Astronomy & Astrophysics manuscript no. draft ©ESO 2016 January 25, 2016 Star formation and black hole accretion activity in rich local clusters of galaxies Matteo Bianconi1, Francine R. Marleau1, Dario Fadda2 1 Institut für Astro und Teilchenphysik, Leopold Franzens Universität Innsbruck, Technikerstraße 25/8, A-6020 Inns- bruck, Austria e-mail: [email protected] 2 Instituto de Astrofisica de Canarias, E-38205 La Laguna, Tenerife, Spain; Universidad de La Laguna, Dpto. de Astrofísica, E-38206 La Laguna, Tenerife, Spain 6 January 25, 2016 1 0 ABSTRACT 2 n Context.Wepresentastudyofthestarformationandcentralblackholeaccretionactivityofthegalaxieshostedinthe a two nearby (z∼0.2) rich galaxy clusters Abell 983 and 1731. J Aims.Weareabletoquantifyboththeobscuredandunobscuredstarformationrates,aswellasthepresenceofactive 2 galactic nuclei (AGN) as a function of the environment in which the galaxy is located. 2 Methods.WetargetedtheclusterswithunprecedenteddeepinfraredSpitzerobservations(0.2mJy@24micron),near- IR Palomar imaging and optical WIYN spectroscopy. The extent of our observations (∼ 3 virial radii) covers the vast ] range of possible environments, from the very dense cluster centre to the very rarefied cluster outskirts and accretion A regions. G Results. The star forming members of the two clusters present star formation rates comparable with those measured in coeval field galaxies. The analysis of the spatial arrangement of the spectroscopically confirmed members reveals h. an elongated distribution for A1731 with respect to the more uniform distribution of A983. The emerging picture is p compatible with A983 being a fully evolved cluster, in contrast with the still actively accreting A1731. - Conclusions. The analysis of the specific star formation rate reveals evidence of on-going galaxy pre-processing along o A1731’s filament-like structure. Furthermore, the decrease in the number of star forming galaxies and AGN towards r the cluster cores suggests that the cluster environment is accelerating the ageing process of the galaxies and blocking t s further accretion of the cold gas that fuels both star formation and black hole accretion activity. a [ Key words. galaxies: clusters: individual (Abell 983, Abell 1731) – galaxies: clusters: general – galaxies: evolution – galaxies: star formation – galaxies: active – infrared:galaxies 1 v 0 1. Introduction more,thehighdensityofgalaxiesintheclusterenvironment 8 promotesfrequentgravitationalencounters,thatinducethe 0 The current paradigm of structure formation predicts that so-calledharassmentprocess.Thedynamicalequilibriumof 6 the galaxy population in clusters is evolving as new mem- the gas is altered and its collapse is facilitated, due to the 0 .bersareaccretedfromthesurroundingfieldregion(Balogh perturbation of the gravitational potential. This leads to 1et al. 1998; Vogelsberger et al. 2014). The dependence of newburstsofSFandhencefurtherejectionofportionofthe 0 the galaxy evolution on the environment in which they are remaininggas,duetostellarwinds.Theefficiencyofgravi- 6 located is proven to be tight. The morphology-density re- tational encounters to trigger new SF episodes increases in 1 lation (Dressler 1980) implies that the environment affects denseenvironmentswithlowvelocitydispersion.Suchcon- : vthe star formation history, color, and structure of galaxies. ditions are found in the filaments, along which the galaxies XiAs a result, the young and active galaxies can be found arefunneledandaccretedinthecluster(Baloghetal.2000; typically in the cluster outskirts, and the passive ones in Diaferio et al. 2001; Okamoto & Nagashima 2003). For ex- r athe cluster core. The processes concurring to this fast age- ample, Fadda et al. (2008) and Biviano et al. (2011) found ing of the galaxies are several. The ram pressure due to a higher fraction of star forming galaxies in the filament the intracluster medium (ICM) (Gunn & Gott 1972; Stein- of Abell 1763, double with respect to the cluster core and hauser et al. 2012) builds on the gas present in the galaxy, outskirts.Wolfetal.(2009),Bivianoetal.(2011),DeLucia as it travels through the cluster. The consequent compres- et al. (2012), Wetzel et al. (2013) showed that long lasting sion of the galaxy gas leads to sudden enhancement of the encounters are more likely to trigger SF in filaments than star formation (SF) but it also favours the gas removal at abrupt processes such as mergers(see also Wijesinghe et al. later stages. Galaxies can also suffer gas losses via gravita- 2012).Hainesetal.(2015)foundthespecificstarformation tionaldisturbance.Larsonetal.(1980)noticedthatthehot rate (sSFR) of massive galaxies in a sample of 30 cluster gaseous halo of the galaxy is stripped as it enters the clus- to be ∼ 30% lower than their counterparts in the field. A ter environment. This process, called galaxy strangulation, subsequent modelling allowed to constrain the quenching prohibits further accretion of gas on the galaxy. Further- Article number, page 1 of 15 A&A proofs: manuscript no. draft Fig. 1. WIRC Ks images of A983 (left panel) and A1731 (right panel), zoomed on the central 7(cid:48) ×5(cid:48) region. The figures are centered on the brightest cluster galaxies. Cluster M [M ] r [Mpc] σ [kms−1] z¯ Members 200 (cid:12) 200 v Abell 983 1.36 ×1015 2.140 1071 0.20 134 Abell 1731 1.92 ×1015 2.408 1201 0.19 91 Table 1. Main properties of the observed clusters. time scale in the range of 0.7−2.0 Gyr, consistent with we selected as targets for our study. ROSAT X-ray images the characteristic accretion time scale of galaxies in clus- ofA983revealtheuniformemissionoftheICM,thattraces ters. Peng et al. (2015) support the scenario in which local the relaxed and virialised