Phenotype Hunting: From a Packet of Seeds to One Keeper Mother
A phenotype hunt is not a lucky dip — it is a structured elimination tournament, and the growers who win it are the ones who wrote things down.
1. What a Phenotype Hunt Is — and What It Is Not
A phenotype hunt means growing a population from one cross under near-identical conditions, scoring every plant against criteria written down before you start, and keeping the best as clonally maintained mothers. It is not breeding — it selects only from what is in the packet.
| Term | Meaning | Changeable? |
|---|---|---|
| Genotype | Code the seed carries | No — fixed at fertilisation |
| Phenotype | Height, structure, timing, resin | Yes — environment shifts it |
| Chemical profile (chemotype) | Cannabinoid and terpene composition | Partly — composition inherited, quantity not |
Per Fundación CANNA’s terpene review, relative terpenoid content is strongly inherited, while total terpene yield per unit of tissue weight is more environmental. The shape of the profile travels with the clone; the intensity may not.
2. Population Size and the Mathematics of Selection
2.1 The Probability of Catching a Top Performer
Raw Genetics says there is no strict minimum; Neptune Seed Bank gives 10 to 100-plus. The familiar “50 to 100” is arithmetic. Treat each seed as an independent draw, a top performer as one in the top p fraction:
P(at least one) = 1 − (1 − p)^N
| Seeds (N) | Top 10% | Top 5% | Top 1% |
|---|---|---|---|
| 10 | 65.1% | 40.1% | 9.6% |
| 20 | 87.8% | 64.2% | 18.2% |
| 50 | 99.5% | 92.3% | 39.5% |
| 100 | 99.997% | 99.4% | 63.4% |
Read the miss rate: with 20 seeds the chance you miss every top-5% plant is 0.95^20 = 35.8%. Inverted, N = ln(0.05) / ln(1 − p) gives the population for 95% confidence: 29 plants for the top 10% (28.43), 59 for the top 5% (58.40), 299 for the top 1%.
2.2 What the Model Assumes
“Top 5%” means top 5% of the population in that packet, not of all cannabis on earth. An F1 from two inbred parents gives uniform offspring whose best plant sits barely above its median — the variance was never there. Segregating F2 and later generations hold it.
2.3 From Seeds to Scorable Females
| Stage (100 regular seeds) | Calculation | Plants |
|---|---|---|
| Germinated at 90% | 100 × 0.90 | 90 |
| Female at 1:1 | 90 × 0.50 | 45 |
| Past vegetative screening | top third | ~15 |
| Finalists re-run, then keeper | — | 3–4, then 1 |
Female count is a binomial draw: from 90 seedlings the standard deviation is √(90 × 0.5 × 0.5) = 4.74, so 45 ± 5 is ordinary. Feminised seed gives about 90, not 45.
3. How Much of What You See Is Genetic?
Selection works only on the heritable share of what you see. A 2026 preprint mapped quantitative trait loci in 155 female F2 plants, clonally grown in two environments.
| Trait group | Region | Variation explained |
|---|---|---|
| Two major cannabinoids, acid forms | Chromosome 7 | ~60% |
| Six monoterpenes, including alpha-pinene and terpinolene | Chromosome 5 | 47–66% |
| Sesquiterpenes | Five chromosomes | Smaller effects |
One population, and a preprint rather than peer-reviewed work. Petit and colleagues (Frontiers in Plant Science, 2020) phenotyped 123 hemp accessions and found broad-sense heritability of 0.94–0.95 for flowering time against 0.05 for fibre weight — grain and fibre accessions, not drug-type cannabis, but flowering time is clearly selectable.
That review also reports between-cultivar myrcene from 16.1% to 80.1% of the terpene fraction, and terpinolene at 37–42% in some mostly-sativa samples but trace in mostly-indica ones — proportions of the terpene fraction, not dry weight.
