INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Recently, it has been shown that changes in the microbial community composition of soils can be brought about by either abusive or improved management practices (1, 11, 12, 26, 37). However, the factors that dictate how such changes in composition occur are generally unknown. Because soil structural properties often change when management practices are modified (10), interest has been shown in the phenomenon of soil aggregation and its influence on the distribution of microorganisms and their activities throughout the soil fabric (2, 13-15, 21, 22, 29). We wished to examine the relationship between soil aggregate size and the distribution and dynamics of soil bacteria under a cover crop management system which has the potential to promote change in soil structural properties (21).

In this particular study, we chose to examine the distribution of an individual bacterial species, Rhizobium leguminosarum bv. trifolii. Despite being recognized primarily for its ability to form symbiotic associations with species of Trifolium, the bacterium is a successful soil saprophyte with the ability to support soil populations ranging from 10⁸ cells g¹ in rhizosphere to <10 cells g¹ in the prolonged absence of the host plant (4). Furthermore, well-established methods for extracting populations of Rhizobium from soil and enumerating them with fluorescent antibodies exist (3, 8). Although several studies have examined the impact of soil pore size on the fate of introduced rhizobial strains (16, 23-25), there are no reports describing the distribution of an indigenous population of Rhizobium sp. or any other bacterial species across different soil aggregate size classes. The presence of red clover (Trifolium pratense L.) in the winter cover crop-summer vegetable crop rotational system at the North Willamette Research and Extension Center, Aurora, Oreg., provided us with an opportunity to compare the dynamics and distribution across aggregates of R. leguminosarum bv. trifolii under conditions in which the host legume was intermittently present each year and the presence of nonhost plants and soil cultivation provided additional physical and biological pressures on the soil ecosystem.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Experimental site description. Soil samples were collected from a vegetable crop rotational experiment initiated in 1989 at the North Willamette Research and Extension Center on a Willamette silt loam (Pachic Ultic Argixeroll). The general characteristics of the site have been described in detail elsewhere (7). Field treatments include three winter management treatments in a vegetable crop rotation that alternates two summer crops, sweet corn (Zea mays L. cv. Jubilee) and broccoli (Brassica oleracea L. botrytis group cv. Gem) grown between May-June and September of alternate years. The three winter treatments are fallow and cover crops of either red clover (T. pratense L. cv. Kenland) or Celia triticale (X Triticosecale wittmack). The cover crops are relayed into the summer crop during late July and are grown until the following April, whereupon they are tilled into the soil. Hereafter, the three winter management treatments will be referred to as fallow, cereal, and legume, respectively. The experimental design is a randomized complete block, split-plot with four replications of each treatment. Winter cover crops are the three main treatments, and three N rates are subplots. In this study, soil samples were taken only from the subplots receiving 0 kg of N ha¹ to avoid the influence of N fertilizer on legume nodulation.

Soil sampling protocol. In June and September 1994 to 1996, soil samples were collected to a depth of 20 cm from each of the four field replicate plots of each of the three winter management treatments. The soil cores were taken with a 2.5-cm-diameter tube auger, gently broken up and mixed by hand in the field, and transported to the laboratory in Ziploc bags. In June 1994, the size of the nodulating population of R. leguminosarum bv. trifolii was examined in whole soil samples from each replicate plot of each of the three treatments. In September 1995 and 1996, the population density of R. leguminosarum bv. trifolii was examined in different size classes of aggregates prepared from soil sampled after summer crops of broccoli (1995) and sweet corn (1996) had been harvested. In June 1996, we examined the population density of R. leguminosarum bv. trifolii in aggregate size classes prepared from soil sampled immediately after the seed bed had been prepared for planting of the summer crop, about 1 month after winter cover crop incorporation. The moisture content of the soil samples collected throughout this study period ranged between 10 and 20% (wt/wt).

