Meaning you need to dedicate 1% of the tree's energy to light, not 0.02%.
They also use a 210 lumen bulb to compare with - but 210 lumens is very very dim. Assuming each tree is planted 20 meters apart, each tree needs to illuminate 400 m^2 of area, which would be 0.5 lux.
For comparison that's about as much light as the full moon provides.
For normal street lighting you want about 18 lux, meaning you actually need a 7000 lumen bulb.
So, now we need 35% of the trees energy delivered as light. But of course half of the light will be emitted upward, so we really need 70% of the trees energy.
Good luck with that.
And we haven't even touched on the fact that the light is not created on the surface of the plant, but rather throughout the tissue, so a lot of it will be lost internally.
The effective output level would have to be sized in such a way that even at the maximum time from the last illumination you're still above some minimum acceptable level which would mean that you'd have to be putting out a much larger amount of light than required early on.
Here is a similar (concept as far as I know) based on a hollow wall with a bacterial soup:
http://www.design.philips.com/philips/sites/philipsdesign/ab...