*2.1. Microbiome and Pollen Composition of Honeybees from Poland (Differences over the Vegetation Season)*

Following NGS analysis, it was possible to assign samples to the time that they were collected by comparing them with vegetation periods of nectar- and pollen-rich plants. From one location, 3 specimens (forager honeybees) were taken, as the representative and consistent number for each group (data adequacy confirmed by the principal component analysis (PCA) analysis of amplicons of the 16S rRNA from bacteria and the ITSs region from fungi and plants, Supplementary Materials Figure S1). Microbiome down to genera analysis enabled division of Polish honeybee samples into 6 sub-groups: PL1, PL2, PL3, PL4, PL5, and PL6 (Supplementary Materials Table S1).

PL1 forager honeybees were collected in April and 16 taxa were identified in total from 16S amplicon analysed and 164 taxa in ITS2 analyses. In PL1 a low number of pollen types was observed (1.81%, SD = 0.512 belonging primarily to the *Betula* sp. and *Urtica* sp. genera) and a high number of fungi (93.83%, SD = 0.554). The low number of pollen types may relate to the low availability of flowering resources at the start of the flowering season, when the bees have to consume the stored feeding solution in the colony, or after feeding

by the beekeeper later in the season. A sugar-based diet can lead to higher yeast numbers as observed in PL1 (35.92%, SD = 3.208) (Supplementary Materials Tables S1 and S2). Furthermore, the pathogenic fungus *Nosema ceranae* was detected in the PL1 (Supplementary Materials Table S3).

PL2 forager honeybees were collected in May and 16 taxa were identified in total from 16S amplicons analysed and 147 taxa in ITS2. *Lactobacillus* spp. were the dominant bacteria in PL2 (43.02%, SD = 1.704) (Supplementary Materials Table S1). At the same time, the percentage of fungi was moderate (9.59%; SD = 4.476) with the prevalence of *Cladosporium* and a small content of fungi from other genera (Supplementary Materials Table S2). The dominant pollen content in the PL2 with 87.54% (SD = 3.886) came from plants from the *Brassicaceae*, mainly unassigned, and speciesof *Raphanus* (Supplementary Materials Table S1B) known for its high protein pollen content [31]. No pathogens were detected in PL2 (Supplementary Materials Table S3).

PL3 forager honeybees were collected in June and 18 taxa were identified in total from 16S amplicons analysed and 177 taxa in ITS2. This group shows a higher number of *γ*proteobacteria from *Gilliamella* and *Snodgrassella* (54.03%, SD = 2.333) (Supplementary Materials Table S1). PL3 honeybees collected pollen from Polygonaceae, which contain moderate amounts of amino acids [48]. In this group, fungal load was moderate (11.25%, SD = 1.847) with *Aspergillus*, *Cladosporium* and *Mycosphaerella* (mainly transferred as bioaerosols by wind in the air) at the level of 3.76% (SD = 1.9486), 1.21% (SD = 0.0961), and 0.81% (SD = 0.624), respectively (Supplementary Materials Table S1B). No pathogens were detected in PL3 (Supplementary Materials Table S3).

PL4 forager honeybees were collected in July and 16 taxa were identified in total from 16S amplicons analysed and 322 taxa in ITS2. This group was differentiated on the basis of the highest fungal DNA loads (87.31%, SD = 1.680) and trace amounts of plant DNA (Supplementary Materials Table S1B). The fungi were mainly *Aspergillus* (15.98%, SD = 0.503), *Cladosporium* (10.91%, SD = 1.048), *Penicillum* (11.07%, SD = 2.190), and *Betsia* (11.20%, SD = 1.572) spp. The spores of the three first genera commonly appeared in the air (transferred as bioaerosols), but the presence of *Betsia* species is suggestive of poor health of the honeybee colony. The pollen mould (*B. alvei*) is a saprophyte that lives on the pollen stored combs, especially in temperate regions [49]. Furthermore, pathogens belonging to *N. ceranae* and neogregarines were detected in the PL4 (Supplementary Materials Table S3). Moreover, a high fungal load and a small amount of plant pollen indicates that honeybees most likely gained sugar based diet. At the same time, PL4 honeybees had high amounts of *γ-*proteobacteria, *Orbales*, *Gilliamella* (35.32%, SD = 0.370) *α-proteobacteria Rhizobiales*, *Bartonella* (27.06%, SD = 0.560), known honeybee gut symbionts (Supplementary Materials Table S1A). *Gilliamella apicola* was found to be a dominant gut bacterium in honeybees and bumble bees, and this bacterium simultaneously utilizes glucose, fructose and mannose, and has the ability to break down other potentially toxic carbohydrates [50].

