Potential short-term losses of N 2 O and N 2 from high 1 concentrations of biogas digestate in arable soils 2

Abstract. Biogas digestate (BD) is increasingly used as organic fertilizer, but has a high potential for NH3 losses. Its proposed injection into soils as a countermeasure has been suggested to promote the generation of N2O, leading to a potential trade-off. Furthermore, the effect of high nutrient concentrations on N2 losses as they may appear after injection of BD into soil has not yet been evaluated. Hence, we performed an incubation experiment with soil cores in a helium–oxygen atmosphere to examine the influence of soil substrate (loamy sand, clayey silt), water-filled pore space (WFPS; 35, 55, 75 %) and application rate (0, 17.6 and 35.2 mL BD per soil core, 250 cm3) on the emission of N2O, N2 and CO2 after the usage of high loads of BD. To determine the potential capacity for gaseous losses, we applied anaerobic conditions by purging with helium for the last 24 h of incubation. Immediate N2O and N2 emissions as well as the N2 ∕ (N2O+N2) product ratio depended on soil type and increased with WFPS, indicating a crucial role of soil gas diffusivity for the formation and emission of nitrogenous gases in agricultural soils. However, emissions did not increase with the application rate of BD. This is probably due to an inhibitory effect of the high NH4+ content of BD on nitrification. Our results suggest a larger potential for N2O formation immediately following BD injection in the fine-textured clayey silt compared to the coarse loamy sand. By contrast, the loamy sand showed a higher potential for N2 production under anaerobic conditions. Our results suggest that short-term N losses of N2O and N2 after injection may be higher than probable losses of NH3 following surface application of BD.


Introduction
Nitrous oxide (N 2 O) is a potent greenhouse gas (Myhre et al., 2013), with agriculture being its single largest anthropogenic source, contributing about 4.1 Tg N 2 O-N yr −1 or 66 % of total gross anthropogenic emissions, mainly as a result of mineral nitrogen (N) fertilizer and manure application (Davidson and Kanter, 2014).The generation of nitrogen gas (N 2 ) is of agronomic interest in terms of nutrient management since such gaseous losses may imply a significant loss of N from the soil-plant system (Cameron et al., 2013;Friedl et al., 2016).However, from an environmental stance, N 2 is in-nocuous and, thus, the preferred type of gaseous N loss from soil (Davidson et al., 2015).In general, the improvement of N use efficiency and thus the decrease in N losses in crop production are paramount in the presence of challenges like food security, environmental degradation and climate change (Zhang et al., 2015).
Digestion residues (biogas digestate, BD) from biogas plants are used as organic amendment in agriculture.However, compared to undigested amendments, digestion results in an increased pH, a higher proportion of ammonium (NH + 4 ) and a narrowed C / N ratio of BD (Möller and Müller, 2012).

These altered chemical properties may promote biochemical
Published by Copernicus Publications on behalf of the European Geosciences Union.
reactions in the soil that are responsible for the formation of gaseous N species like N 2 O, nitric oxide (NO), N 2 and ammonia (NH 3 ) (Nkoa, 2013).
Significant losses of N as NH 3 may occur within the first hours after manure application (Quakernack et al., 2012).To reduce NH 3 losses, the application of BD by injection is recommended, but this measure can simultaneously increase the potential for N 2 O losses compared to surface application (Velthof and Mosquera, 2011;Wulf et al., 2002).On the one hand, high NH + 4 concentrations in the injection band promote nitrification, consuming O 2 and releasing N 2 O (Christensen and Rowe, 1984).On the other hand, increased amounts of C in the injection band also promote respiration, additionally depleting O 2 supply (Dell et al., 2011).Altogether, the conditions during the initial phase after injection of BD foster microsites favourable for microbial denitrification, which also promote the formation of N 2 due to anaerobic conditions (Köster et al., 2015;Webb et al., 2010).
There is a wealth of biotic and abiotic processes in soils that produce N 2 O and N 2 , depending on mineral N content, C availability and temperature, most of which are enhanced by anoxic or at least suboxic conditions (Butterbach-Bahl et al., 2013).The amounts and the relative share of N 2 and N 2 O in the overall gaseous N emissions depend -among other factors -on the degree of O 2 restriction (Firestone and Davidson, 1989).Soil physical and biotic factors (i.e.diffusion permitted by soil porosity in conjunction with water-filled pore space (WFPS) as well as consumption of O 2 by heterotrophic respiration and nitrification) control the aerobic status of a soil (Ball, 2013;Maag and Vinther, 1999;Uchida et al., 2008).In general, clayey soils exhibit a lower gas diffusivity compared to coarse-textured soils.This regularly results in higher denitrification in the former with higher N 2 O emission rates, but also a higher probability for the consecutive reduction to N 2 (Ball, 2013;Gu et al., 2013;Senbayram et al., 2014).
There is a general lack of knowledge about the effects of high BD concentration on gaseous N losses as they might appear after injection into soils and their interactions with O 2 limiting factors like soil texture and WFPS, as well as temperature and heterotrophic respiration.Thus, we applied the helium-oxygen (He-O 2 ) incubation technique (Butterbach-Bahl et al., 2002) in a laboratory experiment to evaluate the effect of the factors suggested above on the emission of N 2 O and N 2 from different soils.Simultaneously, CO 2 flux was determined as an indicator for microbial O 2 consumption, O 2 diffusion and also for the degradability of organic C applied with BD (Blagodatsky and Smith, 2012).We hypothesized that (1) N 2 O and N 2 emissions will increase with WFPS, (2) gaseous N losses will also be affected by BD application rate, i.e. the hypothetical concentration of C and N resulting from injection, and (3) the clayey silt will induce higher gaseous N losses than the coarse loamy sand. 2 Material and methods

