The study area (Figure 1) is part of the Tuktoyaktuk Coastlands physiographic region, which is characterized by thick occurrences of unconsolidated sediments (Rampton, 1988). The landscape is covered by a low arctic tundra vegetation with dwarf birch, willow, grasses and moss. The climate is characterized by long cold winters and short cool summers, with low precipitation. The Tuktoyaktuk area lies within the continuous permafrost zone, with permafrost thicknesses of ∼400 m (Hu et al., 2013). Ground ice, including ice lenses, ice wedges, and extensive bodies of massive ice, is common (Mackay and Dallimore, 1992). The only community in the area is the Hamlet of Tuktoyaktuk, which was originally an Inuvialuit seasonal harvesting site and was established as a harbor by the Hudson’s Bay Company in 1934. The Hamlet presently has a population of ∼900 people (Statistics Canada, 2017).

The coastal environment of the Tuktoyaktuk area is experiencing relatively rapid coastal retreat. The seaward side of Tuktoyaktuk Island, located approximately 1 km from the Hamlet, has eroded at an average rate of ∼2.5 m/a since 1950 (Figure 1; Hynes et al., 2014). The majority of the erosion occurs during wind-induced storm surge events. Although storm surges can occur at any time of the year, the largest erosional impact is during the late summer and early fall when the Beaufort Sea coast is ice free, the dominant storm winds are from the northwest and fetch length is at its maximum (Manson and Solomon, 2007). Increases in mean annual air temperatures from approximately −11°C to −8.5°C have been recorded over the past 5 decades (Figure 2; GISTEMP, 2019). A linear fit to this trend gives an increase of 0.52°C/decade, about twice the global average for this same period (Hansen et al., 2010). Changing permafrost conditions are affecting building foundations and other infrastructure, storm surges are causing flooding, and accelerated coastal retreat is forcing the relocation of buildings (Gruben, 2018; Zingel, 2019).

The permafrost core utilized for our study was collected from a drilling program, conducted by the Geological Survey of Canada in March 2018, to document the properties of the eroding permafrost as a basis to assess the consequences of ongoing warming and erosion of Tuktoyaktuk Island (Figure 1). Four drill holes were completed using a CRREL core barrel (10 cm diameter, Midnight Sun Drilling, Whitehorse, CA, Figure 3A) for frozen sediments and a solid stem auger for unfrozen sediments. The CRREL core barrel was specifically designed to collect permafrost samples with minimal contamination or thermal disturbance to the core. The core samples utilized for this project (Figures 3B,C) were retrieved from a terrestrial borehole positioned at the surface of the Island, ∼6 m above sea level ∼30 m from the eroding coastal bluff (Figure 4A). In August 2018, a thermistor cable was installed in a borehole to measure ground temperatures. The thermistor sensor depths ranged from just below ground surface to a maximum depth of 14 m with a data logger (RBR XR-420, 8 channel) set to record temperatures hourly. In July 2019, the temperature data was recovered and presented here.

Samples used in this study were collected from 10 cm length whole round core sections selected from 10 depth intervals representing a range of in situ ground temperatures (0.6, 0.9, 1.5, 1.7, 2.2, 2.8, 3.4, 3.9, 7.8, and 11.6 m). The drilling was undertaken in March with air temperatures <−25°C and samples were stored in plastic bags and maintained at in situ air temperature until transportation to the Aurora Research Institute in Inuvik, N.W.T. Care was taken to minimize contamination by handling sediment sections with nitrile gloves. Cores were then transferred frozen to the Chesapeake Biological Laboratory (Solomons, MD, USA), and stored at −20°C until further analysis. Given the care to prevent contamination or warming during coring, storage and transport we assume that measured physical and geochemical parameters reasonably reflect in situ conditions of the subsurface permafrost.

Laboratory incubations lasting ∼6 weeks were carried out on sub-samples from four depths. For each depth, the incubations were conducted under anaerobic conditions at different temperatures, and compared to killed (basified) controls to determine if any abiotic processes contributed to CH₄ release. The depths were chosen to document the range of sediment, permafrost and chemical characteristics down core: 0.6 m corresponded to the only sample at the base of the active layer, 1.5 m corresponded to a peak in CH₄ concentrations, 2.8 m corresponded to a peak in DOC concentrations, and 11.6 m was the deepest sample collected which was near the depth of zero annual temperature amplitude.

