The fruit flavor in a natural coffee is mostly made, not moved. Yeasts and acetic acid bacteria spend two or three weeks working through the sugars in the drying cherry, the seed inside keeps metabolizing the whole time under the stress of losing its water, and the roast builds the aroma out of what the two of them leave behind. When De Bruyn and colleagues went looking for what actually crosses from the fruit into the seed during dry processing, what they found was gluconic acid, glycerol and mannitol, in what their paper calls a minor accumulation and a slow but observable migration. Those are microbial housekeeping compounds. None of them tastes like mango.
What is actually happening on a drying bed in Guji?
Ethiopia Buku Abel is the coffee I know best on this question, because we buy it and we publish what we know about it. Buku Abel is a kebele, the smallest administrative unit in Ethiopia, in the Hambela Wamena district of the Guji Zone. The village sits near 2,350 meters and the coffee around it near 2,275 meters, which puts it among the highest planted coffee anywhere. Several hundred smallholders farm a few hectares each, mostly the 74110, 74112 and 74158 selections that Ethiopia's Jimma Agricultural Research Center developed in the 1970s. Harvest runs roughly November through January. Ripe cherry goes to a processing site operated by Esayas Bareso, gets hand-sorted, and goes out whole onto raised beds for two to three weeks, turned often, and hand-sorted again at the end.
Nothing in that sequence is a flavoring step. Nobody adds anything. The cherry loses water, slowly, in the sun, with a seed inside it that is still alive and a population of microorganisms on the outside of it that is very much alive. Everything people mean by the fruit character of a natural happens inside that window, and the useful question is what the organisms and the seed are doing with the sugar while the water leaves.
Does the fruit's flavor soak into the seed?
Almost not at all, and this is the part worth slowing down on, because the explanation most of us grew up on says the opposite. Sprudge puts the common version plainly: natural coffees dry skin-on, "meaning the beans receive flavor from the fruit pulp and sugars as part of their development." Sweet Maria's, who are our own first import partner and still the clearest technical writers on dry processing anywhere, describe it as a transfer in which compounds in the fruit skin bond with the seed and fruited character is infused over the days it takes to dry.
The direction of that is right and the route is not. Barista Hustle asked the obvious question first, and it is a good one: if sugars from the pulp can cross the parchment into the seeds of drying fruit, why has that not already happened on the tree, where the seed sits in wet fruit for months? Their answer, from the measurement work, is that it is not sugar increasing in naturals at all. Knopp, Bytof and Selmar found the hexoses higher in dry-processed green coffee because they fall in washed coffee, not because they climb in naturals, with sucrose barely touched either way. The seed spends its own sugar when it is submerged and wet-fermented. Leave it in the fruit and it spends less. The gap is real and the arrow points the other way.
The SCA's own scientists put the same finding in the plainest available sentence in 25 Magazine: some of the metabolites produced during fermentation "remain on the surface of the beans, and can even survive the entire process," while the green bean carries "all the flavor precursors locked inside them." Surface, not cargo.
Which microbes are working on a drying cherry?
Not the ones most people name, which is why this is worth getting right. A washed coffee ferments underwater and lactic acid bacteria dominate it. A natural does not, and cannot: as the cherry dries, water activity falls to somewhere around 0.7 to 0.8, and Elhalis, Cox and Zhao record lactic acid bacteria as low and sometimes entirely absent in dry processing by the end for exactly that reason. What dominates a drying cherry instead, in De Bruyn's sampling, is acetic acid bacteria — Acetobacter and Gluconobacter — alongside Pichia and other yeasts.
The succession has a clock on it. Enterobacteria climb hard in the first 24 hours. Yeasts take over and hold the population through days 14 to 18, which is most of the drying period at Buku Abel. They eat the sugars in the mucilage and return alcohols, acids, esters, aldehydes and ketones, and the specific compounds that track with fruity cupping notes in that review are ethyl acetate, acetaldehyde, glycerol and ethanol. Then the water runs out. Microbial activity stops somewhere near a water activity of 0.6, and the lot finishes around 11 to 12 percent moisture with the biology switched off.
Is the seed doing anything while it dries?
It is doing a great deal, and it is not germinating. This distinction matters because the literature splits on it and a lot of coffee writing has merged the two. De Bruyn's group found germination triggered by the anoxia of underwater submersion, which is a washed-process condition, while the drying step induced a drought stress response and aerobic respiration in the seed. Bastian and colleagues state it flatly: in the dry method, germination is inhibited.
What the seed does instead is behave like any plant tissue being dried out. Kramer and colleagues tracked dehydrin expression climbing once residual water content fell below 43 percent, with GABA accumulating in more than one wave as the seed passed through different kinds of stress. Bastian's group puts drought stress starting near 45 percent moisture and metabolism shutting down near 25 percent. In between, the seed is running chemistry, and because a natural dries slowly inside its own fruit rather than quickly on its own, it gets longer to run it. Fu and colleagues measured the consequence directly: dry processing allows further metabolism, and it leaves higher sugar levels behind.
That is the second source of everything the roaster later gets to work with, and it belongs to the seed rather than to the fruit.
If the flavor is not absorbed, when is it made?
In the roaster, mostly, which is the same answer we published when we took apart why coffee smells like coffee at all. Vezzulli, Lambri and Bertuzzi ran green and roasted specialty arabica through the same method and identified 51 volatile compounds in the green and 70 in the roasted, the roasted set dominated by furans that simply do not exist in the green bean. Roasting, in their words, generates more than 100 compounds.
