal bags and the 1–2 cm slice in each core was used to extract lipid biomarkers for this study.

Sediment cores with pre-drilled holes on the liners were collected for 16S-rRNA analysis at n= 36 stations (Maciute et al., 2025), where n=11 samples from the Laptev Sea shelf were used in this study. The overlying water was siphoned from the top of the sediment core and 3 ml of sediment was subsampled from three sampling depths in the core (0–1.5, 1.5–3, and 3–4.5 cm). Subsamples were vortexed with 10 mL RNA later (Sigma Aldrich) and stored for 2 h at 4 °C before being frozen at −80 °C. In this study, we only report the top 0–1.5 cm subsection for microbial analyses, as the focus was on aerobic CH₄ oxidizers.

2.2.2 Measurements of dissolved methane concentration in the water column

Water samples were taken from a Rosette sampler with 24×10 L bottles, closed at selected depths. Subsamples were immediately taken through silicon tubes into 60 mL Luer-lock plastic syringes for onboard CH₄ concentration measurements. Briefly, a Restek reservoir bag was used for helium headspace equilibration with an IKA KS 260 benchtop shaker. Before concentration measurements using Gas Chromatography-Flame Ionization Detection (GC-FID), the gases were transferred through stainless-steel Millipore filter holders packed with granular Drierite and quartz wool, sealed using PTFE tape. The CH₄ concentration measurements were made with 5 mL sample injections through one $\mathrm{1}/{\mathrm{8}}^{\prime \prime }$ Porapak Q packed pre-column and a 3 m $\mathrm{1}/{\mathrm{8}}^{\prime \prime }$ Heyesep D column before FID. In this study, the surface water is described as the topmost CTD/Rosette-based water samples (mean = 3.5 m; n=21) and the sub-pycnocline (mean = 37 m; n= 71) as water samples below the largest density gradient (σ_θ) based on the CTD profile (Conductivity, Temperature and Depth). The sub-pycnocline concentrations are here reported as the median and interquartile ranges of all samples between the pycnocline and the bottom water (Fig. 2).

2.3 Measurements of methane-tracing lipid biomarkers and their isotope composition

Bulk organic carbon and stable carbon isotope (δ¹³C-OC) analyses of the 1–2 cm slice in each multicore were made according to standard operating procedures at Stockholm University (e.g., Martens et al., 2020). Briefly, ∼ 10 mg freeze-dried sediment was weighed in silver capsules and acidified with 100 µL 1 M HCl twice for the removal of inorganic carbon. After the first acidification, the sample was stored at 50 °C for 3–4 h, followed by acidification again and reaction at 50 °C overnight. The acidified samples were wrapped in tin capsules and analysed with an elemental analyser coupled to an isotope ratio mass spectrometer (EA-irMS, Delta V Plus, Thermo Scientific, Germany).

2.3.1 Lipid extraction

Around 5 g of freeze-dried sediment was subsampled from one core per station. Before the extractions, internal standards were added to each sample (5α-androstane, tetracosane-d₅₀, triacontane-d₆₂, eicosanoic acid-d₃₉, 2-hexadecanol and stigmasterol-d₅) to calculate recoveries and correct for losses during the laboratory procedure. The lipid extraction and clean-up were a modification of standard operating procedures at Stockholm University (e.g., van Dongen et al., 2008; Vonk et al., 2012; Martens et al., 2020). Briefly, lipids were extracted using accelerated solvent extraction (ASE 350, Dionex), followed by elemental sulphur and water removal from the extracts by addition of HCl-activated Cu and anhydrous NaSO₄. The total lipid extracts were saponified for 2 h at 80 °C following the addition of ∼ 3 mL KOH (6 % in MeOH) to each of the dried samples. Base-catalysed hydrolysis was applied as it has been shown to increase the amount of hopenes through dehydration of diplopterol (Sessions et al., 2013). The saponified neutral lipids (containing hopanoids, alkanes, sterols and alkenones) were collected with liquid-liquid extraction five times with the addition of 1 mL Milli-Q water and ∼ 2 mL hexane: DCM (9:1, $v/v$). After collecting the neutral fraction, the saponified sample was acidified to pH 1 with 6 M HCl, ∼ 2 mL hexane: DCM (9:1, $v/v$) was added, the sample was mixed and the supernatant containing the fatty acid fraction was collected. The fatty acid liquid-liquid extraction was repeated five times in total. In this study, we focused on the neutral fraction, containing hopanes/hopenes/diplopterol.

Following the saponification and separation of the neutral and fatty acid fractions, preparative chromatography was applied to the neutral fraction, with furnace-baked alkaline Al₂O₃ (∼ 0.5 g) in a pre-combusted Pasteur pipette as the stationary phase. The neutral fraction was separated into non-polar hydrocarbons (hopanes/hopenes, F1), alkenones/diplopterol (F2) and polar hydrocarbons (F3) using 4 mL hexane: DCM (9:1, $v/v$), hexane: DCM (1:1) and DCM: MeOH (1:1, $v/v$), respectively. To enable clean chromatograms for δ¹³C measurements, urea adduction was applied to purify the non-polar extracts (Pancost et al., 2008; Inglis et al., 2019). The sample was separated into an adduct (containing n-alkanes) and a non-adduct fraction (containing cyclic and branched hydrocarbons, e.g., hopenes). Before instrument analysis, the diplopterol fraction was silylated for 1h at 80 °C using BSTFA: pyridine (1:1).