For each incubation, approximately 6–9 g of freshly drilled sediment was placed into six pre-combusted (550°C) 10 ml glass serum vials and capped with black butyl rubber septa and aluminum cap. Vials were then flushed with nitrogen (UHP Airgas) to obtain an anaerobic headspace and placed back at −20°C. To obtain replication within treatments, to three of these vials, 3 ml of a 1M KOH solution was added to arrest microbial activity and referred to as the “killed control” (in triplicate). Since flushing the vials with nitrogen may have removed any in situ CH₄, the killed controls act as a way to follow the diffusion of remaining CH₄ to the headspace, which is an abiotic process. The remaining three vials were not amended with base and referred to as the “experimental treatment” to examine the gross potential of microbial production of CH₄ and/or CO₂ during the incubation (also done in triplicate). As the vials were devoid of oxygen, there was no electron acceptor for aerobic microbial CH₄ oxidation, so the CH₄ production rates should be considered a high-end for gross production potential, and not net rates expected under real-world conditions where oxygen would likely be present.

All vials were then placed in temperature-controlled chambers maintained at specific temperatures within 1°C. For 0.6 m, incubations were carried out at −20 and 15°C, to replicate natural range in temperature conditions in the active layer at this site (Figure 4). For 1.5 and 11.6 m, incubations were carried out at −20°C (coldest surface temperatures sediments experience), −5°C (average air temperature), and +5°C (to replicate modest warming comparable to other studies). For 2.8 m, incubations were carried out at the same frozen temperatures (−20 and -5°C); for the warm treatment, 15°C was chosen for comparison to other studies incubating active layer sediments (Treat et al., 2015). All treatment and temperature pairings were carried out in triplicate. During the incubation, vials were stored in the dark and upside down in a small tub of water (or ice) at their respective temperatures to prevent gases from escaping the septa (although independent experiments show this process would be minimal during the incubations (Magen et al., 2014)).

Time points were collected as discrete headspace samples, approximately once a week, to determine the CH₄ and CO₂ concentrations. To do this, 6 ml of nitrogen (UHP Airgas) was injected into the vial to displace 6 ml of the headspace, and this aliquot was injected onto the gas chromatograph described above. Vials never experienced an overpressure of gas. Sample areas were compared to certified standards (Airgas, 30 ppm CH₄, 2000 ppm CO₂) and converted to ppmv in the headspace. Analytical precision was 3%. CH₄ concentrations were averaged between the three triplicates and reported as an average +/− standard deviation. Concentrations were then converted to cumulative production for CH₄ and CO₂ and reported as µmol CH₄ or CO₂ carbon per gram dry weight (µmol CH4−or CO₂−C/gdw) and plotted against incubation time. To calculate a rate of CH₄ generation on a per carbon basis, a line was fitted to the linear portion of the cumulative production time-series and converted to a per carbon basis by multiplying by the %TOC measured for each depth. It should be noted that while CO₂ concentrations were measured in the killed controls, the values were expected to be 0 since base drives gaseous CO₂ into the water phase. The results are still shown as a way to verify our basic assumptions.

When the sediment plugs are first placed in the vials and flushed with N₂ gas, it is possible that CH₄ dissolved in the sediment interstial spaces will diffuse out and be measured as an increase in CH₄ concentrations over time due to the concentration gradient present at the beginning of the incubations. This theoretical diffusion can be estimated using a one-dimension model to calculate how much CH₄ could diffuse from the sediment plug in the incubation vials to the headspace over the time of the incubations (Lapham et al., 2014). We used a diffusion coefficient of 8.6 × 10⁻⁶ cm²/sec for CH₄, which is corrected for temperature, salinity, and pressure in the vials (Riley and Skirrow, 1975). We estimated salinity based on the Cl⁻ concentrations reported in Organic Matter, Sulfate (SO ₄ ²⁻ ) and Chloride (Cl ⁻ ) Contents. The starting CH₄ concentration was the in situ value we measured from the triplicate sub-sections (In situ CH ₄ Content). The results of this model will thus be an overestimate, since CH₄ was most likely lost in the vials during headspace flushing. We assumed the distance for diffusion was 3 cm, the distance in the vials.

