Effort versus Reward: Preparing Samples for Fungal Community Characterization in High-Throughput Sequencing Surveys of Soils

Abstract

Next generation fungal amplicon sequencing is being used with increasing frequency to study fungal diversity in various ecosystems; however, the influence of sample preparation on the characterization of fungal community is poorly understood. We investigated the effects of four procedural modifications to library preparation for high-throughput sequencing (HTS). The following treatments were considered: 1) the amount of soil used in DNA extraction, 2) the inclusion of additional steps (freeze/thaw cycles, sonication, or hot water bath incubation) in the extraction procedure, 3) the amount of DNA template used in PCR, and 4) the effect of sample pooling, either physically or computationally. Soils from two different ecosystems in Minnesota, USA, one prairie and one forest site, were used to assess the generality of our results. The first three treatments did not significantly influence observed fungal OTU richness or community structure at either site. Physical pooling captured more OTU richness compared to individual samples, but total OTU richness at each site was highest when individual samples were computationally combined. We conclude that standard extraction kit protocols are well optimized for fungal HTS surveys, but because sample pooling can significantly influence OTU richness estimates, it is important to carefully consider the study aims when planning sampling procedures.

Full text

RESEARCH ARTICLE
Effort versus Reward: Preparing Samples for
Fungal Community Characterization in High-
Throughput Sequencing Surveys of Soils
Zewei Song
1
*, Dan Schlatter
1
, Peter Kennedy
2
, Linda L. Kinkel
1
, H. Corby Kistler
1,3
,
Nhu Nguyen
2
, Scott T. Bates
1
1Department of Plant Pathology, University of Minnesota, Saint Paul, MN 55108, United States of America,
2Department of Plant Biology, University of Minnesota, Saint Paul, MN 55108, United States of America,
3USDA ARS Cereal Disease Laboratory, Saint Paul, MN 55108, United States of America
*songx208@umn.edu
Abstract
Next generation fungal amplicon sequencing is being used with increasing frequency to
study fungal diversity in various ecosystems; however, the influence of sample preparation
on the characterization of fungal community is poorly understood. We investigated the ef-
fects of four procedural modifications to library preparation for high-throughput sequencing
(HTS). The following treatments were considered: 1) the amount of soil used in DNA extrac-
tion, 2) the inclusion of additional steps (freeze/thaw cycles, sonication, or hot water bath in-
cubation) in the extraction procedure, 3) the amount of DNA template used in PCR, and 4)
the effect of sample pooling, either physically or computationally. Soils from two different
ecosystems in Minnesota, USA, one prairie and one forest site, were used to assess the
generality of our results. The first three treatments did not significantly influence observed
fungal OTU richness or community structure at either site. Physical pooling captured more
OTU richness compared to individual samples, but total OTU richness at each site was
highest when individual samples were computationally combined. We conclude that stan-
dard extraction kit protocols are well optimized for fungal HTS surveys, but because sample
pooling can significantly influence OTU richness estimates, it is important to carefully con-
sider the study aims when planning sampling procedures.
Introduction
Terrestrial ecosystems are inhabited by an astonishing diversity of soil microorganisms [1],
which play critical roles in global geochemical cycles and ecosystem functioning [2,3]. A large
portion of this diversity was essentially hidden from microbial ecologists until DNA-based
techniques eliminated the constraints of culture-dependent community surveys [4]. Since then,
the technology for sequencing DNA has advanced at a staggering rate, with progress currently
allowing for the generation of millions of microbial community sequences from environmental
samples in very short periods of time [5].
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 1/13
OPEN ACCESS
Citation: Song Z, Schlatter D, Kennedy P, Kinkel LL,
Kistler HC, Nguyen N, et al. (2015) Effort versus
Reward: Preparing Samples for Fungal Community
Characterization in High-Throughput Sequencing
Surveys of Soils. PLoS ONE 10(5): e0127234.
doi:10.1371/journal.pone.0127234
Academic Editor: Gabriele Berg, Graz University of
Technology (TU Graz), AUSTRIA
Received: December 12, 2014
Accepted: April 12, 2015
Published: May 14, 2015
Copyright: This is an open access article, free of all
copyright, and may be freely reproduced, distributed,
transmitted, modified, built upon, or otherwise used
by anyone for any lawful purpose. The work is made
available under the Creative Commons CC0 public
domain dedication.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information files.
All the data we used in this manuscripts can be
calculated based on the OTU table in the supporting
materials. All richness data used in statistical analysis
can also be found in the supporting materials. The
data can be found in University of Minnesota Data
Repository through this link: http://conservancy.umn.
edu/handle/11299/170817
Funding: Funding was received from United States
Department of Agriculture (2011-67019-30200 http://
nifa.usda.gov/grants) grant to LLK & HCK; University

