\label{sec-survey_discussion}
\label{ssec-comparison}
Comparison samples:
CoNFIG \citep{gen08} - morphological classifications from NVSS-FIRST; spectroscopic redshifts for \(\sim75\%\)
\citet{kim08} - radio galaxies from four surveys (NVSS + FIRST + WENSS + GB6); morphology is automated, restricted to “simple”, “compact”, and “complex”
morphological similarities; pure ellipticals and spirals from Galaxy Zoo
random points on the sky
We compare the clustering properties of blazars to active radio galaxies since, according to the unified model, these are the parent populations of blazars. As a result, the clustering properties of radio galaxies should be similar to those of blazars, assuming that the orientation of the jet has no effect on the megaparsec-scale environment.
A concern for comparing clustering properties of other galaxy types to blazars is to avoid biases of sample selection wherever possible. One possible bias arises from the redshift distributions of any comparison sample. Results from Figure \ref{fig-bgb_redshift} show a rise in \(B\) values as a function of redshift. If we were comparing our blazars to galaxies with a higher average redshift, the clustering values would be similarly high. We therefore match the redshift distributions of any comparison sample using the following method: firstly, we consider only objects in the same redshift range (\(0.043<z<0.75\)) as for the blazars. Next, we bin the redshift distribution of the blazars with \(\Delta z=0.02\) and count the number of objects per bin. We then truncate the comparison sample at the highest-redshift bin for which the number of blazars is greater than or equal to the number of comparison galaxies. For all bins at lower redshift, we randomly select comparison galaxies within that bin to search for neighbors and compute \(B_{gB}\).
The largest available catalogue of radio galaxies with robust morphological classifications is the CoNFIG sample of \citet{gen08}. CoNFIG consists of 274 bright radio sources at 1.4 GHz, with visually-classified FRI/FRII morphologies available for all galaxies. We searched the SDSS for neighbors of CoNFIG galaxies with confirmed redshifts, which comprises 89% of the total. We were able to compute \(B\) values for 283 radio galaxies, of which 55 were FRI and 184 were FRII. The remaining 44 galaxies were classified by \citet{gen08} as compact sources, with no visibly extended radio lobes.
For all radio galaxies in CoNFIG, the mean \(B\) value is \(164\pm389\), statistically similar to our blazar results. The subsamples of FR galaxies also have similar means, with FRI galaxies at \(150\pm533\) and FRII galaxies at \(175\pm364\). Compact sources showed a similar distribution of \(B\) values, at \(136\pm265\). A K-S test between the FRI and FRII galaxis showed no significant difference in their distribution.
\citet{gen13} specifically measure the richness of radio galaxies using the richness method of \citet{win11}. They found that FR I galaxies existed in richer environments than FR II galaxies.
\citet[][KI08]{kim08} assembled a sample of radio sources with much larger numbers than CoNFIG, but without detailed, visually-classified morphologies. We used their sample of galaxies cross-identified in four radio surveys (FIRST, NVSS, WENSS, GB6) and one optical survey (SDSS; DR6) with spectroscopic redshifts. This yielded 2886 radio galaxies in the SDSS footprint, 1964 of which fell in the redshift range of the blazars. After matching a subsample of the radio galaxies to the blazar redshift distribution, we compute \(B_{gB}\) values for 258 galaxies between \(0.063<z<0.627\).
\(B_{gB}\) values for the \citet{kim08} galaxies were roughly a factor of two higher than the both the blazars and CoNFIG radio galaxies, with a mean value of \(299\pm342\). A higher percentage of these galaxies as compared to blazars also live in overdense regions, with 88% of the KI08 sources having \(B_{gB}\)\(<0\).
Since there are too many galaxies in KI08 to be efficiently classified visually, the authors develop an automated morphological classification for the radio emission. This is based on two criteria: the difference in the amount of 1.4 GHz flux measured in two surveys (FIRST and NVSS) with different synthesized beam sizes, and the ratio of peak to integrated flux from the FIRST survey, which gives a dimensionless concentration index on a 5 scale. Their classes of radio morphology are “complex”, (simple) “resolved”, and (simple) “compact”. The comparison sample we selected contained 111 complex, 110 resolved, and 37 compact galaxies.
Complex radio galaxies from KI08 inhabited the richest environments, with a mean \(B_{gB}\) value of \(356\pm340\). In contrast, the clustering amplitudes for both radio galaxies with simple morphologies were similar, at \(275\pm286\) for compact galaxies and \(251\pm324\) for resolved galaxies.
Why is the clustering amplitude for KI08 twice that of CoNFIG?
Finally, we looked at host galaxies using the GZ1 classifications to see how the morphology-density relation applies to blazar hosts, and whether the standard assumption that blazars are hosted in elliptical galaxies holds true. For optical galaxies, we used classifications from the Galaxy Zoo project \citep[GZ;][]{lin11}, using their clean classifications for ellipticals and spirals. As with the radio galaxies, we analyzed a sub-sample of the roughly 1 million galaxies with GZ classifications that matched the redshift distribution of the blazars. For elliptical galaxies, this allowed us to probe out to a redshift of \(z=0.50\); spiral galaxies are much more limited, with a maximum redshift bin of \(z=0.20\).
The mean \(B\) values for the galaxies selected by optical morphology are significantly different; elliptical galaxies resemble the blazar and radio galaxy distributions, with a mean covariance amplitude of \(168\pm316\). Spiral galaxies have much lower \(B\) values, however, with a larger spread and a mean close to zero (\(12\pm359\)). The K-S test applied to spiral galaxies and the blazar distribution (matching the spiral redshift range) is inconsistent at the 2.7\(\sigma\) level. While the means for both are much closer, it should be noted that a K-S test is also inconsistent between the elliptical galaxy \(B\) distribution and the blazars. This is more puzzling, since blazar hosts are thought to be almost without exception elliptical galaxies \citep{urr95}.