Legume/Maize Intercropping and N Application for Improved Yield, Quality, and Water and N Utilisation in Forage Crops under Arid Conditions in Northwest China


 Aim

Legume/maize intercropping systems have been used in many developing countries. However, the effects of such systems on crop yield, quality, water-use efficiency (WUE), and N-use efficiency (NUE) were differed with the use of different legume crops and fertilizer application rates.
Methods

In the present study, field experiments were carried out in arid areas of Northwest China from 2019 to 2020 with three planting patterns (LM: lablab bean/silage maize intercropping; FM: forage soybean/silage maize intercropping; M: silage maize monoculture) and four N application levels (N1: 0 kg ha−1; N2: 120 kg ha−1; N3: 240 kg ha−1; N4: 360 kg ha−1).
Results

Compared with N1, the N3 and N4 treatments significantly increased fresh and hay grass yield, crude protein yield, crude protein content, and crude fat content of silage maize, legumes, and the whole system, and decreased the content of neutral detergent fibre (NDF) and acid detergent fibre (ADF). Compared with N1, the 2-year average total hay yield of N3 and N4 increased by 60.38% and 56.45%, respectively, and the total crude protein yield increased by 106.71% and 100.00%, respectively. Compared with N1, the N3 and N4 treatments significantly increased WUEB, N content, N uptake, and NUE, and the 2-year average NUE of N3 was 59.52% higher than that of N4. Compared with M, LM and FM increased crude protein content and decreased NDF and ADF content, and the forage quality of LM was higher than that of FM.
Conclusions

Compared with M, LM and FM increased hay yield by 3.70% and 1.72%, crude protein yield by 32.05% and 22.82%, and WUEB by 10.49% and 6.02%, respectively. Among all treatments, LM-N3 had the highest total fresh and hay yield, crude protein yield, N uptake, WUEB, and NUE in two years.


Introduction
The global population is expected to continue to grow, which will result in a signi cant increase in food, feed, and fuel demand by the middle of the 21st century (Raman et al., 2018; Chen et al., 2014). With the rapid development of animal husbandry in China, feed shortage is increasing (Li et al., 2015). Therefore, it is urgent to accelerate the development of the feed production industry in China (Fang, 2018). Maize (Zea mays L.), as a major crop in China, has a wide range of uses, including as food, feed, industrial raw materials, and bioenergy (Chen et al., 2014). With the development of green agriculture, maize will play an increasingly important role in crop production in China (Cui et al., 2018). Therefore, it is important to develop e cient In addition to intercropping, N application is an important factor affecting forage crop yield, quality, and water and N utilisation. At present, the amount of N fertilizer applied per unit area in China has far exceeded the However, applying N fertilizer beyond that needed to meet the N demand of crops will not increase crop yield and N absorption but will reduce NUE (Cui et al., 2010;Li et al., 2016). Xu et al. (2019) reported that a higher N application rate can increase maize yield but may lead to reduced NUE and WUE. Given these previous research results, there remain critical gaps in our understanding regarding the effects of N application on maize yield, WUE, and NUE. Furthermore, information is lacking regarding the arid region of Northwest China. Therefore, the appropriate N application rate for silage maize needs to be elucidated. The objectives of this study are three-fold: (a) can the maize/legumes intercropping increase the biological yield compared with maize monoculture? (b) Between lablab bean and forage bean, which variety will help maize achieve a higher yield in an intercropping system? (c) what is the appropriate nitrogen application rate for the maize-legume intercropping system? We hypothesized that the lablab bean/maize intercropping could increase forage yield and improve quality, WUE, and NUE under appropriate nitrogen application conditions.

Site description
This experiment was carried out in the Linze Grassland Agricultural Experiment Station, State Key Laboratory of Grassland Agroecosystem, Lanzhou University (Fig. 1). The area is located in Linze County, Zhangye City, Gansu Province (100°02′E, 39°15′N). According to the Köppen classi cation (Chen et al., 2013), The region has the three-levels climate classi cation of BSk (Dry-Dry Summer-Cold arid) and lies at an altitude of 1390 m. The annual average temperature was 8.94 ℃ over 30 years , the average annual potential evaporation is 2337.6 mm, and the average annual rainfall is 113.6 mm over 30 years. The rainfall is mainly concentrated in summer and autumn, accounting for more than 60% of the total annual precipitation. In 2019, the annual precipitation of the test station was higher than usual at 167.7 mm (Fig. 2a). The annual precipitation in 2020 was normal at 122.1 mm (Fig. 2b). In 2019, we analysed the soil in the 0-20-cm soil layer before sowing. Soil pH was 7.85, soil bulk density was 1.22 g cm −3 , soil organic matter was 10.62 g kg −1 , soil total N was 0.79 g kg −1 , soil total P was 0.86 g kg −1 , soil total K was 0.53 g kg −1 , available N was 61.6 mg kg −1 , available P was 47.9 mg kg −1 , and available K was 166.7 mg kg −1 .

