Taxonomy and Biodiversity of Sourdough Yeasts and Lactic Acid Bacteria

Abstract

This chapter describes the taxonomy and biodiversity of yeasts and lactic acid bacteria (LAB) isolated from sourdoughs collected around the world. The phylogenetic position of yeast and LAB species found in sourdoughs is discussed, and important problems in taxonomic nomenclature of typical sourdough-associated species are highlighted. An overview is given of factors that may influence the taxonomic composition of sourdough ecosystems, such as geographical location, type of cereals and other raw materials, and technological process parameters. A brief history is provided of the most commonly used approaches to isolate yeast and LAB strains from sourdough samples. Finally, the main techniques for identification of sourdough yeasts and LAB are discussed, emphasizing both on conventional culture-based methods as well as on culture-independent molecular approaches.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

5.1 Taxonomy of Sourdough Yeasts and Lactic Acid Bacteria

5.1.1 Taxonomy of Sourdough Yeasts

Yeasts are microscopic fungi that undergo typical vegetative growth by budding or fission resulting in an unicellular appearance and a sexual reproduction without a within or upon which the resulting spores are formed [1]. From the agro-alimentary and scientific point of view, yeasts are among the most important eukaryotes. Yeast species found in sourdough microbial communities share an adaptation to the specific and stressful environment created mainly by a low pH, high carbohydrate concentrations and high cell densities of lactic acid bacteria (LAB). Such adaptations can be found in species located mostly on one branch of the ­evolutionary tree of fungi, which accommodates the ascomyceteous yeasts. Within this branch, recognized by the current classification in the phylum Ascomycota, as the subphylum Saccharomycotina, class Saccharomycetes, order Saccharomycetales [2], sourdough yeasts belong to different genera. The major sourdough yeasts belong to genera that are currently placed in the family Saccharomycetaceae, although family assignment of yeast genera is still difficult because of a lack of informative data. Basidiomycetous yeasts and dimorphic ascomycetes, also adapted to growth in liquid environments by unicellular growth forms, lack the fermentative abilities that are common to the Saccharomycetales and that are important for growth under oxygen limitations. The taxonomy of the Saccharomycetales, classically based on morphology and physiology, is in the process of being adapted to the increasing knowledge of evolutionary relationships reconstructed from gene sequences, in other words, a phylogenetic system of classification is being developed [2]. This implies a number of name changes. The new genus names have the advantage of reflecting the common genetic background of related yeast species, hereby providing an informative classification in contrast to the former largely artificial classification.

An overview of recent name changes restricted to species that have been obtained from sourdough is given here. The name changes most relevant to the yeasts found in sourdough concern the genera Saccharomyces Meyen ex Reess and Pichia E.C. Hansen emend. Kurtzman. The genus Saccharomyces has been limited to the group of species known as Saccharomyces sensu stricto, including the type species of the genus, Saccharomyces cerevisiae on the basis of multiple gene sequences [3]. The group of species formerly often addressed as Saccharomyces sensu lato has been divided into several genera. The new genus Kazachstania is accommodating the former Saccharomyces exiguus, Saccharomyces unisporus and Saccharomyces barnettii as Kazachstania exigua, Kazachstania unispora, and Kazachstania barnettii, respectively. Saccharomyces kluyveri has been assigned to the new genus Lachancea as Lachancea kluyveri. The genus Pichia has been restricted to species closely related to the generic type species Pichia membranifaciens, including Pichia fermentans [4]. The former genus Issatchenkia has been integrated into the newly defined genus Pichia as its species are located on the same branch as the type species P. membranifaciens on the phylogenetic tree based on multiple gene sequences used for the redefinition of genera. While the species epithet of Issatchenkia occidentalis has been preserved in its new name Pichia occidentalis, a complete name change of Issatchenkia orientalis to Pichia kudriavzevii was necessary as the combination Pichia orientalis had been used for a different species before. The former species Pichia anomala and Pichia subpelliculosa were found to be only distantly related to the generic type species P. membranifaciens and have therefore been assigned to the newly created genus Wickerhamomyces as Wickerhamomyces anomalus and Wickerhamomyces subpelliculosus. A review of the taxonomic considerations including the earlier genus name Hansenula has been given by Kurtzman [5, 6]. Other, only occasionally from sourdough isolated species from the former genus Pichia have been reassigned to new genera, while preserving their species epithet and include Kodamaea ohmeri, Meyerozyma guilliermondii, Millerozyma farinosa, Ogataea polymorpha, Saturnispora saitoi and Scheffersomyces stipitis.

To determine the entities, or taxa,¹ that deserve attention in the sourdough ­context, about 40 original publications were reviewed and the repeatedly reported species are listed in Table 5.1. These publications span the time from the early 1970s until present and it is obvious that not all of them are based on identification techniques that are currently considered as the most accurate. However, most of the six regularly (seven or more reports) encountered species S. cerevisiae (syn.² S. fructuum); Candida humilis (syn. Candida milleri); P. kudriavzevii (syn. I. orientalis, anamorph Candida krusei); K. exigua (syn. S. exiguus, anamorph Candida [Torulopsis] holmii); Torulaspora delbrueckii, anamorph Candida colliculosa; and W. anomalus (syn. P. anomala, Hansenula anomala, anamorph Candida pelliculosa) can be distinguished from each other reasonably well by classical methods based on morphology and physiology as used in the reviewed literature up to the late 1990s. However, comparing studies using phenotypic identification techniques (n  =  19) with those using DNA-based techniques (n  =  23), the incidence of C. humilis has increased markedly in the DNA-based studies, probably at the cost of a decreased frequency of detecting K. exigua. These two phylogenetically closely related species may be mistaken for each other if using phenotypical identification methods. For example, originally S. exiguus was reported from San Francisco sourdough [7], while some strains from this study that were preserved in culture collections, later served for the description of C. milleri, currently a synonym of C. humilis. The sourdough isolate M14 reported as S. exiguus and deposited also as CBS 7901 was suggested to belong to C. humilis after DNA-based analyses [29, 50]. The regularity with which S. cerevisiae, C. humilis P. kudriavzevii, K. exigua, T. delbrueckii, and W. anomalus are encountered in sourdough is an indication of their common association with this substrate. This is not necessarily an exclusive association with sourdoughs and some of the species have to be considered as generalists, able to thrive in a wide range of environmental conditions, as for example W. anomalus [51], while a specific ecological niche of the most frequently encountered species S. cerevisiae has been elucidated [52]. Other species are less frequently detected in sourdough, namely Candida glabrata; P. membranifaciens (anamorph Candida valida); Candida parapsilosis; Candida tropicalis; Candida stellata (syn. Torulopsis stellata); K. unispora (syn. S. unisporus, Torulopsis unisporus); Kluyveromyces marxianus (anamorph Candida kefyr); M. guilliermondii (syn. Pichia guilliermondii, anamorph Candida guilliermondii); and Saccharomyces pastorianus. Finally, 14 species have been mentioned only in single reports and may be considered as rather transient or present fortuitously in the sourdough ecosystem (Table 5.2).

Table 5.1 Overview of original studies that analyzed the yeast diversity of sourdoughs. Only species repeatedly obtained from sourdough are listed. Numbers in species columns indicate the number of obtained isolates. These numbers depend on the isolation method (i.e., one or multiple representatives of the same colony morphotype) and are mostly of value to judge species abundance within a given study. An “x” was used if the isolate number was not specified. Multiple x’s indicate the relative abundance of a detected species if mentioned. Studies that report on more than one sourdough may include sourdoughs of differing yeast species composition. However, a more detailed assignment of yeast species to single sourdoughs was not possible as not all studies reported the necessary data. Synonyms or asporogenic forms (anamorphs) are listed only if used in the cited references or if resulting from recent taxonomic changes

Table 5.2 Yeast species isolated from sourdough mentioned by single reports. None of these reports refer to these species as the sole or dominating yeast found in sourdough. Species considered as rare contaminants (e.g., Schizosaccharomyces pombe, Rhodotorula glutinis, R. mucilaginosa, Endomycopsis fibulinger) or reported as unidentified were not included

