Diversity of bacterial strains in biochar-enhanced Amazon soil and their potential for growth promotion and biological disease control in tomato

The use of bacteria in growth promotion and biological control of plant diseases can minimize environmental contamination caused by the indiscriminate use of pesticides and chemical fertilizers. We aimed to evaluate growth promotion and biological control of Corynespora cassiicola in tomato seedlings mediated by beneficial bacteria isolated from a non-rhizospheric Amazon soil containing different amounts of biochar, and to identify to which groups of bacteria the strains belong. We obtained 200 strains of bacteria from experimental plots containing biochar doses of 0, 40, 80 and 120 t ha-1. Of these, 53 strains were selected by root colonization tests. Based on growth promotion parameters, 25 strains were screened, identified by molecular characterization and evaluated for indoleacetic acid (IAA) production, phosphate solubilization and biological control. The best dose of biochar for colony formation was 40 t ha-1, and a regression model indicated 34 t ha-1 as the optimal dose. The production of IAA was observed in 18 (75%) strains, and two (8%) strains were able to solubilize phosphate. The efficiency in root growth promotion was up to 125%, and the percentage of plant protection ranged from 50 to 59%. Molecular characterization showed that the bacteria used in this study belong to the genera Bacillus and Lysinibacillus.


INTRODUCTION
Tomato (Solanum lycopersicon Linnaeus) is one of the most commonly produced vegetables worldwide (Faostat 2017). In Amazonas state, Brazil, tomato cultivation is hampered by soil acidity and low natural fertility (Cerri et al. 2003) and requires high doses of limestone and chemical fertilizers, which significantly increase the costs of tomato production and discourage farmers. In addition, diseases caused by fungi are a limiting factor in tomato crops. Target spot, caused by the fungus Corynespora cassiicola (Berk and M. A. Curtis) C.T. Wei. is among the most important diseases that affect the aerial parts of tomato seedlings (Mandal et al. 2017). Corynespora cassiicola is non-specific and occurs mainly in the tropics and subtropics (Dixon et al. 2009), where environmental conditions favour disease development (Teramoto et al. 2017). In Brazil, there are still no specific fungicides recommended by the Ministry of Agriculture for the control of target spot in tomato crops. Therefore, the disease is controlled by fungicides recommended for soybean crops, such as Carbendazim Nortox ® and Comet ® , which have benzimidazole and strobirulin, respectively, as active ingredients (Agrofit 2020). These agents can cause harm to the environment, plants, animals and humans (Chaturvedi et al. 2013).
Beneficial bacteria in soils have potential to promote both biocontrol, because some produce natural fungicides to control phytopathogens, and plant growth, because some produce phytohormones (e.g., indoleacetic acid) and siderophores, and solubilize minerals, such as silicates, phosphates and potash (Naureen et al. 2017). Biochar applied to the soil can enhance colonization by beneficial bacteria, including plant growthpromoting bacteria (PGPB) (Bertola et al. 2019). Biochar can improve soil health with or without exposure to contamination by heavy metals and/or organic pollutants (Palansooriya et al. 2019). Biochar improves composting processes, as well as the biochemical properties of compost, by increasing the number of plant growth-promoting rhizobacteria (PGPR), which solubilize phosphate, produce indoleacetic acid (IAA) and degrade protein and cellulose. Biochar can also synergize the use of biofertilizers for promoting sustainable agriculture (Antonius et al. 2015).
Through increases in bacterial abundance and changes in microbial community structure, biochar soil-enrichment can exert a significant role on disease suppression and plant growth promotion, either through direct antagonism or indirectly via induction of systemic resistance in the plant (Jaiswal et al. 2018). Biochar amendments in soil can reduce the severity of bacterial wilt caused by Ralstonia solanacearum on tomato (Lu et al. 2016), induce resistance to the pathogens Botrytis cinerea and Leveillula taurica in both pepper and tomato, and the pest Polyphagotarsonemus latus in pepper (Elad et al. 2010), and improve considerably the growth of tomato plants, as they become more resistant to Fusarium oxysporum and Ralstonia solani (Khalifa and Thabet 2015).
Our aim was to evaluate growth promotion and biological control of C. cassiicola in tomato seedlings mediated by beneficial bacteria from a non-rhizospheric Amazonian soil containing different amounts of biochar, and to identify to which groups of bacteria the strains belong.

