FEMS Microbiology Ecology Advance Access published May 15, 2015
1
Effects of annual and inter-annual environmental variability on soil fungi associated with
2
an old-growth, temperate hardwood forest.
3
Running title: Environmental variability and forest soil fungi
4
David J. Burke1,2*
5
1
The Holden Arboretum, 9500 Sperry Road, Kirtland, OH, USA.
6
2
The Department of Biology, Case Western Reserve University, Cleveland, OH 44106, USA
7
Corresponding author: *The Holden Arboretum, 9500 Sperry Road, Kirtland, OH, USA, e-
8
mail:
[email protected], phone: 440-602-3858, fax: 440-602-8005.
9 10
Tables: 5
11
Figures: 4
12
Supplementary tables:
13
Supplementary figures:
14
14
Abstract
15
Seasonal and inter-annual variability in temperature, precipitation and chemical resources may
16
regulate fungal community structure in forests but the effect of such variability is still poorly
17
understood. In this study, I examined changes in fungal communities over two years and how
18
these changes were correlated to natural variation in soil conditions. Soil cores were collected
19
every month for three years from permanent plots established in an old-growth hardwood forest
20
and molecular methods were used to detect fungal species. Species richness and diversity were
21
not consistent between years with richness and diversity significantly affected by season in one
22
year but significantly affected by depth in the other year. These differences were associated with
23
variation in late winter snow cover. Fungal communities significantly varied by plot location,
24
season and depth and differences were consistent between years but fungal species within the
25
community were not consistent in their seasonality or in their preference for certain soil depths.
26
Some fungal species, however, were found to be consistently correlated with soil chemistry
27
across sampled years. These results suggest that fungal community changes reflect the behavior
28
of the individual species within the community pool and how those species respond to local
29
resource availability.
30 31
Key words: climatic variability, ectomycorrhizal fungi, hardwood forest, soil nutrients, TRFLP,
32
snow cover
33 34
Introduction Soil fungi are important components of temperate forests and play an integral role in
35
organic matter decomposition, carbon (C) and nutrient cycling, and serve as an important food
36
resource for soil organisms (Linderman, 1988; Rayner & Boddy, 1988; Smith & Read, 2008). In
37
addition, soil fungi can affect the distribution of important microbial functional groups, such as
38
those involved in N cycling (Bomberg et al., 2003; Burke et al., 2012). One important group of
39
soil fungi are the mycorrhizal fungi that form mutually beneficial relationships with many forest
40
plants. These fungi colonize plant roots, receive carbon from the plant in the form of sugars and
41
organic acids, and directly benefit the plant by increasing overall nutrient gain as well as plant
42
growth (Smith & Read, 2008). Although a number of studies have explored the environmental
43
factors that govern fungal community composition, diversity and species occurrence in forests
44
(Goodman & Trofymow, 1998; Dickie et al., 2002; Lilleskov et al., 2003; Burke et al., 2009) we
45
still know relatively little about how soil fungi may be affected by global climate change.
46
It may be expected that greater variability in temperature and precipitation associated
47
with global climate change could lead to altered seasonal and inter-annual variability in
48
microbial community composition (Monson et al., 2006), with consequences for the timing and
49
magnitude of many ecosystem processes (Sulkava & Huhta, 2003). Seasonal and inter-annual
50
variability in temperature and precipitation play an important role in regulating nutrient cycling,
51
plant growth, and soil microbial community structure in forests (Groffman et al., 2001b, Tierney
52
et al., 2001; Aerts et al., 2006; Buckeridge & Grogan, 2008). Temperature and precipitation are
53
major controls on microbial growth and reproduction, and may be expected to regulate
54
community structure and species distributions (McArthur, 2006). In addition, temperature and
55
precipitation can affect functional activity of microbes; for example, soil moisture frequently
56
controls production of extracellular enzymes and thus the rate of organic matter decomposition
57
and nutrient cycling (Buckeridge & Grogan, 2008; Steinweg et al., 2012; A’Bear et al., 2014).
