Introduction

Based on numbers of colony-forming units (CFUs) along the plant root, recently, a new, dynamic concept for the spatial distribution of microbial communities in the rhizosphere was proposed: one in which the bacterial populations along the root move in a wavelike fashion with regular intervals instead of the formerly accepted distribution with peaks in the bacterial populations at exudation points along the root [24]. Experiments with wheat roots or an artificial moving nutrient source showed that populations of copiotrophic and oligotrophic bacteria fluctuated along the length of wheat roots or the path of a moving nutrient source [24, 29]. Fourier (or harmonics) analysis demonstrated that these fluctuations in densities of CFUs constituted significant waves with periods of about 30 cm when the root growth or movement rates of the nutrient source were 3–4 cm per day [24, 29].

One hypothesis is that spatial waves along the root are derived from temporal waves in the bacterial populations at any location where soil has been disturbed by nutrients released from a passing nutrient source. An impulse of nutrients would result in exponential growth of microorganisms followed by death because of nutrient limitation or related factors and subsequently by growth from partial reuse of recycled carbon sources supplemented by substrate from soil organic matter and decaying cortex cells. This concept was reinforced when transformation of spatial fluctuations in CFUs to the corresponding temporal scale showed that the period of the oscillations in time was independent of the moving rate of the nutrient source [29]. The frequencies of the oscillations in time would depend on the initial nutrient input and the growth and death rates of the microbial community as a whole [29]. Additional support for this hypothesis came from the detection of significant oscillations in time of bacterial populations (CFUs and direct microscopic counts) after addition of a moving substrate source to soil [34].

An alternative explanation for the fluctuations in densities of microbial communities could be succession of different species within these communities. Such a succession could occur in subsequent waves, each peak being represented by a different community of bacteria, or within individual peaks, so that the waxing phase of each peak, with increasing bacterial numbers, could have a particular microbial composition, and the waning phase, with declining bacterial numbers, could have a different microbial composition. In this last case, the successions within subsequent peaks could be similar. The similarity of fluctuations in Pseudomonas fluorescens populations to the fluctuations observed for total copiotrophic bacteria along roots (Van Bruggen et al., unpublished data) suggests that bacterial communities would be similar from peak to peak.

Variation in rhizosphere microbial communities has been studied by means of metabolic profiles [15, 16], fatty acid profiles [8, 26, 27], or genetic composition [17, 18, 33]. The genetic composition has been studied most frequently using DNA fragment amplification and separation by denaturing gradient gel electrophoresis (DGGE). These DGGE analyses have been carried out for microorganisms isolated on media [5, 8, 10, 11, 26], on whole communities in suspension [11, 15], and on DNA directly extracted from soil [5, 11, 12, 17, 18, 22, 32, 33]. In all of these studies, microbial communities were investigated in single rhizosphere samples comprising the total root length, random root samples [8, 10, 11, 15, 16, 22, 26, 33], or few root sections (for example, root tip, middle section, and root base) [4, 5, 17, 18, 32]. No research on the genetic composition of the microbial community has been conducted in which patterns in microbial composition were studied along the total length of roots.

The research described in this article was conducted to characterize the species composition of bacterial communities and different trophic subpopulations along roots in relation to oscillations in bacterial populations. The goals were to answer the questions (1) if fluctuations in bacterial populations along roots can be explained by fluctuations in complete microbial communities or by fluctuations in subsets of microbial communities and (2) if and how microbial succession occurs along the root in comparison to corresponding locations in the bulk soil. The experiments were conducted in two related soils, which allowed us to address the question (3) whether rhizosphere microbial densities and compositions, biodiversity, and fluctuations differed in soils with different management histories.

Materials and Methods

Experimental Setup

The spatial distributions of CFUs of oligotrophic and copiotrophic bacteria and their species compositions were determined along roots of 4-week-old wheat plants in two soils. The experiment was repeated three times. For each experiment, three root observation boxes per soil type were set up in a greenhouse at Wageningen University. The experiments were initiated in three subsequent weeks to allow for the time needed for sampling and dilution plating of 20–25 root samples per plant [24].

