Identifying the progenitor set of present-day early-type galaxies

14 Pages·2009·1.29 MB·English

A&A503,445–458(2009) Astronomy
DOI:10.1051/0004-6361/200810483 &
(cid:2)c ESO2009 Astrophysics
Identifying the progenitor set of present-day early-type galaxies:
a view from the standard model
S.Kaviraj1,2,J.E.G.Devriendt2,3,I.Ferreras1,S.K.Yi4,andJ.Silk2
1 MullardSpaceScienceLaboratory,HolmburySt.Mary,Dorking,SurreyRH56NT,UK
e-mail:[email protected]
2 DepartmentofPhysics,UniversityofOxford,KebleRoad,OxfordOX13RH,UK
3 ObservatoireAstronomiquedeLyon,9AvenueCharlesAndré,69561Saint-GenisLavalCedex,France
4 CenterforSpaceAstrophysics,YonseiUniversity,134Shinchon,Seoul120-749,Korea
Received30June2008/Accepted9June2009
ABSTRACT
Wepresentacomprehensivetheoreticalstudy,usingasemi-analyticalmodel withinthestandardLCDMframework,ofthephoto-
metricpropertiesoftheprogenitorsofpresent-dayearly-typegalaxiesintheredshiftrange0 <z<1.Weexploreprogenitorsofall
morphologiesandstudytheircharacteristicsasafunctionoftheluminosityandlocalenvironmentoftheearly-typeremnantatz=0.
Inagreement withprevious studies, wefindthat,whilelarger early-typesaregenerallyassembled later,theirluminosity-weighted
stellar agesaretypicallyolder.Indense cluster-likeenvironments, ∼70percent ofearly-type systemsare“inplace”byz = 1and
evolvewithoutinteractionsthereafter,whileinthefieldthecorrespondingvalueis∼30percent.Averagingacrossallenvironmentsat
z∼1,lessthan50percentofthestellarmasswhichendsupinearly-typestodayisactuallyinearly-typeprogenitorsatthisredshift,in
agreementwithrecentobservationalwork.Thecorrespondingvalueis∼65percentinclusters,duetofastermorphologicalevolution
insuchdenseenvironments. Wedevelopprobabilisticprescriptionswhichprovideameansofincludingspiral(i.e.nonearly-type)
progenitorsatintermediateandhighredshifts,basedontheirluminosityandopticalcolours.Forexample,wefindthat,atintermediate
redshifts(z ∼ 0.5),large(M < −21.5),red(B−V > 0.7)spiralshave∼75–95percent chanceofbeinganearly-typeprogenitor,
V
whilethecorrespondingprobabilityforlargebluespirals(M <−21.5,B−V <0.7)is∼50–75percent.Theprescriptionsdeveloped
B
here can beused to address, from theperspective of thestandard model, theissue of “progenitor bias”, whereby theexclusion of
late-typeprogenitorsinobservationalstudiescanleadtoinaccurateconclusionsregardingtheevolutionoftheearly-typepopulation
overcosmictime.Finally,weexplorethecorrespondencebetweenthetrue“progenitorset”ofthepresent-dayearly-typepopulation
–definedasthesetofallgalaxiesthatareprogenitorsofpresent-dayearly-typesregardlessoftheirmorphologies–andthefrequently
used“red-sequence”,definedasthesetofgalaxieswithinthepartofthecolour–magnitudespacewhichisdominatedbyearly-type
objects.Wefindthat,whilemoremassivemembers(M ≤ −21)ofthe“redsequence”tracetheprogenitorsetreasonablywell,the
V
relationshipbreaksdownatfainterluminosities(M ≥ −21).Thus,whiletheresultsofrecentobservational studieswhichexploit
V
theredsequencearevalid(sincetheyarelargelyrestrictedtomassivegalaxies),morecareshouldbetakenwhendeeperobservations
(whichwillprobefainterluminosities)becomeavailableinthefuture.
Keywords.galaxies:ellipticalandlenticular,cD–galaxies:evolution–Galaxy:formation–galaxies:fundamentalparameters
1. Introduction Ferreras et al. 2005).However,a fundamentalfeature of early-
typeformationinthestandardLCDMmodelisthatstellarmass
As“endpoints”ofgalaxymergersequences,early-typegalaxies
that eventually ends up in present-day early-type galaxies is
carryimportantsignaturesofmassassemblyandstarformation
not entirely contained in early-type systems at high redshift.
in theUniverse.Deducingtheirstarformationhistories(SFHs)
Although,early-typesatanyredshiftarealmostguaranteedpro-
thereforecontainsthe keyto understandingnotonly theevolu-
genitors of their counterparts at present-day, looking only at
tion of these galaxies but the evolutionary patterns of galaxies
early-typesathighredshiftintroducesa“progenitorbias”,which
asawhole.Ourviewofearly-typegalaxyformationhasdevel-
becomes increasingly more severe at larger look-back times,
oped over the years, away from the classical “monolithic col-
as the fraction of early-type galaxies becomes progressively
lapse”hypothesis(e.g.Larson1975;Chiosi&Carraro2002)and
smaller andlate-typesystemsbeginto dominatethe progenitor
towardsthehierarchicalassemblyoftheseobjectsthroughmerg-
population(e.g.Franx&vanDokkum1996;vanDokkumetal.
ersandaccretionofsmallergalaxiesovertime,intheframework
2000;vanDokkum&Franx2001;Kavirajetal.2005).
of the currently popular LCDM paradigm of galaxy formation
(e.g.Kauffmannetal.1993,1996;Baughetal.1996;Kauffmann In the current era of large scale surveys, e.g. SDSS
&Charlot1998;Somerville&Primack1999;Coleetal.2000; (Adelman-McCarthy et al. 2006), COMBO-17 (Wolf et al.
Hattonetal.2003;DeLuciaetal.2006;Boweretal.2006,and 2004),MUSYC(Gawiseretal.2006),GEMS(Rixetal.2004),
referencestherein). unprecedentedamountsofdataspanningalargerangeinredshift
A significant body of observational work in the past has (typically0 < z < 1) andenvironmentare becomingavailable,
traced the assembly histories of early-type galaxies by study- allowing us to study statistically significant numbers of galax-
ing only early-type populations at high redshift (e.g. Gladders iesatvariousstagesofevolution.Aquantitativestudyofearly-
et al. 1998; Stanford et al. 1998; Barrientos & Lilly 2003; typeprogenitorswithinthestandardmodelisthereforedesirable
ArticlepublishedbyEDPSciences
446 S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel
to: (a) understandthe difference in the way early-typesare as- includingspecificpartsofthespiralpopulationathighredshift
sembledasafunctionoftheirluminosityandenvironmentand; intothestudyofearly-typeevolution.
