Genetic diversity of Acacia tortilis ssp. raddiana rhizobia in Tunisia assessed by 16S and 16S-23S rDNA genes analysis


S. Ben Romdhane, Laboratoire Interactions Légumineuses Microorganismes (LILM), Institut National de Recherche Scientifique et Technique (INRST), BP 95, 2050 Hammam-Lif, Tunisia. E-mail:


Aims:  In order to understand the genetic diversity of Acacia tortilis ssp. raddiana-rhizobia in Tunisia, isolates from nine geographical locations were obtained and analysed.

Methods and Results:  Characterization using restriction fragment length polymorphism analysis (RFLP) of PCR-amplified 16S rRNA gene and the intergenic spacer (IGS) between the 16S and 23S rRNA genes was undertaken. Symbiotic efficiency of the strains was also estimated. Analysis of the 16S rRNA by PCR-RFLP showed that the isolates were phylogenetically related to Ensifer ssp., Rhizobium tropicii-IIA, and Rhizobium tumefaciens species. Analysis of 16S-23S spacer by PCR-RFLP showed a high diversity of these rhizobia and revealed eleven additional groups, which indicates that these strains are genetically very diverse. Full 16S rRNA gene-sequencing showed that the majority of strains form a new subdivion inside the genera Ensifer, with Ensifer meliloti being its nearest neighbour. Nodulation test performed on the plant host demonstrated differences in the infectivity among the strains.

Conclusion:  Rhizobial populations that nodulate specifically and efficiently Acacia tortilis ssp. raddiana in representative soils of Tunisia is dominated by E. meliloti-like genomospecies.

Significance and Impact of the Study:  This paper provides the first clear characterization and symbiotic efficiency data of rhizobia strains nodulating A. tortilis in Tunisia.


Rhizobia are widespread soil bacteria able to induce the formation of root nodules and to fix nitrogen on cultivated and wild legumes. These rhizobia are of economic importance in low-input sustainable agriculture, agroforestry, and land reclamation. The taxonomy of bacterial endosymbionts of leguminous plants has experienced a profound series of extensions in the recent past (Young 2003). Currently there are five genera of rhizobia in the α-Proteobacteria, Azorhizobium, Bradyrhizobium, Rhizobium, Mesorhizobium, and Ensifer (Young 2003). New lines that contain nitrogen-fixing legumes symbionts include Methylobacterium (Jaftha et al. 2002; Jourand et al. 2004), Devosia (Rivas et al. 2003), Blastobacter (Van Berkum and Eardly 2002) and Ochrobactrum (Ngom et al. 2004; Trujillo et al. 2005) in the α-Proteobacteria; Burkholderia (Moulin et al. 2001), Cupriavidus (Vandamme and Coenye 2004) and Ralstonia (Chen et al. 2001) in the β-Proteobacteria and some unclassified strains in the γ-Proteobacteria (Benhizia et al. 2004) were recently described.

Both desertification and ecosystem degradation problems are common in Mediterranean regions, particularly in the centre and the south of Tunisia, where low precipitation and human activities enhance the erosion and desertification processes. It is widely recognized that indigenous rhizobia play an important role in the dominance of Fabaceae in poor and arid soils (Zahran 2001) but, so far, only few wild legumes have been investigated for their nitrogen-fixing symbionts (Zakhia et al. 2004).

Acacia is widespread in arid regions of Africa and Middle East (Nabli 1989). They are well nodulated under drought stress conditions. Plants of the genus Acacia are pioneer plants, which play an important role for preservation and fertility of poor and eroded soils in Africa. These legumes produce extensive, deep root system, in addition to their potential to fix atmospheric N2 (Nabli 1989). In Tunisia, Acacia tortilis ssp. raddiana is the only wild and native acacia, which grow spontaneously in arid and Saharan areas. This acacia is generally overused by local people, who have no other fuel wood. Regeneration of the tree is made difficult owing to overgrazing, which adds its limiting effects to those of aridity (Nabli 1989). The approach adopted to improve rehabilitation programs and soil fertility is to isolate rhizobia from these areas and obtain data on their diversity and nodulation efficiency.

