Introduction

Gas hydrates are crystalline solids formed from water and gas molecules under high-pressure and low-temperature conditions1. They are abundant in marine sediments along continental margins, typically occurring at water depths greater than 400 m, but in the Arctic, they can remain stable on the seafloor at depths as shallow as ~300 m because of the low bottom-water temperatures2. While there is theoretically no maximum depth limit for the stability of seafloor hydrates because of increasing pressure and consistently low bottom-water temperatures, most discovered outcrops occur at depths shallower than 2000 m on continental slopes, where rapid burial of organic matter leads to the formation of hydrocarbon reservoirs. These hydrocarbon accumulations migrate through faults or low-permeability sedimentary pathways towards the seafloor, feeding gas hydrate systems2,3. Gas hydrates constitute an essential global carbon reservoir, estimated to contain (1–5) × 1015 m3 or ~500–2500 Gt (1015 g) C, and are a potential source of atmospheric methane, a potent greenhouse gas4. The gas within hydrates can derive from biodegradation of sedimentary organic matter, such as in deposits on the Blake Ridge (NW Atlantic) and Cascadia margin (NE Pacific), or be thermogenic in origin, formed by decomposition of organic molecules under high temperature and pressure in deep sedimentary strata5, as found in deep deposits in the Gulf of Mexico4.

Gas hydrate systems are associated with cold seeps, where biogeochemical processes support locally abundant populations of specialised fauna that rely on in situ prokaryotic chemosynthetic primary production6. These chemosynthetic communities are typically dominated by tubeworms, bivalves and gastropods, in association with bacteria capable of metabolising methane, sulphide produced by anaerobic oxidation of methane and higher hydrocarbons coupled with sulphate reduction, and other hydrocarbons7. Seep communities influence local biodiversity, particularly in the relatively species-poor Arctic deep sea8.

Cold-seep communities in the Arctic have been described from the Håkon Mosby Mud Volcano on the western margin of the Barents Sea at 72.0 °N and 1250 m depth9, and from methane seeps associated with subsurface hydrates at Vestnesa Ridge on the continental slope of the Fram Strait at 79.1 °N and 1200 m depth10,11. Methane seepage also occurs on the Svyatogor Ridge, a sediment-covered transform fault on the flanks of the Knipovich Ridge at 79.4 °N and ~1900 m depth. Svyagotor Ridge hosts the deepest cold-seep community found in the Arctic so far12, although its fauna has not been characterised in detail at the time of our analyses13. In shallower waters, exposed hydrate mounds occur at the Storfjordrenna site on the western margin of the Barents Sea at a depth of 350–390 m14 and other methane seeps and mud volcanoes are present from 70 to 800 m depth in areas including the Barents Sea15,16, Beaufort Sea, and canyons on the Norwegian continental margin8. The fauna of these Arctic cold seeps includes siboglinid tubeworms, thyasirid clams, and rissoid snails8,17, and the seeps at shallower depths are often inhabited by abundant populations of species known from non-chemosynthetic habitats8.

Six active deep-sea (>200 m depth) hydrothermal vent fields are currently confirmed above latitude 70 °N. The Soria Moria (500–550 m depth) and Troll Wall (700–750 m depth) sites are 5 km apart at 71 °N on the southern end of Mohns Ridge and are occupied largely by taxa known from non-chemosynthetic habitats18. Vent fauna has not yet been characterised at the Aegir vent field at depth 2600 m and 72.3 °N on Mohns Ridge19, nor at the Jøtul vent field at depth 3020 m and 77.4 N on the Knipovich Ridge20. The fauna at Loki’s Castle at depth 2350 m and 73.6 °N on the northern end of Mohns Ridge21, and the Aurora Vent Field at depth 3888 m and 82.9 °N on the Gakkel Ridge22, includes some taxa not previously recorded at nearby seeps, such as melitid amphipods and cocculinid limpets22,23,24.

Previously, it has been concluded that vents and seeps share relatively few species, although similarities in faunal composition at the level of genera and families suggests evolutionary links such as common ancestry with slope fauna or dispersal from one chemosynthetic system to another25,26 However, this view was likely influenced by the lack of sampling and, more recently, where vents and seeps occur in close proximity and at similar depths higher faunal similarities have been observed27,28. Shared taxa seem especially likely to occur where similar habitats occur on vents and seeps, with sedimented vent sites appearing to share a particularly high number of taxa with those at seeps28. The proximity of lower bathyal vents and seeps in the Arctic raises the possibility of closer connectivity between these ecosystems north of latitude 73°N, depending on whether they share habitat characteristics.

