Galaxy luminosity function per morphological type up to z=1.2

Abstract

We have computed the evolution of the rest-frame B-band luminosity function (LF) for bulge and disk-dominated galaxies since z=1.2. We use a sample of 605 spectroscopic redshifts with I_{AB}<24 in the Chandra Deep Field South from the VIMOS-VLT Deep Survey, 3555 galaxies with photometric redshifts from the COMBO-17 multi-color data, coupled with multi-color HST/ACS images from the Great Observatories Origin Deep Survey. We split the sample in bulge- and disk-dominated populations on the basis of asymmetry and concentration parameters measured in the rest-frame B-band. We find that at z=0.4-0.8, the LF slope is significantly steeper for the disk-dominated population (\alpha=-1.19 \pm 0.07) compared to the bulge-dominated population (\alpha=-0.53 \pm 0.13). The LF of the bulge-dominated population is composed of two distinct populations separated in rest-frame color: 68% of red (B-I)_{AB}>0.9 and bright galaxies showing a strongly decreasing LF slope \alpha=+0.55 \pm 0.21, and 32% of blue (B-I)_{AB}<0.9 and more compact galaxies which populate the LF faint-end. We observe that red bulge-dominated galaxies are already well in place at z~1, but the volume density of this population is increasing by a factor 2.7 between z~1 and z~0.6. It may be related to the building-up of massive elliptical galaxies in the hierarchical scenario. In addition, we observe that the blue bulge-dominated population is dimming by 0.7 magnitude between z~1 and z~0.6. Galaxies in this faint and more compact population could possibly be the progenitors of the local dwarf spheroidal galaxies.

Full text

Page 1

arXiv:astro-ph/0604010v1 2 Apr 2006
Astronomy & Astrophysics manuscript no.
(DOI: will be inserted by hand later)
February 5, 2008
The VIMOS-VLT Deep Survey⋆
Galaxy luminosity function per morphological type up to z = 1.2
O. Ilbert1,7, S. Lauger1, L. Tresse1, V. Buat1, S. Arnouts1, O. Le F` evre1, D. Burgarella1, E. Zucca2,
S. Bardelli2, G. Zamorani2, D. Bottini3, B. Garilli3, V. Le Brun1, D. Maccagni3, J.-P. Picat4, R.
Scaramella5, M. Scodeggio3, G. Vettolani5, A. Zanichelli5, C. Adami1, M. Arnaboldi6, M. Bolzonella7,
A. Cappi2, S. Charlot8,9, T. Contini4, S. Foucaud3, P. Franzetti3, I. Gavignaud4,12, L. Guzzo10,
A. Iovino10, H.J. McCracken9,11, B. Marano7, C. Marinoni1, G. Mathez4, A. Mazure1, B. Meneux1,
R. Merighi2, S. Paltani1, R. Pello4, A. Pollo10, L. Pozzetti2, M. Radovich6, M. Bondi5, A. Bongiorno7,
G. Busarello6, P. Ciliegi2, Y. Mellier9,11, P. Merluzzi6, V. Ripepi6, and D. Rizzo4
1Laboratoire d’Astrophysique de Marseille (UMR 6110), CNRS-Universit´ e de Provence, B.P.8, 13376 Marseille
C´ edex 12, France
2INAF-Osservatorio Astronomico di Bologna, via Ranzani 1, 40127 Bologna, Italy
3INAF-IASF, via Bassini 15, 20133 Milano, Italy
4Laboratoire d’Astrophysique de l’Observatoire Midi-Pyr´ en´ ees (UMR 5572), CNRS-Universit´ e Paul Sabatier,
14 avenue E. Belin, 31400 Toulouse, France
5INAF-IRA, via Gobetti 101, 40129 Bologna, Italy
6INAF-Osservatorio Astronomico di Capodimonte, via Moiariello 16, 80131 Napoli, Italy
7Universit` a di Bologna, Dipartimento di Astronomia, via Ranzani 1, 40127 Bologna, Italy
8Max-Planck-Institut f¨ ur Astrophysik, Karl-Schwarzschild-Str. 1, 85740 Garching bei M¨ unchen, Germany
9Institut d’Astrophysique de Paris (UMR 7095), 98 bis Boulevard Arago, 75014 Paris, France
10INAF-Osservatorio Astronomico di Brera, via Brera 28, 20121 Milano, Italy
11Observatoire de Paris-LERMA, 61 avenue de l’Observatoire, 75014 Paris, France
12European Southern Observatory, Karl-Schwarzschild-Str. 2, 85748 Garching bei M¨ unchen, Germany
Received ... / Accepted ...
Abstract. We have computed the evolution of the rest-frame B-band luminosity function (LF) for bulge and
disk-dominated galaxies since z = 1.2. We use a sample of 605 spectroscopic redshifts with IAB ≤ 24 in the
Chandra Deep Field South from the VIMOS-VLT Deep Survey, 3555 galaxies with photometric redshifts from the
COMBO-17 multi-color data, coupled with multi-color HST/ACS images from the Great Observatories Origin
Deep Survey. We split the sample in bulge- and disk-dominated populations on the basis of asymmetry and
concentration parameters measured in the rest-frame B-band. We find that at z = 0.4 − 0.8, the LF slope
is significantly steeper for the disk-dominated population (α = −1.19 ± 0.07) compared to the bulge-dominated
population (α = −0.53±0.13). The LF of the bulge-dominated population is composed of two distinct populations
separated in rest-frame color : 68% of red (B − I)AB > 0.9 and bright galaxies showing a strongly decreasing LF
slope α = +0.55 ± 0.21, and 32% of blue (B − I)AB < 0.9 and more compact galaxies which populate the LF
faint-end. We observe that red bulge-dominated galaxies are already well in place at z ≃ 1, but the volume density
of this population is increasing by a factor 2.7 between z ∼ 1 and z ∼ 0.6. It may be related to the building-up
of massive elliptical galaxies in the hierarchical scenario. In addition, we observe that the blue bulge-dominated
population is dimming by 0.7 magnitude between z ∼ 1 and z ∼ 0.6. Galaxies in this faint and more compact
population could possibly be the progenitors of the local dwarf spheroidal galaxies.
Key words. surveys – galaxies: evolution – galaxies: luminosity function – galaxies: morphology
Send
olivier.ilbert1@bo.astro.it
⋆Based on data obtained with the European Southern
Observatory on Paranal, Chile.
offprintrequeststo: O. Ilbert,e-mail:
1. Introduction
A central issue to understand galaxy formation is to study
the building up of the Hubble sequence. One approach is
to measure the evolution in numbers and luminosity of
different galaxy types using the luminosity function (LF).

