Introduction

Natural peatlands represent a tremendous sink of atmospheric carbon dioxide (CO2) with an accumulation of about 500 ± 100 Gt of the global soil carbon (Yu 2012). Drainage of peatlands for agriculture and forestry use resulted in an accelerated mineralisation of stocked carbon (C) and led to a substantial loss of C and nitrogen in form of CO2 and nitrous oxide (N2O), respectively, of globally 2–3 Gt CO2-eq per year (Joosten and Couwenberg 2009). During the last four decades, rewetting programmes were initialized in Central Europe aiming at reducing greenhouse gas (GHG) emissions and thus, to recover the nutrient sink and ecological functions of pristine peatlands (e.g. Drösler 2005). It has been shown recently that rewetted peatlands may evolve to highly dynamic systems characterised by an increased mobilisation of nutrients, dissolved organic carbon (DOC) and elevated emissions of methane (CH4) and CO2 as a result of rapid plant dying and decomposition (Zak et al. 2015). Extremely high CH4 emissions from rewetted peatlands have been observed and attributed to enhanced C and nutrient availability via the supply of fresh labile organic substrates resulting from the die-off of plants in formerly non-flooded areas (Wilson et al. 2009; Hahn et al. 2015). Afterwards prevailing plant species revealed a major impact on the CH4 release from rewetted peatlands. First, they are assumed to stimulate CH4 formation due to root exudation of low-molecular organic compounds from living plants (Hines et al. 2008). Second, they provide freshly reduced C from senescent plant biomass (Lawrence et al. 2013). And third, they permit transport of CH4 from the anaerobic root zone directly into the atmosphere, by bypassing the aerobic methane-oxidizing peat layers via their well-developed aerenchyma tissue (Whalen 2005). Concomitantly also CH4 oxidation could occur in the oxygenated zone surrounding the plant roots (Chanton et al. 2002). In addition, the litter quality, in terms of microbial usability of a potential substrate, may differ substantially between plant species (Lai 2009), resulting in differences of CH4 emissions between various plant species (Zak et al. 2015). It was shown that CH4 production by decomposition of plant litter with higher nitrogen to C content is supported whereas plant litter composition rich in lignin and polyphenols suppress CH4 production (Freeman et al. 2001; Thangarajan et al. 2013). Thus, differences in CH4 emissions between rewetted peat sites can be expected due to differences in plant litter quality of the prevalent plant species.

In contrast to CH4, rewetted and natural peatlands are generally characterised by low fluxes of N2O, due to anoxic conditions, which inhibit nitrification processes and cause nitrate limitation at high water saturation (Drösler 2005; Beyer and Höper 2015). Heterotrophic denitrification represents the most important N2O source in water-logged peat soils (Regina et al. 1996). It was observed previously that N2O produced from fertilized nitrate via denitrification was stored in the water phase of water-logged peat soils and later on released to the atmosphere only by falling water tables (Aerts and Ludwig 1997; Tauchnitz et al. 2015). As a consequence temporarily strong N2O bursts can be expected in rewetted peat sites which are characterized by considerably fluctuating water tables. In contrast to CH4 permanently flooded conditions therefore appear to be at least favourable to decrease N2O emissions to near zero (Beyer and Höper 2015).

Most rewetting studies in temperate regions were performed in formerly agriculturally used fens with vegetation dominated by hydrophytes. In temperate regions, however, bogs with an oligotrophic vegetation have scarcely been studied so far. Therefore, the present study aimed at quantifying CH4 and N2O emissions of a drained transition bog (currently undergoing a rewetting process) over a period of four years by using common closed chamber technique. The main objectives of the study are:

  1. (i)

    Examining changes in the CH4 and N2O release in the course of progressing rewetting,

  2. (ii)

    Evaluating differences of the CH4 and N2O emission patterns between (a) permanently flooded conditions and (b) non-flooded conditions with fluctuating water table (mostly below the peat surface),

  3. (iii)

    Identifying main drivers of CH4 and N2O release at the study site.

Materials and methods

Study site

Field experiments were performed in the German National Park Harz Mountains (central Germany) in a transition bog ecosystem (650–670 m a.s.l) in direct vicinity to the Mt. Brocken (1141 m a.s.l.; 51°48′N, 10°37′E). A detailed description of the study site was previously given by Osterloh et al. (2016). Climate conditions are characterised by a long-term (1961–1990) mean annual temperature of 5.3 °C and a mean annual precipitation of 1278 mm (data from German Weather Service).

