Here, we assessed how the morphology, weight and foraging behavior of individual workers are affected by their surrounding landscape. We hypothesized that colonies established in landscapes showing high cover of intensive crops and low cover of flowering crops, as well as low amounts of local floral resources, would produce smaller workers, which would perform fewer foraging trips and collect pollen loads less constant in species composition. We tested these predictions with 80 colonies of commercially reared Bombus impatiens Cresson placed in 20 landscapes spanning a gradient of agricultural intensification in southern Québec, Canada. We estimated weekly rate at which workers entered and exited colonies and captured eight workers per colony over a period of 14 weeks during the spring and summer of 2016. Captured workers had their wing, thorax, head, tibia, and dry weight measured, as well as their pollen load extracted and identified to the lowest possible taxonomic level. We did not detect any effect of landscape habitat composition on worker morphology or body weight, but found that foraging activity decreased with intensive crops. Moreover, higher diversity of local floral resources led to lower pollen constancy in intensively cultivated landscapes. Finally, we found a negative correlation between the size of workers and the diversity of their pollen load. Our results provide additional evidence that conservation actions regarding pollinators in arable landscapes should be made at the landscape rather than at the farm level.

 20 bumble bee commercial quads from Koppert® (with four colonies each) were installed in 20 different landscapes (mean nearest-neighbor: 69km +- 37km) in the south of Quebec region, in Canada. The colonies were installed on May, 3th 2016 and followed every week until their death. Every week, entries and exits were counted for 15 minutes (between 9h00 and 16h00). Also, when possible, a worker was captured, for a maximum of 8 per colony for the whole season. 

Captured workers were conserved in the freezer (-20°C) until processing. Every captured worker had its thorax, head width, head length, wing length, marginal cell length and tibia length. However, when the wing and/or the tibia were damaged, they were counted as NA. Every worker also had its pollen load remove, and 150 representative pollen grains were identified per sample. Finally, every worker were weighted, except workers with missing parts (wing or legs). 

Landscapes were characterized once at the beginning of August (2016) with a radius of 1000m from the quad. The proportions of intensive crops (Maize, soya, cereals, apple, strawberries etc.. ), flowering crops (not using lots of pesticides, such as alfalfa and clover), extensive crops (pasture and hay) were described. Locally, we identified every plant within a 100m radius of the quad.  

For more precise information on the techniques used, don't hesitate to contact me: amelie.gervais.3@ulaval.ca