Correspondence: Antoine Galiana, CIRAD, UMR LSTM, Campus International de Baillarguet, TA A-82/J, 34398 Montpellier Cedex 5, France. Tel.: +33 4 67 59 38 51; fax: +33 4 67 59 38 02; e-mail: firstname.lastname@example.org
The diversity of rhizobia associated with introduced and native Acacia species in Algeria was investigated from soil samples collected across seven districts distributed in arid and semi-arid zones. The in vitro tolerances of rhizobial strains to NaCl and high temperature in pure culture varied greatly regardless of their geographical and host plant origins but were not correlated with the corresponding edaphoclimatic characteristics of the sampling sites, as clearly demonstrated by principal component analysis. Based on 16S rRNA gene sequence comparisons, the 48 new strains isolated were ranked into 10 phylogenetic groups representing five bacterial genera, namely, Ensifer,Mesorhizobium,Rhizobium,Bradyrhizobium, and Ochrobactrum. Acacia saligna, an introduced species, appeared as the most promiscuous host because it was efficiently nodulated with the widest diversity of rhizobia taxa including both fast-growing ones, Rhizobium,Ensifer, and Mesorhizobium, and slow-growing Bradyrhizobium. The five other Acacia species studied were associated with fast-growing bacterial taxa exclusively. No difference in efficiency was found between bacterial taxa isolated from a given Acacia species. The tolerances of strains to salinity and temperature remains to be tested in symbiosis with their host plants to select the most adapted Acacia sp.-LNB taxa associations for further revegetation programs.
Owing to their symbiotic association with rhizobia, nitrogen-fixing woody legumes play a key role in the sustainability of many natural ecosystems and agro-ecosystems, particularly in the dry and arid regions of tropical and sub-tropical zones (Brockwell et al., 2005). Many studies have investigated the nitrogen-fixing status of legume trees from natural forests (Moreira et al., 1998; Diabate et al., 2005; de Faria et al., 2010) and agroforestry systems (Bala et al., 2003; Wolde-Meskel et al., 2005; Wang et al., 2006), particularly dry-zone Acacia species from Africa (Njiti & Galiana, 1996; de Lajudie et al., 1998; Haukka et al., 1998; McInroy et al., 1999; Ba et al., 2002; Odee et al., 2002) and Australia (Marsudi et al., 1999; Lafay & Burdon, 2001). In the case of Algeria, where the Saharan bioclimatic zone covers 89.5% of the total area in addition to 4.78% and 4.12% of arid and semi-arid regions, respectively (Nedjraoui, 2001), Acacia species appear as suitable candidates for thoughtful reforestation programs, with regard to their high resistance to salinity and drought (Brockwell et al., 2005) jointly with their ability to regenerate, stabilize, and fertilize soils through their nitrogen-fixing (Galiana et al., 2002, 2004) and mycorrhizal symbioses (André et al., 2005; Duponnois et al., 2007). In addition, they are considered as excellent sources of fuel, service wood, fodder, tannin, and medicines (Midgley & Bond, 2001). Acacia is a member of the Mimosoideae subfamily and is represented by about 150 species in Africa and 1100 species in Australasia (Maslin et al., 2003). The native Acacia species inventoried in Algeria, mostly distributed in northern and central Sahara, are A. ehrenbergiana, A. laeta, A. nilotica, A. seyal, and A. tortilis (Ozenda, 2004). Acacia species have also been introduced in Algeria, mostly along coastal areas such as Acacia saligna for dune stabilization and A. karroo that has been essentially planted as living hedges (El-Lakany, 1987), both species being exotics native to Australia and southern Africa, respectively. However, the distribution of Acacia species in Algeria remains poorly documented, and a precise cartography is still needed. Similarly, natural rhizobial symbionts of Acacia species have never been inventoried in Algeria so far except the recent publication on A. saligna by Amrani et al. (2010) restricted to bacterial populations from nursery soil. In their natural distribution area, most of the African Acacia spp. are associated with Mesorhizobium (de Lajudie et al., 1998; Haukka et al., 1998; McInroy et al., 1999; Ba et al., 2002; Odee et al., 2002), Ensifer (syn. Sinorhizobium) (de Lajudie et al., 1994; Lortet et al., 1996; Haukka et al., 1998; Khbaya et al., 1998; Nick et al., 1999b; Ba et al., 2002) and, to a lesser extent, Rhizobium (McInroy et al., 1999; Nick et al., 1999a) and Bradyrhizobium (Dupuy et al., 1994; McInroy et al., 1999; Odee et al., 2002). On the other hand, the Australian acacias are mainly associated with Bradyrhizobium (Galiana et al., 1990, 1994; Lafay & Burdon, 2001; Rodriguez-Echeverria et al., 2007; Le Roux et al., 2009; Perrineau et al., 2011). There is a major interest to inventory the biodiversity of introduced and native acacias and their associated rhizobia in Algeria to allow for effective selection of both symbiotic partners for use in reforestation programs. For this purpose, and as reported by Zhang et al. (1991) and Odee et al. (2002) in other arid and semi-arid regions, there is a need for inoculants that are able to survive to the stressful edaphic conditions found in Algeria. Recent field experiments performed on coastal dunes of West Algeria showed that the inoculation at nursery level of A. saligna seedlings with native rhizobia improved the survival of trees several months after transplanting (unpublished data). However, and this applies to the other Acacia species found in Algeria, the beneficial effect of inoculation on tree establishment and growth in field conditions requires the selection of indigenous rhizobia that are adapted and tolerant to high salinity and temperatures.
