Introduction

Terraced paddy fields are one of the most important land use systems for agricultural products in Eastern Asia [1]. However, as a result of shifting economies and the development of high-yield rice varieties, a substantial number of traditional rice fields are no longer cultivated and are abandoned. According to the Korea Institute of Geoscience and Mineral Resources [2], the area of abandoned farmland in Korea was 13,000 ha, and 61 % of Korea’s rice paddy was removed from production during the last century. In Korea, mountainous areas contain 48 % of the abandoned rice paddies [3]. Similarly, by 1993, over 10 % of Japan’s terraced paddy fields, which are also located primarily in the country’s mountainous regions, had been abandoned [4]. Ecological impacts by such changes have drawn much attention, and effects on plant and animal biodiversity have been well documented [5] However, little information is available about the effects on soil microbial properties, which are closely related to wetland functions.

Abandoned rice paddy fields can transform into marsh-type wetlands through natural succession or convert to other types of ecosystems such as grassland or dryland forest grassland, depending on the hydrology. Such dramatic changes in the soil environment can influence the structure, abundance, and activity of microorganisms, which play a critical role in decomposition of organic matter and greenhouse gas emissions in soil [6, 7]. If abandoned rice paddy fields transform into wetlands, they can play a significant role in restoring lost wetland functions to ecosystems including water quality improvement, floodwater storage, biological productivity, and carbon sequestration [8]. Primary succession of plant communities directed toward a climax is not a typical occurrence in wetlands because these ecological systems are inherently dependent on hydrology, and temporal hydrologic variability often causes reversals or setbacks in succession [9]. Succession of plants can be strongly affected by hydrology because terraced abandoned rice paddies consist of a mosaic of different elevations. Therefore, plant species could establish based on the hydrology regime and soil condition and that could produce different developmental stages in the plant community [9]. Different plant species can create wetlands with different ecosystem functions, and they can induce different environmental conditions. These feedbacks would affect the greenhouse gas fluxes and microbial communities in these ecosystems [10]. Park et al. [11] reported that abandoned paddy terraces showed similar plant community across Korea even though environmental conditions such as climate, biogeographic history, and soils were different.

Natural wetlands have large carbon pools and have important roles as natural carbon sinks in global carbon cycles. However, wetlands are also a natural source of CH4, which results from methanogenesis by methanogen, occurring in the sediments. In addition, N2O can also be produced by nitrification in aerobic conditions [11]. 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 [12–14].

We present the results of a study of three wetlands with different plant communities that developed by natural succession in response to changes in hydrology. We investigated greenhouse gas emissions, soil physicochemical characteristics, and microbial communities in rice paddy fields that have transformed into wetlands. The objective of this study was to assess the impact of season and plant species on greenhouse gas flux, emphasizing the interdependency of soil chemical and microbial properties. We hypothesized that different hydrological conditions would produce different plant communities through succession, and as a result, the wetlands would function differently with respect to carbon capture and greenhouse gas mitigation.

Materials and Methods

Site Description and Sampling

The study site was located in a mountainous region, with an elevation ranging from 77 to 87 m, in Ansan, Gyeonggi-do of central Korea (37° 17′10.68″ N, 137° 55′ 25.96″ E; 6353 m2, Fig. 1). Many terraced paddy fields were developed in this region and were abandoned 10–20 years ago [2]. The region has very warm and rainy conditions in the summer and cold and dry conditions during the winter. The main water source was surface water, and the mean precipitation at the study site is 1235.2 mm per year [2]. The rainy season in Korea is spread over the June–September period. Summer rainfall over Korea accounts for 50–60 % of the annual precipitation [2]. Even though the monsoonal climate produces heavy rain during the summer season, the water depth in these sites was kept at a constant level by slopes and dikes, with excess water flowing out downhill. The physicochemical properties of the water were as follows: pH of 5.4–5.7, EC of 75–79 μs cm−1 and DO of 19.6–7.5 mg l−1. Despite the abandoned rice paddy wetlands being relatively lower in altitude than high moors, the contents of calcium (0.45 ± 0.2) and magnesium (1.48 ± 0.6) ions, which are critical limiting factors for Sphagnum spp., were at very low levels, as were those of nitrogen and phosphorous (PO4 −, 0.02 ± 0.0; NO3 −, 0.25 ± 0.3; NH4 +, 0.06 ± 0.1).

