Volume 125, Issue 3 e2019JG005526
Research Article
Free Access

Thaw Transitions and Redox Conditions Drive Methane Oxidation in a Permafrost Peatland

Clarice R. Perryman

Corresponding Author

Clarice R. Perryman

Department of Earth Sciences, University of New Hampshire, Durham, NH, USA

Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, USA

Correspondence to: C. R. Perryman,

[email protected]

Search for more papers by this author
Carmody K. McCalley

Carmody K. McCalley

Thomas H. Gosnell School of Life Sciences, Rochester Institute of Technology, Rochester, NY, USA

Search for more papers by this author
Avni Malhotra

Avni Malhotra

Department of Earth System Science, Stanford University, Stanford, CA, USA

Search for more papers by this author
M. Florencia Fahnestock

M. Florencia Fahnestock

Department of Earth Sciences, University of New Hampshire, Durham, NH, USA

Search for more papers by this author
Natalie N. Kashi

Natalie N. Kashi

Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, USA

Search for more papers by this author
Julia G. Bryce

Julia G. Bryce

Department of Earth Sciences, University of New Hampshire, Durham, NH, USA

Search for more papers by this author
Reiner Giesler

Reiner Giesler

Climate Impacts Research Centre, Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden

Search for more papers by this author
Ruth K. Varner

Ruth K. Varner

Department of Earth Sciences, University of New Hampshire, Durham, NH, USA

Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, USA

Search for more papers by this author
First published: 05 March 2020
Citations: 16


Permafrost peatlands are a significant source of methane (CH4) emissions to the atmosphere and could emit more CH4 with continued permafrost thaw. Aerobic methane-oxidizing bacteria may attenuate a substantial fraction of CH4 emissions in thawing permafrost peatlands; however, the impact of permafrost thaw on CH4 oxidation is uncertain. We measured potential CH4 oxidation rates (hereafter, CH4 oxidation) and their predictors using laboratory incubations and in situ porewater redox chemistry across a permafrost thaw gradient of eight thaw stages at Stordalen Mire, a permafrost peatland complex in northernmost Sweden. Methane oxidation rates increased across a gradient of permafrost thaw and differed in transitional thaw stages relative to end-member stages. Oxidation was consistently higher in submerged fens than in bogs or palsas across a range of CH4 concentrations. We also observed that CH4 oxidation increased with decreasing in situ redox potential and was highest in sites with lower redox potential (Eh < 10 mV) and high water table. Our results suggest that redox potential can be used as an important predictor of CH4 oxidation, especially in thawed permafrost peatlands. Our results also highlight the importance of considering transitional thaw stages when characterizing landscape-scale CH4 dynamics, because these transitional areas have different rates and controls of CH4 oxidation relative to intact or completely thawed permafrost areas. As permafrost thaw increases the total area of semiwet and wet thaw stages in permafrost peatlands, CH4 oxidation represents an important control on CH4 emissions to the atmosphere.

Plain Language Summary

Thawing permafrost, ground that has been frozen for 2 or more years, often results in wet conditions that promote the production of methane (CH4). Methane is a greenhouse gas that is over 30 times more powerful than carbon dioxide at warming the atmosphere. Microorganisms called methanotrophs can consume a substantial portion of this CH4 before it reaches the atmosphere, but the extent of this consumption is not well understood. Methanotrophs are affected by environmental conditions, such as oxygen availability, that change with permafrost thaw. We used laboratory experiments to measure how quickly methanotrophs consume CH4 from sites within a permafrost peatland with little, intermediate, and full permafrost thaw. We found that CH4 consumption increased as permafrost thawed and that sites transitioning between thaw stages had a different potential to consume CH4 than intact and fully thawed sites. Methane consumption was also high in locations with low dissolved oxygen, which suggests that methanotrophs in these sites may tolerate low oxygen conditions in the field. We found that redox potential, a measurement that describes the tendency for a compound to be oxidized or reduced, was a strong predictor of CH4 consumption in partially and fully thawed sites. These measurements highlight the potential for methanotrophs to mitigate CH4 emissions in peatlands with thawing permafrost and provide new insights to modeling CH4 dynamics across the full thaw gradient.

Key Points

  • Potential methane oxidation rates increase across a gradient of permafrost thaw from intact permafrost palsa to submerged fen
  • Methane oxidation rates and controls differed between transitional and end-member thaw stages
  • Redox potential (Eh) explains greater variance in methane oxidation rates than dissolved O2 or water table depth

1 Introduction

Permafrost thaw in permafrost peatlands can cause palsa mounds with intact permafrost to subside into submerged fens (Malmer et al., 2005), increasing methane (CH4) emissions as the water table rises above the peat surface and active layer depth and graminoid vegetation cover increase (Johnston et al., 2014; Malhotra & Roulet, 2015; Turetsky et al., 2002). These changes are associated with shifts in the abundance and activity of CH4-cycling microbes, including aerobic methane oxidizing bacteria (MOB; Singleton et al., 2018; Woodcroft et al., 2018). Methane oxidation may attenuate a significant fraction of potential CH4 emissions from peatlands (Frenzel & Karofeld, 2000; Nielsen et al., 2019), and inhibition and isotopic studies indicate that 0–50% of CH4 in northern peatlands is oxidized by aerobic MOB before emission to the atmosphere (Moosavi & Crill, 1998; Popp et al., 2000; Preuss et al., 2013). Recent work indicates that anaerobic methanotrophs may also attenuate a substantial portion of CH4 emissions in northern peatlands (Miller et al., 2019); however, the significance of this oxidation pathway to peatland CH4 cycling globally is less clear than that of aerobic methanotrophy. Thus, as permafrost is thawing due to atmospheric warming at a global scale (Olefeldt et al., 2016; Turetsky et al., 2019), it is vital to understand the potential of MOB to attenuate CH4 emissions from thawing permafrost. Previous work comparing the abundance/activity of MOB using intact and thawed sites (Liebner & Wagner, 2007; Vaughn et al., 2016) or through in vitro experiments thawing permafrost cores (Mackelprang et al., 2011; Voigt et al., 2019) suggests that permafrost thaw can increase CH4 oxidation. However, these studies focused on comparing end-member thaw states, which omit CH4 oxidation in transitional thaw stages that can cover a considerable area of thawing permafrost peatlands (Malmer et al., 2005; Palace et al., 2018). Given the role of CH4 oxidation in controlling peatland CH4 emissions and the increasing areal coverage of transitional thaw stages in thawing peatlands, it is important to quantify CH4 oxidation in transitional thaw stages to better project CH4 flux from permafrost peatlands.

