Methane emissions from subglacial meltwater of three alpine glaciers in Yukon, Canada

ABSTRACT Subglacial meltwater of land-terminating glaciers in Greenland and Iceland are sources of methane (CH4) to the atmosphere, but sparse empirical data exist about the spatial distribution of subglacial CH4 production and emission from glaciers in other regions of the world. This study presents the first measurements of CH4 emissions from the subglacial meltwater of three outlet glaciers of the St. Elias Mountains in Yukon, Canada. Dissolved CH4 concentrations were highly elevated at 45, 135, and 250 times compared to the atmospheric equilibrium concentration in the meltwater of Dusty, Kluane, and Donjek glaciers, respectively. Dissolved CO2 concentrations were depleted relative to the atmospheric equilibrium. This points to the meltwater being a source of CH4 and a sink of CO2. Stable carbon (13C) and hydrogen (2H) isotopic signatures of the subglacial CH4 were depleted compared to atmospheric CH4 at all sites, indicating both biotic and abiotic sources and possible alteration from bacterial CH4 oxidation in the meltwater. No relation was found between CH4 concentrations in the meltwater and the meltwater chemistry or the size of the glaciers in this study. These findings suggest that CH4 emissions from subglacial environments under alpine glaciers may be a more common phenomenon than previously thought.


Introduction
Meltwaters from subglacial environments of landterminating outlet glaciers from the Greenland Ice Sheet (Christiansen and Jørgensen 2018;Lamarche-Gagnon et al. 2019;Christiansen et al. 2021;Pain et al. 2021) and Iceland (Burns et al. 2018) and, with much lower concentrations, from alpine glaciers in Alaska (Konya et al. 2022) and China (Du et al. 2022) are newly discovered sources of direct methane (CH 4 ) emissions to the atmosphere.Little empirical data exist about these direct CH 4 emissions, their spatial occurrence and distribution in a wider variety of glaciated regions of the world, glacier types, and dynamic behaviors (e.g., surging glaciers).Similarly, little is known about its glacial and biogeochemical drivers.Considering that 11 percent of the Earth's landmass is covered by ice masses (Boyd et al. 2010), this limits our understanding of its regional and global importance for atmospheric carbon budgets and its possible role in the climate system (Wadham et al. 2019).
Although direct emissions of CH 4 have only been shown for five glacial outlets worldwide, it is widely acknowledged that the subglacial CH 4 is most likely of biotic origin produced via methanogenesis (acetoclastic and hydrogenotrophic) by methanogenic archaea that exist at the base of glaciers (Skidmore, Foght, and Sharp 2000;Boyd et al. 2010;Stibal et al. 2012;Burns et al. 2018) and ice sheets (Christner, Montross, and Priscu 2012) where they convert organic carbon to CH 4 under anoxic conditions (Burns et al. 2018;Lamarche-Gagnon et al. 2019;Wadham et al. 2019;Christiansen et al. 2021).Field evidence also points to the fact that bacterial oxidation of this CH 4 to CO 2 happens in the subglacial environment (Michaud et al. 2017;Pain et al. 2021), reducing net CH 4 emissions.In addition, it has been suggested that subglacial CH 4 can originate from other processes, such as the release of trapped crystalline abiotic CH 4 (e.g., formed at high temperatures in the mantle) through mechanical grinding of rocks at the glacier bed (Gill-Olivas et al. 2023) or thermogenic CH 4 (e.g., produced by thermal degradation of organic matter in the upper crust) released from destabilization of CH 4 hydrates below ice sheets (Wadham et al. 2012(Wadham et al. , 2019;;Portnov et al. 2016;Pain et al. 2021).
The subglacial organic carbon can originate from various relict proglacial in situ sources such as soil organic matter, vegetation, peat, and lacustrine and marine sediments, which can be overridden and covered by ice during a glacier advance (Wadham et al. 2008;Stibal et al. 2012;Burns et al. 2018).Glacier retreat and readvance may impact the amounts of organic carbon that can accumulate underneath a glacier and be turned into CH 4 .Though the published studies collectively indicate that microbial methanogenesis and oxidation can occur under various subglacial environments of glaciers and ice sheets, we know very little about the spatial occurrence of subglacial CH 4 and CO 2 processes and their rate-limiting drivers in different glacial settings.
The main aim of this study was to (1) investigate whether glaciers of the St. Elias Mountains had the potential for direct CH 4 and CO 2 emissions from subglacial meltwater and, if present, (2) determine the carbon source(s) of subglacial CH 4 and (3) investigate how bedrock geology, glacier size, and surging history are linked to CH 4 concentrations in the meltwater.
To fulfill these aims, we sampled subglacial meltwater at three land-terminating outlet glaciers of the St. Elias Mountains in the Kluane National Park and Reserve in Yukon, Canada.We measured dissolved CH 4 and CO 2 concentrations in meltwater and air directly in the field and the laboratory and analyzed the isotopic composition of dissolved CH 4 and meltwater chemistry.

