The Kombucha Biofilm: a Model System for Microbial Ecology

Ashley Shade
Gordon and Betty Moore Foundation Fellow of the Life Sciences Research Foundation
Yale University

27 July 2011
Final report on research conducted during the Microbial Diversity course, Marine
Biological Laboratories, Woods Hole, MA

Abstract

Microbial systems are inherently complex and difficult to predict. A strong model
system of naturally-occurring microbial consortia would provide opportunity to address
key questions in microbial ecology. Kombucha is a traditional, fermented tea produced
by a biofilm of yeast and acetic acid producing bacteria. Because it has well-defined
functional products and is maintained easily at the laboratory-scale, kombucha has
potential as a mixed-microbial model system. I provide a first exploration of the
kombucha system for microbial ecology. I first describe the kombucha microbial
consortia using isolation and 16S 454 tag-pyrosequencing. I assessed the temporal
dynamics of kombucha by tracking microbial abundance and key products from
inoculation to culture maturation. Then, I observed the spatial structure and association
of abundant organisms with each other in the cellulose matrix of the biofilm using
confocal microscopy. Finally, I designed an experiment to assess the influence of pH
and tea origin on the consortia and their products. I found that kombucha habours a
dominant Glucoacetobacter population, and uncovered evidence for more resolved
population-level diversity within the biofilm. Cell counts of the kombucha broth peaked
at pH 3 on day 3 after innoculation, indicating that the acid concentration of the peak
may provide a strong environmental filter for the microbes. The biofilm was
heterogeneous in space, containing patches of yeast within a matrix of cellulose and
segregated bacterial cells. Finally, tea origin and pH partially determine the rate of
product evolution in kombucha broth, and tea may inhibit the growth of the consortia.
This work lays a foundation for using kombucha as a model system for microbial
ecology.

! "!
Introduction

Microbial systems are inherently complex. A key challenge is to understand the

biotic and abiotic interactions of microbial systems to the endpoint of prediction (Prosser
et al., 2007). However, many microbial systems found in nature are difficult to
manipulate experimentally and have under-described diversity as well as many
unknown or intractable functions (Jessup et al., 2004). On the other hand, many model
laboratory systems, in microbiology and other fields, are arguably over-simplified,
maintained in controlled conditions generally unrealistic to naturally occurring systems
(e.g. (Carpenter and Url, 2011)). Though both environmental and lab-scale systems are
important for understanding different aspects of microbial complexity, a relatively
simple, authentic model microbial system is desirable for lab scale manipulations. Such
a tractable model would advance theory and achieve progress towards prediction in
microbial ecology.
Kombucha is a beverage of fermented tea consumed traditionally in eastern
Europe and Asia. To prepare kombucha, black tea leaves are seeped in boiling water,
copious amounts of sugar (sucrose) is added, and the mixture is cooled to room
temperature. Then a “mother” biofilm is placed into the tea; this biofilm is sometimes
called a symbiotic culture of bacteria and yeasts (SCOBY), and is comprised of acetic
acid producing bacteria, ethanol fermenting yeasts, and a thick cellulose pellicle. This
starter biofilm, or unpasteurized liquid from an active kombucha culture, is necessary to
begin kombucha production. The biofilm floats at the liquid-air interface and grows
vertically, increasing biomass with cellulose striations as the fermentation matures. After
several days incubation at room temperature, the tea becomes a sweet and sour,
naturally carbonated beverage because of microbial activities. Kombucha has been
both sanctioned and advocated as a health beverage, with no consensus (Dufresne,
2000). However, kombucha consumption is gaining popularity worldwide, and there is
much interest in scaling up the fermentation process to meet food industry demands
(e.g. (Chen and Liu, 2000; Malbasa et al., 2006; Cvetkovic et al., 2008; Jayabalan et al.,
2008)).
The general products of kombucha fermentation are sugars and organic acids.
The yeasts convert sucrose to glucose and fructose during fermentation, as a byproduct
of ethanol fermentation(Blanc, 1996). The acetic acid producing bacteria then convert
fructose into acetic acid (which provides kombucha with its sour flavor) and glucose to
gluconic acid. Accordingly, the pH of the enrichment drops to approximately 2.6 after,
signifying the maturation of the beverage for consumption. If the fermentation is not
stopped or slowed, the gluconic and acetic acid concentrations will continue to increase
to levels of (4 g/100 mL, (Chen and Liu, 2000)), but the beverage will be intolerable for
consumption because of a strong vinegar flavor. An additional byproduct of acetic acid
production by bacteria is cellulose (Iguchi et al., 2000), which provides scaffold to the
biofilm-associated microbes.
Because the kombucha biofilm is a mixed eukaryotic-bacterial microbial
community and has clear functions and products, it may serve as an excellent study
system for microbial ecology. There has been identification of some of the yeast and
bacterial species found in various SCOBYs. It was found that the species composition
of kombuchas vary by biofilm origin, especially with respect to the associated yeast

