Greenhouse gas exchange of rewetted bog peat extraction sites and a Sphagnum cultivation site in northwest Germany

During the last decades an increasing area of drained peatlands has been rewetted. Especially in Germany, rewetting is the principal treatment on cutover sites when peat extraction is finished. The objectives are bog restoration and the reduction of greenhouse gas (GHG) emissions. The first sites were rewetted in the 1980s. Thus, there is a good opportunity to study long-term effects of rewetting on greenhouse gas exchange, which has not been done so far on temperate cutover peatlands. Moreover, Sphagnum cultivating may become a new way to use cutover peatlands and agriculturally used peatlands as it permits the economical use of bogs under wet conditions. The climate impact of such measures has not been studied yet. We conducted a field study on the exchange of carbon dioxide, methane and nitrous oxide at three rewetted sites with a gradient from dry to wet conditions and at a Sphagnum cultivation site in NW Germany over the course of more than 2 years. Gas fluxes were measured using transparent and opaque closed chambers. The ecosystem respiration (CO2) and the net ecosystem exchange (CO2) were modelled at a high temporal resolution. Measured and modelled values fit very well together. Annually cumulated gas flux rates, net ecosystem carbon balances (NECB) and global warming potential (GWP) balances were determined. The annual net ecosystem exchange (CO2) varied strongly at the rewetted sites (from −201.7±126.8 to 29.7± 112.7 gCO2-Cm −2 a) due to differing weather conditions, water levels and vegetation. The Sphagnum cultivation site was a sink of CO2 (−118.8±48.1 and −78.6±39.8 gCO2Cm a). The annual CH4 balances ranged between 16.2±2.2 and 24.2±5.0 gCH4-Cm −2 a at two inundated sites, while one rewetted site with a comparatively low water level and the Sphagnum farming site show CH4 fluxes close to 0. The net N2O fluxes were low and not significantly different between the four sites. The annual NECB was between −185.5±126.9 and 49.9± 112.8 gCO2-Cm −2 a at the rewetted sites and −115.8±48.1 and −77±39.8 g CO2Cm a at the Sphagnum cultivating site. The annual GWP100 balances ranged from −280.5±465.2 to 644.5± 413.6 gCO2-eq.m −2 a at the rewetted sites. In contrast, the Sphagnum farming site had a cooling impact on the climate in both years (−356.8±176.5 and −234.9±145.9 gCO2Cm a). If the carbon exported through the harvest of the Sphagnum biomass and the additional CO2 emission from the decay of the organic material is considered, the NECB and GWP100 balances are near neutral. Peat mining sites are likely to become net carbon sinks and a peat accumulating (“growing”) peatland within 30 years of rewetting, but the GWP100 balance may still be positive. A recommended measure for rewetting is to achieve a water level of a few centimetres below ground. Sphagnum farming is a climate-friendly alternative to conventional commercial use of bogs. A year-round constant water level of a few centimetres below ground level should be maintained.


Introduction
Over many centuries, peatlands have been drained and used for peat extraction, agriculture and forestry worldwide and, in particular, in Germany (Couwenberg, 2011).In the last decades, peat has been extracted by industrial means on more than 30 000 ha in northwest Germany, and since the mid 1980s rewetting and restoration is obligatory on abandoned Published by Copernicus Publications on behalf of the European Geosciences Union.cutover peatlands as regulated by law (Höper, 2007;Höper et al., 2008).
In the long term, rewetting and restoration of the former cutover sites aims at establishing vegetation which is typical for growing, peat-accumulating bogs.Initially, the areas are flooded and natural succession starts.In the short term, flooding may lead to high methane emissions, depending on the water level and the time of year when flooding takes place (Waddington and Roulet, 1996;Le Mer and Roger, 2001;Houghton, 2004).Nevertheless, in the intermediate and long term, methane emissions are supposed to decrease and peat accumulation will lead to CO 2 uptake at the site (Augustin and Joosten, 2007).Little is known about the time dependency of these processes under temperate conditions and the total balance of the greenhouse gases.
About 3 decades ago, restoration programmes on cutover peatlands started in Germany (Höper and Blankenburg, 2000).Today, about 12 000 ha of former cutover sites are part of nature conservation areas and more than 23 500 ha will be rewetted in the future (Höper et al., 2008).Thus, there is a unique potential to study greenhouse gas emissions on sites which were rewetted about 30 years ago and to learn how to manage greenhouse gas emissions on future rewetted sites.
In comparison to other regions in Germany and many other European countries, bogs in Lower Saxony cover a comparatively high proportion of the land surface: almost 60 % of the peatland area or almost 6 % of the total terrestrial area (Höper, 2007).
Recently, Sphagnum cultivation has been proposed as a sustainable alternative to conventional peat extraction and as a climate-friendly use of abandoned cutover bogs (Gaudig et al., 2012).Sphagnum plants are grown under strongly controlled water conditions; the plant material is harvested in a 5-year cycle and used for horticultural purposes.Up to now, Sphagnum cultivation is still undergoing testing and no data on greenhouse gas exchange is available.
To date, studies on the gas exchange in peatlands have mainly been conducted in the boreal region (Alm et al., 1997;Nykänen et al., 1998;Joiner et al., 1999;Tuittila et al., 1999;Höper et al., 2008).In Scandinavian countries, measurements were mostly carried out during the summer months.For the remaining time period, the values are estimated or represent modelled fluxes (Byrne et al., 2004).This procedure can lead to false results (Roehm and Roulet, 2003), especially in temperate regions with mild winters.Most studies in restored peatlands are from recently rewetted sites.Investigations of the gas exchange and the global warming potential (GWP) balance of peatlands with a longer history of rewetting are needed because the gas exchange pattern may change with time (Joosten and Augustin, 2006).In the temperate zone, studies on the greenhouse gas (GHG) exchange in rewetted bogs were published by Nieveen et al. (1998), Drösler (2005), Bortoluzzi et al. (2006) and Beetz et al. (2013).
We present the greenhouse gas exchange of three sites on an abandoned cutover bog rewetted for restoration about 30 years ago and of one bog site rewetted for Sphagnum cultivation 5 years ago.We determined the exchange of CO 2 , CH 4 and N 2 O as well as the net ecosystem carbon balance (NECB) and the GWP balance.
We hypothesize that (a) the GWP balance of the rewetted former cutover area is about neutral when averaged across the three sites because, even 30 years after rewetting, methane emission is not as low and peat accumulation and the resulting uptake of CO 2 are not as high as at natural sites which are slight sinks of greenhouse gases; (b) the three sites of the rewetted former cutover area form a humidity gradient from a slightly lower to a slightly higher water table, which is underlined by its dominant vegetation and which corresponds to a gradient from slightly higher to lower greenhouse gas emissions; and (c) the GWP balance of the Sphagnum cultivation site is negative (cooling effect) because the water table is constantly kept some centimetres below the soil surface throughout the year, leading to low methane emission and optimal conditions for Sphagnum growth and carbon accumulation.
For the determination of gas flux rates, we used the closedchamber method similar to that used in Drösler (2005).To obtain annual balances, modelling and interpolation were carried out.Additional field measurements of driving parameters were conducted.

