Multi-scale identification of driving forces for landscape ecological risk: A case study of the basin containing three plateau lakes in Yunnan, China

• The same driving factor plays different roles in influencing landscape ecological risk across different scales. • Integration of global and local areas for multi-scale driving factors identification. • Local areas are extracted from the landscape ecological risk's spatial-temporal evolution. • Local areas are categorized as deteriorated, improved, and stable. • Multi-scale studies comprehensively and accurately explore the drivers of landscape ecological risk. The continuous development of society, rapid urbanization, industrialization, and climate change have intensified conflicts between fragile ecological systems and limited natural resources. To rationalize land use planning and ensure sustainable landscape development, it is essential to evaluate landscape ecological risk (LER) and identify its drivers. Current approaches for identifying these drivers often focus on a single scale—the global—making it difficult to reveal their effects on LER. In this study, a multi-scale identification perspective in LER driving forces was conducted to solve the problem of too macroscopic of study. First, we developed an LER assessment model for the basin containing three plateau lakes (Fuxian, Xingyun, and Qilu) in the central Yunnan Province and identified local areas (deteriorated, improved, and stable). Second, the driving factors were classified into four categories based on anthropogenic and natural disturbances, with each category containing two indicators. Finally, we employed geodetector to identify the drivers of LER in the global and local areas. From the global perspective, the LER was reduced overall and was most affected by anthropogenic disturbances. The LER of the basin was most affected by elevation from 2000 to 2010, with the annual average temperature in 2020 being the strongest driver. The interactions between topography, climate, regional accessibility, and socioeconomic factors exacerbated the degree of LER. From the local perspective, the deteriorated area in the local areas experienced erosion due to high risk, progressing from southwest to northeast from 2000 to 2020. The LER level in the improved area gradually became complicated and multiple LER levels were intertwined. The stable area did not show significant internal changes, but the surrounding ecological environment improved. Anthropogenic disturbances were most intense in the deteriorated areas, whereas natural disturbances were more intense in the improved and stable areas. The results of extracting local areas for the identification of LER drivers were different from those of the global analysis. Therefore, we believe that relevant studies should be conducted from a multi-scale perspective to provide comprehensive and feasible references for authorities to formulate ecological protection strategies.