Rhizobia and endophytic bacteria isolated from rainforest fragments within an iron ore mining site of the Eastern Brazilian Amazon

Abstract

The aim of the present study was to isolate and evaluate the diversity of rhizobial and endophytic bacterial strains from undisturbed native rainforests within an iron ore mining site of the Serra Norte de Carajás in the Eastern Brazilian Amazon region to assess their biotechnological utility in reclamation of areas. Experiments were conducted to capture strains from samples of the soil of these forests at the sites Arenito II, Noroeste II, and Sul IV using Macroptilium atropurpureum and Mimosa acutistipula var. ferrea as trap host plants. Only M. atropurpureum nodulated, and the different bacterial strains were isolated from its nodules. There was no difference in the number of nodules among the areas, but the Arenito II bacterial community was the most efficient, indicated by the aboveground biomass production and suitable shoot mass/root mass ratio. Fifty-two (52) bacterial isolates were obtained, distributed in five groups, including nodulating and endophytic bacteria: 32 from Arenito II, 12 from Noroeste II, and 8 from Sul IV. The nodulating Bradyrhizobium genus was common to the three areas, whereas Paraburkholderia was found only in Arenito II. The nodD1 gene was amplified in all the strains of both nodulating genera. Strains of the nodulating genus Methylobacterium were also isolated from the three areas; however, they did not nodulate the host of origin, and their nodD1 gene was not amplified. Endophytic strains were also isolated from the genera Paenibacillus, Pantoea, and Leifsonia in Arenito II, Leifsonia in Noroeste I, and Paenibacillus in Sul IV. The greater nodulation and rhizobial and endophytic bacterial diversity observed in Arenito II were probably due to the more suitable edaphic properties of the area. The isolated strains were incorporated in the collection of the Department of Soil Science of UFLA and will be investigated in relation to their symbiotic characteristics with native host plants, as well as their ability to perform other biological processes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s42770-021-00524-0.

Introduction

Diatomic nitrogen (N₂) gas is the most abundant form of N in the biosphere; however, it is not accessible to most organisms. Biological nitrogen (N₂) fixation (transformation of N₂ to NH₃, mediated by some prokaryotes) is the main process by which this element enters in the soil–plant system. N₂-fixing microorganisms (diazotrophs) are found in various phyla of Bacteria and of Archaea, and they can be free-living, associative, or symbiotic [1, 2].

Bacteria representing subclasses α- and β-proteobacteria are able to fix N₂ symbiotically with species of the Leguminosae family and are commonly known as rhizobia or nitrogen-fixing legume-nodulating bacteria (NFLNB) [3, 4]. The role of these bacteria is fundamentally related to nitrogen nutrition of plants; however, they may also bring other benefits, for example, phosphate solubilization [5, 6] and production of growth phytohormones [6] and siderophores. Siderophores not only assist in iron absorption, but can also reduce toxicity from Al [7] and heavy metals [8, 9]. NFLNB can also induce systemic resistance against pathogens [10] and tolerance to abiotic stress [11]. Nodules are the location of functional activity of the NFLNB. Nodules are hypertrophic structures specialized in symbiotic fixation of nitrogen and are also the habitat of other endophytic bacteria, N₂-fixing or otherwise, which can bring benefits to both symbionts, or be opportunists [1, 12, 13]. In general, N₂-fixing bacteria (NFB), in association with plants and/or living saprophytically in the soil, contribute to the operation of Earth ecosystems, participating directly and indirectly in providing ecosystem goods and services [14].

It is estimated that more than 60% of the Amazon forest of South America is in Brazilian territory, and most of Brazilian tropical forests are concentrated in this region. The Brazilian Amazon contains 21 distinct phytophysiognomies [15], where the tropical rainforest constitutes an important hotspot of biodiversity [16], dominated mainly by leguminous plants [16–19]. Various studies related to the nodulation ability of leguminous plants of the Brazilian Amazon have been developed [20–23], and the diversity of the rhizobial communities has been directly evaluated in nodules taken from the field [24, 25] and from soil samples through the use of trap plants [26–30]. In general, these studies had showed a high frequency of nodulating species among Mimosoideae and Papilionoideae and lower frequency among Caesalpinioideae. Nodulation increases with disturbance of the environments; thus, nodules are hardly found in undisturbed forests. Rhizobia diversity was found to be high, and Bradyrhizobium genus predominates among the symbionts with forest species. Most of these studies have been concentrated in the Western Amazon, and it is necessary to expand to other regions to assess the diversity of this important group of soil bacteria and to create and/or strengthen regional germplasm banks for the purpose of diverse future applications. Such regional germplasm banks can be used in studies related to the conservation, use, and management of native nodulating species, as well as in processes of rehabilitation of degraded environments and/or in agricultural production.

Thus, the aim of the present study was to evaluate the diversity and relative efficiency of nodulating and endophytic bacterial communities from undisturbed native rainforests located within three iron ore mining sites in Serra Norte dos Carajás in the Eastern Brazilian Amazon. For that, we conducted two capturing experiments using soil from these forests and the trap plant species Macroptilium atropurpureum and Mimosa acutistipula.