58
Alteration in winter temperatures and especially greater soil freezing associated with reduced
59
snow cover can increase nutrient loss from soil (Fitzhugh et al., 2001) but greater soil freezing
60
could also increase plant growth, possibly through increases in nutrient availability following
61
death of soil microbes.
62
However, most studies that have examined climate change effects on forests have not
63
explicitly examined soil fungal communities, but rather have focused on functional activities or
64
changes in biomass (Campbell et al., 2005; Buckeridge & Grogan, 2008) with mycorrhizal
65
community studies often focused on the effects of increased CO2 (Rygiewicz et al., 2000 and
66
Klamer et al., 2002). Yet, recent observational studies of arctic soil fungi suggest that these fungi
67
may be highly sensitive to changes in their environment including changes associated with
68
climate and soil pH (Timling et al., 2014). Loss of winter snow cover associated with climate
69
warming can alter soil freeze thaw dynamics and affect belowground processes (Tierney et al.,
70
2001) and studies have indicated that distinct microbial communities persist under snow cover,
71
and removal of snow cover may change these communities (Monson et al., 2006). Work
72
examining typical annual variation in mycorrhizal communities has found that some species are
73
stable, with little variation between seasons, while other species are more seasonally abundant
74
(e.g. summer or winter)(Izzo et al., 2005; Koide et al., 2007; Richard et al., 2011). Precipitation
75
has also been found to strongly determine fungal biomass in forest soil (Okada et al., 2011). But
76
additional studies on the effects of inter-annual changes in temperature and precipitation,
77
including variation in snow cover, are necessary to better predict the effects of changing climate
78
and snow cover on forest fungal communities.
79
Of additional interest is how changes in precipitation and temperature with variation in
80
snow cover could affect fungal distribution at different depths within the soil profile. Previous
81
studies have found that soil at deeper depths within the profile can have more consistent and
82
potentially higher temperatures during the winter months than soil at more shallow depths
83
(Henry 2007). This could affect soil freeze thaw cycles, and potentially alter fungal distribution
84
if fungal taxa are sensitive to temperature. A number of studies have found that fungal taxa vary
85
with depth within the soil profile, with some fungi more common in mineral as compared to
86
shallow organic layers (Rosling et al., 2003; Shahin et al., 2013). Nonetheless, whether fungal
87
distribution with soil depth is related to soil temperature dynamics has not been well examined.
88
In this study I hypothesized that 1) annual variability in soil moisture and temperature
89
would affect the abundance and distribution of soil fungi; 2) inter-annual variability in snow
90
cover would alter species distributions between years and 3) communities at deeper depths
91
within the soil profile would be less responsive to annual and inter-annual changes as a result of
92
reduced seasonal variability in moisture and temperature. The way that fungi respond to
93
variability in moisture and temperature could enhance our understanding of how fungi may
94
respond to a changing climate that alters those patterns of variability. I collected soil cores every
95
month for three years from permanent plots established in an old-growth hardwood forest in
96
northeastern Ohio. Soil was subdivided into three depths and used to measure chemistry and
97
determine fungal species distribution using terminal restriction fragment length polymorphism
98
(TRFLP) coupled to a site specific database of fungal sequences. Soil temperature, moisture and
99
snow cover were monitored during the three years of sampling.
100
Materials and Methods
101
Site description and soil sampling
102
In 2006 six long-term sampling plots were established within Stebbins Gulch, a 360
103
hectare mature, mixed-mesophytic forest located within The Holden Arboretum in northeastern
104
Ohio, USA (41°36′N and 81°16′W). The study site was established in an approximately 80
105
hectare, old growth, beech-maple forest dominated by American beech (Fagus grandifolia-
106
~75% of canopy coverage) and sugar maple (Acer saccharum- ~15% of canopy coverage) with a
107
herbaceous understory dominated by Allium tricoccum and Dicentra canadensis. The site has
108
moderately drained silt loam soils with total precipitation averaging 116-cm per year, which
109
includes approximately 287-cm of snowfall. For more detailed information about the field site,
110
see Burke et al. (2009, 2012). Soils are acidic, with pH ranging from 3.5 to 5.6 with a mean pH
111
of 4.0 0.1.
