Crop Productivity and Nitrogen Balance as Influenced by Nitrogen Deposition and Fertilizer Application in North China

In spite of the importance of N management in agricultural production, closing the full nitrogen balance remains a challenge, mainly due to the uncertainties in both fluxes of nitrogen input and output. We analyzed N deposition and its influence on crop productivity and field nitrogen balance based on data from three of 15 years (1990–2005) of experiments in North China. The results showed that the average annual nitrogen deposition was 76, 80, and 94 kg N/ha at Changping, Zhengzhou, and Yangling in a wheat-maize rotation system, respectively. The deposited N could support a corresponding total biomass production (wheat plus maize) of 9.6, 10.6, and 8.8 Mg/ha with a total grain yield of 3.8, 4.8, and 3.7 Mg/ha, however, that did not cause a further decline in soil organic matter. N fertilizer application could increase total biomass (grain) by 244% (259%) and 74% (119%) for wheat and maize, respectively. Under optimal N management, N deposition accounted for 17–21% of the total N inputs, which affected significantly the recovery efficiency of applied N. N deposition showed a significant spatial variation in terms of the fraction of dry and wet depositions. On an annual average, N deposition roughly balanced out N losses due to NH3 volatilization and N2O loss from nitrification and denitrification. NH3 volatilization and NO3-N leaching each accounted for 16–20% of the total N outputs. A system modeling approach is recommended to investigate the spatial variation of N leaching as affected by climatic conditions, and to fully account for the nitrogen fluxes. The N deposition derived from this study can be used as the background N input into the wheat-maize double cropping system for N balance.


Introduction
An improvement of nitrogen (N) management is the key to optimizing crop yields while avoiding costly surplus that 'leak' into the atmosphere and water environment [1].It also impacts on soil organic carbon (SOC) status through increased primary productivity (thus carbon return into the soil) and altering SOC decomposition [2,3].In spite of the importance of N management, closing the full nitrogen balance in agroecosystems remains difficult, mainly due to uncertainties in the flux estimation of input such as N deposition and output, for example N loss through nitrate leaching and NH 3 volatilization.This has led to confusions in understanding the efficiency of applied N fertilizers in different agroecosystems across the regions.
Nitrogen deposition, i.e., input of nitrogen balance from the atmosphere into the soil-plant system, has become an increasingly important part of nitrogen cycling [4,5].It determines the productivity of a cropping system without N fertilizer input, and also influences N recommendation to optimize crop yield with minimum impacts on the environment.It is predicted that global N deposition will increase by between 50 and 100% by 2030 relative to 2000, with the largest absolute increase occurring over east and south Asia [6].However, uncertainty in the estimation of N deposition remains large across the regions.These results derived from sporadic sites did not depict the spatial distribution of nitrogen deposition.It is the only method that the Integral Total Nitrogen Input (ITNI) allows direct determination of the total N deposition [7], and many researchers showed that N deposition determined by ITNI was consistent with the annual average of N uptake by crops in the zero-N plots in the long-term field experiment [8][9][10][11].There is an urgent need and feasible method of quantification of N deposition and its variation based on long-term experimentation, especially in China.N deposition has largely increased in China since 1980 and has had a huge impact on air pollution and ecosystem health [4].Particularly in the agricultural system, quantification of N deposition helps to optimize N management, and also provides the baseline N levels for zero N input treatment for modeling studies of cropping system performance.
The objectives of the present study are to: (1) Report the observed long-term crop productivity (biomass and yield levels) of wheat and maize under a zero and optimized nitrogen input level in North China, (2) investigate the crop nitrogen uptake under different N input levels and the use efficiency of applied N fertilizers across the three regions, (3) analyze N deposition in both wheat and maize growing seasons and its inter-annual variation, and (4) estimate the main inputs and outputs of the nitrogen balance in a wheat-maize double cropping system.

