Genetic Diversity and Characterization of Symbiotic Bacteria Isolated from Endemic Phaseolus Cultivars Located in Contrasting Agroecosystems in Venezuela

Phaseolus vulgaris is a grain cultivated in vast areas of different countries. It is an excellent alternative to the other legumes in the Venezuelan diet and is of great agronomic interest due to its resistance to soil acidity, drought, and high temperatures. Phaseolus establishes symbiosis primarily with Rhizobium and Ensifer species in most countries, and this rhizobia-legume interaction has been studied in Asia, Africa, and the Americas. However, there is currently no evidence to show that rhizobia nodulate the endemic cultivars of P. vulgaris in Venezuela. Therefore, we herein investigated the phylogenetic diversity of plant growth-promoting and N2-fixing nodulating bacteria isolated from the root nodules of P. vulgaris cultivars in a different agroecosystem in Venezuela. In comparisons with other countries, higher diversity was found in isolates from P. vulgaris nodules, ranging from α- and β-proteobacteria. Some isolates belonging to several new phylogenetic lineages within Bradyrhizobium, Ensifer, and Mesorhizobium species were also specifically isolated at some topographical regions. Additionally, some isolates exhibited tolerance to high temperature, acidity, alkaline pH, salinity stress, and high Al levels; some of these characteristics may be related to the origin of the isolates. Some isolates showed high tolerance to Al toxicity as well as strong plant growth-promoting and antifungal activities, thereby providing a promising agricultural resource for inoculating crops.


Phaseolus vulgaris (common bean) is indigenous to
America and is currently the most cultivated legume worldwide after soybean (Velázquez et al., 2005). There are approximately 1,300 wild types of P. vulgaris. The remaining types are distant relatives of the common bean (CIAT, 2001). Approximately 50 wild-growing species are distributed across American countries (Cobley and Steele, 1976). This genus includes between 150 and 200 cultivated species used as food or garden ornamentals. Phaseolus species have been domesticated in different ecosystems ranging from mesic and temperate to warm and cold, humid, or hot and dry and exhibit distinct adaptations and reproductive systems. Phaseolus represents a wide range of habits, including annual and perennial, and some beans (e.g., large lima) may behave as short-lived perennials, bushes, or climbers (Cobley and Steele, 1976;Miklas and Singh, 2007). Several species are important for human or animal consumption, such as common beans (Phaseolus vulgaris L.), lima beans (P. lunatus L.), runner beans (P. coccineus L.), tepary beans (P. acutifolius A. Gray), and year-long beans (P. polyanthus Greenman) (Debouck et al., 1993). Phaseolus is primarily used as a food crop throughout Latin American and African countries. It is considered the core of bean diversity from which wild beans dispersed northwards and southwards to form the two geographically distinct gene pools of Mesoamerica and Andean South America, including the Venezuelan Andes (Gepts, 1998). However, P. vulgaris is a warm-season crop that cannot tolerate frost or extensive exposure to near-freezing or overheating temperatures at any growth stage (CIAT, 2001). Additionally, low P and N availability constrain common bean production, mainly in Hispanic America and Africa (Wortmann and Allen, 1994;Patel et al., 2010). Phaseolus is widely cultivated in Venezuela and mainly grown on erosion-prone slopes or in places with limited fertility. In the highlands, the plants are nodulated by native rhizobial populations without anthropological interventions. However, this diversity remains uninvestigated.
Rhizobia associated with several legumes have been reported in Venezuela (Vinuesa et al., 2005;Marquina et al., 2011;Artigas et al., 2020). Marquina et al. (2011) described rhizobia associated with one variety of P. vulgaris from Inceptisol. Venezuelan Glycine max and Vigna rhizobia were derived from different genera, including novel strain lines within the Burkholderia/Paraburkholderia group (Artigas et al., 2019). However, limited information is currently available on the nodulating bacteria related to endemic legumes in Venezuela, such as Phaseolus spp. Therefore, we herein examined the phylogenetic diversity and physiological characteristics of Phaseolus rhizobia from different Venezuelan ecosystems and topographical regions. The present study focused on two main crop cultivars continuously cultivated in different regions. We also investigated the relationship between the phylogeny and distribution of rhizobia in Venezuela and host specificity between two endemic Phaseolus varieties.

Soil samples and collection sites
Root nodules were collected from P. vulgaris 'Tacarigua' (Black cultivar) and P. vulgaris 'L2234-MGM' (White cultivar) in Aragura and Lara and from the White cultivar in Guárico (Table 1 and Fig. 1).
Soil samples were collected from ten Venezuelan regions (Table  1) and used in another study by Artigas et al. (2019;2020). No inoculants were used in these soil-collected areas; therefore, the strains obtained were considered to be indigenous to Venezuela. The ecosystems of the soil-collected areas and soil types are shown in Fig. 1 and Table 1. Al and pH values in Fig. 1 are reported in Casanova (2005), which were assessed using standard Al/pH-soil methods (Hsu, 1963;Jones, 2001).

