Correspondence: Yahai Lu, College of Resources and Environmental Sciences, China Agricultural University, Beijing 100193, China. Tel./fax: +86 10 6273 3617; e-mail: email@example.com
Soil drainage is one of the most promising approaches to mitigate methane (CH4) emission from paddy fields. The microbial mechanism for the drainage effect on CH4 emission, however, remains poorly understood. In the present study, we determined the effect of short (four drainages of 5–6 days each) and long drainage cycles (two drainages of 10–11 days each) on CH4 emission and analyzed the response of the structure and abundance of methanogens and methanotrophs in a Chinese rice field soil at the DNA level. Rice biomass production was similar between drainage and the practice of continuous flooding. The rate of CH4 emission, however, was reduced by 59% and 85% for the long and short drainage cycles, respectively. Quantitative (real-time) PCR analysis revealed that the total abundance of archaeal populations decreased by 40% after multiple drainages, indicating the inhibitory effects on methanogen growth. The structure of the methanogen community as determined by terminal restriction fragment length polymorphism analysis, however, remained unaffected by drainages, although it varied among rhizosphere, bulk and surface soils. Quantitative PCR analysis of the methanotrophic functional pmoA genes revealed that the total abundance of methanotrophs in rhizosphere soil increased two to three times after soil drainages, indicating a stimulation of methanotroph growth. The CH4 oxidation potential in the rhizosphere soil also increased significantly. Furthermore, drainages caused a shift of the methanotrophic community, with a significantly increase of type II methanotrophic bacteria in the rhizosphere and surface soil. Thus, both inhibition of methanogens and stimulation of methanotrophs were partly responsible for the reduction of CH4 emissions. The methanotroph community, however, appeared to react more sensitively to soil drainage compared with the methanogen community.
Atmospheric methane (CH4) is the second most important greenhouse gas. Its concentration has doubled in the past two centuries. Flooded rice fields are a major source of atmospheric CH4, accounting for 5.3–19.2% of total global emissions (IPCC, 2007). Rice is the staple food for human populations particularly in Asia. The demand for rice production will increase by 65% in the next two decades to meet the needs of the growing population (Nguyen & Ferrero, 2006). A great challenge in rice agriculture, therefore, is to develop strategies to reduce CH4 emission while increasing rice yield (Neue, 1997). Among many agricultural practices such as fertilization and cultivar selection, water management has been recognized as one of most promising approaches for this purpose (Li et al., 2002; Majumdar, 2003).
Several field measurements in the past two decades have shown that intermittent drainages could substantially reduce CH4 emissions from rice fields (Sass et al., 1992; Adhya et al., 1994; Nugroho et al., 1994; Husin et al., 1995; Yagi et al., 1996, 1997; Sigren et al., 1997; Lu et al., 2000a; Wassmann et al., 2000). An empirical model, solely based on statistical relationships between CH4 emissions and environmental parameters, predicted that intermittent drainage would reduce CH4 emissions by 40–48% in Asia (Yan et al., 2005). This estimate is in good agreement with the prediction by a more comprehensive biogeochemical model, DNDC, which showed that adoption of mid-season drainage in China rice fields would reduce CH4 emission by 40–56% as compared with the practice of continuous flooding (Li et al., 2002, 2006). Recently, the DNDC model was revised by emphasizing the importance of electron donor and acceptor dynamics in CH4 production in soil (Fumoto et al., 2008). The revised DNDC revealed that drainage practice would reduce CH4 emissions by 17–82% in Japanese rice soils (Fumoto et al., 2010). The process-based modeling relies on the understanding of key processes involving CH4 production, oxidation and transport from soil to the atmosphere. However, the microbial mechanisms of CH4 production and oxidation and their response to drainages are poorly understood.
Methanogens are considered strictly anaerobic organisms and are less competitive thermodynamically with electron donors than the bacterial reducers of nitrate, Fe(III) and sulfate in soil. Drainage of anoxic soil results in regeneration of Fe(III) and sulfate from their reduced forms by chemical and biological oxidation (Ratering & Conrad, 1998). The Fe(III)- and sulfate-reducing bacteria are then activated upon reflooding of soil, outcompeting methanogens for the common substrates H2 and acetate (Ratering & Conrad, 1998; Kruger et al., 2001; Yuan et al., 2009). This substrate competition mechanism has been adopted in the revised DNDC model, although only the role of Fe(III) is considered (Fumoto et al., 2008).