4. Stage One — Vegetative Screening
4.1 What to Score Before Flowering
Score all plants at the same age, not the same size.
| Criterion | What to measure |
|---|---|
| Vigour | Days to fourth node; height on a fixed day |
| Structure | Internode spacing at nodes 4–6, apical dominance, laterals |
| Resilience | One mild stress applied uniformly; who pests hit first |
| Roots | Mass and colour at transplant — most predictive |
| Clone-ability | Rooting percentage and days to roots |
4.2 Clone Every Candidate Before You Flower Anything
The step beginners skip: harvest a plant with no cutting taken and the genotype is gone. Raw Genetics puts it plainly — cut every plant before you know sex. A full hunt runs near a year, so the library must last.
| Rule | Reason |
|---|---|
| Two cuttings per plant | One rooting failure loses the genotype |
| Label before it leaves the mother | Unlabelled cuttings are unrecoverable |
| Record rooting date and success | This is the clone-ability score |
5. Stage Two — Flower Screening
5.1 Week-by-Week Observation Schedule
| Week | Record |
|---|---|
| 0 | Height at flip, node count, photograph — stretch baseline |
| 1–3 | Weekly height; pre-flower sites; aroma from a rubbed stem |
| 4 | Final height, stretch ratio, bud site count — canopy set |
| 5–6 | Bud formation, first resin, aroma in words, trichome cover |
| 7–9+ | Density by squeeze; open one lower bud; trichome heads; dry weight, trim minutes |
Plants with different flowering times do not ripen together, so harvest each on its own criteria.
5.2 Stretch Ratio and Canopy Planning
Stretch ratio is final height over height at flip: 55 cm to 121 cm gives 121 ÷ 55 = 2.2; 58 cm to 168 cm gives 168 ÷ 58 = 2.9. With 140 cm usable above the pot rim, maximum flip height is 140 ÷ 2.2 = 63.6 cm for the first, 140 ÷ 2.9 = 48.3 cm for the second.
5.3 Mould Resistance, Yield and Bag Appeal
In Thailand, Vietnam, Cambodia, Laos and the wider tropics, mould resistance is not one criterion among ten — in the wet season it decides whether there is a harvest. A dense bud holds interior moisture and rots from the inside, so the best density score can be the worst mould score.
| Observation | Method |
|---|---|
| Interior moisture | Squeeze the largest bud; how long it stays damp |
| Bud interior | Open one large lower bud in week 7 and week 9 |
| Lower canopy after a humid spike | Airflow is worst there: who lesions first |
Log relative humidity and temperature alongside, or a mould score describes your week, not the plant. Score yield as dry weight per square metre of canopy occupied, trim difficulty as minutes per dry gram, and bag appeal separately.
6. Stage Three — Post-Harvest Screening
Plants identical at harvest separate here — hence the live clone library.
| Criterion | What to record |
|---|---|
| Dry-down | Days to stem-snap, identical conditions for all |
| Aroma at jar-up and after cure | 0–10 plus description at 14, 30, 60 days |
| Smoke or vapour quality | Ash, smoothness, flavour at a fixed temperature |
| Effect | Onset, duration, character; blind-coded |
Aroma retention discriminates: two plants can jar up equally loud and sit far apart at 60 days. Label effect scores subjective — no cultivar treats or cures anything.
7. The Scoring System
7.1 Weighted Criteria Table
Score 0 to 10; contribution = score × weight ÷ 10; each column sums to 100.
| # | Criterion | Medical-supply consistency | Aroma focus | Commercial yield |
|---|---|---|---|---|
| 1 | Vigour, roots, structure | 8 | 6 | 10 |
| 2 | Clone-ability | 9 | 6 | 9 |
| 3 | Stress and pest resilience | 9 | 6 | 10 |
| 4 | Flowering time predictability | 11 | 5 | 12 |
| 5 | Stretch ratio consistency | 5 | 3 | 6 |
| 6 | Bud density and structure | 6 | 8 | 10 |
| 7 | Mould resistance | 13 | 9 | 12 |
| 8 | Aroma: character and retention | 11 | 38 | 8 |
| 9 | Yield and trim difficulty | 8 | 6 | 16 |
| 10 | Laboratory profile match | 20 | 13 | 7 |
| Total | 100 | 100 | 100 |
7.2 Worked Example — Two Competing Plants
Two finalists, weighted on the medical-supply column.