! #!
species (Teoh et al., 2004). However, the “vertically” transferred nature of biolfilms
among kombucha producers raises interesting questions about biogeography, selection
and evolution. Additionally, many specific interactions between the yeast and bacteria
are unknown, including the mechanisms of biofilm formation from pelagic cells, the
nature of the symbiosis between the eukaryotic and prokaryotic components (e.g. do
the acetic acid bacteria have an obligate relationship the yeast?), or the cellular and
cellulosic architecture of the biofilm. Genetics of the system (with the exception of
species identifications using common molecular markers, e.g. (Kurtzman et al., 2001;
Dutta and Gachhui, 2007)) is also a dearth. Indeed, much of the basic microbial ecology
of the kombucha system is undocumented. Yet ease of lab-scale culturing and
manipulation (the biofilm grows well and quickly at room temperature with minimal
specialized equipment), kombucha has potential as a naturally occurring, simple, and
authentic model for microbial communities and their interactions.
Here, I provide a first exploration of the utility of kombucha microbial community
as a model system. Using a combination of classical microbiology with molecular tools, I
describe temporal and spatial dynamics of kombucha, from inoculation to culture
maturation. I also conducted an experiment to understand the influence of tea origin and
pH on the microbial products of kombucha.

Materials and Methods

i. Enrichment and isolation conditions

The starter culture of kombucha had been maintained for approximately one year at
home. Teas were from Upton Tea Imports (http://www.uptontea.com). To prepare tea
media, milliQ water was boiled, and 5 g tea per liter was seeped for 10-15 minutes. Tea
leaves were removed from liquid before autoclaving. Afterwards, sucrose was added to
a final w/v concentration of 15%. Oolong (brown) tea (10 mL final volume in capped
culture tubes) was used for the time series observations. Oolong, Ceylon black tea, and
green tea were used for the tea comparison experiment (90 mL final volume in sample
jars covered but not sealed with caps), as well as freshwater minimal medium and
oolong tea amended with MOPS pH 6.2-7 buffer. The freshwater medium broth
contained vitamins while the plates did not. Calcium carbonate plates to identify acid
producers by halo formation was prepared according to (Chen and Liu, 2000): 30 g
glucose, 5 g yeast extract, 3 g peptone, 10 g calcium carbonate, and 30 mL of 95%
ethanol, added after autoclaving, 20 g agar, and sterile milliQ water to 1 L. Tea agar
plates were prepared as above, but 15 g of agar was added prior to autoclaving. Media
was inoculated with 10% v/v of kombucha broth from a previous culture. Liquid cultures
were incubated at room temperature; plates were incubated at 30 C.

ii. pH and organic acids measurements

Acid (pH) was measured daily and the probe was calibrated at each use.
Samples for organic acid analysis on HPLC were prepared from 1 mL of broth.
Cells were pelleted and supernatant filtered through a 0.2 micron Acrodisc syringe filter
(Pall), and then acidified with 100 uL 5 M sulfuric acid. Organic acids analysis was
performed on a Shimadzu LC-2010C System with an RID detector and an Aminex HPX-

! $!
87H column (BioRad). Pump flow was 0.6 mL/min at minimum pressure 0 psi and
maximum 1422 psi, autosampling 20 uL, oven temperature at 60 C, and a 30 minute
program duration. Calibrations for ethanol ranged from 1.0 to 50 mM, acetic acid ranged
from 1.0 to 100 mM, glucose from 1.0 to 500 mM, fructose from 1.0 to 300 mM, and
lactose from to 30 mM. The method was called “Alexa Organic acids- Ashley modified.”

