CH4 production and oxidation processes in a boreal fen ecosystem

CH4production and oxidation processes in a boreal fen ecosystem after long-term water table drawdownK I M Y R J A¨L A¨*,T E R O T U O M I V I R T A w,H E L I J U O T T O N E N*,A N U L I I N A P U T K I N E N*w,K A I S A L A P P I*,E E VA-S T I I N A T U I T T I L A z,T I M O P E N T T I L A¨w,K A R I M I N K K I N E N z,J U K K A L A I N E§,K R I S T A P E L T O N I E M I w and H A N N U F R I T Z E w*MEM-Group,Department of Biosciences,PO Box56,University of Helsinki,00014Helsinki,Finland,w Finnish Forest Research Institute,Southern Unit,PO Box18,01301Vantaa,Finland,z Peatland Ecology Group,Department of Forest Sciences,PO Box27, University of Helsinki,00014Helsinki,Finland,§Finnish Forest Research Institute,Western Unit,Kaironiementie54,39700 Parkano,FinlandAbstractFens,which extend over vast areas in the Northern hemisphere,are sources of the greenhouse gas CH4.Climate change scenarios predict a lowering water table(WT)in mires.To study the effect of WT drawdown on CH4dynamics in a fen ecosystem,we took advantage of a WT drawdown gradient near a ground water extraction plant.Methane fluxes and CH4production and oxidation potentials were related to microbial communities responsible for the processes in four mire locations(wet,semiwet,semidry,and dry).Principal component analyses performed on the vegetation,pH,CH4,and WT results clearly separated the four sampling locations in the gradient.Long-term lowering of WT was associated with decreased coverage of Sphagnum and aerenchymatic plants,decreased CH4field emissions and CH4production potential.Based on mcrA terminal restriction fragment length polymorphism the methanogen community structure correlated best with the methane production and coverage of aerenchymatic plants along the gradient.Methanosarcinaceae and Methanocellales were found at the pristine wet end of the gradient, whereas the Fen cluster characterized the dry end.The methane-oxidizing bacterial community consisted exclusively of Methylocystis bacteria,but interestingly offive different alleles(T,S,R,M,and O)of the particulate methane monooxygenase marker gene pmoA.The M allele was dominant in the wet locations,and the occurrence of alleles O,S, and T increased with drainage.The occurrence of the R allele that characterized the upper peat layer correlated with CH4oxidation potential.These results advance our understanding of mire dynamics after long-term WT drawdown and of the microbiological bases of methane emissions from mires.Keywords:DGGE,greenhouse gas,methanogens,methanotrophs,microbial communities,peatlands,T-RFLPReceived4May2010and accepted2June2010IntroductionIn the Northern hemisphere mires extend over vast areas.They are typically water-saturated ecosystems, where a part of the plant litter production avoids aerobic decomposition and accumulates as peat.The high water table(WT)creates anaerobic microbial ha-bitats where methanogenesis occurs(Rydin&Jeglum, 2006).Hence boreal mires act as methane(CH4)sources into the atmosphere.Owing to climate change,a grow-ing attention has been devoted to these areas,where over one-third of the global terrestrial carbon is stored (Gorham,1991).The rise of earth temperatures may have profound effects on the microbial life of these ecosystems and thus on microbially mediated gas emis-sions(IPCC,2007).Methane,which is a25times more powerful green-house gas than carbon dioxide,is produced microbio-logically in thefinal stage of anaerobic degradation of organic matter(Whalen,2005).Methane is solely produced by methanogens that are classified into Eur-yarchaeota.Methanogens are highly diverse constitut-ing of four classes and six orders of archaea(Garcia et al.,2000;Sakai et al.,2008).Methane-producing archaeal communities have been described from var-ious mire ecosystems(Edwards et al.,1998;Basiliko et al.,2003;Galand et al.,2003,2005a,b;Juottonen et al.,2008;Putkinen et al.,2009)including a range of nutrient conditions(Juottonen et al.,2005)and mire successional stages(Merila¨et al.,2006).Methanogens of Methanosarcinales,Methanomicrobiales,Methano-bacteriales,and Methanocellales(Rice cluster I)have frequently been detected.The CH4emissions are,however,dependent on the activity and abundance of methane-oxidizing bacteriaCorrespondence:Kim