Ecosystem services (ESs) supply-demand balance is pivotal to understanding human-land interactions. However, it remains a core challenge to understand the effects of spatial mismatches on complex ESs interactions for ecological management. This study aims to systematically evaluate these interactions through an Integrated ESs supply-demand framework. Taking the Inner Mongolia grasslands as a model, we quantified the supply, demand, and balance of six key ESs for the 2000-2020 period. Through spatial clustering analysis, we identified ESs bundles and subsequently analyzed the trade-offs and synergies of ESs and their driving mechanisms within each bundle. Our analysis demonstrated a pronounced spatial-temporal heterogeneity and persistent regional imbalances in supply-demand of ESs in the region. While water yield supply declined, carbon sequestration and soil conservation improved. Although grassland carrying capacity and agricultural product output increased, regional disparities persist. ESs interactions exhibit pronounced bundle-specificity, influenced by internal variability within each bundle. The balances of carbon sequestration and food supply are trade-offs at the regional scale and within the central Inner Mongolia bundle, whereas they are synergistic within the water-scarce western bundle. Abiotic conditions dominate supply patterns, while human activities shape demand patterns, jointly shaping spatial-temporal ESs patterns including mismatches. The supply-demand spatial bundle provides operational perspective and precise management units for characterizing ESs interactions. Implementing management strategies targeting specific ESs bundles to mitigate critical trade-offs and enhance synergies offers a scientific pathway for the sustainable management of grassland socio-ecological systems.

Understanding how land-use affects runoff-pathway partitioning is important for controlling sediment and nutrient export from steep agricultural slopes. We established three replicate 30 m × 2 m runoff plots for each of three land-use types—cropland, pastureland, and shrubland—and separately monitored surface runoff, shallow lateral flow intercepted at 30 cm depth (SLF), sediment loss, and nitrogen (N) and phosphorus (P) export in a high-elevation mountainous region of southern China. Nine temporally clustered rainfall events spanning three rainfall-intensity classes were selected from a broader monitoring record. Cropland produced the highest mean runoff volume (8928 mL), compared with shrubland (3640 mL) and pastureland (2125 mL). Surface runoff accounted for 92.7% of the combined runoff from cropland, whereas surface runoff and SLF contributed more evenly in pastureland and shrubland. Under heavy rainfall, SLF exceeded surface runoff in both vegetated land-use types but remained a minor pathway in cropland. Cropland also produced the highest mean surface-runoff-associated sediment loss (1845 g), exceeding pastureland (434 g) and shrubland (238 g). Runoff-pathway redistribution differentiated nutrient export. Phosphorus loss from cropland was closely associated with sediment-bearing surface runoff, whereas SLF was the main pathway for total nitrogen (TN), water-soluble nitrogen (WSN), and nitrate nitrogen (NO₃−-N) loss from pastureland and shrubland. Nutrient concentrations in SLF did not differ significantly among land-use types, indicating that differences in SLF-associated N loss were mainly related to water volume. Targeted conversion of erosion-prone steep cropland to pastureland may therefore reduce surface runoff, sediment loss, and P export but should be combined with N management to limit soluble-N transport through shallow lateral pathways. These findings apply specifically to lateral soil water intercepted at 30 cm depth and should not be extrapolated to deep percolation or groundwater recharge.

