fmicb-08-00821 May 4, 2017 Time: 16:30 # 1
published: 08 May 2017
University of Verona, Italy
University of Burgos, Spain
Universitat Rovira i Virgili, Spain
This article was submitted to
a section of the journal
Frontiers in Microbiology
Received: 09 December 2016
Accepted: 21 April 2017
Published: 08 May 2017
Belda I, Zarraonaindia I, Perisin M,
Palacios A and Acedo A (2017) From
Vineyard Soil to Wine Fermentation:
to Explain the “terroir” Concept.
Front. Microbiol. 8:821.
From Vineyard Soil to Wine
Approximations to Explain the
Ignacio Belda1,2*, Iratxe Zarraonaindia3,4, Matthew Perisin1, Antonio Palacios1,5 and
1Biome Makers Inc., San Francisco, CA, USA, 2Department of Microbiology, Biology Faculty, Complutense University of
Madrid, Madrid, Spain, 3Department of Genetics, Physical Anthropology and Animal Physiology, University of the Basque
Country, Leioa, Spain, 4IKERBASQUE – Basque Foundation for Science, Bilbao, Spain, 5Laboratorios Excell Iberica,
Wine originally emerged as a serendipitous mix of chemistry and biology, where
microorganisms played a decisive role. From these ancient fermentations to the current
monitored industrial processes, winegrowers and winemakers have been continuously
changing their practices according to scientiﬁc knowledge and advances. A new
enology direction is emerging and aiming to blend the complexity of spontaneous
fermentations with industrial safety of monitored fermentations. In this context, wines
with distinctive autochthonous peculiarities have a great acceptance among consumers,
causing important economic returns. The concept of terroir, far from being a rural
term, conceals a wide range of analytical parameters that are the basis of the
knowledge-based enology trend. In this sense, the biological aspect of soils has
been underestimated for years, when actually it contains a great microbial diversity.
This soil-associated microbiota has been described as determinant, not only for the
chemistry and nutritional properties of soils, but also for health, yield, and quality of
the grapevine. Additionally, recent works describe the soil microbiome as the reservoir
of the grapevine associated microbiota, and as a contributor to the ﬁnal sensory
properties of wines. To understand the crucial roles of microorganisms on the entire wine
making process, we must understand their ecological niches, population dynamics, and
relationships between ‘microbiome- vine health’ and ‘microbiome-wine metabolome.’
These are critical steps for designing precision enology practices. For that purpose,
current metagenomic techniques are expanding from laboratories, to the food industry.
This review focuses on the current knowledge about vine and wine microbiomes, with
emphasis on their biological roles and the technical basis of next-generation sequencing
pipelines. An overview of molecular and informatics tools is included and new directions
are proposed, highlighting the importance of –omics technologies in wine research and
Keywords: NGS, wine microbiome, vine health, soil microbiome, metagenomic analysis, bioinformatic tools and
databases, 16S rRNA gene sequencing
Frontiers in Microbiology | www.frontiersin.org 1May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 2
Belda et al. Microbiome in Enology and Viticulture
Wine is a product with high sociocultural interest. In particular,
wines with distinctive autochthonous properties have a great
demand among consumers and collectors, causing important
economic consequences. It is well known that physical (climate)
and biological factors (soil, grape variety and fauna), as well as
viticulture and enological techniques work together to determine
the sensory-characteristics of a wine from a particular region,
establishing the concept of terroir. In this sense it should be
noted that, apart from these factors, recent studies highlight the
contribution of the native vine microbiota in the winemaking
process of wines from a particular region (Knight et al., 2015;
Bokulich et al., 2016). Additionally, results from Burns et al.
(2016),Grangeteau et al. (2017) correlate human-agronomical
practices in vineyards with the soil and grape microbiota and, also
with its later behavior at cellar, reinforcing the interdependence
between the anthropogenic and microbiological basis of terroir.
Microbes transform plant products into socio-economically
important products and fermented beverages, such as wine,
which is an extremely important sector for several countries. For
instance, the International Organization of Wine and Vine (OIV)
estimated in 2015 that the global wine-growing surface area was
7,534,000 hectares, with the biggest producer being Italy (18% of
the global total), followed by France (17.3%) and Spain (13.5%).
Outside the EU, the USA has the highest wine production
followed by Argentina, Chile and Australia (OIV, 2015).
Due to the economic importance of the grapevine, this
crop has received considerable interest among researchers;
although this attention mainly focuses on the plant genome
and transcriptome/metabolome to better understand how the
plant responds to the physical environment, abiotic stresses and
diseases (e.g., the International Grape Genome Program, IGGP).
However, plants cannot be considered a self-contained, isolated
organism, as plant ﬁtness is a consequence of the plant per se and
its associated microbiota (Vandenkoornhuyse et al., 2015). Thus,
a more holistic conception should include plant-microorganisms
and microbe-microbe interactions.
Although the role of microorganisms at cellar stages has
been well investigated, the biological aspect of soils has not
received similar attention, when actually it contains a great
microbial diversity with important roles in plant nutrition and
health (Compant et al., 2010;Bhattacharyya and Jha, 2012).
Next-generation sequencing (NGS) approaches have uncovered
a higher than expected microbial diversity in both vine and
wine and discovering new microbial species, some with unknown
contributions to the organoleptic properties of wines (Bokulich
et al., 2016). Stable diﬀerences among microbial populations of
grape musts have been attributed to grape variety, geographical
area, climatic factors and vine and grape health, leading to the
concept of vine microbial terroir (Bokulich et al., 2014). This fact
has been reinforced at a phenotype-metabolome level by other
works such as Knight et al. (2015),Bokulich et al. (2016), and
Belda et al. (2016). The later observed distinctive and clustered
metabolic proﬁles (production of hydrolytic enzymes) for yeast
strains depending on their geographical origin. It has been also
observed that the origin of these microorganisms in musts is
the microbial consortia of grapes, with the original reservoir of
these microorganisms being vineyard soil (Zarraonaindia et al.,
2015). Thus, the microbiological aspects of wine production
are inﬂuenced by the vineyard and not just by the winery and
The maturation of grapes is a complex process that depends
on numerous factors (Kennedy, 2002). Traditionally, the most
common measured parameters include: sugar concentration,
acidity and aromatic and phenolic maturity. However, soil and
grape microbiological complexity throughout the cycle of the
vine and grape maturation is rarely taken into consideration.
Communities of microorganisms (fungi, yeast and bacteria)
associated with the vineyard play an important role in soil
productivity as well as disease resistance developed by the vine.
It is important to understand the microbial consortia associated
with particular diseases, such as Esca,Eutypa,Botryosphaeria, and
Phomopsis diebacks, and also the dynamics of infection processes
in order to take preventive actions, especially at the most critical
moments (Figure 1). For instance, microbial insights are crucial
for deﬁning strategies for the preparation of new plantings. At
this stage, it could be interesting to improve the microbiological
conditions of the soil by bioremediation and to avoid risk of cross
infection during pruning (Bertsch et al., 2013;Fontaine et al.,
The diversity and number of microorganisms that are
able to establish in an ecological niche in the soil and on
the vine will determine both the grapes’ health and the
variability of microorganisms that will be introduced in the
winery that further aﬀect the fermentation processes and
wine maturation (Barata et al., 2012). Thus, with adequately
managed microbiome information, it could be possible to prevent
fermentation problems, volatile acidity increases, Brettanomyces
contamination and biogenic amines production. Knowing more
about the microbiological conditions of the vineyard allows the
winegrower to think about the reduction of chemical treatments
and performing them only when they are objectively necessary.
Additionally, this knowledge would help the winemaker to
use lower sulfur concentration at cellar stages and even
to decide the type of yeast and dose to be inoculated if
and when necessary (Figure 1). This is valued information
especially considering new enology trends, such as organic
Next-generation sequencing technologies enable the detection
and quantiﬁcation of microorganisms present in vineyard soil,
grapes, as well as its transformation later in winery. The impact of
the microbiological component of terroir and how it contributes
not only to its quality but also in the organoleptic features
of the wine is considerable. This impact also contributes to
the sensory regional distinctiveness and the wine style of the
winery that currently plays an important role in diﬀerentiation
and competitiveness in the worldwide market. If something can
distinguish one vineyard from another, among other factors, it
certainly is its microbial community. In this context, the objective
of this review is to summarize the current knowledge about
the role of microbial communities in viniculture, highlighting
the contributions of NGS technologies and identifying new
Frontiers in Microbiology | www.frontiersin.org 2May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 3
Belda et al. Microbiome in Enology and Viticulture
FIGURE 1 | Current challenges on viticulture and enology assumable by NGS approaches; advisable technical improvements; necessities and
perspectives in data science. NFB means ‘Nitrogen Fixing Bacteria’ and PMB means ‘Phosphate Mobilizing Bacteria’.
THE MICROBIOME OF VINE AND WINE:
Plants host a variety of microorganisms (fungi, yeast, and
bacteria) on and inside organs and their surrounding soil. Among
these inhabitants are both harmful and beneﬁcial microbes
that are involved in crucial functions such as plant nutrition
and plant resistance to biotic and abiotic stresses, hence in
plant growth promotion, fruit yield, disease resistance and
survival (Lugtenberg and Kamilova, 2009;Compant et al., 2010;
Bhattacharyya and Jha, 2012).