Selected soils, sampling of soil cores and biogas digestate
Two soils were selected and both were adjusted to three levels of WFPS and three quantities of BD (Table 1), resulting in 18 factor combinations with three replicates.Temperature was increased from 2 • C during the first 2 days to 15  2 for more details on soil characteristics).After field sampling, the soil cores were dried for 48 h at 40 • C to facilitate adjustment of WFPS.
Both sites have been cultivated with similar crop rotations used as feedstock for biogas production and have been amended with BD for the past 9 years.The crop rotation on the sandy loam consisted of maize (Zea mays L.), rye (Secale cereale L.), sorghum (Sorghum bicolor (L.) Moench), winter triticale (× Triticosecale Wittmack), ryegrass (Lolium perenne L.) and winter wheat (Triticum aestivum L.).The only difference in the crop rotation on the clayey silt was the cultivation of sudan grass (Sorghum × drummondii) instead of sorghum.
The BD used for the incubation was obtained from a biogas plant at "Gut Dalwitz", an organic farm in northeast Germany.The feedstock for the anaerobic fermentation in the plant consisted of 60 % maize, 20 % solid cattle manure, 10 % dry chicken manure and 10 % rye.The digestate was analysed by LUFA Rostock, Germany, and had a pH of 8.3, 2.91 % organic C, 0.16 % dissolved organic C (DOC), 0.54 % Texture and mean values with standard deviations (in parentheses) for carbon (C, n = 9), nitrogen (N, n = 9), pH (n = 3), bulk density (BD, n = 3) and mineral N (NO − 3 and NH + 4 , n = 3) of both soils in 0-10 cm depth after field sampling.
Texture C (mg g −1 ) a N (mg g −1 ) a pH b Bulk density (g cm −3 ) c NO − 3 (mg kg  CS -High BD -55 % CS -High BD -75 % N and 0.27 % NH 4 -N in undried material with a dry matter content of 9.4 %.

Adjustment of WFPS and addition of N
For adjustment of WFPS, the dry and undisturbed soil cores were moistened dropwise.The respective quantities of water were calculated based on the bulk density and an assumed particle density of 2.65 g cm −1 and they were reduced by the expected moisture input from subsequent addition of BD.The soil cores were then mixed with BD and finally repacked to reach nutrient concentrations comparable to those in injection bands.The amounts of added BD were calculated with an assumed injection of 160 kg N ha −1 into the soil, with row spaces of 0.15 m (narrow injection bands with low BD concentration, LOBD) and 0.30 m (wide injection bands with high BD concentration, HIBD).These are common ranges used for injection machinery and correspond to 17.6 and 25.3 mL BD per sample ring.After this procedure, the soil cores were sealed with plastic lids and stored immediately at 2 • C until the beginning of the incubation within a week.Course of incubation and gas measurements with respect to atmosphere and temperature of the headspace after 2 days of preincubation at 2 • C in the He / O 2 gas mixture.Gas concentrations of the headspace were determined on 5 consecutive days, i.e.Monday to Friday in the morning.After the first 2 measurement days, the headspace temperature was increased from 2 to 15 • C. Additionally, after the fourth measurement day, the aerobic helium / oxygen gas mixture in the headspace was replaced by a pure helium atmosphere.

Determination of gas fluxes
The measurements of N 2 , N 2 O and CO 2 fluxes were applied following the He-O 2 method (Butterbach-Bahl et al., 2002;Scholefield et al., 1997).Six soil cores (i.e. the repetitions of two factor combinations at a time, Table 3) were placed simultaneously in special gas-tight incubation vessels inside a climate chamber.Analyses were conducted in the laboratory of the Institute for Landscape Biogeochemistry, Leibniz Centre for Agricultural Landscape Research (ZALF), Müncheberg, Germany.Before flux measurements, the vessels were evacuated to 0.047 bar and flushed with an artificial He / O 2 gas mixture (20.49% O 2 , 345.5 ppm CO 2 , 359 ppb N 2 O, 1863 ppb CH 4 , 2.46 ppm N 2 , rest He) four times consecutively to remove ambient N 2 .Subsequently, the air temperature of the climate chamber was set to 2 • C and a continuous He / O 2 gas flow rate of 15 mL min −1 was applied to the vessel headspace for 72 h to remove residues of N 2 from soil cores by diffusion, including a restricted N 2 production by decreased microbial activity.After this pre-incubation, on the following 2 days the headspace concentration of N 2 O and CO 2 was measured once daily in the morning.To compensate for the lower precision of the detector for N 2 in relation to the detector for N 2 O and CO 2 (Eickenscheidt et al., 2014), N 2 concentrations were measured consecutively three times daily in the morning.Immediately after the last measurement on the second day, the temperature was set to 15 • C and the measurements were continued for another 2 days.Finally, the He / O 2 gas mixture was substituted by pure He and, following 24 h of acclimatization, gas measurements were carried out once again (Fig. 1) to determine the generation of N 2 O and N 2 in a completely anaerobic soil matrix.The latter step is important to get a clue about the actual potential for gaseous N losses after highly concentrated BD application.The settings of the chromatographs for gas analyses are described in Eickenscheidt et al. (2014).Gas fluxes were calculated according to Eq. ( 1): where f is the flux (N 2 and CO 2 : mg m −2 h −1 , N 2 O: µg m −2 h −1 ), M the molar mass in g mol −1 (N 2 : 28, CO 2 : 44, N 2 O: 44), p the air pressure (Pa), v the air flow (L h −1 ), R the gas constant (8.31 J mol −1 K −1 ), T the temperature inside the chamber (K), A the area of the incubation vessel (m 2 ) and dc the difference of gas concentrations (N 2 and CO 2 : ppm, N 2 O: ppb) between inlet and outlet of a vessel.
To enhance the tightness against atmospheric N 2 contamination, the lids of the incubation vessels were permanently purged with helium.We obtained blank values by inserting aluminium blocks into the vessels before each measurement cycle.Since these blank values were usually steady with means of 1.9 (1σ = 0.9) ppm N 2 , 349.6 (1σ = 11.4) ppb N 2 O and 353.9 (1σ = 13.5)ppm CO 2 , we suggest that the vessels were tight.Derived from the blank values, the lowest detectable fluxes were on average 0.427 (1σ = 0 For flux estimation, the blank values were subtracted from the values measured at the respective outlet.Estimated fluxes from the soil cores smaller than the respective blank fluxes of each day were set to zero.

