1. Introduction
The symbiosis between N2-fixing bacteria and plants of the leguminous family has long been exploited in agriculture all over the world. Such bacteria capture atmospheric nitrogen, which is enzymatically reduced to ammonia inside root nodules, and then assimilated by plant tissues, in the form of nitrogenous compounds. In Brazil, the success of the soybean [Glycine max (L.) Merr.] as a cash crop is highly attributed to the benefits obtained from inoculating seeds with elite strains of Bradyrhizobium japonicum, B. elkanii, and B. diazoefficiens, a partnership that is capable of completely meeting the crop’s nitrogen requirement to sustain high yields [1] [2] .
Other beneficial soil microorganisms such as plant growth-promoting rhizobacteria (PGPR) have also been exploited in agriculture. PGPR perform an array of biological processes which are beneficial to plants, such as production of plant-growth hormones, like auxins [3] , gibberellins [4] , cytokines [3] [5] , and ethylene [5] , induction of plant resistance to stresses and diseases [6] , phosphate solubilization [7] , and also N2 fixation [8] . Among such bacteria, those belonging to the genus Azospirillum are outstanding PGPR, having long been employed worldwide as plant inoculants, and, more recently, in Brazil too [9] [10] .
The annual fixation of over 300 kg of N/ha, followed by the delivery of 20 to 30 kg of N/ha to the soil by the association between the soybeans and N2-fixing bacteria are fundamental for the success of the crop in Brazil [1] [2] . However, the continuous development of soybean cultivars, aiming at higher yields, demands an increased supply of N for the crop, which, in turn, implies that investigation must be continued to assure the benefits of N2 fixation to supply the crop. In addition, longer periods of drought and high temperatures, due to the global weather changes, which also affect Brazil, are a threat to the contribution of the biological processes to agriculture. It is known that plants subject to environmental stresses become more susceptible, for example, to pests and diseases, resulting in increased seed treatment with chemicals such as fungicides and insecticides before sowing. This practice, which is a reality for over 90% of the soybean seeds grown in Brazil [1] [11] , may be very deleterious to seed-applied bacteria.
Given the importance of the soybean crop to Brazil, and the awareness of current and potential future limitations to biological N2 fixation by Bradyrhizobium-inoculated soybeans, as well as the alleged benefits of inoculating different crops with Azospirillum, co-inoculation of soybeans with both bacteria sounds like an interesting and promising technology to improve the crop’s performance. In addition, such approach fulfills current paradigms of agricultural, economic, social, and environmental sustainability. One recent study has addressed this issue for the soybean and common bean (Phaseolus vulgaris L.) crops [12] , but in that case, Bradyrhizobium and Rhizobium were applied to the seeds, whereas Azospirillum was inoculated in the planting furrow. Although positive results were obtained, such practice may be of little interest to farmers because it increases planting labor. Therefore, we expanded our studies and here we report the results obtained when both Bradyrhizobium spp. and Azospirillum brasilense were applied to the soybean seeds at sowing.
2. Materials and Methods
Experiments were planted in the field in the 2010/2011 and 2012/2013 cropping seasons. In 2010/2011 trials were planted at two locations in the State of Paraná, Brazil, namely Londrina (23˚11'S, 51˚11'W, 620 m altitude, Köpen-Geiger climate type Cfa), and Ponta Grossa (25˚13'S, 50˚1'W, 880 m altitude, Köpen-Geiger climate type Cfb), both with previously established and naturalized populations of Bradyrhizobium in the soil. In 2012/2013 trials were set at two locations in the State of Goiás, Brazil, namely Rio Verde (17˚47'S; 50˚54'W, 730 m altitude, Köppen-Geiger climate type Aw) and CachoeiraDourada (18˚29'S; 49˚28'W, 450 m altitude, Köppen- Geiger climate type Aw), in soils devoid of naturalized bradyrhizobia.
At each location, 40 days before sowing, soil samples (0 - 20 cm) were collected for analysis of chemical, physical, and microbiological properties, as described before [2] . Soil rhizobial populations at each location (Table 1) were estimated by the most probable number (MPN, with soybean cultivar BMX Potência RR) technique, also as described before [2] . For the estimation of the populations of free-living diazotrophic bacteria, bacteria were count in NFb semi-solid medium [13] .
In the 2010/11 crop season, in both Londrina and Ponta Grossa the soybean cultivar used was BRS 133 (conventional), while in 2012/13 BMX-Potência (RR) was used in Rio Verde and BRS-GO-8360 (conventional) in Cachoeira Dourada.
