www.biogeosciences.net/5/925/2008/ © Author(s) 2008. This work is distributed under the Creative Commons Attribution 3.0 License.

Abstract. In Central Europe, most bogs have a history of drainage and many of them are currently being restored. Success of restoration as well as greenhouse gas exchange of these bogs is influenced by environmental stress factors as drought and atmospheric nitrogen deposition. We determined the methane and nitrous oxide exchange of sites in the strongly decomposed center and less decomposed edge of the Pietzmoor bog in NW Germany in 2004. Also, we examined the methane and nitrous oxide exchange of mesocosms from the center and edge before, during, and following a drainage experiment as well as carbon dioxide release from disturbed unfertilized and nitrogen fertilized surface peat. In the field, methane fluxes ranged from 0 to 3.8 mg m−2 h−1 and were highest from hollows. Field nitrous oxide fluxes ranged from 0 to 574 μg m−2 h−1 and were elevated at the edge. A large Eriophorum vaginatum tussock showed decreasing nitrous oxide release as the season progressed. Drainage of mesocosms decreased methane release to 0, even during rewetting. There was a tendency for a decrease of nitrous oxide release during drainage and for an increase in nitrous oxide release during rewetting. Nitrogen fertilization did not increase decomposition of surface peat. Our examinations suggest a competition between vascular vegetation and denitrifiers for excess nitrogen. We also provide evidence that the von Post humification index can be used to explain nitrous oxide release from bogs, if the role of vascular vegetation is also considered. An assessment of the greenhouse gas release from nitrogen saturated restoring bogs needs to take into account elevated release from fresh Sphagnum peat as well as from sedges growing on decomposed peat. Given the high atmospheric nitrogen deposition, restoration will not be able to achieve an oligotrophic ecosystem in the short term.


Introduction
Due to the high amount of carbon stored in the peatlands of the world and the sensitivity of biogeochemical processes in these ecosystems to climate change, research on matter cycling in peatlands has received considerable interest.Especially the release of greenhouse gases (GHG) as carbon dioxide (CO 2 ), methane (CH 4 ), and nitrous oxide (N 2 O) from peatlands has been the focus of biogeochemical research due to its potential contribution to feedbacks to global warming.Despite large areas of (often degraded) peat bodies in temperate regions, research on peat bogs is mostly from natural boreal sites and focuses on the role of the water table (Roulet et al., 1992;Nykänen et al., 1998;Blodau, 2002).
In temperate Germany, widespread drainage of bogs resulted in a serious decline of peatland area.Today, in NW Germany (Lower Saxony) merely 5% of formerly 2348 km 2 bog area remain undisturbed or in a close to natural state (Schmatzler, 1990).Therefore, protection of the remaining intact peat bogs is accompanied by restoration efforts in moderately degraded bogs.The most important environmental constraints on the successful restoration of these bogs are i) a low water table, a result of previous drainage and climate change, ii) atmospheric N deposition, and iii) strong decomposition of degraded peat.
The importance of water table on GHG release from peat has been discussed extensively (Blodau, 2002).CO 2 evolution follows an optimum function, with highest rates at an intermediate water table (Glatzel et al., 2006).Magnitude and important parameters of CH 4 emission from wetlands are well known (Le Mer and Roger, 2001).Drainage decreases CH 4 release and rewetting does not necessarily lead to an immediate rise in CH 4 release (Tuittila et al., 2000).Jungkunst and Fiedler (2007) stress the role of water table on GHG release and emphasize the climate control on the nature of the relation between water table and GHG release.
Published by Copernicus Publications on behalf of the European Geosciences Union.
S. Glatzel et al.: Greenhouse gases German bog Nutrients that may limit decomposition include nitrogen (N) and phosporus (P) (Güsewell and Freeman, 2003).In Lower Saxony, even "undisturbed" bogs are subject to elevated N deposition of up to 70 kg ha −1 yr −1 (Gauger et al., 2002).At these high N deposition rates, the capacity of Sphagna to take up N is exceeded (Lamers et al., 2000), N concentration in pore water accumulates and plants with high N demand as Molinia caerulea become more competitive (Limpens et al., 2003).An increasing proportion of easily decomposable litter and N enriched Sphagnum tissue enhance decomposition and N mineralization (Lamers et al., 2000;Aerts et al., 1992), facilitating N 2 O and CO 2 release.Generally, in bogs, N 2 O release is most common in disturbed locations influenced by elevated N content (Regina et al., 1996).N 2 O production requires the availability of nitrogen and is highest at high soil moisture, but not inundation (Granli and Bøckmann, 1994).
Decomposition status of peat controls its potential for further decomposition.Examinations by Glatzel et al. (2004) demonstrated a decreasing potential for aerobic and anaerobic CO 2 and CH 4 production with a rising von Post decomposition index.In the Pietzmoor Glatzel et al. (2006) explained increased CO 2 release from Sphagnum hollow peat compared to hummock peat by lower decomposition rates of hollow peat.Alm et al. (1999) remarked that increased NO 3 availability may be due to high decomposition, increasing rates of N 2 O emission from drained peatlands.
In this contribution we intend to add understanding on the influence of these controls on the GHG release of a restoring temperate bog.Previous investigations (Glatzel et al., 2006) have shown the effect of drought on decomposition rates.Specifically, we investigate the influence of a drawdown in water table and peat properties on methane and nitrous oxide release in a restoring peat bog and the influence of nitrogen on decomposition of surface peat.We hypothesize that i) drought decreases the CH 4 and N 2 O release in the bog and rewetting temporarily increases CH 4 and N 2 O release, ii) decomposition of peat controls CH 4 and N 2 O release, and iii) atmospheric nitrogen deposition accelerates decomposition of surface peat.