The objective of this study was thus to characterize the bacteria associated with Acacia spp. in their natural distribution area in Algeria at three different levels: (1) their in vitro tolerances to NaCl and high temperatures to find out putative relationships between these features and the corresponding characteristics of the sampling sites, (2) their molecular diversity based on 16S rRNA gene sequence analysis, and (3) testing the efficiency of the bacterial strains toward their host plant of isolation to highlight possible specificities among Acacia sp.-LNB (Legume-Nodulating Bacteria, Zakhia et al., 2004) taxa associations.
Materials and methods
Soil sampling and sites characteristics
Soil samples were collected from 13 sites distributed in seven districts (Wilayas) in Algeria, within 50 cm around the Acacia trunks and at 20, 40, and 60 cm depth (Table 1, Fig. 1). The sampled sites were distributed from arid to semi-arid zones, with rainfall below 100 mm year−1 for the Saharan regions. Electrical conductivity (CW) and pH of soil samples were measured using a conductivimeter, in fivefold diluted soil stirred during 30 min at 25 °C with five repetitions. Nodules and seeds were collected when present.
Table 1. Geographical origin, ecological characteristics, and properties of sampled soils
Site prospected (Wilaya)
Geographical position (Latitude, longitude, altitude)
Isolation of bacterial strains, culture conditions, and symbiotic status
Most of the root nodules used for bacterial isolation were obtained through a preliminary step of trapping from the collected soil samples, using seeds from the different tree species studied so that each soil sample was used with its associated Acacia species. Direct bacterial isolation from fresh nodules collected in situ was also performed but restricted to all the isolates originating from coastal dunes sites, that is, Bomo-Plage and El Mactaa (Table 2). Seeds were collected at the same time as the soil samples from: (1) five tree species native to Algeria, namely Acacia ehrenbergiana Hayne, Acacia nilotica (L.) Delile, Acacia seyal Delile, Acacia tortilis (Forssk.) Hayne, and Faidherbia albida (Delile) A. Chev., and (2) two introduced species originating from the Oran region, namely Acacia karroo Hayne and Acacia saligna (Labill.) Wendl. Scarification and surface sterilization of seeds were achieved by treatment with concentrated (95%) H2SO4 for 30, 30, 120, 30, and 120 min for the native species, respectively, and 30 and 90 min for the introduced ones, respectively. Seeds were then rinsed thoroughly with sterile distilled water and germinated on 0.8% (w/v) water agar at 28 °C for 2–5 days. For bacterial trapping, the seeds were then transferred in aseptic conditions into Gibson tubes (Gibson, 1980) containing sterile Jensen nitrogen-free medium (Vincent, 1970). Plants were grown in an incubation chamber as described by Diouf et al. (2007). After 1 week of growth, 1 mL of stirred soil suspension was added to each tube. Soil suspension was obtained with 10 g of each soil sample added to 90 mL of sterile buffered saline pH 7 (NaCl 0.15 M, KH2PO4 0.002 M, Na2HPO4 0.004 M) shaken for 1 h. Four replicates were tested for each soil sample. Uninoculated plants served as controls. Plants were observed for nodule formation 6 weeks after germination, and fresh nodules were collected and conserved under desiccated condition in tubes containing CaCl2 crystals. The root nodules were washed by immersion in 70% (v/v) ethanol for 30 s, rinsed with distilled water, then rehydrated for 1 h, individually surface-sterilized by immersion in 35% H2O2 for 15 s to 2 min according to their size, and then rinsed in successive baths of sterile distilled water. After crushing, the nodule homogenates were spread onto yeast extract-mannitol agar (YMA) plates (Vincent, 1970) containing Congo red (0.00125%, w/v). One isolated colony per nodule homogenate was obtained from successive streaking on YMA plates, and purity of colonies was checked by phase-contrast microscopy. Colony shape and color were determined using a magnifying glass. Pure