2.3.2 Hopanoid identification and quantification

Hopanoids were quantified using gas chromatography-mass spectrometry (GC-MS 7820-A, Agilent Technologies USA) with a DB5-MS ultra-inert column (30 m × 250 µ m × 0.25 µm). The GC-MS was run in splitless mode with a starting temperature of 50 °C for 2 min, then 10 °C min⁻¹ to 210 °C, where the temperature increased by 3 °C min⁻¹ to 262 °C for 10 min. Thereafter, the temperature increased by 0.5 °C min⁻¹ to 270 °C and lastly a 3 °C min⁻¹ ramp until 310 °C for 5 min. A post-run at 305 °C for 2 min was used to minimize sample carryover. Identification of hopanoids was done according to published mass spectra in Sessions et al. (2013), Sinninghe Damsté et al. (2014) and Elvert and Niemann (2008). Hopenes were quantified using the C₃₀ hopane external calibration standard (NIST 2266). However, during sample pre-treatment, dehydration of diplopterol and subsequent conversion to diploptene and hop-21-ene were observed within F2. This was likely catalysed by the basic Al₂O₃ column, as the problem resolved when switching to neutral Al₂O₃, which was used to separate the molecules before the final analysis. Diplopterol dehydration commonly occurs during sample workup, even during instrumental analysis (Kannenberg et al., 1995; Sessions et al., 2013). Thus, in this paper, we only present the δ¹³C value of diplopterol when possible and not its concentrations.

2.3.3 Compound-specific isotope analysis of hopanoids

Compound-specific isotope analysis (CSIA) of hopenes/diplopterol was carried out at Stockholm University using gas chromatography coupled to continuous flow isotope ratio mass spectrometry (GC-irMS; Trace-GC, GC-Isolink and Delta V Plus, Thermo Scientific, Germany) on a DB5-MS ultra-inert column (60 m × 0.25 mm × 0.25 µm). The oven temperature program was the same as for GC-MS analyses. Briefly, triplicate 3 µL injections were made for each sample and quantified using external n-alkane standards (n-alkane mixture A7, Dr. Arndt Schindelmann, Indiana University). Before each batch of sample measurements, the precision and linearity of n-alkane standards were monitored to ensure precision ≤ 0.5 ‰ (n= 6). The reproducibility of triplicate sample measurements was 0.1 ‰–2.3 ‰ (mean = 0.8 ‰). The higher standard deviations of some samples were attributed to lower peak areas, causing less precise measurements. To minimize the precision effect from lower peak areas, an acceptance threshold of ≥ 100 mV peak areas was set, at which n-alkane standard measurements yielded a precision of 0.9 ‰. The added trimethylsilyl (TMS; Si-(CH₃)₃) in δ¹³C measurements was corrected by measuring a silylated β-sitosterol standard using GC-irMS and the unsilylated β-sitosterol powder with EA-irMS. The difference in δ¹³C between the silylated and unsilylated standard was used as the δ¹³C of the added carbon to derivatized compounds, where f is the fraction of added carbon from the TMS group and δ¹³C_TMS is the isotopic value of the TMS (Eq. 1).

$$\begin{array}{}\text{(1)}& {\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{Diplopterol}}={\displaystyle \frac{({\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{Silylated}\mathrm{diplopterol}}-f\cdot {\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{TMS}})}{\left(\mathrm{1}-f\right)}}\end{array}$$

2.4 Microbiology

2.4.1 DNA extraction and sequencing

Sediment samples were thawed and prepared to facilitate 16S-rRNA analyses as described in Maciute et al. (2025). Extractions were made from 2.9±0.7 g wet sediment and amplification of the 16S rRNA gene V4 region was conducted with the primers 515 F (Parada et al., 2016) and 806 R (Apprill et al., 2015). Library preparation was conducted using the NEBNext^® Ultra™ II DNA Library Prep Kit with index adapters synthesized in-house by Novogene. The pooled library was sequenced on the Illumina NovaSeq 6000 SP platform with a 2×250 bp paired-end setup. The raw sequencing data have been uploaded to NCBI GenBank (https://www.ncbi.nlm.nih.gov/bioproject/, last access: 20 February 2026) and can be accessed at BioProject: PRJNA1264051.

