Fig. 4

The direct and indirect effects of drivers on soil functional trait stability. a Mean predictor importance of factors on functional trait stability based on random forest analysis. Significant predictors revealed in the random forest analysis are marked with asterisks. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Highly significant predictors (P < 0.01, columns marked with green) were selected for structural equation modeling. b, c Direct and indirect effects of driving factors on functional trait stabilities in low and high resource environments using Partial Least Squares Structural Equation Modeling (PLS-SEM), respectively. The ellipses represent the latent variables, and the rectangles represent the observed variables. The factor on the side of the latent variable is the observed variable, and the value in parentheses indicates the weight of the indicator. The black and red arrows in the PLS-SEM indicate positive and negative relationships, respectively, and gray arrows represent non-significant paths (P > 0.05). Numbers on the arrows are path coefficients, and the path widths represent the strength of path coefficient. OM, organic matter; AP, available phosphorus; AK, available potassium, NO₃⁻-N, nitrate nitrogen, NH₄⁺-N, ammonia nitrogen; (NO₃⁻-N + NH₄⁺-N): TN, the ratio of the sum of nitrate nitrogen and ammonia nitrogen to total nitrogen; AK:TK, the ratio of available potassium to total potassium; AP:TP, the ratio of available phosphorus to total phosphorus; C:N, the ratio of soil organic carbon to soil total nitrogen; C:P, the ratio of soil organic carbon to soil total phosphorus; MAT, mean annual temperature; MAP, mean annual precipitation