Living Soil & Indoor No-Till: What the Evidence Actually Supports

R08 Verified Synthesis

Version: v1.0 — signed off by Danny August 31, 2026 (after ChatGPT’s editorial pass; five calibration micro-edits applied, see changelog) · Date: August 31, 2026 Basis: R08 Claim Ledger v1.0 (R08-C001–C031, signed off 2026-08-31) · Brief v1.0 (frozen) · Sources S-094–S-135 Corpus anchors: R01-C033 (practices, not labels) · R01-C034/C035 (AMF; infiltration/no-till null) · R05-C012/C029/C030 · R02-C025

The question, properly separated

“Does living soil work indoors?” is three questions wearing one coat, and the R08 evidence answers them very differently:

1. Is the biology real? Largely yes — and with direct cannabis evidence.
2. Are the production outcomes demonstrated? Component by component, partially; as an integrated system, not yet demonstrated.
3. Is it commercially competitive indoors? Unknown — not “no,” unknown, with real operations proving existence and no data proving competitiveness.

R08’s keystone finding is the gap between the first answer and the last two: no controlled, replicated, multi-cycle study was identified that compares a persistent indoor no-till cannabis bed against conventional substrate systems on yield, chemistry, sanitation, labor, or economics (C030). Most comparative claims about the performance of indoor no-till as an integrated system — for or against — currently rest on component evidence, adjacent-system evidence, practitioner experience, or extrapolation rather than controlled multi-cycle comparison. The biology is much better demonstrated than the production system.

That’s not a deflating conclusion. It is the map of exactly where the evidence stands, which practices carry it, and what a grower or business can and cannot responsibly claim.

First, words: what system are we even discussing?

Following R01-C033, R08 refused to evaluate labels. Four physically different systems trade under “living soil / no-till” (brief taxonomy):

• (a) Persistent beds — soil stays in place across cycles, roots left to decompose, surface top-dressing only. The strong claim.
• (b) Large minimal-disturbance containers — kept across cycles, top-dressed, some root-zone work.
• (c) Recycled/re-amended media — dumped, mixed, re-amended, “cooked,” reused. This is media reuse, not no-till: mixing shears fungal networks and resets succession.
• (d) Biologically active bottle-fed hybrids — living amendments plus liquid feeding.

Nearly all the direct cannabis evidence lives in (d) and container analogues of (b). Almost none tests (a). Any claim must name its system — and its cycle count: no-till at cycle 2 and no-till at cycle 10 are different regimes, and the evidence rarely gets past the low single digits (C015, C030).

What is well demonstrated

Cannabis soil biology is real, measured, and cannabis-direct. Cannabis forms functional AMF associations under controlled cultivation, with measured effects on growth and cannabinoid outcomes (C001). Those effects are inoculum- and genotype-dependent — in the best consortium trial, three of five cultivars gained floral biomass and rankings flipped between cultivars (C002). And crucially, colonized roots and full-strength mineral feeding coexisted in the same experiment (C003): a biologically active rhizosphere and soluble nutrition are not opposing teams. The organic-vs-salts binary is not how the root zone works.

“Synthetic nutrients kill the soil” fails on the record. Across 107 long-term datasets, mineral fertilization increased average microbial biomass carbon by ~15% — with soil pH steering the direction (C004). Fertilizer regimes genuinely alter community composition; they do not generically sterilize anything. The calibrated verb is “changes,” never “kills” (myth M5).

Organic nutrition can finish an indoor cannabis crop — with a measured yield tradeoff in the controlled comparison R08 located. In the one controlled indoor comparison, organic-fed plants reached harvest but produced ~21% less inflorescence mass than mineral-fed plants; they carried higher CBD concentration yet lower total CBD per plant, and the authors traced the deficit to nutrient acquisition faltering late in flowering (C006). Rate–response work in coir puts organic fertilizer optima around 389 mg/L N vegetative and 212–261 mg/L N in flower, substrate-dependent (C007). So the honest teaching line is double-edged: biology can feed the plant; biological timing can also become the limit exactly when demand peaks. Concentration is not yield, and neither side of the fertility debate gets to quote only its half of the Massuela result.