2.4.2 Bioinformatics

A total of 2 341 395 million sequencing reads, averaging 65 039 reads per sample (36 samples in total) were analysed according to the DADA2 (version 1.28) pipeline (Callahan et al., 2016) with default settings except for the filter commands: truncLen=c(0,0), maxEE=2, truncQ=2, trimLeft=c(22,21); error model: MAX_CONSIST=30; merging of pair-ends: minOverlap=10; and chimera removal: allowOneOff=TRUE, minFoldParentOverAbundance=4. After trimming, filtering, merging pair-ends and chimera removal, a total of 89 148 amplicon sequence variants and 9.2 million sequences were retained, averaging 43 485 reads per sample (minimum 26 424 and maximum 66 969 reads per sample). The sequences in the created ASV table were then assigned to taxonomy using the SILVA database version 132 (Quast et al., 2013) and analysed further as relative abundances (i.e., [counts/∑counts] × 100) in RStudio software. Only the samples collected and sequenced from the Laptev Sea shelf are discussed in this study (n= 11).

2.4.3 Relative abundance of methane-oxidizing bacteria and non-methanotrophic hopanoid producers

Microbial sequencing data were explored for both aerobic methanotrophs as well as other hopanoid-producing bacteria. Aerobic methanotrophs were identified within Gammaproteobacteria (Type I) and Alphaproteobacteria (Type II). Bacteria within phyla that commonly encode shc (squalene-hopane cyclases, i.e., hopanoid producers), such as Cyanobacteria, were assumed to be hopanoid producers. Literature exploration of other organisms with documented production of hopanoids was made using LOTUS (Rutz et al., 2022) and was further used to assess “Other hopanoid producers”. Particularly, the abundance of certain Acidobacteria, Actinobacteria and non-methanotrophic Proteobacteria described in previous research of hopanoid producers (Sohlenkamp and Geiger, 2016; Sinninghe Damsté et al., 2017) was used to explore non-methanotrophic hopanoid producers. The information on all investigated hopanoid producers can be found in the supplementary data (data availability).

2.5 Statistical analyses

Statistical differences were evaluated using a significance level α= 0.05 in RStudio (Version 2024.12.1). One-way Welch's ANOVAs were applied to analyse differences in hopanoid concentrations and isotope signatures between the OLS, MLS and ILS due to the wide concentration ranges and the limited number of samples in the MLS. After Welch's ANOVAs, Games-Howell post-hoc tests were applied to verify where there were significant differences. Comparisons of CH₄ concentrations above and below the pycnocline were made using t-tests assuming unequal variances. All statistical results (F-values, degrees of freedom, n-values, p-values and adjusted p-values) are reported in the Supplement Tables S1–S5.

Sequencing data indicated three microbial types of hopanoid producers (Figs. S1 and S2 in the Supplement) and no evidence for anammox bacteria. Aerobic methanotrophs and heterotrophs were most common and therefore used as the two endmembers for hopanoid source apportionment. The hopanoid source apportionment was calculated through a Bayesian stable carbon isotope mixing model using two endmembers (MOB and heterotrophic bacteria) applied with “simmr” in RStudio (Parnell et al., 2013). Due to a larger isotope fractionation in MOB-I hopanoid synthesis compared to MOB-II (Jahnke et al., 1999), the relative contribution of the different AeOM pathways was estimated using the relative abundance of MOB-I versus MOB-II within each sample (Eq.2). Since 16S-rRNA does not indicate the relative activity of these microbes, this approach is a semi-quantitative estimation of CH₄-related hopanoid production. The associated isotope endmembers and isotope fractionation during hopanoid biosynthesis are displayed in Tables S6 and S7 and further described in Eqs. (2)–(4), where f_MOB-I and f_MOB-II are the fractions of MOB based on the relative abundance of 16S-rRNA. The Δδ¹³C_MOB-I/II and Δδ¹³C_DOC describe the isotope fractionation from CH₄ to hopanoid and dissolved OC to hopanoid, respectively.

$$\begin{array}{}\text{(2)}& {\displaystyle }\begin{array}{rl}{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{MOB}}& =f_{\text{MOB-I}}\cdot ({\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{{\mathrm{CH}}_{\mathrm{4}}}+\mathrm{\Delta }{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\text{MOB-I}})\\ & +f_{\text{MOB-II}}\cdot ({\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{{\mathrm{CH}}_{\mathrm{4}}}+\mathrm{\Delta }{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\text{MOB-II}})\end{array}\text{(3)}& {\displaystyle }{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{Heterotroph}}={\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{DOC}}+\mathrm{\Delta }{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{DOC}}\text{(4)}& {\displaystyle }\begin{array}{rl}& {\mathrm{CH}}_{\text{4-derived hopanoids }\left(\mathrm{\%}\right)}\\ & ={\displaystyle \frac{\left({\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{hopanoid}}-{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{Heterotroph}}\right)}{\left({\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{MOB}}-{\mathit{\delta }}^{\mathrm{13}}{\mathrm{C}}_{\mathrm{Heterotroph}}\right)}}\times \mathrm{100}\end{array}\end{array}$$

3 Results

3.1 Large-scale spatial patterns of dissolved CH₄ concentrations

The dissolved CH₄ concentrations in the Laptev Sea were high compared to the atmospheric equilibrium at ∼ 4 nM. The sub-pycnocline CH₄ concentrations for the 26 studied stations varied widely between 8 and 11 850 nM (median = 278 (IQR = 88–966) nM), with the highest concentrations at station 47 in the Outer Laptev Sea hotspot region (OLS; Fig. 2). The sub-pycnocline CH₄ concentrations were significantly different between all regions (Table S1). The CH₄ concentrations in the OLS (defined as latitudes ≥ 76.1° N and longitudes 125–131° E) were significantly higher compared to the Inner and Mid Laptev Sea (ILS/MLS; defined as latitudes < 73.9° N and 73.9–76.1° N and longitudes ∼ 130° E, respectively; Table S1). In all regions, dissolved CH₄ was significantly lower in the surface water compared to below the pycnocline (Table S2). Additionally, surface water CH₄ was significantly higher in the ILS hotspot region (Table S3) compared to the MLS and OLS (Table S3).

