Diversity and Efficiency of Rhizobia Communities from Iron Mining Areas Using Cowpea as a Trap Plant

Mining is an important economic activity. However, its impact on environment must be accessed, mainly on relevant processes for their sustainability. The objective of this study was to evaluate the diversity and efficiency of symbiotic nitrogen fixing bacterial communities in soils under different types of vegetation in the Quadrilátero Ferrífero: ironstone outcrops, Atlantic Forest, neotropical savanna, and a rehabilitated area revegetated with grass. Suspensions of soil samples collected under each type of vegetation were made in a saline solution to capture rhizobia communities that were then inoculated on cowpea [Vigna unguiculata (L.) Walp.], which was used as a trap plant. The symbiotic efficiency of the communities was evaluated in a greenhouse experiment and the data obtained were correlated to the chemical and physical properties of the soils under each type of vegetation. At the end of the experiment, the bacteria present in the nodules were isolated to evaluate their diversity. The highest numbers of nodules occurred in the treatment inoculated with soil samples from rehabilitated area revegetated with grass and neotropical savanna vegetation, and the lowest numbers were observed in the treatment inoculated with soil samples from ironstone outcrops and Atlantic Forest. In relation to root dry matter, the treatment inoculated with soil samples from Neotropical savanah was superior to those inoculated with soil samples from the other areas; already, in relation to the shoot dry matter, no significant difference among the treatments was observed. The soil properties with the greatest influence on the microbial communities were Al content, considered as high in the Atlantic Forest and neotropical savanna vegetation, as intermediate in the iron outcrops, and as very low in the rehabilitated area revegetated with grass; organic matter, considered as very high in the ironstone outcrops and neotropical savanna, as high in the Atlantic Forest, and as low in the rehabilitated area revegetated with grass; and the pH, with intermediate acidity level in the rehabilitated area revegetated with grass, high level of acidity in the iron outcrops and neotropical savanna, and very high acidity in the Atlantic Forest. After isolation of the nodules, 380 bacterial strains were obtained and separated into 27 groups by cultural characterization analysis. Genetic diversity was evaluated by the 16S rRNA gene partial sequencing of 156 strains, which identified some bacteria belonging to nitrogen-fixing Leguminosae nodulating bacterial genera (Rhizobium, Bradyrhizobium, Burkholderia, and Cupriavidus), some representative of associative bacteria (Bacillus, Paenibacillus, Herbaspirillum, Pseudomonas, and Agrobacterium), and other genera (Brevibacillus, Novosphingobium, Chitinophaga, Dyella, Acinetobacter, and Stenotrophomonas). The highest genetic diversity of bacteria was found in the rehabilitated area revegetated with grass indicated that it was effective in soil rehabilitation


