Nitrogen isotopic signatures and fluxes of N2O in response to land-use change on naturally occurring saline–alkaline soil

The conversion of natural grassland to semi-natural or artificial ecosystems is a large-scale land-use change (LUC) commonly occurring to saline–alkaline land. Conversion of natural to artificial ecosystems, with addition of anthropogenic nitrogen (N) fertilizer, influences N availability in the soil that may result in higher N2O emission along with depletion of 15N, while converting from natural to semi-natural the influence may be small. So, this study assesses the impact of LUC on N2O emission and 15N in N2O emitted from naturally occurring saline–alkaline soil when changing from natural grassland (Phragmites australis) to semi-natural [Tamarix chinensis (Tamarix)] and to cropland (Gossypium spp.). The grassland and Tamarix ecosystems were not subject to any management practice, while the cropland received fertilizer and irrigation. Overall, median N2O flux was significantly different among the ecosystems with the highest from the cropland (25.3 N2O-N µg m−2 h−1), intermediate (8.2 N2O-N µg m−2 h−1) from the Tamarix and the lowest (4.0 N2O-N µg m−2 h−1) from the grassland ecosystem. The 15N isotopic signatures in N2O emitted from the soil were also significantly affected by the LUC with more depleted from cropland (− 25.3 ‰) and less depleted from grassland (− 0.18 ‰). Our results suggested that the conversion of native saline–alkaline grassland with low N to Tamarix or cropland is likely to result in increased soil N2O emission and also contributes significantly to the depletion of the 15N in atmospheric N2O, and the contribution of anthropogenic N addition was found more significant than any other processes.

Nitrous oxide (N 2 O) is a major long-lived anthropogenic greenhouse gas with about 265-298 fold greater potential for global warming in the atmosphere compared to carbon dioxide 1 . It is also an ozone-depleting substance 2 , produced mainly in the soil from nitrification and denitrification processes 3 . Its concentration in the atmosphere has increased to 331 ppb 4 from 270 ppb in the pre-industrial age 5 . This increase of N 2 O in the atmosphere is mainly attributable to rise in anthropogenic nitrogen (N) input to soil 6,7 and this anthropogenic N input to soil increases as more natural ecosystems are converted to croplands. Soil salinity can influence N 2 O flux in different ways. An increase in salinity in a non-saline soil can increase 8 or have no effect on N 2 O emission 9 . Similarly, on naturally occurring saline soils, both decreases 8 and increases 10 in the N 2 O flux have been found in response to increase in the salinity. These results suggest an ambiguous role of salinity in N 2 O emission. Some meta-analyses 11,12 have reported that alkaline soil emits less N 2 O than neutral or acidic soil. In alkaline soil NH 4 may be converted to NH 3 and volatilize to the atmosphere whereas NH 4 is retained in acid soil, favoring N 2 O formation 13 . N loss from alkaline soil may be high in total, but if much of the N is lost in the form of NH 3 there may be less NH 4 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 N 2 O formation processes. So, quantifying N 2 O flux from the saline-alkaline soil may help to increase knowledge on its contribution to soil-atmosphere exchange of N 2 O.

