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

Next Contents


What is a dwarf galaxy? Past definitions always focus on size (e.g., Hodge 1971, Tammann 1994), and the presence of a dark matter halo (e.g., Mateo 1998). Is there any other physical property that distinguishes a dwarf galaxy from bigger galaxies? Are the differences merely due to the amount of baryonic matter that is retained by a system during its evolution? In general, large late-type galaxies sit on the constant central surface brightness ridge defined by Freeman (1970), and appear to have managed to retain most of the baryons they started with. Conversely for galaxies which lie below this limit it seems that the fainter they are, the higher the fraction of baryons they have lost. This could be due to Supernova winds and/or tidal interactions, which are effective when a galaxy lacks a suitably deep potential well to be able to hold onto to its gas and/or metals. The galaxies above this central surface brightness limit are either currently forming stars very actively, such as Blue Compact Dwarfs (BCDs), or they have had very active star formation activity in the past (e.g., Elliptical galaxies).

The definition of a galaxy as a dark matter halo naturally excludes globular clusters, which are believed not to contain any dark matter, and also do not contain complex stellar populations or any evidence of enrichment. The structural properties of globular clusters (see Fig. 1) tend to support the idea that they are distinct from galaxies. This definition also excludes tidal dwarfs, and indeed we do not consider them here as they are more a probe of the disruption of large systems. They are a different category of objects that formed much later than the epoch of galaxy formation. There are also no obvious nearby examples of tidal dwarfs where the resolved stellar population can be accurately studied.

Figure 1

Figure 1. Here are plotted the relationships between structural properties for different types of galaxies (after Kormendy 1985, Binggeli 1994, Kormendy et al. 2008), including as dotted lines the classical limits of the dwarf galaxy class as defined by Tammann (1994). In the upper plot we show the absolute magnitude, MV, vs. central surface brightness, µV, plane, and in the lower plot the MV vs. half light radius, r1/2, plane. Marked with coloured ellipses are the typical locations of Elliptical galaxies & bulges (red), spiral galaxy disks (blue), galactic nuclei (dashed magenta) and large early (spheroidals) and late-type systems (dashed black). Galactic globular clusters are plotted individually as small black points. M 31, the Milky Way (MW), M 33 and LMC are shown as blue open triangles. Some of the BCDs with well studied CMDs are marked as blue solid squares. The peculiar globular clusters omega Cen and NGC 2419 are marked close to the globular cluster ellipse; M 32 in the region of Elliptical galaxies; the SMC near the border of the dwarf class. The Ultra-compact dwarfs (UCDs) studied in the Virgo and Fornax clusters are marked with magenta crosses. Local Group dwarf galaxies are plotted as open pentagons, blue for systems with gas and green for systems without gas. The recently discovered uFds are given star symbols, and the same colour code. For references and discussion see text.

Here we aim to build upon the outstanding review of Mateo (1998) and leave behind the idea that dwarf galaxies are in any way special systems. Many galactic properties (e.g., potential well, metallicity, size) correlate with mass and luminosity, and all types of galaxies show continuous relations in structural, kinematic and population features between the biggest and the smallest of their kind (e.g., see Fig. 1). Part of our aim in this review is to investigate these trends and learn from them. The only justification to segregate dwarf galaxies from other types is to study specific aspects of galaxy formation and evolution on a small scale.

The taxonomy of dwarf galaxies typically opens a Pandora's box. At a very influential conference held at the Observatoire de Haute-Provence in 1993 G. Tammann gave a working definition: all galaxies that are fainter than MB leq -16 (MV leq -17) and more spatially extended than globular clusters (see dotted lines in Fig. 1) are dwarf galaxies (Tammann 1994). This is broadly consistent with the limit of mass at which outflows tend to significantly affect the baryonic mass of a galaxy. This includes a number of different types: early-type dwarf spheroidals (dSphs); late-type star-forming dwarf irregulars (dIs); the recently discovered very low surface brightness, ultra-faint, dwarfs (uFd); centrally concentrated actively star-forming BCDs. The new class of even more extreme ultra-compact dwarfs (UCDs) are identified as dwarf galaxies form spectra but are of a similar compactness to globular clusters (see Fig. 1).

