Research Brief — R02: The Global Cannabis Cultivation Landscape

Report ID: R02 · Wave 1 · Executed 3rd (after R05 pilot, R01 foundation) Brief version: v1.0 — FROZEN on Danny’s explicit confirmation, August 30, 2026 (content = v0.2 verbatim; PROTOCOL v1.3 version lock: both researchers run this identical version, and the critique will record it; any change requires a new version and re-freeze) Date: August 30, 2026 · Drafted by: Claude (editor) · Protocol: v1.3 — first report born under v1.3 (R01’s verification lessons codified) Assigned to: ChatGPT deep research + Gemini deep research (independently, same frozen brief)

Assignment

Produce an evidence review of how, where, and at what scale cannabis is cultivated worldwide — the geographic, systemic, cultural, and market landscape of the crop this entire corpus studies. R05 compared cultivation environments; R01 established the agricultural-systems foundation. R02 answers: who actually grows this plant, where, under what systems and traditions, at what scale, and how reliable is anyone’s data about it?

Boundary with R03: R03 (later) covers legal and illicit production context — policy regimes, enforcement dynamics, and prohibition’s consequences. R02 describes the cultivation landscape itself: geography, systems, scale, traditions, and markets. Where the two unavoidably touch (illicit production is a huge share of global cultivation), R02 reports what and where and how much, and leaves why and with what policy consequences to R03. Note the boundary explicitly wherever you defer.

This is not a market-hype report and not a prohibition polemic. The central intellectual challenge of R02 is that global cannabis production data is among the worst-quality data in agriculture — much of it inferred from seizures, extrapolated by enforcement agencies, or asserted by industry marketing. Treating estimate-quality honestly is the report’s core job, not an obstacle to it.

Standing requirements (PROTOCOL v1.3 — non-negotiable)

• Resolvable identifiers: every citation supporting a substantive claim must carry a resolvable DOI, stable URL, or independently retrievable identifier. An unresolvable citation makes the claim unsourced.
• Citation-identity integrity: for every substantive source, the authors, title, publication year, venue, volume/pages where applicable, and DOI or stable identifier must resolve to the same bibliographic object. A DOI that resolves to a different real paper fails identity verification.
• Exact claim support: a source supporting the general direction of a finding does not automatically support its stated magnitude, population, geography, time period, or causal wording. Preserve these distinctions in candidate claims.
• Corrections inherit the complete verification burden: a replacement citation or corrected statistic remains unverified until its existence, identity, and claim support are checked.
• Candidate claims + stopping rationale: end with a candidate-claims section (15–30 testable statements, each with supporting citations) and a search-stopping rationale explaining why searching concluded.
• Myth-check calibration: “the myth is unsupported” and “the opposite is true” are different findings. Do not overcorrect; unknown is an acceptable verdict.

Methodology requirements

Estimate-provenance discipline — the central trap of this report. For every quantitative production, acreage, or market figure, state HOW the number was produced. Seizure-derived and enforcement-sourced estimates carry known, large, directional biases — never present them with the same confidence as administrative data. A number whose provenance cannot be determined is treated as unsourced regardless of how often it is repeated.

Standardized estimate-provenance record. Every major quantitative estimate must carry these fields (a compact table or labeled inline block is fine — consistency matters more than format):

Terminology discipline. Drug-type/regulated cannabis includes intoxicating THC-dominant and other cannabinoid-rich floral production governed as cannabis in the relevant jurisdiction. Hemp refers to legally defined low-THC production, often — but not exclusively — for fiber, grain, seed, or cannabinoids. Because THC thresholds and product classifications differ by jurisdiction, every dataset must state its legal and product boundary; do not assume “hemp” is biologically or agriculturally uniform. Also keep distinct: licensed/regulated vs. unlicensed/illicit production, and canopy/area licensed vs. production volume (R05-C031 established these are not convertible — carry that discipline globally).

