2010ApJ...720..284M

We present composite 3.6 and 4.5 µm luminosity functions (LFs) for cluster galaxies measured from the Spitzer Deep, Wide-Field Survey for 0.3 < z < 2. We compare the evolution of m* for these LFs to models for passively evolving stellar populations to constrain the primary epoch of star formation in massive cluster galaxies. At low redshifts (z ≲ 1.3), our results agree well with models with no mass assembly and passively evolving stellar populations with a luminosity-weighted mean formation redshift z_f= 2.4 assuming a Kroupa initial mass function (IMF). We conduct a thorough investigation of systematic biases that might influence our results, and estimate systematic uncertainties of Δz_f= ^+0.16_–0.18(model normalization), Δz_f = ^+0.40_–0.05(α), and Δz_f= ^+0.30_–0.45(choice of stellar population model). For a Salpeter-type IMF, the typical formation epoch is thus strongly constrained to be z ∼ 2-3. Higher formation redshifts can only be made consistent with the data if one permits an evolving IMF that is bottom-light at high redshift, as suggested by van Dokkum. At high redshifts (z ≳ 1.3), we also witness a statistically significant (>5σ) disagreement between the measured LF and the continuation of the passive evolution model from lower redshifts. After considering potential systematic biases that might influence our highest redshift data points, we interpret the observed deviation as potential evidence for ongoing mass assembly at this epoch.