Three A. mellifera colonies with swarming potential to survive in the C. oleifera plantation for 14 years were selected as treatment groups in Shengqiao Town, Changning City, Hunan Province, China (112.40 °N, 26.43 °E). The C. oleifera trees are over twenty years old and growing well. Thirty foragers were collected from each colony and kept in three sterile cup cages with ventilation holes. All honey bees were subjected to a 2 h fasting period prior to feeding. Then, adequate C. oleifera honey (42.5°Bé) was added to each cup and fed for 24 h. These cup cages containing worker bees were kept in an incubator at 34.5°C with 75% relative humidity. The honey bees in this group are referred to as the CN group.

The control group comprised the same three swarming potentials from A. mellifera colonies raised in Tianxin District, Changsha City, Hunan Province, China (116.25 °N, 40.01 °E). However, these colonies did not pollinate C. oleifera before. Sixty foragers were collected from each colony and distributed equally between two sterile cup cages, constituting 180 bees in six cup cages. One group comprised three cup cages from three different colonies in six cup cages. Therefore, they were divided into two groups. All honey bees were starved for 2 hbefore feeding. One group was fed adequate C. oleifera honey (concentration, 42.5°Bé), referred to as group N, and the other group was fed adequate 50% sterile sucrose solution, referred to as group CK. Both groups were fed for 24 h. These cup cages containing worker bees were kept in an incubator at 34.5°C with 75% relative humidity.

After 24 h of feeding, the intact guts of honey bees from each of the three groups were dissected, and the honey sacs were removed using sterile forceps and placed in a 1.5 mL sterile centrifuge tube. Each sample consisted of six honey bee guts from three cup cages, with two honey bees originating from the same cup cage. Three biological replicates were used for each group. Two samples were dissected two times to obtain two samples: one for metagenomic sequencing and the other for metabolomic analysis. All samples were collected, frozen in liquid nitrogen for 1 h, and transferred to a –80°C refrigerator for subsequent detection.

Methanol, acetonitrile, ethanoic acid, and water were obtained from Fisher Scientific, Inc. (Waltham, MA, United States).

DNA was extracted from the gut bacterial using an E.Z.N.A. Soil DNA Kit (Omega Bio-tek, Inc., United States) following the manufacturer’s instructions. Sequencing and data analysis were performed by Beijing Allwegene Technology Co., Ltd. Metagenome data quality control, assembly and annotation of foragers gut flora can be found in the Supporting Information.

Shannon index, a measurement index based on information theory, is widely used in ecology. The formula is H=-ΣPilnPi, where H represents the diversity index and Pi represents the relative abundance of the ith species. Shannon index is used to measure the diversity of the community. The larger the Shannon index, the higher the diversity of the species; conversely, the smaller the Shannon index, the lower the diversity of the species.

Simpson index is calculated as C=ΣP_i, where C is the concentration of species (the maximum value is 1), and P_i is the ratio of the number of individuals of the ith taxon in the whole. In general, the Simpson index has a range of values between 0 and 1, with larger values indicating greater diversity and smaller values indicating less diversity.

All metabolite separations were performed using an ultra-performance liquid chromatography (UPLC) system (SCIEX, UK) coupled to Orbitrap Exploris 120 mass spectrometer (Orbitrap, Thermo Fisher Scientific) by Beijing Allwegene Technology Co., Ltd. Specific chromatography, mass spectrometry detection conditions, metabolites data annotation and analysis were as described in the Supporting Information.

All statistical analyses were performed using SPSS software version 23 (IBM Corp, Armonk, NY). Results are expressed as mean ± standard error of the mean (SEM). Differences between groups were tested for statistical significance using Student’s t-test or analysis of variance (ANOVA) and Fisher’s least squares difference (PLSD) test. Significance levels are shown and P < 0.05 is considered to be a significant difference between the two groups.