Studies on microorganisms associated with grapevines have
been centered on the cultivable fungi (mainly yeast) or bacteria
that can have a negative economic impact, compromising
the yield and quality of the grapevine, as well as wine
production. Studies have focused on disease causing pathogens
(Agrobacterium vitis, Xylella fastidiosa, Erysiphe necator,
Phomopsis viticola, Fusarium spp., etc.) and microorganisms of
enological interest. The later species have been grouped into three
classes [reviewed in Barata et al. (2012)]: (1) easily controllable
or innocent species, without the ability to spoil wine when good
manufacturing practices are applied; (2) fermenting species
Frontiers in Microbiology | www.frontiersin.org 3May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 4
Belda et al. Microbiome in Enology and Viticulture
responsible for sugar and malic acid conversion; and (3) spoilage
sensu stricto species responsible for wine alteration The most
widely known cultivable bacteria are acetic acid bacteria (AAB;
e.g., Acetobacter and Gluconacetobacter) and lactic acid bacteria
(LAB; e.g., Lactobacillus, Oenococcus, and Pediococcus). Among
yeasts, Saccharomyces members have attracted most of the
attention as they are the main fermentation agents commonly
used as inocula (e.g., Saccharomyces cerevisiae, S. bayanus, S.
pastorianus, and S. paradoxus among others), while other genera
are the most frequent wine spoilers (e.g., Brettanomyces/Dekkera,
Issatchenkia, Zygoascus, and Zygosaccharomyces).
While culture dependent methods have been useful to
detect and identify microbial organisms associated with
grapevine and grape products, and also to study in vitro
their metabolic properties (Belda et al., 2016), they have
led to a rather biased picture of the microbial community.
These methods neglect the larger, non-culturable fraction
that is believed to be as high as the 95–99% of the
microorganisms present (Amann et al., 1995;Curtis, 2002).
In wine environment, due to the stressful environment
associated to the addition of SO2, high ethanol concentration,
etc., a fraction of the bacteria and yeast enter in a Viable But
Non-Culturable state (VBNC) (Millet and Lonvaud-Funel, 2000;
Divol and Lonvaud-Funel, 2005). At this state cells do not
grow on culture media, however, they are still viable and
maintain a detectable metabolic activity (Yamamoto, 2000)
which may aﬀect fermentation performance as well as ﬂavor.
Examples of such microorganisms include Candida stellata,
Brettanomyces bruxellensis, S. cerevisiae, Zygosaccharomyces
bailii, etc. (Salma et al., 2013). Thus, in order to reach to these
VBNC microbiologists were driven to develop alternative
culture-independent techniques. Particularly, quantitative real
time PCR (qPCR) has been widely used to detect bacteria and
yeast considered to be wine spoilers and that have VBNC strains
responsible for the production of oﬀ-ﬂavors or having a negative
impact on wine, e.g., Brettanomyces spp. (Tofalo et al., 2012).
Nowadays, qPCR is believed to be a rapid diagnostic tool to
detect the presence and quantify the abundance of particular
microorganisms of interest, however, when the objective is
not a targeted species, but rather a whole community analysis,
PCR-DGGE has been the classical method of choice. The later
technique is adequate to approximate the total community
proﬁle and for comparative community structure analysis, but
it has several drawbacks mainly associated to biases related
with species richness estimates and its low sensitivity to detect
low abundance species (Neilson et al., 2013). For instance,
multiple bands could associate with single isolates. In addition,
multiple sequences might be associated with a single band
and preferential ampliﬁcation biases between phylogenetically
diverse members of the community have been shown (Neilson
et al., 2013). Andorrà et al. (2010) compared the population
dynamics of microorganisms of grape must fermentation by
three culture independent techniques (DGGE, direct cloning of
ampliﬁed DNA, and qPCR) with plate counting, and evidenced
that the biodiversity observed in the must and at the beginning
of fermentation was much higher when DGGE or direct cloning
were used. However, the predominance of certain yeast such as
C. zemplinina and S. cerevisiae during fermentation limited the
detection of low abundant species. Thus, while DGGE is believed
to give a quick and non-expensive view of the community, it
skews microbial diversity estimates (David et al., 2014) and
it has a limited use to study diverse environmental samples
dominated by few species (Andorrà et al., 2010). When adding
NGS technique into the detectability comparition of culture
independent techniques to study yeast community in must
and ferments, the studies evidenced that larger numbers of
yeast species were detectable by NGS than by PCR-ITS-RFLP
or DGGE in grape samples. Moreover NGS detected species
in ferment samples that were undetectable with the two later
techniques (David et al., 2014). In addition, Wang et al. (2015)
analyzed Carignan and Granache grape must and fermentation
from three vineyards in Priorat (Spain) and found that NGS
detected all the species identiﬁed by the rest of methods (DGGE,
qPCR and culture dependent), whereas DGGE could just detect
the dominant species of non-Saccharomycetes class. Thus, NGS
showed to be more appropriate to understand must and wine
environment yeast communities (David et al., 2014;Wang et al.,
Next-generation sequencing technologies are providing a
powerful approach to achieve a more complete understanding of
the complexities of microbial communities and their impact on
plant growth, disease resistance/susceptibility, climate adaptation
and environmental remediation. This technology is enabling
researchers to simultaneously obtain information on thousands
of taxa as opposed to targeted approaches that detect only a
taxonomically predeﬁned group. Thus, metagenomics coupled
with new bioinformatics tools, is allowing performance of more
complex multifactorial analyses and is becoming a powerful
strategy in diagnostics, monitoring, and traceability of products.
Its application in viticulture while recent is promising (Table 1),
as accumulating data suggest that there is a much higher
microbial diversity associated both with the plant (Leveau and
Tech, 2010;Pinto et al., 2014;Zarraonaindia et al., 2015) and
the fermentation process (Bokulich et al., 2012;Piao et al., 2015;
Pinto et al., 2015;Portillo and Mas, 2016;Stefanini et al., 2016)
compared to previous culture based studies. Most metagenomics
research in this ﬁeld has focused on microbial monitoring during
fermentation to obtain a detailed description of the relevant
microbial populations associated with grape and must that might
lead to wine spoilage, an advance highly valuable for winemaking.
These NGS-enabled studies reﬂect a wider range of bacteria,
besides the commonly detected LAB and acetic acid species,
able to persist in fermenting musts of various grape varieties
(Bokulich et al., 2012;Piao et al., 2015;Portillo and Mas, 2016;
Stefanini et al., 2016). For instance, the ﬁrst wine-related study
conducted in the wine environment with NGS was conducted by
Bokulich et al. (2012) during botrytized wine fermentation using
16S rRNA gene amplicon sequencing. These authors showed an
array of ﬂuctuating low abundant taxa not traditionally associated
with wine, as well as atypical LAB communities during the
process. Similarly, results from Portillo and Mas (2016) suggested
that AAB are more abundant and dynamic than previously
thought during low or unsulﬁted wine fermentations, and seemed
to be independent of the grape variety. Interestingly, in this
Frontiers in Microbiology | www.frontiersin.org 4May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 5
Belda et al. Microbiome in Enology and Viticulture
TABLE 1 | Research and industrial hallmarks of viticulture and enology led
by NGS approaches.
Enological features addressed by
Interference of microorganisms in plant
Lugtenberg and Kamilova, 2009;
Compant et al., 2010;
Bhattacharyya and Jha, 2012;
Martins et al., 2013;
Vandenkoornhuyse et al., 2015
Microbial diversity in vineyard Leveau and Tech, 2010;
Pinto et al., 2014;
Zarraonaindia et al., 2015
Microbial diversity in wine fermentations Bokulich et al., 2012;
Piao et al., 2015;
Pinto et al., 2015;
Portillo and Mas, 2016;
Stefanini et al., 2016
determining vineyard and wine
Burns et al., 2016;
Grangeteau et al., 2017
Microbial contribution to wine chemistry Verginer et al., 2010;
Bokulich et al., 2016
Terroir markers (microbial zoning) Bokulich et al., 2014, 2016;
Burns et al., 2015;
Knight et al., 2015
study yeast diversity and dynamics during wine fermentation
was assessed in addition to bacteria, evidencing that the genera
Hanseniaspora and Candida were dominant during the initial
and mid- spontaneous fermentation of Grenache grapes while
certain Candida and Saccharomyces species predominated at the
end of the fermentation. Other studies have demonstrated how
diﬀerent fermentation techniques (spontaneous vs. inoculated)
aﬀect the microbial community composition and its succession
during fermentation (Piao et al., 2015), and also how the previous
agronomical practices in the vineyard could play a critical
role in these population dynamics (Grangeteau et al., 2017).
These authors’ results indicated certain phyla are associated
with each particular technique. Interestingly, they observed that
Gluconobacter experienced a notable increase during organic
fermentation, which led the authors to conclude that this might
explain the increased susceptibility to wine spoilage in wines
produced using that technique.
