The requirements for fueling AGNs are more stringent than those for fueling H II nuclei, since angular momentum transport must extend to much smaller radii. Numerical simulations (e.g., Heller & Shlosman 1994) show that a large-scale stellar bar can reduce the angular momentum of the gas only by a factor of ~ 10. A number of ideas have been suggested to further transport the gas to scales of interest for AGNs ( 1 pc) (Lin, Pringle, & Rees 1988; Shlosman et al. 1989; Hernquist 1989; Pfenniger & Norman 1990; Wada & Habe 1992), with the ``bars-within-bars'' mechanism originally proposed by Shlosman et al. (1989) being widely discussed.
The observations, however, seem to indicate that bars have a negligible influence both in the formation and/or fueling of nearby AGNs. Related to this is the obvious fact that in the general galaxy population the fraction of barred galaxies greatly exceeds that of Seyferts. Even when we include emission-line nuclei such as LINERs in the AGN census, barred galaxies still outnumber AGNs by a factor of two (Ho et al. 1996b).
If the proposed mechanisms of angular momentum transport operate in nature, they must do so rather inefficiently. A crucial element of the models which invoke gas dissipation to fuel the nucleus is that the gas must constitute a non-negligible fraction of the total dynamical mass. For example, in the ``bars-within-bars'' models, the gas fraction amounts to ~ 10-20% of the total mass of the disk; likewise, the self-gravitating nuclear disk envisioned by Lin et al. (1988) requires a comparable amount of gas. Since AGNs occur predominantly in early- to intermediate-type hosts, perhaps the centers of these spirals lack sufficient gas for these mechanisms to ``kick in.'' Given that ILRs naturally develop in early-type barred galaxies (Combes & Elmegreen 1993) and that the initial large-scale radial inflow accumulates the gas between the inner and outer ILRs (Athanassoula 1992), it is conceivable that the region interior to the ILRs never attains a sufficiently high gas fraction. The gas content in the central ``cavity'' in barred early-type galaxies may be low enough that the instability mechanisms proposed for further inflow cannot operate.
In a similar vein, one might appeal to a duty-cycle argument. Suppose that the inflow is not continuous, but rather episodic in nature, with the duration between episodes of accretion lasting on the order of the lifetime of the bar. The duty cycle may be related to the timescale for accumulating a critical amount of gas to initiate the instabilities required for the various mechanisms of radial transport. In the absence of conditions which lead to their destruction, bars appear to be long-lived structures (Sellwood & Wilkinson 1993). If the duty-cycle argument holds, it implies that fueling of nearby AGNs on nuclear dimensions must be a slow process.
A third possibility is that the very process of central accretion (or simply the presence of a massive black hole) and the concomitant development of a strong and extended ILR can also lead to the destruction of the bar (Hasan & Norman 1990; Pfenniger & Norman 1990; Friedli & Benz 1993). Such a regulating mechanism has been proposed by Friedli (1994) to account for the scarcity of luminous AGNs in the current epoch. Although this phenomenon may account for the similarity in the frequency of barred and unbarred AGNs, it does not explain why barred and unbarred AGNs currently show the same level of activity. Since the survivability of the bar depends sensitively on the mass of the central concentration relative to the mass of the stellar disk (Friedli 1994), this scenario can be tested by appropriate kinematic observations.
If gas from the kiloparsec-scale region is not the principal source of fuel in AGNs, is there a fuel crisis? To address this issue, let us consider the contribution of stellar sources alone. For an efficiency of conversion between matter and energy of = 0.1, the mass accretion rate required to sustain a luminosity L is = ( c2)-1 L = 0.15( / 0.1) (L / 1045 ergs s-1) M yr-1. While QSOs typically consume 10-100 M yr-1, a rate which may be difficult to accommodate with stellar sources (Shlosman et al. 1990), a Seyfert nucleus such as that of NGC 1068 (Pier et al. 1994) only requires ~ 0.2 M yr-1, and the LINER nucleus in M81 just ~ 5 x 10-5 M yr-1 (Ho, Filippenko, & Sargent 1996a). Fuel may be extracted from stars surrounding the nucleus from their tidal disruption by the central massive black hole (e.g., Hill 1975; Rees 1988; Roos 1992). According to Eracleous, Livio, & Binette (1995), one could expect a stellar disruption once every 100-200 years for a black hole mass of 106-107 M; the tidal debris forms an elliptical accretion disk capable of sustaining the ionizing radiation of low-luminosity sources for several decades. Another source of fuel is expected from the normal mass loss of stars in the vicinity of the nucleus. Ho et al. (1996c) show that if galaxy nuclei have sufficiently high stellar densities, as seems to be the case for some objects recently imaged by the Hubble Space Telescope, an appreciable amount of material is available to sustain the feeble power in nearby AGNs.
L. C. H. is grateful to Ron Buta for providing travel support to attend this meeting. This work was funded by NSF grants AST-8957063 (A. V. F.) and AST-9221365 (W. L. W. S.).