cultures were then cryogenically preserved (−80 °C) in yeast extract-mannitol medium (YEM) adjusted to 30% (v/v) glycerol. The nodulation ability of the bacterial isolates obtained was checked by inoculating seedlings of the corresponding Acacia species grown on sterilized sandy soil (autoclaving at 120 °C for 20 min, three times at 24-h intervals) in 250-mL plastic pots. One week after germination, three plants per Acacia species were inoculated with approximately 2 mL of bacterial suspension at 109 cells mL−1 of each isolate. Uninoculated controls were included. Plants were transferred to greenhouse and fed with a sterilized Broughton & Dilworth N-free nutrient solution (1971) once a week. Five weeks after inoculation, plants were harvested to estimate the infectivity and effectivity of bacterial isolates based on the presence of nodules, nodule color, visual observation of plant vigor, and foliage color. The bacterial isolates were recorded as noninfective (0), symbiotically ineffective (I) when nodulated plants were not more developed than uninoculated control plants and nodules had a white color, and effective (E) when plants were greener and significantly more developed than uninoculated plants while having pink nodules.
Table 2. Origin, phenotypic and genotypic characteristics of new strains
Tolerances of bacterial isolates to salinity and temperature
All tests were carried out in triplicate. Before inoculation, isolates were grown in YEM liquid medium to log phase (corresponding to approximately 109 colony-forming units mL−1). In test plates, inoculation was performed with 10 μL of these cultures. The results were scored after 7-day incubation. Positive results were reported when bacterial colonies were observed under binocular microscope at ×50 magnification. Tolerance to salinity was tested on YEM agar plates containing Nacl concentrations of 170, 340, 510, 680, 860, and 1034 mM incubated at 28 °C. For temperature tolerance, YMA plates were inoculated and incubated at four different temperatures: 35, 40, 45, and 50 °C.
Partial 16S rRNA gene sequencing
Isolated colonies were homogenized into sterile 1.5-mL centrifuge tubes with 50 μL of sterile Milli-Q water; suspensions were centrifuged at 6000 g for 5 min and washed twice, pellets were re-suspended and then stored at −20 °C until PCR use. The universal eubacterial 16S rRNA gene primers FGPS 6 (5′-GGA GAG TTA GAT CTT GGC TCA G-3′) and FGPS 1509 (5′-AAG GAG GGG ATC CAG CCG CA-3′) (Normand et al., 1992) were used, and PCR was performed as follows: an initial cycle of denaturation at 94 °C for 5 min, 36 cycles of denaturation at 94 °C for 30 s, annealing at 56 °C for 30 s, extension at 72 °C for 2 min, and a final extension at 72 °C for 7 min. The PCR products were checked by 0.8% (w/v) agarose gel electrophoresis in Tris/acetate/EDTA (TAE) buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.3). Bands were excised and DNA purified using a QIAquick gel extraction kit (Qiagen, Courtaboeuf, France) and sequenced using primers FGPS6, FGPS1509, and 16S-1080r (5′-GGG ACT TAA CCC AAC ATC T-3′) (Sy et al., 2001). Sequencing reactions were analysed on an Applied Biosystems model 310 DNA automated sequencer using a BigDye Terminator Cycle Sequencing Kit (Perkin Elmer Applied Biosystems). Sequences were corrected using Chromaspro 4 (Technelysium Pty Ltd., Tewantin, Australia) before alignment and analysis of the 16S rRNA gene sequences using Clustal X software (Thompson et al., 1997). Phylogenetic analysis was inferred using the neighbor-joining method (Saitou & Nei, 1987) calculated by the Kimura method (Kimura, 1980). The 16S rRNA gene sequences of the type strains of the various genera used in this study were retrieved from the GenBank/EMBL database and used for cladistic analysis. A bootstrap analysis using 1000 replications was performed. The 16S rRNA gene sequences of 48 new isolated Acacia strains, of which 39 were included in the phylogenetic tree, were deposited in the GenBank database under accession numbers HQ836154 to HQ836191, HQ853452, HQ853453, and JF810499 to JF810506 (see Table 2).
