Annu. Rev. Astron. Astrophys. 2009. 47: 371-425 Copyright © 2009 by Annual Reviews. All rights reserved

4. DETAILED ABUNDANCES OF RESOLVED STARS

The detailed chemical abundance patterns in individual stars of a stellar population provide a fossil record of chemical enrichment over different timescales. As generations of stars form and evolve, stars of various masses contribute different elements to the system, on timescales directly linked to their mass. Of course, the information encoded in these abundance patterns is always integrated over the lifetime of the system at the time the stars studied were born. Using a range of stars as tracers provides snapshots of the chemical enrichment stage of the gas in the system throughout the SFH of the galaxy. This approach also assumes that the chemical composition at the stellar surface is unaffected by any connection between interior layers of the star, where material is freshly synthesised, and the photosphere. This assumption is generally true for main-sequence stars, but evolved stars (giants or super-giants) will have experienced mixing episodes that modified the surface composition of the elements involved in hydrogen burning through the CNO cycle, i.e. carbon, nitrogen and possibly also oxygen.

These studies require precise measurements of elemental abundances in individual stars and this can only be done with high-resolution and reasonably high signal-to-noise spectra. It is only very recently that this has become possible beyond our Galaxy. It is efficient high-resolution spectrographs on 8-10m telescopes that have made it possible to obtain high resolution (R > 40000) spectra of RGB stars in nearby dSphs and O, B and A super-giants in more distant dIs. These stars typically have magnitudes in the range V = 17-19. Before the VLT and Keck, the chemical composition of extra-galactic stars could only be measured in super-giants in the nearby Magellanic Clouds (e.g., Wolf 1973, Hill, Andrievsky & Spite 1995, Hill, Barbuy & Spite 1997, Venn 1999), yielding present day (at most a few 10⁷yr ago) measurements of chemical composition. Looking exclusively at young objects however makes it virtually impossible to uniquely disentangle how this enrichment built up over time.

4.1. Dwarf Spheroidal Galaxies

The first studies of detailed chemical abundances in dSph galaxies are those of Shetrone, Bolte & Stetson (1998), Shetrone, Côté & Sargent (2001, 17 stars in Draco, Ursa Min & Sextans) using Keck-HIRES and Bonifacio et al. (2000, 2 stars in Sgr) using VLT-UVES. These early works were shortly followed by similar studies slowly increasing in size (Shetrone et al. 2003, Bonifacio et al. 2004, Sadakane et al. 2004, Geisler et al. 2005, McWilliam & Smecker-Hane 2005). The total number of stars probed in individual studies remained very low (typically only 3 to 6 in any one galaxy except for Sgr). This was because the stars had to be observed one at a time, and for the most distant dSphs this required exposure times of up to 5 hours per star. Nevertheless from these small samples it was already clear that dSph galaxies follow unique chemical evolution paths, which are distinct from that of any of the MW components (e.g., Shetrone, Côté & Sargent 2001, Shetrone et al. 2003, Tolstoy et al. 2003, Venn et al. 2004a).

Most recently, high-resolution spectrographs with high multiplex capabilities have resulted in large samples (> 80 stars) of high resolution spectra of individual stars to determine abundances in a relatively short time. The FLAMES multi-fiber facility on VLT (Pasquini et al. 2002) has so far been the most productive in this domain. There are a number of FLAMES high resolution spectroscopy studies in preparation, but some results are already published for Sgr and its stream (Monaco et al. 2005, Sbordone et al. 2007, Monaco et al. 2007, 39 stars), Fnx (Letarte 2007, 81 stars), Carina (Koch et al. 2008a, 18 stars) and Scl (Hill et al. in preparation, 89 stars).

These new extensive studies not only provide abundances with better statistics, but they also allow statistical studies over the total metallicity range in each galaxy. This allows for an almost complete picture of their chemical evolution over time, with abundance trends as a function of metallicity for each system. Only the most metal-poor regime in these systems is perhaps still somewhat under-represented in these samples, although this is in part because they are rare (Helmi et al. 2006), and in part because these large samples of abundances have been chosen in the inner parts of the galaxies, where younger and/or more metal-rich populations tend to dominate (Tolstoy et al. 2004, Battaglia et al. 2006). New studies to fill in this lack of measured abundances in low metallicity stars are in preparation (e.g., Aoki et al. 2009). In the following, we will consider groups of elements that give particular insights into dwarf galaxy evolution.

