Changes in peat chemistry associated with permafrost thaw increase greenhouse gas production

Significance

We address the effect of thawing permafrost, and attendant subsidence-induced shifts in hydrology and plant community structure, on CH₄ and CO₂ production potentials and mechanisms driven by changes in organic matter chemical composition in a thawing peatland complex. Advanced analytical characterization of peat and dissolved organic matter along the thaw progression indicated increasingly reduced organic matter experiencing greater humification rates, which were associated with higher relative CH₄ and CO₂ production potentials, increasing relative CH₄/CO₂ production ratios, and shifts from hydrogenotrophic to acetoclastic methanogenesis. The effects of this increase in organic matter reactivity with permafrost thaw could intensify the increases in CH₄ and CO₂ release already predicted due to increasing temperatures, permafrost carbon mobilization, and waterlogging-induced changes in redox conditions.

Abstract

Carbon release due to permafrost thaw represents a potentially major positive climate change feedback. The magnitude of carbon loss and the proportion lost as methane (CH₄) vs. carbon dioxide (CO₂) depend on factors including temperature, mobilization of previously frozen carbon, hydrology, and changes in organic matter chemistry associated with environmental responses to thaw. While the first three of these effects are relatively well understood, the effect of organic matter chemistry remains largely unstudied. To address this gap, we examined the biogeochemistry of peat and dissolved organic matter (DOM) along a ∼40-y permafrost thaw progression from recently- to fully thawed sites in Stordalen Mire (68.35°N, 19.05°E), a thawing peat plateau in northern Sweden. Thaw-induced subsidence and the resulting inundation along this progression led to succession in vegetation types accompanied by an evolution in organic matter chemistry. Peat C/N ratios decreased whereas humification rates increased, and DOM shifted toward lower molecular weight compounds with lower aromaticity, lower organic oxygen content, and more abundant microbially produced compounds. Corresponding changes in decomposition along this gradient included increasing CH₄ and CO₂ production potentials, higher relative CH₄/CO₂ ratios, and a shift in CH₄ production pathway from CO₂ reduction to acetate cleavage. These results imply that subsidence and thermokarst-associated increases in organic matter lability cause shifts in biogeochemical processes toward faster decomposition with an increasing proportion of carbon released as CH₄. This impact of permafrost thaw on organic matter chemistry could intensify the predicted climate feedbacks of increasing temperatures, permafrost carbon mobilization, and hydrologic changes.

High-latitude soils in the Northern Hemisphere contain an estimated 1,400–1,850 petagrams (Pg) of carbon, of which ∼277 Pg is in peatlands within the permafrost zone (1, 2). This quantity of 277 Pg represents over one-third of the carbon stock in the atmosphere (ca. 800 Pg) (3). The fate of this carbon in a warming climate—i.e., the responses of net carbon balance and CH₄ emissions—is important in predicting climate feedbacks of permafrost thaw. Although northern peatlands are currently a net carbon sink, and have been since the end of the last glaciation, they are a net source of CH₄ (4, 5), emitting 0.046–0.09 Pg of carbon as CH₄ per year (4, 6, 7). Due to CH₄’s disproportionate global warming potential (33× CO₂ for 1 kg CH₄ vs. 1 kg CO₂ at a 100-y timescale) (8), this is equivalent to 6–12% of annual fossil fuel emissions of CO₂ (8.7 Pg of C) (9). The thaw of permafrost peatlands may alter their CH₄ and CO₂ emissions due to mobilization of formerly frozen carbon, higher temperatures, altered redox conditions, and evolving organic matter chemistry. Changes in carbon emissions, and in CH₄ emission in particular, could have potentially significant climate impacts.

CH₄ is produced by two primary mechanisms (10–12), distinguishable by δ¹³C values. The reduction of CO₂ with H₂ (hydrogenotrophic production) generally produces CH₄ more depleted in ¹³C (δ¹³C = −110 to −60‰) than CH₄ produced by the cleavage of acetate into CH₄ and CO₂ (δ¹³C = −70 to −30‰) (10, 11, 13–15). Due to the coproduction or utilization of CO₂ during CH₄ production (10–12, 16, 17), δ¹³C_CH4 also depends on δ¹³C_CO2, so we use the more robust parameter α_C (10) to represent the isotopic separation between CH₄ and CO₂. Despite the two production pathways’ stoichiometric equivalence (17), they are governed by different environmental controls (18). Distinguishing these controls and further mapping them is therefore essential for predicting future changes in CH₄ formation under changing environmental conditions. Several studies have suggested that the proportion of CH₄ produced by acetate cleavage relative to CO₂ reduction is likely to increase with increasing pH (19, 20) and organic matter reactivity (12, 14, 15), but direct evidence of the latter is lacking.

