Research Brief — R06: Organic vs. Synthetic/Mineral Fertility in Cannabis Cultivation

Report ID: R06 (Wave 2, systems comparison; executed seventh) · Brief version: v1.0 — FROZEN September 7, 2026 on Danny’s confirmation. Both research runs execute this version and must cite it; the critique stage records it. (v0.9 drafted 2026-09-07; v0.95 same day after ChatGPT’s editorial review — six calibrations, all accepted: equivalence bases declared non-interchangeable; irrigation-water chemistry added as a comparison field; fertilizer chemistry separated from microbial/biostimulant co-treatments; sensory evidence split into four classes; genotype × fertility interaction warning; causal “stress response” wording replaced. v1.0 same day: ChatGPT second read APPROVED FOR FREEZE with no blocking changes; its one optional refinement — root-zone temperature as an inventory field — accepted.) Date drafted: September 7, 2026 · Drafted by: Claude (editor) · Protocol: v1.4 (with the §3 standing quantitative-claim fields adopted 2026-09-07 from R04)

Methodological spine (one sentence): name the input by its chemical form, rate and delivery, never by its label → fix the medium, system and environment → keep the three outcomes distinct (cannabinoid concentration, flower yield, total cannabinoid yield) → compare only on stated nutrient-supply equivalence and a stated functional unit → attribute effects to nutrient availability dynamics and specific compounds, never to “organic” or “synthetic.”

Assigned to: ChatGPT deep research + Gemini deep research (independently, same frozen brief; both runs cite the frozen version)

Assignment

Produce an evidence review of how cannabis is fed — organic/biological fertility (meals, manures, guanos, composts, hydrolysates, amino-acid and humic products, mineralization-dependent programs) versus synthetic/mineral fertility (soluble salts, chelated micronutrients, controlled-release and fertigation programs) — and what the record establishes about their effects on yield, cannabinoid and terpene chemistry, plant health, product safety and compliance, and environmental fate.

This is not a “which is better” report (PROTOCOL research voice). It is a map of what has been measured in cannabis, what is borrowed from other crops, and what is asserted — with the mechanisms that explain differences where differences exist.

R06 sits at the center of the corpus’s handoff web and must inherit, not repeat:

• R01 (organic agriculture): the practices-not-labels license (R01-C033); nitrogen-timing asynchrony as the mechanism behind organic yield gaps (R01-C008, R01-C020); the functional-unit rule — per-area vs. per-product comparisons can reverse (R01-C018, S-045); organic-approved inputs with documented costs (copper, R01-C021); organic systems as open nutrient loops with phosphorus surpluses (R01-C022).
• R08 (living soil): the one controlled indoor organic-vs-mineral comparison (R08-C006, Massuela et al. 2023: mineral ≈+21% inflorescence dry matter; organic higher CBD concentration, lower total CBD yield); the phosphorus response ceiling (R08-C008, 30 vs 59 mg/L dispute preserved); the blinded flushing trial (R08-C026); assay-dependent microbial compliance (R08-C022); the absence of economic and adoption datasets (R08-C029). R06 takes over the flushing question and the organic-timing mechanism as its own.
• R04 (environmental footprint): the only measured nutrient-loss study, in high-CBD hemp (R04-C028: tripling P tripled nothing but leachate); the outdoor LCA’s counter-intuitive fertilizer finding — more fertilizer, lower footprint per kg because fewer plants and less media (R04-C029); “organic/living-soil has a lower footprint” left open as M10.
• R03 (legal & illicit): the contamination backdrop — no market gets a safety halo (R03-C025); the myclobutanil chemistry (R03-C026). R06’s safety questions are about inputs, not markets.
• R05 (environments): nitrogen-heavy fertilization reduced cannabinoid content in the outdoor LCA’s trial (R05-C012).

Tag every load-bearing claim by evidence type (PROTOCOL §3 standing fields): measured / hybrid / modeled / observed-administrative / regulatory / absence / methodological. For R06 the dominant types will be measured trials and absences; label each.