4.1.1 ALPHA ELEMENTS     The -elements abundances that can easily be measured in RGB spectra includes O, Mg, Si, Ca and Ti. Although the -elements have often been considered as an homogeneous group, and their abundances are sometimes averaged to produce a single [/Fe] ratio, their individual nucleosynthetic origin is not always exactly the same. For example, O and Mg are produced during the hydrostatic He burning in massive stars, and their yields are not expected to be affected by the SNII explosion conditions. On the other hand Si, Ca and Ti are mostly produced during the SNII explosion. This distinction is also seen in the observations (e.g., Fulbright, McWilliam & Rich 2007), where Si, Ca and Ti usually track one another, but O and Mg often show different trends with [Fe/H]. It is therefore generally advisable to treat the three -elements which are well probed in dwarf galaxies separately. Fig. 11 shows a compilation of Mg and Ca abundances of individual stars in those dSphs with more than 15 measurements.

Figure 11. Alpha-elements (Mg and Ca) in four nearby dwarf spheroidal galaxies: Sgr (red: Sbordone et al. 2007, Monaco et al. 2005, McWilliam & Smecker-Hane 2005), Fnx (blue: Letarte 2007, Shetrone et al. 2003), Scl (green: Hill et al. in prep; Shetrone et al. 2003, Geisler et al. 2005) and Carina (magenta: Koch et al. 2008a, Shetrone et al. 2003). Open symbols refer to single-slit spectroscopy measurements, while filled circles refer to multi-object spectroscopy. The small black symbols are a compilation of the MW disk and halo star abundances, from Venn et al. (2004a).

The apparent paucity of -elements (relative to iron) in dSph galaxies compared to the MW disk or halo was first noted by Shetrone, Bolte & Stetson (1998), Shetrone, Côté & Sargent (2001), Shetrone et al. (2003), Tolstoy et al. (2003), Venn et al. (2004a) from small samples. Fig. 11 shows this convincingly over most of the metallicity range in each system. However, it also appears that each of these dSphs starts, at low [Fe/H], with [/Fe] ratios similar to those in the MW halo at low metallicities. These ratios in the dSphs then evolve down to lower values than is seen in the MW at the same metallicities.

The ratio of -elements to iron, [/Fe], is commonly used to trace the star-formation timescale in a system, because it is sensitive to the ratio of SNII (massive stars) to SNIa (intermediate mass binary systems with mass transfer) that have occurred in the past. SNIa have a longer time scale than SNII and as soon as they start to contribute they dominate the iron enrichment and [/Fe] inevitably decreases. After that, no SFH can ever again result in enhanced [/Fe], unless coupled with galactic winds removing only the SNIa ejecta and not that of SNII. This is seen as a "knee" in a plot of [Fe/H] vs. [/Fe], see Fig. 11. The knee position indicates the metal-enrichment achieved by a system at the time SNIa start to contribute to the chemical evolution (e.g., Matteucci & Brocato 1990, Matteucci 2003). This is between 10⁸ and 10⁹yrs after the first star formation episode. A galaxy that efficiently produces and retains metals over this time frame will reach a higher metallicity by the time SNIa start to contribute than a galaxy which either loses significant metals in a galactic wind, or simply does not have a very high SFR. The position of this knee is expected to be different for different dSphs because of the wide variety of SFHs. In the data there are already strong hints that not all dSphs have a knee at the same position.

At present the available data only cover the knee with sufficient statistics to quantify the position in the Scl dSph, a system which stopped forming stars 10 Gyr ago, and the knee occurs at [Fe/H] -1.8. This is the same break-point as the two kinematically distinct populations in this galaxy (Tolstoy et al. 2004, Battaglia 2007), see Fig. 9. This means that the metal-poor population has formed before any SNIa enrichment took place, which means on a timescale shorter than 1 Gyr.

