Introduction

Denitrification, which is the microbial reduction of nitrate (NO −3 ) or nitrite via nitric oxide (NO) and nitrous oxide (N2O) to nitrogen gas (N2) (Firestone and Davidson 1989; Ye et al. 1994), plays an important role in the global nitrogen cycle because it removes reactive nitrogen from terrestrial and aquatic ecosystems (e.g., Davidson and Seitzinger 2006; Groffman et al. 2006; Seitzinger et al. 2006). In the absence of denitrification, biological or industrial N2 fixation would eventually deplete the atmospheric N2, and the biosphere would be inundated with nitrate (Robertson and Groffman 2007). Moreover, N2O and NO are primary and secondary greenhouse gases, respectively; and they are also important components influencing the atmospheric chemistry (IPCC 2007). Thus emissions of N2O and NO from upland soils have been intensively investigated (e.g., Bouwman et al. 2002; Stehfest and Bouwman 2006).

Despite the ecological importance of denitrification with N2 as end product, little information regarding N2 emissions from upland soils has been obtained by direct measurements. This is mainly due to methodological constraints that do not allow reliable direct measurements against the high background level of 78 % N2 in the atmosphere (Groffman et al. 2006). So far, indirect measurements of N2 emissions from upland soils have been widely performed with the acetylene (C2H2) inhibition technique (e.g., Ciarlo et al. 2008; Yoshinari et al. 1977) and/or the 15N tracing method (e.g., Mathieu et al. 2006; Ruser et al. 2006). The former technique has been criticized because the denitrification rates can be severely underestimated by even up to a factor of 10 (Bollmann and Conrad 1997), mainly due to a) inhibition of nitrification (Mosier 1980), b) incomplete blockage of the N2O reductase if the soil NO −3 concentrations are low (Simarmata et al. 1993), and c) uneven distribution of the inhibitor (Jordan et al. 1998). The latter method is insensitive to measure N2 emissions, due to a high detection limit of 180 μg N m−2 h−1 or 43 g N ha−1 d−1 (Ruser et al. 2006). Recently, the gas-flow-soil-core technique has been increasingly used for direct quantification of N2 emissions from soils (Butterbach-Bahl et al. 2002; Cárdenas et al. 2003; Dannenmann et al. 2008; De Wever et al. 2002; Scholefield et al. 1997a,b; Wang et al. 2011; Senbayram et al. 2011). Using this method, the soil air is replaced with an artificial N2-free gas, and then direct measurements of N2, N2O and NO emissions are performed. The shortcomings of this technique are mainly associated with the a) long time (20–48 h) needed to firmly establish an N2-free environment, b) low sensitivity of N2 detection (>10 μg N m−2 h−1), and c) strict gas-tightness requirement of the incubation system (Groffman et al. 2006). To overcome these shortcomings, the modification of hardware and an improvement in operational procedures have been recently achieved for the gas-flow-soil-core technique, which include a) using a new design of incubation vessels, b) employing a new micro gas chromatography (GC) equipped with thermal conductivity detector (TCD) to assure a small internal volume to improve the N2 signal quality, c) completing the gas replacement within 25 h via alternating negative pressure and pressurized purging, and d) adopting aerobic, low temperature conditions to maintain the carbon and nitrogen substrates during gas replacement (Wang et al. 2011). These developments have resulted in a better sensitivity in the direct measurement of N2 emissions, with a detection limit of 0.23 μg N2−N kg−1 ds h−1 or 8.1 μg N2−N m−2 h−1 and a currently unachieved reproducibility of measurements. Moreover, these developments also allow for simultaneous measurements of NO, N2O and CO2 emissions with high sensitivities (Wang et al. 2011).

Due to the ecological significance of N2, N2O and NO emissions from soils, quantification of their magnitudes at various timescales (e.g., hourly, daily, annual) and spatial scales (e.g., site, regional) are necessary. Until now, N2 emissions could not be directly measured in the field at any scale. Meanwhile, emissions of N2O and NO can be directly measured at the plot scale and on different timescales, but cannot be easily observed at the regional scale. Therefore, modeling approaches have to be established to quantify the emissions of these gases at different temporal and spatial scales. For simulating the ecosystem nitrogen cycling, it is necessary to parameterize key nitrogen cycling processes such as denitrification. However, due to the already mentioned methodological problems, we are lacking parameter kinetics and a detailed evaluation of the production of N2, N2O and NO for denitrification.

Generally, denitrification in the calcic cambiols as found on the North China plain is regarded as weak and unimportant for the emissions of N2O and NO and losses of nitrate (e.g., Cai et al. 2002; Ju et al. 2009; Wan et al. 2009), although N2 emissions from calcareous soils may be high as shown e.g., by Dannenmann et al. (2008). Nevertheless, data from Liu et al. (2012) shows that during the maize season (June–October) the calcic cambisols often contains very high soil moisture contents, with values around or above 65–70 % water-filled pore space (WFPS) for up to 2–3 weeks. During these periods soils are often weak sources for atmospheric methane (CH4), indicating that predominantly anaerobic conditions prevail. Therefore, denitrification is likely to be important for N2O and NO emissions from these soils under conditions of sufficient supplies of carbon and nitrogen substrates and anaerobiosis (Mei et al. 2009, 2011). Moreover, direct measurement of the denitrification potential and its N2, N2O and NO fractions for calcic cambisols is very important in terms of model parameterization and following model estimation of denitrification on a regional scale. However, few directly measured data on the denitrification potential and the production of N2, N2O and NO during denitrification can be found for this soil type.

