Dynamics of Productivity and Nitrogen Contribution in Mixed Legume/Grass Grasslands in Rain-fed Semi-arid Areas in Northwest China

Abstract

Mixed legume/grass grasslands are the most significant type of artificial grassland in rain-fed semi-arid regions. Understanding the contributions of legumes and grasses to grassland productivity, as well as the nitrogen-sharing mechanisms between them, is crucial to maintaining the sustainability, stability, and high yield of mixed grasslands. In this study, four commonly used cultivated species were selected: smooth bromegrass (Bromus inermis Leyss.), orchardgrass (Dactylis glomerata L.), sainfoin (Onobrychis viciifolia Scop.), and red clover (Trifolium pratense L.). Combinations of two and three species of legumes and grasses were established, with monoculture serving as the control. The results revealed that in all the monocultures and mixed grasslands comprising two or three species, the average dry matter yield (DMY) of mowed grasslands in 2017 was significantly higher than in 2018, while the average DMY of grazed summer regrowth in 2018 surpassed that of 2016 and 2017. Over the period from 2016 to 2018, smooth bromegrass and sainfoin gradually dominated the mixed grasslands, while orchardgrass and red clover exhibited a declining abundance. Over time, the ratio and amount of nitrogen (N) fixation in legumes significantly increased in both the monoculture and mixed grasslands. Similarly, the amount of nitrogen (N) received by grasses also increased significantly in mixed grasslands. However, the proportion of nitrogen fixed by legumes remained below 10% in 2016, 20% in 2017, and 30% in 2018. In contrast, nitrogen transfer from legumes to smooth bromegrass was less than 10%, while in orchardgrass, it was even lower, at less than 2%. The interannual variability in dry matter yield (DMY) and nitrogen contribution in the mixed grasslands of rain-fed semi-arid areas is primarily influenced by forage adaptability and average annual precipitation. Increasing the proportion of grazed forage relative to hay in annual forage consumption should be considered, as more extensive grazing can reduce damage from field rodents and provide higher forage quality at lower costs and energy consumption. To maintain grassland productivity, targeted grazing should be carefully planned and implemented.

1. Introduction

Mixed grasslands are a primary type of artificial grassland, with legume and grass mixtures being the most common combination. In mixed-species grasslands, different types of forages provide distinct biological characteristics and effectively utilize environmental resources to enhance the yield and quality of harvested or grazed forages [1]. Numerous studies indicate that compared to monoculture grasslands, mixed legume/grass grasslands have higher yields [2], more balanced nutritional values [3], and lower weed incidences [4]. The forage yield and nutritional value of mixed grasslands are significantly influenced by establishment and management factors, including species comprised in mixture, utilization, and fertilizer application, particularly nitrogen fertilizer usage [5]. Numerous studies, both domestically and internationally, have focused on mixed legume/grass pastures, primarily examining forage yield [6,7], nutritional value [8,9], soil enhancement [10], interspecific relationships [11], and community stability [12]. Additionally, several studies have investigated biological nitrogen fixation and nitrogen transfer between legumes and grasses in mixed grasslands [13]. For example, the study showed that the competition of Gramineae forage for soil available nitrogen significantly enhanced biological nitrogen fixation in legume forage [14]. Boller and Nosberger (1987) [15] identified three primary limiting factors affecting biological nitrogen fixation in legume/grass mixed grasslands: (1) low biomass of leguminous forage; (2) a proportion of leguminous forage in the mixed grasslands below 50%; and (3) low dependency of leguminous forage on biological nitrogen fixation, less than 70%. In mixed grasslands composed of legumes and grasses, there exists both a competitive and a promotive relationship. Specifically, legumes fix nitrogen from the atmosphere, which not only satisfies their own growth needs but also supplies nitrogen to associated grasses through various mechanisms [14,16,17]. The competition and promotion relationships between legumes and grasses in mixed grasslands significantly influence productivity and stability. This influence is ultimately reflected in changes to forage dry matter yield and the composition of grassland species [17]. Maintaining a balance between these two relationships is crucial for achieving high and stable yields. For artificial mixed grasslands to maximize productivity, it is essential to maintain community stability. The selection of appropriate variety combinations, effective regulation of environmental factors, and prudent management practices are essential for maintaining the stability of an artificial grassland community [18]. Numerous studies have demonstrated that the forage yield in mixed grasslands is higher than that in pure sowing systems, primarily due to niche complementarity [19,20]. This over-yield phenomenon is particularly pronounced when significant differences in morphological structure and function exist among species in mixed grasslands, especially in the context of legume and grass interactions [21,22]. Xinjiang, one of China’s primary pastoral areas, has experienced a decline in natural grassland productivity and severe deterioration of the ecological environment, largely due to unsustainable utilization practices and global climate change [23,24]. With the implementation of the new grazing ban policy, the focus of animal husbandry has shifted from grassland-based systems to confined systems in agricultural regions. However, the shortage of high-quality forage remains a significant barrier to the modernization and large-scale development of animal husbandry in Xinjiang [25]. Large-scale production of high-quality forage is crucial for addressing the forage shortage in the development of animal husbandry in Xinjiang. Producing high-quality forage is vital for maintaining the health and enhancing the productivity of ruminants. The Xiejiagou area in Urumqi City, Xinjiang, is a typical mountain grassland where Kazak herdsmen reside, and mixed grassland represents one of the most important grassland types in this region. Traditional grassland utilization methods primarily involve mowing in mid-August and grazing from September to October.

The following issues are prevalent: (1) The mowing period is too late, typically occurring in mid-August, when most forage has already entered the reproductive growth stage. Consequently, the nutritional value of hay is very low. (2) August and September are characterized by dry and rainless conditions, which hinder forage regeneration after mowing and result in extremely low grazing productivity in artificial grasslands during this period. Extensive management and overutilization have led to low productivity and poor stability in mixed grasslands, making it challenging to ensure sufficient forage supply for livestock during winter. Improving the productivity and stability of mixed grasslands in this area is crucial for the development of its animal husbandry. In response to these issues, this study investigated the existing mixed grassland areas, focusing on forage yield, nutritional value, and utilization methods. The four most commonly used cultivated species were selected: smooth bromegrass, orchardgrass, sainfoin, and red clover. Additionally, two legume/grass combinations and three mixed sowing combinations were established, with single sowing used as the control. Long-term monitoring was conducted to assess changes in productivity and species composition in mixed grasslands. These species were chosen not only for their common use in local production but also for their diverse morphological characteristics, including height, leaf shape, tiller number, and root structure. These characteristics may lead to varying responses to changes in resource utilization [26]. Climatic conditions characterized by low annual average temperatures and low precipitation significantly influence grassland productivity in this region. To mitigate these limitations, it is essential to select forages that exhibit higher drought stress tolerance during grassland establishment. This study aimed to (1) identify the most suitable monoculture or mixed-culture species for summer cutting and autumn-winter grazing in the semi-arid mountainous areas of Xinjiang; (2) assess the productivity, stability, and sustainability of mixed grasslands in the semi-arid rain-fed mountainous region of Xinjiang; and (3) evaluate the role of legume biological nitrogen fixation in the nitrogen cycle within this semi-arid rain-fed mountain ecosystem.