The Molloy Ridge is a slow to ultraslow spreading centre in the Fram Strait, extending north for ~60 km from the Molloy Fracture Zone at ~79.1 °N to the Spitsbergen Fracture Zone at ~79.7 °N29. The seafloor depth of the ridge axis varies from ~5000 m at its southern end, rising to ~1500 m on an Oceanic Core Complex midway along the ridge, and descending to ~4000 m at the northern end29. The formation of the Molloy Ridge began after the opening of the Norwegian-Greenland Sea at ~56 Mya30, and most likely the seafloor spreading at the current Molloy Ridge started at ~20 Mya31.

At the northern end of the Molloy Ridge and in the Spitsbergen Fracture Zone, two large plumes of gas bubbles, described as gas flares, have been detected acoustically, rising ~1770 and ~3355 m above the seafloor, respectively, with the larger plume representing the tallest known worldwide32. From seismic reflection data, these plumes were hypothesised to consist of bubbles of oil-associated thermogenic gas32. The seafloor sources of the plumes, which occur at >3000 m depth, have not been characterised yet.

During the Ocean Census Arctic Deep – EXTREME24 expedition in May 2024, we investigated the seafloor source of the water column gas flare using shipboard instruments and a deep-diving Remotely Operated Vehicle (ROV). We discovered exposed hydrate mounds, named the Freya gas hydrate mounds, inhabited by chemosynthetic fauna at a depth of 3640 m (Fig. 1). These represent the deepest known hydrate deposits worldwide. Methane seepage and crude oil were directly observed and sampled with the ROV, revealing hydrocarbon seepage supporting chemosynthetic life ~1770 m deeper than any other Arctic cold seeps and at depths comparable with the nearby high-Arctic hydrothermal vents in the region20,21,22.

Fig. 1: Overview of the location of high-Arctic (>72 °N) cold seeps and hydrothermal vents.

a regional map of seeps (yellow) and vents (orange): yellow star = Freya gas hydrate mounds; orange star = Jøtul vent field 1 = Vestnesa Ridge seeps; 2 = Prins Karls Forland seeps; 3 = Storfjordrenna gas hydrate mounds; 4 = Bjørnøyrenna seeps; 5 = Leirdjupet Fault Complex seeps; 6 = Borealis Mud Volcano; 7 = Håkon Mosby Mud Volcano; 8 = Loki’s Castle; 9 = Aurora Vent Field. Seabed topography shown is from the Global Multi-Resolution Topography (GMRT) synthesis70. b map of seafloor features observed during ROV dives at the Freya gas hydrate mounds (79.6 °N, depth 3640 m). Detailed bathymetry from MAREANO/Norwegian Mapping Authority71.

Here we present the results from geochemical analysis of hydrates and oil collected from the Freya mounds, which clarifies the origins of the hydrocarbons being released from this site into the overlying ocean. Based on seafloor observations, we identify a sequence of morphological evolution of these hydrate features from inception to collapse. We also characterise the fauna colonising this deep methane seep and compare its taxonomic composition with chemosynthetic communities at other Arctic cold seeps and hydrothermal vents, including the first faunal samples collected from the nearby Jøtul vent field as part of this study. Our results provide insights into the geology and ecology of these habitats and their regional context for understanding patterns of deep-sea biodiversity in the Arctic.

Results

Discovery of the Freya gas hydrate mounds

Shipboard multibeam echosounder (MBES) data confirmed the presence of gas flares above the Molloy Ridge at 79.6930 °N 3.6617°E (Fig. 1a, b), which were originally detected by the Norwegian MAREANO programme33. This location corresponds with the gas flare designated ‘GFA’ in ref. 32. Multibeam backscatter detected two bubble plumes reaching a minimum depth of ~290 m, where the water temperature recorded in the CTD profile was 2.63 °C (Fig. 2).

Fig. 2: Water column characteristics at the Freya gas hydrate mounds.

a Depth profiles of temperature (red) and salinity (blue) measured by CTD, and b is topography as processed with Qimera and acoustic backscatter processed with FMMidwater using the shipboard multibeam echosounder (MBES) at the Freya gas hydrate mounds (79.6 °N, depth 3640 m). The dashed lines show the seafloor depth and the maximum depth of the top of the flare.

