Sound Water and Nitrogen Management Decreases Nitrogen Losses from a Drip-Fertigated Cotton Field in Northwestern China

Drip irrigation systems are becoming more and more mature and are now widely used to improve crop yield and nitrogen use efficiency in Xinjiang, NW China. However, it is not known if leaching is occurring or not and whether leaching will harm the water environment following N fertilization and drip irrigation. The purpose of our study was to estimate the leaching volumes, nitrogen losses, forms of nitrogen losses, and nitrogen loss coefficients under different N fertilization, P fertilization, K fertilization and irrigation regimes. A long-term field experiment was conducted from 2009 to 2015 in Baotou Lake farm in Korla City, Xinjiang, with drip-irrigated cotton (Gossypium hirsutum L.) being grown under different N fertilizer and irrigation regimes. The treatments were designed comprising 0 N, 0 P, and 0 K with an irrigation of 480 mm as the control(N0P0K0W480) and the following three other treatments: (1) 357 kg N·hm−2, 90 kg P·hm−2, 0 kg K2O hm−2, and irrigation of 480 mm (N357P90K0W480); (2) 357 kg N·hm−2, 90 kg P·hm−2, 62 kg K·hm−2, and irrigation of 420 mm (N357P90K62W420); and (3) 240 kg N·hm−2, 65 kg P·hm−2, 62 kg K·hm−2, and irrigation of 420 mm (N240P65K62W420). The results showed the following: (1) the leaching volume was determined by nitrogen fertilization, phosphorus fertilization, and the irrigation amount. In general, the leaching volume was highest under treatment N357P90K0W480. (2) The nitrogen loss was highest under treatment N357P90K0W480. (3) Nitrate nitrogen (NO3–) was the main form of nitrogen lost, followed by ammonium nitrogen (NH4+). (4) The annual nitrogen loss coefficients followed the order of: N357P90K0W480 > N357P90K62W420 > N240P65K62W420 > N0P0K0W480, with values of 0.85, 0.55, 0.30, and 0, respectively. The leaching volume, nitrogen loss, nitrate nitrogen, ammonium nitrogen, and annual nitrogen loss coefficient were lowest under the N240P65K62W420 treatment, except in the N0P0K0W480treatment. These results demonstrate that optimizing the management of water and nitrogen (N240P65K62W420 treatment) can effectively reduce nitrogen losses under drip fertigation and plastic mulching.


Introduction
Nitrogen (N) is one of the main essential plant nutrients and it is taken up by crops throughout the growing season, thereby affecting the plant growth and the yield [1,2]. Nitrogen (N) often comes from urea for field use because of its high N content, low cost, fast conversion, easy dissolution and easy storage [3]. The conversion of urea into ammonium is rapid. Mineralization and nitrification processes convert organic N and NH 4 + into NH 4 + and NO 3 -, respectively, which can be absorbed and utilized by crops. The mineralization and nitrification processes depend on climatic and edaphic factors, such as the soil moisture and temperature [4].
Cotton (Gossypium hirsutum L.) is the main cash crop in Xinjiang, located in northwestern China [5]. Because of arid climate in Xinjiang, drip irrigation is very popular in this area, especially for cotton. Drip fertigation allows the control of the amount of water applied and the concentrations of nutrients supplied to crops, thereby conserving water resources and reducing fluctuation in the concentrations of nutrients in the soil during the crop growing season [6]. The appropriate application of nitrogen fertilization and water can meet the needs of crops and so increase crop yields [7]. However, nitrogen accumulation in the soil, as well as leaching and percolating from the soil profile may occur when the amounts of nitrogen fertilization and water applied exceed the crop's requirements [8]. Furthermore, leaching and percolation can transfer nitrogen (N) from farmlands into aquatic systems [9], thereby leading to water pollution, which is one of the most significant environmental problems at present [10], i.e., non-point source pollution [11].
