Food Microbiology 73 (2018) 351e361

Contents lists available at ScienceDirect

Food Microbiology
journal homepage: www.elsevier.com/locate/fm

Oxygen and diverse nutrients influence the water kefir fermentation

process
David Laureys a, Maarten Aerts b, Peter Vandamme b, Luc De Vuyst a, *
a
Research Group of Industrial Microbiology and Food Biotechnology, Faculty of Sciences and Bioengineering Sciences, Vrije Universiteit Brussel, Pleinlaan 2,
1050 Brussels, Belgium
b
Laboratory of Microbiology, Department of Biochemistry and Microbiology, Faculty of Sciences, Ghent University, K. L. Ledeganckstraat 35, 9000 Ghent,
Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Eight water kefir fermentation series differing in the presence of oxygen, the nutrient concentration, and
Received 3 August 2017 the nutrient source were studied during eight consecutive backslopping steps. The presence of oxygen
Received in revised form allowed the proliferation of acetic acid bacteria, resulting in high concentrations of acetic acid, and
13 January 2018
decreased the relative abundance of Bifidobacterium aquikefiri. Low nutrient concentrations resulted in
Accepted 8 February 2018
Available online 13 February 2018
slow water kefir fermentation and high pH values, which allowed the growth of Comamonas testosteroni/
thiooxydans. Further, low nutrient concentrations favored the growth of Lactobacillus hilgardii and Dek-
kera bruxellensis, whereas high nutrient concentrations favored the growth of Lactobacillus nagelii and
Keywords:
Water kefir
Saccharomyces cerevisiae. Dried figs, dried apricots, and raisins resulted in stable water kefir fermenta-
Yeasts tion. Water kefir fermentation with dried apricots resulted in the highest pH and water kefir grain
Lactic acid bacteria growth, whereas that with raisins resulted in the lowest pH and water kefir grain growth. Further, water
Acetic acid bacteria kefir fermentation with raisins resembled fermentations with low nutrient concentrations, that with
Bifidobacteria dried apricots resembled fermentations with normal nutrient concentrations, and that with fresh figs or
Nutrient a mixture of yeast extract and peptone resembled fermentations with high nutrient concentrations.
Oxygen © 2018 Elsevier Ltd. All rights reserved.

1. Introduction microorganisms of water kefir fermentation are the lactic acid

bacteria (LAB) species Lactobacillus hilgardii, Lactobacillus nagelii
Water kefir is a traditional fermented beverage that is made by and Lactobacillus paracasei; and the yeast species Saccharomyces
adding water kefir grains (the inoculum) to a mixture of water, cerevisiae (Laureys and De Vuyst, 2017). Sucrose is the main sub-
(dried) fruits, and sugar (Gulitz et al., 2013; Laureys and De Vuyst, strate for the water kefir microorganisms and is converted into
2014; Laureys et al., 2017; Marsh et al., 2013; Stadie et al., 2013). water kefir grain EPS, ethanol, carbon dioxide, lactic acid, glycerol,
Usually, this mixture is fermented at room temperature under acetic acid, mannitol, and a variety of aroma compounds (Laureys
anaerobic conditions for two to four days, after which it is sieved to and De Vuyst, 2014; Martínez-Torres et al., 2017).
separate the water kefir grains from the water kefir liquor. The The contents of the vessel wherein water kefir fermentation
water kefir liquor is a slightly sweet, acidic, alcoholic, and sparkling takes place are usually separated from the atmosphere by a rubber
beverage that has a yellowish color and a fruity aroma. The water sealing or water lock (Gulitz et al., 2011; Laureys and De Vuyst,
kefir grains consist of dextran exopolysaccharides (EPS), are 2014; Pidoux, 1989; Stadie et al., 2013). These configurations pre-
translucent, have a brittle structure, and are insoluble in water vent the ingress of atmospheric oxygen, but allow the release of
(Laureys and De Vuyst, 2014; Waldherr et al., 2010). Many different carbon dioxide, thus preventing excessive pressure build-up in the
microorganisms occur on the water kefir grains, whereby the key fermentation vessel. Consequently, the water kefir fermentation
process starts aerobically and becomes gradually anaerobic, as ox-
ygen is consumed and/or flushed out by the carbon dioxide pro-
duced by the yeasts. Oxygen can have an impact on the growth and
* Corresponding author. Research Group of Industrial Microbiology and Food
metabolism of several of the water kefir microorganisms, such as
Biotechnology, Faculty of Sciences and Bioengineering Sciences, Vrije Universiteit
Brussel, Pleinlaan 2, 1050 Brussels, Belgium. on yeasts (Aceituno et al., 2012) and acetic acid bacteria (AAB)
E-mail address: [email protected] (L. De Vuyst). (Guillamo
n and Mas, 2009), suggesting that the presence of oxygen

https://doi.org/10.1016/j.fm.2018.02.007
0740-0020/© 2018 Elsevier Ltd. All rights reserved.
352 D. Laureys et al. / Food Microbiology 73 (2018) 351e361

might influence the microbial species diversity and/or metabolite conditions, the fermentation mixtures were supplemented with 5 g
production during water kefir fermentation. of dried figs (1DF-An), 5 g of dried apricots (1DA-An), 5 g of dried
The water used for fermentation contains calcium ions and raisins (1DR-An), 17 g of fresh figs (1FF-An), or 1 ml of autoclaved
buffer compounds necessary for optimal water kefir grain growth yeast extract-peptone (YP) solution (YP-An). The YP solution was
(D. Laureys, M. Aerts, P. Vandamme, and L. De Vuyst, unpublished prepared by adding 125 g l 1 of yeast extract (Merck, Darmstadt,
results). Other nutrients necessary for water kefir fermentation, Germany) and 125 g l 1 of bacteriological peptone (Oxoid, Basing-
such as amino acids, vitamins, and minerals are provided by the stoke, UK) to ultrapure water (gradient A10 Milli-Q water purifi-
(dried) fruits added to the fermentation mixture. Although fruits cation system; EMD Millipore, Billerica, MA, USA), after which this
are rich in such nutrients, the relatively small amount of (dried) mixture was sterilized by autoclaving (121  C, 2.1 bar, 20 min). The
fruits in the recipe makes the water kefir fermentation medium Schott bottles were equipped with a PTFE water lock for fermen-
relatively poor in nutrients. As (dried) fruits are usually the sole tations under anaerobic conditions (0DF-An, 1DF-An, 2DF-An, 1DA-
source of a variety of important nutrients during water kefir An, 1DR-An, 1FF-An, and YP-An) or were covered with a muslin
fermentation, the amount and/or types of fruits used for fermen- cloth for fermentations under aerobic conditions (1DF-Ae). All
tation might have an impact on the microbial species diversity, bottles were incubated in a water bath at 21  C. The contents of the
substrate consumption, and/or metabolite production during fermentation bottles were mixed by gently turning the bottles at
fermentation. Dried figs are the most common fruits used for water the start and at the end of each backslopping step. Every 3 d, the
kefir fermentation (Gulitz et al., 2011; Laureys and De Vuyst, 2014; backslopping practice was applied for each fermentation bottle,
Pidoux, 1989; Stadie et al., 2013), but raisins, plums, or dates have whereby the water kefir grains were separated from the water kefir
also been used (Reiß, 1990). liquors by sieving, rinsed, after which 50 g of water kefir grains
This study aimed to investigate the influence of the presence of were recultivated in fresh medium with the same composition and
oxygen, the nutrient concentration, and the nutrient source on the under the same conditions as before. This practice was continued
microbial species diversity, water kefir grain growth, substrate for eight backslopping steps.
consumption, and metabolite production during water kefir
fermentation. 2.3. Analyses

2. Materials and methods The pH and the water kefir grain wet mass were determined at
the end of each backslopping step. The water kefir grain dry mass
2.1. Water kefir grain inoculum and prefermentations was determined at the end of backslopping step 8. The viable
counts of the LAB, yeasts, and AAB were determined for the non-
A water kefir grain inoculum was obtained from the household rinsed water kefir grains of the inoculum and the eight fermenta-
water kefir fermentation process of a private person (Ghent, tion series at the end of backslopping step 8. The culture-
Belgium), as described before (Laureys and De Vuyst, 2014). To dependent microbial species diversity of the LAB, yeasts, and AAB
obtain the necessary amount of water kefir grains, the inoculum was determined for the non-rinsed water kefir grains of the inoc-
was cultivated through a series of consecutive prefermentations ulum and the eight fermentation series at the end of backslopping
through backslopping until >1300 g of water kefir grain wet mass step 8. The culture-independent microbial species diversity was
was produced. The prefermentations were performed in glass determined for the water kefir liquors and the non-rinsed water
Schott bottles (1, 2, and 5 l) equipped with a polytetrafluoro- kefir grains of the inoculum and the eight water kefir fermentation
ethylene (PTFE) water lock. They were started by adding 10 g of series at the end of backslopping step 8. The substrate and
sugar (Candico Bio, Merksem, Belgium), 5 g of dried figs (King metabolite concentrations were determined for the water kefir li-
Brand, Naziili, Turkey), and 160 ml of tap water (Brussels, Belgium) quors of the eight fermentation series at the end of backslopping
per 50 g of water kefir grains. The bottles were incubated in a water steps 1 and 8. At the end of backslopping step 8, the water kefir
bath at 21  C. Every 3 d, the backslopping practice was applied, grains were assessed visually.
whereby the water kefir grains were separated from the water kefir The results are presented as the mean ± standard deviation of
liquors by sieving, and then recultivated in fresh medium under the the three independent biological replicates performed for each
same conditions as described above. fermentation series.

