Responses of the methanogenic pathway and fraction of CH4 oxidization in a flooded paddy soil to rice planting

Yuting YANGGuangbin ZHANGJing MAQiong HUANGHaiyang YUKaifu SONGXiaoli ZHUXi MIAO and Hua XU

1State Key Laboratory of Soil and Sustainable Agriculture,Institute of Soil Science,Chinese Academy of Sciences,Nanjing 210008(China)

2University of Chinese Academy of Sciences,Beijing 100049(China)

(Received April 6,2020;revised April 22,2020)

ABSTRACT Rice planting(RP)is significant to methane(CH4)emissions from paddy fields,but its effect on the relative contribution of the acetoclastic methanogenesis to total CH4 production (Fac) and the fraction of CH4 oxidized (Fox) is poorly understood. To quantify the responses of the Fac and Fox to RP, we investigated CH4 fluxes,CH4 production and oxidation potentials,dissolved CH4 concentrations,and their stable carbon isotopes in a flooded paddy soil.The mcrA and pmoA gene copies were also determined by quantitative polymerase chain reaction(qPCR).Compared with the unplanted soil(control,CK),the seasonal CH4 emissions from the planted soil were significantly enhanced,13.6 times,resulting in large decreases in the CH4 concentrations in the soil solution.This indicated that much more CH4 was released into the atmosphere by the RP than was stored in the soils.Acetoclastic methanogenesis became more important from the tillering stage(TS)to the ripening stage(RS)for the CK,with Fac values increased from 17%—20%to 46%—55%.With RP,the Fac values were enhanced by 10%—20%,and it significantly increased the copy numbers of the mcrA gene at the four rice stages(TS,booting stage(BS),grain-filling stage(GS),and RS).Furthermore,the effect of the RP on the abundance of the mcrA gene was highly concurrent with the effect on the Fac values.At the TS,the Fox values at the soil-water interface were around 50%—75%for the CK,being 15%—20%lower than those of the RP in the rhizosphere.It increased to 65%—100%at the GS,but was reduced by 20%—30%after the RP.These differences might be because the copy numbers of the pmoA gene were significantly raised at the TS while lowered at the GS by the RP.This was further demonstrated by the strong correlations between the effect of the RP on the abundance of the pmoA gene and the effect on the Fox values.These findings suggest that RP markedly impacts on the abundances of the mcrA and pmoA genes,affecting the pathway of CH4 production and the fraction of CH4 oxidization,respectively.

Key Words: stable carbon isotope,acetoclastic methanogenesis,CH4 production,CH4 oxidation,mcrA gene,pmoA gene,rhizosphere,soil-water interface

INTRODUCTION

Methane(CH4)is an important greenhouse gas,and its concentration in the atmosphere has increased by 157%since the industrial revolution(WMO,2017).Globally,50%—65%of total CH4emissions come from human activities(IPCC,2013). Rice fields are regarded as a major anthropogenic source of atmospheric CH4,and the global total CH4emissions have reached 36 Tg year-1(IPCC, 2013). The role of rice planting(RP)is critical in the production,oxidation and transport of CH4and finally in the CH4emissions from paddy fields (Jiaet al., 2001; Caiet al., 2009). A large number of observations have shown that RP significantly increases CH4emissions (Tyleret al., 1997; Bileket al.,1999;Jiaet al.,2001,2006).To the best of our knowledge,however, there are few systematic reports available about the effect of RP on the pathway of CH4production and the fraction of CH4oxidized,as well as on the related microbes.

Two major methanogenic pathways are found in paddy soils:acetate fermentation and CO2/H2reduction,making up two-thirds and one-third of the total CH4production,respectively(Conrad,1999).The relative contributions of acetate-(Fac)and CO2/H2-dependent(1-Fac)methanogenesis to the total CH4production change greatly during the rice-growing season and are significantly influenced by many factors,such as soil temperature,water management,and straw amendment(Chidthaisonget al.,1999;Feyet al.,2004;Caiet al.,2009).For instance,it is reported that the 1-Facvalues increase with temperature from about 15%at 10°C to 40%at 37°C(Feyet al.,2004),theFacvalues are higher in the continuously flooded paddy fields than those in the intermittently irrigated fields(Zhanget al.,2012,2013a),and rice straw application greatly increases theFacvalues(Zhanget al.,2013b).Related methanogens are also strongly influenced by the factors mentioned above(Fey and Conrad,2000;Maet al.,2012;Jiet al.,2018).However,the effect of the RP on theFacvalues and methanogens remains unclear.

