SPATIO-TEMPORAL DYNAMICS AND DRIVING MECHANISMS OF ECOSYSTEM SERVICE FUNCTION VALUE: AN INTEGRATED EXPLAINABLE MACHINE LEARNING APPROACH IN THE GANJIANG RIVER BASIN, CHINA

It is imperative to comprehend the variation and identify the driving factors of ecosystem service value (ESV) to enhance regional ecological security of the Ganjiang River Basin (GRB), China.However, existing research methods face the challenge of multiple covariates and nonlinear effects among variables, as well as the "black box" problem inherent in machine learning methods.This study proposed a novel approach by integrating the explainable machine learning approach to identify the driving factors of the four ecosystem service function value (ESFV): provisioning services, regulating services, supporting services, and cultural services.The findings of the study suggest the following: (1) The forestland constituted the primary land use in the study area, and the ESV exhibited a net decline of 3.74 billion yuan from 1990 to 2020.( 2) With regard to vertical spatial evolution, the unit ESV was found to be highest in the 0-3 slope segments, while it was relatively high in the 500-1200 m elevation range.(3) The spatial distribution of ESV demonstrated lower values in the northern regions and higher values in the southern and central regions.(4) The Human Activity Index (HAI) showed the largest substantial impact among the four ecosystem services, with its contribution to the model being predominantly negative.