Received: 16 November 2023 Revised: 23 December 2023 Accepted: 11 January 2024

DOI: 10.1111/1750-3841.16957

ORIGINAL ARTICLE
I n t e g ra t e d Fo o d S c i e n c e

Sourdough starter culture microbiomes influence physical

and chemical properties of wheat bread

Caitlin S. Clark1 Ashley Ohstrom2 M. Laura Rolon2 Molly Smith1

Benjamin E. Wolfe3 Josephine Wee2 Charlene B. Van Buiten1

1 Department of Food Science and Human

Nutrition, Colorado State University, Fort Abstract: Sourdough fermentation is an ancient leavening method that uses
Collins, Colorado, USA wild yeasts to produce carbon dioxide, contributing to bread rise, and bacteria
2 Department of Food Science, The
which produce organic acids. Sourdough starter cultures are known to be diverse
Pennsylvania State University, University
Park, Pennsylvania, USA in terms of the microorganisms they comprise and while specific genera and
3 Department of Biology, Tufts University, species of microorganisms have been identified from starters and associated with
Medford, Massachusetts, USA specific attributes, overarching relationships between sourdough starter culture
microbiomes and bread quality are not well understood. The objective of this
Correspondence
Charlene B. Van Buiten, Department of study was to characterize differences in the physical and chemical properties
Food Science and Human Nutrition, of breads produced with sourdough starter cultures with unique microbiomes.
Colorado State University, Fort Collins,
80523 CO, USA.
Sourdough starter cultures (n = 20) of known microbial populations were used
Email: to produce wheat-based dough and bread, which were analyzed for chemical
[email protected]; and physical properties then compared to their microbial populations in order to
Josephine Wee, Department of Food
Science, Pennsylvania State University, identify relationships between microbial profiles and dough/bread qualities. All
University Park, 16801 PA, USA. Email: samples were also compared to bread produced only with Saccharomyces cere-
[email protected]
visiae (baker’s yeast). Significant differences among pH, titratable acidity, loaf
Funding information volume, crumb firmness, crust color, free amino acids, and organic acids were
Colorado Agricultural Experiment observed when comparing sourdough breads to the yeast-only control (p ≤ 0.05).
Station; National Institute of Food and
Furthermore, bacterial diversity of sourdough starter cultures was correlated
Agriculture, Grant/Award Number:
2023-67017-40251; U.S. Department of with lactic acid and free amino acid in the dough and loaf volume and crumb
Agriculture, Grant/Award Number: firmness of baked breads. No significant correlations were found between fun-
7004809
gal diversity and measured outcomes. These data demonstrate the importance of
considering sourdough starter microbiomes as an ingredient in baked goods and
they contribute to quality and safety outcomes in bread production.

KEYWORDS
bread, food microbiome, lactic acid bacteria, sourdough fermentation

Practical Application: Sourdough starter cultures have diverse and dynamic

populations of bacteria and yeasts, which contribute to the production of bread
products. These populations can influence the physical and chemical properties
of sourdough fermentation and final breads. Understanding of the relationship
between sourdough starter microbiomes and bread quality parameters can lead

1414 © 2024 Institute of Food Technologists. wileyonlinelibrary.com/journal/jfds J. Food Sci. 2024;89:1414–1427.

SOURDOUGH MICROBIOMES AND BREAD QUALITY 1415

to targeted development of sourdough bread products with specific physical and

chemical properties.

1 INTRODUCTION terizing 500 sourdough starter cultures collected from

North America, Europe, and Australasia found that pro-
Sourdough bread is an ancient fermented food made cess parameters and geography were weak predictors of
by the fermentation of dough with “starter cultures” or bacterial and fungal composition of sourdough starter
communities of naturally occurring bacteria and yeasts. cultures. When identifying indicator taxa that were signif-
These microbial communities are responsible for the fer- icantly enriched by age of starters as a process parameter,
mentation of carbohydrates in flour to produce carbon Lactiplantibacillus plantarum subsp. plantarum and Lev-
dioxide (CO2 ) that causes bread dough to rise. In most ilactobacillus brevis dominated younger starters, whereas
commercially-produced conventional doughs and breads, Fructilactobacillus sanfranciscensis and Pediococcus parvu-
CO2 is produced by a single known strain of yeast added lus were more prominent in older starters (Landis et al.,
to the dough specifically for leavening. The yeast used is 2021). S. cerevisiae was noted as the most prevalent yeast
typically Saccharomyces cerevisiae, colloquially referred to species across all samples, and LAB populations were often
as “baker’s yeast.” Sourdough starter cultures, on the other dominated by F. sanfranciscenis or coordinated popula-
hand, comprise wild yeasts, lactic acid bacteria (LAB), and tions of L. plantarum and L. brevis.
acetic acid bacteria (AAB) endogenous to the flour they The contributions of individual microorganisms
are created with and the environments they are main- towards bread quality have been studied in controlled
tained in (Landis et al., 2021). Although these microbial environments, such as the addition of selected LAB
communities produce CO2 to leaven bread similarly to isolates to conventional yeast-only dough fermentations.
S. cerevisiae, they are also known to produce a variety of For example, doughs fermented with a commercial S.
metabolic byproducts that drive quality differences and cerevisiae and supplemented with select strains of F.
sensory attributes between conventional yeast-fermented sanfranciscensis and L. plantarum have been shown to
breads and sourdough breads (de Vuyst et al., 2017). stale slower due to their additional α-amylase, whereas
Organic acids are perhaps the most notable metabolic Lentilactobacillus hilgardii has demonstrated the same
byproducts of sourdough in comparison to conventional benefit due to the production of exopolysaccharides
dough due to their distinct effect on bread flavor. Lactic (Corsetti et al., 2000). While useful in delineating the
and acetic acid are produced by LAB and AAB, respec- functional potential of these isolates, studies designed
tively. These acids create a low pH environment, typically with individual microorganisms eliminate the inherent
between pH 3.5 and 4.3, in comparison to pH 5.0-5.5 as complexities of sourdough cultures as they exist in homes,
observed in conventional yeast-only dough (Arora et al., bakeries, and food processing facilities—as diverse,
2021). From the perspective of food safety and shelf life, dynamic communities of organisms that interact with
this decrease in pH can limit the growth of spoilage one another in a complex growth environment. The
microorganisms over the course of fermentation and dur- importance of the whole microbiome as a diverse and
ing storage (Suhr & Nielsen, 2004). Increased acidity of dynamic community of microorganisms is underscored
sourdough breads also facilitates the hydrolysis of pro- by the apparent antagonistic effects that some cultures
teins and starches by enzymes endogenous to flour and can have on one another. For example, sample doughs
those produced by the starter culture, which can influ- fermented with L. brevis and L. plantarum contained
ence the sensory characteristics of the bread (Arendt et al., greater final concentrations of starch, maltose, and
2007). Enzymatic hydrolysis of dough proteins also liber- fructose when Lactobacillus amylovorus or Pediococcus
ates free amino acids, which can participate in the Maillard pentosaceus were added to the ferment, which is likely
reaction, producing melanoidins and volatile aroma com- caused by competition for substrate and decreased activity
pounds that affect crust color and flavor, respectively of heterofermentative LAB (Siepmann et al., 2019).
(Yıltırak, 2022; Shen et al., 2019). Presently, nearly all of the studies examining the rela-
Sourdough starter cultures are known to have diverse tionship between microorganisms from sourdough starter
microbial profiles. While yeasts, LABs and AABs are cultures and bread quality have focused on yeasts and LAB.
always present, the precise species diversity and abun- However, differences in the production of volatile aroma
dance of organisms can vary widely between individual compounds and dough rise in microscale fermentations
starter cultures. A recent comprehensive study charac- of sourdough starters were shown not to be associated
1416 SOURDOUGH MICROBIOMES AND BREAD QUALITY

