Very deep spectroscopy of the Coma cluster line of sight: exploring new territories

17 Pages·2009·3.37 MB·English

A&A507,1225–1241(2009) Astronomy
DOI:10.1051/0004-6361/200912228 &
(cid:2)c ESO2009 Astrophysics
Very deep spectroscopy of the Coma cluster line of sight:
(cid:2)
exploring new territories
C.Adami1,V.LeBrun1,A.Biviano2,F.Durret3,F.Lamareille4,R.Pelló4,O.Ilbert1,A.Mazure1,
R.Trilling3,andM.P.Ulmer5
1 LAM,OAMP,UniversitéAix-Marseille&CNRS,Pôledel’Etoile,SitedeChâteauGombert,38rueFrédéricJoliot-Curie,
13388Marseille13Cedex,France
e-mail:[email protected]
2 INAF-OsservatorioAstronomicodiTrieste,viaG.B.Tiepolo11,34143Trieste,Italy
3 Institutd’AstrophysiquedeParis,CNRS,UMR7095,UniversitéPierreetMarieCurie,98bisBdArago,75014Paris,France
4 Laboratoired’AstrophysiquedeToulouse-Tarbes,UniversitédeToulouse,CNRS,14Av.EdouardBelin,31400Toulouse,France
5 DepartmentofPhysics&Astronomy,NorthwesternUniversity,2131SheridanRoad,Evanston,IL60208-2900,USA
Received30March2009/Accepted23June2009
ABSTRACT
Context.Environmental effectsareknown tohaveanimportant influence oncluster galaxies,but studiesatvery faintmagnitudes
(R>21)arealmostexclusivelybasedonimaging.Wepresenthereaverydeepspectroscopicsurveyofgalaxiesonthelineofsight
totheComacluster.
Aims.Afteraseriesofpapersbasedondeepmulti–bandimagingoftheComacluster,weexplorespectroscopicallypartofthecentral
regions of Coma, inorder toconfirmand generalize previous results,concerning inparticular thegalaxy luminosity function, red
sequence,stellarpopulationsandthemostlikelyformationscenariofortheComacluster.
Methods.WehaveobtainedreliableVIMOSredshiftsfor715galaxiesinthedirectionoftheComaclustercentreintheunprecedented
magnituderange21≤R≤23,correspondingtotheabsolutemagnituderange−14≤M ≤−12.
R
Results.WeconfirmthesubstructurespreviouslyidentifiedinComa,andidentifythreenewsubstructures.Wedetectalargenumber
of groups behind Coma, in particular a large structure at z ∼ 0.5, the SDSS Great Wall, and a large and very young previously
unknownstructureatz∼0.054,whichwenamedthebackgroundmassivegroup(BMG).Thesestructuresaccountforthemassmaps
derivedfromarecentweaklensinganalysis.Theorbitsofdwarfgalaxiesareprobablyanisotropicandradial,andcouldoriginatefrom
fieldgalaxiesradiallyfallingintotheclusteralongthenumerouscosmologicalfilamentssurroundingComa.Spectralcharacteristics
ofComadwarfgalaxiesshowthatredorabsorptionlinegalaxieshavelargerstellarmassesandareolderthanblueoremissionline
galaxies.R≤22galaxiesshowlessprominentabsorptionlinesthanR≥22galaxies.Thistrendislessclearforfieldgalaxies,which
aresimilartoR ≥ 22Comagalaxies.ThissuggeststhatpartofthefaintComagalaxiescouldhavebeenrecentlyinjectedfromthe
fieldfollowingtheNGC4911groupinfall.Wepresentalistoffiveultracompactdwarfgalaxycandidateswhichneedtobeconfirmed
withhighspatialresolutionimagingwiththeHST.Wealsogloballyspectroscopicallyconfirmourpreviousresultsconcerning the
galaxyluminosityfunctionsbasedonimagingdowntoR=23(M =−12)andfindthatdwarfgalaxiesfollowaredsequencesimilar
R
tothatdrawnbybrightComagalaxies.
Conclusions.SpectroscopyoffaintgalaxiesinComaconfirmsthatdwarfgalaxiesareveryabundantinthiscluster,andthattheyare
partlyfieldgalaxiesthathavefallenontotheclusteralongcosmologicalfilaments.
Keywords.galaxies:clusters:individual:Coma
1. Introduction clusterswaswellconstrainedonlydowntorelativelybrightmag-
nitudes (R ≤ 20 for z ∼ 0 clusters). However, according to
On the pathway toward the use of galaxy clusters to constrain
Cold Dark Matter models of hierarchical structure formation
cosmology,one mustunderstandhow clusters and their galaxy
(e.g.White&Rees1978;White&Frenk1991),thereshouldbe
populations evolve. Until very recently, galaxy evolution in
abundantlow-massdark-matterdominatedhalos presentin the
Universeandthesehalosshouldthereforecontainlowluminos-
(cid:2) Based on observations collected at the European Organisation itygalaxies.Itisthereforeimportanttosamplethefaintandvery
for Astronomical Research in the Southern Hemisphere, Chile (pro- faint cluster galaxy populations. Moreover, in galaxy clusters,
gram: 081.A-0172). Also based on observations obtained with these faint galaxies are of major interest as their evolutionary
MegaPrime/MegaCam, a joint project of CFHT and CEA/DAPNIA, pathsaresometimesdifferentfromthoseofbrightgalaxies:they
attheCanada-France-HawaiiTelescope(CFHT)whichisoperatedby are very sensitive to environmental effects and can be created
theNationalResearchCouncil(NRC)ofCanada,theInstitutNational
via interactions of larger galaxies (e.g. Bournaud et al. 2003).
des Sciences de l’Univers of the Centre National de la Recherche
They also keep their dynamical memory longer (e.g. Sarazin
Scientifique(CNRS)ofFrance,andtheUniversityofHawaii.Thiswork
1986), and as a consequence their spatial distribution can pos-
is also partly based on data products produced at TERAPIX and the
CanadianAstronomyDataCentreaspartoftheCanada-France-Hawaii siblybedifferentfromthatofbrightgalaxies(e.g.Bivianoetal.
TelescopeLegacySurvey,acollaborativeprojectofNRCandCNRS. 1996). With the arrival of large field cameras on medium size
ArticlepublishedbyEDPSciences
1226 C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories
telescopes, we started to reach the faint and very faint galaxy 195.2 195.0 194.8 194.6 194.4
regime.OurteamconcentratedontheComacluster(Adamietal.
2005a,b,2006a,b,2007a,b,2008,2009a,b;Gavazzietal.2009,
and http://cencosw.oamp.fr/COMA/). This cluster is rela- 28.2
tivelynearbyandthisfacilitatedearlysearches(e.g.Wolf1901;
see Biviano 1998and referencetherein;and for recentstudies, 28.1
seee.g.Andreon&Cuillandre2002;Beijersbergenetal.2002;
Iglesias-Páramo et al. 2003; Jenkins et al. 2007; Lobo et al.