INTRODUCTION
The Quadrilátero Ferrífero is in the central area of the state of Minas Gerais, Brazil, and stands out in the national and world scene for its importance as an iron ore producing region.The landscape of this environment is currently composed of fragments of Brazilian environmental hotspots, the Brazilian neotropical savanna (Cerrado) and the Atlantic Forest, which have been intensely transformed by human activities, with emphasis on urbanization and mining, generating impacts such as soil removal and consequent loss of vegetation cover (Jacobi and Carmo, 2008).Soils in this region are mainly ferruginous and generally of low fertility, acidic, and shallow; this has a considerable effect on the vegetation cover, which is composed of plants adapted to these peculiar conditions in the areas of ironstone outcrops (Costa, 2007;Carvalho Filho et al., 2010).
Knowing and evaluating the functions of microorganisms native to these environments is important for selecting strains with biotechnological potential for in situ bioremediation and which can be used as inoculants of species used in revegetation of degraded areas.These microorganisms have an important role in nutrient cycling and in improving nutrient availability, which favors the establishment of plants.They also show potential as indicators of environment changes, such as in pH, Al 3+ contents, and organic matter, which are the factors that most influence the occurrence of microorganisms in the soil (Powlson et al., 1987;Siqueira et al., 1994;Lauber et al., 2008;Jesus et al., 2009).
Most bacteria found in the soil depend on interactions with plants; this interaction favors plant growth by enabling inorganic phosphate solubilization, biological N 2 fixation, production of plant hormones, and production of antifungal compounds, among others factors (Lim et al., 1991;Vessey, 2003;Hara and Oliveira, 2005;Marra et al., 2012;Costa et al., 2013;Rufini et al., 2014;Panizzon et al., 2016).In the case of N 2 -fixing Leguminosae-nodulating bacteria (NFLNB), isolated strains represent essential genetic resources for the selection of strains with biotechnological potential, including revegetation of degraded areas.
Cowpea [Vigna unguiculata (L.) Walp] has been used as a trap plant in studies that evaluate the diversity of NFLNB due to its ability to establish symbiosis with different genera of bacteria, such as Bradyrhizobium, Rhizobium, and Mesorhizobium (Melloni et al., 2006;Guimarães et al., 2012;Costa et al., 2013;Jaramillo et al., 2013;).Identification and selection of NFLNB that establish symbiosis with cowpea and other legumes is important to reduce the use of N fertilizers, which promotes the economic and environmental sustainability of agriculture (Lacerda et al., 2004;Soares et al., 2006;Sousa and Moreira, 2011).
The hypothesis of this work is that diversity and efficiency of rhizobia communities differ significantly in soils under different types of vegetation under influence of mining activities.Thus, the aim of this study was to evaluate the symbiotic, genetic, and phenotypic diversity of cowpea-nodulating rhizobia communities from soils associated with ironstone outcrops, neotropical savanna, Atlantic Forest vegetation, and a rehabilitated area revegetated with grass in the Quadrilátero Ferrífero of Minas Gerais, Brazil, and to verify the influence of soil physical and chemical properties on the microbiota.

Study areas
The collection area was in the municipalities of Nova Lima, in the Technology Center of Ferroso -Miguelão, and in Brumadinho, in the Córrego do Feijão Mine, which belong to Vale S/A.The vegetation of the study site was identified as follows: neotropical savanna, ironstone outcrops, Atlantic Forest, and a rehabilitated area revegetated with grass (Figure 1).According to the Forest Inventory of Minas Gerais of 2009, the vegetation present in this area is called Campo Rupestre (in this study described as ironstone outcrops) and Atlantic Forest.The inventory presents no information on the neotropical savanna area.
The rehabilitated area revegetated with grass had had Atlantic Forest vegetation, which was removed to establish an ore deposit near the railroad loading area.When the mining area was closed, a recovery project was carried out by planting tree species; however, they did not survive to successive fires, and the planted grass (Panicum maximum Jacq) spread throughout the area.To date, the predominant species are Brachiaria decumbens (Brachiaria), Melinis minutiflora (molasses grass), and Panicum maximum Jacq.(Guinea grass).

Sampling and physical-chemical characterization of soil
Soil samples were collected from August 9 to 15, 2015.For soil sampling, two transects at approximately 50 m distance were drawn in each type of vegetation.On each transect, five points were georeferenced, at approximately 50 m distance, resulting in 10 points per type of vegetation.Five subsamples were collected from each point, five meters apart from each other, at the 0.00-0.20 m depth, resulting in a composite sample from each of the 40 georeferenced points.Samples were deposited in sterile plastic bags and polystyrene boxes and taken to the laboratory, where they were stored at 4 °C in a cold chamber until use