Figure 2 Dissolved CH₄ concentrations (nM). (a) Average CH₄ concentrations below the pycnocline. (b) The outer Laptev Sea hotspot region (OLS). (c) Mid-Outer Laptev Sea transect (MLS). (d) The inner Laptev Sea hotspot region (ILS). The bathymetry was plotted using IBCAO version 5.0 (Jakobsson et al., 2024).

3.2 Organic geochemistry

3.2.1 Bulk organic carbon

The total Organic Carbon (OC) content displayed strong variability across the Laptev Sea. The total OC varied between 0.2 % and 2.3 % (mean ± standard deviation; 1.0 ± 0.5 %, n= 25), with the lowest concentrations observed in the the OLS (0.9 ± 0.4, n= 14), and higher concentrations in the MLS (1.0 ± 0.5, n= 5) and ILS (1.3±0.8, n= 6). The δ¹³C-OC ranged from −26.8 ‰ to −23.3 ‰ (−24.7 ± 1.0 ‰, n= 25), with the highest values in the OLS (−23.9 ± 0.4 ‰, n= 14) and slightly lower values in the MLS (−25.0 ± 0.7 ‰, n= 5) and the ILS (−26.1 ± 0.5 ‰, n= 6).

3.2.2 Hopanoid lipid concentrations and spatial distribution patterns

Hopanoids were present at detectable levels across all surface sediments. The sum of C₃₀-hopenes (∑C₃₀-hopenes; diploptene, hop-17(21)-ene and neohop-13(18)-ene) displayed a large range of concentrations across the Laptev Sea (mean ± standard deviation; 27 ± 19 µg gOC⁻¹; n= 25). The highest concentrations were found in the OLS (35 ± 20 µg  gOC⁻¹; n= 15; Fig. 5). Relatively high ∑C₃₀-hopenes concentrations were also present in the ILS, with no significant difference compared to the OLS (18 ± 11 µg gOC⁻¹; n= 6; Fig. 5; Table S4). In contrast, the concentrations of ∑C₃₀-hopenes were significantly lower in the MLS (10 ± 4 µg gOC⁻¹; n= 4; Fig. 5) compared to the OLS, but did not show significantly different concentrations compared to the ILS (Table S4). The large standard deviation of hopanoid concentrations in the OLS and ILS is likely related to the presence of main “hotspot” stations, causing a log-normal distribution of the data. On average, diploptene (hop-22(29)-ene) comprised the majority of C₃₀-hopenes (68 %), with a lower abundance of its diagenetic products (hop-17(21)-ene and neohop-13(18)-ene; 19 % and 13 %). While diploptene dominated the C₃₀-hopene abundance across all regions, it comprised a larger part of the C₃₀-hopenes in the ILS (74 %) compared to the MLS and OLS (52 % and 69 %, respectively). In the MLS in particular, hop-17(21)-ene and neohop-13(18)-ene constituted a larger part of the ∑C₃₀-hopenes (31 % and 18 %).

3.2.3 Hopanoid isotope composition and spatial distribution patterns

The compound-specific isotope composition of δ¹³C-C₃₀-hopenes (diploptene, hop-17(21)-ene and neohop-13(18)-ene) in the Laptev Sea was remarkably depleted compared to the bulk OC. The δ¹³C-C₃₀-hopenes were on average −47.8 ± 7.2 ‰ (mean ± standard deviation; n= 23; Fig. 5). The δ¹³C-C₃₀-hopenes were lowest in the OLS, with a mean ± standard deviation of −52.9 ± 4.3 ‰ across the region (Fig. 5). Hopanoids in MLS and ILS exhibited depleted but relative to the OLS more enriched δ¹³C-C₃₀-hopenes (−42.0 ± 1.9 ‰ and −39.4 ± 3.0 ‰; Fig. 5). No significant differences were observed between the two regions (Table S5), yet both were much more depleted than their regional δ¹³C-OC.