2. Materials and Methods

2.1. Description of the Experimental Site

The field experiment was conducted from 2016 to 2018 at the Xiejiagou Grassland Ecological Experimental Station in Xinjiang Province, northwestern China, located at a latitude of 87°03′4″ N, a longitude of 43°31′4″ E, and an elevation of 1675 m above sea level. The study area is situated on the terraces of the middle and low mountain slopes on the northern side of the Tianshan Mountains. The average annual temperature at the site ranged from 2.1 to 3.3 °C, annual precipitation varied between 100 and 400 mm, and precipitation was unevenly distributed, primarily concentrated from April to October (Figure 1). Annual evaporation ranged from 1141.7 to 1283.3 mm, indicating a short growth period, with a frost-free duration of only 113 to 130 days. Based on previous indicators, areas with annual precipitation of ≤200 mm are classified as dry, while those with 200 to 400 mm are classified as semi-dry [27]. The measured rainfall at the site in 2016, 2017, and 2018 was 389.6, 250.1, and 282.5 mm, respectively. Therefore, this area was classified as semi-arid. Before establishing the artificial grassland, the experimental area contained degraded grassland, predominantly featuring Oxytropis coerulea (Pall.) DC., Polygonum sibiricum Laxm., Potentilla anserina L., and Taraxacum mongolicum Hand.-Mazz. The area suffered significant rodent damage, resulting in a near loss of its value for animal husbandry. The soil was sandy, primarily composed of sand and gravel, and exhibited low fertility. Soil samples were randomly collected from five points in the field at a depth of 0–30 cm before grassland establishment, and their physical and chemical properties were analyzed. The physical and chemical parameters of the 0–30 cm soil layer at the experimental site are shown in

Table 1. The properties of soil physical and chemical in 0–30 cm.

(Index)	pH	Total N (g·kg−1)	Available N (mg·kg−1)	Total P (g·kg−1)	Available P (mg·kg−1)	Total K (g·kg−1)	Available K (mg·kg−1)
Data	8.11	0.62	36.08	0.52	7.46	17.46	231.54

.

2.2. Experimental Design

This experiment employed a randomized block design, featuring monocultures of the four most commonly sown forage species: smooth bromegrass (Bromus inermis Leyss.), orchardgrass (Dactylis glomerata L.), sainfoin (Onobrychis viciifolia), and red clover (Trifolium pratense) as controls. Additionally, four mixed-culture grasslands were established with two-species combinations (one legume and one grass) and two mixed-culture grasslands with three-species combinations. Each treatment was replicated four times across the experimental plots. The plots (15 m²) were sown using a broadcasting seed method. The monocultures of smooth bromegrass, orchardgrass, sainfoin, and red clover were theoretically sown at rates of 37.5 kg ha⁻¹, 22.5 kg ha⁻¹, 30 kg ha⁻¹, and 22.5 kg ha⁻¹, respectively. In the two-species combinations, each species accounted for 50% of the sowing rate, while in the three-species combinations, each represented one-third of the total sowing rate (

Table 2. Seeding rates for different species across various mixed sowing patterns.

	Mixed Mode	Seeding Rates (kg ha⁻1)
		Smooth Bromegrass	Orchardgrass	Sainfoin	Red Clover
1	Sainfoin	——	——	30.0	——
2	Red clover	——	——	——	22.5
3	Smooth bromegrass	37.5	——	——	——
4	Orchardgrass	——	22.5	——	——
5	Smooth bromegrass/Red clover	18.75	——	——	11.25
6	Smooth bromegrass/Sainfoin	18.75	——	15.0
7	Orchardgrass/Red clover	——	11.25	——	11.25
8	Orchardgrass/Sainfoin	——	11.25	15.0	——
9	Red clover/Smooth bromegrass/Sainfoin	12.5	——	10.0	7.5
10	Smooth bromegrass/Sainfoin/Orchardgrass	12.5	7.5	10.0	——

). The sowing of the experiment took place in June 2016. The actual sowing rates were calculated based on the germination rates of each forage species. The germination rates for the smooth bromegrass, orchardgrass, sainfoin, and red clover were 84%, 61%, 50%, and 86%, respectively. These species were chosen for two primary reasons: (1) they are the most commonly used in local production, and (2) they exhibit distinct morphological characteristics, including plant height, leaf type, number of branches/tillers, and root configuration. These various morphological characteristics may influence resource utilization differently [26]. Weed emergence was controlled after sowing, and no irrigation was applied throughout the duration of the experiment. Two PVC tubes, each with a diameter of 30 cm, were inserted into each experimental plot for ¹⁵N isotope labeling on 5 August 2016, 8 August 2017, and 1 August 2018. Urea (CO(¹⁵NH₂)₂) with a total mass of 1.10 g and 10% abundance was dissolved in 400 mL of tap water and uniformly mixed. A 5 mL aliquot of the solution was extracted using a syringe, diluted to 100 mL with tap water, and then uniformly leached into the soil within each PVC tube. This method aimed to assess biological nitrogen fixation in leguminous forages and the associated nitrogen transfer to companion grasses.

2.3. Determination Indicators and Methods

Dry Matter Yield: During the experiment, all the plots were mowed on 1 July 2017 and 28 June 2018. The forage yield of each plot was measured before mowing, which was recorded as the first forage yield (Figure 2A,B, forage yield for cutting); after mowing, fertilizers were applied, and the fertilizers used were diammonium phosphate 120 kg ha⁻¹ (P₂O₅ 46%) and potassium sulfate 100 kg ha⁻¹ (K₂O 50%). Therefore, the field received a total of 21.6 kg N, 55.2 kg P₂O₅, and 50 kg K₂O per hectare per year from these commercial fertilizers. Grazing was carried out from September 30 to October 30 in 2016, 2017, and 2018 under mixed livestock grazing by sheep and beef cattle. The forage yield of the mixed grassland was measured before the start of grazing (the determination dates were September 26, September 28, and September 22 in 2016, 2017, and 2018, respectively), which was recorded as the second crop of forage yield (Figure 2C,D, grazing biomass).

Dry Matter Yield Determination Method: After cutting the forage to ground level in a sample box measuring 1 × 1 m² across all the plots, the fresh weight was recorded by weighing the forage in paper bags. Subsequently, the forage was dried in an oven at 65 °C until a constant weight was achieved, allowing for the calculation of dry matter yield.

Crude Protein Yield Calculation: Dried samples of the second crop forage from 2016, 2017, and 2018 were ground using a plant grinder (FZ102) to determine the nitrogen content of the forage using the Kjeldahl method. This determination was based on the soil agrochemical analysis by Bao Shidan (third edition). The crude protein yield of mixed grassland was calculated as follows: Crude Protein Yield (kg ha⁻¹) = Forage Yield (kg ha⁻¹) × Forage Nitrogen Content (%) × 6.25.

N₂ Fixation Determination: The ¹⁵N isotope dilution method was selected for its superior accuracy in measuring N₂ fixation in situ by legumes [28]. Before determining ¹⁵N abundance and total nitrogen, dried shoot samples were ground into a fine powder using a rotary mill (MM200, Retsch, Haan, Germany). Analyses were conducted using a ¹⁵N isotope ratio mass spectrometer (Mat 253, Finnigan MAT, Bremen, Germany).

2.4. Equations Used in N₂ Fixation Data Analysis

(1) Atom percent excess (¹⁵N atom percent excess, APE) of the sample was calculated by the following equation:

$$\mathrm{A}\mathrm{P}\mathrm{E}=\mathrm{A}-\mathrm{A}\mathrm{o}$$

(1)

where APE is the ¹⁵N atom percent excess of the sample, A is ¹⁵N abundance of the sample, and Ao = 0.365% is the natural abundance of ¹⁵N.

