Relative clustering and the joint halo occupation distribution of red sequence and blue-cloud galaxies in COMBO-17

This paper studies the relative spatial distribution of red-sequence and blue-cloud galaxies, and their relation to the dark matter distribution in the COMBO-17 survey as function of scale down to z similar to 1. We measure the second-order auto- and cross-correlation functions of galaxy clustering and express the relative biasing by using aperture statistics. Also estimated is the relation between the galaxies and the dark matter distribution exploiting galaxy-galaxy lensing (GGL). All observables are further interpreted in terms of a halo model. To fully explain the galaxy clustering cross-correlation function with a halo model, we introduce a new parameter, R, that describes the statistical correlation between numbers of red and blue galaxies within the same halo. We find that red and blue galaxies are clearly differently clustered, a significant evolution of the relative clustering with redshift is not found. There is evidence for a scale-dependence of relative biasing: the linear relative bias factor varies slightly between b similar to 1.7 +/- 0.5 and similar to 2.2 +/- 0.1 on spatial scales between roughly 100 h(-1) kpc and 7 h(-1) Mpc, respectively. The linear correlation coefficient of galaxy number densities drops from a value near unity on large scales to r similar to 0.6 +/- 0.15. Both biasing trends, the GGL and with some tension the galaxy numbers can be explained consistently within a halo model. Red galaxies typically start to populate haloes with masses starting from greater than or similar to 10(12.1 +/- 0.2) h(-1) M-circle dot, blue galaxies from greater than or similar to 10(11.2 +/- 0.1) h(-1) M-circle dot. For the cross-correlation function, one requires a halo occupation distribution (HOD) variance that becomes Poisson even for relatively small occupancy numbers. This rules out for our samples with high confidence a 'Poisson satellite' scenario, as found in semi-analytical models. We compare different model flavours, with and without galaxies at the halo centres, using Bayesian evidence. The result is inconclusive. However, red galaxies have to be concentrated towards the halo centre either by a red central galaxy or by a concentration parameter above that of dark matter. The value of R depends on the presence or absence of central galaxies: if no central galaxies or only red central galaxies are allowed, R is consistent with zero, whereas a positive correlation R = +0.5 +/- 0.2 is needed if both blue and red galaxies can have central galaxies.