PL5 forager honeybees were collected in August and 27 taxa were identified in total from 16S amplicons analysed and 284 taxa in ITS2. PL5 diet was mainly based on *Helianthus* sp. (34.98%, SD = 1.616). PL5 had a higher number of *Lactobacillus* (44.31%, SD = 1.456) and *Gilliamella* (28.46%, SD = 0.129) species (Supplementary Materials Table S1A). The fungal load was medium (59.16%, SD = 2.225) with bioaerosol *Cladosporium* (11.23%, SD = 2.225) and *Mycosphaerella* (6.69%, SD = 1.358) as the main representatives (Supplementary Materials Table S1B). No pathogens were detected in PL5 (Supplementary Materials Table S3).

PL6 forager honeybees were collected in September, and 28 taxa were identified in total from 16S amplicons analysed and 191 taxa in ITS2. PL6 had the highest level of *Firmicutes* and *Lactobacillus* with 65.37%, SD = 0.731, and a slightly higher number of bacteria from the Comensalibacter group (Supplementary Materials Table S1A). PL6 honeybee diet was rich in *Calluna* pollen (88.26%, SD = 0.585), which displays low protein content and is considered to be a poor source of food for bees, and thus destructive

for colony development [51,52]. The fungal load was moderate (11.26%, SD = 0.718) with *Penicillium* as a dominant genus with 0.96%, SD = 0.104 (Supplementary Materials Table S2). Pathogens belonging to *N. ceranae* and neogregarines were detected in PL6 (Supplementary Materials Table S3).

Generally, PCA analysis of stores from spring honeybees (PL1 and PL2), summer honeybees (PL3 and PL4) and autumn honeybees (PL5 and PL6) involved splitting data into five major components which accounted for 100% of the variation. It can be concluded from the PCA analysis that PL3 clearly differed from the others, and was mainly influenced by bacteria (b) and plants (p). PL2 and PL6 were similar to each other, bacteria (b) and plants (p) having the greatest impact as well. PL1 and PL5 were similar to each other. The greatest, albeit low, influence was caused by bacteria (b) and fungi (f). PL4, similarly to PL3, clearly stood out from the others, and was mainly influenced by plants (p) and fungi (f). It could also be generally seen that bacteria had the greatest influence on the variability of the plant pollen-bacteria-fungi system, followed by the plants, and then the fungi. The first two components accounted for over 53% of the variability of the entire system. Positive PC2 values may describe the summer months, and negative PC2 values the months closer to spring and autumn (Supplementary Materials Figure S1).

PC1 and PC3, the two main components, account for approximately 50% of the system variability (Supplementary Materials Figure S1a,b). It can be concluded that positive PC3 also described spring and autumn values, and negative PC3 values described the summer, which clearly differs from the others. Therefore, the third component described the season.

To summarise, we observed a prevalence of *Lactobacillus* and *Bartonella* species in honeybees collected during spring (April PL1, May PL2) and autumn months (September PL6), while in summer months, (June PL3, July PL4, August PL5) microbiome analyses showed the prevalence of *Gilliamella*, which is in agreement with previous findings [4,9,43]. In late July (PL4) the physiology of honeybees changes due to their adaptations to overwintering, and a role from two types of bacterial groups, i.e., *Lactobacillus* and *Giliamella* is suggested to be played in this process. However, increased *Lactobacillus* and *Giliamella* occurrence may simply be a consequence of a protein-rich diet. We observed fluctuations in the microbiome composition correlating with changes in the protein-richness of the pollen available in the environment. During spring and autumn, it is common for honeybees' diet to be based on sugars ingested from honeydew. These high sugar diets can lead to fungal infections observed in April (PL1), May (PL2) and June (PL3) in honeybee samples. Probably, the detected pathogens were present in the honeybee colonies throughout the season but owing to the colony biology and well-balanced diet observed during May (PL4), June (PL3) and August (PL5), infections were less frequent among foragers and may have gone unnoticed in the whole colony screening tests.