Before its release into the atmosphere, a fraction of the produced CH4will be oxidized by the methanotrophic bacteria, mainly in the rhizosphere while partially at the soil-water interface, because of the positive effect of the RP;when no rice is planted,however,the CH4oxidization occurs at the soil-water interface(Le Mer and Roger,2001;Caiet al., 2009). Under RP conditions, previous studies have reported the fraction of CH4oxidized(Fox),both in the rhizosphere and at the soil-water interface,and found that the difference inFoxvalues between them was significant(Zhanget al.,2015b,2016).In addition,theFoxvalues are greatly altered by rice growth and significantly affected by water management,fertilizer application,and rice cultivar(Bileket al.,1999;Krüger and Frenzel,2003;Conrad and Klose,2005;Zhanget al.,2012).Nevertheless,the differences in theFoxvalues and the corresponding methanotrophs between rice planted soil and control soils,with no rice plants(CK)are little understood.

Stable carbon isotope techniques have been widely used to study the mechanisms of CH4emissions in rice-based cropping systems (Tyleret al., 1997; Bileket al., 1999;Krügeret al.,2002;Krüger and Frenzel,2003;Zhanget al.,2016,2017).The values ofFacandFoxcan be quantified simultaneously,based on measurements of the stable carbon isotopic signatures(δ13C)of CH4production,oxidation,and emissions from the paddy fields.Sinceδ13CH4from acetoclastic methanogenesis(-60‰to-50‰)is usually more positive than that from hydrogenotrophic methanogenesis(as negative as-110‰to-90‰),12CH4is oxidized faster than13CH4by methanotrophs, thus leaving more13CH4remained (Whiticar, 1999; Zhanget al., 2013b). Pot and incubation experiments were conducted to estimate the effect of RP on CH4production and oxidation and the abundances ofmcrAandpmoAgenes,in particular on the values ofFacandFox.We hypothesized that RP might increaseFac,Fox,and the abundances of methanogens and methanotrophs.This study is expected to provide new insights into how RP affects the processes of CH4production and oxidation.

MATERIALS AND METHODS

Pot experiment

Soil samples (0—20 cm) were collected in December 2012 from a rice field in Dongxi Town, Jianyang City,Sichuan Province,China(30°23′N,104°34′E).The field is traditionally cultivated with a single rice crop and then kept in flooded fallow after the rice harvest.The soil is derived from purple shale.The main soil properties are as follows:pH(1:2.5 soil to H2O ratio),7.7;total C,25 g kg-1;total N,2.0 g kg-1;andδ13C of soil organic carbon,-26.7‰.Soil samples were air dried and sieved(5-mm mesh size)to remove stones. Fourteen plastic pots(diameter, 25 cm;height,35 cm)were filled with 16 kg of sieved soil each,on May 27,2013.All pots were placed on the terrace outside the greenhouse of the Institute of Soil Science,Chinese Academy Science,and then kept flooded with tap water to maintain a 3—5-cm water layer over the soil surface throughout the rice-growing season.

Two treatments, control (CK) and RP, were laid out with seven replicates,each of which was divided into two groups:three pots were prepared for observing CH4fluxes and the other four pots for measuring CH4production and oxidation potentials.The CK pots contained only flooded paddy soils, with no rice. For the RP pots, 34-d-old rice seedlings(Oryza sativaL.,Huajing 3)were transplanted on June 29 and harvested on November 15.Each pot contained three seedlings.Both treatments were applied with the same fertilizers:urea N,calcium superphosphate P,and potassium chloride K, which were applied on June 28 at 1.9, 0.85,and 0.25 g pot-1as basal fertilizers,respectively.Urea was applied again at a rate of 1.0 N g pot-1, as tillering and panicle fertilizers on July 16 and August 18,respectively.

Gas sampling

The CH4fluxes were monitored using the static chamber technique(Caiet al.,2009).The chambers(30 cm×30 cm×100 cm)were made of plexiglass.To measure the CH4fluxes,the chambers were placed once every 3—5 d,and four gas samples from each chamber were collected using 20-mL vacuum vials,at 15-min intervals between 9:00 and 11:00 in the morning of each sampling day.To determine theδ13C of the emitted CH4(δ13CH4(emission)),samples were taken at 10—30-d intervals,at the same time as the flux observations.Two gas samples per pot were collected using 500-mL bags(aluminum foil compound membrane, Delin gas packing Co.,Ltd.,Dalian,China)with a small battery-driven pump(Zhanget al., 2011a). Soil redox potentials (Eh) and soil temperature at a depth of 10 cm were measured by a portable oxidation-reduction meter(Yokogawa Meter and Instruments Corporation,Japan),after the CH4fluxes.

Water sampling

At the four main rice growth stages(tillering(TS),booting(BS),grain-filling(GS),and ripening(RS)),solutions were collected from 0, 5, and 10 cm soil depths using the Rhizon soil moisture sampler(10 Rhizon SMSMOM,Eijkelkamp Agrisearch Equipment,Giesbeek,The Netherlands)(Zhanget al.,2011a).Three samplers were installed into each pot used for the flux measurements on June 29.The samples(about 5 mL)were first extracted using 20-mL vacuum vials to flush and purge the sampler before sampling.Then,about 10 mL of the solution was drawn into another vial for further analysis. Subsequently, all sampling vials were equilibrated by filling with pure N2gas.After vigorous shaking by hand,gas samples were taken from the headspace of the vials for the analyses of the CH4concentrations and the correspondingδ13CH4.