Scientific Reports
| (2020) 10:21253 | https://doi.org/10.1038/s41598-020-78149-w www.nature.com/scientificreports/ Land-use change (LUC) from natural to semi-natural or artificial ecosystems can have different effects on N 2 O emission [14][15][16] . Specifically, conversion from natural to artificial ecosystems with the addition of N fertilizer significantly increases N 2 O emission while conversion to semi-natural may or may not increase the emission [16][17][18] . LUC directly impacts on soil physical, chemical and biological properties 19,20 , the main factors affecting N 2 O emission 21,22 . N 2 O emission from soil is reduced when pasture is forested 14 , while conversion of rainforest to pasture or plantation leads to an increase in N 2 O emission 15 . A recent study found that the conversion of a conventional agricultural field to bio-energy crops had no effect on N 2 O emission 23 . Therefore, knowing which LUC practice is appropriate in terms of lower N 2 O emissions, and its implementation could mitigate N 2 O emission to the atmosphere and associated impact of climate change. Moreover, various LUC practices 14,15,23 have different or no effect on N 2 O emission, indicating that LUC is rather an indirect cause of N 2 O emission. The main reason for the differences in N 2 O emission due to LUC is probably the alteration of the controlling factors of N 2 O production and reduction processes in the soil. So, quantifying N 2 O flux from LUC, along with soil physical and chemical parameters, would further enable understanding of the main driving factors for N 2 O production and consumption in the soil. N has two stable isotopes i.e., 14 N and 15 N. δ 15 N of a sample is the deviation of the samples' 15 N/ 14 N from the respective isotope ratio of the reference material 24 . Previously, the 15 N in N 2 O emitted from soil has been used to identify the processes for N 2 O production i.e. nitrification and denitrification; however, using only 15 N values in N 2 O may mislead the interpretation 7 as both the processes generally occur in the soils, possibly in different horizons or niche. The 15 N in N 2 O emitted from soil depends on the 15 N content of the substrates i.e. NH 4 and NO 3 , different microbial community composition, pH, temperature and substrate availability 24,25 . Although it is difficult to predict the sources of N 2 O emission using solely 15 [25][26][27] ; while more depleted after N application i.e., − 37.9‰ 28 to − 34.3‰ 25 in fertilized soil. The difference of 15 N in N 2 O is useful to distinguish N 2 O emitted between fertilized and natural soils, and it arises from anthropogenic N addition to soil 25,28 . Moreover, application of N fertilizer leads to high concentrations of NO 3 in the soil, resulting in a decrease in N 2 O reduction to N 2 and therefore a higher N 2 O to N 2 ratio from the denitrification process 29 . The reduction of N 2 O to N 2 through denitrification leads to 1-24‰ 15 N enrichment of the remaining N 2 O 30 . So, differences in the capacity to reduce N 2 O to N 2 between various ecosystems may also influence the 15 To feed the world's growing population requires an additional 2.7-4.9 Mha of cropland per year on average 31 . Due to limited land resources, natural saline-alkaline areas are being reclaimed for producing food 32 . Agricultural soil alone will contribute about 59% of total global N 2 O emissions by 2030 3 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 N 2 O 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 Tamarix 35 ; however, its effect on N 2 O emission is unknown. Though Zhang et al. 36 reported the differences in N 2 O 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 N 2 O 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 N 2 O fluxes 37 . Therefore, we hypothesize that: (1) LUC from native saline-alkaline ecosystem (grassland) to semi-natural (Tamarix) may significantly influence N 2 O 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 15 N in N 2 O 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 N 2 O flux and the 15 N in N 2 O due to anthropogenic N addition and changes in management practices. Therefore, we expect that the 15 N in emitted N 2 O could be used to distinguish N 2 O 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 soil 38 .
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 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 N 2 O 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 15 N-N 2 O 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). 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 N 2 O flux was calculated using the following equation from Li et al. 39 .
, A is the soil surface area occupied by the chamber base (m 2 ), ΔC × Δt −1 is the slope of N 2 O 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 P 0 is standard atmospheric pressure. Annual cumulative emission rate was calculated by interpolating the N 2 O 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 N 2 O 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(ClO 4 ) 2 ] (FINNIGAN PRECON) to remove CO 2 and H 2 O 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 15 N of the N 2 O was measured using an Isotope Ratio Mass Spectrometer (IRMS) (Delta V Plus. Thermo Fisher, Germany). δ 15 N of data reported in this study are in unit of per mill (‰) relative to international standard (atmospheric N 2 ). As the N 2 O in the sample represented the isotopic composition of both atmospheric and soil-emitted N 2 O, the following equation from Snider et al. 40  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): 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 (NH 4 -N) and nitrate (NO 3 -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, NH 4 -N and NO 3 -N concentrations of the filtrate were measured by using a Smartchem140 and a UV spectrophotometer, respectively.