Method of aggregate preparation. Our preliminary studies conducted on aggregate preparation showed that the distribution of soil among different aggregate size classes was not influenced by the soil water content at the time of sieving, provided that the latter was <15% (wt/wt). As a result, the following protocol was developed for preparation of aggregates. Field-moist soil samples were spread onto paper to a depth of approximately 1 cm and allowed to air dry for approximately 7 days in a cold room at 4°C. This treatment effectively lowered the soil water content to 10% (wt/wt). We presumed that slow drying in this manner would lessen any negative impact of drying on the Rhizobium populations. Subsamples (100 g) of soil were placed in the top of a nest of sieves and sieved for 3 min on a Tyler Ro-Tap shaker (Combustion Engineering Inc., Mentor, Ohio) into the following size classes of aggregates: <0.25, 0.25 to 0.5, 0.5 to 1.0, 1.0 to 2.0, and 2.0 to 5.0 mm. Results of preliminary experiments indicated that sieving for 3 min was sufficient to promote good separation of the different size classes (data not shown). The procedure was repeated on another 100-g portion of soil from the same field treatment. Aggregates and whole soil samples were stored in polyethylene bags at 4°C until the time of the analyses. Most-probable-number (MPN) analyses were conducted on aggregates over a 1- to 2-week period after their preparation.

MPN estimates of R. leguminosarum bv. trifolii populations. The population size of Rhizobium organisms was determined by the plant infection-soil dilution method, as described elsewhere (9). Briefly, seeds of red clover (T. pratense cv. Kenland) were surface sterilized, germinated on water agar, and transferred in pairs to sterile test tubes (20 by 2.5 cm) containing 20 ml of N-free mineral nutrient solution solidified with 15 g of Bacto Agar (Difco) per liter. Portions (5 g) of whole soil (or aggregates) were suspended and shaken vigorously in 47.5 ml of a mineral salts solution (per liter, NaCl [0.1 g], CaCl₂ [0.05 g], MgSO₄ [0.2 g], K₂HPO₄ [0.34 g], and KH₂PO₄ [0.16 g] [pH 6.5]). Fivefold dilution series were carried out to achieve final dilutions of 1:(1.56 × 10⁵), 1:(7.81 × 10⁵), and 1:(1.95 × 10⁷) of the fallow, cereal, and legume treatments, respectively. One-milliliter portions of each dilution were pipetted into each of four replicate tubes containing the red clover seedlings. The seedlings were grown in a greenhouse, as described elsewhere (9). After 6 weeks of growth, MPN values were calculated with the MPNES program (36). In June 1994, MPN determinations of the population sizes of nodulating R. leguminosarum bv. trifolii were carried out on samples of soil obtained from each of the four replicate field plots of each of the three treatments (12 soil samples). In September 1995 MPN estimates were carried out on composite samples of whole soil prepared by combining portions of soil from each of the four replicates of a particular field treatment (three samples per date). In the case of aggregates, portions of a specific aggregate size prepared from each of the four field replicates of a treatment were pooled, and an MPN determination was carried out on each of those composite samples (15 samples per date). In June 1996, MPN determinations of population size were made only on composite samples of each of the aggregate size classes from the legume treatment (five samples) and on composite whole soil samples from the three treatments (three samples).

Collection and analysis of isolates from the indigenous soil population of R. leguminosarum bv. trifolii. As many nodules as possible were recovered from plants nodulated by the highest soil dilutions of the first MPN estimate (June 1994). One hundred fifty-seven isolates (one per nodule) were obtained from the three treatments (47 to 62 isolates per treatment). In April 1996, between 8 and 10 red clover plants were recovered from each of the four replicate plots of the legume treatment. Approximately 10 nodules were obtained from each of the plants in a replicate plot, combined, and surface sterilized, and isolates were obtained by standard procedures (35). We attempted to obtain an isolate from each of about 40 nodules per replicate. We were successful in getting 120 isolates into culture with this strategy. The isolates were screened by immunofluorescence with fluorescein-labeled immunoglobulin conjugates (FAs), as described elsewhere (20).

Immunofluorescence enumeration of Rhizobium serotypes in soil. Bacteria were extracted and enumerated from rhizosphere and nonrhizosphere soil, as described elsewhere (8). Population densities of serotype AR18 were determined in aggregate size classes prepared from the four replicates of each of the three treatments in June and September 1996. Serotypes AS6, AS36, and AR6 were enumerated only in aggregates prepared from the four replicates of the legume treatment.

Enumeration of total soil bacteria. Total soil bacteria were enumerated by epifluorescence microscopy with the DNA-specific stain DAPI (4',6-diamidino-2-phenylindole), as described elsewhere (5).

