Department of Biology, Institute of Pharmacy and Molecular Biotechnology, Heidelberg University, Heidelberg, Germany
Correspondence: Michael Wink, Department of Biology, Institute of Pharmacy and Molecular Biotechnology, Heidelberg University, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany. Tel.: +49 6221 544847; fax: +49 6221 544884; e-mail: firstname.lastname@example.org
Lentil is the oldest of the crops that have been domesticated in the Fertile Crescent and spread to other regions during the Bronze Age, making it an ideal model to study the evolution of rhizobia associated with crop legumes. Housekeeping and nodulation genes of lentil-nodulating rhizobia from the region where lentil originated (Turkey and Syria) and regions to which lentil was introduced later (Germany and Bangladesh) were analyzed to determine their genetic diversity, population structure, and taxonomic position. There are four different lineages of rhizobia associated with lentil nodulation, of which three are new and endemic to Bangladesh, while Mediterranean and Central European lentil symbionts belong to the Rhizobium leguminosarum lineage. The endemic lentil grex pilosae may have played a significant role in the origin of these new lineages in Bangladesh. The presence of R. leguminosarum with lentil at the center of origin and in countries where lentil was introduced later suggests that R. leguminosarum is the original symbiont of lentil. Lentil seeds may have played a significant role in the initial dispersal of this Rhizobium species within the Middle East and on to other countries. Nodulation gene sequences revealed a high similarity to those of symbiovar viciae.
Nitrogen is an essential nutrient for all living organisms and necessary for high crop yield and plant quality in agriculture, but only prokaryotes can convert atmospheric nitrogen into forms that are usable to plants. Rhizobia are nitrogen-fixing soil bacteria that are able to enter a mutual symbiosis with leguminous plants that fully or partially satisfy the nitrogen demand of the host plant. During the infection process, rhizobia produce a number of host-specific factors, and thus, it has been assumed that rhizobia have coevolved with their host plants (Perret et al., 2000; Martinez-Romero, 2009). Rhizobium leguminosarum is a cosmopolitan and well-studied species in the genus Rhizobium. The name R. leguminosarum was first proposed by Frank (1889) for all nodule-forming bacteria, and the species currently has three biovars that differ in their host plant specificity (Jordan, 1984).
Lentil (Lens culinaris) is the oldest crop that was domesticated in the Fertile Crescent around 9000 years ago (Zohary & Hopf, 2000; Toklu et al., 2009) and remains popular worldwide for human nutrition and for soil fertility management (Sarker & Erskine, 2006; Sonnante et al., 2009). The region of origin encompasses southeastern Turkey and northern Syria, including the sources of the rivers Tigris and Euphrates (Lev-Yadun et al., 2000) and cultivation spread to Cyprus and, via the Danube, to Europe around 7000 years ago. In Georgia, lentil was cultivated 4000–5000 years ago and was transported to the Indian subcontinent around 2000–2500 years ago (Erskine, 1997; Sonnante et al., 2009). Rhizobia can be carried on the testa of seeds (Perez-Ramírez et al., 1998), allowing them to disperse to different geographical regions along with the seeds.
Nucleotide sequences of the 16S rRNA genes are widely used as genetic markers for bacterial classification but, for a more precise identification and description of closely related bacterial species, multilocus sequence analysis (MLSA) using different protein-coding genes has become the preferred method (Ludwig & Klenk, 2005; Konstantinidis et al., 2006; Martens et al., 2008). Phylogenies inferred from chromosomal genes and plasmid-encoded symbiosis genes of rhizobia are frequently found to be incongruent (Sprent, 1994; Laguerre et al., 1996; Young & Haukka, 1996). Recombination occurs frequently in bacteria and plays an important role in the evolution of bacteria (Vinuesa et al., 2005; Bailly et al., 2006; den Bakker et al., 2008; Tian et al., 2012).