Study area
The St. Elias Mountains stretch along the border of Canada and the United States in southwestern Yukon and southeastern Alaska (Figure 1) and are covered by various glacier types, such as valley glaciers, piedmont glaciers, plateau glaciers, and ice fields with outlet glaciers (Clarke and Holdsworth 2002).The elevation range from sea level to almost 6,000 m.a.s.l.results in a temperature and precipitation gradient and the full spectrum of glacial thermal regimes from cold-based to polythermal and temperate (Clarke and Holdsworth 2002).
The study sites at the termini of three valley glaciers draining the St. Elias Mountains to the east were chosen based on three criteria: (1) different sizes of the catchment to presumably cover different sizes of subglacial areas with anoxic conditions; (2) the absence of a proglacial lake at the terminus to measure and collect samples as close as possible to subglacial outflows as degassing could have taken place, effectively lowering concentrations of dissolved gases at a greater distance; and (3) accessibility by helicopter.The choice was made by visual inspection of aerial photographs and satellite images.From north to south, the chosen sampling locations were Donjek glacier (61°12′32.29″N, 139°32′ 13.88″ W), Kluane glacier (60°58′30.58″N, 139°19′ 56.14″W), and Dusty glacier (60°25′41.52″N, 138°11′ 20.00″ W; Figures 1a, c-d).All three glaciers are classified as surge-type glaciers, indicating a temperate or polythermal regime (Murray et al. 2003).
Donjek glacier is the largest of the three studied valley glaciers when comparing size, length, and volume (Table 1).At the terminus, Donjek glacier flows into the Donjek river valley, where the glacier widens to a lobe.Donjek glacier is an active surge-type glacier with a short and regular surge cycle.The quiescent phase lasts around twelve years before a two-year-long surge (Kochtitzky et al. 2019).Since 1935, eight surges have been identified, the last from 2012 to 2014 (Kochtitzky et al. 2019).Kluane glacier flows into the Donjek river valley upstream of Donjek glacier and is the smallest and thinnest glacier of the study (Table 1).Kluane glacier is classified as a surgetype glacier due to visible fold trains along the ice of the glacier tongue, a morphological characteristic of surging glaciers (Young et al. 2022).Besides the surge of a tributary glacier from 2017 to 2018 (Samsonov, Tiampo, and Cassotto 2021), there is no documentation about previous surges of this glacier (Main et al. 2019).The third study site, Dusty glacier, is the second largest glacier of the study (Table 1).Dusty is a surge-type glacier, as suggested by the presence of folded medial moraines and a surge around 1966 (Clarke and Holdsworth 2002).The periodicity of the surge events of this glacier is unknown (Johnson 1995).
Sedimentary and igneous rocks dominate the bedrock geology of the study area.The majority of the catchment of Donjek and Kluane glaciers is underlain by carbonate rocks, mainly limestone and, to a minor degree, argillite and phyllite, and clastic rocks such as siltstone, sandstone, quartzite, and schist.Plutonic and volcanic rocks such as syenite, granodiorite, diorite, and gabbro are the second most abundant rock types under Donjek glacier but almost absent under Kluane glacier.The bedrock geology of Dusty glacier differs from that of the other two glaciers.Here, carbonate rocks cover only a small area of the catchment, which instead comprises clastic sedimentary and plutonic rocks (Yukon Geological Survey 2020; Figure 1a).

Materials and methods
Collection of samples from upwellings of pressurized subglacial meltwater that can be enriched with dissolved  gases (Lamarche-Gagnon et al. 2019).From a helicopter survey prior to choosing the exact sampling locations, it was observed that upwellings were present at the termini of all three glaciers.On 8 August 2022, all three glaciers were visited for 30 to 120 min, samples were collected, and measurements were performed a few meters from the center of each upwelling.On 11 August 2022, the upwelling at the terminus of Dusty glacier was revisited for a 10-h sampling campaign with hourly sampling.All samples were cooled at +4°C before analyses except on the flights from Canada to Denmark.

Measurements of CH 4 and CO 2
Atmospheric mixing ratios of CH 4 and CO 2 (gCH 4 and gCO 2 ) and dissolved concentrations of CH 4 and CO 2 (dCH 4 and dCO 2 ) were measured with three different methods: (1) ambient atmospheric background measurements with a portable CH 4 /CO 2 gas analyzer; (2) in situ screening tests of dissolved gases with the portable CH 4 /CO 2 gas analyzer; and (3) in situ extracted discrete gas samples for analyses via gas chromatography.

Atmospheric background measurements
Ambient atmospheric mixing ratios of gCH 4 and gCO 2 were measured in situ with a factory-calibrated CH 4 /CO 2 gas analyzer (GLA131-GGA, LGR, sampling frequency 1 Hz with a precision of 0.9 ppb and a flow rate of 2.7 L min −1 ) by placing a tube connected to the inlet of the CH 4 /CO 2 gas analyzer a few meters from the upwelling at all sites and measuring continuously for 5 min.These measurements allowed the detection in the field if the meltwater interacted with the atmospheric mixing ratios of CH 4 and CO 2 by exchange with the atmosphere.These numbers were used to correct concentrations of dissolved gases extracted in the field into an atmospheric headspace as described in method 2.

