Introduction

Rice paddies are an important wet-soil ecosystem type worldwide, but these also emit greenhouse gases (GHG). At the same time, rice is the most important food in the diet of people in many Asian countries including China and Korea. Since the populations of these countries have grown rapidly, the demands for rice have also increased. According to FAO (2012), land used for rice paddy agriculture was about 0.6 × 109 hectares (ha) in 2002 and has steadily increased since then, reaching 0.73 × 109 hectares (ha) in 2011. Because these agricultural lands are usually irrigated and treated with nitrogen fertilizer, they play an important role in GHG emissions, especially methane (CH4) and nitrous oxide (N2O). Agricultural activities are the largest anthropogenic sources of CH4 and N2O, with the latter being a major source of atmospheric N2O (Mosier et al. 1998; Mikaloff Fletcher et al. 2004; IPCC 2007a, b).

Owing to the importance of agricultural areas in GHG emissions, many studies of CH4 or N2O fluxes in agricultural lands have been conducted (Cai et al. 1997; Jiang et al. 2006; Johnson-Beebout et al. 2009). Most of the studies have focused on the estimation of GHG fluxes, and a few have examined management methods for mitigation of GHG emissions (Lindau et al. 1993; Wassmann et al. 1993; Yagi et al. 1996; Boeckx et al. 2005). However, information about microorganisms that affect gas production and removal processes and their ecology in rice paddy soils is lacking. In addition, these studies could give the information about the potential of land use change to influence global carbon and nitrogen cycles.

Under strictly anaerobic conditions, acetate or CO2 is used as an electron acceptor to produce CH4 by specific Archaea named methanogens (Balch et al. 1979; Zinder 1993). CH4 can be oxidized both chemically and microbiologically under aerobic conditions, and this process is mediated by methanotrophic bacteria. Some studies have described the isolation and cultivation of methanogens from rice paddy soils; however, they did not investigate the ecology of methanogens per se (Großkopf et al. 1998; Sakai et al. 2007).

In anaerobic conditions, not only CH4 but also N2O is produced as an intermediate or final product when nitrate (NO3 −) is reduced by denitrifying bacteria. In addition, N2O can also be produced by nitrification in aerobic conditions (Schlesinger 1997). Denitrification is a reductive process in which NO3 − is used as an electron acceptor in anaerobic conditions, while organic carbon is used as an electron donor. Denitrification usually dominates N2O emission in wetlands, and the denitrifier community is controlled by several factors such as moisture, oxygen, pH, and nitrogen and carbon availability (Pihlatie et al. 2004; Wallenstein et al. 2006). While denitrification consists of several reduction steps (chain reaction) by respective reducing enzymes and functional genes (Fig. 1), in this study, we focused on the abundance of specific genes which expressed nitrite reductase (nirK, nirS) and nitrous oxide reductase (nosZ). Nitrite reductase controls the limiting step of denitrification, which produces the first gaseous intermediate. While either N2O or N2 can be produced as the final product of denitrification, nitrous oxide reductase reduces N2O to N2 (Wallenstein et al. 2006). Since the major N2O source is the soil ecosystem, it is important to understand microbial processes of N2O production and spatial distribution to control N2O emission. Many studies have investigated the denitrifier community composition and abundance together with N2O emission (Ma et al. 2008). However, few reports have described the vertical distribution of denitrifying bacteria (Yeomans et al. 1992).

Fig. 1
figure 1

Denitrification pathway, oxidation state and phase of nitrogen, and functional enzymes at each step of denitrification: Nar, nitrate reductase; Nir, nitrite reductase; Nor, nitric oxide reductase; Nos, nitrous oxide reductase (modified from Ye et al. (1994); Wallenstein et al. (2006))

Full size image

Many studies of GHG emission in rice paddy soils have been conducted because of their contribution to global budgets of CH4 and N2O; however, data from Asian countries are still limited. In addition, previous studies were primarily focused on GHG fluxes and their controlling factors and did not consider the microorganisms involved in the processes. In the current study, we examined the (1) abundance of microorganisms involved in N2O emission, CH4 production, and CH4 oxidation in rice paddy ecosystem in various seasons and soil depths, and (2) controlling environmental factors for these microorganisms in an Asian country (Korea).

