Introduction

Nitrogen (N) is the plant nutrient element that most often limits primary production in terrestrial ecosystems and has been introduced into the biosphere mainly as reactive N through the chemical and biological fixation of dinitrogen (N2) (LeBauer and Treseder 2008). Denitrification is the most important process that removes reactive N from the biosphere and returns it to the atmosphere, which includes all or parts of the sequential reduction of nitrate (NO3−) to nitrite (NO2−), nitric oxide (NO), nitrous oxide (N2O) and N2. Whereas N2 is the ultimate end product of denitrification, other intermediate gaseous forms of N, such as N2O, can also be produced through denitrification. The increase in atmospheric N2O concentrations is of growing concern, since N2O has been considered not only to be a potent greenhouse gas (GHG), but also be the most important destroyer of stratospheric ozone in the twenty-first century (Ravishankara et al. 2009). Denitrification is an anaerobic process which occurs mostly in anoxic microsites in unsaturated soils and requires readily available carbon (C) as an electron donor (Weier et al. 1993). Much of this C originates from plant roots either through root exudation or root turnover.

The presence of roots also induces physical and chemical changes in the rhizosphere soil, which subsequently influences N transformation processes such as nitrification and denitrification, and ultimately N2O emissions (Guyonnet et al. 2017). For example, strong competition for soluble N occurs between plant roots and microorganisms (Jones et al. 2013). Further, the continuous uptake of nutrients by plants often leads to the development of strong depletion zones around roots (Kuzyakov and Xu 2013). Root exudate compounds could also alter the structure and activity of soil microbial communities (Shi et al. 2011). For instance, it has been found that many types of plants can secrete organic substances that inhibit nitrification in the rhizosphere (Subbarao et al. 2009, 2013; Sun et al. 2016; Coskun et al. 2017). On the other hand, authors have hypothesized that plant root exudates can stimulate denitrification rates by i) providing organic C for denitrifiers and/or ii) creating O2 depleted microenvironments arising from roots and microbial respiration (Bakken 1988; Hayashi et al. 2015). Furthermore, it has been also suggested that root exudation may also alter/modify fungal and bacterial denitrification in the rhizosphere (Philippot 2002; Broeckling et al. 2008).

Understanding the extent to which plants can influence N2O emissions and alter the N2O/(N2O + N2) product ratio is of particular importance when estimating N budgets, for improving N use efficiency in agriculture and developing effective GHG mitigation strategies. However, there have been relatively few studies on this topic, with those undertaken often coming to contradictory conclusions. For instance, some studies have reported a reduced N2O/N2 ratio with plants (Stefanson 1972, wheat with a sealed Helium system; Vinther 1984, spring barley with acetylene inhibition method), while others have shown no consistent effect (Smith and Tiedje 1979, corn with acetylene inhibition method; Klemedtsson et al. 1987, barley with acetylene inhibition method). Additionally, due to methodological problems associated with the direct measurement of N2 production, most of the previous studies had used the classical acetylene inhibition method, which is now considered unsuitable for quantifying denitrification rates due to a range of inevitable artefacts such as catalytic NO decomposition (Groffman et al. 2006; Nadeem et al. 2013).

Recent decades have seen the development of several continuous flow soil-core incubation systems using a Helium (He) atmosphere to allow a more complete evaluation of denitrification processes in soil (Cardenas et al. 2003; Wang et al. 2011; Senbayram et al. 2018). These systems facilitate a direct and high-accuracy measurement of soil N2O and N2 emissions as compared to the acetylene inhibition approach (Weier et al. 1993) and the 15N isotope labelling approach (Cai et al. 2001). Additionally, new developments in analysis of natural abundance isotopic signatures of N2O can be used to examine the microbial pathways of N2O production (Toyoda and Yoshida 1999). The N2O 15N site preference (SP), i.e. the intramolecular distribution of the N isotopes in the central (α) and peripheral (β) positions in the linear asymmetric N2O molecule, has been proved to be a useful non-invasive tool to differentiate sources of N2O production pathways (Decock and Six 2013).

In this study, we conducted an incubation experiment in a modified robotic, continuous-flow, plant incubation system (PRoFLOW) with a He+O2 + CO2 mixed atmosphere equipped with LED plant light to enable the direct measurement of soil N2O and N2 emissions at high temporal resolution. Additionally, the SP approach was used to distinguish the processes contributing to N2O emissions and their responses to rhizosphere processes. For our experiment, we used an intensively managed moist agricultural soil and a common cereal crop (barley) to examine whether a growing plant in the soil would: i) stimulate N2O and N2 emission; ii) directly affect the N2O/(N2O + N2) product ratio; iii) influence the sources of N2O emission; and iv) whether different N fertiliser (NO3− and NH4+ based) would interfere with the plant impact on N2O and/or N2 fluxes.

