ebook img

A journey from the outskirts to the cores of groups I: Color- and mass-segregation in 20K-zCOSMOS groups PDF

5.1 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 A journey from the outskirts to the cores of groups I: Color- and mass-segregation in 20K-zCOSMOS groups

Astronomy&Astrophysicsmanuscriptno.Presotto˙2011 (cid:13)c ESO2012 10-1-2012 A journey from the outskirts to the cores of groups I: ⋆ Color- and mass-segregation in 20K-zCOSMOS groups V.Presotto1,2,A.Iovino2,M.Scodeggio3,O.Cucciati4,C.Knobel5,M.Bolzonella6,P.Oesch5,7,A.Finoguenov8, M.Tanaka9,K.Kovacˇ5,10,Y.Peng5,G.Zamorani6,S.Bardelli6,L.Pozzetti6,P.Kampczyk5,C.Lo´pez-Sanjuan13, D.Vergani6,24,E.Zucca6,L.A.M.Tasca13,C.M.Carollo5,T.Contini11,12,J.-P.Kneib13,O.LeFe`vre13,S.Lilly5, V.Mainieri14,A.Renzini15,A.Bongiorno8,K.Caputi16,S.delaTorre16,L.deRavel16,P.Franzetti3,B.Garilli3,13, F.Lamareille11,12,J.-F.LeBorgne11,12,V.LeBrun13,C.Maier5,17,M.Mignoli6,R.Pello`11,E.Perez-Montero11,12,18, E.Ricciardelli19,J.D.Silverman9,L.Tresse13,L.Barnes5,R.Bordoloi5,A.Cappi6,A.Cimatti20,G.Coppa8, 2 A.M.Koekemoer21,H.J.McCracken22,M.Moresco20,P.Nair6,andN.Welikala23 1 0 (Affiliationscanbefoundafterthereferences) 2 ReceivedMonthDay,Year;acceptedMonthDay,Year n a J ABSTRACT 9 Context.Studyingtheevolutionofgalaxieslocatedwithingroupsmayhaveimportantimplicationsforourunderstandingoftheglobalevolution ] ofthegalaxypopulationasawhole.Thefractionofgalaxiesboundingroupsatz∼0isashighas60%andthereforeanymechanism(amongthe O manysuggested)thatcouldquenchstarformationwhenagalaxyentersgroupenvironmentwouldbeanimportantdriverforgalaxyevolution. C Aims.UsingthegroupcatalogobtainedfromzCOSMOSspectroscopicdataandthecomplementaryphotometricdatafromtheCOSMOSsurvey, weexploresegregationeffectsoccurringingroupsofgalaxiesatintermediate/highredshifts.Ouraimistorevealif,andhowsignificantly,group . h environmentaffectstheevolutionofinfallinggalaxies. p Methods.Webuilttwocompositegroupsatintermediate(0.2≤z≤0.45)andhigh(0.45<z≤0.8)redshifts,andwedividedthecorresponding - compositegroupgalaxiesintothreesamplesaccordingtotheirdistancefromthegroupcenter.Thesamplesroughlycorrespondtogalaxieslocated o in agroup’s inner core, intermediate, and infall region. Weexplored how galaxy stellar masses and colors - working in narrow bins of stellar r masses-varyasafunctionofthegalaxydistancefromthegroupcenter. t s Results.Wefoundthatthemostmassivegalaxiesinoursample(log(M /M ) ≥10.6)donotdisplayanystronggroup-centricdependenceof gal ⊙ a thefractionsofred/blueobjects.Forgalaxiesoflowermasses(9.8≤log(M /M )≤10.6)thereisaradialdependenceinthechangingmixof [ gal ⊙ redandbluegalaxies.Thisdependenceismostevidentinpoorgroups,whereasrichergroupsdonotdisplayanyobvioustrendofthebluefraction. 1 Interestingly,masssegregationshowstheoppositebehavior:itisvisibleonlyinrichgroups,whilepoorergroupshaveaaconstantmixofgalaxy v stellarmassesasafunctionofradius. 3 Conclusions.These findings can be explained in a simple scenario where color- and mass-segregation originate from different physical pro- 7 cesses.Whiledynamicalfrictionistheobviouscauseforestablishingmasssegregation,bothstarvationandgalaxy-galaxycollisionsareplausible 6 mechanismstoquenchstarformationingroupsatafasterratethaninthefield.Inpoorergroupstheenvironmentaleffectsarecaughtinaction 1 superimposedtoseculargalaxyevolution.Theirmembergalaxiesdisplayincreasingbluefractionswhenmovingfromthegroupcentertomore . externalregions,presumablyreflectingtherecentaccretionhistoryofthesegroups. 1 0 Keywords.cosmology:observations-galaxies:groups:general-galaxies:evolution 2 1 v: 1. Introduction low-densityregions,whereasred,inactive,ellipticalgalaxiesfa- i vorhigh-densityregions. X Thestrikingbi-modalityofthecolor-magnitudeandofthecolor- Thesetwodistinctgalaxyevolutionaryfamiliescanoriginate mass diagrams raises important questions about galaxy forma- r either from a priori differences set at galaxy formation epoch, a tion and evolution. What are the physical processes responsi- the so-called nature scenario, or from environmentally driven ble for the sharp partition into blue cloud and red sequence processes taking place during the galaxy evolutionary history, galaxies? Does the environment play a key role in this pro- theso-callednurturescenario. cess by boosting the transition into the red sequence region? What are the timescales for this transition? There is much ev- The currently accepted ΛCDM model predicts the hierar- idence of correlations between galaxy properties and their en- chical growth of structures: as time proceeds, smaller struc- vironment, the oldest and best known being the morphology- turesmergetoformprogressivelylargerones.Thisprocessim- densityrelation(seeOemler1974;Dressler1980,althoughthe plies that the fraction of galaxies located in groups progres- firstmentionofit datesbackto Hubble).Ingeneral,blue,star- sivelyincreasessincez ∼ 1.5,uptotheLocalUniversevalues, forming, disk-dominated galaxies are located preferentially in wheremostgalaxiesarefoundingroups(Huchra&Geller1982; Ekeetal. 2004; Berlindetal. 2006; Knobeletal. 2009). As a Send offprint requests to: Valentina Presotto consequence,atleastpartofthe observeddeclineofthe global ([email protected]) star-formation rate (SFR) from z ∼ 1.5 until today (Lillyetal. ⋆ BasedonobservationsmadeattheEuropeanSouthernObservatory 1996; Madauetal. 1998; Hopkins 2004) could be accelerated (ESO)VeryLargeTelescope(VLT)underLargeProgram175.A-0839 byenvironmentallydrivenphenomena. 1 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample In this context group environment plays a dominant role ies with environment. To isolate the true environmentaleffect, (Baloghetal. 2004; Wilmanetal. 2005; Iovinoetal. 2010; the analysis must be performed in narrow galaxy stellar mass Pengetal.2010),becauseonlyasmallfractionofgalaxieslive bins.Muchoftheearlierworkperformedatintermediate/highz indenserenvironmentssuchasclustercores.Thereareobserva- wasbasedonincompleteand/orscarcesamplesofgroups,where tionalindicationsthatthecolortransitionfrombluetoredgalax- oftenthestatistics wasnothighenoughto performsuch anac- iesproceedsfasterinagroupthaninthecoevalfieldpopulation, curateanalysisinsmallmassbins. aneffectthatbecomesevidentatredshiftslowerthanz ∼ 1and In this paperwe will study mass - and color segregationin forgalaxiesofmasseslog(Mgal/M⊙)≤10.6(Iovinoetal.2010; groupsoverwide redshiftandgalaxystellar mass rangesusing Kovacˇetal.2010;Bolzonellaetal.2010;Pengetal.2010). thespectroscopicdatafromtherecentlycompletedzCOSMOS- However, the physical phenomena responsible for accel- Bright, a large survey reaching out to z ∼ 1 with a fairly erating the color transition from blue to red galaxies within high and uniformsampling rate (Lillyetal. 