Impact of water table level on annual carbon and greenhouse gas balances of a restored peat extraction area

Peatland restoration may provide a potential afteruse option to mitigate the negative climate impact of abandoned peat extraction areas; currently, however, knowledge about restoration effects on the annual balances of carbon (C) and greenhouse gas (GHG) exchanges is still limited. The aim of this study was to investigate the impact of contrasting mean water table levels (WTLs) on the annual C and GHG balances of restoration treatments with high (ResH) and low (ResL) WTL relative to an unrestored bare peat (BP) site. Measurements of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) fluxes were conducted over a full year using the closed chamber method and complemented by measurements of abiotic controls and vegetation cover. Three years following restoration, the difference in the mean WTL resulted in higher bryophyte and lower vascular plant cover in ResH relative to ResL. Consequently, greater gross primary production and autotrophic respiration associated with greater vascular plant cover were observed in ResL compared to ResH. However, the means of the measured net ecosystem CO2 exchanges (NEE) were not significantly different between ResH and ResL. Similarly, no significant differences were observed in the respective means of CH4 and N2O exchanges. In comparison to the two restored sites, greater net CO2, similar CH4 and greater N2O emissions occurred in BP. On the annual scale, ResH, ResL and BP were C sources of 111, 103 and 268 g C m yr and had positive GHG balances of 4.1, 3.8 and 10.2 t CO2 eq ha −1 yr, respectively. Thus, the different WTLs had a limited impact on the C and GHG balances in the two restored treatments 3 years following restoration. However, the C and GHG balances in ResH and ResL were considerably lower than in BP due to the large reduction in CO2 emissions. This study therefore suggests that restoration may serve as an effective method to mitigate the negative climate impacts of abandoned peat extraction areas.

presently store about a third of the global soil C pool (Gorham, 1991;Turunen et al., 2002). They 17 also provide a small but persistent long-term C sink (between 20 and 30 g C m -2 yr -1 ) (Gorham, 18 1991; Vitt et al., 2000;Roulet et al., 2007;Nilsson et al., 2008). Carbon accumulation in peatland 19 ecosystems occurs mainly due to the slow decomposition rate under the anoxic conditions caused 20 by high water table levels (Clymo, 1983). Within the past century, a large fraction of peatlands 21 has been exploited for energy production and horticultural use. Since commercial peat extraction 22 requires initial vegetation removal and drainage, harvested peatlands are turned into C sources by 23 eliminating the carbon dioxide (CO 2 ) uptake during plant photosynthesis and increasing CO 2 24 emission due to enhanced aerobic decomposition of organic matter. Thus, following the cessation 25 of peat extraction activities, after-use alternatives that mitigate the negative climate impacts of 26 these degraded and abandoned areas are required. 27 Among different after-use alternatives, re-establishment of peatland vegetation, which is essential 28 for returning the extracted peatlands back into functional peat-accumulating ecosystems, has been 29 the ecosystem CO 2 and CH 4 exchanges following peatland restoration. 23 Considering the strong effects of the WTL on plant succession and ecosystem C exchanges, 24 differences in the depth of the re-established WTL baseline (i.e. the mean WTL) due to the 25 varying effectiveness of initial restoration activities (e.g. ditch blocking, surface peat stripping) 26 may have implications for the trajectories of vegetation development and recovery of the C sink 27 function following restoration. To date, only few studies (e.g. Tuittila et al., 1999Tuittila et al., , 2004 have 28 investigated the impact of contrasting WTLs on the subsequent ecosystem C balance within the 29 same restoration site. Understanding the sensitivity of the C balance to differences in the re-30 established WTL baseline is, however, imperative when evaluating the potential of restoration for 1 mitigating the negative climate impacts of drained peatlands. Moreover, estimates of the C sink-2 source strength of restored and unrestored peatlands have been limited to the growing season 3 period in most previous studies (Tuittila et al., 1999(Tuittila et al., , 2000a(Tuittila et al., , 2004Waddington et al., 2010;4 Samaritani et al., 2011;Strack et al., 2014). In contrast, data on annual budgets, which are 5 required to evaluate the full climate benefits of peatland restoration relative to the abandoned peat 6 extraction area, are currently scarce and to our knowledge only reported in a few studies (e.g. Yli-7 Petäys et al., 2007;Strack and Zuback, 2013). 