Study Sites
Three study sites were chosen from the National Fertilization Experiment Network China (Location showed in Figure 1).The experiments commenced in 1990 to investigate the response of crops to fertilizer applications across China.Experimental sites had been cropped for at least 100 years before initiation of the experiment.All the three sites have a semi-arid climate with an annual rainfall of 632, 631, and 533 mm in the experimental period (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) at Changping, Zhengzhou and Yangling, respectively.There is a same crop rotation, i.e., a winter wheat and summer maize double cropping system.Table 1 gives the detailed soil information.and NH3 volatilization.This has led to confusions in understanding the efficiency of applied N fertilizers in different agroecosystems across the regions.
Nitrogen deposition, i.e., input of nitrogen balance from the atmosphere into the soil-plant system, has become an increasingly important part of nitrogen cycling [4,5].It determines the productivity of a cropping system without N fertilizer input, and also influences N recommendation to optimize crop yield with minimum impacts on the environment.It is predicted that global N deposition will increase by between 50 and 100% by 2030 relative to 2000, with the largest absolute increase occurring over east and south Asia [6].However, uncertainty in the estimation of N deposition remains large across the regions.These results derived from sporadic sites did not depict the spatial distribution of nitrogen deposition.It is the only method that the Integral Total Nitrogen Input (ITNI) allows direct determination of the total N deposition [7], and many researchers showed that N deposition determined by ITNI was consistent with the annual average of N uptake by crops in the zero-N plots in the long-term field experiment [8][9][10][11].There is an urgent need and feasible method of quantification of N deposition and its variation based on long-term experimentation, especially in China.N deposition has largely increased in China since 1980 and has had a huge impact on air pollution and ecosystem health [4].Particularly in the agricultural system, quantification of N deposition helps to optimize N management, and also provides the baseline N levels for zero N input treatment for modeling studies of cropping system performance.
The objectives of the present study are to: (1) Report the observed long-term crop productivity (biomass and yield levels) of wheat and maize under a zero and optimized nitrogen input level in North China, (2) investigate the crop nitrogen uptake under different N input levels and the use efficiency of applied N fertilizers across the three regions, (3) analyze N deposition in both wheat and maize growing seasons and its inter-annual variation, and (4) estimate the main inputs and outputs of the nitrogen balance in a wheat-maize double cropping system.

Study Sites
Three study sites were chosen from the National Fertilization Experiment Network China (Location showed in Figure 1).The experiments commenced in 1990 to investigate the response of crops to fertilizer applications across China.Experimental sites had been cropped for at least 100 years before initiation of the experiment.All the three sites have a semi-arid climate with an annual rainfall of 632, 631, and 533 mm in the experimental period (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) at Changping, Zhengzhou and Yangling, respectively.There is a same crop rotation, i.e., a winter wheat and summer maize double cropping system.Table 1 gives the detailed soil information.

Experimental Design and Fertilizer Treatments
Detailed description of the experimental design can be found in Zhao et al. [12].In this study, we only concentrated on nitrogen and productivity and, therefore, only used the data from three treatments: (1) PK-phosphorus (P) and potassium (K) fertilizer application only, (2) NPK-nitrogen (N), P and K fertilizer applications and (3) HF-manure and NPK application.Detailed rates of fertilizers applied were showed in Table 2. Half of the mineral N fertilizer and all of P and K fertilizers, and all the manure were applied as a base application at sowing, and the rest of the mineral N fertilizer was applied as topdressing during the stem elongation stage for both the wheat and maize.The crops were irrigated as needed.Weeds and pests were controlled as necessary.

Crop and Soil Data
The total above ground biomass and grain yield of each crop were recorded every year during the study period.Nitrogen concentration in the biomass and grain was also measured.Nitrogen uptake of each crop was calculated from the biomass weight and the N concentrations.Apparent accumulated Nitrogen recovery efficiency (NRE ac ) [13] and partial factor productivity (PFP) [14] were indices of nitrogen efficiency, which affected crop productivity and N balance.They were calculated as the following equations: PFP(kg grain per kg where N U is total N uptake by crops (both of straw and grain, kg N ha −1 ), N R is the N rate applied (kg N ha −1 ), Y G is the grain yield (kg N ha −1 ), i represents the years of cultivation, and n is the total number of the year and equals to 15 in the present study.Soil organic matter (SOM) and total nitrogen (TN) were also measured in both soil layers of 0-20 and 20-40 cm each year after the maize harvest at each site.These data were used to analyze changes in soil fertility of each treatment.

