ebook img

The Morphological Content of Ten EDisCS Clusters at 0.5 < z < 0.8 PDF

0.44 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview The Morphological Content of Ten EDisCS Clusters at 0.5 < z < 0.8

Accepted forPublicationin ApJ PreprinttypesetusingLATEXstyleemulateapjv.10/09/06 THE MORPHOLOGICAL CONTENT OF TEN EDISCS CLUSTERS AT 0.5 < Z < 0.81,2 V. Desai3,4, J. J. Dalcanton4, A. Arago´n-Salamanca5, P. Jablonka6, B. Poggianti7, S. M. Gogarten4, L. Simard8, B. Milvang-Jensen9, G. Rudnick10, D. Zaritsky11, D. Clowe11, C. Halliday12, R. Pello´13, R. Saglia14, S. White15 Accepted for Publication inApJ ABSTRACT We describe Hubble Space Telescope (HST) imaging of 10 of the 20 ESO Distant Cluster Survey (EDisCS)fields. Each∼40squarearcminute fieldwasimagedinthe F814Wfilter with the Advanced Camera for Surveys Wide Field Camera. Based on these data, we present visual morphological classifications for the ∼920 sources per field that are brighter than I = 23 mag. We use these 7 auto classificationstoquantifythemorphologicalcontentof10intermediate-redshift(0.5<z <0.8)galaxy 0 0 clusterswithinthe HSTsurveyregion. TheEDisCSresults,combinedwithpreviouslypublisheddata 2 from seven higher redshift clusters, show no statistically significant evidence for evolution in the mean fractions of elliptical, S0, and late-type (Sp+Irr) galaxies in clusters over the redshift range n 0.5 < z < 1.2. In contrast, existing studies of lower redshift clusters have revealed a factor of ∼2 a increase in the typical S0 fraction between z = 0.4 and z = 0, accompanied by a commensurate J decrease in the Sp+Irr fraction and no evolution in the elliptical fraction. The EDisCS clusters 6 demonstrate that cluster morphological fractions plateau beyond z ≈ 0.4. They also exhibit a mild 2 correlationbetweenmorphologicalcontentandclustervelocitydispersion,highlightingtheimportance of careful sample selection in evaluating evolution. We discuss these findings in the context of a 1 v recently proposed scenario in which the fractions of passive (E,S0) and star-forming (Sp,Irr) galaxies 8 are determined primarily by the growth history of clusters. 8 Subject headings: galaxies: clusters: general — galaxies: formation — galaxies: evolution 7 1 0 1. INTRODUCTION Although the morphology-density relation is observa- 7 tionally well-established, it is still unclear why the inci- 0 1Based on observations made with the NASA/ESA Hubble dence of early type (E+S0) galaxies is higher in over- h/ SpaceTelescope,obtainedattheSpaceTelescopeScienceInstitute, dense regions of the universe. The options are often which is operated by the Association of Universities for Research couched in terms of “nature” versus “nurture”. These p in Astronomy, Inc., under NASA contract NAS 5-26555. These - observations are associated with proposal 9476. Support for this scenariosdescribe differentpathsto qualitativelysimilar o proposal was provided by NASA through a grant from the Space relationsbetween morphologyandenvironment,and are r TelescopeScienceInstitute. therefore difficult to disentangle. In the nature scenario, t s 2Based on observations obtained at the ESO Very Large Tele- galaxiesthatendupinhighdensityenvironmentsatlow a scope(VLT)andNewTechnologyTelescope(NTT)aspartofthe redshift are more likely to have experienced initial con- : largeprogram166.A-0162(theESODistantClusterSurvey). v 3DivisionofPhysics,MathematicsandAstronomy,320-47,Cal- ditionsleadingtoanearly-typemorphologyupon forma- i iforniaInstitute ofTechnology, PasadenaCA91125,USA tion. Inthe“nurture”scenario,allgalaxies,regardlessof X 4University of Washington Department of Astronomy, Box theirlocaldensitiesatlowredshift,haveidenticalproba- r 351580, SeattleWA98195-1580, USA bilitiesofhavingformedasearlytypes. Themorphology- a 5School of Physics and Astronomy, University of Nottingham, densityrelationisthenthe resultofsubsequentmorpho- Nottingham NG72RD,UK 6Universit´e de Gen`eve, Laboratoire d’Astrophysique de l’Ecole logical alterations as galaxies enter increasingly higher Polytechnique F´ed´erale de Lausanne (EPFL), Observatoire, CH- density environments as structure grows. Many mech- 1290Sauverny, Switzerland anisms that could produce the required morphological 7Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5,35122Padua,Italy transformations have been suggested, including merg- 8Herzberg Institute of Astrophysics, National Research Coun- ers and galaxy-galaxy interactions (Toomre & Toomre cil of Canada, 5071 West Saanich Road, Victoria, BC V9E 2E7, 1972; Icke 1985; Lavery & Henry 1988; Mihos 2004), ha- Canada rassment (Richstone 1976; Moore et al. 1998), gas strip- 9Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen, Den- ping(Gunn & Gott1972;Abadi et al.1999;Quilis et al. mark 2000), strangulation (Larson et al. 1980; Bekki et al. 10NOAO,950NorthCherryAvenue, TucsonAZ85719,USA 2002), and cluster tidal forces (Byrd & Valtonen 1990). 11Steward Observatory, University of Arizona, 933 North The observed evolution of the morphology-density re- CherryAvenue, Tucson,AZ85721, USA 12Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 lation provides an important constraint on models of its Firenze,Italy origin. Observations of the high-density regions of clus- 13Laboratoired’Astrophysiquedel’ObservatoireMidi-Pyr´en´es, ter cores show that while the elliptical fraction has not 14,AvenueEdouardBelin,F-31400Toulouse,France 14Max-Planck-Institut fu¨r extraterrestrische Physik, Giessen- evolved,theS0(Sp+Irr)fractionhasgrown(diminished) bachstrasse,D-85748GarchingbeiMu¨nchen,Germany by a factor of ∼2 over the last 5 Gyr (Dressler et al. 15Max-Planck-Institut fu¨r Astrophysik, Karl-Schwarzschild- 1997; Fasano et al. 2000). However, over the same time Strasse 1, Postfach 1317, D-85748 Garching bei Mu¨nchen, Ger- interval, no evolution is observed at lower densities many (Treu et al. 2003; Smith et al. 2005). The morphology- density relation at larger lookback times (z > 0.5) has 2 Desai et al. justbeguntobeexplored. Smith et al.(2005)usedHub- mag are presented in §4. In §5 we describe our method ble Space Telescope (HST) Wide-Field Planetary Cam- of quantifying the morphological content in each clus- era 2 (WFPC2) data for six clusters at 0.75 < z < 1.25 ter. Our results are presented in §6, and our conclusions to estimate the early-type fraction as a function of local regarding them can be found in §7. galaxy surface density. They extended their measure- Results will be presented for two sets of cosmological ments to low densities (∼1 Mpc−1) by including field parameters, to allow a direct comparison with previous galaxies in HST WFPC2 images of the cluster CL0024 work (Ω = 1, Λ = 0, H = 50 km s−1 Mpc−1) and 0 0 at z = 0.395. Comparing their results to similar studies to provide the corresponding estimate for the currently at lower redshifts (Dressler 1980b; Dressler et al. 1997; standardcosmology(Ω =0.3,Λ=0.7,H =70kms−1 0 0 Treu et al. 2003), they found that the early-type frac- Mpc−1). In the remainder of this paper we will refer to tion has increased steadily since z ∼ 1 in the densest these as the classic cosmology and the WMAP cosmol- regions(∼1000Mpc−2), hasincreasedonlysincez ∼0.5 ogy, respectively. in moderate density regions (∼100 Mpc−2), and has re- mainedconstantsincez ∼1inthelowestdensityregions 2. THEESODISTANTCLUSTERSURVEY(EDISCS) (<10Mpc−2). Postman et al.