Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/science/k20/papers/jenam02.ps Äàòà èçìåíåíèÿ: Tue Jan 7 13:18:07 2003 Äàòà èíäåêñèðîâàíèÿ: Fri Feb 28 06:58:07 2014 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 31

The formation and evolution of field massive galaxies
Andrea Cimatti
INAF ­ Osservatorio Astrofisico di Arcetri, Italy
Abstract. The problem of the formation and evolution of field massive galaxies is
briefly reviewed from an observational perspective. The motivations and the charac­
teristics of the K20 survey are outlined. The redshift distribution of Ks ! 20 galaxies,
the evolution of the rest­frame Ks­band luminosity function and luminosity density
to z ¸ 1:5, the nature and the role of the red galaxy population are presented. Such
results are compared with the predictions of models of galaxy evolution.
Keywords: galaxies; cosmology
1. Introduction
Despite the recent developments in observational cosmology, one of
the main unsolved issues remains how and when the present­day mas­
sive elliptical galaxies (M stars ? 10 11 M fi ) built up and what type of
evolution characterized their growth across the cosmic time.
There are two main proposed scenarios. In the first, such systems
formed at high redshifts (e.g. z ? 2\Xi3) through a ``monolithic'' collapse
accompanied by a violent burst of star formation, then followed by a
passive and pure luminosity evolution (PLE) of the stellar population to
nowadays (Eggen, Lynden­Bell & Sandage 1962; Tinsley 1972; Larson
1975; van Albada 1982). Such a scenario makes some critical and rigid
predictions that can be tested with the observations: (i) the comoving
number density of massive spheroids should be conserved through cos­
mic times, (ii) massive galaxies should evolve only in luminosity, (iii)
old passively evolving spheroids should exist at least up to z ¸ 1 \Xi 1:5,
(iv) there should be a population of progenitors at z ? 2 \Xi 3 charac­
terized by large amounts of gas (and dust) and strong star formation
rates in order to be compatible with the rapid formation scenario and
with the properties (e.g. masses, ages, metallicities) of the present­day
``fossils'' resulting from that formation process.
In a diametrically opposed scenario, massive spheroids formed at
later times through a slower process of hierarchical merging of smaller
galaxies (e.g. White & Rees 1978; Kauffman, White, & Guiderdoni
1993; Kauffmann 1996) characterized by moderate star formation rates,
thus reaching the final masses in more recent epochs (e.g. z ! 1 \Xi 1:5)
(e.g. Baugh et al. 1996, 1998; Cole et al. 2000; Baugh et al. 2002).
As a consequence, the hierarchical merging models (HMMs) predict
c
fl 2003 Kluwer Academic Publishers. Printed in the Netherlands.
cimatti.tex; 7/01/2003; 11:07; p.1

2 Andrea Cimatti
that massive systems should be very rare at z ¸ 1, with the comoving
density of M stars ? 10 11 M fi galaxies decreasing by almost an order of
magnitude from z ¸ 0 to z ¸ 1 (Baugh et al. 2002; Benson et al. 2002).
Several observations were designed over the recent years in order to
test such two competing models.
One possibility is to search for the starburst progenitors expected at
z ? 2 \Xi 3 in the ``monolithic''+PLE scenario. In this respect, submm
and mm continuum surveys unveiled a population of high­z dusty star­
bursts which may represent the ancestors of the present­day massive
galaxies (see Blain et al. 2002 for a recent review).
The other possibility is to search for passively evolving spheroids to
the highest possible redshifts and to study their properties both in clus­
ters and in the field. This latter approach provided so far controversial
results.
Because of their color evolution, fundamental plane and stellar pop­
ulation properties, cluster ellipticals are now generally believed to form
a homogeneous population of old systems formed at high redshifts (e.g.
Stanford et al. 1998; see also Renzini 1999; Renzini & Cimatti 1999;
Peebles 2002 for recent reviews).
However, the question of field spheroids is still actively debated. It is
now established that old, passive and massive systems exist in the field
out to z ¸ 1:5 (e.g. Spinrad et al. 1997; Stiavelli et al. 1999; Waddington
et al. 2002), but the open question is what are their number density and
physical/evolutionary properties with respect to the model predictions.
Some surveys based on color or morphological selections found a
deficit of z ? 1 \Xi 1:4 elliptical candidates (e.g. Kauffmann et al. 1996;
Zepf 1997; Franceschini et al. 1998; Barger et al. 1999; Rodighiero et
al. 2001; Smith et al. 2002; Roche et al. 2002), whereas others did not
confirm such result out to z ¸ 1 \Gamma 2 (e.g. Totani & Yoshii 1998; Benitez
et al. 1999; Daddi et al. 2000b; Im et al. 2002; Cimatti et al. 2002a).
Part of the discrepancies can be ascribed to the strong clustering (hence
field­to­field variations) of the galaxies with the red colors expected for
high­z elliptical candidates (Daddi et al. 2000).
Other approaches made the picture even more controversial. For in­
stance, Menanteau et al. (2001) found that a fraction of morphologically
selected field spheroidals show internal color variations incompatible
with a traditional PLE scenario and stronger than cluster spheroidals
at the same redshifts. Similar results have been obtained with photo­
metric, spectroscopic and fundamental plane studies of field ellipticals
to z ¸ 0:7 \Xi 1 (e.g. Kodama et al. 1999; Schade et al. 1999; Treu et
al. 2002). Such observations suggest that, despite the mass of massive
spheroids seems not to change significantly from z ¸ 1 to z ¸ 0 (Brinch­
mann & Ellis 2000), field early­type systems at z ¸ 0:5 \Xi 1 do not
cimatti.tex; 7/01/2003; 11:07; p.2

