Effect of soil aggregate size on methanogenesis and archaeal community structure in anoxic rice field soil

Authors


*Corresponding author. Tel.: +49 (6421) 178 830; Fax: +49 (6421) 178 809 friedric@mailer.uni-marburg.de

Abstract

In anoxically incubated slurries of Italian rice field soil, CH4 production is initiated after a lag phase during which ferric iron and sulfate are reduced. The production of CH4 was affected by the size of soil aggregates used for the preparation of the soil slurry. Rates of CH4 production were lowest with small aggregates (<50 and 50–100 μm), were highest with aggregates of 200–2000 μm size and were intermediate with aggregates of 2000–15 000 μm size. The different amounts of CH4 accumulated were positively correlated to the concentrations of acetate, propionate and caproate that transiently accumulated in the slurries prepared from different aggregate sizes and also to the organic carbon content. The addition of organic debris that was collected from large-size aggregates to the aggregate size fractions <200 and <50 μm resulted in an increase of CH4 production to amounts that were comparable to those measured in unamended aggregates of 200–2000 μm size, indicating that CH4 production in the different aggregate size fractions was limited by substrate. The distribution of archaeal small-subunit rRNA genes in the different soil aggregate fractions was analyzed by terminal restriction fragment length polymorphism which allowed seven different archaeal ribotypes to be distinguished. Ribotype-182 (consisting of members of the Methanosarcinaceae and rice cluster VI), ribotype-389 (rice cluster I and II) and ribotype-820 (undigested DNA, rice cluster IV and members of the Methanosarcinaceae) accounted for >20, >30 and >10% of the total, respectively. The other ribotypes accounted for <10% of the total. The relative quantity of the individual ribotypes changed only slightly with incubation time and was almost the same among the different soil aggregate fractions. Ribotype-389, for example, slightly decreased with time, whereas ribotype-182 slightly increased. At the end of incubation, the relative quantity of ribotype-182 seemed to be slightly higher in soil fractions with larger than with smaller aggregates, whereas it was the opposite with ribotype-80 (Methanomicrobiaceae) and ribotype-88 (Methanobacteriaceae). Ribotype-280 (Methanosaetaceae and rice cluster V), ribotype-375 (rice cluster III), ribotype-389 and ribotype-820, on the other hand, were not much different among the different soil aggregate size fractions. However, the differences were not significant relative to the errors encountered during the extraction of polymerase chain reaction (PCR)-amplifiable DNA from soil. In conclusion, soil aggregate size and incubation time showed a strong effect on the function but only a small effect on the structure of the methanogenic microbial community.

1Introduction

Rice fields annually contribute about 100±50 Tg of the greenhouse gas CH4 to the atmospheric budget [1]. Methane is produced when flooded rice field soils become anoxic and reduced. After cropping, the fields are drained and the soils become dry and oxic until the next cultivation. The chemical changes after flooding of anoxic paddy soils and the microbiological processes leading to CH4 production have been investigated in detail [2,3]. Vigorous CH4 production is usually not observed before oxidants such as ferric iron and sulfate are largely reduced. However, trace amounts of CH4 are usually produced immediately upon flooding in the presence of these oxidants [4,5].

Soils contain diverse groups of microorganisms, including the methanogenic archaea, which all contribute to the decomposition of organic matter under anoxic conditions [6,7]. Environmental factors such as soil type, rice variety, field management or season are possibly important determinants of the composition of the methanogenic microbial community in rice field soils. Mayer and Conrad [8] reported that the methanogenic population size in Italian rice field soil was almost constant, even during dry fallow periods. Asakawa and Hayano [9] also reported that changes in the populations of H2-CO2-, methanol- and acetate-utilizing methanogens, determined by the most probable number (MPN) cultivation method, were negligible, when investigating Japanese paddy soils under different moisture regime (flooded or non-flooded), crop rotation (rice or wheat), fertilizer treatment and soil depth (0–1, 1–10 and 10–20 cm). Recent studies of archaeal small-subunit (SSU) rDNA retrieved from anoxic Italian rice field soil demonstrated a larger archaeal diversity than suggested by MPN analysis [10–13]. Together with analyses of the archaeal communities on rice roots [10,11,14,15] these authors demonstrated a relatively large diversity of Archaea in rice fields, including members of the families Methanosarcinaceae, Methanosaetaceae, Methanomicrobiaceae and Methanobacteriaceae (taxonomy according to [16]). In addition, members of the kingdom Euryarchaeota have been detected and grouped into novel phylogenetic clusters termed rice clusters I, II, III and V [10,11]. Finally, members of the kingdom Crenarchaeota have been detected and grouped into phylogenetically defined clusters termed rice clusters IV and VI [10–12].

