Spatiotemporal dynamics and driving factors of cultivated land fragmentation across China from 1990 to 2020

Cultivated land fragmentation (CLF) undermines agricultural productivity and food security, yet its spatiotemporal dynamics and driving forces remain poorly understood in diverse regions such as China. This study aims to quantify the spatiotemporal dynamics of CLF across China from 1990 to 2020, using a multidimensional Cultivated Land Fragmentation Index (CLFI) as its measurement, and to identify the dominant natural and anthropogenic factors across different agricultural regions. We developed a multidimensional CLFI integrating land cover data and landscape metrics. CLF patterns were analyzed using the Space Time Cube and Theil–Sen trend estimation. Driving mechanisms were examined through GeoDetector and Multiscale Geographically Weighted Regression (MGWR), using 12 candidate variables representing topography, climate, soil, and human activity. CLFI initially declined during 1990–2000 (65.2% of counties) but reversed after 2000, with 83.8% of counties exhibiting an increase in fragmentation by 2010–2020. Human activity indicators—especially night-time light, impervious surface ratio, and road density—were the most influential nationwide, while elevation and soil conditions remained key constraints in mountainous and ecologically sensitive regions. GeoDetector revealed strong interactive effects (e.g., nighttime light × elevation, q = 0.52), and MGWR identified substantial spatial heterogeneity in driving forces. CLF in China results from a dynamic interplay of anthropogenic expansion and natural constraints, with evident regional divergence. These findings provide critical insights for region-specific land consolidation, ecological zoning, and farmland protection strategies to support sustainable agriculture and food security.