In this study, we tested the hypotheses that (i) organic matter reactivity increases with permafrost thaw due to thaw-induced subsidence and associated shifts in hydrology and plant community (21), and (ii) CH₄ production shifts from hydrogenotrophic to acetoclastic due to this increase in organic matter reactivity. We assessed organic matter reactivity along a distinct chronosequence of permafrost thaw stages with differing plant community and hydrology by performing anaerobic incubations of peat collected along this sequence. We then compared the results to peat and dissolved organic matter (DOM) chemical structure, as described by C/N ratios and Fourier transform infrared (FTIR) spectroscopy of peat and Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) of DOM. Our study specifically addresses the effect of thawing permafrost, and its attendant shifts in hydrology and plant communities, on CH₄ and CO₂ production potentials and mechanisms, via changes in organic matter chemical composition (commonly referred to as organic matter “quality”) in a thawing peatland complex.

Study Site and Habitat Classification

Stordalen Mire (68.35°N, 19.05°E) is a peat plateau in northern Sweden with a peat depth ranging from 1 to 3 m (22). The ground’s thaw state and consequent relationship to the water table determine the plant community, which in turn determines organic matter composition (21). Detailed vegetation surveys and analyses of aerial photographs taken at Stordalen (22–24) have demonstrated that permafrost thaw between 1970 and 2000 has caused dry permafrost hummocks to decrease in area by 10%, giving way to wetter minerotrophic sites, which increased by 17% (24). These changes have been accompanied by shifts in plant community from shrubs and lichens in permafrost hummocks to Sphagnum spp. in wetter sites followed by evolution to tall graminoid species as thaw-induced subsidence increases (22–24). Our site selections are based on these studies, which provide direct evidence that at minimum, six of our sites (Table 1 and Table S1) were once palsa (sites named PHS, PHB, S, and E) or likely palsa (sites named Bog1 and Fen1) underlain by permafrost (22–24). Similar processes are known to occur in other Arctic peatlands, where permafrost thaw creates wet depressions that are colonized first by Sphagnum and then by sedge species as the ground collapses with increasing thaw (25–27).

Table 1.

Sites selected for study, in order of thaw stage; detailed descriptions of each site (with justification their positions in the thaw sequence) are given in Table S1

Site name	Habitat classification	Dominant vegetation	pH
PHS	Collapsed palsa	E. vaginatum, woody species	4.1
PHB	Collapsed palsa	E. vaginatum, floating Sphagnum	4.1
Bog1	Bog	Sphagnum spp.	4.2
SOS	Bog	Sphagnum spp.	4.0
S (triplicate cores)	Bog	Sphagnum spp.	4.2
EOS	Fen	E. angustifolium, Sphagnum spp.	4.8
E (triplicate cores)	Fen	E. angustifolium	5.8
Fen1	Fen	C. rostrata	6.0
Fen2	Fen	E. angustifolium	5.7

For this study, habitats were classified as “palsa,” “collapsed palsa,” “bog,” or “fen.” The palsa, bog, and fen classifications correspond, respectively, to the dry ombrotrophic (I), combined semiwet and wet ombrotrophic (II and III), and tall graminoid minerotrophic (IV) classifications of Johansson et al. (22) and represent a thaw gradient of increasing active layer depth. To these we added the additional designation of collapsed palsa for recently thawed waterlogged thermokarst sinkholes surrounded by palsa. Vegetation in these collapsed palsa sites is generally more diverse than that in the other three habitats, ranging from woody shrubs and Eriophorum vaginatum where the water table is at or just below the peat surface to floating Sphagnum mats where the water table is at or above the peat surface. Over time, vegetation in collapsed palsa may shift toward domination by nonfloating Sphagnum spp. (more bog-like) or sedges (more fen-like). Although permafrost thaw does not necessarily progress through all of these stages—for instance, rapid thaw may create a collapsed palsa sinkhole, which subsequently becomes more like a bog or a fen, whereas more gradual thawing may skip the collapsed palsa stage and cause more gradual changes in vegetation—these stages represent a thaw progression based on the possible shifts in vegetation and hydrology. Following a temporal succession of apparent time since onset of thaw, the habitats are ordered: (i) palsa, (ii) collapsed palsa, (iii) bog, and (iv) fen.