The effect of drainage on methanogens, however, appears more complex. It has been shown that acetate-dependent CH4 production was more strongly suppressed after soil drainage than H2-dependent methanogenesis (Kruger et al., 2001, 2002), suggesting not only that total CH4 production was influenced but also the pathway of methanogenesis changed. It is unclear whether such a complex effect of drainage is related to structure or just activity of methanogens catabolizing these processes.
Methanotrophs inhabiting rhizosphere and surface soil are able to oxidize up to 90% of total CH4 produced in anoxic soil (Denier van der Gon & Neue, 1996; Bosse & Frenzel, 1997, 1998) and hence serve as a biobarrier of CH4 emission into the atmosphere. The activity of methanotrophs relies on the availability of O2, CH4 and nitrogen in soil. It was revealed that CH4 oxidation under field conditions decreased with plant growth and this was attributed to nitrogen limitation in the later stages of plant growth (Kruger et al., 2001, 2002). Drainage will dramatically alter the physicochemical conditions of soil. However, very little is known about the effect of drainage on CH4-oxidizing bacteria. In a preliminary study using rice soil without plants, it was found that drainage expanded the zone of atmospheric CH4 oxidation to a deeper soil layer and had a strong effect on type I methanotrophs (Henckel et al., 2001).
To understand the microbial mechanism for the drainage effect on CH4 emission, we performed a greenhouse microcosm experiment in which three water regimes were established: continuous flooding and intermittent drainage for short cycles (5–6-day drainage) and longer cycles (10–11-day drainage), respectively. The objectives were: (1) to measure and compare CH4 dynamics among three water regimes; (2) to determine the composition and abundance of both methanogenic archaea and methanotrophic bacteria and link them to the dynamics of CH4 emission; and (3) to compare the different effects on the composition and abundance of both methanogenic archaea and methanotrophic bacteria between short and longer drainage cycles, as practiced in Asian rice fields (Yagi et al., 1997; Lu et al., 2000a).
Materials and methods
Soil from the plow layer (0–20 cm) was collected from a rice field in autumn 2007 at the China National Rice Research Institute in Hangzhou (Lu et al., 2000a). Soil was dried, crushed, sieved (2-mm mesh-size), mixed and stored at room temperature. The soil was a clay loam; its characteristics have been described previously (Wu et al., 2009). An Indica rice cultivar (Oryza sativa, Jinzao 47) was used, which shows a high yield potential despite a short growth stage (105 days). Twenty-four cylindrical plastic pots (height 19 cm, diameter 14 cm) were filled with 1.8 kg of soil and preflooded for 7 days. Each pot was then fertilized with urea and K2HPO4 at rates of 30 mg N kg−1 soil, 12.5 mg K kg−1 soil and 5.0 mg P kg−1 soil, respectively. A 16-day-old rice seedling was transplanted in each pot. All pots were flooded for 23 days after transplanting (DAT). Thereafter, eight pots were continuously flooded, maintaining a water depth of 3–5 cm above the soil surface until the end of the experiment (treatment CF). Another eight pots were flooded for 12 days followed by drainage for 11 days, reflooding for 12 days, and then a second drainage for 10 days (treatment LD, long drainage cycle). This scheme of intermittent drainage is popular in the local area (Lu et al., 2000a). The treatment of the last eight pots consisted of four drainage cycles of 5–6 days followed by the same period of reflooding (treatment SD, short drainage cycle). This water management regime is less popular but considered more efficient in mitigating CH4 emission from paddy field (Lu et al., 2000a).
Measurements of CH4 flux and dissolved CH4
CH4 flux was determined on 23, 29, 35, 39, 48, 53, 58, 63 and 69 DAT. At the date of measurement, all pots were placed in a big container filled with water. Pots were covered with bottom-opened plastic chambers (20 cm of diameter; 100 cm of height) with a fan installed at the top of each chamber to mix the gas. The water in the big container was used to close the system. Gas samples were withdrawn using a syringe every 30 min from 10:00 am to 12:00 am. CH4 concentration was analyzed by GC (Shanghai Precision and Scientific Instrument, Shanghai, China). Flux rates of CH4 were calculated from the linear increase of CH4 mixing ratio and expressed in mmol CH4 day−1 m−2.
