Rice (Oryza sativa L.) has been playing an important nutritional role for more than half of the world population, and about 486.67 million ton of rice was consumed worldwide in 2016/2017 (United States Department of Agriculture [USDA], 2018). Brazil is the world's 10^th largest rice producer and it is produced mostly in the wetlands ecosystem of the Southern region under flooded condition (USDA, 2018).

Rice cropping requires considerable amounts of nutrients for development and growth and, nitrogen is particularly important for rice cultivation. Association of this crop with several nitrogen-fixing bacteria might contribute a part of nitrogen required by the plant (Choudhury and Kennedy, 2005).

Nitrogen contribution derived from biological nitrogen fixation (BNF) is a promising alternative  to  circumvent  N losses associated to rice cropping. The BNF occurs in some prokaryotic organisms, such as, bacteria and cyanobacteria which are capable of converting N₂ into NH₃ (Baca et al., 2000; Baldani and Baldani, 2005; Kraiser et al., 2011).

The enzyme nitrogenase which is present in free-living or symbiotic diazotrophic bacteria is extremely versatile as it transforms N₂ into ammonia and catalyzes the reduction of several other substrates such as acetylene. As a consequence, the reduction of acetylene (C₂H₂) to ethylene (C₂H₄) became the basis of indirect estimation of nitrogenase activity (Boddey et al., 1990; Stoltzfus et al., 1997; Moreira and Siqueira, 2006; Mus et al., 2016).

Many genera of diazotrophs such as Azospirillum, Herbaspirillum, Burkhoderia and Sphingomonas have been isolated from rice and have shown positive effects on many rice varieties (Baldani et al., 2000; Tran Van et al., 2000; Baldani and Baldani, 2005; Rodrigues et al., 2006; Rodrigues et al., 2008; Pedraza et al., 2009; Araújo et al., 2013).

These microorganisms colonize the interior and exterior of plant organs exerting beneficial functions (in addition to BNF) to diseases either directly or indirectly by antagonizing the pathogenic fungi through siderophores, chitinase and antibiotics production as well as phosphate solubilization, phytohormones production, etc (Downing et al., 2000; Baldani et al., 2002; Lee et al., 2004; Videira et al., 2009; Figueredo et al., 2010).

Indole acetic acid (IAA) is the most abundant phytohormone that regulates various aspects of plant developmental functions (Teale et al., 2006) reflected by the complexity of their biosynthesis, transport and signaling pathways (Santner et al., 2009). Studies have revealed great variability of auxin production among these strains; also, their inoculation promotes plant growth (Asghar et al., 2002; Khalid et al., 2004; Kuss et al., 2007; Richardson et al., 2009).

Solubilization of phosphates is presented by many bacteria and fungi in soil and in association with different plant species altogether. The metabolic processes involved in P solubilization and mineralization include excretion of hydrogen ions, release of organic anions, siderophores and phosphatases (Richardson, 2001). Solubilization of phosphate compounds is very common in vitro, but it has not been reported frequently because of technical difficulties in uncontrolled environment (Khan et al., 2007; Estrada et al., 2013).

Rice cultivars EPAGRI-109 and BRS TROPICAL are recommended for planting in Rio de Janeiro State and their associations with plant growth promoting bacteria have been previously reported (Kuss et al., 2007; Viana et al., 2015); however, no data is available to diazotrophic bacteria associated with these cultivars in Rio de Janeiro State.

According to some authors, native isolates of diazotrophic bacteria may contribute to greater nitrogen gains   in   Poaceae   since   it    is    locally    adapted   to environmental conditions (Santos et al., 2015; Viana et al., 2015).

The work aimed to isolate, characterize and select diazotrophic bacteria associated with rice cultivars IR-42, EPAGRI-109 and BRS TROPICAL grown in Rio de Janeiro State.

Sampling

Two rice cultivars (O. sativa L.), EPAGRI-109 and BRS TROPICAL, and their rhizospheric soils were used for enumeration and isolation of diazotrophic bacteria. Samples from cultivar EPAGRI-109 were obtained at field condition while those of cultivar BRS TROPICAL were derived from seeds cultivated at greenhouse condition because during the isolation none was planted in the field.