Isolation of rhizobia from Venezuelan soils using endemic
Phaseolus cultivars as trap hosts P. vulgaris 'Tacarigua' (Black cultivar) and 'L2234-MGM' (White cultivar) are used as trap hosts to isolate rhizobia strains. Seeds were surface-sterilized with 70% (v/v) ethanol for 1 min and 3% (v/v) sodium hypochlorite for 2 min, followed by washing four times with sterile distilled water (Artigas et al., 2019). The seeds were pre-germinated in darkness under sterile conditions at 28°C for 48 h on filter paper (Whatman TM No.GF/A, glass microfiber fil-ter, 90 mm) moistened with 5 mL of sterile Milli-Q water (the Direct-Q ® 3 UV system; Merck Millipore) in a sterile Petri dish. Filter paper and water were sterilized with an autoclave at 120°C and 0.2 MPa for 30 min (Sylvester-Bradley et al., 1983) before use. Pre-germinated plants were inoculated with 2 g of each soil suspended in 10 mL of sterile water and then transferred into 300-mL glass jars containing 120 g of sterilized vermiculite (Vermitech). Sterilized N-free nutrient solution (Somasegaran and Hoben, 1994) was added to the jar at 0.6 mL g -1 vermiculite, and this moisture level was maintained throughout the growth period by supplementing with N-free solution. Plants were grown at 28°C for four weeks in a growth room with a 16-h light (5000~7000 LUX)/8-h dark photoperiod. After four weeks, root nodules were harvested and their surface was sterilized with 70% (v/v) ethanol for 30 s, washed with sterile distilled water, and then immersed for 1 min in 5% (v/v) sodium hypochlorite, followed by washing four times with sterile distilled water. Surface-sterilized root nodules were crushed in 500 μL of glycerol solution (50% [v/v]) to obtain bacterial suspensions. An aliquot (10 μL) of each suspension was streaked onto Yeast Mannitol agar (YMA) (Vincent, 1970) and incubated for one week at 28°C. The remaining suspension was frozen at -80°C for further isolation (if necessary). Single colonies were re-streaked onto fresh plates to obtain pure colonies. Strains were phenotypically characterized for their growth rate, texture, and color on YMA plates. All isolates were re-inoculated onto the hosts following the authentication protocol using the modified plant pot experiment type (CIAT, 1988).
Authentication of symbiotic activity and performance P. vulgaris 'Tacarigua' (Black cultivar) was used to examine the symbiotic activity and performance of isolated strains. One hundred and twenty isolates were selected (Table 1) and grown in YM broth at 28°C for 5 days to obtain 10 9 cells mL -1 , as described by Vincent (1970). Cells were collected by centrifugation at 10,000 rpm at 4°C for 5 min, followed by resuspension in TE buffer (1 mM EDTA in 10 mM Tris-HCl [pH 8.0]). Prior to being inoculated, Phaseolus seeds were surface-sterilized with 70% (v/v) ethanol for 30 s and 3% (v/v) sodium hypochlorite for 2 min followed by washing four times with sterile distilled water. One milliliter of each rhizobial cell suspension containing 10 8 cells in TE buffer was then inoculated on each seed of 'Black' P. vulgaris. Inoculated seeds were sown in plant boxes (7.6×7.6×10.2 cm) with 200 g of vermiculite (Vermitech). Sterilized N-free nutrient solution (Somasegaran and Hoben, 1994) was added at 0.6 mL g -1 vermiculite, and moisture was maintained throughout the culture by supplementing with N-free solution. One plant in each plant box was grown at 28°C in a growth chamber (Fli 2000; EYELA, Tokyo Rikakikai) with a 16-h light/8-h dark photoperiod. Plants with no inoculation served as the treatment control (Vincent, 1970).
The entire plant with the root nodules was collected 30 days after the inoculation to assess N 2 fixation activity based on acetylene reduction activity (ARA). ARA was detected using a Shimadzu GC-2014 gas chromatograph (Shimadzu) equipped with a Porapak N column (Agilent Technologies) with a 30-min incubation (Artigas et al., 2020). Root nodule numbers were confirmed. Shoot and root weights were measured after they had been dried at 80°C for 48 h.

Isolation of genomic DNA
Sixty-three isolates were selected based on their nodulation ability. DNA was extracted from isolates grown in YM broth medium at 28°C for 4 days. Before genomic isolation, cells were collected and washed twice with equal volumes of PBS buffer (NaCl 137 mM, KCl 2.7 mM, Na 2 HPO 4 10 mM, and KH 2 PO 4 1.8 mM [pH 7.2]). Total genomic DNA was extracted from isolates using the CTAB method described by Artigas et al. (2019), and DNA concentrations and purities were confirmed using a NanoDrop 2000 UV-vis spectrophotometer (Thermo Fisher Scientific). Total 700 512 1,212 235 120 63 † † Amazonas soil is classified into Entisol and Oxisol; however, the sampling site was Oxisol. † P. vulgaris cultivars differ from those used in the present study.

