Introduction

Methane (CH4) is a greenhouse gas that has the second largest radiating force after carbon dioxide (CO2); CH4 accounts for 78% of total global CO2-equivalent emissions1. As one of the largest anthropogenic sources of CH4 emissions, paddy fields emit 25–54 Tg of CH4 annually2, which is 4–9% of total annual CH4 emissions (IPCC, 2001)3. Therefore, a change in paddy CH4 emissions is of vital significance to global climate change. It is well known that global climate change affects paddy CH4 emissions. For example, elevated atmospheric CO2 and higher soil temperature alone and/or in combination can lead to higher paddy CH4 emissions4,5,6,7. Moreover, elevated ground-level ozone (O3) is a consequence of global climate change8. Ground-level O3 forms from the photochemical oxidation of CH4, carbon monoxide (CO) and volatile organic components (NMVOCs) in the presence of nitrogen oxides (NOX = NO + NO2). However, the impact of elevated ground-level O3 on potential feedbacks of paddy CH4 emissions is largely unknown9, although this information is an indispensable component of a comprehensive understanding of paddy ecosystem responses to global climate change. To the best of our knowledge, only a small number of studies have been conducted, which revealed that elevated ground-level O3 significantly reduces cumulative CH4 emissions from paddy soil9,10 or from peatland11. The underlying mechanisms remain elusive.

In terms of the effects on ecosystem production and function, ground-level O3 is the most important gaseous air pollutant globally12; its concentration has been increasing since the industrial revolution and will continue to increase in the coming years13. Dentener et al.8 suggested that global annual mean surface O3 concentrations will increase by between 1.5 ppb (current legislation scenario) and 4.3 ppb (IPCC SRES A2 scenario) over the period of 2000–2030. Ground-level O3, especially while at an elevated concentration (e.g., over 50 ppb), is phytotoxic with a potential to damage plant photosynthesis, leading to a reduced aboveground biomass and crop yield12,14. Negative effects may also be extended to the belowground biomass of plants. Many reports have indicated that O3 exposure reduces carbon allocation to roots15, the root/shoot biomass ratio16 and root exudates17. O3 stress also indirectly affects soil microorganisms through changes in carbon input to soils15 because soil and vegetation itself can remove most O3 from the atmosphere18. As plants are the main input of carbon and energy to the plant-soil-microbe web, a decrease in the carbon flux from plant to soil due to elevated O319 has been found to adversely influence the diversity of paddy microbes20,21.

Biologically, CH4 production is monopolized by methanogenic archaea, which fall into the phylum Euryarchaeota and form six distinct orders: Methanomicrobiales, Methanocellales, Methanosarcinales, Methanobacteriales, Methanococcales and Methanopyrales. Methanogenic archaeal communities are found to be sensitive to global climate change. Both elevated atmospheric CO222 and higher soil temperatures7 alter the composition of paddy methanogenic archaeal communities and increase their abundance and activity. However, information regarding the responses of paddy methanogens to elevated ground-level O3 is limited. Methanogenesis is the final degradation process of organic matter in paddy fields. Organic matter is first anoxically degraded to small molecules, such as acetate, CO2 and H2, by diverse bacteria. With the help of methanogenic archaea, some of these molecules are further converted into CH423. Therefore, the decrease in carbon input from plant to soil due to elevated ground-level O3 was hypothesized to have a negative effect on methanogenic archaea, especially aceticlastic methanogens, because acetate is quantitatively the most important intermediate in the anaerobic degradation of organic matter and two-thirds of biologically produced methane is derived from the methyl group of acetate24. Thus, we conducted a field experiment in the China Ozone Free-Air Concentration Enrichment (FACE-O3) facility on a rice-wheat rotation system. Changes in paddy methanogenic archaeal composition and abundance in response to elevated ground-level O3 were investigated using culture-independent methods, including 454 pyrosequencing and real-time quantitative PCR. Methanogenic activity was determined by measuring CH4 production during microcosm incubation. Our results contribute to a comprehensive understanding of the impact of global climate change on paddy ecosystems as well as feedback between the two.

Results

Soil characteristics

During rice growth, soil dissolved organic C (DOC) concentrations ranged from 80 ± 6.6 to 103 ± 6.1 mg/kg dry weight soil (d.w.s) (Fig. 1a). According to a pairwise comparison, elevated ground-level O3 significantly decreased the soil DOC concentration under ambient O3 (p < 0.05), from 100 ± 3.6–103 ± 6.1 to 80 ± 6.6–90 ± 4.0 mg/kg d.w.s. The acetate content was measured because it is one of the most abundant low-molecular-weight organic acids in paddy soil and one of the main precursors of CH4 production. As shown in Fig. 1a, elevated ground-level O3 significantly decreased the acetate contents under ambient O3, regardless of the rice growth stage (p < 0.05), from 1.80 ± 0.17–2.12 ± 0.20 to 1.34 ± 0.26–1.49 ± 0.13 mM.

