Disentangling the factors shaping microbiota composition across the plant holobiont

Healthy and asymptomatic plants in nature are colonized by a rich diversity of microbes comprising bacteria, fungi, protists and viruses (i.e. the plant microbiota), forming complex microbial consortia that impact plant growth and productivity. Consequently, plants must not be viewed as autonomous entities but rather as holobionts (a macrobe and its numerous microbial associates), within which all interacting organisms contribute to the overall stability of the system (Vandenkoornhuyse et al., 2015). More than a century ago, Hiltner hypothesized that the resistance of plants towards pathogenesis is dependent on the composition of plant microflora and that root exudates of different plants could support development of different microbial communities (Hartmann et al., 2008). The development of next generation sequencing technologies and associated computational analytical tools now allows the detailed investigation of these important concepts (Bulgarelli et al., 2012; Lundberg et al., 2012). However, despite the fact that the plant microbiota research field shows exponential growth (Fig. 1), most of the studies published so far have focused on one particular microbial kingdom and/or specific host niches. There is consequently a need for a more holistic understanding of the microbial communities associated with different plant compartments and discerning which factors shape these microbial assemblages across the plant holobiont. In this issue of New Phytologist, Coleman-Derr et al. (pp. 798–811) provide a comprehensive analysis of the structure of both fungal and bacterial communities in the rhizosphere, phyllosphere, leaf and root endosphere, as well as proximal and distal soil samples from cultivated and native agaves. Since agaves spp. are adapted to nutrient-poor environments, extreme drought and elevated temperatures, these plants represent important models for the plant microbiota research field because they are likely to host an important reservoir of beneficial microbes that may support their survival. It is now well established that root colonization by soil bacteria is a deterministic and tightly controlled process involving different selective steps. Edaphic factors determine the start inoculum of the soil biome whereas rhizodeposits and host genotype mediate bacterial community shifts from soil communities to host-adapted communities with reduced diversity (Bulgarelli et al., 2013). Therefore, the roots of phylogenetically unrelated plant species assemble overall structurally related bacterial communities belonging to only four major bacterial phyla (Proteobacteria, Actinobacteria, Bacteroidetes and Firmicutes), with Proteobacteria and Acidobacteria community members being consistently enriched and depleted, respectively, in plant roots compared to their surrounding soil biome (Hacquard et al., 2015). Importantly, Coleman-Derr et al. report here that bacterial communities associated with agave plants growing in arid environments also share this taxonomic signature, indicating that the selective forces shaping root microbiota composition at a high taxonomic rank are robust against a wide range of host plants and environmental conditions. In contrast with the bacterial microbiota of plant roots for which the start inoculum can be easily defined, it is assumed that leaf-associated bacterial communities are more subject to stochastic variations due to the multiple and more complex inoculum sources (Maignien et al., 2014). Consistent with this, Coleman-Derr et al. show that leaf endosphere communities show higher variability across seasons than the root endopshere communities. Nonetheless, they could identify core prokaryotic taxa that are shared between endophytic compartments, including microbiota members belonging to Actinobacteria, Bacilli, Alpha-, Beta- and Gamma-proteobacteria that may confer plant fitness benefits during the dry season. Remarkably, despite being harvested 2000 km apart in their respective natural habitats, agave-associated bacterial communities were more impacted by plant compartment than by biogeography of the host (Fig. 2a). Their results suggest that geographically distant bacterial inputs can converge into an overall reproducible taxonomic structure in each agave compartment. It is nevertheless important to note that a substantial fraction of the bacterial, but also the fungal communities, is shared between compartments, raising the possibility of extensive reciprocal relocation of microbiota members between belowground and aboveground plant tissues. ‘Remarkably, despite being harvested 2000 km apart in their respective natural habitats, agave-associated bacterial communities were more impacted by plant compartment than by biogeography of the host.’ Even though less attention has been given to the fungal microbiota of plants, fungal communities appear to be hyper-diverse in both aboveground and belowground plant tissues, and are dominated by two major phyla (Ascomycota and Basidiomycota). However, in contrast with the bacterial microbiota of plant roots, the fungal communities seem to be subject to greater variations that are known to be highly dependant on biogeography, host species and plant compartment, but which may also include a non-negligible stochastic component (Shakya et al., 2013). One key aspect in the experimental design used by Coleman-Derr et al. was to analyse agave communities from distant geographical sites to evaluate whether biogeography of the host similarly impacts fungal and bacterial assemblages. Overall, they found that the major factor driving fungal assemblages in agave plants is the geographic origin of the host, contrasting with bacterial assemblages that are primarily sculpted by plant compartment (Fig. 2a). Their findings indicate that distinct factors shape fungal and bacterial assemblages on plants and also confirm that fungal and bacterial biogeography may differ fundamentally on large scales. Unlike bacteria, but similar to plants and animals, distribution of fungi might be primarily constrained by climate and dispersal limitation, which may favour higher endemism in fungal compared to bacterial populations (Bonito et al., 2014; Talbot et al., 2014). This is consistent with the positive correlation observed between plant and fungal richness, suggesting these two kingdoms respond similarly to climatic and edaphic variables (Tedersoo et al., 2014). It is likely that habitat-specific features drive assemblages of distinct plant-associated fungal communities with high levels of functional redundancies across sites (i.e. conferring plant fitness benefits against a wide range of stresses) but also non-negligible site-specific functional capabilities (i.e. conferring plant fitness benefits against local stresses). Since the colonization of land by ancestral plant lineages 450 million years ago, plants and microbes have interacted with each other (Field et al., 2015). This long co-evolutionary process has likely shaped plant-associated microbial assemblages and selected for beneficial interactions across the plant holobiont. One fundamental question in plant microbiota research is whether cultivation practices can destabilize indigenous and locally adapted microbial communities and whether disturbance can modify prevalence of a particular disease (Fig. 2b). In humans, it has been postulated that hygiene measures aimed at reducing the microbial load in our environments may have instead favoured modern allergic and metabolic diseases, likely due to the loss of our ancestral commensal microbes (Blaser & Falkow, 2009). Although the number of sampling sites selected by Coleman-Derr et al. was not sufficient to draw a general principle regarding how cultivation practices alter microbial community establishment, they found that cultivated agave harbour less prokaryotic diversity than native agave in the rhizosphere and phyllosphere compartments. This loss of prokaryotic diversity was largely explained by the dominance of some bacterial genera belonging to the family Enterobacteriaceae, among which some members are well known pathogens of agave causing ‘soft rot’ disease. The findings of Coleman-Derr et al. are consistent with the idea that agricultural practices that include continuous monoculture, the utilization of germ-free seedlings and homogenous genetic plant material may perturb heritability of native microbial consortia and favour emergence of latent pathogens (Santhanam et al., 2015). Although the structure of both fungal and bacterial communities in the leaves and roots of several plant species has been elucidated, there is still a lack of knowledge regarding how multi-organismal interactions shape microbiota composition across the plant holobiont (Hacquard & Schadt, 2015). More particularly, it remains unclear how competition and cooperation among microbes or between different microbial kingdoms influence microbial community establishment. It is likely that plant-associated microbiota members have evolved sophisticated strategies to interact with each other within complex microbial consortia in order to persist in particular host niches. Understanding plants as holobionts therefore implies careful investigation of these often neglected microbe–microbe interactions (Fig. 2b). Disentangling the interplay between host, microbial and environmental factors is also needed to better understand microbiota niche specialization and adaptation to a specific plant compartment. Large-scale isolation and establishment of reference culture collections of plant-associated microbiota members is a prerequisite for addressing these important concepts and represents a promising way for reconstituting complex microbial ecosystems in vitro with germ-free plants (Lebeis et al., 2015). These reductionist approaches using synthetic microbial communities will have profound implications for the plant microbiota research field, allowing a transition from correlation- to causation-based studies. The author would like to thank Francis Martin for giving him the opportunity to write this commentary, Paul Schulze-Lefert for his continuous support and scientific advice as well as William Chanz Robbins, Stijn Spaepen and Paloma Durán for their critical comments regarding this commentary. The author would also like to thank the funding sources: the Max-Planck Society and the European Research Council.