Evolution of massive galaxies 3
form an entirely homogeneous population, some looking consistent with
the PLE scenario, whereas others with signatures of recent secondary
episodes of star formation (see also Ellis 2000 for a review).
A more solid and unbiased approach is to investigate the evolution
of massive galaxies by means of spectroscopic surveys of field galaxies
selected in the K­band (e.g. Broadhurst et al. 1992), and to push the
study of massive systems to z ? 1. Since the rest­frame optical and
near­IR light is a good tracer of the galaxy stellar mass (Gavazzi et
al. 1996), K­band surveys provide the important possibility to select
galaxies according to their mass up to z ¸ 2. The advantages of the
K­band selection also include the small k­corrections with respect to
optical surveys (which are sensitive to the star formation activity rather
than to the stellar mass), and the minor effects of dust extinction. Once
a sample of faint field galaxies has been selected in the K­band, deep
spectroscopy with 8­10m class telescopes can then be performed to
shed light on their nature and on their redshift distribution. Several
spectroscopic surveys of this kind have been and are being performed
(e.g. Cowie et al. 1996; Cohen et al. 1999; Stern et al. 2000; see also
Drory et al. 2001, although mostly based on photometric redshifts).
In this paper, the main results obtained so far with a new spectro­
scopic survey for K­selected field galaxies are reviewed, concentrating
on the the redshift distribution, the evolution of the near­IR luminosity
function and luminosity density, the very red galaxy population, and
on the comparison with the predictions of the most recent scenarios of
galaxy formation and evolution. H 0 = 70 km s \Gamma1 Mpc \Gamma1
,\Omega m = 0:3
and\Omega \Lambda = 0:7 are adopted.
2. The K20 survey
Motivated by the above open questions, we started an ESO VLT Large
Program (dubbed ``K20 survey'') based on 17 nights distributed over
two years (1999­2000) (see Cimatti et al. 2002c for details).
The prime aim of such a survey was to derive the redshift distribu­
tion and spectral properties of 546 K s ­selected objects with the only
selection criterion of K s ! 20 (Vega). Such a threshold is critical be­
cause it selects galaxies over a broad range of masses, i.e. M stars ? 10 10
M fi and M stars ? 4 \Theta 10 10 M fi for z = 0:5 and z = 1 respectively
(according to the mean M stars =L ratio in the local universe and adopt­
ing Bruzual & Charlot 2000 spectral synthesis models with a Salpeter
IMF). The K s ! 20 selection has also the observational advantage
that most galaxies have magnitudes still within the limits of optical
spectroscopy of 8m­class telescopes (R ! 25).
cimatti.tex; 7/01/2003; 11:07; p.3

4 Andrea Cimatti
The targets were selected from K s ­band images (ESO NTT+SOFI)
of two independent fields covering a total area of 52 arcmin 2 . One of the
fields is a sub­area of the Chandra Deep Field South (CDFS; Giacconi
et al. 2000). Optical multi­object spectroscopy was made with the ESO
VLT UT1 and UT2 equipped with FORS1 and FORS2. A fraction of
the sample was also observed with near­IR spectroscopy with VLT
UT1+ISAAC in order to attempt to derive the redshifts of the galaxies
which were too faint for optical spectroscopy and/or expected to be
in a redshift range for which no strong features fall in the observed
optical spectral region (e.g. 1:5 ! z ! 2:0). In addition to spectroscopy,
UBVRIzJK s imaging was also available for both fields, thus providing
the possibility to estimate photometric redshifts for all the objects
in the K20 sample, to optimize them through a comparison with the
spectroscopic redshifts and to assign a reliable photometric redshift to
the objects for which it was not possible to derive the spectroscopic z.
The overall spectroscopic redshift completeness is 94%, 92%, 87% for
K s ! 19:0, 19.5, 20.0 respectively. The overall redshift completeness
(spectroscopic + photometric redshifts) is 98%.
The K20 survey represents a significant improvement with respect
to previous surveys for faint K­selected galaxies (e.g. Cowie et al.
1996; Cohen et al. 1999) thanks to its larger sample, the coverage of
two independent fields (thus reducing the cosmic variance effects), the
availability of optimized photometric redshifts, and the spectroscopic
redshift completeness, in particular for the reddest galaxies.
3. The redshift distribution of K s ! 20 galaxies
The observed differential and cumulative redshift distributions for the
K20 sample are presented in Fig. 1 (see Cimatti et al. 2002b), together
with the predictions of different scenarios of galaxy formation and evo­
lution, including both hierarchical merging models (HMMs) from Menci
et al. (2002, M02), Cole et al. (2000, C00), Somerville et al. (2001, S01),
and pure luminosity evolution models (PLE) based on Pozzetti et al.
(1996,1998 PPLE) and Totani et al. (2001, TPLE). The redshift distri­
bution can be retrieved from http://www.arcetri. astro.it/ k20/releases.
The spike at z ¸ 0:7 is due to two clusters (or rich groups) at z = 0:67
and z = 0:73. The median redshift of N(z) is z med = 0:737 and
z med = 0:805, respectively with and without the two clusters being
included. Without the clusters, the fractions of galaxies at z ? 1 and
z ? 1:5 are 138/424 (32.5%) and 39/424 (9.2%) respectively. The
high­z tail extends beyond z = 2. The contribution of objects with
only a photometric redshift becomes relevant only for z ? 1:5. The
cimatti.tex; 7/01/2003; 11:07; p.4

