Microbial succession during a composting process as evaluated by denaturing gradient gel electrophoresis analysis

Authors


K. Ishii, Department of Biological Science, Graduate School of Science, Tokyo Metropolitan University, Hachiohji, Tokyo 192–0397, Japan (e-mail: ishii-kousuke@c.metro-u.ac.jp).

Abstract

Microbial succession during a laboratory-scale composting process of garbage was analysed by denaturing gradient gel electrophoresis (DGGE) combined with measurement of physicochemical parameters such as temperature, pH, organic acids, total dissolved organic carbon and water-soluble humic substance. From the temperature changes, a rapid increase from 25 to 58 °C and then a gradual decrease, four phases were recognized in the process as follows; mesophilic (S), thermophilic (T), cooling (C) and maturing (M). The polymerase chain reaction-amplified 16S rDNA fragments with universal (907R) and eubacterial (341F with GC clamp) primers were subjected to DGGE analysis. Consequently, the DGGE band pattern changed during the composting process. The direct sequences from DGGE bands were related to those of known genera in the DNA database. The microbial succession determined by DGGE was summarized as follows: in the S phase some fermenting bacteria, such as lactobacillus, were present with the existing organic acids; in the T phase thermophilic bacillus appeared and, after the C phase, bacterial populations were more complex than in previous phases and the phylogenetic positions of those populations were relatively distant from strains so far in the DNA database. Thus, the DGGE method is useful to reveal microbial succession during a composting process.

Introduction

Recently, there has been growing interest in composting because it is such an important process for the treatment of organic waste, such as municipal solid waste, grass cuttings and manure, with the resultant compost being used as soil fertilizer. Composting turns easily degradable organic matter into stable matter containing a humic-like substance by passing through a thermophilic stage (Gray et al. 1971; Finstein and Morris 1975). This waste treatment is useful in addressing environmental problems such as global warming, due to the accumulation of man-made greenhouse gases in the atmosphere, because the process emits less CO2 than burning. Monitoring of the microbial succession is important in the effective management of the composting process as microbes play key roles in the process and the appearance of some microbes reflects the quality of maturing compost (Macauley et al. 1993). Many early studies dealt with the isolation and description of various microbes in compost by the classical culture method (Webley 1947a; Webley 1947b; Forsyth and Webley 1948; Shilesky and Maniotis 1969; Gray et al. 1971; Kane and Mullins 1973; Finstein and Morris 1975; Strom 1985a; Strom 1985b). Since this method cannot detect non-culturable species, there is a possibility that organisms which are difficult to isolate are predominant. Hardly- or non-culturable organisms may exist, especially in a later phase of the composting process, because easily utilizable substrates have been depleted and only difficult substrates remain. Indeed, information about bacteria in the later phases of the composting process is scant (Miller 1996). In addition, there has been a lack of investigation into anaerobic thermophilic microbes that are difficult to culture, although composting materials might contain anaerobic environments (Miller 1996). Recently, phospholipid fatty acid (PLFA) analysis has been used to characterize microbes in compost (Hellmann et al. 1997; Herrmann and Shann 1997; Carpenter-Boggs et al. 1998; Klamer and Bååth 1998). The results from these studies suggested the rapid changes of different populations that occur during the composting process. However, such results tend to be limited to culturable microbes or large microbial taxa. Thus, new methods, which can follow population changes in all potential organisms at the genus or species level, are required to investigate microbial succession in the composting process.

Denaturing gradient gel electrophoresis (DGGE) analysis of polymerase chain reaction (PCR)-amplified small subunit (ssu) rRNA genes was applied to microbial ecology and this method was able to detect microbes independent of culture (Muyzer et al. 1993). The ssu rRNA was suitable for inferring the phylogenetic position of a wide range of organisms, including bacteria which are difficult to culture and also anaerobes (Woese 1987; Amann et al. 1995). Thus, this method may provide high-resolution information about the succession containing non-culturable microbes in the composting process.

The object of this study was to reveal the microbial succession during a composting process, especially later phases, independent of the culture method. Therefore, we aimed to investigate the succession of microbial populations in the composting process in a laboratory system for 45 d by DGGE analysis.