Site description
The research area, Nordhümmlinger Moore, is located in the northwest part of Lower Saxony in Germany (53 • N, 7.32 • E longitude, about 5 m above mean sea level).The 30year (1951-1980) mean annual temperature and annual precipitation amounts to 8.6 • C and 795 mm (Eggelsmann and Blankenburg, 1990).The warmest month is July (16.4 • C) and the coldest month is January (0.8 • C).Total precipitation is evenly distributed over the 12 months of the year.
The other research area is a peat mining area in the "Westermoor" (about 15 km northeast of Leegmoor).The measurement site ("Sphagnum cultivation") is a 60 m × 20 m test area, which was used agriculturally until 2000, from which peat was subsequently extracted, and which was rewetted in 2004 in order to cultivate peat mosses for harvesting (vegetation: S. papillosum Lindb., S. Cuspidatum Ehrh.ex Hoffm., S. palustre L., S. fallax H. Klinggr., E. angustifolium Honck., E. tetralix L., Juncus effusus L., B. pendula Roth, Drosera, fungi; peat thickness: 195 cm, consisting of 9 cm highly decomposed peat and 186 cm of weakly decomposed peat; Fibric Histosol or Norm-Hochmoor (HHn, AG Boden, 2005)).The water level is kept quite constant, just below ground level, all year round with the aid of a pump.To date, no harvesting has taken place.

Measurements of site factors and environmental controls
The soil identification (soil horizon, material, CaCO 3 content, pH) was conducted according to FAO (2006).The decomposition status was determined according to von Post.H1 and H2, H3 and H4, H5 and H6, H7 and H8 as well as H9 and H10 (von Post) correspond to z1, z2, z3, z4 as well as z5 (humification index, AG Boden, 2005), respectively.Aboveground biomass at "Sphagnum cultivation" was sampled (cut by hand) from the measurement plots down to the original peat and separated into green (green biomass) and brown (dead biomass) plant parts as well as into moss and vascular plants.We determined the dry matter content by drying the samples in an oven at a temperature of 105 • C for two days (until constant weight).Fresh and dry biomass was quantified using a laboratory balance.The dry material was analysed for total carbon and nitrogen with an elemental analyser (Elementar vario plus CNS analyser).
"Molinia", "Eriophorum" and "Sphagnum cultivation" were equipped with tubes inserted into the peat body, close to the collars.Water levels were manually measured with an electric contact gauge during each gas measurement campaign.In addition, at "Eriophorum" and "Sphagnum cultivation" the water levels were continuously (half-hourly) recorded from June 2010 until December 2011 with a Schlumberger MiniDiver.The missing time periods were filled by interpolating between the manual measurements.
Meteorological parameters, such as temperatures (air temperature, soil temperature at 2, 5 and 10 cm depth), photosynthetic active radiation (PAR), air pressure and precipitation were measured and saved half-hourly at the meteorological station near "Sphagnum cultivation".

Measurements and modelling of carbon dioxide exchange
Flux measurements were carried out every 4 weeks, starting September 2009 and ending in December 2011.A temperature-controlled portable closed-chamber technique was applied (see Drösler, 2005;Beetz et al., 2013).The chambers (0.78 m × 0.78 m; height: 0.5 m; equipped with a thermometer, a vent outlet with a rubber tube, a pair of turnable fans and a closed-cell rubber tube on the bottom to ensure airtightness) were connected via a tube with an electric pump and a portable CO 2 gas analyser (Licor LI-820; measurement of gas concentration every 5 s; one gas flux measurement procedure: 1-4 min).Thermal packs were used for cooling (temperature increase during measurement was kept below 1.5 Air pressure, air temperature, soil temperatures at 2, 5 and 10 cm depth (measured with inserting thermometers), and PAR were monitored during gas exchange measurements.Measurements started prior to sunrise and ended in the afternoon, in order to cover the entire range of PAR and temperatures.Per site and measurement day, about 12-18 measurements with opaque and 12-30 measurements with transparent chambers were carried out.
Flux rates were calculated according to Flessa et al. (1998), Drösler (2005) and Beetz et al. (2013), using the linear slope of gas concentration over time inside the chamber.To ensure the quality and representativeness of the slope, the following parameters were tested: (1) linearity of slope, (2) difference of the slope from 0 (slopes not different from 0 were set to 0), (3) variability of the slopes, and (4) constancy of the PAR (cv < 5 %).Negative fluxes denote uptake by the ecosystem; positive fluxes loss to the atmosphere.
NEE was calculated as the difference between gross photosynthetic production (GPP), which has a negative sign and R eco , with a positive sign.
For each measurement day, two models were fitted against the measured data using Microsoft Excel ® Solver.Firstly, the ecosystem respiration (R eco ) was modelled according to Drösler (2005), Elsgaard et al. (2012) and Beetz et al. (2013), using an exponential regression equation (Lloyd and Taylor, 1994) of CO 2 flux against the temperature with the best fit (air; soil at 2 or 5 cm depth).
Secondly, GPP was derived from transparent (NEE) and opaque (R eco ) chamber measurements and modelled using a saturation function (Michaelis and Menten, 1913) and PAR as the input variable (Drösler, 2005;Elsgaard et al., 2012;Beetz et al., 2013).Because of the reduced transmissibility of the chamber Plexiglas, measured PAR was reduced by 5 %.
For the period between 2 measurement days, the flux rates of R eco and NEE were calculated on a half-hourly basis, us-C.Beyer and H. Höper: GHG exchange of bogs ing the weather data and the model parameters of both campaigns.
In order to interpolate the flux rates in a half-hour time step, the model parameters were weighted according to the interval between the time step to be calculated and the measurement days using the following formula: where F i and t i are the flux rate and time at time step i to be modelled and t n and t n+1 are the time (day) of the campaigns n and n + 1. F n and F n+1 are the flux rates calculated with the model parameters of campaigns n and n + 1.
Finally, monthly and annual balances were calculated by accumulating of the half-hourly flux rates.