Estimation of Nitrogen Balance Terms for NPK Treatment
Nitrogen balance was estimated for the NPK treatment at each study site.Nitrogen inputs include N fertilizer application, deposition, asymbiotic N fixation, and N from irrigation and crop seeds.Nitrogen outputs include crop N uptake, loss by NH 3 volatilization, N 2 O from nitrification and denitrification, and nitrate leaching.Nitrogen loss by soil surface runoff is ignored.Nitrogen balance was calculated by the following equation: Nitrogen deposition was considered to be equal to N uptake from the PK treatment according to He et al. [10] when soil total nitrogen kept at a dynamic equilibrium.Nitrogen input from irrigation was calculated by the irrigation amount multiplied by nitrogen content in groundwater, and then detailed information was shown in Table 3.There was no nitrogen fixing plants in the present study and the amount of nitrogen fixed by azotobacter was a regional mean (15 kg N ha −1 year −1 ).Losses of nitrogen volatilization, denitrification, and leaching were also regional average values in the wheat-maize cropping system in China, and the corresponding values were 22%, 3%, and 25% of nitrogen applied, respectively.The detailed information was recorded in Zhao et al. [15].

Productivity-Biomass and Grain Yield
Under zero nitrogen application over 15 years, the average biomass production of wheat and maize was 9.6, 10.6, and 8.8 Mg/ha in Changping, Yangling and Zhengzhou, respectively (Figure 2a).Wheat biomass accounted for 38% of the total.The corresponding total grain yield was 3.8, 4.7, and 3.7 Mg ha −1 at the three sites (Figure 2b).Wheat grain yield was 34% of the total.These figures represent the biomass and grain productivity levels under the conditions with natural nitrogen delivery (no N fertilization applications).
Under the expert-recommended nitrogen input level (NPK treatment, Table 2), the total biomass of the two crops reached 18.6, 25.4, and 24 Mg ha −1 , respectively, at Changping, Yangling, and Zhengzhou, respectively, with corresponding total grain yield of 8.4, 12.5, and 11.7 Mg ha −1 (Figure 2).Wheat biomass and yield accounted for 58% and 45% of the total biomass and yield, respectively.Nitrogen application led to a 244% increase in biomass and 259% in grain yield for wheat, while a 74% increase in biomass and 119% in yield for maize.
Further increase in nitrogen input in the experiment (HF treatment) did not lead to a significant increase in biomass and yield.However, recent records at nearby locations (news from local report) showed much higher biomass and yield for both wheat and maize (Figure 2).The recorded total grain yield was 66%, 63%, and 51% higher than that in the NPK treatment at Changping, Zhengzhou, and Yangling, respectively.It indicates that higher nitrogen application rates than what was recommended (expert-recommendations) may be needed to support the high productivity.
grain yield was 66%, 63%, and 51% higher than that in the NPK treatment at Changping, Zhengzhou, and Yangling, respectively.It indicates that higher nitrogen application rates than what was recommended (expert-recommendations) may be needed to support the high productivity.