(2005)findcompatiblere- The ten clusters included in this study are a subset of sults in their study of the morphology-density relation the ESO Distant Cluster Survey, an extensive program using Advanced Camera for Surveys (ACS) HST imag- toobtainspectroscopyandmultiwavelengthphotometry ing of seven z ∼ 1 clusters. Smith et al. (2005) suggest for20galaxyclustersat0.4<z <1. Detailsofthegoals a nurture scenario to explain the observed evolution in anddatasetsassociatedwiththissurveyaredocumented the morphology-density relation. If structure growth is by White et al. (2005), but we briefly summarize below. hierarchical,thedensestregionsatanyredshiftcollapsed The clusters chosen to be part of EDisCS were theearliest. Thus,ifenvironmentalprocessescanmodify drawn from the Las Campanas Distant Cluster Survey a universal initial mix of morphologies through a trans- (LCDCS) catalog, which consists of optically-selected formationfromlatetoearlytypes,theseprocesseswould clusters identified from fluctuations in the extragalactic have had a longer time to operate in increasingly dense background light (Dalcanton 1996; Zaritsky et al. 1997; regions. In addition, the efficiency of such transforma- Gonzalez et al. 2001). The redshifts of the LCDCS clus- tions could be density-dependent. tercandidateswereinitiallyestimatedfromtheapparent To discover whether environmental processes, rather magnitude of the brightest cluster galaxy (BCG). The than initial conditions, drive the differential evolutionof candidates were divided into an intermediate-redshift the morphology-density relation, we must first demon- group (z ∼ 0.5) and a high-redshift group (z ∼ 0.8). strate that nature scenarioscannot reproducethe obser- Shallow two-color imaging was obtained at the Very vations. If we findthat environmentalfactorsareindeed Large Telescope (VLT) for the 30 brightest candidates important, it is criticalto identify the responsible mech- in each group. A sample of 20 EDisCS clusters was anisms. The current data are insufficient to discrimi- selected based on the presence of a red sequence and nate among the alternative models. For this, we require a BCG of the appropriate magnitude (Gonzalez et al. a full mapping between galaxy properties and environ- 2001). These twenty clusters were then imaged more ment. Towards this goal, we present the morphological deeply at the VLT in BVI (45 minutes) for the z ∼ 0.5 fractionsintenrichclustersat0.5<z <0.8,drawnfrom candidates and in VRI (2 hours) for the z ∼ 0.8 can- theESODistantClusterSurvey16(EDisCS;White et al. didates. These data, covering a 6.5′ × 6.5′ region in 2005). Because strong evolution has been observed in the field of each cluster, are presented in White et al. high-density regions, clusters are a critical environment (2005),andformthebasisoftheweak-shearanalysispre- to probe. The redshift range of our study bridges a gap sented by Clowe et al. (2006). Additional near-infrared between those of Dressler (1980a), Fasano et al. (2000), imagingwasobtainedwithSOFIatthe NewTechnology and Dressler et al. (1997) at low redshift and those of Telescope (NTT) in K for the z ∼ 0.5 candidates and s Smith et al. (2005) and Postman et al. (2005) at high in JK for the z ∼ 0.8 candidates (Arago´n-Salamanca s redshift. The EDisCS clusters represent a significant et al., in preparation). The near-infraredimaging covers increase in the number of well-studied clusters at high approximately6′ × 4.2′ inthe fields ofthe intermediate- redshift. Because the mean age of clusters decreases redshift clusters and 5.4′ × 4.2′ in the fields of the high- with increasing redshift, the scatter in cluster properties redshift clusters. The optical and near-infrared imaging may also increase with redshift. Thus, large samples are wereusedtomeasurephotometricredshifts,asdescribed necessarytodisentangleevolutionfromcluster-to-cluster in Pell´o et al. (in preparation). variations. Finally, we measure morphologiesfrom high- An initial phase of spectroscopyconsisted ofrelatively quality HST ACS images, allowing us to distinguish S0s short exposures of a single slit mask per cluster. One from other early types. As described above, the S0 pop- cluster with an estimated redshift of z ∼ 0.8 was re- ulation appears to have increased at the expense of the vealed as a superposition of weak groups, and was re- Sp+Irr population since z = 0.5. Thus, tracing the S0 jected from the sample. These spectra revealed that population to high redshifts may place important con- the remaining 19 clusters have redshifts in the range straints on their formation. 0.4 . z . 1. Deeper spectroscopic exposures (∼2 hours This paper is organized as follows. In §2 we describe each for galaxies with 18.6 ≤ I(r = 1′′) ≤ 22 mag in theEDisCSsample. In§3wedescribetheHSTACSdata intermediate-redshift clusters; ∼4 hours each for galax- used to measure morphologies. Our classification proce- ies with 19.5 ≤ I(r = 1′′) ≤ 23 mag in high-redshift dure and Hubble types for all galaxies with Iauto < 23 clusters) of 3-5 masks per field were then taken, and are of sufficient quality to obtain information about the 16http://www.mpa-garching.mpg.de/galform/ediscs/index.shtml stellar populations and internal dynamics of the target EDisCS MorphologicalContent 3 galaxies. The spectroscopic observations are described image were then computed using cross-correlation. in Halliday et al. (2004) and Milvang-Jensen et al. (in Combination of the FLT images using the resulting preparation). shifts was accomplished using MultiDrizzle, a Python The ground-based EDisCS imaging and spectroscopy codewrittenbyAntonKoekemoertorununderPyRAF, go a long way towards characterizing both the clusters the Python-based interface to the Image Reduction and themselves and the galaxies within them. At the high Analysis Facility (IRAF). MultiDrizzle automatically re- redshifts of the EDisCS sample, however, only HST can moves cosmic rays and combines dithered images using providethespatialresolutionnecessarytoproviderobust PyDrizzle,whichhasbeendevelopedbytheScienceSoft- morphologies. Motivated by the issues discussed in §1, ware Branch at the Space Telescope Science Institute. we obtained HST imaging for the ten highest redshift For each of the 32 FLT images per cluster, MultiDrizzle EDisCS clusters, listed in Table 1 along with their basic included negative bad pixels in the data quality array, physical parameters. Cluster 9 (cl1232-1250)was drawn subtracted the sky, and separately drizzled and undis- from the EDisCS intermediate-redshift sample, and the tortedeachimage. Next,itcreatedamedianimagefrom remainder were drawn from the high-redshift sample. these separately drizzled images using shifts computed fromtheheadersalongwiththeuser-suppliedrefinement 3. HSTACSDATA shifts described above. This median image, relatively The HST observations were designed to coincide as free of cosmic rays, was compared to the input images closelyaspossiblewiththecoverageoftheground-based to identify cosmic rays. The median image was then re- opticalimaging andspectroscopy,within guide star con- distorted to create cosmic ray masks. These masks were straints. The ground-based optical data cover a 6.5′ × used in the final image combination step using drizzle 6.5′ region around each cluster, with the cluster center and the lanczos3 kernel, which provided optimal noise displaced by 1′ from the center of the region. For ref- properties. erence, the ACS Wide Field Camera has a field of view of roughly 3.5′ × 3.5′. Balancing scientific motives for 4. GALAXYMORPHOLOGIES going deep over the entire field against a limited num- WevisuallyclassifiedallgalaxiesbrighterthanI = auto ber of available orbits, we tiled each 6.5′ × 6.5′ field 23 mag. Here I is the SExtractor(Bertin & Arnouts auto in four pointings, with one additional deep pointing on 1996) AUTO magnitude measured on the I-band VLT the cluster center (taken as the location of the BCG). images, and is an estimate of the total magnitude of a The resulting exposure time per pixel is 2,040 seconds, galaxy, in the Vega system. This magnitude does not except for the central 3.5′ × 3.5′, which has an expo- include anaperture correction,norhas it beencorrected sure time per pixel of 10,200 seconds. The deep central forGalacticextinction. Thelimitwassetbothto ensure pointing probesto lowersurfacebrightness,faintermag- robust classifications and to provide a tractable sample nitudes, and larger galactic radii in the region of the of ∼9200 galaxies. cluster containing the most galaxies. The centers of the Our classifications are most useful if they conform structures ultimately chosen for spectroscopic followup to systems adopted by previous studies. For this rea- inthe cl1103-1245bandcl1227-1138fields aresomewhat son, each classifier trained on the HST WFPC2 images offset from those anticipated at the time the HST ob- and visual morphological catalogs of the 0.3 < z < 0.5 servations were taken. All exposures were taken under MORPHS clusters, using the same procedure described LOWSKYconditionstomaximizeoursurfacebrightness inSmail et al.