Which light spectrum your cannabis plants actually need, and why the answer depends on your setup

Lead: LED manufacturers sell “optimised cannabis spectra” with red peaks, blue boosters and UV modules. The data says something else: spectrum effects depend almost entirely on how bright your room already is. Above 700 micromoles you can file the marketing away. Below that, it gets interesting.

What this is about

The first article in this series covered the big lighting levers: intensity, spectrum, UV, photoperiod. The second one went deep on photoperiod. Today it is the spectrum’s turn, and the finding is more nuanced than it looks.

There is no such thing as “the right cannabis spectrum”. There are conditions under which spectrum has a strong effect, and conditions under which it barely matters at all. If you do not separate the two, you either buy needlessly expensive specialist fixtures or you miss a lever that actually works.

The core insight rests on three studies we read in full: Magagnini et al. 2018 (Italy/Finland), Westmoreland et al. 2021 (USA, Utah State University) and Brousseau et al. 2021 (McGill, a review synthesis).

The spectrum paradox in two studies

Magagnini 2018: spectrum does everything

The Magagnini group compared three light sources in a controlled tent study: an HPS lamp (the standard) and two LED spectra (AP673L, high in red, and NS1, high in blue and UV-A). All three ran at the same PPFD of 450 micromoles.

The results:

THC concentration under NS1 (the high-blue LED): 38% higher than under HPS in the first run, 26% higher in the second.
CBG concentration: 100 to 200% higher under NS1.
CBD concentration: 35% higher under LED.
Flower yield: HPS slightly ahead (bigger, heavier), but the concentration advantage of the LEDs compensated for almost all of it.

At 450 micromoles PPFD, spectrum matters a lot. If you run in that light class, switching to an NS1-style LED can pull noticeably more cannabinoids out of the same plants.

Westmoreland 2021: spectrum does almost nothing

Three years later, the Bugbee group at Utah State University measured the opposite. Both studies are methodologically sound.

Westmoreland and colleagues held PPFD constant at 750 or 900 micromoles across three consecutive studies and varied only the blue fraction, from 4% (HPS) up to 20% (various LEDs).

What they found:

Yield dropped linearly as the blue fraction rose: a 12% drop going from 4% to 20% blue, consistent across all three studies.
Cannabinoid concentration: no significant spectrum effect on THC or CBD.
The white-plus-red LED setup produced 4.6% less yield per square metre than HPS, but 27% more yield per unit of electricity, because LED efficiency (micromoles per joule) was substantially better.

The practical message is the opposite: at higher PPFD, spectrum is a secondary lever, and fixture efficiency matters far more economically.

The resolution: photoreceptor saturation

Both studies are valid within their own setups. The contradiction is only apparent, because they are measuring different things.

Brousseau et al. (2021) supply the mechanistic explanation in their review: photoreceptors in the plant are not yet saturated at low light intensity. As long as cryptochrome (the blue light sensor) and phytochrome (the red light sensor) still have sensitivity headroom, they respond strongly to changes in spectrum. At high light intensity both are already largely saturated, so additional spectral variation changes little in the signalling cascade.

The threshold sits somewhere around 600 to 700 micromoles PPFD. Below it, spectrum tweaks can shift cannabinoid concentrations. Above it, photon quantity dominates.

That same threshold logic also explains why the UV-B findings reproduce so poorly in modern studies. The old Lydon 1987 study measured in a low-light setup, where UV-B worked as a strong stress signal. Under commercial high-output LEDs the plant is already photo-stressed enough, and additional UV-B stops making a measurable difference (Llewellyn 2022).

What this means for three types of setup

Hobby or small CSC room (300 to 500 micromoles PPFD)

Here your choice of spectrum makes a substantial difference. If you are aiming for cannabinoid profiles, you should:

Choose LEDs with a high blue fraction (15 to 25%) plus a red component
Actively avoid HPS: the Magagnini data show 25 to 38% less THC under HPS at the same PPFD
Treat UV-A as optional. It can help, but it is not a must
Consider subcanopy lighting with red plus blue to bring lower flower tiers up (Hawley 2018, cited in Brousseau)

Takeaway: at low installed power, the effort you put into LED spectrum pays off.