Nitrogen Uptake and Recovery of Applied Nitrogen
Without nitrogen application across the three sites, average N uptake was 31 and 52 kg N ha −1 for wheat and maize, respectively.Averaged N uptake by both of wheat and maize was 76, 80, and 94 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively (Figure 3), which represent nitrogen from atmospheric deposition and soil mineralization.
Under the recommended N rate, the average N uptake was 141 and 122 kg N ha −1 for wheat and maize, respectively.Total N uptake of wheat and maize at Zhengzhou and Yangling was significantly higher than that at Changping.Average nitrogen recovery of two crops over 15 years was 71%, 80%, and 83% at Changping, Zhengzhou, and Yangling, respectively (Figure 4).The corresponding partial factor productivity (PFP) was 28, 36, and 33 kg grain per kg N.More N input under the HF treatment increased N uptake for both wheat and maize (157 and 141 kg N ha −1 , respectively), but did not significantly increase the biomass and yield.Both NRE and PFP decreased under HF treatment.They were 61%, 58%, and 78% for NRE and 22, 25, and 29 kg grain per kg N for PFP at Changping, Zhengzhou, and Yangling, respectively.

Nitrogen Uptake and Recovery of Applied Nitrogen
Without nitrogen application across the three sites, average N uptake was 31 and 52 kg N ha −1 for wheat and maize, respectively.Averaged N uptake by both of wheat and maize was 76, 80, and 94 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively (Figure 3), which represent nitrogen from atmospheric deposition and soil mineralization.

Nitrogen Uptake and Recovery of Applied Nitrogen
Without nitrogen application across the three sites, average N uptake was 31 and 52 kg N ha −1 for wheat and maize, respectively.Averaged N uptake by both of wheat and maize was 76, 80, and 94 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively (Figure 3), which represent nitrogen from atmospheric deposition and soil mineralization.
Under the recommended N rate, the average N uptake was 141 and 122 kg N ha −1 for wheat and maize, respectively.Total N uptake of wheat and maize at Zhengzhou and Yangling was significantly higher than that at Changping.Average nitrogen recovery of two crops over 15 years was 71%, 80%, and 83% at Changping, Zhengzhou, and Yangling, respectively (Figure 4).The corresponding partial factor productivity (PFP) was 28, 36, and 33 kg grain per kg N.More N input under the HF treatment increased N uptake for both wheat and maize (157 and 141 kg N ha −1 , respectively), but did not significantly increase the biomass and yield.Both NRE and PFP decreased under HF treatment.They were 61%, 58%, and 78% for NRE and 22, 25, and 29 kg grain per kg N for PFP at Changping, Zhengzhou, and Yangling, respectively.Under the recommended N rate, the average N uptake was 141 and 122 kg N ha −1 for wheat and maize, respectively.Total N uptake of wheat and maize at Zhengzhou and Yangling was significantly higher than that at Changping.Average nitrogen recovery of two crops over 15 years was 71%, 80%, and 83% at Changping, Zhengzhou, and Yangling, respectively (Figure 4).The corresponding partial factor productivity (PFP) was 28, 36, and 33 kg grain per kg N.More N input under the HF treatment increased N uptake for both wheat and maize (157 and 141 kg N ha −1 , respectively), but did not significantly increase the biomass and yield.Both NRE and PFP decreased under HF treatment.They were 61%, 58%, and 78% for NRE and 22, 25, and 29 kg grain per kg N for PFP at Changping, Zhengzhou, and Yangling, respectively.To achieve the productivity recorded at locations nearby the research site, the estimated N uptake was 220 and 219 kg N ha −1 year −1 for wheat and maize, respectively.There was a 64%, 84%, and 53% increase in total uptake N than that under the recommended N rate at Changping, Zhengzhou, and Yangling, respectively.

Nitrogen Deposition in Crop Seasons (Wheat and Maize)
Nitrogen deposition, as the total crop N uptake in the PK treatment, showed significant inter-annual variations, but no significant trend with time at all the three study sites (Figure 5).Average annual N deposition was 76, 80, and 94 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively.Distribution of nitrogen deposition in wheat and maize seasons also showed inter-annual variations.Nitrogen deposition during wheat growing season accounted for 18~50%, 25~67%, and 26~56% of the annual total at Changping, Zhengzhou, and Yangling, respectively, with corresponding average values of 36%, 44%, and 36%.To achieve the productivity recorded at locations nearby the research site, the estimated N uptake was 220 and 219 kg N ha −1 year −1 for wheat and maize, respectively.There was a 64%, 84%, and 53% increase in total uptake N than that under the recommended N rate at Changping, Zhengzhou, and Yangling, respectively.