(1997). Foruniformity,eachclassifierused sensitivity. the same IRAF script to examine and classify EDisCS The ACS calibration pipeline, CALACS v4.3 (6- galaxies. Thisscriptdisplaystwoside-by-sideversionsof June-2003) debiased, dark-subtracted, and flat-fielded a 200× 200squarepixelcutout centeredoneachgalaxy our ACS images “on the fly” when they were requested meeting the magnitude limit described above. One ver- from the HST archives. Known bad pixels and satu- sion is on a log scale between -0.1 – 2 DN/s, while the rateddata werealso flaggedinaccompanyingdata qual- otherisonalogscalebetween-0.1–25DN/s. Together, ity (DQ) arrays. Approximate World Coordinate Sys- these displays allowed classifiers to inspect the galaxies tem(WCS)headerswereprovided. Theresultingimages fromtheir high surface brightness coresto their low sur- werereturnedby the archivewiththe FLT suffix. Inad- face brightness outer features. In general,we found that dition, images that were cosmic-ray split were combined the depth and quality of the EDisCS ACS data were and returned with a CRJ suffix. similartoorbetter thantheMORPHSWFPC2datafor To produce a mosaic, we required precise offsets be- lower redshift clusters. tweenthe 32 FLT images retrievedfor eachcluster. The Classifications were performed by 5 of the authors approximateWCS headersprovidedwere insufficient for (AAS, JJD, VD, PJ, BP). Each classified the galaxies this purpose. We found that the offsets between cosmic- in3–6clusters. Foragivengalaxy,the finalHubble type ray split images were negligible. We therefore computed was based on the classifications of two or more of the the necessary shifts between pointings using the higher authors. First, the Revised Hubble Type was translated signal-to-noiseCRJimages. Theshiftscouldnotbecom- intoaT-typeaccordingtotheschemepresentedinTable puted using the retrieved CRJ files because they had 2. Classifications appended by one question mark were not been undistorted. We therefore drizzled each CRJ givenhalfweight;thoseappendedbytwoquestionmarks image separately, using the approximate WCS headers were given one-quarter weight. If a classifier specified provided. Because eachundistortedCRJ image overlaps twotypes separatedby aslash,the firstwasgiventhree- the central pointing, one of the images centered on the quarters weight and the second was given one-quarter cluster center was chosen as the reference image. Shifts weight. All T types submitted for a given galaxy were between the reference image and each undistorted CRJ then ranked by their frequency. The final Hubble Type 4 Desai et al. was then chosen from the highest ranked T types. In law (i.e. only bulge, no disk). The differential and frac- general, the highest ranked T type was chosen. If there tional cumulative distribution functions for these three were two equally-ranked classifications, one was chosen parameters are plotted in Figure 1. At a statistically randomly. If there were three or more equally-ranked significant level, S0s display larger ellipticity distribu- classifications, the differences in T types were computed tion widths than ellipticals. The presence of a disk in a between them. If there were no gaps smaller than three galaxy will result in a wider range of ellipticities, since T types, the median value was chosen. If there was one the ellipticity of the galaxy changes as the disk begins gap smaller than three T types, one of the two similar to dominate the surface brightness profile. Likewise, the classifications was chosen at random. If there was more B/T fractions for visually-classified ellipticals appear to than one gap smaller than three T types, the median beskewedtohighervaluescomparedtoS0s,asexpected. value of all classifications was chosen. This is confirmed by a two-sided Kolmogorov-Smirnov Themainresultsofthisworkaresensitivetohowaccu- test. Finally, a higher fraction of ellipticals have small ratelygalaxiescanbeplacedintothebroadcategoriesof valuesofχ2,indicatingabetterfittoapurebulgemodel. elliptical (E), S0, and late (Sp+Irr). Although accuracy However, the differences in the χ2 distribution are not in this case is difficult to quantify, we can test how con- significant. The overall results of these tests are that sistently galaxies are placed in the same bin by different visually-classified S0s have quantitatively different sur- classifiers. Using ∼900 galaxies down to I =23 mag facebrightnessprofilesthanvisually-classifiedellipticals. auto in cl1216-1201,all ofwhich were assigneda Hubble type Althoughinsomeindividualcasesitisdifficultforvisual by all classifiers,we computedthe rawfractions for each classifiers to differentiate between Es and S0s, it is clear classifer, uncorrected for the presence of foregroundand that,inastatisticalsense,objectsvisuallyclassifiedasEs background galaxies. The root mean variance for any and S0s form two distinct populations with objectively morphologicalfractionis .0.10,which is comparableto measurablephysicaldifferences. Moresophisticatedtwo- the error computed from Poissonstatistics. dimensionalprofilefitswillbepresentedinSimardetal., There is some controversy surrounding the ability of in preparation. morphologists to distinguish between E and S0 galaxies, Morphological catalogs for the ten highest-redshift especially at high redshifts. For this reason, some pre- EDisCS clusters are available in the on-line version of vious investigators chose to lump E and S0 galaxies to- this paper, with columns described in Table 2. getherintoanearly-typeclass. Foreasycomparisonwith 5. ANALYSIS these works, in the following we have plotted and tabu- The goal of this paper is to quantify the morpholog- lated our results for early types as well as for E and S0 ical content of the EDisCS clusters. In particular, we galaxiesseparately. However,because the S0 population will discuss the overall fractions of E, S0, and late-type isstronglyevolvinginclustersatz <0.5,itisimportant galaxies in our clusters and compare them to the frac- to track this population at higher redshifts. Comparing tionsfoundinclustersspanningarangeofredshifts. Fol- the EDisCS ACS data to the MORPHS WFPC2 data lowing Dressler et al. (1997), we do not include galaxies givesthe impressionthatwecandistinguishS0sfromel- that were unclassifiable (types -6 and 66 in Table 2) in lipticals with the same accuracy for the EDisCS sample our analysis. To facilitate a fair comparison with other aswasobtainedwith the MORPHSsample. Totest this samples,wecomputethemorphologicalfractionsascon- impression,wehaveperformedaquantitativeanalysisof sistentlyaspossiblewithpreviouswork. Inthefollowing thelightdistributionsofEandS0galaxieswithincl1216- subsections,wedescribethekeyelementsofouranalysis. 