Gut microbiota structure differs between honey bees in winter and summer

Lucie Kešnerová; Olivier Emery; Michaël Troilo; Joanito Liberti; Berra Erkosar; Philipp Engel

Interestingly, several Enterobacteriaceae (Klebsiella, Pantoea, Serratia, or Tatumella) were 420 prevalent among nurse samples but absent from nearly all foragers and winter bee samples 421 ( Fig. 2F and Supplementary Fig. S4B) , suggesting a specific association of these bacteria 422 with nurse bees. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/703512 doi: bioRxiv preprint 428 One of the drivers of the observed differences in bacterial load and community 429 composition between winter bees, foragers, and nurses could be diet. Dietary differences 430 between the three types of bees were evident from visual inspection of the dissected guts 431 (Fig. 3A-C) . The rectums of winter bees and nurses appeared yellow indicating the presence 432 of pollen, while those of foragers were translucent. Moreover, the wet weight of the guts 433 was significantly different between the three types of bees (ANOVA F(2,68) = 24.13, P=1.21e-434 8), with foragers having on average two times lighter guts than nurses (Tukey HSD test 435 P=1.14e-6) and winter bees (Tukey HSD test P=6.81e-8) (Fig. 3D) . When plotting normalized 436 16S rRNA gene copy numbers as a function of gut weight, we found that gut weight 437 positively correlated with total microbiota abundances across the three bee types (Fig. 3E) . 438 In order to demonstrate that pollen diet is directly associated with increased bacterial loads 439 in honey bees, we experimentally colonized newly emerged bees with a community of 11 440 bacterial strains representing the seven major bacterial phylotypes of the bee gut 441 microbiota [6]. The colonized bees were kept in the laboratory for ten days and fed ad 442 libitum either sterile sugar water and pollen (SW+P treatment), or sugar water only (SW 443 treatment). We found a significant difference in gut weight between the two treatments 444 (Fig. 3F, Welch's T-Test t=9.433, P=1.452e-11) . While the gut weights of the bees of the SW 445 treatment were comparable to those of forager bees, the gut weights of the bees of the 446 SW+P treatment were markedly higher, exceeding even those of winter bees (Fig. 3D & 447 3F ). We observed a positive correlation between gut weight and microbiota abundance for 448 both the experimentally colonized bees in the laboratory (Fig. 3G) worker bees sampled from the hive (Fig. 3E) . Moreover, differences in bacterial loads of 450 individual community members between the two experimental treatments mirrored, to a 451 large extent, the differences found between nurses, foragers, and winter bees: most 452 phylotypes were more abundant in bees fed pollen as compared to bees fed sugar water 453 only (Fig. 3H, (Fig. 3H) , its abundance was higher in winter bees and nurses as compared to 456 foragers ( Fig. 2A) . Notably, Bartonella was able to colonize only 75% of all bees when pollen 457 was absent. The dependence on pollen for gut colonization was even more pronounced for 458 Frischella. Less than 50% of the experimentally colonized bees of the SW treatment had 459 detectable levels of Frischella, and the loads in bees that were colonized were relatively 460 low. In contrast, bees of the SW+P treatment were all colonized and had relatively high and 461 consistent loads of Frischella (Fig. 3H) . 462 Taken together, these results show that a pollen diet leads to an increase in gut weight and 463 overall bacterial load providing a plausible explanation for some of the di