Materials and methods

Composting material

Garbage composed of mostly food scraps obtained from restaurants was dried by steam. After being adjusted to 40% (w/w) water content by adding water, 20 kg of well-mixed garbage was used for composting.

Composting facilities

Laboratory-scale composting was performed in a 30-l polyethylene bin with a hole drilled in its side 5 cm above the bottom. The composting material, on a stainless mesh fixed 15 cm above the bottom, was aerated through the hole at a rate of 10 l kg−1 h−1 with compressed air. Expanded plastic material adiabatically covered the composting bin. The temperature was measured 20 cm below the compost surface. The water content was determined by the weight loss of a 20-g composting subsample after drying at 105 °C for 20 h and water was added to the composting material in the bin to adjust the water content to 40% (w/w). This water content was decided by the efficiency of mineralization (data not shown).

Water-dissolved component measurement

To each 0·5-g composting sample 1 ml ultra-pure water was added, mixed well and centrifuged at 15 000 g at 4 °C for 10 min. The supernatant fluids were filtered (MILLEX®-GP 0·22-µm filter unit; Millipore, Bedford, MA, USA). The filtrates were used for the analysis of dissolved components during the composting process. The total organic carbon (TOC) in the filtrates after diluting 1000× with ultra-pure water was measured by a TOC-2000 (Shimadzu, Kyoto, Japan). Several organic acids and amino acids in the filtrates were measured by capillary electrophoresis (CIA; Waters, Milford, MA, USA). Briefly, the filtrates were mixed with 5 mol l−1 standard organic and amino acids and diluted 10× with ultra-pure water. The electrophoresis was performed at 200 V in supersaturated sodium tetraborate buffer together with 1/10 volume of CIA-Pak™ OFM Anion BT (Waters). The absorbance at 252 nm was measured for detection. The standard organic and amino acids for measurement were oxalate, formate, fumarate, maleic acid, succinate, α-ketoglutarate, malate, citrate, tartarate, acetate, pyruvate, propionate, lactate, glutamate, butyrate, levurinate, benzoate and asparaginate. In order to measure the dissolved humic substance the absorbance at 440 nm of filtrates was measured. A standard curve was made from humic acid (Wako, Osaka, Japan) solution dissolved in 0·02 mol l−1 NaOH.

Direct bacterial count

Microbes in composting samples stained with 4′6-diamidino-2-phenylindole dihydrochloride (DAPI; Wako) were observed and counted with an epifluorescence microscope (BH2-RFCA; Olympus, Tokyo, Japan) as follows. Freeze-dried composting samples (50 mg) were suspended in 1 ml sodium phosphate buffer (50 mmol l−1; pH 7·6). The suspensions were mixed well and centrifuged at 2000 g for 5 min to collect the cells. The supernatant fluids were discarded and this washing procedure repeated three times. The washed cells were then suspended in 920 µl 50 mmol l−1 sodium phosphate buffer, pH 7·6, mixed with 30 µl DAPI and left for 5 min in the dark. After mixing again, 1 µl of the solution was filtered through a 0·2-µm pore size Nucleopore® filter (Corning, New York, NY, USA). The filter was washed with sterilized water and observed under the epifluorescence microscope using u.v. light. The number of bacteria was calculated by the average count of a random five fields on each filter.

Nucleic acid extraction

To lyse the microbial cells in the composting samples, 0·5 g glass beads (0·1 mm diameter), 0·4 ml 50 mmol l−1 sodium phosphate buffer (pH 6·8), 0·03 ml 20% sodium dodecyl sulphate, 0·05 ml 5 mol l−1 pyrophosphate, 0·6 ml TE (10 mmol l−1 Tris-HCl, pH 8·0, 10 mmol l−1 EDTA)-saturated phenol and 0·1 g freeze-dried composting samples were combined in a 2-ml plastic tube and shaken vigorously (2000 rev min−1) on a beadbeater (Mikrodismembrator U; B. Braun Biotech International, Melsungen, Germany) for 1 min. The tubes were centrifuged at 10 000 g for 5 min and the upper phase collected. The lower phases were extracted twice by the same procedure. Nucleic acids in the collected upper phase were then precipitated by adding 0·1 volumes of 3 mol l−1 sodium acetate (pH 5·3) and two volumes of 99·5% (v/v) ethanol and collected by centrifugation at 10 000 g for 5 min. The nucleic acids were then dissolved in 0·5 ml TE buffer and further purified by the polyethylene glycol (PEG) precipitation method (Selenska and Klingmüller 1991). The resultant solutions were quantified by absorbance at 260 nm.

