Plant-associated microbiomes profoundly influence host interactions with below- and aboveground environments. Characterizing plant-associated microbiomes in experimental settings have revealed important drivers of microbiota assemblies within host species. However, it remains unclear how important these individual drivers (e.g., organ type, host species, host sexual phenotype) are in structuring the patterns of plant–microbiota association in the wild. Using 16s rRNA sequencing, we characterized root, leaf and flower microbiomes in three closely related, sexually polymorphic Fragaria species, in the broadly sympatric portion of their native ranges in Oregon, USA. Taking into account the potential influence of broad-scale abiotic environments, we found that organ type explained the largest variation of compositional and phylogenetic α- and β-diversity of bacterial communities in these wild populations, and its overall effect exceeded that of host species and host sex. Yet, the influence of host species increased from root to leaf to flower microbiomes. We detected strong sexual dimorphism in flower and leaf microbiomes, especially in host species with the most complete separation of sexes. Our results provide the first demonstration of enhanced influence of host species and sexual dimorphism from root to flower microbiomes, which may be applicable to many other plants in the wild.
The plant-associated microbiome is considered part of the extended plant phenotype1,2. Microbiota living on the surface and inside of host plants mediate processes vital to plant fitness, ranging from nutrient acquisition and stress responses, to pollination1,3,4,5. Recent quantitative characterizations of plant-associated microbiomes have focused on microbiota harbored by leaves and roots in greenhouse and common garden settings6,7,8,9,10,11, and have identified important drivers underlying microbiota assemblies in plants aboveground (AG) and belowground (BG).
These drivers include environment-related factors such as soil source6 and experimental site7,8 that may vary in climatic conditions, and host-related factors including organ type9, within-organ compartment10, host species11 and genotype8. However, there has been a lack of comparative data that broaden AG and BG microbiomes to also encompass those associated with plant reproductive organs (e.g., flowers) and sexual phenotype, and generalize relevant findings from a single to multiple host species, especially in the wild. These are, nevertheless, an important first step towards understanding how plants recruit microbiota that may in turn affect host fitness in wild populations, and for determining whether individual drivers identified under experimental settings are also of paramount importance in the wild.
While consensus from studies in experimental or managed field settings is growing that organ type (root, leaf, flower)12 and host plant species11 influence microbial communities, their relative roles have not been explicitly quantified. This is in part because empirical studies profiling plant AG (leaf and flower) and BG (root) microbiomes are often not directly comparable, and vary in terms of environmental settings, microbial community type and anatomical details of plant organs. For example, comparative investigations of leaf microbiomes tend to focus on epiphytic microbiota, and have often involved tens to hundreds of host species in semi-natural or natural habitats13,14. By contrast, those on root microbiomes have typically evaluated microbiota of a few host species11, or different genotypes and root compartments within a single host species6,7,10, often in experimental and/or managed field sites. Studies of flower microbiomes mostly focus on compartment differences, and/or in comparison to other organs, within a single host species in the wild or managed field sites12,15,16. While these studies begin to suggest a defining role of organ type in structuring bacterial microbiota in plants1, the differences among studies make the quantitative inference of respective effects difficult. Quantifying microbiota across multiple organ types (root, leaf, flower) and host plant species in situ will provide a direct evaluation of the extent to which organ type exceeds host species in shaping microbial communities in the wild.
The effect of host species on microbiomes has been examined in controlled settings (in roots)11 and in the wild (in roots and/or leaves)13,17. In wild populations, host species–microbiota association can be influenced by range differences among plant species18. This is because the regional species pool of microbiota that potentially colonize host plants (e.g., those harbored by soil) can vary with broad-scale abiotic environments of plant species19. Additionally, even when ranges overlap and plants occur in sympatry, species can vary in their local habitats, where they also alter the local species pool of microbiota20, by enriching and depleting certain microbes. As a result, host species effect on microbiomes in the wild could be viewed as (residual) host effect, after accounting for that attributable to broad-scale abiotic environments (e.g., temperature, precipitation)21. Such (residual) host species effect represents species–microbiota association attributable to host phenotype (i.e., the outcome of plant genotype and local environment), and local biotic conditions (microbiota) that are modified by plant species over time; this is, by definition, the extended phenotype of host plants. Thus, we may expect host species effect on microbial communities to be stronger in the wild, owing to long-term accumulated feedbacks between plants and local microbiota, than in controlled settings. This knowledge is essential for understanding the variation of plant–microbiota association that can influence plant fitness in nature.
Microbiomes can be also influenced by host sex, as demonstrated in animal systems including humans22,23. Such sexual dimorphism, however, has rarely been explored in plants (but see ref.24), despite the fact that sexual phenotype (female, male or hermaphrodite) is known to influence floral and functional traits, and several ecophysiological processes25,26. As a result, plant sexes may differ in the principles governing microbiota assemblies (i.e., dispersal, habitat filtering and niche partitioning). First, the species pool of colonizing microbes can vary between female and male plants, owing to differential visitation of pollinators that may carry and disperse microbes, as an outcome of differences in floral rewards and attractive traits between sexes27,28. Second, plant sex can influence the niche space available for microbiota. Sexual dimorphism in flower size and longevity26 likely affects the size and dynamics of microbial habitats. Likewise, sexual dimorphism in leaf traits (e.g., trichomes and leaf toughness)29 can potentially define the living environments of leaf microbiota4,30. Finally, sex-differential susceptibility and/or allocation to defense25,31,32 could alter resident microbial communities via microbe–microbe interactions1. Comparisons across AG and BG organs would inform broadly on the potential for sexual differences in microbiomes.
In this study, we aim to quantify root, leaf and flower microbiomes in three closely related wild strawberries in the broadly sympatric portion of their native ranges. These perennial, sexually polymorphic Fragaria33 include the dioecious F. chiloensis, which is a coastal specialist growing in front dunes, and the subdioecious (hermaphrodite, male and female) F. virginiana ssp. platypetala growing in fertile mesic forest edge, as well as their natural hybrid, subdioecious F. ×ananassa ssp. cuneifolia growing in intermediate habitats34. All are the wild relatives of the cultivated strawberry (F. ×ananassa ssp. ananassa). Specifically, we aim to address three key questions concerning the effects of host species, organ type and sexual phenotype in these wild populations: (1) What is the relative importance of host species and organ type in structuring microbial communities? (2) Does the magnitude of the host-species effect on microbiome vary among root, leaf and flower? (3) Do microbiomes differ between host plants of different sexual phenotype?
