Soils were collected from three field sites of the Terrestrial Environmental Observatory (TERENO) (www.tereno.net) from the Eifel/Lower Rhine Valley, Germany, including a coniferous forest (Wüstebach; 50° 30′ 10″ N, 6° 19′ 50″ E), an extensive grassland (Rollesbroich; 50° 37′ 18″ N, 6° 18′ 15″ E) and an arable land (Selhausen; 50° 52′ 10″ N, 6° 27′ 4″ E). The coniferous forest site is located in the low mountain ranges of the Eifel National Park, 630 m above mean sea level, with a tributary of the river Rur basin flowing through it. The site is dominated by Norway spruce (Picea abies (L.) H. Karst). The hillslopes are characterized by Cambisol and Planosol, whereas the riparian zone is dominated by Gleysol and Histosol. The texture of soil in this forest is silty clay loam. The mean annual rainfall is about 1,400 mm, and the mean annual temperature is around 7°C. The vast majority of the precipitation occurred in the form of rain. The grassland site is located in the Northern Eifel region with smooth meadow grassland. Dominant soils at this site are (gleyic) Cambisol, Stagnosol, and Cambisol-Stagnosol with a silt loam texture. The climate is temperate maritime with a mean annual temperature and rainfall at the grassland site of 7.7°C and 1,033 mm, respectively, for the period from 1981 to 2000. The agricultural site is dominated also by (gleyic) Cambisol and (gleyic) Luvisol with a silt loam texture, and regularly cultivated with sugar beet, wheat, and oilseed rape, depending on the year. Mean annual temperature and rainfall at the cropland site are 9.8°C and 690 mm, respectively.

For grassland and arable land, mixed soil samples from five points for each site (~1.5 kg each) were collected from an area of 0.5 hectare from the top 15 cm in January 2016. Due to the strong spatial heterogeneity of soil properties at the forest site (Liu et al., 2016), six soil samples (~3 kg) were collected in January 2016 from the humus-rich layer (Oa horizon, depth 3–5 cm) of five sampling points (F1, F2, F3, F4, and F5) of the forest upland area and in addition from one sampling point of the forest riparian zone (FR) within an area of ~27 hectare of the forested Wüstebach catchment. The riparian zone represented only a small part of the forest, therefore we did not consider spatial heterogeneity of the riparian zone in contrast to the whole forest, and sampled only one point in the riparian zone area. Soil samples were transferred to the laboratory at the day of sampling. In the laboratory, fresh samples (except for the riparian zone sample) were passed through a 2-mm sieve, and coarse plant residues (including roots) and stones were manually removed.

For the anoxic pre-treatment, about 600 g fresh soil from each sampling site, stored at 4°C for 3 days after sampling in open plastic bags, was put into a 1 liter glass bottle and sealed with a rubber plug within a plastic lid. The gravimetric water content (w/w, water/dry soil) of the fresh soils was around 59–108% for the forest, 22% for the grassland, and 10% for arable land. The bottles were then evacuated for 10 min and refilled with helium to 0.4 bar overpressure. This procedure was repeated three times. Then the bottles were incubated with helium as headspace gas at ambient pressure at room temperature for 1 week. For the oxic pre-treatment, another ~600 g fresh soil was put in large open plastic bags and kept under oxic conditions at room temperature for 1 week. All plastic bags were stored in a large plastic box to reduce air flow and further reduce soil water evaporation. All soil samples were freeze-dried immediately after the oxic/anoxic pre-incubations to preserve the chemical status of the soil samples until further treatment. One side effect of freeze-drying could have been that this process led to a disruption of soil aggregates, which would have made more of the substrates soluble when the solution was added, and would have led to an overestimation of the N₂O production from NO2- (via both biotic and abiotic pathways), but an underestimation of the N₂O production from NH₂OH (via abiotic processes). After freeze-drying, half of each oxic and anoxic pre-incubated soil was transferred to 50-ml falcon tubes and sterilized with γ-irradiation (Best Theratronics, Canada) for 14 h (total dose: 11 kGy), and the other half of each soil was kept in ziplock bags at room temperature (21 ± 1°C). The success of the sterilization process was checked by plating soil slurries after the sterilization on R2A medium and incubated for 24 h at 25°C. No growth of bacteria or fungi was observed (data not shown).

