Long-term Ruminant Livestock Distribution Datasets in Grazing Livestock Production Systems in China from 2000 to 2021 (CLRD-GLPS)

Abstract. Understanding the spatial-temporal distribution of grazing livestock is crucial for assessing the sustainability of livestock systems, managing animal diseases, mitigating climate change risks, and controlling greenhouse gas emissions. In China, grazing ruminants are mostly distributed across the vast grasslands in semi-humid and alpine areas. However, existing datasets of gridded distribution of grazing ruminants in China do not distinguish grazing ruminants with other livestock production systems, nor capture their long-term and seasonal dynamics, and tend to overestimate grazing livestock distribution. This study uses the county-level data from the Grassland Ecological Protection Subsidies to differentiate grazing livestock from other forms of livestock rearing. Interpretable machine learning models were used to detect the seasonality of grazing pasture and map the China’s long-term annual ruminant livestock distribution in grazing livestock production systems from 2000 to 2021 (CLRD-GLPS). The model's internal ten-fold cross-validation results (adjusted R²) for cattle ranged from 0.850 to 0.952 and for sheep from 0.780 to 0.836. External validation using province-level livestock meat production data yielded Pearson correlation coefficients of 0.83–0.88 for cattle and 0.92–0.94 for sheep, respectively. The CLRD-GLPS datasets provide more detailed, gridded information on local livestock distribution than census-based data. Compared to actual census data and the GLW datasets, they better capture the spatial-temporal dynamics of livestock distribution. Spatially, the largest cattle numbers on seasonal pastures were in the south-eastern edge of the Qinghai-Tibet Plateau (QTP), while the largest sheep numbers were in north-eastern Qinghai and Xinjiang. Temporally (2000–2021), cattle numbers increased near the Three-River Source National Park and Helan Mountains, while sheep numbers decreased on seasonal pastures on the QTP, with no significant changes on year-round pastures in Inner Mongolia. The datasets provide essential information for understanding the spatial-temporal dynamics of grazing ruminants and formulating relevant livestock management policies, among other applications. Additionally, the research framework developed in this study can serve as a new framework for creating livestock distribution datasets in other regions and livestock production systems.