These above-mentioned studies enhance our understanding
of microbial diversity during fermentation and allow the
identiﬁcation microbial contamination sources. However, as
DNA sequencing approaches detects living as well as dead
microorganisms, it is still not clear to what extent these
microorganisms metabolically are active and capable of aﬀecting
organoleptic properties of wine. The role of the microbiota
inﬂuencing the ﬂavor, color and quality of wine, under a systems
biology perspective, remained elusive until recently.
While soil, weather, farming techniques and grape variety
contribute to the unique qualities of wine, adding distinctiveness
and thus market value, the contribution of the microbiota in
deﬁning terroir is now in the spotlight of scientiﬁc research.
Regionally distinct wines are highly appreciated by consumers
and add value to the industry. In Spain alone there are
90 zones, which produce distinct so-called PDO wines, of
which 69 are Denomination of Origin (DO), 2 are Qualiﬁed
Denomination of Origin (DOCa), 7 are Quality Wine with a
Geographical Indication (Vino de Calidad) and 14 are Single
Estate Wine (Vino de Pago). While the chemosensory distinction
of wines from diﬀerent growing regions has been previously
established [e.g., Loópez-Rituerto et al. (2012)], indigenous
microorganisms associated with grapes were shown to be able
to produce compounds responsible for the regional ﬂavors
of the resulting wine, e.g., VOCs (Verginer et al., 2010). In
addition, Knight et al. (2015) experimentally demonstrated that
wine organoleptic characteristics are aﬀected by the origin and
genetics of wild S. cerevisiae natural strains, providing objective
evidence for a microbial aspect to terroir.Bokulich et al.
(2014) showed that Cabernet Sauvignon must from diﬀerent
growing regions in California could be distinguished based
on the abundance of several key fungal and bacterial taxa.
This diﬀerential must microbiota could potentially inﬂuence
wine properties and contribute to the regionalization of wine.
The later was further proved in Bokulich et al. (2016); these
authors demonstrated that both grape microbiota and wine
metabolite proﬁles were able to distinguish viticultural area
designations and individual vineyards within Napa and Sonoma
Counties in CA, USA. Interestingly, the vineyard microbiota
correlated with the chemical composition of the ﬁnished wines,
hinting at the possibility of predicting wine phenotypes prior to
fermentation. Nevertheless, wine aroma is deﬁned by hundreds of
chemical compounds with diﬀerent natures (i.e., higher alcohols,
esters, fatty acids, terpenes, thiols) causing a broad spectrum of
sensory thresholds, and also suﬀering synergies and antagonisms
(Belda et al., 2017). Thus, looking for microbial signatures
determining wine typicity, the sensorial characterization of wines
should consider not only chromatographic analysis (revealing
the diversity and concentration of aroma compounds), but also
developing serious sensorial or olfactometry analysis to reﬂect
the real perception of wine aroma or, at least, considering odor
activity values (OAVs) to correlate the real inﬂuence of microbial
species in wine aroma, as was addressed by Knight et al. (2015).
While grape and must have been more heavily researched,
Zarraonaindia et al. (2015) further hypothesized that the soil
and its associated microbiota inﬂuences wine characteristics.
First, per these authors’ studies, the aboveground bacterial
community was signiﬁcantly inﬂuenced by soil edaphic factors
such as total carbon, moisture and soil temperature, which would
ultimately impact the quality of grapes due to changes in nutrient
availability for the plant. Second, soil bacterial communities
diﬀered between the sampled vineyards in Long Island, New York
and those diﬀerences were reﬂected in the microbial composition
in vine roots. These root endophytes can shape the microbial
assemblages of aboveground organs by changing the endophytic
microbial loads in grapes. Third, a signiﬁcant input of soil
microorganisms to grapes through epiphytic migration during
harvest was suggested. The later was also evidenced by Martins
et al. (2013), leading Zarraonaindia et al. (2015) to propose
that soil derived microorganisms could have a greater role than
previously anticipated in wine, as they will ultimately end up
in the fermentation tanks. The link between soil microbiota
Frontiers in Microbiology | www.frontiersin.org 5May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 6
Belda et al. Microbiome in Enology and Viticulture
and terroir was further evidenced by Burns et al. (2015) who
identiﬁed distinctive microbial community proﬁles by American
Viticultural Areas (AVA).
NGS MICROBIAL PROFILING: KEY
STEPS, BIASES AND LIMITATIONS
The above summarized studies were conducted on grapevines
and wine and address microbial composition by means
of 16s rDNA PCR amplicon and ITS (Internal Transcribed
spacer) NGS sequencing for elucidating the bacterial and fungi
community, respectively. This marker gene ampliﬁcation and
sequencing method, also called amplicon sequencing, has become
the method of choice to simultaneously detect multiple species
in must and wine environment since 2012 (see Bokulich et al.,
2012, 2016;Pinto et al., 2014, 2015;Knight et al., 2015, among
others). However, the particular experimental question of the
research to be conducted will determine the method mostly
suited to answer to the question. For instance, if the goal is to
track a particular microbial strain or genus from soil to must
to fermentation, then qPCR could be a more appropriate and
has the added beneﬁt of absolute quantiﬁcation (Neeley et al.,
2005). To detect a speciﬁc microbe, primers must be designed to
be highly speciﬁc for the microbe of interest. Often the primer
design can be completed by genome comparison of targeted and
non-targeted strains to ﬁnd a unique gene or region. Another
strategy involves targeting a conserved gene (16S rRNA, gyrB,
rpoB) and making sure the primers mismatch oﬀ-target strains
particularly at the 30end. Single copy genes provide an added
bonus for absolute quantiﬁcation. Microbe quantiﬁcation by
qPCR, however, does not scale easily if the goal is to analyze
more than a few strains while amplicon sequencing is suited to
determine the community.
However, amplicon sequencing is not free of pitfalls,
and diﬀerent biases have been described in multiple steps
of the process; First, DNA extraction method is one of the
key and limiting steps for metagenomic analysis by NGS.
Various approaches have been applied for environmental
DNA extraction, including freeze–thaw lysis (Herrick et al.,
1993), bead beating (Miller et al., 1999;Courtois et al., 2001;
Urakawa et al., 2010;Petric et al., 2011), liquid nitrogen grinding
(Ranjard et al., 1998), ultrasonication (Picard et al., 1992), hot
detergent treatment (Holben, 1994), use of strong chaotropic
agents like guanidinium salts (Porteous et al., 1997), and high
concentration of lysozyme treatment (Hilger and Myrold, 1991).
Furthermore, soil, grapes and wine are complex physicochemical
environmental samples that contain many interfering agents for
molecular analysis such as impurities, phenols, humic acid, fulvic
acid, metal ions and salts, and therefore additional puriﬁcation
steps are necessary which can introduce bias by altering the
original community (e.g., a fraction of the community might
be lost through puriﬁcation, etc.). There are several commercial
kits that could be used to fasten the process, however, the
selection of the best DNA extraction method and kit is not
straightforward as diﬀerent DNA extraction methods can
produce diﬀerent results (Keisam et al., 2016). Unfortunately,
there is no “gold standard” for DNA extraction method and one
should be selected on a case-by-case basis considering the aims,
specimens of the study and scalability (including simplicity,
cost eﬀectiveness, and short handling time) and intended study
comparisons. An additional problem is the introduction of
contaminating microbial DNA during sample preparation.
Possible sources of DNA contamination include molecular
biology grade water, PCR reagents and DNA extraction kits
themselves. Contaminating sequences matching water-and
soil-associated bacterial genera including Acinetobacter,
Alcaligenes, Bacillus, Bradyrhizobium, Herbaspirillum, Legionella,
Leifsonia, Mesorhizobium, Methylobacterium, Microbacterium,
Novosphingobium, Pseudomonas, Ralstonia, Sphingomonas,
Stenotrophomonas, and Xanthomonas have been reported
previously. The presence of contaminating DNA is a particular
challenge for researchers working with samples containing a low
microbial biomass. In these cases, the low amount of starting
material may be eﬀectively swamped by the contaminating DNA
and generate misleading results (Salter et al., 2014).
Second, DNA library preparation, based on fragment
ampliﬁcation through PCR with barcoded primers, is another
step in which it is possible to introduce additional biases.
The choice of primers and targeted variable regions will bias
identiﬁcation and quantiﬁcation (Soergel et al., 2012;Bokulich
and Mills, 2013). Additionally, in any PCR- and primer-based
taxonomic investigation, members of a microbial community
may be omitted, distorted, and/or misrepresented, typically
due to primer mismatches or PCR biases (Acinas et al., 2005;
Hong et al., 2009;Lee et al., 2012;Pinto and Raskin, 2012;
Logares et al., 2014). On the contrary, primers might show
variability in their ampliﬁcation eﬃciency by for example,
favoring certain species ampliﬁcation (Baker et al., 2003;
Sipos et al., 2007;Klindworth et al., 2013). This preferential
ampliﬁcation is thought to be derived from diﬀerent sources
such as primer mismatches, the annealing temperature and PCR
cycle numbers (Sipos et al., 2007). For instance, Sipos et al.
