Where cannabis comes from: botany, names and domestication

Few plants have travelled as far with people as cannabis, and few have collected as many myths along the way. It has been twisted into rope and woven into cloth, pressed for oil and eaten as seed, burned in graves and shrines, listed in the learned medicine of several civilisations, and, in the twentieth century, made the target of one of the largest international control systems ever built around a plant. This article is the first in a series of six that follows that long relationship from the plant's evolutionary beginnings to the legal reforms of the 2020s. It starts at the root: what kind of plant cannabis is, why botanists have argued about its names for more than two centuries, where the genus probably came from, and what genetic evidence says about when and where people first domesticated it.

A companion species

A useful way into the whole story is to stop thinking of cannabis as a "drug plant" and to think of it instead as a multi-purpose "companion species". That phrase sums up the argument of the review on which this series is based. Different societies have valued different parts of the plant in different combinations. The stem gives long, strong fibres. The seeds, botanically small fruits called achenes, are rich in oil and protein. The flowering tops, especially of female plants, carry a sticky resin that contains the compounds now called cannabinoids, among them tetrahydrocannabinol (THC) and cannabidiol (CBD).

Seen this way, cannabis looks less like an exotic outlier and more like other crops that people have bent to several purposes at once. Flax gives both linen and linseed; many cereals give both grain and straw. The difference with cannabis is that one of its products affects the mind, and that fact has coloured how its whole history is told. Careful scholarship tends to show something more ordinary and more interesting: a plant whose fibre and seed mattered to most people for most of history, and whose resin mattered intensely to some people at some times.

The series is organised as follows:

• Origins (this article): botany, naming, centre of origin and domestication.
• Prehistory: the earliest finds of seeds and fibre, from Jōmon Japan to the steppe.
• Antiquity: ritual burning, the Scythians and the first chemical proof of psychoactive use.
• Medieval trade: hashish in Islamic societies, Africa, the Atlantic crossing and hemp as a strategic crop.
• Medicine and prohibition: cannabis in nineteenth-century Western medicine and the building of international control.
• Science and reform: the discovery of cannabinoids and the body's own cannabinoid system, and the reforms of 1996–2026.

Before any of that history can be told, though, two groundwork questions need answering. What exactly are we talking about? And what kinds of evidence can tell us about it?

Getting the words right

Cannabis, hemp and drug-type plants

Discussions of cannabis are notoriously muddled by vocabulary, so it is worth fixing terms at the outset. In this series, "cannabis" means the genus Cannabis and its products in general. "Hemp" means low-THC forms grown for fibre and seed. "Drug-type" cannabis means forms that people have selected for cannabinoid-rich resin.

These are working categories, not watertight boxes, and individual populations can sit awkwardly between labels. But the distinction captures something real: over many generations, selection for fibre and selection for resin can pull populations apart, and, as we shall see, the genomic evidence suggests they did.

One small but important point of language: the familiar labels "Sativa" and "Indica", used in modern commerce to describe strains, are not the same thing as the botanical names Cannabis sativa and Cannabis indica. The overlap of spelling hides a large gap in meaning, and that gap is one of the main threads of this article.

Three kinds of evidence

The second piece of groundwork concerns evidence. The review behind this series insists on a three-way distinction that runs through everything that follows (Long et al., 2017; McPartland, Hegman and Long, 2019):

Each step brings us closer to human intention, but each is rarer. Pollen is abundant and survives well for tens of thousands of years. Seeds and fibres survive only under favourable conditions, such as waterlogged, frozen or very dry deposits. Chemical residues are found less often still, and detecting them requires modern laboratory techniques.

According to the review, confusing these categories is one of the most common sources of error in popular histories. A pollen grain in a lake core may be presented as evidence of "early cannabis use", when it shows only that the plant was growing somewhere in the catchment. A seed in a settlement may be presented as evidence of "cultivation", when it may come from a wild plant someone picked. A length of ancient rope may be presented as evidence of "smoking", when it shows only that someone valued strong fibres. Keeping the categories apart is not pedantry. It is the difference between history and wishful thinking.

A word on dates

Because the evidence discussed below stretches over tens of thousands of years, it helps to understand how scientists write dates. Archaeologists and palaeoecologists commonly use "BP", meaning "before present", where "present" is conventionally fixed at 1950. Many dates come from radiocarbon dating, which measures how much of the radioactive isotope carbon-14 remains in organic material. Living things take up carbon-14 while they are alive; after death it decays at a known rate, with a half-life of roughly 5,700 years. Measuring what remains gives an age.