Soil microbial communities are known to respond to the agricultural management practices and environmental perturbations [17–19]. In soils, bacteria tend to occur immobilized through adsorption to the soil particles [20]. These organisms also play an important role in the formation and stabilization of soil aggregates (groups of primary particles that adhere to each other more strongly than to surrounding soil particles). Tisdall and Oades [21] suggested that soil aggregation is a complex hierarchical process. They classified the organic binding agents, depending on the effectiveness of these agents, at different stages in the structural organization of aggregates into: (a) transient, mainly polysaccharides, often produced by microorganisms as extracellular mucilages or gums; (b) temporary, such as root and fungal hyphae, and (c) persistent, such as humic and phenolic compounds associated with polyvalent metal cations and strongly-sorbed polymers. Disintegration of the bulk soil into aggregates can occur by disruptive forces which include cultivation (tillage/plowing), flooding and erosion (wind and water). The resulting aggregate size distribution in soils is often considered as a function of soil properties and also of the cultivation method [22].

Here, we have studied the initiation of CH4 production upon flooding of dry soil aggregates of different size ranges, together with the archaeal community structure. Different specified aggregate size fractions of dry Italian rice field soil were obtained by rotary sieving. These samples were incubated as anoxic slurries at 25°C and the change of concentrations of CH4, H2, CO2 and volatile fatty acids was analyzed. The archaeal community structure in different aggregate size fractions was studied by terminal restriction fragment length polymorphism (T-RFLP), a molecular genetic fingerprinting technique [12,13,23–25].

2Materials and methods

2.1Soil samples and slurry incubation

Soil samples were collected in 1993 and 1997 from the experimental fields of the Italian Rice Research Institute, Vercelli, and were stored as dry lumps at room temperature. The site descriptions and soil characteristics are presented in previous reports [26,27]. The larger dry lumps were broken manually by slight impounding before passing the samples through stainless steel sieves of 15 000, 2000, 500, 400, 200, 100 and 50-μm mesh size on a mechanical rotary shaker (K. Retsch, Haan, Germany) at a shaking speed of 50 rpm for 5 min. The different-sized fractions of soil aggregates (the fractions passing through the 15 000 μm sieve and collected on the 2000 μm sieve are designated as aggregates 2000–15 000 μm and analogously for the other fractions) were collected and stored in large plastic containers. Total carbon and nitrogen contents of these soil aggregates (Table 1) were determined with a CHN analyzer (Analytical Chemical Laboratory of the Phillips University, Marburg, Germany). For determination of the available iron content, soil samples (0.5 g) were extracted with 4.5 ml of 0.5 M HCl at room temperature for 24 h. After centrifugation at 10 000 rpm for 5 min, the supernatant was mixed with 4-(2-pyridylazol) resorcinol and concentrations of Fe(II) and Fe(III) were measured by ion chromatography using a high performance polymer-coated silica-based cation exchange column [28]. The sulfate content was determined after 2 h extraction at room temperature using 1 ml of sodium carbonate (100 mM) or by centrifuging the aliquots of the soil slurries and filtration through a 0.2-μm membrane filter (Minisart RC15, Sartorius, Göttingen, Germany). The sulfate concentrations were then analysed in a Sykam ion chromatographic system [29].

Table 1.  Contents of total carbon, total nitrogen, available iron and sulfate in different soil aggregate sizes
Soil aggregate size (μm)Carbon (%)Nitrogen (%)Iron (μmol g−1)Sulfate (μmol g−1)
2000–15 0001.46±0.160.17±0.0499.66±11.621.86±0.02
500–20001.36±0.370.12±0.0397.88±6.192.07±0.13
200–5001.87±0.590.20±0.01105.16±2.022.73±0.38
<2001.50±0.340.14±0.06113.27±0.862.24±0.13
50–1001.31±0.140.12±0.0471.22±17.751.94±0.08
<501.26±0.010.15±0.01120.57±6.812.58±0.03

For the preparation and incubation of slurries, 10 g dry weight (d.w.) of different-sized soil aggregates were filled into 120-ml serum bottles and suspended at a weight ratio of 1:1 in anoxic and sterile distilled water. The serum bottles were closed with sterile butyl rubber stoppers, crimped with aluminum caps, and flushed with N2. Serum bottles were incubated in the dark at 25°C without shaking to avoid damage of the methanogenic community [30].