Soil fungi are an important component of terrestrial microbial diversity due to their consid-
erable influence on aboveground biodiversity and primary productivity, symbiotic relation-
ships with most land plants, and as drivers of the global soil carbon cycle [3,6–8]. The number
of publications using high-throughput sequencing (HTS) to characterize fungal communities
in soil is growing exponentially (S1 Fig), and these studies reveal a high degree of diversity and
endemism as well as functional redundancy among these fungi [9–11]. Although general out-
lines for the fungal community HTS workflow have been published [12], few studies have ex-
amined the protocol used in preparing samples for HTS, with some notable exceptions [13–
15]. Such considerations are important, as standardization of HTS procedures will allow for
cross-study comparisons of microbial HTS datasets, which will ultimately benefit many emerg-
ing frontiers in microbial ecology [16].
Modification of sample preparation procedures has been previously shown to influence fun-
gal community characterization in culture-independent non-HTS studies. For example, in-
creasing the amount of soil used in DNA extraction [17–19] or the use of sample pooling
methods [20] can enhance capture of rare species. Additional steps in extraction protocol have
also been shown to increase DNA yield [21–23] as well as the number of species recovered
[24–26]. Conversely, species richness gains can also be achieved through regulating DNA tem-
plate amount to improve PCR efficiency [12,27]. The influence of these different procedural
modifications on our ability to characterize the richness of soil fungal communities, however,
has not been sufficiently examined for the deep sequencing levels now possible using HTS.
To assess the role of modified procedures on species richness capture and address the need
for standardization, we examined sample preparation protocols used in HTS surveys of soil
fungi. Samples were collected at two sites in the Midwestern United States representing prairie
and forest soils. Specifically, we examined the effect of extracting DNA from different volumes
of soil (10 g, 1 g, 0.25 g) and using various extraction protocol modifications (freeze/thaw, soni-
cation, 65°C water bath) on recovered fungal richness. Along with these treatments, we also ex-
amined fungal richness outcomes using varying PCR DNA template amounts (10 ng, 20 ng,
40 ng) and compared different sample pooling methods (physical vs. computational) to assess
their ability to capture fungal community richness and composition at both sites.
Methods
Sample collection in Cedar Creek grant through Cedar Creek Ecosystem Science Reserve of the
University of Minnesota. Sample collection in Cloquet forest grant through Cloquet Forest
Center of the University of Minnesota.
Soil sampling
In fall 2013, soils were collected from two sites: the University of Minnesota Cedar Creek Eco-
system Science Reserve (CCR, 45°24'13" N, 93°11'20" W) and the University of Minnesota Clo-
quet Forestry Center (CFC, 46°40'45" N, 92°31'08" W). These sites represent distinct biome
types, prairie and forest, respectively. CCR has mean monthly temperature range from -10°C
to 22°C with an annual mean temperature at 6.7°C and an annual precipitation of 800 mm.
The soil is highly sandy (93%) of the Zimmerman series [28]. CFC has mean monthly tempera-
tures range from -14°C to 19°C with an annual mean temperature at 4°C and an annual precip-
itation of 761 mm. The soil is a loamy sand of the Omega and Cloquet series [29]. At CCR,
soils were collected from the base of Andropogon gerardii plants growing in monoculture
(n = 2 samples) and polyculture (n = 2 samples). For each sample, three 10 cm deep × 2.5 cm
wide cores were taken from the base of an individual A.gerardii plant and combined in the
field. At CFC, soils were collected under four replicate plots containing six tree species (Pinus
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 2/13
of Minnesota MnDrive (http://mndrive.umn.edu/) grant
to LLK & STB; and National Science Foundation
(Grants EF 12-41895 grant to LLK and Long-Term
Ecological Research Network 0620652). The funders
had no role in study design, data collection and
analysis, decision to publish, or preparation of the
manuscript.
Competing Interests: The authors have declared
that no competing interests exist.

strobus,Larix laricina,Picea glauca,Quercus rubra,Betula papyifera, and Acer saccharum). For
each sample, four 10 cm deep × 2.5 cm wide cores were taken one meter apart and combined
in the field. At each site, four individual plot sample replicates were gathered. All soil samples
were transported in coolers to the lab and stored at -20°C until sieving at 4 mm size to remove
fine roots and any part of small woody debris prior to DNA extraction.
Experimental design and DNA extraction
We designed our experiments to examine the effects of soil volume, DNA extraction modifica-
tions, PCR DNA template amount, and soil sample pooling on observed fungal richness and
community composition. To assess soil volume, DNA was extracted from 10 g, 1 g, and 0.25 g
of soil from each plot (12 samples sequenced per site, 24 samples sequenced in total). The
PowerMax Soil DNA kit (MO BIO Laboratories, Carlsbad, CA, USA) was used for 10 g extrac-
tions. For the 1 g extractions, the PowerSoil kit (MO BIO Laboratories, Carlsbad, CA, USA)
was used to extract DNA from four replicate 0.25 g portions of the same plot soil sample
(0.25 g is the standard quantity recommended for this kit by the manufacturer). DNA concen-
tration for each of the four replicate extractions were then quantified using a BioTek Synergy
H1 plate reader (BioTek, Winoosku, VT, USA) and mixed at equal amount to prepare a single
‘1g’sample. An additional 0.25 g portion of the same plot soil sample was used in the Power-
Soil kit to represent the kit ‘standard’extraction. All extractions were carried out according to
the manufacturer’s instruction. Homogenizing was done, as specified in the kit protocol, by
vortexing 10 min on the MO BIO vortex adapters (MO BIO Laboratories, Carlsbad, CA, USA,
13000-V1-24 for PowerSoil kit, and 13000-V1-50 for PowerMax Soil kit).
In the DNA extraction kit modification experiment, 0.25 g samples of soil were added to the
PowerSoil ‘bead-beater’extraction tubes containing the kit buffer. Samples were then subjected
to three different modification procedures, freeze/thaw, sonication, or heating via water bath
incubation (12 treatment samples sequenced per site, 24 treatment samples sequenced in
total). The freeze/thaw treatment consisted of three rounds of freezing and thawing by sub-
mersing tubes in liquid N
2
for 5–10 seconds followed by an incubation step in a 65°C water
bath for 10 min. The sonication treatment consisted of sonicating tubes for 10 min in a Bran-
son 8200 sonicator (Thomas Scientific, Swedesboro, NJ, USA). Finally, the heating treatment
consisted of incubating tubes in a 65°C water bath for 10 min. After each modification, DNA
extraction proceeded according to the kit protocol instructions. The standard 0.25 g soil extrac-
tion from the soil volume experiment above served as an unmodified control for comparison.
To assess the effects of soil sample pooling and DNA template amount, four individual plot
samples were combined in equal amounts (3 g from each sample) and thoroughly mixed to
form a ‘bulked’sample (hereafter referred to as the ‘physical’pool) for each site. From these
physical pool samples, DNA was extracted from 10 g, 1 g, and 0.25 g soil as describe above.
DNA extracted from each of these three different soil volumes was then used as template in in-
dividual PCR reactions containing different template DNA amounts (10 ng, 20 ng, and 40 ng).
Aliquots were prepared from a standardized 10 ng uL
-1
stock concentration and added to the
PCR reaction at 1 μL, 2 μL, and 4 μL volumes to achieve the desired quantity of DNA template
(9 samples sequenced per site, 18 samples sequenced in total). Along with all of the aforemen-
tioned samples, for each extraction kit that was used (PowerSoil and PowerMax Soil), a soil-
free extraction was conducted using the manufacturer’s protocol to serve as negative controls.
These negative controls allowed us to identify contaminants introduced during the extraction
process and remove them from the final analyses. Finally, a ‘mock’fungal community consist-
ing of DNA extracted from 25 known fungal taxa was also included in the sequencing run to
allow for internal optimization of the OTU clustering method [14].
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 3/13