2007), and its new groups are yet to be described. While the extreme local den- groupcatalog(Knobeletal.2011).Wewillbenefitalsofromthe sities reached within cluster cores enable efficient ram pres- widerangeofphotometricdataavailablefortheCOSMOSsur- sure stripping of the galaxy cold gas on timescales of a few vey (Scovilleetal. 2007; Ilbertetal. 2009; Oeschetal. 2010). Myr (Gunn&Gott 1972; Abadietal. 1999), within the groups Galaxycolorsaretheeasiestparametertomeasureamongthose different physical processes have been proposed. On one hand thatexhibitadistinctivebi-modality,andweselectedrest-frame galaxy-groupinteractionslike’strangulation’,starvationorhalo (U −B)color,bracketingthe4000Åbreak,asagoodindicator gasstrippingcanremovewarmandhotgasfromagalaxyhalo, ofthegalaxyaveragestar-formationhistoriesoverlongertime- efficientlycuttingoffthestarformationgassupply(Larsonetal. scalesthanemissionlineindicatorssuchase.g.,[OII]. 1980;Coleetal.2000;Baloghetal.2000;Kawata&Mulchaey Toshedlightonhowrapidlyandsignificantlystarformation 2008).Alternatively,mergers/collisionsandclosetidalencoun- issuppressedingroupsandtoovercomethelownumberstatis- tersamonggroupmembergalaxiestogetherwithgalaxy-galaxy tics for individual systems (typically 7-8 members per group), harassment at the typical velocity dispersion of bound groups we built stacked groupsby co-addingspatial informationfrom mayalsoresultinstar-formationquenching(Mooreetal.1996). groupmembergalaxies.This strategyenabledus to establish a Thesephysicalprocessesdonotrequireextremelocaldensities, statisticallyreliablesampleandtorevealtrendsofgalaxyprop- andquenchstarformationinamoregradualandgentlewayon ertiesasafunctionofthegroup-centricdistanceandofvarying timescalesofseveralGyr. grouprichnesses. Among the observable effects of these processes are seg- Thepaperisorganizedasfollows:inSect.2wedescribethe regation phenomena, that is, not only differences between data of our analysis, includingthe algorithmchosen to add the group - and field galaxy properties, but also radial trends of groupmembergalaxieswithphotometricredshift.InSect.3we galaxy properties (e.g., colors, morphologies ...) as a func- illustrate the construction of the realistic mock catalogs with tion of distance from the group/cluster center. These phenom- which we tested our algorithms. In Sect.4 we explain how we ena have already been extensively studied in galaxy clusters, stacked group member galaxiesto build a composite group.In where e.g., a strong radial dependence in the star-formation Sects.5and6 wepresentouranalysisanditsresults, whichwe rateisobserved(Hashimoto&Oemler1999;Baloghetal.1999; discussinSect.7.OurconclusionsaresummarizedinSect.8.A Lewisetal. 2002; Baloghetal. 2004; Tanakaetal. 2004). The concordance cosmology is adopted throughoutour paper, with observed quenching of star-formation activity starts at large h = H /70 km s−1 Mpc−1, Ω = 0.25 and Ω = 0.75. All 70 0 m Λ cluster-centric distances and low projected densities, in the magnitudesarequotedintheABsystemthroughout. so-called infalling regions, and even at large radii field star- formation values are not yet reached. This result suggests that galaxy transformationstarts to occur in the infalling filaments, 2. Data which consist of chains of groups in which field galaxies are affected by the group environment which changes their star- It is widely accepted in the literature that classical galaxy forming blue field-like properties into passive, red cluster-like color/morphologytrendsindifferentenvironmentsarebetterin- galaxies. But even if groups seem to be the key environment vestigatedinbinsofmass-volume-limitedsamples(Tascaetal. to search for the nurture scenario, still the observational ev- 2009;Iovinoetal.2010;Cucciatietal.2010;Kovacˇetal.2010; idence for related segregation phenomena is quite scarce and Xueetal. 2010; Cooperetal. 2010; Gru¨tzbauchetal. 2011). holds mainly for the local Universe (Postman&Geller 1984; This strategy enables one to break the degeneracy caused by Mahdavietal. 1999; Tranetal. 2001; Carlbergetal. 2001a,b; therelationshipsbetweengalaxystellarmassesandenvironment Girardietal.2003;Dom´ınguezetal.2002;Wilmanetal.2009; andbetweengalaxystellarmassesandcolors/morphologies. Baietal.2010;Ribeiroetal.2010). The recently completed zCOSMOS-bright survey with its A complication to consider is the strong correlation be- highanduniformsamplingrateoffersuniqueopportunitytoex- tween galaxy properties such as colors, morphologies and plorethepresence/evolutionofthesetrendsoverawiderangeof star formation, with galaxy stellar mass (Cowieetal. 1996; cosmictime. Gavazzietal.1996;Blantonetal.2003;Kauffmannetal.2003; Brinchmannetal. 2004; Baldryetal. 2004) and the additional 2.1.COSMOSandzCOSMOSsurveys correlation between the galaxy stellar mass itself and environ- ment: galaxies in less dense environment tend to be less mas- The COSMOS survey is a large HST-ACS survey,with I-band sivethanthoselocatedindenserenvironment(Hoggetal.2003; exposuresdowntoI = 28onafieldof2deg2 (Scovilleetal. AB Kauffmannetal. 2004; Blantonetal. 2005; Scodeggioetal. 2007). The COSMOS field has been the object of extensive 2009; Bolzonellaetal. 2010). Thus any study performed on multiwavelength ground- and space-based observations span- samples of galaxies containing a wide range of stellar masses ningtheentirespectrum:X-ray,UV,optical/NIR,mid-infrared, cannot distinguish between true environmental effects and ef- mm/submillimeterandradio,providingfluxesmeasuredover30 fectssimplyinducedbythedifferingmassdistributionsofgalax- bands (Hasingeretal. 2007; Taniguchietal. 2007; Capaketal. 2 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample ricredshifts(Ilbertetal.2009).Basedonacomparisonwiththe zCOSMOS spectroscopicredshifts,Ilbertetal. (2009) estimate anaccuracyofσ =0.007×(1+z)forgalaxiesbrighterthan zphot s I = 22.5.Applyingthe ZEBRA code(Feldmannetal. 