Much has been said already about star formation histories in the preceding sections. In the past years, considerable effort has been spent on uncovering star formation histories of nearby dwarf galaxies by photometric means, i.e., via color-magnitude diagrams of the resolved stellar populations of the stars in these galaxies. Basically, very detailed star formation histories are derived photometrically through synthetic color-magnitude diagrams (CMDs) generated from isochrones assuming an initial mass function and a variety of different time-dependent star formation and chemical evolution histories. The synthetic CMDs are compared to the observed CMDs using various statistical techniques until the best match is found (e.g., Tosi et al. 1991; Gallart et al. 1996; Tolstoy & Saha 1996; Dolphin 1997; Holtzman et al. 1997). These methods necessarily have to make a number of assumptions (such as which initial mass function and which binary fraction to adopt), rely strongly on the chosen set of isochrones, and are plagued by the age-metallicity degeneracy, but yield surprisingly similar results in spite of the different approaches of different groups (see contributions in Lejeune & Fernandes 2002).
The information gained from deep CMDs can be complemented by information provided by individual age tracer stars such as Wolf-Rayet stars, carbon stars, RR Lyrae stars, etc. (see Grebel 1997, 1999) to yield a more complete picture of the star formation histories of the galaxies in question. In particular, such tracer populations are valuable when their specific host populations are too sparse to appear prominently in the CMDs. Furthermore, star formation histories vary as a function of position with the oldest populations being typically the most extended ones (e.g., Harbeck et al. 2001), hence ideally one wishes to cover the entire target galaxy.
Our current knowledge about the modes of star formation in LG dwarf galaxies can be summarized as follows: Irregular and dIrr galaxies are characterized by largely continuous star formation with amplitude variations of factors 2-3, governed mainly by internal, local processes. In the more massive irregulars with large gas reservoirs, star formation is likely to proceed for another Hubble time (Hunter 1997). Low-mass so-called dIrr/dSph transition-type galaxies are characterized by currently very low star formation rates and may eventually turn into quiescent dSphs. A detailed review of the evolutionary histories of these galaxies, large-scale and small-scale star formation properties, and their chemical evolution is given in Grebel (2004).
DE and dSph galaxies also tend to have continuous star formation rates, but with decreasing intensity. Some had their peak activity at very early ages, others several Gyr ago. Star formation tends to be longest-lasting in the centers of these galaxies, and in a number of cases age (and possibly metallicity) gradients are observed (Harbeck et al. 2001). Repetitive or episodic star formation, as mentioned earlier, has so far only found in Carina. For more detailed reviews, see Grebel (2000, 2001). A major puzzle is the surprising lack of gas in dSph galaxies, where (with the exception of Sculptor; Bouchard, Carignan, & Mashchenko 2003) only upper limits for neutral and ionized gas could be determined (Gallagher et al. 2003; Grebel et al. 2003, and references therein). Interestingly, the limits for neutral hydrogen lie well below the amounts expected from gas loss from the old red giants in the dSphs. Fornax, the only dSph galaxy with star formation as recent as ~ 200 Myr ago (Grebel & Stetson 1999), surprisingly also appears to be devoid of gas. It is neither understood what caused the gas loss to begin with nor how it is sustained, especially in the distant, isolated dSphs like Cetus and Tucana.
When illustrating star formation histories via population boxes (e.g., Grebel 1997, 1999, 2000), it becomes quickly obvious that no two dwarf galaxies - not even low-mass dSphs - share the same star formation history. Each galaxy needs to be considered as an individual with regard to the time scales of its star formation and the degree of enrichment. A trend of increasing intermediate-age population fractions with increasing distance from the Milky Way among dSphs was first noted by van den Bergh (1994), who attributed it to the possible environmental impact of the Milky Way. Star-forming material might have been removed earlier on from the closer Galactic companions via ram-pressure or tidal stripping, supernova-driven winds or the high UV-flux from the proto-Milky Way. On the other hand, if environment is primarily responsible for gas-poor dSphs, then the existence of the isolated dSphs Cetus and Tucana is difficult to understand. Again, knowledge of the orbits of these galaxies would be very helpful.
We may then turn to the M31 dSph companions, which cover a similar range of distances as the Galactic ones. Interestingly, these dSphs do not show any clear correlation between star formation history and present-day distance to M31. Their lack of a pronounced red clump and of substantial number carbon stars shows a lack of any prominent intermediate-age population regardless of their distance (Harbeck et al. 2001, 2004, 2005). Going one step further, the comparison of the stellar populations in M31's dSphs and of M31's halo shows that the dSphs cannot have been primary building blocks of M31's halo since it was found to be dominated by intermediate-age, comparatively metal-rich populations (Brown et al. 2003). (An old, metal-poor halo population, however, is present as well; Brown et al. 2004).
