Structure of Protocluster Galaxies: Accelerated Structural Evolution in Overdense Environments?
DDraft version November 10, 2018
Preprint typeset using L A TEX style emulateapj v. 11/10/09
INTERNAL STRUCTURE OF PROTOCLUSTER GALAXIES: ACCELERATED STRUCTURAL EVOLUTIONIN OVERDENSE ENVIRONMENTS? Andrew W. Zirm , Sune Toft , and Masayuki Tanaka Draft version November 10, 2018
ABSTRACTWe present a high spatial-resolution
HST /NICMOS imaging survey in the field of a known proto-cluster surrounding the powerful radio galaxy MRC1138-262 at z = 2 .
16. Previously, we have shownthat this field exhibits a substantial surface overdensity of red J − H galaxies. Here we focus onthe stellar masses and galaxy effective radii in an effort to compare and contrast the properties oflikely protocluster galaxies with their field counterparts and to look for correlations between galaxystructure and (projected) distance relative to the radio galaxy.We find a hint that quiescent, cluster galaxies are on average less dense than quiescent field galaxiesof similar stellar mass and redshift. In fact, we find only two (of nine) quiescent protocluster galaxiesare of simliar density to the majority of the massive, quiescent compact galaxies (SEEDs) found inseveral field surveys. Furthermore, there is some indication that the structural Sersic n parameter ishigher ( n ∼ −
4) on average for cluster galaxies compared to the field SEEDs ( n ∼ −
2) This resultmay imply that the accelerated galaxy evolution expected (and observed) in overdense regions alsoextends to structural evolution presuming that massive galaxies began as dense (low n ) SEEDs andhave already evolved to be more in line with local galaxies of the same stellar mass. Subject headings: galaxies: clusters: individual (MRC1138-262) – galaxies: evolution – galaxies: high-redshift – galaxies: structure INTRODUCTION
The internal spatial and velocity distribution of starsis an indicator of the manner in which galaxies haveformed, assembled and evolved. In the local universe,tidal streams, shells and kinematically distinct cores areexamples of archeological clues to past merging and for-mation events (e.g., Peng et al. 2002b; Emsellem et al.2007; van Dokkum 2005; Blanton & Moustakas 2009).Even coarse measures, such as the average stellar surfacemass density within the effective radius (Σ ), correlatewith the star formation rate or the mean stellar age. Gi-ant elliptical galaxies have high stellar mass per unit area(or volume) and show negligible current star formationwhile more diffuse stellar disks and dwarf irregulars areforming stars at sometimes prodigious rates per unit stel-lar mass (specific star-formation rate; sSFR). At higherredshift, analogous relations are already in place (e.g.,Franx et al. 2008). While observationally it remainsdifficult to separate high-redshift galaxies into classicalHubble-types we can now photometrically determine red-shifts, stellar masses and galaxy sizes for large numbersof galaxies at z ∼
2. Such studies (e.g., Zirm et al. 2007;Toft et al. 2007; van Dokkum et al. 2008; Toft et al. 2009;Williams et al. 2010; Mosleh et al. 2011) have found thatquiescent galaxies are in general more dense than theirstar-forming counterparts. Based on observations with the NASA/ESA Hubble SpaceTelescope, obtained at the Space Telescope Science Institute,which is operated by the Association of Universities for Researchin Astronomy, Inc., under NASA contract NAS 5-26555 Dark Cosmology Centre, Niels Bohr Institute, University ofCopenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen, Den-mark; [email protected]; [email protected] Institute for the Physics and Mathematics of the Universe,The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba277-8583, Japan; [email protected]
The origin of this bi-modal distribution of galaxy prop-erties is unclear. It is possible that the quiescent z ∼ z ∼ z ∼ > z ∼ r e ∼ < z ∼ z = 0 is therefore a puzzle.The most massive galaxies in the present-day universeare located at the centers of rich clusters. These galaxyoverdensities were statistically the first to separate fromthe Hubble flow and collapse and are therefore believed tofollow an accelerated timeline for the process of galaxyformation. There is some observational evidence that a r X i v : . [ a s t r o - ph . C O ] O c t Zirm et al.galaxies in the progenitors of clusters, protoclusters, dohave significantly older stars and higher masses thangalaxies in the field at similar redshifts (Steidel et al.2005; Tanaka et al. 2010). Might cluster galaxies, havingformed earlier, be even more dense than field SEEDs?Or, alternatively, the ’fast-forward’ evolution of clustergalaxies may lead to lower density galaxies in protoclus-ters compared to their field counterparts. It is interest-ing, then, to look for dense SEED galaxies in protoclus-ters at redshift z ∼ z = 2 .
