The integrated stellar content of dark matter halos

Measurements of the total amount of stars locked up in galaxies as a function of host halo mass contain key clues about the efficiency of processes that regulate star formation. We derive the total stellar mass fraction f_star as a function of halo mass M500c from z=0.2 to z=1 using two complementary methods. First, we derive f_star using a statistical Halo Occupation Distribution model jointly constrained by data from lensing, clustering, and the stellar mass function. This method enables us to probe f_star over a much wider halo mass range than with group or cluster catalogs. Second, we derive f_star at group scales using a COSMOS X-ray group catalog and we show that the two methods agree to within 30%. We quantify the systematic uncertainty on f_star using abundance matching methods and we show that the statistical uncertainty on f_star (~10%) is dwarfed by systematic uncertainties associated with stellar mass measurements (~45% excluding IMF uncertainties). Assuming a Chabrier IMF, we find 0.012<f_star<0.025 at M500c=10^13 Msun and 0.0057<f_star<0.015 at M500c=10^14 Msun. These values are significantly lower than previously published estimates. We investigate the cause of this difference and find that previous work has overestimated f_star due to a combination of inaccurate stellar mass estimators and/or because they have assumed that all galaxies in groups are early type galaxies with a constant mass-to-light ratio. Contrary to previous claims, our results suggest that the mean value of f_star is always significantly lower than f_gas for halos above 10^13 Msun. Combining our results with recently published gas mas fractions, we find a shortfall in f_star+f_gas at R500c compared to the cosmic mean. This shortfall varies with halo mass and becomes larger towards lower halos masses.