Introduction

Ecosystem services consist of several benefits provided to humans by ecosystems and sometimes transformed into economic profits [52]. Pollination is considered a key ecosystem service [53] involving 75% of globally important crops [31] and also contributing to the maintenance of plant biodiversity [22]. The highest efficiency in crop pollination is achieved when wild pollinators and managed bees cohabit the same ecological niche, co-pollinating crops [32].

The Bombus genus (bumble bees) consists of approximately 250 known species worldwide, mostly proliferating wild, although some of them can be reared for the pollination of several crops like strawberries, peppers, and tomatoes. Within the native bee diversity present in Argentina, eight species of bumble bees were identified [1], and among these, Bombus pauloensis is the most abundant one [3]. Being a generalist pollinator with a well-known life cycle, in South America, B. pauloensis began to be reared and managed as a commercial pollinator to the same extent as Bombus terrestris in Europe. Another bumble bee species of relevance in Argentina is Bombus bellicosus, which is distributed from the North to the Patagonian region [2]. Recent studies identified a connection between climate change and the decline of the B. bellicosus population in Argentina. In fact, B. bellicosus is experiencing a significant decline in colony abundance in its native territory []64, together with a reduction in the area originally populated by the bumble bee itself [29]. Both B. pauloensis and B. bellicosus are known as eusocial bees with nesting attitude in soil cavities covered with a layer of plant debris [34, 47].

Despite their importance in agricultural and natural systems, many wild and managed bee species are suffering a sharp decrease in their population numbers [, 2567]. Pathogen infections represent the main biohazard for bumble bees together with habitat loss due to intensive urbanization or agriculture [20, 65].

Bombus pauloensis can also be affected by exogenous pathogens spilled over by honeybees like Nosema ceranae [28, 18]. However, for N. ceranae infection, there is currently a conflict, since authors like Fernandez de Landa et al. [28] and Plischuk et al.[54] reported a high prevalence and abundance of N. ceranae in South American bumble bees, but, at the same time, Ngor et al. [49] and Gisder et al. [33] have demonstrated for European bumble bees that even high abundance of N. ceranae spores in the gut is not correlated with an actual infection.

The gut microbial communities harbored by bees can enhance both host growth and health [24, 43]. These microbes are involved in specific functions, such as food digestion and nutrient acquisition, regulation of immune responses, and defense against pathogens and parasites [, 771, 4]. The correct balance of the bee gut bacterial communities is critical for maintaining bee health at the individual and colony levels [58]. However, knowledge on bee gut microbial communities is not spread among all bee species, as most studies to date have focused on A. mellifera, and few studies on wild pollinators [39, 50]. However, the positive effects of symbiotic microorganisms on wild pollinators have been reported in recent years. For instance, Hammer et al. [35, 36] and the work by Koch et al. [43] highlighted the crucial role of gut microorganisms in the conversion and activation of nectar metabolites, providing a natural defense against parasites and enhancing the health of B. terrestris. Finally, Steffan et al. [62] detailed in their review how microorganisms benefit both solitary and social bees through pollen digestion (microbes facilitate the breakdown of the resistant outer layer of pollen grains), nutritional supplementation (bee larvae consume microbes present in fermented pollen provisions), pollen preservation (microbes alter the pollen substrate, making it less accessible to competing microbes), and mutual defense.

This study aims to fill gaps in the current knowledge about the gut microbiome composition of two bumble bee species native to South America: B. pauloensis and B. bellicosus. Furthermore, the study aims to evaluate how the prevalence and load of N. ceranae and land use correlate with the microbiome of both social species in Argentina.

Materials and Methods

Some methodological details have been described in Fernandez de Landa et al. [27], which regards the same study areas although focusing on solitary bee microbiome and its correlation with pathogens and land use, and briefly summarized in Fig. S1.