In this study, we performed laboratory experiments for a silt clay calcic cambisol using the gas-flow-soil-core technique (Wang et al. 2011). The objectives were to a) directly measure the denitrification potential and the fractions of nitrogenous gases as products of the process, b) investigate the influences of the NO −3 concentration on denitrification gas emissions, and c) test the hypothesis that denitrification is important for N2O and NO emissions from calcic cambisols under conditions of sufficient supplies of carbon and nitrogen substrates and anaerobiosis as to be found to occur commonly in agricultural lands.

Materials and methods

Soil

The soil used for this experiment was collected (at a depth of 0–15 cm) from a field (34 °55.51′N, 110 °42.59′E, 348 m altitude) cultivated with winter wheat and summer maize in rotation (Shanxi, China). The field site was involved in a study aiming at the quantification of greenhouse gas emissions (Liu et al. 2011, 2012). It exposes to a monsoon climate of the southern temperate zone, with annual mean air temperature and precipitation of 14.8°C and 562 mm, respectively, during 2000–2008 (National Climatic Data Center, http://www.ncdc.noaa.gov/oa/ncdc.html). The field site is dominated by a cinnamon soil (classified by the National Soil Survey Office 1998) or calcic cambisol (classified by WRB 2006), which contains 31.8 ± 0.9 % clay, 38.9 ± 1.8 % silt, 29.3 ± 2.5 % sand, 1.13 ± 0.06 % organic carbon, 0.11 ± 0.005 % total nitrogen and has a pH (H2O) of 8.7 ± 0.05, and a bulk density of 1.17 ± 0.04 g cm−3 (Liu et al. 2011). Bulk soil samples were air dried to a gravimetric moisture content of 18 % (ca. 32 % WFPS), sieved with a 2 mm sieve, mixed and stored at 4°C.

Experimental design

For determining the effect of variation in soil NO −3 on denitrification rates and denitrification products, NO −3 was added to soil samples to obtain initial NO −3 concentrations of approximately 10, 30, 50, 80, 100, 180 and 250 mg N kg−1 dry soil (hereafter referred to as 10 N, 30 N, 50 N, 80 N, 100 N, 180 N and 250 N, respectively). These initial NO −3 concentrations in soils are equivalent to 4.2–107 kg N ha−1. The 10 N treatment, with a background NO −3 concentration of 12.9 mg N kg−1 ds in the soil samples, was free from NO −3 addition and used as the control (Table 1). In all treatments the initial dissolved organic carbon (DOC) concentrations were set at 300 mg C kg−1 dry soil (ds). Hereafter the sum of background NO −3 (or DOC) concentration in soil samples and the NO −3 (or DOC) added prior to each treatment is referred to as the initial NO −3 (or DOC) concentration.

Table 1 Cumulative emissions of Nt (N2+N2O+NO) and carbon dioxide (CO2), and changes in ammonia (NH +4 ), nitrate (NO −3 ), dissolved organic carbon (DOC), soil microbial biomass nitrogen (SMBN) and carbon (SMBC), and the recovery rate of carbon and nitrogen during the entire incubation period under different initial nitrate (NO −3 ) concentrations
Full size table

For all treatments the same soil was used. But due to the length of experiments, we re-started the experiment if a new NO −3 level was investigated (i.e., each treatment was carried out with the same soil, which was not used previously for other treatments, but was taken from the same pool of air dried soil samples stored in our laboratory). For each treatment, two weeks before the start of measurements, the air dried soils were wetted with deionized water to approximately 45 % WFPS and pre-incubated at 4°C. This means, as already said before, that we used a new freshly pre-incubated soil for every NO −3 amendment treatment. The soil moisture content was held stable during pre-incubation, by adding certain amounts of water following daily weighing. For each treatment, the pre-incubated soil was repacked into 12 cores (bulk density: 1.07 g cm−3) using standard stainless steel rings (diameter 5.6 cm, height 4 cm). Four cores were placed into one of the three incubation vessels. Sub-samples taken from the pre-incubated soil were extracted for NO −3 and DOC assay in order to determine the amount of NO −3 and glucose additions. The measured concentrations of NO −3 and DOC varied within a range of 5–19 mg N kg−1 ds and 30–50 mg C kg−1 ds, respectively (shown as C0 in Table 1). Accordingly, 5 mL of solution with appropriate KNO3 and glucose concentrations was sprinkled onto the surface of each soil core to obtain the desired initial NO −3 concentrations (approximately 10–250 mg N kg−1 ds) and a fixed initial DOC concentration (approximately 300 mg C kg−1 ds) (Table 1). Addition of the solution increased the soil moisture content from 45 % to 55 % WFPS.

