5. HOW IMPORTANT ARE INTERACTIONS?

The interacting galaxies of Arp's 1966 catalogue of peculiar galaxies or Arp and Madore's 1987 catalogue of southern peculiar galaxies are convincingly, some of the most impressive objects in the sky, both from a scientific as well as an aesthetic point of view. Galaxy interactions cover a wide range of interaction strength, including minor disturbances due to relatively distant flybys or the presence of smaller companions, strong distortions (e.g. tidal bridges and tails) due to close encounters, larger systems cannibalizing smaller ones, or mergers of systems of similar mass. Interactions have also been implicated to explain the progression of activity in AGN, from the most distant energetic quasars, through radio galaxies and seyferts, to nearby systems with relatively little nuclear activity, i.e. AGN power was enhanced in the past when the galaxian environment was richer. That interactions are indeed taking place is firmly established, both from an observational as well as a theoretical point of view. What is less clear is the frequency with which these events are occurring now and in the past, whether interactions must be factored into a scheme of ``normal'' galaxy evolution, or indeed whether they might even be the dominant driver of evolution. Put another way, to what extent are galaxian properties governed by initial conditions (nature), and to what extent are they driven by environmental effects (nurture)?

A well-established observational effect related to the nature/nurture question is the morphology-density (M-D) relation. The M-D relation states that, both within and outside of clusters of galaxies, the fractional abundance of elliptical and S0 galaxies with respect to spirals increases with increasing space density of galaxies. Also, a giant cD galaxy is sometimes present at the dynamical center of some clusters. The M-D relation, in some form or other, has been known for at least a century (see Roberts and Haynes 1994), but has been quantified more recently over a wide range of density regimes by Dressler 1980 and by Postman and Geller 1984.

Possible explanations for the M-D relation have been summarized by Whitmore 1990. ``Nature'' explanations include a) angular momentum in the protocloud, i.e. the higher angular momentum material in the outer parts of the cloud will have a slower rate of star formation and therefore form disk galaxies after the initial spheroidal component is formed towards the cluster center, and b) strength of the initial density enhancement, i.e. denser material towards the center of the protocluster would form stars at a faster rate during collapse, resulting in more elliptical galaxies towards the cluster center. The ``nurture'', possibilities include a) tidal ``shaking'', in which the internal distributions of galaxies passing through the higher density cluster core are shaken and rearranged resulting in an earlier type galaxy, b) tidal stripping, in which interactions induce stripping of the gas and stars which then fall towards the cluster center, possibly building up a central cD galaxy there, c) ram pressure stripping and gas evaporation, whereby galaxies passing near the center of a cluster are swept by the denser intracluster gas, quenching star formation and producing more S0s at the expense of normal spirals, d) merging of disk galaxies to form ellipticals in the cluster center or e) cannibalism by the central cD galaxy of the numerous galaxies around it. There are various nature/nurture hybrid models as well. The M-D example is presented, not so as to discuss the pros and cons of the various models, but to illustrate that various processes can lead to the same observational result. Unless we can identify specific signatures which can differentiate between these processes, the nature/nurture conundrum will continue to challenge us. For more discussion, see papers in Thuan et al. 1992.

5.2 Interaction-Driven Galaxy Formation

Some of the more interesting work in recent years has focused on environmental effects (nurture) in shaping the morphology of galaxies. In particular, it is becoming apparent that strong interactions and mergers can, in a sense, create galaxies, either by tidally tearing off pieces of the victim to form independent dwarf galaxies, or by tidally randomizing the stellar fields of two merging disk galaxies to create a giant elliptical.

The idea that a tidal remnant or clump resulting from an interaction might eventually develop into an independent dwarf galaxy was originally suggested by Zwicky (1956) and has enjoyed renewed attention lately due to the increasing sophistication of computers and N-body codes. Barnes and Hernquist 1992b, for example, have demonstrated that self-gravitating ``clumps'' with masses up to several x 10⁸ M can form in tidal tails. The theoretical results have received observational support from the detection of massive optical clumps of up to 8 x 10⁹ M (e.g. in NGC 4038 / 9, Mirabel et al. 1992) and also of neutral hydrogen clumps of about 10⁸ M (e.g. IC 2163 / NGC 2207, Elmegreen et al. 1993), which have recently been discovered in tidal tails. Mirabel et al. have shown that the integrated properties of the optical clumps resemble those of dwarf irregular galaxies.

More dramatically, the evidence is mounting that two disk galaxies can merge to produce an elliptical galaxy. Theoretical simulations have shown that elliptical-like remnants result from disk-disk collisions and, observationally, galaxies in various stages of merging also suggest this (see sequence of illustrations in Barnes and Hernquist 1992a). A good example of a late stage merger which is well on its way to becoming a ``classical'' elliptical is the so-called ``atoms for peace'' galaxy, NGC 7252. This galaxy resembles an elliptical at its center in short exposures, but displays highly disturbed morphology and tidal tails in the outer regions which are visible in deeper exposures (Schweizer 1982). Other evidence that this is an elliptical in the making includes, a) chaotic motions observed in the main body, b) shells and ripples similar to those observed in other ellipticals, c) the azimuthally averaged radial brightness profile described by an R^(1/4) law, d) the galaxy falling within the scatter of the Faber-Jackson relation but outside of the scatter of the Tully-Fisher relation, and e) mechanisms which appear to be at work to rid the optical galaxy of its neutral gas (Hibbard et al. 1994). N-body simulations have also successfully reproduced the morphology and velocity field of NGC 7252 as the product of two colliding disk galaxies (Borne and Richstone 1991; Mihos et al. 1993).