2006) AB to 30 bands, Oesch (2011) obtains a similar accuracy (private communication). Inouranalysisweusedphotometricredshiftvaluesobtained bytheZEBRAcode. ForallgalaxiesbrighterthanI =22.5,absoluterest-frame AB magnitudes and stellar masses were obtained using standard multi-colorspectralenergydistribution(SED)fittingtechniques, usingthesecurespectroscopicredshift,ifavailable,orthepho- tometricone.Rest-frameabsolutemagnitudeswereobtainedus- ingtheZEBRAcode(seeFeldmannetal.2006forthedetailsof thecode),whilestellarmasseswereobtainedusingthehyperz- masscode(Pozzettietal.2010;Bolzonellaetal.2010).Fromthe availablestellarpopulationsynthesislibrariesweadoptedthose of Bruzual&Charlot (2003), assuming a Chabrier initial mass function(Chabrier2003). Fig.1. Ra-Dec distribution of the 16623 zCOSMOS-bright galaxies with secure redshift z ≤ 2 (the so-called 20K sam- ple). The area within the red box (149.55 ≤ ra ≤ 150.67 and 1.75 ≤ dec ≤ 2.70) has a nearly uniform sampling rate of 2.2.Thespectroscopicgroupcatalog ∼62%. The group catalog used in this paper is a subset of the 20K group catalog described in Knobeletal. (2011), Knobeletal. 2007;Lillyetal.2007;Sandersetal.2007;Bertoldietal.2007; (see also 2009 for an earlier version of the catalog). The 20K Schinnereretal.2007;Koekemoeretal.2007;McCrackenetal. group catalog consists of 1496 groups with at least two spec- 2010). troscopicmembergalaxies(188withatleastfivespectroscopic The zCOSMOS survey was planned to provide the cru- members).Knobeletal.(2011)usesa“multi-passprocedure”to cial high-quality redshift information to the COSMOS field achieveanimpressivequalityingroupreconstruction,astested (Lillyetal. 2007). It benefitted of ∼ 600 hr of observations at usingrealisticmockcatalogs.Thismethod,whencombinedwith VLTusingtheVIMOSspectrographanditconsistsoftwoparts: the standard friends-of-friends(FOF) algorithm, yields for the zCOSMOS-bright,and zCOSMOS-deep.The zCOSMOS-deep resultinggroupcatalogvaluesofcompleteness(i.e.,fractionof targets ∼ 10000 galaxies within the central 1 deg2 of the real detected groups) and purity (i.e., fraction of non-spurious COSMOSfield,selectedthroughcolorcriteriatohave1.4<∼z<∼ groups)thatareextremelygoodandstableasafunctionofboth 3.0.ThezCOSMOS-brightispurelymagnitude-limitedandcov- redshiftand numberof membersobservedin the reconstructed ersthewholeareaof1.7deg2oftheCOSMOSfield.Itprovides groups. Typical values of these two quantities, for groups re- redshiftsfor∼ 20000galaxiesdowntoIAB ≤ 22.5asmeasured constructedwithfiveormorespectroscopicobservedmembers, from the HST-ACS imaging. The success rate in redshift mea- are around∼ 80%atall redshiftsanddo notdecrease substan- surementsisveryhigh,95%intheredshiftrange0.5<z< 0.8, tiallyforgroupswithlowernumberofobservedmembers.The and the velocity accuracy is ∼ 100 km s−1(Lillyetal. 2009). interloperfraction,i.e.,thefractionoffieldgalaxieserroneously Eachobservedobjecthasbeenassigneda flagaccordingtothe classed asgroupmembers,alwaysremainsbelow∼ 20%atall reliability of its measured redshift. Classes 3.x, 4.x redshifts, redshifts for groupsreconstructed with more than five spectro- plus Classes 1.5, 2.4, 2.5, 9.3, and 9.5 are considered a secure scopicobservedmembers,withonlyaslightincreaseforgroups set, with an overallreliability of 99% (see Lillyetal. 2009 for with lower numberof observedmembers.Anotherpointworth details). noticingisthatthealgorithmtodetectgroupstreatseachgalaxy OurworkisbasedonthethezCOSMOS-brightsurveyfinal as a point in Ra-Dec-redshiftspace, therefore avoiding any in- release:thesocalled20K sample(simply20K hereafter),total- terloper/completenessdependenceongalaxypropertiessuchas ing16623galaxieswithz ≤ 2andsecureredshiftsaccordingto colorsormasses(seeKnobeletal.2011formoredetails). theaboveflagclassification(18206objectsintotal,irrespective ofredshiftandincludingstars). Inthispapertheanalysisisrestrictedtogroupswithatleast Fig.1showsthespatialdistributionofthe20Kgalaxies.The fivespectroscopicallyobservedmembersthatarelocatedwithin redsquare correspondsto the regionwith the highestsampling the high sampling rate box introduced in Fig.1. From now on rate, approximately ∼ 62% of the parent galaxy catalog. Its we will call this sample the spectroscopic group sample: it to- boundariesare 149.55 ≤ ra ≤ 150.67and 1.75 ≤ dec ≤ 2.70. tals178groupsand1437groupmembergalaxiesatz ≤ 1.Our Within this region are 13619 galaxies with secure redshift and choiceenablesustoworkwithgroupsthathavebestvaluesfor z≤1(15730objectsintotal,irrespectiveofredshiftandinclud- purityandinterloperfraction,andtosecureareliabledefinition ingstars)andtheirskydistributionisremarkablyuniform. ofgroupcenterandradius.Thesetwoparametersarecrucialto ForobjectsbrighterthanI =22.5andwithoutsecurespec- buildthecompositegroupandforouralgorithmwhichretrieves AB troscopicredshift,thewealthofancillaryphotometricdatapro- groupmemberswithoutspectroscopicredshiftinformationd(see videdbytheCOSMOSsurveyprovidesgoodqualityphotomet- laterSect.2.4and4). 3 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample 2.3.Thespectroscopicfieldsample Toovercomethisdrawback,we developeda slightlydiffer- entstrategy,whosemainadvantageisthatitassignseachgalaxy To define the field galaxysample, we started by selecting 20K only one spectroscopic group, thus avoiding multiple assign- galaxies located within the high sampling rate box and out- mentsof a galaxytodifferentgroups,andthe needtoadoptan sideanyofthereconstructedgroupsofKnobeletal.(2011).We arbitraryprobabilitycut-offtobypassthisproblem. therefore discarded from this sample galaxies located in pairs, For a detailed description of our algorithm we refer the tripletsandquadruplets,i.e.,membersofthegroupswithfewer readertoAppendixA,whileinSect.3wewillpresentextensive observedmembersarenotconsideredinourscienceanalysis. teststhatweperformedonmockcatalogstocheckthereliability To performthe fairest comparisonbetween groupand field ofthefinalspec+photo-zgroupcatalog. samples, we took into accountthe possibility of spurioustrend Sufficeistosaythatwechosetheselectionfunctiontoiden- introduced by residual group contamination or by the differ- tifyputativephotometric-redshiftmembersinawaytonotonly ent redshift ranges covered by the group/field galaxy samples. keep the fraction of interlopers as low as possible, but also to Galaxieslyingintheclosestproximityofgroupscouldbecon- avoid introducinganyradialdependencyof the interloperfrac- taminatedbyspectroscopicgroupmembersmissedbythegroup- tion.Thelastpointisimportantbecausewewillbelookingfor finding algorithm. In addition, the redshift distribution of the radialdependenciesofgalaxyproperties. spectroscopicgroupcatalogisfarfrombeinguniform,display- As already mentioned in Sect.2.2, we chose a conservative ing prominentpeaks, especially at low redshift where the 20K definition of the spectroscopic group sample, restricting our- fieldofviewlimitsthecosmicvolumeexploredandweneedto selves to only 178 groups detected with at least five spectro- considertheappropriatecoevalfieldpopulation. scopic members within the