In other dSphs the knee is not well defined due to a lack of data, but limits can be established. The Sgr dSph has enhanced [/Fe] up to [Fe/H] -1.0, which is significantly more metal-rich than the position of the knee in the Scl dSph. This is consistent with what we know of the SFH of Sgr, which has steadily formed stars over a period of 8-10 Gyrs, and only stopped forming stars about 2-3 Gyr ago (e.g., Dolphin 2002). The Carina dSph has had an unusually complex SFH, with at least three separate bursts of star formation (Hurley-Keller, Mateo & Nemec 1998), see Fig. 4. The abundance measurements in Carina are presently too scarce to have any hope to confidently detect these episodes in the chemical enrichment pattern (e.g., Tolstoy et al. 2003). It appears to possesses [/Fe] poor stars between [Fe/H] = -1.7 and -2.0, which suggests that the knee occurs at lower [Fe/H] than in Scl. It seems that Carina has had the least amount of chemical evolution before the onset of SNIa of all galaxies in Fig. 11. In the Fnx dSph, another galaxy with a complex SFH, the sample does not include a sufficient number of metal-poor stars to determine even an approximate position of the knee. There are abundances for only five stars below [Fe/H] = -1.2, and only one below [Fe/H] = -1.5. The knee is constrained to be below [Fe/H] < -1.5. From this (small) sample of dSph galaxies, it appears that the position of the knee correlates with the total luminosity of the galaxy, and the mean metallicity of the galaxy. Which suggests that the presently most luminous galaxies are those that must have formed more stars at the earliest times and/or retained metals more efficiently than the less luminous systems.

The abundance ratios observed in all dSphs for stars on the metal-poor side of the knee, tend to be indistinguishable from those in the MW halo. From this small sample it seems that the first billion years of chemical enrichment gave rise to similar enrichment patterns in small dwarf galaxies and in the MW halo. Because the [/Fe] at early times is sensitive to the IMF of the massive stars, if [/Fe] in metal-poor stars in dSphs and in the MW halo (or even the bulge) are similar, then there is no need to resort to IMF variations between these systems. Poor IMF sampling has been invoked as a possible cause of lowering [/Fe] in dwarfs (Tolstoy et al. 2003, Carigi & Hernandez 2008), but these new large samples suggest that this explanation may no longer be necessary, at least in systems as luminous as Sculptor, Fornax or Sagittarius. On the other hand, there is now a hint that the slightly less luminous Sextans (M_V = -9.5) could display a scatter in the /Fe ratios at the lowest metallicities, including [/Fe] close to solar (Aoki et al. 2009). Such a scatter is so far observed only in this purely old system, and suggests a very inhomogeneous metal-enrichment in this system that presumably never retained much of the metal it produced. The true extent of this scatter in Sextans remains to be investigated, and extension to other similar systems is needed before general conclusions can be reached on the mechanisms leading to the chemical homogeneity -or not- of dwarf galaxies.

At later times, in those stars which formed ~ 1 Gyr after the first stars, on the metal-rich side of the knee, the decrease of [/Fe] with increasing metallicity is very well marked. In fact, the end points of the evolution in each of the dSphs investigated has a surprisingly low [/Fe], see Fig. 11. A natural explanation of these low ratios could involve a sudden decrease of star formation, that would make enrichment by massive stars inefficient and leave SNIa to drive the chemical evolution. This sudden drop in star formation could be the natural result of galactic winds which can have a scientific impact on dwarf galaxies with relatively shallow potential wells (see Section 5) or perhaps tidal stripping. In this case, one would expect the metal-rich and low [/Fe] populations to be predominantly young, corresponding to the residual star formation after the sudden decrease. However, the current age-determinations for individual giants in these systems are not accurate enough to probe this hypothesis (e.g., Battaglia et al. 2006).

4.1.2 SODIUM AND NICKEL    Another example of the low impact of massive-stars on the chemical enrichment of dSphs is given by sodium. Fig. 12 shows the compilation of dSphs stars compared to the evolution of Na in the MW. According to stellar current models, Na is mostly produced in massive stars (during hydrostatic burning) with a metallicity-dependent yield. The abundance of Na in metal-poor dSph stars is apparently not different to the MW halo stars at the same [Fe/H], but its abundance at later stages in the evolution is distinct from the MW, dSph producing (or keeping) too little Na to keep on the MW trend above [Fe/H] > -1.