28.0
1997;Milneetal.2007;Smithetal.2008;Terlevichetal.2001;
Trentham 1998). Using CFH12K and Megacam data we now
haveagoodstatisticalviewofthefaintestgalaxiesexistinginthe 27.9
Comacluster(M ∼−9.5).However,thisviewis“only”statisti-
R
calandweonlyhavearoughideaofthebehaviourofindividual
galaxies at these magnitudes. Our previousspectroscopic cata- 27.8
log limit (despite the fact thatit gatheredmost of the literature
data available at thattime, see Adamietal. 2005a) was far too
27.7
brighttoinvestigatethesefaintpopulations.Also,theprecision
on photometricredshifts is far too low to allow any dynamical
analysisorspectroscopiccharacterisation. 27.6
In order to fill this gap, we have obtained deep (21 ≤ R ≤
23)VIMOS/VLTspectroscopyandperformedthespectroscopic
characterisationofthesefaintComaclustergalaxypopulations. Fig.1.u*bandMegacamimagewithareacoveredbytheVIMOSspec-
We describe our new spectroscopic and literature data in troscopy overlayed (in red). Large red circles are the VIMOS galax-
ies inside the Coma cluster. Small green circles are galaxies inside
Sects. 2 and 3. We then present in Sect. 4 the analysis of the
the Coma cluster taken from the literature. Blue contours represent
Comalineofsightintermsofdetectedgroupsandsubstructures.
the X-ray substructures from Neumann et al. (2003). Coordinates are
WederivethedynamicalbehaviouroftheComaclustergalaxies
J2000.
in Sect. 5 and describethespectralcharacteristicsof the Coma
cluster galaxies in Sect. 6. In Sect. 7, we build an ultra com-
pact dwarf galaxy candidate catalog. We discuss in Sect. 8 the detected(asdescribedinAdamietal.2005a):thewestinfalling
luminosity function and color-magnitude-relationof the Coma galaxylayerandsubstructuresclosetoNGC4911.
cluster galaxies. Finally, we summarize our results in Sect. 9, We obtained a reliable redshift (reliability flag ≥2, see the
givingacomprehensivepictureoftheComacluster. following) for 715 objects. Among these, slightly fewer than
Inthispaperweassume H = 70kms−1Mpc−1,Ω = 0.3, 100 galaxiesare partof the Coma cluster. The minimum num-
0 m
ΩΛ = 0.7,adistancetoComaof100Mpc,adistancemodulus berofgalaxiesexpectedinsidetheComaclustergiventhetarget
of 35.00, and a scale of 0.47 kpcarcsec−1. All coordinates are selectionwas70,soourresultsareingoodagreementwithour
givenattheJ2000equinox. predictions.
2. VIMOSspectroscopy 2.2.Datareduction
ThespectrawereextractedusingtheVIPGIpackage(Scodeggio
2.1.Settings
et al. 2005), which was extensively used for the VIMOS VLT
Selectingtargetsonthebasisofnewphotometricredshiftscom- Deep Survey (VVDS) data (e.g. Le Fèvre et al. 2005). This
puted using deep u*BVRI images, we observed three VIMOS package (VIMOS Interactive Pipeline and Graphical Interface:
fields in the Coma cluster in orderto spectroscopicallycharac- VIPGI), is a new software designed to simplify to a very high
terize the faint cluster population and to sample the cluster at degree the task of reducing astronomical data obtained with
unprecedentedmagnitudesofR∼23. VIMOS (Visible MultiObject Spectrograph).The final product
Weobtained∼1000spectraoffaintComalineofsightgalax- wasthewavelength-calibrated1Dspectraofourtargets.
ies (R ∼ [21,23]) using the VLT/VIMOS instrument in 2008 Duetothefactthatourobservationsweremadeathighair-
with exposuretimesof ∼2 h,splitinto five ∼24minindividual masses, differentialrefractionwasnotnegligible.However,the
exposures. Despite the very unfavourable declination of Coma levelofspectraphysicalcurvatureontheCCDsremainedmod-
attheVLTlatitude,wewereabletoobservethreemasksatair- estandwaseasilycompensatedforbyalinearfunctiontodefine
massescloseto1.7,withaseeingoftheorderof1.2arcsec(the theextractionarea.Theslopeofthelinearfunctionwasvisually
relativelylongexposuretimesforR∼23galaxiescompensating adjustedforeachquadrant.
forthehighairmass).Thetargetswerepartlyselectedonapho- We then adopted the same strategy as for the VVDS. Each
tometricredshiftbasis(followingAdamietal.2008)andpartly spectrum was examined independently by two of us (CA and
randomlyinordertoincreasethenumberoftargets. VLB) and redshifts were proposed using well known cross-
Since our aim was to obtain low resolution spectra of very correlation techniques such as the rvsao IRAF package or the
faint nearby galaxies, we used the LR-blue grism (5.3 Å/pix), EZ VVDStool(aswellasVIPGI-includedlineGaussianfitting
providing a S/N between 2 and 5, depending on the galaxy whenpossible).Thedifferentredshiftdeterminationswerecom-
characteristics. Given the redshift range of interest (Coma is paredandafinallistwasestablished,includingqualityflags.
at z ∼ 0.023), it allowed us to efficiently sample lines from When a redshiftdeterminationwas possible, it was flagged
[OII]3727toHα. with a number between 1 and 9, following the VVDS conven-
We chose to observe strategically located regions of the tions (e.g. Le Fèvre et al. 2004) and indicating the reliability
Coma cluster (see Fig. 1), where infalling material has been levelofthemeasurement.Flags2,3,4arethemostsecure,flag1
C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories 1227
Table 1. For the three VIMOS fields observed: coordinates (J2000),
numberofslits(N)andexposuretimes.
Fieldnumber α δ N Exp.time
(deg) (deg) (s)
1 194.72 27.87 322 7200
2 194.69 28.14 298 7184
3 195.04 27.72 306 7098
Fig.3.Redshifthistogramofthespectroscopicdata(binof0.005)avail-
able in the region of interest (literature + VIMOS). The two vertical
dottedlinesrepresentthelimitsweadoptedfortheComacluster.The
inner box shows the redshift histogram (bin of 0.0005) for the Coma
clusteritself.
Fig.2.LogarithmoftheredshiftversusVegaRbandmagnitudeforthe
spectroscopic sample.Filledcirclesaregalaxieswithareliabilityflag
atleastequalto2.Opencirclesaregalaxieswithareliabilityflagequal
to 1. The two horizontal lines show the redshift limits chosen in this
paper for the Coma cluster. The vertical line is the minimum R band
magnitudewewillconsiderforthemostoftheanalyses(R=21).
isanindicativemeasurementbasedonfewsupportingfeatures,
and flag 9 indicates that there is only one secure emission line
correspondingtothelistedredshift.LeFèvreetal.(2005)have
shownwithrepeatedobservationsthattheseflagscorrespondto
Fig.4.Rbandmagnitudehistogramofthespectroscopicdataavailable
the following probabilities of finding the correct redshift: 55% intheregionofinterest(literature+VIMOS)forthegalaxiesinsidethe
forflag1,81%forflag2,97%forflag3,and99.5%forflag4. Comacluster.
