Extended Data Fig. 10: Visualization of the raw data spread underlying the vector (arrow) illustration of all 5 community assembly components associated with biomass loss.

Modelled change (thick vectors; restricted moving average) and raw data change (thin vectors) of arthropod biomass (y axis) and species richness (x axis) per replicate per plot after 7 years in dependence on plant species richness (PSR) and land-use intensity (LUI). Starting at the average community biomass and species richness value in the first year (Baseline), the community assembly components of biomass change are displayed as vectors (arrows) in the order of 1) species richness loss, 2) species identity loss, 3) species richness gain, 4) species identity gain, and 5) abundance change of persisting species (see Fig. 2 for detailed explanations), reaching the predicted comparison community values of absolute biomass and species richness after 7 years. Most arthropod biomass change is associated with species richness change, but the vertical vectors of species identity and abundance change show that species richness change alone does not explain all biomass change. Modelled vectors are based on median predictions from1,000 linear mixed-effects models (see methods and Tables S22, S23). Vectors for high PSR (60 plant species) and low LUI (0.5) plots are colored in green/blue, vectors for low PSR (monoculture) and high LUI (3.5) plots are colored in beige/brown (see legend).