5.1. Stellar halos of disk galaxies
Spiral galaxies usually have extended, low-density stellar halos (Zibetti et al. 2004) whose density decreases with ~ r-3. In the Milky Way the stellar halo consists of old, metal-poor stars and globular clusters on eccentric prograde or retrograde orbits (see Freeman & Bland-Hawthorn 2002 and Helmi 2008 for recent reviews of the Galactic halo). As many of half of the field stars in the halo may have originated in disrupted globular clusters (Martell & Grebel 2010; see Odenkirchen et al. 2001a for an example).
CDM simulations suggest that in galaxies with few recent mergers the fraction of halo stars formed in situ amounts to 20% to 50% (Zolotov et al. 2009). Johnston et al. (2008) propose that halos dominated by very early accretion show higher [/Fe] ratios, whereas those that accreted mainly high-luminosity satellites should exhibit higher [Fe/H].
The detection of substructure (e.g., Newberg et al. 2002; Yanny et al. 2003; Bell et al. 2008) as well as chemical and kinematic signatures (Carollo et al. 2007; Geisler et al. 2007) support the scenario that part of the Galactic halo was accreted. The individual stellar element abundance patterns suggest that such accretion may have mainly occurred at very early times, since the [/Fe] ratios of metal-poor halo stars match the ones found in similar stars in the Galactic dwarf spheroidal and irregular companions (e.g., Koch et al. 2008a). Prominent morphological evidence of ongoing dwarf galaxy accretion has been found not only in the Milky Way (e.g., Ibata et al. 1994), but also around other spiral galaxies (e.g., Ibata et al. 2001; Zucker et al. 2004, Martínez-Delgado et al. 2008).
5.2. Bulges of disk galaxies and formation scenarios
Early-type spirals show prominent bulges, which become less pronounced and ultimately vanish in late-type spirals and irregulars. Classical bulges (found in early-type to Sbc spirals) resemble elliptical galaxies in their properties, are dominated by old, mainly metal-rich stars with a large metallicity spread, show hot stellar kinematics, and follow a de Vaucouleurs surface brightness profile just like typical elliptical galaxies. Pseudobulges (in disk galaxies later than Sbc) resemble disk galaxies, have similar exponential profiles, are rotation-dominated, and may contain a nuclear bar, ring, or spiral. They are believed to form from disk material via secular evolution (Kormendy & Kennicutt 2004). A recent analysis combining data from SST and GALEX found that all bulges show some amount of ongoing star formation, regardless of their type (Fisher et al. 2009), with small bulges having formed 10 to 30% of their mass in the past 1 to 2 Gyr (Thomas & Davies 2006). Extracting a sample of > 3000 nearby edge-on disk galaxies from the SDSS, Kautsch et al. (2006) showed that approximately 30% of the edge-one galaxies are bulge-less disks.
Noguchi (1999) suggested that massive clumps forming at early times in galactic disks move towards the galactic center due to dynamical friction, merge, and form the galactic bulge. This scenario leads to the observed trend of increased bulge-to-disk ratios with increased total galactic masses. van den Bergh (2002) noted that while most of the galaxies observed in the Hubble Deep Fields at z < 1 have disk-like morphologies, most galaxies at z > 2 look clumpy or chaotic. Analyzing such "clump clusters" and "chain galaxies" in the Hubble Ultra Deep Field, Elmegreen et al. (2009) find that the masses of the star-forming clumps are of the order to 107 to 108 M. Bournaud et al. (2007) argue that clusters of such massive, kpc-sized clumps can form bulges in less than 1 Gyr, while the system as a whole evolves from a violently unstable disk into a regular spiral with an exponential or double exponential disk profile on a similarly rapid time scale. While the coalescence of these clumps resembles a major merger with respect to orbital mixing, the resulting bulge has no specific dark-matter component, which distinguishes it from bulges formed via galaxy mergers (Elmegreen et al. 2008).
5.3. Disks and long-term evolutionary trends for disk galaxies
Disks are the primary sites of present-day star formation in spiral galaxies, and it seems likely that they have continued to form stars for a Hubble time. Disks show ordered rotation, and their stars move around the galactic center on near-circular orbits. The rotational velocities greatly exceed the velocity dispersion by factors of 20 or more.
Gas-deficient early-type disk galaxies show little activity at the present time, while gas-rich late-type disks experience wide-spread, active star formation. Star formation occurs mainly in the midplane of the thin disks, in particular along spiral arms, where recent events are impressively traced by giant H II regions. Spiral density waves may induce star formation (see Martínez-García et al. 2009; and references therein), although it has been suggested that this mechanism may contribute less than 50% to the overall star formation rate (Elmegreen & Elmegreen 1986).