Efficiency of bacterial isolates
The most efficient strains identified in the nodulation tests were compared to highlight a possible relationship between strain efficiency and its taxonomic position. As a minimum number of strains was needed for each Acacia species to be tested, only three Acacia species were tested in this experiment, namely A. seyal, A. karroo, and A. saligna. The test was conducted in the same conditions as those used in the nodulation test described earlier, using triplicates, except that plants were harvested here 7 weeks after inoculation for measurements of shoot height and shoot dry weight.
Principal component analysis (PCA) was performed to examine the relationships between soil conductivity and maximal air temperature of the sampled sites (Table 1) and the in vitro tolerances of bacterial strains isolated from the corresponding sites to NaCl and temperature. Computations and graphical display were made using the xlstatTM software package (version 2010.5.04, Addinsoft, Paris, France, http://www.xlstat.com). Factorial correspondence analysis (FCA) was performed to visualize the relationships between rhizobial taxa, as defined from the partial 16S rRNA gene-based phylogeny, and Acacia host species, using xlstatTM software package (see reference above). In the strain efficiency tests, and for each of the three Acacia species studied and each parameter analyzed, all the data collected were subjected to a one-way analysis of variance. When the rhizobium strain factor had a significant effect on a given parameter, the means of the different treatments were ranked into homogeneous groups according to the Duncan's multiple range test (Dagnélie, 1975). The variance analysis was performed using SuperANOVA software (Abacus concepts Inc., version 1989, CA).
Symbiotic status of isolated strains
Of 288 purified bacterial isolates from surface-sterilized nodules, only 84, that is, 29%, were able to renodulate their host plant of isolation. Among these infective isolates, 48 were further characterized molecularly (Table 2). These 48 isolates were selected based on colony morphology, only one representative per morphological type and per origin (host plant, sampling site, and trapping plant), being retained from the 84 nodulating isolates for molecular analysis and estimation of symbiotic status. Five weeks after inoculation, 25 of these 48 strains were found to be efficient, as estimated by qualitative observation based on nodule color, plant vigor, and foliage color in comparison with uninoculated control plants. In these preliminary nodulation tests, the percentage of effective strains varied according to the Acacia species because, among the most represented ones, 100% of the infective strains were effective in A. seyal (10 of 10 strains tested) vs. 55% in A. karroo (six of 11) and only 33% in A. saligna (seven of 21).
Tolerances of bacterial isolates to salinity and temperature
As reported in Table 2, the maximum salinity levels tested varied greatly according to the isolate, that is, from 170 to 1034 mM NaCl. About 20% of the isolates were able to grow at the maximum level tested of 1034 mM NaCl. However, these highly tolerant bacteria were mostly isolated from nonsaline sites except one from Es-Senia (Oran) region that is slightly saline. Isolates from saline sites, especially those from Oran Sebkha, were not particularly tolerant to NaCl under in vitro conditions. In the same way, the in vitro NaCl tolerance of strains originating from the same site varied greatly. Thus, as clearly confirmed by PCA (Fig. 2), no relationship was found between in vitro tolerances of rhizobial isolates to temperature and NaCl, and the edaphoclimatic characteristics of the corresponding sampling sites. Indeed, as shown by direction of variables axes in Fig. 2, no correlation was found between in vitro NaCl tolerance of strains and soil conductivity (r = −0.004) or between in vitro tolerance of strains to high temperatures and maximal air temperature of their original site (r = −0.162). The main contributors to the structure of this PCA biplot, where both F1 and F2 axes represent 60.22% of the total variability, are bacterial isolates from Sebkha Messerghine/Oran (indicated by ‘O3’ symbols), which are logically grouped along ‘Soil CW’ and F1 axes, and those from Aïn Belbel/Adrar (indicated by ‘A’ symbols), grouped along ‘Site Tpre’ axis, both sampling sites being characterized by high NaCl soil concentrations and extreme high temperatures, respectively. Despite their proximity in Fig. 2, ‘Soil CW’ and ‘Tpre tolerance’ variables were slightly correlated (r = 0.247) because F3 axis, representing 21% of the total variability, was not represented. In the same way, the maximum growth temperature varied widely according to the isolate, ranging from 35 to 50 °C with an average of about 40 °C. About 35% of the strains were tolerant to 45 °C while two of them reached a maximal growth temperature of 50 °C. These two latter outstanding strains isolated from A. ehrenbergiana and A. seyal originated from Tamanrasset and from the Sebkha region (Oran), respectively. However, as clearly shown by the PCA, there was no correlation between the tolerance of bacterial strains to high temperatures in pure culture and the maximal temperatures found in their regions of origin. Thus, none of the strains isolated from the hottest regions of Algeria, Adrar in particular, exhibited any particular tolerance to high temperatures in pure culture in comparison with strains from other cooler regions.
