Soil incubation study showed biogas digestate to cause higher and more variable short‐term N2O and N2 fluxes than mineral‐N

Today, a large share of mineral fertilizer is substituted by biogas digestates. Biogas digestates are known to promote N2O production, compared to mineral fertilizer. In particular, the initial phase following fertilizer application is crucial for the N gas release as N2O and also N2. However, this period impact has been rarely investigated, especially not across various field sites. Thus, undisturbed soil cores from two fertilizer types (biogas digestate vs. mineral fertilizer) at five sites with different site characteristics were investigated in a short-term laboratory experiment under N2-free helium–oxygen incubation atmosphere. Across sites, biogas digestate soil cores showed significantly higher absolute N2O fluxes compared to mineral fertilizer soil cores, even though this effect was dominated by samples from one site (Dornburg with the highest biogas digestate fertilization rate). Also relative N2O fluxes showed a similar tendency. On average, absolute and relative N2 fluxes differed between the two fertilizer types, while N2 fluxes were highest at the Dornburg site. A N2O/(N2O+N2) ratio of denitrification below or equal to 0.5 clearly highlighted the importance of N2O reduction to N2 for three of five the biogas digestate soil cores. Soil characteristics like bulk density and water-filled pore space as proxies for gas diffusivity in soil, as well as N availability (NO 3 , NH þ 4 ), significantly affected the N2O and N2 fluxes from the biogas digestate soil cores. While this study presents data on short-term N2O and N2 fluxes, there is a need for further studies in order to investigate the dynamics, the duration of the observed effects and their significance at the field scale.


Introduction
Gaseous nitrogen (N) emissions from agricultural fields are considered problematic both from an agricultural-in terms of N efficiency-as well as from an environmental perspective (Cameron et al., 2013). This is particularly the case for nitrous oxide (N 2 O), which is highly relevant due to its global warming potential (IPCC, 2013), while it is also expected to be the single most important ozone-depleting substance throughout the 21 st century (Ravishankara et al., 2009). N 2 O mainly originates from agricultural soils, where high N fertilizer inputs (mineral and organic fertilizer) are directly related to N 2 O production, resulting in a conversion of 2 to 2.5% of fertilizer N to N 2 O (e.g., Davidson, 2009) and contributing up to 4.8 Tg N y -1 to the anthropogenic N 2 O emissions (IPCC, 2013).
Apart from N 2 O, fertilizer N can be lost in the form of the nonreactive molecular nitrogen gas (N 2 ), which is non-hazardous from an environmental perspective, but still of great agronomic interest in terms of fertilizer management. In addition to N 2 O and N 2 , there is a third significant pathway of gaseous N losses as ammonia (NH 3 ). While for NH 3 there is already a lot of information about the impact of N fertilization available (e.g., Wolf et al., 2014;Seidel et al., 2017;Pietzner et al., 2017), for the other important gas N 2 there is still very limited information on fertilizer-related N 2 losses. This is particularly the case for the application of organic fertilizer in form of fermentation residues, often called biogas digestates (BD), which are increasingly used as organic amendments in agriculture (Charles et al., 2017). BD contain large amounts of organic carbon (C), N, and other nutrients, which promote soil microbial activity. In particular, BD may increase ammonium (NH þ 4 ) concentrations, affect the biological oxygen (O 2 ) consumption, narrow the C/N ratios, as well as elevate pH values in the soil with repeated additions (Mö ller and Mü ller, 2012;Mö ller, 2015). During BD application under typical spring conditions on often wet soils and by using drag hose or injection technique, BD hotspots can stimulate N 2 O production via nitrification and denitrification, resulting in a higher potential of gaseous N losses (Mö ller and Stinner, 2009;Kö ster et al., 2011;Senbayram et al., 2009;. In particular, the modified O 2 availability in combination with high amounts of available organic C provided by BD enhances soil respiration. Moreover, as BD contain mostly water (up to 97% moisture content) (Mö ller, 2015;Charles et al., 2017), their addition rapidly saturates micro-pores, which then enhances the potential for N 2 O reduction to N 2 (Butterbach-Bahl et al., 2013). However, there is no clear picture about the impact of BD application on N 2 formation up to now Fiedler et al., 2017).
In addition to the effect of fertilizer type (mineral vs. organic), N 2 O production and consumption processes in agricultural systems are largely influenced by the soil characteristics at the respective location. N 2 O and N 2 fluxes are significantly influenced by a range of factors, e.g., availability of N, C, and O 2 in the soil (Weier et al., 1993;Conrad, 1996;Senbayram et al., 2012), as well as soil moisture, soil type, soil texture, soil pH, climate, microbial community structure, fertilization, and management (Butterbach-Bahl et al., 2013;Saggar et al., 2013). In particular, soil moisture might be the most important driving factor during BD application in spring which has not been widely investigated with respect to N 2 fluxes (Fiedler et al., 2017). This is likely due to the fact that the determination of N 2 fluxes from soils is still a delicate matter because of high atmospheric background concentrations complicating the detection of concentration changes in amounts relevant to soil processes (Groffman et al., 2006). To overcome this problem, two methods can be used: 15 N tracing technique (i.e., analysis of gas fluxes after addition of 15 N-labelled substrate) and N 2 -free incubation technique (i.e., replacing N 2 by a noble gas like Helium) (Van Groenigen et al., 2015). Since the 15 N tracing technique requires a homogeneous labeling in order to produce precise results (Boast et al., 1988;Arah, 1992), it is particularly challenging for heterogeneous substrates such as BD and thus rather expensive. Hence, the N 2 -free incubation technique provides an alternative for the investigation of a large number of undisturbed soil cores from different sites, where the actual N 2 O and N 2 release can be measured. Generally, the initial phase, i.e., the first days after fertilizer application, is crucial for N 2 O emissions (Senbayram et al., 2009;Kö ster et al., 2011) and previous BD experiments have indicated a similar immediate reaction also for N 2 Fiedler et al., 2017). Since such measurements on BD are up to now not available under field conditions due to the issues described above and still rare under laboratory condi-tions, a short-term laboratory incubation experiment would provide reliable data on N 2 O and N 2 fluxes with respect to fertilizer type and soil conditions from various field sites, which might be further used to supplement field measurements.
By mapping a practical range of parameters, the aim of this laboratory incubation study was to (1) measure short-term N 2 O and N 2 fluxes following the application of biogas digestate (BD) and mineral fertilizer (MIN) in undisturbed soil cores from five different field sites, as well as (2) to analyze the effects of soil and fertilization parameters and their interactions on N 2 and N 2 O fluxes and the N 2 O/(N 2 O+N 2 ) ratio of denitrification. The tested hypotheses were as follows: (1) N 2 O and N 2 fluxes shortly after fertilizer application are higher from BD compared to MIN samples, (2) differences between samples from different sites, covering various soil characteristics in Germany, have a greater impact on N 2 O and N 2 fluxes than the fertilization type (BD, MIN).