Influence of biochar on soil cultivable bacterial populations
The non-rhizospheric soils used were obtained at the Experimental Station for Tropical Fruit Culture of Instituto Nacional de Pesquisas da Amazônia (INPA), located at km 42 of the BR-174 highway, municipality of Manaus, Amazonas state, Brazil. The soil in the area is dystrophic yellow Oxisol with clay texture (> 60%). Soil samples were collected in 2017 from 25-m 2 plots that had been enriched with biochar ( Table  1) at concentrations of 0, 40, 80 and 120 t ha -1 in 2006 (one plot per treatment). A maize/cowpea crop rotation was started on the plots three months after biochar application, when a chemical fertilization (66 kg ha -1 of urea, 177 kg ha -1 of triple superphosphate and 100 kg ha -1 of KCl) was applied. Three samples were collected from each plot at depths between 0 and 10 cm, and mixed into a composite sample. About 300 g of each sample was packed into plastic bags and transported in a polystyrene box to the Phytopathology Laboratory at INPA. Bacterial strains were isolated by the serial dilution methodology (Silva and Romero 2004), with dilution factors varying from 10 -1 to 10 -4 . After dilution, strains were cultivated using solid 523 non-selective culture medium (Kado and Heskett 1970). Colonies were obtained according to the methodology of Silva and Romero (2004), using 10 -4 dilution factor aliquots of 100 μL, which were deposited in Petri dishes containing the culture medium and maintained at 28 °C in biochemical oxygen demand (BOD) during 24-hours of light. The results were expressed in colony forming units per ml (CFU ml -1 ) using the formula: R = a x 10 b CFU ml -1 , where R = result, a = average number of colonies per repetition and b = exponent of the dilution.
The experiment was conducted in a completely randomized design with four biochar treatments (0, 40, 80 and 120 t ha -1 ) and five replications. Each replication consisted of one Petri dish. Fifty colonies were selected for each treatment, totaling  (Kado and Heskett 1970) and the colonies were maintained at 28 ºC during 24 h of light in BOD.

Root colonization of tomato seedlings
After colony growth of the 200 selected strains, a bacterial suspension was obtained by adding 10 ml of saline solution (0.85%) to be used in the microbiolization step. The concentrations of the suspensions were adjusted by dilution, according to the correlation between optical density and number of CFU's, to 0.2 absorbance (Abs.) (540 nm), which corresponds to approximately 10 8 CFU ml -1 .
Untreated Santa Cruz Kada tomato seeds were disinfested by immersion in ethanol (50%) for two minutes, NaCl (2%) for four minutes and washing in sterilized water. The microbiolization was performed according to Silva et al. (2003). After this phase, seeds were sown in tubes containing 523 Kado and Heskett culture medium for 10 days. For each bacterial strain, three tubes containing two seeds each were used. In addition to the 200 strains, we used the rhizobacterium Bacillus cereus Frankland & Frankland, 1887 (UFV-101 strain) as positive control, as it has proven efficiency in root colonization (Romeiro et al. 2010). The presence of a halo around the root was used as an indicator of colonization. Each isolate was categorized as positive or negative for root colonization capacity. A light microscope with a magnification of 100x was used to observe bacterial biofilm formation.