Phylogenetic analysis of bacterial strains
Based on 16S rRNA gene sequence comparisons including reference bacterial sequences, we distinguished 10 phylogenetic groups (Fig. 3) among the 48 new strains studied (Table 2), originating from seven Acacia species and 13 different sites in Algeria. For the sake of clarity, only a limited dendrogram showing the new strains among their closest reference species is presented. Some of the 48 nodulating isolates sequenced were not indicated in the phylogenetic tree because their 16S rRNA gene sequences had 100% similarity with those of representative strains, while only one representative strain per sequence was indicated in the tree (see details in legend of Fig. 3). The 10 groups obtained might be affiliated to five genera: Ensifer, Mesorhizobium, Rhizobium, Bradyrhizobium, and Ochrobactrum. Group 1, closely related to the Ensifer meliloti reference strain, included eight isolates from three Acacia species, namely A. karroo (K1a, K2b, K4a, and K4c), A. saligna (SAB13, SAB17a, and SAB17b), and A. ehrenbergiana (E231c). In this group 1, the four A. karroo strains were isolated from the same geographical origin, that is, Msila forest (Oran), and the three A. saligna strains all originated from Bomo-plage dunes (Oran). The heterogenous group 2 contained four isolates from four different host species, namely A. karroo, A. ehrenbergiana, A. saligna, and A. tortilis, all from different geographical origins that were close to E. fredii (K2d), E. terangae (E60), and E. kostiense (SAKS and T82) reference strains. Isolates in Mesorhizobium distinguished as group 3, a separate sub-branch consisting of two isolates from A. ehrenbergiana (E231a) and F. albida (A121) and group 4, represented by a single isolate from A. saligna (SAB10), in the vicinity of M. mediterraneum. Groups 5, 6, 7, and 8 were affiliated to Rhizobium. Groups 5 and 7 represented single isolates, that is, one from A. saligna (SAAIS) affiliated to the R. tropici clade and the other one from A. seyal (SE3) affiliated to the R. sullae clade, respectively. On the other hand, group 6, consisting of five bacterial isolates, all from A. saligna (SAB3, SAB5, SAB9, SAB12b, and SAB15b) and from the same geographical origin (Bomo-plage Oran), formed a separate clade in the vicinity of the R. galegae–R. huautlense–R. loessense branch. Group 8 containing the R. leguminosarum reference strain LMG14904 was represented by five A. karroo isolates (K31, K32a, K32c, K33b, and K34a) from the same origin (Messerghine Farm-Oran) and five A. seyal isolates (SE23bc, SE25d, SE21a, SE24a, and SE25a) from another origin (Messerghine Sebkha-Oran). In group 9, two A. seyal strains (SE243a and SE243d) isolated from the same site (Labiod-El Bayadh) were identified as Ochrobactrum sp. Lastly, 10 A. saligna isolates originating from two different sites, Bomo-plage dune Oran (SABN1a, SABN1b, SABN1c, SABN2a, SABN2c, SABN4a, SABN4b, and SABN5a) and El Mactaa dune Mostaganem (SA13b and SA15d), were affiliated to Bradyrhizobium and ranked as group 10.
Host specificity of isolated LNB
The FCA showed that each Acacia species had its own specificity in terms of preferential association with taxonomical groups of nodulating bacteria as defined by the phylogenetic analysis mentioned earlier. As shown in Fig. 4, the first factorial plan projection including F1 and F2 axes, which represents about 85% of the total variability, showed that A. saligna was nodulated by the highest diversity of taxonomical groups including four single host taxa, namely, M. mediterraneum, R. tropici, Rhizobium sp., and Bradyrhizobium sp., as well as a multiple host range taxon, namely, E. meliloti, shared with A. karroo and A. ehrenbergiana. In addition to E. meliloti, A. karroo and A. ehrenbergiana were both nodulated by E. fredii while these two host plant species differed in their respective association with R. leguminosarum and Mesorhizobium sp. Lastly, A. seyal was associated with R. leguminosarum and two single host taxa Ochrobactrum sp. and R. sullae. The other Acacia host species prospected were not taken into account in this FCA analysis considering the low number of isolates obtained from each of them.