In another experiment, the soil aggregate fractions of 500–2000 and 200–500 μm were mixed with water at a 1:1 soil and water ratio, stirred briefly and then centrifuged at 10 000 rpm for 5 min on a centrifuge (Universal 30F; Hettich, Tuttlingen, Germany). Pieces of rice roots and straw were collected from the supernatant and then dried at 60°C for 2 days. These native organic substances were added as substrate amendment (150 mg) to the soil slurries prepared from aggregates <50 and <200 μm. The preparation and incubation of the substrate-amended soil aggregate slurries were carried out as described above. The experiments were carried out by preparing in parallel numerous bottles which were subsequently sacrificed in triplicate at each time of sampling.

2.2Analysis of gases

At given time intervals, gas samples were withdrawn from the headspace after vigorously shaking the bottles by hand to allow equilibration between the liquid and gas phases by using a gas-tight pressure-lock syringe. The concentrations of H2, CO2 and CH4 in the gas samples were measured by gas chromatography as previously described [31]: H2 by using a molecular sieve 5-Å column (80–100 mesh, 70 cm length) with HgO-to-Hg vapor conversion detector (RGD2, Trace Analytical, Stanford, CA, USA); CO2 after conversion to CH4 by a methanizer (Ni-catalyst at 330°C, Chrompack-Middleburg, The Netherlands) and CH4 by using an 80-cm-long Propak Q 60–80 mesh column at 50°C on a Shimadzu GC 8A with flame ionization detector (Shimadzu, Japan).

2.3Analysis of volatile fatty acids

Volatile fatty acids were measured at regular time intervals. To collect the liquid samples, the soil slurry in serum bottles was mixed briefly on a vortex mixer, transferred into 2.0-ml microcentrifuge tubes and then centrifuged at 14 000×g for 5 min at 4°C. The supernatant was collected and filtered through 0.2-μm membrane filters (Minisart SRP-15, Sartorius). The filtered samples were stored frozen at −20°C until analysis. The concentrations of the volatile fatty acids were analyzed in a high pressure-liquid chromatograph (Sykam, Gauting, Germany) with refraction index and UV detectors [32].

2.4DNA extraction from soil slurries

The sampling for molecular analysis was carried out with incubated soil slurries and dry soil samples. Slurry samples of 0.5 ml were collected at different intervals after mixing of the serum bottles thoroughly, and then stored in 2.0-ml microcentrifuge tubes at −20°C until analysis. For extraction of total community DNA from dry soil samples, 0.5 g of different soil aggregate size fractions were filled into microcentrifuge tubes, mixed with 0.5 ml of degassed distilled water and incubated for 1 h.

The direct DNA extraction procedure of Moré et al. [33] was adapted with modifications as described by Henckel et al. [34]. The cells in slurry samples were lysed on a beadbeater (BIO 101 Fast Prep, Savant Instruments, Farmingdale, NY, USA) for 45 s at a setting of 6.5 m s−1 in the presence of an SDS solution. Total soil community DNA was extracted, washed, pelleted and resuspended in 100 μl of Tris–EDTA buffer (10 mM Tris base, 1 mM EDTA, pH 8). Then, the DNA extracts were further purified by centrifugation (2100×g, 2 min) through spin columns (Bio-Rad, Munich, Germany) filled with acid-washed polyvinylpolypyrrolidone (Sigma-Aldrich, Steinhem, Germany) in 30 mM potassium phosphate buffer (pH 8.0) as described by Berthelet et al. [35]. The purified extracts were clear and colorless. The nucleic acid concentration of these purified extracts, representing total soil community DNA, was determined either spectrophotometrically (GeneQuant, Pharmacia Biotech, Upsala, Sweden) or fluorometrically by using the PicoGreen dsDNA quantitation kit (Molecular Probes Europe BV, Leiden, The Netherlands).