EPAGRI–109 plants were collected during January 2013 100 days of growth (DAG) and 120 DAG during the vegetative growth stage in Campo dos Goytacazes, RJ, Brazil field.

Seeds of BRS TROPICAL were inoculated with Sp 245 to evaluate if the inoculation would alter bacteria groups associated with rice plants. The seeds of BRS TROPICAL were sown in a randomized experimental design in the greenhouse using 12 pots each containing 16 seeds and 20 kg un-sterile soil of an Ultisol soil collected from Embrapa Agrobiologia experimental field at Seropédica, RJ. The treatments were inoculated and un-inoculated with Azospirillum brasilense strain Sp 245 with 4 replicates.

Isolation of diazotrophic bacteria

EPAGRI-109 and BRS TROPICAL plants were individually wrapped in plastic bags and brought to the Laboratory of Grasses at Embrapa Agrobiologia, RJ for the extraction of soil adhered to roots (rhizospheric soil), shoots and roots. Diazotrophic bacteria were isolated using the serial dilution technique. 10 g rhizospheric soil was transferred to flasks containing 90 ml saline solution of K₂HPO₄ (100 mg L⁻¹), MgSO₄ (50 mg L⁻¹), NaCl, (20 mg L⁻¹), CaCl₂.2H₂O (50 mg L⁻¹), and FeEDTA (16.4 mg L⁻¹). The roots and shoots were washed in tap water and then surface disinfested in chloramine T (1%), chopped into 10-cm pieces and 10 g were transferred to flasks containing 90 ml saline solution. After homogeneization by blending, each sample was diluted up to 10^-6 level with saline solution. Thereafter, 0.1 ml of each dilution extract was used to inoculate vials containing 5 ml JNFb (Döbereiner et al., 1995), Nfb (Döbereiner et al., 1995) and LGI (Magalhães et al., 1983) semi-solid media. The diazotrophs population per gram of fresh tissue was enumerated by the Most Probable Number (MPN) technique according to the McCrady's table (Döbereiner et al., 1995).

Acetylene reduction activity (ARA)

Isolates from EPAGRI-109 were obtained using JNFb, NFb and LGI nitrogen free semi-solid media and those from BRS TROPICAL were obtained using JNFb and NFb semi-solid media. The positive growth was evaluated after the appearance of a typical diazotrophic bacterial pellicle in the subsurface of the medium after incubation for 7 to 10 days at 30°C. The bacteria were purified on potato agar medium (Döbereiner et al., 1995).

ARA was performed to estimate nitrogenase activity following Boddey et al. (1990). Isolates were cultured in NFb, JNFB or LGI semi-solid (without bromothymol blue) and incubated at 30°C, then evaluated after 24, 48, 72 and 96 h in different  flasks.  Flasks  were sealed with rubber caps, injected with 10% acetylene (v/v of gas phase) and incubated for 1 h at 30°C. Aliquots of 0.5 ml of gas samples were analysed through a Perkin Elmer F11 model gas chromatograph with Porapak N column (50 cm, 40°C). Diazotrophic bacteria A. brasilense Sp245, Nitrospirillum amazonense CBAmC and Herbaspirillum seropedicae ZAE94 were used as positive control. Total proteins were determined according to Bradford method (Bradford et al., 1976). Absorbance was recorded through a Labsystems iEMS Microplate Reader MF at 595 nm and protein contents were estimated using bovine serum albumin (BSA) calibration curve (7.5 to 150.0 μg ml^-1) as external standard.

Indolic compounds production

Total indolic compounds produced by isolates were quantified in microplates as described by Sarwar and Kremer (1995). For this test, 1 ml of bacterial culture previously cultured for 24 h in DYGS medium (Rodrigues Neto et al., 1986) was inoculated into 5 ml of DYGS medium supplemented with L-tryptophan at the final concentration of 100 μg ml^-1. The tubes remained in the dark under agitation of 150 rpm at a temperature of 30°C for 48 h. 1 ml aliquots were removed and centrifuged at 5000 ° g for 15 min.

