Introduction

Nitrous oxide (N2O) is a major long-lived anthropogenic greenhouse gas with about 265–298 fold greater potential for global warming in the atmosphere compared to carbon dioxide1. It is also an ozone-depleting substance2, produced mainly in the soil from nitrification and denitrification processes3. Its concentration in the atmosphere has increased to 331 ppb4 from 270 ppb in the pre-industrial age5. This increase of N2O in the atmosphere is mainly attributable to rise in anthropogenic nitrogen (N) input to soil6,7 and this anthropogenic N input to soil increases as more natural ecosystems are converted to croplands.

Soil salinity can influence N2O flux in different ways. An increase in salinity in a non-saline soil can increase8 or have no effect on N2O emission9. Similarly, on naturally occurring saline soils, both decreases8 and increases10 in the N2O flux have been found in response to increase in the salinity. These results suggest an ambiguous role of salinity in N2O emission. Some meta-analyses11,12 have reported that alkaline soil emits less N2O than neutral or acidic soil. In alkaline soil NH4 may be converted to NH3 and volatilize to the atmosphere whereas NH4 is retained in acid soil, favoring N2O formation13. N loss from alkaline soil may be high in total, but if much of the N is lost in the form of NH3 there may be less NH4 available for nitrification and subsequent denitrification. This evidence suggests that in naturally occurring saline–alkaline soil, the influence of both salinity and alkalinity may significantly affect the N2O formation processes. So, quantifying N2O flux from the saline–alkaline soil may help to increase knowledge on its contribution to soil-atmosphere exchange of N2O.

Land-use change (LUC) from natural to semi-natural or artificial ecosystems can have different effects on N2O emission14,15,16. Specifically, conversion from natural to artificial ecosystems with the addition of N fertilizer significantly increases N2O emission while conversion to semi-natural may or may not increase the emission16,17,18. LUC directly impacts on soil physical, chemical and biological properties19,20, the main factors affecting N2O emission21,22. N2O emission from soil is reduced when pasture is forested14, while conversion of rainforest to pasture or plantation leads to an increase in N2O emission15. A recent study found that the conversion of a conventional agricultural field to bio-energy crops had no effect on N2O emission23. Therefore, knowing which LUC practice is appropriate in terms of lower N2O emissions, and its implementation could mitigate N2O emission to the atmosphere and associated impact of climate change. Moreover, various LUC practices14,15,23 have different or no effect on N2O emission, indicating that LUC is rather an indirect cause of N2O emission. The main reason for the differences in N2O emission due to LUC is probably the alteration of the controlling factors of N2O production and reduction processes in the soil. So, quantifying N2O flux from LUC, along with soil physical and chemical parameters, would further enable understanding of the main driving factors for N2O production and consumption in the soil.

N has two stable isotopes i.e., 14N and 15N. δ 15N of a sample is the deviation of the samples’ 15N/14N from the respective isotope ratio of the reference material24. Previously, the 15N in N2O emitted from soil has been used to identify the processes for N2O production i.e. nitrification and denitrification; however, using only 15N values in N2O may mislead the interpretation7 as both the processes generally occur in the soils, possibly in different horizons or niche. The 15N in N2O emitted from soil depends on the 15N content of the substrates i.e. NH4 and NO3, different microbial community composition, pH, temperature and substrate availability24,25. Although it is difficult to predict the sources of N2O emission using solely 15N signatures in the N2O, these values could be used to distinguish between N2O emitted from natural and artificial ecosystems25. N addition in the artificial ecosystems increases the N availability which depletes the 15N in N2O25. So, LUC from natural ecosystems to cropland may not only influence the N2O fluxes but also the 15N in the emitted N2O as N availability is altered. For example, mean 15N in N2O emitted from natural tropical forest, sub-tropical forest and subarctic soil are − 18.0‰, − 14.3‰, and − 13.0‰, respectively 25,26,27; while more depleted after N application i.e., − 37.9‰28 to − 34.3‰25 in fertilized soil. The difference of 15N in N2O is useful to distinguish N2O emitted between fertilized and natural soils, and it arises from anthropogenic N addition to soil25,28. Moreover, application of N fertilizer leads to high concentrations of NO3 in the soil, resulting in a decrease in N2O reduction to N2 and therefore a higher N2O to N2 ratio from the denitrification process29. The reduction of N2O to N2 through denitrification leads to 1–24‰ 15N enrichment of the remaining N2O30. So, differences in the capacity to reduce N2O to N2 between various ecosystems may also influence the 15N in emitted N2O.

