Ask a nursery operator how much of California's cannabis carries hop latent viroid and most will repeat some version of "ninety per cent." That figure comes from one nursery's own paying testing customers, who most likely sent samples in because they already suspected a problem, not from a scientific survey of cannabis at large ^[1]. The number that does come from a large, peer-reviewed prevalence study is smaller, better documented, and comes with sample sizes, provinces and years attached ^[2], and almost nobody repeats it.

Peer-reviewed detection rate: 25.6%; country-wide average, Canada, 2020–2023
Range across that same dataset: 5.3–92%; by province and year, 15,947 samples
Widely repeated industry figure: 90% of facilities; one nursery's client base, not peer-reviewed
First confirmed in cannabis: 2019; two independent reports, California

First confirmed in cannabis in 2019, by two labs at once

Hop latent viroid (HLVd) is a small, circular RNA molecule, about 256 nucleotides long, in the family Pospiviroidae ^[3]. Unlike a virus, it carries no protein coat and no genes of its own: it replicates by hijacking enzymes the host plant already has, which is part of why an infected mother plant can look ordinary for months. It was already known in hop yards before anyone found it in cannabis. Two independent Californian groups confirmed it there for the first time in 2019, within weeks of each other, on separate farms. Bektaş and colleagues found the viroid in every symptomatic plant they sampled from two locations, and in only a few of the accompanying symptomless ones: evidence of an association with stunting, not a population estimate ^[4]. Warren, Mercado and Grace reported the same viroid, in cannabis showing the same stunting pattern, independently, the same year ^[5]. Neither paper claimed to know what share of cannabis carried it.

Growers had already started calling the resulting symptoms "dudding" by the time peer-reviewed reviews began describing it formally: stunted growth, brittle or hollow stems, shorter internode spacing, and thinner, less trichome-dense flower ^[3]. That question, how common dudding actually is, is where the evidence gets thin.

Where the widely repeated number actually comes from

In August 2021, Dark Heart Industries, a Californian clone nursery that sells its own HLVd testing service, announced the results of just over 200,000 tissue tests run on samples from more than 100 licensed California cultivation businesses between August 2018 and July 2021. Ninety per cent of those businesses had at least one positive test at some point, and about a third of the roughly 200,000 individual tests came back positive ^[1]. Those two figures, 90% of facilities and a third of tests, are almost certainly the source of every "90% of cannabis has HLVd" claim now circulating: the two numbers get flattened into one in the retelling, and the caveats disappear entirely.

The figure has since travelled well beyond trade press. A 2023 peer-reviewed review of HLVd in cannabis cites the same 90% figure directly from industry reporting ^[3], which is part of why it now reads like an established scientific fact rather than one company's testing log repeated at second hand.

The one study built to answer the population question

The closest thing to a real prevalence survey is Punja and colleagues' study, published in the Canadian Journal of Plant Pathology, using RT-PCR (reverse-transcription polymerase chain reaction, the standard viroid-detection method) on 15,947 samples submitted from licensed cannabis facilities across nine Canadian provinces between 2020 and 2023 ^[2]. Detection ranged from 5.3% to 92% of samples depending on province and year, and averaged 25.6% country-wide ^[2].

Fig. 1A 17-fold spread between the lowest and highest province-year, in one country, in one dataset. That spread is itself a clue that submission habits, not biology, drive much of the variation.Horus

That is genuinely useful: a large sample, a named peer-reviewed method, a stated period. It is still not a random population survey, for the same structural reason as Dark Heart's figures. These are samples growers or facilities chose to send to a diagnostic lab, which over-represents plants already showing symptoms or already under suspicion, not a blind draw from every licensed cannabis plant in Canada. The 17-fold spread between the lowest and highest province-year is itself a clue: biology rarely varies that much region to region within one country, but how often growers in a given province routinely screen mother stock versus only test when something looks wrong plausibly does.

The four sources below are not interchangeable: two are first-detection reports that never claimed a population rate, one is an unpublished industry figure, and one is the peer-reviewed study built to measure prevalence directly.

Source	Evidence type	Sample size and period	Population sampled	Reported rate
Bektaş et al. (2019) ^[4]	Peer-reviewed, first report	2 California sites, symptomatic and asymptomatic plants	Plants already showing stunting, plus nearby comparisons	Not a prevalence estimate
Warren et al. (2019) ^[5]	Peer-reviewed, first report	California, independent of ^[4]	Plants showing the same stunting pattern	Not a prevalence estimate
Dark Heart Industries (2021) ^[1]	Industry press release	~200,000 tests, 100+ facilities, Aug 2018–Jul 2021	Paying testing clients, California	90% of facilities with 1 or more positive; 33% of tests positive
Punja et al. (2024) ^[2]	Peer-reviewed	15,947 samples, 9 provinces, 2020–2023	Samples submitted to a diagnostic lab, licensed Canadian facilities	25.6% average; 5.3–92% by province and year

Dudding, quantified: what changes when a plant tests positive

Grower forums have described dudding since around the time HLVd was first confirmed in cannabis: a mother plant that just seems weaker than it used to be, thinner stems, less resin, a little less of everything. For years that description was the only evidence available. Punja and colleagues' dataset is the first large, peer-reviewed comparison of confirmed-positive against confirmed-negative plants, and it puts numbers on the folk description: RT-PCR-positive plants showed 12 to 42% lower inflorescence stem length, fresh weight and plant height than RT-PCR-negative plants from the same dataset, along with measurably fewer glandular trichomes and lower THC and terpene content ^[2].