with the predominant yeasts and LAB populations, but water. This was followed by two subsequent feedings of
with the variable population and abundance of AAB 120 and 200 g of 1:1 flour and water after 24 and 48 h,
including Acetobacter (Landis et al., 2021). This suggests respectively.
that nondominant microorganisms can influence impor-
tant quality parameters in baked breads using sourdough
starters and highlights the need to examine microorgan- 2.2 Dough preparation and baking
isms that comprise sourdough starter cultures as whole
communities rather than isolated strains. Motivated by Eight hours after the final propagation step, each active
this need, the objective of the present study was to deter- starter was used to prepare 1 kg bread dough (65% hydra-
mine and compare the physical and chemical properties tion) by combining 200 g of sourdough starter with 400 g
of dough and breads with the diverse sourdough starter King Arthur all-purpose flour, 225 g autoclaved dH2 O, and
microbiomes from which they were prepared. We hypoth- 12 g non-iodized kosher salt. An S. cerevisiae yeast-only
esized that sourdough starter microbiomes would result in dough (henceforth referred to as yeast control) was pre-
differences in physical and chemical properties of dough pared to the same hydration by combining 665 g King
and breads. Arthur all-purpose flour, 430 g autoclaved dH2 O, 12 g non-
iodized kosher salt, and 10 g active-dry yeast (ACH Food
Companies, Inc.).
2 MATERIALS AND METHODS All doughs were prepared in sanitized 5-quart bowls
attached to KitchenAid mixers (KitchenAid) and kneaded
2.1 Starter culture description and for 10 min with a dough hook attachment. Kneaded
preparation doughs were transferred to autoclaved glass bowls, cov-
ered with plastic film, and fermented at 35◦ C and 30%
Twenty unique sourdough starter microbiomes were relative humidity for 24 h. The yeast control dough was
obtained from the citizen science-driven Global Sour- fermented under the same conditions for 2 h per manufac-
dough Project (Landis et al., 2021). These 20 sourdough turer’s instructions, as S. cerevisiae produces CO2 quickly
starter culture samples were selected to represent dis- and requires only a short rise time.
tinct microbiomes based on beta- (Figure S1) and alpha- Each fermented dough was separated into three 150 g
diversity (Table S1, Figure S2) of mapped sequences out loaves and placed in 9 × 6.5 × 15 cm aluminum pans. All
of 500 samples from the Global Sourdough Project (Lan- loaves were fermented for an additional 3 h at the same
dis et al., 2021), allowing for downstream characterization conditions before baking at 180◦ C for 40 min. The remain-
of physical and chemical properties of dough and breads ing dough from each sample was divided into two portions,
(Figure 1). The sourdough starter cultures used in this one for immediate chemical analysis (pH, titratable acid-
study had been stored at −80◦ C at Tufts University since ity, and water activity) and another for lyophilization and
sequencing and were shipped to Colorado State Univer- storage at −20◦ C until organic acid and free amino acid
sity for propagation into active phase as described below. analysis. After baking and cooling for 2 h to room temper-
Although it is possible that the microbial composition ature, analysis of physical attributes was carried out for all
of the samples could be different from what was previ- loaves in triplicate.
ously reported by Landis et al. (2021), the starter cultures
were not re-sequenced after propagation as recent studies
demonstrate the stability of sourdough starter cultures in 2.3 Dough analysis
terms of microbial composition over time (Calabrese et al.,
2022; Minervini et al., 2018). To reduce the potential for 2.3.1 pH and titratable acidity
shifts in microbial composition due to available nutrients
in the flour, starters were fed with same type and brand of Ten grams of each dough were weighed and combined
flour used in Landis et al. (2021) (Gobbetti et al., 2016). with 100 mL of deionized water in a sanitized commer-
Sourdough starters were propagated into active phase cial food blender (Oster) and transferred to a glass beaker
by combining 5 g aliquots of the starter with 10 g of a for mixing on a magnetic stir plate. The pH was recorded
1:1 w/w mixture of autoclaved deionized water (dH2 O) and while mixing using an Accumet AB150 pH meter (Fisher
all-purpose flour (King Arthur Baking Company, Inc.) in Scientific). Titratable acidity of the diluted dough solu-
loosely capped 50 mL polypropylene tubes for 24 h. The tion was measured by titrating with 0.1 N NaOH (Santa
volume of each starter was increased to 375 g for use in Cruz Biotechnology) into the solution until the pH of the
baking by transferring each starter to an autoclaved glass mixture reached 8.4. Titratable acidity was reported as the
jars before the immediate addition of 40 g of 1:1 flour and volume of 0.1 N NaOH used for titration.
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1417

F I G U R E 1 Workflow of sample collection and analysis. (a) Twenty unique sourdough microbiomes were obtained from the citizen
science-based Global Sourdough Project. (b) Raw reads were obtained from NCBI for amplicon analysis. Frozen starter cultures were
propagated from 5 g total material to 200 g total material through four feedings of flour and water (1:1). The starter was combined with flour
and water to produce a dough, which was kneaded and left to ferment. Dough samples were reserved for analysis. The remaining dough was
baked and underwent the indicated analyses immediately after cooling. (c) Water activity, pH, titratable acidity, organic acids, and free amino
acids were measured from dough while loaf volume, crumb firmness, and crust color were measured for baked breads.