Many scholars have done a lot of research on cotton under mulch drip irrigation. Dai and Dong [12] discussed the advantages and disadvantages of applying nitrogen fertilization to grow cotton under drip irrigation and plastic mulching in China. Nitrogen fertilization improves cotton production, and it is especially important during cotton growth, but excessive application is detrimental to the environment [13]. Several studies have investigated the effects of nitrogen application on N leaching during drip-irrigated maize production and the losses of nitrogen from the soil. Jia et al. [14] found that high application rates of nitrogen fertilization and irrigation could increase nitrogen leaching from the soil and lead to the over-consumption of fresh water resources in summer maize (Zea mays L.). Wang and Xing [15] showed that the leaching of nitrate nitrogen from the soil was lower under plastic mulching than without plastic mulching in maize (Zea mays L.), and nitrate nitrogen was the main form of nitrogen leached from the soil. Although various studies have investigated N leaching in maize, most focused on the effects of soil and plastic mulching on nitrogen losses, whereas few considered the losses of nitrogen in leachate water. In addition, it is not clear whether the leaching that occurs under maize production is similar to that with cotton. In addition, most recent studies of N leaching were conducted using laboratory soil columns [16,17], whereas little information is available regarding the N leached in leachate water in long-term field experiments under nitrogen fertilization and drip irrigation in arid regions, especially for cotton. Cotton is one of the most important cash crops in Xinjiang and its production consumes large amounts of N fertilization. Increasing the plant N use efficiency and reducing leaching are an important goal for the development of sustainable agriculture [18].
Therefore, the objectives of this study were: (1) to verify whether nitrogen leaching occurs in cotton fields under drip irrigation in northwest China; (2) to assess how much nitrogen is leached under different nitrogen fertilization and irrigation treatments based on long-term field observations; (3) to examine the nitrogen losses and patterns in the leachate water; (4) to determine the nitrogen loss coefficients under different treatments in order to understand the law of nitrogen leaching in cotton field under drip irrigation in Northwest China, and provide reference for reasonable water and nitrogen management.

Site Description and Soil Properties
Experiments were conducted in each cotton growing season from 2009 to 2015 in different fields at the long-term positioning monitoring base on Baotou Lake farm in Korla City, Xinjiang, China (41.6903 • N, 85.8719 • E). The annual precipitation is 56.2 mm and the evapotranspiration is 2497.4 mm. The effective accumulated temperature is 4252.2 • C and the frost free period is 205 days. The groundwater level is 2 to 2.5 m. Field soil belongs to sandy loam soil in Chinese soil classification, with medium fertility. The bulk density of topsoil (0-20 cm) in the field is 1.23 g·cm −3 . In 2009, the properties of the surface soil at the trial site were: pH = 8.46, organic matter = 7.51 g·kg −1 , total N = 0.45 g·kg −1 , total P = 0.046 g·kg −1 , available phosphorus (P) = 4.99 mg·kg −1 , available potassium (K) = 95.93 mg·kg −1 , NO 3 − -N = 7.27 mg·kg −1 , and NH 4 + -N = 3.61 mg·kg −1 .

Experimental Design and Agronomic Management
For each treatment, it was set up in a factorial design comprising 0 N as urea, 0 P 2 O 5 as calcium phosphate, 0 K 2 O as K 2 SO 4 , and irrigation of 480 mm(N 0 P 0 K 0 W 480 )for the control, and the following three treatments: (1) 357 kg N·hm −2 as urea, 90 kg P·hm −2 as calcium phosphate, 0 K as K 2 SO 4 , and irrigation of 480 mm (N 357 P 90 K 0 W 480 ); (2) 357 kg N hm −2 as urea, 90 kg P hm −2 as calcium phosphate, 62 kg K hm −2 as K 2 SO 4 , and irrigation of 420 mm (N 357 P 90 K 62 W 420 ); and (3) 240 kg N hm −2 as urea, 65 kg P hm −2 as calcium phosphate, 62 kg K hm −2 as K 2 SO 4 , and irrigation of 420 mm (N 240 P 65 K 62 W 420 ) . The nitrogen application rate was set as 357 kg hm −2 and the irrigation amount as 480 mm based on the average yields verified by long-term trials and production in the region [19]. N 357 P 90 K 0 W 480 treatment represented conventional N fertilization and irrigation. N 357 P 90 K 62 W 420 treatment represented conventional N fertilization and optimal irrigation. N 240 P 65 K 62 W 420 treatment represented optimal N fertilization and irrigation. All of the plots received P as calcium phosphate and K as K 2 SO 4 by hand spreading, and they were into the soil using