An ROV survey targeting the seafloor beneath the flares revealed the presence of three gas hydrate mounds, two pit-like collapse features and a few small ridges within an area of ~100 × 100 m at depth from 3570 to 3747 m (Fig. 1b). The ROV’s sonar and visual observations confirmed gas seepage adjacent to the mounds (Fig. 2 and Supplementary Movie 1), thus linking the water column observations to the seafloor hydrates. The temperature measured at 12 m altitude above the seafloor in a CTD profile was −0.63 °C (Fig. 2).

Mound morphology and hydrocarbon (oil and gas) composition

The mounds investigated are conical in shape, ~4–6 m in diameter and ~2–4 m high (Fig. 3a, b). They are covered by a thin layer of soft sediment, occasionally by carbonate slabs, and colonised by siboglinid and maldanid tubeworms that appear to stabilise the surface. Morphological variations of the mounds indicate a development sequence from sedimented domes with no exposed hydrate (Fig. 3a) to mounds with exposed hydrate in the summit (Fig. 3b) and more eroded or decomposed mounds resulting in arches and cave-like structures (Fig. 3c). We also noticed pit-like collapse features ~6–8 m in diameter (Fig. 3e) and several small ridges, rising just a few decimetres off the seafloor and spanning approximately 1–2 m.

Fig. 3: Freya gas hydrate mounds showing different morphologies.

The mounds, made of hydrates, are covered by sediments and frenulate tubeworms forming a ‘Sclerolinum forest’ (a) with occasionally amphipods and caridean red shrimp (b, d). Sometimes, around and at the top of the mounds, there are centimetric carbonate crusts (b). c Shows the position where the sample of gas hydrate for geochemical analyses was taken (yellow star; Supplementary Fig. 1) and the sediment sample used for faunal identification, that on board also revealed the presence of oil. c, d The influence of hydrate buoyancy on mound morphology that leads to structural fractures and alterations in the integrity of the mounds, ultimately resulting in the formation of collapse-like features (e). f Background seafloor.

The hydrate structure hosts visibly trapped gas bubbles. In some portions, the hydrate is yellow and white in white-balanced ROV video images1 (Fig. 3c). Aboard the research ship, as soon as we opened the blade corer, we observed the decomposition of the hydrate, which had already started during the ascent of the ROV, and persisted for several minutes while we were collecting the gas hydrate samples. We collected four hydrate subsamples containing methane (C1, ~66%), accompanied by smaller amounts of ethane (C2, ~8%), propane (C3, ~14%), isobutane (i-C4, ~3%), and normal butane (C4, ~2.3%) yielding an average C1/(C2 + C3) ratio of 3.0 (Fig. 4). The isotopic composition of the gas confirmed an oil-associated thermogenic origin resulting in methane with δ13C of −47‰ (n = 4; 1s = 0.8‰) and δD of −188.5‰ (1s = 1.7‰) and heavy δ13C composition of CO2 of 0.6‰ (1s = 0.2‰) (Fig. 4). The oil present in the hydrate samples shows a characteristic alkane distribution associated with gas condensate, with alkane chain lengths C13 (Supplementary Fig. 2). Steranes and diasteranes indicate a source rock deposited in a fresh/brackish lacustrine environment (tetracyclic polyprenoids-TPP and C26/C25 tricyclic terpanes ratios) with minor marine contribution (24-n-propylcholestane and 4-methylsteroids relatively sparse) (Supplementary Fig. 3). Moreover, the abundant oleanane and ursane compounds are consistent with high angiosperm inputs, with only traces of gymnosperm diterpanes, suggesting to a Miocene or younger source. One oil-impregnated sediment sample collected to study fauna displayed a more open marine organic signature with the presence of immature higher plants (Supplementary Figs. 2 and 3). The maturity proxies indicated a wet gas/pre-oil maturity window (Supplementary Fig. 4).

Fig. 4: Geochemistry of the gas emitted from Freya gas hydrate mounds.

Molecular and isotopic (δ13C, δD) composition of the gas contained in the gas hydrate. Sample data from Freya gas hydrate mounds are reported in yellow stars. For comparison, other high-latitudes cold seeps (location in Fig. 1) are reported: Borealis in ref. 15, Håkon Mosby Mud Volcano38, Prins Karl Forland37, Leirdjupet Fault Complex72, Vestnesa Ridge10, Storfjordrenna and Bjørnøyrenna. Genetic fields of hydrocarbons (CR-CO2 reduction, F—methyl-type fermentation, EMT—early mature thermogenic gas, OA—oil-associated thermogenic gas, LMT—late mature thermogenic gas) after73. a Isotopic composition of methane. b Plot of δ13C-CH4 versus the composition of light hydrocarbon components (C1/(C2 + C3) ratio). Grey arrows indicate the main processes affecting gases’ isotopic and molecular compositions. c Isotopic composition of CO2 (δ13C-CH4) versus methane δ13C-CH4. The combination of the three plots indicates that the methane in the Freya gas hydrate mounds has a thermogenic origin.

