1. Introduction The Altai region of Xinjiang is an important distribution area of natural mowing grasslands in northern China. Its mountain meadows serve not only as a critical source of winter–spring forage reserves, but also as a fundamental basis for maintaining regional livestock stability and ecological security [ 5]. During long-term mowing utilization, aboveground biomass and nutrients are continuously removed. When an imbalance occurs among mowing intensity, nutrient return, and the natural recovery capacity of the plant community, mowing grasslands may undergo degradation [ 6]. Degraded natural mowing grasslands in Altai are characterized primarily by a decline in the proportion of palatable forage species, an increase in forbs, reductions in community cover and aboveground biomass, and a weakening of topsoil nutrient supply capacity [ 7]. These changes not only reduce hay yield and forage quality, but also impair the recovery capacity of the vegetation–soil system [ 8, 9]. Therefore, how to enhance both the productive and ecological functions of degraded natural mowing grasslands through low-disturbance and broadly applicable vegetation restoration measures is an issue that needs to be addressed for the sustainable use of grasslands in this region. Overseeding with forage species is a key measure for restoring natural grasslands [ 10]. Compared with single-species overseeding, overseeding with mixtures of species from different functional groups can improve resource use efficiency through niche complementarity and enhance community structure and forage quality [ 11, 12]. Legume–grass mixtures have drawn particular attention because legumes have high protein content and can improve nitrogen supply through nitrogen fixation, whereas grasses generally possess strong tillering and root expansion capacities, which can increase community productivity and soil resource use efficiency [ 13, 14, 15]. However, the effectiveness of overseeding is not fixed but is jointly influenced by factors such as land-use type, degradation degree, climatic conditions, soil status, and species composition [ 16]. To date, studies on legume–grass mixture overseeding have mostly focused on grazed grasslands or sown pastures, whereas the screening of species mixtures and evaluation of restoration effectiveness for degraded natural mowing grasslands under long-term mowing use in Altai remain relatively limited. To fill the gap in the systematic restoration theory and technology for degraded natural mowing grasslands, this study focused on typical degraded natural mowing grasslands in Altai, Xinjiang. With the goals of optimizing community structure and enhancing ecosystem multifunctionality, and following the principle of selecting native and adaptive species, seven superior legume and grass species were chosen for mixture experiments, taking into account multiple factors such as local climatic adaptability and palatability. A suite of indicators including plant community structure, plant nutrition, and soil physicochemical properties were measured, and an ecosystem multifunctionality (EMF) was integrated and calculated. The objectives of this study were: (1) to compare the effects of overseeding with different legume–grass mixtures on vegetation, forage quality, soil properties, and EMF in degraded natural mowing grasslands; (2) to determine the interannual differences and mixture-specific responses of aboveground EMF (AEMF) and belowground EMF (BEMF) to overseeding; and (3) to elucidate the main pathways through which different legume–grass mixtures drive the synergistic recovery of the vegetation–soil system, and to screen candidate species mixtures suitable for different restoration objectives. 2. Materials and Methods 2.1. Study Area The experimental site is located on a mountain meadow within the administrative area of Qibashileke Village, Tierekti Township, Habahe County, Altay Prefecture, Xinjiang, away from village settlements. The center coordinates of the experimental plot are 48°26′37″ N, 86°33′11″ E, at an elevation of 1199.69 m. These coordinates represent the center point of the experimental plot, rather than the residential area or the village committee of Qibashileke Village. The site has been used long-term as a natural mowing grassland and is part of the local village grassland resources. The area lies within the Altai Mountain forest–steppe ecological functional zone, under a temperate continental climate, with a mean annual temperature of approximately 4.4 °C and a mean annual precipitation of approximately 195.5 mm, while precipitation in mountainous areas can reach 400–600 mm [ 24]. The experimental site is a degraded