The species complex C. humilis/C. milleri frequently detected in sourdough deserves a detailed elucidation. In terms of classification it was placed in the artificial genus Candida, because no sexual reproduction could be observed. Phylogenetically, C. humilis/C. milleri belongs to the same group as the genus Kazachstania [3]. It has, however, not yet been taxonomically placed in this genus, as the phylogenetic reclassification is treating sexually reproducing taxa with priority. The species C. humilis has been described based on a yeast strain associated with South African bantu beer, made from kaffir corn (Sorghum caffrorum) or finger millet (Eleusine coracana) [53]. The species C. milleri has been described to accommodate yeast strains isolated from San Francisco sourdough fermentations and initially assigned to S. exiguus [7, 49]. The basis for this reassignment were significantly higher guanine-plus-cytosine contents in selected San Franscisco sourdough strains compared to the type and other reference strains of S. exiguus and growth stimulation of C. milleri by calcium pantothenate. C. humilis and C. milleri were indicated to be conspecific based on their identical D1/D2 region large subunit (LSU) ribosomal DNA (rDNA) sequences, a locus that in rare cases may not suffice to resolve closely related species [54, 55]. DNA-DNA reassociation of at least 90% between the C. milleri type strain CBS 6897 and strain CBS 2664, the type strain of T. holmii var. acidilactici, a synonym of C. milleri, are of interest in this situation and confirms the conspecificity of these strains [54, 56]. Strain CBS 2664 shows ten substitutions in the internal transcribed spacer (ITS) regions of the rDNA, if compared to the type strain of C. milleri, indicating the degree of ITS divergence within this species. In the sourdough context, C. humilis is often distinguished from its synonym C. milleri by targeting the ITS rDNA region [27]. This is done by restriction fragment length polymorphisms (RFLPs) of the ITS generated by the restriction enzyme HaeIII that is recognizing and cutting the nucleotide sequence GGCC. This site is made unrecognizable by a single nucleotide change from C to T (resulting in GGCT) in a population represented by the C. milleri type strain. As a result of this single nucleotide substitution the RFLP analysis shows two fragments, in contrast to three fragments for strains with the intact recognition site. A comparison of relevant publicly available ITS sequences (CBS 5658: AY046174, CBS 6897: AY188851, SY13: DQ104399; CBS 2664 and CBS 7541: yeast database at www.cbs.knaw.nl/) shows the transitional position of some strains between the type strain sequences of C. humilis and C. milleri, especially CBS 7541, indicative of a continuum of ITS sequence variants between both type strains. Therefore, these two taxa might be best considered as populations of one species. Eventual heterogeneity of the species C. humilis, especially of applied value, should be documented and accompanied by the deposition of the isolates in public culture collections for further study.

5.1.2 Taxonomy of Sourdough Lactic Acid Bacteria

LAB comprise a heterogeneous group of Gram-positive, nonsporulating, strictly fermentative lactic acid-producing bacteria that play an important role in the organoleptic, health-promoting, technological, and safety aspects of various fermented food products. As a result of natural contamination through the flour or the environment or by deliberate introduction via dough ingredients, a wide taxonomic range of LAB has also been found in sourdoughs. In sourdough environments, LAB live in association with yeasts and are generally considered to contribute most to the process of dough acidification, while yeasts are primarily responsible for the leavening. Although also obligately homofermentative LAB have been isolated from sourdoughs, obligately or facultatively heterofermentative LAB species have the best potential and competitiveness to survive and grow in this particular food environment [57, 58].

Initially, classification of LAB was based on morphology, ecology, and physiological characteristics [59]. At a later stage, also chemotaxonomical properties such as cellular fatty acid and cell wall composition were included. In LAB as well as in many other bacterial groups, phenotypic characters are often limited in their taxonomic usefulness for discrimination of closely related species and suffer from poor interlaboratory exchangeability. As a result, differentiation of LAB solely based on phenotypic traits is generally only considered reliable at the genus level. The introduction of DNA-based techniques such as genomic mol % GC, DNA-DNA hybridization, and sequencing of ribosomal RNA (rRNA) genes has brought significant changes to LAB taxonomy [60–62]. Especially the use of rRNA gene sequences as evolutionary chronometers has allowed the elucidation of phylogenetic relationships between LAB species. As a result, comparison of 16S rRNA gene sequences with sequences in public online databases has become a standard approach for identification of unknown LAB isolates. However, the low evolutionary rate of ribosomal genes may compromise differentiation between LAB species exhibiting identical or nearly identical 16S rRNA gene sequences [63–66]. Alternatively, the use of multiple housekeeping genes encoding essential cellular functions has been proposed for sequence-based identification of LAB. For instance, classification of Lactobacillus species based on sequence analysis of the housekeeping genes pheS and rpoA proved to be highly congruent with 16S rRNA gene phylogeny [67].

The LAB species diversity associated with sourdoughs has been reviewed by several authors in recent years [57, 58, 68–70]. On the basis of these reviews, an updated overview of the LAB species most commonly found in fermented sourdough is compiled in Table 5.3. As is the case for many food ecosystems, this overview again highlights that lactobacilli are by far the most frequently recovered LAB species from sourdough ecosystems. The taxonomy of the genus Lactobacillus is extremely complex; according to the April 2011 update, at least 171 species names have so far been proposed in this genus (www.bacterio.cict.fr/l/lactobacillus.html). However, as several of the proposed species names have meanwhile been synonymized, the actual number of phylogenetically unique species is lower. In sourdoughs, more than 55 Lactobacillus species have been identified, of which the large majority are obligately heterofermentative (Table 5.3). Given the taxonomic complexity of this genus, accurate identification of unknown Lactobacillus isolates requires specific expertise, for example the use of methods that offer sufficient taxonomic resolution and the correct interpretation of identification results by comparison with complete and up-to-date databases. In most of the older studies, however, identification of lactobacilli mainly or even exclusively relied on phenotypic approaches with limited taxonomic resolution at species level. Therefore, it is safe to assume that some of the Lactobacillus species previously reported in sourdough environments may have been incorrectly identified at the species or even at the genus level. A typical example is the taxonomic situation in the Lactobacillus plantarum group where discrimination between the ubiquitous sourdough bacterium Lb. plantarum and the phylogenetically highly related Lactobacillus paraplantarum and Lactobacillus pentosus may be problematic when identification methods with insufficient taxonomic resolution are used. In this regard, Torriani and colleagues [71] were among the first to suggest that sequences of housekeeping genes such as recA rather than 16S rRNA gene sequences are recommended to distinguish between members of this phylogenetically tight species group. Likewise, several sourdough isolates initially assigned to Lactobacillus alimentarius may in fact belong to the later described and closely related Lactobacillus paralimentarius due to phenotypic misidentification [72]. Also in this case, it has been shown that molecular fingerprint- or sequence-based methods are required to differentiate between both species [67 73]. In Lactobacillus rossiae, a remarkable intraspecific heterogeneity leading to the identification of several subspecific clusters based on pheS gene sequencing may complicate unambiguous identification of Lb. rossiae strains [77]. Finally, nomenclatural issues may also be a cause for taxonomic confusion. Corrections of originally misspelled specific epithets, such as “Lactobacillus sanfrancisco” [75] (now Lactobacillus sanfranciscensis) and “Lactobacillus rossii” [76] (now Lb. rossiae), and the synonymization of species, such as the recognition of Lactobacillus suntoryeus as a synonym of Lactobacillus helveticus [77], can take a while to be introduced in subsequent taxonomic literature.

Table 5.3 LAB species generally associated with sourdough fermentation or found in fermented sourdoughs^a

Triggered by the introduction of molecular DNA-based taxonomic methods in sourdough microbiology and the growing number of 16S rRNA gene sequences in public databases, in recent years many new Lactobacillus species have been described which were first isolated from a sourdough environment. Since 2000, 13 new Lactobacillus species originally isolated from sourdoughs have been proposed, i.e., Lactobacillus frumenti [78], Lactobacillus mindensis [79], Lactobacillus spicheri [80], Lactobacillus acidifarinae [81], Lactobacillus zymae [81], Lactobacillus hammesii [82], Lb. rossiae [76], Lactobacillus siliginis [83], Lactobacillus nantensis [84], Lactobacillus secaliphilus [85], Lactobacillus crustorum [86], Lactobacillus namurensis [87], and Lactobacillus nodensis [88] (Table 5.3). However, many of these species have only been reported once or are rarely isolated from this type of fermented food, and are represented by only a few strains. Therefore, it is not clear which of these species are really typical for sourdough environments and if so, what their geographical distribution is. In fact, only a few Lactobacillus species such as the obligately heterofermentative Lb. sanfranciscensis and the facultatively heterofermentative Lb. paralimentarius seem to be optimally adapted to this specific environment and are rarely isolated from other sources. Other heterofermentative species such as Lactobacillus brevis and the facultatively heterofermentative Lb. plantarum are also frequently isolated from fermented sourdoughs, but have also been found in many other food and nonfood environments [69].