Capture and efficiency of bacterial communities using cowpea as a trap plant
The experiment was carried out from October to December 2015 under greenhouse conditions and consisted of eight treatments: inoculations with soil suspensions from each type of vegetation, two positive controls [inoculation with strains approved by MAPA as inoculants for cowpea: UFLA 03-84 (Bradyrhizobium sp.) and INPA 03-11B (Bradyrhizobium elkanii)], and two negative controls without inoculation [with high (HN) and low (LN) mineral N concentration].The experimental design was completely randomized, with three replications.
Seeds were disinfested and planted in longneck bottles containing Hoagland and Arnon (1950) nutrient solution, following the method described by Florentino et al. (2009).Seeds were pre-germinated on moist sterile filter paper that was folded, wrapped in aluminum foil, and stored in a growth chamber at 28 °C until radicle emission.Soil samples from each point were resuspended in 0.85 % NaCl solution (Moreira et al., 2010) at a 1:1 ratio, and 1 mL of the suspension was inoculated on each seedling.A nutrient solution with 5.25 mg L -1 of N was used in the inoculated treatments and in the control without inoculation and with low mineral N concentration.A nutrient solution with 52.5 mg L -1 of N was used in the control without inoculation and with high mineral N concentration.
At 30 days after planting, plants were harvested and the following traits were evaluated: number of nodules (NN), nodule dry matter (NDM), shoot dry matter (SDM), root dry matter (RDM), and relative efficiency (RE).For determination of NN, nodules were collected from the roots and counted.Three nodules per plant were selected for isolation.The other nodules were placed in glass jars; shoot and roots were placed in paper bags and kept in a forced air circulation oven at 60 °C until constant weight for determination of NDM, RDM, and SDM.The relative efficiency of each treatment was calculated by the following formula: where RE is the relative efficiency, and SDM is the shoot dry matter.
Data on number of nodules, nodule dry matter, shoot dry matter, root dry matter, and relative efficiency were subjected to analysis of variance (Anova) using the statistical analysis software SISVAR 5.6 (Ferreira, 2011).The NN and NDM data were transformed to square root of (x + 1).The effects of treatments were compared by the Scott-Knott test at 5 % significance (Scott and Knott, 1974).

Relationship between physicochemical properties and biological variables
Chemical and physical properties of soils and biological variables were correlated by Principal Component Analysis (PCA) using the R software (R Development Core Team, 2011).

Isolation and cultural characterization of bacterial strains
Three nodules per plant were collected for isolation of bacterial strains.Nodules were surface disinfected in ethyl alcohol (92.8 %) for 30 s and 3 % H 2 O 2 for 3 min, and washed six times with sterilized distilled water.The reagents and the water used in the nodule disinfection process were changed for each treatment, and the last wash water was plated to evaluate the effectiveness of the disinfection process.Nodules were then macerated and scattered in the form of streaks on plates with culture medium 79 (Fred and Waksman, 1928) with bromothymol blue, in order to obtain isolated bacterial colonies.The plates were stored in an incubator at a constant temperature of 28 °C, and colonies in the culture media were evaluated for 10 days or more.

16S rRNA gene partial sequencing
For genetic identification of the isolates, 16S rRNA gene partial sequencing was performed.Genomic DNA was extracted by the Alkaline Lysis Method (Niemann et al., 1997).Amplification of the 16S rRNA gene followed the procedures described by Guimarães et al. (2012) through use of the primers 27F (GAGTTTGACCTGGCTCAG) and 1492R (GGTTACCTTGTTACGACTT) (Lane, 1991).The Polymerase Chain Reaction (PCR) products were sent to the Macrogen laboratory in South Korea for purification and sequencing.To evaluate the quality of the sequences obtained, the software BioNumerics 7.1 was used (AppliedMaths, Austin, TX, USA).The sequences were subsequently subjected to the BLASTn (Bethesda, MD, USA) for comparison with similar sequences already deposited in the GenBank, National Center for Biotechnology Information (NCBI).The genetic diversity in different vegetation types was also evaluated by the Shannon index, which considers both the richness and abundance of species.For calculation purposes, the different species found in each type of vegetation were considered.

