![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Applied and Environmental Microbiology, February 2006, p. 1198-1206, Vol. 72, No. 2
0099-2240/06/$08.00+0 doi:10.1128/AEM.72.2.1198-1206.2006
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Department of Biological Sciences, State University of New York, Binghamton, New York 13902
Received 23 September 2005/ Accepted 20 November 2005
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-Proteobacteria (Rhizobium, Sinorhizobium, Mesorhizobium, Bradyrhizobium, and Azorhizobium) (37). This changed when Moulin et al. (14) discovered two nodule-forming isolates of the ß-proteobacterial genus Burkholderia on legumes in Africa and South America. They suggested the terms - and ß-rhizobia to distinguish these two phylogenetic lineages of nodule-symbiotic Proteobacteria. Members of two other genera within the ß-Proteobacteria are now known to be legume nodule symbionts. Chen et al. (3) described the novel species Ralstonia taiwanensis as a symbiont of Mimosa pudica in Taiwan. This species was subsequently transferred to the genus Cupriavidus (31). In a study across 14 sites in Taiwan, Cupriavidus taiwanensis was found to be the dominant symbiont associated with the legumes Mimosa pudica and Mimosa diplotricha, and isolates of Burkholderia caribensis also occurred as nodule symbionts in this region (4). Both M. pudica and M. diplotricha are plants endemic tothe neotropics that have been naturalized in Taiwan (1, 4,11, 36). Another recently described ß-proteobacterium (Herbaspirillum lusitanum) was found in Portugal to nodulate Phaseolus vulgaris (28). Data are still limited regarding the symbiotic relationships of rhizobia and mimosoid legumes in their native geographical range. Barrett and Parker (2) found Burkholderia sp. bacteria to be the predominant root nodule symbionts for four mimosoid legumes from Barro Colorado Island, Panama. These included three Mimosa species (M. casta, M. pudica, and M. pigra) of neotropical origin that also occur as invasive weeds in tropical and subtropical locations around the world, and a fourth endemic species, Abarema [Pithecellobium] macradenia. In a study of 20 nodule isolates from nine species of Mimosa native to Brazil and Venezuela, Chen et al. (5) also found that strains of Burkholderia were the predominant symbionts. However, both the number of locations studied and the sample size of nodule isolates within locations remain limited. Further studies are therefore needed to better characterize the diversity of ß-rhizobia in the neotropics and the degree of phylogenetic relationship for bacteria in different regional populations of a given legume host species (2, 5, 13, 17).
In this study, we used 16S rRNA and 23S rRNA gene sequencing, PCR assays, and inoculation experiments to analyze the symbiotic interaction of nodule bacteria with four mimosoid legumes in Costa Rica. Our overall goal was to test whether prior results demonstrating a preponderance of Burkholderia sp. nodule symbionts on two common mimosoid legumes on Barro Colorado Island, Panama (M. pigra and M. pudica), also extended to other locations where these plants grow in the neotropics. The research also addressed the following four issues. First, do local legume populations utilize both - and ß-rhizobia, and if so, are these segregated on different plants, or do they coexist at the scale of an individual plant's root system? Second, how much spatial differentiation in symbiont population composition exists, both among nearby habitats and on a regional scale (comparing sites in Costa Rica versus Panama)? Third, does symbiont population composition differ for the two Mimosa species? Finally, do - and ß-rhizobia provide comparable symbiotic benefits to their legume hosts, in terms of nodule nitrogenase activity and plant biomass gain?
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
An average of 3.3 nodules were taken from each individual plant (range, 1 to 7). Nodules were sampled by carefully excavating around the base of the stem until roots with attached nodules were located. Nodules were stored in vials with desiccant and refrigerated. Within four months, nodules were rehydrated in 0.04 M sodium phosphate buffer (pH 7.0). Nodules were surface disinfected in 100% ethanol (10 seconds) and 3.2% sodium hypochlorite (90 seconds), followed by five rinses in sterile water. One isolate was purified from each nodule as described previously (24). All isolates were grown on yeast extract-mannitol (YM) agar plates.
DNA amplification and sequencing.