In microplate 96-well, 150 μl aliquot of the supernatant was mixed with 100 μl of the Salkowski reagent (1 ml of 0.5 M FeCl₃ in 49 ml of 35% perchloric acid) previously prepared. The samples remained in the dark for 30 min at room temperature and the absorbance reading was done on a microplate reader (Labsystem IEMS Reader MF, Labsystem) to a wavelength of 540 nm. The quantification of indole compounds was evaluated using a calibration curve prepared with serial dilutions of synthetic IAA standards (5 to 100 μg tryptophan ml^-1). Cultures of Gluconacetobacter diazotroficus (PAL5), A. brasilense (Sp245), H. seropedicae (ZAE94) and N. amazonense (CBAmc) were used as positive controls. Concentration was estimated with a standard calibration curve of indole acetic acid (10 to 80 mg ml^-1). Values are presented as mg of indolic compounds per mg of proteins, determined by the Bradford method (Bradford et al., 1976). After the determination of ARA, 20 μl of pellicle and culture medium mix was added into 20 μl 1 M NaOH and 60 μl of sterile distilled water and placed in a water bath at 90°C for 5 min to promote cell disruption. Subsequently, 900 μl of the Bradford reagent was added, the tubes were mixed by vortex and incubated for 3 min at room temperature.

Phosphate solubilization

Phosphate solubilization was evaluated using NBRIP agar medium containing Ca₃ (PO₄)₂ (Nautiyal, 1999). Cells were grown in DYGS liquid medium at 30°C for 24 h at 150 rpm; thereafter, culture optical density was adjusted to O.D_0.9-1.0 at 600 nm. Aliquots (20 μL) of each culture were spotted onto NBRIP solid media, in triplicate. Growth and halo formation (mm) were measured at 48 h intervals during 15 d incubation at 30°C.

Molecular characterization

Total DNA extraction was performed using cetyl trimethylammonium bromide (CTAB) protocol as described in Sambrook et al. (1989). The isolates were inoculated in liquid DYGS medium at 30°C for 24 h under constant agitation and the bacterial culture was transfered to 2 ml microtube centrifuged at 4,000 x g for 8 min. The pellet was suspended in 567 μl T₁₀E₁ (10 mM Tris pH 8.0 and 1 mM NAâ‚‚EDTA) and homogenized vigorously; further, 30 μl of 10% SDS was added and homogenized by inversion with 3 μl of Proteinase K (20 mg ml^-1) and then incubated at 65°C for 20 min.

After 100 μl of 5 M sodium chloride was added and homogenized by inversion, an additional 80 μl of CTAB / NaCl (10% CTAB in 0.7 M NaCl) was added and the incubation was carried out at 65°C for 20 min. 700 μl phenol-chloroform-alcohol isoamyl alcohol (25:24:1) was added and left overnight. The material was centrifuged for 10 min at 4°C at 13,000 × g. The supernatant was transferred to a new microtube and 700 μl chloroform-isoamyl alcohol (24:1) was added and centrifuged for 10 min at 4°C at 13,000 × g. The supernatant was thereafter recovered and transferred to a new 1.5 ml microtube to which 0.6 volume of ice-cold isopropanol was added. The material was incubated at 20°C for 30 min. Afterwards, the material was again centrifuged for 10 min at 4°C at 13,000 × g and the supernatant was discarded. The material was further subjected to centrifugation for 10 min at 4°C to 13,000 x g, with 70% ice-cold ethanol. The material was dried at room temperature and then suspended in 100 μl T₁₀E₁ (10 mM Tris pH 8.0 and 1 mM NAâ‚‚EDTA).

