Denitrification Activity of a Remarkably Diverse Fen Denitrifier Community in Finnish Lapland Is N-Oxide Limited

Peatlands cover more than 30% of the Finnish land area and impact N2O fluxes. Denitrifiers release N2O as an intermediate or end product. In situ N2O emissions of a near pH neutral pristine fen soil in Finnish Lapland were marginal during gas chamber measurements. However, nitrate and ammonium fertilization significantly stimulated in situ N2O emissions. Stimulation with nitrate was stronger than with ammonium. N2O was produced and subsequently consumed in gas chambers. In unsupplemented anoxic microcosms, fen soil produced N2O only when acetylene was added to block nitrous oxide reductase, suggesting complete denitrification. Nitrate and nitrite stimulated denitrification in fen soil, and maximal reaction velocities (vmax) of nitrate or nitrite dependent denitrification where 18 and 52 nmol N2O h-1 gDW -1, respectively. N2O was below 30% of total produced N gases in fen soil when concentrations of nitrate and nitrite were <500 μM. vmax for N2O consumption was up to 36 nmol N2O h-1 gDW -1. Denitrifier diversity was assessed by analyses of narG, nirK/nirS, and nosZ (encoding nitrate-, nitrite-, and nitrous oxide reductases, respectively) by barcoded amplicon pyrosequencing. Analyses of ~14,000 quality filtered sequences indicated up to 25 species-level operational taxonomic units (OTUs), and up to 359 OTUs at 97% sequence similarity, suggesting diverse denitrifiers. Phylogenetic analyses revealed clusters distantly related to publicly available sequences, suggesting hitherto unknown denitrifiers. Representatives of species-level OTUs were affiliated with sequences of unknown soil bacteria and Actinobacterial, Alpha-, Beta-, Gamma-, and Delta-Proteobacterial sequences. Comparison of the 4 gene markers at 97% similarity indicated a higher diversity of narG than for the other gene markers based on Shannon indices and observed number of OTUs. The collective data indicate (i) a high denitrification and N2O consumption potential, and (ii) a highly diverse, nitrate limited denitrifier community associated with potential N2O fluxes in a pH-neutral fen soil.


Introduction
Northern peatlands are important players in the global carbon and nitrogen cycles, and store more than 30% of soil carbon and nitrogen even though they cover only about 3% of the terrestrial surface [1]. Greenhouse gases such as methane (CH 4 ) and nitrous oxide (N 2 O) are produced in and released from northern peatlands soils [2]. High latitude peatlands have been intensively studied with respect to their capacity to emit CH 4 due to the large amount of stored carbon in peat soils (e.g., [3][4][5]). N 2 O has a high global warming potential (approximately 300 times higher than CO 2 ), is a major ozone-depleting substance, and 6% of the greenhouse effect is attributed to N 2 O [6][7][8]. Stored organic N in certain peatlands fuels N 2 O release via coupling of ammonification, ammonia oxidation, and denitrification [9]. Thus, potential N 2 O emissions from northern peatlands are of major interest. Northern peatlands are very diverse ecosystems, including many types of pristine and managed soils. Many studies investigating N 2 O emissions from peatlands have focused on N 2 O emissions from managed peatlands, and only recently N 2 O fluxes from pristine northern peat soils have been investigated [10][11][12][13][14][15]. Pristine northern fens include significant net sources of N 2 O even though emission rates are generally low [10,13]. Negative N 2 O fluxes suggest that peatlands can act as temporary sinks for N 2 O [2,14,16,17]. Understanding of the mechanisms and regulation of N 2 O fluxes in such systems is still incomplete. N 2 O in soils is generally produced during nitrification, denitrification, or chemical processes [18,19]. Denitrification is considered to be the main source of N 2 O in water-saturated soils including peatlands [19,14]. During denitrification, nitrate or nitrite are sequentially reduced via nitric oxide (NO) and N 2 O to dinitrogen (N 2 ) [20]. The reductions are catalyzed by a set of oxidoreductases, namely nitrate reductases (encoded by narG and napA), nitrite reductases (encoded by nirK and nirS), NO reductases (endoded by norBC), and N 2 O reductases (encoded by nosZ) [20]. N 2 O or N 2 can be released into the atmosphere. The ratio of N 2 O to N 2 is determined by in situ parameters such as pH, temperature, as well as nitrate/nitrite and electron donor availability [21].