The percentage of atmospheric N₂ fixation to total N in sainfoin and red clover (%Ndfa) was calculated by the following equation [29]:

$$\%Ndfa=(1-{\displaystyle \frac{\%NdfF}{\%NdfNF}})$$

(2)

where %NdfF is the ¹⁵N atom percent excess in N₂-fixing plants (sainfoin and red clover) and %NdfNF is the ¹⁵N atom percent excess in non-N₂-fixing plants (smooth bromegrass and orchardgrass).

The percentage of N transferred from N₂-fixing plants to non-N₂-fixing plants in total N (%NTrans) was calculated by the following equation [30]:

$$\%NTrans=(1-{\displaystyle \frac{\%NdfNFmix}{\%NdfNFmono}})\times 100$$

(3)

where %NdfNFmix is the ¹⁵N atom percent excess in non-N₂-fixing plants in mixtures and %NdfNFmono is the ¹⁵N atom percent excess in non-N₂-fixing plants in pure sowing.

2.5. Statistical Analyses

All the data were summarized using Excel 2013, graphing was performed using the GraphPad Prism 7.0 software, and statistical analysis was performed using the SAS9.0 software. The PROCGLM program was used to analyze the variance of the dry matter yield under mowing and grazing, the crude protein yield, the crude protein and nitrogen contents, and the ratio and amount of nitrogen fixation and nitrogen transfer.

3. Results

3.1. Forage Biomass

From 2016 to 2018, the dry matter yield (DMY), crude protein content (CP), and crude protein yield (CPY) of the mixed grassland varied significantly with increasing growth years (

Table 3. Variance analysis of forage biomass and nutritional value in mixed grassland.

Block	Dry Matter Yield				Crude Protein Yield		CP
	df	F Value	p Value	df	F Value	p Value	df	F Value	p Value
	3	0.22	0.8841	3	0.62	0.6024	3	0.44	0.7252
Year	2	92.01	<0.0001	2	20.31	<0.0001	2	27.43	<0.0001
Mode	9	21.75	<0.0001	9	21.36	<0.0001	9	19.16	<0.0001
Years × Mode	18	6.01	<0.0001	18	16.73	<0.0001	18	2.29	0.0059

, p < 0.0001). Additionally, significant differences were observed in the forage yield, crude protein yield, and the respective yields of sainfoin, red clover, smooth bromegrass, and orchardgrass (p < 0.001). The results indicated significant differences in DMY, CPY, and CP (p < 0.001). The interaction effects of year and mixed sowing combinations on other indices were statistically significant (p < 0.05). For cutting forages (Figure 3a), the average dry matter yields of the four species under monoculture sowing were 4361.77 kg ha⁻¹ in 2017 and 2383.32 kg ha⁻¹ in 2018; conversely, the average yields of the mixed-culture grassland with two-species combinations were 5048.55 kg ha⁻¹ and 3623.36 kg ha⁻¹, while those with three-species combinations were 7239.95 kg ha⁻¹ and 4793.74 kg ha⁻¹, respectively.

The average dry matter yields of the four forages under pure sowing, as well as in the mixed-culture grassland with two- and three-species combinations in 2017, exceeded those in 2018. For grazing forages (Figure 3b), the dry matter yields of the monoculture and mixed-culture grasslands significantly increased with growth years. For example, the dry matter yields in 2018 were significantly higher than that in 2016 (|t| = 3.75, p = 0.0045) and 2017 (|t| = 4.325, p = 0.0019), but there was no significant difference between 2016 and 2017 (|t| = 0.7015, p = 5007). In the monoculture grassland of sainfoin and smooth bromegrass, the dry matter yield during grazing in 2018 was significantly higher than in 2017 and 2016 (p < 0.001). In the monoculture grassland of red clover and orchardgrass, dry matter yield during grazing in 2018 was marginally higher than in 2017 and 2016; however, this difference was not statistically significant (p > 0.001).

In the mixed grassland with two-species combinations, the dry matter yields for smooth bromegrass/sainfoin and orchardgrass/sainfoin in 2018 were significantly higher than those in 2017 and 2016 (p < 0.01). The dry matter yield for smooth bromegrass/red clover in the mixed-culture grassland in 2018 was significantly higher than the yields in 2017 and 2016 (p < 0.05). No significant differences in dry matter yields for orchardgrass/red clover in the mixed grassland were observed in 2016, 2017, and 2018 (p > 0.05). In the mixed grassland with three-species combinations, the dry matter yields for smooth bromegrass/sainfoin/red clover and smooth bromegrass/sainfoin/orchardgrass in 2018 were significantly higher than those in 2017 and 2016 (p < 0.01). As the number of growth years increased, the crude protein content of the forages decreased. For instance, the crude protein content of the forage in 2018 was significantly lower than the contents in 2016 (|t| = 4.568, p = 0.0014) and 2017 (|t| = 4.006, p = 0.0031) (

Table 4. The crude protein (CP) and crude protein yield (CPY) of the mixed grassland of legumes and grasses.

	CP (%)				CPY (kg ha−1)
	2016	2017	2018	Average	2016	2017	2018	Average
Sainfoin	19.44	18.77	15.95	18.05	441.02	1275	726.32	814.41
Red clover	19.44	18.91	14.20	17.52	205.28	784.1	180.40	389.93
Smooth bromegrass	14.15	10.72	8.63	11.17	194.19	507.1	270.14	323.81
Orchardgrass	12.34	9.92	8.33	10.20	205.55	157.2	161.27	174.68
Smooth bromegrass/Red clover	16.41	17.26	10.56	14.74	204.92	740.1	206.11	383.73
Smooth bromegrass/Sainfoin	18.07	18.08	14.79	16.98	310.89	123.3	635.85	356.70
Orchardgrass/Red clover	12.84	13.34	11.87	12.68	169.52	40.61	140.20	116.78
Orchardgrass/Sainfoin	15.05	14.18	15.06	14.76	326.00	89.60	966.71	460.77
Red clover/Smooth bromegrass/Sainfoin	17.54	22.28	14.37	18.06	273.51	149.2	697.61	373.46
Smooth bromegrass/Sainfoin/Orchardgrass	14.15	17.24	14.17	15.18	282.70	102.9	572.44	319.35
Average	15.94	16.07	12.79		261.36	397.02	455.71
F value	13.26	5.29	27.21		9.81	31.56	10.66
p value	<0.0001	0.0002	<0.0001		<0.0001	<0.0001	<0.0001

). The change in the crude protein yields between 2016 and 2017 was not significant (|t| = 0.165, p = 0.8726). In contrast, the crude protein yields increased significantly. For instance, the average crude protein yield in 2018 was 455.71 kg ha⁻¹, higher than the yields of 397.02 kg ha⁻¹ in 2017 (|t| = 0.3496, p = 0.7347) and 261.36 kg ha⁻¹ in 2016 (|t| = 2.626, p = 0.0275). However, there was no significant difference between the yields of 2017 and 2016 (|t| = 1.092, p = 0.3034).