Medium CSC or small facilities (500 to 700 micromoles PPFD)

Border territory. Spectrum can still contribute something, but fixture efficiency becomes more important. The pragmatic recommendation:

Look for LEDs with good efficiency (above 2.3 micromoles per joule)
Do not over-weight spectrum. White-plus-red mixes with a moderate blue fraction (10 to 15%) are a robust choice
Read the Westmoreland data before committing to any large spectrum investment

Commercial production (above 700 micromoles PPFD)

This class means licensed production, for example medical cultivation or markets outside Germany. Cultivation associations under the German KCanG are non-commercial by law.

Here the logic shifts completely. Spectrum showed no measurable effect on yield or on cannabinoid concentration in these setups (Westmoreland 2021; Llewellyn 2022, who found no additional cannabinoid effect from supplemental UV under high-output LEDs).

What counts instead:

Fixture efficiency (micromoles per joule): the best white-plus-red LED setup returned 27% more yield per unit of electricity than HPS at the same PPFD.
PPFD scaled linearly up to 1,800 micromoles without saturating (Rodriguez-Morrison 2021). Whether your room can use that headroom depends on climate control and CO2.
A low blue fraction is better: 12% less yield at 20% blue than at 4% blue (Westmoreland).
You can file the spectrum marketing promises away. That goes for “cannabis-specific spectra”, “UV boost” and the surcharges that come with them.

Takeaway: in high-output setups, PPFD times efficiency dominates. Spectrum is fine-tuning, not an economic lever.

What spectrum RELIABLY does (independent of PPFD)

Three effects are stable regardless of light intensity, which makes them relevant in commercial setups too:

Plant architecture: a low red-to-far-red ratio (typical of HPS) stretches stems and internodes. LED setups produce more compact plants, which matters for vertical farm layouts.
Leaf area and photon capture: a higher blue fraction reduces leaf expansion, which partly explains the yield loss under blue-heavy spectra.
Energy efficiency per watt: white-plus-red LEDs have the best micromoles-per-joule ratio, simply by physics, because red costs less energy per photon than blue or green.

Stamford et al. (2023) model this with UK electricity prices and arrive at rules of thumb for which LED spectrum recipe pays off at which installed power.

What the studies do NOT say

Worth keeping in mind when you read the literature:

Terpene profiles are barely studied. Rodriguez-Morrison found minimal terpene effects from PPFD, and there is almost no cannabis-specific data on how spectrum affects terpenes.
Cultivar by spectrum interactions are untested. Magagnini used one cultivar, Westmoreland another. That both respond the same way is an assumption, not evidence.
Hardly anyone calculates the economics across harvest cycles. The Stamford analysis is the exception.
UV-B research in cannabis is thin: the old Lydon 1987 study is the only source for the THC and UV-B argument, and nobody has replicated it.

Takeaways for practice

Sort your setup class first: where is your PPFD? Below 500, between 500 and 700, or above that?
Low (below 500): an LED spectrum with a high blue fraction pays off, and HPS is worth avoiding.
Medium (500 to 700): prioritise fixture efficiency and treat spectrum as a bonus, not as the deciding factor.
High (above 700): ignore the spectrum marketing. Push PPFD up towards 1,800 micromoles as far as climate control and CO2 allow, and treat fixture efficiency (micromoles per joule) as the investment metric that counts.
UV-B hardware: stay sceptical in every setup. The evidence for a cannabinoid boost is missing.
Check cultivar compatibility: if you run a cultivar that was not tested in these studies (so, nearly always), run a test batch against a control before you convert the whole room.

Sources

Westmoreland et al. (2021), PLOS ONE. doi:10.1371/journal.pone.0248988
Magagnini et al. (2018), Med. Cannabis Cannabinoids. doi:10.1159/000489030
Brousseau, Wu, MacPherson, Morello, Lefsrud (2021), Front. Plant Sci.. doi:10.3389/fpls.2021.620021
Stamford et al. (2023), HortScience. doi:10.21273/hortsci16823-22
Rodriguez-Morrison et al. (2021), Front. Plant Sci.. doi:10.3389/fpls.2021.646020
Llewellyn et al. (2022), Front. Plant Sci.. doi:10.3389/fpls.2022.974018