The DNA quantification and qualification was assessed by electrophoresis on 0.7% agarose gel. 5 μl of sample was used together with 2 μl of sample buffer (0.25% bromophenol blue and 40% sucrose), along with 3 μl of 1 kb Plus DNA ladder standard (invitrogen® Cat. No. 10787-018). The samples were migrated in 80 v for 90 min in 1X TAE buffer (0.04 M Tris acetate and 1 mM EDTA). The gel was stained with an ethidium bromide solution (0.5 μg ml^-1) and visualized with ultraviolet light on KODAK® Gel Logic 100 Photodocumentator (KODAK Scientific Imaging Systems, Cat. No. 172.8468) and one computer. Gels were analyzed using the KODAK® 1D Image Analysis (KODAK Molecular Imaging Systems, Cat. Nr. 811.2344).

The 16S rRNA gene was amplified using the universal pair of primers: Amp-1: (5′-GAG AGT TTG ATY CTG GCT CAG-3′) and Amp-R (5′-AAG GAG GTG ATC CAR CCG CA-3′) (Wang et al., 1996). 

Amplification was carried out in 50 μl final reaction volume which contained 10%  (final volume) 10X PCR buffer; 3 μl MgCl₂ (50 mM); 1 μl dNTP (200 mM); 1 μl of Taq DNA polymerase (INVITROGEN®) (1.25U μl ^-1); 1 μl of native DNA (50 ng μl) and the volume made up to 50 μl with PCR grade water. The PCR schedule was: one denaturation cycle (95ºC for 5 min), followed by 34 intermittent cycles (94°C for 15 s, 60°C for 45 s and 72°C for 2 min), and one final extension cycle (72°C for 30 min), followed by cooling (4°C for 24 h).

For identification and phylogenetic analysis, PCR products of each isolate were purified using Wizard® genomic DNA purification kit (cat. A1120, PROMEGA Corporation, USA) and then sequenced at the Genomic Laboratory of Embrapa Agrobiologia. Sequences were deposited in the GenBank at accession numbers KT619165 - KT619177.

Inoculation experiment

The 12 most promising isolates (TR-I1, TR-N1, EP3-13J, EP4-2J, EP4-1L, EP3-2L, EP3-16N, EP4-8N, EP3-1L, EP3-7L and EP4-5J) based on plant growth promoting tests (Table 3) were selected to be tested in an inoculation experiment  for growth promotion of rice in the Laboratory of Grasses, Embrapa Agrobiologia, Seropédica, RJ in a completely randomized factorial design (3×20) with four replicates using three cultivars of irrigated rice (IR 42, EPAGRI-109 and BRS TROPICAL) and were inoculated with 12 diazotrofic bacteria isolates (TR -I1, TR-N1, EP3-1L, EP3-2L, EP3-7L, EP3-13J, EP3-14J, EP3-16N, EP4-1L, EP4-2J, EP4-5J, EP4-7J), two bacterial mixtures comprising of MIX 1 (EP4-2J, EP3-13J and EP3-14J) and MIX 2 (TR-I1, TR-N1, EP4-1L, EP4-5J and EP3-14J), positive controls H. seropedicae Z94 (BR11417), A. brasilense Sp245 (BR11005), N. amazonense CBAMC (BR11145) provided by Diabotrophic  Bacteria  Collection  of  Embrapa  Agrobiologa -  CRB Johanna Döbereiner and a commercial inoculant (AZOTOTAL), an uninoculated treatment and a nitrogen treatment with 2.60 mM of nitrogen as ammonium nitrate.

The bacteria were reactivated in 125-ml Erlenmeyer flask containing 50 ml DYGS medium, at 30°C agitation at 150 rpm for 24 h.The experiment was conducted in 120-ml test tubes containing 50 ml Hoagland's agar solution (6 g L^-1), without nitrogen. The tubes were covered with autoclaved cottons. After sterilization and prior to the solidification of the agar, each tube was inoculated with 2 ml pre-grown bacterial culture. The uninoculated treatment was inoculated with sterile culture medium. The seeds were disinfested according to Hurek et al. (1994) and pre-germinated on agar plates (1%) for 48 h. A pre-germinated seed was planted in each test tube. The experiment was terminated at 30 DAP, thereafter shoot and root dry weight were analyzed.