In the monoculture smooth bromegrass, when mixed with red clover, the total biomass decreased by 9.25%, 13.84%, and 34.98% in 2016, 2017, and 2018, while the crude protein content increased by 15.93%, 61.11%, and 22.38%, respectively. When mixed with sainfoin, the total biomass increased by 34.37%, 89.18%, and 62.08%, while the crude protein content increased by 27.70%, 68.77%, and 71.34%, respectively. Similarly, in the mixed-culture grassland with smooth bromegrass/sainfoin, when red clover was added, the total biomass decreased by 14.94%, 3.44%, and 4.84% in 2016, 2017, and 2018, respectively. Protein content decreased by 2.96% and 2.84% in 2016 and 2017, respectively, while the crude protein increased by 23.18% in 2017. When orchardgrass was added, the total biomass increased by 9.57% in 2016 but decreased by 12.98% and 21.10% in 2017 and 2018, respectively. The crude protein content decreased by 21.71%, 4.68%, and 4.22% in 2016, 2017, and 2018, respectively.

For the monoculture orchardgrass, when mixed with red clover, the total biomass decreased by 18.82%, 34.56%, and 40.47%, while crude protein increased by 4.07%, 34.49%, and 42.44% in 2016, 2017, and 2018, respectively. When mixed with sainfoin, the total biomass increased by 28.89%, 64.20%, and 229.72%, and crude protein rose by 21.96%, 43.04%, and 80.74% in 2016, 2017, and 2018, respectively. Similarly, in a mixed-culture grassland comprising orchardgrass and sainfoin, the addition of smooth bromegrass resulted in total biomass declines of 7.83%, 15.59%, and 37.78% in 2016, 2017, and 2018, respectively. The crude protein levels decreased by 6.01% and 5.92% in 2016 and 2018, respectively, but increased by 21.54% in 2017.

For monoculture sainfoin, when mixed with smooth bromegrass, the total biomass decreased by 19.16% in 2016, followed by increases of 23.37% and 11.40% in 2017 and 2018, respectively. The crude protein levels decreased by 7.05%, 3.67%, and 7.24% during the same years. When mixed with orchardgrass, the total biomass decreased by 3.91%, 27.18%, and 41.25%, while crude protein decreased by 22.58%, 24.45%, and 5.56% in 2016, 2017, and 2018, respectively. Similarly, in the mixed-culture grassland consisting of smooth bromegrass and sainfoin, the addition of red clover resulted in total biomass declines of 28.44%, 6.34%, and 24.95% in 2016, 2017, and 2018, respectively. The crude protein levels increased by 16.51% and 57.07% in 2016 and 2017, respectively, but decreased by 4.56% in 2018. In the mixed-culture grassland consisting of sainfoin and orchardgrass, the addition of smooth bromegrass led to total biomass declines of 7.83%, 15.59%, and 37.78%. Crude protein decreased by 6.01% and 5.92% in 2016 and 2018, respectively, while it increased by 21.54% in 2017.

For monoculture red clover, mixing with smooth bromegrass resulted in total biomass increases of 17.66%, 5.93%, and 60.35% in 2016, 2017, and 2018, respectively, while crude protein decreased by 15.59%, 8.72%, and 25.60%. When mixed with orchardgrass, the total biomass increased by 30.44% in 2016, followed by decreases of 4.42% and 8.48% in 2017 and 2018. Correspondingly, crude protein decreased by 33.91%, 29.49%, and 16.43%. In the mixed-culture grassland with red clover and smooth bromegrass, the addition of sainfoin resulted in total biomass increases of 17.66%, 5.93%, and 60.35% in 2016, 2017, and 2018, respectively, while crude protein increased by 6.90%, 29.04%, and 36.03%.

3.2. Interannual Variation Coefficient

During the mowing of forages in 2017 and 2018, the interannual variation coefficient (IVC) of the dry matter yield in the monoculture forages exhibited significant differences. Specifically, red clover demonstrated the highest IVC at 96%, while orchardgrass had the lowest at 6%. In the mixed-culture grasslands with two-species combinations, the IVCs were notably higher for smooth bromegrass/red clover (50%) and orchardgrass/red clover (55%), whereas the values for smooth bromegrass/sainfoin (11%) and orchardgrass/sainfoin (11%) were relatively low. The IVCs for red clover/smooth bromegrass/sainfoin and smooth bromegrass/sainfoin/orchardgrass were 29% and 28.5%, respectively, in the three-species mixed-culture grassland. For the forages utilized for grazing in 2016, 2017, and 2018, the IVC of dry matter yield was lowest for red clover (11.4%) and orchardgrass (12.4%) in monoculture. In contrast, the IVCs for sainfoin (48.3%) and smooth bromegrass (54.2%) in monoculture were relatively high. The IVC for orchardgrass/red clover (12.4%) was the lowest among the two-species combinations in the mixed-culture grassland. The IVCs for red clover/smooth bromegrass/sainfoin (58.7%) and smooth bromegrass/sainfoin/orchardgrass (41.5%) were observed in the three-species mixed-culture grassland. A weak correlation was observed between the number of species and the annual variation coefficients in the mixed grasslands (Figure 4a, mowing: R² = 0.0523, p = 0.5251; Figure 4b, grazing: R² = 0.138, p = 0.2906).

3.3. Proportion of Each Component in Mixed Treatments

The proportion of dry matter yield for each species in mixed sowing reflects its contribution to the productivity of the grassland and indirectly indicates its competitive ability for environmental resources. In the red clover/smooth bromegrass mixed grassland, the proportion of dry matter yield from red clover significantly decreased from 62.8% in 2016 and 51.6% in 2017 to 6.4% in 2018 (p < 0.05, Figure 5A). In contrast, the yield proportion of smooth bromegrass increased significantly to 93.6% in 2018 compared to 37.2% in 2016 and 48.4% in 2017 (p < 0.05, Figure 5A).

In the sainfoin/smooth bromegrass mixed grassland, the proportion of dry matter yield from sainfoin decreased progressively from 2016 to 2018, while that of smooth bromegrass increased during the same period (p > 0.05, Figure 5B). In the red clover/orchardgrass mixed grassland, the proportions of dry matter yield from both red clover and orchardgrass remained relatively stable over the years (p > 0.05, Figure 5C).

In the sainfoin/orchardgrass mixed grassland, the proportion of dry matter yield from sainfoin increased annually, whereas that from orchardgrass decreased (p > 0.05, Figure 5D). From 2016 to 2018, in the sainfoin/red clover/smooth bromegrass mixed grassland, the proportion of dry matter yield from sainfoin significantly increased (p < 0.05, Figure 5E). In contrast, the proportion from red clover sharply decreased from 44.39% to 1.53% (p < 0.05, Figure 5E), while that from smooth bromegrass significantly increased (p < 0.05, Figure 5E).

In the sainfoin/smooth bromegrass/orchardgrass mixed grassland, the proportions of dry matter yield from sainfoin and smooth bromegrass increased gradually (p > 0.05, Figure 5F), while that from orchardgrass significantly decreased from 40% in 2016 to below 5% in 2018 (p < 0.05, Figure 5F).

