Unveiling nonlinear effects of transport development on rural settlement transitions along the “southern Jiangsu - northern shaanxi” transect in China

Abstract

Transport development exerts a significant influence on rural settlement transitions, yet explorations of its nonlinear and threshold effects are limited. This study employed a machine learning approach, specifically the gradient boosting decision tree (GBDT) model, to explore the nonlinear impact of transport development on rural settlement transitions within the “Southern Jiangsu-Northern Shaanxi” (SJNS) transect in China. The assessment of transport development encompasses essential indictors of transport dominance (TD), including density, proximity and accessibility. Concurrently, the rural settlement distribution is characterized by eight landscape metrics, namely, the rural settlement area (RSA), rural settlement number (RSN), patch density (PD), mean patch size (MPS), patch size standard deviation (PSSD), mean shape index (MSI), mean patch fractal dimension (MPFD), and aggregation index (AI). Our findings revealed that regional variations in TD across the transect exhibited a spatial pattern, with elevated values concentrated in the central and eastern sectors and diminished values in the western sector. Notably, areas with elevated values tended to cluster around municipal districts and adjacent areas. Furthermore, during the period from 2000 to 2020, the RSA rate slightly decreased at an average rate of 0.1% per year, while the RSN rate increased at an average rate of 0.3% per year. The RSN was more concentrated in the central sector of the transect than in the eastern and western sectors, with the highest PD observed within the export-oriented economic development zone situated in southern Jiangsu Province, alongside the concentrated agricultural development zone in northern Henan. Finally, the investigation highlights the nonlinear nature of the relationship between transport development and both the RSA and the RSN. Accessibility emerged as a more influential factor than density and proximity. These empirical insights hold substantial value for policy-makers and regional planners, facilitating the refinement of transport development strategies and the cultivation of sustainable land utilization policies within rural areas.