Systematic and Applied Microbiology 34 (2011) 376–384

Contents lists available at ScienceDirect

Systematic and Applied Microbiology

journal homepage: www.elsevier.de/syapm

Genetic diversity of symbiotic elkanii populations recovered from inoculated and non-inoculated Acacia mangium field trials in Brazil

M.M. Perrineau a, C. Le Roux a, S.M. de Faria b, F. de Carvalho Balieiro c, A. Galiana a, Y. Prin a, G. Béna d,e,∗ a CIRAD, Laboratoire des Symbioses Tropicales & Méditerranéennes, Montpellier, France b Embrapa Agrobiologia, Seropédica 23890-000, RJ, Brazil c Embrapa Solos, Jardim Botânico, Rio de Janeiro, RJ, Brazil d IRD, Laboratoire des Symbioses Tropicales & Méditerranéennes, Montpellier, France e Laboratoire de Microbiologie et de Biologie Moléculaire, University Mohammed V Agdal, Morocco article info abstract

Article history: Acacia mangium is a legume tree native to Australasia. Since the eighties, it has been introduced into Received 20 January 2011 many tropical countries, especially in a context of industrial plantations. Many field trials have been set up to test the effects of controlled inoculation with selected symbiotic versus natural coloniza- Keywords: tion with indigenous strains. In the introduction areas, A. mangium trees spontaneously nodulate with Nitrogen fixation local and often ineffective bacteria. When inoculated, the persistence of inoculants and possible genetic Controlled inoculation recombination with local strains remain to be explored. The aim of this study was to describe the genetic Molecular systematics diversity of bacteria spontaneously nodulating A. mangium in Brazil and to evaluate the persistence of Recombination Acacia mangium selected strains used as inoculants. Three different sites, several hundred kilometers apart, were studied, Bradyrhizobium with inoculated and non-inoculated plots in two of them. Seventy-nine strains were isolated from nod- Brazil ules and sequenced on three housekeeping genes (glnII, dnaK and recA) and one symbiotic gene (nodA). All but one of the strains belonged to the Bradyrhizobium elkanii species. A single case of housekeeping gene transfer was detected among the 79 strains, suggesting an extremely low rate of recombination within B. elkanii, whereas the nodulation gene nodA was found to be frequently transferred. The fate of the inoculant strains varied depending on the site, with a complete disappearance in one case, and persistence in another. We compared our results with the sister species , both in terms of population genetics and inoculant strain destiny. © 2011 Elsevier GmbH. All rights reserved.

1. Introduction ria collectively known as . It has been found associated with various Bradyrhizobium strains in different countries and con- Acacia mangium is a leguminous tree native to northern tinents [23,50,55,66], and recently Le Roux et al. [22] suggested Australia, Papua New Guinea and Indonesia. It is naturally found that A. mangium is preferentially nodulated by Bradyrhizobium in tropical rainforests with a mean rainfall ranging from 1500 to elkanii species. It has also been found associated, albeit much 3000 mm per year. A. mangium has been planted widely in the trop- less frequently, with other species, such as Rhizobium sp. [14,65], ics since the early 1980s, first in Indonesia and Malaysia for pulp Mesorhizobium sp. [11], and Ochrobactrum sp. [39]. In South Amer- production, but also for revegetation and rehabilitation purposes. ica, A. mangium has been introduced in several countries (Brazil, Its main qualities are rapid early growth, good wood quality and Colombia, Costa Rica, etc.), and two previous studies identified its tolerance to different soil pH and composition. In 1998, Turn- some isolates nodulating A. mangium in Brazil, such as Bradyrhi- bull et al. [60] estimated the area planted to A. mangium in Asia zobium sp. [35,36]. at 600,000 ha. A. mangium belongs to the Mimosoideae tribe, and Inoculating plants with efficient nitrogen-fixing bacteria is an has the ability to form nitrogen fixing symbiosis with soil bacte- environment-friendly way of improving the economic and nat- ural growth of legumes, especially in poor or degraded soils. In A. mangium, regardless of the symbiotic rhizobial selected, inoc- ∗ ulation improves survival in the field by 10% [28]. Galiana et al. Corresponding author at: LMBM, Faculté des Sciences, Av Ibn Batouta BP 1014, Université Mohammed V - Agdal - Rabat, Morocco. [15] reviewed several inoculation field trials in different countries. E-mail address: [email protected] (G. Béna). They showed that inoculation had a positive effect on A. mangium

0723-2020/$ – see front matter © 2011 Elsevier GmbH. All rights reserved. doi:10.1016/j.syapm.2011.03.003 M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384 377

Table 1 Sites sampled.

Geographical origin Latitude Longitude Use of inoculant strain Type of soil Age of trees Number of trees Number of isolates sampled

01◦41S56◦24W No Subsoil 1–2 years 8 16 Porto Trombetas ◦  ◦  01 41 S5625 W Yesa Tailings tank 1–2 years 10 18 23◦02S48◦38W No Experimental field 11 years 6 10 Itatinga 23◦02S48◦38W Yesa Experimental field 6 years 10 22 Seropédica 22◦45S43◦40W No Experimental field 15 years 9 13