1201,whichisourhighest-redshiftclusterandwhichhas been visually examined by all classifiers. Specifically, 5.1. Magnitude Range we used the ELLIPSE task in the Space Telescope Sci- Earlytypespreferentiallyoccupythe brightendofthe ence Data Analysis System (STSDAS) package of IRAF galaxy luminosity function, while late types dominate to measure the surface brightness of a series of isopho- the faint end (e.g. Blanton et al. 2001; Goto et al. 2002; tal ellipses in each galaxy as a function of semi-major Zucca et al. 2006). Thus, morphological fractions de- axis. For each step, the semi-major axis is increased by pend upon the range of absolute magnitudes sampled. a factor of 1.1. The ELLIPSE task was allowed to vary Morphological fractions in the MORPHS clusters were the ellipticity and position angle of the ellipses within a determined using apparent magnitude cuts in I de- given galaxy. For each surface brightness profile, we fit 814 signedto correspondtoa V-bandabsolutemagnitude of a bulge+disk model of the form M = −20 mag (Dressler et al. 1997), but in actuality V corresponding to M =−19 mag due to a transcription I (r)=I exp(−7.67[(r/r )1/4−1]) (1) V bulge e e error(Fasano et al.2000). However,Fasano et al.(2000) re-analyzedthe MORPHS data with the intended limit- and ing absolute magnitude and additionally analyzed nine clusters with 0.1 < z < 0.25 in the same manner, pro- I (r)=I exp(−r/r ), (2) disk 0 d viding an ideal comparison sample for the EDisCS clus- where the total surface brightness is given by I + ters. We therefore adopt an absolute magnitude limit bulge Idisk. We considered three parameters for which E and of MV = −20 mag. For a given cluster, the limiting apparent magnitude in the I-band is then given by S0 galaxies should display different distributions: 1) the width of the ellipticity distribution in a given galaxy, R, defined as the interquartile range of all values of the Iauto,lim =MV+5log10(dL,pc)−5−(MV−MI)+kI, (3) ellipticity for a given galaxy; 2) the bulge-to-total frac- where d is the luminosity distance of the cluster in L,pc tion B/T; and 3) χ2 goodness-of-fitfrom fitting the r1/4 parsecs (calculated in either the classic or WMAP cos- EDisCS MorphologicalContent 5 mology, as indicated), M −M is the rest-frame color, As described in §2, the EDisCS program includes an V I and k is the I-band k-correction. We adopt the rest- extensive spectroscopic survey. For each cluster, red- I frame color and Cousins I-band k-corrections for an el- shifts were obtained for 6-86 galaxies which both meet liptical galaxy, as presented in Poggianti (1997). Equa- our apparent magnitude limit and lie within 0.6×R 200 tion 3 results in values of I ranging from 21.3 to (WMAP cosmology). We use this spectroscopic sample auto,lim 23.3 mag, the faintest values being slightly fainter than to constrain the morphological fractions in each cluster the limitofourvisualclassifications. Giventhe errorsin in two ways. First, we calculate hard limits on the mor- the k-corrections relative to these differences, we adopt phologicalfractions using both the spectroscopic sample I = 23 mag in these instances. This occurs for and the purely photometric sample, consisting of galax- auto,lim cl1216-1201 when using the WMAP cosmology and for ies that were either not targeted for spectroscopy, or for cl1054-1245, cl1216-1201, and cl1354-1230 when using whichspectroscopyfailedto yielda redshift. This calcu- the classic cosmology. lation does not require that the spectroscopic sample be complete. Second, we estimate the morphological frac- 5.2. Aperture tions using only the spectroscopic sample after applying small completeness corrections. Clusters exhibit a radial gradient in their mor- The upper and lower hard limits on the fraction of phological fractions; the centers of clusters con- galaxies of type i are given by: tain a larger fraction of ellipticals than the out- skirts (e.g. Melnick & Sargent 1977; Goto et al. 2004; Nm(i) Thomas & Katgert 2006). Thus, in comparing the over- f (i)= s (4) all morphological fractions across clusters, one must min Nsm+Npu−Npu(i) chooseaconsistentaperture. TheMORPHSprojecthas and set the standard for the aperture within which to com- pute morphological fractions. Most studies of the mor- Nm(i)+Nu(i) phological fractions of galaxy clusters at other redshifts f (i)= s p , (5) max Nm+Nu(i) have used a similarly-sized aperture. So that we may s p assess the level of evolution between the morphological where the subscripts s and p refer to the spectroscopic studies presented in the literature and those conducted sample and the photometric sample, respectively; and withEDisCSdata,weadheretothisprecedentandadopt the superscripts m and u indicate cluster members and acircularapertureofradius600kpc(classiccosmology). galaxies at unknown (spectroscopic) redshift, respec- Since the radial density profiles of clusters vary, the tively. Eachquantity is a function of galaxy type i. The average galaxy density within a fixed metric aperture fractionofgalaxiesoftypeiisminimizedwhenallofthe will also vary. We therefore also calculated the mor- other types that do not have redshifts are members and phological fractions within an aperture that scales with allofthegalaxiesoftypeithatdonothaveredshiftsare R200,theradiuswithinwhichtheaveragemassdensityis nonmembers. The fraction of galaxies of type i is max- equalto 200times the criticaldensity. The derivedR200 imized when all of the galaxies of other types without values were computed using Equation 8 in Finn et al. redshifts are nonmembers and all of the galaxies of type (2005), and are listed in Table 1 for both the classic and iwithoutredshiftsaremembers. Theresultsofapplying WMAP cosmologies. Unfortunately, ourimaging datais equations 4 and 5 are shown as the grey shaded regions complete out to R200 for only cl1040-1155, cl1054-1146, in Figure 2. cl1054-1245,and cl1354-1230. We therefore use a radius To directly estimate the morphological fractions from of 0.6 × R200, which keeps the analysis area within the the spectroscopic data, we must correct for incomplete- imagingregionforallclustersexceptforcl1227-1138and ness in the spectroscopic sample. As described in the cl1232-1250. The fractionof the analysisregionwhichis Appendix of Poggiantiet al. (2006), the completeness not included in these two clusters is very small, and the of the EDisCS spectroscopy is a weak function of both resulting effect on the morphological fractions is likely apparent I-band magnitude and clustercentric distance. minimal. Using the cluster-by-cluster magnitude and geometric completeness weights W and W calculated in that 5.3. Background Subtraction mag geo work, we estimate the spectroscopic morphological frac- Galaxies within the magnitude range and distance tions f (i) as: spec from the cluster center described in §5.1 and §5.2 lie at a variety of redshifts. We wish to determine the pro- Nsm(i) portions ofE, S0, andlate-type galaxiesof cluster mem- Nspec(i)= X Wmag(j)−1Wgeo(j)−1 (6) bers only. Because the morphologicalmix of field galax- j=1 ies differs substantially from that of cluster populations, morphologicalfractionscomputedwithoutregardtofield Nsm contamination will underestimate the fraction of early Nspec(tot)=XWmag(j)−1Wgeo(j)−1. (7) types andoverestimatethe fractionof late types in clus- j=1 ters. We use four methods, described in detail below, to account for the presence of cluster non-members, check- fspec(i)=Nspec(i)/Nspec(tot) (8) ing for consistency among the resulting morphological The probability of measuring a given morphological fractions. fractionistheproductofthePoissonprobabilityofmea- suring the total number of cluster members times the 5.3.1. Spectroscopic Redshifts binomial probability of measuring the observed number 6 Desai et al. of a given morphology. We use the approximation pre- of observed cluster members of type i (N (i)), and photoz sentedinEquations21and26ofGehrels(1986)toderive finally, the correctedfraction of cluster members of type 1-σ errorestimates basedonthis premise,usingN (i) i (f (i)): spec photoz and N (tot) as inputs. The results of applying Equa- spec tion 8 are shown as hollow squares in Figure 2. N (i)=N (i)+N (i)−N (i) (9) photoz obs m c As described in §2, the faint