Claim Ledger — R05: Indoor vs. Greenhouse/Light-Dep vs. Sun-Grown Cannabis

Report: R05 · Ledger version: v1.0 — SIGNED OFF by Danny, August 28, 2026 · Last updated: August 28, 2026 Grading reference: PROTOCOL.md §3 · Sources cite S-nnn IDs from SOURCE_LIBRARY.md Basis: critique.md §2 convergences + §7 verified sources. Gemini-only unsourced figures excluded pending its §8 follow-up answers.

Claim IDs are permanent. Evidence status ≠ confidence: confidence is in the claim as graded.

Energy & carbon

R05-C001 — Cradle-to-farm-gate emissions for US indoor warehouse cannabis range from ~2,283 to ~5,184 kg CO2e per kg of dried flower, driven primarily by location (grid mix and climate).

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-001 · Researcher agreement: both (identical figures)
• Notes: Modeled LCA, not facility census. Functional unit: kg CO2e/kg dried flower. Low end = hydro-powered mild climates (WA); high end = fossil-heavy grids with high HVAC burden. The single strongest number in R05.

R05-C002 — Fully indoor cultivation carries a substantially higher operational energy and carbon burden per kg of dried flower than outdoor sun-grown production; the magnitude is climate-, grid-, and boundary-dependent.

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-001, S-003, S-004, S-005 · Researcher agreement: both
• Notes: Direction unusually consistent across independent LCAs (US model, Canadian model, outdoor field LCA, 2025 industry analysis). Never quote a single ratio without stating location and boundary.

R05-C003 — Outdoor cultivation-stage carbon reductions relative to indoor are on the order of ~90% in the Canadian comparative model, with some analyses estimating substantially larger factors for specific systems.

• Evidence status: probable · Specificity: direct · Confidence: high
• Sources: S-003, S-004 · Researcher agreement: both (different magnitudes)
• Notes: Modeled. Gemini’s “90–96%” range is consistent but partially rested on a confabulated source; the claim stands on S-003/S-004 alone.

R05-C004 — Modern LED fixtures cut lighting electricity substantially (~40–50% vs. single-ended HPS at equivalent PPFD) but do not close the indoor-vs-outdoor carbon gap, because HVAC and dehumidification loads persist.

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-001, S-005, S-007, S-025 · Researcher agreement: both
• Notes: PG&E field data: HVAC ≈45% of facility energy; flower rooms ≈86%. Verified fixture efficacy (S-025, measured): 1.7 μmol/J (HPS) to 2.5 μmol/J (white+red LED). Gemini’s 2.7–3.2 μmol/J remains unsourced — its replacement citation (“Kusuma/Westmoreland/Bugbee 2020 PLOS ONE e0230755”) also failed verification; real anchors are S-025 and Kusuma, Pattison & Bugbee 2020 (Horticulture Research).

R05-C005 — Replacing HPS with LED can increase winter space-heating demand in cold-climate indoor facilities because canopy-directed infrared heat is removed.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-007 (thermodynamic coupling documented); researcher convergence · Researcher agreement: both
• Notes: Mechanistically solid; magnitude poorly quantified in verified sources.

R05-C006 — Which energy end-use dominates a controlled cannabis facility depends on climate: cooling/dehumidification in hot-humid regions, heating in cold regions.

• Evidence status: probable · Specificity: adjacent · Confidence: high
• Sources: S-021, S-004 · Researcher agreement: chatgpt-only (Gemini implies)

R05-C007 — In sealed indoor facilities, a large majority (~70% measured in one field study) of applied irrigation water is transpired and becomes recoverable HVAC condensate.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-007 · Researcher agreement: both (Gemini’s 80–95% recapture figure unsourced; ledger takes the measured 70%)
• Notes: n=2 facilities. Estimated statewide reuse potential 17.8M gal/yr (CA).

R05-C008 — “Greenhouse” is a family of systems, not one category: passive hoop house, light-deprivation, supplemental-light, and sealed high-tech greenhouses have materially different energy and resource profiles.