Yrja¨la¨,tel.13580919159220,fax13580919159262,e-mail:kim.yrjala@helsinki.fiGlobal Change Biology(2011)17,1311–1320,doi:10.1111/j.1365-2486.2010.02290.xr2010Blackwell Publishing Ltd1311(MOB)that are able to oxidize CH4to CO2in aerobic conditions(Hanson&Hanson,1996).The aerobic layer of peat is roughly defined by the WT position(Bubier& Moore,1994).Low emissions in mires can occur when the CH4never reaches the atmosphere because it is consumed by MOB in the uppermost aerobic layer (Whalen,2005).In this way the CH4flux to the atmo-sphere is the sum of the functionally opposite actions of archaea and bacteria involved in the CH4cycle.The highest MOB activity in mires is observed just above the WT,where CH4and oxygen levels are adequate for CH4 oxidation(Sundh et al.,1994).While estimates of the oxidation efficiency of the produced CH4in different mires vary considerably,estimates of between20%in Carex dominated fens(Popp et al.,2000)and78%in Sphagnum dominated bogs(Yavitt et al.,1988)have been given.The latter phenomena is explained by the fact that MOB inhabit Sphagnum species providing the plant CO2through CH4oxidation under wet conditions (Raghoebarsing et al.,2005;Larmola et al.,2010).MOB are traditionally divided into two taxonomic groups,type I and II,within the Proteobacteria.Type I MOB includes genera like Methylobacter and Methylo-microbium,which belong to the Gammaproteobacteria. The type II MOB,Methylocystis,Methylosinus,Methylo-cella,and Methylocapsa belong to the Alphaproteobac-teria(Hanson&Hanson,1996;Dedysh,2009).These types differ in their carbon assimilation pathway, phylogenetic affiliation,and intracellular membrane arrangement.The methanogenic and methanotrophic communities in Finnish mires have been studied by molecular genetic methods.Methanogens were analyzed by using both functional,mcrA(Galand et al.,2002),and phylogenetic 16S rRNA marker genes(Galand et al.,2003).MOB have been studied by functional(pmoA)marker gene analysis in a pristine and drained fen and bog within a mire complex(Jaatinen et al.,2005).Thefingerprinting of Finnish fen MOB was improved by modifications of the pmoA reverse primer for DGGE analysis protocols(Tuo-mivirta et al.,2009).Methanogens and methanotrophs in mires have,however,rarely been analyzed within the same study(McDonald et al.,1999).The CH4turnover in mires is highly dependent on the ecohydrological conditions.In climate change scenarios where the raise of temperature is31C,the summertime WT in northern mires is expected to drop10–20cm (Gorham,1991;Roulet et al.,1992;Gitay et al.,2001).A WT drawdown of that magnitude(15cm)has occurred in the northern Suonukkasuo fen where a groundwater extraction plant has generated a WT gradient that has gradually changed part of the wet fen into a forested peatland(Jaatinen et al.,2008).Such a persistent change in the WT influences the plant community structure (Weltzin et al.,2000,2003)and may lead to complete replacement with species better adapted to the new conditions(Laine et al.,1995;Strack et al.,2006).The WT gradient was studied combining methane gasflux measurements,environmental data and mea-surements of microbial communities responsible for CH4turnover.This gradient study prompted the for-mulation of hypotheses concerning CH4turnover in relation to detected microbial communities,namely,as the WT drawdown affects the plant cover and stimulate tree growth,the CH4emissions are reduced in the resulting dryer conditions.Peat depth is an important factor determining microbial community structure,both that of methanogens and of methanotrophs.Thus there should be specific types of methane-producing archaea and MOB that are favored by the WT drawdown. Materials and methodsExperimental site and samplingThe study site,Suonukkasuo,is a mesotrophic pine fen located in Rovaniemi,northern Finland(661280N,251510E)within the aapa mire zone.Mesotrophic pine fens(RhSR in the Finnish mire site-type nomenclature of Laine&Vasander(2005))are typically sites where wet lawns and drier hummocks form a mosaic-like vegetation pattern.A groundwater extraction plant on an esker bordering the mire downstream has affected the WT at the study site since1959,resulting in a clear hydrological gradient where the pristine wet fen(location S4)becomes semiwet(location S3),semidry(location S2), andfinally a dry