Growth promotion of tomato seedlings
This experiment was performed at the Von der Pahlen Experimental Station for Vegetable Crops of INPA, and was divided into two steps. First, a preliminary selection was carried out among the positive strains for root colonization. The second step consisted of the confirmation test with the most promising strains. For both steps, the bacterial strains were grown in Petri dishes containing medium 523 (Kado and Heskett 1970), for 24 to 48 h. Tomato seeds (cultivar Santa Cruz Kada) were immersed in the bacterial suspension of each selected strain for a period of 24 h by microbiolization, then were sown in tubes (280 g) containing Vivatto Plus ® substrate (two seeds per tube), with 10 tubes per treatment. Thinning was performed seven days after germination, leaving one seedling per tube. The experimental design was completely randomized. Two controls were used, one negative, with microbiolized seeds in sterilized distilled water, and one positive, with microbiolized seeds in B. cereus (UFV-101), with 10 replicates each.
In the first step, the effect of the bacterial strains on tomato seedling growth was evaluated through plant height (PH), number of leaves (NL), stem diameter (SD) and dry mass of the aerial part (DMAP). In the second step, the most promising strains were reassessed through the same growth parameters (except NL), and also including the dry mass of the roots (DMR) and the total dry mass (TDM). PH was measured from the base to the apical bud of the seedling, using a millimetre ruler. NL was obtained by counting all fully expanded true leaves. SD was measured with a digital caliper (ZAAS). DMAP and DMR were obtained by weighing the aerial part and root dry mass, respectively. Samples were dried in a Digital Stove timer SSD 110L.
The growth promotion efficiency (GPE) was calculated for each variable and isolate selected in the second step using the data of both steps. GPE was calculated as ([G T -G C ]/ G C ) x 100, where G T is the growth parameter for the isolate, and G C is the growth parameter for the negative control, as described by Almoneafy et al. (2014). All evaluations were performed 20 days after sowing.

Indoleacetic acid production and phosphate solubilization
The strains selected in the second step of the growth promotion assay and B. cereus strain (UFV-101) were grown in Petri dishes containing 523 solid medium (Kado and Heskett 1970). After 24 h, they were transferred to test tubes containing 5 ml of TS medium enhanced with hydroxytryptophan (3 g tryptone, 0.1 g soy peptone, 1.6 g NaCl, 0.2 g hydroxytryptophan, and 200 ml sterilized distilled water), and amino acid similar to L-tryptophan produced in capsules containing 50 mg of 5-HTP. After 24 h under 130 rpm stirring, the medium was transferred to 10 ml Falcontype tubes.
The production of indoleacetic acid (IAA) was determined by the colorimetric method, following Bric et al. (1991), using the Salkowski reagent (0.5 M FeCl 3 .6 H 2 O and 35% HClO 4 ), and three replicates per strain. The tubes were centrifuged and 2 ml of the supernatant were placed in assay tubes with 1 ml (2:1) of the reagent. The strains were incubated in the absence of light for 20 min for the reaction to occur. A reddish color in the tube signaled the production of IAA by the bacteria.
The evaluation of solubilization was based on Katznelson and Bose (1959). Strains were cultivated and maintained at 28 °C for 15 days, with three replicates per strain. The colonies that formed a clear halo around them were considered calcium phosphate solubilizers. The diameters of the colonies and solubilization halos were measured to obtain the solubilization index (SI), using the formula: SI = Ø Halo (mm)/ Ø Colony (mm) (Berraquero et al.1976), where Ø = diameter. The bacteria were classified as low (SI < 2), medium (2 ≤ SI < 4) and high solubilizers (SI > 4). According to the starting time of solubilization, the bacteria were classified as precocious (solubilization onset before the third day) or late solubilizers (onset after the third day), and apparent non-solubilizers (that did not show visible solubilization until the 15th day of evaluation) (Hara and Oliveira 2004). VOL. 50(4) 2020: 278 -288 ACTA AMAZONICA