a Inoculated by Bradyrhizobium strains BR3609 and BR6009. growth, whatever the soil type and the country, and improvement seedlings had not been inoculated prior to planting. At two, inoc- was still detectable after 2–3 years. In addition, the strains did not ulation had been carried out in the greenhouse with two selected all display the same effect at the same stage of growth, from the Bradyrhizobium strains, and seedlings were planted in nearby fields. nursery or greenhouse to the field [14]. One main question regard- The genetic diversity of all strains sampled from the fields (both ing legume inoculation, either directly in the field, or in the nursery inoculated and non-inoculated) was analysed for three housekeep- prior to planting/sowing, is not only the survival and persistence of ing genes, recA, dnaK and glnII, selected for their phylogenetic these inoculant strains, but also the possible genetic recombination congruence in the Bradyrhizobium genus [42] and one symbiotic between inoculant and indigenous strains. In the case of perennial nodulation gene, nodA, to test for possible horizontal transfer of the species, the replacement of inoculant strains by local strains over symbiotic island within and between local and inoculated bradyrhi- the life span of the tree is also possible. In two A. mangium field zobia. trials in Ivory Coast in 1994, Galiana et al. [16] recovered an inocu- lated strain 42 months after field transplantation, whereas another 2. Materials and methods one disappeared after 19 months. Martin-Laurent et al. [29] and Prin et al. [49] analysed nodules from A. mangium 4 and 6 months 2.1. Site descriptions after transfer to the field in Singapore and Madagascar respectively. Molecular analyses showed the persistence and the overgrowth of Samples of A. mangium nodules were directly taken from 5 one of the two inoculums to the detriment of the other. In 1990, fields located at three sites in Brazil (Table 1). At Itatinga, we took McLoughlin et al. [30] showed that the success of inoculating soy- samples from two experimental fields (2 km apart), each with a bean with Bradyrhizobium japonicum strains in the US was highly monoculture of A. mangium surrounded by Eucalyptus species. In variable, and that successful inoculation the first year did not ensure one field, A. mangium seedlings had been planted 11 years earlier that the inoculated strain would persist in forming nodules in sub- without any inoculation. In the other one, seedlings had been inoc- sequent years. More recently, Mendes et al. [31] showed that even ulated 6 years earlier in the nursery with the two rhizobia strains 6 years after its inoculation on in Brazil, the B. japonicum BR3609 (SEMIA6387) and BR6009 (SEMIA6404). These two strains serogroup introduced was dominating nodules, occurring in more were originally isolated in Porto Trombetas, from Acacia auriculi- than 50% of the nodules from treatments where it had never been formis (Mimosoideae) and Lonchocarpus costatus (Papilionoideae) inoculated. respectively [12], and were recommended by Embrapa Agrobiolo- Beyond the question of the persistence of a given inoculant gia as efficient with A. mangium. These strains were subsequently strain in soil, its genomic stability and integrity, i.e. whether the described as belonging to the B. elkanii species [4,32]. strain introduced remains genetically isolated in soil, or whether The second site, Porto Trombetas, is exploited as a bauxite mine gene exchanges occur with indigenous strains, is also relevant. One in the State of Pará. Two fields were sampled around the indus- famous study [25,57,58] showed that a Rhizobium loti strain vir- trial site: the first was planted with A. mangium in a rehabilitated tually disappeared 7 years after inoculation, but that its 500 kb bauxite tailings tank (SP-1). This rehabilitation process was per- chromosomal symbiosis island had been transferred to local non- formed in a two-step procedure. In 1998, seeds of leguminous trees symbiotic rhizobia. Nandasena et al. [37,38] also showed that and shrubs, pre-inoculated with N-fixing bacteria and mycorrhizae endemic Australian Mesorhizobium sp. isolates received symbi- (VAM fungi), were hydroseeded (i.e. seeds sown by distribution in otic genes carried on a mobile symbiosis island from inoculant a stream of water propelled through a hose) over the tailings with strains through lateral transfer 5–6 years after inoculation with a the addition of fertilizer [13]. In the second phase in 1999, after Mesorhizobium ciceri strain. Barcellos et al. [2] demonstrated hor- substrate consolidation, A. mangium seedlings were planted after izontal transfer of symbiotic genes from a B. japonicum parental inoculation in the nursery using the two selected strains BR3609 strain to local Ensifer fredii and B. elkanii. The indigenous B. elka- and BR6009. In the second field (1.8 km away from the first), spon- nii strain acquired at least one copy of the nodulation gene nodC, taneous colonization of A. mangium occurred on subsoil, through and the indigenous E. fredii strain received the whole symbi- seed migration from a neighbouring A. mangium reforested area. otic island and maintained an extra copy of its original nifH In both field trials, nodules were collected from 1 to 2-year-old gene. plants. These trees were second generation trees in the inoculated In the context of A. mangium inoculation successes and out- plot. comes in Brazil, the first goal of this study was to analyse and The last site was at Seropédica, a 15 year-old experimental field compare the genetic diversity of bacteria spontaneously nodulat- trial planted with non-inoculated A. mangium trees. ing A. mangium at different sites in Brazil. Our second goal was We randomly sampled from 6 to 10 individual trees in each field, to estimate the persistence and the genetic stability of introduced collecting as many nodules as possible from each (Table 1). selected Bradyrhizobium strains in Brazilian soils, both by search- ing for the original inoculated strains (when applied), but also 2.2. Isolation of bacterial isolates highlighting possible genetic recombination events between local and introduced strains. For this, we studied five bacterial popula- Two nodules per sampled tree, chosen at random, were surface tions nodulating A. mangium, distributed over three sites in Brazil: sterilised in 35% hydrogen peroxide for 4–6 min depending on the Porto Trombetas, Itatinga and Seropédica. At three sites, A. mangium size of the nodules, rinsed and then crushed in sterile water. The 378 M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384

