Annu. Rev. Astron. Astrophys. 2000. 38: 667-715 Copyright © 2000 by Annual Reviews. All rights reserved

5.3. Elliptical galaxies

The simple picture of the passively evolving elliptical galaxy, formed at high redshift in a single rapid collapse and starburst, has held sway since the scenario was postulated by [Eggen et al. 1962] and its photometric consequences were modeled by [Larson1974] and [Tinsley & Gunn 1976]. This hypothesis is supported by the broad homogeneity of giant elliptical (gE) galaxy photometric and structural properties in the nearby universe, and by the relatively tight correlations between elliptical galaxy chemical abundances and mass or luminosity. Observations in the past decade have pushed toward ever higher redshifts, offering the opportunity to directly watch the evolutionary history of elliptical galaxies. Most of this work has concentrated on rich cluster environments, and is not reviewed here except to note that most observers have favored the broad interpretation of quiescent, nearly passive evolution among cluster ellipticals out to z 1 (e.g. [Aragón-Salamanca et al. 1993]; [Stanford et al. 1995, Stanford et al. 1998]; [Ellis et al. 1997]; [van Dokkum et al. 1998, De Propris et al. 1999]).

The "monolithic" formation scenario serves as a rare example of a clearly stated hypothesis for galaxy evolution against which to compare detailed measurements and computations. The alternative "hierarchical" hypothesis is that elliptical galaxies formed mostly via mergers of comparable-mass galaxies that had at the time of merging already converted at least some of their gas into stars (e.g. [Kauffmann et al. 1993]). The argument between the monolithic and the hierarchical camps is not about the origin of galaxies from gravitational instability within a hierarchy of structure, but rather about when gEs assemble most of their mass and whether they form their stars mostly in situ or in smaller galaxies that subsequently merge. At least one observational distinction is clear: In the hierarchical model the number density of elliptical galaxies should decrease with redshift. In the monolithic model the number density should remain constant and the bolometric luminosities should increase out to the epoch of formation.

Measuring the density and luminosity evolution of the field elliptical population, however, has proven difficult, and many different approaches, perhaps complementary but not necessarily concordant, have been used to define suitable galaxy samples. Recent debate has focused on estimates of the evolution the co-moving density of passively-evolving elliptical galaxies in the Canada-France Redshift Survey [Lilly et al. 1995b]. Applying different photometric selection criteria and different statistical tests, two groups [Lilly et al. 1995a, Totani & Yoshii 1998] found no evidence for density evolution out to z = 0.8, whereas another group [Kauffmann et al. 1996], found evidence for substantial evolution. This debate highlights the difficulties inherent to defining samples based on a color cut in the margins of a distribution, where small changes in the boundary, as well as systematic and even random photometric errors in the data, can have substantial consequences for the conclusions.

HST, makes it possible to define samples of distant galaxies morphologically. The multicolor HDF images provide an attractive place to study distant ellipticals, but because of its very small volume and the inherently strong clustering of elliptical galaxies, one must be careful in drawing sweeping conclusions from HDF data alone. Thus, although we focus our attention primarily on the HDF results, they should be considered in context with results from other surveys (e.g. [Driver et al. 1995, Glazebrook et al. 1995, Driver et al. 1998, Treu et al. 1999, Treu & Stiavelli 1999]).

The interpretation of the high-redshift elliptical galaxy counts depends in large measure on a comparison to the local luminosity function (LF) of elliptical galaxies. The basic parameters of published LFs for local elliptical galaxies (or what are sometimes assumed to be elliptical galaxies) ³ span a very wide range in normalization, characteristic luminosity, and faint end behavior. Early HST studies such as that by [Im et al. 1996] concluded that NE or PLE models were consistent with elliptical-galaxy number counts predicted using the local LF of [Marzke et al. 1994].

Searches for distant elliptical galaxies in the HDFs have relied on either color or morphology. In the HDF-N, [Fasano et al. 1998] and [Fasano & Filippi 1998], along with [Schade et al. 1999] for the CFRS and LDSS surveys, all examined the size-luminosity relation for morphologically-selected ellipticals, finding evolution out to z ~ 1 consistent with PLE models. [Kodama et al. 1999] found a well-defined color-magnitude sequence at <z> ~ 0.9, consistent with passive evolution for approximately half the galaxies, but they also noted a substantial "tail" of bluer objects. Similarly, in the [Schade et al. 1999] study, about one third of the sample at z > 0.5 had [OII] line emission and colors significantly bluer than PLE models.