The diversity of rhizobia from pea, faba bean, and vetches has been studied previously (Laguerre et al., 1996; Mutch & Young, 2004; Hou et al., 2009; and many others). It has been concluded that R. leguminosarum is the main nodulating species in the tribe Vicieae (Tian et al., 2010; and references therein), although related species such as R. pisi and R. fabae have also been described (Ramírez-Bahena et al., 2008; Tian et al., 2008). In contrast, there are relatively few studies on rhizobia that nodulate lentil (Hynes & O'Connell, 1990; Moawad & Beck, 1991; Laguerre et al., 1992a, b; Geniaux & Amarger, 1993; Materon et al., 1995; Rashid et al., 2009), and we found three distinct species-level lineages related to R. etli and R. phaseoli (López-Guerrero et al., 2012) in Bangladesh, while R. leguminosarum was absent (Rashid et al., 2012). It is therefore important to examine lentil symbionts from other geographic regions to establish whether lentils are exceptional from other legumes of the tribe Vicieae in having different symbionts. We compared DNA sequences of rhizobia isolated from three countries where the grex pilosae is absent (Barulina, 1930; Sarker & Erskine, 2006) with those of isolates isolated previously (Rashid et al., 2012) from field-grown pilosae (A. Sarker, M. Rahman and M.A. Samad, pers. commun.) lentils in Bangladesh. The aim of this study was (i) to explore the genetic diversity and identity of lentil-nodulating rhizobia, (ii) to evaluate the levels of genetic diversity and population structure of these bacteria from different geographical locations.
Materials and methods
Soil samples, plant growth, and nodule separation
Rhizobia were isolated from nodules of lentils (variety BINA-3, grex pilosae) grown in potted soil under glasshouse conditions and from field-grown lentils (variety Anicia, small green lentil, originally imported from France, W. Mammel, pers. commun.). Field-grown nodules were collected from Lauterach, Baden-Württemberg, Germany. Soil samples were collected from nine locations in Germany, one in Turkey, and two locations in Syria (Supporting Information, Table S1). All soil samples were collected from cultivated soils, except for one that had been collected from a forest in Germany. Soil samples were kept well separated and processed for growing plants under glasshouse conditions within 3–12 days of collection. About 2.5–3.0 kg of soil was transferred to a surface-sterilized plastic pot to grow lentils. One pot per locality was used to grow 2–3 lentil plants for 5 weeks.
Surface-sterilized (1 min in 70% ethanol and 3–5 min in 3% NaOCl) and pregerminated (48 h on 1% water agar) lentil seeds were then placed on potted soil. After germination, a maximum of three plants were cultivated for 5 weeks. Plants were irrigated alternately (i.e. water, then N-free seedling solution, then water, etc.) with sterile water and nitrogen-free seedling solution when needed. After 5 weeks, plants were uprooted carefully, the roots washed with water, dried with tissue paper, and then preserved on silica gel until further processing.
Isolation of Bacteria
For rhizobial isolation, we selected 2–11 pink nodules from lentils of each pot representing a single location, and a maximum of five pink nodules/plant from field-grown lentil. We isolated and purified rhizobia from the selected nodules following standard protocols using CRYEMA (yeast extract mannitol agar medium with congo red) described in Somasegaran & Hoben (1994). Each colony was purified by repeated streaking on CRYEMA medium and preserved at −80 °C with 25% glycerol and at 4 °C on agar slants for further study.
Determination of rhizobial population from collected soil
The number of rhizobial cells in 1 g of collected soil sample was determined following standard protocols described by Brockwell (1963). Fivefold serial dilution with three replicates for each soil sample and four replicates for each dilution were used for plant inoculation. Seeds were surface-sterilized and germinated on agar plates, and the seedlings were transferred to growth medium as described in Rashid et al. (2012). Harvested plants were scored for the presence or absence of nodules. The rhizobial population density of soil samples was determined following the MPN table (Brockwell, 1963; and references therein).