Page 2

2 Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.2
To this effect, large samples of galaxies are required with
a robust estimate of distances, luminosities and morpho-
logical types.
At low as well as at high redshifts, most luminos-
ity functions have been measured using galaxy sam-
ples classified by spectral type (e.g. Madgwick et al. 2002,
de Lapparent et al. 2003) or by photometric type (e.g.
Lilly et al. 1995, Wolf et al. 2003, Zucca et al. 2006). The
direct interpretation of these results in the framework of
a galaxy formation scenario is not straightforward since
galaxies move from one spectral class to another by a pas-
sive evolution of their stellar population. Another way to
define the galaxy type is to define a morphological type
from the measurement of the structural parameters of the
galaxies from image analysis. Even if star formation evo-
lution could affect galaxy morphologies, a morphological
classification is less sensitive to the star formation history
than a classification based on the spectral energy distri-
bution. This classification is also more robust to follow
similar galaxies at different redshifts.
Unfortunately, it is difficult to assemble large sam-
ples of morphologically classified galaxies with measured
spectroscopic redshifts for z > 0.3, as high resolution im-
ages are required to perform a morphological classifica-
tion and a large amount of telescope time is needed to
measure spectroscopic redshifts. This is why the largest
samples of galaxies with both high resolution morphol-
ogy and spectroscopic redshifts are not exceeding ∼ 300−
400 spectroscopic redshifts (e.g. Brinchmann et al. 1998,
Cassata et al. 2005). Larger samples of galaxies are ob-
tained using the photometric redshifts method (e.g.
Wolf et al. 2005, Bell et al. 2005), but a major drawback
of the photometric redshift method is the difficulty to con-
trol the systematic uncertainties which are affecting the
redshift estimate and to quantify the impact of these sys-
tematics on the LF estimate.
This paper presents a study of the evolution of the
galaxy LF as a function of morphological type. We
use the spectroscopic redshifts from the VIMOS VLT
Deep Survey on the Chandra Deep Field South (CDFS;
Le F` evre et al. 2004). The spectroscopic sample selected
at IAB
≤ 24 is twice larger than previous spectro-
scopic samples at similar redshifts. We consolidate the re-
sults using 3555 photometric redshifts estimated from the
COMBO-17 multi-color data. We check that photometric
redshifts are not creating a systematic bias in the LF mea-
surement from a detailed comparison between photomet-
ric redshift and spectroscopic redshift results. The mor-
phological classification is performed using the multi-color
images of the Hubble Space Telescope-Advanced Camera
for Surveys released by the Great Observatories Origin
Deep Survey (Giavalisco et al. 2004). Different methods
to perform the morphological classification of this sample
are presented in the companion paper Lauger et al. (2006)
and compared with a visual classification of the sam-
ple. A single wavelength rest-frame morphological clas-
sification can be applied over the whole redshift range.
Lauger et al. (2006) show that this classification can sepa-
rate two robust classes : bulge- and disk-dominated galax-
ies. This paper presents the measurement of the LF evo-
lution per morphological type, based on this single wave-
length rest-frame morphological classification.
We use a flat lambda (Ωm = 0.3, ΩΛ = 0.7) cosmol-
ogy with h = H0/100 km s−1Mpc−1. Magnitudes are
given in the AB system.
2. Data set description
We use the high-resolution images provided by the
Great Observatories Origin Deep Survey (GOODS,
Giavalisco et al. 2004) on the Chandra Deep Field South
(CDFS) to perform the study of the galaxy morphology.
The images have been acquired with the Hubble Space
Telescope-Advanced Camera for Surveys (HST/ACS).
The field covers 160 arcmin2and has been observed in four
bands F435W, F606W, F775W, F850LP (noted hereafter
B, V , i, z respectively).
We use the spectroscopic
VIMOS VLT Deep Survey (VVDS) on the CDFS
(Le F` evre et al. 2004). The spectroscopic observations
have been conducted with the multi-object spectrograph
VIMOS on the VLT-ESO Melipal. The spectroscopic
targets are selected on a pure magnitude criterion
17.5 ≤ IAB
≤ 24 from the ESO Imaging Survey
(Arnouts et al. 2001). The sample used in this paper
is limited to the area covered by GOODS. Our sample
contains 670 objects (605 galaxies, 60 stars, 5 QSOs) with
a secure measurement of the redshift (confidence level
greater than 80%) and a mean redshift of 0.76.
Multi-color data from COMBO-17 are available on
the CDFS field (Wolf et al. 2004). These data consist in
12 medium-band filters and 5 broad-band filters from
3500˚ A to 9300˚ A. We also use the near-infrared J and
K band data (21600˚ A) from the ESO Imaging Survey
(Arnouts et al. 2001).
redshifts fromthe
3. Photometric redshifts with Le Phare
We apply the code Le Phare1(S. Arnouts & O. Ilbert)
on the COMBO-17 multi-color data completed by the
NIR data from EIS to compute photometric redshifts
for the complete CDFS sample. The photometric red-
shifts are measured with a standard χ2from the best
fit template on multi-color data (Arnouts et al. 2002).
Our set of templates is composed of four observed spec-
tra from Coleman et al. (1980) and one starburst SED
from GISSEL (Bruzual & Charlot 2003). These 5 tem-
plates have been interpolated to increase the accuracy of
the redshift estimate.
Our photometric redshift code significantly improves
the standard χ2method (Ilbert et al. 2006). We compute
the average difference in each band between the observed
apparent magnitudes and the predicted apparent magni-
tudes derived from the best fit template for a restricted
1www.lam.oamp.fr/arnouts/LE PHARE.html

