ABSTRACT
The effect of three plant growth-promoting Bacillus strains inoculated either alone or as a consortium was tested on oat (Avena sativa) growth. The bioinoculants were applied as vegetative cells or endospores at low cell densities on the seeds and their effect was tested in sterile in vitro conditions, pot experiments, and a field trial. The in vitro seed germination assay showed that both individual bacterial inocula and bacterial consortia had positive effects on seed germination. Greenhouse pot experiments with sterile and non-sterile soil showed that consortia increased the total dry biomass of oat plants as compared to single strain inoculation and uninoculated controls. However, the positive impact on plant growth was less prominent when the bioinoculated strains had to compete with native soil microbes. Finally, the field experiment demonstrated that the consortium of vegetative cells was more efficient in promoting oat growth than the endospore consortium and the uninoculated control. Moreover, both consortia successfully colonized the roots and the rhizosphere of oat plants, without modifying the overall structure of the autochthonous soil microbial communities.
INTRODUCTION
Global population growth is putting tremendous pressure on agriculture because of the decrease in arable land and the concomitant increase in the demand for a reliable food supply. Additionally, inappropriate agricultural techniques, such as intensive exploitation of soil, have considerably reduced soil fertility (Souza, Ambrosini and Passaglia 2015). In conventional agriculture, competitive food production is maintained through the continuous use of agrochemicals to enhance soil fertility, crop yield, and to control diseases. However, this has resulted in leaching or runoff of agrochemicals into groundwater, rivers and lakes, creating low oxygen zones where survival of aquatic life is at risk (Muhibbullah, Salma and Chowdhury 2005). Moreover, direct and indirect exposure to hazardous chemicals has deleterious effects on the health of humans and the environment (Muhibbullah, Salma and Chowdhury 2005). Another drawback of agrochemical use is that uptake efficiency by plants decreases over time, requiring regular applications along with increased production cost for farmers. Moreover, for some nutrients, application does not result in higher bioavailability of nutrients. For instance, it has been reported that phosphorous fertilizers precipitate with metal cations and turn into an insoluble form that is no longer bioavailable to plants (Ahemad and Kibret 2014). Therefore, a change in these practices and innovative ways of conducting sustainable food production are a clear need for the future of agriculture.
Many studies have shown the potential of inoculation with plant growth-promoting (PGP) microorganisms to improve agricultural yield (Bhattacharyya and Jha 2012; Ahemad and Kibret 2014; Souza, Ambrosini and Passaglia 2015). More recently, bacterial consortia (mixtures of PGP bacterial strains) were shown to have higher performances as compared to the inoculation of individual species (Baez-Rogelio et al. 2017). For instance, a consortium of PGP bacteria induced the production of defense compounds (such as proline, ascorbate peroxidase or catalase) in response to the biotic stress caused by the phytopathogenic fungus Fusarium oxysporum (Akhtar et al. 2016). Bacterial consortia were also shown to promote plant drought tolerance (Wang et al. 2012). In beans, a consortium of Rhizobium tropici and two strains of Paenibacillus polymyxa increased growth and helped to alleviate the effect of abiotic drought stress as compared to single strain inoculation (Figueiredo et al. 2008).
Bioinoculation of PGP microorganisms also presents many challenges. One of the critical issues that needs to be addressed for application in the field is the survival of bacteria acclimated under laboratory conditions to the harsh conditions in soils. In order to persist in soils, inoculated microorganisms have to compete with autochthonous microbial communities (Souza, Ambrosini and Passaglia 2015). Bioinoculants are also vulnerable to the scarce nutrient availability in soil, as compared to rich nutrient conditions usually provided by growth media, and ultimately declines in the abundance of inoculated bacteria within the soil are frequently reported (Trabelsi and Mhamdi 2013; Souza, Ambrosini and Passaglia 2015). Moreover, the delivery of these microorganisms in an active form is an additional challenge. Application of carrier materials for the protection of bioinoculants (for instance, karnolite, peat or charcoal) has proven to be environmentally unfriendly, as well as costly, making this approach inapplicable as a general practice in agriculture (Arora, Tiwar and Singh 2014). These concerns are the incentive to identify effective bacterial inoculants to be applied in bioinoculation technology.
Among the currently used PGP bacteria, Bacillus is one of the best-studied examples. There are many species of Bacillus that are well known as plant growth-promoters (Kumar, Prakash and Johri 2011). One of the most important characteristics of this genus is that they form endospores, which are a resistant structure that fosters survival. This property can be harnessed in bioinoculation technologies, as it results in long shelf-life of the product before application. Moreover, Bacillus spp., followed by Enterobacter spp. and Pseuodomonas spp., are the most abundant bacteria with PGP traits in the rhizosphere of plants grown in arid conditions because of their high tolerance to extreme environmental conditions (El-Sayed et al. 2014). Bacillus spp. are also used in phytoremediation technologies, increasing biomass production and multi-metal accumulation in plants (Chibuike and Obiora 2014). Bacillus spp. can help plants by not only stimulating their growth through increasing nutrient acquisition (e.g. phosphate solubilization, atmospheric nitrogen fixation, phytohormone and siderophore production (Bhattacharyya and Jha 2012)) but also acting as biocontrol agents against various pests (O'Callaghan 2016; Widnyana and Javandira 2016). Bacillus spp. can also produce a wide range of antiviral, antibacterial and antifungal compounds, which may be important in their interactions with plants and other soil microorganisms. In addition, it has been reported that many species of Bacillus induce systemic resistance in plants against a broad spectrum of phytopathogens (Nui et al. 2011)
In this study, a consortium consisting of three Bacillus strains was used to promote the growth of oat (Avena sativa) plants. The strains were initially tested for physiological traits linked to plant growth-promoting activities. Next, their ability to promote plant growth was assessed in experiments at four levels of complexity (i.e. in vitro seed germination under sterile conditions, greenhouse pot experiments with sterile substrate and non-sterile soil, and finally in a field trial). We hypothesized that based on the complementarity of the plant growth-promoting traits of the strains, the consortium will provide a more robust effect on plant growth, as compared to single strain inoculation. This should represent a real advantage when going from controlled laboratory to field conditions, where the inoculated PGP bacteria will compete with the autochthonous microbial community. Additionally, we also compared two forms of bacterial inocula: endospores and vegetative cells, to assess whether dormancy alters the functionality of the consortium. Finally, we investigated the effect of bioinoculation on native bacterial and fungal communities in the field assay. We hypothesized that a consortium applied directly onto the seeds at low cell densities (ca. 103 cells per oat seed) should not affect the indigenous soil microbial communities.
