The role of plant–pollinator interactions in structuring nectar microbial communities

Journal of Ecology. 2021;109:3379–3395. | 3379wileyonlinelibrary.com/journal/jec Received: 24 January 2021 | Accepted: 21 May 2021 DOI: 10.1111/1365-2745.13726 RESEARCH ARTICLE The role of plant– pollinator interactions in structuring nectar microbial communities Clara de Vega1 | Sergio Álvarez- Pérez2 | Rafael G. Albaladejo1 | Sandy- Lynn Steenhuisen3 | Marc- André Lachance4 | Steven D. Johnson5 | Carlos M. Herrera6 1Departamento de Biología Vegetal y Ecología, Facultad de Farmacia, Universidad de Sevilla, Sevilla, Spain; 2Department of Animal Health, Faculty of Veterinary Medicine, Complutense University of Madrid, Madrid, Spain; 3Afromontane Research Unit and Department of Plant Sciences, University of the Free State, Phuthaditjhaba, South Africa; 4Department of Biology, University of Western Ontario, London, Ontario, Canada; 5Centre for Functional Biodiversity, School of Life Sciences, University of KwaZulu- Natal, Pietermaritzburg, South Africa and 6Estación Biológica de Doñana, Consejo Superior de Investigaciones Científicas, Sevilla, Spain This is an open access article under the terms of the Creat ive Commo ns Attri butio n- NonCo mmerc ial- NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is noncommercial and no modifications or adaptations are made. © 2021 The Authors. Journal of Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society Correspondence Clara de Vega Email: [email protected] Funding information Fundación BBVA; Spanish Ministry of Economy and Competitiveness; Natural Science and Engineering Research Council of Canada; Spanish Ministry of Science and Innovation, Grant/Award Number: RYC2018- 023847- I; National Research Foundation Handling Editor: Ana Pineda Abstract 1. Floral nectar harbours a diverse microbiome of yeasts and bacteria that depend predominantly on animal visitors for their dispersal. Since pollinators visit specific sets of flowers and carry their own unique microbiota, we hypothesize that plant species visited by the same set of pollinators may support nonrandom nectar microbial communities linked together by the type of pollinator. 2. Here we explore the importance of plant– pollinator interactions in the assembly of nectar microbiome and study the role of plant geographic location as a determinant of microbial community composition. We intensively sampled the nectar of 282 flowers of 48 plant species with beetles, birds, longtongued and shorttongued insects as pollinators in wild populations in South Africa, one of the world's biodiversity hotspots, and using molecular techniques we identified nectar yeast and bacteria taxa. The analyses provided new insights into the richness, geographic structure and phylogenetic characterization of nectar microbiome, and compared patterns of composition of bacteria and yeast communities in relation to plant and pollinator guild. 3. Our results showed that plant– pollinator interactions played a crucial role in shaping nectar microbial communities. Plants visited by different pollinator guilds supported significantly different yeast and bacterial communities. The pollinator guild also contributed to the maintenance of beta diversity and phylogenetic microbial segregation. The results revealed different patterns for yeast and bacteria; whereas plants visited by beetles supported the highest richness and phylogenetic diversity of yeasts, bacteria communities were significantly more diverse in plants visited by other insect groups. We found no clear microbial spatial segregation at 3380 | Journal of Ecology de VeGA et Al. 1 | INTRODUCTION From the root tips to the flowers and apical meristems, plants offer diverse physicochemical below- and aboveground habitats for complex communities of microorganisms (Andrews & Harris, 2000; Nelson, 2018; Öpik et al., 2006; Vannette, 2020). Floral nectar has until recently been one of the most neglected microbial habitats. By the secretion of this solution rich in carbohydrates, flowers obtain pollinator services (Nicolson et al., 2007; Willmer, 2011). Floral nectar is also of central importance in hosting a diverse and dynamic microbiome of yeasts and bacteria. Once established, nectar microorganisms may act as ‘ecosystem engineers,’ directly altering the floral environment and consequently the relationships with their large partners. Microbial communities can reduce nectar sugar concentration (de Vega & Herrera, 2012; de Vega et al., 2009; Herrera et al., 2008) modify nectar carbohydrate and aminoacid composition (de Vega & Herrera, 2013; Herrera et al., 2008; Lenaerts et al., 2016; Vannette & Fukami, 2018), warm the flowers (Herrera & Pozo, 2010) and emit volatiles (Rering et al., 2018; Schaeffer et al., 2019) that ultimately alter pollinator foraging preferences and influence pollination success (Herrera et al., 2013; Junker et al., 2014; Schaeffer & Irwin, 2014; Schaeffer et al., 2017; Vannette & Fukami, 2016; Vannette et al., 2013; Yang et al., 2019). Nectar thus constitutes a unique aquatic microenvironment in terrestrial ecosystems that provides a link between macro and microorganisms. Flowers are exposed to a virtual sea of microorganisms that can be passively dispersed by wind, rainfall or even soil among others (Sharaby et al., 2020; Zarraonaindia et al., 2015). However, most of the evidence suggests that microbial dispersal to nectar is not a stochastic, abiotic, passive process. Microbes like yeasts and bacteria depend predominantly on animal visitors, especially insect vectors, for their dispersal and establishment in nectar (Brysch- Herzberg, 2004; de Vega & Herrera, 2012; Herrera et al., 2010; Schaeffer & Irwin, 2014; Vannette & Fukami, 2017). Incoming microorganisms undergo filtering growth processes, priority effects and microbe– microbe interactions that may further shape floral microbiota composition (Álvarez- Pérez et al., 2019; Peay et al., 2012; Tucker & Fukami, 2014). Nectar is colonized by different genera of yeasts and bacteria recovered repeatedly worldwide from this