INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Actinorhizal plants are defined by their ability to form N₂-fixing root nodule symbioses with actinomycetes from the genus Frankia (2). The symbiosis uncouples the plants from a need for soil nitrogen. As a result, the plants and their infective symbionts have radiated into a remarkable variety of niches that include dry tropical soils, temperate wetlands, northern forests, sand dunes, chaparral and matorral, subarctic bogs and tundra, and glacial till (2).

Twenty-five genera of actinorhizal plants have been identified in eight families of angiosperms (2). One family, the Myricaceae in the subclass Hamamelidae, is considered the first actinorhizal family to emerge during the late Cretaceous (17, 19). Other actinorhizal families differentiated independently much later (29). Modern Myrica sp. have successfully adapted to habitats ranging from tropical and temperate dry soil and sand dunes to water-saturated bogs in northern temperate and subarctic regions.

Because of the wide variety of niches occupied by myricaceous plants, it is likely that some Frankia strains have adapted to symbiosis in the context of the prevailing conditions in specific niches. Such specialization can only be revealed by studying the diversity of strains that actually occupy nodules in nature. In greenhouse studies, Myrica spp. have the unusual ability to nodulate with virtually all isolated strains of Frankia tested. For this reason, they are considered to be "promiscuous" hosts, but the extent of their promiscuity in nature has not been addressed (1, 31).

Phenotypic diversity of Frankia strains has previously been studied primarily by isolating and characterizing representative strains from a handful of plant species (2). Nodules from many plants fail to yield isolates, and many isolates grow so slowly that a classical approach to diversity studies that includes isolation, cultivation, and physiological characterization is not practical. Developing an understanding of the broader issues of diversity in actinorhizal nodules requires an approach that circumvents the need to isolate and grow strains in culture. In this study, we used the molecular ecological approach of amplifying partial 16S rRNA gene sequences (rDNAs) directly from root nodules of three species within the Myricaceae as a means to assess the richness and evenness of Frankia strains that inhabit root nodules (richness is the total number of unique sequences found in the nodules; evenness refers to the relative number of times that each unique sequence appears in the nodules sampled). Comptonia peregrina (L.) Coult. (sweet fern), Myrica gale L. (sweet gale, bog myrtle), and Myrica pensylvanica Loisel. (bayberry) were chosen as test plants because they overlap in their geographical distribution but occupy different niches in northeastern North America (24).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Collection of root nodules. To obtain a broad sample of Frankia strains and to avoid bias introduced by local strain dominance (10), root nodules were collected from 96 separate sites, including 30 with C. peregrina, 29 with M. pensylvanica, and 37 with M. gale (Table 1; Fig. 1). C. peregrina grows inland in open dry soil from Georgia and Minnesota to Nova Scotia. M. pensylvanica is distributed from North Carolina to Nova Scotia, particularly in coastal environments, and grows west to the Great Lakes. M. gale grows in bogs and borders lakes from northern New Jersey to Newfoundland and has a circumboreal distribution through Canada, Alaska, northwestern Europe, Scandinavia, and Asia (33). Most of the collection sites for this study were distributed from New Jersey to northern New York to Maine in the northeastern United States, but M. gale nodules were also received from Canada, Sweden, and Scotland and one M. pensylvanica plant was an introduced plant in Washington State (Fig. 1).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Geographical distribution of collection sites for root nodules from the Myricaceae. Numbers indicate the number of nodules taken from each state. The first, second, and third numbers correspond to the number of samples obtained from each state for C. peregrina, M. pensylvanica and M. gale, respectively. Numbers to the right correspond to nodules collected from beyond the geographical range shown.

Amplification of 16S rDNA from nodules. The procedure for nodule dissection and DNA amplification has been described previously (3, 12). Field-collected nodules were frozen or stored in 95% ethanol. Single nodule lobes (1 to 10 mg of plant tissue) were excised from the nodules. After the lobes were washed, the periderm was removed in sterile TEA buffer (10 mM Tris-HCl, 1 mM Na₂ EDTA, 20 mM ascorbic acid [pH 7.6]). The nodule contents were macerated with a scalpel and rinsed in TEA buffer, and Frankia vesicle clusters (mycelial contents from a single infected cell) were washed and collected on nylon screens with Tris-EDTA buffer. Total DNA was extracted by using alkaline lysis and standard ammonium acetate-ethanol precipitation (26). A 380-bp fragment of Frankia 16S rDNA was PCR amplified with primers fD1 (32) and rDB1 (3) in quadruplicate reactions by using a Perkin-Elmer (Norwalk, Conn.) amplification kit. Amplicons were combined, purified on Qiagen (Santa Clarita, Calif.) columns, and cycle sequenced in both directions, and the sequencing reactions were analyzed with an Applied Biosystems (Perkin-Elmer) Prism sequencer, all according to manufacturers' protocols.