Determination of soil pH
Air-dried soil samples (200 g) were first ground and then sieved (2 mm) to remove large particles. From the sieved sample, 10 g was used to determine the pH. From each locality, the soil pH was measured using 0.01 mM CaCl2 following the protocol ISO 2006 (International Standard Organization; www.iso.org) with a HANNA pH meter (HI 98150).
Nodulation and cross-inoculation tests
All isolates were tested for nodule formation with lentils under growth chamber conditions following standard protocol (Somasegaran & Hoben, 1994). A randomly selected set of 30 isolates was used for cross-inoculation tests with Lathyrus sativus and Pisum sativum. Lentil (Teller ‘Linsen’ from Müller's Mühle, commercial grade, from Germany), grass pea (Binamasur-1 from Bangladesh), and pea (variety unknown, commercial grade from Germany) were surface-sterilized using 70% ethanol (1 min) and 3% NaOCl (3–5 min). After sterilization, seeds were washed with sterile water (six times), imbibed in sterile water for 4 h, and then allowed to germinate on 1% water agar for 48 h. Germinated seeds were then transferred to glass tubes (32 mm × 170 mm) containing Fåhreus (1957) medium. Glass tubes containing germinated seed were wrapped with aluminum foil and placed in a growth chamber set to 25 °C temperature with 14-h light/10-h dark cycles for 5 weeks. There were three replications for each isolate, and uninoculated plants were kept as a control.
DNA isolation, PCR, and DNA sequencing
Bacterial isolates were grown at 28 °C for 24–36 h in tryptone-yeast (TY) medium (Beringer, 1974). DNA was extracted following Chen & Kuo (1993) and dissolved in TE buffer. DNA concentration and purity were measured by UV spectrophotometry. PCR amplifications were carried out with about 50 ng of purified DNA. Primer sequences and PCR conditions for the sequencing of four housekeeping genes (16S rRNA gene, recA, atpD, and glnII) and two nodulation genes (nodC and nodD) are provided in Table S1. For sequencing, PCR products were precipitated by addition of one volume of 4 M ammonium acetate with 10 volumes of 100% ice-cold ethanol and pelleted by centrifugation (16 060 g for 30 min at 4 °C). The pellet was washed with 70% ethanol and dried at 65 °C for 10 min. The purified PCR products were dissolved in ultrapure HPLC-grade water before sequencing. Sequencing was performed using an ABI 3730 automated capillary sequencer (Applied Biosystems) with the ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction Kit version 3.1 (carried out by STARSEQ GmbH, Mainz, Germany). Sequences generated in this study are deposited in GenBank under the accession numbers KC679411–KC679680 (Table S4).
Sequences were aligned using clustalw in bioedit version 7.1.3 (Hall, 1999) with manual adjustment. We reconstructed phylogenetic trees using the neighbor-joining (NJ) and maximum likelihood (ML) algorithms in mega version 5 (Tamura et al., 2011). The general time reversible (GTR) model of sequence evolution with gamma distribution was used in ML analysis. Bootstrap support for each node was evaluated with 1000 replicates. Protein-coded housekeeping gene trees (recA, atpD, gln II, and the tree from their concatenated sequence) were rooted with sequences from Rhizobium yuanmingense and 16S rRNA gene tree was rooted with Mesorhizobium loti. Uncorrected genetic distances (p-distance) between different phylogenetic sublineages and the type strain of R. leguminosarum were estimated in mega version 5 (Tamura et al., 2011). Neighbor-net (Bryant & Moulton, 2004) analyses were conducted with the program splitstree4 version 4.12.3 (Huson & Bryant, 2006). Genes were analyzed separately and together in a combined data set.