Treatments and fertilization
Soil core samples for the incubation experiment originate from a larger field experiment, established for the ''EVA'' (Development and Comparison of Optimized Cropping Systems for the Production of Energy Plants under the Variety of Regional Conditions in Germany) project. The project investigated the effects of fertilization with BD on N 2 O emissions at five experimental sites (Ascha, Dedelow, Dornburg, Hohenschulen, Gü lzow) throughout Germany (Tab. 1) by measuring greenhouse gas emissions from continuous cropping of energy maize (Zea mays L.) under different fertilizer regimes from 2011 until 2014 (Hagemann et al., 2017). Two fertilizer treatments were selected for sampling: (1) mineral N fertilizer (granular calcium ammonium nitrate; MIN) and (2) organic N fertilizer (biogas digestate; BD). According to common agricultural practice, BD treatments received an application rate equivalent to 70% mineral fertilizer, which explains the generally higher N fertilization rates in this treatment (Tab. 1). More- 2020  over, the Dornburg site is an exception; here, mineral fertilization was lower due to a planned second MIN application of 100 kg N ha -1 (several weeks after soil sampling), which is the common farming practice on that soil.
The BD originates from local biogas plants near the respective study sites, co-fermenting cattle manure and maize silage (all sites), and a small proportion of rye or barley groats (Dornburg and Gü lzow). Fertilizers were applied using drag hoses (BD except for the Gü lzow site), injection technology (BD in Gü lzow), and a common fertilizer spreader (MIN). A representative sample was obtained from the well-mixed BD at all sites and analyzed in duplicate for dry matter content. pH was measured in the fresh sample using a pH meter. Total organic C and N contents were analyzed by dry combustion (elementary analysis). Phosphorus and potassium content was determined by AAS (Atomic Absorption Spectrometry) following DIN ISE 11885 using an ICP-iCAP 6300 DUO (Thermo-Fisher). NH þ 4 -N content was analyzed photometrically according to DIN 38406-E5-2 using a Segemented Flow Analysis (Skalar Analytics, CFA-SAN).