Prior to rewetting the study site was drained by several ditches and planted with spruces for forestry use in the 1980s. Rewetting was initiated in 2005 and continued in 2009 and 2010 by closing the ditches and redirecting a small stream into the central peatland area.

For emission analyses two different study sites were chosen within the peatland. One site was characterised by permanently flooded conditions (i.e. water table (WT) above the soil surface since August 2009). Consequently, vegetation changed from grass dominated to sedge-grass-moss dominated structures and plant productivity increased simultaneously.

The second study site was rewetted since June 2010, non-flooded and characterised by a fluctuating WT. The vegetation is characterised by an open young spruce stand rich in Vaccinium myrtillus and mosses like Polytrichum formosum and Pleurozium schreberi but rare in wetland plant species. Rewetting caused a die-off of typical acidic forest plant species followed by a spreading of fen indicator species (namely Carex canescens, Agrostis canina, Molinia caerulea and Sphagnum spec.).

The peat soil layers of both sites are 0.7–1.1 m thick. Dry density is in the range of 0.15–0.22 g cm3. Both peat sites are characterised by mesotrophic conditions with C:N ratios ranging from 22 to 29. Nitrate contents are low with a maximum of 10 mg g−1 dry matter of soil. Detected pH values are in a range of 3.4–3.9 and represent strong acid conditions.

Field sampling and analysis

At each sampling site five polyethylene (PE)-tubes (30 cm diameter, 30 cm height) were pushed 15 cm into the peat soil layer acting as the emission chambers frame. From June 2010 to October 2014 gas release was measured weekly and biweekly (in some cases monthly) by the closed chamber method. Gas measurements were omitted during the winter time (December to February) in case of thick snow covering. For gas sampling, a peat soil column was covered with a dark, opaque PE-chamber (31.5 cm in diameter, 15 cm height) and sealed against the atmosphere by a rubber strip. Chamber then remained closed for two hours while gas samples (aliquot of 100 ml) were transferred by a polypropylene syringe via a three-way valve and a needle from the chamber headspace into a septum-capped vial (20 ml) every 20 min (Tauchnitz et al. 2015). Before gas sampling, all vials were flushed with oxygen to prevent contamination from previous samplings. Gas samples were analysed within a few days for CH4 and N2O by a specially configured gas chromatograph (GC 14B, Shimadzu, Japan) equipped with a flame ionization detector (FID for CH4) and an electron capture detector (ECD for N2O) (see Segschneider et al. 1996). Emission rates (E) of CH4 and N2O (µg m−2 h−1) were calculated according to Eq. 1 with ΔC (µg N per norm liter) as the amount of CH4 and N2O emitted into a normalized gas volume between t = 0 min and t = 20, 40, 60, 80, 100 and 120 min, respectively, and h (current champers height) multiplied by 10 and divided by the time in order to correct for the current chamber volume, the covered area and the corresponding closure time. Prior to the calculation of the volume normalized gas concentrations analysed gas concentration was always corrected to standard temperature (i.e. 273 K) and standard pressure (i.e. 1013.25 hPa).

$$E = \Delta C_{t} \cdot h \cdot 10/t$$
(1)

Daily emission rates between the measurement campaigns were obtained by linear interpolation and were used to calculate yearly balances.

Soil temperatures were measured with a puncture thermometer (stainless steel sensor, VWR International GmbH) in 5 and 10 cm depths concomitantly to the gas measurements. In addition to gas sampling WT, redox potential (Eh, Hach Lange GmbH, Germany) as well as oxygen level (O2, Hach Lange GmbH, Germany) was monitored simultaneously by groundwater piezometers placed 20 cm deep in the peat soil layer adjacent to the chambers. Furthermore, water samples were taken and analysed for ammonium (NH4 +, photometric, Thermo Electron Corporation, Germany), nitrate, sulfate (NO3 −, SO4 2−, ion chromatography, Metrohm GmbH & Co. KG Filderstadt, Germany), DOC (Total Carbon Analyser, NPOC method, Shimadzu TOC-VCSH, Japan), iron (Fe, flame atomic absorption spectrometry, Thermo Electron Corporation, Germany) as well as pH (potentiometric, WTW GmbH, Germany) and electric conductivity (EC, conductimetrically, WTW GmbH, Germany). Total N and C contents of plants were determined by high temperature combustion method (Elementar Analysensysteme GmbH, Germany). Total P contents of biomass were analysed by inductively coupled plasma optical emission spectrometry (Ultima 2, Horiba Jobin–Yvon). For analyses of dissolved CH4 concentration, peat soil water was sampled according to Tauchnitz et al. (2015) and measured by a Headspace GC (HSS 86.50, DANI, Italy) coupled to a GC system (GC 14B, Shimadzu, Japan) equipped with a Porapak Q column (80/100 mesh, 6L, 2 mm ID) and an ECD.