Figure 1
figure 1

Concentrations of dissolved organic C (DOC) and acetate in paddy soils (a), copy numbers of methanogenic archaeal and Methanosaeta 16 s rRNA genes (b) and the dynamic curves of methane production in incubated flooded paddy soils (c) under elevated ground-level O3 (FACE-O3) and ambient O3 (Ambient) concentrations.

Data are expressed as the means with standard deviation (SD). There were 12 soil samples in total. Significant differences are indicated by different letters shown above the error bars (p < 0.05) and mean separation was assessed by Tukey's multiple range test. The capital, small and underlined letters indicate different assays.

Full size image

Methanogenic archaeal abundance and activity

Copy numbers of 16 s rRNA genes of paddy methanogens and those of Methanosaeta were measured using qPCR at both rice growth stages under different ground-level O3 concentrations (Fig. 1b). Due to its non-specificity, the primer set of 1106F/1378R targeting methanogenic archaeal 16 s rRNA genes can also detect some non-methanogenic archaea25. However, the strong relationship between the numbers of methanogenic archaeal 16 s rRNA genes and mcrA genes indicates that these primers are suitable for the quantification of methanogenic archaea26 and implies that those non-specific amplified microbes have the same responses to environmental parameters as paddy methanogens. Therefore, although they could overestimate the abundance of paddy methanogenic archaea, primers 1106F/1378R were still used in this investigation. Before quantification, the interference of inhibitory substances in purified DNA extracts on qPCR was tested by a 10-fold dilution series (up to 1000-fold) of one random DNA extract targeting the 16 s rRNA gene fragment of methanogens. The copy numbers derived from the dilution series were not significantly different with numbers ranging from 2.1 × 109 to 2.3 × 109/g d.w.s, which indicated that there were no inhibitory substances in the purified DNA extracts. The copy numbers of the methanogenic archaeal 16 s rRNA gene varied from 1.78 ± 0.23 × 109 to 2.44 ± 0.15 × 109/g d.w.s. There appeared to be a greater abundance of the methanogenic archaeal 16 s rRNA gene at the rice tillering stage than at the rice anthesis stage. A pairwise comparison to methanogenic archaea under ambient O3 conditions suggested that elevated ground-level O3 caused a decrease in the copy numbers of the methanogenic archaeal 16 s rRNA gene from 2.44 ± 0.15 × 109 to 2.11 ± 0.23 × 109/g d.w.s at the rice tillering stage and from 2.19 ± 0.36 × 109 to 1.78 ± 0.23 × 109/g d.w.s at the rice anthesis stage, but these differences were not statistically significant. A similar trend for the Methanosaeta 16 s rRNA gene was found between the two rice growth stages (Fig. 1b). However, significant decreases were observed for Methanosaeta (p < 0.05) in response to elevated ground-level O3 from 7.77 ± 0.78 × 108 to 5.09 ± 0.76 × 108/g d.w.s (decreasing by 34.5%) at the rice tillering stage and from 6.23 ± 0.70 × 108 to 3.51 ± 0.94 × 108/g d.w.s (decreasing by 43.7%) at the rice anthesis stage.

Methanogenic activities were also measured for each soil sample. After a 28-day incubation, CH4 concentrations in the headspace reached their maximum levels and ranged from 12.94 ± 1.59 to 9.58 ± 1.17 μmol CH4/g d.w.s for the soil samples under ambient O3 and from 0.62 ± 0.33 to 4.90 ± 1.93 μmol CH4/g d.w.s for the soil samples under elevated ground-level O3 in the incubated flooded condition (Fig. 1c). Under ambient O3, the methanogenic activities were significantly higher at the rice tillering stage than those in the rice anthesis stage (p < 0.05). By contrast, the methanogenic activities of soil samples under elevated ground-level O3 were significantly higher at the anthesis stage (p < 0.05). At both rice growth stages, however, the methanogenic activities under elevated ground-level O3 were always significantly lower (p < 0.05) than those under ambient O3, with values ranging from 4.62 ± 0.57 × 102 to 3.42 ± 0.42 × 102 μmol CH4/g d.w.s per day under ambient O3 and from 1.75 ± 0.69 × 102 to 0.22 ± 0.12 × 102 nmol CH4/g d.w.s per day under elevated ground-level O3.

