First evidence for cold-adapted anaerobic oxidation of methane in deep sediments of thermokarst lakes

Permafrost contains about 1307 Pg carbon (C), with a substantial amount (450 Pg C; >25% of thaw-susceptible C) found in the Yedoma regions (Walter Anthony et al 2014, Strauss et al 2017). Most of the C is stored in deep layers of undisturbed permafrost soils and organic-rich thermokarst (thaw) lake sediments (Strauss et al 2013, Walter Anthony et al 2014). Thermokarst lakes are estimated to cover 1.3 × 10⁶ km² and store 102 Pg carbon (Olefeldt et al 2016). In the Northern Hemisphere, thermokarst lakes are hotspots of methane (CH₄) emission (Walter et al 2006) through multiple gas transport modes (Sepulveda-Jauregui et al 2015), and are estimated to emit 4.1 ± 2.2 Tg CH₄ per year (Wik et al 2016). Emissions from thermokarst lakes are expected to increase five-fold by 2100 (Schneider von Deimling et al 2015, Walter Anthony et al 2018). For better projections of CH₄ fluxes from high-arctic thermokarst lakes, we have to understand the molecular processes affecting CH₄ during its migration from deeper sediment layers to the atmosphere.

Evidence suggests a portion of CH₄ produced in thermokarst lakes is oxidized to the less potent greenhouse gas CO₂ prior to emission. A previous study of 30 Alaskan thermokarst lakes showed that aerobic CH₄ oxidation takes place in the water column of lakes in the Yedoma region, and is significantly higher in Yedoma lakes than in non-Yedoma thermokarst lakes (Martinez-Cruz et al 2015). Aerobic methanotrophs have also been identified to oxidize CH₄ in near surface sediments (He et al 2012, Martinez-Cruz et al 2017). Even though the process of aerobic CH₄ oxidation seems to be wide spread in thermokarst lakes, little is known about the responsible microorganisms that control the CH₄ mitigation in the oxygenated lake habitats.

Notably, mitigating processes in the form of anaerobic oxidation of methane (AOM) in the deeper anoxic thermokarst lake talik (thaw bulb) sediments have not been previously studied. However, recent findings of AOM-related microorganisms in deep submarine permafrost point towards a potential role in similar habitats (Winkel et al 2018). Hitherto, there is only one study on AOM in permafrost (Winkel et al 2018) besides detection of sequences from anaerobic methanotrophic (ANME) (Kao-Kniffin et al 2015, Shcherbakova et al 2016) in permafrost environments. It is important to understand both aerobic and anaerobic oxidation of methane in this high methane emitting ecosystems to calculate fluxes and completely project C budgets that can be integrate it into new climate models.

AOM in marine environments is commonly performed by (ANME) archaea of the clades ANME-1a/b, −2a/b, −2c, and −3 (Ruff et al 2015). AOM is often coupled to sulfate reduction, hence ANME are found in consortia with deltaproteobacterial sulfate-reducing bacteria of the genera Desulfococcus, Desulfosarcina, or Desulfobulbus (Knittel and Boetius 2009). Beyond marine sediments the occurrence and activity of AOM performed through archaea has mainly been shown in anthropogenic influenced freshwater habitats (Raghoebarsing et al 2006, Vaksmaa et al 2016), while reports for pristine environments are rare (Schubert et al 2011. Gupta et al 2013, Timmers et al 2015, Weber et al 2017). In pristine, terrestrial environments, ANME of the GoM Arc I/AOM-associated archaea/ANME-2d lineage likely perform AOM. Genomes of this cluster originating from Methanoperedens spp. enrichments showed genes for denitrifying-driven (Haroon et al 2013) and iron-driven AOM (Ettwig et al 2016). Marker gene sequences of Methanoperedenaceae were also detected in permafrost environments (Kao-Kniffin et al 2015, Winkel et al 2018), but their activity remains to be proven.

