TEXT

Top Abstract Text References

Bacterial diversity is now accepted as one of the components of soil function, particularly its sustainability. Endemism and clonality, which have an intrinsic spatial dimension, are studied (7, 13, 16, 17), as spatial patterns may have significant implications in soil function. In the case of bacteria, in contrast to other organisms in soil for which hierarchical models of spatial organization have been proposed (1), the spatial arrangement in soil is not well understood yet (11). It was recently shown that only 1 g of soil was needed to represent the prominent bacteria in large homogeneous grassland areas (5).

The distribution of microorganisms at the microscale is rarely explored, although due to the size of bacteria, such distribution obviously represents a functional scale where major interactions take place within the soil structure. Soil is a typical highly heterogeneous medium from both textural and structural standpoints (18), and small-scale heterogeneity of physicochemical characteristics probably results in the patchy arrangement of bacteria observed by Hattori and Hattori (12).

At present, most diversity studies of soil target community changes that follow stress (20); more rarely, a functional or taxonomic group (e.g., Rhizobium) is studied (29). This is probably linked to methodological limits: no specific media have succeeded in providing a reliable means for targeting a defined group among the numerous groups in the soil, and only recently have molecular tools done so. However, the study of spatial heterogeneity of bacterial distribution seems to require the use of a defined bacterial group, as suggested by Felske and Akkermans (5), for interpretation. The genus Nitrobacter, the genus of chemilithotrophic bacteria responsible for the majority of NO₃ formation from NO₂ in soil, is a good candidate for such studies because there is little redundancy in soil (15) and its presence in a culture is easily detected (14). The rrs-rrl intergenic spacer (IGS) of Nitrobacter seems to be appropriate for studying diversity at the subspecies level (19). Nitrobacter-specific primers, which can be used in complex DNA mixtures (10), were designed for the rrs-rrl IGS and for the rrl gene. It has been suggested (5, 6) that fingerprints of rrs ribosomal DNA extracted from soil were indicative of the numerically dominant bacteria (even with 1-g soil samples) but did not take into account the microheterogeneity, particularly the diversity of small specific communities, which requires appropriate sampling techniques (16). It is, however, recognized that the degree of diversity generally depends on sample size (23). The sample size relevant to Nitrobacter spatial arrangement at a microhabitat scale has been studied elsewhere (unpublished data). It was shown that some microsamples less than 500 µm in diameter did not harbor any nitrifiers.

The objective of this work was to evaluate the diversity of the model genus Nitrobacter at different spatial scales down to a spatial scale close to the microhabitat size. Genetic distances, obtained by amplified ribosomal DNA restriction analysis (ARDRA) of Nitrobacter-specific IGS-rrl amplicons, were compared by using "patterns" obtained from samples of soil taken at various levels of spatial organization, including microsamples, clumps, a field, and Nitrobacter species reference strains from large geographical areas.

Soil sampling. The agricultural soil studied was an Alfisol cultivated with maize. The surface soil was a sandy loam (bulk density, 1.3 mg · m³; 17% clay, 39.2% silt, and 40.4% sand; organic carbon content, 1.4%; pH(H₂O) 6.4 [8]) and was weakly structured (24). A 2- to 3-cm³ coherent soil clod (clump a) was gradually subdivided under a binocular microscope into microsamples called volumetric units (VU) by using a sterile scalpel blade. A calibrated grid was used to spot VU that fit into squares that were 50, 100, and 250 µm on each side; these VU are referred to below as size 50, size 100, and size 250, respectively. Each VU, sampled randomly because three-dimensional coordinates could not be used due to the lack of an appropriate technique, was taken up with a sterile plugged glass capillary that had been dipped into sterile, biologically inert silicone oil (SV40) and was transferred to 2 ml of defined culture medium by swirling the tip of the capillary in the medium. About 30 VU of each size were sampled. They were subsequently cultured in NO₂ oxidizer culture medium (final concentration of NO₂, 1 mM) at 28°C in the dark in the wells of 24-well microculture plates (27), until the NO₂ disappeared (about 1 to 8 weeks). The same experiment was carried out with another 2- to 3-cm³ clump of soil (clump b) that was obtained 1 year later at the same spot in the field.

Culturing and analysis of soil microsamples. After NO₂ disappeared from the culture medium (in the presence of NO₂ oxidizers), as determined by the Griess-Ilosvay spot test (14), VU positive for the presence of NO₂ oxidizers were selected from clump a. Six size 250 replicates, seven size 100 replicates, and four size 50 replicates from clump a were transferred into 100 ml of mineral medium and incubated again until NO₂ disappeared in order to obtain enough biomass for DNA extraction. For clump b, four size 250 replicates, five size 100 replicates, and seven size 50 replicates were treated similarly. Total DNA from mixed soil cultures was extracted as described by Rouvier et al. (25). The reference organisms included Nitrobacter winogradskyi, Nitrobacter sp. strain DE30, N. winogradskyi agilis, Nitrobacter sp. strain DE11, Nitrobacter vulgaris Z, 269, and nevada, Nitrobacter sp. strain LL, and Nitrobacter hamburgensis X14. These organisms represented three of the four Nitrobacter species (10, 19, 27), N. winogradskyi, N. vulgaris, and N. hamburgensis.