state of the cluster. On the other quiescentgalaxieswithstellarmassessmallerthan1011M hand, A1731 shows signs of a less homogeneous X-ray sur- (cid:12) (i.e. the vast majority of galaxies) are primarily quenched face brightness. However, the shallow depth of the X-ray asaconsequenceofstrangulation.Tocorrectlyinterpretthe data does not allow to draw secure conclusions on distri- evolutionofgalaxiesandtheiraccretionhistory,itisthere- bution of ICM. With respect to A983, A1731 presents a fore fundamental to cover observationally the entire extent higher density of galaxies in the core, with two bright clus- of galaxy clusters, with the inclusion of the outskirts and ter galaxies which are surrounded by smaller objects (see the possible accretion structures. In addition, it is impor- Figure 1). tant to consider clusters that do not present evidences of In this paper, we present the study of star formation ongoingmergers.Theintensedisturbanceontheclusterdy- and black hole accretion activity in A983 and A1731. This namicscanhideorcanceltheeffectsoftheenvironmenton galaxies and prohibit the study of the secular accretion of study is based on deep infrared Spitzer observations, near- IR Palomar imaging and optical WIYN spectroscopy. This thegalaxies.Afundamentaldifficultyinstudyingtheseob- paper is structured the following way. In Section 2, we jectsisthecontaminationofinterlopers,i.e.foregroundand present in detail the observational program and the data background galaxies. This issue becomes more important reduction. In Section 3, we present our matched photomet- at larger clustercentric radii where the relative fraction of ric and spectroscopic catalogue. In Section 4, we present interlopers is larger. Therefore, spectroscopic redshift mea- the results of the analysis of our dataset. In Section 5, we surements are required. discuss the scenario that can be drawn from our results. Thestarformationrate(SFR)isaninstantaneousquan- In Section 6, we summarise the results and present fu- tity, directly susceptible to the influence of external pro- ture prospects of the project. Throughout this paper, we cesses, and therefore well suited for the study of environ- assumeH =70kms−1Mpc−1,Ω =0.3andΩ =0.7.At mental effects (Fadda & Duc 2002). Robust measurements 0 M Λ the clusters’ redshift, 1 arcsec corresponds to ∼3.2kpc. of the SFR are necessary, and hence must include both the obscured (via the infrared emission) and unobscured (via the UV and optical emission) star formation activity. In addition,recentstudiescorrelatethepresenceandthechar- acteristics of active galactic nuclei (AGN) to their parent galaxiesaswellastotheenvironmentinwhichthesegalax- 2. The data set iesarelocated.Contradictoryscenarioshavebeenproposed so far in the literature, without any clear predominance The main quantitative properties of A983 and A1731 are (Sabater et al. 2013). The cluster environment allows us to obtainedinSection5andsummarizedinTable1.Thearea study the duality of the environmental effects: on the large covered by our photometric and spectroscopic observations scale, influencing the gas supply, and on the local scale, isshowninFigure2.Themaindetailsoftheseobservations regulating the accretion of the AGN via galaxy-galaxy in- are summarised in Table 2 and Table 3. In the following teractions. sections, we introduce and describe the observing strategy ThegalaxyclustersAbell983and1731(hereafterA983 and the subsequent data reduction and analysis that led to and A1731, respectively) are local rich galaxy clusters that the production of the cluster members catalogues. Article number, page 2 of 15 M. Bianconi, F.R. Marleau, D. Fadda: SF and BH accretion activity in rich local clusters of galaxies 6600..6600 WWIIRRCC 5588..8833 WWIIRRCC IIRRAACC 11−−33 IIRRAACC 11−−33 IIRRAACC 22−−44 IIRRAACC 22−−44 MMIIPPSS 2244 MMIIPPSS 2244 MMIIPPSS 7700 MMIIPPSS 7700 MMIIPPSS 116600 MMIIPPSS 116600 6600..0077 5588..3399 0) 0) 0 0 0 0 2 2 J J C ( C ( E E D D 5599..5533 5577..9944 Abell 983 Abell 1731 5599..0000 5577..5500 115544..5500 115555..0033 115555..5566 115566..0099 115566..6622 115577..1155 119999..00 119999..66 220000..22 220000..99 220011..55 220022..11 RA (J2000) RA (J2000) Fig. 2. Coverage of IR and optical data for A983 (left panel) and A1731 (right panel). The open and filled diamonds correspond to the HYDRA observed targets and the spectroscopically confirmed members, respectively. The solid line polygon delineates the WIRC observations. The shaded region marks the IRAC fields. The 3.6 and 5.8 µm data are outlined with a dotted line, and the 4.5 and 8.0 µm data are outlined with a dashed line. The footprints of the MIPS 24 µm, 70 µm and 160µm field are delimited by the dot-dashed, dash-triple dot, and long dashed line, respectively. The large circles mark the WIYN/Hydra field of view. 2.1. Observations: mid to far-IR with Spitzer IRAC and MIPS Point Spread Function (PSF) for the Spitzer observations is quoted from the IRAC and MIPS Data Handbooks1. The standard Spitzer pipeline processes the raw data, A983 and A1731 were observed as part of the Spitzer pro- outputting basic calibrated datasets (BCDs). The correc- gram 20512 (PI: Dario Fadda). A983 and A1731 were se- tions that were applied include dark subtraction, cosmic lected along with another cluster, Abell 1763 (Edwards ray correction, detector linearization, flat field application et al. 2010), as being rich systems and located at a sim- and muxbleeding correction2. The latter artifact consists ilar redshift in regions with extremely low Galactic emis- of an electronic ghosting that can appear on the detector sion, allowing us to measure the SF from the infrared with due to the delay of the detector in returning to its ground a