Soil sampling

Soil (0—20 cm) was destructively sampled from one pot of each treatment at the main rice growth stages(four times in total). For the RP, soil samples were collected by extracting the plants and stripping the soils from the rice roots.Meanwhile,soil cores(0—20 cm)were collected directly for the CK using a stainless-steel corer(Zhanget al.,2011b).These samples were packed into sterile plastic bags,immediately placed on ice,and transported back to the lab within 1 h after destructive soil sampling for the following incubations. Another soil sample was collected from the mixture of each treatment and maintained at-80°C for DNA extraction.

Soil incubations

The soil samples in each treatment were first mixed well,and then six subsamples from the mixture,about 50 g each(dry weight),were taken and transferred promptly into six 250-mL Erlenmeyer flasks separately.Samples in the flasks were prepared into slurries with N2-flushed deionized sterile water at a soil:water ratio (volume/volume) of 1:1. Three flasks were sealed with N2for anaerobic incubation and the other flasks with air headspace were sealed directly for aerobic incubation(Zhanget al.,2011b).All the flasks were sealed with rubber stoppers fitted with a silicon septum that allowed for sampling of the headspace gas.

The potential CH4production was measured under anaerobic incubation.The flasks were flushed with N2consecutively six times through double-ended needles,connecting a vacuum pump to purge the air in the flasks of the residual CH4and O2.They were then incubated for approximately 50 h at 25°C under dark conditions.The concentrations of CH4and CO2in the gas samples were analyzed about 1 and 50 h after heavily shaking the flasks by hand. Simultaneously,theδ13C values of the produced CH4and CO2were determined.The CH4production rates were calculated using the linear regression of the CH4increase with the incubation time.

The CH4oxidation potential was determined under aerobic incubation.Each flask was injected with pure CH4to a concentration of approximately 10 000 μL L-1and incubated in the dark at 25°C with a vibration velocity of 120 r min-1.The CH4depletion was measured by sampling the headspace gas in the flask after vigorous shaking for subsequent analysis.The first sample was collected 30 min after pure CH4was injected.Samples(100 μL each time)were then taken at 2—3 h intervals during the first 8 h of the incubation. Samples were taken the next day at 2 h intervals. The CH4oxidation rates were calculated using linear regression of the CH4depletion with the incubation time.

DNA extraction

During the rice-growing season,there were 8 soil samples collected from the experimental pots in total(4 stages× 2 treatments) and analyzed for the abundance of the methanogen and methanotroph populations.For each soil mixture, three soil subsamples (0.5 g each) were collected and the genomic DNA was extracted using a FastDNA spin kit for soil(MP Biomedicals LLC,Ohio,USA),according to the manufacturer instructions. The extracted soil DNA was dissolved in 50 μL of elution buffer,checked by electrophoresis on 1%agarose,and then quantified using a spectrophotometer(NanoDrop Technologies,Wilmington,USA)(Fanet al.,2016).

Measurements

The CH4concentrations in the gas samples were analyzed with a gas chromatograph(Shimadzu GC-12 A,Kyoto,Japan),equipped with a flame ionization detector.The continuous flow technique and a Finnigan MAT 253 isotope ratio mass spectrometer(Thermo Finnigan,Bremen,Germany) were used for the analysis of theδ13C (Caoet al.,2008;Zhanget al.,2011a).The CO2in the gas samples was directly analyzed, while the CH4in the gas samples was converted into CO2and separated primarily on a PreCon(pre-concentration device)(Thermo Finnigan, Germany).The gas was then piped into a GC equipped with a Pora PLOT Q column(Agilent,USA)(length:25 m;diameter:0.35 mm)at 25°C under 2.0×105Pa,for further separation.The separated gases were finally transferred into the mass spectrometer forδ13C determination.The reference and carrier gases used were CO2(99.999%purity and-23.7‰inδ13C of Pee Dee Belemnite(PDB))and helium(99.999%purity,20 mL min-1),respectively.Results were expressed in the delta notationvs.PDB:δ13C=103(Rsa/Rst-1)withRsaandRst=13C/12C ratios of sample(sa)and standard(st),respectively.The precision of the repeated analyses was±0.2‰when 2.02 μL L-1CH4was injected.

The numbers of the methanogenicmcrAgene copies and the methanotrophicpmoAgene copies were determined using quantitative polymerase chain reaction(qPCR).Fragments of themcrAandpmoAgenes, encoding the methyl coenzyme-M reductase and theαsubunit of the particulate methane monooxygenase,respectively,were amplified using primers according to Haleset al.(1996)and Costello and Lidstrom(1999).Real-time qPCR was performed on a CFX96 optical real-time detection system(Bio-Rad Laboratories Inc.,Hercules,USA).For a detailed description of the methods,please refer to Fanet al.(2016).