Statistical analyses

Statistical analyses were performed using SPSS (vers. 22). Levene-test was used to examine data sets concerning their equality of variances. Since data were not normally distributed the non-parametric Mann–Whitney-U-test was applied to check for significant differences between peat site locations and years, respectively. Spearman correlation coefficients were calculated to test correlations between gas fluxes and environmental as well as nutrient conditions (e.g. WT, NH4 +-, NO3 − -contents, etc.).

Results

Weather and hydrological parameters

Annual precipitation ranged between 974 mm (2011) and 1428 mm (2010) during the study period and was, with the exception of 2010, below the long-term mean (1278 mm) of the study site (Table 1). Annual mean temperature was in a range of 4.6 °C (2010) to 7.3 °C (2014).

Table 1 Annual precipitation (P), air temperature (T) and water table (WT) related to soil surface from the flooded and non-flooded peat sites during the study period (June 2010–October 2014); different letters indicate significant differences (p < 0.05, n = 49, Man-Whitney-U-Test) between years per site (U1) and sites per years (U2)
Full size table

Water tables were significantly different between the two sampling sites. At the flooded peat site a permanently high WT with an annual variation of 2.5 cm (2012) to 4.3 cm (2014) above the soil surface was observed. WT at this site showed very low temporal variations (Table 1). In contrast, the recently rewetted non-flooded peat site showed a significantly lower WT (Table 1) with annual means in the range of 11.7 cm (2011) to 1.8 cm (2014) below the soil surface and high temporal variations. WT decreased markedly during summer with lowest value of 33 cm below the soil surface observed at the end of September 2011. In the same year, lowest annual mean WT coincided with the lowest annual precipitation of the whole study period. Nonetheless, in total an increasing trend of the WT was observed for the non-flooded peat site accompanied by decreasing temporal variations, presumably due to progressive rewetting (Table 1).

Characteristics of the peat soil water

Median redox potentials of the flooded and non-flooded peat site were 187 and 259 mV, respectively, with significant differences between both sites (Table 2).

Table 2 Peat soil water properties from the flooded and non-flooded peat sites during the study period (June 2010–October 2014), different letters indicate significant differences (p < 0.05, Man-Whitney-U-Test) between sites (U)
Full size table

NH4 + concentrations and EC of the peat soil water differed significantly between both sites with higher values at the non-flooded site (Table 2). NO3 − concentrations of the peat soil water were low at both sites with median values of <0.1 mg N L−1 and maximal concentrations of 1.5 mg N L−1 (observed during strong snow melt coinciding with high precipitation in April 2013; Tables 1, 2). DOC concentrations of the peat soil water (Fig. 1) showed no significant differences between both sites with median values of 21 mg L−1 (flooded) and 26 mg L−1 (non-flooded). Most notable was a distinct temporal variation in DOC concentrations with highest values during summer.

Fig. 1
figure 1

Temporal course of soil and water temperature (T) and dissolved organic carbon (DOC) from the flooded and non-flooded peat sites (n = 4) during the study period (March 2010–December 2014); within the DOC diagrams black lines indicate means, box represent 25-, 50 and 75%-quantiles, dots are outliers

Full size image

Nutrient contents of biomass

N contents of biomass at the flooded and non-flooded site were in the range of 1.4–3.9% and 1.1–4.1%, respectively (Table S-1). C:N ratios of biomass ranged between 11 and 31 (flooded) and 11 and 43 (non-flooded) and P contents between 0.1 and 0.4% (flooded) and 0.07 and 0.4% (non-flooded). During the study period N and P contents significantly increased, while C:N ratios of biomass decreased at both site.

CH4 emissions

Highest CH4 emissions were measured at the permanently flooded peat site ranging from 1.5 to 1195 mg C m−2 d−1 (Fig. 2; Table S-2).