lower limit of ∼ 1 M yr−1 . Furthermore, these clusters state, e.g. after the read out of a bright source. We applied were targeted for being(cid:12)at low redshift (z ∼ 0.2), allowing additional corrections to the BCDs, in order to obtain a a wide coverage that extends to approximately to 3 virial higher signal-to-noise (SNR) in the images and to avoid radii. false source extraction when running automated software for object detection. The additional corrections follow the procedure described in Fadda et al. (2006) and Edwards The Spitzer IRAC images were taken with the instru- et al. (2010). For the IRAC bands, we corrected for col- mentsetinmappingmode,allowingthesimultaneousimag- umn pulldown, jailbars, stray light and spurious effects of ing of the field of view in the 4 different channels corre- bright sources. We then applied a superflat to the IRAC sponding to 3.8, 4.5, 5.8 and 8 µm. Each pointing of the channel 3 and 4 BCDs. Specifically, each BCD, after mask- telescope was dithered three times to allow for the detec- ing the bright sources, was divided by its median value. tion and removal of transient phenomena, such as cosmic Then, the median value of all these BCDs was computed rays. The IRAC images cover 39.2×39.2 sq. arcmin. on the and each original BCD was divided by this superflat. This plane of the sky, corresponding to 7.3Mpc×7.3 sq. Mpc. procedures helps in removing background gradient present TheMIPSimagesweretakenusingslowtelescopescanning, inthefinalmosaic.FortheMIPSdata,wecorrectedforjail- suitable for covering large portions of the sky. Through- bars, background discontinuities, and pixel distortions. We out the scanning of the sky, the subsequent frames over- also applied a superflat and an additional flat accounting lap each other, for a more efficient removal of cosmic rays. for the impurities on the cryogenic scan mirror. The MIPS The motions of a secondary mirror in a cryogenic bath, astrometry was corrected by extracting the position of the compensate for the oscillations of the telescope during the sourcesinsetsof25consecutiveframesandmatchingthem scanning. The MIPS images overlap the region covered by to the SDSS DR10 r’-band catalogue (Ahn et al. 2014). the IRAC images, and extend beyond that, reaching a size The IRAC sources were extracted using SExtractor of 40.3 ×55.6 sq. arcmin. (7.5×10.4 sq. Mpc). In the case (Bertin & Arnouts 1996). We measured aperture fluxes us- of A1731, the MIPS 70 and 160 µm present a larger final ing a radius of 3 arcsec for each source with a SNR larger footprint ( ∼ 2880(cid:48)2 and ∼ 2700(cid:48)2, respectively), due to than3.5andmultipliedthefluxofeachchannelbythecor- a shift of ∼18’ that occurred to the pointing of the last responding point source aperture correction (see Table 4), Astronomical Observational Request (AOR). In the 24 µm band, we achieved a flux limit of 0.2 mJy in 29.9 ks. Ta- 1 http://irsa.ipac.caltech.edu/data/SPITZER/docs/sitemap/ ble2summarisesthemainpropertiesofournear-tofar-IR 2 http://irsa.ipac.caltech.edu/data/SPITZER/docs/ dataset.TheFullWidthatHalfMaximum(FWHM)ofthe irac/iracinstrumenthandbook/1 Article number, page 3 of 15 A&A proofs: manuscript no. draft 11000000 880000 11000000 Abell 983 Abell 983 Abell 983 880000 880000 660000 er 660000 er er 660000 b b b m m 440000 m u u u N 440000 N N 440000 220000 220000 220000 00 00 00 1122 1144 1166 1188 2200 1122 1144 1166 1188 2200 1122 1144 1166 1188 2200 J Magnitude H Magnitude K Magnitude s 660000 660000 880000 Abell 1731 Abell 1731 Abell 1731 660000 440000 440000 er er er b b b m m m 440000 u u u N N N 220000 220000 220000 00 00 00 1122 1144 1166 1188 2200 1122 1144 1166 1188 2200 1122 1144 1166 1188 2200 J Magnitude H Magnitude K Magnitude s Fig. 3. TheWIRCJ,H,K numbercountsforA983(top)andA1731(bottom).Thehistogramsshowthenumberofsourcesper s bin of magnitudes J, H, and Ks from left to right, respectively. The overplotted filled histograms show the depth of the archival 2MASS data in the same region of the sky covered by the WIRC observations. Cluster Instrument λ [µm] Date Time [min] Coverage[(cid:48)2] FWHM of PSF [(cid:48)(cid:48)] cent Abell 983 IRAC 3.6 4.5 5.8 8.0 2005 Nov 26 72.9 (each) 1600 1.66 1.72 1.88 1.98 MIPS 24 2006 May 07 498.6 2200 5.9 MIPS 70 2006 May 07 498.6 2035 16 MIPS 160 2006 May 07 498.6 2000 40 WIRC J 1.250 2007 Mar 26-27 61.0 1000 1.3 WIRC H 1.635 2007 Mar 26-27 99.3 1000 1.2 WIRC Ks 2.150 2007 Mar 26-27 67.2 1000 1.4 Abell 1731 IRAC 3.6 4.5 5.8 8.0 2005 Jun 13 72.7 (each) 1600 1.66 1.72 1.88 1.98 MIPS 24 05 Dec 08/06 Jun 12 498.6 2200 5.9 MIPS 70 05 Dec 08/06 Jun 12 498.6 2880 16 MIPS 160 05 Dec 08/06 Jun 12 498.6 2700 40 WIRC J 1.250 2008 Apr 24-26 31.5 950 1.3 WIRC H 1.635 2008 Apr 24-26 31.5 950 1.2 WIRC K 2.150 2008 Apr 24-26 41.0 950 1.4 s Table 2. Near to far infrared observations of A983 and A1731. followingSurace(2005).Wecomputedaperturemagnitudes ment design, using an iterative procedure. The flux was using 9 arcsec radius for extended sources and we applied then measured using apertures with radii of 10, 16 and the extended sources aperture correction, following the in- 20 arcsec for 24, 70 and 160 µm, respectively. Multiplica- structionontheSpitzer’sIRACHandbook,forthegalaxies tiveaperturecorrectionsandcolorcorrectionswereapplied presenting an evident extended structure3 (∼1% of the to- accordingly (see Table 4)4. We checked for the presence of tal number of sources). extended sources in the MIPS 24µm by PSF fitting and TheMIPSpointsourceswereextractedusingStarfinder removing the point sources from the original image. We (Diolaiti et al. 2000). The software allowed us to estimate measuredPetrosianfluxesoftheextendedsources(∼2%of thePSFdirectlyfromtheimage,accountingfortheinstru- 