Nitrogen mineralization and gaseous nitrogen losses from waterlogged and drained organic soils in a black alder ( Alnus glutinosa ( L . ) Gaertn . ) forest

Black alder ( Alnus glutinosa(L.) Gaertn.) forests on peat soils have been reported to be hotspots for high nitrous oxide (N2O) losses. High emissions may be attributed to alternating water tables of peatlands and to the incorporation of high amounts of easily decomposable nitrogen (N) into the ecosystem by symbiotic dinitrogen (N 2)-fixation of alder trees. Our study addressed the question to what extent drainage enhances the emissions of N 2O from black alder forests and how N turnover processes and physical factors influence the production of N 2O and total denitrification. The study was conducted in a drained black alder forest with variable groundwater tables at a southern German fen peatland. Fluxes of N 2O were measured using the closed chamber method at two drained sites (D-1 and D-2) and one undrained site (U). Inorganic N contents and net N mineralization rates (NNM) were determined. Additionally a laboratory incubation experiment was carried out to investigate greenhouse gas and N 2 fluxes at different temperature and soil moisture conditions. Significantly different inorganic N contents and NNM rates were observed, which however did not result in significantly different N 2O fluxes in the field but did in the laboratory experiment. N 2O fluxes measured were low for all sites, with total annual emissions of 0.51 ± 0.07 (U), 0.97± 0.13 (D-1) and 0.93± 0.08 kg N2O–N ha−1 yr−1 (D-2). Only 37 % of the spatiotemporal variation in field N 2O fluxes could be explained by peat temperature and groundwater level, demonstrating the complex interlinking of the controlling factors for N2O emissions. However, temperature was one of the key variables of N 2O fluxes in the incubation experiment conducted. Increasing soil moisture content was found to enhance total denitrification losses during the incubation experiment, whereas N 2O fluxes remained constant. At the undrained site, permanently high groundwater level was found to prevent net nitrification, resulting in a limitation of available nitrate (NO−3 ) and negligible gaseous N losses. N2O flux rates that were up to four times higher were measured in the incubation experiment. They reveal the potential of high N2O losses under changing soil physical conditions at the drained alder sites. The high net nitrification rates observed and high NO −3 contents bear the risk of considerable NO−3 leaching at the drained sites.


Introduction
Black alder (Alnus glutinosa (L.) Gaertn.)forests represent the natural vegetation in many semi-wet and wet regions, mostly on soils with high organic carbon contents (Dilly et al., 1999;Schäfer and Joosten, 2005;Kätzel, 2003).Unlike most other wetlands, alder forests on peatland represent non N-limited ecosystems, due to the ability of alder trees for symbiotic N 2 -fixation of up to 85 kg N ha −1 yr −1 (Dittert, 1992;Augustin, 2003).To date, there are only a few studies on N turnover and N transformation processes Published by Copernicus Publications on behalf of the European Geosciences Union.

T. Eickenscheidt et al.: Nitrogen mineralization and gaseous nitrogen losses
in black alder forests on peatland.First results indicate that these ecosystems show high N-transfer rates between the atmosphere, the alder trees and the soil (Augustin, 2003).Through the symbiotic N 2 -fixation high amounts of easily decomposable N enter the ecosystem and lead to a low C / N ratio of the soil organic matter (Lõhmus et al., 2002).Semiwet peat soils with low C / N ratios are known to be potential hotspots for gaseous and dissolved nitrogen losses via mineralization, nitrification, denitrification and nitrate leaching (Lõhmus et al., 2002;Gundersen et al., 1998;Ollinger et al., 2002;Klemedtsson et al., 2005).Enhanced N 2 O emissions, in particular, are of great interest due to the fact that N 2 O acts as a radiative forcing greenhouse gas (IPCC, 2007) and contributes to the chemical destruction of stratospheric ozone (Crutzen, 1979).The few annual observations of N 2 O emissions from Alnus forest ecosystems on drained or undrained peatlands range between 0.1 and 72.0 kg N ha −1 yr −1 (Augustin et al., 1998;Brumme et al., 1999;Merbach et al., 2001;von Arnold et al., 2005).Most emissions exceed by far the IPCC default emission factor of 0.6 (0.16-2.4) kg N 2 O-N ha −1 yr −1 (IPCC, 2006) for boreal and temperate nutrientrich forest peatlands.The high range requires in-depth process understanding to narrow down the uncertainty and better understand N 2 O formation, not least because of the attempt to reduce the climate impact of drained fen peatland by rewetting and afforestation with black alder trees (Alnus glutinosa (L.) Gaertn.)(Schäfer and Joosten, 2005;Wichtmann and Joosten, 2007).
The production of N 2 O depends mainly on the two microbial processes of denitrification and nitrification (Davidson, 1986).Thus, the availability of soil mineral N and oxygen are the main controlling factors (Del Grosso et al., 2000;Jungkunst et al., 2004).Furthermore, soil temperature is known to be a key variable controlling both processes (Firestone and Davidson, 1989;Smith et al., 1998).Highest N 2 O fluxes are mainly related to incomplete denitrification.This requires the presence of efficient electron acceptors (e.g., NO − 3 ) to prevent the reduction of N 2 O to N 2 (Bremner and Blackmer, 1981;Speir et al., 1995;Jungkunst et al., 2004).Due mainly to analytical difficulties in the determination of microbially produced N 2 , to date the denitrification potential from waterlogged peat soils has rarely been estimated (Watts and Seitzinger, 2000;Mander et al., 2003;Teiter and Mander, 2005;Wray and Bayley, 2007;Roobroeck et al., 2010;Soosaar et al., 2011;Uri et al., 2011).The substitution of ambient N 2 by Helium (He) in laboratory studies possibly represents the most reliable method for the direct and simultaneous determination of N 2 O and N 2 exchange rates (Butterbach-Bahl et al., 2002;Roobroeck et al., 2010;Butterbach-Bahl et al., 2013).
The objectives of our study were to quantify (i) the nitrogen mineralization and nitrogen transformation processes, and (ii) N 2 O emissions and the factors regulating the N 2 O emissions and N 2 O / N 2 ratio from waterlogged and drained black alder forest on organic soil.
We hypothesize that (1) in drained black alder forests high N 2 O emissions occur due to accelerated N turnover and N mineralization, and (2) N 2 O is replaced by N 2 losses during periods of temporarily high water levels.In contrast, NO − 3 limitation in permanently waterlogged peat soils results in negligible N 2 O and N 2 losses.To investigate our hypotheses, we selected three sites in an alder forest along a groundwater table gradient, representative of the small-scale heterogeneity of the forest stand.