Fig. 1
figure 1

Vegetation map within abandoned rice paddies in Ansan (modified Hong and Kim [15])

Full size image

We selected three sites that were dominated by different plants as presented in Fig. 1 [9, 15]. The dominant plant of site 1 was Carex dickinsii (Cyperaceae). This site, which covered about 908 m2, resembled upland sites, because of the low water table (0–5 cm). The dominant vegetation in site 2, which covered about 1500 m2, was Phragmites australis (water table 15–20 cm). Site 3, which covered about 3200 m2, was dominated by a Sphagnum species, identified as Sphagnum palustre. The water table was about 20–25 cm. Gas flux, plants, and soil samples were collected at the three locations that were dominated by different plants for two seasons, winter and summer, in 2012.

Soil samples were collected at random locations using soil cores (50-mm diameter) to a depth of 10 cm through the surface soil in the middle of each wetland after removing the small amount of surface litter. In this study, we collected only surface soils (0–10 cm) to assess the impact of plant species on microbial properties because this is where we expected the greatest differences to be found. Vegetation generally grows slowly over time since abandonment, and over 10 years, such paddy terraces in mountainous valleys would be expected to have similar soil properties deeper in the soil. Three replicate samples, each comprising nine pooled soil cores, were taken at each site per season. Field moist soils were sieved <2 mm, and large pieces of plant material and soil animals were removed before use. After sieving (2 mm), soils were stored at 4 °C, until analysis. All analyses were completed within 4 days of sampling. The soil temperature was measured by a portable thermometer.

Plant samples were collected from triplicate quadrats (1 × 1 m) in each site. Dry mass of plant (Cyperaceae, Phragmites, Sphagnum) was determined by drying a subsample at 75 °C for 3 days.

Soil Characteristics

Soil pH was determined by adding soil to water at a ratio of 1:5 (w/v) and analyzed using a pH meter. Soil moisture content was determined gravimetrically, by drying for 24 h at 105 °C and comparing wet and dry mass of each soil. Organic matter content was determined by loss on ignition, at 600 °C in a furnace (MAS 7000, CEM). Samples for the measurement of extractable dissolved organic carbon (DOC), soil nitrate, and ammonium were prepared by adding deionized water (9 ml) to soil (1 g) and shaking for 10 min. After centrifuging at 10,000 rpm for 5 min, the samples were passed through a 0.45-μm filter and frozen until analysis. Soil nitrate was measured using a nitrate electrode [16]. Ammonium (NH4 +) concentrations were measured using the indophenol blue method [17] with a DR 2000 Auto analyzer at 425 nm. DOC concentrations were analyzed using a TOC analyzer (Shimadzu, Model TOC-5000, Japan).

To measure the concentration of water-extractable phenolic compounds, 1 ml of filtrate was mixed with 1.5 ml of Na2CO3 (50 g l−1) and 0.5 ml of Folin-Ciocalteu reagent and incubated for 2 h at 20 °C, in the dark. Phenol was used for the standard calibration curve, and absorption at 750 nm was measured by a spectrophotometer (FLUO-star OPTIMA, BMG LABTECH).

Greenhouse Gas Flux

To measure greenhouse gas (CO2, CH4, N2O) flux in wetlands, we collected gas samples using a static chamber method. Three acryl chambers (20 cm in diameter and 15 cm in height) were placed close to each soil sampling location. These chambers were inserted about 5 cm into the ground, and there was no vegetation inside the chamber [18]. Each lid had a rubber septum placed in its center. To minimize the effects of chamber installation, we began taking measurements 1 month after installation. The lid was placed on the chamber only during gas sampling occasions. Gas samples were drawn through a septum using air-tight syringes at times 0, 10, 30, and 60 min after the closing of the lid. Previous studies employing static chambers collected gas samples around 1 h [19–21], provided that the reduction rate exhibited linearity. All gas samples were transferred to glass vials, which had been previously evacuated. Gas samples were taken to the laboratory, and the concentration of each gas was analyzed by gas chromatography, with flame ionization detector (FID) and electron capture detector (ECD). The gas flux of each site was calculated by the equation below.

$$ \mathrm{Flux}=\frac{\updelta \mathrm{C}}{\updelta \mathrm{t}}\times \frac{V\times M}{A\times V\mathrm{mol}} $$
(1)

where flux is the gas flux (mg gas m−2 day−1), δC/δt is the gas concentration change rate over time (μl l−1 min−1), V is the volume of the chamber, M is the molecular weight of the gas (CO2, CH4, N2O), A is the basal area of the chamber, and Vmol is the volume of one mole of gas at a certain temperature (m3 mol−1).

