• Open Access

Characterization of culturable bacterial endophytes of switchgrass (Panicum virgatum L.) and their capacity to influence plant growth



Switchgrass (Panicum virgatum L.) is a perennial warm season grass that is native to the plains of North America and is widely grown as a forage, bioenergy or groundcover crop. Despite its importance, a bottleneck in switchgrass production is poor seedling vigor, which as a perennial crop represents an important time for management. Herein, data identify a suite of culturable bacterial microflora extracted from switchgrass, and show their capability to influence host plant growth and development. A total of 307 bacterial isolates were cultured and isolated from surface sterilized switchgrass biomass and sequence identified into 76 strains (subspecies classification), 36 species and 5 phyla. Approximately 58% of bacterial strains, when reintroduced into surface-sterilized switchgrass seeds, were documented to increase lamina length (cm from base to tip after 60 days growth) relative to uninoculated controls. Ecologically, Phylum Firmicutes was the most abundant bacterial classification and encompassed 75% of all isolates. Although the culturable bacterial community studies herein represent an unknown and assumedly minor proportion of the total microbiome, by focusing on culturable bacteria, we delineate functional feedback between the presence of isolated bacteria and switchgrass seedling growth.


Endophytes are defined as microorganisms that successfully colonize and reside within plant tissue and do not cause visible harm to the host (Hallmann et al., 1997; Hardoim et al., 2008; Sturz et al., 2000). As mutualists, endophytes have been shown to elicit functional consequences on their plant host (Schardl et al., 2004), in terms of improved water and nutrient uptake (Pillay & Nowak, 1997), nitrogen fixation (Krause, 2006), assisting with herbivore resistance by producing toxic secondary metabolites (Siciliano, 2001), improving the capacity to resist disease (Choudhary & Johri, 2009) and by promoting plant growth and overall primary productivity (Compant et al., 2005). In addition, plant-endophyte associations have been found to confer partial tolerance to environmental toxins such as heavy metals and chemicals (Barac, 2004; Ma, 2011; Sturz & Christie, 1998). Bacterial endophytes have been isolated and identified from tubers (Sessitsch et al., 2004), stems, fruit, and seeds (Olivares et al., 1996). Considering their potential agro-ecological importance, increasing our understanding of the endophyte community structure at the species level is needed across plant taxonomic boundaries. Examination of bacteria present within a host plant can be achieved either by culture-dependent or culture-independent techniques. Culture independent techniques that employ next generation sequencing technology use bacterial specific probes to amplify target DNA (16S-rDNA) and to provide a depth of understanding of the total diversity. In many cases, however, the probes are limited to identifying unknown species, which remain elusive to study due to their lack of culturability.

Switchgrass (Panicum virgatum L.) is a perennial warm season grass that is native to the plains of North America. Despite its natural habitat being reduced to tiny pockets, it is widely used as a forage crop or as ground cover to control erosion (Blanco & Lal, 2008; Redfearn & Nelson, 2003; Vogel et al., 2002; Keene & Skousen, 2010). It has also been used extensively throughout the United States as a ground cover and in mine site reclamation (Comis, 2006; Teel et al., 2003). One of the emerging uses for switchgrass is as a biofuel crop on marginal or abandoned agricultural land (DeBolt et al., 2009). Vigor can be critical to the successful establishment of switchgrass, but less so if weed pressure is eliminated (Redfearn & Nelson, 2003). Establishment success, particularly defined as high first and second year yields is critical for positive economic outcomes in production settings (Redfearn & Nelson, 2003; Parrish & Fike, 2005). It remains unclear whether endophytic microbial associations are prevalent in switchgrass, and if so which bacteria are present. Of further consideration is whether or not bacterial colonizers could promote change in plant growth, particularly in terms of seedling vigor.

The goal of the present study was to determine whether or not culturable bacteria could be isolated from surface sterilized switchgrass (Panicum virgatum L.) and to subsequently document if isolated microflora were influencing switchgrass growth and development in isolation. Although using culture-dependent techniques to study the bacteria present in the endophytic community will not capture obligate bacteria that are unculturable on media, the main advantage is that diverse bacterial strains can be isolated, identified and characterized for their utility in terms of plant productivity. Therefore, this study focused on culturable bacterial community members as a first step in documenting switchgrass endophytic bacterial species.


