Estimation of Nitrogen Supply for Summer Maize Production through a Long-Term Field Trial in China

: Supplying adequate nitrogen (N) to meet crop demand is critical for enhancing agricultural sustainability. Not only fertilizer N, but also N from other available sources should be considered in N supply capacity. We conducted a 10-year farming experiment using a split-plot design with two different main fertilizer management approaches and three N application strategies as add-on sub-treatments. Based on the experiment, we estimated the total N supply (TN supply ) for the summer maize cropping system, through considering environmental, soil, crop residue, and fertilizer N sources. An appropriate TN supply was established by correlating TN supply with the relative yield (RY), N input and output, and N use efﬁciency (NUE). The results revealed a wide variation in TN supply (from 88 to 755 kg ha − 1 ). The RY, N input, and N output ﬁtted well to TN supply using linear-plateau, linear, and linear-plateau models, respectively. The lower limits of TN supply for achieving the maximum RY and N output were 361 and 358 kg ha − 1 , respectively. The relationship between N input and N output was described as linear-plateau. We determined the slope of the linear curve (55.4%) as the lower limit of NUE, beyond which the upper limit of TN supply was determined to be less than 497 kg ha − 1 . Thus, appropriate TN supply values ranged from 325 to 497 kg ha − 1 for summer maize production, which could ensure enough N supply for higher yields and avoid excessive N input for higher NUE and lower environmental N loss. Our ﬁndings highlight that TN supply can be an alternative indicator for evaluating N management.


Introduction
Approximately 30% of synthetic nitrogen (N) fertilizers produced globally are used in China, with an N use efficiency (NUE) of 25%, compared to 52% in European countries and 68% in the United States of America [1,2]. Excessive application rate and low use efficiency of synthetic N fertilizers have caused serious environmental problems, such as nitrous gases emissions, soil nitrate accumulation, and groundwater pollution [3][4][5]. It is not realistic and is problematic to expect higher crop yields by increasing N fertilizer application rates.
Maize (Zea mays L.) is one of the leading grain crops for global food security [6,7]. In China, a summer maize-winter wheat rotation is a highly intensified cropping system [8], and the total sown area of summer maize (23.7 million hectares) accounts for 57.5% and 24.2% of the whole maize and cereal crops, respectively [9]. Maize production in China is dominated by small-scale farms with high N fertilizer inputs (200-300 kg N ha −1 ), The long-term experiment used a split-plot design, as reported by Huang et al. [16]. The main treatments contained Nutrient Expert (NE), a Nutrient Decision Support System (NDSS) [19] that combines 4R (right source, right place, right time, and right rate) nutrient management together with improved varieties and optimized plant density, and Farmers' Practice (FP), which relies on the field managing practices of local farmers. Three different N addition strategies were used in subplots of each of the main plots, characterized by their treatments with different TNsupply levels, as follows: (1) 0N (zero additional N was introduced).
(2) 2N (N fertilizers were introduced every two out of three years, with no fertilizer application during the first year).
(3) 3N (fertilizer was applied every year; in this case, one cycle lasted three years). The long-term experiment used a split-plot design, as reported by Huang et al. [16]. The main treatments contained Nutrient Expert (NE), a Nutrient Decision Support System (NDSS) [19] that combines 4R (right source, right place, right time, and right rate) nutrient management together with improved varieties and optimized plant density, and Farmers' Practice (FP), which relies on the field managing practices of local farmers. Three different N addition strategies were used in subplots of each of the main plots, characterized by their treatments with different TN supply levels, as follows: (1) 0N (zero additional N was introduced).
(2) 2N (N fertilizers were introduced every two out of three years, with no fertilizer application during the first year).
(3) 3N (fertilizer was applied every year; in this case, one cycle lasted three years). P and K applications were consistent in all three subplots. Thus, six treatments were marked as NE 0N , NE 2N , NE 3N , FP 0N , FP 2N , and FP 3N . The FP fertilizer application rates were calculated using the data obtained for over 100 farmers collected in 2009-2012. A fixed rate was applied from 2013 to 2019 due to the widespread use of compound fertilizers (shown in Supplementary Table S1). We relied on the NDSS to determine the fertilizer rates for the NE treatments (Supplementary Table S1). The system was used to recommend location-specific fertilizer applications with the 4R strategy to help to meet the nutrient demand at different stages of crops and in combination with other optimal agronomic practices [19,25]. For maize, similar fertilizer application methods were applied in both NE and FP: 1/2 of the N, total P, and total K were applied as basal fertilizers and the remaining 1/2 of the N was top-dressed at the V12 stage [26] (lasting from approximately 25 July to 5 August each year). All fertilizers were applied and banded for both basal and top-dressing. We used ZhengDan958 maize variety for FP and NE planted at 60,000 ha −1 density in 2009-2019 and in 2009-2013. In 2014-2019, XianYu335 with higher yield potential was planted for NE treatments at a density of 75,000 ha −1 (Supplementary Table S1).