In the Milky Way, the star formation in the disk was not constant, but shows extended episodes of increased and reduced activity (e.g., Rocha-Pinto et al. 2000), a radial metallicity gradient, a G-dwarf problem, and a large metallicity scatter at all ages (Nordström et al. 2004). The thin disk is embedded in a lower-density, kinematically hotter stellar population consisting of older, more metal-poor stars - the thick disk (Gilmore & Reid 1983; Bensby et al. 2005). The chemical similarity of Galactic bulge and thick disk stars might suggest that the Milky Way does not have a classical bulge (Meléndez et al. 2008).
Dalcanton & Bernstein (2002) showed that thick disks are ubiquitous also in bulge-less late-type disk galaxies, which indicates that their formation is a universal property of disk formation independent from the formation of a bulge. A variety of mechanisms for the formation of thick disks has been proposed, including formation from accreted satellites, gas-rich mergers, heating of an early thin disk by mergers, heating via star formation processes, and radial migration (e.g., Wyse et al. 2006; Brook et al. 2004; Kroupa 2002; Bournaud et al. 2009; Roskar et al.2008). Sales et al. (2009) suggest that the eccentricity distribution of thick disk stars may permit one to distinguish between these scenarios.
In cosmological simulations disk galaxies may form, for example, via major, wet mergers (see, e.g., Barnes 2002; Governato et al. 2009) or without mergers via inside-out and vertical collapse in a growing dark matter halo (Samland & Gerhard 2003).
As noted by van den Bergh (2002) based on an analysis of the Hubble Deep Fields, roughly one third of the objects at z > 2 seem to be experiencing mergers. He suggests that from 1 < z < 2 a transition from merger-dominated to disk-dominated star formation occurred. Moreover, he finds that at z > 0.5, there are fewer and fewer barred spirals. While early-type galaxies assume their customary morphologies relatively early on, 46% of the spirals at 0.6 < z < 0.8 are still peculiar, and with higher redshift, the spiral arm patterns become increasingly chaotic. Also within the class of spiral galaxies there are trends: Only ~ 5% of the Sa and Sab galaxies are peculiar at z ~ 0.7, while almost 70% of the Sbc and Sc types are still peculiar.
Elmegreen et al. (2007) suggest that the formation epoch of clumpy disk galaxies may extend up to z ~ 5. The ones experiencing major mergers may form red spheroidals at 2 z 3, whereas the others evolve into spirals. Elmegreen et al. propose that the the star formation activity in clumpy disks is caused by gravitational collapse of portions of the disk gas without requiring an external trigger.
Regarding environment, Poggianti et al. (2009) find that the fraction of ellipticals remains essentially constant below z = 1, while the spiral and S0 fractions continue to evolve, showing the most pronounced evolution in low-mass galaxy clusters. They attribute this to secular evolution and to environmental mechanisms that are more effective in low-mass environments. At low redshifts, the declining spiral fraction with density is driven by late-type spirals (Sc and later; Poggianti et al. 2008).
Irregular galaxies are gas-rich, low-mass, metal-poor galaxies without spiral density waves, which show recent or ongoing star formation that appears to have extended over a Hubble time (Hunter 1997). Many studies found the H I gas to be considerably more extended than the stellar component in irregulars (e.g., Young & Lo 1997), but more recent, deep optical surveys show that the optical extent of at least some of these galaxies has been underestimated (e.g., Kniazev et al. 2009). All nearby irregulars and dwarf irregulars have been found to contain old populations, although their fractions differ (Grebel & Gallagher 2004). The old populations tend to be more extended than the more recent star formation (e.g., Minniti & Zijlstra 1996; Kniazev et al. 2009) and show a more regular distribution (e.g., Zaritsky et al. 2000; van der Marel 2001).
Irregulars are usually found in the outskirts of groups and clusters or in the field, thus interactions with other galaxies are likely to be rare. Their star formation appears to be largely governed by internal processes and seems to be stochastic. Rather than experiencing brief, intense starbursts, irregulars typically show extended episodes of star formation interrupted by short quiescent periods - so-called gasping star formation (e.g., Cignoni & Tosi 2010). The long-term star formation amplitude variations amount to factors of 2 to 3 (Tosi et al. 1991). For a review of irregulars and dwarf irregulars in the Local Group, for which we have the most detailed data to date, see Grebel (2004).