In situ screening tests of dissolved CH 4 and CO 2
To determine the presence of dCH 4 and dCO 2 in the meltwater with in situ screening tests, 2 L meltwater was collected in 4 L ziplock bags, allowing a headspace with ambient air.The bags were sealed and shaken for 2 min to extract dissolved gases into the headspace.A plastic tube was fitted between the ziplock bag and the inlet of the CH 4 /CO 2 gas analyzer, permitting in situ measurements of headspace CH 4 and CO 2 mixing ratios.The bag was squeezed tightly around the tube to avoid contamination with atmospheric air, but it cannot be excluded that relatively small amounts diluted the headspace after shaking.The best estimate of the headspace volume in the bag was calculated by emptying the headspace using the CH 4 /CO 2 gas analyzer pump while measuring the mixing ratio of CH 4 and CO 2 in the headspace.
Multiplying the measurement duration (13-30 s) by the flow rate (2.7 L min −1 ) of the CH 4 /CO 2 gas analyzer, plus an additional approximately 100 mL (equivalent to 2 s of pumping) left in the bag to avoid water entering the CH 4 /CO 2 gas analyzer (see Figure S1a and b), yielded a headspace of 0.59 to 1.35 L. The measurement duration of 13 to 30 s was long enough for elevated concentrations to stabilize due to the combined high flow and precision of the CH 4 /CO 2 gas analyzer.In situ tests were performed four times at Dusty and Donjek glaciers and three times at Kluane glacier on 9 August 2022, and between two and three times every 4 h at Dusty glacier on 11 August.The dissolved concentration of CH 4 (nmol L −1 ) was calculated using the procedure described in Christiansen et al. ( 2021) assuming a water temperature of 0°C during extraction and subtracting the 5-min average ambient CH 4 atmospheric mixing ratios (Figure 2a) from measured headspace CH 4 concentrations in the bag to account for the additionally added ambient CH 4 .Concentrations of dissolved CO 2 from in situ screening tests were not calculated because the standard method (Weiss 1974) assumes a higher than atmospheric equilibrium concentration in the water sample, which was not the case.

Discrete samples for dissolved CH 4 and CO 2 concentrations
Discrete gas samples were collected to determine the dCH 4 and dCO 2 concentrations more accurately than in the in situ screening tests.A bucket (10 L) was filled with meltwater from the upwelling to allow the collection of duplicate samples for dissolved gases from the same body of water while avoiding potential degassing between replicates due to turbulence.Immediately after filling the bucket, two 60-mL syringes capped with three-way valves were gently submerged in the water with the plunger out.The plunger was inserted into the syringe under water and excess water and gas bubbles were pushed out of the syringe by holding it vertically until 40 mL of water remained.Twenty milliliters of N 2 gas was drawn into the syringes from an N 2 -filled Supel-inert multilayer foil bag, forming a 20-mL N 2 -filled headspace.The three-way valves were closed, and the syringes were shaken for 2 min, avoiding unintentional heating from hands by holding them on the plunger during shaking.A syringe needle was mounted on the valve, and the first 5-mL headspace was discarded.From each syringe, duplicate pre-evacuated 3-mL glass exetainers were filled and overpressurized with 5-mL headspace.On 8 August, four exetainers were filled at the three field sites.At Dusty glacier on 11 August, four exetainer samples were collected every full hour between 10:10 and 20:00.In the laboratory, overpressure was released from the exetainers by inserting a needle through the rubber septum.Three months after sampling, the exetainers were analyzed on an Agilent 7890A gas chromatograph fitted with a flame ionization detector for CH 4 and CO 2 .A methanizer was used to convert CO 2 into CH 4 .The dissolved concentrations of CH 4 (nmol L −1 ) and CO 2 (µmol L −1 ) were calculated using the same procedure as in Christiansen et al. (2021) for CH 4 and Weiss (1974) for CO 2 , assuming a water temperature of 0°C during extraction.

Gas samples for the analyses of the isotopic composition of CH 4
From the outlet of the CH 4 /CO 2 gas analyzer, the measured extracted gases from the in situ screening tests were collected in rinsed and evacuated 2 L foil bags for analyses of stable carbon and hydrogen isotope values (δ 13 C-CH 4 and δ 2 H-CH 4 values, respectively).Each foil bag was filled with the headspace of two to four in situ gas extractions in ziplock bags to a total volume of ~2 L. One foil bag was filled at each field site on 8 August 2022.At Dusty glacier on 11 August, foil bag samples were collected every fourth hour at 12:30, 16:15, and 20:15.Analyses took place in the laboratory of the University of Heidelberg, Germany, four weeks after collection.In this study, the conventional δ-notation in per mill (‰) is used to express the stable isotope ratios of CH 4 relative to Vienna Peedee Belemnite and Vienna Standard Mean Ocean Water for carbon and hydrogen, respectively.
For the measurement of δ 13 C-CH 4 and δ 2 H-CH 4 values, gas samples from the foil bags were transferred to pre-evacuated 12-mL glass exetainers and analyzed using a Delta PLUS XL IRMS (ThermoFisher Scientific) coupled with an HP 6890N gas chromatograph (GC) that was equipped with a CP-PoraPLOT Q capillary column, using a Combustion III Interface.The combustion interface included an oxidation reactor (GC-C) held at a temperature of 960°C for stable carbon isotope ratio measurements and a thermo conversion reactor (GC-TC) maintained at a temperature of 1450°C for stable hydrogen isotope ratio analysis.Before the samples were analyzed in the GC-C/TC-IRMS system, it was introduced via a cryogenic preconcentration unit (for a more detailed explanation, refer to Greule, Huber, and Keppler 2012).The normalization of stable isotope values was performed by using two CH 4 reference gases that were calibrated against International Atomic Energy Agency and National Institute of Standards and Technology (NIST) reference materials, with δ 13 C-CH 4 values of −42.3‰ and −61.4‰ and δ 2 H-CH 4 values of −190.6‰ and −312.7‰.The normalization of the analyzed stable isotope values was performed as described in Paul, Skrzypek, and Fórizs (2007).