Materials and methods

Characterization of the collection areas and sampling

The material of this study consisted of soil samples coming from the Serra dos Carajás (PA), the richest mineral resource region in Brazil and one of the most expressive mineral resource regions in the world [31]. This region is within the Carajás National Forest [Floresta Nacional (FLONA) de Carajás], in the southeastern part of the state of Pará [32]. The climate in the region according to the Köppen-Geiger classification is type “Awi”—tropical rainy with a dry winter, mean annual temperature of 26.2 °C, and rainfall accumulation of up to 1900 mm year⁻¹ [33]. The FLONA de Carajás constitutes an important conservation and sustainable management zone that includes conservation units, forest production, and mining, among other uses [32]. Three areas of undisturbed native rainforest within the iron ore mining sites were chosen to carry out this study: Arenito II (6°5′52.15″S, 50°13′38.42″W), Noroeste II (6°1′33.05″S, 50°11′31.10″W), and Sul IV (6°4′50.53″S, 50°10′1.99″W). All areas are within the Carajás Mineral Complex in the Serra Norte area of the FLONA de Carajás, in the municipality of Parauapebas, PA. Arenito II area corresponded with montane rainforest, while Noroeste II and Sul IV areas corresponded with submontane rainforest. The distance between them ranged from six to 8 km, and they differed in important soil properties (Table 1). In general, the environments were strongly acid soils, with very low P concentration, adequate SOM and K contents, and very high CEC and Fe content (soil analysis interpretation according to Alvarez et al. [34]). However, they differed from each other in Ca, Mg, Al, Mn, B, Zn, and Cu content and texture. High Al content, at toxic levels, stands out in Noroeste II, whereas higher contents of Mn, Zn, and Cu stand out in Sul IV. Arenito II has better soil chemical conditions than the other areas.

Table 1.

Chemical and physical analysis* of soil samples collected from tropical rainforests fragments (Arenito II, Noroeste II, and Sul IV). Means with different letters are significantly different at p < 0.05 by Scott-Knott test

AREA	pHH2O	SOM†	P	K	Ca	Mg	Al	CEC‡	Fe	Mn	B	Zn	Cu	Sand	Silt	Clay
		g kg−1	–mg dm−3–		––––––-cmolc dm−3––––––-				–––––––––––-mg dm−3––––––––––-					–––––––-%–––––––-
Arenito II	4.47	40.00	1.83	67.4	1.87a	0.40ab	0.47b	10.81	198.87	33.43b	0.90a	0.73b	0.23c	32.77a	12.23b	55.00
Noroeste II	4.27	45.33	2.30	68.2	0.30b	0.23b	1.17a	11.87	252.50	39.63b	0.63b	0.90b	2.30b	43.27a	13.07b	43.67
Sul IV	4.30	43.33	2.37	76.6	1.50a	0.53a	0.50b	9.53	169.23	107.87a	0.57b	2.13a	3.20a	11.70b	23.97a	64.33

^*pH in H₂O (1:2.5 v/v), organic carbon (Walkley & Black, 1934), available P, K, Fe, Mn, and Cu (extracted by Mehlich 1-Mehlich, 1953), exchangeable Ca²⁺, Mg²⁺, and Al³⁺ (extracted with KCl 1 mol L¹) (McLean et al., 1958); H + Al (extractor SMP) (Shoemaker, McLean, and Pratt, 1961), granulometry (Bouyoucos, 1951). † SOM = soil organic matter; ‡ CEC = cation exchange capacity

Soil samples were collected during the least rainy season of 2016 (October), selecting three plots (10 × 20 m) per area, with a minimum distance of 30 m between them. Within each plot, five systematically distributed points (in the form of an X) were sampled to form compound samples, with a total of three compound samples per area. The samples were taken to the Soil Biology, Microbiology, and Biological Processes Sector of the Department of Soil Sciences of the Universidade Federal de Lavras (DCS-UFLA), where they were stored at 4 °C until microbiological evaluations.

Experiment for capturing rhizobia and endophyte communities from nodules

Trap plant species were used to capture the bacteria: Macroptilium atropurpureum (DC.) Urb., a promiscuous host widely used in this type of study [27, 35, 36], and Mimosa acutistipula var. ferrea Barneby, a native variety restricted to ferruginous soils of the north of Brazil [37], popularly known as mimosa de canga. The species Mimosa acutistipula (Mart.) Benth., although rarely used in capture experiments [38], is registered as nodule-forming, and there are positive experimental results of nodulation [39, 40]. This situation, along with the availability of seeds of the variety typical to the location of the study, led to its choice as a second trap plant.

Aliquots of 10 g of soil from the each one of all samples of the three areas selected were suspended in sterile saline solution (NaCl 0.85%) in a 1:1 (w/v) ratio and shaken at 120 rpm for 30 min. After that, 1 mL of the soil suspensions was inoculated in recycled long-neck type glass bottles of 500 mL containing Hoagland and Arnon [41] nutrient solution at one-fourth strength and low nitrogen concentration (5.25 mg L⁻¹) together with newly germinated seedlings, using strips of filter paper as a support for root growth. The seeds had previously been exposed to H₂SO₄ (98%) for 30 min for M. atropurpureum and 10 min for M. acutistipula var. ferrea, to break dormancy and for surface disinfection. After disinfection, they were washed and incubated in sterile Petri dishes with moist cotton/filter paper from 48 to 72 h until root emergence, at which time they were transferred to the bottles.