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-001, S-013 (regulatory mixed-light tiers), S-010 · Researcher agreement: both
• Notes: CA’s own licensing distinguishes mixed-light tiers by wattage (0–6, >6–25 W/ft²). A supplemental-light greenhouse trends toward indoor’s profile; a passive hoop trends toward outdoor’s.

R05-C009 — Quantified, subtype-specific greenhouse carbon footprints (per kg flower) are not established in peer-reviewed literature; published LCAs collapse greenhouse types into one category.

• Evidence status: unknown · Specificity: direct · Confidence: high
• Sources: gap documented by both reports · Researcher agreement: both — Gemini formally retracted its subtype ranges as “synthetic extrapolations” (2026-08-25 follow-up) and concurs with this grading
• Notes: Honesty marker. Agreed companion claim: subtype footprints plausibly form a monotonic ladder between unheated passive solar and sealed CEA (probable/adjacent/medium, modeled) — usable in synthesis with that framing, never with specific numbers.

R05-C010 — US industry-wide cannabis emissions were modeled at ~44 Mt CO2e/yr (~595 PJ/yr energy), with a modeled shift to outdoor cultivation reducing emissions by up to ~76%.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-005 · Researcher agreement: chatgpt-only
• Notes: Scenario analysis including illicit production estimates; use as scale illustration, not census. Author is Evan Mills — same author as the 2012 HPS-era baseline (S-006); useful narrative continuity.

Environmental footprint beyond carbon

R05-C011 — A low carbon footprint does not make outdoor cannabis automatically sustainable: water-diversion timing, nutrient losses, fertilizer, potting media, erosion, and habitat impacts can be substantial and site-specific.

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-003, S-008, S-009 · Researcher agreement: both

R05-C012 — In an outdoor cannabis LCA, potting media accounted for up to ~75% of global warming potential and nearly half of fossil-fuel depletion; nitrogen-heavy fertilization reduced cannabinoid content.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-003 · Researcher agreement: neither emphasized (editor’s verification finding)
• Notes: Single study, container-based system. High-value for R06/R17 (media reuse, living soil) and for myth-busting “outdoor is automatically green.”

R05-C013 — The ecological harm of outdoor cannabis water use concentrates in timing: peak irrigation demand coincides with seasonal low-flow in Mediterranean-climate watersheds, threatening salmonid habitat. In Bauer et al.’s four northwestern California study watersheds, estimated (modeled) cannabis water demand ranged from 2–23% of annual seven-day low flow in the least-impacted watershed (Upper Redwood Creek), but reached 34–165% and 36–173% of low flow in two others (Redwood Creek South, Salmon Creek); in Outlet Creek, demand exceeded 100% of the minimum seven-day low flow in years when measured flow approached zero.

• Evidence status: established · Specificity: direct · Confidence: high · Tag: modeled (mapped plants × 22.7 L/plant/day assumption, vs. gauged/modeled flow)
• Sources: S-008, S-009 · Researcher agreement: both
• Notes: Regional (NW California) evidence; off-season storage is the standard mitigation. Demand exceeded low-season streamflow in three of the four watersheds.
• CORRECTED 2026-09-06 (PROTOCOL rules 10 & 13): Original wording: “diversions could reach up to ~23% of seven-day low flow in studied watersheds.” Error: the 23% figure is the ceiling of the least-impacted watershed only, presented as the study’s ceiling — a scope/unsupported-extrapolation calibration error introduced at R05 editing (pilot cycle, PROTOCOL v1.2). Basis: Bauer et al. 2015 Results and abstract (“an estimated flow reduction of up to 23% of the annual seven-day low flow in the least impacted of the study watersheds. Estimates from the other study watersheds indicate that water demand for marijuana cultivation exceeds streamflow during the low-flow period”). Detected: Claude, R04 verification (verify-B B3), 2026-09-06 — proposed, not applied. Independent second check: ChatGPT, 2026-09-06, retrieved the paper independently (PMC4364984; Table 4 and Results) and returned CONFIRMED — PATCH AUTHORIZED with the wording adopted above. Record: verification/R05-C013-correction-2026-09-06.md; AUDIT_LOG R05 row.

