Horsenettle (Solanum carolinense) fruit bacterial communities are not variable across fine spatial scales

Introduction

Plants, like other organisms, host diverse communities of microorganisms that are typically dominated by bacteria (Stone, EA & Jackson, 2018). Bacteria associated with above-ground portions of plants (the phyllosphere) play an important role in host development and health (Gibbons & Gilbert, 2015; Grover et al., 2011; Richardson et al., 2009; Selosse, Baudoin & Vandenkoornhuyse, 2004). Thus, understanding how these bacterial communities are formed and the factors that shape changes in these communities can lead to important insights about plant health. This is especially true when we consider tissues like fruits, which are directly tied to plant fitness (Compant et al., 2011; Eck et al., 2019; Nelson, 2018; Tewksbury et al., 2008). Despite this importance, few studies have examined patterns of bacterial community diversity among fruits (Compant et al., 2011; Miura et al., 2017).

A first step in understanding what shapes plant-associated microbial communities is characterizing variation over time and space. The tendency for communities to become increasingly dissimilar with distance is often referred to as “distance decay” (Finkel et al., 2012; Morrison-Whittle & Goddard, 2015; Peay, Garbelotto & Bruns, 2010), which have been observed in microbial communities (Feng et al., 2019). Viewed through an evolutionary lens, this pattern of community turnover could arise from two distinct mechanisms: local adaptation and dispersal limitation (Bell, 2010). In the first case, communities change concordantly with niche turnover along an abiotic or biotic gradient. In the latter case, community change occurs in the absence of an underlying environmental or resource gradient, exhibiting geographical patterns (Barreto et al., 2014). Studies of phyllosphere microbes have found mixed evidence for these mechanisms. For instance, a recent study of Tamarix leaves found that bacterial community structure was influenced by both geographic distance and environmental heterogeneity (Finkel et al., 2012). Yet, it is challenging to find generalizable patterns, as most studies of phyllosphere communities focus on leaves over large spatial scales, maximizing the potential to detect community differences, even though not all studies find evidence of distance decay. For example, a study of Pinus spp. at distances ranging from 1,800–2,400 km found no evidence that microbial communities became more distinct with increasing distance (Redford et al., 2010). In contrast, a recent study of Magnolia grandiflora leaves at distances up to 452 m found that microbial communities did become more distinct the further away they were (Stone & Jackson, 2016).

While widespread, bacteria may not be evenly distributed over large spatial scales (Bahram et al., 2018; Delgado-Baquerizo et al., 2018). Microbial communities on plants can be regulated by local biotic and abiotic conditions, including plant host species, geographic location, light, temperature, wind, and humidity (Berg et al., 2014; De Vries et al., 2012). In addition, the factors that shape bacterial distribution at global scales, such as soil pH and temperature (Bahram et al., 2018; Delgado-Baquerizo et al., 2018), may be more nuanced or attenuated at local levels (Igwe & Vannette, 2019; Storey et al., 2018). Further, the factors shaping microbial diversity may interact across scales to influence community assembly within particular communities, e.g., rhizosphere versus leaves (Levin, 1992). For example, using genetic mutants, Lebeis et al. (2012) demonstrated that Arabidopsis plants are capable of structuring soil bacterial communities. In contrast, Ottesen et al. (2016) found that in the tomato phyllosphere when host and other biotic factors were ruled out using plastic plants, bacterial communities were similar between the plastic plants and normal controls.

Studies at finer spatial scales are more limited (Stone & Jackson, 2016), and even fewer studies have been conducted in fruit microbial communities (Miura et al., 2017). Fruit microbial communities are likely shaped by a combination of biotic and abiotic factors similar to those that shape the leaf microbial communities. Yet, recent studies have also shown that fruit and flower microbial communities differ from other above ground plant tissues (Ottesen et al., 2013; Rasmann et al., 2012; Shade, McManus & Handelsman, 2013). One possible explanation for this is that these tissues interface directly with pollinators, biotic sources of additional microbes (Russell et al., 2019), especially in obligate outcrossing species (Wei & Ashman, 2018). Since pollinators transfer microbes from one flower to the next (Ushio et al., 2015), they may contribute to homogenizing microbial communities locally at these finer spatial scales.

Fruit and flowers are especially warranted to studies at finer spatial scales, as pollinator interactions are most likely to impact these tissues (Rebolleda-Gómez et al., 2019; Ushio et al., 2015; Zemenick, Vannette & Rosenheim, 2021). Recent studies found that plant communities with a similar floral visitor community have similar microbes (Rebolleda-Gómez & Ashman, 2019; Zemenick et al., 2019). However, they found that while flowers were hubs for arthropod visitation, flowers did not necessarily serve as hubs for microbes, which suggests a degree of host sorting (Zemenick, Vannette & Rosenheim, 2021) which supports the idea that floral visitors shape microbial communities. However, little is known regarding how this could impact fruit microbial communities.

