A meta-analysis of fertilizer-induced soil NO and combined NO+N₂O emissions

Data extraction and compilation

We conducted a detailed review of the literature published in peer-reviewed journals through the year 2016 (cutoff date on June 30, 2016). We extracted data from publications enclosing individual field measurements with only soil NO data or in the cases that both NO and N₂O fluxes were simultaneously measured (Fig. 1, Table S1, Dataset S1). All published data were derived from the Web of Science and Google Scholar, and papers published in the China Knowledge Resource Integrated Database (CNKI) with English abstract, as well as gathering and re-evaluating the older literature cited in the prior review of Stehfest & Bouwman (2006). Different combinations of searching keywords (‘nitric oxide’ OR ‘NO_x’ OR ‘NO’ AND ‘soil’) were used for data extraction. The final database consisted of 520 field measurements derived from 114 publications, of which 81 recent measurements taken from vegetable cropping fields and 17 measurements from rice paddies were not evaluated by Stehfest & Bouwman (2006). In addition, 43 studies that contain 173 simultaneous measurements of NO and N₂O fluxes were available in the database (Table S1, Dataset S1).

We complied with the following criteria to avoid bias in selecting publications. Only field measurement data were included in this analysis, without considering laboratory or pot experimental data, vertical profile monitoring data over vegetation canopy or modeling results. For data from natural ecosystems, the occasional field NO flux measurements without covering the whole experimental period or the number of consecutive measurements less than three date-points were excluded. For data from croplands, the consecutive measurement period covers at least one whole cropping season. The raw data were either obtained directly from tables and texts or extracted by digitizing graphs using the getdata graph digitizer software (version 2.26, http://www.getdatagraph-digitizer.com/download.php). For each literature selected, the following original documented information was compiled: soil NO fluxes, total fluxes of soil NO and N₂O, NO/N₂O and/or NO/(NO+N₂O) ratios, direct emission factors of N fertilizer for NO and/or combined NO+N₂O, location (longitude and latitude), general climate (tropical, subtropical, temperate and cool), experimental duration, land-use type, dominated vegetation, soil properties (texture, pH, mean temperature in 0–20 cm depth, WFPS, soil organic C (SOC) content, total N content, C/N ratio and mineral N), frequency of measurement, number of replicates, as well as fertilizer rate and source (Table S2). If only soil organic matter (SOM) was provided, we converted SOM to SOC using a Bemmelen index value of 0.58 to include as many data as possible.

Data were subjected to a standardization process to allow for comparisons. We calculated the balanced mean values of soil NO and N₂O fluxes as well as their emission factors with the residual maximum likelihood (REML) procedure using genstat release 4.2 to minimize the heterogeneity resulting from missing values and unequal number of observations among reviewed literature (Payne, 2000). Measurements from different sites or different fertilizer rates within a single study were considered as independent field measurements in data collection. In the cases that mean seasonal or annual soil NO and N₂O emissions were not presented directly, hourly or daily fluxes were converted to kilograms of N per hectare for the experimental period. For grassland and forest soils, some nonconsecutive data were simultaneously obtained in both wet and dry seasons; in these cases, their grand mean was derived for the database. Otherwise, for measurements were only taken in a single season (dry or wet) and where temperature performed as the dominant factor driving soil NO and N₂O fluxes with significant correlations, the flux data were further rectified by the annual mean temperature using the temperature-dependent statistical model that was established in an earlier review of Yan et al. (2005). In addition, atmospheric N deposition was not considered in this analysis for most lack of information on actual N deposition rates under experimental field conditions.

In further data compiling prior to analysis, we categorized the soils into five land-use types as paddy field, vegetable field, other cropping upland (including wheat, corn, barley and cotton), grassland and forest based on the experimental land-use mode. Only two studies with three measurements were taken from deserts and three studies with four measurements were available in wetlands, which eventually led to the exclusion of them in our analysis. Besides fertilizer application, other five experimental factors including crop residue incorporation, experimental drought, tillage, biomass burning and atmospheric carbon dioxide concentration ([CO₂]) enrichment were also taken into consideration in this analysis. Soil texture was grouped into three general classes (coarse, medium and fine) due to the inconsistent reporting of soil texture in the literature (e.g., general qualitative description, particle size distribution, soil taxonomical unit).