About 1.4 g of freeze-dried soil with and without γ-irradiation were weighed into 22-ml gas chromatography (GC) vials (VWR International, Darmstadt, Germany), followed by the addition of either H₂O, or NO2- and NH₂OH solutions to reach around 40% water-holding capacity (WHC) and 1 μg N g⁻¹ dry soil (for NO2- and NH₂OH). The water was added to reactivate microbial activity (Morillas et al., 2015). The NO2- concentration in the freeze-dried soil was around 0–0.3 mg kg⁻¹ dry soil. The added N amount corresponded to NO2- content in soil with fertilizer application (Shen et al., 2003; Venterea et al., 2003), and was assumed also realistic for NH₂OH in soils with fertilizer application, as concentrations of 0.3–34.8 μg N kg⁻¹ dry soil had been observed at this forest site (Liu et al., 2014). Solid MnO₂ (Merck, Darmstadt, Germany) was added to the soil with NH₂OH addition to explore the effect of MnO₂ on NH₂OH-to-N₂O conversion in soil with both oxic and anoxic pre-treatment. The added Mn amount amounted to 0.1% (w/w) of soil dry weight, while the natural Mn content of the soil samples of this study ranged between 0.015 and 0.194% (w/w) (Table 1). Each treatment was carried out in triplicate. In total, there were 96 vials used for each soil. All vials were closed gas-tight immediately after addition of the N solution and MnO₂ with butyl septa and aluminum crimp caps (VWR International). Half (48) of the vials were incubated aerobically at room temperature (21 ± 1°C) for 1 h, and the other half (48) were incubated aerobically for 7 h.

The gas in the headspace of the sample vials was analyzed for N₂O using a gas chromatograph (Clarus 580, PerkinElmer, Rodgau, Germany) equipped with an electron capture detector (ECD) and flame ionization detector (FID) for N₂O and CO₂ detection, respectively. The N₂O emission rate was calculated according to the following equation:

where E is the N₂O emission (ng N g⁻¹ dry soil); the factor two is used as a constant to the ratio of N₂O-N to N₂O; C' is the N₂O mixing ratio in the vial headspace (nL L⁻¹); V is the volume of vial headspace (L); V_m is the molar volume of N₂O at standard pressure and room temperature (L mol⁻¹); M is molar mass of nitrogen (g mol⁻¹); W_ds is the mass of the dry soil (g). The instrument was calibrated each day using five different standard gases with 0.25, 0.50, 0.75, 1.00, and 5.00 μL L⁻¹ N₂O, balanced with N₂ (99.999% purity, Linde, Munich, Germany).

Chemical properties of the soils with oxic and anoxic pre-treatment were analyzed before the incubation. The elemental composition of the organic materials was analyzed by using inductively coupled plasma optical emission spectrometry (ICP-OES). Briefly, 100 mg of sample material were mixed with 3 ml HNO₃ and 2 ml H₂O₂, heated in the microwave at 800 W for 30 min. The mixtures were subsequently filled up to 14 ml and diluted 10-fold with deionized water followed by the ICP-OES measurement. Total C and N contents were determined with an elemental analyzer (vario EL Cube, Elementar Analysensysteme GmbH, Langenselbold, Germany).