magnitude limit of the EDisCS spectroscopic survey is I(r = 1′′) = 22 mag for the intermediate-redshift clusters and I(r = 1′′) = 23 N (tot)=N (E)+N (S0)+N (Sp+Irr) photoz photoz photoz photoz mag for the high-redshift clusters. In general, I(r =1′′) (10) is fainter than I . For six of our clusters, all of the auto galaxies meeting the Iauto magnitude and aperture re- fphotoz(i)=Nphotoz(i)/Nphotoz(tot) (11) quirements of our analysis have values of I(r = 1′′) As in 5.3.1, errors were computed using the Gehrels brighter than the spectroscopic survey limit. However, (1986) approximation using N (i) and N (tot). for cl1054-1245, cl1216-1201, cl1232-1250, and cl1354- photoz photoz The morphological fractions computed using Equation 1230, 10-20% are fainter than the spectroscopic flux 11 are shown as hollow circles in Figure 2. limit. Most(&80%)ofthesearelate-typegalaxies. Thus, As described in §2, the area of each cluster with near- it is possible that our direct spectroscopic estimates are infrared imaging is somewhat smaller than that imaged biased slightly towards low late-type fractions. In Fig- intheoptical. Intheclassiccosmology,ananalysisradius ure 2, we compare the morphological fractions derived of 600 kpc extends beyond the near-infraredimaging for from different methods, and find that for cl1054-1245, cl1227-1138. As a result, 19 out of 49 galaxies meeting cl1216-1201, and cl1232-1250, the spectroscopic method themagnitudeandaperturerequirementsforinclusionin doesyieldlowerlate-typefractionsthanthephotometric the morphological analysis lack near-infrared data. Us- redshift or statistical background subtraction methods. ingtheWMAPcosmologyandananalysisradiusof0.6× However, all methods produce late-type fractions that R , a similar problem occurs for both cl1227-1138 (21 are consistent with one another, within the errors. For 200 out of 50 galaxies lack near-infrared data) and cl1232- cl1354-1230, the spectroscopic method results in a late- 1250 (14 out of 222 galaxies lack near-infrared data). type fraction between the photometric redshift method Becausethe P(z)distributions for photometric redshifts and the statistical background method. computed without near-infrared data are broad, they 5.3.2. Photometric Redshifts have low P values, and are somewhat more likely to clust berejectedthangalaxieswithphotometricredshiftscom- Our optical and near-infrared imaging allows the puted using the full filter set. Figure 2 shows that this derivationofphotometric redshifts,asdescribedinPello´ effectdoes notappearto havesystematicallyskewedthe et al. (in preparation). Briefly, two estimates of the morphological fractions for cl1227-1138 and cl1232-1250 redshift probability distribution (P(z)) were computed compared to the other methods employed. for each galaxy. Two independent codes were employed, onedescribedinRudnick et al.(2001)andRudnick et al. 5.3.3. Statistical Background Subtraction (2003) and Hyperz, describedin Bolzonella et al.(2000). An estimate of P(z) can be integrated over a suitable In the absence of spectroscopic or photometric red- interval ∆z (in this case ±0.1) around the cluster red- shift information for each galaxy, we can still estimate shift to obtain the probability P that the galaxy is a themorphologicalfractionsbystatisticallycorrectingthe clust member of the cluster. We used our large spectroscopic number of observed galaxies of a given type to account sampletodetermineP ,theminimumvalueofP for sources that lie in the field: thresh clust requiredfor a galaxyto be considereda cluster member. Membership information derived from each of our two Nfield =ΣfieldA, (12) estimates of P(z) was then combined to determine clus- ter membership. That is, both estimates were required N (i)=N (i)−N P(i) (13) stat obs field to be consistent with cluster membership in order for a galaxy to be considered a cluster member. Weusedthesubset(vastmajority)ofourspectroscopic Nstat(tot)=Nobs(tot)−Nfield (14) sample with both optical and near-infrared imaging to estimate a) the fraction of spectroscopically-confirmed f (i)=N (i)/N (tot). (15) cluster members excluded by our photometric redshifts stat stat stat as a function of morphology and b) the fraction of spec- Here, A is the area of the aperture described in §5.2; troscopicnon-membersthatarephotometricmembersas N (tot) is the total number of galaxies meeting the obs a function of morphology. Since P was calibrated magnitude criterion (§5.1) within that aperture; N (i) thresh obs to include as many cluster members as possible rather is the number of these galaxies which have morphology than to exclude all non-members, the former fraction i; N is the number of the observed galaxies that field tends to be significantly smaller than the latter. Given are expected to be field members; and N (tot) and stat these fractions, we calculated N (i), the expected num- N (i)arethebackground-subtractednumberofgalax- m stat berofcluster membersoftype ithatweremissedby the ies in the aperture and the number of type i, respec- photometric redshifts; and N (i), the expected number tively. The surface density of field galaxies, Σ , is c field of non-members among galaxies of type i that contami- determined by integrating the I-band differential num- nate the photometric redshift member sample. We then ber counts in Table 1 of Postman et al. (1998) down to used this information to compute the corrected number theredshift-dependentI-bandapparentmagnitudelimit EDisCS MorphologicalContent 7 adopted for each cluster, as described in §5.1. We com- ∼2σ level for the S0 and Sp+Irr fractions. We find no puted P(i), the fraction of field galaxies of each mor- correlation between cluster redshift and cluster number phological type, using data from the Medium Deep Sur- that could explain this trend. However,we find that the vey(MDS;Griffiths et al.1994),aHubbleTelescopeKey cluster velocity dispersion is correlated with the cluster Projectthatcatalogedthemorphologiesofintermediate- number, also at the ∼2σ level. Since there is no reason redshift field galaxies down to I ∼22 mag. In partic- for any physicalpropertyof the cluster to correlatewith 814 ular,we used the classificationsof RichardEllis listed in clusternumber,these correlationshintthatthe morpho- Table1ofAbraham et al.(1996). Themagnitudelimits, logicalfractionscorrelatewithclustervelocitydispersion, I , of our morphological analysis vary from clus- a possibility we revisit in §6.2. auto,lim ter to cluster and with cosmology, but generally range 6.1. Evolution of cluster morphological content from 21.3 to 23 mag. The values of P(i) are a function ofIauto,lim,decreasingtofaintmagnitudesforearly-type To assess the degree of evolution in the morphologi- galaxiesandincreasingtofaintmagnitudesforlatetypes. cal content of rich clusters, we searched the literature For those clusters with Iauto,lim >22 mag, we adopt the for similaranalyses ofclusters spanning a broadredshift values of P(i) down to I814 = 22 mag. Based on the range. behavior of P(i) versus I814, this procedure likely over- Thefirstcomparisonsampleconsistsofz <0.5clusters estimates P(E), P(S0), and P(Sp) and underestimates with morphologicalfractions originally measured by dif- P(Sp+Irr) and P(Irr). These trends translate into un- ferentgroupsusing a varietyof methods, butreanalyzed derestimatesoftheE,S0,andSpmorphologicalfractions in a uniform manner by Fasano et al. (2000), hereafter andoverestimatesofthe Sp+IrrandIrrfractions. These F00. The F00 sample includes 55 low redshift clusters mis-estimateswillbemostsevereforcl1054-1245,cl1216- (Dressler 1980a; Dressler et al. 1997); the ten MORPHS 1201, and cl1354-1230,which have the faintest values of clusters at 0.37 < z < 0.5 (Dressler et al. 1997); three Iauto,lim. The morphological fractions computed using clusters at z ∼ 0.3 plus A2218 and A1689 at z = 0.18 statistical