the high sampling rate box intro- To take into accountthese two factors, we furthermore re- ducedin Fig.1box(149.55 ≤ ra ≤ 150.666and1.75 ≤ dec ≤ strictedthefieldsampletogalaxieslocatedwithinvelocitydis- 2.7).Withinthisareaanduptoz=1.0thereare13619galaxies tances2000≤|∆v|≤5000kms−1 fromthespectroscopicgroup withreliablespectroscopicredshiftand11994withanestimated sample, - therefore following the same redshift ditribution of photometricredshift. group member galaxies - and with radial projected distances Ouralgorithmaddsanother684membergalaxieswithpho- R > 4 × R from any group of 20K group catalog, R tometric redshifts to the already existing 1437 spectroscopic fudge fudge beingan estimate of the virialradiusprovidedby Knobeletal. group member galaxies, and from now on this is the group (2011)(seeSect.4.2fordetails).Fromnowonwewillcallthis sample we will use. The final number of groups with more setofgalaxiesthefieldsample,totalling6556galaxiesatz≤1. than10(15)membersafterapplyingouralgorithmistwice(three times) that in the spectroscopic group catalog, i.e., there are We also introduced a complementary set of field galaxies 78(41)groupsinsteadoftheoriginal39(14)groups.Thenumber that we call near-field galaxies. These are 20K galaxieswithin of groupswith more than 20 membersis six times the original the high sampling rate box that do not belong to any of the one:25groupsinsteadoftheoriginalfourgroups. reconstructed groups of Knobeletal. (2011), but with velocity distances |∆v| ≤ 2000 km s−1 and radial projected distances Asafinalpointwenoticethatwerepeatedallanalysespre- sented in this paper considering only galaxies from the spec- R≤4×R fromatleastonegroupofthespectroscopicgroup fudge troscopic group catalog and our results remained entirely un- sample.Thenear-fieldsamplesodefinedtotals1694galaxiesat changed,albeitatalowersignificance. z ≤ 1 and contains, by definition, galaxies located in the close proximityofthespectroscopicgroupsample.InSect.6.3wewill use thissample to check forpossible environmentaleffects ex- 3. ThezCOSMOSmockcatalogs tendingoutside groupradii,e.g.,colordifferencesofnear-field galaxypopulationwithrespecttothegeneralfieldsample. The use of realistic mock galaxy catalogs is important for as- sessing the reliability of the algorithm we adopted to produce the spec+photo-z group catalog and to validate the procedures 2.4.Addingphoto-zs:thespec+photo-zgroupcatalog we chose to define group centers and richnesses (see Sect.2.4 andSect.4). Foreachgroupthenumberofavailablemembergalaxiesdown We took advantage of the 24 COSMOS mock light-cones toI =22.5islimitedbytheincompletesamplingrateof20K. AB provided by Kitzbichler&White (2007). These mock light- To increase this number, we took advantage of the exquisite cones are based on the Millennium DM N-body simulations quality of the photometric redshifts available in the COSMOS ofSpringel(2005)andusesemianalyticrecipesofCrotonetal. field,see Sect.2toincorporatein ouranalysisphotometricred- (2006)asupdatedbyDeLucia&Blaizot(2007)forpopulating shiftsforgalaxiesbrighterthanI = 22.5andwithoutreliable AB thesimulationsvolumewithgalaxies. spectroscopic data. A higher number of group member galax- Fromeachofthese24light-conesweextractedthreediffer- iesenablesonetoimprovecenteringandrichnessestimatesfor enttypesofmockcatalogs: eachgroup,quantitiescrucialtoproperlycenterandrescaledis- tinct groupsto build a composite one (see e.g., Carlbergetal. 1. The 40K mock catalogs: 100%complete to I = 22.5. In AB 1997). thesecatalogsallgalaxiesbrighterthanI =22.5arespec- AB In Knobeletal. (2011) a probabilityapproachwas adopted troscopicallyobservedwitha100%successrate.Weadded to retrieve member galaxies brighter than I = 22.5 and that toeachgalaxyredshiftanerrorof100kms−1toaccountfor AB have no spectroscopic information.To each galaxy a probabil- the typical zCOSMOS spectroscopic redshift error as esti- ity, p ,ofbeingpartofagroupwasassigned,dependingonits matedfromobservations(seeLillyetal.2009). in projectedradialandvelocitydistancefromthegroupcenter(we 2. The 20K mock catalogs: mimicking the 20K zCOSMOS referthe readertothepaperbyKnobeletal.(2011) fora more spectroscopic sample. We applied the same observational detailed description of the adopted method). The drawback of strategyadoptedto selectthe spectroscopiczCOSMOS tar- thisapproachisthateachgalaxymayhavemultipleassociations gets:usingthe slitpositioningalgorithmSPOC on the40K todifferentgroups. catalogs, see Bottinietal. (2005), and accounting for the 4 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample spectroscopicredshiftfailuresbyincludingthesameredshift externalrings,asinSect.4.2,andadoptedexactlythesamemass successrateastherealdata. and redshift limits adopted subsequently in our analysis, see 3. The 20K+photo-zs mock catalogs: mimicking the data set Sect.5. weusedinouranalysis.Thespectroscopicgalaxiesarethose In the top right panel of Fig.2 we show the fraction of in- listedinthe20Kmockcatalogs,whileaphotometricredshift terlopersineachofthesethreeregionsforthe20K/20K+photo- isprovidedfortheremaininggalaxiesofthe40Kmockcat- zs mock groups(black stars and red triangles, respectively),as alogs.Forthephotometricredshiftgalaxysamplewerepro- obtained using the low-z, mass-limited, mock group samples. ducedthephotometricredshifterrorσ =0.007×(1+z). Noticethatthetrendintroducedbythegroup-findingalgorithm zphot s We alsotookintoaccountthepresenceofcatastrophicfail- inthe20Kmockreconstructedgroupsisnotmodifiedbyadding uresinestimatingphoto-z,thatis,theexcessofgalaxieswith photo-zs members. In the same panel the orange squares dis- errorslargerthan∆phot-z= 3σ withrespecttothesim- playthefractionofinterlopersinthe20K samplespectroscopic zphot pleGaussiandistribution.Forthesetofphotometricredshift groupcatalog,estimatedusingtheprobabilities,p ,associated in,i adoptedinouranalysiswewereabletoestimate apercent- toeachobservedgroupspectroscopicmember: ageof∼8%(byusingthe20Ksubsetflagged4.xor3.xand comparing their spectroscopic redshift to their photometric Ntot,obs redshift). Group members with such high values of phot-z PI =1− X pin,i/Ntot,obs, (1) error cannot be retrieved by our algorithm and are a con- i=1 siderablesourceofincompletenessingroupreconstruction. Wechoseafairlyconservativeapproachandalsoconsidered asprovidedbyKnobeletal.