112
Long-term sampling plots were established between 50 and 100-m apart, and measured 4
113
x 4 meters in size. Plots were marked in the center by a 2-m steel pole, driven into the ground,
114
and to which was attached HOBO Pro Series air and relative humidity data loggers (Onset
115
Computer Corporation, Bourne, MA) to record air temperature and humidity every 12 minutes
116
throughout the year. Air temperature loggers were set 1.5 m above the ground. Stowaway Tidbit
117
Temperature Loggers (Onset Computer Corporation, Bourne, MA) were used to record soil
118
temperature and were buried 1-m from the center pole in each plot such that a logger was present
119
between 1-3 cm depth (hereafter 2-cm soil depth), 5-7 cm depth (hereafter 6-cm soil depth) and
120
9-11 cm (hereafter 10-cm soil depth). Stowaway data loggers recorded soil temperature every 8
121
minutes throughout the year.
122
Long-term plots were broken into 16 sampling quadrats, each measuring 1 x 1 meter in
123
size. Two soil cores measuring 2-cm in diameter were collected to a depth of 15-cm every month
124
(around the middle of each month) throughout the year with a metal soil corer from within one
125
randomly selected quadrat. The cores were collected from the corners of each quadrat such that
126
they were at least 50-cm apart (Supplemental Figure 1). These samples were subdivided into 4
127
depths: surface litter layer (Oi), and the three soil depths corresponding to 2-, 6- and 10-cm as
128
defined above. The subdivided samples from each core were composited in the field so that one
129
set of four samples were retained for each plot (24 samples total per month, 6 plots x 4 depths).
130
Samples were kept in a cooler on ice, and transported to the laboratory where they were stored at
131
-70ºC until processing. Soil samples have been collected every month from these plots for the
132
past 7 years, and I present data here from November 1 2006 until October 31 2009. The
133
temperature time series presented here begins on November 1 2006 and continues until October
134
31 2009. Soil samples (excluding litter samples) collected from each plot and depth for the entire
135
2006-2007 year (November 2006-October 2007) and from the entire 2008-2009 year (November
136
2008-October 2009) were analyzed for this study for soil chemistry and fungal community
137
structure (6 plots by 3 soil depths by 12 months by 2 years = 432 samples examined). These
138
years were examined because the winters varied in the depth and duration of snow cover, with
139
the 2006-2007 year having abundant snow fall and continuous snow depth between early January
140
and early April, and the 2008-2009 winter having variable snow cover, with snow cover absent
141
at some times during the winter including during the sample events.
142
Analysis of soil environment
143
Frozen and stored soil samples were thawed and used to measure gravimetric water
144
content which is expressed here as [(g water g FW soil-1) x 100%] (Jarrell et al., 1999) and is
145
meant to represent field fresh soil. Soil for C and N was oven dried and pulverized in a Precellys
146
homogenizer (Bertin Technologies, Montigny-le-Bretonneux, France) and analyzed on an ECS
147
4010 CHNSO elemental analyzer (Costech Analytical, Valencia, CA). Labile soil inorganic
148
phosphorous (Pi; readily available) and organic phosphorous (Po; easily mineralizable) were
149
extracted from pulverized oven dried soil by adding 0.5 M NaHCO3 (pH 8.5) and shaking at 100
150
rpm on an orbital shaker (Lab-Line, Melrose Park, IL) for 30 min (Olsen et al., 1954). Pi was
151
determined using the modified ascorbic acid method (Kuo, 1996) directly on the NaHCO3
152
extracts. Total available P (Pt) was determined by further digestion of NaHCO3 extracts with 1.8
153
N H2SO4 and (NH4)2S2O2 (EPA, 1971) while Po was the difference between Pt and Pi. A total of
154
648 samples (6 plots by 3 soil depths by 12 months by 3 years) collected between November
155
2006 and October 2009 were analyzed for soil chemistry. Temperature and relative humidity
156
data were averaged over the course of the day, and daily averages are represented here. Visual
157
observation of snow cover was also made throughout the winter months, and snow gauges
158
attached to each of the long-term plot center poles were used to estimate the depth of snow cover
159
at that sampling time.
