KYDISC: Galaxy Morphology, Quenching, and Mergers in the Cluster Environment

2.1.1. The Magellan Baade IMACS Observation

The optical images of three Abell clusters (Abell 1146, Abell 3574, and Abell 3659) were obtained in the $g^{\prime}$ and $r^{\prime}$ bands using the IMACS on the 6.5 m Magellan Baade telescope at Las Campanas Observatory (LCO) on a dark night in 2012. We used an f/2 camera on the IMACS with charge-coupled devices (CCDs) consisting of eight mosaic chips covering a circular area 27' in diameter. Descriptions of target clusters and details of the observation are shown in Table 1. Four nearby fields of Abell 3574, the closest cluster in our data, were targeted and covered one-fourth of the area, including 0.85 virial radii of the cluster.

Table 1.  Summary of Deep Photometry

Cluster ID		R.A.(J2000)	Decl.(J2000)	Instrument	Filter	texpa	N of Dithering	Date
		hh:mm:ss	dd:mm:ss			s
Abell 1146		11:01:29.9	−22:46:19	IMACS	$g^{\prime}$	2500 (100)	5	2012 May 16
(z ∼ 0.142)					$r^{\prime}$	2500 (100)	5
Abell 3574	−1	13:48:54.2	−30:23:01		$g^{\prime}$	1250 (30)	5
(z ∼ 0.015)					$r^{\prime}$	1250 (30)	5
	−2	13:47:31.7	−30:26:09		$g^{\prime}$	1250 (30)	5
					$r^{\prime}$	1250 (30)	5
	−3	13:48:44.4	−30:43:59		$g^{\prime}$	1250 (30)	5
					$r^{\prime}$	1250 (30)	5
	−4	13:47:19.3	−30:48:42		$g^{\prime}$	1250 (30)	5
					$r^{\prime}$	1250 (30)	5
Abell 3659		20:02:27.4	−30:04:50		$g^{\prime}$	2500 (100)	5
(z ∼ 0.090)					$r^{\prime}$	2500 (100)	5
Abell 116		00:55:49.05	00:45:01.1	MegaCam	$u^{\prime}$	3320	5	2012 Jul 26
(z ∼ 0.067)					$g^{\prime}$	2940	7	2012 Aug 21
					$r^{\prime}$	2940	7	2012 Aug 16
Abell 646		08:22:09.53	47:05:40.1		$u^{\prime}$	3320	5	2013 Jan 14
(z ∼ 0.127)					$g^{\prime}$	2940	7	2013 Jan 14
					$r^{\prime}$	2940	7	2013 Jan 14
Abell 655		08:24:50.00	46:51:40.1		$u^{\prime}$	3320	5	2013 Jan 14
(z ∼ 0.127)					$g^{\prime}$	2940	7	2013 Jan 14
					$r^{\prime}$	2940	7	2013 Jan 14
Abell 667		08:28:10.00	44:48:38.2		$u^{\prime}$	3320	5	2012 Oct 20
(z ∼ 0.144)					$g^{\prime}$	2940	7	2012 Oct 21
					$r^{\prime}$	2940	7	2013 Feb 02
Abell 690		08:39:16.12	28:57:02.7		$u^{\prime}$	3250	5	2013 Feb 08
(z ∼ 0.080)					$g^{\prime}$	2940	7	2013 Feb 08
					$r^{\prime}$	2940	7	2013 Feb 13
Abell 1126		10:53:50.90	16:57:01.7		$u^{\prime}$	3320	5	2013 Jan 17
(z ∼ 0.084)					$g^{\prime}$	2940	7	2013 Jan 14
					$r^{\prime}$	2940	7	2013 Jan 16
Abell 1139		10:59:08.67	01:42:26.7		$g^{\prime}$	2940	7	2013 Apr 17
(z ∼ 0.040)					$r^{\prime}$	2940	7	2013 Apr 17
Abell 1278		11:30:31.60	20:41:40.9		$u^{\prime}$	3320	5	2013 Feb 14
(z ∼ 0.134)					$g^{\prime}$	2940	7	2013 Apr 09
					$r^{\prime}$	2940	7	2013 Apr 10
Abell 2061		15:21:18.87	30:36:15.5		$u^{\prime}$	3250	5	2013 May 12
(z ∼ 0.078)					$g^{\prime}$	2940	7	2013 May 12
					$r^{\prime}$	2940	7	2013 May 13
Abell 2249		17:09:46.67	34:32:53.2		$u^{\prime}$	3250	5	2013 May 13
(z ∼ 0.085)					$g^{\prime}$	2940	7	2013 May 14
					$r^{\prime}$	2940	7	2013 Jun 05
Abell 2589		23:24:01.50	16:56:23.0		$u^{\prime}$	3320	5	2012 Jul 25
(z ∼ 0.041)					$g^{\prime}$	2940	7	2012 Jul 25
					$r^{\prime}$	2940	7	2012 Jul 26