Sampling and Measurements
At wheat harvest, the N concentration of crop residue was determined by harvesting three 2 m 2 areas in the middle of each plot. Three rows in the middle of each plot were manually harvested. The grains were dried to determine the yield (moisture content of 14%). Five succussive plants located in one spot were sampled to determine the nutrient concentrations. Before sowing and after harvesting, the soil was sampled at 0-5, 5-10, 10-20, 20-30, 30-60, and 60-100 cm layers, after which mineral N content was analyzed. The stock, foliage, and grains were separated and dried at 70 • C until no weight change was observed. The total N contents were obtained by the Kjeldahl automated method after the samples were subjected to wet digestion by the H 2 SO 4 and H 2 O 2 mixture [27]. For the residual mineral N content analysis, the fresh soil was extracted using 1 M KCl [28], and then analyzed by a high-resolution AA3 instrument (manufactured by Bran + Luebbe, Norderstedt, Germany).

Agronomic Indices of Nitrogen Use Efficiency
Four indicators of NUE were calculated as follows [29]: Agronomic efficiency of N (AE N , kg kg −1 ) = (N yield − 0N yield )/N rate , where N up and 0N up indicate N uptake from 3N and 0N treatments in NE or FP plots (kg ha −1 ), N yield and 0N yield indicate grain yields from 3N and 0N treatments in NE or FP plots (kg ha −1 ) respectively, and N rate were the corresponding N application rates in NE or FP plots (kg ha −1 ).

Total Nitrogen Supply and Nitrogen Budgets
To narrow the variations of the maize yield caused by climate differences during the 2000-2019 period, the relative grain yields (RY) were obtained using the formula below [16,30]: where Y treatment and Y max are normal and maximum grain yields of the treatments in the same year (in kg ha −1 ). The TN supply (kg ha −1 ) was calculated as follows: TN supply = N soil + N env + N straw + N fert , (6) where N soil is the mineral N content in the 0-1 m soil layer before planting, N env is the environmental N from irrigation and precipitation during the maize growing season, N straw is the returned-straw N from the preceding wheat crop (also equal to the N in the aboveground wheat straw), and N fert is the N fertilizer applied during the maize growing. N soil is the theoretical end-product of mineralization and immobilization of soil organic matter, environmental N, crop residue, or previously applied fertilizer [14,31], and TN supply and N soil comprise the residual N from the previous seasons. NUE (in %), a popular indicator, was used to assess the crop system N efficiency, which was established by the EU Nitrogen Expert Panel [32]. In this case, crop uptake was the only N output, while inorganic, organic, and soil mineral N were the only inputs [15,16,33].
N output = N uptake , N input = N env + N straw + N fert + ∆N soil , (9) where N uptake is the N content in the maize grain and straw (in kg ha −1 ) and ∆N soil is the change in soil mineral nitrogen content, which is the difference between residual soil N content in the samples collected before maize sowing and after maize harvesting (in kg ha −1 ). RY, N output, and N input were plotted vs. TN supply using SAS software (1993). Linear and linear-plateau models were used to fit the corresponding data [34]. The variance analyzed differences among the obtained data and the least significant difference (LSD 0.05 ) tests via SPSS version 19.0 software (SPSS Inc., Chicago, IL, USA).