Methodology requirements

Inputs are named by chemistry, rate and delivery, never by label (the spine). “Organic” and “synthetic” are regulatory and marketing categories, not agronomic treatments. For every fertility program or product discussed, record: nutrient forms actually delivered (nitrate, ammonium, urea, amino acids/peptides, protein meals requiring mineralization; phosphate salts vs. rock phosphate vs. bone/guano; sulfate/chloride/chelate micronutrients), rates (mg/L in solution or g/m² or kg/ha), delivery (top-dress, fertigation, pre-amended medium, foliar), and release dynamics (immediate, controlled-release, mineralization-dependent with the temperature and biology that controls it). Many “organic” liquid products contain substantial soluble mineral fractions; many “mineral” programs include organic chelators and biostimulants. Where a source does not disclose the chemistry, say so and grade the claim accordingly.

Nutrient-supply equivalence is the comparison rule. No two fertility treatments may be compared on any outcome unless the review states whether they were matched on (i) total applied nitrogen (and ideally P and K), (ii) medium and container volume, (iii) environment and genotype, (iv) duration and harvest criterion, and (v) the availability schedule (how much N was plant-available when). Equal total applied N, equal solution EC, equal label rate, and equal plant-available N are four different equivalence bases and are never interchangeable: a trial matched on one is unmatched on the others, and the review names which one was used. Trials comparing an organic program at label rate to a mineral program at label rate are comparing two rates, not two philosophies; say so. Where treatments were not matched, report the result with the mismatch named — never harmonized by the researcher’s own arithmetic unless the arithmetic is shown.

Three outcomes, kept distinct (the R08-C006 lesson). Cannabinoid concentration (% or mg/g), flower yield (g/plant, g/m², g/m²/cycle), and total cannabinoid yield (mg/plant or g/m²) move independently and frequently in opposite directions under nutrient stress. Every result names which outcome it reports. A treatment that raises cannabinoid concentration while reducing biomass is not evidence of greater cannabinoid productivity; report both numbers, and where stress is proposed as the mechanism, grade that mechanism separately rather than assuming it. Terpene outcomes get the same discipline (concentration vs. total, and which compounds).

Functional-unit discipline (R01-C018, R05-C032, R04 spine). Per-plant, per-m², per-cycle, per-kg-flower and per-g-cannabinoid results are not interchangeable. Environmental and efficiency comparisons must state both per-area and per-product framings where both exist; R04-C029 is the standing example of a per-kg ordering that reverses the per-plant intuition.

Irrigation-water chemistry is a comparison field. Source water (municipal, well, RO, rain), alkalinity, starting EC, Ca/Mg, and sodium/chloride alter pH behavior and nutrient availability enough that a reported fertility effect can be a water-treatment interaction. Every trial in the inventory records its irrigation-water chemistry where disclosed and is flagged water-undisclosed where not; comparisons across trials with different water chemistry name the difference.

Fertilizer chemistry is separated from microbial and biostimulant co-treatments. Many “organic” programs bundle inoculants, humic/fulvic acids, kelp, amino acids, or enzymes with the nutrient source; some mineral programs bundle biostimulants too. Where a treatment contains such co-inputs, an observed difference is not attributed to nutrient source alone: the review names the co-inputs, states whether the design isolated them, and otherwise grades the result as a program comparison, not a nutrient-source comparison.

Genotype × fertility interaction. A nutrient regime that performs differently in one cultivar may not generalize even within drug-type cannabis. Single-genotype trials are labeled as such wherever they are cited, and synthesis conclusions drawn from them carry the label explicitly; multi-genotype trials report whether an interaction was tested.

Mechanism before category. Where organic and mineral programs differ in outcome, the review attributes the difference to a named mechanism with its own evidence — mineralization lag vs. late-flower demand, ammonium:nitrate ratio effects, salinity/EC, pH drift, phosphorus oversupply, micronutrient chelation and availability, microbial competition for N, root-zone oxygen — and grades the mechanism separately from the outcome. Mechanisms borrowed from other crops are adjacent or general, never direct.

Practice–effect separation. Split each practice (flushing, “bloom boosting,” compost teas, amino-acid feeding, foliar programs, run-off/EC management) from its claimed effect and grade each separately. Adoption is evidence about the practice, not the effect. Unmeasured effects stay unverified — a status, not a grade of weakness.

Safety claims name the compound, the input, and the assay. Heavy metals (Cd, As, Pb, Hg, Cr) in phosphate rock, guanos, manures, biosolids-derived products, and chelated micronutrient salts; copper in organic-approved fungicides; microbial loads from manures and teas; pesticide contamination of “organic” inputs. Every safety claim names the compound, its source input, the matrix tested (input, medium, plant tissue, flower, smoke), the jurisdictional action limit, and the assay. Regulatory claims carry jurisdiction and as-of date (California DCC/CDFA, Health Canada, Colorado MED, OMRI/CDFA-OIM listing, Canada’s organic-cannabis standard, absence of USDA organic for cannabis).

