• Bacterial community;
  • 16S rRNA;
  • DGGE;
  • global change;
  • carbon dioxide;
  • FACE


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The response of total (DNA-based analysis) and active (RNA-based analysis) bacterial communities to a pCO2 increase under field conditions was assessed using two perennial grasses: the nitrophilic Lolium perenne and the oligonitrophilic Molinia coerulea. PCR- and reverse transcriptase-PCR denaturing gradient gel electrophoresis analysis of 16S rRNA genes generated contrasting profiles. The pCO2 increase influenced mainly the active and root-associated component of the bacterial community. Bacterial groups responsive to the pCO2 increase were identified by sequencing of corresponding denaturing gradient gel electrophoresis bands. About 50% of retrieved sequences were affiliated to Proteobacteria. Our data suggest that Actinobacteria in soil and Myxococcales (Deltaproteobacteria) in root are stimulated under elevated pCO2.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Since the beginning of the industrial revolution, the atmospheric CO2 concentration (pCO2) has been rapidly increasing, affecting the global climate and the functioning of oceanic and terrestrial ecosystems (Bazzaz & Sombroek, 1999; Fuhrer, 2003). Much research has focused on the consequences of elevated pCO2 on plant physiology and growth, as well as on vegetation structure. Elevated pCO2 enhances the net photosynthesis, the shoot and root biomass, and the litter input relative to ambient pCO2 condition (Sowerby et al., 2000; Zak et al., 2000; Ainsworth et al., 2003), particularly in C3 plants (Long et al., 2004).

Under current ambient atmospheric conditions, up to 50% of the assimilated carbon is translocated to the belowground (Kuzyakov & Domanski, 2000) providing carbon and energy sources easily available for soil biota. Under elevated pCO2, greater input (Darrah, 1996) and qualitative changes (Hodge et al., 1998) in carbon released into the rhizosphere are likely to impact the soil microflora (Jones et al., 1998). For instance, the effects of a pCO2 increase were described on arbuscular mycorrhizal fungi (Gamper et al., 2004), on relative frequency (Marilley et al., 1999) and on phenotypic structure of Pseudomonas (Roussel-Delif et al., 2005; Tarnawski et al., in press). Firstly, CO2-induced alterations in carbon supply could modify microbial processes that are directly dependant on carbon input, particularly decomposition and nutrient cycling (Hu et al., 1999). Secondly, elevated pCO2 could alter the structure of the microbial community due to qualitative changes in carbon supply under these conditions. In turn, the selection or counterselection of plant-deleterious (Chakraborty et al., 2000) or plant-beneficial microorganisms (Gamper et al., 2004; Tarnawski et al., in press) would have feedback effects on plant growth and physiology, because of a shift in microbial balance. In particular, this might enhance plant growth by increasing nutrient acquisition from previously unavailable pools (Hu et al., 1999).

In order to understand how soil–plant systems respond to elevated pCO2, the response of the microbial community has to be characterized and the populations involved in this response have to be identified. As most microbes are in an ‘inactive’ state in soils (Hu et al., 1999), whole community parameters (i.e. DNA- and fatty acid-based analyses) are probably less sensitive than those measuring some component of the active microflora. Recently developed molecular approaches such as stable isotope probing (Radajewski et al., 2000) or RNA-based analysis (Felske & Akkermans, 1998; [19]Koizumi et al., 2003) may be appropriate.

The development of molecular techniques in microbial ecology, including those based on the small subunit ribosomal RNA gene sequence as molecular marker, provides a significant advantage in studying microbial communities in terms of richness and structure, allowing monitoring of changes in microbial communities in a large number of samples (Muyzer et al., 1993; [20]Fromin et al., 2002). Molecular fingerprinting techniques such as denaturing gradient gel electrophoresis (DGGE) generate snapshots of the bacterial community, displayed as patterns related to the presence of dominant populations. Moreover 16S rRNA transcripts can be targeted. As the ribosome content of cells depends on their activity level, profiles obtained after reverse transcriptase (RT)-PCR on environmental 16S rRNA are therefore weighted according to the actual activity of the related populations (Wagner, 1994). By comparing DNA- and RNA-based profiles, it is then possible to highlight the dominant active members of the community (Felske et al., 1998). These fingerprinting analyses generate a large amount of data, which should benefit from the development of numerical ecology (Legendre & Legendre, 1998; Fromin et al., 2002). Ordination methods can be used to compare DGGE patterns with each other as multivariate responses to environmental variables (Ter Braak, 1986; Borcard et al., 1992).

The aims of the present study were to investigate whether elevated pCO2 influence the bacterial community and to highlight and identify the most affected populations and their potential metabolic role in these soils. The response of soil and root-associated microflora to high atmospheric pCO2 content was assessed by molecular fingerprinting of total and active bacterial communities. Two hemicryptophytic perennial grasses were used as model plants: Lolium perenne and Molinia coerulea. These plants have different trophic requirements (nitrophilic and oligonitrophilic Vazquez de Aldana & Berendse, 1997), allowing testing of the importance of the functional type regarding the plant response to an elevation of atmospheric pCO2 (Lüscher et al., 1998). A shift in the bacterial community structure under high pCO2 was revealed by DGGE profiling after direct PCR (total) and RT-PCR (active community) amplification of 16S rRNA genes from soil and root samples. Multivariate statistical analyses were used to highlight responsive bacterial groups.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study site and plant material

Lolium perenne and Molinia coerulea were grown under field conditions and current ambient (36 Pa, C for control) vs. elevated pCO2 (60 Pa, T for treated) in the Free Air CO2 Enrichment (FACE) facilities at Eschikon, Switzerland (Hebeisen et al., 1997). The atmosphere of treated plots was enriched with CO2 during the growing season and daytime only. Lolium perenne cv Bastion (L.) was grown as a monoculture on three control (LC) and three CO2-treated (LT) replicate plots from May 1993. The plants were grown on the autochthonous soil, a fertile Eutric Cambisol (FAO classification). The shoots were harvested four times a year. The LC and LT plots received 14 g m−2 year−1 N as NH4NO3, at the beginning of the season, and then after each cut, except the last (this amount was demonstrated to be limiting for plant growth during the FACE experiment; Daepp et al., 2000). Molinia coerulea plants (M) originated from a littoral meadow on the south shore of Lake Neuchâtel (Cudrefin, Switzerland). The local soil, a Gleysol, Typic Haplaquoll, contained about 4.7% clay, 9.5% silt and 85.8% sand, with a pH[H2O] value of 8.4 (Hamelin et al., 2002). Plants with undisturbed root systems were taken with their surrounding and underlying soil, and transferred to the FACE facilities in September 1999. About 0.7 m2 of littoral meadow with reconstructed soil profile below the root horizon (total depth: 35 cm) was installed in one control plot (MC) and one CO2-treated plot (MT). The plants were neither cut nor fertilized.