DNA was purified by heating bacterial cells for 5 min at 95°C in lysis buffer followed by a single chloroform extraction (19). PCR experiments used standard protocols with 32 cycles of 94°C (20 seconds) and 58 to 61°C (10 seconds; temperature varying for different primer pairs) and 72°C (40 seconds), with a final extension of 3 min at 72°C. Primers UP6N and 23Sr#2 (26) were used to screen for length variation in the 5' 23S rRNA intervening sequence (IVS) region. Thirty-one isolates chosen to represent all IVS length variants, sites, and host legume species were sequenced in the 5' 23S rRNA IVS region (390 to 496 bp), using an Applied Biosystems model 310 automated sequencer with protocols recommended by the manufacturer. These data were then used to design specific primers for two subsequent rounds of PCR screens in order to discriminate genotypes among the remaining isolates.
In the first round, we used PCR primers designed to differentiate bacteria by genus (Table 1). In the second round we used genotype-specific primer pairs to differentiate two genotypes (designated A and C) within the genus Burkholderia, and three genotypes (designated G, H, and I) within the genus Cupriavidus (Table 1). To analyze primer specificity, representatives of each bacterial genus and of genotypes A, C, G, H, and I were tested with each primer pair in all possible combinations. Each isolate yielded an amplification product only with the primer pair(s) expected from its 23S rRNA sequence. Thus, the primers had adequate specificity to differentiate bacterial lineages in this sample.
|
To determine the degree of statistical support for branches in the phylogeny (10), 1,000 bootstrap replicates of the data were analyzed (100 replicates only for maximum likelihood analyses). The resulting trees were nearly identical for maximum parsimony, neighbor joining, and maximum likelihood analyses. In the few cases where a branch inferred by one method was not confirmed by the other methods, the branch generally had low bootstrap support. To be conservative, these questionable branches were collapsed to a multifurcation in the tree diagram (Fig. 1), as were all branches for which maximum parsimony, neighbor joining, and maximum likelihood analyses all yielded bootstrap support values of <50%.
|
For the Ralstonia/Cupriavidus portion of the tree, the reference strains included Ralstonia solanacearum (NC 003295), R. picketti (AY268180, AF012421), Cupriavidus metallidurans (AAAI02000005), C. necator (formerly Ralstonia eutropha) (AADY01000005.1) and C. taiwanensis strains LMG19424 and LMG19425 (AF300324, AF300325). A portion of the 23S rRNA gene was sequenced in strains LMG19424, STM815, Br3467 and Br3469 to provide data comparable to that for the other strains (strain LMG19425 was not available and so was excluded from the analysis of 23S rRNA sequence variation). Four other ß-proteobacterial taxa with completely sequenced genomes were used as outgroups for all phylogenetic analyses: Chromobacterium violaceum (NC 005085), Neisseria meningitidis (NC 003112), Bordetella pertussis (NC 002929) (28), and Bordetella parapertussis (NC 002928).
In a separate analysis, five -rhizobial isolates (three Rhizobium, and two Bradyrhizobium) were also sequenced for a nearly full-length portion of the 16S rRNA gene to permit characterization of their relationships to other -rhizobia using BLAST searches.
Symbiotic phenotypes.
Seeds of M. pigra and M. pudica were scarified and surface disinfected in concentrated sulfuric acid for either 10 min (M. pigra) or 26 min (M. pudica). Seedlings were planted in pots using an artificial soil mix known to be free of nodule bacteria and then inoculated with approximately 109 cells of a particular isolate grown in YM broth. Three separate experiments were performed. First, five isolates representing different ß-proteobacterial genotypes were tested (four plants per isolate for M. pigra and five plants per isolate for M. pudica). Plants were grown in a greenhouse for 32 days with precautions to avoid bacterial contamination (35). Uninoculated control plants grown simultaneously in the same room were found to be completely free of nodules, indicating that contamination was negligible. Also, for each inoculant strain, one bacterial colony was reisolated from a surface-disinfected nodule, and its 23S rRNA IVS region sequence was determined to verify the identity of the nodule occupants. For all five strains, the sequence from the reisolated culture was identical to that of the inoculant strain, showing that the inoculant strains were the ones responsible for nodule development on these plants. At harvest, nodules were counted and each plant's root system was analyzed for acetylene reduction activity using a Hewlett-Packard 5890 series 2 gas chromatograph as described previously (24).