Soil analyses after incubation
After incubation, the soil cores were stored at 2 • C until they were extracted with 0.1 M KCl solution (the soil-toextract ratio was 1 : 4; standardized extraction method of the commissioned laboratory at Leibniz Centre for Agricultural Landscape Research e.V.) and analysed for NH + 4 and nitrate (NO − 3 ) using spectrophotometry according to DIN ISO 14256 with the continuous flow analyser CFA-SAN, Skalar Analytical B.V., the Netherlands, and for DOC by combustion according to DIN ISO 10694 with the analyser RC 612, Leco Instruments GmbH, Germany.

Statistical analysis
All statistical analyses were done using R statistical software version 3.2.3(R Core Team, 2016) with the data of the measuring days under a He-O 2 atmosphere.Data from the vessels with the factor combination of 35 % WFPS and LOBD with clayey silt were omitted due to technical reasons during sample preparation.For the final period of pure He headspace, some gas concentration data could not be documented.For loamy sand, this affects all WFPS levels with LOBD (N 2 and N 2 O), the treatment 75 % WFPS with 320 kg N h −1 (N 2 O and CO 2 ) and for the clayey silt the treatment 35 % WFPS without amendment (N 2 O and CO 2 ).
To account for repeated measurement of vessels, linear mixed effect models were applied with the R package "lmerTest" version 2.0-33 (Kuznetsova et al., 2016) for fluxes of each gas type.The three pseudo-replicated fluxes from the N 2 measurements of each vessel were averaged for each day to obtain the same number of observations as for N 2 O and CO 2 fluxes.The fixed structure of models included soil type, WFPS, amount of digestate, temperature, NO − 3 and DOC contents after incubation as well as the fluxes of N 2 O (in the model for N 2 ) and CO 2 (in the models for N 2 , N 2 O and N 2 / [N 2 +N 2 O] product ratio).Soil NH + 4 was omitted since it showed high autocorrelation with the amount of BD applied.The individual soil cores in the vessels were set as random effect (nested within the week of incubation and with allowance for a variable slope of the effect each day) with regard to lack of independence of consecutive measurements.The model responses for N 2 , N 2 O and CO 2 were log transformed (ln[value + 1]) since gas fluxes from soils usually show lognormal distributions (Kaiser et al., 1998).The function "step" was used for automatic backward selection of models based on AIC (Akaike's information criterion).The skewness (γ ) was calculated with the R package "moments" version 0.14 (Komsta and Novomestky, 2015) to check residuals for normal distribution and |γ | ≤ 2 was assumed as appropriate (West et al., 1995).For mixed-effect models, p values of the ANOVA (type 2) were calculated based on Satterthwaite's approximation.
Cumulated gas fluxes were estimated with a bootstrap method using function "auc.mc" of the R package "flux" version 0.3-0 (Jurasinski et al., 2014).In short, the fluxes for the period of aerobic headspace were cumulated in 100 iterations, while for each run two fluxes were omitted randomly.Then, the resulting data were used to calculate means and standard deviations.126.7 mg (LOBD) and 253.4 mg (HIBD).The NO − 3 content of BD was negligible.In general, the NH + 4 content of the soils after incubation increased with digestate application, with lower amounts detected in the clayey silt.Nitrate was found almost exclusively in the latter soil (Fig. 2).
The amounts of measured DOC increased with the application rate of BD, but with higher magnitudes for the loamy sand than for the clayey silt (Table 4).

CO 2 fluxes
CO 2 fluxes showed clear differences between the soils: under all combinations of temperature and oxygen, the fluxes were always larger from loamy sand compared to clayey silt (Table A1).In general, mean fluxes from loamy sand increased with the amount of digestate during each of the different periods regarding temperature and headspace aerobicity, but showed no obvious pattern with WFPS.There was no clear trend of fluxes with the amount of amendment, but a slight trend of decreasing fluxes with increasing WFPS could be seen for the clayey silt.However, the predictive power of WFPS on CO 2 -C fluxes was minor since it was eliminated during stepwise regression fitting.By contrast, soil type, amount of digestate, temperature and the DOC content after the incubation had significant (p < 0.01) effects (Table 5).