Giventhesenumberswesimplychosetoignoreallobjectswith
flag1.
Wealsogaveasecondintegerflagbetween0and3(spectral
flag): 0 for absorption line only spectra, 1 for spectra showing Le Fèvre et al. (2005) estimated the redshift uncertainty to be
both emission and absorptionlines, 2 forspectra showingonly of the order of 280 kms−1 from repeated VVDS redshift mea-
emissionlines,and3foractiveobjects(i.e.showingbroadband surements.We performedthesameexercisebyconsideringthe
emissionlinesorverystrongBalmerlinesascomparedtooxy- literature redshifts included in our VIMOS sample (6 galaxies
genlines). between R = 17 and 19.8) or observed twice during our spec-
We finally note that our spectra are not photometrically troscopicrun(3galaxiesbetweenR = 20.8and23),andwitha
calibrated. reliabilityflagstrictlygreaterthan1.Giventhereliabilityflags,
we expect to have 8 galaxies out of 9 providinga good agree-
ment. In practice, 7 show a good agreement.One galaxy has a
2.3.VIMOSspectroscopicsample 0.008differenceinredshift,butcheckingtheliteraturespectrum
Table1givesthebasicdetailsofthespectroscopicVIMOSob- (fromtheSDSSsurvey)andapplyingourredshiftmeasurement
servations.Outofthe926slits,833objectsprovidedaspectrum methods,we estimate a redshiftsimilar to the onewe obtained
with a tentative redshift measure. Among these 833, 118 were fromourownspectrumofthesamegalaxy(theSDSSautomated
flagged with a reliability flag of 1, 298 with a reliability flag measurepipelinemisinterpretedtheedgeofanabsorptionlineas
of 2, 187 with a reliability flag of 3, 187 with a reliability flag anemissionline).
of4,and43withareliabilityflagof9. Foranothergalaxythereisastrongredshiftdiscrepancybe-
Afterexcludingobjectswithareliabilityflagequalto1,we tweenourmeasurementandthatoftheliterature.However,our
wereleftwith715galaxies,outofwhich250hadaspectralflag spectrumcorrespondstoaflag4(z=0.1468)galaxywithstrong
of 0, 288 a spectral flag of 1, 168 a spectral flag of 2, and 9 a emission lines, whereas the spectrum from the literature is not
spectralflagof3. availableforinspection,makingourpresentdeterminationmore
Figures2−4giveageneralideaofthepropertiesofthesam- reliablethanthepreviousone.Figure5showstheresultingred-
ple:distributionoftheredshiftsasafunctionoftheR-bandmag- shift comparison, corresponding to a statistical uncertainty of
nitudes,redshifthistogram,andmagnitudehistogram. 196 kms−1, slightly lower than the VVDS estimate. We also
Wealsoinvestigatedthevelocityuncertaintyresultingfrom show in this graph the galaxies observed twice in our spectro-
our measurements. With a similar instrumental configuration, scopicruns.Despitethelowstatistics,theseobjectsdonotseem
1228 C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories
4. GroupsalongtheComaclusterlineofsight
fromthespectroscopicsample
WenowpresentouranalysisofthegroupsfoundalongtheComa
lineofsight.
First, we haveto definethe boundariesofthe Coma cluster
intermsofvelocity.FromAdamietal.(1998),weknowthatthe
Coma cluster galaxy velocity dispersion σ increases with de-
v
creasingluminosity,asexpectedfromdynamicalconsiderations.
Anaturalwaytofixclusterboundariesistolimittheclusterve-
locityrangewithin±3σ .Assumingthemeanσ forthefaintest
v v
galaxiesinAdamietal.(1998)(1200kms−1 forR ∼ 17.5),we
limittheclustertotheredshiftrange[0.011;0.035].
Fig.5. Difference between literature and VIMOS redshift measure- Second, in order to determine galaxy orbits, we limit the
ments as a function of redshift. The galaxy with the 0.008 difference spectroscopic sample to the areas where photometric redshifts
(seetext)isshownbeforeandafterremeasuringitsredshift.Thecircled arealsoavailable(inordertobeabletocomputeadensityprofile
objectsarethethreegalaxiesobservedtwotimesinourspectroscopic onthephotometricredshiftbasis).Finallywerestrictthemagni-
runsandarefainterthanR=20.8(seetext). tuderangeofthespectroscopicsampletotheR=[21,23]range
wherethesamplingrateisthemosthomogeneous.
toshowadifferentbehaviourcomparedtobrighterobjectsinthe In order to search for galaxy groups along the Coma clus-
ter line of sight, we applied the Serna-Gerbal (SG hereafter:
literature.
Serna & Gerbal 1996) method to our total redshift catalogue
(VIMOS+literature).Thishierarchicalmethodwasalreadyap-
3. Complementarydata plied in one of our early studies (Adami et al. 2005a) and we
referthe readertothis paperformoredetails. Briefly,itallows
3.1.Spectroscopicdatafromtheliterature
galaxy subgroups to be extracted from a catalogue containing
A spectroscopiccatalogbasedondatatakenfromtheliterature positions, magnitudes, and redshifts, based on the calculation
was compiledby one of us (RT) from the NED and SDSS (in- of their relative (negative)binding energies. Note that this cal-
cludinguptoDR6)databases.Weextractedfromthiscatalogall culation takes into account the mass to luminosity M/L ratio
galaxiesinourVIMOSregionthatarenotalreadypresentinour chosen by the user as an a priori input value of the M/L ra-
VIMOSdata.WealsoaddedafewnewredshiftsfromtheSDSS tio in the structure. The group mass derived later is estimated
DR7.Theresultingliteraturecatalogisthereforemorecomplete fromthegroupbindingenergyandvelocitydispersion,anddoes
thantheonewe usedinAdamietal. (2005a).Theseadditional not depend upon M/L, which proves empirically (e.g. Covone
datawereincludedinFigs.2−4.Figure3showstheComaclus- etal.2006)toactmainlyasacontrastcriterion.Resultsdonot
ter itself plus several structures along the line of sight for ex- depend strongly on this factor, since a variation of a factor of
ampleatz ∼ 0.16,0.38,or0.52.Figure4showsthemagnitude two in this parameterdoes not significantly change the results.
gapbetweenspectroscopicdatafromtheliteratureandournew Qualitatively,lowinputvaluesof M/Lallowustodetectstruc-
VIMOS spectroscopic data. This gap has no consequences on tureswithlowbindingenergies,whilehighvaluesof M/Lonly
mostofthefollowinganalyses,whicharebasedonR ≥ 21ob- allowthedetectionofthemajorstructures.
jects.However,wehavetobearinmindthemagnitudedistribu- The output of the SG method is a list of galaxies belong-
tionofthespectroscopicsamplewhendiscussingpossiblestruc- ingtoeachgroup,aswellasinformationonthebindingenergy
turesalongthelineofsight. andmassofthegroupitself.Wewillconsiderherethatagroup
consistsofatleast3members.