2.2. Fermentations 2.4. pH, water kefir grain wet and dry mass, and water kefir grain
growth determinations
The water kefir grain mass, obtained through the series of pre-
fermentations described above, was rinsed and used to start eight The pH, the water kefir grain wet mass, the water kefir grain
series of water kefir fermentations differing in the presence of growth, and the water kefir grain dry mass were determined as
oxygen, the nutrient concentration, and the nutrient source for described before (Laureys and De Vuyst, 2014), except for the fact
fermentation. Rinsing of the grains was performed with 2 l of tap that the water kefir grains were not rinsed with saline.
water (Brussels, Belgium) per 50 g of water kefir grains. Each
fermentation series was performed in independent biological 2.5. Microbial enumerations
triplicates. All fermentations were carried out in 250-ml Schott
bottles. They were started with 10 g of sugar (Candico Bio), 160 ml The viable counts of the presumptive LAB and AAB were
of tap water (Brussels, Belgium), and 50 g of rinsed water kefir determined as described before (Laureys and De Vuyst, 2014),
grains. To study the influence of oxygen, the fermentation mixtures except for the fact that an additional antibiotic, amphotericin B
were supplemented with 5 g of dried figs and incubated under (final concentration of 0.0025 g l 1; Sigma-Aldrich, Saint Louis, MO,
anaerobic (fermentation series 1DF-An) or aerobic conditions (1DF- USA), was added to the de Man-Rogosa-Sharpe (MRS) and modified
Ae). To study the influence of the nutrient concentration under deoxycholate-mannitol-sorbitol (mDMS) agar media. The viable
anaerobic conditions, the fermentation mixtures were supple- counts of the presumptive yeasts were determined on yeast
mented with 0 (0DF-An), 5 (1DF-An), or 10 g of dried figs (2DF-An). extract-peptone-dextrose (YPD) agar medium supplemented with
To study the influence of the nutrient source under anaerobic chloramphenicol (final concentration of 0.1 g l 1; Sigma-Aldrich),
 D. Laureys et al. / Food Microbiology 73 (2018) 351e361 353

as described before (Laureys and De Vuyst, 2014). Germany), according to the instructions of the manufacturer, and
the DNA solutions were adjusted at approximately 50 ng ml 1.
2.6. Culture-dependent microbial species diversity analyses The culture-independent microbial community profiles were
obtained by amplifying selected genomic fragments in the total
The culture-dependent microbial species diversity analyses of DNA with the universal prokaryotic primer pair (V3), the LAB-
the LAB, AAB, and yeasts on the water kefir grains were determined specific primer pair (LAC), the Bifidobacterium-specific primer pair
by randomly picking up 10e20% of the total number of colonies (Bif), and the universal eukaryotic primer pair (Yeast); and sepa-
from the respective agar media with 30e300 colonies. Isolates rating the PCR amplicons through denaturing gradient gel elec-
were subcultivated on their respective agar media until the third trophoresis (DGGE), as described before (Laureys and De Vuyst,
generation and used for dereplication by matrix-assisted laser 2014). Selected bands in the community profiles were cut from
desorption/ionization time-of-flight mass spectrometry (MALDI- the gels and identified through sequencing, as described before
TOF MS) fingerprinting, as described before (Spitaels et al., 2014). (Laureys and De Vuyst, 2014).
The fingerprint peptide patterns obtained were clustered numeri-
cally by means of the BioNumerics software version 7.50 (Applied 2.9. Substrate and metabolite concentration determinations
Maths, Sint-Martens-Latem, Belgium). Representative bacterial
isolates within each cluster were identified by sequencing part of Samples were prepared as described before (Laureys and De
their 16S rRNA gene from genomic DNA, as described before Vuyst, 2014). The concentrations of sucrose, glucose, and fructose
(Laureys and De Vuyst, 2014). Representative yeast isolates within were determined through high-performance anion exchange
each cluster were identified by sequencing part of their 26S large chromatography with pulsed amperometric detection (HPAEC-
subunit (LSU) rRNA gene and internal transcribed spacer (ITS) re- PAD), as described before (Laureys and De Vuyst, 2014), except that
gion from genomic DNA, as described before (Laureys and De Vuyst, 100 ml of cell-free supernatant was added to 400 ml of ultrapure
2014). water, and 100 ml of this dilution was added to 900 ml of deprotei-
nization solution. The concentrations of D- and L-lactic acid and
2.7. Exopolysaccharide production acetic acid were determined through high-performance liquid
chromatography with ultraviolet detection (HPLC-UV), those of
All bacterial isolates were grown on MRS agar medium sup- glycerol and mannitol through HPAEC-PAD, those of ethanol
plemented with 10 g l 1 of sucrose at 30  C for 7 d to visually assess through gas chromatography with flame ionization detection (GC-
their EPS production capacity. FID), and those of the aroma compounds through static headspace
gas chromatography with mass spectrometry detection (SH-GC-
2.8. Culture-independent microbial species diversity analyses MS), as described before (Laureys and De Vuyst, 2014).

The culture-independent microbial species diversity analyses of 2.10. Carbon recovery

the bacteria and yeasts in the water kefir liquors and on the water
kefir grains were determined after preparing total DNA extracts The carbon recovery at the end of a backslopping step was
from the cell pellets of the water kefir liquors and 0.2 g of crushed calculated as described before (Laureys and De Vuyst, 2014),
water kefir grains, respectively. whereby the total amount of carbon including that of the fruits
An optimized protocol was used for DNA extraction, as follows. added to the fermentation was taken into account. The mono- and
The pellets of the liquors and the grains were resuspended in 1 ml disaccharide contents (m m 1) of dried figs (48%), dried apricots
of TES buffer [6.7% (m v 1) sucrose, 50 mM Tris-base, 1 mM EDTA, (53%), dried raisins (59%), and fresh figs (16%) were obtained from
pH 8.0], after which the suspensions were centrifuged (21,000  g, the national nutrient database for standard reference (release 26,
20 min, 4  C) and the supernatants were discarded. The resulting http://ndb.nal.usda.gov/). Those of the YP solution were assumed to
pellets were resuspended in 1 ml of sorbitol buffer [1.2 M sorbitol, be 0%.
50 mM Tris-base, pH 7.5] supplemented with 30 mM 2-mercap-
toethanol, 200 U of lyticase (Sigma-Aldrich), and 15 U of Zymolyase 2.11. Statistics
(Sigma-Aldrich). These suspensions were incubated at 30  C for 1 h,
after which they were centrifuged (10,000  g, 10 min) and the An ANOVA was performed in R 3.2.0 to test for differences be-
supernatants were discarded. Then, the pellets were resuspended tween the eight water kefir fermentation series, followed by a se-
in 1 ml of sorbitol buffer, after which the suspensions were ries of post-hoc pairwise comparisons with Fisher's least significant
centrifuged (10,000  g, 10 min) and the supernatants were dis- difference (LSD) test (de Winter, 2013). All statistical tests were
carded. Finally, the pellets were resuspended in 400 ml of STET performed with a significance level of 0.05.
buffer [8.0% (m v 1) sucrose, 50 mM Tris-base, 50 mM EDTA, 5.0% (v
v 1) Triton X-100, pH 8.0] supplemented with 12.5 U of muta- 3. Results
nolysin (Sigma-Aldrich) and 20 mg ml 1 of lysozyme (Merck).
These suspensions were incubated at 37  C for 1 h. A pinch of acid 3.1. Influence of oxygen
washed glass beads (Sigma-Aldrich), 40 ml of 0.2 g ml 1 of sodium
dodecyl sulfate, and 50 ml of 2 mg ml 1 proteinase K solution For all water kefir fermentation processes performed, the
(Merck) were added. These suspensions were vortexed for 60 s and methods used did not accurately control the oxygen level, as the
incubated at 56  C for 2 h. The suspensions were heated until 65  C concentrations of oxygen in the liquors or headspaces of the Schott
and supplemented with 100 ml of 5 M NaCl and 80 ml of 10% (m v 1) bottles were not measured during the fermentation processes.
cetyl trimethylammonium bromide in 0.7 M NaCl at 65  C, vor- However, the inoculation of 50 g of water kefir grains per 160 ml of
texed, and incubated at 65  C for 10 min. Finally, the suspensions liquor ensured high microbial counts from the start of the
were supplemented with 600 ml of chlor- fermentation, as shown before (Laureys and De Vuyst, 2014, 2017).
oform:phenol:isoamylalcohol (49.5:49.5:1.0), vortexed, and Further, as the 250 mL Schott bottles were filled with 210 mL of
centrifuged (13,000  g, 5 min). The DNA obtained was purified total volume, a headspace of only 40 mL favored anaerobiosis.
with the Nucleospin® tissue 96 kit (Macherey-Nagel, Düren, Hence, although oxygen was initially found in the headspace of the
354 D. Laureys et al. / Food Microbiology 73 (2018) 351e361