Water chemistry
On 8 August, meltwater samples for the determination of pH, electrical conductivity, alkalinity, major ions, dissolved organic carbon (DOC), and dissolved inorganic carbon were collected in 50-mL PVC tubes from the same meltwater-filled bucket as the dissolved gas samples rinsed three times with sample water and ensuring that no bubbles were present after sampling.Samples were not filtered or acidified in the field; hydrochemical data should be treated as conservative estimates because dissolution of particulate sediment during transport and storage could have increased the concentration of dissolved elements.Due to the fast dissolution of Ca 2+ and Mg 2+ , the reported values of these elements are estimated to be elevated compared to in situ concentrations (Slatt 1972).As a consequence, measured pH values, which were likely elevated due to dissolution of particulate Ca 2+ post sample collection, will not be reported.
A total of 150 mL was collected at Dusty glacier and 300 mL at Kluane and Donjek glaciers.On 11 August at Dusty glacier, one sample of 50 mL was collected each hour, except at 12:00, 16:00, and 20:00, corresponding to the timing of foil bag samples, when 110 mL was collected in one 50-mL and four 15-mL PVC tubes.
Water samples were analyzed in the laboratory of the Department of Geoscience and Natural Resource Management, University of Copenhagen.For samples collected on 8 August, three 50-mL samples were pooled into a 150-mL sample at the Dusty glacier site, and six 50-mL samples were pooled into 300 mL at Kluane and Donjek glaciers.For samples with a volume of >110 mL, electrical conductivity and alkalinity were measured simultaneously with an 855 Robotic Titrosampler, which was available for all samples collected on 8 August and three samples collected at Dusty glacier on 11 August at 12:00, 16:00, and 20:00.Prior to multielement analysis by inductively coupled plasma-mass spectrometry (ICP-MS) the samples were acidified by adding 200 µL concentrated nitric acid of subboiling quality to a 10-mL sample.Analysis was done on the iCAP-Q ICP-MS in KED mode using He as the collision gas.For quantification, external calibration and internal standards were used to compensate drift and matrix effects.The NIST 1643f and NIST 1640a reference materials were analyzed to evaluate calibration verification and accuracy.DOC was measured with a TOC-V CPH total organic carbon analyzer with a TNM-1 total nitrogen measuring unit.Ion chromatograph (IC) analyses for Cl − , NO 3 − , and SO 4 2− were performed on a 930 Compact IC Flex.HCO 3 − concentration was calculated from alkalinity (Rounds and Wilde 2012).The NH 4 + concentration was measured with continuous flow analysis using a Seal AA500 Autoanalyzer.

Comparison of CH 4 and CO 2 concentrations in air and meltwater between Donjek, Kluane, and Dusty glaciers
The effect of the CH 4 degassing (and hence emission) and net CO 2 uptake in meltwater on the atmospheric concentration was clearly visible at the Donjek site, where atmospheric mixing ratios of gCH 4 and gCO 2 were highly variable and consistently higher than the atmospheric background for CH 4 (~2 ppm) and lower for CO 2 (~415-420 ppm), respectively (Figures 2a,b).For the Dusty and Kluane sites, the background mixing ratios were stable and did not differ for gCH 4 and only a few parts per million for gCO 2 , which could be due to turbulent mixing by wind.
Discrete samples extracted by the headspace method showed the lowest dCH 4 concentrations in the meltwater of Dusty glacier (164 ± 5 nmol L −1 ), followed by Kluane glacier (482 ± 9 nmol L −1 ) and Donjek glacier (922 ± 26 nmol L −1 ; Table 2), corresponding to 45, 135, and 250 times the atmospheric equilibrium of dissolved CH 4 .dCO 2 concentrations were comparable between the sites and had a larger standard deviation between duplicate samples.The lowest dCO 2 concentrations were measured at Dusty glacier (5.5 ± 1.5 µmol L −1 ), followed by Donjek glacier (6.0 ± 1.7 µmol L −1 ) and Kluane glacier at (6.3 ± 0.9 µmol L −1 ).Average dCH 4 concentrations from discrete samples were in the same range as average concentrations from in situ screening tests, deviating only 2.1 percent, for all calculated dCH 4 concentrations (nmol L −1 ; Table 2).An example of an in situ screening test for dissolved CH 4 and CO 2 at each site is plotted in Figures S1a and S1b, and CH 4 and CO 2 mixing ratios during all performed in situ screening tests are provided in and Table S1.
Dissolved gas concentrations of discrete samples collected over an 10-h period at Dusty glacier on 11 August 2022 (Figure 3) were of comparable magnitude to samples collected three days before at this site (Table 2).dCH 4 concentrations varied over the 10 h, starting at 113 ± 2 nmol L −1 at 10:10 and increasing to a maximum of 199 ± 8 nmol L −1 at 17:00 before decreasing slightly to around 176 ± 6 nmol L −1 at the end of the measurement period at 20:00.Average dCO 2 concentrations varied little, between 4 and 6 µmol L −1 , throughout the measurement period but showed considerable variability between duplicate samples.

