Microbial ecology of fermented foods

Abstract

Over thousands of years, a wide variety of food fermentations in human diets have developed in different geographical regions, cultures, and traditions. The desirable conversions of food components that occur during fermentation rely on a diverse array of microorganisms, which interact with one another and with their environments. Notably, many yeast, bacterial, and filamentous fungal taxa are key members of the microbiomes of fermented food products. Strategies employed by these microorganisms to ensure their survival typically include maximising reproductive rates, and this affects patterns of microbial succession and key metabolic dependencies. The principles of microbial ecology can be used to understand growth, activity, and interactions in these diverse communities. In this review, we summarize the historical significance and factors influencing key microbial functions in palm wine, wine, sourdough, and tempeh, selected for their varied raw materials and fermentation techniques. Despite differences, in fermented foods abiotic and environmental influences from natural or human-controlled sources act to define a food’s microbial ecology and therefore the final product’s properties. Modern production of traditional and novel foods by microbes should consider the principles of microbial ecology to understand and control fermentation outcomes.

Understanding the role of microbial ecology in fermented foods can underpin the production of these popular foods in food systems globally.

Introduction

Microbial ecology considers the relationships between microbes and their environment. In the context of fermented foods, we are focussed on how microbes grow, how microbial communities form and interact with one another, and how these communities interact with other living aspects of their situation and respond to abiotic fluctuations. The ecosystems of fermented products are constructed, controlled, and consumed by humans to produce unique foods and beverages for human culture, health, and nutrition.

Fermented foods are human’s earliest attempts at preserving ingredient quality, enhancing nutritional content, and extending lifespan of foodstuffs. The earliest examples of fermented foods come from the preservation of grain through fermentation to make bread, beer, and a variety of wines, with many cultures practicing food preservation through fermentation through time (Samuel 1996, Shevchenko et al. 2014, Wang et al. 2016, McGovern et al. 2017). Of course, foods are deeply related to human culture with many fermented foods becoming stalwarts of religious and cultural practices, with an example of fermented grapes (wine) into Christian religious rites (Kreglinger 2016) and fermented palm sap (palm wine) into Balinese religious and cultural ceremonies (Wijaya et al. 2024). We now know that the process of fermentation restricts access of pathogenic or intoxicating bacteria to food, often by reduction of pH, ethanol, and other inhibitory substances (Caplice and Fitzgerald 1999, Fleet 2007) thereby improving food safety. There are a wide range of foodstuffs converted by fermentation, which have been summarized comprehensively (Tamang Jyoti et al. 2020, Gänzle 2022) and have received further attention as modern society seeks authentic, healthy, plant-based foods (Hernández-Velázquez et al. 2024, Mukherjee et al. 2024).

While microbes transform raw foodstuffs into food, with or without human intervention, many principles of microbial ecology are at play. This review considers those principles by relying on principles of ecology for this applied science before considering interactions between microbes in the fermentation ecosystem. An overview of the fermenting microbes, together with known and unknown interactions is given, with examples given from recent reports and emerging areas of research. The review concludes with a consideration of how the principals of microbial ecology can be applied to fermented foods and how this will continue to help development of the food in the modern food system. The rise of novel foods, fermentation systems, and the resurgence of traditional food means that we can learn from microbial ecosystems in food and apply this knowledge for meeting the challenges of feeding the increasing global population with appropriate food sources.

The term ‘fermentation’ with relationship with food now has a wide definition. A fundamental biochemical definition exists, where compounds are converted to further compounds in an anaerobic environment, as an ATP-generating process where organic compounds are electron donors and receptors (Kim and Gadd 2019). A textbook definition invariably gives conversions of glucose, anaerobically, to ethanol (and yielding ATP) as an example. However, fermented foods can be made using aerobic fermentation (De Roos and De Vuyst 2018). The expanding definition of fermentation and food now includes ‘precision fermentation’ to mean the production, by microbes, of specific ingredients for food (such as rennet, heme), which are purified and used in food. The broadest definition and the one that we rely on here is that ‘fermented food’ are foods made through desired microbial growth and enzymatic conversions of food components (Marco et al. 2021).

There is a broad array of foods obtained using natural or spontaneous fermentation (Tamang Jyoti et al. 2020), and as food systems become bigger with the need for industrial production to be reproducible, starter cultures are often used. Here, an inoculum of a single strain of a desired species is added to conduct the fermentation. While simplified community structures are likely at play in these fermentations, we note that there is considerable variance in the effective fermentation populations [an example is wine fermentation where indigenous populations persist even when yeast is inoculated—a classic example is given by Heard and Fleet (1985)]. The microbes chosen for inoculation have been selected by humans for efficient conversions and short fermentation times, often from isolates from naturally occurring fermentations. These norms mean that our application of ecological principles to fermented foods is somewhat curtailed given these parameters and timeframes.