Immediately following the solution additions, the vessels were sealed gas tight (see Wang et al. 2011) and submerged into a water bath for temperature control as well as reducing possible gas diffusion between atmosphere and vessels containing the soil cores. Then the soil atmosphere was purged at 2°C, using an N2-free atmosphere of 20 % oxygen (O2) in helium (He). During the aerobic condition at low temperature, the consumptions of the added carbon and nitrogen substrates is negligible as was shown by Wang et al. (2011). Hereafter we refer to the period replacing the soil atmosphere with an N2-free gas at low temperature the “gas exchange phase”. The gas exchange phase was set to last for 30 h, during which the replacement of soil atmosphere was carried out by alternating negative pressure (by evacuation for 2 min using an vacuum pump to approximately 30 kPa) followed by pressurized purging with N2-free atmosphere for 2 min at a flow rate of 200 mL min−1 (Wang et al. 2011). This phase was followed by gas emission measurements carried out for approximately 15 h under the aerobic condition at 2°C. Then an anaerobic condition was established by flushing the vessel headspace with pure He. After 48 h of anaerobic incubation at 2°C, the temperature was increased. It took 40 min to achieve the target temperature of 25°C (Fig. 1). Incubation conditions of 25°C and 55 % WFPS were chosen since these conditions are typical in soils of the field during the maize growing season (Liu et al. 2012). In our experiment anaerobic incubations were provided by replacing the soil atmosphere with pure He. The O2 concentrations were monitored with an electron capture detector (ECD) and TCD (see below). Both detectors demonstrated that no O2 was available under the chosen incubation conditions. In the gas exchange phase, N2 could not be measured, whereas the emissions of other gases (N2O, NO and CO2) were measured once every 5–10 h. Following complete gas exchange, the N2, N2O, NO and CO2 emissions were simultaneously measured once every 8 h. Gas measurements for a treatment were stopped when the emissions of nitrogenous gases declined to around the detection limits of the employed gas-flow-soil-core system.

Fig. 1
figure 1

Dynamics and magnitudes of nitrogen gas (N2), nitrous oxide (N2O), nitric oxide (NO) and carbon dioxide (CO2) emissions for the different soil nitrate (NO −3 ) addition treatments (10–250 mg N kg−1 ds). Definitions of the treatment codes are referred to in the text and Table 1. Time zero represents the beginning of the incubation, i.e., following the closure of the vessels and the equilibration of the water bath temperature to 2°C. T = 0 directly follows the addition of NO −3 and setting of soil dissolved organic carbon (DOC) concentrations to 300 mg C kg−1 ds by the addition of the respective amounts of glucose. The means of three replicate experiments ± standard error are shown

Full size image

Gas sampling and analysis

The semi-dynamic chamber method described by Wang et al. (2011) was applied to determine the emissions of the individual gases (N2O, NO, CO2 and N2). To take an air sample for detecting N2O, NO and CO2 concentrations, a gas containing 20 μL N2 L−1 (ppmv) in He was allowed to continuously flush the vessel headspace at 20 mL min−1 for 4 min. The out-flowing air during the last 3 min was collected with a 60-mL syringe. Then we halted the headspace flushing for 11 min before initiating the collection of the next sample. This procedure was repeated 5 times, i.e., the vessel headspace air for simultaneous detection of N2O, NO and CO2 was sampled at an interval of 15 min over a 1-hour period. Out of each 60 mL gas sample, 20 mL was used to analyze the N2O and CO2 concentrations using a GC (Agilent GC 6820, Shanghai, China) equipped with an ECD and a flame ionization detector (FID) fitted with a convertor to reduce CO2 into methane. Pure N2 (99.999 %) was used as the carrier gas for the analysis of both components. For the N2O analysis we applied the DN-CO2 method, in which a buffering gas of 10 % CO2 in pure N2 flowing through the ECD cell at approximately 2 mL min−1 was used for N2O detection. Details regarding the GC configurations for analysis of the two gases can be found in Zheng et al. (2008) and Wang et al. (2010). The remaining 40 mL of the gas sample was diluted to a volume of 2 L by injecting it into a gas-tight Tedlar membrane bag (Delin Gas Packing Co. Ltd., Dalian, China) pre-filled with pure N2 (99.999 %). Then the NO concentrations were measured with a chemiluminescent analyzer (42i NO–NO2–NOX, Thermo Environmental Instruments Inc., USA). To detect the N2 emission, we used a sampling interval of 45 min over a 3 h period. The procedures of the detector calibration with a standard gas and flushing all tubes prohibited a shorter interval of measuring N2 concentration in the vessel headspace. The N2 concentrations were automatically detected by online analysis following injection of the out-flowing air (immediately after flushing the headspace of the vessel with 20 ppmv N2 in He for 3.8 min at a flow rate of 20 mL min−1) into a micro GC equipped with a TCD (Agilent micro GC 3000, USA). Then we halted the headspace flushing for 41 min before initiating the collection of a new sample. This procedure was repeated 5 times. The measured N2 concentrations were corrected by subtracting the inherent N2 leakage rate of the system, which was 0.4 ppmv h−1 on average. Details regarding the GC configurations used for the N2 analysis can be found in Wang et al. (2011).

All instruments were regularly calibrated with standard gases (AP BEIFEN Gases Industry Co. Ltd., Beijing, China). The standard gas concentrations of N2 and CO2 were 20 ppmv (in He) and 354 ppmv (in N2), respectively. Over the wide ranges of N2 (20–2000 ppmv) and CO2 (300–3000 ppmv) concentrations in the air samples, the detector always shows significant linear response, with determination coefficient of the linear regressions (R2) greater than 0.999. A series of N2O standard gases (with 0.352, 5, 20, and 200 ppmv N2O in N2) was employed to obtain calibration curves for routine use.