Taxonomic distribution of methanogenic archaea in flooded paddy soils

A total of 87,688 sequences were obtained (Table S1). Of these sequences, 91.3% were affiliated with methanogenic archaea. Pyrosequencing revealed that the paddy methanogenic archaeal community was dominated by two classes, namely Methanomicrobia (81.1%) and Methanobacteria (10.2%). With higher resolution, we found that Methanosarcinales were most abundant (50.5%), followed by Methanomicrobiales (17.6%) and Methanocellales (13.0%), at the order level. At the family level, the dominant methanogenic archaea were found to be aceticlastic groups, including Methanosaetaceae (32.0%) and Methanosarcinaceae (14.1%), followed by Methanocellaceae (Rice cluster I) (13.0%) and Methanobacteriaceae (10.2%). At the genus level, the dominant methanogenic groups were Methanosaeta (32.0%), Methanosarcina (14.0%), Methanobacterium (10.1%) and Methanocella (7.0%) (Fig. 2).

Figure 2
figure 2

The 100% stacked column chart of relative abundances of the dominant methanogenic genera derived from 16 s rRNA genes in all paddy soils combined and in each soil sample.

*The value of each genus percentage is the mean of triplicates in Table S1.

Full size image

The taxonomic distribution further allowed us to track overall shifts in the structure of the paddy methanogenic archaeal community in response to elevated ground-level O3 (Fig. 2). Elevated ground-level O3 significantly decreased the relative abundance of the most dominant genus, Methanosaeta, from 39.1% to 29.6% at the rice tillering stage (p < 0.05) and from 31.4% to 23.5% at the rice anthesis stage (p < 0.05). In addition, examining several other dominant genera revealed that the percentage of Methanocella was also significantly decreased (p < 0.05) and the percentages of Methanosarcina and Methanoregulaceae were potentially decreased under elevated ground-level O3 at the rice tillering stage. By contrast, at the rice anthesis stage, relative abundances of the genus Methanosarcina and the order Methanocellales were significantly increased in response to elevated ground-level O3 (p < 0.05).

Methanogenic archaeal diversity and richness

The phylogenetic diversity (PD) and Chao1 indices provide estimations of the microbial diversity and richness among different samples. PD and Chao1 consistently indicated that elevated ground-level O3 significantly (p < 0.05) decreased both the diversity (from 37.8 ± 0.9 to 31.3 ± 2.5) and the maximum richness (from 1,891 ± 111 to 1,160 ± 302) of the paddy methanogenic archaeal community at the rice tillering stage (Table 1). Consistently, DGGE fingerprinting analysis found decreases in the intensity of several DGGE bands, including bands 14 and 18, in response to elevated ground-level O3 (Fig. S1). At the rice anthesis stage, elevated ground-level O3 had no influence on paddy methanogenic community diversity and richness.

Table 1 Paddy methanogenic archaeal phylogenetic diversity and Chao1 indices under different ground-level O3 concentrations at the two rice growth stages
Full size table

Shifts in methanogenic archaeal assemblages

Variations in the paddy methanogenic archaeal community of different samples were statistically evaluated using a non-metric multidimensional scaling (NMDS) plot of the weighted pairwise UniFrac community distances (Fig. 3). Significant shifts in the assemblage of these functional guilds were observed between elevated ground-level O3 and ambient O3, as well as between the two rice growth stages (p < 0.05) and were confirmed by the ANOSIM results (Table S2). The distances between different samples along the NMDS axis 1 further imply that elevated ground-level O3 concentration has an even greater influence than the rice growth stage. Furthermore, compared to ambient O3 conditions, it could be inferred that elevated ground-level O3 had the opposite influences on methanogenic archaeal community composition at the rice tillering and anthesis stages.

Figure 3
figure 3

Methanogenic archaeal community compositional structure in soils as indicated by a non-metric multi-dimensional scaling plot of the weighted pairwise UniFrac community distances between different soil samples.

Full size image

Changed methanogenic archaeal OTUs

To demonstrate the effects of elevated ground-level O3 on paddy methanogenic community composition, a Venn diagram and response ratios were generated to identify the changes in overall and taxa-specific OTUs. Using a subset of 3,500 sequences per sample, a Venn diagram was constructed that calculated the overlap among methanogenic assemblages in different samples. We found that 115 OTUs were shared by all the soils and respectively accounted for 33.2%, 30.7%, 31.3% and 31.8% of OTUs under elevated ground-level O3 or ambient O3 at the rice tillering and anthesis stages. Furthermore, unique OTUs for each soil sample were as follows: 114 OTUs (32.9%), 107 OTUs (28.6%), 113 OTUs (30.7%) and 90 OTUs (24.9%) (Fig. 4). At the rice tillering stage, 177 OTUs were shared and accounted for 51.2% and 47.3% of total OTUs under elevated ground-level O3 and ambient O3, respectively. At the rice anthesis stage, 196 OTUs were shared by the elevated ground-level O3 (53.3%) and the ambient O3 (54.1%). By contrast, the ambient O3 at the two rice growth stages shared 208 OTUs accounting for 55.6% (tillering) and 57.5% (anthesis) of total OTUs.