Study Areas

Three different sampling sites were selected in the Buenos Aires province before starting the study because of their different land uses and levels of anthropization: (i) an Actinidia deliciosa–producing farm (Santa Paula’s, 37° 56′ 0.69″ S; 57° 40′ 40.53″ W); (ii) a natural reserve (Reserva Natural Paititi, 37° 54′ 47.774″ S, 57° 48′ 44.806″ W); and (iii) a plant nursery (Vivero Antoniucci, 38° 1′ 42.014″ S, 57° 37′ 59.374″ W). The area with the highest human exploitation and impact was Santa Paula’s Farms (SP), where 91.5% of the farm area was employed for intensive cultivation, mainly kiwis. The Natural Reserve Paititi (NRP) was characterized by the lowest human impact, considering that almost 20% of the area is a nature reserve characterized by abundant natural flora (with a large presence of Eryngium regnelli, Baccharis tandilensis, and Lathyrus pubescebs). Furthermore, 49% of the NRP is employed for extensive agriculture. The plant nursery Vivero Antoniucci (VA) showed a large amount of floral resources for pollinators, although the majority of them are cultivated plants not native to the site. Therefore, considering the level of anthropization, VA is at an intermediate stage between SP and NRP. Analyzing the total area taken into consideration for the study (3 km radius, corresponding to double the known flight radius of native bumble bees), it was observed that VA has 15% high-density residential areas, 4% urban reserves (biological corridors), and over 75% of intensive agricultural areas.

Spatial Characterization

The spatial characterization of the three sampling areas was carried out using the Google Earth Engine platform and the open access software QGIS (https://qgis.org/). Landsat 8 satellite images were used with the atmospheric corrections made under the name “LANDSAT/LC08/C01/T1_TOA,” and only images with a cloudiness lower than 20% were considered. A representative image was constructed from the median of the images filtered in Google Earth Engine, which allowed working with a single image containing information from the entire sampling period. The constructed image was exported and processed with QGIS tools.

Moreover, spatial characterization was based on land use information available in the Geographic Information System repository of the province of Buenos Aires “urbasig” (https://www.urbasig.gob.gba.gob.ar). This information consists of official reports from the government of the province of Buenos Aires, detailing the portions of territory used for different human activities. Three points corresponding to the three sampling zones were added to the final images in QGIS, from which a 3-km radius buffer was constructed according to the foraging area of the bumble bee species. Figure 1 summarizes the images detailed above.

Fig. 1
figure 1

Spatial characterization. The central figure depicts the distribution of the three sampling sites, where A = Natural Reserve Paititi (NRP), B = Santa Paula’s Farm (SP), and C = Vivero Antoniucci (VA) (total distance between A-B and B-C > 10 km, and distance between A-C > 20 km). The three figures on the left side represent, in a yellow scale, the different land uses for each site (A-C) as declared in the national geographic information system database. On the right margin, color references are provided along with the corresponding land use types

Full size image

Bumble Bee Sampling

Bees were collected from December (2019) to March (2020); however, for downstream DNA sequencing, only bees sampled at the end of January were used. During this period, five collection campaigns were performed on the three chosen sites. During each sampling day, three transects of 70 m were delimited in each sampling site and bumble bees were collected in a time interval of 30 min per transect, during sunny days between 9 a.m. and 1 p.m. For this study, a total of ten female bumble bee workers per site were sampled directly from flowers through a homemade bee vacuum collector. Collected bees were placed individually in plastic vials at –20 °C until further analysis. Bee identification was carried out under a stereo microscope with 40 × magnification.

DNA Extraction from Bumble Bee Gut, Pathogen Detection, and Real-Time PCR Analysis

Total genomic DNA was individually extracted from bumble bee guts using the High Pure PCR Template Preparation kit (Roche Diagnostics). The amount of extracted DNA is reported in Supplementary table S1. A PCR screening for bee pathogens Nosema ceranae, Nosema apis, Nosema bombi, Crithidia bombi, Lotmaria passim, Apicystis bombi, and Apis mellifera Filamentous Virus (AmFV) was carried out with specific primers reported in Supplementary Table 2, on all the extracted DNA samples. Only the detected pathogens were then quantified with StepOne™ Real-Time PCR System (Applied Biosystems) relying on the same primers reported in Supplementary Table 2, according to Fernandez de Landa et al. [27] and [40], with a minor modification: PowerUp™ SYBR™ Green Master Mix (Applied Biosystems, Life Technologies) was used as universal master mix, avoiding the initial step of Uracil-DNA glycosylases activation at 50 °C. DNA samples were diluted 1:10 prior to loading in the reaction plate. qPCR was used also to determine the total amount of bacteria (Eubacteria), according to [12]. The target amplicons melting temperature, annealing temperature and amplicon size are reported in Supplementary table S2, along with the primer pair sequences.