Soil porewater was sampled on 23, 30, 36, 40, 49, 54, 59, 63 and 69 DAT (sampling was skipped when soil was under drainage). One Rhizon sampler (Eijkelkamp, Giesbeek, the Netherlands) was vertically inserted into each pot at the depth of 1–11 cm under the soil surface. A vacuumed sampling tube was attached to the Rhizon sampler. After about 3 mL of liquid sample was sucked in, the sampling tube was detached from the sampler. The concentration of CH4 in porewater was analyzed immediately after the procedure, as described previously (Lu et al., 2000b).
Soils were sampled twice at the maximum tillering stage (47 DAT) and the early grain filling stage (69 DAT). Before destructive sampling, the surface soils (0–0.5 cm) were carefully removed using a sterile spoon. Then the whole soil core with roots was removed from the pot. The soils which could be detached easily from rice roots were considered the bulk soil. The roots with tightly attached soils were washed thoroughly with sterile deionized water and the suspensions were centrifuged at 2000 g for 5 min. The precipitated soils were taken as the rhizosphere soil. All soil samples were frozen and stored at −20 °C. The rice shoot biomass was determined by oven drying at 75 °C for 3 days.
Measurement of CH4 oxidation potential
To measure CH4 oxidation potential, about 5.0 g fresh rhizosphere soil was mixed with 5 mL sterile demineralized water and placed in a sterile glass bottle (125 mL). The headspace was supplemented with about 50 000 p.p.m.v. CH4. Bottles were incubated for 100 h at 25 °C in the dark. Depletion of CH4 in the headspace was followed by GC–flame ionization detector analysis after thoroughly shaking the bottles. The potential CH4 oxidation rates were calculated from the slopes of regression lines of CH4 depletion and expressed in μmol CH4 day−1 g−1 dw of soil.
DNA extraction, cloning, sequencing and phylogenetic analysis
For genomic DNA extraction, the frozen soil was thawed and about 0.5 g soil (wet weight) was placed in a 2-mL bead-beating tube containing 0.7 g of 0.1-mm beads. After adding 660 μL phosphate buffer (pH 8.0) and 220 μL TNS buffer (0.5 M Tris-HCl, pH 8.0, 0.1 M NaCl, 10% SDS w/v), the tube was shaken for 45 s at 6.5 m s−1 in a bead beater followed by centrifugation at 20 000 g for 4 min (Lu et al., 2005). Further procedures of DNA purification and precipitation were carried out with a Wizard DNA clean-up system (Promega, Madison, WI). Finally, DNA pellet was suspended in 0.2 mL elution buffer (10 mM Tris-HCl, pH 8.5) and stored at −20 °C.
PCR amplification was carried out using archaeal primer set Ar109f/Ar915r for 16S rRNA genes (Lueders et al., 2004) and methanotrophic proteobacterial primer set A189/mb661 for pmoA genes (Horz et al., 2001, 2005). The 50-μL reaction mixture contained 1 μL of DNA template (10 : 1 dilution of original extracts), 5 μL of 10 × buffer, 3 μL of 25 mM MgCl2, 1 μL of a 10-mM concentration of the dNTPs, 0.5 μL of each primer (50 μM) and 2.5 U of Taq DNA polymerase (TaKaRa, Dalian, Liaoning, China). One clone library each for archaeal 16S rRNA genes and methanotroph pmoA gene was constructed from the rhizosphere soil of SD treatment at the maximum tillering stage (47 DAT) using pMD19-T vector (TakaRa) according to the manufacturer's instructions. The randomly selected clones were sequenced with ABI 3730xl sequencer using BigDye Terminator cycle sequencing chemistry, with 42 clones for archaeal 16S rRNA gene and 48 clones for pmoA gene (Applied Biosystems, Foster City, CA) (Peng et al., 2008).
The sequences were aligned using the clustal x program (Thompson et al., 1997) and some related culture and environmental sequences from GenBank database were included for comparison. The phylogenetic trees were constructed using a neighbor-joining program following the protocol described previously (Peng et al., 2008; Ma et al., 2010). The sequences obtained in the present study have been deposited in GenBank database under the accession numbers: GU134452–GU134493 for archaeal 16S rRNA genes and GU134404–GU134451 for pmoA.
