Biology

Where the Vine Was Domesticated: What Ancient DNA Has Settled, and What It Has Not

Georgia, one origin, 8,000 years ago is the tidy story. What 3,500 genomes, 19,000 pips and 50 ancient seeds say about the wine grape's domestication.

The usual account says the grapevine was domesticated once, in the South Caucasus, about 8,000 years ago, and that everything we drink descends from that event. The account is incomplete rather than wrong. Three kinds of evidence bear on it, and each answers a different question: chemical residues in pottery date early winemaking, the shape of charred pips dates the arrival of cultivated vines in a region, and DNA from modern and ancient vines traces where the domesticated genome came from. Only the genomics is still in open dispute.

Residues in pottery

Grape wine leaves tartaric acid in porous pottery, and red wine also leaves the anthocyanin malvidin. In 2017 Patrick McGovern's team reported tartaric, malic, succinic and citric acids in jar sherds from Gadachrili Gora and Shulaveris Gora in Georgia, dated to about 6,000–5,800 BCE. Grape pollen and starch on a jar of the same type and date corroborated the result. In 2011 Hans Barnard's team had reported malvidin in a pressing installation and storage jars at the Areni-1 cave in Armenia, around 4000 BCE, still the earliest excavated winery. Each is a single-site result, well dated, and neither has been seriously challenged.

Together they establish winemaking in the South Caucasus by the early sixth millennium BCE. They do not show where the plant was domesticated. Residue chemistry cannot tell a wild grape from a cultivated one.

Pip shape

Wild grape pips are small and roundish, with short stalks. Domesticated pips are longer, with longer stalks. Geometric morphometrics turns that difference into a measured outline, and the Montpellier group around Laurent Bouby and Jean-Frédéric Terral has applied it to 19,377 archaeological pips from France, classified against a reference set of 80 wild and 466 cultivated accessions. Their 2024 synthesis in Scientific Reports finds wild-type grapes exploited for millennia, then a shift to domesticated forms around 600–500 BCE. The shift coincides with Greek and Etruscan contact on the Mediterranean coast.

The authors state the limit of the method themselves. Pip shape records a domestication syndrome, not ancestry. A wild-looking pip may come from a wild vine or from a first-generation cross.

The single-origin studies

Where the domesticated genome came from, and how many times, is a question for genomics. The modern baseline is a 2011 paper in PNAS by Sean Myles and colleagues. Genome-wide markers on more than a thousand accessions supported a Near Eastern origin for Vitis vinifera subsp. vinifera, a weak domestication bottleneck, and introgression from local wild sylvestris as the crop moved into Europe. The paper also established a point every later study has confirmed. Modern varieties sit in a dense network of first-degree relationships, because a few favoured plants, propagated by cuttings, were the parents of hundreds of cultivars.

Sara Freitas and colleagues strengthened the introgression side in 2021. From a hundred resequenced genomes they concluded that one central domestication was followed by hybridization with wild vines across Western Europe. The signal was strongest in a subset of Iberian varieties carrying large introgressed tracts, which the authors read as adaptive borrowing from locally adapted wild populations. Hua Xiao and colleagues reached a similar conclusion in 2023 from 345 samples: a single domestication, then continuous gene flow from European wild grapes into wine grapes over roughly the last two thousand years. That gene flow carried useful aroma genes, and also a higher load of deleterious variants.

The 2023 dual-domestication paper

In March 2023 a consortium of about seventy authors led by Yang Dong published the largest grapevine dataset so far in Science: 3,525 accessions after filtering, 2,503 cultivated and 1,022 wild. The reconstruction built on it has several parts, and they do not carry equal weight.

The best-supported part concerns the wild vines. Wild sylvestris splits into an eastern and a western ecotype, separated in the Pleistocene, and the eastern ecotype divides again into a Levantine group and a Caucasus–Iran group. The cultivated groups line up with them. A Western Asian cultivated group (CG1, mostly table grapes) shares its ancestry with Levantine wild vines; a Caucasian cultivated group (CG2, mostly wine grapes) shares its ancestry with Caucasus wild vines. Each cultivated group is closer to its own regional wild population than to the other cultivated group, and the authors read this as two independent domestications. Their models put both splits at a median of about 11,000 years ago, with a 95 percent interval of roughly 10,500 to 12,500 years. That is three thousand years earlier than the conventional date, at the start of farming in the Near East.