Root, leaf and flower microbiomes of Fragaria
These Fragaria plants and associated microbiota experienced distinct abiotic environments in Oregon where they are in broad sympatry (Fig. 1a,b), as revealed by the principal component analysis (PCA) of seven climatic variables35 and elevation data of the seven populations. The first two principal components (PC1 and PC2, denoted as PC1.clim and PC2.clim) accounted for 92% of the abiotic variation. Amplicon sequencing of the V4 region of 16S small subunit ribosomal RNA (rRNA) gene, with peptide nucleic acid (PNA) clamps36, generated substantial chloroplast and mitochondrial sequences (averagely 63% of reads per sample). After removing the plant contaminants, low-frequency OTUs and low-coverage samples, OTU observations averaged 7896 per sample (median = 2796). Individual microbial communities were normalized to the same observations for each sample using the median per-sample depth (of 2796), while keeping per-sample OTU relative abundances unchanged37. The resultant microbial community matrix consisted of 1577 OTUs from 27 root, 22 leaf and 23 flower samples of the three host species, representing the core set of microbiota (comprising both epiphytic and endophytic taxa) in our samples.
The root, leaf and flower microbiomes contained predominantly bacterial taxa (with one Archaea OTU), among which Proteobacteria, Bacteroidetes and Actinobacteria were the most abundant phyla (Fig. 1c–e). Root microbiomes possessed more OTUs than leaf and flower microbiomes (Figs 1f and S1). The majority of leaf (94%) and flower (98%) OTUs were also found in roots (Fig. 1f).
Organ type structures plant-associated microbiota in the wild
Organ type predicted species α-diversity of microbial communities (Shannon diversity, F = 70.33, df = 2, P < 0.001; Fig. 2c; Table S1), after controlling for abiotic environments (PC1.clim and PC2.clim), host species and sex (see Methods). Shannon diversity (in terms of the least-squares mean) was highest in root microbiomes, and significantly decreased in leaf (t = 9.20, df = 51, P < 0.001) and flower microbiomes (t = 10.91, df = 51, P < 0.001; Fig. 2c). Likewise, root microbiomes harbored significantly higher phylogenetic α-diversity (Fig. 2d,e; Table S1), using the metrics38 that both scale positively with (e.g., phylogenetic diversity, PD; Fig. 2d) and are insensitive to species richness (e.g., abundance-weighted mean phylogenetic distance, MPD; Fig. 2e). Within AG microbiomes, microbial α-diversity was comparable between leaves and flowers (Fig. 2c–e).
Despite substantial overlap in OTUs, AG and BG microbiomes exhibited distinct community structures (Figs 2a,b and S2). Hierarchical clustering analysis (Fig. 2a) clearly separated root and leaf/flower microbiomes regardless of host species, and also showed that OTUs of high frequency in roots were not the same ones in leaves and flowers. The nonmetric multidimensional scaling (NMDS) of Bray–Curtis dissimilarity also identified AG and BG organs as the primary source of variation in compositional β-diversity of microbial communities (Fig. 2b). Similar to Bray–Curtis NMDS, the separation of root and leaf/flower microbiomes along the dominant ordination axes was revealed by principal coordinates analyses (PCoAs) of phylogenetic β-diversity metrics (abundance-weighted UniFrac distance, and inter-community MPD, betaMPD; Fig. S2).
Consistent with the qualitative inference from NMDS and PCoAs, permutational multivariate analyses of variance (PERMANOVA) showed that organ type accounted for the largest source of variation in microbial β-diversity among all tested predictors (Table S2), when using Bray–Curtis (19.4% of variation, F = 7.924, df = 2, P = 0.001) and UniFrac distance (21.1%, F = 8.838, P = 0.001). PERMANOVA of betaMPD agreed with the other two β-diversity metrics that microbial community structures were significantly affected by organ type (F = 2.130, df = 2, P = 0.001; Table S2), albeit organ type (the main effect) and its interaction with species accounted for a similar amount of variation (6.2% and 7.0%, respectively).
Supporting the PERMANOVA results, generalized linear models (GLMs) with family-wise error rate (FWER) control in mvabund39 (denoted as FWER-GLMs) also showed that organ type strongly predicted multivariate abundances of microbial communities (deviance = 19,793, df = 2, P = 0.001; Table S2). In addition to organ type, host species (deviance = 19,793, P = 0.001) and sex (deviance = 12,619, P = 0.006) were also identified as having significant impact on microbial communities, unlike in PERMANOVA (Table S2). The difference between Bray–Curtis PERMANOVA and FWER-GLMs suggested that OTUs responding to host species or sex likely had low or modest variability in abundances between groups of interest, which failed to be detected by less sensitive PERMANOVA40, but collectively these OTUs contributed to differential microbial communities. By contrast, OTUs of large among-organ variability were likely involved in distinguishing AG and BG microbiomes, as detected by both PERMANOVA and FWER-GLMs.
FWER-GLMs that detected significant microbial community differentiation caused by organ type also identified the responsible OTUs (Table S3). These differentially abundant OTUs (N = 120) attributable to organ type accounted for a small portion (8%) of the overall OTUs (Fig. S3; Table S3). We further assessed the effect size (i.e., fold change [log2], log2FC) and sign (i.e., depleted/enriched) of differentially abundant OTUs in leaf or flower microbiomes relative to root microbiomes, controlling for all other factors using edgeR41 GLMs with false discovery rate control (FDR-GLMs). As a result, 414 OTUs were identified as differentially abundant between leaf and root microbiomes, and 404 OTUs between flower and root microbiomes (Figs S3, S4 and Table S3), more than three times the OTUs identified by FWER-GLMs. This distinction was primarily caused by stringent FWER relative to FDR control.
