Dynamical virial masses of Lyman-break galaxy haloes at z=3

Abstract

We improve on our earlier dynamical estimate of the virial masses of the haloes of Lyman-break galaxies (LBGs) at redshift z=3 by accounting for the effects of seeing, slit width, and observational uncertainties. From an analysis of the small number of available rotation curves for LBGs we determine a relation Vc7=(1.9+/-0.2)sigma between circular velocity at a radius of 7kpc, and central line velocity width. We use this relation to transform the measured velocity widths of 32 LBGs to the distribution of circular velocities, for the population of LBGs brighter than R=25.5. We compare this distribution against the predicted distribution for the 'massive-halo' model in which LBGs pinpoint all of the highest mass dark matter haloes at that epoch. The observed LBG circular velocities are smaller than the predicted circular velocities by a factor >1.4+/-0.15. This is a lower limit as we have ignored any increase of circular velocity caused by baryonic dissipation. The massive-halo model predicts a median halo virial mass of 10^12.3 Msol, and a small spread of circular velocities. Our median estimated dynamical mass is <10^(11.6+/-0.3) Msol, which is significantly smaller; furthermore, the spread of our circular velocities is much larger than the massive-halo prediction. These results are consistent with a picture which leaves some of the most-massive haloes available for occupation by other populations which do not meet the LBG selection criteria. The median halo mass recently estimated by Adelberger et al. from the measured clustering of LBGs is 10^(11.86+/-0.3) Msol. Our dynamical analysis appears to favour lower masses and to be more in line with the median mass predicted by the collisional starburst model of Somerville et al., of 10^11.3 Msol. [abridged]