Root observation boxes were constructed from PVC irrigation pipes (100 cm long, 12 cm wide) cut longitudinally into halves, with perforated PVC sheets on the bottom. One linear irrigation tube (1.6-cm diameter) with Netafim® Woodpecker button drippers at 20-cm intervals was attached at the back on the inside along the length of each box (four drippers per box) to maintain uniform soil moisture at various depths in each box. The front of each box was covered by plastic foil and by a Plexiglas sheet to prevent disturbance of the roots at harvest time.

Soils

Two sandy soils were used: one from a conventional farm and one from a farm that had been organic for 14 years. Both farms were located at Noordberg in Heelsum, the Netherlands, with the conventional farm on top of a hill and the organic farm in an eroded valley. The soils will be designated as the conventional soil and the organic soil, although it should be noted that next to management type, also the soil histories were different. The sand and silt contents were 91.4 and 5.6% for the conventional soil and 91.9 and 5.4% for the organic soil, respectively. The organic matter contents were 3.0% for the conventional and 2.9% for organically managed soils, respectively. There were stubbles of barley in the conventionally managed field at the time of sampling (March 2000), whereas in the organically managed field, a Triticale crop had just been seeded. The conventionally managed soil had been limed with CaCO3 residues from the sugar beet industry at 8 tons/ha and had been fertilized with 70 kg N/ha in the form of calcium ammonium nitrate. The organically managed field had not been limed and had been fertilized with cow manure plus straw. The original pH water was 6.7 and 4.9 for the conventional and organic soil, re-spectively. To equalize the pH, 50 mg of ground hydrated lime [Ca(OH)2] per 100 g dry soil was added to the organic soil and mixed thoroughly. After incu-bation, the pH of the organic soil was raised to 5.9. The original field moisture of both soils was 13.1% (dry weight). Soils were dried and sieved (∅ 8 mm) before storage. At the start of these experiments, both soils were rewetted to original field moisture content and stored in plastic bags in the greenhouse for 7 days prior to use. Nitrogen (N), Phosphate (P) and Potassium (K) determinations were performed according to standard protocols by spectrophotometric analysis by segmented flow analyzer or with an element analyzer. The total N and P contents were 933 and 732 mg/kg for the conventional soil and 833 and 683 mg/kg for the organic soil, respectively. The total available N, P, and K contents were 19.7, 6.2, and 125.0 mg/kg for the conventional soil and 16.4, 6.2, and 34.6 mg/kg for the organic soil, respectively.

Plant Culture

Triticum aestivum L. var. Trawler seeds (not treated with fungicides) were incubated on moist filter paper for 3 days. Two pregerminated wheat seeds were planted in each box. The Plexiglas was returned and covered by aluminum foil to maintain darkness. The pots were tilted at an angle of 60° with Plexiglas downward.

Irrigation water was supplied for 1 min per day at a rate of 1 L/h per dripper at a water pressure of 1 bar. The temperature was maintained at 20°C during the day and 15°C at night. Natural light was supplemented with high-pressure sodium lamps (Son-T, 150 W, Philips, Eindhoven, The Netherlands) for 14 h per day. Relative humidity was maintained at 60%.

Root Observations and Sampling

Root growth was monitored weekly as reported previously [24]. Four weeks after planting, one root per soil type was selected for sampling. As much as possible, individual roots were selected that were straight, not intermingled with other roots, and about 75 cm long. Root sections (1.5 cm long) together with small amounts of soil (see Dilution Plating from Soil) were cut. Alternate sections were used for colony enumeration by dilution plating or direct DNA extractions, both about 20–25 samples per box. Upper parts of the roots were not sampled when the cortex was decaying. Lateral roots were counted and removed from the main root sections. The rhizosphere was considered to consist of the root surface and tightly adhering soil. Small bulk soil samples at 3- to 5-cm distance from the main roots at corresponding depths to the root sections were also collected for bacterial isolations, determination of soil moisture content, and for direct DNA extractions. All samples were collected in preweighed Eppendorf tubes, immediately placed on ice, and processed directly afterwards (plating and moisture content samples) or frozen at −80°C (for DNA extractions).