Amongst the C₃₀-hopanoids, δ¹³C-diploptene and diplopterol displayed the lowest values in the OLS (−59.5 ‰ and −61.0 ‰; Fig. 3). Hop-17(21)-ene and neohop-13(18)-ene likewise demonstrated low δ¹³C, yet neohop-13(18)-ene was more enriched in ¹³C in the OLS (Fig. 3). However, in the MLS/ILS diploptene and diplopterol in particular were more enriched in δ¹³C, up to a maximum of −34.1 ‰ and −29.0 ‰, respectively (Fig. 3). On the contrary, δ¹³C of hop-17(21)-ene exhibited lower values compared to diploptene in the ILS (−41.9 ± 4.6 ‰ and −37.6 ± 2.2 ‰; Fig. 3). While diplopterol was also present in the ILS samples, coelution with other molecules hindered accurate concentration and isotope measurements. Hence, δ¹³C-diplopterol was not quantified in the ILS. In summary, δ¹³C-C₃₀-hopanoids were generally low across the Laptev Sea, especially in the OLS. However, in the ILS, hop-17(21)-ene was more depleted in comparison to other hopanoids (Fig. 3).

Figure 3 (a) Overview of the measured δ¹³C of individual C₃₀ hopanoids across each region. The shaded green zone indicates the range of δ¹³C-heterotrophic hopanoids by production from DOC after isotope fractionation during hopanoid synthesis (Tables S6 and S7; Salvado et al., 2016; Schwartz-Narbonne et al., 2023). The light blue indicates the measured δ¹³C-OC in this study. (b) Schematic of δ¹³C of carbon substrates used for hopanoid production and the associated isotope fractionation (Δδ¹³C; Tables S6 and S7; Jahnke et al., 1999; Salvadó et al., 2016; Steinbach et al., 2021; Schwartz-Narbonne et al., 2023; Brussee et al., 2026). Semi-quantitative estimations of the Methane Oxidizing Bacteria (MOB; orange) in each region are the δ¹³C-hopanoid endmembers, where the isotope fractionation of MOB-I and II, respectively, was weighted against the relative fraction of MOB (Eq. 2). The pink displays the δ¹³C-CH₄ in each investigated region (Table S6). (c) Measured δ¹³C-C₃₀ hopanoids compared against the hopanoid and carbon source endmembers in each region.

3.3 Bacterial consortia of hopanoid producers

Aerobic methanotrophs, here referred to as Methane Oxidizing Bacteria (MOB), were present in 9 out of the 11 sediment samples probed for 16S-rRNA. Both type I (Gammaproteobacteria; MOB-I) and type II (Alphaproteobacteria; MOB-II, Fig. 4; Fig. S1) MOB were detected in those 9 samples. Type I MOB constituted the majority of the MOB in the OLS compared to MOB-II (mean ± standard deviation; 1.2 ± 1.8 % of all sequences, n= 5; 0.10 ± 0.08 %, n= 3; Fig. 4). Specifically, the highest relative abundance of MOB-I was present at station 47 (4.3 %: Fig. 3). While the relative abundances of MOB were overall low in the MLS, MOB-II were more abundant (0.04 ± 0.01, n= 2 and 0.01 ± 0.01 %, n= 2; Fig. 4). On the other hand, in the ILS MOB I and II were of similar relative abundances (0.18 ± 0.15 %, n= 2 and 0.17 ± 0.15 %, n= 3), yet almost a factor ten lower than in the OLS. The most common MOB-I were of the order Methylococcales, specifically of the genera Methyloprofundus and Marine methylotrophic group 2 (Fig. S1). Other common type I MOB in the Laptev Sea were pLW-20, Milano-WF1B-03, Milano-WF1B-42 and pItb-vmat-59 (Fig. S1). Crenothrix was only detected in the ILS (Figs. 4; S1) and IheB2-23 and Cycloclasticus only at the OLS stations (Fig. S1). Candidates of Type II MOB were of the family Methyloligellaceae, with one identified genus known to produce hopanoids (Methyloceanibacter). However, it should be recognized that not all Methyloligellaceae are necessarily methanotrophs. In addition to MOB, heterotrophic and autotrophic hopanoid producers were present in the Laptev Sea, particularly in the ILS (Fig. S2). Taken together, MOB-I constituted the largest part of the bacterial hopanoid producers in the OLS, with almost ten times higher relative abundance compared to the MLS/ILS. Hopanoid producers in the ILS/MLS were a mix of MOB-I/II and non-methanotrophic hopanoid producers, primarily heterotrophic bacteria (Fig. 4).

Figure 4 Relative abundance of hopanoid producers in the surface sediments (%), based on 16S-rRNA. (a) Type I MOB. (b) Type II MOB. (c) Other hopanoid producers. The bathymetry was plotted using IBCAO version 5.0 (Jakobsson et al., 2024).

