Source apportionment of methane escaping the subsea permafrost system in the outer Eurasian Arctic Shelf

Significance Extensive release of methane from sediments of the world’s largest continental shelf, the East Siberian Arctic Ocean (ESAO), is one of the few Earth system processes that can cause a net transfer of carbon from land/ocean to the atmosphere and thus amplify global warming on the timescale of this century. An important gap in our current knowledge concerns the contributions of different subsea pools to the observed methane releases. This knowledge is a prerequisite to robust predictions on how these releases will develop in the future. Triple-isotope–based fingerprinting of the origin of the highly elevated ESAO methane levels points to a limited contribution from shallow microbial sources and instead a dominating contribution from a deep thermogenic pool.


S1: Isotope systematics and endmember pools in ESAS
Stable isotope source signatures (δ 13 C-CH4 and δD-CH4) reflect the source material and any isotopic fractionation related to the respective methane formation processes, leading to typical δ 13 C vs and δD ranges for microbial, thermogenic and abiotic CH4 as summarized below in Figure S1 (1,2). Observed stable isotopic signatures in nature often originate from a mix of sources; CH4 from gas hydrates, oceanic seep and mud volcanoes may be mixtures of CH4 from both microbial and thermogenic origin. Furthermore, isotope signatures can be altered by kinetic fractionation during diffusion and during microbial or abiotic oxidation (1,3). Such fractionation can be useful in deducing the extent of processes such as degradation yet may complicate source tracing (see (4) for examples).
Natural abundance radiocarbon composition ( 14 C-CH4) gives information about the age of the carbon atoms in the CH4 molecule. Fossil (e.g. thermogenic) sources usually contain no detectable radiocarbon ( 14 C-CH4 < -1000, corresponding to >60 000 years), whereas contemporary CH4 production will reflect the age of the substrate e.g. modern plant material or aged peat. The  14 C-CH4 is not influenced by fractionation effects (it is by definition corrected by normalizing the isotope ratio measurements to a standard  13 C-CH4 value). Source interpretations can be, however, difficult if the  14 C-CH4 signals reflects a mixture of sources. In addition, recent methanogenesis may use an old (radiocarbon depleted) carbon source.
Our literature review of isotope-based studies in the Siberian Arctic revealed deviations of observed stable isotopic signatures and derived isotopic endmembers from the generic ranges shown in Fig. S1. For example,  13 C-CH4 values have been observed to be more depleted than usual (5)(6)(7). This seems a typical pattern for permafrost regions, as CH4 originating under low temperatures experiences very slow fractionation during its formation (4). For a more reasonable classification of methane sources to the ESAS bottom waters, the (stable and also radiocarbon) isotopic signatures were grouped into four different endmember pools: 1. Modern microbial pool: This represents CH4 produced in Holocene sediments from the decomposition of terrestrial and marine organic material. This pool also includes production within the water column where relevant. Expected stable isotope signatures are -120 <  13 C-CH4< -60 ‰ (5-7), and -350‰ < D-CH4 < -170 ‰ (1,7). Radiocarbon signatures would be similar to those for organic carbon in the ESAS ( 14 C-CH4 = -232 ±147, (8,9)).

Old microbial deep pool:
This represents preformed CH4 held at depth in the ESAS sediment, originally formed as wetlands and Yedoma deposits and then inundated at end of the last glacial period. Methane in this pool can derive either from destabilizing gas hydrates or from rising, subsurface gas. Expected stable isotopic signatures are -120 <  13 C-CH4< -60 ‰ and -350‰ < D < -170 ‰) Radiocarbon signatures would be similar to pool 2 or older.
Although the isotopic endmember signatures for these three microbial pools can differ from each other locally, depending on exact environmental conditions during formation and exposure time to different alteration processes, the variation within one these pools can be equally large. It is thus not possible to distinguish those pools by their stable isotopic signature alone, thus we have decided to not assign them each distinct signatures, but rather a representative range for all of them.

Thermogenic pool:
This represents CH4 originating from abiotic production at elevated pressure and temperature. The methane is transported to the sediment surface through bedrock and sediment fractures and talliks. Stable isotopic signatures for thermogenic methane are enriched compared to microbial signatures, with their  13 C > -55 ‰ and D more positive than -275‰ (1). Signatures in the ESAS are expected to mainly represent natural gas, resulting in ranges of -52 <  13 C-CH4 < -37 ‰ and -210‰ < D -CH4< -160 ‰ (4,11). Radiocarbon signatures from this pool have  14 C-CH4 = -1000 ‰ (i.e. no detectable radiocarbon).

Radiocarbon blank characterization and correction
Blank characterization of the  14 C-CH4 purification system was performed by attaching an empty sample trap to the system and processing it in the same way as a sample. The size of the blank was determined by manometric quantification that had been calibrated using carbon content measurements from NOSAMS. This procedure was performed before every sample in the beginning, then after a small subset of samples once the blanks became more reproducible. Two blanks (from the beginning and end of the sample progressing period) were sent to NOSAMS for radiocarbon content analysis. The average of that value ( 14 Cblank= -637±108) was used for blank correction of the samples, using the following equation: These equations are solved for  14 C-CH4 and  14 C-CO2: For calculation of  14 C-CH4 from samples without duplicate,  14 C-CO2 in equation (6) was set to the average value of  14 C-CO2 derived from the duplicate samples ( 14 C-CO2 = -976 ± 29). The value showed no significant variation between stations and throughout the water column (see Table S1).

Error assessment
The error of the disentangled  14 C-CH4 signals is calculated using error propagation: As the errors in the CH4 and CO2 fractions are connected to each other, their combined error was calculating by using the span of  14 C-CO2 calculated with a realistic error in each fraction (1%). For single samples, (∆ 14 − 2 ) is given by the error in the average  14 C-CO2 value derived from the duplicate samples.

Figure S1: Classification of stable isotope signatures of methane sources and pathways
Stable isotope data from vertical and horizontal profiles at the two main seep stations are shown in the context of typical literature-based isotope classification ranges. Isotopic ranges for methane formation pathways are based on (1,2,12). Isotopic fractionation slopes are added for microbial oxidation (3,13), abiogenic oxidation (14), and diffusion (15). Data from station 13 and 14 are coloured according to their dissolved methane concentration, showing clearly a deviation from the "classical" ranges for lower concentrations that correspond to samples taken further away from the seep source.