Site assessment and plant selection on reclaimed coal strip mines

The western Kentucky coal fields are located within the native distribution region for switchgrass along with other warm season grass species. Around 300,000 ha of abandoned coal mines exist in Kentucky and these represent a target for reclamation and enhancing ecosystem services. Warm season grasses such as switchgrass have frequently been used for reclamation. Herein, we identified two sites (referred to as site 1 and site 2) and selected switchgrass plants that displayed vigorous growth habit. Specifically, 20 switchgrass plants were collected separately in July 2010 from two reclaimed strip-mining sites in western Kentucky (USA), where they were established as a monoculture during reclamation (approximately 20 years ago). Selection site 1 was located between the GPS coordinates of W:087 25′ 04,13′′ to 087 25′ 07, 31′′ longitude and N:037, 15′ 46, 21′′ to 037, 15′ 50, 60′′ latitude; site 2 was located between the GPS coordinates of W:087 32′ 18, 09′′ to 087 32′ 19, 14′′ longitude, N:037, 12′ 55, 44′′ to 037, 12′ 55, 81′′ latitude. Given that these were reclaimed mine sites, vigorous plant growth was a qualitative and subjective measure, and related to plants that were growing more vigorously compared to other switchgrass specimen. Inherent variability in soil structure existed in discreet spatial areas (within 10 m) due to coal and rock remnants on the surface. Rock and surface coal remained at both sites (Fig. S1). Site 1 displayed far more variability, which may represent less uniformity of the soil during reclamation. Therefore, we sampled both switchgrass plants and soils at four representative subsites within site 1. Representative soil samples directly surrounding the sampled plants were collected within the rhizosphere (at a depth of 0–15 cm and within 30 cm of the plant stem/root interface). Four soil samples were collected from site 1 and one from site 2, and analyzed for pH, organic matter and other soil chemical parameters by the University of Kentucky Regulatory Service Soil Testing Lab (Soil & Plant Analysis Council, 2000). In selecting the most vigorous plants, we visually scored the average upper end plant height and selected individuals that met or exceeded typical plant growth (>100 cm). Understanding that natural genetic variability plays a major role in plant fitness, we used this primarily as a starting point for making selections in a diverse field environment.

Isolation of bacteria from within shoot, root and seed tissue

Shoot (collectively the leaf and stem) and root segments of approximately 1–1.5 cm in length were hand cut from the collected whole switchgrass plants. Stems were sectioned at the base, middle, and crown and leaves were transversely sectioned from tip to base. These segments were washed with deionized water (dH2O) to remove remnant soil and debris. Segments were then rinsed with 95% ethanol (EtOH) for 2 min and then immersed into a solution of 30% Clorox (household grade) bleach in sterile dH2O for 20 min and then serially rinsed five times in sterile dH2O. Seeds were excised from the lemma and palea, surface sterilized as above, then cut into two parts. All the cut pieces were placed separately on plates with YPDA medium containing yeast, peptone, dextrose, and agar (Clontech Laboratories Inc., Mountain View, CA, USA). Nystatin (Fisher Scientific, Bridgewater, NJ, USA) was added to the YPDA medium to a final concentration of 100 μg/ml to prevent fungal growth. The plates were then incubated in a growth chamber at a temperature of 26 °C for 3–5 days and the colony morphotypes were examined. A single colony was isolated from each formation and cultured separately (Long et al., 2008).