3 http://irsa.ipac.caltech.edu/data/SPITZER/docs/ 4 http://irsa.ipac.caltech.edu/data/SPITZER/docs/ irac/iracinstrumenthandbook/30/ MIPS/mipsinstrumenthandbook/50/ Article number, page 4 of 15 M. Bianconi, F.R. Marleau, D. Fadda: SF and BH accretion activity in rich local clusters of galaxies Cluster Pointing Date FOV Centre [R.A., Dec. (J2000)] Integration time [sec] Abell 983 1 2006 Apr 28 10:23:08.087 +59:46:58.00 3×1200 2 2006 Apr 29 10:23:31.913 +59:43:58.00 3×1200 3 2006 Apr 30 10:23:20.009 +59:48:28.00 3×1200 4 2008 Jan 14 10:23:51.566 +59:48:49.00 2×2200 Abell 1731 1 2006 Apr 29 13:22:09.007 +58:08:32.16 3×1200 2 2006 Apr 28 13:23:05.127 +58:11:26.16 3×1200 3 2006 Apr 28 13:22:59.819 +58:10:44.16 3×1200 4 2006 Apr 29 13:23:05.129 +58:10:44.16 3×1200 5 2006 Apr 30 13:24:39.957 +58:13:56.16 3×1200 6 2006 Apr 30 13:23:54.371 +58:08:44.16 3×1200 Table 3. Details of the spectroscopic observations with HYDRA at WIYN. the total number of sources) from the residual image using reach for these observations was M∗ +2, where M∗ is the SExtractor. magnitudeatthekneeoftheluminosityfunction(Edwards et al. 2010). For a cluster of galaxies at z∼ 0.2, a typical valueisKs∗ =15.6(DeProprisetal.2003).Themajorityof 2.2. Observations: near-IR with Palomar/WIRC the cluster galaxies, being passive, present typical near-IR Measuring the near-IR emission is essential in order to es- color indices of J−Ks = 1.7 and H−Ks = 0.8 (Fukugita et al. 1995). Hence, our near-IR photometry reaches (at timate the stellar mass in galaxies. This is because most of the galaxy stellar mass is locked up in the evolved popula- 3σ) a depth of J∼ 19.5, H∼ 18.5, and Ks∼ 17.5, extend- ingapproximatelytwomagnitudesdeeperthanthe2MASS tionwhichemitsmostofitslightintheK-band(Kauffmann archival data (Figure 3). & Charlot 1998). Therefore, deep observations in the near- IR are essential to obtain a robust estimate of the galaxy Inordertocompareourphotometrywith2MASS,aper- stellar mass. We obtained near-IR images of the central ture magnitudes were extracted from the mosaicked J, H 60(cid:48)×90(cid:48) regionofAbell983andAbell1731usingtheWide and Ks images, using SExtractor and a 4 arcsec aperture InfraRed Camera (WIRC) on the Palomar 200 inch tele- radius. This aperture was chosen to match the one used in scope. 2MASS. In Figure 4, we present the comparison between our WIRC aperture magnitudes and the 2MASS ones. We A983 was imaged during two nights of observations on 2007March26-27.TheimagesofA1731wereobtaineddur- measuredahigherdispersion(1σ)ofthedifferencebetween the two magnitudes at the faint end but limited to about ingasecondrunon2008April24-26.Thenightswerepho- tometric,withseeingbetween0.9and1.4"(seeTable2.Fig- 3.5%ofthecorrespondingWIRCmagnitude.Thisvaluede- ure1showstheKsimagesoftheclustercentralregionsand creases towards higher magnitudes (< 15) to about 2.5%. The increasing scatter at the faint end is also due to the Figure2showsthefootprintsoftheinstrument.Theobserv- increase of the noise in the 2MASS data, that are close to ingstrategyforthetworunswasdifferent.ToobserveA983, the detection limit. we scanned the field by moving the field of view along in We then proceeded with the measurement of the Pet- three strips with fixed declination. The J images were ob- rosian magnitudes from our mosaicked images using SEx- tained with an exposure time of 40 s. In H and Ks, two tractor. These magnitude, used for the final catalogue, are coadds of 30 s and four co-adds of 10 s each, respectively, measured on circular apertures within Petrosian radii. The were taken to avoid saturation. The total integration times Petrosian radius is defined as the radius at which the local for J, H, and Ks were 61, 99, and 67 minutes, respectively. surface brightness is a factor of 0.2 times the mean surface For the second run, A1731 was imaged using a 7-position surfacebrightnessinsidetheradius(Petrosian1976).Inor- dithering pattern centered on 9 different subfields. In this der to be able to detect both faint and extended sources, caseweusedanexposuretimeof30sforJandHandthree coaddsof13sinthecaseofKs.Hence,thetotalintegration we set a det_minarea of 5 and det_minthresh of 2.0. We time was 31.5 minutes for J and H and 41 minutes for Ks. used a deblend_mincont of 0.00005 to detect the smallest object that might be close to the largest galaxies. Dark frames were also obtained for each integration time. The same data reduction technique was applied to all threebandsusingapipelinedevelopedandkindlyprovided 2.3. Observations: optical spectra at WIYN byTomJarrett.Amediandarkframewassubtractedfrom the data frames, and correction terms for the flux nonlin- WeobtainedopticalspectraoftheMIPS24µmsourceswith earitywereappliedtocorrectthebias.Amedianskymade fluxgreaterthan0.3mJyandopticalmagnitudeslessthan of a maximum of 10 frames was calculated and subtracted 20.5 in r’ band in two different runs using the Hydra in- from the data frame before flat fielding. The astrometry strument mounted at the WIYN telescope at Kitt Peak was checked using a list of known stars from the Two Mi- National Observatory (KPNO). In total, we observed six cron All Sky Survey (2MASS; Skrutskie et al. 2006) and and four overlapping pointings for A1731 and A983, re- correctedbyaccountingfortherotationaloffsetofthetele- spectively (see Figure 2 and Table 3). scopeandcompensatingforthedistortionoftheWIRCin- During the first run on 2006 April 28-30, we obtained strument.SWarp(Bertinetal.2002)wasusedtomosaicthe spectra of all 24 µm sources in A1731. We also observed framestogether.Fluxcalibrationwasperformedrelativeto threeWIYNfieldsofA983targeting8µmsourcessincethe the2MASScatalog.Themagnitudelimitthatweaimedto MIPSobservationswerenotyetavailableatthetimeofob- Article