Denitrification is not Necessarily the Main Source of N2O from Rewetted Fens

Drained agricultural peatlands are being increasingly rewetted for global warming mitigation. This creates novel ecosystems, with unclear effects on nitrogen cycling. Therefore, we aim to understand the impact of rewetting on nitrous oxide (N2O) production and its sources. Soil samples from pairs of sites differing in water regime (drained [D] and rewetted [W]) and peatland type (coastal fen [C], percolation fen [P] and alder forest [A]) in North-Eastern Germany were analyzed for microbial production pathways of N2O using the dual-isotope method with four tracers (H218O, N18O3−, 15NO3−, 15NH4+) in a laboratory incubation experiment. Unexpectedly, the largest N2O fluxes were found for rewetted sites. In four sites, denitrification dominated N2O production (80—90%). Only CW and AD displayed almost equal contributions of N2O from NO3− and NH4+, showing also largest maximum contributions of nitrifier denitrification (44–48%). Nitrification contributed less than 8% in all soils. Less than 20% of N2O was from nitrification-coupled denitrification. Soil samples with high initial water content, requiring drying prior to preincubation, displayed largest emissions, irrespective of peatland type or field water regime. Interestingly, if field conditions were dry and water was added for the preincubation, the contribution of nitrifiers to N2O production was increased, in line with larger concentrations of NO3−. The results confirm the enhancing effect of drainage on N2O fluxes. However, they also indicate a legacy effect of previous conditions on sources of N2O. Overall, short-term changes in water content had strong effects on fluxes, but not sources of N2O.


Introduction
Drainage of peatlands started a few hundred years ago for activities like agriculture, peat extraction and forestry (Joosten and Couwenberg 2001). Drainage generally leads to aerobic decomposition and thus, to greenhouse gas emissions (Canadell et al. 2007;Page et al. 2002;Wösten et al. 1997).
In Europe, peatlands account for about 5 to 6% of the land area, and more than 60% of them are drained (Drösler et al. 2008). In Germany, even 95% of peatlands are drained, causing 5% of Germany's total anthropogenic greenhouse gas emissions (46 million tons of carbon dioxide equivalents per year) (Hahn-Schöfl 2015). Of these peatland emissions, 80% originate from fens (Höper 2007): their larger nutrient contents compared to bogs made them preferred drainage targets (Timmermann et al. 2016).
Peat mineralization leads to the release of carbon dioxide and nitrous oxide (N 2 O) (Gelbrecht et al. 2008). N 2 O is a long-lived greenhouse gas with an average concentration of about 331 ppb in the atmosphere (Tian et al. 2020). In the stratosphere, its decomposition products are involved in ozone destruction (Crutzen 1991;Ravishankara et al. 2009).
There is a range of processes and pathways producing N 2 O in soils (Butterbach-Bahl et al. 2013). They can take place simultaneously in different soil microsites, making them difficult to distinguish (Heil et al. 2015;Stein 2019;Wrage-Mönnig et al. 2018) and important to understand in order to develop N 2 O mitigation strategies. In wet fens, denitrification ( Fig. 1) is usually considered to be the main source of N 2 O Lohila et al. 2010). However, especially under drained conditions, also nitrification ( Fig. 1) can contribute to N 2 O production (Martikainen et al. 1993;Regina et al. 1996). Another important pathway is nitrifier denitrification, where NO 2 − is reduced to N 2 O and potentially N 2 as in denitrification, but by autotrophic ammonia oxidizers (Kool et al. 2007;Wrage et al. 2004). Furthermore, there are various other pathways producing N 2 O, like heterotrophic nitrification, co-denitrification or fungal denitrification. Various methods exist to distinguish among these sources, but none covers all processes and pathways.
So far, the effect of rewetting on overall N 2 O emissions and on soil sources of N 2 O is not well understood. Research suggests that rewetting causes an overall reduction in N 2 O emissions (Jordan et al. 2016;Wilson et al. 2016). Thus, direct comparisons of drained and rewetted peatlands demonstrated that drained sites showed larger N 2 O emissions (Davidsson et al. 2002;Vybornova et al. 2019). However, Gelbrecht et al. (2008) observed that while rewetting of drained fens to a water table level of 0.3 -0.8 m above ground strongly decreased N 2 O emissions, a fluctuating groundwater level (-0.3 m ± 0.3 m above ground level) led to their increase (Berendt et al. 2022). Studies systematically investigating sources of N 2 O from (rewetted) fens under controlled conditions are missing.
Therefore, the aim of this study was to improve our understanding of the influence of fen rewetting on N 2 O production and its sources under controlled laboratory conditions. We incubated soil of pairs of drained and rewetted sites of three different fen types, using the dual-isotope method according to Kool et al. (2011). With this method, it is possible to distinguish among nitrification, nitrifier denitrification and denitrification as sources of N 2 O. We chose this method as we suspected that nitrifier denitrification might be important under the conditions encountered. We hypothesized that a) peat from rewetted sites would show smaller N 2 O fluxes than from drained ones, b) that the average water table height in the field would be the main influencing factor for N 2 O emissions as it determines both peat mineralization (and thus substrate availability) and microbial community composition, and c) that denitrification would be a larger source of N 2 O on rewetted sites than on drained ones.

