Evidence for cosmic downsizing from the evolution of clustering of 24 mu m-selected galaxies

We investigate the clustering properties of two complete samples of 24 mu m-selected brighter than F-24 mu m = 400 mu Jy respectively set at redshifts z = [0.6-1.2] (low-z sample) and z >= 1.6 (high-z sample) in the overlapping region between the SWIRE and UKIDSS UDS surveys. The inferred correlation lengths are r(0) = 15.9(-3.4)(+2.9) Mpc in the case of z greater than or similar to 1.6 sources and r(0) = 8.5(-1.8)(+1.5) Mpc for the low-z sample, showing that the high-z population is sensibly more strongly clustered. Comparisons with physical models for the formation and evolution of large-scale structure reveal that the high-z sources are exclusively associated with very massive (M greater than or similar to 10(13) M-circle dot) haloes, comparable to those which locally host groups-to-clusters of galaxies, and are very common within such (rare) structures. Conversely, lower-z galaxies are found to reside in smaller halos (M-min similar to 10(12) M-circle dot) and to be very rare in such systems. On the other hand, mid-IR photometry shows that the low-z and high-z samples include similar objects and probe a very similar mixture of dust-enshrouded AGN and star-forming galaxies. While recent studies have determined a strong evolution of the 24 mu m luminosity function between z similar to 2 and z similar to 0, they cannot provide information on the physical nature of such an evolution. Our clustering results instead indicate that this is due to the presence of different populations of objects inhabiting different structures, as active systems at z less than or similar to 1.5 are found to be exclusively associated with low-mass galaxies, while very massive sources appear to have concluded their active phase before this epoch. Finally, we note that the small-scale clustering data seem to require steep (p proportional to r(-3)) profiles for the distribution of galaxies within their halos. This is suggestive of close encounters and/or mergers which could strongly favour both AGN and star formation activity.