The experiment also included the following treatments: control with low N concentration (5.25 g L⁻¹), C($\downarrow$ N); full N concentration (52.5 g L⁻¹), C(+ N); and low N concentration, inoculated with strains that efficiently form nodules in accordance with the host plant, specifically C(UFLA 04–212), which is Bradyrhizobium brasilense (Costa et al., 2017), for M. atropurpureum, and C(BR 3460), which is Burkholderia sp., for M. acutistipula var. ferrea. Although the latter strain had not been tested on the species before, it was expected to be able to nodulate it since it was effective on another species of the genus [42].

The experiment was conducted for 40 days (January to February 2018) in a greenhouse in a completely randomized experimental design with three replications. The mean maximum temperature recorded in the period was 28.4 °C, and the mean minimum temperature was 17.7 °C.

Relative efficiency of the nitrogen-fixing bacterial communities

After the established time and before collection of the plant material, the relative chlorophyll level, referred to as the SPAD value, was evaluated using the portable Minolta SPAD-502 device. The number of nodules per plant (NN) was also determined by random collection of two nodules from each one of the replications, whenever present, for later isolation of the bacteria. The shoots and roots of the plant and the rest of the nodules were then collected, which were dried in a forced air laboratory oven at 65 °C until constant weight for determination of nodule dry weight (NDW), shoot dry weight (SDW), and root dry weight (RDW). This last quantity was added to the NDW. In addition, based on the evaluations already mentioned, total dry weight (TDW) and the RDW/SDW ratio were determined.

In addition to nodulation and plant biomass production, the relative efficiency of the bacterial communities in nitrogen fixation was estimated from the nitrogen sufficiency index (NSI) [43]. This index is obtained from measurement of the relative chlorophyll level and calculated by the following equation: $=\frac{{\mathit{SPAD}}_X}{{\mathit{SPAD}}_{C(+N)}}$, where SPAD_X refers to the mean reading of the chlorophyll meter of the different treatments and SPAD_C(+N) to the mean reading of the chlorophyll meter under the condition of adequate nitrogen content. The value of the index will be 1.0 if the reading of the treatment and the control C(+ N) are equal and will decrease as the efficiency of biological N₂ fixation (BNF) deceases.

Isolation, purification, and bacterial culture characterization of the bacterial strains

The nodules collected for isolation were surface disinfected with ethyl alcohol (92.8%) for 30 s, submerged in H₂O₂ (3%) for 3 min, and washed 6 times with sterile distilled water. After that, they were crushed, and the liquid extracted was spread on a Petri dish with YMA culture medium [44], pH = 6.8, containing bromothymol blue as a pH indicator. The nodules used here were retained after bacterial isolation, dried, and included in determination of NDW.

The inoculated Petri dishes were incubated at 28 °C in a microbiological incubator, and, after growth, individual colonies were successively subcultured until obtaining pure cultures. After purification, the strains were grouped according to their culture characteristics, considering modification or lack of modification of pH (acid, alkaline, or neutral) and growth time. Growth was described as fast, intermediate, slow, and very slow for the intervals of 0–3 days, 4–5 days, 6–9 days, and 10 days or more, respectively, according to Moreira et al. [24].

Molecular identification and phylogenetic analysis

The strains were identified through partial sequencing of the 16S ribosomal DNA gene, using the primers 27F (5’ AGAGTTTGATCCTGGCTCAG 3’) and 1492R (5’ GGTTACCTTGTTACGACTT 3’) [45]. The DNA was extracted from fresh bacterial cultures by alkaline lysis, according to Niemann et al. [46], and amplified through the PCR technique [47]. The extraction procedure used 5 μL of DNA, 1 μL of dNTP Mix (10 mM), 5 μL of PCR buffer (10X), 2.5 μL of MgCl₂ (50 mM), 1 μL of each primer (10 mM), 0.4 μL of Taq DNA polymerase (5U μL⁻¹, Cellco Biotec do Brasil Ltda), and ultrapure water (Milli-Q) until reaching a final reaction volume of 50 μL. PCR consisted of initial denaturation for 5 min at 85 °C, followed by 30 cycles of denaturation, annealing, extension of 1 min at 95 °C, 5 min at 55 °C, 1 min at 72 °C, and final extension of 10 min at 72 °C. Amplification was checked by electrophoresis in 0.8% agarose gel of 5 μL of the reaction products in Tris–EDTA (1X) buffer for 45 min, using Neotaq Brilliant Green Plus DNA Stain (0.04 μL mL⁻¹ of gel) for visualization of the bands. The rest of the amplified material (45 μL) was lyophilized and sent to the Wemseq laboratory (Curitiba, PR, Brazil), where it was purified (EXOL/SAP-Fermentas) and sequenced through Sanger sequencing [48] using the 3500xL Genetic Analyzer device (Thermo Fisher). The sequences obtained were assembled and edited using the BioNumerics 7.6 software (Applied Maths, Austin, TX, USA), analyzed for detection of possible chimeras with the DECIPHER tool [49], and, if present, removed. The strains isolated were then identified by comparison of sequences with the type of material deposited in the database of the NCBI (National Center for Biotechnology Information) with assistance of the BLAST (Basic Local Alignment Search Tool) [50]. The sequences obtained were deposited in the GenBank [51] under accession numbers from MK848616 to MK848687.

