2.10.2. Proportion of spiral, S0, and elliptical galaxies
Elliptical and S0 galaxies are more common than spiral galaxies in the inner portions of regular compact clusters, while the opposite is true in irregular clusters and in the field (Table 1). Many explanations have been proposed for the origin of this systematic variation in galactic content. In general, these theories fall into two broad classes. In the first class, the proportion of galaxy types is set by the conditions when the galaxies form, and once the galaxies form they do not alter their morphology. Thus, in regions that are or will become regular clusters, more ellipticals are formed. In the second group of theories, galaxies may form with the same distribution of morphological types everywhere, but physical processes that depend on environment cause galaxies to alter their morphology. That is, in compact regular clusters, spirals become S0s or ellipticals, and S0s become ellipticals. Often an analogy is made between the generation of variations in the character of galaxy populations and human populations; in the first case, galaxy morphology is determined at conception (the 'heredity' hypothesis), while in the second case it is influenced primarily by the 'environment' in later life.
From the point of view of the environment theories, the primary difference between spiral galaxies and S0 or elliptical galaxies is that spirals contain much more gas. As a result, spirals have active star formation, and the gas allows shocks, which delineate the spiral structure. The basic idea behind most of the environment theories, first suggested by Spitzer and Baade (1951), is that spiral galaxies would become S0 galaxies if their gas were removed. A number of mechanisms have been proposed to remove the gas from spiral galaxies in clusters; these are reviewed in detail in Section 5.9.
Spitzer and Baade (1951) suggested that collisions between spiral galaxies in the cores of compact clusters remove the interstellar medium from the disks of these galaxies. Subsequent increases in the estimates of the distance scale to clusters have had the effect of seriously reducing the efficiency of this process (Section 5.9). It certainly could strip some of the spirals in a cluster. However, if the cluster has a significant amount of intracluster gas, there are several other processes which are more efficient.
Gunn and Gott (1972) suggested that S0 galaxies are formed when spiral galaxies lose their interstellar medium through ram pressure ablation on intracluster gas. This process has now been studied fairly extensively, and is reviewed in Section 5.9. In general, these studies indicate that a galaxy will be stripped almost completely during a single passage through the core of a cluster if the intracluster gas density exceeds roughly 3 × 10-4 atoms/ cm3.
Another mechanism for removing gas from galaxies, which can operate even when the galaxies are moving slowly through the intracluster gas, is evaporation (Cowie and Songaila, 1977; Sections 5.9 and 5.4.2). Heat is conducted into the cooler galactic gas from the hotter intracluster gas, and if the rate of heat conduction exceeds the cooling rate, the galactic gas will heat up and flow out of the galaxy (equation 5.118). The evaporation rate can be significantly reduced if the conductivity saturates (Section 5.4.2), or because of the magnetic field (Section 5.4.3). If the conductivity is not suppressed by these effects, this mechanism can play an important role in stripping gas from galaxies.
The X-ray observations that are the primary subject of this book have shown that many clusters have intracluster gas densities high enough to make ram pressure stripping effective (Section 5.9; equation 5.115). In addition, there is a strong inverse correlation between the X-ray luminosity and the fraction of spiral galaxies in a cluster (Bahcall, 1977c; Melnick and Sargent, 1977; Tytler and Vidal, 1979; equation 4.9). The spiral fraction also decreases as the velocity dispersion of the cluster increases, as required for ram pressure ablation, and the spirals that are observed in X-ray clusters are on average at large projected distances from the cluster center (Gregory, 1975; Melnick and Sargent, 1977), where the intracluster gas density is presumably much lower than in the cluster core. Although not directly connected with stripping from spiral galaxies, the X-ray observations of M86 (an elliptical galaxy in the Virgo cluster) suggest that it is currently undergoing ram pressure ablation (Forman et al., 1979; Fabian et al., 1980).
Ram pressure ablation or evaporation removes the gas from cluster galaxies, and this prevents ongoing star formation. Thus one would expect that even when spiral galaxies do occur in a cluster, they will have less gas than field spirals and less active star formation. Studies of the 21 cm radio line of hydrogen from spiral galaxies in clusters indicate that they have considerably less gas than field spirals, and that the amount of gas present increases with distance from the cluster center (Section 3.7). Spirals in clusters also show weaker optical line emission than those in the field, which also suggests they have less gas (Gisler, 1978); but see Stauffer 1983); the same is apparently true of cluster ellipticals and S0s (Davies and Lewis, 1973; Gisler, 1978). Moreover, many of the spirals in clusters have poorly defined spiral arms; they are classed as 'anemic' spirals by van den Bergh (1976). These anemic cluster spirals are intermediate in appearance and in color between field spirals and S0s, which suggests that they have less active star formation than the field spirals. Cluster spirals may also be smaller than field spirals (Peterson et al., 1979). Finally, Gallagher (1978) and Kotanyi et al. (1983) present possible examples of spiral galaxies currently undergoing stripping.