Note.

^aThe number in parentheses refers to the exposure time of short-exposure images for the Magellan data.

Download table as:  ASCIITypeset image

All clusters were observed with two different exposures, long and short, for detection of low surface brightness features and a reliable photometry without light saturation, respectively. For long exposures, we targeted five dithered points to recover the gaps in the mosaic and reduce instrumental artifacts such as scattered light in the course of data reduction. The dithering is primarily necessary for the detection of low surface brightness features because faint features can be exposed only in the images where a variation in the background is well removed. Integrated exposure times are 1250 s for Abell 3574 and 2500 s for the others. The center of bright galaxies is saturated in the images with long exposure time, and therefore we took an additional image for each cluster with short exposure for an accurate estimation of magnitudes.

Calibration frames including bias and dome flats were taken during the daytime. Ten sky flats were acquired at twilight and dawn on the same day with appropriate exposure time according to sky flux at the moment. Five standard fields in the southern standard catalog⁹ (Smith et al. 2005) were also observed on the night for each filter in various airmass.

2.1.2. The CFHT MegaCam Observation

2.2.1. Preprocessing

2.2.2. Super-sky Subtraction

2.2.3. Image Coaddition

We have updated the world coordinate system (WCS) of the Magellan data based on stars in the USNO-B catalog (Monet et al. 2003) having low proper motions and similar magnitudes of our target galaxies. Five dithered images were coadded into a single deep image using the tasks in the MSCRED package in IRAF.

For the CFHT data, we calculated astrometric and photometric solutions using the Software for Calibrating AstroMetry and Photometry (SCAMP; Bertin 2006). The photometric catalogs from Source Extractor (SExtractor; Bertin & Arnouts 1996) and SDSS were used as input and reference catalogs, respectively. The seven dithered images with 36 CCDs each were combined using SWarp (Bertin et al. 2002). Images were resampled to have a fixed pixel scale of 0185 pixel^–1 using the distortion map from SCAMP. SWarp coadds multi-extension images with the WCS information and flux scales between each dithered image from SCAMP. Finally, we obtained a deep image for each cluster and band. The depth of the $r^{\prime}$-band images is estimated to be 27 and 26.7 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$ based on 3σ of sky rms in a 1 arcsec² bin for the Magellan and CFHT, respectively.

Figure 1 shows the difference in R.A. and decl. between KYDISC and SDSS. The mean difference and standard deviation are approximately equal to one-fourth and 1 pixel size (0185), respectively.

Zoom In Zoom Out Reset image size

In our study, we adopted the auto-magnitude system from SExtractor. For the Magellan data, we measured an instrumental magnitude mainly from short-exposure images to avoid light saturations in long-exposure images. Additionally, we added some galaxies that are located in CCD gaps to our sample from long-exposure images. We confirmed that the target galaxies in the clusters observed using CFHT were free from light saturation, and therefore we measured their instrumental magnitudes from coadded images. The $u^{\prime}$- and $g^{\prime}$-band images were remapped to have the same WCS information as the $r^{\prime}$-band image, which allows detecting objects in the $r^{\prime}$ band and then measuring values in the $u^{\prime}$ and $g^{\prime}$ bands. That is, the flux of each source was measured with the same aperture in all bands.