#### *2.2. Microbiome and Pollen Composition of Honeybees from UK, Greece, Spain, and Thailand*

Forager honeybees from the UK were collected on the roof of the Fogg Building at Queen Mary University, London (UK1) and in the garden of the Natural History Museum, London (UK2), in July 2019. The mean day temperature in the first half of July was 21.93 ◦C (71.47 ◦F) with average humidity equal to 65% [53]. Their bacterial microbiota was mainly composed of *Firmicutes* (*Lactobacillus*), *γ-proteobacteria* (*Orbales*, *Gilliamella*), *α-proteobacteria* (*Rhizobiales*, *Bartonella*), *γ-*proteobacteria (*Neisseriaceae*, *Snodgrassella*), and *Actinobacteria* (*Bifidobacterium*) (Supplementary Materials Table S2A). In total, 42 and 93 taxa were identified (species level) from the 16S amplicon analysis, and 70 and 94 taxa (species level) in the ITS2 in the UK1 and UK2 groups, respectively. The UK1 sample contained modest amounts of fungi (12.59%) with the prevalence of plant pollen (87.41%) derived from Apiaceae (*Ammi*), Fabaceae (*Astragalus*), Resedaceae (*Reseda*), and Fabaceae (*Styphnolobium*). UK2 honeybees foraged on Hydrangeaceae (*Hydrangea*), and Bignoniaceae plants (Supplementary Materials Table S2B). The UK2 sample was dominated by fungi (90.80%), mainly from the Myxotrichaceae, which is reported to be a common hive fungus in Europe [54,55]. Moreover, the fungal pathogen *N. ceranae* was detected in the UK2 (Supplementary Materials Table S3).

Forager honeybees from Greece (GR1, GR2) were collected in November from two colonies inhabiting the garden of the Agricultural University of Athens. The mean day temperature in the first half of November was 18.53 ◦C (65.35 ◦F) with average humidity of 25% [56]. In total, 80 and 31 taxa were identified from the 16S amplicon analysis, and 42 and 96 taxa in the ITS2 in the GR1 and GR2 group respectively. Although the colonies were close, their microbiota and food preferences differed. GR1 microbiota were mainly Actinobacteria *(Bifidobacterium)*, Firmicutes *(Lactobacillus)*, and Cyanobacteria which reached 54.41%, 26.84%, and 11.66%, respectively. Cyanobacteria indicated some colony health problems most probably connected with the contamination of water used by the honeybee colony (Supplementary Materials Table S2A). Moreover, pathogens belonging to *N. ceranae* and neogregarines were detected in the GR1 (Supplementary Materials Table S3). The fungal load was miniscule (7.94%) and contained mainly fungi present in the air transferred as bioaerosols by wind, such as Mycosphaerella and *Cladosporium*. GR1 honeybees foraged mainly on Eudicotyledonae plants. GR2 microbiome contained α-proteobacteria (Rhizobiales, *Bartonella*), Firmicutes (*Lactobacillus*), and γ-proteobacteria (Orbales, *Gilliamella*) with 49.29%, 25.84%, and 9.78%, respectively. The amount of fungi was miniscule (6.03%) with the dominant taxon of Mycosphaerella reaching 1.09%. GR2 honeybees foraged mainly on Oleaceae (*Ligustrum*), Hydrangeaceae (*Hydrangea*), Myrtaceae (*Myrtus*), and Scrophulariaceae (*Buddleja*) (Supplementary Materials Table S2B).

Forager honeybees from Spain (ES1, ES2) were collected in November from experimental colonies located near Marchamalo. The mean day temperature in the first half of November was 18 ◦C (64.40 ◦F) with average humidity equal to 51% [57]. In total, 34 and 25 taxa were identified from the 16S amplicon analysis, and 31 and 13 taxa in the ITS2 in the ES1 and ES2 groups, respectively. ES1 contained γ-proteobacteria (Orbales, *Gilliamella*), and Firmicutes (*Lactobacillus*), at the level of 36.63%, and 33.11%, respectively (Supplementary Materials Table S2A). These honeybees most probably gained sugar based diet, since pollen DNA was hardly detected (Araliaceae (*Hedera*) 0.12% for ES1 and 0.04% for ES2). ES1 and ES2 had dominant fungal fraction containing spores transferred as bioaerosols by wind in the air, such as *Penicillium*, *Cladosporium* and *Mycosphaerella* (Supplementary Materials Table S2B). Additionally, pathogens belonging to *N. ceranae* and neogregarines were detected in ES2 (Supplementary Materials Table S3).

Forager honeybees from Thailand consisted of both: western honeybee (*Apis mellifera* TAI1 and TAI2) and Asian honeybee (*Apis cerana* TAI3 and TAI4). In Eastern Asia, these two bee genera inhabit the same locations, resulting in the transfer of pathogens from *Apis cerana* to *Apis mellifera*, as described for *Varroa destructor* and *Nosema ceranae*. Thai samples were collected in February, the best month in the year for honeybee colonies in Thailand. The mean day temperature in the first half of Februry was 33.33 ◦C (91.99 ◦F) with average humidity equal to 59% [58].

