Microbial ecologies of novel misos

Here we tell the story of our work over the past years on the microbial ecologies of novel misos. We summarise findings we have published in two journal papers: a more systematic paper (Kothe et al 2024b) that forms the core of the story—what we’ll call the ‘big study’—and an initial, more exploratory paper (Kothe et al 2024a) to support and contrast those findings—what we’ll call the ‘small study’. These papers both build on research conducted as part of Josh’s PhD.

Table of Contents

These experiments might not only be
creating new flavours—they might also
be shaping microbial life, without these
chefs willing or even realising it.

i. Origins

This story begins some years back, probably in 2012, when Josh first started working at Nordic Food Lab—a former non-profit institute based in Copenhagen that conducted open-source research on the edible diversity and gastronomic potential of the Nordic region. Here’s some of the backstory, as told by Josh:

In my first months at Nordic Food Lab I remember tasting dozens, maybe even hundreds of ingredients and products I had never tasted before. Many were fermented. It was around the time the lab, and Restaurant Noma to which it was affiliated, were beginning to explore fermentation more directly. The chefs and researchers I was working with, along with those at some other restaurants at the time, were starting to systematically adapt diverse fermentation techniques to local ingredients to make flavours that had never existed before, which they could use to further develop and specialise their dishes. One way we might describe this approach is as a kind of ‘translated fermentation’, in the same way that a book can be translated from one language into another.

These chefs and fermenters were developing this kind of fermentation for the pursuit of flavour. That alone was fascinating, as was its results. But as I learned more about the microbes that were involved in these processes—how quickly they reproduce and can adapt to new environments—the more it became clear to me that these experiments might not only be creating new flavours; they might also be shaping microbial life, without these chefs willing or even realising it. And if that were happening, it could be having biological and social implications far beyond their kitchens.

That idea—was this flavour-driven transformation of microbial life happening, if so, how, and why did it matter?—became the seed of my subsequent PhD research. And when it came time to figure out which fermented products to study, my collaborators and I all thought miso would be ideal. Yellow pea miso, or ‘peaso’, was one of the first translated fermentations developed at the lab and Noma back in 2011/2012, followed over the years by dozens of other kinds, all using traditional Japanese techniques adapted to Danish ingredients to produce new (incidentally plant-based) sources of umami with respect to traditional Danish cuisine. And, in addition to being one of the restaurant’s first and by then most variegated ‘translated’ fermentations, as relatively complex fermentations involving multiple stages and, at different points, various bacteria, multicellular fungi, and yeasts, misos offered a complex and relatively unexplored microbiological richness to test the theory.

Josh started by sampling some misos already made at Noma (explored in the small study), and proceeded to design a more systematic experiment conducted in the Noma Fermentation Lab (the big study). In both studies, we used DNA sequencing to investigate the microbial ecologies of these misos. Here we describe some of the patterns that we found.

ii. Experimental design

In the small study, we sampled six finished misos made in the Noma fermentation lab, to get some preliminary data on novel miso ecology. This was the first time anyone had used metagenomics to study miso ecology, so our aim was exploratory. In the big study, our aim was to investigate different factors that could shape the microbial ecology. So we designed an experiment, using eight different substrates, with three replicates (‘triplicates’) for each one, to investigate the effects of four different factors: (1) the ingredients used as the proteinous substrate; (2) nixtamalisation, a technique from Mexico whereby grains are soaked and cooked in limewater to improve their nutrition, texture, flavour and aroma; (3) the length of fermentation; and (4) geography. The latter was based on six misos we received from Inua, then a sister restaurant to Noma, in Tokyo. We did not have starting samples or triplicates for these misos, so the discussion of geographical effect is more tentative.

In both studies, we used metagenomics to identify the microbes in the samples (who is there) and conduct functional analyses (what can they do). In the big study we also used metabarcoding, as a preliminary analysis, before doing metagenomics later to corroborate and refine the analysis. We used the metabarcoding data for statistical analysis of the misos’ diversity and the relative effects of the four factors.

iii. Miso preparation

We tested a range of different misos, made using a variety of ingredients and preparation techniques. The misos from Noma focussed on Nordic ingredients; the misos at Inua featured Nordic ingredients (eg. yellow pea) as well as Japanese ones. Figure 1 outlines the misos tested in each study.

