Characterization of depth-related changes in bacterial community compositions and functions of a paddy soil profile


  • Jing Huang,

    1. Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author
  • Xiafang Sheng,

    Corresponding author
    • Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author
  • Linyan He,

    1. Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author
  • Zhi Huang,

    1. Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author
  • Qi Wang,

    1. Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author
  • Zhendong Zhang

    1. Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing, China
    Search for more papers by this author

Correspondence: Xiafang Sheng, Key Laboratory of Agricultural and Environmental Microbiology, Ministry of Agriculture, College of Life Science, Nanjing Agricultural University, Nanjing 210095, China. Tel.: +86 25 84395125;

fax: +86 25 84396326; e-mail:


Depth-related changes in bacterial community structures and functions were analyzed in a paddy soil profile using denaturing gradient gel electrophoresis (DGGE) and a metabolic profiling technique (BIOLOG ECO plates). Canonical correspondence analysis (CCA) was used to analyze the correlations between the relative abundance of bacterial groups and soil-available elements. DGGE and sequencing analysis revealed 12 classes and one unknown bacterial group. At the family level, Comamonadaceae and Moraxellaceae dominated through the soil profile, while Acidobacteriaceae and Nitrospiraceae dominated in the deepest layer. In addition, Streptococcaceae dominated and was only observed in the deeper layers. Metabolic profiles revealed the greatest carbon source utilization capacity in the surface layer, and no significant differences between upper and deeper soil layers. The carbon sources utilized by microorganisms were different among the different layers. CCA indicated that soil-available Mn, Ca, Cu, Al, and K concentrations were positively correlated with the relative abundance of Comamonadaceae, Moraxellaceae, Streptococcaceae, Microbacteriaceae, Nocardioidaceae, and Nitrospiraceae in the profile. The results showed that the paddy soil profile harbored diverse bacterial communities and experienced depth-related changes in community structure and carbon source utilization. The bacterial communities and functions might be shaped by the soil edaphic characteristics along the soil profile.


Microorganisms inhabiting the surface and subsurfaces of soil profile are playing important roles in biogeochemical processes and are vital in maintaining soil ecological function (Torsvik & Ovreas, 2002; Lambais et al., 2008; Bardhan et al., 2012). Previous studies of microbial community diversity and distribution in soil have been predominantly limited to the surface horizons where the densities and activities of microorganisms are highest (Asakawa & Kimura, 2008; Upchurch et al., 2008; Lopes et al., 2011). As a consequence, the subsurface soil horizon has not been investigated in detail (Ekelund et al., 2001; Griffiths et al., 2003; Hansel et al., 2008).

Rice is the most important food for more than 50% of the world's population, and China is the world's largest rice producer (Köegel-Knabner et al., 2010), and paddy soils are very important for rice production in China. The unique water management of paddy fields brings about leaching of iron, manganese, inorganic substances, nutrient salts, and organic substances from plowed soil layer and enrichment of the subsoil (Kimura et al., 2004; Watanabe et al., 2010). These processes of pedogenesis are specific to paddy soil and result in unique paddy soil profiles (Watanabe et al., 2010). Although the studies on the microbial communities along soil profiles have been reported (Fierer et al., 2003; Hansel et al., 2008; Eilers et al., 2012), we know little about the structures and functions of bacterial communities residing in the subsurface paddy soil horizons (Watanabe et al., 2010).

In general, the depth-related differences in physicochemical and structural characteristics of paddy soil profile can provide many microenvironments in which complex microbial populations can evolve (Ranjard & Richaume, 2001). The diversity and function of microbial communities are therefore expected to be structured in relation to depth (Hansel et al., 2008; Eilers et al., 2012). It is important to investigate the depth-related changes in the microbial communities and functions in the paddy profiles to understand the roles that microorganisms play in the soil biochemical processes.

The objectives of this study were to characterize the soil bacterial communities and functions and to correlate the community structures to the soil parameters along a paddy soil profile.

