Direct and Indirect E ﬀ ects of Urease and Nitriﬁcation Inhibitors on N 2 O-N Losses from Urea Fertilization to Winter Wheat in Southern Germany

: Urea (U) is the most important nitrogen (N) fertilizer in agriculture worldwide, and as N fertilizer can result in large gaseous losses of NH 3 and N 2 O. Thus, urease inhibitors (UIs) and nitriﬁcation inhibitors (NIs) have been coupled with U fertilizers to mitigate NH 3 and N 2 O emissions. However, it is still unclear whether adding NIs and / or UIs to U stimulates other pollutants, while reducing one pollutant. Furthermore, part of the NH 3 deposition to earth is converted to N 2 O, leading to indirect N 2 O emission. To estimate direct and indirect e ﬀ ect of UIs and NIs on the N 2 O-N and NH 3 -N losses from U; therefore, we analyzed multi-year ﬁeld experiments from the same site during 2004 to 2005 and 2011 to 2013. The ﬁeld experiments with U fertilization with or without UI (IPAT, N-isopropoxycarbonyl phosphoric acid triamide) and NI (DCD / TZ, Dicyandiamide / 1H-1, 2, 4-Triazol) in winter wheat and with calcium ammonium nitrate (CAN) were conducted in southern Germany. Fluxes of NH 3 or N 2 O emissions were determined following each split N fertilization in separate experiments on the same site. Our results showed that U with NIs considerably reduced N 2 O emissions, and adding UIs decreased NH 3 emissions. However, the e ﬀ ect on N 2 O emissions exerted by (U + UIs) or (U + UIs + NIs) was inconsistent. In contrast to the treatment of (U + UIs + NIs), the addition of NIs alone to U stimulated NH 3 emission compared to treatment with U. When 1% indirect N 2 O emission from NH 3 (IPCC emission factor (EF4)) was considered to estimate the indirect N 2 O emission, total N 2 O emissions from (U + NIs) were approximately 29% compared to that from U alone and 36% compared to that from (U + UI), indicating that indirect N 2 O emission from NH 3 induced by NIs may be negligible.


Introduction
Over the past century, a large quantity of mineral nitrogen (N) fertilizer was produced using the Haber-Bosch process that converts atmospheric N 2 to ammonia (NH 3 ) in order to meet the rapid increase in population for food production [1]. In mineral N fertilizer production, the use of urea (U) as an N fertilizer has dramatically increased during the past decades due to its advantages of a high N content, low cost, and safe and easy transportation. Urea is now the most important N fertilizer in the world, accounting for 55% of mineral N fertilizer production in 2018, and U is expected to increase to 70% during the next five years [2]. However, there are problems resulting from large gaseous losses using U as an N fertilizer, which are due to the increase in pH resulting from U hydrolysis.
In Germany, U is currently the second most used N fertilizer after calcium ammonium nitrate (CAN) with a market share of 25% [3]. Although there is consensus regarding the higher ammonia emissions resulting from U fertilization compared to those of CAN, there are contrasting reports regarding N 2 O losses of these fertilizers [4]. Thus, there is a need to further compare N 2 O emissions from U to those of CAN.
To mitigate NH 3 and N 2 O emissions from U fertilization, urease and nitrification inhibitors have been coupled with U fertilizers. Urease inhibitors (UIs) slow the hydrolysis of U into NH 3 or NH 4 + .
In addition to incorporating U into soil, the addition of UIs is another best practice to avoid NH 3 volatilization, and had a significant effect on NH 3 emissions and reduced NH 3 losses by 26-83% and 24-32%, respectively, of the applied urea-N on average [5,6]. Starting from 2020, the new German fertilizer ordinance requires the application of either UIs or the immediate incorporation of U into soils.
The main working principle of UIs is slowing down the hydrolysis of U, and thus giving more time for the U to increase its dispersion into the soil [6]. Nitrification is a key process of N transformation with two steps: the ammonia oxidizing bacteria such as Nitrosomonas or Nitrosospira convert NH 4 + to nitrite (NO 2 − ), and then the nitrite oxidizing bacteria such as Nitrobacter or Nitrospira convert NO 2 − to NO 3 − [7]. NIs act by depressing the activity of the principal exponent of this first step (Nitrosomonas spp.); thereby, delaying the oxidation process.
Depending on the soil conditions and seasons, UIs or NIs can delay hydrolysis of U or nitrification of NH 4 + from week to months.