Nitrogen Deposition in Crop Seasons (Wheat and Maize)
Nitrogen deposition, as the total crop N uptake in the PK treatment, showed significant inter-annual variations, but no significant trend with time at all the three study sites (Figure 5).Average annual N deposition was 76, 80, and 94 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively.Distribution of nitrogen deposition in wheat and maize seasons also showed inter-annual variations.Nitrogen deposition during wheat growing season accounted for 18~50%, 25~67%, and 26~56% of the annual total at Changping, Zhengzhou, and Yangling, respectively, with corresponding average values of 36%, 44%, and 36%.
Rainfall during wheat growing season (from Oct to Dec and from Jan to May) accounted for 3~39%, 20~59%, and 30~62% of the annual rainfall at Changping, Zhengzhou, and Yangling, respectively, with corresponding average values of 21%, 35%, and 42% (Figure 6).There was no close relationship between the fraction of wheat season N deposition and the fraction of wheat season rainfall to their annual totals.
However, in 12 years at Changping and 11 years at Zhengzhou out of the total 15-year experimental period, the fraction of wheat season N deposition was higher than the fraction of wheat season rainfall, indicating that dry deposition was most significant at the two locations (Figure 7).While at Yangling, in 12 out of the 15 years, the fraction of wheat season N deposition was lower than the fraction of wheat season rainfall, implying more significant wet deposition at this site.Rainfall during wheat growing season (from Oct to Dec and from Jan to May) accounted for 3~39%, 20~59%, and 30~62% of the annual rainfall at Changping, Zhengzhou, and Yangling, respectively, with corresponding average values of 21%, 35%, and 42% (Figure 6).There was no close relationship between the fraction of wheat season N deposition and the fraction of wheat season rainfall to their annual totals.However, in 12 years at Changping and 11 years at Zhengzhou out of the total 15-year experimental period, the fraction of wheat season N deposition was higher than the fraction of wheat season rainfall, indicating that dry deposition was most significant at the two locations (Figure 7).While at Yangling, in 12 out of the 15 years, the fraction of wheat season N deposition was lower than the fraction of wheat season rainfall, implying more significant wet deposition at this site.Rainfall during wheat growing season (from Oct to Dec and from Jan to May) accounted for 3~39%, 20~59%, and 30~62% of the annual rainfall at Changping, Zhengzhou, and Yangling, respectively, with corresponding average values of 21%, 35%, and 42% (Figure 6).There was no close relationship between the fraction of wheat season N deposition and the fraction of wheat season rainfall to their annual totals.However, in 12 years at Changping and 11 years at Zhengzhou out of the total 15-year experimental period, the fraction of wheat season N deposition was higher than the fraction of wheat season rainfall, indicating that dry deposition was most significant at the two locations (Figure 7).While at Yangling, in 12 out of the 15 years, the fraction of wheat season N deposition was lower than the fraction of wheat season rainfall, implying more significant wet deposition at this site.

Soil Fertility Changes
Soil organic matter (SOM), total soil nitrogen (TN), and available nitrogen (AN) showed no significant difference between treatments (Table 4), except for treatment HF.It indicates that N input under NPK treatment did not lead to increased soil fertility as compared to the PK treatment.Furthermore, SOM and TN in 0-40 cm soil profiles from PK treatment were not significantly different across the 15 years at all sites except in the 0-20 cm soil profile at Yangling.This implies

Soil Fertility Changes
Soil organic matter (SOM), total soil nitrogen (TN), and available nitrogen (AN) showed no significant difference between treatments (Table 4), except for treatment HF.It indicates that N input under NPK treatment did not lead to increased soil fertility as compared to the PK treatment.Furthermore, SOM and TN in 0-40 cm soil profiles from PK treatment were not significantly different across the 15 years at all sites except in the 0-20 cm soil profile at Yangling.This implies that the input of carbon and nitrogen into the soil balanced the output even under zero nitrogen application.Note: Different letters indicate significant difference at p < 0.05 level between treatments in same site.