Figure 4
figure 4

Venn diagrams comparing the OTU memberships among different soil samples.

The OTU size in parenthesis under each sample name was calculated from the subset of 3,500 sequences per soil sample. The percentage of unique OTUs was reported in parenthesis.

Full size image

Response ratios were calculated based on the sequence size of each genus (Fig. 5). The 95% confidence interval (CI) at the rice tillering stage ranged from −0.01 to −0.17 and did not overlap with 0, which indicates that elevated ground-level O3 brings a significant negative influence on the methanogenic archaeal community composition at the rice tillering stage (p < 0.05). Compared to ambient O3, a total of 12 genera were significantly (p < 0.05) decreased and 6 genera were significantly (p < 0.05) increased under elevated ground-level O3 at the rice tillering stage. The dominant genera, namely Methanosaeta, Methanosarcina and Methanocella, significantly decreased under elevated ground-level O3 (p < 0.05). At the rice anthesis stage, the 95% CI, ranging from 0.19 to 0.05, revealed a significant positive effect of elevated ground-level O3 on methanogenic archaeal community composition (p < 0.05): under elevated ground-level O3, 12 genera were significantly increased and 7 genera were significantly decreased (p < 0.05). For example, Methanosarcina, Methanocella and Methanocellaceae clones isolated from paddy soil were significantly increased by elevated ground-level O3 (p < 0.05). The results of the Venn diagram and response ratios are consistent with the general trends of the taxonomic distributions of paddy methanogenic archaea (Fig. 2), PD index, Chao1 index (Table 1) and NMDS plot (Fig. 3). In summary, elevated ground-level O3 influenced the phylogenetic composition of paddy methanogenic archaeal community and significantly decreased their diversity at the rice tillering stage (p < 0.05).

Figure 5
figure 5

Significant changes in genera under elevated ground-level O3 according to the response ratio method at a 95% confidence interval.

Full size image

Discussion

The community structure of paddy methanogens has been previously studied25,27,28. To display a high resolution of the paddy methanogenic archaeal community composition, 454 pyrosequencing technology was utilized in the present investigation. We obtained a total of 87,688 sequences, of which an overwhelming proportion was highly related to methanogens. These sequences provided detailed information on the methanogenic archaeal community structure in paddy soil and revealed its potential shift in response to elevated ground-level O3. Such information extends the current knowledge of paddy methanogenic archaeal communities, which is derived from analyses of traditional genetic fingerprinting, clone library and culture-dependent assays.

Detailed information on the paddy methanogenic community

In general, the paddy methanogenic archaeal community is dominated by two aceticlastic groups of Methanosaetaceae (32.0%) and Methanosarcinaceae (18.5%) followed by hydrogenotrophic Methanocellaceae (Rice cluster I) (13.0%) (Fig. 2 and Table S1). This information supports the findings from denaturing gradient gel electrophoresis (DGGE) fingerprinting based on the biomarker of either the 16 s rRNA gene25 or the functional gene mcrA28. Furthermore, the present data provided the respective quantitative percentage in the community. Among low-molecular-weight organic acids, acetate is the most abundant in paddy soil29 and its concentration can exceed 10 mM in anoxic rice soil30. An abundant acetate content can readily stimulate the metabolism of aceticlastic methanogenic archaea, such as Methanosaeta and Methanosarcina. Consequently, approximately two-thirds of biogenic methane is derived from the methyl group of acetate by aceticlastic methanogens31. Moreover, Methanosaeta may even be the principal methane producer on Earth32. Ranked third, the family Methanocellaceae (Rice cluster I), which was previously recognized as the uncultured archaeal group Rice Cluster I, also plays a key role in methane production. RNA-based stable isotope probing analysis strongly suggests that the members of this family are the most active for metabolizing rice root exudates and that they contribute greatly to paddy methane production33. The percentage of Methanocellaceae (Rice cluster I) observed in the present investigation (13.0%) was substantially lower than the 32.0% reported by Conrad et al.34 but higher than the 7.0% reported in data from a rice field in Fukuoka, Japan28. This variation might result from different atmospheric temperatures. Conrad et al.35 found that higher temperatures can increase the percentage of Methanocellaceae (Rice cluster I) in paddy methanogenic archaeal communities. The latitude of Hainan Island, China in the report of Conrad et al.34 is lower than that of the location of the present experiment. The mean annual temperature of Hainan Island (26°C) is higher than ours (16°C). Thus, the percentage of Methanocellaceae (Rice cluster I) in the present investigation is dramatically lower than the percentage reported in their study. Similarly, the present data are higher than that (7.0%) of data from a rice field in Fukuoka, Japan28. Such differences should cause the ecological significance of acetoclastic methanogenic archaea in paddy CH4 formation to be reconsidered, possibly requiring further examination.

