Modelling α-Diversity and Abundance of Soil Bacterial and Fungal Communities in Tibetan Alpine Grasslands Using Climate Data and Normalized Difference Vegetation Index

High-precision models enable the visualization of spatiotemporal patterns of soil microorganisms in the Qinghai- Xizang Plateau, a region highly sensitive to global change. However, large-scale, high-resolution data on soil bacterial and fungal communities, particularly in relation to the ecological impacts of fencing, remains relatively scarce, thereby limiting studies on the spatiotemporal distribution patterns of soil fungi and bacteria. This study constructed models of soil fungal and bacterial abundance and α-diversity (total abundance, major phylum abundance, primary functional group abundance, and species/phylogenetic/functional α-diversity) in the alpine grasslands of the Qinghai-Xizang Plateau, based on observational data and utilizing the random forest method. Under fencing conditions, the variations in soil fungal and bacterial abundance and α-diversity were collectively explained by temperature, precipitation, and radiation, accounting for 46%–88% of the variation; while under grazing conditions, the inclusion of NDVI allowed these four factors to explain 47%–92% of the variation. The relative biases between simulated and observed values for fungal and bacterial abundance and α-diversity were lower than 10.83%, and the simulated values explained the variation in observed values ranging from 75% to 100%. The slopes of the linear regressions between observed and simulated values ranged between 0.81 and 1.00. Consequently, the random forest models constructed in this study demonstrated high accuracy and they can predicted the total abundance, major phylum abundance, primary functional group abundance, and α-diversity of soil microorganisms in alpine grasslands. These models can serve as a robust foundation for future related research.