Among the techniques developed to detect DNA polymorphisms in many different organisms including bacteria, PCR-RFLP of 16S rRNA genes is one of the quickest and easiest (Laguerre et al. 1994). This genomic technique has been applied for characterizing rhizobia at the level of species and genera (Laguerre et al. 1994; Lafay and Burdon 2001). The 16S rDNA sequence analysis, which is a conserved gene, supports the well-established subdivision of rhizobia into species and genera (Young and Haukka 1996). However, DNA sequences in the 16S-23S spacer are more discriminating and known to exhibit a great deal of sequence and length variation (Normand et al. 1996). These variations are used to differentiate genera, species, and strains of prokaryotes (Normand et al. 1996).

Very little is known about the bacterial symbionts of Acacia spp., especially A. tortilis in Tunisia (Ba et al. 2002). Our objective was to re-examine the genetic diversity among rhizobia isolated from root nodules of A. tortilis ssp. raddiana and to find their phylogenetic positions within the family Rhizobiaceae by using 16S and IGS rDNA analysis.

Materials and methods

Bacterial strains and growth conditions

Strains of rhizobia were isolated from root nodules of A. tortilis grown in different regions in Tunisia. The collected nodules were kept in closed containers over silica gel at room temperature until their isolation in the laboratory. Forty isolates were obtained from nine different locations in Tunisia (Fig. 1).

Figure 1.

Map of Tunisia showing the sites prospected for nodulation of Acacia tortilis ssp. raddiana. 1, Ariana; 2, Gabes; 3, Belkhir; 4–9, Bouhedma sites.

Isolations were made according to the procedure described by Vincent (1970) using yeast-mannitol agar (YMA) supplemented with crystal violet (Collins and Lyne 1985), which inhibited the growth of gram positive contaminating-bacteria. Cultures used for further study were purified from single colonies on YMA agar plates pure cultures (or as the dominant colony type) after incubation at 28°C. All the rhizobia were maintained on YMA medium (Vincent 1970) and stored with 25% glycerol at −80°C.

Plant nodulation test and symbiotic efficiency

All isolates were tested for nodulation ability on A. tortilis-plants. The Acacia seeds were surface sterilized with concentrated sulphuric acid (H2SO4) for 90 min and rinsed with sterile water. The seeds were germinated for 72 h on semi-solid agar medium (8 g l−1) in a growth chamber at 28°C, and the seedlings were placed aseptically in Gibson tubes supplemented with a nitrogen free plant nutrient solution (Gibson 1980). Each tube was inoculated with a rhizobial suspension from an early stationary-phase culture. Uninoculated plants were used as controls. Three replicates were prepared to each treatment. After a month, the plants were harvested and the number of nodules was estimated. The nitrogen fixation ability of the strains was estimated from the pink colour of the nodules and the dark green colour of leaves compared to control plants that were not inoculated.

Morphological tests

Colony shape and colour were determined by using a magnifying glass. Cell dimensions and morphology were examined on living cells by phase-contrast microscopy, in order to confirm strain purity.

16S RNA gene amplification

The prokaryotic specific primers used for 16S rRNA gene amplification were fD1 (5′-AGAGTTTGATCCTGGCTCAG-3′) and rD1 (5′-AAGCTTAAGGTGATCCAG-CC-3′) (Weisburg et al. 1991). PCR amplification was carried out in a 25 μl reaction volume-containing template DNA (10–50 ng), reaction buffer, 1x freeze-dried marble (Ready-to-go PCR beads, Pharmacia Biotech) containing 1·5 U Taq polymerase, 10 mmol l−1 Tris–HCl, 50 mmol l−1 KCl, 1·5 mmol l−1 MgCl2, 0·2 mmol l−1 dNTP and 1 μmol l−1 of each of the primers. PCR amplification was performed with a Perkin–Elmer model (GeneAmp PCR System 2400). The PCR temperature profile used was 95°C for 5 min followed by 35 cycles consisting of 95°C for 30 s, 50°C for 1 min, 72°C for 1 min, with a final extension step at 72°C for 1 min. Reaction efficiency was estimated by horizontal agarose-gel electrophoresis (1% w/v) and coloured in an aqueous solution of ethidium bromide (1 mg ml−1).