Statistical analysis

The normality (Lilliefors test) and homogeneity of variance (Bartlet's test) of the data were processed with SAEG 8.0 program (Euclydes, 1983). Analysis of variance was performed using the SISVAR 5.0 program (Ferreira, 2003) and a comparison of means by the Scott-Knott test at a 5% probability (Scott and Knott, 1974).

Isolation of diazotrophic bacteria

The MPN counts of cultivar EPAGRI-109 varied from 10⁴ cells g root frw ^-1 and 10⁶ cells g frw^-1 in rhizospheric soil while the shoot values were 10³ to 10⁴ cells g root frw^-1. Interestingly, root associated bacteria of un-inoculated plants of BRS TROPICAL of the greenhouse were similar to EPAGRI-109 grown under field condition. In addition, number of bacteria in shoots and rhizospheric soil were higher in plants inoculated with A. brasilense Sp245 (Table 1).

The microbial population intimately associated with the rice plants was of 10⁶ cells g frw^-1 in cultivar EPAGRI 109 and in un-inoculated BRS TROPICAL plants. When BRS TROPICAL plants were inoculated with A. brasilense Sp245 the MPN values for shoots grossly doubled over in un-inoculated plants. Pedraza et al. (2009) also observed bacterial MPN increased on phyllosphere of rice cultivars in response to inoculation of Azospirillum strains. The inoculation treatment affected the bacterial counts in shoots and rhizosphere because A. brasilense Sp245 effectively colonized and established at these sites and altered the population of diazotrophs.

We obtained isolates in shoots (10), roots (8) and rhizospheric soil (9) from BRS TROPICAL and only isolates in shoots (9) and roots (3) from cultivar EPAGRI-109 (Table 2).

Hardoim et al. (2011) observed that some rice cultivars have a strong influence on the population composition of bacterial communities in their rhizosphere, selecting similar bacterial communities while other genotypes select divergent bacterial communities. Both bacterial adaptation and plant genotype contribute to the formation of bacterial communities associated with rice roots. This behavior could also be explained by the development stage which influences different exsudates production in roots (Chaparro et al., 2014).

Reduction of acetylene activity

The isolates were screened for in vitro eficiency of nitrogen fixation in semi-solid media (NFb, JNFb and LGI) (Figure 1A and B). Isolates obtained from cultivar EPAGRI-109, fixed 0.36 to 1462.0 nmol C₂H₄ mg protein^-1 (Figure 1A). The isolates EP4-2J and EP4-3J presented ARA of 1021 and 1462 nmol C₂H₄ mg protein^-1, respectively. The ARA values were higher than those of positive control, that is, A. brasilense Sp245 (702 nmol C₂H₄ mg protein^-1), H. seropedicae Z94 (305 nmol C₂H₄ mg protein^-1) and N. amazonense CBAmC (53 nmol C₂H₄ mg protein^-1). The EP3-8J, EP4-4J, EP4-6J, EP4-7J, EP4-9N and EP4-10N isolates could not grow in semi-solid medium and were not evaluated by ARA. Isolates obtained from BRS TROPICAL cultivar showed ARA range of 3.16 to 416 nmol C₂H₄ mg protein^-1 (Figure 1B).

None of them presented ARA comparable to A. brasilense Sp245 (601 nmol C₂H₄ mg protein^-1) but TR-I1 presented ARA more than double to that of H. seropedicae Z94 (198 nmol C₂H₄ mg protein^-1). We selected 3 strains obtained from BRS TROPICAL (TR-I1, TR-N1 and TR-N7) and 15 from EPAGRI-109 (EP3-1L, EP3-2L, EP3-3L, EP3-4L, EP3-6L, EP3-7L, EP3-13J, EP3-14J, EP3-15N, EP3-16N, EP4-1L, EP4-2J, EP4-3J, EP4-5J and EP4-8N) based on the best ARA results. These isolates were further evaluated for indolic componds production and phosphate solubilization.

