From flower to flavour
A chocolate's flavour is the end of a chain that starts with a flower on a tree trunk. Here is the chain, link by link, with what has been measured at each and what has not.
How to read this page
Each step has two lists. Measured holds findings from individual studies, each linked to a summary that gives the study’s design and its limits. Not shown holds what those studies do not establish, which is often what a reader most wants to know.
The studies were read as published abstracts unless their summary says otherwise. Each measured one link, on its own trees or beans. The chain as a whole is an arrangement of separate results, and the last step says what follows from that.
1. The flower
Cacao flowers grow straight out of the trunk and older branches, from permanent sites called flower cushions. Each flower is small, and its pollen and stigma sit behind structures that an insect has to get past.
Measured
- Where a cushion forms depends on the age and position of the piece of branch it sits on. Heavy shade, cutting out 90% of the light, greatly reduced the number of cushions that formed. (Wibaux et al. 2025)
- Three-dimensional scans identified a side opening between the petal hood and the filament as the narrowest point an insect must pass, which limits pollination to very small visitors. (Wolcott et al. 2023)
- The flower's scent is made of straight-chain hydrocarbons, unusual among flowers. Laboratory-reared midges were only weakly attracted to it. (Arnold et al. 2019)
Not shown
How much shade is too much on a working farm. The shade finding is about cushions on young trees of two genotypes, not about pods.
What actually draws a pollinator to a cacao flower.
2. Pollination
Pollen has to be carried from one flower to another, usually on another tree. Tiny biting midges of the family Ceratopogonidae are the pollinators usually named.
Measured
- The midges are not the only candidates, and the role of the other insects seen at flowers is unknown. (Toledo-Hernández, Wanger and Tscharntke 2017)
- On farms in Central Sulawesi the commonest flower visitors were ants and flies, and not one ceratopogonid midge was caught. Forest and agroforest around a farm, and canopy cover over it, brought more visitors. (Toledo-Hernández et al. 2021)
- Putting rotting cacao husks or banana stems on the ground, as breeding sites, increased the number of young fruit in two studies. (Forbes and Northfield 2017; Vanhove et al. 2020)
Not shown
Which insects pollinate cacao in any given region, and in what proportion. A visitor is not a pollinator, and no study read here counted pollen delivered by species.
That the gains from rotting litter came through midges. Neither study counted them.
3. Compatible pollen
Many cacao trees cannot set fruit with their own pollen, or with pollen from a tree of the same type. Pollen arriving is not enough; it has to be the right pollen.
Measured
- Whether a tree can pollinate itself is controlled by two regions of the genome, for which diagnostic markers now exist. (Lanaud et al. 2017)
- In native Peruvian cacao, pollinating flowers by hand with a tree's own pollen succeeded 0.5% of the time. Crossing with another genotype was three to eight times more successful, depending on the pollen donor. (Vansynghel et al. 2023)
Not shown
How often incompatible pollen is what limits yield on ordinary mixed farms.
Genetics studies, one by one · The ten genetic groups of cacao
4. Fruit set, and what the tree keeps
A pollinated flower becomes a small pod called a cherelle. A tree sets many more cherelles than it ripens: most wilt and blacken on the trunk in their first weeks.
Measured
- In one clone, 43 pollen grains on a stigma gave half the maximum fruit set, and more than 115 gave no further gain. Seed number went on rising with pollen up to about 238 grains. (Falque et al. 1995)
- Hand-pollinated trees set more young pods than naturally pollinated ones and then shed more of them, so that mature pods did not differ. The authors concluded that the tree's supply of assimilate regulates fruit set and that wilting is how it adjusts. (Valle, De Almeida and Leite 1990)
- On three Ivorian farms with poor soils, 50–75% more young pods did not become significantly more mature pods. (Vanhove et al. 2020)
- In an Indonesian field experiment, by contrast, hand pollination raised yield per tree by 51% when a small share of flowers was pollinated and by 161% when all were, where fertiliser and insecticide gave no comparable gain. (Toledo-Hernández et al. 2020)
Not shown
Why the same intervention multiplied yield in one place and only multiplied wilted pods in another. The likeliest reading, that pollination limits yield only where the tree can feed the extra fruit, is the authors' inference and has not been tested in one experiment.
Whether a tree pushed to a higher yield by hand pollination can sustain it.
5. What kind of tree it is
The seeds inherit their chemistry from the tree and from the pollen parent. Cacao falls into ten genetic groups, and most farm trees are mixtures of several.
Measured
- The species divides into ten genetically distinct groups, not the three trade types, and Criollo is one small, uniform group among them. (Motamayor et al. 2008; Cornejo et al. 2018)
- Floral aroma in Nacional cacao has a genetic component. (Colonges et al. 2021)
- The pulp differs by variety before any fermentation: pulp of two varieties carried stronger floral and fruity odorants than pulp of CCN-51. (Chetschik et al. 2018)
Not shown
A flavour for each genetic group. None of the genetics studies tasted anything by group.
How much of a chocolate's character is the tree and how much the processing. One review found that 37% of fermentation studies did not even record the variety used.