Cored sediments consisted of bedded glaciofluvial sands with interbeds of gravel and silt, thought to have been deposited during the waning stages of the late Wisconsinan glaciation (∼12,000 years before present; Rampton, 1988). The stratigraphy consisted of a near surface organic-rich interval (containing rootlets) that extended to ∼60 cm depth, ice-bonded fine sands with occasional pebble and silt-rich layers and intervals of massive ice with >80% excess ice by volume (Figure 4B). Thermistor data shows that sediment temperature fluctuates annually to a depth of ∼11.5 m, with a mean annual ground surface temperature of approximately −5°C. The active layer is interpreted to be within the upper 0.6 m based on where the maximum temperature crosses 0°C (Figure 4C). Similar ground temperature measurements from a nearby site (∼15 km east of Tuktoyaktuk 69.4736°N, 132.6285°W) carried out in 1995 (Dallimore et al., 1996; Figure 4C), show that mean annual ground temperatures may have increased 3–4°C in the past 25 years.

The in situ sediment physical properties varied with depth (Figure 4D) reflecting changes in the geology at the site (mainly grain size) and ground ice content. Permafrost sediment porosity averaged 0.48 ± 0.09 down core (Figure 4D). Gravimetric water content was ∼1.0 g/gdwt (grams dry weight) near the surface and at ∼2 m, decreased to ∼0.25 for the remainder of the core (Figure 4D). We observed an ice-rich interval at 3.5 m sediment depth (Figure 4B), which corresponded to an increase in both gravimetric water content and porosity. There was no indication of gas bubbles in the ice. Wet bulk density mirrored the other physical parameters (Figure 4D). Compared to unfrozen settings, the permafrost samples had very low bulk densities and high porosities, especially those with elevated ground ice content.

TOC concentrations (given in weight %) in the uppermost surface sample (∼0.6 m deep) was 2% in one replicate and 7% in the other two replicates, indicating heterogeneity of material in the active layer (Figure 5A). TOC then decreased to ∼0.5% by 1 m, exhibited a sub-surface peak at 2.8 m (3%) and decreased again to 0.5% for the remainder of the core. Similar to TOC, DOC concentrations were highest in the surface sample (60 mg/L) and exhibited a sub-surface peak of 25 mg/L at 2.8 m depth (Figure 5A). DOC concentrations remained at 11–17 mg/L for the remainder of the core. Specific UV absorbance SUVA₂₅₄ values of the DOM throughout the core ranged from 1.3 to 8.9 m⁻¹ L mg⁻¹ (Figure 5A). The values were variable down core with two small sub-surface peaks which corresponded to the highest CH₄ values (highest SUVA value of 8.9 ± 2.5 m⁻¹) and the TOC/DOC sub-surface peak (4.3 ± 2.2 m⁻¹ L mg⁻¹).

δ¹³C-TOC values ranged between −26.7 and −7.4‰ and average -18.6 ± 5.9‰ (Figure 5B). The highest δ¹³C-TOC value of −7.4‰ was found at the deepest depth. δ¹⁵N-TON values ranged between 2–5‰ for the whole core with a sub-surface peak of 8.6‰ at ∼1 m. TON concentrations were highest at the surface (0.4%) and decreased to 0.06% at 3.4 m and was undetected at our deepest depth (data not graphed). C:N values ranged between 15–24 for the whole core (Figure 5B). SO₄ ²⁻ and Cl⁻ concentrations were between 10 and 20 mM in the upper 3 m with large variability within replicate samples (Figure 5C), likely indicating variable ice concentrations in subsamples. Below 3 m, concentrations decreased to ∼2 mM SO₄ ²⁻ and 6 mM Cl⁻.

CH₄ concentrations in the core samples exhibited a sub-surface peak at ∼1.5 m of 0.28 mg CH₄ per kg sediment (Figure 5D). Above this sub-surface peak, the two shallowest samples collected within or just below the modern active layer (0.6 and 0.9 m depth) had the lowest CH₄ concentrations, despite having the highest TOC values. Beneath the sub-surface peak, CH₄ concentrations decreased and became constant at ∼0.03 CH₄ per kg sediment down to 11.6 m in the core (Figure 5D).