(b)togaugetheroleofnon-early-typeprogenitorsinearly-type Theplanofthispaperisasfollows.InSect.2,wedescribe
evolution,especiallyforstudiesthatfocusathighredshift. the salient features of the model used in this study. In Sect. 3,
Acentralthemeofthisworkistousea modelwhichaccu- we quantifythe morphologiesof the galaxiesthat make up the
ratelyreproducesthephotometricpropertiesofearly-typegalax- progenitorset. We map the propertiesof elliptical, S0 and spi-
iesandtheirobservedevolutionoveralargerangeinredshiftto ralprogenitorsasa functionofredshiftandexploredifferences
study, in detail, the properties of the progenitors of early-type betweenprogenitorsasafunctionofthemassandenvironment
galaxiesat presentday.The processofmappingthe progenitor of theellipticalremnantatpresentday.InSect. 4 wefocusex-
populationanditsevolutionwithredshiftalsoprovidesarealis- clusively on spiral progenitors and compare their photometric
tic pictureofprogenitorbiaswithinthe frameworkof thestan- propertiestothegeneralspiralpopulation,toderiveprobabilis-
dardmodel. tic prescriptions for including spiral progenitors in early-type
studies. Section 5 traces the contribution of galaxies in dense
While van Dokkum & Franx (2001) have developed and
regions at high redshift to cluster early-types at present-day.
studied the issue of progenitorbias, their study employedphe-
Finally, in Sect. 6 we explore the correspondencebetween the
nomenologicalSFHs, with thesimple (andperhapsunrealistic)
true progenitor set of present-day early-type galaxies and the
assumption that morphological transformations occur abruptly
∼1.5 Gyr after the cessation of star formation in a particu- “red-sequence”,whichissometimesusedasaproxyforthepro-
genitorsetin observationalstudies(e.g.Bell etal. 2004;Faber
lar galaxy.Their work was an extension of previous ideas (e.g
etal.2007).
Boweretal.1998;Shioya&Bekki1998)thattookintoaccount
Notethatthroughoutthisstudyweproviderest-framemag-
the potentialforellipticalgalaxiestohavecomplexstar forma-
tion histories, but did not explore the effect of morphological nitudesforallmodelgalaxies.Unlessotherwisenoted,thefilters
usedareinthestandardJohnsonsystem.Notethatcolourfigures
transformationsontheevolutionoftheearly-typepopulationat
areavailableintheonlineversionofthispaper.
highredshift.
While some recent observationalstudies (e.g.Holden et al.
2005; van de Ven et al. 2003; van Dokkum & van der Marel 2. Themodel
2007) have used the results of van Dokkum & Franx (2001)
to take the effects of progenitor bias into account, other stud- Thesemi-analyticalmodelusedinthisstudyisGALICS,which
combineslarge-scalecosmologicalN-bodysimulationswithan-
ies (Blakeslee et al. 2003) have used the entire galaxy popula-
alytical recipesforthe evolutionof baryonswithindark matter
tion, without reference to morphology,in an attempt to reduce
haloes(Hattonetal. 2003).GALICSmakespredictionsforthe
the bias in their conclusions that would occur if only elliptical
overallstatisticalpropertiesofgalaxypopulations,withanem-
galaxieswereusedintheanalysis.
phasisontheirpanchromaticspectralenergydistributionsacross
OurstudyrefinesandextendstheresultsofvanDokkum& awidewavelengthrange(UV toinfrared/submillimetre).Inthis
Franx (2001) by studying early-type progenitorswithin a real-
sectionwe describethesalientfeaturesofthemodel,including
istic and well-calibrated semi-analytical framework, in which
theimplementationofthecosmologicalsimulationanddetailsof
mass assembly and morphological transformations can be fol-
thesemi-analyticsthatdeterminethegrowthofbulges.Wealso
lowed more accurately in the context of the LCDM paradigm.
highlightthesuccessesofGALICSintermsofreproducingthe
WhilethevanDokkum&Franx(2001)studydoesattempttoal-
photometricpropertiesofearly-typegalaxiesintheUV-optical
leviatetheeffectsofprogenitorbias,itssimplicity,especiallyin
spectralrangesacrossawideredshiftrange.
theprescriptionusedformorphologicaltransformations,makes
Before we begin our analysis we briefly describe the def-
amorerealistictreatmentofthisissue,intheframeworkofthe
inition of galaxy morphology in the model used in this study.
standardmodel,verydesirable.Usingtheentiregalaxypopula-
Galaxy morphologyin the modelis determinedby the ratio of
tion withoutreference to morphologyis also not ideal because
theB-bandluminositiesofthediskandbulgecomponentswhich
thatimpliesthatallgalaxiesathighredshiftarepotentialprogen-
correlates well with Hubble type (Simien & de Vaucouleurs
itors of present-dayearly-typegalaxies.While this approxima-
1986).Amorphologyindexisdefinedas
tion may be reasonably robust for the most massive early-type
(cid:2) (cid:3)
galaxiesinthedensestregionsoftheUniverse,ourstudyshows −L
thatthisisgenerallynotapplicable. I =exp B (1)
L
D
In thispaperwe study the evolutionof the “progenitorset”
ofearly-typegalaxies,definedasthesetofallgalaxiesatagiven such that a pure disk has I = 1 and a pure bulge has I = 0.
redshiftthatare,regardlessofmorphology,theprogenitorsofan Following Baugh et al. (1996), ellipticals have I < 0.219, S0s
early-typeatz=0.Notethat“early-types”aredefinedasgalax- have0.219< I <0.507andspiralshaveI >0.507.
ies that are elliptical or lenticular. See Sect. 2 for a description
of how galaxy morphologiesare defined in the model. We ex-
2.1.Darkmattersimulation
ploretheevolutionoftheprogenitorsetwithredshift,asafunc-
tionoftheluminosityandenvironmentoftheearly-typeremnant The dark matter “backbone” used in GALICS has been gen-
whichisleftatpresent-day.Wepayparticularattentiontospiral erated by simulating a LCDM model in a cube with a co-
progenitorsinthemodel,sincethesemighthavebeenexcluded moving size of 100 h−1 Mpc with parameters Ω = 0.33,
m
fromsomestudiesofearly-typeevolutioninthepast,byvirtue ΩΛ = 0.667and σ8 = 0.88.The amplitudeof the powerspec-
oftheirmorphology,eventhoughtheyformanimportantpartof trumiscomputedusingthepresent-dayabundanceofrichclus-
theprogenitorset.Bycomparingtheproperties(opticalcolours ters (Eke et al. 1996)and initial conditionsare extracted using
andluminosities)ofspiralprogenitorstothegeneralspiralpop- theGRAFICcode(Bertschinger1985).Thesimulationcontains
ulation, we provide a means of correcting for progenitor bias, 2563darkmatterparticles.Themassofeachdarkmatterparticle
whichisconsistentwiththepropertiesofthestandardmodel,by is∼8×109 M(cid:6) andthespatialresolutionachievedis∼29.3kpc.
S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel 447
has acquired half its bulge mass during mergers and the other
halfthroughdiskinstabilities.Notethatweonlyconsidergalax-
ies with a bulge component. Pure disk galaxies (i.e. those that
haveneverbeenunstableorundergoneamerger)representonly
∼1.4percentofthebrightpopulation.
Severalimportantconclusionscanbedrawnfromthisfigure.
