Thanks for your message! If the foreground branch/es are specified
a priori, based on a biological hypothesis (or various hypotheses), there is no need for FDR. If you do not have any
a priori hypothesis and you have just decided to go through all the branches on the tree one by one, then there is indeed a need to correct for multiple testing. If the latter, you may use FDR to do so as you suggest -- you can also read more details about multiple testing in
Anisomova and Yang 2007.
Hope this is somewhat helpful!