Landrace-provenance discipline. Distinguish: locally maintained farmer populations · documented regional accessions · formally sampled genetic populations · historical cultivar names · contemporary commercial products marketed as “landrace.” Claims of origin, antiquity, genetic distinctness, or genetic erosion require corresponding historical, ethnobotanical, accession, or population-genetic evidence. Breeder lore and seed-catalog descriptions may be recorded as claims, never treated as verification. Attribute traditional knowledge to identifiable communities and sources where the evidence permits — this section must not become a geographic strain-name tour.

Scale-category discipline. “Smallholder,” “craft,” “SME,” and “industrial” vary dramatically across countries. Use jurisdictional definitions where available; otherwise describe measurable characteristics — area, labor model, capital intensity, license type, household dependence — without forcing farms into universal labels.

Comparator & functional-unit discipline (inherited from R01/R05): state denominators on any cross-country or cross-system comparison; date-stamp everything (this landscape changes yearly); flag when a finding is jurisdiction-specific.

Geography: genuinely global scope. North America and Europe will dominate the regulated-market data; do not mistake data visibility for cultivation dominance. Major cultivation regions with weaker measurement must receive proportionate attention without presuming their rank.

Key questions

A. The map

1. Where is cannabis cultivated at scale today? Map the major producing regions — North America (regulated + legacy), Latin America (Mexico, Colombia, Paraguay), North Africa (Morocco’s Rif), Southwest Asia (Afghanistan, Lebanon), sub-Saharan Africa, South/Southeast Asia, Europe (indoor + Albania et al.), Oceania — with the best-available characterization of each region’s dominant systems (outdoor field, greenhouse, indoor; smallholder vs. industrial) and honest data-quality grading per region.
2. What do UN (UNODC World Drug Report), national, and academic sources actually claim about global production scale — and what is the provenance and credibility of each estimate? If no defensible global production figure exists, establish that as a finding (extending R05-C031 worldwide).

B. Systems & scale structure

3. What is known about the distribution of cultivation systems globally — open-field, greenhouse/light-dep, indoor — and how does it differ between regulated markets, legacy/illicit markets, and traditional producing regions?
4. What is the scale structure — smallholder/peasant cultivation (how many households depend on it, where), craft/SME, and industrial/corporate — and how is regulation shifting it? Include what’s known about cultivation as a livelihood (e.g., Rif, Afghan, sub-Saharan contexts) at descriptive level (policy analysis → R03).
5. Home cultivation: what’s known about its prevalence where legal (Canada, US states, Uruguay, Malta, Germany, ACT) and estimated where not?

C. Traditions & landraces

6. Survey the major traditional cultivation cultures and landrace regions (Hindu Kush, Moroccan Rif, Lebanese Bekaa, Jamaican, Thai, Colombian, Mexican, Malawian, South Indian lineages) — cultivation methods, historical depth, and current status under market change. Descriptive and evidence-based; romanticize nothing, but document what these traditions actually developed (this seeds R09–R12 genetics/botany work).
7. What is happening to landrace diversity under legalization and hybridization pressure — is genetic erosion documented or asserted?

D. Cultivation-facing market structure (tightly scoped)

Cover ONLY cultivation-relevant market structure: producer counts and active production capacity · farm-gate/wholesale prices · cultivation oversupply and contraction dynamics · product allocation at the producer level · cultivation-linked exports by product type and mass. Explicitly deferred (out of scope): retail consumer segmentation, company valuations, general market forecasts, employment estimates beyond cultivation itself, detailed regulatory causation, and enforcement/illicit-market policy analysis (→ R03).

8. Regulated-market cultivation structure where data is good: licensed area, producer counts, farm-gate/wholesale price trends, oversupply dynamics (CA, CAN, CO, etc. — extends R05’s economics claims internationally).
9. What share of cultivation output flows to flower vs. extraction, and medical vs. adult-use vs. export, where measurable at the producer level?
10. International trade in cultivated product: the emerging legal export landscape (Colombia, Portugal, Israel, Africa→EU medical flows) vs. traditional illicit flows — descriptive mass/type/route only; policy → R03.