2.5PCR amplification of archaeal SSU rDNA

Archaeal SSU rRNA genes were specifically amplified from total soil community DNA extracts using the primers Ar109f (5′-ACK GCT CAG TAA CAC GT-3′) and Ar912r (5′-CTC CCC CGC CAA TTC CTT TA-3′), modified from Großkopf et al. [10]. For PCR amplification, 20 ng of purified DNA extract was added to a final volume of 100 μl of reaction mixture containing 10 mM Tris–HCl, 1.5 mM MgCl2, 50 μM of each deoxynucleoside triphosphate, 0.3 μM of each archaeal SSU rRNA gene-primer and 2.5 U of AmpliTaq DNA polymerase (PE Applied Biosystems, Weiterstadt, Germany). The reverse primer, Ar912r, was labeled at the 5′ terminus with 6-carboxyfluorescein. The PCR thermal profile included the following steps using a GeneAmp PCR system 9700 (PE Applied Biosystems): an initial denaturation for 5 min at 94°C, then 27 cycles of 1 min denaturation at 94°C, 1 min annealing at 52°C and 1 min extension at 72°C, finished by a 3 min final extension at 72°C. PCR amplicons were analyzed by electrophoresis in 1% (w/v) agarose gels and then purified using the QIAquick PCR Purification Kit (QIAGEN, Hilden, Germany). The DNA concentration of the PCR amplicons was determined spectrophotometrically at 260 nm.

2.6T-RFLP analysis

The PCR amplicons were analyzed by T-RFLP [12,13,23,25]. PCR amplicons (75 ng), 2.5 U of TaqI restriction enzyme (Promega, Mannheim, Germany), 1.0 μl of the appropriate incubation buffer supplied by the manufacturer and 1 μg of bovine serum albumin were made up to a total volume of 10 μl and incubated at 65°C for 2 h. The digested fluorescently-labeled SSU rDNA fragments were size-separated on an automated DNA sequencer (Model 373A, PE Applied Biosystems). The RFLP pattern of the 5′-terminal SSU rDNA fragments of each sample was determined in comparison to that of the internal standard by using GeneScan analysis software (version 2.1, PE Applied Biosystems). This software package estimates the length of T-RFs and integrates the fluorescence emission intensity of individual T-RF peaks. Based on the principle that the relative proportion of the integrated fluorescence of each T-RF corresponds to the proportion of each amplicon in the PCR products, the relative abundance of amplicons was estimated as the ratio between the integrated fluorescence of each of the T-RFs and the total integrated fluorescence of all T-RFs [36]. Thus, the percentage distribution of different ‘ribotypes’ (i.e. T-RFs) within the archaeal community structure in a particular soil sample was determined. The major ribotypes consisted of several phylotypes, which were identified by calculating the theoretical lengths of T-RFs from 267 aligned archaeal 16S rRNA sequences by using the ARB software [37], from various clone libraries of archaeal SSU rDNA retrieved from Italian paddy soil (Table 2). The detailed description of the clone libraries are provided elsewhere [10–14,25].

Table 2.  The major ribotypes and the respective phylotypes detected in Italian rice field soil
Ribotype: length of T-RFPhylotypeNumber of positive clones
80Methanomicrobiaceae9
88Methanobacteriaceae18
182Methanosarcinaceae and rice cluster VI73 and 22
280Methanosaetaceae and rice cluster V34 and 5
375rice cluster III6
389rice cluster I and II55 and 3
820undigested DNA (clones belonging to rice cluster IV and Methanosarcinaceae)12 and 4
The major ribotypes are classified based on the theoretical lengths of T-RFs from the aligned SSU rRNA archaeal sequences of clone libraries and also from the T-RFLP analysis using SSU rDNA clone libraries constructed from DNA retrieved from Italian paddy soil. These clone libraries comprise a total of 267 clones described by [10–14,25]. The degree of conservation in restriction site positions in the SSU rDNA was about 90%[25].