Calculations

The relative contribution of the acetate-dependent methanogenesis to the total CH4production(Fac)was calculated using the mass balance,assuming that methanogenesis was from the acetate fermentation (CH4(acetate)) and H2/CO2reduction (CH4(H2/CO2)) (Tyleret al., 1997; Bileket al.,1999):

whereδ13CH4(acetate)andδ13CH4(H2/CO2)are theδ13C of CH4produced from acetate and H2/CO2,respectively,andδ13CH4(initial)is theδ13C of CH4newly formed.

The isotope fractionation factor for the acetate(εacetate/CH4orαacetate/CH4)or H2/CO2(αCO2/CH4)producing CH4was defined by Hayes(1993):

whereδ13Cacetate-methylis theδ13C of the methyl group of acetate.Moreover,the fraction of the CH4that was oxidized(Fox)was estimated according to the methods of Tyleret al.(1997)and Bileket al.(1999):

whereδ13CH4(final)is theδ13C of the remaining CH4after it has undergone oxidization andαoxis the isotope fractionation factor associated with CH4oxidation.

Statistical analysis

Statistical analysis was conducted using the SPSS 18.0 software for Windows (SPSS Inc., Chicago, USA).Differences between the two treatments were determined through a one-way analysis of variance and the least significant difference test.Relationships were assessed using correlation analysis.Significant differences and correlations were set atP ＜0.05.

RESULTS

CH4 emissions and the corresponding δ13CH4

Relatively low CH4fluxes were observed in the CK throughout most of the rice-growth season,whereas much higher CH4fluxes were measured in the RP(Fig.1a,P ＜0.05). Seasonal cumulative CH4emission for the RP was 50.2 g m-2,13.6 times greater than that of the CK.Soil Eh ranged from-300 to 30 mV,and there was no significant difference in the seasonal means (-195v.s.-178 mV)between the two treatments(Fig.1a). Theδ13CH4(emission)first increased to-50‰at the TS,then there was a rapid decrease to-60‰, but it increased again before the rice harvest(Fig.1b).Seasonally,the CH4emissions were slightly13C-depleted for the RP(-58‰),relative to the CK(-56‰),though no significant differences were observed.

During 34—36 and 59—61 d after rice transplanting(DAT),the CH4fluxes and correspondingδ13CH4(emission)were observed simultaneously at 4-h intervals(Fig.1c—f).The CH4fluxes peaked at 13:00 and 17:00,and a much lower flux peak was observed for the RP than the CK(P ＜0.05).The lowest values ofδ13CH4(emission)(-75‰to-65‰on 34—36 DAT and-70‰on 59—61 DAT) for the CK were measured at 13:00 and 17:00.For the RP,theδ13CH4(emission)was lowest(-65‰on 34—36 DAT)mainly at 17:00.Significant negative relationships were observed between the diurnal variation of the CH4fluxes and theδ13CH4(emission)(r=-0.909 to-0.738,P ＜0.01),except for the RP on 59—61 DAT.Soil temperature was 29.9—38.3 and 26.2—35.4°C on 34—36 and 59—61 DAT,respectively(Fig.1c,e).Diurnal variations of the soil temperature were significantly correlated with the CH4fluxes(r=0.601—0.762,P ＜0.05),except for the CK on 34—36 DAT.

CH4 concentration and the corresponding δ13CH4

Compared with the CK(1.9—30.8 μmol L-1),the CH4concentration in the soil water solution at a depth of 0 cm was significantly decreased by the RP at the four rice growth stages(Fig.2a,P ＜0.05).At depths of 5 and 10 cm,the CH4concentrations were much lower with the RP (5.8—126.1 μmol L-1)than the CK(149.3—221.5 μmol L-1)at the BS,GS,and RS(Fig.2b,c,P ＜0.05).At all four rice growth stages,the CH4emissions were negatively correlated with CH4concentrations at the 5 and 10 cm depths (r=-0.722 to-0.786,P ＜0.05).

The dissolved CH4was generally13C-enriched in the two treatments, withδ13C values increasing from about-70‰to-50‰at the TS to-60‰to-40‰at the RS(Fig.2d—f).As a whole,the seasonal meanδ13C values of CH4in the soil water solution at a depth of 10 cm were between-64‰and-62‰,being relatively lower than those at the depths of both 5(—57‰to-56‰)and 0 cm(—50‰to-46‰).No significant effects were observed for the meanδ13C values of the CH4by RP.

Fig.1 Seasonal(a and b)and diurnal(c—f)variations of CH4 emission and the corresponding stable carbon isotopic signatures(δ13C)of emitted CH4(δ13CH4(emission)),soil Eh,and soil temperature during the rice-growing season.Values are means with standard deviations shown by the vertical bars(n=3).RP=rice planting;CK=control with no rice;DAT=days after rice transplanting.