Material
Soil (0 -20 cm) was collected from the six study sites (pairs of drained (D) and rewetted (W) sites on a coastal fen (C), percolation fen (P) and alder forest (A)) of the WETSCAPES project (Jurasinski et al. 2020) and stored cool (8-10 °C) until the start of the experiment. One week after soil sampling -which was used for preliminary tests to determine the water content and water-holding capacity (WHC) -the preincubation started. For more information about the study sites and the soil properties, see Supplementary Material and Jurasinksi et al. (2020).

Incubation Experiment
The WHC was determined for each soil according to Vengadaramana and Thairiyanathan (2012) using a funnel with filter paper (Whatman No. 1) instead of a perforated tin box.
After a two-day preincubation with 50 g soil (dry mass) in 750 ml Weck jars (n = 5) at room temperature (between 20 and 22 °C) and with a water content of 85% WHC, the main incubation was started by adding isotopic tracers dissolved in distilled water to reach 95% WHC and mixing the dissolved tracers into the soil with a glass rod. All treatments received equal amounts of mineral N in form of 7.14 mg of ammonium nitrate (NH 4 NO 3 ). These conditions were chosen as a compromise between creating comparable conditions for all sites and not changing site conditions too much, while being able to add isotopic tracers. Incubations were carried out according to the dual-isotope method (Kool et al. 2011). In brief, the method used treatments (TR) with the following isotopic tracers: H 2 18 O (TR1), N 18 O 3 − (TR2), 15 NO 3 − (TR3) and 15 NH 4 + (TR4), with the ammonium and nitrate tracers enriched at 10 at% and H 2 O enriched at 1 at%. In contrast to the initial method, the soil samples were not homogenized Fig. 1 Major pathways of N 2 O production: nitrifier nitrification, nitrifier denitrification, fertilizer denitrification and nitrification-coupled denitrification. The difference between fertilizer denitrification and nitrification-coupled denitrification is the different source of the nitrate used from either external sources or nitrification or dried in order not to destroy the peat properties, unless the peat was too wet initially: soil that had a larger water content was dried to approximately 85% WHC at room temperature before the start of the pre-incubation. This was the case in three soils: AW, PW and PD. Especially the site AW was completely flooded at sampling. The additional water was included in the calculation of water content, resulting in AW having a calculated water content of 120% WHC. The jars were closed directly after tracer addition with air-tight lids containing a septum.

Gas Measurements
At 3 h, 6 h and 24 h after tracer application, gas samples were taken with a 20 ml syringe and transferred into evacuated exetainer vials for analyses of N 2 O concentration and its isotopic enrichments. The gas samples were analyzed with a TraceGaspreconcentrator (Elementar, Langenselbold, Germany) coupled to an isotope ratio mass spectrometer (IRMS, IsoPrime 100, Elementar, Langenselbold, Germany). For calibration, we used two working standards (0.9 and 1.8 ppm N 2 O in synthetic air, δ 15 N 0.15 and 0.02‰, δ 18 O 40.66 and 40.32‰, respectively) calibrated against the standards of the laboratory of the Department of Environmental System Science, ETH Zürich (Verhoeven et al. 2019). At the time these experiments were carried out and samples were measured, no official reference materials existed for N 2 O (Mohn et al. 2022) and also no N 2 O with known enrichment in 15 N in the atom% range expected with tracer addition was available. We regularly measured isotopically enriched as well as natural abundance 15 N in solids (see below), finding the IRMS linear over this range. Therefore, we assumed linearity also for N 2 O. The working standards were run at the start and end of each run and in duplicate every 20 samples. For calibration of the sample peak ratios, an N 2 O reference gas (100% N 2 O, Air Liquide, Germany) was run with every sample. Afterwards, the ratios were corrected for drift and span via the working standards. Stability (≤ 0.01‰) and linearity (≤ 0.02‰) of the IRMS were measured by injection of 10 N 2 O reference gas pulses of similar or varying amount, respectively. Determination of external precision for 15 N in N 2 O was done using at least four samples of our 1.8 ppm N 2 O working standard per run and was on average 0.22‰.

