Quantitative mineralogical compositions were determined by powder X-ray diffraction and are reported in weight-% (wt%; Table 2). Powders were produced by pulverizing broken chunks (i.e., bulk-rock) and micro-drilling of cut slabs using a hand-held micro-drill. The powders were analysed with a Bruker D8 Advance diffractometer using Cu K α radiation at a 2ϴ scanning angle of 3°–75° (step size of 0.02°, 1 s per step). Minerals were identified by automatic and manual peak search using the Bruker DIFFRAC EVA3.1 software; quantification was performed applying Rietveld refinement with the TOPAS 5 software. Repeated internal standard measurements yielded a detection limit and uncertainty of ±2 wt%.

Thin sections (6.5 x 5 cm) were prepared from ca. 1.5 cm thick cut slabs that were embedded in epoxy resin. The thin sections were examined using standard petrographic microscopy (Figure 2). Samples for stable carbon and oxygen isotopes (δ¹³C and δ¹⁸O) were drilled from the surface of cut slabs with a hand-held micro-drill. Approximately 200 µg of the obtained powders were reacted with anhydrous phosphoric acid in a GasBench II preparation line; released CO₂ gas was analysed with a ThermoScientific Delta V Advantage isotope ratio mass spectrometer. The δ¹³C and δ¹⁸O values are reported in per mill (‰) relative to the Vienna–Pee Dee Belemnite standard (V–PDB). The data were calibrated against standard materials and the uncertainties for δ¹³C and δ¹⁸O were not higher than ± 0.2‰.

From 17 seep carbonates the U–Th ages of 55 fibrous cement samples were determined. Sample powders between 0.2 and 9.3 mg were obtained from polished cut slabs using a hand-held micro-drill. After powder digestion in 8 M HNO₃ U and Th were separated, concentrated, and analysed as described previously (Crémière et al., 2016). In brief, after digestion, centrifuging, and detritus dissolution, U and Th were pre-concentrated by iron co-precipitation and separated via column ion chromatography (Edwards et al., 1987). Isotope ratios were measured with a Neptune Plus multicollector inductively coupled plasma mass spectrometer. Argon and nitrogen were used as carrying gases to minimize oxide formation. Uranium–Th ages were calculated using an in-house spreadsheet, applying the ²³⁰Th and ²³⁴U decay constants of Cheng et al. (2013). Initial ²³⁰Th correction was based on mean continental crust values. Sample ages are given in thousand years before present (ka BP; BP = before 1950). Samples with ²³⁰Th/²³²Th activity ratios ≤2 were rejected for data interpretation. The initial U isotope signature of seep carbonate tends to reflect seawater values or slight enrichment in ²³⁴U relative to seawater if precipitation occurs under more restrictive conditions from pore fluids (Teichert et al., 2003). Therefore, analyses where (²³⁴U/²³⁸U)_i sample <1.14 or (²³⁴U/²³⁸U)_i sample >1.18 were assumed to represent an open-system behaviour with respect to U and/or Th, and were rejected for interpretation, too (see Supplementary Figure S1).

Aragonite is the main carbonate mineral (22–90 wt%), followed by moderate to low contents in magnesium-calcite (Mg-calcite; 3 to 63 wt%) and calcite (up to 22 wt%; Table 2). Quartz, feldspar, and clay minerals are the major non-carbonate components, ranging between 3 and 17 wt%. Micro-drilled fibrous aragonite in samples HH1029 and HH1077 is completely devoid of non-carbonate components (Table 2). Micrite in sample HH1029 has more Mg-calcite (63 wt%) than aragonite (22 wt%), whereas micrite in HH1077 is mainly aragonite (77 wt%) with only trace Mg-calcite content (below 2 wt%).