3Results

3.1Ecophysiological function of different soil aggregate sizes

In the anoxically incubated slurries of different-sized aggregates from rice field soils, the production of detectable levels of CH4 (assessed on a exponential scale) began as early as after 1–2 days of incubation (data not shown). The onset of CH4 production only on day 2 of incubation was typical for the fractions with soil aggregates <100 μm. CH4 accumulation on a linear scale showed that during an incubation of 25 days less CH4 was produced if small aggregates (<100 μm) were used compared to other aggregate fractions (Fig. 1A). The CH4 production potential among these aggregate fractions exhibited clear differences after about 10 days incubation and could be grouped into three levels. The highest level of accumulation was apparent in the aggregates of 200–500 and 500–2000 μm size. The lowest level was observed with aggregates of 50–100 and <50 μm size. The other two aggregate size fractions (2000–15 000 and <200 μm) showed intermediate levels of CH4 accumulation.

Figure 1.

Temporal change of production of CH4 and volatile fatty acids in different-sized aggregates from Italian rice soil; (A) CH4 partial pressure; (B) acetate concentration; (C) propionate concentration; (D) caproate concentration; the data are means±S.D. of triplicate measurements. For clarity error bars are not shown for the volatile fatty acids, but typically were in the order of CV=<20%.

In order to relate the production potential of different aggregates to the availability of substrates, we measured the total carbon and nitrogen contents in the different soil aggregate size fractions (Table 1) and followed the change of H2 partial pressure (data not shown) and of dissolved volatile fatty acids (Fig. 1). Spearman rank correlation analysis showed that higher amounts of CH4 accumulated were positively correlated with the soil carbon content (r2=0.83). H2 partial pressures increased to high values of 6–14 Pa within 1 day of incubation. Within 3–4 days, the H2 partial pressures rapidly declined and stabilized at values of 0.1–0.5 Pa in all samples except those with the smaller soil aggregate sizes (<100 μm), in which they stabilized around 0.8–2.0 Pa H2. Similar trends in the accumulation of H2, CH4 and CO2 were observed when the soil aggregate size fractions were prepared from soil collected in 1993 and 1997 (data of 1997 are not shown).

Acetate, propionate, caproate, lactate and butyrate were the only volatile fatty acids detectable at concentrations >10 μM. Butyrate was detected only on day 1 (data not shown). Similarly, lactate was >20 μM only on day 1 (data not shown). Acetate (Fig. 1B), propionate (Fig. 1C) and caproate (Fig. 1D), on the other hand, transiently accumulated during the incubation and exhibited different accumulation patterns among the different aggregate size fractions. The maximum concentrations of acetate (r2=0.94), propionate (r2=0.82) and caproate (r2=0.93) were correlated with the amount of CH4 accumulated in the different soil aggregate fractions. Acetate reached concentrations of 1–5 mM (Fig. 1B). In the soil aggregate fractions <100 and 2000–15 000 μm, the maximum acetate concentrations that were reached after 15 days of incubation were relatively low (<2 mM) and decreased thereafter only slightly. In the soil aggregate fractions of 200–2000 μm, on the other hand, much higher acetate concentrations (5 mM) were reached after 10 days incubation, which thereafter decreased to <200 μM acetate until the end of incubation (Fig. 1B).

On a visual inspection, the slurries of the most active aggregate sizes (200–500 and 500–2000 μm) showed root pieces and other organic debris. Increased amounts of native organic substances in these soil aggregates were also evident from the soil carbon content (Table 1). Collection of these native organic substances and addition to the smaller soil aggregate size fractions (i.e. <200 and <50 μm) resulted in increased amounts of CH4 accumulated. Accumulation of CH4 in the amended small aggregate size fractions was then comparable to those in the unamended large aggregate size fractions of 200–500 and 500–2000 μm (data not shown).

3.2Archaeal community structure of different soil aggregate sizes

The total community DNA extracted from slurries prepared with different soil aggregate sizes were used for T-RFLP analysis. An example of such an analysis is shown in Fig. 2A. The T-RFLP analysis identified seven major T-RFs representing seven ribotypes (Table 2). Analysis of the archaeal SSU rDNA sequences of 267 clones retrieved from Italian rice field soil allowed the assignment of the different ribotypes to corresponding phylotypes which had been identified by the construction of phylogenetic trees [25]. Unfortunately, the individual ribotypes were not strictly represented by only one phylotype (Table 2). For example, ribotypes-182, -280 and -389 each comprised two different phylotypes which possibly vary in proportion to each other from sample to sample. However, each ribotype was reproducibly retrieved from duplicate analysis using the same DNA, both in quality and in its relative proportion (determined as peak area) to the total ribotypes (data not shown). The relative proportion of each ribotype to the total was also independent of the number of PCR amplification cycles which were varied between 22 and 44 cycles (Fig. 2B). Only ribotype-389 and -820 exhibited slightly decreased frequencies at PCR cycles >27. Therefore, we routinely used 27 PCR cycles for T-RFLP analysis. Under these conditions, the analysis of triplicate soil samples generally showed an accuracy of better than ±5–10% ribotype frequency (Fig. 2C).