number, page 5 of 15 A&A proofs: manuscript no. draft 11..00 11..00 11..00 Abell 983 Abell 983 Abell 983 00..55 00..55 00..55 J2MASS 00..00 H2MASS 00..00 Ks2MASS 00..00 − − − C C C WIR WIR WIR J H Ks −−00..55 −−00..55 −−00..55 −−11..00 −−11..00 −−11..00 1100 1122 1144 1166 1188 1100 1122 1144 1166 1188 1100 1122 1144 1166 1188 J H Ks WIRC WIRC WIRC 11..00 11..00 11..00 Abell 1731 Abell 1731 Abell 1731 00..55 00..55 00..55 J2MASS 00..00 H2MASS 00..00 Ks2MASS 00..00 − − − C C C WIR WIR WIR J H Ks −−00..55 −−00..55 −−00..55 −−11..00 −−11..00 −−11..00 1100 1122 1144 1166 1188 1100 1122 1144 1166 1188 1100 1122 1144 1166 1188 J H Ks WIRC WIRC WIRC Fig. 4. The difference between the WIRC aperture magnitudes with the 2MASS aperture magnitudes, plotted against the IRC aperture magnitudes, for A983 (top) and A1731 (bottom). Each panel presents the sources in our field of view with a 2MASS counterpart and after the removal of close pairs. Channel Aperture Aperture Color (2007).Consideringtheavailable101fibers,onaverageten radius ["] corr. corr. fibers were not usable and ten others were assigned to sky IRAC1 3 1.14 ... observation at each pointing. The number of 24 µm target IRAC2 3 1.14 ... sourcesperconfigurationwaslessthanthenumberofavail- IRAC3 3 1.25 ... able fibers. Therefore, these free fibers were positioned on IRAC4 3 1.42 ... r’-bandsourceswithno24µmcounterpart(onaverage8per MIPS24 10 1.167 1.041 configuration). We were able to obtain a total of 281 and MIPS70 16 2.044 1.089 406 spectra for A983 and 1731, respectively. Of these, only MIPS160 20 3.124 1.043 10 and 25 spectra, respectively, did not have identifiable lines or continuum features. Table 4. Aperture and color corrections applied to the IRAC We added archival SDSS DR10 spectra to our WIYN and MIPS sources. data. We selected the SDSS sources that are located in the cluster regions and extending up to ∼ 8 Mpc (∼ 4r ) in 200 clustercentric distance. Our aim was to increase the num- servation. During the second run on 2008 Jan 14 we com- berofsourceswithspectroscopicdataattheoutskirtofthe pleted the observations of the 24µm sources in A983 which cluster.ThelinefittingwasperformedwithanIDLcodefor were not already observed as 8µm emitters by observing both the WIYN and SDSS spectra. The code includes the a further 69 sources. To obtain spectra between 4000 and Markwardt package algorithm for curve fitting. The region 9000 Å we used the red cable and grating [email protected] at an of the local continuum is selected by hand and fit with a angle of 21 degrees, centered at 6500 Å, and integrated straightline.Thelineisfittedontopofthecontinuumwith for 3×1200 s. This gives a dispersion of 2.6 Å pixel−1, a aGaussianfunction.Incaseofblendedlines,thecodehan- spectral coverage of 5400Å, and a resolution of 5.7ÅT˙he dles the fit of multiple Gaussian functions. The measured package DOHYDRA5 was used for the extraction and reduc- flux corresponds to the set of Gaussians that minimizes tionofthespectraincludingthewavelengthcalibration,ap- the χ2 value of the fitted lines. The redshift of each line is plication of the flat-field and fiber throughput correction, allowed to vary and the final redshift assigned to the spec- and the sky subtraction. Cosmic rays were removed using trumisthemeanoftheredshiftofeachline.Theabsorption the la_cosmic task (van Dokkum 2001). The scripts were features are included in the spectral templates. Due to the run for each configuration and for each night. More details level of noise, the fit to the absorption features was possi- on the reduction methods can be found in Marleau et al. ble for only 4% of the total sample of spectra. Accounting fortheabsorptionfeatureshelpsinrecoveringthetotalline 5 F. Valdes 1995, Guide to DOHYDRA, available at flux that is otherwise underestimated. This effect is more http://iraf.noao.edu/tutorials/dohydra/dohydra.html relevant at the high order of the Balmer series (H , H ), δ γ Article number, page 6 of 15 M. Bianconi, F.R. Marleau, D. Fadda: SF and BH accretion activity in rich local clusters of galaxies where the fraction of the absorbed flux to the emission line Column Format Description flux is higher. The H and H lines are dominated by the 1 i6 Catalog Number α β emissionandtheabsorptionaccountsfor1%oftheflux.In 2 a12 R.A. (J2000) case of the H line, this translates to an average SFR of 3 a12 Dec. (J2000) α 0.05M yr−1. The emission line fluxes are used to estimate 4 f10.3 MIPS24 ap 10"(µJy) (cid:12) the star formation rate and their ratio is a stringent diag- 5 f10.3 MIPS24 ap 10" error (µJy) nostics of the presence of AGN. The average flux density 6 f10.2 GALEX NUV (µJy) limit that we achieved was 10−17ergcm−2s−1Hz−1. 7 f10.2 GALEX NUV error(µJy) We compared our line fluxes to the SDSS photometry. 8 f10.3 u(cid:48) (µJy) For each WIYN fiber configuration, we measured the r’- 9 f10.3 u(cid:48) error (µJy) bandmagnitudesusingtheSDSSDR10filterresponsefunc- 10 f10.3 g(cid:48) (µJy) tion. A two arcsec aperture was used, corresponding to the 11 f10.3 g(cid:48) error (µJy) diameter of the SDSS fiber. These factors were computed 11 f10.3 r(cid:48) (µJy) for each spectrum with a SDSS counterpart. The median 13 f10.3 r(cid:48) error (µJy) value of this flux correction factor was then calculated and 14 f10.3 i(cid:48) (µJy) the flux of each spectrum, including those without a SDSS 15 f10.3 i(cid:48) error (µJy) counterpart,wascorrectedaccordingly.Thiscorrectionfac- 16 f10.3 z(cid:48) (µJy) tor ranged from 0.78 to 1.22 for both clusters. In addition, 17 f10.3 z(cid:48) error (µJy) weappliedanaperturecorrectiontoeachHαfluxmeasure- 18 f10.3 J Petrosian (µJy) ment. This aperture correction was obtained by taking the 19 f10.3 J Petrosian error (µJy) ratio of the fluxes extracted from the SDSS r’-band image 20 f10.3 H Petrosian (µJy) using a 2 and a 10 arcsec diameter aperture. The latter 21 f10.3 H Petrosian error (µJy) aperturewaschosenasitcorrespondstotheaveragesizeof 22 f10.3 K Petrosian (µJy) s our sources. 