Soil core sampling and analysis
Soil samples were taken within a few hours following fertilizer application in spring 2012 (second year after the beginning of the experiment), prior to sowing and crop germination. In case of organic fertilization, the applied BD was incorporated using harrows or rotary cultivators within less than 4 h. Five intact soil cores per treatment were taken from a depth of 2 to 8 cm using steel cylinders (volume 250 cm 3 ). Samples were taken randomly within the plot (MIN treatment) or within the drag hose or injection line (BD treatment). Soil temperature in 2 to 8 cm depth at time of sampling was recorded. Cylinders were immediately sealed, weighed, cooled to 2°C, transported to the laboratory, and kept at that temperature until gas flux measurements were made.
Following soil incubation, soil subsamples (20 g fresh soil) were extracted with 0.0125 M CaCl 2 solution [1:4 (w/v)] by shaking for 1 h. The extracts were then filtered and analyzed for NH þ 4 -N and NO À 3 -N concentrations using spectrophotometry according to VDLUFA (2002) with a continuous flow analyzer (Skalar Analytics, CFA-SAN). Soil moisture content was determined gravimetrically after drying a soil subsample at 105°C until constant weight. Bulk density and water-filled pore space (WFPS) were calculated based on sample volume, dry weight and gravimetric water content. Soil pH was determined in 0.01 M CaCl 2 solution [1:5 (w/v)] using a glass electrode. An aliquot of each sample was milled and analyzed for soil organic carbon (C org ) and total nitrogen (N t ) by dry combustion using an elemental analyzer (Leco Instruments, TruSpec CNS).

Laboratory incubation and gas flux measurements
Simultaneous measurements of N 2 O and N 2 fluxes were conducted using the N 2 -free helium-oxygen (He-O 2 ) incubation method (Butterbach-Bahl et al., 2002), classified as a flow-through steady-state system according to Livingston and Hutchinson (1995). Soil samples were placed in special gastight incubation vessels inside a climate box and incubated at 10°C and the respective field moisture, similar to field conditions during fertilization and sampling. The incubation vessels feature double seals back-purged with pure He to inhibit N 2 diffusion from the atmosphere. Each measurement series consisted of five vessels with samples, an empty incubation vessel (blank), and a carrier gas measurement (control), which allows correcting for possible N 2 leakage. In order to remove ambient N 2 , vessels were evacuated with moderate suction pressure (0.047 bar) and subjected to four subsequent flushing sequences, first with an artificial He-O 2 gas mixture (20.5% O 2 , rest He) followed by an artificial He-O 2 gas mixture with traces of CO 2 , CH 4 , N 2 O, and N 2 (20.9% O 2 , 340 ppm CO 2 , 1,800 ppm CH 4 , 0.34 ppm N 2 O, 4 ppm N 2 , rest He) (Eickenscheidt et al., 2014;Fiedler et al., 2017). An artificial He-O 2 gas mixture was used, because the admixture of ambient concentrations of CO 2 , CH 4 and N 2 O allows the measurement of possible negative flows of these gases, while the admixture of 4 ppm N 2 improves the precision of measurements of low N 2 concentrations. The approach was based on the assumption that N 2 originally contained in the soil air was completely removed once the N 2 concentrations had reached a constant level. Depending on the soil type, this took from one to two days (clay substrates) after the evacuation rinse cycle. Thus, a He-O 2 gas flow rate of 15 mL min -1 to the vessel headspaces was established and continued for 24 h to 48 h until the N 2 concentrations were constant (1 to 2 days). On the following day, the N 2 O concentrations in all vessel headspaces and the control were measured once and N 2 concentration three times per day. Concentrations of N 2 O were analyzed using a GC (Shimadzu, GC-14B) with an electron capture detector (ECD), while N 2 concentrations were analyzed with a micro-GC (Agilent Technologies, 3000 Micro GC) with a thermal conductivity detector (TCD).
After checking for equal gas concentration values for control and blank, the gas flux from the soil samples was calculated using Eq. (1), based on the concentration difference of the respective gas in the incubation vessel headspace over time, taking the control value (concentration in the carrier gas) into account: where F is the flux (μg m -2 h -1 ), M is the molar mass of N 2 or N 2 O, respectively (μg mol -1 ), r the atmospheric pressure (Pa), V is the volume of the vessel, Δc is the difference between in-and outflow gas concentration (mol), v is the air flow (m 3 h -1 ), A is the surface area of the soil sample (m 2 ), R the gas constant (m 3 Pa K -1 mol -1 ), t is the time over which the concentration change was observed, and T the incubation temperature (K).