Terminal restriction fragment length polymorphism (T-RFLP) analysis
T-RFLP analyses were used to determine the diversity of archaeal and methanotrophic communities. PCR amplification used the same primers as indicated above except that the reverse primer for archaeal 16S rRNA genes and the forward primer for pmoA genes were labeled with 6-carboxyfluorescein (Horz et al., 2001, 2005; Lueders et al., 2004). The digestion of PCR products, and purification and size separation of the digested products followed the protocols described previously (Ma et al., 2010). T-RFLP patterns were analyzed with genemapper 4 software (Applied Biosystems) by peak height integration of different terminal restriction fragments (T-RFs). The percent fluorescence intensity represented by a single T-RF was calculated relative to the total fluorescence intensity of all T-RFs.
Quantitative (real-time) PCR analysis
Quantitative PCR of archaeal 16S rRNA genes and pmoA genes was done in a 7500 real-time PCR system (Applied Biosystems) using the primer pair Ar364/Ar934 for archaea (Kemnitz et al., 2005) and A189/mb661 for methanotrophs (Kolb et al., 2003). DNA was extracted from soil samples (500 mg) using Fast DNA Spin Kit for Soil (BIO 101; Q-Biogene, Heidelberg, Germany) according to the manufacturer's protocol. Quantitative PCR was performed in a total volume of 25 μL containing 12.5 μL of Power SYBR®Green PCR Master mix kit (Applied Biosystems), 0.66 μM archaea primers or 0.66 μM pmoA primers, 200 ng bovine serum albumin and 2 μL of a 25 : 1 dilution of DNA sample. The thermal cycles and fluorescence signal acquisition followed the protocols described in Kemnitz et al. (2005) and Kolb et al. (2003), respectively. The DNA standards were prepared from the cloned sequence of archaeal 16S rRNA gene and pmoA gene, which were further amplified with vector-specific primers. The PCR products were purified with a UNIQ-10 column kit (Sangon Biotech, Shanghai, China). The obtained PCR products were quantified with the PicoGreen dsDNA quantification kit (Invitrogen, Eugene, OR) and then converted into the copy number of DNA molecules per unit volume ranging from 1.7 × 101 to 1.7 × 108 copies μL−1. Three replicates of each measurement were done.
The ordination analysis of T-RFLP patterns of archaeal 16S rRNA gene and pmoA gene was done using canoco for Windows 4.5 software (Microcomputer Power, Ithaca, NY). Linear model approaches, i.e. the principle component analysis and redundancy analysis (RDA), were used for the analysis. The significance of water regime and soil compartment effect on the structure of methanogenic and methanotrophic communities was analyzed using partial RDA with the permutation test (1499 replicate runs) as implemented in canoco. In addition, anova was also performed with the sas program (SAS Institute, Cary, NC), and one-way anova was used to test the significance among water regimes.
Plant growth and CH4 dynamics
The dry weight of rice shoots increased from 10 g per plant at 47 DAT (the maximum tillering stage) to 30 g per plant at 69 DAT (the early grain filling stage) (Fig. 1). The production of shoot biomass was similar to the plants under field conditions, indicating no significant restriction of rice growth in the microcosms used. Short and long drainage cycles did not influence shoot biomass production.
CH4 emission rate ranged from 4.4 to 36.2 mmol CH4 day−1 m−2 within the experimental period (Fig. 2a). The emission rate in the CF treatment increased and reached the maximum of 36.2 mmol CH4 day−1 m−2 at 39 DAT. Thereafter, the flux leveled off between 27.3 and 35.7 mmol CH4 day−1 m−2 until the end of the experiment. The emission rate in the LD treatment was slightly lower but not statistically different from CF treatment during the early period. However, the emission rate in the LD treatment significantly decreased in the later stages (after 55 DAT). The SD treatment resulted in a more pronounced reduction in CH4 emission compared with the LD treatment (Fig. 2a).
Dissolved CH4 in soil porewater in the CF treatment maintained a relative stable concentration over the experimental period, averaging 80.1±16.1 μmol L−1 (Fig. 2b). Drainage significantly decreased the dissolved CH4 concentration at 36 and 49 DAT. However, the concentration increased after reflooding of soil, and finally reached the level of permanently flooded soil. This recovery appeared faster for the SD (at a rate of 5.4–17.3 μmol day−1 L−1) than LD treatment (at a rate of c. 2.0 μmol CH4 day−1 L−1).
CH4 oxidation potential was measured only at 47 and 69 DAT for the rhizosphere soil. It ranged from 8.5 to 16.4 μmol CH4 day−1 g−1 dw for all samples (Fig. 3). Drainage of soil significantly increased CH4 oxidation potential at both sampling times. There was no difference between the two drainage treatments.