Mathematical estimates of diversity. Diversity indices were from Magurran (18). The Alpha and Shannon indices give particular weight to species richness within a sample, with higher values reflecting greater richness. The Berger-Parker (reciprocal) and Simpson (reciprocal) indices are weighted towards dominance, with a lower number representing greater dominance in the sample.

Dendrogram construction. To construct trees, 16S rDNA sequences were initially aligned by using Pileup in the Genetics Computer Group package (14). The alignments were reformatted in ClustalW (30) and checked by eye. All sites in the 380-bp region were included in the analysis. Dendrograms were constructed by using PHYLIP (15). A distance matrix was made by using the maximum likelihood option in DNADIST with a transition/transversion ratio of 2. A neighbor-joining tree was then constructed in NEIGHBOR. The tree was viewed as a cladogram in TREEVIEW (22).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Reproducibility and amplification. By using only healthy nodule lobes, all attempts to amplify DNA from nodules were successful in this study. We avoided problems with errors due to mispriming and Taq misincorporation by using quadruplicate amplifications from the template DNA. The consistency of the data was supported by the observation that random substitutions in conserved regions of the 16S rDNA molecule were not found.

Analysis of diversity. Variability between Frankia 16S rDNA sequences was used as a measure of diversity. A neighbor-joining tree relating the sequences from this study plus published reference sequences was constructed and is shown in Fig. 2. The four groups designated I to IV in Fig. 2 have been identified as four clades in previous studies on the phylogenetic relationships of Frankia (21).

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Neighbor-joining tree of partial 16S rDNA from Frankia strains. The outer taxonomic unit numbers correspond to the sequence numbers in Table 1, and the frequency of occurrence of a sequence in C. peregrina, M. gale, or M. pensylvanica is shown by the number of bullets in the columns to the right of the tree. Groups I to IV correspond to the major lineages of Frankia identified by Normand et al. (21). Group I Frankia strains nodulate actinorhizal plants from the hamamelid families Betulaceae, Casuarinaceae and Myricaceae; Group II Frankia are typical inhabitants of nodules from the families Elaeagnaceae and Rhamnaceae and some hamamelid genera; group III Frankia strains occupy nodules of the families Rosaceae, Coriariaceae, and Datiscaceae plus Ceanothus of the family Rhamnaceae; group IV Frankia strains are a loosely associated group of Nod and Fix actinomycetes. The tree was rooted with the 16S rDNA sequence from Geodermatophilus obscurus subsp. obscurus (accession no. L40620).

Estimates of diversity. To gain additional insight into the diversity of 16S rRNA sequences in these root nodules, several indices were computed by using the information shown in Fig. 2 (Table 2); each unique sequence was treated as a strain of Frankia. Indices that were weighted towards strain richness (as assessed by the total number of unique sequences) consistently reflected the higher diversity of strains in M. pensylvanica nodules relative to C. peregrina and especially M. gale nodules. Indices that weighted evenness (as assessed by the relative number of each unique sequence) reflected the dominance of a few strains in M. gale, the lower dominance of strains seen in C. peregrina, and the relatively greater evenness found in M. pensylvanica nodules.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   Variability in abundance-based coverage estimates (ACE values) as samples accumulate. Data are from Fig. 2. Sobs is the mean total number of strains observed in the pooled randomized samples at each accumulation level (open diamonds). ACE (closed diamonds) values are the mean estimators computed by using the Sobs values at each accumulation level. Sample accumulation refers to the number of nodules used. As samples are added, the estimated total richness increases to a maximum. (a) C. peregrina; (b) M. gale; (c) M. pensylvanica.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Many Frankia strains have not been grown in culture, and others grow slowly once isolated (2). Therefore, the diversity of strains that actually live in root nodules in natural systems has been difficult to address. Several studies have used 16S rDNAs of Frankia as a way to group strains for phylogenetic analyses (21). The 380-bp portion used here is located at the 5' end of the gene and contains about 45% of the known variable sites found among available Frankia 16S rRNA genes (12). Trees constructed by using this region place Frankia strains into the same major groups as those found with longer 16S rDNA sequences, albeit with lower resolution (12). Strains differing at two or more positions in the region have recognizably different total protein profiles, colony morphologies, and carbon source utilization patterns (10). While the portion of the 16S rDNA used here has phylogenetic information, we were primarily interested in using it as a relatively rapid means to estimate the natural diversity of Frankia populations. In that context, two nodule inhabitants that have the same sequence can be viewed as being the same, or very closely related, strains. This view is supported by the observation that the inhabitants of five nodules from M. gale, growing in separate locations, that had identical short 16S rDNA sequences also had identical glutamine synthetase I gene (glnA) sequences (data not shown).