High latitude peatlands are likely to be strongly affected by increasing temperatures due to climate change [3,4,9,22]. Global warming might reduce the water table in northern peatlands and influence the amount of N 2 O released from the soil [9]. A constantly lowered water table increases N 2 O fluxes from nutrient rich peat, whereas fluxes from nutrient-poor peat remain largely unaffected [12]. Dissimilar denitrifier communities are related with dissimilar N 2 O fluxes [15]. Detailed knowledge about the microbial catalysts involved in N 2 O turnover in northern peatlands is scarce. Thus, the aim of this study was to assess denitrification in a pH neutral pristine fen. The main objectives were to (i) assess in situ N 2 O emissions of a pH-neutral fen soil, (ii) determine depth-related N 2 O production and consumption capacities of fen soil, and (iii) link differences in the denitrifier community composition to physiological differences of the denitrification in the peat soil over two depths.

Sampling site and soil parameters
Puukkosuo fen is located in northeastern Finland (66°22'38''N, 29°18'28''E) at an elevation of 220 m above sea level. The mean annual air temperature is (-0.43±0.09)°C, and mean annual precipitation approximates (772±12) mm (average of years 1966 to 2011, measured at Oulanka research station). The fen is meso-eutrophic and water saturated. Vegetation consists mainly of mosses (Sphagnum spp.) and grasses (e.g., Carex spp.). Four replicate soil cores from layers 0 to 20 cm and 20 cm to 40 cm were taken on July 28th 2010. Soil temperatures on the day of sampling were 17.2°C in surface soil and 15.1°C in deeper soil layers (below 15 cm). Samples were transported on ice to the laboratory and stored at 4°C for microcosm analyses or at -80°C for nucleic acid extractions. Microcosm experiments were conducted within 2 weeks after sampling. Nitrate, nitrite and ammonium concentrations, soil pH, soil moisture content, total carbon (TC), dissolved organic carbon (DOC) and total nitrogen (TN) were determined from pooled soil samples as described previously [23]. Permission to access and sample Puukkosuo fen was granted by Metsähallitus (www.metsa.fi) on 12th of July 2010.

Assessment of in situ gas emissions
In situ gas emissions of unfertilized soil and soil supplemented with either nitrate or ammonium were determined in closed poly(methyl methacrylate) (PMMA) chambers. Chambers were placed onto the soil surface and surrounded by metal collars, which had been inserted into the soil for a few centimeters to ensure that the chambers were gas tight. The transition between the plexiglas chamber and the metal ring was sealed with a rubber band to avoid exchange of gases from the chamber with the surrounding air. Before the installation of the gas chambers, 2l of fen pore water with 20 mM of added nitrate or ammonium was applied homogeniously onto the soil surface in 4 replicate treatments each and unsupplemented controls received pure fen pore water. Gas samples (5 ml per sampling timepoint) were taken from gas outlets and injected into gas tight evacuated containers (Exetainer, Labco Limited, High Wycombe, UK) at the start of the experiment, after 0.5, 1 and 3 hours.

Assessment of denitrification potentials in soil microcosms
Denitrification potentials of pH-neutral fen soil (0 to 20 cm and 20 to 40 cm) were assessed in unsupplemented and nitrate-, nitrite-or N 2 O-supplemented anoxic microcosms as described earlier [14,15,23]. Supplemental nitrate and nitrite ranged from 0 to 1000 μM, while supplemental N 2 O ranged from 0 to 4 μM. Acetylene blockage was used to distinguish between total N 2 O production and total denitrification as described earlier [15,24]. Incubations were conducted at 20°C in the dark. N 2 O production rates and apparent kinetic parameters [Michaelis-Menten constants (K M ) and maximum reaction velocitites (v max )] were determined as described [14]. Michaelis-Menten regressions obtained for different incubation conditions were compared using the "extra sum of squares" principle to test for significant differences between the regressions [25]. Obtained values for K M and v max were compared by t-tests.