To feed the world’s growing population requires an additional 2.7–4.9 Mha of cropland per year on average31. Due to limited land resources, natural saline–alkaline areas are being reclaimed for producing food32. Agricultural soil alone will contribute about 59% of total global N2O emissions by 20303 as fertilizer application will need to increase by about 35–60%33. Therefore, it is important to quantify, and develop measures to mitigate increases in N2O fluxes resulting from the conversion of natural saline–alkaline grassland to cropland. Furthermore, Tamarix chinensis (Tamarix), a salt tolerant native species of shrub, is commonly used for the restoration of saline–alkaline soil in coastal areas in China (semi-natural ecosystem)34. Local governments have launched a coastal ecological restoration program promoting the planting of Tamarix35; however, its effect on N2O emission is unknown. Though Zhang et al.36 reported the differences in N2O emission from various natural vegetation in saline–alkaline coastal areas, the impact of LUC from natural to semi-natural or artificial ecosystems on the dynamics of N2O emissions from saline–alkaline soil is unknown. Moreover, different plant species have been reported to modify the soil characteristics in varying ways, resulting in significant changes in N2O fluxes37. Therefore, we hypothesize that: (1) LUC from native saline–alkaline ecosystem (grassland) to semi-natural (Tamarix) may significantly influence N2O flux, along with soil environmental variables (soil temperature, soil moisture, ammonium, nitrate), because of the difference in plant species but have no effect on the 15N in N2O emitted from the soil because there is no addition of anthropogenic N and (2) LUC from native saline–alkaline ecosystem (grassland) to artificial (cropland) may influence both N2O flux and the15N in N2O due to anthropogenic N addition and changes in management practices. Therefore, we expect that the 15N in emitted N2O could be used to distinguish N2O emitted between unfertilized (natural and semi-natural ecosystems) and fertilized (cropland) ecosystems but not between different unfertilized ecosystems (grassland and Tamarix).

Methods

Site description

The study was carried out from April 2017 to June 2018 at the Haixing experimental station of the Center for Agricultural Resources Research (CARR), Institute of Genetics and Developmental Biology (IGDB), Chinese Academy of Sciences (CAS). This site is located near the Bohai sea in Haixing county (117°33′5″ E, 38°09′59″N) of Hebei province, China (Fig. 1). The site has a semi-humid monsoon climate with more than 75% of precipitation occurring during the rainy season, i.e. from July to September. The mean annual precipitation is 582 mm. The groundwater table is at 0.9–1.5 m depth. The soil in this area is classified as solonchak (18.1% clay and 7.8% sand). The salt content in the area ranges from 3 to 20 g kg−1 soil38.

Figure 1
figure 1

Map and an aerial photo of the study site. 1, 2 and 3 in the aerial photo represent the positions of the grassland, cropland, and Tamarix, respectively. The map was created using ArcGIS (v10.3.1) (ESRI Inc.) and Google Earth.

Full size image

In 2008, the native grassland was converted to Tamarix and cropland with the aim of reclamation of the saline–alkaline soil. The Tamarix stand was left to grow naturally after plantation. For this reason, we consider it as a semi-natural ecosystem. The cropland (artificial ecosystem) has permanent plots 7.25 m × 7.25 m in size, which were left fallow after conversion until 2014. During the fallow period, the cropland plots were irrigated (180 mm per year) around early January with saline groundwater. The irrigated water freezes from January to late February or early March as air temperatures are mostly below 0 °C. The salinity of the irrigated groundwater was 7–27 g l−1 38. This practice of irrigation reduces the salinity in the soil and decrease the salt stress on subsequently planted cotton seedlings38. Since 2014, during each March, the cropland has been covered with plastic film until the sowing of the cotton to reduce the evapotranspiration38. The cropland received 400 kg N ha−1 year−1 applied during May every year (200 kg N ha−1 organic fertilizer + 200 kg N ha−1 diammonium phosphate) before sowing cotton since 2014. During this experimental period, cropland was fertilized on 7th May 2017 and 6th May 2018 and irrigation occurred on 10th Jan 2018. The irrigated water had melted completely by 21st Feb 2018. Other details of the three ecosystems are reported in Table 1.