Figure 1: Misos tested in each study, categorised by ingredients and preparation. All misos used barley as the kōji substrate, except those marked with an asterisk (*), which used rice. S=standard preparation; N=nixtamalised.

All misos were prepared according to the same standard recipe.¹ 

Ingredients

• 3 part protein-rich substrate

• 2 parts kōji (Aspergillus oryzae grown on pearled barley or rice)²

• 4% salt of combined weight

Method

1. Soak, rinse, and boil protein-rich substrate until al dente. Cook the nixtamalised samples in a solution of 0.1% calcium hydroxide (CaOH) and let soak overnight in the cooking liquid, then rinse and drain.

2. Grind or blend protein-rich substrate with kōji, then add salt, and mix well.

3. Pack the mixture into a clean vessel. Sprinkle salt over the top layer, before weighting down and covering the mixture.  

4. Ferment at 22-30℃ for 3 months.

iv. Factors influencing microbial ecology

All of the misos contained A. oryzae (as we would expect, as it was added), as well as Bacillus amyloliquefaciens, possibly also present in the kōji starter and an important bacterium found in many plant-based fermentations that, like A. oryzae, helps degrade starch and create flavour. Beyond these similarities, the miso ecologies varied, to different degrees, according to the four factors studied: substrate, nixtamalisation, time, and geography. These are not the only factors that could be shaping the misos’ microbial ecology; others could include the kōji starter and preparation method, the salt type and concentration, the environment, the equipment, the fermenters’ bodies, the fermentation conditions and length, and more. But these four felt like a good place to start.

Substrate

The ingredients used to make each miso appear to be the most influential factor shaping microbial ecology, particularly for bacteria. This effect may be explained by the different microbes already present in each substrate before fermentation, and each substrate’s different molecular composition in turn favouring certain microbes over others.

Three main clusters of bacterial composition were found across the misos in the big study, grouped based on the dominant species:

1. Pediococcus pentosaceus was dominant in the nixtamalised samples, plus the fava bean and habanero-barley misos. This species is commonly associated with raw ingredients like beans and cereals and is well-known in fermented plant foods.

2. Co-dominance of the species P. pentosaceus and Staphylococcus pasteuri was found in the Gotland lentil, yellow pea and fava bean misos. S. pasteuri has also been found in douchi, a Chinese product of fermented black soybeans.

3. Staphylococcus epidermidis was dominant in the traditional soybean (control) misos. This species is more commonly associated with the human skin microbiome than fermented foods, and may have originated from producers' hands. We also detected this species in the small study.³

Many other bacterial species were found in all the misos.

For fungal composition, most misos were unsurprisingly dominated by A. oryzae from the kōji, though some misos showed co-dominance with other species of the Saccharomycetales family. It should be noted here that dominance of A. oryzae DNA does not necessarily mean dominance of A. oryzae living in the miso. In theory, the salt, anoxic conditions, and dropping pH of the miso mixture would all select against A. oryzae, and in favour of bacteria like lactic acid bacteria (LAB) and salt-tolerant yeasts. This is a good example of where culture-dependent methods could complement the culture-independent ones, by offering more direct evidence of who is actually living in and transforming the miso through their metabolism.

Nixtamalisation

The microbial ecologies of the nixtamalised misos were more similar to one another than to any of the other misos, with a higher abundance of P. pentosaceus and Lactobacillus plantarum compared to the non-nixtamalised ones. We found a similarly distinct microbial ecology in the nixtamalised yellow pea miso in the small study, containing P. pentosaceus and even a potentially new species of Exigobacterium, which we discuss further below. These taxa likely became co-dominant because they were better adapted to the misos’ specific alkaline and specific nutritional environmental conditions.