16S-23S spacer amplification

Primers FGPS1490 (Navarro et al. 1992) and FGPL132′ (Ponsonnet and Nesme 1994) were used to amplify the IGS regions. The conditions for 16S-23S intergenic region amplification were the same as those used for 16S gene amplification, except that the annealing steps took place at 55°C.

Restriction fragment analysis

Eight microliter aliquots of PCR products were digested with restriction endonucleases (Pharmacia Biotech) as specified by the manufacturer in a total volume of 20 μl. Digestion was performed by four enzymes (HaeIII, MspI, CfoI and RsaI) for 16S rDNA and two enzymes (HaeIII and MspI) for 16S-23S spacer, used for their highly level of discrimination (Laguerre et al. 1994). The restriction fragments were separated by horizontal electrophoresis in TBE buffer with 2·5% (w/v) MetaPhor (FMC Bioproducts, Rockland, Maine, USA).

16S rRNA gene sequencing

The 16S rRNA gene of the isolate A1, chosen as a representative strain, was amplified. The amplified fragments were purified with the QIAEX II kit (Qiagen Inc., Chatsworth, CA, USA) and sequenced by the dideoxy chain termination method of Sanger et al. (1977). The six primers used for full sequencing of the 16S rRNA gene were fD1, FGPS485-292, FGPS1047-295, FGPS505′-313, FGPS910′-270 and rD1 (Weisburg et al. 1991).

DNA sequence analysis

The 16S rDNA sequences were aligned and analysed using Clustal X software. Phylogenetic analysis was inferred by using the neighbour-joining method (Saitou and Nei 1987) calculated by the Kimura method (Kimura 1980). The resulting tree was drawn with the Njplot software of Perrière and Gouy (1996).

Nucleotide sequence accession number

The newly determined 16S rDNA sequence was deposited in the GenBank Data Library under accession number DQ092342.


Isolation and morphological characteristics

Forty strains (Table 1) were isolated from root nodules of A. tortilis spp. raddiana, the only native acacia tree in Tunisia. Approximately half of rhizobial isolates had the same colony morphology and growth rate on YMA medium. They were fast growing rhizobia and formed transparent to creamy colonies with 2–4 mm in diameter after 3 days incubation on Petri YMA plates.

Table 1. Acacia tortilis strains used in this study
Geographic origin and climateIsolate16S groups (PCR-RFLP)IGS groups (PCR-RFLP)16S rDNA analysisNodulation test and NFA
  1. A, Acacia; E, Ensifer; NI, not identified; NFA, nitrogen fixation ability; E, effective nodulation; I, ineffective nodulation; NN, no nodule formation; PN, occasional formation of pseudonodules.

Ariana, humidA1215Ensifer meliloti-likeE
A1315E. meliloti-likeE
A622Rhizobium tumefaciensNN
A23111E. meliloti-likeE
A5415E. meliloti-likeE
A5115E. meliloti-likeE
A34418Rhizobium tropicii-IIAPN
A522R. tumefaciensNN
A25413R. tropicii-IIANN
Bouhedma-S1, aridA26814NIPN
A22410R. tropicii-IIANN
A1013E. meliloti-likeE
Bouhedma-S2, aridA111E. meliloti-likeE
A211E. meliloti-likeE
A311E. meliloti-likeE
A5211E. meliloti-likeE
Bouhedma-S3, aridA46519NIPN
A33418R. tropicii-IIAPN
A1928R. tumefaciensNN
A2028R. tumefaciensNN
Bouhedma-S4, aridA32717NIPN
A28116E. meliloti-likeNN
Bouhedma-S5, aridA2119E. meliloti-likeNN
A1828R. tumefaciensNN
A35418R. tropicii-IIANN
A813E. meliloti-likeE
A915E. meliloti-likeE
A1415E. meliloti-likeE
A713E. meliloti-likeE
A5315E. meliloti-likeE
Bouhedma-S6, aridA1114E. meliloti-likeNN
Gabes, aridA1727R. tumefaciensNN
A411E. meliloti-likeE
Belkhir, aridA24412R. tropicii-IIAI

