OUP user menu

Effects of selected root exudate components on soil bacterial communities

Shengjing Shi, Alan E. Richardson, Maureen O'Callaghan, Kristen M. DeAngelis, Eirian E. Jones, Alison Stewart, Mary K. Firestone, Leo M. Condron
DOI: http://dx.doi.org/10.1111/j.1574-6941.2011.01150.x 600-610 First published online: 1 September 2011

Abstract

Low-molecular-weight organic compounds in root exudates play a key role in plant–microorganism interactions by influencing the structure and function of soil microbial communities. Model exudate solutions, based on organic acids (OAs) (quinic, lactic, maleic acids) and sugars (glucose, sucrose, fructose), previously identified in the rhizosphere of Pinus radiata, were applied to soil microcosms. Root exudate compound solutions stimulated soil dehydrogenase activity and the addition of OAs increased soil pH. The structure of active bacterial communities, based on reverse-transcribed 16S rRNA gene PCR, was assessed by denaturing gradient gel electrophoresis and PhyloChip microarrays. Bacterial taxon richness was greater in all treatments than that in control soil, with a wide range of taxa (88–1043) responding positively to exudate solutions and fewer (<24) responding negatively. OAs caused significantly greater increases than sugars in the detectable richness of the soil bacterial community and larger shifts of dominant taxa. The greater response of bacteria to OAs may be due to the higher amounts of added carbon, solubilization of soil organic matter or shifts in soil pH. Our results indicate that OAs play a significant role in shaping soil bacterial communities and this may therefore have a significant impact on plant growth.

Keywords
  • rRNA-DGGE
  • organic acids
  • Pinus radiata
  • rRNA-PhyloChip
  • microarray
  • root exudates

Introduction

Plants release significant amounts of photosynthetically derived carbon (C) into the rhizosphere through root exudation, making root exudates a key factor in increasing microbial abundance and activity in the rhizosphere compared with bulk soil (Lynch, 1990; Kapoor & Mukerji, 2006). Low-molecular-weight C compounds present in root exudates, including sugars, organic acids (OAs) and amino acids, are readily assimilated by microorganisms and are proposed to play a primary role in regulating microbial community dynamics in the rhizosphere (Bais et al., 2006; Weisskopf et al., 2008). However, direct studies of the influence of root exudates on soil microbial communities are technically challenging because of the complexity of the rhizosphere environment and difficulty in sampling (Biedrzycki & Bais, 2009). As a consequence, direct causal links between low-molecular-weight C compounds in exudates on plant rhizosphere microbial communities have not been demonstrated conclusively (Weisskopf et al., 2008; Micallef et al., 2009).

The role of root exudate compounds in the structure and activity of soil bacterial communities has been investigated by the application of root exudate compound (REC) solutions to soil in microcosm studies (Griffiths et al., 1999; Baudoin et al., 2003; Paterson et al., 2007; Henry et al., 2008). Although such experiments cannot reproduce completely the complex rhizosphere environment, they offer an approach for investigating the roles of exudates in shaping the rhizosphere microbial community and provide information on potential plant mechanisms for the control of soil microbial diversity and function (Baudoin et al., 2003). Studies using artificial root exudates have demonstrated the importance of exudates in rhizosphere processes (Nielsen & van Elsas, 2001; Henry et al., 2008).

Until now, most of the studies investigating REC solutions on soil microbial communities have focused on the exudate profiles of agricultural plants, such as maize (Benizri et al., 2002; Baudoin et al., 2003; Henry et al., 2008) or perennial ryegrass (Paterson et al., 2007), where sugars are the dominant fraction, with lesser emphasis on OAs and other compounds. Conversely, low-molecular-weight OAs are a more significant component in root exudates of forest trees, where they are typically present at two to three times the C content of sugars (Smith, 1976; Grayston et al., 1996). Therefore, low-molecular-weight OAs in forest tree root exudates may play important roles in shaping rhizosphere microbial communities. As soil microbial communities may be influenced by both the composition and the concentration of exudate compounds (Griffiths et al., 1999; Baudoin et al., 2003), the impact of REC solutions based on those of forest trees may differ from those of maize or ryegrass.

Previously, we characterized the OA and sugar composition of root exudate solutions collected in situ from Pinus radiata D. Don (radiata pine) grown in a rhizotron system (Shi et al., 2011). This study aims to determine the effects of artificial REC solutions, based on those collected in situ, on the community composition and diversity of metabolically active soil bacteria in a soil microcosm experiment.