Excess phosphorus is wasted or worse. Cannabis floral yield saturates somewhere between ~30 and ~59 mg/L P depending on the study — a real disagreement R08 preserves rather than averages — and pushing root-zone P to 90 mg/L just loads tissue without yield or cannabinoid benefit (C008). The reviewed cannabis evidence does not support pushing phosphorus substantially above an adequate range simply to increase flower yield or cannabinoid output — organic or synthetic. Whether high P suppresses AMF in cannabis specifically, at what threshold, remains unmeasured (C005) — the general-agriculture mechanism is probable, the cannabis number does not exist.

Media reuse works — as engineering, not magic. The adjacent-CEA record is genuinely encouraging: peat reused through three commercial strawberry crops with yield equal to virgin substrate and no measured physical deterioration; coir reused for lettuce without yield loss; reused peat in lilies shifting physically without hurting production (C013). But every successful case was actively managed — nutrients re-budgeted because reused media arrive pre-loaded, fertigation endings tuned to control carryover, and pathogens handled with steam (C014). Reuse is a system you operate, not a property soil acquires. And steam — the workhorse sanitation tool of reuse horticulture — is definitionally unavailable to a living bed you refuse to disturb.

What is genuinely two-sided

The persistent root zone is a reservoir — of everything. Continuity preserves fungal networks and, where documented, antagonist-specific disease suppression; continuity also preserves pathogen propagules, pest refugia, and accumulated nutrients (C016). Root pathogens are emphatically not a soil-specific problem — Pythium, Globisporangium, and Fusarium were recovered from coco coir and rockwool at commercial marijuana facilities (C017) — so “inert media = clean media” fails alongside “living media = dirty media.” Which way the reservoir nets out for cannabis across many cycles is exactly the measurement nobody has published (C030).

The sharpest new finding is about what you can’t clean. Hop latent viroid — 12–42% losses in floral biomass and secondary metabolites where it takes hold, transmissible by seed (58–80% in one study) and by tools (C018) — turns out to be chemically unremediable in infected tissue: in full-text verified testing, no agent tried (bleach at 5–20%, hypochlorous acid, Virkon, Zerotol) degraded viroid RNA inside roots or leaves; the effective treatments are matrix-dependent surface protocols — 10–20% bleach for sap-contaminated tools, less in dilute aqueous settings (C019). For a persistent bed that keeps old root systems on purpose, this is the failure mode that forces the reset conversation: an infected bed’s root fragments could not be chemically remediated in place with anything tested. Sanitation-by-reset is the one tool system (a) gave away by design.

Compliance depends on the measuring stick. Non-selective total-yeast-and-mold plating counts the benign and beneficial organisms a living system deliberately cultivates; pathogen-specific qPCR regimes do not. The same flower can pass in a qPCR jurisdiction and fail under a 10⁴ CFU/g TYMC cap (C022). How often living-soil flower actually fails, nobody has published — the circulating failure-rate percentages were withdrawn as unsourced this cycle.

What is not demonstrated (and what that means)

Quality. No verified evidence attributes superior potency, terpene character, sensory quality, or consumer preference to living-soil practice as a system (C025). Fertility and microbes can shift concentrations and ratios within genetic architecture (C008, C009); attributing “better weed” to the bundled practice package is marketing until a controlled comparison says otherwise. This cuts against living-soil boosters and against nobody else — the salt-fed side has no symmetric quality evidence either.

Economics. No defensible dataset was identified for adoption share, for a causal living-soil price premium, or for a full-denominator cost comparison (C029). Commercial existence is settled — certified organic living-soil producers operate at licensed scale (C028) — but existence is not competitiveness, and under the wholesale price compression documented in R02-C025/R05-C029, the differentiation case that would justify a less throughput-efficient system remains a hypothesis awaiting market data.