The carbonates comprise porous intraformational breccias of micrite-cemented sediment clasts, cemented by fibrous aragonite crusts (Figure 2; Supplementary Figures S1–S6). Relatively earlier formed micrite cemented clasts and subsequently formed fibrous aragonite cement comprise the volumetric dominant microfacies. The micrite cements hemipelagic sediment rich in silt-sized quartz and feldspar grains (Figures 2A,D). Fibrous aragonite forms mm-to cm-thick cement crusts, partly surrounding clasts and filling the pore space between micrite-cemented sediment and bioclasts (Figures 2B, C; Supplementary Figures S1–S6). Samples from Storfjordrenna show abundant tube-worm fossils in both, micrite and fibrous cement (Figures 2A, B), relatively few articulated bivalve shells, and centimetre-sized grains of ice-rafted debris (IRD; Supplementary Figures S1–S3). Likewise, the Storbanken High samples display some bivalve shells and IRD grains (Supplementary Figures S3D, E). Samples from the Loppa High contain many articulated bivalves with abundant inter- and intraparticle fibrous cement (Supplementary Figure S5). Macroscopically, fibrous cement in the Loppa High samples can be distinguished into whitish and grey aragonite. Whitish aragonite appears as incoherent layer of inclusion-rich, cryptocrystalline cement between micrite and fibrous aragonite (Figure 2E; Supplementary Figures S6). Under UV light cryptocrystalline aragonite exhibits intense fluorescence, whereas subsequent clear fibrous aragonite is non-fluorescent. Likewise, dark inclusion-rich clotted aragonite and brownish botryoidal aragonite display strong fluorescence (Figure 2F). Occasionally, pockets of a few square centimetre size containing cemented IRD are intercalated within cemented clasts and bivalves (Supplementary Figures S6).

Carbonate δ¹³C and δ¹⁸O values were analysed for micrite (n = 47) and fibrous cement (n = 47; Figure 3; Supplementary Figures S1). Overall, δ¹³C values range from −43.3 to −18.4‰ while the δ¹⁸O values range from 3.1‰ to 5.6‰. Micrite yielded slightly lower δ¹³C (median ±SD: −32.2‰ ± 4.8‰) and slightly higher δ¹⁸O (4.8‰ ± 0.5‰) values than fibrous cement (δ¹³C = −31.2‰ ± 5.1‰; δ¹⁸O = 4.6‰ ± 0.5‰). Samples from Storbanken High exhibit relatively lower δ¹³C values (−37.8‰ ± 3.7‰; n = 19) than the Loppa High (−31.9‰ ± 3.6‰; n = 51) and Storfjordrenna (−27.0‰ ± 2.5‰; n= 30) samples. Storfjordrenna samples show relatively higher δ¹⁸O values (5.0‰ ± 0.3‰) than the samples from Storbanken (4.9‰ ± 0.2‰) and Loppa High (4.3‰ ± 0.5‰).

From the 55 micro-drilled U–Th samples, four have been discarded for interpretation due to relatively low data quality and resulting high uncertainty, and three are coral skeleton material (Figure 4; Supplementary Figure S1–S6; Supplementary Figure S1). Samples C1 through C7, HH1029, and HH1077, were collected from the gas hydrate pingo mound 3 (GHP 3; Serov et al., 2017) at Storfjordrenna and yielded exclusively mid to late Holocene ages, ranging from 6.3 ± 2.3 (ka ± 2σ) to 1.2 ± 0.1 ka. The two samples from the Storbanken High area yielded one late Pleistocene 13.5 ± 2.4 ka (one subsample from C11) and three mid-Holocene ages from 6.6 ± 0.3 to 5.4 ± 0.1 ka (three subsamples, C14). Samples PR1810 and PL954 from the Loppa High area exhibit late Pleistocene and early to mid Holocene ages, ranging from 13.4 ± 0.7 to 4.1 ± 1.3 ka. The three coral samples exhibit Holocene ages from 10.4 ± 1.3 to 4.2 ± 1.3 ka.