160
DNA purification and amplification
161
DNA was extracted from soil using a bead beating protocol. Soil with visible root tissue
162
removed (up to 200-mg fresh weight) was placed in a 1.5-ml bead beating tube containing 500-
163
mg of 400 μM glass beads (VWR, West Chester, Pennsylvania, USA) and 750 µl 2% CTAB
164
(cetyltrimethyl-ammonium bromide). Samples were then beaten for 40s in a Precellys
165
homogenizer at 6500 rpm and purified by phenol/chloroform extraction and precipitation with
166
20% polyethylene glycol 8000 in 2.5 M NaCl with incubation at 37oC. DNA was suspended in
167
200 µl TE buffer and stored at –20oC. To amplify soil fungi, the internal transcribed spacer 2
168
region (ITS2) was targeted using primers 58A2F and NLB4 (Martin & Rygiewicz, 2005; Burke
169
et al., 2005). From previous work, these ITS2 primers have been found to more consistently
170
amplify some fungal species as compared to ITS1 primers, and the region also provided the best
171
discrimination between species when used in conjunction with terminal restriction fragment
172
length polymorphism (TRFLP) typing of fungal communities (Burke et al., 2009). Primers were
173
labeled either with the fluorescent dye 6FAM (58A2F) or HEX (NLB4) and PCR was carried out
174
in 50μl reaction volumes using 1-µl of purified DNA (approximately 100 ng), 0.2 µm of primers,
175
2.0 mM MgCl, 0.2 mM dNTP, 0.15 mg ml-1 Bovine Serum Albumin, and 1.0 units FastStart Taq
176
DNA polymerase (Roche Diagnostics Corporation, Indianapolis, USA) on an PTC 100 Thermal
177
Cycler (MJ Research, Boston, USA). An initial denaturation step of 5 minutes at 94 C was
178
followed by amplification for 35 cycles at the following conditions: 30 seconds at 94 C, 60
179
seconds at 60 C, 90 seconds at 72 C. A final 5-minute extension at 72 C completed the
180
protocol. Although the goal was to target all soil fungi, for the purposes of this study I assume
181
that soil patterns also reflect root tip colonization patterns for ECM fungi.
182
Identification of fungal species in soil samples
183
Labeled PCR product was cut with endonucleases AluI and HaeIII (Promega, Madison
184
WI, USA) and TRFLPs were completed through the Cornell Bioresource Center using an
185
Applied BioSystems 3730xl DNA Analyzer. TRFLPs were analyzed with PeakScanner software
186
version 1 (http://www.appliedbiosystems.com). For each core, 3 useful TRFLP profiles were
187
generated (there is a conserved AluI restriction site in the 28S rDNA that generates little
188
variability for TRFLP), and these profiles were used to identify fungal species using the program
189
Fragsort (Sciarini & Michel, 2002; http://www.oardc.ohio-state.edu/trflpfragsort/index.php)
190
which compared the patterns to a site specific database as previously described and detailed
191
(Burke et al., 2009). The database we used for identifying fungal species for this study contained
192
398 identified fungal isolates separated from sporocarps, mycorrhizal root tips, and soil clones
193
representing 187 species. Use of this database is skewed toward Basidiomycetes and may under
194
represent Ascomycetes present within soil samples. Although plants associated with arbuscular
195
mycorrhizal fungi are present at the field site, the database does not contain arbuscular
196
mycorrhizal taxa and cannot be used to identify them in the soil samples. The identification of
197
fungal species then is limited by the database itself, and is not intended to include all fungi
198
present within soil samples. The strength of the approach however is that we can identify soil
199
fungi to species in most instances, permitting a more precise examination of changes in fungal
200
community structure and fungal distribution.