Figure 2.

Community fingerprint determined by T-RFLP analysis; (A) original electropherogram obtained with a soil aggregate fraction of 200–500 μm after 25 days of incubation; (B) relative peak area (%) of the major ribotypes determined by using different numbers of PCR cycles; (C) relative peak area (%) of the major ribotypes determined in triplicate soil samples using 27 PCR cycles; bars give S.D.

The seven archaeal ribotypes were also detected in dry soil samples prior to incubation as anoxic slurries (Fig. 3; incubation time=zero). Ribotypes-182, -389 and -820 were the most abundant ribotypes and made up 16–23, 43–48 and 13–18%, respectively, of the total ribotypes detected in the various dry soil aggregate sizes. The narrow range of percentages indicates that the composition of the archaeal community by the dominant ribotypes was relatively constant at the beginning of anoxic incubation. The other ribotypes were detected at proportions of <10%.

Figure 3.

Relative distribution of the major ribotypes, expressed as percent of the total archaeal gene frequency, in different-sized aggregate fractions during anoxic incubation: (A) ribotype-80 (open symbols), -88 (closed symbols) and -389 (open symbols), representing putative CO2-reducing methanogens; (B) ribotype-182 and -280, representing putative acetoclastic methanogens together with phenotypically-unknown archaea; (C) ribotype-375 and -820, representing phenotypically-unknown archaea. For details on the phylotype-relatedness of different ribotypes, refer to Table 2.

During anoxic incubation, the proportional contribution of the individual ribotypes changed only slightly with incubation time and also was only slightly different in the different soil aggregate size fractions. These changes and differences were not statistically significant when compared to the error (±5–10% ribotype frequency) encountered during T-RFLP analyses of replicate soil samples (Fig. 2C). However, there seemed to be tendencies. For example, the relative proportion of ribotype-80 to the total increased from about 0.5–1.5% in dry soil aggregates to 3–6% in 25-day-old slurries (Fig. 3A). The proportion of ribotype-88 on the other hand decreased from initially 10% to about 5% after 6 days incubation (Fig. 3A). Subsequently, the relative proportion of ribotype-88 tended to increase with higher values for the larger soil aggregate sizes (Fig. 3A). Noteworthy is the relatively high proportion of ribotype-182 which stayed relatively constant with incubation time at about 15–25% among the small-sized aggregates (<100 μm), but tended to increase to >40% among the larger-sized aggregates (Fig. 3B). Ribotype-280 contributed only little (4–9%) to the total archaeal ribotypes and stayed relatively constant with incubation time (Fig. 3B). Ribotype-375, on the other hand, also contributed little (<6%) but tended to increase with incubation time (Fig. 3C). Ribotype-389 was the most abundant ribotype and tended to slightly decrease with time (Fig. 3A). The same was observed with ribotype-820, which was also relatively abundant (Fig. 3C).

4Discussion

The production of CH4 is accomplished by methanogenic archaea which usually disproportionate acetate into CH4 plus CO2, or reduce CO2 with H2 to CH4. In the present study, we demonstrated that the production of CH4 from different-sized aggregates from Italian rice field soil was less if small aggregates (<100 μm) were used, thus confirming earlier results [8]. Furthermore, we showed that anoxic slurries prepared from different soil aggregate sizes exhibited systematic differences in the transient accumulation of metabolic intermediates that were produced and consumed during the methanogenic degradation of soil organic matter. Thus, the amounts of CH4 produced correlated with the maximum concentrations of acetate, propionate and caproate that transiently accumulated in the different aggregate size incubations. More than 80% of the differences in CH4 production were explained by the differences in accumulated volatile fatty acids, which are direct or indirect methanogenic precursors [3]. Apparently, syntrophic microorganisms, which typically degrade volatile fatty acids, such as propionate, butyrate and other longer chain volatile fatty acids, did not fully couple with their syntrophic H2-oxidizing partner organisms as indicated by different degrees of accumulation of these fatty acids in the different aggregate size fractions, whereas CH4 production increased. On the other hand, the composition of the methanogenic archaeal community, which was determined by T-RFLP analysis, was not significantly different among the different soil aggregate sizes and was also relatively constant with time. Thus, differences in volatile fatty acid accumulation between aggregate size fractions were not affected by methanogenic population shifts.