DNA amplification and sequencing
Primer sets for the 16S rRNA gene are described in Young et al. (1991) and Artigas et al. (2019), and primers for the genes of DNA recombinase A (recA), ATP synthase (atpD), and glutamate synthase (glnA) are described in Gaunt et al. (2001) for αproteobacteria and Baldwin et al. (2005) for β-proteobacteria. The nifH primer set described in Laguerre et al. (2001) and the nodD gene primer set described in Risal et al. (2012) and Zézé et al. (2001) were used. Amplification was performed using the thermal cycler (GeneAmp PCR system 9700; Applied Biosystems) described by Artigas et al. (2020). PCR products were examined using a 1.5% (w/v) agarose gel with 0.5× TBE buffer (10×: 1 M boric acid, 0.02 M EDTA·2Na, and 1 M Tris-HCl base [pH 8.0]) mixed with 0.5 μg mL -1 ethidium bromide. Bands with the predicted sizes were then excised from gels, and DNAs were purified using a Fast-Gene ® agarose gel/PCR extraction kit (Nippon Genetics).
According to the manufacturer's protocols, PCR products were sequenced using the ABI Prism 3500 Genetic Analyzer (Applied Biosystems). The sequences obtained were aligned using the Clus-talW method and then compared in the GenBank database (https:// www.ncbi.nlm.nih.gov/genbank/) using the online software BLAST algorithm-based sequence alignment. Phylogenetic trees were constructed by Genetyx version 11 and MEGA version 12.0 (Tamura et al., 2013) based on a neighbor-joining analysis and using the bootstrap method with the Maximum Composite Likelihood model without topology. Multilocus sequence typing (MLST) was conducted based on 16S rRNA and housekeeping genes (Artigas et al., 2019).
Accession Numbers: The sequences obtained for the different genes found in the present study have been deposited in the DNA Databank of Japan (DDBJ) under the following accession numbers: LC585433-LC585495 for 16S rRNA sequences, LC585496-LC585558 for the atpD gene, LC585559-LC585621 for the recA  Artigas et al. (2020) using Google Earth software ver. Pro. Red plants indicate the areas at which Phaseolus (Black cultivar) are continuously cultivated. pH and Al values were reported by Casanova (2005), and regions without Al values indicate that Al was not detected from soils in these areas. REDBC and INIA-Venezuela reported average temperatures. gene, LC585622-LC585684 for the glnA gene, LC585685-LC585747 for the nifH gene, and LC585748-LC585810 for the nodD gene.

Abiotic stress tolerance profiles of Phaseolus rhizobia
Isolates were initially grown in YM broth at 28°C for 5 days, and 5 μL of cell suspensions at 10 8 cells L -1 was transferred onto YMA plates or broth followed by an incubation at 28°C for 5-10 days with different stress conditions, such as high temperature, alkalinity, acidic pH, high salinity, and a high concentration of Al at different pH levels, as described in Artigas et al. (2019). The temperature tolerance of isolates was based on their ability to grow under the following temperatures: 25, 28, 35, 40, and 45°C on YMA plates, with 28°C being set as the control. The ability to grow at different pH levels was examined at pH 4.5, 5, 6.8, 8, 9, or 10, with pH 6.8 being set as the control (Somasegaran and Hoben, 1994), and pH in YMA plates was adjusted with 0.5 M HCl or 0.5 M NaOH. In the salinity tolerance test, YMA was supplemented with NaCl at 0 (control), 1, 2, 3, or 4% (w/v). Al tolerance was evaluated with 0 (control), 0.1, 0.5, 1, or 2 mM of AlCl 3 ·6H 2 O (Wako Pure Chemical) under acidic (pH 4.5) or neutral (pH 6.8) conditions. After 5 days, colony-forming units (CFUs) were calculated by plate counting under stress conditions. The growth of isolates was estimated relative to the control treatment (non-stress) as follows: no growth; weak growth (10-20% of the control); good growth (30-60% of the control); and excellent growth (similar to/the same as the control) (Somasegaran and Hoben, 1994;Marquina et al., 2011). These experiments were performed in triplicate for each isolate.

Indole-3-acetic acid (IAA) production
Each strain was inoculated into YM broth containing 100 mg L -1 L-tryptophan and incubated at 28°C for 5 days in darkness. Cell suspensions were then centrifuged at 10,000 rpm for 15 min, and IAA concentrations in the supernatants were measured using the Salkovski colorimetric technique (Glickmann and Dessaux, 1995) by measuring absorbance at 530 nm with a spectrophotometer (Ultrospec 3300 pro; Amersham Biosciences). These experiments were performed in triplicate for each strain.

Antifungal profile of Phaseoli isolates
Bacterial isolates were grown in YM broth medium at 28°C for 5 days. Each cell suspension at 10 7 cells mL -1 was applied to test antifungal activity using the Kirby-Bauer disk diffusion susceptibility modified test protocol described in Hudzicki (2009) (MAFF No. 328090). Two controls were set: a plate with an isolate without a pathogen and a plate with a pathogen without an isolate. The zone inhibition diameter and colony size were measured, and the percent inhibition of the growth of the test pathogen was calculated. The experiment was performed in triplicate for each rhizobia strain and pathogen.

Statistical analysis
Dunnett's test was performed using StatSoft 12.0.