To understand if CH₄ mitigation by AOM occurs in proximity of the permafrost thaw front at cold, near-zero temperatures, we sought AOM and the associated organisms by analyzing sediment samples from a depth profile in a well-characterized thermokarst lake, Vault Lake (Heslop et al 2015, Sepulveda-Jauregui et al 2015, Martinez-Cruz et al 2017). Sediment pore water CH₄ concentrations and ¹³C–CH₄ ratios were determined along a ∼6 m sediment core, which extended through the talik into the underlying thawing permafrost. Further, we conducted potential AOM rate measurements of sediments using ¹³C–CH₄ isotopic tracer experiments and analyzed the in situ microbial community using 16S rRNA gene sequencing. We quantified anaerobic methanotrophic communities via quantitative PCR of the functional marker gene methyl-coenzyme M reductase, subunit alpha (mcrA) with specific primer for the Methanoperedenaceae cluster (Vaksmaa et al 2017). Additionally, ebullition bubbles emitted from the drilling boreholes and natural ebullition events were collected at Vault Lake and analyzed for CH₄ concentrations, isotopes (δ¹³C–CH₄ and δD), and radiocarbon age (¹⁴C–CH₄).

2.1. Sampling and physicochemical analysis

2.2. Concentration and stable isotope signatures of CH₄ measurements

2.3. Gas composition and isotopes from natural ebullition events and borehole bubbling

2.4. Anaerobic oxidation of methane (AOM) stable isotope incubation

2.5. DNA extraction, 16S rRNA gene Illumina HiSeq sequencing and analysis

2.6. Quantification of bacterial 16S rRNA genes, ANME-2d mcrA and methanogens mcrA

2.7. Calculating of CH₄ fraction that got oxidized based on ¹³C–CH₄ changes

We used highest changes in stable isotope signatures of dissolved CH₄ in the cores to calculate the fraction of produced CH₄ that got oxidized (f_ox,i). Therefore we used equation after Liptay et al (1998):

$$\begin{eqnarray*}&&{{\boldsymbol{f}}}_{{\boldsymbol{ox}},{\boldsymbol{i}}}=\displaystyle \frac{{{\boldsymbol{\delta }}}_{{\boldsymbol{o}}}-{{\boldsymbol{\delta }}}_{{\boldsymbol{p}}}}{{\bf{1000}}\,* \,({\propto }_{{\boldsymbol{ox}}}-\,{\propto }_{{\boldsymbol{trans}}})}\end{eqnarray*}$$

where δ_o and δ_p are the δ¹³C values (in ‰) of CH₄ in different horizons of the taberite layers. α_ox and α_trans are fractionation factors for AOM and CH₄ transport, respectively. Percentage of oxidized CH₄ is than calculated. (f_ox × 100).

We assumed CH₄ to be transported mainly by diffusion and therefore used a soil fractionation factor of α_trans = 1.001 after Preuss et al (2013). Different fractionation factors (α_ox) for AOM in freshwater system were used mainly for sulfate-dependent AOM (α_ox = 1.031) (Schubert et al 2011), iron-dependent AOM (α_ox = 1.030) (Norði et al 2013), and nitrate-dependent AOM (α_ox = 1.032) (Norði and Thamdrup 2014).

3.1. Pore water and bubble methane analysis

Radiocarbon dating of CH₄-rich bubbles from the borehole that reflects the transitional permafrost (mean CH₄: 79.82 ± 24.8%, CO₂: 0.81 ± 1.1%) revealed a C age of 21.1 ± 0.08 kyr, in comparison to CH₄ of bubbles from natural ebullition events (mean CH₄: 73.6 ± 23.1%, CO₂: 0.53 ± 0.4%) that showed a mixture of old and young carbon (mean: 9.7 kyr, range 2.2–28.5 kyr table 1).

Table 1.  Composition of CO₂, CH₄ and N₂ in gas bubbles and isotopic signatures of CH₄.

