Seafloor imagery was collected in 2022 during the CAGE22-3 cruise (Ferré et al., 2022). During the second ROV dive of the cruise, three sub-areas (Area 01, Area 02, Area 03) were surveyed using high-resolution video acquisition for photogrammetric mosaicking (Fig. 1C). They were selected based on intense seep activity previously identified by screening the water column for gas flares (high backscatter) using R/V-mounted multibeam echosounder (MBES) data (Ferré et al., 2024). ROV Ægir 6000, a 150 Hp work-class vehicle equipped with various samplers and sensors and operated by the Norwegian Marine Robotics Laboratory (NORMAR) at the University of Bergen (UiB), was deployed with an EM 2024 Kongsberg MBES mounted on it and a dedicated photogrammetry sledge. The MBES was used to create a high-resolution microbathymetric map of the surveyed region (Fig. 1). Mounted on the ROV, this MBES captures detailed topographic data at sub-meter to decimetre scales, offering essential context for interpreting seafloor features in the imagery and improving the spatial precision of photogrammetric outputs. The photogrammetry sledge included a downward-facing Imenco^® Spinner II Shark HD zoom camera mounted perpendicular to the seafloor. Illumination was provided by two high-output strobes (>2500 W) and a pair of deep-sea power lasers spaced 14 cm apart, ensuring consistent lighting and enabling scale calibration. ROV video acquisition, with the downward-facing camera, was conducted along parallel transects spaced 2 m apart, with the ROV maintaining a steady speed of 0.15 m s⁻¹ and an altitude of 2.5 m. This configuration resulted in a field of view slightly greater than 2 m and ensured sufficient image overlap for photogrammetric reconstruction. These acquisition parameters were validated in previous field campaigns (Fallati et al., 2023; Panieri et al., 2024). In total, four ∼ 35 m transects were collected in Area 01, five ∼ 30 m transects in Area 02, and six ∼ 60 m transects in Area 03. ROV positioning for both MBES and photogrammetric surveys was achieved using the HIPAP 501 USBL (Ultra Short Baseline) system, ensuring high-precision localisation. These positional data were later integrated into the geospatial processing workflow to scale and georeference the resulting models accurately. ROV videos were processed by automatically extracting one frame per second using VLC Media Player's Scene Video Filter, producing high-definition (1920 × 1080 px) PNG images sorted by survey area. These images were imported into Agisoft Metashape Professional 2.0 (Agisoft 2018) and processed following a standard Structure from Motion (SfM) workflow (Fallati et al., 2020; Lim et al., 2020; Montes-Herrera et al., 2024; Price et al., 2019). After camera alignment, dense point clouds were generated to produce ultra-high-resolution digital elevation models (DEMs) and orthomosaics. The spacing between the laser pointers served as a ground-truth scale bar for each model. To georeference the models, USBL-derived coordinates and timestamps were linked to selected video frames, primarily those at the ends and centres of each transect. These data points were further cross-validated against geomorphic features visible in both the ROV-derived microbathymetry and the SfM-generated DEMs. This multisource integration enabled the accurate spatial placement of the models within the WGS 84/UTM Zone 33N coordinate system. Substrate semi-automatic classification was performed using Object-Based Image Analysis (OBIA) in the eCognition Developer (Trimble) environment. This approach segments the orthomosaics into meaningful image objects based on spectral, textural, and spatial characteristics (Benz et al., 2004; Blaschke, 2010; Hay and Castilla, 2008; Hossain and Chen, 2019), allowing for the supervised classification of distinct substrate classes, such as fine-grained sediments mixed with gravels and pebbles, microbial mats, boulders, and benthic organisms. The interpretation of the observed hardgrounds as MDAC pavements was based on previous extensive studies at this site which reported low δ13C values for these materials ranging from -38.0 ‰ to -22.2 ‰ and the incorporation of biomarkers typical of AOM-performing microbial consortia (Crémière et al., 2016; Sauer et al., 2017).

