1 Introduction

The severe consequence of nitrous oxide (N2O) emissions from agricultural soils has raised environmental concerns, as N2O is a potent greenhouse gas (GHG), contributing to global climate change and to the depletion of stratospheric ozone (Arias et al. 2021). Emissions of N2O from agricultural soils are driven by N fertiliser N inputs (Shcherbak et al. 2014), and direct and indirect N2O emissions caused by N fertiliser addition to agricultural land account for 52% of global anthropogenic N2O emissions (Tian et al. 2020). Production of N2O can either occur via nitrification or via denitrification mediated pathways, while the reduction of N2O to N2 via denitrification assumed to be the main sink for N2O in agricultural soils. Denitrification (N2O + N2) is often assumed to be the dominant pathway of N2O production (Harris et al. 2021; Takeda et al. 2021a; Wang et al. 2016a), and a major pathway of N loss from agro-ecosystems (Butterbach-Bahl et al. 2013; Saggar et al. 2013). However, methodological difficulties regarding N2O source partitioning (Zaman et al. 2021), N2 quantification (Friedl et al. 2020; Groffman et al. 2006) and upscaling in space and time (Groffman et al. 2009) limit available experimental evidence. Pathways of N2O production and the ratio between the GHG N2O and environmentally benign N2 remain therefore two key uncertainties for N cycling in agro-ecosystems across different scales (Scheer et al. 2020), and for their representation/simulation in biogeochemical models (Del Grosso et al. 2020).

Sugarcane (Saccharum spp.) is produced under conditions favouring N2O and N2 production via denitrification: A wet and warm climate, high N substrate availability due to N fertiliser inputs (Takeda et al. 2021a) and readily available labile carbon (C) through litterfall and/or the retention of residues (Thorburn et al. 2012; Wang et al. 2016b). Sugarcane produces large amounts of harvestable biomass, averaging typically around 70 to 90 Mg fresh stalk ha−1 year (Singels et al. 2014; Takeda et al. 2022). Up to 200 kg of N ha−1 and more are applied in intensively managed sugarcane systems (Schroeder et al. 2010; Thorburn et al. 2017), to sustain productivity, and to account for the large amounts of N removed through harvest (Keating et al. 1997). Fertiliser N uptake efficiencies in sugarcane systems are reported range from 20 to 40%, and N fertiliser loss can be as high as 60% of the N applied (Chapman et al. 1994; Prasertsak et al. 2002; Takeda et al. 2021b; Vallis et al. 1996). The close proximity of sugarcane production areas to the Great Barrier Reef, and their large contribution (46–65%) to the total inorganic N load (Bartley et al. 2017) released into the catchment have led to the introduction of N fertiliser regulations in Australia, capping the total amount of N that farmers apply can in sensitive areas (Department of Environment and Science Office of the Great Barrier Reef 2019). However, N fertiliser in sugarcane systems is applied early in the season, when plant N uptake is still low. Together with flood (furrow) irrigation, and/or early season storms, this mismatch of N availability and plant N need creates ideal conditions for N2O production in sugarcane soils, when high soil nitrate (NO3−) availability and spikes in soil water content produce peak N2O emissions (Takeda et al. 2021a). The majority of N2O emissions therefore occurs within 2 to 3 months after N fertiliser application, and annual N2O emissions can exceed 10 kg of N2O–N ha−1 and can account for up to 5% of applied N fertiliser (Degaspari et al. 2020).

Similar to other agro-ecosystems, an exponential response of N2O emissions to increasing N fertiliser rates has been demonstrated for a tropical sugarcane system (Takeda et al. 2021a) and the observed exponential increase of N2O was attributed to increasing NO3− availability, potentially inhibiting the reduction of N2O to dinitrogen (N2). However, the lack of N2 measurements hindered the verification of this hypothesis. The effect of NO3− availability on the last step of denitrification remains a major unknown for N cycling in sugarcane soils due to the limited experimental evidence that includes the direct quantification of N2 (Warner et al. 2019; Weier et al. 1998). As experimental (Wang et al. 2020) and modelling approaches (Thorburn et al. 2010) estimating N2O and N2 emitted rely on the calibration of the N2O:N2 ratio, establishing its response to NO3− availability in sugarcane soils can contribute to an improved quantitative process understanding for denitrification, and help to constrain overall N2O and N2 losses within biogeochemical models.

