Title: Solving small-scale clustering problems in approximate lightcone mocks

Abstract: Realistic lightcone mocks are important in the clustering analyses of large galaxy surveys. For simulations where only the snapshots are available, it is common to create approximate lightcones by joining together the snapshots in spherical shells. We assess the two-point clustering measurements of central galaxies in approximate lightcones built from the Millennium-XXL simulation, which are constructed using different numbers of snapshots. The monopole and quadrupole of the real-space correlation function is strongly boosted on small scales below 1 Mpc/h, due to some galaxies being duplicated at the boundaries between snapshots in the lightcone. When more snapshots are used, the total number of duplicated galaxies is approximately constant, but they are pushed to smaller separations. The effect of this in redshift space is small, as long as the snapshots are cut into shells in real space. Randomly removing duplicated galaxies is able to reduce the excess clustering signal. Including satellite galaxies will reduce the impact of the duplicates, since many small-scale pairs come from satellites in the same halo. Galaxies that are missing from the lightcone at the boundaries can be added to the lightcone by having a small overlap between each shell. This effect will impact analyses that use very small-scale clustering measurements, and when using mocks to test the impact of fibre collisions.