Distribution of satellite galaxies in high redshift groups
aa r X i v : . [ a s t r o - ph . GA ] J un Draft version October 24, 2018
Preprint typeset using L A TEX style emulateapj v. 03/07/07
DISTRIBUTION OF SATELLITE GALAXIES IN HIGH REDSHIFT GROUPS
Yougang Wang , Changbom Park , Ho Seong Hwang ,Xuelei Chen Draft version October 24, 2018
ABSTRACTWe use galaxy groups at redshifts between 0.4 and 1.0 selected from the Great Observatories OriginsDeep Survey (GOODS) to study the color-morphological properties of satellite galaxies, and investi-gate possible alignment between the distribution of the satellites and the orientation of their centralgalaxy. We confirm the bimodal color and morphological type distribution for satellite galaxies atthis redshift range: the red and blue classes corresponds to the early and late morphological typesrespectively, and the early-type satellites are on average brighter than the late-type ones. Further-more, there is a morphological conformity between the central and satellite galaxies: the fraction ofearly-type satellites in groups with an early-type central is higher than those with a late-type centralgalaxy. This effect is stronger at smaller separations from the central galaxy. We find a marginallysignificant signal of alignment between the major axis of the early-type central galaxy and its satel-lite system, while for the late-type centrals no significant alignment signal is found. We discuss thealignment signal in the context of shape evolution of groups.
Subject headings: dark matter – galaxies:halos – galaxies:structure – large-scale structure of universe– methods : statistical INTRODUCTION
In a cold dark-matter dominated universe, galaxiesform within dark matter halos, and smaller halos formfirst, subsequently these may grow larger by accretingmaterial and/or by merging with other halos. As a re-sult, satellite galaxies are distributed within the darkmatter halo of galaxy groups. Under the assumptionthat there is an unbiased distribution between the satel-lites and dark matter halo, the position of satellitescan be used to determine the shape of the dark mat-ter halo (Carter & Metcalfe 1980; Plionis et al. 1991;Fasano et al. 1993; Basilakos et al. 2000; Orlov et al.2001; Plionis et al. 2004, 2006; Wang et al. 2008),and their kinematics could be used to estimate thehalo mass (Zaritsky et al. 1993, 1997; McKay et al.2002; Brainerd & Specian 2003; Katgert et al. 2004;van den Bosch et al. 2004; More et al. 2009). Or, if aspatial bias between the distribution of satellites andthe underlying dark matter distribution is found, itwould be a very important clue for us in the study ofgalaxy formation theory. The orientation of the satel-lites may also provide useful information on its forma-tion and evolution process. High-resolution simulationshave shown that subhalos tend to align with the ma-jor axis of their host halos (Knebe et al. 2004, 2008a,b;Libeskind et al. 2005; Wang et al. 2005; Zentner et al.2005; Agustsson & Brainerd 2006; Kang et al. 2007;Faltenbacher et al. 2008; Knebe et al. 2010). Such ef-fects could be examined with observations of satellites atdifferent redshifts, to give a more comprehensive test of Key Laboratory of Optical Astronomy, National Astronomi-cal Observatories, Chinese Academy of Sciences, Beijing 100012,China; E-mail: [email protected] School of Physics, Korea Institute for Advanced Study,Dongdaemun-gu, Seoul 130-722, Korea CEA Saclay/Service d’Astrophysique, F-91191 Gif-sur-Yvette,France Center of High Energy Physics, Peking University, Beijing100871, China the theoretical model.The morphology and color of the satellites are di-rectly related to formation history of the host group. Ithas long been known that galaxies exhibit a bimodal-ity in color and morphology: morphologically early-typegalaxies which are typically red and have little or noongoing star formation, and morphologically late-typegalaxies, typically blue with active star formation. Itis well known that galaxy morphology depends on lo-cal density environment. Hubble & Humason (1931)found a larger population of ellipticals and lenticularsin galaxy clusters, and subsequent studies revealed theconnection between galaxy morphology and environ-ment in low redshift clusters of galaxies (Oemler 1974;Weinmann et al. 2006; Park & Hwang 2009), nearbygalaxy pairs (Park et al. 2007, 2008), isolated galaxy-scale satellite systems (Ann et al. 2008), and galaxypairs at high redshifts (Hwang & Park 2009). Thecolor bimodality is noted by numerous studies at bothlow redshifts (Strateva et al. 2001; Blanton et al. 2003;Baldry et al. 2004; Kauffmann et al. 2004) and at highredshifts of z ∼ . r p .
