Evolution of the Early-Type Galaxy Fraction in Clusters since z = 0.8
Luc Simard, Douglas Clowe, Vandana Desai, Julianne J. Dalcanton, Anja von der Linden, Bianca M. Poggianti, Simon D. M. White, Alfonso Aragon-Salamanca, Gabriella De Lucia, Claire Halliday, Pascale Jablonka, Bo Milvang-Jensen, Roberto P. Saglia, Roser Pello, Gregory H. Rudnick, Dennis Zaritsky
aa r X i v : . [ a s t r o - ph . C O ] O c t Astronomy&Astrophysicsmanuscript no. ms c (cid:13)
ESO 2018May 30, 2018
Evolution of the Early-Type Galaxy Fraction in Clusters since z =0.8 ⋆ Luc Simard , Douglas Clowe , Vandana Desai , Julianne J. Dalcanton , Anja von der Linden , , Bianca M. Poggianti ,Simon D. M. White , Alfonso Arag´on-Salamanca , Gabriella De Lucia , , Claire Halliday , Pascale Jablonka , BoMilvang-Jensen , , Roberto P. Saglia , Roser Pell´o , Gregory H. Rudnick , , Dennis Zaritsky National Research Council of Canada, Herzberg Institute of Astrophysics, 5071 West Saanich Road, Victoria, British Columbia,Canada Ohio University, Department of Physics and Astronomy, Clippinger Lab 251B, Athens, OH 45701, USA California Institute of Technology, MS 320-47, Pasadena, CA 91125, USA University of Washington, Department of Astronomy, Box 351580, Seattle, WA 98195-1580, USA Kavli Institute for Particle Astrophysics and Cosmology, PO Box 20450, MS 29, Stanford, CA 94309, USA Max-Planck-Institut f¨ur Astrophysik, Karl-Schwarschild-Str., 1, Postfach 1317, D-85741 Garching INAF - Astronomical Observatory of Padova, Italy School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD, UK INAF - Astronomical Observatory of Trieste, via Tiepolo 11, I-34143 Trieste, Italy INAF - Osservatorio Astronomico di Arcetri, Largo Enrico Fermi 5, 50125 Firenze, Italy Observatoire de l’Universit´e de Gen`eve, Laboratoire d’Astrophysique de l’Ecole Polytechnique F´ed´erale de Lausanne (EPFL),1290 Sauverny, Switzerland Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen, Denmark The Royal Library / Copenhagen University Library, Research Department, Box 2149, DK-1016 Copenhagen K, Denmark Max-Planck Institut f¨ur extraterrestrische Physik, Giessenbachstrasse D-85748 Garching, Germany Laboratoire d’Astrophysique de Toulouse-Tarbes, CNRS, Universit´e de Toulouse, 14 Avenue Edouard Belin, 31400-Toulouse,France The University of Kansas, Department of Physics and Astronomy, Malott room 1082, 1251 Wescoe Hall Drive, Lawrence, KS,66045, USA NOAO, 950 North Cherry Avenue, Tucson, AZ 85719, USA Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ, 85721Accepted for publication in A&A
ABSTRACT
We study the morphological content of a large sample of high-redshift clusters to determine its dependence on cluster mass andredshift. Quantitative morphologies are based on PSF-convolved, 2D bulge + disk decompositions of cluster and field galaxies on deepVery Large Telescope FORS2 images of eighteen, optically-selected galaxy clusters at 0 . < z < .
80 observed as part of theESO Distant Cluster Survey (“EDisCS”). Morphological content is characterized by the early-type galaxy fraction f et , and early-typegalaxies are objectively selected based on their bulge fraction and image smoothness. This quantitative selection is equivalent toselecting galaxies visually classified as E or S0. Changes in early-type fractions as a function of cluster velocity dispersion, redshiftand star-formation activity are studied. A set of 158 clusters extracted from the Sloan Digital Sky Survey is analyzed exactly as thedistant EDisCS sample to provide a robust local comparison. We also compare our results to a set of clusters from the MillenniumSimulation. Our main results are: (1) The early-type fractions of the SDSS and EDisCS clusters exhibit no clear trend as a function ofcluster velocity dispersion. (2) Mid- z EDisCS clusters around σ =
500 km / s have f et ≃ z EDisCS clusters have f et ≃ ∼
25% increase over a time interval of 2 Gyrs. (3) There is a marked di ff erence in the morphological content ofEDisCS and SDSS clusters. None of the EDisCS clusters have early-type galaxy fractions greater than 0.6 whereas half of the SDSSclusters lie above this value. This di ff erence is seen in clusters of all velocity dispersions. (4) There is a strong and clear correlationbetween morphology and star formation activity in SDSS and EDisCS clusters in the sense that decreasing fractions of [OII] emittersare tracked by increasing early-type fractions. This correlation holds independent of cluster velocity dispersion and redshift eventhough the fraction of [OII] emitters decreases from z ∼ . z ∼ .
06 in all environments. Our results pose an interesting challengeto structural transformation and star formation quenching processes that strongly depend on the global cluster environment (e.g., adense ICM) and suggest that cluster membership may be of lesser importance than other variables in determining galaxy properties.
Key words.
Galaxies : fundamental parameters, Galaxies : evolution, Galaxies: clusters: general ⋆ Based on observations obtained in visitor and service modes at theESO Very Large Telescope (VLT) as part of the Large Programme166.A-0162 (the ESO Distant Cluster Survey). Also based on ob-servations made with the NASA / ESA Hubble Space Telescope, ob-tained at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., un-der NASA contract NAS 5-26555. These observations are associatedwith proposal 9476. Support for this proposal was provided by NASAthrough a grant from the Space Telescope Science Institute. Simard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters
1. Introduction
Our current paradigm for the origin of galaxy morphologies restsupon hierarchical mass assembly (e.g., Steinmetz & Navarro,2002), and many transformational processes are at workthroughout the evolutionary histories of galaxies. Some de-termine the main structural traits (e.g., disk versus spheroid)while others only influence properties such as color and star-formation rates. Disk galaxy collisions lead to the formation ofelliptical galaxies (Spitzer & Baade, 1951; Toomre & Toomre,1972; Farouki & Shapiro, 1982; Negroponte & White, 1983;Barnes & Hernquist, 1992, 1996; Mihos & Hernquist, 1996),and the extreme example of this process is the build-upof the most massive galaxies in the Universe at the coresof galaxy clusters through the accretion of cluster mem-bers. Disks can also be transformed into spheroidals by tidalshocks as they are harassed by the cluster gravitational po-tential (Farouki & Shapiro, 1981; Moore et al., 1996, 1998).Harassment inflicts more damage to low luminosity galaxies be-cause of their slowly rising rotation curves and their low den-sity cores. Galaxies can be stripped of their internal gas andexternal supply through ram pressure exerted by the intraclus-ter medium (Gunn & Gott, 1972; Larson, Tinsley, & Caldwell,1980; Quilis, Moore, & Bower, 2000), and the result is a“quenching” (or “strangulation”) of their star formation thatleads to a rapid reddening of their colours (also see Martig et al.,2009). The task of isolating observationally the e ff ects of a givenprocess has remained a major challenge to this day.Many processes a ff ecting galaxy morphologies are clearlyenvironmentally-driven, and galaxy clusters are therefore ideallaboratories in which to study all of them. The dynamical state ofa cluster, which can be observationally characterized by measur-ing mass and substructures, should be related to its morpholog-ical content. For example, the number of interactions / collisionssu ff ered by a given galaxy should depend on local number den-sity and the time it has spent within the cluster. Dynamicallyyoung clusters with a high degree of subclustering should con-tain large numbers of galaxies that are infalling for the first time.More massive clusters will contain more galaxies, but they willalso have higher galaxy-galaxy relative velocities that may im-pede merging (Lubin et al., 2002). Spheroidal / elliptical galaxieswill preferentially be formed in environments where the bal-ance between number density and velocity dispersions is op-timal, but it is still not clear where this optimal balance lies.Cluster masses can be estimated from their galaxy internal veloc-ity dispersion (Rood et al., 1972; Dressler, 1984; Carlberg et al.,1997; Tran et al., 1999; Borgani et al., 1999; Lubin et al.,2002), through weak-lensing shear (Kaiser & Squires, 1993;Schneider & Seitz, 1995; Hoekstra, Franx, & Kuijken, 2000;Clowe et al., 2006) or through analysis of their hot X-ray emit-ting atmospheres (e.g., Allen, 1998), and it will be used hereas the main independent variable against which morphologicalcontent will be studied.The morphological content of high-redshift clusters is mostoften characterized by the fraction f E + S of early-type galax-ies they contain (Dressler et al., 1997; van Dokkum et al., 2000;Fasano et al., 2000; van Dokkum et al., 2001; Lubin et al., 2002;Holden et al., 2004; Smith et al., 2005; Postman et al., 2005;Desai et al., 2007; Poggianti et al., 2009b). The bulk of the dataavailable so far is based on visual classification. “Early-type”galaxies are defined in terms of visual classifications as galax-ies with E or S0 Hubble types. A compilation of early-typefractions taken from the literature (van Dokkum et al., 2000)shows a dramatic increase of the early-type fractions as a function of decreasing redshift from values around 0.4 − z ∼ f E + S is in fact a function of both lookback time(redshift) and projected galaxy density. They find f E + S staysconstant at 0.4 over the range 1 < t lookback < Σ <
10 Mpc − . For high density environments( Σ = − ), f E + S decreases from 0.9 to 0.7. At fixedlookback time, f E + S varies by a factor of 1.8 from low to highdensities at t lookback = t lookback = ff erence between low and high density environmentsthus increases with decreasing lookback time. Both studies in-dicate that the transition between low and high densities occursat 0 . R ( R is the projected radius delimiting a sphere withinterior mean density 200 times the critical density at the clus-ter redshift, see Equation 1). Postman et al. (2005) also find that f E + S does not change with cluster velocity dispersion for mas-sive clusters ( σ >
800 km / s). The data for one of their clustersalso suggest that f E + S decreases for lower mass systems. Thistrend would be consistent with observations of f E + S in groupsthat show a strong trend of decreasing f E + S versus decreasing σ (Zabludo ff & Mulchaey, 1998). Finally, f E + S seems to correlatewith cluster X-ray luminosity at the 2-3 σ level (Postman et al.,2005).Recent works on stellar mass-selected cluster galaxy sam-ples (Holden et al., 2007; van der Wel et al., 2007) paint a dif-ferent picture. The fractions of E + S0 galaxies in clusters,groups and the field do not appear to have changed signifi-cantly from z ∼ . z ∼ .
