In the field environment, interactions and mergers of galaxies are thought to be a prime candidate for driving evolution in galaxy populations. Major mergers of spiral galaxies may lead to the transformation of spirals into ellipticals (e.g., Toomre 1978; Schweizer 1982): the violent relaxation associated with a merger effectively destroys the galactic disks and creates a kinematically hot, r1/4-law spheroid, while the concurrent intense burst of star formation may process the cold ISM of a spiral galaxy into the hot X-ray halo of an elliptical. These processes are not totally efficient, however, and leave signatures behind that identify the violence of the merging process: diffuse loops and shells of starlight, extended H I gas, significant rotation in the outskirts of the remnant, and dynamically distinct cores (see, e.g., the review by Schweizer 1998). While it is an open question as to what fraction of ellipticals formed this way, it is clear that at least some nearby field ellipticals - Centaurus A being a notable example (Schiminovich et al. 1994) - have had such violent histories.
Mergers have also been proposed as a mechanism to drive the formation of S0 galaxies. In this case, the scenarios are varied. For S0s with very large bulge-to-disk ratios, reaccretion of gas after a major merger (either from returning tidal material or from the surrounding environment) may rebuild a disk inside a newly formed spheroid. In unequal-mass mergers of disk galaxies, disk destruction is not complete, and the resulting remnant retains a significant amount of rotation (Bendo & Barnes 2000; Cretton et al. 2001) and may be identified with a disky S0, particularly if a significant amount of cold gas is retained by the system to reform a thin disk (Bekki 1998). Finally, minor mergers between spirals and their satellite companions can significantly heat, but not destroy, galactic disks (Toth & Ostriker 1992; Quinn, Hernquist, & Fullagar 1993; Walker, Mihos, & Hernquist 1996), while simultaneously helping to "sweep the disk clean" of cold gas via a gravitationally induced bar driving gas to the nucleus (Hernquist & Mihos 1995). The resulting disks have many similarities to disky S0s (Mihos et al. 1995): thickened disks, little or no spiral structure, cold gas, or ongoing star formation. These different scenarios vary mainly in the proposed strength of the interaction - from major to minor mergers - and it has been proposed that this parameter may, in fact, determine the ultimate morphological classification of galaxies all the way from early-type ellipticals to late-type spirals (e.g., Schweizer 1998; Steinmetz & Navarro 2002).
Applying these arguments to cluster populations, it seems that building cluster ellipticals through a wholesale merging of spirals within the established cluster environment is a difficult proposition. Clusters ellipticals are an old, homogeneous population showing little evolution since at least a redshift of z 1 (e.g., Dressler et al. 1997; Ellis et al. 1997). Within the cores of massive clusters, merging has largely shut off due to the high velocity dispersion of the virialized cluster (Ghigna et al. 1998). The accretion of merger-spawned ellipticals from infalling groups may still occur, and these will be hard to identify morphologically as merger remnants - the combination of cluster tides and hot ICM will strip off any tell-tale tidal debris and sweep clean any diffuse cold gas in the tidal tails (recall Fig. 4) or low-density reaccreting disk. However, the small scatter in the color-magnitude relation and weak evolution of the fundamental plane of cluster ellipticals (see, e.g., the review by van Dokkum 2002) argues that such lately formed ellipticals likely do not contribute to the bulk of the cluster elliptical population. This does not mean that mergers have not played a role in the formation of cluster ellipticals. In any hierarchical model for structure formation, galaxies form via the accretion of smaller objects. Luminous cluster ellipticals may well have formed from mergers of galaxies at high redshift, in the previrialized environment of the protocluster. However, at these redshifts (z >> 1), the progenitor galaxies are likely to have looked very different from the present-day spiral population.
Unlike the rather passive evolution observed in cluster ellipticals, much stronger evolution is observed in the population of cluster S0s. The fraction of S0s in rich cluster has increased significantly since a redshift of z 1, with a corresponding decrease in the spiral fraction (Dressler et al. 1997). Can the same collisional processes that have been hypothesized to drive S0 formation in the field account for the dramatic evolution in cluster S0 populations? S0 formation scenarios that rely on reaccretion of material after a major merger (disk rebuilding schemes) seem difficult to envision wholly within the cluster environment. While mergers are possible in infalling groups, the combination of tidal and ram pressure stripping will shut down reaccretion and ablate any low-density gaseous disks that have survived the merger process. For example, it is unlikely the H I disk in Centaurus A (Nicholson et al. 1992), likely a product of merger accretion, would survive passage through the hot ICM of a dense cluster. Satellite merger mechanisms trade one dynamical problem for another - because the mergers involve bound satellite populations there is no concern about the efficacy of high-speed mergers, but, instead, the issue is whether or not satellite populations can stay bound to their host galaxy as it moves through the cluster potential. And, of course, this mechanism relies on the very local environment of galaxies, which does not explain why S0 formation would be enhanced in clusters.
None of the proposed merger-driven S0 formation mechanisms appear to work well deep inside the cluster potential. On the other hand, these processes should operate efficiently in the group environment, where the encounter velocities are smaller and cluster tides and the hot ICM do not play havoc with tidal reaccretion. The group environment may create S0s and feed them into the accreting cluster, but if there is wholesale transformation of cluster spirals into S0s in the cluster environment, it needs to occur via other mechanisms.
Other cluster-specific methods for making S0 galaxies have been proposed, including collisional heating and ram pressure stripping of the dense ISM (Moore et al. 1999; Quilis, Moore, & Bower 2000) and strangulation, the stripping of hot halo gas from spirals (Larson et al. 1980; Bekki, Couch, & Shioya 2002). While these models, by design, explain the preferential link between clusters and S0 galaxies, they are not without problems themselves. While the effects of ram pressure stripping on the extended neutral hydrogen gas in cluster galaxies is clear (e.g., van Gorkom, this volume), its efficacy on the denser molecular gas is unclear. For example, the H I deficient spirals in the Virgo cluster still contain significant quantities of molecular gas (e.g., Kenney & Young 1989), while studies of the molecular content of cluster spirals show no deficit of CO emission (Casoli et al. 1998). If the molecular ISM survives, it is unclear why star formation should not continue in these disks. Strangulation models suffer less from concerns of the efficacy of ram pressure stripping, since it is much easier to strip low-density halo gas than a dense molecular ISM, although it must be noted that there currently is little observational evidence for hot halos in (non-starbursting) spiral galaxies. In addition, neither of these methods leads to the production of a luminous spheroid - the S0s that might be produced in these ways would have low bulge-to-disk ratios.
Ultimately, S0s are a heterogeneous class, from bulge-dominated S0s to the disky S0s seen in galaxy clusters, and it should not be surprising that a single mechanism cannot fully account for the range of S0 types (e.g., Hinz, Rix, & Bernstein 2001). Whether there is a systematic difference between cluster and field S0s is unclear, an issue fraught with selection and classification uncertainties. What is clear is that, even in clusters, S0s often show evidence for accretion events, similar to that observed in the field S0 population (see, e.g., the discussion in Schweizer 1998). It is likely that many of these S0s were "processed" via mergers in the group environment before being incorporated into clusters.