3.4. Relationship Between DMY of Legume and Grass Composition and Total DMY in Mixed Grassland

The analysis of the relationship between the biomass of legume and grass species and the total biomass in the six mixed grasslands revealed a significant linear correlation between the legume biomass and total biomass in the years 2016, 2017, and 2018 (Figure 6, 2016, Y = 0.4288x + 274.1, R² = 0.2559, p = 0.0117; 2017, y = 0.8039x − 295.0, R² = 0.8371, p < 0.0001; 2018, y = 0.8791x − 1069, R² = 0.9000, p < 0.0001; total, R² = 0.8672, p = 0.0001). As the years of growth increased, the R² value and slope of the linear equation relating the leguminous forage biomass to the total biomass rose, while the R² value and slope for the gramineous forage biomass decreased. This trend indicates an increased contribution of the legume forage to the total biomass and a decreased contribution of the grass biomass. Notably, the linear relationship between the grass biomass and total biomass was highly significant in 2016 (Figure 6a, y = 0.5715x − 274.2, R² = 0.3799, p = 0.0013) and significant in 2017 (Figure 6b, y = 0.1928x + 307.9, R² = 0.1928, p = 0.0318) but was not significant in 2018 (Figure 6c, y = 0.0578x + 1409, R² = 0.0533, p = 0.2777). The three-year average was also significant (Figure 6d, total, R² = 0.3494, p = 0.0001).

3.5. The Ratio and the Amount of Nitrogen Fixation in Legume

The nitrogen contents of sainfoin (

Table 5. Nitrogen content of forages in mixed grassland, amount and ratio of nitrogen fixation by legumes, and transfer nitrogen from legumes to grasses.

Mixed Mode	Nitrogen Content (%)			Proportion of Nitrogen Fixation (%)			Amount of Nitrogen Fixation (kg ha−1)
	2016	2017	2018	2016	2017	2018	2016	2017	2018
Sainfoin
Sainfoin	3.11 ± 0.18	3.00 ± 0.23	2.55 ± 0.14	5.30 ± 1.74	14.56 ± 2.88	20.04 ± 3.04	2.85 ± 2.21	8.51 ± 2.20	23.18 ± 2.57
Smooth bromegrass/Sainfoin	3.09 ± 0.24	3.05 ± 0.89	2.50 ± 0.21	9.06 ± 1.81	18.96 ± 2.77	28.29 ± 6.14	2.85 ± 1.90	10.69 ± 5.50	21.83 ± 21.26
Orchardgrass/Sainfoin	2.94 ± 0.21	2.23 ± 0.76	2.67 ± 0.14	6.55 ± 1.71	17.94 ± 3.24	30.72 ± 4.28	2.32 ± 0.92	6.59 ± 2.45	37.74 ± 16.58
Red clover/Smooth bromegrass/Sainfoin	2.90 ± 0.09	4.02 ± 0.79	2.51 ± 0.13	8.50 ± 1.98	15.49 ± 3.34	32.54 ± 4.44	1.49 ± 0.35	9.60 ± 3.76	23.94 ± 4.12
Smooth bromegrass/Sainfoin/Orchardgrass	2.81 ± 0.28	3.09 ± 1.15	2.59 ± 0.16	9.56 ± 1.28	20.20 ± 1.96	27.54 ± 2.64	2.34 ± 0.46	4.88 ± 4.18	16.18 ± 9.21
	df	F value	p value	df	F value	p value	df	F value	p value
Block	3	1.53	0.2201	3	1.55	0.2164	3	1.85	0.1535
Mode	4	1.77	0.1533	4	7.34	0.0001	4	1.48	0.2268
Year	2	6.12	0.0047	2	204.22	<0.0001	2	47.48	<0.0001
Mode × year	8	2.67	0.0183	8	2.74	0.0157	8	1.55	0.1697
Red clover
Red clover	3.11 ± 0.16	3.03 ± 0.80	2.27 ± 0.03	5.29 ± 1.69	10.84 ± 1.56	24.90 ± 3.57	1.63 ± 0.53	3.52 ± 1.60	7.19 ± 1.27
Smooth bromegrass/Red clover	2.91 ± 0.22	3.28 ± 0.82	2.56 ± 0.02	8.53 ± 2.05	12.99 ± 3.77	29.48 ± 4.88	1.98 ± 1.09	2.24 ± 0.75	1.03 ± 1.30
Orchardgrass/Red clover	2.70 ± 0.11	2.47 ± 0.39	2.35 ± 0.08	8.87 ± 2.83	15.90 ± 4.71	27.67 ± 6.44	1.25 ± 0.59	1.95 ± 1.59	2.69 ± 2.08
Red clover/Smooth bromegrass/Sainfoin	2.89 ± 0.13	2.47 ± 0.79	2.47 ± 0.79	10.41 ± 2.08	14.12 ± 3.63	30.34 ± 2.24	1.09 ± 0.55	1.13 ± 1.28	0.43 ± 0.85
	df	F value	p value	df	F value	p value	df	F value	p value
Block	3	0.98	0.4151	3	0.31	0.8198	3	0.34	0.795
Mode	3	1.76	0.1738	3	3.61	0.0233	3	14.42	<0.0001
Year	2	4.65	0.0167	2	123.76	<0.0001	2	4.64	0.0167
Mode × year	6	0.92	0.492	6	0.38	0.8889	6	6.19	0.0002
Mixed Mode	Nitrogen Content (%)			Proportion of Nitrogen Transferred from Legume (%)			Amount of Nitrogen Transferred from Legume (kg ha−1)
	2016	2017	2018	2016	2017	2018	2016	2017	2018
Smooth bromegrass
Smooth bromegrass	2.27 ± 0.18	1.72 ± 0.12	1.38 ± 0.03	--	--	--	--	--	--
Smooth bromegrass/Red clover	2.20 ± 0.35	2.37 ± 0.77	1.62 ± 0.16	1.24 ± 0.65	4.98 ± 1.52	6.76 ± 1.41	0.13 ± 0.1	0.61 ± 0.36	2.03 ± 0.22
Smooth bromegrass/Sainfoin	2.26 ± 0.27	2.49 ± 0.73	2.02 ± 0.07	2.31 ± 0.98	5.82 ± 2.46	5.77 ± 1.22	0.16 ± 0.12	1.08 ± 0.89	1.93 ± 0.59
Red clover/Smooth bromegrass/Sainfoin	2.34 ± 0.23	3.06 ± 0.40	2.05 ± 0.18	4.18 ± 0.92	8.90 ± 1.56	9.89 ± 1.65	0.24 ± 0.11	1.25 ± 0.65	3.50 ± 1.42
Smooth bromegrass/Sainfoin/Orchardgrass	2.12 ± 0.35	2.99 ± 0.83	2.06 ± 0.10	2.89 ± 0.68	6.82 ± 2.11	9.64 ± 2.98	0.12 ± 0.06	1.23 ± 0.80	2.11 ± 1.77
	df	F value	p value	df	F value	p value	df	F value	p value
Block	3	0.92	0.4375	3	3.04	0.0428	3	1.77	0.1719
Mode	4	5.6	0.001	3	12.55	<0.0001	3	2.14	0.1145
Year	2	14.91	<0.0001	2	52.47	<0.0001	2	34.05	<0.0001
Mode × year	6	2.02	0.0674	6	1.03	0.4223	6	1.1	0.3816
Orchardgrass
Orchardgrass	1.73 ± 0.13	1.59 ± 0.09	1.33 ± 0.10	--	--	--	--	--	--
Orchardgrass/Sainfoin	1.64 ± 0.30	2.43 ± 0.61	1.62 ± 0.16	2.14 ± 0.86	6.91 ± 1.01	6.28 ± 1.33	0.35 ± 0.11	1.31 ± 0.23	1.86 ± 0.69
Orchardgrass/Red clover	1.76 ± 0.22	2.38 ± 0.73	1.77 ± 0.15	1.24 ± 0.84	3.36 ± 1.27	6.12 ± 0.84	0.18 ± 0.15	0.27 ± 0.21	0.66 ± 0.46
Smooth bromegrass/Sainfoin/Orchardgrass	1.69 ± 0.09	1.90 ± 0.24	1.74 ± 0.30	3.20 ± 1.37	6.96 ± 1.34	7.98 ± 1.60	0.44 ± 0.25	0.27 ± 0.32	0 ± 0
	df	F value	p value	df	F value	p value	df	F value	p value
Block	3	0.27	0.8489	3	28.82	<0.0001	3	1.44	0.255
Mode	3	3.60	0.0235	2	0.41	0.6676	2	29.75	<0.0001
Year	2	8.32	0.0012	2	6.72	0.0048	2	7.75	0.0025
Mode × year	6	1.79	0.1308	4	1.52	0.2273	4	9.58	<0.0001