Raw radiocarbon ages are not quite calendar years, because the amount of carbon-14 in the atmosphere has varied over time. Scientists therefore calibrate them against independent records, such as tree rings, to produce calendar dates, written "cal BP" or "cal BCE". Even after calibration, each date carries a range of uncertainty. When you read that something dates to "about 10,000 cal BP", the word "about" is doing honest work.

The family tree: Cannabaceae and the hops connection

Where cannabis sits

Botanists group plants into families of related genera. Cannabis belongs to the family Cannabaceae. On current phylogenetic understanding, that family also includes Humulus, the hops used to flavour and preserve beer, and several genera that were formerly placed in a separate family, the Celtidaceae, among them the hackberries (Celtis). Fossil evidence suggests that the Cannabis lineage diverged from its relatives many millions of years ago (McPartland, 2018).

The phrase "on current phylogenetic understanding" deserves a word of explanation. A phylogeny is a family tree of species, reconstructed from shared characteristics. In the past those characteristics were mostly visible features: the shape of flowers, fruits and leaves. Since the late twentieth century, botanists have relied increasingly on DNA sequences, comparing the same stretches of genetic code across many plants and asking which groups share the most recent common ancestor. The results sometimes overturn older groupings based on appearance, which is how genera once housed in the Celtidaceae came to be placed alongside cannabis and hops. Family boundaries are therefore scientific hypotheses, revised as evidence accumulates, rather than fixed facts.

Pollen twins

That close kinship causes a real practical problem. The pollen grains of Cannabis and Humulus are hard to tell apart under the microscope. As a result, many fossil pollen studies report such grains only under collective labels such as "Cannabis/Humulus" or simply "Cannabaceae" (McPartland, Hegman and Long, 2019).

For anyone trying to trace the deep history of cannabis, this is a serious obstacle. Imagine trying to map the historical range of a bird species using records that lump it together with a close cousin under a single name. If the two species have different habitats, though, you can make an informed guess: a record from a reed bed is more likely to be the wetland species; a record from a mountain forest is more likely to be the woodland one.

Probabilistic criteria

That is essentially the approach McPartland and colleagues took. They developed probabilistic ecological criteria for assigning ambiguous pollen to one genus or the other (McPartland, Hegman and Long, 2019). The details are technical, but the principle is the one in the bird analogy: because the two genera tend to favour different environments, the ecological setting in which an ambiguous grain was found offers a clue to its identity.

The key word is probabilistic. The method does not claim certainty for any single grain. It weighs the evidence and assigns pollen to the more likely genus. Applied across many records, such judgements can reveal broad patterns even if individual assignments might be wrong. This is a common strategy in science when direct identification is impossible: accept some uncertainty at the level of the single case in exchange for a usable picture at the level of the whole. It also means the resulting maps are only as good as the assumptions about where each genus prefers to grow, a point worth remembering when we come to the question of where cannabis began.

The species problem

If the family is settled, at least for now, the question of how many species the genus contains is not. Cannabis taxonomy has been disputed since the eighteenth century, and the dispute has never been purely academic.

Linnaeus and Lamarck

The modern system of naming plants with two-part Latin names, a genus followed by a species, was established by the Swedish naturalist Carl Linnaeus. In 1753 he named Cannabis sativa. The second word, sativa, means "cultivated" or "sown", a sign that Linnaeus knew the plant as a crop familiar from European fields. The "L." often written after the name in formal botanical writing simply records Linnaeus as its author.

In 1785 the French naturalist Jean-Baptiste Lamarck described a second species, Cannabis indica. As the name suggests, he associated it with India, and he regarded it as distinct from the European crop. With that, the central question of cannabis taxonomy was posed: is there one species of cannabis, or more than one?

This might sound like a simple matter of looking carefully. In practice it is very hard, for reasons that apply to many crops. Cannabis has been carried by people across much of the Old World, selected for different products in different places, allowed to escape into the wild and to cross back again with cultivated plants. The result is a cloud of populations that vary enormously in height, leaf shape, branching, chemistry and flowering time, but which interbreed readily. Deciding where one species ends and another begins in such a cloud depends partly on the evidence and partly on what you think a species is.