Materials and methods

Soil

In vegetable fields, much more organic and inorganic fertilizers are applied per unit area as compared to non-vegetable cropping systems causing serious environmental problems. Thus, the soil used in the incubation study was sampled from a vegetable field (Chinese cabbage-Brassica rapa L.) near the Changshu Agro-ecological Experimental Station of the Chinese Academy of Sciences, Jiangsu Province, China (31°33′5”N, 120°42′38″E).. The vegetable field (mainly leafy vegetables, e.g. pak choi and Chinese cabbage) is planted 4–5 times a year and receives circa 1044 kg N ha−1 yr−1 as a combination of mineral fertiliser and manure. The soil (clay 22.6%, silt 42.3%, sand 35.1%) was a typical Wushan Soil (Hydragric Anthrosol according to FAO 1998) developed from lacustrine sediments of the Taihu Lake with a pH (0.01 M CaCl2) of 6.1, containing 1.3% organic C. During soil sampling, the top 2 cm soil layer was removed for homogeneity (removing fresh plant residues from the top layer) and the soil was collected from a depth of 2–10 cm. Subsequently, the soil was air-dried and sieved through a 4 mm mesh and then stored at 4 °C. Prior to performing the experiment, the soil was re-wetted to ca. 85% water holding capacity (WHC; equivalent to 28% gravimetric water content) and placed in the experimental vessels (0.9 kg dry soil) for 21 days to allow the soil to equilibrate and to reduce the initial mineral N content. The NO3− and NH4+ concentrations were 0.82 ± 0.43 and 2.94 ± 0.47 mg N kg−1 soil, respectively, at the time of seeding.

Robotic continuous flow plant-soil incubation system

The incubation experiment was performed in a modified robotic continuous flow plant incubation system (PRoFLOW) using a He (80%) + O2 (20%) + CO2 (400 ppm) mixed atmosphere under LED plant light supply at Thünen Institute of Climate-Smart Agriculture in Braunschweig, Germany (Fig. 1; Fig. S1) (Senbayram et al. 2018). Transparent acrylic glass cylinders with an inner diameter of 140 mm and 150 mm height were used as incubation vessels. At the bottom of each vessel, a polyamide filter membrane (EcoTech, Bonn, Germany - hydrophilic; pore size 0.45 μm) was used for adjusting the soil moisture and sampling soil water. The experiment consisted of six treatments (n = 3): i-ii) non-fertilised control treatment unplanted (B-CK) or planted with barley (P-CK) with no N addition; iii-iv) applied with 230 mg KNO3-N kg−1 soil (equivalent to 136.5 kg N ha−1) unplanted (B-KN) or planted with barley (P-KN); and v-vi) applied with 230 mg (NH4)2SO4-N unplanted (B-AS) or planted with barley (P-AS). Briefly, soil was packed into each vessel with a bulk density of 1.25 g cm−3 (equivalent to 0.9 kg dry soil per pot). After 3 weeks of pre-incubation (85% WHC), in the respective treatments (P-CK, P-KN and P-AS), 12 barley seeds (Hordeum vulgare L. c.v. Onia; KWS SAAT SE, Einbeck, Germany) were sown into each vessel. The incubation vessels were then sealed and the atmospheric air in the vessels replaced by a pure He/O2 mixture (to remove any CO2, NO, N2O or N2 in the soil pores or headspace) by applying a vacuum from the top and filling with He/O2 mixture in three cycles that were completed within 8 h. Subsequently, the headspace of each vessel was flushed continuously with a gas mixture of He, O2 and CO2 at a flow rate of ca. 25 ml min−1. The temperature of the incubation room was set to 20 °C during the pre-incubation period and 59 days of incubation. After germination (14 days after seeding), N fertilisers were applied as a solution (50 ml) from the top of each vessel using the outlet opening. The same procedure was also applied to the non-fertilised control treatment (B-CK, and P-CK) using distilled water (50 ml) in place of the N fertiliser. The final soil moisture at seeding was calculated as ca. 85% WHC. The airflow from each plant/soil cuvette was directed sequentially to a gas chromatography system (GC-2014-Shimadzu Scientific Instruments) by two multi-positional micro-electric valves (VICI, Houston, USA), where the gas sample was analysed by a He ionization detector (HID) for N2, and O2 and an electron capture detector (ECD) for N2O quantification. Three-standard gases (containing i) 10 ppmv N2 and 0.4 ppmv N2O, ii) 50 ppmv N2 and 1.0 ppmv N2O, and iii) 100 ppmv N2 and 100 ppmv N2O inHe) were measured at the end of each cycle for calibration. Each gas sampling cycle (19 pots (including one empty vessel to check for any background noise or contamination), 3 reference gases and carrier gas ((He (80%) + O2 (20%) + CO2 (400 ppm)) - to check for any potential leaks)) was completed in 5.5 h. Outlet gas concentration of each pot was measured three times a day during the plant growth period, and at least one time a day during the post-harvest period. N2 and N2O emission rates were calculated using the commonly employed dynamic flux chamber approach (Senbayram et al. 2018) and the equation:

$$ {E}_m=\left({C}_o\hbox{--} {C}_e\right)Q/A, $$
(1)

where Em (kg N2O-N or N2-N ha day−1) is the daily flux rate, Co and Ce (kg N2O-N or N2-N m−3) are the outlet concentration of the sample pot and empty vessel respectively, Q is the flushing flow rate (m3 d−1), and A is the enclosed emission area (ha). Here, an empty vessel was measured at the end of each cycle to ensure that there is no leak in the system. The outlet N2 concentration of the empty vessel was below 2 ppm N2 throughout the experiment (see Supp. Figure 2). Light was supplied with LED labs (100 W, B.E.S.T. Agro GmbH, Germany), keeping the light intensity in a 14 h photoperiod at a minimum of 350 μmol m−2 s−1 photosynthetically active photon flux density (PPFD) at the top of the plant canopy. The plant shoots were harvested by cutting at ground level 24 days after sowing and their dry weight determined after oven drying (80 °C, 48 h). The root system was left in the soil at harvest. The total N content of the plant dry matter was determined using a NA 1500 elemental analyser (Carlo Erba, Milano, Italy).

Fig. 1
figure 1

Simplified diagram of the robotised continuous flow incubation system (PRoFLOW) used in the experiment. The system consists of 18 airtight acrylic glass cylinders and is controlled by an Arduino-based microcontroller unit (Arduino Mega 2560 R3 equipped with two 16-position relays). This control unit adjusts the position of VICI valves, gives signals to the GC (start/stop method) and the computer (start and stop data acquisition)

Full size image

Mineral N analysis

Soil samples were collected from each vessel at plant harvest (24 days after sowing) with a soil core sampler (2 cm in diameter) and at the end of the incubation period (36 days after harvest). The soil samples were extracted with 2 M KCl (1:5 w/v) by shaking for 1 h. The KCl extracts were subsequently filtered through a Whatman 602 filter paper and stored at −20 °C until analysis. The concentrations of NH4+ and NO3− in soil extracts and soil solution were measured using a continuous flow colorimetric autoanalyser (Smartchem 200S/N1104238, WESTCO, France).

Isotope analysis and N2O source partitioning

Additional gas samples for isotopic analysis were taken from each incubation vessel by attaching pre-vacuumed 120-ml serum bottles to the outlets in flow-through mode for around 2 h (Wu et al. 2017). The N2O δ15Nbulk, δ15Nα and δ18O isotope signatures were then determined by analysing m/z 44, 45 and 46 of intact N2O+ molecular ions, and m/z 30 and 31 of NO+ fragment ions (Toyoda and Yoshida 1999) on an isotope ratio mass spectrometer (DELTA V PLUS, Thermo Fisher Scientific, Bremen, Germany). Pure N2O (Westfalengas; purity >99.995%) was used as internal reference gas. The SP value of the produced N2O (SP0), i.e. prior to its partial reduction to N2, was calculated using a Rayleigh-type model, assuming that isotope dynamics followed closed-system behaviour (Lewicka-Szczebak et al. 2014). The model can be described as follows:

$$ {\mathrm{SP}}_{\mathrm{N}2\mathrm{O}-\mathrm{r}}={\mathrm{SP}}_0+{\eta}_{\mathrm{r}}\ \ln \left(\frac{\mathrm{C}}{{\mathrm{C}}_0}\right) $$
(2)

In this equation, SPN2O-r is the SP value of the remaining substrate (i.e. N2O), SP0 is the SP value of the initial substrate, ηr is the net isotope effect (NIE) associated with N2O reduction, and C and C0 are the residual and the initial substrate concentration (i.e. C/C0 expresses the N2O/(N2O + N2) product ratio). For the SP source partitioning approach, the end-member values (SPfD) were defined as 37‰ for nitrification and fungal denitrification, and − 5‰ (SPD) for bacterial denitrification and nitrifier denitrification (Toyoda et al. 2017) (see Supplementary material for further details of the calculation).

