Defoliation: what the trials actually found
The two controlled trials on cannabis pruning found no significant yield or cannabinoid gain, and the heaviest treatment tested finished lowest of the three.
On this page
Search a grower forum for defoliation and someone will tell you it added 20 to 30% to their yield. Search the peer-reviewed literature for a controlled trial that tested that claim on cannabis and you will find exactly two, not decades of them, and neither found the yield or cannabinoid-concentration gain growers describe. One of them found the most heavily pruned treatment it tested finished lowest of the three compared.
- CBD yield spread, 9-week harvest
- 1,133.9–1,431.6mg/plant
- control, topped and lollipop treatments; not statistically significant, Crispim Massuela et al. 2022
- CBD concentration by treatment
- 7.9–9.0%
- not statistically significant across any pruning treatment
- Yield change from true leaf-strip defoliation
- not significant
- either cultivar, Danziger & Bernstein 2021
- Cultivars and trials with a controlled defoliation test
- 3cultivars across 2 trials
- as of this review
Three names, three different cuts: what these trials actually tested
The word "defoliation" covers at least three different cuts in grower conversation, and the two trials below do not test the same one. Sorting out which study removed what, before comparing numbers, is most of the honest answer here.
| Source | Term used | What was actually removed |
|---|---|---|
| Crispim Massuela et al. (2022) [1] | "Lollipop" | The two lowest side branches, cut at 27 days after planting, plus the next two lowest branches at 36 days: a lower-canopy branch strip, not a leaf strip |
| Danziger and Bernstein (2021) [2] | "Defoliation" | 85% of the plant's leaves, top leaves kept: the closer of the two treatments to what most growers mean by "defoliation" |
| Danziger and Bernstein (2021) [2] | "BBLR" (bottom branches and leaves removal) | All leaves and secondary branches from the bottom third of the plant, which the paper itself notes is "termed 'Lollipopping' in the cannabis jargon" [2] |
Read past the label before you compare a trial's result to your own plan. Massuela's "Lollipop" treatment is a branch strip at the base of the plant, closer to what this library calls lollipopping than to a whole-plant leaf strip. Danziger and Bernstein's "Defoliation" is the true leaf-strip most growers picture, and their separate "BBLR" treatment is the one their own paper equates with lollipopping. Conflating the two, as forum shorthand often does, is exactly how "defoliation" trials end up seeming to disagree with each other.
Massuela et al.: topping beat lollipopping on biomass, not on CBD yield at nine weeks
Crispim Massuela and colleagues grew a CBD-dominant (chemotype III) cultivar indoors at the University of Hohenheim, Germany, at a density of 14.4 plants/m², and applied one of three treatments to three replicate plants each: an untouched control, topping (an apical cut at the tenth node of the main stem), or the lollipop lower-branch strip described above [1]. Plants were harvested at four points, 5, 7, 9 and 11 weeks into flowering, with the 9-week harvest identified as chemically optimal for this cultivar.
Averaged across all four harvest times, topping produced significantly more dry matter than either alternative: 18.5 g of inflorescence dry weight per plant against 16.3 g (control) and 15.7 g (lollipop), and 9.0 g of leaf dry matter against 7.6 g and 6.9 g, with the inflorescence-weight difference reaching statistical significance [1]. These are small research-pot yields, not a representative commercial harvest, but the direction is clear on raw biomass: topping outgrew both alternatives.
That biomass advantage did not carry through to CBD yield at the 9-week harvest specifically, where each treatment had only three replicate plants rather than the twelve pooled across all four harvests. Topping still came out numerically highest at 1,431.6 mg of CBD per plant, control sat at 1,234.3 mg, and the lollipop treatment was lowest at 1,133.9 mg, a 26% gap between the top and bottom treatment that did not clear statistical significance at a 0.05 threshold [1]. Read that the way the sample size demands: with three plants per treatment, a real 20–30% difference and pure noise can look identical on the page. This trial cannot rule either one out, and it did not find one either.
Concentration didn't move either
Total CBD concentration, the share of dried flower weight that is CBD rather than the total milligrams per plant, was equally unmoved: 8.5% in the control, 7.9% in the lollipop treatment, 9.0% in topped plants, none of it statistically distinguishable [1].
Danziger and Bernstein: eight cuts, two cultivars, and a different question entirely
Danziger and Bernstein ran a considerably larger comparison: eight architecture treatments applied to two THC-dominant cultivars, "Himalaya" (a taller, sativa-leaning line) and "Fuji" (a shorter, indica-leaning line, both chemotype I, THC 10–16%, CBD under 0.1%), five replicate plants per treatment per cultivar, grown in a licensed commercial greenhouse in Israel [2]. Cannabis branches in a fixed, predictable order, a main stem, primary branches off it, secondary branches off those, each eventually capped by its own inflorescence once the plant switches to flowering [3], which is what makes "primary branch removal" and "secondary branch removal" distinct, meaningful categories rather than vague synonyms for defoliation.