nodule contents were spread onto YMA Petri dishes [62] supple- Table 2 mented with bromothymol blue as pH indicator. After purification Diversity indices for each locus analysed. by several subculturings of isolated colonies, isolates were grown Locus nRL S − ◦ and stored at 80 C in YEM broth supplemented with 20% mineral recA 80 21 540 119 0.03499 oil to reduce bacterial proliferation. dnaK 80 19 662 88 0.02565 glnII 80 19 590 111 0.03859 2.3. Plant Renodulation Assay (Koch’s postulate) nodA 80 25 487 244 0.19183 n: number of sequences obtained; R: number of different alleles; L: length of the All isolates were tested to confirm their capacity to nodulate A. sequence; S: number of polymorphic nucleotide sites; : nucleotide diversity. mangium in individual monoxenic culture tubes [61]. A. mangium seeds (lot 19297, CSIRO Australian Tree Seed Centre) were dipped and subjected to methods implemented in RDP3 [19], similarly to into boiling water for 2 min, surface-sterilised for 5 min in 35% the previous search for intragenic recombination. Secondly, incon- hydrogen peroxide, rinsed in sterile distilled water for 2 min and gruences between phylogenetic trees obtained from each locus germinated on 10 g l−1 (w/v) water agar Petri dishes for 3 days ◦ separately were tested with the Shimodeira–Hasegawa test imple- at 28 C in the dark. Plants were transferred in CYG Germination mented in PAUP4 [59]. Maximum likelihood trees were obtained Pouches (Mega International, United States) to sterile individual for each locus separately using PAUP4. The best molecular evo- tubes containing 10 ml of a nitrogen-free solution [6]. After a few lution model for each dataset was chosen using Modeltest [47], ␮ days, healthy plants were inoculated with 500 l of a five-day- following the log ratio test. One hundred bootstrap replicates old YEM liquid bacterial culture. Nodulation results were visually were used to estimate node reliability. We also included several checked after 30 days. sequences obtained from GenBank in the phylogeny. We included the best hit for each different allele blasted on the databank, 2.4. DNA extraction, amplification and sequencing but also sequences of each described Bradyrhizobium species and genospecies, whenever available. We also reconstructed a phy- Total genomic DNA of each isolate was extracted as described logeny with the concatenated dataset of the three core genes (recA, in [10]. Partial dnaK (heat shock chaperone protein) and glnII dnaK and glnII). This analysis also included sequences retrieved (glutamine synthetase II) genes were amplified using primers from four full genomes already sequenced, Bradyrhizobium japon- TSdnaK4, TsdnaK2, TSglnIIf and TSglnIIr as described by [56]. icum USDA110, Bradyrhizobium sp. ORS278, Bradyrhizobium sp. The recA (DNA recombination protein) gene fragment was ampli-   Btai1 and the closely related species Rhodopseudomonas palustris, fied by primers TS2recAf (5 -GCCCTGCGTATCGTCGAAGG-3 ) and strain CGA009. We also included several species that had been TS2recAr (5-CGGATCTGGTTGATGAAGATCACC-3), with an anneal- ◦ typed for the same three loci. We thus included sequences from ing temperature of 59 C, modified from [56]. The partial nodA B. elkanii, B. canariense, B. yuanmingense and B. liaoningense in the (acyltransferase) gene was amplified for all strains, using primers phylogenetic tree. nodA1F or nodAf.brad and primer nodAr.brad from [8,9]. PCR amplification was performed with a Thermo Electron Px2 2.6. Nucleotide sequence accession numbers Thermal Cycler in a 25 ␮l total volume. The reaction mixture contained each deoxynucleotide triphosphate (100 ␮M), primers The sequence accession numbers for the four gene frag- (0.4 ␮M each), MgCl (1.5 mM), 0.5 U of Taq DNA polymerase 2 ments of the two inoculant strains and 79 isolates are listed in (Promega, Charbonières, France), the buffer supplied with the supplementary material (Table S1). Sequence alignments are avail- enzyme and 50 ng of genomic DNA. The PCR protocol was as fol- ◦ able upon request. lows: initial denaturation (94 C for 5 min); 35 denaturation cycles (30sat94◦C), annealing (30 s at primer specific hybridization tem- perature), and extension (40 s at 72 ◦C), and final extension (3 mn at 3. Results 72 ◦C). Amplification products were visualized on 1% agarose TAE gel stained with ethidium bromide. PCR fragments were sequenced 3.1. Genetic diversity of isolates and recovery of inoculant strains by subcontracting to Genoscreen, France, using ABI chemistry on ABI3730 sequencers. From 10 to 22 isolates were obtained from each site (Table 1). We obtained a total number of 79 isolates renodulating their orig- 2.5. Nucleotide sequence analyses inal host, A. mangium, confirming their symbiotic status. Although not firmly assessed using statistical tests, all plant–isolate combina- Multiple nucleotide sequence alignments for the four loci were tions were visually checked as being effective for nitrogen fixation performed with ClustalX [21] and manually corrected using Gene- (plants were always greener and bigger than the non-noculated doc [40]. A single sequence data file, including one sequence of each negative control). The 81 strains in the study (79 isolates + the two different allele obtained, was built using Mothur [53]. The number BR3609 and BR6009 inoculant strains) were all sequenced for the of sequences (n), number of different alleles (R), length of sequence four loci (no missing data). (L), number of polymorphic sites (S) and nucleotide diversity () were computed at each locus using DnaSP [24]. 3.2. Nucleotide diversity Each locus was analysed for intragenic recombination. We explored this issue using several alternative methods implemented The sequence lengths obtained ranged from 487 (nodA)to in RDP3 [19], using RDP [26], Chimaera [48], Geneconv [44], MaxChi 662 bp (dnaK)(Table 2). The number of different alleles was rather [54], BOOTSCAN/RECSCAN [27] and 3Seq [5]. A first analysis run was similar between loci, from 19 (dnaK and glnII)to25(nodA), the latter performed using each alignment individually. Only recombination being the most polymorphic with the shortest sequences. These dif- events for which at least two methods gave a positive result were ferences were reflected in the nucleotide diversity index () which kept for further analyses (using phylogenetic tree comparisons) to was 7.5 higher for nodA than for dnaK. Of the four alignment matri- firmly assess their accuracy (as per [64]). ces obtained, only nodA displayed indels, one 3 bp and one 6 bp long. Recombination between loci was tested taking two approaches. All three other loci lacked indels in the alignment obtained, even Firstly, a concatenated data file including all four loci was created when we included other sequences from the databank. M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384 379

100 Bradyrhizobiumsp. genosp. VIII B. denitrificansLMG8 Bradyrhizobiumsp. genosp. VI B. canarienseBTA-1 T B. japonicumUSDA110 B. betae LMG21987T B. japonicum USDA6T Bradyrhizobiumsp. genosp. IV B. liaoningense LMG18230T BR10109 It B. yuanmingenseCCBAU10071 T Bradyrhizobiumsp. genosp. V BR3706 Ser 83 BR3652 PT 91 100 BR3688 PT BR3657 PT Bradyrhizobiumsp. genosp. IX Bradyrhizobiumsp. genosp. X Bradyrhizobiumsp. genosp. VII BR3609 BR3716 Ser B. elkanii LMG6134T BR10116 It BR6009 90 BR3656 PT BR3672 PT 83 BR3687 PT (BR3653 PT, BR3669 PT, BR3684 PT) BR3709 PT BR3671 PT 99 BR10113 It BR10115 It (BR10114 It, BR3750 It) 90 BR3679 Ser BR3676 Ser BR3648 PT Bradyrhizobiumsp. genosp. XI BR3640 PT (BR3642 PT, BR3644 PT) BR3740 PT 0.1

Fig. 1. Maximum likelihood recA phylogeny. The two strains used as inoculants are in bold. One sequence per different allele was included in the dataset. Sequences from other species or genospecies were retrieved from GenBank. Bootstrap values were estimated based on 100 replicates. Only values higher than 75% have been kept in the figure. Isolates shaded in grey were sampled from non-inoculated plots. Isolates in brackets shared the same allele but with a different multilocus haplotype. (PT: Porto Trombetas, It: Itatinga, Ser: Seropédica).

Combination of the four loci resulted in 28 different geno- ers, involving a recA transfer. The two strains BR3648 and BR3656 types (Table S1). Fifteen were only represented by a single isolate, shared the same alleles for both dnaK and glnII but their recA allele whereas others were represented by 2–8 isolates. The main geno- differed for 22 different nucleotide sites, resulting in a completely type in terms of frequency was haplotype 1 (identical to strain different phylogenetic position (see Fig. 1). When including nodA BR6009, 20 isolates) which was used for inoculation at Itatinga. Of in the analyses, several examples of lateral transfer were revealed. the two strains used for inoculation in the nursery, only the latter Eleven nodA alleles clustered in the same well-supported clade, was recovered from our samples, at high frequency in the inocu- whereas they were dispersed in several clades based on the three lated plot in Itatinga (82%, 18 out of 22 isolates), but also twice from other loci. The nodulation locus, and possibly the cluster of genes the non-inoculated plot at the same site (2 out of 10 isolates). This involved in nodulation specificity, had undergone a much higher strain was never recovered from the inoculated field trial at Porto rate of recombination than the other loci, even though nodulation Trombetas, and strain BR3609 was never detected at all among the genes have always been found located on the chromosome in the 79 isolates. Bradyrizhobium strains [18].