Polymerase chain reaction

The PCR amplifications were performed in 50-µl volumes containing approximately 100 ng template DNA, 1× Ex Taq™ Buffer (Takara Shuzo, Shiga, Japan), 200 µmol l−1 dNTP, 25 pmol of each primer and 1·25 units Taq polymerase (TaKaRa Ex Taq™; Takara Shuzo). The PCR cycling was performed using a PC800 (Astec, Fukuoka, Japan). The temperature program was as follows: 25 cycles at 94 °C for 0·5 min, 45 °C for 1 min and 72 °C for 1·5 min with final extension steps at 72 °C for 15 min. The PCR products were quantified by absorbance at 260 nm.

Polymerase chain reaction primers

A eubacterial 16S rRNA-targeted primer pair (341F with GC-clamp, 907R) was used for PCR amplification in this study (Muyzer et al. 1997). The sequences were as follows: GC-clamp, 5′-CGCCCGCCGCGCCCCGCGCCCGTCCCGCCGCCCCCGCCCG-3′; 341F, 5′-CCTACGGGAGGCAGCAG-3′ and 907R, 5′-CCGTCAATTCCTTTRAGTTT-3′. The 341F without GC-clamp and 907R primers were used for sequencing reactions.

Denaturing gradient gel electrophoresis

The PCR products were analysed by DGGE according to Muyzer's protocol (Muyzer et al. 1997) followed by sequencing. Denaturing gradient gel electrophoresis was performed with a D-gene system (BioRad Laboratories, Hercules, CA, USA). Similarly sized PCR products were separated on a 1·5-mm thick vertical gel containing 8% (w/v) polyacrylamide (37·5 : 1 acrylamide : bisacrylamide) and a linear gradient of the denaturants urea and formamide, increasing from 0% at the top of the gel to 80% at the bottom. The 100% denaturant contained 7 mol l−1 urea and 40% (v/v) formamide. The PCR products (10 µl) were applied to individual lanes in the gel. Electrophoresis was performed in a buffer (diluted 100×) of ready-made 50× Tris/Acetic acid/EDTA Buffer (BioRad Laboratories) and 200 V was applied to the submerged gel for 4 h at 60 °C. After electrophoresis, the gels were stained in an aqueous ethidium bromide solution (0·5 mg l−1) and photographed on a u.v. (302 nm) transillumination table with a Polaroid camera (CE-600; Nihon Polaroid, Tokyo, Japan). The photographs were scanned and the image data downloaded into a computer. The computerized images were then inverted to negative images. Small pieces of selected DGGE bands were excised from the DGGE gel with Pasteur pipettes and DNA fragments in the gels were washed and directly reamplified with the same primer. The PCR products were confirmed by DGGE as a single band or not and, if isolated, then sequenced. Before sequencing, the PCR products were purified by a Qiaquick PCR purification kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions.

Sequencing and phylogenetic analysis

Sequencing reactions were carried out with an ABI PRISM™ BigDye™ Terminator Cycle Sequencing Ready Reaction Kit (Perkin Elmer, Foster City, CA, USA) according to the manufacturer's instructions. The products were then analysed by an automatic sequencer (model 377 A; Applied Biosystems, San Jose, CA, USA). Sequences were compared with the compilation of 16S rDNA genes available in the database (DDBJ, EMBL and GenBank) by MPsrch (Smith and Waterman 1981) through the DNA bank at the Ministry of Agriculture, Forestry and Fisheries of Japan Internet site. Sequences were then aligned with a CLUSTAL W Ver. 1.7 (Thompson et al. 1994) and distances determined by a neighbour-joining algorithm (Saitou and Nei 1987) with the same software. Phylogenetic trees were drawn with TREEVIEW (Page 1996).