CH4 production and oxidation potentials and related microbes

The potential of the CH4production ranged from 0.25 to 9.7 μg g-1d-1and from 0.09 to 17.3 μg g-1d-1for the RP and CK,respectively(Fig.3a).At the TS and BS,the CH4production potential in the CK was significantly higher than that in the RP, whereas it was substantially promoted by the RP at the GS and RS (P ＜0.05). The CH4oxidation potential for the CK ranged from 26.5 to 57.6 μg g-1d-1, but for the RP in comparison, it was significantly decreased to 13.5—29.8 μg g-1d-1,at the four rice growth stages (Fig. 3b,P ＜0.05). Taking the CH4production and oxidation potentials together,the RP,relative to the CK,significantly increased the ratio of the oxidation to production at the BS but decreased it at the GS and RS(Fig.3c).

The copy numbers of themcrAgene in the CK were relatively stable at about 1.0×109—2.0×109copies g-1dry soil,but they were significantly lower(P ＜0.05)than those of RP (about 3.5× 109—5.0× 109copies g-1dry soil) (Fig. 3d). The copy numbers of thepmoAgenes for the CK and RP were 0.18× 109—1.3× 109and 0.66×109—1.3×109copies g-1dry soil,respectively(Fig.3e).Compared with the CK,the copy numbers of thepmoAgene were significantly increased at the TS and BS but greatly decreased at the GS by the RP (P ＜0.001). Overall, the RP strongly increased the copy number ratio of thepmoAgene to themcrAgene at the BS,while they decreased it at the GS and RS,as compared to the CK(Fig.3f).Statistical analysis shows that the copy number ratios of thepmoAgene to themcrAgene were positively(r=0.748—0.761,P=0.004—0.005)correlated with the potential ratios of oxidation to production(Fig.S1,see Supplementary Material for Fig.S1).

Fig.2 CH4 concentrations(a—c)and the corresponding stable carbon isotopic signatures(δ13C)of CH4 (d—f)in soil water solutions at soil depths of 0(a and d),5(b and e),and 10(c and f)cm,respectively,at the four rice growth stages.Values are means with standard deviations shown by the vertical bars(n=3).Bars with different letters between the two treatments in each rice growth stage represent significant differences(P ＜0.05).RP=rice planting;CK=control with no rice;TS=tillering stage;BS=booting stage;GS=grain-filling stage;RS=ripening stage;DAT=days after rice transplanting.

Produced δ13CH4 and δ13CO2

The CH4anaerobically produced in the paddy soils was generally enriched in13C,except at the BS,withδ13C values that increased from-68‰to-57‰for the CK and from-63‰to-52‰for the RP(Table I).Compared to the CK,the RP always produced CH4that was13C-enriched. The produced CO2was gradually depleted in13C from the TS to GS and became13C-enriched at the RS(Table I).The RP,relative to the CK,generally produced13C-depleted CO2.The value of the approximation of the apparent fractionation(αapp) from CO2to CH4(Feyet al., 2004) was between 1.048 and 1.063 in the CK and was relatively lowered with the RP(Table I).

Pathway of CH4 production and the fraction of CH4 oxidized

TheFacvalues gradually increased from about 20%at the TS to 45%—55%at the RS for the CK and from 35%—45%to 65%—75%for the RP(Table II).Compared with the CK,the RP increased theFacvalues by about 10%—20% over the four rice growth stages.Variation in the ratios of theFacvalues with the RP to CK was similar to the change in copy number ratios of themcrAgenes(Fig.4a),and there was a significant correlation between them (Fig. 4b,r= 0.730,P=0.007).

About 45%—75%of the CH4was oxidized at the soilwater interface at the TS in the CK (Table III). TheFoxvalues increased to about 65%—100%at the GS,followed by a sharp decrease to approximately 45%—70%at the RS.For the RP,approximately 63%—94%of the produced CH4was oxidized in the rhizosphere at the TS,and then it decreased to about 40%—60%at the RS.Compared with the CK,the RP increasedFoxvalues by 15%—20%at the TS,but decreased the values by 20%—30% at the GS. The ratios of theFoxvalues with the RP to CK were strongly related to the copy number ratios of thepmoAgene over the four rice growth stages(Fig.4c,d,r=0.677,P=0.016).

Fig.3 Production and oxidation of CH4 in soil incubations and the corresponding microbes at the four rice growth stages:CH4 production potential(a),CH4 oxidation potential(b),the ratios of oxidation to production(c),the abundance of mcrA gene(d),the abundance of pmoA gene(e),and the copy number ratio of pmoA gene to mcrA gene(f).RP=rice planting;CK=control with no rice;TS=tillering stage;BS=booting stage;GS=grain-filling stage;RS=ripening stage.Values are means with standard deviations shown by the vertical bars(n=3).Bars with different letters between the two treatments at each rice growth stage represent significant differences(P ＜0.05).