Dilution Plating from Soil

The rhizosphere samples (on average, 0.143 ± 0.069 g dry weight) and bulk soil samples (on average, 0.320 ± 0.100 g dry weight) were weighed precisely and were suspended for colony counts in 0.9 mL of sterile deionized water. Bacterial suspensions were vortexed for 20 s and sonicated in an ultrasonic cleaner (Bransonic 12, Branson Cleaning Equipment Co., Shelton, CT) for 5 min. Tenfold serial dilutions of these suspensions were plated in 6-fold (high carbon medium) or 9-fold (low carbon media) for isolationof copiotrophic and oligotrophic bacteria, respectively [24]. The high nutrient medium contained 0.5 g MgSO4·7H2O, 0.5 g KNO3, 1.3 g K2HPO4·3H2O, 0.06 g CaNO3·4H2O, 2.5 g glucose, 0.2 g enzymatic casein hydrolysate, and 15.0 g Bacto agar per liter and 100 ppm sterile cycloheximide. The total amount of carbon was estimated at 1000 mg C/L of medium. The low nutrient medium was similar to the high carbon medium but 100-fold diluted and contained Agar Noble (Difco Laboratories, Detroit, MI) instead of Bacto agar. After incubation for 60 h (copiotrophs) or 2 weeks (oligotrophs) at 25°C in the dark, bacterial colonies were counted and colony-forming units (CFUs) were calculated per gram dry soil.

To determine the biodiversity and fluctuations in different trophic groups of bacteria, polymerase chain reaction (PCR) DGGE (see next paragraph) was performed on bacterial colonies. Therefore, all bacterial colonies were scraped from the serial high- and low-substrate agar dilution plates. Per sample, between 100 and 200 colonies were collected from different (six or more) replicate plates, and concentrated suspensions with copiotrophic and oligotrophic bacteria, respectively, were saved in the −80°C freezer. Oligotrophic plates overgrown with spreading bacteria were not sampled to avoid bias.

DNA Extraction and PCR DGGE Analyses

All samples of experiment II were analyzed for DNA composition. Bacterial DNA was extracted directly from 25 samples of rhizosphere soil per root (±0.1 g) and 25 bulk soil samples (±0.3 g) of corresponding depth after 4 months of storage in the −80°C freezer. DNA was also extracted from suspensions with copiotrophic and oligotrophic bacteria after 4 and 8 months, respectively, of storage at −80°C. Thus, DNA compositions of copio- and oligotrophic bacteria cannot be compared, but treatment effects can be compared. A protocol based on cell lysis by bead beater treatment ([19], modified by [3]) was used for the extraction of nucleic acids from rhizosphere, bulk soil, and suspensions of copiotrophic and oligotrophic bacteria. The rhizosphere and bulk soil samples were further purified as described by Duineveld et al. [4].

PCR products were generated with the eubacterial primers U968GC and L1401r [12] using a touchdown scheme [23] for 30 thermal cycles. The recovery of the approximately 450-bp-long products was checked by standard agarose electrophoresis and ethidium bromide staining. Samples were randomized on the DGGE gels.

DGGE was performed using the Dcode™ system (Bio-Rad Laboratories, Hercules, CA). We used 6% acrylamide gels (37.5 acrylamide: 1 bisacrylamide) with a 45–60% denaturing gradient as defined by Muyzer et al. [20] to separate the generated amplicons (100% denaturant is 7 M urea and 40% formamide) and an 8% acrylamide stack with no denaturing agents. The gels were poured from the top in the Dcode template and prepared with Gelbond PAG film (Amersham Pharmacia Biotech AG, Uppsala, Sweden) to one side, using a gradient maker and a Heidolph Pumpdrive set at 4 mL/min. Electrophoresis was performed in 0.5× Tris–acetate–EDTA buffer for 16 h at 100 V at a constant temperature of 60°C.