As has been stated throughout the years (Kormendy et al. 2008 and references therein), a morphological classification is only useful if it incorporates a physical understanding of the processes involved. However, at present this understanding is not complete and hence structural parameters and their relations may give us clues to the underlying physics. But we also have to be careful not to over-interpret these global measures, especially when we cross over from structurally simple to more complex systems (e.g., from early type spheroidals to late type disk-halo star-forming systems). This requires care to establish a meaningful comparison of the same properties of such different systems. This has most commonly been done using basic parameters such as surface brightness and absolute magnitude and physical size of the systems. In Fig. 1 we show these familiar relations. These kinds of plots were first made by Kormendy (1985), and have been used to great effect by Binggeli (1994) and more recently by Belokurov et al. (2007b).

Fig. 1 illustrates how dwarfs compare with all other galaxies with no real evidence of a discontinuity, as already noted by [Kormendy 1985]. From a comparison of the absolute magnitude (MV) and central surface brightness (µV) of galaxies (upper plot in Fig. 1), the early and late-type dwarfs (from Irwin & Hatzidimitriou 1995, Mateo 1998, Whiting, Hau & Irwin 1999, Hunter & Elmegreen 2006) appear to fall along similar relations, overlapping with BCDs and other larger late-type systems (from Hunter & Elmegreen 2006) as well as faint spiral galaxy disks and those galaxies defined as spheroidals by Kormendy et al. (2008). This means systems which resemble late-type galaxies in their structural properties but are no longer forming stars. The uFds are clearly separated but arguably follow the same relation as the other dwarfs (from Simon & Geha 2007, Martin, de Jong & Rix 2008). There are clear distinctions in Fig. 1 between elliptical galaxies (from Faber et al. 1997, Kormendy et al. 2008) and other types, with the exception of spiral galaxy bulges. Similarly there are also clear distinctions between Globular clusters (from [Harris 1996]) and any other type of galaxy, with the exception of galactic nuclei. The position of M 32 in the Elliptical galaxy region is consistent with it being a low-luminosity Elliptical galaxy (e.g., Wirth & Gallagher 1984) and not a dwarf galaxy, or even a tidally stripped larger system. There is evidence that omega Cen, with its clear spread in Main Sequence Turn-offs (MSTOs), Red Giant Branch (RGB) sequences and chemical abundances, maybe be the stripped central remnant of an early-type system (e.g., Lee et al. 1999, Pancino et al. 2000, Bekki & Freeman 2003), and its position in Fig. 1 is consistent with that of galactic nuclei. It is interesting to note that omega Cen and nuclei also lie in the same region as the UCDs (from Evstigneeva et al. 2008).

The Magellanic Clouds move in and out of the dwarf galaxy class, which is not surprising as at least the Small Magellanic Cloud (SMC) lies near the boundary of the luminosity definition of dwarf class (see Fig. 1). The fact that the Magellanic Clouds are interacting with each other and our Galaxy makes it more difficult to determine their intrinsic properties. The Large Magellanic Cloud (LMC) appears to be similar to low luminosity spiral galaxies, such as M 33, in terms of mass, luminosity and size. The SMC on the other hand more resembles the larger dIs in the Local Group (e.g., NGC 6822, IC 1613), with similar mass, luminosity and metallicity of star-forming regions.

In the lower plot of Fig. 1 the varying physical size scales of different galaxy types and globular clusters are shown by plotting MV against the half-light radius r1/2, after Belokurov et al. (2007b). In this plot there is a clear (and unsurprising) trend for more luminous galaxies to be larger. The Ellipticals clearly fall on a distinct narrow sequence (which is a projection of the fundamental plane). Dwarf galaxies, i.e., BCDs, late-type and spheroidal galaxies fall along a similar, although offset, tilted and more scattered relation to the Elliptical galaxies. "Classical" Local Group dSphs clearly overlap with Irregular and BCD types. The uFds appear in a somewhat offset position. This is perhaps due to difficulties in accurately measuring the size of such diffuse objects, or it may be a real difference with other dwarf galaxies.