particular ecological niche (Álvarez Pérez & Herrera, 2013; Belisle et al., 2012; Brysch- Herzberg, 2004; Canto et al., 2017; Fridman et al., 2012; Jacquemyn, Lenaerts, Tyteca, et al., 2013; Pozo et al., 2011) and some general patterns have emerged. The dominance of bacterial species within the Proteobacteria, especially species of Acinetobacter, Rosenbergiella and Pseudomonas (Aizenberg- Gershtein et al., 2013; Álvarez- Pérez & Herrera, 2013; Álvarez Pérez et al., 2013; Morris et al., 2020; Sharaby et al., 2020), and yeast species within the genus Metschnikowia (Belisle et al., 2012; Brysch- Herzberg, 2004; Pozo et al., 2012; Schaeffer et al., 2015) is evident in the nectar microbiota. The repeated occurrence of the same group of species identified using either cultivationdependent or cultivationindependent methods is indicative of an adaptive strategy and equilibrium in the nectar communities. While the initial research focus was to discover which microorganisms were the main inhabitants of floral nectar, the more interesting question that has arisen subsequently is why they are found in certain floral environments and not others. Individual flowers harbour diverse microbial communities that may or may not be similar to those found on other flowers. But as pollinators visit specific sets of flowers and carry their own unique microbiota (Brysch- Herzberg, 2004; da Costa Neto & de Morais, 2020; Ganter, 2006; Koch & Schmid- Hempel, 2011; Lachance et al., 2001; Martinson et al., 2011; Rosa et al., 2003), it has been suggested that flowers and pollinators share microbes (Brysch- Herzberg, 2004; Keller et al., 2020; McFrederick et al., 2017) and that host plant species visited by the same set of pollinators may support nonrandom or even unique nectar microbial communities that are linked together by the type of pollinator (de Vega et al., 2017). However, our understanding of how plant– pollinator assembly processes might ultimately be responsible for nectar microbial community structure remains limited. Moreover, most studies that focused on estimates of the composition and abundance of microorganisms associated different geographical scales for bacteria, and only the phylogenetic similarity of yeast composition was correlated significantly with geography. 4. Synthesis. Interactions of animal vector, plant host traits and microbe physiology contribute to microbial community assemblages in nectar. Our results suggest that plants visited by the same pollinator guild have a characteristic nectar microbiota signature that may transcends the geographic region they are in. Contrasted patterns for yeast and bacteria stress the need for future work aimed at better understanding the causes and consequences of the importance of plants and pollinators in shaping nectar microbial communities in nature. KEYWORDS beetle, floral microbiome, insect, multitrophic interactions, nectar bacteria, nectar yeasts, pollination, South Africa | 3381 Journal of Ecolog y de VeGA et Al. with nectar appear to be biased toward plant species that are visited by bees and/or hummingbirds. Additionally, geographic bias is skewed towards temperate regions, with North America and Europe accounting for the majority of studies, despite preliminary evidence that tropical areas may sustain a higher nectar microbial biodiversity (Álvarez Pérez et al., 2012; Canto et al., 2017; Oliveira Santos et al., 2020). There has been little research identifying the biotic factors structuring assemblages of nectar microbial communities in tropical biodiversity hotspots with enormous functional diversity of plant– pollinator interactions. Through recent investigations in tropical regions, we are gaining a clearer perspective of the role of other floral visitors, such as beetles, in vectoring microorganisms to nectar (de Vega et al., 2012, 2017; Lachance et al., 2001, 2003). Thus, intimate symbiotic relationships of microbes with different plants and insects have been suggested, but the fundamental nature of these interactions remains an open question. In this study we focus on South Africa, containing three of the world's biodiversity hotspots— the Cape Floristic Region, the Succulent karoo and the Maputaland- Pondoland- Albany hotspot— to explore how plant– pollinator assembly processes may ultimately influence nectar microbial community structure. Since nectar microbes, plants and pollinators can be inextricably linked, we hypothesize that plants visited by the same pollinator guild possess a unique nectar microbiota signature that could transcend the geographic region that they inhabit. We analyse a diverse set of plants with birds, Hymenoptera, Lepidoptera, Diptera and Coleoptera as main pollinator groups in the Cape and KwaZulu- Natal regions. The study of similar habitats in ecologically disparate settings scenarios is particularly relevant to test the hypothesis that variation of nectar microbial communities is correlated with differences in pollinator composition. Our specific goals were: (a) to explore broadscale differences in nectar microbial communities among plant species visited by different pollinator groups, (b) to ascertain whether the resulting patterns are similar for bacterial and yeast communities and (c) to test whether geographic location is one of the main determinants of nectar microbial community composition. Our findings may have broad implications for a better understanding of the distribution of nectar inhabitants in different plant species visited by different pollinator guilds worldwide, and provide a framework for identifying areas for future research. 