Study area and experimental design
The study was carried out in a drained 45-year-old black alder (Alnus glutinosa (L.) Gaertn.)forest stand (2.2 ha) on organic soil (Freisinger Moos, 48 • 22 N, 11 • 40 E; 449 m a.s.l.).According to the climate station in Weihenstephan, located 10 km northeast of the site, the 30-year mean annual temperature was 7.5 • C and the mean annual precipitation was 787 mm .At a German Level II monitoring area, located at 5 km distance from the alder forest, annual stand N deposition amounted to 17.67 kg N ha −1 yr −1 with a NH + 4 -N : NO − 3 -N ratio of 46 : 54 in 2011 (data were collected and analyzed by the Bavarian State Institute of Forestry).The black alder stand was planted on grassland in 1965 and has not been managed.In 2009, stem density was approximately 1031 trees per hectare, with a mean height of 19.7 m and a mean diameter at breast height of 21 cm (S.Röhling, personal communication, 2012).Approximately 15 % of the forest area consists of undrained hollows whereas most parts are drained by ditches.In October 2009, we selected three sites which differed in groundwater table (Table 1) as well as in their physical and chemical soil properties relevant for N 2 O processes (e.g., bulk density, carbon and nitrogen content) (Table 2).One site was situated in an undrained hollow (named U).The other two sites were situated in the drained part of the forest (named D-1 and D-2) along a gradient of drainage depth.Soils at the drained sites were classified as sapric histosols whereas the undrained site was classified as fluvisol (according to the World Reference Base for Soil Resources (WRB), 2006). Te surface organic layer at the drained sites was a L-mull humus, and at the undrained site a submerged mud layer.At the site U, the dominant species in the herb layer were Phragmites australis and Carex acutiformis.Site D-1 was dominated by Impatiens glandulifera, Impatiens parviflora, Deschampsia cespitosa and Circaea x intermedia, whereas site D-2 was dominated by Carex acutiformis, Circaea x intermedia, Rubus caesius and Carex elongata.
At each site, three PVC-collars for greenhouse gas (GHG) measurements (inside dimensions 75 × 75 cm) were permanently inserted 10 cm into the soil.The collars of a site were grouped closely to each other (distance of 0.8 to  We measured fluxes of N 2 O every second week from December 2010 to January 2012 using the static manual chamber method with a chamber height of 0.5 m (Livingston and Hutchinson, 1995).At site U, in periods where the forest floor vegetation grew higher than the chamber height, extensions (same dimension as chamber) were installed between the collar and chamber (white, opaque; volume varied between 309 and 927 L).A detailed description of chamber dimension and configuration is given in Drösler (2005).Four gas samples were taken at four regular time intervals after chamber closure (enclosure time 60 min, in the case of two extensions 120 min).The samples were collected in 20 mL glass vials, each sealed with a butyl rubber septum.The vials were flushed with chamber air for 30 s using a portable micro pump (KNF Neuberger GmbH, NMP015B), so that the air in the vials was exchanged 32 times.In addition, the pump was used to build up an overpressure of approximately 550 mbar to protect the sample against fluctuations in atmospheric pressure during storage.Gas analyses were carried out with a gas chromatograph (Perkin and Elmer, Clarus 400 GC or Clarus 480 GC) equipped with a headspace auto sampler (Perkin and Elmer, TurboMatrix 110), a Porapack 80/100 mesh column and an electron capture detector (ECD) for N 2 O (ECD temperature 380 • C).Gas flux rates were calculated from the linear change in gas concentration over time considering chamber air temperature and atmospheric pressure.Gas fluxes were accepted when the linear regression was significant (P ≤ 0.05).In the case of small N 2 O fluxes, fluxes were also accepted if the coefficient of determination was ≥ 0.90 and the regression slope was between −1 and 1 ppb min −1 .The cumulative annual mean exchange rate was calculated by linear interpolation between the measurement dates.

Soil sampling and laboratory analysis
Total carbon (C tot ), organic carbon (C org ) and total nitrogen (N tot ) were analyzed by the AGROLAB Labor GmbH (Bruckberg, Germany) on three mixed soil samples per site.Each sample was composed of nine individual samples collected close to each collar at two soil depths (0-10, 10-20 cm).After drying for 72 h at 40 • C, soil samples were sieved to 2 mm to remove stones and living roots.
During every gas flux measurement, one mixed soil sample consisting of nine individual samples was collected at two soil depths (0-10, 10-20 cm) for each site for the determination of mineral N (N min = NH + 4 -N + NO − 3 -N) contents.
Samples were immediately cooled in an ice box until further processing.Mineral N was extracted after shaking 40 g of fresh soil with 160 mL CaCl 2 (0.0125 M) for one hour.The extracts were filtered through a 4-7 µm filter paper (Whatman 595 1/2) and the first 20 mL of the extract were discarded (VDLUFA, 1997).The solution was frozen at −20 • C until analysis, which was conducted by the AGROLAB Labor GmbH (Bruckberg, Germany).A subsample of 20-30 g was used to determine the gravimetric water content, which was taken into account for the calculation of mineral N concentrations.