Depletion effects dominated AG microbiomes relative to BG microbiomes (Fig. S3). In flower microbiota, FDR-GLMs identified 22 OTUs as significantly enriched and 382 OTUs as significantly depleted (Fig. S3a). The enriched OTUs concentrated in the three dominant phyla. Thirteen out of the 22 enriched OTUs in flower microbiomes, with large effect size (typically of ≥5 log2FC), overlapped those identified by FWER-GLMs as responsive to organ type (Fig. S3a). These large-effect, enriched OTUs were primarily from Sphingomonas, Hymenobacter, Janthinobacterium, Pseudomonas, Methylobacterium and Salinibacterium (Table S3). The 382 OTUs that were significantly depleted from flower microbiomes ranged across diverse phyla (Fig. S3a; Table S3), among which 95 OTUs were also identified by FWER-GLMs as significantly influenced by organ type (Fig. S3a). These 95 OTUs had (log2) fold changes averaged −4.50 (Table S3), among which OTUs from Streptomyces, Bradyrhizobium and Steroidobacter showed the largest effect sizes (in absolute values). Flower and leaf microbiomes were similar in enriched and depleted OTUs (Figs S3, S4 and Table S3).
Host species influence increases from root to leaf to flower microbiomes
As plants harbored distinct AG and BG microbiomes, host species effect (see definition in Introduction) was assessed for root, leaf and flower microbiomes separately, while controlling for abiotic environments and host sex. Host species did not predict microbial α-diversity for any organ-associated microbiomes, nor did abiotic environments (Table 1). Different from α-diversity measures, the extent to which host species overlapped in OTUs changed from BG to AG microbiomes (Fig. 3a–c). In root microbiomes, the three Fragaria species overlapped substantially in OTUs, accounting for 71–82% of individual host species OTUs (Fig. 3a); but this OTU sharing dropped to 20–58% in leaf and 27–45% in flower microbiomes (Fig. 3b,c), suggesting that microbial communities could be more similar among host species in roots, compared to leaves and flowers.
In support of this inference, constrained PCoAs of β-diversity among organ-specific microbial communities (Figs 3d–i and S5) detected diminishing community separation by host species from AG to BG microbiomes, while controlling for all other factors. For flower microbiomes (Figs 3f,i and S5), F. chiloensis (F.chilo) associated microbial communities were segregated with the other two species along the first axis of constrained PCoAs for all three β-diversity metrics, whereas those of F. virginiana ssp. platypetala (F. virg) and the hybrid F. ×ananassa ssp. cuneifolia (F.cunei) were separated by the second axis for compositional β-diversity (Bray–Curtis) metric. In leaf microbiomes, community separation was only seen when using Bray–Curtis metric between the hybrid F.cunei and two parental species along the second axis (Fig. 3e); when using phylogenetic β-diversity metrics, host species overlapped in microbial communities (Figs 3h and S5). This microbiome overlapping among host species was most pronounced in roots.
Consistent with the constrained PCoAs results, PERMANOVA (Table 1) showed that host species explained the largest source of variation in flower microbiomes among all tested predictors (Bray–Curtis, 12.9% of variation, df = 2, F = 1.507, P = 0.049; UniFrac, 20.6%, F = 2.920, P = 0.004; betaMPD, 13.1%, F = 1.575, P = 0.006). Further evidence came from FWER-GLMs that host species strongly predicted multivariate abundances of flower microbiomes (deviance = 2120, df = 2, P = 0.030; Table 1). In leaf microbiomes, UniFrac and betaMPD PERMANOVA corroborated the inference from constrained PCoAs that host species did not affect the phylogenetic community structures of leaf microbiota (F = 0.827, df = 2, P = 0.6 and F = 1.001, P = 0.4, respectively; Table 1). Yet, the subtler microbial community separation by host species in leaves relative to flowers (Fig. 3e,f) was captured by FWER-GLMs (deviance = 1771, df = 2, P = 0.031; Table 1), albeit not by less sensitive PERMANOVA (Bray–Curtis, F = 0.966, df = 2, P = 0.5). By contrast, in root microbiomes both FWER-GLMs and Bray–Curtis PERMANOVA (Table 1) supported that host species did not predict multivariate abundances of root microbial communities (deviance = 4657, df = 2, P = 0.056 and F = 1.131, P = 0.3, respectively), as well as the phylogenetic community structures (UniFrac PERMANOVA, F = 0.887, P = 0.5; betaMPD, F = 0.959, P = 0.8). Across root, leaf and flower microbiomes, abiotic environments (PC1.clim and PC2.clim together) explained a similar amount of variation in microbial community structures as did host species (PERMANOVA, Table 1).
FWER-GLMs and FDR-GLMs were used to identify the OTUs underlying microbial community differentiation caused by host species, for flower and leaf microbiomes separately. Surprisingly, no OTUs were reported as differentially abundant among host species in flowers or leaves (Tables S4, S5). However, the effect size estimation by FDR-GLMs indicated that many OTUs exhibited relatively large fold changes (log2FC in absolute value ≥ 5) among host species, accounting for 37% of flower and 53% of leaf OTUs (Tables S4, S5); this was likely attributable to many host species-specific OTUs and limited OTU overlapping among all three host species (Fig. 3b,c). The presence of non-significant, large-effect OTUs also suggested considerable variation in OTU abundances within host species given relatively small sample sizes, which perhaps limited the power to detecting OTU-level but not yet community-level differences among host species.
Sexual dimorphism is present in flower and leaf microbiomes
Sexual phenotype predicted α-diversity of microbial communities in flowers, but not in leaves and roots (Table 1). In flower microbiomes, sex comprised the second largest source of variation in Shannon diversity (16.4% of variation, F = 5.207, df = 1, P = 0.038; Table 1) and MPD (19.7%, F = 5.776, P = 0.030), although not in PD (4.8%, F = 0.943, P = 0.3). Compared to the main sex effect, species-specific sex effect on flower microbial α-diversity was even stronger, explaining the largest source of variation (Shannon diversity, 24.3%, F = 3.857, df = 2, P = 0.045; MPD, 27.5%, F = 4.030, P = 0.040). Specifically, males harbored higher microbial α-diversity than females in F.chilo (Shannon diversity, F = 5.207, df = 1, P = 0.038; MPD, F = 5.776, P = 0.030; Fig. S6) that also has the most pronounced sexual dimorphism (see Discussion); but in the other two species, inter-sexual differences were not significant (Fig. S6).