In addition, mineral N and the quality and quantity of soil DOM were analyzed to determine the effects of anoxic pre-treatment on the DOM dynamics. The mineral N (NH4+ and NO3-) contents were analyzed with ion chromatography (ICS-3000 for NO3-, DX-500 for NH4+; Dionex, Sunnyvale, CA, USA). NH4+ and NO3- were extracted with 1 M KCl (dry soil: solution = 1:10 w/w) and shaken for 24 h. Soil pH was measured by shaking soil with 1 M KCl (dry soil: solution = 1:10 w/w). NO2- was extracted according to the method of Homyak et al. (2015) and measured by ion chromatography. NH₂OH was extracted and measured according to the method of Liu et al. (2014). Dissolved organic carbon (DOC) and dissolved total nitrogen (DTN) were extracted with deionized water (dry soil: water = 1:2.5 for grassland and arable land soils, and 1:5 for forest and riparian soils) by shaking for 1 h at 200 rpm. DOC and DTN were then analyzed with a TOC-TN analyzer (Shimadzu Corp., Kyoto, Japan). In addition, for characterization of aromatic substances the absorbance of the DOC extract at 254 nm (A₂₅₄) was determined with UV-VIS spectrometry (DU 800, Beckman Coulter, Inc., United States) and a path length of 1 cm.

The effects of NO2- and NH₂OH on N₂O emission were calculated by subtracting N₂O emission after water addition only (as control) from the N₂O emission in response to NO2- and NH₂OH addition. Contribution of abiotic processes to formation of N₂O from NH₂OH and NO2- was calculated as follows:

where R means the contribution (%) of abiotic conversion of NH₂OH to N₂O and of NO2- to N₂O; E₁′ (ng g⁻¹) represents the N₂O production after NH₂OH or NO₂- addition to soil after γ-irradiation; E₀′ (ng g⁻¹) represents the N₂O production after H₂O addition to soil after γ-irradiation; E₁ (ng g⁻¹) represents the N₂O production after NH₂OH or NO₂- addition to soil without γ-irradiation; E₀ (ng g⁻¹) represents the N₂O production after H₂O addition to soil without γ-irradiation.

Analysis of variance (ANOVA) was used to test the main factors, i.e., soil type, oxic, and anoxic pre-treatment, N addition and γ-irradiation, and their interactions for significance (P < 0.05) of their effect on N₂O production using the R software package (version 3.4.3). Box-Cox transformation of N₂O data was performed before the ANOVA test. Fisher's Least Significant (P < 0.05) Difference test was used to test means of the effects for significant (P < 0.05) differences. Spearman's rank correlation analysis was performed between variables C, N, C/N, Fe, Mn, pH, DOC, DTN, A₂₅₄, NH4+ and NO3- for the oxic and anoxic pre-treatment using Origin Pro V. 2015.

NO2- addition increased N₂O production in the grassland soil significantly (P < 0.05), whereas it had only minor effect in the other soil samples (Figure 1A). In the grassland soil with oxic pre-treatment, 30.4% of the NO2- had been converted to soil N₂O within 7 h, assuming that all the N₂O came from the added NO2-. Even with water addition only, grassland soil which was pretreated under oxic conditions showed a strong rewetting effect, with a large N₂O production of 512.1 μg N kg¹ dry soil after 7 h (Table S1, supplementary information). For the forest soils, the N₂O formation after NO2- addition amounted to about 40 μg N kg⁻¹ dry soil after 7 h, which was 13% of the grassland soil N₂O production. In general, the N₂O production after NO2- addition from the riparian and arable soils after oxic pre-treatment was significantly (P < 0.05) lower than from the other soil samples.

In contrast, NH₂OH addition induced the highest N₂O production in the arable soil, followed by the grassland soil and the forest soil from sampling point F5 for soil samples with oxic pre-treatment. The conversion of NH₂OH to N₂O ranged from 12 to 47% for the arable, grassland and F5 forest soils within 7 h, assuming that all the N₂O came from the added NH₂OH. NH₂OH addition had only a minor effect on the other forest soil samples during the whole incubation period (Figure 1B). Compared to the effect of NO2-, NH₂OH had a larger effect on N₂O production in the arable soil and in forest soil F5. Moreover, N₂O was produced very quickly in the first hour after NH₂OH addition.

