Fig. 3

The relative importance of covariates for SOC stock prediction at various soil depths. a Comparisons of variable importance (%) for the different factor groups on the estimated SOC stock for each soil depth on the Qinghai Plateau. The variable importance values were determined by the recursive feature elimination (RFE), Boruta, fscaret and mlr methods. The proportion of variable importance (%) indicates the proportion of the sum of the relative importance ranking of the top 25 environmental variables in each factor group. The values in brackets indicate the number of variables where the relative importance of variables ranked in the top 25. Standardized variable importance of paleoclimate (i.e., the paleotemperature and the paleoprecipitation) and modern climate (i.e., the modern temperature and the modern precipitation) (b) and the human footprint (c) for estimated SOC stock on the Qinghai Plateau at various soil depths. Note that the importance of modern climate is the sum of the values of the modern precipitation and the modern temperature by layer. SOC represents soil organic carbon