Calculation of emissions

Since the headspace air was diluted by the flushing process during sampling, as described above, the measured concentrations had to be corrected before they were used in the calculation of emissions. The correction was based on Eqs. 1–2 (see Wang et al. 2011 for details).

$$ C_i^{*} = C_i^m - \sum\nolimits_1^i {{e^{{bi}}} \cdot \left[ {{{\text{e}}^{{\left( {\frac{{ - {v_{{in}}}}}{{{V_{{head}}}}} \cdot t} \right)}}} - 1} \right]} \quad {\text{when}}\,C_i^m > {C_{{in}}} $$
(1)
$$ bi = \ln \left( {\left| {C_i^m - {C_{{in}}}} \right|} \right) + \frac{{{v_{{in}}}}}{{{V_{{head}}}}} \cdot t $$
(2)

where \( C_i^{*} \) is the corrected concentration of a gas (ppmv); t is the sampling time during headspace flushing (min); i is the series number of measurements for determining an emission (i = 1,2, …, 5); C m is the measured concentration of the gas (ppmv); C in is the inlet gas concentration during sampling (ppmv); V head is the headspace volume (mL); v in is the inlet gas flow rate during sampling (mL min−1); and bi is a parameter specific for each concentration measurement.

The emission of a gas was determined using Eq. 3 (Wang et al. 2011).

$$ F = \frac{{{V_{{head}}} \cdot \left( {\Delta C_i^{*} - \Delta {C_L}} \right) \cdot M}}{{{M_{{ds}}} \cdot MV}} \cdot \frac{{273}}{{273 + T}} $$
(3)

where F is the emission of a gas (μg N or C h−1 kg−1 ds); \( \Delta C_i^{*} \) is the change rate in the corrected concentration of a gas (ppmv h−1); ΔC L is the inherent leakage rate of the system for the gas (N2: 0.4 ppmv N2 h−1; all other gases: negligible); M is the weight of pure nitrogen or carbon per mole in N2, N2O, NO or CO2 (28, 28, 14 and 12 g mol−1, respectively); M ds is the dry weight of the incubated soil (g) in an vessel; MV is the molar volume of the gas at 273 K and 1,013 hPa (L mol−1); and T is the incubation temperature (°C). The detection limits of our gas-flow-soil-core system were 0.23, 0.02 and 0.08 μg N h−1 kg−1 ds (or 8.1, 0.6 and 2.7 μg N m−2 h−1 for the soil cores height of 4 cm) for N2, N2O and NO, respectively, and 1.9 μg C h−1 kg−1 ds (or 67.2 μg C m−2 h−1) for CO2.

Soil analysis

At the beginning and end of incubation for a treatment, we measured the concentrations of NO −3 , ammonium (NH +4 ), DOC, and microbial biomass carbon and nitrogen (hereafter referred to as SMBC and SMBN, respectively) in the incubated soils. For analysis of the inorganic nitrogen content, 20 g of soil was extracted for 1 h, using 100 mL of deionized water for NO −3 or 100 mL of a 2 M potassium chloride (KCL) solution for NH +4 (Keeney and Nelson 1982). The NH +4 concentrations were determined colorimetrically using an ultraviolet spectrophotometer (UNICO, UV-2802, Shanghai, China), while the NO −3 contents were determined by ion chromatography (Metrohm 790 IC, Switzerland). The DOC was extracted by shaking 20 g of the soil sample for 1 h with 100 mL of deionized water. The extracts were centrifuged at 6000 rpm for 15 min and decanted, and the supernatant was then filtered through a 0.45 μm polyethersulfone membrane filter (Membran, Germany) before analysis with a C/N analyzer (multi NC 3000, Analytik, Jena, Germany). The SMBC or SMBN content was determined from the difference in the concentrations of DOC or dissolved nitrogen extracted from the fumigated and non-fumigated soils (7 g was extracted using 35 mL of 0.5 M potassium sulfate) (Sparling and West 1988).

Statistics

The software package SPSS Statistics Client 19.0 (SPSS, Beijing, China) was used for data analysis (ANOVA and GLM regressions). Graphical outputs were obtained from Origin 8.0 (Origin Lab Ltd., Guangzhou, China).

Results

Emission dynamics of nitrogenous gases and CO2

The emission dynamics of nitrogenous gases and CO2, as affected by the different levels of initial NO −3 concentrations, are displayed in Fig. 1, while the times of the emission peaks of the nitrogenous gases are listed in Table 2.

Table 2 Magnitude and timing of peak nitrogen gas (N2), nitrous oxide (N2O), nitric oxide (NO) emissions for the different nitrate addition treatments (10–250 mg N kg−1 ds)
Full size table

During aerobic incubation at 2°C, the N2O and NO emissions were generally close to their detection limits, whereas the N2 emissions were slightly higher than its detection limit. The CO2 emissions were initially at 400–500 μg C h−1 kg−1 ds and decreased to approximately 120 μg C h−1 kg−1 ds.

The N2 emissions during the anaerobic incubation at 2°C were in the range of 20–100 μg N h−1 kg−1 ds, while N2O emissions were at the same magnitude as NO emissions, but by a factor of 3–10 lower than N2 emissions. The CO2 emissions were 100–260 (mean: 160) μg C h−1 kg−1 ds. The emissions of all nitrogenous gases during the conditions of anaerobiosis and marginally low temperature were significantly increased by 1–2 orders of magnitude compared to the aerobic conditions (p < 0.01), while the soil respiration, as indicated by CO2, showed no significant difference.