Sediment incubations were carried out at four depths (0.6, 1.5, 2.8, and 11.6 m) that had varied in situ geochemistry and physical properties. As shown on Figure 4C, these depths also experience different seasonal permafrost temperature regimes. CH₄ concentrations in the killed control treatments slightly increased over time for all the depths except 1.5 m, which decreased over time (data points in Figure 6). When a simple 1-D diffusion model was applied to the incubation results, for all depths except the 1.5 m, the slight increase was explained by simple diffusion of in situ CH₄ out of the sediment plug upon incubation set-up (model lines in Figure 6). The results from these controls gave us confidence that any values above those measured in the killed controls could be interpreted as the gross potential for microbial CH₄ production stimulated by the incubations.

For all depths of the incubations, CH₄ and CO₂ concentrations increased over time in all treatments (including some of the frozen treatments; Figure 7). The gross production rates were faster at the higher temperatures than at the lower temperatures. The cumulative gross production rates increased linearly at first, indicative of the growth phase, and then plateaued, indicative of sustained production. The active layer incubation (i.e., 0.6 m) had the lowest CH₄ production rates of any incubation at +15°C (0.0012 µmol CH₄ per gdw, Figure 7A). At −20°C, the rate is 0.000012 µmol CH₄ per gdw (Figure 7A). This active layer also had the highest CO₂ production rates (Figure 7B), with the CO₂:CH₄ ratio reaching 10,000 at the end of the experiment (Figure 7C). The incubation of sediment with the highest in situ CH₄ concentration (i.e., 1.5 m) exhibited the highest gross CH₄ production rates of any incubation (0.15 µmol CH₄ per gdw at 5°C and 0.03 µmol CH₄ per gdw at −5°C; Figure 7D). The CO₂ production rates in this sample were similar for the +5°C and −5°C treatments at about 3 µmol CO₂ per gdw (Figure 7E), resulting in CO₂:CH₄ ratios less than 100 for all temperatures and a pattern where ratios were higher at the colder temperatures (Figure 7F). The incubation of sediment from the sub-surface organic carbon peak (i.e., 2.8 m) had the highest CO₂ production rates at frozen temperatures compared to the other depths (averaging µmol CO₂ per gdw at all temperatures, even in the 15°C treatment; Figure 7H). Gross potential CH₄ production rates reached as high as 0.024 µmol CH₄ per gdw at +15°C in this sample (Figure 7G). Rates decreased with colder temperatures but remained above the killed controls even at −20°C. The CO₂:CH₄ ratios in this sample showed a similar pattern as the above depth but were an order of magnitude higher (Figure 7I). In the deepest permafrost sediment incubation (i.e., 11.6 m), gross potential rates of CH₄ production were near the diffusion line for the frozen temperatures, indicating no microbial production, and there was a ∼25 days lag phase before rates increased to 0.015 µmol CH₄ per gdw in the 5°C incubation (Figure 7J). CO₂ production rates were the lowest observed in the incubations (Figure 7K), resulting in CO₂:CH₄ ratios that showed no pattern with time (Figure 7L) and that were negative at times (data not shown). Negative ratios mean that CH₄ production exceeded CO₂ production.

While most of the permafrost sediment depths analyzed had relatively low organic carbon content (TOC: ∼0.5 weight%; DOC: ∼10 mg/L), as might normally be expected for glaciofluvial sediments, we observed some depths with high TOC (2–7 weight%) and DOC (20–60 mg/L) contents (Figure 5). The active layer had the highest TOC and DOC concentrations, and a secondary peak was observed at ∼3 m depth. This depth may represent an organic-rich (not peat) paleo-active layer formed under warmer temperature conditions in the past. Indeed, other studies conducted in this area have documented the presence of older and deeper organic horizons thought to have formed during the Hypsithermal warm period in the early Holocene (Burn, 1997).