4 Discussion

4.1 Indications of widespread enhanced methane cycling across the Laptev Sea displayed by lipid biomarkers

Depleted δ¹³C-C₃₀ hopanoids in surface sediments indicate large-scale enhanced CH₄ cycling across the Laptev Sea. As the Laptev Sea water and surface sediments are well oxygenated (Stepanova et al., 2017; Brüchert et al., 2018; Maciute et al., 2025), the CH₄ signal in hopanoids is most likely related to aerobic processes. The Laptev Sea is frequently ventilated through storms, resulting in a wide variety of dissolved CH₄ concentrations dependent on the time of sampling relative to storm ventilation of the water column CH₄ (e.g., Shakhova et al., 2014). In contrast, the CH₄ signal from δ¹³C-C₃₀ hopanoids accumulated in sediments indicates a CH₄ signal averaged over several years. Thus, δ¹³C-C₃₀ hopanoids provide means to assess long-term CH₄ cycling patterns compared to the ephemeral dissolved CH₄ concentrations in the Laptev Sea. Hence, the δ¹³C-C₃₀ hopanoids help to unearth CH₄ dynamics in a region that is hard to access and thereby void of time-series monitoring of CH₄. The proxy-derived CH₄ signal in the isotopes of these hopanoids also corroborates previous suggestions of the Laptev Sea as a region of widespread CH₄ cycling in the water column (e.g., Semiletov, 1999; Shakhova et al., 2010, 2014; Steinbach et al., 2021).

4.2 Spatial variations of methane concentrations and hopanoid-isotope patterns

4.2.1 Intensive CH₄ cycling in the Outer Laptev Sea hotspot region

The high abundance of very depleted δ¹³C-C₃₀ hopanoids in the OLS highlights ubiquitous enhanced CH₄ cycling through AeOM across this region (Fig. 5). High concentrations of C₃₀ hopenes in the OLS hotspot region are broadly consistent with previous observations of diploptene from surface sediments sampled in this area in 2011 and 2016 (Grinko et al., 2020, 2025). Beyond earlier studies, our findings of depleted δ¹³C-C₃₀ hopanoids in this region confirm a methanotrophic source of the hopanoids. The presence of very low δ¹³C-C₃₀ hopanoids in the OLS is a strong indication of MOB-I, which is further corroborated by 16S-rRNA (Figs. 4 and 5). Indeed, the application of a stable isotope mixing model (Eq. 4) suggests that 59 ± 4 % of hopanoids in the OLS are related to MOB-I and II. As the mixing model is weighted against the abundance of MOB-I and II and not their respective activity, these results should be regarded as semi-quantitative. Nonetheless, it supports the hypothesis of CH₄ incorporation in hopanoids. Additionally, diplopterol is slightly more depleted in ¹³C in the OLS compared to other C₃₀ hopanoids; this might reflect increased depletion in ¹³C of hopanoids for each step in the biosynthetic pathway (i.e., ¹³C isotope depletion from squalene to diploptene, to diplopterol; Summons et al., 1994). In short, very low values of δ¹³C-C₃₀ hopanoids together with high CH₄ concentrations and a high relative abundance of MOB-I indicate the OLS as a region of extensive CH₄ emissions and associated AeOM (Figs. 2b and 5).

4.2.2 AeOM-biomarkers tracing methane cycling in the Mid-Outer Laptev Sea transect

In the MLS, lower yet readily detectable C₃₀ hopene concentrations suggest less MOB biomass present in this region. Nonetheless, the depleted δ¹³C-C₃₀ hopenes signals and presence of MOB in the MLS regime indicate hopanoids tracing CH₄ cycling (Figs. 4 and 5), with the source apportionment indicating that MOB contributed to 55 ± 15 % of the hopanoid production. In situ hopanoid production in the MLS region or surface water transport of hopanoids from the ILS hotspot are possible causes. Given that CH₄ concentrations are higher in the sub-pycnocline waters of the MLS than in the surface water, where transport from the ILS would be centered and that a larger fraction of hopanoids comes from MOB compared to in the ILS, it seems likely that most hopanoids in the MLS are locally produced. Prior assessments have considered the MLS as a region of continuous subsea permafrost and comparatively low CH₄ concentrations (Shakhova et al., 2010; Bukhanov et al., 2023). In contrast, here we show a proxy-derived record of CH₄ in a region that was previously conceived as being of low CH₄ cycling. However, additional modern calibrations of these hopanoid proxy tools in regions with lower CH₄ concentrations are necessary to confirm whether this indicates an onset of enhanced CH₄ cycling, or if it is regarded as a “background signal” of CH₄-related hopanoids in marine systems. Taken together, we display a record of biomarker proxy-derived CH₄ in the MLS, most likely dominated by a more diffuse regional CH₄ source in the MLS. The presence of low δ¹³C-C₃₀ hopenes across the Laptev Sea indicates the potential to trace proxy-derived CH₄ in all investigated regions and displays important factors to consider when interpreting such biomarker records.

Figure 5 (a) Sub pycnocline CH₄ concentrations, (b) the total concentration of C₃₀ hopenes (diploptene, hop-17(21)-ene and neohop-13(18)-ene; µg gOC⁻¹) and (c) the average δ¹³C-C₃₀ hopenes (‰) in Laptev Sea surface sediments. The bathymetry was plotted using IBCAO version 5.0 (Jakobsson et al., 2024).