Phylogenetic analysis of the strains from the three areas studied was performed using the Molecular Evolutionary Genetics Analysis (MEGA) version X software [52]. First, multiple alignment of the sequences obtained and of type strains of greatest similarity was carried out through the Multiple Sequence Comparison by Log-Expectation (MUSCLE) algorithm, with unweighted pair group method with arithmetic mean (UPGMA) clustering [53]. The phylogenetic tree was then constructed using the neighbor joining (NJ) method, with the Tamura-Nei evolutionary distance model [54]. In addition, the analysis included Bradyrhizobium type strains representing the supergroups B. elkanii, B. jicamae, and B. japonicum [55]. The reliability of the nodes formed was estimated through bootstrapping [56] with 1000 interactions and represented when the values obtained were ≥ 70%, corresponding to the probability of ≥ 95% that the clade formed is real, according to Hillis and Bull [57].

Analysis of nodulation ability of potentially nodulating strains

After identification, strains belonging to genera considered to be nodule-forming were analyzed regarding the presence of the nodD1 gene through PCR, using the primers nodD7f (5’ GATYGTCATGAAATCKGAGAG 3’) and aboxp2 (5’ TCGATAGAANACATCCACACGAT 3’) designed by Sterner and Parker [58]. The reaction mix was the same used for the amplification of 16S rRNA gene, and the PCR consisted of initial denaturation of 30 s at 90 °C, followed by 35 denaturation cycles, annealing, extension of 30 s at 94 °C, 1 min at 53 °C, 30 s at 72 °C, and final extension of 5 min at 72 °C. The presence of the gene was checked through electrophoresis in agarose gel.

Strains that did not have the nodD1 gene amplified were tested in relation to their ability to nodulate the host of origin in an experiment performed in a way similar to the capture experiment, inoculating a bacterial suspension estimated at 1.5 × 10⁸ ufc mL⁻¹ per plant.

Data analysis

Before statistical analyses, the normality and homoscedasticity of the quantified data were checked through the Jarque–Bera and Levene tests, respectively. The $\frac{1}{x}$ transformation was necessary for SDW, RDW, and TDW and $1/(x+1)$ for NN to satisfy the premises of the parametric analyses. However, only SDW, RDW, and RDW/SDW satisfied both premises and were compared through ANOVA. The remaining variables achieved only normality and were compared through Welch’s t-test for unequal variances. In both cases, when significant differences were detected, the variables were compared through the Scott-Knott post hoc test at p < 0.05. Lastly, the bacterial culture and genetic diversity of the microbial communities obtained were evaluated through the species richness, Shannon, and Simpson diversity indices, according to Magurran et al. [59].

Results

Relative efficiency of the nitrogen-fixing bacterial communities

Nodulation was absent for the controls C(+ N) and C($\downarrow$ N) in both species, and nodulation occurred for the control C(UFLA 04–212) in M. atropurpureum, indicating that the experimental conditions were both aseptic and appropriate for nodulation to occur. In the experiment with Mimosa acutistipula var. ferrea, there was no nodulation in the control C(BR 3460) or in the treatments with inoculum soil. Therefore, the results and subsequent evaluations were limited to strains captured with M. atropurpureum.

Nodulation was obtained in M. atropurpureum from soil samples from the plots of the three forest areas evaluated, without significant differences among them in NN or NDW, or with the control C(UFLA 04–212) (Table 2). However, there was wide variation in the nodulation from Arenito II, which was highest (75) in one of its plots. In the rest of the areas, nodulation was low for all the plots.

Table 2.

Mean values ± standard error (SE) of nodule number (NN), nodule dry weight (NDW), root dry weight (RDW), shoot dry weight (SDW), total dry weight (TDW), ratio root dry weight: shoot dry weight (RDW/SWD) and indirect measurement of chlorophyll content (SPAD value) of M. atropurpureum in the experiment of bacteria capture from tropical forest fragments (Arenito II, Noroeste II, and Sul IV)

Treatment	NNns	NDWns	RDW	SDW	TDW	RDW/SDW	SPAD value	NSI
	-	….….…………….…………….g…………….…………….….…				-	-
C(+ N)	-	-	0.53 ± 0.04a	0.87 ± 0.0045a	1.4 ± 0.04a	0.61 ± 0.05b	37.75 ± 3.42a	-
C($\downarrow$ N)	-	-	0.16 ± 0.0029b	0.12 ± 0.01c	0.28 ± 0.0038b	1.41 ± 0.1a	13.82 ± 0.14b	0.37 ± 3.8 × 10−3b
C(UFLA 04–212)	55.5 ± 0.29	0.04 ± 0.0027	0.13 ± 0.0027c	0.21 ± 0.02b	0.33 ± 0.01b	0.63 ± 0.07b	34.18 ± 1.32a	0.91 ± 0.03a
Arenito II	25.56 ± 24.56	0.04 ± 0.02	0.17 ± 0.03b	0.26 ± 0.07b	0.43 ± 0.08b	0.77 ± 0.23b	23.71 ± 6.66b	0.65 ± 0.17b
Noroeste II	1.22 ± 0.78	0.00057 ± 0.00028	0.14 ± 0.01c	0.13 ± 0.01c	0.26 ± 0.01b	1.12 ± 0.11a	17.18 ± 1.64b	0.46 ± 0.04b
Sul IV	1.61 ± 0.96	0.01 ± 0.01	0.17 ± 0.0012b	0.15 ± 0.01c	0.32 ± 0.01b	1.12 ± 0.06a	19.96 ± 3.23b	0.53 ± 0.09b