Table 1 Management practices, dominant vegetation and some physical and chemical soil parameters of the three ecosystems.
Full size table

Gas sampling

In each ecosystem, four closed static chambers were randomly placed. The chambers were made of polyvinyl chloride (PVC) and measured 60 × 20 × 40 cm (L × B × H) and each chamber contained a fan to homogenize the air. The chambers were fitted with a thermometer and a sampling tube with a three-way stopcock. Both sampling tube and thermometer were sealed where they passed through the surface of the chamber to prevent leakage. Five 40-ml gas samples were taken for N2O concentration analysis at 20 min intervals using a glass syringe, while two 160-ml gas samples were taken at 0 and 80 min and stored in glass bottles for δ 15N-N2O analysis. Gas was sampled between 8:00 AM to 12:00 PM. Sampling was done twice to thrice in a month during March to September (warm season) while once in a month during October to February (cold season).

N2O concentration measurement, flux calculation, and 15N isotope determination

The concentration of N2O was measured using gas chromatography (Agilent GC-6820, Agilent Technologies Inc., Santa Clara, CA, USA) equipped with 63Ni electron capture detection (ECD) in the laboratory of CARR, IGDB, CAS, Shijiazhuang, Hebei. The concentrations of N2O were calculated based on the measured peak areas relative to the peak areas measured from reference standards which were run twice before and after every fifteen gas samples.

The N2O flux was calculated using the following equation from Li et al.39.

$${\text{F}} = {\text{ M }} \times {\text{ V }} \times {\text{ A}}^{{ - {1}}} \times \, \Delta {\text{C}} \times \Delta {\text{t}}^{{ - {1}}} \times { 273 } \times \, \left( {{273 } + {\text{ T}}} \right)^{{ - {1}}} \times {\text{ P }} \times \, \left( {{\text{P}}^{0} } \right)^{{ - {1}}} \times { 6}0$$
(1)

where F is the N2O flux (μg N2O-N m−2 h−1), M is the molecular weight of N2O-N, V is the volume of the chamber (m3), A is the soil surface area occupied by the chamber base (m2), ΔC × Δt−1 is the slope of N2O accumulation in the chamber with the time change (10–6 min−1), T is the air temperature (°C) inside the chamber, P is the atmospheric pressure (hPa) on the sampling time and P0 is standard atmospheric pressure.

Annual cumulative emission rate was calculated by interpolating the N2O flux from four replicate chambers during measured days and the interval between sampling days. While calculating annual emission rate, it was assumed that there was no emission of N2O from 10 Jan to 21 Feb, 2018 in the cropland because of the frozen irrigated water on the surface (up to 18 cm thickness which was shrinking when the temperature rising). This assumption might underestimate the annual cumulative emissions. However, for grassland and Tamarix the rate for the whole year was calculated.

The gas samples (160 ml) were passed through a chemical trap [NaOH + Mg(ClO4)2] (FINNIGAN PRECON) to remove CO2 and H2O using a helium flow of 10–15 ml min−1. Using stainless steel trap, the gas sample was passed through liquid nitrogen. After this cryofocusing step, the gas sample passed into a GC (FINNIGAN GC). Finally, the δ 15N of the N2O was measured using an Isotope Ratio Mass Spectrometer (IRMS) (Delta V Plus. Thermo Fisher, Germany). δ 15N of data reported in this study are in unit of per mill (‰) relative to international standard (atmospheric N2). As the N2O in the sample represented the isotopic composition of both atmospheric and soil-emitted N2O, the following equation from Snider et al.40 was used to calculate the δ 15N of soil-emitted N2O.

$$\delta^{{{15}}} {\text{N of soil - emitted N}}_{{2}} {\text{O}} = (\delta^{{{15}}} {\text{N}}_{{\text{measured x}}} {\text{C N}}_{{2}} {\text{Omeasured}} - \delta^{{{15}}} {\text{N}}_{{\text{atmosphere x}}} {\text{C}}_{{{\text{atmosphere}}}} /\left( {{\text{C N}}_{{2}} {\text{O}}_{{{\text{measured}}}} - {\text{C N}}_{{2}} {\text{O}}_{{{\text{atmosphere}}}} } \right)$$
(2)

where δ 15N measured and C N2O measured are the δ 15N and concentration of the N2O sample at time 80 min after the closure of the chamber, while the δ 15N atmosphere and C N2O atmosphere are the δ 15N and concentration of the sample at time zero (immediately after the closure of the chamber). When the fluxes were lower than 10 N2O-N µg m−2 h−1, the 15N analyses were excluded from the results due to errors introduced with lower fluxes.