While the more massive irregulars are rotationally supported and show solid-body rotation, low-mass dwarf irregulars are dominated by random motions. Star formation ceases at lower gas density thresholds than in spirals (e.g., Parodi & Binggeli 2003), and the global gas density of the highly porous interstellar medium has been found to lie below the Toomre criterion for star formation (van Zee et al. 1997). Turbulence may create local densities exceeding the star formation threshold (e.g., Stanimirovic et al. 1999). Low-mass dwarf irregulars without measurable rotation show less centrally concentrated star formation and have lower star formation rates (Roye & Hunter 2000; Parodi & Binggeli 2003). In most irregulars, star formation occurs within the galaxies' Holmberg radius and within three disk scale lengths (Hunter & Elmegreen 2004).
In contrast, in blue compact dwarf (BCD) galaxies the highest star formation rates are found, and star formation occurs mainly in the central regions (Hunter & Elmegreen). (We do not discuss BCDs and other gas-rich dwarfs in more detail here. For an overview of different dwarf types and their properties, see Grebel 2003).
Once believed to be chemically homogeneous, there is now evidence of metallicity variations at a given age in several irregulars (e.g., Kniazev et al. 2005; Glatt et al. 2008). This suggests that local processes dominate the enrichment and that mixing is not very efficient. Irregulars follow a fairly well-defined metallicity-luminosity relation, which however is offset from that of early-type dwarfs covering the same luminosity range. Surprisingly, the offset is such that the continuously star-forming irregulars and dwarf irregulars have lower metallicities at a given luminosity than the inactive early-type dwarfs (e.g., Richer et al. 1998), a discrepancy that holds even when comparing stellar populations of the same age (Grebel et al. 2003). Taken at face value, this may imply that the enrichment of irregulars was less efficient and slower than that of early-type dwarfs. BCDs and in particular extremely metal-deficient galaxies continue this trend and appear to be too luminous for their present-day, low metallicities even when compared to normal irregulars (Kunth & östlin 2000; Kniazev et al. 2003).
5.5. Star formation "demographics"
Lee et al. (2007) investigate the star formation "demographics" of star-forming galaxies out to 11 Mpc combining H and GALEX UV fluxes. Their sample includes spirals, irregulars, and BCDs. Lee et al. identify three different star formation regimes:
1. Galaxies with maximum rotational velocities Vmax > 120 km s-1, total B-band magnitudes of MB -19, and stellar masses 1010 M are mainly bulge-dominated galaxies with relatively low specific star formation rates and increased scatter in these rates. Also the mass-metallicity relation changes its slope in this regime (Panter et al. 2007), and supernova ejecta can be retained (Dekel & Woo 2003). Bothwell et al. (2009) find that the H I content of these massive galaxies decreases faster than their star formation rates, leading to shorter H I consumption time scales and making the lack of gas a plausible reason for the observed quenching of star formation activity.
2. Galaxies with ~ 120 km s-1 > Vmax > 50 km s-1 and -19 < MB < -15 comprise mainly late-type spirals and massive irregulars. Lee et al. (2007) suggest that spiral structure acts as an important regulatory factor for star formation. They find that the galaxies in this intermediate-mass regime exhibit a comparatively tight, constant relation between star formation rate and luminosity (or rotational velocity). The star formation rates show fluctuations of 2 to 3, and the current star formation activity is about half of its average value in the past. Bothwell et al. (2009) argue that the galaxies in this regime evolve secularly. They show that the star formation rates decrease with the galaxies' H I mass, and that the H I consumption time scales increase with decreasing luminosity.
3. Below Vmax = 50 km s-1 and MB > -15, dwarf galaxies, particularly irregulars, dominate. At these low masses, the star formation rates exhibit much more variability ranging from significantly higher (e.g., in BCDs) to significantly lower (e.g., in so-called transition-type dwarfs with properties in between dwarf irregulars and dwarf spheroidal galaxies, see Grebel et al. 2003) star formation activity than in the higher-mass regimes. Overall, there is a general trend towards lower star formation rates. Stochastic intrinsic processes, feedback, and the ability to retain gas play an important role here. Bothwell et al. (2009) find that for many of the galaxies in the low-mass regime the H I consumption time scale exceeds a Hubble time (in good agreement with the results of Hunter 1997).
Bothwell et al. show that the H I consumption time scales have a minimum duration of more than 100 Myr. They argue that this minimum duration corresponds to the gas mass divided by the minimum gas assembly time, i.e., the free-fall collapse time.