NGS Analysis for the Gut Microbiome and Bioinformatics

The microbiome analysis on sampled bumble bees was performed with an NGS approach on an Illumina MiSeq sequencer and was based on the 16S rRNA gene, regions V5–V7. Libraries were prepared according to Alberoni et al. [8] with some modifications: briefly, KAPA Hi-Fi PCR Master Mix (Roche diagnostics) was used to amplify the target DNA with a maximum of 25 PCR cycles. The primers used are reported in Supplementary table S2 and allow the discrimination of the 16S rRNA gene amplicons of gut bacteria (about 470 bp) from amplicons deriving from pollen plastids (about 720 bp) according to Hanshew et al. [38]. The obtained PCR products were purified using magnetic beads (AMPure kit, Beckman Coulter) and E-Gel Size Select II 2% gel (Cat. Number G661012, ThermoFisher, Milan, Italy) according to Fernandez de Landa et al. [27].

Raw reads were analyzed with Qiime2 [15]. DADA2 [19] plugin was used for reads joining, denoising, and chimera check. With qiime feature-table filter-samples, samples with less than 20,000 reads had been removed. Reference reads were assigned using qiime feature-classifier classify-sklearn using full-length sequence classifier of Silva138 Database [57]. Bar plot and microbiome data at different taxonomic levels were obtained using qiime taxa barplot plugin. The tree was generated using qiime phylogeny align-to-tree-mafft-fasttree to use the rooted tree for alpha and beta diversity analysis. qiime diversity alpha-rarefaction script was used to obtain rarefaction curves.

Statistics

Linear models (LMs) have been used to compare the pathogens loads within species and the different sites. Since B. bellicosus was present only in the Paititi natural reserve, this bumble bee was excluded from the analysis. A model was fitted in which the response variables were the counts N. ceranae. The sample site was included as an explanatory variable to compare the effect of land use on gut microbiota composition of B. pauloensis; univariate and multivariate analyses were used. To perform univariate analysis, each genus of the gut microbiota, with at least 1% relative abundance, was tested for normality (Shapiro–Wilk test) and homogeneity of variance (Levene test). Depending on the assumptions of each genus, univariate comparisons were performed using ANOVA or Kruskal–Wallis followed by the Tukey test or Dunn test, respectively. Permutational multivariate analyses of variance (PERMANOVA) were performed based on the Bray–Curtis dissimilarity index applied to fourth-root transformed data (to reduce the weight of the most abundant genus) and 9999 permutations. All variables included in the PERMANOVA analysis were assessed for homogeneity assumption using the betadisper function from the vegan package [10]. A nonmetric multidimensional scaling (nMDS) based on Bray–Curtis distances and 20 minimum and 200 maximum random starts were also performed to corroborate the obtained results in PERMANOVA analysis [8]. These analyses were performed using “vegan” R package version 4.2.1. To analyze the relation between the pathogen N. ceranae and the gut microbiota, a Spearman correlation between the pathogen and each microbial genus encountered in the microbiome was calculated. Finally, PCA analysis was performed using packages FactoMineR and factoextra, taking into consideration 21 taxa at the species and/or genus level. All tests were two-tailed with a significance level of p ≤ 0.05.

The statistical analysis of alpha and beta diversity NGS data was carried out with QIIME2 using vectors and matrixes of script “qiime diversity core-metrics-phylogenetic.” Alpha-index data were analyzed using R to check normality and variance distribution. Faith_PD and observed_features indexes being not normally distributed were analyzed using “qiime diversity alpha-group-significance” plugin. Evenessindex, with a normal distribution of data, was analyzed using R software with ANOVA analysis and posthoc test using lsmeans function. For beta diversity analysis, “Qiime diversity beta-phylogenetic” and “qiime diversity beta-group-significance” plugins were used (considering Generalized Unifrac and Bray–Curtis dissimilarity indexes).

The p-value considered for statistical analysis was adjusted using Bonferroni correction for three comparisons when comparing sampling sites (reference p-value 0.0167) and for six comparisons when comparing the four different groups of bumble bees collected in the different sites (reference p-value 0.0083). The p-values listed in the “Results” section are already corrected. Moreover, the degree of freedom (df) was calculated according to parametric and non-parametric statistics. On the contrary, with parametric statistics, the degree of freedom was calculated as the total number of samples, − 1 per experimental group.