TABLE I Anaerobically produced stable carbon isotopic signatures (δ13C) of CH4 and CO2 (δ13CH4 and δ13CO2, respectively) in the paddy soils and the corresponding approximation of the apparent fractionation from CO2 to CH4 (αapp)at the four rice growth stages

TABLE II Relative contributions of the acetate-dependent methanogenesis to the total CH4 production(Fac)a) at the four rice growth stages

Fig.4 Temporal variations in the ratios of mcrA gene copy numbers and Fac values(a)and pmoA gene copy numbers and Fox values with RP to CK over the four rice growth stages(c)and the corresponding relationships between them(b and d,respectively).Values are means with standard deviations shown by the vertical bars(n=3)in a and b.RP=rice planting;CK=control soil with no rice;TS=tillering stage;BS=booting stage;GS=grain-filling stage;RS=ripening stage;Fac =relative corbribution of the acetate-dependent methanogenesis to the total CH4 production;Fox =fraction of CH4 oxidized.

DISCUSSION

CH4 production,oxidation,and emission

The RP significantly increased CH4emissions by 13.6 times(Fig.1a),which were in the range of the previous field investigations(4.5—29.0 times)(Tyleret al.,1997;Bileket al.,1999).Jiaet al.(2001)found that CH4emissions were much higher with the RP,relative to the CK at the rice tillering and panicle initiation stages,although the production of the CH4was far lower. This significant enhancement was believed to be the positive effect of plant-mediated transportation on CH4fluxes from below ground to the atmosphere. In the current study, compared with the CK, far more CH4was released into the atmosphere by RP rather than stored in the soils, although the CH4production potential was substantially lower at the TS and BS(Fig.3a).Therefore,the RP decreased the CH4concentrations in soil water solution(Fig.2a—c).This could be further demonstrated bythe negative relationships between the CH4emissions and the CH4concentrations. Dannenberg and Conrad (1999)found that soil microcosms planted with rice exhibited lower dissolved CH4concentrations, but higher CH4emission fluxes, compared with the unplanted paddy microcosms.Besides, a large number of studies have shown that high CH4concentrations can enhance the growth and activity of the CH4-oxidizing bacteria (Arifet al., 1996; Cai and Mosier,2000;Xuet al.,2003).The significantly greater CH4concentrations in soil water solution for the CK(Fig.2b,c)could strengthen the oxidizing activity of the methanotrophic bacteria and the potential for CH4oxidation.To some extent,the substantially reduced CH4oxidation potential with the RP(Fig.3b)might explain the much higher CH4emissions than with the CK.

TABLE III Fraction of the CH4 oxidized(Fox)a) in the rhizosphere for rice planting(RP)and at the soil-water interface for control soil with no rice(CK)at the four rice growth stages

During the periods of 34—36 and 59—61 DAT,however,the CH4fluxes were significantly higher in the CK than the RP,at 13:00 and 17:00,leaving a contrary trend in the diurnal emissions,as compared to the seasonal emissions(Fig.1a,c,e).Based on the ratio of the daily average values of the CH4emissions to the measurement of the CH4flux at a given time,Maet al.(2007)and Zhenget al.(1998)have demonstrated the optimum sampling time for the CK(around 18:00)and RP(around 8:00—11:00 or 19:00),respectively.Considerable seasonal measurements of the CH4fluxes were carried out at around this sampling time(8:00—11:00)with or without RP worldwide(Caiet al.,2009 and references therein).In the present study,the seasonal variation of the CH4emissions for both treatments was observed between 9:00 and 11:00 in the morning of each sampling day,which was in the range of the optimum sampling time for the RP.However,this sampling for the CK was not considered to be the best, causing the measurements might not be very representative.Using the static-chamber method,there was another uncertainty,that one could not sample the gas continuously throughout the day(Caiet al., 2009; Minamikawaet al., 2015), and as a result,the greater flux peaks for CK than RP was missed at 13:00 and 17:00 in the measurement of seasonal emissions.Nevertheless,this kind of flux peak for the CK was measured occasionally,and most of the time,it was far lower than that with the RP (Fig. 1c, e). In a field experiment with straw addition, Nouchiet al. (1994) found that the CH4fluxes reached their maximum during the 12:00—18:00 period and they were significantly higher in the unvegetated plot than the vegetated plot, although the seasonal CH4emissions were lower.

Pathway of CH4 production

The production of CH4in paddy fields is mainly from acetate fermentation and CO2/H2reduction(Sugimoto and Wada, 1993; Conrad, 1999, 2005), and the relative contributions of the two methanogenic pathways change with the rice growth (Krügeret al., 2001, 2002; Zhanget al.,2012,2013a,b).In the present study,a high contribution of hydrogenotrophic methanogenesis(about 60%—80%)was initially observed at the TS,but the acetoclastic methanogenesis gradually became dominant towards the RS(at approximately 66%—77%).Previous studies in Italy and China also found that acetoclastic methanogenesis was dominant later in the season (Krügeret al., 2001, 2002; Zhanget al.,2013a,b).In addition,the RP increasedFacvalues by 10%—20%for the different values of theδ13CH4(acetate)andαCO2/CH4(Table II).For porewater,with theδ13CH4as theδ13C values of the CH4from the original production zone,Tyleret al. (1997) and Bileket al. (1999) estimated that acetoclastic methanogenesis was dominant(51%—81%)over the whole rice-growing season,and found that RP enhanced the contribution of hydrogenotrophic methanogenesis later in the season.The derivation method for theδ13C values of produced CH4may explain the differences in theFacvalues between this study and the previous two reports(Krügeret al.,2002;Zhanget al.,2016).