MATERIALS AND METHODS
Bacterial strains and culture conditions
The bacterial strains Bacillusthuringiensis 1312 (BT1) and B. thuringiensis 1310 (BT2) were isolated from soils of the Atacama Desert, Chile, during a sampling campaign carried out by the laboratory of microbiology, University of Neuchâtel in 2011. Bacillus licheniformis (BL) was isolated from soil at Agroscope Liebefeld, Bern, Switzerland. The strains were cryopreserved in 60% glycerol and when required for the experiment, the strains were pre-cultured in nutrient agar (NA, Carl Roth, Karlsruhe, Germany) media.
Physiological characterization of bacterial strains
Dinitrogen fixation assay
A loop of an overnight bacterial culture on NA was streaked onto a nitrogen-free medium (Döbereiner 1980), which was then incubated at 30°C for 48–72 h. Bacteria were cultured over three generations on the nitrogen-free medium and the capacity to fix N2 was assessed by the growth of bacterial strains in the third-generation plates.
Casein solubilization (proteolysis)
Casein solubilization was assessed based on a modified protocol from Frazier and Rupp (1931). This assay was used as a proxy for the ability of the selected Bacillus strains to solubilize organic nitrogen from proteins. The medium was composed of 5% skimmed milk powder (Migros, Zürich, Switzerland), 1.2% malt extract powder (SIOS, Wald, Switzerland) and 1.5% agar (Merck, Darmstadt, Germany). A loop of an overnight grown bacterial culture on NA was streaked onto the casein-solubilizing medium. The plates were incubated for 48–72 h at 30°C. A transparent halo observed around a bacterial colony indicated that the strain was able to solubilize casein.
Siderophore production assay
To assess the ability of the three bacterial strains to chelate iron, a siderophore production test was performed as described in Schwyn and Neilands (1987). Briefly, the initial medium has a blue coloration resulting from the complexation of ferric iron with chrome azurol S (CAS, Thermo Fischer Scientific, Waltham MA, USA) and hexadecyltrimethylammonium bromide (HDTMA, Merck, Darmstadt, Germany). When a strong iron chelator (such as a siderophore) removes iron from the dye complex, the color of the medium turns to yellow. A loop of an overnight bacterial culture on NA was inoculated onto the medium. The plates were incubated at 30°C and a change of color was assessed after 48–72 h.
Auxin-like phytohormone production assay
To check for the production of auxin-like compounds, a modified version of the protocol proposed by Bric, Bostock and Silverstone (1991) was used. In this assay, the Angle medium (Angle, Mcgrath and Chaney 1991) was supplemented with 5 mM Tryptophane (Merck, Darmstadt, Germany), as a precursor for auxin biosynthesis. The test was performed in 96-well microplates with 270 μL of medium per well. A loop of an overnight bacterial culture from NA medium was inoculated in the three microplate wells for each corresponding bacterial strain. The microplate was then incubated at 25°C for 72 h in the dark. After 72 h, one drop of Salkowski's reagent was added to each well and placed in dark for 45 min. A pink to red coloration indicates the production of auxin-like compounds.
Bacterial compatibility
Compatibility of the strains developing as a consortium was tested by growing them together on a favorable medium (NA, Carl Roth, Karlsruhe, Germany). Each bacterial strain was refreshed overnight in 25 mL nutreint medium and 2 µL inocula containing 106 bacterial cells of each strain was placed on a Petri dish containing NA medium. Each strain was inoculated 1 cm apart from other strains in the Petri dish and the plates were incubated at 30°C for 72 h. The experiment was performed in four replicates. Inhibition was assessed visually using overall culture images and close-ups of overlapping areas of the expanding colonies. An additional approach using CRISPR-Cas9 technology to tag the strains with fluorescent proteins was tested in order to visualize and count the cells individually, but this was unsuccessful until now.
Preparation of bacterial inoculants
Seed bioinoculation with the three selected bacterial strains was performed using either vegetative cells or endospores, and both either as individual cells or as a consortium.
Vegetative cells treatment
Pure cultures of the three selected bacterial strains on nutrient agar (NA, Carl Roth, Karlsruhe, Germany) were used to individually inoculate 50 mL of nutrient broth (NB, Carl Roth, Germany) medium which was further incubated overnight at 30°C and 150 rpm (HT incubator, Infors AG, Bottmingen, Switzerland). An overnight bacterial culture of each strain was transferred into a sterile 50 mL tube and centrifuged (2–16PK, Sigma, Osterode am Harz, Germany) at 2150 g for 3 min. The resulting pellet was washed five times with sterile physiological water. After this, the optical density of each bacterial suspension was measured at 550 nm (Genesys 10S UV-VIS spectrophotometer, Thermo Fisher Scientific, Waltham MA, USA) and adjusted to reach a final concentration of 106 colony forming units (CFU) mL−1. The resulting suspension of vegetative cells was used for experiments with plants (in vitro assay of seed germination, greenhouse pot experiments and field experiment). This resulted in four different treatments for experiments with vegetative cells: VBT1 = B. thuringiensis 1312, VBT2 = B. thuringiensis 1310, VBL = B. licheniformis and vegetative cells consortium = VM.