Grasslands are widely recognized as important sinks of methane (CH4). However, their CH4 uptake capacity has been increasingly weakened driven by human activities and changes in precipitation regimes. In particular, the rapid expansion of livestock grazing can lead to substantial increases in CH4 emissions. For the Eurasian grasslands that have been subject to long-term grazing, how well they currently function and will continue to function as CH4 sinks under the changing precipitation regimes remains uncertain. Here we conducted an 11-year grazing-gradient experiment to assess the combined effects of grazing intensity and precipitation variability on grassland CH4 fluxes. We measured soil CH4 uptake in a manipulative grazing experiment in both wet and dry years, estimated livestock-derived CH4 emissions, and examined a range of biotic and abiotic drivers, including vegetation attributes, soil properties, and microbial biomass, for short- (prior to the onset of the lagged response) and long-term (the entire experimental period) influences on the dynamics of CH4 uptake in temperate meadow grasslands. Over the long term, CH4 fluxes are jointly regulated by precipitation and grazing, with overgrazing amplifying the suppressive effect of rainfall on CH4 uptake. Soil CH4 uptake showed a lagged response to grazing, which may have been triggered by extreme rainfall events. While moderate grazing sustained the long-term CH4 uptake, it fails to offset livestock-derived CH4 emissions. In the short term, CH4 fluxes are primarily governed by grazing, whereas in the long term they are jointly regulated by precipitation and grazing intensity. These results offer a new perspective for understanding the source-sink CH4 dynamics of grasslands in the context of ongoing climate change.

Grassland ecosystems are profoundly influenced by land management practices, yet the long-term mechanisms linking plant diversity and community stability remain unclear. In this study, we conducted three-year observational study to assess how enclosure, grazing, and mowing affect plant community dynamics through species turnover, niche overlap, and environmental drivers in the Hulunbuir temperate meadow steppe of Inner Mongolia. Using 198 permanent quadrats monitored over three consecutive growing seasons, we quantified α-, β-, and γ-diversity and assessed stability via biomass variability. Enclosure was associated with higher species richness and lower biomass variability, suggesting enhanced community stability under reduced disturbance, whereas grazing was associated with lower species richness and greater temporal and spatial variability in community structure, potentially linked to intensified species turnover and soil compaction. Mowing generally showed intermediate patterns, reflecting moderate alteration of competitive dynamics and community composition. Our analyses further revealed that soil physical properties and nutrient availability—particularly soil bulk density(SBD), total nitrogen (TN), and organic carbon (OC)—were key environmental factors associated with variation in plant diversity and stability across management regimes. Structural equation modelling based on observational data indicated that these environmental factors may influence diversity and stability both directly and indirectly, with pathways differing among management types. Our findings indicate that grassland management practices modulate diversity–stability relationships in a management-dependent manner, likely through their effects on species turnover, niche structure, and soil–plant feedbacks. These results highlight the importance of context-specific management strategies for sustaining grassland stability under ongoing environmental change. temperate meadow steppe; plant diversity; community stability; enclosure; grazing; mowing; environmental factor drivers

Monitoring sheep behaviour in natural pastures holds significant implications for precision livestock farming and animal welfare. Although numerous studies have developed algorithms for behaviour recognition in sheep, limitations remain regarding practical applicability and effective model deployment. In this study, motion sensors were employed to collect behavioural data from grazing sheep. A behavioural dataset is constructed encompassing four key behaviours (grazing, standing, walking, and lying down) acquired at three sampling frequencies: 20 Hz, 10 Hz, and 4 Hz. A deep learning architecture based on recurrent neural networks (RNNs) is proposed to classify these behaviours. The model's performance is evaluated using different RNN modules, and the impact of sampling frequency on classification accuracy is systematically explored. The results indicated that sampling frequency had a measurable effect on behaviour classification. The model trained on 20 Hz data achieved the highest accuracy. However, satisfactory performance was also obtained at 4 Hz, with only a 2.37% decrease in classification accuracy compared to the 20 Hz model. This finding suggests that lower-frequency data can be effectively used for behaviour recognition while significantly reducing the power consumption of wearable devices. Furthermore, incorporating data balancing techniques during model training was found to improve classification performance, particularly for underrepresented behaviours. The model combines accurate sheep behaviour prediction with daily pattern analysis, demonstrating high potential for the intelligent monitoring of grazing sheep.

