Fate of methane in canals draining tropical peatlands

Abstract

Tropical wetlands and freshwaters are major contributors to the growing atmospheric methane (CH₄) burden. Extensive peatland drainage has lowered CH₄ emissions from peat soils in Southeast Asia, but the canals draining these peatlands may be hotspots of CH₄ emissions. Alternatively, CH₄ oxidation (consumption) by methanotrophic microorganisms may attenuate emissions. Here, we used laboratory experiments and a synoptic survey of the isotopic composition of CH₄ in 34 canals across West Kalimantan, Indonesia to quantify the proportion of CH₄ that is consumed and therefore not emitted to the atmosphere. We find that CH₄ oxidation mitigates 76.4 ± 12.0% of potential canal emissions, reducing emissions by ~70 mg CH₄ m⁻² d⁻¹. Methane consumption also significantly impacts the stable isotopic fingerprint of canal CH₄ emissions. As canals drain over 65% of peatlands in Southeast Asia, our results suggest that CH₄ oxidation significantly influences landscape-scale CH₄ emissions from these ecosystems.

Canals draining vast areas of peatland in Southeast Asia may be hotspots for methane emissions. Perryman et al. surveyed dozens of canals in Indonesia and found that methane-eating microbes reduce emissions from these waterways by over 50%.

Introduction

Wetlands and freshwaters contribute ~30–55% of global CH₄ emissions¹, with significant emissions from tropical ecosystems^2–4. Rising CH₄ emissions from tropical wetlands due to temperature and rainfall anomalies have contributed substantially to the growing atmospheric CH₄ burden^5–8. In addition to climate change, ongoing disturbances to tropical wetlands like deforestation, drainage, fertilizer application, and slash and burn agriculture, as well as rewetting and restoration efforts, stand to impact their contribution to the global CH₄ budget^9–14. However, the impact of tropical wetland disturbance on CH₄ cycling is not well understood.

In Southeast Asia, wetland disturbance has heavily impacted peatlands, destabilizing the large pool of soil carbon stored in the peat soils of this region^15,16. Peatlands in Southeast Asia have undergone extensive drainage for oil palm plantations, timber, and other agriculture through the construction of canals^17,18 that lower the water table and therefore lower CH₄ emissions from peat soils^19–21. Instead, the CH₄ produced in peat soils is transported into canals via lateral flow^22,23, increasing the relative importance of drainage canals as a source of CH₄ emissions^24,25. Canals can represent over 50% of peatland CH₄ emissions in Southeast Asia²⁶, but estimates of the magnitude of drainage canal CH₄ emissions vary by several orders of magnitude^27,28. Given that drainage increases the importance of aquatic carbon fluxes from tropical peatlands²⁹, and the large uncertainty around canal CH₄ emissions, greater understanding of the key controls and mechanisms driving canal CH₄ emissions is needed to constrain their role in tropical peatland CH₄ budgets.

One process that strongly influences freshwater CH₄ emissions is microbial oxidation of CH₄ to carbon dioxide. In other tropical freshwaters (e.g., rivers, lakes) CH₄ oxidation attenuates CH₄ emissions by 40 to nearly 100%^23,30–32. The fraction of CH₄ transported into canals from drained peatlands that is oxidized instead of emitted is highly uncertain, as are the factors that mediate CH₄ oxidation in drainage canals. For example, both aerobic and anaerobic methanotrophic microbiota are found in tropical freshwaters^32–35. As variation in canal water depth and discharge can impact dissolved oxygen in canals^36,37, examining the relationship between CH₄ oxidation and dissolved oxygen could inform how CH₄ oxidation in canal waters may vary over space and time. Constraining the importance of CH₄ oxidation in canals draining tropical peatlands is a key step to improving our understanding of the processes controlling CH₄ emissions from these ecosystems, especially in the densely drained peatlands of Southeast Asia where canals can have a disproportionate impact on landscape-scale CH₄ emissions.

Here, we address how much CH₄ transported from drained tropical peatlands into canals is oxidized instead of emitted to the atmosphere. We quantified the percent of CH₄ oxidized in 34 canal reaches that drain peat soils under varying land uses across West Kalimantan, Indonesia (Fig. 1, Supplementary Data 1) through shifts in the δ¹³C composition of CH₄ during incubation experiments of canal waters and from field observations of in situ canal CH₄ concentration and δ¹³C-CH₄ (Fig. S1). We find that 47.3-91.3% of CH₄ transported into canals from drained peatlands is oxidized instead of emitted. The fraction of CH₄ that is oxidized is influenced by factors including dissolved oxygen, vegetation, and canal water depth. Overall, our results suggest that CH₄ oxidation substantially attenuates CH₄ emissions from canals, and as a result, may be a significant control of landscape-level CH₄ emissions from drained peatlands in Southeast Asia.

Fig. 1. Drainage canal waters were collected in West Kalimantan, Indonesia to measure methane oxidation.