16. Broad and narrow-band imaging, both in theoptical and near-infrared, of the field surrounding thepowerful radio galaxy MRC 1138-262 ( z = 2 .
16) haveidentified more than 100 candidate companion galaxies.There are surface-overdensities of both line-emitting can-didates (Lyman- α and H α ), X-ray point sources, sub-mmselected galaxies and red optical–near-infrared galaxies(Pentericci et al. 2002; Kurk 2003; Kurk et al. 2004b;Stevens et al. 2003; Croft et al. 2005). Fifteen of theLy α and 9 of the H α emitters have been spectroscopi-cally confirmed to lie at the same redshift as the radiogalaxy (Kurk et al. 2004a). By obtaining deep imagesthrough the NICMOS J and H filters, which effec-tively span the 4000˚A-break at z = 2 .
16, we have identi-fied a large surface overdensity of red galaxies consistentwith a forming red sequence (Zirm et al. 2008). In thispaper we present a more detailed analysis of the massesand morphologies of galaxies in this field. The article isorganized as follows: in Section § ?? we describe the dataand their reductions, in Section § ?? we present the photo-metric redshifts, stellar population models and morpho-logical fits, in Section § ?? we present the internal stellarmass densities and other derived properties and finally in § ?? we discuss these results in the context of galaxy evo-lution models. We use a (Ω Λ , Ω M ) = (0 . , . H = 73km s − Mpc − cosmology throughout. At z = 2 .
16 onearcsecond is equivalent to 8.4 kpc. All magnitudes arereferenced to the AB system (Oke 1974) unless otherwisenoted. OBSERVATIONS AND DATA REDUCTIONS
NICMOS Imaging
The NICMOS instrument on-board
HST is capable ofdeep near-infrared imaging over a relatively small field-of-view (51 (cid:48)(cid:48) × (cid:48)(cid:48) ). In the case of MRC 1138-262, weknow that galaxies are overdense on the scale of a fewarcminutes (Kurk et al. 2004b; Croft et al. 2005) and arethus well-suited for observations with NICMOS camera 3on HST . We used 30
HST orbits to image seven overlap-ping pointings in both filters and one additional pointingin H alone. These observations reach an AB limitingmagnitude ( m σ ; 10 σ , 0 . (cid:48)(cid:48) m σ = 24 . J and m σ = 25 . H .The same field was imaged in the g ( m σ = 27 . I ( m σ = 26 . HST aspart of a Guaranteed Time program (
HST archive, the IRAF task pedsky and the dither/drizzle package to combine the images in a mosaic. The dither offsets were calculated using im-age cross-correlation and were refined iteratively. Align-ment of the pointings relative to each other was accom-plished using a rebinned version of the ACS I imageas a reference. The final mosaic has a pixel scale of 0 . (cid:48)(cid:48) H -band image for de-tection within SExtractor (Bertin & Arnouts 1996). Weused a 2 . σ detection threshold with a minimum con-nected area of 10 pixels. We also corrected the NICMOSdata for the count-rate dependent non-linearity (de Jong2006). Total galaxy magnitudes were estimated by usingthe MAG AUTO values from SExtractor. We show theoutline of the NICMOS mosaic in Figure 1 along withthe positions of the radio galaxy (yellow star) and star-forming (blue circles) and quiescent (red circles) proto-cluster galaxies.The J − H colors were determined by runningSExtractor (Bertin & Arnouts 1996) in two-image modeusing the H image for object detection and isopho-tal apertures. The J image was PSF-matched to the H band. We also incorporated the two ACS bands(Miley et al. 2006), the Spitzer IRAC bands, U n (Zirmet al. 2008) and V bands from Keck/LRIS, z and R from VLT/FORS2 (Kurk et al. 2004b,a), H band fromNTT/SOFI and J and Ks from Subaru/MOIRCS (Ko-dama et al. 2007). The assembly of the merged multi-band catalog is detailed in Tanaka et al. 2010. FIREWORKS Survey Data and Literature Sample
The FIREWORKS data are described in detail inWuyts et al. (2008). In brief, the survey is K S -bandselected to 5 σ depth of 24.3 (AB) over an area of 113 (cid:3) (cid:48) .In addition to the deep K S band data there is high-quality imaging in each of the U , B , V , I , i , z , J , H , thefour Spitzer /IRAC bands and the 24 µ m Spitzer /MIPSband. The combined multi-band catalog has been used tomeasure precise photometric redshifts, galaxy sizes (Toftet al. 2009) and to model the spectral energy distribu-tions to derive stellar masses, ages and star-formationrates (Damen et al. 2009). We have made three cuts tothe FIREWORKS sample to ensure that we are makingappropriate field-to-cluster comparisons. First, since weare comparing galaxy sizes (densities) we require thatthe galaxies are bright enough to have a reliable sizemeasurement in these ground-based data. Based on thecomparison of size measurements from VLT/ISAAC andHST/NICMOS for the same galaxies, Toft et al. (2009)found that at K ∼ . − . K < . . < z < . < −
11 yr − ). Wenote that after these cuts, the stellar mass distributionremains similar to our protocluster galaxy sample.For further comparison to our protocluster field data,we have compiled a sample of z ∼ HST /NICMOS Camera 3 as we have, Mancinirotocluster Galaxies at z = 2 . D e c ( J ) Figure 1.