Page 3

Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.23
Fig.1. Difference between VVDS spectroscopic redshifts
and photometric redshifts (∆z) as a function of the spec-
troscopic redshift.
sample of 67 bright galaxies (IAB < 20) with a spec-
troscopic redshift. These differences never exceed 0.2 and
have an average value over all filters of 0.06 magnitudes.
We correct the predicted apparent magnitudes from these
systematic differences. This method of calibration corrects
for the small uncertainties existing in the filter transmis-
sion curves or in the calibration of the photometric zero-
points.
The comparison between photometric redshifts and
spectroscopic redshifts is shown in Fig.1. We only use the
most secure spectroscopic redshifts with a confidence level
greater than 95% (Le F` evre et al. 2005). The fraction of
catastrophic errors (∆z/(1+z) > 0.1) in photometric red-
shifts estimates is 1.1% and the accuracy of the measure-
ment is σ∆z/(1+z)= 0.046. 295 stars are removed from the
sample which satisfy simultaneously a morphological cri-
terion (SExtractor CLASS STAR parameter greater than
0.975) and a χ2criterion (χ2(gal) − χ2(star) > 0). Up to
z = 1.2 and for IAB≤ 24, the sample contains 3555 pho-
tometric redshifts of galaxies associated with ACS/HST
images.
4. Morphological classification
We
tion
ages
Lauger et al. 2005,
centration of light is defined as the ratio between the
radii which contain 80% and 20% of the total flux of
the galaxy, respectively. The asymmetry is obtained by
measure
(C)
(Abraham et al. 1996,
asymmetry
parameters
(A)
the
and
ACS/HST
Conselice et al. 2000,
concentra-
on im-
Menanteau et al. 2006). Thecon-
computing the difference pixel per pixel of the original
image and of its 180◦rotation. We adopt these parameters
to define our morphological classes since this classification
is automatic, quantitative and reproducible.
Importantly, and thanks to the multi-color coverage
of the ACS/HST images, we can measure the parameters
A and C in the same rest-frame B-band from z = 0 up
to z ∼ 1.2 (Cassata et al. 2005). In this way, we reduce
the effect of a morphological k-correction which gives a
more patchy appearance to the galaxies at higher red-
shift when observations are restricted to one band (e.g.
Kuchinski et al. 2000, Burgarella et al. 2001).
To relate the quantitative parameters A-C to the
Hubble sequence, we calibrate our morphological classes
in the A-C plane with a visual classification of galaxies.
We adopt the empirical criterion A = 0.0917C − 0.2383
to separate the bulge-dominated population from the disk-
dominated population (solid line of Fig.2). This criterion is
chosen to maximize the separation between E/S0 and spi-
ral/irregular classified galaxies. We show in the compan-
ion paper Lauger et al. (2006) that the bulge-dominated
population defined with A ≤ 0.0917C − 0.2383 contains
8.9% of late spiral/irregular galaxies and that the disk-
dominated population includes 8.3% of E/S0 galaxies
down to IAB ≤ 24. The bulge-dominated area contains
also 21.2% of early spiral galaxies but the visual differen-
tiation between faint early spiral galaxies and lenticular
galaxies is strongly subjective.
5. Galaxy luminosity function with ALF
We derive the LF using the Algorithm for Luminosity
Function (ALF) described in Ilbert et al. (2005). ALF in-
cludes 4 standard estimators of the LF which are the
1/Vmax, C+, SWML and STY. Combining these 4 esti-
mators allows us to check the robustness of our estimate
against spatial inhomogeneities, absolute magnitude bin-
ning, or spectral type incompleteness (Ilbert et al. 2004).
We measure k-corrections from a procedure of template
fitting on the multi-color data (Ilbert et al. 2005), using
either the photometric redshift or the spectroscopic red-
shift according to the sample used.
We measure the LF in the rest-frame B Johnson band.
At the average redshift z ∼ 0.76 of this I-selected sample,
this choice of the rest-frame B-band limits the model de-
pendency of the absolute magnitudes (Ilbert et al. 2005),
and minimizes any possible biases due to the mix of spec-
tral types (Ilbert et al. 2004).
6. Results
The values given in this section are obtained using the
photometric redshift sample to increase the accuracy of
the measurements. However, we systematically check that
the results obtained with spectroscopic redshifts and pho-
tometric redshifts are fully consistent.

