Smoothly rising star formation histories during the reionization epoch

Cosmological hydrodynamic simulations robustly predict that high-redshift galaxy star formation histories (SFHs) are smoothly rising and vary with mass only by a scalefactor. We use our latest simulations to test whether this scenario can account for recent observations at z ≥ 6 from WFC3/IR, NICMOS and IRAC. Our simulations broadly reproduce the observed ultraviolet (UV) luminosity functions and stellar mass densities and their evolution at z = 6-8, all of which are non-trivial tests of the mean SFH. In agreement with observations, simulated galaxies possess blue UV continua owing to young ages (50-150 Myr), low metallicities (0.1-0.5 Z ⊙ ) and low dust columns [E(B - V) < 0.05]. Our predicted Balmer breaks at z = 7, while significant, are ≈0.5 mag weaker than observed even after accounting for nebular line emission, suggesting observational systematic errors and/or numerical resolution limitations. Observations imply a near-unity slope in the stellar mass-star formation rate relation at all z = 6-8, confirming the prediction that SFH shapes are invariant. Dust extinction suppresses the UV luminosity density by a factor of 2-3, with suppression increasing modestly to later times owing to increasing metallicities. Current surveys detect the majority of galaxies with stellar masses exceeding 10 9 M ⊙ and few galaxies less massive than 10 8.5 M ⊙ , implying that they probe no more than the brightest ≈30 per cent of the complete star formation and stellar mass densities at z ≥ 6. Finally, we demonstrate that there is no conflict between smoothly rising SFHs and recent clustering observations. This is because momentum-driven outflows suppress star formation in low-mass haloes such that the fraction of haloes hosting observable galaxies (the 'occupancy') is 0.2-0.4 even though the star formation duty cycle is unity. This leads to many interesting predictions at z ≥ 4, among them that (1) optically selected and UV-selected samples largely overlap; (2) few galaxies exhibit significantly suppressed specific star formation rates; and (3) occupancy is constant or increasing with decreasing luminosity. These predictions are in tentative agreement with current observations, but further analysis of existing and upcoming data sets is required in order to test them more thoroughly.