We selected nine sites spanning the thaw progression from collapsed palsa to fen (Table 1 and Table S1). Intact palsa sites were not included in this study because their dry, aerobic status prevents meaningful comparison with other sites based on anaerobic incubations and DOM measurements. Previous measurements of CH₄ and CO₂ fluxes from palsas have shown that they produce negligible amounts or even consume ambient CH₄ (23, 28, 29) but are a net source of CO₂ (28).

Results

Thaw Increases Potential CH₄ and CO₂ Production.

Anaerobic incubations (Fig. S1) reveal significant increases in potential CH₄ and CO₂ production and CH₄/CO₂ ratios along the thaw progression (P < 0.0001 for all of these trends) (Fig. 1 and Fig. S2).

Fig. 1.

CH₄ and CO₂ production rates in anaerobic incubations (±SE between replicates), which represent relative organic matter lability as potential decomposition rates. Samples are grouped into categories and placed in hypothesized order of permafrost thaw progression as described in Table 1 and Table S1. (A) CH₄ production rates, (B) CO₂ production rates, and (C) ratios of CH₄/CO₂ production.

Thaw Shifts Methanogenesis to the Acetoclastic Pathway.

δ¹³C and α_C of incubation CH₄ and CO₂ are shown in Fig. 2. Higher α_C (and more negative δ¹³C_CH4) indicates a higher proportion of hydrogenotrophic methanogenesis, whereas lower α_C (and less negative δ¹³C_CH4) implies more acetoclastic methanogenesis (10, 12). The incubations fall into two groups (Fig. 2C) with a significant (P < 0.0001) separation in α_C values: collapsed palsa and bog peat with α_C between 1.075 and 1.092, and fen peat with α_C between 1.048 and 1.062. This result suggests a distinct shift from hydrogenotrophic to acetoclastic methanogenesis along the thaw progression associated with increasing organic matter lability (Fig. 1 A and B) and increasing pH (Table 1). δ¹³C_CO2 (Fig. 2B) also increases along the thaw progression, consistent with the increasing CH₄/CO₂ ratio (Fig. 1C) and reflecting an increasing proportion of methanogenesis-derived CO₂ (17).

Fig. 2.

Carbon isotopic composition of CH₄ and CO₂ produced in anaerobic incubations (±SE between replicates): (A) δ¹³C_CH4, (B) δ¹³C_CO2, and (C) α_C, defined as (δ¹³C_CO2 + 1,000)/(δ¹³C_CH4 + 1,000). α_C and δ¹³C_CH4 imply a shift in methanogenesis from hydrogenotrophy to acetoclasty along the thaw gradient associated with increasing pH and organic matter lability. Increasing δ¹³C_CO2 is consistent with increasing CH₄/CO₂ production ratios (Fig. 1C) (17).

Peat C/N Ratios Reflect Plant Community Shifts.

C/N ratios (by weight) of the peat used in the incubations are highest in bogs, intermediate in collapsed palsas, and lowest in fens (Fig. 3A). This trend is consistent with shifts in plant litter quality along the thaw sequence. Living plant samples of dominant Stordalen species collected at the Marcell Experimental Forest (northern Minnesota) showed a similar pattern in C/N ratios, from E. vaginatum (common in intact palsas, collapsed palsas, and bogs; C/N = 39 ± 24), through Sphagnum spp. (common in bogs; C/N = 46 ± 18), Eriophorum angustifolium (common in fens; C/N = 19 ± 0.4), and Carex spp. (common in fens; C/N = 25 ± 3).

Fig. 3.

Peat chemistry changes with thaw. (A) C/N weight ratios (±SE) of preincubation peat. (B) Ratios of deep (24–35 cm) to surface (<10 cm) peat HI for each site category (±SE), representing humification rates, for several wavenumbers (Fig. 4B) with respect to polysaccharides (1,030 cm⁻¹). The bog category excludes site S (which had no samples below 21 cm), and the fen category excludes site E (which had Sphagnum remains at depth).