, 2016, df = 4, F = 1.54, p = 0.2401; 2017, df = 4, F = 2.39, p = 0.0969; 2018, df = 4, F = 0.77, p = 0.5616) and red clover (2016, df = 3, F = 4.27, p = 0.0288; 2017, df = 3, F = 1.29, p = 0.3226; 2018, df = 3, F = 0.41, p = 0.7459) did not differ significantly between the monocultures and mixed cultures. In contrast, the nitrogen contents of smooth bromegrass (2016, df = 4, F = 0.33, p = 0.8536; 2017, df = 4, F = 2.20, p = 0.1181) and orchardgrass (2016, df= 3, F = 1.16, p = 0.3640; 2017, df = 3, F = 2.52, p = 0.1075) were also not significantly different between the monocultures and mixed cultures in both 2016 and 2017. However, a significant difference was observed in other comparisons between the monocultures and mixed cultures (smooth bromegrass, 2018, df = 4, F = 26.08, p < 0.0001; orchardgrass: 2018, df = 3, F = 4.31, p = 0.0279).

The nitrogen fixation ratio of sainfoin under mixed culture was significantly higher than that under monoculture (Table 5, 2016, df = 4, F = 4.39, p = 0.0150; 2017, df = 4, F = 2.68, p = 0.0720; 2018, df = 4, F = 4.98, p = 0.0093). The nitrogen fixation ratio of red clover under mixed culture was only significantly higher than that under monoculture in 2016 (2016, df = 3, F = 3.82, p = 0.0393), but there were no significant differences in 2017 and 2018 (2017, df = 3, F = 1.38, p = 0.2965; 2018, df = 3, F = 1.11, p = 0.3815).

With the exception of significantly lower nitrogen fixation in red clover under mixed culture than under monoculture in 2018 (Table 5, 2018, df = 3, F = 17.94, p < 0.0001), the nitrogen fixation amounts in sainfoin (2016, df = 4, F = 0.63, p = 0.6461; 2017, df = 4, F = 1.50, p = 0.2527; 2018, df = 4, F = 1.52, p = 0.2468) and red clover (2016, df = 3, F = 1.19, p = 0.3560; 2017, df = 3, F = 2.14, p = 0.1483) were not significantly different between the monocultures and mixed cultures in 2016 and 2017.

3.6. The Ratio and Amount of Nitrogen Transfer from Legume to Grass in Mixed Grassland

There was a significant difference in the proportion of nitrogen transferred from the associated legumes to smooth bromegrass when mixed with different legumes (Table 5, 2016, df = 3, F = 8.90, p = 0.0022; 2017, df = 3, F = 2.98, p = 0.0737; 2018, df = 3, F = 4.51, p = 0.0243). The proportion of nitrogen transferred from the associated legumes to orchardgrass when mixed with different legumes was not significantly different in 2016 (df = 2, F = 3.46, p = 0.0766) and 2018 (df = 2, F = 2.53, p =0.1343), and a significant difference was found only in 2017 (2017, df = 2, F = 11.50, p = 0.0033).

When smooth bromegrass was mixed with different legumes, the amounts of nitrogen transferred from the legumes were different, but the difference was not significant (Table 5, 2016, df = 3, F = 1.27, p = 0.3299; 2017, df = 3, F = 0.72, p = 0.5570; 2018, df = 3, F = 1.59, p = 0.2435). With the exception of 2016 (2016, df = 2, F = 2.04, p = 0.1858), the amount of nitrogen transferred from different legumes to orchardgrass showed a significant difference in the other study years (2017, df = 2, F = 21.85, p = 0.0004; 2018, df = 2, F = 15.64, p = 0.0012). The positive linear relationship between legume biomass and the ratio and amount of nitrogen fixation by legumes was significant (Figure 7a, the amount: y = 0.006982x − 2.442, R² = 0.8615, F = 155.5, p < 0.0001; the ratio: y = 0.00321x + 12.73, R² = 0.186, F = 5.712, p = 0.0247). There was a positive linear relationship between grass biomass and the ratio and amount of nitrogen transfer from legumes (Figure 7b, the amount: R² = 0.149, p = 0.084; the ratio: R² = 0.7109, p < 0.0001).

With the increase in growth years (from 2016 to 2018), the ratio and amount of nitrogen fixation in sainfoin and red clover in both monocultures and mixed cultures increased significantly (p < 0.01). For example, the average nitrogen fixation ratios of sainfoin were 7.79%, 17.43%, and 27.83% in 2016, 2017, and 2018, respectively, and the nitrogen fixation amounts were 2.37 kg ha⁻¹, 8.06 kg ha⁻¹, and 24.57 kg ha⁻¹, respectively. The corresponding values for red clover were 8.28%, 13.46%, and 28.10% and 1.49 kg ha⁻¹, 2.21 kg ha⁻¹, and 2.83 kg ha⁻¹. Similarly, with the increase in growth years (from 2016 to 2018), the amount of nitrogen transferred from the associated legumes to smooth bromegrass and orchardgrass in the mixed-culture grassland also increased significantly (p < 0.01); for example, the proportions of nitrogen transferred to smooth bromegrass from legumes were 2.66%, 6.63%, and 8.01% in 2016, 2017, and 2018, and the amounts were 0.16 kg ha⁻¹, 1.04 kg ha⁻¹ and 2.39 kg ha⁻¹, respectively. The corresponding values for orchardgrass were 1.02%, 1.21%, and 1.26% and 0.32 kg ha⁻¹, 20.62 kg ha⁻¹, and 0.84 kg ha⁻¹, respectively.

4. Discussion

4.1. Productivity and Stability of Mixed Grassland

Moisture is a critical factor influencing biodiversity, productivity, and stability in grassland ecosystems [31,32]. Studies indicate that water shortages can reduce species richness and dry matter yield in grassland plant communities by 37% and 47%, respectively [33]. Additionally, nitrogen and phosphorus are major limiting factors for grassland ecosystem productivity and forage quality [34,35]. In arid and semi-arid regions, the effects of water and nutrients on grassland productivity are interrelated. During drought years, water limitations were predominant; however, in rainy years, these limitations shifted to nutrient restrictions [32,36]. Forage yield in perennial grasslands within arid and semi-arid environments is largely dependent on climatic factors, particularly rainfall and temperature during the growing season. It is also influenced by the environmental conditions at the testing site and the growing season’s duration [37].