We use a nominal M/L ratio in the R band of 400, and we
3.2.u*,B,V,RandIbandCFHTimagingdata
also search for less strongly linked galaxy groups inside the
The B, V, R, and I data are fully described in Adami et al. Coma cluster and inside the z ∼ 0.5 large scale structure (see
(2006a).We givehereonlythe salientpoints.A mosaicoftwo thefollowing)assuminglowerM/Lratiosof100.
fieldswasobservedwiththeCFH12Kcamerainfourbands(B, The SG methodreveals the existence of 76 “groups”along
V, R and I) covering a 52 × 42 arcmin2 field. The total field the line of sight (assuming M/L = 400), including the Coma
isapproximatelycenteredonthetwodominantclustergalaxies clusteritself.Amongthem,52groupshaveN ≥4members,and
(NGC4874andNGC4889).WederivedtotalKronVegamag- 44groupshaveN ≥5members.InFig.6weshowthelog10(N)
nitudes(Kron1980)fromtheseimages.Thecompletenesslevel versusredshiftforthe76groups.ThereisnorelationbetweenN
inRisclosetoR∼24.Theseeingconditionswereallcloseto1 andthegroupredshifts.
arcsec.Alltheseimagingdataareavailableat The main caveat of this analysis is that we do not have a
http://cencosw.oamp.fr. completespectroscopiccatalog.We willdedicatethefollowing
The u* data including the previous field are described in sectiontothispoint.
Adamietal.(2008).Theywereobtainedbetween2006and2007
with the CFH Megacam camera in a field of view of 1 deg2
4.1.Samplingofthetotalspectroscopiccatalog
with an average seeing of 1.1 arcsec. The total exposure time
was9.66h.WederivedtotalKronABmagnitudesfromthisim- InordertofindallthestructuresalongtheComaclusterlineof
agereducedusingSCAMPandSWARPtools(Bertinetal.2002; sight, the SG methodwould requirethe galaxycatalogueto be
Bertin2006).Figure2showsasubareaofthetotalu*image. 100% complete, which is obviouslynever the case. The detec-
These data thereforeprovidea catalog of objectsin the u*, tionefficiencydependsonthecompletenessofthespectroscopic
B,V,RandIbandscompletedowntoR∼24. catalogandwemustthereforeanalyzethiscompletenessinthe
C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories 1229
partsofthespectroscopicareaonlypresentasparsedistribution
of z≤0.2backgroundgroups.The location of these detections
cannot be explained by the spectroscopic incompleteness (see
Fig.7)alone.Evenifnotuniform,thissamplingisnotpreferen-
tiallyhighinthePFA.
Partof the detected groupsare includedin the SDSS Great
Wall(Gottetal.2005)andarecoincidentwiththePFA,confirm-
ing the existence of a complexstructurebetween z ∼ 0.07and
0.085.Wealsodetectanewlargegalaxystructureatz ∼ 0.054
(theBackgroundMassiveGroup,orBMGhereafter).Thisstruc-
ture, undetectedwith SDSS data alone, is mainly populatedby
faint galaxies and is very rich (sampled with 58 spectroscopic
redshifts). It is probably not virialized, because applying the
virial theorem would lead to a mass of ∼2.7×1015 M(cid:6). Such
Fig.6.NumbersofgroupmembersfoundbytheSerna–Gerbalmethod
(inlogarithmicscale)versusredshiftforthe76N ≥3detectedgroups. amajormassconcentrationwouldevidentlyshowupinX-rays,
andnothingisdetectedbyNeumannetal.(2003)atthislocation.
Thisdoesnotmeanthatthisstructurehasanegligiblemass
however, and it could significantly contribute to the mass con-
zonethatweconsidered.Forthis,weuseourCFHTR-bandpho-
centration detected with a weak lensing analysis by Gavazzi
tometric catalog (http://cencosw.oamp.fr/). We pixelize
et al. (2009) at the same location (see Fig. 8). If we define the
the VIMOS field in several subregions of 0.2 × 0.2 deg2 and
PFA ascenteredatα = 194.772,δ = 27.914andenclosedina
compute the percentage of objects with a redshift available in circle of 5.5(cid:7) radius, we can estimate a mass from the Gavazzi
our spectroscopic catalog (VIMOS + literature), as a function
etal.(2009)weaklensinganalysisof(4.2±0.8)×1013 M(cid:6) (if
of R band magnitude. Figure 7 gives the sampling rate in the
at z = 0.023), or 2 × 1013 M(cid:6) (if at z = 0.055). The Adami
VIMOSareaforseveralmagnituderanges.
etal.(2005a)G8andG9Comasubstructuresareprobablycon-
tributing to the PFA mass detected by weak lensing. However,
4.2.SubstructuresintheComacluster G8andG9arenotdetectedinX-raysandtheirmaximummass
canbeinferredtobe∼0.5×1013M(cid:6),thelowestmassComasub-
Assuming an M/L of 400, well adapted to find highly gravita- structure emitting in X-rays (the group attached to NGC 4911,
tionallyboundstructures,wedetecttheComaclusterasasingle seeNeumannetal.2003).GroupsG8andG9canthereforeac-
structure(z ∼ 0.0234sampledwith382redshifts),alongwitha countfor∼25%ofthePFAestimatedmass.The75%remaining
minoradditionalsubstructure(sampledinspectroscopybyonly couldthencomefromthez∼0.054BMGstructure,leadingtoa
3galaxies). massoftheorderof(1−3)×1013 M(cid:6).Thismassishighenough
We theninvestigatehowthe maindetectedstructurecanbe to justify the qualificationof massive, butsmall enoughto jus-
split in several substructures. For this, we redo the same SG tify theundetectedX-rayemission.Thelargedifferenceofthis
analysis using this time a M/L ratio of 100.Results are shown mass estimate with the virial one clearly argues in favor of an
in Fig. 8. We essentially confirmin the consideredarea the re- unvirializedstructure.Withacrossingtimeof∼2.9×108 years
sults of Adami et al. (2005a), and detect 7 subtructures. Three atz = 0.054,thisplacestheBMGstructureinaveryearlyevo-
are directly linked to NGC 4874 (G1 of Adami et al. 2005a). lutionarystage.
