Coffee Fermentation Explained: What Anaerobic, Co-Ferment & Carbonic Maceration Actually Mean

11 min read · Updated July 20, 2026

A pour-over drinker's guide to what actually happens when coffee ferments: what the microbes do, what anaerobic, SIAF, carbonic maceration, and co-ferment mean, and what the research supports.

Anaerobic. Carbonic maceration. Self-induced anaerobic fermentation. 120-hour co-ferment. Modern coffee bags describe fermentation in language borrowed from wine, brewing, and lab microbiology, and it’s easy to read a single word like anaerobic as if it were a flavor. It isn’t. Fermentation is not one process, and no label on its own tells you how a coffee will taste.

This is the deep end. If you want the plain map of washed, natural, honey, and the rest, start with our guide to coffee processing methods. This piece is about what happens inside the tank: what the microbes do, what the popular terms mean, what the research supports, and how to shop and brew without being fooled by a buzzword.

Why coffee is fermented in the first place

A coffee bean is the seed of a fruit. Under the skin and pulp sits a sticky, sugar-rich layer called mucilage. Left alone, that layer makes seeds clump together, slows drying, and can grow mold. For most of coffee’s history, fermentation solved that physical problem: producers let the microbes already living on the fruit eat through the mucilage so it could be washed off, leaving a clean seed ready to dry. Fermentation consultant Lucia Solis, a former winemaker who has worked with coffee producers since 2014, describes this as fermentation’s original job: a practical way to get a clean, stable seed, not a trick for making a coffee taste like passion fruit.

It helps to separate three reasons a producer might ferment:

• For function: remove the mucilage and get to drying.
• For flavor: manage microbes and conditions to change the eventual cup.
• For consistency: make processing more predictable across tanks and harvest days.

Modern specialty coffee often chases all three at once, which is why the labels have gotten complicated.

What the microbes are actually doing

Coffee fermentation is a small ecosystem, not a single reaction. It runs on a shifting cast of yeasts, lactic-acid bacteria, acetic-acid bacteria, and other microbes. Studies routinely identify organisms from genera such as Saccharomyces, Pichia, Hanseniaspora, Candida, Lactiplantibacillus (formerly Lactobacillus), Leuconostoc, and Bacillus, along with Enterobacteriaceae like Klebsiella and Erwinia. Not all of them are desirable; some are just present. Yeasts and some bacteria tend to lead early, and lactic-acid bacteria rise later.

As they work, these microbes produce and modify acids (lactic, acetic, citric, malic), alcohols, esters, aldehydes, and other aromatic compounds, and they change the sugar and pH environment around the seed. The cup you eventually taste is not fermented juice trapped in a bean. Several things happen at once:

1. Some microbial compounds migrate into the seed.
2. Fermentation changes the chemistry around the seed.
3. The living seed reacts to acidity, temperature, and oxygen.
4. Drying decides which compounds remain.
5. Roasting turns those precursors into an entirely new set of aromas.

Reviews find strong evidence that fermentation changes coffee’s chemistry and flavor, but tracing one microbe to one specific tasting note is still hard to generalize.

A single harvest, four ways

The clearest demonstration of how much microbes matter comes from a 2026 video by James Hoffmann and Lucia Solis, filmed in Guatemala. They took one lot of coffee and processed it four ways:

• Mechanically demucilaged: a machine scrubs the mucilage off with friction and water, skipping most of the deliberate fermentation. It’s the closest thing to a low-fermentation control.
• Wild (spontaneous): left about 48 hours with only the microbes already on the fruit and equipment.
• Lactobacillus: the tank is inoculated with a selected lactic-acid bacterium.
• Yeast: the tank is inoculated with a selected Saccharomyces yeast.

This is a tasting demonstration, not a controlled experiment with published scores, and that’s the point: same farm, same variety, same day, four microbial paths, four different cups. Microbial selection is a real processing variable, the way choosing a yeast strain is in wine or beer. “Wild” isn’t truly random either, since temperature, ripeness, water, tank cleanliness, and the local microbe population all steer which organisms win.