Net N mineralization
In 2011, net ammonification, nitrification and N mineralization rates (NNM) were determined in situ in four different periods in time (March, June, September and December).At each study site and period, three pairs of intact soil cores were taken randomly from the upper 20 cm of the soil layer using a cylindrical auger (diameter of 45 mm).In the case of flooding at the undrained site, disturbed soil samples were used.One core of each pair was sealed in low density polyethylene (LDPE) bags (19 µm) and returned into the soil for incubation (t1), which lasted 21 days (20 days in December) (e.g., Adams et al., 1989).The other soil cores were used as initial samples (t0).Mineral N contents were determined as described above after soil cores were divided into 0-10 and 10-20 cm sections.Net ammonification, net nitrification and net N mineralization were calculated according to the method described by Uri et al. (2003)

Laboratory incubation experiment
To examine the magnitude of potential total denitrification losses from the undrained and drained alder forest, a laboratory incubation experiment was performed; the experimental conditions were supposed to reproduce field conditions as closely as possible in respect to their temperature range and mean WFPS.

Soil core sampling and WFPS adjustment
In June 2011, eight intact soil cores with a diameter of 7.3 cm and a height of 6 cm were collected from a soil depth of 9-15 cm close to each collar of site D-2 for incubation in the laboratory (in total, 24 soil cores per site).Four samples per collar were immediately weighed and stored at 4 • C. The remaining four intact soil cores per collar were used to estimate the actual water-filled pore space (WFPS).This information was used to adjust the WFPS to 70 % and 100 % in the cores stored for laboratory incubation (six cores for 70 % and six cores for 100 % per site).
For the soil core sampling at site U, it had to be considered that the drying of the soil cores in order to adjust a WFPS of 70 % would lead to distinct shrinking of the waterlogged organic soil in the cores.To avoid this, three cubic blocks (dimensions 40 × 25 cm, height 20 cm) were collected with plastic containers from which the bottom plate was removed beforehand.For transport, storage and drying to 70 % WFPS, the containers were placed on perforated metal plates.Six intact soil cores for site U were collected from these soil blocks after they reached the required WFPS level.Additionally, six soil cores per collar were directly collected in the field and treated in the same way as the samples from site D-2 (four samples for determining actual WFPS and two samples for incubation at 100 % WFPS level per collar).All soil cores were stored at 4 • C until they were placed in the incubation vessels.
After incubation the WFPS of the soil cores was determined to test the accuracy of the adjusted WFPS.The samples from site U and D-2 had exactly 100 % WFPS as well as 70 % WFPS at D-2 (on average 70.4 ± 2.2 % WFPS).For site U, however, WFPS had been adjusted to 83.0 ± 0.5 % instead of 70 %.We suspect that, due to the shrinking of the soil sample blocks, the bulk density must have changed; therefore the calculation of the required water content was not as accurate as needed to adjust a WFPS of 70 %.

Determination of gas fluxes
For measurements of N 2 , N 2 O, CO 2 and CH 4 fluxes, we applied the helium-oxygen (He-O) method (Butterbach-Bahl et al., 2002;Mander et al., 2003;Roobroeck et al., 2010;Uri et al., 2011) using four different incubation temperatures (0, 5, 15 and 25 • C).Five replications per site and moisture content were simultaneously placed in special gas-tight incubation vessels inside a climate chamber.The sixth one served as a reserve.Analyses were conducted in the laboratory of the Institute of Landscape Matter Dynamics, Leibniz Centre for Agricultural Landscape Research (ZALF), Müncheberg, Germany.Four substitution sequences with moderate evacuation (0.047 bar), followed by flushing the vessels with an artificial He/O 2 gas mixture (20.58 % O 2 , 347.8 ppm CO 2 , 1.780 ppm CH 4 , 0.290 ppm N 2 O, 3.04 ppm N 2 , rest He), were conducted to remove ambient N 2 .Subsequently, the air temperature of the climate chamber was set to 0 • C and a continuous He/O 2 gas flow rate of 15 mL per minute was adjusted to the vessel headspaces, followed by a 24 h period to establish a new flow equilibrium.From each vessel, we measured the N 2 O, CO 2 and CH 4 headspace concentration once and the N 2 concentration three times.After gas flux measurements were done, we immediately increased the incubation temperature to the next level for 20 h and continued the measurements.Soil core samples lost approximately 5.0 ± 0.9 % and 2.7 ± 0.3 % of their initial water content at 100 % and 70/83 % WFPS during the incubation procedure of five days.Concentration of N 2 were analyzed by a micro-GC (Agilent Technologies, 3000 Micro GC), equipped with a thermal conductivity detector (TCD).Gas chromatograph settings were as follows: TCD temperature 60 • C, sample inlet 60 • C, molsieve capillary column (14 m), oven tempera-ture 60 • C, carrier gas He 6.0 (1 mL min −1 ).Concentrations of trace gases were analyzed by a GC (Shimadzu, Duisburg, Germany, GC-14B) equipped with a flame ionization detector (FID) for CH 4 and an ECD for N 2 O and CO 2 .GC settings were as follows: FID temperature 310 • C, ECD temperature 310 • C, column Porapack 80/100 mesh, oven temperature 60 • C, carrier gas N 2 6.0 (13 mL min −1 ).Background N 2 concentrations varied between 3.5 and 4.5 ppm (ca. 3 ppm originated from the artificial He/O 2 gas mixture and 1 ppm from diffusion into the incubation measuring device).Flux rates were calculated from the actual gas concentration of the continuous flow rate from the vessels' headspaces after subtraction of a blank value from a vessel without a soil core, which is equivalent to concentrations from the artificial He/O 2 gas mixture.The lowest detectable flux rates were In order to obtain the temperature effect on gas flux rates, Q 10 values were calculated from the following equation: where Q 10 is the temperature quotient; R 1 and R 2 are gas flux (CO 2 , CH 4 , N 2 O, N 2 ) rates (mg C or N m −2 h −1 ; µg C or N m −2 h −1 ); and T 1 and T 2 are temperature ( • C).