A fourth one (undetected in Adami et al. 2005a) is detected Several other smaller structures are detected between z =
northofNGC4874andcoincideswith thenorthernpartofthe 0.035and0.2inthePFA.Thisleadsustosuspecttheexistence
west X-ray substructures of Neumann et al. (2003). The last ofa(minor)lineofsightfilamentjoiningtheComaclusterand
three are detected south of NGC 4874. One of them is along the new z ∼ 0.054 large galaxy structure, and then extending
the NGC 4839 infalling path, and coincides with the G8/G9 towardtheSDSSGreatWallandbeyond.
groups(seeTable2andFig.4inAdamietal.2005a),whilethe Suchahighlevelofstructuresintheimmediatebackground
lasttwowerenotdetectedinAdamietal.(2005a).Thesethree vicinity of the Coma cluster could have a significant effect on
groupsalsoseemtobevisibleintheOkabeetal.weaklensing theclusterluminosityfunctiondeterminationsinthisregion,es-
study (private communication) and are included in the south- timatedforexamplefromstatisticalarguments(seeAdamietal.
westmassextensiondetectedbyGavazzietal.(2009),whoused 2007a,b),becauseasignificantpartofthesenearbybackground
thesamephotometricdataasinthepresentpaper. groupswere not detected in this early study. We will therefore
determineinthefollowingaluminosityfunctiononlybasedon
4.3.Structuresatz≤0.2notbelongingtoComa ourspectroscopicdata.
We note that most of these structures(e.g. the BMG or the
BesidestheComaclustersubstructures,wearemorespecifically SDSSGreatWall)donotprominentlyappearinFig.3because
interested in the z ≤ 0.2 structures (see also Gutiérrez et al. theyarenotfullyvirializedandthereforetheirredshiftdistribu-
2004)becausetheycannotbeefficientlyremovedfromtheback- tionisnotverycompact.
ground using only photometric redshifttechniques (see Adami
etal.2008).
4.4.SamplingtheComaback-infallinggalaxylayers:nature
WegiveinFig.8andinTable2thelocationsandcharacter-
ofgalaxyhaloes
istics of the z ≤ 0.2 structures, excluding the structures which
aremembersoftheComacluster. Our spectroscopic sample contains a number of active objects,
We immediately notice a concentration of structures in a which can be used to study the foreground gaseous clouds
small zone south-west of NGC 4874, which we will call the through the absorption features that imprint the spectrum (e.g.
Putative Filament Area (PFA) in the following, while other Ledoux et al. 1999). This method has led to the discovery of
1230 C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories
195.2 195.0 194.8 194.6 195.2 195.0 194.8 194.6
28.2
28.2
28.1
28.1
28.0
28.0
27.9
27.9
27.8
27.8
27.7
27.7
27.6
27.6
5% 10% 15% 20% 2% 4% 6% 8% 10%12%
195.2 195.0 194.8 194.6 195.2 195.0 194.8 194.6
28.2 28.2
28.1 28.1
28.0 28.0
27.9 27.9
27.8 27.8
27.7 27.7
27.6 27.6
2% 4% 6% 8% 10%12% 1% 3% 5% 7% 9%
Fig.7.CompletenesslevelinpercentageintheVIMOSarea(redlimitedregions)intheRmagnituderanges[18,20](upperleft),[19,21](upper
right),[20,22](lowerleft),and[21,23](lowerright).CoordinatesareJ2000.
extended halos around field galaxies (e.g. Bergeron & Boissé MgI line. The respective equivalent widths are 2.8 and 2.5 Å,
1991)andoftheintergalacticmedium(thewellknownLyman- andthe detectionlevelsare3σ. Theabsorptionredshiftscorre-
alpha forest, see e.g. Croft et al. 2002, for a detailed review), spond to z ∼ 0.0496and 0.0595,just behind the Coma cluster
and has allowed to study the internal regions of high redshift and very close to the mean redshift of the BMG structure dis-
galaxies (throughthe so-called Damped Lyman-alphasystems, cussed in Sect. 4.3. The only possible identification is MgI, as
see Khare et al. 2007). We tried to detect the gaseous regions any otheridentificationwouldeither lead to a negativeredshift
just behind the Coma cluster with the same method. However or to a redshift higher than that of the active object. However,
thecaseisparticular,becauseoftheverylowredshiftofComa MgIisalowionizationline,whicharisesonlyinlowtempera-
andofthe wavelengthrangeofourspectroscopicobservations, turehaloesaroundfieldgalaxies,anddefinitelynotinhightem-
whichimpedesusfromusinghigh-redshiftbackgroundtargets, peratureintraclustergas.Thisthereforesuggeststhatthe BMG
since it would have been impossible to disentangle blue rest- intrastructuremediumdoesnothaveahightemperature,consis-
frame wavelength lines at high redshift from redder restframe tentwithitssupposedlyunvirializeddynamicalstatus.
wavelength lines at low redshift. We therefore looked for low
redshift (z ≤ 0.25) active objects in our spectroscopic data. At We have searched for galaxies located within 100 kpc of
thoseredshifts,theonlystrongabsorptionlinesthatcouldarise the line of sight to the VIMOS active object, as this corre-
fromgaseoushalosorinterstellarmedium,andbedetectablein spondstothemeasuredsizeoflowionizationhaloesaroundfield
ourspectraareMgIandCaH&K.Otherusualabsorptionlines galaxies(Fig. 10).Thereis a galaxyatredshift0.0482,located
likeMgIIorCIVhaverestframewavelengthsintheUV,andare ∼50h−1 kpc away from the active object, plus several galaxies
notdetectableintheopticalpartofalowredshiftspectrum.One potentially at z ≤ 0.2 (from photometric redshift estimates by
objectissuitableforourstudyataredhiftof0.2312andlocated Adamietal.2008).Giventhatprobablymostfieldgalaxiesare
spatiallyclosetotheBMG.ThespectrumofthisVIMOSobject surroundedbyagaseoushaloofradius∼90kpc(Kacprzaketal.
isdisplayedinFig.9andisprobablyaSeyfert2galaxy. 2007),wesuggestthatoneormoreofthesegalaxies(partofthe
AscanbeseeninthemagnifiedpartoftheVIMOSspectrum, BMGstructure)alsohasitsownhalo.Theintrastructuremedium
there is an unidentified double line, shortward of the intrinsic oftheBMGthereforeallowsagaseoushalotosurviveinatleast
C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories 1231
195.2 195.0 194.8 194.6 194.4 Table 2. Characteristics of the structures at z ≤ 0.2 not belonging to
Coma(inorderofincreasingredshift).