Materials and methods

Soil microcosms

Horizon A of a Templeton silt loam soil (Immature Pallic Soil; NZ Soil Bureau, 1968) under pasture was collected from Lincoln University Canterbury, New Zealand (3% clay, 50% silt, 47% sand), and selected chemical properties of the soil are listed in Supporting Information, Table S1. The soil was moistened to 80% of field capacity and sieved (1.4 mm mesh size) before being packed into Perspex containers (4.5 cm diameter). These microcosms held the equivalent of 50 g dry soil and were preincubated at 18 °C in the dark for 14 days (maintained at 80% of field capacity) to allow the microbial communities to stabilize before the establishment of REC treatments. Because nitrogen (N) is often the most limiting nutrient for microbial growth when C is added to soil (Merckx et al., 1987; van Veen et al., 1989), 0.45 mg N g−1 dry weight soil was added to each microcosm as 1 mL of a filter-sterilized NH4NO3 solution (64 g L−1) and mixed well on the last day of preincubation.

Six REC solutions (S, sugars; SQLM, sugars, quinic, lactic and maleic acids; SQ, sugars and quinic acid; SL, sugars and lactic acid; SM, sugars and maleic acids; and QLM, quinic, lactic and maleic acids) were prepared as outlined in Table 1. The composition and concentration of sugars and OAs in these solutions were based on a previous study (Shi et al., 2011) for root exudate solutions collected in situ from soil-grown P. radiata. The ratio of C content of sugars and OAs was 1 : 2, which is consistent with profiles of tree root exudates reported by others (Smith, 1976; Grayston et al., 1996). The pH of all REC solutions was adjusted to 5.5, and REC solutions and deionized water as a control (Con) were filter-sterilized and frozen in 1.5 mL aliquots for daily application.

View this table:
1

Composition and rates of supply of sugars and organic acids in REC solutions

Treatment codeSugars (mg g−1 dry soil day−1)Organic acids (mg g−1 dry soil day−1)Daily C input (mg C g−1 dry soil day−1)
Con000
SGlucose – 0.08300.1
Sucrose – 0.079
Fructose – 0.083
SQLMGlucose – 0.083Quinic acid – 0.1520.3
Sucrose – 0.079Lactic acid – 0.166
Fructose – 0.083Maleic acid – 0.161
SQGlucose – 0.083Quinic acid – 0.4570.3
Sucrose – 0.079
Fructose – 0.083
SLGlucose – 0.083Lactic acid – 0.4980.3
Sucrose – 0.079
Fructose – 0.083
SMGlucose – 0.083Maleic acid – 0.4830.3
Sucrose – 0.079
Fructose – 0.083
QLM0Quinic acid – 0.1520.2
Lactic acid – 0.166
Maleic acid – 0.161

REC solutions were applied daily to soil microcosms at a C concentration up to 0.3 mg C g−1 soil dry weight day−1 (Table 1) for 15 days, to avoid a large, single C addition. The daily C input was within the range reported by others (Griffiths et al., 1999; Baudoin et al., 2003; Henry et al., 2008). The solutions applied were uniformly mixed with the soil after each addition to avoid concentration gradients. The working solution consisted of 1 mL of each thawed REC solution mixed with sufficient sterile water to maintain the soil moisture at 80% of field capacity throughout the experiment. Four replicate microcosms prepared for each treatment were incubated in a random arrangement at 18 °C in the dark for 15 days.

Soil pH and dehydrogenase activity analysis

Soil pH was measured for each microcosm on completion of the experiment. Dehydrogenase activity, as an indication of the overall microbial activity, was assessed using triphenyltetrazolium as an artificial electron acceptor as described by Alef (1995).

RNA extraction and cDNA preparation

Microbial RNA was extracted from 0.5 g of soil samples after vortexing in Lysing Matrix B tubes (Q-Biogene, CA) in a hexadecyl-trimethyl-ammonium bromide buffer using a phenol–chloroform purification protocol (Clough et al., 2009) modified from Griffiths. (2000). Nucleic acid samples were treated with DNase (TURBO DNA-free Kit, Applied Biosystems, CA) before reverse transcriptase (RT)-PCR and the removal of DNA was validated by control PCR. cDNA was synthesized from RNA samples using SuperScript III reverse transcriptase (Invitrogen, CA) according to the manufacturer's instructions using the reverse primer 1492R (Brodie et al., 2007). A negative control using DNase/RNase-free distilled water (Invitrogen) instead of an RNA sample was included in each RT-PCR run.