a rotator before sowing. Granular urea was applied where 20% of the total N was spread into field by hand before sowing for the three treatments, except for N 0 P 0 K 0 W 480 , and the remaining 80% of N was solubilized and applied with irrigation water according to the proportions used in the different treatments after planting (Table 1). For cotton, the first and last drip irrigation is only irrigation without fertilization. For the N 357 P 90 K 0 W 480 treatment, the remaining 80% of solubilized N was applied in the second, third, and fourth irrigation events with 36.8%, 21.6%, and 21.6%, respectively. For the N 240 P 90 K 62 W 420 treatment, the remaining 80% of solubilized N was applied in the second, third, fourth, and fifth irrigation events with 32%, 20%, 16%, and 12%, respectively. For the N 240 P 65 K 62 W 420 treatment, the remaining 80% of solubilized N was applied in the second, third, fourth, and fifth irrigation events with 32%, 20%, 16%, and 12%, respectively. Irrigation time was based on the dates, and these dates were critical periods for cotton growth. In generally, cotton was planted in late April, then the first irrigation started at the beginning of June only with irrigation, the last irrigation was carried out in mid-August only with irrigation. The remaining six irrigations were usually irrigated once a week, and may be shorter than once a week with irrigation and fertilization. The second irrigation was the bud stage of cotton. The third irrigation was the full bud stage of cotton. The fourth irrigation was the initial flowering stage of cotton. The fifth irrigation was the full flowering stage of cotton. The sixth irrigation was the flowering stage of cotton. The seventh irrigation was the cotton boll opening period. The specific irrigation times and fertilization rates are shown in Table 1. The treatments were randomized into a complete block design with three replicates. The size of each plot was 4.5 × 7.4 m. In the experiment, cotton (variety Xinluzao 38) was planted in late April. Drip irrigation pipes and the plastic mulch were applied before planting using a custom built tractor-drawn seeder. Seeds were seeded with a film of four rows, and the distance between rows was 60 cm and the distance between seeds within a row was 30 cm. The plastic mulch is composed of high-density, airtight transparent polyethylene film, covering two rows.
Standard management practices were used to control weeds and pests using herbicides and pesticides.

Installation of Experimental Device and Analysis of NO 3 − andNH 4 + Levelsin Underground Leachate Samples
The NO 3 − and NH 4 + concentrations were determined in samples of leachate water obtained underground using a drainage collector. The drainage collector comprised a special polyvinyl chloride dish (0.30 m long × 0.60 m wide × 0.08 m high) and a 25 l polyvinyl chloride bucket, which were connected by plastic pipes (Figure 1). A plastic pipe extended from the water storage bucket as a water intake pipe. A set of drainage collectors was buried in each plot. First, a concrete enclosure was prepared by forming a rectangular boundary with a length of 150 cm and width of 80 cm in the middle of the plot. Next, the soil profile was dug to a depth of 90 cm in the designated area. The bottom of the section was cut into an inverted trapezoid where the periphery was slightly higher than the middle (by 5 cm). The leaching barrel was then placed vertically into a small round depression. Finally, the soil was backfilled layer by layer according to the reverse order that the soil was excavated and compaction was applied up to the original bulk density while backfilling. In our study, leaching water was taken nine times every year after one time of spring irrigation and eight times for drip irrigation. Leaching water samples were collected on the third day after each irrigation was applied. Although the annual precipitation of 56.2 mm, which mainly concentrated in April and was divided into several rainfall events, then absorbed by soil. In addition, each experimental plot had a ridge to prevent fertilizer from channeling and store water. The drip irrigation started in June every year, while the rainfall only fell in April and the intensity of the rainfall was small, so the leaching samples did not occur and we neglect the leaching from rainfall events.
The leachate volume was measured using a graduated cylinder measuring 1 l. The nitrate and ammonium concentrations in the leachate water were determined using colorimetric procedures with a continuous flow analytical system (TRAACS 2000). The total nitrogen concentration in the leachate water was determined by potassium persulfate oxidation-ultraviolet spectrophotometry.