Means with different letters are significantly different at p < 0.05 by Scott-Knott test

In plant biomass production (RDW, SDW, TDW), the control C(+ N) was highly superior to the rest of the treatments. The TDW was low and similar among the areas and the controls C($\downarrow$ N) and C(UFLA 04–212); however, RDW and SDW, also low, varied in some cases. Arenito II and UFLA 04–212 had SDW greater than the control without N, indicating their efficiency. The RDW/SDW ratio differentiated C(+ N), C(UFLA 04–212), and Arenito II from C($\downarrow$ N), Noroeste II, and Sul IV. The SPAD reading differentiated the controls C(+ N) and C(UFLA 04–212) from the rest of the treatments (Table 2). The nitrogen sufficiency index (NSI) was 0.37, 0.91, 0.63, 0.46, and 0.53 for the treatments C($\downarrow$ N), C(UFLA 04–212), Arenito II, Noroeste II, and Sul IV, respectively. The NSI value obtained for C(UFLA 04–212), near 1, corroborated that the control strain used is efficient in N₂ fixation. Among the communities isolated from the three areas evaluated, Arenito II showed higher NSI, although the values obtained were low for the three areas.

In general, the bacterial community of Arenito II stood out as the most efficient of the three because it had greater aboveground biomass production and an adequate shoot biomass/root biomass ratio.

Bacterial culture and genetic characterization of the bacterial communities isolated

A total of 52 bacterial strains were isolated: 32 in Arenito II, 12 in Noroeste II, and 8 in Sul IV. They were distributed in six bacterial culture groups: fast growth and acidification of the medium (RA), fast growth and alkalinization of the medium (RB), fast growth and without modification of the pH of the medium (RN), intermediate growth and acidification of the medium (IA), intermediate growth and alkalinization of the medium (IB), and slow growth and alkalinization of the medium (SB). In general, the IB culture group predominated, followed by RB; however, there was variation within the areas. In Arenito II, five culture groups were identified, with predominance of IB and RA in equal proportion. For Noroeste II and Sul IV, four culture groups were recorded, with predominance of SB and IB, respectively (Fig. 1a).

Fig. 1.

Distribution of strains isolated by a cultural groups and b bacterial genera (b) from the fragments of tropical forest: Arenito II, Noroeste II, and Sul IV. Cultural groups, rapid growth and acidification of the medium (FA), rapid growth and alkalization of the medium (FB), rapid growth and no change in pH of the medium (FN), intermediate growth and acidification of the medium (IA), intermediate growth and alkalization medium (IB), and slow growth and alkalization of the medium (SB). Diversity indexes, richness (S), Shannon (H), and Simpson (1-D) for the cultural groups and genera found

Quality sequences of the 16S rRNA gene from 447 to 1293 bp were obtained, identifying, on the genus level, all the strains isolated with similarity of 95.99% to 100% with sequences of type strains deposited in the GenBank (Table 3). Eighteen (18) of the 52 strains identified proved to be equal to more than one type strain (Table 3, Table S1), reaffirming the insufficient resolution of the 16S rRNA gene for identification on the species level. This was most evident for the strains of the Bradyrhizobium genus.

Table 3.

Bacterial strains isolated from three tropical forest fragments (Arenito II, Noroeste II, and Sul IV) at Serra Norte de Carajás, their cultural characteristics at 79 medium, identification by partial sequencing of 16S rRNA gene fragments (length); accession number at GenBank, top-hit taxon (name and accession number of type strain with great similarity (SI), and amplification of NodD1 in potential nodulating genera