Population genetic analyses
Parameters such as recombination events and gene flow were measured with dnasp version 5.10.01 (Rozas et al., 2010). The population structure was evaluated with structure version 2.3.3 (Pritchard et al., 2000; Falush et al., 2003) from housekeeping gene sequences. The most likely number of clusters (K = 1–10) was determined under an ‘admixture’ model, 20 000 ‘burn-in’ and 50 000 sampling iterations following the procedure described by Evanno et al. (2005), which gave the highest peak against eight assumed populations. With an estimated K = 8, five extra long runs of 20 000 ‘burn-in’ and 100 000 sampling iterations were performed. Hierarchical analysis of molecular variance (amova; Excoffier et al., 1992) was conducted in Arlequin version 3.5 (Excoffier et al., 2005).
Recombination and mutation analyses
Levels of recombination, mutation rates, and 50% majority rule consensus trees (with and without recombination) were estimated from housekeeping genes using clonalframe version 1.1 (Didelot & Falush, 2007). Three independent runs were performed with a 100 000 ‘burn-in’ and 300 000 sampling iterations. Satisfactory MCMC convergence was judged following the criterion of (Gelman & Rubin, 1992).
We made use of three approaches to estimate the level of recombination in samples from Germany, Turkey, and Syria: (i) minimal intragenic recombination events (Rm) were detected and compared with expected values of coalescence simulations based on 10 000 genealogy replications at 95% confidence level (Hudson et al., 1992; Rozas et al., 2010) analyzing single genes and the combined data set in dnasp version 5.10.1 (Rozas et al., 2010) (ii) the Shimodaira–Hasegawa (S-H) test (Shimodaira & Hasegawa, 1999) was performed to compare ML tree topologies for phylogenetic congruence as implemented in tree-puzzle version 5.2 (Schmidt et al., 2003); and (iii) recombination rates were determined by the relative impact of recombination as compared with point mutations in the genetic diversification of the lineages (r/m proportion; Guttman & Dykhuizen, 1994) and the relative frequency of the occurrence of recombination as compared with point mutation in the history of the lineage (ρ/θ proportion; Milkman & Bridges, 1990); these analyses were carried out in clonalframe version 1.1 as described before.
Bacterial isolation, soil pH, and rhizobial population density
A total of 98 rhizobial colonies were isolated from lentil nodules representing 12 localities in three countries: Germany (N = 78), Turkey (N = 12), and Syria (N = 8), and different genes were sequenced from 58 isolates (Tables S1 and S2). In addition, we included seven previously isolated strains from Bangladesh (Rashid et al., 2012) in the analyses for species delineation and comparison. A single colony was isolated from each of the selected nodules. Isolates identifier and the corresponding localities are documented in Table S1. Soil pH ranged around neutral (pH 6.5–7.4), with the exception of one forest soil sample from Heidebuckelweg, Heidelberg that presented an acidic pH of 4.8 (Table S1). Rhizobial population density varied across different localities in Germany, from 114 cells g−1 soil in Bürstadt (Hessen) to 2.18 × 103 cells g−1 soil in Ostrach (Baden-Württemberg; Table S1).
Nodulation, cross-inoculation, and symbiotic effectiveness test
Nodulation efficiency test showed that all 98 isolates were able to form nodules with lentil within 3–4 weeks after inoculation under growth chamber conditions. In a cross-inoculation test, a set of 30 randomly selected isolates were able to form nodules with both Lathyrus sativus and Pisum sativum under the same growth conditions. All isolates produced dark pink nodules, and plant leaves were darker green compared to the noninoculated controls, demonstrating that all isolates were symbiotically effective.
Phylogenetic analyses based on housekeeping gene sequences
We amplified the 16S rRNA gene (about 1500 bp length) and obtained sequences of about 1100–1350 bp from 38 rhizobial isolates originating from three different countries. blast searches indicated high similarities (99–100%) to R. leguminosarum. We recovered very similar topologies using different tree reconstruction methods, that is, NJ and ML. Phylogenetic analyses based on 16S rRNA gene sequences revealed that all isolates from the three different geographical origins were closely related to R. leguminosarum but separate from the Bangladeshi isolates (Supporting Information, Fig. S1).
