Comparing the pollen-provision microbial communities of closely related specialist and generalist bees will help address how diet breadth affects microbe exposure and establishment. To help understand how diet breadth affects bee nest microbial communities, we characterized the pollen provision microbial communities of a pair of closely related bees collected across four sites: Osmia lignaria, a polylectic bee that prefers to forage on orchard trees in the family Roseaceae; and O. ribifloris, a specialist on blueberries and its relatives with Berberis serving as an alternate host. We address the hypothesis that, due to the importance of floral transmission for microbes found in the pollen provisions of megachilid bees, diet breadth affects pollen provision microbial composition. We predicted that the specialist bee would harbor fewer species of microbes and show less variation across sites compared to the generalist bee.We collected pollen provisions from the nest cells of two bee species native to the area of Logan, Utah, USA: Osmia lignaria propinqua and Osmia ribifloris biedermannii in Spring 2015. We chose four locations near Logan that varied in floral composition and interaction with other bee species: Site 1 was a cultivated floral garden located in a suburban area with a combination of native and introduced floral species, plastic planters wholesale including Rubus idaeus L. and Calluna vulgaris .
Site 2 was a bee-tight screenhouse that contained older, in-ground plantings of R. idaeus and Berberis fremontii , along with potted C. vulgaris. Sites 3 and 4 were natural areas of Logan Canyon that contained limestone cliffs and a mixed forest, and were approximately 1836 m apart, which is likely outside of the flight range of Osmia species. We note that the locations differed in the composition of their pollinator communities: Site 1 had a diverse pollinator community present, including numerous honey bee and bumble bee colonies. Site 2 was a research screenhouse that had no honey bees nor bumble bees present for at least five preceding years. Sites 3 and 4 had diverse pollinator communities without honey bees present, and we did not collect any O. ribifloris from Site 4. See supplemental Table S1 for the number of each bee species collected at each site. There were also differences in the sources of sampled Osmia nests between locations: We sampled natural Osmia populations from Sites 1, 3, and 4, while we sampled nests made by commercially acquired bees for Site 2. In this location, O. lignaria were sourced from Watts Solitary Bees and O. ribifloris were purchased from NativeBees.com . Within each location, we sampled the pollen provisions from cavities within wooden nesting blocks that were lined with paper straws and placed in each location for one week. As both species of bee will nest in these cavities at the same time, we were able to concurrently sample newly made nests. We used 1–6 nesting straws per bee species, then removed the straws from the nest blocks, numbered them sequentially, and X-rayed the straws to confirm that eggs were present and had not yet hatched.
Once the straws were collected, we carefully slit open the straws, removed the pollen provisions from each initially formed cells , then sterilely removed the eggs from nest cells before DNA extraction. We carefully excluded cell wall partitions from the collection of the pollen provision. Each cell was treated as a separate sample.We extracted DNA from the pollen provisions based on a modified protocol from Engel et al. 2013, Rothman et al., and Pennington et al. 2018. We used a Qiagen DNeasy Blood and Tissue kit for the DNA extractions, with slight modifications. We sterilely transferred entire pollen provisions to a 96-well tissue lysis plate , then added 100 µL glass beads, one 3.2 mm steel-chrome bead, and 180 µL buffer ATL to eachwell. We bead-beat the mixture at 30 Hz for three minutes, turned the plates over, then bead-beat for three more minutes. We incubated the mixture at 50 ◦C overnight, then followed the rest of the manufacturer’s protocol to finish the DNA extraction. We used the extracted DNA to prepare 16S rRNA gene libraries as in McFrederick and Rehan 2016, Rothman et al. 2019, and Pennington et al. 2017. Briefly, we used a dual-indexing approach to build an amplicon construct consisting of the universal primers 799F-mod3 and 1115R, a unique 8-mer barcode, and the Illumina adapter sequence as in Hanshew et al. 2013. We built the libraries in two rounds of PCR amplification: First, we used 4 µL of template DNA, 0.5 µL of 10 µM forward and reverse barcoded primers, 10 µL water, and 10 µL of 2× Pfusion DNA polymerase with an annealing temperature of 52 ◦C for 30 cycles. Next, we cleaned the PCR product with a MoBio UltraClean PCR cleanup kit . We then used the cleaned amplicons as template for another PCR reaction, with the following conditions: 1 µL template DNA, 0.5 µL of 10 µM primers PCR2F and PCR2R, 13 µL water, and 10 µL 2× Pfusion DNA polymerase with an annealing temperature of 58 ◦C for 15 cycles.