4.2.3 Varying hopanoid sources in the Inner Laptev Sea hotspot region

High concentrations of CH_4, depleted δ¹³C-C₃₀ hopenes values and the presence of MOB highlight the possibility of tracing CH₄ cycling also in the ILS. However, while concentrations of C₃₀ hopenes were relatively high and not significantly different from those in the OLS, the δ¹³C-C₃₀ hopenes in the ILS are less depleted. This can be interpreted in different ways, but is likely a combination of three mechanisms: (1) the relative contribution from MOB-I and II, (2) isotopic dilution from non-methanotrophic sources and (3) lower activity of AeOM in the ILS compared to the OLS. In contrast to the OLS, a higher relative abundance of MOB-II is present in relation to MOB-I in the ILS. Type II MOB can produce hopanoids more enriched in ¹³C compared to MOB-I due to their ability to also assimilate CO₂ into biomass (Jahnke et al., 1999). Methyloceanibacter constituted the only genus of MOB-II in the ILS (Fig. S1) and have to our knowledge only been isolated from marine systems (Takeuchi et al., 2014, 2019; Vekeman et al., 2016). Thus, the MOB-II present in our study is likely an in situ signal of AeOM rather than of coastal influence as was found in the Kara Sea/Yenisei River (De Jonge et al., 2016), despite MOB-II generally dominating terrestrial ecosystems (Hanson and Hanson, 1996; Inglis et al., 2019). Noteworthy, δ¹³C-C₃₀ hopenes values ∼ −40 ‰ has in contemporary peatland settings been interpreted as a strong indication of enhanced CH₄ cycling related to MOB-II (Inglis et al., 2019). In fact, even δ¹³C-diploptene as high as −35 ‰ has been interpreted as an indication of AeOM in the Cariaco Trench (Freeman et al., 1994), due to co-occurring autotrophic biomarkers with a δ¹³C composition of > 4 ‰ higher than diploptene (Freeman et al., 1994). Therefore, the less depleted δ¹³C-C₃₀ hopenes, yet in similar concentrations to the OLS, can be an indication of MOB-II and shows that a proxy-derived CH₄ signal in coastal regions needs thorough system knowledge to depict the source.

The higher δ¹³C-C₃₀ hopenes in the ILS compared to the OLS can also be explained by a dilution of hopanoid production from non-methanotrophic bacteria. Hopanoids (e.g., diploptene, hop-17(21)-ene) are directly synthesized or produced as degradation products of bacteriohopanepolyols (BHPs; Rohmer et al., 1980; Mackenzie et al., 1981). A mixture of hopanoid sources is to be expected, as ∼ 10 % percent of bacteria can synthesize hopanoids (Ourisson et al., 1979; Fischer et al., 2005; Racolta et al., 2012; Belin et al., 2018). Relatively high concentrations of bacteriohopanetetrol (BHT) have previously been found in the ILS and have been attributed to terrestrial sources (Bischoff et al., 2016). This study thereby indicates that a dilution from non-methanotrophic terrestrial sources may contribute to the higher δ¹³C-C₃₀ hopenes, due to the vicinity of the Lena River delta and higher relative abundance of non-methanotrophic hopanoid producers compared to MOB. This is further strengthened by δ¹³C-OC around ∼ 26 ‰, indicating a larger terrestrial loading to the sediments of the ILS. Hopanoid source apportionment displayed that 77 ± 7 % of hopanoids in the ILS are related to non-methanotrophic hopanoid synthesis and 16S-rRNA data indicate that the “other hopanoid producers” are partly of terrestrial origin (Figs. 4; S2). Consequently, the presence of hopenes more enriched in ¹³C is also an indication of an additional contribution from non-methanotrophic hopanoid sources, rather than reflecting lower CH₄ cycling in this region. It should also be noted that the high CH₄ concentrations in the ILS originate from an old biogenic source in the coastal region (Sapart et al., 2017) and not riverine input of CH₄ (Shakhova and Semiletov, 2007; Shakhova et al., 2010, 2014, 2017).

Despite high CH₄ concentrations in the ILS, higher δ¹³C-C₃₀ hopenes compared to the OLS hotspot region could alternatively also indicate less active AeOM. Incubation experiments of ILS surface sediments and water show the possibility of AeOM (Bussmann et al., 2017, 2021; Tikhonova et al., 2021), yet at lower rates compared to the OLS (Tikhonova et al., 2021). Taken all together, the relative enrichment of δ¹³C-hopenes in the ILS compared to the OLS hotspot is likely a combination of: (1) a higher relative input of hopanoids from MOB-II, (2) non-methanotrophic bacteria produced in situ and in terrestrial settings and (3) lower activity of AeOM compared to the OLS. Nonetheless, the depleted δ¹³C-hop-17(21)-ene does indicate the presence of AeOM and thereby traces CH₄ cycling (Fig. 3).