(2007)’ studies evidenced that A. hydrophila and P. ﬂuorescens
were preferentially ampliﬁed over both Bacillus strains when
the 63F primer was used (which contained three mismatches
against DNA isolated from the Bacillus strains), while the 27F
primer ampliﬁed all templates without bias. Interestingly, the bias
introduced by primer mismatches was reduced at lower annealing
Multiple primer pairs are available for marker genes, and
each pair is associated with its own taxon biases. Marker
gene databases are frequently updated, and the updated
information can include new microbial lineages with suboptimal
or poor binding to existing PCR primers; to maximize
taxonomic sensitivity in light of these new data, primers
may need to be periodically redesigned. A recent example in
the literature is the modiﬁcation of the most common 16S
primers used 515f and 806r to remove know biases against
Crenarchaeota/Thaumarchaeota and the marine and freshwater
Alphaproteobacterial clade SAR11 (Apprill et al., 2015;Parada
et al., 2016).
Klindworth et al. (2013) evaluated the coverage and phylum
spectrum for bacteria and archaea of 175 primers and
Frontiers in Microbiology | www.frontiersin.org 6May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 7
Belda et al. Microbiome in Enology and Viticulture
512 primer pairs in silico for three amplicon size classes
(100–400, 400–1000, >1000 bp), demonstrating the diﬀerences
in coverage and speciﬁcity among the studied primers. Besides,
this information represents a valuable guideline for selecting
primer pairs that could minimize the bias in PCR-based microbial
diversity studies. In the same way, probeBase1is an additional
online resource, providing the opportunity to evaluate the in
silico hybridization performance of oligonucleotides, as well as
ﬁnding suitable hierarchical probes that could target an organism
or taxon of interest at diﬀerent taxonomic levels (Greuter et al.,
The ideal marker gene should have conserved regions that
ﬂank variable regions. The conserved regions allow primer design
to amplify multiple taxons at ones. Ribosomal rRNA genes ﬁt
this description and have been widely used for identiﬁcation
of bacteria/archaea (16S) and fungi (ITS) (Gilbert et al., 2010).
However, ribosomal RNA genes show copy-number variation,
with very disparate number of copies per taxa (from one in
many species to up to 15 in some bacteria and to hundreds in
some microbial eukaryotes) biasing conclusions related to the
abundance of the organisms.
To evaluate the entire microbial community in the speciﬁc
case of the wine ecosystem, it is necessary to strike an
appropriate balance between amplifying all members of every
taxon (high coverage) and obtaining the highest taxonomic
resolution possible, e.g., to be able to discriminate among closely
related species (Figure 1). Each marker shows diﬀerences in its
discrimination power at intra-genera as well as at intra-species
level. Thus researchers must have that in mind when designing
their project, in order to choose the most appropriate molecular
marker to answer their particular question/s. For instance, primer
pair 515f/806r is the most widely used for targeting the V4
region of for bacteria/archaea (Parada et al., 2016), and this
combined with Illumina sequencing has been used to characterize
the microbiomes of numerous environments (Caporaso et al.,
2012), vine and wine environments among them. Data from
high diverse environments, as Sakinaw Lake, showed species
resolution level from 49.4% of the 16S V4 sequences classiﬁed
compare with 74.5% using full 16S. Although the relative
classiﬁcation diﬀerences at the sequence level do not directly
translate to diﬀerences in community representation (Singer
et al., 2016). However, vine and wine samples have the added
diﬃculty in that mitochondrial and chloroplast DNA can be
ampliﬁed with these V4 region primers and thus grapevine
plastid sequences overwhelm the sequencing. Researcher have
two ways to avoid this problem: design primers that mismatch
mitochondrial/chloroplast sequences or add blocking reagents
that bind these sequences (Lundberg et al., 2013). Besides, the
V4 domain of the 16s rDNA gene is considered to be the most
suitable marker for capturing the bacterial community in wine,
as it is able to reliably discriminate LAB to genus-level (Bokulich
et al., 2012). However, in fermentative systems, some species of
LAB are considered wine spoilers while others exhibit malolactic
activity, thus it might be essential to reach to species level
(Bokulich and Mills, 2012) and/or strain level, in order to have
a more comprehensive view of the community. Unfortunately,
currently available amplicon sequencing markers are unable to
capture that level of resolution in all taxa. These limitations could
be overcome by combining several techniques such as genera
speciﬁc T-RFLP or qPCR and amplicon sequencing.
Third, important sources of artifacts are also derived from
the High-throughput sequencing technology chosen. While
pyrosequencing introduces homopolymer errors (indel error),
Illumina sequencing has average substitution errors at 0,0086
sequencing rate (Schirmer et al., 2015). Sequencing platforms
also show a disparity in sequencing depth (number of reads per
run) and read length. Illumina MiSeq is the most commonly
used sequencer for amplicon sequencing due to its high coverage
with a total nucleotide sequenced of 15GB allowing sequencing
the abundant and rare community giving a deep view of
the community composition. However, llumina sequencing is
characterized by a short variable region sequencing (2 ×300 bp
vs. 700 bp in 454). Currently, nearly full-length rRNA gene
sequencing is possible with PacBio and Nanopore technologies
(Benitez-Paez et al., 2016;Schloss et al., 2016).
Finally, one of the biggest limitation of amplicon sequencing
techniques relays on its inability to address a functional
characterization of the microbial communities. There are many
desired microbial functions in winemaking, mainly related to
alcoholic and malolactic fermentations, and diversity of genes
related to those functions may inﬂuence winemaking more than
just taxonomic diversity. In addition, closely related strains with
highly similar 16S rRNA gene or ITS sequences contain diﬀerent
fermentation-related genes (Knight et al., 2015) and thus that
strain diversity remains hidden in current amplicon sequencing
studies. Single-cell genomics emerges as a potential strategy
that could help to obtain a deeper knowledge into species-
strain level diversity. This strategy is powerful when the targeted
organism is dominant or high abundant in a low species richness
ecosystem. However, in highly diverse ecosystems or when the
species to be targeted is low abundance, it may require a higher
sorting throughput, speciﬁc labeling with ﬂuorescent probes or a
previous cultivation step, all of which could contribute to biases.
Alternatively, shotgun metagenomic sequencing would also
reveal functional genes in addition to rRNA genes, allowing a
more comprehensive genomic and functional representation
through whole-genome sequencing (WGS) of complete
communities, but the cost and the number of reads needed
to estimate the environmental population is high compared
to PCR-based approaches. Even more in wine samples, as a
very deep sequencing is required to detect microbes due to
an overabundance of plant DNA (Zarraonaindia et al., 2015),
making this method costly for a large number of samples.
Metatranscriptomics is emerging as a powerful technology
for the functional characterization of microbial communities
that can reveal both the taxonomic composition and active
biochemical functions of the detected organisms. These approach
is of especial interest in wine environment, as amplicon
sequencing is not able to discriminate among living or dead
organisms, nor the metabolically active or inactive organisms.
However, the high sequencing depth needed and the high cost
associated with the sequencing of each sample limits the number
Frontiers in Microbiology | www.frontiersin.org 7May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 8
Belda et al. Microbiome in Enology and Viticulture
of samples that could be surveyed within a project currently.
In addition, challenges associated with this technique include
among others, the lack of established reference genomes to
annotate the short reads generated in the sequencing and the high
computational eﬀort needed for the analyzes. Being a technique
still in its infancy, new analysis tools and standardized pipelines
are under development. In this context, the next section aims
to summarize critical concepts and sources of biases in NGS
BIOINFORMATICS AND PREDICTIVE
METHODS TO UNCOVER THE
Along with the relative ease with which thousands of organisms
can be detected in samples via 16S/ITS sequencing, a whole host
of bioinformatics approaches have been developed to extract
meaningful results from the large datasets that are generated.
The bioinformatics challenge comes in at least two parts
(I) preprocessing the datasets into a collection of representative
reads (or operational taxonomic unit – OTU) that can be
associated with databases of known species and (II) associating
the collection of species inferred in a sample (known and newly
detected) with properties of the sample in order to study the
relationships between the microbiome and the terroir.
In the ﬁrst stage, the large amounts of raw sequencing reads
are processed (trimming adaptor sequences, merging forward
and reverse sequences, ﬁltering on read quality) before ﬁnally
being dereplicated into a collection of unique sequences. There
is a lot of software available to perform these tasks and they
often are part of packages that oﬀer an entire processing pipeline
(USEARCH, vsearch, FASTX-Toolkit) (Edgar, 2013;Rognes
et al., 2016). The unique sequences are then clustered according
to sequence similarity, choosing a relatively arbitrary cutoﬀ at
97% identity (Seguritan and Rohwer, 2001), resulting in a set of
OTUs that are each assumed to be originating from a speciﬁc
organism. In other words, OTUs are proxies for microbial species
in the sample (Schloss et al., 2009;Caporaso et al., 2010).