In a second experiment, 17 isolates identified as Rhizobium based on PCR assays (Tables 1 and 2) were used to inoculate plants of M. pudica only (n = 4 seedlings per isolate). Nodule numbers per plant and acetylene reduction activity were measured after 28 days. In order to compare plant responses to - and ß-rhizobia in a single experiment, M. pudica seedlings were inoculated with two Rhizobium strains, one Burkholderia strain (genotype C) and one Cupriavidus strain (genotype H) in a final experiment. Fifteen plants were inoculated per strain, and plant biomass was measured after 41 days. A subset of five plants per group was also analyzed for acetylene reduction activity.
|
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The two isolates from Pithecellobium arboreum yielded a 538-bp PCR product for the 23S rRNA IVS region, while the isolate from Pentaclethra macroloba yielded a 510-bp product. Both of these length variants are common among strains of Bradyrhizobium in Central America (16-18, 20). Amplicons from the three strains were sequenced, and BLAST searches indicated that they were indeed closely related to various Bradyrhizobium strains sampled from other legume species in both Costa Rica and Panama (16, 17).
Seven distinct sequences were observed among 20 isolates with the 432-bp length variant initially sequenced, and BLAST searches indicated that 10 isolates were affiliated with the genus Burkholderia, and the others were strains of Cupriavidus. PCR assays with two 23S rRNA primer pairs specific to sequence motifs characteristic of each genus (Table 1) were performed on all 176 isolates having the 432-bp length variant. Two-thirds of the isolates (117) exhibited a PCR product only with the Burkholderia-specific primer pair, and the remaining 59 isolates yielded a PCR product only with the Cupriavidus primers.
Further PCR assays were performed to discriminate genotypes within the two ß-proteobacterial genera. First, the 10 Burkholderia isolates displayed three distinct sequences. Two of these genotypes (differing by 22 bp) proved to be identical to nodule-symbiotic Burkholderia genotype A (four isolates) or genotype C (five isolates) from Barro Colorado Island, Panama (2). The remaining sequence (from isolate Tpig4.4) was a close relative of genotype C, differing by two nucleotide substitutions (here designated genotype C1).
Two primer pairs specific to genotypes A and C developed in a prior study (2) were used to screen all 117 isolates identified as Burkholderia based on their amplification with the burk23f/burk23r primers. Note that the genotype C primers (mp52f/mp52r [Table 1]) matched both the original genotype C and the closely related variant sequence C1, so it was not possible to discriminate between these strains by this PCR assay. Eighty of the Burkholderia isolates yielded a PCR product only with the genotype A primers, and 32 isolates yielded a product only with the genotype C primers. The remaining five Burkholderia isolates amplified with neither primer pair and therefore appeared to represent novel genotypes. The 23S rRNA IVS regions were sequenced in these five isolates, and they all proved to have the same sequence. This genotype (designated genotype E) was divergent from that of the other Burkholderia strains, displaying 30 and 38 nucleotide substitutions relative to genotypes C/C1 and A, respectively.
Four genotypes (designated G, H, I1, and I2) were detected among the 10 Cupriavidus isolates initially sequenced. Genotypes I1 (3 isolates) and I2 (one isolate) differed by only one nucleotide substitution, while 5 to 17 differences existed for all other pairwise comparisons among genotypes. Primers targeting unique sequences in the 23S rRNA IVS region were designed to allow discrimination of three of these genotypes among the 59 isolates classified as Cupriavidus based on the prior PCR assay (Table 1) (the primers did not permit differentiation of genotypes I1 versus I2). Twenty-five isolates yielded a PCR product only with the genotype G primers, 24 isolates yielded a product only with the genotype H primers, seven isolates amplified only with the genotype I1/I2 primers, and three isolates amplified with none of the three primer pairs.
Sequencing of one of these three isolates (MApud3.4) indicated that it corresponded to a novel genotype (designated genotype J). It appeared to be a recombinant between two other Cupriavidus genotypes, being identical to genotype G in the 5' half of the 23S rRNA IVS region and identical to genotypes H and I1/I2 in the 3' portion. A combination of primers tpig6F (G) and rma10r (H) could therefore be used to selectively amplify genomic DNA of this genotype. All three of the Cupriavidus isolates that failed to amplify with the G, H, or I1/I2 primer pairs (Table 1) exhibited a PCR product with the mixed-genotype tpig6F (G)/rma10r (H) primer pair and were therefore assigned to genotype J.