bottles for the anaerobic treatments, the oxygen consumption for Gluconobacter roseus/oxydans and Acetobacter indonesiensis, and
this set-up was, therefore, relatively fast. Furthermore, high mi- the main AAB species in the aerobic fermentation series was Ace-
crobial counts also ensured a high production rate of carbon di- tobacter fabarum.
oxide at the start of the fermentation processes, which flushed out a
substantial part of the low oxygen concentrations from the
3.1.4. Culture-independent microbial species diversity
fermentation bottles rapidly. Finally, the present study wanted to
At the end of backslopping step 8, the rRNA-PCR-DGGE com-
approach the daily practice as much as possible, which means the
munity profiles obtained with the four different primer pairs used
use of a water lock to mimic anaerobic conditions and exposure to
(V3, LAC, Bif, and Yeast) were similar for the three independent
air when mimicking aerobic conditions.
biological replicates performed for each fermentation series (data
not shown).
3.1.1. pH and water kefir grain wet and dry mass The main bands in the community profiles of the inoculum, the
The pH, the water kefir grain growth (based on wet mass), and aerobic fermentation series, and the anaerobic fermentation series
the water kefir dry mass were similar in the aerobic and anaerobic obtained with the Yeast primer pair were attributed to S. cerevisiae
fermentation series at the end of backslopping step 1 (Table 1). and D. bruxellensis (Fig. 3). The relative intensities of the bands
Except for the dry mass, they decreased slightly over the course of attributed to S. cerevisiae were higher than those of the bands
the eight backslopping steps in the aerobic fermentation series attributed to D. bruxellensis. The relative intensities of the bands
(Fig. 1; Tables S1 and S2). attributed to D. bruxellensis were higher in the liquors than on the
grains. Additionally, a band with weak relative intensity, which was
3.1.2. Microbial enumerations attributed to Candida smithsonii, was detected in the liquors of the
The viable counts of the LAB and yeasts, and the ratios of the aerobic and anaerobic fermentation series.
viable counts of the LAB to the yeasts were similar in the anaerobic The main bands in the community profiles of the inoculum, the
and aerobic fermentation series (Table 2). The viable counts of the aerobic fermentation series, and the anaerobic fermentation series
AAB were significantly higher in the aerobic fermentation series obtained with the V3 and LAC primer pairs were attributed to Lb.
than in the anaerobic ones. paracasei, Lb. hilgardii, and Lb. nagelii. Further, a band in the com-
munity profiles obtained with the V3 and Bif primer pairs of the
3.1.3. Culture-dependent microbial species diversity water kefir liquors and water kefir grains of the inoculum and the
The culture-dependent species diversity of the yeasts, LAB, and aerobic and anaerobic fermentation series was attributed to Bifi-
AAB in the aerobic and anaerobic fermentation series was similar dobacterium aquikefiri. The relative intensities of these bands were
and more or less comparable to the inoculum (Fig. 2). Two yeast similar for the inoculum and the anaerobic fermentation series, but
species were found, whereby the relative abundances of lower for the aerobic ones. In the community profiles obtained with
S. cerevisiae were always higher than those of Dekkera bruxellensis. the V3 primer pair, several bands that were attributed to the taxon
The main LAB species were Lb. paracasei, Lb. hilgardii (67% of the Acetobacteraceae were detected in the water kefir liquors but not in
strains produced EPS), and Lb. nagelii. Additionally, Lactobacillus the water kefir grains of the aerobic fermentation series. These
harbinensis strains were found in the water kefir grain inoculum bands were not detected in the liquors and grains of the inoculum
and in the anaerobic fermentation series. The main AAB species in and the anaerobic fermentation series either. The limited length of
the inoculum and anaerobic fermentation series were the amplified 16S rRNA gene fragments (±210 bp) from these bands

Table 1
Characteristics of eight water kefir fermentation series differing in the presence of oxygen, nutrient concentrations, and nutrient source [anaerobic control fermentation with
dried figs (1DF-An); aerobic fermentation with dried figs (1DF-Ae); anaerobic fermentation with low (0DF-An) and high (2DF-An) amounts of dried figs; and anaerobic
fermentation with dried apricots (1DA-An), dried raisins (1DR-An), fresh figs (1FF-An), and a mixture of yeast extract and peptone (YP-An)] at the end of backslopping step 1.
Significant differences (p < 0.05) between the series are indicated with different superscripts (a, b, c, d, e, and f).

Characteristic 0DF-An 1DF-An 1DF-Ae 2DF-An 1DA-An 1DR-An 1FF-An YP-An

Water kefir grain growth (%) 65.9 ± 2.8 ab 63.4 ± 1.9 b 62.7 ± 0.4 b 58.0 ± 2.5 c 69.2 ± 1.1 a 51.7 ± 4.0 d 56.0 ± 3.3 c 55.3 ± 1.4 cd
pH 3.54 ± 0.01 b 3.46 ± 0.05 cd 3.43 ± 0.01 ce 3.47 ± 0.03 c 3.64 ± 0.04 a 3.41 ± 0.01 de 3.39 ± 0.02 e 3.34 ± 0.01 f
Sucrose (g l 1) 1.0 ± 0.1 bc 1.1 ± 0.2 ab 1.1 ± 0.1 ab 1.2 ± 0.1 a 1.1 ± 0.1 ab 0.9 ± 0.1 cd 0.8 ± 0.1 d 0.5 ± 0.1 e
Glucose (g l 1) 0.2 ± 0.1 bc 0.3 ± 0.3 b 0.1 ± 0.1 bc 0.3 ± 0.3 bc 0.0 ± 0.1 bc 1.0 ± 0.4 a 0.1 ± 0.1 bc 0.0 ± 0.1 c
Fructose (g l 1) 13.0 ± 0.1 a 7.8 ± 4.0 bc 5.6 ± 2.5 cd 7.3 ± 3.1 c 2.6 ± 4.1 de 12.2 ± 1.5 ab 3.7 ± 2.3 ce 0.0 ± 0.1 e
Total residual carbohydrates (g l 1) 14.1 ± 0.2 a 9.3 ± 4.2 b 6.7 ± 2.5 bc 8.7 ± 3.3 b 3.8 ± 4.1 cd 14.1 ± 1.9 a 4.6 ± 2.4 bd 0.6 ± 0.1 d
Ethanol (g l 1) 7.65 ± 0.08 c 15.23 ± 1.76 b 15.78 ± 1.31 b 20.60 ± 1.87 a 15.51 ± 0.55 b 16.09 ± 1.27 b 14.08 ± 0.81 b 16.01 ± 0.04 b
Lactic acid (g l 1) 1.32 ± 0.04 d 2.67 ± 0.14 bc 2.50 ± 0.09 c 3.47 ± 0.19 a 2.89 ± 0.10 b 2.45 ± 0.19 c 2.67 ± 0.35 bc 2.44 ± 0.12 c
Acetic acid (g l 1) 0.58 ± 0.07 e 1.18 ± 0.31 ac 1.46 ± 0.34 a 1.41 ± 0.24 a 1.26 ± 0.21 ab 0.98 ± 0.19 bcd 0.75 ± 0.21 de 0.79 ± 0.18 ce
Glycerol (g l 1) 1.12 ± 0.06 f 2.07 ± 0.17 bc 1.94 ± 0.06 c 2.73 ± 0.15 a 2.06 ± 0.07 bc 2.14 ± 0.15 b 1.71 ± 0.11 d 1.50 ± 0.01 e
Mannitol (g l 1) 0.75 ± 0.03 de 0.95 ± 0.06 bc 0.90 ± 0.06 c 1.06 ± 0.01 ab 0.83 ± 0.05 cd 1.19 ± 0.06 a 0.67 ± 0.18 e 0.45 ± 0.01 f
2-Methyl-1-propanol (mg l 1) 4.59 ± 0.30 d 9.23 ± 1.53 bc 11.05 ± 1.92 b 13.28 ± 1.75 a 9.03 ± 1.15 bc 9.89 ± 1.23 bc 8.69 ± 0.41 c 14.91 ± 0.46 a
Isoamyl alcohol (mg l 1) 31.97 ± 2.73 e 50.65 ± 2.86 c 55.65 ± 3.62 bc 64.48 ± 5.71 a 54.79 ± 5.62 bc 50.52 ± 3.67 c 42.24 ± 2.26 d 57.61 ± 1.43 b
2-Phenylethanol (mg l 1) 8.03 ± 1.87 8.33 ± 3.80 8.91 ± 1.49 8.83 ± 4.01 7.70 ± 1.28 7.32 ± 1.87 11.27 ± 0.30 12.67 ± 1.66
Ethyl acetate (mg l 1) 6.86 ± 0.83 e 13.77 ± 1.49 bc 17.74 ± 1.22 a 16.87 ± 1.05 a 14.65 ± 0.59 b 14.23 ± 1.00 bc 6.74 ± 0.87 d 13.76 ± 0.19 cd
Isoamyl acetate (mg l 1) 0.019 ± 0.002 e 0.075 ± 0.026 cd 0.064 ± 0.016 cd 0.120 ± 0.028 b 0.094 ± 0.029 bc 0.053 ± 0.008 d 0.056 ± 0.007 d 0.204 ± 0.010 a
Ethyl hexanoate (mg l 1) 0.036 ± 0.029 f 0.085 ± 0.024 bcd 0.058 ± 0.017 def 0.126 ± 0.019 a 0.097 ± 0.027 ac 0.076 ± 0.010 ce 0.050 ± 0.004 ef 0.113 ± 0.013 ab
Ethyl octanoate (mg l 1) 0.230 ± 0.062 d 0.540 ± 0.054 c 0.568 ± 0.075 c 0.823 ± 0.184 b 0.651 ± 0.189 bc 0.765 ± 0.060 b 0.537 ± 0.060 c 1.173 ± 0.049 a
Ethyl decanoate (mg l 1) 0.025 ± 0.016 c 0.344 ± 0.231 bc 0.233 ± 0.141 bc 0.668 ± 0.197 b 0.537 ± 0.245 bc 0.570 ± 0.048 b 0.525 ± 0.288 bc 3.456 ± 0.714 a
Glycerol/ethanol (mmol/mol) 110 ± 4 a 103 ± 13 ab 93 ± 6 b 100 ± 12 ab 100 ± 7 ab 100 ± 5 ab 91 ± 1 b 70 ± 1 c
Lactic acid/ethanol (mmol/mol) 133 ± 4 abc 135 ± 13 abc 122 ± 7 cd 129 ± 5 bd 143 ± 11 ab 117 ± 6 d 145 ± 11 a 117 ± 6 d
Acetic acid/ethanol (mmol/mol) 58 ± 7 ad 60 ± 18 abc 71 ± 17 a 52 ± 5 ad 63 ± 14 ab 47 ± 12 bd 41 ± 10 cd 38 ± 9 d
Acetic acid/lactic acid (mmol/mol) 439 ± 41 b 439 ± 95 b 580 ± 117 a 405 ± 50 bc 436 ± 69 b 401 ± 83 bc 279 ± 49 c 322 ± 59 bc
D-lactic acid (% of total) 45.0 ± 1.1 44.6 ± 0.5 45.1 ± 0.4 45.2 ± 0.6 45.6 ± 1.1 44.1 ± 0.6 46.4 ± 0.9 44.8 ± 0.7
Carbon recovery (%) 97.4 ± 1.0 c 99.1 ± 0.8 bc 98.5 ± 1.3 bc 97.2 ± 1.8 c 98.1 ± 0.2 bc 100.1 ± 1.6 b 90.4 ± 1.2 d 103.2 ± 0.9 a
 D. Laureys et al. / Food Microbiology 73 (2018) 351e361 355