Analysis of Microbial Activity

The activities of four hydrolases (β-glucosidase, N-acetyl-b-glucosaminidase, phosphatase, and arylsulfatase) were measured. Hydrolase enzyme activities were measured by providing substantial amounts of methylumbelliferyl (MUF)-linked substrates. Two grams of peat soil was suspended in 10 ml of acetate buffer (50 mM, pH 5.0), and 1 ml of this sample suspension was used to assay enzyme activities. To measure hydrolytic activity, 0.83 ml of substrate solution, each containing 200 mM β-glucoside, 800 mM phosphate, 800 mM sulfate, and 400 mM N-acetyl-b-glucosaminide, was added to the suspension. The concentration of each substrate in solution was calculated to obtain maximum potential enzyme activity, which exceeds the saturation concentration in peat soils [22]. The mixture of sample suspension and substrate solution was incubated for 60 min at 25 °C and centrifuged at 5000×g for 5 min to stop enzyme activity. A fixed volume (300 ml) of supernatant was transferred to a 96-well black plate, and absorbance was measured (conditions: emission, 460 nm; excitation, 355 nm; FLUO-star OPTIMA, BMG LABTECH). MUF-free substrate was used as a negative control, and hydrolytic enzyme activity was expressed as nanomole MUF gram per dry soil per minute [23].

Abundance of Microorganisms

Microbial DNA was extracted from about 0.5 g of soil, using a Power Soil DNA isolation kit (MoBio, USA), following the manufacturer’s instructions. To check for sample contamination, a parallel DNA extraction procedure was performed with an unused filter, which was used as “blank” for further analysis. The quantities of eubacteria, methanogens, denitrifiers, and methanotrophs that would produce or consume greenhouse gases were measured by quantitative polymerase chain reaction (qPCR), using an I-Cycler™ (Bio-Rad, USA) and SYBR Green as a detection system (Bio-Rad, USA). Each reaction was performed with specific primer sets for each group (Supplemental Table 1). The amplification followed a three-step polymerase chain reaction (PCR) for both bacterial 16S ribosomal RNA (rRNA) gene and fungal internal transcribed spacer (ITS) regions: 40 cycles with denaturation at 94 °C for 25 s, primer annealing at 50 °C for 25 s, and extension at 72 °C for 25 s.

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. 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 for the first step in methane oxidation. For denitrifying bacteria, we targeted two different genes, nirS, which catalyze the reduction of nitrate to nitric oxide, and nosZ encoding the catalytic subunit of N2O reductase. These genes are related to the first and last steps of gaseous nitrogen production. The amplification followed a three-step PCR [28–30]: 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). Two independent real-time PCR assays were performed for each soil DNA extract. Standard curves were created using tenfold dilution series of plasmids containing the target genes from environmental samples. DNA concentration was described as the gene copy number per gram of dry soil.

Microbial Community Structure

For comparing microbial community structure of each site, we utilized terminal restriction fragment length polymorphism (T-RFLP) analysis of soil DNA. DNA samples were amplified by PCR using the fluorescently labeled forward primer 27F (5′ [6FAM]-AGAGTTTGATCCTGGCTCAG-3′) and the unlabeled reverse primer 927R (5′-CCGTCAATTCCTTTRAGTTT-3′), which target bacterial 16S rRNA genes [24]. For fungal communities, PCR was performed using the fluorescently labeled forward primer ITS1F (5′ [Hex]-CTTGGTCATTTAGAGGAAGTAA-3′) and the unlabeled reverse primer ITS4 (5′-TCCTCCGCTTATTGATATGC-3′), which target the fungal ITS region of the rRNA gene [25–27]. Archaea, methanogen, methanotroph, and denitrifier were amplified using PCR, with primer pairs described in Supplemental Table 1.