We used the southern standard stars observed on the night for standardization of the Magellan data. We selected 106 standard stars from the five standard fields that have a small magnitude error, less than 0.02 mag, both in the catalog and in the measurement. The magnitudes of the standard stars were measured using SExtractor with a 14'' circular aperture, which is the same aperture as the reference catalog from Smith et al. (2005). Standard magnitudes were acquired from instrumental magnitudes using the following steps. First, we estimated the airmass term from the slope of the least-squares fit of the magnitude difference between the catalog and the instrument against airmass. Then, we calculated the color term from the least-squares fit to the magnitude difference after applying the airmass term to the standard equation. The zero points are the median values of the magnitude difference between the catalog and this observation after applying the airmass and color terms to the standardization equation. The standard deviation of the magnitude difference between the catalog and this observation is around 0.02 mag.

We adopted standardization coefficients from the Elixir pipeline by the CFHT team for the CFHT magnitudes. We estimated the photometric accuracy by comparing the KYDISC magnitudes with those from the SDSS (Figure 2). Stars and galaxies brighter than 20th mag in the $u^{\prime}$, $g^{\prime}$, and $r^{\prime}$ bands were utilized for this comparison. The average differences between the KYDISC and the corresponding SDSS point spread function (PSF) magnitudes for stars are −0.0157, 0.0008, and 0.0002 in the $u^{\prime}$, $g^{\prime}$, and $r^{\prime}$ bands, respectively. For galaxies, the average differences between the KYDISC and SDSS Petrosian magnitudes are −0.036, −0.017, and −0.012 in the $u^{\prime}$, $g^{\prime}$, and $r^{\prime}$ bands, respectively. The magnitude differences between KYDISC and SDSS are small, and we also could not find a dependency of magnitude differences on colors. We conclude that the KYDISC magnitude system reasonably agrees with the SDSS magnitude system.

Zoom In Zoom Out Reset image size

Galactic foreground extinction was corrected using the dust map from Schlafly & Finkbeiner (2011) assuming a Galactic extinction law with A_V/E(B − V) = 3.1 (Fitzpatrick 1999). We conducted k-correction following the algorithm described in Chilingarian et al. (2010) and Chilingarian & Zolotukhin (2012) on magnitudes and colors.

2.4.1. The Magellan IMACS

2.4.2. Du Pont WFCCD

2.4.3. WIYN Hydra

2.4.4. Radial Velocity

The $r^{\prime}$ deep images were primarily used for both ellipse fit and galfit. We utilized a short-exposure image when we detected light saturation of a target galaxy in a long-exposure image.

4.1.1. The Ellipse Fitting

4.1.2. Two-dimensional Decomposition Using galfit

We visually classified galaxy morphologies based on the Hubble classification and merger features using a set of seven cutout images (Figure 4). Figure 4(a) is a large FOV cutout image. Figures 4(b)–(d) are $r^{\prime}$-band cutout images with different contrasts that enabled us to give a focus on both the outskirts and inner region of a galaxy. Figure 4(e) is a residual image from the ellipse fit that often reveals asymmetric features. Figure 4(f) is a residual image from galfit that helps to detect light beyond the physical models. Figure 4(g) is a color-composite image for detecting features with distinct colors from a galaxy body (e.g., collisional rings, dust lanes). We used $g^{\prime}$ and $r^{\prime}$ color-composite images for four clusters (Abell 1139, Abell 1146, Abell 3574, and Abell 3659) and $u^{\prime}$, $g^{\prime}$, and $r^{\prime}$ color-composite images for the other clusters.

Zoom In Zoom Out Reset image size

Based on a set of seven cutout images, five researchers have independently performed a visual inspection without knowing any of a galaxy's information, such as redshift, magnitude, size, clustercentric distance, and bulge-to-total ratio (B/T), which can give some bias to judgment. Finally, the visual class of each galaxy is determined to be the most frequent decision by five researchers. Table 5 presents the significance of the classification (N_S), the number of researchers who made the same decision as the final classification. Out of 1409 galaxies, 66 and 48 galaxies show the same frequency (N_S = 2) in two classes for the Hubble and merger classifications, respectively. In those cases, the final class was determined by following the decision of one researcher whose classification shows the highest agreement with the final classification: 84% and 92% for the Hubble and merger classifications, respectively.

