Biased reionization and non-Gaussianity in redshifted 21-cm intensity maps of the reionization epoch

Spatial dependence in the statistics of redshifted 21-cm fluctuations promises to provide the most powerful probe of the reionization epoch. In this paper we consider the second and third moments of the redshifted 21-cm intensity distribution using a simple model that accounts for galaxy bias during the reionization process. We demonstrate that skewness in redshifted 21-cm maps should be substantial throughout the reionization epoch and on all angular scales, owing to the effects of galaxy bias which leads to early reionization in overdense regions of the intergalactic medium (IGM). The variance (or power spectrum) of 21-cm fluctuations will exhibit a minimum in redshift part way through the reionization process, when the global ionization fraction is around 50 per cent. This minimum is generic, and is due to the transition from 21-cm intensity being dominated by overdense to underdense regions as reionization progresses. We show that the details of the reionization history, including the presence of radiative feedback are encoded in the evolution of the autocorrelation and skewness functions with redshift and mean IGM neutral fraction. The amplitudes of fluctuations are particularly sensitive to the masses of ionizing sources, and vary by an order of magnitude for astrophysically plausible models. We discuss the detection of skewness by first-generation instruments, and conclude that the Mileura Wide-field Array-Low-Frequency Demonstrator will have sufficient sensitivity to detect skewness on a range of angular scales at redshifts near the end of reionization, while a subsequent instrument of 10 times the collecting area could map out the evolution of skewness in detail. The observation of a minimum in variance during the reionization history, and the detection of skewness would both provide important confirmation of the cosmological origin of redshifted 21-cm intensity fluctuations.