consumption
According to Schaufler et al. (2010) laboratory incubation experiments are limited by the fact that sample storage and incubation procedure decrease the substrate (e.g., NO − 3 ) supply and thereby falsify the results.The amount of N min in the solution from soil cores was not ascertained in the present study.However, a rough estimation of available NH + 4 and NO − 3 was performed on the basis of t0 samples from the net N mineralization experiment conducted, which started three days before soil core samples were taken (Table 3).In the present study, significant amounts of gaseous N losses are only expected in the case of complete denitrification with N 2 as the end product.Thus only samples incubated at 100 % WFPS were potentially affected by NO − 3 limitation.According to our estimation, samples from site U at 100 % WFPS were presumably permanently NO − 3 limited at each temperature level during the incubation procedure, which reflects the natural site conditions.However, for samples from site D-2 at 100 % WFPS no NO − 3 limitation was expected.For soil core samples with 70/83 % WFPS, it has to be kept in mind that not only N consumption but also nitrification takes place.However, we can not totally rule out that reduced NO − 1.27 6.02Estimated mean N 2 exchange from soil cores at 4

Statistical analysis
Statistical analyses were conducted using R 2.12.1 (R Development Core Team, 2013).The assumption of normality of residuals was tested using the Lilliefors or Shapiro-Wilk test and by plotting the quantile-quantile plots.Homogeneity of variances of residuals was checked with the Levene or Breusch-Pagan test and by plotting the residuals against the fitted values.If necessary, data were transformed (Box-Cox) prior to analyses.We used analysis of variance (ANOVA) (for NNM rates) or the nonparametric Kruskal-Wallis rank sum test (for WFPS, NO − 3 , NH + 4 ) to compare means of samples.In case of significant differences between the means, we used Tukey's honest significant differences (TukeyHSD) or the nonparametric pairwise Wilcoxon rank sum test with Bonferroni correction for multiple comparisons.For testing two independent sample means (for WFPS, NO − 3 , NH + 4 ), we used the nonparametric Mann-Whitney U test.
Due to temporal pseudoreplication of time series data (N 2 O field measurements) and repeated measurements of vessels in the incubation experiment, we applied linear mixed effect models (Crawley 2007;Eickenscheidt et al., 2011;Hahn-Schöfl et al., 2011).For N 2 O field measurements we set up a basic model with site type as a fixed effect and the spatial replication (individual plot) nested in time as a random effect.For the incubation experiment (N 2 O, N 2 , CO 2 , CH 4 ), the fixed structure included the temperature, the site type and the soil water-filled pore space as well as all possible interaction terms.Non significant terms were removed from the fixed structure.The individual vessel was set as a random effect.We extended the respective basic model by a variance function when heteroscedasticity was observed.In case of significant serial correlation in data, a first-order temporal autoregressive function was included in the model.Autocorrelation was tested with the Durbin-Watson test and by plotting the empirical (partial) autocorrelation structure (Eickenscheidt et al., 2011).The model extension was proven by the Akaike information criterion (AIC).For multiple comparisons, we conducted Tukey contrasts using the general linear hypothesis function from the "multcomp" package (Hothorn et al., 2013).We used nonlinear regressions to explain field N 2 O fluxes and NNM rates.The model structure for mean N 2 O fluxes included a linear term for the soil temperature in −2 cm soil depth and a cubic term for the mean groundwater (GW) level.Serial autocorrelation were not taken into account in the regression model.The nonlinear model for NNM rates was optimized by using the program Table Curve 3-D (version 4.0) (nonlinear equation is shown in Fig. 4).We accepted significant differences if P ≤ 0.05.Results in the text are given as means ±1 standard error.

Site characteristics
In 2011, the forest air temperature in 20 cm height ranged from −12.3 to 27.8 • C with an annual mean of 9.1 • C. Soil temperature in −2 cm soil depth averaged 9.6 • C at the drained sites and was slightly lower with 9.2 • C at the undrained site.The topsoil at the undrained site U consisted of a mud sediment while the topsoil at the drained sites D-1 and D-2 consisted of highly degraded fen peat.Total organic carbon and the total nitrogen concentrations in the 0-10 cm and 10-20 cm soil layers were two to three times higher at site U than at sites D-1 and D-2 (Tables 1 and 2).Due to much higher bulk density, however, the total N stock in the upper 20 cm of the soil was more than 50 % higher at D-1 and D-2 than at U. The C / N ratios of all sites were narrow (between 11 and 16), indicating nitrogen-rich conditions.No relation between soil properties and N fluxes was determined so that the results and discussion focus on groundwater table and climate variables.
The three sites showed similar dynamics in their annual hydrographs but on different levels (Fig. 1).The mean annual groundwater level was +4 cm at U, −40 cm at D-1, and −47 cm at D-2.At site U, WFPS remained permanently close to 100 %.At the drained sites, WFPS was close to 70 % (D-1) and 63 % (D-2) in the 0-10 cm layer and around 80 % in the 10-20 cm layer with some temporal variation (Table 1).