2 | MATERIALS AND METHODS 2.1 | Study site and study species The composition and diversity of microbial communities naturally occurring in the floral nectar of animalpollinated plants were studied in the KwaZulu- Natal province and in the Cape Region of South Africa (Figure S1). Nectar samples were collected from 282 flowers (from 282 different individual plants) of 48 plant species belonging to 16 angiosperm families (Table S1). The study was conducted in 10 ecologically diverse localities separated by 20 to 1,250 km. Localities differed in climatic conditions, plant species composition and soil type, covering sites near the coast to high altitudes in the Drakensberg mountains (up to 1,800 m a.s.l.; Table S2). Our previous investigations in the study area suggested that microbes carried by beetles can differ from those transported by other insects (de Vega et al., 2012, 2017). The major groups of floral visitors were grouped as nectarfeeding birds (mainly Nectariniidae sunbirds; henceforth BI), Coleoptera (mainly in the subfamilies Cetoniinae and Hopliinae and also the families Scarabaeidae, Chysomelidae, Melyridae, Nitidulidae, Staphylinidae and Tenebrionidae; BE) and other insects (IN), both shorttongued (ST) insects (including Hymenoptera and Diptera mainly visiting flowers with easily accessible pollen and nectar) and longtongued (LT) insects (including Hymenoptera, Lepidoptera and Diptera usually in tubular flowers; Figure 1; Table S1). Plants were classified into main floral visitor groups using a combination of literature review, floral traitbased assignment and personal observations of visitors in the field. See Table S1 for detailed information. 2.2 | Nectar sampling and microbial cultures Open flowers exposed to pollinator visitation were collected in natural populations (M ± SE nectar samples per species = 5.9 ± 0.6). The mean number of nectar samples per species is similar in other nectar microbial surveys throughout the world (Brysch- Herzberg, 2004; Canto et al., 2017; Fridman et al., 2012; Jacquemyn, Lenaerts, Tyteca, et al., 2013; Pozo et al., 2011). The main criteria used for selecting flowers were that no recently opened flowers were sampled, flower collection should be done with good weather conditions, and whenever possible flowers should have floral visitors or signs of have been visited. Flowers were carried immediately to the laboratory in a cooler and nectar collection was done within a few hours (<4 hr) after sampling. Nectar was carefully extracted with sterile glass microcapillaries (Drummond Scientific Co.), immediately diluted in 500 µl of sterile, ultrapure deionized water (Purite Select; Purite Ltd), and stored at 4℃. Twentyfive microlitres of nectar dilutions were streaked on YGC (yeast extract glucose chloramphenicol) agar plates (Fluka, Sigma Aldrich) and trypticase soy agar (TSA; Panreac) for yeast and bacterial isolation, respectively. Cultures were incubated at room temperature for 7 days. A representative colony of each different morphotype was purified by repeated streaking on solid medium (Álvarez- Pérez & Herrera, 2013). Yeasts were preserved at −80℃ in 10% glycerol (Sigma- Aldrich) or using the Microbank system (Pro- Lab diagnostics), and bacterial isolates were stored at −20℃ in lysogeny broth (LB; Difco) containing 25% v/v glycerol. It can be argued that our study relies only on culturable microorganisms identified by Sanger sequencing of purified cultures. However, most nectar microbial communities worldwide have been studied using culturedependent methods (79.5% of the published manuscripts; C. de Vega, unpubl. review), and more importantly 3382 | Journal of Ecology de VeGA et Al. studies based on massive DNA sequencing have only extended the tail of rare species recovered by culturing. The diversity of the nectar inhabitants is dominated by a few culturable yeast and bacterial taxa (Morris et al., 2020; C. de Vega, unpubl. review). 2.3 | DNA extraction and sequencing Yeast genomic DNA was extracted by freezing in liquid nitrogen a loopful of cells suspended in 500 μl of sterile MilliQ water for 5 min, followed by a 5- min heat shock at 100℃ (Baleiras Couto et al., 2005). For every yeast isolate, the divergent D1/D2 domain of the largesubunit rRNA gene was amplified by PCR using primers NL- 1 and NL- 4, following methods of Kurtzman and Robnett (1998). Reaction mixtures and amplification conditions were as described in de Vega and Herrera (2012). Genomic DNA of bacteria was isolated by boiling bacterial colonies in 500 µl of ultrapure deionized water at 100℃ for 20 min. Cell debris was removed by centrifuging at 8,000 g for 2 min. The bacterial 16S rRNA gene was amplified using the universal primer 1492R (Reysenbach et al., 2000) and the eubacterialspecific primer 27F (Braker et al., 2001) as described in Álvarez- Pérez et al. (2012). The amplified DNA was cleaned with ExoSAP- IT (USB). Sequencing was performed by Macrogen (The Netherlands) and at ‘Estación Biológica de Doñana’ in an ABI PRISM 3130xl DNA automatic sequencer (Applied Biosystems). Sequences were assembled and edited using Sequencher 4.9 (Gene Codes Corporation, Inc.). Consensus sequences were aligned using MAFFT online service (Katoh et al., 2019). Analyses were carried out for yeast and bacteria separately. 2.4 | Phylogenetic analyses Phylogenetic trees were constructed for yeast and bacteria to test whether the microbial communities recovered from plants visited by distinct floral visitors were phylogenetically different (see below in phylogenetic characterization of microbial communities). Phylogenetic reconstructions were carried out using Bayesian inference and maximum likelihood (ML) approaches. For the Bayesian inference we used MrBayes v.3.2.7 (Ronquist et al., 2012) on XSEDE via the CIPRES Science Gateway (http://www.phylo.org/). We performed two independent runs for 5 × 106 generations with four chains each and trees sampled every 1,000 generations. Parameter values from each run and convergence of runs was assessed with Tracer v.1.6 (http://tree.bio.ed.ac.uk/softw are/trace r/). We allowed MrBayes to sample across the general time reversible (GTR) nucleotide substitution model space using reversible jump Markov chain Monte Carlo (rjMCMC) with the function nst = mixed (Huelsenbeck et al., 2004). The first 25% of trees of each run were discarded as burnin and a 50% majorityrule consensus tree was constructed. Maximum likelihood analyses were implemented in PhyML v.3.0 (Guindon et al., 2010) using the Smart Model Selection (http:// www.atgcmontpellier.fr/phyml/) based