Overall N Balance at the Three Sites
The estimated overall nitrogen balance for the NPK treatment at the three study sites was shown in Table 5.Total inputs from NPK were 419, 462, and 480 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively.The corresponding total outputs were 370, 462, and 494 kg N ha −1 year −1 .It turns out that the N deposition roughly balances out the N loss due to NH 3 volatilization and N 2 O loss from nitrification and denitrification.N deposition accounted for 17-21% of the total N inputs, while NH 3 volatilization and NO 3 -N leaching each accounted for 16-20% of the total N outputs.The balance was not closed at Changping and Yangling due to uncertainties in the estimations of N fluxes.There are large uncertainties associated with the estimation of NH 3 , N 2 O and nitrate leaching.

Discussion
Apparent N recovery efficiency is an important index of nitrogen balance in agriculture, and shows the overall N uptake by the crop as a fraction of the applied N during the growing season.In a previous paper, Liu et al. [13] analyzed the difference between NRE and NUF (nitrogen utilization efficiency) from NPK treatment at the same site that could be contributed to the residual effect of N fertilizer, and has been discussed.Furthermore, the apparent N recovery is closely related to the N deposition at the three study sites (Figures 3-5).Both were lowest at Changping and highest at Yangling.However, the difference of partial factory productivity (PFP) was different (Figure 4).The results could suggest that the N rate from the NPK treatment at Changping and Zhengzhou should be decreased but there may be no change in yield due to the lower recorded yield in Changping and higher N deposition at Yangling.
Previous studies showed that increased anthropogenic emissions of both nitrogen oxide compounds (NO X ) and the reduced nitrogen compound (NH 3 ) have resulted in an increase in nitrogen deposition globally [16,17].Although, rates of N deposition have been stabilized in the US and Europe since the late 1980s or early 1990s with the implementation of strict legislation to limit atmospheric pollution [18], global N deposition is predicted to increase by between 50 and 100% by 2030 relative 2000, with the largest absolute increase occurring over east and south Asia [6].Liu et al. [19] reviewed nitrogen deposition in China from 1980 to 2007 and showed a substantial increases since the 1980s due to increased NH 3 and NO X emissions.Zhang [20] re-estimated the total N deposition in China to be 15 Tg N year −1 in the 2000s compared with 7.4 and 12.0 T g N year −1 in the 1980s and 1990s, respectively.The evidence shows that the rate of total N deposition has increased in China.These results, however, were derived using data from sporadic sites in China and did not depict the spatial distribution of nitrogen deposition.Liu and Tian [21] further reported that the total deposition rates of wet and dry deposition peaked over the central south China with maximum values of 63.53 kg N ha −1 year −1 , while total nitrogen deposition in north China was in a range of 1.08-18.65kg N ha −1 year −1 .They reported the trend in NO x deposition to be positive in the southeast and southwest of China, negative in northwest and central parts of the region, and zero in the northwest.Our results, i.e., the N deposition as the crop N uptake under PK treatment, showed that N deposition increased significantly with time at Yangling, and did not have a significant trend at Changping and Zhengzhou (Figure 3).This is consistent with Liu and Tian [21].Changping and Zhengzhou are located on the border between zones with a negative trend and a no trend, while Yangling is located in the zone with a positive trend as described by Liu and Tian [21].Another major part of total nitrogen deposition is reduced N (NH 3 ), and its relatively short lifetime (hours to days) [22].In addition, it is mainly deposited near the location where they were emitted [17].Zhang [20] reported that the reduced N dominated the total N deposition with 2.5 times more than oxidized N deposition, and concluded that the total N deposition was derived mainly from agriculture sources in the North China Plain.Rates of N applied in the present study have always remained stable.The above evidence seems to confirm the increasing trend of total N deposition at Yangling, while no trend at Changping and Zhengzhou.
For seasonal distribution of nitrogen deposition, Zhang et al. [23] reported that 60% of bulk deposition occurred from June to September and that there was a positive relationship between inorganic N deposition and precipitation in the Beijing area.This result seems to indicate that the amount of wet deposition was higher than that of dry deposition.However, our results only indicated some relationship between N deposition and rainfall at Yangling (Figure 7), but no significant relationships at Changping and Zhengzhou, where the wheat (dry) season N deposition was higher than that in the maize (wet) season.The inconsistency in the results could be explained by the method used in Zhang et al., where they measured only a part of the total deposition [10].Our results imply that besides wet deposition, dry deposition could be another major component of total N deposition, especially during wheat growing season.Liu and Zhang [24] reported that a sand storm happened on April 16-17 in 2005 and the amount of nitrogen from the event during 2 h was 6 × 10 3 t in Beijing.