Research site
The study site was the Pietzmoor (Lower Saxony; NW, Germany; 53 • 06 N; 9 • 50 E).It is part of the nature reserve Lüneburger Heide.The bog is located on the eastern edge of the closed occurrence of rainfed bogs in NW Germany.Mean annual precipitation is 790 mm; mean annual temperature is 8 • C. The examination period was March to September 2004.Atmospheric N deposition is ca.22 kg ha −1 yr −1 (Fottner et al., 2004).Today, the Pietzmoor is moderately degraded.Manual peat extraction at the edges of the Pietzmoor was conducted between the 16th century and 1960.Deep drainage ditches, constructed in the 19th century further degraded the bog, resulting in increased growth of birch (Betula sp.) and pine (Pinus sp.).Since 1970, when restoration efforts began, drainage ditches have been closed and trees cut.This resulted in formation of a recent superficial acrotelm with Sphagnum spp.growing in many hollows.Hummocks are still dominated by Empetrum nigrum, Calluna vulgaris, and Eriophorum vaginatum.

Field CH 4 and N 2 O flux determination
Between March and August 2004, CH 4 and N 2 O fluxes were determined 14 times employing a closed chamber method (Hutchinson and Livingston, 1993) at 10 locations within the Pietzmoor bog.Of the 10 previously installed collars (covering 0.068 m −2 ), five collars were installed in the center and at the edge of the bog.Six collars covered hummocks and four collars covered hollows.Hummocks were 20 cm elevated compared to hollows.They dominate the site resulting in a high coverage of Calluna vulgaris, which is typical for dry degraded peatland sites (Rathert, 2004).Due to the protection status of the site it was not possible to construct a system of boardwalks.However, CH 4 fluxes are generally low and we did not find stepwise nonlinear concentration increases in our measurements that would have been a sign for ebullition events (e.g.Chanton and Whiting 1995).Hence, we assume no risk of severe disturbances in our measurement setup.
Among the hollow collars, two were vegetated by Sphagnum fallax, one hosted a small Eriophorum vaginatum tussock and one contained no living vegetation.Among the hummock collars, three were vegetated by Calluna vulgaris, one contained a big and one a small Eriophorum vaginatum individuum, and one was inhabited by lichens.These collars covered the range of microsites in the bog previously determined by Rathert (2004).
For gas flux determination, gas samples from the closed chamber were sampled by syringe five times in 5 minute intervals and transported to the laboratory in Göttingen.On the evening of the day of sampling, the 60 mL syringes were attached to an autosampler coupled to a Shimadzu GC-14B gas chromatograph and a set of four different calibration gas cocktails (described by Loftfield et al., 1997).Precision of analysis was 0.4% for CH 4 and 1.0% for N 2 O.As no saturation effects were found, fluxes were calculated from the linear slope of the concentration change over time (Lessard et al., 1994) taking into account the headspace temperature and the coefficient of determination for each regression.