Reaction mixtures (50-μl reactions) contained approximately 100 ng of template DNA, Taq DNA polymerase (2.5 U), PCR buffer (1×; 10 mM KCl, 10 mM (NH4)2SO4, 20 mM Tris–HCl, 2 mM MgSO4, 0.1 % Triton X-100, pH 8.8), deoxynucleoside triphosphates (0.2 mM each), forward and reverse primers (0.2 μM each), and bovine serum albumin (8 μg). DNA of each sample was amplified using an MJ Research thermal cycler PTC 100 (MJ Research, Waltham, MA), and the following program was applied for analyzing both bacterial and fungal communities: 5 min 94 °C, followed by 35 cycles of 94 °C for 1 min, 50 °C for 1 min and 72 °C for 1 min and 30 s, and a final extension step at 72 °C for 10 min. For each sample, the PCR products of two reactions were pooled and purified using a MoBio PCR purification kit (MoBio) according to the manufacturer’s instructions.

For T-RFLP analysis, approximately 300 ng of purified PCR product was added to a reaction mixture (final vol. 25 μl) containing 10 U of restriction endonuclease HhaI (Promega, Madison, WI) and incubated at 37 °C for 4 h. Digests were desalted and aliquots (1 μl) were used for T-RFLP analysis. Terminal fragment size analysis was performed using ABI 377 DNA Analyzers (Applied Biosystems) in conjunction with GeneScan software (Applied Biosystems).

Statistical Analysis

Data were analyzed using SPSS 21.0 software for two-way ANOVA. The test was carried out on the greenhouse gas flux, soil characteristics, soil enzyme activity data, and quantity of microbes to determine the effects of season and wetland types (between the groups). Standard deviations are shown as numerals in the tables and as error bars in the figures. Microbial community structure and diversity were determined using terminal restriction fragment (T-RF) profiles. The fragment data were transformed into proportional abundance values (%) relative to the total fragment peak area in each sample. Non-metric multidimensional scaling (NMS) ordination was performed using the T-RF data set to compare community structure between sites and seasons. We also calculated the Shannon indices [30], based on the complete linkage clustering data and T-RFs.

Results

Soil Chemistry, Greenhouse Gas Flux, and Plant Biomass

Soil temperature, soil water content, soil organic matter, pH, DOC, phenolics, NH4 +, and NO3 − were measured for two seasons (Table 1). The soil pH of all sites of all seasons was below 7, ranging from 4.7 to 6.1, and the lowest soil pH was from the summer in Sphagnum sites. The concentrations of NH4 + and NO3 − varied from 0.04 to 0.1 mg NH4 g−1 dry soil and from 0.006 to 0.04 mg NO3 g−1 dry soil. Organic matter was influenced by site × season interactions, as indicated by the two-way ANOVA. Moisture content increased in Phragmites sites and Sphagnum sites during summer but decreased with in Cyperaceae sites during summer. The moisture contents were strongly influenced by both sites and sites × season interaction. The concentrations of DOC using both the measurement techniques were higher in the summer at Cyperaceae, while they were lower in the summer at Phragmites sites and Sphagnum sites (Table 1).

Table 1 The physicochemical characteristics, enzyme activities, and greenhouse gas flux in soils
Full size table

Greenhouse gas fluxes (CO2, CH4, and N2O) in abandoned rice paddy wetlands in summer were higher than those in winter (Table 1). CO2 flux was significantly affected by the season. In particular, CO2 emissions of Phragmites sites were highest in summer. According to the correlation between temperature and gas emission, CO2 emission was positively correlated with temperature (data not shown). Dissolved and gas carbon were affected by season in each wetland (Fig. 2). Carbon was released in gaseous form than in dissolved form in soils from the summer (Fig. 2b). CH4 and N2O fluxes were not different by season or sites (plant species). N2O ranged from −0.0005 to 0.007 μg N2O m−2 day−1, and CH4 ranged from −1.2 to 0.08 mg CH4 m−2 day−1. Negative CH4 and N2O flux values indicate consumption.

Fig. 2
figure 2

Dissolved and gaseous carbon released in a winter and b summer. Values are represented on a mass C basis. Different letters denote variation among sites, between dissolved and gaseous forms

Full size image

Soil moisture content also differed among the sites that were dominated by different plant species and was negatively correlated with DOC (R = −0.809, P < 0.01, data not shown). DOC was also negatively correlated with organic matter (R = −0.596, P < 0.05) and moisture content, while organic matter was positively correlated with moisture content (R = 0.495, P < 0.05).