The application of 15N urea in field experiments allows splitting N2O emissions into soil and fertiliser derived fractions. Recent research demonstrated that the response of soil-derived N2O emissions to N rates determined the magnitude of total N2O emissions (Takeda et al. 2022), highlighting the need to understand N sources for N2O production as affected by N addition. An increase of the soil derived fraction of N2O emissions with increasing N inputs has been attributed to positive priming of soil organic matter mineralisation due to added N (Schleusner et al. 2018; Xu et al. 2021). Positive priming of mineralisation is assumed to be evidenced by an increase in CO2 emissions (De Rosa et al. 2018). However, N inputs may impact on plant growth, and thus root growth and respiration, hindering the verification of a positive priming effect under field conditions. Emissions of N2O from sugarcane are usually attributed to denitrification (Takeda et al. 2021a), as pulses of N2O coincide with high soil water content driven by rainfall and irrigation. Nevertheless, the application of 15N urea (Takeda et al. 2022) hinders the attribution of N2O to either nitrification or denitrification, and experimental data on N2O source partitioning from sugarcane soils is still lacking.

The objective of this study was therefore to investigate the response of (a) magnitude and product stoichiometry of denitrification (N2O and N2), (b) N2O derived from nitrification and denitrification and (c) soil and fertiliser derived N2O from a tropical sugar cane soil to NO3− availability in a soil microcosm study. Simulating a single wetting event with subsequent drying of the soil, we aimed to test the following hypothesis: (a) Emissions of N2 are the main product of denitrification; (b) denitrification is the dominant pathway of N2O production and (c) priming of soil organic matter mineralisation causes increased N2O emissions from the soil N pool.

2 Materials and methods

Emissions of N2O and N2 from a sugarcane soil were monitored using the 15N gas flux method (Friedl et al. 2020) in a soil microcosm incubation study over 19 days. At the beginning of the experiment, soil microcosms were wetted close to saturation, and left to dry over the period of the experiment. Treatments (n = 4) included fertilisation with NO3−–N equivalent to 25, 50 and 100 μg N g−1 soil. The wetting of the air-dried sugarcane soil mimicked furrow irrigation, a common practice in the region leading to the saturation of the topsoil. The treatments sought to simulate the range of soil NO3− contents found during a trial on a commercial sugarcane farm in the Burdekin region (19° 37′ 4″ S, 147° 20′ 4″ E), QLD, Australia (Takeda et al. 2021a). Previous tests (data not shown) showed that N2O peak emissions occurred within 9 days after the wetting event. A period of 19 days was therefore deemed to be sufficient to account for N2O and N2 emissions triggered by the wetting pulse.

The soil at the site is classified as Luvisol, and the texture in the topsoil (0–10 cm) is 35% clay, 26% silt and 39% sand. The soil has total N content of 0.08%, a total organic C content of 1.6%, and the soil pH is 6.9.

2.1 Soil microcosm study

For the microcosm study, soil samples were randomly collected (0–10 cm, n = 4) from the site, pooled, air-dried and sieved to 4 mm. Equivalents of 8 g oven dry soil were then weighed into 50-ml centrifuge tubes and fertilised with 1 ml K15NO3− solution (60% atom excess) at the respective rate of 25, 50 and 100 μg N g−1 soil. The soil was compacted to a bulk density of 1 g cm−3 and additional water was applied to achieve an initial water filled pore space (WFPS) of 95%. Soil microcosms were incubated open in an incubator at a temperature of 25 °C over 19 days and the decrease in soil water content (from 95 to 50% WFPS) was estimated gravimetrically during the incubation.

Gas samples were taken on days 1, 2, 3, 5, 8, 9, 11, 15, 17 and 19 after fertilisation using a gastight syringe. Specific background atmosphere samples were taken above the soil microcosms before closure with Suba-seals (Sigma-Aldrich). Headspace gas samples were taken 3 h after closure and were transferred into pre-evacuated vials with a double wadded Teflon/silicon septa cap (Labco Ltd., Buckinghamshire, UK). Samples were analysed for N2O and CO2 concentration by gas chromatography (Shimadzu GC-2014), and for 15N2O and 15N2 using an isotope ratio mass spectrometer (Sercon Limited, 20–20, UK). Fluxes of N2O and CO2 were calculated based on the slope of the assumed linear increase in gas concentration during the closure period, corrected for temperature and air pressure. The 15N enrichment of the NO3− pool undergoing denitrification (ap) and the fraction of N2 and N2O emitted from this pool (fp) were calculated following the equations given by Spott et al. (2006) detailed in the Supplementary materials. Multiplying the headspace concentrations of N2O and N2 with the respective fp values gave N2O and N2 produced via denitrification (referred to as N2Od and N2) and enabled the product ratio of denitrification (RN2Od) to be calculated as N2Od/(N2 + N2O). Cumulative fluxes were calculated by linear interpolation between sampling days and expressed in µg N2, N2Od or N2On –N and CO2–C emitted g−1 soil.