Characterization of Phaseolus rhizobia isolated from different Venezuelan soils
The ecosystems, soil types, and histories of legume cultivation at the locations at which soils or nodules were sampled are shown in Table 1, and pH, Al concentrations, and temperatures at these sites are shown in Fig. 1. In Aragua, soils and nodules were sampled from fields with and without fertilizer for Phaseolus cultivation. Seven out of the ten sampling sites had a cultivation history of Phaseolus or other legumes or Fabaceae vegetation, such as Acacia, Mimosa, or Inga. These sites included the Andes (Trujillo and Mérida) and Floodplain (Apure) with acidic soils and high concentrations of exchangeable Al (Fig. 1).
A total of 1,212 root nodules were collected from two endemic P. vulgaris cultivars: 700 nodules from 'Tacarigua' (Black cultivar) and 512 nodules from 'L2234-MGM' (White cultivar), with 173 root nodules being collected from P. vulgaris grown in the four field locations and 1,039 nodules being harvested from pot cultivations inoculated with the sampled soils (Table 1). The Aragua Valley ecosystem is located in north-central Venezuela; legumes of various genera, such as Vigna, Canajus, Phaseolus, and Glycine, were cultivated with or without fertilizer. This sampling site pro-duced a large number of root nodules, which accounted for 24% of all root nodules obtained in the present study. Mérida (Andes-Temperate) showed the second highest nodulation production (17% of the total). 56 root nodules were obtained in the Amazonas (rainforest); this site is located in the Guiana Highlands, in which crop production has traditionally been performed (e.g., cucumber, tomato, and coriander). In Apure (Floodplain), where nutrients such as N and P are deficient, fewer root nodules were obtained than in most of the other sites. In Lara and DC, no nodule was obtained with the 'Black' cultivar in pot isolation. The Lara region is classified as a dried savanna with a xerophilic ecosystem in which soils are sandy with a low nutrient supply.
All root nodule homogenates from 1,212 nodules were streaked onto YMA. However, bacteria were isolated from 235 (19% of the total) nodule suspensions. Among them, 120 strains representative of each site based on their phenotypes were selected and inoculated into P. vulgaris 'Tacarigua' (Black cultivar). Only 63 isolates produced nodules (Table 1), and were subjected to a phylogenetic analysis.

Phylogenetic analysis and distribution of Venezuelan rhizobia
MLST results obtained using the 16S rRNA gene and housekeeping genes (Fig. 3) were similar to those of the phylogenetic tree based on the 16S rRNA gene (Fig. 2). Venezuelan isolates were clustered into two bacterial groups: α-proteobacteria (GI) and β-proteobacteria (GII) with two Pseudomonas species in the out-group (γproteobacteria). Most of the Venezuelan isolates (94% of the total) were identified as α-proteobacteria (Fig. 2).
The GI cluster was divided into five sub-groups, GIA to GIE (Fig. 3A). GIA consisted of 37 Venezuelan isolates and 12 reference strains, including important Latin American strains, such as R. mesoamericanun CCGE501, R. phaseoli ATCC14482, R. etli CFN42, R. tropici CIAT899, and R. laguerreae FB206, and were further divided into eight subgroups. The Rhizobium isolates in GIA were widely distributed in Venezuelan regions, except in Falcón (Table 2). Rhizobium was predominant in Mérida, DC, and Aragua without fertilizer (Table 1 and 2). GIB consisted of the reference strains Agrobacterium fabrum and R. pusense and six Venezuelan isolates from Trujillo, Amazonas, Guárico, and Lara (Fig. 3B). GIC contained two references of Ensifer and four Venezuelan isolates from Falcon (Table 2 and Fig. 3B). GID grouped Mesorhizobium reference species and two Venezuelan isolates from Guárico; this is the first study to identify Mesorhizobium in Venezuelan cultivars (Fig. 3B). The isolates classified as Ensifer and Mesorhizobium showed more biogeographic specificity than Rhizobium and Bradyrhizobium (Table 2). GIE consisted of ten isolates and Bradyrhizobium reference strains (Fig. 3B) ence strains of β-proteobacteria, such as Burkholderia and Paraburkholderia (Fig. 3B), with VGW7B and VLaW4 being related to Paraburkholderia phymatum. The other isolates, VMP6 and VAmP8, were classified as Burkholderia sp. (Table 2).

Phylogenetic analysis based on nodD gene sequencing
To construct a phylogenetic tree based on nodD gene sequences, Azorhizobium caulinodans ORS571 was selected as the symbiotic out-group (Fig. 4). Fifty-nine isolates were grouped as GI with different genera and further sub-grouped as GIA to GID (Fig. 4). GIA consisted of 38 isolates divided as follows: 26 isolates closely related and eight isolates slightly related to Rhizobium sp. (HQ670661.1)/R. etli CFN42 (U80928.5), and four isolates not related to any reference strain. GIA isolates included VAFW5, which was in the Agrobacterium group based on MLST (Fig. 3B). In contrast to MLST, no isolate was classified as R. pusense based on nodD. GIB contained VFP1, VFP6, and VFP4, classified as Ensifer sp. in MLST, which were closely related to the Ensifer mexicanus reference strain based on the nodD gene. GIC grouped Mesorhizobium reference strains, and no Venezuelan isolate was included in this group.
The Bradyrhizobium group (GID) was divided into two sub-groups (Fig. 4). The first group contained 9 Venezuelan isolates and reference strains of B. embrapense, B. yuanmingense, B. liaoningense, and B. japonicum. VGP2 and VGW2 from Guárico, classified as Mesorhizobium based on MLST, were classified into the B. japonicum group based on the nodD gene sequence. VGP6 and VGP9 were also in this sub-group with the nearest relationship with B. embrapense SEMIA 6208. The second sub-group contained 9 Venezuelan isolates and the reference strains of B. elkanii and B. pachyrhizi (Fig. 4). Five of these isolates were related to R. pusense in MLST (Fig. 3B). VFP9 was close to Ensifer references in MLST; however, this isolate was related to B. elkanii based on the nodD gene. GII consisted of four isolates and Paraburkholderia reference strains (Fig.  4). These four isolates were also classified with βproteobacteria in MLST (Fig. 3B). In the phylogenetic tree based on nodD sequences, three isolates in GII were clustered with P. phymatum.