Evolution of massive galaxies 5
Figure 1. Fig. 1a -- Top panels: the observed differential N(z) for Ks ! 20 (his­
togram) compared with the PLE model predictions. Bottom panels: the observed
fractional cumulative redshift distribution (continuous line) compared with the same
models. The shaded histogram shows the contribution of photometric redshifts. The
bin at z ! 0 indicates the 9 objects without redshift. The left and right panels
show the models without and with the inclusion of the photometric selection effects
respectively. Sc and Sp indicate Scalo and Salpeter IMFs respectively. Fig. 1b --
same as Fig. 1a, but compared with the HMM predictions. Right panels: the M02
model with the inclusion of the photometric selection effects.
fractional cumulative distributions displayed in Fig. 1 (bottom panels)
were obtained by removing the two clusters mentioned above in order
to perform a meaningful comparison with the galaxy formation models
which do not include clusters (PLE models), or are averaged over very
large volumes, hence diluting the effects of redshift spikes (HMMs).
No best tuning of the models was attempted in this comparison, thus
allowing an unbiased blind test with the K20 observational data. The
model predicted N(z) are normalized to the K20 survey sky area.
Fig. 1a shows a fairly good agreement between the observed N(z)
distribution and the PLE models (with the exception of PPLE with
Salpeter IMF), although such models slightly overpredict the number
of galaxies at z ¸ ? 1:2. However, if the photometric selection effects
present in the K20 survey (Cimatti et al. 2002b) are taken into account,
the PLE models become much closer to the observed N(z) thanks to
the decrease of the predicted high­z tail. According to the Kolmogorov­
Smirnov test, the PLE models are acceptable at 95% confidence level,
with the exception of the PPLE model with Salpeter IMF.
On the other side, all the HMMs underpredict the median redshift
(z med =0.59, 0.70 and 0.67 for the C00, M02 and S01 models respec­
cimatti.tex; 7/01/2003; 11:07; p.5

6 Andrea Cimatti
Figure 2. The observed cumulative number of galaxies between 1 ! z ! 3 (continu­
ous line) and the corresponding poissonian \Sigma3oe confidence region (dotted lines). The
PPLE (Scalo IMF) and the M02 models are corrected for the photometric biases.
tively), overpredict the total number of galaxies with K s ! 20 by
factors up to ¸50% as well as the number of galaxies at z ! 0:5, and
underpredict the fractions of z ? 1\Xi1:5 galaxies by factors of 2\Xi4 (Fig.
1b). Fig. 1b (bottom panels) illustrates that in the fractional cumulative
distributions the discrepancy with observations appears systematic at
all redshifts. The Kolmogorov­Smirnov test shows that all the HMMs
are discrepant with the observations at ? 99% level. The inclusion
of the photometric biases exacerbates this discrepancy, as shown in
Fig. 1b (right panels) for the M02 model (the discrepancy for the C00
and S01 models becomes even stronger). The deficit of high­redshift
objects is well illustrated by Fig. 2, where the PPLE model is capable to
reproduce the cumulative number distribution of galaxies at 1 ! z ! 3
within 1­2oe, whereas the M02 model is always discrepant at – 3oe level
(up to ? 5oe for 1:5 ! z ! 2:5). This conclusion is not heavily based
on the objects with only photometric redshifts estimates, as the mere
presence of 7 galaxies with spectroscopic redshift z ? 1:6 is already
in substantial contrast with the predictions by HMMs of basically no
galaxies with K s ! 20 and z ? 1:6.
cimatti.tex; 7/01/2003; 11:07; p.6