Biological control
This experiment was conducted in a greenhouse at EEH, aiming to control target spot caused by the pathogen C. cassiicola using the strains selected in the second step of the growth promotion assay. The INPA 2839 C. cassiicola strain was cultivated in potato-dextrose-agar medium (PDA). On the tenth day of growth, which coincided with the twentieth day of Santa Cruz Kada cultivar tomato seedling growth, the spore suspension was prepared with 1.1 x 10 5 spores ml -1 and applied to the seedlings using an atomizer. Subsequently, seedlings were placed in a humid chamber for 24 h. The biocontrol agents (bacterial strains) were added 20 days before the pathogen following the protocol used in the growth promotion trials. The experimental design was completely randomized, with 10 replicates for each bacterial strain and three controls, which consisted of (i) plant + B. cereus UFV-101+ C. cassiicola (positive control); (ii) plant + C. cassiicola (negative control); and (iii) plant + water (negative control).
The severity of the disease induced by C. cassiicola was evaluated on alternate days starting on the second day after inoculation with the pathogen and ending on the tenth day. Three leaflets per replicate were evaluated with the aid of an adapted Horsfall-Barratt diagrammatic scale (Oliveira et al. 2006) and classified according to the proportion of injured area as follows: 0 -no symptoms; 1 -< 1% injured area; 2 -1.1 to 3%; 3 -3.1 to 6%; 4 -6.1 to 12%; 5 -12.1 to 25%; 6 -25.1 to 50%; and 7 -> 50%.
The area under the disease progression curve (AUDPC) = Σ ((Y i +Y i+1 ) /2) (t i+1 -t i ) was calculated for each strain and control from the data obtained for severity, where Y = intensity of the disease, t = time, and i = number of evaluations in time (Campbell and Greaves 1990). The protection percentage (%) was estimated by the relationship: (1-x/y), where x = AUDPC of the treated plants, and y = AUDPC of the inoculated controls (Li et al. 1996).

Statistical analysis
The CFU count data were compared among treatments with regression analysis using a quadratic regression, which best fitted the data. The frequencies of positive and negative strains for root colonization were compared among soil treatments with a Chi-square test, and those for growth promotion, phosphate solubilization, indole acetic acid production and resistance induction were compared with a Fischer's exact test. The growth promotion variables were compared among strains using ANOVA, except PH and SD, which were submitted to a Skott-Knott test, and NL and DMAP, which were submitted to a Dunn test by the Kruskal-Wallis non-parametric analysis. Homogeneity of variance was assessed by Cochran's Q test and normality by the Shapiro Wilk test. The AUDPC means among replicates were compared among bacterial strains and controls using the Skott-Knott test and a significance level of 5%. All analyses were performed using ASSISTAT 7.7 beta (Silva and Azevedo 2016).
For the PCR reactions with both primers, the following concentrations were used: 100 ng total DNA, 0.2 pmol of each primer, 1X enzyme buffer (100 mM Tris-HCl (pH 8.8 at 25 °C)), 2 mM MgCl 2 , 0.4 mM dNTPs and 1.25 units of Taq DNA Polymerase. The reaction took place in a final volume of 25 μl. Amplification conditions included an initial denaturation at 95 °C for 5 min, 35 denaturation cycles at 94 °C for 1 min, annealing at 50 °C for 1 min and extension at 65 °C for 8 min, with a final extension at 65 °C for 16 min. At the end of the cycles, the reaction was maintained at 10 °C/∞. After amplification, fragments were separated in 1.5% agarose gel with electrophoresis and visualized (L-PIX CHEMI Molecular Imaging).
PCR products generated from the P027F/1492R primer pair were treated with polyethyleneglycol (20% PEG) and sequenced using the BigDye™ Terminator v3.1 kit on the 3500 Genetic Analyzer (Applied Biosystems™) according to the manufacturer's recommendations. The consensus sequence was obtained manually based on the sequencing of the F and R strands, and new sequences generated in this study were deposited in GenBank (http://www.ncbi.nlm.nih) under accession numbers MH547253 to MH547275.
The results obtained from the ERIC1F/ERIC2R primer pair reaction were analyzed using the PAST® Program (version 2.17c; Hammer et al. 2000), after binary data transformation and the construction of a 0-1 matrix, where 1 indicates the presence, and 0 the absence of a band. The bands generated for each strain were compared and their similarities estimated by the Jaccard coefficient, which was obtained by the unweighted pair group method with arithmetic mean (UPGMA) algorithm, and the strains were grouped and plotted using a similarity dendrogram (Sneath and Sokal 1973).
Dataset construction was completed with 16S rRNA region sequences from the strains obtained in this study. Sequences were obtained from GenBank (http://www.ncbi. nlm.nih) using the BLASTn tool. The sequences were aligned VOL. 50(4) 2020: 278 -288 ACTA AMAZONICA with the MAFFT online service (Katoh et al. 2017) and manually adjusted in MEGA 7.0 (Tamura et al. 2013).
Phylogenetic analyses were performed using the maximum likelihood (ML) and Bayesian inference (BI) methods. Partial deletion was used for the treatment gaps and missing data in the ML analysis. The 95% cut off and non-parametric bootstrap measurements were done with 1000 replicates, and the tree was generated and visualized in MEGA 7.0. Bayesian inference was based on the model selected by PAUP*4 and Mrmodeltest2 v2 (Posada 2003) through an alignment including all sites. The analysis was allowed to run for ten million generations, with the first 25% of trees discarded as burn-ins using the tool MrBayes v. 3.6, which is available on the CIPRES platform (https://www.phylo.org/). Posteriori probabilities (PP) and tree topologies were visualized with Figtree v. 1.1.2 (Rambaut 2009).
The identity analysis between the sequences was performed in the SDT v.1.2 program (Sequence Demarcation tool) by means of an array containing sequences of the strains obtained in this study and sequences obtained from GenBank (http: //www.ncbi.nlm.nih) by the BLASTn tool. The alignment MAFFT algorithm was selected to calculate the identity values, and the similarity of the phylogenetic relationships was estimated with the neighbor component using two cut off values, one at 99% and the other at 78%. These cut-off values represent the species demarcation thresholds (Kim et al. 2014).