Grazing exclusion is widely applied for restoring ecosystem structure and function in degraded grasslands caused by livestock overgrazing. While numerous exclusion experiments have been established to evaluate restoration effectiveness in degraded grasslands, the role of historical grazing intensity (i.e., initial grassland conditions before livestock removal) in regulating recovery effects remains largely overlooked. In our study, we selected three representative grazing intensities (LG: light grazing, MG: moderate grazing and HG: heavy grazing) and established enclosures within each grazing treatment to exclude livestock grazing. Over a three-year field investigation, we explored the responses of plant characteristics, functional group composition, and soil nutrient stocks to grazing exclusion across different historical grazing intensities. In the MG and HG plots, grazing exclusion significantly increased plant diversity and biomass inputs, consequently promoting the accumulation of soil organic carbon stock (SOCs) and total nitrogen stock (TNs). In contrast, grazing exclusion removed livestock-mediated nutrient returns and led to community simplification in the LG plot, which in turn decreased plant diversity and ultimately caused TNs loss. Random forest and variation partitioning analysis indicate that initial grazing intensity and exclusion duration are the critical factors influencing subsequent changes in SOCs and TNs after grazing exclusion. The significant negative correlations between the relative change of SOCs and TNs and their initial values further highlight the importance of initial grassland conditions in regulating soil nutrient stocks. These findings highlight that grassland restoration strategies should take into account historical disturbance intensity and local ecological conditions to ensure the effectiveness of restoration measures.

Mowing is a primary practice in temperate L. chinensis meadows, which are severely degraded due to frequent mowing, overgrazing, and other factors, necessitating restoration and sustainable management. The natural recovery of these grasslands hinges on their germinable soil seed banks, which form the basis for future productivity. Thus, germinable soil seed banks are critical for restoring overexploited meadows. In this study, we conducted germination experiments on 135 soil samples from various depths to comprehensively analyze the germinable seed bank under different mowing regimes. The main results were as follows: (1) the germinable soil seed bank density decreased significantly with a mowing event per year (C1), and the number of perennial grass seeds and upper grass seeds also decreased under the mowing event per year; (2) the size of the germinable soil seed bank increased under the other mowing regimes (control area without mowing or grazing, CK; mowing event every 2 years, C2; mowing event every 3 years, C3; and mowing event every six years, C6) relative to that under once-a-year mowing. With increasing soil depth, the number of germinable soil seeds decreased significantly. Most of the seeds in the germinable soil seed banks were distributed in the 0–2 cm soil layer, accounting for approximately 80% of the total, and at depths of 5–10 cm, the number of seeds of upper grasses was greater than that of perennial grasses. (3). During the mowing event each year, the seed bank of germinable soil seeds significantly decreased. Mowing every 2 years provides a one-year interval for natural vegetation growth, allowing for greater retention of seeds in the germinable soil seed bank. Mowing every 6 years significantly reduces the disturbance frequency, providing ample time for plant reproduction and resulting in the accumulation of germinable seeds in the soil.

Grassland ecosystems are particularly sensitive to human disturbances due to their relatively simple structure and limited resource availability. However, the responses of ecosystem carbon and water exchanges to grazing, the dominant human activity in grasslands, remain insufficiently understood. During growing seasons in 2023 and 2024, a grazing gradient experiment was conducted in a meadow steppe of northern China, incorporating four intensity levels: no grazing (CK), light grazing (LG), moderate grazing (MG), and heavy grazing (HG). Using the static chamber method, we assessed ecosystem carbon fluxes (GPP: gross primary production; ER: ecosystem respiration; NEE: net ecosystem CO2 exchange; Rh: soil heterotrophic respiration), water exchanges (ET: evapotranspiration; EP: soil evaporation), and resource use efficiencies (CUE: carbon use efficiency; WUE: water use efficiency). Results indicated that light grazing significantly enhanced NEE, CUE, and WUE compared to other treatments. In contrast, increasing grazing intensity markedly reduced carbon and water fluxes in MG and HG plots. Under grazing stress, aboveground biomass (AGB) was the primary determinant of GPP and ET changes, while ER was mainly influenced by soil microclimate and nutrients. GPP emerged as the key driver of NEE, CUE, and WUE variations. These findings highlight the contrasting roles of biotic and abiotic factors in regulating ecosystem functions and provide comprehensive evidence that light grazing could benefit carbon sequestration and resource use efficiency in the meadow steppe. Our study can offer practical and theoretical support for determining appropriate grazing intensity and promoting the sustainable management of grasslands in northern China.