The chemistry of nixtamalisation suggests a potential mechanism for this hypothesis. The alkaline treatment employed during nixtamalisation converts the hemicellulose components of the cell walls of grains and legumes into soluble gums, which improves the release and accessibility of nutrients such as vitamins, minerals and other bioactive compounds. The resulting simpler carbon sources and the compounds created by the treatment can promote the growth of Pediococcus, known for its ability to thrive at various pH levels. In contrast, the standard misos show a higher abundance of Staphylococcus. Species of this genus can secrete enzymes to break down complex sugars and starches, thus able to utilise a wider range of carbon sources than other taxa, which may have helped it outcompete species such as Pediococcus in the non-nixtamalised miso environment.

Time

A clear overall pattern of fermentation time emerged in the big study, where we had both start and end samples. The bacterial composition in most samples exhibited a pronounced shift over time, moving from Enterobacteriaceae-dominated communities in the starting samples to ones containing primarily LAB and members of the Staphylococcaceae family after three months of fermentation, with some variation between substrates. Other studies focusing on fermented plant-based substrates, such as coffee and sourdough, have also consistently demonstrated a similar finding. Starting communities are usually very bacterially diverse, with taxa likely introduced from the ingredients, equipment, fermenters’ bodies and surrounding environment, including Enterobacteriaceae (typically associated with gut microbiomes), and the fermentation process yields a more selective microbiota. Bacterial diversity in all misos correspondingly decreased over the fermentation period, as a result of this selection.

Fungal diversity, on the other hand, generally increased over the fermentation, shifting from kōji-derived A. oryzae-dominated communities in the starting samples to more diverse communities of fungi better adapted to the changing environment in the misos, as discussed above in the ‘Substrate’ section. This increase in fungal diversity over time illustrated in the statistical analyses may not be so apparent in the metabarcoding and metagenomics data, as a result of the abundant A. oryzae and bacterial DNA respectively overwhelming that of the other fungi.

Geography

The big study was the only one with geographical variation. For this study, geography was defined as the physical location where the fermentation took place: Copenhagen or Tokyo. Of the four factors that we tested, this one appears to have the weakest influence on the bacterial communities of the misos, with a weak overall effect and no distinct patterns in composition. For fungal communities, the signal was stronger than the other factors, but still comparably weak as for the bacteria. However, since only one type of miso was common between those made at Noma and Inua, and replicates weren’t made of the Inua misos, only limited inferences can be made from this study about the geographical effect.

The misos in the small study appeared more diverse than those from Noma in the big study, despite both being made at Noma. This difference suggests that another scale of geography—the environmental conditions of the kitchen where the misos were prepared—may have a greater influence on microbial ecology than larger scales like city. The misos made at Noma from the small and big studies were made in two different locations, at different times, under the creative direction of two different people, at different moments in the restaurant’s stylistic evolution. In the small study, the misos were made in 2017, in Noma’s original fermentation lab, which at the time was housed in a few shipping containers behind the restaurant, and more open to the environment. This was under the direction of a chef with a more tinkering approach, at a moment in the restaurant’s history when slightly more variability was tolerable. The misos in the current study, meanwhile, were made in 2018, after the restaurant had moved to a new location in a brand-new building (‘Noma 2.0’), in a purpose-built fermentation lab with a more controlled environment. This was under the direction of a different chef, with a more controlled, scientific approach, and at a moment in the restaurant when consistency was becoming ever more important. These different micro-geographic circumstances and approaches could have shaped levels of diversity in the misos accordingly. A similar difference in micro-geography may have been at play in the higher diversity of the misos at Inua compared with those at Noma 2.0. Finally, such a micro-geographic perspective may also help explain the apparent overall greater diversity of all the misos across both studies compared to what is currently documented in the (admittedly scant) miso literature. The fact that these misos were all made in restaurant kitchens, where many kinds of fermentation are produced alongside each other, could be one reason why these novel misos exhibit such relatively high microbial diversity.

v. Reflections

Beyond the relative contribution of different factors to shaping microbial ecology, these studies offer some more general glimpses into the science of translated fermentations: for example around microbial terroir, hand taste, possible mechanisms of strain adaptation, and the emergence of novel species. 