03 for galaxies with massesgreater than 4 × M ⊙ . The mass-selected early-type fractionremains around 90% in dense environments ( Σ >
500 galMpc − ) and 45% in groups and the field. These results showthat the morphology-density relation of galaxies more massivethan 0.5M ∗ has changed little since z ∼ . /
902 supercluster (Wolf et al., 2007; Lane et al.,2007). Local environment appears to be more important togalaxy morphology than global cluster properties, and whilethe expected morphology-density and age-morphology relationshave been observed, there is no evidence for a morphology-density relation at a fixed age. The time since infall within thecluster environment and not density might thus be the more fun-damental parameter dictating the morphology of cluster galax-ies. A number of e ff orts have been made on the theoretical sideto model the morphological content of clusters. Diaferio et al.(2001) used a model in which the morphologies of cluster galax-ies are solely determined by their merger histories. A merger be- imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 3 tween two similar mass galaxies produces a bulge, and a newdisk may form through the subsequent cooling of gas. Bulge-dominated galaxies are in fact formed by mergers in smallergroups that are later accreted by clusters. Based on their model,they reach the following conclusions: (1) the fraction of bulge-dominated galaxies inside the virial radius should depend on themass of the cluster, and it should show a pronounced peak forclusters with mass of 3 × M ⊙ followed by a decline forlarger cluster masses. (2) The fraction of bulge-dominated galax-ies should be independent of redshift for clusters of fixed mass,and (3) the dependence of morphology on cluster mass shouldbe stronger at high redshift than at low redshift. Lanzoni et al.(2005) use the GALICS semi-analytical models and find thatearly-type fractions strongly depend on galaxy luminosity ratherthan cluster mass. By selecting a brighter subsample of galax-ies from their simulations, they find a higher fraction of ellip-ticals irrespective of the cluster mass in which these galaxiesreside. This trend is particularly noticeable in their high-densityenvironments. Observations and these earlier models clearly donot agree in important areas, and a comparison between themwould clearly benefit from a larger cluster sample size. More re-cently, the Millennium Simulation (MS; Springel et al., 2005)has provided the highest resolution model thus far of a large(0.125 Gpc ), representative volume of the Universe. Improvedtracking of dark matter structure and new semi-analytical pre-scriptions (De Lucia & Blaizot, 2007) allow the evolution of thegalaxy population to be followed with higher fidelity and bet-ter statistics than in the otherwise similar work of Diaferio et al.(2001). We will use cluster catalogues from the MS later in thispaper for comparison with our observational data.Our understanding of high-redshift cluster galaxy popu-lations in terms of their evolution as a function of redshiftand their cluster-to-cluster variations has been hampered bythe lack of comprehensive multi-wavelength (optical, near-infrared and X-ray) imaging and spectroscopic studies oflarge, homogeneously-selected samples of clusters. Many e ff ortsare underway to improve sample sizes (Gonzalez et al., 2001;Gladders & Yee, 2005; Willis et al., 2005; Postman et al., 2005).One of these e ff orts is the European Southern ObservatoryDistant Cluster Survey (“EDisCS”; White et al., 2005). TheEDisCS survey is an ESO large programme aimed at the studyof a sample of eighteen optically-selected clusters over the red-shift range 0.5-0.8. It makes use of the FORS2 spectrograph onthe Very Large Telescope for optical imaging and spectroscopyand of the SOFI imaging spectrograph on the New TechnologyTelescope (NTT) for near-infrared imaging. A number of paperson star formation in clusters (Poggianti et al., 2006, 2009a) andthe assembly of the cluster red sequence (De Lucia et al., 2004,2007; S´anchez-Bl´azquez et al., 2009; Rudnick et al., 2009) havebeen so far published from these data. In addition to the coreVLT / NTT observations, a wealth of ancillary data are also be-ing collected. A 80-orbit program for the Advanced Camera forSurveys (ACS) on the Hubble Space Telescope was devoted tothe i -band imaging of our ten highest-redshift clusters. Detailsof the HST / ACS observations and visual galaxy classificationsare given in Desai et al. (2007) and the frequency and propertiesof galaxy bars is studied in Barazza et al. (2009). X-ray observa-tions with the XMM-Newton satellite of three EDisCS clustershave been published in Johnson et al. (2006) with more clustersbeing observed. H-alpha observations of three clusters have beenpublished in Finn et al. (2005) with more clusters also being ob-served. Finally, the analysis of Spitzer / IRAC observations of allEDisCS clusters is in progress (Finn et al., in preparation). This paper presents the early-type galaxy fractions ofEDisCS clusters as a function of cluster velocity dispersion,redshift and star-formation activity. A set of local clusters ex-tracted from the Sloan Digital Sky Survey (SDSS) is usedas a comparison sample. Early-type fractions were measuredfrom two-dimensional bulge + disk decompositions on deep, op-tical VLT / FORS2 and HST / ACS images of spectroscopically-confirmed cluster member galaxies. Section 2 describes theEDisCS cluster sample selection and the imaging data. Section 3describes the procedure used to perform bulge + disk decomposi-tions on SDSS, VLT / FORS2 and HST / ACS images. Section 4presents early-type fractions for the EDisCS clusters with a de-tailed comparison between visual and quantitative morphologiesand between HST- and VLT-derived early-type fractions. It alsoincludes early-type fractions for the SDSS clusters. Changesin EDisCS early-type fractions as a function of cluster veloc-ity dispersion, redshift and star-formation activity are studiedin Section 5. Finally, Sections 6 and 7 discuss our results andtheir implications for the morphological content of clusters. Theset of cosmological parameters used throughout this paper is( H , Ω m , Ω Λ ) = (70, 0.3, 0.7).
2. Data
The sample selection and optical / near-infrared imaging data forthe EDisCS survey are described in details in Gonzalez et al.(2002), White et al. (2005) (optical photometry) and Arag´on-Salamanca et al. (near-IR photometry; in preparation).Photometric redshifts for the EDisCS clusters are presentedin Pell´o et al. (2009), and cluster velocity dispersions mea-sured from weak-lensing mass reconstructions are given inClowe et al. (2006). Spectroscopy for the EDisCS clusters is de-tailed in Halliday et al. (2004) and Milvang-Jensen et al. (2008).Clusters in the EDisCS sample were drawn from the LasCampanas Distant Cluster Survey (LCDCS) candidate catalog(Gonzalez et al., 2001). Candidate selection was constrained bypublished LCDCS redshift and surface brightness estimates.Candidates were selected to be among the highest surface bright-ness detections at each redshift in an attempt to recover someof the most massive clusters at each epoch. Using the esti-mated contamination rate for the LCDCS of ∼ − / FORS2 imaging in an e ff ort to obtain twenty (10 at z ∼ . z ∼ .
8) confirmed clusters.The z ∼ . I B and V B , and the z ∼ . I B and R sp . These filters are the stan-dard FORS2 ones. V B and I B are close approximations to theBessell (1990) photometric system while the R sp is a special fil-ter for FORS2. Final cluster candidates for deeper VLT imag-ing were selected on the basis of color and surface density ofgalaxies on the sky (White et al., 2005). The image quality onthe final stacked images ranged from 0 . ′′ . ′′
8. As describedin White et al. (2005), deep spectroscopy was not obtained fortwo cluster candidates (1122.9-1136 and 1238.5-1144), and wetherefore did not include them here. The main characteris-tics (positions, redshifts, velocity dispersions and radii) of theEDisCS cluster sample used in this paper are given in Table 1. R is the projected radius delimiting a sphere with interiormean density 200 times the critical density at the cluster redshift,and it is used throughout this paper as an important fiducial ra- Simard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters dius. R values in Table 1 were calculated using the equation: R = . σ / s 1 p Ω Λ + Ω m (1 + z ) h − Mpc (1)where h = H /
100 and σ cluster is the cluster velocity disper-sion measured using spectroscopically-confirmed cluster mem-bers (Carlberg et al., 1997; Finn et al., 2005). Cluster masseswere calculated using the equation: M cl = . × (cid:18) σ / s (cid:19) p Ω Λ + Ω m (1 + z ) h − M ⊙ (2)as in Finn et al. (2005).In practice, the redshift distributions of high- z and the mid- z samples partly overlap as can be seen from Table 1. We use only spectroscopically-confirmed cluster members tocalculate our cluster early-type fractions. Deep multislit spec-troscopy of the EDisCS was obtained with the FORS2 spectro-graph on VLT. Spectra of >
100 galaxies per cluster field wereobtained with typical exposure times of two and four hours forthe mid-z and high-z samples respectively. Spectroscopic tar-gets were selected from I -band catalogues. This correspondsto rest-frame ∼ ±
400 Å at the redshifts of the EDisCSclusters. Conservative rejection criteria based on photometricredshifts were used in the selection of spectroscopic targets toreject a significant fraction of non-members while retaining aspectroscopic sample of cluster galaxies equivalent to a purely I -band selected one. We verified a posteriori that these crite-ria excluded at most 1% of the cluster galaxies (Halliday et al.,2004; Milvang-Jensen et al., 2008). The spectroscopic selection,observations and spectroscopic catalogs are presented in detail inHalliday et al. (2004) and Milvang-Jensen et al. (2008). As de-scribed in Halliday et al. (2004), cluster redshifts and velocitydispersions were iteratively calculated using a biweight scale es-timator for robustness. Cluster members were defined as galaxieswith redshifts within the range z cluster ± σ cluster where z cluster isthe median redshift of all cluster members. In addition to our ground-based imaging, a 80-orbit program(GO 9476, PI: Dalcanton) for the Advanced Camera for Surveys(ACS) on the Hubble Space Telescope (HST) was devoted to the i -band imaging of our ten highest-redshift cluster fields. Detailsof these observations are given in Desai et al. (2007). Briefly, theHST observations were designed to coincide as closely as pos-sible with the coverage of the ground-based optical imaging andspectroscopy, within guide star constraints. The VLT / FORS2images cover a 6 . ′ × . ′ ′ from the center of the region. Forreference, the ACS WFC has a field of view of roughly 3 . ′ × . ′ . ′ × . ′ . ′ × . ′
5, which had an exposure time per pixel of 10200 seconds. The deep central pointing probes to lower sur-face brightness, fainter magnitudes, and larger galactic radii inthe region of the cluster containing the most galaxies. All expo-sures were taken under LOW SKY conditions to maximize oursurface brightness sensitivity. An image mosaic was created foreach cluster using the CALACS / Multidrizzle pipeline, and thefinal sampling of the multidrizzled image mosaics was 0 . ′′
3. Quantitative Galaxy Morphology
The source catalogs and segmentation images for the EDisCSclusters were created using the SExtractor (“Source Extractor”)galaxy photometry package version 2.2.2 (Bertin & Arnouts,1996). The SExtractor source detection was run on the combineddeep FORS2 images in “two-image” mode using the I-band im-age as the reference detection image for all the other passbands.The detection threshold was 1.5 σ bkg , and the required minimumobject area above that threshold was 4 pixels. The convolutionkernel was a 7 × / galaxy separation based on the SExtractor “stellar-ity” index was attempted. Every source was fit with a bulge + diskmodel, and unresolved sources such as stars could easily be iden-tified as output models with zero half-light radius.As SExtractor performs source detection and photometry, itis able to deblend sources using flux multi-thresholding. Thisdeblending technique works well in the presence of saddlepoints in the light profiles between objects. Each SExtractor pre-deblending “object” consists of all the pixels above the detectionthreshold that are spatially connected to one another. This groupof pixels may or may not include several real objects. The multi-thresholding algorithm assigns the pixels between two adjacentobjects and below the separation threshold based on a probabil-ity calculated from bivariate Gaussian fits to the two objects. Noassumption is made regarding the shape of the objects in this sta-tistical deblending technique. We used a value for the SExtractordeblending parameter DEBLEND − MINCONT of 0.0005. Thisvalue is subjective , and it was found through visual inspectionof several EDisCS cluster images to provide good object sep-aration. Even though the value of DEBLEND − MINCONT wasdetermined subjectively, it provides an unequivocal definition ofan object in the EDisCS catalogs. It was only determined once,and the same value of DEBLEND − MINCONT was consistentlyused for all EDisCS cluster images as well as for all the reliabil-ity tests of Section 3.2.5.
This work uses GIM2D (Galaxy IMage 2D) version 3.2, a2D decomposition fitting program (Simard et al., 2002), tomeasure the structural parameters of galaxies on the EDisCSVLT / FORS2 and HST / ACS images. GIM2D is an IRAF / SPPpackage written to perform detailed bulge + disk surface bright-ness profile decompositions of low signal-to-noise (S / N) images IRAF is distributed by the National Optical AstronomyObservatories, which are operated by the Association of Universitiesfor Research in Astronomy, Inc., under cooperative agreement with theNational Science Foundation.imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 5
Table 1.
Main characteristics of the EDisCS cluster sample: IDs, positions, redshifts, number of spectroscopically-confirmed mem-bers, velocity dispersions and radii. Clusters with HST imaging are identified by the superscript “h” in their ID.
