Optimizing N Application for Forage Sorghum to Maximize Yield, Quality, and N Use Efﬁciency While Reducing Environmental Costs

: Investigating the responses of forage crop yield, quality, and nitrogen (N) use efﬁciency to different N application rates is beneﬁcial for guiding proper N fertilization regimes and for reducing reactive N environmental pollution. A ﬁeld experiment was conducted to investigate the effects of different N application rates on above-ground dry matter yield, forage quality, crop N uptake, N use efﬁciency (NUE), and ecosystem economic beneﬁts (EEBs) of forage sorghum cultivated on the Longdong Loess Plateau in 2019 and 2020. Five N application rates were tested, namely 0, 80, 160, 240, and 320 kg · ha − 1 (referred to as N 0 , N 80 , N 160 , N 240 , and N 320 , respectively). The maximum above-ground dry matter yield (22.3 t · ha − 1 in 2019 and 18.0 t · ha − 1 in 2020) was obtained at an N application of 160 kg · ha − 1 . Forage sorghum crude protein (CP) content increased signiﬁcantly with increasing N application rates (the CP content at N 320 was 7.4% and 8.6% in 2019 and 2020, respectively). In contrast, neutral detergent ﬁber (NDF) and acid detergent ﬁber (ADF) were only affected by high N application rates (NDF and ADF were signiﬁcantly higher in N 320 compared with N 0 and N 90 ). The relative feed value (RFV) was signiﬁcantly higher in N 0 compared with N 320 . Crop N uptake was signiﬁcantly higher in N 160 compared with N 0 (25.7% increase to 249.4 kg · ha − 1 in 2019 and 40.5% increase to 247.4 kg · ha − 1 in 2020, respectively). NUE decreased linearly as N rates increased, but NO 3 − –N residue (0–200 cm), reactive N loss (Nr loss), and greenhouse gas (GHG) emissions increased. Private proﬁtability and EEB were the largest at N 160 (private proﬁtability at N 160 was 514.2 USD · ha − 1 , and EEB at N 160 was 392.7 USD · ha − 1 ). Above-ground yield and optimum forage quality must be maximized, while simultaneously safeguarding farmer income and reducing environmental pollution from N fertilizers. Therefore, the optimum N application rate for forage sorghum cultivation in the dry areas of the Loess Plateau is recommended at 160 kg · ha − 1 .


Introduction
Sorghum is the fifth largest food crop in the world. Global sorghum production in 2021 was 6216.7 × 10 4 t, of which about 44% was used as forage, while the Americas and Asia are the major producing regions accounting for 83.8% and 10.9% of global forage sorghum production during 2000-2014, respectively (data source: official website of Food and Agriculture Organization of the United Nations). The forage sorghum industry fulfills two primary functions, namely production and ecology. On the one hand, forage sorghum is an important food source for livestock, and its various nutrients and crude fibers contained are irreplaceable by other feeds and grains. On the other hand, the rotation of sorghum with legumes (e.g., soybean) effectively improves soil fertility and nitrogen (N) stocks and aids in soil and water conservation [1]. Therefore, ensuring that enough forage is available is the key to the high-quality development of the livestock industry. Human

Experimental Design and Management
The experiment was conducted in a randomized complete block design with five N rates of 0, 80, 160, 240, and 320 kg·ha −1 (expressed as N0, N80, N160, N240, and N320). Each treatment was conducted in triplicate. Urea (N, 46%) was applied in two applications, with 30% of the N fertilizer applied as a basal fertilizer and 70% applied at the jointing stage. Superphosphate (P2O5, 16%) and potassium sulfate (K2O, 51%) were used as basal fertilizer for all plots with a single application of 120 kg·ha −1 P2O5 and 150 kg·ha −1 K2O, respectively.
The forage sorghum cultivar 'F10' was used. 'F10' is a promising variety with salinity and drought tolerance. The experimental plot area was 24 m 2 (4 m × 6 m). The planting density was 67,500 plants·ha −1 with a row spacing of 0.5 m and a plant spacing of 0.3 m. In 2019, forage sorghum was sown on 26 May and harvested on 20 October. In 2020, sorghum was sown on 19 April and harvested on 18 September. In both years, ploughing was performed after crop harvest. The predecessor crop to sorghum in 2018 were winter wheat and the two years of experiments were conducted in the same plot. No irrigation or pesticides were used during the experimental period.