(2011).Thesevalueshavebeen catastrophic errors of 10% in the 20K+photo-zsmock cat- calibrated in Knobeletal. (2011) using simulations, and there- alogs,byrandomlypermutingthephotometricredshiftsfor fore are by constructionagree well with the interloper fraction 10% of the galaxies while keeping the galaxyra-dec fixed. estimatedfrom20Kmockgroups.Inturn,becauseaddingphot- coordinates. z members does not alter the trends significantly, these values agree well with the interloper fraction for the 20K+photo-zs Weappliedtothe20Kmockcatalogsthesamegroupfinding mock groups. The picture does not change when plotting the algorithmusedforthe20Ksample(seeKnobeletal.2009).We same quantitiesfor the high-z, mass-limited,mock groupsam- then selected groups with at least five spectroscopic members ples,orwhenselectingsubsetsofgroupsaccordinge.g.,totheir located within the high sampling rate box introduced in Fig.1 richness.ThereforewealwaysusedthePIvaluesobtainedfrom and applied to the 20K+photo-zsmock catalogs the algorithm spectroscopic group members to estimate the interlopers’ con- describedinSect.2.4. tamination as a function of the distance from the group center The COSMOS mock light-conesprovidedark matter halos forourspec+photo-zgroupcatalog,seeSect.6.2. IDsthatcaneasilybeusedtoidentifyrealgroupsandrealgroup members, i.e., the set of galaxies located within the same dark matterhaloineachmockcatalog(seealsoKnobeletal.2009). 4. Buildingthestackedgroup Ifwedefinecompletenessastheratioofthereconstructedgroup members in the 20K/20K+photo-zsmock catalogs to the total To exploregalaxypropertiesas afunctionofthe distancefrom number of real group members in the 40K mock catalogs, the the center of the group, we needed to build ensemble systems, improvement introduced by our algorithm is shown in the top becausethescarcityofindividualgroupmemberspreventsade- leftpanelofFig.2.Thispanelshowsthedistributionofthecom- tailedanalysisofeachgroup.Inthissectionweillustrateindetail pletenessforallgroupsofthe20Kmocksandforthoseobtained the steps of building the so-called stacked-group: a composite afterapplyingouralgorithmtothe20K+photo-zsmocks(black groupobtainedbyspatiallyco-addingallgroupmembergalax- dot-dashedlineandred-solidline,respectively).Wewereableto ies(simplySG fromnowon). improvethe median completeness from 67% of the 20K mock The two main ingredients to build a SG are precise re- catalog up to 90% in the 20K+photo-zsmocks: the number of centeringand scalingof allavailable groups.Itis thereforeex- groupsthatare 100%completeis threetimes largerthanusing tremelyimportantthateachgroupcenterandrichnessisdefined onlythe20Kmocks.Asaconsequence,thegrouprichness,de- as reliably as possible, so that the trends we are searching for fined as the numberof membersbrighterthan an adopted rest- arenotsmoothedout.Wewilldiscussprecisedefinitionofboth frame absolute magnitude cut-off, is also easier to recover in quantitiesinthissection.Weremindthereaderthatfromnowon a reliable way. For more than half of the cases the richness as anynumberquoted,unlessexplicitlystated,includesbothspec- measuredforreconstructedgroupsinthe20K+photo-zsmocks troscopic and photometric redshift group member galaxies, as equalsthe same quantityas obtainedfromthe 40K mockcata- obtainedfromthealgorithmdiscussedinSect.2.4andwhichwe logs. describeinmoredetailinAppendixA. Our algorithmachievesthis remarkableresultwhileadding anegligiblefractionofinterlopermembers,i.e.,galaxiesthatdo 4.1.Groupcentering notsharethe same darkmatterhalo in the 40Kmock catalogs. Forhalfofthegroupsweaddedlessthan3%ofnewinterlopers A goodgroupcenter definitionis essentialto our scienceanal- withrespecttothetotalmembersinthe20K+photo-zscatalog, ysis, because we will be searchingforradialtrendsthatcan be thereforeattaining the same interlopersfraction as in the spec- easily erased by errorsin groupcentering.After addingphoto- troscopicgroupcatalog. zs, as discussed in the previous section, 50% of the groups in We also checkedforanydependenceof the interloperfrac- oursamplepossessmorethanninememberswhichmakesgroup tiononthenormalizedgroup-centricdistanceR /R ,R center definition more robust. However the simple methods of gal fudge fudge beingan estimate of the virialradiusprovidedby Knobeletal. estimating group centers, such as the median of members co- (2011 see Sect. 4.2 for details). For this test we divided each ordinates,provideonlyroughestimatesofthegroupcenter,es- group into a central part and two concentric intermediate and pecially for the numerically poorer groups. We therefore tried 5 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample Fig.2.Summaryoftheresultsobtainedwithouralgorithm.Topleft:completenessdistribution(seetextfordefinition)forthe20Kmocks(black dot-dashedline)andforthe20K+photo-zsmockgroups(redsolidline).Topright:fractionofinterlopersasafunctionofnormalizedgroup-centric distanceforthe20K/20K+photo-zsmockgroups(blackstarsandredtriangles,respectively).Orangesquaresrefertothefractionofinterlopers, PI,forrealdata,ascalibratedonthemocks,seetextfordetails.Bottomleft:fordifferentmocksasindicatedonx-axis,themediandistancetothe centralgalaxyoftheVWcenter(violettriangles)andthemediancenter(cyanstars).Bottomright:DistributionofthedistanceoftheVWcenter tothecentralgalaxypositionforthe20Kmocks(blackdot-dashedline)andforthe20K+photo-zsmocks(redsolidline). an alternativestrategy,taking into accountsky-projectedgroup We used our set of mock catalogs to test the advantagesof galaxydensities. this center definitionwith respectto simpler ones,like the me- Using the 2D-Voronoiareas as proxy for density measure- dian of the member galaxies coordinates (median center from ment,wedefinedtheVoronoi-weightedcenter(VWcenterfrom nowon).We assumedthepositionofitscentralgalaxyasfidu- nowon)as cialcenterforeachgroup,asprovidedbythemocks. ra = PiN=1rai/AV,i, dec = PiN=1deci/AV,i. (2) In the bottom left panel of Fig.2 we show the median dis- VW N 1/A VW N 1/A tanceoftheVWcenter(violettriangles)andthatofthemedian Pi=1 V,i Pi=1 V,i center (cyan stars) for each of the three mock catalogs defined where AV,i is the 2D-Voronoi area associated to the i-th in Sect.2.4. Error bars show the rms among mock catalogs ex- galaxymember,thatis, the projectedareacontainingallpoints tracted from the different24 light-cones.We note that the VW closer to the i-th galaxy than to any other member galaxy. We centerprovidesabetterestimateofthecenterwithrespecttothe usedgalaxieslocatedoutside3×Rgr (whereRgr istheradiusof mediancenter onaverage.Furthermore,the VW centers, when the minumumcircle containingall groupmembers)andwithin appliedtothegroupswhithphoto-zsaddedusingourouralgo- 1×σzphot toavoiddivergenceof2D-Voronoiareasforgalaxies rithm,arenearlyindistinguishablefromthoseobtainedwhenall locatedattheperipheryofgroups. membersdowntoI =22.5possessspectroscopicredshift.The AB This way galaxies that are located in group denser regions medianvalueofthedistanceoftheVWcentersfromthegroup willhavea smallerAV,i andtheywill weighmore,while those centralgalaxyis40h−1 Kpc for20K+photo-zsmocks,withan thatareinlessdenseregionswillaffectthecenterdetermination improvementofnearly7040%incenteringwithrespecttothe20K less.Thismethodthusprovidesacenterforthegroup,whichis mocks. locatedbydefinitionintheareaofgreatestgalaxyover-density, and is not affected by the details of the spatial distribution of InthebottomrightpanelofFig.2weshowthedistancehis- galaxies at the outskirt. For a similar approach see Diazetal. togram of the VW center to the central galaxy position for the (2005). 