23 f10.3 K Petrosian error (µJy) s 24 f10.2 IRAC1 ap 3" (µJy) 25 f10.2 IRAC1 ap 3" error (µJy) 2.4. Archival data photometry 26 f10.2 WISE1 (µJy) 27 f10.2 WISE1 error (µJy) We obtained archival data to complement the wavelength 28 f10.2 IRAC2 ap 3" (µJy) coverage of our observations. We retrieved the five optical 29 f10.2 IRAC2 ap 3" error (µJy) bands u’ g’ r’ i’ z’ from SDSS DR10, that are needed to 30 f10.2 WISE2 (µJy) measureastellarmassestimateforeachgalaxy,usingspec- 31 f10.2 WISE2 error (µJy) tralenergydistribution(SED)fitting.Wealsoincludedthe 32 f10.2 IRAC3 ap 3" (µJy) near UV (NUV) band from GALEX, containing the emis- 33 f10.2 IRAC3 ap 3" error (µJy) sion from newly formed stars. Additionally, we retrieved 34 f10.2 IRAC4 ap 3" (µJy) the archival WISE 3.6 and 4.6µm data. These bands were 35 f10.2 IRAC4 ap 3" error (µJy) used to check the photometry of the IRAC 1 and 2 chan- 36 f10.2 MIPS70 ap 16" (µJy) nels.The1σdispersionofthedifference,withrespecttothe 37 f10.2 MIPS70 ap 16" error (µJy) IRAC magnitude, between the IRAC and WISE photome- 38 f10.2 MIPS160 ap 20" (µJy) try ranged from 5% to 3.5% from low to high magnitudes, 39 f10.2 MIPS160 ap 20" error (µJy) respectively. 40 f9.3 MIPS24 SNR in 6" aperture Table 5. The IR Source Catalog columns. 3. Cross matching of the source catalogues We produced a final photometric catalogue using as refer- 4. Data analysis ence the positions of the MIPS 24µm sources with a SNR 4.1. The cluster membership inthefirstAiryring(∼6arcsec)largerthan3.Thematch- ing distance used corresponds to half of the first Airy ring The peculiar velocity of each galaxy was computed with (3 arcsec). The matching algorithm associates the position respect to the mean cluster velocity v¯ =¯zc, where z¯is the of the 24µm sources to sources detected in the NUV, the redshift of the cluster and c is the speed of light. The red- five SDSS optical bands, the three near-IR bands, the four shift of the cluster was computed as the mean value of the mid-IR bands, the other two far-IR bands and the spec- redshifts in the range 0.19 <z< 0.21. We used the shift- troscopicredshift.Thespatialdistributionofthesourcesis ing gap algorithm, as described in Fadda et al. (1996) to sparse in the field. Nonetheless, the FWHM of the MIPS determine the cluster membership of the galaxies we ob- 24µm is sufficiently large to hide close pairs of sources. A served. The shifting gap method makes use of both galaxy comparison with the IRAC images, which have a higher velocities and clustercentric distances. The clustercentric resolution, revealed that this problem affected only a small distance to each cluster galaxy was measured from the fractionofthetotalnumberofsources(∼1%).Thespectro- brightestclustergalaxy(BCG).Thegalaxiesweregrouped scopically confirmed members (see Section 4.1) of the two in overlapping and shifting bins of 500kpc from the cluster clusters were not affected by this issue, since all of them center (or wide enough to contain at least 15 or 20 galax- have a unique counterpart in all the wavebands. Table 5 ies each for A983 and 1731, respectively). Then, gaps of summarises the entries of our IR selected source catalogue. 1000kms−1 and800kms−1 forA983and1731respectively, Article number, page 7 of 15 A&A proofs: manuscript no. draft 6000 6000 Abell 983 Abell 1731 4000 4000 2000 2000 s s m/ 0 m/ 0 k k −2000 −2000 −4000 −4000 −6000 −6000 0 1 2 3 4 0 2 4 6 8 Mpc Mpc Fig. 5. The observed galaxies with a spectroscopic redshift in the range 0.15<z <0.25, plotted in the clustercentric distance – velocity space. The filled dots correspond to the cluster members selected with the shifting gap method by Fadda et al. (1996). The right and left panels correspond to A983 and 1731, respectively. were searched in the observed galaxy velocity distribution. 22 22 Abell 983 Abell 1731 These gaps mark the separation between the velocity dis- tribution of the cluster members and the external galaxies. 11 11 This procedure rejects galaxies (interlopers) that have ve- aloncditiAes17b3ig1g,errestpheacnti1v0el0y0.kTmhse−p1raonceddu80re0kwmass−it1e,rfaotredA9u8n3- OIII]/H)β 00 OIII]/H)β 00 g([ g([ til the number of cluster members converged. The advan- Lo Lo tage of this statistical method is the independence from −−11 −−11 any physical assumption on the dynamical state of cluster. The shifting gap method gives us a total of 134 and 91 −−22 −−22 members for A983 and 1731, respectively (see Figure 5). −−22 −−11 00 11 22 −−22 −−11 00 11 22 Theprocedurewasruniterativelyuntilconvergenceonthe Log([NII]/Hα) Log([NII]/Hα) number of selected members. As a comparison, we run also Fig. 6. Emission line diagnostic diagrams using [NII]/Hα and the algorithm of Mamon et al. (2010). This code evaluates [OIII]/Hβ line ratios for A983 and A1731 in the left and right clustermembershipfromthedistance-velocityspace,based panel,respectively.ThedottedlineseparatestheHIIstarform- ing regions, below, from the AGN, above, following the mod- onmodelsofmassandvelocityanisotropyofclusterhaloes ellingbyKewleyetal.(2001).TheselectedAGNcandidatesare obtainedfromacosmologicalsimulation.Thismethodpro- plotted as red asterisks. The triangle identifies the source with- ducedus105membersand77membersforA983and1731. out and [NII] emission lines. As a comparison, we plotted also This code outputs estimates for the virial mass, the veloc- theKauffmannetal.(2003)cutasdashedline,thatselectsless ity dispersion and r , and are quoted in Table 1. The fiercely star forming galaxies. 