8 Furthermore, the full ecosystem greenhouse gas balance (GHG) also includes emissions of 9 nitrous oxide (N 2 O), a greenhouse gas with an almost 300 times stronger warming effect relative contrasting WTLs on the annual C and GHG balances of a restored peatland and ii) to assess the 25 potential of peatland restoration for mitigating the C and GHG emissions from abandoned peat 26 extraction areas. Our hypotheses were that i) the C and GHG balances are improved in Res-H 27 relative to Res-L since the increased net CO 2 uptake, as a result of reduced peat mineralization 28 and greater water availability enhancing gross primary production, outweighs the increase in CH 4 29 emissions under high WTL conditions and ii) the C and GHG balances of the two restoration 30 datalogger (Campbell Scientific Inc., Logan, UT, USA). In addition, continuous 30-min records 23 of the WTL relative to the soil surface were obtained with submerged HOBO Water Level 24 Loggers (Onset Computer Corporation, Bourne, MA, USA) placed inside perforated 1.0 m long 25 PVC pipes (Ø 5 cm; sealed in the lower end). 26 The on-site meteorological measurements were complemented by Estonian Weather Service data 27 to obtain complete time series of Ta, PPT and PAR over the entire year. Hourly means of Ta and 28 daily sums of PPT were obtained from the closest (~20 km away) Viljandi meteorological station. 29 Global radiation (hourly sums) data from the Tartu meteorological station (~40 km away) was 1 converted to PAR based on a linear correlation relationship to on-site PAR. 2 In addition, manual measurements of soil temperature (depths 10, 20, 30 and 40 cm) were 3 recorded by a handheld temperature logger (Comet Systems Ltd., Rožnov pod Radhoštěm, Czech 4 Republic) and volumetric soil water content (depth 0-5cm) using a handheld soil moisture sensor 5 (model GS3, Decagon Devices Inc., Pullman, WA, USA) during each sampling campaign. 6 Furthermore, groundwater temperature, pH, redox potential, dissolved oxygen content, electrical 7 conductivity as well as ammonium (NH 4 + ) and nitrate (NO 3 -) concentrations were measured in 8 observation wells (Ø 7.5 cm, 1.0 m long PVC pipes perforated and sealed in the lower end) 9 installed at each sampling location using YSI Professional Plus handheld instruments (YSI Inc., 10 Yellow Springs, OH, USA). In addition, soil samples (0-10 cm depth) in three replicates were 11 taken from each of the treatments and analyzed for pH as well as total C, total N, P, K, Ca and S 12 contents at the Tartu Laboratory of the Estonian Environmental Research Centre. Three 13 additional samples were taken from the same depth to determine bulk density in each treatment. 14 Mean values for these soil properties are summarized in Table 1. 15

Vegetation cover estimation 16
To assess the effect of vegetation development on greenhouse gas fluxes, vegetation cover (%) 17 and species composition were recorded inside each of the flux measurement collars (see section 18 2.4) in late spring. In each collar, the cover was estimated visually for each species and rounded 19 to the nearest 1%. Bryophyte, vascular plant and total vegetation cover were computed as the sum 20 of their respective individual species coverages. 21

Net ecosystem CO 2 exchange, ecosystem respiration, gross and net primary 22 production measurements 23
To evaluate the impact of WTL on the net ecosystem CO 2 exchange (NEE) in the restored Res-H 24 and Res-L treatments, flux measurements were conducted biweekly from May to December 2014 25 at three sampling locations within each replicate plot (i.e. 6 locations per treatment) using the 26 closed dynamic chamber method. At each sampling location, a collar (Ø 50 cm) with a water-27 filled ring for air-tight sealing was permanently installed to a soil depth of 10 cm. NEE 28 measurements were conducted in random plot order (to avoid diurnal effects) using a clear 1 Plexiglas chamber (95% transparency; h 50 cm, V 65 L) combined with a portable infra-red gas-2 analyzer (IRGA; EGM-4, PP Systems, Hitchin, UK). The chamber was equipped with a sensor to 3 measure photosynthetically active radiation and air temperature (TRP-2, PP Systems, Hitchin, 4 UK) inside the chamber. Ambient air temperature was also recorded with an additional 5 temperature sensor placed on the outside of the chamber. Cooling packs placed inside the 6 chamber were used to avoid a temperature increase inside the chamber during measurements. The 7 chamber was also equipped with a low-speed fan to ensure constant air circulation. After every 8 NEE measurement, ecosystem respiration (RE) was determined from a subsequent measurement 9 during which the transparent chamber was covered with an opaque and light reflective shroud. 