The Fastest Galaxy Evolution in an Unbiased Compact Group Sample with WISE
aa r X i v : . [ a s t r o - ph . GA ] J a n Draft version August 7, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
THE FASTEST GALAXY EVOLUTION IN AN UNBIASED COMPACT GROUP SAMPLE WITH
WISE
Gwang-Ho Lee , Ho Seong Hwang , Jubee Sohn , and Myung Gyoon Lee Department of Physics and Astronomy, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea School of Physics, Korea Institute for Advanced Study, 85 Hoegiro, Dongdaemun-gu, Seoul 02455, Republic of Korea and Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, MA 02138, USA
Draft version August 7, 2018
ABSTRACTWe study the mid-infrared (MIR) properties of galaxies in compact groups and their environmentaldependence using the
Wide-field Infrared Survey Explorer (WISE) data. We use a volume-limitedsample of 670 compact groups and their 2175 member galaxies with M r < − .
77 and 0 . < z < . WISE µ m with a signal-to-noise ratiogreater than 3. Among the 1541 galaxies, 433 AGN-host galaxies are identified by using both opticaland MIR classification scheme. Using the remaining 1108 non-AGN galaxies, we find that the MIR[3 . − [12] colors of compact group early-type galaxies are on average bluer than those of clusterearly-type galaxies. When compact groups have both early- and late-type member galaxies, the MIRcolors of the late-type members in those compact groups are bluer than the MIR colors of cluster late-type galaxies. As compact groups are located in denser regions, they tend to have larger early-typegalaxy fractions and bluer MIR color galaxies. These trends are also seen for neighboring galaxiesaround compact groups. However, compact group member galaxies always have larger early-typegalaxy fractions and bluer MIR colors than their neighboring galaxies. Our findings suggest that theproperties of compact group galaxies depend on both internal and external environments of compactgroups, and that galaxy evolution is faster in compact groups than in the central regions of clusters. Subject headings: galaxies: evolution – galaxies: groups: general – galaxies: interactions INTRODUCTIONCompact groups of galaxies contain several galaxieswithin just a few tens of kilo-parsec scales. This makescompact groups have extremely high galaxy number den-sities (the median galaxy number density of log( ρ/ [ h − Mpc ])= 4 .
36, Sohn et al. 2016), higher than the galaxynumber densities of cluster and group environments(the median galaxy number density of log( ρ/ [ h − Mpc ])= 1 .
97 for clusters and groups in the A2199supercluster, Lee et al. 2015; Sohn et al. 2015). Thevelocity dispersions of compact groups are much smaller( <
879 km s − with a median value of 194 km s − witha standard deviation of 156 km s − , Sohn et al. 2016,Table 7) than those of clusters ( ∼ − − with a median value of 778 km s − and a standarddeviation of 208 km s − , Rines et al. 2013). Thesetwo characteristics make compact groups an idealenvironment for frequent interactions and mergers be-tween galaxies (e.g., Rubin et al. 1990; Rodrigue et al.1995; Amram et al. 2007; Coziol & Plauchu-Frayn2007; Plauchu-Frayn & Coziol 2010; Gallagher et al.2010; Konstantopoulos et al. 2012; Sohn et al. 2013;Vogt et al. 2015).Numerical simulations suggested that galaxies in acompact group should merge into a single ellipti-cal galaxy within a few Gyrs, causing the compactgroup to disappear (Barnes 1985, 1989; Mamon 1987).Gallagher et al. (2010) also reached a similar conclusionthrough their study of star cluster age-dating and neu-tral hydrogen content in the Hickson compact group 31; [email protected], [email protected] the compact group should merge into a single ellipticalwithin 1 Gyr (see also Rubin et al. 1990). Kroupa (2015)suggested that the abundance of compact groups shoulddecrease as the redshift decreases from z ≃ . z = 0 inthe ΛCDM universe. However, Sohn et al. (2015) foundthat the abundance of compact groups changes littlewith redshifts at 0 . < z < .
22, using a spectroscop-ically complete sample of 332 compact groups. Kroupa(2015) also suggested an alternative scenario that com-pact groups do not merge, which predicts a constantabundance of compact groups at z < .
1. However, thisalternative scenario is based on the assumption of a uni-verse without dark matter.Diaferio et al. (1994) suggested that compact groupsreplenish themselves with new members from the sur-rounding environment, thereby extending their life-times to the current epoch. The finding of Sohn et al.(2015), constant abundance of compact group at 0 . 22, can be explained by this replenishmentmodel. This replenishment model is supported byseveral observational findings that showed that many( > m r < 21) of Sloan Digital Sky Survey (SDSS) Data Re-lease 6 (DR6, Adelman-McCarthy et al. 2008), including77,088 compact group candidates and their 313,508 ten-tative member galaxies. Their sample is the largest onecurrently, but it is highly contaminated with interlop-ers ( > . < z < . 19 using SDSS DR12 spectroscopicdata supplemented by additional redshifts from the lit-erature and from FLWO/FAST observations. They ap-plied a friends-of-friends algorithm to identify the com-pact groups without using the isolation criterion. As a re-sult, their catalog successfully contains nearby ( z < . Spitzer IRAC (3 . − . µ m) color space with a statisticallyevident gap, so-called “canyon”, between star-forminggalaxies with MIR red colors and quiescent galaxieswith MIR blue colors. Recently, Zucker et al. (2016)newly identified the canyon galaxies in the WISE color-color diagram. This gap may suggest the acceleratedgalaxy evolution in compact groups because the gapis not seen in comparison galaxy samples from field, from interacting pairs, and from the center of the Comacluster (Walker et al. 2010, 2012, 2013). Cluver et al.(2013) found that compact group galaxies in the gapmostly show enhanced warm H emission from the ob-servations with the Spitzer Infrared Spectrograph, whichcould be caused by shock heating (Appleton et al. 2006;Cluver et al. 2010). This result implies that shock heat-ing may be responsible for rapid evolution of galaxies incompact groups.However, these results are mainly based on the cat-alogs of compact groups selected using Hickson’s crite-ria, which can introduce a sample bias. We thereforerevisit the issues on the comparison of compact groupsgalaxies with cluster and field galaxies using the unbi-ased sample of compact groups from Sohn et al. (2016).In this paper, we use the Wide-field Infrared Survey Ex-plorer ( WISE , Wright et al. 2010) mid-infrared (MIR)data to study the environmental effects on compactgroup galaxies and the relation between compact groupsand their surrounding environments. The MIR data areuseful indicators of mean stellar ages (Piovan et al. 2003;Ko et al. 2009), especially for compact group galaxiesthat are mainly dominated by old stellar populations.Red-sequence galaxies with small amounts of young ( < µ m with asignal-to-noise ratio (S/N) greater than 3. This sam-ple is larger than those of previous MIR studies. Forexample, the samples of Walker et al. (2012) and sev-eral previous studies (Johnson et al. 2007; Bitsakis et al.2010, 2011; Walker et al. 2010) have less than 50 com-pact groups, the sample of Walker et al. (2013) has 99compact groups and 348 member galaxies, and the sam-ple of Zucker et al. (2016) has 163 compact groups and567 member galaxies.Section 2 describes the compact group sample andcomparison samples of cluster/field galaxies. In Section 3we compare the properties of galaxies in compact groupswith those in other environments, and investigate howenvironment affects the properties of galaxies in compactgroups. In Section 4 we discuss the environmental effectson galaxy evolution in compact groups and the relationbetween compact groups and their surrounding environ-ments. In Section 5 we summarize our results and presentconclusions. Throughout, we adopt flat ΛCDM cosmo-logical parameters: H = 70 km s − Mpc − , Ω Λ = 0 . m = 0 . DATASohn et al. (2016) constructed a catalog of compactgroups using the spectroscopic sample of SDSS DR12galaxies with m r < . 