Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~sperello/ApJ647_L99Rev.pdf Äàòà èçìåíåíèÿ: Wed Sep 12 20:00:34 2007 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 03:15:20 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: millennium

Revision of ApJ 647, L99 [2006] for Erratum
A Preprint typ eset using L TEX style emulateap j v. 08/13/06

THE INFLUENCE OF MASS AND ENVIRONMENT ON THE EVOLUTION OF EARLY-TYPE GALAXIES
Sperello di Serego Alighieri
INAF ­ Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50122 Firenze, Italy

Barbara Lanzoni
INAF ­ Osservatorio Astronomico di Bologna, Via Ranzani 1, 40127 Bologna, Italy

and Inger JÜrgensen
Gemini Observatory, 670 North A'ohoku Place, Hilo, HI 96720 Revision of ApJ 647, L99 [2006] for Erratum

ABSTRACT We report on a uniform comparative analysis of the fundamental parameters of early­type galaxies at z 1 down to a well defined magnitude limit (MB -20.0 in the field and MB -20.5 in the clusters). The changes in the M/LB ratio from z 1 to today are larger for lower mass galaxies in all environments, and are similar in the field and in the clusters for galaxies with the same mass. By deriving ages from the M/LB ratio, we estimate the formation redshift for early-type galaxies as a function of galaxy mass and environment. We find that the age of early-type galaxies increases with galaxy mass (downsizing) in all environments, and that cluster galaxies appear to have the same age within 5% as field galaxies at any given galaxy mass. The first result confirms similar ones obtained by other means, while the second one is controversial. The most recent incarnation of the hierarchical models of galaxy formation and evolution is capable of explaining the first result, but predicts that cluster galaxies should be older than field galaxies. We also find a total lack of massive early­type galaxies (M > 3 â 1011M ) with a formation redshift smaller than 2, which cannot be due to selection effects. Subject headings: cosmology: observations -- galaxies: elliptical and lenticular, cD -- galaxies: evolution -- galaxies: formation -- galaxies: high redshift
1. INTRODUCTION

Early­type galaxies (ETG) contain most of the visible mass in the Universe (Renzini 2006) and are thought to reside in the highest density peaks of the underlying dark matter distribution. Therefore, understanding their evolution is crucial for understanding the evolution of galaxies and structures in general. In the 3-dimensional space of their main parameters (the effective radius Re , the central velocity dispersion , and the average surface 2 luminosity within Re , I e = L/2 Re ), ETG concentrate on a plane thus called the Fundamental Plane (FP, Djorgowski & Davis 1987; Dressler et al. 1987). This implies that, besides being in virial equilibrium, ETG show a striking regularity in their structures and stellar populations (e.g., Renzini & Ciotti 1993), which allows, at least at a first order, to use their main observables for deriving the galaxy mass and M/L ratio. For instance, assuming R1/4 homology, the mass is given by (Michard 1980; see also Cappellari et al. 2005): M = 5Re 2 /G. (1)

passive luminosity evolution (see Renzini 2006 and references therein). If ascribed to differences in the stellar populations, the observed changes in the M/L ratio can be used to infer the ages of ETG. We report on a comparative analysis of the best data on the fundamental parameters of ETG at z 1, the highest redshift for which these data are currently available, obtained from recent spectroscopic observations with 8-10m class telescopes, complemented with deep imaging with the Hubble Space Telescope. Using the Universe as a time machine and profiting from the large leverage provided by the redshift, we infer ages for ETG and analyse them as a function of galaxy mass and environment. We assume a flat Universe with m = 0.3, = 0.7, and H0 = 70 km s-1 Mpc-1 , and we use magnitudes based on the Vega system.
2. BACKGROUND

Moreover, the slope of the FP can be interpreted as a systematic variation of the M/L ratio along the plane by a factor of 3 (e.g., Ciotti, Lanzoni & Renzini 1996). At high redshift the FP is known to stay thin, and its intercept shows an offset with respect to the local one that corresponds to a change in M/L consistent with
Electronic address: sp erello@arcetri.astro.it

Recent studies (di Serego Alighieri et al. 2005; Treu et al. 2005; van der Wel et al. 2005) of the FP of ETG in the field at z 1, in the rest-frame B-band, down to relatively faint luminosities (MB -20.0), and hence small masses, demonstrate that, in addition to the offset, the FP at z 1 also shows a different slope. This implies that the galaxy M/LB ratio evolves with redshift in a way that depends on the galaxy mass. By comparing the M/LB ratio of field ETG to that of massive (M > 1011 M ) ETG in clusters, a faster evolution of M/LB for the less massive galaxies has been derived, and it is interpreted as a manifestation of downsizing, i.e. the tendency of smaller galaxies to have later or more