Experimental Design and Management
The experiment was conducted in a randomized complete block design with five N rates of 0, 80, 160, 240, and 320 kg·ha −1 (expressed as N 0 , N 80 , N 160 , N 240 , and N 320 ). Each treatment was conducted in triplicate. Urea (N, 46%) was applied in two applications, with 30% of the N fertilizer applied as a basal fertilizer and 70% applied at the jointing stage. Superphosphate (P 2 O 5 , 16%) and potassium sulfate (K 2 O, 51%) were used as basal fertilizer for all plots with a single application of 120 kg·ha −1 P 2 O 5 and 150 kg·ha −1 K 2 O, respectively.
The forage sorghum cultivar 'F10' was used. 'F10' is a promising variety with salinity and drought tolerance. The experimental plot area was 24 m 2 (4 m × 6 m). The planting density was 67,500 plants·ha −1 with a row spacing of 0.5 m and a plant spacing of 0.3 m. In 2019, forage sorghum was sown on 26 May and harvested on 20 October. In 2020, sorghum was sown on 19 April and harvested on 18 September. In both years, ploughing was performed after crop harvest. The predecessor crop to sorghum in 2018 were winter wheat and the two years of experiments were conducted in the same plot. No irrigation or pesticides were used during the experimental period. plants in each plot using a straightedge and vernier caliper. Plant NDVI and LAI were measured using a Green Seeker handheld optical sensing instrument (DELTRAN, Deland, FL, USA) and LAI-2000 (LI-COR, Lincoln, NE, USA). Five plants were randomly selected from each plot at the jointing, heading, flowering, filling, and harvesting stage, and the five plants were subdivided into stems, leaves, and ears, cut into lengths of 3-5 cm and weighed respectively. They were oven-dried at 105 • C for 30 min, then at 75 • C until a constant weight was obtained, and finally weighed again. The forage sorghum was harvested at the milk stage. Fresh yield was measured for each plot and above-ground dry matter yield was obtained after measuring water content. The moisture content of the above-ground portion of forage sorghum at the harvest stage in 2019 and 2020 was 69.9% and 69.4%, respectively. Dry samples were crushed and passed through a 0.425 mm mesh, and N concentration, neutral NDF, ADF, and ASH were determined using the Kjeldahl, Van Soest's, and high-temperature scorching methods [20], respectively. One mixed sample was analyzed for each plot.
Quality indices for animal feed, including dry matter intake (DMI), dry matter digestibility (DMD), and relative feed value (RFV) were calculated using the following formulas [21,22]:

Evaluation of NUE
The N recovery rate (RE) and partial-factor productivity of applied N (PFP) were used to evaluate NUE, and these indicators were calculated as follows [23]: where U N is N uptake (kg·ha −1 ) with N fertilizer, U 0 is N uptake (kg·ha −1 ) without N fertilizer, Y N is above-ground dry matter yield (kg·ha −1 ) with N fertilizer, and N is N fertilizer input (kg N·ha −1 ).

Soil Sample Collection and Measurement
Soil samples were collected between a depth of 0-200 cm at 20 cm intervals on presowing dates, as well as three growth stages (jointing, flowering, and harvesting). Soil samples were collected, and sealed in plastic bags, from two randomly selected locations per plot and stored in a refrigerator until required for analysis. Soil NO 3 − -N content was determined using an automatic discontinuous chemical analyzer (Smart-Chem 450, French). The NO 3 − -N residue (NR) in each soil layer was calculated as follows [24]: where NC i is the soil NO 3 − -N content of the ith soil layer (mg·kg −1 ), BD i is the soil bulk density of the ith soil layer (g·cm −3 ), SD is soil depth (cm), and 0.1 is a conversion factor.