Page 4

4 Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.2
Fig.2. Distribution in the A-C diagram of the eye-
ball classified galaxies with a spectroscopic redshift. The
solid circles correspond to galaxies visually classified as
elliptical-S0, the solid squares to early spirals, the open
circles to late spirals, the star to irregulars and open tri-
angles to mergers. The solid line is the empirical criterion
A = 0.0917C−0.2383 we have adopted to separate bulge-
and disk-dominated populations. The dotted line corre-
sponds to the criterion A = 0.0917C − 0.2083.
6.1. A blue bulge-dominated population
We observe a blue population of bulge-dominated
galaxies as already observed by e.g. Im et al. (2001),
Menanteau et al. (2004), Cross et al. (2004). For both the
spectroscopic and photometric redshift samples, the (B −
I)0
ABrest-frame colors of the bulge-dominated population
present a bimodal distribution (see upper panel of Fig.3).
We split the bulge-dominated population into two sub-
samples separated in rest-frame color by (B −I)0
according to this bimodality. We measure 32%/68% of
blue/red galaxies at z = 0.4−0.8, respectively. This frac-
tion of blue bulge-dominated galaxies is similar to the
proportion of 30% obtained by Cross et al. (2004) using
a rest-frame color criterion (U −V )AB> 1.7 and a similar
selection of IAB≤ 24.
To investigate farther the structural properties of this
blue population, we use the Petrosian radius r(η = 0.2)
(see Lauger et al. 2005) and the angular distance to mea-
sure the physical size of the galaxy. Fig.3 (lower panel)
shows the galaxy size distribution for blue and red bulge-
dominated galaxies. The blue population, with an average
size of 5.8 h kpc is more compact than the red population
with an average size of 8.2 h kpc. The hypothesis that
these two samples are extracted from the same popula-
tion is rejected at 99.9% by a Kolmogorov-Smirnov test.
AB= 0.9
6.2. Shape of the LF versus morphology
We investigate the dependency of the LF shape on the
morphological type. This analysis is performed in the red-
shift bin 0.4 − 0.8, a good compromise maximizing the
number of galaxies and covering a large absolute magni-
tude range. The LFs of the disk- and bulge-dominated
populations are shown in the middle panels of Fig.4
and the corresponding Schechter parameters are given in
Table 1. The LFs obtained with the photometric and the
spectroscopic redshift samples are in very good agreement
and no systematic trend in the shape is observed when us-
ing photometric redshifts.
The LF of the disk-dominated population presents a
steep slope (α = −1.19 ± 0.07) which contrasts with
the decreasing slope measured for the bulge-dominated
population (α = −0.53 ± 0.13). This population of disk-
dominated galaxies is the dominant population of galax-
ies at z ∼ 0.6. From the integration of the LF up to
MBAB− 5log(h) = −17, the disk-dominated population
represents 74% of the whole galaxy population.
The slope measured for the bulge-dominated popu-
lation (α = −0.53 ± 0.13, see Table 1) is steeper than
previous LF measurement based on spectral type mea-
surements (e.g. α = 0.52 ± 0.20 for Wolf et al. 2003,
α = −0.27 ± 0.10 for Zucca et al. 2006). The presence of
the faint blue bulge-dominated population explains this
difference. The blue bulge-dominated population is com-
posed of faint galaxies representing 67% of the bulge-
dominated galaxies for −19.5 < MBAB−5log(h) but only
7.1% for MBAB−5log(h) < −19.5. As the LF slope is ex-
tremely steep, the proportion of the observed blue galax-
ies is strongly dependent on the considered limit (here
MBAB− 5log(h) ≤ −17). On the contrary, the red bulge-
dominated population is composed of bright galaxies and
represents 92% of the bulge-dominated population for
MBAB−5log(h) ≤ −19.5. The density of red bulge galaxies
decreases toward fainter magnitudes with a strongly de-
creasing LF slope α = 0.55±21. Between −19.5 < MBAB−
5log(h) ≤ −17, the red bulge-dominated population rep-
resents only 5.7% of the whole population. The shape of
the LF is in agreement with the measurement done by
Cross et al. (2004) for red E/S0 galaxies (α = 0.35±0.59,
M∗
BAB− 5log(h) = −19.8 ± 0.5 mag at 0.5 < z < 0.75).
This selected sample of red bulge-dominated galaxies cor-
responds to the classical E/S0 population, composed of
red and bright galaxies with a strongly decreasing LF
slope. Blue and red bulge-dominated populations clearly
exhibit different properties and need to be analysed sepa-
rately.
6.3. Evolution of the LF
Figure 4 presents the evolution of the disk-, red bulge-
and blue bulge-dominated populations from z = 0.05 up
to z = 1.2. The Schechter parameters are given in Table 2.
Since the LF slope is not constrained at z > 0.8, we set α
to the value measured at z = 0.4 − 0.8.

