Figure legends
Fig. 1 Overview of the study area and sampling design. a) Most of the study area is used for agriculture fields (grey). Forests (dark green) make up to 13%. Grasslands (light green) are mainly intensified, wet grasslands or fallows. b) For each sampling site the cover of arable fields and landscape heterogeneity was calculated for different radii (60m - 3000m) around the site. c) dry grasslands are found mainly on smaller hills or slopes, often surrounded by arable fields.
Fig. 2 Wild bee (a) and hoverfly (b) responses to arable field cover (black) and landscape heterogeneity (red) across multiple scales (60m-3000m). The graphs show the parameter estimates of the models for both predictors on both response variables species richness (solid line) and abundance (dashed line) for each specific radius. Thick lines refer to models, in which the parameter had a statistical clear (p<0.05) effect on the response variable. The triangles refer to the scale of effect, i.e. the scale at which the landscape predictor has the largest effect (lowest AIC, see Fig. 3).
Fig. 3 Effect of arable field cover (a, b) and landscape heterogeneity (c, d) on wild bee (a, c) and hoverfly (b, d) species richness (SR). The graphs show the relationships at the scale of effect, i.e. the radius at which the landscape parameter has the largest effect on the response variable (compare Fig. 1). The different colors refer to the three sampling campaigns. The dotted lines in the hoverfly plots refer to analyses (GLM) that included only the third sampling campaign.
Fig. 4 Effects of local plant community attributes on wild bees (a) and hoverflies (b). The figure shows parameter estimates and confidence intervals of the averaged models for species richness (black) and abundance (grey). The local plant community was recorded during pollinator samplings within a circle (r=5m) around the traps. Please note that for hoverflies, only the third sampling campaign was analyzed (see Methods). SR= Species richness; FD = Functional diversity; CWM = community weighted mean.