DAA analysis to compare B. bellicosus and B. pauloensis in the NRP site was performed by filtering the otu table and using the plugins qiime composition ancombc and qiime composition da-barplot.

Results

For the differential abundance analysis of Bombus bellicosus vs Bombus pauloensis, a total of 346 individuals of wild social bees were collected in this study. Among them, two Bombus species were identified: Bombus pauloensis (n = 323) and Bombus bellicosus (n = 23). While Bombus pauloensis was widely distributed across all study sites, allowing the collection of at least ten samples per site (analyzed bees, n = 30), B. bellicosus was only observed and sampled in the NRP site (analyzed bees, n = 23). Comparisons regarding bumble bee species and parasite load were performed only for individuals sampled at NRP since it was the only location where B. bellicosus were sampled. On the other hand, comparisons for the sites and the parasite load were performed only for B. pauloensis samples, since it was the only species present in all the study sites.

The Identified Pathogens in the Analyzed Bumble Bees and Their Spatial Distribution in the Sampling Sites

All the collected wild bumble bees were uninfected by A. bombi, Ascosphaera spp., L. passim, N. apis, and AmFV when analyzed in PCR. However, the samples resulted positive for N. bombi and C. bombi but the amplicon melting temperature did not correspond with the expeted one, therefore the species identification can not be considered reliable. Quantitative PCR showed an N. ceranae prevalence of 82.5% (results are available in Supplementary Table S1). The N. ceranae units (NcU) were statistically different within the two species of Bombus: Log 4.76 (± 1.23) and 4.02 (± 0.50) N. ceranae units (NcU) for B. pauloensis and B. bellicosus, respectively (ANOVA, p < 0.05, df: 1, Fig. 2A). No statistically significant differences were highlighted comparing the different sites (Fig. 2B).

Fig. 2
figure 2

Box plot reporting the intensity of N. ceranae related to host species and sampling sites. A Comparison of N. ceranae units (NcU) with respect to the B. bellicosus (n = 10) and B. pauloensis (n = 10) sampled in NRP or B the environment. (*) Indicate statistically significant comparisons (p < 0.05, df: 30)

Full size image

Alfa and Beta Diversity Analysis

A total of 3,627,070 paired sequences were obtained from next-generation sequencing (NGS). Out of 40 sequenced samples, 32 had enough sequences allowing the downstream bioinformatic analysis: 8 samples of B. bellicosus and 24 samples of B. pauloensis. The obtained ASVs allowed a taxonomical assignation of bacterial taxa up to the genus level. The NGS results obtained for the analyzed bumble bees are summarized in Supplementary Table S3 at the phylum level, Supplementary Table S4 at the family level, and Supplementary Table S5 at the genus level. Only those bacterial genera with a relative abundance higher than 1% were considered for the analysis (according to [8] and [27]). Thus, the downstream analysis focused on 22 bacterial genera, while the less abundant taxa were grouped and categorized as “others.”

The alpha diversity analysis with Faith_PD, Eveness, and Observed_Features indexes underlined significant differences among bumble bee species and sampling sites (Fig. 3A–D; Fig. S2A-B).

Fig. 3
figure 3

α-Diversity indexes. Comparing B. pauloensis of the three sampling sites (n = 24) and B. bellicosus of NRP (n = 8) with A Eveness and B Faith_PD. Comparing the different sampling sites Vivero Antoniucci (VA) vs Reserva Natural Paititi (NRP) vs Santa Paula (SP) with C Eveness and D Faith_PD. p-values: non-parametric tests, degrees of freedom typically relate to the number of data points involved in the analysis. For instance, in the Wilcoxon signed-rank test, which assesses differences between paired samples, the degrees of freedom are determined by the number of pairs minus one; *p < 0.05, ***p < 0.01

Full size image

Considering the two Bombus species in NRP, only the Evenness index showed a significant difference (p < 0.05, df: 1). Also comparing B. pauloensis in the three sampling sites, only the Evenness index turned out to be significant (p < 0.01, df: 2, Fig. 3C). Since the difference between the sampling sites for B. pauloensis is not significant for Faith_PD and Observed Features, for these indexes, we can take into consideration the entire dataset (B. pauloensis and B. bellicosus) to compare the two species. Considering the entire dataset, the two species differ in the phylogenetic composition of the microbiota (Faith-PD, p < 0.01; df: 1, Fig. 3B) and for the number of observed features (p < 0.01; df: 1, Fig. S2A).