| Criterion # | Weight | A raw | A weighted | B raw | B weighted |
|---|---|---|---|---|---|
| 1 | 8 | 5 | 4.0 | 9 | 7.2 |
| 2 | 9 | 9 | 8.1 | 6 | 5.4 |
| 3 | 9 | 8 | 7.2 | 6 | 5.4 |
| 4 | 11 | 9 | 9.9 | 6 | 6.6 |
| 5 | 5 | 7 | 3.5 | 6 | 3.0 |
| 6 | 6 | 5 | 3.0 | 9 | 5.4 |
| 7 | 13 | 9 | 11.7 | 4 | 5.2 |
| 8 | 11 | 4 | 4.4 | 10 | 11.0 |
| 9 | 8 | 6 | 4.8 | 9 | 7.2 |
| 10 | 20 | 9 | 18.0 | 7 | 14.0 |
| Total | 100 | 71 | 74.6 | 72 | 70.4 |
Unweighted, B totals 72 against A’s 71, so a hunter counting ticks keeps B. Under medical-supply weights A wins: B’s weak mould resistance (7) and laboratory match (10) sit under the heaviest weights.
| Weight set | A | B | Winner | Margin |
|---|---|---|---|---|
| Unweighted raw sum | 71 | 72 | B | 1 |
| Medical-supply consistency | 74.6 | 70.4 | A | 4.2 |
| Aroma focus | 62.4 | 80.7 | B | 18.3 |
| Commercial yield | 71.0 | 72.3 | B | 1.3 |
Nothing about the plants changed. Only the question changed.
7.3 Why the Weights Must Match Your Goal
A medical-supply hunt weights repeatability: peak potency barely matters if it is not reproducible. An aroma hunt weights the profile and its survival through cure; a yield hunt weights output and labour. Fix the weights before scoring, and read the margin — 1.3 out of 100 is no result.
8. Data Discipline — Tags, Photographs and Logs
| Field | When |
|---|---|
| Plant identifier | Germination — everything hangs off this |
| Cross, breeder, seed lot | Germination — e.g. a NextGen Genetics lot code |
| Position in room; container and substrate batch | The largest confounder |
| Clone taken, rooted, success | Vegetative — the clone-ability score |
| Height, aroma, resin notes | Fixed weekly day, from the flip |
| Harvest week, dry weight, trim minutes | Harvest onward |
| Cure notes at 14, 30, 60 days; laboratory data | Post-harvest |
Hold inputs identical: whichever line you use — BioBizz, Grotek, CANNA, Advanced Nutrients — every plant gets the same product, strength and day.
Numbering: one scheme such as H14-07, on a tag attached to the plant not the pot, carried unchanged onto the clone tag, jar, photograph filename and laboratory label.
Photographs need fixed conditions: same distance, angle and neutral light, tag in frame, weekly plus macros at weeks 6, 8 and harvest. Magenta light ruins colour comparison.
Environmental logs let you claim you compared plants, not microclimates: temperature and relative humidity at canopy height in two or more spots, plus irrigation volume per plant.
9. Confounders — Why One Plant Is Weak Evidence
One unreplicated plant per genotype confounds genotype with position, and you cannot separate them afterwards — the information was never collected.
| Confounder | Mitigation |
|---|---|
| Position — centre gets more light, steadier air | Randomise, rotate, log positions |
| Light distribution — falls off at edges | Map it; put finalists in like zones |
| Container size — more volume, bigger plant | Identical pots, one substrate batch |
| Watering — hand watering differs per plant | Measure volume per plant |
The fix is replication: re-run finalists from library clones, three replicates each in randomised positions, then compare the spread within a genotype against the difference between genotypes. If those match you measured your room; if a plant only wins in the centre, it was the position.