background subtraction and Equation 15 are (Couch et al. 1998); and nine clusters at 0.1 < z < 0.25 shown as hollow triangles in Figure 2. Examination of (Fasano et al. 2000). Using this large sample of uni- this figure does not reveal any obvious biases with re- formly analyzed clusters at z <0.5, Fasano et al. (2000) specttoothermethods thatcouldbe attributabletoour confirmed the MORPHs finding that the S0 fraction in P(i) estimates. clusters has doubled since z ∼ 0.5, while the late-type As in §5.3.1 and §5.3.2, errors were computed us- fraction has decreased by a similar factor. ing the Gehrels (1986) approximation and Nstat(i) and We also compare our results with that of Nstat(tot). Postman et al. (2005), who we will refer to as P05. They computed the morphological fractions within the 6. RESULTS virial radii of 7 clusters at 0.8 < z < 1.27. This is the We have estimated the morphological fractions of 10 only large study of clusters at these redshifts in which EDisCS clusters at 0.5 < z < 0.8 using four differ- S0 galaxies are separately classified. As previously ent background-subtraction techniques (absolute limits discussed, the S0 population is an important one to from spectroscopic redshifts, direct estimates from spec- track at high redshift. troscopicredshifts,photometric redshifts, andstatistical In Figure 3, we show the evolution at z < 0.8 of subtraction),within aperturesof different radii(600 kpc the morphological fractions within rich clusters, using and 0.6×R ), and for two different cosmologies (the EDisCS data in conjunction with the F00 sample de- 200 classic cosmology: Ω = 1, Λ = 0, H = 50 km s−1 scribedabove. Allpointswerecomputedwithin600kpc, 0 0 Mpc−1; and the WMAP cosmology: Ω = 0.3, Λ = 0.7, usingthe classiccosmology. We do notinclude measure- 0 H = 70 km s−1 Mpc−1 ). Figure 2 shows how the dif- ments made onthe Postmansample in this plotbecause 0 ferent methods compare for an aperture of 0.6 ×R they were computed within R using the WMAP cos- 200 200 in the WMAP cosmology. The direct estimate methods mology. As discussed previously, the F00 sample shows are consistent with the absolute limits derived from our a systematic decrease in the S0 fraction from 55% at spectroscopy,andwithoneanother. Inaddition,thereis z = 0 to 20% at z ∼ 0.5. Where the redshift range no systematic trend for one estimate to produce higher of the EDisCS clusters overlaps with that of F00, the or lower fractions than any other. For each cluster we morphological fractions derived from the two samples therefore adopt a single method: direct estimate from are generally consistent. However, the longer redshift spectroscopy, photometric redshifts, or statistical back- baseline affordedby the additionof the EDisCS data re- ground subtraction. The adopted method, which differs veals that the S0 fraction is actually flat over the range fromclustertocluster,ischoseninthefollowingway. For 0.4<z <0.8. This could mean that z ∼0.4 is a special a given cluster, median E, S0, and Sp+Irr fractions are epoch after which the S0 fraction in cluster cores begins selected from among the estimates of all three methods. to grow. The coincidence of this special epoch with the Themethodselectedmostofteninthisprocessistheone redshift where the two samples overlap raises the ques- adoptedfor the cluster. The E,S0,andSp+Irrfractions tion of whether the samples were analyzed in different estimated using the chosen method for each cluster are ways. However, Section 5 describes our extensive at- shown in Tables 3 and 4, and are the quantities plotted tempts tocontrolsystematicsinthe EDisCSsampleand in Figures 3 through 5. to conform to the analysis presented by F00. Alterna- The morphological fractions in Figure 2 appear to be tively, it is possible that we are missing some S0-rich correlated with cluster number, which is determined by clusters at z ∼ 0.4 which would reveal that the growth right ascension. Spearman and Kendall rank correlation of the S0 fraction in clusters at z < 0.4 is slower than testsshowthatmarginalcorrelationsaredetectedatthe suggested by the present data. Indeed, the scatter in 8 Desai et al. the morphologicalfractions at z ∼0.4 is smaller than at within rich clusters at 0.4 < z < 1.25. Figures 3 and 4 either higher or lower redshifts. indicate that the scatter in these fractions is large. How The F00+EDisCS data are consistent with either a muchofthis scatteris duetoacorrelationbetweenmor- bend in the morphological fractions at z ∼ 0.4 or a phologicalfractionsandclustermass? Suchacorrelation smooth continuation of the trends observed at z < 0.5, may be expected in either nature or nurture scenarios. butwithaflatteroverallslopethansuggestedbytheF00 In the former it is due to the fact that the most massive data alone. More observations at z ∼ 0.4 and z > 0.8 clusters collapsed at earlier times. In the latter it could are required to distinguish between these two possibili- be due to a higher efficiency of morphological transfor- ties. As discussed above, P05 have analyzed the mor- mations in more massive clusters. phological fractions, including the S0 fraction, in clus- P05 found that the E, S0, and E+S0 fractions within ters at z > 0.8. In Figure 4, we show the morphological R of seven z ∼ 1 clusters increase with increasing 200 fractions of the EDisCS and P05 clusters as functions of bolometric cluster X-ray luminosity, although the corre- redshift. Allestimates weremadeusing the WMAP cos- lations are significant only at the .3σ level. However, mology,but the EDisCSfractionswerecomputedwithin they find no correlation between the E+S0 fraction and 0.6×R (see §5.2), while the P05 fractions were com- the X-raytemperature orthe cluster velocitydispersion, 200 puted within R . P05 find that the E and E+S0 frac- perhaps because of small number statistics. In Figure 5 200 tions decrease out to 0.6×R , while the S0 fraction we plot the morphological fractions of the EDisCS and 200 staysroughlyconstantandtheSp+Irrfractionincreases. P05 samples against cluster velocity dispersion. From Between 0.6×R and R , all the fractions are flat. thisfigureitisapparentthatclusterswithlargervelocity 200 200 According to Table 4 of P05, the E, S0, early, and late dispersions harbor a higher fraction of early type galax- type fractions computed within R are factors of ap- ies and fewer late type galaxies. Spearman and Kendall 200 proximately0.79,0.96,0.85,and1.15timesthefractions rankcorrelationtestsshow thatthese visualimpressions computed within 0.6 × R . However, no corrections are statistically significant at the .3σ level. 200 have been made to the points shown in Figure 4. Whileitistemptingtointerpretthiscorrelationasone Figure 4 shows that the strong evolution in morpho- betweenmorphologicalcontentandclustermass,the ve- logical fractions seen at z < 0.4 does not continue at locitydispersionofaclusterisdirectlyrelatedtoitsmass z > 0.4. Indeed, the E+S0 (Sp+Irr) fraction appears only if the cluster is virialized. The velocity dispersion to be larger (smaller) in the higher-redshift P05 sample mayoverpredictmassiftheclusterisexperiencingsignif- comparedto the EDisCSclusters. However,aSpearman icant merging events. The degree of substructure in five rank correlation analysis shows that there is no statisti- of the clusters in this work (cl1040-1155, cl1054-1146, cally significant evidence for any evolution over the en- cl1054-1245, cl1216-1201, and cl1232-1250) has been tire redshift interval 0.4<z <1.25, or in the individual studiedindetailbyHalliday et al.