Due to the overlapping SP signatures between nitrification and fungal denitrification as well as between bacterial denitrification and nitrifier denitrification, distinguishing the N2O produced by those pathways based on SP values is impossible (Lewicka-Szczebak et al. 2014; Toyoda et al. 2017). Thus, fD-SP and ffD-SP represent the contribution of bacterial denitrification+nitrifier denitrification and nitrification+fungal denitrification, respectively, to the total N2O release calculated based on the SP0 values. However, in the B-KN and P-KN treatments, the specific experimental conditions were set up to favour denitrification, i.e. i) N was applied in the form of NO3−; ii) initial soil NH4+ content was below the detection limit (<3 mg NH4+-N kg−1 soil) with constantly low NH4+ content during the incubation (Table 1); and iii) high soil moisture (85% WHC). Therefore, the contributions of nitrification and nitrifier denitrification were assumed to be negligible in KNO3 treated soils (see Discussion). Thus, only the most plausible scenario (heterotrophic bacterial denitrification vs. fungal denitrification) was considered for the SP0 source partitioning calculation in the B-KN and P-KN treatments.

Table 1 Soil nitrate (NO3−) and ammonium (NH4+) concentrations (mg N kg−1 dry soil), plant biomass (kg dry matter ha−1) and plant N content at plant harvest (day 24) and at the end of the experiment (day 60) in non-amended control (B-CK = unplanted soil, and P-CK = planted soil), (NH4)2SO4 amended (B-AS = unplanted soil, and P-AS = planted soil), and in KNO3 amended (B-KN = unplanted soil, and P-KN = planted soil) treatments (n = 3). Means denoted by a different letter in the same column differ significantly according to the Tukey’s HSD post-hoc tests at α = 0.05. The constant value of 0.58 can be used to convert the mineral N concentration unit to kg N ha−1
Full size table

Calculations and statistical analysis

Cumulative emissions and the share of bacterial and fungal denitrification was calculated by linear interpolation between measured N2O emissions and SP0 values. Differences in cumulative N2O, N2 emissions, the N2O/(N2O + N2) ratio and soil mineral N content and the interactions were examined using a two-way analysis of variance (ANOVA-significant difference post-hoc tests at a 5% significance level) by SPSS 21 (SPSS Inc., Chicago, IL, USA).

Results

Soil moisture, mineral N and plant N

At plant harvest (day 24), the soil water content (initial ca. 85% WHC) was slightly lower in the planted soil (78.4 ± 1.6% WHC) than in the bare soil (82.9 ± 3.3% WHC) (P < 0.05; data not shown). The concentrations of soil NH4+ and NO3− in the B-CK and P-CK treatments were below 5 mg N kg−1 soil throughout the incubation period (Table 1). The soil NO3− concentrations in the B-KN treatment were significantly higher than in all other treatments at harvest (day 24). Here, the soil NO3− concentrations were two-fold higher in the B-KN compared to the P-KN treatment. No significant difference in the NH4+ concentration was found between the B-AS and the P-AS treatment at plant harvest.

At the end of the post-harvest phase, the soil NH4+ concentrations decreased markedly in the P-AS and B-AS treatments compared to the concentrations at harvest date, being more pronounced in the P-AS treatment (Table 1). In contrast, the soil NO3− and NH4+ concentrations in the B-KN and the P-KN treatments were similar to those at plant harvest. At the end of the incubation, the soil NO3− and NH4+ concentrations were the highest in the B-KN and the B-AS treatments, respectively. No significant difference was found for plant dry matter yield among treatments, whereas the plant N concentrations were significantly higher in the P-AS and P-KN treatments compared to the P-CK treatment (Table 1).

Emissions of N2O and N2

The daily N2O fluxes in the B-CK and P-CK treatments were relatively stable and low (<3 g N ha−1 d−1) during the entire experimental period (Fig. 2a, b), which was similar in all other treatments prior to the mineral N amendment. The N2O flux increased immediately after (NH4)2SO4 addition in both, planted and unplanted soil, at the same order, and then decreased gradually until plant harvest. Similarly, application of KNO3 triggered N2O fluxes in both, planted and unplanted soil; however, the increase was more dramatic compared to the (NH4)2SO4 treatments and was more pronounced, especially in the P-KN treatment. The emission of N2O peaked at day 19 in the P-KN treatment (with 298 ± 116 g N ha−1 d−1 maximum daily flux) and then decreased gradually, whereas N2O fluxes in the B-KN treatments increased gradually until plant harvest (with 142 ± 67 g N ha−1 d−1 maximum daily flux).