2.3.2 Water activity comparison to standard curves generated with 0–100 mM

of lactic acid and acetic acid standards, respectively. All
The water activity (Aw ) of each dough was measured using samples were injected in technical triplicate. The molar
a calibrated Aqualab 3TA water activity meter (Decagon ratio between lactic and acetic acid was used to calculate
Devices). For each sample, the bottom of a plastic sample the fermentation quotient of each bread (Arora et al., 2021;
boat was coated with 10 g of dough and then inserted into de Luca et al., 2021).
the water activity meter. The meter was auto-adjusted to
room temperature for measurement.
2.3.4 Free amino acid concentration

2.3.3 Organic acid concentration Free amino acids were measured in lyophilized dough
according to previously established methods (Friedman,
Lactic and acetic acid were extracted according to a pre- 2004) using a BioTek UV–Vis Plate Reader (BioTek).
viously established method (de Luca et al., 2021). Three Free amino acid concentration was determined using a
independent extractions of dough were performed. Stan- standard curve of L-leucine (Thermo Fisher).
dards of lactic acid (VWR) and acetic acid (Millipore
Sigma) were prepared at concentrations of 100, 50, 25,
12.5, 6.25, and 3.125, and 50 mM butyric acid was used 2.4 Bread analysis
as an internal standard. Standards and filtered sourdough
samples were injected into a Dionex ICS-5000+ high per- 2.4.1 Crumb firmness
formance liquid chromatography (HPLC) system (Thermo
Scientific) equipped with an Aminex HPX-86H column Two 12.5 mm slices were removed from the center of each
(300 × 7.8 mm; Bio-Rad Laboratories) kept at 50◦ C. The bread loaf for texture analysis according to AACC Method
analytical conditions used were as follows: 20 µL injection 74-09. The slices were stacked to form a 25 mm-thick test
volume, 0.45 mL/min flow rate, and 5 mM H2 SO4 as an sample and the force of compression was measured using
eluent. Lactic and acetic acid were identified within each a TA-XT2 texture analyzer (Stable Microsystems) equipped
chromatogram based on retention time. Concentrations of with a 25.4 mm diameter acrylic probe. The bread was com-
lactic and acetic acid were calculated based on peak area in pressed to 6.2 mm and the peak load (mN) was recorded
1418 SOURDOUGH MICROBIOMES AND BREAD QUALITY

using Texture Expert Exceed Version 2.64 software (AACC analysis (PCoA) ordination plot was generated for bacteria
International, 2022). and fungal communities (Figure S1). Hierarchical clus-
tering analysis (HCA) of fungal and bacterial ASVs was
constructed based on Bray–Curtis dissimilarities using the
2.4.2 Crust color Ward algorithm (ward.d2) at k = 5 for bacteria and k = 6 for
fungi using the ggplot2 package FactoMineR version 3.4.4.
A 1.25 cm sample of crust from the end of each loaf was Alpha diversity was calculated using four separate mea-
placed into the sample cup of a calibrated HunterLab Col- sures, Chao1 and seChao1 (two different estimates of total
orFlex Colorimeter (Model 45/0). Due to spatial variation species richness), Shannon (estimate of species richness
in color across a loaf of bread, multiple 1.25 cm samples of and the evenness of species distribution), and InvSimpson
crust were obtained from three independent locations on (estimate of the richness in a community with uniform
the loaf. CIE L* was measured in triplicate for each sample evenness) (Figure S2). To visualize correlation between
(Troadec et al., 2022). measured physical and chemical properties of dough and
baked breads, Shannon alpha-diversity metric was selected
as it considers both species richness and evenness (Figures
2.4.3 Loaf volume S3 and S4).
Taxonomic composition for mapped reads above 1% rel-
Loaf volume was assessed via rapeseed displacement ative abundance was plotted using the R package ggplot2
according to AACC Method 10–50.01 (AACC Interna- v3.3.5 (Wickham, 2016). The new and old taxonomic
tional, 2022). nomenclature of the genus Lactobacillus was used in this
study for ease of comparison and consistency of classifi-
cation with Landis et al. (2021). The new nomenclature
2.5 Microbiome analyses is based on the reclassification of the genus Lactobacillus
into 25 genera by Zheng et al. (2020) and validated with
Raw sequence reads were retrieved from the publicly avail- EMBL and NCBI databases (Schoch et al., 2020). Pearson
able dataset (NCBI BioProject Accession PRJNA589612) correlation analyses for physical and chemical character-
and analyzed with the R package DADA2 v3.14 pipeline istics of dough and baked sourdough breads with specific
following standard protocol for bacterial 16S rRNA V4 taxa were conducted by combining sample variables, abun-
gene region (515f/806r) and the fungal ITS gene region dance, and taxonomic rank at the genus level using the
(ITS1f/ITS2) (Callahan et al., 2016). Low-quality sequence psmelt function in the phyloseq package and visualized
reads were filtered, low-quality bases were trimmed, error using the corrplot package in R. All microbiome analyses
rates were calculated, and amplicon sequence variants were conducted in R v.4.1.0 (R Core Team, 2021).
(ASVs) were inferred from the remaining sequence reads.
Paired-end reads ASVs were merged into contigs and ASVs
with <251 bp or >253 bp were removed for bacterial 16S 2.6 Statistical analysis
reads. ITS reads were not selected based on length due
to biological variation of fungal ITS gene region. Remain- All statistical analyses were conducted in Prism 10 (Graph-
ing ASVs were assigned taxonomy using the Silva database Pad Software) or in R v4.3.2 (R Core Team, 2021). Quality
v132 for bacteria (Quast et al., 2013) and the UNITE v8.3 parameters of dough and bread were compared using one-
database for fungi (Nilsson et al., 2019). ASVs assigned to way analysis of variance and Tukey’s honestly significant
mitochondria, chloroplasts, and reads unassigned at the difference post hoc test in Prism 10. Values of p < 0.05
phylum or class levels were filtered out from the bacte- were considered significant. Principal component anal-
rial dataset. ASV tables were rarefied to 1800 bacterial ysis (PCA) and hierarchical clustering analysis (HCA_
reads per sample and 7000 fungal reads for downstream were used to evaluate overarching similarities in dough
analysis. and bread properties for each starter. HCA was performed
Beta diversity was estimated based on Bray–Curtis based on pairwise Euclidean distances followed by Ward’s
distances using rarefied bacteria and fungal ASVs as a mea- minimum variance clustering (hclust, method = ward.d2)
sure of how similar or different sourdough starter micro- using the ggplot2 package FactoMineR. The strength of
biomes are between 20 samples. Bray–Curtis dissimilarity associations between microbial diversity and quality out-
is calculated based on occurrence data or abundance. To comes was calculated by Pearson correlation analysis using
visualize inter-sample distances, a principal coordinate the corrplot package in R version 0.92.
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1419

