My graduate student, Daegene Koh, and I have just submitted a new paper (arXiv) on extending semi-numerical reionization models to include ionizing photons from the first stars and galaxies. These objects are often neglected because they may contribute a small fraction of the photon budget to reionization. However, we show that they do play a role during the start of reionization and alter the topology, i.e. biasing toward smaller ionized bubbles, at redshifts greater than ~10. With this new extension, we can explore the effects on reionization from the smallest star-forming dark matter halos and perhaps make predictions for upcoming 21cm detectors that will directly probe the Epoch of Reionization.
See the movie below to visually inspect the differences, where the left panel includes low-mass halos, whereas the right does not. The box size is 100 comoving Mpc, and the movie runs from redshift 25 to 6.