We investigate the evolution of high redshift seed black hole masses at late times and their observational signatures. The massive black hole seeds studied here form at extremely high redshifts from the direct collapse of pre-galactic gas discs. Popu
lating dark matter halos with seeds formed in this way, we follow the mass assembly of these black holes to the present time using a Monte-Carlo merger tree. Using this machinery we predict the black hole mass function at high redshifts and at the present time; the integrated mass density of black holes and the luminosity function of accreting black holes as a function of redshift. These predictions are made for a set of three seed models with varying black hole formation efficiency. Given the accuracy of current observational constraints, all 3 models can be adequately fit. Discrimination between the models appears predominantly at the low mass end of the present day black hole mass function which is not observationally well constrained. However, all our models predict that low surface brightness, bulgeless galaxies with large discs are least likely to be sites for the formation of massive seed black holes at high redshifts. The efficiency of seed formation at high redshifts has a direct influence on the black hole occupation fraction in galaxies at z=0. This effect is more pronounced for low mass galaxies. This is the key discriminant between the models studied here and the Population III remnant seed model. We find that there exists a population of low mass galaxies that do not host nuclear black holes. Our prediction of the shape of the black hole mass - velocity dispersion relation at the low mass end is in agreement with the recent observational determination from the census of low mass galaxies in the Virgo cluster.