Materials and Methods

Site Description

The study site included rice paddies near Haean-myeon in South Korea, which is located in Kangwon Province, near the demilitarized zone (38°17′05″N, 128°08′34″E). Haean-myeon is in a valley surrounded by Gachil-mountaintop, Daewoo Mountain, Dosol Mountain and Daeam Mountain (1,242, 1,179, 1,148 and 1,304 m, respectively). The height of the lowest valley in this region is about 450 m, and the heights of the surrounding mountains are from 1,000 to 1,300 m. In addition, these mountains have a well-developed piedmont. Total area of this watershed is 44.7 km2, and land use is classified as three types. Rice paddies are found at elevations of less than 500 m and there are no natural wetlands surrounding rice paddies. Radish, potato, and other legumes are cultivated in dry fields at 500–750 m elevation. Temperate woody forests are present above 750 m elevation. The region is characterized by a temperate monsoon climate with a rainy season from late June to August (Berger et al. 2013).

Soil Sampling and Analysis

We collected seasonal soil samples in a rice paddy (38˚17′05″N, 128˚08′34″E) during four seasons in May, August, November 2010 and February 2011; spring, summer, fall and winter, respectively. Soil sampling was performed using three replicates of soil cores at randomized places in the rice paddy, at depths of 0–70 cm. We divided soil cores into 10-cm sections and sampled in spring, summer, and autumn. In winter, three replicates of upper-layer soils was collected only (<10 cm) because the lower layer soils were frozen.

Fresh soils were used for all chemical analysis. Soil water content (WC) was measured by oven-drying at 103 °C for 24 h (Smith 2000), and organic matter (OM) content was determined by loss-on-ignition. Soil pH was determined using a pH meter after the soil was mixed with distilled water and the particles were allowed to settle (soil:water = 1:2). Extractable dissolved organic carbon (DOC) content was measured by extracting soil with distilled water, filtering through a 0.45-μm filter, and then performing the analysis with a TOC analyzer (TOC-VCHP, Shimadzu, Kyoto, Japan). Ammonium (NH4 +) content was determined using the indophenol blue method after extraction from the soil with a 2 M KCl solution. NO3 − content was extracted using 0.5 M K2SO4 and measured using a colorimetric method (Anderson and Ingram 1993).

Abundance of Microorganisms

Subsamples from collected soils were stored at −80 °C in a deep freezer until molecular analysis. DNA was isolated using the UltraClean Soil DNA Isolation Kit (MoBio Laboratories, Solana Beach, CA, USA). Soil samples of 0.3 to 0.5 g were used for DNA extraction and there was no replication of DNA extraction in one sample since three replicates of core samples were collected. Methanogens and methane-oxidizing bacteria were analyzed for the presence of CH4-related microorganisms, and denitrifying bacteria were analyzed for the presence of N2O-related microorganisms. Abundances of methanogens, methanotrophic bacteria, and denitrifying bacteria were determined using quantitative polymerase chain reaction (Q-PCR). We amplified these microorganisms by using a CFX96™ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) with SYBR Green as the detector (Bio-Rad, Hercules, CA, USA). Each reaction mixture (30 μl) contained specific primer sets (Table 1). We targeted mcrA and pmoA genes for calculating the abundance of methanogens and methane-oxidizing bacteria, respectively. These genes are required for the last step in CH4 production and in the first step for methane oxidation. For denitrifying bacteria, we focused on three different genes, nirK, nirS, and nosZ. These genes are related to the first and last steps of gaseous nitrogen production. The amplification followed a 3-step PCR (Heid et al. 1996; Kolb et al. 2003; Henry et al. 2004): 40 cycles with denaturation at 95 °C for 25 s, primer annealing at 64.5 °C for methanogens and 55 °C for methane-oxidizing bacteria for 25 s, and extension at 72 °C for 25 s. Primer annealing temperatures were 65 °C for nirS, 56 °C for nosZ, and 63 °C (9 cycles) and 58 °C (39 cycles) for nirK. Two independent real-time PCR assays were performed for each soil DNA extract. Standard curves were created using 10-fold dilution series of plasmids containing the target genes from environmental samples. We also calculated the ratios of nosZ abundance to nirK and nirS.