Field management and research design
A randomized block experimental design was used to set up three planting patterns: lablab bean/silage maize intercropping (LM), forage bean/silage maize intercropping (FM), and silage maize monoculture (M). Four N application levels were set for each planting pattern: no N application (N1: 0 kg ha −1 ), low N application (N2: 120 kg ha −1 ), medium N application (N3: 240 kg ha −1 ), and high N application (N4: 360 kg ha −1 ). There were 12 treatments in the experiment, and each treatment had three replicates. There were 36 plots in the experiment, and the plot area was 77.0 m 2 (length × width: 7.7 m × 10 m). A 1.2-m-wide isolation belt was set to prevent water leakage between communities. Silage maize was sown on 26 April 2019 and 2 May 2020 according to the planting density of 85,000 plants ha −1 (the spacing was 21cm). The maize was sown in wide and narrow rows with a spacing of 70 cm and 40 cm. Two forage bean or lablab bean seeds were planted in one planting spot between two maize plants. In the N1 treatment, 138 kg ha −1 of phosphate fertilizer was applied before sowing; in the N2 treatment, 138 kg ha −1 of phosphate fertilizer and 120 kg ha −1 of N fertilizer were applied before sowing; in the N3 and N4 treatments, the basal fertilizer was the same as that in N2, and 120 kg ha −1 of N fertilizer was applied at the jointing stage in the N3 treatment, and 120 kg ha −1 of N fertilizer was applied at the six-and 12-leaf stages in the N4 treatment. The irrigation amount of each treatment was 400 mm, and 50% irrigation was conducted at the jointing (25 June) and silking (30 July) stages in 2 years.

Sampling and measurements
2.3.1. Hay yield and crude protein yield In the maize harvest period, a 6-m 2 quadrat was randomly selected in each plot, and the fresh weight of legume plants and silage maize was weighed to calculate the fresh grass yield of maize and legume crops. After that, the legume and maize were oven dried at 65℃ for more than 48 hours to a constant weight. Then, the dry weight was measured to calculate the hay yield of legumes and maize. The CP yield was calculated according to the hay yield and CP content of legumes and maize. The formula is shown below.

Forage quality
The N content of legumes and maize was determined by the Micro Kjeldahl method, and the CP content and CP yield (Y C ) were determined based on the following formulae (Bagheri et al., 2015): The crude fat content of legumes and maize was determined by a Near Infrared Analyzer (Foss-Nirsd 2500, FOSS, Denmark). Acid detergent bre (ADF) and neutral detergent bre (NDF) were determined using the

Soil water content and soil water storage
In the sowing and harvest periods, three points were randomly selected from each plot to drill for soil samples between two maize plants in the same row. We sampled the soil every 20 cm in the 0-200 cm soil layer and stored it in an aluminium box. Samples were over dried at 105 ℃ for more than 48 hours to a constant weight, and then weighed to calculate the soil water content (SWC) and SWS. The formulae are as follows: 3 4 where W is the weight of wet soil (g), D is the weight of dry soil (g); h i (cm) is the depth of soil layer i, ρ i (g cm −3 ) is the bulk density of soil layer i, b i (%) is the soil mass moisture content of soil layer i, and n is the number of soil layers. where ET is the evapotranspiration, P is the precipitation (mm), I is the irrigation amount (mm), SWS S is the SWS before sowing (mm), SWS H is the SWS during harvest (mm). Y (kg ha −1 ) is the total hay yield.

N content, N absorption, and N-use e ciency
After measuring the dry weight, the whole legume and maize plants were crushed and stored in self-sealing bags. The N content of crushed samples was measured by an Automatic Kjeldahl N determinator (K-375, Buqi, Switzerland), and the N absorption and NUE were calculated according to the hay yield of legumes and maize (Sparks et al., 1996). The calculation formulae are as follows: where NA N is the N absorption of the N application treatment, NA NN is the N absorption of no N application, NA A is the amount of N application, and NC is the N content.

Statistical analysis
Excel 2010 was used for data processing and mapping, data were analysed using a residual test method before statistical analysis, and the data met the assumption of homogeneity of variances and followed normal distribution. SPSS 13.0 (SPSS Inc., Chicago, IL, USA) was used for analysis of variance, and Tukey's method was used for multiple comparisons between different treatments. The signi cance level was set as P < 0.05.