Endospore treatment
A synthetic medium was used to induce sporulation of the bacterial strains (sporulation medium; Donnellan, Nags and Levinson 1964). Each bacterial strain was refreshed in 50 mL NB. Overnight bacterial cultures were centrifuged (2–16PK, Sigma, Osterode am Harz, Germany) at 7673 g for 3 min. The resulting pellet was washed five times with physiological water. At the end, the pellet of each bacterial strain was suspended in 100 mL of sporulation medium and incubated for 3 weeks at 30°C and 200 rpm. After 3 weeks, endospore formation was verified by phase contrast microscopy. Vegetative cells observed in the suspension were killed by a heat shock at 70°C for 15 min for B. thuringiensis 1312 (BT1) and B. thuringiensis 1310 (BT2) and at 80°C for 10 min for B. licheniformis (BL). Afterward, endospores of each bacterial strain were collected by centrifugation at 7673 g for 10 min. The resulting pellet was suspended in 2 mL of sterile distilled water. The quantities of spores were estimated using the Neubauer Chamber cell counting method and dilutions were made and adjusted to reach a final concentration of 105 spores mL−1. The endospore suspensions were then used in the in vitro assay of seed germination, greenhouse pot experiments, as well as in the field experiment. This resulted in four different treatments for experiments with endospores: SBT1 = B. thuringiensis 1312, SBT2 = B. thuringiensis 1310, SBL = B. licheniformis and endospore consortium = SM.
Seed inoculation with bacterial inoculants
Oat seeds (A. sativa) were soaked in 5% sodium hypochlorite (NaClO) solution for 2 min and then thoroughly rinsed five times with sterile distilled water. Sterilized seeds were incubated for 30 min and placed on a shaker at 150 rpm in the eight respective bacterial treatment suspensions (i.e. VBT1, VBT2, VBL, VM, SBT1, SBT2, SBL and SM) under sterile conditions in order to allow for the cells or spores to adhere to the seeds. Control treatment consisted in surface sterilized seeds treated in the same way, but using sterile distilled water instead of bacterial suspensions.
Bacterial adhesion onto seeds
To assess bacterial adhesion onto seeds (each of the eight different treatments and the control), three seeds were randomly sampled in order to count bacterial cells or endospores adhering at their surface by flow cytometry. Seeds of each individual treatment were placed in sterile Eppendorf tubes containing 2 mL of sodium hexametaphosphate (NaPO3)6 and shaken vigorously with a vortex mixer for few seconds. For the vegetative cell treatment, 10 µL of SYBR Green was added to fluorescently label the cells. This treatment was not applied for endospore treatments because SYBR Green is unable to stain dormant cells (Zheng, Xiong and Wu 2017).
In addition to flow cytometry, scanning electron microscopy (SEM) was used to assess adhesion of vegetative cells or endospores onto the seeds. Uninoculated control seeds and seeds treated with both consortia (vegetative cells or endospores) were prefixed in 2.5% glutaraldehyde (Merck, Darmstadt, Germany) for 1 h followed by washing twice with sterile distilled water. Secondary fixation was carried out by using 4% osmium tetroxide (OsO4, Merck, Darmstadt, Germany) for 1 h, followed by two washing steps with sterile distilled water. Then, stepwise dehydration in graded alcohol was carried out by incubating samples for 15 min at each of the following ethanol concentrations (50%, 75%, 90% and 100%; Merck, Darmstadt, Germany). After this, the samples were dipped in tetramethysilane (C4H12Si; TMS, Merck, Darmstadt, Germany) for 20 min with a second step of 1 h until complete TMS evaporation. The samples were mounted onto SEM stubs covered with graphite and coated with gold with a sputter coater (Bal-Tec sputter coater SCD 005). The seeds were then examined using high-vacuum SEM (Philips ESEM XL30 FEG, Philips, Amsterdam, Netherlands) at an acceleration of 10 KeV and a working distance of ∼8.6 mm.
In vitro seed germination assay
In order to determine the effect of the different bacterial treatments on oat seed germination, an in vitro assay was performed. Five seeds inoculated with each of the eight bacterial treatments were grown aseptically on sterile filter paper. Filter papers were kept moist with sterile distilled water and incubated at 22°C in the dark. Uninoculated seeds were used as control. Experiments were replicated five times such that 25 seeds were assessed for each treatment. After 10 days, the number of germinating seeds was counted, and percentage of seed germination was calculated based on the number of seeds that germinated divided by the total number of seeds placed in Petri dishes.
Greenhouse pot experiments
Pot experiments were conducted in a greenhouse located at the University of Neuchâtel, Switzerland, to study the effect of each of the eight bacterial treatments (i.e. VBT1, VBT2, VBL, VM, SBT1, SBT2, SBL and SM) on growth of oat plants in soil. Two different experiments were carried out, either with a sterile substrate or a non-sterile soil, in order to compare plant growth-promoting activities of the different type of inoculation (single strain inocula versus consortium, as well as vegetative cells versus endospores). This approach was used to observe how the inoculants perform in a simple (sterile substrate) and a more complex (non-sterile soil) environment. The sterile substrate was composed of 25% peat, 25% Seramis® Pflanz-granulat (clay granules; Seramis, Mogendorf, Germany) and 50% sand (mixed carbonate and silicate; Coop, Basel, Switzerland). A thin layer of this substrate was sterilized by five successive autoclave cycles. Sterile substrate was then added in sterile pots (122 g in 130 mL pots). For the non-sterile treatment, soil was collected from a grassland nearby the University of Neuchâtel, Switzerland. The soil corresponded to a calcaric Cambisol (WRB 2015). Pots corresponding to nine treatments (eight bacterial treatments and an uninoculated control treatment) were arranged randomly in the greenhouse. One seed was planted in each pot. For the sterile substrate, 7 replicates per treatment were performed, while for non-sterile soil 10 replicates were performed. A larger number of replicates in the second case were considered given the potential for larger variability in the results when using soil as substrate. Pots were irrigated regularly with 0.5× concentrated Murashige and Skoog (MS, Duchefa Biochemie, Haarlem, Netherlands) solution for the sterile artificial substrate and tap water for the non-sterile soil. The numbers of seedlings that germinated and eventually grew into plants that were used for the experiment are indicated in Table S1 (Supporting Information). Plants from each individual treatment were harvested after 45 days. Plants (shoots, roots, flowers and pods) were dried at 60°C for 72 h and their total dry weight determined using a fine-scale balance.