Elevated ground-level O3 significantly influences the paddy methanogenic community at the rice tillering stage

As initially hypothesized, the paddy methanogenic community was negatively influenced by elevated ground-level O3, especially at the rice tillering stage. Due to the indirect influence of elevated ground-level O3 on soil microorganisms via plants, responses of methanogenic archaeal community were studied at the rice tillering and anthesis stages. It is well known that at the rice tillering stage, paddy soil has the highest methanogenic activity, leading to a seasonal peak of CH4 flux36 because rice root activity reaches its maximum at this stage. However, we found that at the rice tillering stage, elevated ground-level O3 significantly decreased paddy methanogenic activity (p < 0.05) (Fig. 1c), phylogenetic diversity and richness (Table 1) and potentially the overall abundance of methanogens (Fig. 1b). Correspondingly, the community composition was shifted under elevated ground-level O3 (Figs. 2, 3 and 5). Samples under elevated ground-level O3 had the largest number of unique OTUs at the rice tillering stage (Fig. 4). One possible mechanism could be that elevated ground-level O3 decreases the availability of carbon sources for methanogens. Rai et al.37 have reported high inhibition in root biomass under elevated O3 due to phytotoxicity. Lower amounts of organic matter partitioning to roots, in turn significantly decreases rice root activity38. Correspondingly, elevated ground-level O3 reduces carbon inputs from plants, such as root exudates15. Jones et al.39 reported a reduction of up to 55% in DOC contents under elevated O3 concentrations. We also consistently found significant decreases in DOC concentrations under elevated ground-level O3 (p < 0.05) (Fig. 1a). A decrease in carbon sources, which are the fuel for microorganism metabolism, changes both the structural and functional aspects of soil biodiversity40,41. In a previous investigation, we found that anoxygenic phototrophic purple bacteria negatively respond to elevated ground-level O3 due to the decrease in carbon bioavailability21. Similarly, a negative effect on paddy methanogenic archaeal communities was found in the present investigation.

The negative response of the paddy methanogenic community resulted from the responses of dominant groups; this observation was more obvious for several dominant methanogenic groups. At the rice tillering stage, both the relative and absolute abundances of aceticlastic Methanosaeta were significantly decreased (p < 0.05) (Figs. 1b and 2), as was the response ratio (Fig. 5). These findings are supported by the DGGE fingerprinting profile (Fig. S1) and the phylogenetic identification (Fig. S2). Acetate is one of the most abundant low-molecular-weight organic acids in paddy soil29 that is derived both from fermentation42 and from root exudation43. Therefore, inevitably, the acetate content was significantly decreased under elevated ground-level O3 (p < 0.05) (Fig. 1a). In rice fields, acetate levels increase to micromolar levels after sulfate consumption44. When acetate reaches micromolar levels, aceticlastic methanogens can use acetate for methanogenesis and dynamically maintain acetate at the micromolar level. The decrease of the dynamic equilibrium of the acetate content due to elevated ground-level O3 in this investigation could significantly and negatively influence aceticlastic methanogenesis and aceticlastic Methanosaeta (p < 0.05), although the exact mechanism remains unknown. Methanosaeta might be outcompeted under these negative conditions by other aceticlastic taxa because it is slow growing45. Similar phenomena were observed for aceticlastic Methanosarcina (Fig. 2, Table S1 and Fig. 5). Interestingly, the extent of Methanosaeta responses was greater than that of Methanosarcina. A possible explanation is that Methanosaeta is a specialist that uses only acetate, unlike Methanosarcina, which prefers methylated compounds, such as methanol and methylamines, to acetate32. In addition to aceticlastic methanogens, members of the Methanocellaceae family, which are the most metabolically active methanogens for rice root exudates33, were also significantly decreased (p < 0.05) under elevated ground-level O3 at the rice tillering stage (Fig. 2, Table S1 and Fig. 5). As mentioned above, these methanogenic groups play important roles in CH4 production in paddy soil. Consequently, the overall methanogenic activity at the rice tillering stage was completely inhibited by elevated ground-level O3 (Fig. 1c). Furthermore, rhizodeposition is regarded as the main origin of CH4 produced in rice fields46. The decrease in DOC could also reduce the methanogenic substrate for methanogens. Collectively, adverse effects on the diversity and functional behavior of methanogens at the tillering stage would lead to a decrease in cumulative CH4 emissions during the entire rice growing season, which could be the underlying microbial mechanism that explains the observations of Zheng et al.9 and Bhatia et al.10.

