Inside every active starter lives a self-regulating microbial ecosystem — wild yeasts, lactic acid bacteria, organic acids, and enzymatic activity that no commercial yeast can replicate. Here is exactly what is happening, and why it matters.
Commercial bread is a solved engineering problem. Controlled variables, predictable timelines, standardized yeast. Sourdough is the opposite: a living negotiation between flour, water, microbes, temperature, and time, each batch shaped by invisible decisions made by organisms that have been cultivating bread since the Neolithic. Understanding what is actually happening inside a mature starter is one of the most demanding problems in applied food science — and one of the most rewarding.
A mature sourdough starter is not a simple mixture of flour and water. It is a stable, competitive microbial ecosystem that has reached ecological homeostasis through a process of natural selection lasting days to months. The organisms that survive and dominate in this environment are those best adapted to its specific conditions: high organic acid concentrations, low oxygen, fluctuating pH, and cyclical nutrient availability.
The most common misunderstanding about sourdough is that it uses "wild" strains of Saccharomyces cerevisiae — the same species as commercial baker's yeast. In most wheat-based starters, the dominant yeast is actually Kazachstania humilis (formerly classified as Candida humilis and Candida milleri), which has fundamentally different metabolic properties. Crucially, it cannot metabolize maltose — the primary sugar released by amylase activity on damaged starch in flour. This is not a weakness; it is what enables the LAB-yeast symbiosis.
The flora varies by geography, flour type, and handling. Studies of San Francisco sourdough (the benchmark for American artisan bread) consistently find Kazachstania humilis codominating with Lactobacillus sanfranciscensis (now reclassified as Fructilactobacillus sanfranciscensis). Rye starters typically host different LAB species, especially Lactobacillus pontis and L. panis, producing markedly different flavor profiles.
The reason this particular LAB-yeast pairing is so stable comes down to a precise metabolic division. Amylases in flour degrade starch to maltose. F. sanfranciscensis consumes the maltose (which K. humilis cannot use) and produces lactic acid, acetic acid, CO₂, and ethanol. Meanwhile, K. humilis consumes fructose and glucose (the simple sugars released alongside maltose) and produces the CO₂ needed for leavening. Neither organism competes with the other for primary carbon sources. The lactic acid produced by the LAB also creates an acidic environment that suppresses competing organisms — including S. cerevisiae, which is actually less acid-tolerant than this native consortium.
"The sourdough microbiome is not accidental — it is the product of competitive exclusion, metabolic complementarity, and thousands of generations of co-evolution under the specific selective pressure of human baking practice."
After De Vuyst & Neysens, 2005The central flavor decision in sourdough baking is the ratio of lactic acid to acetic acid. These two organic acids are produced by the same bacteria but through different metabolic pathways, under different environmental conditions, in different proportions. Understanding how to shift this ratio is the difference between a mild, yogurt-tangy crumb and a sharp, vinegary punch.
Lactic acid (pKa 3.86) is the primary product of homofermentative and heterofermentative LAB metabolism under warm, wet, short-fermentation conditions. When stiff dough at cool-room temperature is fermented for 8–12 hours, lactic acid accumulates faster than acetic acid. The resulting flavor is smooth, milky, and gently sour — the profile of a San Francisco pain au levain or a well-made bâtard fermented at 24°C (75°F).
Chemically, lactic acid is produced via the Embden-Meyerhof-Parnas pathway in homofermentative LAB (producing only lactic acid from glucose) and as one product of the phosphoketolase pathway in obligate heterofermenters like F. sanfranciscensis. At warmer temperatures (above 28°C / 82°F), LAB favor NADH re-oxidation through lactic acid production — the more thermodynamically efficient route.
Acetic acid (pKa 4.75) is more volatile and more perceptually intense than lactic acid at equivalent concentrations. Acetic acid production is favored under cold temperatures (4–10°C / 39–50°F), low hydration (stiff starter, 65–75% hydration), and extended fermentation times. Cold retarding a shaped loaf for 12–18 hours in the refrigerator dramatically shifts the acid profile toward acetic acid, producing the aggressive sourness associated with dark rye sourdoughs and some San Francisco-style miches.
Beyond organic acids, heterofermentative LAB also produce ethanol (which contributes to ester formation during baking), CO₂ (which supplements yeast-produced leavening gas), and trace quantities of diacetyl, acetaldehyde, and longer-chain volatile organic compounds. These molecules, even at sub-threshold concentrations, contribute to what sensory scientists describe as the "roasted grain complexity" of properly fermented sourdough — absent entirely from commercial yeast bread.
The structural transformation of gluten during sourdough fermentation is one of the most consequential — and least discussed — aspects of what makes sourdough bread superior. The acid environment, proteolytic enzymes, and extended time all work together to produce a gluten network that is fundamentally different from that in commercial yeast bread.
As pH drops during fermentation, the acidic environment activates wheat's endogenous proteases — specifically aspartic proteases that function optimally at pH 3.5–4.5, precisely the range achieved in mature sourdough. These proteases selectively cleave specific peptide bonds in gliadins and glutenins, partially hydrolyzing the protein network. This is not damage — it is controlled modification. The result is a more extensible dough with better gas-holding capacity and improved oven spring.