Anoxic pre-incubation increased soil NH4+ concentration up to 7-fold, with the largest NH4+ concentration (195 mg N kg⁻¹ dry soil) in the grassland soil (Table 1), and decreased NO3- concentration in most of the soil samples (except forest sample F5) to nearly zero. The quality (reflected in the A₂₅₄ value) and quantity of DOM (reflected in the concentrations of DOC and DTN) varied substantially after anoxic pre-incubation between the different soil samples, with 5–42% higher DOC content compared to soil samples with oxic pre-incubation. The A₂₅₄ value followed a trend very similar to DOC, indicating that more aromatic substances were available in dissolved form after anoxic pre-incubation. The difference in DTN between the different treatments was not as pronounced as for DOC and A_254.

Anoxic pre-treatment of the soil samples had a significant (P < 0.05) effect on soil N₂O emission after the addition of NH₂OH and NO2-. For the grassland and arable soil, anoxic pre-incubation resulted in a significant (P < 0.05) increase of N₂O production during the first hour after NO2- addition (Figure 1C, Table 2). For the forest soil, anoxic pre-incubation increased N₂O production from samples F1, F3, F5, and FR after NO2- addition compared to the oxic pre-treatment, but decreased the effect of NO2- on N₂O production in F2 and F4. In terms of NH₂OH, anoxic pre-treatment had a negative effect on the N₂O production after NH₂OH addition in most of soil samples, especially in those with a large NH₂OH effect after oxic pre-incubation, i.e., grassland and F5 forest soils, but had a relatively small, but significant (P < 0.05) effect on the N₂O production after NH₂OH addition in arable soil (Table 2). Anoxic pre-incubation decreased N₂O production only by 12% in arable soil, but by 79% in grassland soil after NH₂OH addition after 7 h of incubation (Figure 1D, Table 2).

Abiotic pathways contributed to 9–39% of soil N₂O production within 7 h after NO2- addition from the different soils after oxic pre-incubation, but contributed to 73–93% of soil N₂O production within 7 h after NH₂OH addition in the arable land, grassland and the F5 soil (Figure 2, Table 3). For the soil samples with anoxic pre-incubation, abiotic pathways contributed to 14–49% of NO2--induced N₂O production after 7 h, but contributed to 67–99% of N₂O production only after NH₂OH addition in the grassland, arable, and F5 soil in the same time period. In general, abiotic pathways played a more important role in N₂O production after NH₂OH addition compared to NO2- addition in the tested soils.

Correlation analysis showed that soil Mn, C, N, DOC, and A₂₅₄ content as well as pH were important variables responsible for soil N₂O formation from NH₂OH in the tested soils (Table 4). The NH₂OH-to-N₂O conversion was positively and significantly correlated with soil Mn content and pH, but negatively and significantly correlated with soil C, N, and DOC content, and A₂₅₄. Soil N₂O production after NO2- addition was found to be significantly (r = 0.93, P < 0.05) correlated with NH4+ content, and only marginally (r = 0.69, P = 0.06) correlated with soil Fe content after anoxic pre-treatment. No significant correlation was observed between the NO2--to-N₂O conversion and any soil properties after oxic pre-treatment.

The addition of MnO₂ increased the NH₂OH-to-N₂O conversion in all soil samples after oxic or anoxic pre-incubation (Figure 3). However, addition of MnO₂ increased the NH₂OH-to-N₂O conversion more in the soil with oxic pre-incubation (especially during the first hour after NH₂OH addition) compared to the soil with anoxic pre-incubation. Only N₂O emission from soil F3 with anoxic pre-incubation was largely affected by the addition of MnO₂ (as high as 2.5 mg kg⁻¹ N after 7 h), which disappeared completely after γ-irradiation (data not shown). The NH₂OH-to-N₂O conversion of the grassland soil, F5 and other forest soil samples after anoxic pre-incubation also increased after MnO₂ addition, but was still much lower than with NH₂OH addition only after oxic pre-incubation.