Grain Yield and Agronomic Indices of NUE
There were no significant differences in grain yield between the NE 3N and FP 3N treatments (Table 1). NE management could maintain a high crop yield with reduced fertilizer N application. Compared with FP, NE increased the recovery efficiencies of the N applied by 10.3 percentage points, and improved agronomic efficiency, physiological efficiency, and partial productivity of the N applied by 54.8%, 10.3%, and 28.8%, respectively ( Table 2).

Relationships between Total N Supply and Relative Yields, N Output, and Input
The best fit of the relationship between TNsupply and RY was obtained using a linearplateau model (shown in Figure 2a). The minimum TNsupply capable of providing the maximum RY (equal to 95.0%) was 361 kg ha −1 . After that, the RY increased linearly (with a slope equal to 0.084). The relationship between TNsupply and N input was also linear (Figure 2b). Moreover, 63.4% (the slope) of the TNsupply was added to the system, while 36.6% was predicted to remain. The best fit of the N output plotted vs. TNsupply was obtained by the linearplateau model (see Figure 2c). The minimum TNsupply level needed for maximum N output was 358 kg ha −1 , and the maximum N output (maize N uptake) was 164 kg ha −1 .

N Input vs. N Output and NUE
The N input and output demonstrated a linear-plateau relationship (Figure 3). The minimum input that could achieve the maximum N output (161 kg ha −1 , similar to the values observed in Figure 2c) was 160 kg ha −1 . Using the NUE conceptual framework developed by the EU Nitrogen Expert Panel, we added the upper and lower limits of NUE in Figure 3. The slope of the linear curve was determined to be the lower limit of NUE (55.4%), and the upper limit was 90.0%. We added two auxiliary lines that represented those NUE values in Figure 3. The points where the NUE lines intersected the plateau line were the thresholds of N input within the NUE targets (55.4-90.0%), which were 179-291 kg ha −1 . Considering that N input tendency on TN supply could be expressed as y = 0.634x − 24.3 (see Figure 2b), the TN supply ranged from 321 to 497 kg ha −1 , within which the system would achieve a sustainable NUE value.

N Input vs. N Output and NUE
The N input and output demonstrated a linear-plateau relationship (Figure 3). The minimum input that could achieve the maximum N output (161 kg ha −1 , similar to the values observed in Figure 2c) was 160 kg ha −1 . Using the NUE conceptual framework developed by the EU Nitrogen Expert Panel, we added the upper and lower limits of NUE in Figure 3. The slope of the linear curve was determined to be the lower limit of NUE (55.4%), and the upper limit was 90.0%. We added two auxiliary lines that represented those NUE values in Figure 3. The points where the NUE lines intersected the plateau line were the thresholds of N input within the NUE targets (55.4-90.0%), which were 179-291 kg ha −1 . Considering that N input tendency on TNsupply could be expressed as y = 0.634x − 24.3 (see Figure 2b), the TNsupply ranged from 321 to 497 kg ha −1 , within which the system would achieve a sustainable NUE value.

Suitable Total N Supply
Based on the correlation between the TNsupply and the RY, N input, and N output, the suitable TNsupply level that could maintain grain yields and achieve increased NUE with reduced environmental N loss was 361-497 kg ha −1 (Figure 4).

Suitable Total N Supply
Based on the correlation between the TN supply and the RY, N input, and N output, the suitable TN supply level that could maintain grain yields and achieve increased NUE with reduced environmental N loss was 361-497 kg ha −1 (Figure 4).

Different Sources of N
Different sources of N can provide bioavailable N for crop uptake [35,36], and to- The minimum TN supply level should be 361 kg ha -1 (ensuring N supply for high yield), and the maximum TN supply level should be 497 kg ha -1 (obtaining the higher NUE and less N loss to the environment).