After the target incubation temperature of 25°C was achieved (>95 h), the emissions of nitrogenous gases immediately increased, but the time to reach the maximum varied among the treatments. Under the 10 N condition, the emissions of all of the nitrogenous gases peaked at the first measurement that was performed within 3 h after the target incubation temperature was archived, and then quickly declined over 28 h (i.e., 126 h after the incubation began) to 1–2 μg N h−1 kg−1 ds of N2 and to near the detection limits of N2O and NO. For 30 N, 50 N, 80 N and 100 N, the N2O and NO emissions in the first few hours were higher compared to the N2 emissions. Maximum N2O and NO emissions were observed by the second measurement (i.e., approximately 103 h after the start of the incubation or 8 h following the temperature increase), which were followed by the N2 peak. The lag time between the appearance of the N2O and N2 peaks was approximately 2 h for 30 N, 12 h for 50 N, and 27 h for 80 N and 100 N. The dynamic emission patterns of all nitrogenous gases observed under 180 N and 250 N conditions showed two peaks. This differed markedly from the patterns of the other treatments that showed single peak of each nitrogenous gas. The N2O and NO peaks were simultaneously observed, with their first peaks appeared at 103 h, for all treatments excluding 10 N. The second N2O peak was higher than the first one by 1–4 folds (p < 0.001) and appeared at 150 h (180 N) or 183 h (250 N) after the incubation started. Meanwhile, the second NO emission was much smaller than the first one. Each N2 peak followed those of N2O and NO. The first N2 peak was much higher than the second one by 2–6 folds (p < 0.001). Subsequently, N2 was the sole end product of denitrification.

During the anaerobic incubation at 25°C, as Fig. 1 shows, the emissions of nitrogenous gases peak in the following sequence: NO ≤ N2O<N2. The time (y, in h) required for the first appearance of an N2 emission peak to occur after raising the temperature increased linearly with the logarithm of the initial NO −3 concentration (x, in mg N kg−1 ds), which could be fitted with y = 13.9 ln (x) –33.0 (R2 = 0.92, p < 0.001). The lag time (y, in h) of the first/single N2 peak, relative to appearance of the first/single N2O peak, was approximately 2–35 h, also showing a trend of a linear increase with the logarithm of initial NO −3 concentrations, with y = 12.0 ln (x) −32.2 (R2 = 0.87, p < 0.05). As Figs. 2a–c illustrate, the maximum emissions of the nitrogenous gases were significantly affected by the initial soil NO −3 concentrations, with the relationship for either N2 or N2O being very well described by Michaelis-Menten functions (p < 0.01) and for NO by a log normal curve (p < 0.05).

Fig. 2
figure 2

The dependences of maximum nitrogen gas (N2), nitrous oxide (N2O), nitric oxide (NO) emissions (a–c) and their cumulative emissions over the entire incubation period (d–f) on initial soil nitrate (NO −3 ) concentrations (10–250 mg N kg−1 ds). The data displayed here are the means of three replicate experiments ± standard error

Full size image

With regard to the CO2 emissions after the temperature increase, the first and in some cases the single peak appeared synchronically with the first/single N2O peak, while the second peak appeared at the same time as or slightly earlier than the maximum N2 emission. Towards the end of the incubations, the CO2 emissions declined and remained constant at approximately 100–300 mg C h−1 kg−1 ds.

Cumulative emissions of nitrogenous gases and CO2

The cumulative emissions of the N2, N2O, NO and N2+N2O+NO (hereafter referred to as Nt) from all the treatments ranged from 8.0–123.4, 1.4–74.4, 0.9–10.1 and 10.3–208.9 mg N kg−1 ds, respectively (Table 1 and Fig. 2). As Fig. 2d illustrates, the cumulative Nt and N2 emissions were linearly and positively correlated with the initial soil NO −3 concentrations (Nt: R2 = 0.995, p < 0.001; N2: R2 = 0.96, p < 0.001). Such a relationship exhibited the first order kinetics, which indicated the phase of Michaelis-Menten kinetics occurring at low substrate concentrations. The cumulative emissions of NO obtained in all treatments very well fitted the Michaelis-Menten kinetics (Fig. 2f), whereas the cumulative N2O emissions depended exponentially upon the initial NO −3 concentrations (Fig. 2e). With respect to the total cumulative nitrogenous gas emissions due to denitrification, the fractions of the N2, N2O and NO emissions to the denitrification potential were in the range of 51–78 %, 14–36 % and 5–22 %, respectively (Fig. 3). These fractions well supported the hypothesis that denitrification is significant for N2O and NO emissions from calcic cambisols under conditions of sufficient supplies of carbon and nitrogen substrates and anaerobiosis. The cumulative CO2 emissions during the entire incubation period varied from 68.1 to 161.2 mg C kg−1 ds among the treatments (Table 1), with a positive relationship with the initial NO −3 concentrations (R2 = 0.88, p < 0.01).