Outline of the NICMOS mosaic. The red and blue points mark the locations of the quiescent and star-forming cluster galaxiesrespectively. The yellow star is the radio galaxy MRC1138-262. et al. (2010) used
HST /ACS imaging, van Dokkum et al.(2008) studied
HST /NICMOS Camera 2 imaging whileCassata et al. (2010) use imaging from the WFC3/IRchannel on
HST .We have attempted to translate these published stel-lar mass estimates to the same IMF (Salpeter) and tothe same stellar population synthesis model set (Maras-ton 2005). We have used the analyses of Salimbeni etal. (2009; see their Fig. 1) to derive mean correctionsbetween model sets. The adopted IMF also affects thederived star-formation rates. The offsets in this quan-tity are similar in magnitude to the systematic shift inderived stellar mass (e.g., Erb et al. 2006; Nordon et al.2010), so the specific star formation rate (i.e., the ratio ofstar-formation rate to stellar mass) should be effectivelyunchanged. ANALYSIS
Here we combine the multiband photometric catalogand the NICMOS high spatial-resolution imaging to de-rive physical parameters for individual galaxies. We paredown the total NICMOS galaxy sample to those whichhave a high quality-of-fit for the photometric redshift (us-ing EAZY, Brammer, van Dokkum, & Coppi (2008)), thespectral energy distribution (using FAST, Kriek et al.(2009)) and 2D surface-brightness profile fit (using GAL-FIT, Peng et al. (2002a)). This reduces the galaxy sam-ple from the H -band detected total of 711. We furtherrestrict our attention to those galaxies which most likelylie within the known protocluster (see Section § Photometric Redshifts
The thirteen filter photometric catalog (Tanaka et al.2010) was used to determine galaxy photometric red-shifts. We used the public code, EAZY, to fit a setof model templates to each galaxy’s photometric data(Brammer, van Dokkum, & Coppi 2008). We required that each galaxy have at least 5 colors measured for thephotometric redshift fit. The set of SED templates weused included both galaxy spectral energy distributionsand a narrow emission line spectrum. EAZY uses all lin-ear combinations of the input templates to find the bestphotometric redshift fit. For each fit, EAZY producesthe full redshift probability distribution (see Figs. 2 - 4).We are most interested in the galaxies detected in therelatively small ( ∼ (cid:3) (cid:48) ), but deep, NICMOS H -bandarea. Therefore, we have only included sources detectedin the H NICMOS images. For the target redshiftof z = 2 .
2, the primary strong spectral feature coveredby the photometric data is the 4000˚A break. We notethat even with 13 bands of imaging, photo- z s are notsufficiently precise to determine whether a galaxy is in-side the cluster or not. There are 12 spectroscopicallyconfirmed (emission line) protocluster members withinthe NICMOS mosaic. Of these, four Hα and one Ly α emitters have well-determined photometric redshifts (theremaining members are generally too faint to have de-tections in enough bands). Four of the five photo- z s arearound z ∼ .
1, ranging from 1 . .
1. There is oneclear outlier, the Hα emitter with z phot = 0 . Stellar Population Modeling
Using the calculated best-fit photometric redshifts, weused FAST (Kriek et al. 2009) to fit stellar populationmodel templates to the rest-frame photometry. Thesetemplates consist of a grid of models drawn from theMaraston (2005) set. We chose to use the Salpeter IMF,exponentially declining star-formation histories with τ varying between 10 and 10 years and A V between0 and 3 magnitudes. FAST calculates the best-fittingmodel template among the grid and thereby derives aluminosity-weighted mean stellar age, stellar mass, star-formation rate and extinction for each galaxy. FAST alsooutputs the 1 σ error estimates for each of these fit pa- Zirm et al. r e = n = λ obs [ ◦ A ]0.11.010.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.010.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) Figure 2.
