FEMS Microbiology Ecology Advance Access published June 10, 2015
1 1
Microbial ecology in a future climate: effects of temperature and moisture on microbial communities
2
of two boreal fens
3 4
Krista Peltoniemi1, Raija Laiho2, Heli Juottonen1, Oili Kiikkilä1, Päivi Mäkiranta1, Kari Minkkinen3, Taina
5
Pennanen1,Timo Penttilä1, Tytti Sarjala2, Eeva-Stiina Tuittila4, Tero Tuomivirta1 & Hannu Fritze1
6 7
1
Natural Resources Institute Finland (Luke), Vantaa, Finland; 2 Natural Resources Institute Finland (Luke),
8
Parkano, Finland; 3Department of Forest Sciences, University of Helsinki, Helsinki, Finland; and 4School of
9
Forest Sciences, Faculty of Science and Forestry, University of Eastern Finland, Joensuu, Finland
10 11
*Correspondence: Krista Peltoniemi, Natural Resources Institute Finland, Jokiniemenkuja 1, FI-01370
12
Vantaa, Finland
13
Tel.: +358 29 532 5585; fax: +358 29532 2103; e-mail:
[email protected] 14
Running title: Environmental effects on peatland microbes
15 16 17
Keywords: peatland, climate warming; water-level drawdown; microbial communities; fungi; mycorrhizae
2 17 18
Abstract
19 20
Impacts of warming with open-top chambers on microbial communities in wet conditions and in conditions
21
resulting from moderate water-level drawdown (WLD) were studied across 0–50 cm depth in northern and
22
southern boreal sedge fens. Warming alone decreased microbial biomass especially in the northern fen.
23
Impact of warming on microbial PLFA and fungal ITS composition was more obvious in the northern fen
24
and linked to moisture regime and sample depth. Fungal specific PLFA increased in the surface peat in the
25
drier regime and decreased in layers below 10 cm in the wet regime after warming. OTUs representing
26
Tomentella and Lactarius were observed in drier regime and Mortierella in wet regime after warming in the
27
northern fen. The ectomycorrhizal fungi responded only to WLD. Interestingly, warming together with
28
WLD decreased archaeal 16S rRNA copy numbers in general, and fungal ITS copy numbers in the northern
29
fen. Expectedly, many results indicated that microbial response on warming may be linked to the moisture
30
regime. Results indicated that microbial community in the northern fen representing Arctic soils would be
31
more sensitive to environmental changes. The response to future climate change clearly may vary even
32
within a habitat type, exemplified here by boreal sedge fen.
3 33 34
Introduction
35 36
Peatlands such as bogs and fens store at least one third of terrestrial carbon (C) as partially decomposed
37
vegetation (peat) and play an important role in global C cycling (Page et al., 2011). Sequestration of C as
38
peat depends on several factors but is ultimately facilitated by a sufficiently high and persistent water table
39
that can maintain anoxic conditions and thus slow down microbial decomposition. Recent and predicted
40
climate warming may accelerate the decomposition rate in northern peatlands and create a large and positive
41
feedback to atmospheric carbon and global climate (Dorrepaal et al., 2009; Vicca et al., 2009).
42
Palaeoecological studies suggest, however, that the response of peatlands to an increase in mean annual
43
temperature is dependent on the local moisture regime; warm and wet periods in the past were associated
44
with rapid expansion of northern peatlands, unlike in warm but dry conditions (Weckström et al., 2010).
45
Approximately half of all peatlands occur in northern Eurasia (Joosten, 2010) where the climate is expected
46
to become warmer and gradually wetter (Arzhanov et al., 2012; Monier et al., 2013), with a broader range of
47
summer rainfall among regions (ROSHYDROMET, 2008; Screen, 2013) and some peatlands expect to
48
suffer drought more frequently (Gorham, 1991; Roulet et al., 1992). As such, projections for the response of
49
northern peatlands to climate warming should be developed under wetter as well as drier scenarios.