Distribution of bacteria across host legumes.
Classification of isolates based on differential PCR amplification results indicated that Mimosa pigra and M. pudica clearly had distinct symbiont communities (Table 2). Isolates identified as Burkholderia, Cupriavidus, and Rhizobium occurred in nodules of both Mimosa species. However, pooling across sites and across genotypes within bacterial genera showed that the overall abundances of the three bacterial genera differed significantly for the two Mimosa species (log-likelihood ratio test, G = 25.49, 2 df, P < 0.001). Burkholderia predominated among nodule isolates from M. pigra (86% of isolates), while there was a more uniform distribution of the three bacterial genera on M. pudica. Cupriavidus and Burkholderia spp. were nearly equal in prevalence (38% and 37% of isolates, respectively), with Rhizobium spp. being only slightly less common (25%).
Genotypes within Burkholderia and Cupriavidus were also differentially distributed across the two Mimosa species. For M. pigra, all but one of the Burkholderia isolates (98%) belonged to genotype A. This genotype occurred on M. pudica, but had a lower frequency (29% of this plant's Burkholderia isolates), with genotype C/C1 predominating instead (61%). With respect to Cupriavidus, the only genotype that occurred in association with M. pigra was genotype G. Mimosa pudica harbored a higher diversity of Cupriavidus genotypes, with nearly equal relative abundances of two genotypes (G and H) together with lower numbers of genotypes I1/I2 and J.
Variation among sites.
The two host legume species differed in the degree of spatial variation in symbiont population composition as well as in overall bacterial diversity (Table 2). M. pigra showed little spatial differentiation, with Burkholderia genotype A predominating at both Costa Rican sites. This same genotype also predominated in samples from an M. pigra population 440 km away in Panama (2). A log-likelihood ratio test indicated that there was no evidence for variation in the frequency of Burkholderia genotype A at the three sites (G = 2.51, 2 df, P > 0.25).
For M. pudica, by contrast, a different group of bacteria predominated at each of the three Costa Rican sites sampled. The majority of bacteria sampled from the Manuel Antonio National Park site were strains of Cupriavidus. The Tres Piedras and Hacienda Baru sites were both dominated by Burkholderia spp., but different Burkholderia genotypes were most common in the two locations (Table 2). Burkholderia spp. also predominated in a Panamanian population of M. pudica, but the most common genotype in the Panamanian M. pudica population (designated Burkholderia genotype D) (2) was not found in Costa Rica. If genotypes within Burkholderia and Cupriavidus are pooled to allow a comparison of bacterial prevalences for M. pudica at the three Costa Rican sites and in Panama, a log-likelihood ratio test showed that there was highly significant variation in the relative abundance of bacterial genera across the four locations (G = 87.43, 6 df, P < 0.001).
Bacterial diversity within individual plants.
Isolates from the two Mimosa species (Table 2) were also used to characterize patterns of bacterial distribution among separate nodules on individual plants. Isolates from at least two nodules were available for 20 of the 22 M. pigra and 41 of the 43 M. pudica plants sampled. For 70% of the plants with multiple nodules (43/61), a single bacterial genus occupied all nodules obtained. However, 33% of these plants (14/43) had nodules occupied by two or more genotypes of the same bacterial genus.
Among the 18 plants that harbored more than one bacterial genus in different nodules, Rhizobium spp. coexisted with one or both of the ß-rhizobial genera (Burkholderia or Cupriavidus) on 10 plants, and the remaining eight plants involved coexistence of Burkholderia and Cupriavidus spp. To analyze whether these values differ from the null expectation that different bacterial genera are distributed at random across plant individuals, data were partitioned by host legume and were also analyzed separately by site.
The overall proportion of nodules within a site occupied by Burkholderia (pB), Cupriavidus (pC), or Rhizobium (pR) spp. was calculated. For each category n of nodules sampled per plant (with n ranging from 2 to 7), the expected fraction of plants that had all nodules occupied by only one bacterial genus was estimated as (pB)n + (pC)n + (pR)n. This sum was then multiplied by the observed sample size of plants within the site that had n nodules analyzed, to obtain the expected number of plants with no generic diversity of nodule bacteria, given that n nodules were sampled. Adding these values for all nodule number classes yielded the expected total number of plants with only a single bacterial genus represented in their nodules.