Results
3.1. Yield of fresh grass, hay, and crude protein As shown in Figs. 3 (a, b), 4 (a, b), and 5 (a, b), under the same planting mode for 2 years, the fresh grass, hay, and CP yields of legumes treated with N3 and N4 were signi cantly higher than those treated with N1. The average value showed that the yield of fresh grass, hay, and CP of legumes treated with N2, N3, and N4 was signi cantly higher than that of those treated with N1, and there was no signi cant difference between N3 and N4. At N3 and N4 levels, the fresh grass and hay yield of legumes under LM treatment were signi cantly higher than those under FM, and the CP yield of legumes under LM treatment was signi cantly higher than that under FM at N3 level. The average value showed that the fresh grass, hay, and CP yields of legumes under LM treatment were signi cantly higher than those under FM treatment.
As shown in Figs. 3(c-f), 4(c-f), and 5(c-f), under the same planting mode, the yields of fresh grass, hay, and CP of maize and in total under N3 and N4 treatments were signi cantly higher than those under N1, but there was no signi cant difference between N3 and N4. The average value showed that the fresh grass and hay yields of maize and in total under N2, N3, and N4 treatments were signi cantly higher than those under N1, in which the average total fresh grass yield increased by 27.21%, 60.38%, and 56.45%, respectively; the average total hay yield increased by 26.34%, 58.27%, and 52.94%, respectively; and the average total CP yield increased by 45.94%, 106.71%, and 100.00%, respectively. Under the same N application level, except for the total CP yield in 2020, there was no signi cant difference in the fresh grass, hay, and CP yields of maize and in total. The average value showed that the total fresh grass, total hay, and total CP yields of LM were signi cantly higher than those of M, with an average increase of 11.97%, 8.93%, and 32.05%, respectively.
Among the 12 treatments, LM-N3 obtained the highest 2-year average total fresh grass, hay, and CP yields.
3.2. Crude protein, crude fat, NDF, and ADF content As shown in Table 1, under the same planting mode, the crude protein (CP) and crude fat (CF) content in maize and in total treated with N3 and N4 were signi cantly higher than those under N1, and the CP content of legumes treated with N3 and N4 was signi cantly higher than that under N1. The average value showed that the CP content in maize, legumes, and in total under N2, N3, and N4 was signi cantly higher than that under N1, and the average content of total CP over 2 years increased by 16.55%, 30.94%, and 31.65%, respectively, compared with that under N1. The CF content in maize and in total under N3 and N4 was signi cantly higher than that under N1, and the average total CF content over 2 years increased by 18.47% and 22.49%, respectively, compared with that in N1. Under the same N application condition, there was no signi cant difference in the CP content of maize and legume, as well as the total CF content of maize and legume, but the total CP content of LM treatment was signi cantly higher than that of M. The average value showed that the total CP content of LM was signi cantly higher than that of M, which increased by 20.81% over 2 years. Among all treatments, LM-N3 obtained the highest 2-year average maize, legume, and total CP content. Table 2 shows that in 2020, under LM and M planting modes, the NDF content of maize and in total under the N1 treatment were signi cantly higher than those under N4. Under the same planting mode for 2 years, the NDF content of legumes under the N1 treatment was signi cantly higher than that under N3 and N4. The average value showed that the NDF and ADF contents of maize and in total under the N1 treatment were signi cantly higher than those under N3 and N4, and the NDF content of legumes under the N1 treatment was signi cantly higher than that under N3 and N4. The 2-year average total NDF content under N1 increased by 10.50% and 14.35%, respectively, compared with that under N3 and N4, and the total ADF content increased by 14.85% and 21.04%, respectively. Under the same N application level, there was no signi cant difference in NDF and ADF contents between maize and in total under different planting methods over 2 years. The average value showed that the NDF and ADF contents of legumes under the FM treatment were signi cantly higher than those under LM. The 2-year average ADF and NDF contents under the LM-N3 treatment were lower than those under LM-N1 and M-N1, but there was no signi cant difference.
3.3. Soil water storage, evapotranspiration, and water-use e ciency As shown in Table 3, under the same planting method, there was no signi cant difference in SWS and eld ET before sowing and during harvest under different N application treatments. The average value showed that there was no signi cant difference in SWS and eld ET between N application treatments. In 2019, the harvest water storage under the N1 and N2 treatments was signi cantly higher than that of N3 and N4, while in 2020, the harvest water storage of the N4 treatment was signi cantly higher than that of N2 and N3. Under the same N application level, there was no signi cant difference in SWS and eld ET between sowing and harvest under different planting methods. The 2-year average water storage before sowing and harvest and eld ET of the LM-N3 treatment were lower than those of FM-N4, LM-N1, and M-N4, but there was no signi cant difference.
Under the same planting mode, the WUE (WUE M ) and population WUE (WUE B ) of maize treated with N2, N3, and N4 were signi cantly higher than those under N1, but there was no signi cant difference between N3 and N4. The average value showed that the WUE M and WUE B of N2, N3, and N4 over 2 years were signi cantly higher than those of N1.