CH 4 and N 2 O release from mesocosms
This experiment was set up to test the first hypothesis.Twelve undisturbed peat cores (diameter 15 cm) were sampled by cutting the peat at the outside of tube and simultaneously pushing the tube above the cut peat until average 23 cm of peat were inside the tube.All cores were taken from hollows, six in the center and six from the edge.The peat cores were transferred into 30 cm high mesocosms that enabled sampling of percolating water and gas concentrations from a 7 cm headspace.Peat cores were watered in three day intervals with artificial Schneverdingen rain (diluted ammonium nitrate solution set to a pH of 4.5, equivalent to an amount of 790 mm yr −1 and 20 kg dry and wet N deposition ha −1 yr −1 ).As suggested by Blodau et al. (2004), a two month equilibration phase preceded the experiment.During the equilibration phase, the water table was set to 7 cm below ground, which is a compromise between flooded conditions that were found in the hollows and much drier conditions in the hummocks.The cores were stored at 20 • C close to windows, allowing a natural night and day regime.Vegetation (Sphagna and small herbs, no large plants) continued to grow during the experiment.
The experiment consisted of three phases.The predrainage phase preceded the drainage phase.During this phase, the six manipulated cores were subjected to free drainage (restricted to 100 mL d −1 ) without applying low pressure.At the control cores water table remained close to the peat surface.During the second phase (drainage phase), the manipulated cores were subjected to free drainage.The third phase (post-drainage phase) began by closing the drainage at experimental cores and the daily addition of 40 mL artificial Schneverdingen rain until the water table was back to 7 cm below ground.The pre-drainage phase lasted 5 to 8 days, the drainage phase until the elimination of standing water lasted 5 to 6 days and the regeneration of high water table (post drainage phase) took 12 to 14 days.
During the experiment, we determined gas fluxes from all cores as described above (except for a 30 s sampling interval due to the small headspace) daily.Following the experiment, carbon (C) and N concentration of peat from all cores was determined.This was done by drying peat at 45 • C from all horizons, milling it to 0.25 mm and analysis by combustion at 900 • C in a LECO CN-Analyzer (LECO, St. Joseph, MI, USA).The C and N concentration of all horizons were averaged to 0-15 cm depth.We also estimated the von Post humification index at all cores.These examinations on peat properties enabled us to test the second hypothesis.

CO 2 evolution from incubated disturbed samples
In order to test the third hypothesis, we sampled peat from 0-10 cm depth from Calluna hummocks and Sphagnum hollows in the Pietzmoor.Approximately 20 g of peat were set to 75% water content, which yields intermediate rates of CO 2 evolution (Glatzel et al., 2006) and placed in 400 mL jars in triplicate.All samples were additionally moistened by one mL of liquid.The fertilized samples received 0.036 M ammonium nitrate solution (equivalent to 50 kg N ha −1 ), and the unfertilized control samples received plain water.The incu- bation experiment was conducted using the method by Isermeyer (1952) following the experimental design described Glatzel et al. (2006).Briefly, evolved CO 2 was absorbed by 20 mL of 0.1 M NaOH adsorption inside the jars.Sampling of NaOH placed in small containers) following 1, 3, 6, 11, 17, 28 and 42 days of incubation and titration with 0.1 M HCl allows the calculation of CO 2 evolved since the preceding sampling date.

Ancillary measurements and statistical procedures
We measured air temperature and precipitation at a weather station located 2 km from the field site and installed an air temperature logger 20 cm above the surface of the bog.We determined water table by previously installed wells at the bog center and edge 14 times between March and August 2004 in hollows at the center and the edge of the Pietzmoor.All data sets were tested for normal distribution using the Kolmogorov-Smirnov test.Data on N 2 O release and day of year (Fig. 3) was normal distributed, so Pearson's correlation coefficient was calculated.The other data was generally not normally distributed, and n was generally small, so correlation analyses were carried out using Spearman's rho test and differences between data subsets were analyzed using the Wilcoxon test employing the Statistica 6.1 software package (Stat Soft, 2004).The outcome of these tests was the basis for rejection or acceptance of the hypotheses.