Sustainability, honestly bounded. Media reuse is a real carbon lever outdoors, where potting media contributed up to ~75% of global-warming potential in the one outdoor field LCA identified (Desaulniers Brousseau et al. 2024). Indoors, electricity/HVAC dominate so strongly (2,283–5,184 kg CO₂e/kg) that media are a comparatively minor share of total climate burden — and by modeled inference, a ~2% yield penalty would erase the entire media-reuse saving (C027). “Reusing soil = greener grow” is true in some systems and immaterial in the one this report is about, unless yield holds.

Operations. Bed volumes, irrigation set-points, reset triggers, drift trajectories, cover-crop HVAC penalties: all of it currently exists as practitioner rules-of-thumb and modeled inference; no cannabis-CEA measurements were identified for these parameters (C031). R08 preserves that lore as lore — a hypothesis inventory for the learning library and for anyone who eventually runs the missing experiments — without laundering it into findings.

The myths, calibrated

M10 deserves its paragraph, because it is the cycle’s best demonstration of why calibration rules exist. The archive taught “30%+ compost in soil mass.” The attempt to refute it cited a paper that — on verification — reported optimal growth at 50–100% biosolids compost in its own ornamental context. So neither “30% is right” nor “30% is wrong” survives; what survives is that no universal compost percentage is defensible, formulation is evaluated by measured properties (porosity, aeration, EC, buffering) against a crop and a feedstock, and a percentage without its denominator (mass? volume? at what moisture?) is not even a specification (C011). That is better science than either original position — and it rewrites a line Danny’s handouts carried for eight years.

What this means for the teaching corpus

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

For Grow Your 6: Ch 4’s soil-building chapter inherits C010/C011 (properties, not recipes — and the 30% line gets corrected in print); Ch 6 inherits C006–C008 and C024 (feeding and pH, with the buffering claim calibrated); Ch 9 inherits C016–C021 (prevention is real and specific, not automatic); Ch 11 — the no-till reset chapter — inherits the whole failure-mode framework, and its title question gets the honest answer: the question is no longer “reuse or replace?” as a doctrine. Reuse while the monitored biological, chemical, physical, and phytosanitary state remains acceptable; reset when the evidence says the system is no longer recoverable or fit for purpose. (“You never throw out living soil until the soil tells you to” survives as the teaching hook — the monitoring list is what “telling” means.)

What would change these conclusions

R08’s gaps are a research agenda in order of leverage: (1) the missing experiment — replicated multi-cycle beds vs. conventional substrate under one roof, measuring yield/area/time, chemistry, pathogens, labor, and cost (C030); (2) longitudinal bed monitoring — EC, nutrient, physical-property, and microbial trajectories across 5–10 real cycles, even observationally (C015, C031); (3) cannabis AMF-suppression thresholds under realistic fertility (C005); (4) living-soil compliance outcomes by assay type (C022); (5) any causal price-premium measurement (C029). Wave 4’s soil-ecology reports (R13–R16) feed the mechanisms; the system experiment would resolve the keystone. Until then, the corpus’s line holds: the biology is demonstrated; the system is a hypothesis being lived by its practitioners.

Changelog

• v1.0 (2026-08-31) — Danny sign-off. Five calibration micro-edits from ChatGPT’s review applied: “not yet demonstrated” replaces “no” for system-level outcomes; the sweeping “everything confident said” sentence narrowed; Massuela framed as measured yield tradeoff (not “cost”); the phosphorus conclusion made misquote-resistant; “rounding term” → “comparatively minor share”; the Ch 11 reset principle stated as monitor-based engineering (with “until the soil tells you” retained as teaching hook only); “a living bed must die” bounded to the tested agents.
• v0.1 (2026-08-31) — Editor’s draft from ledger v1.0.