N content, N absorption, and N-use e ciency
As shown in Table 4, under the same planting mode, the N content and absorption of maize, legumes, and the whole system under N3 and N4 were signi cantly higher than those under N1, and there was no signi cant difference between N3 and N4. The average value showed that the N content and absorption of maize, legumes, and in total under N3 and N4 were signi cantly higher than those under N1. Under N3 and N4, the average total N content increased by 31.39% and 31.39% compared with that under N1, respectively, and the average total N absorption increased by 107.07% and 100.46% compared with that under N1, respectively.
There was no signi cant difference in N content and absorption of maize under the same N application level over 2 years. The average value showed that the total N content and absorption under LM and FM were signi cantly higher than those under M. Under LM and FM, the average total N content increased by 21.52% and 32.26%, respectively, the total N absorption increased by 16.03% and 22.85%, respectively, and the total N absorption under LM was signi cantly higher than that under FM. Among all treatments, LM-N3 obtained the highest 2-year average total N content and absorption.
As shown in Fig. 6, under the same planting mode, the NUE in maize, legumes, and in total under N2 and N3 was signi cantly higher than that under N4. The average value showed that the NUE in maize, legumes, and in total under N2 and N3 was signi cantly higher than that under N4, and the average total NUE over 2 years increased by 37.30% and 59.52%, respectively, compared with that under N4. Under the same N application level, there was no signi cant difference in the NUE of maize under different planting methods in 2019. The average value showed that the total NUE of FM and LM was signi cantly higher than that of M, with an increase of 33.09% and 27.34%, respectively, and the total NUE of LM was signi cantly higher than that of FM. Among all treatments, LM-N3 obtained the highest 2-year average NUE of maize, legumes, and in total. . This may be because legume/maize intercropping can get a higher soil coverage than pure maize and the weeds were reduced by the limited light availability (Fischer et al., 2020). In addition, compared with maize monoculture, maize yield under cowpea/maize and soybean/maize intercropping increased by 25% and 22%, respectively (Latati et al., 2014; Ghaffarzadeh et al., 1994). As a feed crop with high nutrition and good palatability, forage soybean can be mixed with maize to improve crop quality and yield. When the ratio of legume:maize is 1:3 and 1:2, the population yield is 15.5% and 16.4% higher than that of maize monoculture, respectively (Zhou et al., 2015). Lablab bean is a forage crop with high protein content, high temperature resistance, frost resistance, and late maturity. Studies have reported that intercropping lablab bean with maize can improve crop quality but reduce maize yield (Muna et al., 2011), which was con rmed by our study results. In the 2-year experiment, the yield of maize under legume/maize intercropping was lower than that under maize monoculture, but the total yield of the intercropping system increased due to the increase in legume crop yield. The yield of the lablab bean/maize intercropping system was higher than that of the forage soybean/maize intercropping system. N-xing legumes/maize intercropping can improve the protein content of feed crops. In addition, a higher feed quality will show lower ADF and NDF contents. Legume/maize intercropping signi cantly reduces ADF and NDF contents compared with maize monoculture, which may be because it reduces the proportion of maize or increases the proportion of legumes in the total feed yield, thus improving forage quality (Javanmard et al., 2020). The results of this study were consistent with those of the above studies. The intercropping of lablab beans, forage beans, and maize increased the CP and crude fat content of mixed forage crops, reduced the content of ADF and NDF, and improved the nutritional quality. The nutritional quality of lablab bean/maize intercropping was higher than that of forage bean/maize intercropping. Anil et al. (2000) also showed that legume/maize intercropping reduced the contents of dry matter, starch, NDF, and ADF of forage crops and increased the content of CP. This may be due to the high content of protein, Ca, and P in leguminous crops in legume/maize intercropping systems, which is conducive to improving the quality of mixed silage (Titterton et al., 1997).
In addition to the planting mode, N application is also an important factor affecting the nutritional quality of forage crops. Increased N application has been shown to reduce the grain saturation of crops and increase the protein concentration of grains (Kassu et al., 2021). In addition, increasing N application increased the protein content of maize crops and decreased the starch content, which improved the nutritional quality of maize The WUE in our 2-year intercropping system was signi cantly higher than that of the monoculture system, and the WUE of lablab bean/maize intercropping was higher than that of forage bean/maize intercropping. In addition, studies have shown that WUE under maize/peanut 1:5 intercropping was 83.2% higher than that of monocultured maize (Choudhary et al., 2016). This is because mixed planting or intercropping can improve the coverage of the ground surface and reduce ineffective evaporation from the soil, thus improving the WUE of the system (Yin et al., 2015). In addition, legume/maize intercropping helps to increase the abundance of Nxing bacteria in the soil, thus improving the ability of crops to absorb N from the soil (Wang et al., 2021). Correlativity study shows that symbiotic N 2 xation increases because legume and maize root interaction signi cantly increases legume nodulation and maize avonoids (signalling compounds of rhizobia) exudation . This is consistent with the results of our study. The NUE of the legume/maize intercropping system was signi cantly higher than that of maize monoculture. This may be because mixed planting or intercropping changes the spatial structure, thus affecting the competition between crops for resources, and then affecting the utilisation of N fertilizers (Yang et al., 2014).
In addition to the planting mode, N application also affected the WUE and NUE of forage crops. Studies have shown that in the legume/maize intercropping system, reducing the N application rate at the jointing stage of maize from 135 kg ha −1 to 45 kg ha −1 can prolong the N accumulation time of pea and increase the total N accumulation (Hu et al., 2020). Rational application of N fertilizer can improve NUE. If a large amount of N fertilizer is applied, then the yield input ratio of N fertilizer will decrease and the NUE will be reduced (Ju et al., 2007). The results of our experiment were similar to those reported by the above studies. Compared with no N application, N application signi cantly increased the N content and absorption of maize, leguminous crops, and the whole system, and the NUE after treatment with 240 kg ha −1 of N was signi cantly higher than that after treatment with 360 kg ha −1 N. Studies have reported that increasing N fertilizer can signi cantly increase the yield of forage crops, reduce eld ET, and improve crop WUE (Xu et al., 2018; Xu et al., 2020). The results of our study con rmed this. Compared with no fertilization, high N fertilization (360 kg ha −1 ) and medium N fertilization (240 kg ha −1 ) increased the total hay yield, but there was no signi cant difference in eld ET, which signi cantly improved the WUE of maize and the population. In all treatments, the highest WUE and NUE were obtained under the application of 240 kg ha −1 N in the lablab bean/silage maize intercropping system.
This model provides a management strategy suitable for silage maize forage production in the arid area of Northwest China. The results of this study can provide guidance for farmers and farm managers this and similar regions. However, this study is limited in that it was carried out for only 2 years, and the annual rainfall, temperature, and other climatic conditions may differ over longer periods. Therefore, our results and conclusions have certain limitations, and we will carry out longer term research in the future.