N Loss Analysis
The growing season annual nitrogen loss (L N ) and N loss coefficient (θ N ) for each different treatment was calculated as: where L N represents the annual nitrogen loss, c i represents the nitrogen concentration in the leachate, v i is the volume of the leachate, i is the sampling time and m is the total number of leachate.
where θ N represents the N loss coefficient, L N,A represents the annual nitrogen loss under A treatment, L N,CK represents the annual nitrogen loss under N 0 P 0 K 0 W 480 treatment, and M N,A represents the N applied as fertilization under A treatment.

Statistical Analysis
Differences in annual leaching volumes among different treatments were analyzed by one-way parametric analysis of variance using SPSS version 11.5 (SPSS, Chicago, IL, USA). Line charts were prepared to assess the changes in the annual nitrogen loss, annual nitrate nitrogen, and ammonium nitrogen losses under different treatments. The relationships between the leaching volumes and N, P, and K fertilization and irrigation regimes were analyzed by linear regression using SPSS version 11.5 (SPSS, Chicago, IL, USA).

Irrigation and Leachate Volumes
The experimental plots had a continental arid climate with annual precipitation of 56.2 mm, mainly concentrated in April, divided into several rainfall events and rainfall intensity was small, which was much less than the irrigation regimes, and the drip irrigation started in June every year, thus we did not consider the rainfall. The annual irrigation input regimes were 480, 480, 420, and 420 mm for the N 0 P 0 K 0 W 480 , N 357 P 90 K 0 W 480 , N 357 P 90 K 62 W 420 , and N 240 P 65 K 62 W 420 treatments, respectively (Table 1). For the four treatments of this experiment, the leachate volumes were greater when the irrigation regimes were higher in generally. Figure 2 shows the trends change among years. In general, the volume of leachate under N 0 P 0 K 0 W 480 was significantly higher compared with the other treatments in 2009, 2011 and 2015. No fertilization was applied under N 0 P 0 K 0 W 480 treatment, so the growth of cotton was poor and less water was needed, and thus the leachate water volume was high [20]. The leachate volumes were significantly higher under treatment N 357 P 90 K 0 W 480 compared with the other N fertilization treatments because the irrigation regimes were greater. The irrigation regimes were the same for treatments N 357 P 90 K 62 W 420 and N 240 P 65 K 62 W 420 but the trends in the leachate volumes differed, probably because the growth of cotton varied and the cotton yields under both treatments differed among years according to the amount of fertilization applied (Table 3) [21].

Effects of Fertilization on Nitrogen Losses in Leachate Water
The nitrogen losses in the leachates under the four treatments are presented as breakthrough curves in Figure 3. The nitrogen losses in the leachate water ranged from 4.92 to 7.07 kg·hm −2 in the first year. The nitrogen losses in the leachate water were largest during 2010 under all four treatments. Over the seven years, the maximum annual nitrogen losses in the leachate water followed the order of: N 357 P 90 K 0 W 480 > N 240 P 65 K 62 W 420 > N 357 P 90 K 62 W 420 > N 0 P 0 K 0 W 480 , with values of 6.89, 5.57, 5.36, and 4.79 kg·hm −2 , respectively. In general, the nitrogen loss in the leaching water was lower under the N 357 P 90 K 62 W 420 treatment compared with the other N fertilization and irrigation treatments. According to the overall trends determined for each treatment, the nitrogen losses in the leachate water were highest under treatment N 357 P 90 K 0 W 480 . For theN 0 P 0 K 0 W 480 treatment irrigated without fertilization, but the residual N in the soil was leached with irrigation, and the N loss in the leached water showed the same trend. Compared with N 0 P 0 K 0 W 480 , the losses of nitrogen ranged from 0.78 to 2.11 kg hm −2 in the N 357 P 90 K 0 W 480 and N 240 P 65 K 62 W 420 treatments, and between 0 and 0.57 kg hm −2 in the N 357 P 90 K 62 W 420 treatment. Figure 4A shows the annual losses of NO 3 − in the leachates from the four treatments. The nitrate nitrogen losses in the leachate water ranged from 1.56 to 4.62 kg·hm −2 in the first year. The highest annual NO 3 − losses occurred in 2010 coinciding with maximum total nitrogen losses. Similar trends were observed for all four treatments. The ammonium nitrogen losses in the leachates under all four treatments are shown as break through curves in Figure 4B. The ammonium nitrogen losses in the leachates ranged from 0.35 to 1.54 kg hm −2 in the first year. The peak losses of NH 4 + in the leachate water were 1.17 and 1.24 kg hm −2 in N 357 P 90 K 62 W 420 and N 357 P 90 K 0 W 480 , respectively in 2010, and they decreased in the leachate water samples in the next few years. Compared with N 0 P 0 K 0 W 480 without nitrogen application, the losses of NH 4 + ranged from 0.04 to 1.19 kg·hm −2 in the N 0 P 0 K 0 W 480 treatment, and the N 357 P 90 K 62 W 420 treatment resulted in NH 4 + losses of 0 to 0.27 kg hm −2 .