Although the LAB microbiota of sourdoughs is clearly dominated by lactobacilli, other less predominant or subdominant LAB species may also be found, including members of the genera Weissella, Pediococcus, Leuconostoc, Lactococcus, Enterococcus and Streptococcus (Table 5.3). Of these, specific species of Weissella, Pediococcus and Leuconostoc are particularly well adapted to survive and grow in plant-derived materials [57, 89]. The taxonomy of the latter three genera is much less complex than in Lactobacillus, and their presence in sourdoughs is restricted to only a few species. Weissellas are obligately heterofermentative LAB of which a number of species produce dextran, the best-documented exopolysaccharide formed by heterofermentative LAB. In sourdoughs, the dextran-producing species Weissella cibaria and Weissella confusa are most frequently found. Both taxa are positioned together on one of the four phylogenetic branches in this genus based on 16S rRNA gene analyses [90], but can be differentiated using restriction analysis of the amplified 16S rDNA [91], randomly amplified polymorphic DNA-PCR (RAPD-PCR) [92], and PCR targeting the ribosomal ITS [26]. Within the facultatively heterofermentative pediococci, the species Pediococcus acidilactici and Pediococcus pentosaceus are most commonly found in sourdoughs. Differentiation between these two biochemically and phylogenetically related species can be achieved by fingerprinting methods such as ribotyping [93, 94], restriction analysis of the amplified 16S rDNA (16S-ARDRA) [95], RAPD-PCR [96, 97], and by sequence analyses of the 16S rRNA gene, the ribosomal ITS regions and the heat-shock protein 60 gene [98]. In the obligately heterofermentative genus Leuconostoc, the majority of sourdough isolates so far identified belong to Leuconostoc mesenteroides and Leuconostoc citreum. In the former species, further taxonomic distinction is made at subspecies level between Le. mesenteroides subsp. mesenteroides, Le. mesenteroides subsp. dextranicum and Le. mesenteroides subsp. cremoris (www.bacterio.cict.fr/l/leuconostoc.html). The three subspecies can be separated by RAPD-PCR fingerprinting [99].

5.2 Microbial Species Diversity of Sourdoughs

5.2.1 Influence of Geography

5.2.1.1 The Origin of Sourdough

Historically, sourdough production started as a conditio sine qua non to process cereals for the production of baked goods [100]. Indeed, thousands of years ago the first bread production must have been based on spontaneous wild lactic acid fermentation whether or not associated with yeasts and with little or no leavening. Leavening could not have been very pronounced because of the use of barley (Hordeum vulgare) and ancient grains [such as spelt (Triticum aestivum subsp. spelta), emmer (Triticum turgidum subsp. dicoccum), and kamut (T. turgidum subsp. turanicum)] in early times and no addition of yeast for leavening. This form of flat (sour) bread production is still daily practice in many countries of the world, in particular in African countries and the Middle East. Leavening must have been an accidental discovery when yeasts from the air or the flour were allowed to ferment the cereal dough mixture extensively. However, it was only from the late nineteenth century onwards that yeast starter cultures were introduced for bread production from wheat (T. aestivum and Triticum durum) flour, first by using brewing yeasts as remnants of beer production followed by intentionally produced commercial bakers’ yeast, S. cerevisiae. Consequently, sourdough bread must have been consumed for a long time. Also afterwards, bread production in countries relying on cereals other than wheat such as rye (Secale cereale) had still to be supported by lactic acid fermentation. Rye bread baking requires dough acidification to inhibit the abundant α-amylase in the rye flour and to make rye starch and pentosans more water-retaining to form a good dough texture since not enough gluten is present in rye. Hence, various (rye) breads from Germany, Central European countries, and Scandinavia are based on sourdough. In the USA, sourdough was the main base of bread supply in Northern California during the California Gold Rush, because of the easy way to store it and to keep it active for daily bread production. Today, San Francisco sourdough bread is commercially produced in the San Francisco area and it remains a part of the culture of the San Francisco bay area. In the early 1970s, the responsible sourdough bacterium was identified as Lb. sanfrancisco, now Lb. sanfranciscensis, named according to the area where it was discovered [75]. Notice that this LAB species is actually identical to Lb. brevis var. lindneri (now Lb. sanfranciscensis), which was found to be responsible for various sourdough breads produced in Europe [17]. Also, it was shown that these sourdough LAB species occur in a stable association with the yeasts C. humilis (syn. C. milleri) and K. exigua (syn. S. exiguus), respectively [7, 101]. Nowadays, sourdough is used for its technological (dough processing and bread texture, flavor, and shelf life) and nutritional effects [57, 58, 69, 102, 103]. Moreover, sourdough products are appreciated for their traditional value, gastronomic quality, and natural and healthy status [104].

5.2.1.2 The Origin of Sourdough Variation

Thanks to the fact that craftsmanship has determined bakery practice for a long time, a huge variety of bakery products, in particular those based on sourdough, exists, which may differ considerably from region to region. Most of these products, including breads, cakes, snacks, and pizzas, originate from very old traditions. For instance, in Italy, numerous different types of sourdough breads exist, often called according to the name of the region, such as Altamura bread and Pugliese bread [101]. Also, seasonal varieties exist, which are traditionally produced on the occasion of religious festivities. For instance, Panettone cake in Milan and Pandoro in Verona are made for Christmas, while Colomba is a Milanese cake made for Easter. Alternatively, sourdough-based products such as crackers, French baguettes, and Italian ciabattas are much more common, although the original sourdough recipe has been replaced by the faster growing bakers’ yeast that is in only few cases allowed to ferment longer to enable contaminating LAB to develop for flavor formation (pre-doughs or type 0 sourdoughs). Fortunately, numerous bakery sourdoughs have been kept alive for tens of years through backslopping procedures, i.e., repeated cyclic re-inoculation of a new batch of flour and water from a previous one with a so-called “sour” during refreshment of the flour-water mixture by the baker, thereby assuring the quality of the baked goods produced thereof. It turned out that these traditional sourdoughs harbor a mixture of distinctive yeast and LAB strains, which may be held responsible for the typical organoleptic quality of the breads made thereof, as backslopping results in a prevalence of the best-adapted strains [57, 68, 69]. This diversity of natural sourdough starters likely accounts for the variety of artisan sourdoughs produced by bakeries, whether or not typical for a certain geographical region. Alternatively, sourdough starter cultures, comprised of one or more defined strains, are commercially available now. In addition to Lb. sanfranciscensis strains, commercially available strains of other LAB species include Lb. brevis, Lactobacillus delbrueckii, Lactobacillus fermentum, and Lb. plantarum, albeit that not all strains in use are competitive enough to dominate the sourdough fermentation processes that have to be started up [69]. These starter cultures are used for rapid acidification of the raw materials and flavor formation upon fermentation. Also, industrial manufacturers produce dried sourdough powders that are used as nonliving flavor ingredients in industrial bread production.

5.2.1.3 Region-Specific Sourdoughs and Their Associated Microbiota

Whereas it was initially thought that a relationship could be seen between the presence of certain LAB species and the geographical origin of a particular sourdough, it turned out through systematic and detailed taxonomic investigations that the species diversity of both LAB and yeasts of local sourdoughs has nothing to do with the geography of the sourdough production process (Tables 5.1 and 5.4; 57, 68, 69). For instance, the typical sourdough bacterium of San Francisco sourdough bread of the San Francisco bay area, Lb. sanfranciscensis, has been found in various wheat sourdoughs throughout Europe, and hence its (unique) presence should be ascribed to other factors, which are mainly based on the fermentation technology and practical conditions applied [121, 152–154]. In various countries, such as in Italy, several studies have been focused on region-specific sourdoughs (Table 5.4). However, no clear-cut relationship could be shown between for the region typical sourdoughs and their associated microbiota. In contrast, Italian sourdoughs harbor simple to very complex communities of LAB species depending on the final products examined, among which Lb. brevis, Lb. (par)alimentarius, Lb. plantarum, Lb. sanfranciscensis, Lb. fermentum, P. pentosaceus, and W. confusa are widespread. Similarly, Belgian bakery sourdoughs have been analyzed extensively and are characterized by LAB consortia of Lb. brevis, Lb. hammesii, Lb. nantensis, Lb. paralimentarius, Lb. plantarum, Lb. pontis, Lb. sanfranciscensis, and/or P. pentosaceus [105, 106, 155]. Sourdoughs with both stable large and stable restricted species diversities may occur [106]. Also, Lb. rossiae seems to have a wide distribution in sourdoughs, as has been shown through its isolation from sourdoughs in Central and Southern Italy, Belgium, and elsewhere [74, 105, 106, 156–158]. LAB species are responsible for the acidification of the dough and contribute to flavor formation. Besides LAB species, a large variety of yeast species are found in sourdough ecosystems (Tables 5.1 and 5.2). S. cerevisiae has been found in almost every sourdough study (38 out of 42 reviewed) regardless as to whether or not bakers’ yeast is added (14 studies mention S. cerevisiae, although they indicate that no bakers’ yeast was added). A single sourdough usually harbors only one or two yeast species at a given time, among which C. humilis (and K. exigua) and P. kudriavzevii occur most frequently [40]. Yeasts are responsible for the leavening of the dough and also contribute to flavor formation.