Fig. 2
figure 2

Temporal course of the CH4 emissions and water table (WT) from the flooded and non-flooded peat sites (n = 5) during the study period (June 2010–October 2014), dashed line in the WT diagram represent soil surface; black lines in the emission diagrams indicate means, box represent 25-, 50 and 75%-quantiles, dots are outliers

Full size image

In addition, a pronounced inter-annual and spatial variation of CH4 emissions was found. Highest CH4 release was generally observed during summer associated with high temperatures. Overall, CH4 emissions at the flooded site showed no significant trend during the whole study period (Fig. 2; Table S-2). Significantly lower CH4 release was observed at the non-flooded site where emission rates largely tended towards zero. Nonetheless, also higher rates could be detected (up to 571 mg CH4–C m−2 d−1), but mainly at the end of the study period where WT fluctuations diminished and WT remained constantly close the soil surface (Fig. 2).

Mean annual CH4 emissions of the permanently flooded and non-flooded site accounted for 505 ± 414 kg C ha−1 a−1 and 68 ± 60 kg C ha−1 a−1, respectively.

N2O emissions

N2O emissions showed significant differences between both peat sites (p < 0.05, n = 49, Man-Whitney-U-test). High N2O release rates of maximal 4178 µg N m−2 d−1 were found at the non-flooded peat site (Fig. 3; Table S-3) within the first years of the study with highest emissions during summer coinciding with low WT and high temperatures.

Fig. 3
figure 3

Temporal course of the N2O emissions and water table (WT) from the flooded and non-flooded peat sites (n = 5) during the study period (June 2010–October 2014), dashed line in the WT diagram represent ground surface; black lines in the emission diagrams indicate means, box represent 25-, 50 and 75%-quantiles, dots are outliers

Full size image

Nonetheless, N2O emissions showed a decreasing trend associated with increasing WT (with exception of the sampling date 25.07.2013). Furthermore, decreasing inter-annual variation of N2O emission was observed. High N2O release rates occurred in 2010, 2011, and 2012 with annual median values of 193, 836 and 415 µg N m−2 d−1, respectively. By contrast, markedly lower N2O loss was observed in the study years 2013 and 2014 which were, however, only significant in 2014 (p < 0.05, n = 49, Man-Whitney-U-test).

At the permanently flooded peat site N2O emissions were generally low with a maximal N2O release of 200 µg N m−2 d−1. For 63% of the sampling dates no significant N2O release (i.e. >24 µg m−2 d−1) could be detected, whereas an uptake of N2O was observed at numerous sampling dates. Contrary to the non-flooded peat site N2O release of the flooded site showed very low inter-annual and spatial variations. Median annual N2O values indicate no significant trend throughout the study period (Fig. 3). Annual N2O emission of the permanently flooded site accounted for −0.03 ± 0.1 kg N ha−1 a−1. At the non-flooded site significantly higher emissions of 1.6 ± 1.5 kg N ha−1 a−1 were found (p < 0.05, n = 4, Man-Whitney-U-test).

Discussion

CH4 emission rates and associated drivers

Highest CH4 emissions of up to 1195 mg C m−2 d−1 were observed at the permanently flooded peat site (Fig. 2; Table S-2). Similar results for flooded peatlands have been reported elsewhere (Augustin and Joosten 2007; Hahn-Schöfl et al. 2011; Hahn et al. 2015). High emissions are caused by rapid plant die-off and related decomposition processes along with the flooding and thus, providing fresh and easily degradable organic substrates and therefore enhanced C and nutrients availability for methanogenic microorganism (Wilson et al. 2009; Hahn et al. 2015). It was stated elsewhere that the first stage of litter break down during flooding subsequent to die-back of plants resulted in high mass loss of plant nutrient stock (e.g. Aerts and De Caluwe 1997), hydrolysis of polyphenols and other organic substances of mostly low molecular weight, such as carbohydrates and amino acids (Maie et al. 2006). This is supported by the increased mobility of DOC observed during the rewetting process at this site (Osterloh et al. 2016). A positive correlation between CH4 emissions, which are known as a good proxy for biomass production (Strack et al. 2004), and DOC concentrations in peat soil water were additionally found at the flooded peat site (Table 3).

Table 3 Spearman correlation coefficients between methane (CH4) and nitrous oxide (N2O) emissions from the flooded and non-flooded peat sites and relevant parameters
Full size table

It was also shown recently that rewetted sites at the studied peatland are characterised by a more unbalanced DOM quality with higher fluorescence and humification indices (HIX), indicating a higher level of aromatic DOM and a stronger humification compared to pristine sites (Herzsprung et al. 2017; Table 4). These findings are well in agreement with Kalbitz et al. (2000), who reported changes of DOM quality during peat degradation with higher amounts of aromatic moieties in degraded peatlands. A high content of aromatic components might contradict with the idea of enhanced C availability via easily degradable organic substrates derived from submerged plants which was previously discussed (e.g. Zak et al. 2015). However rewetted sites are characterised by large seasonal and spatial variations in DOM quality (Herzsprung et al. 2017), indicating a high dynamic of rewetted systems.