(2004)usingDressler- EDisCS and P05 samples. Additional clusters in this Shectman tests (Dressler & Shectman 1988). They de- redshiftrangearenecessarytorevealthepresenceofany tect substructure in cl1232-1250 and cl1216-1201 with weak correlation. more than 95% confidence. For cl1040-1155 and cl1054- How does the selection requirement for the EDisCS 1245, not enough spectra were available to provide firm clusters to exhibit a red sequence affect the interpreta- evidence for substructure. No evidence for substructure tion of Figures 3 and 4? If the red sequence takes time was found for cl1054-1146. They note that the two clus- to build up, it may be expected that this requirement tersshowingclearevidenceofsubstructurealsohavethe selects for clusters that are also dynamically evolved. In largest velocity dispersions in the EDisCS sample, and fact,theEDisCSsampleincludesclustersthatareclearly caution against using the velocity dispersions for these non-spherical, as well as clusters that display significant twoclustersasaproxyformass. Furtheranalysisonthe substructure (see §6.2). The red sequence requirement remaining EDisCS clusters is ongoing (Milvang-Jensen may also be expected to select for clusters with high et al., in preparation). E and/or S0 fractions. Figure 3 demonstrates that the What is the situation for the z < 0.5 sample, EDisCS clusters do not contain large E fractions com- where evolution is observed? Velocity dispersions paredtoeithertheoptically-selectedF00sampleatlower were available in the literature for 14 of the clusters redshiftsorthe(primarily)X-ray-selectedP05sampleat in the F00 sample (Couch & Sharples 1987; Gudehus higherredshifts. Thus,eithersuchabiasisweak,oritis 1989; Girardi & Mezzetti 2001; De Propris et al. 2002; sharedbythecomparisonsamples. Giventhecorrelation Bettoni et al. 2006). Figure 6 shows the morphological between galaxycolor andmorphology,any clusters lack- fractions for these 14 F00 clusters as a function of clus- ingaredsequencewouldlikelyhavesignificantlydifferent tervelocitydispersion. Without alargernumber ofdata morphologicalfractions from the EDisCS, F00, and P05 points it is difficult to make a conclusive statement re- samples, leading to increased scatter in Figures 3 and garding the existence of a correlation. Spearman and 4. While both the incidence of such clusters and their Kendall rank correlation tests indicate that none of the morphologicalcontent are currently impossible to quan- morphological fractions are significantly correlated with tify, in the next subsection we evaluate the dependence clustervelocitydispersion. However,removaloftwoclus- of morphological fractions on another cluster property: ters (A389 and A3330) with unusually high S0 fractions velocity dispersion. for their velocity dispersions results in a 2σ detection of a correlation between S0 fraction and cluster velocity 6.2. Correlation between Morphological Fractions and dispersion. Additional velocity dispersions for the F00 Cluster Velocity Dispersion sample would greatly aid an assessment of any correla- In the previous subsection, we argued that there is tion, which is necessary for understanding the observed no systematic evolution in the morphological fractions trends between morphology and redshift. EDisCS MorphologicalContent 9 Comparing the x-axes of Figures 5 and 6, we see that (P06). Based on these data, P06 put forth a model in the F00 sample at z < 0.5 includes some clusters with which passive galaxies devoid of ongoing star formation very high velocity dispersions (σ > 1200 km s−1), while are made up of two separate populations: “primordial” the z > 0.5 EDisCS+P05 sample does not. Figure 7 galaxies whose stars all formed at z > 2.5 and galax- shows the cluster velocity dispersions of the EDisCS, ies whose star formation lasted until later times but was F00, and P05 samples as a function of lookback time. ultimately “quenched” due to entering a cluster-like en- Theclusterswiththehighestvelocitydispersionsinthese vironment (> 1014M⊙). In this model, the primordial samples lie at low redshift. Unfortunately, velocity dis- galaxiesmaybeidentifiedwiththeellipticalgalaxiesthat persions are unavailable for a significant fraction of the make up ∼30% on average of the galaxy populations in lowredshiftsample. Theseareurgentlyneededtounder- clusters at z < 1, and perhaps some of the S0 galaxies standthe extentto whichsampleselectionis responsible (see below). This identification is consistent with both for the apparent evolution in the morphological content the lack of evolution in the elliptical fraction at z < 1 of clusters. and with the old ages inferred for elliptical stellar popu- lations. It is tempting to identify the quenched galaxies 7. DISCUSSIONANDCONCLUSIONS with S0s. However, the fraction of quenched galaxies in Usinghighresolutionimagingaffordedbythe ACSin- the model increases at z < 0.8, while the S0 fraction strumentonboardthe Hubble Space Telescope,wehave only increases at z < 0.4. These timescales are consis- shown that the morphological content of rich clusters tent if it takes roughly a billion years for a galaxy to with velocity dispersions in the range σ = 200−1200 resemble an S0 after the cessationof star formation in a kms−1 showsno evidencefor evolutionoverthe redshift spiralgalaxy (Dressler et al. 1999; Poggiantiet al. 1999; range 0.4 < z < 1.25 (see Figures 3 and 4). Although Tran et al.2003). However,theformationoftheS0pop- the scatter is significant, typical morphologicalfractions ulation observed in clusters at 0.4 < z < 1 cannot be for clusters in this redshift range are 0.3, 0.15, and 0.55 explained in this way. Perhaps these are “primordial” for E, S0, and Sp+Irr galaxies,respectively. In contrast, S0s that assumed an S0 morphology upon or soon after studiesofclustersatlowerredshiftshaveshownthatthe formationat z >2.5. If so, the S0 galaxiesin clusters at elliptical fraction remains constant between z = 0 and z > 0.4 should be, on average, older and more massive z ∼0.5,whiletheS0fractiondecreasesbyafactorof∼2 thanclusterS0satz <0.4. Anotherpossibilityisthatat and the Sp+Irr fraction increases by a similar amount least some late types may transform into S0s in systems over the same redshift range (e.g. F00). Our data show less massive than 1014M⊙, perhaps by an entirely dif- thattheobservedevolutioninclusterS0andSp+Irrpop- ferent mechanism than operates in > 1014M⊙ systems. ulations does not continue beyond z ∼ 0.4, at least for The existence of S0s in groups (e.g. Hickson et al. 1989) the velocity dispersions probed in this study. suggests that this may be the case. How do our results concerning the global evolution of If a significant fraction of S0 galaxies in nearby clus- galaxy morphologies within rich clusters translate into ters werequenched, what did they look like beforehand? a statement regarding the evolution of the morphology- TheparalleldeclineintheSp+IrrfractionastheS0frac- density relation? Although the centers of cluster cores tion increases between z = 0.4 and z = 0 suggests that are regions of high local galaxy density, our morphologi- some subset of late type galaxies transformed into S0s calfractionswerecomputedwithinsizableapertures(see subsequent to quenching. For the EDisCS clusters, we §5.2), so the average environment we are probing is of computed the fractions of Sp and Irr galaxiesseparately moderate density. For example, the average galaxy sur- (see Tables 3 through 4). As with the other morpholog- face density within a radius of 0.6 × R (Ω = 0.3, ical fractions at z >0.4, the Sp and Irr fractions do not 200 0 Λ = 0.7, H = 70 km s−1 Mpc−1) for the EDisCS clus- varysystematicallywithredshift. Moreover,theIrrfrac- 0 ters used in this analysis ranges from ∼40–175 Mpc−2. tionsintheEDisCSclustersareverysmall. Weconclude The morphologicalfractions of the F00 andP05 clusters that Sp galaxies and not Irr galaxies are the precursors were likely computed using galaxies inhabiting environ- of the quenched S0s in local clusters. mentsofsimilaraveragedensity. Theseresultsmaypoint In the P06 model, both nature and nurture (within to a lackof evolutionin the morphologydensity relation clusters) play a role in the