A Novel Approach for High-Frequency in-situ Quantification of Methane Oxidation in Peatlands

Abstract

Methane (CH₄) oxidation is an important process for regulating CH₄ emissions from peatlands as it oxidizes CH₄ to carbon dioxide (CO₂). Our current knowledge about its temporal dynamics and contribution to ecosystem CO₂ fluxes is, however, limited due to methodological constraints. Here, we present the first results from a novel method for quantifying in-situ CH₄ oxidation at high temporal resolution. Using an automated chamber system, we measured the isotopic signature of heterotrophic respiration (CO₂ emissions from vegetation-free plots) at a boreal mire in northern Sweden. Based on these data we calculated CH₄ oxidation rates using a two-source isotope mixing model. During the measurement campaign, 74% of potential CH₄ fluxes from vegetation-free plots were oxidized to CO₂, and CH₄ oxidation contributed 20 ± 2.5% to heterotrophic respiration corresponding to 10 ± 0.5% of ecosystem respiration. Furthermore, the contribution of CH₄ oxidation to heterotrophic respiration showed a distinct diurnal cycle being negligible during nighttime while contributing up to 35 ± 3.0% during the daytime. Our results show that CH₄ oxidation may represent an important component of the peatland ecosystem respiration and highlight the value of our method for measuring in-situ CH₄ oxidation to better understand carbon dynamics in peatlands.

1. Introduction

Northern peatlands are an important component of the global carbon (C) cycle, as they are sinks of carbon dioxide (CO₂) and store about one third of the global soil organic C stock [1]. However, northern peatlands also emit the potent greenhouse gas methane (CH₄) at a rate that depends on the balance between CH₄ production and oxidation of CH₄ to CO₂. Detailed knowledge of in-situ CH₄ production and oxidation dynamics is thus key for understanding the contribution of CH₄ from northern peatlands to the atmosphere. Furthermore, as these processes are sensitive to climatic factors [2,3,4], this understanding is even more crucial in order to accurately predict the role of northern peatlands for the atmospheric radiative forcing under a changing climate.

No studies to our knowledge have measured in-situ CH₄ oxidation continuously with high temporal resolution in predominantly methanogenic systems, e.g., peatlands, in the field. Furthermore, current studies do not estimate the contribution of CO₂ resulting from CH₄ oxidation to total ecosystem respiration (ER, abbreviations are listed in

Table 1. List of abbreviations used in the article.

Term	Abbreviation
Concentration of headspace CH4 at time 0 in the incubation experiment for the fractionation factor	[CH4]0
Concentration of headspace CH4 at time t in the incubation experiment for the fractionation factor	[CH4]t
Ecosystem respiration	ER
Fractional contribution of CH4 oxidation to heterotrophic respiration	fCH4
Kinetic fractionation factor	α
Heterotrophic respiration	Rh
Isotopic 13C signature	δ13C
Isotopic 13C signature of ecosystem respiration	δ13CER
Isotopic 13C signature of headspace CH4 at times 0 in the incubation experiment for the fractionation factor	δ13CCH4,0
Isotopic 13C signature of headspace CH4 at times t in the incubation experiment for the fractionation factor	δ13CCH4,t
Isotopic 13C signature of heterotrophic respiration	δ13CRh
Isotopic 13C signature of pore water CH4	δ13CCH4,pw
Isotopic 13C signature of organic matter	δ13COM
Net ecosystem exchange	NEE
Organic matter	OM
Permil fractionation factor	Δ
Relative CH4 oxidation %	%CH4oxi
Soil organic matter	SOM

), and thus our understanding of the importance of this process in relation to the other CO₂ component fluxes is limited. Most peatland CO₂ models assume that all heterotrophic respiration (R_h) comes from soil organic matter (SOM) mineralization [5,6,7,8] while ignoring the contributions from CH₄ oxidation. Meanwhile, those models that estimate in-situ CH₄ oxidation lack data for validation [9]. This could potentially result in false parametrization and model predictions and overestimation of SOM mineralization with impacts on the modeling of the C balance. Another important limitation of previous studies is that they do not measure CH₄ oxidation continuously and with high temporal resolution, and as a result we know little of the temporal variation of CH₄ oxidation on both diurnal and seasonal time scales. This might further affect model prediction and estimation, particularly if input data is measured only during daytime and/or during the peak growing season. These issues therefore highlight the need for high temporal resolution CH₄ oxidation data in order to support process-based model development and to further improve our understanding of the peatland C cycle.