Figure 12. Sodium (above) and nickel (below) in the same four dSphs as in Fig. 11, compared to the MW. Sgr (red: Sbordone et al. 2007, Monaco et al. 2005, McWilliam & Smecker-Hane 2005), Fnx (blue: Letarte 2007, Shetrone et al. 2003), Scl (green: Hill et al. in prep; Shetrone et al. 2003, Geisler et al. 2005) and Carina (magenta: Koch et al. 2008a, Shetrone et al. 2003). Open symbols refer to single-slit spectroscopy measurements, while filled circles refer to multi-object spectroscopy. The small black symbols are a compilation of the MW disk and halo star abundances, from Venn et al. (2004a).

Sodium and nickel under-abundances have also been remarked upon by Nissen & Schuster (1997, 2009) in a fraction of halo stars which also display low abundances, thereby producing a [Na/Fe] - [Ni/Fe] correlation. This correlation is tentatively explained as the common sensitivity of both elements to neutron-excesses in supernovae. Fnx is the most striking example that seems to follow the same slope as the Na-Ni relationship in the MW, but extending the trend to much lower [Na/Fe] and [Ni/Fe] values (Letarte 2007), see Fig. 12. Nickel, unlike sodium, is also largely produced in SNIa (Tsujimoto et al. 1995), so the Ni-Na relation can in theory be modified by SNIa nucleosynthesis, especially in the metal-rich populations of dwarfs where the low [/Fe] ratios point towards a strong SNIa contribution.

4.1.3 NEUTRON-CAPTURE ELEMENTS    Despite their complicated nucleosynthetic origin, heavy neutron capture elements can provide useful insight into the chemical evolution of galaxies. Nuclei heavier than Z ~ 30 are produced by adding neutrons to iron (and other iron-peak) nuclei. Depending on the rate (relative to decay) at which these captures occur, and therefore on the neutron densities in the medium, the processes are called either slow or rapid (s- or r-) process. The s-process is well constrained to occur in low to intermediate-mass (1-4 M) thermally pulsating AGB stars (see Travaglio et al. 2004 and references therein), and therefore provide a contribution to chemical enrichment that is delayed by ~ 100-300 Myrs from the time that the stars were born. Thus s-process elements can in principle be used to probe star formation on similar timescales to [/Fe]. The r-process production site is clearly associated with massive star nucleosynthesis. The most plausible candidate being SNII, although the exact mechanism to provide the very large neutron densities needed is still under debate, (e.g., Sneden, Cowan & Gallino 2008 and references therein). This means that r-process elements should contribute to the chemical enrichment of a galaxy with very little, if any, delay. Obviously they need pre-existing Fe-peak seeds and are therefore not primary elements such as elements. One complication arises from the fact that most neutron-capture elements (through their multiple isotopes) can be produced by either the s- or the r- process, such as yttrium (Y), barium (Ba) or lanthanum (La). Among the few exceptions is europium (Eu), which is almost exclusively an r-process product.

Fig. 13 compares Ba and Eu abundances in four dSph galaxies and in the MW. At first glance, the Eu evolution in dSph galaxies resembles that of their respective -elements (see Fig. 11), as expected for an r-process originating in massive stars. In the MW, the Ba and Y are dominated by the r-process for [Fe/H] -2.0 (e.g., Simmerer et al. 2004, Johnson & Bolte 2002), while the s-process dominates at higher metallicities (e.g., more than 80% of the solar Ba is of s-process origin).

Figure 13. Neutron-capture elements Y, Ba & Eu in the same four dSphs as in Fig. 11, compared to the MW. Sgr (red: Sbordone et al. 2007, Monaco et al. 2005, McWilliam & Smecker-Hane 2005), Fnx (blue: Letarte 2007, Shetrone et al. 2003), Scl (green: Hill et al. in prep; Shetrone et al. 2003, Geisler et al. 2005) and Carina (magenta: Koch et al. 2008a, Shetrone et al. 2003). Open symbols refer to single-slit spectroscopy measurements, while filled circles refer to multi-object spectroscopy. The small black symbols are a compilation of the MW disk and halo star abundances, from Venn et al. (2004a).