Plant infection test and symbiotic efficiency

We tested the infectivity and symbiotic efficiency of 40 strains of rhizobia on A. tortilis spp. raddiana. Only half of the strains could induce root nodules on their plant host, appearing after 2 weeks of inoculation with diameters of 1–4 mm (Table 1). About 42·5% of the inoculated strains give pink-nodules and the leaves of the corresponding plants were dark-green, while the control noninoculated plants were yellow–green.

PCR-RFLP analysis of amplified 16S rDNA genes

The 16S rDNA were amplified and we have obtained a single band of about 1500 bp for all the strains.

Aliquots of PCR products were digested with four restrictions enzymes and separated by electrophoresis. The length of PCR product estimated by summing the sizes of the restricted fragments was shorter than or equal to 1500 bp. The patterns of these isolates were compared, by computer-simulated RFLP analysis, with those of reference strains published sequences (Neyra et al. 1998). A total of nine different combinations, each corresponding to one genotype, were identified among the 40 strains analysed by RFLP in this study (Table 1). Three groups correspond to reference species and six groups did not match with any of the reference species into the family of Rhizobiaceae. Results obtained showed that the majority of the strains showed restriction profiles closer to those of Ensifer meliloti species (47·5%). About 15% of the strains showed restriction patterns identical to those of Rhizobium tropici-IIA. 15% of the strains were related to Rhizobium tumefaciens species. Finally, they were 22·5% of the isolates which were not identified and remained unclassified.

16S-23S spacer analysis

To investigate further the genetic differences among the 40 rhizobia isolated from A. tortilis ssp. raddiana in humid and arid areas of Tunisia, we analyse the 16S-23S spacer by PCR/RFLP. The electrophoresis of the undigested PCR products showed that the majority of the isolates possess one band. The length of the IGS amplified region was between 1000 and 1350 bp (Fig. 2). Therefore, we were able to identify 18 groups in this natural population of rhizobia.

Figure 2.

Electrophoresis of PCR products obtained with the universal primers FGPS1490-72 and FGPL 132′-38, which target the ribosomal IGS of Rhizobium strains that nodulate Acacia spp.

The amplification products were digested separately with endonucleases MspI and HaeIII for all the isolates (Fig. 3). The representative isolates within each cluster showed the same patterns, but different patterns were observed among different clusters. After comparing the restriction patterns obtained by the two endonucleases, we have obtained 20 groups (Table 1).

Figure 3.

Examples of RFLP of PCR-amplified 16S-23S rRNA genes digested with HaeIII and separated by electrophoresis in 2·5% (w/v) metaphor gel. Numbers 1, 5, 6, 13, 18 and 19 are IGS groups. M, molecular mass marker (100 bp ladder) from Pharmacia Biotech; the smallest band of the marker is 100 bp.

Sequence analysis

We sequenced full length of 16S rRNA gene for A1 strain, which belonged to the biggest group (group 1 of the 16S rDNA) of the collection. The sequence was aligned and compared with the 16S rDNA sequences of other members of the family Rhizobiaceae available in the GenBank database. Figure 4 is a dendrogram, which shows the phylogenetic relationships of these unclassified rhizobia and the previously named species of Rhizobiaceae. The tree showed that the majority of identified strains form a new subdivion inside the genera Ensifer, with E. meliloti being its nearest neighbour. This strain showed a 16S rDNA sequence similarity of 99·23% with E. meliloti, 99·22% with Ensifer medicae and 99·09% with Ensifer arboris type strains. DNA–DNA hybridization would be needed to verify whether these strains represent separate species or not.

Figure 4.