The Intergovernmental Panel on Climate Change IPCC [8] has recommended nitrification inhibitors (NIs) as a potential mitigation option for agricultural N 2 O emission. Recent meta-analyses have indicated that NI decreased N 2 O emissions by 31-48% across diverse agricultural ecosystems, including upland, grassland, and paddy fields [9,10]. The use of NIs, however, prolongs the retention of NH 4 + in soil, which could increase NH 3 emission [11]. The application of the NI dicyandiamide (DCD) has been found to increase NH 3 emissions by 18-29% [12,13] and by up to 38% [14]. However, Ni et al. [5,15] reported that there was no effect of NIs on NH 3 emissions from U fertilization. Therefore, there is now a concern regarding if (U + UIs) or (U + NIs) results in a reduction in the loss of one pollutant but leads to an increase in another. In contrast to (U + NI), several recent studies have indicated the existence of inconsistencies in the impact of (U + UIs) on N 2 O emission [16]. For instance, after a meta-analysis of 113 field experiments, Akiyama et al. [9] found that UI tended to reduce N 2 O emissions by 10% on average, though this was not significant. In soils from New Zealand, van der Weerden et al. [17] found that NBPT (N-(n-butyl) thiophosphoric triamide) had no effect on the overall N 2 O emissions from U. For reduction of both NH 3 and N 2 O emissions, a combination of UIs and NIs for U, i.e., (U + UIs + NIs), has been recommended [18]. Beyond reducing N losses via inhibitors, recent studies have found that NI uses can also simplify fertilization in intensive crop production to save labour and machinery costs, mainly by allowing a reduction in split N application by three to two times, while without affecting the agricultural crop yield [19]. According to the IPCC guidelines [8,20], approximately 1% (a range of 0.2-5%) of the emitted NH 3 is converted to N 2 O after its deposition (IPCC emission factor EF4, (kg N 2 O-N [kg NH 3 -N + NO x -N volatilized] −1 )). Furthermore, although the studies on the effects of NIs on NH 3 and N 2 O emissions are available in literature [16,20], there is still lack of information about their effects in cereal crops grown under temperate field conditions. To evaluate the potential of NIs to mitigate agricultural greenhouse gas emissions, this indirect N 2 O emission should therefore also be included.
Therefore, the main objectives of this study were to investigate the effect of NI and UI on N 2 O and NH 3 emissions from winter wheat with U in southern Germany, to evaluate whether adding NIs and UIs to U stimulates NH 3 or N 2 O emissions, while they reduce one pollutant, and to estimate indirect effect of urease and nitrification inhibitors on the N 2 O-N losses of uU based on the IPCC emission factor EF4.

Experimental Site and Design
Five-year field experiments with winter wheat, from 2004 to 2005 and from 2011 to 2013, which was sown at the beginning of October with a seed density of 350 seeds m −2 and was harvested at the end of July, were conducted at the experimental station of the Technical University of Munich at Dürnast in Germany (11 • 41 60" E, 48 • 23 60" N). The experimental site is characterized by a temperate climate with a mean annual temperature of 7.8 • C and an average annual precipitation of 821 mm (14.2 • C and 557 mm from March to September, respectively). Daily meteorological data were obtained from a weather station from Deutscher Wetterdienst (DWD), which is approximately 500 m from the experimental site.
The soil is a mostly homogeneous Cambisol of a silty clay loam texture (IUSS Working Group WRB, 2007). Based on the guideline of VDLUFA (Association of German Agricultural Analytic and Research Institutes) for soil sampling and analysis [21], 16 representative soil samples are required for an experimental site, and were collected from the experimental site (0-0.3 m) and analyzed ( Table 1). The sampling area was homogenous. As a rule in Germany, there is no requirement to have standard deviation for soil analysis, since the samples are already composited. Soil pH was determined following Schofield and Taylor [22] (0.01 M CaCl 2 ). Soil textural analysis was conducted in compliance with ISO (2002). Organic carbon (C org ) and total-nitrogen (N total ) were analyzed using the Dumas procedure (combustion at 1100 • C) [21]. At the beginning of spring, all plots received a compound fertilizer consisting of 25 kg P ha −1 , 100 kg K ha −1 , 10 kg Mg ha −1 , and 40 kg S ha −1 , respectively. The fertilizer was surface-applied using a plot fertilizer distributor. The nitrogen (N) fertilizer treatments and the strategies for fertilizer application are listed in Table 2. Before the first N split application, soil NO 3 -N in the top layer (0-60 cm) was determined and was around 20 kg N ha −1 . The UI, IPAT (N-isopropoxycarbonyl phosphoric acid triamide), at 0.4% w/w (SKW Stickstoffwerke Piesteritz GmbH, Lutherstadt-Wittenberg, Germany) and the NI, DCD/TZ (DCD/TZ: Dicyandiamide/1H-1,2,4-Triazol), at 2% w/w (SKW Stickstoffwerke Piesteritz GmbH, Lutherstadt-Wittenberg, Germany) were