Reactive N Loss and Footprint Calculations
Nr loss and NF were calculated using the following equations [25]: Agronomy 2022, 12, 2969 5 of 20 NF kg N·t −1 = Nr loss/Y N (8) where i represents agricultural input (N fertilizer and other inputs), Rate i is the agricultural materials input application rate, and F i represents the Nr emission factor during the production and transportation of agricultural products. Rate i and F i values for this study are listed in Tables 1 and A1 [26][27][28]. Y N represents the total above-ground dry matter yield with the application of N fertilizer (t·ha −1 ). N 2 O direct -N, NO 3 -N, and NH 3 -N are Nr losses due to N 2 O emissions, NO 3 − leaching, and NH 3 volatilization during N fertilizer application. Nr losses, represented by N 2 O direct -N, NO 3 -N, and NH 3 -N from N fertilizer, were calculated according to fitted models based on previously published reports ( Figure A1) [29][30][31][32][33][34][35][36][37][38]. The models were: where N is the N application rate (kg N·ha −1 ).

GHG Emissions and CF Calculations
GHG emissions and C footprints (CF) were calculated using the following equations [25]: where i represents agricultural input (N fertilizer and other inputs), Rate i is the application rate of agricultural materials input, and G i represents the GHG emission factor during the production and transportation of agricultural products. Rate i and G i values used in this study are listed in Tables 1 and A1 [26][27][28]. The global warming potential of N 2 O is 265 times that of CO 2 on a mass basis [39]. The factor used to convert N 2 O-N to CO 2 was 44/28. N 2 O total -N represents the total N 2 O-N loss from direct and indirect pathways. The indirect N 2 O-N emissions were the sum of 1% NH 3 -N and 2.5% NO 3 -N [40]. Y N represents total above-ground dry matter yield with the application of N fertilizer (t·ha −1 ).

N Fertilizer-Derived Ecosystem Economic Benefits
Accounting for ecosystem and human health costs, the estimated N-derived yield benefits (B Y ), private profitability, and EEB were calculated using the following equations [25]: Private profitability $·ha −1 = B Y -N cos t -L cos t (16) EEB $·ha −1 = B Y -N cos t -L cos t -E cos t -H cos t (17)  where Y N represents total above-ground dry matter yield with the application of N fertilizer, and Y 0 represents the total above-ground dry matter yield without N fertilizer. S Price was 0.25 USD·kg −1 and represents the forage sorghum price for silage [25]. The N cost and L cost are the N fertilizer and labor costs associated with N fertilizer application. They were calculated by multiplying the amount of N or labor applied by the corresponding price (N fertilizer = 0.64 USD·kg −1 ; single-person labor cost = 1.47 USD·h −1 ; data source: http://zdscxx.moa.gov.cn/month/nycsc3/zlsc317#nycsc3/ (accessed on 20 August 2022). E cost represents ecosystem damage costs caused by Nr losses, while H cost represents the human health costs caused by various Nr losses resulting from N fertilizer application. These were calculated as follows [41]: where C GHG , C eu , and C acid represent GHG emission, water eutrophication, and soil acidification damage costs [42]. CO 2 is the total GHG emissions from N fertilizer production, transportation, and application. The CO 2 market price was 0.0204 USD·kg −1 [43]. The eutrophication impact restoration cost for NO 3 -N and NH 3 -N was 1.12 and 0.24 USD·kg −1 , respectively [44]. The soil acidification damage restoration cost for NH 3 -N was 1.87 USD·kg −1 [44,45]. The costs of eutrophication and soil acidification damage per kg of N fertilizer were 0.0018 and 0.021 USD·kg −1 [45], respectively. The human health cost per unit of N 2 O total -N, NO 3 -N, and NH 3 -N were 0.30, 0.20, and 3.30 USD·kg −1 [46], respectively.

Statistical Analysis
Agronomic traits, forage quality, dry matter yield, crop N content, crop N uptake, NUE, NO 3 − -N accumulation, Nr loss, CF, and NF were compared using Analysis of Variance (ANOVA) and the Duncan method (p = 0.05) in SPSS (26.0, SPSS Inc., Chicago, IL, USA). The data in the graphs are expressed as mean ± standard error. GraphPad Prism (9.1.1, GraphPad Software, Inc., San Diego, CA, USA) was used to generate the graphs.