Sampling was performed at three sampling dates (21 June 2001, 7 May 2002 and 15 July 2002). Two subsamples were collected, one for DNA extraction and another for RNA extraction (RNase-free sampling conditions). For L. perenne, one to three control (C) and CO2-treated (T) plots were sampled at each date (Table 1). For M. coerulea, only technical replicates could be sampled as only one plot per pCO2 condition was available. Sampling of L. perenne was always performed just before a cut. For each sampling, three soil cores (about 3 cm diameter, 10–12 cm depth) including root systems, were taken and pooled for analysis. Two fractions were recovered: the non-adhering soil, obtained by shaking roots (S) and the root itself (R, for rhizoplane-endorhizosphere) after thorough washing of root-adhering soil. Soil and root samples were immediately placed in FastRNA™ matrix tubes for RNA and FastDNA™ matrix tubes for DNA (Bio101, QBiogene, Inc., Basel, Switzerland) and instantaneously frozen in liquid nitrogen.

Table 1.   DNA and RNA yields (μg g−1 soil or root fresh weight) obtained from soil and root samples of Lolium perenne and Molinia coerulea growing under ambient or elevated pCO2 content
Sampling datepCO2 treatmentType of sampleNucleic acid typeAverage yield±SD (no. replicates)
Lolium perenneMolinia coerulea
  1. Standard deviation (±SD) is indicated when replications were performed. The number of replicates is indicated in parentheses.

21 June 2001ControlSoilDNA2.56.2
RNA7.8±0.5 (3)8.8
RNA8.5±0.2 (3)7.4
7 May 2002ControlSoilDNA10.34.4
RNA8.0±0.8 (2)9.8±5.7 (2)
RNA29.210.5±0.7 (2)
RNA8.9±7.5 (2)7.9±2.4 (2)
RNA28.8±3.2 (2)9.8±1.5 (2)
15 July 2002ControlSoilDNA6.2±1.0 (3)5.8±1.2 (2)
RNA8.3±1.5 (3)7.2±0.2 (2)
RootDNA6.3±2.8 (3)7.6±1.0 (2)
RNA9.3±1.4 (3)8.9±0.2 (2)
TreatedSoilDNA3.3±0.8 (3)6.6±0.4 (2)
RNA6.9±3.0 (3)6.1±2.3 (2)
RootDNA7.7±1.4 (3)8.2±1.8 (2)
RNA5.6±1.2 (3)11.7±2.1 (2)

DNA extraction and purification

DNA extraction and purification were performed on about 0.5 g of fresh root or soil material. A bead-beating apparatus (FP120 FastPrep™ cell disruptor, Savant Instruments, Inc., Hotbrook, NY) was used in combination with the FastDNA Spin Kit for Soil (Bio101) according to Borneman et al. (1996), except that 500 μL of DNA lysate were purified using 500 μL of Binding Matrix (Bio101). The final DNA extracts were quantified using GeneQuant RNA/DNA calculator (Amersham Pharmacia Biotech, Cambridge, UK) and stored at −20°C before use.

RNA extraction and purification

From sampling until cDNA synthesis, all RNA handling was performed under RNase-free conditions. Aqueous solutions were treated with 0.1% diethyl pyrocarbonate (DEPC). Glassware was heated to 200°C overnight and plastic material soaked overnight in a 0.1 N NaOH/1 mM EDTA solution, before rinsing with RNase-free water. The working area and materials reserved for RNA handling were treated with RNase-AWAY solution (Molecular BioProducts Inc., San Diego, CA).

Total RNA was extracted and purified using a combination of FastRNA™ tubes with Green Caps (Bio101) and RNeasy® Plant Kit (Qiagen AG, Basel, Switzerland). The samples were put on ice between the extraction steps. In each FastRNA™ tube containing about 150–500 mg of frozen sample, 450 μL of RLT Buffer (Qiagen) were added. The mixture was shaken for 10 s at 6 m s−1 using the FastPrep™ cell disruptor. This step was repeated once after cooling tubes for 5 min on ice. Borneman & Triplett (1997) found this 20-s period of bead beating to be optimal for maximum cell lysis and minimum RNA shearing. The tubes were then centrifuged for 5 min at 13000 g and the supernatant was loaded on QIAshredder Spin Columns (Qiagen) and then processed as recommended by the manufacturer. DNA was removed using DNase (Qiagen) according to the manufacturer's protocol. The final RNA extracts were eluted in 100 μL 10-mM Tris pH 7.0, quantified using GeneQuant (Amersham Pharmacia), and stored at −80°C before use. PCR amplification and DGGE were performed directly on each RNA extract to detect DNA contamination. In a few cases the presence of DNA was detected in the RNA extract, in which case the corresponding band positions were then discarded for further analysis.

Reverse transcription of total RNA

Reverse transcription reactions were performed using ImProm-II™ Reverse Transcription System (Promega Corp., Madison, WI) with random hexamer primers in a thermo-cycler model PTC-200 (MJ Research Inc., Watertown, MA). A total of 3.5 μL of RNA extract (55–70 ng depending on the sample) was mixed with 1 μL of primers (10 mM), and 0.5 μL of RNasin® Ribonuclease Inhibitor. This mixture was incubated at 70°C for 5 min for an optimal contact between RNA and primers, and chilled on ice until the reverse transcription mix was added. This mix was then combined with (final concentrations) 1 × ImProm-II™ Reaction Buffer, 0.05 U μL−1 RNasin, 6 mM MgCl2, 0.5 mM each dNTP, 5% (v/v) ImProm-II™ Reverse Transcriptase and DEPC-treated nanopure water in a final volume of 20 μL. The reaction consisted of annealing at 25°C for 5 min, extension at 42°C for 1 h and inactivation of reverse transcriptase at 70°C for 15 min. The resulting cDNA was used immediately for PCR or stored at −20°C. Positive and negative control reactions were performed as recommended by the manufacturer.