Molecular characterisation of fen denitrifier communities
Nucleic acids were extracted from homogenized pooled fen soil of both soil layers as previously described using a bead-beating protocol [15,26]. DNA yields were 4 to 12 μg DNA per gram (fresh weight) of soil. A 260 /A 230 values approximating 0.94 to 1.56 indicated DNA with moderate to low humic acid content. The structural genes narG, nirK, nirS, and nosZ were amplified using the primer pairs narG1960f (TAY GTS GGS CAR GAR AA)/narG2650r (TTY TCR TAC CAB GTB GC) [27], F1aCu (ATC ATG GTS CTG CCG CG)/R3Cu (GCC TCG ATC AGR TTG TGG TT) [28], cd3aF (GTS AAC GTS AAG GAR ACS GG)/R3cd (GAS TTC GGR TGS GTC TTG A) [28], and nosZF (CGC TGT TCI TCG ACA GYC AG)/nosZR (ATG TGC AKI GCR TGG CAG AA) [29], respectively, and subjected to barcoded pyrosequencing as previously described [15,23]. Barcodes used to identify sequences after pyrosequencing were ACTGCG and AGTATG for 0 to 20 cm and 20 to 40 cm fen soil, respectively. Pyrosequencing was performed at the Göttingen Genomics Laboratory using the Roche GS-FLX 454/Titanium technology as previously described [15,23]. Pyrosequencing and PCR errors of the obtained reads were corrected using the AmpliconNoise pipeline [30] and sequences were clustered at species-level (i.e., for narG, nirK, nirS, and nosZ, respectively), and 97% threshold similarities using Qiime as previously described [23,31]. Species-level threshold similarities were determined from pairwise comparisons of 16S rRNA gene similarities and structural gene similiarities of cultured denitrifiers [32]. Such OTUs indicate a minimal estimate of species-level diversity, i.e., is likely to underestimate "real" species-level diversity. Phylogenetic trees with cluster representatives were constructed in MEGA 5.0 [33]. Alpha-and beta-diversity measures were calculated in Qiime from rarified OTU tables as described [23,34] to allow statistical comparison of the structural gene diversity from both soil layers. Rarified OTU tables were generated in Qiime by randomly subsampling original OTU tables 100 times at depth of 1000, 1500, 2500, and 500 sequences for narG, nirK, nirS, and nosZ, respectively. OTU representative sequences of narG, nirK, nirS, and nosZ were deposited at EMBL under accession numbers HE995549 to HE995577. Complete sequence data sets were deposited deposited in the European Nucleotide Archive (ENA) under the study accession number ERP008864.

Soil parameters
Soil moisture content of Puukkosuo fen soil was 90% in both soil layers (Table 1). Soil pH in water was 6.8 and 6.9 in 0 to 20 cm and 20 to 40 cm fen soil, respectively. Nitrate was below the detection limit of 5.8 μg g DW -1 (Table 1). Values for carbon and nitrogen contents appeared to be marginally higher in 20 to 40 cm than in 0 to 20 cm fen soil, but C/N ratios and DOC concentrations were similar in both soil layers (Table 1).

In situ gas emissions of fen soil
During gas chamber measurements, only minor amounts of N 2 O accumulated in gas chambers placed on unsupplemented fen soil on average (Fig 1). Increases of about 10 ppb in N 2 O mixing ratio were observed in two of the four replicate gas chambers, while decreases in N 2 O mixing ratio were observed in the other two replicate gas chambers (-1 to -17 ppb decrease in mixing Pore water n.a. 5 < 2.5 n.a. 5 < 10.9 n.a. 5 < 1.4 n.a. 5 n.a. 5 8,8 n.a. 5 n.a. 5 ratio). Nitrate-addition initially lead to accumulation of N 2 O in the gas chambers. However, this accumulation of N 2 O was restricted to the first 30 minutes after nitrate-addition, and initially accumulated N 2 O was subsequently consumed after 30 minutes (Fig 1). Ammonium likewise led to accumulation of N 2 O in the gas chambers, however this initial accumulation of N 2 O was slower than after nitrate-addition (Fig 1). Moreover, initially accumulated N 2 O was subsequently consumed after the first hour.

Denitrification potentials in fen soil microcosms
In anoxic microcosms, unsupplemented fen soil from both soil layers produced only minor amounts of N 2 O in the absence of acetylene, and initially produced N 2 O was subsequently consumed (Fig 2). However, N 2 O production was significantly higher in anoxic microcosms when N 2 O-reductase was blocked by acetylene (Fig 2). N 2 O mixing ratios increased from 0.04 ± 0.004 to about 35 ± 3.5 ppm within the first 94 hours in acetylene-amended microcosms with 0 to 20 cm fen soil, and the concentration of N 2 O plateaued out after the first 94 hours  (Fig 3 B). Stimulation of N 2 O production with nitrate was smaller than with nitrite, and N 2 O production in nitrate-supplemented microcosms was less than 25% of that in nitrite-supplemented microcosms (Fig 3 A and S1 Fig). In microcosms with fen soil from 0 to 20 cm, N 2 O production in acetylene-amended microcosms was in a similar magnitude for all supplemented nitrate concentrations !100 μM. N 2 O production in microcosms with fen soil from 20 to 40 cm was highest when 50 μM nitrate were supplied, and decreased with increasing nitrate concentrations, indicating that denitrifiers in fen soil were saturated at low nitrate concentrations, and were inhibited by higher nitrate concentrations (Fig 3 A). In nitrite-supplemented microcosms, N 2 O production rates increased with increasing nitrite concentrations in both soil layers (Fig 3 A). N 2 O consumption was likewise stimulated by increasing N 2 O concentrations (Fig 3 B). N 2 O production and consumption capacities were higher in 0 to 20 cm fen soil than in 20 to 40 cm fen soil.