Data analysis
Normal distribution was rejected for most variables at the 0.05 significance level using the Shapiro-Wilk test. Differences in WFPS, bulk density, N fertilization rate (N fert ) , soil NH þ 4 -N and NO À 3 -N concentration, N 2 and N 2 O flux, and the N 2 O/(N 2 O+N 2 ) ratio of denitrification between treatments within site were analyzed using the Mann-Whitney U-Test; differences among sites within treatments using the Kruskal-Wallis H-Test combined with post-hoc Student-Newman-Keuls test. Pairwise associations between either N 2 flux , N 2 O flux, or N 2 O/(N 2 O+N 2 ) ratio of denitrification and the soil and fertilization parameters were characterized using the distribution-independent coefficient Spearman's rho (r). The influence of all variables on the N 2 and N 2 O flux, as well as the N 2 O/(N 2 O+N 2 ) ratio of denitrification was determined using generalized linear model (GLM) analysis with stepwise elimination. Initial included variables were treatment (MIN and BD), site, application, WFPS, bulk density, N fert , soil NH þ 4 -N and NO À 3 -N concentration, as well as bivariate interactions between these factors. All statistical analyses were conducted using SPSS 19.0.1, while figures were created using SigmaPlot 13.0.

Biogas digestate characteristics
BD dry matter and C content varied between 6.6 and 8.4%, and 2.6 and 3.8%, respectively, with Ascha and Gü lzow featuring the highest values for both parameters (Tab. 3). All BD were slightly basic (pH range 7.3 to 8.1). Fresh matter N content ranged from 0.4 to 0.5%, with NH þ 4 -N being the dominant N compound (50 to 60%).

Soil characteristics
Site-specific average soil sample bulk density ranged from 1.1 to 1.6 g cm -3 , with no significant difference in average bulk density between samples with two fertilization types (Tab. 2). BD samples from Gü lzow had the highest bulk density (1.64) and also a particularly high WFPS value of 75 6%. On average, the WFPS of all BD samples (45 4%) significantly exceeded that of MIN samples (27 2%), although this trend was not significant for samples from Ascha. Except for one site, soil pH CaCl2 was significantly higher after BD fertilization. Fertilization with BD also significantly increased soil C org concentration for samples from all sites. While average soil N t concentration was also significantly higher for BD than for MIN samples, this trend was only significant for samples from Dornburg and Gü lzow. Regardless of fertilization type, the Gü lzow samples featured consistently the lowest soil C org and N t concentrations.
Across all sites, BD samples exhibited significantly higher average NH þ 4 -N and lower NO À 3 -N concentrations than MIN samples (Tab. 2), but the latter difference was not significant for all sites. After BD fertilization, NH þ 4 -N concentrations were highest in samples from Dornburg and Gü lzow, which had the highest N fertilization rate (N fert ). In contrast, Dedelow samples featured the highest NH þ 4 -N concentrations after MIN fertilization. MIN samples from Dedelow also had the highest overall NO À 3 -N concentrations, followed by MIN and BD samples from Dornburg.

N 2 O and N 2 fluxes
The measured absolute N 2 O flux ranged from 37 to 1,221 μg N m -2 h -1 and from 23 to 140 μg N m -2 h -1 for BD and MIN samples, respectively (Fig. 1a). Across all sites, BD samples showed a significantly higher absolute N 2 O flux than MIN samples. However this effect was caused by samples from only two sites, particularly from Dornburg where BD samples exhibited a 27-fold higher N 2 O flux than MIN samples. Between sites, relative N 2 O fluxes were different for the BD treatment. However, a trend towards a difference in relative N 2 O fluxes between the BD and MIN treatment can be assumed from Tab. 4, but could not be confirmed with non-parametric statistics setting a significance level (p 0.05).
The absolute N 2 fluxes across all sites were significantly higher for BD than for MIN samples (Fig. 1b). Relative N 2 fluxes were not significantly different across soil sites, but on average between the two fertilizer types (Tab. 4). The site-specific differences of absolute N 2 flux between fertilization types were not significant, due to high between-sample variability (coefficient of variation 42 to 100%). Although differences among sites were generally less pronounced than differences between fertilization types, between-site variability of the N 2 O and N 2 flux was considerably higher for BD samples (Fig. 1)   In addition, the proportion of N 2 O relative to the total N 2 O+N 2 fluxes [also known as the N 2 O/ (N 2 O+N 2 ) ratio of denitrification] was calculated as an important assessment indicator. The N 2 flux was generally higher than the N 2 O flux, thus resulting in low average values for the N 2 O/(N 2 O+N 2 ) ratio of denitrification. The N 2 O/(N 2 O+N 2 ) ratio of denitrification was lower for BD than for MIN samples (Fig. 1c). Samples from Dedelow were an exception to this trend and showed similar high N 2 O/(N 2 O+N 2 ) ratios of denitrification (> 0.93) for both fertilizer types. However, due to the high variability, differences between the sites were not significant. There was no significant effect of fertilization on site-specific N 2 O/(N 2 O+N 2 ) ratios of denitrification, but particularly BD samples from Gü lzow and Ascha had considerably lower N 2 O/(N 2 O+N 2 ) ratios (0.23 to 0.29) than the respective MIN samples (0.63 to 0.81).