iii. Molecular tools

DNA was isolated from two independent kombucha biofilms using the UltraClean
Soil DNA Isolation kit (MoBio), as per manufacturer’s instructions with minor
modifications. The biofilms were from 10 mL 20 day Irish Breakfast and a 13 day
Oolong cultures. Biofilms were cut with a sterile razor blade into small pieces and bead
beat for 1 minute. The optional 65 C incubation was performed to increase yield.
16S sequencing of isolates was performed using bacterial primer set 8F and
1492R with Promega Master Mix as per manufacturer’s instructures. Thermocycler
conditions were: 5 min boil (95 °C), 5 min denaturation 95 °C, 30 cycles of 95 °C 30 s
denaturation, 46 °C 30 s annealing, 72 °C 1.5 m extension, followed by a final extension
for 5 m at 72 °C and a hold at 4 °C. PCR clean-up was performed using MinElute
columns as per manufacturer’s instructions (Qiagen). Twenty ng/uL were submitted for
sequencing
Tag-pyrosequencing was performed on two independent biofilms. The first
biofilm was 20 days old and grown in Irish Breakfast tea, the second was 13 days old
and grown in oolong tea. Both cultures were 10 mL and ~15% sucrose.
SU rRNA genes were amplified with barcoded primers that also incorporated the
roche 454 Ti adapter sequences, with the barcode on the forward primer. Each
barcode was 9 nt long and all the barcodes we used are in the mapping file for the 454
data on dropbox. The primer targets 515F and 907R on the E. Coli 16S gene. The
forward primer was (X denotes barcode, lowercase is the linker between barcode and
16S primer):

5'-
CGTATCGCCTCCCTCGCGCCATCAGXXXXXXXXgaGTGYCAGCMGCCGCGGTAA-3'

The reverse primer was (lowercase denotes linker between adapter and 16S primer):
5'-CTATGCGCCTTGCCAGCCCGCTCAGggCCGYCAATTCMTTTRAGTTT-3'

Phusion HF polymerase (2X MasterMix) was used to amplify the gene with 8% DMSO
and 0.5 uM primers in the final reaction volume. Template was normalized to 15 ng/uL.
Touchdown PCR program annealing temperature was from 58-58 C for the first 10
cycles, followed by 12 cycles of three-step PCR (denaturation, annealing, elongation)
and 10 cycles of two-step PCR. DNA was quantified using the PicoGreen and pooled
~125 ng each PCR product. That pool was concentrated down 100 uL and gel purifed
using the Montage Kit (Millipore). The gel-purified pool was sequenced by the Penn
State sequencing facility.
Analyses of the 454 data were performed using the QIIME workflow (Caporaso et
al., 2010), with quality control for minimum length 400 bp, maximum length 600 mp, no
primer mismatches, and a minimum quality score of 25. Clusters were defined at 97 and

! %!
100% sequence identity using CD-HIT (Li and Godzik, 2006). Taxonomic assignments
were made with RDP. Species abudnace distribution models were fit using the vegan
package in the R environment for statistical computing (R Development Core Team,
2010)(Team", 2010).

iii. Microscopy
Dead cell staining was performed on live biofilm cross sections (manually
prepared with a sterile razor blade) using propidium iodide (Invitrogen) as per
manufacturer’s instructions.
Broth from kombucha cultures were fixed overnight in 1% final concentration of
paraformaldehyde at 4 °C or 1-4 hours at room temperature. 150 uL of the broth was
filtered through 0.22 polycarbonate filters (Nucleopore, Pall). Filters were store at -20
until further processing. Visualization of cells by DAPI staining were performed on slices
of filters by washing in 1 ug/mL of DAPI solution for 5-15 minutes, washing in sterile
milliq water for 10 minutes, washing in ethanol for 5 minutes, and then air drying in the
dark. Filters were mounted in a 75:25 Citifluor: VectraShield solution and visualized with
epifluorescence microscopy. CARD-FISH was performed according to (Pernthaler et al.,
2004) using the Alf968 probe for Alpha-Proteobacteria and Alexa 488 tyramide.
Confocal images were from a Zeiss LSM-700 microscope.

iv. Experimental design

The tea and pH experiment included five treatments with three replicates each,
followed for eleven days. The treatments were different media, including oolong,
Ceylon, and green teas, oolong tea buffered with MOPs, and FWC liquid media. Acid
was measured daily and 1 mL of broth collected for HPLC analysis and stored at -20
until further processing. On the final day, depth of the biofilm at the thickest point was
measured, and samples were preserved for future molecular analysis of community
composition.