For Mimosa pigra, the number of plants harboring only a single bacterial genus was about 46% higher than expected under the null hypothesis that bacteria are distributed at random. However, a 
2 goodness-of-fit test indicated that this difference was not significant (2 = 5.32, 2 df [representing the two sites], 0.05 < P < 0.10). For M. pudica, more than twice the expected number of plants harbored only a single bacterial genus, implying a highly significant deviation from a random distribution of bacterial genera across plants (2 = 26.95, 3 df, P < 0.001). This can be interpreted as evidence for a clumped distribution of bacteria at the scale of individual plants, so that replicate nodules from an individual have lower diversity than expected based on the local population prevalence of the three rhizobial genera.
Because this analysis was based on sampling a single isolate per nodule, the results may underestimate the true frequency with which bacterial genera coexist on individual plants. Separate strains can occupy a single nodule (23), which would not be detectable by our isolate sampling scheme. Future research should characterize patterns of within-nodule diversity. Nevertheless, our results are sufficient to establish that individual plants interact regularly with strains of both - and ß-rhizobia.
16S rRNA phylogenetic relationships.
A nearly full-length 16S rRNA gene sequence was obtained for four Burkholderia and five Cupriavidus strains (representing all 23S rRNA sequence variants detected) to analyze their relationships to other strains of ß-Proteobacteria (Fig. 1). The 16S rRNA sequences confirmed inferences from 23S rRNA data about the generic identity of these strains in all cases, with each of the nine Costa Rican strains showing a clear relationship to reference strains of either Burkholderia or Cupriavidus. Both the Cupriavidus lineage and the Burkholderia lineage had high bootstrap support (Fig. 1).
Within the Burkholderia lineage, the Costa Rican strains were included among a well-supported group (94 to 100% bootstrap values) of legume symbionts from other locations (Panama, Brazil, French Guiana, and South Africa). This clade also included B. fungorum (Fig. 1), which is not known to be a nodule symbiont (7). The four Costa Rican strains had affinities to three of the four Burkholderia genotypes (designated A, B, C, and D) detected in a prior study of Mimosa nodule symbionts in Panama (2). First, Costa Rican strain Hpig15.6 proved to be identical to Panamanian strain Mpig8.9 (genotype A; these strains were also identical for the 23S rRNA IVS region). Strain Hpud12.1 (genotype E) was 99.2% similar to Panamanian strain Mcas7.1 (genotype B). Strains Hpud10.4 and Tpig4.4 (which differed by only one nucleotide substitution) were 99.4% similar to Panamanian strain Mpud5.2 (genotype C). Two reference strains from Brazilian Mimosa plants (Br3467 and Br3469) did not have a close relationship to any of the Costa Rican Burkholderia strains (5), showing at least 45 to 50 nucleotide differences from all of the Central American Burkholderia genotypes.
Among the five Cupriavidus strains sequenced, two strains that were closely related based on 23S rRNA data (Tpud27.6 and MApud8.1, representing genotypes I1 and I2, respectively) proved to have identical 16S rRNA sequences. These strains showed 10 to 12 differences from Costa Rican genotypes G, H, and J. These five strains formed a group together with the two C. taiwanensis strains from M. pudica in Taiwan (3). Each of the two Taiwanese strains showed a closer similarity to a Costa Rican strain than they did to one another. C. taiwanensis strain LMG 19424 differed at only three nucleotide sites from Costa Rican isolates Tpud27.6 and MApud8.1, while C. taiwanensis strain LMG19425 had four differences relative to strain Tpig.6a (Fig. 1).
