Colony metrics, vg expression and survival
Colonies at both sites fed pollen patty consumed the entire patty each week. Populations of adult bees did not differ between sites (F1,19 = 0.58, p = 0.45), but were larger in fed than unfed colonies (F1,19 = 5.43, p = 0.03) (Fig. 1). At both sites, populations in fed colonies peaked in October with about eight combs of bees (20,000 bees). In unfed colonies, populations peaked in September and October with 4.5 combs of bees (about 11,250 bees). By November, populations in unfed colonies dropped, and were about half the size as in October. Fed colonies had a minimal decline from October to November. Brood areas also were similar between sites (F1,17 = 0.01, p = 0.90). Fed colonies had larger brood areas than unfed (F1,17 = 19.9, p = 0.0003). Brood production did not decline until November in fed colonies, but in unfed colonies, brood rearing began to decline in September. Despite differences in colony sizes, stored pollen did not differ between sites (F1,16 = 0.008, p = 0.93) or between fed and unfed colonies (F1,16 = 0.61, p = 0.44).
Vg expression levels were significantly higher in fed than in unfed colonies at both sites, but were significantly higher only at site 2 (site 1: S1 = 1.67, n = 15, p = 0.20; site 2: S1 = 5.4, n = 15, p = 0.02). Fed colonies survived significantly longer than unfed (χ2 = 8.17, df = 1, p = 0.004) (Fig. 1). Of the 16 colonies established in each treatment group at the start of the study, six of the fed colonies survived the winter and were alive in March compared with one unfed colony.
Comparisons between mite populations in fed and unfed colonies were made in two ways; on the basis of individual bees or cells with purple-eyed pupae and for entire colonies. The number of mites per 100 bees did not differ between locations (F1,17 = 0.17, p = 0.69) or between fed or unfed colonies (F1,17 = 1.02, p = 0.33). The average number of mites per colony from July to December also did not differ between sites (F1,26 = 0.05, p = 0.82) or between fed and unfed colonies (F1,26 = 1.01, p = 0.31) (Fig. 2). All fed colonies had ≤ 100 mites (0.20–0.90 mites per 100 bees) in July and August, but this increased to more than 1500 mites in the fed colonies (3–9 mites per 100 bees) by October and November. In the unfed colonies, we estimate < 25 mites per colony in July and August (< 1 mite per 100 bees), and over 800 mites per colony in October and November (7–13 mites per 100 bees).
The proportion of sealed brood cells infested by varroa did not differ between sites (F1,16 = 1.56, p = 0.23), or between fed and unfed colonies (F1,16 = 0.16, p = 0.69). Similarly, the average number of infested brood cells per colony did not differ between sites (F1,29 = 0.81, p = 0.37), but did differ between treatments (F1,29 = 6.4, p = 0.02) with fed colonies having more infested cells than unfed. In July, we found less than one mite per 100 cells in fed and unfed colonies (< 30 infested sealed brood cells per colony). By November, fed and unfed colonies averaged 14% cell infestation rates or about 300 (unfed) to 900 (fed) infested sealed brood cells per colony.
We rarely captured FWM at colony entrances in August or September. The frequency of collecting FWM increased throughout October and peaked in late November when 6–9% of the foragers from fed and about 3% from unfed colonies had mites on their bodies. The proportion of FWM did not differ between sites (F1,33 = 2.45, p = 0.13), fed and unfed colonies (F1,33 = 0.19, p = 0.66), and whether FWM were entering or leaving hives (F1,33 = 0.007, p = 0.95).
Multiple regressions were conducted to determine factors affecting the growth of phoretic mite populations (i.e., mites per 100 bees) and numbers of infested sealed brood cells. Growth of phoretic mite populations was significantly correlated with: combs of bees, the proportion of infested brood, and average proportion of FWM captured at the hive entrances (Table 1). The proportion of infested sealed brood cells was correlated with mites per 100 bees and FWM.
Honey bees typically are attacked by multiple viruses simultaneously (Carrillo-Tripp et al. 2015; de Miranda et al. 2010), and this was the case in our samples. We detected ABPV, BQCV, IAPV, KBV, SBV, and DWV in at least some of the bees during each sampling interval. The number of viruses we detected did not differ between sites (F1,9 = 0.16, p = 0.70) or feeding treatment (F1,9 = 0.0056, p = 0.94), and increased significantly from July to December (F5,5 = 15.8, p = 0.004) (Fig. 3).