Although conceptually simple, this step poses major
challenges both computationally and in terms of biases that
might potentially bleed into subsequent analysis. First of all, for
large sets of sequences, all against all pairwise alignments would
be prohibitive, e.g., 1 million of unique sequences (commonly
encountered), would require 1000 billion pairwise comparisons.
This has led to comprehensive bioinformatics pipelines for
OTU clustering, including the software pipelines mentioned
above (USEARCH, vsearch, swarm), which all rely on clever
heuristics (Edgar, 2013;Eren et al., 2013;Mahé et al., 2014, 2015;
Tikhonov et al., 2015;Rognes et al., 2016) in order to accelerate
this process at the expense of perfectly accurate clustering. The
second challenge is to avoid biases that can occur during OTU
clustering. The biases can be multifold; (a) diﬀerent biological
species might have the same sequence and therefore be grouped
into one set, (b) sequencing errors or ampliﬁcation errors
(including chimeric reads) or untrimmed sequences can group
sequences that have the same origin into separate groups. The
ﬁrst issue will underestimate biological diversity whereas the
latter will overestimate it. Together these scenarios will corrupt
the accurate representation of the real biological makeup of the
terroir. This highlight again that it is important to quality trim
and ﬁlter the raw sequences to minimize the risk of including
artifacts in environmental data sets.
Finally, the curated OTUs are subjected to phylogenetic
assignment, which aims to identify what species or genus an
OTU most likely belongs to. This is achieved by comparing
them with taxonomically classiﬁed sequences at databases, such
as GreenGenes (for bacteria community characterization), SILVA
(bacteria and eucaryotes) and Unite (for Fungi) among others.
Again, a range of software is available (Qiime, UTAX, SINTAX,
stampa) (Caporaso et al., 2010;Edgar, 2013, 2016). This stage
is again a source of biases, partly because OTUs can represent
multiple species, there is ambiguous assignment, and because too
small diﬀerences that do exist could be ignored by these methods.
For instance, in the case of Oligotyping, a single base pair can
diﬀerentiate ecological strains (Eren et al., 2013). Furthermore,
and more generally, reference databases are themselves based
largely on predicted species rather than experimentally cultivated
species and can thus bias taxonomic assignment. Additionally,
diﬀerent reference databases would yield diﬀerent taxonomic
assignments as a function of completeness and quality of the
database (McDonald et al., 2012). Notably, if a given species
is not represented within the database, sequences derived from
that species would receive an incorrect assignment or remain
unclassiﬁed. This is aggravated for wine and soil associated
microbial sequences ﬁeld, where reference databases lag behind
human-associated microbes. Increasing and curating robust
databases is a key goal for the scientiﬁc community (Figure 1).
There are also other methods allowing comparison of amplicons
derived from functional genes in which we might not know
percent identities that correspond to taxonomic levels, but
in some cases, are optimal to reﬂect geographical (and thus,
environmental) distance (Haggerty and Dinsdale, 2017). In
relation to wine related samples, cluster free methods show the
potential to deﬁne the microbial terroir at the strain or sub-OTU
level (Tikhonov et al., 2015;Eren et al., 2016).
Equipped with a dataset of biological entities in the terroir
(genus- or species-level), the second bioinformatics challenge
concerns associating the microbiome to the properties of the
terroir. Depending on the aim, this can be more or less
diﬃcult. One goal is to use microbial community data to
classify soil samples into types and geographical regions and,
therefore, deﬁne the microbial terroir. Recently, Bokulich et al.
(2016) demonstrated the power of this approach for classifying
Californian regions and fermentation metabolites based on
microbial abundances in musts. However, if species or even
strain information is required to establish an association between
microbiome and speciﬁc wine making properties, then the
taxonomic assignment is essential and can make or break an
analysis depending on the resolution it achieves and the biases
it can prevent.
Apart from the nature of the question, generally, the structure
of OTU abundance data poses some challenges that need to
be carefully taken into account. Because the species can occur
Frontiers in Microbiology | www.frontiersin.org 8May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 9
Belda et al. Microbiome in Enology and Viticulture
in very diﬀerent abundances (often spanning several orders of
magnitude), the collection of species across samples can greatly
vary. This leads to a very sparse dataset, which is deﬁned as
a dataset with many zero values. These zero values can be
problematic as they could entertain multiple hypotheses; for
instance, a zero count in a sample could be because a species
is not present, or because it just has not been detected. This
can lead to biased comparisons between samples. One way to
deal with this is to use distance metrics that do not consider
these situation (e.g., Bray-Curtis) or that speciﬁcally include
a phylogenetic tree that allows to relate species information
into meaningful groups. Preprocessing of OTU data from raw
counts to a value that makes samples comparable to each other
is the next step. This is also referred to as normalization and
there are number of analytical choices available (Segata et al.,
2011;Paulson et al., 2013) depending on whether low-abundance
species or high abundance species should have more of an
impact in the analysis inquestion. For instance, counts can be
converted to frequencies (divide the number of reads by the
total number of reads in the sample). The performance of
these techniques given OTU table peculiarities has been tested
elsewhere (McMurdie and Holmes, 2014;Weiss et al., 2016).
This is also a crucial step when applying machine learning
With a preprocessed dataset available, the probe community
level diﬀerences between samples, can be studied with supervised
and unsupervised machine learning techniques. Unsupervised
learning categorizes samples based on OTU abundances without
prior knowledge of the sample phenotypes. Principle component
analysis (PCA or more commonly PCoA) and clustering
algorithms can be used to gain a high level view over
diﬀerences in samples. These analyses are largely exploratory
and provide visual evidence of community diﬀerences. If
information regarding the terroir is available, or there are
some clearly deﬁned groups that are to studied, supervised
learning techniques can be applied to for instance classify new
samples based on past community characterizations (Bokulich
et al., 2016). Distinct wine regions, types and tastes make
wine related samples well-suited for these classiﬁcation methods
(Statnikov et al., 2013). Diﬀerent software packages are available
to perform these methods and can be more or less adapted to
the study of metagenimics problems (vegan, phyloseq, Qiime,
Another method to extract knowledge from microbiome data
is to consider it as a network of interactions between individual
strains. Aside from the impact of single strains in plant health
(pathogens, symbionts) and wine characteristics, or spoilage
potential, these strains impact wine production not in isolation
but instead as members of complex microbial communities.
Much research now focuses on these community level eﬀects that
can impact plant phenotypes such as ﬂowering time (Wagner
et al., 2014).
These predictive technologies allow to make initial inferences
about whether these diﬀerentially abundant single OTUs cause
certain phenotypes. However, they will requires further testing,
likely with pure culture treatments. One excellent example
of going from correlation to causation is the use of pure
fungal and oomycete cultures in a common garden to conﬁrm
single strain eﬀects on the overall microbial community
structure associated with Arabidopsis thaliana (Agler et al.,
Deﬁning the microbial terroir with bioinformatics is only
an early step to understand how microbes shape each step in
winemaking. Wine imparts its taste and smell via metabolites,
many derived from the grapes and many derived from or
modiﬁed by microbes. Identifying which microbes inﬂuence
these processes is key to deﬁning how they aﬀect the sensory
proﬁle of wines. As we add genomic sequences to our
reference database we will be able to leverage annotated
sequences to predict metabolic capacity for each microbe.
Genome scale metabolic models (GSMM) combined with ﬂux
balance analysis allows for analysis of metabolic outputs given
a set of inputs (Varma and Palsson, 1994). Furthermore,
GSMMs can expand to community-level models (Zomorrodi
and Maranas, 2012;Khandelwal et al., 2013;Louca and Doebeli,
2015) to uncover how microbes synergistically create complex
wine metabolite proﬁles. Going forward, it will be critical
not only to deﬁne which microbes created your favorite
wine but also how their metabolisms shaped the taste of
that wine. Thus, viticulture will beneﬁt very much from the
generation of commercial platforms that enable studying vine
and wine microbiome and wine metabolome. Currently such
platforms are already in place, with WineSeqR
Makers, Inc.)2allowing wine microbiome characterization
through NGS and Wine ScreenerR
wine metabolome analysis by nuclear magnetic resonance.
These tools are based on robust databases and allow both
producers and regulatory councils from appellations of origins
to establish ‘standard proﬁles’ for their wines, and better
understand the microbial and chemical bases of their distinctive
In this article, the impact of NGS technologies in vine
and wine microbiology has been reviewed. Regarding the
importance of microbiome in viticulture and enology, the role
of microorganisms in the chemical and nutritional properties
of vineyard soils, crop health and yield, and also in the
later fermentation performance and wine ﬂavor are the main
challenges to explore using –omics tools. For that purpose,
certain technical aspects should be improved at laboratory
stages, such as universal DNA&RNA extraction protocols to
avoid biases, and improved sequencing approaches to increase
microbiome resolution and quantiﬁcation. It is also important
to develop robust and curated databases to improve taxonomic
assignments (Figure 1). Finally, it is time to develop big data
works, using statistical data-mining and machine learning tools
to solve, in a holistic systems-biology view, the above-mentioned
challenges in wine industry.
