Cannabis Plant Anatomy: What the Microscopy Actually Shows About Where Cannabinoids Come From

Cannabinoids are made and stored almost entirely in glandular trichomes on the surface of female floral bracts. Mass spectrometry imaging of leaf cross sections has shown directly that these compounds are not an integral part of leaf tissue at all: they sit on the surface, in the glands.

The stalked and sessile glandular trichomes are not simply large and small versions of the same structure. They differ in secretory cell number, autofluorescence, and terpene profile, and current evidence supports stalked trichomes developing from sessile-appearing precursors.

Nearly every claim a patient encounters about cannabis quality rests on an anatomical assumption. Whether sugar leaf is worth keeping, whether trichome color tells you anything, whether roots are medicinal, whether the resinous structures are calyxes or bracts. Some of those assumptions are supported by microscopy and some are horticultural habit that hardened into fact.

There is also a straightforward consumer protection point. The percentage tables that assign cannabinoid content to roots, stems, nodes, and pistils are not drawn from any published dataset. Several versions circulating online openly admit this in a footnote and then present the numbers anyway.

The definitive localization work comes from laser microdissection and imaging mass spectrometry rather than from bulk extraction of plant parts.

In 2013, Nizar Happyana, Oliver Kayser, and colleagues published a study in Phytochemistry in which they isolated glandular trichomes by laser microdissection across an eight-week flowering period and analyzed as few as 25 collected cells by liquid chromatography with mass spectrometry, with cryogenic nuclear magnetic resonance used for confirmation. Tetrahydrocannabinolic acid, cannabidiolic acid, and cannabigerolic acid were the most abundant compounds in every sample analyzed. Their decarboxylated forms were present at significantly lower levels, which is the expected finding in living tissue. Cannabichromene and cannabinol appeared only as minor components, and only in intact capitate-stalked trichomes and their heads from eight-week plants.

One result from that study is easy to miss and worth keeping. Cannabinoids were detected not only in the glandular heads but in the stems of capitate-stalked trichomes, which implies the biosynthetic and storage machinery is distributed across the secretory unit rather than confined to the resin head.

A 2023 study in Phytochemical Analysis by Marcus Lorensen, Christian Janfelt, and colleagues at the University of Copenhagen used desorption electrospray ionization and matrix-assisted laser desorption ionization mass spectrometry imaging to map cannabinoids and flavonoids across cannabis leaves. Imaging of sugar leaf cross sections at 20 micrometre resolution confirmed that the cannabinoids were not an integral part of the leaf tissue itself but originated from the trichomes on its surface. Cannabigerolic acid was associated specifically with capitate-stalked trichomes, while THCA and CBDA appeared in both capitate-stalked and smaller glandular trichomes.

The most cited anatomical advance of the past decade is a 2019 study in The Plant Journal from Samuel Livingston, Jonathan Page, Lacey Samuels, and colleagues at the University of British Columbia, which used two-photon microscopy and intrinsic autofluorescence to distinguish trichome types functionally rather than by size.

Stalked glandular trichomes showed blue autofluorescence that correlated with high cannabinoid levels, contained 12 to 16 secretory disc cells, and produced strongly monoterpene-dominant terpene profiles. Sessile trichomes on mature flowers and vegetative leaves showed red-shifted autofluorescence, contained eight secretory disc cells, and had terpene profiles that were considerably less monoterpene-dominant. Both the autofluorescence patterns and the disc cell counts supported a developmental model in which stalked trichomes arise from apparently sessile precursors rather than forming as a separate lineage.

Transcriptomic analysis of isolated floral trichomes in the same study found strong expression of cannabinoid and terpene biosynthetic genes, uncharacterized genes co-expressed with CBDA synthase, and two previously unknown monoterpene synthases. The authors framed stalked trichomes as metabolically specialized for monoterpene production, which gives a concrete anatomical basis for why the aroma of cured flower concentrates where it does.

A 2023 study in the Journal of Cannabis Research by Zamir Punja and colleagues at Simon Fraser University added developmental detail using light and scanning electron microscopy on two genotypes across three to eight weeks of flowering. Glandular heads ranged from 40 to 110 micrometres in diameter and stalk lengths varied enormously, from 20 to 1,100 micrometres, produced by differential elongation of epidermal and hypodermal cells. Sessile-capitate forms predominated at three weeks; significantly more stalked-capitate trichomes were present at six weeks, and more on the lower bract surfaces than the upper.

Search for cannabis anatomy and you will find a table assigning cannabinoid, terpene, and flavonoid percentages to roots, stems, nodes, fan leaves, sugar leaves, trichomes, pistils, colas, and bracts. Trichomes are usually given 60 to 70 percent cannabinoids, sugar leaves 10 to 15, nodes 1 to 2, roots under 1.

Those numbers do not come from any published study. Several versions of the table say so directly, in a footnote below the numbers, noting a lack of comprehensive studies and citing nothing. They then get reproduced without the footnote.

What the literature does support is different in kind. It supports statements about localization, that cannabinoids are synthesized and stored in glandular trichomes on floral bracts and are surface features rather than tissue constituents. It supports relative statements, that stalked trichomes carry higher cannabinoid levels and more monoterpene-dominant profiles than sessile ones. It supports absolute statements about specific compounds in specific tissues, such as the root analysis discussed below.

What it does not support is a clean percentage split across nine plant structures, because nobody has published the quantitative survey that would produce one. Confidence intervals, chemovar dependence, and growth-stage dependence would all be required, and the trichome data already show that maturity varies within a single bract.

Cannabis roots have their own literature, and it says something more specific than a percentage. A 2017 review in Cannabis and Cannabinoid Research by Natasha Ryz, David Remillard, and Ethan Russo assembled the available phytochemistry and concluded that cannabis roots are not a significant source of THC, CBD, or other known phytocannabinoids.