Current methods for estimating CH₄ oxidation in peatlands include laboratory incubations [10,11,12], often in combination with oxidation inhibitors [13,14,15], stable isotope techniques [10,16,17,18,19], methane profiles [20] and gas push-pull tests [21] which all have their strengths and weaknesses. For instance, the disadvantage of incubations is that they may estimate oxidation potentials instead of actual in-situ rates. Meanwhile, oxidation inhibitors are intrusive, do not allow for repeated measurements, and may partly inhibit CH₄ production [22,23]. The use of natural abundance stable isotope techniques is promising, although these techniques have traditionally been based on manual sampling which limits the spatial and temporal resolution of these measurements. In addition, natural abundance approaches depend on reliable estimates of the fractionation factors for CH₄ oxidation and diffusion [18,24]. Thus, there is a need for new methods that overcome these limitations and allow in-situ measurements of CH₄ oxidation at high temporal scales to better parameterize C and greenhouse gas dynamics in peatlands.

This study aims at developing a method for continuous high-frequency estimates of peatland CH₄ oxidation and the proportion of R_h that emanates from CH₄ oxidation. In spring 2014, we established an experimental setup in the field with automated flux chambers connected to a Picarro G1101-i isotopic CO₂ analyzer (Picarro Inc., Santa Clara, CA, USA) in an oligotrophic, minerogenic mire in northern Sweden. We measured massfluxes and isofluxes of ER and R_h (from plots with all photosynthetic biomass removed). By combining the isotopic signature (δ¹³C) of R_h, organic matter (OM) and pore water CH₄ in a two-source mixing model, we were for the first time able to partition the CO₂ originating from heterotrophic respiration into CO₂ resulting from CH₄ oxidation and mineralization of OM.

2. Materials and Methods

2.1. Site Description

The field site is an oligotrophic, minerogenic mire, Degerö Stormyr (64°11′ N, 19°33′ E), located near the town of Vindeln, Västerbotten County, Northern Sweden. The average annual temperature and average annual precipitation of the World Meteorological Organization (WMO) reference normal period 1961–1990 is 1.2 °C and 523 mm respectively (

Table 2. Climate and soil properties at Degerö Stormyr.

Properties	Value
Mean annual temperature	1.2 °C [25]
Mean annual precipitation	523 mm [25]
Growing season mean water table level	−4.1 cm [30]
Average depth of peat layer	3–4 m [28]
Peat C:N ratio 1	68.9 ± 1.9

¹ 0–30 cm depth.

) [25]. Long-term net ecosystem exchange (NEE) is −58 g C m⁻² year⁻¹ [26], average growing season CH₄ emission rates are ca. 1 to 5 mg CH₄ m⁻² h⁻¹ [27], and the net ecosystem carbon balance is ca. −20 to −27 g C m⁻² year⁻¹ [28]. Approximately half of the precipitation comes as snow and snow cover lasts for about six months (November to April). The peat layer is on average 3 to 4 m deep and the growing season water table generally varies between ca. 0 and 25 cm [26,28]. The vegetation consists mainly of Sphagnum majus Russ. C. Jens., Sphagnum balticum Russ. C. Jens., and Sphagnum lindbergii Schimp. Ex Lindb, Eriophorum vaginatum L., Trichophorum cespitosum L. Hartm., Vaccinium oxycoccos L., Andromeda polifolia L., Rubus chamaemorus L. [29].