20Kmocksasablackdot-dashedline andforthe 20K+photoz 6 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample mocksasa redsolid line.Theimprovementin groupcentering Vice versa in the external regions of groups the contamination obtainedwhenaddingphoto-zmembersisquiteobvious. byfieldgalaxies,onaveragebluerandlessmassivethangroup WethereforeadoptedtheVWmethodtodefinethecenterof galaxies,willtendtointroducespurioussegregationtrends,and eachgroup. weneedtoaccountforthemcarefully. 4.2.Grouprescaling 5. Analysis Theprocedureofstackinggroupsofdifferentsizesandmasses intoanensemblesystemrequiresrescalingofindividualgalaxy 5.1.Selectingmassvolume-limitedsamples group-centric-distances.Instudiesofgalaxyclusters,projected cluster-centric-distances R are generally rescaled with R or Wefocusedouranalysisontworedshiftbins:0.2≤z≤0.45and vir R , whose estimate is proportionalto cluster velocity disper- 0.45 < z ≤ 0.8,where we definedtheclassical volume-limited 200 sion σ , that is, a proxy of cluster mass (Carlbergetal. 1997; samplestakingintoaccounttheluminosityevolutionofindivid- v Bivianoetal. 2002;Katgertetal. 2004).However,the problem ualgalaxies.FollowingZuccaetal.(2009),weadoptedalinear isnottrivialwhendealingwithgalaxygroups,wheretheuncer- evolution with redshift: MB∗ev = −20.3− 5 log h70 − 1.1 z to taintiesintheestimateofvelocitydispersions,masses,size,and parametrizetheevolutionofMB∗ oftheluminosityfunction.The the group dynamical state in general are larger, because of the correspondingevolvingcut-offmagnitudesareMcut−off = MB∗ev+ smallnumberofgroupmembers. 2.1(+0.8)forthelow(high)redshiftbin.For0.2 ≤ z ≤ 0.45the IntheliteraturetherearedifferentapproachesinrescalingR volume-limitedsample consistsof829outof 1128totalgalax- forgroups,using1)thevirialradiusR orR ,2)anestimateof ies,belongingto79groups.For0.45<z≤0.8itconsistsof510 vir 200 thermsofthepositionofmembergalaxiesR ,and3)sometimes out of 660 total galaxies, belonging to 64 groups (see Tab.1). H radialdistancesarenotrescaledatall(seeGirardietal.2003for The total volume-limited field sample consists of 1869(2893) adetailedreview). galaxies,whilethenear-fieldvolumelimitedsampleconsistsof Ourgroupsspanawiderangeofsizes,andthereforearescal- 683(612)galaxiesforthelow(high)redshiftbin.Intheleftpanel ing of physical distances seemed unavoidable. We decided to of Fig.3 we show the MB∗ versusredshiftdistributionof the to- useR asthescalingfactor,providedbyKnobeletal.(2011). tal galaxysample (blackpoints) and thatof both spectroscopic fudge Thisfudgequantity,asmanyotheronescorrelatingwiththeob- and photometric redshift group member galaxies (red points). served group richness, was estimated and calibrated using our Thecyansolidlineandthevioletdashedlinecorrespondtothe realistic mock catalogs. In brief, given an observed group at magnitudecut-offsdefiningthelow-andhigh-redshiftvolume- redshift z with richness N, defined as the number of members limitedsamples.Inthefollowing,thegrouprichnessN foreach brighterthananadoptedrest-frameabsolutemagnitudecut-off, groupisdefinedasthenumberof(phot+spec-z)membergalax- its R correspondsto the mean R amongallreconstructed iessurvivingtothemoreconservativeabsoluterest-framemag- mockfudggreoups wjth the same N andvriredshift (see Knobeletal. nitudecut-off:MB∗ev+0.8,unlessexplicitelystated.Thisquantity 2011 for more details on how this quantity is calculated). The correlates,albeitwith a largescatter, with the mass of the halo quantityR correlateswithM ,anestimateofthemass wherethegroupresidesandthereforeisagoodproxyforit(see fudge halofudge ofthegroupwell,whichadditionallyshowsitsrelevanceforour Knobeletal.2009). analysis (see Knobeletal. 2009, 2011for more detailson how The flux-limited target definition of zCOSMOS-bright, boththesequantitiesareestimated). IAB ≤ 22.5,translatesintoaB-bandrest-frameselectionatz ∼ Because our goal is to distinguish property of galaxies lo- 0.8.Thereforethe20K galaxysample,whenrest-frameB-band cated in regions with different physical properties rescaling by selectionisadopted,isfreefromsignificantcolor-dependentin- R ,aquantityrelatedtoR ,suitsourneedswell.Indeed,the completenessin(U−B)rest-framecolorsuptoz∼1.However fudge vir virialradiusisa scalingfactorformanytimescalesofdifferent the (U − B) rest-frame color completeness in the B-band rest processes such as the crossing time, the relaxation time or the frameselectiondoesnotimplycompletenessinmassselection: mergingtime(Boselli&Gavazzi2006;Weinmannetal.2006). theB-bandrest-frameselectionisbiasedtowardblue,low-mass Allgalaxiesthatareinsidethevirialradiusareexperiencingthe galaxies,whilemissingthecorrespondingred,equallylow-mass group potential effects either for the first time or many times. ones. Environmentaltrendsobservedin samplesselected using In contrast, those galaxies that are outside the virial radius are rest-frameB-bandmagnitudescouldthereforebesimplythere- amixedpopulationofbothin-fallinggalaxiesandgalaxiesthat sultsofthisincompletenesscoupledwithdifferentgalaxymass oncepassedthroughthevirialradiusbutnowareintheoutskirts, distributionsindifferentenvironments(Bolzonellaetal.2010). theso-calledback-splashpopulation(Gilletal.2005). To separate true environmental effects from mass-driven Beforestackinggroups,wethereforerescaledeachmember ones,weusedinouranalysismassvolume-limitedsamples,that galaxydistance to the VWcenter,R , with the corresponding is, samples complete down to a fixed galaxy mass cut-off. To gal R ofitsgroup.Belowwewilluseonlyscaleddistances,R, obtainthem, we followedthe same approachasin Iovinoetal. fudge unlessotherwisespecified. (2010).Inbrief,wefirstcalculatedthelimitingstellarmassfor Weaddafinalcaveat:whendiscussingourresults,weshould each galaxy in the 20K sample, i.e., the stellar mass it would take into account projection effects. We observed the 2-D pro- haveatitsspectroscopicredshift,ifitsapparentmagnitudewere jection of a 3-D distribution of member galaxies. Assuming a equaltothelimitingmagnitudeofoursurvey:log(M (z ))= lim gal spherical distribution, this implies that e.g., the inner observed log(M )+0.4(I −22.5).Wethenusedtheseestimatedlim- gal AB regionincludesgalaxieslocatedintheoutergroupshellsthatare iting masses to define, in bins of (U − B) rest-framecolors for locatedalongthelineofsightofthegroupcentralpart.Hencea eachredshiftbin,themassM belowwhich85%ofgalaxies cut−off fractionof galaxiesobserved,in projection,in the innerregion of that colorlie. We fitted M to obtain a color-dependent cut−off actually belongs to the outskirts. These projection effects will masslimitcut-off.ThevalueofM forthereddestgalaxies cut−off tendto smooththeradialtrendswearelookingfor,so thatany ineachredshiftbinistheonethatweusedasthelimitingmass observedtrendisalowerlimitfortherealtrendpresentin3-D. forthatbin. 