The ecosystem that a fermented food inhabits is human defined, constructed, and controlled. Defined starting materials (for example milk, grapes, and wheat flour) and known incubation temperatures and timeframes are sourced from traditional practices, which are closely monitored by humans. Microbial ecosystems in wider environmental situations may reach a steady state with occasional perturbations (for example, a soil ecosystem may be steady over days and weeks, with rain, temperature fluctuations, or ingress of plant growth perturbing the system). In contrast, the fermented food ecosystem is often batch culture, with a defined ‘beginning’ and ‘end’. Given that humans deliberately control abiotic parameters withing fermented foods (salt, moisture, temperature, and pH), the ecosystem is more defined but does abide by some similar principles. In the next section, we consider microbial ecological principles; development, diversity, and stability of microbial communities, and how interactions between members of these communities take place.

Principles of microbial ecology for fermented foods

Ecological principles in communities of fermented foods are considered in this section. We discuss the development of microbial communities, population selection, and succession, together with the abiotic factors that tend to influence these processes. We consider the diversity and stability of microbial communities and the known interactions between a microbial community.

Microbial communities are an assemblage of individual cells and populations that inhabit an ecosystem. In the context of fermented foods, the definition and delineation of the community is somewhat easier than in other ecological niches, given that humans are determining the environment and extent of action of the food. Population selection has classically been seen as individual cells or organisms existing along an r to K gradient (Atlas and Bartha 1993). Here, r strategists rely upon high reproductive rates for continued survival, while K strategists depend upon physiological adaptations to the ecosystem. In a fermented food ecosystem, most microbes exhibit ‘r-like’ strategies within the community. For example, Saccharomyces cerevisiae has a high reproductive rate and prevails when resources are not limited through growth on sugar-rich substrates to numerically dominate—a pattern which is characteristic of many fermented food ecosystems. K strategists are rare, but examples exist, such as Brettanomyces bruxellensis in a wine fermentation environment, which reproduces slowly and has a stable and permanent population (Harrouard et al. 2023). While seen as a spoilage yeast based on the unpleasant aromas produced, the ‘K-like’ strategy of this yeast demonstrates the ecological diversity of this fermented food.

Succession of microbial communities is a feature of many fermented foods. Individual populations of a community fill niches within the ecosystem, but over time, some populations can be replaced by other populations that are better adapted to fill a functional role (Atlas and Bartha 1993; p. 131). The types and functions of these interrelationships impose stability of the community over time, from the primary populations of pioneer colonisers through to a stable assemblage of a climax communities (Atlas and Bartha 1993; p. 132) While the climax community is a state of general stability, this state is often not reached in fermented foods, as the food may be seen as spoilt or rotten. A fermented food example of succession with microbial adaptations to allow new populations to develop is in the production of beer by the traditional fermentation mode (for example lambic or coolship beer styles) (Bokulich et al. 2012). These beers are fermented over months to one or several years, often in wooden barrels and experience temperature fluctuations (Bokulich and Bamforth 2013). The beers contain diverse communities of yeasts and bacteria which create characteristic and prized flavours and aromas (De Roos and De Vuyst 2022). A recent comprehensive study shows several phases of growth, where enterobacteria predominate and is followed by the main fermentation stage where yeasts predominate, followed by an acidification process before maturation phase (Vermote et al. 2023). Multiphasic and successional fermentation is a feature of this beverage and moves to stability where the climax community is achieved—and the beer is seen to be ready for consumption.

In environmental microbial communities, successional processes are a consequence of the initial microbes modifying the habitat in a way that permits new populations to develop. Both autogenic and allogenic succession processes exist (Atlas and Bartha 1993) and there are examples of these types in fermented foods. An example of autogenic succession where the creation of anaerobic conditions from the action of facultative anaerobes allows the development of obligate anaerobic microbes to grow. The principal fermenting yeast, S. cerevisiae conducts this work in many fermented food systems. Saccharomyces cerevisiae can create energy from fermentation when oxygen is available (Crabtree effect), which creates a microaerophilic environment for subsequent microbes (Perez-Samper et al. 2018); for example lactic acid bacteria (LAB) in cereal grain and grape wine fermentations. In contrast, allogenic succession occurs where the fermentation environment is altered by abiotic factors such as seasonal changes. An example here would be the response of a microbial community to the addition of salt in fermented foods; an excellent example is during cheese fermentation and maturation (Louw et al. 2023). Here, the interaction between salt and water availability in the cheese rind are strong predictors of microbial community diversity, while geographic location of the cheese production has a limited effect (Wolfe et al. 2014). Another example are the microbial communities present during the mud pit fermentation of baijiu, a popular Chinese drink which is produced from a fermentation of grain (usually sorghum), which then undergoes solid-state distillation. A recent study shows the seasonal dynamics of fungi and bacteria during fermentation and demonstrated while there were different successional features in three seasons, bacteria were found to be more responsive to temperature (Chai et al. 2024). However, this is clearly a complex system as a related fermentation system showed fungi responsive and bacteria nonresponsive to temperature fluctuations (Liu et al. 2023). The nature of succession requires a notion of stability of the microbial community, which is often linked to diversity of species present.