The carbonate δ¹³C values are as low as previously reported values of seep carbonates sampled from the south-west and north-west Barents Sea (Crémière et al., 2016; Crémière et al., 2018; Yao et al., 2021; Argentino et al., 2022), and indicate a predominantly hydrocarbon-derived carbon source. Seepage gas samples from Storfjordrenna and Loppa High show a thermogenic composition and yielded δ¹³C_methane values of −47‰ and−48% V–PDB, respectively (Crémière et al., 2016; Serov et al., 2017). Weniger et al. (2019) extracted gas from surface sediments at Storfjordrenna and report a thermogenic composition with slightly higher average δ¹³C_methane of −41.3‰ ± 2.7‰ V–PDB (n=4). In the same study, Weniger et al. (2019) show a thermogenic gas composition with average δ¹³C_methane of −50.8‰ ± 5.2‰ V–PDB (n=18) for surface sediment bound gas from the Olga Basin, which borders along the southern flank of the Storbanken High. To the best of the authors’ knowledge there are no published methane δ¹³C values for seepage or bound gas from the Storbanken High seeps. Given that the Storbanken High and the Olga Basin share the same reservoir rocks (Lundschien et al., 2023; Serov et al., 2023), a thermogenic origin for the gas seeping at Storbanken High is assumed.

The seep carbonates are relatively ¹³C-enriched compared to the methane. This is interpreted as the result of mixing ¹³C-depleted dissolved inorganic carbon (DIC) with ¹³C-rich DIC during carbonate precipitation, yielding higher carbonate δ¹³C values relative to the parent methane (Formolo et al., 2004; Peckmann and Thiel, 2004; Roberts et al., 2010; Himmler et al., 2015). Seep carbonate δ¹³C values represent a DIC mixture of seawater (δ¹³C ≈ 0‰), DIC produced during organiclastic sulfate reduction (≈ −25‰), and DIC produced by AOM/or and oxidation of higher hydrocarbons (δ¹³C below −20‰; e.g., Irwin, et al., 1977; Formolo et al., 2004; Sassen et al., 2004; Smrzka et al., 2019). Whereas the highest δ¹³C_carbonate value of −18‰ (Figure 3; Supplementary Figure S1) can be explained by a DIC mixture of seawater and organiclastic sulfate-reduction, the rest of the δ¹³C_carbonate values between −43‰ and −24‰ point to a DIC mixture produced by the anaerobic oxidation of the thermogenic methane and/or non-methane hydrocarbons (Formolo et al., 2004; Sassen et al., 2004; Smrzka et al., 2019). The relatively lower values indicate a higher contribution of AOM-derived DIC, whereas the relatively higher values indicate significant contribution DIC resulting from the anaerobic oxidation non-methane hydrocarbons (Formolo et al., 2004).

Aragonite is the dominant carbonate phase, except for the magnesium-calcite rich micrite in sample HH1029 (Table 2). It is not unusual for seep carbonates to mainly comprise aragonite and partly include Mg-calcite (e.g., Greinert et al., 2001; Crémière et al., 2018; Lu et al., 2021). At hydrocarbon seeps aragonite formation is likely favoured over calcite by relatively high sulfate concentrations (Burton, 1993). Bioturbation of the sediment by bivalves and bubble ebullition during episodes of strong gas efflux likely eased entrainment of sulfate-rich seawater into the shallow subsurface, supporting sulfate-driven AOM near the sediment–water interface (e.g., Borowski et al., 1999; Aloisi et al., 2002; Luff et al., 2004). In contrast, during episodes of weak seepage with low gas flux, AOM was likely mediated deeper in the sediment at relatively lower pore water sulfate concentrations, and Mg-calcite precipitation was favoured (e.g., Greinert et al., 2001). Increased magnesium incorporation into the calcite lattice is plausible under low flux conditions, when abundant AOM-produced hydrogen sulfide (HS⁻) supports de-hydration of Mg²⁺ ions, thus increasing the pore fluid Mg²⁺/Ca²⁺ ratio (Zhang et al., 2012; Lu et al., 2021). At the same time, the increased alkalinity resulting from AOM-produced bicarbonate (HCO₃ ⁻) also supports carbonate precipitation.

