Data for: Effects of short-term managed honey bee deployment in a native ecosystem on wild bee foraging and plant-pollinator networks

Honey bees (Apis mellifera L.) are important agricultural pollinators, and there is increasing demand for forage habitat for managed colonies. However, there is also evidence that pasturing honey bee colonies within natural landscapes may negatively affect wild bees through resource competition.

To assess resource competition between managed honey bees and wild bees, we conducted repeated, short-term deployments of honey bee colonies within Florida forests coinciding with seasonal wildflower bloom, and compared wild bee foraging with and without honey bee colonies present over multiple seasons.

We recorded over 2,000 bee visits including 196 pairwise bee-plant interactions. Deploying honey bee colonies was associated with a reduction in wild bee foraging rates, and honey bee and wild bee foraging rates were significantly, negatively correlated. Honey bees disproportionately visited resources with high floral density. Honey bee foraging preferences differed significantly from genera with small-bodied (Lasioglossum, Perdita, Augochlorella), and/or specialist species (Perdita, Andrena), as well as with Megachile, but overlapped with genera including larger-bodied (Bombus, Habropoda, Osmia, Xylocopa) and/or generalist species (Bombus, Xylocopa, Agapostemon). Deploying honey bee colonies did not significantly affect plant-pollinator network metrics.

These results illustrate that short-term honey bee colony deployment can negatively affect wild bee foraging and that competition may be greater for certain genera, particularly larger-bodied bees or those with generalist diets though less for smaller-bodied and/or specialist bees. Our short-term, low-density deployment treatments may have precluded significant effects on network metrics, and likely underestimate the effects of typical higher density and longer-term honey bee deployment.

Methods
One location was selected for the deployment of honey bee colonies at two forest sites, ACF and GSF, and six 50 m2 observation plots (2 m x 25 m) were systematically chosen at 250 m increments along each of two 1.25 km long transects running away from the honey bee deployment location for a total of 12 plots per forest and 24 plots total. 

The study was repeated four times during fall 2018 (late Sept. – Nov., site ACF only), spring 2019 (early March – April, both sites), summer 2019 (early June – July, both sites), and fall 2019 (early Oct. – Nov., both sites). Sampling in each season was done during peaks of local wildflower abundance for a period of ~ 6 weeks. During each season, each plot was visited on three separate observation days before honey bee colonies were deployed. After deployment and a minimum 48-hour acclimation period, we resumed observations for a total of six observation days per plot while honey bees were deployed. After removing the colonies from both ACF and GSF sites and another 48-hour acclimation period, we conducted three additional observations in each plot. At the end of each temporal replication, we had visited each plot a total of six times without, and six times with, honey bee treatment colonies present. 

Plots were visited at each site between the hours of 09:30 - 14:00 to capture highest estimated peak bee activity. We alternated the visitation sequence of plots at each forest site across observation days. We used the scan sampling protocol to record plant-pollination interactions within each plot (Vaissière et al., 2011). In this protocol, each flower is briefly but fully scanned; thus, total sampling time varies with total flower density but each flower is observed for a standardized amount of time (~1 second). For each bee encountered foraging on a flower, we recorded the plant type to the lowest possible taxonomic level. We then captured each wild bee observed with a standard entomological aerial net for taxonomic identification with the exception of bees that could be visually identified to species in the field. In cases where the bee specimen was neither successfully collected nor identified to species, we recorded the bee to the lowest taxonomic level possible. Sampling was only conducted during at least partial sun and temperatures above 15º C. Specimens were frozen for further identification and initially identified by R.E. Mallinger using Discover Life, a key to bees of Florida, USA (Pascarella and Hall 2006), and a reference collection of specimens from the same region identified by bee taxonomist John S. Ascher and entomologist Glenn Hall. Subsequent confirmation and identification of select bees was conducted by John S. Ascher including species of Lasioglossum (Dialictus) (identified with reference to the revision by Gibbs, 2011), Perdita, and Hylaeus. Voucher specimens are stored in the laboratory of Dr. Rachel Mallinger.

Additionally, for each plant species in bloom within plots, the observer counted the number of inflorescences, recorded inflorescence type (“raceme”, “panicle”, “composite”, etc.), and identified the flowering plant to the lowest possible taxonomic level using available plant lists for each site, local plant guides, and research-grade (i.e. validated by two identifiers) identifications of photographs posted to iNaturalist.