Inoculation treatments comprised either inoculation with Bradyrhizobium spp. alone (I) or co-inoculation with Bradyrhizobium spp. and Azospirillum brasilense (I + Azo). All inoculants employed were liquid commercial
![]()
Table 1. Soil chemical properties, texture and population of soybean bradyrhizobia at the 0 - 20 cm layer of the field sites where the experiments were performed. Analyses performed before sowing.
aSB, sum of bases; BS, bases saturation = [(K + Ca + Mg)/Tcec] × 100, where Tcec = K + Ca + Mg + total acidity at pH 7.0 (H + Al).
products available in Brazil. Inoculant concentrations were determined by spread-plating on yeast-manitol-agar medium, in the case of Bradyrhizobium, whereas Azospirillum was counted by spread-plating on NFb [13] and RC [14] solid media. The concentration of Bradyrhizobium in the inoculant was estimated at 2.60 × 1010 CFU/mL in 2010/11, and at 7.33 × 109 CFU/mL in 2012/13. The concentration of Azospirillum in the inoculant was 2.47 × 108 CFU/mL in 2010/2011 and 2.97 × 108 CFU/mL in 2012/2013. Both products were in accordance with Brazilian regulations [15] , which require >109 cells/g or mL, in the case of Bradyrhizobium, and >108 cells/g or mL, in the case of Azospirillum. Bradyrhizobium inoculant contained strains SEMIA 5079 (B. japonicum) and SEMIA 5080 (B. diazoefficiens), and was applied to seeds before planting at an amount necessary to provide 1.2 × 106 cells of Bradyrhizobium/seed. The A. brasilense inoculant contained strains Ab-V5 and Ab-V6, and was applied to the seeds to provide 1.2 × 105 cells/seed. Seed inoculation was performed by mixing both inoculants with the seeds and leaving it to dry in the shadow for 1 h.
Two control treatments were included in each experiment. The first was a non-inoculated (NI) control, and the second was a NI control that received N fertilizer (NI + N). N fertilizer was applied as urea (46.6% N), at 200 kg N/ha, split in two applications of 50% each, one at sowing and the other at the R2 growth stage (fully open flower below one of the uppermost nodes of the main stem with a fully developed leaf).
Experiments were set in a completely randomized block design, with six replicates, and plots measured approximately 24 m2. All plots were separated by 0.5 m-long rows, and 1.5 m-wide terraces to avoid cross contamination due to surface run-off, which might carry bacteria and or fertilizers. At all locations, base fertility was corrected by adding 300 kg/ha of a 0-28-20 NPK fertilizer. No N fertilizer was applied, except for the NI + N treatment plots. During the V4 vegetative growth phase (four nodes of the main stem above the unifoliate node with fully developed leaves) plants received foliar sprays of 20 g Mo/ha and 2 g Co/ha. Weeds were controlled with herbicides and pest control was accomplished by the utilization of both chemical and biological insecticides. Final plant population was of about 360,000 plants/ha.
Thirty-five to 50 days after sowing (DAS), five plants were collected from each plot for evaluation of nodulation (number and biomass), plant biomass, N content, and total N accumulated in the shoots, as described before [2] . Grain yield was determined at the end of the crop’s cycle, from a central portion of each plot of (5.6 to 8 m2). Seeds were cleaned and weighed, and grain yield was estimated after correction of seed weights to 13% moisture.
All data were submitted to tests that assess variable normality and variance homogeneity, followed by an analysis of variance (ANOVA). Duncan’s test was employed to compare means in cases where statistical significance had been detected by the ANOVA F test. For all analyses, the Statistica version 7.0 software was employed.
3. Results
In Londrina, in the first season, neither inoculation nor co-inoculation resulted in significant increases in nodulation, whereas in Ponta Grossa, even though the soil had 2.14 × 104 cells of soybean bradyrhizobia/g soil, both treatments significantly improved nodule number and dry weight relative to the non-inoculated control (Table 2). It was interesting to notice that although co-inoculation did not increase nodulation relative to inoculation
![]()
Table 2. Nodule number (NN), nodule dry weight (NDW), shoot dry weight (SDW), total shoot N content (TSNC) at 35 days after sowing (DAS) in Londrina and at 40 DAS in Ponta Grossa, and grain yield (Yield) of soybean plants (BRS 133, conventional), in response to seed inoculation consisting of single inoculation with Bradyrhizobium, or of co-inoculation with Bradyrhizobium and Azospirillum. Experiments performed in 2010/11 in soils containing ˃104 cells of soybean bradyrhizobia/g soil.
aMeans (n = 6) followed by different letters on the same column are significantly different from one another (p £ 0.05, Duncan test).
with bradyrhizobia alone, no incompatibility between inoculants applied to the seeds was observed. Shoot dry weight was significantly increased by N fertilization and inoculations, relative to the non-inoculated control, in both sites, but shoot total N content was not statistically different in the co-inoculation treatment in Londrina, and in the single inoculation with Bradyrhizobium in Ponta Grossa in comparison to the non-inoculated control (Table 2).