What roots do contain is a different chemical class. The review lists the triterpenoids friedelin at 12.8 mg per kilogram and epifriedelanol at 21.3 mg per kilogram, the alkaloids cannabisativine at 2.5 mg per kilogram and anhydrocannabisativine at 0.3 mg per kilogram, carvone and dihydrocarvone, and sterols including sitosterol at 1.5 percent, campesterol at 0.78 percent, and stigmasterol at 0.56 percent. Pliny the Elder described a decoction of the root for joint stiffness and gout in the first century, and the authors argue the phytochemistry provides some support for re-examining those traditional uses.

That is a more useful statement than any percentage. It says the root is not a cannabinoid source and may be an interesting source of something else, which is a research question rather than a consumer decision.

Stems, nodes, and pistils have received far less attention. The imaging work suggests the honest answer for any non-glandular tissue is that whatever cannabinoid is measured there is most plausibly contamination from adjacent trichomes rather than synthesis in place, since the leaf cross sections showed no cannabinoid within the tissue itself.

Three points where horticultural convention and botanical description diverge.

First, terminology. In the peer-reviewed microscopy literature, the resin-bearing structures of the female inflorescence are consistently described as bracts. The word calyx, common in cultivation writing, does not map onto that usage. The structure people point at when they say calyx is generally a bract, and the resin they are admiring is on its surface.

Second, sex. Sex determination in Cannabis sativa is governed primarily by an XX and XY chromosome system, with monoecious and hermaphroditic expression arising from interactions between the sex chromosomes and autosomal factors, modulated by hormonal and environmental influence. A 2026 review in Plant and Cell Physiology by Ewa Dubas and colleagues summarizes how this plasticity is exploited commercially: silver-based compounds such as silver thiosulfate induce staminate flower formation on genetically female plants, and ethephon promotes pistillate flowers on genetically male plants, both acting through ethylene signalling. Hermaphroditism is therefore not simply a defect but an expression of a regulated and manipulable system.

Third, the trichome color cue used to time harvest. Punja and colleagues documented that maturation of stalked-capitate glandular heads is accompanied by brown color development, reduced ultraviolet autofluorescence, senescence, and dehiscence, so the visual change is real and reflects something happening to the gland. They also documented that trichome formation is asynchronous, leaving trichomes at different stages of maturity on a single bract at the same time, and that post-harvest handling and drying themselves altered trichome morphology. A visual reading is a rough population summary, not a measurement.

The localization evidence is strong and convergent. Laser microdissection, two-photon autofluorescence microscopy, scanning electron microscopy, and two independent mass spectrometry imaging modalities all point to glandular trichomes on floral bracts as the site of cannabinoid synthesis and storage. When four different techniques agree, the conclusion is durable.

The quantitative distribution evidence is weak to absent. No published study has measured cannabinoid, terpene, and flavonoid content across every anatomical structure of the plant with the replication needed to state percentages. Anything presented in that format should be treated as an illustration rather than data.

Most of this work rests on a small number of chemovars. The Simon Fraser trichome study used two genotypes for its detailed time course and five more for the drying observations. Genotype effects on trichome number, stalk length, and maturation timing were significant within that small sample, which means genotype-specific findings should not be generalized to the species.

Sample preparation is a live issue in this field. Trichomes are fragile, they dehisce, they stick together, and the Simon Fraser group showed directly that drying alters their morphology. Every measurement of trichome structure is a measurement of a trichome that has been handled.

Autofluorescence as a proxy for cannabinoid content is elegant but indirect. The University of British Columbia work correlated blue autofluorescence with high cannabinoid levels; correlation across trichome types is not the same as quantification within one.

None of this work establishes the percentage of total plant cannabinoid content contributed by any given structure, supports the anatomy charts that assign such percentages, or shows that any non-glandular tissue synthesizes cannabinoids.

It also says nothing about clinical effect. Anatomy determines where the compounds are and how they should be handled. It does not determine what they do in a patient, and no part of this literature should be read as evidence for or against any therapeutic use.

Cannabis is one of the better-studied trichome systems in plant biology, alongside tomato, mint, and hops, and the work described here sits comfortably in that tradition. The transcriptomic identification of previously unknown monoterpene synthases in stalked trichomes is a normal, unglamorous advance of the kind that eventually makes engineered production possible.

For medicine, the practical consequence of trichome-based synthesis is variability. A compound produced in fragile surface glands that mature asynchronously, dehisce on handling, and change with drying is a compound whose concentration in a finished product depends heavily on process. That is an argument for extracted and standardized preparations wherever dose consistency matters clinically, and it is an anatomical argument rather than a pharmacological one.

The thing I want patients to take from plant anatomy is narrow: the medicine is on the outside of the flower, in structures small enough that a rough grind and a careless container cost you real material. That single fact explains more about product variability than any strain chart.

The rest of it I treat as interesting rather than actionable. Whether a structure is a bract or a calyx does not change anybody’s care. What does bother me is watching numbers with no source get repeated under a physician’s byline, and I would rather publish a page that says we do not have that measurement than publish the table everyone else copies.

Cannabinoids and most terpenes are made and stored in glandular trichomes on the surface of female floral bracts, with stalked trichomes carrying the highest cannabinoid levels and the most monoterpene-dominant profiles. Roots contain triterpenoids and sterols but no meaningful cannabinoid content. The percentage-by-plant-part charts in circulation have no published source.

The finding to carry forward is anatomical and well supported: a surface gland system, fragile and asynchronously maturing, that holds essentially all of the material of interest. The claim to set aside is any numerical breakdown of cannabinoid content by plant structure, which does not exist in the literature regardless of how confidently it is presented.