Alongside an untouched control and true defoliation (85% of leaves removed, top leaves kept), the trial tested BBLR, defoliation combined with BBLR, removal of every primary branch, removal of every secondary branch, and two topping protocols:
| Treatment | What it removed |
|---|---|
| Control | Nothing |
| Defoliation | 85% of leaves, top leaves kept |
| BBLR ("lollipopping") | All leaves and secondary branches, bottom third of the plant |
| BBLR + defoliation | Both of the above, combined |
| Primary branch removal | Every branch off the main stem, weekly, through most of growth |
| Secondary branch removal | Every branch off the primary branches |
| Single prune (topping) | The growing tip, once, on day 0 |
| Double prune | Topping, then 5 cm off each of the six remaining branches |
The two branch-removal treatments cost the most yield, by a wide margin. Removing every primary branch cut shoot weight by 50% in "Himalaya" and produced a 98% yield loss in that cultivar; removing every secondary branch cut inflorescence weight by 59% in "Himalaya" and 29% in "Fuji" [2]. Defoliation alone, BBLR alone and both topping protocols did not significantly change the trial's dried, trimmed yield in either cultivar; the one exception was the combined BBLR-plus-defoliation treatment, which cost "Fuji" a small but statistically significant amount of yield, a result not seen in "Himalaya" or in either single treatment alone [2]. Stripping most of a plant's leaves cost this trial next to nothing in yield on its own; combining that strip with a branch strip, or amputating branch structure outright, did.
Cannabinoid concentration shifted in small, specific ways rather than across the board. Both defoliation treatments raised CBDVA, a minor cannabinoid, by 9–19%, while every other treatment reduced it by 6–23% [2]. Most treatments, except the two branch removals, raised THCA by 5–12% [2]. None of this is a change a grower would notice by eye or taste.
The trial's real subject was not yield, it was standardisation: how much a cannabinoid's concentration varies between the top of a plant and the bottom, a question this library covers in more depth in training and plant architecture: effects on yield uniformity. Danziger and Bernstein scored this as the share of sampled flowers within ±15% of the plant's own average concentration, averaged across every cannabinoid detected, into a "plant uniformity score" [2]. In "Fuji", BBLR scored highest (77% of samples within that band, against 67% for the control); in "Himalaya", a single topping scored highest (69%, against 50% for the control) [2]. Defoliation on its own landed in the middle of the pack on this composite score in "Fuji", but scored second only to single pruning in "Himalaya", another sign the two cultivars did not respond alike.
One comparison in the paper isolates what defoliation itself is doing. Control and defoliated plants are structurally identical, no branches removed from either, so the only thing that differs between them is microclimate: light and airflow reaching the lower canopy. On that comparison, defoliation nudged "Fuji's" uniformity score down by less than half a percentage point, essentially no change, but raised "Himalaya's" by 10.4 percentage points [2]. The authors also measured why: defoliation was one of only three treatments, with primary-branch removal and the BBLR-plus-defoliation combination, that significantly increased light reaching the base of the plant in both cultivars [2]. More light lower down evened out cannabinoid concentration in one cultivar and barely registered in the other.
The light-penetration and humidity argument: measured in part, mostly untested
Growers who defoliate usually give one of two reasons: more light reaches lower bud sites, or better airflow keeps humidity down in a dense canopy. Danziger and Bernstein's light-at-the-base measurements back the first claim directly: defoliation significantly increased photosynthetically active radiation reaching the plant base in both cultivars, alongside primary-branch removal and the combined BBLR-plus-defoliation treatment [2]. That part is measured. What is not measured, in this trial or the Massuela trial, is whether that extra light at the base actually filled out lower bud sites enough to change total yield: in this study it did not move inflorescence weight either way [2].
The airflow-and-humidity argument has even less behind it in the controlled literature. Dense, poorly ventilated canopies are a recognised risk factor for Botrytis and other flower-rot pathogens (see Botrytis and bud rot prevention), and removing leaves that trap moisture against a bud site is a reasonable, low-cost precaution on that logic alone. But neither controlled trial here measured humidity or disease incidence. This stays a plausible mechanism with no cannabis-specific trial behind it yet, not a proven yield or quality benefit.
What the evidence does not support
None of this backs the large claims circulating among growers. "Defoliation adds 20 to 30% to yield" has no controlled cannabis trial behind it, light or heavy. The one trial that tested the most aggressive lower-canopy technique found it produced the lowest CBD yield of the three compared, not the highest; while that specific gap did not reach statistical significance, it runs opposite to what the claim predicts [1]. Had heavy pruning been reliably adding yield, a trial built to detect exactly that would be expected to show the effect running the other way.
A defensible default, and where to check before you cut
Whatever you decide, any pruning method still has to sit inside your own jurisdiction's rules on home and licensed cultivation; check the law section before you plan a season around a technique. If you want the measured airflow benefit without stripping a full canopy, lollipopping removes far less material for a similar clean-up at the base. If you are weighing defoliation against other training methods for your own space, choosing a training method for your space works through the trade-offs, and the aggressive, twice-repeated version of this technique gets its own honest look in schwazzing: the two-strip method and what backs it.
As of September 2026, these two remain the only controlled trials to test pruning or leaf removal against an untouched cannabis control and report yield or cannabinoid data. This page gets revisited the day that changes, not before.
Sources
- Crispim Massuela D, Hartung J, Munz S, et al. (2022). Impact of Harvest Time and Pruning Technique on Total CBD Concentration and Yield of Medicinal Cannabis. Plants 11(1):140 Accessed 2026-09-26.
- Danziger N, Bernstein N (2021). Plant architecture manipulation increases cannabinoid standardization in 'drug-type' medical cannabis. Industrial Crops and Products 167:113528 Accessed 2026-09-26.
- Spitzer-Rimon B, Duchin S, Bernstein N, et al. (2019). Architecture and florogenesis in female Cannabis sativa plants. Frontiers in Plant Science 10:350 Accessed 2026-09-26.