A Study area with drainage canals shown as dark blue lines. B shows location of study within insular Southeast Asia. Canal sample locations marked in red points. C–F Zoomed in view of green, purple, teal, and orange boxes in panel A, respectively, showing sample locations. The base map layers in Panels A–F are available from OpenStreetMap (openstreetmap.org/copyright), available under the Open Database License. G Examples of canals in varying land use context and canals with and without aquatic vegetation. H Overview of study methods to estimate CH₄ oxidation in drainage canals. Created in BioRender. Perryman, C. (2024) BioRender.com/c12r203.

Results and Discussion

CH₄ consumption and isotopic fractionation observed during incubations

To confirm CH₄ oxidation occurs in drainage canals, we incubated water from 13 canals at in situ dissolved CH₄ and oxygen concentrations and measured the change in dissolved CH₄ and δ¹³C-CH₄ over time. On average, 53.8 ± 25.6% of the initial CH₄ was consumed over the incubation period (17.6–99.7%) and δ¹³C-CH₄ increased by 19.8 ± 17.7‰ (2.1–67.8‰, Fig. 2A, Table S1). The increase in δ¹³C-CH₄ observed across incubated waters confirmed the loss of CH₄ was from microbial oxidation, as CH₄ oxidation leaves residual CH₄ enriched in ¹³C^38,39. Methane oxidation rates were variable across incubated waters (0.03–5.6 μmol CH₄ L⁻¹ d⁻¹, Table S1) and were strongly influenced by initial CH₄ concentration (Fig. S2). Neither CH₄ production nor δ¹³C-CH₄ depletion was observed in the canal waters during the incubation period.

Fig. 2. Methane consumption and resulting stable isotope fractionation in incubated canal waters.

A Across incubated waters, δ¹³C-CH₄ increased as the percent of initial CH₄ consumed increased. Each data point shows the mean change over ~50 hours of incubation ± standard error of replicates (Table S1). B Histogram of α_ox values calculated from incubation data. Source data are provided as a Source Data file.

From these data we calculated the first empirically derived isotopic fractionation factors for CH₄ oxidation⁴⁰ (α_ox) in peat-draining freshwaters. Ecosystem-specific values for α_ox are critical to estimating the percent of CH₄ that is oxidized rather than emitted from the natural environment^41,42. Mean α_ox was 1.022 ± 0.009 across the incubated canal waters (range: 1.002–1.039; Fig. 2B). The range of α_ox encompasses past observations from northern and temperate freshwaters incubated under in situ dissolved CH₄ and oxygen concentrations and temperature^42,43, as well as results from incubations of soil from subtropical rice paddies⁴⁴ (α_ox of 1.025–1.033) that are often used in estimates of CH₄ oxidation in tropical freshwaters^30,32. While CH₄ oxidation rates varied with initial CH₄ concentration, we did not observe a correlation between α_ox and initial CH₄ concentration, nor α_ox and CH₄ oxidation rate (Fig. S2). Recent work in temperate lakes identified temperature, pH, and dissolved O₂ as potential controls on α_ox⁴³. Of these factors, α_ox was only weakly positively correlated with the initial dissolved O₂ present in each of the incubated waters (p = 0.07). α_ox did not vary between surface and bottom waters of the subset of canals sampled at two depths for incubation experiments (1.024 ± 0.006 vs. 1.023 ± 0.012, n = 4 canals). As we did not find significant environmental correlates of α_ox, we used the mean value to estimate in situ CH₄ oxidation as discussed below.

Oxidation mitigates the majority of drainage canal CH₄ emissions

We find that the majority of CH₄ transported into canals from drained tropical peatlands is oxidized instead of emitted to the atmosphere. Using the laboratory-derived α_ox values, measurements of in situ canal water δ¹³C-CH₄ from 34 canal reaches (Supplementary Data 1), and measurements of source porewater δ¹³C-CH₄ (Supplementary Data 2), we estimated that CH₄ oxidation consumes 76.4 ± 12.0% of CH₄ transported into canals (range: 47.3–91.3%). Considering the standard deviation of α_ox shifts the mean percent oxidized by ~10%, ranging from 65.5 ± 12.5% to 89.3 ± 8.9% (Figure. S3). Similarly, considering the standard deviation of the porewater source δ¹³C-CH₄ measurements, the mean percent oxidized could range from 68.2 ± 16.1% to 82.4 ± 8.9% (Fig. S3). Our estimate of the fraction of CH₄ transported into canals that is oxidized instead of emitted is consistent with Somers et al. (2023), who estimated that 70% of CH₄ was oxidized in a canal draining a tropical peatland in Brunei using a reactive transport model. These results are also consistent with past work indicating that oxidation consumes ~80% of CH₄ in blackwater rivers in the Amazon³² that have low pH and high concentrations of aromatic-rich, humic-like dissolved organic carbon like the canals in our study region^29,45. Compared to previous work in lotic systems that use a similar approach as employed in our study, we find that oxidation mitigates a higher proportion of potential CH₄ emissions in drainage canals than in headwater streams in temperate forests⁴⁶ (55.6 ± 2.8%) or boreal peatlands⁴⁷ (~60%).