Chemical and physical analysis of soils
The interpretation of soil chemical analysis was performed based on the recommendations of the Soil Fertility Commission of the State of Minas Gerais (Alvarez et al., 1999) (Table 1).Potassium content in the Atlantic Forest (75.60 mg dm -3 ), neotropical savanna (72.60 mg dm -3 ), and rehabilitated area revegetated with grass (88.20 mg dm -3 ) was classified as intermediate.In ironstone outcrops, low K + content (56.80 mg dm -3 ) was observed, classified as intermediate.Phosphorus content was very low and statistically similar among the different types of vegetation [low in ironstone outcrops (1.59 mg dm -3 ), rehabilitated area revegetated with grass (1.66 mg dm -3 ), neotropical savanna (1.36 mg dm -3 ), and Atlantic Forest (2.15 mg dm -3 )].Sulfur content in all types of vegetation was classified as very good, and the rehabilitated area revegetated with grass (45.14 mg dm -3 ) and neotropical savanna (36.29 mg dm -3 ) exhibited the highest concentrations.Manganese content was considered as high in all types of vegetation, reaching the highest values in the neotropical savanna (112.29 mg dm -3 ), rehabilitated area revegetated with grass (104.00 mg dm -3 ), and ironstone outcrops (88.88 mg dm -3 ).
Zinc contents were classified as high in the ironstone outcrops (3.29 mg dm -3 ) and neotropical savanna (3.13 mg dm -3 ), and as good in the rehabilitated area revegetated with grass (1.60 mg dm -3 ) and in the Atlantic Forest (1.92 mg dm -3 ).Boron content was classified as low in the ironstone outcrops (0.26 mg dm -3 ), neotropical savanna (0.20 mg dm -3 ), and Atlantic Forest (0.20 mg dm -3 ), and as very low in the rehabilitated area revegetated with grass (0.15 mg dm -3 ).Cupper content was considered as high in the rehabilitated area revegetated with grass (2.14 mg dm -3 ), as low in the ironstone outcrops, and as good in the neotropical savanna and Atlantic Forest.Magnesium content was considered as low in all types of vegetation, and no difference was observed (p<0.05) between the vegetation areas.Calcium content was considered as low in the rehabilitated area revegetated with grass (0.75 cmol c dm -3 ), neotropical savanna (0.91 cmol c dm -3 ), and Atlantic Forest (0.99 cmol c dm -3 ), and as intermediate in the ironstone outcrops (1.28 cmol c dm -3 ).The soils collected have loam texture in the ironstone outcrops and rehabilitated area revegetated with grass, and clayey texture in the neotropical savanna and Atlantic Forest.
Soil under the rehabilitated area revegetated with grass had a pH(H 2 O) 5.60, indicating an intermediate level of acidity.Soils of the ironstone outcrops and neotropical savanna had mean pH 4.72 and 4.97, respectively, indicating high acidity.In the Atlantic Forest environment, the mean pH value indicates high acidity soil (4.21).Exchangeable acidity was considered as high in the neotropical savanna (1.56 cmol c dm -3 ) and in the Atlantic Forest (1.90 cmol c dm -3 ).In the ironstone outcrop soil, the exchangeable acidity value was classified as intermediate (0.85 cmol c dm -3 ).
Rev Bras Cienc Solo 2017;41:e0160525 Soil under the rehabilitated area revegetated with grass had a low mean value for exchangeable acidity (0.09 cmol c dm -3 ).The mean values of potential acidity found in the ironstone outcrops, neotropical savanna, and Atlantic Forest vegetation were classified as very high (12.64, 15.46, and 12.26 cmol c dm -3 , respectively).This promotes high potential CEC, which indicates that the soil has the ability to retain more nutrients, despite the low pH.In the rehabilitated area revegetated with grass, the potential acidity value was 1.94 cmol c dm -3 , which was considered as low.Soils under the ironstone outcrops and neotropical savanna vegetation had very high organic matter, with values of 7.58 and 8.30 dag kg -1 , respectively.The soil of the Atlantic Forest vegetation had a value of 4.94 dag kg -1 , which is classified as high.The rehabilitated area revegetated with grass had the lowest value, 1.38 dag kg -1 .The ironstone outcrops and neotropical savanna had the highest values for organic matter (7.58 and 8.30 dag kg -1 , respectively), followed by the Atlantic Forest (4.94 dag kg -1 ) and rehabilitated area revegetated with grass (1.38 dag kg -1 ).
Rev Bras Cienc Solo 2017;41:e0160525 The soils of the Quadrilátero Ferrífero are mostly derived from itabirite (a metamorphic BIF), which explains their high iron concentration.The critical level for iron is 45 mg dm -³ (Alvarez et al., 1999).In this study, chemical analysis showed that the lowest value was 124.7 mg dm -³ .This result corroborates the soil characterization made by Carvalho Filho et al. (2010) in this region, which describes the soils of the inner face of the Serra da Moeda as shallow and generally very stony that are derived from itabirite, with great iron concentration.
Capture and symbiotic efficiency of bacterial communities using cowpea as a trap plant In the controls without inoculation with low and high N concentration, the nodulation was negative and reference strains nodulated normally, indicating that there was no contamination and that the experimental conditions were favorable to symbiosis.
In relation to number of nodules, the nodulation ability of the bacterial communities obtained from inoculations prepared from soil suspensions from rehabilitated area revegetated with grass and from neotropical savanna vegetation areas was higher than that from Atlantic Forest and from ironstone outcrops (Table 2).The rehabilitated area revegetated with grass had the highest value for nodule dry matter, followed by neotropical savanna.The area of neotropical savanna stood out for root dry matter, surpassing the other areas.The values of shoot dry matter and relative efficiency did not differ statistically between the areas.Nodules were observed in all treatments; the rehabilitated area revegetated with grass had the highest value, followed by neotropical savanna, Atlantic Forest, and ironstone outcrops.