Over the past 10 years, the evidence in favour of ellipticals forming from merging spirals has been increasing (e.g. see Schweizer 1986). Some have also argued (e.g. Efstathiou 1990) that few galaxies were not shaped in some way by galaxy interactions. There have been some lingering problems with this scenario, however. For example, in the merging scenario, one would expect the specific frequency of globular clusters in both spirals and ellipticals to be similar, whereas in fact, the specific frequency in ellipticals is sometimes much higher (e.g. M87). The fact that there are as many red (and therefore, presumably, early type) field galaxies at redshifts of about 1 as there are today (Lilly, private communication) tends to argue against substantial merger-driven evolution over this time frame. Also, the systematic trend for early type galaxies in rich clusters to be bluer with increasing redshift (e.g. Aragón-Salamanca et al. 1992) argues for a single burst of star formation at high z, followed by steady, passive evolution. Some of these problems seem less compelling, however, in the light of new HST results which appear to suggest the formation of globular clusters during encounters (e.g. NGC 1275) and that spectacular galaxy-galaxy interactions are occuring, even at redshifts of 1. Therefore, the evidence continues to mount that interactions can be significant drivers of galaxy formation and evolution. Whether interactions are the dominant drivers overall and at what redshifts they are most important are still open questions. For further discussion of the importance of interactions, see Ellis (1993).

5.3 Subtle Interactions

The above considerations assume, of course, that we know which galaxies are interacting and which are not. Various indicators have been used in the past to identify the interacting galaxies, virtually always based on some optical criteria, for example, the existence of tails and/or bridges, or other ``disturbed'' or ``peculiar'' morphology. Typically, catalogues such as the Atlas and Catalogue of Interacting Galaxies (Verontsov-Velyaminov 1959, 1977), or the Atlas of Peculiar Galaxies (Arp 1966) are used, and/or samples gleaned from visual inspection. Such studies will only select the strongest, most obvious interactions. In other attempts to include weaker interactions, some investigators have studied samples of galaxies with specific peculiar properties not based on optical morphology, like nuclear starbursts or Seyfert activity, and searched for correlations with the presence of companions. For example, there is good evidence that nuclear starbursts are fuelled by infalling gas from the surrounding interstellar medium as a result of an external tidal disturbance. For reviews of interaction-driven nuclear activity, see Heckman (1990) or papers in Shlosman (1994).

The question could be raised, then, as to whether an interaction might be strong enough to significantly affect the evolutionary history of the galaxy, yet not so strong as to produce obvious morphological peculiarities. As an example, I refer again to the NGC 5775 / 4 pair. By virtue of their apparent proximity, it might be supposed that these galaxies would be interacting, and it is now clear from HI observations (see Fig. 4) that this is indeed the case (Irwin 1994). However, no obvious optical distortions are visible in either galaxy (but note the very faint optical bridges between the galaxies on the Palomar Observatory Sky Survey blue print), no particularly strong nuclear activity (starburst or AGN) is present, and the system had not previously been identified as ``interacting'' in the literature. Yet a significant amount of gas (possibly up to 10 M yr^-1, though this is rather uncertain) may be transferring onto NGC 5775 from its companion. Over a typical interaction timescale, this is certainly sufficient to affect the star formation, and hence the evolution, of both galaxies. Enhanced overall star formation at the present epoch is indeed implied from the high IR luminosity of NGC 5775.

What is the best way of confirming that an interaction might be taking place if the optical morphologies are relatively normal? One possibility is IR brightness, as implied above. For example, enhanced IR luminosity is known to be correlated with a high incidence of optical distortions (and by implication with interactions) in more distant IRAS galaxies (Rowan-Robinson 1991 and references therein). Also, enhanced nuclear IR emission might be expected in interactions (even those which do not result in gas transfer between galaxies) because of the starbursting which can be triggered there. However, given that there may be reasons other than interactions as to why IR emission is enhanced (for example, star formation induced by nuclear outflow, a contribution to IR luminosity from sources not related to star formation, or a higher abundance of dust in the ISM), some other direct evidence, such as the HI connection between NGC 5775 / 4 will probably be required. Note that neutral hydrogen imaging, though time consuming, is probably the best way to directly confirm that an interaction is occurring. Not only is the necessary velocity information obtained, but this component appears to be the most sensitive tracer of bridges, tails, or other tidally induced distortions. This is because the gaseous component can diverge from the stellar component due to shocks (see Barnes and Hernquist 1992a) and also because the original HI distribution in the galaxy is the most extensive to begin with (see Section 3.1). For example, for NGC 5775 / 4, the ratio of emission in a typical intergalactic region to the peak brightness of NGC 5775 is about 1/1000 in the optical, 1/100 for the radio continuum, and 1/10 for HI. Other good examples of interactions which are most obvious in HI are the Leo Triplet (see Haynes et al. 1984) and the M82 group (see Scoville et al. 1994).