Field Sampling
Three large-scale crop production strawberry fields in Southern Ontario were selected. Since crop field size was the presiding selection criteria, fields had differing crop varieties, surrounding habitat, and pollinator-friendly additions, or lack thereof. Fields were within 120 km of each other and within a latitudinal gradient of 0.15 degrees. The narrow latitudinal gradient was chosen to increase the similarity in flowering time and temperature fluctuations. All fields had a seven-day spraying rotation; however, pesticides used are largely unknown and therefore were not considered. Each field was sampled weekly between May 1st and August 31st of 2018 when the crop reached at least 20% bloom. Bloom percentage was assessed by walking up a row from the field edge and counting the number of flowers on each side, to a depth of 50 flowers (100 flowers total); this was repeated in the field centre, and numbers were averaged. Sampling took place from 09:30 to 16:00 and consisted of five, hour-long periods followed by a 30-minute break. Each sampling event was divided into a 30-minute active period and a 30-minute observation period with five sampling events in a day.
During active sampling periods non-bee flower-visitors (excluding Lepidoptera and Drosophila spp.) were collected directly into sterile vials and set aside for lab work. Lepidoptera were excluded from these collection samples, because the scales from their wings would disrupt the quantification of pollen found on individuals.Drosophila spp. were excluded because they were far too numerous to capture without affecting the capture rate of other specimens. Samples were placed in a small cooler containing freezer packs at the end of each sampling to minimize grooming behaviour and regurgitation. The remaining 30 minutes of each sampling hour were for observational sampling, where all flower-visiting insects were identified to the lowest confident taxonomic unit. This approach provided a non-lethal sampling method to determine the insect community (including bees) and reduce collector bias. A small number of bees were sporadically collected, to provide comparative data on pollen sources and their abundance. Because this sampling was not standardized, the abundance of these collections cannot be considered representative of true populations. Only observational data are representative of bee abundance at each site. The order of active and passive sampling portions of each hour periods were randomly selected. Sampling periods rotated between edge habitat (the edge of the crop, to 50 m into the interior) and interior habitat (at least 60 m into the crop), while randomizing whether the first period was interior or exterior. Measurements were collected for wind speed, rainfall, humidity, and temperature using AcuRite weather station and solar radiation using a TES 1333R Solar Power Meter, every half-hour. These environmental measurements were taken every 90 minutes during sampling (five per sampling day) and averaged for the analysis.