We cleaned and normalized the libraries with a SequalPrep Normalization kit , pooled 5 µL of each library, then cleaned and concentrated the libraries with a MoBio UltraClean PCR cleanup kit . Lastly, we used an Illumina MiSeq to sequence the libraries at 2 × 300 cycles in the UC Riverside Genomics Core Facility.Other studies have found that bee species identity, foraging patterns, and geography can affect the microbes found in pollen provisions. McFrederick and Rehan showed that the pollen provision microbiome of Ceratina calcarata co-varies with pollen usage across habitats, especially when considering fungi. Voulgari-Kokota et al. showed that bee species and foraging patterns drive the bacterial communities found in pollen provisions. Our findings add to the consensus of these other studies that floral transmission and pollen usage influence the composition of the pollen provision microbiome. Our study also extends these previous studies by adding an understanding of how diet breadth does and does not affect the pollen provision microbial community in these two species of Osmia. Whether the different bacterial variants found in the host species and sites across studies is due solely from different transmission networks, filtering of microbes via differential floral and pollen provision chemistry, or a combination of both requires further study. As in other bee species’ pollen provisions, Osmia spp. pollen provisions contain a wide diversity of microbes. In agreement with these previous studies, the bacteria identified here can also be found in the environment, in both flowers and soil. For example, four of the 18 ASVs that were found at all sites in our study belonged to the genus Acinetobacter. This genus is ubiquitous in the environment and has been found in association with plants and animals, in floral nectar, in sewage, and in water and soil. The small fragment of the 16S rRNA gene that we use for Illumina sequencing rarely allows species-level resolution of these bacteria, so we are unable to determine whether these taxa were sourced from the mud that the bees used to partition their brood cells or from the nectar they mixed into their pollen provisions. Acinetobacter, however, has been reported in association with pollen provisions of several different bee species, and it is therefore not surprising that it occurs at all study sites. One conspicuous group of bacteria that are found in many pollen provision microbiomes but were uncommon here are bacteria from the Apilactobacillus micheneri clade. These bacteria have been found in the pollen provisions of other megachilid species in North America and Europe, but it is becoming clear that they are unevenly distributed across species. For example, in Germany, Apilactobacillus spp. are abundant in the pollen provisions of Megachile spp., plastic plant pot at variable abundances in O. caerulescens provisions, at low abundance in Heriades truncorum provisions, and absent in O. bicornis and O. leaiana provisions. In Texas , we detected these same lactobacilli at high relative abundances in pollen provisions of Osmia chalybea, Osmia subfasciata and Megachile policaris. Here, we report that lactobacilli are present only at very low relative abundances in the pollen provisions of O. lignaria and O. ribifloris, and that the A. micheneri clade lactobacilli are absent. As these bacteria have been isolated from flowers in both Texas and California, we hypothesize that either foraging preferences or pollen provision chemistry drives the presence or absence of these lactobacilli, and understanding the apparently cosmopolitan phenomenon of uneven distribution of lactobacilli across wild and solitary bee species should be a priority for pollen provision microbial community studies.
Many of the bacteria that we identified in O. lignaria pollen provisions have also been found in association with O. lignaria adults. Cohen et al. found that adults that had been foraging in the environment had a different or more variable microbiome compared to bees that emerged under sterile conditions in the lab, again supporting the importance of environmental transmission for the wild and solitary bee microbiome. Many of the bacteria found in adult O. lignaria adults are found in the environment and in the pollen provisions that we studied here. For example, Massilia is a root-colonizing soil bacterium that has been found in O. bicornis nests. This bacterium may be found in adult O. lignaria and their pollen provisions due to the adult’s habit of collecting mud to build partitions between brood cells in their nests. Pantoea is a plant-associated microbe that is abundant in the environment but has also been reported in association with other solitary bees and honey bees. Surprisingly, Cohen et al. found lactobacilli associated with adult O. lignaria, and additionally found that the abundance of flowers at a site positively correlated with the relative abundance of lactobacilli associated with adult O. lignaria. Future studies examining the adult and pollen provision microbiome of O. lignaria will help unravel how the environment shapes the microbiome of these separate but connected niches. The sole bacterial family that we found to be differentially abundant by host species was Micrococcacea. Micrococcacea is a diverse family that occurs in the environment, includes a commensal but opportunistic human pathogen, and has been classified as a pathogen honey bees. As Micrococcacea represented around 4% of the reads in seemingly healthy O. ribofloris nests, it is unlikely that this bacterium is pathogenic to Osmia. The differential abundance between the two Osmia species may mean that it is somehow important for bee health, but may also be due to the differential abundance between sampling site too. Along this thought, we found that many bacterial families were differentially abundant between geographical location. This provides further support to the hypothesis of environmental and floral transmission for solitary bee microbes as has been shown in previous studies.Insects pollinate many plant species, including several major crops. Bees are the single most important insect pollinator group and can be a limiting factor for the success of plant reproduction. Consequently, there is strong inter- and intra-specific competition among plants for the attention of pollinators. With respect to insect-pollinated crops, pollinator visitation is required to obtain maximal seed and fruit production. Consequently, pollination facilitates higher yields even when a crop plant is self-compatible. Bumblebees are important pollinators of tomato and other Solanum species that utilize an unusual pollination system called ‘buzz-pollination’. Buzz-pollinated flowers provide excess pollen as a reward to foraging bumblebees that feed it to their developing larvae. Although domesticated tomato is to a large extent ‘self-fertilizing’, buzz-pollination by bumblebees or by manual application of mechanical vibration ‘wands’ is required for maximal seed production, which in turn promotes increased fruit yield . Cucumber mosaic virus , one of the major viral pathogens of tomato, is a positive sense RNA virus that encodes five proteins including the 2b protein, which is a viral suppressor of RNA silencing. Bees do not transmit CMV but the virus is vectored by several aphid species. Virus infection causes dramatic changes in plant host metabolism . CMV-induced metabolic changes include qualitative and quantitative alterations in the emission of volatile compounds and in certain host species this makes infected hosts more attractive to aphid vectors.It is not known if the virus-induced alterations in host volatile emission that influence aphid behavior can also affect plant-pollinator interactions. Most bee-plant interaction studies have focussed on the effects of visual cues. Therefore, the influences of floral and non-floral volatiles on bee-mediated pollination are less well understood.