The permafrost organic matter observed in the core samples suggested possible terrestrial and marine sources, based on the ¹³C and ¹⁵N signatures of organic matter (Figure 5B). Terrestrial ecosystems (dominated by C3 plants) typically have ¹³C values that range from −23 to −34‰, whereas marine systems range from −10 to −25‰ (Ehleringer et al., 1993). Values measured in our core are slightly enriched in ¹³C, compared to sediments from the Lena River Delta ranged from −29 to −26‰ (Knoblauch et al., 2013). The ¹³C-organic matter in the surface samples was −26‰, suggesting a predominantly C3 terrestrial source, whereas more positive ¹³C values of deeper layers indicate mixed C3 terrestrial and marine sources. The ¹³C-organic matter value of −7.4‰ for the deepest layer does not reflect either of these sources. Such heavy values have been measured when the terrestrial system is dominated by C4 plants, which are typically tropical and warm-temperate plants. Certainly tropical plants are not supported at this site; however, unconsolidated Tertiary strata, which were deposited during much warmer climate conditions, do occur at shallow depth in the Tuktoyaktuk area, and reworked Tertiary pollen is often seen in Quaternary sediments (Ritchie, 1974). This may be a possible source for the enriched isotope values observed in this sample.

The stable isotope values give us a bulk view of the organic matter available to the microbial community, but microorganisms typically access carbon from the dissolved pool. Therefore, to compliment the stable isotope work, we also assessed the quality of the DOM with optical methods (Supplementary Material). We focus the discussion on SUVA₂₅₄, which relates to the aromatic content of the molecules in the dissolved pool (Traina et al., 1990; Weishaar et al., 2003; Fellman et al., 2008). We recognize that this parameter has limitations in that there is no way to distinguish between a low concentration of highly absorbing material and a high concentration of low absorbing material. Furthermore, permafrost pore waters may have a very different chemical composition than the organic matter isolates analyzed in previous studies, so the relationship to DOM aromatic content may be different here. But, as a generalization, the higher the aromatic content, the less “labile” and more “refractory” the organic carbon is for microbial processes. Given the debate in the literature now with such labels as “labile” and “refractory” (Hansell, 2013) we use this measurement here only as an indicator of organic matter quality and not a definitive measurement. The permafrost sediment layer with the highest in situ CH₄ concentration (Figure 5A), the highest potential for CH₄ production (Figure 7D), and some of the lowest CO₂ production potential (Figure 7E), had the highest SUVA₂₅₄ value (Figure 5A). This indicates that the labile carbon to fuel microbial heterotrophic or fermentative respiration (i.e., producing CO₂) is less available at this depth, and that methanogens are likely using more refractory carbon to produce CH₄. Likewise, the deeper permafrost sediment layer with a subsurface TOC peak also had relatively high SUVA₂₅₄ values (Figure 5A), the second highest CH₄ production potential (Figure 7G), and modest CO₂ production potential (Figure 7H). By contrast, the sediment from the active layer had low SUVA₂₅₄ values (i.e., was more labile) and the lowest and highest CH₄ and CO₂ production potentials, respectively (Figures 7A, B). Overall, these patterns support the observation that permafrost carbon quality, in addition to quantity, plays an important role in determining greenhouse gas production potential (Schädel et al., 2014; Kuhry et al., 2020).

We detected in situ CH₄ in all frozen core samples with values ranging from 0.03 to 0.28 mg CH₄ per kg sediment (Figure 5A). The highest value was measured in the organic-rich layer just below the modern active layer. These values are consistent with published values from other permafrost settings with low organic contents. For example, Alaskan discontinuous permafrost sediments contained ∼0.001–22 mg CH₄ per kg sediment (Kvenvolden and Lorensen, 1993), and silty sediments with ground ice near Prudhoe Bay, Alaska, had 0.2–11 mg CH₄ per kg sediment (Rasmussen et al., 1993). Higher values have been observed in organic-rich active layer sediments with lacustrine vegetation (15 mg CH₄ per kg sediment (Rivkin, 1998); in the active layer of a permafrost floodplain and within a polygon center (6.5 and 8,656 mg CH₄ per kg sediment, respectively Ganzert et al. (2007)), and Lena Delta permafrost setting (4.8 mg CH₄ per kg sediment (Wagner et al., 2007)). Our study adds to the body of work on this topic and demonstrates that in situ CH₄ is common in Arctic permafrost in a variety of sediments and geologic settings.

