The above values were calculated for each inoculated microbe individually and for all five species collectively. For comparison with real nectars and to test our inoculum in artificial solutions, we also added 1μL of inoculum to 10μL of 30% m/m sucrose and an artificial nectar containing sugars and peptone in strip tubes. Tubes were sealed and incubated at 25°C for 24h, then processed identically to actual nectar samples.We estimated concentrations of hydrogen peroxide , a known antimicrobial reactive oxygen species found in some nectars , in the nectar of separate, noninoculated flowers of most sampled plant species . Peroxide values from noninoculated nectar represent initial conditions which would be experienced by microbes arriving in flowers. To assess the contribution of floral morphology, we scored floral phenotypes of all plant species on the basis of 28 binary traits used in past studies to represent pollination syndromes in multivariate space using Bray–Curtis dissimilarity. We determined trait states through a combination of observation and reference with the Jepson eFlora . We also encoded other traits of particular interest such as inflorescence density and corolla fusion.We conducted analyses in R . Using package lme4 , we constructed linear mixed effect models with nectar volume, total and by-species CFU density and CFU Shannon–Wiener diversity index as dependent variables. As independent variables, we included nectar volume and temperature extrema, blueberry production and plant species as a random intercept effect. We obtained type III sums of squares, F– and p-values and Kenward-Roger degrees of freedom using function ‘Anova’ in package car . We inspected model residuals for normality and variance inflation factors to assess multicollinearity. We also created separate linear models with either plant species or nectar peroxide concentration as a fixed effect, as peroxide data were not collected for three species .

For linear models in which we included a quadratic predictor, we conducted a likelihood ratio test comparing the goodness of fit of the models with and without the quadratic term. To test if microbial community composition differed by plant species and temperature extrema, we used function ‘adonis’ in package vegan to perform a permutational multivariate analysis of variance. We used function ‘betadisper’ to examine multivariate homogeneity of dispersions across plant species. Community composition was visualised using nonmetric multidimensional scaling ordination, and we tested for significant microbe species vectors using function ‘envfit’. As above, a separate analysis was conducted with peroxide concentration as a predictor variable. To test for co-occurrence between microbe species, we generated Pearson correlation matrices on CFU densities, for both our entire dataset and for each plant species individually, and visualised matrices using package ‘corrplot’ . To estimate plant phylogenetic relationships among sampled plant species, we used the function ‘phylo.maker’ in package V.PhyloMaker2 using the reference plant phylogeny GBOTB.extended.TPL. Using this tree, we tested for a phylogenetic signal of nectar volume, CFU densities and Shannon diversity using function ‘multiPhylosignal’ in package picante with 10,000 simulations. To test for relationships between plant phylogenetic relatedness and multivariate microbe community composition, we created a pairwise distance matrix of plant phylogenetic relatedness using function ‘cophenetic.phylo’ in package ape . We compared this distance matrix to a Bray– Curtis dissimilarity matrix of the mean CFU densities of each microbe by plant species using a Mantel test via function ‘mantel’ in package vegan, calculating Spearman’s ρ with 10,000 permutations. We also created a Bray–Curtis dissimilarity matrix of plant species based on floral trait data and compared this to the two aforementioned matrices.

We controlled for the effect of plant phylogenetic distance on pollination syndrome using a partial Mantel test via function ‘mantel.partial’. We generated correlograms for all Mantel tests using the function ‘mgram’ in package ecodist . Figures were created using package ggplot2 and tree plots using ggtree and custom function ‘ggtreeplot’ .In this study, we observed shifts in the composition of a synthetic microbe community inoculated into the floral nectar of 31 flowering plant species, mainly predicted by plant species and temperature. Host species-dependency of plant microbiomes is consistent with previous observational studies of nectar , pollen , phyllospheres and roots . Our manipulative study complements this work by leveraging a phylogenetically diverse array of plant taxa, highlighting the role of plant host identity as a driver of microbe community assembly outside of dispersal and priority effects. Specifically, we provide experimental evidence that nectar microbiomes become distinct across plants even when initial community composition is the same.Several factors can impact nectar microbe community assembly even when controlling for dispersal , including filtering by host plants and interactions among microbe species . Our study shows support for both processes. In certain plant species , few microbe taxa established and were at low densities. Furthermore, community assembly in certain plant species was more stochastic than in others ; it may be that environmental stress generates stronger selection and more uniform communities . Interestingly, the highest uniformity in community composition we documented was in our two experimental control solutions . This suggests community stochasticity is higher in actual flowers, likely due to variation in nectar properties across plants. Previous work in a single plant species suggests dispersal positively contributes to nectar microbe beta diversity via priority effects. Our study suggests that even in the absence of dispersal, nectar microbe beta diversity may be more constrained in some plant species than in others.