Dynamics of archaeal community
The phylogenetic analysis of archaeal 16S rRNA genes showed that the archaeal community consisted of Methanocellales, Methanosarcinaceae, Methanosaetaceae, Methanomicrobiaceae, Methanobacteriaceae and the uncultured Euryarchaeota RC-III, RC-V, LDS cluster and crenarchaeotal group 1.1b and group 1.3 (Supporting Information, Fig. S1). The hydrogenotrophic Methanocellales (Rice Cluster-I) (Sakai et al., 2008) were predominant, accounting for 47.6% of total clones sequenced (Table 1), followed by Methanosaetaceae (7.1%), Methanobacteriaceae (7.1%), Methanosarcinaceae (4.8%) and Methanomicrobiaceae (2.4%). A similar composition of the archaeal community in this soil has been reported previously (Peng et al., 2008; Wu et al., 2009).
Table 1. Assignment of T-RFs and analysis of the clone library of archaeal 16S rRNA genes
No. of clones
Related T-RFs (bp)
Crenarchaeotal group 1.3
Crenarchaeotal group 1.1b
Nine major T-RFs were identified in the T-RFLP profiles of archaeal 16S rRNA genes. These T-RFs could be assigned to different archaeal lineages according to the sequence information in this and the previous study (Peng et al., 2008): 84 bp to Methanomicrobiaceae; 92 bp to Methanobacteriaceae; 284 bp to Methanosaetaceae; 393 bp to Methanocellales; 186 bp to mainly Methanosarcinaceae but occasionally also crenarchaeotal RC-group 1.1b; 381 bp to RC-III; 77 and 771 bp to RC-V; and 738 bp to crenarchaeotal RC-group 1.3 (Table 1).
T-RFLP profiles displayed only minor differences between rhizosphere and bulk soil but distinct patterns from surface soil (Fig. 4a). The relative abundances of 77-, 738- and 771-bp T-RFs characteristic for uncultured archaeota increased in the surface soil (Fig. 4a). By comparison, T-RFLP profiles in the rhizosphere and bulk soil were dominated by T-RFs of 92 and 393 bp, which together accounted for over 55% of the total fluorescence frequency. This enrichment of hydrogenotrophic methanogen T-RFs was more significant in rhizosphere than in bulk soil. The spatial differentiation of methanogenic community is clearly displayed in the two-dimensional ordination plot (Fig. 4b).
However, despite a clear spatial difference there was no significant difference in T-RFLP profiles between the first (47 DAT) and second stage (69 DAT) or among the three water treatments.
Dynamics of methanotrophic community
The phylogenetic analysis of methanotrophic pmoA gene fragments showed that the methanotroph community in the soil consisted of the type I genera Methylococcus, Methylocaldum, Methylomicrobium and Methylobacter and type II genera Methylocystis and Methylosinus (Fig. S2, Table 2). Two clones (pSD26 and pSD34) were clustered with pmoA2 gene fragments, which were identified as novel pmoA genes encoding particulate methane monooxygenase enzymes with high substrate affinity (Yimga et al., 2003; Baani & Liesack, 2008). Several sequences (pSD10, 15, 19, 22, 30 and 44) did not have closest matches in the NCBI database (Fig. S2).
Table 2. Assignment of T-RFs and analysis of the clone library of pmoA genes
No. of clones
Related T-RFs (bp)
Seven T-RFs were detected in the T-RFLP profiles of pmoA genes (Fig. 5a). In silico analysis of sequence data showed that 244 bp could be assigned to type II methanotrophs Methylocystis and Methylosinus, while 79, 437 and 505 bp were related to type I methanotrophs Methylococcus/Methylocaldum, Methylomicrobium and Methylobacter, respectively. T-RF of 226 bp represented a novel restriction fragment (Horz et al., 2001), which probably indicated environmental sequences distantly related to the Methylococcus/Methylocaldum cluster (Fig. S2). T-RFs of 208 and 353 bp had no sequence matches in the clone library.