We quantified the direct emission factors of N fertilizer for NO (EF_NO) and a combination of NO and N₂O emissions (EF_c) partitioned by various ecosystems and fertilizer sources. For studies without directly reporting the EF_NO or EF_c but with available data, we estimated them based on the gas flux data from both the control and fertilized treatments and total N input rate over the experimental period. Specifically, the EF_NO is defined as NO emission from N fertilizer plots minus the emission from unfertilized control plots expressed as a percentage of N applied. Similarly, the EF_c is calculated by the difference in the total of NO and N₂O emissions between N fertilizer treatments and controls expressed as a percentage of N input in this analysis.

Data analyses

In this study, data derived from fertilized soils (48 studies with 219 paired measurements) and from studies exposed to other available experimental factors (57 measurements from 23 studies) were separately analyzed using meta-analysis procedure to examine the general response of soil NO emissions to fertilizer and other typical experimental factors. The means of soil NO flux from treatment (X_t) and control (X_c) groups were used to calculate effect size in the form of natural log-transformed response ratio (RR). The standard deviations of both treatment and control were included as a measure of variance:

(1)

where X_t and X_c are means of NO fluxes for the treated and control groups, respectively. Its variance (v) is estimated as:

(2)

where n_t and n_c are the sample sizes for the treatment and control groups, respectively; s_t and s_c are the standard deviations for the treatment and control groups, respectively.

We conducted a weighted meta-analysis using RRs, where mean effect size for each category was calculated using a categorical random effects model. Groups with less than two paired observations were excluded from this study to meet the criteria for meta-analysis. The overall mean effect size and 95% confidential interval (CI) of each grouping category generated by bootstrapping (9999 iterations) were calculated with metawin version 2.0 statistical software (Rosenberg et al., 2000). Mean effect sizes of X_t were considered to be significantly different from those of X_c if the 95% CIs did not overlap with the line RR = 0, and significantly different from one another if their 95% CIs did not overlap.

In addition to the meta-analysis procedure, fitting of data to Gaussian distribution function was carried out using the sigmaplot version 12.0 Software (Systat Software, Inc., San Jose, CA, USA), and frequency distribution of RR was plotted to reflect variability of among individual studies with the following Gaussian function (i.e., normal distribution) (Luo & Zhou, 2006):

(3)

where y is the frequency of RR values within an interval, x is the mean of RR for the given interval, μ and σ² are the mean and variance across all RR values, respectively, and a is a coefficient indicating the expected number of RR at x = μ. One-way anova was performed to evaluate the differences in soil NO emissions and emission factors among ecosystem types, fertilizer sources and other experimental factors. We conducted a Pearson correlation to identify key factors controlling soil NO and N₂O emissions.

Sensitivity analyses

We performed sensitivity analyses to test the robustness of our meta-analysis on response of soil NO emissions to fertilizer and other experimental factors. We removed the outlier studies to perform the same meta-analysis procedure and compared the results with those of the original meta-analysis. Furthermore, we conducted the same meta-analysis procedure by excluding datasets without variances reported, and then repeated the comparisons with the results of original meta-analysis.