Historically, producers knew fermentation was finished by feel. They rubbed the parchment coffee between their hands, and when it stopped feeling slippery and turned rough like gravel, the mucilage was gone and the coffee was ready to wash. That endpoint meant “clean enough to wash,” not “maximum flavor.” Producers fermenting for flavor now also track time, temperature, pH, and aroma, but even those numbers aren’t universal recipes: 48 hours at 16°C is a different process than 48 hours at 30°C.

”Anaerobic” is not a flavor

“Anaerobic” is the term most likely to mislead you. Solis points out that it’s nearly redundant: fermentation is, by definition, a metabolic process that runs without oxygen, so almost all fermentation is already “anaerobic” in the strict biochemical sense.

That’s true at the level of the chemistry, but don’t stretch it into “oxygen doesn’t matter.” Coffee fermentation is an open, mixed-microbe process, and how much oxygen is available decides which microbes dominate and which compounds form. That’s exactly why research treats oxygen availability as a distinct, controllable variable. An open tank, a sealed tank that fills with its own CO2, and a tank actively flushed with CO2 are three different environments.

So “anaerobic” is best read as loose shorthand for a sealed or oxygen-restricted process, and the word alone tells you almost nothing. An “anaerobic” bag could be:

• whole cherries or depulped coffee,
• spontaneous or inoculated,
• dry or submerged in water,
• self-generated or injected CO2,
• 24 hours or 200 hours,
• with or without added fruit, yeast, or other ingredients.

Those aren’t interchangeable, and none of them is a taste.

SIAF and carbonic maceration

Two sealed-tank terms show up constantly. Self-induced anaerobic fermentation (SIAF) puts coffee in a sealed vessel and lets the microbes’ own respiration use up the oxygen and build CO2, so the environment turns anaerobic on its own. Studies of SIAF find microbial communities and flavor profiles that shift with fermentation time, and our processing methods guide notes research where the technique raised sensory scores in some lots. A sealed tank is an environmental control, though, not a quality certificate.

Carbonic maceration is borrowed from winemaking, where whole, intact grapes ferment inside their own skins in a CO2-filled vessel before ordinary yeast fermentation takes over. Coffee producers adapted the term for whole cherries in a CO2-rich sealed tank, and it was popularized after Sasa Sestic used a version of it to win the 2015 World Barista Championship. In practice, coffee’s use of the term is loose: many coffees labeled “carbonic maceration” are closer to ordinary sealed whole-cherry fermentation than to strict wine-style carbonic maceration. Trade explainers are clear that there’s no enforced coffee standard.

Inoculated yeast and bacteria

Instead of relying on whatever is on the fruit, a producer can add a chosen microbe. This is the coffee version of a brewer pitching a specific yeast. The evidence here is some of the strongest in the field, and also the most conditional. A 2020 multi-region study found that yeast inoculation shifted flavor and raised scores by as much as five points, but the best organism depended on the process: Saccharomyces cerevisiae performed best for pulped-natural coffees, while Torulaspora delbrueckii did better for naturals. A separate 2021 study of Mundo Novo coffee found selected yeasts stayed dominant through a roughly 27-hour fermentation and produced the highest scores for sweetness, complexity, and aftertaste, again depending on whether the coffee was processed as natural or pulped natural.

The takeaway: a starter culture can improve or differentiate a coffee, but “yeast inoculated” doesn’t guarantee a better cup. There are many strains of S. cerevisiae, just as there are many coffee varieties, and strain, dose, temperature, and endpoint all matter. The species name alone still hides a lot.

Co-ferments, infusions, and the transparency question

Adding fruit to a tank is a co-fermentation: the fruit brings extra sugar, water, and its own microbes, which changes the fermentation. Adding something like cinnamon, which has no sugar and can suppress microbes, works more like an infusion, closer to direct flavoring. Our processing methods guide covers the co-ferment controversy and the competition bans in depth. Two distinctions matter most here.

First, adding a mango doesn’t reliably make a coffee taste exactly like mango; the fruit acts partly as fuel and microbial seeding, not a flavor injection. Second, Solis has warned that very specific, candy-like flavors, the watermelon-bubblegum kind, sometimes come from added flavor products rather than real fruit. Treat that as an industry caution, not a measured statistic, since the research doesn’t establish how common undisclosed flavoring is. The strongest position isn’t purity but transparency: added fruit isn’t illegitimate, inoculated yeast isn’t “fake terroir,” and the real problem is an undisclosed or deliberately vague addition.

