Implications for unified schemes from the quasar fraction and emission-line luminosities in radio-selected samples

We use a principal components analysis of radio-selected (3CRR, 6CE and 7CRS) active galactic nuclei (AGN) data sets to define two parameters related to low-frequency (151-MHz) radio luminosity L-151 and [O III] luminosity L-[O III]: a parameter alpha encoding the L-151-L-[O III] correlation and a parameter beta encoding scatter about this correlation. We describe methods for constructing generalized luminosity functions (GLFs) based on alpha, beta, redshift and schemes for unifying quasars and radio galaxies. These GLFs can be used to generate radio luminosity functions (RLFs) which improve on those of Willott et al. (2001a), mostly because they incorporate scatter and are therefore much smoother. Luminosity-dependent unified schemes (e.g. a receding-torus scheme) have been invoked to explain the low quasar-to-radio galaxy fraction at low alpha and the differences in emission-line luminosities of radio quasars and radio galaxies. With the constraints of the 3CRR, 6CE and 7CRS data sets and radio source counts, our GLF approach was used to determine whether a receding-torus-like scheme is required if there are two populations of radio sources: one at low alpha, consisting of 'starved AGN'; the other at high alpha, consisting of 'Eddington-tuned AGN'. Because of the overlap between these two populations and the effects of the beta parameter, schemes with or without a receding torus can produce a low quasar fraction at low alpha and differences in [O III] luminosity between radio galaxies and quasars. The receding torus may be a physical process important in one or more populations of radio sources, but this is not yet proved either by the quasar fraction or the emission-line properties of radio-selected samples.