Alan Dressler

Astronomers study variations in galaxy properties with environment because they seek to understand the process of galaxy formation, which, for the vast majority of galaxies, occurred long ago. The dependence of galaxy type on environment, first noted by Edwin Hubble and Milton Humason in 1931, provides an opportunity to learn about events that took place in a largely unobservable past.


Galaxies cluster; although a small fraction live in splendid isolation, the average galaxy is found in a low-density, marginally bound group. The typical density of such a group is only 1 galaxy for a volume of the order of 20 cubic megaparsecs (a sphere with diameter ~ 10 million light-years); therefore, encounters between galaxies are relatively rare. However, the sky is punctuated by rich clusters of several hundred galaxies in regions 100-1000 times as dense as this. These rich clusters, which contain only ~ 5-10% of all galaxies, are surrounded by extensive plateaus of modestly enhanced galaxy density, called superclusters, that reach like vast, lacy bridges to neighboring clusters.

The existence of groups and clusters is itself an important clue to the conditions of galaxy formation. Gravity, which unlike electromagnetism is a one-pole force, relentlessly gathers matter together. After the universe had cooled sufficiently from the Big Bang for the first segments of galaxies to form, fluctuations in density, perhaps arising from quite specific physical processes, were continually amplified by gravity. Only energy released from star formation could have temporarily resisted this process of ``hierarchical clustering'' by which smaller lumps aggregate into larger and larger ones. Models of stellar structure and energy generation show that a star more than 100 times as massive as the Sun would blow itself apart, so it is understood why hierarchical clustering ended with billions of stars per galaxy mass rather than a few supermassive stars. What physics prevented the coalescence of cluster galaxies into a single supergalaxy?

The answer appears to lie in the play-off between the time scale for collapse of the protogalaxy and the cooling time of its largely gaseous mass. Galaxies of the size we see today are the largest objects that could have radiated away significant amounts of their binding energies by dissipation in the gaseous component. As also happens in the birth of a star, dissipation enables a galaxy to contract to a higher density than specified by its initial size and temperature. The densities and temperatures associated with protoclusters, on the other hand, result in cooling times much longer than their dynamical times. Consequently, galaxies retain their individual identities and the effect of clustering is to collect galaxies into high-density regions rather than to build massive supergalaxies. Thus, the existence of galaxy groups and clusters is indicative of the role of gas dynamics and hierarchical clustering in the process of galaxy formation.


The most fundamental property of a galaxy is its mass; however, astronomers usually substitute absolute luminosity because its determination requires only a single photometric measurement and a redshift (as an estimate of the galaxy's distance). (Studies of the internal dynamics of galaxies indicate a range in mass-to-light ratio of about a factor of 5 compared to a range in luminosity of the order of 1000, moderately justifying this substitution.) The number distribution of galaxy luminosities (the luminosity function) is thus used in place of the more fundamental mass distribution.

How does the luminosity function vary with environment? In cataloging thousands of rich clusters, George Abell found little or no variation from cluster to cluster; subsequent investigations determined a similar luminosity function for ``field galaxies,'' those in low-density regions. That galaxies in what are today strikingly different environments could have formed with approximately the same distribution in luminosity (roughly equated to mass) suggests that the process of galaxy formation took place at an early epoch, when environmental differences were considerably smaller. This is an argument that galaxies formed before groups and clusters were well developed.

The second major property of a galaxy is its morphology. Averaged over all environments, spiral and irregular galaxies account for ~ 70% of all galaxies, the balance being elliptical and SO galaxies. In the typical environment (that of loose groups), elliptical and SO galaxies account for only 10-20% of the population. Hubble and Humason, and later (1936) Fritz Zwicky, noted the startling difference that rich clusters are mainly composed of elliptical and SO galaxies. In addition to their unusual population, rich clusters are also notable for their unusual environment: Cluster galaxies exist in a relatively crowded space 1000 times more dense than the locale of a typical galaxy. Moreover, the core volume of a rich cluster is usually permeated by extremely hot plasma with a pressure much higher than that of a galaxy's interstellar gas. It is reasonable to speculate that some property of this extreme cluster environment favored the formation or evolution of less-common galaxy types - a clue to discovering the processes that produced morphological variation.