Small and Cronquist: one species, divided by chemistry

The most influential modern attempt to settle the matter came in 1976, when the Canadian botanist Ernest Small and the American botanist Arthur Cronquist published what they called a "practical and natural" taxonomy (Small and Cronquist, 1976). The title signals their ambition. A natural classification reflects the real structure of variation in the plant. A practical one is usable by people who need to identify and regulate plants.

Their solution was to treat the genus as monotypic, meaning it contains a single species, C. sativa. Within that species they recognised subspecies according to whether the plant had been selected for low or high intoxicant content. In other words, chemotype, the plant's chemical profile and especially whether THC was dominant, became a criterion for classification. Small and Cronquist described variation within the genus as "continuous for all characters" (Small and Cronquist, 1976). That phrase is the heart of the case for a single species: if every feature you measure grades smoothly from one form to another, with no clean gaps, it is hard to justify drawing a species boundary anywhere.

The decision to use chemistry as a taxonomic criterion was significant. Traditional plant classification relies heavily on visible structures. Using THC dominance aligned botanical categories with a feature that mattered greatly to society and to law, which made the system practical in exactly the sense its authors intended. But it also meant that a classification of the plant was partly a classification of how humans had used it. That invites the question of whether such categories describe the plant's evolution or its social history.

McPartland: "almost meaningless"

Later reviews have revisited the problem. The cannabis researcher John McPartland has been one of the most persistent contributors. In a 2018 review of cannabis systematics at the levels of family, genus and species, he argues that widespread interbreeding of commercial "Sativa" and "Indica" strains has made those vernacular labels "almost meaningless" botanically (McPartland, 2018).

It is important to be clear about what this claim does and does not say. It does not say that cannabis plants are all alike; they are obviously not. It says that the labels used in the commercial market, which loosely sort plants into "Sativa" and "Indica" on the basis of appearance or reputed effects, no longer track any coherent botanical lineage. Decades of crossing have shuffled the genetic deck so thoroughly that a plant sold under one label may share much of its ancestry with a plant sold under the other.

Scientific names versus everyday names

In an earlier paper with Geoffrey Guy, McPartland documented the conflicts between scientific and vernacular naming in cannabis, and the cultural bias in both (McPartland and Guy, 2017). Everyday vocabulary and botanical nomenclature can use overlapping words in incompatible ways. Botanical names, meanwhile, were not coined in a vacuum: they reflect the knowledge, assumptions and geographical vantage points of the European naturalists who created them. A plant named for India by a French naturalist in the 1780s carries the imprint of how Europeans of that era understood the world's geography and the origins of its useful plants.

Clarke and Merlin's four groups

A different way out of the impasse was proposed by Robert Clarke and Mark Merlin in their 2013 book Cannabis: Evolution and Ethnobotany (Clarke and Merlin, 2013). Rather than trying to fit the plant into the inherited species names, they proposed an alternative scheme of four groups:

• narrow-leaf hemp
• broad-leaf hemp
• narrow-leaf drug
• broad-leaf drug

The appeal of this scheme lies in its two independent axes. One axis concerns use: whether a population was selected for fibre and seed (hemp) or for resin (drug). The other concerns form, captured by leaf shape. Crossing the two gives four combinations. The scheme acknowledges that the hemp/drug distinction reflects human selection, while recognising that there are also broad patterns of variation in the plant's form that do not map neatly onto use. It is, in a sense, a way of describing the real diversity of cannabis without forcing it into the narrow channel of the Linnaean and Lamarckian names.

No scheme has won universal acceptance, and readers should be wary of any source that presents one as settled. What the debate shows, above all, is that the categories we use to describe cannabis are themselves historical products, shaped by the evidence available, by the concerns of the people doing the classifying, and by centuries of human selection on the plant itself.

When botany went to court

The species problem might sound like a quarrel for herbarium specialists, but it has had very practical consequences. In the twentieth century, courts were asked whether statutes naming "Cannabis sativa" covered plants labelled indica (McPartland and Guy, 2017).

The logic is easy to follow. If a law prohibits a named species, and there is more than one species in the genus, then a defendant might argue that the plant in question belongs to a different species and falls outside the law. Whether that argument succeeds depends on the taxonomic question: is C. indica a separate species, or merely a form of C. sativa? A monotypic taxonomy such as Small and Cronquist's, with all cannabis in one species, closes the gap. A taxonomy with several species opens it.

The episode illustrates a recurring theme of this series: questions about cannabis that look purely scientific turn out to have legal and political weight. How the plant came under legal control is the subject of the fifth article.