Fig. 3
figure 3

Effects of different starting soil nitrate (NO −3 ) concentration on the denitrification fraction during the entire incubation period. The means of three replicate experiments ± standard error are given

Full size image

Ratios of cumulative gas emissions

Regarding the cumulative emissions during anaerobic incubation at 25°C, the N2:N2O molar ratios measured in all treatments were 1.7–5.3 (Table 3), showing no obvious simple relationship against initial NO −3 concentrations. The NO:N2O molar ratios for 180 N and 250 N were less than 0.5 while those for the other treatments were 1.0–1.5 (Table 3), indicating the influence of initial nitrogen substrate concentration to some extent. The CO2:Nt molar ratios for 80 N, 100 N, 180 N and 250 N were in a range of 0.9–1.1, and those for the other treatments with lower NO −3 application rate were 1.9–7.9 (Table 3). The CO2:Nt molar ratios (y) of all treatments were negatively dependent upon the initial NO −3 concentrations (x, in mg N kg−1 ds). This dependence could be described with a power function as y = 32.8 × −0.71 (R2 = 0.88, p < 0.001).

Table 3 Molar ratios of cumulative nitrogenous gas and carbon dioxide (CO2) emissions during anaerobic incubation at 25°C
Full size table

Nitrogen and carbon balance

In addition to the cumulative Nt and CO2 emissions, Table 1 provides an overview of the NH +4 , NO −3 , SMBN, SMBC and DOC contents at the beginning and end of incubation for the various soil NO −3 treatments. As can be deduced from Table 1, the soil microbial biomass carbon to nitrogen ratios fell within the range of 4.5–7.5 (mean: 5.5). The biomass carbon and nitrogen of soil biomass, as well as their ratios showed no significant change between the beginning and end of incubation in any of treatments. However, significant changes in the soil NO −3 contents were observed (p < 0.01), with the NO −3 pool being almost completely depleted. A small but significant (p < 0.01) increase in the soil NH +4 concentrations was observed. The observed cumulative Nt emissions equaled 80–88 % (average: 84 %) of the reduction in soil NO −3 (expressed as N-RRn values in Table 1) and 80–108 % (average: 88 %) of the total changes in (NH +4 + NO −3 +SMBN) - N (expressed as N-RRf values in Table 1). Additionally, the observed CO2–C emissions corresponded to 32–64 % (average: 50 %) of the decrease in soil DOC or 35–88 % (average: 59 %) of the changes in DOC plus SMBC.

Discussion

Effect of initial NO −3 concentrations on peak emissions of denitrification gases

Following induction of anaerobiosis and an increase in incubation temperature from 2°C to 25°C, the first/single peak of N2 emissions appeared later than the first/single peak of N2O or NO (Fig. 1 and Table 2). The lag period between the appearance of N2O and N2 peaks (2–27 h) was comparable to the 6–12 h observed by Mathieu et al. (2006) or the 19 h lag period reported in the study of Meijide et al. (2010). Dendooven and Anderson (1994) suggested that the lag time between the appearance of the N2O and N2 peak resulted from the different lag times for the synthesis of the enzymes of the denitrification chain involved in the production and consumption of N2O under anaerobic conditions. Another possible explanation for the lag time is the preferential acceptance of electrons from NO −3 compared with N2O (Firestone and Tiedje 1979). The latter explanation is well supported by our result that the lag period length was increased linearly with the logarithm of the initial NO −3 concentration for denitrification (p < 0.05).

Only a few previous studies, in which the gas-flow-soil-core technique was applied, have simultaneously measured the emissions of all denitrification gases from soils (De Wever et al. 2002; Weier et al. 1993; Scholefield et al. 1997b), while others usually exclude NO, although this gas constitutes an obligatory intermediate product in the reaction sequence (Conrad 1996; Firestone and Davidson 1989; Russow et al. 2009). In this study, Michaelis-Menten kinetics fitted the dependence of the maximum N2 or N2O emissions from soil denitrification on initial NO −3 levels very well (Fig. 2a–b), but the maximum NO emissions showed an inconsistent pattern (Fig. 2c). The maximum NO emissions was stimulated by the initial NO −3 concentration at low levels (10–50 mg N kg−1 ds) but inhibited under conditions with higher initial NO −3 levels (>50 mg N kg−1 ds) (Fig. 2c). Similar results were reported by Ludwig et al. (2001), who found that the rate of NO emission not only depends on the overall denitrification rate but is also strongly affected by parameters that influence the proportion of NO relative to the terminal products of denitrification, i.e., N2O and N2. However, such a relationship between the peak NO emission and soil NO −3 concentration could not be demonstrated by Shannon et al. (2011), who worked with Pseudomonas mandelii inoculated into anoxic soil. These authors did not observe any significant change in NO reductase activity during incubations with different starting concentrations of soil NO −3 (>10 mg N kg−1 ds). The further study is still needed to explain the reason for the inhibitory effect of high initial NO −3 concentration on NO emission from denitrification.

The maximum N2 emissions observed in our treatments varied from 600 to 2300 μg N h−1 kg−1 ds. In comparison, previous studies using the gas-flow-soil-core technique for denitrification measurements have reported much higher values in some cases but lower values in others. For instance, Swerts et al. (1996a) observed 3.6-fold higher maximum N2 emissions, i.e., 200 versus 55 mg N d−1 kg−1 ds (2300 μg N h−1 kg−1 ds) in our study, from a clay silt loam for the initial NO −3 level of 100 mg N kg−1 ds. The higher maximum N2 emissions were most likely due to the higher initial carbon substrate concentration (950 versus 300 mg C kg−1 ds in our study). Cárdenas et al. (2003) demonstrated maximum N2 emissions that were one magnitude lower (2.6 versus 31 kg N ha−1 d−1 in our study) from a grassland soil with an initial NO −3 content of 50 kg N ha−1 and DOC content of 360 kg C ha−1, which were comparable to the nitrogen and carbon substrate concentrations in our 50 N treatment. The different magnitudes in peak N2 emissions might have been caused by differences in the soil properties such as soil pH (5.7 versus 8.5 in our study).