Seed inoculation with Bradyrhizobium promoted significant increases in grain yield when compared to the non-inoculated controls without N-fertilizer in both Londrina (+4.5%) and Ponta Grossa (+10.7%), even though the soils had population of soybean bradyrhizobia (Table 2). However, yield of single inoculated plants was statically lower than the plants receiving 200 kg N/ha in both sites. However, seed co-inoculation with Azospirillum resulted in additional significant increases in grain yield in comparison to single inoculation with Bradyrhizobium at both sites, with yields similar to that of the N-fertilizer treatment. Compared to the non-inoculated control, overall grain yield increase due to co-inoculation was of 11.8% in Londrina and of 17.1% in Ponta Grossa (Table 2).
In the second year of tests, nodulation and plant growth at both locations were seriously affected by unfavorable dry weather conditions that occurred right after sowing and between sowing and the first sampling. Thereafter, weather conditions improved and rainfall became more regular, allowing plants to recover growth, reflecting in reasonable yields, although not attaining the potential expected for the region (Table 3). Nodule number was very low in both Rio Verde and Cachoeira Verde, attributed to the dry conditions, especially critical if we consider that the areas were cropped with soybean for the first time, showing less than 10 cells of soybean bradyrhizobia/g soil. When Bradyrhizobium and Bradyrhizobium + Azospirillum were applied to the seeds, significantly higher nodule biomass relative to both non-inoculated controls, either with or without N-fertilizer, could be observed in Rio Verde, and in relation to the controle with N-fertilizer in Cachoeira Dourada. However, no significant differences were observed for dry weight or total N accumulated in shoots in response to either inoculation or N fertilization (Table 3).
At the final harvest, in Rio Verde, but not in Cachoeira Dourada there was a slight response when N fertilizer was applied, in comparison with both the non-inoculated and the inoculated with Bradyrhizobium treatments (Table 3). At both locations, there were no significant responses to inoculation with Bradyrhizobium, probably due to water stress at critical phases of root infection, nodule initiation and early stages of N2 fixation. Co-in- oculation with Azospirillum, on the other hand, resulted in significant yield increases at both locations. When compared to inoculation with Bradyrhizobium alone, yield increases due to co-inoculation of 212 kg/ha and 565 kg/ha were observed in Rio Verde and Cachoeira Dourada, respectively. At both locations, co-inoculation treatments yielded significantly more than the non-inoculated control without N-fertilizer, and in CachoeiraDourada it was even superior to the N fertilized treatment with 200 kg N/ha. In comparison to the non-inoculated control without N-fertilizer, the increases due to the co-inoculation were of 7.1% in Rio Verde and of 15.4% in CachoeiraDourada (Table 3).
![]()
Table 3. Nodule number (NN), nodule dry weight (NDW), shoot dry weight (SDW), total shoot N content (TSNC) at 40 days after sowing (DAS) in Rio Verde and at 50 DAS in CachoeiraDourada, and grain yield (Yield) of soybean plants (cultivars BMX-Potência (RR) and BRS-GO-8360 (conventional), respectively) in response to seed inoculation consisting of single inoculation with Bradyrhizobium, or of co-inoculation with Bradyrhizobium and Azospirillum. Experiments performed in 2012/13 in soybean bradyrhizobia population lower than 10 cells of soybean bradyrhizobia/g soil.
aMeans (n = 6) followed by different letters on the same column are significantly different from one another (p £ 0.05, Duncan test).
4. Discussion
Several authors have addressed the issue of co-inoculating legumes with N2-fixing and plant growth-promoting rhizobacteria (PGPR). There have been tests with a variety of species, e.g. alfalfa (Medicago sativa L.) [16] and common bean [12] . Notwithstanding, most studies have only focused the effects of the PGPR on the early aspects of the symbiotic relationship, and very few field studies. In a recent publication [12] , we report the benefits of co-inoculating soybean and common bean with their respective rhizobia and Azospirillum brasilense, but in that case, rhizobia were applied to the seeds and Azospirillum was placed in the planting furrow.