F I G U R E 2 Taxonomic classification of sourdough starter microbiomes. Stacked bars represent the (a) bacterial and (b) fungal microbial
genera present in sourdough starter microbiomes with greater than 1% relative abundance of 20 sourdough starter cultures. Note that the
group labeled Lactobacillus comprises 25 distinct genera including but not limited to Fructilactobacillus, Furfurilactobacillus,
Secundilactobcillus, Schleiferilactobacillus, Levilactobacillus, and Lentilactobacillus.

3 RESULTS AND DISCUSSION Hierarchical clustering of starter cultures based on

Bray–Curtis dissimilarities of bacterial (Figure 3a) and
3.1 Taxonomic composition of twenty fungal (Figure 3b) ASVs was used to identify groups of
sourdough starter microbiomes samples with similar community composition. Interest-
ingly, 12 of the 20 samples were clustered within the same
To build upon the Landis et al., 2021 study, high quality groups for both bacterial and fungal communities. For
sequenced raw reads of 20 sourdough starter microbiomes example, sample 16 with 549, 506 with 521 and 35 with
were retrieved from NCBI and reanalyzed to provide a 461 appear together as pairs within the same clusters. In
baseline for correlation of sourdough microbiomes with addition, samples 1, 130 and 542 appear to be grouped
physicochemical properties of dough and baked breads. together similar to 24, 56, and 118 for both bacterial and
Taxonomic compositions of the sourdough starter micro- fungal communities. Although the identification of simi-
biomes above 1% relative abundance were visualized in a larities in community composition does not suggest causal
bar plot (Figure 2a,b). interactions, clustering allows for exploratory hypothesis
On average, the 20 sourdough starters included in this generation on how sourdough starter microbiomes with
study represented 5 bacterial genera and 22 fungal genera similar community composition could be driving func-
per sample. Notably, and perhaps unsurprisingly, bacterial tional outcomes of dough and baked breads. For example,
populations tended to be dominated by lactobacilli, which one hypothesis would be that sourdough starter samples
comprised over 90% of the total bacterial reads in 10 of that appear to group together would also result in similar
the 20 starters. In this instance, lactobacilli refer to the functional outcomes such as nutritional or metabolomic
25 distinct genera reclassified from the former Lactobacil- profiles.
lus genus (Zheng et al., 2020). In addition to lactobacilli,
ASVs of Acetobacter, Enterobacteriaceae, Gluconobacter,
and Pediococcus were also found in all 20 samples. Com- 3.2 Fermentation method and starter
parison of alpha diversity measures between bacterial and culture influence chemical characteristics
fungal populations showed greater richness and evenness of dough
in fungal communities compared to bacteria (Figure S2).
However, Saccharomyces was observed as the most domi- The chemical characteristics of each fermented dough
nant genus of fungi in 18 of the 20 sourdough starters with were evaluated as pH, titratable acidity, water activity, and
a relative abundance above 50% of all ASVs in 10 starters free amino acids. All sourdough samples had a lower pH
and greater than 90% in one (sample 35). (p ≤ 0.05) and a greater titratable acidity (p ≤ 0.05) than
1420 SOURDOUGH MICROBIOMES AND BREAD QUALITY

F I G U R E 3 Clustering of 20 sourdough starter cultures based on (a) bacteria and (b) fungal amplicon sequence variants (ASVs). Cluster
dendrograms were constructed based on Bray–Curtis dissimilarities using the Ward algorithm (ward.d2). Groups represent distance between
individual samples displayed as height (branch length). To identify discrete groups, k = 5 was used for bacteria ASVs and k = 6 was used for
fungal ASVs.

F I G U R E 4 Chemical properties of sourdough and yeast control dough. Analyses performed included (a) pH, (b) titratable acidity, (c)
water activity (Aw ), and (d) free amino acids. All tested parameters demonstrated significant differences between sourdough and yeast
doughs. Values are expressed as mean ± standard deviation. Different letters within graphs indicate significant differences (p ≤ 0.05).

the yeast control dough (Figure 4a,b), which had a pH of much less than the differences observed in indices of acid-
5.15 ± 0.05 and titratable acidity of 3.63 ± 0.59 mL NaOH. ity and pH; the maximum observed water activity (475;
Sourdough samples ranged from pH 3.62–3.93 and titrat- 0.975 ± 0.002) was only 2.36% greater than the minimum
able acidity of 7.17–11.17 mL NaOH. Significant differences measurement (542; 0.952±), although the maximum pH
were observed between sourdough samples based on both (yeast; 5.15 ± 0.05) was 29.77% greater than the minimum
pH and titratable acidity. pH (120; 3.62 ± 0.02), and the maximum titratable acidity
Sourdough starter microbiomes did not greatly affect (120; 11.17 ± 0.76 mL NaOH) was 67.46% greater than the
dough water activity as only 6 out of 20 sourdough samples minimum measurement (yeast, 3.63 ± 0.59 mL NaOH).
had a significantly lower water activity than the yeast con- The free amino acid content of dough was mea-
trol, which had a water activity of 0.971 ± 0.001 (Figure 4c). sured as a marker of proteolysis. All but 2 sourdough
Similar to pH and TA, differences did exist among sour- samples (16, 461) resulted in greater concentration of
dough samples. Notably, the magnitude of difference was total free amino acids compared to the concentration
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1421

F I G U R E 5 Organic acid content and fermentation quotients of sourdough and yeast control dough. Analyses performed included (a)
lactic acid content, (b) acetic acid content and (c) fermentation quotient, which signifies [lactic acid]:[acetic acid]. Values are expressed as
mean ± standard deviation. Different letters within graphs indicate significant differences (p ≤ 0.05).