2.2. Experimental Setup

The experimental setup was established in spring 2014 and consists of four replicate blocks containing three plots (1 × 1 m) each with different treatments, resulting in a total of 12 plots. Thus, each treatment/measurement type has four replicates. Each plot is equipped with an automated chamber (45 × 45 cm, 15 cm high) for flux measurements. A detailed description of the automated chamber system is provided by Järveoja et al. 2018 [30]. Briefly, two plots in each block are undisturbed where one is used for measurements of NEE using a transparent chamber and the other for measurements of ER using a dark chamber. In the third plot within each block, the aboveground vegetation, including the green parts (i.e., ~upper 5 cm) of the Sphagnum mosses, was removed in autumn 2013, and a 30 cm deep trench was cut along the plot sides using a handheld saw to prevent root activity inside the plots. These plots were used for measurements of CO₂ from heterotrophic activity (R_h) with a dark chamber. CH₄ fluxes were also measured at all plots. Each automatic flux chamber has measurements of air temperature 10 cm above the peat surface, and soil temperature at 2 and 10 cm depth. A water level sensor is also placed in each block.

2.3. Measurements of Mass and Isotopes of CH₄ and CO₂

Fluxes of CO₂ and CH₄ mass as well as the δ¹³C signature of the CO₂ concentrations were measured in the period of 18 to 27 July 2014. Massfluxes of CO₂ (NEE, ER, R_h) and CH₄ were measured using a greenhouse gas analyzer (model GGA-24EP, Los Gatos Research (LGR) Inc., San Jose, CA, USA) connected in a closed loop to the chambers. Isofluxes and massfluxes, used for Keeling plots, were measured using a Picarro G1101-i (Picarro Inc., CA, USA) placed downstream of the LGR. Analytical precision for in-situ carbon isotope analyses using the Picarro 1101-i instrument was 0.2‰ based on repeated analysis of known isotopic standards. An external pump was connected to the loop to provide continuous airflow. Chamber closure time lasted 18 min and was preceded and followed by one-minute flushing of the tubes with ambient air before onset of next measurement. Measurements took one hour per block and thus four hours for one round of measurements. The mean time of each four-hour measurement round is used to designate the measurement time point for a four-hour mean flux. For example, the measurement round taking place from 00.00 to 04:00 a.m. is reported for the time point 02:00 a.m.

2.4. Isotopic Signature of Soil Organic Matter and Pore Water CH₄

In order to obtain the δ¹³C signature of soil organic matter (δ¹³C_OM), eight peat cores from 0 to 30 cm depth, which represents the area of highest potential CH₄ production and oxidation activity [31], were collected from the mire in October 2015. The cores were cut into 2 cm sections and freeze-dried. The 2 cm sections were then ground in a ball mill and analyzed for δ¹³C signature on an elemental analyzer (Flash EA 2000, Thermo Scientific, Bremen, Germany) coupled to a continuous flow isotope ratio mass spectrometer (IRMS, Delta V Advantage, Thermo Scientific, Bremen, Germany). The standard deviation based on analysis of standards was <0.15‰ for δ¹³C.

On 3 and 27 August 2015, pore water was collected at 20 and 30 cm just outside the chamber frames in the non-vegetated plots and the vegetated plots used for dark measurements, as well as one location in the middle of the four blocks (n = 30). 2 mL of pore water was sampled using a syringe and transferred to N₂ flushed vials. Subsequently, 2 mL of gas was removed from the vials in order to equalize the pressure. The samples were stored at 5 °C until analysis for δ¹³C signature of CH₄ (δ¹³C_CH4,pw) on a Precon (Thermo Scientific, Bremen, Germany) and a gasbench (GasBench II, Thermo Scientific, Bremen, Germany) connected to a continuous flow IRMS (Delta V Advantage, Thermo Scientific, Bremen, Germany) on 28 and 29 September 2015. The standard deviation based on analysis of standards was <0.3‰ for δ¹³C.

2.5. Incubation Experiment to Determine the Fractionation Factor for CH₄ Oxidation

Four cores of 10 × 10 cm and 20 cm deep were collected in the Degerö mire on 8 October 2017. The cores were divided into four depths 0–5, 5–10, 10–15 and 15–20 cm below live vegetation, put in zip lock bags, brought back to the lab and placed in a freezer (−18 °C). Samples were kept frozen for two months and then preincubated at 4 °C for a month. After preincubation, 10 g field moist peat material from each sub core was transferred to 160 mL airtight glass bottles. Three replicates were made of each sample (to be incubated at three different temperatures) giving a total of 48 bottles (i.e., one sample per layer per core for each temperature). In addition, nine blanks (bottles containing 10 mL water as an analogue for field moist peat) were prepared. All bottles had ambient air inside and were given 0.05 mL pure CH₄ to feed the methanotrophs, and placed at 5 °C.

