The gut microbiome defines social group membership in honey bee colonies

Abstract

In the honey bee, genetically related colony members innately develop colony-specific cuticular hydrocarbon profiles, which serve as pheromonal nestmate recognition cues. Yet, despite high intracolony relatedness, the innate development of colony-specific chemical signatures by individual colony members is largely determined by the colony environment, rather than solely relying on genetic variants shared by nestmates. Therefore, it is puzzling how a nongenic factor could drive the innate development of a quantitative trait that is shared by members of the same colony. Here, we provide one solution to this conundrum by showing that nestmate recognition cues in honey bees are defined, at least in part, by shared characteristics of the gut microbiome across individual colony members. These results illustrate the importance of host-microbiome interactions as a source of variation in animal behavioral traits.

Fig. 1. Gut microbial communities differ between foragers from different colonies.

(A to C) Forager honey bees across different colonies differ in overall gut microbial community (A) and CHC profile (B), but not in taxonomic richness (C). (D) Individual gut microbial taxa vary in abundance between forager bees from different colonies. (E) Forager bee gut microbial community is determined by colony environment. (A), (C), and (E) shown as nonmetric multidimensional scaling plots depicting BC dissimilarity between samples. (D) Identified microbial taxa shown as stacked bar plots depicting the average relative abundance of honey bee–associated microbes within forager guts from each colony. Bold letters in legends denote statistical significance, P < 0.05 (permutation MANOVA followed by FDR pairwise contrasts). ns, not significant (ANOVA).

Fig. 2. Sister bees inoculated with different honey bee gut microbial communities develop different CHC profiles.

(A and B) Antibiotic-treated bee guts (A) show less growth when plated than those treated with a control sugar solution (B). (C) Sister bees differ in CHC profile when they are treated with antibiotics versus control. (D and E) Sister bees differ in gut microbial community (D) and CHC profile (E) when inoculated with live, full microbiome inoculum versus heat-killed inoculum. (F to H) Sister bees differ in gut microbial community (F) and CHC profile (G) and when they are inoculated by older bees from two different colonies, but not when the older bees are first treated with antibiotics (H). Data are shown as nonmetric multidimensional scaling plots depicting BC dissimilarity (C, E, G, and H) or weighted UniFrac (D and F) between samples. Bold letters in legends denote statistical significance, P < 0.05 (permutation MANOVA followed by FDR pairwise contrasts).

Fig. 3. Gut microbiome plays a role in recognition in honey bees.

(A) G. apicola (black arrow) and L. quercina (green arrow) are easily culturable on LB plates in the lab. (B and C) Bees inoculated with G. apicola differ in gut microbial community (B) and CHC profile (C) from their sister bees inoculated with L. quercina. (D) Bees inoculated with G. apicola are able to distinguish sister bees inoculated with the same phylotype versus those inoculated with L. quercina. (E) Bees inoculated with L. quercina cannot distinguish between those inoculated with the same phylotype versus those inoculated with G. apicola. (F) Colony 1 and 2 bees inoculated with G. apicola or L. quercina differ in CHC profile based on inoculation treatment and colony of origin. (G) Colony 1 bees inoculated with G. apicola accept colony 1 and 2 bees inoculated with G. apicola and reject colony 1 and 2 bees inoculated with L. quercina. (A), (C), and (F) are depicted as nonmetric multidimensional scaling plots depicting BC dissimilarity between samples. Bold letters in legends denote statistical significance, P < 0.05 [permutation MANOVA (A, C, and F) or Pearson’s chi-square (B, D, E, and G) followed by FDR pairwise contrasts]. *P < 0.05 (Pearson’s chi-square).

Fig. 4. Strain level diversity is associated with differences in CHC profile and recognition in honey bees.

(A) G. apicola coding gene tuf has more single-nucleotide polymorphisms (SNPs) between forager bees from different colonies than between those from the same colony. (B) Sister bees inoculated with different strains of G. apicola develop different CHC profiles. (C) Bees inoculated with G. apicola strain 2 or 3 can discriminate sister bees inoculated with the same strain from those inoculated with a different strain. Statistics in (A) using Kruskall-Wallis followed by Dunn’s test pairwise contrasts are shown as violin plots with points representing median and lines represent interquartile region. Statistics in (B) using permutation MANOVA followed by FDR pairwise contrasts are shown as nonmetric multidimensional scaling plots depicting BC dissimilarity between samples. Statistics in (C) using Pearson’s chi-square. Bold letters in the graph (A) and legends (B), and asterisks (C) denote post hoc statistical significance (P < 0.05).