Fig. 2

Effects of plant diversity on arthropods across trophic levels. Direct and indirect effects of plant species richness (orange), leaf trait functional diversity and composition (yellow), and structural characteristics (brown) of the forest and grassland plant communities on arthropod abundance and species richness (blue) based on path model results. a Forest herbivores (χ² = 6.6, DF = 11, P = 0.796), b forest predators (χ² = 13.4, DF = 17, P = 0.723), c forest parasitoids (χ² = 7.7, DF = 12, P = 0.814), d grassland herbivores (χ² = 25.3, DF = 20, P = 0.259), e grassland predators (χ² = 14.8, DF = 15, P = 0.495), f grassland parasitoids (χ² = 27.2, DF = 21, P = 0.302). Trait compositions 1 and 2 represent the first two axes of a principal components analysis (PCA) on community-weighted means of five leaf traits; functional diversity is the mean pairwise dissimilarity (based on Rao’s Q) of these traits among study plots. Vertical stratification (based on Rao’s Q) and horizontal variability of plant structure (based on Moran’s I) represent variability in plant height (grassland) or the first two axes of a PCA on the variability of tree height and crown projection area (forest) within the study plots. Positive and negative pathways and their corresponding standardized path coefficients in c and d are indicated in blue and red, respectively. Solid lines show significant relationships (P ≤ 0.05 based on 1000 bootstrap draws; scaled by their standardized effects), dotted lines show non-significant pathways (see Supplementary Tables 4–9 for full results). For clarity, covariances between structural and functional diversity (see Fig. S4) were not plotted but are shown in Supplementary Tables 4–9. Percentage values show the explained variance of endogenous (dependent) variables. The animal icons (from www.openclipart.org) are licensed for use in the public domain without copyright (Creative Commons Zero 1.0). Source data are provided as a Source Data file