Dynamics of Plant Height, Stem Diameter, LAI, NDVI, and Dry Matter Accumulation
During the forage sorghum growth period, plant height, stem diameter, LAI, and NDVI generally increased at first, whereafter they remained either stable or decreased ( Figure 2). In 2019, forage sorghum plant height, stem diameter, and NDVI were not significantly different between treatments (p > 0.05). In 2020, plant height, stem diameter, and NDVI at N 160 were 3.3 m, 2.3 cm, and 0.8, respectively, which were 7.7%, 4.7%, and 9.3% significantly higher compared with N 0 (p < 0.05), respectively. During the two-year harvest stages, LAI peaked in N 160 with values of 2.7 in 2019 and 2.4 in 2020, respectively, which were 5.1% and 12.3% higher compared with N 0 , respectively (p < 0.05).  Above-ground total dry matter accumulation (TDMA) either gradually increased, or first increased and then decreased, with increasing N application rates under the different growth stages (Figure 3). At the heading, flowering, and filling stages in 2019, aboveground TDMA values in N 160 were 8.6, 11.8, and 18.6 t·ha −1 , respectively, which were 113.6%, 28.8%, and 14.0% significantly higher compared with N 0 (p < 0.05). At the harvest stage in 2019, above-ground TDMA was the highest in N 160 (22.3 t·ha −1 ), but it did not differ significantly from the other N application treatments (p > 0.05). At the heading and flowering stages in 2020, above-ground TDMA values in N 320 increased significantly by 65.3% and 79.8%, respectively, compared with N 0 (p < 0.05). Moreover, N 160 had the highest above-ground TDMA (18.0 t·ha −1 ) at the harvest stage in 2020, which was 23.8% significantly higher compared with N 0 (14.5 t·ha −1 ) (p < 0.05). During the harvest stage in 2019 and 2020, stem and ear dry matter accumulation (DMA) was highest in N 160 (15.6 and 13.2 t·ha −1 for stems; 3.3 and 1.79 t·ha −1 for ears). However, leaf DMA did not differ significantly among all N application treatments during the two years (p > 0.05). and 2020 (f−j). The different lowercase letters in the same parts (leaf, stem, and ear) represent significant difference at p < 0.05 among the different N treatments. Different capital letters represent significant differences at p < 0.05 in the total above−ground parts of dry matter accumulation among the different N treatments. and 2020 (f−j). The different lowercase letters in the same parts (leaf, stem, and ear) represent significant difference at p < 0.05 among the different N treatments. Different capital letters represent significant differences at p < 0.05 in the total above−ground parts of dry matter accumulation among the different N treatments.

NDF, ADF, ASH, CP, and RFV
At the respective harvest stages during the two years, NDF was the highest in N 320 (58.0% in 2019 and 59.7% in 2020) and significantly higher compared with N 0 and N 80 ( Table 2; p < 0.05). ADF was also the highest in N 320 , which was 12.6% and 13.5% significantly higher in 2019 and 2020 compared with N 0 (p < 0.05). N application did not significantly affect ASH in either of the years (p > 0.05). At the harvest stage in 2019 and 2020, CP values did not differ significantly between N 160 , N 240 , and N 320 (p > 0.05), but they were significantly higher compared with N 0 (p < 0.05). During the two-year experiment, there were no significant differences in DMI, DDM, and RFV at N 0 , N 80 , N 160 , and N 240 (p > 0.05). In 2019 and 2020, DMI, DDM, and RFV values were the lowest in N 320 , and were all significantly lower compared with the other treatments (N 0 , N 80 , N 160 , and N 240 ; p > 0.05).
In the same year, different lowercase letters within a column indicate significant differences at p < 0.05 among the different N treatments.

Above-Ground Crop N Uptake
Stem and leaf N content decreased as the growth period progressed (Figure 4). At the harvest stage in 2019 and 2020, stem N content increased overall with increasing N rates, and the stem N content in N 320 increased significantly by 44.9% in 2019 and 40.8% in 2020 compared with N 0 (p < 0.05). Moreover, leaf N content values were the lowest in N 0 at the harvest stage in 2019 and 2020 (26.2 and 28.8 mg·g −1 ) and were significantly lower compared with N 320 (28.1and 31.6 mg·g −1 ). At the harvest stage for both years, total N uptake tended to first increase and then decrease with increasing N application rates.  Different lowercase letters in the same growth stages for the same part (leaf, stem, and ear) represent significant differences at p < 0.05 among the different N treatments. Different capital letters in the same growth stages indicate significant differences at p < 0.05 in total crop N uptake of forage sorghum among the different N treatments.