	Natural ebullition events			Bubbles released from deep boreholes following drilling
	mean	stdev	n	mean	stdev	n
CO2 (%)	0.53	0.38	35	0.81	1.12	9
CH4 (%)	73.61	23.11	35	79.82	24.79	9
N2 (%)	25.40	22.43	23	12.72	8.26	5
δ13C CH4 (‰)	−67.67	3.42	20	−72.67	4.35	4
δD CH4 (‰)	−382.90	12.43	20	−390.41	15.96	4
14C–CH4 (fM)	0.30	0.26	9	0.07		1
14C–CH4 (yrs BP)	9,715	2,190–28,500a	9	21,068	80	1

^aThe range of ¹⁴C–CH₄ measurements for natural ebullition events was given to account for the large variability.

δ¹³C–CH₄ and δD-CH₄ values of gas bubbles (n = 4) emitted from the base of the drilling boreholes (e.g. transitional permafrost facie) were highly depleted (−72.7 ± 4.4‰ and −390.4 ± 16.0‰, respectively, table 1). Bubbles collected from natural ebullition events showed slightly enriched δ¹³C–CH₄ values (−67.7 ± 3.4‰) but were still highly depleted in δD-CH₄ (−382.9 ± 12.4‰, table 1). Mean values are plotting in the mixed zone of microbial methane production (figure S1).

The transitional permafrost, located at the base of the radially expanding talik directly above the thaw boundary, showed the highest in situ sediment pore water CH₄ concentrations (mean 134 ± 8 μM) with depleted δ¹³C–CH₄ between −75‰ and −73‰ (figure 1(a)), similar to borehole methane values. The recently-thawed taberite showed a slight enrichment in δ¹³C–CH₄ (−69 to −50‰) and a decrease in CH₄ concentration (figure 1(a)). In striking contrast, the taberite layers, located above the recently-thawed taberite, were characterized by large δ¹³C–CH₄ fractionation (−63 to −40‰) and low CH₄ concentrations (down to 28 μM, figure 1(a)). Based on the range of AOM fractionation factors (1.030–1.032, Schubert et al 2011, Norði et al 2013, Norði and Thamdrup 2014) reported for freshwater systems, our calculations (Liptay et al 1998) of large isotopic changes (becoming more enriched in δ¹³C–CH₄) - expecting CH₄ to diffuse upwards - suggested that 41% to 83% of the in situ CH₄ fraction in the taberite and upper recently-thawed taberite layers (311 to 532 cm, figure 1(a)) got oxidized. With a few exception (right above the taberite), CH₄ concentration in the lacustrine and organic-rich facie scattered around 43 ± 19 μM with δ¹³C–CH₄ of −65.5 ± 1.8‰ (figure 1(a)).

Zoom In Zoom Out Reset image size

3.2. Potential AOM rates and possible electron acceptors

3.3. Microbial community structure

Microbial community analyses revealed high archaeal diversity (highest invers Simpson 6.33 to 7.90, table S3) at the permafrost thaw front (i.e. transitional permafrost and base of recently-thawed taberite) that cluster together with archaeal community of the surface sediment (figure S2(a)) and included many representative OTUs related to methanogens (Methanosaeta 12%–68%, Methanosarcina 9%–16%, and Methanoregulaceae 3%–18%). The transitional permafrost showed the highest occurrence of Planctomycetes sequences of all bacterial communities (figure 2). Other bacterial sequences fell into dominant taxa of Alphaproteobacteria (Caulobacteraceae, Rhizobiales, Rhodospirillaceae, Sphingomonadace), Betaproteobacteria (Burkholderiales), Saccharibacteria (TM7), Firmicutes (Bacillales), Actinobacteria (Actinomycetales, Acidimicrobiales), and Chloroflexi (Ellin6529). The bacterial communities of the transitional permafrost cluster together with most surface sample (10–20 cm) (figure S2(b)). In contrast to the archaea, the bacterial communities of all taberite layers cluster together (figure S2(b)) and show an increase in taxa of Betaproteobacteria (Rhodocyclaceae/Dechloromonas) and different Firmicutes. For detailed microbial descriptions, please see the supplementary information.