What the research actually shows

It helps to weigh the evidence by strength rather than treat every claim as equal:

• Well supported: fermentation measurably changes coffee’s microbiology, chemistry, and flavor.
• Supported but context-dependent: selected starter cultures can raise scores and, maybe more usefully, improve consistency.
• Emerging: SIAF and carbonic maceration can create distinctive profiles.
• Weak: claims that a specific co-ferment ingredient reliably produces a specific flavor.
• Weakest: universal promises about thermal shock and other under-defined techniques.

A 2023 review in Food Research International argued that fermentation is better described by its underlying variables, fruit treatment, oxygen availability, added water, and starter culture, than lumped under washed, natural, or honey, and concluded that no single process reliably produces higher quality across varieties, climates, and farms. Longer isn’t automatically better, either. Fermentation-time experiments suggest much of the gain arrives around 48 hours, with diminishing returns and rising risk beyond that. “200-hour fermentation” is like saying food was “cooked for eight hours” without saying whether it was simmered, smoked, or left in the sun.

In practice, fermentation’s most useful payoff may be consistency rather than a record score: reproducing a good profile, avoiding defects, managing hot or cold weather, and differentiating lots from a single farm. The literature also has real limits, since most studies use one farm, a few varieties, one season, and a professional cupping panel, so a strong result under one protocol shouldn’t become “anaerobic coffee tastes better.”

How to shop and brew experimental coffees

Buy the disclosure, not the buzzword. A useful process description answers at least five questions: what was fermented (whole cherry, depulped, parchment)? what was the environment (open, sealed, submerged, CO2-flushed)? who did the fermenting (spontaneous microbes, or a selected yeast or bacteria)? what was added (nothing, fruit, spice, flavoring)? and what was controlled (time, temperature, drying)? “72-hour anaerobic” is less informative than it looks.

Don’t infer quality from complexity. A clean washed coffee can be more interesting than an aggressively processed one, and a carbonic maceration can taste muddled. Fermentation can’t manufacture the qualities of good green coffee out of poorly grown or badly dried material.

Taste blind when you can. Process language primes expectations. A 2026 study in Foods found that the label itself changed perception: a “fermentation” label raised expected acidity and lowered liking, while a “carbonic maceration” label raised curiosity without hurting acceptance. The effect was specific to the words used. A good home experiment: brew two coffees without looking at their processing, note aroma, sweetness, acidity, and clarity, then reveal the labels and see whether the word changed your read.

For brewing, there’s no validated “anaerobic recipe.” Start with your normal one so the coffee, not a new method, is the variable. A sensible pour-over baseline is 16 g coffee to 256 g water, 92–95°C, a moderate grind, and limited agitation, tasting as it cools. For an intense, boozy, or candy-like co-ferment, a slightly cooler temperature or coarser grind can bring it back into balance; if a cup turns hollow and aromatic but flavorless, go finer or hotter. These are brewing heuristics, not fermentation laws. Our guide to extraction covers the why.

Track it in BeanBench

The fastest way to learn which fermentation styles you actually like is to write it down. In BeanBench you can record each coffee’s process and fermentation details alongside your grind, temperature, and tasting notes. Over a dozen or so bags, patterns in your own history start to show: maybe you rate clean washed coffees higher than fruit co-ferments on average, or find co-ferments more aromatic but less clean.

Read those as associations in your own log, not laws. A handful of bags is a small sample, and your ratings are tangled up with roaster, origin, price, and roast, so the useful framing is “across your logged coffees you tended to prefer X,” with the sample size in view, rather than “anaerobic lowers your ratings.” If you want to explore on purpose, a gentle ladder runs from spontaneous washed, to yeast-inoculated washed, to sealed, to sealed whole-cherry, to carbonic maceration, to a clearly labeled fruit co-ferment, changing one thing at a time. From here, dial in with a recipe, revisit the processing methods primer, or read up on coffee varieties next.