Measurements of methane and nitrous oxide exchange
Measurement campaigns of N 2 O and CH 4 exchange were held in intervals every 2 weeks, beginning in September 2009 and ending in December 2011.
The chambers were identical in construction to the opaque chambers used for R eco (CO 2 ) but not ventilated.A measuring procedure lasted 1 h; every 20 min a gas sample was transferred from the headspace of the chamber to evacuated glass bottles (60 mL).Gas samples were analysed in the laboratory using the gas chromatograph Perkin Elmer Auto System, with ECD (electron capture detector) and FID (flame ionization detector) detectors (Beetz et al., 2013).
During gas exchange measurement, air temperature and soil temperatures at 2, 5 and 10 cm depth (measured by inserting thermometers) were monitored.
Flux rates were calculated according to Drösler (2005) and Beetz et al. (2013), using the linear slope of gas concentration over time inside the chamber.The slopes of gas concentration were tested for difference from 0. Slopes not different from 0 were set to 0.
Hourly flux rates over the whole research period were obtained by linear interpolation between the measurement campaigns and used to calculate annual balances.

Net ecosystem carbon balance and global warming potential
To obtain a complete carbon balance of an area of peatland, all fluxes of carbon must be considered (Chapin III et al., 2006).In addition to CO 2 flux rates, CH 4 flux rates, dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), carbon monoxide (CO) and volatile organic C (VOC) are factored in the NECB.Values of DOC, DIC, CO and VOC were assumed to be negligible and were not considered.
A widely used technique to establish the climatic impact of the GHG exchange at each site, expressed as CO 2 equiv-alents, is the global warming potential (GWP) methodology (IPCC, 1996).In general, the global warming potential over a time span of 100 years (CO 2 : CH 4 : N 2 O = 1 : 21 : 310; IPCC, 1996) is taken (Drösler, 2005).Positive values represent efflux of CO 2 equivalents into the atmosphere.

Statistical analyses
Unless otherwise stated, Microsoft ® Excel was used.Average values are arithmetic means.
The error analysis of the CO 2 fluxes was conducted by calculating the standard error for each calibrated regression model.Analogously to the interpolation of the half-hourly gas fluxes, we interpolated the standard errors.The monthly and annual standard errors were calculated using the law of error propagation.
For the CH 4 and N 2 O fluxes, we calculated the standard error of the measurements of each measurement campaign and interpolated between the measurement campaigns in the same way as was done with the interpolation of the fluxes.The annual standard errors were calculated using the law of error propagation.
Significant linearity of the slopes was proved by a test of linearity according to Huber (1984).To test whether slopes are significantly different from 0, the t test was performed (Kreyszig, 1973;Neter et al., 1996).The variability of the slopes was determined by calculating the standard deviation of the residuals (s yx ).For the variability of the PAR we calculated the coefficient of variability (cv %).
Correlation and regression analyses were conducted with the coefficient of determination (quadrate of Pearson correlation coefficient = R 2 ) and tested for significance with the t test.
Significant differences between the annual gas exchange balances were tested with the permutation test "diffmean" (1000 permutations) using R script 0.97.237(version 2.15.2) (simba package).

Site factors
Peat at "Molinia" was highly decomposed (H10, von Post scale) and had a very low pH (Table 1).The uppermost horizons of "Eriophorum" and "Sphagnum" are built up of reduced bog peat and had a very low decomposition status.The peat of "Eriophorum" consisted mainly of herbs, while at "Sphagnum" we found Sphagnum peat.At "Sphagnum cultivation", the uppermost horizon (reduced bog peat, Sphagnum peat) also had a similar decomposition status but a higher pH.There was no CaCO 3 at the sites.
The vegetation at "Sphagnum cultivation" was grown on top of the original peat (consisting of highly decomposed peat) and built up a horizon of 15 cm of slightly decomposed Sphagnum material since the beginning of Sphagnum culti- vation.Analysis of aboveground biomass was conducted in May of the last measurement year at "Sphagnum cultivation".The dry mass of the vegetation consisted primarily of Sphagnum.The total carbon stock of the aboveground biomass was 715.8 ± 57.2 g m −2 .

Environmental controls
The annual mean of the water table was 16.1 and 10.8 cm below ground in 2010 and 2011, respectively (Fig. 1, Table 1).The summer mean (May-October) was 34 and 21 cm below ground in 2010 and 2011, respectively.At "Eriophorum" the water table was higher and more stable over the course of the year (mean of 2010 and 2011: 4.4 and 3.8 cm above ground, respectively; summer mean: 4.9 and 2.5 cm below ground, respectively)."Sphagnum" was located about 10 cm lower than "Eriophorum"; thus, the mean was 14.4 and 13.8 cm above ground in 2010 and 2011, respectively."Sphagnum cultivation" had a different water regime because the water table was regulated (annual mean in 2010 and 2011: 6.1 and 9.2 cm below ground, respectively; summer mean: 6.3 and 8.5 cm below ground, respectively).
According to the weather station near Westermoor, air temperatures were between −10 and 27 • C, soil temperature at a depth of 5 cm ranged from −0.2 to 22 • C and precipitation was highest on 7 September 2011, with 20 mm (Fig. 1).

