Evidence for Merger-Driven Activity in the Clustering of High Redshift Quasars

Recently Shen et al. (2007) found that high-z quasars from SDSS are very strongly clustered, with a bias of ~14 at z=4. White et al. (2008) used this result to show that there must be an extremely tight relation between quasar luminosity and halo mass, with an upper limit to the scatter of 0.3 dex. The basic reasoning behind this conclusion is that, if there were a larger scatter, then many quasars in (very abundant) low-mass halos would have high enough luminosity to make it into Shen's sample, however the observed number density of quasars is too low to allow for this. Wyithe & Loeb say that such a tight scatter is difficult to believe since the scatter in the relationship between black hole mass and bulge velocity dispersion is also 0.3 dex, and one might expect that this relationship is tighter and more direct then the relationship between halo mass and quasar luminosity.

Now Wyithe & Loeb have revisited this issue, using a somewhat more flexible model than was used by White et al. For instance, White et al. assumed that quasar luminosity is proportional to halo mass, whereas physical arguments suggest that it should be proportional to halo mass to a higher power. Also, Wyithe and Loeb allow for an arbitrary boost in the clustering of halos that host quasars. Such a boost might be expected if those halos have special properties, for instance if they have just merged.

This figure shows the joint likelihood distributions of various parameters in the Wyithe & Loeb model. I won't bother to explain all of the parameters, so just look at the upper right plot. This shows contours of F (the amount that the bias is boosted by) vs. gamma (the slope of the halo mass vs. quasar luminosity relation). Models where F=1 are highly disfavored. This suggests that you can't explain the observed quasar number density and clustering using a standard clustering model, but that some other ingredient must come into play.

This conclusion is related to some other recent results, as I mentioned here, however those results may be subject to systematic observational uncertainties. Perhaps the Shen et al. measurement is on firmer ground, but I haven't looked at that paper in detail. But it does seem that the very high clustering measurement is in contrast with the measurement presented by Adelberger & Steidel (2005).