on the Akaike Information Criterium (Lefort et al., 2017). Trees were estimated with the GTR nucleotide substitution model. Branch support was assessed by 1,000 bootstrap resamples. Mucor mucedo (GenBank accession number JN206480) and Deinococcus radiodurans (Y11332) were used as outgroups for yeasts and bacteria, respectively. The resulting trees were drawn and edited with iTOL 5.6.3 (Letunic & Bork, 2016). Bayesian and ML approaches provided very similar and wellresolved trees, and only Bayesian results are shown. FIGURE 1 Examples of flowers visited by different pollinator guilds in South African plants. (a) Longtongued tabanid fly Philoliche aethiopica on Watsonia lepida; (b) Shorttongued bee on Aloe dominella; (c) Malachite sunbird on Burchellia bubalina; (d) Atrichelaphinis tigrina on Protea simplex [Colour figure can be viewed at wileyonlinelibrary.com] | 3383 Journal of Ecolog y de VeGA et Al. 2.4.1 | Operational taxonomic units (OTUs) designation Bacterial and yeast alignments were clustered into operational taxonomic units (OTUs) at 97% sequence similarity with MOTHUR v. 1.44.0 (Schloss et al., 2009). OTUs were assigned to a taxonomic identity by querying sequences against the GenBank database using the BLAST algorithm (Altschul et al., 1990) and the EzBioCloud server (https://www.ezbio cloud.net; last accessed 01 April 2020). Yeast and bacterial taxonomic information was obtained from the fungus nomenclature database Mycobank (http://www.mycob ank. org) and NCBI’s microbial database, respectively. 2.5 | Statistical analyses All analyses were conducted using R version 3.6.3 unless otherwise stated. 2.5.1 | Microbial incidence and cooccurrence Presence– absence of each OTU in each nectar sample (1/0 matrix) was used in this analysis to describe microbial incidence. To assess whether the incidence of yeasts and bacteria differs in nectar samples by pollinator type we fitted GLMMs with a binomial error distribution and including plant species identity as random factor. Models were fitted with the packages lme4 (Bates et al., 2015) and car (Fox & Weisberg, 2019). Additionally, in a subsequent analysis plant family was included in the same model as above as a fixed factor. Pearson correlations were used to examine relationships between yeast and bacteria cooccurrence in flowers using presence/ absence matrices (coded as categorical binary variable 1/0). This analysis informed about the cooccurrence of yeast and bacteria in the same nectar sample. We included both all nectar samples and only those containing any microbe. 2.5.2 | Structure and phylogenetic characterization of microbial communities Permutational multivariate analysis of variance (PERMANOVA) was conducted to test for differences in microbial communities among plant species visited by different pollinator groups. Colonyforming unit (CFU) counts of nectar bacteria could not be performed to estimate the number of viable microbial cells of each particular OTU in each nectar sample because of the high morphological similarity between the colonies of different OTUs. For this and subsequent analyses input data are based on counts of different OTUs recovered (confirmed by sequencing) in each set of replicate nectar samples for each plant species (replicated incidence data). We transformed the replicated incidence OTU table to a relative proportional data table with the decostand function (method = ‘total’) of the vegan package (v. 2.5- 7; Oksanen et al., 2019). Analysis of microbial community composition was conducted with the decostand and vegdist functions of this package. Two different analyses, including and excluding unique OTUs (those that appeared in a single nectar sample across all plant species) were carried out. The number of permutations was set at 9,999 and all other arguments used the default values. Pairwise comparisons were performed with the package rvaidememoire (v. 0.9- 79; Hervé, 2020) and correcting significance values for multiple testing with the ‘fdr’ option. Longtongued and shorttongued insectassociated communities were populated by virtually the same microbial taxa; yeast communities associated to plants visited by LT and ST insects were compositionally similar (p = 0.441), and the same arose for bacteria (p = 0.776). Consequently, the LT insects and ST insects categories were then merged in the same group (IN) for further analyses. Although our data suggest that as a group, plants visited by nonbeetle insects share similar nectar microbial communities, this does not mean that the degree of specialization of the plant species associated with these insects is similar. In South Africa, many of the plant species visited by nonbeetle insects, whether ST or LT, have highly specialized pollination systems involving a single species of insect pollinator (see Table S1). Beta diversity was measured using the unweighted UniFrac (uwUniFrac) distance to test whether the microbial communities recovered from plants visited by distinct floral visitors were phylogenetically different. Unweighted UniFrac distance considers presence/absence of OTUs (see McMurdie & Holmes, 2013 for details), and uses both the tree topology and the branch lengths, to test the hypothesis that there has been more unique evolution within each environment than would be expected by chance (Lozupone et al., 2006; Lozupone & Knight, 2005). The distance between communities is measured as the fraction of the branch length of the phylogenetic tree that is unique to one of the communities. The analyses were done separately for yeasts and bacteria. To calculate the uwUnifrac distance, two files are used as input: (a) the rooted OTU phylogenetic tree obtained for yeasts and bacteria using MrBayes as explained above, and selecting one representative of each OTU, and (b) a file linking presence/absence of OTUs to the environment information, using the three pollinator guilds (BE, BI, IN) as environment information. Distances were calculated with the UniFrac function of the package phyloseq (v. 1.34.0; see McMurdie & Holmes, 2013 for details). PERMANOVA analyses were conducted as above to test whether the microbial communities inhabiting the nectar of plants visited by different pollinator guilds were phylogenetically different using the uwUniFrac distance matrix. Pairwise dissimilarities between microbial communities were evaluated by comparisons adjusted for multiple testing (via ‘fdr’). 