Fungal ITS amplification and Illumina sequencing
PCR amplification of the ITS1 region of the nuclear ribosomal RNA gene was conducted using
the general fungal primers ITS1-F and ITS2 adapted for Illumina sequencing [13]. Using other
primer combinations mays improve amplification of certain fungal groups [30]; however, test-
ing differences among primer pairs was not the aim of this study. Each 20 μL PCR reaction
consisted of 10 μL 2x Roche FastStart PCR Master (Roche, Indianapolis, IN, USA), 0.35 μLof
each forward and reverse primers (10 μM), 1.2 μL MgCl
2
(25 mM), and 2 μL template DNA
(20 ng, except in the DNA template amount experiment where 10 ng and 40 ng were also used,
see above). Triplicate PCR reactions were performed for each sample using three separate an-
nealing temperatures (50°C, 53°C, and 55°C)[14]. Thermocycling conditions included an ini-
tial denaturation step of 95°C for 10 min, 30 cycles of 95°C for 30s, annealing at 50°C, 53°C, or
55°C for 20 s, and extension at 72°C for 30 s, and a single final extension at 72°C for 8 minutes.
Triplicate PCR reactions at each annealing temperature were pooled, purified using the Agin-
court AMPure XP PCR purification kit (Beckman Coulter Inc., Brea, CA, USA). Purified PCR
products were quantified using the PicoGreen dsDNA assay (Life Technologies, Grand Island,
NY, USA) following the manufacturer’s instructions, and 30 ng of DNA from each sample was
combined into a single pool for Illumina MiSeq sequencing at the University of Minnesota Bio-
medical Genomics Center.
Data processing
Low quality ends of reads (<q20), primer sites, and adapter sites were trimmed from forward
MiSeq Illumina reads (match error rate of e = 0.2) using Cutadapt [31]. Trimmomatic [32] was
then used to trim from both ends (quality threshold set at 20) and filter out reads less than
125 bp long. Sequences with ambiguous bases and homopolymers greater than 9 bp were re-
moved in Mothur [33]. All sequences files were then combined into a single fasta file with valid
QIIME labels [34]. To generate a de novo database of operational taxonomic units (OTUs), se-
quences were dereplicated and singletons were removed in USEARCH [35]. Sequences were
then clustered at 97% similarity using the UPARSE algorithm implemented in USEARCH,
which included chimera detection and filtering. Subsequently, to reduce the number of spuri-
ous OTUs, a second round of clustering was conducted at 95% similarity using the UCLUST al-
gorithm in QIIME, as Nguyen et al. (2014) found that ‘chaining’OTU clustering algorithms
generated more accurate OTU bins in a similarly processed dataset [14]. The representative
OTU list after the ‘chaining’clustering were filtered against a customized UNITE database [36]
with UCLUST, but at a 75% similarity level to remove non-fungal sequences. Any OTU that
failed in the third clustering round were removed from the de novo OTU list and blasted
against another custom database that includes fungal voucher sequences and the UNITE data-
base [14]. OTUs that hit the custom database and passed a quality filter (match length >0.85,
similarity level >0.75%, and kingdom ‘Fungi’identity in the taxonomic string) were added
back to the de novo OTU list. Mapping the original sequence file to the de novo OTU sequence
list generated an OTU table. Finally, the abundances of each OTU found in negative controls
were subtracted from all samples prior to rarefaction following Nguyen et al. (2014)[14].
Statistical analysis
All samples were rarefied to an equal sampling depth of 18,778 sequences, and observed fungal
OTU richness, Chao1, Simpson as well as Shannon indexes for each sample were calculated
using the QIIME alpha_diversity.py command. To assess differences among treatments, a se-
ries of one-way ANOVAs were applied on OTU richness with Tukey’s multiple comparisons
in the R statistical software (www.r-project.org). Diversity estimates, such as Chao 1 was not
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 4/13