Sediment samplings near the mounds and adjacent sediment were hindered by the presence of a stiff layer of coral debris and/or hardgrounds partially exposed at the seafloor. Moving away from the mounds, we managed to obtain cores for biogeochemical investigations and decided to focus on these areas for samplings and orthomosaicking. Six push cores, three blade cores, and one gravity core were collected during the expedition CAGE22-3 (Ferré et al., 2022) (Fig. 1C). The blade corer consisted of a 32 cm-long frame with a rectangular base of 25 × 10 cm. It was pushed into the seabed using the ROV manipulator arm and a locking system was activated to avoid sample loss. This technique is especially suitable for coarse-grained sediments. We collected three blade cores CAGE22-3-KH-01-BlaC-01, CAGE22-3-KH-01-BlaC-02, and CAGE22-3-KH-01-BlaC-03, hereafter named BlaC-01, BlaC-02 and BlaC-03, respectively. BlaC-01 was collected as a reference core for geochemical interpretations from a seafloor spot barren of chemosynthetic fauna (Ferré et al., 2022). BlaC-02 and BlaC-03 were collected from microbial mats. We also conducted push coring using a 60 cm-long cylindrical tube (8 cm inner diameter) made of fiberglass. The push corer was pushed into the sediment by the ROV during samplings and enabled us to obtain longer cores. We collected six push cores: CAGE22-3-KH-01-PusC-01, CAGE22-3-KH-01-PusC-02, CAGE22-3-KH-01-PusC-04, CAGE22-3-KH-01-PusC-05, CAGE22-3-KH-01-PusC-07 and CAGE22-3-KH-01-PusC-08, hereafter named PusC-01, PusC-02, PusC-04, PusC-05, PusC-07 and PusC-08, respectively and all collected from microbial mats. We also report the data from gravity corer CAGE22-3-KH-01-GC-01 (GC-01), which was collected using a 6 m-long iron barrel hosting a PVC liner with an inner diameter of 10 cm. On deck, the cores were split into 1 m-long sections (or shorter depending on the recovery), capped, labelled, and stored at 4 °C. Pore fluid analyses were conducted on all the cores, while PusC-08 and BlaC-01 were additionally selected for sediment geochemistry and foraminifera analyses. Preliminary microbiological analyses were only performed on PusC-08 after discovering a biofilm at 10 cm, suggesting strong biogeochemical and microbiological gradients.

Pore water extractions were conducted in a cold room (4 °C) immediately after coring operations. We sampled every 2 cm using rhizon soil moisture samplers (5 cm-long filter; 0.15 µm mesh) and 10 mL sterile plastic syringes. We split the pore water samples into two aliquots: one aliquot for dissolved inorganic carbon (DIC) analyses was transferred into 2 mL glass scintillation vials to which we added 10 µL of HgCl₂ saturated solution. DIC samples were then stored in the dark at 4 °C; the other aliquot for sulfate analysis was transferred into 5 mL Eppendorf tubes and stored at -20 °C. DIC concentrations and isotopic composition (δ13C) were measured on a Thermo Fisher Scientific MAT 253 IRMS coupled to a Gasbench II at the Stable Isotope Laboratory (SIL) of the Department of Geosciences, UiT. Isotopic values were normalized to Vienna Pee Dee Belemnite standard (VPDB) using three in-house calcite reference materials covering a δ13C range from -48.95 ‰ to 1.96 ‰. DIC concentration was calculated by comparing the IRMS peak areas for the samples with calibration curves made from two NaHCO₃(aq) stock solutions. Repeatability precision (1 s) of δ13C and DIC concentration based on three duplicate samples was better than or equal to 0.23 ‰ and 1.2 mM, respectively. The sulfate concentration was measured via ion chromatography at TosLab in Tromsø (NO). Accuracy and precision were estimated based on repeated measurements of certified materials. The measured values agreed within the uncertainty of the certified value and are commonly associated with a precision of 15 % (relative standard error RSE). Bulk sediment samples were collected from the bottom of push cores and from depths ranging from 5 to 15 cm in blade cores using a cut-off syringe. Around 5 mL of sediment were transferred to 20 mL serum vials containing two glass beads and 5 mL of 1 M NaOH. The vials were immediately closed with a rubber septum and aluminum crimp seal and stored upside-down at 4 °C. Methane concentration was measured at the SIL laboratory using a Thermo Fisher Scientific GC Trace 1310 gas chromatograph equipped with a Rt-Alumina BOND/Na₂SO₄ PLOT column and calibrated using 100 and 1000 ppm Calgaz and Linde calibration gases. Air was measured for quality control throughout the session. Injected volumes ranged from 10 to 1000 µL and measurement temperatures were set to 50–150 °C.

We conducted down-core foraminiferal investigations in PusC-08 (microbial mat microhabitat) and in reference core BlaC-01. The isotopic composition of foraminifera was used to trace methane-derived carbon incorporation and/or MDAC precipitation as secondary overgrowth. This geochemical proxy is widely used in paleo-seep reconstructions (Argentino et al., 2021, 2024; Boretto et al., 2026; Schneider et al., 2018; Yao et al., 2020). We combined foraminifera geochemistry with pore water data to determine whether isotopic anomalies in foraminifera are associated with modern or ancient Sulfate-Methane Transition Zones (SMTZ). Push core PusC-08 and blade core BlaC-01 were sliced onboard every cm. For each sample, we weighed around 50 g of dry sediment and wet-sieved it through 63 and 125 µm sieves. We dried the residue at 50 °C and hand-picked benthic foraminifera with identification at genus or species level. Picking was conducted from the fraction >125 µm using a stereomicroscope. From each aliquot we selected 5 to 15 individuals for isotope analysis. The dominant species in most of the samples were Cibicidoides lobatulus and Cassidulina spp. Some of the samples were barren of these target foraminifera, thus resulting in a total of 41 and 13 samples for PusC-08 and BlaC-01, respectively. Isotope analyses were conducted at the SIL laboratory using a Thermo Fisher Scientific MAT 253 IRMS with a Gasbench II. Precision on two replicate reference materials (n=6) was better than 0.1 ‰ for both carbon and oxygen. All isotope results were normalized to Vienna Pee Dee Belemnite standard (VPDB) using three in-house reference materials ranging in δ13C and δ18O from -48.95 ‰ to 1.96 ‰ and from -18.59 ‰ to -2.15 ‰, respectively.