At such densities, interactions with the plasma and even direct encounters with other cluster members are likely to be important events in the life of a galaxy. In 1951, Walter Baade and Lyman Spitzer suggested that SO galaxies might be formed when spiral galaxies cross in the dense environment of a cluster, stripping each other of their Interstellar gas and losing their ability to form new stars. To be effective, this process requires a relatively rapid encounter (as would be common in a rich cluster) because a low-speed collision would probably result in the merging or disruption of the original spirals. A related idea, published by James E. Gunn and Richard Gott in 1972, held that a rapidly moving spiral would be stripped of its gas by ram pressure from the hot intracluster plasma.

In the mid-1970s Alar Toomre suggested a model for the formation of ellipticals that, like these SO production models, imagined a metamorphosis of spiral galaxies in later life, this time by disruptive, largely nondissipative mergers with other galaxies. However, the prevalence of ellipticals in rich, dense clusters (where the encounter velocities are ~ 1000 km s-1) is somewhat problematical in this model because at these speeds, much greater than the orbital speeds of stars in the galaxies, mergers are less likely than more elastic encounters. This suggests that if mergers were responsible for forming cluster ellipticals, they occurred relatively early, when the rich cluster was a collection of poorer, less-dense groups, each with a lower characteristic speed of member galaxies. Furthermore, the fact that the stellar density at the center of an elliptical galaxy is typically much higher than in a spiral suggests that a significant amount of dissipation must have occurred, indicating that the process occurred at an earlier time when galaxies were more gaseous. Both of these arguments suggest that if mergers were important in forming elliptical and perhaps SO galaxies, they might have occurred so early as to be considered part of the formation process.

Elliptical galaxies are very similar to the bulges of disk galaxies. It is therefore natural to extend the merger picture to suggest that the spheroidal components of disk galaxies also formed by mergers. However, a counterargument holds that the chaotic addition of even a small amount of mass would disrupt the thin disk found in most spiral galaxies. Although a possible aid in explaining the evolution of SO galaxies, the fact that many spirals with relatively large bulges still retain thin disks might be hard to understand unless these dish themselves formed after the bulges. If disk disruption is a serious problem for the model of building bulges by accretion, the importance of mergers in the formation of ellipticals is brought into question as well.

Perhaps the best evidence for an environmentally induced merger process is the unique presence of giant galaxies in clusters. William W. Morgan identified a class that he called cD galaxies which are surrounded by extensive stellar envelopes. Many of these have extraordinary luminosities and masses, well above those found for galaxies in low-density environments. The view is commonly held (and supported by some admittedly disputed evidence) that cD galaxies have cannibalized neighboring galaxies, accreting by nondissipative mergers unfortunate companions that stray into their domain. Here again, however, both observation and theory suggest that this process was more important in the past when clusters were coalescing from smaller associations of galaxies with lower relative speeds.

The remarkably tight dependence of galaxy type on local density shown in Fig. 1 can be interpreted as evidence that local processes, perhaps early in the lifetime of the galaxy, played an important role in determining galaxy type. Over four orders of magnitude of increasing local density, the fraction of spirals decreases steadily as the fraction of SOs and ellipticals increases. This means that most elliptical and SO galaxies exist outside of rich clusters, even though they are more prevalent in rich clusters. A mechanism like ram-pressure stripping by a hot plasma does not then offer a complete model for what makes a galaxy a dormant SO instead of a star-forming spiral, because it accounts for only a small fraction of the population. Low-velocity encounters between spirals, as would be typical in the lower-density regions where most SOs are found, would be more likely to result in a merger than the mere removal of interstellar gas. However, a better description of the relation between the luminous baryonic component of a galaxy and its massive, nonluminous halo is needed to evaluate fully such models.