First, it is clear that the majority of elliptical galaxies develop
their bulges during mergers rather than disk instabilities. This
result may seem natural, as GALICS is an implementation of
the hierarchical galaxy formation scenario. However, the main
point here is that, for the first time, we do indeed compare the
relativecontributionofmergerstootherprocessesintheforma-
tionofspheroids.Thefactthatmodelellipticalgalaxiesdevelop
theirbulgesmainlythroughmergersratherthanthroughdiskin-
stabilitiesisanoutputofourmodel–wedonotsimplyassume
thatellipticalgalaxiesonlyformduringmergersbutinsteadal-
low the different physical processes we believe are relevant to
the formation process to compete. In that sense, one can con-
sidertheresultsweobtainasademonstrationthattheformation
ofellipticalsisindeedhierarchical.
Second, elliptical and lenticular galaxy formation appears
to be somewhat different, since GALICS predicts that most
lenticular galaxies develop their bulges through disk instabili-
ties. Changing the modelling of disk instabilities would affect
the plots in a similar way for ellipticals and lenticulars, leav-
Fig.1. The ratio between stellar and gas masses transferred from the ingthedifferenceintheirrespectivebehavioursunchanged.The
disktothebulgeduringmergers(M )andthroughdiskinstabilities factthatthemodellenticularsacquiretheirmorphologythrough
merg
(M ). A galaxy witha value of 1 for this ratiohas acquired half its diskinstabilitiesisaresultwhichiscommoninnumericalsim-
inst
bulgemassduringmergersandtheotherhalfthroughdiskinstabilities. ulations of galactic dynamics (Combes et al. 2000) and gives
Thisplotisrestrictedtomodel galaxiesbrighter than M(B) = −18.9. ussomeconfidencethatoursimplephysicalmodellingbroadly
Notethatthe(artificial)peaksat4and–4areforratiosMmerg/Minst =∞ capturesthemechanism.
and0respectively.Theverticaldashedlinesrepresentmedianvalues.
Finally,thelowerpanelsofFig.1indicatethatspiralsform
their bulge components mainly through disk instabilities. The
globalpictureformorphogenesisthatweobtainfromGALICS
The minimum mass of galaxies (due to the minimum number
is satisfactory as it compares well to more detailed numerical
of particles used in the friend-of-friends group finder and the
baryonfractionused)is∼2×1010 M(cid:6). simulations.Moreover,weshowthatwhilstan“isolatedobject”
approximation can be justified to describe spiral galaxies (and
some lenticulars), it cannot be used to properly model the for-
2.2.Thegrowthofbulgesinthesemi-analyticalmodel mation of the vast majority of ellipticals in the framework of
CDMstructureformation.
2.2.1. Mergersvs.diskinstabilities
GALICS includes two processes which lead to the formation
andgrowthofspheroidalcomponentsingalaxies:mergers(both 2.2.2. Anoteonthetreatmentofdiskinstabilities
major and minor) and gravitational disk instabilities1. While
mergersdirectlyrevealthehierarchicalnatureofgalaxyforma- The disk instability model employed is similar to that of
tion, disk instabilities can result in the formation of bulges in vandenBosch(1998).Rotationalequilibriumisenforcedatthe
a “monolithic” manner (i.e. in objects classified as isolated at diskhalfmassradius–inotherwords,thecircularvelocityVtot
our resolutionlimit). The relative contributionof each of these arising from the presence of total mass of gas, stars and dark
twoprocessesinthebuild-upofbright,localgalaxybulgesthus matterenclosedinthehalfmassradiussphereiscalculated.Itis
yields a natural estimate of how hierarchical galaxy formation thencomparedtothecircularvelocityVc ofthediskalone,and
trulyis. if this latter is smaller than a critical value (obtained by mass
For each of our galaxies we measure the stellar and gas weighting the value of 0.7 for a pure gas disk and 0.52 for a
masses transferred from the disk to the bulge during mergers purestellardisk),theminimumamountofmassofgasandstars
(M ) and/or disk instabilities (M ). In Fig. 1, we show the requiredtore-establishstabilityistransferredtothedisk.
merg inst
relative contributions of mergers (both major and minor) and
disk instabilities to the mass build-up of the spheroidal com-
ponents of local galaxies brighter than M(B) = −18.9. This is 2.2.3. Post-mergermorphology
shownasthedistributionoftheratio M /M ,computedfor
merg inst
each model galaxy. A galaxy with a value of 1 for this ratio Intheliteraturemergersaretypicallymodelledbytakingthera-
tiooftheprogenitors,addingthestarsofthelightergalaxytothe
1 A disk is considered to be stable if Vc < 0.7 Vtot, where Vc is the diskoftheheavieroneifthemassratioislessthan fbulge ∼ 0.3
circular velocity of the disk and V is the rotational velocity of the or destroy the disk and form a bulge if the ratio is higher (e.g.
tot
disk-bulge-halosystem(e.g.vandenBosch1998). Walkeretal.1996;Coleetal.2000;Somerville&Primack1999;
448 S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel
Table1.FreeparametersthataffecttheevolutionofthebaryonicUniverse,seeSect.2.3formoredetails.
Parameter Description Fiducialvalue Sourceofconstraints
β Inverseofstarformationefficiency 50 Kennicutt(1998)
(cid:4) Inverseofmassloadingforfeedback 0.1 Martinetal.(2002)
χ Galaxymergerpowerlaw 3.333 Numericalsimulationse.g.Walkeretal.(1996)
Ω Baryonfraction 0.02h−2 2HabundanceinQSOabsorptionlines(Tytleretal.1996)
B
ψ S-Smergingnormalisation 0.017 Makino&Hut(1997)
ς Recyclingefficiency 0.3 SettofiducialvalueinHattonetal.(2003)
κ Ratioofburst-to-bulgeradius 0.1 SettofiducialvalueinHattonetal.(2003)
Dust ISMExtinctionandemissionduetodust – Guiderdonietal.(1987),Desertetal.(1990)
IMF Initialmassfunction – Kennicutt(1983)
Kauffmannetal.1999)2.Sinceagalacticdiskcanbedisrupted well-calibratedrecipestodescribegalaxyevolution.Thesesemi-
byanencounterevenwhentheinterloperislessmassivethanthe analytic recipesare drivenbyfree parameterswhich determine
disk itself, this simple prescription reproduces this behaviour. the evolution of the baryonic Universe. While the parameters
However, the sharp cut-off at f ∼ 0.3 between totally dis- thatdrivethebaryonicevolutioninthefiducialGALICSmodel
bulge
rupting or not disrupting the morphology of the galaxy seems arediscussedinSect.7ofHattonetal.(2003),webrieflyrevisit
somewhatunrealistic. themhereanddiscusstheirpotentialimpactontheanalysisthat
ThemodelimplementedinGALICSisconstructedinterms followsinthepaper.InTable1welistthefreeparametersinthe
of a smooth function (X), which models the fraction of disk modelandindicatehowtheirvaluesareconstrainedinthisstudy.
material remaining in the disk after the merger as a function The first section of the table indicatesparametersthat havethe
of the mass ratio of the two progenitors. The previous studies greatestimpactontheevolutionofthegalaxypopulationandthe
(described above) effectively use a step function, in the sense bottomsection indicatesparametersthatplay a minorornegli-
that the material kept in the disk is either 0 or 1 dependingon giblerole.