PCR amplification

PCR amplification of the V3 region of 16S rRNA gene was performed in two steps. The whole 16S rRNA gene was first amplified using the forward GM3f (5′-AGAGTTTGATCMTGGC-3′) and the reverse GM4r (5′-TACCTTGTTACGACTT-3′) Bacteria primers (Muyzer & Ramsing, 1995). The PCR reaction mix contained (final concentrations) 1X Thermophilic DNA Buffer, 3 mM MgCl2, 0.25 mM dNTPs, 0.25 μM of each primer (MWG Biotech AG, Ebersberg, Germany), and 0.05 U μL−1 of Taq DNA polymerase (Promega). A total of 2 μL of DNA or cDNA extract were added as template for the PCR. The final reaction volume was adjusted to 20 μL. The reaction mixtures were subjected to 26 amplification cycles in a thermo-cycler. The first heat denaturation step was performed at 94°C for 4 min 30 s. Cycles consisted of heat denaturation at 94°C for 1 min, primer annealing at 56°C for 30 s with a touchdown of 1°C every 2 cycles for a total of ten cycles, and extension at 74°C for 1 min. The mixture was maintained at 74°C for 10 min for the final extension. The forward 338f (5′-ACTCCTACGGGAGGCAGCAG-3′) and reverse 520r (5′-ATTACCGCGGCTGCTGG-3′) universal primers (Ovreas et al., 1997) were used for nested amplification of the V3 region of the 16S rRNA gene to increase the amplification yield and to obtain a fragment size suitable for DGGE analysis. A 40-bp GC-clamp (Muyzer et al., 1993) was added on the forward primer for DGGE analysis. The nested-PCR mix was prepared as for the first PCR except 5 μL of PCR-amplified 16S rRNA were added as template and the final volume was adjusted to 50 μL. The first heat denaturation step was performed at 94°C for 5 min. The reaction mixtures were then subjected to 31 amplification cycles. Cycles consisted of heat denaturation at 94°C for 1 min, primer annealing at 65°C for 30 s with a touchdown of 1°C per cycle for ten cycles, and extension at 74°C for 1 min. The mixture was maintained at 74°C for 10 min for the final extension. The PCR products were checked for size and yield on 1% agarose gels in comparison to the Low DNA Mass Ladder (Invitrogen).

DGGE analysis

Denaturing gradient gel electrophoresis analysis of 16S rRNA genes and cDNA amplicons was performed using the D-code electrophoresis system (Bio-Rad Inc., Hercules, CA). About 600–800 ng of PCR products were loaded directly on a 8% (w/v) polyacrylamide gel (acrylamide-bisacrylamide 37.5 : 1) with a linear gradient from 30% to 60% denaturants (100% correspond to 40% formamide plus 7 M urea). The strains used to build the reference DGGE pattern were ordered as follows after migration (Fig. 1): Pseudomonas fluorescens ATCC 27663, Bacillus subtilis ATCC 14893, Flavobacterium capsulatum DSM 30196, Rhizobium meliloti DSM 1981, Arthrobacter globiformis DSM 20124 and Thermus filiformis NCIMB 12588. The gels were run at 60°C and 150 V for 5 h in 1 × TAE buffer. They were stained with 0.01% SYBR Green (Molecular Probes, Leiden, the Netherlands) in 1 × TAE at 4°C in the dark for 20 min, then UV photographed with the Multi-Analyst package (Bio-Rad).


Figure 1.  Example of 16S rRNA gene-based denaturing gradient gel electrophoresis profiles obtained from soil of Lolium perenne plots cultivated under ambient and elevated pCO2. The first six patterns are rDNA-based profiles and the last six are rRNA-based profiles from three replicate plots for each pCO2 condition for the third sampling date. Ref stands for the reference pattern.

Download figure to PowerPoint

Data handling and statistical analysis of DGGE patterns

The gel images were normalized regarding the band positions of the reference patterns for each gel, then the sample patterns were compared using GelCompar software (Applied Maths, Kortrijk, Belgium). The fingerprints obtained were codified in terms of migration length and relative intensity (pi) of each band (i) within the profile (total intensity of the profile Σpi=1). Two datasets were obtained: (1) DNA data matrix, containing relative intensity of bands for DNA-based profiles and (2) RNA data matrix containing relative intensity of bands for RNA-based profiles. For both plants, co-amplification of amyloplastic DNA and RNA was often observed for root samples (the position of plant 16S rRNA fragment after DGGE analysis was checked using plant axenic DNA extract as PCR template). Corresponding bands were discarded for further analyses.

The percentage of similarity between DNA- and RNA-based profiles obtained from both pCO2 conditions (ambient/treated) was calculated using the Steinhaus coefficient. It was computed from the relative intensity of DGGE bands (set between 0 to 1), by the sum of the lower frequency observed for each pair of common bands (Legendre & Legendre, 1998). It gives more weight to intense common bands than to weak common bands. All calculations were done using the R 1.9.0 package (R Development Core Team, 2004).

The data matrices were then transformed for ordination analyses. Bands appearing only in one DNA- or RNA-based profile were discarded from dataset due to their low representation within profiles to avoid a rare species effect. Band intensities from the remaining data (64 bands for RNA matrix and 55 bands for DNA matrix) were normalized using the log-transformation: pi′=ln(pi+1).

Canonical Correspondence Analysis (CCA) was applied on both transformed data matrices to represent the influence of explanatory variables on the one hand, and to evidence the most ‘impacted’ bands on the other hand. Data were initially submitted to variation partitioning analysis (Borcard et al., 1992) using a series of partial CCA to display variability of the patterns constrained by factors of interest. Three sets of explanatory variables were employed: (1) sampling date (environmental conditions at sampling time) and replication; (2) plant–soil system (M. coerulea/L. perenne); and (3) root presence combined with pCO2 treatment (ambient/elevated). The significance of the results was tested with the Monte Carlo permutation test. The whole process was based on computations made with R 1.9.1 (R Development Core Team, 2004). In a second step, in order to focus on the effects of pCO2 and roots on DGGE profile variability, part of the variability explained by sampling date and replication was subtracted by using these as co-variables for partial CCA. Data obtained from L. perenne and from M. coerulea were separated to distinguish key DGGE bands for each plant–soil system.