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. 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).
Soil NH 4 was significantly (p < 0.01) higher in the grassland compared to the cropland and Tamarix. Overall, median NH 4 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. www.nature.com/scientificreports/ Grassland and cropland showed higher temporal variation in soil NH 4 than the Tamarix during the sampling period (Fig. 3b). After fertilization of the cropland, there was a peak in NH 4 content. Soil NO 3 was significantly different (p < 0.001) among all three ecosystems. The order of soil NO 3 was: cropland > Tamarix > grassland. The median concentration of NO 3 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

Relationship between soil environmental variables and N 2 O flux. Spearman correlation analysis
showed various relationships between N 2 O flux and soil environmental variables measured at three studied ecosystems (Table 2). In grassland, there was no significant relationship between N 2 O flux and any of the measured soil parameters. In the cropland, the analysis showed significant positive correlations of N 2 O flux with soil temperature, NH 4 content, and NO 3 content. There was no significant correlation between N 2 O emission and WFPS in the cropland. Analysis of the Tamarix results showed that there were significant positive correlations of N 2 O flux with soil temperature and NO 3 content, while there was a negative relationship with WFPS.

N isotopic signature of soil-emitted N 2 O.
There was a significant difference (p < 0.01) in the 15 N isotopic signature of soil-emitted N 2 O 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 15 15 N 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 N 2 O emission reached its highest peak, the values returned to the normal range (Figs. 5, 6). There was no significant relationship between measured parameters and 15 N in soilemitted N 2 O.

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 N 2 O flux and its 15 N within the same climatic conditions and soil type. LUC is associated with changes in various land cover types as a result of different management www.nature.com/scientificreports/ practices, which then can lead to changes in soil physical, chemical 41 and biological properties 20 . The changes in these soil properties can alter soil greenhouse gas emissions 16,19 . Soil humidity, temperature, NH 4 content and NO 3 content are the major soil parameters that influence N 2 O emission from soil 21,36,42 . 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 N 2 O flux from the different ecosystems.
In the grassland, no studied soil parameters were significantly correlated to N 2 O flux, which may have been due to limited NO 3 content. The relatively high NH 4 content and low NO 3 in grassland soil indicates inhibition of nitrification process, causing low N 2 O emissions. The positive correlation between soil temperature and N 2 O emission in the cropland and Tamarix, observed in our study is consistent with other studies 36,43 and can be explained by the increase in microbial activity with an increase in temperature 44 . WFPS higher than 80% is favorable for N 2 O reduction to N 2 22 . Low N content along with higher WFPS and frequent N 2 O uptake results reported in the grassland site indicate that denitrification is a dominant process of N 2 O emission. Optimum WFPS for N 2 O emissions ranges from 60 to 80% 22 , and there have been reports of significant positive to negative or no relationship between WFPS and N 2 O emission [45][46][47] . Increase in soil moisture has a greater effect when dry soil is wetted 48 . So, higher WFPS (around 90%) in grassland and cropland might not be limiting factor controlling N 2 O emissions in our study. We only observed significant relationship between WFPS and N 2 O flux in Tamarix. The negative relationship might be due to excessive WFPS than that is required for optimum N 2 O production 49 . NH 4 and NO 3 are the main substrates for nitrification and denitrification 50,51 . Significant positive relationships between N 2 O emission and both NH 4 and NO 3 have previously been demonstrated 42 indicating that coupled nitrification-denitrification contributes to N 2 O formation in the soil 50 . Similarly, in the current study positive relationships were found between N 2 O flux and NH 4 and NO 3 content in the cropland; however, only with NO 3 in the Tamarix. It can be difficult to identify the N 2 O formation process responsible or the emissions i.e. either nitrification or denitrification, as both processes can occur simultaneously in the soil 50 . The results showing a range of both positive and negative relationships between various soil environmental parameters and N 2 O flux indicate that N 2 O formation processes have complex interactions with these soil parameters.