Evidence type and evidence quality, graded separately (R04 standard). Type tags per §3. Quality on four axes: relevance (does it answer the question on the medium, system and outcome asked?), representativeness (genotype, scale, environment, single site?), transparency (can inputs and rates be reconstructed?), and reported uncertainty (replication, variance, statistics). Opacity disqualifies: a yield or potency figure from an undisclosed nutrient program is estimate-lore.

Provenance for quantitative estimates — 11 fields for R06. Every major quantitative result carries the standard 10-field record (geography, period, population/genotype, metric, functional unit, system/medium, boundary, method, directional bias, confidence) plus an eleventh field for this report: “input chemistry + rate + equivalence basis” — what was actually applied in what form and amount, which co-inputs accompanied it, the irrigation-water chemistry, and on which of the four bases the treatments were matched (equal total N, equal EC, equal label rate, equal plant-available N — or unmatched). The field is conditional: for a single-program study with no comparison it reads “not applicable — single treatment, chemistry: […]”. A blank eleventh field on any organic-vs-mineral comparison means the comparison is uninterpretable and is labeled equivalence-orphaned.

Jurisdiction- and date-stamping on every regulatory, listing, and compliance claim. Prefer primary sources: peer-reviewed controlled trials in cannabis and hemp; then controlled-environment horticulture (tomato, leafy greens, ornamentals) as adjacent; state and federal agency documents; utility and extension trials; then industry white papers and journalism, labeled as such.

Key questions

A. Definitions, taxonomy, and the regulatory landscape (required output category)

1. Build an input taxonomy table: for each input class used in cannabis (soluble mineral salts; chelated micronutrients; controlled-release fertilizers; protein meals; guanos; manures and composts; fish and plant hydrolysates; amino-acid and humic/fulvic products; microbial inoculants marketed as fertility; rock phosphate and other mined minerals) — nutrient forms, typical analysis, release dynamics, what mineralization requires, and what is unknown about its composition variability.
2. What does “organic” legally mean for cannabis by jurisdiction (no USDA organic; California OCal; Canada’s Cannabis Regulations and the organic standard; state programs elsewhere; OMRI and CDFA-OIM listing for inputs), and what does each standard actually constrain about nutrient inputs vs. what marketing implies?

B. Head-to-head evidence in cannabis and hemp

3. Inventory every controlled trial comparing organic/biological and mineral fertility in drug-type cannabis or hemp, as a table: genotype(s) and whether interaction was tested, system and medium, irrigation-water chemistry, root-zone temperature where reported (mineralization and root activity are temperature-sensitive), treatments with chemistry and rates, co-inputs (microbial/biostimulant) and whether isolated, equivalence basis (which of the four), duration, replication, and the three outcomes (yield, concentration, total cannabinoid yield) plus terpenes where measured. Extend R08-C006 (Massuela et al. 2023) with anything published since or missed; report whether any trial matched treatments on applied N.
4. What is the state of the nitrogen dose-response literature in cannabis (rate, form, timing) irrespective of organic/mineral framing — including N-form (ammonium vs. nitrate ratio) and the stage-specific demand curve — and how does it explain organic-timing effects (R01-C008/C020 mechanism applied to cannabis)?
5. Phosphorus and potassium: consolidate the P ceiling and dispute (R08-C008) and any K, Ca, Mg, S and micronutrient dose-response work; is there evidence that “bloom booster” ratios raise yield or cannabinoid yield at any level above the P ceiling?

C. Nutrient availability dynamics in cannabis systems

6. Mineralization in containers and beds: what is measured about N release from meals, manures and hydrolysates under indoor temperatures and irrigation regimes, and does late-flower demand outrun supply (the mechanism Massuela et al. proposed)? Are there any cannabis-specific mineralization or nutrient-budget studies, or is this entirely adjacent?
7. Root-zone chemistry: EC, pH drift, salinity and sodium/chloride accumulation under mineral programs; ammonium toxicity; microbial immobilization of N under high-carbon amendments — what is measured in cannabis?
8. Organic-N uptake: does cannabis take up amino acids or peptides directly, and does any cannabis evidence support the “amino-acid feeding” product class beyond general plant-science literature?