Efficiency of the isolated LNB according to taxonomic affiliation
Strain efficiency tests were performed using a selection of isolates representative of the bacterial taxa found in association with each Acacia species. As noted earlier, some of the bacterial strains tested, namely, SE243c, SE21b, SE24a, SE25a, K1b, K34a, K33b, and K32c (Tables 3a–c), were excluded from the phylogenetic tree because their 16S rRNA gene sequences were identical to those of other strains already represented in the tree (see details in legend of Fig. 3). Moreover, these experiments were focused on Acacia species that had a sufficient number of isolates available. Seven weeks after inoculation and growth, all plants of the three Acacia species studied were nodulated for all bacteria isolates tested. All the bacteria tested were efficient in A. seyal and A. saligna (Tables 3a and c, respectively) while a few nodulating strains were not effective in A. karroo (i.e. K31, K32c, and K4c) as shown by the absence of significant difference in height and dry weight of shoots between inoculated and uninoculated control plants (Table 3b). In A. seyal, the most effective isolate in terms of shoot growth was a Rhizobium sp. strain (SE243c), followed by one Ochrobactrum sp. strain (SE243a), while another Ochrobactrum sp. strain (SE243d) along with four R. leguminosarum strains (SE21a, SE21b, SE24a, and SE25a) were ranked in a same group according to the Duncan multiple range test at P = 0.05 when shoot dry weight was only considered. The lowest dry weight was observed for a strain of R. leguminosarum (SE25d). In A. saligna, the seven strains tested had statistically similar growth whatever their taxonomic position and geographical origin. In A. karroo, the most effective isolate was an E. fredii strain (K2d), for which inoculated plants had a shoot dry weight about three times higher than uninoculated control plants, followed by a very efficient R. leguminosarum strain (K34a) and a third group of efficiency – as defined by the Duncan multiple range test applied to shoot dry weight mean at P = 0.05 – containing three E. meliloti and one R. leguminosarum strains (K1b, K1a, K2b, and K33b, respectively). On this same Acacia species, an E. meliloti isolate was poorly effective (K4a) while two R. leguminosarum (K32c and K31) and one E. meliloti (K4c) strains were totally inefficient.
Table 3. Efficiency of a selection of new strains on (a) Acacia seyal, (b) Acacia karroo and (c) Acacia saligna grown under greenhouse conditions, 7 weeks after plant inoculation
Bacterial species was determined based on the positioning of a given strain in the phylogenetic tree (Fig. 2).
Means were obtained from tree replicates per strain tested. Means followed by different letters in a same column are significantly different according to the Duncan multiple range test at P =0.05.
The in vitro tolerance of rhizobial strains to NaCl in pure culture varied greatly regardless of their geographical and host plant origins but was not correlated with the salinity of the soils from which they were isolated. However, soil salinity was assessed through soil conductivity (CW) that takes into account all types of salts while NaCl might not be dominant in soils of the different sampled sites, even though the latter is the major soluble salt in highly saline soils throughout the world (FAO, 1976). The absence of correlation between laboratory assay and isolation conditions could also be due to the fact that rhizobia are located in microniches at rhizospheric level and inside nodules and are thus protected from soil constraints (Trabelsi et al., 2010). Overall, the in vitro tolerance of strains to salinity was quite high because about 20% of them grew at the maximum level tested of 1034 mM NaCl (corresponding to 6% NaCl w/v). Our findings are in agreement with previous studies reporting that rhizobia isolated from woody legumes in general and from dry-zone Acacia species in particular tolerated such high salinity levels (Zahran, 1999; Mohamed et al., 2000; Diouf et al., 2007; Essendoubi et al., 2007). In accordance with previous reports (Zahran, 1999), we found Bradyrhizobium strains less tolerant, although originating from saline ecosystems in our study (coastal dunes), than fast-growing Rhizobium, Ensifer, and Ochrobactrum strains that tolerated up to 1034 mM NaCl. In the same way, while the PCA showed that there was no correlation between the tolerance of bacterial strains to high temperatures in pure culture and the maximal temperatures found in their regions of origin, the new isolated strains exhibited a quite high tolerance to high temperatures with about one-third of isolates tolerant to 45 °C and even two strains that grew at 50 °C. There was no link either between the tolerance of rhizobial strains to high temperatures and their taxonomic position because all LNB taxa were indiscriminately represented among these highly tolerant strains. Such high proportion of strains tolerant to high temperatures, as found in our study, is unusual in rhizobial populations isolated from arid-zone tree species (Zhang et al., 1991; Swelim et al., 1997; Zerhari et al., 2000).