Table 1 Primers used for real-time PCR to measure the abundance of soil microorganisms
Full size table

Statistical Analysis

To determine depth differences over time differences in relation to the abundances of methanogens, methane-oxidizing bacteria, denitrifying bacteria and soil properties, soil was measured at seven depths and four seasons over a 1 year period. The data were analyzed using two-way analysis of variance (ANOVA). As only the surface data were available in winter, two-way ANOVA was conducted only for the data of spring, summer and autumn. In case two-way ANOVA revealed that interactive effects were not significant at the 0.05 level, we used one-way ANOVA thereafter for investigating the differences among the seasons or depths. We used Tukey’s test as a post-hoc analysis for the significant difference at the 0.05 level. Normality (Kolmogorov–Smirnov statistics) and homogeneity of variance were evaluated before the ANOVA analysis. In addition, in case there is no difference among the lower layers, t-test was conducted to compare the average values between upper-layer and lower-layer over seasons. In addition, a correlation analysis was employed to determine the relationship between environmental variables and abundances of microorganisms. To attempt to model the relationship between methane oxidation rates and other environmental factors, we used multiple liner regression models, especially stepwise regression models. Stepwise regression is a semi-automated process of building a model by successively adding or removing variables based solely on the t-statistics of their estimated coefficients. All analyses were conducted using SPSS 18.0 (2011).

Results

Soil Properties

Soil pH, water content, organic matter content, NH4 +, NO3 −, and DOC contents of samples divided from replicated soil cores were measured and average values were compared along soil depths and seasons (two-way ANOVA; Tables 2 and 3). Soil water content was higher in the upper soil layer in summer (0–10 cm, P < 0.05; Tables 2 and 3). Soil pH was less in winter than in other seasons. NH4 + content in the soil increased in summer (P < 0.0005; Tables 2 and 3) but did not show any differences among depths. NO3 − content varied by seasons or depths and tended to increase in summer in the upper soil layer. Organic matter content did not show any significant difference among either seasons or depths (P > 0.05). Higher DOC value was recorded in autumn than in spring (P = 0.007).

Table 2 Soil properties of a rice paddy near Haean-myeon in South Korea
Full size table
Table 3 Results of two-way ANOVA of soil properties at all depths in a rice paddy near Haean-myeon in South Korea
Full size table

Methanogens and Methanotrophic Bacteria

The abundance of methanogens in autumn was higher in the upper layer (0–10 cm), and lower in lower layer soils (10–70 cm). This trend was not observed in the other seasons, but two-way ANOVA revealed no spatial and temporal differences in the abundance of methanogens in other seasons (mcrA) (P > 0.05, Table 4). In contrast to methanogen abundance, methane-oxidizing bacteria (pmoA) abundance differed by depth and season interaction (P < 0.05, Fig. 2, Table 4). Abundance of methane-oxidizing bacteria in the upper soils was higher than lower layer in summer and autumn. We conducted correlation analysis using two different approaches. First, we investigated the relation between soil properties and abundance of methanogens and methane-oxidizing bacteria in each season. No significant relationship for both methanogens and methane-oxidizing bacteria was detected in spring and autumn (P > 0.05). However, in summer, abundance of these microorganisms was related to certain soil properties such as organic matter content (r = 0.475, P = 0.029 for mcrA and r = 0.562, P = 0.008 for pmoA), pH (r = −0.446, P = 0.043 for mcrA and r = −0.565, P = 0.008 for pmoA), and inorganic nitrogen content. Second, we conducted the same analysis with independent whole-season data from upper soils (0–10 cm) and the lower layer soils (10–70 cm). There were no significant correlations between environmental variables and methanogens in upper vs. lower layer soils (P > 0.05). However, the results of the correlation analysis for methane-oxidizing bacteria were different between upper and lower layer soils. Quantities of methane-oxidizing bacteria were positively correlated with organic matter content (r = 0.671, P = 0.017) and NH4 + (r = 0.745, P = 0.008) (Fig. 3) in the upper layer soils, whereas methane-oxidizing bacteria abundance was significantly related to pH (r = −0.307, P = 0.023) and NO3 − (r = 0.442, P = 0.001) in lower layer soils. In addition, stepwise linear regression was performed for both pmoA and mcrA genes with environmental factors. Methanogen abundance did not exhibit correlations with soil properties, but methane-oxidizing bacteria abundance showed a relationship with NH4 + (P < 0.0005).