Two types of markers were used on the gels: a synthetic mixture of amplified 16S rDNA of bacterial clones with known different denaturing characteristics and two randomly chosen reference mixtures of 16S rDNA fragments amplified on mixtures of the copiotrophic or oligotrophic CFUs, respectively. The synthetic marker consisted of fragments from 16S rDNA clones Eubacterium halii A07, Fusobacterium prausnitzii-like A10, Butyrivibrio fibrisolvens-like A11, Eubacterium plautii-like A27, Clostridium celerecrescens-like A54, Ruminococcus obeum-like A57, Eubacterium formicigenerans-like A71 [36], and Bifidobacterium XII provided by Hans Heilig, Department of Microbiology, Wageningen University, The Netherlands.

Gels were stained with Bio-Rad's Silver Stain (Bio-Rad Laboratories) according to the manufacturer's protocol, but using the protocol for gels > 1 mm thick instead of 0.5–1 mm to compensate for the barrier formed by the Gelbond. After staining, the gels were preserved for at least 1 h in Cairn's preservation solution of 25% ethanol (v/v) and 10% glycerol (v/v), covered by a permeable cellophane sheet (Amersham Pharmacia Biotech AG) and dried overnight at 60°C.

Statistical Analysis

CFUs in the rhizosphere and bulk soil were compared using one-sided paired t-tests for each soil and experiment separately. CFUs were checked for correlation with soil moisture contents for the different soils and numbers of lateral roots in each experiment with Pearson's correlation coefficients. Densities of CFUs per gram of dry soil were subjected to harmonics analysis (also called time series or Fourier analysis) for each root separately as described previously [24, 29]. For this analysis, observation points need to be equidistant. Because there were some irregularities in the data set due to missing observations (less than 3%), a cubic spline interpolation procedure was used to estimate values at these locations. Before harmonics analyses, the data were smoothed using Hann's three-point window [25]. Significant harmonics, their period, phase, and amplitude were determined at 0.1 and 0.05 significance levels. Characteristics of the harmonics were compared for the different soils.

Scanned images of the DGGE gels were analyzed with Phoretix 1D (NonLinear Dynamics Ltd, Newcastle-upon-Tyne, UK), and the relative density and mobility of the amplified fragments—which we will call amplicons—were calculated. Data of different DGGE gels were compared by referring to the synthetic marker and reference mixtures. 16S rDNA amplicons detected by DGGE were considered to represent dominant bacterial groups. To be detectable, an amplicon has to make up at least 0.1–1% of the total community [5, 12, 20]. The DGGE banding patterns were analyzed in the following ways.

(1) Discriminant analysis. The log10-transformed intensities of DGGE amplicons of oligotrophic bacteria, copiotrophic bacteria, and direct DNA extractions were subjected separately to both stepwise and canonical discriminant analyses (SAS Institute, Inc., Cary, NC). Depending on the number of amplicons in each data set relative to the number of observations (and thus the degrees of freedom), the data were first split in smaller data sets with one third of the variables (bands). The most significant data of stepwise and canonical analysis of the separate sets were then combined in a new data set and analyzed again with stepwise and canonical discriminant analyses. Grouping of observations in classes for discriminant analysis was based on management type, location (rhizosphere versus bulk soil), increasing population densities versus decreasing ones, and peaks versus valleys in colony numbers along the root.

(2) Harmonics analysis. The intensities of individual amplicons were, in turn, analyzed along the root with harmonics analyses [24]. As described above for the CFU data, the data were smoothed and detrended, and significant harmonics and their characteristics were determined at 0.1 and 0.05 significance levels. A distinction was made between amplicons synchronous and asynchronous with the CFU harmonics (in phase and out of phase) and amplicons restricted to the root tip or stem base area. All other types of distributions, including harmonic distributions with other numbers of harmonics, were clustered in a group “miscellaneous.”

(3) Biodiversity. Bacterial diversity in the samples was estimated in two different ways: as S, species richness (the number of DGGE fragments detected disregarding their relative intensities), and H, the Shannon index of bacterial diversity. The Shannon diversity index was calculated as H′ = −P i log P i based on the relative band intensities, as formulated by Eichner et al. [6]. P i was defined as n i /N, where n i is the area of a peak in intensity and N the sum of all peak areas in the lane profiles. Harmonics analysis on the two indices of biodiversity was performed as described above. The biodiversity in the rhizosphere and bulk soil was compared using one-sided paired t-tests. Cross-correlations at 0.1 and 0.05 significance levels between CFUs and bacterial diversity were determined for both rhizosphere soil and bulk soil samples.