Zoom In Zoom Out Reset image size

The archaeal communities of the taberite layer cluster together were almost exclusively represented by ANME-2d sequence of the Methanoperedenaceae (figure 1(c)). Phylogenetic analysis of the Methanoperedenaceae-related OTUs showed an affiliation with two different freshwater clusters (figure 3) mainly comprised of sequences from lake, river, permafrost, iron-rich mat habitats and a Methanoperedens sp. that uses nitrate and iron as electron acceptor (Ettwig et al 2016). Quantification of Methanoperedenaceae (ANME-2d) using specific primers for their functional mcrA, revealed 5.7–6.7 × 10⁵ ± 0.7–1 × 10⁵ copy numbers g⁻¹ wet weight. While bacterial 16S rRNA gene copy numbers in the taberite layer showed similar values (table S2), methanogenic mcrA copy numbers were below the detection limit (<10² copy numbers g⁻¹ wet weight, table S5) and methanogenic 16S rRNA sequencing revealed only poor contribution to the overall archaeal abundance (<0.02%). Additional analysis of a methane seep and a mud volcano from two other lakes revealed a low abundance (∼1%) of Methanoperedenaceae sequences (figure S3).

Zoom In Zoom Out Reset image size

Vault Lake is a thermokarst lake in the discontinuous permafrost region near Fairbanks Alaska, USA. It is a typical first generation thermokarst lake formed by melting of permafrost ground ice. It has massive ice wedges (figure 4), and steep eroding bluffs and several CH₄ seeps that indicate actively deepening. Radiocarbon dating from borehole ebullition clearly reveals a Pleistocene age (∼ 21 kyr) of C–CH₄ falling into the typical range of point sources from thermokarst lakes (Walter et al 2007). In contrast, the age of natural ebullition seeps showed a mixture (2 to 28 kyr) of different ages most likely reflecting the production from different aged C sources and, therefore, ebullition from different sediment depths (figure 4) falling into the range of background and point sources (Walter et al 2007).

Zoom In Zoom Out Reset image size

Bubbles from the borehole showed highly depleted δ¹³C–CH₄ and δD that point towards biological CH₄ production (Whiticar 1999). Even though δ¹³C–CH₄ values indicate hydrogenotrophic methanogenesis, the highly depleted δD values are atypical and might indicate an unusual substrate (e.g. short chain alkanes, Borrel et al 2019), be an effect of higher partial pressure (Bilek et al 2001), or reflect a different methanogenic pathway (e.g. methylotrophic methanogenesis, Sorokin et al 2017). Pore water CH₄ from the transitional permafrost, which reflects ice-bearing sediments with a large quantity of unfrozen water in the inter-pore space, had similar δ¹³C–CH₄ values to the borehole ebullition gas. Since the transitional permafrost also showed the highest CH₄ concentration, it indicates that this might be a horizon of high in situ CH₄ production due to degrading permafrost providing methanogens with substrates. Another possible explanation might be that previously produced CH₄ was entrapped during permafrost formation in the Pleistocene and released through thawing. However, based on the cold, dry steppe conditions during Yedoma formation at this site (Heslop et al 2015) and low abundance of methanogens in undisturbed Yedoma sediments (Bischoff et al 2013, Rivkina et al 2016), we interpret the high concentrations of pore water dissolved CH₄ to be the result of present-day methanogenesis at the thaw front. The depleted δ¹³C–CH₄ showed similar δ¹³C–CH₄ values (−70.9 to −73‰) to hydrogenotrophic methanogenesis of peatlands (Avery et al 1999, Galand et al 2010, McCalley et al 2014), however, missing δD values from the pore water analysis prevents further interpretation in comparison to methane from the borehole bubbles.