6.0 pH 80 Water kefir grain growth (%)

60
5.0

40
4.0
20

3.0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
3.8 80

60
3.6

40

3.4
20

3.2 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Backslopping step Backslopping step
Fig. 1. The pH and water kefir grain growth for eight water kefir fermentation series differing in the presence of oxygen, nutrient concentration, and nutrient source (at the end of
each backslopping step): anaerobic [1DF-An ( )] and aerobic [1DF-Ae ( )] fermentation series with dried figs (top); anaerobic fermentation series with low [0DF-An
(◊ee)], normal [1DF-An ( )], or high [2DF-An (Aee)] nutrient concentration (top); and anaerobic fermentation series with dried figs [1DF-An ( )], dried apricots
[1DA-An ( )], raisins [1DR-An ( )], fresh figs [1FF-An ( )], and a mixture of yeast extract and peptone [YP-An (Xee)] (bottom). For differences of sig-
nificance, see Tables S1 and S2.

Table 2
Characteristics of eight water kefir fermentation series differing in the presence of oxygen, nutrient concentration, and nutrient source [anaerobic control fermentation with
dried figs (1DF-An); aerobic fermentation with dried figs (1DF-Ae); anaerobic fermentation with low (0DF-An) and high (2DF-An) amounts of dried figs; and anaerobic
fermentation with dried apricots (1DA-An), dried raisins (1DR-An), fresh figs (1FF-An), and a mixture of yeast extract and peptone (YP-An)] at the end of backslopping step 8.
Significant differences between the series are indicated with different superscripts (a, b, c, d, e, and f).

Characteristic 0DF-An 1DF-An 1DF-Ae 2DF-An 1DA-An 1DR-An 1FF-An YP-An

1 f de de bc e cd b
Yeasts (log cfu g ) 6.8 ± 0.1 7.5 ± 0.1 7.4 ± 0.1 7.7 ± 0.1 7.3 ± 0.1 7.6 ± 0.1 7.8 ± 0.1 8.2 ± 0.2 a
Lactic acid bacteria (log cfu g 1) 8.2 ± 0.1 bc 8.6 ± 0.1 a 8.6 ± 0.1 a 8.5 ± 0.1 a 8.0 ± 0.2 cd 7.9 ± 0.1 de 8.2 ± 0.1 b 7.8 ± 0.1 e
Acetic acid bacteria (log cfu g 1) 4.8 ± 0.3 b 3.3 ± 0.7 ef 6.3 ± 0.4 a 3.6 ± 0.1 de 4.5 ± 0.2 bc 4.0 ± 0.2 cd 2.8 ± 0.2 fg 2.3 ± 0.3 g
Lactic acid bacteria/yeasts (cfu/cfu) 25.9 ± 1.2 a 12.6 ± 1.2 b 13.4 ± 1.2 b 6.3 ± 1.3 c 5.0 ± 1.7 cd 1.9 ± 1.4 de 2.9 ± 1.3 ce 0.3 ± 1.5 e
Water kefir grain growth (%) 23.0 ± 1.4 e 47.4 ± 5.7 bc 37.9 ± 2.3 d 52.0 ± 3.3 b 62.5 ± 2.9 a 43.7 ± 4.1 c 20.3 ± 1.6 e 8.3 ± 1.2 f
Water kefir grain dry mass (%) 17.1 ± 0.2 a 14.1 ± 0.1 bd 14.1 ± 0.3 cd 13.9 ± 0.2 cd 14.3 ± 0.2 bc 14.5 ± 0.3 b 14.1 ± 0.2 cd 13.8 ± 0.2 d
pH 5.64 ± 0.09 a 3.46 ± 0.02 cd 3.33 ± 0.01 e 3.51 ± 0.02 c 3.71 ± 0.0 b 3.44 ± 0.02 d 3.40 ± 0.03 d 3.42 ± 0.01 d
Sucrose (g l 1) 27.1 ± 0.6 a 0.8 ± 0.2 b 0.6 ± 0.2 b 0.6 ± 0.1 b 1.0 ± 0.1 b 0.8 ± 0.1 b 0.1 ± 0.1 c 0.1 ± 0.1 c
Glucose (g l 1) 3.0 ± 0.1 a 0.1 ± 0.1 c 0.1 ± 0.1 c 0.1 ± 0.1 c 0.1 ± 0.1 c 0.5 ± 0.3 b 0.0 ± 0.0 c 0.0 ± 0.0 c
Fructose (g l 1) 12.7 ± 0.6 a 2.9 ± 4.6 c 2.8 ± 2.0 c 0.1 ± 0.1 c 3.4 ± 3.3 c 7.5 ± 2.9 b 0.1 ± 0.1 c 0.0 ± 0.1 c
Total residual carbohydrates (g l 1) 42.8 ± 0.1 a 3.8 ± 5.0 c 3.5 ± 2.3 c 0.8 ± 0.2 c 4.4 ± 3.4 bc 8.8 ± 3.2 b 0.3 ± 0.2 c 0.1 ± 0.0 c
Ethanol (g l 1) 0.3 ± 0.02 e 21.3 ± 3.37 c 18.8 ± 1.67 cd 28.2 ± 0.70 a 16.2 ± 1.81 d 20.0 ± 0.64 c 24.5 ± 0.99 b 26.1 ± 0.39 ab
Lactic acid (g l 1) 0.11 ± 0.01 e 2.83 ± 0.41 c 2.26 ± 0.15 d 3.99 ± 0.18 a 2.63 ± 0.20 c 2.23 ± 0.12 d 3.35 ± 0.13 b 2.06 ± 0.01 d
Acetic acid (g l 1) 0.11 ± 0.01 d 1.19 ± 0.08 bc 7.88 ± 1.47 a 1.24 ± 0.06 bc 1.16 ± 0.10 bc 1.52 ± 0.31 b 0.79 ± 0.11 bd 0.47 ± 0.03 cd
Glycerol (g l 1) 0.11 ± 0.01 e 1.96 ± 0.06 b 1.61 ± 0.18 d 2.42 ± 0.05 a 1.89 ± 0.08 bc 2.42 ± 0.12 a 1.87 ± 0.24 bc 1.72 ± 0.05 cd
Mannitol (g l 1) 0.10 ± 0.01 c 0.42 ± 0.12 bc 0.58 ± 0.32 b 0.69 ± 0.07 b 0.47 ± 0.07 bc 2.34 ± 0.29 a 0.76 ± 0.62 b 0.00 ± 0.00 c
2-Methyl-1-propanol (mg l 1) 0.24 ± 0.02 d 13.51 ± 3.14 bc 12.66 ± 0.80 bc 17.56 ± 0.55 a 10.82 ± 2.60 c 11.60 ± 0.76 c 17.75 ± 1.61 a 15.19 ± 0.38 ab
Isoamyl alcohol (mg l 1) 1.25 ± 0.18 e 56.12 ± 4.90 c 42.95 ± 1.79 d 68.06 ± 2.63 ab 62.65 ± 5.79 bc 44.99 ± 1.55 d 62.66 ± 6.51 bc 70.71 ± 1.83 a
2-Phenylethanol (mg l 1) 0.08 ± 0.04 c 7.72 ± 1.80 ab 4.94 ± 1.19 b 7.58 ± 2.19 ab 10.24 ± 2.51 a 5.91 ± 2.49 b 6.71 ± 0.61 b 9.98 ± 0.92 a
Ethyl acetate (mg l 1) 0.02 ± 0.00 e 16.07 ± 1.36 c 19.64 ± 2.62 ab 14.88 ± 0.90 c 17.38 ± 2.92 bc 22.40 ± 1.27 a 15.99 ± 2.80 c 10.83 ± 0.44 d
Isoamyl acetate (mg l 1) 0.001 ± 0.000 f
0.170 ± 0.081 cd 0.080 ± 0.007 df 0.387 ± 0.081 b 0.114 ± 0.035 de 0.060 ± 0.005 ef 0.226 ± 0.096 c 0.573 ± 0.034 a
Ethyl hexanoate (mg l 1) 0.001 ± 0.000 e
0.163 ± 0.049 c 0.051 ± 0.009 de 0.264 ± 0.027 b 0.146 ± 0.044 c 0.092 ± 0.009 d 0.222 ± 0.035 b 0.371 ± 0.019 a
Ethyl octanoate (mg l 1) 0.005 ± 0.001 e
1.570 ± 0.650 c 1.304 ± 0.38 cd 2.252 ± 0.314 b 0.841 ± 0.221 d 0.920 ± 0.039 d 1.819 ± 0.315 bc 3.057 ± 0.102 a
Ethyl decanoate (mg l 1) 0.015 ± 0.008 e
1.345 ± 0.999 cd 1.400 ± 0.075 cd 2.429 ± 0.588 b 0.756 ± 0.464 de 0.438 ± 0.108 e 1.945 ± 0.196 bc 4.502 ± 0.682 a
Glycerol/ethanol (mmol/mol) 321 ± 23 a 71 ± 11 c 64 ± 4 cd 64 ± 2 cd 88 ± 7 b 91 ± 2 b 57 ± 9 cd 49 ± 1 d
Lactic acid/ethanol (mmol/mol) 331 ± 37 a 103 ± 7 bc 92 ± 3 c 108 ± 2 bc 126 ± 14 b 85 ± 3 c 105 ± 2 bc 60 ± 1 d
Acetic acid/ethanol (mmol/mol) 337 ± 32 a 43 ± 5 b 325 ± 78 a 34 ± 2 b 56 ± 8 b 58 ± 10 b 25 ± 3 b 14 ± 1 b
Acetic acid/lactic acid (mmol/mol) 1022 ± 30 b 422 ± 37 c 3511 ± 799 a 310 ± 16 c 441 ± 22 c 677 ± 99 bc 234 ± 26 c 227 ± 15 c
D-lactic acid (% of total) 46.4 ± 2.9 45.0 ± 1.2 44.2 ± 2.6 42.4 ± 1.2 44.3 ± 1.1 42.7 ± 0.3 43.1 ± 2.5 42.8 ± 0.6
Carbon recovery (%) 98.6 ± 0.5 d 101.8 ± 2.8 c 103.2 ± 1.0 bc 102.9 ± 0.8 bc 97.4 ± 0.7 d 104.7 ± 0.7 ab 96.7 ± 2.3 d 105.7 ± 0.2 a
356 D. Laureys et al. / Food Microbiology 73 (2018) 351e361