Nitrogen availability and net N mineralization
In 2011, extractable NO − 3 contents in both soil depths investigated differed significantly among the three sites (all P < 0.03) (Table 1).At the undrained site, nitrate was not detectable most of the time, whereas the drained sites showed high NO − 3 contents (Fig. 2a), in particular for the soil depth of 10-20 cm.However, differences between the two soil depths were not significant.
In contrast to NO − 3 , significantly higher NH + 4 contents were measured at the undrained site than at the drained sites (Table 1).NH + 4 contents did not significantly vary between soil depths.
In line with the high contents of extractable inorganic N compounds, NNM rates were also high at the drained sites (Fig. 3a and b).For both soil depths, the drained sites had significantly (all P < 0.005) higher NNM rates than the undrained site, whereas differences between the drained sites were not significant.The NNM rates decreased with soil depth, but differences were not statistically significant.The mean NNM rates of the four time periods investigated and the upper 20 cm of the soil were 23 ± 7 mg N m −2 d −1 (U), 179 ± 35 mg N m −2 d −1 (D-1) and 142 ± 32 mg N m −2 d −1 (D-2).Assuming that the time periods are representative of the annual dynamics, mean annual NNM rates are estimated to be 84 ± 26 kg N ha −1 yr −1 (U), 653 ± 128 kg N ha −1 yr −1 (D-1) and 518 ± 117 kg N ha −1 yr −1 (D-2).The NNM rates of the upper 10 cm displayed a clear seasonality with high rates during the summer months and low rates during the winter months.However, only in June a significantly (all P < 0.001) higher NNM rate was observed compared to December and March.At the drained sites, net nitrification amounted to nearly 100 % of the NNM, whereas ammonification was the dominating process at the undrained site.Negative net nitrification rates (NO − 3 consumption) were detected at the undrained site in June and September in 0-10 cm soil depth.The NNM in the upper 10 cm of the soil depended on the WFPS and the mean temperature in −5 cm soil depth (Fig. 4).

Field N 2 O fluxes
Annual N 2 O emissions were unexpectedly low at all three sites (Fig. 5) with 0.51 ± 0.07 kg N ha −1 yr −1 (U), 0.97 ± 0.13 kg N ha −1 yr −1 (D-1) and 0.93 ± 0.08 kg N ha −1 yr −1 (D-2).The differences in N 2 O fluxes between the sites were not statistically significant.At the drained sites the highest N 2 O emissions were observed during the late summer and autumn, whereas at the undrained site highest emissions occurred during the dry period in May (Figs. 1 and 5).The temperature at −2 cm soil depth and the mean groundwater level represented the best explanation for mean N 2 O fluxes; nevertheless only 37 % of the spatiotemporal variation could be explained (P < 0.001) (Fig. 6).Site-specific regression analyses explained 55 % of temporal N 2 O variation at D-1 (P < 0.001), 32 % at D-2 (P < 0.01) and 20 % at U (P <0.04).

Effect of temperature and soil moisture on
greenhouse and dinitrogen gas exchange under laboratory conditions

CO 2 exchange
CO 2 exchange indicates heterotrophic soil respiration in the incubation.No significant effect of soil moisture and site type on CO 2 exchange was detected (Fig. 7a), indicating comparable soil microbial activity in all incubated treatments.A statistically significant (P < 0.0001) quadratic effect of tem-perature on CO 2 emissions was observed in all treatments with Q 10 values between 1.82 and 17.80 mg C m −2 h −1 10 • C −1 .

CH 4 exchange
No significant CH 4 emissions were observed in any treatment (Fig. 7b).Incubations at 100 % WFPS showed consistent near zero CH 4 fluxes or slight CH 4 uptake without significant effects of site type and temperature.At 70/83 % WFPS the uptake of CH 4 was significantly (P < 0.0001) higher at all temperature levels and for both sites compared  to 100 % WFPS.CH 4 uptake increased significantly with temperature only in incubated samples of the drained site (P < 0.001).

N 2 O exchange
N 2 O exchange differed in response to site type, WFPS and temperature.Significantly (P < 0.0001) higher N 2 O fluxes at both WFPS levels were measured from samples of the drained site (D-2) compared to samples from the undrained site (U) (Fig. 7c).At the drained site, Q 10 of N 2 O emissions was 2.17 µg N m −2 h −1 10 • C −1 at 70 % WFPS for the temperature range of 0 to 25 • C and 2.12 µg N m −2 h −1 10 • C −1 at 100 % WFPS for 0 to 15 • C. For both WFPS levels, samples of the undrained site (U) only showed a weak, but significant (all P < 0.03) temperature response for the temperature range of 0 to 15 • C. The Q 10 values of N 2 O emissions for the undrained site are 2.07 µg N m −2 h −1 10 • C −1 at 100 % WFPS for 0 to 15 • C and 1.51 µg N m −2 h −1 10 • C −1 at 83 % WFPS for 0 to 25 • C.
For both sites a decline in nitrous oxide emissions at 25 • C and 100 % WFPS was observed.However, this pattern was also detected at 83 % WFPS for the undrained site.N 2 O emissions observed at the undrained site U were on the same order of magnitude as the field measurements, whereas the drained site D-2 samples showed up to four times higher N 2 O fluxes during the laboratory incubation than in the field at 70 % WFPS.