Fig. 2
figure 2

Daily emissions of N2O, N2, and site preference (SP0) values during the incubation period (60 days) in KNO3 amended (B-KN = unplanted soil, and P-KN = planted soil), ammonium-sulphate amended (B-AS = unplanted soil, and P-AS = planted soil), and in unamended control (B-CK = unplanted soil, and P-CK = planted soil) treatments (n = 3). Emission of N2 in control soils (B-CK and P-CK) were below detection limits and therefore the data was not presented. Error bars show the standard error of each treatment (n = 3). The arrows show the time of fertiliser application and the green dotted line denotes the point at which the plants were harvested. The legend is the same for all panels

Full size image

During the post-harvest period, N2O emissions in both, the B-AS and the P-AS treatment, increased gradually, whereas the observed N2O emission increased more rapidly in the planted compared to the unplanted soil. In contrast, in the P-KN treatment, the N2O flux remained constant for 1 day and then sharply decreased to almost background levels after harvest, whereas N2O fluxes in the B-KN treatment remained high until day 29 and then decreased gradually (Fig. 2c, d). Overall, the cumulative N2O emissions followed the order: P-KN > B-KN > P-AS>B-AS>B-CK > P-CK (Table 2).

Table 2 Cumulative emissions of N2O, N2, NO and CO2 during plant growth (11–24 days after sowing) and during the whole incubation period (11–60 days after sowing) in non-amended control (B-CK = unplanted soil, and P-CK = planted soil), (NH4)2SO4 amended (B-AS = unplanted soil, and P-AS = planted soil), and in KNO3 amended (B-KN = unplanted soil, and P-KN = planted soil) treatments (n = 3). Means denoted by a different letter in the same column differ significantly according to Tukey’s HSD post-hoc tests at α = 0.05. “–“indicates below detection limit
Full size table

Fluxes of N2 in the non-fertilised treatments were below detection limits. Similar to N2O, the N2 fluxes immediately increased to detectable levels after application of (NH4)2SO4, with the effect being more pronounced in the unplanted soil than in the planted soil (Fig. 2c, d). During the plant growth period, the emissions of N2 were slightly higher in the B-AS treatment (4–12 g N ha−1 d−1) than in the P-AS treatment (1–7 g N ha−1 d−1). During the post-harvest period, N2 emissions in the P-AS treatment increased gradually until day 46 and decreased afterwards. In the B-AS treatment, however, the N2 flux increased sharply during the post-harvest period and then decreased gradually towards the end of the experiment. In the KNO3 treated soil, N2 emissions followed almost the same trend as the N2O fluxes, i.e. they increased over time, being more pronounced in the planted compared to the unplanted soil and then decreased over time. The peak in N2 fluxes equated to a rate of 245 ± 81 g N ha−1 d−1 in the P-KN treatment and 95 ± 59 g N ha−1 d−1 in the B-KN treatment.

During the plant growth period, the cumulative total N flux (N2O + N2) in the P-KN treatment was almost three times as high as in the B-KN treatment, while it was twice as high as when taking the whole incubation period (pre- and post-harvest phase) into account (Table 2). Until plant harvest, the N2O/(N2O + N2) product ratio ranged between 0.33 and 0.67, being the highest in the B-KN and P-KN treatments and the lowest in the B-AS treatment (N2 fluxes were below detection limits in the B-CK and P-CK treatments). Here, the ratio was significantly lower in the B-AS treatment compared to the P-KN and B-KN treatments (P < 0.05); however, over the whole incubation period no significant differences in the N2O/(N2O + N2) product ratio were observed between any treatments.

N2O SP0 values and source partitioning

During the plant growing phase, the N2O SP0 values over all treatments ranged from −1.4 to 13.2‰, being the lowest in the P-CK treatment (−1.4‰ ±2.7) and the highest in the B-AS treatment (13.2‰ ±1.2; Fig. 1). The N2O SP0 values were more or less constant before harvesting in all treatments (except for a slight increase in the B-KN treatment), indicating relatively stable N2O sources in each treatment. No plant effect was observed in soils treated with (NH4)2SO4. However, in the KNO3 amended planted soil (P-KN treatment), N2O SP0 values were significantly higher than in the unplanted soil. After harvest, the SP0 values sharply decreased in the B-AS, P-AS and P-KN treatments, whereas the SP0 values slightly increased in the P-CK and B-KN treatments.