500 kpc for a set of 1489 host galaxies with 3079 satel-lites from the 2dFGRS. This result is in a good agree-ment with similar studies carried out on the SDSS data(Brainerd 2005; Azzaro et al. 2007; Bailin et al. 2008;Agustsson & Brainerd 2010). Yang et al. (2006) andWang et al. (2008) found a strong dependence of thealignment signal on the color of the central and satel-lite galaxies using groups in the SDSS Data Release 2(DR2) and Data Release 4 (DR4) respectively. Thesestudies found that the alignment signal is strongest be-tween red central galaxies (hereafter ‘centrals’) and redsatellites, while the satellites of blue centrals were con-sistent with being distributed isotropically. The align-ment strength is also a function of the group mass, withstronger alignment signal in more massive groups. Otherforms of alignment have also been studied, and some sig-nificant signals are detected. These include the align-ment between neighboring clusters (Binggeli 1982; West1989; Plionis 1994; Wang et al. 2009), between bright-est cluster galaxies (BCGs) and their parent clusters(Carter & Metcalfe 1980; Binggeli 1982; Struble 1990;Niederste-Ostholt et al. 2010), between the orientationof satellite galaxies and the orientation of the cluster(Dekel 1985; Plionis et al. 2003), and between the ori-entation of satellite galaxies and the orientation of theBCG (Struble 1990; Faltenbacher et al. 2007).For obvious reasons, most studies on satellite distribu-tion have been limited to low redshifts, typically z . . . < z < .
5) Luminous Red Galax-ies (LRGs) extracted from the SDSS DR4 and their sur-rounding structures to explore the presence of alignmenteffects. They confirmed that such alignment effect wasalso present at z ∼ .
5. Okumura et al. (2009) inves-tigated the correlation between the orientation of giantellipticals by measuring the intrinsic ellipticity correla-tion function of 83,773 SDSS LRGs at redshifts 0.16-0.47and also found a positive alignment between pairs of theLRGs up to 30 h − Mpc scales.Deep galaxy redshift surveys such as the Great Ob-servatories Origins Deep Survey (GOODS) (Giavalisco2004) now enable us to study the satellite distributionat even higher redshifts. Another aim of our paper is todetect the alignment between the distribution of satellitegalaxies and the orientation of their central galaxy at red-shifts beyond the local universe by using GOODS data.We construct a group catalog using a Friends-of-Friends(FoF) method, then we study the alignment signals withthis sample. Compared with the previous studies, theredshift range in our sample is 0 . z .
0, whichincludes many high redshift groups ( z ∼ OBSERVATIONAL DATA SET galaxy sample
The galaxy sample used here is selected byHwang & Park (2009). Here we give a brief descriptionof the data sample, and we refer the reader to Hwang &Park for more details.We used a spectroscopic sample of galaxies in GOODS.GOODS is a deep multiwavelength survey covering twocarefully selected regions including the Hubble DeepField North (HDF-N, hereafter GOODS-North) and theChandra Deep Field South (CDF-S,hereafter GOODS-South). Total observing area is approximately 300arcmin and each region was observed by NASA’s GreatObservatories ( HST , Spitzer and
Chandra ), ESA’s
XMM-Newton , and several ground-based facilities.
HST obser-vations with Advanced Camera for Surveys (ACS) wereconducted in four bands: B (F435W, 7200s), V (F606W,5000s), i (F775W, 5000s), and z (F850LP, 10,660s).Among the sources in the ACS photometric catalog,Hwang & Park (2009) selected 4443 (2197 in GOODS-South, 2246 in GOODS-North) galaxies whose reliableredshifts are available. In our analysis, a volume-limitedsample of 1332 galaxies with M B ≤ − . . ≤ z ≤ . B -band absolute magnitude M B of galaxies is computed based on the ACS photom-etry with Galactic reddening correction (Schlegel et al.1998) and K-corrections (Blanton & Roweis 2007). Theevolution correction (an increase of 1 . M B per unit red-shift) was also applied to the rest-frame M B (Faber et al.2007). Morphology Classification
Hwang & Park (2009) visually inspected the individual
Bviz band images and
Bvi color images of the galaxies ina volume-limited sample. The galaxy is divided into twomorphological types: early types (E/S0) and late types(S/Irr). Early-type galaxies are those with little fluctua-tion in the surface brightness and color and possess goodsymmetry in morphology, while late-type galaxies showinternal structures and/or variations in the color images. GROUPS OF GALAXIES