Where did cannabis begin? Reading the pollen record

The "Central Asia" tradition

Biogeographers have traditionally placed the origin of cannabis loosely in "Central Asia" (McPartland, Hegman and Long, 2019). The phrase has been repeated so often that it has the ring of established fact. But "Central Asia" is a vast and vaguely bounded region, and the phrase was a loose characterisation rather than a tested conclusion. Wild-looking cannabis today is a poor guide by itself, because some apparently wild stands may be descended from cultivated plants that escaped, a process known as feralisation. A map of where cannabis grows untended now is partly a map of where people once grew it.

How a pollen synthesis works

To test the traditional view, McPartland, Hegman and Long (2019) turned to the fossil pollen record. Their approach, a synthesis, is worth understanding, because it is quite different from a single excavation.

Pollen is one of the most durable biological materials in nature. Each grain has a tough outer wall that resists decay, especially in waterlogged, oxygen-poor environments such as lake beds and peat bogs. Every year, plants shed pollen into the air, and some of it settles into these deposits, layer upon layer. Palynologists, the scientists who study pollen and spores, drill long cores of sediment, slice them into thin samples, extract and count the pollen grains in each, and date the layers, often by radiocarbon. The result is a kind of vegetation diary for the landscape around the site, showing which plants were abundant at which periods.

Cannabis is well suited to this kind of study in one respect. As a wind-pollinated plant, it releases large quantities of pollen, which travels on the air and is likely to end up in sediments. In another respect it is difficult, because of the resemblance to hops discussed above.

A single pollen core tells you about one place. A synthesis gathers the published results of many such studies, re-examines how cannabis-type pollen was reported in each, and assembles the results into a map across space and time. McPartland, Hegman and Long (2019) synthesised 155 fossil pollen studies from Asia. Applying their probabilistic ecological criteria to separate likely Cannabis from likely Humulus, they looked for where and when cannabis pollen first appears and how its distribution changes.

The north-eastern Tibetan Plateau

The synthesis concluded that the genus probably originated on the north-eastern Tibetan Plateau (McPartland, Hegman and Long, 2019). This is a high, cold and dry region on the margins of the great plateau that dominates the interior of Asia, today largely within the Chinese province of Qinghai and neighbouring areas. It is not far from the broad "Central Asia" of tradition, but it is a much more specific claim, and it is grounded in a systematic body of evidence rather than general impression.

The same study found that cannabis pollen was present in India by about 32,600 years ago (McPartland, Hegman and Long, 2019). That date lies deep in the last Ice Age, long before farming anywhere in the world. It is a reminder that cannabis was spreading across Asia as a wild plant, by natural means, for tens of thousands of years before any human began to cultivate it.

Europe: native rather than imported?

A companion synthesis of European pollen records, together with a critical review of archaeological evidence, reached a conclusion that challenged another piece of received wisdom. McPartland and Hegman (2018) argued that cannabis was indigenous to Europe long before people arrived, and that cultivation there began in the Copper or Bronze Age. This challenges the older view that humans carried the plant westward from Asia.

What pollen cannot tell us

It is essential to keep in mind the limit flagged at the start. Pollen data cannot by themselves show human use, because wild cannabis is an abundant member of the steppe and montane vegetation of Central Eurasia (Ren et al., 2019). A rise in cannabis pollen in a core might reflect people clearing land and growing the plant. It might equally reflect a change in climate that favoured wild cannabis, or the opening of disturbed ground by natural processes. Pollen is an excellent witness to where the plant grew, and a very poor witness to what people did with it.

The syntheses also inherit the uncertainties of the probabilistic method used to separate cannabis from hops, so the conclusion about the north-eastern Tibetan Plateau is best read as the most probable answer given the current evidence and methods, not as a final verdict.

Domestication: reading the genome

What domestication means

Domestication is not a single event like the invention of a tool. It is a gradual process in which people, by choosing which plants to harvest, keep, sow and protect, change the plants' genetic make-up over many generations. Domesticated plants often differ from their wild ancestors in predictable ways. In cereals, for example, seeds that stay on the plant instead of scattering are a classic sign, because they make harvesting easier. Domestication is therefore something that happens to a population over time, and it can be hard to say exactly when gathering wild plants turned into managing them, and when managing them turned into creating a distinct domestic lineage.

How whole-genome resequencing works

The most detailed genomic study to date was published by Ren and colleagues in 2021 (Ren et al., 2021). To understand what they found, it helps to understand what they did.