2.5.3 | Rarefaction, richness and (phylo) diversity assessment Rarefaction methods were applied to evaluate the adequacy of sampling by assessing whether the cumulative number of OTUs reached 3384 | Journal of Ecology de VeGA et Al. an asymptote. Species richness and evenness measurements can only be meaningfully compared among different communities when sample sizes are equal and thus sequences were extrapolated at the double sample size of the highest reference sample (N = 206; see results; Colwell et al., 2012; Chao et al., 2014). Samplebased rarefaction curves were computed for pollinator groups (BE, BI, IN, see above) using 10,000 randomizations without replacement. Microbial alpha diversity in each group was calculated as the number of OTUs (OTU richness), rarefied Chao richness, Simpson (1/D) and Shannon observed and estimated diversity indices with the package inext v.2.0.20 (Hsieh et al., 2016). iNext allows incidencefrequency data for diversity calculations. Input data for each assemblage consist of OTUs sample incidence frequencies (row sums of each incidence matrix). For each diversity measure, iNext uses the observed incidence data to compute diversity estimates and the associated 95% confidence intervals for rarefaction and extrapolation (Hsieh et al., 2016). To estimate observed and extrapolated samplebased phylogenetic diversity, Faith's phylogenetic diversity estimator was calculated with the inextpd v.0.3.1 package (Hsieh et al., 2016). Estimators were considered significantly different among floral visitor groups whenever the 95% confidence interval created by resampling did not overlap. 2.5.4 | Geographic structure The correlation between community dissimilarities and geographic distances matrices were tested with Mantel tests in vegan. As community distance measure we used both the Euclidean distances and the uwUnifrac distances which considers the phylogenetic relatedness. Additionally, patterns of geographic segregation were also assessed by grouping samples into two geographic origins (Cape and Drakensberg; see Table S2 and Figure S1) and further analysed with a PERMANOVA in vegan. Analyses were conducted including and excluding unique OTUs. 3 | RESULTS 3.1 | Incidence of microbial taxa and OTUs designation Nectar microbial communities occurred frequently in the study area, with 85.4% and 60.4% of plant species harbouring detectable yeasts and bacteria, respectively (Figure 2). For five plant species, we did not detect either yeast or bacterial communities (Albuca nelsonii, Aloe dominella, Cyrtanthus contractus, Dipcadi viride and Tulbaghia natalensis). For these species, the observation of nectar drops under the microscope confirmed the absence or extremely infrequent presence of microbial cells. The incidence of microbes in nectar samples differed between floral pollinator assemblages both for yeasts (Wald's χ2 = 23.542, p < 0.001) and bacteria (Wald's χ2 = 6.740, p = 0.034), with plant species visited by Coleoptera showing the highest percentage of flowers containing microorganisms. However, when we include FIGURE 2 Diversity and composition of nectar yeast and bacteria genera across the plant species visited by different pollinator guilds sampled in the Kwazulu- Natal and Cape region of South Africa. White bars indicate absence of yeast/bacteria communities in the nectar of a particular plant species. Five plant species that did harbour neither yeast nor bacterial communities are not included in the figure (Albuca nelsonii, Aloe dominella, Cyrtanthus contractus, Dipcadi viride and Tulbaghia natalensis). The yeast genera Metschnikowia is separated in ls (largespored, including M. caudata, M. drakensbergensis and M. proteae) and ss (smallspored, including M. koreeensis, M. rancensis and M. reukaufii) [Colour figure can be viewed at wileyonlinelibrary.com] | 3385 Journal of Ecolog y de VeGA et Al. plant family in the models the differences remained significant for yeast (Wald's χ2 = 9.648, p = 0.008), but not for bacteria (Wald's χ2 = 0.050, p = 0.975), suggesting that plants visited by different pollinator guilds have a similar probability of incidence of bacterial communities in nectar once the plant family is accounted for. Yeasts and bacteria coexisted only in 17% of the nectar samples. The frequency of cooccurrence of yeasts and bacteria across all nectar samples was not significant (N = 282, R = 0.082, p = 0.171). However, when only nectar samples containing microbes were considered, the regression showed a highly significant negative correlation (N = 182, R = −0.526, p < 0.001). DNA sequencing yielded a total of 217 yeast and 113 bacterial sequences which clustered into 35 and 29 OTUs, respectively (Table S3). Across plant species, we recorded 1– 13 OTUs (M ± SE = 2.58 ± 0.37; Figure 2). Individual nectar samples were colonized by 1– 5 OTUs. 3.2 | Nectar microbial communities structure by pollinator type Pollinator guilds help predict the composition of nectar microbial communities. Plants visited by either birds, beetles or by other insects supported significantly different yeast (F = 3.667, p < 0.001) and bacterial communities (F = 5.104, p < 0.001; Table 1). Pairwise comparisons of nectar microbial communities of plants visited by different pollinators were all significant (Table 1). Analyses including unique OTUs yielded similar results (Table S4). The pollinator guild also contributed to the maintenance of beta diversity and nectarassociated phylogenetic microbial segregation. Significant pvalues for Unifrac analyses were found when testing yeast (F = 5.305, p < 0.001) and bacteria (F = 5.769, p < 0.001) communities indicating that microbial communities are significantly clustered phylogenetically by pollinator type (Table 1; Figures 3 and 4). Pairwise comparisons performed with uwUnifrac distance both for yeast and for bacterial communities revealed significant differences for the comparisons of BEIN communities (p = 0.001 and p = 0.003 for yeast and bacteria) for BEBI communities (p = 0.001 and p = 0.023 for yeast and bacteria, respectively) and for BIIN communities (p = 0.023 and p = 0.024 for yeast and bacteria, respectively). Analyses including unique OTUs yielded similar results (Table S4). 