Statistical analysis. At each sampling time, the distributions of AR18 and total soil bacteria across aggregate size classes from the three treatments were analyzed by repeated-measures analyses of variance (ANOVA); aggregate size was the repeated term (28). To compare the distribution of different Rhizobium serotypes across aggregates from the legume treatment, individual ANOVA were conducted on each serotype by comparing the densities found in the five aggregate size classes. Main effects were separated by Fisher's least significant difference test at P of 0.05.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of antigenically distinguishable serotypes in the soil population. Serological analysis of the Rhizobium isolates recovered from nodules on field-grown plants and from terminal dilutions of the MPN revealed that antigenically distinguishable serotypes existed in the Willamette silt loam. Forty-one of 120 isolates from field nodules reacted positively with an immunoglobulin conjugate to serotype AR18. In addition, a substantial percentage of isolates recovered from the MPN terminal soil dilutions of fallow (38%), cereal (19%), and legume (23%) treatments belonged to serotype AR18 (Table 1). Immunofluorescence analysis confirmed the existence of serotypes AS36, AS6, AR6, and AR18 directly in soil samples recovered from the fallow, cereal, and legume treatments (Table 2). Populations of serotypes AS6 and AS36 existed at similar densities in nonrhizosphere soil in all three treatments, whereas the population densities of serotypes AR6 and AR18 were 5 to 10 times greater in nonrhizosphere soil from the red clover treatment than from fallow and cereal soil. Although the rhizosphere/nonrhizosphere ratios for Rhizobium serotypes and the total soil bacterial population were always higher in the legume treatment than in the cereal treatment, the ratios in the former varied widely among serotypes, ranging between 2.8 and 41 (Table 2). Statistically significant differences between rhizosphere and nonrhizosphere populations were observed for AR6, AR18, and total bacteria in the legume treatment and for AR18 and total bacteria in the cereal treatment.

Distribution of R. leguminosarum bv. trifolii across aggregate size classes determined by the MPN procedure. MPN analysis of whole soil showed that the population density of Rhizobium organisms in soil of the legume treatment was greater than in soil from either the fallow or cereal treatments (Fig. 1). The magnitude of this difference varied from year to year and ranged between 2- and 50-fold. Aggregates recovered from the legume treatment carried Rhizobium population densities that ranged from 19 to 284 and from 5 to 77 times greater than the populations from the corresponding aggregate size classes of the fallow and cereal treatments, respectively (Fig. 2). In September 1995, however, there were two exceptions in the cereal treatment. Size classes of 0.25 to 0.5 mm and 1.0 to 2.0 mm contained Rhizobium densities (6.0 × 10⁴ g of soil¹) similar to their counterpart size classes in the legume treatment (8.2 × 10⁴ to 10.2 × 10⁴ g of soil¹).

View larger version (35K): [in this window] [in a new window]   FIG. 1.   Population densities of R. leguminosarum bv. trifolii in whole soil samples taken from the three winter cover crop treatments, as determined by an MPN procedure. For June 1994, the error bars represent the standard errors of MPN determinations conducted on each of the four field replicates of a treatment. For the other sampling times, MPN determinations were conducted on single composite samples of soil prepared by combining the four field replicates of each treatment. In the latter cases, the error bars represent the upper limits of the confidence intervals (P < 0.05).

View larger version (39K): [in this window] [in a new window]   FIG. 2.   Population densities of R. leguminosarum bv. trifolii in different soil aggregate size classes prepared from the three winter cover crop treatments, as determined by an MPN procedure. Determinations were conducted on a single composite sample of a specific aggregate size class produced by combining portions of that size class which had been prepared from each of the four field replicates of the respective field treatment. Missing data account for the absence of a value for the 1.0- to 2.0-mm size class of fallow treatment (September 1995). Error bars represent the upper limits of the confidence intervals (P < 0.05).

Distribution of Rhizobium serotypes across aggregate size classes, as determined by immunofluorescence. In June and September 1996, the population density of AR18 varied significantly as a function of aggregate size (Fig. 3 and Table 3). Furthermore, the interaction between aggregate size and treatment was significant, indicating that AR18 distribution among aggregate size classes differed among the treatments (Table 3). For example, in June, a significantly greater density of AR18 was found in the aggregate size class of 0.5 to 1.0 mm of the fallow treatment than in the 0.25- to 0.5-mm and 2.0- to 5.0-mm size classes of the same treatment (Fig. 3). Similarly, the 1.0- to 2.0-mm size class of the legume treatment contained a significantly greater density of AR18 than the size classes of <0.25, 0.25 to 0.5, and 2.0 to 5.0 mm. However, in September, AR18 densities in all of the size classes of >0.5 mm of the fallow treatment were greater than those found in the <0.5-mm size classes. In contrast, in the legume treatment, one of the smallest size classes (0.25 to 0.5 mm) contained a significantly higher density of AR18 than was found in the size classes of <0.25, 1.0 to 2.0, and 2 to 5 mm. In June 1996, although there was a trend for the lowest densities of serotypes AS6, AS36, and AR18 to be found in the <0.25-mm size class of the legume treatment, the population densities did not differ significantly across the different aggregate size classes (Fig. 4). Nevertheless, in September a shift had occurred in population distribution, with the highest density of AS36 being found in the size class of 0.25 to 0.5 mm and the highest densities of AR6 in the 0.25- to 0.5-mm and 0.5- to 1.0-mm size classes.