Fragment/strain	Identity	Cultural group†	Length (pb)	Accession number	Top-hit taxon (type strain) ‡				NodD1
					Name	Strain	Accession number	SI	amplification
								(%)
Arenito II
UFLA04-858	Bradyrhizobium sp.	IB	1238	MK848618	Bradyrhizobium pachyrhizi Ramírez-Bahena et al. 2009	PAC48	NR_043037	99.43	+
UFLA04-859	Bradyrhizobium sp.	IB	1237	MK848619	"	"	"	99.68	+
UFLA04-860	Bradyrhizobium sp.	FB	1240	MK848620	"	"	"	99.68	+
UFLA04-861	Bradyrhizobium sp.	IB	1200	MK848621	"	"	"	99.75	+
UFLA04-862	Bradyrhizobium sp.	IB	1240	MK848622	"	"	"	99.76	+
UFLA04-863	Bradyrhizobium sp.	IB	1240	MK848623	"	"	"	99.76	+
UFLA04-872	Bradyrhizobium sp.	IB	1243	MK848632	"	"	"	99.52	+
UFLA04-879	Bradyrhizobium sp.	IB	1237	MK848639	"	"	"	99.6	+
UFLA04-880	Bradyrhizobium sp.	IB	1233	MK848640	"	"	"	99.92	+
UFLA04-883	Bradyrhizobium sp.	FB	1244	MK848642	"	"	"	99.6	+
UFLA04-884	Bradyrhizobium sp.	IB	1200	MK848643	"	"	"	99.92	+
UFLA04-868	Bradyrhizobium sp.(6)	IB	1119	MK848628	"	"	"	99.55	+
UFLA04-869	Bradyrhizobium sp. (6)	IB	1256	MK848629	"	"	"	99.52	+
UFLA04-870	Bradyrhizobium sp. (6)	IB	1186	MK848630	"	"	"	99.75	+
UFLA04-867	Bradyrhizobium sp. (2)	IB	1237	MK848627	Bradyrhizobium stylosanthis Delamuta et al. 2016	BR 446	NR_151930	99.51	+
UFLA04-864	Leifsonia sp.	FA	1229	MK848624	Leifsonia shinshuensis Suzuki et al. 2000	DB 102	NR_043663	98.86
UFLA04-865	Leifsonia sp.	FA	1266	MK848625	"	"	"	99.53
UFLA04-882	Methylobacterium sp. (2)	FN	730	MK848683	Methylobacterium tarhaniae Veyisoglu et al. 2013	N4211	NR_109649	99.73	‒
UFLA04-885	Methylobacterium sp. (2)	FN	705	MK848684	"	"	"	99.57	‒
UFLA04-878	Paenibacillus sp.	FA	1203	MK848638	Paenibacillus soli Park et al. 2007	DCY03	NR_043708	96.85
UFLA04-881	Paenibacillus sp.	FA	1179	MK848641	"	"	"	97.29
UFLA04-886	Paenibacillus sp.	IA	493	MK848661	Paenibacillus aestuarii Bae et al. 2010	CJ25	NR_116365	97.57
UFLA04-887	Paenibacillus sp. (2)	IA	1146	MK848644	"	"	"	95.99
UFLA04-866	Pantoea sp.	FA	1288	MK848626	Pantoea agglomerans (Beijerinck 1888) Gavini et al. 1989	JCM1236	NR_111998	99.46
UFLA04-871	Pantoea sp.	FA	1293	MK848631	"	"	"	99.15
UFLA04-873	Pantoea sp.	FA	1248	MK848633	"	"	"	99.36
UFLA04-874	Pantoea sp.	FA	1251	MK848634	"	"	"	99.2
UFLA04-875	Pantoea sp.	FA	1285	MK848635	"	"	"	99.53
UFLA04-876	Pantoea sp.	FA	1252	MK848636	"	"	"	99.52
UFLA04-877	Pantoea sp.	FA	1293	MK848637	"	"	"	99.38
UFLA04-856	Paraburkholderia sp.	FA	1244	MK848616	Paraburkholderia sabiae (Chen et al. 2008) Sawana et al. 2014	Br3407	NR_043180	99.2
UFLA04-857	Paraburkholderia sp.	FA	1276	MK848617	"	"	"	99.29
Noroeste II
UFLA04-890	Bradyrhizobium sp. (3)	SB	1239	MK848647	Bradyrhizobium erythrophlei Yao et al. 2015	CCBAU 53,325	NR_135877	99.68	+
UFLA04-897	Bradyrhizobium sp. (3)	SB	1245	MK848653	"	"	"	99.76	+
UFLA04-892	Bradyrhizobium sp.	IB	1192	MK848648	Bradyrhizobium pachyrhizi Ramírez-Bahena et al. 2009	PAC48	NR_043037	99.83	+
UFLA04-899	Bradyrhizobium sp.	IB	1204	MK848654	"	"	"	99.75	+
UFLA04-893	Bradyrhizobium sp. (6)	IB	1201	MK848649	"	"	"	99.33	+
UFLA04-894	Bradyrhizobium sp. (6)	SB	1156	MK848650	"	"	"	100	+
UFLA04-895	Bradyrhizobium sp. (6)	SB	1164	MK848651	"	"	"	99.57	+
UFLA04-896	Bradyrhizobium sp. (6)	SB	1152	MK848652	"	"	"	99.83	+
UFLA04-888	Leifsonia sp.	FA	1240	MK848645	Leifsonia shinshuensis Suzuki et al. 2000	DB 102	NR_043663	99.19
UFLA04-889	Leifsonia sp.	FA	1197	MK848646	"	"	"	99.42
UFLA04-891	Methylobacterium sp.	FN	690	MK848685	Methylobacterium indicum Chaudhry et al. 2016	SE2.11	NR_135210	99.86	‒
UFLA04-898	Methylobacterium sp.	FN	690	MK848686	"	"	"	99.86	‒
Sul IV
UFLA04-902	Bradyrhizobium sp.	IB	1248	MK848657	Bradyrhizobium pachyrhizi Ramírez-Bahena et al. 2009	PAC48	NR_043037	99.12	+
UFLA04-900	Bradyrhizobium sp. (6)	FB	1145	MK848655	"	"	"	99.91	+
UFLA04-901	Bradyrhizobium sp. (6)	IB	1153	MK848656	"	"	"	99.91	+
UFLA04-904	Bradyrhizobium sp. (6)	FB	1238	MK848658	"	"	"	99.43	+
UFLA04-905	Bradyrhizobium sp. (6)	IB	1198	MK848659	"	"	"	99.58	+
UFLA04-906	Bradyrhizobium sp. (6)	IB	1153	MK848660	"	"	"	99.22	+
UFLA04-903	Methylobacterium sp.	FN	457	MK848687	Methylobacterium indicum Chaudhry et al. 2016	SE2.11	NR_135210	99.78	‒
UFLA04-907	Paenibacillus sp.	IA	447	MK848662	Paenibacillus aestuarii Bae et al. 2010	CJ25	NR_116365	97.54

