This is unlikely to be H2O-undersaturated melting, as this
normally involves breakdown of all the micas. It seems far more
likely to be wet melting of a plagioclase-, quartz-,
muscovite-bearing rock that does not have much K-feldspar.
The fact that fibrolite is present, as well as retrograde
muscovite may mean that you got a prograde reaction involving
muscovite breakdown, which 'simply' ran backwards in the retrograde
history (a quasi-closed system). The T may never have reached a level
consistent with major incongruent biotite breakdown. So, melting
would have been something like:
Bt (trace) + Ms + Pl + Qtz + H2O = Als + Grt (trace) + melt.
Most of the biotite would remain stable on the retrograde path, as
would the garnet because some of the melt was probably lost (as
suggested by Tracy). This would mean that, on the retrograde path you
remaining melt + Als = Kfs (trace) + Pl + Qtz + sericite + Bt (trace)
If this seems silly, it's only a model.
Have a good weekend,