Correlations between H α equivalent width and galaxy properties at z = 0.47: Physical or selection-driven?

ABSTRACT The H α equivalent width (EW) is an observational proxy for specific star formation rate (sSFR) and a tracer of episodic, bursty star-formation activity. Previous assessments show that the H α EW strongly anticorrelates with stellar mass as M−0.25 similar to the sSFR – stellar mass relation. However, such a correlation could be driven or even formed by selection effects. In this study, we investigate how H α EW distributions correlate with physical properties of galaxies and how selection biases could alter such correlations using a z = 0.47 narrow-band-selected sample of 1572 H α emitters from the Ly α Galaxies in the Epoch of Reionization (LAGER) survey as our observational case study. The sample covers a 3 deg2 area of COSMOS with a survey comoving volume of 1.1 × 105 Mpc3. We assume an intrinsic EW distribution to form mock samples of H α emitters and propagate the selection criteria to match observations, giving us control on how selection biases can affect the underlying results. We find that H α EW intrinsically correlates with stellar mass as W0∝M−0.16 ± 0.03 and decreases by a factor of ∼3 from 107 M⊙ to 1010 M⊙, while not correcting for selection effects steepens the correlation as M−0.25 ± 0.04. We find low-mass H α emitters to be ∼320 times more likely to have rest-frame EW>200 Å compared to high-mass H α emitters. Combining the intrinsic W0–stellar mass correlation with an observed stellar mass function correctly reproduces the observed H α luminosity function, while not correcting for selection effects underestimates the number of bright emitters. This suggests that the W0–stellar mass correlation when corrected for selection effects is physically significant and reproduces three statistical distributions of galaxy populations (line luminosity function, stellar mass function, EW distribution). At lower stellar masses, we find there are more high-EW outliers compared to high stellar masses, even after we take into account selection effects. Our results suggest that high sSFR outliers indicative of bursty star formation activity are intrinsically more prevalent in low-mass H α emitters and not a byproduct of selection effects.