measured in the yeast control dough (Figure 4d). The was not detected in the yeast control, whereas lactic acid
control dough contained the lowest concentration of free concentrations ranged from 184.9 to 245.32 mM across dif-
amino acids (12.85 ± 1.47 mg•100 g−1 ), whereas sourdough ferent sourdough starters. Lactic acid concentration was
samples ranged 15.31–24.48 mg 100 g−1 . Differences in free significantly higher in all sourdough samples in com-
amino acid content were also observed between sourdough parison to the yeast control, though differences did exist
samples. between individual sourdough samples (Figure 5a). Mea-
The concentrations of lactic and acetic acid in each surement of acetic acid concentrations revealed that only
dough were measured by HPLC and used to calculate the five sourdough samples produced a significantly greater
fermentation quotient of each bread (Figure 5). Lactic acid amount of acetic acid than the yeast control. However,
1422 SOURDOUGH MICROBIOMES AND BREAD QUALITY

significant differences in acetic acid concentration were tion has been reported in the literature and is thought to
still observed across sourdough starter samples (Figure 5b). result from acid-induced depolymerization of the gluten
The fermentation quotient of sourdough, or the ratio of network in sourdough breads, leading to decreased gas
lactic acid to acetic acid, is typically referred to as an indi- retention (Arendt et al., 2007; Thiele et al., 2004). Another
cator of quality, including flavor and shelf life (Arora et al., contributing factor may be over-proofing. Most sourdough
2021; de Angelis et al., 2007; Rocken et al., 1992). Varia- recipes used by consumers and industry producers rec-
tions in the concentrations of each organic acid described ommend adjusting proofing time on the basis of CO2
above resulted in differences in the fermentation quotient production (i.e., doubling time). Indeed, the rate of CO2
calculated for each sample (Figure 5c). The sourdough production is known to vary across each sourdough starter
samples with the lowest fermentation quotients (samples culture (Landis et al., 2021) and over-proofing sourdough
130, 523, 1, 542, 441, 24; FQ = 2.13–3.94) did not differ has been shown to contribute to decreased loaf volume
significantly from the fermentation quotient of the yeast (Verdonck et al., 2023). Proofing time was controlled to
control (FQ = 0). The highest fermentation quotient was 24 h for all samples in the present study to reduce con-
observed for samples 122 and 549 (FQ = 7.77–11.39), both of founding variables, which may help to explain differences
which contained the highest levels of lactic acid and low- between the sourdough samples when compared to the
est levels of acetic acid. Sourdough has been reported to yeast control, but should nonetheless be considered a
typically contain 15–150 mM lactic acid (median: 75 mM) limitation when translating the findings of this study to
and 1–50 mM acetic acid (median: 20 mM), and the commercial applications. Another potential contributing
typical range of fermentation quotient for sourdough is factor to the difference in loaf volume and crumb firmness
0.25–20 (median: 4.4) (Arora et al., 2021). The calculation is the colony forming units (CFU/g) or microbial load
of the fermentation quotient demonstrates that, despite present in dough that could influence rate of dough rise.
the concentrations of lactic and acetic acid both being Crust color of each bread was measured on the basis of
greater than what is typically observed in a fresh, non- lightness versus darkness as indicated by CIE L* scores
lyophilized dough, the samples used in this experiment are (Figure 6c); a greater score indicates a lighter sample.
representative of typical sourdough fermentations. The yeast control was darker than nine sourdough sam-
ples, and only one sourdough sample was significantly
darker than the control. Significant differences existed
3.3 Fermentation method and starter among sourdough samples as well. Leavening method has
culture influence physical characteristics been shown to influence crust color, with yeast fermenta-
of bread tion producing a darker crust than sourdough-fermented
breads. This has been attributed to the effect of each
All baked bread products were evaluated on the basis of fermentation method on the Maillard reaction, with sour-
physical characteristics, including crumb firmness, loaf dough fermentation favoring the generation of Strecker
volume, and crust color. Crumb firmness and loaf volume aldehydes and yeast fermentation favoring the formation
relate directly to leavening and CO2 production over the of melanoidins (Troadec et al., 2022). Although our data
course of fermentation, whereas crust color can be affected supports the finding that leavening method does affect
by pH and production of reducing sugars and free amino crust color, our observed differences between sourdough
acids. samples suggest that crust color differences may rely more
Crumb firmness is a measure of the force required to on the microbial communities involved in fermentation
compress a slice of bread a given distance, indicating soft- than simply comparing fermentation methods on the basis
ness of a samples. A greater maximum force required for of “yeast versus sourdough.”
compression indicates a stronger, or less soft, crumb. In
the present study, all but three sourdough breads (20, 130,
and 523) resulted in greater crumb strength than the yeast 3.4 Identification of relationships
control bread, as determined by greater force required between quality indices and sourdough
to compress the crumb 6.2 mm, and little variation was starter microbial composition
observed between sourdough samples (Figure 6a).
All sourdough samples also differed significantly HCA of sourdough starters based on the physical and
from the yeast control in terms of loaf volume chemical properties of the dough and baked breads was
(Figure 6b). Although the volume of the control was used to identify groups of samples with similar physical
360.56 ± 46.96 cm3 , sourdough samples ranged from and chemical features (Figure 7a). The yeast control clus-
107.71 to 223.41 cm3 . Sourdough fermentation resulting in tered separately from all 20 sourdough samples, which
decreased loaf volume in comparison to yeast fermenta- were clustered into 5 groups of 3–5 samples each. Three
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1423

F I G U R E 6 Physical properties of baked sourdough breads compared to yeast control. Analyses performed included (a) crumb firmness,
(b) loaf volume, and (c) crust color. Values are expressed as mean ± standard deviation. Different letters within graphs indicate significant
differences (p ≤ 0.05).

groups of samples clustered in the same groupings as breads were assessed and visualized by PCA, and the rela-
observed for fungal communities, whereas five groups tionships between these features and alpha diversity were
of samples clustered in the same groupings as observed evaluated using Pearson’s correlation test (Figure 7b–e).
for bacterial communities. Within these groups, 3 pairs PCA demonstrates the overarching differences in the
of samples—24 with 118, 130 with 542, and 506 with sourdough-fermented samples in comparison to the yeast
521—were shown to cluster together across all clustering control and each other. Sourdough samples were first
analyses associated with this study. compared alongside the yeast control (Figure 7b,c). Loaf
The physical and chemical features driving differentia- volume, pH, titratable acidity, and lactic acid concentra-
tion of the 20 sourdough samples used for dough and baked tion differentiated the control from the sourdough samples
1424 SOURDOUGH MICROBIOMES AND BREAD QUALITY