50
Ecosystem respiration and gross primary production are influenced by both temperature and moisture
51
but their responses to warming and drying of peatlands may vary, such that the net exchange of C may
52
increase, decrease or remain the same (Sullivan et al., 2008; Chivers et al., 2009). This variation seems to be
53
related to plant community composition and microtopography, factors that are in turn created by the local
54
hydrologic regime (Laine et al., 1995; Strack et al., 2006; Riutta et al., 2007). Major changes in vegetation
55
due to climate change may greatly affect the C sequestration potential of northern peatlands (Tahvanainen,
56
2011; Loisel & Yu, 2013). The role of microbial decomposers within this dynamic and variable system has
57
yet to be established, although earlier studies indicate that changes in plant community composition drive
58
successional changes in aerobic microbial communities via shifts in litter quality (Jaatinen et al., 2007;
59
Peltoniemi et al., 2009).
4 60
Although microbes perform the bulk of organic decomposition and play a major role in ecosystem
61
respiration, little is known of the likely effects of climate warming on the diverse soil microflora.
62
Temperature regulates the rate of microbial metabolism and thereby all biogeochemical cycles driven by
63
microbes (Bradford, 2013). An increase in temperature accelerates microbial CO2 production in mineral
64
soils (Vanhala et al., 2008), and can reduce soil quality because labile substrates are depleted faster than
65
more recalcitrant ones (Davidson and Janssens, 2006). Furthermore, the availability of labile C to microbes
66
is lower in experimentally warmed soils (Hartley et al., 2007; Bradford et al., 2008; Curiel Yuste et al.,
67
2010), and this has been suggested as one of the means by which microbes and microbial communities
68
become locally adapted (Bradford et al., 2013). In contrast to the stimulating effect of a warmer temperature,
69
bacterial activity in a sub-Arctic heath decreased after long-term warming (Rinnan et al., 2011), and the
70
density of fungi producing phenol-oxidase was also reduced in a fen after warming (Jassey et al., 2011).
71
Thormann et al. (2004) observed how microbial decomposition became dominated by fungi rather than
72
bacteria as the temperature increased in boreal peatlands. Although it is a common knowledge that moisture
73
is one of the important factors affecting the temperature sensitivity of the microbial processes (Davidson and
74
Janssens, 2006) temperature driven effects on microbial communities in peatlands under different moisture
75
regimes have not been investigated.
76
Our earlier research has shown that even a moderate water-level drawdown (WLD) can induce changes
77
in the microbial communities of boreal peatlands (Jaatinen et al., 2007, 2008; Peltoniemi et al., 2009). In
78
natural mesotrophic fens, fungal decomposers showed a positive response to WLD (Jaatinen et al., 2007)
79
and ectomycorrhizal (ECM) fungi appeared in community profiles (Jaatinen et al., 2008). While WLD
80
appears to be the main factor regulating the succession of microbial communities, increasing temperature
81
can promote species turnover and affect microbial respiration (Conant et al., 2011). Interactions between
82
abiotic and biotic factors do not appear to be simple, and some evidence suggests that warming and drying
83
may have interactive, non-additive effects on microbial processes (A’Bear et al., 2014). For example, while
84
warming had a slight effect on PLFA composition in a dry heath soil, the same increase had little or no
85
effect in a wet heath soil (Rinnan et al., 2007, 2008). Different microbial groups may also vary in the extent
86
to which they are affected by climate, e.g., when field soil was warmed experimentally, the abundance of
5 87
fungi increased while bacterial biomass decreased regardless of moisture level (Castro et al., 2010). Even
88
though microbial communities are key elements of the peatland C cycle, the interactive effects of
89
temperature and moisture on their activity, abundance and structure are poorly understood.