Hydrology and vegetation vary across transitional thaw stages (Malhotra et al., 2018; Malhotra & Roulet, 2015) and influence redox potential (Eh) and rates of CH4 transformation and emission (Street et al., 2016; Svensson & Rosswall, 1984). For instance, in intermediately thawed bogs, rates of CH4 oxidation are predicted to be highest at the oxic-anoxic interface near the water table, where both CH4 and O2 are available for MOB (Kettunen et al., 1999; Moore & Dalva, 1997; Nedwell & Watson, 1995); however, isotopic evidence and the relative abundance of MOB lineages suggest substantial CH4 oxidation in peat that is frequently below the water table (Singleton et al., 2018). Whereas in fully thawed sites with higher graminoid abundance, aerenchymous vegetation can enable CH4 oxidation below the water table by releasing O2 through roots (Popp et al., 2000; Watson et al., 1997). Previous work that compares intermediately thawed bogs and fully thawed, saturated sites with aerenchymous vegetation indicates higher MOB activity in fully thawed sites (Singleton et al., 2018; Whalen & Reeburgh, 2000). However, comparing only the intermediately thawed and fully thawed stages disregards the variability of vegetation, hydrology, and redox conditions found in transitional thaw stages that are abundant and expanding areas of permafrost peatlands (Johansson et al., 2006; Malmer et al., 2005) contributing to landscape-level CH4 flux (Malhotra & Roulet, 2015). Furthermore, while aerobic CH4 oxidation is primarily controlled by O2 and CH4 availability (Hornibrook et al., 2009; Lofton et al., 2014), the availability of alternative terminal electron acceptors (TEAs) may impact MOB by providing substrates for obligate methanotrophs in oxygen-limited environments (Kits et al., 2015; Skennerton et al., 2015), inhibiting methanogenesis (Gao et al., 2019; Knorr & Blodau, 2009; Nedwell & Watson, 1995), and buffering redox potential as other microbes oxidize and reduce TEAs (Fiedler et al., 2007). As several hydrological (Knorr et al., 2009; Street et al., 2016), vegetation (Rupp et al., 2019; Svensson & Rosswall, 1984), and geochemical (Yu & Böhme, 2007) factors affect redox conditions in partially and fully thawed sites, characterizing the effect of these interacting factors on CH4 oxidation in transitional and end-member thaw stages may improve the ability to predict rates of in situ oxidation.

Quantifying CH4 oxidation rates across a full permafrost thaw gradient with changing redox conditions is crucial to better understand drivers of methanotrophy and to predict changes in CH4 oxidation with future permafrost loss. In this paper, we characterize variation in CH4 oxidation rates and redox chemistry across a thaw gradient comprised of eight thaw stages in a permafrost peatland undergoing rapid permafrost degradation. We used laboratory incubations to quantify CH4 oxidation rates across sites ranging from intact to collapsing palsas, Sphagnum spp.-dominated semiwet and dry bogs, and submerged sites dominated by sedge species. Incubation-derived rates of CH4 oxidation quantify oxidation potential under oxygenated conditions and therefore can overestimate in situ rates; however, these measurements are useful as they can provide insight into where MOB are active across the thaw gradient. We assessed changes in redox chemistry, including dissolved CH4 and O2, concentrations of alternative TEAs (e.g., sulfur and iron), and Eh, in porewater and the relationship between redox chemistry and CH4 oxidation. We hypothesized that CH4 oxidation would increase with increasing thaw despite more reducing in situ conditions (lower Eh, dissolved O2, and higher water table) due to high CH4 availability and the oxidizing capacity of the rhizosphere in sedge-dominated fully thawed sites.

2 Materials and Methods

2.1 Site Description

Stordalen Mire (68°21′N, 19°02′E) is a thawing permafrost peatland complex located 11 km east of Abisko, Sweden. Mean annual temperature (1912–2009) in Abisko is −0.5 °C with approximately 300 mm of annual precipitation (Olefeldt et al., 2012). Smoothed average annual temperature trends indicate an increase of 2.5 °C between 1913 and 2006 (Callaghan et al., 2010), and since 2000, the mean annual temperature consistently exceeded 0 °C (Kuhn et al., 2018). The landscape at Stordalen Mire is highly heterogeneous, representing sites that range from intact permafrost to no permafrost, containing palsas, semiwet Sphagnum spp. moss dominated sites, thaw ponds, submerged sedge-dominated sites, and shallow post-glacial lakes (Burke et al., 2019; Malmer et al., 2005; Wik et al., 2013). Methane emissions are variable across the mire, ranging from −1.3 ± 0.21 mg CH4 · m−2 · day−1 in intact palsa to 370.2 ± 52.1 mg CH4 · m−2 · day−1 in open water sedge sites with no permafrost (Malhotra & Roulet, 2015). Increases in semiwet Sphagnum spp. and sedge cover have increased CO2 uptake and CH4 emissions at Stordalen (Bäckstrand et al., 2010; Christensen et al., 2004). While Stordalen is a net CO2 sink, increases in CH4 emissions have intensified the mire's radiative forcing between 1970 and 2000 (Bäckstrand et al., 2010; Johansson et al., 2006).

2.2 Field Sampling and Measurements

Peat cores and environmental measurements were collected at eight sites representing the full thaw gradient at Stordalen Mire. Sites were classified as either palsa, bog, or fen by their vegetation, thaw depth, and water table depth. Site names reflect site classification (P for palsa, B for bog, and F for fen) and are numbered sequentially within each classification to differentiate between sites. Sites P3 and BF are transitioning thaw stages; P3 is a collapsing palsa, and BF is a bog-fen transition with submerged Sphagnum spp. and sedges. While several classification schemes have been used for work at Stordalen Mire, we adopted the classification used by previous microbial ecology (McCalley et al., 2014; Singleton et al., 218; Woodcroft et al., 2018) and modeling work (Deng et al., 2014, 2017) rather than remote sensing (Johansson et al., 2006; Malmer et al., 2005; Palace et al., 2018) as we expect our data will be of greater interest to these fields (see Table S6 in the supporting information for how our classifications relate to others). Table 1 describes vegetation and abiotic factors at each site. The eight sampling sites are a subset of the sites used in Malhotra and Roulet (2015) to measure carbon fluxes and Malhotra et al. (2018) to measure litter decomposition at Stordalen Mire.