Soil Extractions
After 24 h, soil KCl extractions (150 ml 1 M KCl per 40 g soil, 1 h shaking, filtration over Whatman No. 1 filter paper) were carried out and extracts prepared for 15 N isotopic analyses of NH 4 + and NO 3 − using microdiffusion (Brooks et al. 1989).The samples were then measured on an elemental analyzer (vario PYRO cube, Elementar, Germany) coupled to the above IRMS. The external precision for 15 N in solid samples, determined as the standard deviation of 7 to 20 natural abundance samples of sulfanilamide during one run with samples intermixed was on average over the lifetime of the used source 0.16‰. As internal standards, we used sulfanilamide and wheat flour. These were calibrated against IAEA-600 and IAEA-NO-3 for 15 N, as well as IAEA-311 for samples enriched in 15 N. Isotopic values are reported in at% excess for the tracer study.
Calculations and Statistics N 2 O fluxes were calculated based on linear regressions of the gas concentrations over time.
Calculation of sources was done according to Kool et al. (2011). According to this method, N 2 O produced from NH 4 + is divided into nitrification-coupled denitrification (NCD), nitrifier nitrification (NN) and nitrifier denitrification (ND) using 15 N and 18 O as tracers. The frequently used 15 N tracer method was not able to differentiate between pathways related to nitrification (nitrifier nitrification, nitrificationcoupled denitrification and nitrifier denitrification). Here, we also used 18 O as a tracer to quantify the O exchange in the different pathways. The method yields maximum and minimum values per pathway. In the results, we present the maximally possible amounts of these production pathways for reasons of clarity. A one-way ANOVA was used to check for differences in variables among sites (α ≤ 0.05). Data were tested for normality using the Shapiro-Wilk-Test and for equal variances with the Brown-Forsythe-Test. If the requirements for ANOVA were not fulfilled, the Kruskal-Wallis-Test was performed. The Tukey-or Holm-Sidák-Test were used as post-hoc tests. Statistical analyses were performed with SigmaPlot 13.0.

NH 4 + and NO 3 − Concentrations
In all soils, there was less NH 4 + at the end of incubations than NO 3 − (Fig. 2). With the exception of the coastal wetland, drained sites contained significantly more NO 3 − and less NH 4 + than the rewetted one (p ≤ 0.001). CD also had the largest NH 4 + concentration of all sites, 25.8 mg NH 4 + -N kg −1 (p ≤ 0.001), and CW the significantly largest NO 3 − concentration (Fig. 2), 67.5 mg NO 3 − -N kg −1 (p ≤ 0.001). As expected, in TR3 and TR4, considerable 15 N enrichments in mineral nitrogen were measured at the end of the incubations (data in the supplement). In TR3, enrichments in 15 N-NO 3 − ranged from 1.4 to 3.0 at%. There were no enrichments of NH 4 + in this treatment. TR4 showed smaller enrichments of 0.6 -1.4 at% for 15 N-NH 4 + . Furthermore, enrichments in 15 N-NO 3 − of between 0.5 and 1.1 at% were also detected in this treatment (data in the supplement).