Of course, the most direct test of the hypothesis that spiral galaxies evolve to become elliptical galaxies is the observation of spiral galaxies in high redshift clusters which may be undergoing this transformation. Unfortunately, the image sizes of galaxies in high redshift clusters (with ground based telescopes) are so small that the galaxies cannot be directly classified. However, Butcher and Oemler (1978a, 1984a, b) showed that a number of moderately high redshift (z 0.4) clusters apparently contained a high proportion of blue galaxies. The blue galaxies lie at larger projected distances from the cluster center than the redder galaxies. When redshift effects were removed, these blue galaxies had colors indistinguishable from those of nearby spiral galaxies. No such population of blue galaxies occurs in nearby compact clusters (Butcher and Oemler, 1978b). These blue galaxies in high redshift clusters probably contain substantial quantities of gas and may be undergoing star-formation; they may indeed be spiral galaxies. If so, we would have fairly direct evidence that galactic populations evolve in rich clusters.
There are, however, a number of problems associated with this interpretation of the 'Butcher-Oemler effect'. First, Mathieu and Spinrad (1981) and van den Bergh (1983b) have argued, based on the photometry and positions of the blue galaxies in the Butcher-Oemler cluster about 3C295, that the cluster is actually relatively poor, and that the blue galaxies are primarily background and foreground field galaxies; they claim that the actual cluster members have a more normal color distribution. Redshifts and spectra now exist for a reasonably large sample of galaxies in the Butcher-Oemler clusters (Dressler and Gunn, 1982; Dressler, 1984; Dressler et al., 1985); they show that many of the blue galaxies are cluster members, although many are foreground galaxies. The blue galaxies that are cluster members do have colors which might suggest that they are spirals. However, many of them have spectra and surface brightnesses that are very different from those of present day normal spiral galaxies; they appear to consist of Seyfert-like active galaxies (Henry et al., 1983; Dressler et al., 1985) and galaxies that have recently undergone or are undergoing very large bursts of star formation (Butcher and Oemler, 1984b). There is some evidence from deep optical images that the blue galaxies are indeed disk galaxies (Thompson, 1986). The present rather uncertain situation concerning the Butcher-Oemler effect has been reviewed recently by Dressler (1984). Second, the effect may not be universal; high redshift clusters are known which appear not to contain such a blue population (see, for example, Koo, 1981). Third, X-ray observations show that the Butcher-Oemler clusters contain significant quantities of intracluster gas (Section 4.8). The ram pressure ablation time-scales are thus rather short ( 109 years). It seems unlikely that the clusters formed such a short time before they were observed; how then have the spirals survived? It is possible, given the fact that the blue galaxies lie in the outer parts of the clusters, that they have not passed through the cluster core yet. Another possibility, suggested by Gisler (1979), is that the rate of formation of interstellar gas in galaxies was higher at these redshifts. This could make it considerably more difficult to strip the galaxies through ram pressure ablation. (However, also see Livio et al. (1980).) Finally, the redshifts of the clusters observed by Butcher and Oemler are not that large; why is it that similar clusters are not found at the present time?
Norman and Silk (1979) and Sarazin (1979) have suggested that nearly all the gas in a cluster was initially in the form of gaseous haloes around galaxies. These haloes would be stripped rather slowly by collisions until sufficient gas built up in the cluster core for ram pressure ablation to become effective. In this way, the stripping of spiral galaxies could be delayed for a significant fraction of a Hubble time, in agreement with the Butcher-Oemler effect. Larson et al. (1980) argued that the gaseous disks of galaxies are supplied by the continual infall of gas from the gaseous haloes proposed by Norman and Silk (1979) and Sarazin (1979). In this scenario, a galaxy will be transformed from a spiral into an S0 following the removal of its gaseous halo; star formation will then exhaust the interstellar medium in the disk in a few billion years. The Butcher-Oemler clusters might be in the midst of this transformation.
