Statistical Analysis
Bees rarely extended their proboscis during the PER assays (phase 1), so no statistical analysis was performed. For the choice tests (no CR in phase 2), we calculated the two-tailed binomial
probability of bees choosing one of the two offered feeders (P=q=0.5). If foragers choose a feeder at random (H01), equal numbers of individuals should select the feeder scented with the odour that was previously presented during PER assays (S1) and the feeder with alternative odour (S2) (the robability that a bee would choose S1 or S2¼0.5).