Multi‐Objective Spatial Optimization of Cropland Layouts in Highly Urbanized Regions: Evidence From Dongguan, China

Cropland protection in highly urbanized regions requires reconciling stable quantitative‐qualitative targets with the functional optimization of extremely fragmented spatial structures. This study develops a patch‐level multi‐objective spatial reconfiguration framework of cropland layouts based on the Non‐dominated Sorting Genetic Algorithm‐II (NSGA‐II) to balance fragmentation mitigation and land conversion costs under constraints of quantity balance, cropland quality non‐degradation, spatial feasibility, and adjustment priorities. The method integrates multi‐scenario cropland adjustment scales with three patch‐level adjustment modes, that is, internal filling, edge expansion, and patch connection. Validation in Dongguan, a representative highly urbanized city in China, demonstrates the framework's effectiveness in generating spatially realistic layouts through a well‐distributed Pareto front. Results indicate that while larger adjustment scales substantially improve landscape aggregation, land conversion costs escalate non‐linearly; the highest‐intensity scenario incurred costs over six times those of the smallest. Across all scenarios, edge expansion served as the primary cost‐effective pathway for consolidation, while internal filling contributed to resolving patch perforations and stabilizing core areas. Patch connection remained indispensable for bridging isolated fragments and ensuring systemic landscape integrity. However, sensitivity analysis showed that optimization outcomes were more responsive to the connection distance than to the relative cost coefficients, and increasing the connection distance did not necessarily improve the overall landscape structure. These findings emphasize that coupling patch‐level mechanisms with multi‐objective optimization provides a robust, actionable tool for precise spatial planning and targeted cropland policy in high‐density urban environments.