Canal water dissolved CH₄ concentration and δ¹³C-CH₄ across our study region supports our finding that CH₄ oxidation limits CH₄ release from canals. Dissolved CH₄ concentration and δ¹³C-CH₄ in canal waters ranged from 0.05 to 31.6 μM and −71.9 to −34.1‰, respectively (Fig. 3B), and dissolved CH₄ decreased with increasing δ¹³C-CH₄ (R² = 0.43, p < 0.001, Fig. S4). Previous observations in tropical river networks³² also observed a negative relationship between the concentration of CH₄ in river waters and δ¹³C-CH₄. In these rivers δ¹³C-CH₄ also had a positive relationship with gene markers for methanotrophic bacteria, indicating that variation in CH₄ concentration and δ¹³C-CH₄ is influenced by CH₄ oxidation. The consistent relationship between CH₄ concentration and δ¹³C-CH₄ observed across the drainage canals in our study and these tropical rivers supports the idea that differences in dissolved CH₄ concentrations between canal reaches are influenced by CH₄ oxidation.

Fig. 3. Survey of drainage canal CH₄ concentrations and δ¹³C-CH₄ reveal the impact of CH₄ oxidation on canal CH₄ emissions.

A Curve showing the relationship between canal water δ¹³C-CH₄ and estimated percent CH₄ oxidized across the mean (black line) and ± 1 standard deviation (shaded region) of the laboratory derived α_ox value. B, C Surface water δ¹³C-CH₄ and dissolved CH₄ concentration across the studied canals (n = 34). D Estimates of the percent of CH₄ oxidized versus estimated diffusive CH₄ flux across the studied canals. For panels B–D each dot represents a canal. The shaded region of panel D represents the 95% confidence interval associated with the linear relationship. Dissolved CH₄ concentration and estimated diffusive CH₄ flux are shown on a log₁₀ scale in panels C and D. Source data are provided as a Source Data file.

It is unlikely that CH₄ concentration in canal waters is dictated only by the amount of CH₄ originally transported into canals from the surrounding landscape, including CH₄ produced in peat soils and canal sediments. Methane produced in ombrotrophic tropical peat soils is highly depleted in ¹³C^23,48. Unlike in lakes where δ¹³C-CH₄ in littoral sediments and adjacent groundwater can differ by more than 10‰⁴⁹, porewater δ¹³C-CH₄ has not been shown to differ between canal bottoms and adjacent peat soils²². Porewater δ¹³C-CH₄ collected from 6 profiles (40 to 150 cm depth) located alongside canal waters in our study region had a mean δ¹³C-CH₄ of −85.0 ± 5.9‰, which was consistently more depleted than any observed canal δ¹³C-CH₄ value (Supplementary Data 1, Supplementary Data 2). Porewater δ¹³C-CH₄ varied more between sample depths within each profile than between profiles collected across the landscape, suggesting source δ¹³C-CH₄ is similarly depleted in ¹³C throughout the study region. Methane production in the water column could also influence canal water CH₄ concentration and δ¹³C-CH₄. However, this is unlikely to explain our results because we did not observe net CH₄ production in any of the laboratory incubations of canal waters, as CH₄ concentration decreased and δ¹³C-CH₄ increased in all incubated waters (Fig. 2A, Table S1). If canal water CH₄ concentration were influenced solely by the total amount of CH₄ produced and then transported into canal waters, we would expect canal water δ¹³C-CH₄ to be similarly depleted across canals and not vary systematically with dissolved CH₄ concentration. Given that CH₄ concentrations varied ~600-fold alongside a ~40‰ range in δ¹³C-CH₄, our results indicate that CH₄ oxidation has a significant influence on canal water CH₄ concentration and δ¹³C-CH₄.

As CH₄ oxidation was a major control of canal water CH₄ concentration, diffusive CH₄ emissions were also strongly influenced by the percent of CH₄ oxidized. Diffusive CH₄ emissions estimated from dissolved CH₄ concentration (Supplemental Text 1) ranged from 1.0 to 761.8 mg CH₄ m⁻² d⁻¹ (mean = 72.2 ± 151.2; median = 18.0) and decreased as the percent of CH₄ oxidized increased (R² = 0.54, p < 0.001; Fig. 3D). Measurements of CH₄ emissions from floating chamber deployments at a subset of study sites (n = 12 canals, mean = 94.9 ± 142.3 CH₄ m⁻² d⁻¹, median = 33.0, Supplementary Data 1) also indicated a negative relationship between CH₄ emissions and the percent of CH₄ oxidized (Fig. S5). By back-calculating what diffusive CH₄ flux would be in the absence of oxidation, we estimate that CH₄ oxidation reduces drainage canal CH₄ emissions by a mean of 136.8 ± 154.1 mg CH₄ m⁻² d⁻¹ (range: 9.9–684.2). Given the skewed distribution of dissolved CH₄ concentrations that underlie this estimate, the median value of 72.1 mg CH₄ m⁻² d⁻¹ (IQR: 36.2–173.6) may be a more robust estimate of the emissions attenuated by CH₄ oxidation. Overall, our results provide evidence suggesting that CH₄ oxidation mitigates the majority of potential CH₄ emissions from canals on the landscape.