Evolution of massive galaxies 7
Figure 3. The rest­frame Ks­band Luminosity Function in three redshift bins. Data
points derive from 1=Vmax analysis. Solid curves: the Schechter fits derived from
maximum likelihood analysis (thin solid lines are the fit assuming local ff parameter).
Dotted curve: the local Ks­band LF of Cole et al. (2001). Open circles: spectroscopic
redshifts, filled circles: spectroscopic + photometric redshifts.
4. The evolution of the luminosity function
The luminosity function of galaxies has been estimated in the rest­
frame K s ­band and in three redshift bins (z mean =0.5,1,1.5) (Pozzetti
et al. 2002), using both the 1/V max (Schmidt 1968; Felten 1976) and the
STY (Sandage, Tammann & Yahil 1978) formalisms. The LF observed
in the first two redshift bins is fairly well fit by Schechter functions.
A comparison with the local K s ­band LF of Cole et al. (2001) shows
a mild luminosity evolution of LF(z) out to z = 1, with a brightening
of about ­0.5 magnitudes from z = 0 to z = 1 (Fig. 3). Similar results
have been found by Drory et al. (2001), Cohen (2002), Bolzonella et al.
(2002) and Miyazaki et al. (2002) (see also Cowie et al. 1996).
The study of the LF by galaxy spectral or color types shows that red
early­type galaxies dominate the bright­end of the LF already at z ¸ 1,
and that their number density shows only a small decrease from z ¸ 0
to z ¸ 1 (Pozzetti et al. 2002). This is consistent with the independent
study of Im et al. (2002) based on morphologically selected spheroidals.
cimatti.tex; 7/01/2003; 11:07; p.7

8 Andrea Cimatti
Figure 4. Left: the Ks­band LF compared to hierarchical merging model predictions.
Right: the Ks­band LF compared to PLE model predictions.
Fig. 4 shows a comparison of the observed luminosity function with
PLE and HMM predictions. The PLE models describe reasonably well
the shape and the evolution of the luminosity function up to the highest
redshift bin, z mean = 1:5, with no evidence for a strong decline of the
most luminous systems (with L ? L \Lambda ). This is in contrast, especially in
the highest redshift bin, with the prediction by the HMMs of a decline
in the number density of luminous (i.e. massive) systems with redshift.
Moreover, hierarchical merging models (namely M02 and C00) result
in a significant overprediction of faint, sub--L \Lambda galaxies at 0 ! z ! 1:3.
This problem, also evident in the comparison with N(z), is probably
related to the so called ``satellite problem'' (e.g.Primack 2002).
However, it is interesting to note that at z ¸ 1 the HMMs seem
not to be in strong disagreement with the observations relative to the
bright end of the galaxy luminosity function. Thus, the key issue is
to verify whether the bright L ? L \Lambda galaxies in the K20 survey have
the same nature of the luminous galaxies predicted by the HMMs, in
particular for their mass to light ratios (M stars =L).
Fig. 5 compares the R \Gamma K s colors and luminosity distributions
of galaxies with 0:75 ! z ! 1:3 (a bin dominated by spectroscopic
redshifts) as observed in our survey to the predictions of the GIF 1
simulations (Kauffmann et al. 1999). Such a comparison highlights
that a relevant discrepancy is present between the two distributions:
real galaxies with MK \Gamma 5logh 70 ! \Gamma24:5 in the K20 sample have a
1 http:// www.mpa­garching.mpg.de/GIF/
cimatti.tex; 7/01/2003; 11:07; p.8

Evolution of massive galaxies 9
Figure 5. Left panel: R \Gamma Ks colors vs. rest­frame absolute Ks magnitudes for
z = 1:05 GIF simulated catalog (small dots) and data (circles) at 0:75 ! z ! 1:3
(zmean = 1) (empty and filled circles refer to z ! 1 and z ? 1 respectively). Verti­
cal dashed line represents approximately the completeness magnitude limit of GIF
catalog corresponding to its mass limit (see text). Right panel: Color distribution of
luminous galaxies (MKs \Gamma 5 log h70 ! \Gamma24:5) observed (dotted line) and simulated
(continuous line), normalized to the same comoving volume.
median color of R \Gamma K s ¸ 5, whereas the GIF simulated galaxies have
R \Gamma K s ¸ 4, and the two distributions have very small overlap. Given
that red galaxies have old stellar populations and higher M stars =L
ratios, the apparent agreement with HMM predictions of the z ¸ 1
bright end of the luminosity function (Fig. 4) is fortuitous and probably
results form an underestimate of the M stars =L present in the same
models. This is equivalent to say that the number density of massive
galaxies at z ¸ 1 is underpredicted by HMMs, and the predicted colors,
ages and star formation rates do not agree with the observations.
5. The evolution of the luminosity density
Tracing the integrated cosmic emission history of the galaxies at dif­
ferent wavelengths offers the prospect of an empirical determination of
the global evolution of the galaxy population. Indeed it is independent
of the details of galaxy evolution and depends mainly on the star
cimatti.tex; 7/01/2003; 11:07; p.9