Relationship between soil physicochemical properties and biological variables
Results of the principal component analysis between soil physicochemical properties and biological variables explained 48 % of the total variance (PC1: 33 % and PC2: 15 %).These results, together with the correlation matrix, allowed better understanding of the relationship between the physicochemical properties of the soils collected in the different environments and the bacterial communities (Figure 2).The physicochemical properties of the soils, together with shoot dry matter, nodule dry matter, root dry matter, number of nodules, and relative efficiency, were correlated.Analysis of PC1 shows that number of nodules and nodule dry matter are directly correlated with pH, base saturation, Mn, and Cu and inversely correlated with exchangeable acidity, potential acidity, effective CEC, CEC at pH 7, Al saturation, and organic matter (Table 3).
By analyzing the spatial distribution of the points in the PCA, it can be inferred that the soil under the rehabilitated area revegetated with grass is moving closer to neotropical savanna conditions, which indicates possible recovery.The soil under ironstone outcrops is Means followed by the same letter in the columns do not statistically differ by the Scott-Knott test at 5 % probability.
Rev Bras Cienc Solo 2017;41:e0160525 isolated, possibly due to physicochemical conditions, and is very different from the other ecosystems analyzed in the study, especially in relation to Fe contents.Iron contents were high in all environments, but were even higher in this area.The points of Atlantic Forest and neotropical savanna overlap, and the same characteristics may influence the soil conditions.

Isolation and cultural characterization of plant growth promoting bacteria
A total of 380 bacterial strains grown in solid culture medium were obtained from the experimental isolation of cowpea nodules.The rehabilitated area revegetated with grass and the neotropical savanna exhibited 161 and 125 isolates from the 10 collection sites, respectively.The ironstone outcrops exhibited 29 isolates, and only sites 5, 6, and 7 exhibited nodules.The Atlantic Forest exhibited 65 isolates, and only sites 1, 2, 5, and 9 did not exhibit nodulation.Cultural groups which had the greatest number of representatives had a slow growth rate; they alkalinized the medium and produced little mucus (SALL, 23 %); they had intermediate growing time and maintained the characteristics of the medium neutral and with little production of mucus (INL, 13 %); and fast growing, which alkalinize the medium and had little mucus production (FALL, 12 %) (Figure 3), with the highest cultural diversity observed in the neotropical savanna and Atlantic Forest (18 groups), followed by the rehabilitated area revegetated with grass (15 groups) and ironstone outcrops (9 groups).Of the 380 isolates, 35 % showed slow growth time, 3 % very slow, 29 % intermediate, and 33 % fast.Regarding pH changes in the culture medium, 53 % alkalized, 12 % acidified, and 35 % retained the neutral medium.Mucus production by the isolates was also evaluated, and it was abundant in 4 %, moderate in 15 %, little in 80 %, and scarce in 1 %.