Controls on CH₄ oxidation in drainage canals

Of the studied controls on CH₄ oxidation, dissolved oxygen and aquatic vegetation had the most significant influence on the percent of CH₄ oxidized in canals as determined by canal water δ¹³C-CH₄. We found that the percent of CH₄ oxidized increased and dissolved CH₄ concentration decreased with the concentration of dissolved oxygen at the canal water surface (0-10 cm; p < 0.05, Fig. 4A, Table S2). The relationship between dissolved oxygen and CH₄ oxidation is consistent with oxidation mediated by aerobic methanotrophic bacteria, as has been observed in other stream and river networks^32,46. While all canals had low dissolved oxygen (0.2 to 2.3 mg L^-1), methanotrophic bacteria of the order Methylococcales have been shown to have the genetic potential for survival and methanotrophic activity in low oxygen environments⁵⁰. Abundant Methylococcales have been identified in hypoxic tropical freshwaters where paired measurements of dissolved CH₄ concentration and δ¹³C-CH₄ indicate ongoing CH₄ oxidation^34,35. Our results further support the idea that aerobic CH₄ oxidation occurs in tropical freshwaters with low dissolved oxygen.

Fig. 4. Controls on CH₄ oxidation in drainage canals.

A Dissolved oxygen in the surface waters (0–10 cm) of drainage canals versus the percent of CH₄ oxidized. Each point represents a canal (n = 34). B Boxplot of the percent of CH₄ oxidized in open water (light blue, n = 19) and vegetated (green, n = 15) canals. Within each box the black lines represent median values and the height of the boxes represent the interquartile range. Error bars extend up to 1.5 times the interquartile range. The number in each box represents the mean ± 1 standard deviation of the percent of CH₄ oxidized for each group. Source data are provided as a Source Data file.

Furthermore, the percent of CH₄ oxidized was higher in vegetated canals than those with open water (p = 0.01, Fig. 4B). Vegetation may enhance CH₄ oxidation via radial oxygen loss from roots^51,52 or via oxidation by epiphytic methanotrophs in submersed plants⁵³. Although we did not observe a significant difference in dissolved oxygen based on the presence of aquatic vegetation (p > 0.05, Table S3), oxygen delivered to the water column by aquatic vegetation is likely rapidly consumed by methanotrophs or by competing aerobic heterotrophs as deposition of more labile organic carbon by aquatic vegetation could stimulate heterotrophic respiration in canal waters²⁹. Lower CH₄ concentrations and more enriched δ¹³C-CH₄ in vegetated canals could alternatively be explained by plant-mediated emissions⁵⁴, which reduce CH₄ concentration and enrich the δ¹³C of residual CH₄ due to the isotopic fractionation of plant-mediated transport⁵⁵. The deposition of labile organic matter from vegetation could also stimulate acetoclastic methanogenesis, which like CH₄ oxidation would contribute towards larger δ¹³C-CH₄ in vegetated canals³⁹. However, acetoclastic methanogenesis likely contributes little to the δ¹³C-CH₄ in vegetated canals because hydrogenotrophic methanogenesis has been identified as the dominant pathway in the ombrotrophic tropical peatlands of Southeast Asia²³ and the Americas^48,56. Disturbance in peatlands in Southeast Asia has been observed to increase the abundance of plant functional types associated with acetoclastic methanogenesis, like graminoids, but this shift does not appear to increase the abundance of acetoclastic methanogens⁵⁷. While we cannot rule out the possible influence of acetoclastic methanogenesis on canal water δ¹³C-CH₄, the lower dissolved CH₄ concentration in vegetated canals (p = 0.02, Table S3) lends more support to the idea that vegetation enhances CH₄ oxidation rather than acetoclastic CH₄ production in canals.

Given that higher dissolved oxygen and the presence of aquatic vegetation were observed in canals with a shallower water depth (Fig. S6), canal water depth may indirectly mediate CH₄ oxidation in drainage canal waters. Overall, dissolved oxygen in the surface water of canals (0–10 cm) decreased with the depth of water present in the canal (Kendall’s τ = −0.41, p < 0.05, Fig. S6). Dissolved CH₄ concentration, and therefore estimated diffusive emissions, also had a weak but significant positive correlation with canal water depth (τ = 0.26, p = 0.03, Table S2). This result contradicts previous findings in drainage ditches in temperate peatlands where CH₄ emissions had a weak negative correlation with depth⁵⁸, but these differing results may be explained by how well canal waters are mixed and aerated. For example, while we observed CH₄ oxidation in canals where dissolved oxygen is low (<2.5 mg L⁻¹) at the surface, dissolved oxygen may become depleted at depth^29,45 to below the concentration needed for aerobic methanotrophs with high oxygen affinity. As such, CH₄ oxidation may be limited to the surface waters of deeper canals, while in shallower canals oxidation may occur throughout the water column. Our study also only explicitly considered diffusive emissions. Measurements of CH₄ ebullition from canals could further clarify the role of water depth in shaping net canal CH₄ emissions, as ebullitive emissions vary with water depth⁵⁹. Altogether, our results suggest that shallower, vegetated canals may attenuate a higher percentage of CH₄ emissions through CH₄ oxidation.

