Horizontal and vertical variation of soil clay content and its controlling factors in China.

Detailed information on the variation of soil clay content at different soil depths are important for water repellency, management of soil fertility, and crop growth. However, studies on the variation of soil clay content at different depths at large scales are limited. In this study, 1446 soil profiles covering most areas of China were extracted from the World Soil Information Service (WoSIS) database and classified into 0-20, 20-60, and 60-100 cm depths using the equal-area quadratic splines. Random forest (RF) and Shapley additive explanations (SHAP) were used to investigate the relationship between soil clay content and environmental factors (climate, topography, land use, and soil types). The results showed that soil clay content and its variability increased with soil depth. The RF model predicted soil clay content at the three soil depths. The mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R), and Lin's concordance correlation coefficient (CCC) ranged from 6.95 % to 8.74 %, 8.90 % to 11.30 %, 42 % to 46 %, and 56 % to 59 %, respectively. Mean annual temperature (MAT), mean annual precipitation (MAP), solar radiation (solarR), and elevation were the key factors that controlled the variation of soil clay content at depths of 0-20, 20-60, and 60-100 cm. In addition, the main effect values of SHAP can identify environmental thresholds consistent with significant areas by capturing the relationship between soil clay content and MAT, MAP, solarR, and elevation. Significant differences were found in soil clay content on both sides of the threshold at MAT of 15 °C, MAP of 800 mm, solarR of 15,000 Jm yr, and elevation of 1200 m. The results provide clues to the management of national soil security under global and regional climate change.