3.3 | Yeast communities richness and diversity assessment Rarefaction curves of yeast OTU richness were close to reaching an asymptote for BE and IN groups (Figure 5) with samples having a Good's coverage value close to 1 (Table 2), indicating that our sampling effort was sufficient. Based on the Chao estimator of estimated richness, we detected in average 95.7% of expected OTUs in BE yeast communities (25 OTUs), 61.9% in BI communities (10 OTUs) and 69.4% in IN communities (9 OTUs; Table 2). The group of plants visited by beetles supported a significantly higher level of OTU richness and phylogenetic diversity than found in the other pollinator groups (Table 2; Figure 5). Estimated Shannon and Simpson indices showed an increase in diversity from the IN group (5.526 ± 1.058; 3.508 ± 0.619) to the BI group (10.896 ± 3.313; 6.948 ± 2.276) and the BE group (16.243 ± 1.218; 10.935 ± 1.092; Table 2). Ascomycetes were the largest phylum of yeasts recorded, representing 87.6% of the total yeast isolates and 21 OTUs (Figure 3), while Basidiomycetes represented only 12.4% of the isolates. Seventy three percent of Ascomycetous OTUs were unique to BE plants, including all species of Wickerhamiella, Hanseniaspora, Wickerhamomyces and Candida species in the Lodderomyces clade. Interestingly phylogenetic clades of the nectar specialist Mestchnikowia, the dominant genus in nectar, were separated by pollinator visitor type. Distinct clades were evident for beetlevisited flowers (largespored species: Metschnikowia drakensbergensis, M. caudata and M. proteae) and insect/birdsvisited flowers (smallspored species M. koreeensis, M. reukaufii and M. rancencis; Figures 2 and 3). Taphrina inosithophila was exclusively isolated from BI plants. Species of Basidiomycetes did not show a clear relationship to any pollinator group. Euclidean distance uwUnifrac distance Fdf p F df p Yeast Floral pollinators 3.667 2<0.001 5.305 2<0.001 Beetle versus insects 0.001 0.001 Beetle versus birds 0.001 0.001 Birds versus insects 0.042 0.023 Bacteria Floral pollinators 5.104 2<0.001 5.769 2<0.001 Beetle versus insects <0.001 0.003 Beetle versus birds 0.014 0.023 Birds versus insects 0.023 0.024 TABLE 1 Differences between nectar microbial communities of plants visited by different pollinator guilds, using Euclidean and unweighted Unifrac distances. ‘Insects’ = insects other than beetles 3386 | Journal of Ecology de VeGA et Al. 3.4 | Bacterial community richness and diversity assessment In bacteria, the OTU richness estimators only reached the plateau phase for the BE group (Figure 5). This suggests that additional sampling efforts would be required to cover total bacterial species richness in the BI and IN groups, as pointed by the Chao estimator (Table 2). Rarefaction curves for bacterial communities differed from those for yeast communities. The IN group showed the highest Shannon and Simpson estimated indices (23.747 ± 9.401; 7.971 ± 2.941), followed by the BI group (10.045 ± 5.842; 6.357 ± 5.248) and the BE group (6.579 ± 0.788; 5.126 ± 0.644; Table 2). Floral nectar of plants visited by nonbeetle insects have the highest phylogenetic diversity of bacteria (Table 2). FIGURE 3 Bayesian phylogenetic tree with the 217 yeast strains isolated from floral nectar in our study system in South Africa, based on the analysis of the D1/D2 domain of the largesubunit rRNA gene. Inner circle represents leaves labelled with colours corresponding to their plant pollinator guilds; isolates from beetlevisited plants (red), insectpollinated (other than beetle) plants (blue) and birdpollinated plants (green). The outer ring represents taxonomic class. Branch width is proportional to the posterior probability [Colour figure can be viewed at wileyonlinelibrary.com] | 3387 Journal of Ecolog y de VeGA et Al. Nectarassociated bacteria fell into three phyla as follows: Proteobacteria (89.4% of isolates, including α- , β- and γ- Proteobacteria, 20 OTUs), Actinobacteria (7.1%, 6 OTUs) and Firmicutes (3.5%, 3 OTUs; Table S3). The most frequently isolated genera were Pantoea, Asaia, Pseudomonas and Enterobacter, accounting for 60% of total isolates (Figures 2 and 4). The type of floral visitor clearly structured the nectar bacterial communities. All bacterial isolates recovered from BE flowers belonged to the α- and γ- Proteobacteria and several genera as Asaia, Enterobacter or Rahnella were exclusively isolated from BE flowers. In the genus Pseudomonas, a wellsupported phylogenetic clade was exclusively isolated from BE flowers (P. psychrotolerans) and another clade was isolated exclusively from IN flowers (P. palleroriana and P. extremorientalis; called Pseudomonas sp.1 and sp.2 in Figures 2 and 4). All genera in the β- Proteobacteria and almost all genera in the Actinobacteria were isolated exclusively from IN flowers, including among others the genera Paraburkholderia, Methylobacterium or Leifsonia, while the phylum Firmicutes was isolated both from IN and FIGURE 4 Bayesian phylogenetic tree with the 113 bacterial strains isolated from floral nectar, in our study system in South Africa based on of 16S rRNA gene sequences. Inner circle represents leaves labelled with colours corresponding to their plant pollinator guilds; isolates from plants visited by beetle (red), insectpollinated (other than beetle) plants (blue) and plants visited by birds (green). The outer ring represents taxonomic class. Branch width is proportional to the posterior probability [Colour figure can be viewed at wileyonlinelibrary.com] 3394 | Journal of Ecology de VeGA et Al. Letunic, I., & Bork, P. (2016). Interactive tree of life (iTOL) v3: An online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Research, 44(W1), W242– W245. https://doi. org/10.1093/nar/gkw290 Lozupone, C., Hamady, M., & Knight, R. (2006). UniFrac – An online tool for comparing microbial community diversity in a phylogenetic context. BMC Bioinformatics, 7, 371. Lozupone, C., & Knight, R. (2005). UniFrac: A new phylogenetic method for comparing microbial communities. Applied and Environmental Microbiology, 71, 8228– 8235. https://doi.org/10.1128/AEM.71.12. 