Land use and seasonal precipitation cycles can both influence canal water depth and therefore dissolved oxygen. While we did not observe a significant impact of peatland land use on CH₄ oxidation nor other parameters including dissolved oxygen (Table S4), peatland water table, which directly influences canal water levels, has been shown to vary significantly between land use types⁶⁰. Canal water depth also varies 2- to 5-fold throughout the year in response to precipitation (Fig. S7), and reduced precipitation and flow during drier months may facilitate oxygen depletion by limiting turbulent mixing and re-aeration of canal waters^27,61. Accordingly, past studies have reported higher canal CH₄ emissions during dry periods^27,28. While our study was not conducted during pronounced wet or dry periods, the dissolved CH₄ and oxygen concentrations measured in our study fall within the range observed across Southeast Asia under varying land uses and seasons^{22,28,45,62,63} (Table S5). As such, we anticipate that water column CH₄ oxidation is prevalent across canals draining degraded peatlands in Southeast Asia.

Influence of oxidation on CH₄ emissions and their ¹³C in drained tropical peatlands

Our observations of canal CH₄ emissions estimated from dissolved CH₄ concentration (72.2 ± 151.2 mg CH₄ m^-2 d^-1) and collected using floating chambers (94.9 ± 142.3 mg CH₄ m^-2 d^-1) are within range of past observations from Indonesia^27,28,64 and Malaysia²⁶ where mean emissions range from 2.8 to 1073 mg CH₄ m^-2 d^-1 (Table S6). The IPCC CH₄ Emissions Factor for canals in tropical peatlands of 618.9 mg CH₄ m^-2 d^-1 (2259 kg CH₄ ha^-1 y^-1) was based on the only reported data²⁷ at the time of the 2013 Wetlands Supplement⁶⁵ This emission factor now represents the high end of field estimates to date among a still small number of existing studies and should be reconsidered to more accurately inventory the anthropogenic (e.g., from land use change) component⁶⁶ of CH₄ emissions from degraded tropical peatlands.

Despite high oxidation efficiencies, drainage canals can still emit large amounts of CH₄. For example, in canals where ~50% of the CH₄ transported from peatlands is oxidized we observe emissions >200 mg CH₄ m^-2 d^-1 (Figs. 3,  S5). The canals in this study were primarily situated in smallholder agricultural systems (Supplementary Data 1), and the mean estimated diffusive CH₄ emissions from canals presented here are 30x larger on a per area basis than mean peat soil CH₄ emissions from smallholder agriculture fields in West Kalimantan⁶⁷. Thus, while CH₄ oxidation plays a critical role in attenuating canal CH₄ emissions, canals can still contribute significantly to landscape-level CH₄ emissions from drained peatlands in Southeast Asia.

Beyond the rate of emissions, the δ¹³C signature of CH₄ emitted from tropical wetlands and freshwaters are critical for constraining their contribution to the global CH₄ budget, as δ¹³C-CH₄ values underpin source partitioning by atmospheric inversion models. Using a floating chamber to capture CH₄ emitted from a subset of the studied canals, we found that the mean δ¹³C-CH₄ was -64.7 ± 10.5‰ (Fig. 5A). Canal CH₄ emissions generally decreased as emitted δ¹³C-CH₄ increased (Fig. S8), therefore the flux-weighted mean δ¹³C-CH₄ was more negative at -69.0 ± 5.7‰ (Fig. 5A). Past observations of the δ¹³C signature of tropical wetland CH₄ emissions⁶⁸ indicate a range of -64‰ to -53‰. Our results suggest that the δ¹³C signature of CH₄ emissions from drainage canals, and potentially drained peatlands in Southeast Asia as whole due to the contribution of canals to landscape-scale CH₄ emissions, is more negative than prior measurements from tropical wetlands. As such, implementing a distinct δ¹³C-CH₄ source signature for Southeast Asian peatlands may improve top-down estimates of their CH₄ emissions.

Fig. 5. The isotopic composition of CH₄ emissions from tropical peatland drainage canals.

A Probability density estimates of the δ¹³C of CH₄ emitted from canal waters showing the unweighted (blue) and flux-weighted (purple) distributions of emitted δ¹³C-CH₄. The dashed black lines show the range of δ¹³C of tropical wetland CH₄ emissions reported in ref. ⁶⁸. B The percent of CH₄ oxidized in drainage canal waters (estimated from dissolved δ¹³C-CH₄ using α_ox = 1.022) versus the δ¹³C of CH₄ emitted from the corresponding canal. Each point represents a canal (n = 12) and error bars show the mean ± 1 standard deviation if replicates were collected at a canal. The shaded regions represent the 95% confidence interval associated with the linear relationship shown in panel B. Source data are provided as a Source Data file.

