3.2. Interactions and Seyfert activity
Interactions and mergers have long been suspected of triggering high-luminosity AGN such as QSOs (e.g., Disney et al. 1995; Bahcall et al. 1997), although many of such AGN seem to lie in entirely undisturbed elliptical systems. In fact, Dunlop et al. (2003) show that the host galaxy properties of radio-loud and radio-quiet AGN are indistinguishable from those of quiescent but otherwise comparable galaxies, and Floyd et al. (2004) find no correlation between the luminosity of a quasar and the presence of any morphological disturbance in the host.
Seyfert activity is known to occur in interacting and merging galaxies, and several rather spectacular examples are well known (for instance NGC 2992, or a number of the closest ULIRGs such as Mrk 273). To check statistically whether there is a connection between interactions and the occurrence of this type of nuclear activity, authors have considered the numbers of companions to Seyfert galaxies as compared to non-active control galaxies (e.g., Fuentes-Williams & Stocke 1988; de Robertis, Yee, & Hayhoe 1998; Schmitt 2001), or, alternatively, have searched for different fractions of Seyfert or AGN activity among more or less crowded environments (e.g., Kelm, Focardi, & Palumbo 1998; Miller et al. 2003). The conclusion from this substantial body of work must be that no unambiguous evidence exists for a direct connection between the occurrence of Seyfert activity and interactions. Some earlier work did report claims of a statistical connection, but this work was unfortunately plagued by poor control sample selection (see Laurikainen & Salo 1995 for a detailed review), and most early studies, but also some of the recent ones, are not based on complete sets of redshift information for the possible companion galaxies. In addition, Laine et al. (2002) have shown that the bar fraction among both the Seyfert and non-Seyfert galaxies in their sample is completely independent of the presence of companions (interacting galaxies were not considered by Laine et al.). We thus conclude that interactions and Seyfert activity may well be linked in individual cases, but that as yet the case that they are statistically connected has not been made convincingly.