N 2 O fluxes
The mean N 2 O fluxes from the loamy sand in the He-O 2 headspace were virtually zero, independent of temperature and WFPS as well as the amount of BD application (Fig. 3, Table A2).In contrast, the emissions of the clayey silt increased with temperature and were highest at 15 • C with intermediate WFPS and amount of BD, i.e. 6.2 mg N 2 O-N m −2 h −1 at 55 % with LOBD, respectively.Surprisingly, at 15 • C, increasing the amount of BD up to HIBD did not increase the observed N 2 O efflux; rather, it decreased the efflux significantly (p < 0.05, Tukey's HSD) at 55 % and also, but not significantly, at 75 % WFPS (Fig. 3, Table A2).According to the linear mixed model for N 2 O fluxes in aerobic conditions, WFPS, amount of digestate, temperature, DOC content of soil after incubation and CO 2 fluxes had significant (p < 0.001) effects on N 2 O flux (Table 5).
Under anaerobic headspace conditions, the overall highest mean N 2 O flux was observed from the clayey silt at 35 % WFPS with HIBD (11.7 mg N 2 O-N m −2 h −1 ).The same soil showed a tendency of decreasing N 2 O fluxes with increasing WFPS and amendment.In the loamy sand, the pure He atmosphere induced increasing mean N 2 O fluxes (up to 1.3 mg N 2 O-N m −2 h −1 ) with increasing WFPS (Fig. 3, Table A2).Thus, the anaerobic headspace only induced a change in the loamy sand by increasing emissions.

N 2 fluxes
From the loamy sand, no or only small rates of N 2 were detected at both temperatures under a He-O 2 atmosphere (Fig. 4, Table A3).The clayey silt showed mean fluxes of up to 1.4 mg N 2 m −2 h −1 at 2 • C (all incubations with 75 % WFPS) and up to 3.8 mg N 2 m −2 h −1 at 15 • C (75 % WFPS with LOBD), but no fluxes in all BD treatments with 35 % WFPS.Put simply, temperature had a small effect on N 2 emissions from the sandy loam, with no consistent influence of WFPS and the amount of BD.In contrast, the clayey silt emitted increasing fluxes with increasing temperature and WFPS.However, the application rise from LOBD up to HIBD at 15 • C resulted in slightly, but not significantly (p > 0.05, Tukey's HSD) decreased fluxes (Fig. 4, Table A3).The summary of the linear mixed model for N 2 fluxes under aerobic conditions revealed significant effects (p < 0.05) of soil type, WFPS, the amount of digestate, temperature, DOC content after incubation and N 2 O flux (Table 5).
After switching the atmosphere to pure He, the N 2 fluxes from the sandy loam increased more than 60-fold.In contrast to aerobic conditions, all measured factor combinations showed mean fluxes of up to 35.1 mg N 2 m −2 h −1 (55 % with 320 kg N ha −1 ) (Fig. 2, day 5 in Table A3).Mean fluxes from clayey silt increased only up to 9.3 mg N 2 m −2 h −1 in amended treatments.Thus, the loamy sand exhibited a much more intense reaction under anaerobic headspace conditions.
No clear trend of the product ratio of N 2 / (N 2 + N 2 O-N) was found for incubations of the loamy sand.However, there was a clear distinction between the ratios for this soil under aerowww.soil-journal.net/3/161/2017/SOIL, 3, 161-176, 2017  bic and anaerobic atmospheres: while the ratios were close to zero in the former, they were close to 1 in the latter (Fig. 5).
In contrast, in the clayey silt the ratios increased with WFPS and were affected by digestate amendment under both the aerobic and the anaerobic atmospheres, in which the highest ratios (up to 0.8) were found in treatments without digestate and at least 55 % WFPS.The digestate-amended treatments showed ratios around or above 0.5, with the exception of the 35 % WFPS treatments, which had ratios close to zero.According to the linear mixed model, the product ratio under aerobic conditions was affected significantly (p < 0.01) by soil type and the amount of digestate (Table 5).