pine dominated mesotrophic peatland forest (MtkgII,location S1).The average long-term WT values for the ground-frost free periods of the years2001–2003and2005expressed as distances from soil surface were26,21,15,and9cm for the locations S1–S4,respectively.The differences in WT between locations were statistically significant,except between locations S1and S2(P50.066).Thefirst-and third-quartile ranges of the WTs were21–30,13–30,10–21,and5–13cm for the locations S1–S4, respectively.The momentary WT at the time of sampling in 2006was32,50,47,and23cm,respectively.The hydrological gradient is reflected in vegetation composition as reported in Jaatinen et al.(2008).Sampling was conducted in September2006.Intact triplicate peat cores were taken from the four locations using a corer(size4Â6cm)to a depth of70cm.Depth samples were prepared by dividing peat cores at10cm intervals.The sam-ples between0and50cm depth were used for estimation of the methanotroph activity and community analyses and the samples between depths20–70cm for the respective analyses for methanogens.For convenience we call the sam-ples between0and10cm as0cm,between10and20cm as10cm and so on.To measure the peat dry weight subsam-ples were kept at1051C for over12h and subsequently weighted.1312K.Y R J A¨L A¨et al.r2010Blackwell Publishing Ltd,Global Change Biology,17,1311–1320Field CH4measurementsGas efflux measurement plots(n53)were installed at each location in June2001(Dahlin et al.,2003).Collars made of metal tubes(diameter31.5cm)were inserted15–30cm deep in peat.An aluminum groove was attached on the top of each collar.A closed chamber(height30cm)equipped with a fan was placed on the groovefilled with water before sampling to seal the chamber.At each sampling occasion,four successive gas samples of30mL were taken from the chamber with plastic syringes during a35-min measurement period at10-min intervals(5,15,25,35min).The gas samples were then taken to laboratory and analyzed for CH4concentrations within24h from sampling by a gas chromatograph(GC) equipped with aflame ionization(FI)detector.CH4fluxes were calculated from a linear change of CH4concentrations during the sampling period of35min.Obvious ebullition events were deleted from the data.The methane emission data was also recalculated to CO2equivalents of radiative forcing to enable rough estimates of carbon balance in WT draw down at the Suonukkasuo mire gradient.Measurements of potential CH4production and oxidationMeasurements of potential CH4production with incubation at 151C were carried out as described in Juottonen et al.(2008). Briefly within few hours from sampling,15mL of peat was transferred to120mL infusion bottles containing30mL of N2-flushed distilled H2O,and incubated at1151C.Four times during the incubation,subsamples were taken from the head-space for analysis of the methane concentration with a GC. Potenial CH4oxidation was carried out as in Jaatinen et al. (2005).Briefly peat samples were kept at41C for a week and transferred to151C for2days before the gas measurements with a GC.The potential CH4oxidation rate was measured at 151C using6mL subsamples of fresh peat in120mL infusion bottles spiked with4.13m mol CH4.Gas samples were repeat-edly taken over time for5days and in the case of fast oxidation rates the experiment was repeated with gas measurements taken over30h.See Jaatinen et al.(2005)for the GC conditions. Analysis of methanogen communities by mcrA-TRFLPMethanogen communities were analyzed by a terminal restric-tion fragment length polymorphism(T-RFLP)approach that, based on in silico analysis of existing mcrA sequences from northern peatlands and previous work(Merila¨et al.,2006),is well-suited for analysis of peatland methanogens.DNA for methanogen community analysis was extracted from0.25g of wet mass peat with PowerSoil DNA Isolation Kit(MoBio Laboratories,Carlsbad,CA,USA).Methyl coenzyme M re-ductase(mcrA)gene fragments were amplified with the ML primers of Luton et al.(2002).The reactions(50m L)contained 1ÂPCR buffer with2mM MgCl2(Biotools,Madrid,Spain), 200m M dNTPs,0.4m M of both primers,1U of DNA polymerase (Biotools),and1–2m L of undiluted DNA extract as template. The reaction conditions were initial denaturation(941C,3min) followed by31cycles of941C45s,521C1min,721C1min30s,and afinal elongation(721C,7min).In PCR for T-RFLP,theforward primer was50-labelled with6-carboxyfluorescein(FAM).For T-RFLP,approximately25–50ng of PCR products wasdigested with3U of Msp I(Fermentas,Vilnius,Lithuania)at371C overnight.The T-RFLP was carried out as described inJuottonen et al.