Mid- z clustersID RA a DEC a z b Age of Universe N mem c σ d R M cl f (2000.0) (2000.0) ( × t ) (km / s) (Mpc) (10 M ⊙ )(1) (2) (3) (4) (5) (6) (7) (8) (9)1018.8-1211 10:18:46.8 − + − − + − − + − − + − h − + − − + − − + − − + − − + − z clustersID RA DEC z Age of Universe N mem σ R M cl (2000.0) (2000.0) ( × t ) (km / s) (Mpc) (10 M ⊙ )(1) (2) (3) (4) (5) (6) (7) (8) (9)1037.9-1243 h − + − h − + − h − + − h − + − h − + − h − + − h − + − h − + − h − + − a Cluster BCG Coordinates (J2000) b Cluster redshift measured from EDisCS spectroscopy c Number of cluster members confirmed by EDisCS spectroscopy d Cluster velocity dispersion measured from EDisCS spectroscopy e From equation 1 f From equation 2 of distant galaxies in a fully automated way. GIM2D is publiclyavailable, and it has been used extensively in a wide range ofdi ff erent projects so far. The fitting model used for the two-dimensional bulge + disk de-compositions of EDisCS galaxies is the same as the one used bySimard et al. (2002). It consists of a “bulge” component with ade Vaucouleurs profile and of an exponential “disk” component.We put “bulge” and “disk” between quotes to emphasize that thisconventional nomenclature does does not say anything about theinternal kinematics of the components. The presence of a “disk”component does not necessarily imply the presence of an actualdisk because many dynamically hot systems also have simpleexponential profiles. The fitting model had ten free parameters:the total galaxy flux F , the bulge fraction B / T ( ≡ ff ective radius r e , the bulgeellipticity e ( e ≡ − b / a , b ≡ semi-minor axis, a ≡ semi-majoraxis), the bulge position angle of the major axis φ b on the image(clockwise, y-axis ≡ r d (also denoted h in the literature), the disk inclina-tion i (face-on ≡ φ d on the image, thesubpixel dx and dy o ff sets of the model center with respect tothe input science image center. The sky background is not a freeparameter of the fits (see Section 3.2.3). The S´ersic index for the bulge profile is fixed at a value of n = φ b and φ d were not forced tobe equal for two reasons: (1) a large di ff erence between theseposition angles is a signature of strongly barred galaxies, and (2)some observed galaxies do have bona fide bulges that are notquite aligned with the disk position angle.The smooth bulge + disk model used here is obviously a sim-ple approximation. After all, many real galaxies will exhibitmore than two structural components such as nuclear sources,bars, spiral arms and HII regions. Even in the presence of onlya bulge and a disk, the ellipticity and / or the position angles ofthese components might be functions of galactocentric distance.The bulge + disk model is a trade-o ff between a reasonable num-ber of fitting parameters and a meaningful decomposition of dis-tant galaxy images. No non-parametric or parametric quantita-tive classification system is perfect. Any classification systemwill su ff er from biases inherent to its basic definition. However,provided a given quantitative system is clearly defined before itsuse, its results will be readily reproducible in their successes and failure by other investigators.The exact shape of bulge profiles remains under debate (e.g.,Balcells et al., 2003, and references therein). Locally, there is ev-idence that the bulges of late-type spiral galaxies may be betterfit by an n = n = Simard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters galaxies with n = B / T ≤ . ffi cult, if not impossible, to determine the S´ersicindex of distant bulges even with the spatial resolution of theHubble Space Telescope as demonstrated by an extensive setof tests on HST images of the high-redshift cluster CL1358 + n is more important forbulge-dominated galaxies, and n = n = GIM2D disk + bulge decompositions are performed on thumb-nail (or “postage stamp”) images extracted around the objectsdetected by SExtractor rather than on the entire science imageitself. The area of the thumbnail images is given by the isophotalarea of the object. Here, all thumbnails were chosen to have anarea 5 times larger than the 1.5 σ bkg isophotal area. Each thumb-nail is a square image with sides of length p × isophotal area .The first thumbnail is extracted from the science image itself,and the local background calculated by SExtractor is subtractedfrom it so that it should have a background mean level close tozero. The second thumbnail is extracted from the SExtractor seg-mentation image. The GIM2D decompositions were performedon all pixels flagged as object or background in the SExtractorsegmentation image. Object areas in the segmentation image aresharply delineated by the location of the isophote correspondingto the detection threshold because SExtractor considers all pixelsbelow this threshold to be background pixels. However, preciousinformation on the outer parts of the galaxy profile may be con-tained in the pixels below that threshold, and fits should there-fore not be restricted only to object pixels to avoid throwing thatinformation away. Pixels belonging to objects in the neighbor-hood of the primary object being fit are masked out of the fittingarea using the SExtractor segmentation image. The flux fromthe primary object that would have been in those masked areasin the absence of neighbors is nonetheless properly included inthe magnitude measurements given in this paper because mag-nitudes were obtained by integrating the best-fit models over all pixels. Special care must be paid to the determination of the local skybackground level b and dispersion σ bkg as sky errors are thedominant source of systematic errors in bulge + disk decompo-sitions of distant galaxies. As an example, overestimating thebackground sky level will lead to underestimates of the galaxytotal flux, half-light radius and bulge fraction as a result of strongparameter covariances. Even though the SExtractor local back-ground was subtracted from each galaxy thumbnail image, anadditional (residual) background estimate db was computed andused by GIM2D to correct for any systematic error in the initialSExtractor sky level estimate. In order to compute db , GIM2Dused all the pixels in the science thumbnail image flagged asbackground pixels (flag value of zero) in the SExtractor seg-mentation image. GIM2D further pruned this sample of back-ground pixels by excluding any background pixel that is closer than five pixels (1 . ′′ ff er zone ensures that the flux from all SExtracted objects inthe image below all the 1.5 σ bkg isophotes does not significantlybias the mean background level upwards and artificially inflate σ bkg . A minimum of 7500 sky pixels was imposed on the areaof the sky region. In cases where the number of sky pixels inthe input science thumbnail image was insu ffi cient, the origi-nal science image was searched for the 7500 sky pixels nearestto the object. For the EDisCS fits, background parameters werere-calculated with GIM2D before fitting, and the residual back-ground levels db were then frozen to their recalculated values forthe bulge + disk fits. The shape of the Point-Spread-Function (PSF) on theVLT / FORS2 and HST / ACS images varies significantly as afunction of position, and these variations must be taken into ac-count when point-spread-functions for the bulge + disk decompo-sitions are generated. For both sets of images, we used the stand-alone version of the stellar photometry program DAOPHOT II(Stetson, 1987) to construct spatially-varying PSF models forthe EDisCS cluster images. For each cluster and for each pass-band, we selected “clean”, point sources (detection flag of zeroand stellarity index of 0.8 or greater) from the SExtractor sourcecatalog. The positions of these point sources were fed to theDAOPHOT routine PSF to be modelled as the sum of a Gaussiancore and of an empirical look-up table representing correctionsfrom the best-fitting Gaussian to the actual observed values.Both the gaussian core parameters and the look-up table wereallowed to vary linearly as a function of x and y positions on theimage. Finally, the PSF model was used to create a PSF at theposition of each galaxy to be fit. The PSF images were 2 . ′′ Following the same procedure as in Simard et al. (2002), weperformed an extensive set of simulations to test the reliabil-ity of our sky background estimates and of the best-fit param-eter values recovered through bulge + disk fits on both sets of im-ages. 2000 smooth galaxy image models were created with struc-tural parameters uniformly generated at random in the followingranges: 20 . ≤ I ≤ .
0, 0 . ≤ B / T ≤ .
0, 0 ≤ r e ≤ . ′′ . ≤ e ≤ .
7, 0 ≤ r d ≤ . ′′
0, and 0 ◦ ≤ i ≤ ◦ . The bulgeS´ersic index was held fixed at n = ◦ for all simulations,and the bulge and disk sizes were uniformly generated in the logof the size ranges above. Each simulation was convolved with aPSF computed from one of the images with a FWHM typical ofthe VLT / FORS2 ( ∼ . ′′
8) and HST / ACS ( ∼ . ′′
05) observations.The same PSF was used in both creating and analyzing the sim-ulations, so the results will not include any error in the structuralparameters due to PSF mismatch. Poisson deviates were used toadd photon noise due to galaxy flux into the simulations. Thenoisy images were then embedded in a 20 ′′ × ′′ section of oneof the real I − band images to provide a real background for thesimulations. In addition to sky photon noise and detector read-out noise, the real background noise includes brightness fluctu-ations of very faint galaxies below the detection threshold. Thisprocedure thus yields realistic errors that include the e ff ect ofsky errors. The simulations were SExtracted exactly in the same imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 7 way as real EDisCS sources (see Section 3.1). Science and seg-mentation thumbnails extracted from the simulations were ana-lyzed with GIM2D following exactly the same steps as for thereal galaxies (see Section 3.2).Figures 1 and 2 show maps of errors on the galaxy totalmagnitude I , galaxy intrinsic half-light radius r hl and galaxybulge fraction B / T for the VLT / FORS2 images. The left-handpanels show the mean parameter errors as a function of inputgalaxy magnitude and size, and the right-hand panels show the1 σ parameter random error as a function of input galaxy mag-nitude and size. The lower number in each cell is the num-ber of simulated galaxies created for that cell. Most system-atic errors are directly related to surface brightness as magni-tudes and sizes of low surface brightness sources are inherentlyharder to measure. This fact is borne out by the trends in the er-rors shown in Figure 1. Decreasing surface brightness follows aline going from the lower left-hand corners to the upper right-hand ones. The top panels of Figures 1 show that systematicerrors on I start to become significant ( ∆ I ≃ .
2) fainter than I = r hl also increases signifi-cantly beyond this magnitude. It is important to note that I = ff ected.Figure 2 shows that systematic errors on B / T are smallest overthe region I ≤ . , − . ≤ log r hl ≤ . ff ects of PSF mismatch errorsbecause we used the same PSF for creating simulated imagesand for their analysis. However, we were able to check that theseerrors were not significant because we fitted both galaxies and stars on our real VLT / FORS2 images. The measured intrinsicradii of the stars clustered at zero, and this would not have beenthe case should PSF mismatch errors have been important.
4. Early-Type Galaxy Fractions
The bulk of the previous work on the morphological contentof high-redshift clusters is based on the visual classificationof galaxies, and this section compares visual and quantita-tive morphological classification. Visual classifications for 9200galaxies in EDisCS clusters with HST images are presentedin Desai et al. (2007). As shown by previous works (Im et al.,2002; McIntosh et al., 2002; Tran et al., 2003; Blakeslee et al.,2006), quantitative and visual morphologies can be best linkedtogether by focussing on three structural parameters: bulge frac-tion B / T , image smoothness S and bulge ellipticity e . The imagesmoothness, S , is defined as: S = R T + R A (3)where R T and R A are defined in Equation 11 of Simard et al.(2002). These two indices quantify the amount of light in sym-metric and asymmetric residuals from the fitting model respec-tively, and they are expressed as a fraction of the total galaxymodel flux. S is typically measured inside a radius that is amultiple of the galaxy half-light radius. Using our HST / ACSmeasurements, we found no di ff erences between image smooth-ness within one and two galaxy half-light radii. We thereforeuse image smoothness inside two half-light radii (and denoteit S / FORS2 images with their lower spatial resolution. We can choose selection criteria on B / T , S and e that yield the bestmatch to the visual classifications, and the particular choices arenot important as long as the same selection criteria are appliedto both local and high-redshift clusters.We divide the visually-classified EDisCS into T = − − / a) and “others” ( T > / ACSstructural parameter measurements, we find that E and S0 galax-ies have similar B / T distribution with the S0 distribution beingskewed towards slightly lower B / T , but e distributions are dif-ferent. It is therefore possible to di ff erentiate between E and S0galaxies on the basis of these two parameters. S0 and S0 / a galax-ies have similar e distributions but di ff erent B / T and S distribu-tions. Given that the bulge ellipticity e cannot be reliably mea-sured on the VLT / FORS2 images, we restrict on selection crite-ria to B / T and S
2. Figure 3 shows S B / T for the four vi-sual types of galaxies. S B / T and S / S0 galaxies selected while minimizing the contamina-tion from Sa-Irr galaxies. After several iterations, we settled on B / T ≥ .
35 and S ≤ .