added to the urea. According to the fertilizer recommendation, the total N supplied was 180 kg N ha −1 , and based on the common fertilization practices, there were 3× split applications for fertilizer treatments without NI during the growing season and 2× split applications for treatments with NI since NI is able to maintain soil NH 4 + for a longer time (Table 2). For the simultaneous comparison of the effect of the fertilization treatments on N 2 O emission under field conditions, the five fertilization treatments were randomly assigned to 10 plots (2 × 10 m) in 2004 and 2005. In 2012, since the treatment (U + UI) was omitted, only four treatments were randomly assigned to 12 plots. For studying the effect of fertilization treatments on NH 3 emission, 15 plots (2 × 10 m) in 2011 and 2013 were randomized with three replicates, and additional 3 plots (2 × 10 m) were as control treatment with zero N fertilizer in 2011 and 2013. The distance between the neighboring plots was 2 m. In this study, the measurements of N 2 O and NH 3 emission were not parallel. This is because, in our long-term study on N 2 O and NH 3 emission, fertilization treatments with UI or NI varied with years. The comparable fertilization treatments including U, U + NI, U + UI, and U + NI + UI altogether were present only in the years from 2004 to 2005, and 2010 to 2012 (Table 2). Furthermore, our long-term studies on N 2 O emissions have shown that averaged N 2 O emissions over many years were consistent (back to the 1990s, e.g., [23][24][25][26]). In particular, NH 3 emission from U from 2001 to 2010 were much less variable with years and were limited to a small range of 1-3.4% of urea-N applied during the growing seasons of winter wheat [6,[25][26][27][28][29]. A simulation study on modeling NH 3 emissions across Germany [30] showed that NH 3 emissions (% urea N applied) were in a range of 0-10% in strong contrast to the averaged 24% for European arable soils assumed by EMEP/EEA [31].

N 2 O Flux Measurements
The measurements of N 2 O fluxes were carried out using the closed-chamber method according to Hutchinson and Mosier [32] (Figure 1) and are described briefly as follows. Before the first-time sampling, four metal rings with an inner diameter of 0.3 m and a height of 0.15 m were installed in a depth of 0.08 m in the middle of each plot ( Figure 1a). In order to sample the same area, the rings within the plots remained until the last measurements were completed. Nitrous flux rates were determined by placing a closed chamber over the permanently installed rings in each plot and taking three gas samples from the chamber's atmosphere with sampling intervals of 20 min, i.e., at 0, 20, and 40 min after opening the each chamber were collected between 11:00 a.m. and 12:00 p.m. An overview of the N 2 O measurements in a 3-year study in 2004, 2005, and 2012 is summarized in Table 3. For each measurement day, the N2O emission flux rate (FN) (µ g N2O-N m −2 h −1 ) was calculated for each chamber based on a linear model as reported by Flessa et al. [34]: where V is the volume of the chamber in L; ΔC is the difference of the N2O concentration (µ L L −1 ) before closure (t0) and at the end of closure (t1 or t2); A is the ground area enclosed by the chamber in m 2 ; Δt is the time interval between t0 and t1 or between t1 and t2 in h; and ρN is the mass percentage of N in N2O (1.25 µ g N µ L −1 ).
The N2O emission flux rate at a given day was averaged from those at t1 and at t2. To avoid the high spatial heterogeneity within a plot, four measurements of N2O fluxes per plot were averaged at each time. The area of the closed-chamber covered 0.01131 m 2 , and the volume was 1.787 dm −3 ( Figure  1b). Thus, the measured area (4 × 0.01131 m 2 ) in each plot accounted for approximately 0.23% of the area of a plot (20 m 2 ).
Flux rates of each plot were integrated over three or two periods of N2O-N emissions to calculate cumulative N2O-N emissions (g N2O-N ha −1 ), i.e., the amount of N2O-N emissions between two subsequent split N fertilizations was estimated by "averaged N2O-N emission rate (g ha −1 d −1 ) × duration (days)".

NH3 Measurements
Ammonia emissions were measured by a combination of the Dräger Tube Method (DTM) with modified passive samplers [35][36][37][38] (Figure 2a,b). Briefly, the passive sampler gives semi-quantitative  At sampling time, a volume of 20 mL of gas was collected using an air-tight syringe from the headspace ( Figure 1b) and injected into a vacuum serum tube fitted with butyl rubber stopper to keep the gas under pressure. Gas samples were analyzed for N 2 O within a week using an automated gas chromatograph (Varian Star 3400, Varian, Walnut Creek, California, USA) equipped with electron capture and thermal-conductivity detectors. Computer software Star Chemstation (Varian, Walnut Creek, California, USA) was used to calculate N 2 O concentrations. N 2 O content was determined using a gas chromatography system (Varian Star 3400, Varian, Walnut Creek, California, USA). Two calibration curves, i.e., for both low and high concentrations, were used [33].