Measurement of soil parameters

Soil temperature at 5 cm depth was taken using a thermometer inserted into the soil. Each day after the gas sample collection, soil samples (0–20 cm) were collected from the area nearby the chambers. Thermo-gravimetric technique (oven-drying) method was used to measure the soil moisture content. Water filled pore space (WFPS) was calculated using a formula as stated in Eq. (3):

$${\text{WFPS }}\left( \% \right) = \left( {\text{SWC x BD}} \right)/{1} - \left( {{\text{BD}}/{\text{PD}}} \right) \times {1}00\%$$
(3)

where SWC is soil water content (g g−1), BD is bulk density (Mg m−3), and PD is particle density (2.65 Mg m−3).

For soil pH and electrical conductivity (Ec), 10 g of air dried (< 2 mm) soil sample was weighed and mixed with 25 and 50 ml of deionized water, respectively. Then the mixture was mechanically shaken for 1 h. pH was determined in a suspension with a pH meter (METTLER TOLEDO FE20) at 1:2.5 soil–water ratio. Ec was measured using an Ec meter (METTLER TOLEDO SG7) with 1:5 soil–water ratio at room temperature. Soil ammonium (NH4-N) and nitrate (NO3-N) concentrations were measured using the KCl extraction method. For this, 10 g of fresh soil was mixed with 50 ml of freshly prepared 1 M KCl and the mixture was shaken for one hour, then it was filtered through Whatman 42 filter paper. Then, NH4-N and NO3-N concentrations of the filtrate were measured by using a Smartchem140 and a UV spectrophotometer, respectively.

Statistics

Data were not normally distributed for all variables. Several possible transformations were tried without success. As our main objectives were to examine differences in N2O fluxes and 15N in soil emitted N2O in different ecosystems, we conducted the Kruskal Wallis ANOVA (analysis of variance) followed by the Mann Whitney test. The same analysis was used for other measured soil parameters. Similarly, differences in annual cumulative flux between ecosystems were computed through the Kruskal Wallis ANOVA followed by the Mann Whitney test. Spearman correlation analysis was applied to examine the relationships among the measured variables and N2O flux. When p values were less than 0.05 the differences was considered significant. All the figures and statistical analyses were computed in Origin Pro 8 (Origin Lab Ltd., Guangzhou, China).

Results

Soil environmental variables

The pattern of soil temperature was consistent with the air temperature (Fig. 2a,b). Soil temperature at 5 cm soil depth showed a clear and similar seasonal variation (high in summer and low in winter) in all ecosystems. The lowest temperature was − 4 °C reported in January while the highest temperature was 42 °C in July. Soil temperature at 5 cm depth at grassland was similar to the cropland and Tamarix. While the Tamarix had significantly (p < 0.05) lower soil temperatures than the cropland. The median soil temperature was 24.5 °C, 25.3 °C and 23.5 °C in the grassland, cropland and Tamarix, respectively.

Figure 2
figure 2

Daily average air temperature and precipitation at the study site during the study period (a) and soil temperature at 5 cm depth taken at the time of gas sampling (b). Error bars represent mean ± standard error (SE) (n = 4).

Full size image

The overall WFPS of the Tamarix was significantly less (p < 0.001) than the grassland and cropland. The median value of WFPS in the grassland was 89.6% (ranging from 66.9 to 99.95%), cropland was 90.4% (ranging from 73.32 to 99.97%) and Tamarix was 76.2% (ranging from 44.4 to 97.0%). As water table was around 0.9–1.5 m, normally WFPS exceeded 70% in all ecosystems (Fig. 3a).

Figure 3
figure 3

Soil water-filled pore space (WFPS) (a), Soil NH4 (b), and NO3 (c) of the top 20 cm soil. The arrows represent fertilizer application event. Each point represents athematic mean of n = 1–4 ± SE.

Full size image

Soil NH4 was significantly (p < 0.01) higher in the grassland compared to the cropland and Tamarix. Overall, median NH4 concentration in the grassland was 0.55 mg kg−1 (ranging from 0.006 to 4.0 mg kg−1), 0.35 mg kg−1 (ranging 0.006–6.4 mg kg−1) in the cropland and 0.31 mg kg−1 (ranging from 0.01 to 1.2 mg kg−1) in the Tamarix. Grassland and cropland showed higher temporal variation in soil NH4 than the Tamarix during the sampling period (Fig. 3b). After fertilization of the cropland, there was a peak in NH4 content.