Selection of indicative bands

Various bands were selected according to the number of profiles in which they were detected (frequent bands were preferred to rare ones) and their position on CCA plots (bands distant from the graphic axes origin were preferred to others). Among these bands, we considered that those which were close to the centroïds of the explanatory variables corresponded to indicative populations that were most influenced by these factors. Five indicative bands (corresponding to position 225 for L. perenne samples and 151, 224, 246 and 249 for M. coerulea samples; Table 2) were excised from different DGGE patterns to check the corresponding sequence homogeneity.

Table 2.   Characterisation of bands obtained from 20 and 11 DNA- and 39 and 27 RNA-based denaturing gradient gel electrophoresis (DGGE) profiles for Lolium perenne and Myxococcales coerulea respectively. The bands were selected for their variation in intensity within profiles according to the origin of the sample, to the pCO2 content and to the nucleic acid type. The total length of the DGGE gel corresponds to 500 pixels. The representativity of a given band was expressed as the percentage of DNA- or RNA-based profiles in which the band was detected. Changes in band intensities induced by pCO2 elevation are expressed as increase (+%) or decrease (−%) in band intensity. For each band, several clones were checked, the number of clone affiliated to the same restriction type is mentioned in parentheses. One clone of each type has been sequenced, accession numbers of the sequences are indicated after the affiliation group. An asterisk was added next to the DGGE band name when the sequences was retrieved from RNA profile.
DGGE band descriptionCorresponding populations
NamePosition (pixels)OriginRepresentativity among profiles (%)Effect of pCO2 elevation on band relative intensities (%)Closest phylogenetic group (affiliation based on Blast)
DNARNARootSoilRestriction type ARestriction type B
Lolium perenne
L1*101Soil013−100+100Gammaprot./Pseudomonadales (4/8); AJ851090Actinobacteria (4/8); AJ851091
L2*112Root518−100Uunaffiliated (3/3); AJ851092 
L3128Root2028−11+100Bacteroidetes (7/7); AJ851093 
L4133Root4041+82−15Betaprot./Burkholderiales (1/4); AJ851094Gammaprot./Enterobacteriales (3/4); AJ851095
L5*164Root2015+81+444Betaprot./Burkholderiales (3/8); AJ851096Deltaprot./Myxococcales (5/8); AJ851097
L6*168Root2036+35−3Gammaprot./Enterobacteriales (1/8); AJ851098Deltaprot./Myxococcales (7/8); AJ851099
L7179Soil2018−13+12Bacteroidetes (2/7); AJ851100Bacteroidetes (5/7); AJ851101
L8*190Root2526−86+48Alphaprot./Rhizobiales (2/5); AJ851102Deltaprot./Myxococcales (3/5); AJ851103
L9225Soil5062+136+1021unaffiliated (4/4); AJ851104 
L10*225Root5062+136+1021unaffiliated (8/8); AJ851105 
L11*225Soil5062+136+1021Actinobacteria (6/6); AJ851106 
L12*225Root5062+136+1021Alphaprot. (8/8); AJ851107 
L13*231Root1526+1264−1Deltaprot./Myxococcales (5/5); AJ851108 
L14*237Soil2533+197+100Actinobacteria (1/4); AJ851109unaffiliated (3/4); AJ851110
L15*290Root2536−54+4Deltaprot./Desulfurellales (3/6); AJ851111Deltaprot./Myxococcales (3/6); AJ851112
L16293Soil3013+74−58unaffiliated (1/8); AJ851113Actinobacteria (7/8); AJ851114
L17298Soil58U−56unaffiliated (1/3); AJ851115Alphaprot. (2/3); AJ851116
Myxococcales coerulea
M1*98Soil011U+96Bacteroidetes (3/8); AJ851117unaffiliated (5/8); AJ851118
M2140Root4611+82+317unaffiliated (4/4); AJ851119 
M3145Root3644+61+32unaffiliated (1/8); AJ851120Actinobacteria (7/8); AJ851121
M4151Root7337+105−66Gammaprot./Enterobacteriales (1/5); AJ851122Firmicute/Bacillales (4/5); AJ851123
M5*151Root7337+105−66Gammaprot./Enterobacteriales (2/3); AJ851125Deltaprot./Myxococcales (1/3); AJ851124
M6*172Soil5544+121+297unaffiliated (3/8); AJ851126Betaprot./Burkholderiales (5/8); AJ851127
M7*207Soil926−100−38Betaprot./Burkholderiales (3/5); AJ851128Betaprot./Burkholderiales (2/5); AJ851129
M8*224Soil5541+635−70Betaprot./Burkholderiales (3/8); AJ851130Actinobacteria (5/8); AJ851131
M9*224Root5541+635−70Bacteroidetes (6/8); AJ851132unaffiliated (2/8); AJ851133
M10*233Soil2741+4−53Firmicutes (1/3); AJ851134Deltaprot./Myxococcales (2/3) ; AJ851135
M11237Root97+100+100Deltaprot./Myxococcales (1/6); AJ851136Alphaprot. (5/6); AJ851137
M12246Root1815−35−100Deltaprot./Myxococcales (3/7); AJ851138Deltaprot./Myxococcales (4/7); AJ851139
M13*246Root1815−35−100unaffiliated (1/8); AJ851140Deltaprot./Myxococcales (7/8); AJ851141
M14249Root2737+267+1740unaffiliated (1/3); AJ851142Alphaprot./Rhodospirillales (2/3); AJ851143
M15*249Soil2737+267+1740unaffiliated (1/4); AJ851144Deltaprot./Myxococcales (3/4); AJ851145
M16*265Root2726−100−97Chloroflexi (1/4); AJ851146Alphaprot./Rhodospirillales (3/4); AJ851147
M17*303Soil1833+30+549Alphaprot./ Rhodospirillales (2/4); AJ851148unaffiliated (2/4); AJ851149
M18*332Soil1819+100+241Chloroflexi (1/5); AJ851150Actinobacteria (4/5); AJ85115

