Verified Synthesis — R02: The Global Cannabis Cultivation Landscape
Version: v1.0 — SIGNED OFF by Danny 2026-08-31
Date: 2026-08-31 · Editor: Claude · Researchers: ChatGPT, Gemini
Ledger: ledger.md (v1.0, 38 claims) — every substantive claim cites a ledger ID inline like [R02-C02]
Executive summary
R02 asked a deceptively simple question: where is cannabis grown, under which systems, at what scale? The answer that survived verification is that cannabis is nearly everywhere and measured almost nowhere. Cultivation is documented in at least 155 countries covering roughly 96% of the world’s population — but that famous statistic describes geographic reach, not hectares or tonnes [R02-C001]. When you ask for the number everyone assumes exists — how much cannabis the world actually produces — the honest answer is that no scientifically defensible global figure exists, for area or for output. Every published global total is an aggregation of incompatible methods: seizure records multiplied by guessed interception rates, eradication counts multiplied by assumed yields, wet plants counted alongside dried flower and sieved resin [R02-C002, C005]. This extends the central lesson of R05 — canopy is not production — to the entire planet.
What does exist is a mosaic of unequal evidence. A handful of regulated markets (Canada federally; California, Oregon, Colorado at state level; Germany for imports) publish genuine administrative data, mandatory and auditable — but only for their licensed slice [R02-C020, C026–C028]. A small number of traditional producing regions have real crop surveys, now aging into history: Morocco’s Rif was measured properly in 2003 (134,000 ha, ~3,070–3,080 t potential resin, ~96,600 farming households) [R02-C006], Afghanistan between 2009 and 2012 (with extraordinary irrigated resin yields near 145 kg/ha) [R02-C010]. Almost everywhere else, what circulates as production data is enforcement exhaust — seizures, eradication sweeps, “reported origin” tags — which measures detection, not agriculture [R02-C005, C014].
The report’s second major thread is what happened inside the measured places. Regulated markets built more capacity than their closed borders could absorb: wholesale prices collapsed across every mature market [R02-C025], Canada destroyed over 1.68 million kilograms of unsold cannabis in five years [R02-C026], and California shed more than half its cultivation licenses while its licensed output still grew [R02-C027]. Meanwhile, legalization nowhere eliminated illicit cultivation — in California the compliance regime itself pushed legacy smallholders out of the legal market [R02-C030].
The third thread is provenance — of plants and of numbers. The world’s traditional cannabis cultures are real and deep, from Rif kif terraces to Bekaa hashish to Malawi cob-curing [R02-C004, C017]. But the names those traditions lend to modern products are not genetic certificates: strain names fail genotyping as identities [R02-C033], the one well-documented landrace-erosion case (the Rif’s Beldia, displaced by European hybrids) does not license a global extinction narrative [R02-C007, C034], and the most famous number in cannabis advocacy — “the world’s most valuable cash crop” — traces to a 2006 calculation that multiplied eradication counts by an assumed yield and a retail price, then compared the result against farm-gate values for corn [R02-C036]. R02’s deepest finding is a discipline: never compare cannabis numbers until population, product, period, unit, and method match.
What the evidence establishes
Global structure. Production is structurally bifurcated into herbal cannabis — ubiquitous, grown under every system — and resin/hashish, hyper-clustered in North Africa and Southwest Asia with distinct yields, processing traditions, and trade routes [R02-C004]. Any “largest producer” sentence that doesn’t say which product and which years is unanchored [R02-C035].
The measurement hierarchy. Four families of estimation exist, and they are not interchangeable: mandatory track-and-trace (accurate for legal canopy, blind to everything else, and not error-free — California documented thousands of per-plant data-entry corrections), satellite crop surveys (excellent for a place and period), eradication extrapolations, and seizure multipliers (directional at best) [R02-C005, C022]. Licensed area ≠ harvested area ≠ production: Health Canada’s own monthly series shows why even good administrative numbers must be month-stamped (Dec 2024: 1,453,902 m² indoor / 596 ha outdoor; an adjacent month: 1,412,077 m² / 689 ha) [R02-C020].
