Significant sedge-mediated methane emissions from degraded tropical peatlands

The fire-degraded tropical peatland study area lies 20 km south of the coast in Belait District, Brunei Darussalam (4°28ʹ40'' N, 114°18ʹ19'' E; figure 1). The region experiences an equatorial-humid climate with two seasons, separated by two transitional periods (two maxima and two minima) [27]. The first maximum and second minor maximum lasts from late October to January and from May to June, respectively, corresponding to the wet period (ibid). The lowest minimum and minor minimum occur from February to March and June to August, respectively, corresponding to the dry period (ibid). Average rainfall during the wet and dry period (for 1966—2006) is 1939.4 ± 51.6 mm and 924.8 ± 46.3 mm, respectively (ibid). The average monthly temperature is 27.2 °C ± 0.4 °C, which varies annually between 23.4 °C and 30.8 °C [27, 28].

Zoom In Zoom Out Reset image size

We identified our study area from LANDSAT imagery using two criteria. Firstly, the area had burnt multiple times (n = 7; 1998, 2005, 2009–2010, 2014–2016) over the last few decades and had never recovered back to secondary PSF state [24], and therefore the study area was covered with pioneering and flood-tolerant plant species such as ferns and sedges. Secondly, due to the disruption of original hydrology by the construction of an access road and drainage canal, the area closer to the drainage canal remained relatively water-saturated all-year round and hence covered with sedges.

Our fire-degraded tropical peatland study area stretches 2 km laterally along the road with drainage canal. To cover this spatial variability, we established three sites (B1—3) at an interval of 600—800 m along the road, adjacent to the drainage canal. Each site had two plots of 5 m × 5 m, i.e. 2 m and 10 m away from the drainage canal (figure 1). The choice of these specific plots was based on the presence of sedges all year-round, hence allowing continuous CH₄ measurements during the dry and wet period.

The study area also stretches 300—500 m east-west, towards the forest-edge. Hence, to capture the spatial variability away from the drainage canal, we selected a longer transect (T4; described in Lupascu et al 2020), where plots (of 5 m × 5 m) were at an interval of 100 m from the drainage canal towards the forest (B2—T4; figure 1).

In each plot, we recorded environmental variables, pore-water physicochemical properties, CH₄ emissions, and peat characteristics (supplementary information (available online at https://stacks.iop.org/ERL/16/014002/mmedia)). These obs-ervations were recorded over four fieldwork campaigns, i.e. in August 2018, October 2018, January 2019, and July 2019.

2.2.1. Environmental variables

2.2.2. Physicochemical analysis of pore-water

The vegetation at our site was mainly composed of vascular plants. A 1 m × 1 m quadrat [n = 47 in total including both dry (n = 22) and wet (n = 25) period] was used to estimate vegetation per cent cover (number of small 10 cm × 10 cm grids occupied by plant species within 1 m × 1 m quadrat), height, and the total number of plant species in the quadrat [29]. The assessment was conducted on five transects (T1—5; figure 1), where measurements were recorded at 2 m, 5 m, 10 m, 50 m, 100 m, 200 m, and 300 m from the drainage canal. We used this non-uniform sampling design since we observed a noticeable shift in plant communities (more sedges and fewer ferns) within this short interval compared to plots towards forests (50 m, 100 m, 200 m, and 300 m; figure 1(c)) [30].

To capture the seasonal variation, vegetation composition assessments were conducted during the wet (in January 2019) and dry (in July 2019) period. However, due to a fire event in February 2019, which burnt two transects (T3—4), nearly four months before the July 2019 assessment, we analyzed these data separately from the 'previously' burnt transects (T1, T2, T5—burnt in March 2016).

We installed 3—4 permanent PVC collars (of 23 cm diameter) per plot for CH₄ flux measurements on the peat surface (without vegetation). These collars were installed in June 2018, two months before the first measurement (August 2018), and inserted 15—20 cm deep to exclude any root-regrowth. For the total CH₄-flux measurement, i.e. with vegetation (sedges + peat), 9—12 PVC collars (of 10 cm diameter) per plot were inserted few centimeters deep in peat to avoid root damage but to ensure proper sealing for flux measurement.

