Carbon-enhanced metal-poor stars in dwarf galaxies

We investigate the frequency and origin of carbon-enhanced metal-poor (CEMP) stars in Local Group dwarf galaxies by means of a statistical, data-calibrated cosmological model for the hierarchical build-up of the Milky Way and its dwarf satellites. The model self-consistently explains the variation with dwarf galaxy luminosity of the observed: (i) frequency and [Fe/H] range of CEMP stars; (ii) metallicity distribution functions; (iii) star formation histories. We show that if primordial faint supernovae dominated the early metal-enrichment, then CEMP-no stars enriched by the first stellar generations should be present in all dwarf galaxies, with similar number of stars and CEMP fractions at [Fe/H] < -4. We demonstrate that the probability to observe a star that is carbon-enhanced within a given [Fe/H] range strongly depends on the luminosity of the dwarf galaxy and, on average, it is an order of magnitude lower in 'classical' Sculptor-like dwarf spheroidal (dSph) galaxies (P <= 0.02) than in the least luminous ultra-faint dwarfs (P approximate to 0.1). In addition, we explain why it may be easier to find CEMP-no stars at [Fe/H] approximate to -2 in classical dSph galaxies than in ultra-faint dwarfs. These are consequences of the dramatic variation in the fraction of stars at [Fe/H] < -4 with galaxy luminosity: >= 40 per cent for galaxies with L < 10(5) L-circle dot, and <= 0.2 per cent for L > 10(7) L-circle dot. We present model predictions for the low-Fe tail and CEMP fraction of stars in dwarf galaxies, with particular emphasis on the Sculptor dSph, that can be used to shed light on the properties of the first stars.