77. The completeness of thespectroscopic data in SDSS is low for bright galaxies at m r < . TABLE 1SDSS-related Physical Parameters of Compact Group Galaxies Host Group a Galaxy ID R.A. Decl Redshift M br u − r c Morph d V1CG001 1237648705657307198 198.233322 1.007515 0 . ± . − . ± . 21 2 . ± . 10 2V1CG001 1237648705657307347 198.229294 1.010990 0 . ± . − . ± . 01 2 . ± . 06 1V1CG001 1237648705657307315 198.218872 1.019821 0 . ± . − . ± . 01 2 . ± . 04 1V1CG002 1237661126155436166 139.939529 33.745014 0 . ± . − . ± . 06 2 . ± . 01 1V1CG002 1237661126155436169 139.922531 33.738174 0 . ± . − . ± . 05 2 . ± . 01 2V1CG002 1237661126155436164 139.945221 33.749741 0 . ± . − . ± . 01 2 . ± . 01 1V1CG003 1237661136886431890 154.733276 37.285831 0 . ± . − . ± . 01 1 . ± . 03 2V1CG003 1237661136886497282 154.743576 37.300205 0 . ± . − . ± . 01 1 . ± . 01 2V1CG003 1237661136886497353 154.747711 37.308155 0 . ± . − . ± . 01 1 . ± . 02 2 Note . — The full table is available in the online journal. A portion is shown here for guidance regarding its form and content. a Group ID from Table 5 of Sohn et al. (2016). b The k z =0 . -, and evolution-corrected r -band Petrosian absolute magnitudes. c The SDSS extinction- and k -corrected model magnitudes. d Galaxy morphology. 1 indicates an early-type galaxy, while 2 indicates a late-type galaxies. the literature (Hill & Oegerle 1993, 1998; Wegner et al.1996, 1999; Slinglend et al. 1998; Falco et al. 1999; seealso Hwang et al. 2010) and from FAST observations atFLWO (Sohn et al. 2015). They applied the friends-of-friends algorithm with a projected linking length of∆ D = 50 h − kpc and a radial linking length of | ∆ V | = 1000 km s − , and constructed a magnitude-limited ( m r < . 77) sample of 1588 compact groupswith each consisting of three or more member galaxiesat 0 . < z < . 19. This new catalog contains 18 timesas many systems and reaches three times the depth ofthe catalog of Barton et al. (1996), which is also basedon the friends-of-friends algorithm.Sohn et al. (2016) also constructed two volume-limitedsubsamples: the V1 sample of galaxies with M r < − . . < z < . M r < − . 77 and 0 . < z < . .These volume-limited samples contain 670 and 297 com-pact groups that are independently identified throughthe friends-of-friends algorithm. Unlike systems in themagnitude-limited sample, the volume-limited sampleshave systems with a median stellar mass independent ofredshift. Therefore, the volume-limited samples allow usto study the properties of compact groups without anysample bias that could be introduced in the magnitude-limited sample. Details for the catalogs and the com-pact group selection are described in Sections 2 and 3 ofSohn et al. (2016).In this study, we use the V1 sample of 670 com-pact groups and their 2175 member galaxies with M r < − . 77 and 0 . < z < . ALLWISE source catalog using amatching tolerance of 3 ′′ , corresponding to about half ofthe FWHM of the PSF at 3.4 µ m. Among the 2175 galax-ies, 2067 (95%) are matched with ALLWISE sources. Weuse the profile-fitting magnitudes of the sources at fourMIR bands (3.4 µ m, 4.6 µ m, 12 µ m, and 22 µ m). Whenwe investigate the MIR properties of galaxies, we use Sohn et al. (2016) used M r < − 19 + 5log h and M r < − 20 +5log h with h = 1 (i.e., H = 100 km s − Mpc − ). However, weadopt h = 0 . H = 70 km s − Mpc − ), which results in − 19 + 5log h ≃ − . 77 and − 20 + 5log h ≃ − . http://wise2.ipac.caltech.edu/docs/release/allwise/ only galaxies detected at 12 µ m with S / N > 3. Becausethe ALLWISE magnitudes with S / N < / N > µ m. If we adoptthe S/N cut of 5, the sample size is reduced from 1541to 863 galaxies by 44%. However, we confirm that ourconclusions do not change much even if we use this smallsample.To compare the physical properties of compact groupgalaxies with those in other environments, we usetwo comparison samples of cluster and field galaxies.Hwang et al. (2012a) constructed a sample of 129 re-laxed Abell clusters at 0 . < z < . 14 using thespectroscopic sample of the SDSS DR7 (Abazajian et al.2009). The cluster sample also includes the galaxies at R < R ( R is the virial radius of the cluster) and∆ V = | V gal − V cl | < − ( V gal and V cl are ra-dial velocity of a galaxy and systematic velocity of thecluster). From this sample, we selected 2433 galaxies at R < R as the cluster galaxy sample, and 6312 galax-ies at 5 R < R < R as the field galaxy sample.Park & Hwang (2009) showed that the fraction of early-type galaxies decreases with increasing clustercentric ra-dius at R < R . However, the early-type galaxy frac-tion is nearly constant at R > R (see their Figure4), suggesting that the galaxies at 5 R < R < R can be considered to be field galaxies. The galaxies inthe two comparison samples also satisfy the criterion of M r < − . 77 and 0 . < z < . r -band Petrosian absolute magni-tude ( M r ), u − r color, and morphology. The M r was computed using the Galactic reddening correction(Schlegel et al. 1998), K -corrections (Blanton & Roweis2007), shifted to z = 0 . 1, and evolution correction, E ( z ) = 1 . z − . 1) (Tegmark et al. 2004). The u − r color was computed using extinction- and K -corrected(to z = 0 . u - and r -band model magnitudes. Galaxymorphology data are mainly from the Korea Institute forAdvanced Study (KIAS) DR7 value-added galaxy cata-log (Choi et al. 2010). Galaxies in this catalog are mor-phologically classified into early (E/S0) and late types(S/Irr) based on an automated scheme with u − r color, Lee, G.-H., et al. Compact groupgalaxiesCluster galaxiesField galaxies -2 -1 0 1 2[3.4]-[12] (AB)1.01.52.02.53.0 u - r ( A B ) (a)0.00.51.0 N / N m a x max (b) Fig. 1.— (a) u − r versus [3 . − [12] diagram for the 1541 compactgroup galaxies detected at 12 µ m. Contours represent the numberdensity distribution. In panels (b) and (c), we display the u − r and [3 . − [12] distribution for the compact group galaxies withthe filled histograms. For comparison, we also plot the MIR colordistribution of the 1461 cluster galaxies (open histogram with thicklines) and for the 5084 field galaxies (open histogram with thinlines). the g − i color gradient, and the i -band concentrationindex (Park & Choi 2005). For galaxies not included inthe KIAS DR7 value-added galaxy catalog, Sohn et al.(2016) classified them into early and late types throughvisual inspection. In Table 1, we list the group ID, galaxyID, R.A., declination, redshift, M r , u − r , and morphol-ogy of the 2175 compact group galaxies. RESULTS3.1. Galaxy Color Distributions Figure 1 shows the u − r and [3 . − [12] distributionsof the 1541 compact group galaxies detected at 12 µ m.The mean u − r error is 0.07, while the mean [3 . − [12]error is 0.12. In this diagram, compact group galaxies aredistributed from the top-left corner to the bottom-rightcorner because galaxies with bluer optical colors tend tohave redder MIR colors.In panels (b) and (c), we plot the u − r and [3 . − [12]distributions of compact group galaxies. For compari-son, we also plot the distributions of cluster and fieldgalaxies. As shown in panel (b), u − r histograms forthe three galaxy samples peak at u − r ≃ . 6. However,the fraction of galaxies with u − r > . . ± . . ± . . ± . u − r is 2 . ± . 02 in the compact group galaxy sample,which is also larger than the 2 . ± . 