Discussion
4.1 Increased BD application rate did not increase N 2 O and N 2 losses, probably due to inhibitory effect of high NH + 4 concentrations In the loamy sand, the higher NH + 4 content measured after the incubation cycle compared to the calculated NH + 4 appli-cation rates may result from heterogeneity of BD itself (Andruschkewitsch et al., 2013).By contrast, the considerable lower values after incubation in the clayey silt could be attributed to a higher fixation of NH + 4 as NH 3 by clay minerals, enhanced by the increased pH of BD (Kissel et al., 2008).
The overall N 2 O fluxes corresponded well with those from other studies with similar incubation conditions and application rates of BD in terms of nitrogen per hectare (Köster et al., 2015;Senbayram et al., 2012;Severin et al., 2015).However, the latter studies assumed a distribution of BD into soil using a cultivator, which implies a smaller concentration of BD than we actually applied.Although we observed differences in N 2 O emissions between soils, soil type was not confirmed as a significant effect.Nevertheless, WFPS and temperature, which are well-known controls of N 2 O generation (Maag and Vinther, 1999), showed significant influences.Both are physical (by gas diffusion) and biological (by increased metabolic activity and consequently increased O 2 consumption by respiration) drivers for O 2 availability (Ball, 2013;Maag and Vinther, 1999).Accordingly, CO 2 flux (resulting from respiration of O 2 ) generally increased with tem-subd$mean[subd$Nlevel == "0N"] 0 2 4 6 8 10 12 14 q q q q q q q q q No BD Low BD High BD 2 °C q q q Mon subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Tues subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Wed subd$mean[subd$Nlevel == "0N"] q q q q q q 15 °C He/O 2 gas mixture q q q Thurs subd$mean[subd$Nlevel == "0N"] 0 2 4 6 8 10 12 14 Fri q q q q q He only 15 °C Loamy sand subd$mean[subd$Nlevel == "0N"] 0 2 4 6 8 10 12 14 q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 WFPS (%) subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] 35 55 75 0 2 4 6 8 10 12 14 q q q q q q q q Clayey silt Figure 3. Mean N 2 O fluxes (mg N m −2 h −1 ) from loamy sand and clayey silt incubated under different water-filled pore spaces (WFPSs, %) with different amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD).The first to the fourth days of the incubation were measured in an aerobic He-O 2 headspace (with 2 days at 2 • C followed by another 2 days at 15 • C) while on the fifth day measurements were conducted in an anaerobic headspace with pure He (at 15 • C).Error bars show standard deviations; if bars are not visible, they are smaller than the symbols of the means.Under an aerobic atmosphere, N 2 O fluxes from loamy sand were negligible, while fluxes from clayey silt showed an increase with temperature, especially with higher WFPS and intermediate amounts of digestate.Under an anaerobic atmosphere, mean fluxes from loamy sand increased slightly, but significantly (Tukey's HSD, p < 0.05).The fluxes from clayey silt showed no significant differences (Tukey's HSD, p < 0.05) compared to the day before, with the exception of 35 % WFPS, at which mean flux increased strongly in the treatment with 320 kg digestate N ha −1 .
subd$mean[subd$Nlevel == "0N"] 0 1 2 3 4 5 q q q q q q q q q No BD Low BD High BD 2 °C q q q Mon subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Tues subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Wed subd$mean[subd$Nlevel == "0N"] q q q q q q 15 °C He/O 2 gas mixture q q q Thurs subd$mean[subd$Nlevel == "0N"] 0 10 20 30 40 50 Fri q q q q q q He only 15 °C Loamy sand subd$mean[subd$Nlevel == "0N"] 0 1 2 3 4 5 q q q q q q q q 35 55 75 mg N 2 m −2 h −1 subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 WFPS (%) subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] 35 55 75 0 10 20 30 40 50 q q q q q q q q Clayey silt Figure 4. Mean N 2 fluxes (mg m −2 h −1 ) from loamy sand and clayey silt incubated under different water-filled pore spaces (WFPSs, %) with different amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD).The first to the fourth days of the incubation were measured in an aerobic He-O 2 headspace (with 2 days at 2 • C followed by another 2 days at 15 • C) while on the fifth day measurements were conducted in an anaerobic headspace with pure He (at 15 • C).Error bars show standard deviations; if bars are not visible, they are smaller than the symbols of the means.The dotted horizontal lines depict the average blank value; single flux rates lower than the respective lank value were set zero.Under an aerobic atmosphere, N 2 fluxes from loamy sand were zero or rather negligible, while fluxes from clayey silt show a distinct increase with WFPS and higher fluxes at 15 • C.Under an anaerobic atmosphere, mean fluxes from loamy sand increased by orders of magnitude, while the fluxes from clayey silt increased as well, but more gently compared to the sand.
perature and was also identified as significant by regression selection.
The mean N 2 fluxes of up to 0.5 (loamy sand) and 3.8 mg N m −2 h −1 (clayey silt) at 15 • C (Fig. 5, Table A3) were considerably smaller than the mean fluxes of up to 13.0 mg m −2 h −1 observed by Köster et al. (2015) during the first 5 days of their incubation.Although the amount of BD in terms of applied N (250 kg ha −1 ) was comparable, Köster 1.0 q q q q q q q q No BD Low BD High BD 2 °C q Mon subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Tues subd$mean[subd$Nlevel == "0N"] q q q q q q q q q Wed subd$mean[subd$Nlevel == "0N"] q q q q q q 15 °C He/O 2 gas mixture 1.0 q q q q q q q 35 55 75 q q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] q q q q q q q q 35 55 75 WFPS (%) subd$mean[subd$Nlevel == "0N"] q q q q q q q 35 55 75 subd$mean[subd$Nlevel == "0N"] 35 55 75 0.0 0.2 0.4 0.6 0.8 1.0 q q q q q q q Clayey silt Figure 5. Mean N 2 (N 2 + N 2 O-N) product ratio from loamy sand and clayey silt incubated under different water-filled pore spaces (WFPSs, %) with different amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD).The first to the fourth days of the incubation were measured in an aerobic He-O 2 headspace (with 2 days at 2 • C followed by another 2 days at 15 • C) while on the fifth day measurements were conducted in an anaerobic headspace with pure He (at 15 • C).Error bars show standard deviations; if bars are not visible, they are smaller than the symbols of the means.For the loamy sand, there was a clear distinction of the ratios between aerobic and anaerobic atmospheres: while the ratios tended to 0 in the former, they tended to 1 in the latter, regardless of temperature or amount of digestate.For the clayey silt, ratios increased with WFPS and were highest from the unamended treatments under both the aerobic and the anaerobic atmospheres.
et al. ( 2015) used a higher WFPS of 90 %, which may have increased the generation of N 2 .In contrast to N 2 O emission rates, the observed N 2 fluxes depended not only on WFPS but also on soil type (Table 5).This is most likely due to the direct influence of soil structure on diffusivity and the resulting supply with O 2 (Balaine et al., 2016;Butterbach-Bahl et al., 2013).N 2 O flux also showed a significant effect during regression selection for N 2 .N 2 O is the direct precursor of N 2 in denitrification; thus, the flux of the latter depends on the availability of the former.However, temperature showed no significant effect.N 2 / (N 2 + N 2 O) product ratios were significantly determined only by soil type and WFPS: while no clear trend was observable for the loamy sand, there was a pronounced effect in the clayey silt (Fig. 4).We attribute the lack of a trend in loamy sand to generally adverse conditions for the formation of N 2 O and N 2 , i.e. a sufficient supply of O 2 (see Sect. 4.2).Conversely, the influence of WFPS apparently mirrored favourable conditions in the clayey silt (Table 5).Simultaneously, with increasing WFPS, the reduction of N 2 O as an alternative electron acceptor under reduced O 2 supply accelerates (Tiedje, 1988).Accordingly, no or rather small fluxes of the investigated gaseous N species were generally found in our presumably well-aerated treatments with 35 % WFPS.
In our study, one treatment (clayey silt, 55 % WFPS, LOBD) showed exceptionally large mean N 2 O fluxes of up to 7.1 mg N m −2 h −1 (Fig. 3, Table A2).This could be evidence that the injection of such commonly applied amounts of BD-N (i.e.160 kg N ha −1 ) may favour much larger losses of N 2 O compared to an even distribution of BD in a soil surface due to larger substrate concentration in injection slits.However, with higher amendments (i.e.HIBD), we observed partially significant (p < 0.05, Tukey's HSD) reductions of N 2 O and a decreasing tendency of N 2 emissions (Tables A2, A3).In line with this, the amount of BD showed a significant effect during the regression selection on N 2 O, but not on N 2 fluxes (Table 5).A coherent reasoning for the rather smaller emissions of highly amended HIBD treatments might lie in an inhibitory effect of NH 3 on nitrification.Accordingly, Kim et al. (2006) found a selective inhibition of NO − 2 oxidation in the presence of 14 to 17 mg NH 3 -N L −1 .Our calculated application rates in the treatments with HIBD amounted to at least 253.4 mg NH + 4 -N (kg soil) −1 for the clayey silt (Fig. 3), which corresponds to 13.0 mg NH 3 -N (kg soil) −1 at 15 • C when applying the pH of the BD and assuming all extractable NH + 4 -N to be in solution (Emerson et al., 1975).Hence, we consider this inhibitory effect as the reason for the missing increase in N 2 O and N 2 .Nevertheless, because we mixed the BD with the soil, one would expect a lower in situ NH 3 fixation by clay minerals in tubular injection slits (Kissel et al., 2008), resulting in probably lower N 2 O and N 2 fluxes from clayey soils due to a more marked inhibitory effect.
High NH + 4 loads in conjunction with an increased pH favour NO − 2 accumulation because NO − 2 -oxidizing bacteria are less resilient against high concentrations of NH 3 than NH 3 -oxidizing bacteria (Anthonisen et al., 1976).This NO − 2 should have protonated then partly to toxic and unstable HNO 2 , which drives biological and chemical production of SOIL, 3, 161-176, 2017 www.soil-journal.net/3/161/2017/NO and N 2 O for detoxification (Venterea et al., 2015).Although we did not determine NO − 2 , we suggest a dominant role of nitrifier denitrification, i.e.NO − 2 reduction, in the generation of N 2 O during our experiment, especially during the anaerobic headspace conditions at the end of the incubation, resulting in the relatively small NO − 3 recovery in both soils.Accordingly, coupled nitrification-denitrification and bacterial denitrification have been found to dominate the production of N 2 O directly after application of BD (Köster et al., 2011;Senbayram et al., 2009).However, N 2 O-N losses were clearly larger than N 2 losses under aerobic headspace in the clayey silt.This indicates that much of the N gas loss was driven by processes other than canonical denitrification.Under the above-mentioned conditions, NO-N losses may exceed N 2 O losses (Venterea et al., 2015), making it important to take NO measurements into account in future studies.
Supposing that 15 % of NH + 4 -N is volatilized as NH 3 within the first 10 h after surface application of BD (Quakernack et al., 2012), the losses from the NH + 4 amounts we applied would average to 80 mg NH 3 -N m −2 h −1 (LOBD) and 160 mg H 3 -N m −2 h −1 (HIBD).The actual losses of up to 11.7 mg N 2 O-N m −2 h −1 at 30 % WFPS in the clayey soil (Table A2) or of up to 35.1 mg N 2 m −2 h −1 at 55 % WFPS in the sandy loam (Table A3) from our HIBD treatments add up to 117 mg N 2 O-N and 351 mg N 2 , respectively, for the same period.Hence, increased N 2 O and N 2 emissions following injection of BD might effectively cause higher N losses compared to a surface application and deserve closer attention in future.