Soil NO fluxes across ecosystems

When summarizing all field measurements, the global mean of soil NO-N fluxes was estimated to be 4.06 kg ha⁻¹ yr⁻¹, and it significantly differed among ecosystem types (Fig. 2a). On average, soil NO-N fluxes were the greatest in vegetable croplands, with a grand balanced mean of 9.75 kg ha⁻¹ yr⁻¹. The mean of soil NO-N fluxes was relatively comparable among other upland croplands (2.83 kg ha⁻¹ yr⁻¹), grasslands (2.37 kg ha⁻¹ yr⁻¹) and forests (2.74 kg ha⁻¹ yr⁻¹), while paddy fields had the lowest NO-N fluxes (0.11 kg ha⁻¹ yr⁻¹). For the controls without N fertilizer application, background emissions of soil NO-N averaged 0.05, 0.97, 1.02 and 1.24 kg ha⁻¹ yr⁻¹ for paddy field, vegetable field, other cropping upland and forest soils, respectively, while grasslands had the largest background emissions of soil NO-N (1.67 kg ha⁻¹ yr⁻¹).

Fertilizer effect on soil NO fluxes

The datasets were homogenous for fertilizer effect on soil NO fluxes with a normal distribution pattern of effect size in the meta-analysis (Fig. 3a). When averaged across all field measurements, N fertilization significantly increased NO emissions by 25%, with a 95% confidence interval (CI) of 20–32%, while NO response varied with N fertilizer types (Fig. 3b). When pooling data at different N fertilizer types, synthetic N fertilizer application induced the greatest increases in soil NO fluxes (SN, +38%), compared with organic N (ON, +20%) or mixed N fertilizer (Mixed, +18%). Relative to synthetic N fertilizer alone, synthetic N fertilizer combined with nitrification inhibitors substantially reduced soil NO emissions by 81% (SN+NI, Fig. 3c). Among chemical N fertilizer varieties, soil NO emissions had the largest response to ammonium nitrate (AN, +59%) and the smallest response to compound fertilizer (CF, +9%), relative to urea (U, +40%) or controlled-release urea (CRU: +36%).

Direct emission factors of NO and combined NO+N₂O

Soil NO and combined NO+N₂O emissions were significantly and positively correlated with N fertilizer input rate when pooling all field flux measurements across fertilized soils (Table 1, NO-N: r = 0.78, P < 0.001; (NO+N₂O)-N: r = 0.68, P = 0.01). When averaged all available data, the global mean direct emission factors of N fertilizer for NO (EF_NO) and combined NO+N₂O (EF_c) were estimated to be 1.16% and 2.58%, with CIs of 0.71–1.61% and 1.81–3.35%, respectively, while they were significantly different among various ecosystems and N fertilizer types (Table 2; Figs 2b and 3c). On average, the EF_NO was significantly greater in forests and vegetable fields relative to grasslands, but there was no significant difference in EF_c among them (Table 2). The EF_NO were significantly higher in forests (2.39%) than in grasslands (0.87%), while there was no significant difference in EF_c between them (3.82% vs. 3.36%). Within the croplands, vegetable fields showed the highest EF_NO (1.71%) and EF_c (4.13%), while the EF_NO (0.01%) and EF_c (0.08%) were the lowest in paddy fields.

The EF_NO was linearly and significantly correlated with EF_c (Fig. S1), suggesting that both NO and N₂O emissions were simultaneously increased by N fertilizer application. However, both EF_NO and EF_c significantly differed with N fertilizer sources (Fig. 3c). Relative to organic N fertilizer, the EF_NO was significantly greater for synthetic N fertilizer, but there was no pronounced difference in EF_c between them. Relative to organic or synthetic N fertilizer application alone, the EF_c was significantly lower for organic N combined with synthetic N fertilizer (Mixed, Fig. 3c). Moreover, synthetic N fertilizer combined with nitrification inhibitors application greatly decreased EF_NO and EF_c by 82% and 39% as compared with synthetic N fertilizer being applied alone, respectively. Among chemical N fertilizer varieties, ammonium nitrate induced the highest EF_NO (2.93%) and EF_c (5.97%), while the EF_NO (0.45%) and EF_c (0.73%) were the lowest for controlled-release N fertilizer. On the other hand, the EF_NO was negatively correlated with soil pH (Fig. S2), suggesting that N fertilizer-induced NO emissions would be greater in soils with lower pH.