Peat Becomes More Labile Across the Thaw Gradient.

To investigate the structure of the peat organic matter, we used FTIR spectroscopy to examine: (i) surface peat (<10 cm, average ∼3 cm); (ii) near-surface peat used in incubations, sampled just below the water table, examined preincubation (<22 cm, average ∼11 cm); and (iii) deeper peat (24–35 cm, average ∼25 cm) (Table S2). Whereas the spectra are generally similar between samples, the relative intensity of particular bands differs (Fig. 4), revealing the nature of chemical changes during peat development and humification (30). Along the thaw progression, carboxylic acid bands [1,720 cm⁻¹ (31–34); Fig. 4A] weaken, likely because the fens’ higher pH (Table 1) maintains organic acids as nonvisible carboxylate anions. Surface fen peat had more abundant polysaccharides [1,030–1,080 cm⁻¹ (35)] and less abundant lignins [1,513–1,515 cm⁻¹ (33)], aromatics [1,600–1,650 cm⁻¹ (31)], and aliphatics [2,850 and 2,920 cm⁻¹ (31)] than deep fen peat, indicating more cellulose (O-alkyl-C) plant material in the former vs. more decomposed, humified structures in the latter (Fig. 4B; less-pronounced differences were seen in bog and collapsed palsa peat).

Fig. 4.

Average peat FTIR absorption spectra, with between-site SEs as shaded areas. Spectra are stacked (i.e., absorbance = 0 at each apparent baseline) and sized to the same vertical scale. (A) Organic acids in incubated peat samples (preincubation) decrease with thaw from collapsed palsas (n = 2) to bogs (n = 3) to fens (including both sedge only and Sphagnum + sedge; n = 4). (B) In fens (excluding site E, which had Sphagnum remains at depth), humification leads to spectral changes between surface (n = 3) and deep (n = 3) peat. References for wavenumbers: see Results, Peat Becomes More Labile Across the Thaw Gradient.

To further identify the FTIR spectral differences between peat in different habitats, we calculated ratios of aliphatic, aromatic, and phenolic moieties to polysaccharides (Table S2), defined as humification indices (HI) (36) because they tend to increase with decomposition in soils. Across the sites, the surface and near-surface peat lacked consistent differences in HI; however, in the deep peat, fen HIs (except site E) were higher than collapsed palsa and bog HIs, suggesting more advanced decomposition in deep fen peat (Table S2). (In fen site E, remnant Sphagnum at depth resulted in an HI similar to deep bog samples, because HI reflects both source plant material and decomposition.) Consistent with increased decomposition with depth, most HIs were higher in deep than in surface peat (Table S2; see also Fig. 4B). This change can be quantified as a humification rate by calculating the ratio of deep to surface HI (24–35 cm vs. 0–10 cm), with higher ratios implying higher rates of humification through the soil column (i.e., faster decomposition over time). Humification rates with respect to all wavenumbers increased along the thaw progression, with the greatest increases occurring for the transformation of polysaccharides into aliphatic moieties (2,920 and 2,850 cm⁻¹) (Fig. 3B).

DOM Chemistry Changes with Thaw.

To investigate the structure of DOM, pore water samples spanning the thaw progression were analyzed by FT-ICR MS, of which representative bog (SOS site, 31 cm) and fen (E site, 25.5 cm) samples are shown (Fig. 5 and Fig. S3; Table 2). Peaks were assigned molecular formulas, taking into account C, H, O, N, and S (total 21,114 compounds; 11,527 bog and 9,587 fen), which were classified based on their N and S content and H/C and O/C ratios (Table 2). Sixty-four percent of compounds (13,436/21,114) were present in both samples (i.e., matching). To visualize differences in DOM composition, compounds that were unique to each sample (i.e., nonmatching) are plotted by their H/C vs. O/C ratios and molecular size (Fig. 5).

Fig. 5.

Analysis of molecular formulas found in DOM in representative bog (site SOS, 31 cm, green circles) and fen (site E, 25.5 cm, purple diamonds) samples. (A) van Krevelen diagram of nonmatching formulas (those not shared between samples) in bog vs. fen DOM, labeled with the specific compound classes expected to occur in each region (37). Each point represents one molecular formula present in one sample, plotted by its O/C and H/C ratio. (B) Molecular size distribution of nonmatching formulas. Smaller molecules in fen DOM indicate more advanced decay compared with bog DOM (38).