Classical texts in microbial ecology refer to diverse communities being inherently stable and able to withstand change. It is worth noting that due to food fermentations being controlled by and ended by cultural practices of humans, a ‘stable’ end state is rarely seen due to the ‘batch’ culture nature of food fermentations, where there is a clear beginning, middle and end. Yet, the diversity, stability and productivity of a food fermentation can be measured and here we consider some of the major drivers for these processes in the examples below. Species diversity has been measured by culture-dependent assays to calculate and identify major species but of course these methods are enhanced by DNA-dependent measures of diversity. Advancements in the applications, precision, and reduction of costs of next-generation sequencing has meant the data available to examine microbial fermentation is increasing (Carlino et al. 2024). Care should be taken when selecting amplicon markers to gain accurate measures of taxonomic diversity (Rué et al. 2023), yet combining studies and reanalysis of published work can lead to broader conclusions regarding the microbial ecology of food (see for example Sumerta et al. 2025). As a final comment, microbial community stability and interactions are not considered in this review but have been comprehensively assessed in earlier reviews (Ivey et al. 2013, Auchtung et al. 2025).

Diverse microbes are present in fermented foods

The action of fermentation in foods is primarily undertaken by the heterogeneous subgroup of bacteria known as the LAB and fungi, which accordingly are the main microbial groups studied in metagenomic-based surveys of fermented foods (De Filippis et al. 2017, Gänzle 2022). Fermented foods often act as dietary sources of live microorganisms, and while the total content of microbes found in fermented foods varies, commercially available dairy products typically contain 10^5–7 LAB per ml or g (Rezac et al. 2018). LAB are ubiquitous microbes, and the predominantly phenotypic basis of their classification is due to their many shared physiological traits, notably the metabolism of lactic acid (Giraffa 2014). Some LAB inhabiting fermented foods are highly specialized to a particular host and/or ecological niche, whereas others are ‘nomadic’, and this broader distribution is accompanied by relatively higher levels of phylogenetic diversity (Duar et al. 2017). Most widely dispersed LAB associated with fermented foods belong to the family Lactobacilliaceae, with some members of the Enterococcaceae and Streptococcaceae found in fermented vegetable, dairy, meat, and seafood products (Zheng et al. 2020, Gänzle 2022).

Other less common food fermentation microorganisms include Gram-positive bacteria from the genera Bacillus, Lentibacillus, Brevibacterium, Propionibacterium, Staphylococcus, and Kocuria. Gram-negative bacteria found in fermented foods tend to be either acetic acid bacteria (Acetobacter, Gluconacetobacter) or those associated with plants from the family Enterobacteriaceae. Aside from S. cerevisiae, which is the fermenting yeast found in the vast majority of fermented foods where a yeast is present, other yeast genera found in fermented foods and beverages include Blastobotrytis, Brettanomyces, Debaromyces, Issatchenkia, Kluyveromyces, Maudiozyma, Monosporozyma, Torulaspora, and Zygosaccharomyces (Gänzle 2022, Liu et al. 2024). Furthermore, nonunicellular fungi are also used in the production of fermented meat (chorizo, salami), dairy (brie, camembert), soy (soy sauce, miso, tempeh, and sufu), and alcoholic rice beverages (sake, Shaoxing wine). These are most commonly from the genera Aspergillus, Penicillium, and Rhizopus (Gänzle 2022).

Multiple factors are associated with the taxonomic, genetic, and functional diversity of food fermentation microbiomes. First, the diversity of fermented food microbiomes is strongly shaped by positive, negative, and neutral patterns of interaction (Ivey et al. 2013). Notably, various bacteria inhabiting fermented fruit, vegetable, and dairy products can antagonize pathogenic bacteria either via the production of bacteriocins or other antimicrobial activities (Ivey et al. 2013, Tamang Jyoti et al. 2016). In addition to phylogenetic diversity, genome size is associated with lifestyles of LAB found in fermented foods, with those that are not strongly host-adapted typically having larger genomes (Duar et al. 2017). The optimal growth temperature of LAB in food fermentation is also often associated with a species’ typical habitat; vertebrate-adapted lactobacilli usually have higher growth temperature optima than invertebrate-adapted lactobacilli and those that are not adapted to a particular host (Duar et al. 2017). Furthermore, in fermented dairy products, mesophilic LAB display a strong association with cheeses ripened over an extended period of time, whereas thermophilic LAB are associated with earlier stages of cheese-making represented by milk, curds, and fresh cheeses (De Filippis et al. 2017).

Beyond yeasts, fungi, and bacteria, the microbiomes of fermented foods are also usually partially composed of viruses. Tens of thousands of viral sequences have been detected in fermented seafood and vegetable products, although the family-level complexity of viral communities in fermented foods is relatively low compared to those found in human, marine, and soil environments (Park et al. 2011). Furthermore, it has also been demonstrated that within similar fermented foods that are produced in geographically distinct locations, the diversity of viral communities is associated with geographical origin (Jung et al. 2018). Among fermented food viruses, bacteriophages play a particularly important ecological role in microbial communities by causing microbial death or integrating themselves into other microbial genomes, which can both affect population structure and composition (Zhang et al. 2025).