Upper panels, left to right: Galaxy image, GALFIT best-fit model and image residuals after subtraction of the model. Lowerpanel: Photometry and best-fit SED model from FAST (red circles and solid black line Kriek et al. 2009). The shaded regions representthe photometric redshift probability distribution (upper scale) centered at rest-frame 4000˚A. rameters. We show the derived masses and their errorsfor the protocluster galaxies in Table 1. We note thatthe star-formation rates from SED fitting are equivalentto a dust-corrected rest-frame UV SFR and that none ofour quiescent protocluster galaxies have significant de-tections in the MIPS 24 µ m image (20693, PI: Stanford).Since we are only concerned with differentiating the qui-escent and star-forming galaxies, our results are sensitiveonly to catastrophic errors in these SFR determinations. NICMOS Galaxy Sizes and Morphologies
NICMOS camera 3 provides good angular resolutionover its field-of-view (PSF FWHM ≈ . (cid:48)(cid:48) H ≤ . H -band mosaic. We have used our own error map asinput to GALFIT for properly weighting the image pix-els and have masked all neighboring objects. A modelpoint-spread function was created for each of these galax-ies individually by generating a TinyTim simulated PSF(Krist 1993) at the galaxies’ positions in each exposureand then drizzling these PSFs together in exactly thesame fashion as for the data themselves (see Zirm et al.2007). We then executed several different runs of GAL-FIT. We ran fits holding the S´ersic index constant at n = 1 and 4, using a single model PSF for all galaxies, using a stellar PSF instead of the model(s) and holdingthe sky value fixed at zero. For all fits we restricted theS´ersic index, n , to be between 0.5 and 5. The range ofoutput fit values for all these different runs gives us anidea of the variance of the derived parameters due tomodel assumptions. We choose the best fit from theseruns by applying the F-test to the resultant χ ν values.The primary source of systematic offsets in galaxy pro-file and size fitting is the estimation of the local sky value.If the sky is underestimated the galaxy size can be overes-timated, particularly for small faint galaxies. Therefore,we compare our fits where the sky level is a free param-eter with those where we explicitly fix the sky to zero.Many of the “zero-sky” fits fail to converge, for those thatdo converge and have comparable chi-squared values tothe corresponding free fits, we can compare the output r e determinations. It does not appear to be the case thatequally good fits are obtained with and without fittingthe sky. In those five cases where an F-test shows thezero-sky fit to be better, the sizes agree within the errors.Furthermore, none of these where the zero-sky result iscomparably good are for any or our protocluster galax-ies. We note that the GALFIT sky values while non-zero are consistently several orders-of-magnitude smallerthan the values corresponding to galaxy pixels. This sen-sitivity of the fit parameters to even slight variations inthe sky highlights the importance of fitting the local skyrotocluster Galaxies at z = 2 . r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) Figure 3.
Fig. 1 cont. along with the galaxy parameters (even in sky-subtracteddata).We also note that additional scatter to the derived r e values introduced by using a stellar rather than modelPSF is about 10 −
20% and therefore comparable to thescatter on the single-fit measurements themselves.
Sample Selections
We use the surface-brightness profile and stellar popu-lation fits along with the photometric redshifts to definethree sub-samples of the NICMOS-detected galaxies. Wedetail these in order of increasing restriction. The initialsample is defined by a single H -band limit of 26 . GALFIT Sample
We ran GALFIT on the full H -limited galaxy sam-ple. Using the distribution of GALFIT χ ν values for thefits we identify galaxies with good quality-of-fit ( χ ν < r e (= √ ab ) for these well-fit galaxies. We use thisGALFIT sample in our analysis of the dependence of theSersic index, n , on radial position within the cluster (seeleft panel of Fig. 7). We note that we have re-normalizedthe input sigma (error) maps such that the best fits have χ ν ∼ Stellar Populations Sample
To select galaxies with good constraints on both thestellar mass and star-formation rate we have selected an-other sub-sample for the initial 711 galaxies. For targetswith both good photo-z fits and narrow redshift proba-bility distributions (ODDS > .
90, meaning 90% of theprobability distribution is contained within the ∆ z = 0 . χ ν < .
0) based on the SEDfits (for example, see Figs. 2-4).
Probable Protocluster Galaxies
Finally, we have identified a sample of candidate pro-tocluster members using photometric redshifts. In lieuof spectroscopic redshifts, which are difficult to obtainfor z ∼ P ( z ) >
20% at the protocluster red-shift ( z = 2 . sSFR < −
11 yr − ). We showthe galaxy cutouts, the best-fit Sersic model, the modelsubtraction residuals, the broad-band SED fit and photo-metric redshift probability distribution for these galaxiesin Figures 2-4. Zirm et al. r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) r e = n = λ obs [ ◦ A ]0.11.0 R e l a t i v e F λ z phot = ( M )= z R e l a t i v e P ( z ) Figure 4.
Fig. 1 cont.