Effect of initial NO −3 concentrations on cumulative emissions of denitrification gases

In this study, the cumulative N2 emissions showed a linear and positive dependence upon the initial NO −3 concentrations (Fig. 2d). Scholefield et al. (1997b) have also observed a similar linear relationship for NO −3 applications lower than 150 kg N ha−1 (approximately 150 mg N kg−1 ds). When the NO −3 addition rate was further increased to 200 kg N ha−1 (approximately 200 mg N kg−1 ds), however, Scholefield and his colleagues observed a reduced N2 production, which was inconsistent with our results (Fig. 2d).

Scholefield et al. (1997b) reported that the linear increase in N2O emissions from denitrification correlated with the NO −3 application rates of 0–200 kg N ha−1 (approximately 0–200 mg N kg−1 ds). We, however, observed an exponential increase between cumulative N2O emission and initial NO −3 concentration (Fig. 2e). The increase in N2O emissions at very high NO −3 concentration was likely caused by two mechanisms. One of these mechanisms was high NO −3 inhibition of N2O reductase (Blackmer and Bremner 1978). This mechanism would lead to a reduction in N2 emissions, as reported in the literature (e.g., Ruser et al. 2006; Scholefield et al. 1997b), or a slowed-down increase in N2 emissions. The latter effect could be seen in our study; at the highest NO −3 application rate the N2 emissions were only slightly higher compared to the 180 N treatment (Fig. 2d). The other mechanism was carbon substrate limitation of denitrification. In general, when the supply of the organic carbon substrate, as a reductant, prevails over the availability of oxidants, such as nitrate, complete denitrification occurs, and N2 is the main product. Otherwise, carbon substrate limitation occurs, resulting in incomplete denitrification, with N2O being generated as the main product (e.g., Hutchinson and Davidson 1993; Senbayram et al. 2011; Zumft 1997). This mechanism would easily explain the observed patterns of nitrogenous gas production at high soil NO −3 concentrations, i.e., following the increase of incubation temperature to 25°C, a small N2O peak followed by a high N2 peak was observed. The N2 peak was again followed by a second much higher N2O peak before finally the N2 production became the dominating end product of denitrification (Fig. 1f–g). With regard to the carbon substrate supply, this result can be interpreted as follows: at the beginning, the carbon substrate was not limiting, but the diffusion of NO −3 to the sites of active denitrification started to become limiting, resulting in a peak of N2 emissions. The second peak of N2O emissions can therefore be interpreted as resulting from the carbon substrate becoming limited, which resulted in incomplete denitrification, since NO −3 supply by diffusion was not limiting. Finally, the NO −3 diminished, and at the reduced rates of denitrification, the carbon substrate supply was not limiting anymore. Consequently, N2 was the main end product of denitrification and not N2O (Fig. 1f–g). However the above mentioned two mechanisms remain speculative since due to methodological restrictions we did not measure NO −3 and DOC concentrations or enzyme activities during the incubation.

Molar ratios among denitrification gases and CO2

We did not observed significant relationships between the molar ratios (1.7–5.3) of N2:N2O emissions from denitrification with initial soil NO −3 concentrations. This finding differs from the results of Scholefield et al. (1997b), who observed an exponential decrease (p < 0.05) in N2:N2O molar ratios (from 3.0 to 0.5, adapted from data in the literature) with increases in soil NO −3 concentrations (from 25 to 200 kg N ha−1, corresponding to approximately 25 to 200 mg N kg−1 ds). The ratios of N2:N2O for periodic emissions associated with denitrification were also measured in other previous studies that reported values of 0.2 to 2.8 for grassland soils (Cárdenas et al. 2003; Swerts et al. 1996a) and 1.7 to 3.5 for soils of cultivated lands (Schlesinger 2009; Senbayram et al. 2011). In comparison with the N2:N2O ratios associated with periodic emissions, the range of these ratios related to the instantaneous emissions were much wider. For instance, the N2:N2O ratios during the late anaerobic incubation period in our experiment were usually greater than 100, and ratios of 5–200 have also been reported for arable soils in Uzbekistan (Scheer et al. 2009) and forest soils in southern Germany (Dannenmann et al. 2008). These examples indicate that N2:N2O ratios can vary significantly with soil nitrate and carbon substrate availability, redox potential, soil properties, and denitrifier activity (e.g., Ciarlo et al. 2008; Weier et al. 1993; Philippot et al. 2011; Morley and Baggs 2010).