Table 4 Humification index (HIX) and humic-like fluorescence of peat soil water in 20 cm depth of pristine and rewetted peat sites during sampling time of May 2013–April 2014; different letters indicate significant differences (p < 0.05, Man-Whitney-U-Test) between sites (U)
Full size table

A positive correlation was found between CH4 emissions and EC, as well as DON concentrations of peat soil water at the permanently flooded site, confirming the positive effect of nutrient availability on microbial activity (Table 3). Note that decomposing microorganisms depend on N for their anabolism, thus increasing N concentrations induced by higher litter input suggest enhanced microbial activity (Tremblay and Benner 2006). Therefore, CH4 production induced by fresh plant litter input during the rewetting processes might differ substantially between various plant species depending on their litter quality (Lai 2009). Zak et al. (2015) obtained a 70-times higher CH4 production under the hydrophyte Ceratophyllum demersum (C:N ratio: 14, N content: 2.8%, P content: 0.68%) compared to Carex riparia (C:N ratio: 35, N content: 1.6%; P content: 0.06%). In the present study the C:N ratios, as well as the N and P contents of main plant species at the flooded site (e.g. Carex canescens, Table S-1) are in the range of those reported by Zak et al. (2015). Also Bohdálková et al. (2013) presented highest CH4 emissions from an oligotrophic peat bog in the Czech Republic liked to the occurrence of Carex. The high CH4 production rate under Carex species (and other vascular plants) could be attributed to their relatively fast decomposition rates compared to moss species (Lahio 2006). In agreement with other studies (e.g. Aerts and Ludwig 1997) we observed a high spatial variability of CH4 emissions with coefficients of variation often exceeding 100% within the replicates (Fig. 2; Table S-2). Similar to our results a high inter-annual variation of CH4 emissions was recognized in recent rewetting studies (Koebsch et al. 2013; Günther et al. 2015). Overall, highest CH4 release occurred at high temperatures during summer months, as supported by the positive correlation between CH4 emissions and soil temperature at the permanently flooded site. Also Audet et al. (2013) and Beyer and Höper (2015) found increased CH4 production rates under warmer conditions. Additionally, a significant negative correlation between CH4 emissions and SO4 2− concentrations of peat soil water was observed at the flooded peat site (Table 3). This finding confirms previously published results by Audet et al. (2013). CH4 is produced under strictly anaerobic conditions by methanogenic archaea species as a result of fermentation of organic compounds (acetate) or reduction of CO2 by H2 (Le Mer and Roger 2001). The negative effect of SO4 2− can be explained by the competition between SO4 2− reducing bacteria, using SO4 2− as an electron acceptor, and methanogens regarding the utilization of H2 and acetate (Conrad 2007). Note that methanogenesis might be also inhibited in the presence of NO3 − and iron (Fe(III)) as electron acceptors (Le Mer and Roger 2001). However, we could not confirm this latter suggestion. Contradicting, a positive correlation between CH4 emissions and Fe concentrations of peat soil water was found, which was assumed rather a result of the interaction between DOC and Fe due to the formation of chelate complexes (Osterloh et al. 2016).

Significantly lower CH4 emissions were detected at the non-flooded peat site with fluctuating WT compared to the permanently flooded site (Fig. 2). Note that the WT level is one of the key factors of CH4 emissions in peatlands determining the thickness of the aerobic soil layer where CH4 oxidation takes place (Lai 2009; Wilson et al. 2009). In contrast to Jungkunst et al. (2008) who found a non-linear correlation between WT position and CH4 emissions, no significant correlation was detectable in the present study at the non-flooded peat site (Table 3). Nevertheless, higher CH4 emissions coincide with higher WT throughout the whole study period (Fig. 2). Differences between the flooded and non-flooded peat site were found to be concerning CH4 concentrations within the peat soil water (Table 5). In addition, no significant correlation between temperature and relevant hydro-chemical parameters to CH4 emission could be seen at the non-flooded site with fluctuating WT, in contrast to the permanently flooded site (Table 3), supporting the assumption of differences in CH4 production conditions between the both sites. A very effective way to channel large amounts of CH4 from the anaerobic zone directly into the atmosphere, i.e. bypassing the aerobic methane-oxidizing peat layers, is a plant mediated transport through the aerenchyma of aquatic vascular plants (Whalen 2005). However, such an effect, mainly induced by wetland plants (reed plant species), are not relevant for the here presented data due to their absence on both study sites.