Soil microbial communities play a crucial role in maintaining grassland ecosystem functions and are strongly influenced by livestock grazing. However, the long-term responses and driving mechanisms of soil microbial communities to grazing intensity gradients, remain largely unexplored. In this study, we investigated the mechanism of different grazing intensities (i.e., ungrazed, light, moderate and heavy grazing) affect the diversity and composition of soil bacteria and fungi in the Hulunbuir Leymus chinensis meadow steppe. Using a Bipartite network to represent indicative species shifts, bacterial community presented a clear succession along the grazing intensity gradient, likely linked to soil abiotic conditions (e.g. soil temperature, silt). In contrast, fungal community exhibited a more discrete shift along the grazing intensity gradient, challenging the traditional view that fungal community is more stable under disturbance. The shifts in fungal community were closely related to the vegetation composition and aboveground biomass, reflecting a typical bottom-up resource-related regulation, which were more dynamic than changes caused by abiotic conditions along the grazing intensity gradient. Interestingly, indicator analysis showed that higher grazing intensity shifted bacterial and fungal composition towards more oligotrophic (e.g. Dothideomycetes, Sordariomycetes, Leotiomycetes, and Chloroflexi, Thermoleophilia) and less copiotrophic (e.g. Saprotrophs, Bacteroides and subgroup_6). This shift reflects the depleted substrate and is consistent with the observed inhibition of ecosystem respiration, implying lower organic matter decomposition. The distinct patterns of bacteria and fungi responses provides novel insights into the mechanisms, through which grazing alters soil bacterial and fungal communities with potential long-term consequences, including future growth-limiting resource and soil environment conditions to withstand future disturbances, which affect soil bacterial and fungal communities differently and consequently modulate soil organic carbon turnover. Moreover, the different substrate affinity of copiotrophic and oligotrophic groups altered available and recalcitrant C decomposition, which may change soil carbon cycling and stocks.

The northern Chinese grasslands, serving as the cradle of traditional animal husbandry and pivotal ecological bulwarks, performing crucial ecological roles, including climate regulation, biodiversity conservation, hydrological replenishment, carbon sequestration, and amelioration of wind and sand erosion. The arid and semi-arid climatic conditions of the northern grasslands impose inherent limitations on their resilience to disturbances and their capacity for regulatory response. In recent decades, the escalating livestock load, compounded by the ramifications of global climate change, has precipitated varying levels of degradation, desertification, and salinization, thereby diminishing the ecosystemic functionality of these grasslands. The spatial heterogeneity of the grassland flora is a precursory indicator of grassland degradation and desertification. Consequently, the investigation of the spatial heterogeneity patterns of various grassland vegetation under the influence of long-term experimental grazing intensities is of paramount importance. Such research is instrumental in delineating the optimal grazing intensities for diverse grassland types, in the scientific delineation of grassland production and utilization paradigms, in the formulation of rehabilitation strategies for degraded grasslands, and in the enhancement of grassland ecosystem services. This, in turn, is essential for safeguarding the ecological security of China's northern frontier. The transformation of vegetation spatial patterns profoundly impacts the microclimatic conditions within plant community patches, affecting both biotic and abiotic processes. This study leverages a long-term controlled grazing experiment to investigate how varying grazing intensities influence the spatial configuration and response traits of four distinct types of grassland plant patches in northern China: desert steppe, typical steppe, meadow steppe, and alpine meadow. High-resolution drone imagery and advanced machine learning algorithms were employed for this analysis. The findings reveal a two-tiered impact based on the type of grassland and the level of grazing intensity. At the species level, increased grazing intensity leads to the consistent dominance of perennial tufted grasses in desert steppes, along with an expansion of bare soil patches and enhanced connectivity between these two elements. In typical steppes, the formerly dominant Leymus chinensis patches wane in both dominance and connectivity, while perennial bunchgrass and subshrub patches increase in connectivity. Similarly, in meadow steppes, Leymus chinensis loses dominance, with degenerated forb patches gaining prominence and connectivity. In alpine meadows, Stipa aliena patches gradually cede dominance to Kobresia humilis, which also shows increased connectivity. From a landscape perspective, desert steppe vegetation exhibits diminished spatial heterogeneity with increased grazing intensity. In contrast, typical steppe vegetation reaches peak spatial heterogeneity under light grazing. Meadow steppe vegetation heterogeneity tends to increase, while alpine meadow vegetation maintains relatively stable spatial heterogeneity regardless of grazing intensity. These observations suggest that different grassland types respond variably to grazing pressures under diverse climatic conditions. This study underscores the efficacy of high-resolution drone remote sensing and machine learning technologies in assessing the impact of grazing intensity on grassland spatial heterogeneity. It also highlights the importance of a detailed plant patch classification system in understanding the pivotal role of grazing in reshaping vegetation spatial patterns. The insights gained contribute to a better understanding of the complex dynamics between grazing practices and grassland ecosystems, offering valuable information for developing sustainable grazing management strategies.