10 Andrea Cimatti
formation history of the universe (Lilly et al. 1996, Madau, Pozzetti
& Dickinson 1998). Attempts to reconstruct the cosmic evolution of
the comoving luminosity density have been made previously mainly in
the UV and optical bands, i.e. focusing on the star formation history
activity of galaxies (Lilly et al. 1996, Cowie et al. 1999).
Our survey offers for the first time the possibility to investigate it in
the near­IR using a LF extended over a wide range in luminosity, thus
providing new clues on the global evolution of the stellar mass density
(Pozzetti et al. 2002). Using the observed local luminosity density de­
rived from Cole et al. (2001), the K s ­band luminosity density up to z ¸
1:3 is well represented by a power law with ae – (z) = ae – (z = 0)(1+z) 0:37 .
Compared to the UV and optical bands, the near­IR luminosity density
evolution is much slower (fi = 3:9 \Xi 2:7 from 0.28 to 0.44 ¯m by Lilly et
al. 1996 and fi = 1:5 at 0.15, 0.28 ¯m by Cowie et al. 1999,
for\Omega m = 1).
The slow evolution of the observed K s ­band luminosity density suggests
that the stellar mass density should also evolve slowly at least up to
z ¸ 1:3. This is in agreement with a recent analysis by Bolzonella et
al. (2002) (see also Cowie et al. 1996 and Brinchmann & Ellis 2000).
The analysis of the stellar mass function and its cosmic evolution is in
progress and will be presented elsewhere.
6. Extremely Red Objects (EROs)
Extremely Red Objects (EROs, R \Gamma K ? 5) are critical in the context
of galaxy formation and evolution because their colors allow to select
old and passively evolving galaxies at z ? 0:9.
For a fraction of EROs (70% to K s ! 19:2) present in the K20
sample it was possible to derive a spectroscopic redshift and a spectral
classification (Cimatti et al 2002a). Two classes of galaxies at z ¸ 1
contribute nearly equally to the ERO population: old stellar systems
with no signs of star formation, and dusty star­forming galaxies.
6.1. Old EROs
The colors and spectral properties of old EROs are consistent with –3
Gyr old passively evolving stellar populations (assuming solar metal­
licity and Salpeter IMF), requiring a formation redshift z f ? 2:4. The
number density is 6:3 \Sigma 1:8 \Theta 10 \Gamma4 h 3 Mpc \Gamma3 for K s ! 19:2, consistent
with the expectations of PLE models for passively evolving early­type
galaxies with similar formation redshifts (Cimatti et al. 2002a). HMMs
predict a significant deficit of such old red galaxies at z ¸ 1, ranging
from a factor of ¸ 3 (Kauffmann et al. 1999) to a factor of ¸ 5 (Cole et
cimatti.tex; 7/01/2003; 11:07; p.10

Evolution of massive galaxies 11
Figure 6. The average rest­frame spectra (smoothed with a 3 pixel boxcar) of old
passively evolving (top; zmean = 1:000) and dusty star­forming EROs (bottom;
zmean = 1:096) with Ks Ÿ 20 (Cimatti et al. 2002a).
al. 2000). Preliminary analysis of recent HST+ACS imaging shows that
old EROs have indeed spheroidal morphologies with surface brightness
profiles typical of elliptical galaxies.
6.2. Dusty star­forming EROs
The spectra of star­forming EROs suggest a dust reddening of E(B \Gamma
V ) ¸ 0:5--1 (adopting the Calzetti extinction law), implying typical
star­formation rates of 50­150 M fi yr \Gamma1 , and a significant contribution
(? 20 \Gamma 30%) to the cosmic star­formation density at z ¸ 1 (see also
Smail et al. 2002). A recent analysis based on their X­ray emission
provided a similar estimate of the SFRs (Brusa et al. 2002).
The comoving density of dusty EROs is again ¸ 6 \Theta 10 \Gamma4 h 3 Mpc \Gamma3
at K s ! 19:2. The GIF simulations (Kauffmann et al. 1999) predict
a comoving density of red galaxies with SFR ? 50 M fi yr \Gamma1 that is a
factor of 30 lower than the observed density of dusty EROs.
Such moderate SFRs suggest that the far­infrared luminosities of
dusty star­forming EROs are generally below L F IR ¸ 10 12 L fi , and
would then explain the origin of the low detection rate of EROs with
Ks ! 20 \Xi 20:5 in submm continuum observations (e.g. Mohan et
cimatti.tex; 7/01/2003; 11:07; p.11

