2000A&A...363..851D

This paper predicts self-consistent faint galaxy counts from the UV to the submm wavelength range. The stardust spectral energy distributions described in Devriendt et al. (1999A&A...350..381D, Paper I) are embedded within the explicit cosmological framework of a simple semi-analytic model of galaxy formation and evolution. We begin with a description of the non-dissipative and dissipative collapses of primordial perturbations, and plug in standard recipes for star formation, stellar evolution and feedback. We also model the absorption of starlight by dust and its re-processing in the IR and submm. We then build a class of models which capture the luminosity budget of the universe through faint galaxy counts and redshift distributions in the whole wavelength range spanned by our spectra. In contrast with a rather stable behaviour in the optical and even in the far-IR, the submm counts are dramatically sensitive to variations in the cosmological parameters and changes in the star formation history. Faint submm counts are more easily accommodated within an open universe with a low value of Ω₀, or a flat universe with a non-zero cosmological constant. We confirm the suggestion of Guiderdoni et al. (1998MNRAS.295..877G) that matching the current multi-wavelength data requires a population of heavily-extinguished, massive galaxies with large star formation rates (∼500M_☉/yr) at intermediate and high redshift (z≥1.5). Such a population of objects probably is the consequence of an increase of interaction and merging activity at high redshift, but a realistic quantitative description can only be obtained through more detailed modelling of such processes. This study illustrates the implementation of multi-wavelength spectra into a semi-analytic model. In spite of its simplicity, it already provides fair fits of the current data of faint counts, and a physically motivated way of interpolating and extrapolating these data to other wavelengths and fainter flux levels.