Materials and methods

Soil sampling

The soil samples were collected in a paddy field (the rice has been harvested) in Yanting, Sichuan Province (31°16′N, 105°27′E) in October 2011. Yanting paddy soil was classified as purple-orthic primosols developed from purple sand shale. No visual pedological characteristics were observed in the Yanting soil profile, so the depth increments for soil sampling used were 0–10 cm (P1-layer), 10–20 cm (P2-layer), 20–40 cm (P3-layer), 40–60 cm (P4-layer), 60–80 cm (P5-layer), and 80–100 cm (P6-layer) for the soil profile. Three replicate soil samples (about 600 g of each sample) were collected from each layer by digging vertically from the surfaces at three randomly chosen locations within one layer. Soil sampling tools were sterilized by flame sterilization between sample collection, and the soil samples were placed in sterile bags to prevent cross-contamination. Each soil sample was thoroughly mixed and sieved (2 mm). The soil samples were stored at 4 °C for physical–chemical analyses and at −20 °C for molecular analysis.

Analysis of physicochemical properties

The moisture content was measured by gravimetric method using fresh samples (Angel et al., 2010). Briefly, soil samples were dried in an oven at 105 °C to constant weight. The moisture content of the soil samples was determined based on the fresh and dry weights of the soil samples. For determination of organic material (OM) content and soil pH, subsamples were air-dried and sieved to < 2 mm. Soil pH was measured with a digital pH meter by mechanically shaking soil (10 g) in 1 : 2.5 (soil: H2O w/v) water suspensions at 200 r.p.m. for 1 h. OM content was determined titrimetrically. The urease activity in the soil was determined by the method of Wang et al. (2006). The invertase activity in the soil was measured according to the methods of Schinner & Von Mersi (1990) and Stemmer et al. (1999). Different particle-size soil fractions (2000–200, 200–63, and < 63 μm) were collected with wet sieve method using the corresponding mesh size according to Zhang et al. (2007) with some modification. The clay fraction (< 2 μm) included in < 63 μm was separated by centrifugation at 250 g for 5 min. The pellets were collected and assigned to 63–2 μm. The supernatants (containing < 2 μm fraction) after three centrifugations together were concentrated and dried to weigh the clay fraction. To determine the available elemental concentrations, 2.5 g of soil sample was mixed with M3 soil test extractant (Mehlich, 1984) by shaking at 25 °C, 200 r.p.m. for 15 min. The filtrates were then used to analyze the major element (Fe, Si, Mn, Mg, Ca, Cu, Al, and K) concentrations by ICP-OES (Inductively coupled plasma-optical emission spectrometer, Optima 2100 DV; Perkin Elmer).

DNA extraction and PCR–DGGE

Approximately 0.5 g of soil samples were used to isolate the total community DNA using the ENZA at soil DNA extraction kit (Omega Bio-Tek, Inc., Norcross, GA) according to the manufacturer's instructions. Highly variable V3 regions of microbial 16S rRNA gene were amplified from all samples by a nested PCR approach (Ye et al., 2009; Cleary et al., 2012). Briefly, the 16S rRNA genes were amplified using the bacterial universal primers 27F/1492R (Lane, 1991) in the first round of amplification, followed by a second round of amplification using bacterial denaturing gradient gel electrophoresis (DGGE) primer set 341FGC/534R (Ellis et al., 2003).

The reaction mixture in a volume of 25 μL contained 1 μL of DNA templates, 12.5 μL TaKaRa (TaKaRa Biotechnology Co., Ltd, Dalian, China) Ex TaqMix, and 0.2 μM of each primer. The conditions of PCR amplification of bacterial 16S rRNA gene for the first round were as follows: an initial denaturing step of 5 min at 95 °C followed by 28 cycles of 45 s at 95 °C, 45 s of annealing at 55 °C, and a 90 s extension at 72 °C, and a final extension step at 72 °C for 7 min. The second round PCR amplification was performed using an initial denaturing step of 5 min at 95 °C followed by 30 cycles of 30 s at 95 °C, 30 s of annealing at 55 °C, and a 45 s extension at 72 °C, and a final extension step at 72 °C for 7 min. PCR products were checked with electrophoresis on agarose gels.