Identification of the responsive populations

Selected DGGE bands were excised. Corresponding DNA was eluted overnight at 4°C with 15 μL nuclease-free water and the electrophoresis – excision – elution cycle was repeated to ensure that the recovered DNA corresponded to a given DGGE band. Then the recovered DNA was amplified with a 338f – 520r primer set using 15 amplification cycles (heat denaturation at 94°C for 1 min, primer annealing at 65°C for 30 s and extension at 74°C for 1 min). PCR products were purified and cloned into pGEM-T and introduced into Escherichia coli XL1 by electroporation. The sequence homogeneity within a single DGGE band was checked on obtained clones with HaeIII restriction patterns of T7-SP6 PCR amplicons. For each band, one clone was sequenced for each restriction profile type. The phylogenetic affiliation of corresponding organisms was achieved by BLAST analysis (Altschul et al., 1997). Sequences were deposited at EMBL under the accession numbers AJ851090 to AJ851151.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Extraction yields and extracts purity

The average DNA and RNA extraction yields were respectively 6.0 (SD±2.5) and 8.1 (SD±2.4) μg g−1 fresh weight for soil and 7.1 (SD±2.7) and 12.2 (SD±8) μg g−1 fresh weight for root samples (Table 1). The quality and quantity of the extracts were always sufficient for PCR and RT-PCR reactions, irrespective of the plant-soil system and the pCO2 treatment.

DGGE pattern description

DNA and RNA extracts obtained from the same sample, after PCR and RT-PCR amplification, generated contrasting DGGE profiles (Fig. 1). Band intensity was more uniform within DNA-based patterns, whereas RNA profiles displayed a clear dominance of a few bands. Soil patterns displayed smeared areas, probably representing clusters of low intensity bands, whereas root profiles often presented sharp bands.

For both plant–soil systems, the similarities between DNA- and RNA-based profiles for a given sample (Fig. 2) were generally below 50%. The DNA and RNA profile similarities were higher for ambient than for elevated pCO2 plots (eight out of nine for root fraction, and five out of nine for soil fraction), indicating an influence of pCO2 on community profiles.


Figure 2.  DNA-/RNA-based profile similarity compared for ambient and elevated pCO2 conditions (calculated with Steinhaus coefficient). (•) Soil samples; (○) Root samples. Sampling dates are indicated with numbers (1 for June 2001, 2 for May 2001 and 3 for July 2002) and replicates with small letters (a; b; c). Samples scattered on the left upper part of the plot indicate that their DNA and RNA patterns are more similar for treated plots compared to control plots. Samples scattered on the right lower part of the plot indicate that DNA and RNA patterns are more similar for ambient plots than for treated plots.

Download figure to PowerPoint

Sources of the DGGE profile variability

The variation partitioning analysis (Fig. 3) allowed the relative influence of: (1) replicate plots and sampling date; (2) plant–soil system; and (3) elevated pCO2 and root proximity on total (DNA-based) and active (RNA-based) bacterial community profiles to be shown. This analysis first revealed the high percentage of unexplained variance (71–79%, Fig. 3). The remaining 21% and 29% of the variance were significantly explained by each of the identified descriptors. The descriptors displaying the highest part of explained profile variability were the sampling date and plots. Globally, 15.1% of DNA- and 11.9% of RNA-based pattern variation were attributed to these descriptors, which integrate numerous environmental conditions (e.g. temperature, soil water content) varying in time and space. The plant–soil system explained 8.8% (DNA-) and 6.3% (RNA-based patterns) of the variability. The pCO2 treatment alone did not influence microbial fingerprints; however, the combined influence of root vicinity and pCO2 treatment significantly accounted for DNA- (6.9%) and for RNA-based (5.4%) profile variation, suggesting that elevated pCO2 impacts bacterial communities through the roots.


Figure 3.  Variation partitioning for data obtained from DNA- and RNA-based denaturing gradient gel electrophoresis patterns. The variation partitioning was tested for significance with 999 permutations using the Monte Carlo test for each set of descriptors: (1) plant–soil system; (2) pCO2 treatment and root proximity; and (3) sampling date and replicate plots (*P<0.05; **P<0.01; ***P<0.001).

Download figure to PowerPoint

Canonical Correspondence Analysis (Fig. 4) allowed us to ordinate response variables in a single ordination plane, constrained only by root and pCO2 treatment after removing part of the pattern variability explained by date and plot replication, and by the soil-plant system. Bands influenced by elevated pCO2 condition were generally associated with the root fraction. This is in agreement with data from similarity coefficients (Fig. 2). Root influence (as shown by strong correlation of the corresponding centroïd to CCA axis 1) was more important than pCO2 influence (more correlated to CCA axis 2) (Fig. 4). Changes induced by pCO2 increase were observed for both DNA- (P=0.021) and RNA-based (P=0.026) community profiles for M. coerulea. A similar trend was observed for metabolically active communities associated with L. perenne (P=0.087).


Figure 4.  Biplots of partial Canonical Correspondence Analysis of data obtained from DNA- and RNA-based denaturing gradient gel electrophoresis (DGGE) patterns for Lolium perenne and Molinia coerulea datasets. Constrained axes 1 and 2 were used for graphic representation. DGGE bands selected for sequencing are indicated with a letter (M or L).

Download figure to PowerPoint

Choice of characteristic bands and affiliation of corresponding populations

A total of 17 and 18 characteristic bands were selected for L. perenne and M. coerulea, respectively, based on the DGGE bands representativity (relative intensity and frequency of occurrence among all DNA- and RNA-based profiles, Table 2) and on CCA graphic representation (Fig. 4). On average, the intensity of all selected bands within a given profile represented 26% of the total intensity (data not shown). Twenty-seven (L. perenne) and 35 (M. coerulea) sequences were retrieved from these selected bands. Clones obtained from the same excised band corresponded to one or two restriction types. In most cases, sequences from a single DGGE band were affiliated to the same group or to related groups (e.g. bands L7, M7 and M12, Table 2). However sometimes sequences from a single band were affiliated to different phylogenetic groups (e.g. bands L1 and M8). Some sequences obtained from co-migrating bands for different samples were affiliated to related groups (bands M12 and M13), whereas some others were affiliated differently (bands L11 and L12). Generally, sequences obtained from both plant–soil systems displayed similar affiliations. Among the 27 sequences retrieved for L. perenne, 52% were affiliated to Proteobacteria (Alpha-, Beta-, Gamma-, Delta-), of which 36% were related to Myxococcales (Deltaproteobacteria). Other sequences were affiliated to Actinobacteria (15%), Bacteroidetes (11%), and others were unaffiliated. Similar proportions were observed for the 35 sequences obtained from M. coerulea. The average intensity of selected bands doubled under elevated compared to ambient pCO2 (Table 2). The average intensity of bands corresponding to Actinobacteria increased under elevated pCO2 (root: +77%, soil: +291% for L. perenne; root: +265%, soil: +68% for M. coerulea). Myxococcales-related bands displayed higher intensities under elevated pCO2, particularly in root (+248%) compared to soil (+79%) for L. perenne, and in soil (+254%) compared to root (+67%) for M. coerulea.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Comparison between total and active 16S rRNA gene community profiles