CH₄ flux was measured using a portable Greenhouse Gas Analyzer (GasScouter, Picarro Inc. Santa Clara, CA, USA) connected to a transparent chamber. The chamber was made of a clear acrylic pipe of 11 cm internal diameter and variable height up to 200 cm resulting in a maximum total volume of 19.1 l. Gas collected inside the chamber was mixed using a battery-operated fan, before passing through 1.5 m tubing (3 mm internal diameter) to the gas analyzer and circulated back to the chamber (figure S2). To estimate the sedge-mediated CH₄ emissions, measurements without vegetation (peat-only) was recorded adjacent to measurements with vegetation (i.e. sedge + peat) and subtracted from it to estimate the net CH₄ contribution from sedges (n = 16—63 per field campaign). All measurements were recorded between 09:00 and 17:00 h. The chamber was left closed for up to 5 min, and CH₄ flux (F) was calculated as per the equation below:

$$\begin{equation*}F = \frac{{s\, \times \,V\, \times \,t}}{{R\, \times \,{T_{{\text{air}}}}\, \times \,A}}\end{equation*}$$

where F is in mgCH₄ m⁻² h⁻¹; s is the linear regression slope of CH₄ concentration change over time in ppm per second (accepted with r² > 0.85 and p < 0.05) [31, 32]; V is the volume of the chamber, tubing, and the internal cavity of the gas analyzer in liters; t is the conversion factor from seconds to hours, µg to mg, and 16 µg µmol⁻¹ of CH₄; R is the universal gas constant (8.205 × 10⁻⁵ l atm mol⁻¹K⁻¹); T_air is the air temperature inside the chamber (K); A is the surface area of the chamber (m²).

With each flux measurement, soil temperature (T) and soil water content (SWC) were measured beside the collar using a digital thermometer (15–077, ThermoFisher Scientific, USA) and soil moisture meter (Dynamax TH300, USA) at a depth of 5 cm.

All the statistical tests were performed using Origin-Pro (2019), and significance was accepted at p < 0.05. The data was checked for normality using the Shapiro–Wilk test. Accordingly, non-parametric tests, namely Kruskal–Wallis ANOVA and Mann–Whitney U tests, were used for data comparison. Spearman correlation coefficients were calculated to evaluate the relationship between CH₄ emission and number and height of sedges, environmental variables, and pore-water properties. All the values are reported as mean ± s.e. unless stated separately as mean ± s.d.

Environmental variables were recorded from August 2018 to July 2019 with a total of 337 observation values (n = 173 for the dry period and n = 164 for the wet period). Air temperature over the sampling period showed an overall annual mean of 26.9 °C ± 1.6 °C with an average monthly value of 27.1 °C ± 1.5 °C in the dry period and 26.7 °C ± 1.7 °C in the wet period (figure 2(a)). Similarly, peat soil temperature showed a mean of 27.4 °C ± 1.0 °C for the study period, but with no significant difference between dry (27.4 °C ± 0.9 °C) and wet (27.3 °C ± 1.1 °C) period (figure 2(b)).

Zoom In Zoom Out Reset image size

Total annual precipitation over the sampling period was 1447.8 ± 12.5 mm, with 505.9 ± 10.2 mm and 941.9 ± 14.3 mm during the dry and wet period, respectively. Precipitation increased from August 2018 (75.1 mm) until January 2019 (190 mm); however, it remained infrequent from February 2019 to July 2019 (figure 2(c)).

The WT was affected by precipitation patterns, with higher values during the wet period (−7.1 ± 10.2 cm), compared to the dry period (−26.0 ± 10.0 cm; figure 2(c)).

Soil water content varied from 78.8 ± 13.4% to 91.6 ± 14.4% during the dry and wet period, respectively.