DGGE analysis was performed with the DCode Universal Mutation Detection System (Bio-Rad, Hercules, CA). 20 μL PCR products of each sample were loaded onto 8% acrylamide gel with a 45–75% denaturing gradient (urea and formamide) as described by Das et al. (2007).

Analysis of DGGE band pattern

In DGGE fingerprint profile, each band was designated as an operational taxonomic unit. The Shannon index of diversity (H′), determined as H′ = −∑(ni/N)ln(ni/N), where ni is the height of the peak and the N is the sum of all peak heights of the densitometric curve, was employed to estimate the diversity of PCR–DGGE ribotypes of each sample. The richness of the samples was estimated based on the number of bands per lane. Values obtained for three DNA samples extracted from the three soil samples per layer were averaged, and diversity indices were subjected to analysis of variance (anova). The means were compared with the Duncan's test using the spss 17.0 based on the presence or absence of bands at certain positions in each lane. DGGE profiles were grouped by the clustering, by utilizing the Dice coefficient of similarity (LaPara et al., 2002).

DGGE banding profiles for bacterial communities were digitized after average background subtraction for the entire gels. The data process referred to Costa et al. (2006). Standardization of bands automatically identified by quantity one was carefully checked manually further, and correction was done when needed. A table containing band positions and their corresponding relative intensities (peak areas) was obtained for further statistical analyses. The relative intensity of a specific band was transformed according to the sum of intensities of all bands in a pattern.

Recovery of bands from DGGE gels and sequence analysis

Bands representing ribotypes of interest were excised with a scalpel from the gel, put into sterile microcentrifuge tubes containing 50 μL of sterile double-distilled water and stored at 4 °C overnight. About 1–2 μL of the resulting elution suspensions was used as templates for a PCR with the same conditions mentioned above to re-amplify 16S rRNA gene fragments from the excised band with the V3 primer without GC-clamp. The re-amplified PCR products were ligated into pMD19-T vector (TaKaRa) and transformed into competent cell (Escherichia coli DH5α) following the supplier's instructions. Clones containing inserts were selected and sequenced with an ABI model 3730 DNA sequencer (GenScript, Nanjing, China). The sequences were BLAST-assisted searches of the NCBI database, and the closest match of known phylogenetic affiliation was used to assign the bands to taxonomic groups. The sequences determined in this study have been deposited in the EMBL and GenBank database under accession numbers HE866980HE867024.

Bacterial community metabolic (BIOLOG) profiles

Metabolic profiles of soil microbial communities were assessed using BIOLOG® 96-well Eco-Microplates (Biolog Inc., Hayward, CA) with 31 different carbon sources and a negative control (water), repeated three times in each microplate. From each soil sample, the microbial cells were extracted as described by Hitzl et al. (1997). One hundred and fifty microlitre of a soil suspension [the soil suspension had a density of 3 × 103–4 × 104 cells mL−1 based on dilution-plate method incubated in Luria–Bertani's (LB) agar at 28 °C for 3 days] was dispensed into each of the 96 wells, and then the microplates were incubated at 28 °C in the dark for 7 days. Color development (reflecting carbon utilization) in the wells was followed by absorbance measurements at 590 nm every 24 for 168 h using the Microstation ELX808BLG (Biolog TEK).

Analysis of BIOLOG data was as followed by an average well color development (AWCD) method: AWCD = ∑(C−R)/31, where, C is color production within each well (optical density measurement), R is the absorbance value of the plates control well (Garland & Mills, 1991). Diversity was evaluated by calculating Shannon's diversity index (H′): H′ = −Σpi.lnpi, where pi is ratio of activity (optical density reading) on the i th substrate to the sum of activities on all substrates. The functional structure of microorganisms was characterized by classifying treatments according to their substrate utilization patterns using principal component analysis (PCA). The diversity indices and PCA data presented were based on 96-h incubation readings (Garland et al., 2001). Above statistical analyses were done using spss 17.0 software and canoco for Windows 4.5.

