Ever since Hubble (1936) published his famous `Sequence of Nebular Types' (a.k.a. tuning-fork diagram) the question has been: What determines the position of a galaxy along this sequence? And why are galaxies at one end of the sequence disk-shaped and at the other end ellipsoidal? Was this shape dichotomy imprinted during an early collapse phase of galaxies, or did it arise through subsequent evolution?
Work begun several decades ago by Zwicky (1956), Arp (1966), Alladin (1965), and Toomre & Toomre (1972, hereafter `TT'), among others, has led to growing evidence that gravitational interactions between neighbor galaxies do not only explain some of the most striking `bridges' and `tails' observed in disturbed galaxy pairs, but also tend to lead to galactic mergers that often trigger bursts of star formation and clearly represent important phases of galaxy building (Larson 1990; Barnes & Hernquist 1992; Kennicutt et al. 1998).
Before reviewing some of the evidence for interactions and mergers being a significant driver of galaxy evolution, it seems wise to agree on some terminology and point out biases.
To be called a `merger', a galaxy pair or single galaxy should show at least clear morphological signatures of an advanced tidal interaction, such as significant distortions, major tails, and ripples or `shells' (for a review, see Schweizer 1998). A stronger case for merging can usually be made when kinematic signatures are available as well, such as opposite tail motions, counter-rotating parts, or tail material falling back onto a remnant. As figure 1 illustrates, much recent progress in this area is due to the upgraded Very Large Array's ability to map the line-of-sight motions of neutral hydrogen (H I) in tidal features in great detail (Hibbard 1999).
Figure 1. Neutral hydrogen distribution and kinematics of NGC 4038 / 4039. Left: H I contour lines (white) superposed on an optical photograph; right: H I position-velocity plot, with declination along y-axis and line-of-sight velocity along x-axis (from Hibbard 1999).
The main bias in studies of gravitational interactions has been toward major mergers, which involve two galaxies of nearly equal mass. Such mergers are highly destructive and tend to lead to spectacular morphologies, whence they can be observed from the local universe out to redshifts of z 2 and beyond. Minor mergers involving galaxies with mass ratios of, say, m/M = 0.1-0.5 are less spectacular and often require verification via some kinematic signature (esp. in the remnant phase). Hence, such interactions and mergers have been studied mainly in nearby galaxies and out to z 0.5. Finally, although satellite accretions leading to mass increases of a few percent or less may be relatively frequent, they are the most difficult to detect and have been studied only in the Local Group, and even there nearly exclusively in our Milky-Way galaxy. Thus, our knowledge of growth through accretions and minor mergers is severely limited.
Because of its dissipative nature gas plays a disproportionately large role in galaxy interactions. Even at the present epoch, the vast majority of galaxies contain significant amounts of cold gas (Roberts & Haynes 1994). During tidal interactions and mergers this gas tends to be driven toward the centers of galaxies through gravitational torques exerted on it by tidally induced stellar bars (e.g. Barnes & Hernquist 1996). The ensuing shocks and energy dissipation allow the gas to get compressed, leading to intense bursts of star formation, globular-cluster formation, and the feeding of nuclear activity. Starbursts and active galactic nuclei in turn drive galactic winds and jets, which can have profound effects on the chemical evolution of galaxies (Heckman 2000).
Some of these processes can now be reproduced by modern N-body simulations that include gas hydrodynamics. Barnes (1999) shows a beautiful sequence of two gas-rich disk galaxies merging. Whereas their stars end up in a three-dimensional pile not unlike an elliptical galaxy with considerable fine structure, more than half of the cold gas from the input disks gets funneled to the center of the remnant into a region only about 0.5 kpc in diameter, while the initially warm gas (T 104 K) gets heated to X-ray temperatures (~ 106 K) and forms a pressure-supported atmosphere of similar dimensions as the stellar pile. The time scale for this transformation from two disk galaxies to one merged remnant is remarkably short: about 1.5 rotation periods of the input disks or, when scaled to component galaxies of Milky-Way size, about 400 Myr.
The rapidity of this equal-mass merger is due to strong dynamical friction. We should keep this in mind when trying to understand the formation of elliptical galaxies in dense environments. Claims have been made that cluster ellipticals formed in a rapid monolithic collapse because their present-day colors are rather uniform. Yet, experts agree that age differences of 3 Gyr cannot be discerned from broad-band colors of galaxies 10-15 Gyr old. A time interval of 3 Gyr may seem short when we struggle with logarithmic age estimates, yet it is long when compared to the merger time scale. About eight major mergers of the kind simulated by Barnes could take place one after another during this time interval, and 12 Gyr later all their remnants would look nearly the same color and age. Hence, claims about monolithic collapses and a single epoch of elliptical formation are to be taken with a grain of salt. There was time for many major mergers of juvenile disks during the first few Gyr after the big bang, and most cluster ellipticals could have formed through such mergers without us knowing it from their present-day colors.
The following review of evidence for interactions being a driver of galaxy evolution begins with accretions in the Local Group, continues with minor mergers and the damage they inflict on disk galaxies, moves on to major mergers forming ellipticals from wrecked disks, and ends with a brief description of what we have learned from first glimpses of high-redshift mergers.