Moving from the transitional permafrost towards recently-thawed taberite, the archaeal community had increasing proportions of methanogens supporting recently found CH₄ production in the same sediment layers (Heslop et al 2015). CH₄ production occurred most likely from precursors such as acetate and formate as well as H₂/CO₂. Acetate and formate are known to occur in high concentration in Pleistocene sediments (Strauss et al 2015, Mitzscherling et al 2017), especially in Yedoma regions (Ewing et al 2016). CO₂ concentrations measured at the same depth showed values between 120 to 1000 mM (A. Sepulveda-Jauregui, unpublished data). Some bacterial taxa (Planctomycetes and Alphaproteobacteria) in the transitional permafrost, which are often found in peatlands, can be related to aerobic lifestyles and could point to relicts from conditions during the Pleistocene, while many taxa such as Burkholderiales, Saccharibacteria, Myxococcales, Actinomycetales, and Chloroflexi are related to species that are known to anaerobically hydrolyze plant polymers as the first step of organic matter degradation (figure 4). Interestingly, the bacterial communities of the transitional permafrost cluster together with the sample nearest to the surface (10–20 cm, figure S2(b)) that are influenced by fresh organic matter input from lake primary production during summer (Martinez-Cruz et al 2015).

The slight increase in δ¹³C–CH₄ values in the pore water CH₄ in the recently-thawed taberite might be due to availability of other substrates such as acetate and could also explain the increase in Methanosaeta (figure 1(c)), which solely uses acetate as substrate (Michał et al 2018). In all taberite layers, we also detected an increase in bacterial groups that are known fermenters (Betaproteobacteria/Rhodocyclaceae/Dechloromonas, Firmicutes/Clostridiales) and syntrophs (Ruminococcaceae, Syntrophobacterales, Clostridiaceae) known to degrade monomers into acetate, hydrogen, carbon dioxide, and formate as precursors for CH₄ production (figure 4). Compared to organic decomposition processes in typical active layers (Tveit et al 2012), where fresh material is broken into polymers and monomers and then hydrolyzed and fermented into methanogenic precursors with depth, permafrost organic matter in thermokarst lake talik is decomposed by metabolic processes that occur in the inverse depth order (figure 4). For detailed microbial descriptions, please see the supplementary information.

The taberite facies also showed the most fractionated δ¹³C–CH₄, with enriched CH₄ up to 40‰ pointing towards oxidation (Whiticar 1999) and reaching similar values as reported for AOM in other freshwater sediments (Schubert et al 2011, Norði et al 2013, Norði and Thamdrup 2014). Our calculated fraction of CH₄ that was oxidized in the taberite facie is consistent with previously reported CH₄ measurements and isotopic values from other Alaskan lakes that also assumed CH₄ oxidation, but were mainly attributed to aerobic oxidation (Elder et al 2018). Our highest measured potential AOM rates in the taberite suggests this layer is a hotspot for AOM. The archaeal sequence analysis supports this findings as we almost exclusively found Methanoperedenaceae known for nitrate-driven (Haroon et al 2013) or iron-driven AOM (Ettwig et al 2016, Cai et al 2018). To rule out any sequencing artifact, we analyzed the quantity of the functional mcrA, which revealed similar copy numbers to the entire bacterial community and were in the range of river sediments (Vaksmaa et al 2016). Methanogens in this horizon were below the detection limit (<10² copy numbers), even though long-term incubations (>115 days) indicated a reactivation of methanogens (Heslop et al 2015). Measurements for potential electron acceptors revealed no detectable nitrate, in the taberite layers, which might indicate low concentrations in the pore water that could not support AOM or high turnover of nitrate. Interestingly, high nitrite concentrations in the interface of transitional and recently-thawed permafrost could point towards incomplete denitrification of AOM, since Methanoperedenaceae sequences were also found in lower abundances in these layers (figure 1(c)). We also detected unusually high sulfate concentrations in the taberite, while potential sulfate-reducing microorganisms were low in abundance (up to 0.7% relative abundance, data not shown). It is therefore not clear if sulfate is used as electron acceptor coupled to AOM. Other electron acceptors for AOM in freshwater sediments are metal oxides of iron (Weber et al 2017) and manganese that could support Fe(III)-dependent AOM (Ettwig et al 2016, Cai et al 2018). No pore water iron concentration or reactive iron solid phases have been measured for the Vault Lake sediment. However, pore water measurements in other thermokarst lakes showed high Fe(III) concentrations at the thaw boundary (Winkel, unpublished data). Therefore, iron could be involved in AOM by Methanoperedenaceae, as recently been shown (Ettwig et al 2016, Cai et al 2018), and might play a role in deeper sediments (Egger et al 2015).