T-RFLP profiles showed obvious differences among three soil compartments (Fig. 5a). In bulk soils, T-RFLP profiles displayed a monotonous pattern, exclusively dominated by 244-bp T-RF (accounting for over 83% of total fluorescence intensity). T-RFLP patterns in the rhizosphere and surface soil were more diverse. In particular, the number and abundance of T-RFs characteristic for type I methanotrophs significantly increased in the rhizosphere and surface soil compared with bulk soil. T-RFLP patterns also differed slightly between the rhizosphere and surface soil. The 79-bp T-RF was more abundant in the surface soil, whereas the 226-bp T-RF was mainly detected in the rhizosphere soil.
The sampling time had minor effects on T-RFLP profiles. For example, the 226-bp T-RF in the rhizosphere soil was more abundant at 47 DAT than at 69 DAT (P<0.001), whereas the 437-bp T-RF in the surface soil showed an obvious increase at 69 DAT compared with 47 DAT (P<0.001). However, most samples remained unaffected by the sampling time.
The drainage treatments showed no effect on T-RFLP patterns in the bulk soil (Fig. 5a, A). However, in the rhizosphere soil at the second stage, drainage significantly increased the 244-bp T-RF compared with the CF treatment (P=0.0028). The most significant influence of drainage was found in the surface soil (Fig. 5a, C), where the 244-bp T-RF markedly increased (P<0.001) while 79-bp T-RF decreased (P<0.001) after soil drainage. However, there was no significant difference between the LD and SD treatments (Fig. 5a, B). In agreement, the correspondence analysis showed that both soil compartments (trace=0.485; P<0.001) and drainage treatments (trace=0.122; P<0.001) influenced T-RFLP profiles of pmoA genes (Fig. 5b), while sampling time showed no significant effect.
Quantification of archaeal 16S rRNA genes and bacterial pmoA genes
Rhizosphere soil samples were used for quantitative (real-time) PCR analysis of archaeal 16S rRNA genes and bacterial pmoA genes. The copy numbers of archaeal populations ranged from 4.2 × 108 to 2.4 × 109 copies g−1 soil in the samples (Fig. 6a). The copy number increased at the second stage relative to the first (P<0.001). At the second stage the copy numbers in drainage treatments decreased by 40% compared with the CF treatment (P=0.001). But at the first stage the copy numbers showed a slight increase in drainage treatments (P=0.017). The pmoA gene copy number ranged from 2.2 × 107 to 3.0 × 108 copies g−1 soil (Fig. 6b). The number was greater at the second stage than first stage (P<0.001). The drainages resulted in an increase of two to three times the pmoA gene copy number compared with the CF treatment.
Our study showed that CH4 emission was substantially reduced while rice growth was unaffected by drainage. Estimates by integration of seasonal emissions indicate that long and short drainage cycles reduce CH4 emissions by 59% and 85%, respectively. This estimate is in a good agreement with the previous long-term field experiment in the same rice soil (from 1995 to 1998), showing that mid-season drainage (a single long drainage) and intermittent drainages (multiple short drainages) reduced CH4 emission by 44% and 74%, respectively, as compared with continuous flooding practice (Lu et al., 2000a). The concentration of dissolved CH4 in soil porewater markedly decreased after soil drainage but increased again after reflooding, indicating the recession and recovery of methanogenesis along with the water regime management. The CH4 oxidation potential, on the other hand, was promoted by the intermittent drainages. Therefore, the decrease of CH4 emission was the integrated result of the soil drainage effect on both methanogenesis and CH4 oxidation. By determining the structure and abundance of archaeal 16S rRNA genes and methanotrophic pmoA genes, we further showed that drainages caused complex reactions in methanogenic archaea and methanotrophic bacteria, both contributing to the decrease of CH4 emission.
Fingerprinting of archaeal populations by T-RFLP analyses revealed that the community structure of methanogens differed among different soil compartments. The surface soil displayed a higher complexity of archaeal community, probably due to its dynamic environment preventing the establishment of a stable community. The relative abundance of hydrogenotrophic methanogens increased in the bulk soil, and this was more significant in the rhizosphere soil. The enrichment of hydrogenotrophic methanogens in rhizosphere has been detected previously (Conrad et al., 2006; Wu et al., 2009). Labeling the rice plant with 13CO2, in situ RNA-SIP demonstrated that hydrogenotrophic methanogen Methanocellales was responsible for methanogenic activity using the plant-derived carbon (Lu & Conrad, 2005). The mechanism is currently unclear. However, the enrichment of hydrogenotrophic methanogens in the root environment agrees with the finding that CH4 production in the washed root material is primarily via a hydrogenotrophic pathway (Conrad & Klose, 1999; Chin et al., 2004).