01 of the clustergalaxy sample and the 2 . ± . 01 of the field galaxysample . These suggest that compact group galaxiesare dominated by optical red-sequence galaxies much likecluster galaxies, which is consistent with the findings of The errors in the fractions and in the mean values are thestandard deviations derived from 1000-times resamplings. N / N m a x (a) CG galaxies Total (1541)Non-AGN (1108)AGN fraction -2 -1 0 1 2[3.4]-[12] (AB)0.00.51.0 N / N m a x (b) Non-AGN Cluster galaxies (R Fig. 2.— (a) [3 . − [12] color distribution for the all 1541 (filled)and the 1108 non-AGN (hatched) compact group galaxies. Squaresrepresent the fraction of AGN as a function of [3 . − [12]. Theerror bars indicate the 1 σ deviation from 1000-times bootstrapresamplings. (b) Comparison of the color distributions betweenthe 1108 non-AGN compact group galaxies (hatched), the 1137non-AGN cluster galaxies at R < R (filled), and the 617 non-AGN cluster galaxies at R < . R (open). Walker et al. (2013) and Zucker et al. (2016).Optical red-sequence galaxies, located in a narrow u − r range of 2 . − . 2, have MIR colors ranging from[3 . − [12] ≃ − . . − [12] ≃ 0. Lee et al. (2015) di-vided optical red-sequence galaxies into MIR blue galax-ies (i.e., [3 . − [12] < − . 55) and MIR green galaxies(i.e., − . ≤ [3 . − [12] < − . . The MIR bluegalaxies are composed of stellar populations older than10 Gyr, while the MIR green galaxies have small ( ∼ . − − 10 Gyr) stellar populations (Piovan et al.2003; Ko et al. 2013; Lee et al. 2015; Ko et al. 2016).Optical colors trace star formation on timescales up to1 Gyr (Schawinski et al. 2014). Therefore, the MIRgreen galaxies are not distinguishable from the MIR bluegalaxies in optical colors because both lie in the opticalred-sequence (Walker et al. 2012, 2013; Ko et al. 2013;Lee et al. 2015; Zucker et al. 2016). Thus, we need touse MIR colors instead of optical colors to investigatewhether galaxies in compact groups experienced differ-ent star formation histories from those in clusters, ornot.In panel (c) of Figure 1, we compare the MIR colordistributions between the three different galaxy sam-ples. The cluster galaxy sample shows a blue peak at[3 . − [12] ≃ − . . − [12] ≃ . . − [12] is − . ± . 03 for thecompact group galaxy sample, − . ± . 04 for the clus-ter galaxy sample, and 0 . ± . 02 for the field galaxysample. These results suggest that compared to clustergalaxies, compact group galaxies are more dominated by This criterion is determined from the decomposition of the[3 . − [12] color distribution with three Gaussian functions usingthe Gaussian mixture modeling (Muratov & Gnedin 2010, see Fig-ure 2 in Lee et al. 2015). We stress that this MIR green colorselection criterion is not identical to the “MIR green valley” one inSection 3.2. alaxy Evolution in Compact Groups 5MIR blue galaxies: the fraction of MIR blue galaxies is32 . ± . 2% in the compact group galaxy sample and23 . ± . 1% in the cluster galaxy sample. This suggeststhat the mean stellar ages of compact group galaxies areon average older than those of cluster galaxies.On the other hand, because compact groups favorstrong galaxy interactions, the presence of active galacticnuclei (AGN) should be expected (Coziol et al. 1998a,b;Mart´ınez et al. 2010; Tzanavaris et al. 2014). Becausethe contribution of AGN to the MIR fluxes of the hostgalaxies is not negligible, we identify the AGN in oursample and remove them. First, we identify opticallyselected AGN using the scheme given by Kewley et al.(2006) who classified star-forming galaxies from AGN(Seyferts and LINERs) and composite galaxies in the[NII]/H α versus [OII]/H β diagram (see their Figure 1).We remove 281 AGN and 149 composite galaxies fromthe sample of 1541 compact group galaxies. Second,we identify MIR selected AGN using the classificationmethod introduced by Mateos et al. (2012) who definedthe AGN wedge in the WISE [3 . − [4 . 6] versus [4 . − [12]color-color diagram (see their Figure 2). We find 8MIR selected AGN in the compact group galaxy sample.Among the 8 MIR selected AGN, 5 are also optically se-lected AGN. In total, 433 galaxies (28%) are classifiedas AGN-host galaxies among the 1541 compact groupgalaxies.We compare the MIR color distribution of all 1541compact group galaxies with that of 1108 non-AGN com-pact group galaxies in Figure 2(a). The difference be-tween the two distributions is remarkable at − . . [3 . − [12] . . WISE infraredtransition zone, − . . [3 . − [12] . . 02, defined byAlatalo et al. (2014) . They showed that Seyferts andLINERs are representatives of the zone, which is consis-tent with our result.In Figure 2(b), we compare the MIR color distribu-tion of non-AGN compact group galaxies with that ofnon-AGN cluster galaxies. In this comparison, we use1137 non-AGN cluster galaxies ( R < R ) after remov-ing 324 AGN-host galaxies (22%) from the 1461 clustergalaxy sample. To statistically examine the difference inthe color distributions of the two samples, we compute p -values from the Kolmogorov-Smirnov (KS) test ( P KS )and the Anderson-Darling (AD) k-sample test ( P AD ).The P KS and P AD indicate the probability of rejectingthe null hypothesis that the two samples are drawn fromthe same parent sample. The two tests on the two colordistributions give the P KS and P AD of ≪ . 01, rejectingthe null hypothesis at a significance of > σ level.In Figure 2(b), we use another comparison sample of617 cluster galaxies located at R < . R . In com-parison with the R < R cluster galaxy sample, the R < . R cluster galaxy sample displays the MIRcolor distribution more colsely to that of the compactgroup galaxy sample ( P KS = 0 . 076 and P AD = 0 . < . σ level). Thefractions of MIR red (i.e., [3 . − [12] ≥ − . 3) galaxies inthe compact group galaxy sample is 35 . ± . . ± . 8% in the R < . R cluster This color range is converted from the 0 . < [4 . − [12] < . galaxy sample, but smaller than the 43 . ± . 5% in the R < R cluster galaxy sample. The mean [3 . − [12]for the compact group galaxy sample ( − . ± . 04) issimilar to the − . ± . 06 for the R < . R clus-ter galaxy sample, but smaller than the − . ± . R < R cluster galaxy sample. The early-type galaxy fraction in the R < . R cluster galaxysample (66 . ± . . ± . . ± . 5% in the R < R cluster galaxy sample.These results suggest that compact groups are composedof galaxy populations similar to those in cluster centralregions.However, we find that the fraction of MIR green (i.e., − . ≤ [3 . − [12] < − . 3) galaxies is smaller in thecompact group galaxy sample (28 . ± . R < . R cluster galaxy sample (34 . ± . R < . R cluster galaxy sample in thecomparison with the compact group galaxy sample.Figure 3(a) shows again the MIR color distributions ofthe 1108 non-AGN compact group galaxies and of the617 non-AGN cluster galaxies. In Figure 3 we dividethese two galaxy samples into early-type and late-typegalaxy samples. In Figure 3(b) we compare the MIRcolor distribution of the 687 compact group early-typegalaxies with that of the 407 cluster early-type galaxies.The P KS and P AD values of < . 001 reject the null hy-pothesis for the two color distributions at a > σ signifi-cance. We find that the fraction of MIR green galaxies issmaller in the compact group early-type galaxy sample(37 . ± . . ± . . − [12] valueis smaller in the compact group early-type galaxy sam-ple ( − . ± . 03) than in the cluster early-type galaxysample ( − . ± . < . σ ), compared to thecase for early-type galaxy samples. However, the mean[3 . − [12] value is smaller in the compact group late-typegalaxy sample (0 . ± . 