7.2 Evolutionary Scenarios and Connections
Like all other galaxies, dwarf galaxies evolve: the extremely gas poor dEs must have originally contained gas to form their observed stellar populations. Moreover, many BCGs have unsustainable star formation rates and therefore represent a transient stage, unless they form stars with a very different IMF. Thus, evolutionary connections between different dwarf types must exist, unless some initial conditions, determined the future evolution of different types of galaxies. Even if links exist, there might be several different, physically distinct channels in the evolution of dwarfs. Such a discussion necessarily becomes speculative, and we invite the readers to make their own judgement.
Several evolutionary scenarios that link different types of dwarfs have been discussed over the years. One can think of two different types of evolution, either internal or passive where the evolution of a galaxy proceeds, according to its physical (initial) conditions or through external effects such as mergers, or interactions with galaxies or intergalactic matter. In the latter case, the environment will be a key parameter. Possibly, cluster and field galaxies evolve in a similar manner, but the clock runs faster in a high density environment. Since dEs are found in clusters or as companions to field giants, their evolution is likely related to their environment.
Lin and Faber (1983) suggested that the dSph satellites of the Milky Way were dIs that had lost their gas by ram pressure stripping (see also van den Bergh 1994). It was later pointed out that this could not be the general explanation since, on average, dEs have higher surface brightness than dIs at a given luminosity (Bothun et al. 1986, Ferguson and Binggeli 1994). However, this is mainly a problem for relatively luminous dEs (Skillman and Bender 1995). Searle et al. (1973) and Thuan (1985) proposed that BCGs originate in low surface brightness dIs, an idea that was further elaborated by Davies and Phillipps (1988) where they suggested that dIs evolve into dEs after a number of bursts in the BCG stage. The star formation is regulated by continuous gas infall from the halo, but this model includes no physics to explain the suggested behaviour. One can also imagine a cyclic BCG-dE scenario where a starburst in a BCG gives rise to a superwind, which expells the gas and halts star formation. The expelled gas later cools and falls back on the galaxy, creating a new starburst. Silk et al. (1987) proposed that the BCG phenomenon could be explained by gas expelled from dwarfs at high redshift, now accreting on dwarf ellipticals (see also Babul and Rees 1992). However, this can hardly work for the field BCGs since there are seemingly very few dEs to accrete onto. Gas loss through supernovae driven winds has been a popular mechanism for forming dEs (Larson 1974, Vader 1986, Dekel and Silk 1986), but although outflows are observed in some nearby dwarfs (Sect. 3.3) they do not appear capable of clearing a galaxy of its ISM. Galaxies that have managed to retain gas until the present epoch are unlikely to loose it now, and become dEs (Ferrara and Tolstoy 1999). Skillman and Bender (1995) point out that the majority of Local Group dEs/dSph formed most of their stars in an initial burst, while dIs have had more extended SFHs. However, there has been quite some progress in recent years in unveiling the SFHs of Local Group dwarfs, and the distinction between SFHs of dIs and dEs has been somewhat blurred (Sect. 4.1, 4.2; Grebel 1998). Despite being gas poor, most Local Group dEs have significant young or intermediate populations, and in active SF phases they would appear as BCGs or dIs depending on the extent of these episodes. The best example is probably Carina (Smecker-Hane et al. 1994) which went through a major SF episode some Gyrs ago, demonstrating that dI-dE transitions may have occurred fairly recently. A thorough discussion on the the origin of dEs is given by Ferguson and Binggeli (1994), see also Skillman and Bender (1995). Meurer et al. (1992) argued that BCGs like NCG 1705 could evolve into nucleated dEs. Similarly, it has been suggested that compact galaxies at higher redshifts (z = 0.5 to 1) are the progenitors of the present day dEs (Koo et al. 1995). Since outflows seem rather inefficient, the only way for field BCGs to evolve into dEs (of some kind) is through total gas consumption, which requires very high star formation efficiencies. Field dEs appear too rare, and field Es typically too massive to be the successors of most BCGs, but it may apply for the most massive BCGs, especially those seemingly involved in mergers.