Page 5

Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.25
M∗
AB(B) − 5log(h)
(mag)
φ∗
TypeNb
α
(×10−3h3Mpc−3)
disk
bulge
red-bulge
blue-bulge
892
261
178
83
-1.19+0.07
-0.53+0.13
+0.55+0.21
-2.00
−0.07
-20.22+0.15
-20.20+0.19
-19.53+0.14
-20.95+0.63
−0.15
12.39+2.18
7.48+1.04
−1.11
7.44+0.56
−0.56
0.16+0.16
−0.09
−2.01
−0.13
−0.20
−0.21
−0.15
−0.79
Table 1. Schechter parameters for the rest-frame B-band LF in the redshift bin 0.4 − 0.8 and the corresponding 1σ
error. The parameters without associated errors are fixed. Values are given for the photometric redshift sample.
Fig.3. Top panel: the distribution of the (B −I)0
frame colors for the bulge-dominated population. Bottom
panel: galaxy size distribution for the red bulge-dominated
population (dotted line) and for the blue bulge-dominated
population (solid line). For both panels, the thick lines
correspond to the photometric redshift sample and the
thin lines to the spectroscopic redshift sample.
ABrest-
The results obtained with the photometric redshift
sample are fully in agreement with the results obtained
with the spectroscopic redshift sample (see the open stars
of Fig.4). This comparison gives confidence that our mea-
surement of the LF evolution is not due to systematic
trends in photometric redshift estimates.
The LF of the disk-dominated galaxies evolves only
mildly over the redshift range up to z = 1.2. The slope
of the disk-dominated galaxies in our sample is compara-
ble to the local values obtained by Marinoni et al. (1999)
(α = −1.10 ± 0.07 for S-Im eyeball classified galaxies) or
by Nakamura et al. (2003) (α = −1.12 ± 0.18 for S-Im
type defined with concentration parameter). No signifi-
cant evolution of Φ∗is measured between z = 0.05 and
z = 1.2 when testing for a pure density evolution (setting
the α − M∗parameters, see Table.2). We measure only a
small brightening of 0.4 magnitude when testing for pure
luminosity evolution (setting the values of α−Φ∗). Leaving
M∗and φ∗free (setting only the slope), we measure a
brightening of ∆M∗ ∼ 0.5 magnitude with no significant
evolution of Φ∗between z = 0.05 and z = 1.2.
The density of the red bulge-dominated population de-
creases at high redshift z = 0.8 − 1.2. We measure a de-
crease in density by a factor 2.7 between z ∼ 0.6 and
z ∼ 1 for a pure density evolution (setting the M∗− α
parameters). A pure luminosity evolution of the LF is
not a good fit of the non-parametric data (see Fig.4).
Leaving M∗and φ∗free (setting only the slope), we mea-
sure a small brightening of 0.2 mag with a decrease of
Φ∗by a factor 2 between z ∼ 0.6 and z ∼ 1, consis-
tent with a pure density evolution. Due to the difficulty
to sample the bright part of the red-bulge dominated LF
at z = 0.05 − 0.4, no strong conclusions can be drawn
in this low redshift range. The evolution of the LF for
the red bulge-dominated population is opposite to the in-
crease in density observed by Cross et al. (2004) between
z = 0.5 − 0.75 and z = 0.75 − 1 but is in agreement
with the result obtained by Ferreras et al. (2005) on the
same field. Our results are fully consistent with the results
from the VVDS based on spectroscopic redshifts and spec-
tral type classification (Zucca et al. 2006). The observed
evolution of the red bulge-dominated population in our
analysis remains small in comparison to the strong de-
crease in density of the elliptical galaxies measured by
Wolf et al. (2003) using spectral type classification and
photometric redshifts. Our results instead show that the
population of E/S0 galaxies is already mostly in place at
z ∼ 1.
The LF slope of the blue bulge-dominated population
remains extremely steep in all the redshift bins. This pop-
ulation strongly evolves. To quantify the LF evolution of
this population, we first set the M∗− α parameters to
the z = 0.4 − 0.8 values and look for an evolution in Φ∗.
We measure an increase in density of a factor 2.4 between
z = 0.4 − 0.8 and z = 0.8 − 1.2. Using the same proce-
dure, we set the α − Φ∗parameters to the z = 0.4 − 0.8
values and look for an evolution in M∗. We measure a
brightening of 0.7 magnitude between z = 0.4 − 0.8 and
z = 0.8 − 1.2. The same trend is also present between
z = 0.05−0.4 and z = 0.4−0.8. It is unlikely than a large