In western honeybee (*Apis mellifera* TAI1 and TAI2), in total 29 and 22 taxa were identified from the 16S amplicon analysis, and 50 and 94 taxa in the ITS2 in the TAI1, TAI2 groups, respectively. TAI1 contained high loads of Firmicutes (*Lactobacillus*) and *Bartonella* (Supplementary Materials Table S2A), and a high quantity of fungi such as *Aspergillus* (66.42%), Saccharomycetaceae (9.09%), and *Peniclilium* (6.53%) (Supplementary Materials Table S2B). A trace amount of plant pollen was detected in TAI1 (*Pterocarpus* with 0.16%, *Mimosa* 0.08%) indicating sugar based diet to have been the main source of forage (Supplementary Materials Table S2B). Bacterial microbiota of TAI2 was mainly composed of *Arsenophonus* bacteria (γ-proteobacteria, Enterobacteriales) 92.11% of which were insects' intracellular symbionts. *Arsenophonus* species showed a broad spectrum of symbiotic relationships varying from parasitic son-killers to coevolving mutualists [59]. Moreover, pathogens belonging to *N. ceranae* and neogregarines were detected in TAI1 (Supplementary Materials Table S3). TAI2 honeybees foraged mainly on Asteraceae pollen

(91.33%) and contained only a miniscule amount of fungi transferred as bioaerosols by wind in the air, as *Cladosporium* 4.32%.

From Asian honeybee (*Apis cerana*) in total 19 and 42 taxa were identified from the 16S amplicon analysis, and 59 and 85 taxa in the ITS2 in the TAI3 and TAI4 groups, respectively. *Apis cerana* samples contained 70.74% of *Lactobacillus* genus for TAI3, 27.17% for TAI4, *Gilliamella*, 33.45% for TAI4, and *Snodgrassella* with 3.60% and 13.13% for TAI3 and TAI4, respectively. Fungi present in the samples were related with the air bioaerosols, including *Aspergillus* 42.77% for TAI3, *Cladosporium* 59.03% and *Pleosporales* 30.38% for TAI4. The amount of plant pollen was miniscule (*Pterocarpus* with 0.42% for TAI3, *Mimosa* 0.15% for TAI4) indicating sugar based diet as the main forage for Thai *A. cerana* bees (Supplementary Materials Table S2). Moreover, pathogens belonging to *N. ceranae* and neogregarines were detected in TAI4 (Supplementary Materials Table S3).

PCA analysis of the honeybees' stores split data into five major components which accounted for 100% of the variation. PC1 and PC2 components accounted for nearly 57% of the variation, respectively 32.83% and 23.65 (Figure 3, figures a and b should be considered simultaneously).

**Figure 3.** Loading plot (**a**) and score plot (**b**) of the principal components' analysis (PC1 and PC2) carried out on the analytical data of the taxonomy detected in the world bees (Thailand A.c.—*A. cerana*). Small letters on loading plot (a): b—data obtained from bacteria NGS analysis, f—data obtained from fungal NGS analysis, p—data obtained from plant NGS analysis.

The PCA analysis allowed us to determine the differences between bees from different countries. Four different areas were distinguished. Clear differences can be seen between bee samples from Greece, UK, Spain, Poland, and the two samples of bees from Thailand (Figure 3b).

PCA analysis shows Greek samples to markedly differ from the others, which is mainly influenced by bacteria (b) and plants (p). During the sample gathering, the weather conditions in Greece were most similar to those of Spain. Therefore, there must have been reasons for the Greek samples differences other than the weather. Most probably, nutrition was of the greatest importance in this case. The PCA analysis findings for UK bee samples also clearly differ from the others, mainly due to bacteria (b), then plants (p) and finally fungi (f). One can also distinguish samples from Spain and Poland from the other samples, which is mostly influenced by bacteria (b) and fungi (f), however, these are similar to the results for the samples from Thailand which are mostly influenced by bacteria (b), plants (p) and fungi (f). Bacteria, and, to a lesser extent, plants and fungi can be said to have the greatest influence on the variability of the system (Figure 3a,b).

Furthermore, ANOVA analysis confirms the correlation between the health status of honeybees and some of their bacterial microbiota (Supplementary Materials Table S4 and [60]). Bacterial groups such as Firmicutes (*Lactobacillus*); γ-proteobacteria, *Orbales*, *Gilliamella*, *γ-*proteobacteria, Neisseriaceae, *Snodgrassella*, Enterobacteriaceae, and other unidentified bacteria had significantly different loads in healthy and in *Nosema ceranae* and neogreagarine infected honeybees. The load of fungi was always higher in infected honeybees (*p* = 0.002429) whereas the load of plant pollens was always higher in healthy honeybees (*p* = 0.030446).