Methane and nitrous oxide
At "Molinia" and "Sphagnum cultivation", the annual CH 4 balances were low and not significantly different between the two sites (Table 2).At "Eriophorum" flux rates were higher, and the highest methane emissions occurred at "Sphagnum".In 2011, the CH 4 emissions of "Eriophorum" and "Sphagnum" were not significantly different.The CH 4 emissions in 2010 and 2011 were not significantly different at any of the sites.
The annual N 2 O balances did not differ significantly between all sites in both years (Table 2).While "Sphagnum" was a weak sink for nitrous oxide in both years, the other sites were a sink in one and a source in the other year.Highest emissions occurred at the dry sites and lowest emissions or uptake at the wetter sites.The annual values did not differ significantly between the years, except for those at "Eriophorum".

Carbon dioxide
The regressions between measured and modelled flux rates for R eco and NEE of each measurement campaign at "Molinia" (Table 3) were significant in all cases (P < 0.1).
At the other three sites, in a few cases, mostly in winter, the difference between the lowest and the highest temperature was too small for modelling the R eco (Tables 4-6).At "Eriophorum", this was the case on 25 November 2009, 15 December 2010 and 14 December 2011; at "Sphagnum" it was the case on 31 March, 21 April, 18 August and 15 December 2010.In these cases the model parameter E 0 , which repre- sents the ecosystem sensitivity parameter (K), was set to 0 and the model parameter R ref , which is R eco at the reference temperature (283.15K), was replaced by the mean value of the measured values.This is a conservative way to obtain flux rates.The measurements on 9 February 2011 at "Sphagnum" had to be discarded because it was not possible to calibrate both models (R eco and NEE) with the available data.At "Sphagnum cultivation" the regressions between measured and modelled flux rates for R eco of the measurements on 29 September and 27 October 2009 were not significant; on 8 June 2011, gas fluxes at "Sphagnum cultivation" did not increase with increasing temperatures; and on 14 December 2010, there was almost no change in the soil temperature and the air temperature was below 0 • C for the whole day, which did not result in an appropriate relationship.In these cases it was possible to pool the results of two measurement campaigns together to achieve significant regressions because, in contrast to "Molinia", "Eriophorum" and "Sphagnum" the long-term controls at "Sphagnum cultivation" remained similar between the two pooled measurement campaigns.The measurement-campaign-specific regressions between measured and modelled flux rates for the NEE were significant in all cases (P < 0.1).
At each site the regression between all modelled and measured values for R eco and NEE were always significant (R 2 between 0.88 and 0.98, P < 0.0001) and almost followed the 1 : 1 line (Fig. 2).The standard errors for the R eco and the NEE at "Molinia" were 0.36 and 1.45 µmol CO 2 -C m −2 s −1 , respectively.At "Eriophorum", standard errors of 0.70 and 1.32 µmol CO 2 -C m −2 s −1 , respectively, were de-  termined.The standard errors at "Sphagnum" amounted to 0.39 and 0.83 µmol CO 2 -C m −2 s −1 , respectively, and to 0.41 and 0.53 µmol CO 2 -C m −2 s −1 at "Sphagnum cultivation", respectively.The gas fluxes at "Molinia" and "Eriophorum" were generally higher than at the other sites; consequently the standard errors were usually higher, at least for the NEE.
The annual development of R eco showed a strong seasonal pattern at all sites, with high values in summer and low values in winter (Fig. 3).On average over the 2 years, the highest monthly cumulated R eco occurred in July (192.2 ± 18.3,173.5±15.4,94.2±2.5 and 87.1±9.5 g CO 2 -C m −2 , at "Molinia", "Eriophorum", "Sphagnum" and "Sphagnum cultivation", respectively).The lowest monthly cumulated R eco was determined in December.
R eco and GPP showed characteristically seasonal patterns at all four sites (Fig. 3)."Molinia" and "Eriophorum" demonstrate a similar annual development for the most part of the year.However, at the beginning of the vegetation period, CO 2 fluxes started to increase much earlier at "Eriophorum" than at the other sites.The annual pattern of "Sphagnum" and "Sphagnum cultivation" differed strongly from the development of "Molinia" and "Eriophorum"."Sphagnum" and "Sphagnum cultivation" both revealed much lower R eco and GPP.By comparing the annual development of the R eco with the development of the temperature, it appeared that there is a lag in the development of the vegetation in spring at all sites with the exception of "Eriophorum".In late summer and autumn, when temperatures were still high, the R eco had already dropped.In July 2010, the GPP at "Sphagnum" went against the general trend and dropped to lower values.Subsequently, it increased again.Comparing the annual development of the GPP with the development of the PAR, a delay is visible, similar to the delay in the R eco (see above  In summer, the gross uptake of CO 2 through GPP outbalanced the release through R eco , while during the colder months the net fluxes were near 0 or net emissions occurred (Fig. 3).At "Eriophorum" and "Sphagnum" the highest cumulated monthly net uptake of CO 2 occurred on average in June (−49.5±20.4 and −47.8±16.2g CO 2 -C m −2 , respectively), whereas at "Molinia" the highest average uptake of CO 2 was detected in July (−74.8±2.4 g CO 2 -C m −2 ) and, at "Sphagnum cultivation", in August (−25.0±16.1 g CO 2 -C m −2 ).There was a gradient from highest net uptake at "Molinia" to lowest uptake at "Sphagnum cultivation" during the summer months of June until August.During the remaining part of the year, "Molinia" emitted CO 2 ; "Eriophorum" also emitted CO 2 but less than "Molinia"."Sphagnum" and "Sphagnum cultivation" sequestered CO 2 , but "Sphagnum cultivation" sequestered more CO 2 .
The net CO 2 sequestration (NEE) at "Sphagnum" and "Sphagnum cultivation" was about the same and not significantly different (Table 2).The uptake was higher in 2010 than in 2011."Molinia" and "Eriophorum" were net sinks in 2010 and net sources in 2011.At all sites, the annual NEE balance was significantly different between the 2 years due to a higher R eco in 2011 as compared to 2010.On average, a gradient from "Molinia" with a smaller uptake to "Sphagnum cultivation" with a higher uptake was apparent.
The standard errors of the annual CO 2 balances were high, compared to the annual balances, especially at "Molinia" in 2011.In addition, the difference between the 2 years was high, particularly at "Molinia" and "Eriophorum", where the standard errors of the annual balances were much higher than the mean values."Sphagnum cultivation" showed the most stable values.

