Landscape Patterns and Drivers of Farmland Evolution in Basin Margin Mountainous Areas—A Case Study of Sichuan Basin, China

ABSTRACT Rapid urbanization has resulted in increasingly complex and diverse farmland landscape patterns in mountainous areas. The southern margin of Sichuan Basin, noted for its prominent urban–rural‐nature gradient, serves as a typical area for analyzing farmland landscape evolution. In this study, an identification system was developed to classify and identify the typology of farmland evolution, including farmland fallow (FF), farmland abandonment (FA), farmland loss (FL), and farmland reclamation (FR). Multiscale geographically weighted regression (MGWR) was employed to analyze the drivers and spatial differentiation mechanisms of farmland evolution. The results revealed the following. (1) The areas of FR (45.39%), FF (29.62%), and FA (21.81%) in the region were relatively large, and the FL area (3.17%) was relatively small. Compared with 2000–2010, the areas of FL and FA increased, and the areas of FF and FR decreased from 2010 to 2020. (2) FL, FR, FA, and FF presented similar landscape patterns during the two periods. FF and FA showed lower fragmentation in lowland plain areas (LPA) compared to mid‐high mountainous areas (MHMA). In contrast, FL had higher fragmentation in LPA, while FR had greater fragmentation in low mountainous and hilly areas (LMHA); (3) The landscape pattern and drivers of farmland evolution exhibit clear urban‐rural‐nature gradients. From the LPA to the MHMA, the dominant farmland evolution typologies are sequentially FL, FR, and FF‐FA. Concurrently, the primary driving factors shift from socioeconomic factors (urbanization and policies) to more natural factors (terrain and ecological conservation). This study proposes an integrated framework for managing farmland in mountainous regions, considering regional policy trade‐offs and synergies. It can guide the sustainable use and protection of farmland, while supporting rural revitalization and agroecological sustainability.