90
Fens are minerotrophic peatlands dominated by graminoids that receive nutrient-rich water from
91
surrounding mineral soils. They are considered to be more vulnerable to disturbance than bogs, especially
92
with respect to WLD (Laine et al., 1995; Komulainen et al., 1999; Jaatinen et al., 2007). Our experiment
93
considered two alternative future climate scenarios and their effects on fen microbes, i.e., a higher mean
94
temperature in a wetter or drier moisture regime. We studied the soil microbes in pristine (i.e., wetter) sites
95
and sites after moderate WLD (i.e., drier) of two boreal fens that had received an increased temperature
96
treatment for three years prior to sampling. We hypothesized that temperature-driven changes in the
97
microbial community are dependent on moisture regime. In particular, we studied changes in microbial (i)
98
activity, (ii) abundance, (iii) community composition, and iv) occurrence of fungi.
99 100
Materials and methods
101 102
Study site, experimental design and sampling
103 104
Experimental sites intended to simulate future climate scenarios were established in 2008 in two locations of
105
the boreal zone to test the impacts of warming on plant-soil processes in a variable moisture regime. Sites
106
were located in Orivesi (the southern fen, Lakkasuo, 61°48' N 24°19' E) and Kittilä (the northern fen,
107
Lompolojänkkä, 67°60' N 24°12' E). In the southern fen from 1981 to 2011, the mean annual temperature
108
and precipitation were ca. +3.5 °C and 700 mm, respectively, and the accumulative temperature sum (+5 °C
109
threshold) was ca. 1050 degree days (d.d.). Correspondingly, values for the northern fen were -1.4 °C, 511
110
mm and 700 d.d. The sites included pristine plots as well as plots slightly drained by shallow ditches dug in
111
2008. Ditching lowered water levels approximately 6 and 3 cm on an average in the southern and the
112
northern fen, respectively (Table 1); the differences were greater during the growing season.
6 113
Even though both sites were classified as sedge fens, there were some different patterns in their
114
vegetation. In both locations, the field layer was characterized by sedges such as Carex lasiocarpa
115
(especially the southern fen), C. rostrata (especially the northern fen), C. chordorrhiza and C. limosa along
116
with dwarf shrubs Andromeda polifolia and Vaccinium oxycoccos (more abundant in the northern fen) and
117
the herb Menyanthes trifoliata, and in the northern fen additionally Equisetum fluviatile and Comarum
118
palustre. In the moss layer Sphagnum papillosum and S. flexuosum were abundant in the southern fen with
119
small patches of S. fallax, S. subfulvum and S. subnitens, while in the northern fen the moss layer was more
120
patchy and consisted mainly of S. fallax, S. riparium and S. flexuosum, along with some S. jensenii.
121
Chemical analyses were performed prior to the start of the experiment from a separate set of peat core
122
samples from 5–15 and 15–25 cm below the moss layer. They showed that the northern fen was higher in
123
calcium, potassium, magnesium and phosphorus and the southern site had more total nitrogen (Table S1 in
124
Supplementary material). Carbon and nitrogen were determined from an air-dried peat sample with a LECO
125
CHN-1000 analyzer, and the concentrations of other elements with an ICP-emission spectrometer (ARL
126
3580) using dry ash dissolved in hydrochloric acid.