12 Andrea Cimatti
al. 2001; see also Smail et al. 2002). However, the fraction of dusty
ultraluminous infrared systems may be higher in ERO samples selected
at fainter Ks­band magnitudes (e.g. Wehner et al. 2002).
6.3. Clustering
Taking advantage of the spectroscopic redshift information for the two
ERO classes, we compared the relative 3D clustering in real space
(Daddi et al. 2002). The comoving correlation lengths of dusty and
old EROs are constrained to be r 0 ! 2:5 and 5:5 ! r 0 ! 16 h \Gamma1 Mpc
comoving respectively, implying that old EROs are the main source of
the ERO strong angular clustering. It is important to notice that the
strong clustering measured for the old EROs is in agreement with the
predictions of hierarchical clustering scenarios (Kauffmann et al. 1999).
7. Summary and discussion
The high level of completeness of the K20 survey and the relative set
of results presented in previous sections provide new implications for a
better understanding of the evolution of ``mass­selected'' field galaxies.
(1) The redshift distribution of K s ! 20 field galaxies has a median
redshift of z med ¸ 0:8 and a high­z tail extended beyond z ¸ 2. The
current models of hierarchical merging do not match the observed me­
dian redshift because they significantly overpredict the number of low
luminosity (hence low mass) galaxies at z ! 0:4 \Xi 0:5, and underpredict
the fraction of objects at z ? 1 \Xi 1:5. Instead, the redshift distributions
predicted by PLE models are in reasonable agreement with the observa­
tions. It is relevant to recall here that early predictions of the expected
fraction of galaxies at z ? 1 in a K s ! 20 sample indicated respectively
ú 60% and ú 10% for a PLE case and for a (then)
standard\Omega m = 1
CDM model (Kauffmann & Charlot 1998). This version of PLE was
then ruled out by Fontana et al. (1999). The more recent PLE models
and HMMs consistently show that for z ? 1 the difference between the
predictions of different scenarios is much less extreme. These results
come partly from the now favored \LambdaCDM cosmology which pushes most
of the merging activity in hierarchical models at earlier times compared
to ÜCDM and SCDM models
with\Omega m = 1 (structures form later in
a matter­dominated universe, thus resulting in an even lower fraction
of galaxies at high­z), and partly to different recipes for merging and
star formation modes, which tend to narrow the gap between HMMs
and the PLE case (e.g. Somerville et al. 2001; Menci et al. 2002). In
this respect, the observed N(z) provides an additional evidence that
the universe is not
matter­dominated(\Omega m ! 1).
cimatti.tex; 7/01/2003; 11:07; p.12

Evolution of massive galaxies 13
(2) The rest­frame K s ­band luminosity function shows a mild lumi­
nosity evolution up to at least z ¸ 1, with a brightening of about 0.5
magnitudes. Significant density evolution is ruled out up to z ¸ 1.
Current hierarchical merging models fail in reproducing the shape and
evolutionary properties of the LF because they overpredict the number
of sub­L \Lambda galaxies and predict a substantial density evolution. PLE
models are in good agreement with the observations up to z ¸ 1.
(3) At odds with the HMMs, the bright­end of the LF at z ¸ 1 is
dominated by red and luminous (hence old and massive) galaxies.
(4) The rest­frame K s ­band luminosity density (hence the stellar mass
density) evolves slowly up to z ¸ 1:3.
(5) Old passive systems and dusty star­forming galaxies (both at z ¸ 1)
equally contribute to the ERO population with K s ! 19:2.
(6) The number, luminosities and ages of old EROs imply that massive
spheroids formed at z ? 2:4 and that were already fully assembled at
z ¸ 1, consistently with a PLE scenario.
(7) Dusty EROs allow to select (in a way complementary to other sur­
veys for star­forming systems) a population of galaxies which contribute
significantly to the cosmic star formation budget at z ¸ 1.
(8) HMMs strongly underpredict the number of both ERO classes.
Overall, the results of the K20 survey show that galaxies selected
in the K s ­band are characterized by little evolution up to z ¸ 1,
and that the observed properties can be successfully described by a
PLE scenario. In contrast, HMMs fail in reproducing the observations
because they predict a sort of ``delayed'' scenario where the assembly of
massive galaxies occurs later than what is actually observed. We recall
here that the discrepancies of HMMs in accounting for the properties of
even z = 0 !¸ 1 early­type galaxies have been already emphasized in
the past (e.g., Renzini 1999; Renzini & Cimatti 1999). Moreover, among
low­redshift galaxies there appears to be a clear anti­correlation of the
specific star formation rate with galactic mass (Gavazzi et al. 1996;
Boselli et al. 2001), the most massive galaxies being ``old'', the low­mass
galaxies being instead dominated by young stellar populations. This is
just the opposite than expected in the traditional HMMs, where the
most massive galaxies are the last to form. The same anti­correlation
is observed in the K20 survey at z ¸ 1.
It is important to stress here that the above results do not necessarily
mean that the whole framework of hierarchical merging of CDM halos
is under discussion. For instance, the strong clustering of old EROs
and the clustering evolution of the K20 galaxies (irrespective of colors)
seem to be fully consistent with the predictions of CDM models of large
scale structure evolution (Daddi et al. 2001; Firth et al. 2002; Daddi et
al. in preparation).
cimatti.tex; 7/01/2003; 11:07; p.13

