We studied the effects of external application of individual and mixed Bacillus species on the sprouting broccoli root endophytic bacterial community using 454 pyrosequencing. Since over 99% of bacteria in a range of environmental samples have been shown to be unculturable , we took a culture independent approach, wherein total bacterial DNA from each sample was extracted and 16S rRNA amplicons were sequenced to analyze what is in effect the entire root bacterial communities from differentially treated and untreated plants. Primers equivalent to those chosen have previously been shown to amplify over 90% of all species tested and to show high coverage of almost all phyla . We found that the addition of these ubiquitous bacterial species to plants changed the diversity, evenness, and composition of their indigenous endophytic bacterial communities, in a context-specific manner, upholding our original hypothesis.
The relative abundance of the genus Bacillus itself, however, was very low in all plants despite its external addition to seeds and soil samples. Bacillus is one of the most abundant bacteria in the soil  and rhizosphere ; however, despite its ubiquitous nature and two external applications, its abundance was comparatively lower than that of the majority of predominant endophytic bacteria. Bacillus has been found to be a significant contributor to the endosphere within many other species such as rice, maize, potato, grapevine, coffee, and coconut . It is important to note that this observation may be specific to the soil type used in this study but our observations are similar to those in the related plant species, Brassica napus, where Bacillus was only found in low relative abundance in most cultivars . In contrast, another common endophyte, Pseudomonas , was the dominant genus in all samples and this also mirroring the situation found in B. napus . Apart from Pseudomonas, the predominant genera in the sprouting broccoli endosphere were Lysobacter, Acidovorax, and Rhizobium, again, all commonly observed endosphere genera [40, 41]. This lack of recruitment of the added Bacillus inocula to the endosphere suggests that the Bacillus species may have instead exerted their effect on endophyte recruitment in the rhizosphere, most likely via competition with other rhizosphere bacteria. However, it is critical to note that the overall structure and composition of microbial community, plant-associated factors, and environmental conditions may have shaped the outcome of Bacillus-mediated effects on bacterial endophytes. A thorough investigations analyzing most of these components in real time would provide a comprehensive understanding of these intricate interactions.
There were a number of common effects of Bacillus supplementation on endosphere composition, notably, a marked reduction in the relative abundance of Lysobacter and Acidovorax, perhaps suggesting a general competition between Bacillus spp. and these genera possibly through antagonism in earlier stages when the majority of bacteria colonize roots . The competition for nutrients and niches may have prevented several other genera from colonizing germinating seeds and seedlings. On the other hand, there was an increase in the relative abundance of Acinetobacter. It may, therefore, be that Acinetobacter was previously inhibited due to antagonistic activities of Lysobacter and Acidovorax. However, the most dramatic effects were species specific. Addition of B. amyloliquefaciens resulted in a large decrease in the relative abundance of the most common endophyte, Pseudomonas. There was also a loss of Rhizobium, with these changes being accompanied by an increase in the relative abundance of a wide range of genera, particularly Dyadobacter, Variovorax, Tahibacter, and Sphingomonas. This outcome suggests a significant impact of B. amyloliquefaciens on the bacterial community within the rhizosphere from which the endosphere community is recruited, while again suggesting that the genera Dyadobacter, Variovorax, Tahibacter, and Sphingomonas may have previously failed to proliferate within roots, possibly due to antagonistic activities of Pseudomonas and Rhizobium.
Apart from numerous bacterial interactions in rhizospheres and endospheres , a variety of extraneous factors such as root metabolites, plant growth stage, native rhizosphere microbial community [42, 43], competence for nutrients and niche, and soil type may have played important roles in determining relative abundances of different taxa in each group and thus in shaping the overall endophytic community structure and dynamics [5, 14, 44].
Curiously, the effects of B. amyloliquefaciens supplementation were not observed with the mixed inoculum which also contained B. amyloliquefaciens. Such a phenomenon has been observed in studies exploring the effects of bacterial inocula on plant resistance to pests [21, 45, 46]. Effects of single versus multiple species inocula have been proposed to depend on the abundance of the added species among the indigenous microbial community at the target site. Addition of single indigenous species may boost that species at the expense of others and it may, in fact, be this decrease in other key species which affects the outcome. Conversely, addition of multiple indigenous species will boost all of those species equally and, if this mixed inoculum now also includes the key species too, the outcome would not be observed . Here, Bacillus is a ubiquitous component of the rhizosphere and, thus, it is very possible that several or even all of the added species are indigenous. Addition of B. amyloliquefaciens would likely boost that species at the expense of others and it may be that it is a decrease in one of the other trialed Bacillus species which impacts the ultimate recruitment to the endosphere. Conversely, addition of multiple Bacillus species will boost all equally and so the effects of reduction of whichever species was key would not be seen.