whetherthemassratioisgreaterorlesserthan0.3respectively. Whilethemodelisdrivenbyseveralfreeparameters,those
InGALICSthissmoothfunctionisdefinedas: that have the greatest impact on the spectro-photometricprop-
⎡ ⎛ ⎞ ⎤ erties of the galaxy population are: (a) the star formation effi-
X(R)=⎢⎢⎢⎢⎢⎣1+⎜⎜⎜⎜⎜⎝χ−1⎟⎟⎟⎟⎟⎠χ⎥⎥⎥⎥⎥⎦−1, (2) ciency; (b) the efficiency of mass loading for supernova feed-
R−1 back; and (c) the merging law that determines morphological
transformationsduringgalaxyinteractions.We exploreeach of
where R represents the mass ratio of the heavier to the lighter these in turn and briefly discuss the sensitivity of the modelto
progenitorand χ is the critical value of that ratio i.e. the value theseparameters.
that X = 0.5. Since step functionswith f ∼ 0.3 have been
bulge
foundtogivegoodresultsinthepast,thefiducialvalueadopted Starformationefficiency: the star formation efficiency (SFE)
inGALICSisχ=1/0.3.SeeFig.3inHattonetal.(2003)fora determines the rate at which cold gas is converted into
visualisationofX(R)forthisfiducialvalueofχ.Wereferread- stars. For example, decreasing the SFE results in less cold
erstoSect.5inHattonetal.(2003)fordetailsofprescriptions gas being converted into stars, moves the peak of the star
relatedtomergingemployedbyGALICS. formation rate closer to present day and affects both the
Galaxiesaremodelledwiththreecomponents:thedisk,the shape and normalisation of the z = 0 galaxy luminosity
bulgeandtheburst.Thebursthasthesamegeometry(andthere- function.Inourfiducialmodel,theinverseoftheSFE(β)is
forestarformationlaw)asthebulgebutitsscaleradiusis10% set equal to 50 (i.e. the SFE is ∼2%), following Kennicutt
ofthatofthebulge.Duringamerger,afractionofXofbothgas (1998, see their Eq. (7)), who combined Hα, HI, CO and
andstarsoriginallyinthediskremaininthediskwhiletherest far-infrared measurements in spiral and infrared-selected
aretransferredtotheburst.Starsformed/existingintheburstare starburstgalaxiestoprovideaparametrisationoftheglobal
rapidlytransferredtothebulge.Asthecentralburstformsstars star formation rate in the local Universe. Figure 6 in their
(typically over shorttimescales than the “quiescent” mode due study indicates that this parametrisation is accurate over a
toitssmallerscaleradius),thenewstellarcomponentgetstrans- wide gas density range, from gas-poor spiral disks to the
ferred to the bulge. As it subsequently runs out of the gas de- coresofthemostluminousstarburstgalaxies,implyingthat
positedbythemerger,thebursteventuallydisappears.Werefer the value for this parameter that is used in the model is
readerstoSect.5ofHattonetal.(2003)forfurtherdetails.Note reasonably well-constrained. Furthermore, the present-day
that,forhighervaluesofR(i.e.mergerswithahighmassratio) B and K-band galaxy luminosity functions predicted by
almost all disk material remains in the disk, while equal mass GALICS are very consistent with those observed by the
mergersresultinrapidbulgeformation,mimickingtheprescrip- 2dF survey (Cross et al. 2001),indicating that the effect of
tionsemployedintheliterature. the adopted value (β = 50) is consistent with the observed
distributionofgalaxyluminositiesintherealUniverse.
2.3.Adiscussionoffreeparametersinthemodel Massloadingforsupernovafeedback: for a given stellar
mass formed, a certain fraction is contained in massive,
Inevitably, any study that exploits a semi-analytical model
fast-evolving stars that become supernovae (SN), injecting
such as the one presented here, relies on simplified, albeit
kinetic energy into the interstellar medium. The efficiency
2 Inkeepingwiththeliterature,wedefineamergeras“major”when ofthisSNdriven“wind”dependsonboththeporosityofthe
themassratiobetweenmergingobjectsrangesbetweenoneandathird, ISM (see Silk 2001) and the “mass loading factor”, which
andas“minor”whenitissmaller. describes the entrainment of interstellar gas by the wind
S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel 449
(seee.g.Silk2001,2003).Forexample,increasingthemass radius of the bulge). Reducing κ would make the burst re-
loading efficiency will produce more feedback, heat more gionssmaller,shorteningthestarformationtimescales(and
cold gas and reduce the amount of fuel available for star vice-versa). However, typical star formation timescales (a
formation. This will, in turn, affect the galaxy luminosity fewtensofMyrat most)inthe burstregionsarenegligible
function at present-day.Following Martin et al. (2002),we comparedtothetimescalesoverwhichgalaxypropertiesare
assumethatthemassloadingfactoris∼10,whichresultsin being “viewed” in the model (several Gyr). Hence, chang-
the outflow rate of a starburst to be of the order of the star ingthisparameterdoesnotaffecttheobservedcoloursofthe
formationrate (see Sect. 4.2 in Hattonet al. 2003formore galaxypopulationpredictedbythemodelandthusleavesour
details).Thustheinverseofthemassloadingfactor((cid:4))isset conclusionsunchanged.
to0.1.Asmentionedbefore,thereproductionofthegalaxy The dust recipe used in the model is calibrated using
luminosity functions in the B and K bands – which relies the Milky Way, the Large Magellanic Cloud and Small
in part on (cid:4) – are well reproducedby the fiducial GALICS MagellanicCloud and a few localspirals for which the ex-
model. tinction curve, gas content and metallicity have been mea-
sured (see Guiderdoni & Rocca-Volmerange 1987, for de-
Mergerlaw: a third parameter that has a significant impact tails). While a key assumption is that the dust properties
ongalaxyevolutionisthemergerlawthatdeterminespost- of galaxiesare invariantwith redshift,tests ofthis assump-
mergermorphology.Thislawdeterminesthemorphological tion requires data at high redshift which are not yet avail-
mix of the Universe at z = 0 and therefore has a direct able, leaving us very little room to further calibrate our
impact on the early-type progenitor set in the model. The dust recipe. Finally, the fiducial GALICS model uses the
prescriptionusedtodeterminepost-mergermorphologyhas Kennicutt(1983)InitialMassFunction(IMF).Wenotethat
alreadybeendescribedindetailaboveinSect.2.2.3.Hatton the dispersion in the predicted properties of the early-type
et al. (2003) indicates that the predicted morphological populationin the localUniversedoesnotvarysignificantly
mix in GALICS is (within counting errors) consistent withothersimilarIMFssuchasSalpeter(1955)–seeTable1
with that seen in the real Universe. Note, however, that inKavirajetal.(2005).
the comparisons presented in Sect. 8.6 of Hatton et al.