N 2 O Source Determination
All sites produced at least half of the N 2 O from labelled NO 3 − , i.e. from denitrification (Fig. 4). Interestingly, rewetting produced no clear patterns concerning the contribution of the different sources to N 2 O production. AD and CW produced the smallest amount of N 2 O from NO 3 − (with 52.3 ± 17.1% and 56.1 ± 14.0%, respectively) compared to the other sites (p = 0.001 -0.008). The contribution of denitrification to N 2 O production was almost identical for both sites of the percolation fen (84.4 ± 11.4% in PW; 85.8 ± 4.9% in PD, respectively) and AW (81.0 ± 7.4%). These values were not significantly different (p = 0.841 between PW and PD; p = 0.590 between PW and AW, and p = 0.269 between PD and AW). In CD, N 2 O was formed almost entirely from NO 3 − under the conditions tested, representing with 90.2 ± 5.2% the largest contribution and showing significant differences to CW and AD (p = 0.001), but not to the other sites (p = 0.054 -0.690).
The largest maximum contributions of ND to the production of total N 2 O were estimated for CW (43.9 ± 14.0%) and AD (47.7 ± 17.1%) (p = 0.710, Fig. 5). At the remaining four sites, the maximum amounts of ND were between 10 -20% (p = 0.054 -0.841), with significant differences to CW and AD (p ≤ 0.001 to p = 0.009).
At CW, the maximal contribution of NCD was equal to that of ND and significantly larger than that of all other study sites (p ≤ 0.001 to p = 0.008). The smallest maximal contributions of NCD were calculated for CD and AD (4.4 ± 7.6%, 7.6 ± 10.8%, respectively), showing a significant difference between CD and PD (p = 0.042), with PW, PD and AW having intermediate values for the maximal contribution of NCD (Fig. 5).
NN did not contribute to N 2 O production from CW, PW and PD under the conditions tested (Fig. 5). For the other sites, the maximally possible contributions from NN were  also small (5.4 ± 3.6%, 3.8 ± 5.6%, 7.6 ± 10.8%, respectively), showing no significant differences among sites (p = 0.503 -0.841).