The beta diversity analysis with Generalized Unifrac and Bray–Curtis dissimilarity indexes gave also significant results. PERMANOVA analysis on the Bray–Curtis dissimilarity index resulted in significant results considering both species and sites (Fig. S3A-S3B). When the sampling sites are considered, the pairwise comparison indicated that SP was significantly different vs NRP and VA (p < 0.05, df: 1). Considering the four different groups of bumble bees analyzed (B. pauloensis in three different sites and B. bellicosus in one site), the Bray–Curtis dissimilarity index evidenced a significant divergence comparing BellicosusNRP vs PauloensisSP groups (p < 0.01, df: 1), underlining a compositional dissimilarity between the two microbial communities. Moreover, Generalized Unifrac analysis shows a significant difference comparing PauloensisSP to PauloensisNRP and PaulonesisVA (p < 0.05, df: 1), suggesting an impact of the anthropic environment on the bumble bee microbiome) (Fig. S4A-D, and Supplementary Table S6).

B. pauloensis Gut Microbiome

NGS sequences showed that the gut microbiome of B. pauloensis is mainly composed of four different phyla, namely Pseudomonadota (relative abundance 69.83%), Bacillota (21.38%), Actinomycetota (5.17%), and Bacteroidota (3.47%) (Fig. S5A). At the family level, Neisseriaceae (42.31%) and Lactobacillaceae (20.86%) resulted the most abundant, followed by Pseudomonadaceae (7.31%), Orbaceae (7.00%), Bifidobacteriaceae (4.84%), Weeksellaceae (3.43%), Morganellaceae (3.21%), Enterobacteriaceae (2.66%), Anaplasmataceae (1.82%), and Pectobacteriaceae (1.27%) (Fig. S5B). Finally, at the genus level, the gut microbiome of B. pauloensis is mainly composed of bacteria of the genera Snodgrassella (42.30%) and Lactobacillus (20.86%). With a lower relative abundance, the genera Bifidobacterium (4.82%), Pseudomonas (7.13%), Wolbachia (1.82%), and Gilliamella (3.32%) were detected (Fig. 4A–C). Apibacter (3.34%) and Arsenophonus (3.21%) were also present, although their prevalence in B. pauloensis was sporadic (7 out of 32 individuals for both Apibacter and Arsenophonus).

Fig. 4
figure 4

Relative abundance of the gut microbiota of B. pauloensis and B. bellicosus at a genus level. (A) Microbiome profiles for B. pauloensis in SP; (B) microbiome profiles for B. pauloensis in VA; (C) microbiome profiles for B. pauloensis in NRP; and (D) microbiome profiles for B. bellicosus in NRP

Full size image

B. bellicosus Gut Microbiome

The bacterial community of the sampled individuals of B. bellicosus was composed of the same four phyla detected in B. pauloensis, although their relative abundance was different: Pseudomonadota were the most abundant (83.49%), followed by Actinomycetota (5.88%), Bacteroidota (5.4%), and Bacillota (4.22%) (Fig. S6A). At the family level, as observed in the microbiota of B. pauloensis, the family with the highest relative abundance was Neisseriaceae (45.90%), followed by Pseudomonadaceae (14.70%), Erwiniaceae (8.33%), Orbaceae (7.81%), Weeksellaceae (5.38%), Bifidobacteriaceae (4.91%), Lactobacillaceae (3.57%), Morganellaceae (1.02%), and Rhodocyclaceae (1.05%) (Fig. S6B). At the genus level, Snodgrassella was the most abundant (45.88%), but unlike B. pauloensis, Pseudomonas (14.70%) was the second most abundant genus, followed by the genera Gilliamella (5.35%), Bifidobacterium (4.88%), Rosenbergiella (2.29%), Apibacter (5.34%), Lactobacillus (3.56%), Pantoea (3.77%), and Wolbachia (0.43%) (Fig. 4D). Although Lactobacillus presented low relative abundance values, it was the only bacterial genus found in 100% of the analyzed samples of B. bellicosus. On the other hand, Snodgrassella, Gilliamella, and Pseudomonas were detected in 88.90% of the samples analyzed.