Table 5 Dissolved methane (CH4) concentrations in peat soil water in 30 and 60 cm depth from the flooded and non-flooded peat sites during sampling time of September 2010; different letters indicate significant differences (p < 0.05, Man-Whitney-U-Test) between sites per depth (U)
Full size table

In 2010 (beginning of the study period), the non-flooded peat site showed significantly lower CH4 concentrations in depths of 30 and 60 cm compared to the permanently flooded peat site. Already Chanton and Whiting (1995) described that at high rates of methanogenesis supersaturation of CH4 might occur in the porewater of deeper anaerobic peat layers and hence, gas bubbles are formed when the partial pressure of all dissolved gases in solution exceeds the hydrostatic pressure in soil water. These bubbles accumulate within the peat profile and can be released suddenly in case of a rapid drop of atmospheric pressure, a lowering of hydrostatic pressure (e.g. via decreasing WT), a rise in temperature or mechanical stress by e.g. walking on the peat surface (Tokida et al. 2007). It is well known that the process of ebullition may contribute significantly to the total CH4 fluxes in peatlands (Tokida et al. 2007). It can be assumed that the high inter-annual and spatial variation of CH4 emissions observed in our study largely resulted from ebullition events.

N2O emissions rates and associated drivers

Generally, low N2O emissions were observed at the permanently flooded peat site, which is in line with results from previous studies reporting negligible N2O release from rewetted peatlands (Drösler 2005; Beyer and Höper 2015; Günther et al. 2015). Note that heterotrophic denitrification should be the major process of N2O formation in peatlands (Regina et al. 1996). In fact, a weak positive correlation between N2O release at the flooded peat site and NO3 − concentration in the peat soil water was observed indicating that N2O formation mainly depended on the availability of NO3 − as a source for heterotrophic denitrification (Table 3). It could therefore be assumed that the process of denitrification is still active at this peat site, even at very low soil pH, as long as NO3 − is available (Tauchnitz et al. 2015). Actually Rückauf et al. (2004) stated that N2 should be the predominant product of denitrification in water-saturated peatlands. This could explain the extremely low N2O release rates observed at the permanently flooded peat site. The permanent flooding could limit the mobility of N2O in the sense of moving towards the atmosphere and hence N2 production is favoured (i.e. accumulation and consumption of N2O) over N2O release (Jørgensen et al. 2012). Previous studies already revealed an accumulation of N2O in peat soil water after NO3 − addition, which was then released to the atmosphere only at falling WT (e.g. Tauchnitz et al. 2015). Conversely, Krüger et al. (2014) found a positive response of rising WT on N2O emission during rewetting of a temporally dry wetland in Southern Africa. These high N2O emissions after short-term changes in WT were explained by a flushing out of N2O which had accumulated in deeper zones. In contrast to the flooded site significantly higher N2O emission rates were observed at the non-flooded site associated with lower WT and considerable WT fluctuations (Table 1). Similar results concerning a significant response of N2O release on WT fluctuations were reported previously (Aerts and Ludwig 1997; Tauchnitz et al. 2015). It was pointed out that N2O fluxes of water-logged soils are (besides the availability of NO3 −) mainly controlled by the balance of production and consumption of N2O and by convective or diffusive transport (e.g. Well et al. 2003). As mentioned above N2O accumulates in the saturated zone when NO3 − was added to a rewetted peat soil. This accumulation then occurs under constant high WT conditions due to a transport limitation towards the atmosphere. A falling WT leads then to a release of accumulated N2O based on a significant decrease of diffusional transport limitations, i.e. a direct connection of the soil gas phase to the atmosphere (e.g. Tauchnitz et al. 2015). A significant negative correlation between WT and N2O emissions was detected at the non-flooded peat site (Table 3), which proves the impact of falling WT on N2O release. However it is also known that N2O emission of water-logged soils can be promoted by plants through rapid gas phase diffusion in the aerenchyma of wetland plants (reed plant species) (Jørgensen et al. 2012). But such species were not present at our study site.