(2003)usereasonablysmallsurveys(e.g.theStromlo-APM Notwithstandingthelargesetoffreeparametersinherenttoany
redshiftsurvey;Lovedayetal.1996)comparedtothescale semi-analyticalanalysis, we note thatthe fiducialvaluesof the
of modern surveys such as the SDSS. Future papers will parameters in the model are calibrated either through observa-
present more robust comparisons between the morpholog- tional data (e.g. the SF efficiency) or through numerical simu-
ical mix predicted by GALICS and that observed in the lations(e.g.themergerlaw).Mostimportantly,thefullsuiteof
SDSS e.g. through comparison of the GALICS predictions calibrated parameters that drives the model predictions repro-
tovisually-inspectedmorphologiesoftheentireSDSSDR6 ducesafundamentalsetofspectro-photometricpropertiesofthe
measuredbythe“GalaxyZoo”project(Lintottetal.2008). observedgalaxypopulationintheUniverse.GALICSproduces
goodagreementtothegalaxyluminosityfunctionsobservedby
Minorparameters: we complete our description of the free the 2dF survey in the B and K bands (Cross et al. 2001). The
parameters by briefly discussing minor parameters that do (B−V)coloursofspiralgalaxiescloselyfollowtheobserveddata
not produce a measurable impact on the analysis. The of Buta etal. (1994),bothin termsof averagevaluesandscat-
normalization for the satellite-satellite merging law (ψ) is ter.SatisfactoryfitstotheFaber-Jacksonrelationforearly-types
constrained using Makino & Hut (1997). Satellite-satellite (and the Tully-Fisher relation for disks) are obtained and pre-
mergingismuchrarerthandynamicalfrictionmerging(i.e. dictionsshowgoodagreementwiththeearly-typeFundamental
mergingwithcentralsinahalo)andthusthisparameterplays Plane, with the model predicting the observed morphological
anegligibleroleintheanalysis. mixofthelocalUniversewithareasonabledegreeofaccuracy.
The recyclingefficiency(ς) describesthe fractionof gas As we describein detailin the nextsection,GALICShasbeen
originally expelled from the halo by feedback which is re- specifically tested against a wide variety of multi-wavelength
accreted over time as the dark matter (DM) halo grows in (UV-optical) early-type galaxy properties across a wide range
mass. We note that Somerville & Primack (1999) set this in redshift (0 < z < 1.5), making it the ideal tool to map the
parameter to 0 (i.e. any gas expelled is lost from the sys- early-typeprogenitorsetovertheredshiftrange0<z<1.
tem altogether) while Cole et al. (2000) set it to 1. Hatton Wenowturnbrieflytotheimpactofthefreeparameterson
etal.(2003)useafiducialvalueof30%,torepresentaninter- oursubsequentanalysisandexplorethepotentialuncertaintyin
mediatebehaviourcomparedtothetwo extremesdescribed ourresults,giventhefreedomthatwemayhaveinthevaluesof
above,whichisplausiblyclosertothetruth.Wenote,how- the parametersin the model. Recalling that the primaryaim of
ever,thattheeffectofthisparameterisveryweakbecauseat thepaperistomapthephotometricpropertiesoftheearly-type
the mass resolutionof the model,verylittle gasis expelled progenitor set, we note that the parameters that potentially af-
fromDM halos. The re-accretedgasis alwayssignificantly fecttheanalysisthemostare:(a)thosethatareresponsiblefor
lessthantheamountof“pristine”gasaccretedortheamount the morphological mix of the Universe (since this determines
of gas that is ejected outof the galaxyand into the halo in the composition of the progenitor set at any given time); and
thefirstplace. (b) those thatstronglyaffectgalaxyluminosities/colours(since
InGALICSstarformationdrivenbydiskinstabilitiesand theprobabilisticprescriptionsdesignedtoincludelate-typepro-
mergers takes place in a central “burst” region, which is genitorsaregivenasafunctionofluminositiesandcolours).
modelled as having the same morphology as the bulge but As describedabovethe key parameterthatdrivesmorphol-
withafractionofitsradius(seeSect.4inHattonetal.2003 ogy is the merger law, in particular the transition mass ra-
foramoredetails).The“bursttobulge”radius(κ)–settoa tio (∼1:3) at which morphological transformations takes place
fiducialvalueof0.1–dictatesthesizeoftheburstregionin (i.e.disksaredisruptedandbulgesform).However,themerger
modelgalaxies(i.e.an orderofmagnitudesmaller thanthe recipe is constrained reasonably robustly through numerical
450 S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel
simulations and the galaxy population predicted by the model early-type galaxiesin the nearby (0 < z < 0.11)Universe that
isconsistentwiththemorphologicalmixofobservedUniverse, havebeenimagedbybothGALEXandtheSDSSDR3.
indicatingthatthecompositionoftheprogenitorsetispredicted Givenitsgoodreproductionofearly-typephotometryacross
with an acceptable level of accuracy across our target redshift a widewavelengthrangeanditssuccessfulpredictionofearly-
range. typecolourevolutiontohighredshifts(z ∼ 1.23),GALICSisa
Inasimilarvein,galaxyluminosities/coloursaredrivenbya usefulandwell-calibratedtoolwithwhichtofollowtheprogeni-
suite of parameters including the SF efficiency, SN feedback, torsetofearly-typegalaxiesandtheevolutionofthatprogenitor
IMF and dust recipes. This set of parameters is constrained set in the redshiftrange 0 < z < 1. We note that the emphasis
throughcomparisonofthefiducialmodeltothepropertiesofob- inthisstudyisnottofocusontheproperties(e.g.starformation
servedgalaxies.Whilewehave,inprinciple,freedomtochange andassemblyhistories)ofearly-typeremnantsatz=0(whichis
these parameters,calibrationstomultipleobservationalfactsin themajorthrustofprevioussemi-analyticalstudiesofelliptical
therealUniverse,suchastheonesdescribedintheprevioussec- galaxies),buttofocusonthepropertiesoftheprogenitorgalax-
tion,severelyreducethisfreedominpractice.Inotherwords,if iesovertheredshiftrangecoveredbythebulkoftherecentand
individual parameters are altered arbitrarily then the reproduc- forthcomingobservationalsurveys(0<z<1).
tion of the galaxy properties in the model would fail. In this
sense, the analysis presentedin the paper is stable, in thatthey
aredrivenbyvaluesofthefreeparametersthatarebasedeither 3. Dissectingtheprogenitorset:morphologies
on observations/hydrodynamicalsimulations and which repro- ofprogenitorsandepochsoflastmergers
duce the galaxy properties in the present-day Universe with a
Early-typegalaxieshavean assortmentof star formationhisto-
reasonabledegreeofaccuracy.
ries(SFHs). Centralto thisstudyistheepochatwhichthe last
merger,thatfinally createsthe early-typeremnant,takes place.