E. Synthesis targets

11. Where is the landscape data solid, where is it estimate-theater, and what would a defensible global picture actually require?
12. Fact-check the 6–10 most repeated public claims — e.g., “cannabis is the world’s most valuable cash crop,” “Afghanistan/Morocco is the world’s largest producer,” “most legal weed is grown indoors,” “there are X million people employed in cannabis cultivation” — with the calibration rule. For “legalization killed the illicit market,” test ONLY its measurable cultivation-landscape component: where jurisdictions estimate licensed vs. unlicensed supply, is there evidence that unlicensed production disappeared, contracted, persisted, or changed form? (Policy interpretation → R03.)

F. Corpus linkage layer

13. For major findings, tag connections to existing corpus claims (especially R05-C031 production-share gap, R05-C002 environment footprints by region, R01-C004/C005 adoption-data lessons) and downstream reports (R03 legal context, R09–R12 genetics, R21–R24 implementation). Present as a compact linkage table: finding · linked claim/report · relationship (extends / confirms / tensions / seeds).

Known controversies to handle carefully

• “World’s largest producer” claims rotate among Morocco, Afghanistan, and Mexico depending on source and year — usually seizure-inference. Trace each to provenance.
• “Most valuable cash crop” traces to decades-old advocacy arithmetic — check its lineage before repeating or refuting.
• Post-2022 Afghanistan: the drug-economy shifted dramatically under Taliban bans (opium collapse); what actually happened to cannabis there is genuinely uncertain — date-stamp hard.
• Industry market-size reports (billions-by-2030 genre) are marketing instruments; usable only as labeled market positions.
• Hemp contamination: many “cannabis acreage” figures silently include hemp — audit every large number for this.

Source & citation requirements

• Prefer: regulated-market administrative datasets (state/provincial/national); UNODC and government reports with provenance stated; peer-reviewed geography/economics/ethnobotany literature (e.g., the academic drug-geography corpus); FAO/EMCDDA-style institutional syntheses. Industry reports and journalism only as labeled positions or leads.
• Full citations (authors, year, title, publisher, DOI/URL) for every substantive claim. No orphan statistics — especially here.
• Report ranges and disagreements between sources, not averaged-away conclusions; date-stamp aggressively.
• Explicitly list data gaps — in this report, the gap map may be the most valuable deliverable.

Candidate claims for editorial review

Conclude with 15–30 concise, evidence-based claims phrased as testable statements with supporting citations, for R02 Claim Ledger review (R02-C001+). Do not assign final evidence status or confidence; the editor will.

Search-stopping rationale

End with a short statement of why searching concluded — which evidence lanes reached diminishing returns.

What “done” looks like

A structured report answering A–F with cited evidence and per-estimate provenance labels, a stated-limitations section, a top-10–15 strongest-sources list with reliability notes, an explicit data-gap map, the candidate-claims section, the corpus-linkage table, and the stopping rationale. In this report above all others: a well-documented “nobody knows” beats a confident number every time.

Changelog

• v1.0 FROZEN (2026-08-30) — Danny’s explicit freeze confirmation; content = v0.2 verbatim. Launch version.
• v0.2 (2026-08-30) — Editorial revisions per ChatGPT review: upgraded to PROTOCOL v1.3 (citation-identity integrity, exact claim support, corrections-inherit-burden codified as standing requirements); standardized estimate-provenance record schema; drug-type/hemp definition refined (jurisdictional boundaries, CBD-dominant regulated cultivars included); landrace-provenance discipline added (with traditional-knowledge attribution); scale-category discipline added; Section D tightened to cultivation-facing market structure with explicit deferrals; “legalization killed the illicit market” rescoped to its measurable cultivation component; data-visibility wording fixed. Freeze pending Danny’s explicit confirmation.
• v0.1 (2026-08-30) — initial draft.