Multivariate statistical analysis of potentially toxic elements in soils under different land uses: Spatial relationship, ecological risk assessment, and source identification.

The Qinghai-Tibet Plateau is one of the most fragile and susceptible areas to climate change and human disturbances in the world. Here, a total of 48 soil samples were obtained from areas of different land uses within a typical basin in eastern Qinghai-Tibet, China. The selected potentially toxic elements (PTEs, including Cd, Cr, Cu, Pb, and Zn) contents were analyzed to explore their spatial patterns, ecological risks, and then the effects of land use types on these elements were assessed by self-organizing map (SOM) and random forest regression (RFR) models, and the main sources were revealed using positive matrix factorization (PMF) model. Results showed that mean concentrations of selected PTEs in surface soils were higher than local background values and those of subsurface soils. The low-degree ecological risk was obtained with comparatively high risks in the north and south of the study area. The results of the SOM and RFR models revealed that land use types affected the redistribution of PTEs in surface soil. The PMF model demonstrated that these PTEs were mainly derived from natural sources (46.7%), traffic emissions (31.2%), and industrial and agricultural inputs (22.1%). Natural sources were the essential contributors for these soil PTEs, especially for Cr. In addition to natural sources, traffic sources made great contributions for Cd, Pb, and Zn elements, while the enrichment of Cu was mainly related to industrial and agricultural activities.

Influence of soil properties, topography, and land cover on soil organic carbon and total nitrogen concentration: A case study in Qinghai-Tibet plateau based on random forest regression and structural equation modeling.

The effects of topography, land cover type, and soil physicochemical properties on the distribution of soil organic carbon (SOC) and total nitrogen (TN) have drawn growing attention recently, but little is known about how these features-associated interactions impact SOC and TN. To elucidate how these interactions affect the preservation of carbon and nitrogen in soils, we used data-driven models (random forest regression and structural equation modeling) to identify the dominant environmental factors affecting the distribution of SOC and TN in two different soil layers (0-20 and 20-40 cm) of the Qinghai-Tibet plateau. In addition, an algorithm based on random forest ("Boruta") was chosen to identify the relevant influencing factors and partial dependence was used to depict the two most important factors. We found that rather than land cover type, environmental properties, such as soil physicochemical characteristics and altitude had the most significant effects on the distribution of SOC and TN. Our findings indicate that elevation and TN are the two most important factors influencing SOC in the surface and subsurface soil layers. Moreover, total potassium (TK) impacts TN content in the surface soil layer, but only in a specific range of concentrations, which could be attributed to anthropogenic activities such as applying nitrogen and potassium fertilizers to increase the yields of local food crop, Tibetan hulless barley. These findings provide a scientific perspective on soil nutrient preservation.