06) than in the cluster late-typegalaxy sample (0 . ± . . ± . . ± . M r . We use M r instead of stel-lar masses because the former is available for all galaxiesin both compact group and cluster galaxy samples. Wefind that the null hypothesis is rejected in the comparisonfor early-type galaxies with − . ≥ M r > − . 77 atthe & . σ level (panel h; P KS = 0 . 03 and P AD = 0 . − . ≥ M r > − . P AD rejects the null hypothesis at the 2 σ level,but P KS does not. The other cases show lower signifi-cance of rejection. This indicates that the difference inthe MIR color distribution between the compact groupgalaxies and the cluster galaxies is mainly attributed to Lee, G.-H., et al. AllAll N / N m a x (a) Total(1108/617) P KS =0.072P AD =0.0350.00.51.01.5 N / N m a x (d) -19.77 ≥ M r > -20.77(427/309) P KS =0.126P AD =0.1790.00.51.01.5 N / N m a x (g) -20.77 ≥ M r > -21.77(407/226) P KS =0.333P AD =0.045-2 -1 0 1 2[3.4]-[12] (AB)0.00.51.01.5 N / N m a x (j) M r ≤ -21.77(274/ 82) P KS =0.119P AD =0.085 Early Types (b) Total(687/407) P KS <0.001P AD <0.001(e) -19.77 ≥ M r > -20.77(196/165) P KS =0.785P AD =0.478(h) -20.77 ≥ M r > -21.77(261/170) P KS =0.030P AD =0.001-2 -1 0 1 2[3.4]-[12] (AB)(k) M r ≤ -21.77(230/ 72) P KS =0.130P AD =0.166 Late Types (c) Total(421/210) P KS =0.027P AD =0.066(f) -19.77 ≥ M r > -20.77(231/144) P KS =0.154P AD =0.329(i) -20.77 ≥ M r > -21.77(146/ 56) P KS =0.589P AD =0.220-2 -1 0 1 2[3.4]-[12] (AB)(l) M r ≤ -21.77( 44/ 10) p KS =0.842p AD =0.444 Fig. 3.— Comparison of [3 . − [12] color distributions for compact group galaxies (filled histograms) with those of cluster galaxies (openhistograms). The middle column is for early-type galaxies, and the right-hand column is for late-type galaxies. The left-hand column isthe sum. The second row is for faint galaxies with − . ≥ M r > − . 77, the third row is for galaxies with − . ≥ M r > − . 77, andthe bottom row is for bright galaxies with M r ≤ − . 77. We list p -values from the KS and AD k-sample tests for the two distributions(for compact group galaxies and for cluster galaxies) in each panel. TABLE 2 WISE -related Physical Parameters of Compact Group Galaxies Host Group Galaxy ID [3 . − [12] a (S/N) b log( L /L ⊙ ) c AGN d M e . log( M star /M sun ) f V1CG001 1237648705657307198 − . ± . 00 -9.9 − . ± . 00 0 − . ± . 00 0 . +0 . − . V1CG001 1237648705657307347 0 . ± . 00 1.4 8 . ± . 00 0 − . ± . 03 9 . +0 . − . V1CG001 1237648705657307315 − . ± . 00 0.2 8 . ± . 00 0 − . ± . 04 10 . +0 . − . V1CG002 1237661126155436166 − . ± . 05 26.5 8 . ± . 02 1 − . ± . 02 10 . +0 . − . V1CG002 1237661126155436169 0 . ± . 03 44.9 8 . ± . 01 1 − . ± . 02 10 . +0 . − . V1CG002 1237661126155436164 − . ± . 05 23.2 8 . ± . 02 0 − . ± . 02 11 . +0 . − . V1CG003 1237661136886431890 0 . ± . 11 9.3 8 . ± . 05 0 − . ± . 03 10 . +0 . − . V1CG003 1237661136886497282 2 . ± . 05 31.7 9 . ± . 01 1 − . ± . 04 10 . +0 . − . V1CG003 1237661136886497353 0 . ± . 10 10.6 8 . ± . 04 0 − . ± . 03 10 . +0 . − . Note . — The full table is available in the online journal. A portion is shown here for guidance regarding its form and content. a The ALLWISE profile-fitting AB magnitudes. − . 99 indicates no ALLWISE photometry. b The signal-to-ratio at 12 µ m. − . ALLWISE photometry. When (S/N) ≤ 3, we assign zero errors to [3 . − [12]and log( L /L ⊙ ). c µ m luminosities. − . 99 indicates no ALLWISE photometry. d AGN classification (0: non-AGN, 1: optical AGN, 2: MIR AGN, and 3: optical+MIR AGN) e µ m absolute magnitudes. − . ± . 000 indicates no ALLWISE photometry. f Stellar masses calculated using H = 70 km s − Mpc − . 0.00 means no measurement. alaxy Evolution in Compact Groups 7 Compact Groups -2-1012 [ . ]-[ ] ( A B ) (a) Total ± ± ± -2-1012 [ . ]-[ ] ( A B ) (d) Early Types ± ± ± -2-1012 [ . ]-[ ] ( A B ) L µ m /L sun (g) Late Types ± ± ± Clusters (R < 0.5R ) (b) Total ± ± ± (e) Early Types ± ± ± L µ m /L sun (h) Late Types ± ± ± Field (c) Total ± ± ± (f) Early Types ± ± ± L µ m /L sun (i) Late Types ± ± ± Fig. 4.— MIR color vs. luminosity diagrams for compact group galaxies (left), cluster ( R < . R ) galaxies (middle), and field galaxies(right). Orange and blue dots represent early- and late-type galaxies, respectively. We divide the galaxies into three classes followingthe classification scheme of Lee et al. (2015): MIR star-forming sequence galaxies (above the long-dashed lines), MIR blue cloud galaxies(below short-dashed lines), and MIR green valley galaxies (between the two lines). Middle and bottom rows are for early- and late-typegalaxies in different environments. We list the fractions of different MIR classes in each panel. Lee, G.-H., et al.the galaxies with − . ≥ M r > − . 77. We exam-ine galaxy properties using the luminosity-limited sub-samples in the following analysis, but find no significantdifference from the results based on the total sample.We therefore present the results in the following analysisbased on the total sample for better statistics.3.2. MIR Color-Luminosity Diagram In Figure 4 we investigate the [3 . − [12] versus 12 µ mluminosity distribution of galaxies in compact groups,clusters, and field. In the [3 . − [12] versus 12 µ m lu-minosity diagram, galaxies are divided into three classesfollowing the classification scheme of Lee et al. (2015):MIR star-forming sequence galaxies (above the inclineddashed lines ), MIR blue cloud galaxies ([3 . − [12] < − . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . 2% and 6 . ± . . ± . 9% and 17 . ± . [3 . − [12] = log( L /L sun ) × . − . 51. See Section 5.1 ofLee et al. (2015) for details. fraction of MIR star-forming sequence galaxies in com-pact groups is consistent with the case that the star for-mation quenching for late-type galaxies occurs more ef-fectively in compact groups than in clusters and in thefield. We note that the results in this section do notchange significantly if we use − . ≤ [3 . − [12] < − . WISE -related parameters and AGN classes ofcompact group galaxies.3.3. Environments of Compact Group Galaxies To investigate the environmental effects on galaxyproperties in compact groups, we use three kinds ofenvironmental parameters. The first parameter isthe number of neighboring galaxies ( N nei ) of compactgroups. Sohn et al. (2016) calculated N nei around com-pact groups within a comoving cylinder defined by pro-jected group centric radius R = 1 Mpc and ∆ V = | V gal − V gr | = 1000 km s − , where V gal and V gr are a radialvelocity of a galaxy and a mean velocity of member galax-ies in a compact group. In the N nei count, Sohn et al.(2016) used galaxies in the V2 sample ( M r < − . . < z < . N nei for 670 compact groups using galaxies in the V1 sample( M r < − . 77 and 0 . < z < . N nei count. We do notassign N nei to 59 compact groups close to the lower andupper redshift limits because their N nei can be underes-timated, resulting in that we assign N nei to 611 compactgroups.The second environmental parameter is the fractionof early-type member galaxies in compact groups ( f E ).Bitsakis et al. (2010, 2011, 2015) used this f E to dividecompact groups into dynamically old and young systemsbased on the assumption that dynamically old systemsare dominated by early-type galaxies that could formthrough repeated interactions between member galaxies.Thus, f E is the parameter reflecting the internal environ-ment of compact groups.The third environmental parameter is the rest-framevelocity dispersions of compact group member galaxies( σ CG ). We adopt the σ CG from Table 5 in Sohn et al.(2016). Sohn et al. (2016) found that compact groupswith low σ CG ( . 