Phylogenetic tree based on 16S rDNA complete sequences of Acacia raddiana strains and references type strains of Rhizobiaceae. Numbers into brackets following strains names are Genbank accession numbers or strain designation. Bootstrap values were calculated from 1000 trees and the levels of support for the presence of nodes above a value of 600 are indicated.


In this research, we characterized 40 nodules isolates from A. tortilis ssp. raddiana in representative soils of Tunisia. We have used PCR/RFLP analysis of 16S rDNA to characterize these natural rhizobia. We found a high diversity among rhizobial strains. Our results corroborated several studies, which revealed a high heterogeneity in the populations of rhizobia nodulating Acacia spp. (Ndiaye 1996; De Lajudie et al. 1998; Khbaya et al. 1998; McInroy et al. 1999; Mohamed et al. 2000; Lafay and Burdon 2001; Odee et al. 2002; Toledo et al. 2003).

We have used PCR-RFLP of the 16S rDNA to characterize all the isolates of our collection. The results obtained have discriminated nine groups covering three genera; genus Ensifer was the most represented in our collection. To our knowledge, there are many data showing that a high proportion of tree-nodulating rhizobia in Africa, particularely Acacia spp., are more closely related to Ensifer species (Ndiaye 1996; Khbaya et al. 1998; Ba et al. 2002).

Full 16S rRNA gene-sequencing of the strain A1 which belong to the biggest group (group 1) have showed that the majority of the identified strains that nodulate A. tortilis ssp. raddiana in Tunisia belonged to the genus Ensifer within they can form new species. This strain showed a 16S rDNA sequence similarity of 99·23% with E. meliloti, 99·22% with E. medicae and 99·09% with E. arboris reference strains but occupied separate phylogenetic position within the Ensifer genus. Similar results were found by Zakhia et al. (2004) on rhizobia nodulating wild legumes in Tunisia. DNA–DNA hybridizations and GC% are needed to clarify their taxonomic status inside Rhizobiaceae family. Our results thus confirm and extend the large diversity of fast-growing A. tortilis-rhizobia within the Ensifer-Rhizobium branch.

Our results showed that the majority of rhizobial strains were related to E. meliloti-like species (47·5%) and only 15% of the strains were related to R. tropici-IIA. However, rhizobia nodulating A. tortilis in Morocco were related to E. meliloti and Ensifer fredii species (Khbaya et al. 1998). Ndiaye (1996) have characterized, by SDS-PAGE, a collection of A. tortilis-rhizobia from diverse countries in Africa and he had found that they were related to Ensifer teranga, E. fredii, R. tropici, Mesorhizobium huakuii and Mesorhizobium plurifarium species. Strains from Tunisia, Senegal and Mauritania were grouped into Ensifer terangae lineage and into two novels groups (Ndiaye 1996). Ba et al. (2002) studied eight Tunisian strains that were more related to E. medicae. However, De Lajudie et al. (2003) found that these rhizobia were related to E. terangae. Recently, Zakhia et al. (2004) investigated one A. tortilis strain and was found to be related to E. meliloti. This high diversity of A. tortilis-rhizobia in Tunisia seemed to be related to the soil-origin of plant host.

Our results together with past surveys of nodule bacteria in Tunisia (Aouani et al. 2001; Jebara et al. 2001; Mhamdi et al. 2002; Zakhia et al. 2004; Zribi et al. 2004) suggest that Tunisian soils are dominated by fast growing rhizobia, especially by E. meliloti and E. medicae within they can form new genomospecies.