127
Experimental sites were established in a split-plot design, i.e., each plot with either ambient or lowered
128
water level was divided into six subplots; three of which received no temperature manipulation (i.e.,
129
controls) and the other three received seasonal artificial warming. Warming the surface air and soil of the
130
subplots was realized passively with small open-top chambers (OTC), widely used to simulate climate
131
warming (Marion et al., 1997; Hollister & Webber 2000). We used hexagonal OTCs made of clear plastic
132
sheet, 60 cm tall and each side was 76 cm long (max. diagonal distance of 131 cm). OTC side panels were
133
inclined 60% to improve the transmittance of solar radiation and help trap heat, and were placed on 10 cm
134
tall supports that allow ventilation of the air space and reduce disturbance to the soil surface. Because our
135
experiment took place on groundwater-fed fens, OTCs had little or no effect on the water level of each
136
subplot (i.e., they did not induce WLD as such; Aerts, 2006). During the growing season, OTCs increased
137
average daily air temperature within the subplot by ca. 1.5 °C at 15 cm above the peat surface. With respect
138
to controls, the effective temperature sum in the OTCs increased by ca. 320 and 200 degree days in the
139
southern fen and the northern fen, respectively. In 2011, the average daily temperature immediately below
7 140
the moss layer was approximately 0.8 °C higher and at 5cm depth 0.3 °C higher under the OTCs compared
141
to control subplots. Notable differences were not observed in soil layers 5 cm below the surface.
142
Sample cores were taken with a 6×8 cm box-corer from each subplot in September 2011. Sampling was
143
conducted to the depth of 20–50 cm; some sample cores were so wet that only first 20 cm of peat could be
144
obtained. Cores were divided into sub-samples at 10 cm intervals from the peat surface (L1: 0–10, L2: 10–
145
20, L3: 20–30, L4: 30–40 and L5: 40–50 cm) (Table 1). In total, 100 sub-samples from two experimental fen
146
sites were obtained from 3–5 depths from pristine and WLD plots of three OTC treatments and their
147
controls. 71 of the sub-samples were from layers L1–L3.
148 149
In-growth mesh bag experiment for harvesting ECM fungi
150 151
We used in-growth mesh bags (Wallander et al., 2001) to study the ECM fungal community. Bags were
152
filled with 150 g acid-washed quartz sand (particle size 0.5–1.5 mm, SP Minerals OY AB, Finland) and
153
buried horizontally just below the living moss layer in each subplot during June–October of 2010 and June–
154
October of 2011. Bags were harvested after three months and transported to the laboratory where they were
155
opened, immersed in water, and mycelia were removed with the aid of a stereo-microscope and stored at -20
156
°C prior to molecular analysis. The ergosterol concentration (µg g-1 dry weight) in mesh bags was
157
determined using a modification of a method developed by Nylund and Wallander (1992), consisting of
158
ethanol extraction, saponification with KOH (Merck) and further extraction in pentane (Merck) with two
159
replicate samples from each mesh bag. Analysis of ergosterol was performed with High-Powered Liquid
160
Chromatography (HPLC) with a Merck Hitachi UV–VIS detector (280 nm wavelength) and a LiCrospher®
161
100 RP-18 column with methanol as eluent.
162 163
Microbial basal respiration and growth rate measurements
164 165
Within 24 h of collection, peat samples were transported to the laboratory and sub-samples were transferred
166
to +14 °C for two days prior to the respiration measurements. The basal respiration rate (BR) was measured
8 167
as the amount of CO2 evolved after 24 h and 48 h from 6 ml of fresh peat soil placed into a cut-end syringe
168
and inserted into a 120 ml incubation bottle as described by Pietikäinen and Fritze (1995). Peat soil pH was
169
determined in distilled water (1:3, vol/vol) after BR measurements from the same bottles.
170
Bacterial and fungal growth rates and fungal ergosterol were determined for L1 and L2 (0–20 cm) which
171
showed the largest microbial biomass according to PLFA. For the measurement of bacterial growth rate, a
172
small sub-sample of fresh peat (equivalent of 0.7 g of loss on ignition) was mixed with 100 ml of ultrapure
173
water and shaken at 200 rpm for 1 hour. After centrifugation (750×g, 10 min) and filtration through quartz
174
wool, 1.4 ml of the supernatant was transferred to microcentrifuge tubes. An aliquot (3.5 µl, 0.1 MBq) of
175
methyl-3 H-thymidine (740 GBq mmol-1, Moravek Biochemicals) was added and samples were incubated for
176
2 hours at 22oC. Samples were then washed to remove excess tracer and the measurement of radioactivity
177
was described in detail by Bååth et al. (2001). Relative bacterial growth is expressed as radioactivity (DPM,
178
disintegrations per minute) in a sample.