Table 4 Results of two-way ANOVA of abundance of microorganisms at all depths in a rice paddy near Haean-myeon in South Korea
Full size table
Fig. 2
figure 2

Average abundance of methane-oxidizing bacteria at spring (black), summer (gray), and autumn (blank) in the upper-layer (<10 cm) and lower layer (>10 cm). Different letters denote significant differences at P < 0.05 level based on two-way ANOVA (F = 1.433; P = 0.1889 in depth × season interaction). Because there is no difference among lower layers (10–70 cm), we draw the average value over lower layers

Full size image
Fig. 3
figure 3

Correlation analysis results between upper layer soils (0–10 cm; n = 12) and soil properties in a rice paddy near Haean-myeon, South Korea. The closed circle represents the relationship with organic matter content and the open circle, the relationship with nitrate. In addition, the solid line and dotted line represent the linear regression with organic matter content and nitrate, respectively

Full size image

Denitrifying Bacteria

The abundance of denitrifying genes (nirK, nirS, and nosZ) tended to vary across soil depths (Table 4; Fig. 4). The most abundant gene was nirK, the copy numbers are in the order of 1 × 1011, whereas nosZ exhibited the lowest copy number (in the order of 1 × 106 to 1 × 108). Because the soil was frozen, we collected the upper-layer of soil only in winter (Fig. 5). As such, two-way ANOVA without winter data was performed to compare seasons and depths. The abundance of nirS and nosZ varied with depth (P < 0.0005), and season * depth interaction was insignificant (P > 0.05). We found the average abundance of each gene over spring, summer, and autumn was higher in the upper-layer (<10 cm) than lower layer (>10 cm) by t-test (t = 3.080, P = 0.003 for nirK; t = 5.152, P < 0.0005 for nirS; t = 2.976, P = 0.017 for nosZ; Fig. 4). We also looked into the ratios of nosZ abundance to nirK that is much abundant than nirS. The nosZ/nirK ratios were two to five times greater in upper-layer than all-layer averages at each season (data not shown).

Fig. 4
figure 4

Average abundance of denitrifying bacteria genes (nirK, nirS, and nosZ) over spring, summer, and autumn in the upper-layer (<10 cm) and lower layer (>10 cm). At the lower-layer, nirS was not detected in two samples. Different letters denote significant differences between upper-layer and lower-layer at P < 0.05 level based on T-test (F = 3.054, P = 0.003 for nirK; F = 3.144, P < 0.0005 for nirS; F = 11.128, P = 0.017 for nosZ)

Full size image
Fig. 5
figure 5

Average abundance of denitrifying bacteria genes (nirK, nirS and nosZ) at spring (black), summer (light gray), autumn (dark gray) and winter (blank) in the upper-layer (<10 cm) and whole layer. Different letters denote significant differences among seasons at P < 0.05 level based on Tukey’s post-hoc test following one-way ANOVA

Full size image

Over all seasons, soil pH showed a negative correlation with the abundance of all the genes (P < 0.0005, r = −0.437 for nirK, r = −0.473 for nirS, and r = −0.525 for nosZ). Water content in the soil was also correlated with nirK (r = 0.258, P = 0.036) and nosZ (r = 0.355, P = 0.003) abundance. OM (r = 0.338, P = 0.005), NH4 + (r = 0.304, P = 0.013), and NO3 − (r = 0.386, P = 0.002) showed a correlation with only nosZ expression. In summer, soil pH, OM, NH4 +, NO3 −, and DOC showed correlation with all genes (P < 0.05).