10 CO 2 concentrations, PAR, temperature, pressure and relative humidity were recorded by the 11 IRGA system every 4.8 s over a 4-min or 3-min chamber deployment period for NEE and RE 12 measurements, respectively. Since the aim of this study was to assess the atmospheric impact of 13 restoration, all fluxes are expressed following the atmospheric sign convention in which positive 14 and negative fluxes represent emission to and uptake from the atmosphere, respectively. 15 Gross primary production (GPP) was derived from the difference between NEE and RE (i.e. GPP 16 = NEE -RE). In addition, an estimate of net primary production (NPP) was derived from the 17 difference between NEE and heterotrophic respiration (Rh; see section 2.5) (i.e. NPP = NEE -18 Rh). 19 RE estimates during the non-growing season months of March to April 2014 and January to 20 February 2015 were determined from closed static chamber measurements (described in section 21 2.6). Air samples collected during these measurements were analyzed for their CO 2 22 concentrations on a Shimadzu GC-2014 gas chromatograph with an electron capture detector 23 (ECD). These RE estimates also represented non-growing season NEE for all treatments. 24 In the BP treatment, RE was determined by measurements using a separate closed dynamic 25 chamber set-up as described below in section 2.5. Due to the absence of vegetation, GPP as well 26 as NPP were assumed to be zero and NEE subsequently equaled RE in the BP treatment.

Flux calculation 24
Fluxes of CO 2 , CH 4 and N 2 O were calculated from the linear change in gas concentration in the 25 chamber headspace over time, adjusted by the ground area enclosed by the collar, volume of 26 chamber headspace, air density and molar mass of gas at measured chamber air temperature. The 27 linear slope in case of the dynamic chamber measurements was calculated for a window of 25 28 measurement points (i.e. 2 min) moving stepwise (with one-point increments) over the entire 1 measurement period after discarding the first two measurement points (i.e. applying a 9.6 sec 2 'dead band'). The slope of the window with the best coefficient of determination (R 2 ) was 3 selected as the final slope for each measurement. In the static chamber method, the linear slope 4 was calculated over the four available concentration values. 5 All dynamic chamber CO 2 fluxes with a R 2 ≥ 0.90 (p < 0.001) were accepted as good fluxes. 6 However, since small fluxes generally result in a lower R 2 (which is especially critical for NEE 7 measurements), dynamic chamber fluxes with an absolute slope within ±0.15 ppm s -1 were 8 always accepted. The slope threshold was determined based on a regression relationship between 9 the slope and respective R 2 values. For static chamber measurements, the R 2 threshold for 10 accepting CO 2 , CH 4 and N 2 O fluxes was 0.90 (p < 0.05), 0.80 (p < 0.1) and 0.80 (p < 0.1), 11 respectively, except, if the maximum difference among the four concentration values was less 12 than the gas-specific GC detection limit (i.e., < 20 ppm for CO 2 , < 20 ppb for CH 4 and < 20 ppb 13 for N 2 O), in which case no filtering criterion was used. Based on these quality criteria 11% of 14 NEE, 9% of RE, 21% of Rh, 33% of CH 4 and 6% of N 2 O fluxes were discarded from subsequent 15 data analysis. 16

Annual balances 17
To obtain estimates for the annual CO 2 fluxes, non-linear regression models were developed 18 based on the measured CO 2 flux, PAR, WTL and Ta data following Tuittila et al., (2004). As a 19 first step, measured GPP fluxes were fitted to PAR inside the chamber using a hyperbolic 20 function adjusted by a second term which accounted for additional WTL effects (Eq. 1): 21 (1) 23 24 where GPP is gross primary production (mg C m -2 h -1 ), PAR is the photosynthetically active 25 radiation (µmol m -2 s -1 ), α is the light use efficiency of photosynthesis (i.e. the initial slope of the 26 light response curve; mg C µmol photon -1 ), A max is maximum photosynthesis at light saturation 27 (mg C m -2 h -1 ), WTL is the water table level (cm), WTL opt is the WTL at which maximum 1 photosynthetic activity occurs and WTL tol is the tolerance, i.e. the width of the Gaussian response 2 curve of GPP to WTL. 3 Secondly, RE fluxes were fitted to Ta using an exponential function (Eq. 2): 4 (2) 6 7 where RE is ecosystem respiration (mg C m -2 h -1 ), Ta is air temperature (°C), R 0 is the soil 8 respiration (mg C m -2 h -1 ) at 0 °C and b is the sensitivity of respiration to Ta. Both GPP and RE 9 were modeled with hourly resolution using hourly PAR, WTL and Ta as input variables.  