From theory and modelling, and increasingly also from observations, it is clear that bars can remove angular momentum from gaseous material, and thus drive it from the disk into the central kpc-scale regions of a galaxy (Section 2). In contrast, the evidence that this centrally condensed gas directly and immediately leads to AGN or starburst activity remains rather elusive. The relevant results, reviewed more in depth elsewhere in this paper as indicated below, have been summarised in Table 1, where bars and interactions have been labelled as primary indicators for links between the inflow-provoking mechanisms and the possibly resulting AGN or starburst activity. Also listed in Table 1 is a small number of so-called secondary indicators, which have received attention as outlining possible links between inflow and activity, but which may well be a result of one of the primary indicators. The information summarised in Table 1 can be related to the content of this paper as follows:
Feature | Seyferts/LINERs | Starbursts |
Primary indicators | ||
Bars | yes (but 2.5) | yes (but not in general?) |
Interactions | no | yes (but extremes only?) |
Secondary indicators | ||
Nuclear bars | no | N/A |
Rings | yes, some | yes (nuclear rings at least) |
Starbursts can be provoked by bars or interactions, at least in some cases (Section 2, 1, 3.1).
There is an increasing body of evidence that Seyfert activity preferentially occurs in barred host galaxies, but the effect is a statistical one, and not very pronounced (Section 2.2).
There is no convincing evidence that AGN hosts are interacting more often than non-AGN (Section 3.2).
Nuclear bars (Section 2.2) have great theoretical potential for bringing fuel very close to the centre of a galaxy (the "bars within bars" scenario, Shlosman et al. 1989) but have so far not lived up to that potential in terms of their detections in imaging surveys, where no higher nuclear bar fractions have been found in Seyferts as compared to non-Seyferts. As far as we aware, the possible statistical connections between nuclear bars and starbursts have not yet been studied.
There is some interesting evidence that rings, both nuclear and non-nuclear, may be related to the presence of low-luminosity AGN activity (Section 5). This issue needs further exploration, but in any case the rings will most likely have formed under the influence of either a bar or another form of non-axisymmetry in the gravitational potential of the host, hence the inclusion of rings as secondary indicators in Table 1.
Finally, there is a direct link between starburst activity and the presence of nuclear rings, since small nuclear rings with significant massive star formation can be classified as starburst, and since many starbursts might in fact be circumnuclear rather than nuclear, albeit with small ring radii of tens to hundreds of parsec (e.g., González-Delgado et al. 1998), or possibly even smaller. Statistical links between inner/outer rings and starburst activity have not yet been explored.
Since we have known for quite some time that net radial gas inflow must be accompanied by the loss of significant quantities of angular momentum, and that the kind of deviations from axisymmetry in the gravitational potential of the host galaxy set up by bars and interactions is well suited to lead to such angular momentum loss (see reviews by Shlosman et al. 1989, 1990; Shlosman 2003), we must be missing some part of the puzzle. One possibility is that we are not looking at the right things at the right time: the spatial- as well as the time-scales under consideration may not be correct. So far, the spatial scale considered observationally has been from tens of kpc down to, roughly, a few hundred parsec. Whereas this range may be wholly adequate for the study of a major starburst, which can span up to a kpc, it may well be wholly inadequate for AGN fuelling, which is expected to be related to accretion to a SMBH, on scales of AUs. As far as timescales are concerned, what has been considered in the studies reviewed here is generally a rather long-lived phenomenon influencing kpc-scale regions (bar or galaxy-galaxy interaction). Starburst or AGN activity, on the other hand, occurs on essentially unknown timescales (somewhere around 106 - 108 years could be expected for most AGN or starbursts). If the starburst or AGN activity is indeed short-lived, and possibly also periodic, the connection between the presence of activity at the currently observed epoch and any parameter of the host galaxy is not necessarily straightforward (as pointed out, e.g., by Beckman 2001).
The fact that we see any correlations at all, such as those of bar fractions with the presence of starburst or Seyfert activity, indicates that bars and interactions do have a role, presumably by establishing a gas reservoir in the central kpc region. In the coming years, we must start to disentangle the effects of gravitationally induced gas inflow, which brings gas to the inner kpc region at least, from those of possible other mechanisms which can transport the gas further in, and from the time scales and duty cycles of the activity. We seem to have reached the limits of purely morphological studies of the central regions of active and non-active galaxies (e.g., Laine et al. 2002), and must start to worry about the effects of the AGN or starburst on their immediate surroundings as we push the observations to smaller spatial scales, of tens of parsecs. One must, hence, move on to careful studies of the gas and stellar kinematics and dynamics. Integral field spectroscopy (e.g., Bacon et al. 2001), especially when used in conjunction with adaptive optics techniques, should allow a good deal of progress here, giving simultaneous high-resolution mapping of the distributions of stellar populations and dust, as well as of the gas and stellar kinematics. In combination with detailed numerical modelling, this could lead to the detection of the dynamical effects of, e.g, nuclear bars on gas flows which may be more directly related to the fuelling process of starbursts and/or AGN.
Acknowledgements I am indebted to my collaborators on the various aspects of the work described here, especially John Beckman, Shardha Jogee, Seppo Laine, Reynier Peletier, and Isaac Shlosman. Valuable comments by conference participants have helped improve this paper.