Furthermore, we find that the variation in δ¹³C of CH₄ emissions from the canal water surface (−86.9 to -44.3‰) was largely explained by the percent of CH₄ oxidized in canal waters (R² = 0.68, p < 0.05; Fig. 5B). Previous studies have identified oxidation, alongside variation in methanogenic pathways and wetland vegetation, as one potential explanation for latitudinal differences in the δ¹³C of wetland CH₄ emissions^68–70. Our results indicate that once CH₄ produced in peat soils is transported into canals, both the magnitude and the isotopic signature of canal CH₄ emissions are strongly influenced by CH₄ oxidation.

In summary, we demonstrate that CH₄ oxidation can substantially attenuate CH₄ emissions from canals draining peatlands in Southeast Asia. We estimate that CH₄ oxidation mitigates >50% of potential CH₄ emissions from canals across West Kalimantan, Indonesia. As landscape-scale measurements of CH₄ exchange in drained tropical peatlands indicate that canal networks contribute disproportionately to emissions from these ecosystems²⁰, our results suggest that CH₄ oxidation influences emissions not only from drainage canals but from degraded peatlands in Southeast Asia as a whole. Our results also have implications for peatland CH₄ emissions in response to land use change, including peatland restoration efforts. For example, we find that oxidation attenuates more CH₄ emissions from shallower canals that have higher dissolved oxygen concentrations. As such, efforts to rewet drained peatlands in Southeast Asia through canal blocking may impact CH₄ oxidation and therefore canal CH₄ emissions through changing canal water depth. Given the extensive networks of drainage canals in Southeast Asia and their substantial contribution to peatland CH₄ emissions, land use changes impacting CH₄ oxidation in canals will be reflected in the contribution of peatlands in Southeast Asia to the global CH₄ budget.

Methods

Field sampling

Drainage canals in lowland peatlands were sampled in Kubu Raya and Mempawah Districts, West Kalimantan, Indonesia. Canals were sampled in Kubu Raya in May 2023 and Mempawah in April 2024. This region has an equatorial rainfall pattern with no clear wet and dry season⁷¹. There is heavy rainfall year-round, but the driest months of the year usually occur in July or August. We sampled waters from canals of different sizes (5 to 90 cm water depths, 0.5 to 6 m canal widths), canals with (n = 15) and without aquatic vegetation (n = 19), and canals situated on peatlands under a variety of land uses. Smallholder mixed agriculture is the most represented land use in this study, but the sampled canals also include areas in smallholder plantations (pineapple and oil palm), industrial oil palm plantations, and open undeveloped land (i.e., deforested and/or burned areas), as well as 1 canal in a degraded forest, to capture the heterogeneity of drainage canals in the region. At each canal, we measured the canal dimensions as well as water temperature (°C), pH, dissolved oxygen (mg L^-1), conductivity (μS cm^-1), and redox potential (Eh, in mV) using a Hanna Instruments HI9829 multiparameter meter. A summary of the canals included in this study is available in Supplementary Data 1.

To measure the isotopic composition of source CH₄, we collected porewater profiles at 6 locations adjacent to a subset of the sampled canals. As shallow porewater is the primary source of discharge to drainage canals²², porewater was collected from 4-5 depths between 40 cm and 150 cm pending water table depth. Porewater was collected using a portable piezometer made of 3/8” stainless steel tubing housing 1/4” polyethylene tubing equipped with a coarse polypropylene screen to prevent collection of coarse debris (SedPoints, M.H.E. Products). Porewater samples were stored in 12 mL glass Exetainer^TM vials (Labco Ltd.) without headspace and acidified in the field to a pH of less than 2 using 1.5 M HCl.

Canal CH₄ concentration and δ¹³C

We collected surface water samples for analysis of dissolved CH₄ concentration and δ¹³C-CH₄ at all canals. Canal water samples were collected approximately 5 cm below the water surface and stored in 12 mL glass Exetainer^TM vials (Labco Ltd.) without headspace. Canal waters collected for assessment of in situ dissolved CH₄ concentration and δ¹³C-CH₄ were acidified in the field to a pH of less than 2 using 1.5 M HCl. Canal water samples collected in 2023 (including incubations described below) and all porewater samples were analyzed at the Stable Isotope Facility at UC Davis via a Delta V Plus IRMS following headspace equilibration. Samples collected in 2024 were analyzed at Stanford University via a Picarro G2210-i cavity ring down spectrometer following headspace equilibration. Reference standards with CH₄ mixing ratios and δ¹³C-CH₄ of 10 ppm/-45.5‰ and 30 ppm/-69.0‰ were run before and after sample analysis on the Picarro G2210-i to check for accuracy and instrument drift. Dissolved CH₄ concentrations were calculated considering the mixing ratio of CH₄ in the equilibrated headspace, using the ideal gas law, and in solution, following Henry’s Law, using the neonDissGas package⁷².