Overall, the aragonite-dominated mineralogy is congruent with carbonate precipitation near the sediment–water interface. This is also supported by the macrofossil content, including the chemosynthesis-based siboglinid tube worms Oligobrachia sp. (Sen et al., 2018; Åström et al., 2018), unidentified articulated bivalves, and a presumably cool water coral (Figure 2, Supplementary Figures S1–S6). Moreover, molecular fossils including ¹³C-depleted lipid biomarkers diagnostic of AOM-mediating methanotrophic archaea and sulfate-reducing bacteria have been found in Storfjordrenna samples HH1029 and HH1077, corroborating that microbial-mediated sulfate-driven AOM induced carbonate precipitation (Yao et al., 2021). Although no lipid biomarker data are available for the Storbanken High and Loppa High samples, they contain abundant microbial textures, including clotted micrite and cryptocrystalline aragonite (Figures 2C–F; Riding, 2000; Zwicker et al., 2018). Both textures exhibit strong fluorescence under UV radiation, indicative of high organic matter content resulting from in situ organomineralization (Neuweiler et al., 2000). Given that increased carbonate alkalinity at methane seeps supports the lithification of microbial biofilms, it is feasible that the fluorescent textures observed in the Storbanken High and Loppa High samples represent mineralized AOM hotspots (Himmler et al., 2018; Zwicker et al., 2018). Remarkably, the strong fluorescent textures in samples PR1810-002 and PL945-C2 (Figures 2E,F) correspond macroscopically to whitish aragonite, similar to whitish and cryptorcrystalline aragonite with high AOM-biomarker contents described in modern and ancient seep carbonates (Leefmann et al., 2008; Hagemann et al., 2013).

Regarding the reconstructed ice margins, all U–Th dates are consistent with carbonate formation after local ice retreat and protracted methane flux (e.g., Crémière et al., 2016; Serov et al., 2023). The oldest samples C11 (Storbanken High; SH) and PR1810-001C (Loppa High; LH) yielded similar ages of 13.5 (SH) and 13.4 (LH) ka. The ages indicate relatively high gas flux and associated seep carbonate formation at these two distant sites during the Bølling–Allerød interstadial. At this time, influx of relatively warmer Atlantic seawater promoted rapid disintegration of the grounded Barents Sea ice sheet, leaving large areas in the west and south-west Barents Sea ice free (e.g., Rasmussen and Thomsen, 2021; Sejrup et al., 2022). The influx of warm water likely accelerated the dissociation of shallow subsurface gas hydrates, resulting in increased methane emissions at the seafloor (Crémière et al., 2016; Andreassen, et al., 2017). This is congruent with the SH carbonate δ¹⁸O values, indicating a contribution of hydrate-derived ¹⁸O-enriched water during carbonate precipitation (Figure 3).

Except for sample PL954–C2 (≈4.7 and 4.4 ka), the rest of the non-coral LH samples have late Pleistocene and early Holocene U–Th ages between ≈12.6 and 7.5 ka. Overall, the LH dates are in agreement with previously reported seep carbonate U–Th dates from the south-west Barents Sea and northern Norwegian Sea, together indicating episodic post–LGM hydrocarbon gas release and seep carbonate formation between 17.5 and 1.6 ka (Crémière et al., 2016; Sauer et al., 2017; Crémière et al., 2018; Argentino et al., 2022). Sauer et al. (2017) suggested that in northern Scandinavia, seismic activities and resulting fault reactivation and changes in reservoirs pore pressure could have triggered gas seepage around ≈11 and 4 ka. Remarkably, the majority of the new LH U–Th ages between 12 and 10 ka plus the two ages of sample PL954–C2 (≈4.7 and 4.4 ka), are in agreement with this scenario.