2

di Serego Alighieri, Lanzoni & JÜrgensen M/LB and age obtained by evolutionary population synthesis models (Maraston 2005, see also http://wwwastro.physics.ox.ac.uk/maraston/SSPn/ml/), and assuming that the model stellar mass is proportional to the dynamical mass obtained from equation (1). Since the M/L­age relation depends both on the stellar initial mass function (IMF) and on the metallicity, we adopt a Kroupa (2001) IMF, which is known to better reproduce the characteristics of low and high redshift galaxies. Moreover we estimate the galaxy metallicity from the observed velocity dispersion following Thomas et al. (2005; see also Annibali et al. 2006), and we assume that this relationship does do not evolve with redshift, as in the case of passive evolution. Given the values of M/LB and metallicity for every galaxy in our sample, ages have been inferred by means of a spline interpolation of the population synthesis model results. Then, the lookback time to formation has been derived by using the Universe as a time machine and exploiting the large leverage provided by the considerable distance of the observed ETG. The uncertainties on the age estimates have been computed taking into account the known errors in M/LB , as well as the uncertainties on the estimated metallicities, due to the known errors in the velocity dispersion measurements and to the observed scatter in the metallicity vs. velocity dispersion relation (Thomas et al. 2005). The resulting formation epochs of ETG are shown in Figure 3 as a function of galaxy mass, both for the cluster and for the field environment. The estimated ages for the two brightest cluster galaxies (#1567 in RXJ0152.7-1357, and #563 in RXJ1226.9+3332 JÜrgensen et al. (2006)), are 23.4 ± 2.6 and 16.4 ± 0.9 Gyr, respectively, and are not included in Fig. 3 (see below for a discussion about these large ages). A clear and important result is the lack of young massive ETG. In particular all ETG with M > 3 â 1011 M have a lookback time to formation larger than 10 Gyr and have a formation redshift larger than 2. Clearly this cannot be the result of a selection effect, since relatively young massive ETG could not escape from the available surveys. Confirming the analysis of the evolution of M/LB given at the beginning of this section, we find that more massive galaxies are older than lower mass ones in all environments, and that cluster galaxies have the same age within 5% as field galaxies with the same mass, in the whole mass range (see Fig. 4). A similar dependence of the age on the mass has already been obtained by an analysis of the absorption line indices of a sample of local ETG (Thomas et al. 2005). However Thomas et al. (2005) find that ETG in clusters are older than those in the field by about 2 Gyr. Given the number of ob jects in the samples that we have examined and the errors in the estimate of their age, we should have seen such a systematic age difference, if it were present in the data that we have used. We argue that using the Universe as a time machine should be more powerful than "archaeology" on local galaxies, since galaxies are caught closer to the action. Interestingly, also the Coma ETG show the downsizing effect, and their formation redshifts are very consistent with those of z 1 ETG (JÜrgensen 1999). This suggests that the z 1 samples examined here are not much affected by the progenitor bias (van Dokkum & Franx 2001; di Serego Alighieri et al. 2005). Thus, our results suggest that the first ETG to form are the

prolonged star formation histories than the massive ones (Cowie et al. 1996). Very recently the high-z FP of the ETG has been studied in two clusters (RX J0152.7-1357 at z=0.835 and RX J1226.9-3332 at z=0.892), reaching a similarly faint limiting absolute magnitude (MB -20.5), also in the rest-frame B-band (JÜrgensen et al. 2006). This has pointed out that, also in the clusters, the slope of the FP changes with redshift, a manifestation of downsizing even in high density environments. Unfortunately, because of an error in the calibration of the galaxy luminosities used by JÜrgensen et al. (2006), the photometry for the two clusters should be offset to brighter luminosities by a factor (1 + z ), with respect to the values published in that paper. Correcting for this error corresponds to an offset in log L to brighter luminosities with log (1+z ), which is 0.26 and 0.28 for RX J0152.7­1357 (z = 0.835) and RX J1226.9+3332 (z = 0.892) respectively. Therefore the cluster data, which we have used in the published version of this letter (ApJ 647, L99), should be changed and we present here a corrected version of our original letter (see also the Erratum to ApJ 647, L99, published in ApJ 652, L145).
3. THE FORMATION EPOCH OF CLUSTER AND FIELD
EARLY­TYPE GALAXIES