Page 6

6 Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.2
fraction of this blue bulge-dominated population contains
misclassified spiral galaxies at high redshifts as the visual
inspection of the UV rest-frame images at z ∼ 1 do not
show star formation in the disc of these galaxies. If a large
fraction of these blue bulge-dominated galaxies is includ-
ing misclassified spiral galaxies at z < 0.7 as claimed by
Ferreras et al. (2005), the observed evolution of the blue
bulge-dominated galaxies should even be stronger than
what we have reported here.
Since A and C are quantitative parameters and since
the visual classification is strongly subjective, the sepa-
ration between bulge- and disk-dominated galaxies in the
A − C plan is not a sharp limit. The area enclosed be-
tween A ≤ 0.0917C − 0.2383 (solid line of Fig.2) and
A ≤ 0.0917C − 0.2083 (dotted line of Fig.2) contains a
mix of different visual types. To quantify the impact of
the adopted criterion on our conclusions, we recompute
the LFs using the criteria A ≤ 0.0917C − 0.2083 rather
than A ≤ 0.0917C − 0.2383. This less conservative crite-
rion increases the contamination of blue late spiral galax-
ies in the bulge-dominated area. As expected, the LF nor-
malization of the blue bulge-dominated galaxies increases
by a factor 1.6-2. We have specifically used a conservative
criterion throughout this paper (A ≤ 0.0917C − 0.2383)
to limit the fraction of late spiral/irregular galaxies in the
bulge-dominated area. We have checked that only 12% of
the blue-bulge dominated galaxies have been visually clas-
sified as late spiral or irregular galaxies. The LFs of the
disk-dominated and of the red bulge-dominated galaxies
remain unchanged and are little sensitive to the adopted
criterion.
7. Discussion and conclusions
We derive the rest-frame B-band LF of galaxies classi-
fied in morphological types up to z = 1.2 in the CDFS
using the VVDS spectroscopic sample of 605 galaxies,
3555 photometric redshifts from COMBO-17 multi-color
data and the HST/ACS images from GOODS. We de-
fine bulge- and disk-dominated populations on the basis
of A-C parameters measured in the rest-frame B-band
(Lauger et al. 2006). We show that the LF of the bulge-
dominated population is the combination of two popu-
lations: a red and bright population making 68% of the
bulge-dominated sample, and a blue population of more
compact galaxies for the remaining 32% of the population.
We observe a strong dependency of the LF shape on the
morphological type. In the redshift range 0.4 < z < 0.8,
we measure a shallow slope α = +0.55 ± 0.21 for the
red bulge-dominated population while the disk-dominated
population shows a very steep slope α = −1.19 ± 0.07.
The blue bulge-dominated population dominates the faint-
end of the bulge-dominated LF. We emphasize that with-
out morphological information, the blue bulge-dominated
population can not be separated from the late spec-
tral type population even if a composite fit of the LF
(de Lapparent et al. 2003) is computed as an alternative
when visual morphologies are not available.
We observe a small evolution of the disk-population.
This is unexpected as the irregular galaxies included in our
disk-dominated class are expected to evolve strongly (e.g.
Brinchmann et al. 1998). As our one-wavelength A-C pa-
rameters are not efficient to isolate the irregular or merger
galaxies from the late spiral galaxies (see Fig.2), this small
evolution of the disk-dominated population could possibly
be explained by cosmic variance in this small field, by the
fact that spiral galaxies dominate the population of irreg-
ular galaxies, and by the fact that we are insensitive to
the morphological k-correction effect.
We measure an increase in density of the red, bright
(brighter than M∗at z > 0.8), bulge-dominated popu-
lation with the age of the Universe. The observed evolu-
tion of the red bulge-dominated LF could be interpreted
as evidence for the build-up of massive elliptical galaxies
from merging and accretion of smaller galaxies in a hier-
archical scenario. Our results indicate that the population
of E/S0 galaxies is already in place at z ∼ 1, in agree-
ment with e.g. Lilly et al. (1995), Conselice et al. (2005),
Zucca et al. (2006). As the field used in this paper is 160
arcmin2and includes large structures (Adami et al. 2005),
cosmic variance could possibly play a role and affect our
results. To investigate this point, we compare the global
LF measured on the CDFS field with the LF in the
VVDS-0226-04 field (Ilbert et al. 2005) which covers an
area 10 times larger. Correcting the LF normalizations in
the CDFS field to match the normalization of the global
LF in the VVDS-0226-04 field, we find that the evolu-
tion, although less pronounced, still shows an increase
of the red bulge-dominated density with the age of the
Universe. The effect of the density-environment relation
(e.g. Dressler et al. 1997) on the observed evolution is dif-
ficult to assess. One approach to address this uncertainty
is to compute the combined LF per morphological type
and environment.
We also observe a very strong evolution of the blue
bulge-dominated population corresponding to a brighten-
ing of 0.7 magnitude (or an increase in density by a fac-
tor 2.4) between z ∼ 0.6 and z ∼ 1. The nature of this
population remains unclear. Mergers expected in the hi-
erarchical scenario could create a burst of star formation
explaining the blue color of these galaxies. We have ob-
served some signs of disruption for a significant fraction
of these galaxies. These galaxies could be also a dwarf
population undergoing a strong burst of star formation
in the galaxy core, which could be interpreted as a bulge
component (Im et al. 2001). The smaller size of this blue
population in comparison to the red one as well as the
faint absolute magnitude distribution of this population
seem to favor this hypothesis. Considering that the evo-
lution of the blue bulge-dominated population produces a
fading in luminosity, we can relate this evolution to the
strong decrease of the star formation rate observed from
z ∼ 1. This blue population could be the progenitor of
the local population of dwarf spheroidal galaxies under-
going strong star formation at z ∼ 1. Another possibility
to investigate is the presence of an AGN in the galaxy