28.2
α δ N Tentativemass Redshift
(2000.0) (2000.0) (ifvirialized)
28.1 (deg) (deg) (M(cid:6))
194.9867 27.77410 3 6.67E+13 0.0074
194.7974 27.93770 5 3.9E+12 0.0364
28.0
194.8937 27.81673 3 2.0E+12 0.0373
194.7462 27.87338 5 2E+10 0.0381
27.9 194.8206 27.95107 58 2.73E+15 0.0542
194.8296 27.84444 8 1.27E+14 0.0720
194.8266 27.95877 30 1.88E+14 0.0831
27.8
194.8543 27.92563 3 1.8E+12 0.0945
194.7782 28.16868 5 7E+11 0.0986
27.7 194.7517 28.17983 3 5E+11 0.0952
194.9777 27.78710 3 7.2E+12 0.1054
194.8250 27.90883 3 6E+11 0.1078
27.6 194.7706 27.82888 5 8.8E+12 0.1153
194.7916 27.89518 5 5.3E+12 0.1175
194.7713 27.94355 4 1.6E+12 0.1197
Fig.8. u* band image overlayed with X-ray substructures from 194.8791 27.85061 9 1.19E+13 0.1272
Neumann et al. (2003: blue contours), and with the mass map from 194.8550 27.92140 4 1.86E+13 0.1305
Gavazzi et al. (2009: green contours), VIMOS fields (red area), and 194.6917 27.89147 3 1E+08 0.1326
galaxy groups at z ≤ 0.2and outsidetheComa cluster (red:z ≥ 0.1, 194.8280 27.82573 3 4.18E+13 0.1374
green:z ≤ 0.1).ComasubstructuresfromtheM/L = 100SGanalysis 194.8205 28.02470 6 8.23E+13 0.1427
areshownasblackcircles.PFAislocatedaroundα = 194.77degand 195.0567 27.73113 6 1.9E+12 0.1458
δ=27.91deg.CoordinatesareJ2000. 194.7197 28.09023 6 3.7E+12 0.1534
194.7230 27.89113 39 8.5E+12 0.1590
194.7263 27.99955 4 8.13E+13 0.1734
194.9545 27.85410 10 2.7E+12 0.1780
194.8757 27.91133 3 1.1E+12 0.1796
oneof these galaxies.Thisisin goodagreementwith the early 195.0393 27.76123 3 3E+10 0.1838
evolutionarystageoftheBMG. 194.9579 27.83376 7 4.68E+13 0.1854
194.8118 27.89033 6 3E+11 0.1888
194.8523 28.01927 6 2.3E+12 0.1900
4.5.Structuresatz∼0.5 194.6847 27.93407 3 8E+11 0.1925
Inordertocharacterizethevicinityofthez ∼ 0.5galaxystruc-
turedetectedinAdamietal.(1998and2000),wearealsointer- 5.1.Jeansanalysis
estedinthestructuresnearthatredshift.AssuminganM/Lratio
Inordertodeterminetheorbitsofapopulationofgalaxiesina
of 100, we find with the SG method a very extended structure
cluster,weneedthreeingredients:
(coveringthewholefieldofview)atz∼ 0.52andsampledwith
29spectroscopicredshifts(Fig.11),whichwedonotdetectwith – themassprofileofthecluster,M(r),whereristhe3-Ddis-
an M/Lratioof400.Amoreconcentratedstructureisfoundin- tancefromtheclustercenter;
sidethisgalaxylayer(sampledwith6galaxies),locatedontop – the projected number density profile of the galaxies, N(R),
oftheAdamietal.(1998and2000)clustercandidateatz∼0.52. whereRistheprojecteddistancefromtheclustercenter;
We therefore have a z ∼ 0.52 compact structure of galaxies in – the line-of-sight velocity dispersion profile of the galax-
thiszone.Wecannotestimatearobustmassbecauseofthelow iesσ(R).
sampling.However,thefactthatwedetectthisstructureneither
intheweaklensingmassmapofGavazzietal.(2009)norinthe We choose to take M(r) from the literature, scaling all param-
X-raymapsofNeumannetal.(2003)andFormanetal.(private eters to the presently assumed cosmology. Geller et al. (1999)
communication)doesnotsupportaverymassivestructure. haveappliedthecaustictechniquetodeterminethemassprofile
oftheComaclusterouttolargeradii.Thistechniqueisclaimed
toprovideaclustermassprofilewithlittledependenceontheas-
sumedorbitalvelocityanisotropy.Theyhavefitthemassprofile
5. Galaxyorbits withaNFWmodelwithconcentrationc =r200/rs =8.3−+11..27and
massM200 =1.14×1015M(cid:6).AnotherstudybyŁokas&Mamon
ItisimportanttocharacterizetheorbitsoffaintgalaxiesinComa (2003)givesforthedarkmatterprofilec=9andatotalmassat
asthisisapowerfultooltoputconstraintsontheirorigin.While the virialradiusof M200 = 1.19×1015M(cid:6). These are notvery
bright galaxy orbits are relatively well known in clusters (e.g. differentfromtheGelleretal.estimates,andsupportourinitial
Biviano & Katgert2004), nothingis knownaboutthe orbitsof choice.
faintdwarfgalaxies.Wewillselectinthefollowingallgalaxies In both Geller et al.’s and Łokas & Mamon’s analyses, the
withmagnitudesbetweenR=21and23andhavingameasured Coma cluster center is taken to be the position of NGC 4874.
spectroscopicredshift.Weassumeasafirstguessthatthedwarf Since webaseouranalysisontheirmassprofile,wemustcon-
galaxypopulationishomogeneousinordertoperformtheJeans sistently assume the same center throughout our analysis (i.e.
analysis. alsoforN(R)andσ(R)).
1232 C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories
Table 3. Characteristics of the substructures detected in the Coma 195.2 195.0 194.8 194.6 194.4
cluster.
28.2
α δ N Tentativemass Redshift
(2000.0) (2000.0) (ifvirialized)
(deg) (deg) (M(cid:6)) 28.1
194.9353 27.83393 3 3.85E+11 0.0117
194.8504 27.96517 7 4.83E+10 0.0128
28.0
194.8732 27.95258 9 2.83E+12 0.0162
194.6830 27.84417 4 2.49E+12 0.0193
194.8887 27.95417 3 1.17E+12 0.0236 27.9
194.7610 28.15490 3 1.36E+13 0.0283
194.8395 27.82490 4 5.08E+12 0.0342
27.8
27.7
27.6
Fig.11. u* band image withtheVIMOS area(red area) withthe z ∼
0.52 galaxy structure members (green circles) overlayed. Coordinates
areJ2000.
Fig.9.VIMOSspectrumofaz=0.2312activeobject(black)detected
inour survey.Thenoisespectrumisshowninred.Thepossiblefore-
groundMgIlinesatz=0.0496andz=0.0595areindicatedinred.
27.79
27.78
27.77
0.0482
27.76 0.2312
27.75
Fig.12. Number densityprofile N(R ). 1σerrorbarsareshown. The
n
27.74 solidlineisthebest-fitcoreprofile,thedashedlineisthebest-fitKing-
modelprofile(i.e.acoreprofilewithslopea=−1).
195.02 195.00 194.98
cosmological term, v≡(V −V )/(1+V /c), where we have
los c c
Fig.10.Vicinityofthez=0.2312VIMOSactiveobject.Thelargeblue takenVc =7090kms−1 fromGelleretal.
circleshowsa100kpcradiusatz=0.055.Thesmallbluecircleshows We havecheckedthatifwetakeŁokasandMamon’svalue
thepositionoftheactiveobject.Thegreencircleistheonlygalaxywith forM200insteadofGelleretal.’s,theresultsoftheJeansanalysis
aknownspectroscopicredshiftinthefield.Redcirclesaregalaxiesat areessentiallyunchanged.
photometricredshiftsz≤0.2.CoordinatesareJ2000. In order to compute N(R ) one must be aware of the prob-
n
lemsduetoincompleteness.Onecannotusethesampleofspec-
troscopically confirmed members, since it is only poorly com-
We then consider the value of M given by Geller et al. plete. We therefore consider here the sample of dwarfs whose
200
to define the scaling radiusr and the scaling velocityv ≡ membership is based on their photometric redshift z < 0.2.