Read that precisely. It is a comparison between plants found positive and negative within a large submitted-sample dataset, not a randomised trial where researchers deliberately infected half a set of identical clones and left the other half untouched. It is real, peer-reviewed, and a long way ahead of an anecdote. It is not the same thing as controlled experimental proof that HLVd alone, with every other variable held constant, causes exactly that range of loss.

Testing: what a result actually tells you

RT-PCR and RT-qPCR (quantitative RT-PCR, which also estimates how much viroid RNA is present) are the standard detection methods, and viroid distribution inside an infected plant is uneven enough that sampling matters as much as the assay itself. Guidance in the peer-reviewed literature recommends multiple leaf samples, old and new growth, taken from lower to upper stem, because a single sample from the wrong tissue can come back negative in a plant that is genuinely infected ^[3].

Timing compounds the problem. In a 2025 inoculation study, Punja and colleagues introduced HLVd-infected sap into healthy stems and then tracked where the viroid became detectable: root tissue tested positive within two to three weeks, but foliage did not test positive until four to six weeks after inoculation ^[6]. A cutting taken, and tested, immediately after a mother plant picks up a new infection can genuinely test negative today and positive next month, with no change in how the plant was handled in between.

Sources: [3] Adkar-Purushothama, Sano and Perreault (2023), Viruses 15(3):681, for sampling across old and new leaves, lower to upper stem, and uneven viroid distribution; [6] Punja et al. (2025), Plants 14(5):830, for detection times after stem inoculation. Generic RT-PCR / RT-qPCR workflow, not any one laboratory’s protocol.

Fig. 2One negative result covers one sample, one tissue and one point in time. It does not clear the whole plant, and it does not clear a sibling cutting.Horus

A negative result, then, covers exactly what it tested: one tissue sample, from one plant, at one point in time. It does not confirm the whole plant is clean, and it says nothing about a sibling cutting taken from different tissue, at a different time, from the same mother.

Why this page sits between two sections

HLVd is a genetics problem as much as it is a disease. The transmission mechanism and symptom biology belong to the diseases section's hop latent viroid profile, and the sampling walkthrough belongs to how to test your plants for viruses and viroids; this page does not repeat either. What belongs here is the screening decision a genetics or nursery programme has to make before any of that, which is covered in full in the genetics section's clean-stock cluster, in clean-stock programmes and HLVd testing. If a mother plant you already keep seems to be declining for no obvious reason, do mother plants degrade? covers the genetic-drift myth this same viroid is usually the real explanation for.

None of that decision actually depends on knowing whether the true rate where you operate is 5% or 90%. A single infected mother converts every future cutting taken from it, so the screening rule is the same regardless of the exact number: test every incoming clone or candidate mother before it joins the main stock, during quarantine rather than after; sample multiple leaves, old and new, given how unevenly the viroid distributes; retest existing mothers on a fixed interval tied to your own propagation cycle, not once and never again; and rogue, or send for specialist meristem-tip remediation, anything that comes back positive rather than waiting for a "high enough" prevalence figure to justify the cost.

A nursery running 20 mother lines can see why the exact number matters less than it first seems to. Apply the Canadian average of 25.6% and you would guess roughly five lines are infected. Apply the Californian client-base figure of a third and you would guess seven. Neither guess tells you which five, or which seven, and a facility that has never tested cannot rule out either number. The practical conclusion is the same whichever figure you start from: test all 20, not the fraction a population rate suggests should be affected.

Genetic-stock record-keeping is already mandatory in some licensed markets, which is worth using rather than fighting. Malta's ARUC cultivation standards already require seeds and clones to be botanically identified and traceable, with seed receipts notified to the regulator; adding an HLVd result to a record that already has to exist costs little beyond the test. Check the law section for what your own jurisdiction actually requires before assuming this applies to you.

A narrow page, on purpose

Most pages in this section answer a broad question with a workable range, such as how yield scales with light or roughly what a gram costs to grow. This one is different. A single, widely repeated fact turned out to trace back to one company's client list, and the closest thing to a genuine answer is a Canadian diagnostic dataset that is explicit about what it does and does not represent, not a clean national percentage. The same caution belongs on cultivar names: another place a widely repeated number turns out to rest on less than it seems.

Showing that gap between what circulates and what has actually been checked is worth doing even, especially, when the honest answer is "we don't fully know, and here is exactly why." Revisit this page if a randomised, facility-level prevalence survey is ever published. Until then, this is the most complete comparison available of what has actually been measured against what has merely been repeated.