Nitrogen Leaching and Leachate Volumes
Nitrogen that is not taken up by crops due to the excessive application of irrigation or nitrogen is likely to be leached. Nitrogen leaching depends on the irrigation treatment, fertilizer type, and fertilizer rate, and there is a strong interaction between the irrigation treatment and fertilizer rate [22]. In our study, increasing the irrigation by 14.3% increased N leaching under all of the fertilizer treatments, and by almost two times (Figure 3). This result showed that the amount of irrigation accounted for a large part in leaching. The D-value of N leaching (Figure 3) was larger under N 357 P 90 K 62 W 420 and N 357 P 90 K 0 W 480 treatments in 2013, while there was significant difference in yield (Table 3), which may be attributed to a higher K rate. Til et al. [23] found that mineral K fertilization acts positively on irrigation [24]. Thus, the leaching under N 357 P 90 K 62 W 420 treatment was smaller than that under N 357 P 90 K 0 W 480 treatments. Then, the N loss under N 357 P 90 K 62 W 420 treatment was smaller than that under N 357 P 90 K 0 W 480 treatments. So, the difference of cotton yield among different treatments in 2013 may be due to the difference of nitrogen loss. The total N loss under N 357 P 90 K 62 W 420 was higher than that under N 357 P 90 K 0 W 480 in 2010, possibly because the absorption of N by cotton was less than in N 357 P 90 K 0 W 480 treatment. The leachate volume per year ranged from 1.48 to 2.67 mm, and it depended on the fertilization rate. There were significant differences in the total leaching amounts among the four treatments ( Figure 2). The leachate volume per year was significantly higher with an annual irrigation amount of 480 mm compared with 420 mm. The leachate volumes also differed significantly with the same amount of irrigation (N 0 P 0 K 0 W 480 and N 357 P 90 K 0 W 480 , and N 357 P 90 K 62 W 420 and N 240 P 65 K 62 W 420 ), possibly because the cotton growth was poor, the root water uptake was decreased, thereby leading to greater transpiration and less leaching in N 0 P 0 K 0 W 480 compared with N 357 P 90 K 0 W 480 in some years [20]. The same reasons could explain the variations under the N 357 P 90 K 62 W 420 and N 240 P 65 K 62 W 420 treatments. The leachate volumes also significantly differed with the same quantity of fertilizer (N 357 P 90 K 0 W 480 and N 357 P 90 K 62 W 420 treatments), which could be attributed to differences in the amount of irrigation or the higher plant growth and water uptake at a higher K rate. Furthermore, this result indicates that irrigation at 480 mm hm −2 may have been excessive. These results are consistent with those obtained by Barton, Wan, and Colmer [22]. Parmodh et al. [25] found that the leaching volumes varied with the different plant time for onion, which due to the depth of the onion roots. These results suggest that proper timing of sowing for vegetables may help reduce leaching. This conclusion may be suitable for vegetables. For the cotton, we should set different sowing times for cotton to verify it. Some previous studies of N leaching found that 30% to 50% of the urea-based N fertilizer applied may be lost and as much as 60% under favorable environmental conditions, and irrigation was recommended to reduce the losses due to volatilization [4,26]. Other studies mainly focused on the nitrogen loss dynamics in the soil after large rainfall events or irrigation [27,28]. By contrast, we investigated the amounts and properties of the leachate water after irrigation. By analyzing the data obtained in several consecutive years, we found that the irrigation of 480 and 420 mm have resulted in leaching. Further research is required to ensure that the cotton yield will not decline by: (1) setting up different total irrigation amount, and adjusting each irrigation amount during the cotton growth according to the potential evapotranspiration; and (2) determining the dynamic changes in nitrogen and ammonia volatilization after irrigation, and understanding the changes in nitrogen in leachate water to identify reasonable water and fertilizer management practices in arid areas where irrigation is required.