Food fermentation microorganisms also display diverse functional traits based on genetic variation. According to their fermentation products and characteristics, LAB can be divided into three groups. Obligately and facultatively heterofermentative LAB are able to produce lactic acid, acetic acid, formic acid, ethanol, and CO₂, whereas obligately homofermentative LAB only produce lactic acid (Giraffa 2014). While yeasts do not fall under similar broad divisions, species that occupy different food environments often exhibit genetic specialisations to nutrients found in those environments. Strains of S. cerevisiae can vary in terms of the acquisition of nutrients and responses to stressors such as ethanol and sulphur (Flörl et al. 2025). In addition to S. cerevisiae, Kluyveromyces marxianus and Torulaspora delbrueckii isolates inhabiting dairy-based foods can have genes involved in lactose and/or galactose metabolism. Likewise, S. cerevisiae and T. delbrueckii isolates found in bread usually exhibit increased numbers of the MAL genes responsible for the metabolism of maltose (De Guidi et al. 2023). Certain important sourdough-related yeasts including S. cerevisiae and Pichia kudriavzevii may even display pathogenic behaviours depending on strain identity (Flörl et al. 2025).

The following sections of this mini-review each cover the description and cultural context of four fermented foods. For each case study, we describe the abiotic factors influencing their microbial communities and their constituent structures and interactions.

Case studies in fermented foods

This section introduces and explores four fermented beverages and foods to explore effects of raw material composition, inoculum sources, ecosystem composition, and functional properties through four fermented beverages and food. These case studies are representative of reflect broad variations in food fermentation technologies in terms of substrate composition, geography, and process techniques.

The first product discussed here is fermented palm sap, or palm wine. This is a traditional alcoholic beverage made based on traditional and local knowledge across tropical and subtropical countries in South Asia, South–East Asia, Central America, and Central Africa (Hebbar et al. 2018, Astudillo-Melgar et al. 2019, Djeni et al. 2022, Sumerta et al. 2025). It contributes to the development of regional socio-economic and culture-tradition dynamics, and is involved in activities such as weddings, offerings, and rituals where it is often served as the main refreshment (Djeni et al. 2022, Wijaya et al. 2024, 2024). Palm wine is made by fermenting palm sap, which is collected by tapping palm trees. There are seven genera of palm tree that are tapped for sap production across palm wine-producing countries including Cocos spp., Borassus spp., and Elaensis spp. (Sumerta et al. 2025). Their sap is also used for many valuable derivatives, including palm sugar, alcoholic beverages, and palm vinegar (González and De Vuyst 2009, Somawiharja et al. 2018, Limtong et al. 2020, Zongo et al. 2020, Wijaya et al. 2024, Sumerta et al. 2025). The tapping is performed by cutting the apical structure of the palm tree, the blossom tip, or shoot tip, and letting the sap collect overnight, although specialized harvesting practices vary by geographical region (Sumerta et al. 2025). Fresh palm sap is translucent, has a white-yellowish colour, and is sweet in flavour (Chandrasekhar et al. 2012). This sweetness is due to its sugar content, which is largely made up by sucrose, followed by fructose and glucose (Naknean et al. 2010, Hebbar et al. 2018, Limtong et al. 2020). Palm sap also contains organic acids, vitamins, proteins, phenolics, and minerals, which serve as a rich nutrient source for fermenting microorganisms (Barh and Mazumdar 2008, Naknean et al. 2010, Xia et al. 2011, Zongo et al. 2020, Das and Tamang 2023).

Grape wine, typically referred to simply as wine, also has a rich history and is one of the oldest fermented beverages, holding deep cultural significance. Archaeological and chemical findings suggest that winemaking dates back to the Neolithic era, with some of the earliest evidence coming from the South Caucasus and the Zagros Mountains over 7000 years ago (McGovern et al. 2003). Over the centuries, various production techniques have emerged, shaped by cultural traditions, leading to the diverse array of wines produced today (Mills et al. 2008). Over time, it has been recognized that wine quality is not only influenced by the winemaking process, but also by factors related to the soil, the grapes, the winery equipment, and, thus, by geographical origin. This has in turn led to the concept of ‘microbial terroir’, which highlights how local microbial communities play a vital role in defining a wine’s character and uniqueness (Bokulich et al. 2014).

Sourdough bread is a cereal-based fermented food produced by the leavening of dough with a sourdough starter. This is a fermented mixture of ground cereal grains and water containing a community of yeasts and predominantly LAB, although acetic acid bacteria are also often present (Landis et al. 2021). While there are several types of sourdough starters, traditional starters are the main focus of sourdough microbial ecology research (De Vuyst et al. 2014). The use of sourdough starters to leaven bread dates back several thousand years, with evidence of bread containing yeasts and LAB found in middle-late Bronze age Egypt and a cemetery from 500 to 300 bce in Xinjiang, China (Carbonetto et al. 2018). Moreover, cereal-based food fermentations are among the oldest fermented foods according to available archaeological evidence (Gänzle 2022). Sourdough fermentation remained the primary method of bread leavening until the advent of technologies to produce pure yeast cultures in the 19th century (Bigey et al. 2021, Gänzle 2022).