For each measurement day, the N 2 O emission flux rate (F N ) (µg N 2 O-N m −2 h −1 ) was calculated for each chamber based on a linear model as reported by Flessa et al. [34]: where V is the volume of the chamber in L; ∆C is the difference of the N 2 O concentration (µL L −1 ) before closure (t 0 ) and at the end of closure (t 1 or t 2 ); A is the ground area enclosed by the chamber in m 2 ; ∆t is the time interval between t 0 and t 1 or between t 1 and t 2 in h; and N is the mass percentage of N in The N 2 O emission flux rate at a given day was averaged from those at t 1 and at t 2 . To avoid the high spatial heterogeneity within a plot, four measurements of N 2 O fluxes per plot were averaged at each time. The area of the closed-chamber covered 0.01131 m 2 , and the volume was 1.787 dm −3 ( Figure 1b). Thus, the measured area (4 × 0.01131 m 2 ) in each plot accounted for approximately 0.23% of the area of a plot (20 m 2 ).
Flux rates of each plot were integrated over three or two periods of N 2 O-N emissions to calculate cumulative N 2 O-N emissions (g N 2 O-N ha −1 ), i.e., the amount of N 2 O-N emissions between two subsequent split N fertilizations was estimated by "averaged N 2 O-N emission rate (g ha −1 d −1 ) × duration (days)".

NH 3 Measurements
Ammonia emissions were measured by a combination of the Dräger Tube Method (DTM) with modified passive samplers [35][36][37][38] (Figure 2a,b). Briefly, the passive sampler gives semi-quantitative results of NH 3 emission (NH 3 -N ppm, sum), which properly reflects the relative difference between treatments. For the absolute flux values (kg NH 3 -N ha −1 ) from passive sampler measurements (Figure 2b), a transfer coefficient is required. The DTM (Figure 2a), a calibrated dynamic chamber method, is employed to obtain a transfer coefficient for converting the semi-quantitative NH 3 emission of the passive sampler to quantitative units (kg N ha −1 ) [37,38]. For N fertilizer treatments, a control treatment without N fertilizers should be included as well. The DTM is a variant of dynamic chamber technique and adjusted to actual meteorological conditions by a calibration equation [36]. The principle is firstly obtaining primary flux during a short measurement interval (~1 min) in the chamber with a fast reading indicator of NH 3 concentration, and then scaling the primary flux into actual ambient flux using temperature and wind speed outside of the chamber through an empirical calibration equation. This method has been shown to be well-suited for NH 3 measurements in multi-plot filed trials [35][36][37][38]. To measure NH3 emission in the field, a passive sampler with 20 mL of absorption solution (0.05 M H2SO4) was installed in the center of each plot at 0.15 m above plant canopy [35][36][37][38]. The gaseous NH3 in the air was absorbed by H2SO4 solution in the passive samplers to form (NH4)2SO4. The absorption solution from the passive samplers was exchanged daily for all treatments, except during rainy days. On rainy days, the solution in samplers was replaced at the day after the rain stopped. An overview of NH3 measurements in a 2-year study in 2011 and 2013 is shown in Table 4.  To measure NH 3 emission in the field, a passive sampler with 20 mL of absorption solution (0.05 M H 2 SO 4 ) was installed in the center of each plot at 0.15 m above plant canopy [35][36][37][38]. The gaseous NH 3 in the air was absorbed by H 2 SO 4 solution in the passive samplers to form (NH 4 ) 2 SO 4 . The absorption solution from the passive samplers was exchanged daily for all treatments, except during rainy days. On rainy days, the solution in samplers was replaced at the day after the rain stopped. An overview of NH 3 measurements in a 2-year study in 2011 and 2013 is shown in Table 4. Daily except during rainy days * * (i) On rainy days, the passive solution samplers were replaced at the day after the rain stopped. (ii) Between two subsequent N fertilizer applications, the measurements were stopped after NH 3 emission was zero for three days.