For cotton, the amount of water needed to irrigate varies among different regions. Liu et al. (2011) found that the amount of irrigation acts negatively on irrigation water productivity, with the higher irrigation the smaller water efficiency [29]. Thevs et al. (2015) found the amount of cotton water requirement was 500-660 mm·hm −2 for cotton [30], Deng et al. (2013) illustrated the optimal amount of irrigation was 585 mm·hm −2 [31], and Shen et al. (2013) reported 670 mm·hm −2 average water requirement of cotton [32]. For the Kullu area in Southern Xinjiang, the average water requirement of cotton was 480-525 mm·hm −2 . In our study, eight irrigation events were set up during the whole cotton growth period. Thevs et al. (2015) found that irrigation frequency showed no significant effect on cotton yield [30]. Hunsaker et al. (1998) identified increasing irrigation frequency could increase cotton yield [33], but Bordovsky et al. (2012) found reducing irrigation frequency could increase cotton yield [34]. Although a large number of studies have been conducted on cotton, most of them focused on the use of water and fertilizer management to improve cotton yield. There are few studies on whether unreasonable water and fertilizer management could produce leaching, and whether it causes pollution to surface and groundwater. Therefore, long-term monitoring is necessary to collect the subsurface leaching of drip irrigated cotton fields under different water and fertilizer treatments.

Nitrogen Loss Patterns
Abundant irrigation and/or precipitation can greatly increase the leaching of NO 3 − , especially from sandy soils [35]. Parmodh et al. (2012) found that irrigation was the most important factor affecting leaching [25]. In our study, we determined the concentrations of NO 3 − and NH 4 + in the leachate water samples. As expected, the ammonium nitrogen losses in the leachate water remained at relatively low levels (<4 kg hm −2 ) compared with the losses of nitrate nitrogen, which accounted for most of the N leaching losses. Similar results were obtained by Mahboubeh and Mohsen [36] in column experi-ments. The nitrate nitrogen, ammonium nitrogen, and nitrogen losses were highest under N 357 P 90 K 0 W 480 treatment, probably because this treatment employed the highest amounts of irrigation and N fertilizer. The losses varied considerably in the early years, possibly because the soil structure was disturbed when the leaching bucket was installed and the recovery from the soil disturbance was slow. The losses of nitrogen, nitrate nitrogen, and ammonium nitrogen varied under the continuous inputs of nitrogen fertilizer, but they all tended to stabilize in2011, 2012 and 2013, but there was a significant increase of nitrate leaching in 2014 and 2015, probably because of the lower temperature with the weaker transpiration, the less water needed, and the more leaching, vice versa. Nitrate was still the main form of nitrogen lost under all four treatments. However, the total nitrogen loss was greater than the sum of the nitrate nitrogen loss and ammonium nitrogen loss under four treatments, and we only focused on the inorganic nitrogen absorbed and utilized by plants directly, whereas we did not consider the dissolved organic nitrogen. Maybe the differences in organic N are much higher than the differences in ammonium or nitrate or a considerable proportion of urea is leached without being mineralized. Gao et al. studied the dynamic changes in nitrogen leaching under different fertilization treatments and found that nitrate nitrogen (NO 3 − ), dissolved organic nitrogen, and ammonium nitrogen (NH 4 + ) were the main forms of nitrogen in the leachate water, where they accounted for 72.1%, 26.2%, and 1.7% of the total nitrogen leached, respectively [37]. In future research, we will investigate the ammonia losses, which accounted for some of the inorganic nitrogen losses, in order to provide a reference for more reasonable water and fertilizer management. Jariani et al. (2020) found that DON also was the dominant N form in stormwater runoff [38]. In addition, DON has been regarded as a crucial component of the terrestrial nitrogen (N) cycle [39]. This conclusion suggests that in terms of N loss, we should consider not only inorganic N, but also organic N.