There are suggestions that the decomposition process leads to the development of an aggregate hierarchy [38]. Much of the organic matter in soil is particulate, not evenly distributed and physically protected from microorganisms by adsorption onto inorganic clay surfaces and by entrapment in aggregates. The highest concentrations of volatile fatty acids and the highest rates of CH4 production were observed in intermediate soil aggregate sizes (<200–2000 μm), whereas lower values were found in both the largest (>2000 μm) and the smallest (<100 μm) soil aggregate fractions. Less production of CH4 from small soil aggregates is likely due to less available substrates. Diffusion limitation could have been responsible for reduced microbial activities in the larger size aggregates, but it is unknown to which degree size aggregates remain physically intact after flooding of the soil. Elliott [39] showed that organic matter associated with microaggregates was more recalcitrant than organic matter associated with macroaggregates. Less production of CH4 from large soil aggregates may also be due to less available substrate. Indeed, the intermediate soil aggregate fractions contained the highest amounts of organic debris indicating that this material controlled the accumulation of volatile fatty acids as well as the subsequent CH4 formation. This conclusion was confirmed by stimulated CH4 production in the small-size fractions when native organic debris from the intermediate soil fractions was added.

The ecology of methanogenic archaea in soils is little explored and there is a need to establish the link between the community structure (speciation) and function (metabolic activity). We have used T-RFLP as a molecular fingerprinting technique for the purpose of a direct analysis of the archaeal community structure in the different-sized soil aggregates. This technique enabled us to compare unique T-RFs (ribotypes) generated from the digestion of PCR amplicons with those derived from SSU rRNA gene sequence database in order to make phylogenetic inference. In addition, the relative amount of the individual ribotypes in the PCR amplicons (‘gene frequencies’; [23,36]) were quantified. Our results showed that soil aggregate disruption, cell lysis for extraction of total community DNA, amplification of archaeal SSU rDNA by PCR, restriction and separation of the terminally-labeled fragments were all efficient and reproducible and yielded community fingerprints which were qualitatively and quantitatively consistent.

Thus, we found highly reproducible PCR yields and T-RFLP patterns from replicate aliquots of the same DNA extract. By using the SSU rDNA of Methanosarcina barkeri for PCR reactions and subsequently measuring the dsDNA concentration of amplicons fluorometrically using the ‘PicoGreen’ method [40], we found that the relative proportion of integrated fluorescence of the restriction fragment peak linearly increased with increasing concentrations of amplicon digests (unpublished results). Similarly, duplicate sets of PCRs using the same DNA from a single soil community gave highly reproducible community fingerprints with the same relative distribution of ribotypes. We also found that the relative distribution of the different ribotypes was not significantly affected by the number of PCR amplification cycles if varied between 24 and 32. The kinetic bias effect reported by Suzuki et al. [36] therefore did not seem to pose a problem in the PCR reaction using rice field soil community DNA. Different soil replicates, on the other hand, created a somewhat larger error. Nevertheless, the error of the total procedure using replicate soil samples was only in the order of ±5–10% ribotype frequency, similar to that recently described for T-RFLP analysis of the community of Bacteria in rice field soil [24].