Different effects of soil diffusivity on N 2 O and N 2 fluxes
Apparently, the tested factors affected the N 2 O and N 2 fluxes from both soils in a different way.A specific soil characteristic that exhibits such a fundamental control on biogeochemical processes such as denitrification is the diffusivity for O 2 (Ball, 2013;Letey et al., 1980;Parkin and Tiedje, 1984), which is a main soil characteristic responsible for the appearance of anaerobic microsites.In general, diffusivity integrates the soil porosity, i.e. pore continuity and size as well as WFPS, which control both soil N 2 O and N 2 emissions (Balaine et al., 2016;Ball, 2013;Letey et al., 1980).Soils with a coarser texture like the loamy sand have a higher proportion of macropores and thus a higher gas diffusion compared with fine-textured soils like the clayey silt we used (Groffman and Tiedje, 1991).This lets us expect conditions that are more favourable for N 2 O and N 2 generation in the latter due to relatively poor diffusion characteristics and, thus, a smaller O 2 supply.Actually, although we incubated the soils at comparable levels of WFPS and BD amendments, the apparent lower diffusivity led to larger N 2 O and N 2 production in the treatments with the clayey silt in relation to the loamy sand.The role of the distinct diffusivities of both soils is corroborated by our observations of the gas fluxes in anaero- bic headspace.With switching the He-O 2 atmosphere in the headspace to pure He, the denitrification potential can be tested because anaerobicity eliminates respiration processes that use O 2 as an electron acceptor (Parkin and Tiedje, 1984).
We acknowledge DNRA and anammox, for example, as possible additional sources of N 2 O and N 2 under such conditions but we were not able to quantify their contribution.
The anaerobic headspace induced a considerable increase in N 2 O fluxes in the loamy sand, but not in the clayey silt.Concurrently, the N 2 fluxes increased in both soils, but pronouncedly, i.e. more than 60-fold, in the sandy loam.These observed changes resulting from oxygen deprivation imply that, during the previous aerobic conditions, the diffusivity of the sandy loam was too high to allow for a sufficient establishment of anaerobic microsites, while the clayey silt ensured a moderate diffusional constraint to maintain suboxic conditions.In general, only N 2 O fluxes from treatments with negligible fluxes during the previous aerobic period increased under anaerobic conditions, including all treatments with loamy sand (Fig. 3, Table A2).At the same time, there was a reduction of N 2 O fluxes in most clayey silt treatments.However, a closer look reveals that virtually all of the latter treatments showed increased N 2 flux rates.Hence, there was an enhanced reduction of N 2 O to N 2 , which is reflected in the increased N 2 / (N 2 + N 2 O) product ratio (Fig. 5) and points to intensified reduction of N 2 O due to the lack of oxygen (Parkin and Tiedje, 1984).The much larger N 2 fluxes from the loamy sand compared to the clayey silt might have been caused as well by poor NO − 3 availability (Fig. 2) and a high availability of C (Table 4), which promoted the reduction of N 2 O to N 2 (Senbayram et al., 2012).Further, we found no evidence for any shortage of substrate in the clayey silt during the subsequent anaerobic headspace conditions.However, the cumulated fluxes of both N 2 and N 2 O amounted to a maximum absolute loss of 9.4 (1σ = 0.3) mg N per kg soil in the clayey silt with LOBD and 55 % WFPS, which was 7.4 % www.soil-journal.net/3/161/2017/SOIL, 3, 161-176, 2017 of the calculated NH + 4 -N applied with BD (Fig. 2).Conversely, the N 2 / (N 2 +N 2 O) product ratios increased only slightly (Fig. 5) and, in contrast to the loamy sand, there were still significant N 2 O fluxes in the clayey silt (Fig. 3).This points to still-sufficient stocks of NO − 3 in the latter (Senbayram et al., 2012).In fact, the NO − 3 stock was greater in the clayey silt than in loamy sand after incubation (Fig. 2).Thus, we suggest that the gas fluxes were unaffected by the change to anaerobic headspace in the clayey silt due to already low O 2 concentrations as a result of poor diffusivity.In conclusion, distinct gas diffusivities of both soils can be proposed as the main reason for the differing N 2 O and N 2 fluxes.
In interaction with soil diffusivity, respiration affects the aerobicity of a soil matrix by concurrent consumption and formation of O 2 and CO 2 as well.Depending on microbial C availability, respiration could be indicated by DOC, though not all DOC may be readily degradable (Cook and Allan, 1992).Generally, DOC content after our incubation increased with application rate of BD (Table 4), but DOC content was always smaller in clayey silt.This might reflect a stronger sorption of C and thus a lower availability for respiration in the clayey silt compared to loamy sand (Kaiser and Guggenberger, 2000).If we compare DOC concentrations with cumulated flux rates of CO 2 over the period of aerobic headspace, we find a good regression fit (R 2 = 0.91, p < 0.001) for both soils (Fig. 6), indicating a sufficient availability of C from BD for respiration and, thus, implicitly also for denitrification (Reddy et al., 1982).Moreover, as increased DOC enhanced respiration (Table A1), it consequently affected O 2 consumption and, thus, also the emergence of anaerobic microsites (Azam et al., 2002).Accordingly, there is also a good correlation between cumulated CO 2 and N 2 O + N 2 fluxes for the same period from the clayey silt (R 2 = 0.93, p = 0.001), when the treatments with 35 % WFPS (which showed virtually no N emissions) are omitted (Fig. 7).However, there was no such a correlation for the loamy sand.This confirms the interactive effect of diffusivity (induced by both the soils and WFPS) and C availability on the emissions of N 2 O and N 2 , which, nevertheless, interacted with the inhibitory effect of high NH + 4 loads on nitrification (see Sect. 4.1).