R05-C014 — Reliable measured per-plant water figures for cannabis are lacking: the widely cited 22.7 L/plant/day is an assumption from a 2010 Humboldt County ordinance, drawn from sources ranging 3.8–56.8 L/plant/day.

• Evidence status: established (about the state of the literature) · Specificity: direct · Confidence: high
• Sources: S-008 · Researcher agreement: conflict, partially resolved — Gemini’s follow-up asserts its 5.5–12.0 L/plant/day figure is Dillis et al.’s metered measurement; this pass confirmed the Dillis papers exist (S-009 and the 2023 permitted/unpermitted demand study) but could NOT confirm the specific per-plant rates from abstracts. Verify against Dillis full text at synthesis stage before upgrading.
• Notes: Any per-plant or per-kg water number in downstream content must carry this caveat.

R05-C015 — Supplemental greenhouse lighting can increase total crop evapotranspiration while improving water-use efficiency per gram of flower (~37% reduction in ET per gram at higher light in one controlled trial).

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-010 · Researcher agreement: chatgpt-only
• Notes: One cultivar (‘Suver Haze’), one facility. Teaching point: absolute use and efficiency can move opposite directions.

R05-C016 — Unshielded greenhouse supplemental lighting creates nighttime light pollution; California regulation requires cultivation lighting to be shielded from sunset to sunrise.

• Evidence status: probable (ecological impact) / established (regulatory requirement) · Specificity: general / direct · Confidence: medium
• Sources: S-013 · Researcher agreement: both
• Notes: Cannabis-specific quantitative ecological studies are limited.

R05-C017 — Documented pesticide and watershed contamination from cannabis cultivation is concentrated in illicit/trespass operations; these findings must not be generalized to licensed production.

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-019, S-020 · Researcher agreement: both

Product quality

R05-C018 — Cultivation environment can materially alter cannabis secondary-metabolite profiles even in genetically identical clones.

• Evidence status: probable · Specificity: direct · Confidence: high
• Sources: S-002 · Researcher agreement: both
• Notes: The directional claim is safe; specific universal rankings are not (C019–C021).

R05-C019 — In the one matched-clone commercial comparison available, outdoor samples showed greater terpene abundance and diversity (notably sesquiterpenes: β-caryophyllene, α-humulene, α-bergamotene, α-guaiene, germacrene B), unique C4/C6 cannabinoid homologues, and fewer oxidized cannabinoids than indoor samples; primary cannabinoid levels were broadly comparable.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-002 · Researcher agreement: both (Gemini’s author attribution corrected — Zandkarimi et al., critique §7)
• Notes: Two cultivars, commercial (non-standardized) conditions, media confounded with environment (living soil outdoor vs. artificial media indoor). Supports “environment matters,” not “outdoor is always better.”

R05-C020 — The claim that indoor cannabis is inherently higher in THC than sun-grown is not established by direct evidence independent of genetics, management, harvest timing, and lab variation.

• Evidence status: unknown · Specificity: direct · Confidence: high
• Sources: S-002, S-022 · Researcher agreement: both
• Notes: Retires the strong version of Danny’s 2016 “full control = medical quality” teaching (archive claim #1). Market premium ≠ chemistry (C022).

R05-C021 — The claim that sun-grown cannabis universally has richer terpene profiles is not established; one small confounded study supports the direction.

• Evidence status: unknown · Specificity: direct · Confidence: high
• Sources: S-002 · Researcher agreement: conflict (Gemini “True”; ChatGPT “plausible, not universal” — editor sides with ChatGPT, critique §3.5)

R05-C022 — Indoor-labeled flower commands a persistent market price premium in many jurisdictions; this demonstrates market perception and grading conventions, not biochemical superiority.