3.1. Mean metallicities and the metallicity-luminosity relation
Galaxies generally obey a metallicity-luminosity relation such that more luminous (and potentially more massive) galaxies are more metal-rich than faint galaxies with presumably more shallow potential wells. DIrrs and dSphs differ in their metallicity-luminosity relations (e.g., Binggeli 1994). Originally this was discovered when comparing nebular H II region oxygen abundances of dIrrs with stellar metallicity estimates of gas-poor dwarfs such as dSphs. However, in making this comparison, very different populations are compared, and very different metallicity tracers are used. Nebular oxygen abundances cannot easily be translated into mean stellar metallicities and vice versa. Mean stellar metallicities usually rely on the measurement of iron or of an element that has been found to be an excellent tracer of iron in certain populations, such as the near-infrared Ca II triplet (Armandroff & Da Costa 1991; Rutledge, Hesser, & Stetson 1997; Cole et al. 2004). These mean stellar metallicities are usually quoted as [Fe/H] or [Me/H] (to indicate that primarily metallicity as opposed to solely Fe is meant).
In order to avoid the uncertain O to Fe conversion, Richer, McCall, & Stasinska (1998) published a metallicity-luminosity relation based entirely on nebular O measurements: H II region abundances in dIrrs and planetary nebula (PN) abundances in dEs and dSphs. The offset between the resulting relations remained; however, one still compares different populations with each other. H II region abundances trace the present-day abundances of the most recently formed population of stars. PNe trace primarily intermediate-age populations with ages of at least several 100 Myr if not several Gyr. Furthermore, PNe have only been detected in two dSphs to date (Fornax and Sagittarius). Obviously one cannot expect to measure H II regions in the gas-deficient dSphs.
In order to compare not only mean stellar metallicities in dIrrs and dSphs, but also the metallicities of the same populations (i.e., of stars of similar age), we instead concentrated on old Population II giants, which as mentioned earlier have been detected in all LG dwarf galaxies. We used (1) old red giants in dSphs and in the outskirts of dIrrs (where old populations dominate), (2) spectroscopic abundances wherever available (from own Keck LRIS measurements and literature data from studies conducted at ESO, NOAO, and Keck), and (3) photometric abundances elsewhere (from comparison with globular cluster fiducials applied to our own deep HST data from a WFPC2 snapshot survey and archival or literature data). While the degree of homogeneity of the resulting data set is not ideal, it is the best currently available one and entirely based on well-calibrated empirical indicators.
The resulting metallicity-luminosity relationship shows that even when confined to old populations, there is a considerable offset between dSphs and dIrrs (Grebel et al. 2003). At the same galaxy luminosity, the old populations of dSphs are more metal-rich than those of dIrrs. This indicates that in contrast to dIrrs, dSphs must have experienced fairly rapid early enrichment. Together with various other factors, these evolutionary differences make normal dIrrs unlikely progenitors of dSphs (see also Binggeli 1994). DIrr/dSph transition-type galaxies, on the other hand, seem fairly plausible progenitors as explained in more detail in Grebel et al. (2003).
3.2. Detailed abundance ratios
Mean spectroscopic metallicities of individual stars provide crucial constraints on otherwise photometrically derived star formation histories and allow one to break the age-metallicity degeneracy - one of the prime areas of study with the new large telescopes.
Another very important area for large telescopes is high-resolution spectroscopy to measure individual abundance ratios. In particular, the determination of Fe, -, r-, and s-process element abundance ratios makes it possible to measure the modes and rates of star formation spectroscopically: the relative contribution of supernovae of Type II vs. Ia, and that of AGB stars, at different times during the evolution of the target galaxy. Importantly, this research permits one to compare the abundance ratios of different types of dwarf galaxies with those measured in various Galactic components and to constrain the building block scenario from the chemical point of view.
The [/Fe] ratios in dSphs (and dIrrs) at a given [Fe/H] are lower than those measured in the Galactic halo, indicating either low star formation rates in the dwarfs, loss of metals, or a larger contribution from supernovae of Type Ia. This is strong evidence against present-day dSphs as the dominant contributors to the build-up of the Galactic halo (Shetrone et al. 2001). For a discussion of the implications of recent results for nucleosynthesis and galaxy evolution, see also Venn et al. (2004). Extending these kinds of measurements and adding kinematic data as well is rapidly becoming one of the major research areas for the world's largest telescopes such as SALT, nicely complemented by ongoing and future space missions such as HST, JWST, and Gaia.
I would like to thank Joanna Mikolajewska for organizing a very stimulating conference and for her patience while this contribution was finished.