Different Sources of N
Different sources of N can provide bioavailable N for crop uptake [35,36], and together, these sources constitute the total N supply. When all sources of N are fully considered, achieving NUE together with sustainable development can be accomplished.
In north-central China, smallholder farmers often apply excessive amounts of N fertilizer to ensure high yields, and the overuse of N fertilizer leads to lower NUEs and higher mineral N accumulation levels in the soil [5,37]. In this study, the average RE N (28.5%), AE N (6.2 kg kg −1 ), PE N (22.3 kg kg −1 ), and PFP N (31.9 kg kg −1 ) of FP treatment were below the common values, which were 30-60%, 10-30 kg kg −1 , 30-50 kg kg −1 , and 40-70 kg kg −1 , respectively [29]. Our data also showed that much greater mineral N accumulation occurred within the 0-100 cm soil profile under the FP 3N treatment (168-448 kg ha −1 ) compared with the other treatments, which agrees with Lu et al. [11]. The soil N accumulation level was strongly correlated with nitrate leaching and is a valuable indicator of leaching tendency [4,38]. Moreover, the study region is characterized by sudden heavy rains and irrigation events (including flood irrigation) during the summer maize growing season, increasing the risk of N leaching [13,39]. Thus, any accumulated deposits of mineral N need to be depleted to reduce leaching, and the most effective method for achieving this is absorption by crops [40,41]. Thus, the cumulative soil N is an important N source.
Precipitation and irrigation are other N sources. Our field data showed that maize received a mean of 28 kg N ha −1 from these sources during the growing period. Researchers have reported that increasing amounts of N deposition occurred in China (from 13.2 kg N ha −1 in the 1980s to 21.1 kg N ha −1 in the 2000s) [42], and north-central China is a hotspot, as it currently has the greatest N deposition rate of 52.2 kg N ha −1 year −1 [22]. Returned N from crop residue can be utilized after decomposition by soil microorganisms. Returned crop residue N accounts for 37.4% of the total N needed for maize in China [24]. However, few studies have calculated crop residue N when recommending N fertilizer application rates.
N soil represents the end-product of mineralization and immobilization of soil organic matter, crop residue, environmental N, or previously applied fertilizer. Nitrogen accessibility depends on N mineralization and immobilization after N enters the soil, which can strongly affect N management [14,43]. Therefore, the TN supply also includes N remaining from the previous growing season, in combination with N soil .

Essence of the Suitable Total N Supply
The TN supply is a representation of all the sources of N available in a cropping system and an indicator of the N supply capacity. In this study, the TN supply provides an improved way for better understanding crop system N supply capacity. Correlations between the TN supply and RY, N input, and N output are shown in Figure 4. The TN supply is a mixture of environmental, soil, crop residue, and fertilizer N, and indicates the support capacity for the system. With TN supply values above 361 kg ha −1 , the RY values were relatively high. Additionally, we observed a linear-plateau dependency between RY and TN supply . In addition, 63.4% of the TN supply (as N input) entered the crops. Additionally, not all N input was assimilated by the crops: some proportion was lost or converted to the organic N pool, which affected the NUE. The N input/output linear-plateau correlation with the 55.4% linear slope was the lower-limit target NUE. We then estimated the sustainable TN supply to be 325-497 kg ha −1 .
The EU Nitrogen Expert Panel suggested a NUE target value of 50-90% [32]. In fact, the compositions of the N supply in China and Europe were different. In Europe, manure is used more, which is different from that of the current maize system in China [44]. The lower-limit target NUE should be higher in the current system due to the lower N availability in manure than in inorganic fertilizer. Zhang et al. [15] also reported that the optimal NUE in the winter wheat-summer maize rotation system should be 60% according to a similar quantification method of NUE. In this study, at N input below 160 kg ha −1 , the NUE was stably maintained at 55.4% (the slope), except for the intercept term. As N input increases, the NUE gradually decreases [16]. Therefore, it is acceptable that 55.4% is used as the lower-limit NUE.