Results and Discussion

i. Composition
Isolation uncovered four different colony morphologies from the kombucha,
including yeast and bacteria (Figure 1). All isolates were submitted for 16S sequencing,
but only isolate Kmbch_06 had a quality sequence and was preliminarily identified as a
Gluconacetobacter species. Notably, not all bacterial isolates produced acid, as
assayed by halo formation on the calcium carbonate plates.
Tag pyrosequencing resulted in 36,414 quality sequences (SCOBY 1 = 18,699
tags, SCOBY 2 = 17,715 tags), and 131 OTUs at the 97% sequence identity clusters
using CD-HIT. Of these OTUs, 26 were non-singleton OTUs and 8 of these were
assigned to the Gluconacetobacter genus. Species abundance distributions revealed a
few dominant OTUs (all Gluconacetobacter) with many singletons in both SCOBYs
(Figure 2). Few OTUs were identified that were not either Gluconacetobacter or
belonging to the Acetobacteraceae family. These included a few chloroplast (likely form
the tea), and also OTUs identified with Alteromonadaceae, Halomonadacae
Halomonas, Spirochaetacea Spirochaeta, Staphylococcaceae Staphylococcus,

! &!
Streptococcaceae Streptococcus, Bacillariophyta, and Desulfobacteraceae. These
OTUs may represent flanking rare members of the kombucha community.
To further probe population-level diversity within the Gluconacetobacter, OTUs
were clustered at 100% sequence similarity, resulting in 2854 OTUs and 2697 of these
were identified as Gluconacetobacter. Alignments of the tag-sequences revealed that
many of these OTUs were different in a single base pair within the tag-sequences
(which were 414 bp in length). A one to one plot of the independent SCOBY samples
revealed that the abundances of Gluconacetobacter OTUs were very consistent across
samples, suggesting that the patterns in population-level diversity are not due to
random sequencing error (Figure 3). This also suggests that there is low compositional
variability between mature SCOBY communities. Consideration of the ten most
abundant Gluconacetobacter OTUs supported this consistency across samples (Table
1).
A local database of the 100% identity OTUs was created, and isolate 16S
sequences of was BLASTed against this to determine if one of the abundance OTUs
had been isolated. Surprisingly, Kmbch_06 isolate, the acid producer with small
colonies in Figure 1c and 1g had 100% identity in the overlapping region to OUT 1973,
which was a singleton sequence in the dataset.
These results show that while there is low bacterial community diversity in
kombucha, there is strong evidence for high population-level diversity. The differences
in phenotypes, especially isolating bacteria that produces cellulose and another that
produces acid, suggests some complementation of roles of Gluconacetobacter in the
biofilm.

ii. Temporal and spatial dynamics of kombucha

CARD-FISH with Alf968 probe was used to determine the proportion of Alpha-
Proteobacteria in the kombucha broth. Samples used for this analysis were replicate 10
mL cultures of 15% Oolong, sacrificed daily. No unlabeled cells were detected,
suggesting that the majority of cells were alphas. Therefore, DAPI staining was used to
count cells from the time series and identify Eukaryotes by nuclear staining (Figure 4).
Total cells in the broth peaked on day 3 after inoculation, suggesting some inhibition of
growth of planktonic populations, possibly because of increased acid or other
antimicrobial products.
A series of confocal images through the horizontal axis of the biofilm was
observed using DIC imaging with propidium iodide dead cell staining (Supplemental
Material: Confocal Z stacks movie). This revealed high spatial heterogeneity within
the biofilm, as there were patches of dead Acetobacter-like cells, unstained
Acetobacter-like cells throughout the cellulose matrix, and patches of unstained yeast
cells on a different plane. These results suggest that the yeast and Gluconacetobacter
are not physically associated with each other within the biofilm, but that the biofilm may
have specific compartments for functions (e.g. ethanol fermentation by yeast).

iii. The influence of pH and tea type on kombucha function

I designed an experiment to better understand the temporal dynamics of the
microbial products in kombucha given different tea origins and acidities. Also I asked
whether enrichment on tea was essential for kombucha microbes by inoculating

