The miraculous biodiversity bubbling in your sourdough starter

I made it through four weeks of Los Angeles’ COVID-19 shelter in place order before I climbed aboard the isolation sourdough bandwagon. It took more effort than expected just to stay on.

I followed a protocol provided on the King Arthur Flour website to ferment my own starter from flour mixed with water. Twice a day, from March into April, I transferred a bit of starter to a fresh jar, and added fresh flour and water, with no apparent progress. I moved the starter around my little kitchen, searching for a sufficiently warm nook. I restarted the whole process from scratch. I ran through two five-pound bags of all-purpose flour I scrounged at a neighborhood grocery store (everyone else in Los Angeles had taken up baking, too), and when, visiting Costco on a friend’s membership, I lucked into a 25-pound sack, I didn’t pause to wonder where I’d find space for it in the pantry.

And then, finally, one afternoon, I checked the jar in the oven (heat off, light on, a nice steady 70°F) and saw that the starter had surged above the rubber band I put around the jar to mark its beginning volume. I baked bread the next day.

I am no manner of microbiologist, but I knew in a general way that sourdough starter is a microbial culture, and the rising is, of course, due to the metabolism of those microbes — carbohydrates in, carbon dioxide out. I also knew enough to recognize the process of establishing a starter as something like a serial passage protocol: sample the microbes in a batch of flour, give them periodic infusions of fresh growth medium, and let them duke it out for supremacy until the population growth rate was enough to leaven a batch of bread. What I am, though, is a molecular ecologist — and so I naturally started wondering what was known about the population and community genetics bubbling in that blue glass jar.

When I first dove into the research on the diversity of sourdough microbes, I thought it would be a story of evolution — adaptation by one or a few microbes present in the flour used for the initial fermentation, with the serial passaging selecting for genotypes that grow the fastest. In fact, the process is understood more properly as community assembly, with many microbial species setting up shop in a starter and jostling for position within a community that eventually stabilizes to produce the leavening effect a human baker wants. This made sense as soon as I thought about it for more than the time it took to formulate a Google Scholar query. No working baker keeps sourdough starter in sterile isolation, and the new growth medium added during the serial passaging — “backslopping” is the term of art — is no more sterile than the flour used at the very beginning. The point being, there are all sorts of opportunities for new microbes to join a starter culture. Mature starter seems to stabilize because it’s a specialized environment that, if engineered correctly, limits the kinds of microbes that can establish.

The leavening effect of a sourdough culture is in carbon dioxide produced by yeast, but the yeast can’t digest all the sugars in flour. That’s taken care of by bacteria, usually lactic acid bacteria, which metabolize those other sugars via anaerobic lactic acid fermentation, with byproducts the yeast can metabolize in turn. In terms of cell count, the bacteria outnumber the yeast by a factor of about 100:1. At that broad level, the process of establishing a “spontaneous” sourdough like mine is really impressively predictable — an initially diverse microbial community becomes rapidly dominated by one or a few yeast species and a larger diversity of lactic acid bacteria, with pH depressed by the bacteria’s acidic waste products. That low pH is why sourdough is sour.

Cell counts for the major microbial taxa in a rye sourdough starter, and pH of the starter culture, over 11 feeding (“refreshment”) cycles. Detail from Figure 1, Ercolini et al. 2013

Within those broad outlines, the microbial community of starters can get bewilderingly diverse. The yeast can be Saccharomyces cerevisae, the same species cultivated for granulated baker’s yeast, but it can also be numerous other taxa, including Candida humilis and Kazachstania exigua. Many of those other yeast species are introduced to the starter with the flour used in the original fermentation, and are actually facilitated by the byproducts of lactic acid metabolism — but S. cerevisae isn’t so lactic-acid-friendly. It’s thought that it often ends up in sourdough starters not because it arrives with the initial ferment but because sourdough starters tend to be produced and maintained in bakeries, where processed baker’s yeast is all over the place.

The bacteria in a starter are even more potentially diverse — species in the genus Lactobacillus are most common, but lactic acid bacteria comprise a whole order of potential acid-tolerant anaerobes. Lactobacillus sanfraciscensis was isolated from a San Francisco sourdough culture that dated back more than a century, and it’s thought to be responsible for the specific taste and texture of that local product, though it turns out to be geographically widespread. It shows signs of adaptation to the sourdough environment — the L. sanfraciscensis genome is unusually small compared to other lactic acid bacteria, and shows signs of recent gene inactivation and loss, which may reflect selection for rapid growth through the serial passages of backslopping.

