Thousands of high-redshift galaxies at z ⇠ 6 are now being routinely discovered using HST (e.g. McLure et al. 2011), but the bright-end of the galaxy luminosity function at these redshifts can only be constrained using ground-based NIR surveys over much wider fields (Bowler et al. 2015). At z ⇠ 7 the bright-end of the galaxy luminosity function appears to follow the dark matter halo mass function suggesting that quenching is yet to occur at these redshifts (Bowler et al. 2014). However, the brightest z ⇠ 6 galaxies are found to su↵er from strong cosmic variance. Statistical samples of luminous galaxies at z ⇠ 5 − 7 over multiple fields, would enable us to accurately constrain the epoch of onset of mass quenching in high-redshift galaxies. Clustering measurements would allow the luminosity function to be connected with the dark matter halo mass function, providing stringent constraints on galaxy formation models. We have used the Bowler et al. (2015) z ⇠ 6 galaxy luminosity function to estimate that there will be at least 800 ultra-luminous star-forming galaxies in our survey region at significantly higher luminosities than currently explored by UltraVISTA. These ultra-luminous high redshift galaxies can only be discovered via large ground-based surveys and will be ideally suited for detailed follow-up with JWST, E-ELT, VLT-MOONS and ALMA.
The selection of high-redshift star-forming galaxies and quasars requires deep optical data blueward of the Lyman break as well as observations in more than one filter redward of the break to distinguish true highredshift galaxies from low redshift contaminants. Thus this science can only be achieved by combining deep optical data with deep NIR surveys with VISTA. DES now provides the necessary optical coverage. To fully exploit the depth and area of the DES z-band observations (zAB < 26.5), VISTA J-band observations are needed down to J ⇠25 (deeper than VIDEO). This is sufficiently deep to separate the spectral energy distributions of true high-redshift galaxies from low-redshift interlopers. The deep KS-band observations from VEILS will similarly allow full exploitation of the Spitzer 3.6μm and 4.5μm data in these fields from DeepDrill, enabling the selection of the most luminous z ⇠ 6 − 8 star-forming galaxies where the presence of strong optical nebular emission lines results in extreme Spitzer colours (e.g. Roberts-Borsani et al. 2015).