Relevance and implications
Our aim was to estimate the effect of differing soil environmental conditions on gaseous N losses -and not to draw conclusions about the long-term dynamics of N 2 and N 2 O emissions after BD application in concentrations similar to injection.In another laboratory study at a WFPS of 65 %, Senbayram et al. ( 2009) measured only one peak within 2 days without a repeated increase later, regardless of the amount of applied BD.Thus, we assume that a single peak shortly after application holds also true for our incubation.We also assume the measurements after only 24 h of anaerobicity in the headspace to be representative for the emission potential since in similar studies Wang et al. (2011Wang et al. ( , 2013) ) showed that the emission of N 2 and N 2 O peaked within less than 24 h after switching their headspace from aerobic to anaerobic conditions.
In summary, as hypothesized, N 2 O and N 2 emissions as well as the N 2 / (N 2 O+N 2 ) product ratio increased with WFPS, most probably due to restricted supply of O 2 .Contrary to our second hypothesis, the gaseous losses of N 2 O and N 2 did not increase with the application rate of BD.This indicates an inhibitory effect of high NH 3 and NH + 4 concentrations on nitrification, which are typically found in BD.At the same time, the N 2 / (N 2 O+N 2 ) product ratio tended to decrease with application rate as supposed, probably due to a copious supply with NO − 2 and NO − 3 from oxidized BD-NH + 4 .Confirming our third hypothesis, the fine textured clayey silt induced larger gaseous N losses and a higher N 2 / (N 2 O+N 2 ) ratio than the coarse loamy sand by the apparent distinct diffusivities of both soils.Overall, there was a larger potential for formation of N 2 O in the fine-textured clayey silt compared to the coarse loamy sand after the application of high concentrations of BD as they may appear after injection.However, the loamy sand showed a large potential for N 2 formation under anaerobic headspace conditions.
Since coupled nitrification-denitrification N losses from injected BD seem to be massive in this study, the short-term emissions of N 2 O and N 2 after injection appear to offset the reduced NH 3 -N losses that would have hypothetically arisen from surface application.Further investigations are needed regarding the dynamics and the duration of the observed effects and their reliability for field conditions.Table A2.Mean N 2 O-N fluxes with standard deviations in mg m −2 h −1 from the loamy sand and the clayey silt, treated with different waterfilled pore spaces (WFPSs, %), amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD) and different temperature regimes ( • C) under aerobic (He-O 2 ) and anaerobic (He) atmospheres.The column "Day" denotes the consecutive measuring days of the respective incubation cycle.Different letters after fluxes indicate significant differences (Tukey's HSD, p < 0.05) within each soil and measuring day.Zeros as last digits were omitted.
Day Atmosphere Temperature ( Figure1.Course of incubation and gas measurements with respect to atmosphere and temperature of the headspace after 2 days of preincubation at 2 • C in the He / O 2 gas mixture.Gas concentrations of the headspace were determined on 5 consecutive days, i.e.Monday to Friday in the morning.After the first 2 measurement days, the headspace temperature was increased from 2 to 15 • C. Additionally, after the fourth measurement day, the aerobic helium / oxygen gas mixture in the headspace was replaced by a pure helium atmosphere.