The affiliation of strains associated with introduced and native Acacia spp. in Africa to Rhizobium, Agrobacterium, Ensifer, Mesorhizobium, and Bradyrhizobium is well documented (de Lajudie et al., 1994, 1998; Khbaya et al., 1998; McInroy et al., 1999; Odee et al., 2002; Diouf et al., 2007, 2010; Amrani et al., 2010). In this study, A. seyal was found to be associated with R. leguminosarum, R. sullae, and Ochrobactrum lineages, in contrast to previous studies that reported that A. seyal was preferentially associated with Mesorhizobium in Senegal (de Lajudie et al., 1998; Diouf et al., 2007, 2010) and to Ensifer spp. in Ethiopia (Wolde-Meskel et al., 2005). To date, only one study performed on A. mangium has reported the ability of Ochrobactrum (Brucellaceae family) to form nitrogen-fixing nodules in legume trees (Ngom et al., 2004). More recent studies also reported the isolation of infective and effective Ochrobactrum strains in annual and shrubby legume species (Trujillo et al., 2005; Zurdo-Piñeiro et al., 2007; Imran et al., 2010). In our study, symbionts of A. karroo were identified as E. meliloti, E. fredii, and R. leguminosarum while this Acacia species was reported to be associated with R. tropici in Kenya (McInroy et al., 1999), S. terangae biovar acaciae in Senegal (de Lajudie et al., 1994), but E. fredii in Morocco (Khbaya et al., 1998). Similarly to A. seyal and A. karroo, A. ehrenbergiana was shown to be only associated with fast-growing rhizobia. Although only a few strains were isolated from this latter species, they were quite diverse because they belonged to three well distinct taxa, namely, E. meliloti, E. terangae, and Mesorhizobium sp. Symbiotic status of this species remains to be assessed because rhizobium diversity related to this Acacia species has been poorly studied. Among other poorly represented species, A. tortilis was found associated with Ensifer strains as in other studies, that is, E. fredii in Ethiopia (Wolde-Meskel et al., 2005), E. meliloti in Morocco (Khbaya et al., 1998), and several Ensifer spp. in Tunisia (Romdhane et al., 2006), while F. albida, known to be widely nodulated by slow-growing rhizobia, associated with fast-growing Mesorhizobium, as already reported (Odee et al., 2002). Acacia saligna appeared as the most promiscuous host plant among all Acacia species investigated here considering its ability to nodulate with the widest diversity of rhizobial taxa, including both fast-growing ones, namely, Rhizobium, Ensifer, and Mesorhizobium, and the slow-growing Bradyrhizobium. Such promiscuity has rarely been reported in other legume species. Other studies also showed that A. saligna was nodulated either by fast- and slow-growing rhizobia, either with a predominance of slow-growing ones (Marsudi et al., 1999) or in equal proportions (Amrani et al., 2010) but always with a lower diversity than in our survey. Partial sequencing of the 16S rRNA gene does not allow the identification of Bradyrhizobium taxa to the species level, and at least 16S-23S rRNA-ITS gene sequencing is requested (Willems, 2006). We thus considered majority of the new strains as undifferentiated Bradyrhizobium sp. Many of them clustered in the vicinity of B. betae, similarly to some A. saligna strains described by Amrani et al. (2010) through 16S rRNA gene sequencing. Concerning the fast-growing strains, our finding that A. saligna is associated with R. tropici confirmed the report of Marsudi et al. (1999), although the latter also found R. leguminosarum. Among other Rhizobium species, we did not detect any R. gallicum strain like Amrani et al. (2010), but a cluster of five A. saligna strains isolated from Bomo-plage that formed a separate Rhizobium sp. subgroup (group 6 in Fig. 3) that was not related to any known referenced type strain, R. galegae/R. loessense being the closest ones. Ensifer meliloti strains were also isolated from A. saligna by Amrani et al. (2010), while in addition, we found strains that clustered with Ensifer kostiense. Lastly, to our knowledge, no Mesorhizobium strain, precisely M. mediterraneum, had been isolated from A. saligna before this study.