200 200 p
(GM /r )1/2, so that we can work in the space of normal- We computed a binned N(R ) by counting galaxies within cir-
200 200 n
ized radii r ≡ r/r (R ≡ R/r in projection), and ve- cularannulianddividingthesecountsbytheeffectivearea,i.e.
n 200 n 200
locities v ≡ v/v , where v are the line-of-sight velocities thetotalareaofannuliexcludingthemaskedregions(definedin
n 200
with respect to the cluster mean velocity, corrected for the Adamietal.2006a).TheresultingN(R )isshowninFig.12.
n
C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories 1233
Fig.13.Velocitydispersionprofile.1σerrorbarsareshown.Thesolid Fig.14.ResultsoftheJeansanalysis:χ2 valuesfordifferentβ(cid:7)values.
lineistheJeanssolutionintheassumedNFWmassprofileforβ(cid:7)=1.8, Thesolidlineshowstheresultsobtainedusingthebest-fitcoreprofile
obtainedusingthebest-fitcoreprofileforN(Rn).Thedashedlineisthe forN(Rn),thedashedlineshowstheresultsobtainedusingthebest-fit
JeanssolutionintheassumedNFWmassprofileforβ(cid:7) =1.3,obtained King-modelprofileforN(Rn).Theverticallinesindicatethebest-fitβ(cid:7)
usingthebest-fitKing-modelprofileforN(R ).Thedash-dottedlineis values.Thehorizontallinesrepresenttheχ268%and90%limits.
n
theisotropicJeanssolution(β(cid:7) =1)intheassumedNFWmassprofile,
obtainedwiththebest-fitcoreprofileforN(R ).
n
would obtain marginally acceptable solutions (90%confidence
level) butonly forverylow concentrations,c < 0.3,whichare
We fit N(Rn) with a core model profile, N(Rn) = N0[(1+ excludedbyallanalysesofthemassprofilesnotonlyofComa,
(Rn/Rc)2]−a, with three free parameters. Note that N0 does not butofanygalaxycluster.
entertheJeansequationsolution(itcancelsout),sotheeffective WecanthereforeconcludethatdwarfgalaxiesinComahave
number of interesting parameters is two. The best fit obtained radially anisotropicorbitsevenclose to the cluster center. This
with N0 = 11544r2−020, Rc = 1.2r200 anda = −0.1 is shownin isatvariancewithanyothertypeofgalaxy(Biviano&Katgert
Fig.12.Clearly,thedensityofdwarfgalaxiesisalmostconstant 2004). Late-typegalaxiesdomovealongradiallyelongatedor-
withRn(atleastinthiscentralregion). bitsbutfarfromthecenter.Itistemptingtointerprettheseresults
For the determination of σ(Rn) we consider the spectro- inevolutionaryterms.
scopic sample. We do not need to consider incompleteness, as Dwarfgalaxiescouldbetheremnantsofthosegalaxiesthat
longasthevelocitydispersionisindependentofthegalaxymag- fallintoComawithradialorbits.Theirradialorbitsdrovethem
nitude.Thisislikelytobethecasesincedwarfgalaxiesarevery very near the cluster center where they were morphologically
low-massobjectsandthetimescaleofdynamicalfrictionneeded transformedbysomephysicalmechanismthatis effectiveonly
toslowdownsuchlightobjectsisbeyondtheHubbletimeina nearthecenter(e.g.tidaleffectsrelatedtotheclusterpotential).
clusterlike Coma.We thendetermineσforthegalaxiesincir- Galaxies that we still see today as giant spirals would then be
cularannuli.TheresultingbinnedprofileisshowninFig.13. those thatdid not pass verynear the Coma centerand so man-
aged to survive. Hence their orbits cannot be radially strongly
elongated in the central regions (or their pericenters would be
5.2.Results
small).
We arenowinthepositiontosolvetheJeansequation.Ideally, Unfortunately, these results depend on the solution for
we could perform this analysis splitting our sample in several N(R ). The density profile is not well determined because it is
n
spectromorphologicaltypes,butthestatisticsaretoolowforthis onlywellknownneartheclustercenter.Wehavetriedtoassess
purpose.We notehoweverthatoursampleisdominatedbyab- the systematics by forcing the slope of the core model to the
sorption line galaxies (with or without emission lines) and by valuea = −1,whichcorrespondstothetraditionalKingmodel
relativelyredobjects(∼75%oftheconsideredsampleisredder (King 1962). The fit is still acceptable, with a larger core ra-
thanthe2-σlowerboundoftheColorMagnitudeRedSequence, dius than before, R /r = 1.7 (see Fig. 14). Using this new
c 200
seeFig.23).Ourresultswillthereforemainlyapplytoreddwarf N(R )andthesameM(r)asbefore,weobtainabest-fitconstant
n
galaxies. anisotropy solution of the Jeans equation β(cid:7) = 1.3 ± 0.2 (see
Given M(R )andN(R )wesearchfortheconstantvalueof Figs.12and14).Thissolutionisconsistentwiththeoneprevi-
n n
β≡1−(σ/σ)2thatprovidesthebest-fittoσ(R ),whereσ and ouslyfound,butisalso(marginally)consistentwithisotropy.
t r n t
σ are respectivelythe tangentialand radialcomponentsof the Withthecurrentdata-set,wehavemarginallysignificantev-
r
velocitydispersion.Aconstantβmodelshouldbeadequatehere idencethatthedwarfgalaxiesfollowradiallyanisotropicorbits
giventhatthesampledregiondoesnotextendoverawideradial nearthecenterofComa.Extendingthephotometricdata-setto
range. We preferto expressthe results in termsof β(cid:7) ≡ σ/σ. largerradiiwouldbeveryusefultobetterconstrainthenumber
r t
Thebestfitisgivenbyβ(cid:7) =1.8±0.3,significantlydifferentfrom density profile slope, which seems to play a critical rôle in the
theisotropiccaseβ(cid:7) =1.ThiscanbeseeninFig.14. solutionoftheJeansequation.
Ifweweretoforceisotropywithβ(cid:7) =0,wewouldobtainan Moreover,ifwearbitrarilysplitthedwarfgalaxysamplein
unacceptable solution. On the other hand, if we force isotropy two parts (below and above R = 22.3) a puzzling behaviour
andleavethe concentrationofthe massprofilefreetovary,we appears: dwarf galaxies with R ≤ 22.3 have a mean velocity
1234 C.Adamietal.:VerydeepspectroscopyoftheComaclusterlineofsight:exploringnewterritories
of 6721 ± 350 kms−1, while dwarf galaxies with R ≥ 22.3 more massive (regardingtheir stellar mass) than blue galaxies.
haveameanvelocityof7846±272kms−1.Thetwovaluesap- We findhereforfaintdwarfgalaxiesthesamewellknownten-
pear quite different and a Kolmogorov-Smirnovtest leads to a denciesobservedforgiantgalaxies.
