The presence of excess scatter in the Ly-α forest at z ∼ 5.5, together with the existence of sporadic extended opaque Gunn-Peterson troughs, has started to provide robust evidence for a late end of hydrogen reionization. However, low data quality and systematic uncertainties complicate the use of Ly-α transmission as a precision probe of reionization’s end stages. In this paper, we assemble a sample of 67 quasar sightlines at z > 5.5 with high signal-to-noise ratios of >10 per ≤15 km s−1 spectral pixel, relying largely on the new XQR-30 quasar sample. XQR-30 is a large program on VLT/X-Shooter which obtained deep (SNR > 20 per pixel) spectra of 30 quasars at z > 5.7. We carefully account for systematics in continuum reconstruction, instrumentation, and contamination by damped Ly-α systems. We present improved measurements of the mean Ly-α transmission over 4.9 < z < 6.1. Using all known systematics in a forward modelling analysis, we find excellent agreement between the observed Ly-α transmission distributions and the homogeneous-UVB simulations Sherwood and Nyx up to z ≤ 5.2 (<1σ), and mild tension (∼2.5σ) at z = 5.3. Homogeneous UVB models are ruled out by excess Ly-α transmission scatter at z ≥ 5.4 with high confidence (>3.5σ). Our results indicate that reionization-related fluctuations, whether in the UVB, residual neutral hydrogen fraction, and/or IGM temperature, persist in the intergalactic medium until at least z = 5.3 (t = 1.1 Gyr after the big bang). This is further evidence for a late end to reionization.

Motivated by recent measurements of the number density of faint AGN at high redshift, we investigate the contribution of quasars to reionization by tracking the growth of central supermassive black holes in an update of the Meraxes semi-analytic model. The model is calibrated against the observed stellar mass function at z ∼ 0.6 − 7, the black hole mass function at z 0.5, the global ionizing emissivity at z ∼ 2 − 5 and the Thomson scattering optical depth. The model reproduces a Magorrian relation in agreement with observations at z < 0.5 and predicts a decreasing black hole mass towards higher redshifts at fixed total stellar mass. With the implementation of an opening angle of 80 deg for quasar radiation, corresponding to an observable fraction of ∼23.4 per cent due to obscuration by dust, the model is able to reproduce the observed quasar luminosity function at z ∼ 0.6 − 6. The stellar light from galaxies hosting faint AGN contributes a significant or dominant fraction of the UV flux. At high redshift, the model is consistent with the bright end quasar luminosity function and suggests that the recent faint z ∼ 4 AGN sample compiled by Giallongo et al. (2015) includes a significant fraction of stellar light. Direct application of this luminosity function to the calculation of AGN ionizing emissivity consequently overestimates the number of ionizing photons produced by quasars by a factor of 3 at z ∼ 6. We conclude that quasars are unlikely to make a significant contribution to reionization.

The transmission of Lyman-α (Lyα) in the spectra of distant quasars depends on the density, temperature, and ionization state of the intergalactic medium (IGM). Therefore, high-redshift (z &gt; 5) Lyα forests could be invaluable in studying the late stages of the epoch of reionization (EoR), as well as properties of the sources that drive it. Indeed, high-quality quasar spectra have now firmly established the existence of large-scale opacity fluctuations at z &gt; 5, whose physical origins are still debated. Here we introduce a Bayesian framework capable of constraining the EoR and galaxy properties by forward-modelling the high-z Lyα forest. Using priors from galaxy and CMB observations, we demonstrate that the final overlap stages of the EoR (when &gt;95 per cent of the volume was ionized) should occur at z &lt; 5.6, in order to reproduce the large-scale opacity fluctuations seen in forest spectra. However, it is the combination of patchy reionization and the inhomogeneous UV background that produces the longest Gunn-Peterson troughs. Lyα forest observations tighten existing constraints on the characteristic ionizing escape fraction of galaxies, with the combined observations suggesting $f_{\rm esc} \approx 7^{+4}_{-3}$ per cent, and disfavoring a strong evolution with the galaxy’s halo (or stellar) mass.