We observed seasonal differences in pore-water properties where mean values for pH, Temperature, EC, TDS, Salinity, and DOC, except DO, were overall higher during the dry period (August 2018 and July 2019) compared to the wet period (October 2018 and January 2019) (table 1). However, only pH and Salinity differed significantly (p < 0.05), with a decrease in pH from 4.1 ± 0.2 to 3.1 ± 0.3, and in Salinity from 0.032 ± 0.003‰ to 0.024 ± 0.001‰ during the dry and wet period, respectively. Comparing the data from before the fire event in February 2019 with our July 2019 sampling, we observed an increase for all parameters except for DO.

Pore-water ions concentration remained stable for the first three sampling campaigns between August 2018 and January 2019 (table 2) with Cl⁻ ranging from 3.53 to 4.31, SO₄ ²⁻ from 0.21 to 0.34, PO₄ ³⁻ from 0.26 to 0.39, Na⁺ from 2.30 to 3.71, K⁺ from 0.87 to 0.96 mg l⁻¹. Ion concentrations were significantly (p < 0.05) higher in our July 2019 sampling compared to previous months. Specifically, when compared to Jan 2019 values, the concentrations of Cl⁻ and SO₄ ²⁻ were five times, PO₄ ³⁻ concentration was eight times, and Na⁺ and K⁺ concentrations were three and seven times higher, respectively (table 2).

During our wet (January 2019) and dry (July 2019) period assessment, we recorded only one species of sedge (Scleria sumatrensis) and three species of ferns (Blechnum indicum, Nephrolepis hirsutula, Nephrolepis biserrata).

We observed that the vegetation cover change was spatially significant (p < 0.001) compared to seasonal variability, where the mean sedge cover increased from 5% to 9% in plots far from the drainage canal to 23% to 31% in plots near the drainage canal (figures 3(a) and (b)).

Zoom In Zoom Out Reset image size

We also analyzed the vegetation data separately for previously (in March 2016) and recently (in February 2019) burnt transects to observe the effects of recent fire episodes on the number of sedges. However, similar to the above-mentioned, we found that the change in the number of sedges was spatially significant (p < 0.001) compared to seasonal variability or fire episode. Therefore, using the average of both the dry and wet period, we found that the number of sedges along the transect declined from 28 sedges per m² in plots near canal ( plots at 2 m, 5 m, and 10 m) to 14 sedges per m² in plots towards forest (plots at 50 m, 100 m, 200 m, and 300 m) on previously burnt transect (figure 4(a)). Similarly, it declined from 43 sedges per m² in plots near canal to 5 sedges per m² in plots towards forest on recently burnt transect (figures 4(b)).

Zoom In Zoom Out Reset image size

The average height of sedges decreased from 1.6 ± 0.3 (n = 125) m to 1.3 ± 0.2 (n = 110) during the wet and dry periods, respectively.

We found that the sedges in degraded tropical peatlands are a significant source of CH₄, transporting approximately >70% of the total CH₄ emissions directly into the atmosphere (tables S1 and S2). This sedge-mediated transport varied seasonally as well as spatially during the sampling period.

Over the sampling period, sedge-mediated CH₄ emissions contributed to an annual mean of 3.77 ± 0.47 mgCH₄ m⁻² h⁻¹, compared with a total flux of 4.61 ± 0.44 mgCH₄ m⁻² h⁻¹ (n = 180) (Inset figure 5). CH₄ emissions significantly (n = 180; p < 0.001; Kruskal–Wallis ANOVA, H = 67.84, DF = 3) varied over the sampling period following rainfall patterns, and hence WT levels. Lower values were recorded during the dry period with 1.60 ± 0.47 and 0.67 ± 0.08 mgCH₄ m⁻²h⁻¹ for August 2018 and July 2019, respectively. The peak in CH₄ emission was observed in the wet period with 3.52 ± 0.83 and 5.96 ± 0.96 mgCH₄m⁻²h⁻¹ in October 2018 and January 2019, respectively (table S1). These net emissions were approximately five times higher compared to the dry period.