Controls on N 2 O and N 2 fluxes
N 2 O flux from MIN samples was significantly but weakly correlated with N t and NH þ 4 -N concentrations and (for relative N 2 O flux) with bulk density, while absolute and relative N 2 O flux from BD samples were strongly correlated with C org , N t , and NO À 3 -N concentrations (Tab. 5). GLM analysis showed NH þ 4 -N concentrations, site, WFPS, bulk density, and their interactions to be the most important factors for N 2 O flux (Tab. 6), which reflects the observed significant differences in the N 2 O flux between soil cores from different field sites. Other significant factors influencing the N 2 O flux were application type (i.e., injection vs. drag hose) and fertilization treatment.
While no correlations between absolute or relative N 2 flux and the control parameters were observed for MIN samples, the N 2 flux from BD samples was significantly correlated with NH þ 4 -N, pH, C org , N t (only the absolute N 2 flux), and WFPS. Overall, the main factor complexes controlling the N 2 flux were (1) fertilization treatment and its interactions with NH þ 4 -N concentration, followed by (2) WFPS, application type, and their interaction, while the relative importance of site was considerably smaller than for N 2 O (Tab. 6).
Like for N 2 , the N 2 O/(N 2 O+N 2 ) ratio of denitrification from BD samples was significantly correlated with soil NH þ 4 -N concentration, pH, and WFPS (Tab. 5). GLM analysis revealed that the N 2 O/(N 2 O+N 2 ) ratio of denitrification is mainly controlled by site, bulk density and NH þ 4 -N concentration (Tab. 6), and the interaction of NH þ 4 and WFPS, whereas the influence of treatment, WFPS, N t content and their interactions with soil parameters was of far lesser relevance than for N 2 O and N 2 fluxes. 2020   Note: Uppercase letters denote significant differences between treatments within site or across-site averages (non-parametric Mann-Whitney U-Test); lowercase letters denote significant differences among sites within treatment (non-parametric Kruskall-Wallis H-Test. post-hoc Student-Newman-Keuls test).

Discussion
A single application of mineral or organic fertilizer during sowing in spring is common practice in maize cultivation. Fertilization thus usually takes place under conditions favorable for N 2 O and N 2 emissions, i.e., high soil moisture levels in combination with increasing soil temperatures (Butterbach-Bahl et al., 2013). As a consequence, a major part of the annual N 2 O emissions is directly related to fertilization and often emitted within hours to days after fertilizer application (Heintze et al., 2017), together with a less known amount of N 2 , which in our experiment was found to be up to 2,773 μg N m -2 h -1 (average) within the first two to four days. The present experimental set-up allowed studying short-term N 2 O and N 2 fluxes shortly after fertilizer application with respect to different soil characteristics in Germany, but not aimed to study the dynamics after BD and MIN application on different field sites. In order to study N 2 O and N 2 dynamics, a longer incubation period (up to several weeks) with highfrequency measurements would have been necessary.