Isotopic (δ 13 C-CH 4 and δ 2 H-CH 4 values) composition of subglacial CH 4
The stable carbon and hydrogen isotope values of dCH 4 at Dusty glacier during all four sampled time points were the most enriched of the three investigated sites and lie in a very narrow range with δ 13 C-CH 4 values from −47.5‰ to −45.5‰ and δ 2 H-CH 4 values between −197‰ and −195‰ (Figure 4).Compared to Dusty glacier, both δ 13 C-CH 4 and δ 2 H-CH 4 values were more depleted at Kluane and Donjek glaciers, which also contain a higher coverage of sedimentary rocks (Figure 1).dCH 4 at Kluane glacier displayed the most depleted δ 13 C-CH 4 and δ 2 H-CH 4 values of −56.6‰ and −250‰, respectively, whereas at Donjek glacier stable isotope values were less depleted, with δ 13 C-CH 4 and δ 2 H-CH 4 values of −51.3‰ and −248‰, respectively (Figure 4).

Water chemistry
The most abundant anion in the meltwaters of all three glaciers was HCO 3 − and cation Ca 2+ .HCO 3 − concentrations were 139, 45, and 79 mg L −1 , and Ca 2+ concentrations were 15, 21, and 26 mg L −1 in the meltwaters of Dusty, Kluane, and Donjek glaciers, respectively.DOC concentrations were highest in the meltwater of Dusty (9.6 mg L −1 ), followed by Donjek (8.3 mg L −1 ) and Kluane .The five most abundant cations in the meltwater of all glaciers after Ca 2+ were S > Mg 2+ > K + > Na + > Si (Table 3).The highest concentration of all major cations was measured in the water from Donjek glacier, followed by Kluane and Dusty glaciers.Overall, the chemical composition of the meltwater was similar between the three glaciers, with some deviation in concentrations, but due to nonfiltering and acidification in the field as mentioned above, concentrations should be regarded as approximate estimates.