179
For the measurement of fungal growth rate, 0.5 g of fresh peat was placed in a test tube with 1.5 ml of
180
ultrapure water, and 0.13 µmol (0.3 MBq) 14C-acetate solution (1-14C-acetic acid, sodium salt, 2.2 GBq
181
mmol-1, Moravek Biochemicals) and 0.35 µmol of 1 mM non-radioactive acetate were then added. After
182
incubating the mixture for 20 hours at 20 °C, formalin was added and samples were centrifuged, and the
183
supernatant was discarded. The ergosterol was then extracted as in Bååth (2001) and measured with HPLC
184
and 14C-ergosterol with a radioactivity monitor (Berthold, LB 506 C-1). Fungal growth is expressed as DPM
185
14
C-acetate per ergosterol per g dry soil.
186 187
Phospholipid fatty acid (PLFA) analyses of peat soil
188 189
Peat samples were kept at 4 °C between sampling and PLFA analyses. Dry weights were determined after
190
drying sub-samples overnight at 105 °C. Moisture content of the sub-samples ranged between 87−95 % after
191
drying. The phospholipid extraction and analysis of PLFAs were carried out as described by Frostegård et
192
al. (1993). The green vegetation part (mostly mosses and fresh litter) was excluded from the analysis since
193
the plant-derived PLFAs would interfere with the interpretation of microbial signatures. Briefly, fresh peat
9 194
soil corresponding to 1.5 g of dry peat was extracted with chlorophorm:methanol:citrate (1:2:0.8) buffer and
195
lipids were separated into neutral lipids, glycolipids and phospholipids in a silic acid column. Phospholipids
196
were subjected to a mild alkaline methanolysis and the fatty acid methyl esters were detected by gas
197
chromatography using a flame ionization detector and 50-m HP-5 capillary column. Thirty-nine different
198
PLFAs were identified from each sample and they were expressed as mole percentage (mol % =area % of a
199
single PLFA from the area sum of all identified PLFAs).
200
The following PLFAs were considered to be predominantly of bacterial origin (i15:0, a15:0, 15:0, i16:0,
201
16:1ω9, 16:1ω7t, i17:0, a17:0, 17:0, cy17:0, 18:1ω7 and cy19:0) and chosen as an index of bacterial
202
biomass (Frostegård & Bååth, 1996). The amount of 18:2ω6 was used as an indicator of fungal biomass
203
because it is suggested to mainly be of fungal origin in soil (Federle, 1986) and is known to correlate
204
strongly with the amount of ergosterol (Frostegård & Bååth, 1996). PLFAs 10Me16, 10Me17, 10Me18 are
205
considered to be of actinobacterial origin (Kroppenstedt, 1985). The two deepest sample layers (L4 and L5:
206
30–50 cm) were discarded from the final analysis since their PLFA content was constantly at or below the
207
detection limit.
208 209
PCR-DGGE community fingerprinting and fungal sequence analyses
210 211
A small amount of each sample was kept frozen at -20 °C for DNA analysis. DNA was extracted from L1
212
(0–10 cm) and L2 (10–20 cm), which contained most of the fungal biomass according to PLFA analysis.
213
Extractions were conducted according to the slightly modified protocol of the PowerSoil DNA extraction kit
214
(MO BIO Laboratories Inc., CA, USA). Briefly, cell disruption and homogenization were conducted with a
215
FastPrep instrument (3 × 20 s at 5.5 m s-1). After homogenization, extraction tubes were incubated at 65 °C
216
for 30 mins. Mycelia recovered from in-growth mesh bags were subjected to DNA extraction as described
217
by Vainio et al. (1998) and Jaatinen et al. (2008) with slight modifications in that some DNA extracts
218
needed an extra purification step with PEG-solution (20 % PEG in 2.5 M NaCl) where DNA:PEG ratio was
219
10:6 and some extracts were precipitated with 3 M sodium acetate (DNA:sodium acetate ratio 10:1) and
220
99,5 % EtOH (DNA:EtOH ratio 1:2) to obtain sufficient DNA for further analyses.