Soil NO3 was significantly different (p < 0.001) among all three ecosystems. The order of soil NO3 was: cropland > Tamarix > grassland. The median concentration of NO3 in the grassland was 1.0 mg kg−1 (ranging 0.004–14.0 mg kg−1), 65 mg kg−1 (6.4–209 mg kg−1) in the cropland and 12.3 mg kg−1 (ranging from 2.6 to 34.30 mg kg−1) in the Tamarix. For some sampling dates, NO3 was below the limit of detection in the grassland soil. Fertilizer application in the cropland led to a peak in NO3 content in the soil (Fig. 3c).

N2O fluxes and annual cumulative emission

Among the 24 sampling occasions, 9 occasions were found negative fluxes in the grassland, but in the cropland and Tamarix there were always positive fluxes (Fig. 4). Overall, N2O fluxes were significantly different (p < 0.001) among the ecosystems. The median N2O flux was 4.0 N2O-N µg m−2 h−1 (ranging from − 22.0 to − 1.1 for negative flux and 2.8 to 117.7 4 N2O-N µg m−2 h−1 for the positive flux, over the study period), 25.3 N2O-N µg m−2 h−1 (ranging from 2.0 to 678.04 N2O-N µg m−2 h−1) and 8.2 N2O-N µg m−2 h−1 (ranging from 0.5 to 179.0 N2O-N µg m−2 h−1) from the grassland, cropland and Tamarix, respectively. The peak fluxes in the cropland occurred after the application of fertilizer (Fig. 4). In 2017, after fertilization the N2O peak lasted for two weeks. While in 2018, on the day of fertilization there was a small increase, then the highest peak occurred in the 4th week after fertilization. Results for February 2018 and the 3rd week after fertilization in 2018 are not reported because it was noted that there were unusually high concentrations of N2O (4 times higher than usual atmospheric concentration) in all samples taken at time zero, which may have led to errors in interpretation of results. For two of the sampling points, high N2O emissions from Tamarix were observed. This occurred during the decomposition of a large number of pill-bugs that had died at the site (the reason for the pill-bug deaths is unknown).

Figure 4
figure 4

N2O flux from grassland, cropland, and Tamarix. Each point represents the arithmetic mean and standard error of four replicates. Arrows pointing upward indicate fertilization events on the cropland, while those pointing downward indicate the presence of dead pill-bugs in the Tamarix or irrigation and covered by the plastic film in the cropland (left to right). Red color arrows represent specific events in cropland and blue for Tamarix.

Full size image

The annual cumulative N2O emissions were significantly different (p < 0.05) among all three ecosystems. The annual cumulative N2O emissions increased in the order of cropland > Tamarix > grassland. Cropland emitted 3.5 kg N2O-N ha−1 year−1 (ranging from 2.7 to 3.9 kg N2O-N ha−1 year−1) about 1.7 times more than the Tamarix, which emitted 1.3 kg N2O-N ha−1 year−1 (ranging from 0.9 to 1.6 kg N2O-N ha−1 year−1), and 7 times more than the grassland (0.5 kg N2O-N ha−1 year−1, ranging from 0.3 to 0.7 kg N2O-N ha−1 year−1) (Fig. 5).

Figure 5
figure 5

Box plot for annual N2O emissions (n = 4). Different letters indicate significant difference (p < 0.05) and square represents mean values.

Full size image

Relationship between soil environmental variables and N2O flux

Spearman correlation analysis showed various relationships between N2O flux and soil environmental variables measured at three studied ecosystems (Table 2). In grassland, there was no significant relationship between N2O flux and any of the measured soil parameters. In the cropland, the analysis showed significant positive correlations of N2O flux with soil temperature, NH4 content, and NO3 content. There was no significant correlation between N2O emission and WFPS in the cropland. Analysis of the Tamarix results showed that there were significant positive correlations of N2O flux with soil temperature and NO3 content, while there was a negative relationship with WFPS.

