Spatial heterogeneity patterns along the human footprint gradient and their ecological implications: A case study in South Korea

• Human footprint creates spatial heterogeneity across urban, transitional, and natural zones. • GLCM and LISA reveal nonlinear relationships between landscape structure and human impact. • Machine learning detects complex and nonlinear patterns between heterogeneity and ecology. • Transitional zones support high biodiversity despite limited protected area coverage. • Findings highlight the need to enhance connectivity of transitional zones in conservation planning. Assessing spatial heterogeneity patterns along human footprint (HF) gradients is important for understanding biodiversity conservation in human-modified landscapes. This study investigates how human footprint intensity shapes spatial heterogeneity patterns and their ecological implications in South Korea by integrating multiple heterogeneity metrics and machine learning approaches. We quantified HF heterogeneity through local indicators of spatial association and gray-level co-occurrence matrix indices and identified three distinct zones through Self-Organizing Map clustering: urbanized (low heterogeneity), transitional (high heterogeneity), and natural (moderate heterogeneity) areas. The relationships between heterogeneity metrics and ecological indicators—forest patch number, mammal species richness, and protected area coverage—exhibited complex nonlinear patterns. While forest fragmentation increased with HF intensity, mammal species richness showed local increases in transitional zones despite declining protected area coverage, supporting the heterogeneity-diversity relationship. Machine learning models incorporating multiple heterogeneity metrics significantly improved predictive performance compared to HF intensity alone, with Random Forest achieving the highest accuracy. These findings indicate that moderate levels of human activity create complex landscape mosaics that can sustain biodiversity, whereas intensive urbanization leads to homogenization. Our results suggest that landscape planning should consider spatial heterogeneity patterns, particularly in transitional zones, to maintain ecological connectivity. This study provides a quantitative framework for assessing human impacts on ecosystems through spatial heterogeneity analysis, contributing to more effective conservation strategies in human-modified landscapes.