Weather and water table
The field season was warmer and wetter than the long term mean (1989 to 2004).Between March and August 2004, we recorded 427 mm precipitation as opposed to a long term mean of 381 mm.Mean temperature during the field season was 14.  start of the field season, water table was close to the surface (Fig. 1).Following a rather dry spring, frequent precipitation led to a rise in water table until early July.In July and August, water table dropped to 25 cm, but rose again in late August.In the center of the bog, water table responded more quickly than at the edge.

Field CH 4 and N 2 O fluxes
Field CH 4 fluxes ranged from 0 to 7.8 mg CH 4 m −2 h −1 and averaged 1.2 mg CH 4 m −2 h −1 .Spatial variability of CH 4 fluxes was high, so we were not able to detect significant differences between the mean CH 4 flux from hummocks and hollows and between the collars in the center and at the edge of the bog (Fig. 2), although there was a tendency for elevated CH 4 release in hollows and at the center of the bog.As the water table at the center was not lower than at the edge, the absence of a significant difference between CH 4 release at the two sites is not surprising.
Although N 2 O fluxes in the field were generally low, and often 0 at some collars, we detected a N 2 O release of up to 574 µg m −2 h −1 .We found no N 2 O uptake.There was no difference in N 2 O release between hummocks and hollows, but at the edge, nitrous oxide release was higher than at the center (Fig. 1) despite the lack of a difference in water table.
During the course of the season, CH 4 fluxes rose from 0.5 mg m −2 h −1 to 2 mg m −2 h −1 (at some hummocks) and 4 to 8 g m −2 h −1 (at some hollows).This trend could not be noticed at all collars.There was no seasonal trend of N 2 O fluxes, except for the collar vegetated by a large Eriophorum vaginatum tussock.There, N 2 O fluxes decreased linearly with the course of the season (Fig. 3).

Methane
Methane release from the cores was higher than from field sites, averaging 8.2 mg CH 4 m −2 h −1 .During the pre-drainage phase, there was no difference in CH 4 flux between the control cores and the manipulated cores.
During this phase, methane fluxes were between 0.1 and 84.5 mg m −2 h −1 and averaged 7.6±9.1 to 8.7±11.7 mg CH 4 m −2 h −1 .(Fig. 4) During drainage, the control cores remained at the CH 4 release level, emitting −0.1 to 138 mg CH 4 m −2 h −1 and averaging 9.4±11.7 mg CH 4 m −2 h −1 .Methane release of the manipulated cores dropped to 0 to 3.1 mg CH 4 m −2 h −1 with a mean release of 0.3±0.4mg CH 4 m −2 h −1 .During the third phase, CH 4 emissions from the control plots remained at 0 to 99.6 mg CH 4 m −2 h −1 with an average value of 8.3±12.9mg CH 4 m −2 h −1 .During the postdrainage phase, CH 4 emissions from the manipulated cores remained at the level of the drainage phase emitting 0 to 11.2 mg CH 4 m −2 h −1 and a mean CH 4 release of 0.3±0.3mg m −2 h −1 .In summary, CH 4 release of the manipulated cores remained at close to zero even when the water table reached the original position.

Nitrous oxide
Variability of emissions of N 2 O from the cores was higher than the variability of CH 4 emissions.During the first phase, N 2 O release from the control cores was 0 to 1571 µg m −2 h −1 (Fig. 5).Previous to drainage, the manipulated cores released 0 to 2255 µg N 2 O m −2 h −1 .Thus,

Properties of the peat cores
Simple measures of surface peat point towards stronger decomposition of peat in the center of the bog (Table 1): C and N content in the top 15 cm of the peat cores from the center of the bog were significantly higher than from the edge of the bog.There was no significant difference in the C/N ratio from the cores sampled at the center to the ones sampled at the edge of the bog, but cores from the edge tended towards  a higher C/N ratio.As evidenced by the von Post index, peat from the bog center was more humified than peat at the bog edge.