In addition to the above mentioned negative correlation between WT and N2O emissions, a positive relationship between soil (and water) temperature and N2O emissions was found for the non-flooded peat site (Table 3). It is known from other studies (e.g. Silvan et al. 2002) that microbial processes (such as denitrification) are favoured by warm conditions and falling WT. This framework may support NO3 − formation via nitrification processes, causing high N2O release during summer at simultaneously wet conditions. However, increased N2O emissions during freeze–thaw-cycles as shown e.g. by Jungkunst et al. (2006), could not be confirmed here. Moreover, a significant positive correlation between N2O release and DOC concentrations of the peat soil water was observed at the non-flooded peat site (Table 3). Since most denitrifying bacteria couple organic C oxidation and NO3 − oxidation to gain energy (Knowles 1982), it is obvious that a high supply of readily biodegradable organic C would favour denitrification. The impact of a high availability of organic C on N2O release was already shown in other studies (e.g. Eickenscheidt et al. 2014). The positive relationship between N2O emission and dissolved organic N (DON) concentrations of peat soil water at the non-flooded peat site furthermore indicates a stimulation of microbial activity due to increasing amounts bioavailable N (Glatzel et al. 2006).

Nitrate contents of the peat soils at both peat sites were low and showed no significant differences among each other. Furthermore no differences in NO3 − contents of peat soil and peat soil water during the study period could be identified (unpublished data), which may indicate an enhanced NO3 − availability via nitrification. We therefore suggest that NO3 − mainly originates from atmospheric N deposition. This assumption is supported by the atmospheric NO3 − deposition of 18 kg N ha−1 a−1 observed at the study site (Tauchnitz et al. 2010). Hill et al. (2016) also found a positive correlation between potential denitrification and atmospheric NO3 − input for a bog watershed. Another pathway might be the NO3 − transport via slope water into the bog (Tauchnitz et al. 2010).

Evolution of CH4 and N2O emission in the course of rewetting

The annual CH4 emission accounted for 68 ± 60 and 505 ± 414 kg C ha−1 a−1 at the non-flooded and flooded peat site, respectively. Thus, a global warming potential over 100 years (GWP100) of 1428 (non-flooded site) and 10605 kg CO2-eq. ha−1 a−1 (flooded site) can be derived considering a 21-fold higher global warming potential of CH4 compared to CO2 (IPCC 1996). The results are in line with previous studies of rewetted temperate peatlands showing a wide range of −0.75 to 3525 kg C ha−1 a−1 (Meyer et al. 2001; Hahn et al. 2015) with the highest CH4 release observed in a flooded coastal fen in Northeast Germany (Hahn et al. 2015).

At the permanently flooded peat site CH4 release remained high during whole study period but revealed a small decreasing trend within five years after flooding. Similar high CH4 emissions, recently after flooding were found in fens in Northeast and Northwest Germany where CH4 was presumably emitted from a newly formed litter layer derived by dead plant residues (Meyer et al. 2001; Günther et al. 2015; Hahn et al. 2015). In contrast, Drösler (2005) showed markedly lower CH4 fluxes of a recently flooded bog heathland in South Germany which in part remained below common CH4 emissions of natural bog sites. After Höper et al. (2008) degradation of fen mires after flooding occurs in three different phases with different characteristics. The first initial phase is characterised by extremely high CH4 emissions associated with a low net CO2 uptake, resulting in a negative climate effect. In the second phase a strong reduction of CH4 emissions occurs accompanied by an increase of CO2 uptake and thus, a slightly positive climate effect. Whereas in the third phase low CH4 release and low net CO2 uptakes can be expected (Asaeda et al. 2002). Until yet little is known about the duration of those phases because most rewetting studies are based on data gathering shortly after rewetting. Beyer and Höper (2015) observed high CH4 emissions with inter-annual variations at two flooded sites of a bog in Northwest Germany even 30 years after rewetting. Drösler (2005) reported that after 10 years of renaturation the optimal conditions for low greenhouse gas emissions had not been achieved. Besides that, Günther et al. (2014) found that after 15 years of rewetting the net greenhouse gas balance of a fen in Northeast Germany was similar to those of pristine fens. Note that the present study represents the initial phase of rewetting.

In contrast to the permanently flooded peat site, a small increasing trend of CH4 emissions was observed at the non-flooded site with fluctuating WT, associated with an increasing WT and observed dieback of plants, which was also reported by Hahn et al. (2015).

Different approaches were discussed to mitigate CH4 emissions of flooded fen peatlands. When keeping the WT on a constant level, with a threshold value for the mean annual water level of about 10 cm below the soil surface, high CH4 emissions during rewetting could be prevented (Höper et al. 2008). Other wetland studies have demonstrated that an aerobic layer of 20–30 cm is sufficient to ensure low CH4 emissions (Koebsch et al. 2013) which are, however, unsuitable for establishing Sphagnum species. Zak et al. (2015) suggested the removal of upper degraded peat layers in order to minimize nutrient availability. As another possibility with the same goal biomass harvesting, which could be used then for energetic purposes, was proposed (Günther et al. 2015).