The NH 4 + concentration from each passive ampler was subsequently analyzed in the laboratory using the standard distillation method. Briefly, the samples collected from passive samplers were firstly mixed with 10 mL 10 M NaOH solution. During distillation, NH 3 was released and, as the gas cooled down, NH 4 OH formed was collected in an Erlenmeyer flask. After the distillation was completed, titration with 0.005 M H 2 SO 4 was done to determine the amount of NH 3 in the distillate. 1 mL 0.005 M H 2 SO 4 corresponded to 0.14 mg N. The DTM requires much fewer frequent measurements than the passive sampler measurements. However, a recent study [37] suggested that the range of valid calibration conditions has to be kept in mind when applying the DTM. To obtain reliable transfer coefficients for the same treatment at given time, the measurements for the transfer coefficients by the DTM were carried out simultaneously with the passive sampling for all treatments during the growing seasons. Ambient air was sucked through four soil chambers (height of 0.105 m, diameter of 0.115 m, and surface area of 0.0415 m 2 ) using a manual pump (Drägerwerk AG & Co., KGaA, Lübeck, Germany). NH 3 concentrations in the air passing through the chamber system of the DTM were instantaneously measured with indicator tubes (Drägerwerk AG & Co., KGaA, Lübeck, Germany). Due to the low air-exchange rate in the chambers, the NH 3 -flux values need to be corrected by means of a calibration approach proposed by Pacholski et al. [37]. For more details on the DTM method and the calculation of transfer coefficients see Gericke et al. [37] and Pacholski [38].
The cumulative NH 3 -N emissions (kg NH 3 -N ha −1 ) were the sum of daily NH 3 -N emissions from the beginning up to the end of the NH 3 measurements.

Estimation for Considering NH 3 as an Indirect N 2 O Emission
Estimation for considering NH 3 as an indirect N 2 O emission was based on the IPCC guidelines [8,19], i.e., 1% of the emitted NH 3 (N 2 O-N + 1% × NH 3 -N) is converted to N 2 O after its deposition.

Statistical Analysis
Depending on the number of replicates of treatments, Tukey's (HSD) post hoc (>2 replicates) and the nonparametric (Kruskal-Wallis-H-test) (2 replicates) statistical tests were used to test for differences between the fertilization treatments in terms of N 2 O and NH 3 emissions using SPSS software (Proc. Mixed, SPSS ver. 26, IBM). The statistical significance of all tests was evaluated at the 0.05 level.

N 2 O Flux Pattern and Cumulative N 2 O-N during Growth Periods under Different Fertilization Treatments
The spatial and temporal variability of N 2 O fluxes with weather conditions was notably high after the N applications (Figures 3-5). For all three years, our results showed a general pattern of N 2 O fluxes, such as an increase in the N 2 O flux rate followed both N fertilization and rainfall events (Figures 3-5). For example, the first peak of N 2 O flux rate was 100-600 g N 2 O-N ha −1 d −1 on 9 April 2004, 10-50 g N 2 O-N ha −1 d −1 on 11 April 2005, and 15-60 g N 2 O-N ha −1 d −1 on 9 April 2012, respectively. In particular, the first highest peak of N 2 O in 2005 appeared two weeks after the first N application following a heavy rainfall. Similar patterns of N 2 O emissions from arable soils with increased flux rates following N fertilization and rainfall have been reported, e.g., for winter wheat [39] and for oilseed rape [40,41]. This type of pattern is induced by the main drivers of N 2 O emission from N fertilization and precipitation, and thus could be explained by enhanced denitrification due to (i) an increase in the availability of nitrate as a substrate for N 2 O production [33]; (ii) formation of anaerobic conditions as a result of lower gas diffusivity in soil water, and thus reduced O 2 diffusion into the soil combined with O 2 consumption by soil microbes [42][43][44]; and (iii) soil drying and rewetting. As the content of available organic matter in soils may increase with soil drying, organic substrates for soil microorganisms, which may promote denitrification and thus N 2 O release, may be highly available when the soil is rewetted [39]. A meta-analysis study [45] showed that rewetting can increase N 2 O fluxes by nearly five times in cropland following a drying period. Figure 3a shows that, although N 2 O emission in (U + NI) and (U + UI + NI) treatments was lower at most measurement dates compared with the treatment of U alone, N 2 O emission on 9 April 2004 was higher in (U + NI) and (U + UI + NI) treatments than that of U alone, which could still not be explained. Furthermore, it was unusual to observe the negative N 2 O flux on 15 May 2005. This may have been due to a longer dry period, resulting in no N 2 O emissions on the day when the measurement was taken.     Table 5). Early reports for different winter cereal fields in Germany showed that annual N2O emissions ranged from 1700 g to 4000 g N2O-N ha −1 a −1 [39,47,48]. However, N2O emission postharvest or during winter may account for approximately 50% of the total annual N2O emission [47]. Thus, N2O emissions during the winter wheat growth season from our experiment in southern Germany were of the same order of magnitude as previously reported.