In contrast, microbial community β-diversity was not influenced by the main effect of sex (Table 1); this pattern was consistent across organ types, β-diversity metrics and statistical models. However, species-specific sex effects on community structures were observed in flower and leaf microbiomes by FWER-GLMs (Table 1), although such interaction effects failed to be captured by less sensitive PERMANOVA.
Because small sample sizes limited the detectability of differentially abundant OTUs, we focused on phylum-level variation in relative abundances between intraspecific male/hermaphrodite and female hosts for flower and leaf microbiomes. In flowers (Fig. 4), males/hermaphrodites harbored proportionately more Bacteroidetes (in all three host species, P < 0.001 in proportion tests) and less Proteobacteria (all P < 0.001) than females, after controlling for FDR (alpha = 0.05). For flower Actinobacteria, the relative abundances were also higher in males/hermaphrodites but only in two species (F.chilo and F.cunei, P < 0.001). In leaves (Fig. S7), sex differences in the three dominant phyla were variable among host species.
Our results support the hypothesis that organ type has a prominent role in structuring plant-associated microbiomes across host species1,4. In the three wild strawberries in their native environments, organ type not only predicts species and phylogenetic α-diversity of plant-associated microbiomes, but also explains the largest source of variation in compositional and phylogenetic β-diversity, while controlling for the effects of their broad-scale abiotic environments, host species and sex. In other words, the root microbiome of one host species is expected to be more similar to that of a different host species than to its own leaf and/or flower microbiome, potentially owing to niche-specific selection for adapted microbiota in different plant organs1,4,9.
In line with earlier studies on root microbiomes11, Proteobacteria (45%, relative abundance), Actinobacteria (22%) and Bacteroidetes (14%) are the dominant bacterial phyla in Fragaria, with Firmicutes to a lesser extent (6%). However, these previous studies often identified substantial influence of soil type on root microbiomes in both greenhouse and manipulated field settings, stronger than the effects of host genotypes6,42 and plant species11. By contrast, our study observed convergence in root microbiomes, despite distinct soil habitats where the three Fragaria species grow34.
The discrepancy with previous studies on the relation of root microbiomes with soil habitats likely has two reasons. First, root microbiomes quantified here comprise microbiota in association with absorptive fine roots (or first-order roots), as compared to those associated, for example, with the primary root11,42 or the whole root system6,10 in Arabidopsis and rice. It is possible that metabolically active root parts impose stronger filtering for colonizing microbes and thus cause stronger deviation from source soil microbiota, relative to metabolically inactive parts. Nevertheless, this hypothesis requires further investigation relating root traits to root microbiomes. In fact, several root samples in this study comprising old segments of fine roots, which were perhaps no longer metabolically active, formed a separate cluster different from the other root samples (Fig. 2a). Second, root microbiomes here did not consider low-frequency OTUs, owing to sequencing depth constraint and the use of normalized microbial community matrix and abundance-weighted diversity metrics. Although we cannot rule out the possibility that the low-frequency OTUs, which did not pass bioinformatic filtering or were not retrieved by sequencing, may indeed differ with soil habitats, their influence on microbial community structure and function remains an open question.
Fragaria leaf and flower microbiomes shared most of their OTUs with root microbiomes. This is in line with the idea that the sources of microbial assemblies in phyllosphere4 involve colonizing microbiota from soil by rain splash, wind and the visits of ground-dwelling herbivores and pollinators, as well as endophytes migrating from root to AG organs43. Consistent with grapevine AG and BG microbiomes in managed field settings12, we also found that leaf and flower microbiomes shared proportionally more OTUs with root microbiomes than with each other, indicating soil microbiota as a common species pool for microbial assemblies associated with different plant organs. Despite substantial OTU overlap, AG and BG microbiomes differed significantly in community structures, as has been observed in leaf and root microbiomes of Arabidopsis in both controlled and wild settings9,44. Such AG–BG microbiota differentiation in Fragaria was attributable to many depleted bacterial taxa in phyllosphere, especially Bradyrhizobium and Steroidobacter. These two bacterial genera were also found enriched in grapevine root microbiome (relative to AG microbiomes), where they probably mediate essential processes including nitrogen fixation in roots in vineyards12. The enriched taxa in Fragaria leaves and flowers such as Methylobacterium, Sphingomonas and Pseudomonas have been detected in other plant species, and their abilities to withstand more hostile habitats of phyllosphere have been implicated3,4.
Although in wild populations plant species–microbiota association can be influenced by abiotic environments, our study showed that host species differences in microbiomes were not just the by-product of their broad-scale variation in abiotic environments, but that host species explained a substantial amount of variation of organ-specific microbial communities, after accounting for that explained by abiotic environments.
The effect of Fragaria host species was strongest in flower microbiomes followed by leaf, and weakest in root microbiomes. Such enhanced host influence from root to leaf microbiomes has been detected in other plants at both genotype8 and species levels17. In common garden experiments of Boechera stricta8, phylogenetic community structure of leaf bacterial microbiota was clustered in accordance with host genotype but not for root microbiota. Similar to our finding of Fragaria species influence on bacterial microbiomes, host species was found as significantly affecting fungal endophytic communities in leaves but not in roots, among three common grass species in their wild populations17. If these patterns hold true across plant taxa, we may expect relaxed phylogenetic conservatism of plant microbiome traits (e.g., community structure) from BG to AG, owing to the paralleling phylogenetic signal of host plant functional traits.
