2009MNRAS.397..802H

Transformation of discs into spheroids via mergers is a well-accepted element of galaxy formation models. However, recent simulations have shown that the bulge formation is suppressed in increasingly gas-rich mergers. We investigate the global implications of these results in a cosmological framework, using independent approaches: empirical halo-occupation models (where galaxies are populated in haloes according to observations) and semi-analytic models. In both, ignoring the effects of gas in mergers leads to the overproduction of spheroids: low- and intermediate-mass galaxies are predicted to be bulge-dominated (B/T ∼ 0.5 at <10¹⁰M_☉, with almost no `bulgeless' systems), even if they have avoided major mergers. Including the different physical behaviour of gas in mergers immediately leads to a dramatic change: bulge formation is suppressed in low-mass galaxies, observed to be gas-rich (giving B/T ∼ 0.1 at <10¹⁰M_☉, with a number of bulgeless galaxies in good agreement with observations). Simulations and analytic models which neglect the similarity-breaking behaviour of gas have difficulty reproducing the strong observed morphology-mass relation. However, the observed dependence of gas fractions on mass, combined with suppression of bulge formation in gas-rich mergers, naturally leads to the observed trends. Discrepancies between observations and models that ignore the role of gas increase with redshift; in models that treat gas properly, galaxies are predicted to be less bulge-dominated at high redshifts, in agreement with the observations. We discuss implications for the global bulge mass density and future observational tests.