Negative influences of elevated ground-level O3 were still observed at the rice anthesis stage

The negative influence of elevated ground-level O3 on the paddy methanogenic archaeal community was alleviated at the rice anthesis stage. For example, a positive effect of elevated ground-level O3 was observed on the community composition at the rice anthesis stage (Fig. 5): the OTU size increased to 368 (Fig. 4) and 12 methanogenic genera were significantly increased (p < 0.05) (Fig. 5). Moreover, the methanogenic activity at the rice anthesis stage was significantly higher than that at the rice tillering stage (p < 0.05) (Fig. 1c). All of these phenomena may have been due to the increase in the relative abundances of the genus Methanosarcina and the order Methanocellales (Table S1, Figs. 2 and 5). Both Methanosarcina and Methanocella can tolerate micro-oxygen47 and Methanocella is potentially the most oxygen-tolerant methanogen48. For similar reasons, the relative abundances of these genera were increased at the rice anthesis stage after mid-season drainage. However, the abundances of overall methanogens and Methanosaeta continued to decrease under elevated ground-level O3 at the rice anthesis stage (Fig. 1b). Moreover, 7 genera were significantly decreased (p < 0.05) (Fig. 5) and the overall methanogenic activity was still significantly lower (p < 0.05) (Fig. 1c) under elevated ground-level O3. Therefore, elevated ground-level O3 still potentially negatively influences the paddy methanogenic archaeal community at the rice anthesis stage.

The observed methanogenic archaeal responses imply that a continuously elevated ground-level O3 would retard the rate of the increase in methane emissions and further influence global climate change. In view of the cumulative effect of global climate change49, all of these results indicate that a continuously elevated ground-level O3 would negatively influence paddy methanogenic archaeal diversity and the ecological behaviors of the community and finally reduce CH4 production9,10. Furthermore, methanotrophs, which utilize CH4 as their sole source of carbon and energy, would also be influenced. Specifically, with the accumulation of O3 stress, the decrease in the availability of methane in paddy soil could make high-affinity methanotrophs more active to assimilate atmospheric CH4 for survival and would increase the sink of atmospheric CH4. Global climate change encompasses multiple aspects, such as elevated atmospheric CO2, elevated ground-level O3 and global warming. Both elevated atmospheric CO2 and higher soil temperatures can significantly increase paddy CH4 emission5. Therefore, continuously elevated ground-level O3 would retard the rate of the increase in methane emissions, which are implicated in global climate change. Consequently, decreased paddy CH4 emission would be expected to mitigate global warming potential (GWP) of greenhouse gases because CH4 has approximately 25 times higher GWP than CO2. Moreover, the reduction in CH4 emission may slow the increase in tropospheric O3 concentration because CH4 is one of the major O3 precursors.

Methods

Site description

As previously described by Feng et al.21, the FACE-O3 system was established in Jiangdu County, Jiangsu Province, China (119°42′0″E and 32°35′5″N). This site has been in continuous cultivation for over 1,000 years with a rice-wheat rotation. The soil is classified as stagnic anthrosol. The relevant soil properties are as follows: 9.2% sand (1–0.05 mm), 65.7% silt (0.05–0.001 mm), 25.1% clay (<0.001 mm), 1.2 g/cm3 bulk density, 15.0 g/kg soil organic C (SOC), 1.59 g/kg total N, 1.23 g/kg total P, 10.4 mg/kg available P and pH 6.8. The station sits in the subtropical climatic zone with a mean annual precipitation of 900–1,000 mm, mean annual temperature of 16°C, an average daily integral radiation of 12.3 MJ/m2, a total annual sunshine time of more than 2,000 h and a frost-free period of more than 230 days.

FACE-O3 system description

The FACE-O3 project began on 1 July 2007 and has been carried out for 4 years. The FACE-O3 system has been described in detail by Tang et al.50. Briefly, this system has six plots, of which three were under elevated O3 (hereinafter called FACE-O3) and three were under ambient O3 (hereinafter called Ambient). Each plot had an area of approximately 240 m2. The target O3 concentration in FACE-O3 was 50% higher than that in Ambient and was around the average of 60 ppb during the entire rice-growing season. Any one of the FACE-O3 was separated from the other plots by at least 70 m to avoid cross-contamination. Each FACE-O3 had a dedicated system for measurement and control of O3 concentration with an O3 concentration analyzer (Thermo Electron 49i, Thermo Scientific Co., USA) and a data logger-controller (Campbell CR10X, Campbell Scientific Co., USA). The O3 concentration analyzers were calibrated against a transfer standard (Thermo Electron 49i-PS, Thermo Scientific Co., USA) on a monthly basis. The O3 fumigation began at 9:00 a.m. and continued until sunset but was discontinued when the following occurred: (1) leaves were wet, in which case a leaf-wetness sensor was used to shut down the fumigation, or (2) ambient O3 was lower than 20 ppb. When the target O3 was higher than 250 ppb, the set-point O3 was fixed at 250 ppb to prevent the plants from being exposed to extraordinarily high O3. In the Ambient plots, plants were grown under ambient O3 without the ring.