Biological community composition

More than 20 faunal morphospecies were observed at the methane hydrate site, as detailed in Table 1. The upper surfaces and periphery of the hydrate mounds are conspicuously colonised by dense aggregations of the sessile siboglinid polychaete Sclerolinum cf. contortum (Fig. 5a), termed the ‘Sclerolinum forest’, a refinement of the ‘tubeworm forest’ concept introduced by34, and more dispersed maldanid polychaetes (Fig. 5b) in soft sediments.

Fig. 5: Fauna of the Freya gas hydrate mounds.

a In situ hydrate mound fauna, including Sclerolinum forest. b Tube-dwelling maldanid polychaete. c Melitid amphipod. d Ampharetid polychaete. e Stauromedusa Lucernaria cf. bathyphila. f Rissoid and skeneid gastropods on a maldanid polychaete tube. g Thyasirid bivalve.

Invertebrates, including melitid amphipods (Fig. 5c), caridean shrimps, pycnogonids, and nemertean worms, were found in association with the Sclerolinum forest and maldanid polychaete tubes. Other polychaetes sampled from sediments at the mounds include an ampharetid species (Fig. 5d). Stauromedusae (Fig. 5e), identified as Lucernaria cf*. bathyphila*, were observed within the Sclerolinum forest on the methane hydrate mounds and among the maldanid polychaete tubes. Smaller specimens of the stauromedusa were also found in samples of the Sclerolinum forest collected by the ROV.

High densities of rissoid and skeneid microgastropods, each 2–3 mm in size, were noted in samples of the Sclerolinum forests and attached to maldanid tubes (Fig. 5f). The shell of the rissoid gastropod morphospecies was coated in orange precipitate, whereas the skeneid gastropod morphospecies featured a hyaline shell revealing light-coloured soft parts and white gonadal tissue at its apex. The same habitats also commonly hosted a buccinid gastropod, with juveniles smaller than 1 mm found in the Sclerolinum forests and larger specimens observed on Sclerolinum and maldanid tubes.

Dead thyasirid bivalves were observed on the sediment surface at the mounds, while live specimens (Fig. 5g) were retrieved using ROV push cores and scoops next to the Sclerolinum forest. Additionally, a smaller bivalve species with a maximum shell size of 1.5 mm and a black precipitate coating was found in the same area. Other taxa observed at the Freya mounds include the stalked sponge Caulophacus cf. arcticus and the fishes Lycodes cf. frigidus and Lycenchelys cf. platyrhina.

Comparison of Freya fauna with other Arctic seeps and vents

In addition to the discovery and investigation of the Freya gas hydrate mounds, our expedition described the fauna from the Jøtul hydrothermal vent field. This vent field is situated 266 km south of the Freya mounds at a depth of 3020 m on the Knipovich Ridge (Fig. 1a). The fauna at the Jøtul vents includes the siboglinid tubeworm Sclerolinum cf. contortum with melitid amphipods, caridean shrimp, skeneid and rissoid snails, which are also present in the fauna at the Freya mounds (Supplementary Data 1). At the family level, the fauna that we sampled at the Jøtul hydrothermal vents shows a 59% Sørensen Index similarity with the Freya mound fauna (Fig. 6).

Fig. 6: Family-level faunal similarity at high-Arctic (>72 °N) cold seeps and hydrothermal vents.

a Regional map of seeps (yellow) and vents (orange): yellow star = Freya gas hydrate mounds; orange star = Jøtul vent field, sites: 1 Vestnesa Ridge seeps; 2 = Prins Karls Forland seeps; 3 = Storfjordrenna gas hydrate mounds; 4 = Bjørnøyrenna seeps; 5 = Leirdjupet Fault Complex seeps; 6 = Borealis Mud Volcano; 7 = Håkon Mosby Mud Volcano; 8 = Loki’s Castle; 9 = Aurora Vent Field. Seabed topography shown is from the Global Multi-Resolution Topography (GMRT) synthesis70. b Dendrogram of faunal similarity between sites from hierarchical single-linkage agglomerative clustering based on Sørensen Index. Data analysed from this study and published literature for sites (76 families at 8 sites; for data sources, please see Supplementary Table 1; data for Storfjordrenna and Bjørnøyrenna are combined because separate inventories are unavailable in the literature). c Two-dimensional ordination of faunal similarity between sites from non-metric multidimensional scaling (nMDS) based on Sørensen Index. Bubble diameters represent site depths (starred yellow bubble = Freya gas hydrate mounds; starred orange bubble = Jøtul vent field).