Six days after addition of CH₄, two replicate batches of bottles were placed at 10 and 15 °C respectively, while one replicate batch remained at 5 °C. After an additional three days, the bottles were flushed with technical air and given 0.12 mL pure CH₄, thereby raising headspace to app. 1000 ppm. Immediately after injecting CH₄, 0.5 mL of the headspace was sampled and transferred to 12 mL vials containing helium. Further headspace samples were taken at six, 12, 24, and 48 h and additionally at 96 h for the 0–5 cm interval in order to trace the oxidation rate (i.e., the decrease in headspace CH₄ concentration over time, and the concurrent change in δ¹³C signature of the CH₄). For ¹³C isotope analysis of CH₄ in the gas samples, a Finnigan MAT PreCon unit (Thermo Scientific, Bremen, Germany) was used for automated sample conversion and concentration. Briefly, sample CO₂ was removed by chemical adsorption succeeded by Pt-catalyzed oxidation of the CH₄-component to CO₂ that was subsequently trapped by duplicated cryogenic (liquid N₂) focusing. The isotopic analysis took place upon separation on a GC (HP 6890, equipped with 25 m long PoraPlot Q fused-silica column (32 mm i.d.), operated at 40 °C) coupled in continuous flow-mode to a Finnigan MAT Delta PLUS isotope ratio mass spectrometer. At the 24-h sampling, an additional 0.5 mL of the headspace was taken out and transferred to a 22 mL vial, where the concentration of CH₄ was determined on a gas chromatograph in order to preliminarily assess the oxidation rate and hence the required incubation time (data not shown). By the end of the experiment, the peat samples were dried for 48 h at 60 degrees and weighed for determination of dry weight.

2.6. Flux Calculation and Estimation of Flux Isotopic Signature

Massfluxes of NEE, ER, R_h and CH₄ were calculated from the linear change in gas concentration within the chamber headspace over time using the ideal gas law [30]. The linear slope was determined based on 10 concentration records over a 1 min 40 s calculation window (each record representing a 10 s mean of the 1 Hz sampling) moving stepwise (with one-point increments) over the chamber closure period. From these individual slopes, the one with the highest coefficient of determination (R²) was selected as the final slope for each flux measurement. All fluxes with an R² ≥ 0.90 (p < 0.001) were accepted giving a total of 428 CO₂ flux measurements and 437 CH₄ flux measurements over the ten-day period.

The δ¹³C source signature of respired CO₂ (corresponding to the δ¹³C signature of the source material) was determined using the Keeling plot approach [32]. The Picarro G1101-i logged measurements approximately every four seconds. From these data, one-minute averages were generated. Disregarding the first one minute average, we used linear regression analysis of δ¹³C and 1/[CO₂] for the remaining 17 min, with the y-axis intercept corresponding to the ¹³C signature of respired soil CO₂. Intercepts were excluded for regressions with slopes not significantly different from 0 (p > 0.05) and if the slopes were between −0.25 ppm min⁻¹ and 0.25 ppm min⁻¹ due to the uncertainty in Keeling estimates associated with very small fluxes.

2.7. Calculation of Fractionation Factor for CH₄ Oxidation

To account for preferential use of ¹²C over ¹³C by methanotrophs [33,34], we used a kinetic fractionation factor for CH₄ oxidation (hereafter referred to as “fractionation factor” or α). The fractionation factor describes the fractionation against the heavy isotope, where α = 1 means no fractionation and α > 1 means that fractionation is occurring with product becoming depleted in the heavy isotope and the substrate becoming enriched. α was calculated by the following equations [24] based on data from the incubation experiment:

$$\mathsf{\alpha }=\frac{1}{\left(\mathrm{m}+1\right)},$$

(1)

where m is:

$$\mathrm{m}=\frac{{\mathsf{\delta }}^{13}{\mathrm{C}}_{\mathrm{CH}4,\mathrm{t}}-{\mathsf{\delta }}^{13}{\mathrm{C}}_{\mathrm{CH}4,0}}{\ln \frac{{\left[{\mathrm{CH}}_4\right]}_{\mathrm{t}}}{{\left[{\mathrm{CH}}_4\right]}_0}},$$