Lastly, tempeh (or tempe) is a plant-based fermentation product originating from Indonesia (Romulo and Surya 2021, Watanabe et al. 2023, Rizzo 2024, Wicaksono et al. 2024a). Soybean is the most popular ingredient for producing tempeh, but some regions in Indonesia also use tofu dregs and other legumes, such as peanuts or jack beans, with some varieties mixed with shredded coconut meat left over from coconut oil production. There are many tempeh variants depending on the main ingredient used; tempe gembus (jack beans), tempe mlanding (white leadtree beans), tempe benguk (velvet beans), tempe bungkil kacang (peanut), and tempe bongkrek (soybean mixed with coconut dregs) (Romulo and Surya 2021). Tempeh has been a part of Indonesian cuisine, especially Javanese cuisine, for decades. It is served as an appetiser, snack, or main course, and more recently as a dessert. It is believed that tempeh-making practices were brought to Indonesia by Chinese traders in 1000 ad and established around 1600–1800 ce in Central and East Java (Romulo and Surya 2021, Sjamsuridzal et al. 2021, Tamang Jyoti et al. 2022). The key steps of the tempeh production process are soaking the soybeans, dehulling, starter inoculation, and a 48-h fermentation (Kadar et al. 2020, Romulo and Surya 2021). Now, tempeh is widely consumed in Indonesia and has gained global popularity, especially as plant-based meat substitutes increase in popularity in Western countries.

Raw materials and inoculum sources

The properties of fermented products are partially determined by their raw materials, which sustain the fermenting microorganisms. A substrate’s nutrient composition can select for some microbes over others due to differing metabolic preferences and limitations. These materials may also carry their own microbial ecosystems that are modified by human-led fermentation processes.

This is the case for palm wine, whose fermentation is initiated by naturally occurring microbial communities without additional inoculum, which often results in uncontrolled and unstable palm wine quality. The microbes inhabiting the palms, the species of palm tree, insect visitation, and tapping technique can all alter the composition of the microbiota driving the fermentation process (Djeni et al. 2020, 2022, Sumerta et al. 2025). Insects such as bees and flies are often found in the liquid of the collection vessels, and the use of tapping tools and human hands can transmit fermenting microbes to palm sap. An integrated effect is also simultaneously contributed by ecological factors, e.g. climate and weather to shape microbial structure, implying product differentiation (Bokulich et al. 2016, Astudillo-Melgar et al. 2019, Liu et al. 2019). For example, sap tapping in certain seasons can reduce the production of palm sap and affect microbial load and sap acidity (Ziadi et al. 2014). Microorganisms inhabiting plants are sensitive to changing environmental conditions (Waldrop and Firestone 2006, Ding and Melcher 2016, He et al. 2024) as microbes coexist within plant microecosystems (Classen et al. 2015). If temperature variation is high, the microbial proteomic profile can be altered (Greenblum 2024), suggesting that macroecological factors can play a significant role in fermentation behaviour (Liu et al. 2019).

Like palm sap, grape must is a sugar-rich environment in which different species of yeasts, bacteria, and moulds coexist. Yeasts can metabolise grape sugars into ethanol, carbon dioxide, and other secondary metabolites (Fleet 2003). The genus Saccharomyces, especially S. cerevisiae, usually dominates alcoholic fermentation when ethanol levels increase and nutrients become limited. However, especially at the beginning of fermentation, populations of several non-Saccharomyces yeast genera, such as Hanseniaspora spp., Starmerella spp., and Metschnikowia spp. can develop and influence the wine’s flavour profile (Jolly et al. 2014).

Traditional sourdough starters are defined by the natural or ‘spontaneous’ fermentation that gives rise to their microbial communities, wherein the microbes present are likely to have come from a baker’s hands or the equipment and surrounds of a kitchen or bakery (Minervini et al. 2015, Comasio et al. 2020, Reese et al. 2020). Contrary to popular belief, yeasts and LAB in sourdough starters are unlikely to originate from flour (Gastrow von et al. 2022). These communities are sustained through a process known as ‘backslopping’, which is regular refreshment or refeeding of the ‘mother’ or ‘chief’ starter with new flour and water (De Vuyst et al. 2014, Carbonetto et al. 2018). Over time, this process leads to the stabilization of the microbial community, whose constituents acidify, leaven, and flavour the dough, contributing to the desirable properties of sourdough bread (De Vuyst et al. 2014). Different sourdough production techniques varying by temperature, the proportion of water in the starter, the starter’s age and refreshment frequency, pH, and flour nutritional composition all affect the starter’s microbial composition and physicochemical properties. Taken together, these are now considered superior predictors of a sourdough’s microbial ecology and functional properties compared to previous theories of a geographic or terroir-like influence (Van et al. 2017, Landis et al. 2021). When sourdoughs are prepared with traditional materials, namely wheat or rye flour, maltose is the principal small carbohydrate available for consumption by the sourdough microbiota (Carbonetto et al. 2018). The yeasts found in sourdough starters are also able to access glucose and fructose that is either naturally present in the dough or the product of hydrolysis of dietary fibres found in flour, namely fructans (Struyf et al. 2017, 2018, Carbonetto et al. 2018).