The ratio of N 2 O to (N 2 + N 2 O) was below 30% and 40% for all supplied nitrate concentrations in microcosms with 0 to 20 cm and 20 to 40 cm fen soil, respectively (S2 Fig), indicating that more than half of the N 2 O produced from nitrate was further reduced to N 2 in fen soil. The ratio of N 2 O to (N 2 + N 2 O) was below 30% in microcosms with fen soil from 0 to 20 cm when nitrite concentrations were 100 μM or smaller and increased to about 75% for higher nitrite concentrations. In microcosms with 20 to 40 cm fen soil, the ratio of N 2 O to (N 2 + N 2 O) was between 50% and 100% for all supplied nitrite concentrations (S2 Fig  was a major product of denitrification in that soil layer when nitrite was provided as electron acceptor. Initial nitrite-dependent N 2 O production rates of fen soil microcosms amended with acetylene followed apparent Michaelis-Menten kinetics, as did nitrate-dependent N 2 O production rates of fen soil microcosms from 0 to 20 cm depth and N 2 O-dependent N 2 O consumption rates in both layers (Fig 3). The Michaelis-Menten kinetics differed significantly between the different treatments and soil layers (p 0.03 for all comparisons). Apparent maximal reaction velocities (v max ) were highest for nitrite-dependent N 2 O production, followed by N 2 O-dependent N 2 O consumption rates. v max was lowest for nitrate-dependent N 2 O production ( Table 2). v max values for nitrate and nitrite dependent N 2 O production, as well as N 2 O-dependent N 2 O consumption were significantly higher in 0-20 cm than 20-40 cm fen soil (p < 0.001, and p < 0.001, as well as p = 0.02, respectively). Apparent Michaelis-Menten constants K M were about 60 to 140 times lower for N 2 O consumption than for nitrite dependent N 2 O production in 0 to 20 cm fen soil (p = 0.003), indicating a high affinity of fen denitrifiers for N 2 O (Table 2).

Phylogenetic analysis of denitrifiers in high latitude peatlands
Approximately 14 000 denoised quality-filtered sequences of the structural gene markers narG, nirK, nirS, and nosZ were utilized in total for further analyses. Forward and reverse reads for nirK and nirS showed a sufficiently long overlap (amplicon lengths of approximately 470 and 410 bp, respectively) to allow combined assessment of forward and reverse reads per gene for further analyses. Only forward reads of narG and nosZ were analyzed, as the overlap of forward and reverse reads was not sufficient to allow a combined analysis of forward and reverse reads (amplicon lengths approximately 670 and 700 bp for narG and nosZ, respectively), and previous studies indicate that results obtained from forward and reverse reads of narG and nosZ are similar [15,23]. More than 99% of sequences generated from amplicons of a certain gene specific (i.e., narG, nirK, nirS, nosZ) primer set were specific amplicons of the target gene. All library coverages were greater than 99% at species-level DNA sequence dissimilarities of 33%, 17%, 18%, and 20% for narG, nirK, nirS, and nosZ, respectively, and varied from 80% to 97% at 3% sequence dissimilarity ( Table 3), indicating that the number of sequences generated was sufficient.