Statistical analysis

One-way analysis of variance (anova) and the Tukey's test (< 0.05) were performed to compare the mean values for different samples at the 95% confidence level. All the statistical analyses were carried out using spss 13.0 for Windows (SPSS Inc., Chicago, IL). Canonical correspondence analysis (CCA) was performed to ordinate the spatial compositions of the bacterial community to the measured soil parameters. Monte Carlo reduced model tests with 499 unrestricted permutations were used to statistically estimate the significance of the first canonical axis and of all canonical axes together (Zhang et al., 2011).


Changes in physicochemical properties with depth

The concentrations of available Al and Fe decreased with depth. However, the concentrations of available K, Mg, and Mn exhibited irregular fluctuation and did not display any obvious depth-related effects. In addition, the available Si content did not show significant change among the different horizons of the soil profile (Supporting Information, Table S1). The pH ranged from 7.58 in horizon of P2 to 8.09 in horizon of P4. No significant changes in pH values between upper horizons of P1 and P2 and between deeper horizons of P5 and P6 were observed (Table S1). The content of OM was higher in the upper horizons of P1 and P2, significantly decreased about four- to sixfolds in horizon of P4 (Table S1). However, no significant changes in OM content were observed between the deeper horizons (Table S1). Significant differences in the contents of fine sand (63–200 μm) and sand (200–2000 μm) were observed along the soil profile, however, no significant differences were observed in the contents of clay and silt among six horizons of the soil profile (Table S1). The horizon P3 contained the most fine sand, but the least sand, and the horizon P2 contained the most sand. The sand/silt/clay ratios for P1, P2, P3, P4, P5, and P6 were 22 : 5 : 1, 37 : 4 : 1, 30 : 4 : 1, 37 : 8 : 1, 35 : 7 : 1, and 43 : 8 : 1, respectively. The groundwater table depth of the paddy soil profile was ≥ 100 cm. Surface horizon of P1 had the maximum moisture content (23%). Horizon of P3 had the lowest moisture content (14.5%). The moisture content increased with depth from horizons of P3–P6. The urease and invertase activities in the soils were higher in the upper horizons of P1 and P2, significantly decreased with depth (Table S1). However, no significant changes in urease and invertase activities were observed between the deeper horizons (Table S1).

DGGE analysis of soil microbial community

As shown in Fig. 1, the soil samples from the surface layer of P1 showed a more complex DGGE pattern than those from the subsurface layers, indicating the presence of a higher number of different bacterial taxa in the surface soil samples (Fig. 1a). Besides the shared bands among the lanes, the special bands were found in the DGGE profiles of soil samples among the different horizons (Table S2). Cluster analysis, which was performed based on the band patterns on the DGGE gel, revealed the differences among the soil samples at different horizons of the paddy soil profile (Fig. 1b). The profiles of the soil samples were separated into two major clusters: profiles of samples from horizons of P1, P2, and P6 into one and profiles of samples from horizons of P3, P4, and P5 into another group. The soil samples at different clusters showed the low similarity (29%) of the bacterial community. The Shannon diversity (H) and the richness index (S) values to evaluate the bacterial diversity decreased with depth (Fig. 1c). The Shannon diversity (H) of the bacterial communities significantly decreased with depth among the upper layers of P1, P2, and P3, while the Shannon diversity did not significantly change among the deeper layers of P4, P5, and P6.

Figure 1.

DGGE profile of the paddy soil profile (a) and corresponding cluster (b) and diversity indices (c) analyses of triplicate DGGE bacterial community profiles. Data not annotated with the same letter for diversity indices are significantly different (95% confidence interval).