environmental dependence between z = 0.8 and z = 0.4, with a subsequent in- of star formation and therefore morphology. The nur- creaseinthe S0populationandadecreaseinthe Sp+Irr turecomponentwithinclustersisconsistentwiththeob- population. Larger samples in this redshift range are servation that the relation between star formation and needed to understand the scatter in the morphological environment in nearby clusters is not solely a reflec- fractions, and to rule out weak evolution. However, our tion of the relation between stellar age and environ- current data are consistent with recent studies of the ment, as would be expected in a pure nature scenario morphology-densityrelationatz ∼1 (Smith et al.2005; (Christlein & Zabludoff 2005). The P06 model is also Postman et al. 2005). compatible with observations which show that the SFR- Severalstudiesofnearbygalaxiesindicatethattheob- density relation out to z ∼ 1 extends to very low local served relation between morphology and environment is densities, comparable to those found at the virial radius a reflection of a more primary relation between star for- of clusters and even outside clusters (Lewis et al. 2002; mation rate (SFR) and environment (Kauffmann et al. G´omez et al.2003; Cooper et al.2006). The low density 2004; Christlein & Zabludoff 2005; Blanton et al. 2005). relations could be set up by the dependence of galaxy Thestarformationpropertiesofcluster,group,andfield properties on initial conditions, or perhaps a different galaxies in the EDisCS spectroscopic sample have been mechanism creates S0 galaxies in low-density environ- measured and compared to a low redshift sample from ments, or a combination. the Sloan Digital Sky Survey by Poggiantiet al. (2006) Although the model forwarded by P06 is an attrac- 10 Desai et al. tive framework within which to view our observational and cluster velocity dispersion is itself a function of red- results on the morphological content of rich clusters, it shift. Largerclustersampleswithrobustvelocitydisper- is not necessarily the only model which could explain sions and morphologies are needed to determine which all of the data accumulating on the evolution of galaxy oftheseoptionsismostlikely. Ifthemorphology-density properties with environment. Furthermore, it does not relationis drivenby environmentalprocessesinclusters, identify a mechanism responsible for the “quenching”, suchmeasurementsareessentialfordetermininghowthe although it does specify a timescale (3 Gyr) and a mass efficiencyofthesetransformationsdependsonclusterve- scale (1014M⊙) of a workable mechanism. locitydispersion,which,asdiscussedin§6.2isrelatedto In addition to analyzing the evolution of the morpho- both the cluster mass and its dynamical state. logicalcontentingalaxyclustersouttoz =1.25,wehave also studied the dependence of the morphological con- tentonthe clustervelocitydispersion(see Figures5and 6). We find that the early and late-type fractions in the We greatly appreciate the timely and cheerful assis- EDisCS and P05 cluster samples correlate with cluster tance of Galina Soutchkova in planning the HST obser- velocitydispersionatastatisticallysignificantlevel. This vations. We also thank the referee, Manolis Plionis, for correlation highlights the importance of understanding feedback which improved the paper. VD acknowledges global cluster properties in samples where evolution is funding from the Graduate Student Researchers Pro- observed, such as the F00 sample. Unfortunately, only gram. This work was supported by NASA grant HST- limitedvelocitydispersioninformationisavailableforthe GO-09476.01. JJDwaspartiallysupportedbytheAlfred F00clusters. The existinginformationfor14clustersin- P. Sloan Foundation. The Medium Deep Survey catalog dicates that they have a higher average velocity disper- is based on observations with the NASA/ESA Hubble sionthaneithertheEDisCSorP05clusters. Inaddition, SpaceTelescope,obtainedattheSpaceTelescopeScience the morphological fractions of these 14 clusters do not Institute, which is operated by the Association of Uni- correlate strongly with cluster velocity dispersion. It is versities for Research in Astronomy, Inc., under NASA possible that additional velocity dispersions for the F00 contract NAS5-26555. The Medium-Deep Survey anal- sample will reveal an underlying trend, that additional ysis was funded by the HST WFPC2 Team and STScI studies at high redshift will prove the trend observed grantsGO2684,GO6951,GO7536,andGO8384toProf. in the higher redshift EDisCS+P05 sample to be spuri- RichardGriffithsandDrKavanRatnatungaatCarnegie ous, or that the relation between morphological fraction Mellon University. REFERENCES Abadi,M.G.,Moore,B.,&Bower,R.G.1999,MNRAS,308,947 Halliday,C.etal.2004,A&A,427,397 Abraham, R. G., Tanvir, N. R., Santiago, B. X., Ellis, R. S., Hickson,P.,Kindl,E.,&Auman,J.R.1989,ApJS,70,687 Glazebrook, K.,&vandenBergh,S.1996, MNRAS,279,L47 Icke,V.1985,A&A,144,115 Bekki,K.,Couch,W.J.,&Shioya,Y.2002,ApJ,577,651 Kauffmann, G., White, S. D. M., Heckman, T. M., M´enard, B., Bertin,E.,&Arnouts,S.1996,A&AS,117,393 Brinchmann, J., Charlot, S., Tremonti, C., & Brinkmann, J. Bettoni, D., Moles, M., Kjærgaard, P., Fasano, G., & Varela, J. 2004,MNRAS,353,713 2006, A&A,452,811 Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980, ApJ, 237, Blanton,M.R.etal.2001,AJ,121,2358 692 Blanton, M. R., Eisenstein, D., Hogg, D. W., Schlegel, D. J., & Lavery,R.J.,&Henry,J.P.1988,ApJ,330,596 Brinkmann,J.2005,ApJ,629,143 Lewis,I.etal.2002, MNRAS,334,673 Bolzonella,M.,Miralles,J.-M.,&Pello´,R.2000,A&A,363,476 Melnick,J.,&Sargent, W.L.W.1977,ApJ,215,401 Byrd,G.,&Valtonen, M.1990,ApJ,350,89 Mihos,J.C.2004,inClustersofGalaxies:ProbesofCosmological Christlein,D.,&Zabludoff, A.I.2005,ApJ,621,201 StructureandGalaxyEvolution,277–+ Clowe,D.etal.2006,A&A,451,395 Moore,B.,Lake, G.,&Katz,N.1998,ApJ,495,139 Cooper,M.C.etal.2006,MNRAS,370,198 Poggianti,B.M.1997,A&AS,122,399 Couch, W. J., Barger, A. J., Smail, I., Ellis, R. S., & Sharples, Poggianti, B. M., Smail, I., Dressler, A., Couch, W. J., Barger, R.M.1998,ApJ,497,188 A.J.,Butcher,H.,Ellis,R.S.,&Oemler,A.J.1999,ApJ,518, Couch,W.J.,&Sharples,R.M.1987,MNRAS,229,423 576 Dalcanton, J.J.1996, ApJ,466,92 Poggianti,B.M.etal.2006,ApJ,642,188 DePropris,R.etal.2002,MNRAS,329,87 Postman,M.etal.2005,ApJ,623,721 Dressler,A.1980a,ApJs,42,565 Postman,M.,Lauer,T.R.,Szapudi,I.,&Oegerle,W.1998,ApJ, —.1980b,ApJ,236,351 506,33 Dressler,A.etal.1997,ApJ,490,577 Quilis,V.,Moore,B.,&Bower,R.2000,Science,288,1617 Dressler,A.,&Shectman, S.A.1988,AJ,95,985 Richstone,D.O.1976,ApJ,204,642 Dressler,A.,Smail,I.,Poggianti,B.M.,Butcher,H.,Couch,W.J., Rudnick,G.etal.2001, AJ,122,2205 Ellis,R.S.,&Oemler,A.J.1999,ApJS,122,51 —.2003, ApJ,599,847 Fasano, G., Poggianti, B. M., Couch, W. J., Bettoni, D., Smail, I., Dressler, A., Couch, W. J., Ellis, R. S., Oemler, A. J., Kjærgaard,P.,&Moles,M.2000,ApJ,542,673 Butcher,H.,&Sharples,R.M.1997, ApJS,110,213 Finn,R.A.etal.2005, ApJ,630,206 Smith, G. P., Treu, T., Ellis,R. S., Moran, S. M., & Dressler,A. Go´mez,P.L.etal.2003,ApJ,584,210 2005,ApJ,620,78 Gehrels,N.1986,ApJ,303,336 Thomas,T.,&Katgert,P.2006,A&A,446,31 Girardi,M.,&Mezzetti,M.2001,ApJ,548,79 Toomre,A.,&Toomre,J.1972,ApJ,178,623 Gonzalez,A.H.,Zaritsky,D.,Dalcanton,J.J.,&Nelson,A.2001, Tran, K. H., Simard, L., Illingworth, G., & Franx, M. 2003, ApJ, ApJS,137,117 590,238 Goto,T.etal.2002,PASJ,54,515 Treu,T.,Ellis,R.S.,Kneib,J.,Dressler,A.,Smail,I.,Czoske,O., Goto, T., Yagi, M., Tanaka, M., & Okamura, S. 2004, MNRAS, Oemler,A.,&Natarajan,P.2003,ApJ,591,53 348,515 White,S.D.M.etal.2005,A&A,444,365 Griffiths,R.E.etal.1994,ApJ,435,L19 Zaritsky, D., Nelson, A. E., Dalcanton, J. J., & Gonzalez, A. H. Gudehus,D.H.1989, ApJ,340,661 1997,ApJ,480,L91 Gunn,J.E.,&Gott, J.R.I.1972,ApJ,176,1 Zucca,E.etal.2006,A&A,455,879

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.