Studies examining root and leaf traits45,46 have revealed stronger phylogenetic conservatism in roots, with root trait variation largely among deeply diverged plant lineages and little variation among species of low taxonomic ranks. The three Fragaria congeners in this study have a relatively short divergence history (originated ~1 Mya) and identical creeping herbaceous life form33, and thus they likely have similar fine root traits. This perhaps explains the resemblance of root microbiomes among host species in our study. Compared to root traits, interspecific differentiation in leaf traits (e.g., coriaceousness, leaf thickness) have been seen among the three Fragaria species in the wild (Fig. 1a) and greenhouse34; this may underlie the more heterogeneous leaf microbiomes relative to root microbiomes among host species. Relative to root and leaf traits, floral traits (e.g., color, size, scent and reward) are exceedingly diverse in plants47,48, and thus flower microbiomes are predicted to be distinct among host species3. Interestingly, although Fragaria flowers are morphologically similar in shape and color (Fig. 1a; see also ref.33), microbiome divergence among host species was yet strongest in flowers, suggesting that other floral traits such as scent, and pollen and nectar rewards might be critical in shaping species-specific flower microbiome. For plant genera with highly diversified floral traits, host species influence on flower microbiomes may be even stronger than what we observed in Fragaria.
Intriguingly, we found that the degree of sexual dimorphism in microbiomes coincided with the degree of sexual dimorphism in the three host species. In wild strawberries, F. chiloensis has the most complete separation of sexes than other Fragaria species, and has perhaps the greatest sexual differentiation in floral and other traits49. In F. chiloensis, male flowers are typically larger in petal size than female flowers49, which likely in part explains higher α-diversity of flower microbiomes in males. Although comparable studies on sexual dimorphism in plant microbiomes are lacking, it is noteworthy that culturable nectar-dwelling yeasts appeared to be higher in richness in male than female flowers of Silene latifolia in the wild24.
Relative to α-diversity, microbial community structure seems more sensitive to host sex, as many bacterial phyla were found differentially abundant between sexes in flower and leaf microbiomes across species. Sexually dimorphic leaf traits, which may underlie the observed differences in leaf microbiomes between sexes, have been detected in F. chiloensis and F. virginiana ssp. platypetala in common gardens (T-L Ashman and N Wei, unpubl. res.). For both species, males/hermaphrodites possess higher leaf nitrogen content and specific leaf area than females; but the degree of sexual dimorphism in leaf traits is still higher in F. chiloensis. Although similar data are not available for the hybrid F. ×ananassa ssp. cuneifolia, we suspect that it can be similar to F. virginiana ssp. platypetala considering their morphological resemblance34. Nevertheless, exactly how these sexually dimorphic traits affect microbiomes is not clear, but deserves additional research.
To conclude, our study provides the first characterization of microbiomes associated with the close wild relatives of the cultivated strawberry. We show, for the first time, enhanced host species influence and sexual dimorphism from root to flower microbiomes in wild populations. While these findings await similar investigations to generalize how plants control microbiota assemblies in the wild, it is important to recognize that such patterns of host species–microbiota association in situ affect plant interactions with AG and BG environments and plant fitness. Moreover, our results of sex-differential microbiota expand the understanding of sexual dimorphism in plants, and also highlight the needs for future research on the underlying mechanisms and on relating these differences to sex-specific fitness. Overall, findings from the wild, like ours here, strengthen those from experimental settings, and together they have broad implications for understanding this extended phenotype of plants.
Fragaria microbiome collection
The three Fragaria species (F. chiloensis, F. virginiana ssp. platypetala and F. ×ananassa ssp. cuneifolia) are widely distributed in western North America50. In Oregon, where they occur in sympatry but not in the same microhabitats34, we collected microbiota samples from seven wild populations over a 6-day period in May 2016: two populations of F. chiloensis (Salishan, ‘SAL’: 44.919°N, 124.027°W; Strawberry Hill, ‘SH’: 44.254°N, 124.112°W); two of F. virginiana ssp. platypetala (Willamette National Forest, ‘WNF’: 44.638°N, 121.941°W; Fisherman’s Bend Recreation, ‘FBR’: 44.755°N, 122.515°W); three of the natural hybrid F. ×ananassa ssp. cuneifolia (Marys Peak, ‘MP1’: 44.497°N, 123.546°W; ‘MP2’: 44.507°N, 123.569–123.579°W; Corvallis, ‘COR’: 44.506°N, 123.285°W). From each population, we randomly selected two female and two male-fertile (male or hermaphrodite) plants that were at least 2 m apart from each other, and collected root, leaf and flower samples from each plant. However, at COR we only sampled roots and leaves from three plants, as this population passed flowering. In total, our collection comprised 78 samples for the three species with organ type of root (N = 27), leaf (N = 27) and flower (N = 24).
For each plant, one flower (~2.5 cm in diameter), and the central leaflet (~2 cm in width) of a healthy trifoliate leaf that showed no evidence of herbivory and pathogen infection, were collected separately using ethanol-rinsed forceps and put directly into 750 µL Xpedition Lysis/Stabilization Solution in a ZR BashingBead Lysis tube (Zymo Research, Irvine, CA). Roots of the same plant were unearthed by the assisting person with ethanol-rinsed gloves, and were shaken vigorously to remove attached soil. Then five segments (~5 cm in length each) of fine roots, including some rhizosphere soil particles, were severed using sterile forceps and stored in the same manner. These samples were transferred to a −20 °C freezer within six hours after field collection, and shipped with dry ice to the University of Pittsburgh for DNA extraction.
Our leaf and flower samples contained both epiphytic and endophytic microbiota. The root samples also included some rhizosphere microbiota, in addition to rhizoplane and endosphere microbiota. For simplicity, we refer to these organ-associated microbiomes as root, leaf and flower microbiomes.
DNA extraction and 16S rRNA amplicon sequencing
Samples were homogenized using a TissueLyser II (QIAGEN, Germantown, MD), and genomic DNA was extracted using Xpedition Fungal/Bacterial DNA MiniPrep (Zymo Research) under a sterile laminar flow hood. The same extraction procedure was conducted on three negative controls without plant samples. The amplification and sequencing of 16S rRNA (the V4 region) were performed at the Environmental Sample Preparation and Sequencing Facility at Argonne National Laboratory. In brief, the V4 region was amplified using 515F-806R primer pair following the Earth Microbiome Project protocol (http://www.earthmicrobiome.org/protocols-and-standards/16s/) with 12-bp barcodes. Peptide nucleic acid (PNA) clamps designed from Arabidopsis thaliana36 were added in amplification to reduce Fragaria plastid contamination. The three negative controls failed in PCRs and generated primarily primer dimers. The 16S rRNA amplicons of the 78 samples were sequenced using a 1/5 lane of 2 × 151 bp on an Illumina MiSeq instrument.