narG sequences were assigned to 7 species-level OTUs in total. 7 and 4 OTUs were detected in 0 to 20 cm and 20 to 40 cm of fen soil, respectively (Table 3). narG community composition was similar in both sampled soil layers (Fig 4 A). Three OTUs had a relative abundance greater than 1%. Of those OTUs, OTU 1 dominated narG in fen soil (about 60% in both soil layers). About 40% of narG belonged to OTUs 2 and 3 (Fig 4 A). OTU 2 was more abundant in 0 to 20 cm than  in 20 to 40 cm fen soil (relative abundances of 33% and 9%, respectively), whereas OTU 3 was more abundant in 20 to 40 cm fen soil (23% vs. 6% in 0 to 20 cm fen soil ; Fig 4 A). Most of the OTUs were only distantly related to narG of cultured organisms or environmental sequences (i.e., sequence dissimilarities of OTU representatives were 10-23%) ( Table 4 and S3 Fig). Sequences of OTUs 1, 2, and 3 affiliated with narG of Alphaproteobacteria, Actinobaceria, and Deinococci, respectively, more specifically they were related to narG of uncultured bacteria and to those of Oligotropho carboxidovorans, Salinispora arenicola, and Marinithermus hydrothermalis, respectively (Table 4 and S3 Fig). Observed narG diversity was higher at 97% threshold similarity than at species-level threshold similarity (Table 3). At 97% threshold similarity, 359 and 230 OTUs were detected in 0 to 20 cm and 20 to 40 cm fen soil, respectively (Table 3). Shannon diversity, species evenness indices, and the observed number of OTUs calculated from rarified OTU tables indicated significantly higher diversity in 0-20 cm than 20-40 cm fen soil at 97% and species-level threshold similarity (Table 3). Beta-diversity measures indicated greater differences in community composition at 97% than at 67% threshold similarity (Table 3). nirK were assigned to 24 species-level OTUs in total. 23 and 17 OTUs were detected in fen soil from 0 to 20 cm and from 20 to 40 cm, respectively (Table 3). Community composition differed significantly between the soil layers (Fig 4 B). OTU 2 dominated nirK in fen soil from 0 to 20 cm (about 60%), while OTU 1 dominated nirK in fen soil from 20 to 40 cm, respectively (about 70% ; Fig 4 B). Similarities of OTU representative sequences to nirK of cultured organisms ranged from 75-100% (Table 4). Most OTUs were related to Alphaproteobacterial nirK. OTUs 1, 2, and 3 were related to nirK of Brucella canis, Rhizobium etli, and Castellaniella sp., respectively (Table 4 and S4 Fig). Further OTUs were related to nirK of Bosea sp., Afipia sp., or uncultured bacteria (Table 4 and S4 Fig). nirS were assigned to 25 species-level OTUs in total. 22 and 23 OTUs were detected in fen soil from 0 to 20 cm and from 20 to 40 cm, respectively (Table 3). Differences in community composition of nirS from the soil layers were more pronounced than those of nirK (Fig 4 B and 4 C). nirS of fen soil was dominated by OTUs affiliated to Beta-and Gammaproteobacterial nirS. However, about 26% of detected nirS from 20 to 40 cm affiliated with Alphaproteobacterial nirS (S5 Fig). nirS of OTU representatives were only distantly related to nirS of cultured organisms (i.e., similarities ranged from 74-84%, Table 4). Many OTUs of both soil layers were related to nirS of uncultured wetland or marine sediment bacteria, and distantly related to nirS of e.g., Thiobacillus denitrificans, Dechloromonas sp., and Arthrobacter sp. (Table 4 (Table 3). Chao1 richness estimates of nirS did not differ significantly at species-level similarity thresholds, amounting to about 24 in both soil layers, while Shannon diversity as well as species evenness were significantly higher in the lower soil layer (Table 3). On the contrary, Shannon diversity, species Evenness, and Chao1 richness estimates of nirK and nirS calculated from rarified OTU tables based on 97% threshold similarity were consistently higher in 0 to 20 cm than 20-40 cm fen soil (Table 3).
