Fig. 3: BQCV loads related to plant species traits and pollinator roles in the network.

a–d, Relationships of BQCV load of virus-positive wild pollinator individuals with floral resource overlap with honeybees (CWM number of honeybee visits on visited flowers) (P < 0.0001, corrected P < 0.0001) (a), weighted betweenness as a measure of centrality in the network (P = 0.009, corrected P = 0.027) (b), proportion of dish-bowl flowers among visited flowers (P < 0.0001, corrected P = 0.001) (c) and corolla length (mm) of visited flowers (CWM) (P = 0.045, corrected P = 0.082) (d). Z-tests were used to derive P values from averaged LMMs and P values were corrected with the FDR correction. Averaged model predictions of the best models (ΔAICc < 2) are plotted with 95% Bayesian credible intervals (shaded areas). All explanatory variables present in the best models (ΔAICc < 2) except the ones for which relationships with response variables are plotted were fixed at their mean values. Points show raw data. Figure created using Procreate (https://procreate.com/).