Table 2 Spearman correlation analysis between soil environmental variables and N2O flux in different ecosystems.
Full size table

15N isotopic signature of soil-emitted N2O

There was a significant difference (p < 0.01) in the 15N isotopic signature of soil-emitted N2O between the three ecosystems (Fig. 6). The difference between grassland and Tamarix was at the level of p < 0.01 while between grassland and cropland was at the level of p < 0.001, suggesting N addition has strong effect on depletion of 15N in N2O. N2O emitted from cropland was more depleted in 15N while N2O emitted from grassland was less depleted. The median 15N values in emitted N2O were − 0.18 ‰ (ranging from − 41.0 to 5.8‰, n = 14), − 25.3 ‰ (ranging from − 68.3 to 4.6 ‰, n = 63) and − 13.7 ‰ (ranging from − 50.5 to 3.0‰, n = 32) for the grassland, cropland and Tamarix, respectively. Due to problems with the IRMS, results from the beginning of the experiment are not included. In the grassland, due to low and negative fluxes of N2O, it was not always possible to calculate 15N values in soil-emitted N2O. Emitted N2O was more depleted in 15N in April in the grassland while in the Tamarix it was during the pill-bug decomposition period. In the cropland, it was just after the application of N fertilizer and this continued for about three weeks after the fertilization, then in the fourth week, when N2O emission reached its highest peak, the values returned to the normal range (Figs. 5, 6). There was no significant relationship between measured parameters and 15N in soil-emitted N2O.

Figure 6
figure 6

15N isotopic signature of soil-emitted N2O from studied ecosystems. Each point represents arithmetic mean of 1–4 replicates with standard errors. Arrow represents fertilizer application event in the cropland.

Full size image

Discussion

At our experimental site, we had a unique opportunity to investigate the impact of land-use change (LUC) from natural to semi-natural and artificial ecosystems on N2O flux and its 15N within the same climatic conditions and soil type. LUC is associated with changes in various land cover types as a result of different management practices, which then can lead to changes in soil physical, chemical41 and biological properties20. The changes in these soil properties can alter soil greenhouse gas emissions16,19. Soil humidity, temperature, NH4 content and NO3 content are the major soil parameters that influence N2O emission from soil21,36,42. With the change in the land use, it was observed that these soil parameters were significantly influenced at our study site, which may have led to the differences in N2O flux from the different ecosystems.

In the grassland, no studied soil parameters were significantly correlated to N2O flux, which may have been due to limited NO3 content. The relatively high NH4 content and low NO3 in grassland soil indicates inhibition of nitrification process, causing low N2O emissions. The positive correlation between soil temperature and N2O emission in the cropland and Tamarix, observed in our study is consistent with other studies36,43 and can be explained by the increase in microbial activity with an increase in temperature44. WFPS higher than 80% is favorable for N2O reduction to N222. Low N content along with higher WFPS and frequent N2O uptake results reported in the grassland site indicate that denitrification is a dominant process of N2O emission. Optimum WFPS for N2O emissions ranges from 60 to 80%22, and there have been reports of significant positive to negative or no relationship between WFPS and N2O emission45,46,47. Increase in soil moisture has a greater effect when dry soil is wetted48. So, higher WFPS (around 90%) in grassland and cropland might not be limiting factor controlling N2O emissions in our study. We only observed significant relationship between WFPS and N2O flux in Tamarix. The negative relationship might be due to excessive WFPS than that is required for optimum N2O production49. NH4 and NO3 are the main substrates for nitrification and denitrification50,51. Significant positive relationships between N2O emission and both NH4 and NO3 have previously been demonstrated42 indicating that coupled nitrification–denitrification contributes to N2O formation in the soil50. Similarly, in the current study positive relationships were found between N2O flux and NH4 and NO3 content in the cropland; however, only with NO3 in the Tamarix. It can be difficult to identify the N2O formation process responsible or the emissions i.e. either nitrification or denitrification, as both processes can occur simultaneously in the soil50. The results showing a range of both positive and negative relationships between various soil environmental parameters and N2O flux indicate that N2O formation processes have complex interactions with these soil parameters.