Phylogenetic analyses based on three individual housekeeping genes, recA (415 bp from 58 isolates s), atpD (472 bp from 55 isolates), and glnII (598 bp from 57 isolates; Figs S2–S4), and their concatenated sequence (Fig. 1) recovered five sublineages (IVa–IVe) with high bootstrap support (70–90%). However, the recovered tree topologies differed in phylogenetic analyses, that is, some isolates changed their positions between different sublineages depending on the analyzed gene, revealing phylogenetic incongruence among the loci (Tables S2 and S8). Defined by high bootstrap support, two new sublineages (IVb and IVc) differ from sublineages of R. leguminosarum described previously from various hosts and geographic regions (Tian et al., 2010). Phylogenetic analyses and genetic distances of three protein-coding genes (Valverde et al., 2006) with respect to the type strain of R. leguminosarum (Fig. 1, Figs S2–S4 and Table S5) suggest that all isolates belong to R. leguminosarum.
Species delineation and recombination visualization using neighbor-net analysis
Neighbor-network analysis based on the combined data set (recA-atpD-glnII) showed a reticulate structure, in which we identified five sublineages (Fig. 2). Two sublineages (IVb and IVc) are distinguishable from other described R. leguminosarum sublineages (Tian et al., 2010). In neighbor-network analysis, by a long edge, lentil isolates from Bangladesh (lineages I, II, and III) differed significantly from German, Turkish, and Syrian isolates, and they did not form any reticulate structure among themselves (Fig. S5). The isolates GLR7, GLR45, TLR14, and TLR10, which lay outside the main phylogenetic clusters (Fig. 1), had unique positions in the network with a high level of reticulation, indicating that they are potentially recombinant for one or more genes.
Genetic diversity analyses using structure
With five long runs in structure, we determined the optimal number of clusters K to be 8 and found admixture among populations (Fig. 3a and b). However, we obtained admixed structures in isolates GLR7, GLR23, GLR31, GLR33, GLR40, GLR45, GLR67, GLR74, TLR7, TLR10, and TLR14. In contrast, the new lineages from Bangladesh apparently did not show any admixture (Fig. 3a and b).
Detection of minimum recombination events and gene flow using DnaSP
Minimum recombination events found in single gene sequences and concatenated sequences of three genes are shown in Table 1. The three housekeeping genes, recA, atpD, and glnII, reveal 12, 14, and 16 recombination events, respectively, and concatenated data showed 43 recombination events. High values (Nm = 4.34) of gene flow were found between Turkish and Syrian isolates, along with nonsignificant KST values (0.039), while low values (Nm = 1.91) for the same parameters were found between Germany vs. Turkey and Syria (Table S6).
Table 1. Minimum recombination events on different genes
Gene size (bp)
P < observed Rm
Rm, minimal intragenic recombination events; N, number of samples; bp, base pair.
Differentiation between geographical groups
amova of the concatenated sequence of protein-coding genes (recA-atpD-gln II) from samples from Germany, Turkey, and Syria indicated the existence of significant differences between geographical regions (Table S6), although the percentage of variation remained low among populations (11–20%) compared to the variation found within populations (80–90%). However, there were no significant differences between Turkish and Syrian isolates (P > 0.05; Table S6). Overall, amovas showed that German samples differed significantly from Turkish and Syrian samples, with Turkish samples the most different (Table S6).
Relative impact of recombination and point mutation
We determined the relative effect of recombination vs. point mutations from recA, atpD, and gln II gene sequences, which were used to make concatenated sequence. The r/m was 1.52 and the ρ/θ was 0.269, suggesting a greater importance of recombination over mutation for explaining the observed genetic diversity (Table 2). The reconstructed dendrograms from ClonalFrame analyses revealed much shorter times to the most recent common ancestor (TMRCA) when considering recombination (TMRCA = 0.220 (0.130–0.352), Fig. 4, Table S7) than when assuming no recombination (TMRCA = 1.110 (0.342–3.073) Fig. S6, Table S7). The tree topology also differed between these phylograms. For example, the position of the sublineage e and polytomies in the sublineages a and e (Fig. 4) are well resolved assuming no recombination (Fig. S6). However, we obtained more polytomies in the dendrogram when considering recombination (Fig. 4).