Table 1. Peat Sampling Site Descriptions
Site Vegetation descriptiona Thaw depth in July (cm) Water table depth (cm) Temperature at 10 cm (°C) pH Upper core depth (cm) Lower core depth (cm)
P1 Intact palsa, Dicranum elongatum, Vaccinium uliginosum 24 ± 2.7 - 5.6 ± 0.7 - 0–7 7–14
P2b Thawing palsa, D. elongatum, Eriophorum vaginatum 35 ± 2.4 - 9.6 ± 0.4 - 0–7 7–14
P3 Collapsed palsa, saturated Sphagnum fuscum 38 ± 2.0 - 11.1 ± 1.1 - 0–7 7–14
B1b Sphagnum lawn 67 ± 14.5 −7 ± 2.2 8.8 ± 0.8 3.50 ± 0.03 0–8 8–15
B2 Sphagnum lawn, Betula nana 59 ± 21.5 - 7.6 ± 1.7 - 0–7 7–15
BF Submerged Sphagnum lawn, E. vaginatum, S. cuspidatum 53 ± 1.4 2 ± 0.7 11.7 ± 1.1 3.80 ± 0.15 0–10 10–18
F1b E. angustifolium, open water > 100 6 ± 0.3 11.4 ± 1.4 3.40 ± 0.19 0–10 10–20
F2 C. rostrata, open water 65 ± 4.9 11 ± 1.3 11.5 ± 0.2 4.90 ± 0.08 0–10 10–20
  • Note. Values are reported as mean ± standard deviation (n = 3). Water table depth is reported as the distance between the water surface and peat surface. Positive values indicate fully submerged peat; negative values indicate that the water table is below peat surface. Hyphens indicate no water saturation in the peat profile. P = palsa; B = bog; BF = bog-fen transition; F = fen.
  • a Vegetation descriptions adapted from Malhotra and Roulet (2015).
  • b Sampled for incubations with variable headspace CH4; sampling depths: 5–12 cm (P2), 6–13 cm (B1), and 2–10 cm (F1).

We extracted one set of triplicate cores ranging in depth from 15 to 20 cm from each site and split each core into two sections. For sites where the water table fell within the core depth, we split cores at the water table. Additional sets of triplicate peat cores were collected at sites P2, B1, and F1 for additional incubation experiments in 2017 (15-cm depth) and in 2018 (only B1 and F1, 30-cm depth split into three depths). Thaw depth at the time of sampling was measured within 30 cm from each core location using a 1-m metal rod inserted into the peat surface until reaching the depth of permafrost. Thaw deeper than 1 m was reported as >100 cm.

In sites with surface water, we measured Eh, pH, and dissolved O2 in situ in accordance with peat sampling depths at each coring location in 2017. For each measurement, the respective electrode was inserted into the peat to 10 cm and then 20 cm below peat surface to collect measurements. Redox potential was measured with an Orion Star A221 oxidation-reduction potential meter paired with a gel-filled oxidation-reduction potential/automatic temperature compensation electrode (9179BNMD) with a platinum redox sensor and a silver/silver chloride internal reference system. Calibration and field measurements of redox potential were conducted using the “relative mV” mode that converts raw voltage into the corresponding Eh (in mV) values that would be obtained using a standard hydrogen electrode. Redox measurements were further corrected for pH by standardizing to a pH of 7 by subtracting 58 mV for every pH unit below 7 (Wetzel, 2001). Dissolved O2 was measured with a Hanna Instruments portable dissolved oxygen meter (HI9142), and pH was measured with a YSI Model 60 handheld pH meter. The electrode used to measure redox potential was calibrated using a 220-mV standard solution. During redox calibration, an offset was calculated as the difference between the value of the standard and the final reading during probe calibration. The values reported for porewater redox measurements were corrected by the calculated Eh offset. Dissolved O2 and pH meters were calibrated with a two-point calibration using standard solutions provided by the manufacturers. Final measurements were recorded after a settling period of several minutes and/or when the corresponding meter stabilized on a final reading. All probes used to collect porewater measurements were calibrated prior to each sample collection day. All field sampling occurred in mid-July 2017 and 2018.

2.3 Incubations

We stored peat cores at 5 °C in sample bags without headspace air between field sampling and incubation. All incubations were started within 24 hr of coring. For incubations, peat cores were weighed in sample bags and then sealed in 350-ml glass jars. Peat filled approximately half of the jar volume. The jar lids were fitted with a 1/8″ brass bulkhead union (Swagelok Company, Solon, Ohio, USA) and silicon septa for headspace gas sampling. Blank controls indicated that the incubation vessels did not leak (Figure S1a in the supporting information). Incubation jars were spiked with 9 ml of 8.0% CH4 at the beginning of the incubation period to achieve an initial headspace concentration of approximately 4,000 ppm CH4. Pilot incubations indicated that at a headspace concentration of 4,000 ppm, CH4 uptake had reached saturation with respect to CH4 (Figures S1bS1d). Incubations ran for 24 hr at 10 °C in dark incubators.

We conducted an additional two sets of incubations to assess the response of CH4 oxidation to changes in headspace CH4 concentration both across the thaw gradient and with depth. The first set examined how the response to increasing CH4 concentration varied between palsa (P2), bog (B1), and fen (F1) sites. This incubation had six different initial headspace concentrations ranging from 200 to 8,000 ppm CH4. The second set of incubations assessed how the response to increased CH4 varied with depth in B1 and F1 and had five different initial headspace concentrations ranging from 200 to 15,000 ppm CH4 in accordance with elevated in situ dissolved CH4 at depth in these sites. These incubations also ran for 24 hr at 10 °C in dark incubators.