7 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample Fig.3.Leftpanel:redshiftdistributionofthezCOSMOS-brightgalaxies(black).Redpointsrepresentgroupmembergalaxieswith spectroscopicandphotometricredshiftsasobtainedfromouralgorithm.Thecyansolidlineandthevioletdashedlinecorrespond tothetwodifferentmagnitudecut-offsadoptedtodefinethevolume-limitedsamplesofthelow-andhigh-redshiftbinrespectively, seetextfordetails.Centralandrightpanel:(U −B)rest-framecolorversusmassforthelowestandthehighestredshiftbin.The bluedashedlinecorrespondstothecolor-dependentM ,whiletheredsolidlinecorrespondstothefixedM forourmass cut−off cut−off volume-limitedsample.Thereddot-dashedlineshighlightthemassrangesadoptedinthemass-segregationanalysis,seeSect.6.5 fordetails.Thecyandottedlinecorrespondstotheseparationbetweenredandbluegalaxies(seetextforitsprecisedefinition). InthecentralandrightpanelofFig.3weshowthe(U −B) Table 1. Number of volume-limited and mass-volume-limited rest-framecolorversusthestellarmassforthelowestandhigh- (spec+phot-z)groupmembergalaxies.Inbracketswereportthe est redshift bin respectively. The blue dashed line shows the number of spectroscopic-only group members. The number of color-dependent M , while the red solid line shows the groupscontainingthesegalaxiesislistedinthelastcolumn. cut−off valuechosentodefinemass-limitedsamples:log(M /M ) ≥ gal ⊙ M = 9.8 and log(M /M ) ≥ M = 10.56 for the Redshift Vol-lim Vol-Mass-lim cut−off gal ⊙ cut−off lowestandhighestredshiftbins. Ngals Ngals Ngr To define the mass-dependentcolor cut separating the blue 0.2≤z≤0.45 829(570) 571(410) 79 0.45<z≤0.8 510(391) 265(200) 64 andredgalaxies,weperformeda robustfitoftheredsequence asafunctionofthegalaxystellarmassinthehigh-zbin,wherea largenumberofobservedgalaxiesdisplaysaprominentandwell definedredsequence.Thecolorcutwasthenobtainbyshifting Before moving to a detailed study of the group member thefittinglineby2·rms ,whererms ∼ 0.08isthedisper- galaxiesproperties,itisinterestingtocomparethegeneralprop- red red sionoftheredgalaxiesalongtheredsequence.Weadoptedthe ertiesofgroupsinlow-andhigh-redshiftbins,tohighlightany same color cut for the low-z bin. Numerically,the stellar mass redshift-dependenttrendinthegroupsampleweusedinoursci- dependentcolorcutis enceanalysis. In Fig.4 we compare from left to right R - the virial (U−B)=0.094·log(M /M )+0.05, (3) fudge gal ⊙ radius estimate (Knobeletal. 2011), M - the mass of the fudge anditisshownbythecyandottedlinesinFig.3. groupcalibratedwiththemocksasinKnobeletal.(2009),and We tested thatourresultsdo notchangeif we applya con- thegrouprichness,N,asdefinedinsection5.1,forlow-(black stantcolorcut,(U −B)=1,toseparateredandbluegalaxies,a dot-dashedline)andhigh-(redsolidline)redshiftgalaxygroups. simplerdefinitionthatcorrespondsequallywelltothedipofthe The KS test always rejects with more than 99.99% confidence bimodaldistribution. thehypothesisthatpropertiesoflowandhighredshiftgroupsare For the lowest redshift bin the final group mass-complete drawnfromthesamedistribution.Inthelow-redshiftbinonthe samplecontains571galaxies,whileforthehighestredshiftbinit meanwedealwithsmaller,lessmassiveandpoorergroupsthan contains265galaxies.Themass-completefieldsamplesconsist thoseinthehighest-redshiftbin.Thisisnotanunexpectedresult of 743(728)galaxiesforthe lowest(highest)redshiftbin, while giventhatzCOSMOSisaflux-limitedsurveyandthereforethe thenear-fieldsamplesconsistof293(211)galaxiesforthelow- observedpopulationofbothgalaxiesandgroupsvarieswithin- est(highest)redshiftbin. creasingredshift.Asaconsequence,thegroupdetectionworks onlyonprogressivelybrighter/moremassivegalaxiesmovingto higherredshifts.Weshallneedtotakeintoaccountthesediffer- 5.2.Low-zandhigh-zstacked-groups enceswhendiscussingourresults. We definea subsetofricher Foreachofthetworedshiftbinsdefinedintheprevioussection, groupsforthelow-redshiftbinusingrichnessN,definedforthis 0.2 ≤ z ≤ 0.45 and 0.45 < z ≤ 0.8, we proceeded to build bin as the number of member galaxies surviving the evolving the correspondingSG. Notice thatwhile forcenteringpurposes magnitude cut-off: M = M∗ + 2.1 (see Sect. left panel cut−off Bev we used all spec-z and phot-z galaxies available in our group of Fig.3). We adopted a separation of N ≤ (>)12 to distin- catalog,irrespectiveof their mass andB-band rest-framelumi- guishbetweenpoor(rich)groups,avalueroughlycorresponding nosity,forouranalysiswewilluseonlygalaxieswithinthemass toM ≤(>)13.3,sothatrichlow-zgroupsarevirtuallyindis- fudge volume-limitedsamplesasdefinedinTable1. tinguishableinmassdistributionfromthehigh-redshiftsample. 8 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample Fig.4.Comparisonofthegeneralpropertiesofgroupsinthe0.2 ≤ z ≤ 0.45(blackdot-dashedline)and0.45 < z ≤ 0.8(redsolid line)redshiftbin.Fromlefttorightwe compareR (theestimateofthevirialradius),logM (theestimateofthemassof fudge fudge thegroup),bothfudgequantitiesarecalibratedwiththemocksasdefinedinKnobeletal.