Table 2. Recombination effect inferred by ClonalFrame (confidence intervals are shown between parentheses)
R, recombination rate; r/m, relative impact of recombination as compared with point mutation; ρ/θ, relative frequency of the occurrence of recombination as compared with point mutation; θ, mutational rate.
Symbiotic gene analyses
The maximum likelihood analyses of nodulation gene sequences recovered five groups (based on nodC from 38 isolates and nodD from 23 isolates) from Germany, Turkey, and Syria (Fig. 5 and Fig. S7). In nodC gene analysis, group A contained lentil isolates from Germany, Turkey, and Syria. This group clusters with previously described strains isolated from Peru, Spain, and United Kingdom from different members of the legume tribe Vicieae. This group includes the isolate BLR195, which was isolated from lentils in Bangladesh and belongs to lineage III, suggesting a clear case of horizontal transfer of nodulation genes between lineage III and IV. In same tree (nodC), group B and D contained isolates from Germany only, while groups C and E contained isolates from Turkey (except GLR10) and Syria, respectively. Among seven groups (A–E from Germany, Turkey, and Syria, and groups I–II from Bangladesh), the group C from Turkey and the group II from Bangladesh differed considerably from existing nodC sequences of R. leguminosarum symbiovar viciae strains. Moreover, two isolates with unique nodC sequences, GLR2 and SLR2, showed significant differences to existing strains.
We sequenced nodD gene from 24 isolates (11 from Germany, six from Turkey, six from Syria, and one from Bangladesh) to compare with the nodC gene analysis to determine whether they were congruent or not. The reconstructed ML tree from the nodD gene sequences showed similar tree topologies (except for the isolates GLR17 and SLR 4) to the one based on nodC gene sequences (Fig. S5). The exception of GLR17 and SLR4 isolates may arise from internal rearrangement of the nod region by recombination. Group A corresponds to the previously described nodD type II/nodD type g (Laguerre et al., 2003; Mutch & Young, 2004; Tian et al., 2010 and reference therein) from faba bean rhizobia from different geographical locations (Jordan, Spain, Canada, and UK). Group B showed similarity to the previously described nodD type III from faba bean rhizobia from Europe (France and UK) and China (Tian et al., 2010). Although Turkish isolates form a separate group (group C) in the nodC gene tree, this group showed similarity to a previously described nodD type I from the Middle East and China, suggesting rearrangement of nodD gene within the nod region by recombination. Although the nodC group from Syria (E) was close to previously described strains, in the nodD gene tree, this group formed a strong separate group from existing nodD groups. The isolate GLR17 had an identical sequence to a distinct strain found in France from pea rhizobia. Although Bangladeshi isolates formed a strongly separate group in nodC gene, members of this group were very close to the previously described nodD type IV (Tian et al., 2010) in nodD gene. In terms of topology, the nodulation gene trees differed from trees based on chromosomal genes, suggesting horizontal/lateral gene transfer events of nodulation genes among different lineages and sublineages.
Genetic diversity of chromosomal genes and species delineation of lentil rhizobia
We sequenced three protein-coding housekeeping genes (recA, atpD, and glnII) that contain valuable information for determining biogeographic patterns (Palys et al., 1997; Lan & Reeves, 2001; Vinuesa et al., 2005) and identified five sublineages (IVa–IVe) among the 58 isolates from three countries (Turkey, Syria, and Germany). All sublineages belonged to R. leguminosarum, and two sublineages (IVb & IVc) did not show any great similarity to the sublineages that were described earlier within R. leguminosarum (Tian et al., 2012). We found phylogenetic incongruence in reconstructed trees from chromosomal genes, suggesting an influence of recombination on housekeeping genes. Consistent with this, results from different analyses (S-H test, recombination analyses, estimation of TMRCA) showed a substantial level of recombination among sublineages (Fitch, 1997; Shimodaira & Hasegawa, 1999; Bryant & Moulton, 2004; Tian et al., 2010).