The T-RFs obtained by T-RFLP analysis of our dry rice field soil samples had sizes in the following order of dominance: 389>182>820>280>80>88≥375 bp. Based on the size of the T-RFs, representing individual ribotypes, all the major phylogenetic archaeal groups were identified that had previously been detected in Italian rice field by either isolation and cultivation or by cloning and sequencing, i.e. those belonging to either Methanosarcinaceae, Methanosaetaceae, Methanomicrobiaceae, Methanobacteriaceae or one of the taxonomically undefined archaeal rice clusters I to VI [10–12,14]. The resolution of T-RFLP analysis is generally limited to that of high order taxa due to the variable conservation of restriction site positions in 16S rDNA. In addition, the production of T-RFs of identical size by different taxa with a given restriction enzyme [41] or multiple ribotypes due to 16S rDNA sequence heterogeneity within one taxon [42,43] can pose problems in attributing a phylogenetic position to a particular ribotype. Thus, we were unable to differentiate between members of the Methanosarcinaceae and rice cluster VI, between members of the Methanosaetaceae and rice cluster V and between rice cluster I and rice cluster II. Since the phylogenetic differentiation of the individual ribotypes is based on the sequence information presently available, we can not exclude that ribotype-80, -88, -375 and -820 consisted of more than the one phylotype indicated in Table 2.

Our data suggest that there may be slight temporal changes in the relative quantities of ribotypes, but the dominance of individual ribotypes did not change. The euryarcheotal rice clusters I and II, which are represented by ribotype-389, dominated in the beginning of incubation. Their contribution slightly declined with incubation time with the concurrent increase of ribotype-182 representing members of the Methanosarcinaceae and crenarchaeotal rice cluster VI. These changes suggest that the populations of individual species within the archaeal community changed during the incubation period. Microbial isolates representing rice cluster I and II do not exist and, therefore, the phenotype of these archaeal populations is unknown. Indirect evidence suggests that they represent methanogenic archaea which perhaps utilize H2 or ethanol as electron donors [14]. The genus Methanosarcina, on the other hand, is known for acetoclastic methanogenesis. If the temporal change in the frequency of the two dominant ribotypes (ribotype-182 and -389) is real, the results suggest that acetoclastic methanogens may have slowly replaced CO2-reducing methanogens during the course of incubation. A change in the activity from predominantly H2/CO2-utilizing to acetoclastic methanogenesis has recently been concluded from inhibition studies with anoxic rice field soil [4]. Our results of the process-oriented experiments support this view. While CH4 production began as early as 1–2 days after incubation, the accumulated acetate did not start to decrease before day 10, being consistent with the view that the acetoclastic methanogenic populations increased with time. Similar population increases of acetoclastic Methanosarcina or Methanosaeta species were recently reported for cellulose-degrading anaerobic enrichment cultures inoculated with rice field soil [13] and during sequential reduction processes after flooding of rice field soil [25]. However, a statistically significant proof for the temporal change of the dominant methanogenic populations will only be possible when the individual ribotypes detected by T-RFLP analysis will be determined with an even better accuracy to that in the present study. Further refinement in the procedures for cell lysis, DNA extraction, DNA purification [44] and quantification of DNA from the soil environment [40,45] will possibly enhance the accuracy.

Besides the possible slight change in the populations of H2/CO2-utilizing and acetoclastic methanogenic populations, the frequency of the third-most dominant ribotype-820, representing undigested DNA and clones of crenarchaeotal rice cluster IV and of members of the Methanosarcinaceae, also seemed to decrease slightly with incubation time. We have presently no clue what the physiological consequences of this decrease might be. All the other ribotypes made up <10% of the total.

In general, our assessment of relative frequency of the different ribotypes revealed a relatively stable archaeal community structure both in time and among the different aggregate sizes. Especially with regard to the different soil aggregate fractions, the stable archaeal community patterns were quite surprising. The different soil aggregate fractions revealed quite different patterns of production of volatile fatty acids and CH4; large differences in the archaeal community structure have been reported for rice field soil incubated at different temperatures [12]; different community structures were found for the flora of the bulk soil [10] and the rice rhizoplane [11]; and different structures were also observed among rice field soils sampled from different rice-growing regions (unpublished results). Other studies have shown the presence of phenotypically and phylogenetically distinct members of both Euryarchaetoa and Crenarchaetoa in different soil types [46–48]. Our results, however, clearly show that the predominant representatives of Archaea form a relatively stable part of the microbial communities in the Italian rice field soil. However, we did not estimate the total number of species (‘species richness’) and it may well be that differences occur among the different species belonging to the same ribotype.

Acknowledgements

We thank B. Wagner, M. Klose, J. Scheld, S. Ratering and P. Dunfield for sharing their expertise. B.R. was supported by a BOYSCAST fellowship of the Department of Science and Technology, Government of India.

Ancillary