View larger version (61K): [in this window] [in a new window]   FIG. 3.   Comparison of the distribution of R. leguminosarum bv. trifolii serotype AR18 across different aggregate size classes prepared from the three winter cover crop treatments in June and September 1996. Within each cover crop treatment and time, bars (mean values) not headed by the same letter (a, b, or c) are significantly different at P of <0.05.

View larger version (64K): [in this window] [in a new window]   FIG. 4.   Comparison of the population densities of four R. leguminosarum bv. trifolii serotypes (AS6, AS36, AR6, and AR18) in different aggregate size classes prepared from the legume winter cover crop treatment in June and September 1996. For each serotype and time, bars (mean values) not headed by the same letter (a, b, or c) are significantly different at P of <0.05.

Distribution of total soil bacteria across aggregate size classes. The densities of total soil bacteria in aggregates from the fallow, cereal, and legume treatments were determined in June and September 1996 (Fig. 5). At both sampling times, the total soil bacterial population varied significantly as a function of aggregate size and the distribution across aggregate size classes was influenced by the treatment (Table 3). Two distinct patterns of distribution were observed across aggregate sizes. In the fallow treatment (June and September) and in the legume treatment (September), the two extremes of aggregate size classes (<0.25 and 2.0 to 5.0 mm) contained the lowest densities of total bacteria. In the cereal treatment (June and September) and in the legume treatment (June) the <0.25-mm size class contained significantly lower bacterial densities than all larger size classes.

View larger version (63K): [in this window] [in a new window]   FIG. 5.   Comparison of the population densities of total soil bacteria in different aggregate size classes prepared from the three winter cover crop treatments in June and September 1996. Within each cover crop treatment and time, bars (mean values) not headed by the same letter (a, b, or c) are significantly different at P of <0.05.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Heterogeneous distribution of microbial biomass and its activities in soil aggregates of different sizes has been observed previously (13, 22, 29). However, our data are the first of their kind to illustrate that the population of a soil-borne bacterial species can be heterogeneously distributed in soil fabric. Some features of the distribution of R. leguminosarum across the different soil aggregate size classes were particularly interesting. While it is logical to expect that the population of nodulating Rhizobium organisms would be greater in a legume cover crop soil than in either the cereal or fallow soil, aggregate size classes in which the population densities of nodulating Rhizobium were similar in the cereal and legume treatments were identified. These data indicate the existence of microniches in the legume cover crop soil in which the Rhizobium population is stimulated to different degrees by the presence of the legume.

Data supporting the idea that microaggregates (<0.25 mm) are a less favorable habitat than macroaggregates for microbial activity exist in the literature (13, 31, 34). Microaggregates are thought to represent the primary nuclei of the aggregation process and to be relatively isolated from current biological events occurring in soil, and they contain a low level of organic matter which is complex and somewhat recalcitrant to microbial attack (34). Nonetheless, other reports have shown that microaggregates can be as biologically active as macroaggregates (2, 17, 18, 29). Although some of our microscopic data support the idea that aggregates of <0.25 mm are less favorable habitats for microbial growth in our soil because they contain either the lowest or one of the lowest DAPI (Fig. 5) and immunofluorescence (Fig. 4) counts, some of our Rhizobium MPN data contradict that notion (Fig. 2). For example, in September 1995 and 1996, we observed that the <0.25-mm size class of the legume treatment contained densities of nodulating Rhizobium organisms that were equal to or greater than those found in larger size aggregate size classes. These data might indicate that the population of nodulating Rhizobium in the <0.25-mm size class responds more favorably to growth conditions in this niche than the majority of the soil bacterial population. Further experiments are required, however, to investigate the possibility that the <0.25-mm size class of aggregate simply provides better physical protection over the summer months than larger size classes for nodulating Rhizobium organisms. Furthermore, we cannot exclude the possibility that the nodulating populations in the larger size classes of aggregates might have been artificially deflated during the processing of September soil for aggregate preparation.