Biological diversity in food and agriculture, and the resilience of agroecosystems and sustainability are vital for food security. However, economic development and urbanization have led to landscape fragmentation, eroding biodiversity and threatening its conservation. This study assessed the resilience of agroecosystems in three distinct communities in the Beijing-Tianjin-Hebei region of China: WangJinzhuang (Shexian County), Yuershan Farm (Fengning County), and Qingguang Village (Tianjin). The assessment applied the Integrated Valuation of Ecosystem Services and Tradeoffs with the Global Biodiversity Model for Policy Support (InVEST-GLOBIO) model to evaluate mean species abundance (MSA) and the Socio-Ecological Production Landscapes and Seascapes (SEPLS) framework to assess the resilience of agroecosystems. Between 1992 and 2022, WangJinzhuang's MSA declined slightly from 0.445 to 0.444, Yuershan Farm saw a larger drop from 0.129 to 0.110, while Qingguang Village improved from 0.037 to 0.051. WangJinzhuang had the highest MSA and resilience score (3.9), followed by Qingguang Village (3.84), while Yuershan Farm had the lowest (3.28). The findings highlight higher the resilience of agroecosystems in WangJinzhuang and Qingguang Villages' self-sufficient systems compared to Yuershan Farm. To promote agrobiodiversity conservation, locally tailored policies and strategies must include: adopting eco-friendly farming; optimizing farming systems for regional planning; enhancing agro-ecotourism, and establishing agrobiodiversity conservation compensation mechanisms. Addressing community-specific challenges, while fostering agroecosystem resilience and sustainable development, highlights the critical interplay between human livelihoods and ecological systems, paving the way for long-term, adaptive management in rural and transitional landscapes.

Grassland ecosystem functions are severely threatened by intensified grazing. Fine roots, which contribute over 33

Wind erosion poses a significant challenge to agricultural sustainability in Northern China’s arid regions. This study investigated the effectiveness of alfalfa grassland versus conventional cropland in controlling wind erosion across nine study sites in three agroecological regions. Using Sentinel-2 satellite imagery and the Revised Wind Erosion Equation (RWEQ) model, we analyzed vegetation cover duration and quantified soil wind erosion from 2018 to 2020. The results showed that alfalfa grassland extended vegetation cover by 80 days annually compared to cropland, with most extension occurring in spring. Alfalfa grassland demonstrated superior erosion control, reducing soil losses by 50% (24.02 versus 50.70 t/ha/yr) and increasing soil retention threefold (1.52 versus 0.59 t/ha/yr) compared to cropland. The northwest region experienced the highest erosion rates, while management practices significantly influenced alfalfa’s soil conservation effectiveness. Multiple regression analysis revealed vegetation cover and annual precipitation as primary factors affecting wind erosion. These findings suggest integrating alfalfa into crop rotations could effectively enhance soil conservation in Northern China’s wind erosion-prone regions.