Here, we hypothesized that bacterial communities associated with fruit that are spatially closer to one another will be more similar. This is because they are more likely to share a similar environment and be exposed to the same pollinator, since pollinators are known to visit multiple flowers. To address this hypothesis, we sampled the fruit of Carolina horsenettle, Solanum carolinense, a native herbaceous perennial weed that is an obligate outcrossing species, at a fine scale (under 450 m) and compared bacterial community composition at different distances to determine if we observed distance decay in the bacterial communities associated with fruit.

Materials & Methods

Sampling, processing, and DNA extraction

A total of 23 fruit samples, from nine different sampling locations of S. carolinense, were collected from around Blacksburg, VA, USA on November 15th, 2017 (Fig. 1, Table S1). The use of the site was approved by the Department of Parks and Recreation of the town of Blacksburg. Sample locations ranged in distance from 2 to 450 m apart, and the total number of individual plants sampled ranged from 2 to 5 individuals/site. Since, S. carolinense is a perennial species that can spread via underground rhizomes generating clusters of clonal plants it is possible that multiple of S. carolinense in a collection site are genetically identical. One fruit sample per S. carolinense ramet was collected aseptically in 50 ml falcon tubes and stored at −80 °C until DNA extraction. Prior to DNA extraction, samples were sliced using EtOH flame sterilized forceps and scalpels so that the sample included both surface and internal microbes. The samples were then disrupted with beads. DNA was extracted from samples using the DNeasy PowerSoil^® Kit (Qiagen USA catalog no. 12888-100).

Results

To describe the composition of the bacterial communities associated with fruit tissue from each sampling location, we estimated relative abundances of different bacterial orders (Fig. 2). The orders Enterobacteriales, Sphingomonadales, Sphingobacteriales, Rhizobiales, Cytophagales, and Micrococcales were found to make up the greatest proportion of the relative abundance found at each sampling site (Fig. 2). Interestingly, some differences in abundance were found across sites. The orders Micrococcales, Caulobacterales, and Betaproteobacteriales were present in some sites and not found at others. Micrococcales was found at sites 2, 4, and 5, and Caulobacterales was found at sites 2, 5, and 7. Both Micrococcales and Caulobacterales were found at relatively low abundances compared to other prominent taxa. Betaproteobacteriales was found at relatively low abundances at all the sites except site 1. Flavobacteriales was also found at relatively low abundance but found at all the sites sampled. The most common bacterial genera by relative abundance were: Aureimonas, Chryseobacterium, Dyadobacter, Hymenobacter, Methlobacterium, Mucilaginibacter, Neorhizobium, Pantoea, Pedobacter, Pseudomonas, Rosenbergiella, Sphingomonas, Spirosoma, and Tatumella; these all had a relative abundance of ≥ 0.1%.

Bacterial community analysis

Shannon diversity, observed richness, and Faith’s phylogenetic diversity were then calculated for each sample location using abundance data to determine richness and evenness (Fig. 3). Although there was some separation of mean values in alpha diversity across sites, none of these differences were significant for observed richness (p = 0.2173, DF = 7, chi-squared = 9.5226) or Faith’s phylogenetic diversity via Kruskal–Wallis rank sum test (p = 0.1879, DF = 7, chi-squared = 10.012). The high values observed in mean alpha diversity likely result from high variation within sites (Fig. 1).

Similarly, Bray–Curtis dissimilarity of S. carolinense fruit did not reveal any differences in bacterial community structure across sampling locations (PERMANOVA, p = 0.711, DF = 22, R² = 0.3033). PCoA analysis revealed that the first two components of variation explained a combined 29.1% (18.3% and 10.8% respectively) of the variation in bacterial community composition, however we did not find any significant clustering of samples by site (Fig. 4). For example, samples from the same location (color in Fig. 4) tended to be dispersed across the axes of variation. Likewise, no differences in bacterial community composition among sampling locations was identified using the Jaccard index (Figure S1, PERMANOVA, p = 0.688, R² = 0.30896). These findings were supported by a Mantel test of Bray–Curtis dissimilarity index and calculated spatial distances that failed to detect a relationship between bacterial diversity and distance (Fig. 5 Pearson’s, Mantel statistic r: 0.03343, p = 0.302, DF = 22). Together, these results suggest that fruit bacterial diversity is not differentiated at this spatial scale.

