How sensitive are predicted galaxy luminosities to the choice of stellar population synthesis model?

We present a new release of the GALFORM semi-analytical model of galaxy formation and evolution, which exploits a Millennium Simulation-class N-body run performed with the Wilkinson Microwave Anisotropy Probe 7 cosmology. We use this new model to study the impact of the choice of stellar population synthesis (SPS) model on the predicted evolution of the galaxy luminosity function. The semi-analytical model is run using seven different SPS models. In each case, we obtain the rest-frame luminosity function in the far-ultraviolet, optical and near-infrared (NIR) wavelength ranges. We find that both the predicted rest-frame ultraviolet and optical luminosity function are insensitive to the choice of SPS model. However, we find that the predicted evolution of the rest-frame NIR luminosity function depends strongly on the treatment of the thermally pulsating asymptotic giant branch (TP-AGB) stellar phase in the SPS models, with differences larger than a factor of 2 for model galaxies brighter than M-AB(K) - 5 log h < -22 (similar to L-* for 0 <= z <= 1.5). We have also explored the predicted number counts of galaxies, finding remarkable agreement between the results with different choices of SPS model, except when selecting galaxies with very red optical-NIR colours. The predicted number counts of these extremely red galaxies appear to be more affected by the treatment of star formation in discs than by the treatment of TP-AGB stars in the SPS models.