Relative frequency of 16S rRNA sequences from different bacterial taxa in sourdoughs based on rye (R), Triticum durum (D), or Triticum aestivum (A), from an initial timepoint (0) to the end of a 10-day backslopping propagation (Figure 3, Ercolini et al 2013)

However, in spite of the fact that L. sanfraciscensis beats out other Lactobacillus species in tests of direct competition, there are lots of other Lactobacillus species, and lactic acid bacteria more generally, found in sourdoughs. This suggests that at least some starter diversity is due not to the superiority of particular bacterial species in particular starter conditions, but just dispersal limitation — the dominant members of a starter community are the ones locally available to join a starter, and there may be some priority effects once the starter community is stabilized. One well-cited community genetic study of sourdoughs based on different types of flour traced this stabilization, in which Lactobacillus sequences increased in frequency over the course of a 10-day “traditional” backslopping protocol.

As reliable as the result of sourdough cultivation seems to be (well, except for me), the fact that a typical starter is continuously receiving new community members from the environment seems like a challenge for attempts to link community composition to bread quality. It makes intuitive sense that geographic variation in the bacteria available to join sourdoughs, and the flours used to feed them, could contribute to regional differences in sourdough bread, though. There’s at least one ambitious project underway to survey global starter diversity and link it to bread qualities. The very latest word from that work is that, even when provided with flour from the same source and a common recipe, different bakers produce sourdoughs containing distinctive communities — and many of the microbes in starters likely came from the bakers’ own hands. The authors of that study baked bread from their experimentally produced starters, and found associations between bread sourness and the diversity of starters’ lactic acid bacteria, though not the diversity of their yeast.

Proportion of the microbial community in nine sourdough starters reconstructed as derived from different possible sources: dust and tap water from the bakery, the hands of the baker, or the flour used to make and feed the starter. From Figure S4, Reese et al. 2020.

Sourdough is everyday magic in a very real sense: mix just two incredibly simple ingredients and let them sit, and you’ll conjure a nascent starter. It’s also a pretty impressive feat of applied ecology. Assembling an ecological community from scratch is supposed to be a massive challenge, requiring the understanding of all sorts of emergent properties and nonlinear feedbacks. But with sourdough, a baker creates a fresh habitat and cultivates what grows there to guide succession toward a community that produces a near-universal basic foodstuff. Imagine a set of instructions short enough to fit on a Post-it Note that would let you turn a cleared field into an orchard, without ever deliberately placing a seed or pulling a weed. On a microbial scale, that’s what any of us can do with a little flour and water.

And, well, some patience.


De Vuyst L and P Neysens. 2005. The sourdough microflora: biodiversity and metabolic interactions. Trends in Food Science and Technology. 16:43-56; doi: 10.1016/j.tifs.2004.02.012

De Vuyst L, S Van Kerrebroeck, H Harth, G Huys, HM Daniel, and S Weckx. 2014. Microbial ecology of sourdough fermentations: diverse or uniform? Food Microbiology, 37:11-29; doi: 10.1016/

Ercolini D, E Pontonio, F De Filippis, F Minervini, A La Storia, M Gobbetti, and R Di Cagno. Microbial ecology dynamics during rye and wheat sourdough preparation. Applied and Environmental Microbiology 79 (24) 7827-7836; doi: 10.1128/AEM.02955-13

Reese AT, AA Madden, M Joossens, G Lacaze, RR Dunn. 2020. Influences of ingredients and bakers on the bacteria and fungi in sourdough starters and bread. mSphere 5:e00950-19; doi: 10.1128/mSphere.00950-19

Ripari, V., Gänzle, M.G. and Berardi, E., 2016. Evolution of sourdough microbiota in spontaneous sourdoughs started with different plant materials. International Journal of Food Microbiology 232:35-42; doi 10.1016/j.ijfoodmicro.2016.05.025

Vogel RF, M Pavlovic, MA Ehrmann, A Wiezer, H Liesegang, S Offschanka, S Voget, A Angelov, G Böcker, and W Liebl. 2011. Genomic analysis reveals Lactobacillus sanfranciscensis as stable element in traditional sourdoughs. Microbial Cell Factories 10(Suppl 1):S6. doi: 10.1186/1475-2859-10-S1-S6

About Jeremy Yoder

Jeremy B. Yoder is an Assistant Professor of Biology at California State University Northridge, studying the evolution and coevolution of interacting species, especially mutualists. He is a collaborator with the Joshua Tree Genome Project and the Queer in STEM study of LGBTQ experiences in scientific careers. He has written for the website of Scientific American, the LA Review of Books, the Chronicle of Higher Education, The Awl, and Slate.
This entry was posted in community ecology, DNA barcoding, ecology, evolution and tagged , . Bookmark the permalink.