10 221
Fungal ITS regions were PCR amplified using GC-clamped ITS1F (Gardes & Bruns, 1993, Muyzer et
222
al., 1993) and ITS2 primers (White et al., 1990). PCR reactions were conducted in Bio-Rad Thermal cycler
223
(Bio-Rad Laboratories, Hercules, USA) using 50 μl reaction mixtures containing 10× reaction buffer (100
224
mM Tris/HCl, pH 8.8; 15 mM MgCl2; 500 mM KCl; 1 % Triton X-00), 0.25 U or 0.5 U of DNA polymerase
225
(Biotools B&M Labs S.A, Madrid, Spain), 200 μM of each dNTP (Thermo Fisher Scientific Biosciences,
226
Germany), 0.5 mM of both primers (MWG Biotech AG, Edersberg, Germany) and 1 μl of DNA extract.
227
Thermal cycling parameters were for ITS as described in Korkama et al. (2007).
228
Fungal communities from peat soil and in-growth mesh bags were analyzed by denaturing gradient gel
229
electrophoresis (DGGE) using the INGENYphorU-system (Ingeny, Netherlands) with denaturing gradient of
230
25–60 % made of 7.5 % weight per volume (w/v) acrylamide/bisacrylamide (37:5:1), urea and formamide.
231
Gels were run at 70 V and 60 °C for 16 h, stained with SYBR Gold II (Molecular Probes, Eugene, OR,
232
USA) and visualized with blue light on a Dark Reader transilluminator (Clare Chemical Research Inc.,
233
Dolores, CO, USA). DGGE bands from each operational taxonomic unit (OTU) representing a distinct
234
mobility pattern were selected for sequencing, excised, re-amplified with 20 or 25 cycles of PCR and
235
purified with the GeneJET PCR purification kit (Thermo Fisher Scientific Biosciences, Germany). Excised
236
ITS bands that derived from peat soil samples were coded with a unique number-letter combination. Excised
237
ITS bands were sent to a commercial sequencing service (Macrogen Europe, Amsterdam, The Netherlands)
238
and resulting sequences were compared to reference sequences on the public databases of
239
GenBank/EMBL/DDBJ (MEGABLAST algorithm) and the International Nucleotide Sequence Databases
240
(INSD) in UNITE (Kõljalg et al., 2013) with BLASTN and FASTA3 algorithms for ITS sequences. Similar
241
sequences were aligned with Geneious 6.1.2 (Biomatters, New Zealand) and novel high-quality sequences
242
were deposited to GenBank databases (KJ588522 to KJ588613).
243 244
Quantitative PCR
245 246
PCR products obtained with fungal ITS1F and ITS2 (White et al., 1990; Gardes and Bruns, 1993), bacterial
247
16S rRNA 1055F and 1392R (Olsen et al., 1986; Woese, 1987; Stahl et al., 1988; Lee et al., 1993) and
11 248
archaeal 16S rRNA Arch967f and Arch1060R primer pairs (Amann et al., 1990; Reysenbach 6 Pace, 1995;
249
Riley-Buckley, 2001) were cloned. Standard curves were constructed with plasmids containing
250
corresponding inserts, taking into account the molecular mass of the plasmid, including the insert, and the
251
plasmid concentration. We ran qPCR (Rotor-Gene 6000, QIAGEN, Netherlands) with Maxima TM SYBR
252
Green qPCR Master Mix (2×) (Thermo Fisher Scientific Biosciences, Germany) in a 20 µl final reaction
253
volume containing 1µl template, 0.3 µM of each primer and 1× qPCR master mix. Fluoresence was
254
measured at the end of each extension step. Each qPCR run was carried out under the following conditions:
255
initial denaturation at 95°C for 10 min; 40 cycles denaturation at 95°C for 15 sec, annealing at 55°C (ITS
256
and archaeal 16S rRNA) or 68°C (bacterial 16S rRNA) for 15 (ITS) or 30 (16S rRNA) sec, extension at
257
72°C for 15 (ITS), 20 (archaeal 16S rRNA) or 30 sec (bacterial 16S rRNA); and final extension at 72°C for
258
10 min. All samples were replicated and lack of PCR inhibition was verified through 1:10 dilution. The copy
259
numbers in samples were calculated based on comparison to threshold cycle values of the standard curve
260
and are given per gram of soil (dry weight).