N Fertilizer Application Recommendations under Suitable N Supply Levels
To achieve a balanced N supply and crop demand, the TN supply indicator should be used to manage N balance. However, full use of this indicator to provide recommendations is somewhat challenging.
All four N sources should be considered when the TN supply is used to make fertilizer recommendations. Residual N accumulation should be above 100 kg ha −1 (the soil buffer capacity level) in summer maize-winter wheat rotations, over which the nitrate leaching increases exponentially [10,16,25,38]. This value could be a recommended one for mineral N contents during N management in soils. Moreover, the environmental N should be 28 kg ha −1 , which is close to the value reported by Liu et al. [22] in the same region. Additionally, the previous wheat residue N content ranged from 2.9 to 143 kg ha −1 [45]. The specific value was influenced by the N delivery conditions [25,30,41]. We chose the upper quartile (48 kg ha −1 ) and the median (37 kg ha −1 ) of 2142 observations for the wheat residue N contents under excessive and suitable supply conditions based on suggestions provided by Chuan et al. [45]. Based on the above information and the appropriate TN supply level, the fertilization recommendation was determined under different N supply conditions by measuring N soil levels, which are shown in Figure 5 and explained below:

Conclusions
The TNsupply, including different N sources, is a robust indicator that can be used for synchronizing N supply and crop demand, which is valuable for guiding the scientific application of N fertilizer in the summer maize cropping system. In this study, the RY, N input, and N output plotted as functions of TNsupply values were fitted using linear-plateau, linear, and linear-plateau models, respectively. The suitable TNsupply ranged from 361 to 497 kg ha −1 , within which the crop can achieve higher grain yields, higher NUE, and less environmental N loss. Considering the high amounts of N accumulation in the soils of north-central China, sustainable N management should be considered, such as soil N, environmental N input, crop residue, and the minimum fertilizer applications, with regards to N for crop demand, and the residual soil N should be at a safe level. This research would provide a guideline for crop management involving crop residue post-utilization, improvements of soil quality, and environmental considerations.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1: Variety, seed density, and fertilizer rate of summer maize in NE and FP treatments. (i) Excess supply condition: The N soil plus N straw (48 kg ha −1 ) and N env (28 kg ha −1 ) surpassed the minimum TN supply requirements (equal to 361 kg ha −1 ). Thus, N soil needs to be above 285 kg ha −1 . Under this condition, N fert should be zero, and the priority of the N management strategy is to deplete the accumulated residual N by crops.
(ii) Conventional supply condition: The N soil ranged from 100 to 285 kg ha −1 , and the N straw should be 48 kg ha −1 under this situation. The optimal N fert rate should equal to the minimum TN supply (361 kg ha −1 ) minus the N straw , N env , and N soil . In addition, the maximum N fert rate cannot exceed the maximum TN supply (497 kg ha −1 ) minus the N straw , N env , and N soil . Under these conditions, the goal of the N management strategy is to reduce the accumulated residual N to benign contents to maintain sufficient N.
(iii) Optimal supply condition: The N soil level was below 100 kg ha −1 , and the N straw level should be 37 kg ha −1 under this situation. The optimal N fert should equal the minimum TN supply (361 kg ha −1 ) minus the N straw , N env , and N soil . In addition, the maximum N fert rate cannot exceed the maximum TN supply (497 kg ha −1 ) minus the N straw , N env , and N soil . Thus, this is the minimum amount of synthetic N fertilizer to achieve high yields and high NUE with low soil N contents.

Conclusions
The TN supply , including different N sources, is a robust indicator that can be used for synchronizing N supply and crop demand, which is valuable for guiding the scientific application of N fertilizer in the summer maize cropping system. In this study, the RY, N input, and N output plotted as functions of TN supply values were fitted using linear-plateau, linear, and linear-plateau models, respectively. The suitable TN supply ranged from 361 to 497 kg ha −1 , within which the crop can achieve higher grain yields, higher NUE, and less environmental N loss. Considering the high amounts of N accumulation in the soils of north-central China, sustainable N management should be considered, such as soil N, environmental N input, crop residue, and the minimum fertilizer applications, with regards to N for crop demand, and the residual soil N should be at a safe level. This research would provide a guideline for crop management involving crop residue post-utilization, improvements of soil quality, and environmental considerations.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/agronomy11071358/s1, Table S1: Variety, seed density, and fertilizer rate of summer maize in NE and FP treatments.