Incubations

We collected canal waters at a subset (n = 13) of the drainage canals for incubation experiments. Surface waters (~5 cm) were collected for all canals included in the incubation experiments, and at 5 of the canals we collected water from ~10 cm above the canal bottom using gas-tight tubing and a hand pump. Collecting these deeper canal waters for incubation experiments enabled us to account for any variability in isotopic fractionation of CH₄ oxidation with depth in the water column that could impact our estimates of oxidation efficiency. For canal waters collected for incubation experiments, we collected waters as described above but only field acidified samples for the initial incubation time point. Incubations occurred in the dark at room temperature (~25 °C) for 3 days. Duplicate samples for each canal (and depth, if applicable) were acidified every ~24 hours to pH <2 using 1.5 M HCl to stop CH₄ oxidation. All incubated waters were analyzed at the Stable Isotope Facility at UC Davis. Dissolved CH₄ concentration was calculated as described above.

Dissolved CH₄ concentration fell below the limit of quantification within 72 hours, as such incubation results only consider data from the first 2 days. One of the 5 deeper canal waters was omitted as CH₄ was not detectable after 24 hours. We calculated potential oxidation rates as the change in CH₄ concentration over the total incubation time. We also calculated the fractionation factor of CH₄ oxidation, or α_ox, from the CH₄ mixing ratios (in ppm) and δ¹³C-CH₄ of the incubated waters using a simplified Rayleigh model⁴⁰:

$$\ln \left(\frac{{\mathrm{CH}}_4}{{\mathrm{CH}}_{4,0}}\right)=\frac{{\mathrm{\alpha }}_{\mathrm{ox}}}{1-{\mathrm{\alpha }}_{\mathrm{ox}}}*\ln \left(\frac{1000+{\mathrm{\delta }}^{13}{\mathrm{C}-\mathrm{CH}}_4}{1000+{\mathrm{\delta }}^{13}{\mathrm{C}-\mathrm{CH}}_4,0}\right)$$ 1

Plotting Eq. (1) with ln(1000 + δ¹³C-CH₄) on the x-axis and ln(CH₄) on the y-axis produces a line with a slope of (α_ox/1-α_ox). As such, we calculated the slope as the difference in ln(CH₄) between the initial and final time points over the difference in ln(1000 + δ¹³C-CH₄) over the same time and then solved for α_ox.

Canal CH₄ emissions

We used a floating chamber to manually collect chamber headspace gasses at 12 of the sampled canals to assess CH₄ emissions and emitted δ¹³C-CH₄. A 20 cm diameter/2.1 L floating chamber was deployed on the canal water surfaces for 6 minutes in 2023 and 12 minutes in 2024. Chamber deployment time was increased in 2024 to ensure sufficient CH₄ accumulation for analysis via Picarro CRDS. The floating chamber was not held in place, but due to low canal water flow (stagnant to ~0.1 m s^-1) that chamber did not travel during flux measurement. Three 15 mL gas samples were collected from the chamber headspace over the deployment time via a sampling syringe and injected into a pre-evacuated 12 mL glass Exetainer^TM vial (Labco Ltd.). Floating chamber headspace gas samples from 2023 were analyzed at the Stable Isotope Facility at UC Davis and from 2024 at Stanford University, as described above. Methane emissions were calculated as the linear increase in chamber headspace CH₄ mixing ratio over the measurement period and converted from ppm CH₄ min^-1 to mg CH₄ m^-2 d^-1 using the Ideal Gas Law and the floating chamber dimensions. Fluxes were accepted if the linear increase in CH₄ over time met the standards of R² > 0.9 and p < 0.05. Emitted δ¹³C-CH₄ was determined via a Keeling plot approach, in which the δ¹³C of CH₄ emissions is the y-intercept of a linear regression of the inverse mixing ratio of CH₄ versus the δ¹³C-CH₄ of the corresponding sample^73,74.

We calculated gas transfer velocity (k, m d^-1) using data from the subset of canals where paired floating chamber CH₄ fluxes and canal water CH₄ concentrations were collected using Eq. (2):

$$\mathrm{Flux}=\mathrm{k}\left(\mathrm{C}{\mathrm{H}}_{4-\mathrm{canal}}-\mathrm{C}{\mathrm{H}}_{4-\mathrm{eq}}\right)$$ 2

Where CH_4-canal is the concentration of CH₄ in canal water, CH_4-eq is the CH₄ concentration at equilibrium the atmosphere (CH_4-eq), and flux is the rate of CH₄ emissions measured using the floating chamber. We used the median k value from the floating chamber deployments to estimate diffusive fluxes across all sampled (n = 34) canals. While applying a uniform value introduces uncertainty into the estimates of diffusive fluxes, conditions across the study region are characterized by high canal water temperature, low canal flow velocity (~0.1 m s^-1), and low windspeed. As such, factors that strongly influence CH₄ degassing (e.g., solubility and turbulence) should have minimal variation relative to the ~600-fold variation in canal water CH₄ concentration across study sites. Values were normalized to k₆₀₀ for literature comparison. See Supplementary Text 1 for further discussion of approaches to estimate k.