Soil NO and its link to N₂O regulated by experimental and environmental factors

When averaged across all observations, soil NO emissions were significantly affected by experimental factors, including crop residue incorporation, drought, tillage, burning and atmospheric [CO₂] enrichment (Fig. 4a). Soil NO emissions were significantly decreased by crop residue incorporation (−9%) and atmospheric [CO₂] enrichment (−6%), in contrast to an increase in soil NO emissions due to biomass burning (+12%), experimental drought (+14%) or soil tillage (+30%).

Environmental factors such as climate and soil properties also played an important role in regulating soil NO emissions (Table 1; Fig. 4b). When pooling data at different climate zones, soil NO fluxes were the greatest in subtropical regions and the lowest in cool regions (Fig. 4b). Soil NO fluxes depended on soil properties, including soil pH, soil texture, soil organic carbon (SOC) content and soil C/N ratio. When experimental soils were divided into acid/neutral (pH ≤7) and alkaline soils (pH >7), soil NO emissions were significantly greater in acid or neutral soils than in alkaline soils (Fig. 4b). Soil NO fluxes depended on soil textures, decreasing from the coarse and medium soils to fine soils. Given that soil SOC content of 1.5% or soil C/N ratio of 10 was adopted as the threshold for further subgrouping, soil NO fluxes were greater in carbon-enriched soils (SOC content >1.5%) relative to carbon-poor soils (SOC content <1.5%), or in soils with lower C/N ratios (≤10) relative to soils with higher C/N ratios (>10).

The NO-N/N₂O-N or NO-N/(NO+N₂O)-N ratio was negatively correlated with soil water content (WFPS, %) (Table 1, Fig. 5a), suggesting that nitrification instead of denitrification might dominate soil NO emissions irrespective of soil or ecosystem type. In addition, soil NO fluxes were linearly increased with soil temperature (Table 1, Fig. 5b). The combined NO+N₂O emissions were positively correlated with soil mineral N content (Table 1, Fig. 5c). However, no significant differences were found between cropping and noncropping soils in terms of above correlations. Across all observations, soil NO fluxes tended to increase with the extension of experimental duration (Fig. 5d), suggesting that soil NO emissions might have been underestimated in short-term field studies. For a systematic review, an overall schematic response of soil NO to its controlling factors is presented in Fig. 6.

Soil NO fluxes varying with land-use type

In this meta-analysis, the vegetable fields were highlighted from upland croplands category due to its substantially high NO emission rates. On average, soil NO fluxes were the greatest in vegetable fields and the lowest in paddy fields (Fig. 2a). The higher NO emissions from vegetable croplands have been recently documented in field studies (Akiyama & Tsuruta, 2003; Mei et al., 2009; Pang et al., 2009; Deng et al., 2012; Yao et al., 2015). Typically, intensified vegetable production is characterized by excessive N fertilizer input rate, especially in greenhouse vegetable cropping systems to maintain high yield. It is generally believed that soil NO emissions were strongly correlated with N applied rates (Smith et al., 1997; Bouwman et al., 2002a), and thus, high N input rate would incur substantial soil NO emissions in vegetable croplands (Fang & Mu, 2006; Yao et al., 2015). In addition, higher soil NO fluxes might be also due to frequent irrigation in vegetable cropping fields because soil NO fluxes have been found to be greatest when dry soils are rewetted (Homyak et al., 2016). Relative to vegetable croplands, soil NO fluxes were relatively lower in other upland croplands, mainly due to lower N fertilizer input rate and less irrigation in terms of per unit area. Soil NO fluxes were comparable between grasslands and the other upland croplands. Although synthetic N fertilizer is seldom applied in grasslands and forests, organic N input by animal waste could be occasionally used in such seminatural or natural ecosystems. Indeed, the ‘mining’ of soil N following land-use change from natural forest or grassland to cultivated agricultural land has been of great concern to its important role in global N budget (Davidson, 2009; Smith et al., 2012).