Whereas RNA-based profiles highlight active bacterial populations at the time of sampling, DNA-based profiles display the most abundant bacterial populations, independently of their current activity. DNA- and RNA-based profiles for a given sample generally shared less than 50% of pattern similarity (Fig. 2), as previously observed (Muyzer & Smalla, 1998; Kowalchuk et al., 1999; Duineveld et al., 2001).

Field-induced variation

The FACE system currently provides the most realistic way to estimate how plants will respond to elevated pCO2 in their native environment, avoiding the modification of natural air flow induced by other CO2 enrichment systems (Long et al., 2004). However, field experiments imply large variations in environmental conditions within time and space (e.g. soil heterogeneity, root distribution, temperature, precipitation). A high unexplained variation was observed, as in most ecological studies (Borcard et al., 1992; Ritz et al., 2004). The largest part of the profiles variability was explained by time of sampling and plots (Fig. 3). Such a high percentage of fingerprint variability between sampling dates is consistent with the short-term response of microbial populations to environmental changes.

Plant–soil system influences

Bacterial communities associated with the ecologically contrasting perennial grasses L. perenne and M. coerulea (Vazquez de Aldana & Berendse, 1997) were different: 6.3% (RNA) and 8.8% (DNA) of the variability of community profiles could be significantly explained by the plant–soil system (Fig. 3). The soil characteristics are different for L. perenne and M. coerulea swards: they allow different bacterial communities to settle (Latour et al., 1996).

We expected the response of soil microbial communities to elevated pCO2 to be dependent on the plant type, through rhizodeposition (Wardle et al., 2004), whereas the same bacterial groups were found to be influenced by pCO2 in the rhizosphere of both plants. This suggests that bacterial communities associated with the two plants (both being perennial hemicryptophytic grasses) responded similarly to the pCO2 increase, despite the functional differences between the two host plants (nitrophilic for L. perenne vs. oligonitrophilic for M. coerulea).

Influence of elevated pCO2

The direct influence of an atmospheric pCO2 increase on soil bacterial communities is probably negligible because of the naturally high pCO2 concentrations in the soil atmosphere (200–3500 Pa) (Gobat et al., 2004). However, bacterial communities were significantly modified by the combined effect of pCO2 treatment and root vicinity (Fig. 3). The effect of pCO2 enrichment on soil bacterial communities is likely mediated by the plant through quantitative and qualitative changes in rhizodeposition (Paterson et al., 1996; Hodge et al., 1998).

Changes induced by high pCO2 were more pronounced on active than on total bacterial communities at root vicinity. Similarities between DNA- and RNA-based profiles were higher under ambient than under elevated pCO2, except for L. perenne soil samples (Fig. 2). This could reflect a more stable state of the bacterial community under ambient pCO2, whereas low similarity between total and active communities under high pCO2 would reflect a shifting state of the bacterial community due to fluctuations in the metabolic activity of specific populations (Montealegre et al., 2000), because of root-mediated modification in trophic fluxes due to higher pCO2 (Hodge et al., 1998).

Total community analysis frequently failed to indicate pCO2-induced changes (Griffiths et al., 1998; Jones et al., 1998; Insam et al., 1999; Ebersberger et al., 2004) and soil microbial biomass often seemed unaffected under elevated pCO2 (Zak et al., 2000). By contrast, pCO2-induced changes could be observed when targeting specific functional or taxonomic groups (Jones et al., 1998; Ronn et al., 2003; Roussel-Delif et al., 2005; Tarnawski et al., in press). The activity of some functional groups such as fungal cellulose decomposers (Jones et al., 1998) and simple carbohydrate consumers (Hodge et al., 1998) was shown to be enhanced under elevated pCO2. Belowground responses of microbial communities to global change could generate feedback effects on aboveground biota, such as plant physiology and diversity (Van der Heijden et al., 1998; Jackson et al., 2002). The identification of responsive groups is necessary to understand putative feedbacks on the functioning of soil-plant systems.

Key populations

The bacterial groups responding to root vicinity and pCO2 increase were highlighted by sequencing selected DGGE bands from total or metabolically active communities. A high proportion (11%) of the selected bands corresponded to sequences affiliated to Actinobacteria. They were generally retrieved from the active fraction of the soil bacterial community, regardless of the plant-soil system studied. Actinobacteria are known to be soil engineers using soil organic matter as their main carbon source (Ensign, 1992). Their dependence on plant exudates is therefore weak and Actinobacteria may be less affected by root-mediated perturbations. The importance of Actinobacteria in terms of abundance and activity in soils was demonstrated (Gremion et al., 2003), including under elevated pCO2 (Billings & Ziegler, 2005).

Myxococcales were identified as the most responsive group to either pCO2 increase or root influence (19% of retrieved sequences). The relative intensity of corresponding bands increased under elevated pCO2. Myxococcales are known to be cellulolytic organisms (Reichenbach & Dworkin, 1992). Cellulolytic fungi were shown to be favoured under elevated pCO2 conditions (Jones et al., 1998), as shown for Myxococcales in the present study. Root growth and exudation are increased under elevated pCO2 (Zak et al., 2000), leading to a greater availability of cellulose (Robinson et al., 1997), and a stimulation of cellulolytic organisms.