Stellar Mass Density
The surface (volume) mass density in individual galax-ies is a fundamental property which seems to correlatedirectly with the absence of star formation (e.g., Kauff-mann et al. 2003; Franx et al. 2008). To measure thisquantity requires accurate total mass estimates basedon either stellar velocity dispersion or, much more com-monly, the stellar mass from SED fits to the broad-bandphotometry. Along with the resolved surface-brightnessprofile to determine the galaxy size we can calculate themass density. We must make the assumption that lighttraces mass and that there are no strong gradients in thestellar mass-to-light ratio, i.e., we measure
M/L for theintegrated galaxy light and assume that value applies tothe resolved profile in a single broad-band image.For our galaxies, we calculate the average surface massdensity (Σ in M (cid:12) kpc − ) within the (circularized) ef-fective radius ( r e ) as follows:Σ = M (cid:63) / πr e (1)We present our measurements in Table 1. RESULTS
Distribution of Internal Surface Mass Densities
We have used the combination of photometric red-shifts, stellar population modeling and surface-brightness profile fits to calculate internal surface mass densitiesfor our protocluster sample. We have also added datafrom the literature and from FIREWORKS to constructa well-populated density versus stellar mass diagram inFigure 5. If the published data had a measured star-formation rate in addition to the stellar mass we haverestricted the points plotted to those with low sSFRs(quiescent; log sSFR < −
11 yr − ). In cases where thestar-formation rate was not quoted explicitly, we onlyplot those galaxies which are described as “quiescent”by the authors. This distribution for both our protoclus-ter candidates (yellow circles and squares for quiescentand star-forming) and the field sources from the litera-ture (black squares) and FIREWORKS (blue squares) isshown in Figure 5. For the comparison field sample wehave restricted to FIREWORKS galaxies brighter than K = 21 . z between 1.9 and 2.6. Wehave made the same redshift cut for the literature sam-ple. This redshift range approximately corresponds to a1 Gyr epoch centered on the protocluster redshift.The mean density of the protocluster sample (log < Σ > = 9 .
9) is 0.5 dex lower than that for the fieldsample. For the stellar mass range of our protoclustersample, 10 . M (cid:12) < M (cid:63) < . M (cid:12) , where the massdistributions are similar, we can calculate the distribu-tion of the surface mass densities irrespective of totalstellar mass. We have also used the KS test to calcu-late the probability that the (quiescent) protocluster androtocluster Galaxies at z = 2 . M fl )10 Σ ( M fl k p c − ) Figure 5.
Surface stellar mass density (Σ) vs. total stellar mass for individual galaxies. The blue and black squares are from theFIREWORKS survey (Toft et al. 2009) and other literature (van Dokkum et al. 2008; Cassata et al. 2010; Mancini et al. 2010; Saracco,Longhetti, & Andreon 2009) respectively. The large blue squares are the
K > . × M (cid:12) ). The shaded regions are the local relations for early-type (light red) and late-type galaxies (lightblue). Note that most of the protocluster members have lower densities than their field counterparts. field densities are drawn from the same parent distribu-tion. A fiducial value of P KS <
5% may be consideredsufficient to reject the null hypothesis that they are fromthe same parent. In this case P KS ∼
5% and is there-fore a relatively strong constraint. We have perturbedour measured densities and re-calculated P KS for 10000trials. We show the distribution of P KS in Figure 6. Forthe black (yellow) histogram 30% (60%) of the realiza-tions fall below P KS = 5%. Both histograms have tailsto higher probabilities. Radial Dependencies
In order to assess possible radial gradients in the galaxyproperties within the cluster, with respect to the radiogalaxy, we have constructed the histograms shown in Fig-ure 7. For each of these four physical galaxy parameters:S´ersic n value, surface mass density (Σ), stellar mass( M (cid:63) ) and specific SFR, we have made a single cut inthe galaxy sample and plotted two radial histograms forabove (orange) and below (blue) the chosen cut value.The left panel of Fig. 7 shows the histograms for the fullgalaxy sample appropriate to each parameter, i.e., theGALFIT sample for the Sersic n value. For the rest,the galaxies must also be in the stellar population sam-ple. The right panel shows the histograms for our proto-cluster galaxy sample (both quiescent and star-forming).We have presented these two analyses because while thepresence of field galaxies in the larger samples will diluteany result, the statistics are poor for the quiescent pro-tocluster galaxy sample. Furthermore, this field exhibitsa factor of six surface overdensity of red galaxies (Zirm ( P KS )05001000150020002500 N Figure 6.