Results

Root Growth

The average root extension rates were 2.95 (±0.25) cm/day in the organic soil and 2.39 (±0.21) cm/day in the conventional soil at 15/20°C, slightly more than half the rate observed previously in California at 15/25°C [15]. The plants were growing satisfactorily, although some symptoms of Fusarium root rot appeared close to the root base in the last week of each experiment and especially in the conventional soil. Rhizosphere and soil samples were collected only from the healthy parts of the roots; infected areas with decaying cortex close to the root base were avoided.

CFUs Along the Roots

The numbers of copiotrophic and oligotrophic CFUs were significantly higher in the rhizosphere soil than in the bulk soil for both soils and in all three replications of the experiment, thus showing a strong rhizosphere effect (Table 1). Overall organic and conventional rhizospheres did not differ significantly in the number of both oligotrophic and copiotrophic CFUs, but the organic bulk soil samples did contain significantly more oligotrophic and copiotrophic bacteria. The CFU numbers were neither correlated with soil moisture content nor with the number of lateral roots in the sampled root sections (data not shown).

Table 1 Geometric means with standard errors of copiotrophic and oligotrophic CFUs in rhizosphere and bulk soils and the statistical analyses for rhizosphere effect and differences between soil types (three experiments combined)
Full size table

Distributions in the CFUs along the root showed significant harmonics in the rhizosphere soil from both farms in all experiments (for an example, see Fig. 1). The densities were high at the root tip, declined, rose to high levels again around 25 cm from the root tip, declined again, and increased again one or two times after similar intervals. In the organic soils, four peaks were observed over 75-cm root length after 4 weeks of root growth, whereas three peaks were observed in the shorter (maximum 65 cm) roots in the conventional soil. The periods of the harmonics in distance along the roots were similar for both conventional and organic soil.

Figure 1
figure 1

Harmonic curves and experimental data of colony forming units (CFUs / g dry soil) of copiotrophic (A) and oligotrophic (B) bacteria in wheat rhizosphere and bulk soil from an organic farm plotted against distance (cm) from the root tip (experiment II).

Characteristics of the harmonics:

Trophic group of bacteria

Harmonic numbera

Amplitude × 108 CFUsb

Phase (cm)c

Period (cm)d

Contribution to variance (%)

F-estimatede

Copiotrophics

3

1.54

−5.94

25

29.7

3.42**

4

1.74

9.26

19

38.3

4.40**

Oligotrophics

3

1.33

−2.94

25

23.0

2.76*

4

1.65

9.20

19

35.4

4.24**

  1. aNumber of oscillations.
  2. bHalf the difference between maximum and minimum CFU numbers in an oscillation.
  3. cRelative position to root tip of the first full oscillation.
  4. dDuration of one oscillation.
  5. eSignificance levels: * = 0.10 and ** = 0.05.

Full size image

Bacterial Community Composition

On the DGGE gels, differences in presence and intensity of amplicons were clearly visible for DNA directly extracted from organic and conventional soil, both bulk and rhizosphere soils, as well as for DNA from oligotrophic and copiotrophic colonies (Fig. 2). Combining DGGE data from the organic and conventional soils, in total, 33 copiotrophic amplicons and 52 oligotrophic amplicons could be distinguished on gels of DNA extracted from colonies on C-rich and C-poor agar plates, indicating more diversity among the oligotrophic than the copiotrophic bacteria. On gels with DNA directly isolated from soil, a total of 94 dominant eubacterial amplicons could be recognized, more than in DNA extracted from either oligotrophic or copiotrophic colonies.

Figure 2
figure 2

Photograph of a 45–60% denaturing gradient gel with randomized samples of total 16S rDNA from organic rhizosphere soil (OR), organic bulk soil (OBS), conventional bulk soil (CBS), and marker lanes with synthetic marker (M1) and a mixture of oligotrophic and copiotrophic DNA (M2) (experiment II).