Conclusion
Compared with M, LM and FM treatments signi cantly increased the total fresh hay yield, CP yield, WUE B , total N content and absorption, and NUE of forage crops over 2 years. Compared with monoculture, intercropping signi cantly increased the content of CP, reduced the content of NDF and ADF, and the forage quality of LM was better than that of FM. Compared with N1, the N3 and N4 treatments signi cantly increased the total fresh and hay yield, CP yield, CP content, and crude fat content, and decreased the contents of NDF and ADF.
In addition, the N3 and N4 treatments signi cantly improved WUE B , N content, N absorption, and NUE compared with N1, and the 2-year average NUE under N3 was signi cantly higher than that under N4. Among all treatments, LM-N3 obtained the highest 2-year average total fresh and hay yield, CP yield, CP content, and NUE, and the contents of NDF and ADF were low. Therefore, the application of 240 kg ha −1 N (LM-N3) in a lablab bean/silage maize intercropping system is a reasonable model for forage crop production in the arid area of Northwest China.   and P × N represent planting mode, N application, and interaction between them, respectively; * P < 0.05 level, ** P < 0.01; ns, no signi cant difference.      Crude protein yield of legume, maize, and intercropping systems under different treatments in 2019 and 2020. Note: M, silage maize monoculture; FM, intercropping of forage soybean and silage maize; LM, intercropping of lablab bean and silage maize; N1, 0 kg ha −1 N; N2, 120 kg ha −1 N; N3, 240 kg ha −1 N; N4, 360 kg ha −1 N. P, N, and P × N represent planting mode, N application, and interaction between them, respectively; * P < 0.05 level, ** P < 0.01; ns, no signi cant difference.