1. In response to anthropogenic environmental change, the cues that animals use throughout their lifecycle to optimize fitness may become unreliable, resulting in an ecological trap.

2. Here we investigated whether commercial bumble bee (Bombus impatiens) colonies managed for early spring crop pollination act as ecological traps for wild nest-searching Bombus queens by subverting their natural nest usurpation behavior.

3. An average of 10 dead wild queens were recovered from each standard colony during the two-week period of the experiment, but colonies with queen excluders were successful in preventing wild queen deaths. The use of queen excluders did not impact colony performance in terms of resident queen survival, colony reproduction, colony weight gain or worker body size.

4. Sites where wild nest-searching queens were small had higher rates of failed usurpation, suggesting smaller-sized queens are disproportionately at risk from failed usurpation. Furthermore, sites where commercial colonies without queen excluders were introduced for spring crop pollination had fewer bumble bee visits to a later-blooming crop compared to sites without commercial colonies.

5. Synthesis and applications. Our findings reveal a novel mechanism by which commercial colonies can negatively impact wild bumble bee populations and their pollination services. At the same time, we demonstrate a simple and inexpensive risk mitigation tool – a queen excluder – was 100% effective at eliminating this risk without compromising colony performance. Commercial colonies used outdoors during the bumble bee nest-searching period should be fit with queen excluders to prevent negative impacts on wild pollinator communities and their services.

Do commercial colonies function as an ecological trap and can queen excluders prevent this effect?

To investigate whether wild bumble bee (Bombus spp.) queens enter and are killed by commercial bumble bee colonies, we placed two commercial Bombus impatiens colonies (BioBest; Romulus, Michigan, USA) at eight apple orchards in the Finger Lakes region of New York state on May 13th, 2020, shortly before apple bloom. Permission from farm managers was granted for all research and collections from privately owned land and ethical approval for the use and collection of the bumble bees was not required. All orchards except one were conventionally managed and orchards ranged in size from 2 to more than 30 acres. Each colony contained ~75 workers, worker brood, and a laying queen, and was fitted with two flight openings that could be set to be open or closed with a rotating slide. We confirmed that each colony contained only one queen and no queen cells and marked the resident queen with a paint mark. In all colonies from this supplier, one opening allowed bumble bees to enter and exit the colony freely, whereas a second opening allowed bumble bees to enter the colony but generally not exit. At every site, one of the two colonies was fitted with a queen excluder over the enter/exit flight opening, reducing the opening from 16 mm to 5 mm and allowing workers but not queens to pass both in and out of the colony (Fig. 1).  The second colony at each site was maintained in the standard configuration and did not use a queen excluder therefore serving as a control colony. Because the supplier only provides queen excluders on the enter/exit flight opening, the enter-only flight opening of the queen excluder colony remained closed at all times to prevent wild queens from entering the colony. In contrast, for standard control colonies, both flight openings were set to open. Colonies were placed on orchard edges within weather-resistant quad boxes (Fig. 1b) for the duration of apple bloom and were recovered on May 26th. To compare numbers of dead queens that accumulated in excluder versus standard colonies deployed at the same site over the same time period, at the end of the experiment colonies were anesthetized with CO2 for 1-2 minutes and all dead queens and non-B. impatiens arthropods were removed and counted.

Do queen excluders impact colony performance or demographics?

We studied the effects of a queen excluder on colony performance and demographics to determine whether there were impacts of using queen excluders in terms of colony productivity, investment in workers or on resident queen survival. Within 24 hours after colonies were recovered, final colony weight measurements were recorded and compared to initial colony weight measures taken in the 24-hour period before colonies were deployed. We recorded whether resident colony queens were dead or alive to assess whether queen excluders influence resident queen survival. Additionally, we assessed colony reproduction by counting the number of queen cells and taking pictures of the colony to later count the number of workers. To determine whether a queen excluder influenced the body size of the workers that would defend against colony invaders, each colony was agitated and 15 workers that came out to attack the agitators were captured. We measured the body size of the attacking workers by recording their intertegular distance (ITD) using Zen software connected to a Zeiss Stemi 305 microscope (Carl Zeiss Microscopy, LLC).             

Do queen body size and wild queen abundance influence the number of dead queens that accumulate in a colony?

To determine whether queens that entered and died in a colony differed in size from the average wild queens at a site, we haphazardly collected five wild B. impatiens queens that were foraging or nest searching at the eight sites between May 29th and June 1st. We also looked at whether dead queens differed in body size from resident colony queens that had retained their paint mark by the end of the experiment. We measured the inter-tegular distance (ITD) of the dead wild, live wild and resident colony queens using the same software and microscope described above. Though dead wild queens had dried somewhat, the majority were still pliable and prior studies have found little difference in ITD between fresh and dried Bombus specimens (Hagen & Dupont, 2013). Therefore, we do not expect size differences as measured by ITD to be impacted by differences in sample moisture content. 

Additionally, we evaluated whether local B. impatiens queen abundance affected the number of queens that invaded a colony and died. We visited each site three times during apple bloom to measure local queen density by slowly walking at a consistent pace 25 m transects along the orchard edge and counting nest-searching, foraging, and flying B. impatiens queens that were identified visually within 5 meters in both directions. During the first round of site visits, two transects were walked at each site. Due to low counts observed across all sites during the first round, this was increased to six transects during subsequent rounds.

Does presence of commercial colonies during nest-founding impact bumble bee visitation to a later-blooming crop?

To evaluate the impact of commercial B. impatiens colonies on crop visitation under realistic conditions, we explored visitation patterns from a large field study that had been conducted in the prior two years which used colonies without queen excluders from Koppert Biological Systems (Howell, Michigan, USA) in strawberry fields during bloom on eight farms in May 2018 and five farms in May 2019. For this study we selected farms that grew both strawberry, an early blooming crop, as well as tomato, which blooms later in the season and is visited predominantly by bumble bees (Toni, Djossa, Ayenan, & Teka, 2020). Each farm that received colonies was paired with a nearby farm that also grew strawberries and tomatoes but did not receive commercial bumble bee colonies. In total 16 farms were surveyed in 2018 and 10 farms were surveyed in 2019. Strawberry fields were stocked at the recommended density of two colonies per acre of strawberries resulting in a range of 1-16 colonies per farm.  Colonies were arranged in quads and distributed within or at the edge of the field.

            Because this experiment was designed to measure the impact of commercial bumble bee colonies on visitation to both crops, colonies were relocated following strawberry bloom a short distance to tomato field edges on farms that received commercial colonies. We began visitation surveys on tomatoes in early June during 2018 and late June in 2019. On both farms with and without commercial colonies, we recorded visits to tomato flowers by B. impatiens as well as those from wild bees including other bumble bee species during a 30-minute period while walking through the tomato field. A visit and visitor identity were recorded when a bee contacted a tomato flower. Surveyors counted all visits within their visual field while walking at a consistent pace along a row and attempted not to sample the same area in an individual survey bout. Tomato fields were surveyed three times in 2018 and four times in 2019 from early June to early July. Other Bombus species were combined with other wild bee species for analysis as only 0.5% of visits were attributed to these other Bombus species, and failed usurpation attempts were largely constrained to B. impatiens.