rRNA-denaturing gradient gel electrophoresis (DGGE) bacterial community analysis

Bacterial 16S rRNA genes were amplified from cDNA samples (1 : 10 dilution) using the primer set 341F-GC and 534R (Muyzer et al., 1993). PCR was carried out in 25 μL with 1 μL cDNA template or RT-PCR-negative control, 1 × NH4 buffer (Bioline, Alexandria NSW, Australia), 0.2 mM dNTPs (Promega Corporation, WI), 0.25 μM each primer, 3.75 mM MgCl2 and 1 U Biotaq DNA polymerase (Bioline). PCR amplifications were performed in an iCycler (BioRad, CA) with an initial denaturing step at 95 °C for 4 min, followed by 30 cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 45 s, before a final extension step at 72 °C for 7 min. PCR products were loaded on an 8% polyacrylamide gel (acrylamide/bisacrylamide, 37.5 : 1) with a 35–68% denaturing gradient of urea and formamide (Dcode System, BioRad). Electrophoresis was performed at 60 °C at 200 V for 10 min and then 80 V for 18 h. Acid silver staining was used to visualize the DGGE bands and the gels were scanned, and bands were selected and aligned in TotalLab TL120 (Nonlinear Dynamics, Newcastle upon Tyne, UK).

16S rRNA gene PhyloChip microbial community analysis

For practical reasons, 18 samples were selected for PhyloChip analysis: Con, S, SQLM, SQ, SL and SM (see Table 1) with three randomly chosen replicates of each treatment. Treatment QLM was not included in the PhyloChip analysis as communities from QLM and SQLM treatments appeared to be highly similar in the DGGE analysis. Bacterial 16S rRNA genes were amplified from cDNA samples (1 : 10 dilution) using the primer set 27F.1 and 1492R (Wilson et al., 1990). PCR was carried out in 100 μL containing 4 μL cDNA or RT-PCR-negative control, 1 × NH4 buffer (Bioline), 0.2 mM dNTPs (Promega Corporation), 0.4 μM each primer, 2 mM MgCl2, 0.4 mg mL−1 bovine serum albumin (Promega Corporation) and 4 U Biotaq DNA polymerase (Bioline). PCR amplification was performed in a thermal cycler (BioRad) with an initial denaturing step at 95 °C for 3 min, followed by 25 cycles at 95 °C for 30 s, 53 °C for 30 s and 72 °C for 1 min, before a final extension step at 72 °C for 7 min. PCR products were cleaned (MinElute PCR purification kit; Qiagen Science, CA) and 304 ng was subsequently used for each microarray. Hybridization, fragmentation and biotin labelling of the PCR product were performed as described by Brodie. (2006).

PhyloChips were scanned using a GeneArray Scanner (Affymetrix) and the initial data acquisition and intensity determination were performed using standard Affymetrix software (genechip microarray analysis suite, version 5.1). Background correction and data normalization between chips using internal standards were conducted as described by Brodie. (2006). Probe pairs scored as positive were those that met two conditions as outlined previously (Brodie et al., 2006; DeAngelis et al., 2008). A taxon was considered present when >90% of its assigned probe pairs were positive [probe fraction (pf)>0.9]. A hybridization intensity score (HybScore) was calculated in arbitrary units for each probe set as described previously (Brodie et al., 2006; DeAngelis et al., 2008). A taxon in each treatment was considered present if it occurred at least twice in three replicates. The HybScore values of taxa that were not present in treatments (did not fit the criteria with at least two replicates with pf>0.9) were set to 0, because these taxa are scored as absent in treatments (Yergeau et al., 2009).

Statistical analysis

Statistical analysis of soil pH and dehydrogenase activity across the treatments was conducted by anova using genstat 11 (VSN International Ltd, Rothamsted, UK). For multiple comparisons, treatment means were separated using Fisher's protected LSD at a P=0.05 level.