D. Quality, sensory, and the flush question

Sensory evidence comes in four classes that never collapse into “tastes better”: (a) objective chemistry (terpene/cannabinoid/mineral assays), (b) trained-panel sensory testing, (c) consumer-preference testing (blinded or not — say which), and (d) grower or consumer anecdote. Every quality claim names its class; a class-(a) difference does not establish a class-(b) or (c) outcome, and class (d) is anecdotal by definition. 9. Terpene and minor-cannabinoid responses to fertility regime: what is measured (which compounds, concentration vs. total), and does any study link nutrient program to a sensory outcome — and in which of the four classes? 10. Flushing. Consolidate R08-C026 (the single blinded industry trial) with any peer-reviewed or additional trials: does pre-harvest nutrient withdrawal change flower mineral content, combustion properties, ash color, or sensory preference? Separately: what does flushing do as a discharge event (R04’s flush-as-discharge framing — nutrient solution to drain)? 11. “Smoothness,” “harshness,” and ash color as quality proxies: is there any measured link between flower mineral or nitrate content and smoke chemistry, or is this entirely assertion?

E. Safety, contamination, and compliance

12. Heavy metals: which inputs (phosphate rock, guanos, manures, biosolids products, some chelates and micronutrient salts, some organic amendments) carry Cd, As, Pb, Hg, Cr at levels that have been measured, and is there any cannabis study tracing input metal content to flower residue and compliance failure? Contrast with R03-C025’s finding of higher chromium in legal Canadian samples.
13. Copper: organic-approved copper fungicides (R01-C021) on a combusted crop — residue limits by jurisdiction, and any cannabis measurement.
14. Microbial safety of manures, composts and compost teas in cannabis (R08-C022 assay dependence): pathogen (E. coli, Salmonella, Aspergillus) evidence in cannabis or adjacent crops, and how TYMC vs. qPCR regimes interact with biological fertility.
15. Nitrate and residue: does nutrient program affect nitrate, nitrite, or other combustion-relevant residues in flower — measured anywhere?

F. Environmental fate and footprint of the fertility choice

16. Leaching and runoff by input class: extend R04-C028 (measured P leachate, hemp) with any N or K loss data; drain-to-waste vs. recirculating vs. bed systems; what fraction of applied N leaves the root zone under each program — measured or modeled?
17. Embodied impacts of inputs: mineral N (Haber-Bosch) and P (mining) vs. guano mining and transport, manure and compost production, fish hydrolysate — on compatible boundaries and units (R04 discipline); nitrous-oxide emissions from organic vs. mineral N in containers/beds, if any cannabis or CEA evidence exists.
18. The functional-unit reversal (R04-C029): does any evidence beyond S-003 bear on whether fertilizer reduction lowers or raises per-product footprint in cannabis?

A practitioner-record passage is withheld from the public copy.

H. Synthesis targets

21. The input-chemistry map: every load-bearing result placed on its input chemistry, rate, equivalence basis, medium, and functional unit — the artifact that makes organic-vs-mineral results non-poolable by construction.
22. The three-outcome map (concentration / yield / total cannabinoid yield) by treatment class, with named gaps. Absence of measurement is a finding: do not build a verdict on “organic vs. synthetic quality” because readers expect one.
23. The mechanism table: each documented difference → named mechanism → evidence grade for the mechanism in cannabis vs. adjacent crops.
24. The 8–10 most repeated public claims and what the evidence says (M1–M10 below).

Pre-registered public claims to test

Word all verdicts so absence of evidence remains absence of evidence (“unsupported” ≠ “the opposite is true”; unknown is acceptable):

• M1. “Organically grown cannabis tastes better / smokes smoother” — which of the four sensory classes has evidence, and which compounds would carry it?
• M2. “Synthetic nutrients leave residues that make harsh smoke; white ash proves a clean flush” — measured link between mineral/nitrate content and combustion or ash? (Ties to M6.)
• M3. “Organic yields less” — R08-C006 is one controlled study; is it replicated, and was it an equal-N comparison?
• M4. “A plant can’t tell the difference — an ion is an ion” — the Liebig framing; test against organic-N uptake evidence and mineralization-timing effects. Both the strong form and its denial need evidence.
• M5. “You can’t overfeed or burn plants with organics” — salinity, ammonium, and P-excess evidence in organic programs.
• M6. “Flushing pulls nutrients out of the flower” — R08-C026 plus anything newer; distinguish plant-tissue change from medium change from discharge.
• M7. “Organic is safer — no heavy metals, no chemicals” — input-metal and microbial evidence; and the inverse “mineral is cleaner because it’s controlled.”
• M8. “Synthetic fertilizers kill the soil microbiome” — what is measured in container media and beds vs. broadacre soils; distinguish salt effects from N-form effects from the absence of carbon inputs.
• M9. “High-P bloom boosters increase yield and resin” — R08-C008 ceiling; any evidence above it.
• M10. “Organic/living-soil cannabis has a lower environmental footprint” — inherited from R04 M10 and R01-C018: per-area vs. per-product, and the R04-C029 reversal.