4.2.1. Morphological Classification

We divided galaxies into four groups based on the Hubble classification: ellipticals (E), lenticulars (S0), early spirals (S_E), and late spirals (S_L). We first identified early types (E/S0) showing smooth light profiles without spiral arms. Classifying E and S0 is complicated, and therefore careful and clear criteria are required to minimize subjectivity. When galaxies show only one component in any contrast images, they are classified into E. Lenticular galaxies having both bulge and disk components sometimes reveal edge-on disks on residual images (Figures 4(e) and (f)). S0 galaxies having face-on disks tend to show rapidly (bulge) and more mildly (disk) shrinking components through Figures 4(b)–(d) due to the different light profiles of the two components.

Late-type galaxies having either spiral arms or prominent disk components (≥40%) were divided into early (S_E) and late (S_L) spirals according to the dominance of two components. The S_E types, including Sa and Sb types in the standard Hubble classification, have prominent bulge components that are brighter than disks and whose size is seemingly larger than 20% of galaxy size in large contrast images (Figures 4(c) and (d)). The S_L types are disk-dominated galaxies, including galaxies from Sc to Irr types whose bulge brightness is competitive with or fainter than disk brightness, and bulge size is smaller than 20% of galaxy size, especially in Figures 4(c) and (d). We present sample images for our classification based on Hubble types in Figure 5. Example cutout sets for each morphological type are shown in Appendix A.

Zoom In Zoom Out Reset image size

Galaxy clusters are the site where the most massive galaxies reside; therefore, we found a large fraction of early-type galaxies in our sample (Table 6). The bulge fraction becomes larger from late to early types (Hudson et al. 2010); therefore, galaxy morphology has a close connection to the B/T. We compared our classification with the B/T in the $r^{\prime}$ band from galfit (Figure 6). The mean values of the B/T were 0.78, 0.61, 0.33, and 0.15 for the E, S0, S_E, and S_L classes, respectively. Our result is comparable with that of Kim et al. (2016), who compared the B/T with the visual classification from Khim et al. (2015).