DNA extractions, 16S rDNA gene amplification, sequencing, and species/strain identification

Individual colonies were separately grown in YPD broth medium (Clontech Laboratories Inc.) overnight at 26 °C on a rotary shaker. DNA was extracted using a Zymo Research fungal/bacterial DNA miniprep kit (Zymo Research, Irvine, CA, USA) following the manufacturer's instruction. The 16S rDNA amplification was performed in a 50-μl reaction mixture including 3-μl DNA template (1–20 ng), 100 μm of primers 27f (5′-GAGTTTGATCCTGGCTCA-3′) and 1498r (5′-ACGGCTACCTTGTTACGACTT-3′), which are complementary to the conserved regions at the 5′ and 3′ ends of the 16s rDNA gene of the Escherichia coli at the positions of 9–27 and 1477–1498, respectively (Lane, 1991), 3mm MgCl2, 3 mm dNTPs, 5 μl of Taq buffer, and 1 U Taq DNA polymerase (Fermentas Inc., Glen Burnie, MD, USA). PCR amplification was performed on an icycler PCR machine (Bio-Rad Labortories, Hercules, CA, USA), with the initial denaturation at 94 °C for 5 min, followed by 50 cycles of amplification (94 °C for 1 min, 54 °C for 1 min, 72 °C for 2 min) and an extension step (72 °C for 5 min). The PCR products were purified using a Fermentas GeneJET PCR purification kit (Fermentas Inc.) and quantified by using a nanodrop spectrophotometer and sent to Elim Biopharm Inc. (Hayward, CA, USA) for sequencing. The sequences were examined and edited manually or using Bioedit Sequene Alignment Editor (http://www.mbio.ncsu.edu/BioEdit/bioedit.html). The sequences were subjected to BLASTn searches in the NCBI and BIBI Databases (Devulder et al., 2003; Mignard & Flandrois, 2006) and the top hits were used to identify the most possible taxonomic resolution to species/strain level.

Data organization and statistical analysis

The bacteria species were grouped into their higher taxonomic level (Phylum level) so that the Phylum frequencies could be evaluated according to their tissue distribution and site distribution. The species diversity was calculated using the Shannon diversity index (Bowman et al., 1971). Differences in Shannon index diversity indices of the two bacterial communities were compared using the Student t-test at a 95% confidence level.

Screening for plant growth promoting bacterial isolates

Switchgrass seeds were washed with 95% EtOH in dH2O for 2 min and then washed with a solution of 30% bleach, 5% sodium dodecyl sulfate (SDS) solution in dH2O for 25 min. They were then rinsed with sterilized dH2O five times and incubated at 4°C for 24 h. Individual colonies of switchgrass-associated bacteria were kept in the YPD broth medium in flasks and grown overnight at 26 °C on a rotary shaker to the log phase of the bacteria growth (OD600 = 0.6). Sterilized switchgrass seeds were co-inoculated with cultured bacteria in the flask for 24 h at 26 °C on a rotary shaker. The bacteria-treated switchgrass seeds were sown into pots containing autoclaved Pro-Mix (Premier Horticulture Inc., PA, Quakertown, USA) potting media, and seeds with mock bacterial treatment (YPD broth) sown as a control in separate pots. Bacteria mixed with broth were also surface applied to the pots containing treated seeds to drive overrepresentation of bacteria in the rhizosphere. The pots were kept in a green house with constant temperatures of 28 °C, and 16 h of light followed by 8 h of dark for 60 days. The length of the longest laminas (n = 3) from three independent switchgrass plants was measured per treatment and compared with control plants to establish lamina expansion after 60 post co-inoculation. Lamina lengths were compared by the Student's t-test at a 95% confidence level.