Page 7

Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.27
M∗
AB(B) − 5log(h)
(mag)
φ∗
Typez Nb
α
(×10−3h3Mpc−3)
disk 0.05-0.40
0.05-0.40
0.05-0.40
0.40-0.80
0.80-1.20
0.80-1.20
0.80-1.20
0.05-0.40
0.05-0.40
0.40-0.80
0.80-1.20
0.80-1.20
0.80-1.20
0.05-0.40
0.05-0.40
0.40-0.80
0.80-1.20
0.80-1.20
482
482
482
892
755
755
755
37
37
178
109
109
109
40
40
83
109
109
-1.19
-1.19
-1.19
-1.19+0.07
-1.19
-1.19
-1.19
+0.55
+0.55
+0.55+0.21
+0.55
+0.55
+0.55
-2.00
-2.00
-2.00
-2.00
-2.00
-19.79+0.12
-20.22
-19.54+0.14
-20.22+0.15
-20.21+0.04
-20.22
-20.06+0.07
-19.77+0.40
-19.53
-19.53+0.14
-18.23+0.09
-19.53
-19.76+0.09
-20.95
-19.11+0.22
-20.95+0.63
-20.95
-21.63+0.11
−0.11
12.39
10.95+0.58
15.85+0.91
12.39+2.18
12.39
12.44+0.51
−0.51
15.65+1.06
−0.98
7.44
5.79+1.23
−1.23
7.44+0.56
−0.56
7.44
2.78+0.30
−0.30
3.05+0.29
−0.29
0.03+0.01
−0.01
0.16
0.16+0.16
−0.09
0.38+0.04
−0.04
0.16
−0.58
−0.15
−0.87
−0.07
−0.15
−2.01
−0.04
+0.07
red-bulge
−0.39
−0.21
−0.15
−0.09
−0.09
blue-bulge
−0.20
−0.79
−0.10
Table 2. Evolution of the Schechter parameters for the rest-frame B-band LF and associated 1σ errors. Values are
given for the photometric redshift sample. The parameters without associated errors are set using the values measured
at z = 0.4 − 0.8. At z = 0.05 − 0.4 and z = 0.8 − 1.2, we always provide the parameters for a pure luminosity and
density evolution. When possible, we also provide the parameters for M∗−Φ∗let free to vary. When α is set, the STY
errors on M∗and Φ∗are underestimated.
nucleus as shown by Menanteau et al. (2005) which could
be strongly related to the star formation activity.
An increase in survey areas is clearly necessary to
limit the cosmic variance effects on computing the LFs
of morphologically selected populations, as is on-going
with next generation surveys (e.g. COSMOS). The de-
velopment of quantitative methods to better isolate mor-
phological galaxy classes, in particular merger and irreg-
ular galaxies, and the combination of morphological and
spectral classifications, will also be necessary for further
progress.
Acknowledgements. This research has been developed within
the framework of the VVDS consortium.
This work has been partially supported by the CNRS-INSU
and its Programme National de Cosmologie (France), and by
Italian Ministry (MIUR) grants COFIN2000 (MM02037133)
and COFIN2003 (num.2003020150).
The VLT-VIMOS observations have been carried out on guar-
anteed time (GTO) allocated by the European Southern
Observatory (ESO) to the VIRMOS consortium, under a
contractual agreement between the Centre National de la
Recherche Scientifique of France, heading a consortium of
French and Italian institutes, and ESO, to design, manufac-
ture and test the VIMOS instrument.
References
Abraham R.G., van den Bergh S., Glazebrook K. et al., 1996,
ApJS, 107, 1
Adami C., Mazure A., Ilbert O. et al., 2005, A&A, 443, 805
Arnouts S., Vandame B., Benoist C. et al., 2001, A&A, 379,
740
Arnouts S., Moscardini L., Vanzella E. et al., 2002, MNRAS,
329, 355
Bell E.F., Papovich C., Wolf C. et al., 2005, ApJ, 625, 23
Brinchmann J., Abraham R., Schade D. et al., 1998, ApJ, 499,
112
Bruzual G. & Charlot S., 2003, MNRAS, 344, 1000
Burgarella D., Buat V., Donas J., Milliard B. & Chapelon S.,
2001, A&A, 369, 421
Cassata P., Cimatti A., Franceschini A. et al., 2005, MNRAS,
357, 903
Coleman G.D., Wu C.-C., Weedman D.W., 1980, ApJS, 43,
393
Conselice C.J., Bershady M.A. & Jangren A., 2000, ApJ, 529,
886
Conselice C. J., Blackburne J. A. & Papovich C., 2005, ApJ,
620, 564
Cross N.J.G., Bouwens R. J., Bentez N. et al., 2004, AJ, 128,
1990
de Lapparent V., Galaz G., Bardelli S. & Arnouts S., 2003,
A&A, 404, 831
Dressler A., Oemler A., Couch W.J. et al., 1997, ApJ, 490, 577
Ferreras I, Lisker T., Carollo C.M., Lilly S.J., Mobasher B.,
2005, ApJ, 635, 243
Giavalisco M., Ferguson H.C., Koekemoer A. et al., 2004, ApJ,
600, L93
Ilbert O., Tresse L., Arnouts S. et al., 2004, MNRAS, 351, 541
Ilbert O., Tresse L., Zucca E. et al., 2005, A&A, 439, 863
Ilbert O., Arnouts, S., McCracken, H.J., et al., 2006, A&A,
submitted, astro-ph/0603217
Im M., Simard L., Faber S. M. et al., 2002, ApJ, 571, 136
Im M., Faber S. M., Gebhardt K. et al., 2001, AJ, 122, 750