100 km s − ) show features of ongoinginteractions among member galaxies, but that compactgroups with σ CG & 300 km s − do not. Thus, σ CG isanother parameter that reflects the internal environmentof compact groups.We note that we use all the member galaxies regardlessof 12 µ m detection when we calculate f E and σ CG . InTable 3, we list the ID, R.A., declination, redshift, as wellas three environmental parameters of the 670 compactgroups. Figure 5 shows the relations between the threeenvironmental parameters. We plot only 611 systemswith complete N nei . The N nei histogram peaks at N nei ≃ − 6, and stretches out to more than N nei = 40. The f E mainly has discontinuous values (e.g., 0, 1/3, 2/3, and1) because the majority (495/611, 81%) of our systemshave three member galaxies. The σ CG histogram peaksat ∼ 80 km s − , and stretches out to more than 600 kms − . The median σ CG is ∼ 200 km s − .3.3.1. Environmental Dependence on N nei alaxy Evolution in Compact Groups 9 TABLE 3Environmental Parameters of Compact Groups Group ID a R.A. a Decl. a Redshift a N b mem N c nei f d E σ CG (km s − ) e V1CG001 198.227173 1.012775 0 . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± . ± Note . — The full table is available in the online journal. A portion is shown here for guidanceregarding its form and content. a Group ID from Table 5 of Sohn et al. (2016). b The number of member galaxies in each compact group. c The number of neighboring galaxies ( M r < − . 77) in the comoving cylinder. For 59 compactgroups near the lower and upper redshift limits, N nei are not assigned ( N nei = − d The fraction of early-type galaxies among member galaxies of a compact group. e The rest-frame line-of-sight velocity dispersion of galaxies in each compact group and the 1 σ errorderived from 1000-times bootstrap resamplings, from Table 5 of Sohn et al. (2016). To examine the effects of N nei on galaxy propertiesin compact groups, we investigate how the MIR colordistribution of compact group galaxies varies with N nei .Figure 6 shows the MIR color histograms of the galaxiesin different N nei bins. The left-hand column shows thatthe fraction of MIR red (i.e., [3 . − [12] > − . 3) galaxiesdramatically decreases as N nei increases. The MIR redgalaxies are mainly late-type galaxies. The fraction oflate-type galaxies gradually decreases from 56 . ± . . ± . 6% (panel d) as N nei increases.For comparison, we also plot the MIR color distributionof cluster galaxies as open histograms. We find that thefraction of late-type galaxies in the N nei ≥ . ± . N n e i (c)050100 N (d) (e)0 20 40 60N nei f E (a) 0 400 800 σ CG (km s -1 ) (b) 0 200N(f) Fig. 5.— Relation between the three environmental parametersfor compact groups: (a) f E vs. N nei , (b) f E vs. σ CG , and (c) N nei vs. σ CG diagrams. (d − f) Histograms for the three parameters. columns in Figure 6. The KS and AD k-sample tests forearly-type galaxy samples in the four different N nei bins(e − h) reject the null hypothesis at the < . σ level. Theyalso reject the null hypothesis for late-type galaxy sam-ples in the four different N nei bins (i − l) at the < . σ level. These imply that N nei does not affect the MIRcolors of early- and late-type galaxies in compact groupsdirectly. Thus, the N nei -dependence of MIR colors shownin the left column (a − d) results from the fact that com-pact groups located in denser regions (larger N nei ) tendto have more early-type member galaxies with MIR bluecolors.The middle column in Figure 6 shows that compactgroup early-type galaxies have MIR colors bluer thancluster early-type galaxies regardless of N nei . The colordifference is significant ( & . σ ) in the N nei < 20 com-pact groups, but it is marginal ( ∼ . σ ) in the N nei ≥ N nei < N nei ≥ ∼ − σ . 3.3.2. Environmental Dependence on f E Figure 7 shows the MIR color distributions of compactgroup early-type galaxies in different f E bins: f E < . . ≤ f E < 1, and f E = 1. We find no significant dif-ferences in the MIR colors between the three f E bins.The mean [3 . − [12] value is − . ± . 08 at f E < . − . ± . 04 at 0 . ≤ f E < . 1, and − . ± . 03 at f E = 1, respectively. These mean values agree wellwithin 1 σ errors. The KS and the AD k-sample testsdo not reject the null hypothesis, either.Bitsakis et al. (2011) showed that early-type galax-ies in dynamically young ( f E < . 25) compact groupshave bluer NUV − r colors than those in dynamically old( f E > . 25) groups. However, the number of early-typegalaxies in their dynamically young groups is only three.In our sample, there are no early-type galaxies belongingto the f E < . 25 groups, and only six early-type galax-ies belonging to the f E < . Total N / N m a x Total N / N m a x (a) N nei < 3 (N gal =265) P KS <0.001P AD <0.001f L =56.2% ± N / N m a x (b) 3 ≤ N nei < 8 (N gal =306) f L =43.5% ± KS =0.100P AD =0.146 N / N m a x (c) 8 ≤ N nei < 20 (N gal =261) f L =27.2% ± KS <0.001P AD <0.001 N / N m a x -2 -1 0 1 2[3.4]-[12] (AB)(d) N nei ≥ 20 (N gal =195) f L =15.9% ± KS <0.001P AD <0.001 Early Types (e) N nei < 3 (N gal =116) ClusterCGP KS =0.026P AD =0.004 (f) 3 ≤ N nei < 8 (N gal =173) P KS =0.038P AD =0.016 (g) 8 ≤ N nei < 20 (N gal =190) P KS <0.001P AD <0.001 -2 -1 0 1 2[3.4]-[12] (AB)(h) N nei ≥ 20 (N gal =164) P KS =0.144P AD =0.044 Late Types (i) N nei < 3 (N gal =149) P KS =0.375P AD =0.198 (j) 3 ≤ N nei < 8 (N gal =133) P KS =0.215P AD =0.265 (k) 8 ≤ N nei < 20 (N gal = 71) P KS =0.002P AD =0.005 -2 -1 0 1 2[3.4]-[12] (AB)(l) N nei ≥ 20 (N gal =31) P KS =0.054P AD =0.026 Fig. 6.— Dependence of the [3 . − [12] color of compact group galaxies on the number of neighboring galaxies ( N nei ). In the left-handcolumn, we list the fraction of late-type galaxies ( f L ). In the middle and right-hand columns, we plot early-type galaxies and late-typegalaxies, separately. For comparison, we also plot the color distributions for cluster galaxies (open histograms). N / N m a x N / N m a x (a) f E < 0.5 (N gal = 66) Early Types P KS =0.014P AD =0.005 N / N m a x (b) 0.5 ≤ f E < 1 (N gal =233) ClusterCGP KS =0.016P AD =0.019 N / N m a x -2 -1 0 1 2[3.4]-[12] (AB)(c) f E = 1 (N gal =388) P KS =0.003P AD <0.001 Fig. 7.— MIR color distributions for early-type galaxies in the(a) f E < . 5, (b) 0 . ≤ f E < 1, and (c) f E = 1 compact groups. Forcomparison, we also plot the (filled) histograms for the early-typegalaxies in our cluster sample. that the MIR color distribution for the six galaxies is notdifferent significantly from that for early-type galaxies inthe f E < . f E bins,compact group early-type galaxies have MIR colors bluerthan cluster early-type galaxies. The P KS and P AD are < . > . σ ). This is consistent with the result inFigure 6. N / N m a x N / N m a x (a) f E = 0 (N gal =110) Late Types ClusterCGP KS <0.001P AD <0.001 N / N m a x (b) 0 < f E < 0.5 (N gal =175) P KS =0.001P AD =0.001 N / N m a x -2 -1 0 1 2[3.4]-[12] (AB)(c) 0.5 ≤ f E < 1 (N gal =136) P KS =0.001P AD =0.003 Fig. 8.— MIR color distributions for late-type galaxies in the(a) f E = 0, (b) 0 < f E < . 5, and (c) 0 . ≤ f E < Figure 8 shows the MIR color distributions of compactgroup late-type galaxies in different f E bins: f E = 0, 0 5, and 0 . ≤ f E < 1. For comparison, we also plotthe color distributions for the cluster late-type galaxysample. Late-type galaxies in the f E = 0 compact groupshave MIR colors redder than cluster late-type galaxies.However, late-type galaxies in the f E > P KS and P AD are ≤ . 03, indicating that thecolor differences are significant ( > σ ). On the otheralaxy Evolution in Compact Groups 11 Total N / N m a x Total N / N m a x (a) σ CG < 100 km s -1 KS <0.001P AD <0.0010.00.51.01.5 N / N m a x (b) 100 km s -1 ≤ σ CG < 300 km s -1 KS =0.308P AD =0.1430.00.51.01.5 N / N m a x (c) σ CG ≥ 300 km s -1 -2 -1 0 1 2[3.4]-[12] (AB)0.00.51.01.5 CGs (383)Clusters (617) P KS <0.001P AD <0.001 Early Types (d) σ CG < 100 km s -1 CGs (100)Clusters (407) P KS =0.258P AD =0.156(e) 100 km s -1 ≤ σ CG < 300 km s -1 CGs (299)Clusters (407) P KS =0.003P AD =0.002(f) σ CG ≥ 300 km s -1 -2 -1 0 1 2[3.4]-[12] (AB)CGs (288)Clusters (407) P KS =0.002P AD <0.001 Late Types (g) σ CG < 100 km s -1 CGs (135)Clusters (210) P KS =0.134P AD =0.098(h) 100 km s -1 ≤ σ CG < 300 km s -1 CGs (191)Clusters (210) P KS =0.075P AD =0.216(i) σ CG ≥ 300 km s -1 -2 -1 0 1 2[3.4]-[12] (AB)CGs ( 95)Clusters (210) P KS <0.001P AD <0.001 Fig. 9.