16S rRNA gene partial sequencing
The 16S RNAr gene partial sequencing was performed for 156 of the 380 strains isolated from nodules in the experiment.These strains have representatives in 18 of the 27 cultural groups formed (FNL, IALL, INL, FNAB, FNM, SALL, INAB, SNL, SALM, FALL, FALM, SAL, FAL, SALAB, IAL, SAM, SNS, and FAM).The analyzed sequences ranged from 320 to 1420 base pairs, with 98 % to 100 % similarity with the sequences of strains that have already been deposited in the NCBI GenBank.The table 4 shows the species already known as N 2 fixing nodulating bacteria.In contrast, table 5 shows the species and genera that have not yet been proven to be N 2 fixing nodulating bacteria.Strains belonging to genera which constitute both bacterial types were included in table 4. The sequences determined in this study have been deposited in GenBank under accession numbers MF495721 to MF495861.The most frequent genera were Burkholderia (Moulin et al., 2001), Rhizobium (Frank, 1889), and Bradyrhizobium (Jordan, 1982), representing 60 % of the isolates present in the soils of all types of vegetation.Bradyrhizobium was not found only in ironstone outcrops.
From soil under ironstone outcrops soil, 18 strains were sequenced.Seventy two percent of these strains belong to the genus Burkholderia, comprising associative species such as B. acidipaludis, and legume symbiont species such as B. nodosa (Chen et al., 2007), which is also considered to be a free-living N 2 fixing bacterium.Only one representative of the genus Chitinophaga and one of the genus Rhizobium were identified, and species of the genus Paenebacillus.From soil under the neotropical savanna, 54 strains were sequenced, and the genus Burkholderia (55.5 %) prevailed, with representatives of B. nodosa, B. sabiae, and B. tropica.In addition to this genus, the genera Rhizobium and Bacillus were also identified.Of the 15 isolates sequenced from soil under the Atlantic Forest, most of them belonged to the genus Paenebacillus, followed by Bradyrhizobium, with three representatives; the others are distributed among the genera Rhizobium, Brevibacillus, Chitinophaga, Acinetobacter, Novosphingobium, and Burkholderia, with only one representative in each genus.Of the isolates obtained from the soil under rehabilitated area revegetated with grass, 69 strains were sequenced, most of them belonging to the genus Rhizobium (28.9 %).In this environment, the following bacterial genera were identified: Burkholderia,   (2) NPB: number of pair of bases.
(3) SI: percentage of similarity in the GenBank.
(3) SI: percentage of similarity in the GenBank.The genetic diversity of bacteria isolated from nodules was much higher in the soil under rehabilitated area revegetated with grass, followed by that under the Atlantic forest, neotropical savanna, and ironstone outcrops (Table 6).