Field experiment
To study the effect of both consortia, composed of either vegetative cells or endospores, on oat plants growth, a field experiment was conducted at a designated research field of the University of Neuchâtel, Switzerland. This was carried out in collaboration with the GRAMU association (Groupe d'Aménagement de l'Unine) involved in various projects related to sustainable agriculture and permaculture. Based on the results of the greenhouse pot experiments, we selected only three treatments for the field experiment: an uninoculated control, the vegetative cells consortium (VM) and the endospores consortium (SM). The selection of a limited number of treatments was made to allow for enough replication of the experiments in the field. The total area of the field was 5 m × 1.35 m, which was subdivided into nine plots to get three replicated plots of each treatment (Figure S1, Supporting Information). The position of the plots for each treatment was randomized. Each plot consisted of a length of 40 cm with a consecutive gap of 18 cm in which no treatment was applied. In each subdivided plot, 23 seeds were initially sown. The final numbers of seedlings that germinated and eventually grew into a plant and that were used for the experiment are indicated in Table S1 (Supporting Information). Plots were irrigated regularly, and weeds were manually removed. Plants (consisting of shoots, roots and seeds) were harvested after 85 days and dried at 60°C for 72 h. The total dry weight of the plants was measured using a fine-scale balance and the number of seeds per plant were counted. Physicochemical analyses (pH, water content and water holding capacity, concentrations in bioavailable and total phosphorus, nitrates, ammonium, organic carbon as well as calcium carbonate) were carried out on the soil of the field experiment to characterize its fertility. These measurements were done before and after the experiment for each of the three treatments (control, VM and SM; Table S2, Supporting Information).
DNA extraction
In order to study the effect of bioinoculation on native bacterial and fungal soil communities in the field experiment, different samples (bulk soil, rhizospheric soil and roots) were compared. Bulk soil consisted of soil that was distant from the roots influence. This fraction was analyzed before sowing and after harvesting. Rhizospheric soil consisted of the soil that was adhering to the roots upon sampling (Luster et al. 2009), and was obtained by detaching soil from the roots by vigorously shaking them. Finally, to obtain the root sample, the roots were washed with sodium hexametaphosphate ((NaPO3)6; Merck, Darmstadt, Germany) to remove any remaining soil particles. With this treatment, we assumed that microbes strongly attached to the root surface, as well as root endophytes were present (Reinhold-Hurek and Hurek 2011). DNA extraction was performed in triplicates for each treatment. DNA was obtained from 3 g of soil (bulk and rhizospheric soil) and 0.5 g of root samples. Fungal DNA was obtained by following the protocol of the FastDNA®SPIN kit for soil according to the manufacturer's instructions (MP Biomedicals, Waltham MA, USA; direct DNA extraction). Bacterial DNA was obtained using an indirect DNA extraction method previously developed to enhance the extraction of resistant structures such as endospores (Wunderlin et al. 2013). This method consists of an initial pre-extraction from the environmental samples, resuspending the soil twice in 10 mL of 1% (NaPO3)6 for the initial pre-extraction step (indirect method), followed by a modified DNA extraction protocol with the FastDNA®SPIN kit for soil (MP Biomedicals, Waltham MA, USA), with three successive bead-beating steps in the initial part of the protocol. DNA concentrations were measured with a Qubit Fluorometer using the dsDNA HS Assay Kit (Invitrogen, Carlsbad CA, USA). The concentration of all samples was adjusted to 2 ng μL−1 by diluting with double distilled sterile water.
Microbial community analyses
Amplicon sequencing was performed on an Illumina MiSeq platform, using the services of Fasteris SA (Geneva, Switzerland). For the bacterial community analysis, the regions V3-V4 of the 16S rRNA gene were targeted using universal primers 341F (5’-CCTACGGGNGGCWGCAG-3’) and 805R (5’-GACTACHVGGGTATCTAATCC-3’) (Herlemann et al. 2011). For the fungal community analysis, the 18S rRNA gene was targeted using the primer pair AMV4.5NF (5’-AAGCTCGTAGTTGAATTTCG-3’) and AMDGR (5’-CCCAACTATCCCTATTAATCAT-3’), which have been specifically designed for the analysis of arbuscular mycorrhizal fungal communities (AMF; Glomeromycota) (Sato et al. 2005). In total 36 samples were sent for sequencing of both markers. However, for the bacterial V3-V4 region, sequences were obtained only for 31 samples. The five missing samples consisted in one replicate of the control and two replicates of VM-treated post-harvest bulk soils, as well as two replicates of the root fraction controls.
The Illumina reads were processed with Mothur (Schloss et al. 2009) following the MiSeq standard procedure (Kozich et al. 2013), and using the SILVA NR v123 (16S rRNA) and SILVA NR v132 (18S rRNA) reference databases (Quast et al. 2013) for the alignment of amplicons and the taxonomic assignment of representative OTUs (operational taxonomic unit). Chimeras and singletons were removed prior to OTU clustering. For the 16S rRNA sequencing, average neighbor clustering of the 535 919 retained sequences (79 430 unique sequences) at 97% identity led to the identification of 6475 OTUs. For the 18S rRNA sequencing, clustering the 7 978 155 retained sequences (54 754 unique sequences) at 97% identity using the opticlust algorithm (Westcott and Schloss 2017) led to the identification of 11 466 OTUs. In addition, as a proxy for the detection of the inoculated strains at the end of the experiment, the bacterial sequences were used to query for near to exact matches to the 16S rRNA gene sequences of the inoculated strains. Bacterial sequences were queried using the strains 16S rRNA gene fragment matching the sequenced region. Those sequences were extracted from the draft genomes of the three strains. The Smith–Waterman algorithm (Smith and Waterman 1981) was used to check for near identical matches (identity of 99% or above). The number of sequences matching the queries under this stringent cut-off was used to calculate the relative abundance of the inoculated strains in the different samples.
Although the primers used for analyzing the fungal community were specifically designed for the analysis of AMF (Sato et al. 2005), sequences assigned to other divisions were also found to be highly represented, mostly in the Basidiomycota (Figure S2, Supporting Information). This contrasts with a previous study evaluating the potential of using these primers for targeting the AMF community in root samples, which showed a high specificity of these primers (Van Geel et al. 2014). The differences in the soil compartments analyzed (bulk soil, rhizosphere soil and roots in the present study) and the use of different sequencing technologies with different sequencing depths (Miseq VS 454 pyrosequencing) might partly explain these results. Sequences were submitted to GenBank under the Bioproject accession numbers PRJNA472865 for bacterial and fungal communities.