Among all ecosystem types, paddy fields were shown to have the lowest NO source strength. The lowest NO fluxes in paddy fields might be closely related to the waterlogged soil conditions (Zhou et al., 2010; Gaihre et al., 2015; Zhao et al., 2015). It is generally believed that soil NO is predominantly produced through microbial nitrification rather than denitrification (Russow et al., 2008). The produced NO could be further reduced toward the formation of N₂ under poor aeration conditions due to waterlogging in paddy fields (Russow et al., 2009).

Fertilizer effects on soil NO fluxes

When averaged across all fertilizer sources, fertilizer application significantly increased soil NO emissions by 25% (Fig. 4b). As reported in previous individual studies, linear or nonlinear positive responses of soil NO emissions to N fertilizer input have been recently documented (e.g., Shcherbak et al., 2014; Zhao et al., 2015). Synthetic N fertilizer relative to organic or mixed N fertilizer application was more effective at stimulating NO emissions in this meta-analysis, generally in agreement with previous field studies (Hayakawa et al., 2009). However, Deng et al. (2012) reported that soil NO emissions were greater for manure relative to synthetic N fertilizer treatments in a vegetable cropping system. Nevertheless, optimizing fertilizer types, apart from cutting down N input amounts, would be an effective option for mitigating soil NO emissions, especially in agronomic cropping soils.

Synthetic N fertilizer combined with nitrification inhibitor or controlled-release N fertilizer application decreased soil NO emissions as compared with common synthetic N fertilizer application alone in this meta-analysis, although it is limited by few available data (Akiyama et al., 2000; Cheng et al., 2002; Hou & Tsuruta, 2003; Sanz-Cobena et al., 2012). As proposed by Sánchez-Martín et al. (2010), the lowered pool of NH₄⁺ in soils as a consequence of controlled-release N fertilizer can weaken the nitrification potential driving NO release. Menéndez et al. (2009) found a significant effect of urease inhibitor on reducing NO emissions but greatly depended on soil water status. Among different chemical N fertilizer varieties, ammonium nitrate was most effective at increasing soil NO emissions. Presumably, direct mineral N substrate enrichment greatly facilitates soil NO release relative to other indirect N chemical sources, but more field verifications are needed.

Direct emission factors of NO and combined NO+N₂O

The direct emission factor of fertilizer N for NO (EF_NO) has been reported to show large variabilities, highly dependent on climate, soil conditions and site-specific management practices (Bouwman et al., 2002a; Dobbie & Smith, 2003). By summarizing available data, the EF_NO was the greatest in forests among different ecosystem types (Fig. 2b), suggesting that N input has a high risk of being lost as NO in forests (Venterea et al., 2003). The soil pH might have played an important role in regulating EF_NO in different ecosystems. It is generally believed that nitrite is stable at soil pH >5.5 and most NO can be resulted from chemical decomposition (chemo-denitrification) of NO₂⁻ in acidic soils (Kesik et al., 2006; Heil et al., 2016). By examining the database of this study, soil pH was significantly lower in tropical/subtropical forests than in grasslands (4.6 vs. 6.8), and negative correlation of EF_NO with soil pH suggested that lower pH would benefit for higher N fertilizer-induced soil NO emissions from forests (Table 1; Figs 2 and S2). Overall, the forest soils subjected to long-term atmospheric N deposition would result in high ratio of anthropogenic N loss as NO (Venterea et al., 2003).

The global mean EF_NO was estimated to be 1.16%, with a CI of 0.71–1.61%, in this study. This estimate based on updated data samples, particularly including new 81 recent measurements taken from vegetable cropping fields, is obviously greater than the previous estimates of 0.55% or 0.71% for global agriculture that were based on limited available data (Bouwman et al., 2002b; Yan et al., 2003b, 2005; Stehfest & Bouwman, 2006). When averaged across studies within croplands, vegetable fields had the greatest EF_NO (1.71%) and EF_c (4.13%), which would be largely associated with drying–rewetting episodes created by frequent irrigation in vegetable cropping systems. The EF_NO was significantly greater in vegetable fields than in grasslands, but EF_c did not differ between them, suggesting that relatively higher NO-N/N₂O-N ratio would be obtained in vegetable fields with larger N input rate (Zhang et al., 2016). In addition, the EF_NO was negligible (mean: 0.01%) in paddy fields, mainly ascribed to negligible NO emissions from paddy soils under waterlogging conditions (Zhou et al., 2010).