! '!
freshwater minimal media (FW) with kombucha broth. There were five treatments with
three replicates each: Ceylon black tea, green tea, oolong tea, oolong tea buffered with
MOPS (range 6.5-7.9), and freshwater minimal media. Ninety milliliter enrichments were
used, and broth was removed each day for eleven days to measure pH and organic
acids. On the final day, biofilm thickness was measured as a proxy for cellulose
production.
pH steadily decreased in all of the treatments, but the oolong buffered mops
maintained higher pH than the others (Figure 5). The pH remained lowest in the FW
treatment, I observed that the biofilm formed most rapidly. The different tea types were
similar in their pH dynamics.
Organic acids analysis of non-inoculated control media showed that all of the
media were very similar in initial lactose and fructose concentrations, but that the FW
media had higher initial glucose than the others (Figure 6a). Because all media were
prepared with 15% sucrose, these concentrations of simpler sugars are likely due to
natural breakdown from specific media conditions. Ethanol was produced immediately
and gradually in all of the treatments (Figure 6b). Ceylon tea had the most ethanol
production by the end of the time series, twice as concentrated as the other treatments.
FW ethanol increased most slowly, but by day 10 matched the concentrations of the
other treatments (except for Ceylon). Acetic acid was not produced in significant
quantities until day 5 (Figure 6c), suggesting that a minimum threshold of ethanol
(resulting in sucrose degradation to fructose and glucose) may be required before
Gluconacetobacter could begin acetic acid production. Further, it makes sense that the
Gluconacetobacter would grow more quickly given simpler carbon sources. The most
acetic acid was produced in the FW treatment, and the least in the oolong MOPS-
buffered tea. The former suggests the hypothesis that tea may inhibit growth of
Gluconacetobacter, leading to a decrease in acid production. Lactose concentrations
were most dynamic in time, with a gradual drop until day 3 and then a sharp increase by
day 5, followed by another drop to day 10 (Figure 6d). However, these dynamics were
consistent across all treatments. These data suggests a hypothesis that there may have
been an increase in flanking lactic acid producing bacteria that were later outcompeted
by Gluconacetobacter. Note, however, the generally low concentrations (y axis) in lactic
acid throughout. Analysis of the community composition through the time series would
be necessary to test this hypothesis. Glucose dynamics, however were not consistent
across treatments (Figure 6e). Oolong buffered with MOPs had twice as much glucose
than the other treatments throughout the experiment. Fructose gradually decreased in
all treatments (Figure 6f), though the magnitude was much less in the MOPS buffered
oolong.
Biofilm thickness was consistent across treatments, however the oolong
treatment was much more variable than the others and the FW treatment had a greater
thickness (Figure 7). Future work should measure dry biomass as a better proxy for
cellulose production in the biofilm.

Conclusions
My preliminary explorations of the microbial ecology of the kombucha biofilm
have only created more questions. I have shown that kombucha is dominated by
Gluconacetobacter, and that there is phenotypic and functional diversity in isolate (some

! (!
produce more acid, some produce more cellulose). I have demonstrated by DAPI
counts that the abundance of pelagic community changes in time, and by confocal
microscopy with propidium iodide staining that there is high spatial heterogeneity in the
biofilm. In my “tea experiment”, I showed that pH and tea origin separately play a role in
the dynamics of known products of the kombucha consortia, and also that the tea is not
required for kombucha biofilm formation, as the biofilm in the FW treatment grew even
thicker (more cellulose production) than the other treatments, despite that the acetic
acid production remained low and glucose remained high. An interesting finding was in
the population-level diversity within the Gluconacetobacter OTUs observed with deep
454-sequencing (arguably exhaustive for this simple bacterial consortia). Full 16S
and/or rpoB sequencing of isolates from the kombucha would help to address the
question of how much diversity exists within Gluconacetobacter, and what
environmental or biological factors may drive this diversity. Another interesting unknown
is the identity of yeast species in kombucha; isolation and sequencing of the yeast will
provide a foundation for understanding Eukaryotic- bacterial interactions in the
kombucha consortia.

Acknowledgements
A.S. is a Gordon and Betty Moore Foundation Fellow of the Life Sciences Research
Foundation. Funds for participation in Microbial Diversity were provided by the Life
Sciences Research Foundation fellowship to AS, the American Society for Microbiology
Early Career Development Grant for Postdoctoral Women to AS, and National Science
Foundation grant to the Microbial Diversity course (Buckley and Zinder).