3.3. Recombination and lateral gene transfer 3.4. Phylogenetic analyses

None of the four loci, taken individually, revealed intragenic The four alignments were used separately to reconstruct max- recombination (data not shown). We first combined the three imum likelihood phylogenies (see recA and nodA in Figs. 1 and 2 housekeeping genes loci in a single alignment to detect recom- respectively, dnaK and glnII in supplementary material Figs. S1 and bination between loci. Only a single unambiguous recombination S2). The three core gene phylogenies clustered all strains but one could be detected between one of the three core genes and oth- (BR10109) within the B. elkanii clade, although not with high boot- 380 M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384

Methylobacterium nodulans ORS2060 Burkholderia tuberum STM678 BR3709 PT Bradyrhizobiumsp. CH2493 (Lupinus paraguariensis, Brazil) 100 BR3669 PT 100 BR3609 BR10116 It BR6009 BR3671 PT (BR3652 PT) BR10113 It BR10115 It 70 BR3750 It BR10114 It 90 Bradyrhizobiumsp. CH2509 (Lupinus albescens, Brazil) Bradyrhizobiumsp Mm1-3 (Machaerium milleflorum, Panama) 78 BR10109 It Bradyrhizobiumsp.ORS285 78 94 Bradyhizobiumsp. ANU289 100 B. elkanii USDA76T B. elkanii USDA61 95 97 B. yuanmingense CCBAU10071T 100 B. japonicum USDA6T B. japonicum USDA110 80 Bradyrhizobiumsp. CPAC-M9 (Mucuna aterrima, Brazil) 100 Bradyrhizobiumsp. ORS130 (Acacia albida, Senegal) BR3676 Ser BR3706 Ser BR3648 PT BR3653 PT BR3640 PT BR3679 Ser 91 BR3642 PT BR3740 PT BR3672 PT (BR3656 PT) BR3687 PT BR3716 Ser BR3644 PT BR3657 PT BR3688 PT (BR3684 PT) 0.1

Fig. 2. Maximum likelihood nodA phylogeny. Sequences from various Bradyrhizobium species were included in the dataset. One sequence obtained in this study per distinct clade was used in a blast search to retrieve the closest related sequence in the databank. Host legume species and country of origin are indicated for GenBank sequences. Isolates shaded in grey were sampled from non-inoculated plots. Isolates in brackets shared the same allele but with a different multilocus haplotype. (PT: Porto Trombetas, It: Itatinga, Ser: Seropédica). strap values. Depending on the loci and sequences that could be by all three datasets. All single-gene phylogenies obtained with the retrieved from GenBank, isolates appeared to be related to various three core genes were significantly incongruent one against the B. elkanii strains (USDA76, USDA94, LMG6147), several Bradyrhizo- other, and rejected based on the SH test. bium genospecies (IX, X, VII and XI), or the recently described B. However, as shown in Vinuesa et al. [64], the mean and median pachyrhizi species [51]. The strain BR10109 appeared to be related congruence levels of trees increase with the number of concate- to B. yuanmingense/B. liaoningense in the two glnII and recA trees, nated partitions used to infer them. We thus concatenated the three but not in the dnaK tree, where it was close to B. betae. core genome loci to build up a single matrix and reconstructed a The nodA phylogeny clustered strains in three different clades. maximum likelihood tree (Fig. 3). The glnII tree was congruent (not Strain BR10109 was strongly supported in the same clade as the rejected) with the concatenated dataset, whereas both the recA and photosynthetic ORS285 strain. However, this may have been an dnaK trees were significantly rejected (p = 0.002 each). Neverthe- effect of a Long Branch Attraction artefact. Fourteen haplotypes less, the concatenated tree obtained strongly clustered all strains fell in a single clade together (and 100% bootstrap value) with two but one in the B. elkanii clade, confirming the preponderance of that strains isolated from Acacia albida in Senegal (ORS130) and Mucuna species in our samples. The clades also usually grouped together aterrima in Brazil (CPAC-M9). This clade appeared as the sister clade of a B. japonicum/B. yuanmingense clade. It is worth noting that none of the strains fell in the same clade as a nodA sequence retrieved Table 3 Shimodeira–Hasegawa tests on the four loci and the core gene sequence concate- from any previously described B. elkanii strain. The third group of nated tree. nodA haplotypes clustered in a clade with no described species, with only one sequence retrieved from GenBank, amplified from Locus recA dnaK glnII Conc Bradyrhizobium sp. CH2493, and one strain isolated from Lupinus nodA 0.000 0.000 0.000 0.000 paraguariensis, also in Brazil. recA 0.000 0.000 0.002 dnaK 0.001 0.002 The Shimodeira–Hasegawa (SH) test implemented in PAUP was glnII 0.391 performed on all four trees and datasets (Table 3). We only included phylogenies with “our” strains from this study since none of the SH test using RELL bootstrap (one-tailed test); number of bootstrap repli- cates = 1000. The P value indicates the significance of the difference in −ln L scores “reference” strains from the databank was common to all of the (Diff −ln L) for each locus compared individually or with the concatenated (Conc) four datasets. As expected, the nodA tree was significantly rejected tree. M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384 381

Rhodopseudomonas palustris CGA009 100 Bradyrhizobium sp. BtAi1 Bradyrhizobium sp. ORS278 BR10109 It 100 99 B. yuanmingense SEMIA 6319 B. liaoningense LMG18230T B. japonicum USDA110 77 B. canariense SEMIA 928 BR3716 Ser 73 BR6009 97 BR10116 It B. elkanii USDA76T 100 BR3657 PT BR3652 PT BR3688 PT BR3609 BR3669 PT 100 BR3709 PT BR3653 PT BR3687 PT BR3684 PT BR3671 PT BR3706 Ser 92 BR3648 PT BR3656 PT 95 BR3672 PT BR10113 It 100 BR10114 It 100 BR3750 It BR10115 It 100 BR3676 Ser BR3679 Ser BR3640 PT 100 BR3642 PT 100 BR3740 PT 0.1 BR3644 PT

