Properties of galaxy groups in the Sloan Digital Sky Survey – II. Active galactic nucleus feedback and star formation truncation

Successfully reproducing the galaxy luminosity function (LF) and the bimodality in the galaxy distribution requires a mechanism that can truncate star formation in massive haloes. Current models of galaxy formation consider two such truncation mechanisms: strangulation, which acts on satellite galaxies, and active galactic nucleus (AGN) feedback, which predominantly affects central galaxies. The efficiencies of these processes set the blue fraction of galaxies, f(blue)(L, M), as a function of galaxy luminosity, L, and halo mass, M. In this paper, we use a galaxy group catalogue extracted from the Sloan Digital Sky Survey (SDSS) to determine f(blue)(L, M). To demonstrate the potential power of these data as a benchmark for galaxy formation models, we compare the results to the semi-analytical model for galaxy formation of Croton et al. Although this model accurately fits the global statistics of the galaxy population, as well as the shape of the conditional LF, there are significant discrepancies when the blue fraction of galaxies as a function of mass and luminosity is compared between the observations and the model. In particular, the model predicts (i) too many faint satellites in massive haloes, (ii) a blue fraction of satellites that is much too low, and (iii) a blue fraction of centrals that is too high and with an inverted luminosity dependence. In the same order, we argue that these discrepancies owe to (i) the neglect of tidal stripping in the semi-analytical model, (ii) the oversimplified treatment of strangulation, and (iii) improper modelling of dust extinction and/or AGN feedback. The data presented here will prove useful to test and calibrate future models of galaxy formation and, in particular, to discriminate between various models for AGN feedback and other star formation truncation mechanisms.