Influence of biochar on the soil cultivable bacterial population
There was a significant difference in CFU among treatments according to regression analysis. Biochar stimulated bacterial growth up to 34 t ha -1 and from there on a decrease was observed in the bacteria population ( Figure 1). The estimated CFU for soil without biochar (dose 0) was 6.65 x 10 6 CFU ml -1 . The optimal dose of biochar estimated by the adjusted regression model was 34 t ha -1 , providing a maximum value of 7.55 x 10 6 CFU ml -1 . The adjusted regression model was ŷ = 6.656 + 0.0532x -0.0008x 2 , where ŷ is the estimated CFU value and x the biochar dose. The model explained 91.4 % of the total CFU variation in response to biochar doses (Figure 1).

Root colonization and growth promotion
Seventy (35%) of the initial 200 strains were positive for root colonization (Supplementary Material, Tables S1, S2), of which 53 were selected for the growth promotion test. Among the 53, eight (15%) significantly reduced the growth of tomato seedlings relative to the controls, 20 (37.7%) did not differ significantly from the control, and 25 had significantly higher DMAP (Supplementary Material, Table S3), and were considered the most promising bacterial strains. Among the 25 strains, SD differed significantly from the negative control in eight (32%), DMR differed significantly in ten (40%), and there was no significant difference in PH, DMAP and TDM (Table 2).
There was a significantly higher frequency of strains capable of colonizing the root system in the soils with 0 and 40 t ha -1 biochar (χ 2 = 28.92; p < 0.0001) (Supplementary Material, Table S2). The frequency distribution of the bacterial strains positive for growth promotion did not vary significantly among soil types (Fisher's p = 0.71709) (Supplementary Material, Table S4), i.e., the strains responded in the same way to the growth promotion test, independently of the biochar dose.

Indoleacetic acid production and phosphate solubilization
Eighteen of the 25 strains were found to produce IAA, while only two showed calcium phosphate solubilization capacity: 25T4 and 12T4 (Table 2). These strains formed a solubilization halo and were categorized as precocious, with a low capacity to solubilize phosphate (SI < 2). The two strains were from the treatment with 120 t ha -1 biochar, resulting in a relative frequency of 66.7% positive strains for solubilization of inorganic phosphate in the form of CaHPO 4 (Supplementary Material, Table S5). Accordingly, the frequency distribution of strains positive for phosphate solubilization varied significantly among soil types (Fisher's p = 0.01), i.e., there was an influence of the dose of biochar on the population of phosphate solubilizing bacteria present in non-rhizospheric soil (Supplementary Material, Table S5). It is worth mentioning that only in the soil with the highest dose of biochar (120 t ha -1 ) were found rhizobacteria belonging to the B. megaterium group (strain 25T4; Supplementary Material, Table S6) capable of solubilizing calcium phosphate in inorganic form.
The frequency distribution of the bacterial strains positive for production of indole acetic acid did not vary significantly  Table S7).