— Left: MIR color distributions for compact group galaxies (filled histograms) in the (a) σ CG < 100 km s − , (b) 100 kms − ≤ σ CG < 300 km s − , and (c) σ CG ≥ 300 km s − compact groups. Open histograms represent cluster galaxies. We plot early-typegalaxies in the middle column (d − f) and late-type galaxies in the right column (g − i). hand, we find that MIR colors of late-type galaxies inthe 0 < f E < . . ≤ f E < P KS = 0 . 835 and P AD = 0 . . − [12] values of late-type galaxies in the0 < f E < . . ± . 09) and in the 0 . ≤ f E < . ± . 10) are similar, but smaller than the1 . ± . 09 in the f E = 0 groups.These results suggest that star formation activity oflate-type galaxies is suppressed more efficiently in the f E > f E = 0 groups have comparable orhigher star formation activity than that of cluster late-type galaxies. It is interesting that late-type membergalaxies in compact groups show a hint of star formationquenching only when the compact groups have early-typemember galaxies (see Section 4.3 for detailed discussion).3.3.3. Environmental Dependence on σ CG Figure 9 shows how the MIR colors of compact groupgalaxies depend on σ CG . We divide compact groupsinto low ( < 100 km s − ), intermediate (100 − 300 kms − ), and high ( ≥ 300 km s − ) σ CG systems. In theleft column, we find that the fraction of MIR red (i.e.,[3 . − [12] > − . 3) galaxies decreases as σ CG increases.This is because that the fraction of late-type galaxies issmaller in higher σ CG systems. This trend is similar tothe N nei -dependence in Figure 6.The MIR colors of compact group early-type galax-ies do not depend on σ CG . The MIR color distributionof early-type galaxies in the σ CG < 100 km s − groupsdoes not differ from that of cluster early-type galaxies.However, the MIR colors of early-type galaxies in the σ CG ≥ 100 km s − groups are significantly ( > σ ) differ- ent (bluer) from those of cluster early-type galaxies.The MIR colors of late-type galaxies are bluer in the σ CG ≥ 300 km s − groups than in the σ CG < 300 km s − groups. The P KS and P AD confirm the color differencewith a significance & σ level. Furthermore, the MIRcolors of late-type galaxies in the σ CG ≥ 300 km s − groups are significantly ( > σ ) bluer than those of clusterlate-type galaxies.3.3.4. Which Environment Most Affects the MIRProperties of Compact Group Galaxies? In Figures 6 − 9, we find that the MIR colors of com-pact group early-type galaxies do not depend much onthe three environmental parameters, and that compactgroup late-type galaxies depend on f E and σ CG , but noton N nei . In f E > σ CG ≥ 300 km s − compact groups,late-type member galaxies show bluer MIR color distri-butions. Bitsakis et al. (2011) showed that σ CG tend tobe larger in dynamically old ( f E > . 25) compact groupsthan in dynamically young groups. However, f E and σ CG for our compact groups do not have a significant correla-tion (Figure 5). Spearman’s rank correlation coefficientfor f E and σ CG is 0.28 and the probability of obtainingthe correlation by chance is ≪ . f E , σ CG , and N nei most affects the MIR colors of galaxies,we first consider f E and σ CG simultaneously in Figure10. In the left panels we compare MIR colors of late-type galaxies in the f E = 0 compact groups with thoseof late-type galaxies in the f E > σ CG bins. We find that MIR colors of late-type galaxies in the f E > N / N m a x N / N m a x (a) σ CG < 100 km s -1 P KS <0.001P AD <0.001f E = 0 (58)f E > 0 (77) N / N m a x (b) 100 ≤ σ CG < 300 km s -1 P KS =0.024P AD =0.006f E = 0 (40)f E > 0 (151) N / N m a x -2 -1 0 1 2[3.4]-[12] (AB) (c) σ CG ≥ 300 km s -1 P KS =0.012P AD =0.007f E = 0 (12)f E > 0 (83) (d) f E = 0 P KS =0.153P AD =0.151 σ CG < 100 km s -1 (58) σ CG ≥ 100 km s -1 (52) (e) 0 < f E < 0.5 P KS =0.106P AD =0.117 σ CG < 100 km s -1 (48) σ CG ≥ 100 km s -1 (127) -2 -1 0 1 2[3.4]-[12] (AB) (f) 0.5 ≤ f E < 1 P KS =0.590P AD =0.621 σ CG < 100 km s -1 (29) σ CG ≥ 100 km s -1 (107) Fig. 10.— Left: MIR color distributions for late-type galaxies in the (a) σ CG < 100 km s − , (b) 100 ≤ σ CG < 300 km s − , and (c) σ CG ≥ 300 km s − compact groups. Filled and hatched histograms represent late-type galaxies in the f E = 0 groups and those in the f E > f E = 0, (e) 0 < f E < . 5, and (f) 0 . ≤ f E < σ CG < 100 km s − groups and the σ CG ≥ − groups, respectively. We list P KS and P AD for the two histograms in each panel. N / N m a x N / N m a x (a) N nei < 3 f E = 0 (N gal = 54)f E > 0 (N gal = 95) P KS =0.002P AD <0.001 N / N m a x (b) 3 ≤ N nei < 8 f E = 0 (N gal = 34)f E > 0 (N gal = 99) P KS =0.043P AD =0.001 N / N m a x -2 -1 0 1 2[3.4]-[12] (AB)(c) N nei ≥ f E = 0 (N gal = 8)f E > 0 (N gal = 94) P KS =0.018P AD =0.016 (d) f E = 0 N nei < 6 (N gal = 86)N nei ≥ gal = 10) P KS =0.694P AD =0.578 (e) 0 < f E < 0.5 N nei < 6 (N gal =114)N nei ≥ gal = 61) P KS =0.521P AD =0.781 -2 -1 0 1 2[3.4]-[12] (AB)(f) 0.5 ≤ f E < 1 N nei < 6 (N gal = 71)N nei ≥ gal = 65) P KS =0.272P AD =0.359 Fig. 11.— Left: MIR color distributions for late-type galaxies in the (a) N nei < 3, (b) 3 ≤ N nei < 8, and (c) N nei ≥ f E = 0 groups and those in the f E > f E = 0, (e) 0 < f E < . 5, and 0 . ≤ f E < N nei < N nei ≥ alaxy Evolution in Compact Groups 13 N / N m a x N / N m a x CGsR/R < 0.40.4 ≤ R/R < 0.6 (a)0.00.51.0 N / N m a x -2 -1 0 1 2[3.4]-[12] (AB) CGs (N nei ≤ (b) Fig. 12.— (a) Comparison of MIR color distributions for com-pact group galaxies (filled histograms) and cluster galaxies (openhistograms). Red and blue histograms are for cluster galaxies with R/R < . . ≤ R/R < . 4, respectively. (b) We onlyplot galaxies belonging to N nei ≤ in the f E = 0 groups in all three σ CG bins. The KS andAD k-sample tests reject the null hypothesis at a & . σ level in the three bins. On the other hand, in the rightcolumn of Figure 10 we compare MIR colors of late-typegalaxies in the σ CG < 100 km s − groups with those oflate-type galaxies in the σ CG ≥ 100 km s − groups in thethree f E bins. We find that there is no significant colordifference between the two σ CG samples in all three f E bins. These results suggest that the MIR colors of late-type galaxies in compact groups are more sensitive to f E than σ CG .We conduct a similar analysis to examine whetherthe f E effects on galaxy colors exist when N nei is fixed.Spearman’s rank correlation coefficient for f E and N nei is 0.39, and the probability of obtaining the correlationby chance is ≪ . f E = 0 groups with those in the f E > N nei bins. In all three N nei bins, the MIRcolors are bluer in the f E > f E = 0groups. The AD k-sample test rejects the null hypothe-sis at the & . σ level, and the KS test also rejects thenull hypothesis at the > σ level. On the other hand,the right column shows that there is no significant dif-ference in MIR colors between the late-type galaxy sam-ples in the N nei < N nei ≥ f E among the three environmental pa-rameters, suggesting that the f E is the most importantparameter in determining the MIR colors of late-typegalaxies in compact groups. DISCUSSION4.1. Fast Galaxy Evolution in Compact Groups In Section 3, we find that the MIR colors of early-typegalaxies are bluer in compact groups than in clusters andthe field. This trend also persists when we use severalsubsamples with N nei , f E , and σ CG . We also find thatthe late-type galaxies in the f E > σ CG ≥ 300 kms − ) compact groups have MIR colors bluer than those ofcluster late-type galaxies. These results imply that stel-lar populations in early-type galaxies are on average olderin compact groups than in clusters, and that star for- mation activity of late-type galaxies is suppressed moreefficiently in f E > R < . R ) regionsof clusters.So far, several studies have concluded that galaxyevolution is faster in compact groups than in thefield through various analyses (e.g., Proctor et al.2004; de la Rosa et al. 2007; Bitsakis et al. 2010,2011, 2016; Tzanavaris et al. 2010; Walker et al. 2010,2012; Coenda et al. 2012, 2015; Lenki´c et al. 2016).Proctor et al. (2004) and Mendes de Oliveira et al.