Discussion
The first aim of this study was to investigate whether glaciers of the St. Elias Mountains have the potential for direct emissions of CH 4 and CO 2 from meltwater as has previously been shown for other glacial locations (Burns et al. 2018;Christiansen and Jørgensen 2018;Lamarche-Gagnon et al. 2019;Pain et al. 2021).The meltwater of all three surveyed glaciers was elevated in dCH 4 and depleted in dCO 2 compared to the equilibrium with atmospheric mixing ratios.Though the magnitude of dCO 2 concentrations was comparable between the three glaciers, dCH 4 concentrations varied significantly.dCH 4 concentrations in the meltwater of the Donjek, Kluane and Dusty glaciers (136-899 nmol L −1 ) fell within the range of published values from other known subglacial CH 4 -emitting glaciers (Christiansen and Jørgensen 2018;Lamarche-Gagnon et al. 2019;Pain et al. 2021).dCH 4 concentrations in this study (Table 2) were higher than reported for Kiattut Sermiat in southwestern Greenland (mean ± standard deviation: 9 ± 5 nmol L −1 ; Pain et al. 2021) and fell within the concentration range reported from the two other Greenland glaciers: Russell glacier (dCH 4 110 ± 78 nmol L −1 ; Pain et al. 2021) and Leverett glacier (271 ± 34 nmol L −1 over the first two months of the melt season; Lamarche-Gagnon et al. 2019) and the lower concentration range of previously reported values from the subcatchment of Isunnguata Sermia (1,575 ± 997 nmol L −1 ; Pain et al. 2021), which is often referred to as Point 660 (Lindbäck et al. 2015).The Point 660 site displayed pronounced diurnal and seasonal variations in dCH 4 that exceeded values measured in this study up to one order of magnitude (Christiansen and Jørgensen 2018;Christiansen et al. 2021;Pain et al. 2021).The comparably small diurnal signal of dCH 4 concentrations and stable dCO 2 concentrations in the 10-h time series of Dusty glacier could potentially be an indicator of pressurized meltwater discharge from a confined subglacial conduit with the first point of degassing of dissolved CH 4 at the glacier terminus.
In the meltwater of the Yukon glaciers, dCO 2 concentrations were significantly depleted (5 ± 2 µmol L −1 ) compared to the equilibrium at atmospheric concentrations.In contrast, meltwaters of other known subglacial CH 4 -emitting glaciers were reported to be enriched in dCO 2 throughout the melt season (Isunnguata-Sermia subcatchment, 58 ± 18 µmol L −1 ; Pain et al. 2021) or close to the atmospheric equilibrium (Russell glacier, 29 ± 4 µmol L −1 ; Pain et al. 2021).The meltwater of Kiattut Sermiat was also reported to be depleted in dCO 2 (19 ± 2 µmol L −1 ; Pain et al. 2021) but apparently to a smaller degree than the measurements from the Yukon glaciers as this study suggested.Carbonate and silicate weathering reactions of freshly eroded sediments at the bed of glaciers can consume CO 2 (St. Pierre et al. 2019).The Yukon glaciers cover predominantly carbonate bedrock, and the low concentrations of dCO 2 in the meltwater are therefore attributed to CO 2 consumption from carbonate dissolution and a slower dissolution of atmospheric CO 2 in surface meltwater compared to consumption at the glacier bed.
Measured dCH 4 and dCO 2 concentrations at the glacier margin represent the net concentration.They are the result of the interaction of the combined sources and sinks with the meltwater in between the source area(s) and the point of contact with the atmosphere.Concentrations will presumably vary on diurnal and seasonal timescales in response to changes in surface meltwater inflow and the size and physical configuration of the subglacial drainage system (Burns et al. 2018;Christiansen and Jørgensen 2018;Christiansen et al. 2021;Pain et al. 2021), which are essential parameters for the calculation of the overall magnitude of CH 4 and CO 2 emissions.Therefore, the dCH 4 and dCO 2 concentrations measured in the meltwaters of Donjek, Kluane, and Dusty glaciers can only be considered snapshots in time and are not necessarily representative of diurnal and seasonal average concentrations.
The second aim of the study was to determine the possible sources of the CH 4 based on the δ 13 C-CH 4 and δ 2 H-CH 4 values of dCH 4 from the three studied glaciers.The dual isotopic signatures of dissolved CH 4 are not conclusive as to whether the CH 4 is produced only by microbial methanogenesis of a subglacial organic carbon source, because the isotopic signature values of CH 4 of δ 13 C-CH 4 and δ 2 H-CH 4 are located in the proposed region of thermogenic CH 4 or in the transition to acetoclastic methanogenesis (Figure 4).CH 4 from the subglacial meltwater of Dusty glacier might have originated from a different source or a combination of multiple sources and processes compared to Kluane and Donjek glaciers.The dCH 4 from Kluane and Donjek glaciers was more depleted in 13 C, indicating a dominance of acetoclastic methanogenesis as the source (Figure 4), whereas the stable isotope values of CH 4 from Dusty glacier were more enriched in 13 C and in the same range as reported from the meltwater outlet of Sólheimajökull in Iceland (Burns et al. 2018), indicating a predominantly thermogenic source.The degree of 13 C depletion in the dissolved CH 4 is higher in the catchments with more sedimentary rocks.Though there may be a mechanistic link between bedrock geology and isotopic composition, we have too little data to discuss any links at this moment.
In recent studies (Burns et al. 2018;Lamarche-Gagnon et al. 2019;Pain et al. 2021), it has been argued that the oxidation of subglacial CH 4 , which shifts both δ 13 C-CH 4 and δ 2 H-CH 4 values toward more enriched values (see arrows in Figure 4), results in apparent isotopic signatures resembling a thermogenic origin, which has also been shown in other natural biogenic CH 4producing systems (Coleman, Risatti, and Schoell 1981;Wang, Welander, and Ono 2016).It is likely that the dissolved CH 4 found in the meltwater from glaciers of this study could have been altered by methanotrophy in the more oxygen-rich part of subglacial meltwaters.This would imply that dCH 4 under Dusty glacier was more oxidized than that from Donjek and Kluane glaciers based on the clear separation of the 13 C-CH 4 depletion (Figure 4).However, CH 4 can be potentially produced as a result of abrasion of especially carbon-rich sediments or rocks in the subglacial environment (Gill-Olivas et al. 2023).Because the bedrock under the studied glaciers, especially of Donjek and Kluane glaciers, is dominated by carbonate rocks, the resulting production of CH 4 from abrasion could contribute to the dissolved CH 4 in the meltwater.In addition, the studied glaciers are located in an active tectonic area where thermogenic formation, and subsequent evasion, of CH 4 in subducted and warmed carbon containing sediments could happen near the surface.The isotopic signatures of these alternative CH 4 sources as well as their impact on the isotopic signature of dCH 4 in subglacial meltwater remain unknown.The results of isotopic analyses for the studied glaciers potentially could imply that CH 4 from the subglacial meltwater of Dusty glacier originated from multiple sources which could also be the case for Kluane and Donjek glaciers.
Hence, the occurrence of multiple processes within the subglacial CH 4 cycle complicates an unambiguous characterization of the CH 4 source(s) only based on δ 13 C-CH 4 and δ 2 H-CH 4 values.Thus, future studies should address the extent of microbial methane oxidation in the subglacial meltwaters of the three investigated glaciers and include other geochemical abiotic and thermogenic CH 4 contributions, such as the presence of C 2 and C 3 gases (Bernard, Brooks, and Sackett 1976) and clumped isotopes (Agmuas Adnew et al. 2023) for a more complete overview of processes influencing the CH 4 cycle in these environments and the source of CH 4 .
The third aim of the study was to investigate how bedrock geology, glacier size, and surging history could be linked to concentrations of dCH 4 in the meltwater.In contrast to the known sites of direct subglacial CH 4 emissions in Greenland and Iceland (Burns et al. 2018;Christiansen and Jørgensen 2018;Christiansen et al. 2021;Pain et al. 2021), the studied glaciers in Yukon are different due to their sedimentary bedrock type and, hence, geochemical settings.Compared to other known subglacial CH 4-emitting glaciers, the valley glaciers in the Yukon are among the largest in size.Further, they differ in their dynamic behavior due to their classification as surge-type glaciers.
The differences in the hydrochemical composition of the meltwater between the glaciers can be related to the timing of sampling, the amount of discharge, and the portion of surface melt of the total discharge.The pattern of increasing dCH 4 concentrations measured over a 10-h period at Dusty glacier on 11 August coincided with similar trends for several of the dissolved elements (SO 4 2− , Cl − , NO 3 − , S, Mg 2+ , K + , Na + , and Si).This indicates a similar meltwater control on the subglacial export of these elements (Table S2) but not necessarily a geochemical regulation of CH 4 concentrations.A comparison to the hydrochemical composition of meltwater from other known CH 4 -emitting glaciers in Greenland that are underlain by metamorphic bedrock showed that Donjek, Kluane, and Dusty glaciers differ substantially in cation content, even with the consideration of enrichment post sampling, but were similar in dissolved anions (Figure S2).However, comparable dCH 4 concentrations have been reported from glaciers covering sedimentary (Yukon) and metamorphic bedrock (Greenland), indicating that bedrock geology, as reflected in the geochemical composition of the meltwater, is no valuable indicator for the presence or concentration of dCH 4 .An exception to this seems to be the case for the seepage of thermogenic methane from groundwater springs in recently deglaciated forefields in Svalbard, where the highest concentrations coincided with the presence of shale bedrock (Kleber et al. 2023).For dCO 2 , it was observed in this study and for Kiattut Sermiat (Pain et al. 2021), which is also partly underlain by sedimentary rocks, that dCO 2 concentrations were depleted.This could indicate that CO 2 -driven subglacial weathering of sedimentary rocks impacts dCO 2 to a higher degree than dCH 4 .This study provides too few data points to establish whether the highly different dCH 4 concentrations from Donjek, Kluane, and Dusty glaciers (Table 2) relate to differences in bedrock geology.The apparent correlation of dCH 4 concentrations and dissolved solutes could rather be a sign of a larger area of the glacier bed being flushed by meltwater, increasing solute and dCH 4 concentrations simultaneously.
With  S3).This indicates that areal or subglacial catchment extent alone does not explain the varying magnitudes of dCH 4 concentrations between the glaciers, but differences in CH 4 exports between glaciers will be better constrained by other factors; for example, the size of the subglacial area of anoxia, the ratio of subglacial versus surface melt of the discharge volumes, and the availability of organic matter and other nonbiogenic CH 4 sources.
Different to the known subglacial CH 4 -emitting glaciers to date, the three studied glaciers in the Yukon are all classified as surge-type glaciers (Clarke and Holdsworth 2002;Kochtitzky et al. 2019;Young et al. 2022), raising the question of whether the magnitude of CH 4 emissions could be influenced by glacier dynamics.The high abundance of surge-type glaciers is notable for the St. Elias Mountains, especially on the Canadian side (Clarke and Holdsworth 2002;Sevestre and Benn 2015).Out of 2,356 glaciers, about 6.4 percent show surge-type behavior (Clarke et al. 1986), the third highest accumulation after Svalbard and Pamir (Kochtitzky et al. 2019).In comparison, less than 1 percent of glaciers worldwide are classified as surging glaciers (Sevestre and Benn 2015).The studied Yukon glaciers follow highly different surge cycles.Donjek is the most active surge glacier of the three, followed by Kluane and Dusty glaciers (Clarke and Holdsworth 2002;Kochtitzky et al. 2019;Young et al. 2022).Interestingly, the surge frequency correlates with higher dCH 4 concentrations measured, and our data provide the first observation to suggest a potential link between subglacial CH 4 production and the dynamic behavior of glaciers.A hypothesis could be that the remobilization of subglacial sediment during a surge could establish contact between the subglacial anoxic environment and fresh organic material.Frequent surges could therefore provide a resupply of subglacial organic material sustaining subglacial CH 4 production.Surging further increases abrasion at the glacier bed, providing nutrients and energy sources from crushed sedimentary rocks that could drive methanogenesis (Gill-Olivas et al. 2023).A change in the distribution of ice volume due to surging and the accompanying decrease in basal pressure on the underlying bedrock could further lead to a release of stored gases within the lithosphere along fractures and faults (Vachon et al. 2022;Kleber et al. 2023).Repeated measurements of dCH 4 across a wider range of glacier types, surging and nonsurging, are needed to test this hypothesis.

