Another possible mechanism for removing material - gas and stars - from galaxies is the interaction between the galaxies (e.g. Clemens et al. 2000, Mihos et al. 2005, see Figs. 4, 5 and 6). While the direct stripping effect is mostly not very efficient in clusters due to the short interaction times, the close passage of another galaxy (sometimes called galaxy harassment) can trigger a star burst (Barnes & Hernquist 1992, Moore et al. 1996, Bekki 1999), which subsequently can lead to a galactic wind (Kapferer et al. 2005). But there can be a competing effect: the ISM might be stripped off immediately by ram pressure (Fujita et al. 1999, Heinz et al. 2003) and hence the star formation rate could drop. In any case ISM would be removed from the galaxies.
Figure 4. Image of the interacting galaxies NGC 4490 / NGC 4485 in H I (contours) and optical R band (grey scale). Some of the gas is lost due to the interaction of the galaxies (from Clemens et al. 2000).
Simulations of interactions between galaxies containing an AGN show a complex interplay between star formation and the activity of the AGN itself (Springel et al. 2005).
Figure 5. Very deep observation of the core of the Virgo cluster. Diffuse light is visible between the galaxies which results from stars that have been expelled from the galaxies due to interactions between them (from Mihos et al. 2005).
In order to estimate how likely such interaction events are the number of encounters and mergers needs to be assessed. The number of close encounters that a galaxy experiences within tH = 1010 years was estimated by Gnedin (2003) in the following way. He assumed a galaxy of size Rg = 10 kpc, a virialised cluster with a one-dimensional velocity dispersion of cl = 1000 km s-1, a virial radius of Rcl = 1 Mpc and Ng = 1000 galaxies uniformly distributed within this radius. With a relative velocity of 2 cl and neglecting the gravitational focussing, he finds
i.e. a galaxy is expected to encounter one other galaxy over the course of its evolution. Even though the assumption is very simplifying one sees that an encounter is a relatively frequent event in a cluster. In contrast to this, Gnedin (2003) estimated for the probability to merge with another galaxy
with a galactic velocity dispersion g = 200 km s-1. Hence an actual merger is an unlikely event.
Tidal interactions and merging between galaxies are highly non-linear phenomena that can be partly handled analytically (e.g., Binney & Tremaine 1987, chapter 7). Numerical simulations, however, are required for accurate estimates of mass loss rates and the morphological modification of galaxies.
Figure 6. Simulation of the interaction between two galaxies: distribution of gas (left) and stars (right). Due to the interaction gas and stars can be expelled out to distances of hundreds of kpc (from Kapferer et al. 2005).