2003MNRAS.343..481L

The evolution of the metallicity of damped Lyman α systems (DLAs) is investigated in order to explore several scenarios for the nature of these systems. The observational data on chemical abundances of DLAs are analysed with robust statistical methods, and the abundances are corrected for dust depletion. The results of this analysis are compared with predictions of several classes of chemical evolution models describing a variety of scenarios for DLAs: one-zone dwarf galaxy models, multizone disc models and chemodynamical models representing dwarf galaxies. In order to settle constraints for star formation time-scales and metal production in DLAs, we compare the observational data on the [α/Fe] and [N/α] ratios to the predictions from the models. In DLAs, these ratios are only partially reproduced by the dwarf galaxy one-zone model and by the disc model. On the other hand, the chemodynamical model for dwarf galaxies reproduces the properties of nearly all DLAs. The connection between the gas flow evolution and the star formation rate is the reason for the ability of this model in reproducing the range of abundance ratios seen in DLAs. The comparison of the observed [α/Fe] and [N/α] trends with the predictions of the chemodynamical model is used to derive the formation epoch of dwarf galaxies. The chemodynamical model predicts that dwarf galaxies make a significant contribution to the observed total neutral gas density in DLAs, and that this contribution is more important at high redshifts (z ≳ 2-3). This is consistent with a scenario in which the DLA population is dominated by dwarf galaxies at high redshifts and by discs at lower redshifts. The relation between DLAs and Lyman break galaxies (LBGs) is investigated with chemodynamical models describing LBGs. Our results calls for a smoother progression in the evolutionary history of DLAs and LBGs rather than a sharp dichotomy between the two populations. LBGs and DLAs may constitute a sequence of increasing star formation rate, with the LBGs being systems with typically short star formation time-scales (∼10⁸yr), and the DLAs having slower star formation. We also raise the possibility that we could be missing a whole population of high Hi density column objects, with metallicities intermediate between those of DLAs and LBGs. Finally, we discuss the possibility that relying only on the observations of DLAs could lead to an underestimate of the metal content of the high-redshift Universe.