From Fig. 1 it can be seen that there is no clear separation between dwarf galaxies and the larger late-type and spheroidal systems. The dIs, BCDs, dSphs, late-type and spheroidal galaxies tend to overlap with each other in this parameter space. The overlapping properties of early and late-type dwarfs has long been shown as convincing evidence that early-type dwarfs are the same as late-type systems that have been stripped of their gas (Kormendy 1985). This is quite different from the distinction between Ellipticals and Spirals (and Spheroidals), which show a more fundamental difference (Kormendy et al. 2008). There is no clear break which distinguishes a dwarf from a larger galaxy, and hence the most simple definition does not have an obvious physical meaning, as recognised by Tammann (1994).

Whatever the precise definition of sub-classes, dwarf galaxies cover a large range of size, surface brightness and distance, and so they are usually studied with different techniques with varying sensitivity to detail. Some galaxies are just easier to study than others (due to distance, size, concentration, location in the sky, heliocentric velocity etc.). This also leads to biases in understanding the full distribution of properties of a complete sample (e.g., Koposov et al. 2008). Because of this the properties and inter-relations of the various types of dwarf galaxies are not always easy to understand.

The classic dichotomy is between early and late-type dwarf galaxies. It is not easy to compare the properties of dwarf galaxies which have on-going star formation (e.g., dIs, BCDs), with those that do not (e.g., dSphs). Indeed, the properties which can be measured, and then compared, are often different from one type of galaxy to another. The dSphs do not contain gas and so their internal kinematics can only be determined from stellar velocity dispersions. In gas rich dIs, on the other hand, the internal kinematics can be easily determined from the gas and their distance makes them challenging targets to determine stellar velocities from RGB stars. Likewise abundances in dIs are typically [O/H] measurements in young HII regions whereas in dSphs they are usually [Fe/H] coming from individual red giant branch (RGB) stars over a range of ages.

Galaxies in which the individual stars can be resolved are those which can be studied in the greatest detail. These are primarily to be found in the Local Group, where individual stars can be resolved and photometered down to the oldest main sequence turnoffs (MSTOs). This provides the most accurate star formation histories (SFHs) going back to the earliest times. In the Local Group spectra can be taken of individual RGB stars at high and intermediate resolution, providing information about the chemical content as well as the kinematics of a stellar population. The most accurate studies of resolved stellar populations have been made in Local Group dwarf galaxies which are the numerically dominant constituent (e.g., Mateo 1998).

Historically the first dwarfs to be noticed in the Local Group, leaving out the Magellanic Clouds which are clearly visible to the naked eye, were the early-type dwarf satellites of M 31. M 32 and NGC 205 (M 110) were first catalogued by C. Messier in 1770, NGC 185 by W. Herschel in 1787 and NGC 147 by J. Herschel in 1829. The spatially extended but low surface brightness dIs were first noticed somewhat later, e.g., NGC 6822 (1881, by E.E. Barnard), IC 1613 and WLM (early 1900s, by M. Wolf). In all cases these galaxies were catalogued as "faint nebulae". It was not until the discovery (by H. Leavitt in 1912) and application to NGC 6822 (by E. Hubble 1926) of the Cepheid distance scale that they were realised to be (dwarf) extra-galactic systems. In 1938 H. Shapley discovered the first low surface brightness dwarf spheroidal galaxies, Sculptor (Scl) and Fornax (Fnx), around the MW. From the 1930s onwards extensive observing campaigns led to the compilation of large catalogs of dwarf galaxies extending beyond the Local Group, such as Zwicky's catalogs and the Uppsala General Catalog (UGC) initiated by E. Holmberg.

Over the last 50 years there has been a steady stream of new discoveries of dwarf galaxies in the Local Group, and also in other nearby groups and clusters. The Local Group discovery rate has dramatically increased recently thanks to the Sloan Digital Sky Survey (SDSS; e.g., Adelman-McCarthy et al.2007) and a new class of uFds (e.g., Willman et al. 2005, Zucker et al. 2006b, Belokurov et al. 2007b) has been found around the MW. However there remains some uncertainty about the true nature of these systems.

Thus dwarf galaxies provide an overview of galaxy evolution in miniature which will also be relevant to understand the early years of their larger cousins and important physical processes which govern star formation and its impact on the surrounding interstellar medium. There remain issues over the inter-relations between different types of dwarf galaxies, and what (if any) is the connection with globular clusters. When these relations are better understood we will be a significant step closer to understanding the formation and evolution of all galaxies.

Next Contents