Identification of culturable bacteria from switchgrass plant tissue

A total of 307 bacterial isolates were cultured from switchgrass plants collected from the two mining sites. Through 16S ribosomal-DNA sequencing and NCBI BLAST search we demonstrated that these endophytic bacteria associated with switchgrass belonged to 76 different strains that were classified into 36 distinct species. Further classification showed that the 36 species belonged to five principle bacterial phyla. Phylum Firmicutes were the most commonly isolated bacteria, which included 51 strains, 18 species, and 231 isolates and constituted approximately 67%, 50%, and 75% of the total bacterial species types, strain types, and isolates number, respectively, in this study (Fig. 1a–c). The phylum Firmicutes included Bacillus cereus, B. licheniformis, B. pumilus, B. simplex, B sp., B. thuringiensis, Brevibacillus reuszeri, Brevibacillus sp., Lysinibacillus fusiformis, Microbacterium sp., M. oleivorans, M. testaceum, Micrococcus sp., Paenibacillus polymyxa, Paenibacillus rhizosphaerae, P. sp., Staphylococcus sp., and uncultured Eubacterium (Table S1). Other phyla identified included Proteobacteria, Actinobacteria, Bacteriodetes, Deinococcus-Thermus. The phylum of Proteobacteria is the second most abundant phylum of these isolated bacteria and includes Agrobacterium sp., Burkholderia cenocepacia, Burkholderia gladioli, Pantoea conspicua, Paracoccus sp., Pseudomonas geniculata, P. putida, P. sp., Stenotrophomonas maltophilia, Stenotrophomonas sp., and uncultured α-Proteobacterium (Table S1). The phylum Actinobacteria included Arthrobacter sp., Kocuria kristinae, Microbispora rosea, and Propionibacterium sp., and Bacteriodetes included two genera, the Chryseobacterium sp. and Flavobacterium sp. (Table S1). Lowest abundance was observed in the phyla Deinococcus-Thermus. Specifically, Deinococcus sp., was the only species cultured.

Figure 1.

Taxonomic distribution of endophytic bacteria isolated from switchgrass grown on two mine sites. The distribution of strains (n = 76) (a), species (n = 36) (b) and isolates (n = 307) (c) were categorized into five principle Phyla and displayed as a percentage of the total. All data reflect total data from both mining sites (see Table S1). Distribution of strains beyond the species level was examined (d) where bars represent the occurrence of strains (n = 76) among the 36 different bacterial species isolated from switchgrass. All data reflect total data from both mining sites (see Table S1). Distribution of isolates (e) at the species level was examined. Bars represent the number of isolates (n = 307) categorized among 36 different bacterial species from switchgrass. Data are broken into site 1 and site 2 in each species (see Table S1).

The term strain was used to classify an organism beyond the species level, which was frequently observed, most notably for the Bacillus genera (Fig. 1d, see Table S1 for strain identity). Examination of the distribution pattern within the endophytic bacterial isolates showed that 10 species comprised approximately 80% of the total 307 isolates, with the most abundant being B. thuringiensis, B. sp., B. cereus, Staphylococcus sp., Stenotrophomonas sp., B. pumilus, Lysinibacillus fusiformis, Paracoccus sp., Micrococcus sp., and Paenibacillus sp. (Fig. 1e; Table S1).

Classification of bacterial endophytes associated with sites and tissues

A total of five phyla were distributed in both mining sites. At the phylum level, the Actinobacteria and Deinococcus-Thermus displayed higher relative proportions in site 2 (Fig. 2). A total of 16 species were commonly distributed in both sites, 17 species were unique to site 1, and 3 species were unique to site 2, including Arthrobacter sp., Kocuria kristinae, and Paenibacillus rhizosphaerae (Figs 1e and 3). Together there were 33 species in site 1, and 19 different species in site 2 (Figs 1e and 3).

Figure 2.

Phylum level classification of endophytic bacteria from switchgrass. Individual endophytic bacteria were categorized into Phylum and compared between mining site 1 and 2 to illustrate distribution.

Figure 3.

Abundance of individual isolates (n = 307) broken into a species level classification (n = 36) compared between sites and tissues. During endophyte isolation, the frequency at which an individual was isolated (termed isolate) was counted based on sequence based classification at the species level (resulting in 36 different species). These were then organized into a meta-analysis bar chart, which reveals the abundance of specific species shifting between sites and tissues.

Visual inspection of site 1 revealed a great deal of inherent variation in the local soil environments (Fig. S1). To explore whether soil conditions were different within site 1, we examined pH and organic carbon from the rhizosphere of four sub-sites (individual plants) and found variation in both chemical parameters (Table 1). Since all the switchgrass plants were established at the site at the same time, environmental conditions such as temperature and rainfall were consistent, and all the plants were collected within 100 m2 area, we examined a pairwise comparison of Shannon diversity index vs. pH and organic carbon across the four sub-sites. These data revealed correlation between increasing soil pH and increasing diversity index value (Fig. 4a), but no correlation between bacterial diversity and organic matter content was evident (Fig. 4b). The pH of the soil has been shown to impact soil microbial diversity (Fierer & Jackson, 2006), and therefore a similar positive correlation with the endophytic microbial diversity of switchgrass was not unexpected.