CO 2 evolution from incubated disturbed samples
According to the incubation experiment, N fertilization of surface peat does not control potential CO 2 release.In contrast to sampling depth or peat properties, a wide range of unfertilized and fertilized samples did not differ in the amount of CO 2 release throughout the incubation period.Following 42 days of incubation, unfertilized peat released 43.7±40.1 mg CO 2 per g of dry peat and fertilized peat released 43.0±45.9mg CO 2 per g of dry peat (Fig. 6).

Field CH 4 and N 2 O fluxes
The CH 4 fluxes that we measured in the Pietzmoor are within the range previously reported by many authors and recently reviewed by Le Mer and Roger (2001) and Whalen (2005).
Although the high spatial variability of CH 4 fluxes impedes the interpretation of data, we discuss patterns of CH 4 release.The elevated CH 4 emissions from hollows at our sites are probably due to the proximity to the water table and a shallower aerobic zone of CH 4 oxidation (Pelletier et al., 2007;Strack et al., 2004).Furthermore, some of the hollows are covered with Eriophorum vaginatum.Vascular plants, especially sedges are known for high CH 4 release (Joabsson et al., 1999;Strack et al. 2006) and Eriophorum vaginatum tussocks are CH 4 emission hotspots as they provide substrate for methanogenesis and provide a pathway for CH 4 release (Tuittila et al., 2000;Marinier et al., 2004).The somewhat elevated CH 4 emissions at the center of the bog cannot be explained by water table.However, due to the higher decomposition, field moisture could be higher in the center than at the edge.Only recently, Basiliko et al. (2007)  the water table at 24.5 cm below the surface (Fig. 7).This is in contrast to the well established relationship between CH 4 release and water table (Moore andKnowles, 1989, Moore andDalva, 1993) and the idea of water table acting as an "on-off-switch" for CH 4 emissions.The rather steady increase of maximum CH 4 emissions with a rising water table documented in Fig. 7 resembles the relation described by Jauhiainen et al. (2005).In contrast to data from boreal regions cited above, results from temperate regions more frequently do not confirm a straightforward relation between water table and CH 4 release.Fiedler and Sommer (2002) conclude that the effect of water table on CH 4 release in several peatlands of southern Germany is indirect and controlled by more strongly by redox potential, which is rarely determined.Fiedler at al. (2005) state that the thickness of the oxidative zone above the water table does control CH 4 release, but also report CH 4 emission peaks simultaneously to a slightly falling water table.
We are not able to explain the (insignificantly) elevated CH 4 emission in the center of the bog.Following the reasoning of Glatzel et al. (2004), the low degree of humification of surface peat at the edge of the Pietzmoor as evidenced by the von Post index (Table 1) should favor elevated CH 4 emission at that subsite.Consequently, the von Post index is an insufficient measure for CH 4 release.This is not surprising, because this index cannot take into account the function of living plants as Eriophorum vaginatum as conduit for CH 4 (Joabbson et al., 1999) and the supply of easily degradable compounds in the rhizosphere of Eriophorum vaginatum (Saarnio et al. 2004).
As oligotrophic peatlands are generally N limited, they are usually no sources of N 2 O (Martikainen et al., 1993).Thus, the field N 2 O fluxes reported in this contribution are high compared with these sites.However it must be taken into account that most studies from pristine oligotrophic peatlands are from boreal sites with rather low atmospheric N deposition (Nordin et al., 1998).Our site has a history of drainage, is located in the temperate zone, experiences high atmospheric N input and a rapid fluctuation in water table (Fig. 1), and, at drought conditions, NO 3 -N concentrations of 22±31 mg L −1 (Glatzel et al., 2006).The N 2 O release from the Pietzmoor is higher than the N 2 O release from a restoring peat bog in S Germany, where Drösler (2005) determined an N 2 O emission of 1 to 31 µg N 2 O m −1 h −1 .According to our research, only cultivated or drained peatlands release >100 µg N 2 O m −1 h −1 .Regina et al. (1999) measured N 2 O release of 440 µg N 2 O m −1 h −1 in a Finnish drained tall sedge fen.On the other hand, the same authors found that rewetting reduces N 2 O release from a previously drained birch-pine fen from 50 to 100 µg m −1 h −1 .Cultivated sites on organic soils from NW Finland released 70 to 170 µg N 2 O m −1 h −1 (Maljanen et al., 2001).
Water table did not control N 2 O release (Fig. 8).This is not surprising as the field campaign was rather short and N 2 O release is at its maximum in unsaturated soil (Granli and Bøckmann, 1994).Due to the infamously high spatial and temporal variability of soil N 2 O emissions (Folorunso and Rolston, 1984), the lack of a difference of N 2 O release between hummocks and hollows is not surprising.However, the significantly elevated N 2 O release from the edge of the Pietzmoor compared to its center (Fig. 2) is surprising.Even when there is no difference in water table, nitrous oxide flux from the edge of the bog is elevated.Unfortunately, no N data from peat pore water are available from 2004.However, an increased pore water NO 3 concentration at the edge of the bog is unlikely: In contrast to the center of the Pietzmoor, NO 3 concentrations in the pore water at its edge never exceeded 0.5 mg L −1 between July 2002 and July 2003 (Lemke, 2004).Considering the narrow C/N ratio of surface peat at the center and the edge of the Pietzmoor and the low degree of decomposition at the edge (Table 1), it is possible that the peat itself provided the N source for denitrification.Schiller and Hastie (1996) report N 2 O release from the destruction of surface moss following clearfelling, so it is possible that the moss is the N source.This is in line with the findings by Lamers et al. (2000), who found that, at an atmospheric N deposition rate of 12 to 18 kg ha −1 yr −1 , excess N is accumulated in Sphagnum tissue, stored as free N or Nrich free amino acids.Our C/N ratio of 30 is not far from the threshold C/N ratio of 25 for significant N 2 O emissions reported by Klemedtsson et al. (2005).In Canadian bogs and the Pietzmoor, Glatzel et al. (2004Glatzel et al. ( , 2006) ) found high CO 2 release rates from poorly decomposed surface Sphagnum peat.Since CO 2 release involves N mobilization and moderately dry conditions are accompanied by the strong CO 2 emissions (Glatzel et al., 2006), in phases of moderate dryness, NO 3 could be accumulated that is subject to denitrification and N 2 O release during subsequent wetter phases.The decreasing N 2 O release from the collar with the large Eriophorum vaginatum tussock with the progressing season suggests a competition for excess nitrogen (Silvan et al., 2005).By the end of August, plant uptake of NO 3 keeps N 2 O emission close to 0. This mechanism has been noted by Glatzel and Stahr (2001), where it led to soil N 2 O uptake.It is interesting that this pattern occurred only where the collar was vegetated by a large cottongrass tussock and suggests effective rhizosperic N uptake.It is likely that the wet summer favored rapid plant uptake of NO 3 as high soil moisture was found to be connected to efficient N uptake of Phalaris arundinacea (Rückauf et al., 2004).4.2 CH 4 and N 2 O release from mesocosms Gas fluxes from mesocosms were higher than from the field.This is due to constantly warm temperatures in the laboratory (Regina et al., 1999) and could, despite the two month equilibration phase, also be a consequence of an enduring disturbance effect following field sampling.As disturbance effects are site specific and there is no standard equilibration period, the comparison of the absolute magnitude of gas fluxes from mesocosms is not useful.Thus, the purpose of CH 4 and N 2 O flux determinations from mesocosms is the evaluation of differences between our treatments.