Nucleotide sequence accession numbers

The sequences obtained in this study are available in DDBJ under accession numbers AB029407AB029419.

Results

Physical and chemical changes in a composting process

The temperature increased rapidly after mixing of the compost material, reached a maximum value (58 °C) within 9 d and then gradually declined to ambient level over about 30 d (Fig. 1a). In the early stage, pH increased from 5·3 to 8·3 and stabilized at the maximum after 20 d (Fig. 1a). The changes in these parameters were similar to those in a typical composting process (Gray et al. 1971). This suggested that the composting method in this study was as successful as that using other methods.

Figure 1.

Physical and chemical changes in the composting process. (a) ●, Temperature; □, pH. (b) ●, Acetate; □, lactate. (c) ●, Total dissolved organic carbon; □, organic acids; ▵, soluble humic substance; × , direct bacterial counts

The composting process can be divided into four phases: mesophilic, thermophilic, cooling down and maturing (Gray et al. 1971; Godden et al. 1983). These four phases were recognized in the temperature–time (and pH–time) curves in the present process as follows: mesophilic (0–4 d; S), thermophilic (4–13 d; T), cooling (13–32 d; C) and maturing (32–45 d; M) (Fig. 1a).

The total dissolved organic carbon (TDOC) decreased from 30 to around 15 mg-C g (wet weight)−1 in the first 4 d, rose once on day 9 and then decreased gradually (Fig. 1c). The initial concentrations of organic and amino acids (mg-C g (wet weight)−1) were as follows: formate, 0·1; malate, 0·6; citrate, 1·1; acetate, 1; lactate, 2·1; levurinate, 1·1; glutamate, 0·8 and asparaginate, 0·2. The above organic and amino acids (except for acetate) declined in the first 9 d and were exhausted after 9 d. In contrast, acetate increased in the first 9 d from 0·1 to 0·6 mg-C g (wet weight)−1 and was then exhausted after 13 d (Fig. 1b). The sum of the organic acids measured decreased from 5·6 to 0·3 mg-C g (wet weight)−1 for the first 4 d and they were exhausted after the C phase (Fig. 1c). The water-soluble humic substance began to increase in the C phase and maintained its maximum value of 11 mg-C g (wet weight)−1 until the end of the experiment (Fig. 1c). In contrast to TDOC, direct bacterial counts rapidly increased, although declining to half the value of 4 d after 9 d, then reaching their maximum and stabilizing at a constant order (1011 cell g (wet weight)−1) (Fig. 1c).

Microbial succession analysed by denaturing gradient gel electrophoresis

Since compost contains much humic substance (Gray et al. 1971), PCR amplification of nucleic acids from composting samples is difficult. After the dark-brownish colour of the extracted nucleic acids solution was removed by PEG precipitation, the PCR amplification was successful except for the sample on day 0. The PCR amplification of the sample on d 0 failed because it might contain scarce nucleic acid and various contaminants such as polysaccharides. Denaturing gradient gel electrophoresis analysis of these PCR products with a denaturing gradient ranging from 0 to 80% showed that all the bands were in the denaturing range from 20 to 50%. Thus, we used a DGGE gel range from 20 to 50%.

Although nucleic acid samples were subjected to PCR–DGGE more than three times, no difference in banding pattern was observed among the PCR products from the same sample. This showed reproducibility. The band pattern in DGGE gel showed drastic changes over the composting process and the higher the temperature became the fewer the number of bands (Fig. 2). In addition, the number of bands increased as composting progressed after the C phase. While the four phases were determined by temperature and pH changes, DGGE band patterns were different within the same phases (Fig. 2).

Figure 2.

Inverted image of the denaturing gradient gel electrophoresis gel stained by ethidium bromide. The number under each lane shows the sampling day. The band nomenclature follows the A–B : C pattern. A and B show the existing period from A days to B days. C is the serial number for the bands with the same existing periods

Although DNA fragments from major bands in the DGGE gel were successfully amplified, those from minor bands failed to be amplified or isolated. This may be because the major bands tailed and contaminated the minor bands. Thus, it was difficult to isolate the minor bands from DGGE gel, especially when many bands exist in a lane. The sequences of successfully amplified DNA fragments were determined to deduce the phylogenetic affiliation of microbes in the composting process. These sequence data showed the phylogenetic positions as seen in Fig. 3. Consequently, analysis of the rDNA libraries, along with the results from DGGE on the composting samples, indicated the presence of putative bacteria closely related to known genera. The similarities of these sequences to those of most related strains in DDBJ and existing days of DGGE bands are shown in Table 1.