14 Andrea Cimatti
It is also important to stress that the K20 survey allows to perform
tests which are sensitive to the evolutionary ``modes'' of galaxies rather
than to their formation mechanism. This means that merging, as the
galaxy main formation mechanism, is not ruled out by the present ob­
servations. Also, it should be noted that PLE models are not a physical
alternative to the HMMs, but rather tools useful to parameterize the
evolution of galaxies under three main assumptions: high formation
redshift, conservation of number density through cosmic times, passive
and luminosity evolution of the stellar populations.
Thus, if we still accept the \LambdaCDM scenario of hierarchical merging of
dark matter halos as the basic framework for structure and galaxy for­
mation, the observed discrepancies highlighted by the K20 survey may
be ascribed to how the baryon assembly is treated and, in particular, to
the heuristic algorithms adopted for the star formation processes and
their feedback, both within individual galaxies and in their environ­
ment. Our results suggest that HMMs should have galaxy formation in
a CDM dominated universe to closely mimic the old­fashioned mono­
lithic collapse scenario. This requires to enhance merging and star
formation in massive halos at high redshift (say, z ¸ ? 2 \Xi 3), while in the
meantime suppressing star formation in low­mass halos. For instance,
Granato et al. (2001) suggested the strong UV radiation feedback from
the AGN activity during the era of supermassive black hole formation
to be responsible for the suppression of star formation in low­mass
halos, hence imprinting a ``anti­hierarchical'' behavior in the baryonic
component. The same effect may well result from the feedback by the
starburst activity itself (see also Ferguson & Babul 1998).
In summary, the redshift distribution of K s ! 20 galaxies, together
with the space density, nature, and clustering properties of the ERO
population, and the redshift evolution of the rest­frame near­IR lu­
minosity function and luminosity density provide a new set of ob­
servables on the galaxy population in the z ¸ 1 \Gamma 2 universe, thus
bridging the properties of z ¸ 0 galaxies with those of Lyman­break
and submm/mm­selected galaxies at z – 2--3. This set of observables
poses a new challenge for theoretical models to properly reproduce.
Deeper spectroscopy coupled with HST+ACS imaging and SIRTF
photometry will allow us to derive additional constraints on the nature
and evolution of massive stellar systems out to higher redshifts.
Acknowledgements
The K20 survey team includes: S. Cristiani (INAF­Trieste), S. D'Odorico
(ESO), A. Fontana (INAF­Roma), E. Giallongo (INAF­Roma), R. Gilmozzi
cimatti.tex; 7/01/2003; 11:07; p.14

Evolution of massive galaxies 15
(ESO), N. Menci (INAF­Roma), M. Mignoli (INAF­Bologna), F. Poli
(University of Rome), A. Renzini (ESO), P. Saracco (INAF­Brera), J.
Vernet (INAF­Arcetri), and G. Zamorani (INAF­Bologna).
We are grateful to C. Baugh, R. Somerville and T. Totani for pro­
viding their model predictions. AC warmly acknowledges Jim Peebles
and Mark Dickinson for useful and stimulating discussions.
References
Barger A.J. et al. 1999, AJ, 117, 102
Baugh C.M., Cole S. & Frenk C.S., 1996, MNRAS 283, 1361
Baugh C.M., Cole S. & Frenk C.S., Lacey C.G. 1998, ApJ, 498, 504
Baugh C.M. et al. 2002, in proceedings `The Mass of Galaxies at Low and High
Redshift', Venice 2001, eds. R. Bender, A. Renzini (astro­ph/0203051)
Benitez N. et al. 1999, ApJ, 515, L65
Benson A.J., Ellis R.S., Menanteau F. 2002, MNRAS, 336, 564
Blain A.W., Smail I., Ivison R.J., Kneib J.­P., Frayer D.T. 2002, Physics Reports,
in press, astro­ph/0202228
Bolzonella M. et al. 2002, A&A, in press, astro­ph/0207501
Boselli A., Gavazzi G., Donas J. & Scodeggio M. 2001, AJ, 121, 753
Brinchmann J., Ellis R.S. 2000, ApJ, 536, L77
Broadhurst, T., Ellis, R.S., & Grazebrook, K. 1992, Nature, 355, 55
Brusa M. et al. 2002, ApJL, in press
Cimatti A., Daddi E., Mignoli M. et al. 2002a, A&A, 381, L68
Cimatti A., Pozzetti L., Mignoli M., et al. 2002b, A&A, 391, L1
Cimatti A., Mignoli M., Daddi E. et al. 2002c, A&A, 392, 395
Cohen J.G., Blandford R., Hogg D.W. et al. 1999, ApJ, 512, 30
Cohen J.G. 2002, ApJ, 567, 672
Cole S., Lacey C.G., Baugh C.M. & Frenk C.S. 2000, MNRAS, 319, 168
Cole S., Norberg P., Baugh C.M. et al. 2001, MNRAS, 326, 255
Cowie, L.L., Songaila A., Hu E.M. & Cohen J.G. 1996, AJ,112,839
Cowie, L.L., Songaila A. & Barger A.J. 1999, AJ, 118, 603
Daddi E. et al. 2000, A&A, 361, 535
Daddi E., Cimatti A. & Renzini A., 2000, A&A 362, L45
Daddi E., Broadhurst T., Zamorani G. et al. 2001, A&A, 376,825
Daddi E., Cimatti A., Broadhurst T. et al. 2002, A&A, 384, L1 (Paper II)
Drory N., Bender R., Snigula J. et al. 2001, ApJ, 562, L111
Eggen O.J., Lynden­Bell D., Sandage A. 1962, ApJ, 136, 748
Ellis, R.S., in XIth Canary Islands Winter School of Astrophysics ``Galaxies at High
Redshift'', astro­ph/0102056
Felten J.E. 1976, ApJ, 207, 700
Ferguson H.C. & Babul A. 1998, MNRAS, 296, 585
Firth A.E., Somerville R.S., McMahon R.G. et al. 2002, MNRAS, 332, 617
Fontana A., Menci N., D'Odorico S. et al. 1999, MNRAS, 310, L27
Franceschini A. et al. 1998, ApJ, 506, 600
Gavazzi G., Pierini D. & Boselli A. 1996, A&A, 312, 397
Giacconi R., Rosati P., Tozzi P. et al. 2001, ApJ,551,624
Granato G.L., Silva L., Monaco P. et al. 2001, MNRAS, 324, 757
cimatti.tex; 7/01/2003; 11:07; p.15