The annual N2O emissions of the permanently flooded and non-flooded peat site accounted for −0.03 ± 0.1 and 1.6 ± 1.5 kg N ha−1 a−1, respectively. Thus, a GWP100 of −9.3 and 496 kg CO2-eq. ha−1 a−1 can be expected when considering a 310-fold higher GWP of N2O compared to CO2 (IPCC 1996). The results are in line with previous studies of restored temperate peatlands showing a range of −0.42 to 3.2 kg N ha−1 a−1 (e.g. Meyer et al. 2001; Beyer and Höper 2015) with highest N2O release reported for rewetted fens.

At the permanently flooded peat site no significant trend in N2O emission was observed during the whole study period (five years after rewetting started). Gas measurements started one year after rewetting already at permanent water saturation, which resulted in negligible N2O emissions. These results were confirmed by other rewetting studies who reported very low N2O fluxes and even a net sink function of flooded peatlands (Meyer et al. 2001; Drösler 2005; Günther et al. 2015). The non-flooded peat site with fluctuating WT also showed no significant trend in N2O emission throughout the study period. Nevertheless, annual differences with significantly lower N2O release rates towards the end of study could be observed, most likely due to less intensive WT fluctuations and generally higher WT levels. Particularly in the first two years, parallel to the progressing rewetting, N2O emissions reached their maximum (up to 4178 µg N m−2 d−1), associated with a significantly decreasing WT during summer. These high emissions did not occur as strong three years after rewetting. However, one significant N2O outlier with a release rate of 11454 µg N m−2 d−1 could be observed on 25th of July 2013 (Fig. 3), which was presumably caused by optimal conditions for N2O formation, i.e. high soil temperatures (15.2 °C in 5 cm depth) combined with a WT just slightly below the peat soil surface (−2 cm) and a weak precipitation event (4 mm) shortly before gas measurements.

A weak negative trend of N2O emissions at the non-flooded peat site with lower and fluctuating WT was observed since 2012 (Fig. 3). Drösler (2005) showed lower N2O release rates of maximal 672 µg N m−2 d−1 at a bog in South Germany flooded parallel to the gas measurements. However, the mean water table was thereby 44.5 cm above the peat soil surface with low WT fluctuations. Meyer et al. (2001) reported N2O emissions near zero and even an uptake of N2O at a recently flooded (parallel to gas measurements) shallow grassland fen in Northwest Germany with a maintained WT of −10 cm. At an also lately, but less rewetted site (WT −30 cm below soil surface) they observed slightly higher N2O emissions. Also Augustin and Joosten (2007) described a rewetted grassland fen in Northeast Germany as a weak source of N2O (0.7 kg N ha−1 a−1). In general, it is well known that frequently alternating wet-dry-cycles may enhance N2O emissions (e.g. Harrison-Kirk et al. 2013). Concerning the impact of N2O emissions during rewetting a fast adjustment of permanent high WT near the peat soil surface (i.e. a permanent flooding) should be preferred. This will have, however, negative consequences with respect to CH4 release which will then significantly increase.

Conclusions

The present study showed an increasing trend of CH4 release and decreasing N2O emissions during rewetting. Permanent flooded conditions caused the highest CH4 release. Thus, the present results reveal that not only flooded fens with high nutrient availability of decomposed plant litter but also bog can contribute to very high CH4 emissions when kept under permanently flooded conditions. Approaches to mitigate CH4 emissions from rewetted peatlands as discussed in literature (e.g. removal of upper degraded peat soil layer to minimize nutrient availability or biomass harvesting) are not realizable in nature reserves as studied here. Also the maintenance of a constant WT near 10 cm below the peat soil surface appears to be practically unfeasible due to spatial and seasonal variations at the large scale.

The main objective in the presented rewetting program was the recovery of ecological functions by establishing peat forming plants (Sphagnum) and hence, the long-term C sink function of the peatland. Present results have revealed that five years after rewetting no decreasing trend of CH4 emissions were detected. It could be expected that GHG emissions will decrease in the long term after rewetting. However, in order to generate a better understanding of the functionality of restored peatlands and to improve the planning process of rewetting approaches long-term GHG emission monitoring and the development of vegetation patterns after rewetting are required. Furthermore, the impact of vegetation structure and quality of dissolved organic carbon on CH4 release should be addressed more in detail in further research.