Currently, a major challenge to understand better the role of microbial communities in plant-soil system functioning is to link taxonomic diversity and functions. This requires a prior identification of organisms and their corresponding functions. Some bacterial groups (Actinobacteria, Myxococcales) were identified as key organisms in the response of soil-plant systems to elevated pCO2. This study will be useful further to identify bacterial functions involved in the response of ecosystems to global changes.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This research was supported by the Swiss National Science Foundation (grants number 3100-055899.98 and 31-68208.02). We are also grateful to the Swiss National Centre of Competence in Research (NCCR) ‘Plant Survival’. We thank to Jacob Zopfi for the English corrections.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Ainsworth EA, Davey PA, Hymus GJ, Osborne CP, Rogers A, Blum H, Nösberger J & Long SP (2003) Is stimulation of leaf photosynthesis by elevated carbon dioxide concentration maintained in the long term? A test with Lolium perenne grown for 10 years at two nitrogen fertilization levels under free air CO2 enrichment (FACE). Plant Cell Environ 26: 705714.
  • Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W & Lipman DJ (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acid Res 25: 33893402.
  • Bazzaz F & Sombroek WG (1999) Changements du Climat et Production Agricole ONU (Alimentation et Agriculture). Polytechnica, Paris.
  • Billings SA & Ziegler SE (2005) Linking microbial activity and soil organic matter transformations in forest soils under elevated CO2. Global Change Biol 11: 203212.
  • Borcard D, Legendre P & Drapeau P (1992) Partialling out the spatial component of ecological variation. Ecology 73: 10451055.
  • Borneman J & Triplett EW (1997) Rapid and direct method for extraction of RNA from soil. Soil Biol Biochem 29: 16211624.
  • Borneman J, Skroch PW, Osullivan KM, Palus JA, Rumjanek NG, Jansen JL, Nienhuis J & Triplett EW (1996) Molecular microbial diversity of an agricultural soil in Wisconsin. Appl Environ Microbiol 62: 19351943.
  • Chakraborty S, Tiedemann AV & Teng PS (2000) Climate change: potential impact on plant diseases. Environ Pollut 108: 317326.
  • Daepp M, Suter D, Almeida JPF, Isopp H, Hartwig UA, Frehner M, Blum H, Nösberger J & Lüscher A (2000) Yield response of Lolium perenne swards to free air CO2 enrichment increased over six years in a high N input system on fertile soil. Global Change Biol 6: 805816.
  • Darrah PR (1996) Rhizodeposition under ambient and elevated CO2 levels. Plant Soil 187: 265275.
  • Duineveld BM, Kowalchuk GA, Keijzer A, van Elsas JD & van Veen JA (2001) Analysis of bacterial communities in the rhizosphere of Chrysanthemum via denaturing gradient gel electrophoresis of PCR-amplified 16S rRNA as well as DNA fragments coding for 16S rRNA. Appl Environ Microbiol 67: 172178.
  • Ebersberger D, Wermbter N, Niklaus PA & Kandeler E (2004) Effects of long term CO2 enrichment on microbial community structure in calcareous grassland. Plant Soil 264: 313323.
  • Ensign JC (1992) Introduction to the Actinomycetes. The Prokaryotes (BalowsA, TrüperHG, DworkinM, HarderW & SchleiferKH, eds), pp. 811815. Springer, New York.
  • Felske A & Akkermans ADL (1998) Prominent occurrence of ribosomes from an uncultured bacterium of the Verrucomicrobiales cluster in grassland soils. Lett Appl Microbiol 26: 219223.
  • Felske A, Akkermans ADL & De Vos WM (1998) Quantification of 16S rRNAs in complex bacterial communities by multiple competitive reverse transcription PCR in temperature gradient gel electrophoresis fingerprints. Appl Environ Microbiol 64: 45814587.
  • Fromin N, Hamelin J, Tarnawski S, Roesti D, Jourdain-Miserez K, Forestier N, Teyssier-Cuvelle S, Gillet F, Aragno M & Rossi P (2002) Statistical analysis of denaturing gel electrophoresis (DGE) fingerprinting patterns. Environ Microbiol 4: 634643.
  • Fuhrer J (2003) Agroecosystem responses to combinations of elevated CO2, ozone, and global climate change. Agric Ecosys Environ 97: 120.
  • Gamper H, Peter M, Jansa J, Lüscher A, Hartwig UA & Leuchtmann A (2004) Arbuscular mycorrhizal fungi benefit from 7 years of free air CO2 enrichment in well-fertilized grass and legume monocultures. Global Change Biol 10: 189199.
  • Gobat JM, Aragno M & Mathey W (2004) The Living Soil: Fundamentals of Soil Science and Soil Biology. Science Publishers, Inc., Enfield, NH, USA.
  • Gremion F, Chatzinotas A & Harms H (2003) Comparative 16S rDNA and 16S rRNA sequence analysis indicates that Actinobacteria might be a dominant part of the metabolically active bacteria in heavy metal-contaminated bulk and rhizosphere soil. Environ Microbiol 5: 896907.
  • Griffiths BS, Ritz K, Ebblewhite N, Paterson E & Killham K (1998) Ryegrass rhizosphere microbial community structure under elevated carbon dioxide concentrations, with observations on wheat rhizosphere. Soil Biol Biochem 30: 315321.
  • Hamelin J, Fromin N, Tarnawski S, Teyssier-Cuvelle S & Aragno M (2002) nifH gene diversity in the bacterial community associated with the rhizosphere of Molinia coerulea, an oligonitrophilic perennial grass. Environ Microbiol 4: 477481.
  • Hebeisen T, Lüscher A & Nösberger J (1997) Effects of elevated atmospheric CO2 and nitrogen fertilisation on yield of Trifolium repens and Lolium perenne. Acta Oecologica-Int J Ecol 18: 277284.
  • van der Heijden MGA, Klironomos JN, Ursic M, Moutoglis P, Streitwolf-Engel R, Boller T, Wiemken A & Sanders IR (1998) Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396: 6972.
  • Hodge A, Paterson E, Grayston SJ, Campbell CD, Ord BG & Killham K (1998) Characterisation and microbial utilisation of exudate material from the rhizosphere of Lolium perenne grown under CO2 enrichment. Soil Biol Biochem 30: 10331043.
