Mass downsizing and "top-down" assembly of early-type galaxies

Aims. We present a new analysis of the rest-frame B-band COMBO-17 and DEEP2 luminosity functions (LFs) of early-type galaxies (ETGs) as a function of luminosity and mass. Our aim is to place new stringent constraints on the evolution of ETGs since z similar to 1. Methods. We correct the LF( z) data for the luminosity dimming assuming pure luminosity evolution. However, instead of relying on stellar population synthesis model-dependent assumptions, we adopt the empirical luminosity dimming rate derived from the evolution of the Fundamental Plane of field and cluster massive ETGs. Results. Our results show that the amount of evolution for the ETG population depends critically on the range of luminosity and masses considered. While the number density of luminous (massive) ETGs with M-B(z = 0) < -20.5 ( M > 10(11) M-circle dot) is nearly constant since z similar to 0.8, less luminous galaxies display a deficit which grows with redshift and that can be explained with a gradual population of the ETG "red sequence" by the progressive quenching of star formation in galaxies less massive than similar to 10(11) M-circle dot. At each redshift there is a critical mass above which virtually all ETGs appear to be in place, and this fits well in the now popular "downsizing" scenario. However, "downsizing" does not appear to be limited to star formation, but the concept may have to be extended to the mass assembly itself as the build- up of the most massive galaxies preceeds that of the less massive ones. This evolutionary trend is not reproduced by the most recent theoretical simulations even when they successfully reproduce "downsizing" in star formation.