The vegetation community is one of the most critical factors determining grassland productivity. The diversity of functional traits at the community level is crucial for maintaining the structure and stability of grassland ecosystems. Therefore, understanding grassland community dynamics is essential for comprehending grassland productivity. Currently, a range of ecosystem process models are effectively simulated vegetation dynamics and predicting productivity. However, these process-based models in grasslands do not consider the complex community composition, often only differentiating between C3 and C4 plants. This limitation can lead to biased estimations and predictions of grassland productivity if the effect of plant functional trait diversity is not considered. This study constructed a grassland dynamic vegetation model based on plant functional traits, including different plant functional groups within the framework of the carbon-water-nitrogen cycling processes in grassland ecosystems. After validating the model with the long-term observations, we predicted the community change and the production potential under scenarios that alleviate water and nitrogen limitations. The simulated above-and below-ground biomass had RMSE values of 19.48 g C m(-2 )and 97.31 g C m(-2), respectively. The soil water content during the growing season had an R-2 of 0.689 compared to the observations. The modeling analysis showed that in the Hulunbeier meadow grassland, net primary productivity responded more rapidly to nitrogen alleviation than water alleviation. This is because functional groups responded differently to water and nitrogen resources. The portion of perennial tall grasses(PTG) significantly increased following nitrogen alleviation, while perennial tall forbs(PTF) were more sensitive to soil water alleviation than other plant functional groups. Therefore, fertilization can rapidly increase grass productivity by increasing the biomass of PTG, with a saturation threshold at 20 g N m(-2). However, long-term simulations indicated that the effects of water and nitrogen alleviation on improving grassland net primary productivity gradually weakened overtime. In the equilibrium state, the enhancement of net primary productivity by water alleviation was greater than nitrogen alleviation. Additionally, water alleviation reduced interannual variability in grassland productivity. Across all scenarios, the high canopy functional group consistently dominated community biomass changes due to its competitive advantage in light acquisition. Based on the simulation results of grassland production potential, the net primary productivity of grasslands can reach 427.1 g C m(-2) yr(-1) under natural conditions. Simultaneous alleviation of both soil water and nitrogen resulted in a net primary production potential of 1,323 g C m(-2) yr(-1). The modeled biomass and community dynamics in each scenario were consistent with field nutrient addition experiment, suggests that modeling grassland ecosystem processes based on plant functional traits can effectively estimate and predict grassland productivity. This model provides a new method to simulate interactions between functional groups and ecosystem biogeochemical cycles.

Grazing plays a pivotal role in shaping the carbon dynamics within grassland ecosystems. Although the impact of grazing on soil carbon dynamics has recently become a major focus, the mechanistic drivers of grazing effects on functionally distinct carbon fractions remain unclear. Here, we employed a combined density and particle-size fractionation approach to divide the soil carbon pool into three distinct functional fractions (fPOC: free particulate organic carbon, oPOC: occluded particulate organic carbon, and MAOC: mineral-associated organic carbon), and investigated their responses to four different grazing intensities (Non-grazing: G0.00 (0 AU ha(-1)), light grazing: G0.23 (0.23 AU ha(-1)), moderate grazing: G0.46 (0.46 AU ha(-1)) and heavy grazing: G0.92 (0.92 AU ha(-1)), where 1 AU = 500 kg of adult cattle) and further explored the potential mechanisms involved in the Inner Mongolia meadow grassland. Our results show that POC stock has greater sensitivity to grazing disturbance than MAOC stock. Light to moderate grazing promoted the increase in both fPOC and oPOC stocks (fPOCs, oPOCs) compared to non-grazing, while heavy grazing (G0.92) significantly decreased relative to moderate grazing. In contrast, all intensity levels decreased in mineral-associated organic carbon stock (MAOCs). The structural equation model (SEM) indicates that grazing increases POCs by increasing belowground biomass input and suppressing microbial processes (microbial and enzyme activities). In addition, grazing-induced soil environment deterioration and nutrient depletion inhibit the input of microbial biomass and necromass, ultimately reducing MAOC formation. Furthermore, the distribution of POC increased significantly with grazing intensity, indicating enhanced SOC activity. Overall, our results highlight that grazing-induced shifts in plant above-belowground biomass allocation strategy and microbial enzymatic activity collectively drive soil carbon dynamics.