Statistical analysis
All statistical analyses were computed using R (version 3.4.3) (R Core Team 2014). Plant data from the greenhouse pots and field experiments were analyzed for statistical significance (P < 0.05) using one-way analysis of variance (ANOVA) followed by a Tukey's post hoc test. Comparisons of in vitro seed germination were analyzed with binomial test followed by Tukey's post hoc test. Bacterial and fungal community analyses were performed using phyloseq and vegan packages (McMurdie and Holmes 2013; Oksanen et al. 2017). Bacterial and fungal communities were analyzed by principal coordinates analysis (PCoA), based on Bray–Curtis dissimilarity and Hellinger transformation of the OTU table. OTUs accounting for less than four reads in the whole dataset were removed prior to the analysis. Comparisons between groups (soil compartments and treatments) were performed by permutational multivariate analysis of variance (PERMANOVA) using the adonis function from the vegan package. Analysis was based on the same dissimilarity matrix as described above, with 5000 permutations. Additional pairwise comparisons were performed using the function pairwise.adonis from the package pairwise adonis (Martinez Arbizu 2019), with 5000 permutations and Holm correction. Finally, the effect of treatment was analyzed with and without controlling for the soil compartment (use of the ‘strata’ parameter in the adonis function).
RESULTS
Screening of plant growth-promoting activities of the three bacterial strains
The screening for PGP traits of the three bacterial strains showed that all of the strains possessed all of the tested traits. All three strains were able to grow on nitrogen-free medium, showing their ability to fix atmospheric nitrogen. They were also able to utilize organic nitrogen, as demonstrated by their ability to solubilize casein. Moreover, all of them produced siderophores and auxin-like phytohormone compounds. Furthermore, the three stains were able to grow as a consortium as it was observed that the individual growth of each strain was not affected when grown as a co-culture under laboratory conditions (Figure S3, Supporting Information).
Assessment of bacterial inoculants adhesion onto seeds
The attachment of bacterial cells onto the seeds was quantitatively measured by flow cytometry. The number of cells attached per seed was in the range of 103 cells/seed (Fig. 1A). A slightly higher number of cells was observed on seeds treated with vegetative cells as compared to endospores, with the exception of the endospore consortium. The number of cells on seeds treated with the consortium of either vegetative cells (VM) or endospores (SM) was also assessed qualitatively by SEM observation (Fig. 1B and C). The number of bacterial cells attached to the seeds was higher for the treatment with vegetative cells compared to the seeds treated with endospores (Table S3, Supporting Information). No bacterial cells were observed on the untreated control seeds.
Assessment of the number of cells or spores attaching to A. sativa seeds after bioinoculation. (A) Number of vegetative cells or endospores attached per seed measured by flow cytometry (Control = uninoculated; V = vegetative cells; S = endospores; BT1 = B. thuringiensis 1312; BT2 = B. thuringiensis 1310; BL = B. licheniformis; Consortium = M). SEM images of seeds bioinoculated with either (B) the consortium of vegetative cells or (C) the consortium of endospores. Arrows show bacterial cells or spores attached onto the seeds.
In vitro seed germination assay
The in vitro seed germination assay showed that all three bacterial strains enhanced seed germination. Inoculation solely with vegetative cells of B. thuringiensis 1312 (VBT1) had the highest effect on seed germination (92%), which was significantly (adj P < 0.05) different from the control (52% of seed germination) and other treatments with vegetative cells (Fig. 2A; Table S4, Supporting Information). The vegetative cell consortium (VM) also induced higher germination than the untreated control (72% seed germination), although this difference was not statistically significant. In seeds treated with endospores, a trend of increased seed germination was observed in all treatments. However, only the treatment with B. thuringiensis 1310 (SBT2; 92%) and the consortium (96%) were significantly different from the control (Fig. 2B; Table S4, Supporting Information). Overall, the percentage of germinating seeds increased with inoculation using either individual strains or consortia as compared to the uninoculated control treatment.
Percentage of A. sativa seed germination comparing the effect of seeds treated with individual inocula and consortia of both (A) vegetative cells and (B) endospores. Different lowercase letters denote statistically significant differences among different treatments of bacterial inoculation at the 5% level according to Tukey's post hoc test (n = 5). Control = uninoculated; V = vegetative cells; S = endospores; BT1 = B. thuringiensis 1312; BT2 = B. thuringiensis 1310; BL = B. licheniformis; consortium = M.
Greenhouse pot experiments
The effect of individual strains on the total dry weight of oat plants was not statistically different from the untreated control treatment regardless of the mode of inoculation (vegetative cells or endospores) (Fig. 3). This result was consistent in experiments using both a sterile substrate (Fig. 3A) and a non-sterile soil (Fig. 3B). In contrast, inoculation with a consortium of the three selected Bacillus strains had a positive effect on plant growth, and this was the case for both the vegetative cells and endospore consortia. However, this effect was less prominent in the experiments with non-sterile soil as compared to the sterile substrate (Fig. 3). Inoculation with consortia showed a significant increase (adj P< 0.05; Table S5, Supporting Information) in total dry weight of oat plants compared to uninoculated control plants. No significant difference was observed when comparing the type of consortium (vegetative cells or endospores; Fig. 3; Table S5, Supporting Information).
Effects of individual inocula and both consortia (vegetative cells and endospores) on total dry weight of A. sativa grown in pots using (A) a sterilized substrate (n = 3 to 7; Table S1, Supporting Information) and (B) a non-sterilized soil (n = 8 to 10; Table S1, Supporting Information). The bar represents the average dry weight of the number of replicates per treatment (Table S1, Supporting Information) and error bars indicate standard deviation to this average value. Different lowercase letters denote statistically significant differences among different treatments of bacterial inoculation at the 5% level according to Tukey's post hoc test. Control = uninoculated; V = vegetative cells; S = endospores; BT1 = B. thuringiensis 1312; BT2 = B. thuringiensis 1310; BL = B. licheniformis; consortium = M.