Of all the viruses we screened for, only DWV was detected consistently and varied through time in each group we examined. In purple-eyed pupae, DWV levels were similar between sites in unfed colonies, but differed between sites in the fed colonies, with site 2 having lower median levels than site 1 (Table S1). Based on this difference, data were analyzed separately for each site. At both sites, DWV levels in purple-eyed pupae did not differ between fed and unfed colonies. Using July DWV levels as a calibrator, DWV levels in pupae increased by about 100 fold during the study.
DWV levels in bees collected during sugar shakes (i.e., nurse bees) did not differ between sites for fed and unfed colonies (Table S1). Data for both sites were combined, and analyzed to determine if DWV levels differed between fed and unfed colonies. Supplemental pollen feeding did not affect median DWV levels. DWV levels in nurse bees from fed and unfed colonies at both sites showed an increase throughout the study with December values being 104–105 higher than those in July (Fig. 4).
An analysis of foragers (without mites) at site 1 indicated no difference in DWV titers between incoming and outgoing foragers from either fed or unfed colonies (Table S1). Data were combined for incoming and outgoing foragers, and DWV titers were compared between fed and unfed colonies. There was no difference between fed and unfed colonies in median DWV levels. Similarly, at site 2 there was no difference in DWV titers between incoming and outgoing foragers in either fed or unfed colonies. Data were combined (incoming and outgoing foragers), and comparison of fed and unfed colonies showed no difference in DWV levels. A final analysis was conducted comparing DWV levels in foragers between sites. No differences were found in median DWV levels in foragers between sites. Data from both sites were combined to estimate fold differences in DWV in foragers during the study period. DWV levels were 104 higher in November than in July (Fig. 4).
We compared DWV in purple-eyed pupae, nurse bees and foragers to determine if levels differed as bees aged. We found that levels of DWV in nurse bees at sites 1 and 2 were significantly higher than in purple-eyed pupae (site 1: S1 = 25.60; site 2: S1 = 16.90; both p < 0.0001, n = 40). Comparisons between nurse bees and foragers were made using data from both sites since median DWV levels between sites did not differ. DWV levels in foragers was significantly higher than in the nurse bees (S1 = 7.56, p = 0.006, n = 64).
To determine if DWV levels in FWM differed from foragers without mites, we compared DWV levels in November at both sites. Only November data were used because the frequency of capturing FWM from all hives was greatest at this time. At site 1, median DWV levels did not differ between incoming and outgoing FWM in fed or unfed colonies (Table S2). Similarly, foragers without mites (November data only) at site 1 did not differ between incoming and outgoing bees in fed or unfed colonies. Values for incoming and outgoing foragers were combined to test for difference between fed and unfed colonies. DWV levels in FWM from fed and unfed colonies did not differ. DWV levels also did not differ between November samples of incoming and outgoing foragers without mites at site 1 or fed and unfed colonies. Similar trends occurred at site 2 where there was no difference in median DWV levels between November samples of incoming and outgoing foragers without mites in either fed or unfed colonies.
DWV levels in November samples of foragers either with or without mites did not differ between sites, incoming or outgoing foragers, or fed and unfed colonies, so data were combined, and comparisons were made between foragers with and without mites. There was no difference between sites in DWV levels in foragers either with or without mites. Data from both sites were combined to test for differences in DWV levels between foragers with and without mites. In the combined data, median DWV levels were significantly higher in foragers without mites compared with FWM. DWV levels in FWM at both sites were similar to those in nurse bees during November (site 1: H1 = 1.7, p = 0.19, n = 32; site 2: H1 = 1.9, p = 0.17, n = 20).
A final analysis was conducted to compare fold differences in DWV levels between incoming and outgoing foragers with and without mites. The lowest average virus levels were detected in foragers with mites entering fed colonies, so this value was used as the calibrator to estimate fold differences. The lowest fold increases were in foragers with mites entering fed and unfed colonies. Fold increases in DWV in foragers without mites entering or leaving fed or unfed colonies were about twice as high as FWM entering colonies.
The best-fitting path configuration was determined by significant explanatory variables selected for fit with the AIC (Fisher’s C = 45.111, df = 16, p < 0.001) (Fig. 5). Colony size (combs with bees) and the frequency of capturing FWM significantly affected phoretic mite populations (mites per 100 bees). The proportion of infested pupal cells also was significantly affected by FWM. DWV levels in purple-eyed pupae from uninfested cells were significantly affected by the proportion of cells infested with mites. DWV levels in nurse bees were significantly affected by the size of the phoretic mite population and proportion of cells with mites.