In tempeh, unlike in palm wine, wine, or sourdough starters, moulds play a dominant and crucial role in the fermentation process. These often come from commercial starter cultures, which are used in production at both small and larger scales (Kadar et al. 2020). During fermentation, mould proliferates in a warm and humid environment, forming a mycelial network to bind cooked and de-hulled soybeans together, thereby creating a firm texture. The quality of tempeh can be determined by various factors, such as raw material, pH, moisture, and temperature. However, among those factors, the starter culture used to inoculate the raw materials makes the principal contribution to tempeh’s metabolite profile (Kadar et al. 2020).

Microbial ecosystems

Phenomena such as domination and specialization in microbial communities can determine the flavour and aroma of palm wine. Even though Saccharomyces tends to dominate palm wine microbial communities though specific mechanisms such as direct cell contact and the secretion of chemicals (Perrone et al. 2013, Albergaria and Arneborg 2016), a shift in dominance between Saccharomyces and non-Saccharomyces has been observed over the course of palm wine fermentation, which can be seen in the decrease of diversity indices (Djeni et al. 2022, Das and Tamang 2023). Genetic variations between strains of S. cerevisiae show that its activities are also considerably determined by its microenvironment and interactions with other microorganisms (Jouhten et al. 2016). Palm wine S. cerevisiae strains, unlike grape wine and saké strains, are considerably exposed to natural interactions, and it is unclear whether they can be considered ‘domesticated’ (Fay and Benavides 2005). Their homozygosity is also more evident in wild strains (Ezeronye and Legras 2009). Some genetic admixtures of palm wine’s Saccharomyces strains are also related to different continents, niches, and indicated strain migration through human interference (Ezeronye and Legras 2009, Tra Bi et al. 2019). For example, S. cerevisiae global population demonstrates ecological isolation over all food niches, constructing a very distinct population structure pattern (Legras et al. 2007, Liti et al. 2009). Genetic differentiation shows the domesticated lineages of S. cerevisiae within wild species, linked to environmental adaptation and human-food production history (Legras et al. 2007, Duan et al. 2018, Bigey et al. 2021, Bai et al. 2022).

Saccharomyces cerevisiae is similarly the main yeast responsible for grape wine fermentation, and is thought to have been gradually domesticated through consistent use and environmental adaptation (Goddard et al. 2010). The microbial communities responsible for grape wine, however, contain yeasts and bacteria that both contribute to the transformation of the substrate through their metabolic processes. For instance, Oenococcus oeni, a key lactic acid bacterium in malolactic fermentation, contributes to microbiological and chemical stabilization, deacidification, and sensory impacts, and has evolved to thrive in the challenging conditions of winemaking (Betteridge et al. 2015). Traditional winemaking practices have helped select unique microbial ecosystems closely linked to specific regions and their associated practices (Goddard et al. 2010, Betteridge et al. 2015). Furthermore, the physicochemical conditions of the grape must, such as sugar concentration, pH, temperature, and oxygen levels, are crucial in shaping how microbial succession progresses during fermentation (Cocolin et al. 2011). Temperature affects the competition between different microbial species present and their growth kinetics: lower temperatures are conducive to greater biodiversity and a richer flavour profile, while higher temperatures are conducive to faster but less complex fermentations (Fleet 2003). Since wine fermentation is a dynamic process characterized by microbial succession, the environment during malolactic fermentation is even more selective, requiring tolerance to high ethanol levels and low pH. Typically, LAB genera such as Lactobacillus spp., Pediococcus spp., Leuconostoc spp., and Oenococcus spp. can survive in these environments (Mills et al. 2008). Wines with a pH above 3.5 select for Lactobacillus and Pedoicoccus species both during and after fermentation, while in wines with a pH below 3.5, the bacterial genus that usually takes over and can survive is O. oenis (Mills et al. 2008). Geographic origin, understood as the sum of climatic factors, soil, and grapes, is increasingly recognised as a key abiotic determinant of vineyard-associated microbial ecology. This concept, often referred to as ‘microbial terroir’ has gained empirical support from large-scale metagenomic and amplicon sequencing studies. Bokulich et al. (2014) demonstrated that grape and must microbial communities across California vineyards exhibit clear regional patterns, correlating with factors such as temperature, precipitation, and grape variety. Similarly, Coller et al. (2019), in their investigation of Northern Italian vineyards, found that soil microbial communities were significantly shaped by geographical gradients, including altitude, soil pH, and organic matter content. These regionally distinct microbial signatures persist into fermentation and have been linked to variations in wine chemistry and sensory properties, reinforcing the notion that microbial terroir plays a functional role in defining regional wine styles (Bokulich et al. 2016, Liu et al. 2020).