natural mowing grassland that has experienced long-term combined mowing and grazing use. Locally, mowing typically occurs from 20 to 25 July each year, followed by grazing until approximately 10 September, with no use during the remainder of the year. To ensure uniform initial conditions across treatments, the experiment was arranged on a meadow section with relatively consistent topography, slope, soil conditions, vegetation degradation level, and use history. The total experimental area was approximately 3096 m 2 (86 m × 36 m), within which 33 plots were established. Each plot measured 50 m 2, with 1 m isolation strips between adjacent plots. Prior to overseeding, a background vegetation survey was conducted over the entire experimental area on 17 May 2023. The pre-experiment vegetation survey adopted a quadrat method: three 1 m × 1 m quadrats were randomly placed within each planned plot, all vascular plant species within the quadrats were recorded, and the cover, height, and aboveground biomass of each species were measured. The survey results showed that a total of 15 plant species, belonging to 15 genera and 13 families, were recorded in the original grassland community at the experimental site. Because the site is a moderately degraded grassland under long-term mowing and grazing use, and the survey was confined to the homogeneous area where the experimental plots were established, the recorded species number is lower than the total flora of the region. Only dominant and main companion species are listed in the main text. The dominant species were Carex tristachya, Achillea millefolium, Geum aleppicum, Alopecurus aequalis, and Agrostis alba, whereas the main companion species included Potentilla chinensis, Taraxacum mongolicum, Rumex acetosa, and Geranium pratense. In terms of forage quality, high-quality forage species were relatively scarce, accounting for only 11.34% of the community; the sward quality was poor, with forbs dominating, and degradation was highly evident. According to the national standard “Classification indicators of degradation, desertification, and salinization of natural grasslands” [ 25], the grassland was assessed as moderately degraded. The soil physicochemical properties of the 0–10 cm layer before overseeding were as follows: alkali-hydrolyzable nitrogen content was 233 mg kg −1, available phosphorus content was 12.72 mg kg −1, available potassium content was 252.67 mg kg −1, and soil pH ranged from neutral to slightly alkaline. 2.2. Experimental Design Based on the adaptability to the local cool, arid, and non-irrigated conditions, feeding value and establishment capacity, nitrogen-fixing potential of legumes, community stability and forage yield potential of grasses, and relevance to degraded grassland restoration objectives, seven legume and grass species were selected as mixture components ( Table 1): Onobrychis viciifolia cv. Qitai, Medicago sativa cv. Xinmu No. 4, Trifolium pratense cv. Minshan, etc. The non-overseeded control (CK) was maintained as the original degraded natural grassland vegetation and received only the same basal fertilization and management as the other treatments. The remaining treatments were artificially overseeded mixture treatments. Based on preliminary trial results and the ecological functional differences among species, 10 legume–grass mixtures were established, including treatments with 3, 4, 5, and 6 species. The seeding ratio of legume to grass was fixed at 2:8 for all mixtures, determined according to the establishment performance, community stability, and competitive advantage of grasses observed in the preliminary trial ( Table 2). The experiment was arranged in a completely randomized block design with 11 treatments and three replications, resulting in 33 plots. Each plot measured 48 m 2 (6 m × 8 m), with 1 m-wide isolation strips between adjacent plots. Overseeding was performed on 17 May 2023, using no-till drill seeding with a row spacing of 30 cm. Seeds were untreated, and the seeding rates of the mixtures were calculated on a pure live seed (seed value) basis (see Table 1 for actual seeding rates). Prior to seeding, basal fertilizer was uniformly broadcast across all plots, consisting of 2000 kg ha −1 of locally sourced composted sheep manure and 8 kg ha −1 of diammonium phosphate (N–P 2O 5–K 2O: 18–46–0). No irrigation was applied. The composted sheep manure was a traditional farmyard manure produced by local farmers, without commercial labeling or nutrient content analysis, and was applied uniformly as a basal amendment to all treatments. 