LAB also produce their own protease and peptidase enzymes. Research by Pruß et al. (2014) documented that L. plantarum strains express cell-wall-associated proteinases capable of cleaving proline-rich sequences in gluten proteins — the same sequences associated with celiac disease and gluten sensitivity. This is the biochemical basis for the observation that properly fermented long-ferment sourdough is better tolerated by some (not all) gluten-sensitive individuals.
"Prolonged fermentation with selected LAB strains reduces immunoreactive gliadin peptides by up to 99% in some wheat preparations — a finding with profound implications for both clinical nutrition and bread science."
After Gänzle et al., 2014Despite (or because of) this partial proteolysis, sourdough gluten exhibits a more organized network at the microscopic level. Scanning electron microscopy of sourdough crumb reveals:
Wheat bran is high in phytic acid (phytate), an antinutrient that chelates minerals like iron, zinc, and calcium, preventing their absorption. During sourdough fermentation, the drop in pH activates wheat's endogenous phytase enzyme, which hydrolyzes phytic acid. Studies consistently find 80–90% phytate reduction in whole-wheat sourdough compared to 10–20% in commercial yeast bread. This makes sourdough — particularly whole grain sourdough — meaningfully more nutritious than an equivalent loaf leavened with commercial yeast.
The Maillard reaction is the non-enzymatic browning cascade that produces the hundreds of aromatic compounds responsible for the smell and flavor of baked crust. It occurs between reducing sugars and amino acids under high heat — but sourdough's biochemistry sets up this reaction in ways commercial bread cannot replicate.
Fermentation consumes significant quantities of simple sugars (glucose, fructose) through yeast metabolism. This paradoxically benefits Maillard browning because: (1) the proteolysis of gluten during fermentation releases free amino acids — the other essential Maillard reactant — in higher concentrations than in unfermented dough, and (2) the caramelization of the remaining sugars proceeds faster in the slightly acidic environment of the crust surface.
The result is a crust that browns at a lower internal temperature threshold, develops more quickly in a dutch oven environment (where steam is trapped during the first 20 minutes), and produces a wider spectrum of Maillard products including:
The standard dutch oven technique — covered for 20 minutes at 260°C (500°F), uncovered for 20–25 minutes — exists precisely to manipulate the Maillard window. Steam from the covered pot keeps the crust surface moist and gelatinized, allowing maximal oven spring before the exterior sets. Removing the lid then allows rapid water evaporation from the surface, concentrating the reducing sugars and amino acids needed for Maillard browning into an increasingly thin, hot, dry layer. The "ear" (the raised flap along a scoring line) forms where differential expansion forces gluten to rupture and bloom outward — a structural signature of proper gluten development and adequate fermentation.
Most sourdough failures trace back to one or more of three root causes: starter immaturity, timing errors, or temperature mismanagement. Understanding the underlying microbiology makes diagnosis straightforward.
The most common failure mode. Causes in descending likelihood: (1) Underproofed dough — fermentation was insufficient and the gluten network lacked the organic acid modification needed for proper extensibility; gas expansion is limited by an underdeveloped, unextensible structure. (2) Overproofed dough — proteolysis went too far, the gluten matrix degraded, and gas retention collapsed during baking (produces a dense, wet, almost gluey crumb). (3) Weak starter — insufficient CO₂ production, low leavening power; caused by underfeeding, cold storage without adequate warm activation, or starter too young to have established a stable microbial ecosystem.
Excess sourness is almost always an acetic acid problem. Cold extended fermentation, stiff levain, old/overripe starter — all push the acetic/lactic ratio higher. Solutions: shorten bulk fermentation time, use a younger (4–6h post-peak) starter rather than an old overripe one, increase dough temperature to 25–28°C to favor lactic acid production, and reduce or eliminate cold retard. Reducing hydration paradoxically can help — at lower water activity, LAB metabolism slows and acetic acid accumulates less.
If the loaf spreads sideways in the oven rather than rising, the cause is almost always one of: overproofing (gluten structure degraded, cannot hold gas), insufficient shaping tension (the surface tension built during shaping is what creates directional expansion), or a weak, immature starter. The diagnostic test: check whether your bulk-fermented dough has 50–75% volume increase and passes the poke test (indent springs back slowly but completely). If it passed poke but still spread flat, the shaping needs work. If poke did not spring back, the dough was overproofed before shaping.
| Reference | Year | Focus | Key Finding |
|---|---|---|---|
World J Microbiol Biotechnol |
1994 | LAB ecology in sourdough | Established symbiotic relationship between L. sanfranciscensis and C. humilis; documented maltose-negative yeast as dominant leavening agent in SF sourdough |
Trends Food Sci Technol |
2014 | Sourdough microbiome diversity | Survey of 50+ sourdough cultures across Europe identified over 50 LAB and 20 yeast species; demonstrated geographic and flour-type specificity of starter microbiomes |
Curr Opin Food Sci |
2014 | Fermentation biochemistry & nutrition | Documented phytate reduction (80–90%), protein modification patterns, and reduction of immunoreactive gliadin epitopes during LAB fermentation of wheat flour |
Food Microbiology |
2014 | Gluten proteolysis by LAB | Characterized proteinase activity of L. plantarum strains in sourdough; demonstrated cleavage of proline-rich gluten peptide sequences associated with celiac immunoreactivity |