A very important step in our investigation is to iden-tify the galaxy groups. There are many different tech-niques to identify groups in the local and distant Uni-verse (Yang et al. 2005; Koester et al. 2007; Li & Yee2008; Wen et al. 2009). All of these methods have theirown advantages and disadvantages, which we do not dis-cuss here. We use the FoF method to find the groups.The FoF algorithm adopted here is that of Eke (2004)and Knobel et al. (2009). There are three adjustable pa-rameters in the FoF algorithm: the linking length b , themaximum perpendicular linking length in physical coor-dinates L max and the ratio between the linking lengthalong and perpendicular to the line of sight R . If twogalaxies i and j with comoving distances d ci and d cj sat-isfy the following two conditions, then they are assignedto the same group. The two conditions, respectively, are θ ij ≤ (cid:18) l ⊥ ,i d ci + l ⊥ ,j d cj (cid:19) (1) Fig. 1.—
Four typical groups found by the FoF method. Thethree parameters b , L max and R are set b = 0 . L max = 0 . R = 13, respectively. The filled and open circles represent thecentral and satellite galaxies, respectively. and | d ci − d cj | ≤ l k ,i + l k ,j . (2)where θ ij is the angular separation between galaxy i and j , and the two parameters l ⊥ and l k are the comovinglinking lengths perpendicular and parallel to the line ofsight defined by l ⊥ = min (cid:20) L max (1 + z ) , b ¯ n / (cid:21) (3) l k = R l ⊥ , (4)where L max is the maximum perpendicular linking lengthin physical coordinates and ¯ n is the mean density ofgalaxies. Since we have a volume-limited sample of galax-ies, it is easy to obtain the value ¯ n . For the typical valueof three free parameters b , L max and R , we adopted themas listed in Table 1 of Knobel et al. (2009).Figure 1 shows a few groups found by the FoF method.The three parameters b , L max and R are set as b = 0 . L max = 0 . R = 13, respectively. In eachgroup, the filled and open circles represent the centraland satellite galaxies, respectively. The brightest galaxyin each group is defined as the central galaxy, and therest are called its satellites. As shown in the top rightpanel of Figure 1, in some cases the “central galaxy” de-fined by this way is not actually located in the centralregion of groups. We call them the Deviation-Center-Group (DCG), and the other groups Located-Center-Group (LCG). The criterion for LCG is d a ≤ g a , where d a is the angular distance between the brightest galaxyand the geometrical center of the group, and g a is the an-gular size of the group. The half angular size of the groupis defined as the angular separation between the geomet-rical center of the group and farthest member galaxy.In the determination of the alignment signal, only theLCGs are used. A total of 206 groups are found in ourvolume-limited sample.In Figure 2, we show the redshift distribution of thegroups in our sample. The distribution has a broad peaknear z ∼ . PROPERTIES OF SATELLITES IN GROUPS
Fig. 2.—
Redshift distribution of the groups.
Fig. 3.—
Early-type fraction of satellites galaxies as a functionof projected distance from the central galaxy. The filled circles andcrosses correspond to the early- and late-type central galaxy cases,respectively.
In Figure 3, we present the early-type fraction of satel-lites as a function of projected distance ( r p ) from thecentral galaxy in groups. The filled circles and crossesare for the early- and late-type galaxy cases, respec-tively. The error bars represent 68% (1 σ ) confidence in-tervals, which are determined with the bootstrap resam-pling method. It is seen that the early-type fraction ofsatellites increases as satellites approach their early-typecentral galaxy. Also, the early-type fraction of satellitesin groups with early-type centrals is higher than thosewith late-type centrals, which indicates that the satel-lite galaxies tend to have morphology similar to theircentrals. Similar results have been obtained for galax-ies in groups and clusters by Weinmann et al. (2006), forgalaxy pairs in general environment by Park et al. (2007,2008) in low redshift samples, and by Capak et al. (2007)and Hwang & Park (2009) in high redshift samples. Inthe group with a late-type central galaxy, the early-typefraction of satellites is nearly constant as the distancefrom the central galaxy changes.In Figure 4, we show the rest frame B -band absolutemagnitude M B of the early- (filled circle) and late-type(cross) satellites as a function of the projected distancefrom the central galaxy. Two solid lines represent the Fig. 4.—
Rest frame B -band absolute magnitude M B of theearly- (filled circle) and late-type (cross) satellites as a functionof the projected distance from the central galaxy. Two solid linesrepresent the median values. Fig. 5.—
Color of the early- (filled circle) and late-type (cross)satellites as a function of the projected distance from the centralgalaxy (left panel). Two solid lines represent the median values.Color distributions for the satellite are shown by histograms in theright panel. median values. The early-type satellites are on the aver-age brighter than the late-type satellites.In the left panel of Figure 5, we present the color ofthe early- (filled circle) and late-type (cross) satellitesas a function of the projected distance from the centralgalaxy. V W and i W of galaxies are K-corrected (toz=0) magnitudes (Blanton & Roweis 2007). Two solidlines represent the median values. In the right panel ofFigure 5, the distribution of color of satellites is shown.It can be clearly seen that the early- and late-type satel-lites occupy the red and blue bumps of the bimodal colordistribution, respectively. QUANTIFYING THE ALIGNMENT
