Gut microbiota influences onset of foraging-related behavior but not physiological hallmarks of division of labor in honeybees

Abstract

Gut microbes can impact cognition and behavior, but whether they regulate the division of labor in animal societies is unknown. We addressed this question using honeybees since they exhibit division of labor between nurses and foragers and because their gut microbiota can be manipulated. Using automated behavioral tracking and controlling for co-housing effects, we show that gut microbes influence the age at which bees start expressing foraging-like behaviors in the laboratory but have no effects on the time spent in a foraging arena and number of foraging trips. Moreover, the gut microbiota did not influence hallmarks of behavioral maturation such as body weight, cuticular hydrocarbon profile, hypopharyngeal gland size, gene expression, and the proportion of bees maturing into foragers. Overall, this study shows that the honeybee gut microbiota plays a role in controlling the onset of foraging-related behavior without permanent consequences on colony-level division of labor and several physiological hallmarks of behavioral maturation.

Importance: The honeybee is emerging as a model system for studying gut microbiota-host interactions. Previous studies reported gut microbiota effects on multiple worker bee phenotypes, all of which change during behavioral maturation-the transition from nursing to foraging. We tested whether the documented effects may stem from an effect of the microbiota on behavioral maturation. The gut microbiota only subtly affected maturation: it accelerated the onset of foraging without affecting the overall proportion of foragers or their average output. We also found no effect of the microbiota on host weight, cuticular hydrocarbon (CHC) profile, hypopharyngeal gland size, and the expression of behavioral maturation-related genes. These results are inconsistent with previous studies reporting effects of the gut microbiota on bee weight and CHC profile. Our experiments revealed that co-housed bees tend to converge in behavior and physiology, suggesting that spurious associations may emerge when rearing environments are not replicated sufficiently or accounted for analytically.

Keywords: Apis mellifera; behavioral development; behavioral maturation; microorganisms; social behavior; symbiosis.

Fig 1

The gut microbiota accelerates the onset of foraging-like behavior under an automated behavioral tracking system. (a) Timeline and experimental setup for the automated behavioral tracking experiment. In each experimental replicate, two groups of 100 gnotobiotic bees (either microbiota depleted or colonized) could move freely between two Plexiglass boxes connected by a plastic tube and hosted in separate climate-controlled chambers. (b) Average number of trips between the nest and the foraging arena per bee for each sub-colony in the automated behavioral tracking experiment. (c) Average proportion of time spent in the foraging arena per bee for each sub-colony. (d) Average age at which bees made their first trip to the foraging arena for each sub-colony. Lines connect paired sub-colonies and are colored by experimental replicate. Boxplots show the median and first and third quartiles. Whiskers show the extremal values within 1.5×, the interquartile ranges above the 75th and below the 25th percentile. **P  <  0.01; NS, not significant, as calculated by paired t-test (two sided). The camera and bee icons in panel a were created with BioRender.com.

Fig 2

The gut microbiota does not affect CHC profile. (a) Non-metric multidimensional scaling (NMDS) of Bray-Curtis dissimilarities between CHC profiles in the automated behavioral tracking experiment (MD, n = 88; CL, n = 89). (b) NMDS of Euclidean distances between CHC profiles in the RNA-sequencing experiment, after removal of batch effects from two separate GC-MS runs (MD, n = 31; CL_Bifi, n = 29; CL_13, n = 30; CL, n = 30). (c) NMDS of Bray-Curtis dissimilarities between CHC profiles in the longitudinal experiment (MD, n = 54; CL, n = 54). (d) NMDS of Bray-Curtis dissimilarities between CHC profiles in the single-colony experiment (MD, n = 59; CL, n = 59). Samples are colored by gut microbiota treatment, and shapes indicate nurses and foragers in panels a, b, and d. Samples in panel c are colored by age of the sampled bees, and shapes indicate treatment group.

Fig 3

The gut microbiota does not affect weight gain and hypopharyngeal gland size. Boxplots reporting fresh body weight (a) and gut wet weight (b) by gut microbiota treatment group in the RNA-sequencing experiment (8-day-old bees). Boxplots show the median and first and third quartiles. Whiskers show the extremal values within 1.5×, the interquartile ranges above the 75th and below the 25th percentile. (c) Photographs showing examples of maximally developed and degenerated hypopharyngeal glands and (d) proportion of hypopharyngeal gland sizes across gut microbiota treatment groups in the RNA-sequencing experiment. NS, not significant. (e, f) Fresh body weight growth curves of individual bees colored by gut microbiota treatment group, shown separately for the bee bread (e) and pollen (f) experiments. Values are proportions of initial body weight at the time of gut microbiota colonization (1-day-old bees). Thicker lines represent mean values, and bars indicate SD.

Fig 4

Gnotobiotic bees reared in cages diverge into nurses and foragers showing differences in physiology and behavior. (a) Heatmap of relative abundance of detected CHCs on the cuticle of gnotobiotic bees in the RNA-sequencing experiment (8-day-old bees; n = 120; shown in gray in the annotation column toward the left) and conventional nurses (n = 51) and foragers (n = 9) randomly collected from the same hives in blue and pink, respectively. The dendrogram toward the left shows clustering of CHC profiles based on Euclidean distances using Ward’s criterion. (b) Fresh body weight, (c) gut wet weight, (d) number of Actin copies in the gut, and (e) hypopharyngeal gland size of CHC-classified nurse and forager gnotobiotic bees in the RNA-sequencing experiment (8-day-old bees). (f–i) Boxplots reporting the number of head-to-head interactions (normalized by group size) (f), trips to the foraging arena (g), the age at first foraging trip (h), and the proportion of time spent in the foraging arena (i) over the week of tracking for the gnotobiotic nurses and foragers classified based on their CHC profiles at the end of the automated behavioral tracking experiment (10-day-old bees). ***P  <  0.001; **P  <  0.01; *P  <  0.05; NS, not significant. Numbers at the bottom of boxplots and stacked bars in panels b to i indicate sample sizes. (j) Venn diagram reporting overlap in the brain between the differentially expressed genes (DEGs) associated with the gut microbiota (as identified in reference 7) and those associated with behavioral maturation (CHC-classified gnotobiotic nurses vs foragers). (k) Venn diagram reporting overlap in DEGs in brain region-specific comparisons of CHC-classified gnotobiotic nurses vs foragers. (l) Venn diagram reporting overlap in the gut between the DEGs associated with the gut microbiota (as identified in reference 7) and those associated with behavioral maturation. The brain icons were created with BioRender.com.

Fig 5

Proportions of nurses and foragers across the experiments. Stacked bars report the percentage of CHC-classified gnotobiotic nurses and foragers (based on clustering of Euclidean distances in CHC profiles using the Ward’s criterion) in the RNA sequencing (a), automated behavioral tracking (b), single-colony CHC (c), longitudinal weight gain with either bee bread (d) or pollen diet (e), and time-series CHC (f) experiments. Numbers at the bottom of stacked bars indicate sample sizes. NS, not significant.