2.4.Reproductionoftheearly-typegalaxypopulation After this event, the early-type remnantevolves to present day
withoutfurtherinteractionswithothergalaxies.Inthediscussion
Since our aim is to map the photometric properties of early- thatfollows,werefertothelook-backtimetothis“last-merger”
type progenitors, it is important that the model reproducesthe asthedynamicalageofagalaxy.Notethatthelastmergerstyp-
multi-wavelength properties of the early-type population and icallyhavemassratiosof1:5orhigher.
theirevolutionwithredshift.Awell-calibratedmodelisclearly Asoursubsequentanalysisoftheprogenitorpopulationin-
needed for the progenitor predictions to be reliable. Note that volves environments of model galaxies, an explanation of the
in the redshift interval (z < 1) and for the wavelength range definition of this quantity is necessary. Galaxy environments
(B and V band) studied in this paper, the population synthesis in the model are driven by the mass of the dark matter (DM)
modelsemployedbythefiducialGALICSmodel–STARDUST halo in which they are embedded. At z = 0, DM halo masses
(Devriendt et al. 1999) – provide virtually identical results to greater than ∼1014 M(cid:6) correspond to “cluster” environments,
othercommonlyusedmodels(Yi2003;Bruzual&Charlot2003; whilehalomassesbetween∼1013 M(cid:6)and∼1014 M(cid:6)correspond
Maraston2005). to“groups”.Allotherhalomassescorrespondtothe“field”.At
In additionto the reproductionof the generalgalaxypopu- higher redshifts these definitions do not strictly hold since the
lation, the fiducial GALICS modelhas been specifically tested DMhalopopulationisevolving–forexample,thelargesthaloes
in the context of early-type galaxies, across virtually the en- atz=1arelikelytoberoughlyhalftheirsizeatpresentday(e.g.
tire redshift range over which early-types have been observed van den Bosch 2002).We take this mass accretion history into
(0<z<1.5).GALICSaccuratelyreproducestheopticalcolour account when specifying the environments of galaxies at high
magnitude relations (CMRs) of the early-type population (in redshift.ThemassaccretionhistoryistakenfromvandenBosch
denseenvironments)andtheirevolutionfromz = 0toz ∼ 1.23 (2002,seetheirFig.5).
(Kaviraj et al. 2005). The predicted evolution of both the gra- Thenumberofearly-typesgalaxiesinthesimulation,based
dient and the scatter in the optical CMRs are consistent within onthedefinitionspresentedabove,is5418.279early-typesare
errorswithvariousobservationalstudieswhichuseavarietyof inclusters(∼5%),1081(∼20%)areingroupsand4058(∼75%)
optical colours (see Fig. 8 in Kaviraj et al. 2005). More mas- are in the field. At z = 0 there are 8 clusters identified in the
siveellipticalsarepredictedtobeolder(althoughtheyassemble GALICSsimulationbox.
morerecently)andmoremetal-richthantheirlessmassivecoun- Figure2indicatesthelastmergerredshiftsofthesampleof
terparts (see Fig. 3 in Kaviraj et al. 2005). The star formation early-typesinthemodel,splitbytheenvironmentoftheremnant
historiesofellipticalgalaxiesareshowntobequasi-monolithic, atz=0.Recallthatafterthelastmerger,theearly-typeremnant
which enables the model to maintain the correct gradient and evolves to present day without further interactions with other
scatterovertheentirerangeinredshift(0 < z < 1.27)atwhich galaxies.TheinsetinFig.2showshistogramsofthelastmerger
observationalearly-typestudieshavebeenconducted(seeFig.5 redshifts, again split by environment. As expected, we find
inKavirajetal.2005).Notethatthedecouplingoftheassembly that, in allenvironments,largerearly-typesare assembledlater
historyofellipticalgalaxiesfromtheirstarformationhistoryde- althoughtheirstarsaregenerallyolder(Kavirajetal.2005).This
scribedabovehasalsobeenfoundinothersemi-analyticalwork resultisconsistentwiththefindingsofDeLuciaetal.(2006),al-
(e.g.Kauffmannetal.1996;Baughetal. 1996;De Luciaetal. thoughwenotethattheydidnotconsiderearly-typesindifferent
2006), although they do not study the evolution of the optical environmentsseparately.
colourswithredshift.Finally,GALICSistheonlysemi-analytic Clustergalaxies(atleastthosebrighterthanL∗)havesignif-
modeltohavebeentestedagainstthenewgenerationofUVpho- icantly larger dynamicalages – morphologicaltransformations
tometric data, shortward of 3000 Å, made available by the re- in clustersthereforeproceedmore quicklythan in all otheren-
centGALEXmission(Martinetal.2005).Kavirajetal.(2007) vironments.ThispointismademoreclearlyinFig.3,wherewe
findexcellentquantitativeagreementbetweenthepredictionsof plot the cumulative fraction of early-type galaxies which have
this model and the observed UV-optical photometry of ∼2100 alreadyhadtheirlastmerger.We findthat,onaverage,without
S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel 451
Fig.2. Last merger redshifts of the early-types in the model, split by
environmentoftheremnantatz=0.INSET:Histogramsoflastmerger
redshiftsshowninthetoppanel,splitbyenvironment. Fig.4. Morphologies of progenitors in binary mergers as a function
ofredshift.Non-binarymergersdohappenbutarerare,andonlytake
place,inthemodel,atredshiftsgreaterthanz ∼ 1.5.Mergersatinter-
mediateandhighredshiftaredominatedbypairsofprogenitorswhich
containatleastonespiralprogenitor.
fractionsrespectivelyof progenitorsof differentmorphological
types,intheredshiftrange0 < z < 3andsplitbyenvironment
andluminosityoftheearly-typeremnantatz = 0.Wefindthat,
averagingacrossallenvironments,atz∼1,lessthan50percent
ofthestellarmasswhichendsupinearly-typestodayisactually
in early-type progenitors at this redshift. Faster morphological
transformationsinclusterenvironmentsmeansthatthisvalueis
∼65 percent in clusters at z ∼ 1. As a result, looking only at
early-type galaxies at z ∼ 1 does not take into account almost
half the stellar mass in the progenitor set. In other words, the
mass in the progenitor set doubles between z = 1 and z = 0.
A similar observational result was found by Bell et al. (2004)
andFaberetal.(2007),whousedtheoptical“redsequence”as
Fig.3.Cumulativefractionofearly-typegalaxieswhichhavehadtheir
aproxyfortheprogenitorsetofpresent-dayearly-typegalaxies
lastmergerasafunctionofredshift.
(seealsothediscussioninSect.6below).
Thebiasdoesnotarisesimplybecausesomeprogenitormass
isnottakenintoaccount,butbecausetheageprofileofthemass
referencetoenvironment,only35percentofearly-typegalaxies in progenitors of different morphological types tends to vary.
are“inplace”(i.e.theyevolvewithoutfurtherinteractionswith We illustrate this point in Fig. 7. The top panel shows the av-
othergalaxies)byz=1(blackline).Therestarestill“inpieces”. erage NUV-weighted ages of progenitorsof different morpho-
Intermsofmorphologicaltransformations,clusterenvironments logicaltypes.TheNUV weighting,generatedusingtheGALEX
are special, in that early-typemorphologiesare attained signif- (Martin et al. 2005) NUV filter (2300 angstroms), is heavily
icantly faster in clusters (red curve),with almost 70 percentof dominatedbystarsformedwithinthelast0.5Gyroflook-back
early-typegalaxieshavingundergonetheirlastmergerbyz=1. time.Atallredshifts,early-typeprogenitorshavehigherNUV-
Beforethelastmergeroccurs,themorphologyoftheprogen- weightedages,becausethemassfractioncontributedbyrecent
itorsisnotnecessarilyearly-type.Figure4showsthemorpholo- star formation(RSF) i.e.withinthelast0.5Gyrissmallerthan
gies of progenitorsin binary mergersas a function of redshift. forspiralprogenitors.Thedifferencesbetweenellipticalandspi-
Non-binarymergersdohappenbutarerare,andonlytakeplace, ralprogenitorsaremostpronouncedatlowredshift.Thebottom
inthemodel,atredshiftsgreaterthan1.Inagreementwithpre- panel shows the fraction of the RSF across the progenitor set
viouswork(e.g.Khochfar&Burkert2003)wefindthat,inthe that is contained in each morphological type. This plot has to
localUniverse,mergersbetweenearly-typeprogenitorsmakeup be interpretedin conjunctionwith themassfractionshostedby
less than 20 percent of the merger activity. All other mergers eachmorphologicaltypeasafunctionofredshift.Forexample,
contain at least one spiral progenitor.Mergersinvolvingsolely at z ∼ 0.1, althoughspiral progenitorshost ∼40 percentof the
spiral progenitorsincreasingly dominate at higher redshift and totalRSFintheprogenitorset,theyonlyconstitute∼30percent
dominatethemergeractivitybeyondz = 1(seealsoKangetal. of the mass in the progenitor set (see bottom panel of Fig. 5).