used in our analyses due to the potential biases associated with them for next generation se-
quencing studies [37]; however, we have provided them in the supplementary materials (S1
File.) for general comparison. A Bray-Curtis dissimilarity matrix among samples was calculat-
ed, and community distance was visualized in non-metric multidimensional scaling (NMDS)
plots using the vegan [38] and ggplot2 [39] packages in R. The vegan package was also used to
assess potential community composition differences among treatments by testing with ADO-
NIS. For each site, the physical pooled samples sequenced from the standard 0.25 g soil sample
(using a 20 ng DNA template amount for PCR) were also compared to a corresponding pool of
the individual plot samples (0.25 g soil and 20 ng DNA template) achieved through computa-
tion. For this ‘computational’pool, the number of each OTU recovered from the four individu-
al plot samples were summed, and the resultant composite pool was rarefied at the
18,778-sequence level. We also compared the average occurrence frequency of OTUs between
physical and individual pools. The occurrence frequency of each OTU is defined as the occur-
rence (sequence number 1) in either 9 physical pools or 24 individual pools in the form of
percentage in both site. The average occurrence frequency was calculated as the average of 20
OTUs around each individual OTU starting from the 10
th
(i.e., 1
st
to 20
th
OTU, 2
nd
to 21
st
OTU, ..., 100
th
to 119
th
...). Paired T-test was applied between each pair of the 20-OTUs sub-
set (α= 0.05). All T-tests were adjusted with Bonferroni correction with hypotheses n= 20.
Results and Discussion
High-throughput sequencing of soil fungal communities
After quality filtering and sequence processing, we obtained 2,958,381 sequences, with a range
of 18,778 to 78,509 sequences per sample. Rarefying at the lowest per sample recovery thresh-
old (18,778 sequences), we recovered a total of 2,009 unique fungal OTUs from our 66 soil
samples. These OTUs represented five known phyla (S2 Fig). We found a total of 1,273 and
1,436 OTUs in CCR (Cedar Creek Ecosystem Science Reserve) and CFC (Cloquet Forestry
Center) soils (33 samples at each site), respectively. On average, individual samples contained
~330 OTUs for CCR and ~390 OTUs for CFC. Due to substantial community turnover among
samples at each site, soils at CCR and CFC harbored a very high degree of total fungal richness,
with rarefaction curves failing to achieve an asymptote (S3 Fig). These results agree with other
high-throughput sequencing studies [11,40–42] that consistently reveal extremely high levels
of fungal richness in soils of various ecosystems. In fact, only one study to date has successfully
saturated the rarefaction curve for soil fungi, recovering a total of 1,002 OTUs from a single
biome type [43], a richness level that is considerably smaller than what we detected at our sites.
Treatment effects on fungal alpha diversity
Extracting DNA from larger volumes of soil did not result in significantly higher OTU richness
for either site (Fig 1A and 1B for richness and S1 Table for other indexes), although average
richness did increase slightly with more soil used in the extraction. Extracting DNA from more
soil may be desirable if detecting low abundance taxa is critical. For example, methods for early
detection of soil-borne plant pathogens have been developed that require using large volumes
of soil [17–19]. In support of this approach, we were able to detect potential plant pathogens,
such as Cochliobolus sativus, in our 10 g and 1 g soil extraction samples that were not present
when only 0.25 g of soil were used. However, our data suggest that extracting DNA from small
soil volumes (e.g., the standard 0.25 g recommended in widely-used commercial extraction
kits) is sufficient for most HTS applications in which a general characterization of soil fungal
richness is required.
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 5/13

Including additional steps in the extraction protocol, such as sonication treatments, freeze/
thaw regimes, or period of incubation in a hot water bath, all of which are thought to increase
DNA yields and recovered richness, likewise did not significantly impact observed fungal rich-
ness (Fig 1C and 1D for richness & S1 Table for other diversity indexes). Although previous
studies have found significant effects of DNA extraction method on recovered fungal and bac-
terial richness [21,24–26,44–46], those studies used older molecular techniques (e.g., RFLP and
DGGE) that are likely more sensitive to biases due to the limited depth at which they survey
the communities [47–49]. While our results suggest unmodified protocols from standard DNA
extraction kits is sufficient for general characterizing fungal richness of soil samples in HTS
surveys, we reiterate that study aims must be considered. For example, more stringent DNA
Fig 1. Rarefied OTU richness estimates of soil fungi by treatment. A-B. Amount of soil extracted for CCR
and CFC, respectively (n = 5 for each treatment). C-D. Extraction modification for CCR and CFC, respectively
(n = 4 for each treatment). E-F. Amount of DNA template in each PCR reactions for CCR and CFC,
respectively (n = 3 for each treatment). The same letter above each bar indicates no significant difference
(P0.01) in ANOVA using Tukey’s multiple comparison tests (α= 0.05).
doi:10.1371/journal.pone.0127234.g001
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 6/13