Concluding remarks and perspectives
This study first investigates the presence of dissolved CH 4 and CO 2 in the subglacial meltwater of three outlet glaciers of the St. Elias Mountains in Yukon, Canada.We show that the subglacial meltwaters from three mountain glaciers, Donjek, Kluane, and Dusty glaciers, were sources of CH 4 and sinks of atmospheric CO 2 .dCH 4 concentrations in discrete samples were 45, 135, and 250 times higher in the meltwater of Dusty, Kluane, and Donjek glaciers, respectively, compared to the atmospheric equilibrium and were comparable to concentrations reported from subglacial meltwater rivers in Greenland and Iceland.dCO 2 concentrations were below the atmospheric equilibrium and lower than reported from other known subglacial CH 4 -emitting sites.
The stable carbon and hydrogen isotope values of CH 4 from Kluane and Donjek glaciers were similar to values measured for Greenland glaciers, where δ 13 C-CH 4 and δ 2 H-CH 4 values of meltwater from Dusty glacier showed more 13 C-enriched CH 4 .Thus, CH 4 at Dusty glacier originated either from a different source, such as thermogenic or from a mixture of thermogenic and microbial CH 4 when compared to Kluane and Donjek glaciers.Additionally, potentially occurring microbial methanotrophy might have led to an enrichment in the heavier isotopes.The ambiguous isotopic evidence of the origin of CH 4 from the observed sites is not completely understood and demands further investigation in the future and measurements of excess C2 and C3 gases to better constrain sources and sinks of CH 4 .
Compared to the currently known sources of direct subglacial CH 4 emissions in Greenland and Iceland, the studied glaciers in Yukon are different due to their sedimentary bedrock and hence geochemical settings as well as their classification as surge-type glaciers.No relation between dissolved ions in the meltwater, glacier size, and dCH 4 concentrations could be established.However, the CO 2 sink can be explained by carbonate dissolution.This finding indicates that bedrock geology may not be a determining factor of the presence of methanogenesis underneath glaciers; however, it might slightly influence the measured concentrations and isotopic values.Therefore, it is hypothesized that the residence time and the presence of anoxic unconsolidated sediments with available organic carbon substrates play the most critical role in determining whether CH 4 is produced subglacially.More glaciers need to be included and preferably longer time series established to determine whether there is a link between bedrock geology, glacier size, and subglacial CH 4 and CO 2 production and/or consumption, and more subglacial CH 4 -emitting sites must be found to constrain the physical parameters required for subglacial CH 4 production.
It remains unresolved whether the dynamic behavior of glaciers-for example, surging-could impact the methanogenic potential.In this study, Donjek glacier, the most active surge-type glacier, also shows the highest dissolved CH 4 concentrations in the meltwater.However, only one measurement of dCH 4 concentrations is not enough to answer this hypothesis.Further, more surge-type glaciers need to be investigated for elevated dCH 4 concentrations in the meltwater and longer time series measured to answer whether surgetype glaciers offer favorable conditions for subglacial CH 4 production and whether the surging frequency is positively linked to the presence and magnitude of direct CH 4 emissions from subglacial meltwater.
The discovery of three alpine glaciers as subglacial CH 4 -emitting sites proposes that subglacial CH 4 production might be a much more common phenomenon underneath the world's ice masses than shown through empirical measurements so far.More long-term time series measurements of dCH 4 and dCO 2 are needed to assess whether the subglacial meltwater of outlet glaciers from the St. Elias Mountains is a net source or sink of GHGs to the atmosphere over a melt season and how much attention should be given to such alpine glaciers in the calculation of future carbon budgets.