^† Cultural groups at 79 solid medium: fast-growth acidifiers (FA), fast-growth alkalinizers (FB), fast-growth with no change of medium pH (FN), intermediate-growth acidifiers (IA), intermediate-growth alkalinizers (IB) and slow-growth alkalinizers (SB). ‡ Top-hit taxon name, first name of type strain that appeared in the BLASTn. Other type strains with the same similarity are found in Table S1

The Bradyrhizobium and Methylobacterium genera occurred in all the areas, with predominance of the first. Paraburkholderia and Pantoea were exclusive to Arenito II, whereas Leifsonia was recorded for Arenito II and Noroeste II and Paenibacillus for Arenito II and Sul IV (Fig. 1b, Table 3). In general, the bacterial communities of the forest areas studied had similar bacterial cultures, but they differed numerically and genetically, and Arenito II was the most diverse (Fig. 1, Table 3).

Phylogenetic analysis

The strains identified in the present study were distributed in the phyla Actinobacteria, Firmicutes, and Proteobacteria, with predominance of this last phylum, which included representatives of the sub-phyla α-, β-, and Ɣ-proteobacteria (Fig. 2).

Fig. 2.

Maximum likelihood phylogeny based on the 16S rRNA gene sequences (1142 bp) showing the relationships of bacterial strains from fragments of tropical forest: Arenito II (●), Noroeste II (▲), and Sul IV (■). Bootstrap values greater than 70% are indicated at the nodes. Type strains of Bradyrhizobium included in the analysis and which define the supergroups B. elkanii, B. jicamae, and B. japonicum are highlighted in bold

The 29 strains of Bradyrhizobium obtained separated into two main nodes with high reliability (bootstrap value from 96 to 97%). Twenty-eight of them clustered together with the types of many species, including B. elkanii and B. jicamae. Only the strain UFLA 04–867 segregated from these and joined the types B. japonicum, B. huanghuaihaiense, and B. stylosanthis. Thus, partial sequencing of the 16S rRNA gene allowed discrimination of the strain of the B. japonicum supergroup but not those of the B. elkanii and B. jicamae supergroups (Fig. 2).

Analysis of the nodulation ability of the potentially nodulating strains

Amplification of the nodD1 gene was obtained for all the Bradyrhizobium and Paraburkholderia strains, whereas it was negative for the Methylobacterium strains (Table 3). The inability of nodulation of these last strains mentioned was confirmed in the authentication experiment in the plant species of origin. Therefore, the NFLNB communities of Arenito II were represented by the Bradyrhizobium and Paraburkholderia genera and the Noroeste II and Sul IV areas only by Bradyrhizobium.

Discussion

This study characterized the bacterial cultures, the genetic structure, and the symbiotic association of symbiotic NFB communities and endophytic bacteria from nodules formed from soil bacteria from three undisturbed native rainforest areas within iron ore mining sites in the Amazon region of Pará.

Contrary to expectations, there was no nodulation in M. acutistipula var. ferrea by the control strain C(BR 3460) or by native strains within the soil used as inoculum. This showed its inefficiency in capturing NFLNB in the present study. Although the nodulation ability of the species is widely confirmed in the literature [38–40, 60], the variety used here did not nodulate. Borges et al. [30], in capture experiments using this variety, likewise did not find nodulation. More recently, Silva et al. [61] in a study related to its initial growth (45 days) in different substrates found nodulation only when it was grown in canga vegetation topsoil, with and without adequate nitrogen fertilization. However, when grown in forest topsoil or in mine tailings from the Serra dos Carajás itself, there was no nodulation. These results suggest that the symbiosis in M. acutistipula var. ferrea (also known as mimosa de canga) is highly specific and related to the region (substrate) of occurrence. In fact, Andrews and Andrews [62] indicated that although the promiscuity of the genus is widely reported, this is restricted to a few species with wide distribution and might not reflect the situation for most of the species of Mimosa, which are generally endemic and of restricted distribution.

M. atropurpureum, for its part, allowed capture of NFLNB in the three areas evaluated, once more confirming its efficiency as a trap plant for this type of study. Biomass production in the plants with adequate N fertilization was superior to the other treatments, including the inoculated control. Considering that establishment of symbiosis has a high energy cost for the host [63], the lower growth of nodulated plants may also be a reflection of this, above all during the first stages of growth, the period of evaluation of the experiment. However, the similar RDW/SDW ratio among the C(+ N) and C(UFLA 04–212) controls and Arenito II indicates an adequate balance in their vegetative growth, regardless of the absolute values of both variables.

The strong positive relation between chlorophyll content and N concentration in the leaves is the basis for the prediction of the N status of the crops [43]. In this context, the relative efficiency of the BNF of the native communities and of the inoculated control strain could be inferred from the NSI using the C(+ N) as a reference. The efficiency in BNF was excellent for the control C(UFLA 04–212), in which the inoculated strain was able to satisfy the N demand of the plant in a way similar to C(+ N), whereas the native communities proved to be less efficient. The fact that communities were less efficient does not imply that they do not harbors efficient strains. Usually, rhizobia communities are composed by strains with variable efficiency in relation to a given plant species [27]. Besides, symbiotic relationships of rhizobia communities can vary among host species.