The low number of LNB isolated from the Saharan part of South Algeria, around Tamanrasset in particular, led to a weak representation of the rhizobial diversity in some of the tree species surveyed such as A. nilotica, A. tortilis, and F. albida that were only found in this region. This was mostly because of the inability of many strains isolated from neoformed nodules to renodulate their original host plant after the first step of bacterial trapping from the soil samples collected, these noninfective strains having been further characterized as opportunistic bacteria such as Agrobacterium, Burkholderia or Devosia (results not shown). By contrast, a majority of the strains isolated from the two exotic Acacia species found in the Mediterranean coastal zone of North Algeria, namely A. karroo and A. saligna, did renodulate their original hosts and were identified as rhizobia. The low success of rhizobium isolation from soils of desertic environments could be explained by the very low density of nodulating bacteria that is usually found in such ecosystems, especially at the soil surface around superficial roots of legume trees (Jenkins et al., 1987). Thus, the elevated temperature in the Saharan sampling sites might play a role in rhizobial nonsurvival. On the other hand, the noninfectiveness of nodule isolates has been observed in many studies (Khbaya et al., 1998; de Lajudie et al., 1999; Odee et al., 2002). Different assumptions have been proposed: the infective bacteria might have lost their symbiotic information or at least their nod genes during symbiosis or once purified onto artificial media (Brom et al., 1992), or the true symbiont is unculturable or very slow-growing, and only nonnodulating bacterial endophytes coexisting with rhizobia inside nodules may be isolated (Rosenblueth & Martínez-Romero, 2006; Muresu et al., 2008), which has been clearly established for Agrobacterium (de Lajudie et al., 1999) and Burkholderia (Diouf et al., 2007) and several other genera (Muresu et al., 2008).
Opposite to other Acacia species, such as A. mangium and A. mangium × A. auriculiformis interspecific hybrids (Galiana et al., 1990; Le Roux et al., 2009), the efficiency of rhizobial strains did not vary according to the bacterial taxon in each of the three Acacia species tested, namely, A. karroo, A. saligna, and A. seyal. Our results also highlight the particularly high promiscuity of A. saligna that nodulated and fixed nitrogen efficiently with a wide range of rhizobial taxa. Such promiscuity could also explain the invasive status of A. saligna and to a lesser extent A. karroo in their introduction zones. However, some authors did not find any clear relationship between rhizobial diversity and the invasiveness status of acacias, in particular the Australian acacias introduced as exotic species in different Mediterranean countries such as A. longifolia (Rodriguez-Echeverria et al., 2007). Considering its outstanding promiscuity, A. saligna remains an exception because all Australian Acacia species investigated so far are known to be preponderantly nodulated by Bradyrhizobium spp. (i.e. over 90% of nodule isolates).
The collection of new strains showed high phenotypic and genotypic diversity reflecting no particular relationship between the phylogenetic position of bacterial strains and (1) their tolerances to salinity and temperature, and (2) their efficiency toward the different Acacia species tested. Moreover, despite the variety of host species and wide range of geographical zones investigated across Algeria, we clearly showed that there was no correlation between the in vitro tolerances of strains to salinity and high temperature and the edaphoclimatic characteristics of their regions of origin. This questions the strategies of rhizobium selection based on in vitro phenotypic criteria because, in case of incongruence, we should always tend to favor the choice of indigenous strains isolated from the future planting site to be used as inoculants for a given tree species. The tolerance of strains to these factors remains to be further tested in symbiosis with the host plant. In such perspective, parallel extensive research of tree species provenances tolerant to particular edaphoclimatic conditions is also requested. This involves the in situ exploration of genetic material for each species of interest to establish a broader genetic basis, keeping in mind that the geographical distribution of Acacia spp. in Algeria remains poorly known.
We are thankful to INRF (National Institute for Research in Forestry) and Forestry Department for their assistance during the field collections, and to Diegane Diouf for his welcome at LCM laboratory (IRD/ISRA/UCAD) in Senegal where a part of the work was performed. This study was also supported by AUF (Agence universitaire de la Francophonie) and CIRAD scholarships.