nosZ forward reads were assigned to 10 species-level OTUs in total. 8 OTUs were detected in each soil layer (Table 3). OTU 1 dominated nosZ of fen soil from both soil layers (Fig 4 D). Essentially all nosZ from both soil layers affiliated with Alpha-and Betaproteobacterial nosZ (S6 Fig). Most nosZ sequences from fen soil were distantly related to nosZ of cultured organisms with sequence dissimilarities ranging from 11-27% (Table 4), indicating hitherto uncultured denitrifiers capable of N 2 O reduction in fen soil. nosZ sequences clustered with nosZ of wetland and upland soils, as well as Achromobacter sp., Herbaspirillum sp., and Ralstonia sp. (Table 4 and S6 Fig). Shannon diversity and species evenness calculated from rarified OTU tables at species-level threshold similarity were significantly higher in 0 to 20 cm than in 20 to 40 cm soil, while there was no significant difference in Chao1 richness estimates (Table 3). At  * 100 (ns = OTUs that occur only once, nt = total number of sequences). 2 Number of OTUs observed in non-rarified OTU tables ± standard error. 3 Chao1 richness estimate of rarified OTUs ± standard error. 4 Shannon diversity index of rarified OTUs ± standard error. 5 Species Evenness of rarified OTUs ± standard error. 6 Sørensen similarity index of rarified OTUs ± standard error. 7 Bray Curtis similarity index of rarified OTUs ± standard error. 8 Unweighted Unifrac distance of rarified OTUs ± standard error. 9 Weighted Unifrac distance of rarified OTUs ± standard error.  97% threshold similarity, all diversity estimates calculated from rarified OTU tables were significantly higher in the upper soil layer ( Table 3). The difference in threshold similarity most strongly affected on the number of observed and estimated OTUs, which were similar at species-level threshold similarity (around 8 in both soil layers), but were about 3 times higher in 0 to 20 cm soil at 97% similarity threshold (Table 3). Beta-diversity was higher at 97% than at species-level threshold similarity ( Table 3).
Quantification of narG, nirK, nirS, and nosZ relative to 16S rRNA genes  5). Copy numbers of nirK, nirS, and nosZ were lower than narG copy numbers (Fig 5). Copy numbers of nirS were app. 100x and 10x higher than copy numbers of nirK and nosZ, respectively, in both soil layers (Fig 5). Copy numbers of narG and nosZ were 3 x higher (P < 0.01), and those of nirK were slightly lower in 0 to 20 cm than 20 to 40 cm fen soil (P = 0.1). Ratios of nosZ to narG were similar in both soil layers. Those of nosZ to nirK and nirS were 30 and 3 x higher, respectively, in 0 to 20 cm than 20 to 40 cm fen soil (Fig 5).

Discussion pH neutral fen soil as N 2 O sink
Peatlands are important ecosystems in the northern hemisphere and cover more than 30% of the Finnish land surface [36]. The potential of those peatlands to produce or consume greenhouse gases is of great interest, especially in respect to climate warming which is predicted to have a strong impact on peatlands [37]. N 2 O emissions from natural wetlands are highly variable, and many water-saturated soils are also sinks for N 2 O [2, 16,17]. Many studies demonstrate that undrained, pH-neutral and acidic fens are sources of molecular nitrogen, and act as sinks for N 2 O depending on environmental conditions [10,14,16,[38][39][40]. N 2 O accumulation in gas chamber experiments from Puukkosuo fen were also variable, ranging from 10 ppb to -17 ppb at the time of soil sampling. Fen soil in situ consumed initially produced N 2 O during nitrate or ammonium fertilization experiments (Fig 1). Previous studies show that mainly complete denitrification to N 2 occurs in pristine pH-neutral fens at in situ nitrate concentrations [10,40]. Thus, the absence of in situ N 2 O emission from Puukkosuo fen soil is likely due to complete denitrification to N 2 as the major end product (Fig 1). Even though the amount of stored nitrogen in the soil is high, low concentrations of available nitrate are observed in a northern boreal fen, where denitrification is thus N-limited [10]. Nitrate concentrations in Puukkosuo fen soil were likewise low, nitrite was not detected (Table 1), and nitrate as well as nitrite stimulated denitrification (Fig 3), indicating nitrate-and nitrite-limitation of fen denitrifiers. Microcosms and in situ fertilization with nitrate resulting in temporary in situ emission of N 2 O with subsequent consumption indicated ongoing complete denitrification (Fig 1). Peatland soils are temporarily or permanently water-logged, and oxygen generally penetrates only the uppermost centimeters, leading to oxygen-limitation in lower soil layers. In the absence of oxygen and nitrate, nitrous oxide is a potent sink for electrons released during the oxidation of organic carbon compounds, as the reduction of N 2 O by H 2 is even more exergonic than O 2 reduction by H 2 (N 2 O half-cell potential of E 0 ' (pH 7.0) = 1.35 V; ΔG 0 ' = -339.5 kJ Ã mol -1 ; reviewed in [41]). In situ relevant concentrations of dissolved organic carbon (app. 5 μM glucose equivalents; e.g., [42]) and atmospheric concentrations of     (Table 2), indicating a higher affinity of fen denitrifiers for N 2 O than for nitrate. The assumed absence of oxygen, the observed nitrate-limitation and high N 2 O affinity indicate a strong in situ sink potential of Puukkosuo fen for N 2 O.