Impact of fertilizer type and application (BD vs. MIN)
In agreement with our first hypotheses, average N 2 O and N 2 fluxes were significantly higher from BD than from MIN treatments in soil cores across five sites (Fig. 1). Moreover lower average N 2 O/(N 2 O+N 2 ) ratios of denitrification were found in the BD soil cores across sites, except those from the Dedelow site. These findings are in line with a previous laboratory experiment by Senbayram et al. (2009), who reported that soils treated with BD derived from maize emitted more N 2 O than soils treated with MIN fertilizer. The average value of N 2 O fluxes following BD application was strongly influenced by the sample from the Dornburg site (Fig. 1a), where BD samples exhibited a 27-fold higher N 2 O flux than MIN samples, mainly due to the four times higher N application rate of BD compared to MIN fertilization. Despite slightly different boundary conditions (different biogas digstate, slightly higher temperature with 15°C), a subsequent laboratory study by Fiedler et al. (2017) showed higher N 2 O fluxes overall but compared to Gü lzow significantly higher N 2 O fluxes again for the Dornburg soil cores. On the other hand, Heintze et al. (2017) investigated N 2 O fluxes following BD application (160 kg N ha -1 ) for the Ascha site within a 21-day incubation experiment and found maximum mean N 2 O fluxes of 7.8 5.6 μg N m -2 h -1 . The measured N 2 O fluxes were thus up to ten times lower than in our study (i.e., 139 43 μg N m -2 h -1 ). Differences in N 2 O fluxes between the three related studies might be due to the well-known spatial and temporal variability of N 2 O, particularly because samples were taken in different years and soil cores were treated differently [i.e., drying and rewetting for WFPS adjustment in Fiedler et al. (2017)]. However, N 2 fluxes from the BD treat- ment between the two subsequent studies were found to be similar for the two sites (Dornburg and Gü lzow). As no MIN fertilizer treatment was investigated by Fiedler et al. (2017), a direct comparison between the two studies was not possible. N 2 fluxes were also higher from BD samples than MIN samples; hence, the calculated average N 2 O/(N 2 O+N 2 ) ratio of denitrification was lower for BD compared to MIN samples, even though N 2 fluxes were higher than N 2 O fluxes in both fertilizer treatments. Compared to another incubation study using BD samples with an N 2 release of up to 13 mg N m -2 h -1 within five days of incubation , present mean N 2 fluxes were considerably smaller. This might be explained by a slightly different set-up with BD application after two days of helium incubation, a higher fertilization rate of 250 kg N ha -1 and a significantly higher WFPS of 90%, which could favor N 2 O reduction to N 2 in the study by Kö ster et al. (2015). In the present case, N 2 fluxes were up to ten-fold higher than N 2 O fluxes, which underlined the assumption that a substantial share of N 2 O formed by denitrification was further reduced to N 2 [see also low N 2 O/(N 2 O+N 2 ) ratios of denitrification], although NO À 3 availability was limited (Tab. 2).

Effect of soil conditions from the different field sites
Soil biochemical processes in agricultural systems are influenced by physical and chemical soil conditions at the respective field sites. For fertilized agricultural sites, the impacts of fertilization rate, substrate composition and fertilizer application type have to be added as important influencing factors.