Page 8

8Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.2
Kuchinski L.E., Madore B.F., Freedman W.L., Trewhella M.,
2000, ApJS, 131, 441
Lauger S., Burgarella D. & Buat V., 2005, A&A, 434, 77
Lauger S., Ilbert O., Burgarella D. et al., 2006, A&A, submit-
ted
Le F` evre O., Vettolani G., Paltani S. et al., 2004, A&A, 428,
1043
Le F` evre O., Vettolani G., Garilli, B., et al., 2005, A&A, 439,
845
Lilly S. J., Tresse L., Hammer F., Crampton D., Le F` evre O.,
1995, ApJ, 455, 108
Madgwick D.S., Lahav O., Baldry I.K. et al., 2002, MNRAS,
333, 133
Marinoni C., Monaco P., Giuricin G. & Costantini B., 1999,
ApJ, 521, 50
Menanteau F., Ford H. C., Illingworth G.D. et al., 2004, ApJ,
612, 202
Menanteau F., Martel A.R., Tozzi P. et al., 2005, ApJ, 620,
697
Menanteau F., Ford H.C., Motta V. et al., 2006, AJ, 131, 208
Nakamura O., Fukugita M., Yasuda N. et al., 2003, ApJ, 125,
1682
Wolf C., Meisenheimer K., Rix H.-W. et al., 2003, A&A, 401,
73
Wolf C., Meisenheimer K., Kleinheinrich M. et al., 2004, A&A,
421, 913
Wolf C., Bell E.F., McIntosh D.H. et al., 2005, AJ, 630, 771
Zucca E., Ilbert O., Tresse L. et al., 2006, in press,
astro-ph/0506393

Page 9

Ilbert O. et al.: Galaxy luminosity function per morphological type up to z = 1.29
Fig.4. Evolution of the Luminosity Function for galaxies classified in morphological types from z = 0.05 up to z = 1.2.
Open stars correspond to the LFs measured with the spectroscopic redshift sample using the 1/Vmaxestimator. All
the other results are obtained using the photometric redshift sample. The left panels correspond to the LF of the
disk-dominated population, the middle panels to the red bulge-dominated population and the right panels to the blue
bulge-dominated population. The redshift bin and the corresponding number of galaxies is indicated in each panel.
We adopt the following symbols for the various estimators: circles for the 1/Vmax, triangles for the SWML, squares
for the C+and solid lines for the STY (only at z = 0.4−0.8 where the LF shape is constrained). The STY fit derived
in the redshift bin 0.4 − 0.8 is reported in each panel with dotted lines. The long dashed lines and short dashed lines
correspond respectively to the fit obtained setting α − M∗(pure density evolution) and α − Φ∗(pure luminosity
evolution) at the values obtained at z = 0.4 − 0.8.