Often ecosystems with low N content have a negative flux and low annual N2O emission. The grassland site in our study was like most natural ecosystems21,53, N limited with low atmospheric nitrogen input and densely rooted vegetation and therefore emitted less N2O54. High WFPS with low N content favors denitrification leading to N2O consumption53,55. However, relatively dry ecosystems have also been reported to consume atmospheric N2O56,57,58; however, the possible mechanisms of N2O consumption by soil under dry conditions are not well understood59. N2O uptake has been observed at low NO3 levels (~ 1 mg N kg−1) and NH4 content (< 2 mg N kg−1) levels and high WFPS (90%)60,61. The grassland conditions in the current study were similar to these previous findings that may be the reason for N2O uptake occurring in the grassland in some sampling occasions. It has also been observed that soil under different plant species can have different rates of N2O reduction62 and that N2O consumption rate decreases with increase in soil NO363. In the cropland and Tamarix systems in the present study, NO3 content was significantly higher than in the grassland, which might have resulted in a decrease in the reduction of N2O to N2, leading to the higher emission of N2O. The more depleted 15N values in soil-emitted N2O in the cropland and Tamarix compared to the grassland (Fig. 6) is further evidence of a decrease in the reduction of N2O to N2 in those systems29,30.

Overall, N2O flux in the grassland was low (4.0 N2O-N µg m−2 h−1) with an annual cumulative emission of 0.5 kg N2O-N ha−1 year−1. These findings are similar to those observed in other studies on natural grassland under different climatic conditions on non-saline soils54,59,64,65,66. However, compared to a saline grassland with the same dominant vegetation36 the flux rate in the current study was low. This was possibly due to the low NO3 and NH4 content. When natural grasslands with low N content are converted to cropland, the addition of a large amount of N fertilizer may potentially contribute to high N2O emissions65. Consistent with this, the cropland in the current study emitted about 7 times more N2O than the grassland. The annual N2O emission rate was similar to the IPCC default emission factor, i.e. 1% of applied N fertilizer is emitted as N2O in the agricultural fields67. The observed N2O emission from our cropland was lower than that from non-saline–alkaline soils in the same climatic area under application of the same amount of fertilizer68. Similarly, the N2O flux from some non-saline–alkaline soils, receiving a similar rate of fertilizer, was three times higher than from the cropland in our study43. A saline–alkaline sunflower field, receiving 300 kg N ha−1 year−1, emitted 9.8 kg N ha−1 year−110, which is 3.8 times higher than the emission rate from the cropland in the current study, which had 400 kg N ha−1 year−1 applied. The Tamarix ecosystem emitted 2.6 times more N2O than the native grassland. This increase can be attributed to the higher NO3 content. The increase in NO3 content could also be linked to a lower reduction of N2O to N2 in the Tamarix system because high NO3 inhibits N2O reduction69. Conversion of grassland to tree plantations has a contrasting (increased to no influence) effect on N2O emission17,18. Overall, our results support our hypothesis that conversion of native grassland to cropland or Tamarix ecosystems would lead to change in soil environmental variables and an increase in N2O emission.

When compared with studies involving similar land use or land-use change (Supplement Information S1) our results from the respective ecosystems are within the ranges reported in the literature. This result suggests that saline–alkaline soils may not always have a higher potential for N2O emission, as hypothesized by Ghosh et al.70 and Yang et al.10. For the cropland the fertilizer application rate was higher than other studies in the literature (Supplement Information S1), this is likely to have led to the higher rate of N2O emission from the cropland. In saline–alkaline soil, NH4 can be converted to NH3 and lost to the atmosphere, which may decrease the probability of N2O formation due to nitrification13. Two meta-analyses11,12 reported that alkaline soils emit less N2O compared to natural and acidic soils. Furthermore, high salinity inhibits both nitrification and denitrification processes8,9. These negative effects of both salinity and alkalinity on N2O production processes and emissions further suggest that saline–alkaline soil may not emit more N2O.

It is evident from previous research25,28 that there may be differences in the 15N in soil-emitted N2O between fertilized and unfertilized ecosystems. Therefore, significant differences were expected in the 15N isotopic signatures in soil-emitted N2O between the unfertilized ecosystems (grassland and Tamarix) and the fertilized cropland. As there was no anthropogenic N input in grassland and Tamarix, our expectation was 15N in N2O would be similar in these two ecosystems. However, differences were observed among all three ecosystems. The 15N in N2O emitted from the grassland, cropland, and Tamarix were all within the range reported by other studies25,26,28,62,71. As we can see from Fig. 6 that temporal variability of 15N in soil-emitted N2O was highest in cropland, indicating that N cycling process in the cropland is relatively open. The more depleted 15N in N2O emitted from the cropland implies that N availability can be considered enhanced (due to the high rate of N fertilizer) in the ecosystem25. When nitrogen availability is enhanced, the N2O production process favors larger 15N fractionation, leading to more depleted 15N in N2O from the soil25,72. This phenomenon can lead to difference in the 15N in N2O emitted from the cropland compared to the grassland and Tamarix, as observed in this study. After application of fertilizer the cropland could be considered to have unlimited N availability so the N2O emitted was strongly depleted in 15N, indicating the production of N2O, either by nitrification or denitrification, favored larger 15N fractionation rather than shift from denitrification to nitrification25,28,71,72. Although 15N values in soil-emitted N2O can sometimes be used to predict sources of N2O when combined measurements of 15N values in substrates for N2O production28 and molecular analysis of N2O producing organisms40, with data from this trial was not possible to estimate relative contributions of nitrification and denitrification. Moreover, more powerful tools like 15N site preference (SP) is a good indicator of production pathways24,73,74, which was not used in this study, making difficult to generalize dominant process of N2O production in different ecosystems.

