Effects of Different Diets on Gut Microbiota of Apis cerana (Hymenoptera: Apidae)

TL;DR

This study examines how different diets affect the gut microbiota of Apis cerana bees, finding that natural pollens promote greater microbial diversity compared to substitutes.

Contribution

The study provides new insights into how diet influences gut microbiota diversity and structure in Apis cerana.

Findings

Firmicutes and Proteobacteria are the dominant phyla in the gut microbiota of A. cerana.

Pollen diets increase gut microbiota diversity compared to pollen substitutes.

Different diets significantly alter the structure and abundance of core microbial communities.

Abstract

Pollen is one of the main food sources for honeybees. The honeybee gut microbiota plays a crucial role in maintaining digestive function and host health during long-term coevolution. While the consumption and utilization of pollen have been extensively studied, there is limited information about the effects of pollen on the gut microbiota of Apis cerana. In this study, we used 16S rRNA amplicon sequencing to evaluate the effects of four natural pollens (oilseed rape pollen, camellia pollen, lotus pollen and buckwheat pollen) and two pollen substitutes (Diet 1 and Diet 2) on the hindgut microbiota of newly emerged A. cerana worker bees, following feeding periods of 5, 10 and 15 days. The results showed that Firmicutes and Proteobacteria are dominant in the gut microbiota of A. cerana. A. cerana workers fed with pollen diets had a higher diversity of gut microbiota than those fed with pollen substitutes. There have been significant differences in the gut microbiota structure and relative abundance of the core microbial community among A. cerana workers supplied with different diets. Our results confirm that gut bacterial communities of A. cerana can be influenced by pollen diets and may play an important role in host adaptation.

Figures

Full text

1. Introduction

Apis cerana has evolved over time to achieve long-term adaption to the natural ecological environment in China [1]. It mainly nests in diverse habitats characterized by complex topography, divergent climate and rich flora, resulting in remarkable resilience to external environment changes [2]. As an important part of the natural ecosystem, A. cerana is predominantly distributed in the alpine mountainous regions and excels at utilizing sporadic nectar sources, enabling it to pollinate a wide range of flowering plants and crops in the alpine mountains [3]. A. cerana not only provides ecological and economic benefits to local residents, but also plays a vital role in increasing agricultural productivity and maintaining the stability of natural ecosystems [1,4].

The gut microbiome is crucial in maintaining honeybees’ health and environmental adaptation. As an important symbiont of animals, the microbiome play roles in growth and development [5], digestion and absorption of nutrients [6], and disease prevention in host health [5,7,8]. The midgut of honeybees presents an unstable environment for microbial establishment, whereas the hindgut harbors a substantial microbial community, comprising approximately 10^8^–10^9^ bacterial cells [9,10]. Honeybees harbor a relatively simple but remarkably specialized and consistent gut microbial community. Five species of bacteria are nearly ubiquitous across individual honeybees, including two Gram-negative bacteria Snodgrassella alvi and Gilliamella apicola and Gram-positive bacteria such as Lactobacillus Firm-4 and Firm-5 as well as Bifidobacterium spp. [11,12,13]. The remaining four strains Frischella perrara, Bartonella apis, Parasaccharibacter apium and a specific species group of Gluconobacter, “Alpha 2.1”, are less abundant and exhibit an unstable distribution within the intestinal tract of Apis mellifera. However, these strains are notably abundant, widespread and stable in A. cerana [14].

The gut microbiota is influenced by various factors, including diet [15], host living environment [16] and host genetic background [17]. Diet serves as a predominant factor in shaping gut microbial communities, which is closely associated with environmental conditions [18,19]. Honeybees collect pollen as a source of protein, amino acids, lipids and vitamins for maintaining physiological processes [20]. When pollen is scarce in the external environment, beekeepers usually feed their colonies pollen or pollen substitutes. A previous study demonstrated that the species diversity and evenness of gut microbiota were significantly different between pollen-fed and pollen-free diet-fed honeybees [21]. In bumble bees, the gut bacterial communities are also somewhat shaped by pollen diets [15]. Bees rely on pollen for essential nutrition, and they may require a more diverse bacterial community to effectively digest pollen. This is particularly true for species like A. cerana, which is adapted to complex environments. However, the effects of different pollens and pollen substitutes on the gut microbiota of A. cerana remain poorly understood.