261 262
Statistical analyses
263 264
We explored the effects of warming, moisture regime (wetter pristine or drier after WLD), fen site, sampling
265
depth in the peat profile, pH and their interactions on microbial community composition and activity with
266
multilevel (mixed) models from peat layers L1–L3 (0–30 cm) (n=71), acknowledging the hierarchical
267
structure in the data. Our observations were derived from different depths at each sampling plot, and the
268
sampling plots were clustered within treatment plots, following the split-plot experimental design within
269
each fen. These four hierarchical levels (i.e., depth, sampling plot, treatment plot, site) were included in the
270
basic structure of the models constructed using MLwiN 2.26 (Rasbash et al., 2012). They were effectively
271
applied as random variables except for cases where one or a few were used as a fixed effect (e.g., when
272
specifically analyzing the effect of sampling depth or fen). Statistical significance of differences was
273
evaluated based on a coefficient:standard error ratio ≥2. The -2loglikelihood value was used to compare
274
models of increasing complexity. The restricted iterative generalized least square (RIGLS) algorithm was
12 275
applied, a decision motivated by the size of the data set. RIGLS is formally equivalent to residual maximum
276
likelihood (REML) under normality.
277
Preliminary analyses indicated that treatment effects were complex, often site and/or depth dependent,
278
and did not always show interactive effects. Consequently, we tested treatment effects both separately and in
279
combination, i.e., they were added to the models as one-factorial variables indicating three situations:
280
warming, WLD, and warming+WLD, or as two-factorial variables indicating warming (irrespective of
281
moisture regime) and WLD (irrespective of warming). Model construction began with site and depth as
282
explanatory variables, and complexity was gradually increased by applying treatment effects, first as pure
283
effects, and next in interaction with site, depth, and site+depth.
284
We conducted microbial community analyses from PLFA- and two ITS-derived data sets (from peat soil
285
and in-growth bags) with multivariate analyses using CANOCO 5.0 (ter Braak 6 Šmilauer, 2012). DGGE-
286
derived ITS data were converted to a binary matrix (absence=0, presence=1). We first examined the extent
287
of variation (heterogeneity) in the PLFA and two ITS data sets with detrended correspondence analysis
288
(DCA). According to the length of the main gradient, we applied methods with a linear species response
289
model for the PLFA data and unimodal species response model for the ITS data. PLFA data were then log-
290
transformed and the overall variation in PLFA composition was analysed using principal components
291
analysis (PCA). The PLFA data were also analyzed with a redundancy analysis (RDA) to investigate the
292
relationships between community composition and environmental variables (fen site, sampling depth,
293
warming treatment, moisture regime and their interactions). We used partial canonical correspondence
294
analysis (CCA) for the ITS data to investigate whether the relationships between fungal community
295
composition and environmental variables (sampling year or depth, fen site, warming treatment, moisture
296
regime and their interactions) were significant. The individual or combined effects of each environmental
297
variable in RDA or CCA were tested by extracting the effects of other variables by using them as covariates.
298
The significance of the axes (P