We conducted geochemical investigations of the organic matter in core PusC-08 and BlaC-01 to identify the presence of methanotrophic biomass in the sediment. In combination with pore water data, it is possible to discriminate between modern or ancient AOM-impacted intervals. Negative δ13C_org values lower than the ranges for marine and terrestrial end-members can be used as a proxy for AOM-associated microbial biomass in the sediment (Argentino et al., 2023, 2024; Ghosh et al., 2025; Szymczycha et al., 2025). We analysed a total of 45 sediment samples from PusC-08 and 15 samples from BlaC-01 collected every cm down the cores. Prior to the analyses, the samples were treated with 6 N HCl to remove the carbonate component following the protocol reported in Argentino et al. (2023). Analyses were conducted on a Thermo Fisher Scientific MAT 253 Isotope Ratio Mass Spectrometer (IRMS) coupled to a Flash HT Plus Elemental Analyzer hosted at the SIL laboratory. The δ13C and δ15N values were obtained by normalization to international standards, Vienna Pee Dee Belemnite, VPDB (δ13C) and Air-N₂ (δ15N), using three in-house reference materials ranging in δ13C and δ15N from -40.44 ‰ to -9.14 ‰ and from -4.91 ‰ to 20.73 ‰, respectively. The total nitrogen in the decarbonated material (TN_d) is hereafter assumed to represent organic nitrogen based on the close-to-zero-intercept of TN_d versus total organic carbon (TOC) for TOC = 0 (e.g. Knies et al., 2007). This assumption allowed us to consider TN_d values as mixtures of different organic sources, and the bulk nitrogen isotopic composition (δ15N_d) can be used for organic source interpretations (Argentino et al., 2023; Hu et al., 2020). Repeatability precision (1 s) of four duplicate samples measured in the same analytical session was better than or equal to 1.7 ‰, 0.12 %, 0.66 ‰, and 0.011 % for δ13C_org, TOC, δ15N_d, TN_d, respectively. The C/Nd atomic ratio was calculated using the atomic mass weighted ratio of TOC and TN_d as C/Nd = (TOC / 12.011) / (TN / 14.007).

Sediment samples for DNA analyses were collected from PusC-08 immediately after core recovery using sterile spoons and stored frozen at -80 °C until processing. Samples were collected every cm in the interval 0–10 cm, to characterize any vertical gradient from the seafloor down to the biofilm-rich interval at 10 cm. For DNA extraction, 0.2–0.5 g of sediments was extracted using the DNeasy Power Soil Kit (Qiagen, Carlsbad, CA) and stored at -80 °C. DNA quality was measured by a NanoDrop Spectrophotometer and a Qubit Fluorometer (Invitrogen). The V4 region of the 16S SSU rRNA gene was amplified following the Earth Microbiome Project (EMP) (Gilbert et al., 2014) 16S Standard191Illumina library preparation protocol, using the forward-barcoded 515f (5^′-3^′192GTGYCAGCMGCCGCGGTAA) and 806r (5^′-3^′ GGACTACNVGGGTWTCTAAT). The quality of PCR products was assessed using 1 % agarose gel electrophoresis and GelRed staining, and three replicates per sample were chosen for sequencing. Illumina MiSeq sequencing was performed at the Environmental Sample Preparation and Sequencing Facility at Argonne National Laboratory (Lemont, IL, USA). The amplicon analyses were performed with the QIIME2 environment and using QIIME2 plugins as previously described in Aalto et al. (2022). Briefly, Illumina forward and reverse reads and the corresponding barcode files were imported and demultiplexed using the Earth Microbiome Project paired-end flag. All reads were quality filtered, de-replicated, and chimera-checked using all default parameters in DADA2 v2021.2.0 (Callahan et al., 2016). The reads were merged and amplicon sequence variants (ASVs) were determined using DADA2 v2021.2.0. The DADA2 statistic on sequence reads is provided in Data set S1. A 16S from the SILVA v138.1 database was trained using RESCRIPt (Quast et al., 2013; Robeson et al., 2021). The ASVs were classified with the self-trained classifier database (Yilmaz et al., 2014). Sequences are being processed for archiving with the European Nucleotide Archive and are available under accession number PRJEB96327.