2.3. Measurements and Variables From 7 to 14 August in both 2024 and 2025, vegetation and soil surveys were conducted using a quadrat-based sampling approach. In each experimental plot, three 1 m × 1 m quadrats were randomly selected to determine plant community characteristics, forage nutritional quality, and soil physicochemical properties. 2.3.1. Measurement and Calculation of Community and Functional Group Attributes Within the selected quadrats, the height (natural growth height), cover, and aboveground and belowground biomass of the plant community and functional groups (grasses, legumes, perennial forbs, and annual/biennial forbs) were measured. Cover. Cover was determined using the point-intercept method. A 10 cm × 10 cm grid with a total of 100 intersection points was placed over each 1 m × 1 m quadrat. A fine pin was vertically inserted at each intersection point, and the plant species in contact with the pin were recorded. Species cover was calculated as the proportion of the number of pin-point hits for a given species relative to the total number of pin points. If the pin contacted more than one species at a single point, each contacted species was recorded separately. Total community cover was calculated as the proportion of pin points at which at least one plant was contacted. Aboveground biomass (AGB). All aboveground plant material in each quadrat was clipped at ground level and sorted by species. Samples were returned to the laboratory, weighed fresh, dried at 65 °C to constant weight, and then weighed to determine dry mass. Belowground biomass (BGB). After aboveground biomass harvest, root samples were collected from the same quadrats using a root corer with an inner diameter of 7 cm from the 0–10 cm soil layer. Three replicate cores were collected from each plot and pooled into one composite sample per plot, which was placed in a mesh bag. Samples were thoroughly washed under running water to remove soil particles and impurities until roots were clean. The washed roots were dried at 65 °C to constant weight, and dry mass was recorded as belowground biomass. 2.3.2. Determination of Forage Nutritional Value At each harvest, three 1 m × 1 m quadrats were selected along the diagonal of each plot, and approximately 200 g of fresh forage was collected as a composite sample. All samples were air-dried and ground prior to analysis. The following nutritional variables were determined: crude protein (CP) using the Kjeldahl method, and acid detergent fiber (ADF) and neutral detergent fiber (NDF) using the Van Soest method [ 26]. 2.3.3. Determination of Soil Physicochemical Properties After aboveground vegetation sampling, soil samples were collected from the same quadrat. A soil auger with an inner diameter of 3.8 cm was used to collect soil from the 0–10 cm layer following a five-point “Z”-shaped sampling pattern. The five subsamples from each quadrat were thoroughly mixed to form one composite sample. The soil samples were divided into two portions. One portion was used for the determination of soil physical properties. Undisturbed soil samples were collected using a cutting ring or auger of known volume, weighed fresh, and then oven-dried at 105 °C to constant weight. Soil water content (SW) was calculated from the difference between fresh and dry soil mass, and soil bulk density (BD) was calculated as the ratio of dry soil mass to sampling volume. The other portion was brought back to the laboratory, air-dried naturally in a cool and ventilated place, and freed of plant roots, litter, gravel and other debris. The samples were then ground, homogenized, and passed through a 2 mm sieve, after which they were sealed and stored for subsequent analysis. Soil organic matter (OM) was determined by the potassium dichromate oxidation—external heating method. Soil alkali-hydrolyzable nitrogen (AN) was measured by the alkali diffusion method, with hydrolysis performed using 1.0 mol L −1 NaOH. Soil available phosphorus (AP) was extracted with 0.5 mol L −1 NaHCO 3 solution (pH 8.5) and determined spectrophotometrically by either the molybdenum-antimony anti-colorimetric method or the molybdenum blue colorimetric method. Soil available potassium (AK) was extracted with 1.0 mol L −1 NH 4OAc solution (pH 7.0) and determined by flame photometry. All soil physicochemical analyses were performed following the methods of Klute [ 27] and Sparks [ 28]. 2.4. Data Analysis 2.4.1. Importance Value Calculation Species importance value was used to characterize the relative dominance of each species in the community and was calculated for all plant species recorded within the quadrats. The importance value was derived from relative height, relative cover, and relative aboveground biomass [ 29]: Importance value (IV) I V i = ( R H i + R C i + R B i ) / 3 where: RHi, RCi, and RBi represent the relative height, relative cover, and relative aboveground biomass of the i-th species, respectively. 