This study involved sowing a mixed grassland in June 2016. However, the biomass produced that year was insufficient for mowing and drying hay, resulting in grazing only being conducted at the end of 2016. Forages harvested at the end of June 2017 and June 2018, which were assessed for dry matter production, were primarily utilized as hay for winter livestock feeding. An analysis of the results indicates that the dry matter yields from both monoculture and mixed-culture (two-species and three-species combinations) were significantly higher in 2017 compared to 2018. For instance, the average dry matter yield of the four forages grown in monoculture in 2017 was 1.83 times greater than that in 2018. Additionally, the average yield from the mixed-culture grassland with two-species combinations in 2017 was 1.39 times that of 2018, while the three-species combination yielded 1.51 times more than in 2018. This discrepancy may relate to effective rainfall during forage growth. During the period from April to June, total rainfall was 147 mm in 2017 and 111 mm in 2018.

In 2016, 2017, and 2018, the second crop dry matter yield of regenerated forage was primarily utilized for the mixed grazing of sheep and beef cattle during the autumn and winter months, beginning in October. The dry matter and crude protein yields were measured prior to grazing. Overall, the summer growth average dry matter yields of the four forages from both the pure sowing and mixed cultures (two-species and three-species combinations) were significantly higher in 2018 compared to 2016 and 2017. This increase may be attributed to the 171.4 mm of rainfall recorded from July to September in 2018, surpassing the 172.0 mm in 2016 and 102.8 mm in 2017. Deak et al. (2007) also observed that the impact of annual rainfall on grassland yield distribution was more significant than that of the mixed components [38].

Mixed grasslands characterized by high species diversity generate higher and more consistent biomass yields [39]. The stability of species composition in a mixed grassland comprising legumes and grasses, along with its effects on dry matter yield and nutritional quality, results from competition for resources between these plants (i.e., their ability to coexist). Other studies on grassland stability have demonstrated that the yield stability of mixed grasslands, both spatially and temporally, increases with greater species or functional group diversity [40,41]. This phenomenon occurs because greater species diversity provides a stronger buffering effect against environmental changes.

In this study, we found that the relationships between stability (measured by the interannual coefficient of variation) and species richness in mixed grasslands under mowing and grazing were weak or statistically insignificant. These findings are consistent with the existing literature [42,43].

This may be attributed to the fact that the stability of artificial grasslands is more susceptible to environmental impacts compared to that of natural grasslands. For instance, an analysis of a grassland experiment conducted at Rothamsted Park in the United Kingdom found that species richness had only a weak effect on biomass stability [43]. The authors of the Leps study suggested that community stability is more closely related to the life history strategies of the plant species present than to species richness [44]. Species characteristics and mixed grassland composition may be more significant determinants of yield and yield stability than merely increasing species richness in these systems [38].

Forages exhibit variations in production performance when mixed with other types due to their distinct biological characteristics and environmental adaptability. Deak et al. (2007) indicated that pasture production in mixed grasslands declines as the number of species increases [38]. This phenomenon was also observed in the current experiment (see Table 3). The dry matter yields of the mixed grasslands with the same number of species exhibited significant differences. For instance, in the pastures used for grazing, the difference between the highest yield (orchardgrass + sainfoin) and the lowest yield (orchardgrass + red clover) in the mixed grasslands with two-species combinations was 938.66 kg ha⁻¹ in 2016, 1510.75 kg ha⁻¹ in 2017, and 5276.91 kg ha⁻¹ in 2018. Similarly, in the mixed grasslands with three-species combinations, the differences between the highest and lowest yields were 448.04 kg ha⁻¹, 232.45 kg ha⁻¹, and 826.03 kg ha⁻¹. In the grasslands subjected to mowing, the difference between the highest yield (smooth bromegrass + sainfoin) and the lowest yield (orchardgrass + red clover) for the mixed grasslands with two-species combinations was 3990.40 kg ha-1 in 2017 and 4589.34 kg ha-1 in 2018. Similarly, in the mixed grasslands with three-species combinations, the differences between the highest yield (smooth bromegrass + sainfoin + orchardgrass) and the lowest yield (smooth bromegrass + sainfoin + red clover) were 942.10 kg ha⁻¹ and 657.45 kg ha⁻¹.

Sainfoin is a drought-tolerant perennial legume forage [45] known for its high forage yield, elevated crude protein content, excellent palatability, and substantial nutritional value [46]. Additionally, it contains concentrated tannins (20–40 g/kg dry matter) and polyphenols, which do not induce bloating in grazing livestock, making it particularly suitable for grazing and grassland applications [42].

In this study, the total dry matter yields of the mixed grasslands containing smooth bromegrass and orchardgrass were lower than those of the monoculture grasslands when red clover was included. In contrast, the total dry matter yield of the mixed grasslands was higher than that of the monoculture grasslands when mixed with sainfoin. Similarly, when mixed with smooth bromegrass, the total dry matter yield of the grasslands containing sainfoin and red clover was greater than that of the monoculture grasslands. Conversely, when mixed with orchardgrass, the total dry matter yield of the mixed grasslands was lower than that of the monoculture grasslands. This indicates that the yield persistence in the mixed grasslands is influenced by the compatibility of the species present.

Among the mixed grasslands, the dry matter yield of the smooth bromegrass and sainfoin mixture was relatively high, measuring 1828.38 kg ha⁻¹ in 2016, 2436.60 kg ha⁻¹ in 2017, and 5078.63 kg ha⁻¹ in 2018. The addition of other species, such as red clover or orchardgrass, did not result in increased yield.

In this study, the proportion of orchardgrass and red clover in the mixed grasslands gradually decreased with increasing growth years, indicating a tendency for these species to disappear [47]. This decline is likely associated with their relatively poor drought tolerance. Additionally, selective feeding and trampling by grazing livestock affected the composition and persistence of species in mixed grasslands [48]. For instance, grazing ruminants typically prefer legumes [49]. In our experiments, we did not assess the grazing livestock’s preferences or selectivity. However, we believe that increasing livestock density and reducing grazing duration or employing mixed grazing with large and small livestock will limit selective feeding opportunities.

The crude protein content of forage in the mixed grasslands containing legumes was significantly higher than that in the grass-only grasslands. This increase is primarily attributed to the higher protein content in legumes and the nitrogen transfer from legumes to grasses. In the mixed grassland containing orchardgrass and red clover, the crude protein content of forage significantly decreased when combined with smooth bromegrass. Conversely, the crude protein content significantly increased when mixed with sainfoin. This indicates that the crude protein content in the mixed grasslands depends on the proportion of legumes present. The dry matter content of legumes typically ranges from 35% to 40%, which is optimal for both forage yield and quality [50].

As the growth period increased, the crude protein content of the forage declined; however, the crude protein yield increased significantly. This increase is primarily attributed to the forage yield and nitrogen content. Across all the treatments, the crude protein yield was notably high, ranging from 261.36 to 455.71 kg ha⁻¹. The higher values may indicate plant maturity and forage yield during the second harvest. This observation aligns with the CP yield reported by Jefferson et al. (2004) and may be related to the vegetative growth period during grazing in the second harvest [51].