Table 2. 15 Annual sums of CH 4 and N 2 O fluxes were estimated by scaling their hourly mean and median 16 flux values, respectively, to annual sums. The median flux was used for N 2 O to avoid a positive 17 bias caused by episodic high peak fluxes measured directly after rainfall events. The annual sums 18 were converted to CO 2 equivalents (CO 2 eq) using the global warming potentials (GWP, over a 19 100-year timeframe including carbon-climate feedbacks) of 34 and 298 for CH 4  in Res-H was always higher than in Res-L with the difference varying between 3 and 10 cm. The 19 mean WTL in BP was -46 cm resulting in mean differences of -22 and -15 cm compared to Res-20 H and Res-L, respectively. 21

Vegetation cover and composition 22
The total surface cover, i.e. the fraction of re-colonized surface area, inside the flux measurement 23 collars was higher in the wetter Res-H (63%) than in the drier Res-L (52%) treatment. 24 Bryophytes were more abundant in Res-H (62%) than in Res-L (44%) ( Table 3). The bryophyte 25 cover consisted primarily of Sphagnum species which contributed 98 and 96% in Res-H and Res-26 L, respectively. Vascular plants occurred more frequently in the drier Res-L (14%) than in the 27 wetter Res-H (4%) treatment and were dominated by woody plants (i.e. shrubs and tree 1 seedlings) ( Table 3). The cover of sedges was <1% in both restored treatments. Res-L, respectively. In contrast, NEE remained positive in BP throughout the growing season and 9 followed the seasonal pattern of Ta with maximum emission rates of 104 mg C m -2 h -1 occurring 10 in early August. The annual mean midday NEE in Res-H and Res-L were significantly lower than 11 in BP, but not significantly different between the two restored treatments (Table 4). Res-L than in BP (Table 4). 18 From early June to late August, both the daytime GPP and NPP were lower (i.e. representing 19 greater production) in the drier Res-L than in the wetter Res-H treatment (Figure 2c Res-H and Res-L. The seasonal patterns in NPP followed closely those of GPP, reaching -65 and 24 -68 mg C m -2 h -1 in Res-H and Res-L, respectively. The growing season mean GPP in Res-H (-25 49.3 mg C m -2 h -1 ) was significantly higher than that in Res-L (-65.5 mg C m -2 h -1 ) ( Table 4). The respectively, were observed in early July (restored treatments) and early August (unrestored BP). 10 The growing season mean Rh was significantly lower (by about 50%) in Res-H and Res-L than in 11 BP (Table 4). h -1 ). The mean annual CH 4 exchange was about two times greater in the wetter Res-H than in the 18 drier Res-L treatment, however, the differences among the three treatments were not statistically 19 significant (Table 4).  (Table 4). Meanwhile, the mean N 2 O 25 exchanges in the two restored treatments were significantly lower (by 1-2 magnitudes) compared 26 to the 27.1 µg N m -2 h -1 in BP (Table 4). 27

Biotic and abiotic controls of greenhouse gas fluxes 1
The differences in mean growing season NEE, GPP, NPP and Ra among individual collars (i.e. 2 the spatial variability) were significantly correlated to bryophyte but not to vascular plant cover 3 in Res-H (Table 5). In contrast, spatial variations in NEE, GPP, NPP and Ra were significantly 4 correlated to vascular plant but not to bryophyte cover in Res-L. In addition, RE was significantly 5 correlated to vascular plant cover in Res-L. Meanwhile, the CH 4

Annual carbon and greenhouse gas balances 17