N 2 exchange
N 2 emissions were near zero at 70/83 % WFPS at low temperatures and never exceeded 0.4 mg N m −2 h −1 at high temperatures (Fig. 7d).N 2 emissions at 100 % WFPS were consistently at least 10 times higher than at 70/83 % WFPS (P < 0.0001).N 2 exchange also reacted to site type at 25 • C and to temperature.At the drained site, Q 10 of N 2 emissions was 2.03 µg N m −2 h −1 10 • C −1 at 70 % WFPS for the temperature range of 0 to 25 • C and 1.48 µg N m −2 h −1 10 • C −1 at 100 % WFPS for the temperature range of 5 to 25 • C. At the undrained site, Q 10 of N 2 emissions was 2.13 µg N m −2 h −1 10 • C −1 at 83 % WFPS for the temperature range of 0 to 25 • C. N 2 emissions of the undrained site did not respond to temperature at 100 % WFPS.N 2 emissions increased most strongly from 15 to 25 • C in samples of the drained site at 70 % and in particular 100 % WFPS and in samples of the undrained site at 83 % WFPS.
The high N 2 emissions observed at 100 % WFPS resulted in low N 2 O / N 2 ratios, with mean values of 0.03 ± 0.01 (U) and 0.10 ± 0.02 (D-2).A similarly low N 2 O / N 2 ratio (mean value of 0.12 ± 0.03) were found at 83 % WFPS for the samples from the U site, as a result of the low N 2 O emissions.In contrast, the samples from the drained site D-2 showed a considerably higher N 2 O / N 2 ratio, with a mean value of 0.62 ± 0.11 (note that in the case of zero N 2 fluxes no ratio was calculated).No relationship between temperature and N 2 O / N 2 ratio was observed for either soil moisture contents.