Table 2.

Abundance of different types of formulas based on elemental composition (containing CHO only; CHON only; CHOS only; and CHONS) and compound class (described in the van Krevelen diagram; Fig. 5A)

Sample	No. of formulas	CHO	CHON	CHOS	CHONS	Lipid-like	Protein- and AS-like	Other low-O/C*	Lignin-like	Tannin-like	Condensed aromatics
		%m	%m	%m	%m	%m	%m	%m	%m	%m	%m
		(±1.6)	(±1.2)	(±0.5)	(±0.1)	(±0.0)	(±0.4)	(±0.8)	(±1.3)	(±0.9)	(±0.2)
All formulas:
Bog (SOS)	11,527	88.9	7.6	2.4	1.1	0.0	2.2	0.2	45.9	40.2	2.3
Fen (E)	9,587	79.7	9.2	10.3	0.8	2.6	5.3	4.0	59.9	20.2	1.0
Nonmatching formulas:
Bog (SOS)	4,809	72.7	18.2	3.2	5.9	0.2	2.9	0.3	40.9	32.2	15.4
Fen (E)	2,869	69.3	6.9	22.0	1.9	9.7	7.7	13.9	63.0	4.5	2.7

Abundances are given as %m, which represents the combined relative abundance of all formulas in each class as a percentage of the total abundance of all formulas. SEs for each category are based on a parallel analysis of replicate samples collected at the same depth and time.

^*Peaks in the region with 1 ≤ H/C ≤ 1.6 and 0 ≤ O/C ≤ 0.29, representing low-O compounds similar to lipids but with lower H/C (Fig. 5A).

DOM compositional differences are clear between bog and fen sites. In the SOS/E site comparison (Fig. 5 and Fig. S3; Table 2), this is despite E’s presence of remnant Sphagnum at depth, revealing that DOM at depth (unlike solid peat) is shaped by surface vegetation (39) even when buried remnants of older vegetation are present. Two distinct compound classes can be seen in both samples (Fig. 5A): class 1, occupying most of the plot area and representing various aliphatics, lignins, and tannins, and class 2 in the lower left-hand corner (O/C = 0–0.4, H/C = 0–0.8) representing condensed aromatic structures (37). Within both classes, O/C ratios are lower in fen than in bog DOM. Specifically, fen DOM contained more lipid-, other low-O/C-, protein- and amino sugar- (AS), and lignin-like compounds, whereas bog DOM contained more condensed aromatic and tannin-like compounds (Table 2 and Fig. 5A) (37). Fen DOM also had a lower average molecular size than bog DOM (Fig. 5B).

To characterize the oxidation states and unsaturation of the compounds comprising DOM, the double-bond equivalence (DBE), i.e., the total number of double bonds and aliphatic rings in each molecular formula, was calculated. The DBE distribution in bog DOM is skewed toward higher values compared with fen DOM (Fig. S3A), indicating that bog DOM is more unsaturated overall. To characterize C=C bonds and rings, DBE−O was calculated as the number of oxygen atoms subtracted from DBE under the assumption that most DOM oxygen is bound to carbon by a double bond (40, 41). In contrast with DBE, the DBE−O distribution is lower in bog than in fen DOM (Fig. S3B). This difference between the DBE and DBE−O distributions indicates that bog DOM contains more hydrophilic, oxygen-rich compounds than fen DOM, and that a larger portion of the unsaturation in bog DOM is due to C=O bonds (mostly from carboxylic acid) than in fen DOM (41, 42). The higher oxygen content of bog DOM, as evidenced by both its DBE/DBE−O distributions and its higher O/C ratios relative to fen DOM, indicate that bog DOM has a higher oxidation state than fen DOM.

Discussion

The increase in relative peat CH₄ and CO₂ production potentials (Fig. 1 A and B) along the thaw gradient when incubated under identical temperature and water saturation indicates an increase in organic matter lability. We propose three primary causes for this increase, all of which are tied to the changes in plant community associated with permafrost thaw: (i) increasing pH causes loss of organic acids that would otherwise inhibit microbial decomposition; (ii) decreasing peat C/N ratios increase organic matter quality by providing more abundant nitrogen; and (iii) increasing protein-like (and perhaps also relatively labile lipid-like) compounds in DOM may act as high-quality substrates for microbes. These changes can be directly observed in the chemical structure of peat and DOM along the thaw progression.