DGGE and PhyloChip data were analysed using the primer 5 multivariate software package (PRIMER-E Ltd, Plymouth, UK). Resemblance matrices for community profiles were constructed using DGGE band intensity or HybScore values as indicators of taxon abundance by calculating similarities between each pair of samples using the Bray–Curtis coefficient (Clarke & Warwick, 2001), as done previously (Rees et al., 2004; Wakelin et al., 2008; Micallef et al., 2009). The resemblance matrix was calculated based on square root-transformed DGGE or HybScore data to reduce the influence of dominant bands or species and take into consideration rarer species (Clarke & Warwick, 2001; Sanguin et al., 2006). Nonmetric multidimensional scaling (MDS) ordination was used to interpret multivariate distances between samples and treatments. MDS ordination plots were generated from the lowest stress solution following 100 random restarts. Nonmetric analysis of similarities (anosim), a nonparametric procedure using permutation-based testing, was used to test for a statistical difference between treatments (Clarke & Warwick, 2001).

To determine which soil microbial populations were most responsive to different REC solutions, pair-wise comparisons between various treatments and control or treatment S were assembled into ‘responsive communities’. The criteria for the assignment of a taxon to the responsive community were based on being detected in one treatment only or when a taxon's HybScore value differed significantly (P<0.05) from either the Con or the S treatments. For this analysis, all the HybScore values were log10(1+x) transformed and analysed by a paired t-test using the statistical software r 2.8.1 (R Development Core Team, 2007). Differences in the number of responsive taxa at each taxonomic level across REC treatments were determined using Pearson's χ2 distribution tests for independence, with the null hypothesis that the frequency of distribution was the same across all treatments.

Results

Soil pH and dehydrogenase activity

At the completion of incubation, the pH of soils amended with REC solutions containing OAs (SQLM, SQ, SL, SM and QLM) was 1–3 U greater than that of control soil and differed significantly (P<0.05) across treatments (Table 2). The pH in treatment S (sugars only) did not differ significantly (P>0.05) from that of control soils. Dehydrogenase activity was significantly greater (P<0.05) in soils amended with all REC solutions than in the control, with the exception of treatment SM, where the activity was significantly lower (P<0.05) (Table 2). Microbial activity in microcosms receiving different REC solutions was significantly different across treatments, indicating an effect of REC composition (Table 2).

View this table:
2

Soil pH and dehydrogenase activity in microcosms amended daily with either deionized water (Con) or REC solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid) for 15 days.

TreatmentSoil pHDehydrogenase activity
Con4.68 ± 0.01 a1.67 ± 0.10 b
S4.66 ± 0.01 a2.53 ± 0.07 c
SQLM6.82 ± 0.08 d4.38 ± 0.15 e
SQ6.57 ± 0.01 c3.99 ± 0.11 d
SL7.61 ± 0.03 f5.60 ± 0.03 f
SM5.40 ± 0.01 b1.23 ± 0.06 a
QLM7.48 ± 0.05 e3.87 ± 0.12 d
P-value<0.001<0.001
  • Values are presented as the mean ± 1 SE (n=4). Different letters indicate significant differences (P<0.05) between treatments.

rRNA-DGGE analysis of soil bacterial communities

DGGE profiles indicate clear treatment-specific effects on bacterial community structure (Fig. 1). MDS ordination analysis of DGGE profiles for treatments Con, S, SQLM and QLM showed that the distance (based on similarity) between bacterial communities in the control and S soils was less than those between control soil and the other two treatments (Fig. 2a). This indicates larger shifts in soil bacterial communities amended with the REC solutions containing a mixture of OAs, either in the presence or in the absence of sugars. Differences across those four treatments were, however, statistically significant (anosim, R=1, P<0.05). More similar community profiles in treatments SQLM and QLM were observed in the DGGE profiles (Fig. 1) and the MDS plot (short distance between the two treatments) (Fig. 2a), indicating that sugars had a smaller effect on communities than OAs. The MDS ordination of the DGGE profiles of treatments S, SQ, SL and SM also differed significantly among the treatments (anosimR=1, P<0.05) (Fig. 2b).

1

rRNA-DGGE profiles of bacteria in soils amended daily with either deionized water (Con) or different REC solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid) for 15 days. St, bacterial 16S rRNA gene-DGGE marker (Pectobacterium carotovorum, Variovorax paradoxus and Arthrobacter sp.) (n=4).

2

Nonmetric MDS ordination plots of rRNA-DGGE bacterial communities in soils amended with either deionized water (Con) or S, SQLM and QLM solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid) (n=4). Ordination plots contain samples from each DGGE gel shown in Fig. 1. (a) Con, S, SQLM, QLM; (b) S, SQ, SL, SM.