Zoom In Zoom Out Reset image size

Table 6.  Galaxy Morphology Based on Visual Inspection

Cluster	Hubble Type		Merger Type
		N	OM	PM
Abell 116 (44)	E (11)	0.82 (9)	0.00 (0)	0.18 (2)
	S0 (13)	0.92 (12)	0.00 (0)	0.08 (1)
	SE (11)	0.82 (9)	0.00 (0)	0.18 (2)
	SL (9)	0.33 (3)	0.11 (1)	0.56 (5)
Abell 646 (98)	E (43)	0.91 (39)	0.00 (0)	0.09 (4)
	S0 (37)	0.81 (30)	0.00 (0)	0.19 (7)
	SE (9)	0.56 (5)	0.00 (0)	0.44 (4)
	SL (9)	0.78 (7)	0.00 (0)	0.22 (2)
Abell 655 (180)	E (71)	0.89 (63)	0.00 (0)	0.11 (8)
	S0 (63)	0.78 (49)	0.00 (0)	0.22 (14)
	SE (26)	0.46 (12)	0.04 (1)	0.50 (13)
	SL (20)	0.70 (14)	0.00 (0)	0.30 (6)
Abell 667 (117)	E (42)	0.90 (38)	0.05 (2)	0.05 (2)
	S0 (38)	0.89 (34)	0.00 (0)	0.11 (4)
	SE (19)	0.37 (7)	0.00 (0)	0.63 (12)
	SL (18)	0.50 (9)	0.06 (1)	0.44 (8)
Abell 690 (88)	E (27)	0.89 (24)	0.00 (0)	0.11 (3)
	S0 (37)	0.73 (27)	0.03 (1)	0.24 (9)
	SE (12)	0.58 (7)	0.00 (0)	0.42 (5)
	SL (12)	0.42 (5)	0.08 (1)	0.50 (6)
Abell 1126 (73)	E (22)	0.86 (19)	0.05 (1)	0.09 (2)
	S0 (31)	0.65 (20)	0.06 (2)	0.29 (9)
	SE (15)	0.60 (9)	0.07 (1)	0.33 (5)
	SL (5)	0.60 (3)	0.20 (1)	0.20 (1)
Abell 1139 (52)	E (13)	0.92 (12)	0.08 (1)	0.00 (0)
	S0 (15)	0.87 (13)	0.00 (0)	0.13 (2)
	SE (11)	0.64 (7)	0.27 (3)	0.09 (1)
	SL (13)	0.54 (7)	0.08 (1)	0.38 (5)
Abell 1146 (121)	E (66)	0.83 (55)	0.06 (4)	0.11 (7)
	S0 (37)	0.73 (27)	0.08 (3)	0.19 (7)
	SE (17)	0.53 (9)	0.06 (1)	0.41 (7)
	SL (1)	1.00 (1)	0.00 (0)	0.00 (0)
Abell 1278 (41)	E (16)	0.81 (13)	0.06 (1)	0.13 (2)
	S0 (10)	0.70 (7)	0.00 (0)	0.30 (3)
	SE (10)	0.60 (6)	0.00 (0)	0.40 (4)
	SL (5)	0.20 (1)	0.40 (2)	0.40 (2)
Abell 2061 (250)	E (79)	0.90 (71)	0.04 (3)	0.06 (5)
	S0 (92)	0.82 (75)	0.03 (3)	0.15 (14)
	SE (58)	0.50 (29)	0.09 (5)	0.41(24)
	SL (21)	0.67 (14)	0.00 (0)	0.33 (7)
Abell 2249 (240)	E (94)	0.88 (83)	0.03 (3)	0.09 (8)
	S0 (110)	0.81 (89)	0.01 (1)	0.18 (20)
	SE (22)	0.55 (12)	0.18 (4)	0.27 (6)
	SL (14)	0.71 (10)	0.00 (0)	0.29 (4)
Abell 2589 (67)	E (15)	0.93 (14)	0.00 (0)	0.07 (1)
	S0 (41)	0.85 (35)	0.02 (1)	0.12 (5)
	SE (6)	0.67 (4)	0.00 (0)	0.33 (2)
	SL (5)	0.60 (3)	0.00 (0)	0.40 (2)
Abell 3574 (14)	E (2)	0.00 (0)	0.50 (1)	0.50 (1)
	S0 (3)	0.33 (1)	0.33 (1)	0.33 (1)
	SE (6)	1.00 (6)	0.00 (0)	0.00 (0)
	SL (3)	1.00 (3)	0.00 (0)	0.00 (0)
Abell 3659 (24)	E (5)	0.80 (4)	0.00 (0)	0.20 (1)
	S0 (10)	0.70 (7)	0.10 (1)	0.20 (2)
	SE (9)	0.56 (5)	0.22 (2)	0.22 (2)
	SL (0)	0.00 (0)	0.00 (0)	0.00 (0)

Download table as:  ASCIITypeset images: 1 2

4.2.2. Merger-related Disturbance Identification

Galaxies showing evidence of galaxy mergers were identified from galaxies without any features (N) by focusing on whether they show any disturbance in morphology, including one or more of the following features: merger features, multiple cores, tidal features, shell structures, unusual dust lanes, and asymmetric light distribution. An example cutout image set for each feature is shown in Appendix A.

We divided galaxies showing disturbed features into two subgroups according to the existence of their companions: galaxies with features caused by the ongoing mergers with close pairs within four times $\overline{{r}_{90}}$ (OM) and galaxies with features seemingly related to mergers in the past without relevant pairs for the features (PM). Figure 4(a) was useful for identifying OM by examining the existence of counterparts for the merger-related features. We present sample images of OM and PM in Figure 7.

Zoom In Zoom Out Reset image size

The identification of merger-related disturbance features is admittedly subjective, having a chance to be contaminated by other environmental interactions. Ram-pressure stripping making features along the direction of the path is one of the possible contaminations. However, our post-merger galaxies tend to have asymmetric features that are distinct from the features generated by ram-pressure stripping. We measured the CAS asymmetry parameter on the $r^{\prime}$-band deep images as a sanity check of our classification (Abraham et al. 1996; Conselice 2003). We limited the measurement window to 1.5 times the radii containing 90% of the total luminosity to reduce contaminations. Merger-related galaxies (PM and OM) have a distinct distribution of the asymmetry parameter compared with that of normal galaxies, although the distributions are not separated sufficiently to identify merger-related galaxies based solely on the asymmetry parameter (Figure 8). Late-type galaxies (S_E and S_L) show a larger value of asymmetry than early-type galaxies, likely due to their spiral features, which weaken the ability of the asymmetry parameter to differentiate between normal and merger-related groups.

