[Characteristics of Vegetation and Soil of Degraded Grasslands and Their Relationships in Xizang]

Abstract

To investigate the characteristics of grassland degradation on a regional scale in Xizang, data on grassland degradation from the second grassland survey of Xizang and 12 vegetation and soil indicators from the National Tibetan Plateau Data Center were collected. Using ArcMap, 10 000 random sample points were selected on raster data （excluding non-grassland, desertification, and salinization data, leaving 7 949 valid sample points）. The multi-value extraction to-point method was applied to extract degradation and indicator data for each sample point. The characteristics of degraded grassland vegetation and soil and their relationships were analyzed in Xizang. Moreover, random forest modeling was conducted to predict the trend of grassland ecosystem changes. The results indicated that： ① The grasslands in Xizang were primarily composed of alpine steppe and alpine meadow types, accounting for 45.83% and 41.15% of the valid sample points, respectively. ② With the intensification of grassland degradation, the number of steppe-type species among the 17 grassland types gradually decreased, and the proportion of steppe dominated by species such as Stipa purpurea and Carex moorcroftii decreased, whereas the proportion of miscellaneous grasses and Dasiphora fruticosa increased. ③ As the degree of degradation increased, vegetation indicators generally showed a declining trend, with soil total nitrogen, total phosphorus, total potassium, and organic carbon decreasing, whereas soil pH and bulk density increased, and soil moisture content was not significant. ④ A positive correlation exists between soil moisture content, total nitrogen, total phosphorus, total potassium, organic carbon, vegetation cover, net primary productivity of vegetation, normalized difference vegetation index, aboveground biomass, and habitat quality. However, there was a negative correlation between pH and soil bulk density, and the correlation coefficients among various indicators decreased with the intensification of degradation. ⑤ The random forest simulation results showed that during the degradation process, the contribution rates of soil bulk density and habitat quality both exceeded 12%, with the model prediction accuracy reaching 78%. The study revealed that grassland degradation in Xizang was closely related to soil bulk density and habitat quality, indicating that higher soil bulk density or lower habitat quality may correspond to more severe grassland degradation. This provides a scientific basis for future grassland conservation and management.

Keywords: Xizang grassland degradation; correlation; random forest; soil physical and chemical properties; vegetation indicators.