Consequences of bursty star formation on galaxy observables at high redshifts

Abstract: The star formation histories (SFHs) of dwarf galaxies are thought to be \emph{bursty}, with large -- order of magnitude -- changes in the star formation rate on timescales similar to O-star lifetimes. As a result, the standard interpretations of many galaxy observables (which assume a slowly varying SFH) are often incorrect. Here, we use the SFHs from hydro-dynamical simulations to investigate the effects of bursty SFHs on sample selection and interpretation of observables and make predictions to confirm such SFHs in future surveys. First, because dwarf galaxies' star formation rates change rapidly, the mass-to-light ratio is also changing rapidly in both the ionizing continuum and, to a lesser extent, the non-ionizing UV continuum. Therefore, flux limited surveys are highly biased toward selecting galaxies in the \emph{burst} phase and very deep observations are required to detect all dwarf galaxies at a given stellar mass. Second, we show that a $\log_{10}[\nu L_{\nu}(1500{\rm \AA})/L_{{\rm H}\alpha}]>2.5$ implies a very recent quenching of star formation and can be used as evidence of stellar feedback regulating star formation. Third, we show that the ionizing continuum can be significantly higher than when assuming a constant SFH, which can affect the interpretation of nebular emission line equivalent widths and direct ionizing continuum detections. Finally, we show that a star formation rate estimate based on continuum measurements only (and not on nebular tracers such as the hydrogen Balmer lines) will not trace the rapid changes in star formation and will give the false impression of a star-forming main sequence with low dispersion.