Phylogenetic analysis based on nifH gene sequences
The phylogenetic tree based on nifH gene sequences included A. caulinodans ORS571 as a symbiotic out-group (Fig. 5). The classification of isolates was similar to the nodD gene analysis, with two groups GI and GII. GI was further divided into four sub-groups: GIA to GID. GIA contained 60% of the Venezuelan isolates and Rhizobium reference strains. VAW6 was classified as R. mesoamericanum based on the nifH gene (Fig. 5), similar to MLST (Fig. 3A). Thirty-six isolates were closely related to phaseoli reference strains based on the nifH gene (Fig. 5), and were further divided into three sub-clusters with four references of phaseoli strains. The first sub-cluster included the type strain R. phaseoli ATCC14482; this type strain isolated from Mexico with Phaseolus species was found in different American continent regions and was reported to be closely related to R. etli bv. phaseoli (Ribeiro et al., 2013), which is consistent

VMP6 VAmP8
Paraburkholderia caribensis DSM13236 T (CP026101.1) 3B and 5). Mesorhizobium references were classified as GIC VApP1, VLaP5, VAmP2A, VGW15C, and VAFW5 were without any Venezuelan isolate (Fig. 5). closely related to B. elkanii based on the nifH gene (Fig. 5), GID consisted of 19 isolates and reference strains of while these isolates were classified as R. pusense in MLST Bradyrhizobium (Fig. 5) and were further sub-divided as (Fig. 3B). VAW3, VMiP4, VLaW27, VAFP8, and VMiP5 follows. VGP2 and VGW2 were sub-grouped with the refer-were closely related to B. yuanmingense based on the nifH ence of B. japonicum based on the nifH gene, while these gene (Fig. 5), showing similar classifications to those in isolates were classified as Mesorhizobium sp. in MLST. MLST (Fig. 3B). VGP9 and VGP6 were classified into the VLaW3, VGP2B, and VAFP9 were sub-grouped with B. Bradyrhizobium group; however, since there was no referembrapense SEMIA 6208 based on the nifH gene. VTrW6, ence strain closely related to them, they were classified as