(2)

and δ¹³C_CH4,0 and δ¹³C_CH4,t designates the δ¹³C signature of the headspace CH₄ at times 0 and t, and [CH₄]₀ and [CH₄]_t is the concentration of the headspace CH₄ at times 0 and t. In practice, since we had more than two time points, m was calculated as the slope of a linear regression between the difference in isotopic signatures (δ¹³C_t–δ¹³C₀) and natural logarithm of the fraction between remaining and initial headspace CH₄ concentration (ln([CH₄]_t/[CH₄]₀) [24]. The permil fractionation factor Δ could then be calculated from the α [35]:

$$\Delta =\frac{\mathsf{\alpha }-1}{1000}.$$

(3)

CH₄ oxidation rates were also calculated for the incubations, using linear regression. Only significant fractionation factors (slope different from 0, p < 0.05) with a corresponding flux <0 were included in the analysis.

2.8. Mixing Model

The fractional contribution of CH₄ oxidation and OM mineralization to total R_h was calculated using a two-source mixing model [35]:

$${\mathrm{f}}_{\mathrm{CH}4}=\frac{{\mathsf{\delta }}_{\mathrm{sample}}-{\mathsf{\delta }}^{13}{\mathrm{C}}_{\mathrm{OM}}}{{\mathsf{\delta }}^{13}{\mathrm{C}}_{\mathrm{CH}4,\mathrm{pw}}-\Delta -{\mathsf{\delta }}_{\mathrm{OM}}},$$

(4)

where f_CH4 is the fraction of the heterotrophic respiration contributed by oxidation of CH₄, δ_sample is the δ¹³C signature of the heterotrophic respiration, δ¹³C_CH4,pw is the isotopic signature of dissolved CH₄ in pore water, and δ¹³C_OM is the isotopic signature of the OM. We used the mean δ¹³C signature of OM in 0–30 cm depth and the mean δ¹³C signature of CH₄ in 20 and 30 cm depth for the mixing model. We used the mean fractionation factor for methane oxidation in peat at 0–20 cm depth (Δ = 54.0‰) and across three incubation temperatures (5, 10 and 15 °C) as the statistical test showed no significant effect of neither depth nor temperature. The fractionation factors are subtracted from the δ¹³C_CH4,pw because the oxidation of CH₄ discriminates against the heavy isotope (¹³C) and thus the resulting CO₂ is depleted in ¹³C compared to the source CH₄. In other words, the δ¹³C signature of the CO₂ produced during CH₄ oxidation is more negative than the source CH₄.

The relative contribution of CH₄ oxidation (%CH_4oxi) is calculated as [36]:

$${\%\mathrm{CH}}_{4\mathrm{oxi}}=\frac{{\mathrm{Oxidized}\mathrm{CH}}_4\mathrm{flux}}{{\mathrm{Oxidized}\mathrm{CH}}_4\mathrm{flux}+{\mathrm{measured}\mathrm{CH}}_4\mathrm{flux}}\ast 100,$$

(5)

and the oxidized CO₂ flux is calculated by multiplying f_CH4 and R_h.

2.9. Data Presentation and Statistics

The differences in fractionation factor between the four depths, and three temperatures were tested with a mixed linear model in R version 3.5.1 (R Core Team, Vienna, Austria) using depth and temperature as fixed effects and core as random effect. The model was reduced stepwise using ANOVA.

All data are presented as mean and standard error. Means and errors for the time series of relative contribution of CH₄ oxidation to R_h as well as time series of isotopic signatures of ER and R_h are weighted averages based on the four-hour averages, as some missing values during nighttime would skew an overall average towards daytime values. Figures were made in Sigmaplot 13.0 (Systat Software Inc., San Jose, CA, USA) and R.