8228- 8235.2005 Martinson, V. G., Danforth, B. N., Minckley, R. L., Rueppell, O., Tingek, S., & Moran, N. A. (2011). A simple and distinctive microbiota associated with honey bees and bumble bees. Molecular Ecology, 20(3), 619– 628. https://doi.org/10.1111/j.1365- 294X.2010.04959.x McFrederick, Q. S., Thomas, J. M., Neff, J. L., Vuong, H. Q., Russell, K. A., Hale, A. R., & Mueller, U. G. (2017). Flowers and wild megachilid bees share microbes. Microbial Ecology, 73(1), 188– 200. https://doi. o r g / 1 0 . 1 0 0 7 / s 0 0 2 4 8 - 0 1 6 - 0 8 3 8 - 1 McFrederick, Q. S., Wcislo, W. T., Taylor, D. R., Ishak, H. D., Dowd, S. E., & Mueller, U. G. (2012). Environment or kin: Whence do bees obtain acidophilic bacteria? Molecular Ecology, 21(7), 1754– 1768. https:// doi.org/10.1111/j.1365- 294X.2012.05496.x McMurdie, P. J., & Holmes, S. (2013). phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE, 8(4), e61217. https://doi.org/10.1371/journ al.pone.0061217 Mittelbach, M., Yurkov, A. M., Nocentini, D., Nepi, M., Weigend, M., & Begerow, D. (2015). Nectar sugars and bird visitation define a floral niche for basidiomycetous yeast on the Canary Islands. BMC Ecology, 15( 1 ) , 2 . h t t p s : / / d o i . o r g / 1 0 . 1 1 8 6 / s 1 2 8 9 8 - 0 1 5 - 0 0 3 6 - x Morris, M. M., Frixione, N. J., Burkert, A. C., Dinsdale, E. A., & Vannette, R. L. (2020). Microbial abundance, composition, and function in nectar are shaped by flower visitor identity. FEMS Microbiology Ecology, 96(3). https://doi.org/10.1093/femse c/fiaa003 Msweli, S. M. (2018). Friendly neighbours? Investigating ecological facilitation between Thunbergia atriplicifolia (Acanthaceae) and Exochaenium grande (Gentianaceae) (Doctoral dissertation). University of KwaZulu- Natal, South Africa. Ncube, B., Nair, J. J., Rárová, L., Strnad, M., Finnie, J. F., & Van Staden, J. (2015). Seasonal pharmacological properties and alkaloid content in Cyrtanthus contractus NE Br. South African Journal of Botany, 97, 69– 76. https://doi.org/10.1016/j.sajb.2014.12.005 Nelson, E. B. (2018). The seed microbiome: Origins, interactions, and impacts. Plant and Soil, 422(1– 2), 7– 34. https://doi.org/10.1007/s1110 4 - 0 1 7 - 3 2 8 9 - 7 Nepi, M. (2014). Beyond nectar sweetness: The hidden ecological role of nonprotein amino acids in nectar. Journal of Ecology, 102(1), 108– 115. https://doi.org/10.1111/1365- 2745.12170 Nicolson, S. W., Nepi, M., & Pacini, E. (2007). Nectaries and Nectar. Springer. Nicolson, S. W., & Thornburg, R. W. (2007). Nectar chemistry. In S. W. Nicolson, M. Nepi, & E. Pacini (Eds.), Nectaries and nectar (pp. 215– 264). Springer. https://doi.org/10.1007/978- 1- 4020- 5937- 7 Nicolson, S. W., & Van Wyk, B. E. (1998). Nectar sugars in Proteaceae: Patterns and processes. Australian Journal of Botany, 46(4), 489– 504. https://doi.org/10.1071/BT97039 Oksanen, J., Guillaume Blanchet, F., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P. R., O'Hara, R. B., Simpson, G. L., Solymos, P., Stevens, M. H. M., Szoecs, E., & Wagner, H. (2019). vegan: Community ecology package. R package version 2.5- 6. Retrieved from https:// C R A N . R - p r o j e c t . o r g / p a c k a g e =vegan Oliveira Santos, A. R. D., Lee, D. K., Ferreira, A. G.., do Carmo, M. C. Rondelli, V. M., Barros, K. O., & Lachance, M. A. (2020). The yeast community of Conotelus sp. (Coleoptera: Nitidulidae) in Brazilian passionfruit flowers (Passiflora edulis) and description of Metschnikowia amazonensis sp. nov., a largespored clade yeast. Yeast, 37(3), 253– 260. https://doi.org/10.1002/yea.3453 Öpik, M., Moora, M., Liira, J., & Zobel, M. (2006). Composition of rootcolonizing arbuscular mycorrhizal fungal communities in different ecosystems around the globe. Journal of Ecology, 94(4), 778– 790. https://doi.org/10.1111/j.1365- 2745.2006.01136.x Peay, K. G., Belisle, M., & Fukami, T. (2012). Phylogenetic relatedness predicts priority effects in nectar yeast communities. Proceedings of the Royal Society B: Biological Sciences, 279(1729), 749– 758. https:// doi.org/10.1098/rspb.2011.1230 Pozo, M. I., Herrera, C. M., & Bazaga, P. (2011). Species richness of yeast communities in floral nectar of southern Spanish plants. Microbial Ecology, 61( 1 ) , 8 2 – 9 1 . h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7/ s 0 0 2 4 8 - 0 1 0 - 9 6 8 2 - x Pozo, M. I., Lachance, M. A., & Herrera, C. M. (2012). Nectar yeasts of two southern Spanish plants: The roles of immigration and physiological traits in community assembly. FEMS Microbiology Ecology, 80(2), 281– 293. https://doi.org/10.1111/j.1574- 6941.2011.01286.x Ravenscraft, A., Berry, M., Hammer, T., Peay, K., & Boggs, C. (2019). Structure and function of the bacterial and fungal gut microbiota of Neotropical butterflies. Ecological Monographs, 89(2), e01346. https://doi.org/10.1002/ecm.1346 Rering, C. C., Beck, J. J., Hall, G. W., McCartney, M. M., & Vannette, R. L. (2018). Nectarinhabiting microorganisms influence nectar volatile composition and attractiveness to a generalist pollinator. New Phytologist, 220(3), 750– 759. https://doi.org/10.1111/nph.14809 Reysenbach, A. L., Longnecker, K., & Kirshtein, J. (2000). Novel bacterial and archaeal lineages from an in situ growth chamber deployed at a Mid- Atlantic Ridge hydrothermal vent. Applied and Environmental Microbiology, 66(9), 3798– 3806. https://doi.org/10.1128/ AEM.66.9.3798- 3806.2000 Ronquist, F., Teslenko, M., Van Der Mark, P., Ayres, D. L., Darling, A., Höhna, S., & Huelsenbeck, J. P. (2012). MrBayes 3.2: Efficient Bayesian phylogenetic inference and model choice across a large model space. Systematic Biology, 61(3), 539– 542. https://doi. org/10.1093/sysbi o/sys029 Rosa, C. A., Lachance, M. A., Silva, J. O., Teixeira, A. C. P., Marini, M. M., Antonini, Y., & Martins, R. P. (2003). Yeast communities associated with stingless bees. FEMS Yeast Research, 4(3), 271– 275. https://doi. o r g / 1 0 . 