Soil heavy metals and their associated ecological risks are significant environmental issues, yet comprehensive studies are limited in ecologically sensitive regions such as the Qinghai-Tibet Plateau. This study focuses on Yushu County and utilizes principal component analysis-multiple linear regression (PCA-MLR) and network environmental analysis (NEA) models to identify pollution sources and ecological risk transmission pathways. The results indicated that the heavy-metal content in the soil of the study area predominantly reflects natural background levels, with minimal anthropogenic influence. The PCA-MLR model identified five main factors influencing soil composition: acidic rocks, basic rocks, agricultural/pastoral activities, hydrothermal processes, and crustal backgrounds. The NEA model revealed that cadmium and arsenic posed the highest ecological risks, primarily accumulating in soil microorganisms (67.1 %), while vegetation, herbivores, and carnivores accounted for 25.2 %, 4.6 %, and 3.1 % of the potential risk, respectively. Acidic and basic rock sources were major contributors, with acidic rocks posing the highest risk of arsenic input to soil microorganisms. The environmental capacity of soil to absorb Cd and As is approaching its limit, highlighting the urgent need to address their pollution risks. This study provides critical insights into geogenic heavy-metal risks, offering a framework for sustainable management in fragile ecosystems.