To determine CH4 oxidation rates, 10 ml of headspace gas was sampled with a polypropylene syringe equipped with a three-way stopcock at the start of the incubation period and 4, 8, 12, and 24 hr into the incubation. Incubation jars were injected with 10 ml of ambient air after each sampling to maintain a constant pressure in the incubation jars. The volume of headspace air was corrected to account for the dilution caused by ambient air injections. After the incubation period, the cores were dried (40 °C, 72 hr) and the dry weight was used to normalize CH4 oxidation to per gram dry weight of peat. Headspace gas samples were analyzed for CH4 concentration on a gas chromatograph (GC-FID; Shimadzu GC-14-A) equipped with a flame ionization detector, a 2-m 80/100 mesh HayeSep-Q packed column, and a 200-μl injection loop. The GC-FID was run with a detector temperature of 180 °C, a column temperature of 50 °C, and nitrogen carrier gas at a flow rate of 50 ml/min. The detector response was calibrated with a 2,010 ppm or 2.5 ppm CH4 standard in N2 depending on the initial concentration of the incubations.

Methane oxidation was determined from the linear decrease in CH4 concentration over 12–24 hr of incubation time. Headspace CH4 concentration decreased nonlinearly after 12 hr for some incubations, and the 24-hr sampling point was omitted. We restricted our CH4 oxidation rate calculations to the linear decrease in headspace CH4 because linear uptake indicates substrate saturated conditions under which CH4 is not limiting (Sundh et al., 1995; Whalen et al., 1992), therefore allowing for the measurement of maximum oxidation potential under the given CH4 concentration. While only a subset (<20%) of the incubations had nonlinear changes in headspace concentration within 24 hr, >50% became nonlinear after 24 hr of incubation, and therefore, incubation time was restricted to 24 hr for calculation of oxidation rates for consistency. Oxidation rates (MethOx) were normalized to the dry weight of the corresponding peat cores. Rates were converted from ppm CH4/hr to μg CH4 · gdw−1 · hr−1 for data analysis using:
where ∆ is the change in CH4 concentration (in ppm) per hour, P is pressure in atm, R is the ideal gas constant (in L atm · mol−1 · K−1), T is the incubation temperature in K, V is the headspace volume of the incubation jar in L, and gdw is the dry weight of peat.

2.4 Porewater Chemistry

In saturated sites, porewater samples were collected at 10 and 20 cm below peat surface using a stainless steel sipper. These depths were selected to align with peat sampling depths in sites with porewater. Three replicate water samples were collected at each depth for all porewater analyses. Aliquots (30 ml) of porewater were equilibrated with 30 ml of ambient air in a 60-ml syringe by shaking each syringe for 2 min (McAuliffe, 1971) and then were analyzed using GC-FID for CH4 concentration as described above. Sampling, equilibration, and analysis of porewater for dissolved CH4 were all conducted in the same day. Aliquots of porewater (~10 ml) were filtered through 0.7-μm glass fiber syringe filter then frozen and stored in a 15-ml high-density polyethylene vial for dissolved nitrate and sulfate content. An additional 10 ml was frozen and stored in acid-washed 15-ml high-density polyethylene vials for trace metal analysis. Bulk Eh and pH conditions in the porewater of Stordalen Mire suggest that some dissolved N may exist as in situ NH4+ and dissolved S as SO4−2 or H2S. Though steps were taken to minimize postsampling oxidation, porewater collection and subsequent freezing for storage took place under oxygenated conditions, and accordingly, some nitrification and/or H2S oxidation may have occurred between sampling and analysis.

Dissolved nitrate (NO3-N) and sulfate (SO4-S) concentrations were determined using ion chromatography (IC) following EPA Method 300.1 (U.S. EPA, 1997). Samples were filtered a second time through a 0.45-μm filter prior to IC analysis. Porewater aliquots measured for metal (manganese and iron) contents were acidified with concentrated triple-distilled HNO3 to 2% HNO3, filtered through 0.7- and 0.22-μm polypropylene glass fiber syringe filters, and amended with H2O2 to bring trace metals into solution and remove colloidal organic matter prior to analysis. Porewater Mn and Fe values may include some particulate-bound Mn and Fe that was brought into solution through acidification prior to filtering and therefore are not referred to as dissolved. Full procedural blanks were undertaken to constrain any contamination introduced in sample handling. Syringes and filters were cleaned with 1N HCl, 2M HNO3, and 0.2% HNO3 prior to sample introduction to minimize contamination from the filtering process. Porewater Mn and Fe concentrations were measured with a Nu Instruments AttoM high-resolution inductively coupled plasma mass spectrometer. Instrument drift was corrected by applying an external drift correction from monitor solution analyzed every seven samples. Quality-control reference materials (USGS T193, T221, and T217) were analyzed to assess precision and accuracy. Values for standards fell within 2σ of accepted values (Table S4).

2.5 Statistical Analyses

Univariate statistical analyses were performed using the R Project for Statistical Computing (v. 3.3.3) software. Significance of all statistical analyses were determined at the α = 0.05 level. Oxidation rates and porewater chemistry data were checked for their homoscedasticity and normality using the Bartlett and Shapiro-Wilk tests. Analysis of variance (ANOVA) and the Tukey honestly significant difference test were used to assess the effects of thaw stage and measurement depth on CH4 oxidation rates. Differences in porewater chemistry were assessed using ANOVA for data meeting the assumptions of homoscedasticity and normality and the Kruskal-Wallis test for nonnormal data. To assess the effect of increasing CH4 headspace concentration on methanotrophic activity, linear fits and slopes of CH4 oxidation and CH4 concentration from the incubations in variable CH4 conditions were determined using linear regression.

Partial least squares (PLS) regression was used to identify factors that maximized covariance between porewater chemistry measurements and CH4 oxidation rates from the four wet sites. A PLS model using the nonlinear iterative partial least squares method was constructed in JMP 14.0 using water table depth, Eh, dissolved O2, porewater CH4, pH, dissolved NO3-N and SO4-S, and porewater Mn and Fe as predictor variables, and measured CH4 oxidation rates as the response variable. Water table depth was included as a predictor variable in the PLS model as distance to water table, calculated as the difference between the measured water table depth and the depth of the incubated peat sample, to more accurately describe the relationship between the water table and sample depths. The PLS model was validated using leave-one-out validation, a method best suited for smaller datasets like the porewater chemistry data used in the PLS analysis (24 sampling sites). Predictor variables and CH4 oxidation rates were centered and scaled in the model to minimize the effect of different measurement scales. Pearson's correlation coefficient (r) was used to assess correlations between strongly loading predictor variables and CH4 oxidation rates.