Table 5.4 Nonexhaustive overview of the species diversity of LAB in sourdoughs made from various flours and of different geographical origins. The method of identification of the LAB species is indicated

Single isolations of yeast and LAB species have caused former misinterpretations of their association with certain sourdough-producing regions, not only because of the random isolation itself but also regarding the single habitat (sourdough) explored. Thus, the association of, for instance, Lb. spicheri with German rice sourdoughs [80] might represent accidental discoveries [57, 68, 69]. Instead, the dedicated use of basic raw materials as well as the technological procedures applied rather determines the stability and persistence of the yeast and LAB communities involved in the sourdough fermentation process. Indeed, the presence of Lb. sanfranciscensis in wheat sourdoughs, which was for a long time the sole habitat wherein this LAB species could be found [now, it has been detected in rye sourdoughs (Table 5.4) and the insect gut as well [159]], can be ascribed to its selection by the type of technology applied, i.e., backslopping practices, temperature of incubation of the dough, pH of the dough, and/or microbial interactions [113, 121, 135, 160, 161]. However, the use of certain raw materials, encompassing cereal types and other ingredients such as adjunct carbohydrates, salt, yoghurt, herbs, etc., and operational practices, such as dough yield and refreshment times, may be linked to local traditions, may favor particular microorganisms as a result of trophic and metabolic relationships and interactions (both cooperation and antibiosis), and hence may associate specific LAB and/or yeast species with specific geographical regions. For instance, the use of rye may select for amylase-positive homofermentative Lactobacillus amylovorus, although higher temperatures cause a shift toward the predominance of heterofermentative lactobacilli [120, 162]. The dominance of mainly heterofermentative LAB species in traditional sourdoughs is caused by a highly adapted carbohydrate metabolism, a dedicated amino acid assimilation, and environmental stress responses [57, 58, 68, 163–166]. In particular, maltose, as the most abundant fermentable energy source in dough, is metabolized via the maltose phosphorylase pathway and the pentose phosphate shunt by strictly heterofermentative LAB species such as Lb. sanfranciscensis, Lactobacillus reuteri, and Lb. fermentum. This efficient maltose metabolism coupled to the use of external alternative electron acceptors such as fructose, together with specific pathways such as the arginine deiminase (ADI) pathway, and various environmental stress responses such as response to acidic conditions, increases its competitiveness in the harsh sourdough environment [167–172]. Moreover, maltose-positive LAB species such as Lb. sanfranciscensis often form a stable association with maltose-negative yeast species such as C. humilis, thereby preventing competition for the same carbohydrate sources [58, 164, 165]. Also, the production of specific inhibitory compounds, maintained through backslopping, such as the antibiotic reutericyclin produced by Lb. reuteri, may favor the dominance of this LAB species, as is the case for certain German type II sourdoughs [173].

5.2.2 Influence of Cereals and Other Raw Materials

Cereal flours are not sterile. Their microbiological stability is related to their low water activity. As yeasts and LAB naturally occur on plant materials, cereal flours carry both groups of microorganisms and competitive yeast and LAB species reach numbers above those of the adventitious microbiota upon fermentation of a ­flour-water mixture [57]. However, whether the type of flour mainly directs the growth of sourdough LAB species remains controversial [25, 26, 47, 89, 105, 107, 121, 123, 135]. For instance, whereas it was assumed that rice sourdough fermentation selects for Lb. spicheri [80], this LAB species cannot always be found in rice sourdoughs [123]. This has been ascribed to competitiveness of the microorganisms that are present in the sourdough ecosystem. Indeed, microbial interactions between the spontaneous microbiota and an added sourdough starter culture may lead to the dominance of autochthonous LAB species and/or strains. Among other mechanisms, competitiveness may explain the apparent prevalence of LAB species in specific sourdough preparations, such as evidenced by single reports on Lb. amylovorus in rye sourdoughs [120], Lactobacillus sakei in amaranth sourdoughs [122], and Lb. pontis in teff sourdoughs [47]. Yet, spontaneous sourdough fermentations carried out in the laboratory with flour as the sole nonsterile ingredient indicate that the type and quality (microbiological and nutritional) of the cereal flour used is indeed an important source of autochthonous LAB and yeasts occurring in the ripe sourdoughs [40, 107]. Hence, the flour plays a key role in establishing stable microbial consortia within a short time. In this context, it has been shown that laboratory sourdoughs based on wheat, rye, or spelt, backslopped daily for 10 days at 30 °C, whether or not initiated with a Lb. sanfranciscensis starter culture [as tested in the case of wheat sourdough fermentations in the study of Siragusa et al. [135]], reach an equilibrium of LAB species through a three-step fermentation process: (1) prevalence of sourdough-atypical LAB species (e.g., Enterococcus spp. and Lc. lactis subsp. lactis); (2) prevalence of sourdough-typical LAB species (e.g., species of Lactobacillus, Leuconostoc, Pediococcus, and Weissella); and (3) prevalence of highly adapted sourdough-typical LAB species (e.g., Lb. fermentum and Lb. plantarum) [107–109, 135]. Indeed, it has been shown that the LAB species Lb. fermentum (strictly heterofermentative) and Lb. plantarum (facultatively heterofermentative) dominate several sourdough fermentation processes, irrespective of the type of flour or the addition of starter cultures that are not robust enough [47, 108, 123, 136–138, 142, 172, 174, 175]. Concerning yeasts, C. glabrata and W. anomalus prevail during laboratory sourdough fermentations [40]. Further, it has been shown that previous introduction of flour into the bakery environment helps to build up a so-called house microbiota that serves as an important inoculum for subsequent bakery sourdough fermentations [155]. Indeed, LAB strains adapted to the sourdough and bakery environment (apparatus, air, etc.), which have been shown to be genetically indistinguishable, may be repetitively introduced in consecutive sourdough batches during backslopping. The widespread use of bakers’ yeast may be responsible for the prevalence of S. cerevisiae in bakery sourdoughs [26, 29, 31, 34–36, 39, 40]. However, there are also indications through reliable molecular data of a large strain diversity of S. cerevisiae in single sourdoughs, that suggest an autochthonous wheat flour origin of this yeast species in sourdough too [27, 30, 34]. Supportive of an autochthonous origin of S. cerevisiae is also the presence of this species in rye flour [29]. Yet, during laboratory fermentations with flour as the sole nonsterile ingredient and without added bakers’ yeast, other species such as C. glabrata and W. anomalus emerge [40]. Anyway, the direct environment is another important source of ­(accidental) contamination of the flour by LAB and yeasts. Consequently, hygienic conditions in the sourdough and bakery environments will play a role as well. Finally, microorganisms occurring on cereals and subsequently in sourdoughs may be of intestinal origin, due to fertilization practices on the grain fields, mouse feces or insects in the flour mills, or fecal contamination of the sourdough production environment [70, 175–178]. It may explain the opportunistic presence of Lactobacillus acidophilus, Lactobacillus johnsonii, Lb. reuteri, and Lb. rossiae, which are common gastrointestinal inhabitants.

5.2.3 Influence of Technology

Besides the cereals and other dough ingredients, which are mainly responsible as the source of metabolic activity in the form of flour enzymes and endogenous microorganisms, specific technological process parameters determine the species diversity, number, and metabolic activity of the microorganisms (whether or not added) present in the stable, ripe sourdough. These process parameters include chemical composition and coarseness of the flour, leavening and storage temperature, fermentation time, pH, redox potential, dough yield, refreshment time and number of propagation steps, and interactions between the microorganisms [57, 102, 164, 165, 179].

Different types of sourdough exist, on the basis of the processing conditions and/or technology used for production, with a specific microbiota occurring in each type [57, 180]. Type I or traditional sourdoughs are manufactured by continuous, (daily) backslopping, at ambient temperature (<30 °C), to keep the microorganisms in an active state. Therefore, mother doughs are used as an inoculum for subsequent doughs by addition of the desired amount of dough to a fresh flour-water batch according to defined cycles of preparation. These small-scale sourdough productions are used in traditional (home-made) sourdough bread making. Natural sourdoughs frequently harbor Lb. sanfranciscensis and C. humilis/K. exigua as prevalent LAB and yeast species, respectively. Type II or industrial sourdoughs are produced through one-step propagation processes of long duration (typically 2–5 days) at a fermentation temperature above 30 °C and with high water content. These large-scale sourdough productions result in semifluid preparations, which are used as dough acidifiers or flavor ingredients. Lb. amylovorus, Lb. fermentum, Lb. pontis, and Lb. reuteri are commonly found in type II wheat and rye sourdoughs. Type III sourdoughs are prepared in dried form to be used as nonliving acidifier supplement and flavor carriers for (sourdough) bread production. In contrast to type I doughs, doughs of types II and III require the addition of baker’s yeast for leavening.