Many dIs appear more simple and able to form stars at more or less continuous rates over long time scales. Although the details of their star formation activity are not completely understood, there is less need for evolutionary connections since their star formation activity is typically sustainable over a Hubble time or more. Given the arguments above, dIs are not likely to passively evolve into dEs (unless perhaps on timescales comparable to the age of the Universe or longer). However there might be connections to dEs (especially in clusters) and BCGs (some of which may fade into dIs). Given their large number and richness in gas, LSBGs may serve as important fuel for star formation.
Constraints on BCG evolution from photometric structure: BCGs have on average higher central surface brightnesses than dIs, even after the starburst component has been subtracted (Papaderos et al. 1996b, Marlowe et al. 1999), which argues against links between BCGs and dEs/dIs, unless the structure can change during the bursting phase. (Note however that this may be an artefact due to insuffiently deep data, see Sect. 4.4.1). Papaderos et al. (1996b) propose that in the initial phase a BCG may contract because of accretion, while the successors of BCGs may expand due to gas loss. However the required expansion/contraction necessary to explain structural differences would require at least 50% of mass loss/gain (Marlowe et al. 1999) which is quite unrealistic given the apparent inefficiency of winds, the observed gas mass fractions and large DM content of dIs. Marlowe et al. (1999) also note that the burst, as defined from excess surface brightness over a disc fit, is quite modest, and that similar excess is seen in some dIs (although the underlying surface brightness is lower). Then BCGs may simply be regarded as unusually active dIs, because of higher central mass densities (van Zee 1998c). In the non burst phase, they will be classified as a high surface brightness dIs or as amorphous galaxies or BCGs with low emission line equivalent widths. This may be the case for many BCGs, especially those close to the dI relation in Fig. 10, but hardly for the XBCGs discussed below.
Constraints from luminosity, metallicity, and gas content: Some BCGs are much more metal-poor at a given luminosity than the dIs forming the basis of the metallicity-luminosity diagram. We called such objects ``extreme BCGs'' (XBCGs). Thus, under the conservative assumption that the current burst has not yet affected the observed nebular abundances, a dI would need to brighten by 3 magnitudes to become an XBCG. That amount of brightening is not observed from profile fitting and is unrealistic in view of the observed H I masses as the following example illustrates:
Imagine a typical dI with M / LB = 2 for its stellar population, on which we add a 10 Myr old burst. The burst should have M / LB > 0.05 for a Scalo IMF (Bruzual and Charlot 2000). To accomplish a brightening with 3 magnitudes, the burst has to be 15 times brighter than the underlying galaxy, and hence make up more than one third of the total stellar mass. This would require, for realistic star formation efficiencies of say 10%, that the dI precursor had a neutral hydrogen mass several times larger than its total stellar mass. The observed gas mass fractions are much lower, and therefore XBCGs cannot originate in bursting dIs (see also Bergvall et al. 1998). On the other hand, a LSBGs turning on a burst would need to increase its absolute luminosity with only ~ 1.5 magnitudes to meet the location of XBCGs in the MB - Z diagram. This would simultaneously reproduce the MHI / LB values for BCGs (Bergvall et al. 1998). Thus LSBGs are more likely to be the precursors of XBCGs than dIs. Moreover, Telles and Terlevich (1997) found the colours of the underlying component in BCGs to be consistent with those of blue LSBGs. This argument against dI-BCG evolution does however not apply to BCGs which are found close to, or above, the metallicity-luminosity relation for dIs. These galaxies must definitely have a different history from the XBCGs.