When averaged 173 field simultaneous measurements of NO and N₂O fluxes, the direct emission factor of combined NO+N₂O (EF_c) was estimated to be 2.58%, with a CI of 1.81–3.35%. This value of EF_c is greater than the sum of global mean direct emission factor of NO (1.16%) estimated in this meta-analysis and the global IPCC default emission factor of 1.0% for N₂O (IPCC, 2006). In this meta-analysis, the direct emission factor of fertilizer N for soil N₂O (EF_N2O) was estimated to be 1.42%, with a CI of 1.04–1.76%, higher than the global IPCC default value of 1.0%, derived from over 1000 field N₂O flux measurements prior to the year of 2006 (IPCC, 2006). The higher EF_N2O in this meta-analysis was largely contributed by recent flux measurements showing much greater fertilizer-induced N₂O emissions from grassland and vegetable soils (Dataset S1, Table 2). Our mean EF_N2O estimates in fertilized grasslands (2.49%) and vegetable cropping systems (2.42%) were highly close to the global top-down mean EF_N2O estimates of 2.5% over the period of 1860–2005 (Davidson, 2009). Indeed, the global IPCC bottom-up EF_N2O of 1% has been shown to be much lower than some recent regional/global top-down estimates of 3–5% (Crutzen et al., 2008; Davidson, 2009; Smith et al., 2012; Griffis et al., 2013) or bottom-up estimates of 1.75–1.8% (Grace et al., 2011; Griffis et al., 2013; Shcherbak et al., 2014).

This meta-analysis primarily focused on fertilizer-induced NO and combined NO and N₂O emissions; the estimation of EF_N2O was only based on much smaller sample size (173 field simultaneous measurements of NO and N₂O fluxes). Some EF_NO and EF_C estimates were also limited by few available data in this study and deserve to be further examined with more field measurements (Fig. 2b). Nevertheless, this study first attempted to estimate the bottom-up EFs in forest and grassland given that land-use change from natural forest or grassland to cultivated agricultural land could constitute an important source of global N₂O (Davidson, 2009). In particular, the fertilizer N-induced total emissions of NO and N₂O should be of more concern in future studies in terms of their interrelation in soil N cycling.

Dependence of soil NO emission on experimental and environmental factors

We also examined the response of soil NO emissions to other experimental factors apart from fertilizer application (Fig. 6a). When averaged across all observations, available experimental factors had significant effects on NO emissions, but the size of effect substantially differed. Of which, crop residue incorporation and atmospheric [CO₂] enrichment significantly decreased soil NO emissions, in contrast to significant positive responses to experimental drought, soil tillage and aboveground biomass burning. The negative response of soil NO emissions to crop residue incorporation may be largely attributed to the declined soil oxygen availability for nitrification following crop residue amendment, dependent on specific crop residue type and its C/N ratio (Yao et al., 2009; Zhang et al., 2011).The negative response of soil NO fluxes to elevated [CO₂] was associated with the highly enhanced N uptake by plants grown under elevated [CO₂], which would decrease mineral N availability to soil microbes (Hungate et al., 1997; Hu et al., 2001). Moreover, the decreased surface soil water content under elevated [CO₂] could also constitute a key limiting factor for soil NO emissions (Mosier et al., 2002). However, the current data on soil NO emissions in response to elevated [CO₂] are only limited in grassland soils, and more individual evidence in other extensive ecosystem types is anticipated.