The KS probability distributions for the protoclusterversus field galaxy comparison. The black histogram are derivedfrom 10000 runs with the protocluster densities perturbed at ran-dom within the Gaussian errors. The yellow histogram is the samebut excluding the most dense protocluster galaxy. The verticaldashed line marks P KS = 0 .
05. About 55% of the realizations fallbelow the 5% probability. et al. 2008), so the large sample may not strongly dilutetrends.For each pair of histograms we have run the two-sampleKS test to determine whether the distributions are con-sistent with one another. However, for the full sample Zirm et al.(left) the Sersic n distributions are more dissimilar thanfor the other parameters with a low P KS = 5%. Thislow probability seems to be the result of a relatively flatdistribution of the n < . n galaxies. This hint may imply a scenario inwhich the galaxies are deeper within the gravitationalpotential. We discuss this point further below. In thefuture, with spectroscopic redshifts for red galaxies, itwill be possible to repeat this test with better interloperrejection to see if the discrepancy between histograms issignificant for bona fide protocluster galaxies. DISCUSSION
We have presented the combined analysis of 13 bandphotometry and high spatial-resolution NIR imaging inthe field of a known galaxy protocluster at z = 2 .
16. Incases like this where there is a known, confirmed overden-sity and strong statistical evidence for a dominant contri-bution from protocluster members we can make progressdespite the lack of spectrocopic redshifts. We have iden-tified a robust sample of likely protocluster galaxies. Ourconclusions are tentative and tempered by the followingcaveats. We do not have spectroscopic redshifts for thequiescent cluster galaxies. While we have done our bestto isolate the most probable cluster members, they maystill be field galaxies. The cluster sample is also relativelysmall and the results are therefore more suggestive ratherthan statistically robust. Finally, due to the limited arealcoverage of the protocluster (see Fig. 1), the very massivegalaxies may be underrepresented in the cluster sample.These may also be more dense.Possible stellar population model offsets have beenminimized by converting the literature points to the samemodels and IMF we used to fit the cluster galaxies.
Evolution of Galaxy Structure and Stellar MassSurface Density
From the initial discovery of SEEDs, in general fieldsurveys, the primary question has been what evolution-ary processes affect the SEEDs between z ∼ − z = 0 that bring them in line with the local mass-size re-lation. If the dominant process is galaxy merging, thenwe might expect that most of the full galaxy mergers, asopposed to tidal interactions and harassment, may havealready happened. While if the primary determinant ofgalaxy density is the formation redshift, the young Uni-verse being denser, we might expect that cluster galaxieswill be denser than their field counterparts having formedearlier.In our data we see some indication for a differencebetween the profile shapes and density distributions forprotocluster versus field galaxies (Fig. 5). For our sam-ple of likely quiescent protocluster galaxies their stellardensities are lower and perhaps even the Sersic index ishigher than for similarly selected field galaxies. Fromother studies it appears that the majority of field SEEDshave higher axial ratios (flattened) with n ∼ Cosmic Merger Clocks
