The brightest galaxies at cosmic dawn

Recent JWST observations suggest an excess of z ≳ 10 galaxy candidates above most theoretical models. Here, we explore how the interplay between halo formation time-scales, star formation efficiency, and dust attenuation affects the properties and number densities of galaxies observed in the early Universe. To guide intuition, we calculate the theoretical upper limit on the UV luminosity function (LF), assuming star formation is 100 per cent efficient and all gas in haloes is converted into stars, and that galaxies are at the peak age for UV emission (~10 Myr). This upper limit is ~4 orders of magnitude greater than current observations, implying no formal tension with star formation in Lambda cold dark matter cosmology. In a more realistic model, we use the distribution of halo formation time-scales derived from extended Press-Schechter theory as a proxy for star formation rate (SFR). We predict that the galaxies observed so far at z ≳ 10 are dominated by those with the fastest formation time-scales, and thus most extreme SFRs and young ages. These galaxies can be upscattered by ~1.5 mag compared to the median UV magnitude versus halo mass relation. This likely introduces a selection effect at high redshift whereby only the youngest (≲10 Myr), most highly star-forming galaxies (specific SFR$\gtrsim 30\, \mathrm{Gyr}^{-1}$) have been detected so far. Furthermore, our modelling suggests that redshift evolution at the bright end of the UV LF is substantially affected by the build-up of dust attenuation. We predict that deeper JWST observations (reaching m ~ 30) will reveal more typical galaxies with relatively older ages (~100 Myr) and less extreme specific SFRs ($\sim 10\, \mathrm{Gyr}^{-1}$ for a M_UV ~ -20 galaxy at z ~ 10).