Experimental drought significantly increased soil NO emissions in this meta-analysis. As suggested by Davidson et al. (2008), reduced precipitation may have important feedback effects on climate change by altering soil-atmospheric N oxide gas emissions. It has also been found that dry soil conditions tended to facilitate soil NO emissions over N₂O emissions (Firestone & Davidson, 1989; Davidson et al., 2000). In addition, soil N availability to nitrifying bacteria may be improved in drought-exposed soils, subjected to site-specific soil characteristics (Goldberg & Gebauer, 2009).

Soil tillage relative to no-tillage practice significantly increased soil NO emissions, and our datasets generally presented uniform positive responses in this analysis. The improved soil aeration following tillage events would greatly increase oxygen availability favoring nitrification that dominates soil NO production (Sanhueza et al., 1994; Yamulki & Jarvis, 2002; Yao et al., 2009). Nevertheless, soil tillage effects on soil N oxide emissions were dependent on soil textures, where coarse- and medium- vs. fine-textured soils may weaken nitrification potential in tilled soils (Mummey et al., 1998).

Biomass burning, as a typical management practice in chaparral and tropical savannas, was found to have a significant positive effect on soil NO emissions in this analysis. The burning-induced postfire pulse emissions of NO have been often documented (e.g., Williams et al., 1992; Poth et al., 1995). As suggested by Levine et al. (1996), for example, the increased ammonium substrate for nitrification may account for the stimulated NO emissions following burning of combustible matter. However, the extent of stimulation substantially varied with original soil moisture conditions and vegetation cover types (Anderson & Poth, 1998).

Soil NO emissions also depended on the environmental parameters (Fig. 4b). The subtropical soils presented to most facilitate NO emissions, relative to the lowest NO release rate in cool regions. Our meta-analysis results are well in line with the modeling estimates by Yan et al. (2005), showing that the highest NO emissions occurred in subtropical regions with relatively dry climates favoring soil nitrification. Among soils with different pH values, the acid or neutral relative to alkaline soil conditions significantly increased soil NO emissions. Low soil pH could inhibit NO reduction to N₂O (Šimek & Cooper, 2002), or low soil pH may promote chemical decomposition of nitrite, yielding NO as the primary product (McKenney et al., 1990). As reviewed by Bouwman et al. (2002b), aerobic soil conditions may be more easily reached and maintained for longer periods within aggregates in coarse- than in fine-textured soils, which would in turn facilitate soil NO release with improved oxygen availability and gas diffusion. Besides, soils with relatively higher soil C contents but lower C/N ratios were observed favorable for NO emissions. Soils rich in SOC but with low C/N ratios tend to create more available C and N substrate for nitrifiers or denitrifiers driving NO production.

Future studies on soil NO emission

This meta-analysis provided an insight into fertilizer-induced soil NO emissions and the link to N₂O emissions across various ecosystems, soil types and climatic regions (Fig. 6). However, most field studies were limited within seasonal or annual timescale, and there is an urgent need for continuous long-term (multiyear) measurements to elucidate the temporal variation. In particular, the continuous measurements are extremely limited in forest and grassland ecosystems and few measurements of NO fluxes from croplands are available in South America, India and South-East Asia, directly leading to large uncertainties existed for the current available bottom-up estimates of NO emissions from related soils. On the other hand, although few studies have examined indirect NO emissions or overall accounting of both indirect NO and N₂O emissions from N leaching and runoff in agro-ecosystems, more field measurements at expanded geographic range should be strengthened in future work due to much anthropogenic N loss through the above two major pathways.

Besides field measurements, more condition-controlled experiments are still needed to further explore the mechanisms controlling soil NO emissions, such as closely linking it to the microbial analysis of abundance dynamics and composition of functional microbes involved in NO production from soils. In addition, more credible statistic or process-oriented models should be developed based on updated data sources to minimize the current large uncertainties in global or regional soil NO estimates. Some options such as mixed N fertilizer, synthetic N fertilizer combined with nitrification inhibitors instead of synthetic N fertilizer alone, decreasing N fertilizer application rate and low irrigation frequency could be adopted to mitigate soil NO emissions.