Nitrogen mineralization and transformation processes
For both drained sites, the NNM rates observed (518-653 kg N ha −1 yr −1 ) are at the high end of NNM rates given in the literature.Janiesch (1978) and Janiesch et al. (1991) reported NNM rates of up to 500 kg ha −1 yr −1 (0-15 cm soil depth) and of up to 321.2 kg ha −1 yr −1 (0-10 cm soil depth) for different Alnus glutinosa forests from degraded fen peatlands in northern Germany.The significantly lower NNM rates from the undrained site are in the range of NNM rates reported for other seminatural black alder forests (Janiesch, 1978;Janiesch et al., 1991).
The compiled values from the literature and the results of the present study indicate an increasing NNM rate in black alder forests with increasing degradation of the fen peatland.In forest soils, high NNM rates are generally related to narrow C / N ratios (Gundersen et al., 1998;Ollinger et al., 2002), which are also characteristic for alder swamps (Mäkinen, 1979) and which were observed in the present study.Typically NNM rates for deciduous stands range between 50 to 150 kg N ha −1 yr −1 (Aber et al., 1989).The considerably higher NNM rates measured in the present study compared to those of other deciduous forests can partly be attributed to the high amount of easily decomposable organic nitrogen which becomes incorporated into the forest ecosystem through symbiotic atmospheric N 2 fixation by Frankia actinomycetes (Kätzel, 2003).Dittert (1992) estimated N 2 fixation rates of between 40 to 85 kg N ha −1 yr −1 for a black alder stand in northern Germany.Comparable and even higher values have also been reported for grey alder stands (e.g., Rytter et al., 1991;Lõhmus et al., 2002;Mander et al., 2005;Uri et al., 2011).Additionally, drainage of histosols supplies large amounts of N through mineralization of ancient organic matter (Klemedtsson et al., 2005).However, the distinct differences in the C and N contents of the drained sites did not result in significantly different NNM rates in the present study.Low retranslocation of nitrogen from senescing leaves to the alder tree (Mander et al., 1997;Lõhmus et al., 2002;Uri et al., 2011) causes high contents of N and narrow C / N ratios of the foliage, favoring the fast decomposition of the leaf litter (Struwe and Kjøller, 1990;Rytter et al., 1991;Lõhmus et al., 2002;Uri et al., 2011).The organic surface layer at the drained sites was classified as L-mull humus type, indicating the fast decomposition of the incorporated alder leaves.
Beside substrate chemistry, moisture content together with temperature are considered as the most important abiotic factors influencing biogeochemical transformation processes (Tietema et al., 1992).In general, increasing moisture contents stimulate the biogeochemical processes up to a threshold value where anaerobicity can limit microbial activity (Tietema et al., 1992).Temperature has a similar effect on microbial activity, due to the fact that rates of enzymatic processes generally increase with increasing temperature (Michaelis and Menten, 1913).This finding becomes apparent in the Gaussian relationship observed between NNM rates in 0-10 cm soil depth, mean WFPS and mean temperature in −5 cm soil depth during the incubation period (Fig. 4).Significant effects of soil moisture and temperature on NNM or nitrification rates were also reported by other studies (e.g., Kowalenko and Cameron, 1976;Goodroad and Keeney, 1984;Gonçalves and Carlyle, 1994).The minor reduction in N turnover observed with increasing soil depth furthermore reflects the influence of aeration and temperature on mineralization processes.We attributed the significantly lower NNM rates at the undrained site to the limitation of microorganisms by anaerobicity due to the permanently high water levels.The oxygen deficit at the undrained site also explained why NNM was solely determined by net ammonification.Events of NO − 3 presence and NO − 3 consumption in the upper soil layer suggest that in periods of low water level net nitrification also occurs at the undrained site.In contrast, at the well-aerated drained sites, the NH + 4 produced becomes immediately oxidized to NO − 3 , hence net nitrification entirely controls the NNM.However, using NNM rates as indicators of soil N dynamics bears the risk of considerably underestimating the intensity and complexity of the real N fluxes (Augustin, 2003); thus, further investigation should rather use the 15 N isotope dilution method (e.g., Davidson et al., 1991) for measuring gross rates of all relevant N turnover processes.
The permanently high mineral N contents detected in the probed soil depth probably indicate that the alder forest investigated has reached N saturation, due to the fact that the availability of N exceeds the biotic uptake capacity of the system (Aber et al., 1989).High contents of available NO − 3 in the soil carry the risk of leaching or gaseous losses (Robertson, 1982).Several studies reported NO − 3 losses by leaching from soils under alder stands (e.g., Binkley et al., 1992;Van Miegroet et al., 1992;Compton et al., 2003;Uri et al., 2011).The slightly higher NO − 3 contents but lower NNM rates observed in the soil depth of 10-20 cm compared to the soil depth of 0-10 cm at the drained sites probably indicate leaching of NO − 3 into deeper soil layers.Analysis of groundwater samples from the sites investigated showed nitrate contents of up to 36.2 mg L −1 and dissolved organic nitrogen (DON) contents of up to 2.8 mg L −1 (B.Tiemeyer, personal communication, 2013), confirming the earlier assumption of N leaching.Nitrate leaching into the anaerobic peat profile can result in enhanced denitrification (Regina et al., 1996), wherein the ratio of N 2 O / N 2 depends on the residence time of N 2 O on its way from the production site in the anaerobic subsoil to the atmosphere ( van Groenigen et al., 2005).In the present study, field N 2 O fluxes observed at the drained sites are very low.Different studies reported considerably higher values for N 2 O emissions from drained black alder stands, varying from 5 to 75 kg N ha −1 yr −1 (Mogge et al., 1998;Augustin, 1998;Brumme et al., 1999;Merbach et al., 2001;Hefting et al., 2003;von Arnold et al., 2005).According to Klemedtsson et al. (2005), NO − 3 availability is the main driver of N 2 O emissions from drained histosols.However, despite distinct differences in the NO − 3 contents and NNM rates, significantly different N 2 O emissions between the three sites were not found in the present study.This was in contrast to our hypothesis that accelerated N turnover and N mineralization leads to significantly increased amounts of N 2 O emissions.We cannot rule out, however, that we may have missed high N 2 O events on the drained sites due to our regular measurement intervals.In addition many wetland adapted trees such as alder can mediate N 2 O fluxes from the soil to the atmosphere through the stem (Rusch and Rennenberg, 1998;Pangala et al., 2012;Machacova et al., 2013).This flux was not taken into account in the present study so that this study may underestimate the N 2 O emissions from the alder ecosystem.
In the alder forest investigated, temperature in −2 cm soil depth and the GW level best explained field N 2 O flux rates, whereas no relationship was found between field N 2 O fluxes and the amount of NO − 3 in the soil.It is well known that temperature is a key variable affecting the emission rates of N 2 O because both nitrification and denitrification are microbial processes (Firestone and Davidson, 1989;Maag and Vinther, 1996;Smith et al., 1998;Dobbie and Smith, 2001).Higher soil moisture content increases N 2 O emissions through higher denitrification rates (Wolf and Russow, 2000).In our study, however, temperature and GW only explained 37 % of the spatiotemporal variation of mean N 2 O flux rates.This demonstrates the complex interlinkage of the controlling factors for N 2 O emissions, as reported for agricultural soils by Jungkunst et al. (2006).According to Couwenberg et al. (2011), mean annual GW levels below −15 cm show a wide range of N 2 O emissions in fens; in contrast, Jungkunst et al. (2004) found highest annual N 2 O emissions in a Norway spruce stand (black forest, southwest Germany) at a GW level of −20 cm, whereas higher or lower GW levels decreased the emissions.However in the present study, differences observed in the GW levels were only small between the drained sites, resulting in comparable annual N 2 O emissions.The observed cubic response to the GW level in our N 2 O model reveals maximum N 2 O emissions at a certain GW level.Perhaps it could be assumed that a mean GW level between the levels measured at the observed sites D-1 and U would result in enhanced N 2 O emissions at the present black alder forest.Indeed, Brumme et al. (1999) reported that small seasonal fluctuations of the GW level seem to be more important than the mean GW level.
The overall results from the laboratory experiment lead to the conclusion that the incorrect setting of the WFPS to 83 % only marginally affected the responsiveness of samples from site U compared to 100 % WFPS treatments as well as to the samples from site D-2.Nevertheless, we cannot totally exclude the possibility that the incorrect adjustment lowered the gaseous N losses from these samples.However, in the laboratory experiment, significantly different N 2 O emissions were not observed between the two adjusted soil moisture contents, but total denitrification was considerably higher at 100 % WFPS compared to 70/83 % WFPS for both sites.This was in line with investigations by Scholefield et al. (1997), who found a greater than 50-fold increase in denitrification activity with increasing WFPS from 70 % to 90 %.Beside soil moisture, the dependency of denitrification activity on temperature, as found for field N 2 O fluxes, becomes even more apparent regarding the results from the incubation experiment, particularly for samples from site D-2.Additionally, our results reveal that the production of N 2 increased more with increasing temperature than the production of N 2 O, which was in line with Maag and Vinther (1996).Higher denitrification activity with increasing temperature was also reported by several other studies (e.g., Dobbie and Smith, 2001;Schindelbacher et al., 2004;Schaufler et al., 2010).Butterbach-Bahl et al. (2013) reported that the Q 10 of denitrification exceeds the Q 10 of soil respiration and attributed this to a tight coupling between the microbial C an N cycles.Additionally, denitrification is indirectly affected by the temperature-induced respiratory depletion of soil oxygen concentration (Butterbach-Bahl et al., 2013).
The proportion of N 2 O emissions at 100 % WFPS amounted to ∼ 1 to 6 % and to ∼ 0.7 to 2 % of total N release from denitrification for samples from site D-2 and U, indicating that in the present alder stand losses of N 2 O are only of minor importance compared to N 2 losses at watersaturated conditions.This agrees with findings that, with increasing anoxic conditions, the percentage of N 2 increases until it becomes the major gas evolved (Rolston et al., 1978;Davidson, 1991).Indeed several studies reported that high contents of available NO − 3 result in the inhibition of N 2 O reductase activity (Blackmer and Bremner, 1978;Firestone et al., 1979;Weier et al., 1993;Regina et al., 1996).However, for the present study it is not possible to state conclusively to what extend the probably higher NO − 3 availability inhibits the conversion of N 2 O to N 2 at samples from site D-2.N 2 O fluxes only declined in D-2 samples at water-saturated conditions and 25 • C with simultaneously exponentially increasing N 2 fluxes.While this mechanism supports our hypothesis that N 2 O release is displaced by N 2 losses during periods of temporarily high water levels, the field observations never reached the point in which the decreasing N 2 O : N 2 ratio overcompensated the increasing denitrification rate.
In contrast to the field observations, significant differences in the N 2 O fluxes between the two sites investigated were found at both soil moisture contents in the incubation experiment conducted, mainly due to the distinct differences in the N 2 O reaction to temperature.Our data have shown that denitrification was limited by temperature at the drained site, but limited by other factors at the undrained site, like the availability of labile organic carbon, available NO − 3 or soil moisture (Letey et al., 1980;Knowles, 1982;Weier et al., 1993;Klemedtsson et al., 2005).We can rule out that redox conditions are too anaerobic or microbial activity is less favorable for denitrification in samples from the undrained site in the incubation experiment because no CH 4 emissions were detected from water-saturated soil cores and aerobic and anaerobic CO 2 production was comparable and showed a distinct temperature response at both sites.However, at 100 % WFPS and 25 • C, the CO 2 to N gas emission ratio was 30 for site U and 13 for site D-2, indicating that anaerobic CO 2 production at site U was significantly more decoupled from denitrification activity than at the drained sites (Wang et al., 2011).Thus we attribute the small temperature response of N 2 O and denitrification of the undrained site to NO − 3 limitation.The amount of NO − 3 in the solution from soil cores was not ascertained, but in the field net nitrification could never be detected at the undrained site.However, N 2 O fluxes measured temporarily confirm our assumption that in periods of low water level net nitrification also occurs at the undrained site, albeit at a low level.Thus, we conclude that for most of the year denitrification activity is NO − 3 limited at the undrained site, confirming our hypothesis that NO − 3 limitation in permanently waterlogged peat soils results in negligible N 2 O and N 2 losses.
For N 2 fluxes no comparable values from the field are available in the literature, but N 2 fluxes observed in the incubation experiment are in the range of other studies from drained and undrained fen ecosystems (e.g., Teiter and Mander, 2005;Wray and Bayley, 2007;Mander et al., 2008;Roobroeck et al., 2010;Soosaar et al., 2011).
In the present study less than 0.2 % of the nitrified N was emitted as N 2 O-N at the drained sites.Similar or slightly higher values were reported by Klemedtsson et al. (1988) and Maag and Vinther (1996).In the incubation experiment N 2 O fluxes that were up to four times higher were measured in samples for site D-2 compared to the field fluxes.This indicates the potential of high N 2 O emissions from the drained alder sites under different soil physical conditions.
The low N 2 O and N 2 losses, despite very high N mineralization rates, perhaps indicate that, for most of the year, aerobic or only slightly anaerobic soil conditions are predominant at the drained sites.Thus, a certain proportion of the gaseous N may have been lost as nitric oxide (NO), a byproduct of nitrification and an intermediate product of denitrification (Davidson, 1991).Several studies reported that net nitrification rates and NO − 3 in the soil solution are positively correlated to NO fluxes in well-aerated soils (Skiba et al., 1998(Skiba et al., , 1999;;Gasche and Papen, 1999;Venterea et al., 2003;Eickenscheidt and Brumme, 2012).