200 bias towards late-type galaxies of our sample could lead to 11..00 11..00 an overestimate of the velocity dispersion, as these objects Abell 983 Abell 1731 00..88 00..88 present more elongated orbits with respect to the passive population of galaxies (Biviano & Katgert 2004). All the 00..66 00..66 members selected via the Mamon et al. (2010) method are 5] 00..44 5] 00..44 irnitchlumd.eAdsinthtehsehisfatminpglegaspelmecettehdodviaaptpheearsshtifotibnegmgaopreaclogno-- 6]−[4. 00..22 6]−[4. 00..22 3. 3. servativeinidentifyingclustermembers,specificallyathigh [ 00..00 [ 00..00 clustercentric distances, we keep this larger sample for our −−00..22 −−00..22 analysis. −−00..44 −−00..44 −−11 00 11 22 33 44 −−11 00 11 22 33 44 [5.8]−[8.0] [5.8]−[8.0] Fig.7.AGNselectedwiththeSternetal.(2005)colorselection 4.2. Stellar masses criteria.Eachcolorwasevaluatedbyconvertingthefluxineach of the IRAC channels into Vega magnitudes. Galaxies in A983 and 1731 are plotted in the left and right panel, respectively. We estimated the stellar masses for each cluster member from the SED fitting, using the software MAGPHYS (da Cunhaetal.2008),withtheadditionofthestellarlibraries 4.3. AGN and star forming galaxy separation byBruzual&Charlot(2003).Thewavelengthcoverage(up to16bands)andtheprecisionofthespectroscopicredshift TheSFRvalueswereretrievedusingtwodifferentmethods, ensure the high significance of the fit. As described in Sec- i.e. using the computed luminosity in the infrared bands tion 2.3, the spectroscopic observations targeted also non and the H emission line flux. This two methods assume α IR emitters. For these galaxies, we computed a separate that the emission is dominated by star formation. Hence, SED fit that included only the bands from NUV to Ks. beforeweappliedthesemethods,weidentifiedthoseobjects Article number, page 8 of 15 M. Bianconi, F.R. Marleau, D. Fadda: SF and BH accretion activity in rich local clusters of galaxies 110000 110000 4.3.2. The IR color diagnostic Abell 983 Abell 1731 8800 8800 We used the IR color selection proposed by Stern et al. −1yr] −1yr] (2005)toidentifyobscuredAGNinoursample.Thesources M Ο • 6600 M Ο • 6600 withintheso-called"Sternwedge"aretaggedasAGN.For αH) [corr 4400 αH) [corr 4400 euarceh7)c,lunstoetr,idweentififinedd ounsliyngontheeAeGmNisscioanndliidnaetedi(asgeneosFtiigc-. SFR( SFR( In total we identify 17 AGN candidate in A983 and 13 in 2200 2200 A1731. 00 00 00 2200 4400 6600 8800 110000 00 2200 4400 6600 8800 110000 SFR(IR) [MΟ • yr−1] SFR(IR) [MΟ • yr−1] 4.4. The SFR from the total IR luminosity Fig. 8. The comparision of SFR evaluated from the extinction The use of MAGPHYS coupled with our multiwavelength corrected Hα line flux and the SFR based on the total IR lu- catalogue(upto16bandsfromnear-UVtomid-IR)ensures minosity obtained from the SED fit. The triple dot dash line aconsistentmodellingofthegalaxies,thatincludesthedif- corresponds to the relation with slope m=1. The dot dashed ferentphasesoftheISMandthereprocessedstarformation line corresponds to the linear fit of the data. (m = 0.8±0.02 andm=0.74±0.06forA983andA1731,respectively).Theleft emission. We used the Kennicutt relationship (Kennicutt and right panels refer to A983 and 1731, respectively. 1998) to translate the total infrared luminosity L (from IR 8 to 1000µm) estimated by MAGPHYS into a SFR: SFR[M yr−1]=1.7×10−10L /L . (1) (cid:12) IR (cid:12) whose emission was dominated by an AGN. We used three 4.5. The SFR from the optical spectra independent diagnostics for the detection of the AGN: the first is based on the characteristic ratio of optical emission 4.5.1. The extinction correction lines, the second on identifying broad line AGN and the The emission line fluxes were corrected for the internal ab- third utilises an infrared color diagnostic. sorption of each galaxy due to the ISM. The extinction can be estimated using the Balmer decrement, i.e. com- paring the observed and predicted Balmer line fluxes (Hα 4.3.1. The emission line diagnostic at 6563Åand Hβ at 4861Å). A direct measurement of this decrement was possible for 53 and 66 galaxies out of 108 and 223 in the redshift range 0.15<z<0.25 in A983 and The ratio of the fluxes of specific emission lines is useful A1731 field, respectively. We used the median value of the for discerning the source of the ionizing radiation causing Balmerdecrementofthesesubsetsfortheremaininggalax- such lines. A clear signature of an active galactic nucleus ies in which one of the two lines was not detected. The is a high value for the flux ratio of [NII]/H and [OIII]/H colorexcessE(B-V)ofeachsourceiscomputedbycompar- α β (Baldwin et al. 1981), with respect to more moderate val- ing the ratio of the observed lines H and H , FHα,β, with ues in case of a star forming region. In order to iden- α β o the predicted unobscured value via the equation: tify narrow-line AGN in our spectroscopic sample, we used t[OheIIdI]i/aHgnβovsteircsudsia[gSrIaI]m/Hsαo,f a[NppIIl]y/iHngαtvheerssuesle[cOtiIoInI]/cHritβerainodn E(B−V)= 2.5 log(cid:32)FHoα/FHoβ(cid:33) (2) by Kewley et al. (2001). These cuts result from the mod- k(Hα)−k(Hβ) FHα/FHβ elling of starburst galaxies with stellar population models i i (PEGASE version 2.0), producing the ionizing radiation, where FHα,β is the intrinsic unobscured flux of H and H , i α β and with a detailed self-consistent photoionization model respectively, and k(λ) is the reddening curve as a function (MAPPINGS III). The AGN are modeled as 500kms−1 of the wavelength. Here, the intrinsic unobscured line ratio radiative shocks (Kewley et al. 2001). For A983, we identi- FHα/FHβ is set equal to 2.87, assuming case B recombi- fied 11 and 6 AGN using the two diagrams with [NII] and