Fig. 3. Maximum likelihood phylogeny based on a concatenated dataset including the three loci recA, dnaK and glnII. Each different multilocus haplotype obtained in the study was represented in the dataset. The coding following isolate names gives the site of origin (PT: Porto Trombetas, It: Itatinga, Ser: Seropédica). The two strains in bold were used as inoculants. The external data were taken either from full genome sequences (R. palustris, Btai1, ORS278 and USDA110) or from a combination of several single sequences. strains sampled at same site, suggesting that point mutations were characterized 26 strains trapped by A. mangium in Indonesia, of the main force driving diversification locally, although migration which 12 were identified as B. elkanii and 13 as B. japonicum, the last also occurred, since the sites were represented throughout the phy- strain isolated being related to Mesorhizobium loti. Fremont et al. logeny. [14] also isolated both B. japonicum and B. elkanii strains after trap- ping with A. mangium in Malaysia. Nuswantara et al. [41] isolated 4. Discussion 10 strains from Indonesia, all allegedly belonging to B. elkanii. The lack (or almost) of B. japonicum strains in our samples may, 4.1. Diversity of Bradyrhizobium strains associated with A. however, have resulted from its scarcity in these soils. Parker and mangium in Brazil Kennedy [45] showed that the distributions of the two species B. elkanii and B. japonicum in the northeastern United States, after A. mangium was introduced into Brazil <20 years ago, its area trapping with 13 native legume species, were not similar. How- of origin being confined to northern Queensland in Australia and ever, several studies have demonstrated that Brazilian soil naturally Papua New Guinea. At the 3 study sites, seedlings were introduced harbours B. japonicum strains [36]. These results (and others) rein- without any artificial inoculation, forcing A. mangium to interact forced the question of the virtual absence of B. japonicum in our with local strains. A. mangium is usually described as a promis- samples. This unbalanced ratio between the two species is all the cuous nodulating species, able to associate with a wide range of more surprising in that at one site hydroseeding provided a large bacteria. Indeed several different genera have been isolated from cocktail of strains, including B. japonicum (de Faria, pers. comm.). A. mangium, including B. japonicum, B. elkanii, Rhizobium, Mesorhi- Moreover, at least one strain (SEMIA 6420) included in that cocktail, zobium and Ochrobactrum. In our study, 39 out of the 40 isolates originally isolated from A. mangium in Brazil, was described as B. obtained from the non-inoculated field trials were clustered in japonicum based on the 16S sequence [33]. B. japonicum might have the B. elkanii clade. The 22 strains sampled from inoculated fields been eliminated from this soil due to environmental constraints, (excluding original inoculant strains) were all classified as B. elka- such as the pH or organic matter content. For instance, Giongo et al. nii. Le Roux et al. [22] previously suggested such preference, which [17] showed that these parameters significantly affected the level is not strict symbiotic specificity, however. Indeed, Clapp et al. [11] of microbial diversity in areas that had been used as commercial 382 M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384 crop fields. Whatever the explanation, B. elkanii appears to be bet- acquired symbiotic genes from the inoculants. Inoculation was car- ter adapted to A. mangium in Brazil, at least in terms of survival and ried out in 1999 and 2003 at the Porto Trombetas and Itatinga nodulation, and should be chosen as a priority for inoculation. sites respectively, theoretically leaving enough time for recombi- nation and gene transfer to take place. Several studies have shown 4.2. Genic diversity, recombination and persistence of the that, outside its native area, A. mangium inoculated with native inoculum Australian Bradyrhizobium strains generally gives excellent growth responses compared to the local spontaneous bacterial strains [15], Bradyrhizobium is a complex and blurred genus that has been the although A. mangium promiscuity results in non-inoculated plants subject of many taxonomic and phylogenetic studies [52]. Several being extremely easily colonized by local bacteria in field trials [49]. have focused on phylogenetic congruence between loci on a species In our case, the local strains might have been as competitive and or genus level, estimating the rate of lateral gene transfer over efficient as the inoculant strains, at least in terms of nodulation, the diversification of the various species and genospecies. Nzoué thereby not favouring symbiotic gene acquisition by local strains. et al. [42], Parker and Kennedy [45] or Batista et al. [3] showed that recombination and gene transfer between and within Bradyrhi- 4.3. Artificial inoculation and persistence of the inoculum zobium lineages occurred frequently, to the extent of resulting in statistically incongruent phylogenies. However, all these studies At two of the three sites tested, A. mangium seedlings were highlighted that the results were dependent upon the loci stud- inoculated in the nursery prior to their introduction. We obtained ied, but also the lineage. Vinuesa et al. [63] suggested frequent contrasting results on the persistence of those inoculant strains homologous recombination within but not across lineages, and in the soil, with a high frequency of re-isolation of one inoculant detected variable levels of recombination depending on the lineage strain at Itatinga, but a complete lack at Porto Trombetas. Two major [64]. Parker and Kennedy [45] estimated that lateral gene trans- parameters differed between the two sites. Firstly, at Porto Trombe- fer altered genealogical relationships of different portions of the tas, the stressful environmental conditions (a bauxite tailing tank), genome, with gene transfer between divergent taxa of Bradyrhizo- coupled with the initial hydroseeding including a combination of bium, from B. elkanii to the B. japonicum group. Minamisawa et al. various legume species and rhizobial strains, created a highly com- [34] showed horizontal gene transfer between B. elkanii and B. petitive environment, where introduced strains would appear to japonicum strains in soil, suggesting potential horizontal transfer have been eliminated. Secondly, we took samples directly from the of nod genes between bradyrhizobia and other bacterial popula- inoculated trees in the Itatinga field, whereas at Porto Trombetas tions in the soil environment. In our study, we only detected one we collected nodules from 1 to 2-year-old second generation trees. case of gene recombination between strains, leading to the replace- The latter trees must have reacquired their symbiotic bacteria, since ment of the recA copy in strain BR3648 by another copy from an there was no vertical transmission in the rhizobial symbiotic associ- (unknown) strain. Within-species recombination thus appeared ation, thus going through the competition between strains present to be extremely limited between these strains. Moreover, none in the soils. Nonetheless, this is possible, since we isolated the inoc- of the B. elkanii strains harboured any B. japonicum housekeep- ulant strain twice in nodules sampled in the non-inoculated field ing sequences, suggesting a lack of gene transfer between species. trial at Itatinga, proving that the inoculated strain was able, after Sampling at various sites might also have influenced our results short-distance migration, to settle in a new environment and col- in creating an artificial significant Linkage Desequilibrium value, onize the roots of local A. mangium. The same “invasion” was also due to the admixture of several genetically isolated populations observed in Madagascar [49]. (i.e. equivalent to a Wahlund effect). However, even within each The question of the persistence of the inoculant in soil is crucial population, the LD was maximum, supporting the lack of recombi- for both ecological and agronomic reasons. Whether the introduced nation (data not shown). The fact that we focused on the B. elkanii strain remains in the soil, and whether it has an impact on the species, for which no similar study is yet available, makes it diffi- local indigenous microbial population are important questions that cult to compare our results with others. Our data showed however have been addressed several times for soybean. Although Brutti that B. elkanii would appear to be less subjected to recombination et al. [7] confirmed that B. japonicum strains could be recovered than B. japonicum. The same results were observed in the Sinorhi- 5 years after introduction in an arable soil in Argentina, Obaton zobium (Ensifer) genus, where Bailly et al. [1] suggested a different et al. [43] showed that this stability could be extremely variable rate of recombination between the two sister species S.meliloti and between sites, suggesting that the problem of competition could S. medicae. be solved by repeated inoculation. Loureiro et al. [25] showed that Although only a single lateral core gene transfer event could the introduction of bradyrhizobial strains in a soil that was origi- be detected in our samples, nodulation gene transfer conversely nally devoid of nodulating and effective rhizobia had an impact on occurred extremely frequently. Nodulation genes are known to lie local rhizobial diversity, possibly through acquisition of the ability behind symbiotic specificity, and several studies have shown that to nodulate soybean for local strains. The influence of environmen- nod gene transfers are frequent between rhizobial lineages, even tal factors was also highlighted by Hungria and Vargas [20] in terms between alpha and beta [9]. Whilst nearly all our of nodulation limitation, but also in the selection of stress tolerant strains clustered in the same monophyletic group (in the three core inoculants strains. Our results, even though based on B. elkanii when genes phylogeny), nodA alleles clustered in two divergent clades all other studies have focused on B. japonicum, confirmed that the (plus one single clade allele). It was striking to find that none of the stability and persistence of an inoculated strain is unpredictable. A nodA sequences fell closely related to the two B. elkanii strains we combination of many factors, including competition for both nodu- included in the phylogeny, even though they were closely related lation of hosts and adaptation to environmental factors, needs to to this species in terms of core gene phylogeny. Nodulation gene be considered to estimate the survival of strains in soil. transfers have been well described in the Mesorhizobium model [57,58]. Parker et al. [46] showed nitrogenase nifD gene transfer between Bradyrhizobium lineages, and Barcellos et al. [2] showed Acknowledgements that symbiotic genes were horizontally transferred from a B. japon- icum inoculant strain to indigenous E. fredii and B. elkanii. This work was supported by CIRAD (PhD grant to MM. Per- At the two sites where A. mangium seedlings were introduced rineau.). This work was funded by the French Ministry of the after nursery inoculation, none of the local strains recovered Environment and Sustainable Development (Ecofor/Ecosystèmes M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384 383

