The evolution of early-type field galaxies selected from a NICMOS map of the Hubble Deep Field North

The redshift distribution of well-defined samples of distant early-type galaxies offers a means to test the predictions of monolithic and hierarchical galaxy formation scenarios. NICMOS maps of the entire Hubble Deep Field North (HDF-N) in the F110W and F160W filters, when combined with the available Wide Field Planetary Camera 2 (WFPC2) data, allow us to calculate photometric redshifts and determine the morphological appearance of galaxies at rest-frame optical wavelengths out to zsimilar to2.5. Here we report results for two subsamples of early-type galaxies, defined primarily by their morphologies in the F160W band, which were selected from the NICMOS data down to H-160AB<24.0. A primary subsample is defined as the 34 galaxies with early-type galaxy morphologies and early-type galaxy spectral energy distributions. The secondary subsample is defined as those 42 objects that have early-type galaxy morphologies with non-early-type galaxy spectral energy distributions. The observed redshift distributions of our two early-type samples do not match that predicted by a monolithic collapse model, which shows an overabundance at z>1.5. A test confirms this result. When the effects of passive luminosity evolution are included in the calculation, the mean value of V-max for the primary sample is 0.22+/-0.05, and it is 0.31+/-0.04 for all the early types. A hierarchical formation model better matches the redshift distribution of the HDF-N early types at z>1.5 but still does not adequately describe the observed early types. The hierarchical model predicts significantly bluer colors on average than the observed early-type colors and overpredicts the observed number of early types at zsimilar to1. Although the observed redshift distribution of the early-type galaxies in our HDF-NICMOS sample is better matched by a hierarchical galaxy formation model, the reliability of this conclusion is tempered by the restricted sampling area and relatively small number of early-type galaxies selected by our methods. For example, our results may be biased by the way the HDF-N appears to intersect a large-scale structure at zsimilar to1. The results of our study underscore the need for high-resolution imaging surveys that cover greater area to similar depth with similar quality photometry and wavelength coverage. Although similar in appearance in the H-160 data, the primary and secondary samples are otherwise rather different. The primary sample is redder, more luminous, larger, and apparently more massive than the secondary sample. Furthermore, the secondary sample shows morphologies in the optical WFPC2 images that are more often similar to late-type galaxies than is the case for the primary sample. The bluer secondary sample of early types have a star formation history that can be approximated by a Bruzual Charlot tau model or by a galaxy formed at high redshift with a small, recent starburst. Given the differences in their apparent stellar masses and current luminosities, it would seem unlikely that the secondary sample could evolve into galaxies of the primary sample.