Figure 1 .
Figure 1.(a) Overview map of the study area.Source: Open Street Map, Mosaic of Sentinel-2 satellite images acquired between 23 July and 31 August 2021 at 50-m resolution, coordinate system WGS 84/UTM zone 8 N, glacier outlines as of 2010 (GLIMS Consortium 2005), and bedrock geology (Yukon Geological Survey 2022).(b) Location of study area on the border of Yukon and Alaska.Source: Google Maps.Close-up of sampling areas at (c) Donjek glacier, (d) Kluane glacier, and (e) Dusty glacier at scale 1:50,000.Source: Sentinel-2 satellite images from August 2021 at 10-m resolution, glacier outline as of 2010 (GLIMS Consortium 2005).

Figure 2 .
Figure 2. (a), (b) Atmospheric background mixing ratios of gCH 4 and gCO 2 measured with the CH 4 /CO 2 gas analyzer for a duration of 5 min a few meters from the center of the upwellings at each site.

Figure 4 .
Figure 4. Dual isotope plot showing δ 13 C-CH 4 and δ 2 H-CH 4 values of dissolved CH 4 at Dusty, Kluane, and Donjek glaciers compared to previous studies at other sites from Burns et al. (2018), Lamarche-Gagnon et al. (2019), and Christiansen et al. (2021).Stable isotope values taken from the literature are given as the mean average with standard deviation.Colored areas show the classification of CH 4 sources based on its stable isotope composition (adapted after Whiticar 1999).

Table 1 .
Basic glacier characteristics.Glacier size and length as of 2010 for all glaciers (GLIMS Consortium 2005), elevation in 15 arc seconds resolution (National Centers for Environmental Information 2022), and glacier surface flow velocities for 2017 and 2018 in meters per year (Millan et al. 2022).Note.Average values and standard deviation of elevation and surface flow velocity are shown in brackets.

Table 2 .
Summary of dCH 4 concentrations and dCO 2 mixing ratios from in situ screening tests and dCH 4 and dCO 2 concentrations from discrete samples.
Note.Values given for in situ screening tests were averaged over the number of tests performed per site (column n).Values given for discrete samples are averages of four exetainer samples.Discrete samples from Dusty glacier on 11 August were collected within 30 min of the time given for in situ screening tests.All measured values are presented with standard deviations.

Table 3 .
Results of hydrochemical analyses of meltwater of all three glaciers.Note.The four most abundant anions and six most abundant cations are listed.
areal extents of 448, 324, and 343 km 2 (Consortium 2005) for Donjek, Kluane, and Dusty glaciers, respectively, the glaciers in the Yukon are among the largest of all known CH 4 -emitting glaciers.The smallest glacier with reported subglacial CH 4 production is Kiattut Sermiat (36 km 2 ; Hawkings et al. 2016; Pain et al. 2021), followed by the subcatchment of Isunnguata Sermia (40 km 2 ; Lindbäck et al. 2015), Russell glacier (300 km 2 ; Pain et al. 2021), and Leverett glacier (600 km 2 ; Hawkings et al. 2016).Compared to Donjek, Kluane, and Dusty glaciers, sizes reported for the outlet glaciers of the Greenland Ice Sheet are all estimates of the subglacial catchment area.The areal extent or subglacial catchment area does not correlate with reported dCH 4 concentrations (Figure