On merger bias and the clustering of quasars

We use the large catalogues of haloes available for the Millennium Simulation to test whether recently merged haloes exhibit stronger large-scale clustering than other haloes of the same mass. This effect could help us to understand the very strong clustering of quasars at high redshift. However, we find no statistically significant excess bias for recently merged haloes over the redshift range 2 <= z <= 5, with the most massive haloes showing an excess of at most similar to 5 per cent. We also consider galaxies extracted from a semi-analytic model built on the Millennium Simulation. At fixed stellar mass, we find an excess bias of similar to 20-30 per cent for recently merged objects, decreasing with increasing stellarmass. The fact that recently merged galaxies are found in systematically more massive subhaloes than other galaxies of the same stellar mass accounts for about half of this signal, and perhaps more for high-mass galaxies. The weak merger bias of massive systems suggests that objects of merger-driven nature do not cluster significantly differently from other objects of the same characteristic mass over the range 5 < r < 25 h(-1) Mpc. We discuss the implications of these results for the interpretation of quasar clustering data with respect to quasar duty cycles, visibility times and evolution in the black hole-host mass relation.