Full size image

Individual amplicons could be followed along the root, and their plotted intensities showed distinct distribution patterns. Amplicons with the highest intensity often showed a harmonic distribution synchronous to the harmonics of CFUs over the length of the root, in phase with the CFU harmonics. The synchronously fluctuating amplicons from directly extracted DNA formed 19.2% of all bands and represented 37.4% of the total band intensity. Low-intensity amplicons sometimes fluctuated in antiphase with the CFU oscillations or were limited to certain areas such as to the root tip or the stem base (illustrated in Fig. 3). All other possible distributions were assigned to a “miscellaneous” group (Table 2). This miscellaneous group includes random or little varying fragments. For all DNA-based data sets (extraction from copiotrophic bacteria, oligotrophic bacteria, and directly from soil), the in-phase group of bands had a higher total band intensity than the out-of-phase group.

Figure 3
figure 3

Relative band intensities of single amplicons of 16S rDNA directly extracted from the rhizosphere of wheat grown in organic soil plotted versus distance from the root tip. (A) Amplicon at relative band position 170, distribution synchronous, and in phase with copiotrophic CFUs; (B) amplicon at relative band position 338, synchronous distribution in antiphase with oligotrophic CFUs; (C) amplicon at relative band position 44, a more local miscellaneous distribution, restricted to the top half of the root. For the trends of the harmonic curves for the oligotrophic CFUs and capiotrophic CFUs see Figure 1.

Full size image
Table 2 Groups of amplicon distributions along the root: synchronous with the CFU oscillations (in and out of phase), restricted to the root tip or stem base area, or miscellaneous (all other distributions); and average percentages of the number of amplicons and of band intensities per group for 16S rDNA directly extracted from conventional or organic rhizosphere soil or indirectly extracted from copiotrophic or oligotrophic colonies
Full size table

Discriminant analysis proved a powerful tool to distinguish among amplicon intensity patterns for different treatments, locations, and phases in the harmonic waves of CFUs. This analysis takes into account both presence and absence of amplicons as well as their relative intensity when present. All three methods of DNA extraction (from copiotrophic colonies, oligotrophic colonies, and directly from soil) resulted in significant distinctions among classes of sample origin: organic versus conventional farms and rhizosphere versus bulk soil, or locations along the root where CFUs increased/decreased or where peaks/valleys in CFUs occurred along the length of the root (Table 3). Most of the variation in amplicons and their relative intensities explained by canonical variable 1 separated organic and conventional soil (Figs. 4a–c). Canonical variable 2 separated rhizosphere and bulk soil from the conventional farm very well, but separated the samples from the organic farm less clearly (Figs. 4a, b). Rhizosphere and bulk soil from the organic farm were separated in the analysis by the less discriminating canonical variable 3 (Table 3, graph not shown).

Table 3 Discriminant analysis of intensities of amplicons from the DGGE analysis of 16S rDNA extracted from copiotrophic colonies, oligotrophic colonies, or directly from wheat rhizosphere or bulk soil
Full size table
Figure 4
figure 4

Plots of canonical discriminant coefficients 1 and 2 (can 1 and can 2) based on the DGGE data of 16S rDNA directly isolated from soil. Plot A distinguishes between organic and conventional rhizosphere and bulk soils; plot B discriminates between management type and increasing and decreasing phases in oscillations of oligotrophic CFUs in the rhizosphere; plot C distinguishes between management type and peaks and valleys in oligotrophic CFU waves in the rhizosphere (experiment II).

Full size image

The separation between banding patterns at locations where CFUs increased versus locations where CFUs decreased along the wheat roots in conventional soil was greater than that between peaks and valleys in CFUs in the same soil, whereas the opposite is true for the organic soil (Figs. 4b, c). Taking into account the location of the samples with respect to their distance from the root tip, the data suggest that there was a succession of bacterial species within a wave, and that this succession was largely repeated from wave to wave.