OTU profiling and filtering
Paired-end reads were first joined using PEAR v0.9.651 with an overlap size of ≥20 bp. The successfully merged reads were used for subsequent open-reference operational taxonomic unit (OTU) picking. Sequence demultiplexing and quality filtering (with Phred quality scores of ≥20) were performed using QIIME v1.9.152. The resulting sequences were clustered into OTUs based on a similarity threshold of ≥97% by PyNAST and assigned with taxonomic identification by RDP classifier based on the Greengenes reference database (13_8 release), as implemented in QIIME. After chimera removal using QIIME ChimeraSlayer, aligned OTU representative sequences were used to build a midpoint-rooted phylogenetic tree of these OTUs using QIIME FastTree.
The QIIME-generated OTU table was further filtered before the conversion into a microbial community matrix. First, we filtered out OTUs belonging to chloroplasts and mitochondria. Second, we removed the singletons as well as low-frequency OTUs that accounted for ≤0.01% of the total observations of the entire OTU table. Third, we removed low-depth samples of <100 observations (N = 6, five leaf and one flower samples). Fourth, we normalized the OTU table to the same observations, which were the product of the median per-sample depth and per-sample OTU proportions (or relative abundances)37. The resultant normalized OTU table was used as the microbial community matrix for downstream statistical analyses, because normalization using alternative per-sample depths (e.g., mean or maximum depth) and raw OTU table did not affect the results (data not shown).
Abiotic environments of sampled populations
To account for abiotic effects on microbial communities, we used seven PRISM climatic variables35 of the current (1981–2010) conditions at 30-arcsec resolution, and elevation data, for the seven sampled populations. The seven annual climatic variables include temperature (mean, minimum and maximum), mean dewpoint temperature, precipitation and vapor pressure deficit (minimum and maximum). We conducted a principal component analysis (PCA) of these variables, including elevation, using prcomp() in R v3.3.353. The first two principal components (denoted as PC1.clim and PC2.clim) were taken as the abiotic predictors in the following statistical models.
Statistical analyses of microbial community α-diversity
Species and phylogenetic α-diversity metrics considered Shannon diversity, Faith’s phylogenetic diversity (PD) and abundance-weighted mean phylogenetic distance (MPD), which were calculated using the R package vegan54 and picante55. These α-diversity metrics were transformed (i.e., loge(PD), MPD3) to improve normality, and used as response variables in general linear mixed models (LMMs) using the package lme456. The fixed effects included PC1.clim + PC2.clim + Species + Sex + Organ + Species:Sex + Species:Organ + Sex:Organ; the random effect included individual plants. We did not include populations in random effects for two reasons: first, models that incorporated nested random effects failed to converge given the sample size; second, individuals also captured some of the population variation. For the main effect of each predictor and their interactions, the least-squares means (LS-means) were estimated using the package lmerTest57, and the statistical significance was evaluated by Type III sums of squares (SS). When considering organ type separately, we subdivided the microbial community matrix by organ type and re-estimated α-diversity metrics for organ-specific microbial communities. General linear models (LMs) were fitted with PC1.clim + PC2.clim + Species + Sex + Species:Sex, in which the LS-means and Type III SS were estimated using the package phia58 and car59, respectively.
Statistical analyses of microbial community β-diversity
Compositional and phylogenetic β-diversity metrics considered Bray–Curtis dissimilarity (in vegan), inter-community MPD (betaMPD in picante) and abundance-weighted UniFrac distance in the package GUniFrac60. Visualization of β-diversity metrics used the nonmetric multidimensional scaling (NMDS) in vegan for Bray–Curtis dissimilarity, and principal coordinates analyses (PCoAs) by cmdscale() for UniFrac distance and betaMPD.
These β-diversity metrics were taken as response variables in permutational multivariate analyses of variance (PERMANOVA) using vegan adonis2(). To assess the statistical significance (i.e., the marginal, instead of sequential, effect) of each main effect (or main term), PERMANOVA included PC1.clim + PC2.clim + Species + Sex + Organ. To assess the marginal effect for each interaction term, PERMANOVA included both the above main effects and their interaction terms (Species:Sex + Species:Organ + Sex:Organ). As a complement to distance-based PERMANOVA, generalized linear models (GLMs) with negative binomial errors were conducted using the package mvabund39 to assess how community structures changed in response to the main and interaction terms. The marginal effect of each term was assessed by nested model comparison between a full model and a reduced model with the focal term removed using a likelihood ratio test. OTUs that responded significantly to each model term were identified using univariate likelihood ratio tests with P values adjusted by resampling-based multiple testing implemented in mvabund to control for the family-wise error rate (FWER; alpha = 0.05). Here we referred to mvabund GLMs as FWER-GLMs.
PERMANOVA and FWER-GLMs were also conducted to model microbial community β-diversity for each organ type separately. Visualization of organ-specific β-diversity metrics was performed using constrained PCoAs by vegan capscale().
Differentially abundant OTUs among microbial communities
As a complement to the univariate tests of individual OTU abundances in mvabund, we used the package edgeR41 to estimate the effect size (log2 fold change) and sign (depleted or enriched) of each differentially abundant OTU attributed to individual predictors and their interactions, as well as between different levels within a predictor. Similar to mvabund, edgeR also uses GLMs with negative binomial errors, but it models individual OTU abundances with false discovery rate control (FDR; alpha = 0.05) for multiple testing. Here we referred to edgeR GLMs as FDR-GLMs. FDR-GLMs allow a design matrix accommodating complex experimental structure. Our design matrix followed PC1.prism + PC2.prism + 0 + Group, in which Group contained all combinations of different levels of predictors (Species, Sex and Organ). The model was fitted using glmQLFit() and specific contrasts were made by glmQLFTest(). The P values were adjusted by the Benjamini–Hochberg (BH) correction using p.adjust(). We also conducted differential analyses for organ-specific microbial communities to detect differentially abundant OTUs between species within each organ type.
