The China Long-Term Datasets for Mixed-Farming and Landless Livestock Production Systems (CLD-MLPS and CLD-LLPS)

The China Long-Term Datasets for Mixed-Farming and Landless Livestock Production Systems (CLD-MLPS and CLD-LLPS) provide high-resolution (1 km) annual maps of pigs, cattle, sheep, and goats across mainland China from 2000 to 2021. These datasets were developed to capture long-term changes in non-grazing livestock systems, reflecting the rapid shift from mixed-farming to landless production over the past two decades. Livestock populations were first segmented into mixed-farming and landless systems using intensification rates derived from official livestock census data. Spatial distribution models for each species and system were then constructed using a stacking ensemble of five interpretable machine learning algorithms (Random Forest, Extra Trees, XGBoost, LightGBM, and CatBoost). Model predictions were constrained by cropland extent for mixed-farming systems and enterprise registration records for landless systems to ensure spatial realism. Together, CLD-MLPS and CLD-LLPS provide the first long-term, system-specific livestock distribution datasets for China, offering a valuable foundation for research on agricultural transitions, food system sustainability, and environmental management.