16S rRNA gene PhyloChip bacterial community analysis

Of the possible 8432 bacterial taxa resolvable at the species level on the rRNA-PhyloChip, 1595 taxa (falling within 42 phyla) were detected in at least one sample across all treatments. For further community analysis, taxa represented in at least two of three replicates in any treatment were examined. This eliminated 407 rare taxa, resulting in 1188 taxa from 38 phyla. Forty-six bacterial taxa (belonging to eight phyla) were detected in control soil communities, while 2.5-fold more taxa (from 12 phyla) were detected in treatment S (Fig. 3). The addition of mixtures of OAs and sugars (SQLM, SQ, SL and SM) increased the number of bacterial taxa detected to between 458 and 1072, representing an approximate 10–22-fold and a four to ninefold increase over the control and treatment S, respectively (Fig. 3). The number of bacterial phyla detected in OA treatments (SQLM, SQ, SL and SM) ranged from 29 to 38. Actinobacteria, Proteobacteria and Firmicutes dominated within each treatment (Fig. 3). However, taxa within these three phyla, in addition to Bacteroidetes, were the most abundant (810–3170 taxa) on the PhyloChip.

3

Richness of bacterial communities measured as the number of taxa detected by PhyloChip per phylum in soils amended daily with either deionized water (Con) or different REC solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid). Taxa from different phyla are represented by different colours. The numbers above each bar indicate the number of phyla detected in each treatment.

MDS ordination plots indicated the separation of soil bacterial communities in different treatments (Fig. 4). The stress values of MDS plots were 0, indicating a near-perfect representation of the multivariate distance between communities in two-dimensional space (Clarke & Warwick, 2001). As with the DGGE results, the distance between bacterial communities in control and S soils was smaller than that between control soil and other treatments (Fig. 4a). Thus, the addition of OAs to the sugar treatment resulted in much greater shifts in community structure than with sugars alone. The distances between the control samples and those containing OAs were also considerably greater than those between different OAs (Fig. 4a). To visualize the relationships between these treatments better, a second (subset) MDS was performed (Fig. 4b), showing the differentiation of the SQLM, SQ, SL and SM treatments. There was a strong and highly significant overall effect of REC solutions on communities (anosimR=1; P=0.001) and pair-wise comparisons revealed significant differences between all pairs of treatments (anosimR=1).

4

Nonmetric MDS ordination plots of PhyloChip-assessed bacterial communities in soils amended with either deionized water (Con) or different REC solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid) (n=3). Ordination plots contain samples from (a) all six treatments or (b) only the four treatments SQLM, SQ, SL and SM.

Bacterial taxa that responded most significantly to REC solutions are summarized in Table 3 (the complete list is presented in Table S2). The bacterial community responsive to sugar addition (compared with Con) comprised 101 taxa (88 positive and 13 negative) from 12 phyla. A significantly wider range of bacterial taxa were affected by REC solutions containing OAs (SQLM, SQ, SL and SM), whereby 435–1067 taxa were responsive, which is 4–10-fold more than for the sugars alone (P<0.001; Table 3). A range of 38–241 Firmicutes taxa, mainly belong to Bacilli and Clostridia classes, responded positively to REC solutions containing mixtures of sugars and OAs, while no Firmicute taxon responded positively to the sugar treatment alone and one Firmicute taxon (unidentified) responded negatively to the presence of sugars (Table 3). As with the effect of sugars in treatment S, most responsive taxa responded positively to REC solutions, whereas only a minority of taxa (<2%) responded negatively (Table 3).

View this table:
3

Numbers of bacterial taxa in the main phylogenetic groups found in communities responsive to REC solutions (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid)