The differences between the bacterial communities of the forests evaluated were probably due to their edaphic particularities. The better conditions of Arenito II may have favored greater nodulation (although without significant difference, due to the wide dispersion of the data) and therefore, the largest number of isolated strains obtained, while the high contents of Al in Noroeste II and of Mn, Zn, and Cu in Sul IV may have been limiting factors. However, these characteristics constitute an important selection factor, which can be positively exploited in later studies of evaluation and application of the strains under similar conditions.

Bradyrhizobium (with typical intermediate and slow growth and alkalization of culture medium) was the main nodulating genus from the three areas studied. This genus has been widely isolated from Brazilian Amazon ecosystems, both from trap plant crops [27–30] and directly from nodules from the field [24, 25, 64]. Currently, its phylogeny separates it into seven lines referred to as supergroups: B. japonicum, B. elkanii, B. jicamae, Kakadu, photosynthetics, Soil 1, and Soil 2, where the first five include nodulating genera [55]. In our study, most of the isolated strains were related to those recently separated and defined as the B. elkanii and B. jicamae sister supergroups [55]. Only the UFLA 04–867 strain was included in the B. japonicum supergroup and related to the species B. huanghuaihaiense and B. stylosanthis.

Avontuur et al. [55] analyzed the distribution of the sets of nif and nod genes in the different Bradyrhizobium lines, observing that both sets of genes may be absent, including in the supergroups of B. japonicum, B. elkanii, and B. jicamae. In addition, these authors found that when present, the nod genes are always accompanied by the nif genes, but not the contrary, that is, nodulating Bradyrhizobium spp. are always diazotrophic, but being diazotrophic does not imply that they are nodulating. In our study, all the genus strains obtained amplified nodD1, confirming they are nodulating and, considering the results of Avontuur et al. [55], also diazotrophic.

The Paraburkholderia genus (gen. nov., valid name), which includes nodulating species of Burkholderia sensu lato [65, 66], has been recorded in different ecosystems of the Western Amazon [27, 28]. In our study, the two strains obtained from Arenito II amplified nodD1, thus indicating they are nodulating. In contrast, strains of Methylobacterium, a genus that contains few representatives that nodulate [67, 68], did not amplify it. Although there are numerous factors of nodulation, whose presence and expression are not universal for all nodulating bacteria, nodD is a common transcriptional regulator. Specifically, nodD1 is necessary for the induction of the other nod genes in the presence of the specific signal emitted by the plant [69]. Therefore, its absence in the isolates of Methylobacterium, together with the negative experimental response of nodulation, confirms them as non-nodulating.

The quantity and diversity of nodulating bacteria isolated in the present study are low compared to agroecosystems and/or secondary Amazon forests, where greater diversity has been reported for the disturbed and/or modified areas [26–30], though similar to quantity found in native forests in the region [26, 30]. In fact, the low BNF in native tropical forests is widely recognized, this being attributed mainly to efficient cycling and accumulation of N, as well as to the considerable N/P imbalance in the soil [70–72].

Together with the rhizobia, a wide diversity of endophytic bacteria has been isolated from the nodules [12, 73, 74], commonly finding strains of the genera Paenibacillus, Pantoea, and Leifsonia, just as in our study. In contrast, Methylobacterium spp., also found, have been less common [3, 74]. The coexistence of rhizobia and endophytes from nodules has been indicated as not an accident [12, 75]. The study of their interactions is an important field of research and application, and there have already been positive experimental results from combined inoculation [13, 76].

The bacterial strains isolated during this study are already available in culture collection for further studies of biotechnological use in revegetation of these areas.

Conclusions

Our study shows the diversity of the communities of rhizobia and endophytic bacteria from nodules of M. atropurpureum inoculated with soil samples coming from areas of undisturbed native rainforest located in iron ore mining sites of the Eastern Brazilian Amazon. In general, very low nodulation was recorded in the three areas, with the Arenito II community standing out as most efficient in promoting plant growth. The predominant genera common to the three areas were Bradyrhizobium and Methylobacterium, the first being nodulating and the second endophytic. Representatives of the following genera were also found: Paraburkholderia (nodulating) and Paenibacillus, Pantoea, and Leifsonia (endophytic) in Arenito II, Leifsonia in Noroeste II, and Paenibacillus in Sul IV. The small number of nodules found resulted in a small number of isolates, which may not reflect the enormous diversity expected in the Amazon region. Nevertheless, the areas had considerable diversity, and the isolated bacteria constitute the first nodulating and endophytic strains from nodules isolated from ferruginous soils of the FLONA de Carajás. These strains have been incorporated in the collection of the Soil Biology, Microbiology, and Biological Processes Sector of the DCS-UFLA and will be studied in relation to their symbiotic characteristics with other leguminous species, as well as in relation to their ability to promote plant growth through processes other than biological N₂ fixation for use in rehabilitation of mining areas at this site.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

FMSM and MACC conceived and designed the research. RMRR conducted the experiment and performed the assessments. RMRR, AAG, and JLC carried out the molecular data analysis. RMRR processed the data and wrote the manuscript with support from FMSM, MACC, and JOS. All authors revised and contributed to the manuscript.

Funding

CAPES PEC-PG Grant/Award No., 23038.005401/2016–08; CAPES-PROEX Grant/Award No., 593–2018; CNPq Grant/Award No., 304527/2016–5.

Availability of data and materials

Sequence data that support the findings of this study have been deposited in GenBank with accession codes from MK848616 to MK848687.

Declaration

Conflict of interest

The authors declare no competing interests.