Fig. 1.

Relative effects of estimates (%) of each group of predictors (i.e. land-use history and current land use) and parameter estimates (± SE) resulting from the model averaging procedure on the flower trait diversity of plant community (hypervolume, sd³). P-values of the best selected models for each model parameter are given, *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001. We considered only plants with nectar in the analyses. The flower trait diversity was log-transformed and all variables were scaled. See also Table S6 for details on model selection and Table S7