Figure 3.

Neighbour-joining tree of partial 16S rRNA sequence (approx. 550 bp) recovered by denaturing gradient gel electophoresis bands. Accession numbers appear before the genus name. The neighbour-joining tree was constructed as described in the text. Each sequence, except for those from this study, was obtained from the DDBJ. The numbers on the branches refer to bootstrap values for 1000 times; only those above 500 are shown

Table 1.  Similarity of sequences from denaturing gradient gel electrophoresis bands to those of most related strains in database together with existing days of bands recognized
Band nameMost relative strain in DDBJ (accession no.)Match (%)Band intensities
Days
491320242832384345
  1. +++, Strong; ++, substantial; +, recognizable; no mark, not recognized.

4–4 : 2Staphylococcus piscifermentans (Y15753)99·1++         
4–4 : 1Leuconostoc paramesenteroides (S67831)97·4++         
4–9 : 1Pediococcus acidilactici (M58833)98·2+++        
4–13 : 1Bacillus badius (D78310)99·3+++++++       
4–9 : 2Bacillus sp. (AJ000648)99·1++        
13–13 : 1Virgibacillus proomii (AJ012667)94·7  ++       
13–13 : 2Corynebacterium urealyticum (X84439)95·3  ++       
13–20 : 1Gracilibacillus halotolerans (AF036922)94·4  +++      
20–45 : 1Clostridium filimentosum (X77847)85·6   +++++++
20–45 : 2Alcaligenes sp. NKNTAU (U82826)91·4   +++++++++++++
24–45 : 2Alloiococcus otitis (X59765)90    ++++++++
24–45 : 3Clostridium fervidus (L09187)88·2    +++++++
24–45 : 1Sphingobacterium multivorum (AB020205)83·4    +++++++++++
43–45 : 1Arthrobacter sp. (X93356)86·4        +++

In the S phase (0–4 d), the DNA sequences from DGGE bands 4–4 : 1, 4–9 : 1 and 4–4 : 2 were closely related to the fermenting bacteria Leuconostoc paramesenteroides, Pediococcus acidilactici and Staphylococcus piscifermentans, respectively (97·4, 98·2 and 99·1% similarity for each recovered sequence over 500 nucleotides), concurrently with high concentrations of organic acids, especially lactate (Table 1, Figs 1b and 3). The sequence of DNA 4–13 : 1 was closely related to Bacillus coagulans and B. badius (99·5 and 99·3%) that were closely related to each other (Table 1 and Fig. 3).

In the T phase (4–13 d), when the temperature showed a maximum on day 9, the main DGGE band was only 4–13 : 1 (B. coagulans or B. badius;Table 1 and Fig. 3). Other major bands (4–9 : 2, 13–13 : 1 and 13–20 : 1) in this phase were related to Bacillus sp., Virdibacillus proomii and Gracilibacillus halotolerans, respectively (99·1, 94·7 and 94·4%), so the organisms represented by these bands might belong to the Bacillaceae. The DNA sequence of band 13–13 : 2 was related to Corynebacterium urealyticum (95·3%) (Table 1 and Fig. 3).

In the C phase (13–32 d), DGGE bands in the T phase died out and new bands appeared. The new bands were named 24–45 : 1, 24–45 : 2, 24–45 : 3, 20–45 : 1 and 20–45 : 2, sequences of which were related to Sphingobacterium multivorum, Alloiococcus otitis, Clostridium fervidus, Cl. filimentosum and Alcaligenes sp. NKNTAU, respectively (83·4, 90·0, 88·2, 85·6 and 91·4%). While the similarity between band 24–45 : 1 and S. multivorum was relatively low, it might belong to the Cytophaga–Flavobacterium phylum. Moreover, similarities between bands 24–45 : 3 and Cl. fervidus and bands 20 and 45 : 1 and Cl. filimentosum were relatively low; these bands might belong to the Clostridiaceae and furthermore, the latter might belong to cluster XI or XII defined by Collins et al. (1994). The DGGE pattern and sequence data showed that the organisms in this phase were different from those in the S phase.