Although, N-free semi solid medium was used for diazotyroph isolation, all isolates from EPAGRI-109 were not  able  to  grow  and/or  reduce  acetylene  to  ethylene under the experimental conditions. Some isolates from EPAGRI-109 showed higher ARA than A. brasilense Sp245 (Figure 1A). In contrast, isolates from BRS TROPICAL showed lower ARA than A. brasilense Sp245, although 16S rDNA homology showed that TR-I1 is closely related to this species (Figure 1A, Table 4). Variations of ARA values among the isolates of the  same species have been reported previously by several authors (Staal et al., 2001; Videira et al., 2012; Estrada et al., 2013). Although an indirect technique, ARA was used for indirect nitrogen fixation ability measurement because it is inexpensive and practical but since several factors (such as pH, carborn source, temperature and dissolved oxygen) can interfere with the performance of the isolates it have been used mostly to qualify than quantify their BNF potential.

Production of indolic compounds

Production of indolic compounds varied among the isolates (EP3-13J, EP3-14J EP3-16N and EP4-8N) and positive control strain (H. seropedicae ZAE94). The H. seropedicae ZAE94 and the isolates EP3-13J, EP3-14J EP3-16N as well as EP4-8N produced indole compounds during the first 24 h, with most of them attaining optimum production at 72 h; however, isolate EP4-8N accumulated indole compounds only after 48 h (Figure 2).

Indolic compound production by isolates TR-N1, TR-N7, EP3-1L, EP3-7L, EP3-15N, EP4-1L, EP4-2J, EP4-3J and EP4-5J could be detected after 24 h, in which their maximum production was observed at 48 h and then decreased up to 72 h. Interestingly, TR-I1 and EP3-7L behaved similarly to A. brasilense Sp245 and accumulated small amounts of indolic compounds after 72 h, that is, EP3-2L after 72 h had minimal production after 72 h but A. amazonense CBAmC and isolates EP3-3L and EP3-4L did not produce indolic compounds (data not shown).

 Indolic compounds production by isolates was similar to those previously reported for rice plants isolates, an average production of 1.8 and 53 µg IAA ml^-1 (Kuss et al., 2007; Rodrigues et al., 2008; Araújo et al., 2013). After 48 h, amount of indole accumulation was lesser than after 72 h probably due to degradation of indole compounds as reported by Leveau and Gerards (2008). Based on positive control,  the organisms could be clustered into: Group 1 - Isolates that produced indole compounds comparable to H. seropedicae (EP3-13J, EP3-16N, EP4-2J and EP4-3J); Group 2 - Isolates that produced lower indoles than H. seropedicae ZAE94 but higher than that of A. brasilense Sp245 (TR-N1, TR-N7, EP3-14J, EP3-15N and EP4-5J); and Group 3 - Isolates that produced indoles comparable to A. brasilense Sp245 (TR-I1, EP3-1L EP3-2L, EP3-6L, EP3-7L, EP4-1L and EP4-8N).

Phosphate solubilization

41% of isolates was able to solubilize inorganic phosphates  in  vitro;  the  solubilization  halo  of   isolates varied from 0.33 to 2.45 mm during the four evaluations (Figure 3). These values are in the same range of those observed for A. brasilense Sp245 and N. amazonense CBAmC used as positive control. Phosphates solubilization was higher for G. diazotrophicus PAL5, Burkhoderia tropica Ppe8 and H. seropedicae ZAE94. However, no trend of P solubilization by the isolates could be established until 16^th day of evaluation. Some isolates produced greater solubilization halo on first day and others produced bigger halos of solubilization on the last day of analysis.

Phosphorus  (P)  is  important  in  several  physiological processes of plants, especially in photosynthesis, carbon metabolism and membrane formation, being a structural component of many coenzymes, phospho-proteins, phospholipids, DNA of all living organisms and responsible for the transfer and storage of energy which is used for growth and reproduction (Anand et al., 2016).

Despite this, about 95 to 99% of the phosphorus present in the soils is unavailable to the plants due to the fixation of P that is adsorbed on the mineral particles of the soil or precipitated by the ions Al³⁺ and Fe³⁺ in solution (Wu, 2005; Sharma et al., 2013; Anand et al., 2016).