Soil PFP, RE, and NO3 − -N Residue
In both years, PFP and RE tended to decrease with increasing N rates ( Figure 5). PFP and RE responses towards N rates were fitted to linear equations for both years.  . Different lowercase letters in the same growth stages for the same part (leaf, stem, and ear) represent significant differences at p < 0.05 among the different N treatments. Different capital letters in the same growth stages indicate significant differences at p < 0.05 in total crop N uptake of forage sorghum among the different N treatments.

Soil PFP, RE, and NO 3 − -N Residue
In both years, PFP and RE tended to decrease with increasing N rates ( Figure 5). PFP and RE responses towards N rates were fitted to linear equations for both years. At the jointing stage for both years, NO3 − -N remained at a high level in the top 0-50 cm soil layer but remained at a low level for the 50-200 cm depths (Figure 6). At the flowering and harvesting stages for both years, NO3 − -N tended to leach into deeper soil layers (>50 cm). NO3 − -N residue in the 0-200 cm soil layer was not affected by N rates during the jointing stages in 2019 and 2020 (p > 0.05). NO3 − -N residue at the flowering stages for both years first increased and then decreased with increasing N rates, and NO3 − -N residue was the highest in N240 (470.0 kg·ha −1 in 2019 and 519.9 kg·ha −1 in 2020). At the harvest stages for both years, NO3 − -N residue increased with increasing N rates and peaked at N320 (412.2 kg·ha −1 in 2019 and 307.9 kg·ha −1 in 2020); these values in 2019 and 2020 were 151.3% and 183.2% significantly higher compared with N0, respectively (p < 0.05).

Nr Losses, GHG Emissions, NF, CF, and EEB
The two-year average Nr losses generally increased with increasing N application rates (Figure 7). The Nr losses increased from 7.2 kg N ha −1 in N0 to 119.0 kg N ha −1 in N320, in which about 56.0-70.0% of the Nr losses were derived from NH3 volatilization, followed by NO3 − leaching (21.3-31.1%). Only minor Nr losses were caused by direct N2O emission (4.3-10.0%). Total N2O emission, as well as N production and transportation, were the main contributors to GHG emissions. Among all the treatments, Total N2O emission induced GHG emissions accounted for 48.8-53.8% of the total GHG emissions, while N fertilizer production and transportation induced GHG emissions accounted for 38.5-43.5%.

Plant Growth, Dry Matter Yield, and Nutrient Quality
During the forage sorghum life cycle, as N rates increased, the N-derived benefits initially peaked at N 160 (up to 624.4 USD·ha −1 ) and then decreased; however, the ecological and human health costs continuously increased (Table 3). Private profitability increased with increasing N application rates to 514.2 USD·ha −1 (N 160 ), and then decreased to 56.2 USD·ha −1 (N 240 ) and −55.8 USD·ha −1 (N 320 ). Finally, EEB was similar in trend to private profitability.