The three orthomosaics (Fig. 2.A1, A2, A3), generated at millimetric resolution, cover a total area of 1307 m² and generally show a hard seafloor composed of compact patches of MDAC, colonized by numerous benthic organisms, such as sponges, sea anemones, and sea fans. Outside the MDAC areas, the seafloor is characterized by fine sediments mixed with gravels and pebbles, which dominate the mapped area and include coral rubble.

The high resolution of the orthomosaics enabled the detection and mapping of small, subcircular microbial mat patches, mainly found along the edges of the carbonate crusts (Fig. 2.C1, C2, C3). These patches vary in surface area from 0.002 to 0.17 m², with an average of 0.02 m² and an average perimeter of 0.58 m.

OBIA-derived benthic coverage maps identify the main seafloor classes in the high-resolution orthomosaics. Area 1 (Fig. 2.A1–B1), covering 312 m², primarily consists of fine sediments with gravels, occupying about 267 m². Microbial mats are located near the carbonate crusts (4.6 m²) in the south-central part of the area, with a total surface area of 0.26 m². In the area, two benthic nodes (landers) are present, and well visible in the orhomosaic (Fig. 2.A1) as part of the project LoVeOcean, Lofoten-Vesterålen Ocean Observatory (https://loveocean.no/, last access: 24 August 2022).

Area 2 is smaller (Fig. 2.A2–B2), at 258 m², and features a central, slightly elevated subcircular MDAC patch (42.2 m²), surrounded by fine sediments with gravels and pebbles (211 m²). This area also displays a more complex three-dimensional structure and hosts numerous benthic organisms and small microbial mat patches, which together cover 0.22 m².

Area 3 comprises the largest portion of the mapped seafloor (Fig. 2.A3–B3), spanning 738 m². Most of this area is dominated by fine sediments with gravels (437.7 m²), along with extensive, discontinuous MDAC outcrops (282 m²) concentrated in the northwestern section. Small, subcircular bacterial mats are also visible along the rims of the carbonate crusts, with a total surface coverage of 1.65 m².

All pore water profiles from the 10 investigated cores are reported in the Supplement (Fig. S1) and Fig. 3. Two cores were selected based on completeness of the data (no gaps in either sulfate or DIC profiles) to discuss the subsurface biogeochemistry of microbial mats (Fig. 3). Sulfate concentration in surface sediments (0–1 cm below the seafloor) varies from a normal seawater value ∼ 29 mM (Millero, 2005) to as low as 24.3 mM (PusC-01; presence of a microbial mat at the seafloor). Sulfate values <1 mM are observed in two cores, PusC-02 (Fig. S1) and PusC-04 (Fig. 3), at 24 and 10 cm, respectively. Concentrations of dissolved inorganic carbon in surface sediments (0–1 cm) ranges from 2.2 mM (BlaC-01) to 5.1 mM (PusC-01), with overall increasing downcore trends (Fig. S1). DIC isotopic composition (δ13C) in surface sediment ranges from -22.5 ‰ (PusC-01) to -1.1 ‰ (BlaC-01) (Fig. S1). Isotopic values decrease with depth, with the most negative value δ13C = -29.8 ‰ measured at 6 cm in PusC-04 and at 18 cm in PusC-08, which also corresponds to an inversion in isotopic trends (becoming higher down the core). Methane concentration in samples collected from the bottom of the push cores and at 5–15 cm in blade cores are reported in the Supplement (Table S1). Methane concentrations show high values in pushcores, ranging 1.2 mM (PusC-02) to 3.4 mM (PusC-04). Blade cores contain trace amounts of methane, with measured values of 0.003 mM (BlaC-01; 10 cm depth), 0.010 mM (BlaC-02; 5 cm depth) and 1.1 mM (BlaC-03; 15 cm depth).

The carbon isotopic composition δ13C of Lobatula lobatula, an epifaunal species, and Cassidulina spp., an infaunal genus, shows a marked difference between reference core BlaC-01 and PusC-08. In BlaC-01, δ13C values remain relatively steady throughout the core with L. lobatula ranging from 0.8 ‰ to 1.9 ‰, and Cassidulina spp. ranging from -1.5 ‰ to -1.0 ‰. In push core PusC-08, the δ13C values for both taxa show a wider range and significant depletion. For L. lobatula, values ranged from -18.5 ‰ to 1.7 ‰ and for Cassidulina spp. from -15.0 ‰ to 1.0 ‰. A drop in isotopic values is observed at around 18 cm for both taxa, although more evident in Cassidulina spp. In the blade core, L. lobatula and Cassidulina spp. have δ18O values within the intervals 2.4 ‰–3.5 ‰ and 4.1 ‰–4.3 ‰. In PusC-08, the range of δ18O values are 2.0 ‰–4.1 ‰ and 1.8 ‰–4.6 ‰. BlaC-01 displays rather flat downcore isotope profiles (Fig. 4A), while PusC-08 shows a sharp increase in δ18O and decrease in δ13C starting at 20 cm below the seafloor (Fig. 4B).