As the values ofδ13CH4(acetate)were not available in this study,they were assumed to be-43‰and-37‰in the calculation of theFac(Table II).It is mainly ascribed to theδ13C values of acetate that ranged from-22‰to-16‰(Krügeret al.,2002)followed by an additional isotopic shift of-21‰in the acetate-dependent methanogenesis(Gelwickset al.,1994).Theδ13CH4(acetate)values of-43‰and-37‰have been used repeatedly to estimateFac(Tyleret al.,1997;Bileket al.,1999;Krügeret al.,2002;Zhanget al.,2012,2015a).In addition,the ratio of theδ13CO2andδ13CH4from the anaerobic incubation between produced CO2and CH4can be calculated asαapp=(δ13CO2+1 000)/(δ13CH4+1 000).Theαapp,in theory,is lower than theαCO2/CH4,as reported by investigations from paddy soils (Feyet al., 2004) and freshwater lake sediments(Conradet al.,2011).In this study,theαappwas estimated to be 1.038—1.063(Table I),which is consistent with previous measurements for the paddy soils(Conradet al.,2012;Zhanget al.,2016)but substantially lower(1.057—1.085)than those from lake sediments(Conradet al., 2011). Owing to the lack of measurements and for a direct comparison with previous studies (Krügeret al.,2002),we assumed thatαCO2/CH4was 1.06 and 1.07,in its quantification of theFac.

In the current study,more reasonableFacvalues were found withαCO2/CH4=1.07,as the relative contribution of the acetoclastic methanogenesis to the total CH4production should be positive.Compared to the CK,theFacvalues were always increased by the RP(Table II).Because the RP significantly increased the abundance of themcrAgenes(Fig.3d),which possibly improved the acetoclastic methanogens and their activities(Miet al.,1993;Chawanakulet al.,2009).The finding that the effect of RP on the abundance of themcrAgenes was highly correlated with the effects on theFacvalues,may partially explain this(Fig.4a,b).With RP,more methanogenic substrates such as organic acids(e.g.,acetate and propionate)and amino acids from root exudates or products of decaying roots,might favor the acetoclastic methanogenesis(Ueckertet al., 1990; Conrad and Klose,1999;Luet al.,2000).The methanogenic archaeal community or substrate component responses to the RP are poorly known and thus require further investigation in the future.

Fraction of CH4 oxidized

Before release into the atmosphere,CH4will be oxidized in the rhizosphere and at the soil-water interface in paddy fields (Le Mer and Roger, 2001; Caiet al., 2009). Tyleret al. (1997) first reported the values ofFoxboth in the rhizosphere and at the soil-water interface in Texan paddy fields,by taking soil porewaterδ13CH4asδ13CH4(initial)and rhizosphericδ13CH4(δ13CH4(emission)—plant transport fractionation of-12.2‰±1.4‰)asδ13CH4(final)for the RP,and belowgroundδ13CH4asδ13CH4(initial)andδ13CH4(emission)asδ13CH4(final)for the CK.These findings showed that 19%—56%of the CH4was oxidized in the rhizosphere and 3%—21%at the soil-water interface,indicating that a higher proportion of CH4is consumed by the RP.In the current study,we found that theFoxvalues were as high as 47%—100%in the CK,and they were raised by approximately 15%—20%at the TS but lowered by 20%—30%at the GS with RP(Table III).

There is a large difference in the values ofδ13CH4(initial)cited as the estimatedFoxvalues between Tyleret al.(1997),and the findings in the current study.Theδ13CH4(initial)was from the anaerobically producedδ13CH4in the paddy soils(Table III),which was obviously lower than the measurements(from-56.5‰to-47.9‰)of Tyleret al.(1997).Generally,porewaterδ13CH4or belowgroundδ13CH4is a poor proxy of initially producedδ13CH4because of the isotope effects of both CH4oxidization and transportation (Chantonet al., 1997; Krügeret al., 2002). The CH4in soil water solution was gradually enriched in13C from 10 to 0 cm soil depths(Fig.2d—f),which indicated an oxidation signal in the isotopic signature of the CH4from the belowground soils to upper soils.The values ofFoxwere probably underestimated by considering the porewater or belowgroundδ13CH4asδ13CH4(initial).The lower estimations of the rhizosphericFoxby Tyleret al.(1997)than the results of the present study could demonstrate this.