075 as our definition of an early-typegalaxy. These limits are very similar to those used in previousstudies (Im et al., 2002; McIntosh et al., 2002; Tran et al., 2003).With these criteria, our quantitative selection can be translatedinto visual classification terms as f et = (cid:0) . N E + . N S + . N S / a + . N S a − Irr (cid:1) / N total (4)The coe ffi cients in Equation 4 give the completeness of thequantitative classification in terms of the Desai et al. (2007) vi-sual classes. For example, the adopted B / T and S B / T and S
2. Equation 4 is to be compared to the prescription ofvan Dokkum et al. (2000): f et = (cid:0) N E + N E / S + N S + N S / a (cid:1) / N total (5)where N total is the number of galaxies with M V ≤ − r / profile. Indeed, many early-type galaxies such asdwarf ellipticals have simple exponential profiles (Lin & Faber,1983; Kormendy, 1985), and we have verified through isophotetracing that many galaxies visually classified as early-types andmissed by our selection criteria do have radial surface bright-ness profiles that are exponential and thus consistent with theirmeasured low B / T values.Given N total galaxies brighter than an absolute magnitudelimit M V , lim inside a clustercentric radius R max of which N et areearly-types galaxies, we actually calculate the early-type galaxyfraction by finding the median of the binomial probability distri-bution p ( x ) dx = N total ! N et !( N total − N et )! x N et (1 − x ) N total − N et (6)and we integrate Equation 6 to calculate the lower and upperbounds of the corresponding 68% confidence interval. In thelimit of large N total and N et (not always true for the current clus-ter sample), this converges to the same symmetric error bars aswould be obtained from the propagation of gaussian errors. Simard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters
Fig. 1.
Two-dimensional maps of GIM2D systematic and random galaxy magnitude and half-light radius errors from 2000VLT / FORS2 image simulations.
Top left-hand panel : Systematic error on recovered galaxy total magnitude I rec as a function of input galaxy log half-light radius r hl , input in arcseconds and input galaxy total magnitude I input . The top number in each cell is themean magnitude error ( I rec − I input ), and the bottom number is the number of simulations created in that cell. Top right-hand panel :1 σ random error on I rec ( σ ( I rec − I input )) as a function of log r hl , input and I input . Bottom left-hand panel : Systematic error on recoveredgalaxy intrinsic log half-light radius r hl , rec as a function of input galaxy log half-light radius r hl , input in arcseconds and input galaxytotal magnitude I input . The top number in each cell is the mean log radius error (log r hl , rec − log r hl , input ), and the bottom number isthe number of simulations created in that cell. Top right-hand panel : 1 σ random error on log r hl , rec ( σ (log r hl , rec − log r hl , input )) as afunction of log r hl , input and I input . For each EDisCS cluster with HST / ACS imaging, we have com-puted the fraction of early-type galaxies using our quantitativeHST / ACS morphologies ( B / T ≥ .
35 and and S ≤ . V -band magnitude M V , lim . We varied M V , lim as a func-tion of redshift from − z = − z = M V , lim was made tobe fully consistent with previous work (Poggianti et al., 2006)although it may not be strictly the best choice for late-typegalaxy populations. Our results did not appear to be sensitive tovariations in M V , lim at the level of a few tens of a magnitude.Following Poggianti et al. (2006), our early-type galaxy frac-tions were also computed by weighting each galaxy accordingto the incompleteness of the spectroscopic catalog. This incom-pleteness depends on both galaxy magnitude and clustercentric position. Incompleteness as a function of magnitude was com-puted by dividing the number of galaxies in the spectroscopiccatalog in a given magnitude bin by the number of galaxies inthe parent photometric catalog in the same bin. We used 0.5 magbins here. Incompleteness due to the geometrical e ff ects comesfrom the finite number of slitlets per sky area, and the increasingsurface density of galaxies on the sky closer to the cluster cen-ters. Geometric incompleteness is field dependent as it dependson cluster richness, and we thus computed this incompletenesson a field-by-field basis. We also used four radial bins out to R with a bin width of 0.25 R .The raw and incompleteness-corrected HST-based early-type galaxy fractions are given in Table 2 for a maximum clus-tercentric radius R et of 0.6 R (columns 4 and 5) and R (columns 9 and 10). Most of the corrected fractions do not sig-nificantly di ff er from the raw ones because our spectroscopic imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 9 Fig. 2.
Two-dimensional maps of GIM2D systematic and random galaxy bulge fraction errors from 2000 VLT / FORS2 image simu-lations.
Top left-hand panel : Systematic error on recovered galaxy bulge fraction ( B / T ) rec as a function of input galaxy log half-lightradius r hl , input in arcseconds and input galaxy total magnitude I input . The top number in each cell is the mean bulge fraction error(( B / T ) rec − ( B / T ) input ), and the bottom number is the number of simulations created in that cell. Top right-hand panel : 1 σ randomerror on ( B / T ) rec ( σ (( B / T ) rec − ( B / T ) input )) as a function of log r hl , input and I input .sample is essentially complete down to I ≤
23 ( M V ∼ − z = . de facto free from the magnitude and ge-ometric incompleteness of our spectroscopic sample. Anotherimportant caveat is that they were computed using two di ff erentways to isolate cluster members (photometric redshift and statis-tical background subtraction), and they are thus not restricted tospectroscopically-confirmed members. Nonetheless, the agree-ment between fractions measured from visual and quantitativeclassifications is remarkably good. The largest disagreement isfor 1138.2-1133, but even this case can be considered marginalas it is not quite 2 σ . Quantitative morphologies measured from HST images aremore robust than those measured from ground-based images(Section 3.2.5 and Simard et al. (2002)). Figure 4 shows a directgalaxy-by-galaxy comparison between bulge fraction and im-age smoothness measurements from HST / ACS and VLT / FORS2images. This comparison includes spectroscopically-confirmedmember galaxies from all clusters with HST imaging that arebrighter than M V , lim and within a clustercentric radius of 0.6 R to take into account the e ff ect of crowding. For a given galaxy,the agreement between the two sets of measurements will ob-viously depend on its apparent luminosity and size. The over-all agreement is reasonably good. The scatter in the bulge frac-tion plot is consistent with σ B / T , ACS ∼ σ B / T , VLT ∼ Table 2.
Early-Type Galaxy Fractions Based on HST / ACS Imaging ID R et ≤ . R R et ≤ R N clus N et f et , raw f et , corr f a E / S , phz f b E / S , bkg N clus N et f et , raw f et , corr (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)1216.8-1201 45 23 0.51 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a From Table 14 of Desai et al. (2007) b From Table 16 of Desai et al. (2007)
Fig. 3.
Image smoothness parameter S B / T for di ff erent visual types. The galaxies selected by ourquantitative early-type galaxy criteria ( B / T ≥ .
35 and S ≤ . S FORS and S ACS , but it is not one-to-one. S ACS values increase fasterthan S FORS . This is expected as PSF blurring will be more sig-nificant on the ground-based images, and S ff ects. Part of the scatter is again due tothe use of independent segmentation images.The inclusion of clusters with only VLT / FORS2 imaging al-lows us to extend our analysis to nine additional clusters - animportant consideration given that we seek to probe cluster-to-cluster variations in morphological content. We therefore need toshow that we measure consistent early-type fractions for clusterswith overlapping ACS and FORS2 images. The problem boilsdown to finding the set of limits on B / T FORS and S FORS thatyield FORS2 early-type fractions in agreement with the ACSfractions obtained with B / T ACS ≥ .
35 and S ACS ≤ . M V , lim and within a clustercentric radiusof R . No corrections for incompleteness were applied hereas these corrections would be identical for both cases. We wentthrough many manual iterations until we found satisfactory lim- Fig. 5.
Comparison between early-type galaxy fractions for clus-ters with overlapping VLT and HST imaging. VLT / FORS2and HST / ACS early-type galaxy fractions were computed us-ing galaxies with B / T FORS ≥ .
40 and S FORS ≤ .
05 and B / T ACS ≥ .
35 and S ACS ≤ .
075 respectively. The ACS andFORS2 f et values plotted here are listed in column 4 of Table 2and column 4 of Table 3. Dashed line is the one-to-one line.its on B / T FORS and S FORS . We found FORS2 fractions to bein very good agreement with the ACS ones for B / T FORS ≥ . S FORS ≤ .
05 (Figure 5). This agreement is especiallygood if one considers the fact that we performed our FORS2and ACS bulge + disk decompositions completely independentlyfrom one another, i.e., we did not attempt to use the sameSExtractor segmentation map for both FORS2 and HST images.The limit on B / T FORS is slightly higher than the one on B / T ACS because lower spatial resolution typically leads to a small over-estimate of the bulge fraction. Similarly, the limit on S FORS needs to be more stringent than on S ACS to select the samegalaxies as they will look smoother on the FORS2 images dueto lower resolution. imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 11 , Fig. 4.
Direct galaxy-by-galaxy comparison between bulge fraction (left-hand panel) and image smoothness (right-hand panel)measurements from HST / ACS and VLT / FORS2 images. Filled circles are galaxies classified as early-type on both ACS and VLTimages, asterisks are galaxies classified as early-type only on the VLT images, pluses are galaxies classified as early-type only onthe ACS images, and open circles are galaxies not classified as early-type on either ACS or VLT images, The dashed lines show thecuts used for the definition of an early-type galaxy as discussed in Sections 4.1 and 4.3.Following the procedure described in Section 4.2, we com-puted early-type galaxy fraction for all eighteen clusters usinggalaxies on our FORS2 images with B / T FORS ≥ .
40 and S FORS ≤ .
05. The results are shown in Table 3. The sameincompleteness corrections as in Section 4.2 were applied hereas well. The errors on the early-type galaxy fractions in the tabledo not include errors on R due to correlated errors on cluster σ . We hereafter use our VLT / FORS2 early-type fractions for allEDisCS clusters for the sake of uniformity.
The Sloan Digital Sky Survey (SDSS; Abazajian et al., 2009) of-fers by far the best, “local” ( z < .
1) baseline for a comparisonof early-type galaxy fractions between local and high-redshiftclusters. Clusters similar in mass to EDisCS clusters can be se-lected from spectroscopic SDSS data, and galaxy morphologiescan be measured using GIM2D from SDSS images. We there-fore used SDSS-selected clusters here to construct a local base-line as nearly free of systematics as currently possible given theavailable data.We use the sample of SDSS clusters defined invon der Linden (2007). The basis of this cluster sample isthe C4 cluster catalogue (Miller et al., 2005), and we brieflyrecapitulate here how the von der Linden et al. sample wasselected. Their primary aim was to find the galaxy closest to thedeepest point of the potential well of a cluster. In order to insurethat the clusters would span a large angular extent compared tothe minimum distance of 55 arcsec between fibers, the samplewas restricted to redshifts z ≤ .
1. This first cut resulted in aninitial sample of 833 clusters. A combination of clustercentricdistance, galaxy concentration and colour cuts was used to identify brightest cluster galaxies (BCGs) for these clusters. Forcases where the same BCG was identified for more than onecluster, only the cluster with the density peak was retained, andthe others were deemed to be substructures. This cut rejected101 clusters. Refined velocity dispersion and virial radii werethen computed through an iterative process of velocity cuts. Thisprocess failed for 55 clusters, and these were also rejected. Allremaining clusters were then visually inspected. An additionalset of 35 clusters were rejected at this point as being in theinfall regions of other clusters, and another 17 clusters werediscarded because they had less than three galaxies within 3 σ of the cluster redshift and 1 R of its center. This brought thetotal of SDSS clusters down to 625. Following Poggianti et al.(2006), we applied a final redshift cut to keep clusters in therange 0.04 < z < ff ects, and the upper limit minimizes incompleteness in galaxyabsolute magnitude. Our final SDSS comparison sample thushas 439 clusters.Given that we are interested in probing galaxy propertiesas a function of environment, it is important to ensure that theSDSS and EDisCS samples both cover the same range of envi-ronments. We therefore selected a subsample of SDSS clusterswith a velocity dispersion distribution matching the EDisCS dis-tribution. This match was done by adding SDSS clusters to thesubsample one at a time and keeping only those that maintainedthe EDisCS-SDSS two-sample Kolmogorov-Smirnov probabil-ity above 50%. This is the probability of the maximum di ff erencebetween the normalized cumulative distributions of the EDisCSand SDSS samples. It means that even if the two sampls wereselected at random from the same underlying distribution, theywould di ff er by more than the two observed samples more thanhalf the time. This probability threshold thus yields a SDSS sub- Table 3.