Phylogenetic analysis of bacterial communities

In the present study, 45 representative DGGE bands observed in the patterns of bacterial communities were excised and sequenced. Sequence homologies to known sequences in the NCBI database ranged from 81% to 100%. The soil profile contained a diverse bacterial community, which varied in compositions along the soil profiles (Tables 1 and S3). Phylogenetic affiliations of excised DGGE bands were represented by Gammaproteobacteria (24.4%), Betaproteobacteria (15.6%), Bacilli (11.3%), Deltaproteobacteria (11.1%), Actinobacteria (11.1%), Clostridia (6.5%), Acidobacteria (4.4%), Nitrospirae (4.4%), Alphaproteobacteria (2.2%), Epsilonproteobacteria (2.2%), Spirochetes (2.2%), Chloroflexi (2.2%), and unknown bacterial groups (2.2%) (Table 1). Gammaproteobacteria, Betaproteobacteria, Bacilli, Deltaproteobacteria, and Actinobacteria were the dominant. At the family level, 20 bacterial families were obtained in these samples (Table 1), among which only six families of Comamonadaceae, Moraxellaceae, Desulfovibrionaceae, Microbacteriaceae, Nocardioidaceae, and Nitrospiraceae were detected throughout the paddy soil profile. In addition, three Gram-positive Bacillaceae, Streptococcaceae, and Thermoanaerobacteraceae species were detected in the deeper layers, but not in the upper layers of P1 and P2; Gram-negative Burkholderiaceae, Helicobacteraceae, and Acidobacteriaceae species were not detected in the deeper layers of P3, P4, and P5. Enterobacteriaceae and Pseudomonadaceae species were also not detected in the deepest layer of P6.

Table 1. Distribution and proportion of bacterial community structures based on class level for each soil layer, as assessed by denaturing gradient gel electrophoresis and sequencing, respectively
Bacterial groupLayer of soil profile
Gram-negative bacteria
Sphingomonadaceae 3 ± 0.13 ± 0.21 ± 0.0101 ± 0.024 ± 0.2
Burkholderiaceae 2 ± 0.23 ± 0.70004 ± 0.4
Comamonadaceae 8 ± 211 ± 119 ± 117 ± 0.114 ± 0.48 ± 2
Unclassified Burkholderiales6 ± 0.64 ± 0.60005 ± 1
Enterobacteriaceae 5 ± 13 ± 0.28 ± 0.88 ± 0.36 ± 10
Moraxellaceae 8 ± 211 ± 119 ± 117 ± 0.114 ± 0.48 ± 2
Pseudomonadaceae 2 ± 0.14 ± 0.14 ± 0.24 ± 0.25 ± 0.10
Unclassified Gammaproteobacteria3 ± 0.32 ± 0.1004 ± 0.50
Desulfovibrionaceae 2 ± 0.12 ± 0.14 ± 0.45 ± 0.72 ± 0.12 ± 0.1
Geobacteraceae 4 ± 0.74 ± 0.73 ± 15 ± 0.200
Syntrophaceae 005 ± 0.45 ± 0.39 ± 0.10
Unclassified Deltaproteobacteria11 ± 111 ± 0.32 ± 0.22 ± 0.41 ± 0.0113 ± 2
Helicobacteraceae 4 ± 0.34 ± 0.40000
Acidobacteriaceae 7 ± 19 ± 100010 ± 2
Nitrospiraceae 7 ± 0.58 ± 0.52 ± 0.31 ± 0.011 ± 0.110 ± 0.8
Dehalococcoidaceae 5 ± 0.35 ± 0.92 ± 0.2005 ± 0.4
Spirochetaceae 3 ± 0.23 ± 0.503 ± 0.600
Gram-positive bacteria
Brevibacteriaceae 01 ± 0.015 ± 0.85 ± 0.65 ± 10
Microbacteriaceae 7 ± 1.64 ± 0.52 ± 0.22 ± 0.22 ± 0.35 ± 0.5
Nocardioidaceae 7 ± 0.74 ± 0.83 ± 0.23 ± 0.210 ± 0.88 ± 1.5
Bacillaceae 005 ± 0.48 ± 17 ± 19 ± 1
Streptococcaceae 0013 ± 1.613 ± 0.913 ± 0.60
Thermoanaerobacteraceae 002 ± 0.71 ± 0.11 ± 0.19 ± 1
Unclassified Clostridiales4 ± 0.24 ± 0.10000
Unclassified2 ± 0.101 ± 0.011 ± 0.15 ± 0.60