A genome is the complete set of genetic instructions in an organism, written in a four-letter chemical code. Once a high-quality reference genome exists for a species, researchers can resequence many other individuals, reading their DNA and comparing it with the reference. The most useful differences are single-letter variations scattered across the genome. Individuals that share many such variations are likely to be closely related; those that share few are more distantly related.

By comparing many genomes at once, researchers can sort individuals into genetic groups, reconstruct a family tree of populations, and, with appropriate models, estimate when different lineages split from one another. The approach is a little like using surnames, accents and family recipes to work out how a dispersed community is related, except that the "clues" are vastly more numerous and are analysed statistically.

Ren et al. (2021) resequenced 110 whole genomes. Their samples included wild-growing feral plants, landraces and modern cultivars from around the world. A landrace is a traditional local variety, maintained by farmers over generations and adapted to local conditions, as opposed to a modern cultivar produced by formal breeding. Feral plants are those growing without cultivation, which may include descendants of cultivated plants that escaped. Including all three kinds was important, because older genetic signals are more likely to survive in traditional and untended populations rather than in modern commercial lines.

What the study found

The study's headline conclusion was that C. sativa "was first domesticated in early Neolithic times in East Asia" (Ren et al., 2021). The Neolithic is the period when people began to farm, and its timing varies from region to region; the authors place cannabis domestication at the early end of that period in East Asia.

The study also concluded that all current hemp and drug cultivars diverged from an ancestral gene pool now represented by feral plants and landraces in China (Ren et al., 2021). In other words, the genetic roots of today's hemp and today's drug-type cannabis appear to lie in the same place, and traces of that ancestral population still survive in Chinese feral plants and traditional varieties.

The researchers divided their samples into four genetic groups:

1. Basal: the group closest to the ancestral gene pool.
2. Hemp-type: plants selected for fibre and seed.
3. Drug-type feral: untended plants related to drug-type lineages.
4. Drug-type: plants selected for resin.

They dated the split between hemp-type and drug-type lineages to the period after initial domestication (Ren et al., 2021). This is a significant finding. It supports the idea that selection for fibre and for resin produced separate lineages: people first domesticated cannabis, and then, over time, different communities or different purposes pulled populations in different directions. It matches the "companion species" picture well. The plant did not arrive in human hands already split into a fibre crop and a drug; the split appears to have been made by people.

The authors placed East Asia alongside rice, millets and soybean as a domestication "hot spot" (Ren et al., 2021). This puts cannabis in good company. East Asia was one of the world's great centres of early agriculture, and the idea that cannabis belongs to the same suite of early domesticated plants fits what is known about the region's long farming history. Later articles in this series return to hemp as a fibre, food and paper plant in China.

Reading the 12,000-year estimate

The study's most widely quoted number is that early domesticated ancestors of hemp and drug types diverged from basal cannabis about 12,000 years BP, with a 95% confidence interval of 6,458–15,728 years BP (Ren et al., 2021). It is worth slowing down here, because the way such numbers circulate often strips away the very information that makes them meaningful.

First, this is a model-based estimate, not a date read off an object. No one dug up a 12,000-year-old domesticated plant. Instead, the researchers used the patterns of genetic difference between groups, together with assumptions about how quickly genetic changes accumulate and how many generations pass per year, to work backwards to an approximate time of divergence. Different reasonable assumptions could shift the result.

Second, the confidence interval matters as much as the headline figure. A 95% confidence interval is, informally, a range that reflects the uncertainty in the estimate given the data and the model. Here the range runs from roughly 6,500 to nearly 16,000 years ago: a span of more than 9,000 years. At one end, that would place the divergence well into the period when farming was established in East Asia; at the other, before the end of the last Ice Age. The single figure of 12,000 years is simply a point within that wide band.

Third, "divergence from basal cannabis" is not quite the same as "the moment of domestication". Genetic lineages can begin to separate before or after people start deliberately managing plants, and the relationship between a genetic split and a cultural change is a matter of interpretation.

McPartland's critique: how East is "East Asia"?

The genomic study did not go unchallenged. McPartland questioned the "East Asian" label itself. In a 2021 commentary he noted that "the majority of accessions in their 'basal cannabis' group (9 of 16) came from Xinjiang or Qinghai—that's Central Asia by most definitions, not East Asia" (Beyond THC, 2021).