Potential AOM rates decreased moving towards the surface of the core but started to increase again at 100 cm depth (figure 1(b)), pointing towards a second zone of AOM in near-surface sediments (generalized dashed line, figure 4). AOM activity in surface sediment layers of the Vault Lake was recently ascribed by Martinez-Cruz and colleagues (Martinez-Cruz et al 2017) to an entirely different microbial community, namely aerobic methanotrophs consuming CH₄ under anaerobic conditions. Our data support this hypothesis, since we did not find any ANME or NC10 phylum microorganisms (Ettwig et al 2010) that would support a typical AOM community but found low abundances of aerobic methanotrophs (Crenotrichaceae, Methylophilaceae and Methylocystaceae) similar to Martinez-Cruz and colleagues (Martinez-Cruz et al 2017). However, presently undiscovered anaerobic methanotrophs could also be responsible for the second AOM peak. The δ¹³C–CH₄ (mean −65.5 ± 1.8‰) in the lacustrine silt and organic-rich mud facies were similar to the natural ebullition event bubbles, despite a few layers with low (lacustrine silt) or high concentrations (organic-rich mud).

We found that AOM occurs in thermokarst lake sediments and mitigates diffusive CH₄ emission from these lakes. Previous work suggests that CH₄ primarily escapes from talik sediments as bubbles through secondary pore channels (i.e. bubble tubes) (Sepulveda-Jauregui et al 2015). This is largely due to low hydraulic conductivity of the silt-dominated Yedoma sediments, which inhibits diffusion of dissolved CH₄ and therefore traps dissolved CH₄ in the sediment pore water where it is subject to anaerobic microbial oxidation. As a result, Methanoperedenaceae anaerobically oxidize CH₄ at rates similar to other lake sediments (Knittel and Boetius 2009) effectively filtering for dissolved CH₄ concentrations despite the near 0 °C temperature conditions. Recent findings on AOM in degrading submarine permafrost that also showed natural enrichments in Methanoperedenaceae-related communities (Winkel et al 2018) support the findings of cold-adapted anaerobic methanotrophic communities. Methanoperedenaceae in thermokarst lakes were mainly found in deep sediment layers and mud volcanos that represents windows to the subsurface (Ruff et al 2018). AOM in thermokarst lake sediments may play an important role in mitigating the release of ancient CH₄ as permafrost warms and thaws beneath lakes (Walter Anthony et al 2018). Since our analyses were done on one sediment core and the study lacks certain chemical and physical background data, additional thermokarst lakes need to be investigated in the future to understand if such a mitigation process along the deep permafrost thaw front is widespread. In particular, what kind of electron acceptor is used in the process of AOM should be studied to understand the molecular process in freshwater thermokarst lake environments with many possible electron acceptors such as metal oxides, nitrate and humic acids.

We thank Sam Skidmore for granting access to Vault Lake and Anke Saborowski for laboratory support. Funding for M Winkel and K M Walter Anthony was provided by the National Science Foundation ARCSS-1500931. K Martinez-Cruz was funded by Conacyt (Grant no. 330197/233369). The Helmholtz Young Investigators Group of S Liebner is funded by the Helmholtz Gemeinschaft (HGF) (VH-NG-919).

We deposited sequences of the thermokarst lake permafrost metagenome at the NCBI Sequence Read Archive (SRA) with the Project number BioProject ID# PRJNA381521, accession numbers for archaeal 16S rRNA gene sequences were SRX3047230- SRX3047235 and for bacterial 16S rRNA gene sequences were SRX3047223- SRX3047228.