The types of compounds comprising organic matter help define its lability. For example, undissociated organic acids, particularly sphagnum acid and other Sphagnum-derived phenolics (43, 44), inhibit organic matter decay. The apparent increase in lability across the gradient could thus be due to a decrease in organic acids (Fig. 4A) driven by changing plant inputs and increasing pH (Table 1 and Table S1). However, pH’s effect on decomposition (both directly through impacts on microbial activity and indirectly via shifting organic acids to their less inhibitory anionic forms) is unlikely to be the sole process governing decomposition rates in this system, given Ye et al.’s (45) demonstration that incubating bog peat at fen-typical pH did not stimulate fen-like CH₄ production to the levels seen in fen peat. Phenolic abundance may also decrease along the thaw sequence due to the enzymatic latch mechanism (46), whereby the enzyme phenol oxidase (which degrades phenolic compounds) depends on bimolecular oxygen availability. In fens, vascular plant roots may transport oxygen into the soil, where it activates phenol oxidase and thereby decreases the concentration of decomposition-inhibiting phenolics. Decreasing C/N ratios (Fig. 3A), driven by thaw-associated changes in plant community (from E. vaginatum and Sphagnum spp. to E. angustifolium and Carex rostrata) that lower the organic matter C/N ratio, may also increase organic matter lability by decreasing nitrogen limitation for decomposers.

Whereas the solid-phase peat represents an important substrate for decomposition, a large proportion of CH₄ and CO₂ production uses materials in the dissolved phase (39). This is particularly true in fens due to fen DOM’s high lability (relative to bog DOM), such that the majority of fen respiration products is derived from DOM decomposition (39). We therefore analyzed DOM chemistry and reactivity in addition to peat chemistry and reactivity. The DOM chemical structure was markedly different between bog and fen (Fig. 5 and Fig. S3; Table 2), shaped in part by different plant source material. Decay-resistant tannins from Sphagnum were abundant in the bog, whereas the fen had a higher proportion of comparatively labile protein- and fatty-acid-like compounds possibly originating as sedge root exudates.

In addition to organic matter compositional differences that directly affect its lability, there are also differences indicating the degree of decomposition. Ratios of HI in deep vs. surface peat, which represent decomposition rates through the soil column, indicate that decomposition rates increase along the thaw progression (Fig. 3B). Analogous to the fens’ higher solid-phase humification rates with respect to the accumulation of both lignin- and lipid-like moieties (Fig. 3B), fen DOM also has a higher percentage of dissolved lignin- and lipid-like compounds (some of which are recalcitrant and can accumulate) than bog DOM (Table 2), consistent with more advanced decomposition. Fen DOM also has a smaller average molecular size than bog DOM (Fig. 5B). This result is consistent with the size reactivity model, in which high molecular weight DOM is hydrolyzed into low molecular weight DOM, causing small refractory compounds to accumulate (38). Furthermore, because microbial biomass includes abundant lipid- and protein-like substances, these categories, as well as the unknown low-O/C category (which appears in the same general region of the van Krevelen plot; Fig. 5A), may represent microbial biomass. Higher microbial biomass in the fen is likely associated with higher microbial activity (47), so this result is consistent with the higher CH₄ and CO₂ production potentials (Fig. 1 A and B) and humification rates (Fig. 3B) observed in fens relative to bogs.

The increase in relative CH₄/CO₂ production ratios along the thaw progression has several possible explanations. Anaerobic CH₄ oxidation (48, 49), if present, may be more pronounced in collapsed palsa and bog peat than in fen peat, but because CH₄ oxidation tends to enrich the δ¹³C of residual CH₄ (50) whereas the δ¹³C_CH4 values in bog and collapsed palsa peat were relatively depleted (Fig. 2A), this explanation seems unlikely. More plausible reasons for the observed CH₄/CO₂ ratios involve pH and organic matter oxidation state. Ye et al. (45) demonstrated that CH₄/CO₂ production ratios increase with pH, which suggests that the low pH of the bog and collapsed palsa habitats may selectively inhibit CH₄ more than CO₂ production. Methanogenesis in collapsed palsa and bog peat may also be inhibited by non-O₂ electron acceptors, particularly humic acids (51–54), whereas fen DOM’s lower oxidation state (Fig. 5A and Fig. S3) may translate to lower humic acid electron-accepting capacity. The lower O/C ratios in fen DOM could also support higher relative CH₄/CO₂ production ratios based on its stoichiometry, which supports the direct conversion of more DOM to CH₄ instead of CO₂ (55, 56). Finally, the low CH₄/CO₂ ratios in bogs and collapsed palsas may reflect diversion of reduced decay products into microbial biomass (anabolism) instead of CH₄ (catabolism).