Biodiversity: Species Richness and Shannon Index of Biodiversity

Both measures of bacterial biodiversity in soil, species richness S and Shannon index H, had significant harmonic distributions along the root (Fig. 5). The biodiversity measures oscillated in opposite phase to the bacterial CFUs. Significant negative cross-correlations were found between wavelike fluctuations in biodiversity and harmonics in oligotrophic and copiotrophic CFUs (Table 4), especially along the root in conventional soil. In the bulk soil, no significant cross-correlations were found.

Figure 5
figure 5

Biodiversity determined for 16S rDNA directly extracted from wheat rhizosphere soil from an organic farm plotted against distance from the root tip. Harmonic oscillations of the Species Richness, i.e., numbers of amplicons detected by DGGE (♦, dotted line), and the Shannon index based on DGGE amplicons and their relative abundance (□, continuous line) (Exp. II).

Full size image
Table 4 Cross-correlation coefficients (CC) between oligotrophic or copiotrophic CFUs in the rhizosphere or bulk soil from an organic and conventional farm and bacterial diversity based on the number of amplicons or the Shannon index for amplicon intensities on DGGE gels of 16S rDNA directly isolated from rhizosphere or bulk soil
Full size table

A rhizosphere effect was detected with one-sided paired t-tests for both biodiversity measurements S and H. While bacterial numbers were higher in rhizosphere than in corresponding bulk soil samples along the root, biodiversity was less in the rhizosphere than at corresponding locations in the bulk soil (Table 5). The biodiversity in the rhizosphere of organic and conventional soil did not differ significantly, but the bulk soil biodiversity was higher in the conventional than in the organic soil, supporting the idea that diversity is lower when the density of CFU is higher.

Table 5 Biodiversity as Species Richness S or Shannon index H, both based on amplicons on DGGE gels for 16S rDNA directly isolated from rhizosphere and bulk soil
Full size table

Discussion

We combined “classic” cultivation methods and molecular biological methods based on 16S rDNA PCR and DGGE to study the microbial communities along wheat roots in two different soils in detail. Significant harmonic fluctuations in the numbers of both copiotrophic and oligotrophic bacteria were found along the root. The two subpopulations were positively cross-correlated, and both started with an increased population at the root tip followed by a complete wave approximately every next 25 cm of root. Taking the difference in root growth rate into account, the period of these harmonics was similar to that in two experiments in California [20].

The wavelike distributions of bacterial CFUs along the root were matched by similar wavelike distributions of high-intensity amplicons in DGGE gels with DNA extracted directly from soil or from bacterial suspensions. Sequencing of individual bands would be needed to investigate if wavelike fluctuations are primarily due to growth and death cycles—as we assume—of relatively few fast-growing species that can be isolated on culture media. Other, less intense amplicons fluctuated in antiphase compared to the CFU oscillations, suggesting that these taxonomic units may follow the high-intensity amplicons in the succession of bacteria within a wave of bacterial growth and death. This scenario of a succession of bacterial species within a complete cycle rather than between cycles is supported by the similarity of the communities in the waxing phases of successive waves (and in the waning phases of the successive waves), whereas the waxing and waning phases within the same wave differed from each other. It is also supported by a wavelike distribution along roots of P. fluorescens mixed uniformly into soil (Van Bruggen et al., unpublished data). The periods of the waves in P. fluorescens density (CFUs and direct microscopic counts) coincided with those of total copiotrophic CFUs in previous experiments [24] and the current experiments.

About 80% of the amplicons detected by DGGE in our experiments after direct extraction of DNA from soil were found in both rhizosphere and bulk soils. The majority of the soil bacteria thus appear ubiquitous throughout the soil [13, 18]: it is their relative intensity that varies. The increase of bacterial numbers detected at the root tip can be explained by a combination of rapid proliferation of certain r-strategist species in response to root exudates [13, 22] and, less likely, passive or active migration of bacteria in the soil toward the root [1, 2, 22]. The following decrease in the population observed along the root can be explained by dying off of those pioneer bacteria. This decrease is then followed by a new round of bacterial growth in response to compounds released by dying bacteria supplemented with nutrients from soil and the root [35]. The described events in the rhizosphere correspond well with the theory of r and K selection. During the waxing phases and peaks, r-strategists (more uniform) could take advantage, whereas in the waning phases and valleys, K-strategists (more diverse) dominate. The DGGE data set of copiotrophic, faster-growing “r-strategists” had relatively few amplicons with high intensities. The much slower-growing oligotrophic bacteria or “K-strategists” were more diverse resulting in more DGGE bands. Especially the large number of DGGE amplicons from oligotrophic colonies detected in antiphase with the CFU harmonics supports the notion of a greater diversity of “K-strategists” in the valleys of wavelike fluctuations.