Additional soil microcosms were established for soil mineral N extraction whenever gas samples were taken. Soil microcosms were extracted with 40 ml 2 M KCl. After shaking on a horizontal shaker (150 rpm) for 1 h, extracts were filtered through Whatman no. 42 filter paper, and analysed for inorganic NH4+–N and NO3–N by AQ2 + (SEAL Analytical WI, USA).

2.2 Statistical analysis

Statistical analyses and graphical presentations in this study were conducted using R statistical software version 3.5.2 (R Core Team 2018) with a significant level set at P < 0.05. Models estimating the response curve of cumulative N2O emissions to N rates were compared using R2 and the Akaike information criterion (AIC), root mean square error (RMSE) and the normalised RMSE (nRMSE) (Table S1). The effects of measured explanatory variables (WFPS, NO3− concentration and CO2 as a measure of heterotrophic soil respiration) on the N2O/(N2O + N2) ratio and the fraction of N2O emitted via denitrification fp N2O were assessed with generalised additive models (Wood 2015), which can quantify non-linear relationships between response and explanatory variables.

3 Results and discussion

Emissions of N2 were the main product of denitrification, accounting for > 99% of N2Od + N2 across treatments. Cumulative emissions of both N2O and N2 increased with NO3− addition, yet their curvilinear response differed markedly: Average cumulative N2O emissions ranged from 0.06 ± 0.01 to 0.26 ± 0.03 µg N g−1 soil and increased exponentially with increasing NO3− availability. Emissions of N2 increased from 48.9 ± 6.7 to 98.5 ± 18.4 µg N g−1 soil (Fig. 1) following an exponential increase to maximum. The difference between linear and exponential model regarding fit for cumulative N2 emissions was small, but AIC, RMSE and nRMSE indicated better fit for the exponential model (Table S1). The resulting product ratio of denitrification ratio (RN2Od) increased exponentially with increasing NO3− availability, while the fraction of N2O emitted from denitrification (fp N2O) exhibited a linear increase (Fig. 2). The response of fp N2O reflects increasing NO3− substrate availability for denitrification, which increasingly inhibits the reduction of N2O to N2, as evidenced by the exponential response of RN2Od. Even though this effect was superseded by the magnitude of N2 emissions, these results suggest preferential reduction of NO3− (Blackmer and Bremner 1978), and help explain the exponential increase of seasonal N2O fluxes observed from the same soil in the field (Takeda et al. 2021a). Soil NO3− concentrations decreased over the time of the incubation, averaging at < 1, 27 and 76 µg NO3−N g−1 for the 25, 50 and 100 µg NO3−N g−1 soil treatment at the end of the incubation, respectively. Soil WFPS decreased from 95% in the beginning to 50% at the end of the incubation, resembling a single wetting and drying cycle in response to flood irrigation and/or intensive rainfall. The temporal trend of these parameters indicates no N substrate limitation for N2O and N2 production for the 50 and 100 µg NO3−N treatments. The depleted NO3− concentrations in the 25 µg NO3−N however suggest potential N substrate limitation from day 3 onwards. Average WFPS decreased and was < 70% after day 8 of the incubation. This temporal trend indicates that the drying of the saturated sugarcane soil, and the subsequent supply of oxygen (O2) into the soil matrix limited N2O and N2 production for treatments (Morley and Baggs 2010) where NO3− concentrations were still sufficiently high. Increasingly aerobic conditions in the soil are also likely to explain the diminishing response of N2 emissions to NO3− availability. The response of N2O and N2 emissions to the simulated wetting pulse shows that environmental significant N losses in the form of N2O respond exponentially to excess NO3− in the soil, while overall N loss via denitrification (N2Od+ N2) reflects decreasing anaerobicicity in the soil matrix. This pattern is indicative for N2O and N2 emissions following a single wetting pulse and subsequent drying. Increasing the time of soil saturation in future research could help to expand the response curves for denitrification losses presented in this study here to account for a wider range of environmental conditions.

Fig. 1
figure 1

Cumulative N2O and N2 emissions and the respective amounts of fertiliser and soil derived N2O and N2 from a tropical sugarcane soil across three different NO3− rates (25, 50 and 100 µg N NO3− g−1 soil)

Full size image
Fig. 2
figure 2

The fraction of N2O derived from denitrification fp N2O and the product ratio of denitrification RN2Od calculated as N2Od/(N2 + N2O) from a tropical sugarcane soil across three different NO3− rates (25, 50 and 100 µg N NO3− g−1 soil)

Full size image

The observed N2O:N2 ratios were significantly lower than those reported from field studies in sugarcane systems (Warner et al. 2019; Weier et al. 1998). High soil WFPS and microbial oxygen consumption in response to the wetting of dry soil likely reduced O2 availability, conducive for complete denitrification to N2 (Senbayram et al. 2019). However, these conditions also occur in situ, where irrigation and/or rainfall drive the depletion of O2 and therefore production of N2O and N2 in the soil matrix. The soil pH of 6.9 in the study presented here is higher compared to the those reported by Weier et al. (1998) and Warner et al. (2019), ranging from 5.5 to 5.7. Complete denitrification to N2 is known to be promoted by increasing soil pH (Čuhel and Šimek 2011), and may explain the low N2O:N2 ratio observed in this study. Our findings highlight that the magnitude of N2O emissions is not necessarily a good indicator for the rate of denitrification when conditions promote complete reduction to N2 and suggests that soil pH effects on the N2O:N2 ratio should be accounted for in future experimental work and modelling approaches.