F I G U R E 7 Clustering and comparisons of 20 sourdough starter cultures based on physical and chemical properties of dough and baked
breads. (a) A cluster dendrogram was constructed based on Euclidean distance calculated using the Ward algorithm (ward.d2). Groups
represent distance between individual samples displayed as height (branch length), unique sourdough starter cultures are labeled as
designated sample ID compared to yeast only (designated ‘0′). k = 6 was used for clustering of discrete groups. Principal component analysis
(PCA) (b) loadings and (c) scores for samples including the yeast control and PCA (d) loadings and (e) scores for samples without the yeast
control demonstrate overarching differences between all sourdoughs in comparison to the control versus the characteristics which
differentiate sourdoughs among one another.

along PC1, whereas sourdough samples were differenti- (p ≤ 0.05). Positive correlations were identified between
ated across PC2 on the basis of acetic acid concentration, titratable acidity and lactic acid (p ≤ 0.05) and fermen-
crust color, fermentation quotient, and water activity tation quotient and lactic acid (p ≤ 0.001). The only
(Figure 7b). In the presence of the yeast control, sour- relationship identified between dough chemistry and the
dough samples appeared to cluster on the basis of the physical characteristics of bread was an inverse rela-
quality parameters included in PC2, denoting the strong tionship between acetic acid concentration and crumb
influence of pH and loaf volume in the differentiation firmness (p ≤ 0.05).
of the control from the sourdough samples (Figure 7c). Alpha diversity indices of bacteria and fungi including
When analyzed in the absence of the yeast control, the Chao1, seChao1, Shannon, and InvSimpson were com-
influence of water activity, free amino acid content, loaf pared to each physical and chemical output (Figure 8a,
volume, and crust color were reduced in comparison to Figures S1 and S3). Bacterial diversity was positively corre-
crumb compressibility and all parameters related to acid- lated with lactic acid production (Shannon, p ≤ 0.05), free
ity (Figure 7d). Additionally, the clustering of samples that amino acid concentration (Shannon, p ≤ 0.05; InvSimpson,
was observed in analysis alongside the yeast control is no p ≤ 0.01), and crumb firmness (Chao1, p ≤ 0.05; Shan-
longer observed (Figure 7e), suggesting that the micro- non, p ≤ 0.05). Inverse correlations between loaf volume
bial diversity of the starters used in each fermentation and bacterial diversity were identified on the basis of Shan-
drives diversity in functional outcomes related to dough non (p ≤ 0.01) and InvSimpson (p ≤ 0.01). Fungal richness
and bread quality. within each sample was significantly correlated with only
The underlying relationships between sourdough starter one measured output, pH (Chao1, p ≤ 0.05). No significant
culture microbiota and the physical and chemical char- correlations between fungal alpha diversity as measured
acteristics of the bread they are used to produce were by Shannon diversity index and physical and chemical
investigated using Pearson’s r correlations between each properties were detected (Figure S4).
measured outcome and bacterial and fungal alpha diver- In identifying relationships between specific micro-
sity (Figure 8a) as well as the 20 most abundant bacterial bial genera and the physical and chemical features of
and fungal genera across sourdough samples (Figure 8b,c). sourdough-fermented dough and bread, the contributions
Unsurprisingly, significant inverse correlations were found of bacteria over fungi were again apparent (Figure 8b,c).
between pH and titratable acidity (p ≤ 0.01), acetic acid Within the top 20 most abundant bacterial genera, 12
and fermentation quotient (p ≤ 0.001). An inverse corre- demonstrated significant correlations with 8 of the 10 mea-
lation was also identified between lactic and acetic acid sured physical and chemical features, in comparison to 7
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1425

F I G U R E 8 Comparison of physical and chemical features of dough and bread made from 20 sourdough starter cultures. (a) Correlations
between physical/chemical characteristics of sourdough-fermented dough/bread and starter culture alpha diversity demonstrate the
influence of bacterial diversity on certain physical and chemical parameters in sourdough-fermented dough and bread. Heatmaps
demonstrate correlations between abundance of individual genera of (b) bacteria and (c) fungi (c) with physical and chemical features of
dough and bread (* = p < 0.05, ** = p < 0.01, *** = p < 0.001). Genera are presented in descending order or overall abundance across all 20
samples. Note that the group labeled Lactobacillus comprises 25 distinct genera including but not limited to Fructilactobacillus,
Furfurilactobacillus, Secundilactobcillus, Schleiferilactobacillus, Levilactobacillus, and Lentilactobacillus.

fungal genera correlating with 7 of the 10 measured physi- 4 CONCLUSION

cal and chemical features. Some correlations also appeared
to be complementary—Aw and lactic acid concentrations Others have shown that variation in microbial composition
demonstrated significant correlations with specific fungi of sourdough starters has been correlated with functional
genera and not bacteria, whereas pH, TA, and crumb firm- properties of bread (Landis et al., 2021; Rizzello et al.,
ness were correlated with specific bacterial genera but not 2015; Winters et al., 2019), but the impact of microbiome
fungal genera, resulting in all physical and chemical fea- diversity on food quality is not well understood. Here,
tures appearing to be related at least in part to the relative we demonstrate that 20 unique sourdough starters dif-
abundance of at least one genus of bacteria or fungi. ferentially impact physical and chemical parameters of
Multiple bacterial genera but no fungal genera were dough and baked bread compared to a yeast only control.
associated with different features. For example, free amino These findings provide a foundation for further inves-
acid concentration was inversely associated with the tigation of relationships between sourdough community
abundance of Gluconobacter, Acinetobacter, and Chry- structure, bread nutrient profile, and quality outcomes
seobacterium. Though quantification of free amino acids related to safety, sustainability, and nutrition. Although
was initially selected as a measurement of proteolysis many studies have taken a reductionist approach by iso-
over the course of fermentation, this inverse relation- lating and adding back specific species such as LAB to
ship is likely demonstrative of differences in metabolic identify functions associated with a particular strain, this
requirements of organisms found in starter cultures (Gob- study was motivated by the need to examine microorgan-
betti et al., 1994). The few direct correlations between isms that comprise sourdough starter cultures as whole
bacterial genera and quality measures in combination communities rather than isolated strains. Despite the
with our findings related to bacterial diversity underscore greater alpha diversity of fungal communities compared
the importance of studying whole microbial communi- to bacterial communities within the sourdough ecosys-
ties rather than attributing functionality to single strains tem, bacterial communities were shown to have a greater
of sourdough starter culture microorganisms. Functional effect on measured physical and chemical properties. This
redundancy has been observed in synthetic sourdough study establishes a foundation for future metagenomic
starter cultures made from 5–8 species of bacteria and 1–2 and metatranscriptomic studies to identify and charac-
species of yeast where dominant, subdominant, and satel- terize specific microorganisms and their associated func-
lite species maintain community resilience and stability tional byproducts as they relate to food quality. Increased
(Calabrese et al., 2022). understanding of fermented food microbiomes such as
1426 SOURDOUGH MICROBIOMES AND BREAD QUALITY