Diverse denitrifier communities are associated with denitrification activities in pH-neutral fen soil
Unsupplemented fen soil from both sampled soil layers produced N 2 O in acetylene-amended microcosms, demonstrating the denitrification potential of the fen soil. However, nearly no N 2 O was produced in the absence of acetylene and initially produced N 2 O was subsequently consumed (Fig 2). Nitrate-and oxygen-limitation might select for denitrifiers capable of complete denitrification, and hitherto unknown denitrifiers as well as N 2 O reducers might occur in Puukkosuo fen soil. Indeed, nosZ copy numbers in 0 to 20 cm fen soil were of a similar magnitude as nitrite reductase copy numbers (Fig 5), and newly-discovered nirK/S and nosZ (Table 4) indicate that a high percentage of uncharacterized denitrifiers in that soil layer possessed a complete denitrification pathway. Supplemental nitrate and nitrite resulted in immediate N 2 O production in fen soil after internal nitrate and nitrite were consumed. Stimulation was greater with nitrite than with nitrate in both soil layers (Fig 3 A). This reflects the fact that all denitrifiers "sensu stricto" use nitrite as electron acceptor, while many cultured denitrifiers lack the ability to use nitrate as electron acceptor [20,43]. Stimulation was also greater in the top soil layer (Fig 3 A), reflecting a greater denitrification potential of the top soil. In other wetland and also agricultural soils, denitrification potentials are also highest in the top soil layers (e.g., [14,44]). In 20 to 40 cm fen soil, N 2 O production and total denitrification decreased with increasing nitrate concentrations, indicating substrate inhibition of denitrification at high nitrate concentrations. This finding is in contrast to denitrification potentials reported for deeper horiozonts of agricultural soils, suggesting that the fen denitrifier community of 20-40 cm depth is well adapted to low nitrate concentrations (e.g., [44]). Nitrate and nitrite reduction compete for electrons at high nitrate concentrations, and nitrate reduction is favored over the rest of the denitrification pathway, causing eventually accumulation of nitrite when electron donors are limiting [45].
The ratio of N 2 O to (N 2 +N 2 O) was lower in nitrate-and nitrite-amended microcosms with 0 to 20 cm fen soil than in 20 to 40 cm fen soil when nitrate or nitrite were supplied (S2 Fig), and consumption of supplied N 2 O was about 2-fold higher in 0 to 20 cm than in 20 to 40 cm fen soil (Fig 3 B). Indeed, the ratio of nitrite to nitrous oxide reductases was higher in 20 to 40 cm fen soil than in 0 to 20 cm fen soil (Fig 5), indicating an increased amount of denitrifiers lacking nitrous oxide reductase in the lower soil layer. The ratio of nitrite reductase genes to N 2 O reductase genes is highly variable in soils, and often nitrite reductase copy numbers largely exceed N 2 O reductase copy numbers [15,23,46]. However, non-denitrifying N 2 O consumers were recently shown to be quantitatively important in certain soils [47,48]. Relative abundances of both atypical and typical nosZ assigned to non-denitrifiers and denitrifiers, respectively, are variable in soil metagenomes. Hence, further analyses including both groups are demanded for better understanding of N 2 O reducers in fens [49]. Nevertheless, N 2 O produced in lower layers of fen soil can diffuse upwards and be further reduced to N 2 in upper soil layers, and thus emission of N 2 O into the atmosphere can be reduced [14,16,38]. It is thus hypothesized that also in Puukkosuo fen soil lower soil layers are N 2 O sources while upper soil layers are N 2 O sinks.