To estimate the share of each N2O emitting process on the observed N2O fluxes, source partitioning based on a two end-member model was used. The very low SP0 values in the P-CK and B-KN treatment suggest that almost all of the emitted N2O originated from bacterial denitrification, whereas the significantly higher SP0 values in the (NH4)2SO4 treatments compared to KNO3 treated soils indicated a lower share of bacterial denitrification derived N2O. Assuming that the contribution of nitrification to N2O emissions was minor in the KNO3 amended treatments (due to the constant low soil NH4+ content and high soil moisture), the increase in SP0 values over time in the B-KN treatment would imply an increase in fungal denitrification activity over time. Therefore, the higher SP0 values in the P-KN treatment compared to B-KN treatment from the beginning of the incubation indicated a higher contribution of fungal denitrification in the presence of growing plants. After harvest, the decreasing SP0 values in the B-AS, P-AS and P-KN treatments indicated an increased share of bacterial denitrification in the total N2O production. During the plant growing period the share of bacterial denitrification in P-KN (74% ±2.5) was significantly lower compared to the B-KN treatment (92% ±4.0; Fig. 3).

Fig. 3
figure 3

Contribution of fungal (+nitrification in (NH4)2SO4 treatment) and bacterial denitrification derived N2O emissions to the cumulative N2O fluxes during vegetation period (0–24 days) in (NH4)2SO4 amended (B-AS = unplanted soil, and P-AS = planted soil) treatments (Panel A), and in the KNO3 amended (B-KN = unplanted soil, and P-KN = planted soil) treatments (Panel B). Error bars show the standard error of each treatment (n = 3)

Full size image

Discussion

Plant effects on N2O, N2 emissions and N2O/(N2O + N2) product ratio

In this study, the three-fold higher N2O and N2 fluxes in the KNO3 treated soil containing plants was most likely due to the stimulation of bacterial denitrification, which was confirmed by the low N2O SP0 values. Plant roots can affect denitrification in many ways, however, of most importance is the stimulation of microbial activity, growth of C and N transformations in the rhizosphere (Hayashi et al. 2015; Guyonnet et al. 2017). In cereal plants, typically 5% of the net C fixed in photosynthesis is lost into the soil and enters the soil microbial community (Farrar et al. 2003). In the present experiment, the rhizosphere effect on denitrification can be partly attributed to the possible depletion of O2 due to root and symbiont respiration (e.g. mycorrhizas) and partly to the stimulation of microbial respiration by rhizodeposition (Bakken 1988; Hayashi et al. 2015). However, if O2 depletion was the dominant factor, we would have expected a lower nitrification rate and lower nitrification-derived N2O (i.e. lower SP values) in the (NH4)2SO4 supplied soil containing plants in comparison to the bare soil during the plant growing period, which was not the case (Fig. 1 and Table 1). Therefore, we assume that root exudates and root turnover (delivering additional electron donor to denitrifiers) rather than O2 depletion played a more important role in stimulating denitrification in our study. Nevertheless, more research is required to quantify the extent to which root exudates and O2 depletion are responsible for the observed increase in denitrification and N-derived gas emissions in the root zone.

The effects of plants and the underlying mechanisms regulating the amount of end-product N2 produced during denitrification are not fully understood due to a lack of suitable techniques for accurately quantifying N2 emissions. Using hermetically sealed soil-plant growth chambers with an argon atmosphere, Stefanson (1972) first reported that growing plants (Trifolium subterraneum and Lolium rigidum) decreased the N2O/(N2O + N2) product ratio in the planted treatments, while the ratio increased when NO3− was supplied to the plants. Henry et al. (2008) reported that the composition of root exudates may also affect the N2O/(N2O + N2) ratio of denitrification, where artificial exudates with more sugar appeared to promote more N2O reduction. The latter reflects the general assumption that labile C from root exudates and O2 depletion in the root zone may alter the N2O/(N2O + N2) ratio (Hayashi et al. 2015). On the other hand, it is commonly accepted that NO3− is preferred over N2O during denitrification processes if it is sufficiently available at the denitrifying microsites (Blackmer and Bremner 1978; Baggs et al. 2003; Smith 2010). Our recent studies (Senbayram et al. 2018; Wu et al. 2018) illustrate that the soil NO3− concentration is likely to be the predominant factor that directly regulates the denitrification end products. Conversely, O2 availability and available C appear to mainly influence the N2O/(N2O + N2) ratio when soil NO3− concentrations fall under a so called site-specific threshold value (35–50 mg N kg soil−1) (Weier et al. 1993; Senbayram et al. 2012; Qin et al. 2017). The NO3− concentrations in both the B-KN and the P-KN treatments were constantly higher than the aforementioned NO3− threshold values (>50 mg N kg soil−1; Table 1), which likely explain why no significant effect of growing plants on the N2O/(N2O + N2) product ratio was observed in the KNO3 treated soil.