Conclusions
We studied N turnover processes in organic soils along a water table gradient in a nitrogen-saturated alder forest characterized by high mineral N concentrations.Drainage considerably increased net N mineralization and N turnover processes and shifted the dominant process from net ammonification to net nitrification, resulting in nitrate leaching.Surprisingly, N 2 O emissions in the field remained low at the drained sites, but we cannot rule out that we have missed N 2 O emission peaks.The incubation experiment supported the potential for high N 2 O emissions at the drained alder sites.Temporarily water-saturated conditions enhanced N 2 emissions, whereas N 2 O emissions remained unchanged.In contrast, the permanently high groundwater level at the undrained site prevented net nitrification and resulted in negligible gaseous N losses.

Figure 3 .
Figure 3. Net ammonification, net nitrification and net N mineralization rates (sum of net ammonification and net nitrification rate) for the soil depth of 0-10 cm (a) and 10-20 cm (b).Means (±SE; n = 3) followed by the same letter indicates no significant differences in the net N mineralization rate between the months within one soil depth (two-way ANOVA, TukeyHSD test at P ≤ 0.05).Differences between both soil depths are not significant.

Table 1 .
Groundwater level, water-filled pore space (WFPS), extractable nitrate (NO − 3 ) and ammonium (NH + 4 ) contents of the three sites for the year 2011.Values are given as mean with minimum and maximum in parentheses.Means followed by the same lower-case letters indicate no significant differences between the sites within the year 2011.(Pairwise Wilcoxon rank sum test with Bonferroni correction for WFPS, NO − 3 and NH + 4 at P ≤ 0.05).a Values give the mean data from the water level loggers (15 min log interval).b Values give the mean groundwater level during the gas flux measurements conducted.

Table 2 .
Soil properties of the sites.
Values give means ± standard error.* World Reference Base for Soil Resources.

Table 3 .
Estimation of available N min and NO − 3 consumption during sample storage and incubation time at 100 % WFPS.