16 Andrea Cimatti
Im M. et al. 2002, ApJ, 571, 136
Kauffmann G., White S.D.M., Guiderdoni B. 1993, MNRAS, 264, 201
Kauffmann G. 1996, MNRAS, 281, 487
Kauffmann G.,Charlot S., White S.D.M. 1996, MNRAS, 283, L117
Kauffmann G., Charlot S. 1998, MNRAS, 297, L23
Kauffmann G. et al. 1999, MNRAS, 303, 188
Kodama T., Bower R.G., Bell E.F. 1999, MNRAS, 306, 561
Larson R.B. 1974, MNRAS, 173, 671
Lilly, S.J. et al. 1996, ApJ, 460, L1
Madau P., Pozzetti L. & Dickinson M., 1998, ApJ, 498, 106
Menateau F., Abraham R.G., Ellis R.S. 2001, MNRAS, 322, 1
Menci N., Cavaliere A., Fontana A., Giallongo E. & Poli F. 2002, ApJ, 575, 18
Miyazaki M. et al. 2002, ApJ submitted, astro­ph/0210509
Mohan N.R. et al. 2002, A&A, 383, 440
Peebles P.J.E. 2002, astro­ph/0201015
Pozzetti L. Bruzual A.G. & Zamorani, G. 1996, MNRAS, 281, 953
Pozzetti L. et al. 1998, MNRAS, 298, 1133
Pozzetti L. et al. 2002, A&A, submitted
Primack J.R. 2002, astro­ph/0205391
Renzini A. 1999, in The Formation of Galactic Bulges, ed. C.M. Carollo, H.C.
Ferguson, & R.F.G. Wyse (Cambridge: CUP), p. 9
Renzini A., Cimatti A. 1999, ASP Conf. Ser. 193, 312
Rodighiero G., Franceschini A., Fasano G. 2001, MNRAS, 324, 491
Roche N.D. et al. 2002, MNRAS, in press, astro­ph/0205259
Sandage, A., Tammann, G.A., Yahil, A., 1979, ApJ, 232, 352
Schade D. et al. 1999, ApJ, 525, 31
Schmidt, M., 1968, ApJ, 151, 393
Smail I. et al. 2002, ApJ, in press, astro­ph/0208434
Smith G.P. et al. 2002, MNRAS, 330, 1
Somerville R.S, Primack J.R. & Faber S.M. 2001,MNRAS,320,504
Spinrad H. et al. 1997, ApJ, 484, 581
Stanford S.A., Eisenhardt P.R., Dickinson M. 1998, ApJ, 492,461
Stern D., Connolly A., Eisenhardt P. et al. 2001, in ``Deep Fields'', Proceedings of
the ESO Workshop, Garching, Germany, Springer­Verlag, p. 76
Stiavelli M. et al. 1999, A&A, 343, L25
Tinsley B.M. 1972, ApJ, 178, 319
Totani T., Yoshii Y. 1998, ApJ, 501, L177
Totani T., Yoshii Y., Maihara T., Iwamuro F. & Motohara K. 2001, ApJ, 559, 592
Treu T. et al. 2002, ApJ, 564, L13
van Albada T.S. 1982, MNRAS, 201, 939
Waddington I. et al. 2002, MNRAS, in press, astro­ph/0207663
Wehner E.H., Barger A.J., Kneib J.­P. 2002, ApJ, 577, L83
White S.D.M., Rees M.J. 1978, MNRAS, 183, 341
Zepf S.E. 1997, Nature, 390, 377
cimatti.tex; 7/01/2003; 11:07; p.16