Oilseed rape is the most widely cultivated oil crop in China; camellia is a crucial cash crop in the southern mountainous regions; lotus is extensively planted in wetlands of the middle and lower reaches of the Yangtze River; and buckwheat is widely cultivated on marginal lands such as the Yunnan–Guizhou Plateau. These four pollen types represent regionally significant crops cultivated across diverse Chinese landscapes, and beekeepers practice migratory beekeeping to access these pollen sources during their respective flowering seasons. In the present study, we evaluated the effects of four monofloral pollens—oilseed rape pollen (RP), camellia pollen (CP), lotus pollen (LP) and buckwheat pollen (BP)—and two commercial pollen substitute diets (Diet 1 and Diet 2) on the bacterial communities of A. cerana using high-throughput 16S rRNA sequencing. The newly emerged worker bees were fed with protein diets for 5, 10 and 15 days. The variations in microbiota composition, both between and within groups, were reflected in alpha diversity metrics. Principal coordinate analysis (PCoA) and linear discriminant analysis effect size (LEfSe) confirmed structural differences in the gut microbiota across the pollen diet groups. Furthermore, PICRUSt analysis indicated alterations in most predicted functional categories of the microbial community. This research represents a substantial step forward in elucidating the microbial diversity of A. cerana under different diets, which is the fundamental basis for understanding the host’s biological and ecological characteristics. Furthermore, it sheds light on the physiological adaptations of A. cerana to diverse environmental conditions.

2. Results

2.1. Characteristics of Sequencing Data

We conducted an analysis of gut bacterial communities in A. cerana that were subjected to various diets. After performing a series of procedures with the sequencing results, a total of 4,613,598 high-quality sequences were acquired, and the lengths of 100% of the effective sequence reads were between 410 and 430 bp. In this study, the sequencing data have a relatively high coverage of the real microbial community. The sequencing results of this study represent the actual composition of the microorganisms in the samples (Supplementary Tables S1 and S2).

2.2. Comparison of Bacterial Community

The gut microbiota alpha diversity differed depending on the diet and age of the workers (p < 0.05, Figure 1; Supplementary Table S3). Based on the Shannon and Simpson indices, the alpha diversity of the gut microbiota was significantly higher in the BP group than in the control on day 5 (p < 0.05). Regarding the Shannon index, on day 10, the RP group had significantly higher diversity than the control, the CP and the Diet 1 group. On day 15, a significant difference was detected only between the BP group and the Diet 2 group. There were no significant differences in Ace and Chao 1 indices among all groups on day 5 and day 15. However, significant differences were detected in the Chao 1 and Ace indexes between the CP group and the control on day 10 (Figure 1).

The microbiota community structure of workers supplied with different diets was analyzed at both the phylum and genus level (Figure 2). For the relative abundance of gut microbiota at the phylum level, the top four phyla (Firmicutes, Proteobacteria, Actinobacteriota, and Bacteroidetes) accounted for approximately 90% of the total sequences in the seven groups (Figure 2a–c). Among them, Firmicutes was the absolute dominant phylum (mean ± SD = 48.98 ± 10.07% of total sequences), followed by Proteobacteria (42.44 ± 10.90%), Actinobacteriota (3.80 ± 3.48%) and Bacteroidota (2.99 ± 3.67%). For the workers from the RP, CP, LP and BP groups, the relative abundance of Actinobacteriota showed an overall increasing trend on day 10 and day 15 compared to day 5. The relative abundance of gut microbiota at the genus level displayed wide variations, as shown in Figure 2d–f. Lactobacillus (42.90 ± 11.22%) and Gilliamella (16.94 ± 12.67%) were the two most prevalent genera in all seven diet groups. The relative abundance of Bombella and Apilactobacillus in workers from all diet groups generally showed a downward trend on day 10 and day 15 compared to that on day 5.