The contents of total organic carbon in BlaC-01 range from 0.07 % to 0.58 % and are associated with δ13C_org values from -24.8 ‰ to -22.1 ‰. TN_d contents range from 0.01 % to 0.06 %, with isotopic values δ15N_d between 4.1 ‰ and 5.8 ‰. The resulting C/Nd values range from 8.5 to 16.3. PusC-08 has TOC contents ranging from 0.09 % to 1.78 %, associated with δ13C_org values from -43.4 ‰ to -23.6 ‰. TN_d and δ15N_d varies between 0.01 % and 0.18 %, and between 4.3 ‰ and 5.9 ‰. C/Nd values range from 10.0 to 17.9. A sharp decrease in δ13C_org and C/Nd values is observed between 5–11 cm which matches positive peaks in TOC, TN_d and δ15N_d (Fig. 4B). Similar patterns, though less pronounced, are observed at approximately 17, 25, and 40 cm. During onboard sediment slicing, we observed a macroscopic biofilm at a depth of 10 cm having white color and exhibiting a soft, slimy texture (Fig. S2).

The phylum-level taxonomic composition of 16S rRNA gene–informed ASVs showed a steady, depth-structured shift across the 10 cm sediment column in PusC-08 (Fig. 5A). Triplicates from each depth were largely consistent, with the exception of single outlier replicates at 8 and 9 cm in phylum-level abundances. Across depths, relative abundances changed progressively: Proteobacteria exceeded 50 % at 1 cm but declined with depth, such that by 8 cm the community was largely composed of other phyla. In contrast, Desulfobacterota and Campylobacterota increased from the shallow to deeper layers. Around 7 cm, the profile shifted again, with Halobacterota increasing sharply below this horizon. Looking at the same microbial community on a family level (Fig. 5B), the aforementioned depth-structured shift was less apparent and steady. However, notable shifts could be documented. Thioglobaceae decreased over the first 3 cm of depth whereas Sulfurovaeae increased moving down to 7 cm. Below 7 cm of sediment depth, a clear cut in the community structure could be observed after which organisms of the ANME families could be detected.

Alpha- and beta-diversity patterns indicate a depth-structured reorganization of the microbiome across the push-core. Alpha diversity (observed ASVs) rose steadily from 1 to 6 cm, consistent with increasing richness/niche diversification in the upper layers, then declined below 6 cm, coincident with the depth at which the taxonomic profile began to turn over. Beta diversity resolved three depth-associated clusters: a cohesive group comprising all samples from 1–6 cm; a clearly separated group comprising 8–10 cm; and 7 cm samples positioned between them, consistent with a transitional assemblage. Together, these patterns in PusC-08 point to a stratified microbiome with a boundary around 6–7 cm, above which communities are richer and more similar to one another, and below which communities are less diverse and compositionally distinct.