Raw reads are available from National Center for Biotechnology Information (PRJNA434446).
Müller, D. B., Vogel, C., Bai, Y. & Vorholt, J. A. The plant microbiota: systems-level insights and perspectives. Annu. Rev. Genet. 50, 211–234 (2016).
Bordenstein, S. R. & Theis, K. R. Host biology in light of the microbiome: ten principles of holobionts and hologenomes. PLoS Biol. 13, e1002226 (2015).
Aleklett, K., Hart, M. & Shade, A. The microbial ecology of flowers: an emerging frontier in phyllosphere research. Botany 92, 253–266 (2014).
Vorholt, J. A. Microbial life in the phyllosphere. Nat. Rev. Microbiol. 10, 828–840 (2012).
Bulgarelli, D., Schlaeppi, K., Spaepen, S., Ver Loren van Themaat, E. & Schulze-Lefert, P. Structure and functions of the bacterial microbiota of plants. Annu. Rev. Plant Biol. 64, 807–838 (2013).
Edwards, J. et al. Structure, variation, and assembly of the root-associated microbiomes of rice. Proc. Natl. Acad. Sci. USA 112, E911–E920 (2015).
Peiffer, J. A. et al. Diversity and heritability of the maize rhizosphere microbiome under field conditions. Proc. Natl. Acad. Sci. USA 110, 6548–6553 (2013).
Wagner, M. R. et al. Host genotype and age shape the leaf and root microbiomes of a wild perennial plant. Nat. Commun. 7, 12151 (2016).
Bai, Y. et al. Functional overlap of the Arabidopsis leaf and root microbiota. Nature 528, 364–369 (2015).
Lundberg, D. S. et al. Defining the core Arabidopsis thaliana root microbiome. Nature 488, 86–90 (2012).
Schlaeppi, K., Dombrowski, N., Oter, R. G., Ver Loren van Themaat, E. & Schulze-Lefert, P. Quantitative divergence of the bacterial root microbiota in Arabidopsis thaliana relatives. Proc. Natl. Acad. Sci. USA 111, 585–592 (2014).
Zarraonaindia, I. et al. The soil microbiome influences grapevine-associated microbiota. mBio 6, e02527–02514 (2015).
Kembel, S. W. et al. Relationships between phyllosphere bacterial communities and plant functional traits in a neotropical forest. Proc. Natl. Acad. Sci. USA 111, 13715–13720 (2014).
Redford, A. J., Bowers, R. M., Knight, R., Linhart, Y. & Fierer, N. The ecology of the phyllosphere: geographic and phylogenetic variability in the distribution of bacteria on tree leaves. Environ. Microbiol. 12, 2885–2893 (2010).
Junker, R. R. & Keller, A. Microhabitat heterogeneity across leaves and flower organs promotes bacterial diversity. FEMS Microbiol. Ecol. 91, fiv097 (2015).
Ottesen, A. R. et al. Baseline survey of the anatomical microbial ecology of an important food plant: Solanum lycopersicum (tomato). BMC Microbiol. 13, 114 (2013).
David, A. S., Seabloom, E. W. & May, G. Plant host species and geographic distance affect the structure of aboveground fungal symbiont communities, and environmental filtering affects belowground communities in a coastal dune ecosystem. Microb. Ecol. 71, 912–926 (2016).
Coleman-Derr, D. et al. Plant compartment and biogeography affect microbiome composition in cultivated and native Agave species. New Phytol. 209, 798–811 (2016).
de Vries, F. T. et al. Abiotic drivers and plant traits explain landscape-scale patterns in soil microbial communities. Ecol. Lett. 15, 1230–1239 (2012).
Burns, J. H., Anacker, B. L., Strauss, S. Y. & Burke, D. J. Soil microbial community variation correlates most strongly with plant species identity, followed by soil chemistry, spatial location and plant genus. AoB Plants 7, plv030–plv030 (2015).
Zimmerman, N. B. & Vitousek, P. M. Fungal endophyte communities reflect environmental structuring across a Hawaiian landscape. Proc. Natl. Acad. Sci. USA 109, 13022–13027 (2012).
Markle, J. G. et al. Sex differences in the gut microbiome drive hormone-dependent regulation of autoimmunity. Science 339, 1084–1088 (2013).
Dominianni, C. et al. Sex, body mass index, and dietary fiber intake influence the human gut microbiome. PLoS One 10, e0124599 (2015).
Golonka, A. M. & Vilgalys, R. Nectar inhabiting yeasts in Virginian populations of Silene latifolia (Caryophyllaceae) and coflowering species. Am. Midl. Nat. 169, 235–258 (2013).
Vega-Frutis, R., Munguia-Rosas, M. A., Varga, S. & Kytoviita, M. M. Sex-specific patterns of antagonistic and mutualistic biotic interactions in dioecious and gynodioecious plants. Perspect. Plant Ecol. Evol. Syst. 15, 45–55 (2013).
Barrett, S. C. & Hough, J. Sexual dimorphism in flowering plants. J. Exp. Bot. 64, 67–82 (2013).
Ashman, T.-L., Swetz, J. & Shivitz, S. Understanding the basis of pollinator selectivity in sexually dimorphic Fragaria virginiana. Oikos 90, 347–356 (2000).
Ashman, T.-L., Bradburn, M., Cole, D. H., Blaney, B. H. & Raguso, R. A. The scent of a male: The role of floral volatiles in pollination of a gender dimorphic plant. Ecology 86, 2099–2105 (2005).
Cornelissen, T. & Stiling, P. Sex-biased herbivory: a meta-analysis of the effects of gender on plant-herbivore interactions. Oikos 111, 488–500 (2005).