The average number of clean reads for the nine samples was 43,911,068, with an average clean base of 6.12 G. Q20 (represents a quality value of 20 if a base is 99% correct) had an average of 97.89%, and Q30 (represents a quality value of 20 if a base is 99% correct) had an average of 93.46%, indicating high sequencing quality ( Table 1 ). Figure 1 reveals that the differences in the Shannon and Simpson among the three groups were not significant (F_{2, 6} = 1.80, df = 2, P = 0.24; F_{2, 6} = 1.32, df = 2, P = 0.33; F_{2, 6} = 1.31, df = 2, P = 0.34), indicating that the changes in the diversity of flora in the gut of Apis mellifera foragers were insignificant in CK, N, and CN groups.

PCA revealed significant changes in the structure of the intestinal flora of honey bees in CK, N, and CN groups ( Figure 2A ). Subsequent community bar graphs demonstrated that the top four dominant phyla in CK, N, and CN groups were Proteobacteria (CK, 86.84%; N, 82.67%; and CN, 77.60%), Firmicutes (CK, 11.12%; N, 16.67%; and CN, 19.93%), Actinobacteria (CK, 1.85%; N, 0.47%; and CN, 1.67%), and Bacteroidetes (CK, 0.12%; N, 0.06%; CN, and 0.07%). However, the dominant phyla varied considerably among the three groups ( Figure 2B ).

The top four dominant genus in the CK group were Bartonella (57.52%), Gilliamella (11.22%), Lactobacillus (10.28%), and Snodgrassella (7.04%) ( Figure 2C ). The top four dominant genera in the N group were Bartonella (53.46%), Lactobacillus (14.87%), Gilliamella (14.74%), and Frischella (4.49%). The top four dominant genera in the CN group were Gilliamella (43.43%), Lactobacillus (18.82%), Snodgrassella (14.21%), and Frischella (6.79%).

The top four dominant species in the CK group were Bartonella apis (31.76%), g_Bartonella_s_unclassified (25.38%), g_Lactobacillus_s_unclassified (6.30%), and Gilliamella apicola (6.06%) ( Figure 2D ). The top four dominant species in the N group were Bartonella apis (30.34%), g_Bartonella_s_unclassified (22.78%), g_Lactobacillus_s_unclassified (8.39%), and Gilliamella apicola (8.26%). The top four dominant species in the CN group were Gilliamella apicola (28.15%), g_Gilliamella_s_unclassified (13.79%), g_Lactobacillus_s_unclassified (10.44%), and g_Snodgrassella_s_unclassified (7.89%). Figure 2E illustrates the top 10 dominant species in each group, with Gilliamella apicola abundance being significantly higher (F_{2, 6} = 153.97, df = 2, P < 0.05) in the CN group than in CK and N groups.

Glycoside hydrolases (GH), glycosyl transferases, carbohydrate esterase, carbohydrate-binding modules, polysaccharide lyases, and auxiliary activity families were found in the following order of abundance: CN > N > CK ( Figure 3A ). The abundance of GHs in the CN group was 1.61 times higher than that in the CK group and 1.45 times higher than in the N group. Similarly, the abundance of GH65, GH73, GH29, GH13, GH38, GH4, GH23, GH68, GH32, GH105, GH20, GH43, GH102, GH121, GH5, GH101, GH127, GH3, GH31, GH77, GH130, GH51, and GH30 families in the CN group was elevated ( Figure 3B ).

Figure 4A presents PCA graph scores for the CK, N, and CN groups. Each point represents a sample, the distance between points represents the similarity between samples, and ellipses are 95% confidence intervals. Figure 4A presents that the three groups were independently distributed, with no overlap, indicating intergroup differences among CK, N, and CN groups.

Hierarchical cluster analysis allows for classifying metabolites with the same characteristics into groups. It determines the degree of variation in the content of metabolites with the same characteristics within a group. This analysis revealed differences in gut metabolites of A. mellifera foragers in CK, N, and CN groups. Heat maps exhibited that CK, N, and CN groups were categorized into three different profiles, suggesting that the consumption of C. oleifera honey and the source of the bee species significantly affected the metabolites in the guts of foragers ( Figure 4B ).

Figure 5 reveals that mannanotriose, raffinose, and stachyose were undetected in the guts of foragers in the CK group. However, mannanotriose, raffinose, and stachyose levels were significantly lower in the guts of foragers in the CN group than in those in the N group (P < 0.05). This suggests that the three oligosaccharides accumulated in the guts of the foragers from the CN group.