They concluded that dry nitrogen deposition during winter and spring seasons should not be ignored in northern and especially northwest regions of China.
It has been reported that conventional agriculture has led to a decrease of soil organic matter [25], especially in China [26].However, the results in the present study showed that organic matter and nitrogen from PK treatment did not significantly change in the 0-40 cm soil profile at CP and ZZ, even with a marked increase in the 0-20 cm soil profile at YL, which implies that under zero nitrogen application, stubble and root biomass produced with deposited N from the atmosphere could replenish decomposition of soil carbon and nitrogen under current climate and agricultural management.However, this could only maintain a very low level of soil organic carbon (Table 1).Under the recommended N rate (NPK), crop yield could be maintained at a level significantly higher than that at zero nitrogen treatment (PK).However, the application of N, P, and K fertilizers did not increase the content of SOC, TN, and AN (Table 3).This also implies that the higher residue return to soil was accompanied by a higher decomposition of SOC, at least at the top soil layers.This may indicate an increased SOC decomposition due to increased N applications, consistent with the results in Khan et al. [2] and Mulvaney et al. [3].
The present study showed that the nitrogen budgets were 49, 1 and −14 kg N ha −1 year −1 at Changping, Zhengzhou, and Yangling, respectively.That means total soil N could increase at Changping, almost remain stable at Zhengzhou, and decrease at Yangling.The experimental data, however, showed that total soil N only in 0-20cm soil profiles significantly increased at Yangling and did not remarkably change at Changping and Zhengzhou (Table 4), which was inconsistent with the estimated nitrogen balance.This inconsistency is likely caused by inaccurate estimation of the nitrogen fluxes, particularly the NO 3 -N leaching and NH 3 volatilization.Nitrogen losses through leaching and volatilization are subject to management practices, including the type, time, rate, and placement of N applications and whether rainfall or irrigation occurred after N application [27].In the present study, although management practices were kept almost similar across sites, the climatic conditions, particularly rainfall distribution during the seasons were significantly different (Figure 6).The fraction of maize season rainfall was highest at Changping and lowest at Yangling, which may have led to significant differences in N leaching and/or NH 3 volatilization [28,29].Addiscott [30] also reported that there was the closest positive relationship between the losses of N applied and the rainfall.A systems modeling approach could be helpful in order to fully account for the nitrogen balance.

Conclusions
During the 15 years from 1990 to 2005, the average annual nitrogen deposition at Changping, Zhengzhou, and Yangling were 76, 80, and 94 kg N ha −1 , respectively.For the wheat and maize double rotation system, the deposited N could support a correspoding total biomass production (wheat plus maize) of 9.6, 10.6, and 8.8 Mg ha −1 , and total grain yield of 3.8, 4.8, and 3.7 Mg ha −1 .Without N fertilization, the system could maintain a low level of soil organic matter and nitrogen balance.N fertiliser application with the recommended N rate, increased total biomass by 244% and 74% for wheat and maize, respectively.The corresponding increase in total grain yield was 259% and 119%.Under optimal N management, N deposition accounted for 17-21% of the total N inputs, contributing significantly to crop uptake and influencing the apparent recovery efficiency of the applied N. At the study sites, N deposition showed significant spatial variations in terms of the fraction of dry and wet depositions, which seems to be location-dependent.Annual N deposition roughly balanced out N losses due to NH 3 volatilization and N 2 O loss from nitrification and denitrification.NH 3 volatilization and NO 3 -N leaching each accounted for 16-20% of the total N outputs.A system modeling approach is recommended to investigate the spatial variation of N leaching as affected by climatic conditions, and to fully account for the nitrogen fluxes.The N deposition in both the wheat and maize seasons derived from this study can be used as the background N input of N balance in the wheat-maize double cropping system.