Chemosynthetic habitats found at cold seeps rely on hydrogen sulfide fluxes from the SMTZ which are maintained by AOM-sustained sulfate reduction (Foucher et al., 2007; Levin, 2005; Levin et al., 2016). There is a causative link between magnitude of methane fluxes, depth of the SMTZ, sulfide fluxes and type of chemosynthetic communities at the seafloor (Argentino et al., 2022; Fischer et al., 2012; Levin, 2005; Levin et al., 2016; Sahling et al., 2002). Hydrogen sulfide in aquatic environments is generally viewed as a toxicant able to interrupt cellular respiration due to its lipid solubility (Bagarinao, 1992; Riesch et al., 2015; Tobler et al., 2016). Therefore, at cold seeps, only organisms that have adapted to varying levels of sulfide exposure are present and their distribution follows subsurface geochemical gradients. Microbial mats are found in correspondence of the most intense seepage areas with shallower SMTZs (∼ 5–10 cm below the seafloor), whereas higher organisms (clams, mussels, tubeworms) in symbiosis with sulfur-oxidizing bacteria colonize lower-seepage intensity areas associated with deeper SMTZs (∼ 10–50 cm below the seafloor) (e.g. Fischer et al., 2012; Sen et al., 2018b; Lee et al., 2019; Argentino et al., 2022; Barrenechea et al., 2025). We identified the SMTZs in our cores where a drop in sulfate content matched a decrease in δ13C_DIC and an increase in DIC concentration, evidence of ongoing AOM which consumes sulfate and releases ¹³C-depleted bicarbonate. In the study area, the SMTZ ranges from ∼ 5 cm (PusC-01, PusC-4) to ∼ 25 cm (PusC-02, PusC-07, PusC-08) (Figs. S1, 3). No SMTZs were intercepted in the other cores. Besides BlaC-01 (and probably gravity core GC-01), all the other cores were collected from microbial mats. However, while the SMTZ depths in PusC-01 and PusC-4 are typical of those expected from microbial mat habitats, the other cores from mats showed deeper SMTZs. To interpret the latter case, we considered the morphology and extent of the microhabitats and local biogeochemical and environmental conditions. The high-resolution seafloor imagery and habitat maps show that the sampled mats form small (∼ 200 cm²) and highly heterogeneous (shape, color) features with reduced mat thickness giving color shades from white to light grey (Fig. 2). Chemosynthetic mats can rapidly colonize new substrates (e.g. seep sites, vents, wood, whale carcasses) within a few months (Alain et al., 2004; Girard et al., 2020; Guezennec et al., 1998; Kalenitchenko et al., 2016), as well as being rapidly eroded away or consumed by grazers when the chemical substrate is no longer present (Niemann et al., 2013; Seabrook et al., 2019; Sen et al., 2018b; Thurber et al., 2013). The appearance of the mats might therefore suggest that the fluid supply from underneath has been recently decreasing. It is known that cold seeps can be dynamic environments characterized by spatial and temporal variations in gas seepage distribution (Ferré et al., 2020; Greinert, 2008). A substantial variability in seepage activity as seen in water column backscatter anomalies was reported between 2018 and 2022 at this site (Ferré et al., 2024). Temporal fluctuations in methane fluxes could also explain the concave shape of the sulfate concentration profiles observed in our dataset and deeper SMTZs in some microbial mat microhabitats (Fig. S1). The profile shapes range from concave-up (PusC-04, PusC08?) to concave-down (PusC-07), which is a common indication of rapid increases and decreases in methane flux, respectively (Hensen et al., 2003; Kasten et al., 2003; Minami and Tatsumi, 2012). Strong bottom currents present in this area (Ferré et al., 2022, 2024) might also contribute to the observed reduced surface area of the single microbial mat patches, since mats are sensitive to physical disturbances (Cardoso et al., 2019; Noffke, 2010; Pan et al., 2019). Although the ROV samplings were conducted accurately, targeting the inner part of the mat patches, it is possible that the anomalous profiles reflect a mixed signal due to the sampling area of our coring devices (push corer = 50 cm²; blade corer = 250 cm²) being comparable to the size of the investigated microhabitats. Moreover, the ex-situ pore water extraction using 5 cm long rhizon filters inserted transversally into the core stratigraphy is also yielding bulk geochemical values. In-situ sensors such as the Raman probe (Zhang et al., 2010, 2011), are crucial for resolving challenges associated with cm-scale horizontal variations in geochemical gradients in microhabitats of small extent, such as those mapped in our study. Regarding the spatial distribution of microhabitats, the seep carbonate pavements in the CWC area act as a barrier to upward migrating sulfide-rich fluids, controlling the distribution of mats around the edges of carbonate crusts or in apparently carbonate-free seafloor areas (Fig. 2). Siboglinid tubeworms are common at Arctic cold seeps (Argentino et al., 2022; Sen et al., 2018a, b) and represent the dominant benthic community at nearby seeps in canyons on the lower continental slope off Lofoten–Vesterålen islands (Sen et al., 2019). The general lack of tubeworms in the investigated seepage site must be related to the sedimentary substrate (Bellec et al., 2024; Sen et al., 2019), i.e. hard carbonate pavements and/or coarse sediment (Fig. 2), since tubeworms require softer sediments to settle (Hilário et al., 2011; Southward et al., 2005).

Geochemical anomalies in δ13C of foraminifera in PusC08 appear below ∼ 18 cm, with remarkably negative isotopic values compared to the reference core. Those anomalies match the modern SMTZ as indicated by DIC data and the sharp drop in sulfate concentration right above (Fig. 4). Based on these observations, we ascribe the foraminiferal anomalies to precipitation of methane-derived carbonate as secondary overgrowth on the tests. Oxygen isotopes in PusC-08 show high δ18O values (up to 4.6 ‰) and similarly high values seem recorded at the bottom of the reference core BlaC-01 (Fig. 4A). Previous measurements on pure carbonate crusts from this site (Crémière et al., 2016; Sauer et al., 2017) indicated seep carbonate precipitation in isotopic equilibrium with seawater, with no influence from gas hydrates, at least during Holocene. Additional foraminiferal investigations on longer sediment cores are needed to interpret the oxygen trends and reconstruct the history of seepage.