(2008).The range of fragment lengths includedin the analysis was75–500bp.Minimum peak height thresholdwas100fluorescence units.Results are presented based onrelative peak area.For identification of terminal restrictionfragments(T-RFs),three clone libraries(S420cm;S230cm;S150cm)were constructed as in Juottonen et al.(2008).Insertsfrom clone colonies were reamplified with mcrA primers(33–54clone per library)and screened by RFLP with Msp I.FromRFLP groups with45clones,two to three clones from eachgroup were sequenced,and from groups with o5clones,onemember was sequenced as in Juottonen et al.(2008).T-RF wereidentified based on in silico digestion of the sequences andT-RFLP analysis of clones.Investigation of MOB by DGGE analysis of the pmoA marker geneDNA for MOB community analysis was extracted from ca.0.5g wet mass peat using the FastDNA kit for soil(MPBiomedicals,Solon,OH,USA)according to Yeates&Gillings(1998)modified as in Tuomivirta et al.(2009).A region ofsubunit a of particulate methane monooxygenase(pmoA)wasPCR targeted with the A189f/A621r with a GC-clamp attachedto the reverse primer(see Tuomivirta et al.,2009for details).Also broad specificity A189f/A682r(Holmes et al.,1995)primers were applied for reference(Tuomivirta et al.,2009).Fingerprinting of the MOB diversity was performed bydenaturing gradient gel electrophoresis(DGGE)described inJaatinen et al.(2005).The DGGE method was preferred since ithas a resolution power to detect minute changes in pmo A(Jaatinen et al.,2005;Tuomivirta et al.,2009;Larmola et al.,2010).The DGGE gel photographs were screened for thepresence(1)or absence(0)of pmoA bands using the ALPHAI-MAGER2.1program of the AlphaDigiDoc gel documentation system(Alpha Innotech Corp.,CA,USA).A binary matrix wasgenerated with this data using only the positions of success-fully sequenced bands.Single DGGE bands of interest wereexcised from the gel,purified and sequenced as described inTuomivirta et al.(2009)and Larmola et al.(2010). Sequence analysis and phylogenymcrA and pmoA sequences were compared with databasesequences with BLAST(Altschul et al.,1997)analysis of theNational Center for Biotechnology Information(NCBI).Deduced amino acid sequences of mcrA together with selectedreference sequences were aligned with ClustalW(Larkin et al.,2007).Suitable evolutionary model was selected with ProtTest(Abascal et al.,2005).Maximum likelihood trees were con-structed with PhyML(Guindon&Gascuel,2003)with modelLG1G1F.Bootstrap values were generated from100repli-cates in PhyML.Pairwise distance calculations of nucleotideC H4P R OD U C T I O N A N D O X I D A T I O N P R O CE S S E S1313 r2010Blackwell Publishing Ltd,Global Change Biology,17,1311–1320pmoA sequences were analyzed with MEGA4.0(Tamura et al., 2007).The mcrA and pmoA sequences have been submitted to the EMBL database under accession nos.FN564006–FN564029 (mcrA)and GQ279342–GQ279346(pmoA).Statistical evaluationMethanefluxes between locations were compared by one-way analysis of variance(ANOVA).Significance of pairwise differ-ences was assessed by Tukey’s test.The effect of categorical variables depth,location,and their interaction on potential CH4production and oxidation were compared with general-ized linear models.Tukey’s post hoc test was applied to determine which pairs of means differ significantly.Because location and depth had significant interactions,ANOVA fol-lowed by Tukey’s test was performed separately for each depth.Level of significance in all statistical analyses was P50.05.We applied principal component analysis(PCA)to compare the four locations based on their environmental variables,i.e., long-term and momentary WT,CH4field emission and oxida-tion and production potential,coverage of Sphagnum and aerenchymatic plants,and pH.To examine variation in metha-nogen and MOB communities,wefirst applied detrended correspondence analysis(DCA).Based on the DCA showing rather small compositional variation,(ter Braak&Prentice, 1988)we applied redundancy analysis(RDA)to test which environmental variables best explained variation in methano-gen