MUFASA: The assembly of the red sequence

We examine the growth and evolution of quenched galaxies in the MUFASA cosmological hydrodynamic simulations that include an evolving halo mass-based quenching prescription, with galaxy colours computed accounting for line-of-sight extinction to individual star particles. MUFASA reproduces the observed present-day red sequence reasonably well, including its slope, amplitude and scatter. In MUFASA, the red sequence slope is driven entirely by the steep stellar mass-stellar metallicity relation, which independently agrees with observations. High-mass star-forming galaxies blend smoothly on to the red sequence, indicating the lack of a well-defined green valley at M* ≳ 10^10.5M_⊙. The most massive galaxies quench the earliest and then grow very little in mass via dry merging; they attain their high masses at earlier epochs when cold inflows more effectively penetrate hot haloes. To higher redshifts, the red sequence becomes increasingly contaminated with massive dusty star-forming (SF) galaxies; UVJ selection subtly but effectively separates these populations. We then examine the evolution of the mass functions of central and satellite galaxies split into passive and star-forming via UVJ. Massive quenched systems show good agreement with observations out to z ~ 2, despite not including a rapid early quenching mode associated with mergers. However, low-mass quenched galaxies are far too numerous at z ≲ 1 in MUFASA, indicating that MUFASA strongly overquenches satellites. A challenge for hydrodynamic simulations is to devise a quenching model that produces enough early massive quenched galaxies and keeps them quenched to z = 0, while not being so strong as to overquench satellites; MUFASA's current scheme fails at the latter.