93% probability that the two velocity distributions (below and Dust attenuation and stellar metallicity do not vary signifi-
above R = 22.3) are different. The fainter sample has a mean cantly and are not given in Table 4. Metallicities are compati-
velocity equivalent to that of the G4 group of Adami et al. blewithsolarvalues,ingoodagreementwiththeassumptionof
(2005a:notsampledbytheVIMOSdata).Thisgroupisrelated Adamietal.(2009b).
to the giant galaxyNGC 4911,so we could expectto have the
R ≥ 22.3 dwarf galaxies spatially correlated with the position
of NGC 4911. However, this is not the case: a Kolmogorov- 6.3.Comparisonwithnon-Comaclusterdwarfgalaxies
Smirnovtestdoesnotshowanyevidencefora differentspatial
In this analysis, we classify our dwarf galaxies in four dif-
distribution between the two samples of galaxies. An explana-
ferent classes: low surface brightness galaxies, absorption-line
tionwouldbethattheNGC4911groupislosingitsfaintgalaxy
galaxies,absorption-plusemission-linegalaxies,emission-line
populationalongitstrajectory,spreadingthispopulationallover
galaxies.ThesecorrespondtotheComaLSB,Comaabs,Coma
theComacluster. em+abs, and Coma em lines in Table 4 for the Coma cluster
The velocity dispersionsof the two samples are less differ- andtoNonComaLSB,NonComaabs,NonComaem+abs,and
ent: galaxies with R ≥ 22.3 have 1516+−210739 kms−1 and galax- NonComaemlinesinTable4forthez≤0.1non-Comacluster
ies with R ≤ 22.3 have 1942+−226300 kms−1, higher than the value galaxies.Thesegalaxiesactasa comparisonsample,notbeing
computedin Adamiet al. (1998) of1200kms−1 forR ∼ 17.5. subject to the strong influence of the Coma cluster and being
This is not surprising as we deal here with very low mass ob- atsufficientlylowredshiftnottoexhibitstrongevolutionaryef-
jects,forwhichdynamicalrelaxationmechanismsareveryinef- fects. Figure 16 shows the PDF for the ages of the oldest stel-
ficientinremovinginitialenergyandtheninreducingtheveloc- larpopulationsforthesefourclasses(fromlefttoright)andfor
itydispersion. galaxies inside and outside the Coma cluster (top to bottom).
We observe two trends. First, dwarf galaxies show decreasing
mean ages from absorption-line to emission-line galaxies. We
6. Galaxyspectralcharacteristics note that this trend is also associated with decreasing stellar
masses. Second, the Coma cluster dwarf galaxies seem on av-
6.1.Method
erage older than field dwarf galaxies, even if this trend is not
significant.Conversely,lowsurfacebrightnessgalaxiesseemto
We have produced stacked spectra by computing a weighted
meanoftheindividualcleanspectraavailable,usingtheR-band followtheoppositetrend:theyarebarelysignificantlyolderout-
sidetheComaclusterthaninsideit,wheretheyhaveagessimilar
magnitudeasaweight.Wemeasuredspectralindicesfromthese
toabsorption-linegalaxies.Thereisnosignificantevidenceofan
stacked spectra and fit them, together with the stacked spectral
energydistributionsintheu∗BVRI bands,withalibraryofstel- agetrend,although,takenatfacevalue,themeanageswouldin-
dicateadifferentoriginforatleastpartoftheLSBsinthefield
larpopulationmodels.Theabsorptionindiceswhichwereused
andintheComacluster.Thiswouldbeingoodagreementwith
inthefitareLick_G4300,Lick_Mgb,Lick_Fe5270,Lick_NaD,
aninteraction-inducedoriginforpartoftheComaLSBs(seee.g.
BH_G,andBH_Mgg.Wealsousedthe4000Åbreak:Dn4000.
Adamietal.2009b).
The fit was performed in a Bayesian approach, namely we
Finally, it is interesting to note that the metallicity of the
computed for each stack the probability distribution function
(hereafterPDF)ofeachdesiredparameter,givenaninputlibrary Comaintraclustermediumis0.2Z(cid:6)(Strigarietal.2008),clearly
lessmetal–richthanthedwarfgalaxiesconsideredhere.
of 100000 modelswith uniformcoverageof their physicalpa-
rameters(seeSalimetal.2005;Walcheretal.2008;Lamareille
et al. 2009). These models include in particular complex star
6.4.Influenceoftheintraclustermedium:X-ray
formationhistories.TheIMFisthatofChabrier(2003).
substructures.
We thus derived for each stack the age of the oldest stellar
populations,thestellarmass,thedustattenuation,andthestellar We check in this section if the intracluster medium X-ray sub-
metallicity. structures have an influence on the faint Coma dwarf galaxies
(R≥21).GalaxiesinsideX-raysubstructuresofNeumannetal.
(2003: Coma X line in Table 4) and outside of these substruc-
6.2.PropertiesofComaclusterdwarfgalaxiesasafunction tures(ComanonX lineinTable4)haveverysimilarproperties
ofcolor:red,blue,andred-sequencegalaxies. withoutanysignificantdifferencesgivenouruncertainties.This
In this analysis we consider three different classes of dwarf suggests that whatever influence the cluster substructures may
haveonthefaintdwarfgalaxies,thisdoesnotoccurviathehot
galaxies: red galaxies, red-sequence galaxies, and blue galax-
gasattachedtothesesubstructures.
ies (see Sect. 7.2). Figure 15 shows the spectral energy distri-
butionsandthemeanstellarpopulationmodelfortheseclasses:
red-sequence population, blue galaxies, and red galaxies. This
6.5.Magnitudeversusspectralproperties
figurealsoshowsthePDFfortheagesoftheoldeststellarpop-
ulationsandthestellarmasses. We investigate in this section the possible relation between
Frombluetoredsequence,toredgalaxiesrespectively,the galaxymagnitudesandtheirspectralpropertieswithinthemain
agesoftheComaclusterdwarfgalaxiesrangefromlog(age) = magnituderangecoveredbytheVIMOSdata(R=[21,23]).
9.13to9.53and9.78.Thebluegalaxiesthereforeseemtohave Splitting our Coma sample into R ≤ 22 (the Coma abs
a young stellar population (∼1.3 Gyr) while red galaxies ap- R ≤ 22,andComaem+absR≤ 22linesinTable4)andR> 22
peartobeolderobjects(∼6Gyr).Thestellarmassesrangefrom (the Coma abs R > 22, and Coma em+abs R > 22 lines in
log(M(cid:2)/M(cid:6))=6.21to6.66and6.93.Redgalaxiesaretherefore Table 4), we show that the ages of galaxies of a given class