The consequence for wine comes next. The Caucasian domesticate, which the usual account treats as the ancestor of all wine grapes, mostly stayed where it was. Its cultivars are confined to both sides of the Caucasus, with limited dispersal into the Carpathian basin, and the authors call it a local domestication with a minor impact on grapevine diversification. The wine grapes of Western Europe, the group that holds the Bordeaux, Burgundy and Rhône families, descend instead from the Western Asian lineage. That lineage travelled west with early farmers, gave off a muscat group, a Balkan group and an Iberian group along the way, and then took up further ancestry from western European wild vines.

On this reading, the wine grape of France is a table-grape lineage that became a wine grape in Europe, and the Caucasus contributed early winemaking far more than it contributed genes. The paper also notes that winemaking, dated to about 8,000 years ago on the Georgian evidence, possibly came after domestication rather than causing it.

The western wild ancestry is common ground. Every genome study since 2011 finds substantial ancestry from European wild vines in the western European wine grape, whatever it concludes about origins.

The dispute over method

The 2023 result comes from one very large study and has not been replicated. Its disagreement with the single-origin papers is about method as much as data, as an exchange in PNAS later that year showed. In a letter, José Luis Blanco-Pastor argued that grapevine genomic analyses which keep highly admixed genotypes (varieties that are themselves recent hybrids) and define groups from a phylogenetic tree distort a history that is not tree-shaped. The same choices, the letter argued, shift the estimated dates of divergence and gene flow by large factors. Xiao and colleagues replied, defending their group definitions and their forty demographic models.

The two sides make different modelling choices about admixture, and admixture is the defining feature of this crop. One domestication or two, eleven or eight thousand years ago, remains a live disagreement between one very large study and several smaller ones.

Three kinds of new evidence would settle it. First, more wild vines from the eastern range, sequenced to the same depth as the cultivars, since the dual-origin case rests on each cultivated group matching its local wild relatives. Second, the two 2023 datasets run under a common model, so that the 11,000-year date and the single-event conclusion are compared on equal terms. Third, ancient DNA from the domestication zone itself, Neolithic seeds from Georgia, Iran and the Levant, which does not yet exist at genome scale.

Ancient DNA from France

Where ancient genomes do exist, they have been decisive, but about the end of the westward spread rather than its beginning. In 2019 Jazmín Ramos-Madrigal and colleagues sequenced 28 archaeological seeds from Iron Age, Roman and medieval France and found them closely related to the western European wine cultivars grown today. Identical genotypes turned up at different Roman sites. One seed from Orléans, dated to around 1100 CE, was a genetic match for Savagnin blanc: nine centuries of cuttings from a single plant.

The 2026 study led by R. Noraz in Nature Communications extends the record to 49 pips spanning four thousand years, and its results match the pip morphometrics almost point for point. Bronze Age seeds from about 2300–2000 BCE are purely wild. Domesticated ancestry appears by 625–500 BCE, alongside local wild lineages. Levantine ancestry is present from the Iron Age. Caucasian ancestry is detected only in Roman-period samples, which fits the genomic picture of a Caucasian lineage that arrived late and mattered little.

Clonal propagation shows up early. The same genotype occurs at Saint-Maximin and at Marseille, some 40 kilometres apart, in the mid Iron Age. A Roman genotype occurs in Alsace and at two sites near Montpellier, 560–610 kilometres away. One medieval pip from Valenciennes, dated 1400–1500 CE, is genetically identical to modern Pinot noir.

Forty-nine pips is a small sample, as in all ancient plant DNA work, and the authors say plainly that pre-Roman and post-Roman contexts are under-sampled. Where it overlaps the 19,000-pip morphometric record, the two independent methods agree. The shift to cultivated vines in southern France around 600–500 BCE, and routine vegetative propagation by the Roman period, therefore rest on two methods and tens of thousands of pips.

The vine that reached Bordeaux

For Bordeaux, the consequence is one the history of its grape varieties already implies. The plant the Romans set at Burdigala was not a Caucasian import. It was the end product of several thousand years of westward movement and repeated crossing with the wild vines of Europe. When nineteenth-century growers had to graft it onto American roots, they added one more foreign contribution to a genome that had never been pure.