Regional anchors. Morocco’s 2003 survey is the gold-standard historical baseline, followed by measured contraction (120,500 ha in 2004; 72,500 in 2005; later official estimates ~47,000, then 21,000 ha by 2019) [R02-C006]. The Rif then underwent a documented agronomic revolution: imported European hybrids displaced the traditional Beldia landrace, tripling per-plant resin extraction (≈1.5–2.5% → 4.0–7.5%) at the cost of borehole irrigation, aquifer depletion, and genetic introgression [R02-C007]. Morocco remains Europe’s principal resin source, but its current tonnage is genuinely unknown — modeled output conflicts with seizure volumes [R02-C009]. Afghanistan’s surveys (2009–2012) established the world’s highest documented resin yields [R02-C010]; the EU’s herb is mostly grown inside the EU [R02-C013]; Lebanon’s legalization law has never operated [R02-C012]; Mexico never enacted the commercial framework its court rulings implied [R02-C016]; Thailand swung from mass registration (>1.1M home-grow registrants) toward re-restriction within three years [R02-C018]; Sub-Saharan Africa pairs deep traditional production with a new licensed export tier (Lesotho, South Africa) and no continental quantification [R02-C017].
Regulated-market dynamics. Structural oversupply and price collapse are now the documented norm in closed-loop markets: outdoor flower fell from ~$1,200–1,500/lb to ~$150–400/lb on the US West Coast; Colorado’s benchmark rate fell from $1,721 to $648/lb [R02-C025]. Canada’s tracking system provides the world’s clearest record of agricultural overproduction: 1.2–1.5 million kg held in vaults against ~450–550 t of annual sales, with 400–600 t destroyed per year [R02-C026]. Legal international trade is medical-only and INCB-gated; Canada is the largest exporter (~60 t FY2022–23, rising toward 80–100 t) and Germany the largest importer (31.4 t in 2023, surging past 60–70 t after its April 2024 reform) [R02-C028].
Illicit persistence. Nowhere did legalization end unlicensed cultivation. California’s peer-reviewed record shows compliance costs and local bans excluding the legacy smallholders legalization was supposed to absorb, while trespass grows on public lands continue to poison wildlife with banned carbamates and anticoagulant rodenticides — 70% of sampled northern spotted owls carried rodenticide exposure [R02-C030].
Origins. Genomics places domestication in East Asia ~12,000 years ago, with hemp and drug lineages diverging ~4,000 years ago within a single interfertile species; archaeology supports fiber and seed use by ~10,000–8,000 BCE [R02-C032].
What is probable but not settled
Morocco’s legal ANRAC program (>3,000 permits, ~2,500–3,000 ha, ~4,000 t authorized biomass in 2024) rests on agency announcements without independent audit [R02-C008]. Colombia’s dual structure — consolidated licensed CEA beside armed-group-controlled Cauca production — is well-reported but thinly measured [R02-C015]. The smallholder-livelihood pattern (0.2–1.5 ha family plots financing household survival in the Rif, Afghanistan, Pondoland) is consistent across sources but survey-era [R02-C021]. The chemotype contrast between Beldia and the hybrids that replaced it comes from field reports, not systematic assays [R02-C007 notes]. The US cultivation-employment slice (~100,000–125,000 FTEs of ~412,500 industry jobs) is single-source industry data [R02-C031].
Traditional practice & practitioner knowledge
The traditional processing repertoire is real, regionally distinctive, and largely untested by formal science: Moroccan dry-sieving of rainfed kif; Afghan garda threshing and pressing; Lebanese Red/Blonde sun- and shade-cure distinctions; Himalayan live hand-rubbed charas; Malawi’s cob-curing anaerobic fermentation; Thai stick bamboo curing [R02-C004, C017]. The corpus records these with respect as traditional practice — including the honest note that their sensory and chemical claims await study. Breeder pedigrees (like Jack Herer’s reported Haze × [NL#5 × Shiva Skunk] parentage) live in the same tier: recorded as lore, never as verification [R02-C037].
Disputes & open questions
The two researchers disagreed remarkably little; the report’s real conflicts are between data sources. Health Canada’s area figures differed between the two reports because the series is monthly — both were right for their month, which became its own ledger lesson [R02-C020]. Gemini’s regional depth carried several figures with unresolvable provenance — Paraguay’s “70–80% of regional supply” and “6,000–8,000 hectares” trace to enforcement extrapolations, not surveys, and are ledgered as estimate-lore [R02-C014]; the widely repeated producer-level allocation splits (A-grade 15–25%, extraction 35–50%, etc.) have no resolvable source at all and sit in the ledger explicitly as unresolved [R02-C029]. Verification itself was the story of this report: zero invented sources from either researcher, but seven bibliographic-identity failures caught across Gemini’s report and its correction round — every one a real paper wearing the wrong address (critique §2, §6a).