extraction treatments may be necessary to detect fungi exhibiting resistant structures [50–52];
however, facilitating extraction of DNA from such structures (e.g., resistant spores) is often un-
desirable as the associated fungal taxa may represent transient, rather than living, components
of the community.
As with the other preparation treatments, the amount of DNA template used in PCR reac-
tions (10 ng, 20 ng, or 40 ng) did not significantly influence recovered fungal richness in our
samples either (Fig 1E and 1F for richness & S1 Table for other indexes). While significant ef-
fects on fungal richness were not detected overall, opposing trends were noted between sites,
with slightly lower richness observed for the highest template amount at CCR and slightly
higher richness recorded for higher template amounts at CFC. It is possible that these results
are related to the fact that increasing DNA template amount can adversely affect PCR by also
increasing the concentrations of inhibitors [27,53,54] and Lindahl et al. (2013) noted that dilut-
ing DNA template for PCR steps in HTS library preparation is often required [12]. While we
did not directly measure potential inhibitors in our samples, our results do highlight the often
inconsistent results of increasing DNA templates amounts among different sample types as
well as the need for template optimization in preparing HTS runs.
Sample pooling and treatment effects on fungal community composition
At both sites, physical pools generally had higher richness (Fig 2) and more unique OTUs (S4
Fig) when compared with each individual plot sample. Abundant OTUs (>1% of the total se-
quence abundance, consisted of ~70% sequences in total for both sites) appearing in each indi-
vidual plot sample were also more fully captured in the physical pools at both sites (Fig 3 and
S2 File.). Between the two pooling methods, computational pools of each site had higher rich-
ness measures than physical pools (Fig 2), due mostly to the contribution of unique OTUs
from all four individual plot samples (S4 Fig). A number of OTUs were also recovered in the
computation pool that were not present in the physical pool, while a smaller portion were
Fig 2. Soil fungal OTU richness of individual samples, physical pool, and computational pool. The
shorter bars inside of physical and computational pools indicated the number of unique OTUs that were not
shared between the two pools in each site. All samples used 0.25 g soil with standard DNA extraction and
PCR procedures.
doi:10.1371/journal.pone.0127234.g002
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 7/13

unique to the physical pool alone (Fig 2). Since soil from each of the study sites originated from
a very similar source (see methods), it is unlikely that the differences on unique OTUs could be
due to data process procedures or PCR amplifications. Our findings are similar to Branco et al.
(2013), who found that computational pooling yielded higher fungal OTU richness than physi-
cal sample pool and Dickie et al. (2010), who found different rare taxa in pooled and individual
samples [20,55]. Importantly, the high level of fungal community turnover demonstrated in
this study and others [11,40–42] indicates the difficultly in achieving comprehensive character-
ization of soil fungal communities, even using the deep level of survey possible with HTS.
In addition to differences in richness, physically pooled and individual plot samples for each
site differed significantly in their fungal community structure (ADONIS based on Bray Curtis
and presence/absence distance matrixes respectively, F
4, 28
= 17.71 and 6.10 for CCR, F
4,28
=
81.06 and 6.71 for CFC, P= 0.001 for all tests). Our results are again similar to those of Branco
et al. (2013), who showed significant differences in fungal community structure between
Fig 3. Average occurrence frequency of OTUs in physical pools versus individual samples, and percent of total number of sequence, moving from
abundant (left) to rare (right) taxa. Physical pools are in red and blue points, individual samples are in green points, total number of sequence is in gray
points. The occurrence frequency of each OTU is defined as the occurrence (sequence number 1) in either 9 physical pools or 24 individual pools in the
form of percentage in both site. The average occurrence frequency shown in the plot was calculated as the average of 20 OTUs around each individual OTU.
In the physical pool, red points indicate significant difference (P0.05) in paired T-test between physical and individual pools (α= 0.05), while blue points
were not significantly different. All T-tests were adjusted with Bonferroni correction with hypotheses n= 20. The percent of total number of sequence was
defined as the sum of abundance percentage from 1
st
to current OTU in descending order of abundance.
doi:10.1371/journal.pone.0127234.g003
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 8/13

Fig 4. Nonmetric Multidimensional Scaling Plot of physical pools and individual plot samples from each site. Plots based on either Bray-Curtis
distance matrices generated from rarefied taxon abundances of A. CCR and B. CFC or from presence/absence matrix of C. CCR and D. CFC.
doi:10.1371/journal.pone.0127234.g004
Fig 5. Nonmetric Multidimensional Scaling Plot of samples from CCR and CFC. Plot is based on a Bray-
Curtis distance matrix generated from rarefied taxon abundances.
doi:10.1371/journal.pone.0127234.g005
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 9/13

individual versus physically pooled samples [20]; however, the differences among individual
samples taken at their site were not significant. In our study, significant compositional differ-
ences among individual samples are apparent in the distinct clustering patterns of the NMDS
plots (Fig 4). The importance of individual taxa, rather than their respective abundances, in
driving these clustering patterns (Fig 4) is demonstrated by the consistency of the results be-
tween the two different matrix types (Bray Curtis vs. presence/absence). In addition, the tight
clusters formed by the physical pool samples in the central ‘ordination’spaces of Fig 4, suggests
the averaging effect that physical pooling had on the fungal communities recovered. Overall,
site differences in fungal community structure were much stronger than any other effect ob-
served within the sites (Fig 5).
Conclusions
When considering the effort applied in the sample preparation phase of this survey, we found
that the effort generally outweighed any associated rewards. Specifically, we found that sonica-
tion, freeze/thaw cycling, and heating during DNA extraction did not result in notable gains of
richness or structural characterization of soil fungal communities. Likewise, community-level
results were very similar regardless of the amount of soil used in the DNA extraction protocol
or the amount of DNA template used in the PCR reactions. We did, however, find that pooling
individual soil samples, either physically or computationally, resulted in higher richness. Al-
though computational pooling of samples provides a more comprehensive assessment of local
OTU richness than physical pooling, there are additional associated effort of preparation and
sequencing. We therefore suggest that for complete biotic inventories, collecting a high number
of individual samples at a given site would be preferable, while a single pooled sample from
many sites would be preferable for studies aimed at larger spatial scales. Overall, we conclude
that following standard protocols of widely used DNA extraction kits is sufficient to successful-
ly assess soil fungal community richness and composition in HTS surveys.
Supporting Information
S1 Fig. Number of publications related to fungal HTS studies from 2000 to 2013 found by
searching key words—fungi OR fungal, "high throughput sequencing"—in Google Scholar
(http://scholar.google.com/). The points plotted fit an exponential curve (α= 0.05; P0.01).
(TIF)
S2 Fig. Relative abundance of fungal phyla in Cedar Creek Reserve (CCR) and Cloquet For-
est Center (CFC) soils after rarefaction (n = 619,674 sequences/site).
(TIF)
S3 Fig. Taxon (OTU) accumulation curves in this study. Taxon (OTU) accumulation curves
depicting the rarefied richnessof fungal taxa as a function of increased sample sequencing ef-
fort at CCR (brown) and CFC (green) in the Midwestern United States (sample
depth = 18,778 sequences per sample).
(TIF)
S4 Fig. Number of unique and shared OTUs between individual samples or computational
pool compared with the physical pool. Samples using 0.25 g soil with standard extraction and
PCR procedure were used. Computational pool was generated with the four individual samples
and rarefied at the same level of richness (18,778 sequences).
(TIF)
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 10 / 13