The M phase (32–45 d) was very stable in all respects, temperature, pH, direct counts, TDOC and soluble humic substance (Fig. 1). Relative to earlier phases, the changes in microbial communities reflected by the DGGE pattern were stable and complex (Fig. 2). Only one DGGE band, 43–45 : 1, the DNA sequence of which was related to Arthrobacter sp. (86·4%) appeared as late as 43 d.

Discussion

For DGGE analysis in this study, the 16S rRNA gene was targeted because it has large databases, the contents of which have been increasing, and it is suitable for inferring phylogenetic relationships (Woese 1987; Amann et al. 1995). There is not an accepted value of percentage identity at which two 16S rRNA genes are concluded to belong to the same genus or species. It can be quite different for different genera (Fox et al. 1992). In this study, sequence data obtained from the early phase of the composting process had high similarities to most relative sequences in the database although those from the later phase had relatively low similarities indicating the scarcity of sequence data related to microbes in the later phase so far (Table 1). This suggested that microbes which appeared in the early phase were often isolated and well investigated; on the other hand, those in the later phase were difficult to culture and/or were not well investigated. While bacteria existing after the C phase in the composting process might be difficult to isolate, the phylogenetic positions of those microbes are of use to select media for the isolation of composting micro-organisms.

In the DGGE pattern the number of bands decreased as the temperature rose. This suggests that temperature is a significant factor in determining the relative advantage of some population over another. This supported the contention that temperature is the dominant physicochemical parameter controlling microbial activity during composting (McKinley and Vestal 1984). The band patterns were different between the low temperature phases, mesophilic and maturing, suggesting that the environments in these phases were different. The difference between these environments was observed as being due to TDOC, pH and the concentration of organic acids and humic substance in this study (Fig. 1).

In the S phase, the DNA sequences from the DGGE bands fell within the cluster of Gram-positive fermenting bacteria, concurrent with high concentrations of organic acids (Table 1, Figs 1b and 3). This suggested that the microbes appearing in this phase might have fermenting ability enabling them to be isolated. This was supported by Golueke's suggestion that fermenting bacteria dominated at the beginning of the composting process (Golueke et al. 1954). Usually fermenting bacteria rapidly use easily degradable substrates and proliferate. This was observed in the results of a decrease in TDOC, accumulation of organic acids and an increase in bacterial cell numbers (Figs 1b and 1c).

In the T phase, fermenting bacteria disappeared and different microbes appeared as the pH and temperature increased. The microbes proliferating in this phase were related to bacillus. This was supported by previous studies showing that 87% of isolated thermophilic bacteria in composting were identified as Bacillus spp. (Strom 1985b). Interestingly, when the temperature showed a maximum on day 9, there was only one major DGGE band, named 4–13 : 1, whose sequence was related to that of B. coagulans or B. badius (Table 1 and Fig. 3). Bacillus coagulans isolated from compost by Strom had the ability to grow at 65 °C (Strom 1985b). Also, the acidic environment early in this phase is suitable for B. coagulans (closely related to the sequence 4–13 : 1) which requires a slightly low pH value (6·0) for the initiation of growth (Sneath et al. 1986). These factors support the dominance of B. coagulans in this phase. The increase in pH in this phase might be caused by ammonia production, as evidenced by the strong ammoniacal odour. This suggested that proteolysis occurred in this phase. This was supported by the report of Riffaldi et al. (1986) that proteolytic and ammonia-producing bacteria proliferated when the temperature was high, but were soon extinct. Substantial strains of B. coagulans have a proteolytic ability (Sneath et al. 1986). In summary, the fermentation of carbohydrate interchanged with proteolysis by thermophilic B. coagulans relatives in the T phase of this composting system.

