The tracks of the chemical evolution models in Fig. 1 suggest that the bright star forming z~1.4 galaxies are likely to evolve into the population of less luminous but nonetheless rather massive, metal-rich galaxies that appear in the 0.5 < z < 0.9 galaxy population. The broad range of galaxy morphologies suggests that the metal-enriched reservoirs of star forming gas that we are probing at intermediate redshifts are being mostly consumed to build up both the disk and the bulge components of spiral galaxies.

Our analysis of the metallicity-luminosity relation at 0 < z < 1.5 suggests that the period of rapid chemical evolution takes place progressively in lower mass systems as the universe ages. The Figure shows the signatures of this "downsizing" effect: (a) at z < 0.7 (green symbols) nearly all galaxies with M_B,AB<-20 are fairly close to the low-z [O/H]-M_B relation (with one obvious exception); (b) at 0.7 < z < 0.9, this is true only for M_B,AB<-21.3; and (c) at z~1.4 even the most luminous galaxies are evolved off of the low redshift [O/H]-M_B,AB relation. As the Universe ages, particular signatures of "youth" (e.g., high [OIII]/[OII] or low [O/H]) are seen in progressively less luminous, less massive systems!

Which galaxies contribute to the star formation rate density over the past 8 billion years?

One of the key unanswered questions in the study of galaxy evolution is what physical processes inside galaxies drive the changes in the star formation rates in individual galaxies that, taken together, produce the large decline in the global star-formation rate density to redshifts since z ~ 2 (Lilly et al. 1996, Hippelein, Maier et al. 2003). Studies using the local SDSS sample have argued that the surface mass density may be more important than stellar mass in regulating star formation. Using the SDSS sample Brinchmann et al. (2004) found that the low specific star formation rate (SSFR, star formation rate / unit stellar mass) peak is more prominent at high surface density than at high stellar mass, and therefore concluded that the surface density of stars is more important than stellar mass in regulating star formation.

Figure 2. The shape of the specific SFR (SSFR) versus stellar mass surface density relation for relatively massive zCOSMOS z~0.7 galaxies is very similar to that of local SDSS galaxies (left panel). There is a roughly uniform increase in the average SSFR by a factor of 5-6 that is broadely independent of surface mass density, and which occurs for both late-type (middle panel), and early-type (right panel) galaxies. This emphasizes that galaxies of all types are contributing, proportionally, to the global increase in star formation rate density in the Universe back to these redshifts. Disk galaxies have a SSFR that is almost independent of surface mass density, and the same is probably also true of high Sersic index galaxies once obvious disk systems are excluded (red squares in the right panel).

Using the HST/ACS images of the COSMOS field, plus star formation rate information from emission lines measured in large numbers of zCOSMOS spectra we can study the changes that have occured in the SSFR - surface mass density relation between redshifts approaching z~1 and the present epoch, as sampled by the SDSS studies. The requirement is that we can select comparable samples at the different redshifts, and therefore we derive star formation rates, stellar masses, and structural parameters in a consistent way for both zCOSMOS and SDSS samples, and apply them to samples that are complete down to the same stellar mass at both redshifts.

Figure 3. The SSFR - surface mass density step-function (Fig.1) is clearly due to the change-over of different structural types from disk-dominated low Sersic galaxies (n<1.5) to bulge-dominated high Sersic galaxies (n>2.5), as the surface mass density increases. The cross-over point shifts to higher surface mass density in zCOSMOS compared to SDSS, because of a modest differential evolution in the size-mass relations of disk and spheroid galaxies.