Phylum/class/orderCompared with ConCompared with STotal taxa detected
SSQLMSQSLSMP-valueSQLMSQSLSMP-value
Acidobacteria3 (1)27 (1)291132 (1)<0.00125 (1)2610 (1)31 (2)<0.00139
Actinobacteria69202 (4)198 (4)172 (4)212 (8)<0.001138 (3)133 (3)108147 (29)<0.001247
*Actinobacteria69202 (4)197 (4)172 (4)212 (7)<0.001138 (3)132 (3)108147 (28)<0.001245
#Actinomycetales63180174156186 (3)<0.001121 (1)114 (2)98127 (24)<0.001211
Bacteroidetes115232832<0.00114221731<0.00137
*Bacteroidetes0787100.002787100.62614
*Sphingobacteria0713716<0.0017137160.00217
Chlorobi022030.04722030.0893
Chloroflexi314 (3)18 (3)13 (3)23 (4)<0.00112 (1)15 (1)10 (1)20 (6)<0.00128
Cyanobacteria1813635 (1)<0.001712534 (1)<0.00136
Firmicutes0 (1)120 (1)124 (1)38 (1)241 (1)<0.00112012338241<0.001250
*Bacilli058598106<0.00158598106<0.001107
*Clostridia0494923112<0.001494923112<0.001117
Gemmatimonadetes054240.05354240.4357
Nitrospira112120.80701210.3923
Planctomycetes058580.00458580.33611
Proteobacteria5 (10)193 (8)305 (3)115 (4)349 (8)<0.001197 (1)309 (1)112 (1)346 (3)<0.001418
*Alphaproteobacteria3 (2)99 (5)126 (1)53 (2)149 (5)<0.0019712451147 (1)<0.001164
#Azospirillales012140.04812140.1124
#Bradyrhizobiales034361639<0.00134361639<0.00140
#Caulobacterales07778<0.00177780.7768
#Sphingomonadales1 (1)18 (1)2715 (1)25 (1)<0.001182715250.00130
#Rhizobiales01520031<0.0011518131<0.00132
*Betapreoteobacteria020511850<0.00120511850<0.00162
#Burkholderiales010341033<0.00110341033<0.00142
#Methylophilales011110.2871111NA1
#Nitrosomonadeles065460.00565460.5537
*Gammaproteobacteria012 (2)58 (1)14 (1)54 (2)<0.00112 (1)58 (1)14 (1)54 (1)<0.00183
#Pseudomonadale005120.00705120.027
*Deltaproteobacteria2 (1)38 (1)51 (1)16 (1)65 (1)<0.00137501564 (1)<0.00175
*Epsilonproteobacteria0 (7)22181330<0.00129251330<0.00132
#Campylobacterales0 (7)22181330<0.00129251330<0.00132
Spirochaetes072421<0.00172421<0.00121
Verrucomicrobia215131120<0.00113119180.02120
Total dynamic taxa88 (13)658 (18)793 (12)423 (12)1043 (24)<0.001587 (8)963 (44)718 (5)345 (4)<0.001
  • The influence of different REC solutions was determined by comparing individual treatments with the control (Con) treatment, and the impacts of individual OAs and a mixture of OAs were determined by comparing treatments SQ, SL, SM and SQLM with treatment S. Values indicate the number of taxa that responded positively to the presence of REC solutions, while those in brackets indicate the number of taxa that respond negatively to the presence of REC solutions or compounds. The total taxa detected indicates the total number taxa detected in each phylogenetic group across all six treatments. Taxa are shown mainly to the phylum level, but, where appropriate, some of the phyla are divided into a class (indicated by *) or an order (indicated by #) level. P-values for Pearson's χ2 distribution tests for independence are presented for treatment comparisons within each taxonomic level.

The influence of OAs on bacterial taxa was determined more specifically by comparing the bacterial communities in treatments SQ, SL, SM and SQLM with treatment S (Table 3). Between 349 (maleic acid) and 1007 (quinic acid) taxa were affected, representing 29–81% of taxa detected. In contrast, the addition of sugars only (treatment S) affected 101 taxa (P<0.001; Table 3). Across all treatments, only small numbers of taxa (<5%) responded negatively to OAs [i.e. either were not detected in OA treatments or had a significantly lower (P<0.05) HybScore than those in treatment S] and there was no apparent synergistic effect of the mixture of OAs over each of the OAs when added alone (Fig. 3; Table 3).

The presence of OAs in REC solutions also had a larger influence than sugars on the community at the phylum level as indicated in Fig. 3. For example, only 15 proteobacterial taxa were significantly affected by the sugars added to soil, while 113–349 additional proteobacterial taxa responded to OAs (P<0.001; Table 3). Within this phylum, taxa in Azospirillales, Bradyrhizobiales, Caulobacterales, Rhizobiales and Sphingomonadales from Alphaproteobacteria, Burkholderiales and Nitrosomonadeles from Betaproteobacteria responded positively to OAs (especially quinate and maleate). Although only seven Pseudomonadales taxa in the Gammaproteobacteria were detected across the treatments, they also responded to the presence of OAs in soils, but not sugars, with the exception of treatment SQLM (Table 3). Although greatly influenced by the presence of sugars in soil, >100 additional Actinobacteria taxa were affected by OAs (either individual or as a mixture) treatments (Table 3). As with sugars, OAs mainly impacted the taxa from the order Actinomycetales, while some other taxa belonging to functionally important groups, such as Bacteroidetes and Sphingobacteria from Bacteroidetes, were also influenced by OAs, but not sugars. Taxa responsive to OAs also included some that are common in soil, but not well studied, such as Acidobacteria, Planctomycetes and Verrucomicrobia (Table 3).

