Evidence for TP-AGB stars in high-redshift galaxies, and their effect on deriving stellar population parameters

We explore the effects of using different stellar population models on estimates of star formation histories, ages, and masses of high-redshift galaxies by fitting the SEDs with models by Maraston (hereafter M05) and by Bruzual & Charlot (hereafter BC03). We focus on the thermally pulsing asymptotic giant branch (TP-AGB) phase of stellar evolution, whose treatment is a source of major discrepancy. In this respect, BC03 models are representative of other models whose treatment of the TP-AGB phase is similar. Moreover, M05 and BC03 models adopt stellar tracks with different assumptions on convective overshooting. For our experiment we use a sample of high-z (1:4 ≲ z ≲ 2:7) galaxies, for which rest-frame UV spectroscopy and spectroscopic redshifts are available, along with Spitzer IRAC and MIPS photometry from GOODS. The mid-UV spectra of these galaxies exhibit features typical of A- or F-type stars, indicative of ages in the range ∼0.2-2Gyr,when the contribution of TP-AGB stars is expected to be maximum. We find that the TP-AGB phase plays a key role in the interpretation of the Spitzer data,where the rest-frame near-IR is sampled. Generally, M05 models give better fits than BC03 models and indicate systematically lower ages and lower masses (by ∼60%, on average). Photometric redshifts derived using M05 models are also in better agreement with the spectroscopic ones, especially when the rest-frame near-IR fluxes from Spitzer IRAC are included in the fit.We argue that the different results are primarily a consequence of the different treatment of the TP-AGB phase, although other differences in the input stellar evolution also contribute. This work provides a first direct evidence for a strong contribution by TP-AGB stars to the SED of galaxies in the high-redshift universe (z ∼ 2).