The mean NO:N2O molar ratios of the cumulative emissions during anaerobic incubation conditions at 25°C were in the range of 1.0–1.5 for all of the treatments, excluding 180 N and 250 N, which showed lower ratios of <0.5. This shows that NO is a significant by-product of denitrification. Similar results were reported by Anderson and Levine (1986), who observed a NO:N2O ratio of 3.0 in pure culture of the denitrifier Alcaligenes faecalis in a liquid medium. When the authors pure-cultured the denitrifiers Rhizobium japonicum and Pseudomonas fluorescens in the same medium, they observed NO:N2O molar ratios less than 1.0. Due to the fact that in our experiment, even under strict anaerobic conditions, NO:N2O molar ratios greater than 1.0 were commonly observed under conditions with low to moderate initial NO −3 concentrations; therefore, earlier statements by different authors using the NO:N2O ratio as an indicator to judge whether nitrogen trace gas fluxes are nitrification (NO:N2O ratio >1) or denitrification (NO:N2O ratio <1) dominated must be reconsidered (Del Prado et al. 2006; Scheer et al. 2009).

Theoretically, the molar ratios of the CO2–C emissions to NO–N, N2O–N or N2–N during denitrification are 0.75 (NO), 1.00 (N2O) or 1.25 (N2) (Swerts et al. 1996a). The ratios we observed in the 80 N to 250 N treatments (0.9–1.1) are in good agreement with these theoretical values, whereas higher ratios (>1.9) were detected in other treatments with lower initial NO −3 concentration. Swerts et al. (1996a) and De Wever et al. (2002) also observed comparably high ratios in anaerobic soil incubation. The higher ratios were likely caused by CO2 production in anaerobic process other than denitrification, such as fermentation or microbial iron or sulfate reduction (Swerts et al. 1996a; Achtnich et al. 1995; Yao et al. 1999). The fact that those other microbial processes were significantly contributing to the CO2 production can also be observed from the second peak of CO2 emission in the 10 N treatment, since during this period the nitrogenous gas emissions were already close to the detection limit.

Nitrogen and carbon balances

The changes in soil NO −3 pools between the beginning and end of the incubation for all treatments employed in this study were largely explained by the emissions of nitrogenous gases, with recovery rates (N–RRn: 84 % on average) falling in the range of those reported in other studies, which varied from 42–57 % (Scholefield et al. 1997b) to 91–117 % (Swerts et al. 1996a). The emissions of nitrogenous gases explained the changes in the total soil nitrogen pool (i.e., the sum of NO −3 , NH +4 and SMBN) slightly better (p < 0.01), with an average recovery rate (N–RRf) of 88 % being observed. In view of the uncertainties involved in measurements of soil NO −3 , NH +4 , SMBN and nitrogenous gas emissions from denitrification, these results were encouraging. The soil NH +4 contents significantly increased for all NO −3 treatments. Increases in soil NH +4 concentrations during anaerobic soil incubation have also been reported by Scholefield et al. (1997b) and Meijide et al. (2010). This indicates that under strict anaerobic conditions dissimilatory nitrate reduction to ammonia (DNRA) (Rütting et al. 2011) or anaerobic mineralization of organic matter (Bridgham et al. 1998) can play a significant role. For the former process N2O was also produced (Rütting et al. 2011). Some studies have reported that N2O production from DNRA accounted for 1 % (Cole 1988) to 5–10 % (Smith and Zimmerman 1981) of the NO −3 amount, while the product NH +4 accounted typically for more than 90 % (Bleakley and Tiedje 1982). However, we are not able to finally judge if DNRA is indeed an important process for nitrogenous gas formation unless 15N isotopes are used. This topic deserves further investigation.

The measured CO2 emissions only explained 35–77 % of the losses of soil carbon pool (i.e., DOC plus SMBC). The carbon pool losses that were not from CO2 emissions were attributed to three possible reasons. The first reason was likely fermentation-induced losses. This process could have occurred simultaneously or sequentially with denitrification under anaerobic conditions, competing for DOC and producing volatile fatty acids (Swerts et al. 1996a,b) that were not detected in this study. The second cause was likely due to CO2 dissociation in soil water under conditions with high pH (Ingwersen et al. 2008). The final reason was likely underestimation due to no employing any correction for SMBC using an extraction coefficient (Joergensen 1996).

Conclusions

The dynamics and magnitudes of nitrogen gas (N2), nitrous oxide (N2O), nitric oxide (NO) and carbon dioxide (CO2) emissions from denitrification in a silt clay calcic cambisol during anaerobic incubation were directly measured, using the gas-flow-soil-core technique that has been proven to be a useful and reliable tool to investigate emissions of N2 and the other gases. In the treatments with different initial nitrate (NO −3 ) concentrations, almost all NO −3 were consumed during anaerobic incubation, with 80–88 % of the changes in the soil NO −3 pools being recovered by measuring the emissions of nitrogenous gases. The increases in initial NO −3 concentration significantly enhanced the denitrification potential and the emissions of N2O and N2 as products of the process. The individual products of N2, N2O and NO accounted for the denitrification potential with very narrow fraction ranges in spite of the much wider range of initial NO −3 concentrations. This study strongly supported the hypothesis that denitrification is significant for N2O and NO emissions from calcic cambisols under conditions of sufficient supplies of carbon and nitrogen substrates and anaerobiosis. Our study provides some directly measured parameters for model-simulating the denitrification process of the investigated soil. In addition, the product ratios of NO:N2O in denitrification were more than 1.0 when the initial NO −3 concentrations were at low to moderate levels and vice versa. However, further study is needed to test whether these results are specific to calcic cambisols. To better parameterize the denitrification process of a soil, dynamic monitoring of carbon and nitrogen substrates during incubation is also strongly demanded in further studies using the gas-flow-soil-core technique.