Figure 4.

Graphical representation of the pairwise relationship between the diversity of switchgrass associated endophytic bacteria and soil pH (a), R2 = 0.65 or (b) SOM, R2 = 0.11.

Table 1. Soil chemical analysis from two coal mining sites. In terms of subclassifications based on localized soil samples, site 1 has been broken into four subsites (part 1 to part 4) and site 2 has one site (part 5). Measurements included soil organic matter (%) and pH. Standard deviation from the mean is indicated by plus/minus based on three replicate analyses for each sample
 Site 1Site 2
Part 1Part 2Part 3Part 4Part 5
  1. SOM, soil organic matter.

SOM (%)10.5 ± 5.33.3 ± 0.51.7 ± 0.31.6 ± 0.63. 1 ± 0.8
Soil pH7.5 ± 0.36.5 ± 0.66.1 ± 0.26.9 ± 0.64.3 ± 0.04

Bacterial endophytes were cultured and isolated from different switchgrass tissues. Four distinct phyla- Firmicutes, Proteobacteria, Actinobacteria, and Bacteriodetes were isolated from within all the tissues, except Phylum Deinococcus-Thermus, which was only found in seed and shoot tissues. Of 307 isolates, there were 36 species isolated; of these eight species were distributed in all tissues sampled, including B. cereus, B. pumilus, B. sp., B. thuringiensis, Flavobacterium sp., Paenibacillus sp., Staphylococcus sp., and Stenotrophomonas sp. (Fig. 3). Switchgrass shoot segments were most heavily sampled and had 196 different isolates belonging to 30 species; the roots had 65 isolates belonging to 17 species; the seeds had 46 isolates belonging to 16 species, which meant that there were 64%, 21%, and 15% bacterial isolates distributed in shoot, root and seed tissue, respectively. The most abundant species in all the tissues were Bacillus ssp., Staphylococcus sp. and Stenotrophomonas sp. Two of the bacteria isolated from seed, Arthrobacter sp. and Microbacterium testaceum, both were previously demonstrated to behave endophytically (Sessitsch et al., 2004; Sturz & Christie, 1998; Zinniel, 2002). Certain species were unique in certain tissues, such as Agrobacterium sp., B. licheniformis, Burkholderia cenocepacia, Burkholderia gladioli, Chryseobacterium sp., Kocuria kristinae, Microbacterium oleivorans, Paenibacillus rhizosphaerae, Paracoccus sp., Pseudomonas geniculata, uncultured alpha proteobacterium, uncultured Propionibacterium sp., and Eubacterium sp., which were only isolated from shoot. As a reference tool rather than a categorization, metadata including tissue and site information have been tabulated (Table S1).

Analysis of the impact of co-inoculating bacteria isolated from switchgrass

We sought to isolate culturable bacterial strains that could then be examined for their capacity to impact plant performance. Bacteria are capable of altering plant performance in multiple ways (Krause, 2006; Olivares et al., 1996; Rosenblueth & Martínez Romero, 2006); the most common is by increasing mineralization in the rhizosphere to enhance nutrient availability (Fierer et al., 2007; Govindasamy et al., 2011). With this in mind, it was evident that bacteria, isolated as endophytes, may be rhizosphere bacteria that have the capacity to reside within plant tissue. As noted, we isolated a total of 76 different bacterial strains that were composed of 36 different species. We therefore examined the capacity for each of the 76 strains to modify plant growth by co-inoculating individual strains with surface-sterilized switchgrass seed. By inoculating seed and dispersing bacteria into the soilless media we promoted an overrepresentation of the target bacterial colony. To develop a simple metric to screen all 76 samples, the length of the longest lamella (n = 3 from each of n = 3 plants) was used to establish the average seedling lamina expansion. The average lamina length of mock treated seed was 32.68 cm ± 1.8 (SD), measured after 60 days of growth. Of the 76 strains that were co-inoculated with switchgrass plants, 24 caused a decrease in average lamina length while 52 caused a relative increase. Not all increases were significant and it was found that of the 76 strains, 58% promoted lamella length significantly compared with the mock control (Table S1, P < 0.05, Student t-test) and 9.2% caused a significant decrease. Hence, more than half of the bacteria isolated from surface sterilized switchgrass plants were functional in plant growth promotion. Those that increased switchgrass lamina expansion by more than 25% relative to the mock control included ten members of the genus Bacillus, as well as Flavobacterium sp., Brevibacillus sp., Paenibacillus sp., Paenibacillus polymyxa, Lysinibacillus fusiformis, Pseudomonas sp., Pseudomonas putida, Micrococcus sp. and Burkholderia gladioli (Table S1). By contrast, several bacteria caused greater than 25% reduction in lamina growth, notably Paracoccus sp. Zy-3, Burkholderia cenocepacia strain ccbau, Brevibacillus reuszeri strain DSM 9887, Bacillus thuringiensis strain Kt1-21, and Bacillus sp. YXE3-6 (Table S1). Taken together, these reveal that in isolation, one of two bacteria isolated from surface-sterilized switchgrass material was capable of acting in host plant growth promotion. Although less common, growth-retarding bacteria were also members of the bacterial community isolated from surface sterilized switchgrass.