Mass vs. metallicity: Another indication that dIs like those in the Local Group are not the progenitors of the most metal-poor BCGs comes from a comparison with total mass, rather than absolute blue magnitude. Data for BCGs are scarce, but taking IZw18 (van Zee et al. 1998a), SBS 0335-052 (Papaderos et al. 1998) and luminous BCG from the Östlin (1999a, b) sample and comparing them to Local Group dIs (data from Mateo 1998) it is clear that XBCGs appear to be an order of magnitude more massive than dIs at a given metallicity. Thus the scatter in the metallicity-luminosity diagram is not primarily due to various degrees of star formation affecting the B-luminosity. Interestingly, metal-poor LSBGs (like ESO146-IG14, Bergvall and Rönnback 1995; and some galaxies in the sample of de Blok et al. 1996 and McGaugh 1994) fall on the XBCG mass-metallicity relation. However, three BCGs studied by van Zee et al. (1998) and NGC 1705 (Meurer et al. 1998) fall in the dI range. Quiescent dwarfs (dI/LSBG) from van Zee et al. (1997) overlap with BCGs. The BCGs overlapping with dIs are of the regular type II. This comparison indicates that different types may indeed have a different origin, but we caution that this excursion into studying the mass versus metallicity was based on inhomogeneous data from various sources, using various methods. Thus, it should not be taken at face value, but as a motivation for further studies.
Dwarf mergers? Several investigators have noted the wide range of morphologies displayed by BCGs (see Sect 4.4.1). Telles et al. (1997) note that luminous BCGs (Type I) on average have a more perturbed morphology and Telles and Terlevich (1997) speculate that this may be related to mergers or interactions, but they note that merging is disfavoured, based on the clustering properties of BCGs. As we discussed in Sect. 6.3 this conclusion might be too pessimistic since the present investigations may be incomplete for dwarfs, especially LSBGs. Taylor et al. (1993, 1995, 1996a, 1996b) investigated the H I environment of BCGs and LSBGs finding that 60% and 30% of them respectively have H I companions with faint optical counterparts, with a typical detection limit of 107 M. From the higher fraction of H I companions around BCG, Taylor (1997) argue that LSBGs are not the likely progenitors of BCGs. However, this may be a selection bias in the sense that the more frequent are the H I companions, the more probable mergers will be.
From a kinematical/morphological study Östlin et al. (1999a, b) found that a sample of luminous BCGs (most of them XBCGs) were likely the product of mergers, involving gas rich dwarfs or H I-clouds. A merger can provide the necessary change of the structure of the host and offers a mean for changing the kinematics of galaxies, from rotating (like dIs) to essentially non-rotating (like E and dE). A critical point, especially if some BCGs are to evolve into E/dE, is how efficient the gas consumption may be? Galaxies with a large gaseous disc at moderate column densities are not likely to be able to consume this gas on a short timescale. In mergers, the star formation efficiency might be higher, especially if globular clusters are able to form (Goodwin 1997). Van Zee et al. (1998c) point out that BCGs have higher central H I densities than normal dIs. This is a natural explanation for the high star formation rates, but some mechanism is needed to put the gas where it is, since the gas consumption time scale is rather short. If gas consumption is less efficient, perhaps because gas settles into a rotating disc, the merger remnant would rather evolve into a dI or an amorphous galaxy. If a merging gas rich dwarf contains a rather unpolluted HI halo, this could dilute the metal enriched ISM and lower the observed abundances, to produce XBCGs. Clearly, mergers would be able to explain many properties of BCGs, but of course this does not imply that merging is the general mechanism, especially concerning those with regular morphology. The absence of morphological perturbations in Type II BCGs may however be an observational effect. Since these systems are on the average fainter, their underlying components have low surface brightnesses, and thus morphological irregularities such as tails will have low surface brightnesses and be difficult to detect.
We have fast outlined some ideas about evolutionary connections. It is clear that the overall picture is not yet well understood. There is probably no unified scheme that can explain all dwarfs. The origin and evolution of dE/dSph galaxies is probably related to their environment, and it cannot be ruled out that some dE/dSphs are stripped dIs, although dIs are not likely to passively evolve into dEs. Many BCG-like galaxies may simply be extreme dIs. Some BCGs, especially the XBCGs, may be triggered by mergers involving gas-rich galaxies such as LSBGs. Of course if some BCGs are truly young, their progenitors must be pure gas clouds and not visible as galaxies. Such gas clouds have however not been found, and although the existence of genuinely young BCGs at the current epoch cannot be ruled out, we do not consider that there is any BCG where there is an unambiguous evidence for pure youth (cf. Sect. 4.4.2 and 5).