When ammonium based N fertilizers are used, ammonium oxidation is the first and rate-limiting step in the nitrogen cycle. As far as we are aware, interaction effect of nitrogen form and rhizosphere processes on denitrification and the N2O/(N2O + N2) product ratio has not yet been studied. In the present study, emission rates of N2O and N2 in the (NH4)2SO4 amended planted and bare soil were significantly lower compared to the KNO3 treatments. We ascribe this to the limited nitrification activity under the given experimental conditions. During the plant growth period, the total N flux in the B-AS compared to the P-AS treatment was one-fold higher (significant (p < 0.05) when analysed separately). We attribute this to an increased plant uptake of NO3− or to the enhanced N immobilization due to root exudates inducing growth of the microbial community in the planted soil (Smith and Tiedje 1979; Kuzyakov and Razavi 2019), which depleted the soil NO3− level and decreased the nitrogenous gas emission through denitrification. This information is particularly important and can be used to implement mitigation techniques in planted soils. For example, nitrogen fertilization practices that avoid a NO3− build up in the root zone (e.g. split N application, nitrification inhibitors, and slow release fertilizers) are likely to mitigate both N2O and N2 emissions (Guyonnet et al. 2017; Senbayram et al. 2012).

During the post-harvest phase, the higher total gaseous N fluxes with lower SP0 values suggested that bacterial denitrification (likely triggered by NO3− production through nitrification) was responsible for the observed increase in nitrogenous gas fluxes in both B-AS and P-AS treatments. Moreover, slightly lower soil moisture with higher soil NO3− content at harvest indicates stimulated nitrification activity causing more rapid increase in N2O + N2 emissions in the P-AS compared to the B-AS treatment (Table 1). These suggest that harvesting of the plants (i.e. removal of plant N uptake), stimulated nitrification activity due to a lower soil moisture in the P-AS treatment and the delivery of more NO3− to the denitrifying microsites (opposite effect compared to the rhizosphere effect during plant growth) and thus triggered gaseous N loss.

Plants have also been found to directly emit N2O produced by the plants themselves (Lenhart et al. 2019) or to serve as a conduit for the transport of N2O produced in the soil to the atmosphere (Chen et al. 1999). In this study, the N2O emission rates before and immediately after harvesting showed no immediate decrease in all the planted treatments, indicating that the contribution of plant-emitted N2O was insignificant in this study. This is probably due to the barley plants being in the very early stages of growth prior to harvest, and thus a reduced ability to convey or produce N2O in comparison to mature plants with hollow stems (Chang et al. 1998). It should also be noted that we studied the initial stages of plant development (first basal N application period) where plant N uptake was limited, and where competition between roots and soil microorganisms for N is limited. Our approach was designed to minimise excess root growth in small vessels, and also to better reflect field conditions when a first basal N dressing is applied to cereals. Further studies (with larger vessels) should focus on the later growth stages of the barley crop (e.g. second N dressing period), when more competition between plant roots and microbial activity occurs for both water and nutrient uptake.

Sources of N2O as affected by growing plants

Increasing evidence suggests that actively growing plants in moist soils play a critical role not only on controlling the rate of denitrification but also on the composition of the microbial population (Guyonnet et al. 2017; Langarica-Fuentes et al. 2018). For example, Broeckling et al. (2008) and De Graaff et al. (2010) showed that labile soil C inputs by root exudates could increase the metabolic activity and gene abundance of both fungi and bacteria. Several studies have also reported that the addition of labile C to soil can induce a shift in microbial community structure (e.g. increase the fungal-to-bacterial biomass ratio) leading to enhanced fungal N2O production (Laughlin and Stevens 2002; Hayden et al. 2012; Senbayram et al. 2018; Zhong et al. 2018). Our observations support this as the SP0 values were higher in the P-KN treatment (up to 8.5‰) compared to the B-KN treatment suggesting enhanced fungal denitrification in the presence of plants alongside the delivery of additional labile C substrate to the denitrifying hotspots in the planted soil. This indicates that root exudates not only enhanced the rate of microbial activity by supplying additional electron donors, but also modified the actively N2O producing microbial community (Figs. 2 and 3).