Field experiment
Given the results obtained in the greenhouse experiments, only the consortia were evaluated in the field experiment. The plants inoculated with the vegetative cell consortium (VM) showed a significant increase in total dry weight compared to the uninoculated control (Fig. 4A; Table S6, Supporting Information). The plants treated with the VM consortium also produced significantly more seeds compared to the uninoculated control (Fig. 4B). Inoculation with the vegetative cell consortium (VM) did not show a significantly greater beneficial effect of plant growth promotion compared to the endospores consortium (SM).
Plant growth-promoting effects of both consortia (vegetative cells and endospores) on A. sativa plants grown in the field. (A) Total dry weight of oat plants and (B) total number of seeds produced per oat plant. The bars represents average dry weight and number of seeds of 28 plants (the number of plants in individual plots appears in Table S1, Supporting Information) in each respective treatment. Error bars represent standard deviation to the average value. Different lowercase letters denote statistically significant differences among different treatments of bacterial inoculation at the 5% level according to Tukey's post hoc test (n = 28). Control = uninoculated; VM = consortium of vegetative cells; SM = consortium of endospores.
Microbial community analyses
The effect of bioinoculation with both consortia on the autochthonous bacterial and fungal communities in different soil compartments (pre-sowing and post-harvest bulk soil, rhizospheric soil and roots) was evaluated. PCoA computed from the bacterial communities (Fig. 5) exhibited four significant axes according to the Kaiser–Guttman criterion, while the broken stick model retained only the first axis, indicating that axis 2 of the PCoA must be interpreted with caution. Axis 1 of the PCoA explained 34.4% of the variance and clearly separated two groups of samples: roots from bulk (pre-sowing and post-harvest) and rhizospheric soil (Fig. 5). Samples from the pre-sown bulk soil grouped together, indicating that the community composition was homogenous at the beginning of the experiment. Although there was some variability between samples from the rhizospheric and the post-harvest bulk soils (illustrated by their arrangement along the axis 2), there was no clear separation between samples from these soil compartments based on the PCoA. Comparing the three treatments (i.e. untreated control, inoculation with either the vegetative cell consortium (VM) or endospore consortium (SM)), no prominent difference was observed regardless of the soil compartment considered. A PERMANOVA analysis confirmed that the community was significantly different across the soil compartments, each pairwise comparison between groups being significant (Table S7, Supporting Information). On the contrary, comparison based on the treatment did not show any significant difference (Table S7, Supporting Information).
PCoA based on Bray–Curtis dissimilarity of bacterial community composition in bulk soils (pre-sowing and post-harvest), rhizospheric soil and roots. Percentage of the variation between the samples explained by each axis is indicated on the respective axis. The shape of the symbol represents the different treatments (Control = uninoculated; VM = consortium of vegetative cells; SM = consortium of endospores). Colors represent the different soil compartments. For the post-harvest bulk soil libraries prepared for the control and VM treatment, only two and one replicate yielded sequence information, respectively. The same was true for root samples in the control treatment, in which only two replicates were included in the analysis.
The relative abundance of OTUs assigned to Bacillus spp. was also analyzed (Fig. 6). OTUs affiliated to Bacillus spp. represented only a minor fraction of the community (3–6%). The highest relative abundance was observed in the case of the rhizosphere of VM-treated plants (Fig. 6). A BLAST search with the OTUs corresponding to Bacillus showed that all of them were between 96% and 100% identical to B. thuringiensis or B. licheniformis (Table S8, Supporting Information). Moreover, a refined search was performed in order to identify identical sequences corresponding to the 16S rRNA gene of the inoculated strains. We found a 5–8-fold potential increase of the inoculated strains in rhizosphere (VM treatment), post-harvest and root (SM treatment) compartments, relative to the pre-sown conditions (Figure S4, Supporting Information).
Relative abundance of the 12 OTUs assigned to Bacillus spp. in pre-sowing bulk soil (bulk soil before the experiment), post-harvest bulk soil (bulk soil after the experiment), rhizospheric soil and roots. The different colors represent the different OTUs.
Analysis of the fungal communities yielded a similar trend to the one observed for the bacterial communities. Only the first PCoA axis was significant based on the broken stick model, compared to 11 axes determined using the Kaiser–Guttman criterion. Axis 1 of the PCoA explained 22.7% of the variance and separated the same two groups of community components: root-associated fungi versus soil-associated fungi (including rhizospheric soil; Fig. 7). Similar to the bacterial communities, fungal communities in the pre-sown bulk soil appeared homogenous. Samples from the rhizospheric and post-harvest bulk soils did not show a clear separation according to the soil compartment. Although PCoA tended to group samples belonging to the same soil compartment, some samples fell out of these groups. Moreover, no clear distinction was made based on the different treatments (i.e. untreated control, inoculation with the vegetative cell consortium (VM) or inoculation with the endospore consortium (SM)). Results of the PERMANOVA on the fungal communities are similar to those from the bacterial communities, and confirmed the observations made from the PCoA: communities are significantly different across the soil compartments, while treatment had no significant effect on the fungal community composition (Table S7, Supporting Information).
PCoA based on Bray–Curtis dissimilarity of fungal community composition in bulk soils (pre-sowing and post-harvest), rhizospheric soil and roots. Percentage of the variation between the samples explained by each axis is indicated on the respective axis. The shape of the symbol represents the different treatments (Control = uninoculated; VM = consortium of vegetative cells; SM = consortium of endospores) and the colors represent the different soil compartments.
DISCUSSION
In this study, a consortium of three Bacillus strains showed plant growth-promoting activities in vitro and they were also effective in promoting the growth of oat plants both in pots (sterile substrate and non-sterile soils) and in the field. There are several studies investigating the effect of bacterial inoculation on plant growth, either directly in the field or by comparing their efficiency under controlled conditions (García et al. 2004; Mishra et al. 2009; Ali et al. 2011; Sharma, Ramesh and Johri 2013; Masciarelli, Llanes and Luna 2014; Kuan et al. 2016; Samaniego-Gámez et al. 2016; Yasmin et al. 2016; Hernández-Montiel et al. 2017). However, there are fewer systematic studies that span from the selection of bacterial inoculant to testing their effect on seed germination and also testing their ability to promote plant growth under different conditions, from sterile substrate and non-sterile soil in controlled condition to field trials. Likewise, evaluating the effect of microbial inoculants on native bacterial and fungal communities is seldomly done. Therefore, to the best of our knowledge, this study presents the first thorough investigation that combines both approaches.