Carbon and climate budget
The NECB of "Molinia" and "Sphagnum cultivation" were similar to the annual NEE, because the NECB was mainly determined by the NEE (Table 2).In spite of the higher CH 4 emissions at "Eriophorum" and "Sphagnum", there is still a gradient from "Molinia", with the lowest, to "Sphagnum cultivation", with the highest NECB.At "Eriophorum" and "Sphagnum", the GWP100 balance revealed a different picture than the NECB because methane exerted a greater climatic impact (Table 2)."Sphagnum" was a greenhouse gas source (GWP100 balance) in both years and on average the highest source of all sites."Eriophorum" and "Molinia" were sinks (GWP100 balance) in one year and sources in another year; "Sphagnum cultivation" was a sink in both years.
Changing the time perspective of the GWP assessment from 100 to 500 years leads to the conversion of "Eriophorum" and "Sphagnum" from a GHG source to a sink (Table 2)."Sphagnum cultivation" and "Molinia" become stronger sinks.Due to the short life time span of CH 4 , the climatic impact of this gas decreases from the 100 (GWP100) to the 500 year perspective (GWP500).

Statistical analysis of the relations between control factors and gas exchange
At the sites in the Leegmoor, the annual development of the methane fluxes showed no seasonal trends but, rather, a diffuse pattern (Fig. 4).At "Sphagnum cultivation" a relationship between CH 4 fluxes and temperature as well as between CH 4 fluxes and water table was evident (Fig. 4).Analogous to the rising temperatures in spring and falling temperatures in autumn, the methane emissions increased in spring and dropped in autumn.In 2010, the water table and emissions were higher than in 2011.The relationships between CH 4 fluxes and water levels (R 2 = 0.32) as well as between CH 4 fluxes and soil temperatures (R 2 = 0.59) were significant.Taking the measurements from all sites, it was obvious that at a water level of less than 20 cm below ground level the CH 4 fluxes were around 0 and, at a water level of above 20 cm below ground level, the CH 4 fluxes increased.The highest fluxes could be determined at a water level of around 0.
The correlation analyses revealed a highly significant correlation between CH 4 balance and mean water level (R 2 = 0.93).

Reliability of the research methods
Especially in the cold months it was not possible to establish a good correlation between temperature and R eco because the temperature span was too small or even inexistent during the day.In this case, results from two or three campaigns were pooled or R eco was set constant to the mean values of the measurements.The error in the annual balance due to this procedure is low because the CO 2 exchange is low at low temperatures.Nevertheless, winter conditions are taken into consideration as well as is possible by keeping up a schedule of one campaign every 4 weeks.
Temperatures and PAR were used for the modelling during each measurement campaign, while other controls, such as soil moisture or vegetation, were not considered in the short term (i.e.daily) but only in the long term (i.e.monthly), regardless of the fact that these may also change over the course of the day.Petrone et al. (2003) suppose that soil moisture may be the primary controlling factor of GPP.However, we found coefficients of determination (Pearson) between modelled and measured values of the NEE model well above R 2 = 0.5 (Tables 3-6).Factors having an effect in the long term are accounted for by the repeated measurement campaigns, which result in different model parameters as a consequence of the different field situations.The linear interpolation between measurement campaigns assumed that long-term controls also change linearly, which is certainly not the case.For this reason, we kept the time between two measurement campaigns as short as possible (Beetz et al., 2013).