In this section we interpret the geochemical anomalies measured in the sediment of PusC-08 and discuss the putative origin of a white biofilm found at 10 cm in the same core (Fig. S2). Generally, the bulk organic matter composition of marine sediments includes a mixture of marine algae and vascular plants debris from land (Burdige, 2005; Emerson and Hedges, 1988; Meyers, 1994, 1997). The elemental and isotopic signatures of these two categories have been widely used for assessing the contribution of marine versus terrestrial organic inputs in paleoceanography (e.g. Kang et al., 2007; Knies and Martinez, 2009; Hasegawa et al., 2013; Knies, 2022). Marine-derived organic matter is commonly associated with δ13C values between -23 ‰ and - 16 ‰ (Emerson and Hedges, 1988; Meyers, 1994), δ15N values as high as 7 ‰ (Wada, 1980), and C/N ratios between 4 and 10 (Meyers, 1994). Terrestrial organic matter from C₃ plants which dominate the Arctic regions has lower δ13C values <-25 ‰ (Meyers, 1994; Naidu et al., 2000; Rachold and Hubberten, 1999; Ruttenberg and Goñi, 1997), δ15N values ∼ 0 ‰ (Kienast et al., 2005), and C/N ratios >20 (Emerson and Hedges, 1988; Meyers, 1994). These ranges can vary regionally but they hold true for most marine settings, with minor adjustments of the end-member compositions (Argentino et al., 2023; Knies et al., 2007; Knies and Martinez, 2009). In PusC-08, δ13C_orgreaches values as negative as -43.4 ‰, which cannot be explained by a simple mixture of marine and terrestrial organic matter. Similar values in bulk sediment were reported from cold seeps in the SW Barents Sea (-42 ‰, Argentino et al., 2023), offshore Pakistan (-42 ‰, Yoshinaga et al., 2014) and interpreted as indicative of significant contributions of methanotrophic biomass to the organic matter pool. In fact, methanotrophic microbes inhabiting cold seeps incorporate variable amounts of isotopically-fractionated carbon from methane (Kurth et al., 2019; Stock et al., 2025; Wegener et al., 2008), which is known to carry negative δ13C signals, generally <-50 ‰ (Judd and Hovland, 2007; Milkov and Etiope, 2018; Whiticar, 2020). Therefore, we confidently ascribe the isotopically-depleted δ13C values in PusC-08 between 5.5 and 11.5 cm to higher contents of AOM consortium biomass. The concomitant drop in C/Nd values in that interval supports this interpretation, since bacterial organic matter has low C/N ratios ∼ 5 (Madigan et al., 2017). Differently from other studies reporting a decrease in δ15N in AOM-impacted layers (Argentino et al., 2023; Hu et al., 2020), here we observe a slight increase (Fig. 4). This difference can be due to different nitrogen assimilation pathways by AOM-SRB consortia and/or different nitrogen substrates isotopic composition (Dekas et al., 2009, 2014; Gruber and Galloway, 2008). This AOM-impacted interval hosts a biofilm and is stratigraphically located above the modern SMTZ (Fig. 4). Macroscopic biofilms similar to the one found in our study have been reported by Briggs et al. (2011) from methane seeps at Hydrate Ridge (offshore of Oregon, USA), northern Cascadia Margin (offshore of Vancouver Island, Canada), and the Indian Ocean (offshore India), and by Gründger et al. (2019) from gas hydrate pingoes in the NW Barents Sea. In those studies, the biofilms consist of ANME-SRB consortia performing AOM (Briggs et al., 2011; Gründger et al., 2019), and the reported isotopic values for the pure biofilm masses yielded δ13C as negative as -43 ‰ (Briggs et al., 2011). We did not conduct carbon-nitrogen analyses on the biofilm material to directly compare with previous studies, but we found a similar value (-43.4 ‰) in bulk sediment from the same interval. What leads to the accumulation of macroscopic masses is the occurrence of pockets, cracks or fractures within the sediment, which provide high accommodation space (Briggs et al., 2011; Gründger et al., 2019). In PusC-08 we observed a network of microfractures in the core but no clear correlation with biofilm position was noticed during sediment slicing. The lack of indication of paleo-SMTZs in geochemical profiles of foraminifera (Fig. 4B) enable us to exclude that the biofilm could represent past conditions with higher methane fluxes but rather suggests ongoing activity. The stratigraphic position of the biofilm suggests that there is methane bypassing the SMTZ and being locally consumed within fractures (Briggs et al., 2011).

