Near-infrared spectroscopy of post-starburst galaxies: a limited impact of TP-AGB stars on galaxy spectral energy distributions

We present VLT-ISAAC near infrared (NIR) spectro-photometric observations of 16 post-starburst galaxies aimed at constraining the debated influence of TP-AGB stars on the spectral energy distribution (SED) of galaxies with stellar ages between 0.5 and 2 Gyr, hence critical for high-redshift studies. Post-starburst galaxies are characterised by negligible on-going star formation and a SED dominated by the stellar population formed in a recent (< 2 Gyr) burst. By selecting post-starburst galaxies with mean luminosity-weighted ages between 0.5 and 1.5 Gyr and a broad range of metallicities (based on SDSS optical spectroscopy), we explore the parameter space over which the relative energy output of TP-AGB stars peaks. A key feature of the present study is that we target galaxies at z � 0.2, so that two main spectral features of TPAGB stars (C-molecule band-head drops at 1.41 and 1.77 µm, blended with strong telluric absorption features, hence hardly observable from the ground, for targets at z � 0) move inside the H and K atmospheric windows and can be constrained for the first time to high accuracy. Our observations provide key constraints to stellar population synthesis models. Our main results are: i) the NIR regions around 1.41 and 1.77 µm (rest-frame) are featureless for all galaxies in our sample over the whole range of relevant ages and metallicities at variance with the Maraston (2005) “TPAGB heavy” models, which exhibit marked drops there; ii) no flux boosting is observed in the NIR. The optical-NIR SEDs of most of our post-starburst galaxies can be consistently reproduced with Bruzual & Charlot (2003) models, using either simple stellar populations (SSP) of corresponding light-weighted ages and metallicities, or a more realistic burst plus an underlying old population containing up to approximately 60% of the total stellar mass. In contrast, all combinations of this kind based on the Maraston (2005) models are unable to simultaneously reproduce the smoothness of the NIR spectra and the relatively blue optical-NIR colours in the observations. The data collected in this study appear to disfavour ”TP-AGB heavy” models with respect to ”TP-AGB light” ones.