INTERACTION_DATA.xlsx

We visited each farm at three time points: in early (18^th–30^th June) and late (30^th July–10^th August) summer 2016 when WFS were in bloom, and during the following spring before the onset of WFS flowering (from 25^th March to 9^th April 2017). We recorded flower density and diversity on two to three transects per farm and time point, depending on the flower provision at the time we performed the survey. Transects of 100 m length and 2 m width were selected for high abundance and richness of flowers and insect visitors within the farm. When present, we prioritised WFS in summer. When no WFS were flowering, other transects were performed on other non-cropped field boundary features such as hedgerows. Transect locations were adjusted as necessary across time points. In total, 55 summer transects were performed, including 28 on WFS, and 27 on other field margins with no sown flowers. In spring 2017, we performed transects either on hedgerows including perennial trees and shrubs, on field margins with perennial flowering plants such as<i> Lamium album</i>, or on flowering oilseed rape crop fields. The spring floral data includes only flower diversity, as density cannot be reliably estimated for large flowering trees such as <i>Prunus sp.</i> and <i>Salix sp.,</i> which represent important pollinator resources.