4.1. Global Parameters
The near dichotomy in the distribution of Hubble types for galaxies hosting active versus inactive nuclei (Figure 4) leads to the expectation that the two populations ought to have fairly distinctive global, and perhaps even nuclear, properties. Moreover, a detailed examination of the host galaxies of AGNs may shed light on the origin of their spectral diversity. These issues were examined by Ho, Filippenko & Sargent (2003) using the database from the Palomar survey. The host galaxies of Seyferts, LINERs, and transition objects display a remarkable degree of homogeneity in their large-scale properties, after factoring out spurious differences arising from small mismatches in Hubble type distribution. The various nuclear types have slightly different Hubble distributions, which largely control many of the statistical properties of the host galaxies. Unless this effect is taken into account, one can arrive at erroneous conclusions about the intrinsic differences of the AGN populations. This is a crucial step, one that is often not appreciated. All three classes have essentially identical total luminosities (~ L*), bulge luminosities, sizes, and neutral hydrogen content. Moreover, no obvious differences are found in terms of integrated optical colors or far-IR luminosities and colors, which implies very similar global stellar content and current star formation rates. No clear differences in environment can be seen either. The only exception is that, relative to LINERs, transition objects may show a mild enhancement in the level of global star formation, and they may be preferentially more inclined. This is consistent with the hypothesis that the transition class arises from spatial blending of emission from a LINER and H II regions. The velocity field of the ionized gas within the nuclear region, as measured by the width and asymmetry of the narrow emission lines, is crudely similar among the three AGN classes, an observation that argues against the proposition that fast shocks primarily drive the spectral variations (Section 6.2).
4.2. Nuclear Stellar Populations
The uniformity in the global stellar populations among the three AGN classes extends to circumnuclear and even nuclear scales. The Palomar spectra cover a suite of stellar absorption-line indices and nuclear continuum colors, which collectively can be used to obtain crude constraints on the age and metallicity of the stars within the central 2'' (~ 200 pc). After isolating a subsample that mitigates the Hubble type dependence, Ho, Filippenko, & Sargent (2003) find that Seyferts, LINERs, and transition objects have very similar stellar content. The same holds true when comparing type 1 and type 2 objects, both for LINERs and Seyferts. With a few notable exceptions such as NGC 404 and NGC 4569 (Maoz et al. 1998; Barth & Shields 2000; Gabel & Bruhweiler 2002) or NGC 4303 (Colina et al. 2002), the stellar population always appears evolved. Similar findings have been reported for smaller samples of LINERs, based on both optical (Boisson et al. 2000; Serote-Roos & Gonçalves 2004; Zhang, Gu & Ho 2008) and near-IR spectroscopy (Larkin et al. 1998; Bendo & Joseph 2004). The optical regime is not well suited to detect very young, ionizing stars. However, the Palomar spectra do cover the broad He II 4686 emission bump, a feature indicative of Wolf-Rayet stars commonly seen in very young (3-6 Myr) starbursts. Notwithstanding the difficulty of detecting this feature on top of a dominant old population, it is noteworthy that not a single case has been seen among the sample of over 200 Palomar LLAGNs. By contrast, the Wolf-Rayet bump has been found in a number of the H II nuclei (Sargent & Filippenko 1991; Ho, Filippenko & Sargent 1995), which, as a class compared to the LLAGNs, exhibit markedly younger stars, as evidenced by their blue continuum, strong H and H absorption, and weak metal lines (Ho, Filippenko & Sargent 2003). The general dearth of young, massive stars in LLAGNs presents a serious challenge to proposals that seek to account for the excitation of their line emission in terms of starburst models.
Closer in, on scales 10 pc accessible by HST, Sarzi et al. (2005) studied the nuclear stellar population for a distance-limited subsample of 18 Palomar LLAGNs. Their population synthesis analysis shows that the majority (80%) of the objects have predominantly old ( 5 Gyr), mildly reddened stars of near-solar metallicity, the only exceptions being 3 out of 6 transition objects and 1 out of 4 LINER 2s that require a younger ( 1 Gyr) component. In no case, however, is the younger component ever energetically dominant, falling far short of being able to account for the ionization budget for the central region.