! )!
References
Blanc, P.J. (1996) Characterization of the tea fungus metabolites. Biotechnology
Letters.18: 139-142.
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. Nature. 7: 335-336.
Carpenter, S.R., and Url, S. (2011) Microcosm Experiments have Limited Relevance for
Community and Ecosystem Ecology. Ecology. 77: 677-680.
Chen, C., and Liu, B.Y. (2000) Changes in major components of tea fungus metabolites
during prolonged fermentation. Journal of Applied Microbiology. 89: 834-839.
Cvetkovic, D., Markov, S., Djuric, M., Savic, D., and Velicanski, a. (2008) Specific
interfacial area as a key variable in scaling-up Kombucha fermentation. Journal of Food
Engineering 85: 387-392.
Dufresne, C. (2000) Tea, Kombucha, and health: a review. Food Research International
33: 409-421.
Dutta, D., and Gachhui, R. (2007) Nitrogen-fixing and cellulose-producing
Gluconacetobacter kombuchae sp. nov., isolated from Kombucha tea. International
journal of systematic and evolutionary microbiology 57: 353-357.
Iguchi, M., Yamanaka, S., and Budhiono, A. (2000) Bacterial cellulose—a masterpiece
of nature's arts. Journal of Materials Science 35: 261-270.
Jayabalan, R., Marimuthu, S., Thangaraj, P., Sathishkumar, M., Binupriya, A.R.,
Swaminathan, K., and Yun, S.E. (2008) Preservation of kombucha tea-effect of
temperature on tea components and free radical scavenging properties. Journal of
agricultural and food chemistry 56: 9064-9071.
Jessup, C.M., Kassen, R., Forde, S.E., Kerr, B., Buckling, A., Rainey, P.B., and
Bohannan, B.J.M. (2004) Big questions, small worlds: microbial model systems in
ecology. Trends in Ecology & Evolution 19: 189-197.
Kurtzman, C.P., Robnett, C.J., and Basehoar-Powers, E. (2001) Zygosaccharomyces
kombuchaensis, a new ascosporogenous yeast from 'Kombucha tea'. FEMS yeast
research 1: 133-138.
Li, W., and Godzik, A. (2006) Cd-hit: a fast program for clustering and comparing large
sets of protein or nucleotide sequences. Bioinformatics 22: 1658.
Malbasa, R., Loncar, E., Djuric, M., Klasnja, M., Kolarov, L., and Markov, S. (2006)
Scale-Up of Black Tea Batch Fermentation by Kombucha. Food and Bioproducts
Processing 84: 193-199.
Pernthaler, A., Pernthaler, J., and Amann, R. (2004) Sensitive multi-color fluorescence
in situ hybridization for the identification of environmental microorganisms. Molecular
microbial ecology manual, 2nd ed Kluwer Academic Publishers, Dordrecht, The
Netherlands: 711-726.
Prosser, J.I., Bohannan, B.J.M., Curtis, T.P., Ellis, R.J., Firestone, M.K., Freckleton,
R.P. et al. (2007) The role of ecological theory in microbial ecology. Nature Reviews
Microbiology 5: 384-392.
R Development Core Team (2010) R: A language and environment for statistical
computing. In. Vienna, Austria: R Foundtation for Statistical Computing.
Teoh, A.L., Heard, G., and Cox, J. (2004) Yeast ecology of Kombucha fermentation.
International Journal of Food Microbiology 95: 119-126.

! *!
Figure Legends
Figure 1. Colony and cell morphologies of representative isolates from kombucha. a)
Yeast isolate on calcium carbonate plate with glucose and ethanol; b) Acetobacter-like
isolate on calcium carbonate plate, globs are likely cellulose; c) Acid-producing isolate
on calcium carbonate plate; d) Acetobacter-like isolate on FWC plate without vitamins,
globs are almost 1 mm high and are likely cellulose; e) DIC image reveals cell
morphology of yeast isolate in (a); f) Phase contrast reveals of cell morphology of
bacterial isolate in (b), long rods occurring in pairs; g) Phase contrast image cell
morphology of bacterial isolate from (c), squat rods occurring in pairs; h) Phase contrast
image of cell morphology of bacterial isolate from (d), long rods in pairs or chains with
bulbs, arrow shows likely cellulose fibers. White bars are 10 microns, e-h are 100x.

Figure 2: Species abundance distributions of SCOBY 1 and SCOBY 2. No model

tested (lognormal, preemptive, Zipft, Zipft-Mandelbrodt, Null) fit well to the distribution.