Fig. 2. Culture-dependent species diversity of bacteria and yeasts of the water kefir grains of the inoculum (INO) and the eight fermentation series differing in the presence of
oxygen, nutrient concentration, and nutrient source, at the end of backslopping step 8. The closest known type strains of the sequenced fragments are given. (A) Isolates from MRS
agar media: 1, Lactobacillus paracasei (100% identity; accession no. AP012541); 2, Lactobacillus hilgardii (100% identity; accession no. LC064898); 3, Lactobacillus nagelii (99% identity;
accession no. NR112754); 4, Lactobacillus harbinensis (100% identity; accession no. NR113969); and 5, Leuconostoc mesenteroides (99% identity; accession no. LC071839). (B) Isolates
from mDMS agar media: 1, Gluconobacter roseus/oxydans (100% identity for both species; accession no. NR041049/NR026118); 2, Acetobacter fabarum (100% identity; accession no.
NR113556); 3, Acetobacter indonesiensis (99% identity; accession no. NR113847); and 4, Gluconobacter japonicus/frateurii (100% identity; accession no. NR041445/NR112239). (C)
Isolates from YPD agar media: 1, Saccharomyces cerevisiae [LSU (99% identity; accession no. CP011558) and ITS (99% identity; accession no. KC515374)]; and 2, Dekkera bruxellensis
[LSU (99% identity; accession no. GU291284) and ITS (99% identity; accession no. FJ545249)]. LSU, large subunit rRNA gene; ITS, internal transcribed spacer.

did not allow their species level identification. mass was higher than those of fermentation series 1DF-An and
2DF-An (Table 2).
3.1.5. Substrate consumption and metabolite production At the end of backslopping step 1, the pH of fermentation series
At the end of backslopping step 1, the total residual carbohy- 0DF-An was already significantly higher than the pH values of
drate and metabolite concentrations were similar in the aerobic fermentation series 1DF-An and 2DF-An, and increased over the
and anaerobic fermentation series. The total residual carbohydrate course of the eight backslopping steps (Fig. 1 and Table S1). The pH
concentrations remained similar at the end of backslopping step 8, in fermentation series 2DF-An was always slightly higher than in
but the concentrations of acetic acid were higher, and those of 1DF-An.
ethanol, lactic acid, and glycerol were lower in the aerobic
fermentation series than in the anaerobic ones (Table 2). Further- 3.2.2. Microbial enumerations
more, the concentrations of ethyl acetate were higher and those of The viable counts of the yeasts and LAB on the water kefir grains
the higher esters were lower in the aerobic fermentation series were higher when the amount of dried figs added to the fermen-
than in the anaerobic ones. tation series was higher (Table 2). Further, the ratios of the LAB to
the yeasts decreased when the amount of dried figs added to the
3.2. Influence of the nutrient concentration fermentation series increased.

3.2.1. pH and water kefir grain wet and dry mass 3.2.3. Culture-dependent microbial species diversity
At the end of backslopping step 1, the water kefir grain growth The main yeast species were S. cerevisiae and D. bruxellensis,
(based on wet mass) was similar for the fermentation series 0DF- whereby the relative abundances of S. cerevisiae increased and
An, 1DF-An, and 2DF-An, and it decreased over the course of the those of D. bruxellensis decreased when the amount of dried figs
eight backslopping steps in the fermentation series 0DF-An (Fig. 1 added to the fermentation series increased (Fig. 2).
and Table S2). At the end of backslopping step 8, the water kefir The main LAB species in all three fermentation series were Lb.
grains of fermentation series 0DF-An were larger and their dry paracasei and Lb. hilgardii (82% of the strains produced EPS),
 D. Laureys et al. / Food Microbiology 73 (2018) 351e361 357

Fig. 3. Community profiles of the bacteria and yeasts on the water kefir grains and in the water kefir liquors of the inoculum (INO) and the eight fermentation series differing in the
presence of oxygen, nutrient concentration, and nutrient source, at the end of backslopping step 8. The numbers indicate the bands that were sequenced and the closest known type
strains of the sequenced fragments are given. (A) With the V3 primer pair: 1, Lactobacillus casei/paracasei/zeae/rhamnosus (99% identity for the four species; accession no. LC064894/
AB289229/AB289313/JQ580982); 2, Lactobacillus hilgardii (100% identity; accession no. LC064898); 3, Lactobacillus nagelii/ghanensis (99% identity; accession no. NR112754/
NR043896); 4, Leuconostoc mesenteroides/pseudomesenteroides (99% identity; accession no. LC071839/LC096220); 5, Bifidobacterium aquikefiri (100% identity; accession no.
LN849254), 6, Comamonas testosteroni/thiooxydans (100% identity; accession no. NR113709/NR115741); and 7, Acetobacteraceae (100% identity). (B) With the LAC primer pair: 1,
Lactobacillus casei/paracasei/zeae (99% identity; accession no. LC064894/AB289229/AB289313); 2, Lactobacillus hilgardii/diolivorans (100% identity; accession no. LC064898/
NR037004); and 3, Lactobacillus nagelii (99% identity; accession no. NR119275). (C) With the Yeast primer pair: 1, Saccharomyces cerevisiae (100% identity; accession no. NG042623);
2, Dekkera bruxellensis (100% identity; accession no. AY969049); and Candida smithsonii (99% identity; accession no. AY518525).