In the C phase, the DGGE patterns and sequence data showed that the organisms in this phase were different from those in the S phase. The DGGE bands in the T phase faded out and new bands appeared. These additional bands were related to S. multivorum, A. otitis, Cl. fervidus, Cl. filimentosum and Alcaligenes sp. NKNTAU. The metabolic characters of these microbes could not be inferred from the phylogenetic positions because the phylogenetic distance between the sequence of the band and those of these organisms was, respectively, low (Fig. 3). They might, at least, degrade the remaining complex compounds since easily degradable compounds inducing a temperature increase were lost. The appearance of obligate anaerobes, such as clostridium, in this phase suggested that an anaerobic microenvironment developed within the aggregates of composting materials.

After the C phase, the DGGE band patterns were more stable and more complex than in earlier phases (Figs 2 and 3). Only one DGGE band related to Arthrobacter sp. was added later in the M phase. Most of the species in this genus were known soil bacteria. These results suggested that the environment of the later phases in the composting process was similar to an oligotrophic environment such as that of soil. Since information on bacterial taxa after a temperature drop was scarce in previous studies because of the difficulty of isolation, this provides important documentation on bacterial communities after the C phase in the composting process and could become a guide for isolation.

The advantage of PLFA analysis was its ability to quantify indicators of organisms as a biomass without cultivation. The results obtained by this method mainly showed that fungi and actinomycetes proliferated when the temperature was low and that thermophilic bacteria dominated when it was high (Hellmann et al. 1997; Herrmann and Shann 1997; Klamer and Bååth 1998). The microbial succession in this study partially disagreed with the reports using PLFA analysis with respect to the absence of fungi and the scarcity of actinomycetes detected after the C phase. Amplification of eukaryotic rRNA genes was attempted using eukaryotic specific primer sets and the PCR products were only confirmed from a 4-d sample (data not shown). There are three possible reasons for the absence of fungi and the scarcity of actinomycetes. Firstly, actinomycetes and fungi were generally suitable for drier conditions than bacteria (Scott 1957). The surface of the composting materials in this study was wet because the composting materials were supplied with water every day. Secondly, fungi are favoured by C : N ratios higher than those for bacteria and actinomycetes (Griffin 1985). Generally, garbage has lower C : N ratios than field waste and wastepaper. Thirdly, since the growth rates of fungi and actinomycetes were slower than those of bacteria, they were not suitable for a laboratory-scale composting system that progressed more rapidly than field-scale composting (Godden et al. 1983). In the T phase, acetate was abundant and the temperature was high. These environments are suitable for thermophilic methanogens and former researchers had detected ether lipids using PLFA analysis (Hellmann et al. 1997). We tried to amplify the archaeal rRNA gene using archaea-specific primer sets, but the PCR product was not recognized (data not shown).

The DGGE combining sequence analysis requires the following refinements for a more accurate estimation of microbial succession. Firstly, an excessively complex microbial population is not suitable for analysis because sequence analyses of the vast number of DGGE bands were too laborious and bands on the DGGE gel were often difficult to isolate. Compost is, however, a suitable subject for DGGE analysis of bacterial populations because of its relatively simple community structure as shown in Fig. 2. Secondly, weak bands on DGGE gel were difficult to isolate and reamplify, because they tended to be contaminated by strong bands. The only solution to this problem is to carefully excise the target band as precisely as possible. Furthermore, DGGE profiling cannot provide accurate quantitative information because of the PCR amplification bias of genes extracted from mixed microbial populations (Farrelly et al. 1995; Suzuki and Giovannoni 1996; Polz and Cavanaugh 1998). Some researchers, however, were able to find dominant microbes by using the series of diluted template DNA for PCR amplification (Ferris et al. 1997; Øvreås et al. 1997).

It is unknown whether there are differences in the microbial succession for different composting processes. However, composting has specific conditions, such as a temperature increase, humic substance production and the solid matrix of the substrate. Consequently these conditions must limit organisms to the specific microbes that generally appeared in the composting process. The results presented here will help to elucidate general composting processes. For example, these results are of use to select media for the isolation of composting micro-organisms. Further study is needed to clarify microbial nutritional capacities in the composting process by DGGE combined with the cultivation method.

Acknowledgements

The authors would like to thank Dr Kazuyoshi Suzuki and Ms Maki Itagaki for their advice and help.

Ancillary