Our results showed that the role of NO2- and NH₂OH in N₂O production was strongly dependent on soil properties. The by far largest amount of N₂O was produced in non-γ-irradiated grassland soil after NO2- addition, much higher than in all other soils (Figures 1A,C). About 30% of the added NO2- was converted to N₂O in the grassland soil with oxic pre-incubation after 7 h of incubation, assuming that all the N₂O produced came from the added NO2-. This large and quick N₂O pulse could easily lead to the assumption of abiotic processes, e.g., chemodenitrification, being responsible for the N₂O production upon addition of NO2-. However, γ-irradiation decreased the N₂O production after NO2- addition to soil with oxic pre-incubation by 72% (Figure 2, Table 3), indicating that biotic pathways played a more important role in N₂O production after NO2- addition compared to abiotic processes. It was reported that nitrifier denitrification involving biological NO2- reduction can play an important role in soil N₂O emissions, especially in grassland and arable soil with large nitrifier activity (Wrage et al., 2001, 2004). Thus, the large pulse of N₂O production with NO2- addition in the grassland soil could be due to nitrifier denitrification. In contrast, the smaller effect of NO2- addition on N₂O production in forest soils was at first unexpected, as there was more carbon available in the forest soils than in the grassland soil for biotic (denitrification) and abiotic (chemodenitrification) pathways leading to N₂O production. One possible reason responsible for the smaller N₂O production in the forest soils could be that more NO instead of N₂O was produced as it mostly decomposes to NO and NO₂ at low pH (Davidson, 1992; Venterea et al., 2005). This assumption is supported by the fact that in our study the forest soil was acidic with pH values lower than 3.5 for most of the samples and more NO instead of N₂O was produced in this forest site during our former research (Wei et al., 2017).

In contrast to NO2-, large N₂O production was observed after NH₂OH addition to the arable land, grassland, and the forest soil sample F5, and only negligible amounts of N₂O were produced in the other forest soils (Figures 1B,D). Abiotic reactions played a much larger role in the case of NH₂OH addition compared to NO2- addition. Comparison of the results from the experiments with γ-irradiated and non-irradiated soils revealed that most of the N₂O from NH₂OH was chemically produced, contributing 73–93% to the total conversion of NH₂OH to N₂O for the arable, F5 and grassland soil. We found larger Mn content in soils of the grassland, the arable land and the F5 forest sub-sample (Table 1), and a positive and significant correlation was observed between soil N₂O production in response to NH₂OH addition and Mn content (Table 4). This is in accordance with previous findings, which identified the chemical reaction between MnO₂ and NH₂OH as important factor for abiotic N₂O production in soil (Bremner, 1997; Heil et al., 2015). The contradictory observation that the forest soil with the largest Mn content (F5) had a lower N₂O production upon NH₂OH addition than the grassland and cropland soils can be explained with the inhibitory effect of SOM on the abiotic conversion of NH₂OH to N₂O, as the effect of NH₂OH on soil N₂O emissions had been found earlier to be related to SOM quantity, quality and Mn content, with larger NH₂OH-to-N₂O conversion ratios in soils with higher Mn content and lower soil organic C content or, more specifically, lower content of carbonyl groups to which NH₂OH could bind chemically (Liu et al., 2017b).

Despite their reactivity, the two N intermediates NH₂OH and NO2- may accumulate in soils under anoxic conditions. NH₂OH accumulation in anoxic sediment slurries has been observed in a preliminary experiment (Figure S1, supporting information), while transient NO2- accumulation as well as the absence of NO3- have been reported in soil slurries during anaerobic incubation (Clément et al., 2005). In the present study, the NH4+ concentration, especially of the grassland soil, increased largely with anoxic pre-incubation (Table 1). Soil organic matter quality, transition metal redox state, and pH may change remarkably at this low redox potential (Dassonville and Renault, 2002). According to thermodynamic theory, the following sequential reduction of electron acceptors is observed with decreasing redox potential: O₂, NO3-, MnO₂, Fe₂O₃, SO42-, and CO₂ reduction (Froelich et al., 1979) during respiratory or other dissimilatory processes. Thus, Fe²⁺, Mn²⁺, and fermented organic matter would accumulate during anoxic pre-incubation. Furthermore, the transient occurrence of reactive C substances could have reversed the effects of NH₂OH and NO2- addition, as a transient increase in reactive C rich in carbonyl groups would preferentially react with NH₂OH and decrease N₂O production from abiotic conversion of NH₂OH, while a transient increase in reactive phenolic compounds would lead to preferential reaction with NO2- to produce N₂O chemically.