Figure 6 .
Figure6.Regression between DOC (mg per 100 g soil) measured after the incubation and the respective cumulated CO 2 emissions (g C m −2 ) during the period of aerobic headspace with their standard deviations and confidence interval (95 %).If error bars are not visible, they are smaller than the symbols of the means.Both soils showed increasing emissions with increasing soil DOC contents as well as a good regression fit (R 2 = 0.91, p < 0.001).

Figure 7 .
Figure 7. Regression between cumulated CO 2 emissions (g C m −2 ) and the respective cumulated N 2 O + N 2 emissions (g N m −2 ) from the clayey silt with WFPS > 35 % during the period of aerobic headspace with their standard deviations and confidence interval (95 %).If error bars are not visible, they are smaller than the symbols of the means.The proportional increase in CO 2 and the N gas species shows a good regression fit of R 2 = 0.93 (p = 0.001).
Mean N 2 fluxes with standard deviations in mg m −2 h −1 from the loamy sand and the clayey silt, treated with different waterfilled pore spaces (WFPSs, %), amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD) and different temperature regimes ( • C) under aerobic (He-O 2 ) and anaerobic (He) atmospheres.The column "Day" denotes the consecutive measuring days of the respective incubation cycle.Different letters after fluxes indicate significant differences (Tukey's HSD, p < 0.05) within each soil and measuring day.Zeros as last digits were omitted.Atmosphere Temperature ( • C) WFPS (%) mg N 2 m −2 ,h −1 26.8 ± 1.1 b 6.7 ± 0.8 b

Table 1 .
The examined factors soil texture, water-filled pore space (WFPS) and amount (i.e.concentration) of nitrogen (N) applied with biogas digestate (BD) with their respective levels applied in the present study, resulting in 18 treatments with three replicates each.The temperature was manipulated consecutively during the incubation.

Table 2 .
Characteristics of both soils.

Table 3 .
Chronological order of the incubated factor combinations.
Ammonium and nitrate contents from loamy sand and clayey silt after incubation with different water-filled pore spaces (WFPSs, %) and amounts of digestate (0 mL per sample ring: no BD, 17.6 mL: low BD and 35.2 mL: high BD).Error bars denote standard deviations.In general, the ammonium content increased with digestate application, with lower amounts detected in the clayey silt.Nitrate was found almost exclusively in the latter soil.For comparison, inverted triangles show calculated amounts of applied ammonium, which may differ from actual rates due to heterogeneity of biogas digestate.One treatment (*) was omitted from all analyses due to technical reasons.

Table 4 .
Mean DOC values from soils, measured after incubation, with standard deviations in parentheses for the respective treatments differing in amount of applied biogas digestate (BD) and water-filled pore space (WFPS).

Table 5 .
ANOVA table (type 2, p values calculated based on Satterthwaite's approximation) of the linear mixed-effect models for estimated fluxes of N 2 , N 2 O, N 2 / (N 2 +N 2 O) product ratio and CO 2 in an aerobic He-O 2 atmosphere.Soil type, water-filled pore space (WFPS), amount of digestate, temperature, NO − 3 and DOC content of soil after incubation, and fluxes of N 2 O and CO 2 were set as possible independent variables.The individual soil rings were set as the random effect (nested within the respective week and with the allowance for varying slopes for each day of measurements).The random effect was always significant.
a Variable eliminated during stepwise regression selection.b Variable was not included in the original regression.