Livestock feeding behavior and intake play a crucial role in influencing grassland health and productivity. A comprehensive investigation into livestock feeding behavior and intake can effectively elucidate the interactions and impacts of livestock and grasslands, providing scientific evidence and technical support for the formulation and implementation of sustainable grassland development strategies. Based on a long-term controlled grazing experiment platform conducted over 13 years, the feeding behavior and forage intake of cattle under different grazing intensities were observed and analyzed. Additionally, we used GPS sensors to study cattle grazing behavior trends. Using Mantel's test, we analyzed the relationship between cattle movement distance, forage intake, and environmental factors. The results demonstrated that cattle forage intake decreased with increasing grazing intensity. Forage intake peaked at the end of July and beginning of August, with the highest efficiency observed in August. Moreover, under light grazing intensity, cattle exhibited greater fluctuations in forage intake than those under moderate and heavy grazing intensity. Cattle movement levels increased with higher grazing intensity, and during the period of lush grass growth, cattle displayed significantly higher movement levels than during grass senescence. The accuracy of the behavior determination model based on cattle velocity ranged from 60 to 80 %. Using this model, we found that under heavy grazing conditions, cattle spent significantly more time roaming than under light and moderate grazing. Conversely, under light grazing conditions, cattle spent significantly more time feeding. A negative correlation was identified between cattle forage intake and movement distance. Cattle's forage intake was significantly positively correlated with grass height and grass biomass and significantly negatively correlated with stocking rate and movement distance. Thorough research on live-stock feeding behavior and intake offers scientific evidence and technical support for formulating and implementing sustainable grassland development strategies.

In the face of a series of challenges, such as climate change, population growth, and agricultural intensification, as well as the issue of how to promote sustainable development and guarantee food security, biodiversity, with its unique genetic, ecological, and traditional socio-cultural values, has become an important way to solve this dilemma. Urban biodiversity has continued to decline in recent decades due to rapid urbanization. The agroecosystem health of the Beijing-Tianjin-Hebei region, a typical urban agglomeration economic area, is facing a critical situation. Therefore, assessing the potential of ecosystem diversity in the Beijing-Tianjin-Hebei region and exploring the assessment mechanisms and methods of ecosystem health can provide theoretical support for biodiversity conservation and utilization. In this thesis, the overall ecosystem health of the Beijing-Tianjin-Hebei region was assessed based on the land cover data from 1992 to 2022 and the projected land cover data up to 2032, as well as using the habitat quality indicated by the Fragstats and InVEST models and the landscape pattern index, habitat quality, and mean species abundance (MSA) indicators of the GLOBIO module. The main results are as follows: Habitat quality and mean species abundance (MSA) in the Beijing-Tianjin-Hebei region were observed to show a continuous downward trend over 40 years from a landscape level perspective, and landscape fragmentation due to urbanization was the main reason. Habitat loss and habitat degradation caused by landscape fragmentation led to a decline in biodiversity. The spatial distribution of habitat quality in the Beijing-Tianjin-Hebei region is closely correlated with topography and landscape, being higher in the northwest and lower in the southeast, forming a clear spatial pattern that declined from 0.599 to 0.564 between 1992 and 2032. The mean species richness (MSA) value of the Beijing-Tianjin-Hebei region is significantly affected by infrastructure, especially road construction. With the continuous expansion of the road network, the MSA values in the region generally show a decreasing trend from 0.270 to 0.183 between 1992 and 2032. Based on the above results, it is recommended to carry out several aspects of agrobiodiversity conservation and ecosystem restoration.