Comparison of Microbiomes Between B. pauloensis, B. bellicosus, and Land Use

The comparison among B. pauloensis and B. bellicosus bacterial genera showed significant differences in Lactobacillus that decreased from 20.68% in B. pauloensis to 3.56% in B. bellicosus (p < 0.05, df: 1), respectively. Interestingly, genera Arsenophonus and Sodalis showed a higher prevalence in B. pauloensis while they were completely absent from B. bellicosus (p < 0.05, df: 1). ASVs of the genus Kluyvera were detected only in one B. pauloensis (sample BP23 with 19.27% of relative abundance). On the contrary, the genera Rosenbergiella and Pantoea were detected only in B. bellicosus (p < 0.05, df: 1) with 2.29% and 3.77% of relative abundance, respectively. Therefore, these two genera (Rosenbergiella and Pantoea) are suggested as specific core members of B. bellicosus gut microbiome.

The comparison of the gut microbiome composition of B. pauloensis sampled in the different sites showed significant differences only for minor genera Brevudimonas, Sodalis, and Methyloversatilis when NRP is compared to SP and VA (p < 0.05, df: 1). All the other detected genera showed no statistical difference between the study sites. The DAA analysis (Fig. S7) considering B. bellicosus and B. pauloensis in natural environment (NRP) revealed a depletion of 2 LFC (Log fold change) for the genus Arsenophonus in B. bellicosus and an increase of more than 3 LFC for the genera Microbacterium, Geothermobacter, Geobacter, Methyloversatilis, Comamonadaceae, Clostridium, Propinivibrio, and Dechloromonas.

Nonmetric multidimensional scaling (NMDS) showed a nonmetric fit of r2 = 0.978, as the confidence level on the iteration stress test. The microbiome of both wild bumble bees shows statistically significant differences (PERMANOVA, F = 2.37, r2 = 0.07, and p < 0.05, df: 1) (Fig. 5A). However, the same NMDS analysis demonstrated that the intestinal microbial assemblage of bumble bees did not vary significantly in the different sites (PERMANOVA, F = 1.02, r2 = 0.23, and p > 0.05, df: 1, Fig. 5B) as shown by the overlapping of the confidence ellipses corresponding to the study sites. This finding suggests that the different land uses does not significantly influence the overall composition of the bumble bee intestinal microbiome taking into consideration the two studied species grouped (B. pauloensis of NRP–SP–VA compared to B. bellicosus of NRP).

Fig. 5
figure 5

Variation in the gut microbiota assemblage of B. pauloensis and B. bellicosus. Results of the PERMANOVA analysis for the variation of the gut bacterial assemblage with respect to the A different species studied and B the different land uses

Full size image

The Gut Pathogens Contribute to the Shaping of the Gut Microbiome of B. pauloensis

The microbial taxa detected in the B. pauloensis gut revealed different interactions and correlation confidence levels with N. ceranae, reporting negative significant correlations between Gilliamella (Spearman’s correlation value =  − 0.33), Spiroplasma (Spearman’s correlation value =  − 0.35), Pseudomonas (Spearman’s correlation value =  − 0.43), Wolbachia (Spearman’s correlation value =  − 0.36), and Lactobacillus (Spearman’s correlation value =  − 0.37) (Fig. 6). At the same time, genera Spiroplasma and Rosenbergiella showed a negative correlation with N. bombi (Spearman’s correlation values =  − 0.68 for both genera). Conversely, N. bombi is positively correlated with the genus Methyloversatilis (Spearman’s correlation values = 0.36), while negative correlations were highlighted with the genera Rosenbergiella and Spiroplasma (p > 0.05, df: 1). Like Methyloversatilis, the genus Apibacter exhibited positive and significant correlation with a parasite, in this case C. bombi (Spearman’s correlation value =  − 0.36). Differently to previous studies [48, 56], other gut microbial taxa did not show any significant correlation with C. bombi.