In addition to the concentration of in situ CH₄ in permafrost, the state of CH₄ (i.e., as bubbles trapped in ice, free gas in the sediment, and/or dissolved) is important when evaluating the potential for greenhouse gas release from these environments and for understanding operative geologic processes. Given that all of our core samples were collected in the winter when ground temperatures were well below −5°C, and that they were stored at −20°C until testing, we envisage that the CH₄ detected was residing within a heterogeneous matrix consisting of sediment grains (sand, minor silt and some organics), pore ice and possibly a very small fraction of liquid water [even in sand, small amounts of unfrozen water can occur at negative temperatures (Watanabe et al., 2011)]. By comparing the in situ ice content (as estimated visually and from gravimetric water content and density determinations) vs. CH₄ concentrations, we can estimate if CH₄ existed as bubbles within ice. The highest ground ice contents within the samples investigated were within the active layer and at 3.5 m depth (Figure 4D), where the lowest CH₄ concentrations were measured (Figure 5D). This indicates that the CH₄ is not dominantly residing as bubbles in the ground ice, as coarse-grained sands are almost completely frozen at temperatures slightly below 0°C (Williams, 1967). Thus, we envisage that a significant fraction of the in situ CH₄ may reside within the sediment matrix as free gas, and a small amount dissolved within unfrozen water. While the presence of free gas in permafrost settings is poorly studied, free gas was detected in permafrost research wells in the Mackenzie Delta (Dallimore and Collett, 1995), and has been implicated in the formation of permafrost craters in coastal settings in Siberia (Leibman et al., 2014). Likewise, some active geomorphic features may be associated with free gas release (Mackay, 1965). As finer-grained, organic rich sediments may be more conducive to free gas occurrence, additional studies are needed to constrain the physics and chemistry of CH₄ residing with the pore space of permafrost sediments.

We envisage that a number of coastal processes could cause direct release to the atmosphere of the in situ CH₄ observed in the Tuktoyaktuk Island permafrost. Tuktoyaktuk Island is undergoing rates of coastal erosion of ∼2.5 m per year, mostly during storm events which is significant when compared to other Arctic coastal settings (Lantuit et al., 2012). These storms can cause erosional niche formation creating block failures (Figure 8 upper left photo). When these events occur, the in situ CH₄ can be immediately exposed to the atmosphere on the cliff faces. The permafrost away from the cliff faces can also rapidly thaw over a period of hours or days. Because the thawing is very rapid we speculate that the in situ CH₄ could be released without being oxidized to form CO₂. At Tuktoyaktuk Island coast, 6 m of permafrost is exposed on the exposed cliff edge (see Figure 4A). If we assume an average concentration of 0.15 mg CH₄ per kg sediment in this upper 6 m, and an erosion rate of 2 m/yr, we estimate an annual CH₄ flux rate from the eroding sediments could be ∼3,500 g CH₄/yr/km of shoreline. Eroding coastal areas in the Canadian Beaufort Sea comprise at least 1,000 km in length (Solomon and Gareau, 2003; Manson et al., 2019) thus, if similar sediments were encountered elsewhere in the region, there is potential for atmospheric release of 3.5 × 10⁶ g CH₄/yr. While this is a relatively small amount compared to the ∼1 × 10¹² g CH₄ estimated to be released from the global permafrost reservoir annually (Saunois et al., 2020), it should be noted that the glaciofluvial sediments we have studied from Tuktoyaktuk Island are not particularly organic rich. A full assessment of the potential for release of CH₄ to the atmosphere from this process is only possible with further testing of the different types of eroding sediments along the coast. Other eroding sedimentary deposits commonly found along the Beaufort coast include glacial till, lacustrine and deltaic sediments (Rampton, 1988) which may contain higher in situ CH₄ concentrations and therefore result in increased atmospheric release from event driven erosion.

The observation of in situ CH₄ in permafrost environments raises the question of how this CH₄ was formed and if in situ CH₄ concentrations could be increasing in association with warming of the permafrost due to climate change? Our incubation experiments clearly document that microbial activity remediate permafrost organic matter producing CO₂ and CH₄ with warming under still frozen conditions. Previous studies have documented the presence of viable microorganisms (Ganzert et al., 2007; Jansson and Taş, 2014; Stapel et al., 2016; Altshuler et al., 2017; Margesin and Collins, 2019) and active bacteria (Elberling and Brandt, 2003; Wagner et al., 2007; Mykytczuk et al., 2013) in frozen permafrost. Our incubations support these findings that there is an active, viable population of microorganisms producing CH₄ at below freezing temperatures. We can not, however, speculate on the mechanism driving methanogenesis without further experiments with isotope labeled substrates (as done in Rivkina et al., 2004).