In order to quantify the alignments of objects, we fol-low the method in Brainerd (2005) and compute the dis-tribution functions of the alignment angles, P ( θ ), where θ is the angle between the major axis of the centralgroup galaxy and the direction of a satellite relative tothe centrals. The angle θ is constrained in the range0 ◦ ≤ θ ≤ ◦ , where θ = 0 ◦ (90 ◦ ) suggests that thesatellite lies along the major (minor) axis of the central galaxy.For a given set of the central and satellite galaxies,we first count the total number of central-satellite pairs, N ( θ ), for a number of bins in θ . Next, we construct 200random samples in which we randomize the orientationsof all centrals, and compute h N R ( θ ) i , the average num-ber of central-satellite pairs as function of θ . The randomsamples constructed this way suffer exactly the same se-lection effects as the real sample, so any significant differ-ence between N ( θ ) and N R ( θ ) reflects a genuine align-ment between the orientations of the centrals and thedistributions of their corresponding satellite galaxies.Following Yang et al. (2006) and Wang et al. (2008),we introduce the distribution of normalized pair counts: f pairs ( θ ) = N ( θ ) h N R ( θ ) i . (5)In the absence of any alignment, f pairs ( θ ) = 1, while f pairs ( θ ) > θ implies a satellite is preferentiallyaligned along the major axis of their central galaxy.We quantify the fluctuation using σ R ( θ ) / h N R ( θ ) i ,where σ R is the standard deviation of N R ( θ ), and is es-timated from the 200 random samples. In addition tothis normalized pair count, we also compute the averageangle h θ i . In the absence of any alignment h θ i = 45 ◦ . Ifone finds h θ i < ◦ ( h θ i > ◦ ), it means that the satel-lites are distributed along the major (minor) axis of thecentral galaxy. ALIGNMENT MEASUREMENT
In order to study the alignment signal, the positionangles of the central galaxies are required. We only usethose groups with central galaxy axis ratio b/a < . a and b are the isophotal semi-major and mi-nor axis lengths adopted from the i -band measurementsin the HST/ACS photometric catalog, respectively. Fi-nally, we have 168 central-satellite pairs in the detectionof the alignment signal.Figure 6 shows f pairs for the selected central-satellitesystems. There is a marginally significant signal of align-ment between the orientation of the central galaxiesand the distribution of the satellites. Satellite galax-ies are distributed preferentially along the major axisof their central galaxy. This is also supported by thefact that h θ i = 41 ◦ . ± ◦ .
3, which deviates from thecase of no alignment (i.e. h θ i = 45 ◦ .
0) by 1 . σ . Moreover,a Kolmogorov-Smirnov (KS) test also suggests that anisotropic distribution of satellites in our sample is re-jected with a confidence level higher than 90%. If weremove groups with z > . . σ to 1 . σ .In order to study how the alignment depends on thecentral galaxy properties, we divide our sample intoearly-type central and late-type central cases. Figure 7shows the alignment signals f pairs ( θ ) for the sample withearly- (left panel) and late-type (right panel) centralgalaxies. As can be seen, systems with an early-type cen-tral galaxy shows 2 σ alignment signal. A KS test findsthat the sample with early-type centrals is not isotropicwith confidence level higher than 99%. The distribu-tion of f pairs ( θ ) has an interesting shape, being greaterat both θ ∼ ◦ and θ ∼ ◦ than at θ ∼ ◦ . This could Fig. 6.—
Normalized probability distribution, f pairs ( θ ), of theangle θ between the major axis of the central galaxy and the dis-tribution of satellites. Fig. 7.—
Same as Figure 6, but for different subsamples, dividedby the morphological type of the central galaxy of the group. perhaps be explained by the effect of infall from perpen-dicular filaments on forming galaxies (Brook et al. 2008).Systems with a late-type central galaxy, however, showno significant alignment. Compared with the alignmentsignal detected by Yang et al. (2006)( h θ i = 42 ◦ . ± ◦ . h θ i = 42 ◦ . ± ◦ .
12) in low red-shift groups, there is no significant difference (the confi-dence level for this tiny difference is below 0 . σ ) betweenthe alignment strength in high-z and local groups (for thetotal samples). In other words, no evolution of the align-ment is seen within redshift [0,1]. SUMMARY AND DISCUSSION
Using the FoF group finder, we create a high-redshift(0 . ≤ z ≤ .