Switchgrass, once widely distributed across the plains of the United States, in now limited to few natural stands. Despite this, it has emerged as a forage, reclamation, and bioenergy crop of increasing importance to agriculture (Comis, 2006). The endophytic community of switchgrass is therefore of considerable interest. Surveys of fungal endophytes in switchgrass classified a substantial fungal population (Ghimire et al., 2010; Kleczewski et al., 2012). Furthermore, fungal endophytes appear to have the capacity to modify switchgrass growth performance (Kleczewski et al., 2012). In this study, abundant bacteria were cultured from surface-sterilized switchgrass tissue (Table S1), which is consistent with prior reports of bacterial colonization of plant tissues (Hardoim et al., 2008; Olivares et al., 1996; Pillay & Nowak, 1997; Sessitsch et al., 2004; Siciliano & Germida, 1999). These prior studies have linked bacteria that colonize plant tissues with growth promotion (Compant et al., 2005) or to the soil environment surrounding the host plant (Siciliano et al., 2001). Similarly, Burkholderia phytofirmans strain PsJN has been documented to colonize and to promote the growth of switchgrass (Kim et al., 2012). However, culturable bacteria residing in switchgrass have not been examined in depth nor had the extent to which microflora were capable of influencing switchgrass growth and development in isolation documented. Our data sugges that the subset of bacterial flora identified as colonizing switchgrass were capable of influencing plant growth and development (Table S1). We advance two hypotheses to explain these data. Firstly, increased lamina length, which was observed as arising from greater than 50% of isolates could arise from direct endophytic association. Here, a direct influence on host metabolism plausibly involves the production of phytohormones including indole acetic acid, acetoin, or 2,3-butanediol, all of which have been shown to be produced by the endophytic bacterium Enterobacter sp. 638, which colonizes Populus trees (Taghavi et al., 2010). Indeed, production on exogenous indole acetic acid (auxin) would directly stimulate cellular expansion in host tissues. Alternatively, increased lamina length may arise from an indirect influence via an overrepresentation of individual bacteria residing within the rhizosphere and correspondingly improving mineralization rates. It was found that many strains isolated from switchgrass were previously documented as plant growth promoting rhizobacteria (Hardoim et al., 2008; Kloepper et al., 1980; Siciliano & Germida, 1999; Smalla, 2001). On this note, bacteria isolated from within plant tissue may have a commensal relationship with the host, rather than mutualistic. Indeed, by their isolation on selective media, they do not require the host plant (not obligate) for proliferation. Although it is intriguing that a proportion of switchgrass-associated bacteria were capable of eliciting increased lamina expansion in isolation when added to potting media, this can by no means be inferred to reflect the soil or rhizosphere environment where more complex ecosystem processes exist.

