I did read your paper...and many others. They all had the same graphs vs. z, but not quasar luminosity on the same scale. I realize z represents the data, but actually velocity is even better. The z is a pain because it distorts the scale and luminosity needs a proportional time base for comparison since its value depends on time.
"Saying you 'couldn't find' a plot shows you didn't see my paper! Fig 1 shows the smoothed curve."
The data plotted is really simple and does not really show any recycling. Recycling is such a simple model that if it fit the data, there would be hundreds of papers touting that quantitative fit...there are no papers with quantitative fits for recycling, ergo, recycling just doesn't represent a substantial factor.
As a matter of fact, this plot seems very straightforward. SMBH quasars form at 12.25 Byr, collect galaxies and max in number at 10.3 Byr, decay to zip at 1 Byr, and there we are...in the dip as you say. Or, SMBHs are mostly in their ground states and no longer feeding on stars and gas and not radiating much any more.
It appears that you are the one trying to impose your model by contorting this data. You do not even plot the luminosity function in your paper and yet that is what everyone and their dog is doing in the literature.
There are lots of unanswered questions, that is sure, but recycling is just not consistent with this dataset nor with any other that I have seen. You did not do a quantitative fit in your paper and that weakens your argument. Noone else seems to do a recycle fit either and that further weakens your point. You need to get the numbers to work with more than just words. In any book, a radiated sun a year is just not much mass for a 200 billion star galaxy...