Contrarily to our hypothesis, there was a difference between 15N in soil-emitted N2O within unfertilized (grassland and tamarix) ecosystems. The reason for differences in the 15N in N2O between the grassland and Tamarix may be a difference in N2O reduction capability. It is likely that N2O reduction in the grassland (as evidenced by N2O consumption) enriched the 15N in N2O, so when it was emitted to the atmosphere it was less depleted than N2O emitted from soil in which reduction has not occured29,75. A possible reason for the reduction of N2O being favored in the grassland soil may be the low concentrations of NO369 and high WFPS22. For this reason reduction of N2O to N2 might be more prominent in the grassland compared to the Tamarix. However, it could be a possibility that gross N2O consumption may be masked by higher rates of N2O production76 in the cropland and Tamarix. The 15N isotope content of the substrates (NH4 and NO3) for N2O production were not measured in the current study, which could have provided more insight into the reason for the observed differences between the ecosystems. The 15N differences in the emitted N2O between ecosystems could also be due to variation in the microbial community composition in the soils77. Several factors favor complete denitrification, such as differences in microbial community composition (denitrifiers), presence of denitrification enzymes, high soil water content, high soil pH, a low rate of O2 diffusion and presence of labile carbon55. So differences in those factors should not be ruled out as causes for the differences in the 15N content in emitted N2O between the ecosystems.

The 15N content in atmospheric N2O has been decreasing since the preindustrial age78; however, atmospheric N2O concentration is increasing5. This decrease in the 15N in N2O has been considered to be a result of an increase in the use of chemical fertilizer5,28. Moreover, global decline in the N2O reduction process relative to production might also contribute to the decrease in the 15N29. Our results indicate that the conversion of natural ecosystems to cropland with the addition of anthropogenic N would greatly contribute to the depletion of the 15N in atmospheric N2O by emitting more depleted 15N in N2O along with higher N2O emission rate, which was according to our hypothesis. Moreover, if ecosystems with more reduction capability (such as grassland) are converted to Tamarix that have less reduction capability (assumed due to the absence of measured atmospheric N2O consumption in our study), this would also play a role in the depletion of 15N in atmospheric N2O. Overall, it can be concluded that the addition of anthropogenic N to cropland would contribute more to deplete 15N in atmospheric N2O than any other processes.

Conclusions

Our study showed that LUC from native grassland to Tamarix and cropland on saline–alkaline soil significantly influence soil temperature, soil moisture and NH4 and NO3 contents. The changes in these soil factors, along with the observed correlations between N2O fluxes and the soil parameters, could explain the differences in N2O flux caused by the LUC. Saline–alkaline soil may not always act as a potentially high source of N2O, as our fluxes and annual emissions result are in the usual ranges for the respective ecosystems reported in the literature. The conversion from native grassland to Tamarix ecosystem increased more N2O 2.6 times while cropland increased 7 times. The LUC also influenced the 15N in soil-emitted N2O, greatly depleting it in cropland and moderate in Tamarix compared to native grassland. The differences in the 15N in soil-emitted N2O between the fertilized and unfertilized ecosystems could be attributable to anthropogenic N fertilization. The differences in the 15N in N2O between the unfertilized ecosystems (grassland and Tamarix) could be attributable to the N2O reduction capacity of native grassland. Our results further suggest that the depletion of the 15N in atmospheric N2O since the pre-industrial age could be highly attributable to anthropogenic N addition and to lesser extent to land-use changes where ecosystems with more N2O reduction capacity have been converted to ecosystems with less N2O reduction capacity.