Wright reported a strong positive correlation between the concentrations of N and P in the leaves of various species, thereby suggesting that the uptake of N by plants is affected by the availability of P [40]. However, Liu et al. demonstrated that the field N balance between N fertilization inputs and the plant total N uptake was significantly affected only by water management practices and not by the P rate or any treatment interactions [41]. In our experiments, the contents of various forms of nitrogen in the leachate water differed significantly among the treatment sand they were influenced by the availability of P and N as well as by the water management practices. Clearly, the conversion of nitrogen is a complex process in a field environment. Yuan et al. (2017) found N input from atmospheric deposition was significantly increased, with an average annual growth rate of 6.1% [42]. In this study, we only considered the losses of nitrogen in leachate water under different fertilization application and irrigation quantity treatments. The loss coefficients of nitrogen residues in different layers of the soil profile require further investigation. In addition, in order to figure out the nitrogen balance, we must take nitrogen deposition into account. Yuan S and Peng X B found that Larger increase in N input from atmospheric deposition was observed, with an average annual growth rate of 6.1% [42]. Next, we not only focus on the nitrogen from the soil and plant, but also the atmosphere should be included. Soil salt concentration is a serve problem in southern Xinjiang. Wang et al. (2010) found when the soil salt concentration exceeded 0.58%, the cotton seedlings was low and no yield [43]. Therefore, winter irrigation is a widely effective measure to treat soil salinization. The optimal irrigation amount for winter irrigation in northern Xinjiang is still guided by experience rather than by research, and the leaching would occur [44]. This phenomenon also exists in southern Xinjiang. In this study, we only analyze the average leaching amount all the year, next we should (1) make clear that the leaching volumes after each irrigation (2) compare the leaching volume under winter or spring irrigation in order to develop a reasonable irrigation system.
In particular, a long-term field study is required to determine whether the loss coefficients are similar for nitrogen and phosphorus under the same conditions. In addition, phosphate is much less soluble than nitrate, so that P leaching is considerably lower than N leaching and may lead to eutrophication of surface waters. In our study, we have only examined N element although several P gradients have been set, whether P leaching would occur and the amount of P loss under different treatments with the interannual variation is still unclear. So, P leaching is necessary for our next study. This measure would provide important reference for reasonable water and fertilizer management of cotton field in arid area.

Conclusions
Our results showed that leaching would occur in Xinjiang, northwest China, while the leaching volume was less. Under the same amount of N fertilization, the leachate volume increased by 8% to 52% (N 357 P 90 K 0 W 480 and N 357 P 90 K 62 W 420 ). Under the same irrigation amount, the leachate volume decreased by 11% to 29% (N 357 P 90 K 62 W 420 and N 240 P 65 K 62 W 420 ). The loss of total nitrogen under convention fertilization and irrigation was 6.89 kg hm −2 , which was 2.10 kg hm −2 higher than that under optimizing fertilization and irrigation. The nitrogen losses occurred mainly in the form of nitrate nitrogen and the least was lost in the form of ammonium nitrogen. The nitrogen loss coefficients were generally 0.55 higher under the high irrigation and fertilization treatments than that under the low irrigation and fertilization conditions (N 357 P 90 K 0 W 480 and N 240 P 65 K 62 W 420 ). Further studies are needed to monitor the nitrogen residues in the soil and N deposition in long term positioning station in order to estimate the nitrogen balance after nitrogen fertilization.
Author Contributions: H.M. conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, prepared figures and/or tables, authored or reviewed drafts of the paper, and approved the final draft. T.Y. and X.N. conceived and designed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Z.H. and X.M. conceived and designed the experiments, analyzed the data, authored or reviewed drafts of the paper, and approved the final draft. All authors have read and agreed to the published version of the manuscript.