Figure 3. SCOBY 1 and SCOBY 2 comparison of abundances of tag-pyrosequences

from OTUs identified most closely as Gluconacetobacter using 100% clustering of tag-
pyrosequences with CD-HIT in the QIIME workflow. A linear model revealed a
significant 1:1 relationship, indicating that most OTUs were equally represented in both
biofilms.

Figure 4. Cell counts by DAPI staining through time. a) DAPI and b) CARD-FISH with
alpha-Proteobacteria probe Alf968 and Alexa 488 was used to confirm that all observed
bacterial cells observed belong to alpha-Proteobacteria. Therefore, DAPI was used for
total cell counting and identification of Eukaryotes by nuclear staining. c) Blue bars are
total cell counts (bacteria and Eukaryotes), and red bars are total Eukaryotic cells.

Figure 5. pH dynamics of different treatments to understand the influence of tea origin

and pH on the kombucha microbial consortia. Error bars are standard deviations around
the mean of three replicates per time point per treatment.

Figure 6. Organic acids analysis by HPLC to understand the influence of tea origin and
pH on the kombucha microbial products. a) Initial control concentrations of sugars. No
acetic acid or ethanol was detected in the controls. b) Ethanol; c) acetic acid; d) lactose;
e) glucose; f) fructose concentrations through time. Error bars are standard deviations
around the mean of three replicates per time point per treatment.

Figure 7. Thickness of biofilms formed by the end of eleven days, as a proxy for
cellulose production and biofilm growth. Measurements were at the thickest point in the
biofilm.

! "+!
Table 1. Ten most abundant OTUs identified as Gluconacetobacter, clustered at 100%
tag-pyrosequence identity using CD-HIT. Sequences are available in supplemental
information.
OTU No. sequences SCOBY No. sequences SCOBY Representative sequence
ID 1 2 ID
1095 1011 1037 ASB1_27557
627 860 858 ASB1_27505
1089 690 733 ASB1_27517
1088 696 670 ASB1_27509
517 634 685 ASB1_27522
1101 684 607 ASB1_27641
39 565 635 ASB1_27533
1092 536 483 ASB1_27534
1087 499 503 ASB1_27507
1020 537 434 ASB1_27540

! ""!
Supplemental material: movie of series of biofilm DIC propidium iodide stain, along the
horizontal axis of the biofilm. Propidium idodide stains dead cells. The movie shows
patches of dead Acetobacter-like cells, unstained Acetobacter-like cells throughout the
cellulose matrix, and unstained yeast patches on a separate plane.

! "#!
 !"#$%&'('

!" #" $" ("

%" &" '" )"

!"#$%&')'

!" Species abundance distr. SCOBY 1 #" Species abundance distr. SCOBY 2

10000
10000

1000
1000
Abundance

Abundance

100
100

10
10
1

1
0 10 20 30 40 50 60 0 20 40 60 80

Rank Rank
!"#$%&'*'

1000
Gluconacetobacter OTUs, 100% seq. identity

m = 1.023, p < 2e-16

800
No. seqs per OTU SCOBY 2

600
400
200
0

0 200 400 600 800 1000

No. seqs. per OTU SCOBY 1

!"#$%&'+'

!" #"

$567" 5.'8!"9::"

,"
0'..1234"

$!%"!&'(")*+,-.!/+*"
!"#$%&','

7 Ceylon Oolong
FWC Oolong Mops
Green

5
pH

2 4 6 8 10
Experiment Day
 !"#$%&'-' !" HPLC controls for tea experiment

RID_Lactose
RID_Glucose
RID_Fructose

800
Organic acid concentration (mM)

600
400
200
0

Ceylon FWC Green OolongMops Oolong

#" ," ;"

Control Media
200

200

0.8
Lactose_Avg (mM)
Ethanol_Avg (mM)

Acetate_Avg (mM)
150

150

0.6
100

100

0.4
0.2
50

50

0.0
0

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10

Experiment Day Experiment Day Experiment Day

'" <"
500

450
400

Fructose_Avg (mM)
Glucose_Avg (mM)

400

Ceylon Oolong
300

FWC Oolong Mops

Green
350
200
100

300

2 4 6 8 10 2 4 6 8 10

Experiment Day Experiment Day

!"#$%&'.'

Day 11 biofilm thickness

10

OolongMops
Green
FWC
Oolong
8

Ceylon
SCOBY thickness (mm)

6
4
2
0

OolongMops Green FWC Oolong Ceylon

Media