What we don’t know
This section is R02’s spine. We do not know global production volume or area [R02-C002]. We do not know Afghanistan’s post-ban cannabis trajectory [R02-C011]. We do not know Morocco’s current resin tonnage [R02-C009], Lebanon’s or Paraguay’s or Africa’s actual output [R02-C012, C014, C017], the global indoor/outdoor split [R02-C019], the unlicensed share of any market [R02-C030 notes], global cultivation employment [R02-C031], or a global landrace-erosion rate [R02-C034]. C038 turns these into a requirements list — what the world would have to measure before a defensible global picture could exist.
Then vs. now (archive delta)
The 2016–2020 teaching archive was California to the bone, and its worldview inherited the movement’s favorite numbers. R02 graded that inheritance directly. “The world’s most valuable cash crop” now has its receipts: Gettman’s 2006 arithmetic, eradication counts × 1 lb/plant × $1,600/lb retail, compared against crops measured at the farm gate. It is the corpus’s flagship example of denominator laundering, and the corpus can now teach why it felt true and isn’t supported [R02-C036]. The 2019 Mexico deck’s “worldwide leader” thesis was fact-checked against seven years of regulatory paralysis: not realized [R02-C016]. The pop-history timeline splits cleanly: “cultivated since 8000 BC” survives as an early-fiber inference; “Emperor Shennong, 2700 BC” is legendary attribution — the text is a Han-era compilation from around 200 CE [R02-C032]. And the Jack Herer strain story from the dispensary years becomes the perfect teaching example: a beloved 1990s Dutch polyhybrid named for an activist, exactly the kind of name the landrace-provenance hierarchy exists to classify [R02-C037]. The engine is now fact-checking its own founder’s bookshelf — respectfully, with sources.
Implications for cultivation practice
For the home grower and educator audience, R02’s payoffs are mostly epistemic, and they matter. First: treat every big cannabis number you encounter — production totals, “largest producer” rankings, job counts, cash-crop valuations — as unverified until you can name its population, product, period, unit, and method; most fail the test [R02-C002, C005, C031, C036]. Second: “landrace” seeds bought under geographic names are marketing tier-5 until proven otherwise; genuine landrace conservation is a real concern (the Rif shows how fast displacement happens) but authenticity requires accession history or genotyping, not a name [R02-C033, C034]. Third: the documented collapse of wholesale prices in every mature market is the economic backdrop for everything this corpus later says about growing systems — capacity is cheap, differentiation is not [R02-C025, extending R05-C029]. Fourth: sun-grown and greenhouse production dominate legal physical area even where indoor dominates the shelf — the “indoor is everything” assumption fails administrative data [R02-C019].
Sources
Registered as S-066+ in SOURCE_LIBRARY.md at sign-off. Principal sources: UNODC World Drug Report 2025/2026 + Morocco Cannabis Surveys 2003–2005 + Afghanistan Cannabis Surveys 2009–2012 · Health Canada CTLS (licensed area, market data, exports) · Statistics Canada / Canadian Cannabis Survey · California DCC (Market Outlook 2024, licensing, dashboard corrections) · Colorado MED AMR · OLCC · EUDA EU Drug Market: Cannabis · BfArM import statistics · INCB Narcotic Drugs estimates · Chouvy & Afsahi 2014 (IJDP 25(3):416–423) · Afsahi & Chouvy 2015 · Chouvy & Macfarlane 2018 · Chouvy 2024 · Ren et al. 2021 (Science Advances 7:eabg2286) · Sawler et al. 2015 (PLOS ONE 10(8):e0133292) · Long et al. 2017 · Small 2015 · Clarke & Merlin 2013/2016 · Polson et al. 2023 (Land Use Policy 126:106531) · Dillis et al. 2021 (Land Use Policy 105:105369) · Bodwitch et al. 2019 (California Agriculture 73(3):177–184) · Gabriel et al. 2012/2015/2018 (verified tuples per critique §6a) · Gettman 2006 (claim-lineage) · Herer 1985 (claim-lineage) · FAOSTAT · Vangst/Whitney Economics 2026 · national statutes (Law 13-21 Morocco; Law 178/2020 Lebanon; Law 1787/2016 & Decree 811/2021 Colombia; KCanG 2024 Germany; CfPPA 2024 South Africa; Cannabis Regulation Act 2020 Malawi).
Changelog
- v0.1 — initial draft (2026-08-31), from ledger v1.0 (38 claims)
- v1.0 — signed off by Danny 2026-08-31; derivatives + content extraction released