Figure 1 .
Figure 1.The location of the three experiments.Figure 1.The location of the three experiments.

Figure 1 .
Figure 1.The location of the three experiments.Figure 1.The location of the three experiments.

Figure 2
Figure 2 Biomass (a) and grain (b) yield under different N fertilization treatments at the three study's and recorded high yield in corresponding regions.Note: Recorded grain yield from the regional report, BJ-Beijing, HN-Henan, SX-Shaanxi.Error bar indicates standard error between biomass and yield of wheat plus maize across 15 years, Recorded biomass calculated using recorded grain yield multiplied by harvest index from corresponding experimental sites.

Figure 2 .
Figure 2. Biomass (a) and grain (b) yield under different N fertilization treatments at the three study's and recorded high yield in corresponding regions.Note: Recorded grain yield from the regional report, BJ-Beijing, HN-Henan, SX-Shaanxi.Error bar indicates standard error between biomass and yield of wheat plus maize across 15 years, Recorded biomass calculated using recorded grain yield multiplied by harvest index from corresponding experimental sites.

Figure 2
Figure 2 Biomass (a) and grain (b) yield under different N fertilization treatments at the three study's and recorded high yield in corresponding regions.Note: Recorded grain yield from the regional report, BJ-Beijing, HN-Henan, SX-Shaanxi.Error bar indicates standard error between biomass and yield of wheat plus maize across 15 years, Recorded biomass calculated using recorded grain yield multiplied by harvest index from corresponding experimental sites.

Figure 3 .
Figure 3. Nitrogen in aboveground biomass under different N treatments and estimated nitrogen in biomass with recorded high yields.Note: BJ-Beijing, HN-Henan, SX-Shaanxi.Error bars indicate one standard error.Recorded biomass N based on relationship between grain yield and biomass N reported by Liu et al. [13].

Sustainability 2018 , 13 Figure 3 .
Figure 3. Nitrogen in aboveground biomass under different N treatments and estimated nitrogen in biomass with recorded high yields.Note: BJ-Beijing, HN-Henan, SX-Shaanxi.Error bars indicate one standard error.Recorded biomass N based on relationship between grain yield and biomass N reported by Liu et al. [13].

Figure 5 .
Figure 5. Estimation of Total N deposition across 15 years by total crop N uptake from PK treatment.

Figure 6 .
Figure 6.Monthly average rainfall at the three study sites.

Figure 5 .
Figure 5. Estimation of Total N deposition across 15 years by total crop N uptake from PK treatment.

Figure 5 .
Figure 5. Estimation of Total N deposition across 15 years by total crop N uptake from PK treatment.

Figure 6 .
Figure 6.Monthly average rainfall at the three study sites.

Figure 6 . 13 Figure 7 .
Figure 6.Monthly average rainfall at the three study sites.Sustainability 2018, 10, x FOR PEER REVIEW 8 of 13

Figure 7 .
Figure 7. Relationship between the fraction of wheat season rainfall and the fraction of wheat season N deposition.

Table 1 .
The experimental sites and its initial properties.

Table 3 .
Nitrogen input from irrigation for both wheat and maize.

Table 4 .
Soil organic matter (SOM), total soil nitrogen (TN) and available soil nitrogen (AN) after 15 years of cultivation at the study sites.

Table 5 .
Nitrogen balances in the wheat-maize rotation system at the three study sites.