Discussion
Our results demonstrate that the Xylocopa species examined here,
although incipiently social, host a microbiome that exhibits
similarities to previously characterized social corbiculates in at least
two ways. First, Xylocopa individuals host a specific and
distinctive set of bacterial taxa found consistently in bees that span
geographic sampling locations, sexes, and individuals displaying
different behaviors (nest-caught vs. foraging), unlike most previously
characterized solitary bees (Voulgari-Kokota et al. 2019). Second, many
of the bacterial lineages detected at high abundance in the gut have
been previously described in social corbiculates (Kwong et al. 2017).
Below, we discuss the specific taxa found and the implications for the
effect of sociality on the evolution of the bee microbiome.
The bacterial taxa that we describe here are comprised of primarily
bee-associated taxa or those that associate with diverse insects,
similar to the previously characterized Xylocopa tenuiscapa(Subta et al. 2020). Xylocopa gut samples containedBombilactobacillus (previously Lactobacillus Firm4),Bombiscardovia ( Bifidobacteriaceae), and Lactobacillus
bxid5692 (similar to Lactobacillus Firm5) in nearly every
individual, often with multiple co-occurring ASVs, and typically at high
abundance, similar to previously characterized Apis andBombus species (Kwong et al. 2017). Bombilactobacillus andBombiscardovia are considered core symbionts of adult bumble bees
(Killer et al. 2010, Kwong et al. 2017, Parmentier et al. 2018) and
primarily transmitted via social contact (Billiet et al. 2017). In
addition, the full-length Bombiscardovia sequences inXylocopa are diverse and cluster with B. coagulansisolated from Bombus , or closely related to the newly describedBifidobacterium xylocopae and B. actinocoloniiforme from
European Xylocopa sp; suggesting that this bacterial group has a
larger host range than previously recognized (Kwong et al. 2017,
Alberoni et al. 2019, Hammer et al. 2021). In Apis mellifera ,Bombilactobacillus and Bifidobacterium sp. colonize the
hindgut and are implicated in saccharide breakdown and fermentation (Lee
et al. 2015). Despite similarities to the A. mellifera andBombus microbiome, the Xylocopa microbiomes characterized
here are also distinct in the consistent presence of Apibacter ,
which is found consistently in Asian honey bees (Kwong et al. 2018) but
only sporadically in Bombus. In contrast to Bombus ,Xylocopa appears to mostly lack Snodgrassella— we
detected a single ASV in low abundance. Gilliamella was
frequently detected, but differed between species: X. sonorinahosted Gilliamella more consistently and at higher abundance in
than did X. tabaniformis (Figure 2). In addition, the gut of bothXylocopa species consistently contained Entomomonas at
relatively high frequencies (Supplementary Figure 6), a bacterium
with a highly reduced genome and limited metabolic capabilities which
was previously described in a diverse group of insects including
Diptera, Coleoptera, and other Hymenoptera, including ants andApis (Wang et al. 2020). The ecological role of this genus is not
understood. In X. sonorina , the crop was highly dominated byApilactobacillus (Supplementary Figures S5a; S6a),previously documented in solitary bees and in the crop of social bees
(McFrederick et al. 2018). Apilactobacillus may dominate in the
provisions of some bee species where they have been hypothesized to
inhibit pathogen growth or prevent spoilage of stored pollen (Vásquez
and Olofsson 2009, McFrederick et al. 2018, Kapheim et al. 2021).