N level and substrate composition
As this study was not designed to test a large gradient of fertilizer input, the choice of N fertilization levels was based on typical local management preferences for maize cultivation. Local fertilization rates were based on the assumption that 70% of the N of the biogas digestate will be available for the plant in the year of application. Correspondingly, increased N doses via biogas digestate application were planned from the outset in this case to compensate for this factor, which thus increased the BD compared to the MIN fertilization rate. This resulted in a range of fertilizer N inputs from 202 to 332 kg N ha -1 in the BD treatments (Tab. 2) and likely the lack of any statistical effect of the N fertilization rate (N fert ) on short-term N 2 and N 2 O fluxes (Fig. 1). In particular, the Dornburg site exceeded the compensation by a large margin, which has led to significant differences between the two N fertilizer treatments (232 kg N ha -1 were applied in the BD treatment compared to 74 kg N ha -1 MIN treatment). These differences were reflected in the absolute and relative N 2 O fluxes at this site ( Fig. 1, Tab. 4). Some recent results have shown that N 2 O and N 2 fluxes did not increase with increasing the BD application rate (Fiedler et al., 2017), whereas Sä nger et al. (2014 found an increase in net nitrification rates with increasing application rates, which was linked to the greater availability of NH þ 4 . Moreover, the application of local BD instead of a standardized BD might have masked any site or treatment effects via introduction of uncertainty due to differences in BD quality, but all applied BD originated from co-fermentation of cattle slurry and maize silage featuring similar physical and chemical characteristics. While the dry matter content of the applied BD (6.6 to 8.4%) was slightly higher than reported for mono-fermented maize BD (4.9 to 5.9%; Senbayram et al., 2009;Svoboda et al., 2013), it was representative of cofermented slurry-maize BD (5.0 to 8.5%; Kluge et al., 2008). The observed range of pH and C:N ratios were also typical for mono-and co-fermented maize BD (pH of 7.6 to 8.6 and C:N ratio of 5.2 to 7.6; Senbayram et al., 2009;Chen et al., 2012;Svoboda et al., 2013). Regardless of the small differences in the applied BD fertilizers, digestate properties and therefore also the C and N dynamics are always influenced by the type of substrate input for anaerobic digestion (Sä nger et al., 2014). Due to the usage of BD fertilizers from local farmers, the quantities of BD fertilizers applied were to some extent different for the respective sites, which must be considered when interpreting results. When relative N 2 fluxes were calculated, no site differences within the BD treatment were found (Tab. 4). Nevertheless, the much higher variability of the N 2 fluxes of the BD treatment compared to N 2 fluxes of the MIN is an indication that the quantity and composition of the BD may to have a strong impact on the N 2 fluxes (i.e., the share of N 2 O reduction to N 2 ).
Higher N 2 O fluxes after BD application were likely related to stimulated microbial activity caused by the addition of some easily degradable C and N and the resulting consumption of soil O 2 , which in turn may enhance N losses due to denitrification, in particular also the reduction of N 2 O to N 2 (Petersen and Sommer, 2011;Kö ster et al., 2015;Heintze et al., 2017). Moreover, it can be assumed that the increased amount of NH þ 4 due to BD application, which was also found here, probably enhanced the process of nitrifier-denitrification, i.e., the oxidation of NH þ 4 to NO À 2 and its subsequent reduction to NO and N 2 O by autotrophic ammonia oxidizing microorganisms. Since nitrifier-denitrification particularly occurs in soils with high N availability but low C org , BD application has been found to stimulate this process under low O 2 availability (Eickenscheidt et al., 2014). However, a multitude of additional sources of N 2 O production has to be considered, due to the known wealth of N 2 O production processes and their simultaneous occurrence in soils (Butterbach-Bahl et al., 2013).
In addition to the discussed N 2 O, NO, and N 2 losses, a significant N loss via ammonia (NH 3 ) can also occur within the first hours after organic fertilizer application (Webb et al., 2010;Wolf et al., 2014;Ni et al., 2015), which should be mentioned for the sake of completeness. It is supposed that up to 15% of NH þ 4 is probably volatized as NH 3 within the first 10 h after surface application of BD (Quakernack et al., 2012), which might have also affected the share of N 2 O and N 2 losses.

Soil physical properties
Since the study sites were selected to represent a wide range of typical agricultural soil conditions, this study allowed the evaluation of the study results in light of common agricultural practice throughout Germany while lacking systematic gradients of potential impact factors. It is thus reasonable that the soil cores taken from five study sites represent a variety of soil physical properties regarding differences in soil moisture, soil type and texture, which also had a certain impact on the N gas release (see impact of site in the GLM). In particular, the parameter interaction of study site and bulk density, as well as soil moisture (i.e., WFPS) had a significant impact on the N 2 O fluxes. In the case of bulk density, the greater variability in the BD treatment might be due to the local, strip-like soil loosening during the incorporation of the applied BD using harrows or rotary cultivators and the possibility of not always exact removal of steel cylinders in the drag hose line. The on average higher proportion of WFPS in the BD treatment is due to localized application of high amounts of water in the form of liquid digestate, which will have influenced the N 2 O fluxes more than the N 2 fluxes. It might also be possible that MIN fertilizer granulates were not entirely dissolved at the time of soil core sampling, which might also lead to a certain bias. However, when looking at single impact factors, bulk density and WFPS were important factors affecting the N 2 , N 2 O release as well as the N 2 O/(N 2 O+N 2 ) ratio of denitrification (Tab. 6). For example, for the study site Dedelow, only a limited denitrification activity was previously assumed due to the sandy soil texture and low WFPS contents, which are positive for soil aeration (Del Grosso et al., 2000;Ball, 2013). However, N 2 O/(N 2 O+N 2 ) ratios of denitrification < 0.4 were found for the Dedelow samples, which indicates a great contribution of N 2 O reduction to N 2 . By contrast, high WFPS values (up to 81%) were found for BD in the Gü lzow samples, probably also resulting from the different fertilizer application technique via injection. Although soil moisture conditions were optimal for complete denitrification, N 2 fluxes ranged on a moderate level (460 224 μg N m -2 h -1 ). By using stable isotope labeling, Senbayram et al. (2009) also found denitrification to be the dominant process in soils having 58% and 76% WFPS values. However, for a WFPS value of 58% (Senbayram et al., 2009), a significant contribution of nitrification was revealed, which could be also the case for the other sites in this study (WFPS < 49%; Tab. 2).
Overall, these observations were more pronounced in the BD treatment than in the MIN treatment, which might be additionally promoted by the high water content of the digestate (BD treatment) compared to a dry granulated MIN fertilizer. BD contain mostly water (up to 97% moisture content), which can lead to the short-term saturation of micro-pores (Charles et al., 2017), thus limiting O 2 diffusivity and favoring denitrification (Tiedje et al., 1984). This may apply in particular for the Dornburg site, where in addition to the highest BD fertilization rates, which led to high NO À 3 -N and NH þ 4 -N (Tab. 2), also a great amount of water was added during BD application. Moreover, the Dornburg site has a very clayey silt soil texture, which often corresponds to limiting O 2 diffusivity and the greater potential for the occurrence of anaerobic microsites, which in combination with the effects of BD application could have favored conditions for denitrification and might explain the high N 2 fluxes for the Dornburg samples (Fig. 1).