201
Statistical Analyses
202
Differences in soil chemistry between plots, soil depth, and season were determined using
203
3-Way ANOVA and data were transformed (log10 (moisture, C) or arcsinsqrt (N)) prior to
204
ANOVA to maintain equality of variances. Soil N and water content data are presented as a
205
percent, and both types of data were transformed prior to ANOVA. Chemistry data were
206
analyzed using procedures in SigmaStat 3.5 (Systat Software Inc., CA, USA). To analyze
207
seasonal affects on soil chemistry and fungal communities, samples were grouped into 4
208
categories: winter (December/January/February), spring (March/April/May), summer
209
(June/July/August) and autumn (September/October/November). Differences in fungal richness
210
and Shannon diversity were determined through 3-Way ANOVA. Richness represents the fungal
211
species positively identified in soil cores. Shannon diversity (H') was calculated for identified
212
fungi, where the average TRF peak area was used as a measure of proportional abundance in that
213
core (Burke et al., 2005, 2006). Evenness (E) is represented as Pielou’s J (H'/ln(richness). Both E
214
and H' were calculated using procedures available through PC-ORD 4 (MjM Software, OR,
215
USA). Permutation-based nonparametric MANOVA (PERMANOVA) was used to determine
216
whether depth, season or plot location affected fungal communities and whether sample year
217
affected these relationships. The year sampled was used as a covariate for PERMANOVA
218
analysis which was completed in R using the vegan (v2.0-10) package (Oksanen et al., 2013).
219
Since PERMANOVA found that fungal communities varied significantly between sampled
220
years, we also conducted PERMANOVA on microbial communities separately for each year
221
sampled. Non-metric multidimensional scaling (NMS) procedures available through PC-ORD 4
222
were also used to determine whether fungal communities changed with depth, season or between
223
plots and to visualize these relationships. The “autopilot” mode of PC-ORD was initially used
224
and final results of the ordination confirmed manually. The Sørenson distance with a random
225
starting configuration was used for these analyses. Ordination procedure included 250 runs with
226
real data and 250 runs with randomized data and use of a Monte Carlo test to help select final
227
dimensionality. Additional dimensions of the ordination that did not reduce stress by 5 or more
228
were not considered useful for improving the ordination and the highest dimensionality that met
229
this criterion was used for the final ordination. A total of 500 iterations were used for the final
230
solution. Because the data set includes taxa at the species level, and many fungal species are
231
infrequently encountered, the data set contains many zeros. This has the effect of increasing the
232
stress of the final ordination, which could be reduced by eliminating rare species. However, rare
233
species were included in the data set in order to offer as complete a picture of community
234
structure as possible, even though this results in higher than desired final ordination stress. It
235
should be noted however than NMS ordination generally confirmed PERMANOVA results and
236
provides a useful visualization of fungal community structure despite the higher than desired
237
stress levels. In addition, for NMS, PERMANOVA and diversity estimates the proportional
238
abundance of all detected fungal species was used as an indicator of abundance within each
239
sample (Burke et al., 2005, 2006) and all proportional abundance data were transformed before
240
analysis. In some instances, samples failed to amplify with PCR. In those cases, we calculated
241
the mean fungal abundance for that sample type and used it for the community analysis
242
(PERMANOVA, NMS, diversity). For example, the 2-cm depth averaged across plots would be
243
used for a missing 2-cm sample for that sample time point. Although most samples amplified
244
adequately, 22 from the 2006-2007 year and 11 from the 2008-2009 year did not amplify
245
successfully (out of 432 total samples) and the mean values were used for these samples during
246
analysis. The mean scores were used to maintain a balanced design to meet PERMANOVA
247
requirements. Indicator species analysis using procedures available through PC-ORD 4 was used
248
to determine whether some fungal taxa displayed preference for depth, season or plot. This
249
procedure included a Monte Carlo randomization test to determine whether the distribution
250
pattern of fungal taxa were significant as compared to a randomized data set constructed from
251
observed values. Pearson correlation coefficients were calculated for the named fungal taxa
252
using all samples and time points (n=432; both sampled years) to determine whether significant
253
correlations existed between fungal taxa and soil chemistry. Since intraspecific trait variation is
254
expected to be low, fungal taxa should respond consistently to environmental variability across
255
space and time, justifying using the combined data set to examine taxa correlations with
256
environmental conditions. Proportional abundance was used for these tests with P