Phylogenetic analyses on 23S rRNA IVS region sequences from the ß-proteobacterial strains resulted in trees that were generally similar to the 16S rRNA tree (data not shown), although bootstrap values were lower because this shorter region (395-bp aligned length) had only about 38% the number of polymorphic nucleotide sites as in the 16S rRNA data set. In a prior analysis of Burkholderia strains from Panama, it was found that Burkholderia genotype B, B. tuberum, and a few nonsymbiotic strains (B. multivorans and B. gladioli) showed significantly conflicting topologies for the two rRNA gene regions (2), as revealed by partition homogeneity tests (9). This was interpreted as evidence for lateral gene transfer events (29) modifying the genealogical history of the two rRNA regions (2). Comparison of 16S rRNA versus 23S rRNA trees for the Costa Rican strains confirmed these findings, showing that the Costa Rican strain related to genotype B (Hpud12.1) exhibited the same alteration in phylogenetic placement previously found for Panamanian genotype B (data not shown). Thus, the 16S rRNA tree (Fig. 1) should only be interpreted as a phylogeny of this gene and not viewed as an organism-level phylogeny, because certain strains show an altered pattern of relationship for other portions of the ribosomal gene region.
Sequences for 16S rRNA genes were also obtained from five isolates provisionally classified as Bradyrhizobium (n = 2) or Rhizobium (n = 3) based on 23S rRNA IVS region sequences. The 16S rRNA sequence data verified earlier inferences about generic identity in all cases. The two Bradyrhizobium isolates were each identical to other Bradyrhizobium strains sampled from the tree Dalbergia retusa in the same habitat in a prior study. Isolate TPma1.5 from Pentaclethra macroloba was identical to Bradyrhizobium sp. strain Dr3b.11, while isolate TPar1.1 (Pithecellobium arboreum) was 100% similar to Bradyrhizobium sp. strain Dr4a.7 (18). Strains highly similar to each of these two genotypes also occur in Panama on several other host legume species (14, 16).
Isolate Hpig4.1 from M. pigra was 99.8% similar to Rhizobium genospecies Q (Z94806) from the legume Davesia leptophylla in southeastern Australia (13). Rhizobium isolates Tpud40a and Tpud22.2 from Mimosa pudica were identical to each other, and BLAST searches indicated that the closest known strain in GenBank was Rhizobium etli SEMIA 384 from Brazil (AY904730), which was 98% similar. Wang et al. (33) found a lineage of R. etli to be commonly associated with a species of Mimosa native to central Mexico (M. affinis). A partial 16S rRNA sequence (516 bp) from one of these M. affinis isolates (AY929418) proved to be 97% similar to sequences from Tpud40a and Tpud22.2. Thus, some of the Costa Rican Rhizobium strains appear to be related to an R. etli lineage that is associated with other Mimosa taxa in the region.
Symbiotic phenotypes.
Plants of M. pudica and M. pigra were inoculated with five ß-rhizobial isolates (representing Cupriavidus genotypes G, H, and I1 and Burkholderia genotypes C and E). Representatives of genotype A were shown to nodulate Macroptilium atropurpureum and Mimosa pigra in a previous study (2). All three Costa Rican Cupriavidus genotypes formed effective, nitrogen-fixing root nodules on both M. pudica and M. pigra (Fig. 2), as did the two Burkholderia genotypes. For M. pigra, acetylene reduction values varied widely, ranging from 0.04 to 2.34 µmol min–1 plant–1. Mean acetylene reduction values were in most cases higher for nodules developing on M. pudica, ranging from 1.9 to 4.1 µmol min–1 plant–1 (Fig. 2).
|
To further analyze the finding that M. pudica plants seemed to have lower nodule nitrogenase activity with Costa Rican Rhizobium strains than with ß-rhizobia, a follow-up experiment was done to directly compare plant biomass gains and acetylene reduction rates for plants interacting with these different bacterial lineages (Table 3). Although plants inoculated with the two Rhizobium strains had nodules with detectable acetylene reduction activity, their biomass after 41 days was no higher than that of uninoculated control plants. The Burkholderia genotype C strain and the Cupriavidus genotype H strain resulted in significantly higher plant biomass and acetylene reduction activity than the two Rhizobium strains. Thus, while Mimosa pudica plants associate with significant numbers of both ß-rhizobia and Rhizobium strains in natural environments (Table 2), the ß-rhizobial strains appear to be superior symbiotic partners with respect to nitrogen fixation rate and plant growth enhancement.