Molecular characterization
Of the 25 selected strains, 23 16S-rRNA regions were sequenced successfully. The phylogenetic analysis of 25 sequences was based on 1354 characters, including gaps, of which 30 were obtained from GenBank. The analysis was performed using the best-selected evolutionary model (HKY + G). The topology of the tree obtained was derived from the ML analysis plus the posterior probability values in the main branches containing three genera of the Bacillaceae family with twenty-one taxa (Figure 4). Of the 23 strains obtained, 20 were characterized as Bacillus, 17 belonged to the Cereus group, and three were characterized as Lysinibacillus, more closely related to Lysinibacillus sphaericus and Lysinibacillus macroides. Among the strains of the genus Bacillus that do not belong to the B. cereus group, 46T1 was more closely related to Bacillus altitudinis, and 25T4 and 12T4 to Bacillus megaterium, all with a high level of statistical support.
Based on the pairwise identity data (Figure 5), strain 114T1 showed > 99% identity with L. sphaericus and L. VOL. 50(4) 2020: 278 -288 ACTA AMAZONICA The dendrogram generated from ERIC-PCR had low similarity coefficients in some of the formed groups. Four main groups were formed, with their respective subgroups. Group I was subdivided into three subgroups. Strain 46T1 (subgroup Ia), which was identified as Bacillus sp. (pumilus group), did not group directly with the other strains of group I, corroborating the identification by sequencing of the 16S rRNA gene. Subgroups Ib, Ic and Id were composed exclusively of strains from group B. cereus. It is noteworthy that strains 17T3 and 22T3 showed the formation of only one band and two bands, respectively, which did not allow their correct grouping in the dendrogram, and provided low coefficients of similarity that do not corroborate the information obtained by sequencing the 16S rRNA gene ( Figure 6). Two strains, 3T4 and 4T2, were allocated into group II and identified as Bacillus sp. group B. cereus and Lysinibacillus sp., respectively. This group differed most from the others, as they grouped in the same clade and belonged to different genera, which differs from the information obtained from the 16S rRNA gene sequencing ( Figure 6).
Group III was divided into two subgroups, both containing only Bacillus strains. In subgroup IIIa, only strains belonging to the group B. cereus were grouped, with two groupings (53T1-80T2 and 15T2-6T3) with 100% similarity   macroides. Strains with identity > 99% belonging to the Bacillus clade were grouped in the Cereus group. Strains 8T2 and 3T4 were more closely related to Bacillus thuringiensis, and 6T3 and 30T2 to B. cereus. Five strains shared a 98% identity, and thirteen shared a < 97% identity (Supplementary Material, Table S6). VOL. 50(4) 2020: 278 -288 ACTA AMAZONICA among the strains. Strains 25T4 and 12T4, both identified as Bacillus spp. (megaterium group), also grouped (70% similarity) into subgroup IIIb, as observed in the phylogenetic analysis ( Figure 6).