To investigate the proportion of shared taxa among samples from all sites, we conducted an analysis of the core microbial community and found that just 25 ASVs were present in at least 60% of the samples with a relative abundance > 0.10% (Fig. 6). The top five taxa that dominated the core microbial communities, as determined by a relative abundance of > 1.0% and prevalence > 50% were Aureimonas spp., Pantoea spp., Sphingomonas spp., Hymenobacter spp., and Pedobacter spp. Thus, 0.0135% of ASVs we identified in our analysis were found in the majority of fruit and less than half of these were found with a relative abundance > 0.30%. Finally, we found that Aureimonas spp. persisted with a very high prevalence >90% at relative abundance levels up to 1.0%.

Discussion

The advent of non-culture based microbial community profiling has led to a wealth of information about patterns and scales of community diversity among bacteria that form close associations with plant organs (Compant et al., 2011; Junker et al., 2011; Ottesen et al., 2016; Shade, McManus & Handelsman, 2013). Non-pathogenic fruit-associated bacteria have received less attention, yet these communities have important implications for host fitness. This study provides a first glimpse into the composition of fruit bacterial communities of S. carolinense. This study also is the first to explore patterns of bacterial community structure of fruit in natural populations at fine spatial scales. This is one of the first studies to be conducted in fruit at such a fine spatial scale. We found that bacterial community richness and diversity are similar at distances up to 450m, and there is no correlation between community structure and distance, suggesting no evidence of distance-decay at these fine spatial scales. Factors such as microenvironment or biotic interactions (pollinators, seed predators, etc.) may, therefore, be more important for shaping patterns of bacterial diversity among fruit. It is also possible that a lack of genetic diversity among host plants (S. carolinense can and does reproduce clonally through rhizomes) could account for a lack of distance decay. However, because rhizome growth is generally limited to about 1.25 m (Kiltz, 1930) from the ramet, we would expect to reduce variation at the site level (Imaizumi et al., 2006), which was not apparent in this study. Still, the role of host-genotype level filtering in shaping fruit microbial communities in this system remains an open question.

As with other studies, we found that fruits are colonized by a diverse array of bacteria. In our study, we found that the Enterobacteriales, Rhizobiales, and Cytophagales orders comprised about 63% of the relative abundance across all samples. Within these orders, the most frequently identified genera were Pantoea, Aurantimonas, and Methylobacterium. In other plant hosts, different genera seem to dominate. For instance, fruits of congeneric tomatoes (Solanum lycopersicum) are dominated by the genera Xanthomonas, Rhizobium, and Pseudomonas (Ottesen et al., 2013); the only overlapping genus identified in our study was Sphingomonas. These findings reinforce the importance of host species (Igwe & Vannette, 2019; Knief et al., 2010; Ottesen et al., 2016; Wei & Ashman, 2018) and environment (Knief et al., 2010) in shaping bacterial communities in fruit. Despite the substantial variation in diversity, we identified 25 ASVs that were present in a majority of samples (>60%) (Fig. 6). Subsetting these further we identified five taxa that were present in nearly every sample with a relative abundance > 0.10%, Aureimonas spp., Pantoea spp, Sphingomonas spp., Hymenobacter spp., and Pedobacter spp. These top five taxa have all previously been reported from different plant tissue samples. Aureimonas spp. have been identified in the phyllosphere of Galium album (Aydogan et al., 2016), the bark of Populus spp. (Li et al., 2018), and, recently in Actinidia deliciosa (Ares et al., 2021). Pantoea spp. have been identified as endophytes in the stems of tomatoes, Solanum lycopersicum (Dong et al., 2019). Pantoea spp. can also cause internal fruit rot in Cucurbitaceae (Kido et al., 2008). Sphingomonas spp. have been found in tomato specifically on the lower stems and leaves (Ottesen et al., 2013), can cause disease in Cucurbitaceae (Buonaurio, Stravato & Cappelli, 2001), and can be endophytes that produce gibberellins (Latif Khan et al., 2014). Hymenbobacter spp. have been observed on Hedera spp. (Smets et al., 2016) and in the phyllosphere of Galium album (Aydogan et al., 2016). Pedobacter spp. have been observed in the phyllosphere of Solanum tuberosum, which is closely related to our study species (Manter et al., 2010), as well as the leaves of Arabidopsis thaliana (Qi et al., 2021).

Comparing bacterial community diversity, we found a great deal of variation within and among sites (Figs. 3 and 4). This variation was not partitioned by site or distance (Fig. 5) and is consistent with diverse source pools and environmental filtering (Rebolleda-Gómez & Ashman, 2019). It is important, however, to consider these findings with the caveat that reduced sample sizes negatively impact alpha diversity statistics (Willis, 2019). Thus, while widespread, bacterial taxa may not be evenly distributed over large spatial scales (Bahram et al., 2018; Delgado-Baquerizo et al., 2018).