Plant-level mediation of microbe community assembly outside of pollinator vectoring is consistent with the hypothesis that physical or chemical properties of flowers and nectars differentially inhibit microbegrowth. This phenomenon may serve as an adaptive defence against nectar spoilage , but it is also possible that some nectars could facilitate the growth of particular microbes. Several plants we inoculated belong to genera containing species known to produce antimicrobial nectar metabolites: for example, alkaloids, phenolics, and terpenoids . We suspect that the occurrence of secondary metabolites or other nectar constituents might explain the distinct differences in community structure we observed across plant species. This is supported by our finding that nectar peroxide concentration, which is regulated via nectarin proteins in Nicotiana , was negatively associated with total microbe density across plant species. The effects of peroxide concentration differed depending on microbe species, mirroring trends from in vitro assays , perhaps due to differences in microbial detoxification mechanisms. However, mean peroxide concentration on its own explained very little variation in the dataset. Future work incorporating a much broader diversity of nectar chemicals and compounds in a similarly diverse array of plant species is needed to determine if such a predictive framework exists. Similarity in nectar chemistry among species can be associated with phylogenetic relatedness in certain plant clades . We found plant relatedness was weakly positively associated with similarity in microbe community composition, but not with the densities of any individual microbes. Plant relatedness alone was not sufficient to explain the similarity in microbe community assembly however as this relationship was not monotonic. Within major plant clades, plant species in our study hosted similarly composed microbe communities , but several exceptions are clear. Hierarchical clustering analysis reflected this pattern as some, but not all, plant species of major clades clustered together and congeneric plant species did not necessarily cluster closely. In other plant microbiomes, host plant phylogeny can be a predictor of microbial communities , vary between bacteria versus fungi or show little predictive power . In the latter cases, microbe communities were better predicted by plant traits, implying a weak relationship between plant phylogeny and traits . We found that floral morphological traits were correlated with plant relatedness, but were not predictive of nectar microbe communities , blueberry in container suggesting that key host traits mediating microbial growth were not measured in the current experiment. Floral trait similarity, here approximating pollination syndromes , not predicting variation in microbial composition is contrary to predictions based on floral surveys of open flowers in which pollinator identity or pollination syndrome is a key predictor of nectar microbial communities .

Nevertheless, the microbes used here are common in most geographical regions sampled to date, and we expect that pollinator movement will homogenise microbial populations to some extent within coflowering communities.Interactions among microbes likely influenced community assembly within flowers, and we detected signatures of both facilitation and competition depending on analytical approach. All five species in our synthetic community were capable of coexisting after 24 h at varying densities in artificial nectar in vitro. We detected only positive or neutral correlations between microbe species pairs, similar to Francis, Mueller, and Vannette , in both our pooled dataset and separately within each plant species. In the pooled dataset, Neokomagataea was the only species showing no positive correlations with any other microbe, perhaps due to unknown specificities in its nutrient requirements. At first, this all seems to suggest facilitation among some species pairs , or that competition between microbes at 24 h was insignificant. However, we also observed a unimodal, ‘hump-shaped’ relationship between CFU Shannon diversity and increasing total CFU density across plant species. Shannon diversity increased with CFU density until roughly 102CFU μL−1, after which diversity declined as density increased. Similar unimodal relationships between microbe diversity and productivity have been documented in both artificial and natural aquatic environments . Several underlying mechanisms have been proposed for this relationship, including a shift from abiotic to biotic pressures along the gradient of increasing productivity . We suggest that extreme resource limitation or antimicrobial conditions in some nectars may limit the growth of all microbes in some nectars, whereas the availability of pollen or other nutrients may enable dominance of specific microbes in other nectars. Co-occurrence networks reflect the combined influence of biotic interactions and the environment and may under represent negative, nontrophic interactions relative to empirically observed interactions . Additionally, only 24h post inoculation may represent an early to intermediate time point in community progression, perhaps preceding manifestation of antagonistic interactions . Conversely, the unimodal relationship across plant species suggests that the growth of specific microbes in highly productive environments can effectively reduce community diversity, resulting in competitive exclusion of other microbes. We hypothesise that such competitive dynamics will be more apparent in longer persisting or senescing flowers , but interaction outcomes can also depend on microbes’ phylogenetic relatedness or their local adaptations to flower environments .Consistent with our expectations, microbial growth was correlated with seasonal temperature shifts. Maximum and minimum ambient temperatures over 24h of growth were differentially associated with components of community assembly and species individual densities, further supporting that nectar microbe species differ in their temperature ranges for optimal growth . Notably, increases in daily minimum temperature increased densities of the nectar yeast Metschnikowia and bacterium Acinetobacter, suggesting that their population densities are limited by growth rate. In contrast, high maximum daily temperatures decreased Lactobacillus and Neokomagatea densities, suggesting temperature thresholds for these microbes. These patterns are consistent with previous observations that nectar yeast prevalence was found to be positively correlated with temperature , while high temperatures can negatively impact nectar bacterial diversity . Although we did not detect a significant effect of temperature on nectar volume in the current study , open flowers likely experience increased evaporation affecting nectar composition and secretion . In any case, our observations indicate, similar to other plant– microbe systems , that shifts in temperature extrema over time may alter baseline effects of plant host filtering on nectar microbial communities in predictable ways, such as favouring certain microbe species over others or limiting maximum achievable levels of diversity. The implications of these shifts for plant–pollinator interactions deserve further attention. We also emphasise that in our study, different plant species were necessarily sampled at different times of year due to flowering phenology, so temperature was confounded with other variables like plant host identity, humidity and solar radiation, all of which affect microbial assembly in flowers . Nevertheless, our results suggest that differential response to temperature minima and maxima mediates microbial growth and interactions. We show that plant species host consistent microbial communities, suggesting plant populations could potentially adapt to the presence of specific microbes . Insect populations could also adapt to plant-specific microbial growth, such as perceived volatile cues or acquired microbes . For example, Apilactobacillus is thought to benefit pollinators , thus increasing high temperatures may inhibit the growth of this beneficial microbe.