Plant Growth, Dry Matter Yield, and Nutrient Quality
Appropriate N fertilizer application significantly increased forage sorghum height and stem diameter [3,47]. Specifically, plant height and stem diameter were significantly higher (7.7% and 9.3%) at N 160 compared with N 0 at the harvest stage in 2020, which agreed well with the results of Afzal et al. [48], who reported that forage sorghum height and stem diameter were 25.2% and 71.4% higher at an N application rate of 57.5 kg N·ha −1 compared with the zero N treatment. Several studies have reported that N fertilizer influences LAI [49]. Our study showed that forage sorghum LAI first increased and then decreased with increasing N application rates at the harvest stage during both years. Moreover, the largest LAI values appeared in N 160 at the harvest stage, which were 5.1% and 12.3% significantly higher compared with N 0 , respectively. This may be due to insufficient plant nutrient supply in N 0 (no N applied), which results in fewer and smaller plant leaves. Additionally, excessive N application results in denser leaves, thereby creating inadequate lighting for the lower and middle plant parts, thus leading to premature leaf wilting and decreased LAI.
Sorghum yield can greatly be increased by utilizing moderate N application rates [50,51], but excessive or insufficient N application rates can reduce yield [14]. When N application levels are too high, excessive leaf growth and poor population ventilation is promoted, and the lower and middle leaves do not receive enough light, which negatively affects photosynthetic product formation and leads to yield reduction. In this study, the forage sorghum N requirement was sufficient at a 160 kg·ha −1 N application rate. In contrast, when the N application rate exceeded 160 kg·ha −1 , the above-ground dry matter yield at the harvest stages decreased. Lower NDF and ADF levels improve food digestibility in ruminants, and thus increases nutrient intake. Therefore, lower NDF and ADF levels exhibit better forage quality. A high N application rate (N 320 ) significantly increased NDF and ADF at the harvest stages. Tang et al. [51] similarly concluded that sorghum NDF and ADF increased by 4.5% and 1.4%, respectively, at a 240 kg·ha −1 N application rate compared with no N application. ASH content at the harvest stage was not affected by N input. Qu et al. [52] and Zhang et al. [53] generated similar results and demonstrated limited effects of N application on sorghum ASH, but a more significant impact from genes. Furthermore, Monti et al. [54] reported that ASH was influenced more by P and K fertilizer. For CP content, the results of this study are similar to previous studies, which concluded that sorghum CP content at the harvest stage significantly increased with increased N application rates [14,51]. Highly desirable quality elements, such as DMI, DDM, and RFV, benefit forage quality and thereby enhance the capacity of livestock to utilize forage nutrients [21,55]. In our study, high N application rates (N 320 ) significantly reduced forage sorghum RFV, whereas low and medium N application rates (N 0 , N 80 , N 160 , and N 240 ) did not produce significant differences.

Crop N Uptake, NUE, and NO 3 − -N Residue
An appropriate N application rate can improve crop production capacity and promote plant N uptake. During both growing seasons in this study, forage sorghum N uptake at the harvest stages was significantly higher in N treatments compared with N 0 . This result is consistent with other studies [56,57]. N fertilization may increase crop N uptake by stimulating root growth. Furthermore, the crop N uptake in different N rates at the harvest stage was higher than the amount of N fertilizer we applied. On the one hand, a considerable part of the N absorbed by crops comes from the mineralized N, and the application of N fertilizer increases the soil N source, which can improve soil physicochemical properties and increase the number and activity of soil microorganisms, thus promoting the mineralization of soil organic N [58]. On the other hand, global N deposition has continued to increase in the last hundred years, and the rate of increase is rising [59]. He et al. found that total airborne N inputs to a maize-wheat rotation system on the North China ranged from 99 to 117 kg N ha −1 yr −1 [60]. Therefore, crop N uptake can be higher than the applied N fertilizer. N application increased stem N uptake at the harvest stages, which was similar to previous studies. For example, Cosentino et al. [61] showed that the forage sorghum stem N uptake increased by 69.6%, 58.5%, and 67.0% at 60, 120, and 180 kg·ha −1 N application rates, respectively, compared with the zero N treatment. Abunyewa et al. [62] also found similar results.
PFP and RE are common indicators that express crop NUE in different ways. In this study, PFP and RE decreased significantly with increasing N application. Ju [63] and Abunyewa et al. [62] showed that excessive N application resulted in reduced plant yield increase. This is possibly due to the fact that a high N fertilizer application depresses plant root growth, and the subsequent reduction in root length and absorption area affects root system nutrient uptake rates [64,65].
Soil mineral N content gradually increases with increasing N fertilizer application [66], and NO 3 − -N is the main N form present in the soil, which is used by crops. In both growing seasons, NO 3 − -N residue in the 0-200 m soil layer increased significantly with increasing N application. Scordia et al. [38] similarly concluded that N application rates of 120 and 240 kg·ha −1 significantly increased soil NO 3 − -N residue. Wang et al. [67] also showed that soil NO 3 − -N residue was highly correlated with N application rate. In this study, NO 3 − -N residue in N 320 at the harvest stage in 2020 increased by 74.1% compared with N 160 . This indicates that a two-year successive application of excessive N fertilizer leads to significant NO 3 − -N residue in deeper soil layers. Ju et al. [68] showed that 20.9-48.4% of N fertilizer remains in the soil after the crop is harvested. Since farmers tend to apply large amounts of N fertilizer every year, a large amount of NO 3 − -N remains in the soil profile after crop harvest, and if heavy rains are encountered, NO 3 − -N can move deeper. This NO 3 − -N is unavailable for crop use, ultimately entering the groundwater and atmosphere through leaching, nitrification, and denitrification, thereby causing potential environmental risks [67,69]. The use of cover crops and controlled-release urea can be promising measures to increase crop NUE and reduce the N losses to the environment [70]. In our study, NO 3 − -N residue in 2020 was lower than in 2019, this may be caused by higher rainfall in 2019 than in 2020. High intensity precipitation can cause NO 3 − -N residue in the soil to leach deeper into the soil, resulting in ineffective NO 3 − -N residue below the root zone, which cannot be detected [71]. In addition, NO 3 − -N residue in the 0-200 cm soil layer at harvest stage was higher than the amount of N fertilizer applied in 2019 and 2020. This is understandable since the in-season N fertilizer is not the sole source of NO 3 − -N residue in soil, the NO 3 − -N residue before sowing, mineralization of soil organic N, and increasing global N deposition in agroecosystems also made a big contribution