Foragers are worker honeybees of a similar age, when they collect water, nectar, and pollen as well as supplements necessary for the colony to survive. All forager bees have similar function and physiological processes [61] and similar microbiota. Therefore, they should share similar microbiota. However, studies indicated that forager honeybees have a "contingent microbiome" dependent mainly on the food they forage [2,10,11]. This carries a danger, because with poor food resources, the microbiota will be inappropriate and non-functioning [62]. Currently, many factors influence bee microbiota e.g.,: monocultures, nutritional stress, pesticide exposure and agrochemicals, many of which exhibit antimicrobial properties, and thus contribute greatly to reductions in honeybee stress tolerance and disease resistance, leading to higher honeybee mortality, and a high rate of colony loss [5,63], pathogens which trigger bee malnutrition [64], changes in the composition of their microelements [65] and yeast content [19].

Intestinal pathogens such as microsporidia and neogragarines can strongly interfere with bee microbiota (Supplementary Materials Table S4). During *Nosema*-infection the honeybee intestine is covered by a layer of mature spores which is the cause of deprivation of the physiological function of the bee alimentary tract for food absorption [64]. Recent studies have revealed the *N. ceranae* infection course, showing a spring peak, and a subsequent decline in summer and autumn [19,66,67]. These studies also confirmed the seasonal pattern of *Nosema* infection, as in samples taken during April (PL1) and July (PL4, UK2) *N. ceranae* was detected. However, the presence of the parasite in Autumn samples (GR1, ES2) may indicate the colony health problem, and may pose a threat to overwintering. It is worth emphasizing that *N. ceranae* was found both in *Apis mellifera* (TAI1) and *A. ceranae* (TAI4) Thai samples. Besides *Nosema*-infection in samples PL4, PL6, GE1, ES2, TAI1, TAI4 also harboured other intestinal parasite such as neogregarines. Neogregarines, since their first detection in *A. mellifera* and *Bombus* sp. described in 1992 were linked to declines in bee populations [68–70]. Neogregarines, inhabit the intestines of many invertebrates and lead to the impoverishment of the host's organism. Neogregarines' interaction with their bee hosts had not been deciphered in full detail and still more studies need to be undertaken in regard to nutrient uptake and malnutrition, facilitating susceptibility to other diseases, etc.

Pollination is a crucial process for the maintenance of plant-based food supplies [71]. To maintain the health of honeybees, it is favourable to prevent the spread of disease, prevent exposure to insecticides and pesticides and provide a variety of plants to maintain optimal nutrition and microbiome throughout the season [72]. The use of NGS techniques in the identification of the pollen pool preferentially chosen by honeybees can provide strategies to maintain healthy colonies of bees. This technique can greatly expand and supplement knowledge based on long term observation for the most efficient "pollinatorfriendly" plants [73–75]. This should help to create more effective pollinator beneficial plant species composition for "pollinator-friendly" gardens or to enrich the plant species on flower strips.

#### **3. Conclusions**

Honeybee dietary preferences, developed during the course of evolution, may not be currently favourable for honeybees. It is an urgent issue that should be carefully studied to aid bees to survive in the anthropogenic biosphere.

In our research, the composition of honeybee microbiomes was mainly determined by dietary preferences and forage availability. Even when colonies originated from one apiary, for instance from Spain or Greece, honeybees chose different plants to forage on. UK bees from a single highly urbanised area (London) exhibited particularly high diversity, chose different food sources and were otherwise prone to diseases. Furthermore, honeybees choosing a sugar based diet were more susceptible to pathogens (*Nosema ceranae* and neogregarines). The period when honeybees switch to the winter generation (longer-lived forager honeybees) was the time when the whole colony proved most sensitive to dietary perturbations.

Our findings are in line with limited other reports, that suggested honeybees from varying apiaries make independent decisions on the choice of pollen and nectar they forage upon [76]. As observed, colonies of bees located in close proximity may have different pollen composition in their gut. It remains unclear how bees decide which pollen to forage. However, flower structure, nectar volume, sugar content and composition were indicated to play a role in attracting bees [51,52,73]. Some studies indicated that honeybees primarily chose pollen rich in essential amino acids [51,52,73]. On the other hand, honeybees, because of their considerable energy need and having being weakened by disease, were unable to undertake forage flights to collect good quality pollen, and collected pollen of poorer nutritional quality. It is possible that evolutionary adaptation of bees has failed to benefit them in the modern anthropomorphised environment.

#### **4. Highlights**