3 Results

3.1 Methane Oxidation Rates Increased With Permafrost Thaw

Methane oxidation in the eight studied sites ranged from 0.097 to 4.2 μg CH4 · gdw−1 · hr−1 (Table S1). Across the thaw gradient, CH4 oxidation was significantly different between sites (two-way ANOVA, F7,32 = 38.1; p < 0.001) but not sampling depth (F1, 32 = 0.7; p = 0.406) nor the interaction between site and depth (F7, 32 = 1.1; p = 0.417). As oxidation rates were not different by sampling depth, the rates for the upper and lower cores were merged in each site and one-way ANOVA was used to determine differences between sites (Figure 1). Oxidation rates were lowest in P1, the intact palsa site, and highest and most variable in F1, the open water Eriophorum angustifolium site. The rate of CH4 oxidation in the collapsed palsa transition site, P3, was significantly greater than in the intact palsa P1 (Tukey honestly significant difference, p = 0.009) but not significantly different than in the bog sites. Methane oxidation in peat from the bog-fen transition site, BF, was not significantly different from the bog sites and the fen sites; however, the mean rate of CH4 oxidation in BF falls between the mean oxidation rates in the bog and fen sites. In B1, where the core was split at the in situ water table depth, CH4 oxidation rates were comparable above (0.81 ± 0.16 μg CH4 · gdw−1 · hr−1) and below (1.1 ± 0.20 μg CH4 · gdw−1 · hr−1) the water table. Rates of CH4 oxidation from the unsaturated and saturated depths of B1 are not significantly different from CH4 oxidation in the upper and lower sections of peat of the submerged BF bog-fen transition site nor B2, the dry bog site which lacked porewater down to 20 cm below peat surface.

Details are in the caption following the image
Methane oxidation rates were significantly different by sampling site (one-way analysis of variance, F7,40 = 38; p < 0.001, n = 6). Methane oxidation rates were greatest in fully thawed fen sites and lowest in the intact palsa. The shaded boxes indicate transitioning thaw stages. The boxes represent the 25th and 75th percentiles of the data, the midlines indicate median oxidation rates, and the error bars extend to 1.5 times the interquartile range. The white diamonds represent site means (Table S1). Letters indicate groups by site determined using a Tukey's honestly significant difference post hoc test.

3.2 Response to Variation in CH4 Concentration

Overall, CH4 oxidation rates increased linearly with increasing headspace CH4 concentrations, independent of thaw stage (palsa, bog or fen) or peat depth (Figures 2a–2c). Methane oxidation did not follow Michaelis-Menten kinetics over the range of concentrations used at any site nor depth. All treatments had remaining headspace CH4 at the end of the incubation (Figure S2), and none of the incubations approached atmospheric CH4 concentrations during the incubation period.

Details are in the caption following the image
Methane oxidation rates increase linearly with CH4 headspace concentration in the (a) upper 15 cm of peat across the thaw gradient from sites P2, B1, and F1, as well as in the 0- to 10-cm, 10- to 20-cm, and 20- to 30-cm regions of the peat column in the (b) bog and (c) fen sites. All linear relationships were significant at the p < 0.05 level (Table S2). Each point represents a single measured CH4 oxidation rate calculated as the linear decrease in CH4 concentration over a 12-hr incubation period.

The increase in CH4 oxidation with increasing headspace CH4 concentrations was most pronounced in the fen: the slope of the linear regression an order of magnitude higher than in the palsa and bog (Figure 2a and Table S2). Also, the upper methane oxidation rates in the fen site were fourfold higher than the highest CH4 oxidation rates in the bog and palsa sites (Figure 2a). Methane oxidation rates ranged from 2.4 × 10−3 to 1.1, 1.7 × 10−2 to 4.8, and 1.9 × 10−2 to 19.8 μg CH4 · gdw−1 · hr−1 in the palsa, bog, and fen sites, respectively.

We found no depth-related changes in CH4 oxidation rates with increasing headspace CH4 concentrations for the bog site (Figure 2b); the slope of the regression lines did not differ significantly between the three depths (Table S2). The uppermost oxidation rates from the bog site under the larger range of headspace concentrations was over threefold higher than observed in the smaller range (Figures 2a and 2b). In contrast, the linear increase in CH4 oxidation rates was more pronounced in the 0- to 10-cm surface peat from the fen site as compared to the 10- to 20-cm and 20- to 30-cm depths (Figure 2c and Table S2). The uppermost oxidation rates were consistent between the larger and smaller range of tested concentrations for the fen (Figures 2a and 2c). Similar to the 2017 incubations (Figure 2a), the slope for the 0- to 10-cm fen site was an order of magnitude higher as compared to the 0- to 10-cm bog site (Figures 2b and 2c; Table S2).

3.3 Redox Chemistry in Bog and Fen Sites, and by Depth

Measurements of Eh, dissolved O2, porewater CH4, and alternative TEAs were made in the four sites with porewater (B1, BF, F1, and F2). Overall, dissolved CH4 increased as dissolved O2 decreased with depth (Table 2). Dissolved O2 and Eh also decreased with thaw. There was no consistent trend in the concentration of alternative TEAs with thaw nor depth.

Table 2. Porewater Redox Chemistry in the Bog and Fen Sites
Measurement Depth (cm) B1 BF F1 F2
Eh (mV) 10 84.13 ± 41.39a 56.33 ± 31.81a −56.99 ± 10.46b −51.21 ± 14.25b
20 93.50 ± 27.61 86.33 ± 40.23 10.64 ± 40.39 −47.41 ± 35.24
Dissolved O2 (mM) 10 0.15 ± 0.04a 0.13 ± 0.03a 0.04 ± 0.03ab 0.02 ± 0.01b
20 0.02 ± 0.01 0.06 ± 0.03 0.01 ± 0.2 0.01 ± 0.00
Dissolved CH4 (mM) 10 0.05 ± 0.02a 0.04 ± 0.03a 0.21 ± 0.11a 0.01 ± 0.00b
20 0.18 ± 0.11 0.20 ± 0.11 0.35 ± 0.17 0.01 ± 0.01
Dissolved NO3-N (μM N) 10 0.26 ± 0.11 0.43 ± 0.05 0.44 ± 0.03 0.33 ± 0.16
20 0.29 ± 0.14 0.4 ± 0.08 0.3 ± 0.14 0.24 ± 0.1
Dissolved SO4-S (μM S) 10 0.78 ± 0.56 0.99 ± 0.75 0.31 ± 0.28 3.13 ± 0.16
20 1.03 ± 0.16 1.53 ± 0.97 0.72 ± 0.78 1.03 ± 1.53
Porewater Mn (μM) 10 0.85 ± 0.08b 1.58 ± 0.32a 0.47 ± 0.04b 0.37 ± 0.11b
20 0.51 ± 0.25 1.38 ± 0.18 0.32 ± 0.23 0.42 ± 0.13
Porewater Fe (mM) 10 0.027 ± 0.013c 0.06 ± 0.2b 0.05 ± 0.00bc 0.28 ± 0.16a
20 0.03 ± 0.00 0.06 ± 0.01 0.03 ± 0.01 0.28 ± 0.08
  • Note. Values are reported as mean ± standard deviation of replicate samples (n = 3). Different letters denote a significant difference between sites at the 0.05 level (one-way ANOVA followed by Tukey HSD post hoc test for Eh and multiple comparison of means using the Kruskal-Wallis test with a Bonferroni correction for all others; analysis included both sampling depths). Dissolved oxygen was also significantly different between 10- and 20-cm sampling depths as determined by the Kruskal-Wallis test. Values for each replicate in Table S3 and S4.
  • Abbreviations: ANOVA: analysis of variance; HSD: honestly significant difference.