2.4.2. Forage Nutritional Quality Calculation [ 30] Relative feed value (RFV) RFV = DMI ( % BW ) ୍ଠ DDM ( % DM ) / 1.2 Dry matter digestibility (DDM) DDM ( % DM ) = 88.9 − 0.779 ୍ଠ ADF ( % DM ) Dry matter intake (DMI) DMI ( % DM ) = 120 / NDF 2.4.3. Species Diversity Calculation [ 31] Species richness was expressed as the number of plant species recorded within each quadrat (S). Three 1 m × 1 m quadrats were established in each plot, and all plant species occurring in each quadrat were recorded. Plot-level species richness was calculated as the mean number of species across the three quadrats. Shannon–wiener diversity index (H) H = ∑ i = 1 S P i ln P i where: P i = I V i ∑ i = 1 S I V i is the proportion of the importance value of the i-th species relative to the total importance value of all species in the quadrat. 2.4.4. Ecosystem Function Index Calculation Highly collinear indicators were excluded through Pearson correlation analysis, and 11 core indicators were selected from the 28 measured indicators based on their representativeness of ecological functions and literature conventions for calculating ecosystem multifunctionality. Ecosystem multifunctionality (EMF) was then calculated using the selected core indicators, with aboveground ecosystem functions represented by Height, aboveground biomass (AGB), belowground biomass (BGB), the Shannon–Wiener index (H), and relative feed value (RFV), and belowground ecosystem functions represented by bulk density (BD), soil water content (SW), organic matter (OM), alkali-hydrolyzable nitrogen (AN), available phosphorus (AP), and available potassium (AK) [ 31]. Ecosystem multifunctionality index (EMF) EMF = 1 N ∑ i = 1 N f ( X i ) 2.5. Data Processing All data were organized using Microsoft Office Excel 2013, and statistical analyses were performed using IBM SPSS Statistics 27.0.1. Differences in various indicators among overseeding treatments were tested by one-way analysis of variance (ANOVA), with overseeding treatment treated as a fixed factor and quadrat replicates as the error term. When ANOVA results were significant, post hoc multiple comparisons were conducted using Duncan’s multiple range test, with the significance level set at p 0.05). By 2025, cover reached 100% in the A1, A3, A6, A7, and A10 mixtures, while the remaining mixtures had cover values between 95.33% and 99.33%, none of which differed significantly from CK. Height was greatest in A9, at 63.26 cm, which was significantly taller than CK. From 2024 to 2025, all mixtures increased in both cover and height, with the largest increase observed in A6 ( p 0.05); species richness was significantly higher than CK only in the A2 mixture, reaching 13.33 ( p 0.05). From 2024 to 2025, both indices increased significantly in the A8 and A10 mixtures ( p 0.05). By 2025, bulk density ranged from 0.68 to 1.06 g cm −3, with the lowest value in A7 and the highest in A10; the difference between these two treatments was significant ( p 0.05) ( Figure 5). 3.4. Relationships Among Individual and Multiple Ecosystem Functions No significant correlations were found between the EMF index and any individual indicator ( p > 0.05), nor between the belowground EMF index and any single indicator ( p > 0.05) ( Figure 6). In 2024, soil nutrients were positively correlated with species richness and plant diversity ( p < 0.01). Soil organic matter (OM) and alkali-hydrolyzable nitrogen (AN) were also positively correlated with the biomass proportion and importance value of grasses ( p < 0.01), whereas legume importance value was positively correlated with plant crude protein content and relative feed value ( p < 0.01). The ecosystem multifunctionality indices (total EMF, aboveground EMF, and belowground EMF) showed strong positive correlations with most community-structure and soil-nutrient variables; notably, aboveground EMF was significantly correlated with soil bulk density ( p < 0.01) ( Figure 6a). In 2025, the strong positive correlations between total EMF, aboveground EMF, and belowground EMF and individual indicators generally weakened to moderate or weak levels. Aboveground EMF exhibited the weakest correlations with individual indicators, whereas total EMF was more strongly correlated with plant indicators than with soil indicators ( Figure 6b). 