To identify the dynamic changes in dominant bacteria in different groups more accurately, the relative abundance of core bacteria at the phylum and genus levels were analyzed (Figure 3). At the phylum level, there was no significant difference in the relative abundance of Actinobacteriota, Bacteroidota, Firmicutes and Proteobacteria among the diet groups on day 5 (Figure 3A). On day 10, Actinobacteriota abundance in the RP and BP groups was significantly higher than that in the CP, Diet 1, Diet 2 and control groups (F6, 14 = 3.967, p < 0.05). The abundance of Firmicutes in the CP group was significantly higher than in the control, while the abundance of Proteobacteria was significantly lower than that in the control (F6, 14 = 4.986, p < 0.05; Figure 3A). On day 15, the abundance of Actinobacteriota in the RP and BP groups was significantly higher than that of the Diet 1, Diet 2 and control groups (F6, 14 = 4.251, p < 0.05). The abundance of Proteobacteria in the RP, CP and Diet 2 groups was significantly higher than that of the control (Figure 3A).

At the genus level, the content of Bifidobacterium, Gilliamella and Snodgrassella in the CP feeding group was the lowest among the pollen diet groups (Figure 3B). On day 10 and 15, Bifidobacterium abundance in the RP and BP groups was significantly higher than that of the Diet 1, Diet 2 and control groups (F6, 14 = 4.023, p < 0.05; F6, 14 = 4.430, p < 0.05, Figure 3B). On day 5 and 10, there were no significant differences in the abundance of Snodgrassella. However, the BP group had the highest Snodgrassella abundance on day 15, which was significantly different from that of the CP, LP, Diet 1 and Diet 2 groups (Figure 3B).

We compared the microbiota composition of workers and found significant differences among workers of different diet groups at different ages (day 5: R^2^ = 0.34, p = 0.009; day 10: R^2^ = 0.32, p = 0.002; day 15: R^2^ = 0.31, p = 0.035) (Figure 4; Supplementary Table S4). On day 5, the workers fed with BP possessed significantly different microbiota compositions compared with workers supplied with RP (p = 0.01) (Figure 4a). On day 10, the workers of the control group possessed significantly different microbiota compositions compared with workers supplied with RP (p = 0.001) and Diet 2 (p = 0.048) (Figure 4b). On day 15, there was no significant difference (Figure 4c).

The LEfSe analysis indicated significant differences in the gut bacterial taxa among the diet groups (Figure 5). There were 40 distinct microbial taxa with significant differences identified among the diet groups, with 7 taxa in the RP group, 3 in the CP group, 6 in the LP group, 15 in the BP group, 1 in the Diet 1 group, and 8 in the control group. At the genus level, Gilliamella (LDA = 5.13) and Bombella (LDA = 4.92) were significantly enriched in the control group, and Bifidobacterium (LDA = 4.83) and Lactobacillus (LDA = 4.68) were significantly enriched in the RP group, while Snodgrassella (LDA = 5.17) and Frischella (LDA = 4.48) were significantly enriched in the BP group. Pantoea (LDA = 4.79) and Klebsiella (LDA = 5.14) were significantly enriched in the CP and LP groups, respectively (Figure 5).

2.3. Functional Gene Prediction

The functional genes were involved in metabolism, genetic information processing, environmental information processing, human diseases, cellular processes, and organismal systems. The pathway involved in metabolism was the top function identified with the greatest abundance across all diet groups (Figure 6). The relative abundance of genes involved in metabolism in the control, RP, CP, LP, BP, Diet 1 and Diet 2 groups was 73.19%, 73.84%, 72.65%, 73.86%, 74.41%, 72.22% and 73.97%, respectively. For the secondary functional categories, we identified 44 secondary sub-functional genes, which included amino acid metabolism, carbohydrate metabolism, energy metabolism, etc. (Supplementary Table S5). While direct pathways related to specific human diseases (e.g., cancers, cardiovascular and immune diseases) did not show significant variation among the dietary groups, significant shifts were observed in pathways fundamental to physiological resilience and metabolic health. Specifically, the significant differences in environmental adaptation (p = 0.009) and cell growth and death (p = 0.017) suggest that the gut microbiota in different diet groups possess varying capacities to respond to external stress and regulate cellular turnover. Additionally, the significant variation in energy metabolism (p = 0.043) and the endocrine system (p = 0.012) indicates that the dietary treatments may differentially influence the host’s metabolic efficiency and hormonal regulation potential mediated by the microbiome.