Environment is an important source and factor shaping leaf microbial communities across fine spatial scales. This has been observed by Stone & Jackson (2016) found that bacterial communities associated with the leaves of Magnolia grandiflora exhibited a distance-decay relationship across a similar distance to the present study (1–452 m). Ultimately, Stone & Jackson (2016) conclude that subtle differences in environmental conditions contributed to differences in microbial communities. The lack of distance-decay in the present study may be for several reasons. Our sampling may not have covered a sufficient range nor elevation (17 m) in environmental variation however we did have variation in slope and aspect (north and west facing) (Fig. 1).

Host species and genotype may also play a role in shaping microbial communities associated with the plant (Laforest-Lapointe, Messier & Kembel, 2016; Redford et al., 2010). However, there has been contrasting evidence in Solanum lycopersicum leaves, suggesting that the host plant is not different from a controlled plastic plant (Ottesen et al., 2016). In addition, plant organs may also play a role in the interactions shaping distance decay. Several studies have documented distinct microbial communities across plant tissues (Compant et al., 2011; Junker & Keller, 2015; Ottesen et al., 2013; Shade, McManus & Handelsman, 2013), including different tissues within the same organ (Hayes et al., 2021). This also means that microbial communities associated with particular plant organs may have a different set of organizing rules (Rebolleda-Gómez & Ashman, 2019; Wei & Ashman, 2018; Zheng & Gong, 2019). For instance, flowers and fruits that rely on pollinators, which spread microbes, may not display distance decay at finer spatial scales. Another important consideration is the production of secondary compounds within the fruit. As with other members of the Solanaceae S. carolinense fruit contain steroidal glycoalkaloids that likely impact microbial communities (Milner et al., 2011). Future studies can determine the relative impact of SGAs on microbial communities by isolating the endophytic component of fruit microbial communities. By understanding how and when microbial communities are shaped in fruits and flowers we can learn how this impacts host fitness (Baltrus, 2020; Taylor et al., 2014).

Dispersal is another important component acting across these smaller spatial scales. For instance, Miura et al. (2017) investigated leaves and fruits in vineyards in Chile that were within 35 km of one another and found that at spatial scales <2 km fungal community dissimilarity increased with distance; however, the bacterial communities did not show similar patterns. This may be because fungal spores tend to have more limited dispersal than bacteria (Peay & Bruns, 2014). However, over large distances (1,800–2,400 km) Pinus spp. leaves showed no variation with an increase in distance (Redford et al., 2010), yet there was a distance-decay effect observed in Tamarix spp. (Finkel et al., 2012). We are still in the early phases of understanding how microbial communities are shaped across the landscape, particularly in reproductive tissues. Biotic interactions can drive facilitated bacterial dispersal (Russell et al., 2019). As an obligate outcrossing species, S. carolinense depends on pollinators. Pollinators leave microbial footprints on the flowers that they visit (Ushio et al., 2015), which may influence floral microbial communities, and possibly fruit microbial communities. Thus, the observed similarities among bacterial communities in our study could result from pollinator transmission, though we do not yet know if bacterial communities are transferred from flowers to fruit in this system. In addition to pollinators, seed feeding insects, such as the Lygaeidae, pierce fruit and feed on the developing seeds. This process introduces microbes to fruits (Tewksbury et al., 2008) and may be another source of bacterial dispersal. Indeed, our clustering analysis identified some axes of separation for samples within a site, which could be indicative of these sorts of biotic filters, yet more work needs to be done to test this hypothesis.

Conclusions

In our study, we found that richness, diversity, and community structure of bacterial communities associated with the fruit of S. carolinense are similar at fine spatial scales, suggesting there is no evidence of a distance-decay relationship. Thus, we are left with the question at what point do bacterial communities become distinct in the fruit of Solanum carolinense? In order to address this question, samples will need to be taken at larger spatial scales to determine when these communities become distinct. We also know that environment, pollinator, and host species play a large role in shaping microbial communities (Igwe & Vannette, 2019; Knief et al., 2010; Ottesen et al., 2016; Ushio et al., 2015; Wei & Ashman, 2018), but the degree to which these factors shape the bacterial communities found in Solanum carolinense fruit remains unclear. To address this, sampling over a large spatial scale should be done across a variety of environmental conditions. From that sampling, a comparison of similar environments, microbial communities will help to address the degree that environment plays a role in shaping microbial communities and determine if spatial scale plays a larger role than environment. To address the role of host species in shaping microbial communities, research should be conducted to compare the genotype of host species within a site to determine if in fact hosts that are closer related are more similar with regards to their microbial communities.

Supplemental Information