Soil chemical properties
Regarding the chemical soil properties, large differences between the BD and MIN treatment were present (Tab. 2). BD application increased differences and variability among sites due to higher pH values, higher C org contents, higher NH þ 4 -N and lower NO À 3 -N concentrations in BD samples compared to MIN samples. However, it is not possible to distinguish whether the differences result from the application of the BD in this year or already resulted from the generally good supply of the soils with BD from the previous experimental year. Apart from soil physical properties, N availability (i.e., NH þ 4 and NO À 3 ) is known as an important driver for N 2 O and N 2 processes. Here, it seems that the higher N 2 O and in particular N 2 gas release from the BD compared to the MIN treatment was driven by the interaction of physical (i.e., high soil moisture) and chemical properties (i.e., high NH þ 4 -N concentrations and C contents) (see Tab. 5). Following BD application, the NH þ 4 -N concentration was enhanced, which probably stimulated the N 2 O release via nitrification (Kö ster et al., 2011). However, along with increased amounts of C org following BD application, which promote soil respiration and thus interact with O 2 availability, the potential for denitrification losses (i.e., N 2 O and N 2 fluxes) due to anaerobic soil conditions increased . This might be also the case here, although NO À 3 availability (i.e., substrate for denitrifiers) was limited. However, limited NO À 3 availability at the beginning can also promote the reduction of N 2 O ongoing nitrification in anaerobic microsites, because C org availability for denitrification becomes more and more depleted (i.e., decreasing CO 2 rates) and lower O 2 consumption allow nitrification (Weier et al., 1993;Kö ster et al., 2011). In turn, with increasing NO À 3 availability again, this would then have caused the shift back from N 2 to N 2 O as the main denitrification end product (Senbayram et al., 2009). Due to the focus on shortterm results from different sites within the present N 2 -free incubation experiment, it was not possible to investigate any interactions of NO À 3 and NH þ 4 dynamics.

Conclusions
Despite the limited scope of this experiment, this study provides valuable insights into short-term N 2 O and N 2 fluxes as well as the N 2 O/(N 2 O+N 2 ) ratio of denitrification following the initial phase of fertilizer application (biogas digestate vs. mineral fertilizer) from different field sites. On average, results demonstrated higher absolute N 2 O and N 2 fluxes from biogas digestate samples compared to mineral fertilizer samples. Relative (to fertilizer input) N 2 O fluxes were different between sites, but not across sites, while relative N 2 fluxes were only different across sites. Due to the higher N 2 than N 2 O fluxes, the low N 2 O/(N 2 O+N 2 ) ratios of denitrification highlighted the importance of N 2 O reduction to N 2 , in particular for the biogas digestate compared to the mineral fertilizer soil cores. While variability in N gas release between the investigated soil cores were lower for mineral fertilizer samples, soil conditions like bulk density and WFPS as proxies for gas diffusivity in the soil particularly affected the N 2 O production and reduction rates in the biogas digestate treatment. Nevertheless, the presented results also emphasize the need for further investigations of the dynamics and the duration of the observed effects and their significance under field conditions. However, a first estimate of the short-term N 2 gas release in the field during the initial phase might be possible, when applying the N 2 O/ (N 2 O+N 2 ) ratio of denitrification to field measurements.