Both soils showed a positive rhizosphere effect on numbers of oligotrophic and copiotrophic bacteria along the root in comparison to the corresponding bulk soils, but the biodiversity along the root diminished in comparison to the surrounding bulk soil, particularly in the peaks of the wavelike fluctuations in the rhizosphere. Rhizosphere effects like we detected—a higher bacterial biomass [7, 15], a different community composition [13], or a diminished biodiversity [14, 15, 27]—have been observed before, but not in relation to distance from the root tip. The observed negative cross-correlation between bacterial numbers and biodiversity can be explained if a limited number of fast-growing species are responsible for the peaks. These few abundant species overshadow numerous other species that are close to or under the limit of detection, resulting in lower biodiversity. Steer and Harris [27] and Ogino et al. [21] describe a decline in biodiversity and change in population composition similar to the rhizosphere effect after fertilization of the soil. A disturbance in a soil resulting in the availability of extra nutrients, for example, by a real or artificial root, fertilizer, or organic amendment, obviously enhances a few fast-responding species, which cause a local loss in biodiversity.

Several limitations of DGGE are that it includes DNA from dead cells and assumes equal numbers of ribosomal copies per cell, equal recovery from soil, and equal amplification of all copies, which may not be the case. Still, it is one of the most powerful currently available tools to look at microbial communities [13, 18, 21, 32]. Comparison of relative frequencies of amplicons at certain relative positions in the DGGE gels does give an idea how the frequency of corresponding bacteria changes in the populations that are compared. Moreover, the trends found in amplicons of DNA directly extracted from soil were supported by similar trends in amplicons of DNA extracted from copiotrophic and oligotrophic colonies.

The differences recognized with discriminant analyses between management, location, increasing and decreasing population numbers, and peaks and valleys in population numbers were stronger in the conventional soil than in the organic soil. Thus, the rhizosphere effect was greater in the conventionally managed soil than in the organically managed soil with respect to CFUs, bacterial composition, and bacterial biodiversity. Rhizosphere soil with “root camouflage,” whose microbial communities are more similar to the microbial community in the surrounding soil, may be less attractive to pathogens [9]. In our organic soil, the differences between rhizosphere and bulk soil communities were smaller than in the conventional soil. Some of our other preliminary data show that organic soil may indeed be less conducive to soil-borne pathogens than conventional soil [31]. In the experiments described here, we did find less severe symptoms of the incidental Fusarium infection on wheat roots in the organic soil than in the conventional soil. Enhanced root disease suppression in organically managed soils was demonstrated in many other studies [28, 31].

Never before has the rhizosphere of single roots been so intensively studied as in our experiments. Previously, at the most, a few root sections were investigated [4, 5, 17, 18, 32]. Marschner et al. [17] found that the bacterial community composition changed with time and was specific for the root zones at root tip, subapical zone, and mature root zones at the sites of lateral root emergence. In our experiments, we demonstrated that microbial populations and communities are not directly related to lateral root emergence, but fluctuate in harmonic waves, although some species seem to be specific for certain regions. Following the changes from one bacterial community to the next along the root, bacterial successions with harmonic fluctuations were detected in both soils, the oscillations being clearer in the rhizosphere of the conventionally farmed soil. Interestingly, also plant pathogens show harmonic fluctuations along a root, particularly in a soil low in organic matter [30]. Our approach to studying microbial communities in the rhizosphere could lead to the identification of antagonists or competitors of plant pathogens that are active in the same phase as microbial communities along the root.