2007). Early-type progenitors (elliptical and S0 taken together) con-
Havingprovideda pictureof the mergeractivity withinthe tribute∼60percentoftheRSF-buttheyalsoconstitute∼70per-
progenitorset,itisinstructivetolookatthefractionofthepro- centofthetotalmassintheprogenitorset.Thereforeatz ∼ 0.1
genitorsetwhichismadeupofacertainmorphologicaltypeasa spiral progenitors host 1.5 times the amount of RSF per unit
functionofredshift.Figures5and6showthenumberandmass masss than their early-typecounterparts.At higherredshiftthe
452 S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel
Fig.6.Number(top)andmass(bottom)fractionscontainedinprogen-
Fig.5.Number(top)andmass(bottom)fractionscontainedinprogen- itorsofdifferentmorphological typesintheredshiftrange0 < z < 3,
itorsofdifferentmorphological typesintheredshiftrange0 < z < 3,
splitbytheluminosityoftheearly-typeremnantatpresent-day.
splitbytheenvironmentoftheearly-typeremnant.
early-typeprogenitors.InFig.8weshowtheevolutionoftheB-
balanceofRSFcontainedineachmorphologicaltypemovesto-
bandLFofspiralprogenitors–wealsoshowthespiralprogen-
wardsspiralprogenitors,partlybecausetheyaremorespiralsin
itorsseparatedbytheenvironmentoftheircorrespondingearly-
theUniversethanearly-types.
typeremnantatpresent-day.Thelefthandcolumnillustratesthe
Figure 7 illustrates that an increasingly larger fraction of
evolutionofthespiralLFs–theyellowcurvedenotestheLFof
RSF in the progenitor set is contained in late-type systems at thegeneralspiralpopulation(i.e.progenitors+non-progenitors)
increasingredshift.Inthecontextofcolour-magnituderelations
and the black curvethe LF of spiral progenitorsonly.The LFs
(CMRs),whichareoftenusedtoage-dateearly-typepopulations ofspiralprogenitorswhoseearly-typeremnantsare,atz=0,in
at all redshifts, the exclusion of spiral progenitorsat high red-
clusters, groupsandthe field,are shownin red,greenandblue
shiftbiasestheCMRtowardsreddercoloursanddoesnotgivea
respectively.Theright-handcolumnshowsthefractionofspiral
properindicationoftheageofallthestellarmassthateventually
galaxieswhichare early-typeprogenitorsas a functionof their
constitutes present-day early-type galaxies. This is particularly
B-bandluminosities.Thecolourcodingisidenticaltothatused
trueifbluefilters(e.g.U-bandorshorterwavelengths)areused
fortheleft-handcolumn.
intheage-datinganalysis.
It is apparent that there is a greater preponderance of pro-
genitors amongst larger spirals at all redshifts. For example,
at low redshifts (z < 0.1), 20 to 40 percent of spirals with
4. Thespiralprogenitors
M(B) < −20.5are early-typeprogenitors.Atintermediatered-
Oneoftheaimsofthisstudyistoprovideameansofincluding shifts (0.3 < z < 0.52), these values rise to 30 and 60 percent
spiralgalaxiesobservedathighredshiftwhichmaybeprogeni- respectively.Athighredshift(z∼1)spiralswithM(B)<−21.5
torsintostudiesofearly-typegalaxyevolution,andthuscorrect, have more than a 60 percentprobabilityof beingan early-type
atleastpartially,forprogenitorbias.Wethereforefocusonspiral progenitor, while spirals with −20 < M(B) < −21.5 have be-
progenitorspredictedbythemodelandcomparetheirphotomet- tween a 30and 40 percentchanceofbeingearly-typeprogeni-
ricpropertiestothegeneralspiralpopulation. tors.Thefallingprogenitorfractionstowardslowerredshiftare
partlyduetothechangingmorphologicalmixoftheUniverse.
4.1.Theluminosityfunctionofspiralprogenitors
4.2.Thecolourmagnitudespaceofspiralprogenitors
We begin by studying the luminosity function (LF) of spiral
progenitors. We are interested in studying how the luminosi- While investigating the LFs of spiral progenitors is useful in
ties of spiral progenitors compare to the general spiral popu- indicating the probability that a spiral of a given luminosity
lation and what fraction of spirals, at a given luminosity, are has of being a progenitor, it is also desirable to explore the
S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel 453
At intermediate redshift (z ∼ 0.5), spirals with −21.5 <
M(B) < −20.5and(B−V) > 0.6,have∼30percentprobability
ofbeinganearly-typeprogenitor,whilebluespiralsinthesame
luminosity range have a low progenitor probability. For larger
spiralsattheseredshifts,theprobabilitiesareappreciablyhigher
-redspiralswith(B−V) > 0.7havebetweena75and95per-
centchanceofbeingprogenitors,while50to75percentofblue
spiralsinthisluminosityrangeareprogenitors.Thesituationat
highredshiftz∼1issimilartothatatintermediateredshift.
5. Progenitorevolutioninclusters
Before the advent of large scale surveys, dense regions of the
Universewereoftentargettedforearly-typegalaxystudies,both
atlowandhighredshift(e.g.Boweretal.1992,1998;Stanford
etal.1998;vanDokkumetal.1998,1999,2000,2001).While
studies of dense regions are attractive for a variety of reasons
(e.g. Ellis 2002; van Dokkum 2004), a key benefit is statisti-
calconvenience–clustersprovideaccesstolargehomogeneous
samplesofluminousobjectsatallredshifts.Ithasbeenusualto
“connect”resultsfromclusterstudiesoverlargeredshiftranges
to determine (at least qualitatively) the chronology of galaxy
evolution.
In this section, we investigate progenitors of present-day
cluster early-types, which are themselves in dense regions at
z > 0. The motivationfor this investigationis two fold. Firstly
(andmostimportantly),itprovidesacomparisontothevastlit-
eratureof“cluster”early-typestudies.Secondly,thisversionof
Fig.7. Top: average NUV-weighted ages of progenitors of different theGALICSmodelhasbeenaccuratelycalibratedtomatchthe
morphologicaltypes.TheNUVweightingisheavilydominatedbystars opticalCMRs ofearly-typesin denseregionsfromlow tohigh
formedintheseprogenitorswithinthelast0.5Gyroflook-backtime. redshift(Kavirajetal. 2005),whichimpliesthatthe coloursof
The NUV weighting was generated using the GALEX (Martin et al. early-type progenitors(regardless of morphology)are robustly
2005) NUV filter.Bottom: the mass fraction of recently formed stars reproduced.