DISCUSSION
Our results suggest that biochar enhancement in higher doses reduces the population of cultivable bacteria in nonrhizospheric soil and intere negatively with the survival of bacterial groups capable of actively colonizing the rhizosphere of tomato seedlings, while having beneficial effects on bacteria of the B. megaterium group capable of solubilizing phosphate. However, no selective effect of biochar dose was observed  on rhizobacteria capable of producing indole acetic acid or promoting growth or inducing resistance against C. cassiicola in tomato seedlings.
Despite the negative effects of higher doses of biochar on bacterial populations in general, and on bacteria capable of root colonization in particular, observed in this study, there are a few studies reporting the influence of biochar on specific groups of rhizobacteria that are capable of forming effective plant-rhizobacterial associations that provide plant growth (e.g. Egamberdieva et al. 2016;Nadeem et al. 2017;Ren et al. 2020). Therefore, more studies are necessary, testing a wider range of strains and experimental designs with more independent replicates for each treatment level, in order to further evaluate the effects of biochar enhancement on growthpromoting bacteria for tomato seedlings.
Biochar amendments increase the population densities of soil bacteria and actinomycetes, modify soil fungi/bacteria and fungi/actinomycetes ratios and increase soil microbial activity (Lu et al. 2016). However, higher rates may not be beneficial and can even become detrimental (Zwart and Kim 2012), which seems to have occurred in the soil analyzed here, as our results pointed to an optimal dosage for bacterial development below that of the lowest experimental biochar concentration. The dosage and type of biochar added to the soil can influence the morphology and topography of biofilms, as the binding force, or even biochar colonization, depends on the type of dominant molecule readily available on the surface of the biochar (e.g., phenolic components, silica and metal oxides). These act as chemical signals that induce cell lysis and biofilm formation, and promote interaction pathways that condition the adaptation and survival of bacterial species (Bueno et al. 2018). For example, the growth rate of Bacillus subtilis SL-13 in NB medium increased with the addition of biochar, because it contains nutrients for the growth of these bacteria and the special porous structure of the biochar has a positive effect on the adsorption of bacteria, being a potentially suitable carrier of PGPR for agriculture (Tao et al. 2018).
Molecular analysis showed that the three most promising strains for biological control are species of Bacillus (Cereus group) and Lysinibacillus, and those that showed the best results for growth promotion belong to Bacillus. A study in greenhouse conditions demonstrated that maize plants inoculated with B. subtilis and Lysinibacillus fusiformis in biochar-enhanced substrate presented better growth and nutrient concentration than biochar and bacterial treatments alone (Rafique et al. 2017). These improvements in plant growth were mainly attributed by the authors to phosphate-solubilization by the bacterial strains in the soil, phosphate from the biochar and IAA, cytokinin and gerbilline production.
Bacillus thuringiensis inhibited the growth of the C. cassiicola in-vitro and in-planta (Giau and Quoc 2017).
The production of antimicrobial lipopeptides synthesized in a nonribosomal mode is one of the possible means for Bacillus strains to use their antimicrobial action (Almoneafy et al. 2014). The fresh and dry mass of tomato plants were also enhanced by Bacillus strains (Almoneafyet al. 2014). Bacillus sp., B. amiloliquefaciens, B. pumilus and B. subtilis significantly increased the length, pseudostem diameter, fresh mass and dry mass in Prata Anã banana seedlings (Souza et al. 2017). Plant growth was also increased when a L. sphaericus strain was inoculated on Trigonella feonum-graecum (methi) and Vigna radiata (mung beans) seeds (Sharma and Saharan 2015). Genes coding for secondary metabolites, such as bacillibactin, bacilysin, microcin, bacillaene, difficidin, fengycin, macrolactin and surfactin, were found in bacterial strains that promoted plant growth and controlled multiple diseases (Liu et al. 2017), showing that mechanisms of growth promotion and plant protection may be genetically mediated.
The distribution based on pairwise identity corroborated the data obtained by phylogenetic analyses (ML and BI) for strain identification. However, for the Lysinibacillus clade, the relationship between L. sphaericus and L. macroides, which showed identity > 99%, was not well defined, making identification difficult through the 16S region due to the high homology between these taxa. For the genus Bacillus, the 16S region also had a low phylogenetic resolution (Janda et al. 2007). The 16S sequencing method has limitations, as in the closely related Bacillus anthracis, B. cereus and B. thuringiensis, which have identical 16S rDNA sequences, making their differentiation difficult using only this barcode (Han 2006) .
Strains that showed the highest identity with B. cereus and B. thuringiensis responded to the clustering tendency observed both in the pairwise identity analysis by the neighbor component and in the phylogenetic inferences, despite the low support of PP and bootstrap. According to Petti (2007), an identity of < 97% may indicate the existence of a new species.

SUPPLEMENTARY MATERIAL (only available in the electronic version)
Caniato et al. Diversity of bacterial strains in biochar-enhanced Amazon soil and their potential for growth promotion and biological disease control in tomato.