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our results also provide the first evidence for Cupriavidus nodule symbionts in the New World. The association of Cupriavidus bacteria with invasive populations of Mimosa sp. legumes in Asia is apparently not rare, in view of their prevalence in Taiwan (3) and locations in India (32). M. pigra and M. pudica are native to Central and South America (1, 8, 12) and have become naturalized or invasive in numerous tropical and subtropical regions on other continents (1, 11, 36). Since the Cupriavidus isolates found in this survey occur within the native range of both M. pigra and M. pudica, these symbiotic relationships may be of great antiquity. The Costa Rican Cupriavidus genotypes were <2% divergent from each other and from Taiwanese strains of C. taiwanensis for 16S rRNA genes, suggesting that they may be variants within the same species (25).
Although none of the Costa Rican and Taiwanese strains had identical 16S rRNA genes, the two C. taiwanensis strains from Taiwan studied by Chen et al. (3) (LMG 19424 and LMG 19425) each had a closer relationship to a Costa Rican strain than they did to one another. This is consistent with the interpretation that the Taiwanese Cupriavidus strains represent two separate introductions from a Central American source population. A parallel conclusion was reached by Chen et al. (6) for Burkholderia strains associated with invasive populations of M. pigra in Taiwan, based on sequence data showing that these bacteria appeared to be derived from bacterial strains found in M. pigra populations in South America.
Although some Burkholderia and Cupriavidus genotypes were shared by both Mimosa species, it is clear that M. pigra and M. pudica differed in their overall patterns of symbiont utilization (Table 2). These legumes also differed in the degree of geographical variability of their rhizobial populations. M. pigra showed a high degree of spatial uniformity, with Burkholderia genotype A dominant in both Costa Rican sites as well as in a population on Barro Colorado Island, Panama (2). For M. pudica, there was considerable spatial variation in symbiont population composition, both among nearby habitats in Costa Rica (Table 2) and in broader regional comparisons between these sites and Panama (2). Indeed, a different Burkholderia or Cupriavidus genotype predominated at every site sampled. These results suggest that M. pudica may be more indiscriminant in its choice of symbiotic partners than M. pigra. If so, then local environmental factors and stochastic processes that alter the relative abundance of bacterial lineages among microhabitats may have a bigger impact on nodule bacteria utilization by this plant.
An alternative possibility is that M. pudica populations vary genetically for traits that affect bacterial strain selection (15). Future research should address whether plant genotypes in different locations vary in nodulation preference for particular rhizobial lineages.
The presence of Cupriavidus distinguished the Costa Rican sites from a prior survey of these same Mimosa species in Panama, where most nodules were occupied by strains of Burkholderia (2). However, considering only Burkholderia strains, 16S rRNA sequence comparisons indicated an overall pattern of regional similarity between Costa Rica and Panama. Each of the Costa Rican Burkholderia strains had a close relationship to a particular Panamanian strain (Fig. 1), with no novel lineages detected. Indeed, Burkholderia genotype A isolates from Costa Rica and Panama were completely identical for both 23S rRNA and 16S rRNA sequences. This suggests that bacterial migration across this region has been sufficiently common over recent evolutionary time to introduce the same lineage into both locations. It should be noted, however, that Burkholderia strains from South American M. pigra populations (Venezuela and Brazil) (5) had a distant relationship to the Central American strains (Fig. 1). Thus, on a broader geographic scale, symbiont populations associated with this plant appear to be effectively isolated.
Strains belonging to the genus Rhizobium comprised about 6 and 25% of all nodule isolates from M. pigra and M. pudica, respectively. These occurred regularly on the same individual host plants as ß-rhizobia and were found at all sites. However, the inoculation experiments with M. pudica suggested that these bacteria may be inferior symbiotic partners on average relative to ß-rhizobia. The mechanisms that allow - and ß-rhizobia to coexist as nodule symbionts for a given legume population despite apparent differences in the benefits conferred on plants (Table 3) are of great interest (23, 34).
We plan in future studies to analyze the relative competitive abilities of - and ß-rhizobia regarding their capacity to induce nodule formation and the ability of legume hosts to differentially select -rhizobia versus ß-rhizobia as symbiotic partners in order to address these issues.
|
ACKNOWLEDGMENTS |
|---|
Funding was provided by the Tropical Forestry Initiative and NSF grant DEB-0235766.
|
FOOTNOTES |
|---|
Present address: Ohio State University Herbarium, Museum of Biological Diversity, Columbus, OH 43212.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| J. Bacteriol. | Microbiol. Mol. Biol. Rev. | Eukaryot. Cell | All ASM Journals |
|---|