Paraburkholderia sabiae LMG24235 T
Rhizobium sp.  Bradyrhizobium sp. grouped with Bradyrhizobium in GID for the nifH gene. The remaining isolates were in GII with strains of β-Consequently, the Nif assessment was based on the detecproteobacteria (Paraburkholderia) based on the nifH gene tion of N 2 fixation. Several isolates did not exhibit the abil- (Fig. 5); this is congruent with the phylogenetic analysis ity to fix N 2 because ARA was not detected (Table 2). based on MLST and the nod gene. Three isolates, VAmP8, Ineffective isolates were mainly related to Bradyrhizobium, VMP6, and VGW7B, were closely related to P. phymatum. and two isolates belonged to R. pusense and Ensifer. VMP2 VLaW4, classified as Paraburkholderia in MLST, was was classified as Rhizobium sp. based on MLST, and the GIA GIB Rhizobium phaseoli 1713 (KR262784) nodD and nifH genes showed the highest ARA cally characterized and assessed under different abiotic (89.2±3.2 μM C 2 H 4 h -1 g -1 dry weight of nodules). stress conditions (Table S1). Growth rates were classified into three groups: fast growers (24-72 h, 74% of isolates), Physiological characterization of Phaseolus rhizobia under intermediate growers (96-120 h, 24% of isolates), and slow abiotic stress conditions growers (≥144 h, 2% of isolates). Isolates were classified All 63 genetically categorized isolates were phenotypi-into three types according to color: 14 were white (W), four were transparent (T), and the remaining were whitetransparent (WT). According to texture, four isolates were classified as sticky (SS) and 59 as creamy (C).
Strains showed different growth abilities under high temperature conditions. A total of 99% of isolates grew at 45°C, with weaker, better, or the same growth as the control. All isolates grew at 20°C, whereas those from hot regions (Lara and Falcon) showed weak growth. VAFW14, VGP9, and VLaW3, classified as Bradyrhizobium in MLST, were fast growers (48 h). The majority of isolates utilized sucrose as a carbon source and exhibited the ability to live as free-living bacteria using Ashby medium, which is generally used for free-living diazotrophic bacteria (unpublished data).
Salinity tolerance was assessed by recording growth with NaCl. A total of 98% of isolates grew at low NaCl concentrations (1 and 2%) with better or the same growth as the control. Two isolates, VMiP5 from Alfisol (B. yuanmingense) and VTrP4 (Rhizobium sp.) from acidic soil (Ultisol), were not tolerant of NaCl at 3 or 4%. The growth of isolates from Apure, Guárico, and Falcon was not affected by NaCl, even at 4%. Four isolates obtained from Falcon (Aridisol), 14 out of 15 isolates from Guárico or Lara (Vertisol), and two from Andes (Ultisol) showed high salinity tolerance growing at 4% of NaCl (Table S1).
Venezuelan isolates showed tolerance to different pH conditions from acidic to alkaline; however, their growth was inhibited more by acidic than alkaline conditions. Among the isolates that mainly originated from alkaline soils, such as Falcon, Guárico, and Aragua (without fertilizer), 18% did not grow at pH 4.5, and VTrP4 (Rhizobium sp.) did not grow even at pH 5.0. In contrast, all isolates survived under alkaline conditions at pH 10, with only 2% showing weak growth (Table S1).
The effects of Al on bacterial growth were recorded under two pH conditions. All isolates, except for VTrP4, survived with 2 mM Al at neutral pH. In contrast, only 8% of isolates grew under 2 mM at pH 4.5 (Table S1). Among isolates from soils with neutral or alkaline pH, such as Aragua (without fertilizer), Guárico, and Falcon, growth was severely inhibited by 2 mM of Al at pH 4.5.
The antibiotic resistance profiles of isolates were examined against the following antibiotics, spectinomycin (Spe), streptomycin (Str), kanamycin (Kan), nalidixic acid (Nal), and chloramphenicol (Cp). Some isolates exhibited multitolerance, and VAFP4, VAFW1, VAW10, VAW6, VMP3, and VLaW3 exhibited resistance to all antibiotics (Table  S1). A total of 90% of isolates were resistant to Nal. In contrast, the growth of 86% of isolates was strongly inhibited by Str. Most isolates obtained from Lara, Amazonas, Trujillo, Miranda, and Aragua (without fertilization) were sensitive to Str.
No relationship was observed between antibiotic resistance and pH or Al tolerance (P≥0.05). VTrP4 (Rhizobium sp.) was susceptible to all antibiotics. Furthermore, it did not survive under low pH and high Al. In contrast, VGP6 and VGP9, classified as Bradyrhizobium sp., showed high tolerance to almost all abiotic stress conditions. Other isolates with high performance under abiotic stress were VAP4, VApP10, VApP8, VMP23 (classified as Rhizobium), and VGW2 (Mesorhizobium sp.) (Table S1).

Plant growth-promoting profiles of Venezuelan isolates
A summary of symbiotic performance, physiological profiles as plant partners, antifungal profiles, and phylogenetic groups based on MLST are shown in Table 2. Physiological profiles as plant partners included activities directly or indirectly related to plant growth, such as IAA, P, or K solubilization. Six isolates significantly increased plant biomass with ≥2,000 mg plant -1 , and were classified as Rhizobium (VMiP1, VApP5, VAmW2, VTrW6, and VMP2), except for VAmP8, which was classified as Burkholderia in MLST. Auxin hormone production was dominated by Rhizobium (VAFP10, VAP4, VMP1, VMP8, and VLaP2) and Mesorhizobium (VGP2) isolates with more than 90 μg mL -1 . Fourteen isolates did not produce detectable levels of IAA, but exhibited strong antifungal activity (Table 2). Isolates with the best performance were as follows: VMP1 (Rhizobium sp.) for IAA (198.99 μg mL -1 ), VFP6 (Ensifer sp.) for P solubilization (1.67 index), and VMP23 (R. phaseoli) for K solubilization (1.22 index). Among all isolates, only VGP4 (Rhizobium sp.) and VFP6 (Ensifer sp.) showed Fe chelation (by CAS media, unpublished data).
Furthermore, 46% of isolates exhibited antifungal activity against at least one pathogen at different levels ( Table 2). Twenty-five isolates exhibited activity against P. aphanidermatum, whereas only ten isolates exhibited activity against C. gloeosporioides. VTrP29 (Rhizobium sp.) showed good responses against all fungal pathogens, while VGP9 (Bradyrhizobium sp.) exhibited very high potential against all tested pathogens, mainly P. aphanidermatum, R. necatrix, and H. mompa. VTrP4 (Rhizobium sp.) exhibited activity against four pathogens (P. aphanidermatum, R. solani, P. oryzae, and R. necatrix). These isolates with high antifungal performance exhibited low physiological activities for plant growth promotion and symbiotic performance.