Air temperature ( C) or Precipitation (mm d In this study, a rapid drop in air temperature followed a peak of N 2 O flux on 9 April 2004 and 8 April 2012 (Figures 3 and 6). Since the temperature drop in April lasted only a very short time and the temperate was still above 0 • C, effects of frozen-thawing cycles on N 2 O flux probably did not occur. A number of studies on the effect of frozen-thawing cycles on N 2 O flux in arable soils in Germany have shown that the distinct frozen-thawing cycles occur mainly during winter season, i.e., from January to February or March (e.g., [39][40][41][42][43]46]). As described earlier, the N 2 O flux pattern in this study was therefore probably induced by the main drivers from N 2 O emission resulting from N fertilization and wet-dry fluctuation, i.e., precipitation.
The cumulative N 2 O emission during the wheat growth period ranged from 2133 to 3614 g ha  Table 5). Early reports for different winter cereal fields in Germany showed that annual N 2 O emissions ranged from 1700 g to 4000 g N 2 O-N ha −1 a −1 [39,47,48]. However, N 2 O emission post-harvest or during winter may account for approximately 50% of the total annual N 2 O emission [47]. Thus, N 2 O emissions during the winter wheat growth season from our experiment in southern Germany were of the same order of magnitude as previously reported.  Accumulative N2O emission during the growth season varied considerably according to the tested year ( Figure 6). For example, N2O emission from fertilization with U alone in 2004 was nearly three times higher than that in 2005 and 2012, and that from U with NIs in 2004 was nearly four times higher than that in 2005 and 2012. A consistently higher emission was observed between the first and second applications compared to that during the later growth period (Figure 6). For example, the N2O emissions after the first two N applications were approximately 87-94% in 2004, 55-76% in 2005, and  Accumulative N 2 O emission during the growth season varied considerably according to the tested year ( Figure 6). For example, N 2 O emission from fertilization with U alone in 2004 was nearly three times higher than that in 2005 and 2012, and that from U with NIs in 2004 was nearly four times higher than that in 2005 and 2012. A consistently higher emission was observed between the first and second applications compared to that during the later growth period ( Figure 6). For example, the N 2 O emissions after the first two N applications were approximately 87-94% in 2004, 55-76% in 2005, and 48-64% in 2012, respectively, of the total emission during the growth period of winter wheat. A high variation in N 2 O fluxes with growth season and year has often been reported in field studies with N 2 O measurements [40,49], as well as in modelling approaches from sites with different climatic conditions [50,51]. Despite a uniform management approach (N fertilization and crop type), annual N 2 O emission varied by up to a factor of seven between single experimental years. Different rainfall frequencies, i.e., the interval of drying and rewetting periods, could explain the difference in N 2 O emission between 2004 and 2012 (Figures 3 and 5). Although a similar rainfall frequency, particularly between the first and second split N applications (Figures 3 and 4), the difference in N 2 O emissions between 2004 and 2005 was still considerable.
Between the fertilization treatments with U alone and CAN, N 2 O losses during the growth period of winter wheat were 1394 g N 2 O-N ha −1 and 510 g N 2 O-N ha −1 higher for U alone than for CAN in 2004 and 2012, respectively, whereas there was no difference in N 2 O emissions between these two N forms in 2005 ( Figure 6). Lebender et al. [52] reported there was a difference in N 2 O emissions between U and CAN during the growth period from March to the end of July for winter wheat. The higher N 2 O-N losses from U might be due to an increase in pH from U hydrolysis, leading to a higher N 2 O emission. Bremner and Blackmer [53] reported that nitrification, which produces N 2 O as a by-product, generally increases under higher soil pH (>6) conditions. Denitrification has been reported to occur naturally over a wide range of soil pH (5.0-8.0) [54]. As an alkaline-hydrolyzing N fertilizer, U influences nitrification through a transient rise in pH with subsequent denitrification leading to the formation and release of larger amounts of N 2 O [45,55,56]. A recent study by Wrage-Monnig et al. [57] showed that high pH may favor nitrifier denitrification that accounts for up to 100% of nitrous oxide emissions from ammonium (NH 4 + ) in soils.
A decrease in N 2 O fluxes via the use of NIs has already been confirmed across climates, soil types, and soil characteristics [9,24,33,39,44]. Recent meta-analyses indicated that NIs decreased N 2 O emissions by 31-48% across diverse agricultural ecosystems, including upland, grassland, and paddy fields [9,10]. Weiske et al. [58] reported a reduction of 49% during the cropping season for measurements in southern Germany. Our study confirmed these previous findings. For example, compared to U alone, the cumulative N 2 O emission of U with NIs during the winter wheat growth season was reduced by 24% in 2004, 30% in 2005, and 44% in 2012. Because the number of split N application was reduced from 3 to 2 times, the results of this study further suggest that beyond reducing N 2 O-N losses by NIs, the use of NIs can also simplify fertilization in intensive crop production to save labour and machinery costs, which is in keeping with the results of our previous report [19].