Grassland degradation is primarily influenced by two factors: rodent damage and the overgrazing of livestock. Research has shown that grassland degradation creates conditions that promote rodent damage, which in turn exacerbates further degradation [52]. This study examined severely degraded grassland before establishing a mixed grassland, characterized by reduced vegetation cover, decreased productivity, a lower proportion of edible forage, and significant rodent damage. Following the establishment of the mixed grassland, both vegetation density and cover significantly increased, and no rodent damage was observed in the areas subjected to mowing and grazing during the experimental period. This outcome may stem from grassland restoration and the activities of grazing livestock, which disrupted rodent behavior patterns, resulting in increased avoidance and vigilance, along with decreased play activities. However, it is essential to implement preventive measures against rodent damage through effective grazing management. Studies by La Morgia et al. (2015) and Steen et al. (2005) found that grazing livestock suppress field rodent populations by competing for food, trampling and disturbing their habitats, and using feces and urine to drive them away [53,54]. Therefore, to maintain grassland productivity, it is crucial to carefully plan and implement targeted grazing strategies that increase the portion of grazing consumption in total hay yield. More intensive grazing can reduce rodent damage and produce higher quality forage at lower costs and energy expenditure.

4.2. Nitrogen Utilization of Mixed Grassland

The absorption and utilization of nitrogen by plants are vital for enhancing productivity. Biological nitrogen fixation allows perennial legumes to decrease their reliance on soil nitrogen, thereby potentially improving the sustainability of pasture-based livestock production systems. In both sainfoin and red clover, no significant difference was observed in the nitrogen content between the monocultures and mixed cultures (p > 0.05). However, smooth bromegrass and orchardgrass exhibited significantly higher nitrogen contents in mixed sowing compared to monoculture (Table 4, p < 0.01). This indicates that smooth bromegrass and orchardgrass can acquire more nitrogen in mixed cultures compared to monocultures. The primary reasons are as follows: (1) in mixed cultures, smooth bromegrass and orchardgrass can absorb nitrogen from associated legumes via biological nitrogen fixation; (2) red clover and sainfoin can fulfill their nitrogen requirements through biological nitrogen fixation, thereby minimizing competition with the grasses for nitrogen in the soil.

The nitrogen content of smooth bromegrass mixed with sainfoin was significantly higher than that of smooth bromegrass mixed with red clover (3.17 kg N ha⁻¹ vs. 2.77 kg N ha⁻¹). Similarly, the nitrogen content of orchardgrass mixed with sainfoin was significantly higher than that mixed with red clover (3.52 kg N ha⁻¹ vs. 1.11 kg N ha⁻¹). An analysis of the relationship between nitrogen fixation and legume yields revealed that legume yields significantly influence their nitrogen fixation, a finding supported by prior studies [55,56]. Due to their higher biomass and unique biological nitrogen fixation capabilities, legumes are regarded as excellent green manure crops. They are frequently used to enhance poor soils since they can alleviate nitrogen limitations in various habitat conditions. Over time, legume growth influences soil nitrogen availability and the nitrogen cycle in grassland ecosystems [57,58]. However, inoculation before sowing may be necessary, as poor soil conditions are highly unfavorable for the survival and growth of rhizobia [59]. Generally, biological nitrogen fixation in legumes supplies most of the nitrogen required for their growth. Huss-Danell, Chaia, and Carlsson (2007) observed that the proportion of nitrogen from biological nitrogen fixation in clover typically exceeds 80% throughout the growing season [60]. In red clover, the proportion of nitrogen derived from biological nitrogen fixation ranged from 65% to 67% [56]. Leguminous forages mixed with grasses derive over 80% of their nitrogen from biological nitrogen fixation [61]. Thus, the productivity of these forages determines both the quantity and proportion of nitrogen fixation [62]. However, in our study, the biological nitrogen fixation ratio in legumes was below 30%, significantly lower than in other studies. The primary reason may be the relatively low rainfall in the study area. High contributions of biological nitrogen fixation to the nitrogen yield in mixed pastures can only be achieved when soil conditions are suitable for nitrogen fixation and transfer in a mixed pasture-legume system [3]. Biological nitrogen fixation in legumes within arid zone ecosystems exhibits significant seasonality. During the rainy season, particularly during flowering and fruiting periods, the fixation rate was high, whereas it was significantly lower during the dry season [63].

A negative correlation exists between soil nitrogen supply and its contribution to nitrogen input via biological nitrogen fixation in legumes. However, the content of inorganic nitrogen in the soil does not influence the biological nitrogen fixation (BNF) ratio of legumes in mixed grasslands, as associated grasses have a greater competitive ability to absorb nitrogen from the soil than legumes [3,64]. Grasses in mixed grasslands can absorb both available nitrogen from the soil and nitrogen transferred from the associated legumes. For instance, studies indicate that nitrogen transferred from legumes to grasses can comprise over 60% of the total nitrogen in grasses within mixed legume and grass systems established for two years [65]. Conversely, the proportion of nitrogen transferred from legumes to grasses in mixed grasslands established for three years was less than 5% in this study. This low proportion is primarily due to relatively low rainfall, which resulted in lower soil moisture content and subsequently limited the activity of rhizobia [61,64]. The growth years significantly affected the nitrogen content of sainfoin, red clover, smooth bromegrass, and orchardgrass in both single and mixed sowing conditions (p < 0.05). However, the mixed sowing method significantly influenced the nitrogen content of gramineae forage, specifically smooth bromegrass and orchardgrass, in both single and mixed sowing conditions (p < 0.05). The nitrogen content of sainfoin, red clover, smooth bromegrass, and orchardgrass decreased as growth years increased when sown alone. This suggests that to maintain high productivity and forage nutritional value in arid and semi-arid areas, a certain amount of nitrogen fertilizer should be applied during forage production and management.

5. Conclusions

Climatic conditions characterized by low annual average temperatures and low precipitation generally impact grassland productivity in this region. The most effective strategy to address these limitations is to select forage species that exhibit a high tolerance to drought stress during grassland establishment. Variations in the interannual dry matter yield of mixed grasslands in a semi-arid region primarily depend on average annual precipitation and forage adaptability. In this study, both the monocultures and mixed grasslands with two or three species exhibited significantly higher average dry matter yields in 2017 compared to 2018. Additionally, the average dry matter yield of the summer regrowth in 2018 surpassed that of 2016 and 2017. Over the course of three years, smooth bromegrass and sainfoin increasingly dominated the mixed grasslands, while orchardgrass and red clover were disadvantaged and exhibited a declining trend in abundance. As the growth years increased, the crude protein content of the forage declined, whereas the crude protein yield increased significantly. From 2016 to 2018, both the monocultures and mixed grasslands showed a significant increase in the proportion and quantity of biological nitrogen fixation in sainfoin and red clover. Similarly, the nitrogen uptake by smooth bromegrass and orchardgrass in the mixed grasslands also increased significantly (p < 0.01). However, the proportions of biological nitrogen fixation in sainfoin and red clover were below 10%, 20%, and 30% for the years 2016, 2017, and 2018, respectively. Additionally, nitrogen transfer from legumes to smooth bromegrass was less than 10%, while in orchardgrass, it was even lower, at less than 2%.

We propose the following recommendations for the establishment and management of mixed grasslands in this region: (1) Orchardgrass and red clover are not recommended species for the local mixed grasslands. (2) Management of mixed grasslands in this area is appropriate for both mowing and grazing activities. The optimal mowing period should change from mid-August to late June and early July. This adjustment ensures the nutritional value of the harvested forage and provides sufficient time for forage regeneration in autumn and winter. Early mowing can increase the proportion of forage consumed by livestock through grazing. It also helps suppress field rodent populations, which reduces yield losses and lowers the need for chemical control. Additionally, it can reduce livestock reliance on concentrates, thereby lowering overall livestock production costs. (3) To maintain the high productivity and nutritional value of mixed grasslands, nitrogen fertilizer must be applied during rainy periods of forage growth and regeneration.