Zoom In Zoom Out Reset image size

We also observed spatial variation in sedge-mediated CH₄ emission where plots near the drainage canal (at 2 m, 5 m, and 10 m) emitted higher CH₄ compared to plots farther away from drainage canal, i.e. towards the forest edge (plots at 100 m, 200 m, and 300 m) (figure 6). This spatial variation was significant (n = 180; p < 0.001; Kruskal–Wallis ANOVA, H = 42.48, DF = 5) and was mainly attributed to the higher number of sedges and higher WT in plots closer to the drainage canal. Sedge-mediated CH₄ emissions from plots near the drainage canal (2 m, 5 m, 10 m) emitted a mean of 3.48—5.41 mgCH₄m⁻²h⁻¹, whereas plot farther away (at 100 m, 200 m, 300 m) emitted a mean of 0.63—1.75 mgCH₄m⁻²h⁻¹ (table S2).

Zoom In Zoom Out Reset image size

Overall, the CH₄ fluxes measured during both dry and wet period showed a significant positive correlation with the number (p < 0.001) and height of sedges (p < 0.05), soil water content (p < 0.001) and WT level (p < 0.001), and a significant negative correlation with distance from drainage canal (p < 0.001) (table S3). Among these, the WT level showed the strongest relationship with CH₄ (r² = 0.78; p < 0.001). In addition, pore-water parameters such as pH (p < 0.001) and Salinity (p < 0.05) were significantly correlated with CH₄ emissions (table S4).

Our analysis show for the first time that sedge-mediated CH₄ transport to the atmosphere from degraded tropical peatland contributes approximately 70—80% of the total CH₄ emission at an ecosystem level. This result is similar to plant-mediated CH₄ emissions in northern peatlands, which transports ∼28% to >90% of the total CH₄ emission to the atmosphere at an ecosystem level [18, 20, 33]. In the Tropics, few studies have investigated the tree-mediated CH₄ transport in the intact tropical peatlands of Panama [34, 35], Brunei [36], and Amazon flood plains [37], where such plant-mediated transport contribute in the range of ∼30—92% of the total CH₄ emission to the atmosphere at an ecosystem level.

Considering that degraded tropical peatlands cover approximately one-tenth of the total peatland area in SE Asia (as in 2015 [4]), it is evident that the contribution of sedge-mediated CH₄ emission is an important factor in refining the regional CH₄ budget. Although the percentage of sedge-mediated CH₄ emission remained consistent during the dry (70%) and wet (80%) period, the CH₄ emission values were significantly lower during the dry period (0.78 ± 0.14 mgCH₄m⁻²h⁻¹) when compared to the wet period (4.86 ± 0.66 mgCH₄m⁻²h⁻¹), which is due to significant positive correlation with the WT level (figure 5). In fact, studies have reported that a WT lowering of 20 cm below surface disrupts the optimum redox conditions for methanogenesis as well as exposes the anoxic peat to aerobic decomposition [38–40]. It is worth noting that the CH₄ emitted via sedges may primarily be originated from recently fixed carbon (litter, root exudates) and to lesser extent from bulk peat as confirmed from recent studies [24, 41]. Although this recently fixed C is considered as a short-term cycling C [42], nevertheless the significant CH₄ emission from sedge-mediated transport additionally contribute to the existing sources of CH₄ from degraded tropical peatlands, which are expected to increase in future.

We also observed a significant spatial variation in sedge-mediated CH₄ emission where values were four times higher in the area near the drainage canal (4.33 ± 0.56 mgCH₄m⁻² h⁻¹) compared to the area far from the drainage canal, i.e. towards the forest edge (1.12 ± 0.24 mgCH₄ m⁻² h⁻¹). Although burning did affect the number of sedges along the transect, the differences were not statistically significant. Moreover, this may have been compensated by fast regrowth of sedges in the presence of increased nutrients released after the fire (table 2) as well as constant high soil water content (72—92%) in these plots [43, 44]. Nevertheless, with future frequent flooding and changes in vegetation composition of such fire-degraded tropical peatland areas, these emission values are likely to shift towards the higher end.