Isolation and serotyping. Isolates were obtained from 5-g fresh soil subsampled from a 3-kg sample of soil taken at different places in the same field and subsequently sieved (mesh size, 2 mm) and mixed. Serial 10-fold dilutions of soil were prepared, and 0.5-ml aliquots of each dilution were inoculated into six wells of 24-well microtiter plates filled with 1.5 ml of mineral medium containing NO₂ at a final concentration of 1 mM. The microtiter plates were incubated for 90 days at 28°C. Total disappearance of NO₂ from a well indicated the presence of NO₂-oxidizing bacteria (27). The contents of five wells positive for the presence of NO₂-oxidizing bacteria were transferred to new medium, and pure isolates were obtained as described by Soriano and Walker (28).

Genetic distances between ARDRA patterns for clumps of soil. The detection threshold of ARDRA gels, as tested by mixing DNA of two different strains in various proportions, indicated that the ARDRA pattern of each strain could be recognized if the quantities of DNA were in range of ratios down to 1/20 (data not shown). At larger ratios, the fragments of the minority strain were indistinguishable from the background smear. Besides, as the sample sizes were multiples of one another and the larger samples did not produce complex ARDRA patterns, the patterns observed in microsamples probably were the dominant patterns.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   ARDRA distance dendrogram of the genotypes (as calculated by the method of Saitou and Nei [26]). a, microsamples from clump a; b, microsamples from clump b. The numbers are sample numbers. The values in parentheses are the sizes of the UV (in micrometers). I, isolate. Reference organisms are indicated by boldface type. Serotyping results for microsamples and isolates are indicated when available. Bar = 0.02 nucleotide substitution per site.

Genetic distances between isolates. The 20 isolates obtained from 5 g of homogenized soil obtained at the field site yielded patterns different from the patterns of the reference organisms and from the patterns of the microsamples of soil (Fig. 1). Some isolates were as distant from the reference organisms as the microsamples were. Some isolates came from the same culture well of the most-probable-number count analysis and yielded identical patterns; isolates I21, I22, I24, I27, and I35 produced identical patterns, as did isolates I34, I33, and I19, and isolates I31 and I32.

Serotyping in microsamples. The detailed serotyping results are indicated on the dendrogram in Fig. 1; both serotype analysis and ARDRA were performed for VU. The serotyping data clearly showed that several serotypes coexisted closely in a single VU (Fig. 1). The sera did not cross-react with the reference strains although the possibility that they cross-reacted with non-Nitrobacter autochtonous cells in the soil sample cannot be eliminated. Furthermore, sera were checked against 40 isolates from the same soil on Luria-Bertani medium (data not shown). Only one isolate yielded an equivocal reaction yet did not have a typical Nitrobacter shape. Although the level of discrimination of serotyping, which corresponds to phenotypes, has not been clearly established yet (3, 19), there is at least large phenotypic diversity.

Spatial limit of identical ARDRA patterns. We showed that the diversity of Nitrobacter, in terms of genetic distances based on rrs-rrl studies, was as large in a small clump of soil as the diversity of reference strains from different geographical areas. It was possible to find identical ARDRA patterns, considered in the scope of this study a clone, in microsamples from a clump of soil but not at a larger scale. Our results point out that the size of a clone is difficult to delineate considering the low numbers of VU showing identical patterns (4 and 2) and that quantification of the extent of an individual clone's habitat requires thorough spatial investigation at a smaller scale. The spatial coordinates of microsamples would be needed to prove the existence of a relationship between the spatial distance and the genetic distance for the range of distances explored in this work.

Evolutionary implication and soil status. It is tempting to bring together our results and their evolutionary and soil function implications. Mutations are undoubtedly responsible for diversity. Nitrobacter species are slow growers, but mutations happen between divisions (21). Based on rrs gene analysis, genetic distances indicate that Nitrobacter emerged about 50 to 100 million years ago (22). Considering that the rrs-rrl IGS accumulates 10 times more mutations than the 16S rRNA (10) and given that a universal substitution rate rules bacteria evolution (21), 1% divergence in the IGS would represent 5 million years. The time estimate yielded by such calculations based on the present study (Fig. 1), about 60 million years, corresponds to the lower value for the period mentioned above. This time estimate seems inconsistent with the age of the soil, which is about 200,000 years old (i.e., Quaternary [Riss era]). Besides, the diversity results cannot be separated from the physical status of the soil. The soil studied is of glacial origin, which means that substantial mixing of imported particles occurred. The soil is also cultivated and weakly structured (24). Previous results indicated that this soil has high percolating values, as shown by the mercury porosimetry intrusion method; pores as small as 1.8 µm in diameter were connected. This is much larger than the size of the cells in the soil (2) and allows transportation by water movement for example. These conditions should provide a high potential for migration of cells during rain events and ploughing. These remarks lead to the hypothesis that migration of cells happens at different time and spatial scales on a regular basis.