This is a pointed criticism, and it connects neatly with the pollen evidence. Qinghai lies on the north-eastern Tibetan Plateau, the very region that McPartland's own pollen synthesis identified as the genus's probable centre of origin. Xinjiang, in China's far north-west, borders the countries usually grouped as Central Asia. Both regions lie within the modern borders of China, but whether they count as "East Asia" depends on whether you are drawing a political map or a geographical and cultural one.

The critique and the original study still agree that the genetic roots of cultivated cannabis lie broadly in or near what is now China; the argument is about how precisely to locate them and what to call the region.

Why the genomic picture remains under revision

Two further problems keep the picture under revision. The first is hybridisation. Modern domestic and feral cannabis populations hybridise extensively. Over centuries this blurs the genetic boundaries that researchers try to detect, and it can make a feral population look older or younger, more or less "basal", than it really is. Disentangling ancient patterns from recent mixing is a statistical challenge, and the results depend on the methods used.

The second is sampling. The review behind this series notes that the genomic evidence is complicated by sampling biased toward China and commercial cultivars. Any genetic study can only find ancestry in the populations it includes, so a study that samples many Chinese landraces and many modern commercial varieties, but comparatively few traditional populations from elsewhere, may tilt its conclusions accordingly, even with entirely careful analysis.

A promising way forward is ancient DNA: genetic material recovered directly from archaeological plant remains. Unlike modern samples, ancient specimens are snapshots from a known time and place, unaffected by later hybridisation. The review identifies ancient DNA from archaeobotanical remains as one of the most promising directions for future research.

Fitting the strands together

At first sight, the evidence in this article might seem to point in different directions. The pollen record suggests the genus originated on the north-eastern Tibetan Plateau and had spread widely across Asia and into Europe long before farming. The genomic evidence suggests a single early-Neolithic domestication in East Asia. The archaeological evidence, discussed in the next article, includes very early cannabis seeds from Japan and records that appear at roughly the same time in Europe and East Asia, which some researchers have read as pointing to more than one region of early human use (Long et al., 2017).

These positions are not necessarily contradictory. A coherent picture would run something like this. Cannabis evolved as a wild plant in the interior of Asia and spread naturally over tens of thousands of years. Wild plants may have been gathered for seed and fibre by people in several regions. At some point, probably in the early Neolithic and somewhere in the eastern half of Asia, one lineage was brought under sustained human management and became the ancestor of all today's cultivated cannabis (Ren et al., 2021). Later, as that domesticated cannabis spread and was put to different uses, selection for fibre and for resin drove the hemp-type and drug-type lineages apart.

This reconciliation depends on keeping the kinds of evidence distinct. Pollen speaks to presence. Seeds and fibres speak to use. Genomes speak to ancestry and divergence. Chemical residues, which appear much later in the record, speak to psychoactive use. Each answers a different question, and the apparent conflicts often dissolve once we ask which question each piece of evidence is equipped to answer.

The naming debates fit into this picture too. If hemp-type and drug-type lineages diverged after domestication, and if modern populations have since been crossed and recrossed, then it is no surprise that every attempt to divide cannabis into neat species has run into difficulty. The plant's variation is real, but much of it is the product of human selection layered over populations that interbreed freely. Small and Cronquist's finding that variation is "continuous for all characters", McPartland's description of commercial labels as "almost meaningless", and Clarke and Merlin's use-based groups can all be read as different responses to the same underlying reality.

What we know, and what we don't

It may help to finish with a plain summary of where the evidence stands.

What is reasonably well established

• Cannabis belongs to the family Cannabaceae, together with hops (Humulus) and several genera formerly placed in the Celtidaceae, and its lineage separated from its relatives many millions of years ago (McPartland, 2018).
• Cannabis and hops pollen are hard to distinguish, which complicates the fossil record and requires probabilistic methods to interpret (McPartland, Hegman and Long, 2019).
• Variation within the genus is extensive and, on many measures, continuous, which is why the number of species has been disputed since Linnaeus and Lamarck (Small and Cronquist, 1976; McPartland, 2018).
• Scientific and vernacular names for cannabis conflict, and modern commercial "Sativa" and "Indica" labels have been argued to be botanically "almost meaningless" (McPartland and Guy, 2017; McPartland, 2018).
• Wild cannabis was widespread across Central Eurasia, so pollen alone cannot demonstrate human use (Ren et al., 2019).
• On the best current genomic evidence, today's hemp and drug cultivars share an ancestral gene pool represented by feral plants and landraces in China, and the hemp-type and drug-type lineages split after domestication (Ren et al., 2021).