(2005) found that the properties of compact groups(i.e., stellar ages and the early-type galaxy fraction) aremore similar to those of cluster galaxies than those offield galaxies. However, previous studies have not foundevidence that galaxy evolution is faster in compactgroups than in galaxy clusters.Johnson et al. (2007) and Walker et al. (2010, 2012)showed that compact group galaxies show a strong bi-modal Spitzer IRAC 3 . − µ m color distribution with anevident gap at green colors. This gap is not found in com-parison samples of isolated galaxies, galaxy pairs, and thecenter of the Coma cluster. However, they found thatthe Coma infall region (i.e., 0 . − . R , Jenkins et al.2007) shows the color distribution statistically similar tothat of compact group galaxies. They interpreted theseresults as that the infall region and compact groups havea similarity in environment. To examine their resultswith our data, we select 495 galaxies at R < . R (rep-resenting the central regions of clusters) and 231 galax-ies at 0 . R ≤ R < . R (representing the infallregions) from the cluster galaxy sample. We comparethe MIR color distributions for the two cluster galaxysubsamples with that for the compact group galaxy sam-ple in Figure 12(a). The color distribution for the com-pact group galaxy sample is more similar to that for the R < . R cluster galaxy sample ( P KS = 0 . 211 and P AD = 0 . . R ≤ R < . R clus-ter galaxy sample ( P KS = 0 . 001 and P AD = 0 . N nei ≤ N nei ≤ . R ≤ R < . R cluster galaxysample ( P KS = 0 . 444 and P AD = 0 . N / N m a x star /M sun ) CGClusterField N / N m a x µ m absolute magnitude Fig. 13.— Comparison of 3.4 µ m absolute magnitude distri-butions between compact group (hatched histograms) and clustergalaxies (filled histograms). The top panel is for early-type galax-ies, while the bottom panel is for late-type galaxies. in clusters. The median velocity dispersion for our com-pact groups is ∼ 200 km s − , which is much smallerthan the median value for galaxy clusters, ∼ 800 km s − (Rines et al. 2013). Furthermore, the mean size of com-pact groups is 30.7 ± . ∼ − . µ m absolute magni-tude ( M . ) distribution of early- and late-type galaxiesin compact groups, clusters, and the field. Hwang et al.(2012b) showed that M . can be used as a proxy for stel-lar masses ( M star ) of galaxies. We confirm the correlationbetween the M . and M star using the compact groupgalaxies . The stellar masses in this study are basedon H = 70 km s − Mpc − , and are calculated usingthe LePHARE spectral energy distribution fitting code(Arnouts et al. 1999; Ilbert et al. 2006) with the SDSS ugriz photometric data (see Section 2 of Sohn et al. 2016for details). Table 2 lists M star and M . of compactgroup galaxies.We find that among compact group early-type galaxies,the fraction of massive populations with M . < − . M star & M sun ) is 34 . ± . . ± . 1% for the cluster early-type galaxy sam-ple and the 21 . ± . 3% for field early-type galaxy sam-ple. Similarly, among compact group late-type galaxies,the fraction of massive populations with M . < − . M star & . M sun ) is 44 . ± . . ± . 2% for the cluster late-type galaxysample and the 27 . ± . 8% for the field late-type galaxysample. This result suggests that compact groups are anideal environment for efficient mass build-up of galaxies.4.2. Relation between Compact Group Members andNeighboring Galaxies In Section 3.3, as shown in Figure 6, we find that com-pact groups have larger f E and bluer MIR colors as their log( M star /M sun ) = ( − . ± . × M . − (0 . ± . 20) withrms= 0 . ∼ arnouts/LEPHARE/lephare.html surrounding environments are denser (larger N nei ). Tofurther investigate the effects of the surrounding envi-ronments on the physical properties of compact groupmember galaxies, we compare the properties of neighbor-ing galaxies around each compact group with those of thecompact group member galaxies. We calculate f E , mean u − r ( u − r ), and mean [3 . − [12] ([3 . − [12]) for neigh-boring galaxies in the comoving cylinder of each compactgroup. Figure 14 shows f E , u − r , and [3 . − [12] forneighboring galaxies as a function of N nei . Triangles in-dicate the mean values of the three parameters in N nei bins. As N nei increases, neighboring galaxies have larger f E , redder optical colors, and bluer MIR colors. Forcomparison, we compute f E , u − r , and [3 . − [12] formember galaxies of each compact group. Squares repre-sent the mean values for the three parameters of compactgroup member galaxies.We find that the two mean curves show a similarpattern in all three panels. The N nei -dependence of f E , u − r , and [3 . − [12] for neighboring galaxies isconsistent with the morphology-density or the SFR-density relation (e.g., Dressler 1980; Lewis et al. 2002;Park & Hwang 2009). However, it is interesting that thesimilar trends are also seen for compact group membergalaxies. Since compact groups are very small in size(the mean size is 30 . ± . ± 18 Mpc − , which is much higherthan the mean surrounding density, 2 . ± . 10 Mpc − ,from N nei . Moreover, the internal galaxy number densi-ties of compact groups do not depend on the surroundinggalaxy number densities. Thus, the N nei -dependence ofthe properties of compact group member galaxies can notbe explained by the morphology-density or SFR-densityrelations alone.The similar N nei -dependence for compact group mem-ber galaxies and their neighboring galaxies suggests thatthe properties of compact group galaxies are relatedto those of their neighboring galaxies. If the physicalproperties of compact group galaxies are independent oftheir neighboring galaxies, the two mean curves wouldshow different behaviors. A plausible scenario wouldbe that the neighboring galaxies are sources of membergalaxies in compact groups. This supports the idea ofDiaferio et al. (1994) that compact groups replenish theirmembers from surrounding environments, so that theydo not disappear by mergers within a few Gyrs. Previ-ous studies have supported this replenishment model byshowing that many ( > nei f E (a)5 20 3010N nei f E || 10N nei u - r ( A B ) (b)5 20 30 CG MembersNeighboring nei -2-1012 [ . ]-[ ] ( A B ) (c)5 20 3010N nei -2-1012 Fig. 14.— N nei dependence of (a) f E , (b) u − r , and (c) [3 . − [12] for neighboring galaxies within the comoving cylinder around thecompact groups, and their mean values (triangles) in N nei bins. For comparison, we also plot the mean values of the three parameters forcompact group member galaxies in N nei bins (squares). Error bars represent 3 σ deviation in a given N nei bin. ably results from frequent interactions among membergalaxies. This result also suggests that compact groupsare the most suitable environment for the pre-processing(Zabludoff & Mulchaey 1998).4.3. Hydrodynamic Interactions in Compact Groups We find that the MIR colors of late-type galaxies arebluer in the f E > f E = 0compact groups (Figure 8). This f E dependence is stillfound when N nei or σ CG is fixed (see Figures 10 and11), indicating that among the three environmental pa-rameters, f E could be the most important parameter indetermining the MIR colors of late-type member galax-ies. The [3 . − [12] colors of late-type galaxies arewell correlated with specific SFRs (Donoso et al. 2012;Hwang et al. 2012b). Thus, this finding suggests thatstar formation activity of late-type galaxies is suppressedmore efficiently in the f E > f E = 0 compact groups. Moreover, the suppression ofstar formation activity is stronger in the f E > R < . R ) regions(Figure 8).Bitsakis et al. (2010, 2011) showed that late-typegalaxies in the f E > . 