Redox potential ranged from −87.74 to 132.36 mV and dissolved O2 ranged from 0.006 to 0.19 mM. Redox potential and dissolved O2 decreased with increasing water table depth from B1 (Sphagnum spp. lawn) to F2 (C. rostrata). Both Eh (F3,20 = 22.5; p < 0.001) and dissolved O2 (Kruskal-Wallis, χ2(3) = 10.9; p = 0.012; Table 2) were significantly different across sites. Dissolved O2 was also significantly different by measurement depth (χ2(1) = 5.1; p = 0.024) and was consistently lower at 20 cm than at 10 cm (Table 2). The difference between dissolved O2 concentrations is most pronounced in the bog site B1, where dissolved O2 is 0.15 ± 0.04 mM at 10 cm and 0.02 ± 0.01 mM at 20 cm.

Dissolved CH4 concentrations varied by sampling site (χ2(3) = 15.3; p = 0.002) and ranged from 3.36 × 10−4 to 0.55 mM CH4 (Table 2). Dissolved CH4 concentration was highest in the E. angustifolium site (F1) and lowest in the C. rostrata site (F2). Porewater concentrations at the collapsed palsa site P3 were 0.033 ± 0.028 mM CH4 at 10 cm and 0.20 ± 0.08 mM CH4 at 20 cm, comparable to site B1 and BF. Soil air concentrations of CH4 in the palsa sites from P1 were lower than in P2, 1.82 ± 0.75 and 8.14 ± 6.77 ppm CH4, respectively. Dissolved CH4 concentration increased with depth in bog and fen sites. In 2018, when the deeper cores from bog and fen sites were extracted, conditions were drier in the bog (B1) and porewater CH4 concentrations increased from 0.006 ± 0.005 mM CH4 at 10 cm to 0.46 ± 0.17 mM CH4 at 30 cm. Porewater CH4 concentration in the fen sites (F1) increased from 0.33 ± 0.21 mM at 10 cm to 0.90 ± 0.16 mM at 30 cm.

Dissolved NO3-N concentrations ranged from 0.14 to 0.49 μM N, and dissolved SO4-S concentrations ranged from 0.10 to 3.16 μM S in the four wet sites. Concentrations of dissolved NO3-N and SO4-S were not significantly different by site type nor depth (Table 2). Porewater Mn ranged from 0.14 to 1.92 μM, and porewater Fe ranged from 0.02 to 0.40 mM. Porewater Mn (χ2(3) = 16.1; p = 0.001) and Fe (χ2(3) = 18.8; p = 0.0003) were significantly different by site type (Table 2). Porewater Mn was greatest in the bog-fen transition site (BF), and Fe was highest in F2. While not significantly different across sites, dissolved SO4-S was also highest in F2. High dissolved SO4-S and Fe concentrations in the porewater at this site may limit CH4 production and contribute to low dissolved CH4 concentrations in F2.

3.4 Impact of Redox Chemistry on CH4 Oxidation

The first two PLS factors combined explained 54.9% of the variation in CH4 oxidation rates and 57.1% of the variation in porewater chemistry predictor variables. Factor 1 explained 41.7% of variation in oxidation rates, and Factor 2 explained an additional 13.2% of variation (Table S5). PLS factor loadings reflect variables that maximize variance in the predictor variables as well as those that maximize the relationship between predictors and response variables. Redox potential, dissolved O2, and distance to water table loaded most strongly onto Factor 1, while porewater CH4 loaded most strongly onto Factor 2 (Table S5). Redox potential and dissolved O2 both had negative correlations with CH4 oxidation rates, while distance to water table had a positive correlation with CH4 oxidation rate (Figure 3). Porewater CH4 did not significantly correlate to measured CH4 oxidation rates (r = 0.12, p = 0.58), likely due to very low CH4 concentration in porewater from F2. Porewater Fe also has a strong loading onto Factor 1 like Eh and dissolved O2 (Table S5), but Fe did not significantly correlate with CH4 oxidation (r = 0.24, p = 0.26). High Fe concentrations in site F2 drive the variability in porewater Fe across sites and likely contribute to a strong loading for Fe. High loadings of pH and Mn on Factor 1 reflect the same pattern, as pH is notably higher in site F2 and Mn in site BF. While individual TEAs were not found significant in this analysis, they do contribute to the bulk redox conditions (Eh) in a site, which was found to be a significant predictor of CH4 oxidation.

Details are in the caption following the image
Pearson correlation (r) between Eh, dissolved O2, and distance to water table, which had the highest loadings on the partial least squares factor explaining the greatest variance in oxidation potential, and potential CH4 oxidation rate. Symbols are colored based on site type: B = bog; BF = bog-fen transition; F = fen.

Redox potential, dissolved O2, and distance to water table also had the largest variable importance for projection scores of 1.67, 1.04, and 1.20, respectively, implying that these three parameters are particularly important for predicting CH4 oxidation rates. In order to test the explanatory power of each of these parameters, additional PLS analyses were conducted omitting one at a time. Constructing a PLS model with the distance to water table and dissolved O2 but without Eh explains only 38.4% of the variance in CH4 oxidation rates, 16.5% less than a model including all three. A PLS model including Eh and dissolved O2 and omitting distance to water table explains 53.5% of the variance in oxidation rates, only 1.4% less than the model including all three measures. A model omitting dissolved O2 actually explains slightly more variance in oxidation rates than a model containing all three measurements (56.6% vs 54.9%).