Frontiers in Microbiology | www.frontiersin.org 9May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 10
Belda et al. Microbiome in Enology and Viticulture
IB and AA conceived the work. IB, AP, and AA wrote the
“Introduction” section. IB and IZ wrote “The microbiome of
vine and wine: a review” section. AA, IZ, and MP wrote “NGS
microbial proﬁling: key steps, biases and limitations” section.
IZ and MP wrote “Bioinformatics and predictive methods to
uncover the microbial terroir” section. Finally, IB edited the ﬁnal
version of the manuscript.
This study was funded by WineSeq Project, BiomeMakers Inc.
Acinas, S. G., Sarma-Rupavtarm, R., Klepac-Ceraj, V., and Polz, M. F. (2005). PCR-
induced sequence artifacts and bias: insights from comparison of two 16S rRNA
clone libraries constructed from the same sample. Appl. Environ. Microbiol. 71,
8966–8969. doi: 10.1128/AEM.71.12.8966-8969.2005
Agler, M. T., Ruhe, J., Kroll, S., Morhenn, C., Kim, S. T., Weigel, D., et al. (2016).
Microbial hub taxa link host and abiotic factors to plant microbiome variation.
PLoS Biol. 14:e1002352. doi: 10.1371/journal.pbio.1002352
Amann, R. I., Ludwig, W., and Schleifer, K. H. (1995). Phylogenetic identiﬁcation
and in situ detection of individual microbial cells without cultivation. Microbiol.
Rev. 59, 143–169.
Andorrà, I., Landi, S., Mas, A., Esteve-Zarzoso, B., and Guillamón, J. M. (2010).
Eﬀect of fermentation temperature on microbial population evolution using
culture-independent and dependent techniques. Food Res. Int. 43, 773–779.
Apprill, A., McNally, S., Parsons, R., and Weber, L. (2015). Minor revision to
V4 region SSU rRNA 806R gene primer greatly increases detection of SAR11
bacterioplankton. Aquat. Microb. Ecol. 75, 129–137. doi: 10.3354/ame01753
Baker, G. C., Smith, J. J., and Cowan, D. A. (2003). Review and re-analysis of
domain-speciﬁc 16S primers. J. Microbiol. Methods 55, 541–555. doi: 10.1016/j.
Barata, A., Malfeito-Ferreira, M., and Loureiro, V. (2012). The microbial ecology
of wine grape berries. Int. J. Food Microbiol. 153, 243–259. doi: 10.1016/j.
Belda, I., Ruiz, J., Alastruey-Izquierdo, A., Navascués, E., Marquina, D., and
Santos, A. (2016). Unraveling the enzymatic basis of wine “ﬂavorome”: a
phylo-functional study of wine related yeast species. Front. Microbiol. 7:12.
Belda, I., Ruiz, J., Esteban-Fernández, A., Navascués, E., Marquina, D., Santos, A.,
et al. (2017). Microbial contribution to wine aroma and its intended use for wine
quality improvement. Molecules 22, E189. doi: 10.3390/molecules22020189
Benitez-Paez, A., Portune, K. J., and Sanz, Y. (2016). Species-level resolution of
16S rRNA gene amplicons sequenced through the MinION portable nanopore
sequencer. Gigascience 5, 4. doi: 10.1186/s13742-016-0111-z
Bertsch, C., Ramírez-Suero, M., Magnin-Robert, M., Larignon, P., Chong, J., Abou-
Mansour, E., et al. (2013). Grapevine trunk diseases: complex and still poorly
understood. Plant Pathol. 62, 243–265. doi: 10.1111/j.1365-3059.2012.02674.x
Bhattacharyya, P., and Jha, D. (2012). Plant growth-promoting rhizobacteria
(PGPR): emergence in agriculture. World J. Microb. Biot. 28, 1327–1350.
doi: 10.1007/s11274-011- 0979-9
Bokulich, N. A., Collins, T. S., Masarweh, C., Allen, G., Heymann, H., Ebeler,
S. E., et al. (2016). Associations among wine grape microbiome, metabolome,
and fermentation behavior suggest microbial contribution to regional wine
characteristics. mBio 7, e00631-16. doi: 10.1128/mBio.00631-16
Bokulich, N. A., Joseph, C. L., Allen, G., Benson, A. K., and Mills, D. A. (2012).
Next-generation sequencing reveals signiﬁcant bacterial diversity of botrytized
wine. PLoS ONE 7:e36357. doi: 10.1371/journal.pone.0036357
Bokulich, N. A., and Mills, D. A. (2012). Diﬀerentiation of mixed lactic acid bacteria
communities in beverage fermentations using targeted terminal restriction
fragment length polymorphism. Food Microbiol. 31, 126–132. doi: 10.1016/j.fm.
Bokulich, N. A., and Mills, D. A. (2013). Improved selection of internal transcribed
spacer-speciﬁc primers enables quantitative, ultra-high-throughput proﬁling of
fungal communities. Appl. Environ. Microbiol. 79, 2519–2526. doi: 10.1128/
Bokulich, N. A., Thorngate, J. H., Richardson, P. M., and Mills, D. A. (2014).
Microbial biogeography of wine grapes is conditioned by cultivar, vintage,
and climate. Proc. Natl. Acad. Sci. U.S.A. 111, E139–E148. doi: 10.1073/pnas.
Burns, K. N., Bokulich, N. A., Cantu, D., Greenhut, R. F., Kluepfel, D. A., O’Geen,
A. T., et al. (2016). Vineyard soil bacterial diversity and composition revealed by
16S rRNA genes: diﬀerentiation by vineyard management. Soil Biol. Biochem.
103, 337–348. doi: 10.1016/j.soilbio.2016.09.007
Burns, K. N., Kluepfel, D. A., Strauss, S. L., Bokulich, N. A., Cantu, D., and
Steenwerth, K. L. (2015). Vineyard soil bacterial diversity and composition
revealed by 16S rRNA genes: diﬀerentiation by geographic features. Soil Biol.
Biochem. 91, 232–247. doi: 10.1016/j.soilbio.2015.09.002
Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D.,
Costello, E. K., et al. (2010). QIIME allows analysis of high-throughput
community sequencing data. Nat. Methods 7, 335–336. doi: 10.1038/nmeth.
Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Huntley, J., Fierer, N.,
et al. (2012). Ultra-high-throughput microbial community analysis on the
Illumina HiSeq and MiSeq platforms. ISME J. 6, 1621–1624. doi: 10.1038/ismej.
Compant, S., Clément, C., and Sessitsch, A. (2010). Plant growth-promoting
bacteria in the rhizo-and endosphere of plants: their role, colonization,
mechanisms involved and prospects for utilization. Soil Biol. Biochem. 42,
669–678. doi: 10.1016/j.soilbio.2009.11.024
Courtois, S., Frostegård, Å., Göransson, P., Depret, G., Jeannin, P., and Simonet, P.
(2001). Quantiﬁcation of bacterial subgroups in soil: comparison of DNA
extracted directly from soil or from cells previously released by density gradient
centrifugation. Environ. Microbiol. 3, 431–439. doi: 10.1046/j.1462-2920.2001.
Curtis, T. P. (2002). Estimating prokaryotic diversity and its limits. Proc. Natl.
Acad. Sci. U.S.A. 99, 10494–10499. doi: 10.1073/pnas.142680199
David, V., Terrat, S., Herzine, K., Claisse, O., Rousseaux, S., Tourdot-Maréchal,
R., et al. (2014). High-throughput sequencing of amplicons for monitoring
yeast biodiversity in must and during alcoholic fermentation. J. Ind. Microbiol.
Biotechnol. 41, 811–821. doi: 10.1007/s10295-014-1427-2
Divol, B., and Lonvaud-Funel, A. (2005). Evidence for viable but nonculturable
yeasts in botrytis aﬀected wine. J. Appl. Microbiol. 99, 85–93. doi: 10.1111/j.
Edgar, R. C. (2013). UPARSE: highly accurate OTU sequences from
microbial amplicon reads. Nat. Methods 10, 996–998. doi: 10.1038/nmeth.
Edgar, R. C. (2016). SINTAX, a Simple Non-Bayesian Taxonomy Classiﬁer for 16S
and ITS Sequences. Available at: http://biorxiv.org/content/early/2016/09/09/
074161 doi: 10.1101/074161
Eren, A. M., Maignien, L., Sul, W. J., Murphy, L. G., Grim, S. L., Morrison, H. G.,
et al. (2013). Oligotyping: diﬀerentiating between closely related microbial taxa
using 16S rRNA gene data. Methods Ecol. Evol. 4, 1111–1119. doi: 10.1111/2041-
Eren, A. M., Sogin, M. L., and Maignien, L. (2016). Editorial: new insights into
microbial ecology through subtle nucleotide variation. Front. Microbiol. 7:1318.
Fontaine, F., Pinto, C., Vallet, J., Clément, C., Gomes, A. C., and Spagnolo, A.
(2016). The eﬀects of grapevine trunk diseases (GTDs) on vine physiology. Eur.
J. Plant Pathol. 144, 707–721. doi: 10.1007/s10658-015-0770- 0
Gilbert, J. A., Meyer, F., Jansson, J., Gordon, J., Pace, N., Tiedje, J., et al. (2010).