Estimating percent oxidation

We used a simple box model to estimate the percent of CH₄ transported from drained peatlands into canals that is oxidized and therefore not emitted to the atmosphere. The model calculated the percent oxidized based on the difference in δ¹³C between the source CH₄ (e.g., peat porewater) and CH₄ after oxidation (e.g., in the canal waters) as well as the isotopic fractionation of CH₄ oxidation (α_ox), which we determined via incubations as described above. Oxidation efficiency (f_ox) was calculated using a Rayleigh model for closed systems^42,75:

$$\ln \left(1-{\mathrm{f}}_{\mathrm{ox}}\right)=\left[\ln \left({\mathrm{\delta }}_{\mathrm{source}}+1000\right)-\ln \left({\mathrm{\delta }}_{\mathrm{canal}}+1000\right)\right]/\left[{\mathrm{\alpha }}_{\mathrm{ox}}-1\right]$$ 3

Where δ_source and δ_canal are the δ¹³C-CH₄ of peat porewater and drainage canal waters, respectively, and f_ox is the fraction of CH₄ oxidized. Values of f_ox were multiplied by 100 to convert to the percent of CH₄ oxidized. The closed system approach represents a lower bound on oxidation, as open system models often result in estimates of the percent oxidized >100%.

The results presented in the main analyses and figures are estimates of the percent oxidized based on mean observed values of α_ox (1.022 ± 0.009, from incubations) and δ_source (-85.0 ± 5.9‰, n = 27 measurements from 6 porewater profiles, Supplementary Data 2). To characterize the uncertainty of our estimates due to variability in α_ox and δ_source, we also report how our estimate varies when using ± 1 standard deviation of α_ox or δ_source in Eq. (3). Varying α_ox or δ_source by ± 1 standard deviation changes our estimate of the mean percent oxidized by ~10%.

To estimate the amount of CH₄ emissions attenuated by CH₄ oxidation, we back-calculated the concentration of CH₄ in canal waters based on the f_ox value for each canal:

$$\mathrm{Predicted}\mathrm{C}{\mathrm{H}}_4\mathrm{Concentration}=\frac{\mathrm{Observed}\mathrm{C}{\mathrm{H}}_4\mathrm{Concentration}}{1-{\mathrm{f}}_{\mathrm{ox}}}$$ 4

Using this predicted concentration, we calculated predicted diffusive CH₄ fluxes as described above. We then subtracted the diffusive CH₄ fluxes calculated from observed CH₄ concentrations from the predicted CH₄ fluxes based on the back-calculated concentrations to estimate the CH₄ emissions mitigated by CH₄ oxidation.

Statistical analysis

Statistical analysis and data visualization were performed in R v4.0.3. Data preparation was conducted using the dplyr package⁷⁶. Data and analyses were visualized using the ggplot2⁷⁷ and patchwork⁷⁸ packages. Statistical analyses were performed using the R Core Team stats package. Dissolved CH₄ concentration and estimated diffusive emissions were log₁₀-transformed prior to all statistical analysis to improve normality. If there were replicate measurements taken in a canal, the mean value of replicates was used in statistical analysis and data visualization. Summary statistics were calculated using all observations, including spatial replicates, to report the full range of observations. The level of significance for all analyses was 0.05.

We tested relationships between canal water CH₄ concentration and δ¹³C-CH₄ and between estimated diffusive CH₄ fluxes and the percent of CH₄ oxidized using least squares regression. We used Kendall’s rank correlation to assess the strength and direction of monotonic relationships between dissolved oxygen or canal depth and canal water dissolved CH₄ concentration, δ¹³C-CH₄, and percent CH₄ oxidized. We used a non-parametric correlation for these analyses as relationships between CH₄ oxidation and dissolved oxygen are often non-linear due to substrate saturation and potential inhibitory effects of oxygen above the optimal levels for CH₄ oxidation in freshwater environments⁷⁹. We used one-way ANOVA to assess the impact of vegetation and land use on canal properties and CH₄ cycling.

Author contributions

C.R.P., J.C.B., and A.M.H. designed the study. C.R.P., J.C.B., J.S., D.S.P.A.B., E.D., and Y.A. performed the research. C.R.P. analyzed the data and wrote the manuscript. N.N. and G.Z.A. provided field site and laboratory access. J.C.B., J.S., D.S.P.A.B., E.D., Y.A., A.A., A.G., N.N., G.Z.A., and A.M.H. reviewed the manuscript and contributed to manuscript revisions.

Peer review

Peer review information

Nature Communications thanks Michael Peacock, Paula Reis and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are presented in the manuscript and/or the Supplementary Information. The data used in this study are available at the Zenodo repository under ‘Tropical Peatland Drainage Canal Methane Concentrations, Fluxes, and Isotopic Composition’ (10.5281/zenodo.11155160). Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-54063-x.