Discussion

The microbes responsible for CH4 and N2O releases are closely related to soil properties and possibly, to oxygen availability, which are expected to be substantially modified by global climate change in the future. CH4 production and N2O emission may increase as temperatures rise, but such change would modified by changes in precipitation and activation of methanogens, methane oxidizing bacteria and denitrifying bacteria.

As wetlands are one of the greatest sources of CH4, many studies have been conducted to determine the rates of CH4 emission and abundance of methanogens in wetland ecosystems such as temperate marshes and northern peatlands (Freeman et al. 2002; Kim et al. 2008, 2012). Those studies have revealed seasonal variations as well as key controlling variables such as temperature change and precipitation. Rice paddies are similar to those systems as one type of wetland, but they have several dissimilar properties. For example, rice paddies are exposed to water level fluctuation by artificial irrigation and to high supply of nutrients as fertilizers. As such, information about methanogens in relation to those two variables is warranted to better understand CH4 dynamics in rice paddies. Methanogens and methane-oxidizing bacteria are sensitive to temperature, water content and oxygen availability; therefore, the abundance of methanogens is very variable in rice paddy ecosystems (Großkopf et al. 1998; Chin et al. 1999; Nicol et al. 2003). In spring, rice paddies are flooded with water and are maintained in a submerged state until the rice harvest in autumn. After that the soils in the rice paddy are exposed to the air during winter until the reflooding in the following spring. Therefore, rice paddy soils are exposed to cycles of aerobic and anaerobic conditions. The alternating water conditions of rice paddies may explain why both methanotrophic bacteria and methanogens are well-adapted to these environments, absorb and release methane in anaerobic versus aerobic conditions, respectively. Periodical exchanges between aerobic and anaerobic conditions, which are suitable for methanotrophic bacteria and methanogens, could provide the optimal environment for the survival of both methanotrophic bacteria and methanogens. As methanogens are distributed evenly throughout soil depths and seasons, the potential for methane emissions is always present. IPCC (2007a, b) predicted that Asia area is likely to receive about 7 % more precipitation than present. The result from this study suggests that global climate change such as temperature rise or local increases in precipitation may accelerate methane production and emission substantially from rice paddies, resulting in positive feedback to global climate change. Therefore, the emissions of methane in Asian countries could increase with the predicted climate change for the region.

Methane-oxidizing bacteria abundance in the upper layer of the soil was higher than that in the lower layer of the soil in this study. We speculate that this difference is due to the gas exchange capacity of the soil in these regions, as oxygen cannot penetrate to lower layer of soils. In support of the idea that oxygen limits the density of methane-oxidizing bacteria, Henckel et al. (2000a, b) also recorded that oxygen was the most important factor for methanotrophic bacterial growth in rice paddy soils (Henckel et al. 2000b). Many studies have investigated the relationships among soil depth and methane oxidation rates and the methane-oxidizing bacteria community in the terrestrial forest soils (Whalen et al. 1992; Adamsen and King 1993; Priemé and Christensen 1997; Henckel et al. 2000a). They also reported that a larger amount of methane oxidizing bacteria exist in upper layer of soils where oxygen availability is high same as our result. Our results extend these observations to rice paddy soils, since we find the largest abundance of methane-oxidizing bacteria only in oxygen-rich areas (i.e., soils in the upper layer, or during seasons of drawn down water levels). In addition, methane produced in anaerobic soil could be oxidized near the soil surface, and hence methane emission rates could be much smaller than production rates. Therefore, if the rice paddy soils convert other land use such as grassland or dryland forest, the methane would be consumed.