Our study showed that sedge-mediated CH₄ emission correlated with the WT, temperature, and pore-water properties (tables S3 and S4). WT is the dominant factor in controlling CH₄ emission from tropical peatland [24, 39, 45]. At our site, CH₄ emission during the dry period ranged from 0.67 ± 0.08 to 1.60 ± 0.47 mgCH₄m⁻²h⁻¹ when WT was low (−28.1 cm in July 2019 and −16.4 cm in August) and from 3.52 ± 0.83 to 5.96 ± 0.96 mgCH₄m⁻²h⁻¹ at higher WT during the wet period (−10.7 cm in October 2018 and −4.2 cm in January 2019). We also observed a similar trend along the transect where emission values ranged from 4.33 ± 0.56 to 1.12 ± 0.24 mgCH₄m⁻²h⁻¹ near the drainage canal to the area towards the forest, respectively. These increases in CH₄ emissions were mainly attributed to WT, which increased with the seasonal increase in precipitation from August 2018 to January 2019 (figure 2(c)) as well as along the transect. CH₄ emission at our study is on the higher side compared to other literature reported from degraded tropical peatlands [24, 38, 46, 47], likely because of the difference in WT depths (table S5) as well as previous studies did not include plant-mediated CH₄ transport.

Temperature is known to enhance CH₄ production and emission, where the optimum temperature range is between 25 °C and 40 °C [48, 49]. At our site, CH₄ emissions showed a weak positive correlation with peat soil temperature (r² = 0.08; p < 0.1) and a contrasting negative correlation with pore-water temperature (r² = −0.41; p < 0.1) (tables S3 and S4). These trends corresponds to the observations from other studies that reported WT as a primary factor affecting CH₄ emission in the tropics instead of temperature change [24, 39, 47, 50, 51].

Pore-water properties were similar across measurements recorded before the February 2019 fire event, whereas, in July 2019, we observed noticeably higher pH, EC, TDS, Salinity, DOC, and pore-water ions concentration (tables 1 and 2). This increase could partly be due to the release of anions and cations from soil organic matter combustion from the fire event [52] as well as due to atmospheric deposition of nutrients from biomass burning in tropical peatland areas [53]. In addtion, high Ca²⁺, Mg²⁺ and K⁺ content of ash can increase pH by replacing H⁺ and Al³⁺ ions adsorbed on the negative charge of the soil colloids [52, 54]. This is consistent with our pore-water nutrient analysis, which showed higher K⁺ (p < 0.05), Ca²⁺, and Mg²⁺ concentrations in our July 2019 measurements (table 2). We also observed higher PO₄ ^3- (p < 0.05) and SO₄ ²⁻ concentrations, which are known to oxidize CH₄ and inhibit methanogenesis in soils [55, 56] (table 2), hence explaining the lower CH₄ emissions along with the lower WT levels (figures 2–5).

Methanogenesis occurs at an optimum pH level, i.e. near neutral [48, 57]. At our site, the average pH was acidic (3.6 ± 0.2), as reported for other tropical peatland studies [24, 58, 59], and was negatively correlated (r²= −0.67; p < 0.001, table S4) with CH₄ emissions. However, methanogenesis can occur at low pH with the presence of acid-tolerant methanogens and pH-neutral microsites within the complex peat matrix [60, 61]. Studies from northern regions, however, have reported an inconsistent relationship between pH and CH₄ emissions and further suggested pH as a secondary determinant in CH₄ emission [62]. In addition to pH, we also observed that Salinity had a significant negative correlation (r²= −0.51; p < 0.05, table S4) with CH₄ emissions, which is consistent with the studies reported from wetlands in the saline environment [63, 64].