Tasting terms this bears on: floral, fresh fruit.
The ten genetic groups of cacao · Genetics studies, one by one
6. Fermentation: what happens in the pulp
The beans are heaped or boxed in their sweet, acidic pulp. Microbes from the surroundings consume the pulp in a rough sequence, yeasts first, then lactic acid bacteria, then acetic acid bacteria, producing alcohol, then acid, and heat.
Measured
- The sequence of three microbial groups over about four days is the standard account, and the same few species recur in heaps and boxes across countries. (De Vuyst and Weckx 2016; Camu et al. 2007)
- Yeasts appear to be indispensable: with them suppressed, the beans did not develop chocolate character. With lactic acid bacteria suppressed, the chocolate was judged no different. (Ho, Zhao and Fleet 2014; Ho, Zhao and Fleet 2015)
- Of the choices a producer makes, fermentation time changed the microbes and the bean's volatile compounds most, turning next, and pod storage least. (Hamdouche et al. 2019)
- Box, ground and jute fermentations grew measurably different communities, but across the literature no specific relationship has been found between the microbes present and either the method or the place. (Ghisolfi et al. 2023; Taylor et al. 2022)
- Starter cultures added on farms took over the fermentation, and added yeasts changed the flavour of the chocolate even without dominating. (Lefeber et al. 2012; Crafack et al. 2013)
Not shown
That heap, box or tray is better. Method changes the microbes; no study read here ties that change to the chocolate.
That a starter culture beats a well-run spontaneous fermentation in a blind comparison.
That the laboratory result on lactic acid bacteria holds at farm scale.
Tasting terms this bears on: acidity, acetic acidity, lactic acidity.
Fermentation studies, one by one · The microbiology of cacao fermentation
7. Fermentation: what happens in the bean
The microbes never enter the bean. What reaches it is acetic acid and heat, which kill the seed and break down its internal compartments, so that the bean's own enzymes meet their substrates. Fermentation makes the raw materials of chocolate flavour, not the flavour.
Measured
- The precursors of cocoa aroma are produced from one storage protein of the seed by two of its own enzymes, which work best at different acidities. (Voigt et al. 1994; Voigt et al. 1994)
- Bitter and astringent compounds fall during fermentation, mostly early: in two field studies most of the change in polyphenols came in the first two days. (Albertini et al. 2015; Febrianto and Zhu 2020; Payne et al. 2010)
- Seven heaps with the same microbes gave chocolates that tasted different, and what remained of epicatechin and theobromine set how bitter each was. (Camu et al. 2008)
- Turning a heap let in air, increased acetic acid, and gave a sourer chocolate. (Camu et al. 2008)
Not shown
How long to ferment. Six days suited one set of beans; another genotype had done most of its changing in two days and gained nothing after four.
How much flavanol is lost. Published figures run from under a fifth to over four-fifths.
That fermentation by itself produces chocolate aroma. The precursors only become aroma on roasting.
Tasting terms this bears on: bitterness, astringency, raw, over-fermented, cacao.
8. Drying
Fermented beans are wet and still full of acid. Drying stops the fermentation, lets volatile acid escape and browns the bean. It is where a good fermentation is most easily spoiled.
Measured
- Sun-dried and air-blown beans did not differ in acidity and made chocolate judged favourably. Oven-dried beans in the same trial kept their volatile acids and made chocolate with a strong off-flavour. (Jinap, Thien and Yap 1994)
- Beans left wet in shade for five days did not dry fast enough to stop mould and putrefaction, and fast drying afterwards did not rescue them. (Jinap, Thien and Yap 1994)
- In another trial, drying at 70 or 80 °C after six days of fermentation gave a volatile profile similar to sun drying. (Rodriguez-Campos et al. 2012)
- The mould toxin ochratoxin A arises mainly during drying and storage. (Copetti et al. 2010)
Not shown
A safe temperature for artificial drying. The trials disagree, and they used different beans, dryers and measures.
Tasting terms this bears on: acetic acidity, cheesy, musty and mouldy, smoky.
9. Roasting, and what is tasted
Roasting turns the precursors into aroma. It is the first point at which anything smells of chocolate, and the last at which the maker can change much.
Not shown
A line from any one farm practice to any one tasting note. Each link above was measured in a different study, on different cacao; nobody has followed one lot of beans from the flower to a tasting panel.
Tasting terms this bears on: malty, caramel, honey, roast degree, burnt, metallic bitterness.
The tasting vocabulary, with its own sources · Roasting · Chocolate flavour chemistry
What the chain does not add up to
It is tempting to read the steps as a set of levers: more shade here, a day less fermentation there, a fruitier bar at the end. Nothing above supports that. The studies used different cacao in different countries, and most measured compounds in beans, not flavour in chocolate. Where two studies address the same question, on fermentation length or drying temperature or hand pollination, they often disagree, and the disagreement is itself the finding: the answer depends on the trees, the soil and the beans.
The pages that go with this one: pollination studies, genetics studies, fermentation studies, the ten genetic groups and the tasting vocabulary. All research areas.
Reviewed 8 October 2026.