Spatiotemporal response of vegetation productivity to coseismic landslides: a case study of NPP dynamics following the 2014 Ludian earthquake, China

Coseismic landslides triggered by the 2014 Mw 6.2 Ludian earthquake substantially disturbed vegetation and altered ecosystem productivity in mountainous southwestern China. Using MODIS net primary productivity (NPP) data, a high-resolution landslide inventory, and multi-source environmental datasets, this study assessed post-seismic vegetation responses by integrating landslide area density (LAD), multivariate linear regression, and Geodetector analysis. NPP declined sharply in 2015, particularly in grid cells with LAD > 40%, where productivity decreased from ~725 to ~670 gC/m² and required nearly five years to recover. Regression analyses indicated that precipitation and temperature were positively but weakly associated with NPP, explaining substantially less variance than LAD. Multivariate regression further revealed that LAD exerted the strongest negative effect on NPP (β = –155.20, p < 0.001), exceeding the contributions of both climatic variables and land-use types. Geodetector results demonstrated that LAD had the highest explanatory power for the spatial heterogeneity of NPP, and its interaction with precipitation further enhanced explanatory strength. Among land-cover types, Cropland and shrubland experienced the largest NPP losses and the slowest recovery, whereas forest and grassland exhibited greater resilience. These findings show that coseismic landslides dominate short- to medium-term ecosystem productivity, with important implications for post-earthquake carbon cycling and restoration.