(2009)),andN,asdefinedinSect.5.1. Indeed, while a KS test comparing the distributions of M 6. Results fudge of poor and rich groups defined this way rejects the hypothe- sis that they are drawn from the same distribution with more We will start our analysis by exploring how galaxy colors are than99.99%confidence,theKStestcomparingdistributionsof affected by group environment, irrespective of galaxy position M ofrichlow-zgroupsandofhigh-zgroupsdoesnotreveal withinthegroup(seeSect.6.1).Wewillthenmovetoinvestigate fudge anysignificantdifferencebetweenthetwo. thepresenceofcolorsegregationwithinthegroupenvironment (Sect.6.2),andiftheeffectofthegroupenvironmentextendsto scales somewhat larger than those of the group size itself (see To explore how galaxy population properties change as a function of group-centric distance, we first sorted SG galaxies Sect.6.3).Thankstothehighstatisticofthe20Kwewillalsobe intoincreasingscaleddistancesfromSG centerandthendivided abletoinvestigateifandhowobservedtrendsdependongroup richness and on galaxy stellar mass (see Sect.6.4). Finally we theirdistributionintothreeequipopulatedbinscorrespondingto inner,intermediate,andperipheralSG regions. will search for evidenceof mass segregationinside groupsand howitmightdependongrouprichnessandaffectobservedmass trends(seeSect.6.5). In Table 2 we list the exact radial ranges of each of these three regions, all values are normalized to R . The corre- fudge sponding three median distances are R ∼ 0.15, R ∼ 0.4 and 6.1.F andgalaxystellarmassesingroupsvsfield R∼0.85,respectively,thereforetheseregionscanbeconsidered blue as the groupinner core,intermediate,and moreexternal/in-fall The cumulative galaxy stellar mass distributions of the mass- region. complete group and field samples are shown in the top pan- els of Fig.6, red solid and cyan dot-dashed lines, respectively, Given that the median R is ∼ 500 h−1 Kpc in both the left(right) panels refer to the low(high) redshift bin. The fudge 70 redshift bins, the inner region extends typically up to ∼ 150 KS test rejectsthe hypothesisthat groupand field galaxymass h−1 Kpc. Because the VW center is on the average only ∼ 40 distributions are drawn from the same population with more 70 h−1 Kpc awayfromthegroupfiducialcenter(seeSec.4.1),our than 99.99% confidence for both redshift bins. Group envi- 70 errorincenteringisnegligiblewithrespecttothemedianinner ronment hosts preferentially more massive galaxies than the region size, and should not have a significant impact when ex- fieldone,confirmingwell-knownliteratureresults(Iovinoetal. ploringthegroup-centricdependenceofgalaxyproperties. 2010;Kovacˇetal.2010;Bolzonellaetal.2010). Asaconsequence,toexplorethepresenceofcolortrendsas Theskydistributionofgalaxiesbelongingtothelow-z(left) a function of environment,we need to separate the joint effect andhigh-z(right)compositegroupisshowninFig.5.Pointsare ofmassandenvironmentandtoperformtheanalysisinnarrow coded according to the (U − B) colors of the galaxies, while mass bins of galaxy stellar mass. We adopted a galaxy stellar point dimensions are scaled according to galaxy masses. As a mass bin of 0.4 dex,which is approximatelytwice our errorin referencewedrawdashedcirclescorrespondingtothe division estimating galaxy stellar masses (Pozzettietal. 2010). Bottom betweenthedifferentregionsineachcompositegroup.Wenote panelsofFig.6showF atfixedgalaxystellarmassingroups blue thattheoverallshapeofthecompositegrouphasawell-defined (redcircles)andfield(cyanstars),whileTable3listsindetailthe peakcorrespondingtothecenter,whiletheprojecteddensityde- F valuesand theirerrors.Both at highandlow redshift,the blue creases as we move from the center to the outskirts. A visual bluefractionincreaseswhenmovingtowardlessmassivegalax- inspection of the galaxy sky distribution already shows rough ies.F isalwayshigherinthefieldthaninthegroup,adiffer- blue differencesin masses and colors dependingon the area we ex- encethatdecreasesmovingtomoremassivegalaxies.Themost plore.Inthenextsectionweproceedtoextensivelyanalyzethese massivegalaxies(log(M /M ) > 11.0)donotshowanysig- gal ⊙ trendsandtheirdependenceonintrinsicgalaxy/groupproperties, nificant F evolution with redshift within the error bars. For blue properlyaccountingforpossiblefieldcontaminationeffects. thegalaxieswith10.6≤ log(M /M ) ≤ 11.0,F decreases gal ⊙ blue 9 Presotto,V.etal:SegregationeffectsinthezCOSMOS-20Kgroupsample Fig.5. Sky distribution of the galaxies belonging to the low-z (left) and high-z (right) composite group. Ra-dec positions are expressed in terms of the rescaled distances R. Points are colored according to the (U − B) colors of the galaxies, while point dimensionsarescaledaccordingtothemassesofthegalaxies.Asareferencewedrawdashedcirclescorrespondingtothedifferent central/intermediate/externalregionlimitsineachcompositegroup. Table2.RadialrangeexploredinthethreeSG regions.AlldistancesRarenormalizedtoR . fudge Redshift 1st region 2ndregion 3rd region range R range R range R median median median 0.2≤z≤0.45 R≤0.30 0.15 0.30<R≤0.68 0.47 R>0.68 0.94 0.45<z≤0.8 R≤0.23 0.13 0.23<R≤0.51 0.37 R>0.51 0.74 forbothgroupandfieldenvironmentwhenmovingfromhighto Table3.Observedbluefractionsingroupsandfieldfordifferent lowredshift. galaxystellarmassbins. Notice that on average, groups in the low-z bin are poorer thanthoseinthehigh-zbin.Aswewillshowbelow,seeSect.6.4, Sample0.2≤z≤0.45 group field F dependsonthegrouprichness,F beinglowerinricher blue blue 9.8≤log(M /M )≤10.2 0.33+0.03 0.70+0.03 groups. As a consequence,the decrease of Fblue across the ex- gal ⊙ −0.03 −0.03 ploredredshiftrangeforthesampleofgroupgalaxiesshouldbe 10.2≤log(M /M )≤10.6 0.18+0.03 0.49+0.03 evenmorepronounced. gal ⊙ −0.03 −0.03 At fixed galaxy stellar mass, the migration to the red se- 10.6≤log(M /M )≤11.0 0.22+0.03 0.32+0.03 gal ⊙ −0.03 −0.03 quencehappensearlierinthegroupsandlaterinthefield,sug- gestingthepresenceofphysicalmechanismsabletoremovegas 11.0≤log(M /M )≤11.4 0.11+0.04 0.25+0.07 gal ⊙ −0.03 −0.06 that causes the earlier quenchingof galaxiesin groups. We re- mindthereaderthatanycontaminationofthegroupsampleby Sample0.45<z≤0.8 group field fieldgalaxiesandviceversa,forwhichwearenotapplyingany correction,willonlyrendertheobservedtrendslessprominent. 10.6≤log(Mgal/M⊙)≤11.0 0.28−+00..0034 0.41−+00..0022 The real, corrected, trends therefore would be even more pro- 11.0≤log(M /M )≤11.4 0.11+0.04 0.23+0.03 nounced.ThisresultexcellentlyagreeswithourpreviouszCOS- gal ⊙ −0.03 −0.03 MOS results on groups (Iovinoetal. 2010; Bolzonellaetal. 2010;Pengetal.2010). Thequestionswewilladdressinthefollowingsectionsare: dothegroupmembergalaxiesallsharethesame F valueir- blue respectiveoftheirpositionwithinthegroup?Dothegalaxieslo- catedinthecentralregionofgroupssharethesamemassdistri- butionasthegalaxieslocatedinthegroupoutskirts?Ideally,the firstquestionisbetteraddressedinnarrowbinsofgalaxystellar statistics,whensplittingoursampleaccordingtodistancesfrom mass to avoid the mass-color degeneracy. However, even with groupcenterandthereforewewillstartouranalysisworkingin suchadatasetasthe20K,we arestilllimitedbysmallnumber cumulativemassrangesinthenextsection. 10

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.