A practitioner-record passage is withheld from the public copy.

Standing source & citation requirements (PROTOCOL v1.4 — verbatim)

• 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. If you cannot fully resolve the tuple, flag it explicitly as a partial citation rather than completing it from memory.
• Exact claim support: a source supporting the general direction of a finding does not automatically support its stated magnitude, population, geography, time period, boundary, or outcome measure — 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.
• Source-identity deduplication and corroboration independence (rules 11–12): one paper is one source however many times it is cited; “multiple studies agree” requires demonstrably distinct studies or datasets.
• When testing public claims, “the claim is unsupported” and “the opposite is true” are different findings — do not overcorrect; unknown is an acceptable verdict.
• Standing quantitative-claim fields (§3, adopted 2026-09-07): every load-bearing claim carries an evidence-type tag (measured / hybrid / modeled / observed-administrative / regulatory / absence / methodological), and every quantitative claim carries an inline boundary line — system/medium · functional unit · outcome measure and method · population/genotype — with “N/A — reason” where a field does not apply. For R06, the boundary line also names the input chemistry, rate, and equivalence basis.
• Report ranges and disagreements between sources, not averaged-away conclusions; list data gaps explicitly.

Candidate claims for editorial review

Conclude with 15–30 concise, testable candidate claims with supporting citation(s), each carrying its system/medium/genotype/jurisdiction/period tags and, for magnitudes, the outcome measure, functional unit, input chemistry and equivalence basis, and evidence-type tag. Do not assign final evidence status or confidence; the editor will.

Search-stopping rationale

End with an explicit, bounded statement of why searching concluded: the search cutoff date, the databases and query strings covered, the inclusion criteria applied (language, publication types, date range, crop and system scope), saturation observed, and named question areas where no eligible studies were located within the documented search — distinguishing “not searched” from “searched and none located.” Do not write that the literature is “genuinely absent”; absence is a property of the search, not of the world.

What “done” looks like

A structured report answering A–H with cited, chemistry- and outcome-tagged evidence; the input taxonomy (A1) and the controlled-trial inventory (B3) as distinct outputs; the input-chemistry map (H21), three-outcome map (H22), and mechanism table (H23); calibrated verdicts on M1–M10; a top-10–15 strongest-sources list with one-line reliability and commercial-position notes; an explicit data-gap list; candidate claims; and the bounded search-stopping rationale. Uncertainty findings are as valuable as positive findings.

Editor’s note (repo workflow): outputs land in raw-inputs/chatgpt-YYYY-MM-DD.md and raw-inputs/gemini-YYYY-MM-DD.md; then verification & critique (critique.md), claim extraction (ledger.md, IDs R06-C001+; next source number S-198), synthesis, and content extraction per PROTOCOL v1.4. Archive audit runs before external research. Corpus anchors: R01-C008/C020 (N asynchrony), R01-C018 (functional-unit reversal), R01-C021 (copper), R01-C022 (open loops, P surplus), R01-C033 (practices-not-labels — the license); R08-C005 (no cannabis AMF-suppression threshold), R08-C006 (the one controlled comparison — the keystone R06 extends), R08-C008 (P ceiling and dispute), R08-C022 (assay-dependent compliance), R08-C026 (blinded flush trial), R08-C029 (no economic dataset); R04-C028 (measured P leachate), R04-C029 (fertilizer reduction raised per-kg footprint), R04-C030 (media as carbon term), R04 M10 (open); R05-C012 (N-heavy fertilization reduced cannabinoid content); R03-C025 (no safety halo), R03-C026 (myclobutanil); S-045 (per-area vs. per-mass). Audit backlog reminder: R03 post-signoff audit remains queued (AUDIT_LOG); at most one backlog audit runs alongside this round.