Genetic distribution and relationship of symbiotic genes
The diversity of rhizobia isolated from P. vulgaris has been examined worldwide with various techniques and criteria. Collectively, the present results indicate that many species nodulate Venezuelan common beans; these cultivars were promiscuous hosts. Bacterial cells were only obtained from 19% of the 1,212 nodules collected in the present study (Table 1), indicating that many were unculturable. Furthermore, only 63 out of 120 strains produced nodules when inoculated on the Black cultivar, which may have been due to the specificity of the isolates to Phaseolus species because we only used one type of cultivar to corroborate nodulation activity. Phaseolus rhizobia were widely distributed in Venezuela, including areas without its cultivation history. The highest number of isolates was found in an area with the soil type Inceptisol, which may be because beans have been continuously cultivated in areas with Inceptisol, such as in Aragua (Fig. 1, Table 1).
The geographical distribution and diversity of rhizobia are mainly influenced by the climate and distribution of host legumes; however, other factors in soil parameters, such as pH, salinity, and nutrient content, are important because they affect the distribution of host legumes (Córdova-Sánchez et al., 2011;Chidebe et al., 2018). Graham (1992) reported that crop production and rotation, soil pH, and other factors may affect rhizobial diversity and occupancy. The present results demonstrated that legumes belonging to the Phaseolus genus were nodulated by rhizobia across αand β-proteobacteria, which is consistent with the findings of Andrews and Andrews (2017). However, in most studies, Phaseolus symbionts were reported separately with a maximum of one or two genera. The present study is the first wide exploration of rhizobia associated with two different P. vulgaris cultivars that are endemic in Venezuela. The results obtained showed that rhizobial diversity associated with P. vulgaris distributed in one country, Venezuela, was greater than in other countries, such as Spain, Brazil, Ecuador, and Mexico (Velázquez et al., 2005;Zurdo-Piñeiro et al., 2009;Ribeiro et al., 2013;Verástegui-Valdés et al., 2014).
In the present study, a larger number of Venezuelan isolates associated with Phaseolus were Rhizobium species with predominance related to R. phaseoli (Fig. 3A) than in other countries in which R. tropici or R. etli bv. phaseoli were predominantly associated with Phaseolus species (Martínez-Romero et al., 1991;Eardly et al., 1995;De Oliveira-L et al., 2015). Isolates classified as R. phaseoli in the present study were heterogeneous because they showed different stress tolerance (Crow et al., 1981;Wang et al., 2016;Mwenda, 2017). Previous studies reported that when R. etli bv. phaseoli populations were low, there were high numbers of rhizobia other than R. etli in bean nodules (Martinez-Romero, 2003;Zurdo-Piñeiro et al., 2009;Andrews and Andrews, 2017), which is consistent with the present results.
We detected two Mesorhizobium isolates from Guárico (Vertisol) with low nutrients, such as N and P, which was similar to the distribution of Mesorhizobium in soils with low nutrients in Brazil (Grange and Hungria, 2004). However, Mesorhizobium has been more commonly associated with Astragalus and Glycyrrhiza than P. vulgaris (Andrews and Andrews, 2017).
In the present study, the isolates classified as Bradyrhizobia in MLST were also capable of nodulating Phaseolus, which is supported by other findings on isolates from Mexican, Peruvian, and Brazilian soils (Barcellos et al., 2007;López-López et al., 2013;Matsubara and Zúñiga-Dávila, 2015). Since P. vulgaris from the tribe Phaseoleae is indigenous to the American continent (principally South and Central America), the nodulation of Bradyrhizobium species in P. vulgaris may be ascribed to the adaptation of legumes in Venezuela (Wang et al., 2016). It may also be due to adaptations from inoculants applied through the seeds in other legumes, such as soybean (Barcellos et al., 2007;Nandasena et al., 2007;López-López et al., 2013;Matsubara and Zúñiga-Dávila, 2015).
We also identified isolates that are closely related to Burkholderia with nodulating ability on P. vulgaris, which is consistent with previous findings on isolates from Moroccan and Brazilian soils (Talbi et al., 2010;Ferreira et al., 2012;Dall'Agnol et al., 2016); however, the predominant species of Burkholderia differed in the present study. Our results are consistent with previous findings on isolates from Latin American countries, such as Mexico, Venezuela, and Uruguay, in which Burkholderia was mainly associated with Mimosa species Lammel et al., 2013;Bontemps et al., 2016;Platero et al., 2016). In the present study, species of β-proteobacteria were observed in nutrient-deficient soils, such as Guárico, Lara, and Amazonas. Burkholderia is reportedly distributed to infertile acidic and low nutrients soils (Graham, 1992;De Oliveira-L et al., 2015;Bontemps et al., 2016;Platero et al., 2016), and this also appears to be the case in Venezuela.
Strong correlations were observed between soil types and rhizobia types; Rhizobium predominantly correlated with soil types such as Ultisol (Trujillo and Mérida), Alfisol (Miranda and D.C), Inceptisol (Aragua), and Oxisol (Amazonas). However, several isolates belonging to Burkholderia were also found in Mérida and Amazonas (Table 2 and S1). Ensifer strains were only found in Falcón with Aridisol. In contrast, highly diverse groups were detected in Guárico and Lara with Vertisol, such as Bradyrhizobium, Rhizobium, Mesorhizobium, and Burkholderia.
In the present study, all four strains classified as βproteobacteria in MLST had nodD genes closely related to Paraburkholderia (Fig. 4), suggesting that these genes in the four strains originated from P. phymatum. Our results on the nodD sequences and nodulation activity for βproteobacteria are similar to the findings of other studies on isolates from Brazilian forests and Moroccan soils, describing the symbiotic effectiveness of β-proteobacteria with common beans (Talbi et al., 2010;Lammel et al., 2013).
VGP2 and VGW2 from Guárico classified as Mesorhizobium based on MLST were closely related to B. japonicum based on the nodD gene sequence (Fig. 4), suggesting the horizontal gene transfer of symbiotic islands in these strains. The horizontal gene transfer of the nodD gene between Mesorhizobium and Bradyrhizobium has also been reported (Sullivan and Ronson, 1998;Grange and Hungria, 2004;Barcellos et al., 2007;Nandasena et al., 2007). Furthermore, the present results suggest the transfer of the nodD gene from Bradyrhizobium into Ensifer, Mesorhizobium, and R. pusense isolates, which may be from Bradyrhizobium inoculants applied through seeds in other legumes, such as soybean.
The present results also indicate the transfer of nifH genes from Bradyrhizobium into Rhizobium. nifH genes are regarded as markers of the efficiency of N 2 fixation. In the present study, strains with the nifH gene classified as Bradyrhizobium showed slightly low nodulation, and ARA resulted in a low biomass in plants, such as VAFP9 and VLaW3 with no detectable ARA (Table 2). VLaW4 classified as Burkholderia in MLST was grouped with Bradyrhizobium for the nifH gene, suggesting horizontal gene transfer between different proteobacteria groups. The effectiveness of nodulation may correlate with their origin, cultivation conditions, rhizobia type, and gene transfer, as previously reported (Vargas and Graham, 1988;Martínez-Romero et al., 1991;Hungria et al., 1993;Graham et al., 1994;Eardly et al., 1995;Chidebe et al., 2018).