In the restored Res-H and Res-L treatments, the modelled annual RE estimates were 188.6 and 18 213.2 g C m -2 yr -1 , respectively, whereas in the unrestored BP treatment annual RE was 267.8 g C 19 m -2 yr -1 ( Table 6). The annual GPP was estimated at -78.0 and -110. treatments acted as carbon sources, however, the annual C balance was lower in the restored Res-26 H (110.8 g C m -2 yr -1 ) and Res-L (102.8 g C m -2 yr -1 ) treatments than in the unrestored BP (268.0 27 g C m -2 yr -1 ) treatment. The total GHG balance, including the net CO 2 exchange as well as CH 4 28 and N 2 O emissions expressed as CO 2 eq, was 4.14, 3.83 and 10.21 t CO 2 eq ha -1 yr -1 in Res-H, 1 Res-L and BP, respectively ( Table 6). The GHG balance was driven by the net CO 2 exchange (96 2 to 98%) in all three treatments. The contribution of CH 4   following peatland restoration. Furthermore, the lower RE in the restored treatments relative to 27 BP might also result from the lower temperature sensitivity of Rh, i.e. soil organic matter 28 decomposition, observed in this study which is likely due to greater oxygen limitation in the 29 restored treatments following the raising of the WTL. Thus, our findings highlight the 30 effectiveness of raising the WTL in reducing peat decomposition and CO 2 emissions from 1 drained organic soils. Overall, the potential effects from enhanced anaerobic conditions due to the raised WTL, CH 4 7 oxidation in the moss layer or greater vascular plant substrate supply on the net CH 4 fluxes were 8 small, considering that CH 4 emissions were not significantly different from those in BP which 9 was characterized by a considerably lower WTL and absence of vegetation. Thus, our study 10 suggests that in non-flooded conditions WTL changes following peatland restoration have a 11 limited effect on the CH 4 emissions during the initial few years.  In our study, the annual C source strength of the two restored treatments and the bare peat site 18 was about 1.5 to 2.5 times greater than on the growing season scale. This highlights the 19 importance of accounting for the considerable non-growing season emissions when evaluating 20 the C sink potential of restored peatlands. In comparison, the annual C source strength of the two 21 restored treatments (111 and 103 g C m -2 yr -1 ) was lower than the annual emissions of 148 g C m -22 2 yr -1 reported for a restored cutaway peatland in Canada 10 years following restoration (Strack 23 and Zuback, 2013). Similarly, the C balance of BP (268 g C m -2 yr -1 ) in our study was about half 24 of the 547 g C m -2 yr -1 emitted at the Canadian unrestored site. However, high emissions in the 25 study of Strack and Zuback (2013) were partly attributed to the dry conditions during the study 26 year. Thus, this indicates that restored peatlands are unlikely to provide an annual C sink during 27 the first decade following restoration of peat extraction sites. However, compared to naturally re-28 vegetating peatlands which may require 20-50 years to reach a neutral or negative C balance 29 The similar GHG balances in the two restored treatments Res-H and Res-L suggest that the 4 differences in the mean WTL had a limited effect on the GHG balance within the few years 5 following restoration of the peat extraction area. Moreover, the GHG balances in the restored 6 treatments were driven primarily by the net CO 2 exchange, while the contribution of CH 4 and 7 N 2 O exchanges remained minor in our study. In contrast, 30 years after rewetting of a German 8 bog, high CH 4 emission were reported as the main component of the GHG balance (Vanselow-9 Algan et al., 2015). The same study also reported GHG balances ranging from 25-53 t CO 2 eq ha -10 1 yr -1 which are considerably higher compared to our study. This indicates that the GHG balances 11 of restored peatlands may vary greatly over longer time spans. Moreover, this also suggests the 12 GHG balance of peatland restoration with differing WTL baselines is likely to further diverge 13 over time due to contrasting trajectories in vegetation development and changes in soil 14 biogeochemistry (e.g. pH, nutrient contents and soil moisture dynamics). 15 While the two restored treatments had similar GHG balances, the difference between the GHG 16 balances in restored and BP treatments was considerable. Only three years following restoration, 17 the GHG balance in the restored treatments was reduced to about half of that in BP. This 18 reduction was mainly due to lower annual CO 2 emissions (i.e. lower NEE) in the restored 19 treatments compared to BP likely as a result of increased WTL and vegetation development. In 20 addition, annual N 2 O emissions were also significantly reduced in the restored treatments, 21 although, compared to the differences in the CO 2 balance, the impact of the reduction in N 2 O 22 emissions on the GHG balance was relatively small. Overall, our study suggests that peatland 23 restoration may provide an effective method to mitigate the negative climate impacts of 24 abandoned peat extraction areas in the short-term. However, due to the lack of long-term 25 observations and