10. Laboratory Analysis and the Certificate
| Analysis | Typical method |
|---|---|
| Cannabinoid potency | High-performance liquid chromatography (HPLC) |
| Terpene profile | Gas chromatography with mass spectrometry (GC-MS) — myrcene, limonene, linalool, pinene, caryophyllene, humulene, terpinolene, ocimene |
Two verified points from the American Oil Chemists’ Society review of cannabis testing. A proficiency survey found around 90% of laboratories now use liquid chromatography for potency, because gas chromatography needs an error-prone derivatisation step and its heat drives decarboxylation that inflates potency. And headspace gas chromatography profiled 38 terpenes yet stays semi-quantitative: some terpenes have low volatility and adsorb onto solids.
So terpene figures rank reliably but do not measure precisely: 1.8% total terpenes is not demonstrably better than 1.6% from another sample. Three cautions. The sample is not the plant — that review notes potency figures bounce around on unrepresentative sampling, and laboratory shopping is known. Compare within one laboratory, and replicate before believing a small gap. Score the match, not the maximum.
11. Keeper Maintenance Over Many Cycles
Finding the keeper is half the work. Holding it never ends.
| Task | Interval |
|---|---|
| Mother health check — vegetative, low stress | Weekly |
| Take and root a backup clone | Monthly |
| Pathogen and viroid testing | Before major propagation |
| Full re-run against the runner-up | Annually |
11.1 Drift, Somatic Mutation and Pathogen Load
“Genetic drift” is used loosely for a long-held clone line seeming to decline. Two mechanisms sit underneath.
Somatic mutation. The Cannavigia review attributes clonal degradation to somatic mutations accumulating without sexual reproduction, invoking Muller’s ratchet: irreversible mutations build up with no recombination to purge them. Its triggers are stressors — inconsistent light cycles, wrong substrate pH, over- and under-watering — and the longer a mother is kept and the more cuttings taken, the higher the risk. Replace mothers periodically from middle-region clones, which showed the least variation.
Pathogen load. Hop latent viroid is the cause most often mistaken for genetic decline. Oregon State University Extension publication EM 9570 (Cynthia M. Ocamb, August 2024) reports that infected drug-type cannabis shows reduced flower mass and trichome numbers plus reduced cannabinoid and terpene levels, that it travels in cuttings from infected mothers, and that its symptoms are too poorly known for visual diagnosis. Test mothers before cutting — a viroid is solvable, a mutation load is not.
Re-testing. Keep the runner-up clone alive. Every year or few cycles, run keeper and runner-up side by side in randomised positions, score both on the original rubric, and submit both to one laboratory — comparing against archived scores, not memory.
12. Front End of a Medical-Supply Workflow
This is also the entry stage of cultivar selection for medical supply, where the requirement is not a spectacular batch but batch-to-batch equivalence.
| Home hunt | Medical-supply selection |
|---|---|
| Finds the most impressive plant | Finds the most repeatable plant |
| Notebook or spreadsheet | Auditable records tied to batches |
| Testing optional | Testing per batch, samples retained |
Good agricultural and collection practice, and the pharmaceutical good manufacturing frameworks built on it, rest on traceability and reproducibility — which start with the plant you chose and the records proving why.
13. Share Your Scoring Sheets
Post your rubric — not just the winner, but the weights you chose and why, the criteria you dropped because they measured your room instead of your plants, and the plant you were sure about in week 5 that lost on cure. Tell us your population size, how many reached flower, your stretch ratios, and how mould resistance held up in a real wet season. If you have re-run finalists side by side, say whether the winner stayed the winner — almost nobody publishes that, and it is the most useful thing here.
This article is part of the Asiannabis Community educational series on cannabis genetics and breeding. Content is for educational purposes within jurisdictions where cannabis cultivation is legally permitted.