3. Results and Discussion

3.1. Isotopic Signatures of CO₂ Fluxes

Measurements of mass- and isofluxes of ER and R_h as well as fluxes of NEE and CH₄ from the undisturbed plots in the mire were carried out from 18 to 27 July 2014 (Figure 1 and Figure S1). During this period, the water table level dropped from 0.08 to 0.16 m below mire surface, and the average daily air temperature at 10 cm above mire surface varied between 8.2 and 30.9 °C (Figure S2). The δ¹³C signature of source CO₂ in both the ER and R_h fluxes showed strong diurnal cycles (Figure 1). The δ¹³C signature of ER (δ¹³C_ER) varied between −52 and −22‰ with minimum values occurring during mid-day. The average of δ¹³C_ER for the period was −32.3 ± 1.0‰ (±standard error). The four-hour means of δ¹³C signature of R_h (δ¹³C_Rh) usually peaked at 2 am (Figure 1b) with a maximum δ¹³C_Rh of −22‰, whereas the minimum δ¹³C_Rh was −99‰. Average δ¹³C_Rh was −49.2 ± 2.3‰.

The observed diurnal pattern of δ¹³C_ER indicates that given the less depleted source during night, ER mainly results from OM mineralization, whereas during the day, the more depleted signatures of CO₂ suggest an additional process contributing to ER. This trend is even more apparent in the vegetation-free R_h plots. We suggest that the source of these negative δ¹³C values is the result of methanotrophs oxidizing CH₄ with an average δ¹³C of −67.2‰ (see below) in the peat pore water. The fact that some δ¹³C_Rh values were lower than the average δ¹³C signature of the CH₄ may be due to the uncertainty associated with the estimation of the keeling intercepts (the standard error of the intercept is on average 8.3‰ for R_h) or a small difference in δ¹³C_CH4,pw between the 2014 measurement period and the pore water sampling done in 2015. However, it is also likely due to a large contribution from CH₄ oxidation and the fractionation occurring during the oxidation process, which lowers the δ¹³C signatures of the resulting CO₂ relative to the substrate CH₄ [33].

3.2. Mixing Model

We used a two-source mixing model to quantify the relative contributions of OM mineralization and CH₄ oxidation to total R_h fluxes. The δ¹³C signature of OM (δ¹³C_OM) integrated over 0 to 30 cm depth was −27.4‰ (Figure S3) and was used to represent the δ¹³C signature of OM mineralization in the mixing model. The average δ¹³C signature of the pore water CH₄ (δ¹³C_CH4,pw) was −67.2‰ and represents the δ¹³C signature of CO₂ originating from CH₄ oxidation in the mixing model. We consider the δ¹³C_CH4,pw in 2015 a good representation of the δ¹³C_CH4,pw from 2014 due to little variation between years in these depths below the water table (−73.4 ± 0.5‰ on 8 August 2017 and −68.1 ± 0.6‰ on 26 July 2018). In order to account for the fractionation taking place when CH₄ is oxidized to CO₂, we subtracted the measured fractionation factor Δ = 54.0 ± 3.4‰ (n = 28) from the δ¹³C_CH4,pw (Figure S4). Our fractionation factor is within the range reported in the literature [19,24,33,37,38,39] though slightly on the high end. Over the measurement period (18 to 27 July 2014) CH₄ oxidation contributed 20 ± 2.5% of R_h (Figure 2a) and 10 ± 0.5% of ER (Figure S1, assuming that no additional oxidation occurs in vegetated plots due to production and transport of oxygen by plants). At the same time, if ignoring any difference in transport rate between CH₄ and CO₂, 74% of the produced CH₄ in the R_h plots were oxidized to CO₂ within the assessed time period. We made a sensitivity analysis to assess the effect of the calculated contribution of CH₄ oxidation to R_h using the minimum and maximum measured δ¹³C_OM and δ¹³C_CH4,pw as well as minimum and maximum fractionation factors from the literature (Figures S7 and S8). The analysis showed that the highest and lowest estimates across this period resulted in methane oxidation contributing between 16 and 57% of R_h. Our result (20%) is therefore in the lower range, making it more likely that we underestimate the relative contribution of methane oxidation to soil CO₂ effluxes. Using this novel approach our results showed that during a relatively warm and dry period (Figure S2), CH₄ oxidation likely reduced CH₄ emissions and contributed considerably to R_h in this boreal peatland. Although our approach appears promising in quantifying methane oxidation in real time in the field, our results also stress that more high-frequency measurements are needed to quantify the importance of this process for various mire plant communities and during various stages of the growing season associated with differences in plant phenology, water table levels, and soil temperatures.