Krimm, U., Abanda-Nkpwatt, D., Schwab, W. & Schreiber, L. Epiphytic microorganisms on strawberry plants (Fragaria ananassa cv. Elsanta): identification of bacterial isolates and analysis of their interaction with leaf surfaces. FEMS Microbiol. Ecol. 53, 483–492 (2005).
Kaltz, O. & Shykoff, J. A. Male and female Silene latifolia plants differ in per-contact risk of infection by a sexually transmitted disease. J. Ecol. 89, 99–109 (2001).
Ashman, T.-L. The role of herbivores in the evolution of separate sexes from hermaphroditism. Ecology 83, 1175–1184 (2002).
Liston, A., Cronn, R. & Ashman, T.-L. Fragaria: A genus with deep historical roots and ripe for evolutionary and ecological insights. Am. J. Bot. 101, 1686–1699 (2014).
Salamone, I. et al. Bioclimatic, ecological, and phenotypic intermediacy and high genetic admixture in a natural hybrid of octoploid strawberries. Am. J. Bot. 100, 939–950 (2013).
PRISM Climate Group. Oregon State University. http://prism.oregonstate.edu (2004).
Lundberg, D. S., Yourstone, S., Mieczkowski, P., Jones, C. D. & Dangl, J. L. Practical innovations for high-throughput amplicon sequencing. Nat. Methods 10, 999–1002 (2013).
McMurdie, P. J. & Holmes, S. Waste not, want not: why rarefying microbiome data is inadmissible. PLoS Comp. Biol. 10, e1003531 (2014).
Miller, E. T., Farine, D. R. & Trisos, C. H. Phylogenetic community structure metrics and null models: a review with new methods and software. Ecography 40, 461–477 (2017).
Wang, Y., Naumann, U., Wright, S. T. & Warton, D. I. mvabund – an R package for model-based analysis of multivariate abundance data. Methods Ecol. Evol. 3, 471–474 (2012).
Warton, D. I., Wright, S. T. & Wang, Y. Distance-based multivariate analyses confound location and dispersion effects. Methods Ecol. Evol. 3, 89–101 (2012).
Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).
Bulgarelli, D. et al. Revealing structure and assembly cues for Arabidopsis root-inhabiting bacterial microbiota. Nature 488, 91–95 (2012).
Chi, F. et al. Ascending migration of endophytic rhizobia, from roots to leaves, inside rice plants and assessment of benefits to rice growth physiology. Appl. Environ. Microbiol. 71, 7271–7278 (2005).
Bodenhausen, N., Horton, M. W. & Bergelson, J. Bacterial communities associated with the leaves and the roots of Arabidopsis thaliana. PLoS One 8, e56329 (2013).
Weemstra, M. et al. Towards a multidimensional root trait framework: a tree root review. New Phytol. 211, 1159–1169 (2016).
Valverde-Barrantes, O. J., Smemo, K. A., Blackwood, C. B. & Norden, N. Fine root morphology is phylogenetically structured, but nitrogen is related to the plant economics spectrum in temperate trees. Funct. Ecol. 29, 796–807 (2015).
Stebbins, G. L. Adaptive radiation of reproductive characteristics in angiosperms, I: Pollination mechanisms. Annu. Rev. Ecol. Evol. Syst. 1, 307–326 (1970).
Fenster, C. B., Armbruster, W. S., Wilson, P., Dudash, M. R. & Thomson, J. D. Pollination syndromes and floral specialization. Annu. Rev. Ecol. Evol. Syst. 35, 375–403 (2004).
Ashman, T.-L., Spigler, R. B., Goldberg, M. & Govindarajulu, R. In New insights on plant sex chromosomes (ed Navajas-Pérez, R.) 67–90 (Nova Science Publishers, 2012).
Staudt, G. Systematics and geographic distribution of the American strawberry species: Taxonomic studies in the genus Fragaria (Rosaceae: Potentilleae). (University of California Press, 1999).
Zhang, J. J., Kobert, K., Flouri, T. & Stamatakis, A. PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics 30, 614–620 (2014).
Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336 (2010).
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (2017).
Oksanen, J. et al. vegan: Community Ecology Package. R package version 2.4–3. https://CRAN.R-project.org/package=vegan (2017).
Kembel, S. W. et al. Picante: R tools for integrating phylogenies and ecology. Bioinformatics 26, 1463–1464 (2010).
Bates, D., Machler, M., Bolker, B. M. & Walker, S. C. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48 (2015).
Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest: tests in linear mixed effects models. R package version 2.0–33. https://CRAN.R-project.org/package=lmerTest (2016).
Rosario-Martinez, H. D. phia: Post-Hoc Interaction Analysis. R package version 0.2–1. https://CRAN.R-project.org/package=phia (2015).
Fox, J. & Weisberg, S. An R companion to applied regression. Second edn (SAGE, 2011).
Chen, J. GUF: Generalized UniFrac distances. R package version 1.0. https://CRAN.R-project.org/package=GUniFrac (2012).
We thank Chris Marshall for generous help with QIIME pipeline, Aaron Liston for providing population locations and Anna Freundlich for assistance with sampling, and the Ashman lab members for discussion, and two anonymous reviewers and the editor for comments that improved the manuscript. This work was supported by the National Science Foundation (DEB 1241006) and UPitt Dietrich School of Arts and Sciences.
The authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wei, N., Ashman, TL. The effects of host species and sexual dimorphism differ among root, leaf and flower microbiomes of wild strawberries in situ. Sci Rep 8, 5195 (2018). https://doi.org/10.1038/s41598-018-23518-9
This article is cited by
Honeybees affect floral microbiome composition in a central food source for wild pollinators in boreal ecosystems
Microbial assemblages of Schisandraceae plants and the correlations between endophytic species and the accumulation of secondary metabolites
Plant and Soil (2023)
Insights into the microbiome assembly during different growth stages and storage of strawberry plants
Environmental Microbiome (2022)
Plant Gender Affects Soil Fungal Microbiota Associated with Welwitschia mirabilis, an Unusual Desert Gymnosperm
Microbial Ecology (2022)
Alterations of endophytic microbial community function in Spartina alterniflora as a result of crude oil exposure
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.