3. Discussion

Foraging workers collect floral nectar as the principal carbohydrate source for the colony, while pollen serves as a rich supply of diverse carbohydrates, amino acids, lipids and vitamins [22]. While the consumption and utilization of pollen have been extensively studied, there is limited understanding of its effects on the gut microbiota of A. cerana. The gut microbiota can be regarded as a separate organ with its own metabolic functions, capable of processing indigestible food and producing beneficial products for the host [8]. Pollen is the primary source of dietary protein for honeybees [23]. It also includes complex polysaccharides in its outer coat, which are largely indigestible by bees but can be metabolized by bacterial species within the gut microbiota [18]. Dietary specialization of the honeybee has selected for substantive physiological changes and microbial associations over the course of evolution. Here, we quantified the effects of different pollen types on the establishment of bacterial composition in the gut of A. cerana.

The differences in richness and diversity in the gut microbiota between and within the diet groups were reflected in the alpha diversity analysis. The RP, LP and BP feeding groups had higher gut microbial diversity than the CP, pollen substitute and control groups. The result is consistent with research suggesting that pollen could increase the diversity of gut microbes [24]. The CP feeding group, which was associated with lower caffeine and santonin levels, exhibited reduced microbial diversity compared to other pollen groups. This reduction may be attributed to the effects of caffeine or other compounds on gut microbiota colonization [25]. Our study is consisted with previous studies that the host’s diet has a significant impact on the diversity and richness of gut microbiota [26,27].

Our results reveal that the core bacterial species, which typically constitute the honeybee gut bacterial community, were present in varying degrees in different diet groups. At the phylum level, the dominant bacteria were Firmicutes and Proteobacteria, followed by Bacteroidetes and Actinobacteriota, which is consistent with recent findings [13]. Kim et al. [28] showed that the dominant bacteria in the bees may not change due to external factors such as nutrients. However, our results indicated that while dominant bacteria present across groups, their relative abundance varied with both diets and age of bees. Firmicutes, Proteobacteria and Actinobacteriota are considered beneficial gut inhabitants of animals and involved in immunomodulation, interference with enteric pathogens and the maintenance of healthy homeostasis [29]. Firmicutes and Proteobacteria play important roles in the digestion and absorption of food [30,31]; in particular, Proteobacteria can digest secondary metabolites produced by the host and help maintain the growth and development of insects [32]. The high abundance of Firmicutes and Proteobacteria in our research highlights their functional significance in the digestion process. This supports the broader ecological principle that the nurse bee’s microbiota, often characterized by a distinct abundance profile, enhances pollen-derived nutrient extraction for brood care, whereas the forager’s microbiota prioritizes energy metabolism and pathogen defense. The observed dietary and age-related shifts in phylum-level abundance provide a functional perspective on community composition, indicating that the synchronization between host development and microbial ecology forms a foundation for colony-level health and functional division of labor.

At the genus level, the dominant genera are Lactobacillus and Gilliamella, followed by Bombella, Snodgrassella and Bifidobacterium. Lactobacillus has a protective effect on bees, which can enhance the expression of disease resistance genes, improve the immune response of bees [33] and protect bees from bacteria, yeast and pathogens [34]. The abundance of Lactobacillus in workers supplied with pollen increased with the age of bees, suggesting that pollen diets improved their immune capacity to ensure better growth and development. Bifidobacterium has the capacity to break down semi-fibrous and pectic substances and possesses antioxidant enzymes involved in oxidative metabolism [35]. The high abundance of Bifidobacterium in the four monofloral pollen diets on day 10 and 15 indicates that natural pollens are more beneficial for bees’ metabolism. Additionally, Lactobacillus and Bifidobacterium are closely related to gut homeostasis in bees [36,37]. Gilliamella has potential effects on pathogen defense and nutrient acquisition [38] and participates in the digestion of the host’s carbohydrate-rich diet [39], where it can degrade pectin in pollen cell walls and ferment substances [40]. In addition, Snodgrassella plays an important role in the metabolism and defense of bees [41]. The overall abundance variation of Gilliamella and Snodgrassella was not significant. In previous research, Kešnerová et al. [42] found that the colonization of Snodgrassella was not affected by dietary status, intestinal physicochemical conditions or other bacteria abundance, which was possibly due to Snodgrassella’s niche being host-dependent rather than food-dependent [10,13]. Therefore, the gut microbiota of A. cerana appears to be structured as a stable, host-adapted core that safeguards fundamental gut functions and colonization resistance, with a dynamically regulated layer that fine-tunes physiology in response to conditions, which were likely shaped by long-term evolutionary adaptation.