Furthermore, 22·5% of the isolates had no identity with any of rhizobial strains, but some of them (7·5%) could induce nodule formation. In fact, unknown associations were previously found and several strains may represent new genomospecies nodulating wild legumes in Tunisia (Zakhia et al. 2004). This non-nodule formation can be explained by several hypothesis (i) host specificity of strains to plant genotype (Paffetti et al. 1998). Strains would be better tested on their original host plants. (ii) Nodulation test experiment, used here, was not optimal for these strains. Using sterilized gravel or soil will be tested to verify this hypothesis. (iii) These strains are only accidentally associated with A. tortilis but are in fact better adapted for nodulation of another legume host species. In fact, in the natural environment, A. tortilis grow intermingled with other leguminous plant (perennial or annual) that might utilize nonidentified-rhizobia (Nasr, personal communication). Surveys of how strains are distributed on hosts in natural environments, together with cross inoculation experiments using additional legume species, will be necessary to clarify patterns of symbiotic specificity in these organisms. (iv) These bacteria originally symbiotic have lost their unstable symbiotic gene during laboratory experiments (isolation, purification and conservation) (Laguerre et al. 1993). Sutherland et al. (2000) have found that all strains had lost infectivity during storage. The symbiotic plasmid has been previously shown to be unstable under laboratory conditions (Romero et al. 1991). (v) These strains have acquired symbiotic genes by lateral gene transfer, which were lost during laboratory experiments. Several papers reported that soils contain a large diversity of nonsymbiotic bacteria which can acquire symbiotic properties by lateral gene transfer between bacteria in the soil (Segovia et al. 1991; Sullivan et al. 1995, 1996).

Moreover, several new bacterial strains, belonging to α- β- and γ-proteobacteria, were isolated from the root nodules of tropical legumes (Benhizia et al. 2004; Young 2003).

Further studies will help to classify these symbiotic bacteria and to understand their phylogenetic relationships to other rhizobial species.

Our results showed that 15% of the isolates were related to R. tumefaciens species, but could not induce tumour formation. In fact, several Rhizobium strains lacking their pathogenic gene were isolated from root nodules of tropical legumes in Africa (De Lajudie et al. 1999; Khbaya et al. 1998) and cultivated legumes in Tunisia (Mhamdi et al. 2002, 2005). It seems likely that under soil conditions, R. tumefaciens behaves as a pathogen or a symbiont according to its plasmid content. The symbiotic state could be unstable under laboratory conditions (Mhamdi et al. 2002).

We have used PCR with RFLP analysis to analyse 16S-23S spacer variation among our rhizobial isolates. We found a high diversity in the length of the amplification bands among the rhizobium strains. Therefore, we were able to distinguish 18 groups on the basis of IGS length. Furthermore, restriction digestion of the amplified IGS allowed us to distinguish two other new patterns. A similar result was also found by Paffetti et al. (1996) when investigating several strains belonging to E. meliloti by using PCR with RFLP of the IGS. However, digestion of the amplified 16S-23S spacer with nine restriction enzymes did not allow Khbaya et al. (1998) to distinguish additional groups. PCR-RFLP of the IGS have discriminated 11 supplementary groups than obtained by PCR-RFLP of the 16S rDNA gene. Thereby, our results confirmed the higher discriminating level for the IGS than for the 16S rDNA (Navarro et al. 1992; Normand et al. 1996; Yu and Mohn 2001).

Distribution of rhizobial strains appeared to be independent to the site of origin and to site-climate. These results corroborated with those of Khbaya et al. (1998) and Ndiaye (1996).

Results obtained from nodulation test showed that 50% of the isolates failed to renodulate their host plant. Non-nodulating or erratic-in-nodulation Acacia species have been reported (Odee et al. 1995; Frioni et al. 1998; Mohamed et al. 2000). Most (80%) of our rhizobial strains, which could renodulate their host plant were related to E. meliloti species. Similar results were found by Ndiaye (1996) and Ba et al. (2002) on rhizobia nodulating A. tortilis in Africa.

A collection of A. tortilis spp. raddiana-nodulating bacteria has been characterized, and had revealed a high diversity. Some strains may represent new genomospecies to be further characterized to clarify their phylogenetic positions. These rhizobia can be used to develop efficient inoculants in order to restore acacia forest and then soil fertility in arid and Saharan regions. This study provides the first clear characterization and symbiotic efficiency of rhizobia strains nodulating A. tortilis in Tunisia.


This work was supported by Grant from projet de coopération avec l'Union Européenne MYRISME-CONTRAT ERBIC18 CT97 0197.