History of Coffee.

The conventional history of coffee begins in the mid-fifteenth century, when the Mufti of Aden adopted the beverage to support the Sufi dhikr. From there, the practice spread along the Ottoman trade network — Mecca, Cairo, Damascus, Aleppo, Istanbul — before crossing into Europe via Venetian merchants and Ottoman embassies. For more than two hundred years, Yemen was the world's only commercial source. The timeline below traces those first three centuries: who adopted coffee where, and when.

Yemen is central to the history of coffee as we know it.

For more than two centuries, Yemen was the world's only commercial source of coffee, and the Ottoman authorities went to extraordinary lengths to keep it that way — beans were boiled or partially roasted before export to prevent germination, and seedling transport was forbidden. The monopoly broke in five named events between roughly 1670 and 1727. Each began as a single act of human-scale plant transport — sometimes commercial, sometimes clandestine, occasionally outright theatrical. Every cultivar grown commercially today descends from one of them.

Bourbon and Typica between them are the genetic basis of nearly all commercial arabica grown today. The world's commercial coffee diversity, in other words, descends from perhaps a dozen seeds taken out of Yemen across a few decades — a stark genetic bottleneck that will matter when we look at the genome.

Although every English-language history of coffee focuses on Yemen, the thread does not start there. The earliest extant primary source on coffee — the Umdat al-safwa fi hill al-qahwa by Abd al-Qadir al-Jaziri al-Ansari, written in 996 AH (1588 CE; translated in full into English here for the first time) — is itself a Yemeni Hanbali defense of the beverage against Islamic critics. Even from that pro-Yemen vantage, al-Jaziri distinguishes two epistemic states explicitly: the Yemeni adoption is named, dated, and sourced; the Ethiopian use is older, more diffuse, and already beyond the documentary horizon he can reach.

Three centuries before the genome made it possible to ask the question with molecular tools, an Arabic scholar in Mecca had already given the answer in writing: Yemen acquired coffee in the fifteenth century from a source it could trace to a named individual; Ethiopia had used it for so long that its origins were already, in 1588, beyond recall.

The strongest line of ethnographic evidence for the antiquity of Ethiopian coffee is not what any single informant says about its origins — it is the sheer breadth of preparations the modern Oromo and adjacent peoples maintain side by side. A culture that received coffee as a foreign import six hundred years ago does not develop fifteen distinct named preparations spanning food, beverage, medicine, ritual, and travel rations, with the leaves, husks, beans, and whole fruit each having their own niche. That diversity is a fingerprint of long, continuous use.

What the documentary and ethnographic records suggest, the coffee genome corroborates with molecular evidence al-Jaziri could not have imagined. Whole-genome resequencing places the wild range of Coffea arabica unambiguously in the Ethiopian highlands, and traces every modern commercial lineage — Bourbon, Typica, Mundo Novo — to a Yemeni daughter population that split from that Ethiopian source ~27,000 years ago and continued exchanging genes with it until ~8.5 ka. Ethiopian origin, Bab al-Mandab corridor.

If coffee crossed the corridor, it crossed with people. The maternal DNA of modern Ethiopians and Yemenis records overlapping migration waves whose dates fall almost entirely within the coffee gene-flow window. Four haplogroups elevated in the Oromo and adjacent Cushitic and Omotic Ethiopians — M1a, L3x1, L3h2, N1a1a3 — each have a Yemeni-specific subclade founded inside the coffee window. Human movement and coffee gene flow overlap closely.

The corridor was not always open. The Bab al-Mandab was only a few kilometers across at the last glacial maximum; through the late Pleistocene a person, an animal, or a wind-borne propagule could cross routinely. After 12 ka, sea levels rose more than a hundred meters; by 8 ka the strait widened past the threshold below which casual crossing was plausible.

Coffee in the wild is animal-dispersed — birds and small mammals, not wind. A 7 km strait with intervening islands is well within the daily range of frugivorous birds; a 25 km open-water gap to the remote Yemeni highlands is not. The corridor closes for every dispersal vector at the same threshold, which is why coffee gene flow and human migration cease together at ~8.5 ka.

Once coffee crossed the strait, what happened to it? The same kind of selection that left visible signatures in cereal genomes — concentrated, multi-locus, directional, targeting palatability and architecture — is visible in the cultivated coffee genome. Three regions stand far above the genomic background: caffeine biosynthesis, the RNA-directed DNA methylation machinery, and an older pathogen-resistance cluster. The first two are the central signal. They look like classical domestication anchors, dated by Bayesian coalescent analysis to a window centered on 12 – 14 ka — about 11,500 years before the conventional Sufi domestication horizon.

The age distribution makes the point more sharply still. Each Yemeni-fixed allele carries its own coalescent age. Pooled across high-signal chromosomes, the distribution is bimodal: a recent peak at 5 – 15 ka comprising roughly 28% of fixed alleles, and an older peak at 50 – 200 ka. A 600-year founder event cannot fix alleles that diverged from the wild population 12,000 years ago.

The coffee sweep window does not sit alone. Plotted next to the canonical first-wave domestications of cereals, it occupies the same threshold at the Pleistocene–Holocene transition — and is fully nested inside the African Humid Period, the Holocene's largest reorganization of monsoonal climate.

The gene families involved are not random. Five anchors recur across crops as humans select for them — organ size, plant architecture, shattering, palatability, epigenetic regulation. Coffee's recent-age tail recovers a member of each. It is not a coincidence that coffee selection looks like cereal selection. They were happening at the same time, by the same mechanism, in human-associated landscapes on three continents.

The 12–14 ka selection signal is consistent with three scenarios that are not mutually exclusive. None require the formal cultivation we recognize from fifteenth-century Yemen.

The data cannot distinguish (1) from (2) from (3). What the data can do is establish that something on a human-domestication timescale selected coffee's caffeine and methylation machinery, alongside organ-size and architecture regulators that recur across cereals. The Sufi communities of Yemen received a plant that had already been shaped — and six numbers from the data are difficult to absorb into the conventional 600-year story:

The cultivated coffees of the world descend from a small set of seedlings that left Yemen in the seventeenth and eighteenth centuries. They became Bourbon, Typica, and through them every commercial arabica grown today. Yemen kept much more than it exported.

Montagnon and colleagues genotyped Yemeni coffee in 2021 and found five distinct genetic clusters — only one of which corresponds to the lineage the world received. A second cluster, the New-Yemen group, has never been exported and was never sampled until the work that defined it. Its 24 accessions, drawn from landraces like Bura'i, Udaini, and Dawairi, grow in some of the most extreme conditions any arabica tolerates: 200–700 mm annual rainfall, temperatures reaching 42 °C. The cup record points the same way: recent Yemeni microlots have scored in the highest tier Coffee Review has ever awarded.

Thirty thousand years of selection produced something we still haven't fully tasted.