Several previous studies have shown evidence that pro-tocluster galaxies tend to be more massive and contain older stars than their field counterparts at the same red-shift (e.g., Steidel et al. 2005; Tanaka et al. 2010). Thisadvanced evolution in the cluster environment may alsoextend to the internal structure and dynamics of thegalaxies. At lower redshift, we observe a morphology-density relation, and we may be seeing the beginningsof that relation in the MRC 1138 protocluster. Further-more, the lower densities of the protocluster galaxies sug-gests that the necessary merging has also taken place at aquicker pace than in the field. If we assume then that allgalaxies begin as dense SEEDs at high redshift, we canuse the observed densities as a measure of the “mergerage” of the remnants. A similar idea was put forth byHopkins and Hernquist (2010) to derive the global star-formation history by using the mass profiles of galaxies.They propose that the dense cores of galaxies form earlyin starbursts and the outer parts form in a more quies-cent mode, perhaps in disks. Based on the results fromthis paper we suggest that the resolved mass profiles canbe used further to constrain the redshift of formation andthe subsequent merger history. The ratio of high-densitystellar mass to low-density stellar mass may tell us some-thing about the merger age of a galaxy while the absolutedensity of the highest density stellar components may tellus about the formation redshift. Studies at low-redshifthave already found some correlation between galaxy den-sity and the mean stellar age (e.g., van der Wel et al.2009). With more precise stellar ages for z ∼ REFERENCESBertin, E. & Arnouts, S. 1996, A&AS, 117, 393Blanton, M. R. & Moustakas, J. 2009, ARA&A, 47, 159Brammer, G. B., van Dokkum, P. G., & Coppi, P. 2008, ApJ,686, 1503Cassata, P., et al. 2010, 714, 79Cimatti, A., et al. 2008, A&A, 482, 21Ciotti, L., Lanzoni, B., & Volonteri, M. 2007, ApJ, 658, 65Croft, S., Kurk, J., van Breugel, W., Stanford, S. A., de Vries,W., Pentericci, L., & R¨ottgering, H. 2005, AJ, 130, 867Daddi, E., et al. 2005, ApJ, 626, 680Damen, M., F¨orster Schreiber, N. M., Franx, M., Labb´e, I., Toft,S., van Dokkum, P. G., & Wuyts, S. 2009, ApJ, 705, 617–623de Jong, R. 2006. Correcting the nicmos count-rate dependentnon-linearity. Technical report, STScIDekel, A. & Birnboim, Y. 2008, MNRAS, 383, 119Dekel, A., et al. 2009, Nature, 457, 451Emsellem, E., et al. 2007, MNRAS, 379, 401Erb, D. K., Steidel, C. C., Shapley, A. E., Pettini, M., Reddy,N. A., & Adelberger, K. L. 2006, ApJ, 647, 128Franx, M., van Dokkum, P. G., Schreiber, N. M. F., Wuyts, S.,Labb´e, I., & Toft, S. 2008, ApJ, 688, 770Hopkins, P. F. & Hernquist, L. 2010, MNRAS, 402, 985Kauffmann, G., et al. 2003, MNRAS, 341, 54–69Khochfar, S. & Silk, J. 2006, ApJ, 648, 21Khochfar, S. & Silk, J. 2009, ApJ, 700, 21 rotocluster Galaxies at z = 2 . Khochfar, S. & Silk, J. 2011, MNRAS, 410, L42Kodama, T., Tanaka, I., Kajisawa, M., Kurk, J., Venemans, B.,De Breuck, C., Vernet, J., & Lidman, C. 2007, MNRAS, 377,1717Kriek, M., van Dokkum, P. G., Labb´e, I., Franx, M., Illingworth,G. D., Marchesini, D., & Quadri, R. F. 2009, ApJ, 700, 221Krist, J. 1993, in ASP Conf. Ser. 52: Astronomical Data AnalysisSoftware and Systems II 536Kurk, J. D. 2003. The cluster environments and gaseous halos ofdistant radio galaxies. Ph. D. thesis, Leiden University,P.O. Box 9504, 2300 RA Leiden, The NetherlandsKurk, J. D., Pentericci, L., Overzier, R. A., R¨ottgering, H. J. A.,& Miley, G. K. 2004a, A&A, 428, 817Kurk, J. D., Pentericci, L., R¨ottgering, H. J. A., & Miley, G. K.2004b, A&A, 428, 793Mancini, C., et al. 2010, 401, 933Maraston, C. 2005, MNRAS, 362, 799Miley, G. K., et al. 2006, ApJ, 650, L29Mosleh, M., Williams, R. J., Franx, M., & Kriek, M. 2011, ApJ,727, 5Naab, T., Johansson, P. H., & Ostriker, J. P. 2009, ApJ, 699, 178Nordon, R., et al. 2010, A&A, 518, L24Oke, J. B. 1974, ApJS, 27, 21Oser, L., Ostriker, J. P., Naab, T., Johansson, P. H., & Burkert,A. 2010, ApJ, 725, 2312Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W. 2002a, AJ,124, 266Peng, E. W., Ford, H. C., Freeman, K. C., & White, R. L. 2002b,AJ, 124, 3144 Pentericci, L., Kurk, J. D., Carilli, C. L., Harris, D. E., Miley,G. K., & R¨ottgering, H. J. A. 2002, A&A, 396, 