4. Discussion 4.1. Effects of Overseeding Different Species Mixtures on Plant Community Structure and Stability The degradation of natural mowing grasslands has profound impacts, reducing both grazing capacity and ecosystem productivity while lowering the proportion of palatable forage [ 32]. Selecting suitable species and mixture combinations can fully realize the complementary advantages of forages in overseeded grasslands, facilitate the efficient utilization of environmental resources such as water, nutrients, light, and space, promote plant growth, and thereby increase community biomass and cover, ultimately enhancing the community stability of sown grasslands [ 10, 33]. Our results showed that overseeding effectively improved the plant community structure of degraded natural mowing grasslands, but the stability of community structure varied with species composition [ 16]. Specifically, the most effective mixtures, particularly A5, A6, and A8, were dominated by highly competitive grasses such as Bromus inermis cv. Wusu No. 1, combined with nitrogen-fixing legumes such as Trifolimm pratense cv. Minshan. These mixtures rapidly established a community configuration characterized by grass dominance, legume complementarity, and suppression of forbs, thereby conferring greater resistance to disturbance and higher stability. By contrast, A4 was less effective in suppressing forbs and optimizing functional group composition, suggesting that restoration outcomes remain limited when species mixtures fail to achieve effective niche complementarity [ 34]. Mechanistically, early-stage niche preemption and the initial recovery of productivity were primarily governed by the mass ratio effect [ 35]. In resource-limited degraded habitats, successful restoration at the initial stage depends on pioneer species that can rapidly establish dominance [ 36]. In our study, the grass biomass proportion of A6 reached 85.87% in 2024 and further increased to 90.67% in 2025, indicating that grasses rapidly attained overwhelming dominance in community biomass. This supports the mass ratio hypothesis, which predicts that ecosystem functioning is determined largely by the functional traits of dominant species [ 37]. The high first-year biomass of A5 (475.88 g m −2) can likewise be attributed mainly to the strong competitive ability of grasses such as Bromus inermis cv. Wusu No. 1, whose rapid growth, vigorous tillering, and superior resource capture enabled efficient preemption of light, water, and space. This accelerated the recovery of community productivity, while rapid canopy closure created a suppressive environment for forbs. These findings are consistent with grassland experiments by Sonkoly et al. (2019) [ 36], which showed that productivity is often highest when communities are dominated by perennial grasses with high-performance traits, whereas increased evenness may reduce the productivity gains driven by dominant species. Together, these results highlight the pivotal role of dominant species in the early phase of grassland restoration. However, dominance by competitive grasses alone cannot fully explain the superior performance of the best mixtures in forage quality and long-term sustainability. Both our results and the correlation analysis showed that legume importance value was positively correlated with crude protein content and relative feed value. This suggests that the contribution of legumes lies not in dominating community biomass, but in occupying a distinct functional niche linked to nitrogen fixation and nutritional improvement [ 17, 37]. This pattern is consistent with a functional complementarity effect, whereby different functional groups differ in their resource-acquisition strategies and thereby enhance resource-use efficiency and multifunctionality at the community level [ 18, 23]. In a controlled experiment on sown grasslands, Suter showed that complementary nitrogen-acquisition strategies between grasses and legumes can support both high productivity and multifunctionality under low to moderate nitrogen availability [ 38]. This finding further suggests that functionally informed mixture design can reduce dependence on external nitrogen inputs and promote a more self-sustaining restoration trajectory. Viewed in this context, the strong performance of A5, A6, and A8 indicates that these mixtures were underpinned by a dual mechanism of dominant-species structuring and functional complementarity [ 34, 35]: grasses provided structural stability and a production base, whereas legumes enhanced nitrogen availability and forage quality. Together, these components strengthened community competitiveness, suppressed forbs, and promoted the formation of stable communities. 