To investigate the key environmental factors for methanogens and methane-oxidizing bacteria, we used a correlation analysis between abundance of methanogens and methane-oxidizing bacteria and environmental variables. When the correlation analysis was conducted separately on the soils in the upper- and lower layers, the pmoA gene exhibited different expression patterns. Organic matter and NH4 + contents showed positive correlations with abundance of methane-oxidizing bacteria in the surface soil (Fig. 3). Generally, NH4 + is known to inhibit methane oxidation (Bosse et al. 1993; King and Schnell 1994; Bodelier and Laanbroek 2004). However, our results (Fig. 3) contradict those of previous reports and are instead similar to those of Bodelier and Laanbroek (2004), who observed that the abundance of some methanotrophic bacteria increased when an exogenous nitrogen source was applied. In addition, Mohanty et al. (2006) showed a positive relationship between inorganic nitrogen in rice paddy soils and the abundance of methane-oxidizing bacteria; they concluded that the species comprising the methane-oxidizing bacteria community in rice paddies were different than those of forest soils, and that this phenomenon was a response to inorganic nitrogen (Mohanty et al. 2006). Methane oxidizing bacteria in rice paddy soils might be different with dried soils, so the response for inorganic nitrogen could be different with previous studies. Furthermore, in this study, NO3 − also had a positive correlation with pmoA gene expression in lower layer soils. The effects of NO3 − on methane oxidation are yet to be confirmed. For example, some studies report inhibitory effects on CH4 oxidation (King and Schnell 1994; Dunfield and Knowles 1995), whereas other studies reported no response (Willison et al. 1995). Some studies have even demonstrated that methane oxidation can be stimulated by NO3 − (Jang et al. 2011). Deutzmann and Schink (2011) reported that nitrate could be used as an electron acceptor during anaerobic methane oxidation, particularly for deep sediment soils, especially oligotrophic soils. The NO3 − in our site could also be used as an electron acceptor by methane-oxidizing bacteria, as the organic matter content ranged from 3.9 % to 6.0 %. If this idea is true, it may be that NO3 − and pmoA gene expression is positively correlated. Previous studies reported that inorganic nitrogen inhibit methane-oxidizing bacteria, but this study suggested inorganic nitrogen could stimulate abundance of methane-oxidizing bacteria. Thus, our results suggest that inorganic nitrogen addition may increase methane-oxidizing bacteria. While these results differ from those of some other studies, it may indicate that the fertilization of wet agriculture in rice fields may in some cases increases atmospheric emissions over other landuse types.

We also compared the relative abundance of functional genes (nirK, nirS, and nosZ) along the soil profile, because the variation in gene abundance may be related to denitrification activity (Guo et al. 2011). Denitrifier abundance and community composition may be a good predictor of soil N2O emission in wetland soils (Ma et al. 2011). Although many studies have investigated the response of the denitrifying bacterial community to the climate change in various wetlands or depths (Dowrick et al. 1999; Kim et al. 2008), changes in denitrifier gene abundances with depth have rarely been reported. Palmer and Horn (2012) have reported that the copy number of nirS in surface soil (0–20 cm) tended to be higher than that in deeper soils, while opposite trend was found for nosZ, which is partially in accordance with our findings (Fig. 4). Nitrite reduction (nir) is a rate-limiting and representative step of denitrification that produces the first intermediate of gaseous nitrogen. Nitrous oxide reduction (nos) is the last step of denitrification that reduces N2O to N2 and thereby determines the N2O/N2 ratio of the final denitrification product (Fig. 1). Because the copy number of nirK was 100 times greater than that of nirS (Fig. 4), we compared the ratio of nosZ to nirK rather than to nirS. While every denitrifying gene we analyzed tended to increase in the upper-layer of the soil, the vertical trend of nosZ abundance was the most notable. In addition, the nosZ/nirK ratio was higher in the upper-layer versus the lower-layer of the soil. The denitrification rate could be highest at the surface because the overall abundance of denitrifiers is high, and meanwhile, the relative activity of nitrous oxide reduction is likely highest at the surface, according to the nosZ ratio. In lower layers of the soil, the relative nosZ ratio is lower than that in the upper-layer of the soil, and that relationship may underlie the lower reduction of nitrous oxide there. This trend suggests that N2O emission might originate from activity deeper in the soil profile. This result is accordance with a previous study that investigated the stable isotope signature of N2O in soil-air in a wetland and showed that N2O emission was attributed to lower layer rather than upper layer of the soil profile (Goldberg et al. 2008, 2010a, b). Our result supports the idea that the pattern of N2O emission from soil profile may be related to the distribution of soil microbes in the soil profile.