The significant interaction that was measured between the population size of serotype AR18 and the aggregate size and treatment (Fig. 3 and Table 3) presumably reflects the different nutrient sources and soil conditions imposed upon rhizobia during the spring months in the presence or absence of a plant and by the presence of either a host or a nonhost species. It has been recognized for many years that the plant rhizosphere is a zone where microbial activity is greater than in bulk soil and that legume rhizospheres are particularly active (4, 20). Further work is needed to determine to what extent the different distributions of the Rhizobium populations in the legume and cereal treatments can be linked to plant-associated rhizosphere processes versus differences attributed to soil disturbance and cover crop incorporation and/or the early stages of cover crop residue decomposition. The changes that occurred between June and September in the distribution of all Rhizobium serotypes in the legume treatment presumably reflect the changes occurring in soil conditions over the summer months. During this period, the factors influencing microbial activity are complex, promoted in all treatments by growth of the summer crop and the availability of irrigation water and differentiated among treatments by the presence or absence of decomposing cover crop residues. Further experiments are planned to compare the growth rates and turnover of Rhizobium organisms in the different aggregate size classes of the three treatments at the different sampling times.

Since AR18 isolates were recovered from the nodules of red clover plants in the terminal soil dilutions of all three field treatments in the 1994 MPN study, we can assume that the MPN values obtained for the size of the nodulating population of R. leguminosarum bv. trifolii also reflect (to an approximation) the density of the nodulating population of serotype AR18 in the soil. A comparison of the immunofluorescence direct count (IFDC) microscopy and MPN values of the AR18 population from the legume treatment show that MPN estimates range between 5 and 65% and between 10 and 30% of IFDC estimates from June and September soil samples, respectively. These values agree reasonably well with each other, if some latitude is given for the large confidence intervals normally associated with the MPN procedure and for the fact that a preliminary study with an immunofluorescence cell elongation assay (6) indicated that ~40% of the immunofluorescence-detectable cells of AR18 were viable (data not shown). However, in the case of the cereal and fallow soil samples, the discrepancies between the MPN and IFDC estimates were greater, with MPN estimates of AR18 representing only 5 to 0.5% of the cereal and fallow IFDC counts, respectively. A variety of possibilities might account for the discrepancy between IFDC and MPN estimates in these treatments. First, unrelated bacteria carrying surface antigens similar to rhizobia might be mistakenly enumerated by the FAs, and their population sizes might vary between treatments. For example, Bohlool and Schmidt (3) reported a cross-reaction by an FA raised to Bradyrhizobium japonicum against a soil actinomycete. Second, dead Rhizobium cells might be more numerous in the absence of a host plant and persist in microsites that are inaccessible to predators (25, 27). Indeed, the immunofluorescence cell elongation assay indicated that a significantly lower (P = 0.05) percentage (~20%) of the immunofluorescence-detectable cells of AR18 were viable in the fallow and cereal treatments than in the legume treatment (data not shown). These observations add credence to the idea that protected pore space exists in soil in which nonviable or dormant bacteria can persist (16, 23, 24). Third, IFDC might enumerate a population of nonsymbiotic rhizobia that are antigenically related to the nodulating serotypes. The existence of nonsymbiotic forms of Rhizobium species in soil has been shown on several occasions over the past few years (19, 30, 32, 33), and nonsymbiotic cells have been reported to be 40 times more numerous than the symbiotic forms (30). Nonsymbiotic cells may represent a larger percentage of the population in the absence of a host plant. Further studies are necessary to distinguish between these possibilities and to determine why the percent viability of the AR18 serotype differs among treatments and if the percentage of viable cells varies across the aggregate sizes among the treatments.

Soils are complex environments in which it is generally recognized that changes in crop systems invariably cause changes in soil physical properties that influence microbial activity. It is not clear, however, to what extent physical properties control the distribution of bacteria and influence their growth, activities, and turnover. We hope that this study will stimulate the interest of other microbiologists to gain an understanding of how soil management might influence the activities of soil microorganisms by modifying soil structural properties.