The results showed that the length and biomass of Phragmites were the highest (Supplemental Table 2). The dry masses of Phragmites, Cyperaceae, and Sphagnum were 317, 110, and 247 g m−2, respectively.

Microbial Activity and Community Diversity

The activity of all enzymes differed, depending on sites and seasons (Table 1). Soil microbial activity increased in Phragmites sites and Sphagnum sites during the summer. However, the activities of β-glucosidase and N-acetyl-b-glucosaminidase decreased in Cyperaceae sites during summer.

The soil microbial abundance was strongly affected by season and sites. The abundance of bacteria and fungi was higher in summer although the abundance of archaea in summer was lower in all sites (Fig. 3). The abundance of bacteria was similar among the sites in winter, but the abundance of bacteria was higher in Phragmites and Sphagnum sites than those in Cyperaceae sites (Fig. 3a). Although the abundance of archaea was similar among sites, they were more abundant in winter than in summer (Fig. 3b). Likewise, the abundance of methanogens in winter was higher than in summer (Fig. 4a). In contrast to methanogens, the abundance of methane-oxidizing bacteria (pmoA) in Cyperaceae sites and Phragmites sites did not differ by season (P < 0.05, Fig. 4b). Winter abundance of methane-oxidizing bacteria in the Sphagnum site was higher than in summer. The abundance of denitrifying genes (nirS and nosZ) tended to vary across seasons (Fig. 4c, d). We found that the average abundance of nosZ increased in winter but did not show any differences among the sites (Fig. 4c). However, the abundance of nirS was less in winter than in summer. In addition, the nosZ/nirS ratio was higher in the winter versus in the summer.

Fig. 3
figure 3

Abundance of a bacteria, b archaea, and c fungi over summer and winter. Different letters denote significant differences between sites and season at P < 0.05 level based on two-way ANOVA

Full size image
Fig. 4
figure 4

Abundance of a methanogene (mcrA), (b) methanotrophic gene (pmoA), denitrifying bacteria gene c nosZ and d nirS over summer and winter. Different letters denote significant differences between sites and season at P < 0.05 level based on two-way ANOVA

Full size image

To analyze the effects of season and sites on soil bacterial communities, DNA-based T-RFLP analysis was conducted. Figure 5 shows Shannon’s diversity index (Supplemental Table 3). In both the archaea and fungi, H’ values in summer were higher than in winter. In contrast, the H’ values of the nirS in summer were lower than those in winter. The T-RFLP analysis showed that fungi and pmoA were more diverse and organisms with nirS were less diverse in summer (Fig. 6). Most other measures did not show a strong seasonal pattern. Sphagnum sites had less diverse archaea than the other sites, but the sites were otherwise similar.

Fig. 5
figure 5

Shannon diversity indices of T-RFLP profiles in a winter and b summer. C dominated by Carex dickinsii (Cyperaceae), P dominated by Phragmites australis, S dominated by Sphagnum

Full size image
Fig. 6
figure 6

Non-metric multidimensional scaling (NMS) ordination of a bacterial and b Archaea, c fungal communities, d methanogen, and e denitrifier in rhizosphere soils, determined by terminal restriction fragment (T-RF) profiles. Each axis reflects relative distance between points based on the originally given first axis score, with the automatic orientation of the center, and the stress values were 0.21 (a) and 0.14 (b). C dominated by Carex dickinsii (Cyperaceae), P dominated by Phragmites australis, S dominated by Sphagnum