DISCUSSION
From the soil collected, cowpea captured NFLNB species/genera among others that promote plant growth.According to Moreira and Siqueira (2006), nodulation can be influenced by a factor such as temperature, which can affect several stages in infection, formation, and function of nodules in the case of plant symbioses with NFLNB.During the experiment, the temperature in the greenhouse reached 46 °C, which may have negatively affected nodulation in some treatments.Soil chemical properties may also have affected nodulation of some treatments, especially those related to acidity and low concentration of nutrients, which may negatively affect microbiota (Moreira, 2006;Jesus et al., 2009;Lima et al., 2009).From the 16S rRNA gene partial sequencing, free-living, associative, and symbiotic N 2 -fixing bacteria that act as plant growth promoters were identified, in addition to other genera commonly isolated from nodules.
The diversity of NFLNB in soils of the Quadrilátero Ferrífero of Minas Gerais was evaluated by Costa (2016) that observed the influence of soil chemical properties on the microbiota, corroborating the results obtained in this study.Principal component analysis (PCA) showed that soil chemical properties, such as pH, Al 3+ content, and base saturation, provided more favorable conditions under the rehabilitated area revegetated with grass, possibly due to the influence of soil tillage.Among the soil properties, factors related to acidity, such as Al 3+ content and pH, are those that most directly influence microbial communities, which possibly favored the higher nodulation rate in the treatments inoculated with soil from the rehabilitated area revegetated with grass.
The genus Burkholderia can benefit plant growth in several ways, such as in siderophore production and phosphate solubilization (Vial et al., 2007;Collavino et al., 2010;Marra et al., 2011;Mathew et al., 2014).This genus represented 46 % of the sequenced isolates, occurring in all types of vegetation.The selection of strains, such as those of the genus Burkholderia, which are able to adapt to certain soil conditions, can improve yield under field conditions and reduce the use of N fertilizers and, consequently, agricultural costs (Alves et al., 2016)  The altitude may also favor the predominance of this genus, since it influences humidity and/or temperature (Bontemps et al., 2010). Moreover, Dall'Agnol et al. (2016) state that the presence of Burkholderia (Paraburkholderia) in these environments did not indicate a preference of these genera for acidic conditions, but a tolerance to these conditions, which may represent an important role of these bacteria in maintenance of the ecosystem in these environments, characterized by acid soils with high Al saturation and low N content.
The genus Bradyrhizobium was identified among the isolates obtained from the plants inoculated with bacteria from soils from the rehabilitated area revegetated with grass, Atlantic Forest, and neotropical savanna.Rhizobium was the most frequent genus under the rehabilitated area revegetated with grass; however, it was also identified in other types of vegetation.Bacteria belonging to these genera are known to be N 2 fixers, and they form symbiosis with different leguminous plants, which is of agronomic importance because it benefits plant development (Zahran, 1999;Moreira and Siqueira, 2006).Bacteria belonging to the genus Herbaspirillum usually occur in grasses (Baldani et al., 1996;Olivares et al., 1997) and were identified under the rehabilitated area revegetated with grass.In addition, the genus Brevibacillus was identified, which occurs in grasses, and it also acts as a plant growth promoter (Nakamura, 1991;Shida et al., 1996;Lima, 2009).
Nodule endophytic but non-symbiotic bacteria belonging to the genera Agrobacterium, Pseudomonas, and Terriglobus also occurred in this area (Bai et al., 2002;Mhamdi et al., 2005;Wang et al., 2006;Kan et al., 2007;Li et al., 2008;Muresu et al., 2008).The genus Bacillus was identified under neotropical savanna and the rehabilitated area revegetated with grass, and includes the plant growth promoting rhizobacteria, commonly found in the rhizosphere of plants (Araujo, 2008;Jaramillo, 2010).Some representatives of this genus, in addition to endophytes, have been reported to nodulate siratro and cowpea, together with representatives of the genus Paenibacillus, found in the rehabilitated area revegetated with grass and the Atlantic Forest, but this still needs to be proven (Halverson and Handelsman, 1991;Siddiqui and Mahmood, 1999;McSpadden Gardener, 2004;Silva et al., 2007;Li et al., 2008;Marra et al., 2012;Costa et al., 2013;Jaramillo et al., 2013).
The genus Novosphingobium (formerly Sphingomonas) was isolated under the Atlantic Forest.Members of this genus are able to degrade polycyclic aromatic hydrocarbons and are frequently isolated from petroleum-contaminated soils; they are important for in situ bioremediation (Balkwill et al., 1997;Zhou et al., 2016).The genus Chitinophaga was also found under the Atlantic Forest vegetation.Representatives of this genus have already been isolated from the soil and rhizosphere of the plants (Kämpfer et al., 2006;Kim and Jung, 2007;Lee et al., 2007Lee et al., , 2009;;Weon et al., 2009;Chung et al., 2012;Li et al., 2013).Even in the rhizosphere of leguminous plants, there may be other N 2 fixers, and the rhizosphere of grasses may have a representative number of NFLNB, which was observed in this study, especially in the rehabilitated area revegetated with grass.
The high genetic diversity of bacteria found in the rehabilitated area revegetated with grass was a good indicator of success in its rehabilitation.