The stripping mechanisms described above (ram pressure ablation, evaporation, or the Spitzer-Baade effect) remove the gas from galaxies, but probably would not seriously affect the distribution of the stars because the mass fraction of gas in typical spirals is not terribly large (Farouki and Shapiro, 1980). Unfortunately, simply removing the gas from a spiral would leave behind a thin stellar disk; thus a stripped spiral galaxy might resemble an S0 galaxy, but never an elliptical galaxy. As the fraction of elliptical galaxies is higher in the centers of compact clusters than in the field (Table 1), the difference in galactic population between these two environments cannot result solely from the transformation of spirals into S0s. Moreover, S0s appear to have thicker disks than spirals (Burstein, 1979b). If galactic populations have evolved since formation, the stellar distribution in galaxies must have been modified during this evolution.
Of course, the process of gas removal could directly affect the stellar distribution if, during the removal of the gas, a significant portion were converted into stars which remained bound to the galaxy. This might produce the 'thick disk' component seen in S0s by Burstein (1979b).
Tidal gravitational effects during galaxy collisions can alter the stellar and mass distributions in galaxies. The possibility that massive haloes around galaxies might be removed by tidal collisions in clusters has already been discussed in Sections 2.8 and 2.10.1. Such tidal encounters might also puff up the disks of spiral galaxies and transform them into S0 or elliptical galaxies (Richstone, 1976). Unfortunately, detailed numerical simulations suggest that tidal interactions are not capable of transforming disk galaxies into ellipticals unless they do so before the cluster collapses (Da Costa and Knobloch, 1979; Farouki and Shapiro, 1981).
Another possibility is that elliptical galaxies are formed by the merger of spiral galaxies in clusters (Toomre and Toomre, 1972; White, 1979). There are a number of serious problems with this hypothesis, which have been summarized by Ostriker (1980) and Tremaine (1981). Most of these objections disappear if the mergers occur in subclusters before the formation of the cluster (see White, 1982).
Evidence given in support of the "heredity hypothesis" (the theory that galaxy morphology is determined at the time of galaxy formation) is primarily in the form of evidence against the "environment hypothesis". Specifically, most of these arguments attack the extreme suggestion that the differences in galaxy morphology are determined solely by some environmental mechanism which removes the gaseous content of galaxies.
First, the total fraction of disk galaxies (Sp+S0) / (Sp+S0+E) is not fixed, but varies from regular clusters to the field (Faber and Gallagher, 1976; Table 1). If the only change in galaxy morphology were the conversion of spirals into S0s, this ratio would be a constant. Second, some S0 and E galaxies are sometimes found in low density regions (the field), where ram pressure ablation and other environmental influences should be very weak (Sandage and Visvanathan, 1978; Dressler, 1980b), and these field S0s and Es have the same color distribution as the cluster S0s and Es. Third, Dressler (1980b) has found that distribution of galaxy morphologies correlates most strongly with the local galaxy density, and not as strongly with the global environment (which probably determines the density of intracluster gas). Fourth, the properties of S0 galaxies (colors, bulge-to-disk ratios, gaseous content, etc.) are generally intermediate between Sp and E galaxies (Sandage et al., 1970; Faber and Gallagher, 1976). This would not necessarily be true if S0s were stripped Sps, but would be explained under the "heredity hypothesis" if the nature of galaxy formation were characterized by a single parameter, which is often taken to be the density of the region in which galaxy formation occurs (see below). Fifth, S0 galaxies have larger ratios of bulge-to-disk luminosity than spirals (Burstein, 1979a), and also absolutely larger and more luminous bulges (Dressler, 1980b). This would not be expected if S0s were simply spirals with their gas removed. Finally, S0 galaxies appear to have a thick, boxy component to the disk, which is not present in spirals (Burstein, 1979b). Of course, one could argue that these thick disks actually arise during the process of stripping the gaseous disks from spirals; for example, the stripping process might induce star formation in the gas while it is being stripped, and produce a thick disk of stars supported by large velocity components perpendicular to the disk.