sourdough can be leveraged as a strategy for creating bread Arora, K., Ameur, H., Polo, A., di Cagno, R., Rizzello, C. G.,
products with tailored quality and health benefits. & Gobbetti, M. (2021). Thirty years of knowledge on sour-
dough fermentation: A systematic review. Trends in Food Sci-
ence & Technology, 108, 71–83. https://doi.org/10.1016/j.tifs.2020.12.
AU T H O R CO N T R I B U T I O N S
008
Caitlin S. Clark: Writing―review and editing; inves-
Calabrese, F. M., Ameur, H., Nikoloudaki, O., Celano, G., Vacca,
tigation; formal analysis; writing―original draft; M., Junior, W. J., Manzari, C., Vertè, F., di Cagno, R., Pesole, G.,
methodology. Ashley Ohstrom: Investigation; for- de Angelis, M., & Gobbetti, M. (2022). Metabolic framework of
mal analysis; writing―review and editing; methodology; spontaneous and synthetic sourdough metacommunities to reveal
writing―original draft. M. Laura Rolon: Formal analysis; microbial players responsible for resilience and performance.
writing―review and editing. Molly Smith: Investiga- Microbiome, 10, 148. https://doi.org/10.1186/s40168-022-01301-3
tion; writing―review and editing. Benjamin E. Wolfe: Callahan, B. J., McMurdie, P. J., Rosen, M. J., Han, A. W., Johnson, A.
J., & Holmes, S. P. (2016). DADA2: High resolution sample infer-
Resources; writing―review and editing. Josephine Wee:
ence from Illumina amplicon data. Nature Methods, 13, 581–583.
Conceptualization; investigation; formal analysis; funding https://doi.org/10.1038/nmeth.3869.DADA2
acquisition; project administration; writing―original Corsetti, A., Gobbetti, M., de Marco, B., Balestrieri, F., Paoletti, F.,
draft; writing―review and editing; methodology; Russi, L., & Rossi, J. (2000). Combined effect of sourdough lactic
supervision; resources. Charlene B. Van Buiten: acid bacteria and additives bread firmness and staling. Journal of
Conceptualization; investigation; formal analysis; Agricultural and Food Chemistry, 48, 3044–3051. https://doi.org/
funding acquisition; project administration; writ- 10.1021/jf990853e
de Angelis, M., di Cagno, R., Gallo, G., Curci, M., Siragusa,
ing―original draft; writing―review and editing;
S., Crecchio, C., Parente, E., & Gobbetti, M. (2007). Molec-
resources; methodology; supervision.
ular and functional characterization of Lactobacillus sanfran-
ciscensis strains isolated from sourdoughs. International Jour-
C O N F L I C T O F I N T E R E S T S TAT E M E N T nal of Food Microbiology, 114, 69–82. https://doi.org/10.1016/j.
The authors do not have any conflicts of interest to declare. ijfoodmicro.2006.10.036
De Luca, L., Aiello, A., Pizzolongo, F., Blaiotta, G., Aponte, M., &
D A T A AVA I L A B I L I T Y S T A T E M E N T Romano, R. (2021). Volatile organic compounds in breads prepared
All sequence data (amplicon sequence data for 16S and with different sourdoughs. Applied Sciences, 11, 1330. https://doi.
org/10.3390/app11031330
ITS) were accessed from their original data repository
De Vuyst, L., Van Kerrebroeck, S., & Leroy, F. (2017). Microbial ecol-
(NCBI, BioProject PRJNA589612). The metadata presented
ogy and process technology of sourdough fermentation. Advances
in the figures of this study are available upon request from in Applied Microbiology, 100, 49–160. https://doi.org/10.1016/bs.
the corresponding authors. aambs.2017.02.003
Friedman, M. (2004). Applications of the ninhydrin reaction for anal-
F U N D I N G I N F O R M AT I O N ysis of amino acids, peptides, and proteins to agricultural and
This research was supported by the intramural research biomedical sciences. Journal of Agricultural and Food Chemistry,
program of the U.S. Department of Agriculture, National 52, 385–406. https://doi.org/10.1021/jf030490p
Gobbetti, M., Corsetti, A., & Rossi, J. (1994). The sourdough
Institute of Food and Agriculture (NIFA), Agriculture and
microflora. Interactions between lactic acid bacteria and yeasts:
Food Research Initiative [Grant no. 2023-67017-40251] to Metabolism of amino acids. World Journal of Microbiology and
C.V.B and J.W. C.V.B is also supported by the USDA NIFA Biotechnology, 10, 275–279. https://doi.org/10.1007/BF00414862
Hatch project [Grant no. 7004809] and J.W. is supported by Gobbetti, M., Minervini, F., Pontonio, E., di Cagno, R., & de Angelis,
the Hatch Appropriations under Project no. PEN04699 and M. (2016). Drivers for the establishment and composition of the
Accession no. 1019351. sourdough lactic acid bacteria biota. International Journal of Food
Microbiology, 239, 3–18. https://doi.org/10.1016/j.ijfoodmicro.2016.
05.022
ORCID
Landis, E. A., Oliverio, A. M., McKenney, E. A., Nichols, L. M.,
Ashley Ohstrom https://orcid.org/0009-0007-7678-1746
Kfoury, N., Biango-Daniels, M., Shell, L. K., Madden, A. A.,
Charlene B. Van Buiten https://orcid.org/0000-0001- Shapiro, L., Sakunala, S., Drake, K., Robbat, A., Booker, M., Dunn,
9613-9100 R. R., Fierer, N., & Wolfe, B. E. (2021). The diversity and function
of sourdough starter microbiomes. Elife, 10, 1–24. https://doi.org/
REFERENCES 10.7554/eLife.61644
AACC International. (2022). AACC approved methods of analysis Minervini, F., Dinardo, F. R., Celano, G., de Angelis, M., & Gobbetti,
(11th ed.). AACC International. M. (2018). Lactic acid bacterium population dynamics in artisan
Arendt, E. K., Ryan, L. A. M., & dal Bello, F. (2007). Impact of sour- sourdoughs over one year of daily propagations is mainly driven by
dough on the texture of bread. Food Microbiology, 24, 165–174. flour microbiota and nutrients. Frontiers in Microbiology, 9, 1984.
https://doi.org/10.1016/j.fm.2006.07.011 https://doi.org/10.3389/fmicb.2018.01984
SOURDOUGH MICROBIOMES AND BREAD QUALITY 1427