Commercially available bulk starter cultures to prepare type II and III sourdoughs aim at standardizing the end products through acidification of and flavor formation in the dough [69, 181, 182]. New trends tend to develop starter cultures that lead to improved functional properties other than acidification and flavor formation, such as texture improvement, antibacterial and antifungal activities, and health-­promoting effects [69, 166]. In this context, strain robustness and fitness towards microbial competitors and environmental conditions should be the driving force in the selection of useful starters for sourdough fermentation processes, as it has been shown that autochthonous strains often emerge [69, 135, 138, 142]. However, studies on the industrial exploitation of sourdough starter cultures are scarce [183, 184]. Recently, the use of starter cultures in type I propagated sourdoughs has been investigated [89, 123, 135, 185]. It is of course well known that the fermentation temperature affects the ratio of lactic acid to acetic acid [185, 187]. In general, homofermentative LAB starter cultures are used at high temperature and for short fermentation times (e.g., 37 °C for 36 h) and heterofermentative LAB starter cultures are used at low temperature and for long fermentation times (e.g., 25 °C for 48 h), resulting in sourdoughs with mainly lactic acid and acetic acid, respectively. However, it would be of great value to know the circumstances for the expression of other functional properties that are of added value to sourdoughs [182].

The fermentation temperature, one of the criteria to distinguish type I and II sourdoughs, is essential for the community dynamics and stability of a sourdough microbiota [29, 160, 175, 179, 185, 188, 189]. For instance, spontaneous wheat sourdough backslopping fermentations (type I) carried out at 23 °C for 10 days select for Le. citreum instead of Lb. fermentum that prevails at 30 °C and 37 °C [175]. Similarly, rye fermentations initiated with commercial sourdough starter cultures maintain the presence of Lb. mindensis and Lb. sanfranciscensis at 25 °C (type I), but select for Lactobacillus crispatus and Lb. pontis at 30 °C and for Lb. crispatus, Lb. frumenti, and Lb. panis at 40 °C (both type II) [184]. Whereas Lb. sanfranciscensis prefers long fermentation times at relatively low temperature, conditions that often prevail during type I sourdough preparations, this species grows optimally at 32 °C [159, 189]. However, whereas C. humilis grows optimally at 27–28 °C but does not grow above 35 °C [160, 189], the association of Lb. sanfranciscensis-C. humilis grows optimally at 25 °C and 30 °C and may explain its stability between 20 °C and 30 °C [160, 190]. The abundance of Lb. sanfranciscensis in wheat sourdoughs made at ambient temperature indicates a low competitiveness of other LAB species such as Lb. fermentum that prefers higher temperatures for optimal growth. Similarly, temperature may be responsible for a selection toward Lb. helveticus during Sudanese sorghum sourdough fermentations, which are carried out at 37 °C [148].

For the growth of sourdough LAB, also the pH plays an important role [160, 179, 188, 189]. For instance, Lb. sanfranciscensis cannot grow below pH 3.8–4.0 [160, 189], whereas C. humilis is not influenced by the pH [190]. An optimal pH for growth of around 5.0 has been found for Lb. sanfranciscensis. This pH value corresponds approximately to that observed during the first stage of dough fermentation. However, the growth of lactobacilli is favored over yeast growth at pH values above 4.5 [160]. Hence, the rate of acidification of the dough may determine the level of Lb. sanfranciscensis in the dough. Natural sourdough fermentations displaying higher pH values are often dominated by a different microbiota, encompassing Enterococcus, Lactococcus, Leuconostoc, Pediococcus, Streptococcus, and Weissella, which are commonly present in the cereal flour [57, 133] or during the early fermentation process but die off when a significant pH decrease occurs upon fermentation [89, 108].

Although sourdough fermentation proceeds anaerobically, the presence of oxygen in the beginning of the fermentation and when small amounts of dough (high ratio of surface to volume) are used may favor certain LAB and yeast species [179, 190, 192]. For instance, mild aeration has a positive influence on the competitiveness of Lb. amylovorus DCE 471 [179]. Similarly, P. kudriavzevii can only grow when enough oxygen is available during fermentation [190]. Further, the ionic strength and salt concentration of the dough affects microbial growth [160, 179, 193, 194]. Similarly, the presence of organic acids in and the buffering capacity of the flour influence growth of both yeasts and LAB [160, 179]. In general, sourdough LAB are acid-tolerant and their growth is favored in the presence of salt, as is the case for, for instance, Lb. amylovorus DCE 471 [188, 193]. Alternatively, the growth of C. humilis and S. exiguus is completely inhibited by 4% NaCl; also, the growth of these yeasts is strongly inhibited in the presence of acetic acid and to a much lesser extent by lactic acid [160, 179].

Whereas backslopping practices select for mainly heterofermentative LAB, the amounts of dough used for backslopping and the frequency of the refreshments determine the community dynamics and stability of the sourdough microbiota as well. The amount of backslopping dough defines the initial pH and in this way influences the growth and acidification rates of the LAB species involved [107, 189, 190]. Also, the amount of backslopping dough determines the dough yield and hence the availability of water (water activity of the dough). Short refreshment times may select for rapidly growing LAB species, which in turn depends on the fermentation temperature and influences the acidification rate. In this regard, Lb. fermentum is most competitive at 30 °C and 37 °C with backsloppings every 24 h, while a mixture of Lb. fermentum and Lb. plantarum prevails at 30 °C with backsloppings every 48 h [175]. This may explain why Lb. sanfranciscensis is sometimes missed during laboratory-scale fermentation processes [136, 175]. Also, a short refreshment time seems to favor C. humilis during sourdough fermentation compared to S. cerevisiae [190].

Finally, interactions between LAB and yeasts are an important aspect for the community dynamics and stability of the sourdough microbiota [50, 57, 165, 189, 195, 196]. Interactions encompass both cooperative and antagonistic ones. During some sourdough fermentation processes yeasts cannot develop at all, perhaps because of inhibition of yeast growth by nutritional competition or the presence of inhibitory compounds [47, 123, 175]. In other processes, mutualistic interactions lead to stable associations, not only between LAB species and yeasts (besides Lb. sanfranciscensis/C. humilis, also Lb. sanfranciscensis/K. barnettii, Lb. plantarum/S. cerevisiae, and Lb. brevis/Candida spp.) but also among LAB species (e.g., between Lb. sanfranciscensis and Lb. plantarum or Lb. paralimentarius) [40, 41, 58, 69, 142]. Nevertheless, the competitiveness of LAB and yeasts in sourdough seems to be strain-specific and not species-specific, as has been shown for Lb. sanfranciscensis [135] and Lb. plantarum strains [138] in wheat sourdoughs recently.

5.3 Isolation of Sourdough Yeasts and Lactic Acid Bacteria

5.3.1 Isolation of Sourdough Yeasts

Yeast isolation from mature sourdoughs is relatively uncomplicated as a stable ­sourdough usually harbors a homogeneous yeast population as part of the resident microbiota. However, the follow-up of different developmental stages of a sourdough or the search for minor components requires a strategy that is optimized towards the detection of subdominant components. Detailed information on food yeast isolation was provided by Deak [197]. Here, only a brief discussion of the currently practiced yeast isolation methods from sourdough is given, including some references to more complete methodological resources. After sample homogenization and dilution, yeast growth is suitably effected on solid media. This sequence of manipulations should be performed with minimal delay to avoid the settling of yeast cells and cell death. Diluents, usually distilled water, peptone water, saline, or Ringer solution, may influence the resulting cell counts, with peptone water having resulted in the highest cell recovery [198]. Overviews of classical growth media and isolation techniques for yeasts and foodborne yeasts are given by Yarrow [199] and Beuchat [200], respectively. The cultivation media used in sourdough analyses are rich media containing complex compounds such as peptone (e.g., Sabouraud agar), tryptone (e.g., Wallerstein Laboratory nutrient agar), yeast extract (e.g., yeast extract peptone dextrose agar), malt (e.g., yeast and malt extract agar, wort agar), and potato infusion (e.g., potato dextrose agar), together with an additional component to inhibit bacterial growth. Most often chloramphenicol is used as an antibiotic that can be added to the medium before sterilization without losing its activity. Acidification of the medium is sometimes used to restrict bacterial growth, while this is known to affect growth of some yeasts (namely of the genus Schizosaccharomyces). However, the acidity of the sourdough lets it appear unlikely for acid-sensitive strains to be present in a ripe sourdough. In general, most yeasts show good vegetative growth at room temperature, although some may grow at subzero temperatures and others up to 45 °C [201]. Cardinal growth temperatures of yeasts are species- and strain-specific. The most suitable incubation temperature for sourdough yeasts would be the temperature at which the sourdough is in its most active state. The most frequently applied temperatures are 25–30 °C. First yeast growth can under such conditions usually be observed after 2–3 days, while daily inspection of the plates for 5 up to 10 days is recommended to allow full differentiation of colony morphology and detection of more slowly growing components.