Methane
The variability of CH 4 fluxes from all mesocosms before drainage and from the control was high, but as a consequence, CH 4 release from the control mesocosms was not different from the mesocosms that were to be manipulated.Our finding that a water table drawdown brings CH 4 release to an end confirms the conclusion by Christensen et al. (2003) that water table acts as an "on-off-switch" for CH 4 emissions.Strack and Waddington (2007) report a more differentiated CH 4 release pattern as a result of water table drawdown.They show that CH 4 release from hummocks may rise following a drawdown due to peat subsidence.CH 4 release following drainage to −50 cm also did not decline to zero (Moore and Dalva, 1993), but the peat columns sampled by Moore and Dalva were 80 cm in length.Our experimental design however eliminated the anoxic zone, -although anoxic pockets may have been preserved -, so differences due to a differing capacity for CH 4 oxidation one might have been able to find in the bog could not be detected.It is still interesting that immediately following the beginning of drainage, CH 4 fluxes at all mesocosms declined to close to 0. Also, CH 4 release did not reappear during the third phase.This confirms findings by Freeman et al. (2002) who reports a suppression of CH 4 for >1 month following a drought and Segers (1998) stated that, due slow growth rates, methanogens require a long regeneration period following exposition to oxygen.So we are not able to report a hysteresis in CH 4 release for the falling and rising limb as detected by Moore and Dalva (1993).