Tropicaux Programme) and supported by CNPq Project number [20] Hungria, M., Vargas, M.A.T. (2000) Environmental factors affecting N-2 fixation 492683/2004-2. We are grateful to Elisa Taschen for help in nodu- in grain legumes in the tropics, with an emphasis on Brazil. Field Crops Research 65, 151–164. lation tests and molecular experiments, and Lionel Moulin for very [21] Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., helpful comments on the manuscript. Strains were collected and McWilliam, H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., transported with IBAMA authorization, Licence no. 029/2006. Gibson, T.J., Higgins, D.G. (2007) Clustal W and clustal X version 2.0. Bioinfor- matics 23, 2947–2948. [22] Le Roux, C., Tentchev, D., Prin, Y., Goh, D., Japarudin, Y., Perrineau, M.M., Duponnois, R., Domergue, O., de Lajudie, P., Galiana, A. (2009) Bradyrhizobia Appendix A. Supplementary data nodulating the Acacia mangium × A. auriculiformis interspecific hybrid are spe- cific and differ from those associated with both parental species. Applied and Environmental Microbiology 75, 7752–7759. Supplementary data associated with this article can be found, in [23] Lesueur, D., Diem, H.G., Dianda, M., Leroux, C. (1993) Selection of Bradyrhi- the online version, at doi:10.1016/j.syapm.2011.03.003. zobium strains and provenances of Acacia mangium and Faidherbia albida relationship with their tolerance to acidity and aluminium. Plant and Soil 149, 159–166. [24] Librado, P., Rozas, J. (2009) DnaSP v5: a software for comprehensive analysis of References DNA polymorphism data. Bioinformatics 25, 1451–1452. [25] Loureiro, M.D., Kaschuk, G., Alberton, O., Hungria, M. (2007) Soybean Glycine [1] Bailly, X., Olivieri, I., Brunel, B., Cleyet-Marel, J.C., Bena, G. (2007) Horizon- max (L.) Merrill rhizobial diversity in Brazilian oxisols under various soil, tal gene transfer and homologous recombination drive the evolution of the cropping, and inoculation managements. Biology and Fertility of Soils 43, nitrogen-fixing symbionts of Medicago species. Journal of Bacteriology 189, 665–674. 5223–5236. [26] Martin, D., Rybicki, E. (2000) RDP: detection of recombination amongst aligned [2] Barcellos, F.G., Menna, P., da Silva Batista, J.S., Hungria, M. (2007) Evidence sequences. Bioinformatics 16, 562–563. of horizontal transfer of symbiotic genes from a Bradyrhizobium japonicum [27] Martin, D.P., van der Walt, E., Posada, D., Rybicki, E.P. (2005) The evolutionary inoculant strain to indigenous diazotrophs Sinorhizobium (Ensifer) fredii and value of recombination is constrained by genome modularity. PLoS Genet 1, Bradyrhizobium elkanii in a Brazilian Savannah soil. Applied and Environmental e51. Microbiology 73, 2635–2643. [28] Martin-Laurent, F., Lee, S.K., Tham, F.Y., Jie, H., Diem, H.G. (1999) Aeroponic pro- [3] Batista, J.S., Hungria, M., Barcellos, F.G., Ferreira, M.C., Mendes, I.C. (2007) Vari- duction of Acacia mangium saplings inoculated with AM fungi for reforestation ability in Bradyrhizobium japonicum and B. elkanii seven years after introduction in the tropics. Forest Ecology and Management 122, 199–207. of both the exotic microsymbiont and the soybean host in a cerrados soil. [29] Martin-Laurent, F., Tham, F.Y., Lee, S.K., He, J., Diem, H.G. (2000) Field assess- Microbial Ecology 53, 270–284. ment of aeroponically grown and nodulated Acacia mangium. Australian Journal [4] Binde, D., Menna, P., Bangel, E., Barcellos, F., Hungria, M. (2009) rep-PCR fin- of Botany 48, 109–114. gerprinting and based on the sequencing of the 16S rRNA gene of [30] McLoughlin, T.J., Alt, S.G., Merlo, P.A. (1990) Persistence of introduced Bradyrhi- 54 elite commercial rhizobial strains. Applied Microbiology and Biotechnology zobium japonicum strains in forming nodules in subsequent years after 83, 897–908. inoculation in Wisconsin soils. Canadian Journal of Microbiology 36, 794–800. [5] Boni, M.F., Posada, D., Feldman, M.W. (2007) An exact nonparametric method [31] Mendes, I.C., Hungria, M., Vargas, M.A.T. (2004) Establishment of Bradyrhizo- for inferring mosaic structure in sequence triplets. Genetics 176, 1035–1047. bium japonicum and B-elkanii strains in a Brazilian Cerrado oxisol. Biology and [6] Broughton, W.J., Dilworth, M.J. (1970) Methods in legume–rhizobium technol- Fertility of Soils 40, 28–35. ogy: plant nutrient solutions. In: Somasegaran, P., Hoben, H.J. (Eds.), Handbook [32] Menna, P., Hungria, M., Barcellos, F., Bangel, E., Hess, P., Martínez-Romero, for Rhizobia, NifTAL Project and University of Hawaii, Paia, Hawaii, pp. 245–249. E. (2006) Molecular phylogeny based on the 16S rRNA gene of elite rhizo- [7] Brutti, L., Rivero, E., Basurco, J.C.P., Nicolas, M., Iriarte, L., Abbiati, N., Ljunggren, bial strains used in Brazilian commercial inoculants. Systematic and Applied H., Martensson, A. (1998) Persistence of Bradyrhizobium japonicum in arable Microbiology 29, 315–332. soils of Argentina. Applied Soil Ecology 10, 87–94. [33] Menna, P., Hungria, M., Barcellos, F.G., Bangel, E.V., Hess, P.N., Martinez-Romero, [8] Chaintreuil, C., Boivin, C., Dreyfus, B., Giraud, E. (2001) Characterization of the E. (2006) Molecular phylogeny based on the 16S rRNA gene of elite rhizo- common nodulation genes of the photosynthetic Bradyrhizobium sp. ORS285 bial strains used in Brazilian commercial inoculants. Systematic and Applied reveals the presence of a