4 Discussion

4.1 Methane Oxidation Increases Across a Gradient of Permafrost Thaw

The increase in CH4 oxidation rates across the permafrost thaw gradient (Figure 1) is consistent with previous incubation work in northern peatlands that report greater CH4 oxidation in submerged sites than in unsaturated or intermittently saturated sites (Kettunen et al., 1999; Moore & Dalva, 1997; Raghoebarsing et al., 2005; Whalen & Reeburgh, 2000). As in Sundh et al. (1995), we also observed a stronger response to increases in initial headspace CH4 concentrations in peat from the upper 15 cm of a submerged fen than in a thawing palsa or bog site (Figures 2a–2c), where higher in situ dissolved CH4 in site F1 may foster greater MOB activity than in the upper, aerated portions of the peat in P2 or B1. We also observed high CH4 oxidation in F2 (Figure 1), despite low in situ dissolved CH4 (Table 2). While low dissolved CH4 is unexpected in a site with high CH4 oxidation potential like F2, this site is positioned in an area of Stordalen Mire characterized as an outflow fen with high net DOC export (Olefeldt & Roulet, 2012). The week of porewater sampling, cumulative precipitation was 67.8 mm (SMHI, 2019); therefore, F2 may have been flushed with freshwater prior to sampling resulting in lower than expected dissolved CH4. Our incubation results reinforce previous work that suggests the expansion of sedge-dominated sites as permafrost peatlands thaw may provide a mechanism in rhizospheric oxidation that attenuates CH4 emissions from submerged sites with high rates of methanogenesis (Preuss et al., 2013; Rupp et al., 2019; Watson et al., 1997), which stands in contrast to work that highlights the zone of aerated peat above the water table as the key site of CH4 oxidation (Granberg et al., 1997; Sundh et al., 1994, 1995).

Singleton et al. (2018) examined CH4 oxidation at Stordalen Mire, using metagenome and metatranscriptome sequencing of peat from the active layer of a permafrost palsa, partially thawed bog dominated by Sphagnum spp., and a fully thawed fen dominated by E. angustifolium. Our incubation results support the broad thaw patterns identified by Singleton et al. (2018), who found high expression of pmoA, a key gene for CH4 oxidation, in a fully thawed fen site and observed that transcripts of pmoA comprised a greater proportion of total metatranscriptome reads in a fen at Stordalen Mire than in a bog or palsa. However, unlike Singleton et al. (2018), who observed shifts in MOB abundances within the depths we sampled in both bog and fen sites, our incubation results showed no depth differences in CH4 oxidation rates at any thaw stage. Incubation studies that quantify potential rates of CH4 cycling are needed to make functional assessments for metabolic insights from metagenomic approaches, as measurements of potential reaction rates can be used to validate and upscale modeling of landscape-level C budgets. Together, these results indicate that that MOB in fully thawed sites are responding to increased CH4 availability and possibly oxidizing a substantial fraction of CH4 in situ. These results support that as permafrost thaw expands the area of wet graminoid-dominated sites in permafrost peatlands, MOB represent an important control on CH4 emissions to the atmosphere.

4.2 Methane Oxidation Rates and Controls Differ Between Transitional and End-Member Thaw Stages

Our incubation results indicate that CH4 oxidation varies between transitional thaw stages (Figure 1). Assessing transitional thaw stages provides insights into finer-scale changes in CH4 oxidation as MOB respond to the continuum of vegetation, hydrology, and redox conditions that occur as permafrost thaws. For example, potential CH4 oxidation was higher in a collapsing palsa with saturated peat than an intact palsa with dry heath vegetation, as well as in a submerged bog-fen transition with both Sphagnum spp. and E. vaginatum versus a Sphagnum spp.-dominated bog where the water table was below peat surface (Figure 1). Similarly, Hodgkins et al. (2014) observed differences in CH4 production between sites with just Sphagnum spp., codominated by Sphagnum spp. and E. angustifolium, and with just E. angustifolium or C. rostrata and open water. Previous work focused on end-member thaw stages does not capture these transitioning stages or aggregates them with adjacent thaw stages. This effort to simplify shifts in CH4 dynamics during thaw can obscure important transition points in CH4 oxidation rates.

Furthermore, Malhotra and Roulet (2015) show that the controls of CH4 emissions in transitional thaw stages are not necessarily related to those in end-member or adjacent thaw stages, which may be driven by variability in the abundance, activity, and sensitivity to temperature and moisture of MOB and methanogens across the thaw gradient. The strength and directionality of controls on CH4 oxidation also change across thaw transitions in Stordalen Mire. The strength of the relationship between headspace CH4 concentration and CH4 oxidation varied across thaw stages (Figures 2a–2c). Comparing field conditions and incubation results, in intermediate stages that are more substrate-limited MOB increase their activity with higher dissolved CH4, but in fully thawed stages higher CH4 oxidation coincides with lower dissolved CH4 as MOB attenuate high rates of methanogenesis (Figure S3). Additionally, while overall CH4 oxidation increases with higher water table (Figure 3), in fully thawed sites CH4 oxidation tends to decrease with distance from the water table and thus also away from the moderating effects of the rhizosphere (Figure S3). Consistent with CH4 emissions, controls on CH4 oxidation are not constant across thaw stages. Our results further emphasize the need to incorporate transitional permafrost thaw stages into studies of CH4 dynamics in thawing peatlands, as focusing on end-member thaw states fails to capture the dynamic nature of CH4 dynamics across the full thaw gradient. As transitional thaw stages represent a significant and growing portion of the landscape in thawing permafrost peatlands (Malmer et al., 2005; Palace et al., 2018), further characterization of rates of CH4 transformation in transitional thaw stages is key for validating predictive models of CH4 flux in these systems.