Outline of cropping systems

The rice (Oryza sativa L.) and wheat (Triticum aestivum L. cv. Yangmai 14) fields were managed in a rice-wheat rotation system, which is widely distributed in the subtropical area in China. In this system, wheat is sown in early November and harvested in late May or early June of the following year. Rice seeds are sown on a nursery bed in mid-May and the seedlings are transplanted to fields in mid-June and harvested in mid- to late October. Urea and compound fertilizer are applied as N fertilizer. For wheat cultivation, N fertilizer is applied three times as follows: 50% of nitrogen is basally applied in early November and the remaining nitrogen is top-dressed in mid-February (10%) and mid-April (40%). For rice cultivation, N fertilizer is applied three times as follows: 36% of nitrogen is basally applied in mid-June and the remaining nitrogen is top-dressed in mid-June (24%) and mid-July (40%). Detailed information on rice and wheat cultivation has been described previously51,52.

Sample collection and determination

On 18 July 2010 (rice tillering stage beginning with the appearance of the first tiller and continuing up to the maximum tiller number) and 10 September 2010 (rice anthesis stage beginning with the emergence of the first anthers from the uppermost spikelets on each panicle and continuing for approximately 15 days), soils from three FACE-O3 plots and three ambient plots were collected. In total, there were 12 soil samples (2 stages × 2 different O3 concentration × 3 replicates). Each soil sample was collected at a depth of 0 to 5 cm at six points and fully mixed. Aboveground plant materials, roots and stones were removed before homogenizing the soil samples. Samples for molecular studies were maintained at −40°C until further use. Sub-samples for the microcosm incubation assay were pooled and maintained at 4°C without sieving. Soil dissolved organic C (DOC) was extracted by adding 50 ml of 0.5 M K2SO4 to 10 g of soil, shaking for 1 h and then vacuum filtering through glass fiber filters (Fisher G4, 1.2 μm pore space). The filtrate was stored at −20°C until further analysis. DOC was determined using a TOC-TN analyzer (Skalar, Netherlands). Acetate in the filtrate was analyzed by high-performance liquid chromatography (Dionex, USA).

Soil DNA extraction

For each soil, genomic DNA was extracted from the same amount of moist soil (0.5 g) on the day after sampling using a FastDNA® SPIN Kit for soil (MP Biomedicals, Santa Ana, CA) according to the manufacturer's instructions. The extracted soil DNA was dissolved in 50 μl of TE buffer, quantified using a spectrophotometer and stored at −20°C until further use. A total of 12 DNA samples were used for qPCR and bar-coded pyrosequencing analyses.

PCR and preparation of the amplicon libraries for 454 pyrosequencing

For each soil sample, the following primer set was used to amplify approximately 280 bp of methanogenic archaeal 16 s rRNA gene fragments for sequencing on the 454 GS-FLX pyrosequencing platform: 1106F (TTWAGTCAGGCAACGAGC) and 1378R (TGTGCAAGGAGCAGGGAC)23. The oligonucleotide sequences included the 454 Life Science A or B sequencing adapters (19 bp) fused to the 7-bp bar-coded primer set as follows: Primer B (GCCTTGCCAGCCCGCTCAG) + barcode + forward primer; and Primer A (GCCTCCCTCGCGCCATCAG) + reversed primer. PCR was carried out in 50-μl reaction mixtures with the following components: 4 μl (initial 2.5 mM each) of deoxynucleoside triphosphates, 2 μl (initial 10 μM each) of forward and reverse primers, 2 U of Taq DNA polymerase with 0.4 μl (TaKaRa, Japan) and 1 μl of template containing approximately 50 ng of genomic community DNA as a template. Thirty-five cycles (95°C for 45 s, 56°C for 45 s and 72°C for 60 s) were performed with a final extension at 72°C for 7 min. Triplicate reaction mixtures per sample were pooled, purified using the QIAquick PCR Purification kit (QIAGEN) and quantified using a NanoDrop ND-1000 (Thermo Scientific, USA). The bar-coded PCR products from all samples were normalized in equimolar amounts before pyrosequencing using Genome Sequencer FLX System platform (454 Life Science Branford, CT, USA).