Physiological characteristics and stress tolerance of Phaseolus rhizobia
The abilities of tropical rhizobia other than N 2 fixation currently remain unclear; therefore, the present results add new and essential information on the characteristics of rhizobia distributed in tropical areas. Most of the strains isolated in the present study not only survived, they also adapted well to abiotic stress conditions, such as acidic soils, high temperatures, and high salinity. Previous studies support this result, showing that the most tolerant isolates were α-proteobacteria and their tolerance depended on their soil of origin (Graham et al., 1994;Martinez-Romero, 2003;Marquina et al., 2011;Mhamdi et al., 2015). The present results indicate that Venezuela possesses a great diversity of rhizobia that may be beneficial for agriculture under stress conditions.
In the present study, the strains isolated from soils with Al or acidic pH showed higher tolerance to Al-acidic conditions (Table S1), suggesting that the tolerance of isolates is associated with soil origin and cultivation history, as reported by Piña and Cervantes (1996). Low soil pH is often attributed to Al and Mn toxicity and Ca deficiency (Graham et al., 1981;Graham et al., 1994), and to improve soil conditions, calcium carbonate was added before and during Phaseolus cultivation for the traditional cropping system (Casanova, 2005) in some sampling sites in Venezuela, such as Trujillo. Many studies describe Rhizobium species showing resistance to acidic pH (e.g., R. tropici) (Muglia et al., 2007;Marquina et al., 2011), while Rhizobium species sensitive to low pH have also been reported, including R. meliloti (Tiwari et al., 1992). Acidic pH tolerance in rhizobia depends on the maintenance of intracellular pH (Graham et al., 1994).
Furthermore, most of our isolates were tolerant to alkaline pH, which is supported by previous findings showing that R. phaseoli strains were more competitive than other Rhizobium species under alkaline conditions (Shamseldin and Werner, 2004;Verástegui-Valdés et al., 2014).
Additionally, antifungal activity results (Table 2) suggest that strains have the ability to promote plant health based on their resistance to pathogens. Several isolated strains did not show good potential as inoculants due to low IAA, P, and K; however, these rhizobia exhibited strong antifungal activity against different fungi, which may suppress the disease activity of soil-borne pathogens and reduce disease severity.
VGP6 and VGP9 were genetically unrelated to any reference strains, suggesting that these isolates belong to the Bradyrhizobium group. According to the MLST, nodD, and nifH sequence analyses, these may be novel species or lineages that are fast or intermediate growers. VGP6 and VGP9 may improve biomass, show tolerance to different abiotic stress conditions, and exhibit antifungal activity.
Stress tolerance and physiological abilities in rhizobia isolated from specific areas may be attributed to wild Phaseolus rhizobia. Wild common beans have been reported from Mérida, Portuguesa, Táchira, and Trujillo in the western Andes of Venezuela and some Lara areas (Debouck et al., 1993). Diverse rhizobia with tolerance to stress in these areas may be associated with the native wild legumes.
More factors affected the ability of rhizobia to promote plant growth than N 2 fixation. Our plant assay was performed at a neutral pH and under controlled temperature, light, and humidity; however, in fields with various stressors, isolated rhizobia with stress tolerance may help plants to grow better. Further studies are needed to confirm the effectiveness of strains as inoculants for Phaseolus in field conditions with acidic soils, high Al concentrations, low nutrient supply, or the presence of pathogens.