Early-Type Galaxy Fractions Based on VLT / FORS2 Imaging ID M V , lim R et ≤ . R R et ≤ R N a f et , raw f et , corr N a f et , raw f et , corr (1) (2) (3) (4) (5) (6) (7) (8)1018.8-1211 − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± − ± ± ± ± a Number of cluster members brighter than M V , lim inside R et sample that is very well-matched to the EDisCS clusters. The re-sulting subsample (referred to as “SDSS-C4” hereafter) includes158 clusters, and these clusters are listed in Table 4.We ran GIM2D on SDSS Data Release Seven (DR7;Abazajian et al., 2009) u -, g -, r - and i -band images of objectsin the magnitude range 14 ≤ r petrosian , corr ≤ .
77 with agalaxy spectrum (i.e., with field SpecClass = + disk decompositions were successfullyobtained for 674,693 galaxies (Simard, in preparation). GIM2Dmorphologies for galaxies in our matched SDSS-C4 clusterswere extracted from this large morphological database to com-pute early-type fractions. There are two sources of incomplete-ness that must be taken into account here. The first one is in-completeness versus magnitude. We denote this spectroscopiccompleteness function as C mag ( m ) here, and we compute itaround each cluster position by taking the ratio of the numberof galaxies in the spectroscopic SDSS catalog (database tableSpecPhoto) to the number of galaxies in the photometric SDSScatalog (database table PhotoPrimary) as a function of Petrosian r magnitude. Galaxies around a given position on the sky wereextracted from the database using the SDSS “fGetNearbyOb-jEq” function. The second source of incompleteness comes fromthe spatial sampling of the SDSS fibers on the sky. Fibers can-not be placed closer than 55 ′′ from one another. This means thatregions with a higher surface density of targets could not be sam-pled as completely as regions in the global field. The net resultfor SDSS clusters is a decrease in spectroscopic sampling as afunction of decreasing clustercentric distance R . We can map thespectroscopic completeness versus R by computing the ratio ofgalaxies in the spectroscopic and photometric SDSS catalogs asa function of R . We denote this geometrical completeness func-tion as C geom ( R ) here. Ideally, C geom ( R ) should be computed foreach cluster because it will depend on cluster richness and ap-parent size (and thus indirectly on redshift). However, in prac-tice, there are not enough galaxies in a single cluster to yield C geom ( R ) with acceptable error bars. So, we opted for averagingclusters with the same redshifts and velocity dispersions to com-pute C geom ( R ). We divided the cluster list of Table 4 into threecluster groups: (1) z < .
06, (2) z > . , σ <
800 km / s, and (3) z > . , σ >
800 km / s. The weight W spec ( m , R ) in the spec-troscopic catalog of a galaxy with a r ′ -band magnitude m at aclustercentric R is thus given by the product C mag ( m ) 1 C geom ( R ) , andthe completeness-weighted early-fraction of a SDSS cluster isthen simply: f et (cid:18) M V ≤− . , B / T ≥ . , S ≤ . (cid:19) = X i ∈ [ M V ≤− . , R ≤ R et , B / T ≥ . , S ≤ . W spec ( m i , R i ) X i ∈ [ M V ≤− . , R ≤ R et ] W spec ( m i , R i ) (7)In terms of spatial resolution, the ACS, SDSS and FORS2images have sampling of 0.68 kpc / FWHM at z = . ′′ i ), 1.87 kpc / FWHM at z = . ′′ g ) and 4.5 kpc / FWHM at z = . ′′ I ) respec-tively. Even though the sampling of the ACS and FORS2 im-ages di ff ers by a factor of seven, their limits on B / T and S B / T ) S DS S , g ≥ .
35 and S S DS S , g ≤ . r =
16. If we apply our limits on ( B / T ) S DS S , g and S S DS S , g to galaxies in this catalogue, then we find that the coe ffi cientsof the SDSS-to-visual equivalent of Equation 4 would be 0.88,0.68, 0.14, and 0.014 respectively. Early-type SDSS galaxies aretherefore quantitatively selected with an “e ffi ciency” compara-ble to our selection from the ACS images.The raw fractions of [OII] emitters for the 158 SDSS-C4clusters were calculated by directly querying the SDSS database imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 13 Fig. 6.
Comparison between fractions of [OII] emitters com-puted using emission-line measurements from Brinchmann et al.(2004) and the DR7 release. Filled and open circles are clusterswith σ ≥
600 km / s and σ <
600 km / s respectively.table SpecLine for the [OII]3727 and [OII]3730 equivalentwidths for each confirmed cluster member, adding them togetherand correcting them to rest-frame by dividing by (1 + z ). Thecorrected [OII] fractions were then computed following exactlythe same calculations (and using the same weights, the sameluminosity and clustercentric radius cuts of M V ≤ − . R ≤ . R ) as for the early-type fractions except that theearly-type selection criteria on bulge fraction and image smooth-ness were simply replaced by the Poggianti et al. (2006) cut ofEW([OII]) ≤ − R ≤ . R for the 158 SDSS clus-ters in our local comparison sample. We included only galax-ies brighter than M V , lim = − . ff magnitude corre-sponds to the absolute magnitude limits we used for our distantEDisCS clusters once passive evolution is taken into account(see Section 4.2). Numerical simulations of dark matter haloes populated withgalaxies using semi-analytical models greatly help in the inter-pretation of observational results. We use here the MillenniumSimulation (MS; Springel et al., 2005), and the semi-analytical code described in De Lucia & Blaizot (2007) . The MS followed2160 particles of mass 8.6 × h − M ⊙ within a comoving boxof size 500 h − Mpc on a side with a spatial resolution of 5 h − kpc. Early-type galaxy fractions were computed from these sim-ulated galaxy catalogs using the following procedure. Haloeswere randomly selected at three di ff erent redshifts ( z =
0, 0.41,0.62) so that they were uniformly distributed in log( M ). Thefinal halo sample was 100 haloes at z =
0, 94 haloes at z = z = ∆ M = M bulge − M total (in the rest-frame B -band). Galaxies with ∆ M < B / T FORS ≥ .
40. It is important hereto note that an early-type galaxy in the simulations was definedsolely based on this cut in bulge fraction because the simulationsdo not have the resolution required to model internal fine struc-tures such as asymmetries. Given that real, early-type galaxieswere also selected according to image smoothness, one mightfind the early-type fractions of real clusters to be systematicallylower. For each halo, the fraction of early-type galaxies within0.6 R from the BCG was computed using three di ff erent pro-jections. Furthermore, only galaxies that were within 2 Mpcfrom the BCG along the line of sight were included. The frac-tions were computed using only galaxies brighter than − − − V − band at redshift 0.6, 0.4,and 0.0 respectively to match the limits used for the SDSS andEDisCS early-type galaxy fractions. A galaxy in the simulationwas deemed to be star-forming if its star-formation rate in thelast timestep of its evolution was not equal to zero.Figure 7 shows the resulting model early-type fractions asa function of cluster velocity dispersion, redshift and fractionof star-forming galaxies for the MS haloes. At a given red-shift, there is no dependence of the early-type fraction on clustervelocity dispersion, but the scatter symmetrically increases to-wards both lower and higher fractions leading to a ”wedge-like”distribution towards lower cluster σ ’s. The early-type fractionsof both low and high-mass clusters increase with decreasing red-shift from ∼ z = .
65 to ∼ z =
0. The early-typefractions of massive clusters are anticorrelated with the fractionsof star-forming galaxies: clusters at z =
5. Results
We use here our VLT / FORS2 early-type fractions for all EDisCSclusters for the sake of uniformity. Simulated galaxy catalogs used here are publicly available athttp: // / millennium / ,, Fig. 7.
Early-type galaxies in Millennium Simulation dark matter haloes
Top, left-hand panel:
Early-type galaxy fraction within0 . R versus cluster velocity dispersion at three di ff erent redshifts. Top, right-hand panel:
Early-type galaxy fraction within0 . R versus age of the universe. Blue and red points are clusters with velocity dispersions below and above 600 km / s respectively. Lower, left-hand panel:
Early-type galaxy fraction within 0 . R versus fraction of star-forming galaxies in clusters with σ < / s. Blue points show haloes selected at redshift zero, and all the other haloes are in red. Lower, right-hand panel:
Early-typegalaxy fraction within 0 . R versus fraction of star-forming galaxies in clusters with σ ≥
600 km / s. Figure 8 shows early-type galaxy fractions versus velocity dis-persion for the SDSS and EDisCS clusters. The early-typegalaxy fractions of both cluster samples exhibit no clear trendas a function of σ . Table 5 gives Spearman rank test results forthe SDSS sample and di ff erent EDisCS subsamples. The onlysignificant correlation between early-type fraction and velocitydispersion is found in the high-z EDisCS clusters. It only has a2.5% chance of being due to randon sampling. Such a positivecorrelation was also reported in Desai et al. (2007) for the samecluster subsample, but it disappears when the full EDisCS sam-ple is considered. The lack of a significant correlation agreeswell with the results for the Millennium Simulation in the topleft-hand panel of Figure 7 but disagrees with the earlier the- oretical results of Diaferio et al. (2001) which showed a trendbetween f et and σ . A visual inspection of Figure 8 confirmsthe statistical test results. The mid-z EDisCS clusters do notshow any correlation with σ in contrast to the high-z clusters.In particular, two mid- z EDisCS clusters (CL1119.3-1130 andCL1420.3-1236) with σ ∼
200 km / s have early-type fractionssimilar or higher ( f et ∼ σ relation in the sense that they have a lowfraction of [OII] emitters for their mass. This is consistent withwhat we observe here given that early-type galaxies typicallyhave lower [OII] emission fluxes.Figure 8 does show that there is a marked di ff erence in themorphological content of the EDisCS and SDSS clusters. All imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 15 Table 5.
Spearman Rank Tests Results for Early-Type Fractionversus Cluster Velocity Dispersion
Cluster Sample N cl R s p -value(1) (2) (3) (4)SDSS 158 − − Table 6.
Two-sample Kolmogorov-Smirnov Test Probabilitiesfor Early-Type Fraction versus Cluster Velocity Dispersion
Cluster Sample 2SDSS SDSS EDisCS EDisCSCluster Sample 1 (All) ( σ ≥ σ < σ ≥ σ < σ ≥ σ < EDisCS f et values (with the exception of one cluster) are be-low 0.6, but half of the SDSS clusters are above this value. Thepopulation of early-type galaxies has thus increased significantlyin half of the clusters of all velocity dispersions . An increasein early-type fraction with decreasing redshift may already bevisible when one compares mid- z and high- z EDisCS clusters.Mid- z clusters around σ =
500 km / s have f et ≃ z clusters have f et ≃ ∼
25% in-crease over a time interval of 2 Gyrs. As shown in Figure 7, theearly-type fractions of clusters in the Millenium Simulation alsoincrease with decreasing redshift in clusters of all velocity dis-persions, but there is a lack of simulated clusters with f et < . f et val-ues of simulated clusters is also smaller than in those of realclusters. For simulated clusters at z = σ ≥
600 km / s, σ ( f et ) = σ ( f et ) = f et values is 0.12, the intrinsic scatter would be0.17. This intrinsic scatter is still almost three times the scatterin the simulated clusters.Figure 9 shows SDSS and EDisCS early-type fractions as afunction of the age of the universe (i.e., redshift). The clustershave been divided into two subgroups based on their velocitydispersions. The early-type fractions of massive ( σ >
600 km / s)EDisCS clusters (right panel) are in very good agreement withthe ones in the compilation of van Dokkum et al. (2001) whichalso have velocity dispersions greater than 600 km / s. The clus-ters at low redshift in the van Dokkum et al. (2001) compilationsuggest that there are no local clusters with low early-type frac-tions and hence that all clusters have uniformly increased theirearly-type fraction from z ∼ . to the present day. However, ourSDSS cluster sample shows that this simple picture is not en-tirely true. While half of the SDSS clusters have higher early-type fractions than clusters at high redshift, the other half haveearly-type fractions equal or even lower than the EDisCS clus-ters. The same holds true for the low mass clusters (left-handpanel). The scatter in f et ( < Table 7.