At early times (at [Fe/H] < -1) there seems to be little difference between the various dSphs, and the MW halo in Fig 13. However, there is a hint that at the lowest metallicities ([Fe/H] < -1.8), [Ba/Fe] increases in scatter and starts to turn down. This hint is confirmed in the plot in Section 4.2 which includes other dSphs, although from much smaller samples (Shetrone, Côté & Sargent 2001, Fulbright, Rich & Castro 2004, Aoki et al. 2009). In fact, this scatter and downturn of [Ba/Fe] is a well known feature in the MW halo (François et al. 2007, Barklem et al. 2005 and references therein), where it occurs at much lower metallicities ([Fe/H] < -3.0). So far we have extremely low number statistics for dSphs and these results need to be confirmed in larger samples of low-metallicity stars. These low r-process values at higher [Fe/H] than in the Galactic halo would either mean that the dwarf galaxies enriched faster than the halo at the earliest times or that the site for the r-process is less common (or less efficient) in dSphs. The r-process elements are clearly useful tracers of early time scales, because unlike the -elements (in the halo and in dSphs) they show significant scatter in the lowest metallicity stars. The r-process is thus produced in much rarer events than the -elements and so it can be a much finer tracer of time scales and enrichment (and mixing) processes.

The ratio of [Ba/Eu], shown in Fig. 14, indicates the fraction of Ba produced by the s-process to that produced by the r-process. In dSphs, as in the MW, the early evolution of all neutron-capture elements is dominated by the r-process (this was already noted by Shetrone, Côté & Sargent 2001, Shetrone et al. 2003). In each system, however, the low and intermediate mass AGB stars contribute s-process elements, that soon start to dominate the Ba (and other neutron capture elements) production. The metallicity of this switch from r- to s-process ([Fe/H] ~ -1.8, the same as the [/Fe] knee) is only somewhat constrained in the Scl dSph. This turnover needs to be better constrained in Scl and even more so in other galaxies to provide timing constraints on the chemical enrichment rate. It could reveal the metallicity reached by the system at the time when the s-process produced in AGBs starts to contribute.

Figure 14. Ratios of r- to s- process element production in the same four dSphs as in Fig. 11, compared to the MW. Sgr (red: Sbordone et al. 2007, Monaco et al. 2005, McWilliam & Smecker-Hane 2005), Fnx (blue: Letarte 2007, Shetrone et al. 2003), Scl (green: Hill et al. in prep; Shetrone et al. 2003, Geisler et al. 2005) and Carina (magenta: Koch et al. 2008a, Shetrone et al. 2003). Open symbols refer to single-slit spectroscopy measurements, while filled circles refer to multi-object spectroscopy. The small black symbols are a compilation of the MW disk and halo star abundances, from Venn et al. (2004a).

For the more metal-rich stars ([Fe/H] > -1) there is also a distinctive behaviour of [Ba/Fe] in dSphs (Fig. 13). In the Scl dSph the [Ba/Fe] values never leave the MW trend, but this galaxy also has almost no stars more metal-rich than [Fe/H] < -1. Fnx, on the other hand, and to a lesser extent Sgr, display large excesses of barium for [Fe/H] > -1. This is now barium produced by the s-process, and it shows the clear dominance of the s-process at late times in dSphs.

Fig. 15 shows the trends of [Fe/], and [Ba/] against [/H]. The fact that [/H] keeps increasing significantly after the knee in the Scl dSph demonstrates that even in this system which has no significant intermediate-age population, there was still ongoing star formation contributing enrichment from massive stars well after SNIa started contributing. This is also confirmed by the presence of stars, which have [Fe/H] > -1.8 (the knee), and were therefore formed after SNIa started exploding. In Fnx or Sgr, the very flat, extended and high [Fe/] plateau also shows that massive stars have kept feeding the chemical enrichment all along the evolution, even though they do not dominate the Fe enrichment. As for the s-process, the widely different behaviour of Scl, Fnx and Sgr and the MW are even more striking viewed in this representation than they were in Fig. 13, illustrating the total disconnect of massive stars nucleosynthesis to Ba, and the strong influence of AGB stars at a time when massive stars do not drive the metallicity evolution anymore.