(age <1Gyr old) acrosstheprogenitor set whichiscontained inpro- Itisimportantheretoclarifythedefinitionsof“density”that
genitorsofeachmorphologicaltype.Notethatthebottompanelhasto weusetodefineboth“clusters”atpresent-day,and“dense”re-
be interpreted in conjunction with the mass fractions hosted by each
gionsathighredshift.Asmentionedbefore,model“density”is
morphological typeasafunctionofredshift.Forexample,atz ∼ 0.1,
assumed to be a directfunctionof the mass of the DM halo in
althoughspiralprogenitorshost∼40percentofthetotalRSFinthepro-
genitorset,theyonlyconstitute∼30percentofthemassintheprogen- which galaxies are embedded. At z = 0, a DM halo mass of
itorset(seebottompanel ofFig.5).Early-typeprogenitors(elliptical 1014 M(cid:6) represents the lower limit for a cluster-hosting halo.
andS0takentogether)contribute∼60percentoftheRSF–buttheyalso Observationalstudiesofdenseregionsathighredshiftarelikely
constitute∼70percentofthetotalmassintheprogenitorset.Therefore to contain an assortment of cluster-type haloes of varying oc-
atz∼0.1spiralprogenitorshost1.5timestheamountofRSFperunit cupancies. Furthermore, DM haloes themselves are evolving -
massthantheirearly-typecounterparts. on average,the largesthaloes at z = 1 are likely to be roughly
halftheirsizeatpresentday(e.g.vandenBosch2002).Totake
thesetwopointsintoaccount,weuseavariablelowermasslimit
colour-magnitude (CM) space of the spiral population, so that for “cluster-hosting” haloes at z > 0. At a given redshift this
wecanseparateprogenitorspiralsbetterfromthegeneralpopu- lowerlimitiscalculatedfromtheaveragemassaccretionhistory
lationatagivenluminosity. (vandenBosch2002,seetheirFig.5)appliedtoa1014 M(cid:6)halo.
InFig.9wecomparethe(B−V)coloursofthegeneralspi- Figure10showsthe(B−V)CMRinclustersintheredshift
ralpopulationtothe (B−V) coloursof spiralprogenitors.The range 0 < z < 1. Large diamondsindicate progenitorgalaxies
left-hand column shows the spiral (B − V) CMR from z = 0 - black indicates ellipticals, green indicates S0s and blue indi-
to z = 1. Black dots represent the spiral galaxies and red dots cates spiral galaxies. Small crosses indicate galaxies which do
representspiralprogenitors.Inthe righthandcolumnwe show notcontributetothemassinpresent-dayclusterearly-types.All
the fractionofspiralprogenitorsacrossthe(B−V) CMR. The early-typegalaxiesindenseregionsare,notunexpectedly,pro-
fractionsareindicatedusingthecolourcodingshownintheleg- genitors of cluster early-types at present day. The top panel in
end. Warmer colours indicate a higher progenitor fraction (red Fig. 11 shows the fraction of spiral galaxies in dense regions
impliesaprogenitorfractionof1,blackrepresentsaprogenitor at high redshift (split by luminosity) which are progenitors of
fractionof0andpartsoftheCMspacewithoutanygalaxiesare early-typesatz=0.Thebottompanelshowstheoffsetin(B−V),
notcolour-coded). withrespecttoellipticalprogenitors,oftheS0andspiralprogen-
At z ∼ 0.1, spirals with −21.5 < M(B) < −20.5 have itorgalaxies.Theoffsetsareshownsplitbyluminosity.
∼30 percent chance of being a progenitor. For larger spirals, We find that at high redshift (z ∼ 1), up to 40 percent of
thosewithred(B−V)colours(i.e.(B−V)>0.8)have∼60per- large spirals (−23 < M(V) < −21) are progenitors, whereas
centchanceofbeingaprogenitor,whilethecorrespondingprob- only∼10percentofsmallspirals(M(V)>−21)aremembersof
abilityforbluerspiralsis30to50percent. theprogenitorset.Largespiralsarefourtimesmorelikelytobe
454 S.Kavirajetal.:Early-typeprogenitorsinthestandardmodel
Fig.8. Evolution of the B-band LF of spiral
progenitors. The left hand column illustrates
the evolution of the spiral LFs – the yellow
curve denotes the LF of the general spiral
population(i.e.progenitors+non-progenitors)
and the black curve the LF of spiral progeni-
torsonly.TheLFsofspiralprogenitorswhose
early-type remnants are, at z = 0, in clusters,
groups and the field, are shown in red, green
and blue respectively. The right-hand column
showsthefractionofspiralgalaxieswhichare
early-type progenitors as a function of their
B-bandluminosities.Thecolourcodingisiden-
ticaltothatusedfortheleft-handcolumn.
progenitorsthansmallspirals,regardlessofredshift,inthered- which is dominated by early-type galaxies. In Figs. 12 and 13
shiftrange0 < z < 1.Ellipticalgalaxiesformthereddestlocus theCMspacedominatedbyearly-typegalaxiesisshowningrey
in(B−V).S0galaxiesshowanaverageoffsetof–0.04compared –thisregionisdeterminedbyaprogressiveone-sigmafittothe
totheellipticalpopulation,regardlessofluminosity.Largespiral colours of the early-type population. It therefore contains, on
progenitorsshow an average (B−V) offsetof –0.05compared average,68percentoftheearly-typepopulationwithinit.Large
to the ellipticalpopulationat highredshift, mainly becausethe diamondsindicategalaxieswhicharepartoftheprogenitorset.
scatterintheellipticalcoloursalsotendstobelargeathighred- Galaxieswhicharenotpartoftheprogenitorsetareshownusing
shift.Atlowredshifttheoffsetismorepronounced-largespiral smalldots.Galaxiesintheredsequencearecircled.Itisappar-
progenitorsareupto0.1magbluerin(B−V)thantheelliptical ent that the red sequencemisses blue progenitorgalaxies,both
population. early-type and late-type. Figure 12 shows the evolution of the
red sequence population from low to intermediate redshift and
Fig. 13 the corresponding evolution from intermediate to high
6. The“redsequence”asaproxyfortheprogenitor redshift.
InFig.14wecomparegalaxiesintheactualprogenitorsetto
set
thoseintheredsequence.Thetoppanelshowsthenumberratio
Early-typegalaxiesinclusterstendtopreferentiallypopulatethe betweentheprogenitorsetpopulationandtheredsequencepop-
reddest parts of the colour-magnitude(CM) space. In this sec- ulation,splitbyredshiftandluminosity,whilethebottompanel
tionweinvestigatewhetherthesampleofgalaxies(withoutref- showsthemassratiobetweentheprogenitorsetpopulationand
erence to morphology),within the “red sequence” can be used the red sequence populationsplit by redshiftand luminosity.It
asaproxyfortheprogenitorset.Wedefinetheredsequenceas is apparentthatlargegalaxies(−23 < M(V) < −21)in the red
thegalaxypopulationwhichoccupiesthepartoftheCMspace sequence trace the progenitor set well in terms of number and