Despite its limitations that constrain what we are able to say about geographical effect, the big study suggests that this effect is complicated, and varies depending on the scale of analysis—both biologically, varying in strength across genus, species, strain, and gene, and geographically, varying across different spatial scales and kinds of spaces. How geographical effect and its variation relates to the popular concept of ‘microbial terroir’ depends in turn on what is meant by the latter. If it should mean a unique microbial community, or even a community of unique microbes, associated to a particular place, this is probably less likely to be the case. While there is enormous and yet unplumbed microbial diversity in all planetary environments, this diversity is structured and distributed according to certain patterns. In the case of miso, as with many fermented foods, these patterns seem to have most to do with ‘treatment’—what one does to the food and its fermenting environment, and how these practices shape the food and its microbial ecology. The same has been documented in, for example, cheese rinds.⁴ So, rather than understanding ‘terroir’ to refer to some kind of unchanging, pre-existing source of pure nature, it might make more sense, if we wish to use the concept, to understand it as an historical, natural-cultural happening, in dynamic feedback with human practice.⁵ It might also be best understood as both unfolding and varying across scales, such that micro-geography might often be a better lens through which to explore it.

If microbial terroir is about the links between a fermenting food and its environment, these studies suggest related links between food microbes and human body microbiomes. The presence of S. epidermidis in some of the misos is particularly telling here, since this species is more typically associated with the human skin microbiome than with fermented foods. It may have originated from the producers’ hands, potentially being transferred to the kōji or into the miso mixture itself during preparation. This finding naturally made us think of the Korean concept of son-mat or ‘hand taste’, in which the maker's own microbiome influences the microbiome, and sensory properties, of the final product.

Yet we did not only find S. epidermidis present in some of the misos, but also that some of its strains exhibited adaptation to the different miso environments. In particular, strains of S. epidermidis in the yellow pea misos made at each site were closely related, while another strain found in the maitake-soy miso at Inua was much different: a clear example of substrate effect at the strain level. These genetic adaptations could have arisen through selective pressures exerted by these substrates, a phenomenon also documented in other studies. The genes themselves could have come from the miso substrates, other microbes in the miso communities, and/or pre-existing genetic diversity in local microbial populations. We can’t yet say for sure, and this question would require further study. But we can say how S. epidermidis’ evident adaptation to the miso niche is an illuminating example of the benign, even generative microbial traffic between humans and fermented foods, which might be much more normal in fermented food microbiomes than currently thought. This new way of understanding fermentation microbiomes as part of the ‘extended human microbiome’ also fits nicely into the emerging ‘holobiont’ paradigm in the life sciences, in which organisms are increasingly reconceptualised as complex ecological assembles of ‘host’ and associated microbiota, each shaping the other to give rise to the biological individual.⁶ Aside from gesturing to the messy, enriching symbioses normal rather than exceptional to the web of life, such biological novelty emerging within holobiotic fermentation microbiomes likely also has functional impacts on nutrition and flavour—another promising avenue for further research.

Finding this species in the nixtamalised
yellow pea miso is a perfect example of how
translated fermentation opens up niches for
new microbial communities to emerge.

It is not only new strains that are emerging in novel fermentation. In the small study, we document the possible discovery of a novel species of Exiguobacterium, which we found in the nixtamalised yellow pea miso. This genus was first found in an alkaline potato processing plant, and most of its species are so-called ‘extremophiles’, fond of (to us) extreme environments. The only other time a member of this genus has been found in fermented food was in pozol, a Mexican fermented beverage, made by fermenting nixtamalised corn dough (masa), diluting it with water or milk, and sometimes adding sweeteners like sugar or honey and seasonings like cocoa powder or dried chili. You might be detecting a theme—all of these instances of the genus are from alkaline environments. 