Fig. 6
figure 6

Pathogens and intestinal bacteria correlation. The dot plot shows the correlations between the N. ceranae units (expressed as the logarithm of the absolute quantification of both spores and vegetative forms) and the relative abundances of the significant bacterial genera studied that showed statistical significance (p ≤ 0.05)

Full size image

Discussion

The present work focuses on the structural composition of the microbiome of two bumble bee species (B. pauloensis and B. bellicosus (Fig. S7 and Fig. S8)) widely distributed in South America. This work was carried out in the Pampas region, where a strong anthropic pressure affected the entire ecosystem, destroying the nesting sites and trophic resources available for wild pollinators. Recent works showed how the microbiome of soil, plants, and pollinators are strictly interconnected [51] and dependent on environmental stressors and human activities [13, 26]. Therefore, our research work aimed at correlating the native bumble bee’s microbiome with the land use and gut pathogens load that may lead to dysbiosis [6, 50]. Of particular relevance is the focus on the gut microbiome of B. pauloensis, a bumble bee relevant for commercial pollination purposes. This species is considered the natural and environment-adapted substitute of the European bumble bee species Bombus terrestris (exogenous of South America), and its boosting is necessary to preserve the neotropical region from exogenous invaders. Moreover, our study describes for the first time the gut microbiome composition of B. bellicosus, a vulnerable species [46] and extinct in some South American regions such as Paranà (The IUCN Red List of Threatened Species).

Previous investigations on bumble bee microbiome were mainly focused on European and North American bumble bee species, such as Bombus bimaculatus, B. griseocollis, Bombus impatiens [21], and B. terrestris [16]. These bumble bee species showed a microbiome usually composed of five core microbial taxa: Bifidobacterium, Gilliamella, Lactobacillus, Schmidhempelia, and Snodgrassella [14, 35,36,37]. The gut microbiome of B. pauloensis and B. bellicosus is similar to the microbiome described in other bumble bee species. However, B. pauloensis and B. bellicosus microbiomes showed differences when compared with each other, in both abundance and diversity of core microbial taxa. The most abundant genus in both species was Snodgrassella which had a relative abundance higher than 40%. The second predominant taxon was Lactobacillus in B. pauloensis, whereas Pseudomonas was the second predominant in B. bellicosus. In honeybees, lactobacilli (Firm 4 and Firm 5) are known to produce short-chain fatty acids [73] useful for pollinators as a source of energy, behavioral modulation [70], and chitinases [9]. These lactobacilli are likely closely related to the lactobacilli found in the analyzed bumble bees. However, the low Lactobacillus abundance in B. bellicosus was unexpected. This microbial taxon was probably replaced by Pseudomonas, a controversial opportunistic taxon when detected in the insect gut microbiome [23]. The genus Pseudomonas was recently found to populate the gut microbiome of solitary bees in the same Neotropical region [27, 41], in honeybees populating semi-desertic environment [30], or in bees treated with antibiotics [12]. However, this taxon in bumble bees was recently related to microbiome disruption [66] and therefore to compromised bee health [6]. Since B. bellicosus is an endangered species, our finding seems to confirm the high susceptibility of this pollinator to adverse environmental factors.

Another indicator of the health status of the sampled B. bellicosus is represented by the relative abundance of opportunistic environmental bacteria (here defined as “Other_taxa”), or non-core members which are highly abundant in this bumble bee species (above 10% in B. bellicosus vs below 4% in B. pauloensis). Among the opportunistic bacteria, the order Enterobacteriales was found to be the most abundant. Along with Pseudomonadaceae, also Enterobacteriaceae and Yersiniaceae have been previously correlated with environmental stressors [59], disrupted microbiome [66], or septicemia [17]. Our analysis detected Enterobacteriaceae in nine individuals, seven of them belonging to the NRP environment and two to the VA, often co-occurring with Pseudomonas. Among opportunistic taxa, Erwiniaceae (Pantoea and ZDC) were preferably populating B. bellicosus, while Sodalis (Pectobacteriaceae) showed a higher prevalence in B. pauloensis. Pantoea was recently described as a low-occurring taxa in B. terrestris, and its presence seems to be shaped by the local trophic resources [68]. However, currently, there is a knowledge gap on the role of Rosenbergiella and Sodalis in the bumble bee gut. The incidence of these environmental bacteria is probably indicating interactions between these bumble bees with other pollinators through flowers [42].