The effects of grazing on the cycling of carbon (C), nitrogen (N) and phosphorus (P) in grassland ecosystems are complex. Uncertainty still exists as regards the allocation of C, N and P storage amounts in grazed ecosystems in Inner Mongolia, situated at the eastern end of the Eurasian dryland. Based on the long-term cattle grazing experimental platform in the Hulun Buir meadow steppe of Inner Mongolia, a 3-year (2019-2021) field control experiment was conducted to assess how the grazing intensity influenced the quantities of C, N and P stored in canopy biomass, root, litter and soil compartments. We examined the relationships between the different pools and their regulatory pathways at the ecosystem level across six grazing intensities. In general, grazing increased the aboveground N and P contents but decreased the aboveground biomass C content and nutrient storage amounts in aboveground biomass, roots and litter. The grazing intensity of 0.34 AU ha-1 increased soil organic carbon, total nitrogen and total phosphorus storage amounts, with the soil accounting for 98 % of total reserves on average. Grazing affected soil pH， nutrient contents, above- and belowground biomass and soil environmental factors such as soil bulk density, which in turn affected C, N and P storage in the ecosystem according to the results of the structural equation model; therefore, grazing intensity can be an important factor regulating the input and output of nutrients in the ecosystem. In the future, for adaptive management of grasslands, moderate grazing could effectively increase C, N and P storage in meadow steppe ecosystems and ensure the nutrient balance and long-term sustainable development.

The forage-livestock balance is an important component of natural grassland management, and realizing a balance between the nutrient energy demand of domestic animals and the energy supply of grasslands is the core challenge in forage-livestock management. This study was performed at the Xieertala Ranch in Hulunbuir City, Inner Mongolia. Using the GRAZPLAN and GrazFeed models, we examined the forage-livestock energy balance during different grazing periods and physiological stages of livestock growth under natural grazing conditions. Data on pasture conditions, climatic factors, supplemental feeding, and livestock characteristics, were used to analyze the metabolizable energy (ME), metabolizable energy for maintenance (MEm), and total metabolizable energy intake (MEItotal) of grazing livestock. The results showed that the energy balance between forage and animals differed for adult cows at different physiological stages. In the early lactation period, although the MEItotal was greater than MEm, it did not meet the requirement for ME. MEItotal was greater than ME during mid-lactation, but there was still an energy imbalance in the early and late lactation periods. In the late lactation period, MEItotal could meet ME requirements from April–September. Adult gestational lactating cows with or without calves were unable to meet their ME requirement, especially in the dry period, even though MEItotal was greater than MEm. Adult cows at different physiological stages exhibited differences in daily forage intake and rumen microbial crude protein (MCP) metabolism, and the forage intake by nonpregnant cows decreased as follows: early lactation > mid-lactation > late lactation, pregnant cows’ lactation > dry period. For the degradation, digestion and synthesis of rumen MCP, early-lactation cows were similar to those in the mid-lactation group, but both were higher than those in the late-lactation group, while pregnant cows had greater degradation, digestion, and synthesis of MCP in the lactation period relative to the dry period. For lactating cows, especially those with calves, grazing energy requirements, methane emission metabolism and heat production were highest in August, with increased energy expenditure in winter. Overall, grazing energy, methane emissions and heat production by dry cows were low. In the context of global climate change and grassland degradation, managers must adopt different strategies according to the physiological stages of livestock to ensure a forage-livestock balance and the sustainable utilization and development of grasslands.