Our analysis shows that sedge-dominated degraded tropical peatland areas may contribute to 18.0—329.7 kgCH₄ ha⁻¹ yr⁻¹ (118.0 ± 71.4) (figure 7). This number is substantially higher in comparison to values reported from other tropical peatland land use types [65, 66], as well as compared to recent estimates from the intact PSF in Sarawak [67] and Kalimantan [68] (table S5). It is approximately three times the values reported for intact PSF [66, 67], whereas seventeen times higher when compared to emission values reported for similar land-use types, i.e. cropland and shrubland category [66] (table S5). Moreover, such high variation across different land-use presented here could primarily be attributed to the WT level, which is a well known factor in regulating CH₄ emission in tropical peatlands compared to temperature change [24, 39, 47, 50, 51]. In addition to WT, changes in peat soil characteristics due to fire may play a significant role in affecting CH₄ emissions. Specifically, the higher bulk density at the peat surface (supplementary information), the lower DO, EC, PO₄ ³⁻ and SO₄ ²⁻ concentrations together with higher pH (tables 1 and 2) may lead to optimum conditions for methanogenesis to occur [24]. Furthermore, differences in vegetation cover (sedges and ferns at our sites) may also drive CH₄ emissions, as changes in vegetation type can affect substrate availability for methanogenesis, as reported by recent studies from tropical peatlands in Panama [69, 70].

Zoom In Zoom Out Reset image size

We know that the latest (2015) estimate of degraded open tropical peatlands in SE Asia is approximately 1.42 Mha [4]. However, we do not have a clear sedge cover assessment within this area. At our site, the sedge percentage cover varies from 23—31% near the drainage canal to 5—9% away from the drainage canal, i.e. towards the forest. Due to the logistical difficulties in undertaking this kind of research, only a handful of vegetation assessment studies have been pursued in degraded tropical peatlands in SE Asia [71–73]. Our vegetation composition results are similar to other studies [72, 73], where sedge cover ranges from 15% to >60% during different stages of regeneration and restoration of degraded tropical peatland areas. More importantly, sedges are known to out-compete native seedlings during natural regeneration, which may result in these fire-degraded tropical peatland areas becoming sedge-dominated [11, 74]. As our findings confirm that sedge-mediated transport is a significant source of CH₄, it becomes even more important to account for sedge cover change when planning for effective restoration and regeneration strategies of degraded tropical peatlands within a C accounting context.

Recent studies have established that the fire in tropical peatlands originates commonly in deforested areas, where presence of shrubs such as sedges and ferns makes it more prone to recurrent fires [5, 14, 76–78]. Moreover, as prolonged dry-conditions associated with El-Niño are expected to increase in the future [79], these fires can become even more frequent. One of the consequences of repeated fires and drainage on peatland is land subsidence, resulting from combustion and oxidation losses of peat volume, which leads to an increased risk of flooding [15, 75, 80], and eventually impaired vegetation regeneration [17, 72, 81]. Moreover, these already subsided fire-degraded tropical peatland areas may experience severe flooding where the total flooded area can expand up to five times during the wet period as shown in a 2006 study in Jambi Province, Indonesia, hence expanding the sedge dominated landscapes drastically [17]. Our data also indicates that sedges are more abundant during the wet period and closer to water inundated area near the drainage canal.

Further, the chance of an increase in floods can be higher in these areas, which is compounded by a general increase in mean and extreme precipitation projected in SE Asia by 2100 [82]. Within this context, there is a potential for a vegetation shift from fern-dominated to more flood-tolerant sedge-dominated landscapes, where fire-degraded tropical peatland areas may play an even more significant role within the CH₄ budget at the regional level in the near future.

To summarize, we found that the WT level, pore-water properties, and sedge cover significantly correlated with CH₄ emission. As our results are based on observations from one study site with measurements recorded over four field campaigns in one year period, we acknowledge the uncertainties associated with our analysis and its replicability in similar tropical peatland ecosystems. Therefore, we suggest a spatially distributed study across multiple fire-degraded tropical peatland sites over a longer temporal scale to investigate further these relationships as it can be affected by the factors such as seasonality, fire events, and distance from drainage canal.