A few bacterial taxa only responded to the presence of one particular OA. For example, taxa in phyla Caldithrix, Chlorobi, LD1PA group, OD1, Synergistes, WS3 and WS5 responded positively to quinate and maleate, but not lactate. One taxon in OP8 (AF419671) and one taxon in Termite group 1 (AB089050) responded only to maleate and not to other OAs or a mixture of three OAs (Table S2). However, a large number of taxa (over 350) responded similarly to all four REC solutions SQLM, SQ, SL and SM.

Discussion

Distinct changes in bacterial communities in soil amended with artificial REC solutions were revealed by both DGGE and PhyloChip analyses. The composition of these REC solutions was based on that of exudates collected in situ from roots of radiata pine grown in a rhizotron (Shi et al., 2011). These results indicate the potential importance of exudate compounds in influencing the structure of soil bacterial communities, in general agreement with other studies (Baudoin et al., 2003; Landi et al., 2006; Paterson et al., 2007; Henry et al., 2008). Of particular novelty in the present study is the measurement of changes in the metabolically active component of the bacterial community, based on the extraction and amplification of rRNA (Pennanen et al., 2004; DeAngelis et al., 2010; Hirsch et al., 2010). Furthermore, there was a significantly greater response of the community to OAs applied to soil, either individually or as a mixture, than to the addition of sugars alone. Importantly, these differences were consistently detected by both DGGE and PhyloChip analyses, despite clear differences in the relative taxonomic resolution of the two techniques. The microarray data demonstrated that the addition to soil of REC solutions containing OAs in particular significantly stimulated the growth of a wide range of microorganisms with potentially important ecological roles in soil and the rhizosphere.

Both DGGE and PhyloChip analyses demonstrated that the effects of OAs on soil bacterial communities were considerably greater than for sugars alone. This was reflected in the total detectable taxon richness of communities, the number of bacterial taxa in the ‘responsive communities’ and the total soil microbial activity as determined by dehydrogenase assays. Richness and diversity indices do, however, refer only to those detected as DGGE bands or on PhyloChips and not to the total richness of soil communities. All OA treatments stimulated the microbial community, although significant differences were observed within some taxonomic groups (Table 3). This implies that OAs are not only an important quantitative component of root exudates (Smith, 1976) but also play a potentially important qualitative-biological role in shaping the structure of microbial communities in the rhizosphere. This is likely to be of major significance for tree species such as radiata pine, where organic anions are a predominant constituent of root exudates (Grayston et al., 1996; Shi et al., 2011).

Greater impacts of specific OAs than sugars on soil bacterial communities have been reported by others (Falchini et al., 2003). For example, Eilers. (2010) showed a more pronounced response of microbial communities to the addition of citric acid than glucose or glycine in three different soil types. Landi. (2006) also found that glucose induced fewer changes in the bacterial community than oxalic acid. These authors suggested that this might be due to the nonspecific use of glucose by a large proportion of soil microorganisms, whereas oxalic acid was decomposed by specialized microorganisms. However, the current study showed a common positive response to OAs by a diverse range of microorganisms, rather than highly specific responses by a few distinct bacterial groups. This might be explained partly by the addition of higher amounts of total C content in the OA treatments (Table 1). For example, Griffiths. (1999) reported greater changes in microbial communities with increasing C input. The application of OAs to soil could also contribute to mobilization of soil organic matter (Jones, 1998; Kuzyakov, 2002), benefitting microorganisms through enhanced availability of a highly heterogenic source of soil C. Soil pH changed significantly in all treatments containing OAs, compared with control soils (including the sugar-only treatment), potentially contributing to further shifts in bacterial communities observed for OAs. Changes in soil pH occur through the microbial utilization of OAs in soil (rather than by any direct effect per se) (Gramss et al., 2004; Evangelou et al., 2008). Variation in soil pH has been identified as a major driver of change in the microbial community structure (Fierer & Jackson, 2006; Wakelin et al., 2008; Lauber et al., 2009; Bissett et al., 2010; Osborne et al., 2011). It is also well established that the pH of rhizosphere soil is commonly significantly different from that of bulk soil (Marschner & Römheld, 1983). Therefore, a shift in soil pH due to the metabolism of OAs by distinct groups of microorganisms may have had a wider impact and a more important role in shifting or maintaining the total diversity of microbial communities in the soil. Within the diverse range of taxa stimulated by OAs (Table 3), some may thus be primary utilizers of the acids as a C source, whereas others may have emerged through either succession or benefited from soil organic matter partially mobilized (or mineralized) by enzymes produced by specialist microorganisms (Haichar et al., 2008). For example, various responsive taxa detected in our study (e.g. Actinomycetales, Caulobacterales, Rhizobiales) can use exudate compounds as energy sources and produce soil organic matter mineralizing enzymes (Horwath, 2007; Haichar et al., 2008).