We initially thought of positioning each inoculum at a different place because we believed it might be necessary to apply very large doses of Azospirillum that, in turn, would reduce the concentration of Bradyrhizobium on the seeds. However, our preliminary studies showed that larger doses of Azospirillum may inhibit plant growth, and that the best benefits are obtained when Azospirilllum inoculum size is ten times smaller than the optimal inoculum size for Bradyrhizobium [12] . Therefore, taking into account the concentration of the commercial products containing Azospirillum available in the Brazilian market, we realized it would be possible to apply both bacteria to the same seeds. In the experiments reported here, we aimed at evaluating nodulation, plant growth, and yield increases in response to soybean seed inoculation with both Bradyrhizobium and Azospirillum.
It is not clear, in the literature, whether the benefits of inoculating legumes with rhizobia and azospirilla are only due to an increase in nodulation and N2 fixation, or if other indirect factors are involved. A. brasilense is one of the so-called PGPR, which may stimulate plant growth by means of an array of favorable mechanisms. In fact, the inoculation of common bean, in Egypt, with a mixture containing vesicular-arbuscular mycorrhizal (VAM) fungi, Rhizobium sp., Azospirillum sp., and Bacillus circulans, resulted in gains in plant height, branching, nodulation and plant biomass, when compared to control plants [17] . Stajković et al. [18] obtained similar results when they inoculated common bean with a mixture containing Rhizobium sp., Pseudomonas sp., and Bacillus sp. Bx, which promoted gains in plant biomass and in N and P contents in the shoots. Those authors also observed phosphate solubilization, indole-acetic acid (IAA), ammonia, and siderophore production by Pseudomonas sp. LG in vitro, what might explain common bean growth promotion. Grain yield and yield components, such as the number of pods per plant, the number of seeds per pod, the weight of 100 seeds, as well as protein content in the grains were also positively influenced by the inoculation of common bean with Rhizobium and Pseudomonas fluorescens [19] . Non-leguminous plants may also benefit from inoculation with Azospirillum and other N2-fixing PGPR, e.g. banana (Musa spp.) trees present better root growth, earlier flowering, increased fruit yield, and improved fruit quality when inoculated with PGPR [20] .
These reports evidence that Azospirillum contributes to promote plant growth and yield by means of an array of mechanisms, which, in turn, may explain our field results. Azospirillum is known, for instance, for improving the plant’s resistance to water stress [21] . Such favorable activity might be of utmost importance to plants depending on fixed N2, since it is well documented that water and temperature stresses are among the most limiting factors for biological N2 fixation in the tropics, seriously compromising root infection, nodule formation, and N2 fixation.
In the case of two (Rio Verde and Cachoeira Dourada) of the four experiments reported here, soybean co-in- oculation with Azospirillum seemed to be beneficial under conditions of water stress. Our field trials revealed that the utilization of both bacteria resulted in significant average yield increase of 388 kg/ha (equivalent to 6.5 60 kg bags), or 15.3% relative to when only Bradyrhizobium was employed as inoculant on the seeds. Similar results were obtained when co-inoculation was compared to the non-inoculated, non-fertilized control in both experiments, and even to the N-fertilized (200 kg N/ha) control in Cachoeira Dourada (Table 3 ).
Our results demonstrate that seed co-inoculation with Bradyrhizobium and Azospirillum is agronomically efficient and is beneficial to the crop. We obtained gains in grain yields both in areas where soybean was cropped for the first time, where no soil bradyrhizobia were available, as well as in areas where the crop had been grown before, where there was already an established population of bradyrhizobia in the soil. The strategy of co-in- oculation represents a new biotechnological tool to improve soybean yield without adding any chemical N fertilizers, thus contributing to current practices of sustainability in agriculture.
5. Conclusion
Soybean seeds can be inoculated with both Azospirillum and Bradyrhizobium inoculants, and co-inoculation improves soybean yield without the addition of any chemical N fertilizers, helping reduce costs and keeping pace with current practices of sustainability in agriculture.
Acknowledgements
The microbiology group of Embrapa Soja is also supported by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil), CNPq-Repensa (562008/2010-1). M. Hungria and M.A. Nogueira are also research fellows from CNPq. Approved for publication by the Editorial Board of Embrapa Soja as manuscript number 285/2014.
NOTES
*Corresponding author.