In addition to the core bacteria, a low abundance of non-core bacteria such as Serratia, Apilactobacillus, Klebsiellal and Pantoea was found in bee gut samples. These bacteria made up a relatively small fraction of the overall community, suggesting that they represent opportunistic or transient colonizers [42], which could be linked to the environmental factors or developmental stages within the colonies [43]. Serratia is considered an opportunistic pathogen of insects [44]. Apilactobacillus is a flower-associated bacteria and an obligately fructophilic Lactobacillus [45], which contributes to bee health by inhibiting the proliferation of certain pathogens, such as Serratia marcescens [46]. The presence of these low-abundance, transient bacteria highlights the complex interface between the bee gut and its environment. Their dynamics might serve as a sensitive bio-indicator for hive health or environmental exposure.

The composition of the gut microbiota was different among the workers of the different diet groups. In our study, PCoA revealed a distinct difference in the compositions of gut bacterial communities in the pollen diet groups. The microbiota of bees fed pollen diets clustered separately with the control on day 10 and 15, suggesting that pollen diets affected microbiota composition with the age of bees. Even though bees harbor a consistent core microbiota across the globe, the gut bacterial composition dynamic changes due to environmental factors, especially diet [15]. Frischella was the genus with a significant difference in the BP group, suggesting that Frischella may be a biomarker of BP-feeding bees. Frischella has metabolic capabilities to gain energy via anaerobic fermentation of carbohydrates [47], implying the metabolic capacity of bacteria in BP-feeding bees. PICRUSt2 function prediction analysis revealed a diverse range of genetic variation, genetic information, cellular processes and metabolites. Similar to the research of Khan et al. [48], we found that the aggregation amount of genetic furniture with metabolic functions was the highest, including amino acid metabolism, carbohydrate metabolism, energy metabolism, global and overview maps, the metabolism of cofactors such as vitamins and nucleotide metabolism, and the metabolic differences among various feeds mainly focuses on energy metabolism pathways. Optimizing these functions is vital for preventing metabolic disorders and promoting the general well-being iofn the workers. From a societal safety perspective, it is noteworthy that the abundance of genes related to antimicrobial drug resistance did not differ significantly among the groups. This suggests that the tested diets do not disproportionately select for antibiotic-resistant bacteria, alleviating concerns regarding the potential spread of resistance genes in the environment. Collectively, these functional interpretations imply that while the diets do not induce specific disease-associated pathways, they actively modulate core metabolic and adaptive functions that are essential for maintaining host health and ecological resilience.

4. Materials and Methods

4.1. Sampling of Workers

This study was conducted during November 2023 and January 2024 at the Sericulture and Apiculture Research Institute, Yunnan Academy of Agricultural Sciences (116.33° E, 39.96° N). Four A. cerana colonies headed by sister queens were used in this study. They were maintained under standard management conditions. Combs with sealed brood from these colonies were placed in an incubator (34 °C and 75% relative humidity). The workers emerged within 24 h were collected and mixed. Subsequently, 30 newly emerged workers were randomly selected and transferred to the experimental cage and reared at 30 °C, 50% RH.

We used oilseed rape pollen (RP), camellia pollen (CP), lotus pollen (LP) and buckwheat pollen (BP) in this study. These four monofloral pollens were collected in 2023 using pollen traps installed at the entrance of Apis mellifera hives. The pollens were dried and stored at −20 °C. Before usage, we manually sorted these pollens based on color and appearance to ensure purity of the pollen pellets above 95%. Pollen substitute #1 (Diet 1) and pollen substitute #2 (Diet 2) were purchased from two commercial companies. Diet 1 primarily consists of plant germ, plant protein, yeast powder, lysine and methionine. Diet 2 is formulated with a blend of multiple plant proteins, carrot powder, yeast powder, Lactobacillus, amino acids, minerals, trace elements and a vitamin complex. The two pollen substitutes are powders and claimed to be formulated for A. cerana.

