The Qinghai-Tibetan Plateau (QTP), crucial for water resources and ecosystem services, is largely covered by alpine grasslands supporting millions of livestock and pastoralists. These grasslands are fragile and highly sensitive to climate change and human activities, particularly overgrazing. Historically, sustainable grazing practices prevailed, but recent population growth and intensification of livestock farming have led to widespread grassland degradation. This degradation manifests as a decline in plant and soil quality, altered ecosystem composition, and reduced functionality. Understanding the interactions between climate change, human activities, and grassland health is critical for sustainable development and effective management strategies. This study aims to develop a "stocking rate threshold," a critical level of grazing intensity that triggers extreme grassland degradation. By using a process-based ecosystem model, this study seeks to identify regions at risk and predict the timing of potential degradation under various scenarios, including different grazing intensities and future climate conditions.

Numerous studies have examined the local impacts of grazing on QTP grasslands, focusing on vegetation cover, species diversity, primary production, soil organic carbon, soil respiration, and nutrient dynamics. However, extrapolating these local findings to the regional scale remains challenging, hindering the development of effective, large-scale grazing management strategies. Process-based ecosystem modeling, which integrates biogeophysical, biogeochemical, and ecological processes, offers a promising approach to simulate interactions among vegetation, climate, and human activities, providing a more holistic understanding of ecosystem dynamics. Existing research on QTP grasslands has highlighted the significant impact of overgrazing, climate change, and the effects of government programs aimed at restoring grasslands. However, a lack of comprehensive, region-wide analysis of a stocking rate threshold that predicts degradation remains a gap in the research. This paper seeks to address this gap and provide an early warning system for sustainable grassland management.

This study employs a modified version of the TRIPLEX-GHG process-based ecosystem model, incorporating a grazing module. The model simulates biogeophysical and biogeochemical processes, vegetation dynamics, and greenhouse gas emissions, allowing for the assessment of grassland responses to grazing and climate change. Net primary productivity (NPP) is used as a key indicator of grassland degradation. The stocking rate threshold is defined as the grazing intensity at which NPP falls to 1% of the pre-grazing level, signifying extreme degradation. The study uses multi-year average stocking rate data at the county level (1980-2017), obtained from provincial and prefecture statistical yearbooks, converted into sheep units (SU) per hectare per year. These data are combined with grassland distribution maps and information on grazing seasons to create dynamic maps of stocking rate. Spatial patterns of the stocking rate threshold are then mapped across the QTP using model simulations under varying grazing intensities (0-15 SU ha⁻¹ year⁻¹, incremented by 0.5) with average climate conditions (1980-2017). Future scenarios are also explored using representative concentration pathways (RCPs) 2.6, 4.5, and 8.5, representing different levels of greenhouse gas emissions. The model's performance is validated by comparing simulated NPP and GPP with observed data from field experiments, eddy covariance measurements, and remote sensing datasets (MODIS and AVHRR). Comparisons are made at both the county level and grid scale.

The analysis reveals a spatial gradient in stocking rate across the QTP, generally increasing from northwest to southeast. Stocking rate thresholds also exhibit a similar pattern, reflecting regional differences in climate and productivity. Approximately 80% of QTP grasslands currently have stocking rates below the threshold, but in 55% of these areas, the rate exceeds half the threshold, indicating a potential risk. Model projections suggest that climate change and CO2 fertilization can partially offset the negative impacts of grazing in ~70% of grasslands, but only in areas below the threshold. If the stocking rate remains at the threshold, the model predicts grassland degradation within 20 years in the northwest and over 80 years in the south and southeast. Overgrazed areas show a rapid degradation timeline. The study suggests that maintaining a stocking rate no more than 60% (50-70% range) of the threshold could balance human demands with grassland protection, even under climate change. Future climate scenarios (RCPs) project an increase in stocking rate thresholds, particularly in the south and southeast, highlighting the ongoing need for careful grazing management even under future beneficial climate conditions.

The findings highlight the complex interplay between grazing intensity, climate change, and grassland degradation on the QTP. The defined stocking rate threshold offers a valuable early warning signal for sustainable grazing management. The results emphasize the importance of considering both current and future climate conditions when planning grazing strategies. The significant portion of grasslands with stocking rates exceeding half the threshold underscores the urgency of implementing adaptive management practices. The model's ability to predict the time until degradation provides valuable insights for timely interventions. The study's limitations should be noted, such as the model's assumptions and potential uncertainties related to data availability and model complexity, particularly around compensatory growth and the impact of other human activities beyond grazing.

This study provides a novel approach to assess grassland degradation risk on the QTP using a process-based model and a defined stocking rate threshold. The findings emphasize the spatial heterogeneity of degradation risk and the potential mitigating effects of climate change in areas with sustainable grazing practices. The model's predictive capacity allows for timely interventions to prevent irreversible grassland degradation. Future research should refine the model by incorporating additional factors, such as other human activities, and improve data accuracy to enhance the precision of threshold estimations and predictions.

The study acknowledges several limitations. The model's accuracy depends on the quality of input data, and uncertainties exist in parameter estimations and model representation of complex ecological processes. The focus on NPP as a single indicator of degradation may not fully capture the complexity of grassland ecosystem health. The model does not account for all human activities impacting grassland productivity. The definition of the stocking rate threshold, while useful, is simplified and may not capture the nuances of grassland resilience in all contexts. Further research is needed to fully understand the complex dynamics of QTP grasslands and refine these methods to improve management decision-making.