Fig. 3: Response of different soil enzyme activity to global changing factors applied singly or in different numbers of factors (2, 5, and 8 co-acting factors).

For each soil enzymatic activity [a(1)–e(1), a(2)–e(2) and a(3)–e(3)], effect sizes of single factors (n = 8) and multiple factor groups (2, 5 and 8 factors,50 treatments included in each factor group) were estimated [a(1)–(3)], then the correlations of soil enzymatic activity responses to the normalized factor dissimilarity index are shown in scatter plots [b(1)–d(1), b(2)–d(2) and b(3)–d(3)], Spearman correlation coefficients and significance of correlations are indicated by \(r\) and p, respectively. The statistical test used was two-sided. [e(1)–(3)] show the soil response variability explained by Random Forest models (R²). Added predictors represent the contributions of factor identity effect, number of factors, factor dissimilarity, and factor composition to the model explanatory rate. Points are the mean values and the error bars are their 95% confidence intervals (CI) (permutation n = 1000) (see Supplementary Data. 4).