109Salimbeni, S., Fontana, A., Giallongo, E., Grazian, A., Menci, N.,Pentericci, L., & Santini, P. 2009, in American Institute ofPhysics Conference Series, ed. G. Giobbi, A. Tornambe,G. Raimondo, M. Limongi, L. A. Antonelli, N. Menci, &E. Brocato, Volume 1111 of American Institute of PhysicsConference Series 207Saracco, P., Longhetti, M., & Andreon, S. 2009, 392, 718S´ersic, J. L. 1968, Atlas de galaxias australes Cordoba, Argentina:Observatorio Astronomico, 1968Steidel, C. C., Adelberger, K. L., Shapley, A. E., Erb, D. K.,Reddy, N. A., & Pettini, M. 2005, ApJ, 626, 44Stevens, J. A., et al. 2003, Nature, 425, 264Tanaka, M., De Breuck, C., Venemans, B., & Kurk, J. 2010,A&A, 518, A18Toft, S., Franx, M., van Dokkum, P., F¨orster Schreiber, N. M.,Labbe, I., Wuyts, S., & Marchesini, D. 2009, ApJ, 705, 255Toft, S., et al. 2007, ApJ, 671, 285van der Wel, A., Bell, E. F., van den Bosch, F. C., Gallazzi, A., &Rix, H.-W. 2009, ApJ, 698, 1232van der Wel, A., et al. 2011, ApJ, 730, 38van Dokkum, P. G. 2005, AJ, 130, 2647Dokkum, van Pieter G., et al. 2008, 677, 5Williams, R. J., Quadri, R. F., Franx, M., van Dokkum, P., Toft,S., Kriek, M., & Labb´e, I. 2010, ApJ, 713, 738Wuyts, S., Labb´e, I., Schreiber, N. M. F., Franx, M., Rudnick, G.,Brammer, G. B., & van Dokkum, P. G. 2008, ApJ, 682, 985Zirm, A. W., et al. 2008, ApJ, 680, 224Zirm, A. W., et al. 2007, ApJ, 656, 66
Table 1
Protocluster CandidatesObject Photometric Odds H Line Stellar Specific Effective Effective MassObject Redshift (AB) Mass SFR Radius Radius Density(log( M (cid:12) )) (log(yr − )) ( (cid:48)(cid:48) ) (kpc) (10 M (cid:12) kpc − )Quiescent Protocluster Galaxies312 2.24 1.00 21.98 ± . +0 . − . -99.00 0.23 ± . ± . ± . +0 . − . -11.24 0.14 ± . ± . ± . +0 . − . -11.24 0.23 ± . ± . ± . +0 . − . -99.00 0.19 ± . ± . ± . +0 . − . -99.00 0.06 ± . ± . ± . +0 . − . -11.72 0.09 ± . ± . ± . +0 . − . -99.00 0.14 ± . ± . ± . +0 . − . -12.20 0.15 ± . ± . ± ± ± ± ± ± a
700 2.16 · · · ± · · · · · · ± · · ·
463 2.16 · · · ± · · · · · · · · · · · · · · · · · · ± · · · · · · ± · · · · · · ± · · · · · · · · · · · · · · ·
511 2.16 · · · ± · · · · · · · · · · · · · · ·
516 2.16 · · · ± · · · · · · ± · · ·
575 2.16 · · · ± · · · · · · ± · · ·
988 2.16 · · · ± · · · · · · ± · · ·
536 2.16 · · · ± · · · · · · ± · · · · · · ± · · · · · · ± · · ·
457 2.16 · · · ± · · · · · · · · · · · · · · ·
451 2.16 · · · ± · · · · · · · · · · · · · · ·
275 2.16 · · · ± · · · · · · ± · · ·
897 2.16 · · · ± · · · · · · ± · · ·
300 2.16 · · · ± · · · · · · · · · · · · · · ·
311 2.16 · · · ± · · · · · · ± · · ·
361 2.16 · · · ± · · · · · · ± · · ·
365 2.16 · · · ± · · · · · · · · · · · · · · ·
215 2.16 · · · ± · · · · · · ± · · ·
448 2.16 · · · ± · · · · · · · · · · · · · · ·
435 2.16 · · · ± · · · · · · · · · · · · · · ·
431 2.16 · · · ± · · · · · · · · · · · · · · · Confirmed Line-emitting Galaxies53 2.16 · · · ± · · · · · · ± · · ·
263 2.16 · · · ± · · · · · · ± · · ·
270 2.16 · · · ± · · · · · · · · · · · · · · ·
450 2.16 · · · ± · · · · · · · · · · · · · · ·
945 2.16 · · · ± · · · · · · ± · · ·
289 2.16 · · · ± ± · · · · · · · · · · · · · · ·
561 2.16 · · · ± ± · · · ± ± · · · ± · · · · · · ± · · ·
535 2.16 · · · ± · · · · · · ± · · ·
387 2.16 · · · ± · · · · · · ± · · · a Narrow-band selected objects that are not yet spectroscopically confirmed. rotocluster Galaxies at z = 2 . l og ( N ga l ) Sersic n P KS = l og ( N ga l ) Surface Mass Density P KS = l og ( N ga l ) Stellar Mass P KS =
20 40 60 80Distance r from MRC1138-262 (")110 l og ( N ga l ) sSFR -12.2 P KS = l og ( N ga l ) Sersic n P KS = l og ( N ga l ) Surface Mass Density P KS = l og ( N ga l ) Stellar Mass P KS =
20 30 40 50 60Distance r from MRC1138-262 (")110 l og ( N ga l ) sSFR -12.3 P KS = Figure 7.
Radial (measured from the radio galaxy) distributions of galaxies as a function of four derived physical parameters. Top tobottom: S´ersic n value, surface mass density (Σ), stellar mass ( M (cid:63)(cid:63)