0) group catalogue out of a spectroscopicsample of galaxies in the GOODS fields. We also identifythe morphology of the satellite galaxies visually. Themorphologically early- and late-type satellites occupy thered and blue bumps of the bimodal color distribution,respectively. We then study the early-type fraction, themagnitude-radius relation and the color-radius relationsof the satellite galaxies in these groups. We find that theearly-type fraction of satellites in early-host centrals ishigher than those in late-type centrals. The early-typesatellites are also on the average brighter than the late- type satellites.We measured the alignment between the distributionof satellites and the orientation of their central galaxy.We find a marginally significant alignment signal for thewhole sample and for the subsample with early-type cen-trals. However, we do not find any alignment signal forthe subsample with late-type centrals.It is known that the group catalog strongly dependson the three adjustable parameters b , L max and R in theFoF algorithm. In order to study how these parametersaffect the measurement of the alignment signal, we haveadopted five groups of parameter-sets as listed in Table1 of Knobel et al. (2009) in identifying groups, and cal-culated alignment signals for each case. We find thatthe result of the alignment signal from the groups foundby using the five different sets of group parameters arenearly the same. This indicates that our results on groupcentral-satellites alignment is not sensitive to the choiceof adjustable parameters in the FoF algorithm.The measured alignment signal may be compared withnumerical simulation results. Using N-body simula-tions, Jing & Suto (2002); Wang & Fan (2004) foundthat the non-sphericity of dark matter halos are greaterat higher redshift, so we might expect a stronger align-ment strength in the high-redshift groups. However,in our high redshift catalog we only find an alignmentstrength similar to the local groups.There are several possibilities for this result. First,due to the limited number of pairs available, the sam-ple variance is still large, and the detection is marginal.It is still difficult to draw conclusions from this obser-vation. We have 168 central-satellite pairs in total andfind a 1.6 σ alignment signal. To check how the strengthof alignment signal depends on the sample size, we usethe SDSS DR4 data adopted in Wang et al. (2008) tomake a test. The total number of central-satellite pairsin Wang et al. (2008) is 62212, and the alignment signalis 21 σ ( h θ i = 42 ◦ . ± ◦ . ×
2, 168 × .
4, 2 . . σ , respectively. If the same scaling isapplicable to the high redshif sample, we need to increaseour sample size by at least fourfold to reach a significantdetection of about 3 σ .Second, we have assumed that stronger non-sphericityproduce stronger alignments. To certain extent thisshould be true, as there would be no alignment signalwhen the distribution is spherical. However, at differentredshifts the relation between non-sphericity and align-ment may be different. For the same non-sphericity, thealignment might be weaker at higher redshifts due tosome reason (e.g., less time for dynamical adjustment),thus partly compensated for effects of the stronger non-sphericity.Finally, the predictions by Jing & Suto (2002);Wang & Fan (2004) were based on N-body simulations.Inclusion of baryon cooling effect may affect the shape ofthe halo (Kazantzidis et al. 2004; Debattista et al. 2008),and change the conclusions on the shape evolution ofdark matter halos.We may also compare the observational results of align-ment strength at different redshifts in the literature. Forthe group in the range 0 . ≤ z ≤ . ◦ . Using a sample at (0 . ≤ z ≤ . ∼ ◦ . If we extend this trend tosamples at even higher redshifts, the misalignment anglesshould be larger. With such a trend, a weaker alignmentsignal is expected at higher redshifts.In order to further improve our understanding of thespatial distribution of satellites in high redshift groups,we need a large sample. Some undergoing surveys, suchas the Canada-France-Hawaii Telescope (CFHT) LegacySurvey, may extend the sample size significantly, andbring a definite answer to questions related to the dis-tribution of satellites in the high redshift groups. ACKNOWLEDGMENTS
We sincerely thank the referee for the constructiveand detailed comments and suggestions. This work hasstarted during YGW’s visit to KIAS, and he would liketo express his gratitude to KIAS. CBP acknowledges thesupport of the Korea Science and Engineering Founda-tion (KOSEF) through the Astrophysical Research Cen-ter for the Structure and Evolution of the Cosmos (ARC-SEC). YGW acknowledge the support by the YoungResearcher Grant of National Astronomical Observato-ries. XLC acknowledge the support by the NSFC Dis-tinguished Young Scholar Grant No.10525314. YGWand XLC are also supported by the Ministry of Scienceand Technology under the 973 program (2007CB815401,2010CB833004), and the CAS Knowledge InnovationProgram (Grant No. KJCX3-SYW-N2).