Full size image

Discussion

These research sites had very different plant communities that were developed by natural succession in response to different hydrological regimes of terraces. This study focused on quantifying seasonal fluxes of CH4, N2O, and CO2 from soils with contrasting dominant plant types in wetlands from abandoned terraced rice paddies. We hypothesized that different plant communities with different hydrological conditions would function differently with respect to carbon capture and greenhouse gas mitigation. In this study, Phragmites sites had the highest CO2 emission and highest biomass in summer although greenhouse gas emission was low compared to other ecosystem. These high CO2 emissions related to high microbial activity with high plant biomass [31]. In this study, the biomass of Phragmites was the highest, compared with those of Sphagnum and sedge [15] (Supplemental Table 2). Root exudates are an important source of labile organic matter for soil microbial communities and are released at greater rates during the growing season than when the plants are dormant [32]. In Phragmites wetlands, enzyme activity and fungal and bacterial abundance increased during summer with increasing plant biomass. However, methanogen and denitrifier did not increase in the sites. That means that DOC cannot change N2O or CH4. The methanotrophic activity in submerged freshwater marshes is predominantly found in the oxic rhizosphere of macrophytes [33, 34] and in the oxic surface layer of the sediment [35, 36]. Carbon fixation was extrapolated to annual values to assess whether the wetland was net C sources or sinks. We calculated carbon uptake and loss-based biomass and carbon loss form. The carbon uptake by Phragmites was 151.05 g m−1, but the carbon loss was 1.09 g m−1 year−1 (gases, dissolved forms of C). That means that Phragmites wetlands can be C sinks.

Sphagnum-dominated wetland had the greatest enzyme activity but produced low greenhouse gas flux. This result might be an exceptional case of inhabitation of a Sphagnum in this site. Most distributions of Sphagnum were observed at tussock structures as microtopography by sedges and grasses within a wetland (74 %) and the shaded slope area under Pinus densiflora’s canopy in wetland boundary (26 %) [15]. Plant productivity such as emergent vegetation is generally the largest C flux into ecosystems, and any alteration thereof could drive changes in ecosystem C storage [37]. Plants can increase soil enzyme activity and can also affect the microbial species composition and diversity, by releasing exudates and oxygen into the rhizosphere [38]. Plants can also indirectly mediate enzyme activities in wetlands by controlling the above-ground and below-ground litter production. Sphagnum mosses are the dominant moss species in many peatland ecosystems and play a key role in C accumulation. Sphagnum mosses are excellent peat-forming plants because they decompose at very slow rates and produce metabolites with antimicrobial activities [39]. Microtopography and Sphagnum planted together would be highly efficient for capturing C.

Sphagnum-dominated wetland produced the most CH4 among three kinds of wetlands in this study. The CH4 flux of Sphagnum sites was in the range from 0.0004 to 0.084 mg CH4 m−2 day−1 and slightly higher than those of Phragmites sites and Cyperaceae sites, but these were not significant. These results related to high abundance of methanogen (Fig. 4) with high water table. The water table of Sphagnum sites (10–15 cm) was slightly higher than those of Phragmites sites or Cyperaceae sites. Anaerobic conditions in wetlands are highly favorable for the production of methane [40] and nitrous oxide [41]. Water level plays a key role in controlling CH4 emissions, by determining the interface between aerobic and anaerobic processes. Saturated soil limits the diffusion of atmospheric oxygen into wetland sediment and so also oxygen availability, microbial activities, and decomposition rates, thereby reducing CO2 and increasing the CH4 emission rate [40]. Conversely, decreasing the water table level increases oxygen diffusion and carbon decomposition, thereby causing the emission rate of CO2 to increase and CH4 to decrease [40].

In the Cyperaceae sites, there is not much methane produced, likely because the water table is very low and methane is produced in anoxic conditions. However, Cyperaceae sites produced the N2O flux with high denitrifying bacteria. Cyperaceae sites were at a higher elevation, allowing water to drain freely and making it almost like an upland ecosystem. Drainage can increase nitrous oxide emissions from minerotrophic peatlands [42, 43]. However, mean N2O flux ranged from −0.0002 to 0.007 μg N2O m−2 day−1 which were very low value due to low nutrient concentration (N, P) in this site. Nitrous oxide emissions from nutrient-poor sites are considered to be affected little by drought because they lack nitrate [42, 43], which is a precursor and a major control on denitrification rates [44] and nitrous oxide emissions [45] from organic soils.

As our expectation, microbial communities were changed significantly by plant species although their characteristics were strongly affected by season. This means that the microbial community composition is able to respond to dominant plant species. Plants and soil biota have a strong structural and functional linkage, and plant species can cause high heterogeneity of resource environments and consequently develop diverse microbial groups in ecosystems [46]. Generally, T-RFLP gives a coarse snapshot of community composition. According to the results of T-RFLP, it showed the lack of detectable differences among sites and especially among seasons in this study. Therefore, further research is needed to better understand the microbial community effects using other community profiling tools.