Soil microorganisms play a key role in the soil P dynamics and subsequent soil P availability. Phosphates solubilizers, especially bacteria, increase the solubilization of insoluble phosphorus compounds through the release of phosphatic and phytase enzymes and enzymes that are present in various soil micro-organisms (Vassileva et al., 2000; Anand et al., 2016).

Molecular characterization

Amplification of 16S rDNA from all isolates generated fragments in the range of 1500 base pairs. The sequences varied in the range of 1208 to 1479 bp and Blast analysis revealed their phylogenetic relationship, based on similarity to sequences deposited at GenBank (GenBank accession numbers KT619165 - KT619177). The diazotrophic isolates belong to Alfaproteobacteria (Azospirillum and Nitrospirillum species), Betaproteobacteria (Herbaspirillum and Ideonella species) and Gammaproteobacteria (Pseudomonas species) within Proteobacteria. In addition, Firmicutes (Paenibacillus species) within Bacilli was also identified (Table 3).

Inoculation experiment

The results of diazotrophic bacteria inoculation under axenic conditions showed significant improvement in dry weight of the inoculated rice cultivars. The shoots of EPAGRI-109 were more responsive to inoculation than BRS TROPICAL and IR42 which did not significantly respond to any treatment (Table 4). The EP3-16N significantly improved shoot wt., that is, about 30% higher than the uninoculated plants  and  was  similar  to  that  of nitrogen supplemented cultivar EPAGRI-109.

Root dry weight of the cultivar IR42 increased more with inoculation of the isolate EP4-1L and commercial inoculant AZOTOTAL recording 27 and 31% improvement, respectively. However, growth of bacterized BRS TROPICAL and EPAGRI-109 were not statistically different from the uninoculated plants, rather cultivars were negatively affected for inoculation of some bacteria (Table 4). The results support that in addition to the intrinsic characteristics of the bacteria, interactions between genotype and genotype-environment would directly interfere in the growth promotion efficiency of host plants also (Oliveira, 1994; Reis et al., 2000; Kennedy, 2004; Hungria et al., 2016).

However, the results conform to several studies which have shown that in axenic conditions diazotrophic bacteria was promising for field conditions. The in vitro evaluation allows selection of a large number of strains with potential plant growth promoting ​​functions (Baldani et al., 2000; Araújo et al., 2013).

Evaluation under axenic conditions enabled Araújo et al. (2013) to select plant growth promoting isolates and host cultivars for testing in greenhouse, and found dstinct increases in biomass, that is, by 48 and 21% of Cana Forte and Cana Roxa varieties due to inoculation of N. amazonense strains AR3122, respectively. In addition, some varieties respond more to inoculation than others. The bacteria that promote yield with lower nitrogen fertilizer are greatly relevant for sustainable agriculture.

Application of  inoculant Azototal containing A. brasilense strains Abv5 and Abv6  has been approved by the Ministry of Agriculture Livestock and Food Supply (MAPA) in Brazil for maize, wheat and rice to reduce nitrogen fertilizers by up to 50% which implies in savings of  R$ 1.5 billion per year to farmers.

Thirty-nine isolates of diazotrophic bacteria were isolated from rice cultivars EPAGRI-109 and BRS TROPICAL, about 49% of which occurred from shoots, 28% from roots and 23% in rhizospheric soil. Among them, eight isolates belonged to the genus Azospirillum spp. and two isolates were identified as Herbaspirillum spp.

Paenibacillus spp. strain EP3-16N increased in biomass of plants mainly the cultivar EPAGRI-109 and Nitrospirillum spp. strain EP4-1L in roots of cultivar IR-42. However, more studies are necessary to confirm this capacity in field conditions.

The use of molecular tools must be considered in future studies to elucidate the microorganism diversity and interaction in rice plants.

The authors have not declared any conflict of interests.

The authors thank the Coordination for the Improvement of Higher Education Personnel (CAPES) for the fellowships granted and the Brazilian Agricultural Research Corporation (Embrapa) for the financial support received.