The alpha diversity and evenness indices were comparatively higher in B. amyloliquefaciens- and mixed-treated plants. Although the mixed treatment did not cause such dramatic effects on the major taxa as the B. amyloliquefaciens treatment, both resulted in greatly increased relative abundances of minor taxa. Addition of B. cereus or B. subtilis had no significant effect on alpha diversity or evenness, consistent with them showing little effect on the relative abundances of either dominant or minor taxa. However, assessment of beta diversity showed that they did actually significantly affect the specific composition of the endophytic community. Principal coordinate analysis of the weighted UniFrac distance matrix revealed that all inocula resulted in a significant change in beta diversity. The fact that all inocula result in significant changes to the endophytic bacteria emphasizes the importance of the rhizosphere in shaping the endosphere, as previously suggested [9,10,11]. It also emphasizes the sensitivity of the rhizosphere-endosphere relationship and the possibilities for external manipulation of rhizosphere community via addition of inocula. Along the first principal coordinate, B. amyloliquefaciens and B. cereus showed opposing directions of effect with possibly the intermediate position of the mixed Bacillus-treated sample representing the canceling out of these opposing effects. This supports the theory proposed by Gadhave et al. , referenced above, that addition of mixed species inocula may actually cause a loss of effects that would be observed due to addition of only one component of that mixture.
Since this experiment was a part of a larger study, we have previously published the results of our field experiment  designed in the exact same way, with identical resources, and location, but which instead studied the effects of the Bacillus spp. treatments on foliar-feeding insects and natural enemies. A number of strikingly similar patterns can be drawn from results from both studies. Firstly, Bacillus spp. showed the significant effects on various attributes of bacterial endophytes and field populations of cabbage aphid (Brevicoryne brassicae) in a context-specific manner. Secondly, the mixed treatment appeared to be not as effective as individual treatments in terms of its effects on endophytes and foliage feeders, suggesting the prevalence of interspecies competition within added Bacillus species. Consequently, aphid populations grew rapidly on both control and mixed treated plants, while not on the other individual treated ones. Lastly, B. amyloliquefaciens proved to be the most distinct of all three individual Bacillus treatments. For instance, the rates of parasitism of B. brassicae by the braconid wasp Diaeretiella rapae were significant on B. cereus- and B. subtilis-treated plants, while not on B. amyloliquefaciens-treated ones. This is possibly due to differential manipulation of the PGPR community by B. amyloliquefaciens. Pseudomonas fluorescens, an ubiquitous PGPR, is reported to play important roles in modulating plant volatile emission and in triggering natural enemy responses . Its reduced abundance in B. amyloliquefaciens treatment in the current study could explain the non-significant effects of this treatment on parasitism by D. rapae in the field study . None of the Bacillus spp. treatments produced significant effects on sprouting broccoli biomass. It is, therefore, possible that the pest suppression mediated by different Bacillus spp. was directly associated with a concomitant diversion of resources into the enhanced constitutive and induced plant defenses at the expense of biomass.
Thus, the effects of each bacterial species may be context specific, instead of being widely applicable, and may be the net effect of opposing actions or may even be governed by interspecific competition within the seed inoculum too. As such, this study emphasizes the need to examine effects of inocula on a case by case basis. The fact that the effects of rhizosphere inoculum addition on the endosphere are dependent on the context of other bacterial species in the rhizosphere further emphasizes the importance of testing in the field environment. The published effects of rhizobacteria addition on plant growth and biochemistry, and insect herbivores are likely to be determined indirectly via changes in endophytic bacterial communities through numerous bacterial interactions prior and post colonization . Consistent with our findings, these effects and interactions are often unpredictable, which can lead to variable effects of inoculants on plants, herbivores, and higher trophic levels in the field . A relatively lower endophytic and rhizosphere bacterial diversity in laboratory and green house experiments could be one of the major reasons behind the relatively more consistent performance of inoculants in these conditions than in the field. On the contrary, microbial inoculants often fail to show promising results in the field , possibly due to the inability of added inoculants to compete in the complex rhizosphere environment. A comprehensive understanding of bacterial communities in diverse soils and plants through high-throughput sequencing technologies will help develop inoculants that are better suited to local conditions. This will increase the magnitude of plant growth promotion via inclusion of better suited bacterial species in an inoculant and will help alleviate the inconsistencies in inoculant performance in different conditions. Such an increase in use will help to reduce the fertilizer and pesticide application rates and promote the use of this sustainable approach in agriculture.