This study started with the in vitro screening of plant growth-promoting traits in 15 mesophilic Bacillus strains of different origins from the University of Neuchâtel microbiology laboratory's bacterial collection (data not presented). Among these bacterial strains, three strains, BT1, BT2 and BL, showed positive results for a maximal number of traits and were thus selected for further in planta experiments, irrespective of the fact that two of them originated from Chile (Atacama Desert) and were identified as belonging to the same species. The key element that was considered for the selection was functional redundancy (multiple strains providing the same plant growth-promoting trait), rather than their phylogenetic affiliation. It may be assumed that two strains of a same species (as well as closely related species) are functionally similar. However, it is well known that defining a species in bacteria is difficult, in particular in the genomic era (Konstantinidis, Ramette and Tiedje 2006), and that strains of a similar species may have completely different metabolic capabilities (Wielbo et al. 2010). Furthermore, functional redundancy is defined as the ‘ability of one microbial taxon to carry out a process at the same rate as another under the same environmental conditions’ (Allison and Martiny 2008). Therefore, we hypothesized that using closely related species within a consortium is a trade-off that has the advantage of having some genetic variation, while ensuring functional redundancy in plant promoting traits. While functional redundancy is a concept that is widely recognized in microbial ecology, to the best of our knowledge, it is rarely used with reference to the inoculation of plant growth-promoting microbes in the form of a consortium.
Inoculation of oat seeds with single strains or a consortium of three bacteria increased germination rate as compared to uninoculated seeds. Therefore, direct seed coating seems to have beneficial effects on seed germination. This could be due to the ability of the selected strains to produce auxin-like hormones. There are many reports showing that Bacillus spp. are able to enhance plant growth by producing different plant growth hormones such as gibberellins, indole acetic acid (IAA) and cytokinins (Arkhipova et al. 2005; Radhakrishnan and Lee 2016). It has also been reported that some Bacillus spp. were able to produce more than one growth hormone, with a beneficial effect on the overall physiology of plants (Kumar, Prakash and Johri 2011). Moreover, we choose seed inoculation over foliar spray because the former introduces the required inoculants directly into the soil at the vicinity of the emerging roots. When the roots develop, beneficial bacteria are already present to stimulate plant growth. It has been reported that inoculation of seeds reduces seed-borne diseases caused by soil phytopathogens (O'Callaghan 2016). Furthermore, according to the findings of Ciccillo et al. (2002), the application method and density of the bacterial inoculants play a vital role on plant growth promotion. For instance, incorporation in soil of different concentrations (106, 107 and 108 CFU g−1 soil) of the plant growth-promoting strain Burkholderia ambifaria MCI 7 had a negative effect on plant growth, with the strongest reduction observed at 108 CFU g−1 soil. In the same study, same plant growth-promoting strain had a significant positive effect on the growth of maize plants when applied directly onto seeds at similar concentrations (106, 107 and 108 CFU per seed). However, bacterial inoculum concentration was not a significant predictor of plant biomass (Ciccillo et al. 2002). In our study, we found that oat seeds carried ∼103 cells of inoculated bacteria. Nonetheless, even at such low cell density, a positive effect toward oat growth was observed. This result is in accordance with a previous study where an inoculum of Bacillus amyloliquefaciens BNM122B was used for seed coating of soybean for protection against the plant pathogen Rhizoctonia solani (Correa et al. 2009). In that experiment, the initial concentration of the bacterial suspension was 108 CFU mL−1 and the final concentration of bacterial cells attached per seed was 104 CFU, showing that even at low cell densities the desired outcome can be obtained.
One of the most significant results obtained in the present study is that inoculation with a consortium of bacteria had a more robust effect on oat biomass than inoculation with individual bacterial strains in the pot experiments. Similar observations were also obtained in other studies using combined bioinoculants. For instance, a consortium of two Bacillus strains and Enterobacter cloacae increased tomato seedling growth and helped in disease suppression (Abdeljalil and Renault 2016). Moreover, it has been reported that inoculation with a consortium of R. tropici and two different strains of P. polymyxa elicits plant growth in presence of an abiotic stress (drought), in contrast to inoculation with R. tropici alone (Figueiredo et al. 2008). Finally, (Jetiyanon 2007) showed that a consortium of Bacillus spp. alleviated biotic stress caused by phytopathogens by producing defense-related enzymes.
In the field experiment, the consortia of either vegetative cells or endospores were compared, without including the comparison of individual strains. This was partly due to logistic limitations such as space and time to replicate plots at an agronomic scale. The choice of comparing both life forms rather than single inocula versus consortium was rationalized by the fact that in the greenhouse pot experiments in which nine different treatments were compared, consortia appeared to have more robust and consistent effects than single strains, whatever the cell type used (vegetative cells or endospores). Therefore, in the field experiment, the comparison of the two life forms of Bacillus cells and their effectiveness in plant growth promotion was investigated. This is of particular interest since Bacillus-based commercial products, for instance Kodiak, Rhizovital®42, FZB24® Bacillus subtilis, Serenade® and Quantum-400, contain in most cases spores, which are supposed to enhance the shelf life of commercial products (Radhakrishnan, Hashem and Abd Allah 2017).