whereby the relative abundances of Lb. hilgardii were higher in relative intensities of the bands attributed to S. cerevisiae increased
fermentation series 0DF-An than in fermentation series 1DF-An and those of the bands attributed to D. bruxellensis decreased when
and 2DF-An (Fig. 2). Further, Lb. nagelii and Lb. harbinensis were the amount of dried figs added increased (Fig. 3). Further, a band
only isolated from fermentation series 1DF-An and 2DF-An. with weak relative intensity in the community profiles of the li-
The main AAB species in the three fermentation series were G. quors of fermentation series 1DF-An and 2DF-An was attributed to
oxydans/roseus, A. fabarum, and A. indonesiensis, whereby the C. smithsonii.
relative abundances of A. fabarum increased when the amount of The main bands in the community profiles obtained with the V3
dried figs added to the fermentation series increased (Fig. 2). Glu- and LAC primer pairs for the water kefir liquors and water kefir
conobacter japonicus/frateurii was only isolated from the fermen- grains of fermentation series 0DF-An, 1DF-An, and 2DF-An were
tation series 0DF-An. attributed to Lb. hilgardii, Lb. paracasei, and Lb. nagelii (Fig. 3). The
relative intensities of the bands attributed to Lb. nagelii increased
and those of the bands attributed to Lb. hilgardii decreased when
3.2.4. Culture-independent microbial species diversity the amount of dried figs added increased. In the community pro-
The main bands in the rRNA-PCR-DGGE community profiles files obtained with the V3 and Bif primer pairs, a band attributed to
obtained with the Yeast primer pair for the water kefir liquors and B. aquikefiri was detected in the three fermentation series, whereby
the water kefir grains of fermentation series 0DF-An, 1DF-An, and the relative intensities of the bands were highest in the community
2DF-An were attributed to S. cerevisiae and D. bruxellensis. The
358 D. Laureys et al. / Food Microbiology 73 (2018) 351e361

profiles of the fermentation series 1DF-An. Additionally, in the isolated from the fermentation series YP-An (Fig. 2).
community profiles obtained with the V3 primer pair, a band
attributed to Comamonas testosteroni/thiooxydans was detected in 3.3.4. Culture-independent microbial species diversity
the water kefir liquors and the water kefir grains of the fermenta- The main bands in the rRNA-PCR-DGGE community profiles
tion series 0DF-An, with higher relative intensities in the liquors obtained with the Yeast primer pair of the water kefir liquors and
than on the grains. This band was not detected in the water kefir the water kefir grains of all fermentation series were attributed to
liquors or on the water kefir grains of fermentation series 1DF-An S. cerevisiae and D. bruxellensis. The relative intensities of the bands
and 2DF-An. attributed to D. bruxellensis were highest in the fermentation series
1DR-An, lower in 1DA-An and 1DF-An, and lowest in 1FF-An and
3.2.5. Substrate consumption and metabolite production YP-An (Fig. 3).
The concentrations of the total residual carbohydrates at the end The main bands in the community profiles obtained with the V3
of backslopping steps 1 and 8 were higher and those of the me- and LAC primer pairs of the water kefir liquors and the water kefir
tabolites were lower when the amount of dried figs added to the grains of all fermentation series were attributed to Lb. nagelii, Lb.
fermentation series was lower (Tables 1 and 2). hilgardii, and Lb. paracasei. The relative intensities of the bands
attributed to Lb. nagelii were highest in fermentation series 1FF-An
3.3. Influence of the nutrient source and YP-An, lower in 1DR-An and 1DA-An, and lowest in 1DR-An
(Fig. 3). The relative intensities of the bands attributed to Lb. hil-
3.3.1. pH and water kefir grain wet and dry mass gardii were highest in the fermentation series 1DR-An, lower in
The water kefir grain growth (based on wet mass) was around 1DA-An and 1DF-An, and lowest in 1FF-An and YP-An. The relative
60% for all fermentation series at the end of backslopping step 1. It intensities of the bands attributed to Lb. paracasei were lower in
remained more or less stable over the course of the eight back- fermentation series 1FF-An and 1 YP-An than in the other ones. In
slopping steps for fermentation series 1DF-An, 1DA-An, and 1 DR- the community profiles obtained with the V3 primer pair, a band
An, but decreased slowly in the fermentation series 1FF-An and attributed to Leuc. mesenteroides was detected in the fermentation
fast in the fermentation series YP-An (Fig. 1). The water kefir grain series 1FF-An. In the community profiles with the V3 and Bif primer
growth was highest in fermentation series 1DA-An, followed by pairs, a band attributed to B. aquikefiri was detected in all
1DF-An, and 1DR-An (Fig. 1 and Table S2). The water kefir grain dry fermentation series, whereby the relative intensities of these bands
mass was similar for all fermentation series. The water kefir grains were lowest in fermentation series 1FF-An and YP-An.
were largest in the fermentation series 1DA-An, smaller in 1FF-An,
and smallest in YP-An. 3.3.5. Substrate consumption and metabolite production
The pH at the end of backslopping step 1 was comparable in the The total residual carbohydrate concentrations were always
fermentation series 1DF-An, 1DR-An, and 1FF-An (approximately lowest in fermentation series YP-An and 1FF-An, higher in 1DF-An
3.45), significantly higher in the fermentation series 1DA-An and 1DA-An, and highest in 1DR-An (Tables 1 and 2). The ethanol
(approximately 3.65), and significantly lower in the fermentation concentrations at the end of backslopping step 1 were approxi-
series YP-An (approximately 3.35) (Table 1). The pH values of mately 15 g l 1 for all fermentation series. At the end of back-
fermentation series 1DF-An, 1DA-An, 1DR-An, and 1FF-An slopping step 8, the ethanol concentrations remained at
remained stable over the course of the eight backslopping steps, approximately 16 g l 1 for the fermentation series 1DA-An,
whereas the pH of the series YP-An increased after backslopping increased to approximately 20 g l 1 in 1DF-An and 1DR-An, and
step 4 to become similar to the pH in fermentation series 1DF-An, increased to approximately 25 g l 1 in fermentation series 1FF-An
1DR-An, and 1FF-An (Fig. 1 and Table S1). and YP-An. The concentrations of glycerol were always highest in
the fermentation series 1DR-An. At the end of backslopping step 8,
3.3.2. Microbial enumerations the ratios of the concentrations of glycerol to ethanol were higher
The viable counts of the yeasts were highest in fermentation in fermentation series 1DA-An, 1DR-An, and 1DF-An than in
series YP-An and 1FF-An, lower in 1DF-An and 1DR-An, and lowest fermentation series 1FF-An and YP-An.
in the fermentation series 1DA-An (Table 2). The viable counts of The lactic acid concentrations were always higher in fermen-
the LAB were highest in the fermentation series 1DF-An, lower in tation series 1DF-An, 1DA-An, and 1FF-An than in fermentation
1FF-An, 1DA-An, and 1DR-An, and lowest in YP-An. The ratios of the series 1DR-An and YP-An (Tables 1 and 2). The acetic acid con-
viable counts of the LAB to the yeasts were highest in the centrations were always lower in fermentation series 1FF-An and
fermentation series 1DF-An, lower in 1DA-An, 1FF-An, and 1DR-An, YP-An than in fermentation series 1DF-An and 1DA-An, and were
and lowest in YP-An. highest in the fermentation series 1DR-An at the end of back-
slopping step 8. The ratios of the concentrations of lactic acid to
3.3.3. Culture-dependent microbial species diversity ethanol were always higher in fermentation series 1DF-An, 1DA-An,
The yeast species S. cerevisiae and D. bruxellensis were isolated and 1FF-An than in fermentation series 1DR-An and YP-An, and
from all fermentation series, whereby the relative abundances of D. those of acetic acid to ethanol and acetic acid to lactic acid were
bruxellensis were highest in the fermentation series 1DR-An, lower always higher in fermentation series 1DF-An, 1DA-An, and 1DR-An
in 1DA-An and 1DF-An, and lowest in 1FF-An and YP-An (Fig. 2). than in fermentation series 1FF-An and YP-An. At the end of
The LAB species Lb. paracasei, Lb. nagelii, and Lb. hilgardii (79% of backslopping step 8, the concentrations of the higher alcohols and
these strains produced EPS) were isolated from fermentation series higher esters were lowest in the fermentation series 1DR-An,
1DF-An, 1DA-An, 1DR-An, 1FF-An, and YP-An; Lb. harbinensis was higher in fermentation series 1DF-An and 1DA-An, and highest in
isolated from fermentation series 1DF-An, 1DR-An, and YP-An; and fermentation series 1FF-An and YP-An, whereas the concentrations
Leuc. mesenteroides was isolated from the fermentation series 1FF- of ethyl acetate were opposite.
An (Fig. 2).
The AAB species G. roseus/oxydans, A. fabarum, and A. indone- 4. Discussion
siensis were isolated from fermentation series 1DF-An, 1DA-An, and
1DR-An; G. roseus/oxydans was the only AAB species isolated from The present study showed that the presence of oxygen, the
the fermentation series 1FF-An; and A. fabarum was the only one nutrient concentration, and the nutrient source influenced the
 D. Laureys et al. / Food Microbiology 73 (2018) 351e361 359