Impact of soil vertical physicochemical changes on bacterial community

The relationship between the geochemical parameters of the paddy soil profile samples and the bacterial families was investigated by CCA (Fig. 2). CCA showed the stronger correlation values found between the concentrations of the soil mineral elements and the relative abundance of bacterial groups. For example, the significantly positive correlations were found between the content of soil available Mn and the relative abundance of Comamonadaceae and Moraxellaceae species (r = 0.634–0.651, < 0.01). The significantly positive correlations between the concentrations of soil available elements and the relative abundance of Microbacteriaceae and Nitrospiraceae species (r = 0.688–0.797, < 0.05 for Cu, and r = 0.468–0.606, < 0.05 for Al) and between the concentrations of soil available elements and the relative abundance of Nocardioidaceae (r = 0.606, < 0.01 for K) and unclassified Deltaproteobacteria (r = 0.736, < 0.01 for Cu; r = 0.497, < 0.05 for Fe; r = 0.554, < 0.05 for Al; r = 0.548, < 0.05 for OM) species were also found.

Figure 2.

CCA ordination diagram of environmental factors in relation to bacterial phylogenetic groups in the family levels. Environmental parameters are indicated by lines with arrows, and phylogenetic groups are represented by triangles. The points indicate samples.

Substrate utilization analysis

BIOLOG ECO plates were used to describe the relative capacity of substrate utilization and characterize the metabolic diversity of microbial community in different soil layers. The greatest carbon source utilization was observed in the upper layer of P1, which showed utilization of (86 ± 1)% of the 31 carbon sources. Microbial communities showing the lowest substrate utilization capacities were seen in the layer of P4, which utilized only (21 ± 2)% carbon sources. The intermediary substrate utilization of the other samples from top to bottom (P2, P3, P5, and P6 layer) exhibited (44 ± 2)%, (44 ± 1)%, (35 ± 2)%, and (36 ± 4)%, respectively. There were no significant differences in substrate utilization capacity for microbial communities between P2 and P3 layers and between P5 and P6 layers.

The microorganisms in different layers had different AWCD values and Shannon diversity index (Fig. S1). AWCD of the surface layer increased rapidly in the first 48 h and increased with time. The surface layer had the highest AWCD and Shannon diversity index among the different layers of the soil profile, however, no significant differences in the AWCD and Shannon diversity index were observed among the subsurface layers of the soil profile before 96 h of incubation time.

The PCA loading plot (Fig. 3) showed that the substrates that were more utilized by microorganisms in the layer of P1 than in the other layers included α-d-lactose (C07), Glycyl-l-glutamic acid (C29), itaconic acid (C21), α-ketobutyric acid (C22), glucose-1-phosphate (C14), α-cyclodextrin (C04), phenylethyl-amine (C30), 4-hydroxy benzoic acid (C19), l-threonine (C28), d-galacturonic Acid (C17), and l-phenylalanine (C26), while the substrates including d-galactonic acid γ-lactone (C16), d,l-α-glycerol (C15), d-xylose (C09) and d-glucosaminic (C13) utilized more in P2 and P5 layer. The substrate d-mannitol (C11) was more utilized in P4 layer, while tween 40 (C02) and glycogen (C05) were more utilized in the deepest layer of P6.

Figure 3.

PCA factor loading plot at different depths. The italic numbers ranging from 1 to 31 prefixed with C indicate 31 different carbon sources. C01, pyruvic acid methyl ester; C02, Tween 40; C03, Tween 80; C04, α-cyclodextrin; C05, glycogen; C06, d-cellobiose; C07, α-d-lactose; C08, β-methyl-d-glucoside; C09, d-xylose; C10, I-erythritol; C11, d-mannitol; C12, N-acetyl-d-glucosamine; C13, d-glucosaminic acid; C14, glucose-1-phosphate; C15, d,l-α-glycerol; C16, d-galactonic acid γ-lactone; C17, d-galacturonic acid; C18, 2-hydroxy benzoic acid; C19, 4-hydroxy benzoic acid; C20, γ-hydroxybutyric acid; C21, itaconic acid; C22, α-ketobutyric acid; C23, d-malic acid; C24, l-arginine; C25, l-asparagine; C26, l-phenylalanine; C27, l-serine; C28, l-threonine; C29, glycyl-l-glutamic acid; C30, phenylethyl-amine; C31, putrescine. The points indicate samples.