The CH4oxidation fractionation factorαoxneeds to be quantified to estimate theFoxvalues.Unfortunately,theαoxvalues were not determined in this study. To the best of our knowledge,αoxvalues of 1.025—1.038 were initially measured in methanotroph-enriched cultures at a temperature of 12—35°C(Colemanet al.,1981),which was later widely applied to landfill cover soils(Liptayet al.,1998;Chanton and Liptay, 2000; Chantonet al., 2008). Moreover, the values of 1.025—1.038 have been cited by many authors in quantifyingFoxin paddy fields(Tyleret al.,1997;Bileket al., 1999; Krügeret al., 2002; Krüger and Frenzel, 2003;Zhanget al.,2012,2015b).Theαoxvalues were found to be 1.025—1.033 at 28.3°C in Chinese paddy soils(Zhanget al.,2013a).For better comparability with previous studies,we conducted an alternative calculation usingαox=1.025,1.033,and 1.038.

It was found that theFoxvalues for the CK increased greatly from the TS to the GS,which was possibly related to the same changes in both CH4oxidation potentials and abundance of thepmoAgene(Fig.3b,e).In a Texas unplanted sandy soil,Tyleret al.(1997)found a similar variation pattern ofFoxvalues,though the values(3%—21%)were much lower than our estimations(Table III).In unplanted paddy soils,the CH4produced is released into the atmosphere,mainlyviabubble ebullition and molecular diffusion, which is highly impacted by temperature.In addition,the emission of the CH4viaebullition is little oxidized at the soil-water interface, resulting in the relatively13C-depleted emitted CH4(Chanton, 2005). On 34—36 and 59—61 DAT, much higher flux peaks were observed at 13:00 and 17:00 for the CK, leaving the values ofδ13CH4(emission)far lower than those of the RP(Fig.1d,f).Diurnal variation of the CH4emissions was positively correlated with soil temperature but negatively correlated withδ13CH4(emission)(Fig.1c—f).This demonstrates that the higher the temperature,the faster the CH4emission and the lower theFox,thus causing higher CH4fluxes and the emission of isotopically lighter CH4(Zhanget al.,2015a).

However, theFoxvalues were highest at the TS and decreased gradually towards the RS with RP (Table III).A similar variation pattern was observed by Krügeret al.(2002)and Zhanget al.(2013a,b).Compared with the CK,the RP had a different effect on theFoxvalues from the TS to the GS,probably because the rice roots were very vigorous during TS, with a strong capability for oxygen exudation(Li,2012;Liu and Fu,2014).Moreover,the application of tillering N fertilizer is beneficial for the growth and activity of the methanotrophic bacteria(Krügeret al.,2002;Krüger and Frenzel,2003).Correspondingly,a large fraction of the CH4was oxidized in the rhizosphere in the current study;for the CK,the CH4was mostly transported and released into the atmosphere in the form of bubbling when the temperature was high,leading to a small amount of CH4that could be oxidized at the soil-water interface(Chanton,2005;Zhanget al.,2015a).Therefore,RP increased theFoxvalues relative to the CK at the TS(Table III).

At the GS, however, the roots became aged, causing the oxygen secretion abilities to weaken(Li,2012;Liu and Fu, 2014). The growth and activity of the methanotrophs would be limited by the depletion of N fertilizer,and then the values of the rhizosphericFoxgradually decreased(Krügeret al.,2002;Krüger and Frenzel,2003).Furthermore,the temperature was relatively low in this period.Most of the CH4in the CK is likely to be transportedviamolecular diffusion, producing a higher estimation of theFoxat the soil-water interface,than that in the rhizosphere(Table III).Overall,these changes can be indirectly demonstrated by the higher abundance of thepmoAgene at the TS and the far lower abundance at the GS with the RP as compared to the CK(Fig.3d).It was further verified by the findings in this investigation that the effect of the RP on the abundance of thepmoAgene was highly correlated with the effect onFoxvalues(Fig.4c,d).

CONCLUSIONS

The RP strongly influenced the abundance of themcrAandpmoAgenes throughout the rice growing season,which correspondingly affected the pathway of CH4production and the fraction of CH4oxidized,respectively.Compared with the CK,the RP significantly increased the contributions of the acetoclastic methanogenesis and the abundance of themcrAgene at the four rice growth stages.Statistical analysis showed that the effects of the RP on the abundance of themcrAgene were highly correlated with the contribution of acetoclastic methanogenesis.In addition,the RP significantly enhanced the fraction of CH4oxidized at the TS but decreased it at the GS.This may be related to the finding that the effect of the RP on the abundance of thepmoAgene was well related to the effect on the fraction of CH4oxidized.

CONTRIBUTION OF AUTHORS

Yuting YANG and Guangbin ZHANG contributed equally to this work.

ACKNOWLEDGEMENTS

This work was financially supported by the National Key Research and Development Program of China(No.2017YFD 0300105),the National Natural Science Foundation of China(Nos.41571232,41671241,and 41877325),and the Youth Innovation Promotion Association of Chinese Academy of Sciences(CAS;member No.2018349).We thank LucidPapers(http://lucidpapers.com)for linguistic assistance during the preparation of this paper.Sincere thanks also go to the anonymous reviewers for their helpful comments.

SUPPLEMENTARY MATERIAL

Supplementary material for this article can be found in the online version.