and MOB communities.The significance of the above listed explanatory factors was assessed using Monte Carlo permutation test.The multivariate analyses were carried out with CANOCO for Windows(ter Braak&Smilauer,2002). ResultsMethane emissionsIn the years(2001–2004)preceding the sampling for microbial community analysis,the CH4emissions from location S4,were substantial,averaging 73mg mÀ2dayÀ1(S1).As a result of WT drawdown, the emissions were much smaller in the locations S3 (22mg mÀ2dayÀ1),S2(0.9mg mÀ2dayÀ1),and S1 (1.9mg mÀ2dayÀ1).Similarly to long-term WT the emis-sions differed significantly between all pairs of locations (P50.001–0.004)except for locations S2and S1.The emissions were recalculated to g CO2equivalents of radiative forcing for the years2002–2004(Fig.1a)to enable the comparison with soil respiration in the WT gradient for the years2002–2004(Fig.1b).Methane production and oxidation potentialThe wettest location S4had the highest CH4production potential at depths from20to50cm(P50.001–0.036, Fig.2)in correspondence with theflux rates recorded in earlier years.The S4production was43.2nmol g dry weightÀ1dayÀ1compared with0.4(S3,P50.005),4.0 (S2,P50.010),and0.4(S1,P50.005)at the depth of 20cm.The production potential of the three other dryer locations was much lower and did not differ between the locations at any depth.The laboratory measure-ments showed that location and depth significantly affected the CH4production potential through the WT gradient(P o0.001).The highest methane oxidation potential was ob-served in location S3,561.2nmol g dry weightÀ1dayÀ1, but the sampling depth influenced the oxidation poten-tial(P50.004)in addition to location(P50.033)(Fig.2). In the top layer the potential was significantly higher in S3than in S2,62.2nmol g dry weightÀ1dayÀ1(P50.042) and S1,28.9nmol g dry weightÀ1dayÀ1(P50.031)and second highest in S4,177.7nmol g dry weightÀ1dayÀ1. The dry S1location had the lowest potential oxidation rate.The oxidation potential did not differ between locations in deeper peat layers.Environmental factors characterizing the WT gradient Decrease in long-term WT,coverage of Sphagnum and aerenchymatic plants,low CH4field emissions together with CH4production potential characterized the dry end of the PC1(Fig.3).Decreasing momentary WT and CH4oxidation potential together with increasing pH characterized the peat layers along PC2.PCA per-formed on the vegetation,potential CH4production and oxidation,CH4field emissions,pH,and long-term and momentary(date of sampling)WT depth clearly separated the four sampling locations along thefirst principal axis(PC1in S1).With the provided measure-ments the second principal axis(PC2)separated all samples according to depth.Microbial communities involved in the CH4cycle along the gradientT-RFLPfingerprinting of mcrA representing methano-gens from the WT gradient resulted in altogether12 T-RFs(Table1).In the RDA T-RFs468and471were combined since the large size of T-RFs prohibited reli-able distinction between them.Thefirst RDA axis separated the four locations when the methanogen communities at different depths were analyzed together with environmental factors(Fig.4a).The top layers were most variable through the gradient,but deeper down they became more alike.The methanogen com-munity structure showed highest correlation with methane production and coverage of aerenchymatic plants along the gradient(RD1in S2).The second1314K.Y R J A¨L A¨et al.r2010Blackwell Publishing Ltd,Global Change Biology,17,1311–1320RDA axis showed that the pH influenced the commu-nity structure according to depth.The locations S2and S1at the dry end of the WT gradient sampled from three depths showed lower diversity with six T-RFs than the wet end.The higher CH 4production potential locations showed higher methanogen diversity with 8–10T-RFs.The occurrence of the T-RFs 115bp (Methanocellales),272bp (putatively Methanosarcinaceae),and 298bp (Methanosarcinaceae)correlated most strongly with the wet location S4(Fig.4b).In addition to identified methanogens,T-RF 193bp affiliated with a cluster with no cultured representatives in the phylogenetic tree (S3).The T-RF 278bp,representing Methanomicro-biales-associated Fen cluster,was one of two T-RFs prone to the dry end of the gradient despite of overall decrease in methanogen groups along the gradient.In analysis of MOB communities,the broad specifi-city pmoA primer set A189f/A682r did not produce correct amplicons (Tuomivirta et al .,2009).Five different pmoA alleles (T,S,R,M,and O),all Methylocystis spp according to sequence comparisons,could,however,beFig.1(a)Average methane emission 2002–2004from the four locations forming a water table gradient,from wet fen location S4to dry forested location S1.The CH 4emissions presented in the text are scaled to season and given here as g CO 2equivalents taking into account the IPCC conversion factor of 25for methane.