In the current study, we presume that the enhanced fungal denitrification derived N2O in the presence of plants (only in KNO3 applied soils) was likely due to the available C supplied by root exudates. As shown in our recent study, application of organic C would enhance fungal denitrification over bacterial denitrification specifically in soils with a high NO3− content (Senbayram et al. 2018). On the other hand, with the increase in incubation time, significant decreases in SP0 in P-KN treatment (in parallel to the decreasing trend in N2O + N2 emission) indicates a clear shift from fungal to bacterial denitrification, whereas both emission rates and SP0 values remained constant for a longer period in the B-KN treatment. As fungi lack the N2O reductase enzyme (Laughlin and Stevens 2002; Shoun et al. 2012), the present study suggests that the faster depletion in soil NO3− in the P-KN treatment (converted mainly to N2O + N2) overrode the assumed stimulatory effect of rhizodeposition on fungal denitrification, leading to an increase in bacterial denitrification (decrease in SP0 values). In our previous study, we showed for the first time that labile C has a major impact on fungal denitrification as well as being dependent on soil NO3− level (Senbayram et al. 2018). Our present study supports this view, especially in the planted soils. Therefore, we may conclude that any practices that lower soil NO3− content (e.g. nitrification inhibitors (Wu et al. 2017), using ammonium based fertilizers (Senbayram et al. 2009) and/or split N application (Lebender et al. 2014) may mitigate N2O loss and total gaseous N fluxes in planted moist soils.

We also acknowledge that the SP0 source partitioning approach employed here (especially when NH4+ is used as an N source) provides a rough source estimation of emitted N2O. This is due to the i) overlapping SP0 signals of different processes (Decock and Six 2013); ii) variability of isotopologue enrichment factors of N2O reduction (Lewicka-Szczebak et al. 2014); and iii) variation in SP0 signals between different microbial strains (Toyoda et al. 2017). Nevertheless, assuming that endmember values and enrichment factors were identical in all treatments, there would be still significant differences between treatments and thus our conclusions would be still valid.

N2O and N2 emissions as affected by roots in the post-harvest period

In the post-harvest period, the N2O and N2 fluxes both gradually decreased in the NO3− treatments (P-KN and B-KN), while the NO3− contents show only minor changes when compared to the values at harvest and at the end. This implies that the decrease is most likely due to the depletion of available C in soil that limits denitrification. Previous research indicates that excision of shoots does not cause instant death of cereal roots, but rather they can remain active for 7–14 days after photosynthetic activity has ceased (Marella et al. 2017). Root activity during this period is fuelled by the progressive autolysis of the root cells and exhaustion of any remaining internal C stores. In this scenario, exudation is also expected to decline due to i) a decrease in passive exudation due to the low concentration of solutes in the cytoplasm; ii) a cessation of C losses associated with active root growth; and iii) no more unloading of C from the phloem into the apoplast (Jones et al. 2013; Paterson et al. 2005). The sharp decline of N2O and N2 emission in the P-KN treatment in the post-harvest phase can thus be attributed to the response of reduced supply of root C to the soil microbial community (Dilkes et al. 2004). On the other hand, the small N2O emission increase observed immediately after cutting in both the P-AS and P-KN treatments could be due to a pulse of root exudation following defoliation due to shifts in internal root C partitioning (Paterson et al. 2005).

Removal of the shoots induces senescence of the root system potentially leading to a large input of C into the soil via root turnover. However, the constant low N2O and N2 emission observed in the NO3− treatments during the entire period of post-harvest indicates that root integrity is not yet lost 26 days after plant harvest.

As evidenced by the decreased NH4+, increased NO3− content and extremely low SP values during post-harvest period (Table 1; Fig. 1), the significant increase of N2O and N2 fluxes in the NH4+ treated treatments is likely attributed to the ongoing nitrification, denitrification and/or nitrifier denitrification. N2O and N2 emissions in the P-AS treatment increased more rapidly than in the B-AS treatment, which is possibly due to the NH4+ released by root autolysis. Root autolysis induces the breakdown of cellular proteins to create keto acids for use in respiration, while the cleavage of the amine groups leads to the accumulation of NH4+ within the cell which is subsequently excreted into the soil to prevent cytotoxicity (Bingham and Rees 2008; Saglio and Pradet 1980). This may have provided an additional source of substrate for nitrifier denitrification to produce N2O and N2, as evidenced by the significant decreased SP values in both the P-KN and P-AS treatments in this period (Fig. 2).

Conclusions

Our results clearly show that in moist soils with a moderately high NO3− content (ca. >50 mg N kg−1 dry soil), the root system of an actively growing barley plant stimulates greater losses of both N2O and N2 through denitrification but without affecting the N2O/(N2O + N2) product ratio. The stimulation of microbial activity by rhizodeposition plays a key role in the observed increase in denitrification activity, whereas the N2O SP0 approach suggests growing plants may alter the contribution of fungal-to-bacterial denitrification-derived N2O in NO3− rich moist soils. In conclusion, we suggest that the assessments of N budgets and GHG emissions especially in agricultural ecosystems must pay more attention to the decisive influence of the rhizosphere on N2O and N2 emissions and its interaction with different forms of N fertiliser.