We hypothesized that anoxic pre-incubation would lead to higher N₂O release after NO2- addition and less N₂O release after NH₂OH addition due to the accumulation of more reduced substances. Our results suggest that anoxic pre-incubation increased N₂O production after NO2- addition in most of the soils. The stimulatory effect of anoxic pre-incubation on N₂O production could be due to the increased contribution of N₂O production via chemodenitrification, as more reduced transition metal ions, e.g., Fe²⁺, may accumulate after anoxic pre-incubation, which could have explained the observed very fast abiotic N₂O production. However, since most of the N₂O produced after NO2- addition came from biotic pathways, anoxic pre-incubation may have increased denitrification activity by stimulating denitrifier growth during the anoxic phase, leading to a positive effect of anoxic pre-incubation on N₂O production via NO2- in soils.

Anoxic pre-incubation had an even more pronounced effect on N₂O production after NH₂OH addition, with a significant reduction of N₂O production in the forest soil F5 and the grassland soil (97 and 79%, respectively), in accordance with our hypothesis, but with only a small effect (12%) on the arable soil after 7 h. As a strong oxidant, most of the MnO₂ should have been reduced to Mn²⁺ during the anoxic pre-incubation period according to the large increase in NH4+ (Table 1), which may indicate a soil redox potential that was lower than +200 mV (Froelich et al., 1979). In those soil samples with high C content, the organic carbon can be used by microorganisms that reduce Fe³⁺ or Mn⁴⁺ instead of oxygen (Lovley et al., 2004). The lower effect of the anoxic pre-treatment on the conversion of NH₂OH to N₂O in the arable soil could be attributed to the lower C content in this soil, where less Mn⁴⁺ would be reduced to Mn²⁺. To further explore the effect of MnO₂ on the NH₂OH-to-N₂O conversion ratio, we added 0.1% (w/w) MnO₂-Mn (equal to the Mn content of the grassland soil) to the soil samples with both oxic and anoxic pre-treatment. We hypothesized that this amount of MnO₂ addition would increase the NH₂OH-to-N₂O conversion ratio of the soil samples pre-incubated under anoxic conditions. However, only the grassland soil and forest soils FR and F5 showed a larger Mn effect after anoxic pre-treatment (by comparing Figures 1D, 3B), but the added Mn amount could not make up the reduction in N₂O production caused by the anoxic pre-incubation, despite the large increase of N₂O production from F3 with anoxic pre-incubation (comparing Figures 1B, 3B).

It was reported that large amounts of fermented substances could accumulate during anoxic incubation (Dassonville and Renault, 2002). In the present study, we found more DOC and dissolved aromatic substances (represented as A₂₅₄) in the soil samples with anoxic pre-incubation than with oxic pre-incubation (Table 1). The change of soil DOC quality and quantity could be responsible for the difference in N₂O production after NH₂OH addition to soils with different redox history, as the increase in DOC and aromatic substances after anoxic pre-incubation would increase the likelihood of fast binding of NH₂OH to organic compounds once added to the soil, and lead to a lower availability of NH₂OH for the reaction with MnO₂ to produce N₂O. Therefore, the absence of a MnO₂ addition effect on the NH₂OH-to-N₂O conversion ratio could be due to the accumulation of fermented substances that can quickly react with NH₂OH.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.