Spatiotemporal evolution and associated factors of coupling coordination between socio-economic development and ecological health: county-level evidence from the Central Yunnan Urban Agglomeration, China

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Understanding how socio-economic development aligns with ecological health is essential for the sustainable governance of mountainous urban agglomerations. This study investigates 49 county-level units in the Central Yunnan Urban Agglomeration, China, from 2010 to 2023. Socio-economic development and ecological health were evaluated using a pooled entropy-weighted TOPSIS framework. A baseline coupling coordination degree (CCD-B) and a reconstructed non-overlapping measure (CCD-R) were developed to reduce mechanical overlap between the dependent variable and subsequent explanatory variables. Spatial autocorrelation analysis, two-way fixed-effects and Mundlak models, OLS–GWR–MGWR comparisons, and leakage-controlled XGBoost–SHAP analysis were employed to examine spatiotemporal patterns, spatial variation, and nonlinear associations. Socio-economic development increased markedly but retained a persistent core–periphery structure centered on Kunming, whereas ecological health remained broadly stable, with pronounced interannual fluctuations and stronger spatial polarization after 2021. The mean CCD-B increased from 0.378 in 2010 to 0.516 in 2023, while the proportion of counties classified as severely or mildly imbalanced declined from 75.51% to zero. Nevertheless, 61.22% of counties remained only barely coordinated in 2023. CCD-R preserved the overall temporal trajectory and spatial hierarchy and remained strongly correlated with CCD-B ( r = 0.957). Panel estimates revealed clear within–between differences: road-network density was insignificant within counties but positively associated with coordination across counties, whereas per capita savings showed a negative within-county but positive between-county relationship. MGWR identified localized variation for only selected variables and did not consistently outperform simpler models. Leakage-controlled XGBoost achieved R 2 values of 0.763 under spatial validation and 0.777 under temporal validation. Local budget revenue, per capita savings, road-network density, and nighttime light intensity were the leading predictive contributors, while intangible cultural heritage density and minority population share played smaller contextual roles. These findings support differentiated regional governance while emphasizing that the identified relationships and turning points represent sample-specific associations and should not be given a causal interpretation.