Identifying Key Degradation Factors in Robinia pseudoacacia Shelterbelts using a Comprehensive Degradation Index and Explainable Machine Learning

This dataset supports the research presented in the manuscript entitled "Identifying Key Degradation Factors in Robinia pseudoacacia Shelterbelts of the Historic Abandoned Channels of the Yellow River using a Comprehensive Degradation Index and Explainable Machine Learning".Data Content: The dataset comprises field measurements and calculated indices from 40 sample plots within Robinia pseudoacacia shelterbelts located in the historic abandoned channels of the Yellow River, China.

Data were collected during the 2024 growing season (May to July). It includes 27 variables across four key dimensions for assessing forest degradation: Productivity (e.g., diameter at breast height, tree height, stand volume, biomass), Structural Function (e.g., canopy closure, uniform angle index, competition indices), Disturbance Pressure (e.g., dead branch incidence, insect/disease incidence, disturbance level), and Soil Environment (e.g., soil bulk density, soil organic matter, total nitrogen).Usage and Application: This primary data was used to construct a Comprehensive Degradation Index (CDI) and to identify key drivers of shelterbelt degradation using a machine learning workflow integrating LASSO regression, Random Forest with spatial cross-validation, and SHAP (SHapley Additive exPlanations) analysis.