We make a uniform comparison of these results on the high redshift FP (Fig. 1) and on the consequent variations of the M/LB ratio (Fig. 2), both in the field, by using the samples of di Serego Alighieri et al. (2005) and of Treu et al. (2005), and in the clusters, by using the sample of JÜrgensen et al. (2006). As a reference in the local Universe, we use new data for the Coma cluster (JÜrgensen 1999; JÜrgensen et al. 2006). The figures show that the change in M/LB between ETG at high redshift and the local ones decreases with the galaxy mass and is very similar in the clusters and in the field. However, since the clusters are at a slightly lower redshift, a deeper analysis is necessary to show this more clearly. The usual way to achieve this purpose is to compare the M/LB ratio of the high redshift ETG with the corresponding ratio obtained for massive (M 1011 M ) cluster ETG at the same redshift, as compiled and parameterized by van Dokkum & Stanford (2003). However this analysis is unsatisfactory, since the massive cluster ETG are not necessarily a uniform class, and, by construction, such a procedure prevents one from studying the lower mass cluster galaxies. What is of interest is how the star formation history of ETG, or at least their average stellar age, depends on both galaxy mass and environment. We analyse this by interpreting the changes in M/LB as differences in the ages of the stellar populations1 . While the star formation histories of some ETG could have had multiple episodes of star formation (Treu et al. 2005), we can only estimate luminosity weighted average stellar ages, by using single stellar population models. We therefore infer galaxy ages using the relation between
1 It has b een shown that other p ossible interpretations, i.e. systematic structural changes and partial supp ort by rotation, can only explain a small fraction of the observed differential evolution of M/LB , and that this evolution correlates with the rest-frame U - B colour, thereby providing indep endent evidence for changes in the stellar p opulations (di Serego Alighieri et al. 2005).


Mass Drives Spheroidal Galaxies Evolution

3

Fig. 2.-- The M/L ratio in the B-band as a function of galaxy mass for the ETG samples shown in figure 1 (same symb ols). The dotted line is a fit to the Coma ETG, while the upp er and lower dashed lines represent the MB = -20.0 and MB = -20.5 magnitude limits of di Serego Alighieri et al. (2005) and of JÜrgensen et al. (2006) resp ectively. The changes in M/LB from high redshift to z = 0 decrease with galaxy mass in all environments and are similar in the field and in the clusters.

Fig. 1.-- The Fundamental Plane seen edge­on for lo cal ETG in the Coma Cluster (JÜrgensen et al. 2006) (black crosses), for field ETG at z 1 from the K20 survey (di Serego Alighieri et al. 2005) b oth in the CDFS field (filled black circles) and in the Q0055 field (filled black triangles), for field ETG at z 1 in the GOODS area (Treu et al. 2005) (filled blue squares), and for the ETG in two clusters (JÜrgensen et al. 2006) at z=0.835 (op en red squares) and at z=0.892 (op en red triangles). The dashed line is the b est fit plane to the Coma cluster galaxies. Compared to the lo cal one, the FP at high redshift is offset and rotated in all environments.

most massive ones independently of the environment. Although the absolute ages that we derive are somewhat model dependent, are affected by an approximate metallicity estimate, and obviously depend also on the adopted cosmological parameters, we stress that the trends of age differences between high redshift and local ETG, and between galaxies with different masses and in different environments are much more robust. One of the uncertainties affecting the age estimates derives from the assumption of structural homology when computing masses through equation (1). It is well known that ETG show a systematic departure from homology, both locally (Caon, Capaccioli & D'Onofrio 1993; Guti` errez et al. 2004; Gavazzi et al. 2005) and at z 1 (di Serego Alighieri et al. 2005) and that more precise dynamical masses can be obtained taking these deviations into account, using the S´ ersic (1968) profile, to describe the observed surface brightness distribution, instead of the R1/4 law (Bertin, Ciotti & Del Principe 2002). These mass estimates can be up to 50% higher than those obtained assuming homology for the low mass galaxies, but can also be lower by up to 20% for

Fig. 3.-- The formation ep o ch for the ETG shown in figure 1 (same symb ols), evaluated as explained in the text. The two upward p ointing arrows indicate that the two most massive cluster ETG are out of the figure (their ages amount to 16.4 and 23.4 Gyr). The continuous line shows the median mo del ages obtained by De Lucia et al. (2006) from a semianalytic mo del of hierarchical galaxy evolution, while the dashed lines are their upp er and lower quartiles. More massive ETG form earlier in all environments, and the ages are not influenced by the environment.


