The whole subject of finding and cataloging AGN is dominated by selection effects (see Hewett & Foltz 1994 for an extensive discussion and an excellent summary of the field). The main effects are:
(1) Dilution of the optical/IR brightness and color by starlight. This is a large effect for AGN with optical/IR luminosities less than that of an average galaxy (roughly 1044 ergs s-1) at moderate to high redshifts. This effect has made optical selection of Seyfert galaxies at z > 0.5 a very difficult process, since it reduces the signal from time variability, color, or emission-line survey techniques. This effect is unimportant for X-ray selection above a luminosity of ~ 1041.5 ergs s-1 and for radio selection above a power of 1024 Watts Hz-1. The dilution reduces the equivalent widths of all of the lines, as well as the visibility of the non-thermal radiation in the spectral bands where stars are luminous (i.e., optical/near-IR). The magnitude of this effect clearly depends on angular resolution and aperture and is minimized for HST data. However, very recent work shows that the brightness of the optical nuclei for the hard X-ray-selected sources in the Chandra Deep Field-South are a factor of 20 dimmer than expected based on the ROSAT sources (Grogin et al. 2003). It may also be important for luminous AGN at high redshifts, where most galaxies seem to have copious star formation. 5
(2) Obscuration. Based on X-ray and IR surveys, one finds that many (most?) AGN have large column densities of gas and dust in the line of sight (Fabian & Iwasawa 1999). Models of the X-ray and IR backgrounds (Almaini, Lawrence, & Boyle 1999) suggest that more than 70% of all AGN have high column densities (> 1022 atoms cm-2) in the line of sight, which, for normal dust-to-gas ratios, gives an effective optical absorption of AV > 5, effectively extinguishing the UV/near-IR fluxes. However, the situation is not simple: there are many objects known with (a) high X-ray column densities and luminous UV continua (the most famous being NGC4151), (b) high IR dust emission and low X-ray column densities (e.g., IRAS 1334; Brandt et al. 1997), and (c) very red continua, strong, broad optical lines, and apparently very high X-ray column densities (e.g., Wilkes et al. 2002; for a review, see Comastri et al. 2003). The multivariate distribution functions of each of these is not known at the present time, and thus these effects cannot be corrected for.
In unified models of active galaxies (Antonucci 1993), the physical difference between Seyfert 1 and Seyfert 2 galaxies is that the line of sight to the latter is blocked by optically-thick material in the UV and optical. This accounts for the observed weakness of UV/optical/soft X-ray emission, the lack of short timescale intensity variability, the high optical polarization, and the detailed form of the X-ray spectra. For these objects, the observed UV/soft X-ray continuum is only a small fraction of the emitted radiation, and much of the energy is emitted in the IR. Seyfert 2 galaxies have low soft X-ray luminosities, low optical luminosities, and large IR luminosities. Thus, they are not found in great numbers in soft X-ray or optical color samples.
Both of the above selection biases make it very difficult to directly connect samples derived in different spectral bands and to derive unique values for the bolometric luminosity function and its evolution. It is also not clear if these selection effects are functions of redshift and source luminosity.
Much recent progress has been achieved via Chandra and XMM-Newton observations of optically well-surveyed fields, with some surprising results. The Chandra-selected AGN sample shows that at high X-ray luminosities, almost all AGN show broad optical emission lines, indicating that the effects of obscuration are small, while at low luminosities, the vast majority of AGN show little or no activity in the optical (Steffen et al. 2003). There are also strong indications of evolution in these ratios.
There is another important effect, which, while not a classical selection effect, has important consequences for the nature and completeness of AGN samples. It is clear that the strengths and widths of various UV and optical lines are not random but lie along two eigenvectors (Boroson 2002). These line strengths are strongly correlated with the slope of the X-ray continuum (Brandt & Boller 1998). Thus, in X-ray flux-limited surveys at different energy ranges, or in UV/optical surveys that rely on line widths and fluxes, one will end up with different sets of objects. There also may be evolution in the nature of these eigenvectors. It is speculated that the narrow-line Seyfert 1 galaxies that lie along one of the extrema are radiating near the Eddington limit, and, if so, should be more common at higher redshifts.
5 Recent work by Comastri et al. (2003) shows that X-ray-to-optical luminosity ratios for objects having only narrow optical lines rise as L2-10 keV). Note that these may not be the same as classical Seyfert 2 galaxies; for the purposes of this chapter, I follow the nomenclature of the BeppoSAX High-Energy Large Area Survey (HELLAS) team and call these optically-obscured type 2 AGN. Thus, at high X-ray luminosities, these objects are X-ray bright rather than X-ray dim. Back.