(b)For comparison the average 2002–2004field CO 2emissions (soil respiration)of the same locations are presented (see Jaatinen et al .,2008for measurement details).CH 4 oxidation (nmol g dw –1 day –1)CH 4 production (nmol g dw –1 day –1)Fig.2(a)The potential methane oxidation,(b)the potential methane production in the four locations S4–S1.The depths are given as distance from the surface of the mires.C H 4P R OD U C T I O N A N D O X I D A T I O N P R O CE S S E S 1315r 2010Blackwell Publishing Ltd,Global Change Biology ,17,1311–1320isolated from the gradient by PCR-DGGE and sequen-cing with the A189f/A621r primer set (Table S1).RDA(Fig.5a)of the pmoA alleles showed that their occur-rence was best correlated with CH 4oxidation potential,pH,and vegetation.Similarly to methanogen commu-nities (Fig.4a)the locations from wet to dry wereseparated along the first RDA axis,and the second RDA axis characterized the depth of the peat (Fig.5b).The two wettest locations had similar MOB commu-nities in their whole profile (Fig.5b).The deep layers (30–40cm)of the locations S4,S3,and S2were also characterized by a similar MOB.The MOB community of the dry location S1differed from the other sites.The M allele,dominating the wet end of the gradient (Fig.5a)was identical to a previously detected Finnish fen sequence FJ930087(Tuomivirta et al .,2009)and was most different from the other alleles with 11–12point mutations.The R allele inhabiting the wetter part of the gradient was identical to sequences FJ930090and FJ930091of a Finnish fen (Tuomivirta et al .,2009),AY309209,a North-American fen (S.C.Nold,personal communication),and the sequence GQ121280(Larmola et al .,2010)occurring in methane oxidizing community within Sphagnum mosses.Drying correlated with in-creasing abundances of alleles O,S,and T.The R allele dominated the pristine location and the upper layer of the peat (Fig.5a).DiscussionThe study of methane turnover along a peatland WT gradient showed that drying caused a dramatic reduc-tion in CH 4emissions and in potential CH 4production in agreement with our hypotheses.Despite more oxic conditions the methane oxidation also decreased sub-PCA axis 1P C A a x i s 2Fig.3Principal component analysis (PCA)of environmental factors,emissions,vegetation,water table (WT)for locations S4–S1coupled with CH 4production and oxidation.The direction of the arrow describes the correlation between measured factors.The length of the arrow gives the degree of explanation of each variable in the ordination.The first two principal components together explained 55.3%of the variation.Eigenvalues of the first and second axis were 0.366and 0.186,respectively.Table 1Identification and occurrence of terminal restriction fragments (T-RFs)detected in analysis of methanogens in the mires S4–S1forming a water table gradientT-RF (bp)Methanogen groupOccurrence of T-RF (%)S4S3S2S1105No sequences 32––115Methanocellales 64––193Mx cluster512–243Methanosarcinaceae –––1251Methanocellales131876272Methanosarcinaceae *14651278Fen cluster (Methanomicrobiales)31417069293Fen cluster (Methanomicrobiales)1615314298Methanosarcinaceae 2–––468Methanobacteriaceae9w 13w 13w 8w 471Fen cluster (Methanomicrobiales)489Methanosarcinaceae1–––WetSemiwetSemidryDry*No sequences were detected in this study but identification is based on sequences from other mire sites (H.Juottonen,H.Fritze &K.Yrja ¨la ¨,unpublished results).w Includes T-RFs 468and 471bp.The occurrence is given as a relative abundance of the T-RF phylotype across all sampling depths.T-RFs were identified based on sequenced mcr A clones.1316K.Y R J A¨L A ¨et al .r 2010Blackwell Publishing Ltd,Global Change Biology ,17,1311–1320stantially along the gradient.In agreement with the hypothesis