Soil carbon stocks on the Tibetan Plateau are widely considered to be increasingly threatened by drastic climate warming and intensified livestock grazing. But it remains elusive due to unconstrained model projections. Here we integrate large-scale soil campaigns, soil incubation with paired grazing experiments to project impacts of climate change and grazing on soil carbon stocks in a three-pool soil carbon model. While Tibetan soils will act as a carbon sink, over half of the gains occur in active or unprotected pools, making them vulnerable to extreme events and grazing. Although thermokarst processes may not reverse this trend, continued livestock grazing at current levels, or even a transition to a forage-livestock balanced state, could nearly offset climate-induced benefits. We highlight the critical need to optimize grazing to sustain soil carbon sinks on the Tibetan Plateau, and emphasize the importance of incorporating grazing impacts on soil carbon stocks into Earth system models.
Soil Carbon Database: Integrating 1608 soil samples from large-scale standardized field surveys on the Qinghai Tibet Plateau from 2019 to 2022, as well as 2562 soil carbon data from published studies over the past thirty years, with a total of 4170 observation data covering depths of 0-300cm, including multiple vegetation types such as alpine meadows and forests, and indicators such as carbon storage, bulk density, and pH.
Soil cultivation experiment data: Retrieve relevant literature on "soil carbon", "cultivation", and "Qinghai Tibet Plateau" from WoS and CNKI, screen 19 long-term cultivation experiment data under aerobic conditions lasting more than six months, and extract CO ₂ emission time-series data.
Grazing experiment data: Using keywords such as "soil carbon", "grazing", and "Qinghai Tibet Plateau", WoS and CNKI were searched to select 52 peer-reviewed studies, and 296 depth paired observation data were obtained, including carbon storage, vegetation biomass, soil texture, and other information of grazing and prohibited grazing sites.
Auxiliary dataset: including ESMs climate and NPP prediction data from CMIP6, CRU climate data, MODIS/GIMMS3g NPP data, vegetation distribution map of the Qinghai Tibet Plateau, permafrost types and other geographical environment data, as well as livestock population data from the China Statistical Yearbook.
Soil carbon storage simulation: Using climate, vegetation, and other variables, a hierarchical random forest model is used to construct carbon storage prediction models at different depths. After ten fold cross validation optimization, a spatial distribution map is generated by upscaling a 1km grid, and uncertainty is quantified using Bootstrap method.
Estimation of carbon turnover time: fitting the CO ₂ emission data of the cultivation experiment with a three reservoir carbon decomposition model to obtain the internal turnover time of the carbon reservoir. After temperature correction, using a random forest model to analyze environmental impacts, and then upscaling to the entire plateau.
Future carbon dynamic prediction: Carbon storage and turnover time data are incorporated into the three reservoir model, and after optimizing parameters, carbon storage changes in 2060 are predicted based on SSP2-4.5, SSP5-8.5 scenarios, and ESMs data, combined with thermal karst sensitivity assessment.
Grazing impact analysis: quantifying the effect of grazing on carbon storage using response ratio, identifying key driving factors using a random forest model; Design two grazing scenarios, estimate carbon storage changes based on future NPP data, and analyze their interaction with climate.
The data quality is good.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