The earth microbiome project: meeting report of the “1 st EMP meeting on
sample selection and acquisition” at argonne national laboratory October 6th
2010. Stand. Genomic Sci. 3, 249–253. doi: 10.4056/aigs.1443528
Grangeteau, C., Roullier-Gall, C., Rousseaux, S., Gougeon, R. D., Schmitt-
Kopplin, P., Alexandre, H., et al. (2017). Wine microbiology is driven by
Frontiers in Microbiology | www.frontiersin.org 10 May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 11
Belda et al. Microbiome in Enology and Viticulture
vineyard and winery anthropogenic factors. Microb. Biotechnol. 10, 354–370.
Greuter, D., Loy, A., Horn, M., and Rattei, T. (2016). probeBase-an online resource
for rRNA-targeted oligonucleotide probes and primers: new features. Nucleic
Acids Res. 44, D586–D589. doi: 10.1093/nar/gkv1232
Haggerty, J. M., and Dinsdale, E. A. (2017). Distinct biogeographical patterns
of marine bacterial taxonomy and functional genes. Glob. Ecol. Biogeogr. 26,
177–190. doi: 10.1111/geb.12528
Herrick, J. B., Madsen, E., Batt, C., and Ghiorse, W. (1993). Polymerase chain
reaction ampliﬁcation of naphthalene-catabolic and 16S rRNA gene sequences
from indigenous sediment bacteria. Appl. Environ. Microbiol. 59, 687–694.
Hilger, A., and Myrold, D. (1991). Method for extraction of Frankia DNA from soil.
Agric. Ecosyst. Environ. 34, 107–113. doi: 10.1016/0167-8809(91)90098-I
Holben, W. E. (1994). “Isolation and puriﬁcation of bacterial DNA from soil,” in
Methods of Soil Analysis: Part 2—Microbiological and Biochemical Properties,
eds R. Weaver, P. Bottomly, and S. Angle (Madison, WI: Soil Science Society of
Hong, S., Bunge, J., Leslin, C., Jeon, S., and Epstein, S. S. (2009). Polymerase chain
reaction primers miss half of rRNA microbial diversity. ISME J. 3, 1365–1373.
Keisam, S., Romi, W., Ahmed, G., and Jeyaram, K. (2016). Quantifying the biases
in metagenome mining for realistic assessment of microbial ecology of naturally
fermented foods. Sci. Rep. 6:34155. doi: 10.1038/srep34155
Kennedy, J. (2002). Understanding grape berry development. Prac. Winery
Vineyard 24, 14–23.
Khandelwal, R. A., Olivier, B. G., Röling, W. F., Teusink, B., and Bruggeman, F. J.
(2013). Community ﬂux balance analysis for microbial consortia at balanced
growth. PLoS ONE 8:e64567. doi: 10.1371/journal.pone.0064567
Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M., et al.
(2013). Evaluation of general 16S ribosomal RNA gene PCR primers for classical
and next-generation sequencing-based diversity studies. Nucleic Acids Res. 41,
e1. doi: 10.1093/nar/gks808
Knight, S., Klaere, S., Fedrizzi, B., and Goddard, M. R. (2015). Regional microbial
signatures positively correlate with diﬀerential wine phenotypes: evidence for a
microbial aspect to terroir. Sci. Rep. 5:14233. doi: 10.1038/srep14233
Lee, C. K., Herbold, C. W., Polson, S. W., Wommack, K. E., Williamson,
S. J., McDonald, I. R., et al. (2012). Groundtruthing next-gen sequencing for
microbial ecology–biases and errors in community structure estimates from
PCR amplicon pyrosequencing. PLoS ONE 7:e44224. doi: 10.1371/journal.pone.
Leveau, J., and Tech, J. (2010). “Grapevine microbiomics: bacterial diversity
on grape leaves and berries revealed by high-throughput sequence analysis
of 16S rRNA amplicons,” in Proceedings of the International Symposium on
Biological Control of Postharvest Diseases: Challenges and Opportunities, Vol.
905, Leesburg, VA, 31–42.
Logares, R., Sunagawa, S., Salazar, G., Cornejo-Castillo, F. M., Ferrera, I.,
Sarmento, H., et al. (2014). Metagenomic 16S rDNA Illumina tags are a
powerful alternative to amplicon sequencing to explore diversity and structure
of microbial communities. Environ. Microbiol. 16, 2659–2671. doi: 10.1111/
Louca, S., and Doebeli, M. (2015). Calibration and analysis of genome-based
models for microbial ecology. eLife 4:e08208. doi: 10.7554/eLife.08208
López-Rituerto, E., Savorani, F., Avenoza, A., Busto, J. S. H., Peregrina, J. S. M.,
and Engelsen, S. B. (2012). Investigations of La Rioja terroir for wine
production using 1H NMR metabolomics. J. Agric. Food Chem. 60, 3452–3461.
Lugtenberg, B., and Kamilova, F. (2009). Plant-growth-promoting rhizobacteria.
Annu. Rev. Microbiol. 63, 541–556. doi: 10.1146/annurev.micro.62.081307.
Lundberg, D. S., Yourstone, S., Mieczkowski, P., Jones, C. D., and Dangl, J. L.
(2013). Practical innovations for high-throughput amplicon sequencing. Nat.
Methods 10, 999–1002. doi: 10.1038/nmeth.2634
Mahé, F., Rognes, T., Quince, C., de Vargas, C., and Dunthorn, M. (2014). Swarm:
robust and fast clustering method for amplicon-based studies. PeerJ 2:e593.
Mahé, F., Rognes, T., Quince, C., de Vargas, C., and Dunthorn, M. (2015). Swarm
v2: highly-scalable and high-resolution amplicon clustering. PeerJ 3:e1420.
Martins, G., Lauga, B., Miot-Sertier, C., Mercier, A., Lonvaud, A., Soulas, M.-
L., et al. (2013). Characterization of epiphytic bacterial communities from
grapes, leaves, bark and soil of grapevine plants grown, and their relations.
PLoS ONE 8:e73013. doi: 10.1371/journal.pone.0073013
McDonald, D., Price, M. N., Goodrich, J., Nawrocki, E. P., DeSantis, T. Z.,
Probst, A., et al. (2012). An improved Greengenes taxonomy with explicit ranks
for ecological and evolutionary analyses of bacteria and archaea. ISME J. 6,
610–618. doi: 10.1038/ismej.2011.139
McMurdie, P. J., and Holmes, S. (2014). Waste not, want not: why rarefying
microbiome data is inadmissible. PLoS Comput. Biol. 10:e1003531. doi: 10.1371/
Miller, D., Bryant, J., Madsen, E., and Ghiorse, W. (1999). Evaluation and
optimization of DNA extraction and puriﬁcation procedures for soil and
sediment samples. Appl. Environ. Microbiol. 65, 4715–4724.
Millet, V., and Lonvaud-Funel, A. (2000). The viable but non-culturable state of
microorganisms during storage. Lett. Appl. Microbiol.30, 136–141. doi: 10.1046/
Neeley, E. T., Phister, T. G., and Mills, D. A. (2005). Diﬀerential real-time PCR
assay for enumeration of lactic acid bacteria in wine. Appl. Environ. Microbiol.
71, 8954–8957. doi: 10.1128/AEM.71.12.8954-8957.2005
Neilson, J. W., Jordan, F. L., and Maier, R. M. (2013). Analysis of artifacts suggests
DGGE should not be used for quantitative diversity analysis. J. Microbiol.
Methods 92, 256–263. doi: 10.1016/j.mimet.2012.12.021
OIV (2015). World Vitiviniculture Situation 2015. Paris: Oﬃce international de la
vigne et du vin.
Parada, A. E., Needham, D. M., and Fuhrman, J. A. (2016). Every base matters:
assessing small subunit rRNA primers for marine microbiomes with mock
communities, time series and global ﬁeld samples. Environ. Microbiol. 18,
1403–1414. doi: 10.1111/1462-2920.13023
Paulson, J. N., Stine, O. C., Bravo, H. C., and Pop, M. (2013). Diﬀerential abundance
analysis for microbial marker-gene surveys. Nat. Methods 10, 1200–1202.
Petric, I., Philippot, L., Abbate, C., Bispo, A., Chesnot, T., Hallin, S., et al. (2011).
Inter-laboratory evaluation of the ISO standard 11063 “Soil quality—Method
to directly extract DNA from soil samples”. J. Microbiol. Methods 84, 454–460.
Piao, H., Hawley, E., Kopf, S., DeScenzo, R., Sealock, S., Henick-Kling, T., et al.
(2015). Insights into the bacterial community and its temporal succession
during the fermentation of wine grapes. Front. Microbiol. 6:809. doi: 10.3389/
Picard, C., Ponsonnet, C., Paget, E., Nesme, X., and Simonet, P. (1992). Detection
and enumeration of bacteria in soil by direct DNA extraction and polymerase
chain reaction. Appl. Environ. Microbiol. 58, 2717–2722.
Pinto, A. J., and Raskin, L. (2012). PCR biases distort bacterial and archaeal
community structure in pyrosequencing datasets. PLoS ONE 7:e43093.