Processing of pyrosequencing data

The methanogenic archaeal 16 s rRNA gene data were processed using the Quantitative Insights Into Microbial Ecology (QIIME) 1.4.0-dev pipeline (53; http://www.qiime.org) with the default parameters unless otherwise noted. In brief, sequences were quality trimmed (> 25 quality score and 200 bp in length) and assigned to soil samples based on unique 7-bp barcodes. Sequences were denoised54 and then binned into OTUs using a 97% identity threshold; the most abundant sequence from each OTU was selected as a representative sequence for that OTU. Taxonomy was assigned to methanogenic archaeal OTUs against a subset of the Silva 104 database (http://www.arb-silva.de/download/archive/qiime/). OTU representative sequences were aligned using PyNAST and chimera sequences were removed through QIIME. A phylogenetic tree was then constructed using FastTree255 to support phylogenetic diversity calculations.

The richness of phylotypes was calculated to compare community-level methanogenic archaeal diversity at a single level of taxonomic resolution. We also estimated phylogenetic diversity using Faith's index56, which provides an integrated index of the phylogenetic breadth across taxonomic levels. In this diversity analysis, 87,688 methanogenic archaeal sequences that passed QIIME's quality filtering were included. We obtained between 3,551 and 15,387 sequences per sample for all soil samples (mean = 7,307 and median = 6,801) (Table S1). Because an even depth of sampling is required for beta diversity calculations, we reduced the datasets to the lowest number available to correct for differences in survey effort between samples. Namely, we calculated both diversity metrics using a randomly selected subset of 3,500 sequences per soil sample. This approach allows us to compare general diversity patterns among sites even though it is highly unlikely that we surveyed the full extent of diversity in each community57. The weighted pairwise UniFrac distances58 were calculated for community comparisons using QIIME and were visualized using non-metric multidimensional scaling plots as implemented in PRIMER v659. ANOSIM analyses were conducted using PRIMER v659.

Real-time quantitative PCR

For each soil, real-time quantitative PCR (qPCR) was used to determine the abundances of the methanogenic archaeal 16 s rRNA gene (1106F and 1378R primer set) and key methanogens, namely the Methanosaeta 16 s rRNA gene (1AF (TCYGKTTGATCCYGSCRGAG)60 and Mst 746r (TCCCTTGCCGTCAGGTC)61 primer set). Briefly, the target gene copy number was quantified by qPCR analysis with a C1000Tm Thermal Cycler equipped with a CFX96Tm Real-Time system (Bio-Rad, USA). To generate a qPCR standard curve, a single clone containing the correct insert was grown in Luria-Bertani medium and then plasmid DNA was extracted, purified and quantified. A 10-fold dilution series of the linear plasmid DNA was made to generate a standard curve covering seven orders of magnitude from 102 to 108 copies of template per assay. Assays were set up using the SYBR Premix Ex Taq™ Kit (TaKaRa) with a 25-μl reaction mixture containing 12.5 μl of SYBR® Premix Ex Taq™, 1 μl of the primer set (initial concentration of 10 μM each), 0.5 μl of BSA at 20 mg/ml initial concentration and 1.0 μl of template containing approximately 2–9 ng of DNA. Blanks were run with water as the template instead of soil DNA extract. Specific amplification of the target gene was confirmed by agarose gel electrophoresis of qPCR amplicons showing the expected band of PCR amplicons sizes and a melting curve analysis resulting in a single peak. qPCR was performed in triplicate and amplification efficiencies of 97.4–104% were obtained with R2 values of 0.966–0.977. The final methanogenic archaeal and Methanosaeta 16 s rRNA gene quantities were obtained by calibrating against the total DNA concentrations extracted and soil water content.

Methanogenic activity in paddy field soil

Soil samples, which were collected from each FACE-O3 and ambient plots, were used for the measurement of methanogenic activity in paddy field soils23. Briefly, the soils were passed through a 2-mm mesh sieve and 10 cm3 of wet soils under flooded conditions were transferred into a 120-ml serum bottle sealed with a butyl rubber septum and an aluminum cap. The headspace in the bottle was flushed with N2. The preparations with triplicates for one soil sample were incubated statically at 25°C in a dark room and CH4 production was determined periodically by a gas chromatograph (Varian 3380) with FID. Methanogenic activity was calculated from the concentration during the period with a linear increase of methane.

Statistical analysis

Statistical procedures were performed with the SPSS 13.0 software package for Windows. Data were expressed as the means with standard deviation (SD) and the letters above the error bar indicate significant differences between the results of the different samples. Mean separation was assessed by Tukey's multiple range test. Differences at p < 0.05 were considered statistically significant. Based on the subset of 3,500 sequences per soil sample, a Venn diagram of OTUs among different soil samples was plotted using the R package software (Version 2.12.1) and the response ratios of methanogenic archaeal genera under different ground-level O3 concentrations were analyzed following the statistical method of Luo et al.62.