Home sick: impacts of migratory beekeeping on honey bee (Apis mellifera) pests, pathogens, and colony size

Honey bees are important pollinators of agricultural crops and the dramatic losses of honey bee colonies have risen to a level of international concern. Potential contributors to such losses include pesticide exposure, lack of floral resources and parasites and pathogens. The damaging effects of all of these may be exacerbated by apicultural practices. To meet the pollination demand of US crops, bees are transported to areas of high pollination demand throughout the year. Compared to stationary colonies, risk of parasitism and infectious disease may be greater for migratory bees than those that remain in a single location, although this has not been experimentally established. Here, we conducted a manipulative experiment to test whether viral pathogen and parasite loads increase as a result of colonies being transported for pollination of a major US crop, California almonds. We also tested if they subsequently transmit those diseases to stationary colonies upon return to their home apiaries. Colonies started with equivalent numbers of bees, however migratory colonies returned with fewer bees compared to stationary colonies and this difference remained one month later. Migratory colonies returned with higher black queen cell virus loads than stationary colonies, but loads were similar between groups one month later. Colonies exposed to migratory bees experienced a greater increase of deformed wing virus prevalence and load compared to the isolated group. The three groups had similar infestations of Varroa mites upon return of the migratory colonies. However, one month later, mite loads in migratory colonies were significantly lower compared to the other groups, possibly because of lower number of host bees. Our study demonstrates that migratory pollination practices has varying health effects for honey bee colonies. Further research is necessary to clarify how migratory pollination practices influence the disease dynamics of honey bee diseases we describe here.

180 Varroa, Nosema, BQCV, DWV, and IAPV. To quantify Varroa and Nosema spp., we collected 181 approximately 300 bees from the brood chamber and transferred them to ethanol. To quantify 182 virus prevalence and load, we collected an additional 150 bees from the brood chamber. These 183 samples were stored and shipped to Vermont on dry ice and transferred to -80C for storage prior 184 to analysis.

185
To examine differences in climate and weather conditions experienced by the migratory 186 and stationary groups, we used publicly available NOAA local climatology data collected by 187 weather stations nearest to our field sites (NOAA National Centers for Environmental 188 Information) (Table S1).
189 Varroa mite and Nosema spp. quantification 190 To calculate the number of Varroa mites per 100 bees, ethanol samples were agitated for 191 60 seconds, strained through hardware cloth to separate the mites from the bees, and all mites 192 and bees were counted (Lee et al., 2010). We conducted spore counts to quantify Nosema spp.
193 Although our methods did not differentiate between the two species of Nosema, (N. apis and N. 194 ceranae) previous work has found N. ceranae to be the predominant species in many regions Manuscript to be reviewed 315 an increase in between-group separation with groups becoming more distinguishable from each 316 other. While all groups separated in this third time step, the exposed and migratory groups were 317 less distinguishable from one another compared to the stationary group (Fig. 2B). The linear 318 combinations (LD1 and LD2) yielded a correct classification rate of 75% for stationary colonies 319 but correct classification rates for migratory and exposed colonies were lower, 43.75% and 320 56.25%, respectively. PERMANOVA results indicated statistically significant group separation 321 between isolated, migratory and exposed treatments (F 2,43 = 4.72, P = 0.001).

322
We found no effects of treatment (exposed verses isolated) for any of the parasite or 323 disease response variables (Fig 3). However, Varroa prevalence and load, Nosema prevalence 324 and load, and BQCV significantly increased with time (Table 2). There was a significant 325 treatment × time interaction for both DWV load ( 1 2 = 9.229, P = 0.002; Fig 3B) and DWV 326 prevalence ( 1 2 = 4.94, P = 0.026; Fig. S1) such that DWV in exposed colonies increased at 327 significantly higher rates than the isolated group. There was also a significant treatment × time 328 interaction for FOB ( 1 2 = 9.946, P = 0.0016; Fig 3D) with exposed bees increasing at a 329 significantly higher rate compared to the isolated group. Other interaction terms were not 330 significant (Table 2).

349
The migratory colonies in our study returned from almond pollination with higher BQCV   Schematic of Experimental Design Three sampling events occurred during the experiment. Three experimental groups (isolated stationary group, migratory group, and exposed group) were located in two separate apiaries in North Carolina throughout the experiment: the stationary yard (where all groups begin and the isolated stationary group remained for the duration of the experiment) and the exposed yard (where the exposed group was exposed to the migratory group). Dotted arrows show movement of colonies throughout the experiment. Between sampling events one and two, the migratory colonies were transported to California for almond pollination and back.
Exposed colonies began in the stationary yard and were transferred to the exposed yard prior to sampling event two. Geographic distance between yards are specified in kilometers.

Figure 2
Pathogen community and colony health predicts treatment group membership Linear combinations from discriminant analyses created from all pathogen variables (except BQCV prevalence) and frames of bees for exposed (black), migratory (red) and stationary/isolated (blue) colonies. Axes represent the percentage of between group variance explained.(A) Experiment 1 at sampling event two, migratory and stationary colonies were separated by LD1 while stationary and exposed colonies are clustered. B)Experiment 2 at sampling event three, after the exposed group had been allowed to forage alongside the migratory colonies, exposed and isolated were separated along LD2, while LD1 separated out migratory colonies.The significant PERMANOVA tests for both experiments corroborated the differences between group centroids. Circles represent 70% confidence intervals and are provided to visualize the centroids of each group.