Because the peat was incubated at a higher temperature (22 °C) than generally found at the incubated depths (<15 °C), it is possible that the incubation CH₄/CO₂ production ratios could be higher than field ratios due to rapid consumption of alternate electron acceptors in the higher-temperature, closed incubations (57). However, gas production was linear over the incubations (Fig. S1), and Lupascu et al. (29) found similar relative differences in methanogenesis between Sphagnum and sedge peat from Stordalen when incubated at 4 °C, 14 °C, and 24 °C. The lower CH₄/CO₂ ratios in the early part of the thaw sequence are also consistent with bog DOM’s higher oxidation state and organic oxygen content relative to fen DOM (Fig. 5A and Fig. S3). The incubations thus provide a useful proxy for relative field CH₄/CO₂ ratios.

Combined with the increases in organic matter lability as revealed by the incubations, as well as the more advanced decay state of fen peat and DOM relative to bog and collapsed palsa peat and bog DOM as revealed by FTIR and FT-ICR MS, the shift from hydrogenotrophic to acetoclastic methanogenesis with thaw (Fig. 2) is consistent with the understanding that these shifts are associated with increasing organic matter lability (12, 14, 15). A likely mechanism for this effect is the loss of organic acids along the thaw progression (Fig. 4A) driven by increasing pH (Table 1), because hydrogenotrophic methanogens are generally more acid-tolerant than acetoclastic methanogens (18, 20, 58).

Conclusions

Based on increases in CH₄ and CO₂ production potentials across the permafrost thaw progression, organic matter lability appears to increase significantly with thaw, subsidence, and resulting changes in vegetation (i.e., up to 10–15× more CH₄ + CO₂ production in incubations of fen peat compared with collapsed palsa peat; Fig. 1 A and B). In addition to the overall increase in decomposition potential, the incubations also showed a five- to sixfold increase in the relative proportion of decomposition by methanogenesis (Fig. 1C) along with a shift from hydrogenotrophic to acetoclastic CH₄ production (Fig. 2). Although controlled laboratory incubations and organic matter characterizations of targeted field samples cannot be directly upscaled to absolute field CH₄ and CO₂ production, the increases in in situ organic matter quality and decomposition rates (as indicated by C/N ratios and FTIR of peat and FT-ICR MS of DOM) suggest that the trends seen in the incubations are likely to translate into relative increases in field CH₄ and CO₂ production rates and CH₄/CO₂ ratios along the thaw progression, which are likely to be further enhanced by increased graminoid-mediated gas transport in fens (59).

As the Arctic warms, peat is expected to release more CH₄ and CO₂ as permafrost thaws and exposes more peat, and as peat experiences faster decomposition due to increasing temperatures. The production of CH₄ relative to CO₂ is also likely to increase due to increasing anaerobic conditions caused by peat waterlogging associated with thaw and subsidence. Our results suggest that estimates of CH₄ climate feedbacks based on these effects alone are likely to be conservative: as permafrost thaws, overall carbon release, as well as the proportion released as CH₄, is likely to increase on a per-volume basis as thaw-induced subsidence and shifts in plant community increase organic matter lability and favor increasingly methanogenic conditions.

Materials and Methods

Incubations were performed in the dark under water-saturated, anaerobic conditions with a N₂ headspace at 22 °C. Although this temperature is higher than those likely to occur at ∼10-cm depth in the field, it is common practice to incubate samples at a temperature higher than in situ to observe significant increases in gas production at the timescales used for laboratory experiments (21, 60–63). The results thus represent potential CH₄ and CO₂ production rates (21), which are evaluated relative to one another to assess changes in organic matter lability along the thaw sequence.

For additional details on incubations, C/N ratios, FTIR, and FT-ICR MS, see SI Materials and Methods.