Based on the definition of Hallmann et al. (1997), the internal bacterial biota of a host plant, termed as an endophyte, are ‘bacteria that can be isolated from visually asymptomatic, surface sterilized plant tissues’. Herein, we sought culturable bacteria rather than utilizing pyrosequencing to obtain a complete snapshot of the community profile. Pyrosequencing can lead to a more extensive view of obligate and difficult to culture endophytes (Manter et al., 2010). Culturing individual bacteria and identifying each candidate by traditional sequencing requires laborious isolation, DNA extraction, PCR amplification, and sequence analysis, but also yield pure isolates that can be examined in isolation, as was achieved herein. Interestingly, our data revealed an overrepresentation of the phylum Firmicutes in bacteria isolated from surface-sterilized switchgrass. By contrast, pyrosequencing (Manter et al., 2010) revealed that Proteobacteria and Bacteroidetes were markedly more abundant. It is plausible that a large difference in community profile exists between host taxa, particularly when plants are grown in a highly stressful environment, as is the case for switchgrass plants growing on reclaimed mine sites. However, it is more likely that a culture-based approach underrepresented this phylum. These two hypotheses are not mutually exclusive. Furthermore, via a culture-based approach, we isolated and identified 307 bacteria from switchgrass; these isolates represented 76 taxonomic units. Based on pyrosequencing (Manter et al., 2010) studies that revealed up to 477 taxonomic units in root tissue, the identification of 76 from whole plants was almost certainly a fraction of the total community present.

Soil pH and organic matter levels differed greatly between the two sites (Table 1). Pairwise comparisons between the diversity of switchgrass-derived bacteria and soil pH/organic matter using four sub-sites within site 1 showed that soil pH correlated with endophyte diversity (Fig. 4a), but not soil organic matter. Although it is consistent with pH remaining a central driver for soil bacteria community diversity at the continental scale (Fierer & Jackson, 2006), we are cautious in interpreting a correlation between local soil conditions and endophyte diversity without pyrosequencing, for which future studies are needed.

Sequence homology of the 16S rDNA revealed that B. thuringiensis, B. cereus, and sequences matching uncharacterized Bacillus sp. were the most abundant isolates cultured from within switchgrass tissue (Fig. 1e). B. thuringiensis has been well studied because many strains synthesize the Cry toxin, which displays potent insecticidal activity against lepidopteran and dipteran larva (Höfte & Whiteley, 1989). The presence of B. thuringiensis as switchgrass colonizing may plausibly represent a functional association, but was not mechanistically explored in this study. The second most abundant Phylum in this study was Proteobacteria, which constituted 31%, 24%, and 19% of the total bacterial species types, strain types, and isolates numbers, respectively (Fig. 1a–c). Proteobacteria was also found to be the most prominent phyla in potato root (Solanum tuberosum) by pyrosequencing (Manter et al., 2010). Some genera in this phylum such as Stenotrophomonas form mutualistic endophytic associations with host plants and fix atmospheric nitrogen (Noel et al., 2008; Ramos et al., 2011). Several Stenotrophomonas isolates associated with switchgrass (Table S1) and caused increased lamina length resulting from their co-cultivation in isolation. Interestingly, the recent characterization of a nitrogen-fixing Stenotrophomonas (Ramos et al., 2011) was identified in sugar cane (Saccharum officinarum), which like switchgrass falls into the Panicoideae clade of C4 grasses. Further common isolated bacteria from switchgrass and sugar cane included Burkholderia, Pantoea, Microbacterium, and Pseudomonas (Table S1) (Mendes et al., 2007).

Future work is needed to establish an understanding of the ecological diversity of the switchgrass microbiome, which could be achieved by pyrosequencing samples from the remaining endemic stands of switchgrass as well as regionally diverse forage and biofuel plantings across a seasonal gradient. Of further interest is whether or not bacteria that are capable of residing as endophytes, are also active microbial components of the rhizosphere.


We thank Dr. Tim Phillips, Mr. Rick Gammill, and Dr. Jozsef Stork for assistance with sample collection. The University of Kentucky Soil Analysis Laboratory and Mr. Xia Yu (Department of Mathematics and Statistics, University of Kentucky) are gratefully acknowledged for technical assistance. Supported by the National Science Foundation 0922947, 0937657 (SD).