new insertion sequence upstream of nodA. FEMS Microbiology 29, 315–332. Microbiology Letters 194, 83–86. [34] Minamisawa, K., Itakura, M., Suzuki, M., Ichige, K., Isawa, T., Yuhashi, K., Mitsui, [9] Chen, W.M., Moulin, L., Bontemps, C., Vandamme, P., Bena, G., Boivin- H. (2002) Horizontal tranfer of nodulation genes in soil and microcosms from Masson, C. (2003) Legume symbiotic nitrogen fixation by beta-proteobacteria Bradyrhizobium japonicum to B. elkanii. Microbes and Environment 17, 82–90. is widespread in nature. Journal of Bacteriology 185, 7266–7272. [35] Moreira, F.M.S., Gillis, M., Pot, B., Kersters, K., Franco, A.A. (1993) Character- [10] Chen, W.P., Kuo, T.T. (1993) A simple and rapid method for the prepara- ization of rhizobia isolated from different groups of tropical Leguminosae by tion of Gram-negative bacterial genomic DNA. Nucleic Acids Research 21, comparative polyacrylamide gel electrophoresis of their total proteins. System- 2260–12260. atic and Applied Microbiology 16, 135–146. [11] Clapp, J.P., Mansur, I., Dodd, J.C., Jeffries, P. (2001) Ribotyping of rhizobia nodu- [36] Moreira, F.M.S., Haukka, K., Young, J.P.W. (1998) Biodiversity of rhizobia iso- lating Acacia mangium and Paraserianthes falcataria from different geographical lated from a wide range of forest legumes in Brazil. Molecular Ecology 7, areas in Indonesia using PCR-RFLP-SSCP (PRS) and sequencing. Environmental 889–895. Microbiology 3, 273–280. [37] Nandasena, K.G., O’Hara, G.W., Tiwari, R.P., Howieson, J.G. (2006) Rapid in situ [12] de Faria, S.M., Franco, A.A. (2002) Identificac¸ ão de bactérias eficientes na fixac¸ão evolution of nodulating strains for Biserrula pelecinus L. through lateral transfer biológica de nitrogênio para espécies leguminosas arbóreas. Embrapa Agrobi- of a symbiosis island from the original mesorhizobial inoculant. Applied and ologia Documentos, 158. Environmental Microbiology 72, 7365–7367. [13] Dias, L.E., Franco, A.A., Campello, E.F.C., Faria, S.M., Castilho, A., Henriques, J.C., [38] Nandasena, K.G., O’Hara, G.W., Tiwari, R.P., Sezmis, E., Howieson, J.G. (2007) Daniels, W.L. (2008) Presented at the National Meeting of the American Society In situ lateral transfer of symbiosis islands results in rapid evolution of diverse of Mining and Reclamation, Richmond, VA. competitive strains of mesorhizobia suboptimal in symbiotic nitrogen fixation [14] Fremont, M., Prin, Y., Chauviere, M., Diem, H.G., Pwee, K.H., Tan, T.K. (1999) on the pasture legume Biserrula pelecinus L. Environmental Microbiology 9, A comparison of Bradyrhizobium strains using molecular, cultural and field 2496–2511. studies. Plant Science 141, 81–91. [39] Ngom, A., Nakagawa, Y., Sawada, H., Tsukahara, J., Wakabayashi, S., Uchiumi, [15] Galiana, A., Gnahoua, G.M., Chaumont, J., Lesueur, D., Prin, Y., Mallet, B. (1998) T., Nuntagij, A., Kotepong, S., Suzuki, A., Higashi, S., Abe, M. (2004) A novel Improvement of nitrogen fixation in Acacia mangium through inoculation with symbiotic nitrogen-fixing member of the Ochrobactrum clade isolated from rhizobium. Agroforestry Systems 40, 297–307. root nodules of Acacia mangium. Journal of General and Applied Microbiology [16] Galiana, A., Prin, Y., Mallet, B., Gnahoua, G.M., Poitel, M., Diem, H.G. (1994) 50, 17–28. Inoculation of Acacia mangium with alginate beads containing selected [40] Nicholas, K.B., Nicholas, H.J., Deerfield, D.W.I. 1997 Genedoc: Analysis and Visu- Bradyrhizobium strains under field conditions: long-term effect on plant alization of Genetic Variation, EMBnews, 4 pp. growth and persistence of the introduced strains in soil. Applied and Envi- [41] Nuswantara, S., Fujie, M., Sukiman, H.I., Yamashita, M., Yamada, T., Murooka, ronmental Microbiology 60, 3974–3980. Y. (1997) Phylogeny of bacterial symbionts of the leguminous tree Acacia [17] Giongo, A., Ambrosini, A., Vargas, L.K., Freire, J.R.J., Bodanese-Zanettini, M.H., mangium. Journal of Fermentation and Bioengineering 84, 511–518. Passaglia, L.M.P. (2008) Evaluation of genetic diversity of bradyrhizobia strains [42] Nzoue, A., Miche, L., Klonowska, A., Laguerre, G., de Lajudie, P., Moulin, L. (2009) nodulating soybean Glycine max (L.) Merrill isolated from South Brazilian fields. Multilocus sequence analysis of bradyrhizobia isolated from Aeschynomene Applied Soil Ecology 38, 261–269. species in Senegal. Systematic and Applied Microbiology 32, 400–412. [18] Gottfert, M., Rothlisberger, S., Kundig, C., Beck, C., Marty, R., Hennecke, H. (2001) [43] Obaton, M., Bouniols, A., Piva, G., Vadez, V. (2002) Are Bradyrhizobium japonicum Potential symbiosis-specific genes uncovered by sequencing a 410-kilobase stable during a long stay in soil? Plant and Soil 245, 315–326. DNA region of the Bradyrhizobium japonicum chromosome. Journal of Bacteri- [44] Padidam, M., Sawyer, S., Fauquet, C.M. (1999) Possible emergence of new gem- ology 183, 1405–1412. iniviruses by frequent recombination. Virology 265, 218–225. [19] Heath, L., van der Walt, E., Varsani, A., Martin, D.P. (2006) Recombination pat- [45] Parker, M.A., Kennedy, D.A. (2006) Diversity and relationships of bradyrhizobia terns in aphthoviruses mirror those found in other picornaviruses. Journal of from legumes native to eastern North America. Canadian Journal of Microbiol- Virology 80, 11827–11832. ogy 52, 1148–1157. 384 M.M. Perrineau et al. / Systematic and Applied Microbiology 34 (2011) 376–384