Spearman Rank Tests Results for Early-Type Fractionversus Fraction of [OII] Emitters
Cluster Sample N cl R s p -value(1) (2) (3) (4)SDSS (All) 158 − × − SDSS ( σ < − × − SDSS ( σ ≥ − × − EDisCS (All) 18 − σ < − σ ≥ − Table 8.
Two-sample Kolmogorov-Smirnov Test Probabilitiesfor [OII] Emitter Fraction versus Cluster Velocity Dispersion
Cluster Sample 2SDSS SDSS EDisCS EDisCSCluster Sample 1 (All) ( σ ≥ σ < σ ≥ σ < σ ≥ σ < Figure 7. There is a clear deficit of clusters with low early-typefraction at low redshift in the Millenium Simulation comparedwith our SDSS sample.
The link between star formation and morphological transforma-tion and its evolution as a function of redshift provides moreclues on the processes driving galaxy morphology in local anddistant clusters. The fractions of galaxies with [OII] emissionin the EDisCS clusters were computed as in Poggianti et al.(2006) using the same absolute magnitude limits and the sameprescriptions for correcting magnitude and geometric incomplet-ness, but the clustercentric radius cut was changed to matchthe one used for the early-type fractions in this paper ( R et ≤ . R ). The two datasets are therefore directly comparable.Figure 10 shows f et versus f [ OII ] with our local and distant sam-ples again divided according to velocity dispersion. Table 7 givesSpearman test results between f et and f [ OII ] . There is a strongcorrelation between f et versus f [ OII ] in both SDSS and EDisCScluster samples irrespective of cluster velocity dispersion. TheEDisCS clusters lie within the envelopes defined by the SDSSclusters. There is no o ff set between the zeropoints of the corre-lations at low and high redshift. However, as demonstrated byPoggianti et al. (2006), the star formation activity (parametrizedby f [ OII ] ) has decreased in all environments from z ∼ .
75 to z ∼ .
08. This is confirmed by the K-S test results in Table 8.The probabilities that the EDisCS and SDSS clusters are drawnfrom the same parent f [ OII ] distribution are only 0.026, 0.005and 0.046 for the whole samples, low σ and high σ subsamplesrespectively.The f et versus f [ OII ] values for clusters from the MilleniumSimulation (Figure 7) are quite di ff erent from the observations.Low σ MS clusters at low and high redshifts are confined to high f et and f OII values with no apparent correlation. There is only ahandful of clusters with low values for both f et and f OII . Thehigh σ MS clusters are found in a very limited range of f et and f OII values (0 . < f OII < .
75, 0 . < f et < . , Fig. 8.
Early-type galaxy fraction within 0 . R versus velocity dispersion for SDSS and EDisCS clusters. Both samples have beenmatched in velocity dispersion. Left panel:
SDSS clusters. Only typical error bars are shown in the lower right-hand corner forclarity.
Right panel:
Filled and open circles are mid- z and high- z EDisCS clusters respectively. Errors bars shown in both panels are1 σ errors. Our VLT / FORS2 early-type fractions are used here for all EDisCS clusters for the sake of uniformity. , Fig. 9.
Early-type galaxy fraction versus age of the universe (i.e., redshift) for clusters with σ <
600 km / s (left panel) and clusterswith σ ≥
600 km / s (right panels). SDSS and EDisCS clusters are blue and red respectively, and both samples have been matched invelocity dispersion. Clusters shown in black are from the compilation of van Dokkum et al. (2001) in which open and solid pointshave X-ray luminosities below and over 10 . ergs s − respectively. Our VLT / FORS2 early-type fractions are used here for allEDisCS clusters for the sake of uniformity.
6. Discussion
In order to fully understand possible evolutionary trends ob-served here, it is important to determine how cluster velocitydispersion changes with redshift as a result of the hierarchicalgrowth of structures. Are we looking at similar clusters whenwe focus on the same range of velocity dispersions in the SDSSand EDisCS clusters? Poggianti et al. (2006) looked at the meanchange in σ between z = z = × and 5 × M ⊙ . Their Figure 8shows how σ evolves over that redshift interval. For example,a z = σ =
900 km / s would typically have σ ∼
750 km / s at z = ffi cient to introducebiases in our analysis here. Indeed, selecting clusters with σ ≥
600 km / s, say, at either z = z = N = N <
10 for which velocity disper-sions may be less reliable and found that our results remainedunchanged.Poggianti et al. (2006) proposed a scenario in which twochannels are responsible for the production of passive galaxiesin clusters, and others (Faber et al., 2007; Brown et al., 2007)have proposed a similar scenario for the migration of galaxiesfrom the ”blue cloud” to the red sequence. ”Primordial passive imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 17 , Fig. 10.
Early-type galaxy fraction versus [OII] emitter fraction for clusters with σ <
600 km / s (left panel) and clusters with σ ≥
600 km / s (right panel). SDSS and EDisCS clusters are shown in blue and red respectively, and both samples have beenmatched in velocity dispersion. Only typical error bars are shown for the SDSS clusters in the lower right-hand corner for clarity.Our VLT / FORS2 early-type fractions are used here for all EDisCS clusters for the sake of uniformity.galaxies” are composed of galaxies whose stars all formed atvery high redshift ( z >
2) over a short timescale. These galax-ies have been observed in clusters up and beyond z =
1, andthey largely comprise luminous ellipticals. ”Quenched passivegalaxies” have had a more extended period of star formationactivity, and their star formation has been quenched after theirinfall into dense cluster environments. These quenched passivegalaxies would then su ff er the e ff ects of cluster processes suchas ram pressure stripping, harassment, strangulation and merg-ers to become S0 and earlier type galaxies. A key point of thisscenario is that processes a ff ecting morphology and star forma-tion activity operate on di ff erent timescales as shown recently forthe EDisCS sample by S´anchez-Bl´azquez et al. (2009). Thereis good evidence that star formation is quenched in galaxiesover timescales of 1-3 Gyr after they have entered the clusterenvironment (Poggianti et al., 1999, 2006) whereas morpholog-ical transformation through mergers and harassment can takelonger ( ∼ / FORS2 images do not have su ffi cient spatial resolution todisentangle E and S0 galaxies as mentioned in Section 4.1 to de-termine the exact contribution from each channel. We can there-fore only study the overall production of early-type galaxies, butit should exhibit di ff erent behaviors with cluster global proper-ties depending on the process(es) dominating it. Given our quan-titative definition of an early-type galaxy based on bulge fraction and image smoothness, there are essentially two ways to trans-form late-type galaxies into early-type ones: 1) processes such ascollisions and harassment that can fundamentally alter the struc-ture of a galaxy by forming bulges and / or destroying disks and2) quenching processes that can extinguish star forming regionsresponsible for some of the galaxy image asymmetries and alsocause a fading of the disks.Applying the Poggianti et al. (2006) scenario to our re-sults, the ”threshold” in f et values in our high redshift clusters(Figures 8 and 9) could be explained by a population of pri-mordial passive galaxies that formed at even higher redshifts.Most of our high redshift clusters have early-type fractions inthe range 0.3-0.6 with no correlation with cluster velocity dis-persion. Are these early-type fractions indeed consistent with apopulations of primordial passive galaxies? Calculations donein Poggianti et al. (2006) show that the fraction of galaxies at z = . × M ⊙ at z = ± / f et on clus-ter velocity dispersion.One of our main results is that the early-type fractions ofgalaxy clusters increase from z = . − . z ∼ .
08 inclusters of all velocity dispersions. What kind of morpholog-ical transformation process(es) can lead to such an evolution?Collisions and harassment both depend on galaxy-galaxy inter-actions and the time a galaxy has spent within the cluster en-vironment. Cluster velocity dispersion influences the number ofinteractions and their duration. Higher velocity dispersions inmore massive clusters yield more interactions per unit time N but with shorter durations ∆ t in a given time interval. One mighttherefore expect to see a peak in early-type type fraction at the cluster velocity dispersion where the product N ∆ t is maximized.No such peak is seen in our clusters. Ram-pressure strippingis expected to go as ( n ICM v . gal ) / ˙ M rep (Gaetz et al., 1987) with n ICM , v gal and ˙ M rep being the density of the ICM, the veloc-ity of the galaxies within the ICM and the rate at which galax-ies can replenish their gas respectively. The fraction of passivegalaxies should therefore be a relatively strong function of clus-ter velocity dispersion if quenching by ram pressure stripping isthe dominant process. The number of post-starburst galaxies inEDisCS clusters does correlate with cluster velocity dispersion(Poggianti et al., 2009a), but the uniform increase in early-typefractions at all cluster velocity dispersions observed going fromEDisCS to SDSS clusters is not consistent with the intraclustermedium being the main cause of the changes in cluster morpho-logical content.Even though the early-type and [OII] emitter fractions inEDisCS and SDSS clusters show no correlation with cluster ve-locity dispersion (Poggianti et al., 2006, and this work), thereis a very strong correlation between f et and f OII . This corre-lation is seen at both low and high cluster masses as well asat both low and high redshifts. Morphology and star formationtherefore appear to be closely linked with one another over awide range of environments and times. However, di ff erent struc-tural transformation and quenching processes are thought tooperate over di ff erent timescales (e.g., S´anchez-Bl´azquez et al.,2009). Timescales range from 1-2 Gyr (based on typical clus-ter crossing times) for truncating star formation to 3-5 Gyr fortotally extinguishing star formation in newly accreted galaxies(Poggianti et al., 2006; Tonnesen & Bryan, 2009). Looking atthe evolution of EDisCS cluster red-sequence galaxies over 2Gyr (from z = .
75 to z = z EDisCS clus-ters may be ∼
25% higher than the ones of high- z clusters. Thischange would therefore have taken place over a 2 Gyr intervalin our adopted cosmology. However, the time baseline here be-tween SDSS and EDisCS clusters is almost 6 Gyr, and, unfor-tunately, this is ample time to erase any di ff erence arising fromdi ff erent timescales in the link between morphology and star for-mation.The lack of dependence of morphology and star forma-tion on global cluster properties such as velocity dispersionraises the question of whether changes in galaxy properties aredriven by more local e ff ects or whether they occur outside ofthe cluster environment. Recent work (Poggianti et al., 2008;Park & Choi, 2009; Bamford et al., 2009; Ellison et al., 2009)have re-emphasized the strong link between galaxy propertiesand local galaxy density rather than cluster membership. Galaxyproperties are seen to change at densities around 15-40 galax-ies Mpc − or projected separations of 20-30 h − kpc. Others(e.g., Kautsch et al., 2008; Wilman et al., 2009) have suggestedthat the galaxy group environment might be more conduciveto galaxy transformation. Our observed evolution in early-typefraction as a function of redshift and the strong correlation be-tween morphology and star formation at all cluster masses wouldsupport the idea that cluster membership is of lesser importancethan other variables such as local density in determining galaxyproperties.The properties of simulated clusters from the MilleniumSimulation compare well with those of EDisCS and SDSS clus-ters. Their early-type fractions also show no dependence withcluster velocity dispersion in contrast to previous theoretical work (e.g. Diaferio et al., 2001) but in agreement with observa-tions. However, there is a definite lack of MS clusters with lowearly-type fractions at z = z =
0, andsuch a large discrepancy could only be explained by a signifi-cant population of real bulge-dominated galaxies with relativelylarge asymmetries. It is more likely that bulge formation in thesimulations may be too e ffi cient. The scatter in f et values for thesimulated clusters with σ ≥
600 km / s is also nearly three timessmaller than observed in the real clusters (Section 5.1) whichmay indicate that the models may not include the right mixtureof evolutionary processes at work on real galaxies. High-masssimulated clusters show a correlation between early-type frac-tion and star-forming fraction (albeit over narrower ranges thanobserved), but the correlation is not seen in the low-mass sim-ulated clusters. This may be understood by high mass clustershaving been formed long enough for evolutionary processes tohave had enough time to act on galaxies to modify their proper-ties whereas this is not necessarily the case for low-mass clus-ters. The fact that the correlation is observed in both low- andhigh-mass real clusters may be an indication that processes giv-ing rise to the correlation may be more e ffi cient (or altogetherdi ff erent) than modelled. It is also important to keep in mindhere that the properties of a galaxy in these models are essen-tially driven by the mass of its parent halo.