What is probable but not settled

• That the genus originated on the north-eastern Tibetan Plateau, a conclusion that rests on a synthesis of 155 Asian pollen studies and on probabilistic assignments of ambiguous pollen (McPartland, Hegman and Long, 2019).
• That cannabis was native to Europe long before people arrived, with cultivation beginning there in the Copper or Bronze Age (McPartland and Hegman, 2018).
• That domestication took place once, in the early Neolithic, in East Asia (Ren et al., 2021).

What remains open or contested

• The species question. No taxonomy, whether Small and Cronquist's single species, the older two-species view, or Clarke and Merlin's four groups, commands universal agreement.
• The precise date of domestication. The often-quoted figure of about 12,000 years is a model-based estimate with a 95% confidence interval of 6,458–15,728 years BP (Ren et al., 2021).
• The precise place. Whether "East Asia" is the right label, given that most of the basal accessions came from Xinjiang or Qinghai, has been questioned (Beyond THC, 2021).
• The effects of hybridisation and uneven sampling on the genetic results, which mean the genomic picture remains under revision.

That last group is not a weakness of the field so much as a mark of its honesty. The history of cannabis has suffered from confident claims built on thin evidence. The researchers discussed here have, for the most part, done the opposite: they have tested inherited assumptions, stated their uncertainties and argued in the open.

Where the series goes next

With the plant's identity and origins in view, the next article turns to the earliest physical traces of people and cannabis together: the seeds from Okinoshima in Jōmon Japan, the debate over whether human use began in one region or several, and the role of steppe peoples in carrying hemp across Eurasia. As we shall see, the distinctions between pollen, seed and chemical residue set out at the start of this article are exactly what is needed to make sense of that record.

Sources

Beyond THC (2021) 'Was cannabis first grown in eastern China?', 19 July 2021. Available at: https://beyondthc.com/was-cannabis-first-grown-in-eastern-china/.

Clarke, R.C. and Merlin, M.D. (2013) Cannabis: Evolution and Ethnobotany. Berkeley: University of California Press. doi:10.1525/9780520954571.

Long, T., Wagner, M., Demske, D., Leipe, C. and Tarasov, P.E. (2017) 'Cannabis in Eurasia: origin of human use and Bronze Age trans-continental connections', Vegetation History and Archaeobotany, 26, pp. 245–258. doi:10.1007/s00334-016-0579-6.

McPartland, J.M. (2018) 'Cannabis systematics at the levels of family, genus, and species', Cannabis and Cannabinoid Research, 3(1), pp. 203–212. doi:10.1089/can.2018.0039.

McPartland, J.M. and Guy, G.W. (2017) 'Models of Cannabis taxonomy, cultural bias, and conflicts between scientific and vernacular names', The Botanical Review. doi:10.1007/s12229-017-9187-0.

McPartland, J.M. and Hegman, W. (2018) 'Cannabis utilization and diffusion patterns in prehistoric Europe: a critical analysis of archaeological evidence', Vegetation History and Archaeobotany, 27, pp. 627–634. doi:10.1007/s00334-017-0646-7.

McPartland, J.M., Hegman, W. and Long, T. (2019) 'Cannabis in Asia: its center of origin and early cultivation, based on a synthesis of subfossil pollen and archaeobotanical studies', Vegetation History and Archaeobotany, 28(6), pp. 691–702. doi:10.1007/s00334-019-00731-8.

Ren, G., Zhang, X., Li, Y., Ridout, K., Serrano-Serrano, M.L., Yang, Y., Liu, A., Ravikanth, G., Nawaz, M.A., Mumtaz, A.S., Salamin, N. and Fumagalli, L. (2021) 'Large-scale whole-genome resequencing unravels the domestication history of Cannabis sativa', Science Advances, 7(29), eabg2286. doi:10.1126/sciadv.abg2286.

Ren, M., Tang, Z., Wu, X., Spengler, R., Jiang, H., Yang, Y. and Boivin, N. (2019) 'The origins of cannabis smoking: chemical residue evidence from the first millennium BCE in the Pamirs', Science Advances, 5(6), eaaw1391. doi:10.1126/sciadv.aaw1391.

Small, E. and Cronquist, A. (1976) 'A practical and natural taxonomy for Cannabis', Taxon, 25(4), pp. 405–435. doi:10.2307/1220524.