4.3 Redox Potential Is a Stronger Predictor of Methane Oxidation than Water Table Depth

Higher CH4 oxidation in thawed permafrost versus intact permafrost has also been observed in polygonal tundra (Vaughn et al., 2016) and a peat plateau (Voigt et al., 2019); however, in both cases permafrost thaw led to drier conditions (i.e., less reducing environment). By contrast, our results indicate that more reducing conditions coincide with greater CH4 oxidation rates. Redox potential decreased along a gradient of permafrost thaw and rising water table depth at Stordalen Mire (Table 2), consistent with previous work that reports more reducing conditions with increasing soil moisture, water table depth, and seasonal thaw depth (Street et al., 2016; Svensson & Rosswall, 1984). Our results indicate that Eh, dissolved O2, and distance to water table are all important for predicting CH4 oxidation rates (inferred by PLS; Table S5 and Figure 3), but that redox potential contributes more toward the explained variance in CH4 oxidation rates.

While water table depth can impact rates of CH4 oxidation and emission by creating an interface at which both CH4 and O2 are present (Bubier et al., 1995; Larmola et al., 2010; Moore & Dalva, 1993), the depth of the water table does not solely control redox conditions that affect methanotrophy in partially and fully thawed sites. For example, while sedge-dominated wetlands are associated with high CH4 emissions due to plant-mediated transport that allows CH4 to bypass oxidation (Noyce et al., 2014; Waddington et al., 1996), sedges in submerged, thawed sites also moderate the biogeochemical environment by oxidizing porewater in the rhizosphere and creating favorable conditions for methanotrophy (Elberling et al., 2011; Flessa & Fischer, 1992; Rupp et al., 2019), in additional to providing substrate for MOB by stimulating methanogenesis (Chanton et al., 1995; Hodgkins et al., 2014). The oxidizing effect of the sedge rhizosphere can be seen in sites BF and F1, in which E. vaginatum and E. angustifolium are present (Table 1), where Eh at 20 cm is greater than Eh at 10 cm (Table 2). Additional mechanisms could oxidize submerged sites, including rainwater (Neumann et al., 2017) and photosynthesis in submerged mosses (Knoblauch et al., 2015; Liebner et al., 2011), which are independent of water table depth. Additionally, competition for O2 between aerobic heterotrophs and MOB in the rhizosphere of sedge-dominated sites can deplete available O2 (Ding et al., 2004; van Bodegom et al., 2001). As such, low dissolved O2 could reflect active MOB and aerobic heterotrophs in sedge-dominated sites. As Eh is an integrative measure of the effects of hydrology, vegetation, and environmental conditions that impact CH4 transformations, it may be a stronger measure to use to predict rates of in situ CH4 oxidation.

Redox potential may also be a stronger predictor of CH4 oxidation under more reducing conditions given new insights into the metabolic flexibility of MOB in submerged sites, as it is a more complete measure of the redox environment. Our results contribute toward a body of evidence that suggests that MOB actively oxidize CH4 under moderately reducing, low-O2 conditions. Potential for CH4 oxidation in reduced, low-O2 conditions has been observed in alpine lakes and permafrost thaw ponds, suggesting that MOB are resilient to changing oxygenation regimes (Blees et al., 2014; Crevecoeur et al., 2017). Gammaproteobacterial methanotrophs within the Methylococcaceae family have been found to be abundant and active at depth across Arctic wetland sites (Christiansen et al., 2015; Liebner & Wagner, 2007; Singleton et al., 2018; Tveit et al., 2014), indicating that they are able to oxidize CH4 under micro-aerobic conditions. Singleton et al. (2018) identified genes for dissimilatory nitrate and sulfate reduction in MOB genomes recovered from submerged fen samples, which allow MOB to use alternate TEAs for energy generation in low-O2 environments in order to utilize available O2 for CH4 oxidation. Conversely, methanotroph genomes from bog samples included more provisions for variable CH4 availability than for low-O2 conditions. While there is some evidence of anaerobic CH4 oxidation (AOM) in minerotrophic peatlands, the TEA used for AOM is uncertain (Smemo & Yavitt, 2007) and rates of AOM in boreal peatlands are very low to below detection (Blazewicz et al., 2012); therefore, the importance of AOM in peatlands is still unclear (Smemo & Yavitt, 2011). All but one of the Eh measurements reported in this study (Table S3) from sites with porewater were above the empirical threshold for aerobic CH4 oxidation (−75 mV; Fiedler & Sommer, 2000) and NO3-N and SO4-S, which are likely TEAs for AOM, were at low in situ concentrations (Table 2); therefore, we expect aerobic MOB capable of tolerating low O2 to be the dominant methanotrophs oxidizing CH4 at our submerged sampling sites.

5 Conclusions

Our results indicate that CH4 oxidation potential increases across a gradient of increasing permafrost thaw from intact palsa mounds to submerged fens in an ice-rich peatland. Methane oxidation potential and its apparent controls differed between end-member and transitional thaw stages, highlighting the importance of characterizing transitional thaw stages that comprise an increasingly significant area of thawing peatlands. Furthermore, covariation of CH4 oxidation rates was strongest with redox potential (Eh), followed by dissolved O2 and water table depth. We contribute to the mounting evidence of studies suggesting that MOB in submerged sites are able to survive and oxidize CH4 under microaerobic conditions and conclude that redox potential may be a stronger predictor of CH4 oxidation than dissolved O2 or water table depth alone. Our results emphasize the importance of integrating measurements of CH4 transformations and geochemistry across the full permafrost thaw continuum to better understand the potential of in situ CH4 oxidation to attenuate CH4 emissions in rapidly thawing permafrost peatlands.


This work was supported by the National Science Foundation Graduate Research Fellowship Program under grant DGE145027, the American Geosciences Institute Harriet Evelyn Wallace Scholarship, the American-Scandinavian Foundation Graduate Fellowship, the Geological Society of America Graduate Student Research Grant, the UNH Earth Science Travel Grant, and the U.S. Department of Energy Office of Biological and Environmental research grant DE-SC0016440. Thank you to the Abisko Scientific Research Station and their staff for the use of their facilities and to the McDowell Lab at UNH for performing IC analyses. Special thanks to Patrick Crill for the use of his laboratories and GC-FID, the faculty of the NERU NSF REU (EAR 1063037), and Virginia Rich for supporting the early development of this work; to Jessica DelGreco, Christopher Horruitiner, Sophia Burke, Apryl Perry, Kathryn Bennett, and Peter Tansey for their help with field work; and to Nigel Roulet (McGill University) for supporting initial sampling-site infrastructure. The data associated with this work are stored on the data repository site of the ISOGENIE project and can be found online (https://isogenie-db.asc.ohio-state.edu/datasources).