Often ecosystems with low N content have a negative flux and low annual N 2 O emission. The grassland site in our study was like most natural ecosystems 21,53 , N limited with low atmospheric nitrogen input and densely rooted vegetation and therefore emitted less N 2 O 54 . High WFPS with low N content favors denitrification leading to N 2 O consumption 53,55 . However, relatively dry ecosystems have also been reported to consume atmospheric N 2 O 56-58 ; however, the possible mechanisms of N 2 O consumption by soil under dry conditions are not well understood 59 . N 2 O uptake has been observed at low NO 3 levels (~ 1 mg N kg −1 ) and NH 4 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 N 2 O 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 N 2 O reduction 62 and that N 2 O consumption rate decreases with increase in soil NO 3 63 . In the cropland and Tamarix systems in the present study, NO 3 content was significantly higher than in the grassland, which might have resulted in a decrease in the reduction of N 2 O to N 2 , leading to the higher emission of N 2 O. The more depleted 15 N values in soil-emitted N 2 O in the cropland and Tamarix compared to the grassland (Fig. 6) is further evidence of a decrease in the reduction of N 2 O to N 2 in those systems 29,30 . Overall, N 2 O flux in the grassland was low (4.0 N 2 O-N µg m −2 h −1 ) with an annual cumulative emission of 0.5 kg N 2 O-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 soils 54,59,[64][65][66] . However, compared to a saline grassland with the same dominant vegetation 36 the flux rate in the current study was low. This was possibly due to the low NO 3 and NH 4 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 N 2 O emissions 65 . Consistent with this, the cropland in the current study emitted about 7 times more N 2 O than the grassland. The annual N 2 O emission rate was similar to the IPCC default emission factor, i.e. 1% of applied N fertilizer is emitted as N 2 O in the agricultural fields 67 . The observed N 2 O 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 fertilizer 68 . Similarly, the N 2 O flux from some nonsaline-alkaline soils, receiving a similar rate of fertilizer, was three times higher than from the cropland in our study 43 . 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 N 2 O than the native grassland. This increase can be attributed to the higher NO 3 content. The increase in NO 3 content could also be linked to a lower reduction of N 2 O to N 2 in the Tamarix system because high NO 3 inhibits N 2 O reduction 69 . Conversion of grassland to tree plantations has a contrasting (increased to no influence) effect on N 2 O emission 17,18 . 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 N 2 O 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 N 2 O 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 N 2 O emission from the cropland. In saline-alkaline soil, NH 4 can be converted to NH 3 and lost to the atmosphere, which may decrease the probability of N 2 O formation due to nitrification 13 . Two meta-analyses 11,12 reported that alkaline soils emit less N 2 O compared to natural and acidic soils. Furthermore, high salinity inhibits both nitrification and denitrification processes 8,9 . These negative effects of both salinity and alkalinity on N 2 O production processes and emissions further suggest that saline-alkaline soil may not emit more N 2 O.  28 that there may be differences in the 15 N in soil-emitted N 2 O between fertilized and unfertilized ecosystems. Therefore, significant differences were expected in the 15 N isotopic signatures in soil-emitted N 2 O 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 15 N in N 2 O would be similar in these two ecosystems. However, differences were observed among all three ecosystems. The 15 N in N 2 O emitted from the grassland, cropland, and Tamarix were all within the range reported by other studies 25,26,28,62,71 . As we can see from Fig. 6 that temporal variability of 15 N in soil-emitted N 2 O was highest in cropland, indicating that N cycling process in the cropland is relatively open. The more depleted 15 N in N 2 O emitted from the cropland implies that N availability can be considered enhanced (due to the high rate of N fertilizer) in the ecosystem 25 . When nitrogen availability is enhanced, the N 2 O production process favors larger 15 N fractionation, leading to more depleted 15 N in N 2 O from the soil 25,72 . This phenomenon can lead to difference in the 15 N in N 2 O 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 N 2 O emitted was strongly depleted in 15 N, indicating the production of N 2 O, either by nitrification or denitrification, favored larger 15 N fractionation rather than shift from denitrification to nitrification 25,28,71,72 . Although 15 N values in soil-emitted N 2 O can sometimes be used to predict sources of N 2 O when combined measurements of 15 N values in substrates for N 2 O production 28 and molecular analysis of N 2 O producing organisms 40 , with data from this trial was not possible to estimate relative contributions of nitrification and denitrification. Moreover, more powerful tools like 15 N site preference (SP) is a good indicator of production pathways 24,73,74 , which was not used in this study, making difficult to generalize dominant process of N 2 O production in different ecosystems.