Importance of site and control factors
At all sites the seasonal pattern of R eco and GPP basically followed the development of the temperature and the PAR, respectively (Figs. 1 and 3).It can be assumed that the deviations from these patterns were caused by the vegetation (Lafleur et al., 1997;Buchmann and Schulze, 1999;Tuittila et al., 1999;Wilson et al., 2007;Kivimäki et al., 2008).In spring, when temperatures and PAR are already raised, the vegetation is not yet fully developed, while in late summer and autumn, senescence occurs, and R eco as well as GPP are low although temperatures and PAR are still quite high.In both cases, autotrophic respiration is low and is not outbalanced by heterotrophic respiration, which is more closely related to temperature.In July, temperatures are highest and the vegetation is fully developed; thus, highest R eco occurs during July, which is consistent with Beetz et al. (2013).
The daily GPP of "Molinia" and "Eriophorum", on the one hand, and "Sphagnum" and "Sphagnum cultivation", on the other hand, were similar, due to similar types of vegetation.
In comparison to the other sites, the GPP at "Eriophorum" increased early in spring and decreased late in fall because Eriophorum has a high potential for photosynthesis early in the season and throughout most of the season (Tuittila et al., 1999).Another important driver of the gas exchange is the water level (Tuittila et al., 1999;Waddington and Warner, 2001;Lafleur et al., 2003;Glatzel et al., 2006;Wilson et al., 2007).The strong decrease of GPP in July 2010 at "Sphagnum" was caused by the decline of the water table in connection with warm and dry weather (Fig. 1), leading to a drying-out of the Sphagnum (Titus et al., 1983;Schipperges and Rydin, 1998).In contrast, the other sites were not affected by the dry period in July 2010 because "Molinia" and "Eriophorum" were not dominated by Sphagnum but by species such as Eriophorum, which are less vulnerable to fluctuations of the water table (Tuittila et al., 1999).The clearly visible effect of the dry pe- riod on GPP at "Sphagnum" proves the ability of our model to account for such influencing parameters.
For a meta-analysis we used the data of this paper, our own unpublished data from a bog near Bremerhaven (NW Germany) and published data of rewetted (mostly former peat cut sites) in the temperate zone (Nieveen et al., 1998;Drösler, 2005;Bortoluzzi et al., 2006;Beetz et al., 2013).CO 2 balances of our sites are in line with those of other bogs, but if only published values from German bogs are taken into consideration, it seems that our results fit the results of natural bogs better than those of rewetted bogs: Drösler (2005) and Beetz et al. (2013) found annual balances in the range of −157 to −8 g CO 2 -C m −2 a −1 .Our values ranged from −201.7±126.8 to 29.7 ± 112.7 g CO 2 -C m −2 a −1 (Table 2).In contrast, rewetted bogs in Germany reveal annual balances in the range of −148 to 192 g CO 2 -C m −2 a −1 (Drösler, 2005;Beetz et al., 2013).Water level and vegetation partly explains the variation in the meta-analysis (Fig. 5).However, small differences in mean water level between the sites lead to vanishingly small differences in gas fluxes.Thus, "Sphagnum" and "Sphagnum cultivation" were not significantly different, while "Eriophorum" revealed a slightly lower net CO 2 uptake due to a different type of vegetation, and "Molinia" showed a much lower net accumulation because of different vegetation and a much lower water level in summer.
A rewetted Sphagnum-dominated bog in Finland with a similar water level to "Sphagnum" shows CO 2 balances in the same range (Kivimäki et al., 2008).However, other rewetted bogs in the boreal zone with comparable water levels and vegetation composition reveal much lower annual exchange rates (with −800 to 1644 kg CO 2 -C ha −1 a −1 ) compared to rewetted bogs in the temperate zone, which range between -2960 and 1920 kg CO 2 -C ha −1 a −1 (Tuittila et al., 1999;Yli-Petays et al., 2007;Kivimäki et al., 2008).Also, natural bogs in the boreal zone show lower uptake rates (Lafleur et al., 2003).A comparison of the annual balances of our sites and other sites in the temperate zone with rewetted bogs in the boreal zone indicate that, on average, a net accumulation in both zones takes place, but the average annual uptake rate is much higher in the temperate zone.The high variation of the annual CH 4 emissions between the sites (Table 2) is in line with the results of other studies in rewetted (mostly former peat cut sites) bogs in the temperate zone (Drösler, 2005;Bortoluzzi et al., 2006;Beetz et al., 2013;our own unpublished data) and can be mainly explained by the variation in the mean water level (Fig. 5).The relationship between CH 4 balances and water level was also described by Drösler (2005) and Couwenberg et al. (2011).Therefore, "Eriophorum" and "Sphagnum" revealed significantly higher annual methane emissions than "Molinia" and "Sphagnum cultivation".Our meta-analysis showed a threshold value for the mean annual water level of about 10 cm below ground level.At the rewetted sites, the methane emissions are strongly determined by the water level (R 2 = 0.78, P < 0.001).The Eriophorum (E.vaginatum and E. angustifolium)-dominated rewetted sites show slightly higher CH 4 balances compared to the Ericaceae-, Moliniaor Juncus-dominated sites and Sphagnum-dominated sites with a similar average water level.E. vaginatum and E. angustifolium are plants with aerenchymatous leaves and contribute to the methane emissions because methane is transported through the aerenchyma (Joabbson et al., 1999;Joabbson and Christensen, 2001;Drösler, 2005;Couwenberg et al., 2011).One Ericaceae-dominated site showed unusually high CH 4 emissions (see Fig. 5).This site was a drained Calluna vulgaris heathland, which was rewetted about 10 years ago (Drösler, 2005).The vegetation still consists mainly of C. vulgaris although the water level is too high to support such a type of vegetation in the long term.C. vulgaris is found at places with a lower water level (Poschlod, 1988;Drachenfels, 2011).A comparison between rewetted and natural bogs with the data from our study and literature data (Drösler, 2005, Bortoluzzi et al., 2006;Beetz et al., 2013) reveals that, in contrast to the findings of Augustin and Joosten (2007), CH 4 emissions from rewetted bogs are not higher than from natural bogs.Also, rewetted bogs in the boreal zone show increasing annual CH 4 emissions with rising water levels (Yli-Petays et al., 2007).Compared to the temperate zone the emissions are much higher: Yli-Petays et al. (2007) found CH 4 emissions between 113 and 378 kg CH 4 −C ha −1 a −1 at mean water levels of between 0 and 7.5 cm below ground.Methane emissions of bogs are low compared to fens, rice fields and freshwater ecosystems (Moore and Knowles, 1989;Le Mer and Roger 2001;Höper et al., 2008).
It can be summarized that the GHG exchange patterns depend mainly on temperature, PAR, vegetation and water level, while the state of the bog (rewetted or natural) seems to be of minor importance.
The rewetted, former peat mining site Leegmoor consists of a small-scale mosaic of areas with different water levels; hence the spatial pattern of CH 4 emissions is very heterogeneous.In order to obtain the methane emissions of the whole rewetted area, a mapping of mean water tables and/or vegetation type is needed.The relationship between water table and methane emission can than be used to estimate the overall emission.
By contrast, the Sphagnum farming site is very homogenous on the spatial scale, and, due to the active water management, the water table is well regulated during the whole year.Methane emissions from this site will generally be low if the water level is kept below the land surface.
Annual GPP and R eco are both high and of the same magnitude; consequently, the annual NEE, which represents the small difference between both gross values, is generally close to 0. A change in weather conditions and water table can easily convert a sink into a source and vice versa.This high interannual fluctuation of the CO 2 balances in organic soils, with years releasing CO 2 and others sequestering CO 2 , has been observed by many authors (Tuittila et al., 1999;Lloyd, 2001;Arneth et al., 2002;Roulet et al., 2007;Yli-Petays et al., 2007;Beetz et al., 2013).Natural bogs can be sources in the short term, but, on average over many years, they are sinks.Therefore, measurements should be conducted over several years.
Rewetted and natural bogs generally have low fluxes of nitrous oxide due to anoxic conditions (Byrne et al., 2004;Drösler, 2005;Beetz et al., 2013); this is confirmed in our study (Table 2).