4.3 Large-scale implications for Laptev Sea CH₄ cycling

Since the discovery two decades ago, the Laptev Sea CH₄ dynamics and sources have been increasing. It has been suggested that gas-hydrate-bearing sediments in the region store vast amounts of CH₄ (Soloviev et al., 1987), functioning as a possible CH₄ source upon destabilization (e.g., Shakhova et al., 2010). Thawing subsea permafrost has also been suggested as a source of CH₄ in the ILS (Shakhova et al., 2014; Sapart et al., 2017) and to serve as a conduit for leakage of deep thermogenic CH₄ gas reservoirs in the OLS (Steinbach et al., 2021). Our biomarkers of AeOM across the Laptev Sea indicate the presence of a microbial sink, but the high CH₄ concentrations question its efficiency. Historically, a large focus has been given to the ILS and OLS regions with detected CH₄ ebullition (e.g., Shakhova et al., 2010, 2014; Sapart et al., 2017; Steinbach et al., 2021), but few investigations have paid attention to the MLS as the observations have reported lower CH₄ concentrations in that region, yet still elevated relative to atmospheric partial pressures. Thus, our indications of enhanced CH₄ cycling in the MLS stimulate a few questions. (1) Why has subsea permafrost-related CH₄ release only been observed in the ILS/OLS? (2) Is the CH₄ activity in the MLS linked to ILS CH₄ cycling, or is there a large-scale diffuse source across the MLS?

Many challenges still exist regarding our understanding of the state of current subsea permafrost and associated CH₄ releases. In situ bottom sediment temperatures point towards thawing permafrost in the ILS/OLS CH₄ hotspot regions (Chuvilin et al., 2022; Bukhanov et al., 2023). However, the state of the subsea permafrost and consequent CH₄ releases from these two regions are likely caused by different mechanisms. Inundation of the OLS and accompanied heat exposure from both Atlantic water and underlying fault zones transporting geothermal heat have likely contributed to its warming thermal state (Chuvilin et al., 2022). Indeed, areas of warmer sediment temperatures in the OLS are coinciding with CH₄ seepages and underlying fault zones (Baranov et al., 2020; Chuvilin et al., 2022). In the ILS, heat flux from the Lena River is proposed to be the main mechanism thawing subsea permafrost, but thawing is also likely induced by geothermal heat through the Ust' Lenskii Rift (Chuvilin et al., 2022; Bukhanov et al., 2023). Additionally, the bottom water from the Lena River plume has increased by ∼ 1 °C between 1999–2012 (Shakhova et al., 2014) during which time subsea permafrost has rapidly thawed (Shakhova et al., 2017). Thus, the state of subsea permafrost and lower CH₄ cycling in the MLS can be related to the mechanisms regulating the thermal state of subsea permafrost (i.e., heat flux from geothermal or riverine sources).

The surface water CH₄ concentrations in the MLS are lower than in the ILS and OLS, yet they exceed the atmospheric equilibrium (4 nM) by ∼ 4.6 times (median = 18.6 nM). However, the source of CH₄ in the MLS is not fully clear. Assuming that a regional CH₄ source is present in the MLS, further investigations are necessary, especially if it is related to the subsea permafrost thaw-state. In situ bottom sediment temperatures indicate that the MLS is on the border of discontinuous/continuous subsea permafrost (Chuvilin et al., 2022; Bukhanov et al., 2023) and thus possibly on the “thaw-front”, which could be an indication of the thaw-state regulating the CH₄ emissions from the MLS. If a regional-MLS source of CH₄ is present in this region, the source of these CH₄ emissions and the associated AeOM needs to be further investigated. Taken together, there may be geospatial thaw fronts of the subsea permafrost from the OLS moving inwards and from the ILS moving outwards, thus both over time leading to a shrinking MLS transition zone and growing coverage of the extensive areas with accompanying CH₄ releases.

5 Conclusions

This study demonstrates that δ¹³C-C₃₀ hopanoids can be used as time-integrated proxies to deduce geospatial patterns of CH₄ cycling from marine sediments. A clear signal of CH₄ being incorporated into hopanoids is shown throughout the wide outer Laptev Sea hotspot region, indicating the possibility to also investigate time trends of enhanced CH₄ cycling from this region in chronologically-constrained sediment cores. Additionally, a signal of enhanced CH₄ cycling is discovered in the Mid-Outer Laptev Sea transect (MLS), previously thought to be a region of comparatively low CH₄ cycling. The MLS signal could be derived from the Inner Laptev Sea hotspot (ILS), yet, more likely, from a diffuse source in the MLS itself. In the ILS, high CH₄ concentrations were less well reflected in the molecular fossils, indicating AeOM, yet still indicative of CH₄. This is attributed to a higher relative abundance of MOB-II that utilize both CH₄ and CO₂ for hopanoid production, isotopic mixing with non-methanotrophic bacteria, partly of terrestrial origin, producing hopanoids, but also less active AeOM in the ILS hotspot. This indicates that complementary evidence to δ¹³C-C₃₀ hopanoids and thorough system knowledge is necessary in coastal settings in the vicinity of large rivers. However, low AeOM in a region of high CH₄ concentrations, such as the ILS, could have consequences for the amount of CH₄ emitted to the atmosphere. To the best of our knowledge, this is the first modern calibration of δ¹³C-hopanoids over an extensive marine setting, facilitating improved interpretations of hopanoid sources in contemporary and past climates. Furthermore, this study provides a large-scale geospatial picture of Laptev Sea CH₄ cycling, with considerations for future studies of CH₄ cycling in the circum-Arctic shelf regions.