I am in the opposite side of the situation. I have about 20 trees that I started over the years and have been gradually merging them into a single large tree with several branches. I got tired of discovering the same person in two or more separate trees and realizing that I had already updated the person in one tree and was duplicating my research.
When I first started merging the trees, I figured that I would find a few duplicate individuals in each merge session. I ended up finding over 3000 duplicate individuals when all was said and done.
If you have a stable database that can handle the large tree, my opinion is to keep them in the same tree. You can always set up your filters to separate the website into the three branches, but with one single database file.