1 0 1 6 / S 1 5 6 7 - 1 3 5 6 ( 0 3 ) 0 0 1 7 3 - 9 Schaeffer, R. N., & Irwin, R. E. (2014). Yeasts in nectar enhance male fitness in a montane perennial herb. Ecology, 95(7), 1792– 1798. https:// doi.org/10.1890/13- 1740.1 Schaeffer, R. N., Mei, Y. Z., Andicoechea, J., Manson, J. S., & Irwin, R. E. (2017). Consequences of a nectar yeast for pollinator preference and performance. Functional Ecology, 31(3), 613– 621. https://doi. org/10.1111/1365- 2435.12762 Schaeffer, R. N., Rering, C. C., Maalouf, I., Beck, J. J., & Vannette, R. L. (2019). Microbial metabolites elicit distinct olfactory and gustatory preferences in bumblebees. Biology Letters, 15(7), 20190132. https:// doi.org/10.1098/rsbl.2019.0132 Schaeffer, R. N., Vannette, R. L., & Irwin, R. E. (2015). Nectar yeasts in Delphinium nuttallianum (Ranunculaceae) and their effects on nectar quality. Fungal Ecology, 18, 100– 106. https://doi.org/10.1016/j. funeco.2015.09.010 Schloss, P. D., Westcott, S. L., Ryabin, T., Hall, J. R., Hartmann, M., Hollister, E. B., Lesniewski, R. A., Oakley, B. B., Parks, D. H., Robinson, C. J., Sahl, J. W., Stres, B., Thallinger, G. G., Van Horn, D. J., & Weber, C. F. (2009). Introducing MOTHUR: Open– source, platform– independent, community– supported software for describing and comparing microbial communities. Applied and Environmental Microbiology, 75(23), 7 5 3 7 – 7 5 4 1 . h t t p s : / / d o i . o r g / 1 0 . 1 1 2 8 / A E M . 0 1 5 4 1 - 0 9 Sharaby, Y., Rodríguez- Martínez, S., Lalzar, M., Halpern, M., & Izhaki, I. (2020). Geographic partitioning or environmental selection: What governs the global distribution of bacterial communities | 3395 Journal of Ecolog y de VeGA et Al. inhabiting floral nectar? Science of the Total Environment, 749. https:// doi.org/10.1016/j.scito tenv.2020.142305 Six, D. L. (2012). Ecological and evolutionary determinants of bark beetlefungus symbioses. Insects, 3(1), 339– 366. https://doi.org/10.3390/ i n s e c t s 3 0 1 0 3 3 9 Springer, T. J. (2019). The role of pollinators as ecological drivers of diversification in the Drakensberg mountain centre endemic genus Glumicalyx (Scrophulariaceae, limoselleae) (Doctoral dissertation). University of Witwatersrand, Johannesburg, South Africa. Starmer, W. T., & Lachance, M. A. (2011). Yeast ecology. In C. P. Kurtzman, J. W. Fell, & T. Boekhout (Eds.), The yeasts: A taxonomic study (5th ed., pp. 65– 83). Elsevier. Steenhuisen, S. L., & Johnson, S. D. (2012). Evidence for beetle pollination in the African grassland sugarbushes (Protea: Proteaceae). Plant Systematics and Evolution, 298, 857– 869. https://doi.org/10.1007/ s 0 0 6 0 6 - 0 1 2 - 0 5 8 9 - 5 Suh, S. O., McHugh, J. V., Pollock, D. D., & Blackwell, M. (2005). The beetle gut: A hyperdiverse source of novel yeasts. Mycological Research, 109(3), 261– 265. https://doi.org/10.1017/S0953 75620 5002388 Tucker, C. M., & Fukami, T. (2014). Environmental variability counteracts priority effects to facilitate species coexistence: Evidence from nectar microbes. Proceedings of the Royal Society B: Biological Sciences, 281(1778), 20132637. https://doi.org/10.1098/rspb.2013.2637 Urbina, H., Schuster, J., & Blackwell, M. (2013). The gut of Guatemalan passalid beetles: A habitat colonized by cellobiose– and xylose– fermenting yeasts. Fungal Ecology, 6(5), 339– 355. https://doi. org/10.1016/j.funeco.2013.06.005 Vannette, R. L. (2020). The floral microbiome: Plant, pollinator, and microbial perspectives. Annual Review of Ecology, Evolution, and Systematics, 51, 3 6 3 – 3 8 6 . h t t p s : // d o i . o r g / 1 0 . 1 1 4 6 / a n n u r e v - e c o l s y s - 01172 0- 013401 Vannette, R. L., & Fukami, T. (2016). Nectar microbes can reduce secondary metabolites in nectar and alter effects on nectar consumption by pollinators. Ecology, 97(6), 1410– 1419. https://doi. org/10.1890/15- 0858.1 Vannette, R. L., & Fukami, T. (2017). Dispersal enhances beta diversity in nectar microbes. Ecology Letters, 20(7), 901– 910. https://doi. org/10.1111/ele.12787 Vannette, R. L., & Fukami, T. (2018). Contrasting effects of yeasts and bacteria on floral nectar traits. Annals of Botany, 121(7), 1343– 1349. https://doi.org/10.1093/aob/mcy032 Vannette, R. L., Gauthier, M. P. L., & Fukami, T. (2013). Nectar bacteria, but not yeast, weaken a plant– pollinator mutualism. Proceedings of the Royal Society B: Biological Sciences, 280(1752). https://doi. org/10.1098/rspb.2012.2601 Vásquez, A., Forsgren, E., Fries, I., Paxton, R. J., Flaberg, E., Szekely, L., & Olofsson, T. C. (2012). Symbionts as major modulators of insect health: Lactic acid bacteria and honeybees. PLoS ONE, 7(3), e33188. https://doi.org/10.1371/journ al.pone.0033188 Willmer, P. (Ed.) (2011). Rewards 2: The biology of nectar. In Pollination and floral ecology (pp. 190– 220). Princeton University Press. Yang, M., Deng, G. C., Gong, Y. B., & Huang, S. Q. (2019). Nectar yeasts enhance the interaction between Clematis akebioides and its bumblebee pollinator. Plant Biology, 21(4), 732– 737. https://doi.org/10.1111/ plb.12957 Zarraonaindia, I., Owens, S. M., Weisenhorn, P., West, K., Hampton- Marcell, J., Lax, S., Bokulich, N. A., Mills, D. A., Martin, G., Taghavi, S., van der Lelie, D., & Gilbert, J. A. (2015). The soil microbiome influences grapevine– associated microbiota. MBio, 6(2), https://doi. o r g / 1 0 . 1 1 2 8 / m B i o . 0 2 5 2 7 - 1 4 SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section. How to cite this article: de Vega, C., Álvarez- Pérez, S., Albaladejo, R. G., Steenhuisen, S.- L., Lachance, M.- A., Johnson, S. D., & Herrera, C. M. (2021). The role of plant– pollinator interactions in structuring nectar microbial communities. Journal of Ecology, 109, 3379– 3395. https://doi. org/10.1111/1365- 2745.13726