Network analysis allow conflicting or alternative phylogenetic histories to account for ambiguities caused by recombination, hybridization, gene conversion, and gene transfer (Fitch, 1997). From this analysis, we obtained a clear reticulate structure among different sublineages from three different countries, but a long edge between Bangladeshi isolates, and samples from the other three countries. In other words, network analysis showed a clear difference between lineage IV and the others by a long edge, suggesting that lineage IV belongs to separate species (Bailly et al., 2006). Lentil-nodulating rhizobia from Bangladesh were well separated by long edges from R. leguminosarum (those nodulating lentils in the Mediterranean region and Central Europe) and always formed three lineages distinct from closely related R. etli and R. phaseoli with no incongruence in phylogenetic analyses. This is evidence that the Bangladeshi samples belong to separate, currently undescribed, species. Phylogenetic analyses (ML and BI) of three housekeeping genes and their concatenated sequence found three distinct species-level lineages in Bangladesh without any incongruence. These lineages also showed substantial differences, not only from closely related species, but also among themselves. Moreover, these lineages also showed phenotypic differences from closely related species (Rashid et al., 2012). Based on protein-coding genes (recA, atpD and glnII), lineage IV is genetically similar to the R. leguminosarum type strain (> 94%) and forms a reticulate structure with R. leguminosarum in network analysis suggesting (Bailly et al., 2006; Valverde et al., 2006; Santillana et al., 2008) that these sublineages belong to R. leguminosarum.
Origin and distribution of new lineages in Bangladesh are influenced by symbiosis with lentil grex pilosae
For cultivated lentils, Barulina (1930) proposed six geographical groups, or greges, viz. europeae, asiaticae, intermediae, subspontanea, aethiopicae, and pilosae (Cubero, 1981; Cubero et al., 2009; and references therein). Interestingly, among the six greges of cultivated lentil, three groups are restricted to very specific areas. For instance, pilosae is endemic to the Indian subcontinent, aethiopicae to Ethiopia and Yemen, and subspontanea to Afghanistan. This study and others on lentil rhizobia (Hynes & O'Connell, 1990; Moawad & Beck, 1991; Laguerre et al., 1992a, b; Geniaux & Amarger, 1993; Materon et al., 1995) show clearly that R. leguminosarum is the main symbiont of all lentil greges except pilosae in the Indian subcontinent.
Although Eastern Turkey and Northern Syria are the area of lentil domestication (Barulina, 1930; Ladizinsky, 1979; Lev-Yadun et al., 2000; Zohary & Hopf, 2000; Cubero et al., 2009), the Himalaya-Hindu Kush may corresponds to the center of origin for small seeded microsperma lentils because of the presence of a higher proportion of endemic varieties in this region. Pilosae has a strong pubescence, which is absent in other lentils (Cubero et al., 2009), and a flowering asynchrony, and has not overlapped with any other greges of lentil during the history of domestication and cultivation (Barulina, 1930; Erskine et al., 1994). This grex may also have specific genetic characteristics like nod factor receptors that allow for a successful symbiosis with the new Rhizobium species or lineages found in Bangladesh. Thus, pilosae and their symbionts may have coevolved in the Indian subcontinent. It is therefore possible that we identified new lineages (Rashid et al., 2012) of rhizobia from Bangladesh due to the significant influence of the pilosae grex on their symbiotic partners.