water kefir grain growth, microbial species diversity, substrate production without a decrease of the total residual carbohydrate
consumption, and metabolite production during water kefir concentrations or pH.
fermentation. The water kefir grain growth was initially not affected by the
The most characteristic effect of the presence of oxygen during nutrient concentrations, but insufficient nutrient concentrations
water kefir fermentation was the proliferation of AAB. These obli- resulted in a slow and gradual decrease of the water kefir grain
gately aerobic microorganisms are often present in water kefir growth upon backslopping. This was caused by a lack of nutrients,
(Laureys and De Vuyst, 2014). However, their viable counts can vary as high pH values at low nutrient concentrations excluded its
widely (Franzetti et al., 1998; Gulitz et al., 2011; Laureys and De decrease due to acidic stress. Nutrient concentrations in excess of a
Vuyst, 2014), but they are usually low during water kefir fermen- certain threshold value did not further increase the water kefir
tation because oxygen is only periodically available at the start of grain growth.
each backslopping step, whereas ethanol (an important energy Low nutrient concentrations resulted in high viable counts of
source for AAB) is only available at the end of a water kefir AAB species, which were probably caused by the limited expulsion
fermentation process. Yet, AAB are known to survive low-oxygen of oxygen due to the low metabolic activity of the microorganisms
conditions, even for long periods of time (Bartowsky and in such fermentation processes. High nutrient concentrations
Henschke, 2008; Moens et al., 2014). Alternatively, during the favored the growth of yeasts at the expense of the LAB species, and
current study, distinction between aerobic and anaerobic condi- this was reflected in the ratios of the metabolite concentrations of
tions was achieved by exposing the fermentation mixtures to air the yeasts to those of the LAB. The relative abundances of Lb. nagelii
(aerobic conditions) or protecting them toward air ingress by and S. cerevisiae were high at high nutrient concentrations, whereas
means of a water lock (anaerobic conditions). As the concentrations those of Lb. hilgardii and D. bruxellensis were high at low nutrient
of oxygen were not monitored during the fermentation processes concentrations. This is in line with the low nutrient requirements of
carried out, complete aerobic or anaerobic conditions thus could D. bruxellensis (Uscanga et al., 2000). Furthermore, high nutrient
not be guaranteed. The main AAB found in the present study were concentrations resulted in low ratios of the concentrations of acetic
A. fabarum, G. roseus/oxydans, and A. indonesiensis. The former two acid to ethanol and acetic acid to lactic acid, which may be related
AAB species were reported for water kefir before (Gulitz et al., 2011; to the shift in microbial species diversity. Indeed, Lb. hilgardii
Laureys and De Vuyst, 2014). To our knowledge, this is the first time (obligately heterofermentative) produces more acetate than Lb.
that A. indonesiensis was isolated from water kefir. Further, G. nagelii (obligately homofermentative), and D. bruxellensis produces
roseus/oxydans and A. indonesiensis were more abundant under more acetate than S. cerevisiae (Ludwig et al., 2009; Oelofse et al.,
anaerobic fermentation conditions, whereas A. fabarum was more 2008). Finally, B. aquikefiri, which produces acetic acid upon
abundant under aerobic fermentation conditions. Also, the AAB fermentation of hexoses (Laureys et al., 2016), thrived best under
species were more abundant in the water kefir liquors than on the moderate nutrient concentrations.
water kefir grains, indicating that the liquor was their preferred Low nutrient concentrations allowed the growth of C. testos-
niche. teroni/thiooxydans during water kefir fermentation. This bacterium
The proliferation of AAB in the aerobic fermentation series is widely present in soil and water and on plants, but has not yet
resulted in high acetic acid concentrations and thus low pH values. been reported for water kefir (Bayhan et al., 2013). It is a motile,
This probably caused a slow but gradual decrease of the water kefir obligately aerobic b-proteobacterium that grows at pH 6.0e8.5
grain growth upon backslopping during the aerobic fermentation (Bayhan et al., 2013; Narayan et al., 2010). Its growth during water
processes, as excessive acidic stress decreased the water kefir grain kefir fermentation with low nutrient concentrations was likely
growth during fermentation (Laureys and De Vuyst, 2017). The caused by the high pH values and prolonged presence of oxygen.
lower concentrations of ethanol and lactic acid in the aerobic fer- Further, this bacterium preferred the water kefir liquors above the
mentations probably resulted from their consumption by the AAB water kefir grains, reflecting its obligately aerobic and mobile na-
species (Moens et al., 2014). Further, there were no indications that ture. Under normal water kefir fermentation conditions, C. testos-
S. cerevisiae and/or D. bruxellensis switched to respirational meta- teroni/thiooxydans is not expected, as water kefir fermentation
bolism in the presence of oxygen (Schifferdecker et al., 2014). The normally proceeds under anaerobic conditions whereby the pH
proliferation of AAB species in the aerobic water kefir fermenta- decreases fast below 4.0 (Laureys and De Vuyst, 2014).
tions coincided with higher concentrations of ethyl acetate and Dried figs are the most commonly used source of nutrients
lower concentrations of fruity esters than in the anaerobic ones. during water kefir fermentation. Yet, stable water kefir fermenta-
Similarly, the proliferation of AAB species in wine results in higher tion was also possible with dried apricots and dried raisins, but not
concentrations of ethyl acetate and lower overall fruitiness with fresh figs or a mixture of yeast extract and peptone (YP so-
(Bartowsky and Henschke, 2008; Bartowsky et al., 2003). Further- lution), as this resulted in a gradually decreasing water kefir grain
more, the high concentrations of acetic acid may have caused lower growth. The high and low water kefir grain growth in the
relative abundances of B. aquikefiri in the aerobic fermentation fermentation series with dried apricots and raisins, respectively,
series, as this bifidobacterial species is not inhibited by aerobic was probably caused by the high pH values when dried apricots
conditions (Laureys et al., 2016) or low pH values during water kefir were added and the low pH values when raisins were added. Low
fermentation (Laureys and De Vuyst, 2017). Indeed, high concen- pH values in the fermentation series with YP solution probably
trations of acetic acid may inhibit the growth of certain microor- caused a fast decrease of the water kefir grain growth during the
ganisms; for instance, D. bruxellensis is sensitive to acetic acid first backslopping steps. However, the low pH values in these
concentrations higher than 1 g l 1 (Yahara et al., 2007). fermentation series were not caused by high acid concentrations,
Low nutrient concentrations caused a slow fermentation, underlining that the nutrient source influenced the pH during
resulting in high total residual carbohydrate concentrations, low water kefir fermentation via the release of buffer compounds, as
metabolite concentrations, and high pH values. In contrast, high mentioned above. After the fast initial decrease, the water kefir
nutrient concentrations caused a fast fermentation, resulting in grain growth in the fermentation series with YP solution remained
high metabolite concentrations without a decrease of the total re- low, despite the presence of EPS-producing Lb. hilgardii strains. This
sidual carbohydrate concentrations or pH values. The latter showed showed that an excessive acidic stress caused low water kefir grain
that dried figs supplied both carbohydrates and buffer compounds growth and that the presence of EPS-producing Lb. hilgardii strains
to the water kefir fermentation mixtures, allowing high metabolite was not sufficient for water kefir grain growth (Laureys and De
360 D. Laureys et al. / Food Microbiology 73 (2018) 351e361