3.3. Diurnal Variation

We also observed a diurnal variation in the relative contribution of CH₄ oxidation to total CO₂ efflux from the R_h plots (Figure 2b). Previously, speculations around diurnal variation in peatland CH₄ oxidation have been inferred based on diurnal variation of CH₄ fluxes [40]. However, this study shows diurnal variation the isotopic signature of R_h and thus likely in CH₄ oxidation. During nighttime, the contribution of CH₄ oxidation to R_h seemed negligible, while during the day, CH₄ oxidation appeared to contribute up to 35 ± 3.0%. In order to assess whether the diurnal variation was caused by a change in CH₄ oxidation rather than increased OM mineralization during night, we calculated the CO₂ fluxes associated with the two processes (Figure S5). These fluxes show that even though mineralization like total R_h appears to be higher during night, there still seems to be a distinct diurnal pattern in the CH₄ oxidation. We suggest that the diurnal variation is driven by a combination of changes in soil temperature (on average 7.1 and 2.6 °C difference between min. and max. temperatures in 2 and 10 cm depth respectively, Figure S2) and a decreasing water table. The water table follows a staircase-like trend with lowering during day and plateauing at night (Figure S2) and is likely driven by enhanced evapotranspiration during the day. This would lead to daytime O₂ intrusion into areas in the soil profile with high CH₄ concentration (enhancing CH₄ oxidation rates), as well as to the possible release of CO₂ with a depleted signature (originating from CH₄) due to enhanced diffusion rates [41]. It is important to point out that the diurnal variation in the isotopic signature of CO₂ from vegetated plots may also partly be caused by increased input of O₂ to the rhizosphere by photosynthesizing plants. Although it was beyond the scope of this study, the observed diurnal variation in methane oxidation highlights the need for depth specific O₂ and CH₄ measurements in order to better understand the drivers responsible for temporal variation in methane oxidation in the field. The diurnal variation in δ¹³C signatures of R_h also highlights that bias in the diurnal sampling protocol can influence the results as e.g., only daytime sampling would give values highly biased towards more depleted δ¹³C signatures. It should also be noted, that there was a buildup of CO₂ and CH₄ in the air above the mire surface during each night of the measurement period, which may have led to some overestimation of the flux isotopic signature [42,43] (see Figure S6).

Our results show that CH₄ oxidation may contribute considerably to R_h fluxes, and thus highlights the importance of including this process in peatland CO₂ models in order to better predict and understand peatland C dynamics. Our new approach for measuring CH₄ oxidation creates the opportunity for future studies to provide the necessary data to validate and improve these models. Furthermore, studies estimating ecosystem respiration in boreal peatlands based on partitioning of eddy covariance data assume that OM mineralization and plant respiration are the only sources for respired CO₂ and commonly relate ecosystem respiration to only one factor; namely temperature [44,45]. However, our results show that in peatland ecosystems a considerable amount of R_h fluxes could be derived from CH₄ oxidation which is controlled by additional factors (e.g., water table level [46] and CH₄ and O₂ availability [47]), that differ from the main factors controlling OM mineralization and plant respiration.

3.4. Methodological Limitations

In our two-source mixing model, we assume that fractionation during diffusion of CH₄ and CO₂ would not strongly influence our results for the following reasons (1) our estimates of the source δ¹³C signatures integrate over the soil profile, (2) there was no difference in the δ¹³C_CH4,pw at depths 20 and 30 cm, and (3) CH₄ diffusion in water saturated soil causes negligible fractionation [24] and in addition other non-fractionating transport processes such as pressure gradients and near surface layer air flow might have contributed to the total flux [48]. We also assume that OM mineralization and aerobic CH₄ oxidation are the only two processes influencing the isotopic signature of ¹³CO₂ from the R_h plots. Although it is possible that anaerobic CH₄ oxidation could potentially take place [49,50], we consider this process of minor importance in this nutrient-poor ecosystem. However, if anaerobic CH₄ oxidation was occurring, it would produce CH₄ less depleted in ¹³C due to a lower fractionation factor [50] and thereby, if anything, our estimates of contribution of CH₄ oxidation to total R_h would be underestimated. We also acknowledge that CO₂ is produced during methanogenesis. Theoretically, both hydrogenotrophic (based on carbohydrate fermentation and including the step of H₂ production) and acetoclastic methanogenesis produces equimolar amounts of CO₂ and CH₄, and as the CH₄ is depleted compared to the substrate, the CO₂ must be equally enriched [51]. An estimation by Corbett and colleagues [51] based on a substrate signature of −26‰ and a resulting CH₄ signature of −60‰, suggested a ¹³C signature of 8‰ for the CO₂ produced during methanogenesis [51]. In the current study, we were not able to include the CO₂ from methanogenesis in our mixing model, and therefore our estimation of the contribution of CH₄ oxidation to total R_h is potentially underestimated. This is because the CO₂ from methanogenesis raises the isotopic signature of the total pore water CO₂ pool and emitted CO₂ and thereby causes the contribution from CH₄ oxidation to appear smaller.