7. Summary
We have presented quantitative morphologies measured fromPSF-convolved, 2D bulge + disk decompositions of cluster andfield galaxies on deep VLT / FORS2 images of eighteen,optically-selected galaxy clusters at 0 . < z < .
80 observed aspart of the ESO Distant Cluster Survey. The morphological con-tent of these clusters was characterized by the early-type fractionwithin a clustercentric radius of 0.6 R , and early-type galaxieswere selected based on bulge fraction and image smoothness. Weshowed a very good agreement between quantitative and visualgalaxy classifications. We used a set of 158 clusters extractedfrom the Sloan Digital Sky Survey matched in velocity disper-sion to our EDisCS sample and analyzed exactly in the sameway to provide a robust comparison baseline and to control sys-tematics. We studied trends in early-type fraction as a functionof cluster mass and redshift. We also explored the link betweenmorphology and star formation by comparing early-type frac-tions to the fractions of [OII] emitters in our clusters. Our mainresults are:1. The early-type fractions of the SDSS and EDisCS clustersexhibit no clear trend as a function of cluster velocity dispersion.2. Mid- z EDisCS clusters around σ =
500 km / s have f et ≃ z EDisCS clusters have f et ≃ ∼
25% increase over a time interval of 2 Gyrs.3. There is a marked di ff erence in the morphological contentof the EDisCS and SDSS samples. None of the EDisCS clus-ters have an early-type fraction greater than 0.6 whereas half ofthe SDSS clusters lie above this value. This di ff erence is seen inclusters of all velocity dispersions (i.e., masses) . imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 19
4. There is a strong and clear correlation between morphol-ogy and star formation activity in the sense that decreasing frac-tions of [OII] emitters are tracked by increasing early-type frac-tions. This correlation holds in both low and high cluster massesas well as at both low and high redshift.5. The early-type fractions of clusters drawn from theMillennium Simulation (Springel et al., 2005) using the galaxyformation model of De Lucia & Blaizot (2007) also show noclear dependence on cluster velocity dispersion. However, at z =
0, they are not enough simulated clusters with low early-typefractions compared to the SDSS cluster sample. While high-mass simulated clusters show a correlation between early-typefraction and star-forming fraction (albeit over narrower rangesthan observed), this correlation is not seen in the low-mass sim-ulated clusters in contrast to the real ones.Our results pose an interesting challenge to structural trans-formation and star formation quenching processes that stronglydepend on the global cluster environment (e.g., a dense ICM)and suggest that cluster membership may be of lesser impor-tance than other variables in determining galaxy properties.
Acknowledgements.
We are thankful to the anonymous referee for sugges-tions that greatly contributed this paper. We have benefitted from the gen-erosity of the ESO / OPC. G. R. thanks Special Research Area No 375 of theGerman Research Foundation for financial support. The Millennium Simulationdatabases used in this paper and the web applications providing access to themwere constructed as part of the activities of the German Astrophysical VirtualObservatory. Funding for the creation and distribution of the SDSS Archive hasbeen provided by the Alfred P. Sloan Foundation, the Participating Institutions,the National Aeronautics and Space Administration, the National ScienceFoundation, the U.S. Department of Energy, the Japanese Monbukagakusho,and the Max Planck Society. The SDSS Web site is http: // / .TheSDSS is managed by the Astrophysical Research Consortium (ARC) for theParticipating Institutions. The Participating Institutions are The University ofChicago, Fermilab, the Institute for Advanced Study, the Japan ParticipationGroup, The Johns Hopkins University, the Korean Scientist Group, Los AlamosNational Laboratory, the Max-Planck-Institute for Astronomy (MPIA), theMax-Planck-Institute for Astrophysics (MPA), New Mexico State University,University of Pittsburgh, University of Portsmouth, Princeton University, theUnited States Naval Observatory, and the University of Washington. The DarkCosmology Centre is funded by the Danish National Research Foundation. References
Abazajian, K. et al. 2009, arXiv:0812.0649Allen, S. W. 1998, MNRAS, 296, 392Andredakis, Y. C. 1998, MNRAS, 295, 725Balcells, M., Graham, A. W., Dom´ınguez-Palmero, L., & Peletier, R. E. 2003,ApJ, 582, 79Bamford, S. P. et al. 2009, MNRAS, 393, 1324Barnes, J. E., & Hernquist, L. 1992, ARA&A, 30, 705Barnes, J. E., & Hernquist, L. 1996, ApJ, 471, 115Barazza, F. D. et al. 2009, A&A, 497, 713Bessell, M. S. 1990, PASP, 102, 1181Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393Blakeslee, J. P. et al. 2006, ApJ, 644, 30.Borgani, S., Girardi, M., Carlberg, R. G., Yee, H. K. C., & Ellingson, E. 1999,ApJ, 527, 561Brinchmann et al. 2004, MNRAS, 351, 1151Brown, M. J. I. 2007, ApJ, 654, 858Carlberg, R. G., Yee, H. K. C., & Ellingson, E. 1997, ApJ, 478, 462Clowe, D. et al. 2006, A&A, 451, 395Courteau, S., de Jong, R. S., & Broeils, A. H. 1996, ApJ, 457, 73de Jong, R. S. 1996, A&A, 118, 557De Lucia, G., Poggianti, B. M., Arag´on-Salamanca, Clowe, D., Halliday, C.,Jablonka, P., Milvang-Jensen, B., Pell´o, R., Poirier, S., Rudnick, G., Saglia,R., Simard, L., & White, S. D. M. 2004, ApJ, 610, 77De Lucia, G., & Blaizot, J.. 2007, MNRAS, 375, 2De Lucia, G., Poggianti, B. M., Arag´on-Salamanca, White, S. D. M., Zaritsky,D., Clowe, D., Halliday, C., Jablonka, P., von der Linden, A., Milvang-Jensen, B., Pell´o, R., Poirier, S., Rudnick, G., Saglia, R., & Simard, L., 2007,MNRAS, 374, 809. Desai, V., Dalcanton, J. J., Arag´on-Salamanca, A., Jablonka, P., Poggianti, B.,Gogarten, S. M., Simard, L., Milvang-Jensen, B., Rudnick, G., Zaritsky, D.,Clowe, D., Halliday, C., Pell´o, R., Saglia, R., & White, S. D. M. 2007, ApJ,660, 1151Diaferio, A., Kau ff mann, G., Balogh, M. L., White, S. D. M., Schade, D., &Ellingson, E. 2001, MNRAS, 323, 999Dressler, A. 1984, ApJ, 281, 512Dressler, A., Oemler, A., Jr., Couch, W. J., Smail, I., Ellis, R. S., Barger, A.,Butcher, H., Poggianti, B. M., & Sharples, R. M. 1997, ApJ, 490, 577Ellison, S. L. et al. 2009, MNRAS, in press.Faber, S. M. et al., ApJ, 665, 265Farouki, R. T.& Shapiro, S. L. 1981, ApJ, 243, 32Farouki, R. T.& Shapiro, S. L. 1982, ApJ, 259, 103Fasano, G., Poggianti, B., Couch, W. J., Bettoni, D., Kjaergaard, P., & Moles, M.2000, ApJ, 542, 673Finn, R. A., Zaritsky, D., McCarthy, D. W. Jr., Poggianti, B., Rudnick, G.,Halliday, C., Milvang-Jensen, B., Pell´o, R., & Simard, L. 2005, ApJ, 630,206Fukugita, M. et al. 2007, AJ, 134, 579Gaetz, T. J. 1987, ApJ, 316, 530Gladders, M. D., & Yee, H. K. C. 2005, ApJS, 157,1Gonzalez, A. H., Zaritsky, D., Dalcanton, J. J., & Nelson, A. 2001, ApJS, 137,117Gonzalez, A. H., Zaritsky, D., Simard, L., Clowe, D., & White, S. D. M. 2002,ApJ, 579, 577Gunn, J. E., & Gott, J. R. I. 1972, ApJ, 176, 1Halliday, C. et al. 2004, A&A, 427, 397Hoekstra, H., Franx, M., & Kuijken, K. 2000, ApJ, 532, 88Holden, B. P., Stanford, S. A., Eisenhardt, P., & Dickinson, M. 2004, ApJ, 127,2484Holden, B. P., Illingworth, G. D., Franx, M., Blakeslee, J. P., Postman, M.,Kelson, D. D., van der Wel, A., Demarco, R., Magee, D. K., Tran, K.-V.,Zirm, A., Ford, H., Rosati, P., & Homeier, N. 2007, ApJ, 670, 190Holden, B. P. et al. 2009, ApJ, 693, 617Im, M., Simard, L., Faber, S. M., Koo, D. C., Gebhardt, K., Willmer, ChristopherN. A., Phillips, A., Illingworth, G. D., Vogt, N. P., Sarajedini, V. L. 2002, ApJ,571, 136Johnson, O., Best, P., Zaritsky, D., Clowe, D., Arag´on-Salamanca, A., Halliday,C., Jablonka, P., Milvang-Jensen, B., Pell´o, Poggianti, B. M., Rudnick, G.,Saglia, R., Simard, L., & White, S. D. M. 2006, MNRAS, 371, 1777.Kaiser, N., & Squires, G. 1993, ApJ, 404, 441Kautsch, S. J. et al. 2008, ApJ, 688, 5Kormendy, J. 1985, ApJ, 295, 73Lane, K. P., Gray, M. E., Arag´on-Salamanca, A., Wolf, C., & Meisenheimer, K.2007, MNRAS, 378, 716Lanzoni, B., Guiderdoni, B., Mamon, G. A., Devriendt, J., & Hatton, S. 2005,MNRAS, 361, 369Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980, ApJ, 237, 692Lin, D. N. C. & Faber, S. M. 1983, ApJ, 266, 17Lubin, L. M., Oke, J. B., & Postman, M. 2002, AJ, 124, 1905Martig, M., Bournaud, F., Teyssier, R., & Dekel, A. 2009, ApJ, submittedMcIntosh, D.H., Rix, H.-W., Caldwell, N. 2002, ApJ, 610, 161Milvang-Jensen, B. et al. 2008, A&A, 482, 419Mihos, J., & Hernquist, L. 1996, ApJ, 464, 641Miller, C. J. et al. 2005, AJ, 130, 968Moran, S.M et al. 2007, ApJ, 671, 1503Moore, B., Katz, N., Lake, G., Dressler, A., & Oemler, A. 1996, Nature, 379,613Moore, B., Lake, G., & Katz, N. 1998, ApJ, 495, 139Negroponte, J., & White, S. D. M. 1983, MNRAS, 205, 1009Park, C., & Choi, Y.-Y. 2009, ApJ, 691, 1828Pell´o, R. et al. 2009, A&A, in press.Poggianti, B. M. et al. 1999, ApJ, 518, 576Poggianti, B. M. et al. 2006, ApJ, 642, 188Poggianti, B. M. et al. 2008, ApJ, 684, 888Poggianti, B. M. et al. 2009a, ApJ, 693, 112Poggianti, B. M. et al. 2009b, ApJ, 697, L137Postman, M. et al. 2005, ApJ, 623, 721Quilis, V., Moore, B., & Bower 2000, Science, 288, 1617Rood, H. J., Page, T. L., Kintner, E. C., & King, I. R. 1972, ApJ, 175, 627Rudnick, G. et al. 2009, ApJ, in press.S´anchez-Bl´azquez, P. et al. 2009, A&A, 499, 47Schneider, P., & Seitz, C. 1995, A&A, 294, 411Simard, L. et al. 2002, ApJS, 142, 1Smith, G. P., Treu, T., Ellis, R. S., Moran, S. M., & Dressler, A. 2005, ApJ, 620,78Spitzer, L. J., & Baade, W. 1951, ApJ, 113, 413Springel, V. 2005, Nature, 435, 629 Steinmetz, M., & Navarro, J. F. 2002, New Astronomy, 7, 155Stetson, P. B. 1987, PASP, 99, 191Tonnesen, S., & Bryan, G. L. 2009, ApJ, 694, 789Toomre, A., & Toomre, J. 1972, ApJ, 178, 623Tran, K.-V. H., Kelson, D. D., van Dokkum, P., et al. 1999, ApJ, 522, 39Tran, K.-V. H., Simard, L., Illingworth, G. D., & Franx, M. 2003, ApJ, 590, 238van der Wel, A., Holden, B. P., Franx, M., Illingworth, G. D., Postman, M. P.,Kelson, D. D., Labb´e, I., Blakeslee, J. P., & Ford, H. C. 2007, ApJ, 670, 206van Dokkum, P. G., Franx, M., Fabricant, D., Illingworth, G. D., & Kelson, D.D. 2000, ApJ, 541, 95van Dokkum, P. G., Stanford, S. A., Holden, B. P., Eisenhardt, P. R., Dickinson,M., & Elston, R. 2001, ApJ, 552, 101von der Linden, A., Best, P. N., Kau ff mann, G., & White, S. D. M. 2007,MNRAS, 379, 867White, S. D. M. et al. 2005, A&A, 444, 365.Willis, J. P. et al. 2005, MNRAS, 363, 675Wilman, D. J. et al. 2009, ApJ, 692, 298Wolf, C., Gray, M. E., Arag´on-Salamanca, A., Lane, K. P., & Meisenheimer, K.2007, MNRAS, 376, L1Zabludo ff , A. I., & Mulchaey, J. S. 1998, ApJ, 496, 39 imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 21 Table 4.