Vuyst, 2017). Further, high relative abundances of Lb. hilgardii in the Appendix A. Supplementary data
fermentation series with dried raisins or low nutrient concentra-
tions were not reflected in a high water kefir grain growth, con- Supplementary data related to this article can be found at
firming that the relative abundance of Lb. hilgardii during water https://doi.org/10.1016/j.fm.2018.02.007.
kefir fermentation did not determine water kefir grain growth
(Laureys and De Vuyst, 2017). Low water kefir grain growth resulted References
in small water kefir grains with high viable counts of microorgan-
isms, resulting in a fast fermentation with low total residual car- Aceituno, F.F., Orellana, M., Torres, J., Mendoza, S., Slater, A.W., Melo, F., Agosin, E.,
2012. Oxygen response of the wine yeast Saccharomyces cerevisiae EC1118
bohydrate and high metabolite concentrations, as has been shown grown under carbon-sufficient, nitrogen-limited enological conditions. Appl.
previously (Laureys and De Vuyst, 2017). Finally, it has been shown Environ. Microbiol. 78, 8340e8352.
that water kefir microorganisms are able to ferment various vege- Bartowsky, E.J., Henschke, P.A., 2008. Acetic acid bacteria spoilage of bottled red
wine: a review. Int. J. Food Microbiol. 125, 60e70.
table and fruit juices, which has been explored to develop non-
Bartowsky, E.J., Xia, D., Gibson, R.L., Fleet, G.H., Henschke, P.A., 2003. Spoilage of
dairy, kefir-like beverages (Corona et al., 2016; Randazzo et al., bottled red wine by acetic acid bacteria. Lett. Appl. Microbiol. 36, 307e314.
2016). Bayhan, G.I., Tanir, G., Karaman, I., Ozkan, S., 2013. Comamonas testosteroni: an
The nutrient source had an immediate impact on the substrate unusual bacteria associated with acute appendicitis. Balkan Med. J. 30,
447e448.
consumption and metabolite production during the water kefir Blomqvist, J., Eberhard, T., Schnurer, J., Passoth, V., 2010. Fermentation character-
fermentation processes, and this impact became even more pro- istics of Dekkera bruxellensis strains. Appl. Microbiol. Biotechnol. 87, 1487e1497.
nounced at the end of these processes, probably due to the shift in Corona, O., Randazzo, W., Miceli, A., Guarcello, R., Francesca, N., Erten, H.,
Moschetti, G., Settanni, L., 2016. Characterization of kefir-like beverages pro-
the microbial communities. Indeed, high relative abundances of the duced from vegetable juices. Food Sci. Technol. (N. Y.) 66, 572e581.
obligately heterofermentative Lb. hilgardii coincided with high ac- de Winter, J.C.F., 2013. Using the Student's t-test with extremely small sample sizes.
etate concentrations and high ratios of the concentrations of ace- Practical Assess. Res. Eval. 18, 1e12.
Franzetti, L., Galli, A., Pagani, M.A., De Noni, I., 1998. Microbiological and chemical
tate to ethanol and acetate to lactic acid (Ludwig et al., 2009). The investigations on "sugar kefir" drink. Ann. Microbiol. Enzimol. 48, 67e80.
concentrations of glycerol did not parallel those of ethanol, and the Guillamo
n, J.M., Mas, A., 2009. Acetic acid bacteria. In: Ko €nig, H., Unden, G.,
ratios of the concentrations of glycerol to ethanol were higher Fro€hlich, J. (Eds.), Biology of Microorganisms on Grapes, in Must and in Wine.
Springer, Heidelberg, Germany, pp. 31e46.
when D. bruxellensis was more abundant and S. cerevisiae less. This Gulitz, A., Stadie, J., Ehrmann, M.A., Ludwig, W., Vogel, R.F., 2013. Comparative
was in contrast with literature data, which indicate that S. cerevisiae phylobiomic analysis of the bacterial community of water kefir by 16S rRNA
produces more glycerol than D. bruxellensis (Blomqvist et al., 2010). gene amplicon sequencing and ARDRA analysis. J. Appl. Microbiol. 114,
1082e1091.
The concentrations of ethyl acetate were highest when raisins were
Gulitz, A., Stadie, J., Wenning, M., Ehrmann, M.A., Vogel, R.F., 2011. The microbial
added to the water kefir fermentation process and coincided with diversity of water kefir. Int. J. Food Microbiol. 151, 284e288.
high relative abundances of Lb. hilgardii. Laureys, D., Cnockaert, M., De Vuyst, L., Vandamme, P., 2016. Bifidobacterium aqui-
The fermentations with dried raisins resulted in high relative kefiri sp. nov. isolated from water kefir. Int. J. Syst. Evol. Microbiol. 66,
1281e1286.
abundances of Lb. hilgardii and D. bruxellensis, low relative abun- Laureys, D., De Vuyst, L., 2014. Microbial species diversity, community dynamics,
dances of Lb. nagelii and S. cerevisiae, high total residual carbohy- and metabolite kinetics of water kefir fermentation. Appl. Environ. Microbiol.
drate concentrations, and low metabolite concentrations, thus 80, 2564e2572.
Laureys, D., De Vuyst, L., 2017. The water kefir grain inoculum determines the
resembling the fermentations with low nutrient concentrations characteristics of the resulting water kefir fermentation process. J. Appl.
described above. The fermentations with fresh figs or with a YP Microbiol. 122, 719e732.
solution resulted in high relative abundances of Lb. nagelii and S. Laureys, D., Van Jean, A., Dumont, J., De Vuyst, L., 2017. Investigation of the insta-
bility and low water kefir grain growth during an industrial water kefir
cerevisiae, low relative abundances of Lb. hilgardii and fermentation process. Appl. Microbiol. Biotechnol. 101, 2811e2819.
D. bruxellensis, low total residual carbohydrate concentrations, and Ludwig, W., Schleifer, K.H., Whitman, W.B., 2009. Lactobacillales. In: De Vos, P.,
high metabolite concentrations, thus resembling the fermentations Garrity, G., Jones, D., Krieg, N.R., Ludwig, W., Rainey, F.A., Schleifer, K.H.,
Whitman, W.B. (Eds.), Bergey's Manual of Systematic Bacteriology: Volume
with high nutrient concentrations described above. High relative
Three: the Firmicutes. Springer, NY, USA, pp. 464e735.
abundances of D. bruxellensis resulted in high concentrations of Marsh, A.J., O'Sullivan, O., Hill, C., Ross, R.P., Cotter, P.D., 2013. Sequence-based
ethyl acetate, whereas high relative abundances of S. cerevisiae analysis of the microbial composition of water kefir from multiple sources.
FEMS Microbiol. Lett. 348, 79e85.
resulted in high concentrations of higher alcohols and higher
rrez-Ambrocio, S., Heredia-del-Orbe, P., Villa-Tanaca, L.,
Martínez-Torres, A., Gutie
esters. Hern
andez-Rodríguez, C., 2017. Inferring the role of microorganisms in water
In conclusion, the presence of oxygen allowed the proliferation kefir fermentations. Int. J. Food Sci. Technol. 52, 559e571.
of AAB species during water kefir fermentation, resulting in high Moens, F., Lefeber, T., De Vuyst, L., 2014. Oxidation of metabolites highlights the
microbial interactions and role of Acetobacter pasteurianus during cocoa bean
acetic acid concentrations, and decreased the relative abundances fermentation. Appl. Environ. Microbiol. 80, 1848e1857.
of B. aquikefiri. The nutrient concentrations had an immediate Narayan, K.D., Pandey, S.K., Das, S.K., 2010. Characterization of Comamonas thioox-
impact on the metabolism of the water kefir microorganisms and idans sp. nov., and comparison of thiosulfate oxidation with Comamonas tes-
tosteroni and Comamonas composti. Curr. Microbiol. 61, 248e253.
influenced the microbial species diversity upon backslopping, Oelofse, A., Pretorius, I.S., du Toit, M., 2008. Significance of Brettanomyces and
which in turn influenced the substrate consumption and metabo- Dekkera during winemaking: a synoptic review. S. Afr. J. Enol. Vitic. 29,
lite production. The influence of the nutrient source was similar to 128e144.
Pidoux, M., 1989. The microbial flora of sugary kefir grain (the gingerbeer plant):
that of the nutrient concentration, indicating that different nutrient biosynthesis of the grain from Lactobacillus hilgardii producing a polysaccharide
sources supplied different (amounts of) nutrients to the water kefir gel. J. Appl. Microbiol. 5, 223e238.
fermentation mixtures. Randazzo, W., Corona, O., Guarcello, R., Francesca, N., German a, M.A., Erten, H.,
Moschetti, G., Settanni, L., 2016. Development of new non-dairy beverages from
Mediterranean fruit juices fermented with water kefir microorganisms. Food
Microbiol. 54, 40e51.
Reiß, J., 1990. Metabolic activity of tibi grains. Z. Lebensm. Unters. Forschung 191,
462e465.
Acknowledgements
Schifferdecker, A.J., Dashko, S., Ishchuk, O.P., Piskur, J., 2014. The wine and beer yeast
Dekkera bruxellensis. Yeast 31, 323e332.
The authors acknowledge their financial support of the Research Spitaels, F., Wieme, A.D., Janssens, M., Aerts, M., Daniel, H.M., Van Landschoot, A., De
Council of the Vrije Universiteit Brussel (SRP7, IRP2, and IOF342 Vuyst, L., Vandamme, P., 2014. The microbial diversity of traditional sponta-
neously fermented lambic beer. PLoS One 9, e95384.
projects) and the Hercules Foundation (grant UABR09004). DL was Stadie, J., Gulitz, A., Ehrmann, M.A., Vogel, R.F., 2013. Metabolic activity and sym-
the recipient of a PhD fellowship of the Vrije Universiteit Brussel. biotic interactions of lactic acid bacteria and yeasts isolated from water kefir.
 D. Laureys et al. / Food Microbiology 73 (2018) 351e361 361

Food Microbiol. 35, 92e98. 1.828 involved in granule formation of water kefir. Food Microbiol. 27, 672e678.
Uscanga, M.G.A., Delia, M.L., Strehaiano, P., 2000. Nutritional requirements of Yahara, G.A., Javier, M.A., Tulio, M.J.M., Javier, G.R., Guadalupe, A.U.M., 2007.
Brettanomyces bruxellensis: growth and physiology in batch and chemostat Modeling of yeast Brettanomyces bruxellensis growth at different acetic acid
cultures. Can. J. Microbiol. 46, 1046e1050. concentrations under aerobic and anaerobic conditions. Bioproc. Biosyst. Eng.
Waldherr, F.W., Doll, V.M., Meissner, D., Vogel, R.F., 2010. Identification and char- 30, 389e395.
acterization of a glucan-producing enzyme from Lactobacillus hilgardii TMW