Contrarily to our hypothesis, there was a difference between 15 N in soil-emitted N 2 O within unfertilized (grassland and tamarix) ecosystems. The reason for differences in the 15 N in N 2 O between the grassland and Tamarix may be a difference in N 2 O reduction capability. It is likely that N 2 O reduction in the grassland (as evidenced by N 2 O consumption) enriched the 15 N in N 2 O, so when it was emitted to the atmosphere it was less depleted than N 2 O emitted from soil in which reduction has not occured 29,75 . A possible reason for the reduction of N 2 O being favored in the grassland soil may be the low concentrations of NO 3 69 and high WFPS 22 . For this reason reduction of N 2 O to N 2 might be more prominent in the grassland compared to the Tamarix. However, it could be a possibility that gross N 2 O consumption may be masked by higher rates of N 2 O production 76 in the cropland and Tamarix. The 15 N isotope content of the substrates (NH 4 and NO 3 ) for N 2 O 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 15 N differences in the emitted N 2 O between ecosystems could also be due to variation in the microbial community composition in the soils 77 . 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 O 2 diffusion and presence of labile carbon 55 . So differences in those factors should not be ruled out as causes for the differences in the 15 N content in emitted N 2 O between the ecosystems. The 15 N content in atmospheric N 2 O has been decreasing since the preindustrial age 78 ; however, atmospheric N 2 O concentration is increasing 5 . This decrease in the 15 N in N 2 O has been considered to be a result of an increase in the use of chemical fertilizer 5,28 . Moreover, global decline in the N 2 O reduction process relative to production might also contribute to the decrease in the 15 N 29 . 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 15 N in atmospheric N 2 O by emitting more depleted 15 N in N 2 O along with higher N 2 O 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 N 2 O consumption in our study), this would also play a role in the depletion of 15 N in atmospheric N 2 O. Overall, it can be concluded that the addition of anthropogenic N to cropland would contribute more to deplete 15 N in atmospheric N 2 O 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 NH 4 and NO 3 contents. The changes in these soil factors, along with the observed correlations between N 2 O fluxes and the soil parameters, could explain the differences in N 2 O flux caused by the LUC. Saline-alkaline soil may not always act as a potentially high source of N 2 O, 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 N 2 O 2.6 times while cropland increased 7 times. The LUC also influenced the 15 N in soil-emitted N 2 O, greatly depleting it in cropland and moderate in Tamarix compared to native grassland. The differences in the 15 N in soil-emitted N 2 O between the fertilized and unfertilized ecosystems could be attributable to anthropogenic N fertilization. The differences in the 15 N in N 2 O between the unfertilized ecosystems (grassland and Tamarix) could be attributable to the N 2 O reduction capacity of native grassland. Our results further suggest that the depletion of the 15 N in atmospheric N 2 O 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 N 2 O reduction capacity have been converted to ecosystems with less N 2 O reduction capacity.

Data availability
The datasets of the current study will be available from the corresponding author on reasonable request.