Other than the consideration of growth conditions, the selection of yeast colonies for further characterization and identification is the most important factor influencing the completeness of a diversity survey. Even though sourdough samples often present a homogeneous yeast population, one needs to bear in mind that the colony morphology of different yeast species is often very similar. Each observed morphotype should therefore be sampled more than once. A logical strategy would be to recover a number of colonies of each type that represents a reasonable percentage of the total number of colonies of that particular morphotype. Each morphoptype’s percentage provides important species abundance data in the sourdough if more than one yeast species is isolated. The generation of pure cultures is crucial before characterization and identification of the isolates. The purity should be tested microscopically and on antibiotic-free medium to exclude any carry-over of the accompanying bacterial microbiota.

5.3.2 Isolation of Sourdough LAB

Isolation of LAB from sourdough environments is challenging for three main reasons. First, sourdoughs are complex ecosystems not only in terms of their microbial composition but also in terms of the interactive effects among types of breadmaking processes and ingredients. The utilization of soluble carbohydrates by LAB and, thus, their energy yield are greatly influenced by the associated yeasts and vary according to the type of carbohydrates [195]. However, as many media for selective isolation of LAB incorporate yeast-inhibiting agents such as cycloheximide, pimaricin, and amphotericin B the trophic interaction between LAB and yeasts is in these cases disturbed, which may affect the recovery potential of LAB strains that strongly rely on this association. Secondly, sourdough fermentation is a dynamic process in which fast-acidifying LAB initially dominate the ecosystem and are then gradually replaced by typical sourdough LAB that largely contribute to the organoleptic and textural properties of the end product. Depending on whether the early subdominant LAB and/or the final dominant LAB are the target of the isolation approach, it may thus be necessary to include multiple samples taken at different time points. Finally, the LAB communities in sourdoughs may consist of metabolically very diverse groups, including obligately homofermentative and facultatively or obligately heterofermentative species. As some of these species have specific growth requirements in terms of the incubation medium and conditions (e.g., temperature, pH, atmosphere, etc.), it seems inevitable that different medium formulations and/or sets of incubation parameters are required to cover the entire metabolic LAB spectrum present in a sourdough sample.

Initially, sourdough LAB were mostly isolated on de Man-Rogosa-Sharpe (MRS) medium [202], which is the general medium used for the isolation and enumeration of lactobacilli from fermented food products. The MRS medium contains glucose as the main carbohydrate source. Triggered by growing insights in the species diversity of sourdough-associated LAB, a number of more specialized media have been developed for the selective isolation of typical sourdough species. For the specific detection of Lb. sanfranciscensis, Kline and Sugihara [75] proposed the SourDough Bacteria medium which contains maltose as the carbohydrate source in addition to freshly prepared yeast extract (FYE) to further enhance growth. The Sanfrancisco medium was developed for the isolation and description of Lb. pontis and Lb. mindensis [79, 203]. This medium contains three carbohydrates (maltose, fructose, and glucose), FYE, cysteine and rye or wheat bran. In parallel to the design of new media, several authors also described variations of the original MRS medium ­formulation for isolation of sourdough LAB. Vogel and co-workers [203] proposed a modified MRS medium, referred to as MRS “Vogel”, with higher pH value (6.3), whereas the MRS5 medium [185] contains the three major carbohydrates present in the sourdough ecosystem (i.e., maltose, fructose, and glucose) in addition to cystein and a vitamin mixture. In subsequent studies, the MRS5 medium has been successfully used for the isolation of several novel Lactobacillus species from sourdough such as Lb. spicheri [80], Lb. namurensis [87], and Lb. crustorum [86]. From these recent descriptions, it thus appears that the use of a modified MRS formulation with a lowered pH (<6.0) and supplemented with an additional carbon source such as maltose and/or fructose as well as with amino acids and vitamins under anaerobic conditions is one of the most successful strategies for the isolation of (new) sourdough LAB species. In a recent study, the qualitative and quantitative performance of 11 elective and selective culture media was compared for isolation of lactobacilli from type I sourdoughs [204]. On the basis of the identification results obtained with protein profiling, the largest species diversity was recovered on maltose-containing MRS medium. However, the fact that MRS5 medium allowed the isolation of a specific (but unidentified) subpopulation only found on this medium indicates that there is no single efficient medium for the recovery of all lactobacilli from type I sourdoughs.

5.4 Identification of Sourdough Yeasts and Lactic Acid Bacteria

Traditionally, identification of sourdough microorganisms relied on (selective) culturing, selection and purification of a limited number of isolates, and identification of purified isolates with phenotypic and/or genotypic methods. Although this approach has significantly contributed to our current knowledge of the sourdough-associated yeast and LAB species diversity, the use of culture media holds a number of intrinsic limitations. In a culture-based approach, species with very specific nutrient and growth conditions may only sporadically or even not be recovered which leads to an underestimation of the actual microbial species diversity present in the complex sourdough ecosystem [68]. In contrast, culture-independent techniques that are based on phylogenetic dissection of the metagenomic DNA extracted directly from the sample allow one to unravel the species diversity and dynamics of sourdough yeasts and LAB without the need to isolate and culture its single components. However, also these DNA-based methods have a number of limitations, including poor detection capacity of subdominant species and inadequate taxonomic resolution between phylogenetically closely related species. Depending on the aim of the study, conventional culturing and molecular methods are therefore often combined to obtain a more complete picture of the microbial species diversity of sourdough ecosystems [41, 106].

5.4.1 Culture-Dependent Approaches

5.4.1.1 Yeasts

Identification is the localization of individuals in a classification scheme by means of diagnostic characteristics resulting in the assignment of names. The diagnostic characteristics to be used are provided in the species descriptions and, in the case of yeasts, are collected in the monograph “The yeasts: a taxonomic study,” currently in its fourth edition [1], with the fifth edition about to be released [204]. While the fourth edition still included instructions for the phenotypic identification of yeasts [199], the fifth edition reformulates them as phenotypic “characterization” instead [205]. This is a consequence of the need to use DNA-based methods to recognize the since 1998 twofold increased number of yeast species. Nevertheless, the accurate description of fermentation and assimilation abilities as well as other phenotypic characters of yeasts continues to be of interest in technological, ecological and taxonomic frameworks.

Among the DNA-based methods currently applied to yeast identification the partial sequencing of the large subunit (LSU) ribosomal ribonucleic acid genes occupies a key position [206]. The DNA sequences of the variable regions D1 and D2 located at the 5’ end of the LSU of virtually all known yeast species are documented in the public databases of the International Nucleotide Sequence Database Collaboration (INSDC, including GenBank, the European Molecular Biology Laboratory, and the DNA Data Bank of Japan). The entries of the three submission hubs are bundled, daily updated, and made available for searches by the NCBI (www.ncbi.nlm.nih.gov/nucleotide/). Few distinct yeast species show no or low sequence divergence in the D1/D2 LSU rDNA, can therefore not reliably be distinguished, and require complementary analyses such as Saccharomyces bayanus and S. pastorianus [206], Hanseniaspora meyeri and Hanseniaspora clermontia; Hanseniaspora guilliermondii and Hanseniaspora opuntiae [207], Meyerozyma guilliermondii and Meyerozyma caribbica; Trichomonascus ciferrii and Candida mucifera; K. marxianus and Kluyveromyces lactis [55], Debaryomyces hansenii, Debaryomyces fabryi and Debaryomyces subglobosus [208]. Genetic regions that show in most cases larger divergence than the D1/D2 LSU rDNA and for which substantial sequence records have been accumulated include the ITS region of the ribosomal gene cluster. This region is favored by mycologists as the barcoding locus of fungi, although no common threshold value to distinguish intraspecific from interspecific variation can be defined [209, 210]. No systematic evaluation of ITS sequence variation to answer this question for ascomycetous yeast species exists to date. In comparison to sequencing, RFLP analysis of ITS sequences offers simplified access to partial DNA sequence information. The accessed information is determined by the recognition sites of the applied restriction enzymes and typically includes only a few nucleotides, necessitating a range of restriction enzymes to reliably distinguish the species in question.