In our study, we found that the bacterial inoculum applied in the form of vegetative cells had a more prominent effect in terms of total dry weight and grain yield of oat plants compared to the inoculum containing endospores, even though plants inoculated with endospores still grew larger than the untreated control plants. In a study comparing inoculation with vegetative cells and endospores of Bacillus subtilis EA-CB0575 (at concentration of 1 × 107 and 1 × 108 CFU mL−1, respectively), both the vegetative cells and endospores enhanced total dry weight of banana plants as compared to untreated controls (Posada et al. 2016). This suggests that both cellular forms can be equally used for bioinoculation technology. However, germination of spores in the soil is crucial for the strain to work effectively. In our study, the inoculation with endospores was less effective than the inoculation with vegetative cells and we hypothesized that this was the result of a low germination of the inoculated strains. Petras and Casida (1985) demonstrated that the germination of B. thuringiensis spores in the soil can be poor. They also noticed that despite the fact that spores had the ability to survive in the soil, germination was poor, and spores only germinated when plated on a nutritive medium in the laboratory. Likewise, Crane, Frodyma and Bergstrom (2014) observed that spores of B. amyloliquefaciens were unable to inhibit the growth of Fusarium graminearum (responsible for the fungal disease Fusarium head blight) on wheat. To overcome this problem, the authors suggested using nutrients to induce spore germination (a mixture composed of D-glucose, D-fructose and potassium chloride, along with amino acids, either L-asparagine or L-alanine). Finally, Omer (2010) compared a bioformulation of Bacillus megatherium and free endospore suspensions, finding that in field applications, an inorganic carrier increased the shelf life of the product and the bioformulation of B. megatherium with a phosphate stabilizer enhanced the viability and thus the efficacy of spores, as well as the growth promotion of bean plants, compared to the free endospore suspensions. Therefore, these studies demonstrate that spore germination is not always guaranteed in the field and that investigating this aspect is of interest to allow developing bioformulations with enhanced efficacy in the field.
Besides assessing the beneficial effect of bioinoculation on plant growth, another aspect often neglected in studies dealing with the use of bioinoculants is the impact of this practice on the indigenous soil microflora (Orhan et al. 2006; Noumavo et al. 2013; Kuan et al. 2016; Widnyana and Javandira 2016; Korir et al. 2017; Moustaine et al. 2017). Investigating whether the applied consortium has an impact on the structure of the native microbial community is crucial for allowing further usage of the inoculum in replacement of agrochemicals in sustainable agricultural ecosystems. It is important to consider that bioinoculants could potentially have a major impact on the long-term functional capabilities of the resident microbial communities by changing their overall structure (Trabelsi and Mhamdi 2013). Several studies have reported that bacterial inoculants alter the bacterial community composition after inoculation (Schwieger and Tebbe 2000; García de Salamone et al. 2012; Trabelsi et al. 2012). Although a previous study has shown that the plant growth-promoting rhizobacteria Bacillus sp. SUT1 and Pseudomonas sp. SUT19 did not alter the native fungal communities of Chinese broccoli (Piromyou et al. 2013), more recently, it was reported that inoculation of three Bacillus strains (B. cereus, B. subtilis and B. amyloliquefaciens) changed the endosphere bacterial communities of broccoli (Gadhave et al. 2018). In some cases, drastic shifts were observed, for instance, two biocontrol agents Pseudomonas corrugate IDV1 and Pseudomonas putida RA2, changed the composition of the autochthonous bacterial community from a Gram-positive- to a Gram-negative-dominated population in the rhizosphere of maize (Kozdrój, Trevors and van Elsas 2004). Such effects should not be ignored. One of the main factors explaining drastic changes in the resident bacterial community could be the high cell densities (up to 109 CFU mL−1) used in bioinoculation experiments (Kozdrój, Trevors and van Elsas 2004). In our study, we found that using a low bacterial cell density (103 cells per seeds) for the direct application of the inoculant onto seeds effectively increased the growth of oat plants, with no significant effect on the structure of the native bacterial and fungal communities (Figs 5 and 7). The Bacillus spp. fraction remained a minor fraction of the bacterial community even after the experiment and some of these community members were likely the bioinoculated strains as suggested by the outcome of the BLAST search (Fig. 6; Figure S4, Supporting Information). This again demonstrates that the bioinoculated strains were able to colonize the rhizosphere of oat plants, but not to outcompete native bacteria.
Importantly, the bacterial and fungal communities associated to the roots were found to be highly different from the communities of the other soil compartments (bulk and rhizospheric soils), showing that oat roots had a unique root-associated community. This is in accordance with previous studies investigating the differences in the microbiome of roots and other soil compartments. This was true for the bacterial microbiomes of the roots and the surrounding soil of maize (Niu et al. 2017). Similar observations regarding fungi were reported by Urbina et al. (2018) in a study comparing roots, rhizospheric and bulk soil fungal communities. Finally, both fungal and bacterial communities associated with Populus differed between roots and rhizospheric soil (Gottel et al. 2011).
CONCLUSION
In summary, a consortium of three Bacillus strains with plant growth-promoting activities positively affected oat seed germination and plant growth. This effect was less prominent in non-sterile soil as compared to a sterile substrate, highlighting the importance of competition with the native microbial communities and demonstrating that inoculated consortia can provide more robust plant growth promotion than single strain inocula. In sterile conditions (in vitro seed germination experiment an pot experiment with sterile substrate), consortia of both vegetative cells and endospores performed similarly. Conversely, in non-sterile conditions (pot experiment with non-sterile soil and field trial), vegetative cells showed slightly higher performances in promoting plant growth. Moreover, we also show that direct seed inoculation with bacterial consortia at low cell density did not lead to an alteration of the indigenous bacterial and fungal communities. Indeed, it appeared that the bioinoculated strains actually established in the rhizosphere of oat plants, but without a major effect on native microbial communities. This is of particular importance since a change in the indigenous communities induced by the inoculation could lead to unpredictable and undesirable effects on plant and ecosystem health. Therefore, we conclude that a consortium of three different Bacillus strains could be used as a low-cost and effective alternative for improving growth of oat plants. Finally, even though vegetative cells performed slightly better than endospores, both type of cells may be used to promote plant growth. This study may be helpful in the field of sustainable agriculture and could lead to more effective bacterial inoculants for promoting plant growth and health. Indeed, our results help expand the knowledge on how microbial consortia interact with plants and soil native microbial communities. This promotes a deeper and more comprehensive understanding of the factors required for the development and the formulation of next-generation bacterial inoculants to be used for sustainable agricultural ecosystems.
FUNDING
This work was supported by: the Swiss Government Excellence Scholarships for Foreign Scholars (number 2014.0865 to IH), the University of Neuchâtel Swiss National Science Foundation overheads (grant number 11.1 to SB), the ‘Fondation Pierre Mercier pour la Science’ (to SB), and the U.S. Department of Energy, Office of Science, Biological and Environmental Research Division, under award number LANLF59T (to PJ).
ACKNOWLEDGEMENTS
We would like to thank Geoffrey House for proofing the English language of the revised version of the manuscript.
Conflicts of interest. None declared.