The carbonate ages reveal lag times between local ice retreat and distinct episodes of enhanced gas efflux. It needs to be stressed that the reported ages were exclusively derived from relatively late-stage fibrous aragonite cement and do not constrain the initial seepage stage (see also Crémière et al., 2016; Crémière et al., 2018). In particular at Storfjordrenna, seep carbonate formation occurred ≈13 ka after the grounded ice has had retreated after ≈19 ka (Rasmussen and Thomsen, 2021; Sejrup et al., 2022). The long lag time at Storfjordrenna is surprising, considering that abundant gas hydrates (and probably free gas) had accumulated in the subsurface during the LGM in an approximately 200 m thick GHSZ (Serov et al., 2017). After glacial retreat, the GHSZ totally diminished by ≈13 ka and must have released significant gas amounts from the seabed due to gas hydrate dissociation (Serov et al., 2017). Yet, the carbonate ages do not indicate carbonate formation before 6.3 ka. It could be that the U–Th ages of seafloor-sampled seep carbonates do not reflect the entire post–LGM seepage history, and large ice bergs from the disintegrating ice sheet might have eroded any older seep carbonates from the seafloor (Rasmussen and Thomsen, 2021). El bani Altuna et al. (2021) showed that the average bottom water temperature at Storfjordrenna increased from 2.5°C ± 1°C to 4.3°C ± 1°C between ≈13 and 5 ka, followed by cooling down to down 2.8°C ± 1°C after 4 ka. In addition, their gas hydrate model suggests that a fluctuating GHSZ existed between 13 and 9 ka (i.e., variable thickness of 0 to ca. 100 m below seafloor), but was absent between 9 and 5 ka, resulting in liberation of gas and hydrate-derived water. Remarkably, the gas hydrate modeling also shows a stable GHSZ since 5 ka, which would sequester most of the upward migrating gas. Yet, our U–Th ages show constant seep carbonate formation near the sediment–water interface since 6.3 ka. Regardless of the existence of the GHSZ, the carbonate dates support the notion that the underlying hydrocarbon reservoir actively injects thermogenic methane into the shallow subsurface sediments at this site, probably since the mid to late Holocene until modern times, as put forward by Waage et al. (2019). Given constant gas supply from below and the resulting high methane concentrations, free methane likely bypasses the GHSZ through cracks and faults, eventually escaping at the seafloor.

Seep carbonate δ¹⁸O values depend on the mineralogy, the ambient temperature and the δ¹⁸O of the carbonate-precipitating fluid, and serve as proxy for possible gas hydrate dissociation during carbonate formation (e.g., Aloisi et al., 2000; Greinert et al., 2001; Crémière et al., 2016; Crémière et al., 2018). Assuming seep carbonate formation under post-glacial conditions, i.e., bottom water temperature 1°C and seawater δ¹⁸O = 1‰ V–SMOW (Duplessy et al., 2002), carbonate δ¹⁸O values of 4.7‰ for aragonite and 5.6‰ for magnesium–calcite would be expected (Kim et al., 2007; Crémière et al., 2016). Samples from Storfjordrenna (median δ¹⁸O = 5.0) and Storbanken High (median δ¹⁸O = 4.9) show carbonate δ¹⁸O values slightly higher than 4.7‰, while the Loppa High samples are mostly (but not exclusively) below 4.7‰ (median δ¹⁸O = 4.3; Figure 3). Values above 4.7‰ would indicate either precipitation from fluids below 1°C temperature or relatively enriched in ¹⁸O, while lower values point to precipitation at slightly higher temperatures or fluids depleted in ¹⁸O. Given that the reconstructed bottom water temperature was above 2°C in the Storfjordrenna since the early Holocene (El bani Altuna et al., 2021; Rasmussen and Thomsen, 2021), the seep carbonates most likely precipitated from ¹⁸O-enriched fluids. It is feasible that the ¹⁸O-rich fluids originated from extensive post–LGM gas hydrate dissociation, since hydrate incorporates preferentially ¹⁸O (Davidson et al., 1983; Serov et al., 2023). Considering that warm water influx has triggered abrupt gas hydrate dissociation to form large craters on the seafloor between 15 and 12 ka ago (Andreassen et al., 2017), a similar formation scenario for the ≈6 ka old Storbanken High sample C14 is suggested. Influx of warmer bottom water around 6 ka, as shown by El bani Altuna et al. (2021), likely accelerated gas hydrate dissociation in the shallow subsurface, which would explain the relatively ¹⁸O-erniched carbonates. For the Loppa High samples, most of the carbonate δ¹⁸O values can be explained by precipitation in equilibrium with ambient post–LGM bottom water temperatures. However, as pointed out by Crémière et al. (2018), a contribution of ¹⁸O-rich fluids from the relatively deeper reservoir and/or from gas hydrate dissociation can not be ruled out, considering the complex tectono–sedimentary history and possible fluid dilution effects during carbonate precipitation.