Most mature sourdough microbial communities are described as simple, containing approximately three bacterial and one yeast species (Van Kerrebroeck et al. 2017, Landis et al. 2021). In addition to the species diversity of sourdough bacteria being slightly higher than that of yeasts, bacterial cell counts are usually 10–100 times greater than yeast cell counts (Scheirlinck et al. 2007, De Vuyst et al. 2014). When Fructilactobacillus sanfranciscensis is found in a sourdough, it tends to be the dominant LAB species present, suggesting that in suitable conditions it is a highly competitive member of the sourdough microbiome (Van Kerrebroeck et al. 2017, Comasio et al. 2020, Landis et al. 2021). Saccharomyces cerevisiae is similarly dominant, although this is more likely to be due to its high prevalence in bakery environments as well as its strong suitability to cereal fermentations (De Vuyst et al. 2014, Legras et al. 2018). However, sourdough ecosystems typically encompass stable associations between yeasts and LAB. These are often linked to specific preferences for carbohydrates; the most well-known association is that between Maudiozyma (formerly Kazachstania or Candida) humilis and F. sanfranciscensis (De Vuyst et al. 2014). The basis of this relationship is presumed to be firstly that M. humilis relies on F. sanfranciscensis to provide small sugars, as the former is maltose-negative, and secondly that M. humilis is adapted to the acids produced by F. sanfranciscensis (Jacques et al. 2016, Wittwer et al. 2022). The amount of protein and alpha- and beta-amylases in flour can also affect sourdough yeast and LAB population composition, with a particular effect on fermenting yeast identity (Oshiro et al. 2019). When kept at 37°C, however, sourdough starters tend to contain few to no yeasts, either due to the yeasts’ inability to grow or improved LAB competitiveness at these temperatures (Vrancken et al. 2011). Furthermore, adjustments to the storage temperature between 20°C and 37°C affect the predominance of different LAB species, with Lactobacillus fermentum favoured by higher temperatures, and Leuconostoc citreum and F. sanfrancisensis favoured by lower maximal temperatures (Vrancken et al. 2011, Landis et al. 2021). The duration between sourdough starter refreshments or instances of backslopping also affects microbial community composition. Primarily, the depletion of maltose present in wheat dough occurs to a greater extent as backslopping frequency decreases (Vrancken et al. 2011). The low pH and nutrient exhaustion that characterize sourdough starters many hours from their last refreshment can also favour the emergence of species such as Lactiplantibacillus plantarum, thereby shifting population composition (Vrancken et al. 2011). In the case of spontaneously started sourdoughs, ~10 refreshments are required for lactobacilli to become dominant (Oshiro et al. 2020). Moreover, over the course of sourdough maturation, different groups of LAB predominate in successive ‘waves’ (Oshiro et al. 2019). Among sourdough starters produced domestically, the older a starter is, the more likely it is to contain F. sanfranciscensis (Landis et al. 2021). Further investigations as to the nature of these dominant yeasts and bacteria is needed to fully understand dynamics of fermentation.

The fermentation of tempeh is mainly conducted by mould but is often assisted by bacteria and yeasts. There are several mould species used in tempeh starters, such as Rhizopus arrhizus, R. delemar, R. stolonifera, and R. microsporus (formerly oligosporus). However, since the commercialisation of tempeh production, R. microsporus is the predominant species used in Indonesian tempeh manufacturing (Sjamsuridzal et al. 2021). Genera from the LAB group including Leuconostoc, Limosilactobacillus, Lactobacillus, Enterococcus, Paenibacillus, Azotobacter, and Enterobacter have been found in tempeh (Wicaksono et al. 2024b) and the yeast S. cerevisiae is also often observed (Rizal et al. 2021). There is little information about the nature of interactions between fungi and bacteria in the tempeh ecosystem, and this should be an area of further work.

Functional properties of fermenting microbes

Fermenting yeasts and bacteria are directly responsible for producing the ethanol that is crucial to the production of alcoholic beverages like palm and grape wine. Saccharomyces cerevisiae has a prominent role in palm wine ethanol production and is often assisted by the ethanol-producing bacterium Zymomonas mobilis (Djeni et al. 2022, Sumerta et al. 2025). Non-Saccharomyces yeasts further enhance the flavour and aroma of the product (Li and Liu 2016, Pietrafesa et al. 2020). However, these yeasts have a relatively negative response to the presence of ethanol compared to acid-producing bacteria (Sumerta et al. 2025). These microorganisms—such as Lactococcus spp., Lactobacillus spp., and Acetobacter spp.—concurrently perform acidic fermentation by converting sugars into organic acids, leading to a reduction in pH at the end of fermentation (Alcántara‐Hernández et al. 2010, Astudillo-Melgar et al. 2019). This acidic fermentation is often exploited for palm vinegar production.

At the onset of grape wine fermentation, various non-Saccharomyces yeasts like T. delbrueckii, Metschnikowia pulcherrima, Starmerella bacillaris, and Hanseniaspora uvarum contribute to the development of early aromas, enzyme activity, and metabolic interactions. Coinoculation strategies using these species allow for the enhancement of complexity without compromising fermentation reliability. They decrease ethanol content, decrease S. cerevisiae niche monopolization, decrease volatile acidity and acetic acid concentration, and increase concentrations of terpenols (Ciani et al. 2010, Jolly et al. 2014, Benito 2018, Englezos et al. 2022). In parallel, LAB such as O. oeni and L. plantarum play crucial roles in the deacidification of wine through malolactic fermentation. These bacteria interact metabolically with yeast-derived compounds and can either benefit from or be inhibited by byproducts such as ethanol, SO₂, and medium-chain fatty acids. Therefore, the timing and compatibility of yeast and bacteria inoculation are therefore crucial for achieving successful fermentation and desirable sensory outcomes (Du Toit et al. 2011).