Despite the spatial difference, methanogen communities remained stable in different compartments and were unaffected by drainages. Some studies under field conditions (Kruger et al., 2005; Watanabe et al., 2006, 2007) also indicated that the structure of methanogens did not change over the growing season of rice or even throughout the dry fallow period. The reason for the constant methanogen structure remains unknown. However, it should be noted that the stable structure of methanogen community may not necessarily reflect the metabolically active members (Watanabe et al., 2009).
Despite the stable structure of the archaeal community, we found their abundance changed between two stages and among water regimes. The copy number of archaeal 16S rRNA genes increased with time and this increase was particularly pronounced in the continuous flooding soil, being about five times greater at the second stage than at the first stage. The increase of methanogens is very possibly due to the increase of root exudates with plant growth, which supply substrates for the growth and activity of methanogens (Lu et al., 2000b; Kruger et al., 2005). The increase of methanogen abundance was consistent with the increase of CH4 emissions. At the second stage, when methanogenic activity was most active, a marked decrease of archaeal abundance was observed in drainage treatments compared with the permanently flooded soil. Apparently, this decrease was responsible for the reduction of CH4 emissions from the drainage treatments (Fig. 2a). However, such an effect did not occur in the early stage, probably due to the relatively poor drainage and hence insufficient aeration of soil (soil water content of 80% in the early drainages compared with <70% in the later drainages).
The methanotrophic community displayed a significant shift among soil compartments. In the bulk soil, type II methanotrophs were exclusively dominant, which is consistent with the observation in previous studies (Horz et al., 2001; Macalady et al., 2002; Eller et al., 2005). By comparison, a higher diversity of methanotrophs was detected in the surface and rhizosphere soils. In particular, type I methanotrophs (79, 226 and 437 bp) were significantly enriched. As the rhizosphere and surface soils are the major sites of CH4 oxidation, the enrichment of type I groups is an indication of their functional importance in rice soil (Bodelier et al., 2000; Qiu et al., 2008; Shrestha et al., 2008).
Soil drainage significantly increased the abundance of methanotrophs in the rhizosphere soil at both early and late stages. This enhancement was corroborated by the increase of CH4 oxidation potential. Apparently, the enhancement of methanotrophic activity was partly responsible for the decrease of total CH4 emission. In addition to an increase of methanotroph abundance, drainages also caused a visible effect on methanotrophic composition. However, the effect was detected only in the surface and rhizosphere soil. Surprisingly, it was type II methanotrophs that were significantly stimulated in the rhizosphere and surface soil by soil drainage. This was in contrast to a previous study in Italian rice soil where type I was found enhanced after 8-day drainage (Henckel et al., 2001). However, that experiment was conducted in microcosm without rice plants. In addition, soil properties differed significantly between Chinese soil (clay loam) and Italian soil (sandy loam). All these differences may influence the observed effect of drainage on methanotrophs (Jackel et al., 2001). Nevertheless, the increase of type II methanotrophs in the present study implies that drainage probably activated these organisms. As the total abundance of methanotrophs increased after soil drainage, we assumed that both type I and type II methanotrophs were stimulated, but that a more pronounced effect occurred for type II groups in our soil.
In summary, the present study demonstrated that intermittent drainage not only substantially reduced CH4 emissions but also had complex effects on methanogenic and methanotrophic communities. On the one hand, the growth of methanogens was suppressed, although the structure of methanogenic communities in different soil compartments remained unaffected by drainages. On the other hand, the total abundance of methanotrophs substantially increased and the structure of methanotroph community in the rhizosphere and surface soil where CH4 oxidation was most active also altered after soil drainages. Both the decrease of methanogen populations and the increase of methanotrophs could be partly responsible for the reduction of CH4 emissions. We assume that different responses of both methanogens and methanotrophs should be taken into account if CH4 emissions from rice fields are to be better understood and predicted under various water management scenarios. However, it should be noted that the DNA level analyses in the present study may not be sufficiently sensitive to fully detect the responses of methanogenic and methanotrophic members to intermittent drainages. Further studies using more sensitive methods such as rRNA and mRNA analyses would shape a better understanding of the microbial dynamics in rice field soils.
This study was partially supported by the Natural Science Foundation of China (grant no: 40830534; 40625003) and the Chang Jiang Scholars Program of Chinese Ministry of Education.