Comparing the gut microbiome of B. pauloensis and B. bellicosus with the microbial profile of other American bumble bee species such as Bombus hortulanus and Bombus rubicundus [66], the only difference detected was the absence of the genus Schmidhempelia. However, it is known that the relative abundance of Schmidhempelia drastically decreases during aging [36, 37]. Considering that bumble bees sampled in our study were all foragers, the absence of Schmidhempelia might be associated with aging and therefore coherent with previous findings [36, 37, 68]. The differences between the gut microbiomes of B. pauloensis and B. bellicosus might be influenced by pathogen load. N. ceranae was present with a high prevalence and intensity in the analyzed bumble bees, as already described by Arbulo et al. [11] and Plischuk et al. [55]. Authors such as Ngor et al. [49] and Gisder et al. [33] have demonstrated that even when exposed to doses previously considered optimal for the development of parasitosis (from 6500 to above 100,000 spores), individuals of different bumble bee species from the northern hemisphere (Bombus terrestris and Bombus impatiens) did not exhibit symptoms of infection. Therefore, the detection of DNA in the individuals collected in this study would not be sufficient to indicate an impacting or lethal infection.

In our study, the correlation between the studied pathogen and the gut microbiota of B. pauloensis and B. bellicosus was attempted. The abundance of the genus Gilliamella was found to be negatively correlated with N. ceranae, in contrast with the findings described by Rubanov et al. [60] and Sbaghdi et al. [61], who identified various strains of Gilliamella that were positively correlated with high infections of N. ceranae. However, our findings are consistent with previous studies of Kwong et al. [45] and Zheng et al. [72], which proposed the gut acidification by Gilliamella as a barrier against gut pathogens. Studies correlating the genus Gilliamella with various pathogens are recent, and clear trends or precise explanations for these correlations are not yet established. Therefore, future research is necessary to precisely elucidate the role of Gilliamella in the gut of pollinators and its interaction with parasites. This understanding is crucial for developing comprehensive strategies to support current methodologies aimed at enhancing the immune systems of both honeybees and non-Apis pollinators. Regarding the interaction between N. ceranae and the genus Lactobacillus, recent literature supports the results evidenced here. On one hand, Tejerina et al. [63] observed that the supplementation of A. mellifera colonies located in northern Argentina with preparations of the species Lactobacillus salivarius contributed to the control of both N. ceranae and the ectoparasite Varroa destructor. On the other hand, Yu et al. [69] demonstrated how the supplementation of B. terrestris individuals with Lactobacillus melliventris promoted improved immune systems and reduced parasitosis rates. However, the results obtained here do not explain causality but rather correlation. Future studies should be conducted to investigate and delve deeper into the interaction between species of the genus Lactobacillus and a widely distributed pathogenic agent such as N. ceranae. Finally, to the best of our knowledge, this research represents the first evidence of a negative correlation between Spiroplasma, Wolbachia, and Pseudomonas and the parasitic intensities of N. ceranae. As mentioned previously, the interactions between the bacterial taxa and the microsporidian N. ceranae discussed here do not demonstrate a causal relationship. Therefore, the development of future studies to determine whether these bacterial groups can indeed contribute to the regulation of Nosemosis development in Bombus bees is of great interest. This could lead to the development of new, comprehensive, and holistic strategies for parasite control using biological-based technologies.

Our study provides a first insight into the structural composition of the core microbiome of two widely distributed bumble bee species in South America, B. pauloensis and B. bellicosus. The characterization of the gut microbiome of B. bellicosus, an endangered species, raises the importance of profiling the core gut microbial community and the understanding of microbial dynamics involved in a threatened species. Alpha and beta diversity indexes supported the hypothesis of the possible influence of anthropized environments being the species and site variables significant when comparing B. bellicosus vs B. pauloensis and B. pauloensis in NRP vs B. pauloensis in SP. Bumble bees collected from each sampling site necessarily belong to different colonies due to the distances between the sites (greater than the foraging radius of bumble bees). However, this factor has a minor impact if compared with anthropogenic factors on the changes and variability within bacterial communities. In fact, the microbiome of social corbiculate bees tends to remain constant between populations, despite geographical distance [5, 44]. A limitation of this research is represented by a lack of replicates considering sampling sites (especially anthropized ones). Therefore, future research efforts should focus on elucidating the intricate interactions between bumble bee microbiota, pathogens, and environmental factors at a higher scale, differentiating land use, the effect of distance, and dispersal limitation of microorganisms.