Pair correlations and merger bias

We analytically study the possibility that mergers of haloes are more highly clustered than the general population of haloes of comparable masses. We begin by investigating predictions for merger bias within the extended Press-Schechter formalism and discuss the limitations and ambiguities of this approach. We then postulate that mergers occur whenever two objects form within a (small) fixed distance of each other. We therefore study the clustering of pairs of points for a highly biased population in the linear regime, for the overall mass distribution in the quasi-linear regime and (using the halo model of clustering) in the non-linear regime. Biasing, quasi-linear evolution and non-linear clustering all lead to non-zero reduced (or connected) three- and four-point correlation functions. These higher order correlation functions can in many cases enhance the clustering of close pairs of points relative to the clustering of individual points. If close pairs are likely to merge, then the clustering of mergers may be enhanced. We discuss implications for the observed clustering of luminous z=3 galaxies and for correlations of active galactic nuclei and galaxy clusters.