In comparison with faunal inventories compiled for other seeps and vents from published literature (Supplementary Data 1), the fauna identified to family level at Freya and Jøtul is most similar to the fauna recorded at Loki’s Castle vent field (47% single-linkage Sørensen Index similarity) and Vestnesa Ridge seeps (46% single-linkage Sørensen Index similarity), and least similar to the fauna at the Prins Karls Forland (PKF) seeps (23% single-linkage Sørensen Index similarity; Fig. 6). Several widespread taxa contribute to faunal similarity between sites, including habitat-engineering tubeworms (Siboglinidae: recorded at all sites except the Aurora Vent Field; and Maldanidae: recorded at five out of eight sites) and rissoid snails (recorded at six out of eight sites; Supplementary Data 1).

The proximity of sites, calculated as great-circle distances from latitude and longitude values, does not correlate significantly with Sørensen Index similarities (Spearman rank correlation: r**s = −0.24, p = 0.21, 26 d.f.). However, differences in depth between sites show a significant negative correlation with faunal similarity values (Spearman rank correlation: r**s = −0.47, p = 0.012, 26 d.f.), indicating that depth may be a factor influencing faunal composition.

The number of families recorded at sites varies from 5 at the Aurora Vent Field to 46 at the Håkon Mosby Mud Volcano (Supplementary Data 1), which may result from greater cumulative sampling effort at longer-studied sites. But there is no significant negative correlation between faunal similarity and differences in family richness between sites (Spearman rank correlation: r**s = −0.089, p = 0.65, 26 d.f.), indicating that variation in the number of families recorded at sites does not determine their overall pattern of faunal similarity.

Discussion

The direct evidence of hydrate outcrops at unprecedented depths producing gas flares that rise for more than 3000 m to within 300 m of the ocean surface, confirms the active nature of these features and their potential contribution to carbon cycling in the water column. The presence of another gas flare nearby in the Spitsbergen Transform Fault35 also indicates a likelihood of further methane seep communities at >3000 m depth in the region, possibly associated with gas hydrates. Studying these Arctic ultra-deep gas hydrate systems is crucial to enhance our understanding of the deep carbon cycling and ecosystems influenced by natural hydrocarbon emissions, which is key to fill gaps in Arctic deep-sea biogeography.

Composition and dynamics of the Freya gas hydrate mounds

The Freya gas hydrate mounds contain thermogenic gas primarily composed of methane (C1) and a smaller amount of heavier hydrocarbons (C2–C5). This thermogenic gas is produced from the degradation of organic matter under high heat and pressure conditions and migrates upward through faults in the area, as indicated by previous studies32, acting as conduits from deeper geological strata to shallower sediment layers where gas hydrates form. Geochemical analysis indicates that the oil, and possibly the associated gas, originated from the breakdown of material derived from angiosperms, flowering plants that were abundant in the Arctic during the Miocene epoch36. We draw a first-order correlation with the potential source rock identified for nearby shallow oil seeps of Prins Karls Forland (Fig. 4)37, based on similarities in age and depositional paleo-environments. For Prins Karls Forland, Arctic blooms of the freshwater Azolla fern at 56 Mya25 and later depositions of organic-rich sediments during the Miocene have been suggested26,27. The thermogenic gas contained in the Freya hydrates is distinguished from other known seeps in the Barents Sea that show a microbial-dominated origin, such as Håkon Mosby Mud Volcano38, or mixed origin, such as Vestnesa Ridge10.