Single-copy protein-coding gene sequences also accumulate in the public ­databases and may be used to complement those of D1/D2 LSU and ITS sequences in cases where the ribosomal genes do not allow a conclusive species identification. Such protein-coding genes include the actin gene (ACT1) [55], translation elongation factor 1 alpha (TEF1), the mitochondrial cytochrome oxidase 2 gene (COX2) [211], and the largest and second largest subunits of the RNA polymerase II gene (RPB1, RPB2) [212]. In contrast to the multicopy ribosomal DNA (e.g., D1/D2 LSU, ITS), single-copy nuclear genes may be more difficult to amplify, as the design of universal primers effective in phylogenetically distant species is not always possible and as only single primer binding sites are available in a haploid genome in contrast to hundreds of binding sites for ribosomal genes. Highly variable mitochondrial genes (e.g., COX2) bear the possibility of having been subject to a different evolutionary path than the nuclear genome, in other words, are prone to potential horizontal gene transfers across species borders. As databases for complementary gene sequences are far more restricted than for the D1/D2 LSU, one needs to assure the existence of reference sequences. Databases that allow searching the available sequences for specific strains, such as the yeast database of the Centraalbureau voor Schimmelcultures, The Netherlands (www.cbs.knaw.nl/yeast/BioloMICS.aspx) and the StrainInfo portal (www.straininfo.net/), Ghent University, Belgium, may be used for the selection of complementary sequencing targets. The comparison of a query sequence with reliable type strain sequences is essential, as type strains are the only valid taxonomic reference for a species. Sequence alignments outside the commonly consulted BLAST (Basic Local Alignment and Search Tool) tabulated results are helpful, because type strain sequences are often not recognizable from the sequence entry title line. It is recommended to include type strain sequences of the phylogenetically most closely related species in these alignments to confirm the differentiation of the species in question by the given DNA region.

While the so far discussed methods use genetic information of few or single loci, the techniques commonly known as DNA fingerprinting exploit genetic information that is distributed throughout the genome. A large variety of protocols exists and the more reproducible among them are based on the specific binding of PCR primers to mini- or microsatellite sequences in contrast to arbitrary binding realised in RAPD-PCR. The primer most frequently applied to sourdough yeasts was derived from a ubiquitous minisatellite sequence found in the protein II gene of the bacteriophage M13 [213]. The primer referred to as M13 results from a consensus sequence of 12 partially incomplete repeats. After its use in a PCR assay as the single primer and visualization of the PCR reaction products by agarose gel electrophoresis, usually species-specific banding profiles based on the different lengths of amplifiable sequences enclosed or flanked by M13 minisatellites are observed. The important influence of experimental factors such as the DNA extraction method and PCR and electrophoresis parameters on the resulting profiles implies the need to include type strains for ideally side-by-side-comparisons if a complete identification is to be performed. However, PCR-fingerprinting without type strains may be used to group larger numbers of isolates and to select those that are representative of each group for identification by DNA sequencing [40].

The species S. cerevisiae has been observed to show extensive pheno- and ­genotypic intraspecies diversity (reviewed in [214]). Part of such strain diversity has been traced by molecular analyses to genomic variability associated with Ty-element insertion sites [215]. Ty elements belong to a group of eukaryotic transposable elements that are also called retrotransposons because of some similarities with retroviruses. Ty-elements are flanked by long terminal repeat (LTR) sequences, in turn formed by repeated delta elements. Delta elements are also found in larger numbers independently from LTR sequences and have been used to design and optimize specific PCR primers to amplify the sequences between two delta elements in S. cerevisiae, resulting in an effective strain-typing method for this species [216, 217].

5.4.1.2 LAB

Although largely abandoned and replaced by molecular tools, characterization and identification of sourdough LAB species by phenotypic methods is in some cases still useful, or even mandatory when it concerns new species descriptions. The conventional phenotypic approaches for identification of sourdough LAB species may include physiological and chemotaxonomic tests and determination of major fermentation pathways, carbohydrate utilization patterns, lactic acid configuration, and peptidoglycan types. To determine carbohydrate patterns and enzymatic properties in a faster and more reproducible way, miniaturized biochemical test systems such as the API system (Biomérieux, France) can be used for phenotypic characterization of sourdough LAB species [23]. However, it should be stressed that the identifications obtained by comparison with commercial databases such as those linked to API are only tentative and need verification with other taxonomic methods. A more advanced phenotypic identification approach is offered by chemotaxonomic methods, which are based on the use of analytical methods to detect and characterize one or several chemical cell components. Protein profiling by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) has been used for identification of LAB isolates recovered from Italian [127] and Greek [72] sourdoughs. SDS-PAGE of cellular proteins generally offers sufficient discrimination of LAB isolates at species level but may fail to discriminate between species in the Lb. acidophilus group [218] and the Lb. plantarum group [71], both of which are prominent members of sourdough ecosystems. Although not yet evaluated for sourdough LAB, the use of mass spectrometry (MS) methods is probably the most powerful phenotypic approach currently available for classification and identification of bacteria [219]. One of these methods, matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) MS, allows one to measure peptides and other compounds in the presence of salts and to analyze complex peptide mixtures, which makes it an ideal method for measuring nonpurified extracts and intact bacterial cells. The resulting MALDI-TOF MS spectra can be used to generate identification libraries for simple and high-throughput identification of unknown bacterial isolates. De Bruyne and co-workers [220] constructed such identification libraries for the LAB genera Leuconostoc, Fructobacillus and Lactococcus, and reported that 84% of the ­leuconostocs and fructobacilli and 94% of the lactococci were correctly identified at species or subspecies level. Identification accuracies for the former two groups further increased to 94–98% when machine learning was applied, which indicates the important role played by advanced techniques for the analysis of complex MALDI-TOF MS profiles.

Essentially, molecular approaches for identification of sourdough LAB are either DNA fingerprint- or sequence-based. DNA fingerprinting methods rely on the use of restriction enzyme analysis, the use of specific or random PCR primers, or a combination thereof. Ribotyping, one of the first DNA fingerprinting techniques used in bacterial taxonomy [221], relies on a combination of restriction analysis of total genomic DNA and Southern hybridization to visualize a subset of restriction fragments with labeled rDNA probes targeting conserved domains of ribosomal 16S and 23S rRNA encoding genes. Although ribotyping generally provides high discriminatory power at species to subspecies level, it has been used only sporadically as an identification technique for LAB species. In specific cases, ribotyping has proven particularly useful in the classification and identification of sourdough LAB, for example for discrimination between the genomically and phenotypically highly similar sourdough LAB W. cibaria and W. confusa [63] and for intraspecific differentiation of Lb. sanfranciscensis strains from different sourdoughs [121]. Despite its high resolution at species as well as at strain level, amplified fragment length polymorphism (AFLP) fingerprinting was also only used sporadically for LAB identification purposes. Essentially, AFLP combines the power of restriction fragment length polymorphism with the flexibility of PCR-based methods by ligating primer-recognition sequences (adaptors) to the digested DNA [222]. This whole-genome fingerprinting technique has proved useful to support the description of the sourdough species Lb. hammesii [82], Lb. crustorum [86] and Lb. namurensis [87] and for molecular source tracking of Lb. spicheri, Lb. plantarum and Lb. sanfranciscensis in the production environment of artisan sourdough bakeries [155]. In contrast to the aforementioned methods, RAPD-PCR and repetitive DNA element (rep)-PCR are technically less demanding DNA fingerprinting techniques based on a single PCR step. Both methods are fast, relatively inexpensive, and exhibit a high discriminatory power ranging from genus to intraspecific level, which explains their wide application range for the identification and classification of LAB. RAPD-PCR has been used in multiple studies on sourdough LAB for species identification [129], strain differentiation [80, 89, 120, 126, 127, 131, 148, 158] and strain monitoring purposes [70, 122, 135, 138, 223]. The reproducibility of RAPD-PCR is highly influenced by various factors, such as DNA purity and concentration and minimal differences in the PCR temperature programme [224], for which reason this method is less suitable for interlaboratory comparisons. Because of the use of longer PCR primers complementary to bacterial interspersed repetitive DNA elements such as ERIC, BOX, REP or (GTG)₅ and higher annealing temperatures, rep-PCR protocols are more robust and display a higher level of reproducibility [225]. rep-PCR using the (GTG)₅ primer, i.e., (GTG)₅-PCR, has been found particularly useful for differentiation of sourdough LAB at the (sub)species up to the strain level [74, 105, 106, 155, 226, 227]. For high-resolution differentiation of individual sourdough LAB strains by DNA fingerprinting, however, AFLP fingerprinting and pulsed-field gel electrophoresis (PFGE) are the most powerful. Essentially, PFGE involves the electrophoretic separation of genomic macrorestriction fragments obtained by digestion with rare-cutting enzymes in an alternating electric field. In sourdough studies, PFGE has been used as a typing method to differentiate among strains within Lb. plantarum and Lb. sanfranciscensis [126, 135, 174].