4.2. Effects of Overseeding Different Species Mixtures on Ecosystem Service Functions 4.3. Response Patterns of Ecosystem Multifunctionality Under Overseeding with Different Species Mixtures Previous studies have indicated that ecosystem functions are continuously reorganized as communities develop, and the positive effect of diversity on productivity gradually emerges over time [ 22, 57]. Therefore, the changes observed in the second year suggest that the restored system, driven jointly by environmental conditions and human interventions, is still progressing toward a higher level of multifunctionality. Isbell [ 39] demonstrated in a large-scale grassland experiment that most species contribute positively to at least some functions under certain environmental contexts, and that their importance varies considerably with time, space, functional targets, and environmental conditions. This indicates that the optimal mixtures identified in a study often represent local optimums specific to the particular years, soil conditions, and selected functions. Facing longer time scales and greater climatic uncertainty, reliance on a single species mixture poses risks, whereas maintaining a pool of species with diverse functional traits and different environmental adaptation strategies can enhance system adaptability and effectively buffer against risks [ 58]. Long-term experiments have shown that the diversity–stability relationship strengthens over time, and that ecosystem stability is enhanced through mechanisms such as species asynchrony [ 40]. Therefore, while this study identified superior mixtures such as A5, A6, and A8, it further demonstrates that species mixture composition is a key factor influencing the restoration outcomes of degraded natural mowing grasslands. Appropriate legume–grass mixtures can not only improve the establishment success and productive performance of overseeded communities, but also drive synergistic recovery of the vegetation–soil system through functional complementarity, thereby providing essential support for maintaining ecosystem multifunctionality and achieving sustainable use of degraded mowing grasslands. 5. Conclusions Overseeding with grass–legume mixtures effectively restored degraded natural mowing grasslands in Altai. Legumes and grasses, through niche complementarity, rapidly established stable and productive communities, leading to early increases in aboveground biomass and forage quality, which subsequently drove gradual improvements in soil nutrient cycling and structure. However, the enhancement of the belowground EMF index lagged markedly behind that of the aboveground index and varied among species mixtures. For different land-management objectives, we recommend distinct mixture strategies. The mixture of O. viciifolia cv. QiTai (5%) + T. pratense cv. Minshan (15%) + D. glomerata (15%) + P. pratensis (15%) + B. inermis cv. Wusu No. 1 (50%) exhibited the most comprehensive and sustained performance in improving ecosystem multifunctionality and soil fertility; therefore, it is best suited to areas where integrated ecological restoration is the primary goal. The mixture of T. pratense cv. Minshan (10%) + M. sativa cv. Xinmu No. 4 (10%) + P. pratensis (30%) + B. inermis cv. Wusu No. 1 (50%) demonstrated clear advantages in rapidly suppressing forbs, maintaining the highest aboveground biomass, and increasing soil phosphorus content; thus, it is most suitable for natural mowing grassland where rapid productivity recovery and weed control are priorities. The mixture of M. sativa cv. Xinmu No. 4 (10%) + T. pratense cv. Minshan (10%) + D. glomerata (30%) + B. inermis cv. Wusu No. 1 (50%) was uniquely effective in improving forage nutritional quality and is therefore the preferred option for producing high-quality hay. By contrast, the mixture of O. viciifolia cv. QiTai (5%) + M. sativa cv. Xinmu No. 4 (15%) + D. glomerata (15%) + P. pratensis (15%) + B. inermis cv. Wusu No. 1 (50%) showed the weakest suppression of annual and biennial forbs, which may leave natural mowing grassland vulnerable to persistent weed invasion. Similarly, the mixture of M. sativa cv. Xinmu No. 4 (15%) + O. viciifolia cv. QiTai (5%) + P. pratensis (30%) + B. inermis cv. Wusu No. 1 (50%) performed poorly in driving key belowground functions such as soil organic matter accumulation, suggesting limited long-term restoration sustainability. Neither mixture is therefore recommended for practical restoration. Taken together, the stage-dependent recovery dynamics identified here demonstrate that restoration outcomes must be evaluated from a multidimensional and long-term perspective that integrates vegetation, soil, and hydrological interactions. These findings provide empirical support for restoration planning and adaptive land management in vulnerable ecosystems under global change. More broadly, our results support