S1 Table. One-way ANOVA of alpha diversity indexes and ADONIS in the three treatments
on sequencing sample preparations.
(DOCX)
S1 File. Diversity indexes in different treatments.
(XLSX)
S2 File. OTU tables for CCR and CFC.
(XLSX)
Acknowledgments
We thank the University of Minnesota Supercomputing Institute (MSI) for providing
computing support.
Author Contributions
Conceived and designed the experiments: DS PK LK HCK NN STB. Performed the experi-
ments: ZS DS PK NN STB. Analyzed the data: ZS DS PK NN STB. Contributed reagents/mate-
rials/analysis tools: PK NN STB. Wrote the paper: ZS DS PK LK HCK NN STB.
References
1. Fierer N, Bradford MA, Jackson RB. Toward an ecological classification of soil bacteria. Ecology. 2007;
88: 1354–64. PMID: 17601128
2. Falkowski PG, Fenchel T, Delong EF. The microbial engines that drive Earth’s biogeochemical cycles.
Science. 2008; 320: 1034–1039. doi: 10.1126/science.1153213 PMID: 18497287
3. Wagg C, Bender SF, Widmer F, van der Heijden MGA. Soil biodiversity and soil community composi-
tion determine ecosystem multifunctionality. Proc Natl Acad Sci. 2014; 111: 5266–70. doi: 10.1073/
pnas.1320054111 PMID: 24639507
4. Curtis TP, Head IM, Lunn M, Woodcock S, Schloss PD, Sloan WT. What is the extent of prokaryotic di-
versity? Phil Trans R Soc Lond A. 2006; 361: 2023–37. doi: 10.1098/rstb.2006.1921 PMID: 17028084
5. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. Ultra-high-throughput
microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012; 6: 1621–4.
doi: 10.1038/ismej.2012.8 PMID: 22402401
6. Van der Heijden MGA, Klironomos J. Mycorrhizal fungal diversity determines plant biodiversity, ecosys-
tem variability and productivity. Nature. 1998; 74: 69–72.
7. Van der Heijden MGA, Bardgett RD, van Straalen NM. The unseen majority: soil microbes as drivers of
plant diversity and productivity in terrestrial ecosystems. Ecol Lett. 2008; 11: 296–310. doi: 10.1111/j.
1461-0248.2007.01139.x PMID: 18047587
8. De Vries FT, Manning P, Tallowin JRB, Mortimer SR, Pilgrim ES, Harrison K a, et al. Abiotic drivers and
plant traits explain landscape-scale patterns in soil microbial communities. Ecol Lett. 2012; 15: 1230–9.
doi: 10.1111/j.1461-0248.2012.01844.x PMID: 22882451
9. McGuire KL, Payne SG, Palmer MI, Gillikin CM, Keefe D, Kim SJ, et al. Digging the New York City Sky-
line: soil fungal communities in green roofs and city parks. PLoS One. 2013; 8: e58020. doi: 10.1371/
journal.pone.0058020 PMID: 23469260
10. Kemler M, Garnas J, Wingfield MJ, Gryzenhout M, Pillay KA, Slippers B. Ion Torrent PGM as tool for
fungal community analysis: a case study of endophytes in Eucalyptus grandis reveals high taxonomic
diversity. PLoS One. 2013; 8: e81718. doi: 10.1371/journal.pone.0081718 PMID: 24358124
11. Talbot JM, Bruns TD, Taylor JW, Smith DP, Branco S, Glassman SI, et al. Endemism and functional
convergence across the North American soil mycobiome. Proc Natl Acad Sci. 2014; 111: 6341–6. doi:
10.1073/pnas.1402584111 PMID: 24733885
12. Lindahl BD, Nilsson RH, Tedersoo L, Abarenkov K, Carlsen T, Pennanen T, et al. Fungal community
analysis by high throughput sequencing of amplified markers–a user’s guide. New Phytol. 2013; 199:
288–299. doi: 10.1111/nph.12243 PMID: 23534863
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 11 / 13