Nitrous oxide
N 2 O fluxes from mesocosms declined with drainage, but did not fully recover following drainage.Increasing N 2 O release following drainage has been observed in field and laboratory experiments (Freeman et al., 1992, Martikainen et al., 1993, Regina et al., 1999).Dowrick et al. (1999) found that a moderate drought (with a water table at −8 cm) did not affect N 2 O released compared to waterlogging and that a more extreme drought (like the one that we simulated) causes an exponential increase in N 2 O release with water table depth.On the other hand, Nykänen et al. (2002) determined very low N 2 O release rates although the water table subsided up to −40 cm and one site had been fertilized with 100 kg N ha −1 prior to the experiment.Nykänen et al. (2002) explain the low N 2 O emission despite fertilization with plant uptake and the accumulation of ammonium (NH 4 ) below the root zone.Another reason for this is probably the low background N load of 6 kg ha −1 yr −1 and some capacity of the peat for adsorption of NH 4 .This is a profound difference to N dynamics of boreal bogs compared to temperate bogs in industrialized regions with high atmospheric N deposition an N loaded peat (Lamers et al., 2000).
There is a (non-significant) rise of N 2 O emissions from the manipulated mesocosm in the post-drainage phase.This could be a consequence of nitrification and an accumulation of NO 3 during the drainage phase and denitrification as the water table rises again, explaining the high NO 3 concentration in the pore water of the Pietzmoor during the drought in 2003 (Glatzel et al., 2006).Updegraff et al. (1995) emphasized the relationship between drainage and N mineralization.Regina et al. (1999) elaborate the link between drainage, high NO 3 accumulation and increased N 2 O release as well as lower NO 3 concentrations and N 2 O release as a consequence of rewetting.Van Beek et al. (2004) concluded that in low-land areas, ground water levels tend to control the magnitude of N losses via denitrification.In summary, although we do not know the reason for the rise of N 2 O emissions in the third phase, there is evidence for denitrification following NO 3 accumulation.