naitioniand T = 104K (Osterbrock 1989). The reddening [SII],respectively.OnlytwocandidateAGNarecommonto curve k(λ) was taken from Calzetti et al. (2000) for star- both diagrams. In the case of A1731, we identified 8 and 7 burst galaxies in the wavelength range from 0.12−2.2µm. AGN using both [NII] and [SII] diagrams, respectively (see These quantities allow us to express the extinction as a Figure 6). In this case, only one candidate AGN is com- function of wavelength via: mon to both diagrams. We consider only the AGN selected via the [NII] as the [SII] lines (with observed wavelength ∼ 8000Å) are located towards the end of the observed A(λ)=E(B−V)k(λ), (3) waveband range(∼ 9000Å), where the sky emission lines whereA(λ)isthemeanextinctioninunitsofmagnitudeat dominate. Furthermore, Kewley et al. (2006) showed that aspecificwavelengthλ.UsingtheIDLroutinecalz_unred, the [NII] selection method is more sensitive to low energy wecomputedthedereddenedfluxforeachemissionline.We AGN. In addition to these narrow line emission AGN, we assumedthedefaultvaluefortheeffectivetotalobscuration visually identify 5 and 4 broad emission line AGN in A983 for starburst galaxies R = 4.05, which include the effect V and 1731, respectively. Broad emission line AGN present of extinction, scattering, and the geometrical distribution permitted line width in the range 103 −104kms−1, while of the dust relative to the emitters (Calzetti et al. 2000). narrow emission line AGN show permitted and forbidden ThevalueofA(λ)rangesbetween0.5and3.5anditsmean line width of 102−5×102kms−1 (Hao et al. 2005). is 1.5. Article number, page 9 of 15 A&A proofs: manuscript no. draft 10 10 8 8 IR IR 6 6 Ha Ha N N 4 IR+Ha 4 IR+Ha 2 2 0 0 33 33 Abell 983 Abell 1731 22 22 ] ] 1 1 − 11 − yr yr 11 MΟ • M Ο • [ 00 [ R R 00 F F S S g −−11 g o o −−11 L L −−22 −−22 −−33 99..00 99..55 1100..00 1100..55 1111..00 1111..55 0 2 4 6 8 99..00 99..55 1100..00 1100..55 1111..00 1111..55 0 2 4 6 8 M [M ] N M [M ] N star Ο • star Ο • Fig. 9. The M −SFR relation for the A983 and 1731 on the left and right panel, respectively. The histograms on the top and ∗ right side of each plot show the stellar mass and SFR distribution of the members. The SFR from the total IR luminosity only is plotted as red diamonds, the SFR from the extinction corrected Hα as open blue triangles, the combined SFR as open green squares. The filled blue triangles mark the sources for which the mean extinction correction was used. The light blue shaded area corresponds to the relation with uncertainties found by Noeske et al. (2007). The red dot-dashed and the blue dashed lines mark the SFR limit evaluated from the detection limit of the IR 24µm and optical spectral SNR, respectively. 4.6. The star formation rate from the Hα emission line luminosity to the uncorrected H emission, following Ken- α nicutt et al. (2009). In the limit of complete obscuration WecalculatedtheSFRfromtheHα lineflux(correctedfor (satisfied in the most active galaxies, luminous and ultra- apertureandextinction)usingtheHα-SFRrelationderived luminous infrared galaxies), the SFR was evaluated using by Kennicutt (1998): the total IR luminosity only. The SFR of the galaxies with optical spectral lines only was retrieved from the corrected SFR(Hα)[M(cid:12)yr−1]=7.9×10−42L(Hα)[ergs−1]. (4) Hα emission. Figure 9 presents the M −SFR relation for the spec- Thisrelationappliesespeciallywhenconsideringyoungand ∗ troscopically confirmed members, along with the M and massive stellar populations, under the assumption that the ∗ SFR distributions for each of the two clusters. We over- emission lines are tracers of the ionizing flux from newly plotted as a comparison the relation obtained by Noeske formed stars. Eq. (4) was introduced and calibrated using et al. (2007) who used a sample of field galaxies at reshift a Salpeter initial mass function (IMF) and over the mass of 0.2<z<0.45. The different colours encode the methods range 0.1 < M/M(cid:12) < 100. We estimated star formation used for estimating the SFR. For both clusters, the mean rates in the range 0.1 − 200M(cid:12)yr−1. In agreement with SFRofthemembersiscompatiblewiththemeanSFRofa Marleau et al. (2007), we find that these SFR are strongly coeval sample of field objects (Log(SFR)∼0.8M yr−1 for correlated with extinction. (cid:12) the clusters with respect to Log(sSFR) ∼ 0.5M yr−1 for (cid:12) the field). 5. Results 5.1. SFR from IR vs H 5.2. The effect of the dynamical state of the cluster on SFR α Figure 8 shows the comparison of the SFR obtained from The clusters that we are considering present a clear differ- theIRluminosityandfromtheextinctioncorrectedH line ence in their dynamical state. This difference is evident in α flux. An underestimate of the corrected flux, and hence of thecomparisonoftheclustercentric-velocityplot(Figure5) the SFR computed using the H , can be seen for galax- ofthetwoclusters.A983presentstheclassicaltrumpet-like α ies with SFR > 30M yr−1. As a consequence, the lin- shape of the galaxy distribution, with a clear separation (cid:12) ear fit of the data presents a slope of m = 0.8±0.02 and between the hosted objects and the external ones. A1731 m = 0.74 ± 0.06 for A983 and A1731, respectively. The showsalessuniformdistributionofthemembers.Thecen- extinction correction is more effective for normal galaxies tral high number density of members becomes more sparse thanforhighstarformingdustyones(Calzettietal.2000). at a clustercentric distance of about ∼ 2Mpc ∼ 1r ). 200 In order to properly account for the obscured and unob- TheanalysisofthepeculiarvelocityofA1731galaxymem- scuredstarformation,weaddedtheSFRfromthetotalIR bers allows us to exclude the presence of neighboring clus- Article number, page 10 of 15