Seasonal Effects

We found that the greater magnitude of differences in the summer and that being greatest in the site with the greatest plant biomass suggest that the plants are having an effect, since water levels do not change much, regardless of rainfall amounts. However, we found that seasonal effects were the key factors in determining ecosystem functions in these mountainous abandoned rice paddy wetlands.

Seasonal change likely influenced wetland function by impacting the hydrology and plant biomass [31]. Soil moisture and organic matter content were higher in wetlands dominated by other species than in Cyperaceae-dominated sites. These results may be related to precipitation and elevation. In Korea, summer rainfall accounts for 50–60 % of the annual precipitation [2].

Carbon released from the system was mostly in a dissolved form in winter (Fig. 2) but turned into a gaseous form in summer. Soil CO2 fluxes were generally higher in summer than in winter and had a seasonal pattern that was likely driven more by soil temperature and soil moisture than by other factors. Soil water content limits on soil respiration have also been observed by Suseela et al. [47]. These may be the driving forces behind the increased efflux of CO2, as increasing temperature tends to increase soil microbial metabolism [48]. High CO2 flux usually occurs in the summer, when soil temperature is higher, and soil water content and substrate C availability are adequate, while low emission occurs in the winter, when soil biological activity is minimal, due to near-freezing soil temperatures [49–51]. Our results showed that soil temperature and water content influenced microbial activities and C mineralization, thereby resulting in different CO2 fluxes.

In this study, microbial abundance showed sensitivity to seasonal effects. Seasonal variations in rhizosphere methane oxidation were probably due to seasonality in rhizosphere oxygen concentrations. Availability of O2 for methanotrophs is determined by the delicate balance between the amount of O2 transported and the amount consumed by root respiration and (non-methanotrophic) oxygen-demanding processes in the rhizosphere.

The abundance of denitrifying genes (nirS and nosZ) tended to vary across seasons (Fig. 4c, d). However, the abundance of nirS was lower in winter than in summer. We found that the average abundance of nosZ increased in winter but did not show any differences among the sites (plants) (Fig. 4c). These results might be related to low nitrogen and low organic carbon in mountainous wetlands. Land use changes can influence the production and consumption of gases [14]. In Korea, we have shown that abandoned terraced rice paddy fields transform into wetlands because they receive high amounts of precipitation. However, greenhouse gas flux was limited because of their physicochemical characteristics such as high elevation and low nutrient availability.

Ecological Implications

Overall, CH4 flux was in the range from −1.2 to 0.08 mg CH4 m−2 day−1. These ranges are very low, compared to the CH4 flux of other irrigated rice paddy fields in Asia. Kazuyuki et al. [41] reported that the total emission rates of CH4 during the cultivation period were from 14.8 to 9.49 g m−2 in the continuously flooded condition in Japan. In South Korea, methane emission from irrigated rice has been measured at three sites: Suwon [52, 54], Iksan [55, 56], and Milyang [57, 58]; the mean value of the three sites of the CH4 emission was 10.04 mg CH4 m−2 h−1. The reasons for this are related to the high elevation of the sites and low nutrient availability. Wind, longer sunlight duration, and low nutrient likely negatively affected greenhouse gas production in our study sites. The lower nutrient content and higher acidity of the bog peat would suggest lower methane production and emission [59]. Methane exchange in high-elevation ecosystems may be particularly sensitive to changes in temperature and precipitation. Morse et al. (2012) [60] reported that methane production and consumption in Wyoming subalpine meadow and forest soils during snow-free periods were spatially and temporally controlled by soil moisture. Such knowledge would improve our understanding of the effectiveness of wetland land use on greenhouse gas mitigation in Asia.

Conclusion

Terraced abandoned rice paddies that transform into wetlands with different plant species had similar greenhouse gas fluxes and released similar forms of carbon by their each geological characteristic. Gas flux was very low compared to other wetlands or rice paddies, although gaseous carbon forms and microbial activity significantly increased in the summer. Due to their low greenhouse gas flux, we suggest that the abandoned rice paddy wetlands with high ecological values should be continuously managed, under governmental or municipal conservation. However, more research should be conducted to clarify the general function of mountainous wetlands in eastern Asia.