Velocity-dispersion-matched sample of 158 SDSS clusters in order of decreasing velocity dispersion
SDSS C4 z N σ v R M cl R ≤ . R ID (km / s) (Mpc) (10 M ⊙ ) f et , raw f et , corr f [ OII ] , raw f [ OII ] , corr (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)3004 0.0801 199 1156 ±
61 2.75 2.550 0.69 ± ± ± ± ±
114 2.60 2.117 0.38 ± ± ± ± ±
59 2.30 1.491 0.77 ± ± ± ± ±
71 2.26 1.364 0.56 ± ± ± ± ±
69 2.20 1.281 0.28 ± ± ± ± ±
56 2.18 1.231 0.70 ± ± ± ± ±
102 2.15 1.216 0.50 ± ± ± ± ±
113 2.13 1.174 0.50 ± ± ± ± ±
77 2.08 1.113 0.62 ± ± ± ± ±
54 2.09 1.101 0.69 ± ± ± ± ±
134 2.03 1.024 0.71 ± ± ± ± ±
78 1.97 0.937 0.60 ± ± ± ± ±
54 1.98 0.929 0.60 ± ± ± ± ±
49 1.94 0.885 0.63 ± ± ± ± ±
56 1.92 0.851 0.57 ± ± ± ± ±
54 1.91 0.827 0.68 ± ± ± ± ±
70 1.88 0.795 0.52 ± ± ± ± ±
79 1.83 0.755 0.72 ± ± ± ± ±
99 1.82 0.740 0.24 ± ± ± ± ±
59 1.83 0.730 0.42 ± ± ± ± ±
63 1.78 0.692 0.58 ± ± ± ± ±
97 1.79 0.696 0.40 ± ± ± ± ±
58 1.78 0.687 0.74 ± ± ± ± ±
77 1.77 0.675 0.61 ± ± ± ± ±
79 1.73 0.639 0.43 ± ± ± ± ±
54 1.74 0.631 0.66 ± ± ± ± ±
69 1.72 0.621 0.53 ± ± ± ± ±
119 1.72 0.619 0.50 ± ± ± ± ±
52 1.68 0.564 0.68 ± ± ± ± ±
78 1.66 0.553 0.44 ± ± ± ± ±
84 1.65 0.537 0.50 ± ± ± ± ±
52 1.65 0.535 0.36 ± ± ± ± ±
101 1.63 0.527 0.78 ± ± ± ± ±
68 1.62 0.516 0.70 ± ± ± ± ±
51 1.61 0.513 0.74 ± ± ± ± ±
82 1.62 0.513 0.64 ± ± ± ± ±
74 1.60 0.498 0.76 ± ± ± ± ±
130 1.60 0.497 0.41 ± ± ± ± ±
115 1.58 0.479 0.69 ± ± ± ± ±
84 1.54 0.435 0.44 ± ± ± ± ±
67 1.52 0.424 0.71 ± ± ± ± ±
83 1.50 0.418 0.81 ± ± ± ± ±
73 1.50 0.415 0.56 ± ± ± ± ±
146 1.51 0.416 0.36 ± ± ± ± ±
42 1.51 0.411 0.70 ± ± ± ± ±
62 1.49 0.398 0.60 ± ± ± ± ±
75 1.47 0.388 0.45 ± ± ± ± ±
132 1.48 0.390 0.41 ± ± ± ± ±
79 1.47 0.383 0.43 ± ± ± ± ±
40 1.44 0.361 0.35 ± ± ± ± ±
141 1.45 0.357 0.58 ± ± ± ± ±
54 1.44 0.353 0.69 ± ± ± ± ±
56 1.42 0.350 0.83 ± ± ± ± ±
74 1.41 0.343 0.65 ± ± ± ± ±
74 1.39 0.323 0.41 ± ± ± ± ±
89 1.38 0.320 0.50 ± ± ± ± ±
68 1.39 0.320 0.34 ± ± ± ± ±
54 1.38 0.315 0.59 ± ± ± ± Table 4. continued.
SDSS C4 z N σ v R M cl R ≤ . R ID (km / s) (Mpc) (10 M ⊙ ) f et , raw f et , corr f [ OII ] , raw f [ OII ] , corr (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)3275 0.0699 12 570 ±
133 1.36 0.307 0.32 ± ± ± ± ±
61 1.34 0.288 0.63 ± ± ± ± ±
51 1.33 0.285 0.62 ± ± ± ± ±
72 1.32 0.283 0.45 ± ± ± ± ±
34 1.34 0.283 0.42 ± ± ± ± ±
93 1.32 0.275 0.32 ± ± ± ± ±
96 1.30 0.266 0.55 ± ± ± ± ±
66 1.31 0.267 0.85 ± ± ± ± ±
98 1.29 0.263 0.60 ± ± ± ± ±
52 1.29 0.257 0.72 ± ± ± ± ±
56 1.29 0.255 0.61 ± ± ± ± ±
71 1.27 0.247 0.63 ± ± ± ± ±
82 1.26 0.240 0.70 ± ± ± ± ±
99 1.26 0.241 0.54 ± ± ± ± ±
81 1.25 0.237 0.66 ± ± ± ± ±
67 1.25 0.238 0.59 ± ± ± ± ±
84 1.26 0.238 0.26 ± ± ± ± ±
87 1.24 0.233 0.92 ± ± ± ± ±
109 1.23 0.227 0.44 ± ± ± ± ±
88 1.23 0.225 0.68 ± ± ± ± ±
80 1.23 0.223 0.64 ± ± ± ± ±
92 1.23 0.224 0.29 ± ± ± ± ±
47 1.23 0.218 0.84 ± ± ± ± ±
50 1.22 0.216 0.50 ± ± ± ± ±
83 1.21 0.213 0.68 ± ± ± ± ±
94 1.20 0.212 0.50 ± ± ± ± ±
122 1.21 0.212 0.21 ± ± ± ± ±
110 1.20 0.209 0.68 ± ± ± ± ±
172 1.19 0.207 0.50 ± ± ± ± ±
62 1.21 0.209 0.45 ± ± ± ± ±
155 1.19 0.204 0.44 ± ± ± ± ±
71 1.19 0.203 0.71 ± ± ± ± ±
95 1.17 0.197 0.68 ± ± ± ± ±
67 1.18 0.198 0.74 ± ± ± ± ±
41 1.18 0.197 0.54 ± ± ± ± ±
49 1.15 0.185 0.61 ± ± ± ± ±
62 1.14 0.184 0.50 ± ± ± ± ±
63 1.14 0.182 0.80 ± ± ± ± ±
64 1.12 0.172 0.50 ± ± ± ± ±
180 1.11 0.166 0.37 ± ± ± ± ±
50 1.09 0.159 0.50 ± ± ± ± ±
103 1.07 0.149 0.23 ± ± ± ± ±
80 1.06 0.146 0.70 ± ± ± ± ±
90 1.07 0.145 0.26 ± ± ± ± ±
47 1.06 0.144 0.76 ± ± ± ± ±
86 1.05 0.141 0.42 ± ± ± ± ±
89 1.00 0.124 0.80 ± ± ± ± ±
87 1.01 0.124 0.26 ± ± ± ± ±
91 1.01 0.121 0.61 ± ± ± ± ±
72 0.99 0.118 0.50 ± ± ± ± ±
53 0.98 0.115 0.48 ± ± ± ± ±
41 0.97 0.110 0.53 ± ± ± ± ±
45 0.97 0.110 0.55 ± ± ± ± ±
60 0.96 0.109 0.74 ± ± ± ± ±
112 0.98 0.109 0.39 ± ± ± ± ±
183 0.95 0.102 0.79 ± ± ± ± ±
57 0.92 0.095 0.94 ± ± ± ± ±
55 0.92 0.095 0.50 ± ± ± ± ±
38 0.90 0.087 0.68 ± ± ± ± imard et al.: Evolution of the Early-Type Galaxy Fraction in Clusters 23 Table 4. continued.
SDSS C4 z N σ v R M cl R ≤ . R ID (km / s) (Mpc) (10 M ⊙ ) f et , raw f et , corr f [ OII ] , raw f [ OII ] , corr (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)1341 0.0700 10 370 ±
59 0.88 0.084 0.58 ± ± ± ± ±
109 0.85 0.075 0.61 ± ± ± ± ±
50 0.85 0.075 0.73 ± ± ± ± ±
216 0.84 0.074 0.13 ± ± ± ± ±
84 0.85 0.073 0.44 ± ± ± ± ±
54 0.82 0.066 0.77 ± ± ± ± ±
101 0.82 0.065 0.58 ± ± ± ± ±
97 0.79 0.060 0.56 ± ± ± ± ±
54 0.79 0.059 0.69 ± ± ± ± ±
50 0.79 0.059 0.82 ± ± ± ± ±
66 0.78 0.058 0.79 ± ± ± ± ±
61 0.75 0.052 0.42 ± ± ± ± ±
60 0.75 0.051 0.79 ± ± ± ± ±
97 0.74 0.048 0.29 ± ± ± ± ±
37 0.72 0.045 0.58 ± ± ± ± ±
38 0.72 0.045 0.39 ± ± ± ± ±
147 0.71 0.044 0.31 ± ± ± ± ±
40 0.72 0.044 0.50 ± ± ± ± ±
52 0.70 0.043 0.56 ± ± ± ± ±
146 0.70 0.041 0.32 ± ± ± ± ±
48 0.68 0.038 0.61 ± ± ± ± ±
85 0.66 0.033 0.39 ± ± ± ± ±
80 0.64 0.031 0.69 ± ± ± ± ±
64 0.62 0.028 0.64 ± ± ± ± ±
44 0.62 0.028 0.56 ± ± ± ± ±
56 0.61 0.027 0.50 ± ± ± ± ±
48 0.58 0.024 0.61 ± ± ± ± ±
55 0.58 0.023 0.71 ± ± ± ± ±
57 0.57 0.022 0.50 ± ± ± ± ±
107 0.56 0.021 0.29 ± ± ± ± ±
66 0.53 0.018 0.50 ± ± ± ± ±
59 0.51 0.016 0.71 ± ± ± ± ±
53 0.50 0.015 0.61 ± ± ± ± ±
31 0.50 0.015 0.69 ± ± ± ± ±
77 0.48 0.014 0.50 ± ± ± ± ±
47 0.48 0.014 0.77 ± ± ± ± ±
41 0.46 0.012 0.39 ± ± ± ± ±
57 0.34 0.005 0.71 ± ± ± ± ±
11 0.34 0.005 0.50 ± ± ± ± ±
61 0.30 0.003 0.84 ± ± ± ± ±
20 0.29 0.003 0.71 ± ± ± ± From von der Linden (2007)4