Complexity on small scales: The metallicity distribution of the Carina dwarf spheroidal galaxy

The Carina dwarf spheroidal galaxy is the only galaxy of this type that shows clearly episodic star formation separated by long pauses. Here we present metallicities for 437 radial velocity members of this Galactic satellite. The metallicities and radial velocities were measured as part of a Large Programme with the Very Large Telescope at the European Southern Observatory, Chile. We obtained medium-resolution spectroscopy with the multiobject spectrograph FLAMES. Our target red giants cover the entire projected surface area of Carina. Our spectra are centered at the near-infrared Ca II triplet, which is a well-established metallicity indicator for old and intermediate-age red giants. The resulting data sample provides the largest collection of spectroscopically derived metallicities for a Local Group dwarf spheroidal galaxy to date. Four of our likely radial velocity members of Carina lie outside this galaxy's nominal tidal radius, supporting earlier claims of the possible existence of such stars beyond the main body of Carina. We find a mean metallicity of [Fe/H] similar to -1.7 dex on the 1997 metallicity scale of Carretta and Gratton for Carina. The formal FWHM of the metallicity distribution function is 0.92 dex, while the full range of metallicities is found to span approximately -3.0 dex < [Fe/H] < 0: 0 dex. The metallicity distribution function might be indicative of several subpopulations distinct in metallicity. There appears to be a mild radial gradient such that more metal-rich populations are more centrally concentrated, matching a similar trend for an increasing fraction of intermediate-age stars (see the 2001 work of Harbeck and coworkers). This, as well as the photometric colors of the more metal-rich red giants, suggests that Carina exhibits an age-metallicity relation. Indeed, the age-metallicity degeneracy seems to conspire to form a narrow red giant branch despite the considerable spread in metallicity and wide range of ages. The metallicity distribution function is not well matched by a simple closed-box model of chemical evolution. Qualitatively better matches are obtained by chemical models that also take into account infall and outflows. A G dwarf problem remains for all these models.