a management pathway based on nature-mimicking, function-oriented vegetation reassembly to enhance land-based ecosystem services and climate resilience, and provide a practical technical option for advancing the goal of land degradation neutrality under SDG 15. Author Contributions Conceptualization, J.Y., X.Z. and Z.D.; methodology, J.Y., P.Z. and H.X.; software, J.Y. and C.S.; validation, J.Y., P.Z., H.X. and C.S.; formal analysis, J.Y. and P.Z.; investigation, J.Y., H.X. and C.S.; resources, X.Z. and Z.D.; data curation, J.Y. and P.Z.; writing—original draft preparation, J.Y.; writing—review and editing, X.Z., Z.D., P.Z., H.X. and C.S.; visualization, J.Y. and C.S.; supervision, X.Z. and Z.D.; project administration, X.Z. and Z.D.; funding acquisition, X.Z. and Z.D. All authors have read and agreed to the published version of the manuscript. Funding This research was funded by Xinjiang Forestry and Grassland Administration grant number XJLYKJ-2024-21. APC: The APC was funded by Xinjiang Forestry and Grassland Administration. Data Availability Statement The original contributions presented in the study are included in the article material. Further inquiries can be directed to the corresponding authors. Acknowledgments The authors thank the relevant departments of the Xinjiang Forestry and Grassland Administration for their coordination and assistance. Conflicts of Interest The authors declare no conflicts of interest. References Figure 1. Changes of community functional group structure. Different letters indicate significant differences between treatments ( p < 0.05); * denote significant differences at p < 0.05. Figure 1. Changes of community functional group structure. Different letters indicate significant differences between treatments ( p < 0.05); * denote significant differences at p < 0.05. Figure 2. Changes in plant community characteristics. Different letters indicate significant differences between treatments ( p < 0.05); *, **, and *** denote significant differences at p < 0.05, 0.01, and 0.001, respectively. Figure 2. Changes in plant community characteristics. Different letters indicate significant differences between treatments ( p < 0.05); *, **, and *** denote significant differences at p < 0.05, 0.01, and 0.001, respectively. Figure 3. Changes of plant nutritional quality. Different letters indicate significant differences between treatments ( p < 0.05); * and ** denote significant differences at p < 0.05 and 0.01 respectively. Figure 3. Changes of plant nutritional quality. Different letters indicate significant differences between treatments ( p < 0.05); * and ** denote significant differences at p < 0.05 and 0.01 respectively. Figure 4. Changes of soil physical and chemical properties. Different letters indicate significant differences between treatments ( p < 0.05); *, **, and *** denote significant differences at p < 0.05, 0.01, and 0.001, respectively. Figure 4. Changes of soil physical and chemical properties. Different letters indicate significant differences between treatments ( p < 0.05); *, **, and *** denote significant differences at p < 0.05, 0.01, and 0.001, respectively. Figure 5. Changes of ecosystem multifunctionality (EMF) index, aboveground EMF index and belowground EMF index. Different letters indicate significant differences between treatments ( p < 0.05); * and ** denote significant differences at p < 0.05 and 0.01 respectively. Figure 5. Changes of ecosystem multifunctionality (EMF) index, aboveground EMF index and belowground EMF index. Different letters indicate significant differences between treatments ( p < 0.05); * and ** denote significant differences at p < 0.05 and 0.01 respectively. Figure 6. Correlation coefficients among vegetation characteristics, soil properties and ecosystem multifunctionality (EMF) index. ( a): Cover, community coverage; Height, community height; AGB, community above-ground biomass; BGB, community below-ground biomass; Gr–AGB%, grass above-ground biomass ratio; Gr–IV, grass importance value; Lg–AGB%, legume above-ground biomass ratio; Lg–IV, legume importance value; Pf–AGB%, perennial forb above-ground biomass ratio; Pf–IV, perennial forb importance value; Af–AGB%, annual and biennial forb above-ground biomass ratio; Af–IV, annual and biennial forb importance value; H, Shannon-Wiener index; R, Margalef index; CP, crude protein; ADF, acid detergent fiber; NDF, neutral detergent fiber; RFV, relative feed value; BD, bulk density; SW, soil water content; OM, organic matter; AN, alkaline hydrolysable nitrogen; AP, available phosphorus; AK, available potassium. ( b): Cover, community coverage; Height, community height; AGB, community above-ground