• Evidence status: traditional-practice (market convention) · Specificity: direct · Confidence: high (that the premium exists)
• Sources: S-015, S-016 (cost/price context); Gemini’s 40–80% figure unsourced · Researcher agreement: both
• Notes: Perception drivers: bag appeal, density, THC-number shopping, historic scarcity. Physical/commercial-quality research (blinded grading, matched genetics) is essentially absent — flagged data gap.

R05-C023 — No cultivation environment is inherently “cleaner”: indoor excludes some outdoor pathways (drift, ambient spores, wildlife) but has its own (humidity-pocket Botrytis/PM, recirculating-system pathogen persistence, pesticide use documented in illicit indoor grows); no robust cross-environment contamination-rate dataset exists that controls for differing state testing rules.

• Evidence status: unknown (for any universal ranking) · Specificity: direct · Confidence: high
• Sources: S-017, S-018, S-023 · Researcher agreement: conflict on outdoor pass rates (both unsourced); agreement on “no ranking”
• Notes: Belgian illicit indoor study: pesticides in 64.3% of plant samples (S-017).

R05-C024 — Cannabis readily accumulates heavy metals; contamination risk tracks growing media, soil history, fertilizers, and water — not the cultivation environment per se.

• Evidence status: probable · Specificity: direct · Confidence: high
• Sources: S-002 context, general phytoremediation literature; 2026 reclaimed-water study (unverified, cited by ChatGPT) · Researcher agreement: both
• Notes: Outdoor in-ground on legacy-agriculture soils = the characteristic risk case; inert media largely avoid it.

Yield, economics & risk

R05-C025 — Light deprivation induces flowering by artificially extending the uninterrupted dark period while natural days are still long, decoupling flowering from season and enabling multiple annual harvests; the achievable count depends on cultivar, climate, propagation schedule, crop duration, and facility controls.

• Evidence status: established (mechanism) · Specificity: direct · Confidence: high
• Sources: S-013 (regulatory definition), photoperiodism literature, both reports · Researcher agreement: both
• Notes: Supersedes archive claim #10 (“up to 5×”) — the mechanism survives; the fixed multiplier doesn’t. Typical commercial practice: 2–4 cycles/season. Characteristic failure mode: light leak during dark period → hermaphroditism/reversion.

R05-C026 — Yield comparisons across cultivation environments are routinely misleading: results depend on plant density, cultivar, cycle length, training, light level, and above all the functional unit (per plant / per m² / per cycle / per year / total vs. saleable flower).

• Evidence status: established · Specificity: direct · Confidence: high
• Sources: S-012 · Researcher agreement: both
• Notes: Outdoor maximizes per-plant size; indoor maximizes annual per-area via density + cycles. Neither number transfers.

R05-C027 — Industry surveys consistently report the same cost ordering — outdoor cheapest, greenhouse intermediate, indoor most expensive per pound (2021 medians ≈ $214 / $256 / $472/lb; 2018 ≈ $305 / $490 / $635/lb) — but these are dated, jurisdiction-sensitive market observations, not biological constants.

• Evidence status: probable · Specificity: direct (industry survey data) · Confidence: high (for the ordering) / low (for the specific figures today)
• Sources: S-015, S-016 · Researcher agreement: both
• Notes: Cost dominated by energy prices, labor, taxes, licensing, testing, security — separate regulatory economics from cultivation biology (brief D11).

R05-C028 — Cannabis profitability can flip sign with yield, price, labor, and planting-material assumptions within the same facility: a 2026 New York greenhouse analysis estimated −$1.48/ft²/yr for an autoflower strategy vs. +$7.18/ft²/yr for photoperiod at the same wholesale price.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-011 · Researcher agreement: chatgpt-only
• Notes: More crop cycles ≠ more profit. Highly assumption-sensitive; cite the journal version.

R05-C029 — High-OpEx indoor production is disproportionately exposed to wholesale price compression, while low-cost outdoor/light-dep operations remain viable at prices that render indoor unprofitable.

