Key points are not available for this paper at this time.
We present a semi-empirical model to infer the atomic and molecular hydrogen content of galaxies as a function of halo mass and time. Our model combines the SFR–halo mass–redshift relation (constrained by galaxy abundances) with inverted SFR–surface density relations to infer galaxy H I and H2 masses. We present gas scaling relations, gas fractions, and mass functions from z = 0 to z = 3 and the gas properties of galaxies as a function of their host halo masses. Predictions of our work include: 1) there is a ∼ 0.2 dex decrease in the H I mass of galaxies as a function of their stellar mass since z = 1.5, whereas the H2 mass of galaxies decreases by 1 dex over the same period. 2) galaxy cold gas fractions and H2 fractions decrease with increasing stellar mass and time. Galaxies with M? 10 10 M are dominated by their stellar content at z 6 1, whereas less-massive galaxies only reach these gas fractions at z = 0. We find the strongest evolution in relative gas content at z 1.5. 3) the SFR to gas mass ratio decreases by an order of magnitude from z = 3 to z = 0. This is consistent with lower H2 fractions; these lower fractions in combination with smaller gas reservoirs correspond to decreased present-day galaxy SFRs. 4) an H2-based star-formation relation can simultaneously fuel the evolution of the cosmic star-formation and reproduce the observed weak evolution in the cosmic H I density. 5) galaxies residing in haloes with masses near 1012 M are most efficient at obtaining large gas reservoirs and forming H2 at all redshifts. These two effects lie at the origin of the high star-formation efficiencies in haloes with the same mass.
Popping et al. (Tue,) studied this question.