Pinto, C., Pinho, D., Cardoso, R., Custodio, V., Fernandes, J., Sousa, S.,
et al. (2015). Wine fermentation microbiome: a landscape from diﬀerent
Portuguese wine appellations. Front. Microbiol. 6:905. doi: 10.3389/fmicb.2015.
Pinto, C., Pinho, D., Sousa, S., Pinheiro, M., Egas, C., and Gomes, A. C. (2014).
Unravelling the diversity of grapevine microbiome. PLoS ONE 9:e85622.
Porteous, L., Seidler, R., and Watrud, L. (1997). An improved method for purifying
DNA from soil for polymerase chain reaction ampliﬁcation and molecular
ecology applications. Mol. Ecol. 6, 787–791. doi: 10.1046/j.1365-294X.1997.
Portillo, M. D. C., and Mas, A. (2016). Analysis of microbial diversity and dynamics
during wine fermentation of Grenache grape variety by high-throughput
barcoding sequencing. Food Sci. Technol. LEB 72, 317–321. doi: 10.1016/j.lwt.
Ranjard, L., Poly, F., Combrisson, J., Richaume, A., and Nazaret, S. (1998). A single
procedure to recover DNA from the surface or inside aggregates and in various
size fractions of soil suitable for PCR-based assays of bacterial communities.
Eur. J. Soil Biol. 34, 89–97. doi: 10.1016/S1164-5563(99)90006-7
Rognes, T., Flouri, T., Nichols, B., Quince, C., and Mahé, F. (2016). VSEARCH: a
versatile open source tool for metagenomics. PeerJ. 4, e2584. doi: 10.7717/peerj.
Frontiers in Microbiology | www.frontiersin.org 11 May 2017 | Volume 8 | Article 821
fmicb-08-00821 May 4, 2017 Time: 16:30 # 12
Belda et al. Microbiome in Enology and Viticulture
Salma, M., Rousseaux, S., Sequeira-Le Grand, A., Divol, B., and Alexandre,
H. (2013). Characterization of the Viable but Nonculturable (VBNC) state
in Saccharomyces cerevisiae.PLoS ONE 8:e77600. doi: 10.1371/journal.pone.
Salter, S. J., Cox, M. J., Turek, E. M., Calus, S. T., Cookson, W. O., Moﬀatt,
M. F., et al. (2014). Reagent and laboratory contamination can critically impact
sequence-based microbiome analyses. BMC Biol. 12:87. doi: 10.1186/s12915-
Schirmer, M., Ijaz, U. Z., D’Amore, R., Hall, N., Sloan, W. T., and Quince, C.
(2015). Insight into biases and sequencing errors for amplicon sequencing
with the Illumina MiSeq platform. Nucleic Acids Res. 43:e37. doi: 10.1093/nar/
Schloss, P. D., Jenior, M. L., Koumpouras, C. C., Westcott, S. L., and
Highlander, S. K. (2016). Sequencing 16S rRNA gene fragments using the
PacBio SMRT DNA sequencing system. PeerJ. 4:e1869. doi: 10.7717/peerj.
Schloss, P. D., Westcott, S. L., Ryabin, T., Hall, J. R., Hartmann, M., Hollister,
E. B., et al. (2009). Introducing mothur: open-source, platform-independent,
community-supported software for describing and comparing microbial
communities. Appl. Environ. Microbiol. 75, 7537–7541. doi: 10.1128/AEM.
Segata, N., Izard, J., Waldron, L., Gevers, D., Miropolsky, L., Garrett, W. S.,
et al. (2011). Metagenomic biomarker discovery and explanation. Genome Biol.
12:R60. doi: 10.1186/gb-2011- 12-6-r60
Seguritan, V., and Rohwer, F. (2001). FastGroup: a program to dereplicate
libraries of 16S rDNA sequences. BMC Bioinformatics 2:9. doi: 10.1186/1471-
Singer, E., Bushnell, B., Coleman-Derr, D., Bowman, B., Bowers, R. M., Levy, A.,
et al. (2016). High-resolution phylogenetic microbial community proﬁling.
ISME J. 10, 2020–2032. doi: 10.1038/ismej.2015.249
Sipos, R., Székely, A. J., Palatinszky, M., Révész, S., Márialigeti, K., and
Nikolausz, M. (2007). Eﬀect of primer mismatch, annealing temperature
and PCR cycle number on 16S rRNA gene-targetting bacterial community
analysis. FEMS Microbiol. Ecol. 60, 341–350. doi: 10.1111/j.1574-6941.2007.
Soergel, D. A., Dey, N., Knight, R., and Brenner, S. E. (2012). Selection of
primers for optimal taxonomic classiﬁcation of environmental 16S rRNA gene
sequences. ISME J. 6, 1440–1444. doi: 10.1038/ismej.2011.208
Statnikov, A., Henaﬀ, M., Narendra, V., Konganti, K., Li, Z., Yang, L., et al.
(2013). A comprehensive evaluation of multicategory classiﬁcation methods for
microbiomic data. Microbiome 1:11. doi: 10.1186/2049-2618-1-11
Stefanini, I., Albanese, D., Cavazza, A., Franciosi, E., De Filippo, C., Donati, C., et al.
(2016). Dynamic changes in microbiota and mycobiota during spontaneous
‘Vino Santo Trentino’fermentation. Microb. Biotechnol. 9, 195–208.
Tikhonov, M., Leach, R. W., and Wingreen, N. S. (2015). Interpreting 16S
metagenomic data without clustering to achieve sub-OTU resolution. ISME J.
9, 68–80. doi: 10.1038/ismej.2014.117
Tofalo, R., Scirone, M., Corsetti, A., and Suzzi, G. (2012). Detection of
Brettanomyces spp. in red wines using Real-Time PCR. J. Food Sci. 77, 545–549.
Urakawa, H., Martens-Habbena, W., and Stahl, D. A. (2010). High abundance of
ammonia-oxidizing Archaea in coastal waters, determined using a modiﬁed
DNA extraction method. Appl. Environ. Microbiol. 76, 2129–2135. doi: 10.1128/
Vandenkoornhuyse, P., Quaiser, A., Duhamel, M., Le Van, A., and Dufresne, A.
(2015). The importance of the microbiome of the plant holobiont. New Phytol.
206, 1196–1206. doi: 10.1111/nph.13312
Varma, A., and Palsson, B. O. (1994). Stoichiometric ﬂux balance models
quantitatively predict growth and metabolic by-product secretion in wild-type
Escherichia coli W3110. Appl. Environ. Microbiol. 60, 3724–3731.
Verginer, M., Leitner, E., and Berg, G. (2010). Production of volatile metabolites
by grape-associated microorganisms. J. Agric. Food Chem. 58, 8344–8350.
Wang, C., García-Fernández, D., Mas, A., and Esteve-Zarzoso, B. (2015). Fungal
diversity in grape must and wine fermentation assessed by massive sequencing,
quantitative PCR and DGGE. Front. Microbiol. 6:1156. doi: 10.3389/fmicb.2015.
Wagner, M. R., Lundberg, D. S., Coleman-Derr, D., Tringe, S. G., Dangl, J. L., and
Mitchell-Olds, T. (2014). Natural soil microbes alter ﬂowering phenology and
the intensity of selection on ﬂowering time in a wild Arabidopsis relative. Ecol.
Lett. 17, 717–726. doi: 10.1111/ele.12276
Weiss, S., Van Treuren, W., Lozupone, C., Faust, K., Friedman, J., Deng, Y., et al.
(2016). Correlation detection strategies in microbial data sets vary widely in
sensitivity and precision. ISME J. 10, 1669–1681. doi: 10.1038/ismej.2015.235
Yamamoto, H. (2000). Viable but nonculturable state as a general phenomenon of
non-spore-forming bacteria, and its modeling. J. Infect. Chemother. 6, 112–114.
Zarraonaindia, I., Owens, S. M., Weisenhorn, P., West, K., Hampton-Marcell, J.,
Lax, S., et al. (2015). The soil microbiome inﬂuences grapevine-associated
microbiota. MBio 6, e02527-14. doi: 10.1128/mBio.02527-14
Zomorrodi, A. R., and Maranas, C. D. (2012). OptCom: a multi-level optimization
framework for the metabolic modeling and analysis of microbial communities.
PLoS Comput. Biol. 8:e1002363. doi: 10.1371/journal.pcbi.1002363
Conﬂict of Interest Statement: The authors declare that the research was
conducted in the absence of any commercial or ﬁnancial relationships that could
be construed as a potential conﬂict of interest.
Copyright © 2017 Belda, Zarraonaindia, Perisin, Palacios and Acedo. This is an
open-access article distributed under the terms of the Creative Commons Attribution
License (CC BY). The use, distribution or reproduction in other forums is permitted,
provided the original author(s) or licensor are credited and that the original
publication in this journal is cited, in accordance with accepted academic practice.
No use, distribution or reproduction is permitted which does not comply with these
Frontiers in Microbiology | www.frontiersin.org 12 May 2017 | Volume 8 | Article 821