In terms of the geothermal setting of the cores from Tuktoyaktuk Island, we note the geothermal environment of the sediments is within the interval that experiences annual seasonal fluctuations in temperatures with warming and cooling that ranges from perhaps >10°C at 1 m depth to only a 1–2°C at 10 m depth. At in situ temperatures, our results demonstrate that CH₄ and CO₂ is potentially produced when the upper intervals of sediment experience seasonal warming. Indeed the in situ CH₄ concentrations we observed likely result from slow but repetitive seasonal formation of biogenic gas, that is then trapped within the ice bonded sediment matrix.

Our experiments may also have implications when considering future climate warming and its impact on in situ greenhouse gas concentrations and geomorphic processes. While the pace of warming of subsurface permafrost is clearly somewhat slower than the pace of atmospheric warming, our work suggests that temperature increases of only a few degrees can enhance activity of the in situ microbial communities, generating greenhouse gases. This creates the potential for build-up of CH₄ and CO₂ under warming permafrost conditions than might be expected from seasonal temperature variations alone. This points to the potential for subsurface biogenic gas accumulations which are continually being charged or pressurized over time. Such conditions may create the potential for the in situ pressure regime to exceed the confining pressure of the permafrost cover. This in turn could cause the trapped gas to leak to the surface through cracks in the permafrost or even cause more catastrophic pressure release as is inferred for the formation of permafrost craters described in Siberia (Leibman et al., 2014). There is also the potential for the formation of intrapermafrost CH₄ gas hydrate within the permafrost interval at depth (Dallimore and Collett, 1995).

At incubation temperatures above freezing, the amount of CH₄ produced in the Tuktoyaktuk Island samples, in terms of moles CH₄ per weight sediment or petagrams CH₄ per gram carbon (Figure 9) was comparable to rates observed in incubations with sediments from other settings with higher organic carbon contents (Roy Chowdhury et al., 2015; Treat et al., 2015). These results suggest that glaciofluvial sediments found along the coast of the Beaufort Sea, which might not normally be considered as candidates for high greenhouse gas production upon thawing, could be contributing significant amounts of modern microbial greenhouse gases in association with ongoing climate warming. A key consideration for our experiments however is that they were conducted under anaerobic conditions. If sediment processes associated with thawing change the in situ biogeochemical environment from anaerobic to aerobic conditions, of course there is potential for oxidation of CH₄ to form CO₂.

We anticipate a number of settings, found within a landscape such as Tuktoyaktuk Island, that experience greenhouse gas release from coastal processes. The first, and most widespread setting, is in association with a deepening of the active layer due to atmospheric warming along the coast (Figure 8A). In this case, CH₄ and CO₂ can be generated under anaerobic conditions similar to our experiments at the base of the active layer by microorganisms that already exist within the permafrost but become more active under thawed conditions. In addition, if new labile organic material thaws at the base of the active layer, there is potential for new microbial communities to become active and generate CH₄. If CH₄ and CO₂ concentrations at the base of the active layer are substantive there is potential for free gas migration through the active layer quickly without oxidation. However, as observed in other settings, it is likely that at least some of the CH₄ is oxidized to CO₂ in the upper portion of the active layer which is likely to be oxygenated (Oh et al., 2020 and references within). The second candidate geologic setting relate to the possible mobility of organic material through an unstable landscape as it thaws and is re-deposited in new sedimentary environments (Figures 8B–E). At Tuktoyaktuk Island, this might include terrestrial landslide deposits, beach sediments or shallow nearshore settings on the north side of the Island, which are in oxygenated waters. Our experiments, conducted under anoxic conditions do not directly address this environment, however new studies (Tanski et al., 2019) are beginning to better characterize these conditions demonstrating potential for CO₂ production and potential atmospheric release. However, we note that in Tuktoyaktuk Harbor on the south side of Tuktoyaktuk Island, deep water pockets allow for anoxic conditions to form year round creating conditions similar to our incubation experiments (Figure 8E; Lawrence et al., 1993). Local CH₄ and CO₂ production may possibly be quite vigorous in this setting. And finally, as the coastal setting is inundated with sea level rise, organic matter can accumulate in the offshore setting and host similar microbial processes to release CH₄ and CO₂ to the overlying waters (Figure 8D).