drying decreased the diversity of the metha-nogen communities and a group of archaea,the Fen cluster methanogens,was characteristic of dryer condi-tions.As hypothesized the depth appeared to be an important factor for the community structure of metha-nogens and of methanotrophs.The depth effect in WT drawdown brought about a peculiar shift in methano-troph populations showing specific pmoA allele types of Methylocystis bacteria distributed to specific locations in the gradient.The dry end of the WT gradient was best described by low coverage of aerenchymatic plants and low CH 4emissions as shown by statistical analysis.Low emis-sions were coupled with low potential CH 4production and oxidation.In agreement with our results from the gradient,experimental field studies in Canada (Strack et al .,2004)and in Ireland (Laine et al .,2009)have shown decrease in CH 4emission as a result of lowered WT.The CH 4oxidation potential in turn correlated positively with the WT of the time of sampling and the coverage of Sphagnum .Very recently methane oxidation was re-ported from 23different Sphagnum species originatingfrom a Finnish boreal mire site and the activity was related to a Methylocystis signature similar to the R allele of this study (Larmola et al .,2010).The correlation of the oxidation potential with Sphagnum coverage might be a direct regulation instead of a covariation with more pristine conditions.This phenomenon explains the correlation of the oxidation potential with Sphagnum coverage.The methanogen community structure according to T-RFLP analysis changed by decrease in diversity and decreasing production along the gradient.WT gradients have mainly been studied in meso-and microcosms (see however Talbot et al .,2010),and as far as we know reports from natural gradients combined with microbial community analysis of the CH 4cycle are lacking.Sampling depth has been shown to affect the metha-nogen community structure in mires (Galand et al .,2002).The RDA analysis of the methanogen commu-nities together with environmental factors showed ni-cely how the locations differ at the surface,but become alike in the deepest sampling layer where the condi-tions in the gradient apparently are more similar (Fig.4a).The discrepancy of location S2to the general trend may well be explained by larger pH variation at this location (results not shown)and lowest emissions telling that this location is the most disturbed one in the gradient.The methanogen group most strongly associated with the pristine end of the gradient was Methanosarcina-ceae.In the drier locations,Methanosarcinaceae T-RFs were less abundant or absent.This finding was surpris-ing considering that many members Methanosarcina-ceae are versatile methanogens that are able to use several different substrates for growth (Garcia et al .,2000)and possess enzymes for detoxification of oxygen,which could favor their occurrence in the dry end of the gradient,which is the opposite of what was observed.Only two methanogenic groups were typical to the dry end of the gradient,and they were affiliated with the Fen cluster which has commonly been found in Finnish mires (Juottonen et al .,2008).One member of this cluster,Methanoregula boonei ,was isolated from an acidic bog and has the lowest pH optimum known for a methanogen (Bra ¨uer et al .,2006).It is tempting to say that the Fen cluster/Methanoregula-type hydrogeno-trophic methanogens are well adapted to the conditions of boreal fens characterized by changing WT and low pH.All detected MOBs were by pmoA sequence analysis found to be of the Methylocystis genus,but differed by minute changes,point mutations,in the gene (Support-ing Information Table S1).This suggests a phylogeneti-cally very narrow MOB population of this fen.RDA showed that the occurrence of the five alleles wasRDA axis 1R D A a x i s 2RDA axis 1R D A a x i s 2Fig.4(a)Redundancy analysis (RDA)of methanogenic term-inal restriction fragments (T-RFs)representing methanogen com-munities at different depths and environmental factors.(b)RDA of T-RFs detected in the water table gradient.The length of arrow describes how well the T-RF phylotype correlates with locations in the gradient.C H 4P R OD U C T I O N A N D O X I D A T I O N P R O CE S S E S 1317r 2010Blackwell Publishing Ltd,Global Change Biology ,17,1311–1320。