• Evidence status: probable · Specificity: direct · Confidence: medium
• Sources: S-015, S-016, S-011 (cost-structure basis); Gemini’s specific $/lb thresholds unsourced · Researcher agreement: both

R05-C030 — Each cultivation system has a characteristic failure-mode profile: outdoor risks weather, autumn rain/frost, wildfire smoke, and pests; light-dep risks blackout-mechanism failure and light leaks; indoor risks power/HVAC failure with rapid catastrophic loss.

• Evidence status: traditional-practice · Specificity: direct · Confidence: high (that these are the recognized failure modes)
• Sources: both reports, practitioner consensus; smoke-taint mechanism is a viticulture transfer (Kennison/Parker) · Researcher agreement: both
• Notes: Little formal comparative crop-loss data — the profile is practitioner knowledge, honestly labeled.

Landscape & methodology

R05-C031 — No defensible current breakdown of North American cannabis production volume by cultivation environment exists; available data (license counts, licensed area) cannot be converted to production shares.

• Evidence status: unknown · Specificity: direct · Confidence: high
• Sources: S-013, S-014 (what exists); gap per both reports · Researcher agreement: conflict RESOLVED — Gemini’s follow-up (2026-08-25) concedes Metrc reports license counts and canopy area only, and concurs with unknown status
• Notes: Health Canada: licensed indoor 1.41M m² vs. outdoor 6.89M m² — area ≠ output. Leaves archive claim #3 (“most industrial outdoor uses light-dep”) unresolved.

R05-C032 — Comparative cannabis sustainability and production results are sensitive to the functional unit; per plant, per m², per cycle, per year, and per kg saleable flower can produce different rankings, and figures must not be pooled across units.

• Evidence status: established · Specificity: general · Confidence: high
• Sources: S-001, S-012, methodology sections of both reports · Researcher agreement: both

R05-C033 — Published cannabis carbon figures use different lifecycle boundaries (operational-only vs. embodied-inclusive; cradle-to-gate vs. full system) and cannot be averaged into a single “cannabis carbon footprint.”

• Evidence status: established · Specificity: general · Confidence: high
• Sources: S-001, S-005, S-006 vs. modern work · Researcher agreement: both
• Notes: Mills 2012 (≈4,600 kg CO2e/kg, HPS-era assumptions) is the canonical example of a stale figure still quoted as current.

Excluded pending sourcing (tracked, not ledgered)

• Gemini’s production-share percentages (Metrc) → C031 stays unknown
• Gemini’s greenhouse-subtype CO2e ranges → C009 stays unknown
• Gemini’s per-plant/per-kg water figures → C014 governs
• RII benchmark efficiency figures (0.9→1.6–2.1 g/kWh) — plausible; obtain actual RII report
• Gemini’s 2.7–3.2 μmol/J LED efficacy — superseded by verified 1.7–2.5 μmol/J (S-025); its replacement citation also failed verification (second-round confabulation, logged for postmortem)
• Gemini’s Dillis-attributed 5.5–12.0 L/plant/day — papers real, figure unconfirmed; full-text check at synthesis
• LeafLink/Cannabis Benchmarks 40–80% premium quantification

Gemini follow-up resolutions (received 2026-08-25, after v0.1 draft)

Gemini formally: confirmed Zandkarimi authorship; fully retracted “Mettu et al. 2023”; retracted subtype carbon ranges as synthetic extrapolations; concurred with unknown for production share; supplied replacement lighting citations (one of which — “Kusuma/Westmoreland/Bugbee 2020 PLOS ONE e0230755” — again failed verification; Rodriguez-Morrison/Llewellyn/Zheng 2021 FPS is a real paper family, title imprecise). All resolutions incorporated above.

Changelog

• v1.0 (2026-08-28) — Danny’s sign-off. 33 claims canonical; downstream products may now cite R05-Cnnn IDs.
• v0.1 (2026-08-25) — 33 claims drafted from critique §2 convergences + §7 verified sources; Gemini follow-up resolutions incorporated.