13. Smith DP, Peay KG. Sequence Depth, Not PCR Replication, Improves Ecological Inference from Next
Generation DNA Sequencing. PLoS One. 2014; 9: e90234. doi: 10.1371/journal.pone.0090234 PMID:
24587293
14. Nguyen N, Smith D, Peay K, Kennedy P. Parsing ecological signal from noise in next generation ampli-
con sequencing. New Phytol. 2014; doi: 10.1111/nph.12923
15. Vishnivetskaya TA, Layton AC, Lau MCY, Chauhan A, Cheng KR, Meyers AJ, et al. Commercial DNA
extraction kits impact observed microbial community composition in permafrost samples. FEMS Micro-
biol Ecol. 2014; 87: 217–30. doi: 10.1111/1574-6941.12219 PMID: 24102625
16. Knight R, Jansson J, Field D, Fierer N, Desai N, Fuhrman J a, et al. Unlocking the potential of metage-
nomics through replicated experimental design. Nat Biotechnol. 2012; 30: 513–20. doi: 10.1038/nbt.
2235 PMID: 22678395
17. Woodhall JW, Webb KM, Giltrap PM, Adams IP, Peters JC, Budge GE, et al. A new large scale soil
DNA extraction procedure and real-time PCR assay for the detection of Sclerotium cepivorum in soil.
Eur J Plant Pathol. 2012; 134: 467–473. doi: 10.1007/s10658-012-0025-2
18. Than DJ, Hughes KJD, Boonhan N, Tomlinson JA, Woodhall JW, Bellgard SE. A TaqMan real-time
PCR assay for the detection of Phytophthora “taxon Agathis”in soil, pathogen of Kauri in New Zealand.
Andrea V, editor. For Pathol. 2013; 43: 324–330. doi: 10.1111/efp.12034
19. Boine B, Renner AC, Zellner M, Nechwatal J. Quantitative methods for assessment of the impact of dif-
ferent crops on the inoculum density of Rhizoctonia solani AG2-2IIIB in soil. Eur J Plant Pathol. 2014;
doi: 10.1007/s10658-014-0506-6
20. Branco S, Bruns TD, Singleton I. Fungi at a Small Scale: Spatial Zonation of Fungal Assemblages
around Single Trees. PLoS One. 2013; 8: 1–10. doi: 10.1371/journal.pone.0078295
21. Fredricks D, Smith C, Meier A. Comparison of six DNA extraction methods for recovery of fungal DNA
as assessed by quantitative PCR. J Clin Microbiol. 2005; 43: 5122–5128. doi: 10.1128/JCM.43.10.
5122 PMID: 16207973
22. Karakousis A, Tan L, Ellis D, Alexiou H, Wormald PJ. An assessment of the efficiency of fungal DNA ex-
traction methods for maximizing the detection of medically important fungi using PCR. J Microbiol Meth-
ods. 2006; 65: 38–48. doi: 10.1016/j.mimet.2005.06.008 PMID: 16099520
23. Francesconi A, Kasai M, Harrington SM, Beveridge MG, Petraitiene R, Petraitis V, et al. Automated and
manual methods of DNA extraction for Aspergillus fumigatus and Rhizopus oryzae analyzed by quanti-
tative real-time PCR. J Clin Microbiol. 2008; 46: 1978–84. doi: 10.1128/JCM.02246-07 PMID:
18353931
24. Martin-Laurent F, Philippot L. DNA extraction from soils: old bias for new microbial diversity analysis
methods. Appl Environ Microbiol. 2001; 67: 2354–2359. doi: 10.1128/AEM.67.5.2354 PMID: 11319122
25. Roh C, Villatte F, Kim B, Schmid RD. Comparative study of methods for extraction and purification of
environmental DNA from soil and sludge samples. Appl Biochem Biotechnol. 2006; 134: 97–112.
PMID: 16943632
26. Sagova-Mareckova M, Cermak L, Novotna J, Plhackova K, Forstova J, Kopecky J. Innovative methods
for soil DNA purification tested in soils with widely differing characteristics. Appl Environ Microbiol.
2008; 74: 2902–7. doi: 10.1128/AEM.02161-07 PMID: 18344341
27. Wilson I. Inhibition and facilitation of nucleic acid amplification. Appl Environ Microbiol. 1997; 63: 3741–
51. PMID: 9327537
28. Carol Adair E, Reich PB, Trost JJ, Hobbie SE. Elevated CO
2
stimulates grassland soil respiration by in-
creasing carbon inputs rather than by enhancing soil moisture. Glob Chang Biol. 2011; 17: 3546–3563.
doi: 10.1111/j.1365-2486.2011.02484.x
29. Dijkstra F a., West JB, Hobbie SE, Reich PB. Antagonistic effects of species on C respiration and net N
mineralization in soils from mixed coniferous plantations. For Ecol Manage. 2009; 257: 1112–1118. doi:
10.1016/j.foreco.2008.11.014
30. Bellemain E, Carlsen T, Brochmann C, Coissac E, Taberlet P, Kauserud H. ITS as an environmental
DNA barcode for fungi: an in silico approach reveals potential PCR biases. BMC Microbiol. 2010; 10:
189. doi: 10.1186/1471-2180-10-189 PMID: 20618939
31. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads.EMBnet.jour-
nal. 2011; 17: 10–12. http://dx.doi.org/10.14806/ej.17.1.200
32. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinfor-
matics. 2014; 30: 2114–2120. doi: 10.1093/bioinformatics/btu170 PMID: 24695404
33. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-
source, platform-independent, community-supported software for describing and comparing microbial
communities. Appl Environ Microbiol. 2009; 75: 7537–41. doi: 10.1128/AEM.01541-09 PMID:
19801464
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 12 / 13

34. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows
analysis of high-throughput community sequencing data. Nat Methods. 2010; 7: 335–6. doi: 10.1038/
nmeth.f.303 PMID: 20383131
35. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010; 26:
2460–1. doi: 10.1093/bioinformatics/btq461 PMID: 20709691
36. Kõljalg U, Nilsson RH, Abarenkov K, Tedersoo L, Taylor AFS, Bahram M, et al. Towards a unified para-
digm for sequence-based identification of fungi. Mol Ecol. 2013; 22: 5271–5277. doi: 10.1111/mec.
12481 PMID: 24112409
37. Dickie IA. Insidious effects of sequencing errors on perceived diversity in molecular surveys. New Phy-
tol. 2010; 188: 916–918. doi: 10.1111/j.1469-8137.2010.03473.x PMID: 20854395
38. Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, et al. vegan: Community Ecol-
ogy Package. R package version 2.0–7. 2013;
39. Wickham H. ggplot2: elegant graphics for data analysis. Springer New York; 2009.
40. Jumpponen A, Jones KL, David Mattox J, Yaege C. Massively parallel 454-sequencing of fungal com-
munities in Quercus spp. ectomycorrhizas indicates seasonal dynamics in urban and rural sites. Mol
Ecol. 2010; 19 Suppl 1: 41–53. doi: 10.1111/j.1365-294X.2009.04483.x PMID: 20331769
41. Tedersoo L, Nilsson RH, Abarenkov K, Jairus T, Sadam A, Saar I. Methods 454 Pyrosequencing and
Sanger sequencing of tropical mycorrhizal fungi provide similar results but reveal substantial methodo-
logical biases. New Phytol. 2010; 188: 291–301. doi: 10.1111/j.1469-8137.2010.03373.x PMID:
20636324
42. Xu L, Ravnskov S, Larsen J, Nilsson RH, Nicolaisen M. Soil fungal community structure along a soil
health gradient in pea fields examined using deep amplicon sequencing. Soil Biol Biochem. 2012; 46:
26–32. doi: 10.1016/j.soilbio.2011.11.010
43. Taylor D, Hollingsworth T, McFarland J. A first comprehensive census of fungi in soil reveals both
hyperdiversity and fine-scale niche partitioning. Ecology. 2013; 84: 3–20.
44. He J, Xu Z, Hughes J. Pre-lysis washing improves DNA extraction from a forest soil. Soil Biol Biochem.
2005; 37: 2337–2341. doi: 10.1016/j.soilbio.2005.04.016
45. Carrigg C, Rice O, Kavanagh S, Collins G, O’Flaherty V. DNA extraction method affects microbial com-
munity profiles from soils and sediment. Appl Microbiol Biotechnol. 2007; 77: 955–64. doi: 10.1007/
s00253-007-1219-y PMID: 17960375
46. Lakay FM, Botha A, Prior BA. Comparative analysis of environmental DNA extraction and purification
methods from different humic acid-rich soils. J Appl Microbiol. 2007; 102: 265–73. doi: 10.1111/j.1365-
2672.2006.03052.x PMID: 17184343
47. Muyzer G. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analy-
sis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl Environ Microbiol. 1993;
59: 695–700. PMID: 7683183
48. Sekiguchi H, Tomioka N, Nakahara T, Uchiyama H. A single band does not always represent single
bacterial strains in denaturing gradient gel electrophoresis analysis. Biotechnol Lett. 2001; 23: 1205–
1208.
49. Pilloni G, Granitsiotis MS, Engel M, Lueders T. Testing the limits of 454 pyrotag sequencing: reproduc-
ibility, quantitative assessment and comparison to T-RFLP fingerprinting of aquifer microbes. PLoS
One. 2012; 7: e40467. doi: 10.1371/journal.pone.0040467 PMID: 22808168
50. Sánchez B, Rodriguez M. Development of an efficient fungal DNA extraction method to be used in ran-
dom amplified polymorphic DNA–PCR analysis to differentiate cyclopiazonic acid mold. J Food Prot.
2008; 71: 2497–2503. PMID: 19244904
51. Jang CS, Lim JH, Seo MW, Song JY, Kim HG. Direct Detection of Cylindrocarpon destructans, Root
Rot Pathogen of Ginseng by Nested PCR from Soil Samples. Mycobiology. 2010; 38: 33–8. doi: 10.
4489/MYCO.2010.38.1.033 PMID: 23956622
52. Goldschmidt P, Degorge S, Merabet L, Chaumeil C. Enzymatic treatment of specimens before DNA ex-
traction directly influences molecular detection of infectious agents. PLoS One. 2014; 9: e94886. doi:
10.1371/journal.pone.0094886 PMID: 24936792
53. Wintzingerode FV, Göbel UB, Stackebrandt E. Determination of microbial diversity in environmental
samples: pitfalls of PCR-based rRNA analysis. FEMS Microbiol Rev. 2006; 21: 213–229. doi: 10.1111/
j.1574-6976.1997.tb00351.x
54. Wu J, Jiang X, Jiang Y, Lu S, Zou F, Zhou H. Effects of polymerase, template dilution and cycle number
on PCR based 16 S rRNA diversity analysis using the deep sequencing method. BMC Microbiol.
BioMed Central Ltd; 2010; 10: 255. doi: 10.1186/1471-2180-10-255
55. Dickie IA, Bolstridge N, Cooper JA, Peltzer DA. Co-invasion by Pinus and its mycorrhizal fungi. New
Phytol. 2010; 187: 475–84. doi: 10.1111/j.1469-8137.2010.03277.x PMID: 20456067
Standard Sample Preparations Suffice for Next Gen Amplicon Library
PLOS ONE | DOI:10.1371/journal.pone.0127234 May 14, 2015 13 / 13