Globular Cluster (GC) systems of most early-type galaxies show bimodal optical color distributions, yet their Luminosity Functions (LFs) look like simple Gaussians with turn-over magnitudes used for determination of distances and the Hubble constant. Taking color distributions and LFs of Milky Way halo GCs as a reference for the apparently universal old metal-poor GC populations and assuming their intrinsic widths to be constant in time at the values of the Milky Way halo GC populations, I study for which combinations of age and metallicity a secondary GC population formed in some violent star formation event in the history of a galaxy may or may not be detected in the observed GC color distributions and LFs (U- through K-bands). Significant differences are found among the diagnostic efficiencies in various wavelength regions with the NIR predicted to be able to reveal the presence of GC subpopulations that may perfectly hide within one inconspicuous optical color peak. If, and only if, some broad age -- metallicity relation is assumed, as expected e.g. for secondary GCs forming in spiral galaxy mergers from a pre-enriched ISM, models indicate single-peak GC LFs even in GC systems with bimodal color distributions. We point out the importance of having multi-band information (including NIR) to independently constrain ages and metallicities of GC subpopulations that will shed light on both cluster and galaxy formation scenarios. Models presented here are being used for the interpretation of multi-wavelength data from our ASTROVIRTEL project on GC systems in galaxies of various types, luminosities and environments as well as on young star cluster systems in interacting galaxies and mergers.

18 pages, 7 figures, accepted for A&A