Rhizobium leguminosarum is the original symbiont of lentils
In agreements with other studies (Moawad & Beck, 1991; Laguerre et al., 1992a, b; Geniaux & Amarger, 1993; Moawad et al., 1998; Tegegn, 2006), we found R. leguminosarum in the center of origin of lentil and in countries where lentil had been introduced (e.g. Germany). We not only found the same R. leguminosarum species but also the same chromosomal genotype (e.g. sublineage IVa) in three different countries. Hence, it could be assumed that R. leguminosarum is the original symbiont of lentils. The dispersal of rhizobia with legume seeds is a well-accepted hypothesis (Perez-Ramírez et al., 1998; Aguilar et al., 2004; Álvarez-Martinez et al., 2009). This mode is considered to be the most important among the indirect ways of rhizobium dispersal (Hirsch, 1996; and references therein). Lentil is the oldest crop, has remained popular from ancient times until now, and is found all over the world. Thus, it could be assumed that lentil seeds might have played a significant role in initial dispersal of R. leguminosarum symbiovar viciae to different countries. However, of the three distinct lineages, one lineage (lineage I) is found all over Bangladesh (Rashid et al., 2012), suggesting that it may have been distributed with pilosae seeds.
Phylogenetic incongruence between chromosomal and nodulation genes
Rhizobial species diversity should be described not only based on genetic markers located on chromosomes but also on plasmid-borne nodulation genes (Graham et al., 1991; Amarger et al., 1997; Wang et al., 1999; Laguerre et al., 2001; Silva et al., 2005). The nodC and nodD genes determine host range, host promiscuity, and the relationships between host plants and rhizobia (Laguerre et al., 2001; Zeze et al., 2001; Iglesias et al., 2007). In our study, phylogenetic analyses based on nodulation genes confirm that these isolates belong to symbiovar viciae, and that they have nodC and nodD genotypes which have previously been described from Europe, Middle East, and China. We also detected new groups within this symbiovar. There was no congruence between chromosomal and nodulation genotypes, which may be due to frequent lateral transfer of nodulation genes between different rhizobial chromosomal genotypes (Sprent, 1994; Young & Haukka, 1996). However, phylogenetic analyses based on nodulation genes (nodC and nodD) revealed similar tree topologies correlated mostly with ecological regions. Probably, as a consequence of both host and soil microhabitats (Sprent, 1994), we detected new nodulation genotypes (or groups) that were supported by both nodulation genes. These results support the hypothesis that plasmid-borne characters in bacteria change rapidly according to their adaptation to particular environment. Nonetheless, nodulation genes from the Europe and the Middle East did not show great similarities to the isolates from Bangladesh, suggesting that the latter have an independent origin on the Indian subcontinent. A similar hypothesis has been proposed for the origin of R. etli (Aguilar et al., 2004).
By analyzing lentil-nodulating rhizobia from four countries on two different continents, we found four different lineages of rhizobia, of which three are new. These three new lineages of rhizobia are endemic to Bangladesh and have probably coevolved with the lentil grex pilosae. The presence of common genotypes of R. leguminosarum with lentil in different countries suggests that R. leguminosarum is the original symbiont of lentil. Additional research is needed to further examine the genetic diversity and population structure of lentil-nodulating rhizobia in different geographical regions.
Erasmus Mundus Mobility with Asia (EMMA) provided a PhD scholarship to MHR, and additional funding was provided by the DAAD-Stipendien-und Betreuungsprogramm (STIBET) grant, Heidelberg University. We are grateful to Dr. Martin Krehenbrink for fruitful discussion while writing this manuscript. We thank Dr. Holger Schäfer, Heidi Staudter, Annika Heinemann, Vanessa Erbe, Johannes Reiner, Beate Waibel, Razan Hamoud, Shirin Hamoud, Rasha Abou Aleinein, Dr. Tamer Albayrak, Dr. Sabine Grübe, Woldae Mamel, and Markus Santhosh Braun for their help in collecting samples, and Hedwig Sauer-Gürth and Heidi Staudter (IPMB) and Michael Schilbach (COS) for technical assistance. This work has been performed according to German laws.