CONCLUSIONS
The chemical properties of the soils that influenced biological properties were pH, sum of bases, and aluminum content.
Burkholderia, Rhizobium, and Bradyrhizobium were the most common genera.
The presence of soils with high levels of acidity may have favored high occurrence of the genus Burkholderia.
The greatest genetic diversity among vegetation types was found in the rehabilitated area revegetated with grass.

Figure 1 .
Figure 1.Map showing the collection sites in the municipalities of Brumadinho, MG and Nova Lima, MG in the neotropical savanna (NS), ironstone outcrops (IO), Atlantic Forest (AF) and rehabilitated area revegetated with grass (RA).

Figure 2 .
Figure 2. Principal component analysis (PCA) relating soil physical and chemical properties and biological variables (SDM: shoot dry matter, RDM: root dry matter, NDM: nodule dry matter, NN: number of nodules, and RE: relative efficiency) of the different types of vegetation in the Quadrilátero Ferrífero.Soil chemical properties: H+Al: potential acidity, SB: sum of bases, t: effective CEC, OM: organic matter, rem-P: remaining phosphorus, V: base saturation, m: aluminum saturation, and T: CEC pH 7.

Table 1 .
Chemical and physical properties of the soils collected in ironstone outcrops, rehabilitated area revegetated with grass, neotropical savanna, and Atlantic Forest vegetation areas at the Ferrous Technology Center -CTF Miguelão and in the Córrego do Feijão Mine, Vale S/A

Table 3 .
Principal component analysis of physical, chemical, and biological properties of soils under different vegetation types in the Quadrilátero Ferrífero environments PC: Principal component.The values in bold which are greater than or equal to 0.7 indicate strong correlation for interpretation of the behavior of the principal components.Rev Bras Cienc Solo 2017;41:e0160525 Atlantic Forest; c Rehabilitated area revegetated with grass; d Ironstone outcrops; e Rehabilitated area revegetated with grass and neotropical savanna; f Neotropical savanna and Atlantic Forest; g Rehabilitated area revegetated with grass, neotropical savanna and Atlantic Forest; h Rehabilitated area revegetated with grass, neotropical savanna, Atlantic Forest, and ironstone outcrops; and i Rehabilitated area revegetated with grass, ironstone outcrops, and Atlantic Forest.

Table 5 .
Identification of bacterial strains isolated from cowpea nodules based on 16S rRNA gene sequencing
Dall'Agnol et al. (2016)nalyzed the nodulation and biological N 2 fixation of Mimosa species in the neotropical savanna and Caatinga biomes in Brazil.The authors suggest that Burkholderia spp.prefer soils with high acidity, which was corroborated in the present study since this genus occurred more frequently under ironstone outcrops.Some researchers have suggested the name Paraburkholderia as a new name for part of the species in the genus Burkholderia.This genus occurred more frequently in a study carried out byDall'Agnol et al. (2016), and, according to the authors, this may be associated with the properties of neotropical savanna soils, such as their pH, high Al contents, and low fertility.

Table 6 .
Diversity of rhizobia and other bacteria isolated from nodules formed in cowpea after inoculation by suspensions of soil from different vegetation types of the Quadrilátero Ferrífero, MG, Brazil