Fig. 2: Individual assessments of the relationship between crop functional diversity features and TFI_diff for each pesticide group.

Each column represents one category of pesticide use for their y-axis: A Total TFI; B Herbicide TFI; C Fungicide TFI; D Insecticide TFI; E Other TFI. Scatter plots with linear regressions represent TFI_diff as a function of the proportion of the diversity feature, restricted to cropping systems that include the feature. Each dot corresponds to one cropping system, with the green/red dots representing significant decreasing/increasing regressions and the grey dots representing non-significant regressions. In addition to linear regressions, binary comparisons of box plots were conducted (except for N.Families and N.Sowing.seasons) to identify potential non-linear effects. Box plots represent the distribution of TFI_diff for cropping systems with vs. without the corresponding feature (e.g. with vs without temporary grasslands; see Supplementary Table 3 for details). The asterisks above box plots indicate the statistical significance level of the t-test comparing the cropping systems with the functional features and those without, based on P values: *P ≤ 0.05; ** P ≤ 0.01; ***P ≤ 0.001. Slopes of regressions and P values are presented above each sub-figure. Each row represents a crop diversity feature for their x-axis: T.grassland, presence and proportion of temporary grassland; Intercrop, presence and proportion of intercrops; Cover crop, presence and proportion of cover crops; N.Families, the number of botanical families; Poaceae, presence and proportion of Poaceae; Faba., presence and proportion of Fabaceae; Bras., presence and proportion of Brassicaceae; Other, presence and proportion of other botanical families; N.Sowing.Seasons, the number of sowing seasons; Spring, presence and proportion of spring-sown crops; Summer, presence and proportion of summer-sown crops; Autumn, presence and proportion of autumn-sown crops; Winter, presence and proportion of winter-sown crops.