Since publishing the small study and offering this tentative finding, we have checked other microbial genomic databases, and found that the species we detected might not be genetically different enough from any other to constitute a new one.⁷ Such is the way with scientific discovery. We are looking further into it. Whether or not it turns out to be an entirely new species, our finding this species in the nixtamalised yellow pea miso, the first time this genus has been detected in any miso, is a perfect example of how translated fermentation—in this case, marrying techniques from Japan and Mexico with ingredients from Denmark—opens up niches for new microbial communities to emerge.

vi. Further directions

Many further directions emerge from this miso research, including:

• more direct and thorough testing of geographical effect, at different biological and geographic scales, with robust design and replicates;

• more sampling of the misos over time to investigate changes over the course of the fermentation

• additional analyses—physical (eg. more detailed pH measurements), nutritional, enzymatic, metabolomic, and sensory—to investigate specific mechanisms underlying the dynamics of microbial composition and genetics in miso fermentation, and to identify the functional roles of individual microorganisms and microbial interactions in shaping the sensory and nutritional properties of the final product; and

• investigating all these questions for traditional misos as well, so that we have something to compare the more novel products to—for example, to be able to more robustly compare levels of diversity.

And that’s just for misos—we still have much more to learn about the microbial ecologies of translated fermentations of all kinds.

vii. Read the original research

Caroline Kothe, Christian Carøe, Florent Mazel, Pablo Cruz-Morales, Naher Mohellibi, Joshua Evans (2024b). ‘Novel misos shape distinct microbial ecologies: opportunities for flavourful sustainable food innovation’. Food Research International.

Caroline Kothe, Jacob Agerbo Rasmussen, Sarah Mak, M Thomas Gilbert, and Joshua Evans (2024a). ‘Exploring the microbial diversity of novel misos with metagenomics’. Food Microbiology.

Joshua Evans (2022) ‘Taste Shaping Natures: a Multiplied Ethnography of Translated Fermentation in the New/er Nordic Cuisine’. Oxford University Research Archive. 

For more of our published scientific work, see publications.

Contributions & acknowledgements

Eliot prepared the first draft and overall structure of the article based on discussions with Josh. Josh then expanded it and developed the story further, bringing in more details from the two papers. Josh, Caroline and others performed the original research that it is based upon.

Josh and Hugh Allen photographed the misos at Noma. Josh took the landscape ones and Hugh took the portrait one.

Eliot created the Venn diagram graphic.

Related posts

Endnotes

[1] The standard Noma miso recipe is documented in: David Zilber and René Redzepi (2018), The Noma Guide to Fermentation, Artisan, New York, USA.

[2] Per standard recipe in: David Zilber and René Redzepi (2018), The Noma Guide to Fermentation, Artisan, New York, USA.

[3] It should be noted that all these Staphylococcus species are safe, and that other Staphylococcus species have been found in miso, such as S. gallinarum and S. kloosii.

[4] Ben Wolfe et al. (2014), ‘Cheese rind communities provide tractable systems for in situ and in vitro studies of microbial diversity’, Cell. 

[5] Amy Trubeck (2009), The Taste of Place: A Cultural Journey into Terroir, University of California Press, USA; Philippe Prévost et al (2014), ‘Terroir, a concept for action in territorial development’, Articulated and transversal food systems for food security; Adrien Rigobello and Josh Evans (2024), ‘Design terroir: An eco-social, relational, bioregional approach to design’, SSRN. Design Research Society Conference 2024.

[6]  Rob Dunn et al. (2020), ‘The Internal, External and Extended Microbiomes of Hominins’, Frontiers in Ecology and Evolution; Seth Bordenstein and Kevin Theis (2015), ‘Host Biology in Light of the Microbiome: Ten Principles of Holobionts and Hologenomes’, PLOS Biology; Kevin Theis et al. (2016), ‘Getting the Hologenome Concept Right: an Eco-Evolutionary Framework for Hosts and Their Microbiomes’, mSystems. 

[7] There seem to be four isolates of an unnamed species of Exiguobacterium (sp902362975 / sp018617385) registered on gtdb with an ANI to our species of 97.27%, isolated from the human gut, a ruminant gastrointestinal tract, the Jet Propulsion Laboratory cleanroom, and an undisclosed source on Sakhalin Island. Thanks to Elisa Caffrey at the Sonnenburg Lab for taking the initiative to check this for us, and for uncovering this tantalising and bewildering list of sites for purportedly one microbe.