Moreover, the observed yellow colour of the hydrates exposed at the seafloor is ascribed to oil-sustaining and/or encrusting bacteria, as observed in the Gulf of Mexico39. We incorporated the gas composition of Freya hydrates into a thermodynamic model of the hydrate stability zone (see Method section for full parameters). Our model results indicate a subsurface stability zone approximately 248 m thick, suggesting significant potential for gas hydrate accumulation in the sediment. This estimate aligns with previous predictions of a stability zone up to 250 m thick on the flanks of Molloy Ridge40. Despite significant progress in understanding the distribution and concentration of gas hydrates41,42, a major challenge remains in evaluating gas hydrates as an energy resource and their role in global climate change, resulting from the uncertainty surrounding the size of the resource. In addition, since the 1980s, the Greenland Sea has experienced a noticeable warming, with temperatures rising from approximately −1.30 to −0.85 °C by the 2020 s43. In the Fram Strait, influenced by both Greenland Sea Deep Water and Eurasian Basin Deep Water, temperatures have fluctuated between −1.20 and −0.95 °C in the 1980s, warming to around −0.85 °C by the 2020s43. While we cannot completely rule out the impact of global warming on the Molloy gas hydrate, the complexity of these changes suggests multiple influencing factors and important aspects are associated with the methane’s role in supporting local ecosystems

Gas hydrate dissociation contributes to methane seepage into the deep ocean, potentially reaching the upper mesopelagic zone. Where gas hydrates are stable, gas bubbles released are typically coated in a hydrate skin that inhibits their dissolution. Although bubbles lose this coating and dissolve rapidly as they ascend beyond the gas hydrate stability zone44, the presence of oil can lead to the formation of oil-coated bubbles, which were shown to travel through a 3400 m high water column in the Gulf of Mexico45. In ROV video observations (Supplementary Movie 1), we noted numerous trains of bubbles ascending through the water column from localised areas on the seafloor that had visible patches of clear hydrate directly beneath them. Some of these bubbles exhibited unusual flat shapes while rising, which we attribute to the formation of oil and gas hydrate coatings46.

Water column temperature above the Freya mounds increased from 0 °C at ~1000 m to 2.60 °C at ~300 m depth (Fig. 2), thereby crossing the boundary of the gas hydrate stability zone (~297 m)47. The minimum depth at which we observed a bubble plume in multibeam backscatter signals was ~290 m, therefore consistent with gas bubbles losing their hydrate coating and dissolving rapidly as they rise above hydrate stability conditions. The rise height previously reported for a gas flare at this site was ~1770 m (the ‘GFA’ flare in ref. 32). Our MBES shows a truncated bubble plume at ~3350 m above the seafloor, much higher than this previous measurement and suggesting that the plume reaches even higher levels. It has been previously suggested that methane is generated along the Spitsbergen Transform Fault immediately north of the slow/ultraslow spreading Molloy Ridge and released through boundary faults of the deep sediment-filled Spitsbergen Transform Fault depression32.

The visual identification of gas hydrate mounds and ridges in different stages of evolution at Freya (from sedimented domes to mounds and arches of exposed hydrate and, finally, pit-like collapse features; Fig. 3) provides a snapshot of distinct features observed simultaneously. While these features are interpreted as representing different stages of evolution, this interpretation is based on their morphology and spatial distribution rather than direct temporal observations. This suggests continual processes of formation and dissociation, consistent with hydrates being dynamic and metastable systems48.

Hydrate dissociation releases gas and freshwater into the surrounding environment as its crystal lattice breaks down. This may physically disturb fauna that have colonised hydrate mounds, particularly removing substratum occupied by sessile taxa. Availability of methane may also be reduced locally once most or all of the hydrates have dissociated from a structure. The pit-like collapse features that we suggest form where the sedimented mounds have collapsed as a result of hydrate dissociation have patches of depauperate fauna dominated by taxa such as Stauromedusae and motile species, in contrast to the Sclerolinum forest and maldanid tubeworms occupying the mounds of intact hydrates. Hydrate mounds may, therefore, represent a successional deep-sea habitat, with faunal composition changing as a result of disturbance from hydrate dissociation and subsequent waning in methane supply at the individual mound scale.

Diversity and biogeography of fauna at the Freya gas hydrate mounds

Siboglinid tubeworms are one of the dominant taxa at the Freya mounds and may function as ecosystem engineers, with their tubes providing three-dimensional structure colonised by filamentous bacteria and grazers such as gastropods. Siboglinids have a widespread distribution at other Arctic seep and vent sites (Supplementary Data 1): the frenulate Oligobrachia occurs at shallower cold seeps in the region17, and the monoliferan Sclerolinum is the biomass dominant at the Håkon Mosby Mud Volcano7, also occurs at the Loki’s Castle[23](https://www.nature.com/articles/s41467-025-67165-x#ref-CR23 “Eilertsen, M. H. et al. Diversity, habitat endemicity and trophic ecology of the fauna of Loki’s Castle vent field on the Arctic Mid-Ocean