CO 2 evolution from incubated disturbed samples
The purpose of laboratory incubations is the isolation of confounding factors and the absolute values obtained by this type of experiment do not approximate field fluxes.Still, Moore and Dalva (1997) suggested that integrated potential production rates and field fluxes might be similar.In any case, CO 2 productions rates from peats do not differ strongly and can be compared (Glatzel et al., 2004).
The large variability of CO 2 release within the unfertilized and fertilized peat is due to the wide range of peat samples used for the experiment, involving poorly as well as strongly decomposed peat as well as hummock and hollow peat.The absence of any N limitation at optimal peat moisture shows that there is no N limitation of decomposition.Thus, the high N deposition rates in the region do not necessarily directly enhance peat decay, but favor N accumulation in the bog (Lamers et al., 2000).Besides the consequences on CH 4 and N 2 O release discussed above, a change in species composition is to be expected in case of persistent high N deposition and drought stress.Specifically, the competitiveness of Sphagnum spp.(Lamers et al., 2000, Limpens et al., 2003, Tomassen et al., 2003), Calluna vulgaris (Heil and Bruggink, 1987), and Erica tetralix (Aerts and Berendse, 1988) suffers facing atmospheric N deposition and N mineralization due to water table subsidence in favor of Molinia caerulea (Lamers et al., 2000, Limpens et al., 2003, Tomassen et al., 2003, Heil and Bruggink, 1987, Aerts and Berendse, 1988) and Betula pubescens (Tomassen et al., 2003).
We were not conclusively able to accept the second part of the first hypothesis -rewetting temporarily increases CH 4 and N 2 O release.One of the additional factors that modify the response of CH 4 and N 2 O fluxes to water table is the degree of decomposition.We add additional evidence to the notion that the von Post humification index can be used to explain N 2 O release, but not CH 4 release from restoring bogs.A large variation of the humification index occurs within small areas.In the strongly decomposed center with scarce Sphagnum coverage, N 2 O release is lower than at the poorly decomposed edge with fresh N-rich Sphagnum.Thus, the second hypothesis, -decomposition controls CH 4 and N 2 O release -can be accepted for N 2 O, but not for CH 4 .CH 4 production appears to be controlled more strongly by plant mediated factors as the CH 4 conduit and root exudates.N 2 O emission could be enhanced when N-rich plant tissue is available for decomposition.
Our work also examined the effects of N addition to surface peat and leads to the rejection of the third hypothesisatmospheric N deposition accelerates the decomposition of surface peat.
The ongoing restoration process in the Pietzmoor aims at the restoration of peatland ecosystems including reestablishment of vegetation cover, especially Sphagnum mosses, and of the hydrological regime (Rochefort and Lode, 2001).Finally, the return of its functions e.g.accumulation of carbon and nutrient cycling is aspired.Realistically, this is only possible when aiming at developing an eutrophic ecosystem rather than restoring an oligotrophic one.
Another goal of peatland restoration is the net reduction of the release of CO 2 equivalents.This contribution shows that -under conditions of high rates of atmospheric N deposition -it is important to avoid frequent water table fluctuations that may increase N 2 O release.Especially in periods when NO 3 uptake by vegetation is not strong (late autumn to early spring), a high water table must be maintained.At this point, we are not able to judge for how long a water table drawdown with subsequent restoration of high water table will decrease CH 4 release.A very low water table may decrease CH 4 and CO 2 efflux, but likely damages peat forming vegetation (Glatzel et al., 2006) and may favor growth of species adapted to a fluctuating water table as Molinia caerulea.For this reason, our present state of knowledge suggests that the reduction of the net release of CO 2 equivalents in N loaded temperate peatlands depends on a high water table.

Fig. 1 .
Fig. 1.Water table at the center and at the edge of the research site in the Pietzmoor from March to August 2004.

Fig. 2 .
Fig. 2. Methane (CH 4 ) and nitrous oxide (N 2 O) release from hummocks and hollows at the center and the edge of the Pietzmoor, Germany.Shown are means and standard deviations of CH 4 and N 2 O release of 14 field gas flux determinations from March to August 2004 (CH 4 : (A), N 2 O: (B)) originating from five center (three hummocks and two hollows) and five edge (also three hummocks and two hollows) locations (CH 4 : (C), N 2 O: (D)).Significantly different values (Wilcoxon test) are marked with different letters.

Fig. 3 .
Fig. 3. Nitrous oxide (N 2 O) release from an Eriophorum vaginatum dominated hummock from April to August 2004 and its relation to sampling date at the edge of the Pietzmoor, Germany (n=14).

Fig. 4 .
Fig. 4. Methane (CH 4 ) release from peat cores before, during, and following the drainage experiment (open bars).Unmanipulated control cores are black.Mean values and standard deviation from three replicates are shown.Significantly different values (Wilcoxon test) are marked with different letters.

Fig. 5 .
Fig. 5. Nitrous oxide (N 2 O) release from cores before, during, and following the drainage experiment (open bars).Unmanipulated control cores are black.Mean values and standard deviation from three replicates are shown.Significantly different values (Wilcoxon test) are marked with different letters.

Fig. 6 .
Fig. 6.Carbon dioxide (CO 2 ) release from unfertilized peat and peat fertilized with ammonium nitrate from the Pietzmoor, Germany, following a 42 day incubation.Mean values and standard deviation from 17 samples are shown.Significantly different values (Wilcoxon test) are marked with different letters.

Table 1 .
Carbon (C) and nitrogen (N) concentration and von Post humification index in the top 15 cm of peat cores used for the water table manipulation experiment from the Pietzmoor, Germany.Mean values and standard deviation from six replicates are shown.Significantly different values (Wilcoxon test) within one line are marked with different letters.
table [cm below surface] Nitrous oxide release and water table from the Pietzmoor, Germany during the 2004 field season.Shown are individual flux determinations from 10 collars and 14 sampling dates from March to August 2004.Water table was corrected for elevation of hummocks by adding 20 cm distance from the bog surface.