  • Hu S, Firestone MK & Chapin S (1999) Soil microbial feedbacks to atmospheric CO2 enrichment. Trends in Ecol Evol 14: 433437.
  • Insam H, Baath E, Berreck M, Frostegard A, Gerzabek MH, Kraft A, Schinner F, Schweiger P & Tschuggnall G (1999) Responses of the soil microbiota to elevated CO2 in an artificial tropical ecosystem. J Microbiol Meth 36: 4554.
  • Jackson RB, Banner JL, Jobbagy EG, Pockman WT & Wall D (2002) Ecosystem carbon loss with woody plant invasion of grasslands. Nature 418: 623626.
  • Jones TH, Thompson LJ, Lawton JH, et al. (1998) Impacts of rising atmospheric carbon dioxide on model terrestrial ecosystems. Science 280: 441443.
  • Koizumi Y, Kojima H & Fukui M (2003) Characterization of depth-related microbial community structure in lake sediment by denaturing gradient gel electrophoresis of amplified 16S rDNA and reversely transcribed 16S rRNA fragments. FEMS Microb Ecol 46: 147157.
  • Kowalchuk GA, Naoumenko ZS, Derikx PJL, Felske A, Stephen JR & Arkhipchenko IA (1999) Molecular analysis of ammonia-oxidizing bacteria of the beta subdivision of the class Proteobacteria in compost and composted materials. Appl Environ Microbiol 65: 396403.
  • Kuzyakov Y & Domanski G (2000) Carbon input by plants into the soil. J Plant Nutr Soil Sci 163: 421431.
  • Latour X, Corberand TS, Laguerre G, Allard F & Lemanceau P (1996) The composition of fluorescent pseudomonad populations associated with roots is influenced by plant and soil type. Appl Environ Microbiol 62: 24492456.
  • Legendre P & Legendre L (1998) Numerical Ecology. 2nd English Edition. Elsevier, Amsterdam.
  • Long SP, Ainsworth EA, Rogers A & Ort DR (2004) Rising atmospheric carbon dioxide: plants FACE the future. Annu Rev Plant Biol 55: 591628.
  • Lüscher A, Hendrey GR & Nösberger J (1998) Long-term responsiveness to free air CO2 enrichment of functional types, species and genotypes of plants from fertile permanent grassland. Oecologia 113: 3745.
  • Marilley L, Hartwig UA & Aragno M (1999) Influence of an elevated atmospheric CO2 content on soil and rhizosphere bacterial communities beneath Lolium perenne and Trifolium repens under field conditions. Microb Ecol 38: 3949.
  • Montealegre CM, van Kessel C, Blumenthal JM, Hur HG, Hartwig UA & Sadowsky MJ (2000) Elevated atmospheric CO2 alters microbial population structure in a pasture ecosystem. Global Change Biol 6: 475482.
  • Muyzer G & Ramsing NB (1995) Molecular methods to study the organisation of microbial communities. Water Sci Technol 32: 19.
  • Muyzer G & Smalla K (1998) Application of denaturing gradient gel electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial ecology. Antonie Van Leeuwenhoek 73: 127141.
  • Muyzer G, de Waal EC & Uitterlinden AG (1993) Profiling of complex microbial populations by DGGE analysis of PCR-amplified genes coding for 16S ribosomal RNA. Appl Environ Microbiol 59: 695700.
  • Ovreas L, Forney L, Daae FL & Torsvik V (1997) Distribution of bacterioplankton in meromictic lake Saelenvannet, as determined by denaturing gradient gel electrophoresis of PCR-amplified gene fragments coding for 16S rRNA. Appl Environ Microbiol 63: 33673373.
  • Paterson E, Rattray EAS & Killham K (1996) Effect of elevated atmospheric CO2 concentration on C-partitioning and rhizosphere C-flow for three plant species. Soil Biol Biochem 28: 195201.
  • R Development Core Team (2004) A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Vienna, Austria.
  • Radajewski S, Ineson P, Parekh NR & Murrell JC (2000) Stable-isotope probing as a tool in microbial ecology. Nature 403: 646649.
  • Reichenbach H & Dworkin M (1992) The Myxobacteria. The Prokaryotes (BalowsA, TrüperHG, DworkinM, HarderW & SchleiferKH, eds), pp. 34163487. Springer, New York.
  • Ritz K, McNicol W, Nunan N, et al. (2004) Spatial structure in soil chemical and microbiological properties in an upland grassland. FEMS Microb Ecol 49: 191205.
  • Robinson CH, Michelsen A, Lee JA, Whitehead SJ, Callaghan TV, Press MC & Jonasson S (1997) Elevated atmospheric CO2 affects decomposition of Festucavivipara (L.) Sm. litter and roots in experiments simulating environmental change in two contrasting arctic ecosystems. Global Change Biol 3: 3749.
  • Ronn R, Ekelund F & Christensen S (2003) Effects of elevated atmospheric CO2 on protozoan abundance in soil planted with wheat and on decomposition of wheat roots. Plant Soil 251: 1321.
  • Roussel-Delif L, Tarnawski S, Hamelin J, Philippot L, Aragno M & Fromin N (2005) Frequency and diversity of nitrate reductase genes among nitrate-dissimilating Pseudomonas in the rhizosphere of perennial grasses grown in field conditions. Microb Ecol 49: 6372.
  • Sowerby A, Blum H, Gray TRG & Ball AS (2000) The decomposition of Lolium perenne in soils exposed to elevated CO2: comparisons of mass loss of litter with soil respiration and soil microbial biomass. Soil Biol Biochem 32: 13591366.
  • Tarnawski S, Hamelin J, Jossi M, Aragno M & Fromin N (in press) Phenotypic structure of Pseudomonas populations is altered under elevated pCO2 in the rhizosphere of perennial grasses. Soil Biol Biochem.
  • Ter Braak CJF (1986) Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 11671179.
  • Vazquez de Aldana BR & Berendse F (1997) Nitrogen-use efficiency in six perennial grasses from contrasting habitats. Funct Ecol 11: 619626.
  • Wagner R (1994) The regulation of ribosomal RNA synthesis and bacterial cell growth. Arch Microbiol 161: 100106.
  • Wardle DA, Bardgett RD, Klironomos JN, Setälä H, van der Putten WH & Wall DH (2004) Ecological linkages between aboveground and belowground biota. Science 304: 16291633.
  • Zak DR, Pregitzer KS, King JS & Holmes WE (2000) Elevated atmospheric CO2, fine roots and the response of soil microorganisms: a review and hypothesis. New Phytol 147: 201222.