Nilsson, R. H., Larsson, K. H., Taylor, A. F. S., Bengtsson-Palme, J., Troadec, R., Nestora, S., Niquet-Léridon, C., Marier, D., Jacolot, P.,
Jeppesen, T. S., Schigel, D., Kennedy, P., Picard, K., Glöckner, F. Sarron, E., Regnault, S., Anton, P. M., & Jouquand, C. (2022). Effect
O., Tedersoo, L., Saar, I., Kõljalg, U., & Abarenkov, K. (2019). The of leavening agent on Maillard reaction and the bifidogenic effect
UNITE database for molecular identification of fungi: Handling of traditional French bread. Food Chemistry, 393, 133387. https://
dark taxa and parallel taxonomic classifications. Nucleic Acids doi.org/10.1016/j.foodchem.2022.133387
Research, 47, D259–D264. https://doi.org/10.1093/nar/gky1022 Verdonck, C., de Bondt, Y., Pradal, I., Bautil, A., Langenaeken, N. A.,
Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Brijs, K., Goos, P., de Vuyst, L., & Courtin, C. M. (2023). Impact
Peplies, J., & Glöckner, F. O. (2013). The SILVA ribosomal RNA of process parameters on the specific volume of wholemeal wheat
gene database project: Improved data processing and web-based bread made using sourdough- and Baker’s yeast-based leavening
tools. Nucleic Acids Research, 41, 590–596. https://doi.org/10.1093/ strategies. International Journal of Food Microbiology, 396, 110193.
nar/gks1219 https://doi.org/10.1016/j.ijfoodmicro.2023.110193
R Core Team. (2021). R: A language and environment for statistical Wickham, H. (2016). ggplot2: Elegant graphics for data analysis.
computing. The R Project for Statistical Computing. Springer.
Rizzello, C. G., Cavoski, I., Turk, J., Ercolini, D., Nionelli, L., Winters, M., Panayotides, D., Bayrak, M., Rémont, G., Viejo, C. G.,
Pontonio, E., de Angelis, M., de Filippis, F., Gobbetti, M., & di Liu, D., Le, B., Liu, Y., Luo, J., Zhang, P., & Howell, K. (2019).
Cagno, R. (2015). Organic cultivation of Triticum turgidum subsp. Defined co-cultures of yeast and bacteria modify the aroma, crumb
Durum is reflected in the flour-sourdough fermentation-bread and sensory properties of bread. Journal of Applied Microbiology,
axis. Applied and Environmental Microbiology, 81, 3192–3204. 127, 778–793. https://doi.org/10.1111/jam.14349
https://doi.org/10.1128/AEM.04161-14 Yıltırak, S., Kocadağlı, T., Evrim Çelik, E., Ozkaynak Kanmaz, E.,
Rocken, W., Rick, M., & Reinkemeier, M. (1992). Controlled pro- & Gökmen, V. (2022). Effects of sprouting and fermentation on
duction of acetic acid in wheat sour doughs. Zeitschrift Fuer the formation of Maillard reaction products in different cereals
Lebensmittel-Untersuchung Und – Forschung, 195, 259–263. https:// heated as wholemeal. Food Chemistry, 389, 133075. https://doi.org/
doi.org/10.1007/BF01202806 10.1016/j.foodchem.2022.133075
Schoch, C. L., Ciufo, S., Domrachev, M., Hotton, C. L., Kannan, S., Zheng, J., Wittouck, S., Salvetti, E., Franz, C. M. A. P., Harris,
Khovanskaya, R., Leipe, D., Mcveigh, R., O’Neill, K., Robbertse, H. M. B., Mattarelli, P., O’Toole, P. W., Pot, B., Vandamme, P.,
B., Sharma, S., Soussov, V., Sullivan, J. P., Sun, L., Turner, S., Walter, J., Watanabe, K., Wuyts, S., Felis, G. E., Gänzle, M. G., &
& Karsch-Mizrachi, I. (2020). NCBI taxonomy: A comprehensive Lebeer, S. (2020). A taxonomic note on the genus Lactobacillus:
update on curation, resources and tools. Database, 2020, baaa062. Description of 23 novel genera, emended description of the genus
https://doi.org/10.1093/database/baaa062 Lactobacillus Beijerinck 1901, and union of Lactobacillaceae and
Shen, Y., Tebben, L., Chen, G., & Li, Y. (2019). Effect of amino acids on Leuconostocaceae. International Journal of Systematic and Evolu-
Maillard reaction product formation and total antioxidant capac- tionary Microbiology, 70, 2782–2858. https://doi.org/10.1099/ijsem.
ity in white pan bread. International Journal of Food Science and 0.004107
Technology, 54, 1372–1380. https://doi.org/10.1111/ijfs.14027
Siepmann, F. B., Sousa de Almeida, B., Waszczynskyj, N., & Spier, M.
R. (2019). Influence of temperature and of starter culture on bio- S U P P O RT I N G I N F O R M AT I O N
chemical characteristics and the aromatic compounds evolution Additional supporting information can be found online
on type II sourdough and wheat bread. Lebensmittel-Wissenschaft in the Supporting Information section at the end of this
& Technologie, 108, 199–206. https://doi.org/10.1016/j.lwt.2019.03. article.
065
Suhr, K. I., & Nielsen, P. V. (2004). Effect of weak acid preservatives
on growth of bakery product spoilage fungi at different water activ-
ities and pH values. International Journal of Food Microbiology, 95, How to cite this article: Clark, C. S., Ohstrom,
67–78. https://doi.org/10.1016/j.ijfoodmicro.2004.02.004 A., Rolon, M. L., Smith, M., Wolfe, B. E., Wee, J., &
Thiele, C., Grassl, S., & Gänzle, M. (2004). Gluten hydrolysis and
Van Buiten, C. B. (2024). Sourdough starter culture
depolymerization during sourdough fermentation. Journal of
microbiomes influence physical and chemical
Agricultural and Food Chemistry, 52, 1307–1314. https://doi.org/10.
1021/jf034470z properties of wheat bread. Journal of Food Science,
89, 1414–1427.
https://doi.org/10.1111/1750-3841.16957