Correlations between biogeochemical parameters from pore waters and carbon-nitrogen geochemistry with 16S rRNA ASVs at the genus level reveal a clear division of the microbial community into two distinct groups, driven primarily by TOC, TN_d, SO42-, and δ13C_org (Fig. 6). The first group, comprising archaeal taxa, is associated with organic-rich, nitrogen-rich, sulfate-depleted environments and plays a key role in methane oxidation and carbon fixation. In contrast, the second group, consisting largely of bacterial taxa, thrives in sulfate-rich, inorganic carbon-enriched conditions and is involved in sulfate reduction and sulfur cycling, highlighting the complementary roles of these groups in driving key biogeochemical processes. The first group includes the top 5 genera: Candidatus_Altiarchaeum, ANME-1b, SCGC_AAA011-D5, MSBL2, and NKB15. Among these, Candidatus_Altiarchaeum is known to act as a carbon dioxide sink, contributing to carbon fixation (Probst et al., 2014). Similarly, SCGC_AAA011-D5 belongs to the Nanoarchaeota archaeon, which is also adapted to anaerobic conditions. MSBL2 (classified as Methylocella) is a methanotroph, suggesting its involvement in methane oxidation. Together, these taxa likely contribute to carbon fixation, methane cycling, and organic matter degradation. The negative correlation with δ13C_org suggests that these taxa, particularly ANME-1b, use isotopically light organic matter, likely incorporating the strongly negative isotopic signature of methane-derived carbon into their biomass, further supporting their potential link to AOM. The second group is more diverse, with many taxa positively correlated with SO42- and δ13C_org, suggesting adaptation to sulfate-rich environments and involvement in sulfur cycling, likely including sulfate-reducing bacteria (SRB). Other genera showing strong inverse correlations with sulfate concentration in pore water and with δ13C_org correspond to sulfate-reducing Deltaproteobacteria which assimilates bicarbonate released by organoclastic sulfate reduction in marine sediments, i.e. MSBL2 (Vuillemin et al., 2022). The positive correlation with δ13C could indicate preference for isotopically heavier carbon sources, potentially linked to sulfur-based metabolic pathways. Other correlations relate to genera which are found in “normal” marine sediments unaffected by seepage. Future metagenomic analyses will be conducted on samples from PusC-08 for identifying the key microbial players and their ecological roles.

Along the Norwegian coast, several coral fields have been mapped over the last decades by MAREANO (https://www.mareano.no, last access: 17 December 2025; Hovland, 2008), with some reef associations with hydrocarbon seeps. Among them, the Hola Trough became a case study to investigate the relationships between corals and seeps and monitor their response to environmental perturbations such as ocean acidification. Our study is part of a larger set of investigations conducted within the EMAN7 project. Here we did not focus the analyses on corals, but on the cold seep environment over which they superimpose. Still, we can discuss our results in light of recently-published studies from EMAN7 (Ferré et al., 2024; Sert et al., 2025), and previous work in the same area (e.g. Crémière et al., 2016; Hovland, 1990; Hovland and Thomsen, 1997; Sauer et al., 2015, 2017), to draw some lines of evidence about the geological and biogeochemical influence of seeps on CWCs. During ROV Dives conducted in 2022 (Ferré et al., 2022) microbial mats and MDACs were frequently observed in direct contact (either adjacent to or at their base) with coral mounds shown in Fig. 1b (Ferré et al., 2024), although in our orthomosaics and habitat maps covering the broader seepage environment (Fig. 2) corals are absent. Based on ROV imagery and the generated datasets, we found that CWCs in this area are mostly associated with chemosynthetic substrates (MDAC, microbial mats) and bubbling (Ferré et al., 2024). This is true also for the site to the north-east of the three mosaics, where hundreds of corals occur inside an intense gas emission area (Ferré et al., 2024). From a sediment perspective, we do not expect any difference in methane-related biogeochemical processes getting closer or further from CWCs, with AOM-associated sulfide fluxes sustaining chemosynthesis at the seafloor in spots where the MDAC “seal” is broken. Sub-bottom profiles crossing the two main mounds and reported in Ferré et al. (2024) revealed that the CWCs occur on top of topographic highs, likely related to glacial deposits (Buhl-Mortensen et al., 2012), which overly gas accumulations. Gas seeps identified outside of the profile coverage also show a correlation with small positive structures in the seafloor morphology. Our microbathymetric maps revealed a relatively flat morphology dominated by carbonate pavements. There is a strong topographic control for both gas migration in the shallow subsurface and coral occurrence. The presence of widespread MDAC deposits close to or at the seafloor explains the gas migration paths, which concentrate the gas toward topographic highs. Long-lasting seepage of deep-sourced thermogenic gas (Crémière et al., 2016; Sauer et al., 2015, 2017) resulted in the formation of widespread hard substrate, while the elevated structures shaped by glacial activity provided better exposure to strong bottom currents that supply food to the corals. At the CWC mounds, cold seeps influence the chemical and isotopic composition of dissolved organic carbon (Sert et al., 2025). Whether there is any seep-derived carbon uptake from corals remains unclear. Future results from ongoing coral transplantation experiments at this site will offer insights into this and other biochemical processes.