Microorganisms in paddy soil may be responsible for the decomposition and translocation of OMs and element (carbon, iron, and manganese) cycling in the paddy soil profile (Köegel-Knabner et al., 2010; Watanabe et al., 2010). However, there was little information available on whether certain bacterial taxa were restricted to specific soil depths and how the bacterial communities and functions changed within a given soil profile (Eilers et al., 2012).

The diversity typically decreased for bacterial communities with depth in a soil profile (Hansel et al., 2008; Will et al., 2010; Eilers et al., 2012). In our study, the DGGE profiles showed the similar depth-related changes in the diversity of bacterial communities (Fig. 1, Table 1). Despite the great diversity of bacterial populations in the paddy soil profile, it was clear that within the soil profile, certain Gram-positive and Gram-negative bacterial populations were dominant only in certain subsurface layers (Table 1). Blume et al. (2002) showed that microbial communities generally shift from greater Gram-negative dominance at the soil surface to greater Gram-positive dominance at deeper soil depths. Members of Proteobacteria, Firmicutes, and Nitrospira are metabolically diverse bacterial groups and are commonly found in soil habitats (Fierer et al., 2003; Asakawa & Kimura, 2008; Hansel et al., 2008). With the BIOLOG system, Braun et al. (2006) and Griffiths et al. (2006) showed that the capability of utilizing diverse substrates decreased with soil depth. Metabolic profiles revealed a decreased capacity for carbon source utilization down the paddy soil profile (Fig. S1), however, similar carbon source utilization capacity for communities was observed between the deeper layers. While the carbon source species utilized by communities were different among the different layers of the paddy soil profile. Furthermore, the bacterial community and function had the similar changes along the soil profile (Figs 1c and S1). The bacterial community and function diversities were significantly higher in the surface layers than in the subsurface layers, but no significant changes among the deeper layers. Furthermore, the difference in bacterial community resulted in the difference in the carbon source species utilized by communities along the paddy soil profile (Table 1, Fig. 3).

Soil physicochemical changes along the soil profiles are likely to be the primary factors controlling the nature and properties of the bacterial communities residing in the soil profile (Ranjard & Richaume, 2001; Fierer et al., 2003). In the soil profile, the significant depth-related differences (< 0.05) in the available Al and Fe contents, pH, organic matter, and moisture contents, and the ratio of sand in the soil were observed (Table S1). The above multiple soil factors may be responsible for the observed changes in bacterial community composition and function through the paddy soil profile (Fig. 2) (Agnelli et al., 2004; Preem et al., 2012). Furthermore, the bacterial communities and metabolic profiles can be shaped by carbon quantity and quality through the profiles (Eilers et al., 2012). In this study, organic matter quantity and quality may explain the differences in the communities and metabolic profiles along the paddy soil profile. Furthermore, the effects of soil type and rice cultivars as well as groundwater level on the bacterial community and function diversity through paddy soil profiles need to be further investigated.

This study provided the characterizations of bacterial community and function diversity in the paddy soil profile developed from purple sand shale. Our study demonstrated that the paddy soil profile harbored highly diverse bacterial communities and functions. Different depth-related change patterns in bacterial community structure and functional diversity for substrate utilization were observed in the paddy soil profile. The paddy soil edaphic characteristics might shape the bacterial communities and functions throughout the soil profile. These findings may have important implications for a greater understanding of the different indigenous bacterial communities within these environments and their relevance to the biological processes.


Support was provided by the Chinese National Natural Science Foundation (41071173).