We used the method of plant removal and plot trenching for estimating R_h [52]. As is the case for other methods measuring R_h, this approach has some shortcomings. For instance, the removal of the vegetation causes a reduction in supply of rhizodeposits and possibly a lower input of O₂ to the soil (due to elimination of downward plant mediated transport of O₂). For CH₄ oxidation, the latter is mostly relevant when the water table is high and thus limits the diffusion of O₂ from the atmosphere into soil, as was not the case during our measurement campaign. The decrease in concentration of both O₂ and rhizodeposits may however be counteracted to some extent by lateral transfer with moving water, and we consider this issue of minor importance in our study. The plant removal in the R_h plot also eliminated CH₄ transport by plants, and thus the CH₄ fluxes from these plots were on average 16% lower than from the vegetated plots. In addition, the potential for CH₄ oxidization in the vegetation-free plots might also be somewhat lower to due to the removal of the upper moss layer with its associated methanotrophic communities [53]. However, the oxidation in these plots was on average 74% of potential fluxes, which is marginally higher than the 70% oxidation in vegetated plots, and thus we would argue that vascular plant CH₄ transport and moss-associated CH₄ oxidation plays only a minor role for net CH₄ oxidation in our experiment.

In this study we used an average of the measured fractionation factors across peat depth and incubation temperature as we found no statistical effect of these parameters. We did, however, find a correlation between oxidation rate and fractionation factor (Figure S9), although we were not able to include this information into the mixing model at this point, as we are using the model to estimate the oxidation rate. We acknowledge however that the fractionation factor can vary in response to different parameters such as CH₄ concentration and CH₄ oxidation rate [54,55]. The fractionation factor may be positively correlated with CH₄ starting concentration of in incubations [54] highlighting the importance of matching the CH₄ starting concentration of incubations with conditions found in the field. The fractionation factor is also influenced by the fraction of active methanotrophs [55] and thus a decreasing water table could initially influence the fractionation factor when exposing potentially dormant methanotrophs to optimal conditions. However, according to our sensitivity analysis (Figure S8) this factor alone cannot explain the observed diurnal variation in CH₄ oxidation. We encourage future studies for improvements on this method by including this component.

4. Conclusions

In this study, we present a new method for continuous, high-frequency in-situ quantification of CH₄ oxidation in peatlands. Previous studies from wetlands and lakes have quantified CH₄ oxidation using the δ¹³C (and δD) of pore water or lake water CH₄ [19,33,37] and CH₄ flux [56]. However, our method is the first, to our knowledge, that uses high temporal resolution isotopic measurements of ¹³CO₂ in R_h fluxes based on automated chamber measurements on vegetation-free plots to quantify the relative contribution of CH₄ oxidation to the heterotrophic and ecosystem respiration fluxes in the field. Thus, our approach creates an unprecedented opportunity to study the temporal dynamics and controls of CH₄ oxidation in peatland ecosystems. In addition, we observed a diurnal pattern in the δ¹³C signatures of heterotrophic respiration suggesting a high contribution of CH₄ oxidation to CO₂ fluxes during daytime and a negligible contribution during nighttime. Overall, our novel approach of directly measuring the isotopic composition of R_h at high temporal resolution provides unique insight into the effect of CH₄ oxidation on CO₂ and CH₄ fluxes, which is crucial for further developing process-based models and improving our understanding of peatland C dynamics.