Nr losses, Private Profitability, and EEB
Increased N fertilizer application has played an irreplaceable role in improving crop yields, as well as the economic benefits for farmers. However, the overuse of N fertilizer has resulted in increased Nr losses, which has negatively impacted the global N cycle and has caused numerous environmental problems. To achieve sustainable land use, farmland Nr losses must be understood. NO 3 − leaching, N 2 O emission, and NH 3 volatilization are the main Nr loss pathways from farmland [72]. In this study, NO 3 − leaching, N 2 O emission, and NH 3 volatilization increased with increasing N application. Zhang et al. [25] and Yao et al. [73] reached similar conclusions: Nr losses increased with increasing N application. Therefore, a suitable N fertilizer application rate must be selected to prevent major environmental problems.
EEB and private profitability reached a maximum value at N 160 . Therefore, when accounting for dry matter yield, a 160 kg·ha −1 N application rate can maximize forage sorghum above-ground dry matter yield while maintaining a high private profitability and EEB. This study also demonstrated that using common urea (CU) results in higher ecological and health costs. Zhang et al. [25] showed that the ecological and health costs of controlled-release urea and urea blends (BU) were 14.7% and 20.1%, respectively, lower than CU. Moreover, BU can be applied at once, which reduces the labor cost by half. Therefore, widespread BU use can be considered in future agricultural production.
Both in-season N fertilizer application and soil basic fertility condition before sowing have notable influences on yield performance and Nr losses, thus influencing private profitability and EEB. This two-year N application study was performed on a typical cropland with history of farmer conventional fertilization and management. The basic fertility of the cropland used in this study has good representatives of soil fertility condition in the area. So the yield response to fertilizer N application and Nr losses, private profitability, and EEB can possibly vary significantly when the same study is performed on sites with different soil basic fertilities. This two-year successive study was conducted on the same typical cropland to investigate the cumulative effect of two seasons of N application on yield performance, quality, and N-use efficiency in forage sorghum production. This is essentially important in optimizing N application for forage sorghum to maximize yield, quality, and N use efficiency while reducing environmental costs.

Conclusions
This study indicated that forage sorghum above-ground dry matter yield and N uptake initially increased and then decreased with increasing N application (both reached a maximum value at the 160 kg·ha −1 N rate, with a two-year average of 20.1 t·ha −1 and 248.4 kg·ha −1 , respectively). Similarly, forage sorghum CP content at the 160 kg·ha −1 N rate was significantly higher compared with the zero-N treatment, but NDF, ADF, and RFV were not significantly different. Nr losses, GHG emission, NF, and CF continuously increased with increasing N application. Private profitability and EEB were maximized at the 160 kg·ha −1 N rate. Simply increasing the N fertilizer amount will not increase yield, but will indeed raise the economic costs for farmers and negatively impact the environment. Therefore, we propose that 160 kg·ha −1 is the ideal N application rate for forage sorghum in the dry region of the Loess Plateau. This will ensure optimum yield and forage quality, as well as improve farmer income and reduce environmental pollution.

Acknowledgments:
We would like to thank the State Key Laboratory of Grassland Agro-ecosystems, Lanzhou University for the use of its facility. And we are also very appreciative of Juncheng Li's help in field sampling.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1. Nr losses and GHG emission for production and transportation of various agricultural inputs in the forage sorghum production system.