From scale conflict to coordination: Multi-scale spatiotemporal correlation of cultivated land conversion to non-agricultural uses and implications for adaptive land-use governance in mountainous Yunnan, China

Le résumé fourni par la source

Identifying factors associated with cultivated land non-agriculturalization (CLN) is essential for food security and sustainable development, yet most existing studies neglect spatiotemporal non-stationarity and scale dependence, limiting both theoretical understanding and policy precision. To address this gap, this study develops an integrated "spatiotemporal–scale–mechanism" analytical framework and applies it to Yunnan Province, China, a region characterized by pronounced topographical and ecological heterogeneity. Using complete annual panel data for 2000–2022, we fitted one unified Geographically and Temporally Weighted Regression model at each of two nested administrative scales—prefecture and county—and summarized the resulting location- and year-specific coefficients across three development periods. Three principal findings emerge: (1) CLN exhibits marked spatiotemporal non-stationarity, with driver magnitude and direction varying across both space and time; (2) a clear scalar shift in driver dominance is confirmed — macroeconomic factors dominate at the prefecture scale while biophysical endowments prevail at the county scale, providing empirical evidence for the ecological fallacy; and (3) cross-scale analysis documents a governance transition from scale conflict toward scale coordination, reflected in declining spatial mismatches and expanding 'oasis within hotspot' patterns. These findings demonstrate that CLN governance cannot be adequately informed by stationary or single-scale analyses. More effective cultivated-land protection requires multiscale, adaptive, and spatially differentiated strategies that account for both macro-level development pressures and locally specific environmental constraints, while recognizing that the identified relationships are associational rather than causal.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.