Well, here's the thing. I read that as well, but it isn't saying there is any harm between 24-50%, or that the enhanced growth disappeared, is it? I've read other studies that say green light's effect is also related to the R:FR ratio, and compared to the sun all broad spectrum LEDs have very little FR compared to nature. They are using mono leds in different ratios, where the situation is a little more complicated with cobs.
I just broke out the spreadsheet, and grabbed some SPD curves from Nichia cobs, and here's what it looks like. Blue is 400-500nm, Green is 500-600nm, Red is 600-700nm:
"Green" is usually broadly defined as 500-600nm, which of course includes a lot of yellow as well. Does that affect growth negatively? Maybe you should cut the percentages in half to get closer to green percentages like the study you quoted, since a bunch is yellow light. In which case you'd want much more green than you would think!
NFDLJ130B 3000k 80 CRI
View attachment 3680673
NFDLJ130B 5000k 80 CRI
View attachment 3680672
NFDLJ130B 3000k 90 CRI
View attachment 3680671
So yeah, even 3000k 90 CRI has way too much green, maybe, depending on how you interpret the data. That was interesting.
@alesh might have already done all of this, I just chose Nichia because they actually have tables of SPD data, so I could skip the digitizing.
My 4000k Citizen at 80 CRI actually looks like this:
View attachment 3680674
And who knows, maybe that is terrible! These really need alesh's Cree data for most people to find it applicable. Maybe some pie charts, and more fine-grained color differentiations, especially between green and yellow. But these are numbers I had close at hand, and I am being lazy.