Sequence-based analysis approaches for identification of sourdough LAB have long relied on the use of 16S rRNA genes, and this has become a standard approach to obtain a first preliminary view of the taxonomic diversity among a set of unknown isolates recovered from a sourdough ecosystem [72, 80, 148]. In many of these studies, only partial 16S rRNA gene sequences are determined and used in comparisons with public sequence databases. In many cases, the use of partial sequences will only allow a tentative identification, of which the reliability is likely to improve when the entire 16S rRNA gene is sequenced [228]. Despite its established use as a standard method for identification of LAB species, 16S rRNA gene sequencing does not allow differentiation of phylogenetically closely related species [63–66, 71]. The growing availability of whole-genome sequences has triggered the search for alternative genes that offer a higher taxonomic resolution than the 16S rRNA gene. The use of protein-encoding genes or so-called housekeeping genes essentially combines the technological advantages of 16S rRNA gene sequencing and the taxonomic resolution offered by a number of fingerprinting methods. Sequencing of one or preferably multiple of these genes as taxonomic markers is a crucial step forward in the development of standardized and globally accessible methods for the identification of LAB. Housekeeping genes such as pheS (encoding the phenylalanyl-tRNA synthase) and rpoA (encoding the DNA-dependent RNA polymerase alpha-subunit) display higher divergence rates than the 16S rRNA gene, and allow discrimination between closely related LAB species with almost identical 16S rRNA gene sequences [67, 229]. Several studies on sourdough LAB species diversity have used such protein-encoding genes as phylogenetic markers in a single-locus sequence approach. In conjunction with (GTG)₅-PCR fingerprinting, the pheS gene has been successfully used for the identification of LAB species from sourdough fermentations at laboratory scale [107] and from Belgian artisan bakery sourdoughs and their environment [105, 106, 155], as well as for unraveling the intraspecific diversity in the sourdough species Lb. rossiae [74]. Settanni and co-workers [156] used the recA gene, encoding a protein essential for repair and maintenance of DNA, in a multiplex PCR assay to discriminate between the phylogenetically highly related Lb. plantarum, Lb. pentosus and Lb. paraplantarum in sourdough ecosystems. The recA gene has also been used in combination with the 16S rRNA gene to unravel the identity of LAB isolates recovered during wheat flour sourdough type I propagation [135]. Sequences derived from the tuf gene, which encodes the elongation factor Tu, have revealed a higher discriminatory power compared to 16S rRNA gene sequences and have been used to support the delineation of the new sourdough species Lb. secaliphilus [85]. Although single-locus sequence analysis approaches are now commonly used within specific LAB groups, it has been argued that the phylogenetic information obtained from only a single gene may be influenced by lateral gene transfer (LGT) and may lead to incorrect identifications. To compensate for possible LGT events, it has been suggested that multilocus sequence analysis (MLSA) of at least five housekeeping genes from diverse chromosomal loci and with wide distribution among taxa is required to reliably distinguish a species from related taxa [230]. After a more thorough evaluation, however, Konstantinidis and co-workers [231] concluded that three genes are sufficient to anticipate the possible effects of LGT in MLSA-based identification schemes. For LAB, MLSA based on the combined sequence analysis of the genes atpA, pheS and rpoA has been successfully explored for species identification of enterococci [229], lactobacilli [67], leuconostocs [232], and pediococci [233]. For sequence-based differentiation of LAB at strain level, multilocus schemes typically include six or seven housekeeping genes. The resulting multilocus sequence typing (MLST) approach has so far mainly been applied to study community structure, evolution and phylogeography of bacterial pathogens [234]. A few MLST schemes have been specifically developed for Lactobacillus species, including Lb. casei [235, 236], Lb. plantarum [237] and Lb. salivarius [238].

5.4.2 Culture-Independent Approaches

The first approaches used to identify sourdough microorganisms independent of culturing relied on the use of oligonucleotide probes targeting ribosomal gene sequences specific for individual species or groups of species. The majority of these probe-based methods made use of partial 16S rRNA gene sequences that were identified as molecular signatures unique to specific LAB species [203, 239]. Gradually, the relatively laborious probe hybridizations were replaced by faster community PCR assays using species-specific oligonucleotide primers. The success of both approaches strongly depended on rigorous in silico probe or primer design and required in vitro and in vivo validation using taxonomically well-characterized type and reference strains and spiked sourdough samples, respectively. Species-specific PCR primers complementary to signature sequences in the 16S or 23S rRNA gene or in the 16S–23S rRNA intergenic spacer region have been applied for the culture-independent identification of sourdough LAB species [26, 156, 161, 240–242]. By combining multiple sets of primers, several typical sourdough LAB species can be simultaneously detected. In this way, Settanni and co-workers [156] developed a two-step multiplex community PCR assay that enabled rapid identification of up to 16 Lactobacillus species in sourdough samples. The introduction of real-time PCR technology has allowed one to further increase the sensitivity of PCR-based identification assays and enables the simultaneous detection and quantification of food microorganisms [243]. For this purpose, SYBR Green-based real-time PCR assays based on the detection of the pheS gene have been used for source tracking of Lb. plantarum and Lb. sanfranciscensis in traditional sourdoughs and their production environments [155].

Whereas probe- and primer-based identification approaches offer the specificity and selectivity required to detect and monitor specific LAB in sourdough samples, they were not designed to offer a complete picture of the predominant LAB species diversity or to reveal new or unknown LAB species diversity in sourdough ecosystems. In contrast, community fingerprinting methods such as denaturing gradient gel electrophoresis (DGGE) and temporal temperature gradient gel electrophoresis (TTGE) do not require prior knowledge of the ecosystem’s diversity and are universally applicable to study the species diversity and dynamics of complex bacterial communities in food environments [244]. The universal use of DGGE fingerprinting is based on the sequence-dependent separation of a mixture of equally sized PCR amplicons generated from a common taxonomic marker such as the 16S rRNA gene. For the design of PCR primers, the V1, V3, and V6-V8 hypervariable regions of the 16S rRNA gene are most commonly used. Taxonomic information on individual members of the sample community can be obtained by band position analysis provided that an identification database is available, clone library analysis, sequencing of excised and purified DGGE bands or hybridization using species-specific probes. Major drawbacks of DGGE fingerprinting include its inability to detect subdominant (i.e., <1%) community members and the fact that a single strain or species may be represented by multiple bands in the DGGE profile due to heterogeneous rRNA operons and/or heteroduplex molecules. Either using universal or group-specific 16S rRNA gene primers, DGGE has been widely applied to inventorize LAB communities in sourdoughs [41, 47, 106, 128, 245, 246] and to investigate the dynamics, adaptation, and source of predominant sourdough LAB communities [34, 39, 80, 107, 123, 142, 155, 185]. Likewise, primers targeting the 26S LSU rDNA have been used for DGGE fingerprinting analysis of sourdough yeast communities [29, 46]. To maximally cover the microbial species diversity present in a sourdough ecosystem, a number of DGGE studies have combined the use of 16S and 23S rRNA gene primers to determine in parallel the predominant LAB and yeast composition of sourdough samples [33, 39, 41, 47, 123, 142, 245]. Compared to DGGE, TTGE has been used to a much lesser extent for culture-independent analysis of the sourdough microbiota [112]. In many of the cited studies, the sequence heterogeneity of the multicopy 16S rRNA gene is mentioned as an important limitation in DGGE, as this may lead to an overestimation of the LAB species diversity. The degree of overestimation can be estimated by scoring individual bands by position analysis with a reference database and/or by band sequencing. Alternatively, single-copy genes that do not exhibit this heterogeneity such as rpoB have been evaluated for DGGE fingerprinting of LAB species during food fermentations [247].

Microarray technology represents one of the most recent culture-independent approaches to study the diversity and identify individual members of the sourdough microbiota. Phylogenetic microarrays, containing partial 16S rRNA gene sequences as targets, are ideally suited for this purpose but are currently not available for sourdough microbiota. Alternatively, a functional gene microarray can be used when the original annotation information allows one to link the responding oligonucleotides to the original species. Weckx and co-workers [108, 109] used a LAB functional gene microarray to analyze time-related RNA samples that represented the ­metatranscriptome of sourdough fermentations maintained by daily backslopping. The resulting set of hybridization data allowed one to monitor the LAB community dynamics in the sourdough ecosystem and to identify the major LAB species that contributed to the establishment of a stable ecosystem through its three successive phases.