Evaluation of the effectiveness of the methods for bog revitalization
All sites in Leegmoor revealed negative NECB on average over the 2 measurement years (Table 2).Thus, the aim of carbon accumulation is achieved.On the other hand, "Eriophorum" and "Sphagnum" have, on average over the 2 years, a small positive GWP100 balance, which means that they have a warming effect on the climate.This can be attributed to the methane emissions.
A meta-analysis of the data of our research sites and other data of rewetted (mostly former peat cut sites) bogs in the temperate zone (Drösler, 2005;Bortoluzzi et al., 2006;Beetz et al., 2013;our own unpublished data) shows that inundated bogs are generally GHG sources in terms of the GWP100 balance, due to the high methane emissions (Fig. 5).At a mean water level between 0 and 20 cm below ground level, most rewetted bogs seem to be GHG sinks in terms of the GWP100 balance.In contrast, rewetted bogs in Finland with water levels between 7.5 and 0 cm below ground have high GWP100 balances of up to 10 600 kg CO 2 −eq.ha −1 a −1 due to high CH 4 emissions (Yli-Petays et al., 2007).At a lower mean water level, an increase in GHG emissions is expected due to higher CO 2 emissions.A comparison between natural and rewetted bogs with the data of our study sites and published data (Drösler, 2005;Bortoluzzi et al., 2006;Beetz et al., 2013) shows that natural bogs do not have lower emissions or higher accumulation rates of GHG than rewetted bogs.
In contrast, agriculturally used drained bogs are huge net GHG sources.Sites used for cropland and grassland in the temperate zone in Europe show NECBs between 434 and 1150 g CO 2 -C m −2 a −1 (Höper, 2007;Elsgaard et al., 2012;Beetz et al., 2013).In addition, high nitrous oxide releases at cropland sites might occur.Petersen et al. (2012) observed emissions of 6100 mg N 2 O-N m −2 a −1 at a Danish bog used for cropland.
In conclusion, rewetted former peat mining areas may become net carbon sinks and a growing are of peatland within 30 years of rewetting.In single years, a net loss of carbon may occur, but, in the long term, a small accumulation of carbon takes place or, at least, the carbon balance is 0. Nearnatural conditions are therefore required.A high water table (above ground level) leads to the release of high amounts of methane, resulting in a net warming effect (positive GWP100 balance) instead of a cooling effect.Rewetted bog areas such as Leegmoor always have a heterogeneous spatial distribution of areas differing in water level and vegetation.A high spatial variability in GHG fluxes was also observed by Fest et al. (2009) and Merbold et al. (2013).Moreover, the water level shows interannual variation.On one hand, there are always zones with a high water level where a positive GWP100 balance due to high CH 4 emissions has to be expected.On the other hand, there are always places with low water levels and positive GWP100 balances due to carbon dioxide emissions.Therefore, the water level should be kept a few centimetres below ground in most of the area and inundation should be avoided if possible.

Evaluation of the effectiveness of the methods for
Sphagnum farming "Sphagnum cultivation" shows the highest accumulation of carbon, compared to the other sites (Table 2)."Sphagnum cultivation" and "Sphagnum" have a similar NEE, but the net carbon accumulation (NECB) at "Sphagnum cultivation" is higher due to lower methane emissions.The main difference in the environmental controls between "Sphagnum cultivation" and "Sphagnum" are the water level dynamics (Fig. 1).
The water level at "Sphagnum cultivation" is kept quite constant at a level which is unfavourable for high CH 4 emissions.
At "Sphagnum cultivation" the carbon stock of the biomass grown on the old peat layers revealed that, on average, −102.3±8.2 g C m −2 a −1 had been accumulated.This is similar to the annual NECB.The good fit of the values confirmed the results of the gas flux measurements and modelling.The roots of vascular plants in the peat layer are not included in the analysed biomass.However, the proportion of vascular plants is low (about 15 % of all of the biomass).
"Sphagnum cultivation" was a GHG sink in terms of the GWP100 balance in both examination years, and, with −295.8±73.8g CO 2 -eq.m −2 a −1 , it was the strongest sink of GHG on average over the 2 years compared to Leegmoor (Table 2, Fig. 5).The annual average water level is about 6-9 cm below ground, which is quite unfavourable for peat mineralization but obviously deep enough for the oxidation of most of the methane produced.
At "Sphagnum cultivation" no biomass had been harvested up to now.If the carbon which will be exported by the harvest and its mineralization to carbon dioxide due to the use of the biomass for horticultural purposes are taken into account, the NECB and GWP100 balance will be near neutral because almost all of the biomass built up during the years of Sphagnum farming is removed.Nevertheless, commercial Sphagnum farming would be by far the most convenient use of bogs.Conventional commercial uses of bogs, such as use as cropland, grassland or for peat mining, cause high emissions of GHG.
The results indicate that keeping the water table constant year-round just a few centimetres below ground level leads to a neutral GWP balance.Providing this, a conversion from conventional farming with deep drainage to Sphagnum farming would lead to a great reduction in climate impact.

Can the results be generalized?
"Molinia" was a net carbon sink in one year and a source in another year (Table 2: −74.1±91.4 and 11.0 ± 138.9 g CO 2 -C m −2 a −1 , respectively).Beetz et al. (2013) and Yli-Petays et al. (2007) observed the same in similar rewetted bogs.A literature survey of the data of our research sites and other data (Drösler, 2005;Bortoluzzi, et al., 2006;Beetz et al., 2013;our own unpublished data) shows that the NECB and the GWP100 balances of our sites (Table 2) are similar to the balances of other rewetted bogs in the temperate zone (Fig. 5).Most rewetted bogs in the boreal zone show lower NECB and GWP100 balances (Tuittila et al., 1999;Yli-Petays et al., 2007;Kivimäki et al., 2008).A time-dependent effect of gas flux dynamics is not visible in the temperate and in the boreal zone.The gas exchanges of natural and rewetted sites are similar.