Many studies have reported that the ratio of gaseous denitrification products (N2O/N2) increases when the soil is acidic because N2O reductase that reduces N2O to N2 is inhibited by acidification (Firestone et al. 1980; Čuhel et al. 2010). When acidity is high (low pH) and/or organic carbon and inorganic nitrogen availability is limited, denitrifying activity would decline and overall gaseous emission (N2O + N2) may decrease (Šimek and Cooper 2002). However, in this study, the abundance of the denitrifying genes had a negative correlation with pH, meaning that high abundance was observed in acidic soil. Because soil pH was not extremely low and the variation of soil pH was not great among soil samples (pH: range 5.3–6.7; mean = 6.2 ± 0.3), we suggest that pH does not strongly control the denitrifying bacteria community in this rice paddy site. However, it may be more likely that the soil was slightly acidified by nitrogen input because nitrogen fertilization or deposition as well as organic matter decomposition typically acidifies soil. Actually, pH also had a negative correlation with ammonium (NH4 +) and nitrate (NO3 −) levels in the soil. Although there was no nitrogen added to fertilize the rice paddy during the research period, there could be nitrogen input from the atmospheric wet deposition (inorganic nitrogen contents increased in summer after monsoon rains; Table 2). The general observation was that denitrifier abundance was correlated with nitrogen contents in the rice paddy, suggesting that the denitrification rate generally increases when higher nitrate concentrations are reached (Henrich and Haselwandter 1997).

The abundance of denitrifiers is known to be influenced both by temperature and carbon supply. The vertical trend of the denitrifier distribution was most distinctive in summer (Data not shown) when denitrifier abundance reached the highest level in upper soil layers. In addition, strong correlations were observed in summer between denitrifier abundance and other environmental factors such as inorganic nitrogen and organic carbon content. Organic carbon substrate availability is one of the most important determinants of denitrifier community abundance (Kandeler et al. 2006). In this research, the abundance of denitrifiers showed positive correlation with OM and DOC in summer. The abundance of denitrifiers is likely to be strongly influenced by carbon substrate in summer, considering that the denitrification rate is high in the growing season (Henrich and Haselwandter 1997). Global climate change will increase mean temperature, the duration of growing season and CO2 concentrations in the atmosphere (IPCC 2007a, b), which would increase carbon supply (Freeman et al. 2004). As a positive correlation between organic matter and denitrifier abundance was found in summer, global climate change may stimulate N2O emission from rice paddies through enhanced denitrification. (Heimann and Reichstein 2008).

The abundance of nirK and nosZ tended to increase considerably in the winter (Fig. 5). It is reported that nirK-harboring denitrifiers are more responsive to conditions in rice paddy soil than nirS-harboring bacteria (Yoshida et al. 2009, 2010). It has been observed that N2O emission increases rapidly when soil is thawing (Goldberg et al. 2010a; Wolf et al. 2010). In addition, Sharma et al. (2006) discovered that denitrification-associated genes were expressed at high levels during thawing and suggested that newly available organic substrates from disrupted soil aggregates may induce microbial activity, thereby increasing denitrification. Our winter sampling time was in late February, when soil was just thawing. Carbon availability might increase at this time by freeze/thaw cycle (Christensen and Christensen 1991) and this may explain the increased denitrifier abundance in winter. Follow-up studies are required to investigate in detail the microbial response to frost and thawing, together with dynamics of N2O emission and soil properties.

Conclusions

We analyzed the abundance of microorganisms related to CH4 and N2O emissions in a rice paddy across different soil depths and seasons. We concluded that microorganisms related to GHGs emissions respond to periodic shifts in environment such as temperature and water management in rice paddies. In addition, soil properties such as oxygen availability, pH, organic matter and inorganic nitrogen contents are also crucial factors for these organisms. As the abundance of microbes did not significantly change across seasons and irrigation-drainage period, the microorganisms related to GHG emissions could adapt well to future climate change such as temperature rises or precipitation patterns change. However, the changes of soil properties by future climate changes could lead to changes of microorganisms. A longer growing season induced by global climate change may affect the abundance of various types of microorganisms in different ways. This study was limited to observations over 1 year where we analyzed the quantities of microorganisms. Such limitations could be overcome by performing additional longer-term monitoring and measurement of actual gas fluxes in the rice paddy environment.