Discussion
In contrast to the first hypothesis, N 2 O fluxes were larger from the rewetted sites than from the respective drained ones (Fig. 3). This was remarkable, as considerably larger fluxes are normally expected from drained peatlands than from wet ones (Augustin et al. 1998). In this experiment, however, all soils were incubated at the same water content, making adaptations in water content necessary at the beginning of the incubation. When comparing the change in water content between field conditions at sampling and the start of the incubation, it is striking that the sites that had to be dried before the incubation all showed substantive N 2 O emissions (Fig. 7a). This reinforces that drainage increases N 2 O fluxes, even if some water was added again to start the incubation. The N 2 O fluxes of the other sites, where hardly any drying was required or even a considerable amount of water had to be added, were almost negligible. This indicates that further wetting of the soils did not lead to larger N 2 O fluxes, but drying of the soils just before the addition of water, i.e. quick reduction in water content, did. This is in line with studies showing increasing N 2 O emissions with fluctuating water regimes (Gelbrecht et al. 2008;Jørgensen and Elberling 2012) and suggests that drying causes the onset of emissions, even lasting into concurrent wetter conditions. This is important for the management of rewetted sites, where fluctuating water regimes are more usual than in pristine fen peatlands (Kreyling et al. 2021). For further information on calculations, see text When regarding the water table level of the sites previous to sampling (Fig. 6), it was evident that although rewetted sites usually had a higher water table level than drained ones, seasonal fluctuations were large, in line with other findings on fens (Kreyling et al. 2021). The largest fluctuations in the water table level were found on AD, where it dropped to more than -2.5 m in summer 2019 (Fig. 6), reflecting the drought conditions in that year. However, other sites also showed large variations in water table level among the seasons, fluctuating up to 1 m. Even on all rewetted sites, the water table level was more than 0.25 m below the surface in summer. Based on the findings of this current incubation study, such drying could cause increased fluxes of N 2 O. In field measurements at those sites, larger N 2 O emissions were measured particularly at PW and AW in summer 2018 (Berendt et al. 2022). At that time, the water table level for AW was more than 60 cm below surface, resulting in large N 2 O emission during that season.
Despite the low water level in the field and rewetting of the soil before incubation, emissions from AD and CW (as well as CD, were no change in water content had to be carried out) were very small. This is remarkable, as many studies showed large emissions from drained sites (Augustin et al. 1998;Merbach et al. 2001). Nevertheless, there are also some studies that reported small fluxes from drained alder sites (Eickenscheidt et al. 2014). Based on the results seen here, short-term decreases in water content seem to be more important for N 2 O emission events (Dinsmore et al. 2009; Jørgensen and Elberling 2012) than long-term site conditions, even if substantial overall changes in water content occur over time.
In order to incubate all soils under the same conditions, we used a moisture intermediate between all soils, meaning some soils had to be air-dried and others wetted for preincubation. As our results show, the soils that had to be dried the most showed the largest N 2 O emissions. It is likely that the soils would have produced considerably smaller N 2 O emissions without prior drying. This is a methodological effect, but it also shows the large impact of short-term drying of fens on N 2 O production.
Here, we only concentrated on some production pathways of N 2 O (denitrification, nitrifier nitrification, nitrifier denitrification and nitrification-coupled denitrification) based on the dual isotope method chosen. Nevertheless, there are many other pathways that can produce N 2 O. So far, these are not captured by the present methods and efforts should be taken to find a method that differentiates all known major sources of N 2 O, potentially in a combination of isotope approaches.
Denitrification was an important source of N 2 O, but not in all cases more important in rewetted than in drained sites (Fig. 4). In contrast to our second hypothesis, the largest contribution of denitrification with more than 90% was found for the drained site CD. When comparing the change in water content before the start of the incubation with the amount of N 2 O from denitrification (Fig. 7b), it was noticeable that the sites that had to be dried before incubation (AW, PW and PD) showed a larger amount of denitrification. Since these sites were also very moist before incubation, they were probably tending towards denitrification (Lohila et al. 2010). In contrast, soil samples of the sites CW and AD were drier than 85% WHC before the start of the preincubation. These two sites produced smaller amounts of N 2 O from denitrification. Thus, the soils seem conservative in the main source of N 2 O, despite short-term changes in conditions before the incubation. This is in line with results from an acidic fen experimentally dried or flooded, which did not show large reactions to experimental conditions in terms of N 2 O production or denitrifier community structure (Palmer et al. 2016). Another recent study showed that the predominant N 2 O production pathway of a (mineral) soil  (Ji et al. 2020). Thus, a (molecular) fingerprint of the dominant N 2 O production pathway(s) of a soil might help to better understand its behavior in changing conditions.
The sites CW and AD showed small NH 4 + concentrations in relation to NO 3 − , with almost half of the N 2 O produced from NH 4 + . The small NH 4 + concentrations indicate that nitrification was fast here in relation to mineralization, and also to denitrification, as large NO 3 − concentrations as well as 15 N enrichments of NO 3 − in incubations with added 15 NH 4 + suggest. Since N 2 O emissions were relatively small in CW and AD, either little N 2 O was produced or the N 2 O produced was largely reduced to N 2 . For CW, N 2 O produced from NH 4 + could originate from either ND or NCD. In contrast, up to 50% of the N 2 O produced in AD from NH 4 + originated from ND. These results were very surprising since we expected that most of the N 2 O in peatlands would be produced via denitrification at a WHC of 95%. Interestingly, there are other studies reporting a remarkable contribution of nitrification at 80% water-filled pore space (Pihlatie et al. 2004), but most of the studies showed that denitrification was the dominant process of N transformation in the soil under water-saturated conditions (Wolf 2000).
Again, the predominant conditions and thus predominant microbial pathways might play a role here. Probably, the dry conditions in the field prior to incubations led to this large contribution of nitrification processes, even after over 24 h at wetter conditions. Even at water contents of 95% WHC, peat soils can still have dry pores, as pores can be very large, draining quickly, making peat a dual-porosity medium (Rezanezhad et al. 2016). Large contributions of nitrifiers to N 2 O production in rewetted fens were also shown by Masta et al. (2022). In all soils studied here, either ND or NCD could explain N 2 O production from NH 4 + , with negligible potential contributions of NN as these showed no or extremely small contributions (smaller than 10%) to the production of N 2 O (Fig. 5). Thus, pure nitrification does not seem to play a large role for N 2 O production in these soils.

Conclusion
Our results suggest that contrary to our hypothesis, a categorization into drained and rewetted fen sites cannot be used as an indicator for the microbial production pathways of N 2 O: as largest contributions of denitrification to N 2 O production were observed on a drained site. Short-term reductions in water content immediately prior to incubation resulted in largest N 2 O emissions, not rewetting of soil that had been comparatively dry in the field for a longer time. Thus, such quick drainage appears to stimulate N 2 O production more than lower long-term water table levels. Interestingly, all sites showed contributions to N 2 O production from both nitrification and denitrification processes, with water addition to field-dry peat soils leading to large contributions of nitrification pathways to N 2 O emissions.
Interestingly, although short-term changes in water content overruled longer-term conditions in the field in terms of N 2 O fluxes, its sources were determined by longer term conditions and predominant microbial communities. This is interesting for the management of rewetted peatlands: It could enable a fingerprint of microbial communities to help predict N 2 O dynamics and develop an informed management of rewetted peatlands. For this, the stability of such communities over time needs to be investigated. Furthermore, the results underline that short-term changes in water content of rewetted peatlands need to be reduced to minimize N 2 O emissions.