Conclusions
Our study showed that former cutover bogs which had been rewetted 30 years ago were net carbon sinks and, therefore, peat accumulating sites, at present.The high water level led to a slowdown of the oxidation processes and to the accumulation of plant material as the base for peat formation.Nevertheless, the GWP100 balance is slightly positive due to methane emissions under inundated conditions, and the bog therefore has a small warming impact on the climate.Thus, in order to promote carbon accumulation, the water level should be high, whereas, in order to achieve a climate cooling effect, inundation should be avoided.However, in practice an exactly regulated water level is impossible or too expensive to achieve.In addition, the ground is never flat.Rewetted bogs always show a mosaic of comparatively dry and wet places.Thus, it can be concluded that, firstly, the rewetted area should be levelled in order to achieve an appropriate water level as far as possible, and, secondly, sites for the measurement of GHG fluxes should be placed in such a way as to represent the spatial heterogeneity.Due to the interannual variation of weather conditions, rewetted and natural bogs may be net carbon sources in single years; however, in the long term, they function as sinks.Measurements of GHG fluxes should be conducted over more than 1 year to account for temporal variability.
Temperature, PAR, water level and type of vegetation were identified as the main driving forces of GHG exchange in rewetted bogs.Thus, modelling approaches should consider these drivers.A measurement-campaign-based modelling procedure which considers driving forces in the short term (e.g.temperature, PAR) as well as in the long term (e.g.water level, phenology) seems to be an appropriate method, leading to flux rates with a high temporal resolution.
This study shows that Sphagnum farming is a climatefriendly way to use former peat extraction sites.Nevertheless, bogs used as grassland are also suitable for Sphagnum farming if the nutrient-rich topsoil is removed and the surface is levelled, as long as nutrient-poor water is available for rewetting.These sites represent a high potential in area and in the reduction of greenhouse gas emissions in northwest Germany.Sphagnum farming results in a win-win situation, because the production has a near-neutral climate impact, provides an alternative renewable material for peat, which is a nonrenewable resource, and results in an economic use of land.

Figure 1 .
Figure 1.(a) Water level of the examination sites; (b) yearly development of temperatures and precipitation at the weather station near "Sphagnum cultivation", September 2009 until December 2011.

Figure 2 .
Figure 2. Modelled versus measured fluxes of NEE at the measurement sites.Regression equations and coefficients of determination (Pearson) (R 2 ).

Figure 3 .
Figure 3. Yearly development of daily R eco , GPP and NEE of the measurement sites (left axis).Soil temperature and PAR of the measurement sites (right axis of the upper and middle diagram, respectively).

Figure 4 .
Figure 4. Yearly development of CH 4 flux at "Eriophorum", "Sphagnum" and "Sphagnum cultivation" (left axis).Mean of the three collars; error bars are standard errors.Yearly development of water level and soil temperature at a depth of 5 cm (right axis).

Figure 5 .
Figure 5. Yearly gas exchange balances versus mean water level of rewetted bogs in temperate zone: categories according to dominant vegetation (Ericaceae-, Molinia-, or Juncus-dominated; Eriophorum-dominated; Sphagnum-dominated).(a) Yearly NEE balance versus mean water level; (b) yearly CH 4 balance versus mean water level; (c) yearly GWP100 balance versus mean water level.Only data from rewetted bogs in the temperate zone are included; the data used come from this paper, own unpublished data, Nieveen et al. (1998), Drösler (2005), Bortoluzzi et al. (2006) and Beetz et al. (2013).Please note: in general, values of flux rates and water level are given in the text or in tables of the corresponding publications.If values of water level are not given, but only presented in the form of a figure, then mean water level was estimated from the figures in the corresponding publications.

Table 1 .
Soil properties, mean water level and vegetation at the examination sites.
a According to FAO (2006); b von Post scale; c below ground.

Table 2 .
Annual and average balances for R eco , NEE, CH 4 -C, N 2 O-N exchange, NECBs (net ecosystem carbon balances) and GWP (global warming potential for the time spans of 100 and 500 years) balances; m: mean; SE: standard error.Letters indicate that balances are not significantly different.

Table 3 .
Parameters for the R eco and NEE models of "Molinia"; E 0 : activation energy like parameter (K); R ref : respiration at the reference temperature (µmol CO 2 -C m −2 s −1 ); temp: best fit temperature for R eco model.GP max : maximum rate of carbon fixation at PAR infinite (µmol CO 2 -C m −2 s −1 ); alpha: light use efficiency (µmol CO 2 -C m −2 s −1 /µmol m −2 s −1 ); R 2 : coefficient of determination (Pearson) between modelled and measured values.SE: standard error of the model (µmol CO 2 -C m −2 s −1 ); n: number of samples.Max. and min.values are printed in bold.Eventually measurement campaigns were pooled together.

Table 4 .
Parameters for the R eco and NEE models of "Eriophorum"; E 0 : activation energy such as parameter [K]; R ref : respiration at the reference temperature [µmol CO 2 -C m −2 s −1 ]; temp: best fit temperature for R eco model.GP max : maximum rate of carbon fixation at PAR infinite [µmol CO 2 -C m −2 s −1 ]; alpha: light use efficiency [µmol CO 2 -C m −2 s −1 /µmol CO 2 -C m −2 s −1 ]; R 2 : coefficient of determination (Pearson) between modelled and measured values.SE: standard error of the model [µmol CO 2 -C m −2 s −1 ]; n: number of samples.Max. and min.values are printed in bold.Eventually measurement campaigns were pooled together.

Table 5 .
Parameters for the R eco and NEE models of "Sphagnum"; E 0 : activation energy such as parameter [K]; R ref : respiration at the reference temperature [µmol CO 2 -C m −2 s −1 ]; temp: best fit temperature for R eco model.GP max : maximum rate of carbon fixation at PAR infinite [µmol CO 2 -C m −2 s −1 ]; alpha: light use efficiency [µmol CO 2 -C m −2 s −1 /µmol CO 2 -C m −2 s −1 ]; R 2 : coefficient of determination (Pearson) between modelled and measured values.SE: standard error of the model [µmol CO 2 -C m −2 s −1 ]; n: number of samples.Max. and min.values are printed in bold.Eventually measurement campaigns were pooled together.

Table 6 .
Parameters for the R eco and NEE models of "Sphagnum cultivation"; E 0 : activation energy such as parameter [K]; R ref : respiration at the reference temperature [µmol CO 2 -C m −2 s −1 ]; temp: best fit temperature for R eco model.GP max : maximum rate of carbon fixation at PAR infinite [µmol CO 2 -C m −2 s −1 ]; alpha: light use efficiency [µmol CO 2 -C m −2 s −1 /µmol CO 2 -C m −2 s −1 ]; R 2 : coefficient of determination (Pearson) between modelled and measured values.SE: standard error of the model [µmol CO 2 -C m −2 s −1 ]; n: number of samples.Max. and min.values are printed in bold.Eventually measurement campaigns were pooled together.