Since Xylocopa are incipiently social rather than classically
eusocial (Gerling et al. 1981), their distinctive microbiome raises
questions about the mechanisms required for effective social
transmission of the microbiome. Dominant Xylocopa females will
feed newly emerged nestmates via trophallaxis, and allow consumption of
the stored provision (Gerling et al. 1989, Ostwald et al. 2021, Vickruck
and Richards 2021). In this way, Xylocopa exhibit behavioral
similarities to Apis , which engage in trophallaxis, andBombus , which do not engage in trophallaxis yet feed from shared
food resources and engage in coprophagy (Näpflin and Schmid-Hempel
2016). In addition, Xylocopa individuals migrate among nests,
including those of non-kin (Ostwald et al. 2021, Vickruck and Richards
2021). We hypothesize that these behaviors, as well as the relatively
long lifespan (1-2 years), and large body size of Xylocopa could
help explain the maintenance of specialized microbial taxa. Differences
between Xylocopa species in microbiome composition and richness
also suggest areas for future study. Although we lack details onX. tabaniformis sociality, this species is likely incipiently
social like X. sonorina (Breed 1976) yet differs in breeding
systems: males patrol flowers for mating, while X. sonorina males
host non-resource based territories (Marshall and Alcock 1981). This
could explain the differences in sex-specific microbiomes between
species. Additional information on social structure or nesting biology
may inform reduced sequence variant richness and more variable crop
community in X. tabaniformis compared to X. sonorina .
In previous work, the microbiome of halictid bees that nest socially,
including those that exhibit trophallaxis and eusociality (Kapheim et
al. 2016), did not differ substantially from those that nest
non-socially (McFrederick et al. 2014, Rubin et al. 2018), and resembled
solitary bee microbiomes (Voulgari-Kokota et al. 2019), which are
primarily environmentally acquired (McFrederick et al. 2012, Kapheim et
al. 2021). The primary difference between solitary and social halictids
was the abundance of Sodalis , an insect endosymbiont (Rubin et
al. 2018). Combined, these studies suggest that not only social
behaviors but perhaps additional other biological differences may be
required for the maintenance of a distinctive bee microbiome.
Both Xylocopa species displayed microbiome differences among
collection locations, suggesting population-level differentiation in
microbiome composition. Indeed, many ASVs of core taxa were
location-specific (Supplementary Figures 7, 8, 9), while only a few
dominant ASVs (e.g. Lactobacillus bxid5692 ASV 9) was detected at
all locations in both bee species. Sampling location explained between
9-16% of variation in bacterial composition, more than previous studies
examining geographic signatures in honey bees (Ge et al. 2021) stingless
bees (Liu et al. 2021), and even some solitary Osmia (Rothman et
al. 2020). Although long-read sequences likely enable us to detect such
patterns (Supplementary Table 1), distinctive Xylocopa sociality
and patterns of microbial transmission may also contribute to geographic
structuring.
As expected, we found that restricting our analysis to the V4 region
resulted in fewer taxa detected, a loss of phylogenetic resolution
(Supplementary Figure S7), and the loss of genetic information that
could distinguish bacterial taxa between host species and among
geographic locations (Supplementary Table S4). Due to reduced sequence
length, ASVs that previously distinguished species and locations were
collapsed into a single ASV (Supplementary Figure S 9), yet species and
geographic location could still largely be distinguished using V4
sequences only (Supplementary Table S1). This comparison suggests that
first-generation sequencing may be able to detect drivers of microbial
community composition, but fail to show the extent of strain-level
differentiation that exists among populations and species. However, we
caution that our comparative approach does not account for realistic
primer bias or sequencing bias (Quail et al. 2012, Tedersoo et al. 2018)
and as a result, may overestimate the similarity of these regions and
their ecological inference. Nevertheless our data support the conclusion
that long reads enable enhanced ecological insights into the
strain-level composition and evolution of the microbiome, suggesting
that despite its greater cost, this approach may be warranted when
strain-level information may differentiate populations or closely
related species (Tedersoo et al. 2021).
Overall, our results provide evidence that the microbiomes of species
with simple social groups can have characteristics typically associated
with the more complex eusociality of the corbiculate bees. Further work
will be necessary to determine the role that specific features of
sociality, such as trophallaxis, play in shaping the microbiome in the
earliest stages of social evolution, and uncover the functional
consequences of a specialized microbiome.