The results of Ho, Filippenko & Sargent (2003) have been disputed by Cid Fernandes et al. (2004), who obtained new ground-based spectra for a subset of the Palomar LINERs and transition objects. González Delgado et al. (2004), in a study similar to that of Sarzi et al. (2005), further analyzed STIS spectra of some of these. An important improvement of their ground-based data is that they extend down to ~ 3500 Å, covering the 4000 Å break and the higher-order Balmer lines, which are sensitive probes of intermediate-age (~ 107 - 109 yr) stars. While LINERs are predominantly old, roughly half of the transition nuclei show significant higher-order Balmer lines. Again, there are virtually no traces of Wolf-Rayet features. These authors propose that the ionization mechanism of transition sources must be somehow linked to the intermediate-age stellar population.
I disagree with their assessment. Figure 1 of Cid Fernandes et al. (2004) clearly shows that, as in the parent Palomar sample, the Hubble type distribution of the LINERs is skewed toward much earlier types than that of the transition objects. Any meaningful comparison of the stellar population, which strongly depends on Hubble type, must take this into account. The strategy employed by Ho, Filippenko & Sargent (2003) is to restrict the two-sample comparisons to a relatively narrow range of Hubble types (Sab-Sbc). Within this domain, LINERs and transition objects (as well as Seyferts) have statistically indistinguishable stellar indices and continuum colors. Given the limited size of Cid Fernandes et al.'s sample, it is not possible to adopt the same strategy. To minimize the Hubble type bias, I examined the subsample of 30 spiral galaxies in their study. Out of 17 transition objects, 15 (88%) contain intermediate-age stars according to their Table 3; but so do 10 out of the 13 (77%) LINERs in this subgroup. This simple exercise underscores the importance of sample selection effects, and leaves me unconvinced that transition objects have a younger stellar population than LINERs. True, both classes evidently do contain detectable amounts of intermediate-age stars - a qualitatively different conclusion than was reached in the Palomar survey, whose spectral coverage was not well-suited to detect this population - but the fact remains that in a relative sense all three LLAGN classes in the Palomar survey have statistically similar populations. If the poststarburst component is responsible for the nebular emission in LLAGNs, we might expect the intensity of the two to be correlated, by analogy with what has been found for higher luminosity AGNs (Kauffmann et al. 2003). I searched for this effect in the final sample presented in González Delgado et al. (2004), but did not find any correlation. Clearly we wish to know what factors drive the spectral variations in LLAGNs; whatever they are (Section 6), they are unlikely to be related to stellar population.
4.3. Influence of Bars and Environment
Numerical simulations (e.g., Heller & Shlosman 1994; see review in Kormendy & Kennicutt 2004) suggest that large-scale stellar bars can be highly effective in delivering gas to the central few hundred parsecs of a spiral galaxy, thereby potentially leading to rapid star formation. Further instabilities result in additional inflow to smaller scales, which may lead to increased BH fueling and hence elevated nonstellar activity in barred galaxies compared to unbarred galaxies. As discussed in Section 3.1, the Palomar sample is ideally suited for statistical comparisons of this nature, which depend delicately on issues of sample completeness and the choice of control sample. Ho, Filippenko & Sargent (1997d) find that while the presence of a bar indeed does enhance both the probability and rate of star formation in galaxy nuclei, it appears to have no impact on either the frequency or strength of AGN activity. Bearing in mind the substantial uncertainties associated with sample selection, as well as the method and wavelength used to identify bars (Laurikainen, Salo & Buta 2004), other studies broadly come to a similar conclusion (see review by Combes 2003), although Maia, Machado & Willmer (2003) claim, on the basis of a significantly larger and somewhat more luminous sample drawn from the Southern Sky Redshift Survey, that Seyfert galaxies are preferentially more barred.
In the same vein, dynamical interactions with neighboring companions should lead to gas dissipation, enhanced nuclear star formation, and perhaps central fueling (e.g., Hernquist 1989). Schmitt (2001) and Ho, Filippenko & Sargent (2003) studied this issue using the Palomar data, parameterizing the nearby environment of each object by its local galaxy density and the distance to its nearest sizable neighbor. After accounting for the well-known morphology-density relation, it was found that the local environment, like bars, has little impact on AGNs, at least in the low-luminosity regime sampled locally. These findings broadly agree with the results from SDSS (Miller et al. 2003; Li et al. 2008). Kauffmann et al. (2004), Wake et al. (2004), and Constantin & Vogeley (2006), in fact, report a drop in the fraction of high-luminosity AGNs for dense environments.