The most useful number in that paper is a smaller one. When they asked what separates a green natural from a green washed coffee before either is roasted, the discriminating compounds were 2,3-butanediol, pinene and octanol, with linalool higher in naturals and hexanal higher in washed lots. 2,3-butanediol is a microbial fermentation product. Linalool is a terpene the plant makes. Neither is a cargo of fruit esters. Whatever a natural is carrying into the roaster, it is not the cherry's aroma.
There is a harder piece of evidence pointing the same way, and it comes with a caveat I would rather state than bury. Zhang and colleagues tracked volatiles through drying and watched total esters inside the bean collapse from 42.84 to 0.54 micrograms per gram of dry weight over 48 hours at 40 degrees. Drying destroys esters faster than anything is adding them. The caveat is that those beans were mechanically peeled and degummed first, so that experiment is a washed-style drying and not a true natural. It shows the direction; it does not close the question.
Why do three naturals on one shelf taste nothing alike?
Because if the mechanism were absorption, three coffees that dried in fruit ought to converge on fruit, and ours do not. Three of the five coffees we can sell today carry the same natural processing record on their product pages. Buku Abel cups 93 and publishes as mango, papaya, berries, chocolate, chai, cinnamon and star anise. Organic Peru Decaf cups 88.4 and publishes as dried strawberry, dark chocolate, berry jam, sarsaparilla and banana milk. El Salvador Finca San Luis cups 86.8 and publishes as cherry wine, earthy cacao, dried date, baking chocolate and Brazil nut. Buku Abel and Finca San Luis carry the identical drying record as well as the identical processing record, and one of them is a mango coffee while the other is a nut and cacao coffee.
Our own explainer on what natural process coffee is makes the same point from the shelf and stops there, at the conclusion that variety and growing region predict the cup and the processing word does not. The mechanism above is why that conclusion holds rather than merely being an observation about four bags. Microbes and a stressed seed are transforming whatever sugar, acid and precursor that particular cherry brought them, in whatever quantity that plant in that place put there. The drying bed sets how long they get. It does not set what they are working with.
Publishing a cup score on every coffee is what makes that argument checkable, and it costs us something to keep doing it. An 86.8 sits on the shelf with its number printed next to a 93, on the same scorecard, in the same roastery, and there is no version of that where the lower number reads well. We publish it anyway, because a six-point spread inside one processing word is the most honest thing we can show a customer about what that word is worth.
What would actually settle this?
A study nobody appears to have run: a compartment-resolved measurement of a true natural, tracking specific compounds across the parchment into the endosperm over a real drying curve, rather than inferring it from the outside. Until that exists, reasonable people are still apart on how much gets in.
Barista Hustle, having demolished the sugar half themselves, hold that microbially made esters are absorbed into the seed even though the fruit's own aromas are not. Anja Rahn goes considerably further and argues that diffusion of fermentation precursors is the main influence on cup quality, putting the parchment's resistance at somewhere between 30 and 60 percent rather than at a wall. Elhalis and colleagues, reviewing the whole territory, hedge every version of the migration claim they make: metabolites are "believed to" migrate. That hedge is doing honest work and I am not going to unpick it for them.
So the claim I will actually stand behind is narrower than the headline. The cherry's own finished flavors are not what you taste. Small microbial molecules do get in, in minor and measured amounts. The seed's own stressed metabolism matters at least as much as anything crossing the parchment. And the roast makes most of the aroma out of all of it afterward. We buy green coffee, which means every decision described here was taken at origin months before the lot reached Pensacola, and I was not standing on the bed at Buku Abel watching it happen. What we can tell you is which decisions were recorded, what the science says they do, and what the coffee did on our own cupping table.
Questions we get about natural processing
Does natural process coffee absorb flavor from the fruit?
Very little. The measurement work finds minor amounts of microbial compounds like gluconic acid, glycerol and mannitol reaching the seed, described as a slow but observable migration, and essentially none of the fruit's own aroma compounds. The fruit character you taste is built by microbes working on the cherry's sugars, by the seed's own metabolism while it dries, and then by the roast.
Is fermentation the same thing as rotting?
No. Fermentation on a drying cherry is a controlled succession, mostly yeasts and acetic acid bacteria, that ends on its own when water activity falls to around 0.6. Spoilage is what happens when drying stalls and the wrong organisms get an extra week of moisture, and it shows up in the cup as sour, vinegary or musty. Turning a bed frequently is how a producer keeps the first thing and not the second.
Why does a natural taste sweeter than a washed coffee?
Largely because washed coffee is less sweet rather than because natural coffee gains sugar. Knopp, Bytof and Selmar found that the hexose sugars fall during wet processing while they hold or rise in dry processing, with sucrose barely affected either way. The seed spends its own sugar when it is submerged. Left inside the fruit, it spends less.
How long does a natural take to dry?
At Buku Abel, two to three weeks on raised beds, turned frequently and hand-sorted at the end. Drying ends when the lot reaches roughly 11 to 12 percent moisture, at which point microbial activity has already stopped and the seed's own metabolism has shut down.
Does the roast level change how fruity a natural tastes?
Yes, and more than most people expect, because the aroma is largely made in the roaster rather than carried in. Green and roasted coffee analyzed by the same method give 51 and 70 volatile compounds respectively, and the roasted set is dominated by furans that are absent in the green. The drying bed supplies precursors; the roast decides what they become.
The coffee this was written about
Ethiopia Buku Abel is the natural described throughout: heirloom selections from the Guji Zone, dried whole on raised beds, cupping 93, and published as mango, papaya, berries, chocolate, chai, cinnamon and star anise. It is available in all three sizes, and we roast to order every Wednesday in Pensacola.