Interestingly, a small percentage of bacterial taxa in our study also responded negatively to the REC solutions. This negative impact may result from the direct inhibition of microorganisms due to the presence of particular compounds. It is more likely, though, that these taxa were outcompeted by the rapid growth of other microorganisms better able to tolerate changes in soil environmental factors such as soil pH or more able to access soil resources (e.g. Paterson et al., 2007). Indeed, the use of soil microcosms, with the daily addition of substrates and routine mixing and incubation under optimal temperature and moisture conditions, may have favoured select groups of fast-growing heterotrophic microorganisms. The rapid growth and proliferation of such microorganisms in the microcosms would enhance their detection in DGGE profiles and the strength of hybridization on microarrays. Various studies have indicated a relationship between growth rate and rRNA abundance in cells that allows dominant and metabolically active microorganisms to be more highly represented in community surveys (Binder & Liu, 1998; Hirsch et al., 2010). However, exceptions of the linear link between rRNA abundance and growth rate have also been reported (Pernthaler et al., 2001; Schmid et al., 2001) and caution is needed when interpreting data from rRNA-based assays. Irrespective of this, REC solutions, especially those with OAs, caused shifts in a diverse range of bacterial taxa, many of which are closely related to beneficial groups of bacteria that have either ‘biofertilizing’ or ‘biocontrol’ effects on plant growth (Richardson et al., 2009). Such bacteria may contribute to biogeochemical cycling of nutrients, including N (e.g. Rhizobiales, Burkholderia, Nitrosomonadeles, Nitrospira), phosphorus (e.g. Actinomycetales) and sulphur (e.g. Chlorobi). Other more specific plant growth-promoting bacteria such as Bacillus pp. produce phytohormones that are able to stimulate ectomycorrhizal colonization of roots (Frey-Klett et al., 2007) or are antagonistic (e.g. Pseudomonadale and Bacilli) toward plant pathogens and pests (Emmert & Handelsman, 1999).

Although significant effects of OAs on the composition of soil bacterial communities were demonstrated, it is important to recognize that the soil microcosms and artificial REC solutions used are a representation only of the mechanisms that may contribute to microbial diversity in the rhizosphere. Clearly, further studies are required to establish links between such controlled systems and the characterization of the rhizosphere in undisturbed soils to better understand the interactions between root exudates and microbial communities. However, such interactions are complex and their investigation remains a significant challenge. Nonetheless, investigation of the role of artificially constructed REC solutions in influencing microbial communities as shown here, combined with new techniques for in situ sampling of rhizosphere soils (Shi et al., 2011), provides new opportunities for this to be achieved. In addition to the role of low-molecular-weight OAs in directly mediating rhizosphere processes (e.g. Jones, 1998), OAs potentially play a more important role than sugars in shaping the structure of microbial communities within the rhizosphere.

Supporting Information

Table S1. Characteristics of the soil used in this study.

Table S2. Numbers of bacterial taxa in each of the phylogenetic groups found in the responsive communities in relation to REC solutions or compounds (S, sugars; Q, quinic acid; L, lactic acid; M, maleic acid; see Table 1 for descriptions).

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Acknowledgements

This work was funded by the New Zealand Tertiary Education Commission and a travel grant from the Bio-Protection Research Centre to Shengjing Shi for a 4-week visit to Berkeley. This work was also funded in part by a Seaborg Fellowship to KMD, and in part under DOE-LBNL contract DE-AC02-05CH11231. We thank Drs Darren Smalley (AgResearch, New Zealand) and Christian Walter (Scion, New Zealand) for helpful discussions during experiments. We also thank Dr Susan Worner (Lincoln University) for assistance with the primer 5 statistical package and Dr Steve Wakelin (AgResearch, New Zealand) for helpful comments on the manuscript.

References

View Abstract