25 compact groups tend to havesmaller specific SFRs than those in the f E < . 25 com-pact groups. Their result seems similar to our result. Toverify their result with our sample, we divide our com-pact group sample into the f E = 0, 0 < f E ≤ . 25, and f E > . 25 compact groups. The large values ( > . P KS and P AD cannot reject the null hypothesis fortwo MIR color distributions for late-type galaxies in the0 < f E ≤ . 25 compact groups and for late-type galax-ies in the f E > . 25 compact groups. However, the P KS = 0 . 056 and P AD = 0 . 008 reject the null hypothesisfor two MIR color distributions for late-type galaxies inthe f E = 0 compact groups and for late-type galaxiesin the 0 < f E ≤ . 25 compact groups at a significance & σ . These suggest that the f E = 0 and 0 < f E ≤ . f E > f E = 0 compact groups. The differ-ent f E criterion between this study and Bitsakis et al.probably results from different compact group samples. An important point here is that the star formationactivity of late-type galaxies in compact groups is sup-pressed when the groups contain early-type members.This suppression of star formation activity cannot beexplained by gravitational interactions among membergalaxies alone. Park et al. (2008) and Park & Choi(2009) found that galaxy properties strongly dependon the distance and morphology of the nearest neigh-bor galaxy, using spectroscopic samples drawn from theSDSS data. They showed that when a galaxy is locatedwithin the virial radius of its nearest neighbor galaxy, thegalaxy tends to have a morphological type similar to thatof the neighbor galaxy. However, this phenomenon doesnot manifest if the distance to the neighbor is greaterthan the virial radius of the neighbor. They suggestedthat the phenomenon is due to hydrodynamic interac-tions, and that the effects of hydrodynamic interactionsare significant to change the morphology and star for-mation activity of galaxies when the galaxies are locatedwithin the virial radii of their neighbor galaxies. Thesizes of compact groups (the mean value is 30.7 kpc) aresignificantly smaller than virial radii of member galax-ies (i.e., 430 kpc and 340 kpc for early- and late-typegalaxies with M r = − . 77, Park & Hwang 2009). Thisimplies that compact group member galaxies are alreadylocated within the virial radius of each other, and thatthey interact hydrodynamically each other.Park & Hwang (2009) focused on the case of galaxieslocated within the virial radii of massive galaxy clusters,and found that even in cluster environments the hydrody-namic interactions with early-type neighbor galaxies arethe main drivers of star formation quenching of late-typegalaxies. They also found that the effects of the clusterhot gas on the star formation quenching of galaxies at R > . R is insignificant compared to the effects ofhydrodynamic interactions with neighbor galaxies. Un-like cluster environments, hot gas of compact groups isnot in a hydrostatic equilibrium state, and it is likely tobe associated with the individual galaxies or brightestgalaxies (Desjardins et al. 2013, 2014), suggesting thathydrodynamic effects from hot gas are insignificant incompact groups. Therefore, we conclude that the sup-pressed star formation activity of late-type galaxies inthe f E > SUMMARY AND CONCLUSIONSWe study the MIR properties of galaxies in com-pact groups and their environmental dependence usinga volume-limited sample of 670 compact groups andtheir 2175 member galaxies with M r < − . 77 and0 . < z < . N nei , f E , and σ CG to represent the internal and external envi-ronment of the compact groups. Our key findings of thisstudy are summarized as follows.1. The MIR colors of compact group early-type galax-ies are on average bluer than those of cluster ( R < . R ) early-type galaxies regardless of N nei , f E ,and σ . This suggests that early-type galaxies incompact groups are on average older than those ofcluster galaxies.2. The MIR colors of the late-type galaxies in the f E > f E > N nei increases, compact groups have larger f E ,redder optical colors, and bluer MIR colors. Thesetrends are also seen for neighboring galaxies aroundcompact groups. Considering the extremely highgalaxy number densities in compact group envi-ronments, this similar N nei -dependence for com-pact group member galaxies and their neighboringgalaxies is not well explained by the morphology- and SFR-density relations. This result can be ex-plained by the scenario that neighboring galaxiesare sources of member galaxies in compact groups,supporting the replenishment model suggested byDiaferio et al. (1994).4. At a given N nei , compact group members alwayshave on average larger f E and bluer MIR colorsthan neighboring galaxies. This suggests that com-pact groups are not simply aggregates of capturedneighboring galaxies, and that compact group en-vironments play a critical role in accelerating mor-phology transformation and star formation quench-ing for the member galaxies.5. In the f E = 0 compact groups, the MIR colorsof late-type galaxies are on average redder thanthose of cluster late-type galaxies. However, in the f E > REFERENCESAbazajian, K. N., Adelman-McCarthy, J. K., Ag¨ueros, M. A., etal. 2009, ApJS, 182, 543Adelman-McCarthy, J. K., Ag¨ueros, M. A., Allam, S. S., et al.2008, ApJS, 175, 297-313Alam, S., Albareti, F. D., Allende Prieto, C., et al. 2015, ApJS,219, 12Alatalo, K., Cales, S. L., Appleton, P. N., et al. 2014, ApJ, 794,L13Amram, P., Mendes de Oliveira, C., Plana, H., Balkowski, C., &Hernandez, O. 2007, A&A, 471, 753Appleton, P. N., Xu, K. C., Reach, W., et al. 2006, ApJ, 639, L51Arnouts, S., Cristiani, S., Moscardini, L., et al. 1999, MNRAS,310, 540Barnes, J. 1985, MNRAS, 215, 517Barnes, J. E. 1989, Nature, 338, 123Barton, E., Geller, M., Ramella, M., Marzke, R. O., & da Costa,L. N. 1996, AJ, 112, 871Bitsakis, T., Charmandaris, V., Le Floc’h, E., et al. 2010, A&A,517, A75Bitsakis, T., Charmandaris, V., da Cunha, E., et al. 2011, A&A,533, A142Bitsakis, T., Dultzin, D., Ciesla, L., et al. 2015, MNRAS, 450,3114 Bitsakis, T., Dultzin, D., Ciesla, L., et al. 2016, MNRAS, 459, 957Blanton, M. R., & Roweis, S. 2007, AJ, 133, 734Choi, Y.-Y., Han, D.-H., & Kim, S. S. 2010, Journal of KoreanAstronomical Society, 43, 191Cluver, M. E., Appleton, P. N., Boulanger, F., et al. 2010, ApJ,710, 248Cluver, M. E., Appleton, P. N., Ogle, P., et al. 2013, ApJ, 765, 93Coenda, V., Muriel, H., & Mart´ınez, H. J. 2012, A&A, 543, A119Coenda, V., Muriel, H., & Mart´ınez, H. J. 2015, A&A, 573, A96Coziol, R., & Plauchu-Frayn, I. 2007, AJ, 133, 2630Coziol, R., Ribeiro, A. L. B., de Carvalho, R. R., & Capelato,H. V. 1998, ApJ, 493, 563Coziol, R., de Carvalho, R. R., Capelato, H. V., & Ribeiro,A. L. B. 1998, ApJ, 506, 545de Carvalho, R. R., Ribeiro, A. L. B., & Zepf, S. E. 1994, ApJS,93, 47de la Rosa, I. G., de Carvalho, R. R., Vazdekis, A., & Barbuy, B.2007, AJ, 133, 330Desjardins, T. D., Gallagher, S. C., Tzanavaris, P., et al. 2013,ApJ, 763, 121Desjardins, T. D., Gallagher, S. C., Hornschemeier, A. E., et al.2014, ApJ, 790, 132Diaferio, A., Geller, M. J., & Ramella, M. 1994, AJ, 107, 868 alaxy Evolution in Compact Groups 17alaxy Evolution in Compact Groups 17