Figure 1

Figure 1. The morphology-density relation [reproduced from Dressler (1980) Ap. J. 236 351]. The fractions of E, S0, and spiral plus irregular galaxies are shown as functions of the projected local density (in galaxies per square megaparsec). The upper histogram shows the number distribution of galaxies found in these environments, for a sample of over 6000 galaxies in 55 rich clusters. The fraction of spiral galaxies falls steadily for increasing local density, compensated by a corresponding rise in the fraction of elliptical and S0 galaxies. Identical trends have been found to hold for galaxies in poorer groups, but the dependence weakens or disappears in groups where the crossing time is comparable to the age of the universe.

As in the case of the luminosity function, the very slow change in the morphological mix over orders of magnitude in density suggests that differentiation took place at an early epoch before the density contrast was very large. The morphology-density relation could even be evidence for a process that formed galaxies of different types ab initio, with little or no alteration later in the galaxy's life. In this picture, the formation of high-density spheroids, such as the bulges of disk and elliptical galaxies, is the result of more efficient star formation in denser environments. Crosstalk between the early fluctuations that formed galaxies and the much longer wave perturbations that formed clusters would then account for the fact that denser galaxies are more commonly found in denser environments. However, an elliptical or a disk galaxy with a dominant spheroid could form even in a globally low-density region if there had been a small-scale fluctuation of unusually high amplitude. Here again, the nonluminous halo and variations in its properties (e.g., density, angular momentum) with environment could play the key role.

Recent work suggests that the luminosity functions for different galaxy types are different but that each may show little or no variation with environment. If so, it is even more curious that these luminosity functions play off against the morphology-density relation in such a way as to make the summed luminosity function nearly universal. However, it is important to remember that the mass-to-light ratio for spirals is significantly lower than that of elliptical or SO galaxies. Thus, the universality of the luminosity function may be little more than an accident masking a more fundamental dependence of the mass function on environment.


In the 198Os astronomers began observational studies of the dependence of galaxian properties on environment as a function of cosmic time. clusters of galaxies can be recognized at great distances corresponding to look-back times of 5-10 billion years, and the identification of these clusters as the ancestors of present-day clusters allows for a statistical comparison of their populations. It has not yet been possible to determine morphological types for these distant galaxies (due to limitations in image detail imposed by the atmosphere on observations with Earth-based telescopes) and observations with the Hubble Space Telescope are eagerly awaited to correct this deficiency. In the meantime, however, it has been possible through spectrophotometry to detect in distant galaxies the level of star formation, closely related to galaxy type at the present epoch.

The surprising result is that, in contrast to the relatively dormant elliptical and SO galaxies found in today's clusters, rich clusters at z ~ 0.5 contained a much higher fraction of galaxies with vigorous star formation. This implies that the ancestors of some elliptical and SO galaxies (or the component parts that preceded them) were much more active in star formation only a few billion years ago and/or that strong bursts of star formation were once common in the spiral galaxies of these clusters. It is not yet understood to what extent this is related to the cluster environment rather than just a reflection of galaxy evolution with cosmic time. It is certain, however, that adding the dimension of time to the observational data base will help distinguish between alternate models for the evolution of different galaxy types as a function of environment.

Additional Reading
  1. Binggeli, B., Sandage, A., and Tammann, G.A. (1988). The luminosity function of galaxies. Ann. Rev. Astron. Ap. 26 509.
  2. Dressler, A. (1984). The evolution of galaxies in clusters. Ann. Rev. Astron. Ap. 22 185.
  3. Giovanelli, R. and Haynes, M.P. (1986). Morphological segregation in the Perseus-Pisces supercluster. Ap. J. 300 77.
  4. Gott, J.R. (1977). Recent theories of galaxy formation. Ann. Rev. Astron. Ap. 15 235.
  5. Schweizer, F. (1986). Colliding and merging galaxies. Science 231 227.
  6. Strom, S.E. and Strom, K.M. (1979). The evolution of disk galaxies. Scientific American 240 (No. 4) 72.
  7. See also Clusters of Galaxies, Component Galaxy Characteristics; Galaxies, Disk, Evolution; Galaxies, Elliptical, Origin and Evolution; Galaxies, Formation; Intracluster Medium.