If galaxy morphology is determined by the conditions at the time of galaxy formation, what is the mechanism and to what conditions is it responsive? Most theories of galaxy formation assume that galaxies form by the collapse of initially gaseous matter (Eggen et al., 1962; Gott, 1977). The stellar bulge components of galaxies all have distributions similar to the de Vaucouleurs profile (equation 2.14), which suggests that they have relaxed violently (Section 2.9) during the collapse. As discussed in Section 2.9, this implies that the collapse takes place on nearly a free-fall time (equation 2.32). If, prior to this collapse, most of the gas in the galaxy were converted to stars, these stars would act as a collisionless, dissipationless fluid, and only violent relaxation would occur. This would produce an ellipsoidal distribution of stars-that is, an elliptical galaxy. On the other hand, if star formation were ineffective, and most of the collapsing material remained gaseous, it would dissipate through radiation much of its energy, while maintaining its net angular momentum, and collapse to form a disk. With this hypothesis one can understand why the galaxies that contain significant quantities of gas are the disk-dominated spirals, and why the galaxies that lack gas are ellipticals and S0s.
Moreover, if galaxies initially have little gas compared to their stellar content, it is easier for them to remain free of gas. First, the stripping processes discussed above (ram pressure ablation, evaporation, etc.) are more effective if the density of interstellar gas is low. Second, even in the absence of such external mechanisms for removing gas, a galaxy with a high ratio of the density of stars to gas can clean itself of interstellar medium through the formation of a galactic wind (Mathews and Baker, 1971). The interstellar gas in a galaxy is heated by the stars, through supernovae, stellar winds, ionizing radiation, and the motion of mass-losing stars at high velocity through the ambient gas. If the gas density is sufficiently high compared to the stellar density, the gas will be able to cool efficiently and the energy input from stars will be radiated away. Conversely, if the gas density is low compared to the stellar density, the gas will heat up until thermal velocities in the gas exceed the escape velocity from the galaxy, and the gas leaves the galaxy in a transonic wind. Thus, if a galaxy starts with a small proportion of gas to stars, it can keep itself free of gas. Since the standard hypothesis is that E and S0 galaxies start with little gas, one can understand how they have managed to stay relatively gas free.
If it is the efficiency of star formation during the collapse of a protogalaxy which determines its morphology, what determines the efficiency of star formation? Why does it depend on the location of the protogalaxy? Two attempts to answer these questions have received particular attention. First, Sandage, Freeman, and Stokes (1970) argued that the efficiency of star formation in a protogalaxy is determined by the specific angular momentum content of the gas. If the angular momentum content of the gas is high, the collapse of protostars will be halted or delayed by centrifugal forces. Thus one would expect protogalaxies with a high angular momentum content to form spirals, and those with low angular momentum to form ellipticals. This hypothesis is in agreement with the observation that the specific angular momentum of disks significantly exceeds that of ellipsoidal components of galaxies.
Alternatively, Gott and Thuan (1976) have argued that the efficiency of star formation during the collapse of a galaxy is set by the density in the protogalaxy. Star formation requires that gas cool. Cooling processes generally involve two-body collisions. Therefore it is reasonable to assume that the cooling rate increases with density (see equation 5.23 for example). In protogalaxies with a sufficiently high initial density, the gas will be largely converted to stars during the collapse. If the initial density is sufficiently low, star formation is not effective and the gas collapses to a disk.
In Section 2.9 evidence was presented which indicated that the sequence of cluster morphology, from the field to irregular clusters to regular clusters, was a dynamical sequence resulting from increasing initial density. Thus regular, compact clusters formed from regions of high density, and irregular clusters from lower density regions. If density is also the factor that determines galaxy morphology, the relationship of galaxy morphology and cluster morphology can be understood.
In summary, it remains controversial whether galaxy morphology is determined primarily by conditions at the time of galaxy formation, or whether galaxy morphology evolves in response to the environment after formation. It seems unlikely that all galaxies have identical forms at birth, and that all the variation in galaxy morphology is due to environment. It seems reasonable that environment has played some role in determining galaxy morphology; surely, somewhere at least one spiral galaxy has blundered into a core of a rich, compact cluster and been stripped. Thus I believe both mechanisms have probably significantly affected galaxy morphology. Note that two other possibilities further obscure the distinction between these two hypotheses. First, galaxies may have formed before clusters; then, the galaxies might have evolved in an environment different from that observed today (i.e., Roos and Norman, 1979). Second, the formation of disks in galaxies might be a slow and ongoing process (Larson et al., 1980); then, there is no real distinction between the heredity and environment hypotheses, at least to the origin of disks.
Ultimately, the Hubble Space Telescope (Hall, 1982) will permit structural studies of galaxies in and out of clusters at large redshifts. These studies will show whether morphological evolution has occurred in galaxies, at least over the last half of the age of the universe.