The fraction of N2O derived from nitrification, calculated as 1-fp N2O, decreased from 0.66 to 0.38 with increasing NO3− availability. The 15N gas flux method identifies the sources of N substrate for N2O production, meaning that > 35% of the N substrate for N2O production came from nitrification, and not from the soil NO3− pool. Emissions of N2O via nitrification-mediated pathways are commonly associated with aerobic conditions in the soil matrix. However, the majority of N2O emitted via nitrification-mediated pathways occurred under conditions that are usually associated with denitrifying enzyme activity: high soil WFPS, likely amplified by the wetting of dry soil, creating anaerobic conditions in the soil matrix (Giles et al. 2012). The production of N2O from N substrate derived from nitrification and anaerobic conditions suggest the reduction of nitrite (NO2−) produced via nitrification by either autotrophic nitrifiers (nitrifier–denitrification), or via heterotrophic denitrifiers (Bakken and Frostegard 2017; Friedl et al. 2021), as important pathways of N2O production in response to wetting pulses.

We used GAMs to evaluate to what extent the product ratio of denitrification and fp N2O, the fraction of N2O derived from denitrification could be predicted by soil WFPS, N substrate availability and heterotrophic soil respiration (CO2), similar to modelling of the N2O/(N2O + N2) ratio described recently by Wang et al. (2020). The variables showed little explanatory power for the daily variation of the N2O/(N2O + N2) ratio (Table 1), suggesting that additional variables, such as soil O2 availability, or measured soil gas diffusivity need to be taken into account to describe the N2O/(N2O + N2) ratio as a function of measured variables. However, WFPS (P < 0.001), NO3− concentration (P < 0.001) and heterotrophic soil respiration (P < 0.001) were significant predictors for fp N2O, explaining 90.6% of the deviance of fp N2O (Table 1). As these are standard variables measured in the field, the use of this model is a viable approach to attribute field N2O data to either nitrification or denitrification based on a soil specific and laboratory-based model of fp N2O.

Table 1 Influence of measured explanatory variables (WFPS, NO3− concentration and CO2 as a measure of heterotrophic soil respiration) on the N2O/(N2O + N2) ratio and the fraction of N2O emitted via denitrification fp N2O assessed with generalised additive models (Wood 2015)
Full size table

Soil N contributed with > 50% to N2O and N2 emissions and emissions of N2O and N2 derived from the native soil pool increased with increasing NO3− availability. This effect can indicate priming of organic N mineralisation in response to N addition and subsequent denitrification (Kuzyakov et al. 2000; Roman-Perez and Hernandez-Ramirez 2021). However, CO2 emissions did not differ between treatments (Figure S3), indicating no changes in mineralisation with increasing NO3− availability. Even though absolute amounts increased, the soil-derived fraction of N2O and N2 decreased slightly (Fig. 1). Our findings therefore denote a positive yet apparent priming effect: Similar to in situ observations at the same site (Takeda et al. 2022), the results indicate the increase in soil derived N2O and N2 driven by pool substitution and/or an overall increase in denitrification with increasing NO3− availability  (Kuzyakov et al. 2000), increasing utilisation of both soil and fertiliser N by denitrifiers.

4 Conclusions

Partitioning soil- and fertiliser-derived N2O and N2 highlights the importance of native soil N for overall magnitude of denitrification (N2 + N2O) in sugarcane systems. The exponential response of N2O to N addition is likely driven by preferential NO3− reduction, increasing N2O even when increasing NO3− levels show only a diminishing effect on the rate of denitrification (N2O + N2). Considering 3 kg N2O–N ha−1 lost season−1 in the field (Takeda et al. 2021a), the low N2O/(N2 + N2O) ratio indicates N2 emissions as a potential significant N loss pathway, demanding in-situ quantification of N2O and N2 emission in future research to establish the significance of these N losses for the overall N budget in sugarcane systems. The high explanatory power of soil parameters for the fraction of N2O from denitrification fp N2O highlights the potential to predict the contribution of N2O production pathways at the field scale using laboratory calibrated models.