biomass; BGB, community below-ground biomass; Gr–AGB%, grass above-ground biomass ratio; Gr–IV, grass importance value; Lg–AGB%, legume above-ground biomass ratio; Lg–IV, legume importance value; Pf–AGB%, perennial forb above-ground biomass ratio; Pf–IV, perennial forb importance value; Af–AGB%, annual and biennial forb above-ground biomass ratio; Af–IV, annual and biennial forb importance value; H, Shannon-Wiener index; R, Margalef index; CP, crude protein; ADF, acid detergent fiber; NDF, neutral detergent fiber; RFV, relative feed value; BD, bulk density; SW, soil water content; OM, organic matter; AN, alkaline hydrolysable nitrogen; AP, available phosphorus; AK, available potassium. Figure 6. Correlation coefficients among vegetation characteristics, soil properties and ecosystem multifunctionality (EMF) index. ( a): Cover, community coverage; Height, community height; AGB, community above-ground biomass; BGB, community below-ground biomass; Gr–AGB%, grass above-ground biomass ratio; Gr–IV, grass importance value; Lg–AGB%, legume above-ground biomass ratio; Lg–IV, legume importance value; Pf–AGB%, perennial forb above-ground biomass ratio; Pf–IV, perennial forb importance value; Af–AGB%, annual and biennial forb above-ground biomass ratio; Af–IV, annual and biennial forb importance value; H, Shannon-Wiener index; R, Margalef index; CP, crude protein; ADF, acid detergent fiber; NDF, neutral detergent fiber; RFV, relative feed value; BD, bulk density; SW, soil water content; OM, organic matter; AN, alkaline hydrolysable nitrogen; AP, available phosphorus; AK, available potassium. ( b): Cover, community coverage; Height, community height; AGB, community above-ground biomass; BGB, community below-ground biomass; Gr–AGB%, grass above-ground biomass ratio; Gr–IV, grass importance value; Lg–AGB%, legume above-ground biomass ratio; Lg–IV, legume importance value; Pf–AGB%, perennial forb above-ground biomass ratio; Pf–IV, perennial forb importance value; Af–AGB%, annual and biennial forb above-ground biomass ratio; Af–IV, annual and biennial forb importance value; H, Shannon-Wiener index; R, Margalef index; CP, crude protein; ADF, acid detergent fiber; NDF, neutral detergent fiber; RFV, relative feed value; BD, bulk density; SW, soil water content; OM, organic matter; AN, alkaline hydrolysable nitrogen; AP, available phosphorus; AK, available potassium. Table 1. Grass species composition of experimental plots and some management and quality indicators. Table 1. Grass species composition of experimental plots and some management and quality indicators. Species Name Ecological Economic Class Group Seeding Rate/ (kg·hm −2) Growing Period/ (d) Dactylis glomerata Loosely tufted grasses 20 180 Poa pratensis Loosely tufted grasses 14 180 Trifolium pretense cv. Minshan Stoloniferous legume forages 14 180 Medicago sativa cv. Xinmu No. 4 Fine-stemmed legume forages 14 180 Onobrychis viciifolia cv. QiTai Fine-stemmed legume forages 54 180 Elymus dahuricus Loosely tufted grasses 20 180 Bromus inermis cv. Wusu No. 1 Rhizomatous grasses 20 180 Table 2. Grass combination. Table 2. Grass combination. Grass Combination A1 Medicago sativa cv. Xinmu No. 4 10% + Trifolimm pratense cv. Minshan 10% + Dactylis glomerata 10% + Poa pratensis 10% + Bromus inermis cv. Wusu No. 1 50% + Elymus dahuricus 10% A2 Onobrychis viciifolia cv. QiTai 5% + Medicago sativa cv. Xinmu No. 4 5% + Trifolimm pratense cv. Minshan 10% + Dactylis glomerata 20% + Poa pratensis 20% + Bromus inermis cv. Wusu No. 1 40% A3 Medicago sativa cv. Xinmu No. 4 10% + Trifolimm pratense cv. Minshan 10% + Dactylis glomerata 15% + Poa pratensis 15% + Bromus inermis cv. Wusu No. 1 50% A4 Onobrychis viciifolia cv. QiTai 5% + Medicago sativa cv. Xinmu No. 4 15% + Dactylis glomerata 15% + Poa pratensis 15% + Bromus inermis cv. Wusu No. 1 50% A5 Onobrychis viciifolia cv. QiTai 5% + Trifolimm pratense cv. Minshan 15% + Dactylis glomerata 15% + Poa pratensis 15% + Bromus inermis cv. Wusu No. 1 50% A6 Trifolimm pratense cv. Minshan 10% + Medicago sativa cv. Xinmu No. 4 10% + Poa pratensis 30% + Bromus inermis cv. Wusu No. 1 50% A7 Medicago sativa cv. Xinmu No. 4 15% + Onobrychis viciifolia cv. QiTai 5% + Poa pratensis 30% + Bromus inermis cv. Wusu No. 1 50% A8 Medicago sativa cv. Xinmu No. 4 10% + Trifolimm pratense cv. Minshan 10% + Dactylis glomerata 30% + Bromus inermis cv. Wusu No. 1 50% A9 Trifolium pretense cv. Minshan 20% + Poa pratensis 30% + Bromus inermis cv. Wusu No. 1 50% A10 Medicago sativa cv. Xinmu No. 4 20% + Poa pratensis 30% + Bromus inermis cv. Wusu No. 1 50% CK Control Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. © 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license.