Figure 15. Trends of iron and neutron-capture elements as a function of (PUT IN THE ALPHA SYMBOL HERE) elements in the same four dSphs as in Fig. 11, compared to the MW. Sgr (red: Sbordone et al. 2007, Monaco et al. 2005, McWilliam & Smecker-Hane 2005), Fnx (blue: Letarte 2007, Shetrone et al. 2003), Scl (green: Hill et al. in prep; Shetrone et al. 2003, Geisler et al. 2005) and Carina (magenta: Koch et al. 2008a, Shetrone et al. 2003). Open symbols refer to single-slit spectroscopy measurements, while filled circles refer to multi-object spectroscopy. The small black symbols are a compilation of the MW disk and halo star abundances, from Venn et al. (2004a).

4.2. Ultra-Faint dwarf galaxies

Individual stars in the uFds that have recently been discovered around the MW have so far been little observed at high spectral resolution. This is probably due to the difficulty in confirming membership for the brighter stars in these systems. However, several groups are currently following up confirmed members (typically selected from lower resolution Ca II triplet observations) to derive abundances. So far, Koch et al. (2008b) have observed two RGB stars in Herc (M_V ~ -6.6) and Frebel et al. (2009) are following up RGB stars in the even fainter uFds UMa II and Coma (both with, M_V ~ -4.). The latter study confirms that uFds do contain very metal-poor stars, [Fe/H] < -3, (as found by Kirby et al. 2008), unlike the more luminous "classical" dSphs (Helmi et al. 2006). It also appears that these uFds extend the metallicity-luminosity relation down to the lower luminosities (Simon & Geha 2007), see Section 3.3.

The two stars in Herc seem to have particularly peculiar abundance patterns, with high Mg and O abundances (hydrostatic burning in massive stars), normal Ca, Ti abundances (explosive nucleosynthesis in massive stars), and exceedingly unenriched in Ba (Koch et al. 2008b). On the other hand, elemental ratios in the extremely metal-poor stars in the two fainter dwarfs UMa II and Coma (Frebel et al. 2009) are remarkably similar to the MW halo extremely metal-poor stars. Fig. 16 compares Mg and Ba measurements in faint dSphs, with all more luminous dSph stars that have [Fe/H] -2 (including a new sample of 6 very metal-poor stars in Sextans by Aoki et al. 2009) and the MW. In fact, only Sextans seems to have scattered and low [Mg/Fe] ratios, while other dSphs and uFds all show similar [Mg/Fe] enhancements. The similarity between stars with metallicities below [Fe/H] -2.5 in the MW and faint dwarfs is seen also in other light elements, such as Na, Sc, Cr, Mn, Ni or Zn. This may also be true of more luminous dSphs, see Section 4.1.

Figure 16. Mg ( element) and Ba (s-process element) abundances in dSphs, uFds and the Galactic halo. The magenta symbols are abundances of stars in uFds as measured by Frebel et al. (2009), for three stars in UMa II (squares) and Coma (circles) and by Koch et al. (2008b) (triangles) in Herc. These are compared to the trends derived from Fig. 13 for Scl, Fnx and Sgr as well as individual stellar abundances for all very metal-poor stars ([Fe/H] < -2) in dSphs (Fulbright, Rich & Castro 2004, Sadakane et al. 2004, Venn et al. 2004a, Koch et al. 2008a, Aoki et al. 2009), black open squares), and the MW from the compilation by Venn et al. (2004a) and complemented by Cayrel et al. (2004), François et al. (2007). The dSph trends were derived by a simple 10 points running average on the data for each dSph galaxy with a sufficient statistics (more than 20 measurements).

The overall similarity between all the most metal-poor stars for element ratios up to the iron-peak can be taken as an indication that star formation and metal-enrichment, even at the earliest times, and even in the smallest systems, has proceeded in a similar manner. This may lead to the net yield of the very first stars. The very low dispersion found in abundance ratios of these elements in Galactic extremely metal-poor stars (EMPS), down to metallicities of [Fe/H] ~ -4 came as a surprise (Cayrel et al. 2004): since it was thought that one or a few SN II were sufficient to enrich the gas to those metallicities, the expectation was that among EMPS the variety of metal-production sites (SN II of different masses) would appear as dispersed abundance ratios. We are now adding to this puzzle the fact that these well defined abundance ratios are also achieved by considerably smaller halos.

The only discrepancy among the most metal-poor stars concerns the r-process element Ba, that stands out below the MW halo distribution both for faint and somewhat more luminous dSph galaxies. The most extreme low Ba abundances are found so far in Herc (Koch et al. 2008b) and Draco (Fulbright, Rich & Castro 2004), where only upper limits were detected.