The molecular ecological approach used here has allowed us to estimate the strain richness and evenness of Frankia strains in the root nodules of three species from the plant family Myricaceae. We found that species (or strain) evenness differs dramatically in naturally occurring root nodules. The major groups I to IV in Fig. 2 relate closely to the major groups of plants that participate in the actinorhizal symbiosis. Group I includes strains that are typically found in, and are isolated from, nodules of plants in the hamamelid families Betulaceae, Casuarinaceae, and Myricaceae and is composed of more than one major subgroup (21, 34). Group II includes symbionts and isolates from plants in the families Elaeagnaceae and Rhamnaceae, as well as some hamamelid families (20). Group III contains symbionts from plants in the families Coriariaceae, Datiscaceae, Rosaceae and from Ceanothus of the Rhamnaceae (3, 12, 25). Group IV contains Frankia-like organisms whose role in symbiosis is unclear (25); the latter have been isolated from a variety of nodules but are noninfective when tested on actinorhizal plants. Isolates that represent groups I, II, and IV but not group III have been obtained.

M. pensylvanica was unusual in the sense that Frankia from groups I, II, and IV were found in its nodules. The species can thus be considered to be promiscuous in nature (31) and possibly a reservoir for different major lineages of Frankia. The existence of such reservoir plants may explain why certain actinorhizal species nodulate wherever they are planted, regardless of whether plants from the same group have previously grown in the vicinity (2). Previous studies on Frankia strains isolated from M. pensylvanica nodules have also indicated that many strains were present although the relationship of one strain with another could not be determined at the time (4, 5).

In contrast to M. pensylvanica, we found widely collected M. gale nodules to be primarily occupied by a limited set of strains from group I despite the observation that M. gale seedlings are infected by Frankia strains or crushed nodule inocula originating from many actinorhizal plants in the greenhouse (31). The major strain was not found in other myricaceous nodules examined despite its apparent abundance in M. gale. Thus, it seems likely that this strain has specialized in infecting M. gale in the context of its environment. The low diversity of strains may explain why isolating Frankia from M. gale nodules is often very difficult. In a previous study, only 30 of 3,000 (1%) M. gale nodule lobes prepared yielded isolated Frankia strains (28). Interestingly, one of the isolates obtained in that study had characteristics of group II strains. Therefore, M. gale exhibits promiscuity but at a very low frequency in nature.

C. peregrina nodules have an intermediate level of diversity with all strains grouping within group I. Some sequences are common since three accounted for half of the total number of sequences obtained. It is possible that additional sampling would show a distinct preference for one or a few sequences within Comptonia nodules. Although M. pensylvanica and C. peregrina share a predilection for colonizing dry sites, C. peregrina apparently does not commonly associate with Frankia strains from group II, at least at a sufficiently high frequency to be detected in our sample. This latter observation suggests that C. peregrina is genetically disposed to maintain a more restrictive level of symbiont specificity than M. pensylvanica.

Estimates of the total strain richness shown in Table 2 suggest that dozens of Frankia strains can inhabit myricaceous root nodules. Although that conclusion is likely to be correct, the estimates for individual species must be interpreted with caution since they depend on having a sufficient sample for the analysis. Calculating the ACE values by using random addition of samples (Fig. 3) indicates that the total number of strains that might be found to infect M. gale and C. peregrina in the field is likely to be higher than that shown in Table 2 since the ACE continues to increase with sample addition. On the other hand, the ACE for M. pensylvanica stabilizes, suggesting that the diversity of 16S sequences for these nodules is reflected in the sample used. The low evenness (dominance) observed in M. gale and C. peregrina requires that a much larger sample set be used to estimate total species richness. It is likely that new strains would continue to emerge as more samples were analyzed to the extent that, if enough nodules could be examined, the total species richness in these plants may approach that of M. pensylvanica.

The characteristics of the niches occupied by the plants may largely explain the differences in diversity observed. C. peregrina and M. pensylvanica inhabit sandy soils or sand dunes that experience prolonged periods of dryness. Such conditions may be more conducive to the accumulation of frankia spores rather than vegetative hyphae whose growth would be limited by lack of moisture and organic matter. Thus, bayberry and sweet fern roots may be exposed to a high diversity but low overall populations of infective cells with the result that symbioses are formed with a greater variety of strains. In contrast, M. gale typically grows in and around wet habitats such as bogs and ponds, where the soil is water saturated and often acidic (27). Such factors are known to affect the ability of some Frankia strains to reinfect their host plants (6). Thus, the dominant Frankia strains in M. gale nodules may be those capable of living in an environment characterized by low oxygen and acidic pH. It would be interesting to determine if those strains have specialized in nodulating M. gale to the exclusion of other actinorhizal species.

The extent to which the observed levels of diversity stem from the ecology of Frankia populations in the soil in addition to plant and microbial genetic factors is unknown. Nevertheless, the variability we have observed shows that the diversity of Frankia strains that occupy field-grown nodules in the Myricaceae differs at the plant species level rather than at the genus or family levels. The common occurrence of identical sequences in nodules from different plants at widely separated sites indicates that the host ranges of some groups of Frankia strains in the field can be quite broad and that some strains are cosmopolitan (Table 1). Their distribution may best be described in terms of a geographical mosaic, and their interactions with their hosts take place within the constraints of each part of the mosaic.