A key functional property of the sourdough microbial ecosystem is the acidification of the starter dough (De Vuyst et al. 2014). Values range from 3.4 to 4.9 with an average of 4.1, which forms a key distinction from ‘standard’ bread leavened only with baker’s yeast, which typically has a pH of slightly over 5.0 (Oshiro et al. 2019, Arora et al. 2021, Clark et al. 2024). Dough acidity increases with the duration of sourdough fermentation, and a low pH favours eventual dominance of lactobacilli in sourdough starters (Oshiro et al. 2020, Arora et al. 2021). This phenomenon is likely to be bidirectional, as in simulated sourdoughs with relatively high pH values (5.5–6.7), lactic acid production is observed to increase (Oshiro et al. 2020). The production of organic acids and other compounds by sourdough fermenting microbes is known to an increase in shelf-life through the inhibition of spoilage fungi and retardation of starch retrogradation (Arora et al. 2021).

Tempeh is rich in diverse nutritional substances with hypoglycaemic, antioxidative, and blood lipid control properties (Abdurrasyid et al. 2023, Watanabe et al. 2023, Rizzo 2024). The high content of protein, isoflavones, and dietary fibres in tempeh that contribute to its hypoglycaemic effect may be generated or modified by the fermentation process. Microbial community composition has also been shown to improve beta-glucan production in tempeh (Rizal et al. 2022). Isoflavone aglycones and hydroxylated compounds—which are associated with the antioxidative capabilities of tempeh—are also produced during fermentation (Watanabe et al. 2023 ). The bioavailability of minerals and vitamins is increased by fermentation (Kadar et al. 2020, Romulo and Surya 2021, Rizzo 2024,). Furthermore, tempeh fermentation increases the concentration of several proteins and lowers fat content, helping to optimise protein and caloric intake (Rizal et al. 2022).

Conclusion and outlook

Although most defined food fermentation processes were developed long ago, they continue to be a mainstay of modern human diets worldwide. Studies in the field of microbial ecology have revealed the complex and often interdependent biological and enzymatic contributions of the bacteria, fungi, and viruses responsible for the desirable properties of these fermented foods. Moreover, the industrialization and consequent standardization of fermented food production has led to advances in our ability to precisely control and modulate the activities of microorganisms to prepare fermented foods with organoleptic, nutritional, and safety properties that are optimal and consistent. This research and technological progress yield new insights into food fermentation phenomena that can be applied in various ways to the design of novel food. As human lives continue to be shaped by global forces that affect the modes and efficacy of processes within the agricultural and food manufacturing industries, such new products and the techniques involved in their production are likely to become important to the problem of feeding the world’s population (Table 1).

Table 1.

Overview of the fermented food case studies outlined. The raw plant materials are given, along with the known microbial inocula to create the complex ecosystems shown as cartoons. The functional properties that the microbes give to the food are detailed in the final column. Detailed references are given in the appropriate section in the text.

	Raw materials	Inoculum source	Microbial ecosystem	Functional properties
Palm wine	Palm sap	Humans, sap collection vessels, palm trees, soil, and insects,	Yeasts (Saccharomyces cerevisiae), Zymomonas mobilis	Ethanol production Flavour compounds Acidification Antimicrobial activity
Sourdough bread	Flour, water	Humans, bakery equipment, and surrounds	Yeasts (Saccharomyces cerevisiae, Maudiozyma humilis), LAB (Fructilactobacillus sanfranciscensis, Lactobacillus fermentum, Lactiplantibacillus plantarum)	Dough rise Flavour compounds Acidification
Wine	Grapes	Humans, grape vines, maturation vessels, insects, birds, and soil	Yeasts (Saccharomyces cerevisiae), LAB (Oenococcus oeni, Lactobacillus spp., Pediococcus spp., Leuconostoc spp.)	Ethanol production Flavour compounds Deacidification Enzyme activity Bioprotection
Tempeh	Soybeans	Humans, soybean hulls, water, and fermentation vesels	Bacilli (Bacillus subtilis), Enterobacteriaceae, mycelial fungi (Rhizopus spp.)	Increase digestibility Flavour compounds Phytate reduction Antimicrobial activity

Conflicts of interest

None declared.

Funding

This work has been supported by a Research Training Program scholarship provided by the Australian Government and the University of Melbourne to A.E.W. This research was funded by the Italian Ministry of University and Research (MUR) under the National Recovery and Resilience Plan (PNRR)—Mission 4, Component 2, Investment 1.4, within the framework of the National Centre for the Development of New Technologies in Agriculture (Agritech), Project MUR CN00000022 (CUP D13C22001330005), Spoke 9. I.N.S. is supported by Australia Awards from the Australian Government (the views and opinions do not represent the views of the Australian Government).