The analysis of denitrification-specific gene markers indicated a higher diversity of these genes in 0 to 20 cm than in 20 to 40 cm fen soil (Table 3). Detected narG and nosZ were more similar in 0 to 20 cm and 20 to 40 cm fen soil than nirK and nirS (Fig 4), indicating that nitrite reductases show a higher variability in fen soil than nitrate and N 2 O reductases. Nitrite reductase community composition is highly variable in other types of peatland soils, including permafrost affected systems, while variations in nitrate and N 2 O reductase community composition are much less pronounced [15,23]. Indeed, the distribution of nitrite reductases is more heavily impacted by changes in environmental conditions than those of nitrate or N 2 O reductases [50][51][52]. Nitrite reductase genes from fen soil were affiliated with Proteobacterial nirK/S (Table 4 and S4 Fig and S5 Fig). For nirS, sequences related to Rhodanobacter/Bradyrhizobium were detected (S5 Fig). Such sequences are also detected in other peatland soils such as permafrost affected tundra and palsa peat soils [15,23]. Proteobacteria-affiliated sequences of narG and nosZ (Table 4 and S3 Fig and S6 Fig) further support that Proteobacteria play an important role for denitrification in this pH-neutral fen soil. Denitrification-associated genes related to Proteobacteria are also found in acidic fen soils or permafrost-affected peatlands [14,15,23], indicating that Proteobacteria represent general peatland denitrifiers. Sequences of narG were also affiliated with Actinobacterial narG (10-30%; S3 Fig). Actinobacteria are common in soils, include many genera capable of nitrate reduction, and are in general considered to be more tolerant to extreme environmental conditions such as low pH or low temperature [53][54][55]. Actinobacteria and Actinobacteria-affiliated gene markers are frequently detected in a variety of peatlands including acidic fen soils, permafrost-affected tundra and palsa peat soils [14,15,23]. However, in those more extreme environments, Actinobacteria often dominate the narG communities, indicating that Actinobacteria are further important players involved in nitrate reduction and potentially denitrification in pH-neutral fen soil [15,23]. The nitrate reducer community in pH-neutral fen soil also contained a substantial portion of Deinococciaffiliated narG (Fig 4A and S3 Fig.), which are not detected in the above mentioned more extreme habitats such as acidic fens, frost-affected tundra and palsa peat soils [15,23]. Soil pH is a driver of the general microbial community structure [56]. Denitrifier diversity in pH-neutral fen soil is high when compared to more acidic pristine peatland soils [14,15,23], suggesting that soil pH likewise plays an important role in shaping denitrifier communities.
Denitrifier diversity and quantity is routinely underestimated due to choice of primer sets, e.g., gram-positive denitrifiers escaped detection in many studies [46,50,57]. Soil metagenomes might represent an alternative strategy to obtain a more complete picture of denitrifier diversity in soils. However, the low abundance of denitrification associated genes on denitrifier genomes (i.e., app. 1%; most of the genes on denitrifier genomes are associated with other functions than denitrification like anabolism, motility, etc.) in combination with a low number of denitrifiers compared to total number of prokaryotes in soil (app. 1%) limits their detection by metagenomics [49,58,59]. However, metagenomes are extremely useful for the design of denitrification gene specific primers. Although amplicon based approaches combined with next generation sequencing depend on the choice of primers, such approaches currently provide a cost-effective way for the detection of a large denitrifier diversity.
The collective data indicate that (i) a core nitrate reducer/denitrifier community might be common to all kinds of (northern) peatlands, (ii) some nitrate reducers/denitrifiers are unique in pH-neutral fen soil, possible due to the lack of environmental stress that might be induced by acidic pH, (iii) denitrifier communities are from upper and lower layers are dissimilar as indicated by apparent Michaelis-Menten kinetics and structural gene marker analyses, and (iv) pH-neutral fens are a strong potential sink for atmospheric N 2 O. The tree was calculated based on translated amino acid sequences. OTUs were grouped at a species-level threshold dissimilarity of 17%. Numbers preceeding sequence names refer to sequence accession numbers of reference sequences from public databases. Values given in parentheses show the relative abundances of each OTU in 0 to 20 cm (left) and 20 to 40 cm (right) fen soil. Grey boxes indicate reference sequences belonging to the same phylogenetic group. The percentages of replicate trees that produced the observed clustering of taxa in the bootstrap test (10 000 replications) are shown next to the branches. Bootstrap supports below 50% are not displayed. nirK of Nitrosomonas sp. C-56 was used as outgroup to root the tree. The tree was calculated based on translated amino acid sequences of nosZ forward reads. OTUs were grouped at a species-level threshold dissimilarity of 20%. Numbers preceeding sequence names refer to sequence accession numbers of reference sequences from public databases. Values given in parentheses show the relative abundances of each OTU in 0 to 20 cm (left) and 20 to 40 cm (right) fen soil. Grey boxes indicate reference sequences belonging to the same phylogenetic group, white boxes indicate single taxa not belonging to the major phylogenetic group. The percentages of replicate trees that produced the observed clustering of taxa in the bootstrap test (10 000 replications) are shown next to the branches. Bootstrap supports below 50% are not displayed. nosZ of Haloarcula marismortui ATCC 43049 was used as outgroup to root the tree. (TIF)