Experimental Design and Field Methodology
At each field site, we placed a pair of B. vosnesenskii colonies on a given date. Within the pair, we randomly assigned colonies to one of two food treatments: early supplementation (during the first 20 days in the field, early-pulse) or late supplementation (at 21-40 days in the field, late-pulse). In each treatment, food pulses were intended to represent elevated resources occurring earlier or later during the growth phase of colony dynamics (based on our past experience withB. vosnesenskii, cf. Crone & Williams 2016; Kerr et al.2019; Malfi et al. 2019). Colony weight at field placement did not differ by experimental food treatment (F1,26 = 2.04,P = 0.17).
During pulses, a colony received (1) ad libitum access to artificial nectar (BioGluc®, BioBest Canada Ltd.), which was delivered directly to the colony via a reservoir with an access point located inside of the colony box (Fig S1), and (2) pollen in the form fresh frozen honey bee pollen mixed with diluted artificial nectar at a ~1:1 mass ratio, and encased in USDA certified organic beeswax (Koster-Keunan Inc., CT, USA), delivered to the colony on the first and 10th day of the 20-day pulse. The total weight of pollen provided to a colony was proportional to the colony weight at the time of each feeding in order to provide colonies 50% of their estimated needs for a 10-day interval (see Supporting Information, Appendix 3 ). We used this limit because superabundance of pollen in the nest can reduce foraging activity (Pelletier & McNeil 2003). To determine whether and how foraging was affected by supplementation, we recorded the number of return trips per hour for both colonies during site visits throughout the experiment and analyzed this foraging activity information in relation to food treatment (seeSupporting Information, Appendix 4 ). At the end of the 20-day pulse, access to food supplements was terminated.
We weighed colonies in the field every 10 days in order to track their growth across the season. We assumed that a colony’s peak weight would be associated with its switch to reproduction (Crone & Williams 2016) and pulled them from the field once they had declined in weight for two consecutive visits. This ensured that brood structure of the nest would be intact, allowing for final counting of reproductive output. Declining colonies are commonly invaded by insects and fungi that quickly destroy the nest structure. Upon removal from the field, colonies were frozen (-20º C) until they could be dissected in the laboratory. Gyne production was determined by counting the number of queen brood cells present during dissections of the remnant brood structure. Queen cells can be reliably distinguished from worker/male cells based on their larger size (Williams et al. 2012). At the time of dissection, the majority of gynes had already eclosed; for the few colonies where some gynes were still in cocoons, both eclosed and uneclosed gynes were tabulated in the total gyne count.