recent reports of potential high CH 4  We found that differences in the re-established WTL strongly affected the vegetation 2 communities following restoration of the abandoned peat extraction area. Furthermore, the 3 difference in vegetation cover and composition was identified as the main control of within-and 4 between-site variations in GPP, NPP and plant respiration. We therefore conclude that variations 5 in WTL baselines may have important implications for plant-related CO 2 fluxes in restored 6 peatlands. In contrast, differences in the WTL baseline had only small effects on the net CO 2 7 exchange due to the concurrent changes in plant production and respiration in the wet and dry 8 restoration treatments. Moreover, since CH 4 and N 2 O exchanges were also similar in the two 9 restored treatments, this study suggests that differing water table levels had a limited impact on 10 the C and GHG balances three years following restoration. Furthermore, we observed a 11 considerable reduction of heterotrophic respiration in the restored treatments which advocates 12 rewetting as an effective method to reduce aerobic organic matter decomposition in drained 13 peatlands. In contrast, our study suggests that the effects of rewetting on CH 4 fluxes were 14 negligible three years following restoration. However, rewetting reduced the N 2 O emissions by 1-15 2 magnitudes which indicates a high potential of peatland restoration in reducing the N 2 O 16 emissions commonly occurring in drained peatlands. Three years following restoration, the C and 17 GHG balances of the restored treatments were reduced by approximately half relative to those of 18 the abandoned bare peat area. We therefore conclude that peatland restoration may effectively 19 mitigate the negative climate impacts of abandoned peat extraction areas; however, longer time 20 spans may be needed to return these sites into net C sinks.   Table 2. Parameters for the gross primary production (GPP) and ecosystem respiration (RE) 1 models in restoration treatments with high (Res-H) and low (Res-L) water table level and bare 2 peat (BP); α is the quantum use efficiency of photosynthesis (mg C µmol photon -1 ), A max is the 3 maximum rate of photosynthesis at light saturation (mg C m -2 h -1 ); WTL opt is the WTL at which 4 maximum photosynthetic activity occurs; WTL tol is the tolerance, i.e. the width of the Gaussian 5 response curve of GPP to WTL; R 0 is the soil respiration (mg C m -2 h -1 ) at 0 °C, b is the 6 sensitivity of respiration to air temperature; numbers in parenthesis indicate standard error; Adj. 7   Table 4. Means of measured CO 2 fluxes (mg C m -2 h -1 ) including net ecosystem exchange (NEE), 1 ecosystem respiration (RE), gross primary production (GPP), net primary production (NPP), 2 autotrophic respiration (Ra) and heterotrophic respiration (Rh) as well as means of measured 3 methane (CH 4 ; µg C m -2 h -1 ) and nitrous oxide (N 2 O; µg N m -2 h -1 ) fluxes in restoration 4 treatments with high (Res-H) and low (Res-L) water table level and bare peat (BP). Negative and 5 positive fluxes represent uptake and emission, respectively. Numbers in parenthesis indicate 6 standard error; different letters indicate significant (P < 0.05) differences among treatments. 7 Table 5. Correlation coefficients of vegetation (bryophytes and vascular plants) cover (%) with mean growing season CO 2 fluxes 1 including the net ecosystem CO 2 exchange (NEE), ecosystem respiration (RE), gross primary production (GPP), net primary 2 production (NPP) and autotrophic respiration (Ra) and with mean growing season methane (CH 4 ) and nitrous oxide (N 2 O) fluxes in 3 restoration treatments with high (Res-H) and low (Res-L) water table level. Total vegetation represents the sum of bryophyte and 4 vascular plant cover; significant correlations are marked with asterisks ( * indicates P < 0.05 and ** indicates P < 0.01).  Table 6. Growing season (GS; May 1 to October 31) and annual (A) sums of the carbon 1 balance components (g C m -2 ) including gross primary production (GPP), ecosystem 2 respiration (RE), net ecosystem exchange (NEE) of CO 2 , and methane (CH 4 ) fluxes as well as 3 of the greenhouse gas (GHG) balance components (t CO 2 eq ha -1 ) including NEE, CH 4 and 4 nitrous oxide (N 2 O) exchanges (using global warming potentials of 34 and 298 for CH 4 and 5 N 2 O, respectively) in restoration treatments with high (Res-H) and low (Res-L) water table  6 level and bare peat (BP). Negative and positive fluxes represent uptake and emission, 7 respectively. 8