Ir imaging has made it attractive to pursue gEs at redshifts z > 1. In the HDF-N, [Zepf 1997] and [Franceschini et al. 1998] used ground-based K-band data, selecting samples by color and morphology, respectively, and concluded that there was an absence of the very red galaxies that would be expected if the elliptical population as a whole formed at very large redshift and evolved passively. In particular, Franceschini et al. highlighted the apparently sudden disappearance of HDF ellipticals beyond z > 1.3, which suggests that either dust obscuration during early star formation, or morphological perturbation during early mergers, was responsible. [Barger et al. 1999] made an IR study of the HDF-N flanking fields, selecting objects by color without reference to morphology. To K 20, they found few galaxies with I - K > 4, the expected color threshold for old ellipticals at z 1. However, other comparably deep and wide IR surveys have reported substantially larger surface densities of I - K > 4 galaxies ([Eisenhardt et al. 1998, McCracken et al. 2000]), raising concerns about field-to-field variations. [Menanteau et al. 1999] studied the optical-to-IR color distribution for a sample of ~ 300 morphologically early-type galaxies selected from 48 WFPC2 fields, including the HDF-N and flanking fields. They too find an absence of very red objects and a generally poor agreement with predictions from purely passive models with high formation redshifts, although the sky surface density agrees reasonably well with the sorts of models that matched the older MDS and HDF counts, i.e., those with a suitably tuned local LF.

On HST, NICMOS has provided new opportunities to identify and study ellipticals at z 1. Its small field of view, however, has limited the solid angle surveyed. Spectroscopic confirmation of the very faint, very red elliptical candidates identified so far will be exceedingly difficult, but would be well worth the effort. [Treu et al. 1998], [Stiavelli et al. 1999] and [Benítez et al. 1999] noted several very red, R^1/4-law galaxies in the HDF-S NICMOS field, identifying them as ellipticals at 1.4 z_phot 2. Given the small solid angle of that image, this suggests a large space density, and Benítez et al. have proposed that most early type galaxies have therefore evolved only passively since z ~ 2. Comparably red, spheroidal galaxies are found in the NICMOS map of the HDF-N ([Dickinson et al. 2000a]), with photometric redshifts in the range 1.2 z 1.9. Their space density appears to be well below that of HDF ellipticals with similar luminosity at z < 1.1, in broad agreement with [Zepf 1997] and [Franceschini et al. 1998], although the comparison to z = 0 is again limited primarily by uncertainties in the local gE LF. Curiously, few fainter objects with similar colors are found in the HDF-N, although the depth of the NICMOS images is adequate to detect red ellipticals with L ~ 0.1L^* out to z 2. [Treu & Stiavelli 1999] find that PLE models with high (z ~ 5) and low (z ~ 2) formation redshifts over- and under-predict the observed counts, respectively, of red elliptical-like objects in 23 other NICMOS fields.

The HDF-N NICMOS images from [Dickinson et al. 2000a] and [Thompson et al. 1999] are deep enough to have detected red, evolved elliptical galaxies out to at least z ~ 3 if they were present, eliminating concerns about invisibility because of k-corrections (e.g., [Maoz 1997]). The only plausible z > 2 candidate is an HDF-N "J-dropout" object, whose colors resemble those of a maximally old gE at z ~ 3 [Lanzetta et al. 1998, Dickinson et al. 2000b, Lanzetta et al. 1999]. Other z > 2 HDF objects which some authors have morphologically classified as ellipticals (cf. [Fasano et al. 1998]) are blue, mostly very small, manifestly star forming "Lyman break" objects. Few if any appear to be passively evolving objects that have ceased forming stars, so the connection to present-day ellipticals is more speculative.

The collective evidence surveyed above suggests that mature, gE galaxies have been present in the field since z ~ 1 with space densities comparable to that at the present era. At 0.4 < z < 1, however, they exhibit an increasingly broad range of colors and spectral properties, which suggests a variety of star formation histories over the proceeding few billion years (or alternatively errors in classification). The statistics seem to favor a substantial decline in their space density at z >> 1, although the well-surveyed sightlines are small and few, and clustering might (and in fact, apparently does) cause large variations from field to field. Therefore, the conclusion has not been firmly established. Moreover, it is very difficult to achieve uniform selection at all redshifts, regardless of the criteria used (photometric, morphological, or both), and the existing samples of objects with spectroscopic (or at least well-calibrated photometric) redshifts are still small. Thus even the deceptively simple task of comparing elliptical galaxy evolution to the simple PLE hypothesis remains a stubborn challenge.

³ even locally, classification uncertainties may be partly responsible for the widely diverse measurements of the elliptical galaxy LF (see [Loveday et al. 1992, Marzke et al. 1994, Marzke et al. 1998, Zucca et al. 1994, 2000]). Back.