Nurturing Lyman break galaxies: observed link between environment and spectroscopic features

Abstract

We examine the effects of magnitude, colour and Lyα equivalent width (EW) on the spatial distribution of z ∼ 3 Lyman break galaxies (LBGs) and report significant differences in the two-point autocorrelation functions. The results are obtained using samples of ∼10 000–55 000 LBGs from the Canada–France–Hawaii Telescope Legacy Survey Deep Fields. We find that magnitude has a larger effect on the autocorrelation function amplitude on small scales (≲1 |$h^{-1}_{70}$| Mpc, the one-halo term) and that colour is more influential on large scales (≳1 |$h^{-1}_{70}$| Mpc, the two-halo term). We find the most significant differences between autocorrelation functions for LBGs with dominant net Lyα EW in absorption (aLBGs) and dominant net Lyα EW in emission (eLBGs) determined from ≳95 per cent pure samples of each population using a photometric technique calibrated from ∼1000 spectra. The aLBG autocorrelation function has a higher two-halo amplitude than the full LBG sample and has a one-halo term departure from a power-law fit near ∼1 |$h^{-1}_{70}$| Mpc, corresponding to the virial radii of MDM ∼ 1013 M⊙ dark matter haloes. In contrast, the eLBG autocorrelation function has a one-halo term departure at ∼0.12 |$h^{-1}_{70}$| Mpc, suggesting parent haloes of MDM ∼ 1011 M⊙ and a two-halo term that exhibits a curious ‘hump’ on intermediate scales that we localize to the faintest, bluest members. The aLBG–eLBG cross-correlation function exhibits an anticorrelation component that reinforces different physical locations for a significant fraction of aLBGs and eLBGs. We introduce a ‘shell’ model for the eLBG autocorrelation function and find that the form can be reproduced assuming that a significant fraction of eLBGs have a shell-like spatial distribution. Based on the analysis of all LBG subsamples, and considering the natural asymmetric distribution of LBGs on the colour–magnitude diagram, we conclude that aLBGs are more likely to reside in group-like environments hosting multiple luminous (i′ < 26.4) LBGs whereas eLBGs are more likely to be found on group outskirts and in the field. Because Lyα is a tracer of several intrinsic properties, including morphology, the results presented here imply that the mechanisms behind the morphology–density relation at low redshift are in place at z ∼ 3 and that Lyα EW may be a key environment diagnostic. Finally, our results show that the LBG autocorrelation function amplitude is lower than the true average as a result of the spatial anticorrelation of the spectral types. This result holds broad consequences for all autocorrelation functions measured for any population that contains members residing in different environments as the average amplitude, and hence the inferred average dark matter mass, will always be underestimated.