We prepared feed paste by mixing pollen pellets and pollen substitutes with 50% (w/w) sucrose solution. Newly emerged workers of the treatment groups were supplied with protein pastes of RP, CP, LP, BP, Diet 1 and Diet 2 separately. The control group was fed with only 50% sucrose solution. Three biological replicates were set up per group, with each replicate comprising the hindguts of three bees. Each time, workers in the rearing cage of the treatment groups were provided with 2 g of protein diets and 1.5 mL of 50% sucrose solution simultaneously, which ensured that neither were fully consumed within 24 h. The protein diets and sucrose solution were replaced daily. On day 5, 10 and 15, we randomly selected 10 workers from each cage.

4.2. DNA Extraction

The sampled workers were anesthetized on ice for 5 min. Then, they underwent surface sterilization using 75% ethanol. Subsequently, the worker was carefully dissected and its hindgut was transferred into a 1.5 mL centrifuge tube using sterile forceps. The hindguts of ten workers were pooled together and immediately frozen in liquid nitrogen and stored at −80 °C until genomic DNA extraction.

Total genomic DNA of samples was extracted using a Stool DNA Kit (Tiangen Biotech, Beijing, China). The target V3-V4 region of the bacterial 16S rRNA gene was amplified using specific primers (forward primer 338F: 5′-ACTCCTACGGGAGGCAGCA-3′; reverse primer 806R: 5′-GGACTACHVGGGTWTCTAAT-3′) [49]. PCR was performed using KOD-FX polymerase (Toyobo, Osaka, Japan) in a 10 μL reaction volume according to the manufacturer’s instructions. The PCR reaction mixtures contained 0.05 μL of template DNA, 0.3 μL Vn F (10 μM), 0.3 μL Vn R (10 μM), 5 μL KOD FX Neo Buffer, 2 μL dNTPs (2 mM each), 0.2 μL KOD FX Neo and 2.15 μL ddH_2_O. PCR amplification was conducted according to the following program: 95 °C for 3 min; 25 cycles of denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s, and extension at 72 °C for 40 s; final extension at 72 °C for 7 min. The purified PCR products were quantified and the libraries were prepared with a VAHTSTM Universal Plus DNA Library Prep Kit for Illumina (Vazyme, Nanjing, China). Sequencing was conducted on the NovaSeq 6000 platform (Illumina, San Diego, CA, USA).

4.3. Diversity Analysis and Statistical Analysis

The bioinformatics analysis of this study was conducted on the BMK Cloud platform. Raw reads were initially quality-filtered with Trimmomatic [50] (v0.33), followed by primer removal using Cutadapt [51] (v1.9.1). Paired-end reads were assembled with USEARCH [52] (v10), and chimeric sequences were removed with UCHIME [53] (v8.1). We chose the OTU-based approach to align with methodologies used in studies on the bee microbiome and to ensure direct comparability with the existing literature [14,48]. High-quality reads were subsequently clustered into operational taxonomic units (OTUs) at 97% similarity using USEARCH [50] (v10.0), and OTUs with an abundance below 0.005% were discarded. Taxonomic annotation was performed in QIIME2 [54] using a Naïve Bayes classifier trained on the SILVA database [53] (release 132) with a confidence threshold of 70%. Alpha diversity was assessed using four indices (Shannon, Simpson, Chao1, and Ace) computed with QIIME2 (v2020.6) and R software (v 4.2.3). Beta diversity was evaluated via PCoA in QIIME2. Additionally, LEfSe [52] was employed to identify significantly different taxa among groups, using an LDA score threshold of 4.0. Microbiota functions were predicted by annotating pathways against the KEGG database using PICRUSt2. Statistical analyses were generated using Microsoft Excel 2016 and IBM SPSS Statistics v25 (IBM Corp., Armonk NY, USA). Data visualizations were constructed using GraphPad Prism 9 (GraphPad Inc., La Jolla, CA, USA).

5. Conclusions

In summary, the present study outlines an investigation of the structure of the gut microbiota of A. cerana fed different pollen diets. It has been consistently observed that host diet and age have an impact on shaping the bacterial communities. Gut bacterial communities were influenced by host diets and may play an important role in host adaptation. However, there are still many limitations in our study. Future studies should aim to combine the effects of diet on gut microbes and host growth and development and evaluate the functional role of microbiota in affecting the growth and physiology of A. cerana, enabling the selection of diets that are more beneficial to A. cerana. Functional feed formulations should aim to directly optimize the honeybee gut microbiota and regulate bee growth. Altogether, our study provides a theoretical basis for the study of gut symbiotic bacteria in A. cerana.