4

di Serego Alighieri, Lanzoni & JÜrgensen Very recently the largest high resolution simulation of the growth of cosmic structure in the hierarchical formation scenario (the Millennium Run, Springel et al. 2005) has been used to study how the ages of ETG depend on environment and on galaxy mass (De Lucia et al. 2006). In this model, since merging of smaller galaxies is an important ingredient for the formation of ETG, the galaxy formation time, which is when most of its stars formed, and the galaxy assembly time, which is when stars assembled in the single galaxy that we observe, are considered separately (De Lucia et al. 2006). Our dating based on changes in the M/LB ratio relates to when the stars formed, rather than to when they assembled. The semianalytic hierarchical model of De Lucia et al. (2006) is able to reproduce the already known result, i.e. that the formation times are earlier for more massive ETG, although the downsizing effect is considerably steeper in the model than in the data (see Fig. 3), but clearly predicts that cluster galaxies should be older than field galaxies, which is not what we observe.

Fig. 4.-- Histogram of the average lo okback time to formation p er mass bin for the high redshift ETG in the field and in the clusters (hatched). The error bars (dotted for the clusters) show the standard deviation due to the galaxy­to­galaxy variations in each mass bin.

the high mass galaxies (di Serego Alighieri et al. 2005). Unfortunately we do not have S´ ersic indices for all the ETG examined here, but we have checked on the K20 field samples of di Serego Alighieri et al. (2005) that the ages estimated by taking non homology into account do not vary substantially from those given in Fig. 3 and 4, computed using eq. (1). Since the brightest cluster galaxies are known to deviate from the R 1/4 profile, and if the influence of dark matter increases in high mass galaxies, these factors could lead to an overestimate of the ages of the most massive ETG in the cluster sample. The influence of selection effects is shown by the dashed lines in Fig. 2, which represent the magnitude limit of the K20 field samples of di Serego Alighieri et al. (2005) and of the two high redshift clusters of JÜrgensen et al. (2006). These samples are affected by selection only for M < 4 â 1010 M , while the different slope in the high redshift samples compared to the local one is clearly visible also for larger masses, thus cannot be totally due to selection effects (see also van der Wel et al. (2005)).

We thank Sandro Bressan, Claudia Maraston and Alvio Renzini for useful advice and suggestions. The data on high redshift field ETG were obtained at the European Southern Observatory, Chile (ESO Programme 70.A-0548) and at the W.M. Keck Observatory on Mauna Kea, Hawaii. The spectroscopic data for the high redshift clusters were obtained at the Gemini Observatory (GN-2002B-Q-29, GN-2004A-Q-45), which is operated by AURA, Inc., under a cooperative agreement with NSF on behalf of the Gemini partnership: NSF (US), PPARC (UK), NRC (Canada), CONICYT (Chile), ARC (Australia), CNPq (Brazil) and CONICET (Argentina). Based on observations made with the NASA/ESA Hubble Space Telescope.

REFERENCES Annibali, F. 2006, A&A, in press (also astro-ph/0609175) Bertin, G., Ciotti, L. & Del Princip e, M. 2002, A&A, 386, 149 Caon, N., Capaccioli, M. & D'Onofrio, M. 1993, MNRAS, 265, 1013 Capp ellari, M. et al. 2005, MNRAS, 366, 1126 Ciotti, L., Lanzoni, B., Renzini, A. 1996, MNRAS, 282, 1 Cowie, L.L., Songaila, A., Hu, E.M., & Cohen, J.G. 1996, AJ, 112, 839 De Lucia, G., Springel, V., White, S.D.M., Croton, D. & Kauffmann, G. 2006, MNRAS, 366, 499 di Serego Alighieri, S. et al. 2005, A&A, 442, 125 Djorgovski, S. & Davis, M. 1987, ApJ, 313, 59 Dressler, A. et al. 1987, ApJ, 313, 42 Gavazzi, G. et al. 2005, A&A, 430, 411 Gebhardt, K, et al. 2003, ApJ, 596, 239 Guti` errez, C.M. et al. 2004, ApJ, 602, 664 JÜrgensen, I. 1999, MNRAS, 306, 607 JÜrgensen, I. et al. 2006, ApJ, 639, L9, and Erratum Kroupa, P. 2001, MNRAS, 322, 231 Maraston, C. 2005, MNRAS, 362, 799 Michard, R. 1980, A&A, 91, 122 Renzini, A. 2006, ARA&A, 44, 141 Renzini, A. & Ciotti, L. 1993, ApJ, 416, L49 Springel, V. et al. 2005, Nature, 435, 629 Thomas, D. et al. 2005, MNRAS, 621, 673 Treu, T. et al. 2005, ApJ, 633, 174 van der Wel, A. et al. 2005, ApJ, 631, 145 van Dokkum, P. & Franx, M. 2001, ApJ, 553, 90 van Dokkum, P. & Stanford, S.A. 2003, ApJ, 585, 78