[46] Parker, M.A., Lafay, B., Burdon, J.J., van Berkum, P. (2002) Conflicting phylogeo- serradella in soils of Western Australia and South Africa. Applied and Environ- graphic patterns in rRNA and nifD indicate regionally restricted gene transfer mental Microbiology 71, 7041–7052. in Bradyrhizobium. Microbiology 148, 2557–2565. [57] Sullivan, J.T., Patrick, H.N., Lowther, W.L., Scott, D.B., Ronson, C.W. (1995) Nodu- [47] Posada, D., Crandall, K.A. (1998) MODELTEST: testing the model of DNA substi- lating strains of Rhizobium loti arise through chromosomal symbiotic gene tution. Bioinformatics 14, 817–818. transfer in the environment. Proceedings of the National Academy of Sciences [48] Posada, D., Crandall, K.A. (2001) Evaluation of methods for detecting recom- of the United States of America 92, 8985–8989. bination from DNA sequences: computer simulations. Proceedings of the [58] Sullivan, J.T., Ronson, C.W. (1998) Evolution of rhizobia by acquisition of a 500- National Academy of Sciences of the United States of America 98, 13757–13762. kb symbiosis island that integrates into a phe-tRNA gene. Proceedings of the [49] Prin, Y., Galiana, A., Le Roux, C., Meleard, B., Razafimaharo, V., Ducousso, M., National Academy of Sciences of the United States of America 95, 5145–5149. Chaix, G. (2003) Molecular tracing of Bradyrhizobium strains helps to correctly [59] Swofford, D.L. (1993) PAUP – a computer program for phylogenetic inference interpret Acacia mangium response to inoculation in a reforestation experiment using maximum parsimony. Journal of General Physiology 102, A9–A19. in Madagascar. Biology and Fertility of Soils 37, 64–69. [60] Turnbull, J.W., Midgley, S.J., Cossalter, C. (1998) Tropical Acacias planted in [50] Prin, Y., Nelson, E., Leroux, C., Galiana, A. (1998) Genetic diversity of nitro- Asia: an overview. In: Turnbull, J.W., Crompton, H.R., Pinyopusarerk, K. (Eds.), gen fixing rhizobia associated with Acacia mangium in its native area: North Recent Developments in Acacia Planting, Australian Center for International Queensland, Australia. In: Elmerich, C.e., et al. (Eds.), Biological Nitrogen Fix- Agricultural Research, Canberra, pp. 14–28. ation for the 21st Century, Kluwer Academic Publishers, Netherlands, pp. [61] Vergnaud, L., Chaboud, A., Prin, Y., Rougier, M. (1985) Pre-infection events in the 675–676. establishment of Alnus-Frankia symbiosis. Development of a spot inoculation [51] Ramirez-Bahena, M.H., Peix, A., Rivas, R., Camacho, M., Rodriguez-Navarro, D.N., technique. Plant and Soil 87, 67–78. Mateos, P.F., Martinez-Molina, E., Willems, A., Velazquez, E. (2009) Bradyrhizo- [62] Vincent, J.M. 1970 A Manual for the Practical Study of the Root-nodule Bacteria, bium pachyrhizi sp. nov. and Bradyrhizobium jicamae sp. nov., isolated from Blackwell Scientific Publ., Oxford. effective nodules of Pachyrhizus erosus. International Journal of Systematic and [63] Vinuesa, P., Leon-Barrios, M., Silva, C., Willems, A., Jarabo-Lorenzo, A., Evolutionary Microbiology 59, 1929–1934. Perez-Galdona, R., Werner, D., Martinez-Romero, E. (2005) Bradyrhizobium [52] Rivas, R., Martens, M., de Lajudie, P., Willems, A. (2009) Multilocus sequence canariense sp. nov., an acid-tolerant endosymbiont that nodulates endemic analysis of the genus Bradyrhizobium. Systematic and Applied Microbiology genistoid legumes (Papilionoideae: Genisteae) from the Canary Islands, along 32, 101–110. with Bradyrhizobium japonicum bv. genistearum, Bradyrhizobium genospecies [53] Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., alpha and Bradyrhizobium genospecies beta. International Journal of System- Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B., atic and Evolutionary Microbiology 55, 569–575. Thallinger, G.G., Van Horn, D.J., Weber, C.F. (2009) Introducing mothur: Open- [64] Vinuesa, P., Rojas-Jimenez, K., Contreras-Moreira, B., Mahna, S.K., Prasad, B.N., Source, Platform-Independent. Community-Supported Software for describing Moe, H., Selvaraju, S.B., Thierfelder, H., Werner, D. (2008) Multilocus sequence and comparing microbial communities. Applied and Environmental Microbi- analysis for assessment of the biogeography and evolutionary genetics of ology 75, 7537–7541. four Bradyrhizobium species that nodulate on the asiatic continent. [54] Smith, J.M. (1992) Analyzing the mosaic structure of genes. Journal of Molecular Applied and Environmental Microbiology 74, 6987–6996. Evolution 34, 126–129. [65] Wang, F.Q., Wang, E.T., Zhang, Y.F., Chen, W.X. (2006) Characterization of rhi- [55] Souvannavong, O., Galiana, A. (1990) Acacia mangium – rhizobium symbiosis: zobia isolated from Albizia spp. in comparison with microsymbionts of Acacia selection and propagation of the host plant and the microorganism. In: Taylor, spp. and Leucaena leucocephala grown in China. Systematic and Applied Micro- D.A., MacDicken, K.G. (Eds.), Research on Multipurpose Tree Species in Asia, biology 29, 502–517. Bangkok. pp. 216–222. [66] Zhang, X.P., Harper, R., Karsisto, M., Lindstrom, K. (1991) Diversity of rhizo- [56] Stepkowski, T., Moulin, L., Krzyzanska, A., McInnes, A., Law, I.J., Howieson, J. bium bacteria isolated from the root nodules of leguminous trees. International (2005) European origin of Bradyrhizobium populations infecting lupins and Journal of Systematic Bacteriology 41, 104–113.