In contrast, the effect of (U + UIs) and/or the interactive effect of (U + UIs) and (U + NIs) on N 2 O emissions from U fertilization was inconsistent. A meta-analysis of the effect of UIs on N 2 O emissions from U fertilization by Akiyama et al. [9] showed that (U + UIs) did not reduce N 2 O emissions. Figure 6 shows that N 2 O emissions from (U + UIs) or (U + UIs + NIs) decreased, increased, or did not change compared to those of U alone. Compared to U alone, the cumulative N 2 O emission of (U+ UIs) remained nearly unchanged in 2004 and increased by 36%, while that for (U + UIs + NIs) decreased by 11% in 2004, 13% in 2005, and 47% in 2012. This result is probably observed because unlike the nitrification process, U hydrolysis is not directly related to N 2 O emissions. The use of UIs only delays U hydrolysis, and all U will eventually be hydrolyzed to become NH 4 + . With the use of NIs, if plant uptake of N in the form of NH 4 + does not significantly increase, a similar amount of NH 4 + will eventually undergo nitrification and subsequent denitrification compared to U alone [9].

Cumulative NH 3 Emissions during Growth Periods under Different Fertilization Treatments and Added Inhibitor Effect on NH 3 Emissions
The cumulative NH 3 emissions from a two-year study in 2011 and 2013 are shown in Figure 7. The cumulative NH 3 emissions from U alone were 12.7 kg N ha −1 in 2011 and 7.1 kg N ha −1 in 2013, which corresponded to an emission factor of 11.5% and 3.9% of the applied N during 2011 and 2013, respectively (Table 6). On average, during the two-year study, NH 3 -N losses amounted to 7.7% of the applied urea N, which was similar to recent findings (i.e., 8%) reported by Ni et al. [5]. Generally, the emissions observed in this study were relatively low compared to the NH 3 losses of up to 64% reported in the literature [13] and of 16% assumed for the calculation of the national emissions inventory for Germany [31]. Among many factors affecting soil NH 3 emissions, air temperature, precipitation, and soil moisture and pH are the key factors. However, the major reasons leading to low NH 3 emission in southern Germany may be due to low pH values of the investigated soils together with increased cation exchange capacity, and low temperature prevailing in April and frequent precipitation events after N application. However, the 7.7% of the applied urea N from NH 3 -N losses was higher than the 0.1-2.7% in southern Germany reported by Schraml et al. [6]. This discrepancy may be attributable to the higher pH at the experimental site of this study (pH = 6.7).
Atmosphere 2020, 11, x FOR PEER REVIEW 10 of 19 emissions observed in this study were relatively low compared to the NH3 losses of up to 64% reported in the literature [13] and of 16% assumed for the calculation of the national emissions inventory for Germany [31]. Among many factors affecting soil NH3 emissions, air temperature, precipitation, and soil moisture and pH are the key factors. However, the major reasons leading to low NH3 emission in southern Germany may be due to low pH values of the investigated soils together with increased cation exchange capacity, and low temperature prevailing in April and frequent precipitation events after N application. However, the 7.7% of the applied urea N from NH3-N losses was higher than the 0.1-2.7% in southern Germany reported by Schraml et al. [6]. This discrepancy may be attributable to the higher pH at the experimental site of this study (pH = 6.7). Compared to the NH3 loss associated with U treatment, NH3 loss from CAN has been less intensively studied in the past [4]. The results in this study showed an emission factor of 4.5% of the applied N based on the average of a two-year study (Table 6), which was lower than the NH3 emission from U. Ni et al. [5] reported that the average NH3 losses from CAN were approximately 2.1% of the applied N, which was near the EEA emission factor of 2%. The lower NH3 emissions of CAN could be explained by the lower NH4 + concentration in CAN and its considerably smaller effect on soil pH.  3). A mean comparison between fertilization treatments from Tukey´s HSD-test indicates a significant difference at p < 0.05. The same letters indicate groups that were not significantly different from one another. Compared to the NH 3 loss associated with U treatment, NH 3 loss from CAN has been less intensively studied in the past [4]. The results in this study showed an emission factor of 4.5% of the applied N based on the average of a two-year study (Table 6), which was lower than the NH 3 emission from U. Ni et al. [5] reported that the average NH 3 losses from CAN were approximately 2.1% of the applied N, which was near the EEA emission factor of 2%. The lower NH 3 emissions of CAN could be explained by the lower NH 4 + concentration in CAN and its considerably smaller effect on soil pH.
