The shape of Model B resembles the SFR history expected for the ``monolithic collapse'' model of galaxy evolution, although expectations for hierarchical galaxy formation models are not ruled out (, see discussion in ; see also the contribution of G. Kauffmann to these Proceedings). Thus the monolithic-versus-hierarchical dilemma is still unsolved by our current knowledge of the SFR history. Values of the SFR density at higher redshifts (z 5) will be able place more definite constraints on the galaxy evolution scenario.
The final question we want to ask is what fraction of the total SFR density the Lyman-break galaxies represent at each given redshift. And how much of the SF is so deeply buried in dust that its accounting is missing. The obscuration curve discussed in Section 3 is technically valid only for UV-bright star-forming galaxies; it cannot, obviously, correct for objects which are missing from the sample because they are too dusty. On the one hand, Model B is only slightly in excess of the obscuration-corrected SFR density calculated from the z > 2 galaxies (by a negligible amount within the observational uncertainties), and the SFR history of Model B is perfectly sufficient to reproduce the observed CIB. It appears that the fraction of SF missed by considering the Lyman-break galaxies only is relatively small. On the other hand, a number of considerations invite to take this as a preliminary statement. We know that at low redshift a fraction of the star formation is deeply buried in dust, and is obscured even at IR wavelengths. The same could happen at high redshift, and the SCUBA sources seem to suggest that large dust contents are not impossible in high-z galaxies. The angular density of the SCUBA sources is about 1/2-1 of that of the z ~ 3 galaxies, and are spread over a (possibly) much larger redshift range than the Lyman-break galaxies, namely over ~ 5-10x larger cosmological volumes. The SCUBA sources are then ~ 5-20% of the Lyman-break galaxies by number density, but are forming stars with SFR 300-500 M yr-1. Thus the SCUBA galaxies could still add 25-100% to the SFR density of the obscuration-corrected Lyman-break galaxies, although an assessment of the AGN contribution is still missing.
Because of their characteristics, the two populations, the UV-bright Lyman-break galaxies and the FIR-bright SCUBA sources, are likely to be complementary, rather than overlapping. At the level of current knowledge, it appears that about 50-80% of the SF in the early phases of the Universe is accounted for by the obscuration-corrected Lyman-break galaxies; the remaining 20-50% of the SF may be contained in FIR-bright sources. However, more investigation of the nature, luminosity distribution and redshift placement of the SCUBA sources is needed before these figures can be taken at face value.
I am indebted with C. Steidel, M. Giavalisco, and M. Dickinson for making their most recent results on the Lyman-break galaxies available to me prior to publication, for discussions, and for a critical reading of the manuscript. I would like to thank the Organizing Committee for inviting me to this stimulating meeting and for financially support my stay at the Ringberg Castle.