4.3. Dwarf irregulars

The dIs are all (except the SMC) located at rather large distances from the MW and so far, the only probes that could be used to derive chemical abundances in these objects were HII regions and a few super-giant stars. Both types of probes allow a look-back time of at most a few 10 Myr, and this is the end-point of a Hubble-time's worth of chemical evolution for any galaxy. This limitation makes it difficult to gather relevant information to constrain the chemical enrichment over time in these systems. However, abundances in HII regions and super-giants (see references in Table 1) are useful to understand how dIs fit in the general picture of dwarf galaxies, and how they compare to larger late-type galaxies. First, they give the present day metallicity of these systems, and all are more metal-poor than the MW disk young population, in agreement with the metallicity-luminosity relation (see for example van Zee & Haynes 2006 for a relation based on dIs within 5 Mpc), and range between 12 + log(O/H) ~ 8.1 (e.g., NGC 6822, IC 1613) to 12 + log(O/H) ~ 7.30 (Leo A), or [O/H] ~ -0.6 to -1.4. Both HII regions and super-giants typically agree on the oxygen abundances of the systems, within the respective measurement uncertainties (Venn et al. 2003, Kaufer et al. 2004), with little metallicity dispersion within a galaxy, and no spatial gradient (e.g., Kobulnicky & Skillman 1997, van Zee, Skillman & Haynes 2006, van Zee & Haynes 2006). This holds even in the most metal-poor galaxies, and suggests a very efficient mix of metals across the galaxy despite the clumpiness of the ISM and ongoing star-formation. The shear within these systems is expected to be very low, and this has been taken as an indication that mixing occurs in the gaseous hot phase, before the gas cools down to form new stars (e.g., van Zee, Skillman & Haynes 2006).

A to M type super-giants have a further interest as they provide the present-day [/Fe] ratios in dIs, which are not accessible from HII regions where typically only light elements (e.g., He, N, O, Ne, S, Ar) can be measured, and and no iron (nor any other element that would trace SNIa).

The first dI where abundances of stars were measured was of course the SMC in our backyard. The largest samples to date with abundances in SMC are of super-giants which can be found in Hill, Barbuy & Spite (1997, K-type stars), Luck et al. (1998 F-type stars) or Venn (1999 A-type stars). Similar studies in more distant dIs needed efficient spectrographs on 8-10m telescopes, and at the expense of observing for many hours a few stars detailed abundances have been observed in A-type super-giants out to distances of 1.3 Mpc. This work has been pioneered by K. Venn and collaborators using A-type stars (Venn et al. 2001, 2003, Kaufer et al. 2004) in NGC 6822, Sextans A and WLM. There has also been a more recent study using M-type stars in IC 1613 by Tautvaisiene et al. (2007).

Fig. 17 illustrates the observed low [/Fe] in these systems, and compares them to the observed trends of older populations (RGB stars) in dSph galaxies, as defined in Figs. 16 & 11. These low [/Fe] are expected in galaxies that have formed stars over a long period of time, however; they clearly occur at much lower metallicities than in larger systems such as the MW or the LMC, pointing towards an inefficient metal-enrichment of the galaxy (low star formation and/or metal-losses through winds). It is interesting to see in Fig. 17 how dIs actually prolong the trends of dSph galaxies, not only for elements but also for neutron-capture elements. From these diagnostics, dSphs are entirely consistent with being dIs that lost their gas at a late stage of their evolution. The Fnx dSph and the SMC, which are both dominated by intermediate-age populations, are also quite similar in their chemical enrichment, except that Fnx ran out of gas (or lost its gas) and stopped star formation about 10⁸yr ago.

Figure 17. Mg ( element) and Ba (s-process element) abundances in individual super-giants in the dI galaxies Sex A (Kaufer et al. 2004), SMC (Venn 1999, Hill, Barbuy & Spite 1997, Luck et al. 1998), NGC 6822 (Venn et al. 2001) and WLM (Venn et al. 2003), compared to the trends derived from Fig. 13 for Scl, Fnx and Sgr (see Fig. 16 for details on the trends). For the SMC, the points are in fact the mean and dispersion of the samples studied in the references given (~ 6-10 A, K and Cepheids super-giants respectively).