Addition of the UIs to U caused a significant reduction in NH 3 emissions and reduced NH 3 losses by 26-83% and 24-32%, respectively, of the applied urea-N [5,6]. Our study showed that (U + UIs) reduced NH 3 losses by 52% in 2011 and 54% in 2013, respectively, compared to those observed with U treatment alone. Similarly, (U + UIs + NIs) reduced NH 3 losses by 50% in 2011 compared to those of U alone, but there was a considerably greater reduction in 2013, i.e., 89%, compared to that of the NH 3 losses from U alone. This finding may suggest that (U + UIs + NIs) might be a potential approach to minimizing the negative environmental effects of U application under agro-ecological conditions.
In contrast to (U + UIs + NIs), the addition of (U + NIs) in this study stimulated NH 3 emissions compared to treatment with U alone ( Figure 6 and Table 6). The cumulative NH 3 emission from (U + NIs) was highest in both 2011 and 2013 among all N fertilization treatments, which corresponded to an emission factor of 14.1% and 5.9% of the applied N, respectively (Table 6). This result may be due to the retention of NH 4 + in the soil during the use of NIs, which could increase NH 3 emissions [11].
The application of an NI (DCD) has been found to increase NH 3 emissions by 18-29% [11,13], and by up to 38% [14]. However, previous studies have found inconsistent results, e.g., stimulating [13], neutral [5,15], and retarding [59] effects of NIs on NH 3 emissions, which may also depend on soil properties. For example, the study [45] showed that the effect of NIs was positively correlated with soil pH and negatively correlated with CEC.

Estimation of NH 3 as an Indirect N 2 O Emission Based on the Default Value of EF4
According to the IPCC guidelines [8], approximately 1% of the emitted NH 3 is converted to N 2 O through nitrification and denitrification processes [20], which is referred to as an indirect N 2 O emission from NH 3 deposition (IPCC emission factor EF4, (kg N 2 O-N (kg NH 3 -N + NO x -N volatilized) −1 )). Direct N 2 O emissions and the estimation of indirect N 2 O emissions under treatments of U with inhibitors and U alone are shown in Table 7. Although the current studies on N 2 O and NH 3 emission were not parallel, and the indirect N 2 O emission derived from NH 3 was not measured, further evidences that may support the assumption of estimation above are that the experiments for NH 3 measurements were conducted in the same sites, and that the results of NH 3 were in close agreement with long term experiments done on these sites [6,24,29,60,61].
The results in Table 7 demonstrate that the indirect N 2 O emission from NH 3 was lowest from CAN and (U + UI). Although (U + NI) caused a higher indirect N 2 O emission, NIs decreased overall N 2 O emissions, i.e., totally 1.4 kg N 2 O-N ha −1 from direct and indirect N 2 O emissions, compared to 3.9 kg N 2 O-N ha −1 from (U + UI) and 4.9 kg N 2 O-N ha −1 from U alone (Table 7), which was only locally deposited NH 3 on the same agricultural field and did not include further deposition. Lam et al. [16] suggested that the beneficial effect of NIs in decreasing direct N 2 O emissions can be outweighed by an increase in NH 3 volatilization. In contrast, our study may indicate that considering indirect N 2 O emissions from NH 3 induced by NIs may be negligible for winter wheat in southern Germany.

Conclusions
This field study on winter wheat in southern Germany confirms that U added with NIs greatly reduced N 2 O emissions, and additive UIs decreased NH 3 emissions. Compared to U alone, the cumulative N 2 O emissions of U with NIs during the winter wheat growth season decreased by 24-44%, and UIs added to U reduced NH 3 losses by as much as 52-54%. The difference in the effect of UIs or (UIs + NIs) on N 2 O emissions from U fertilization may suggest that the combined use of UIs and NIs might be a potential approach to minimizing negative environmental effects of U application under agro-ecological conditions. Furthermore, in contrast to the treatment of (U + UIs + NIs), (U + NIs) stimulated NH 3 emissions compared to the treatment with U alone. According to the IPCC guidelines for emission factor EF4 [8,20], approximately 1% (a range of 0.2-5%) of the emitted NH 3 is converted to N 2 O after its deposition. The results of the indirect N 2 O emissions from NH 3 estimated based on the emission factor EF4 by IPCC showed that total N 2 O emissions from (U + NIs) were approximately 2.8-3.5 times lower than those from U and (U + UI), which may suggest that indirect N 2 O emission from NH 3 induced by NIs may be negligible.