Zoom In Zoom Out Reset image size

Our post-merger sample can be contaminated by fly-by interactions. Fly-by encounters are expected to generate similar features as minor mergers even without actual collisions (Sinha & Holley-Bockelmann 2015). The frequency and outcome of fly-bys in the cluster environment are less informed and should be constrained by simulations. However, fly-bys can be considered subcategories of galaxy interactions. In this study, we assumed fly-bys bring a similar outcome as minor mergers.

4.2.3. Spatial Resolution

We investigated galaxy morphology in our sample clusters as a function of the rest-frame cluster velocity dispersion (σ_cl), which is a proxy for the cluster mass. The fraction of each morphology is clearly related to the cluster velocity dispersion (Figure 9). Less massive clusters (e.g., σ_cl < 600 km s⁻¹) show a similar portion (∼25%) of all morphological types, whereas massive clusters (e.g., σ_cl > 800 km s⁻¹) are dominated by early-type galaxies. We also examined the correlation between cluster velocity dispersion and morphological contents based on the B/T from galfit. In Figure 10, we divide galaxies into four subgroups according to their B/T: 0.7 < B/T; 0.4 < B/T < 0.7; 0.2 < B/T < 0.4; B/T < 0.2. The fraction of B/T subgroups increases or decreases according to σ_cl in common with that of visual morphology. We employed the Pearson correlation to evaluate the statistical significance of the correlations. In Figures 9 and 10, the R coefficient (R_p) and p-value from the Pearson correlation are presented in the top right corner of each panel. The p-value represents the probability of detecting a correlation by random shuffling; therefore, the lower the p-value, the more significant the correlation. The correlations between visual morphology and σ_cl have p-values lower than 0.05, except for S0 galaxies. Also, we found low p-values in the correlations between the fraction of B/T subgroups and σ_cl. Our results suggest that morphological contents of clusters based on both visual classifications and B/Ts significantly correlate with cluster velocity dispersion of local clusters.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We investigate the fractions of early-type galaxies (E and S0) and galaxies having B/T > 0.4 as a function of clustercentric coverage in Figure 11. Large values of the Pearson R coefficient (i.e., R_p > 0.7) and small p-values (i.e., p-value <0.05) indicate that both early-type and bulge-dominant galaxy fractions strongly correlate with cluster velocity dispersion, especially for the sample covering a large clustercentric distance (R/R₂₀₀ > 1). Moderate correlations are found within R/R₂₀₀ < 0.6, and the correlation becomes stronger (large values of R_p and slope) and significant (small p-value) toward large clustercentric coverage until R/R₂₀₀ = 1.5. Almost identical trends were detected for the limiting coverage in the range 1.5 < R/R₂₀₀ < 2. These results imply that morphological transformations happen within the virialized region of clusters, causing the dominance of early-type (or B/T > 0.4) galaxies in the cluster core (R/R₂₀₀ < 0.6), which dilutes the morphological dependence on cluster mass.

Zoom In Zoom Out Reset image size

A small sample size of the small geometrical coverage can also weaken the morphology–σ_cl correlation in Figure 11, and the effect of sample size should be considered. For each geometrical coverage, we randomly selected the same number of galaxies at R < 2 R₂₀₀ and measured R_p for the (E+S0)–σ correlation (left panels of Figure 11). We repeated the random selection 1000 times and calculated the median values of R_p ($\overline{{R}_{{\rm{p}}}}$) and the standard deviation of the distribution (${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$). In Table 9, (${R}_{{\rm{p}}}-\overline{{R}_{{\rm{p}}}}$) shows the difference in R_p between geometrical and random selections. The R/R₂₀₀ < 1.5 sample shows the similar R_p to that from the random selection, and the difference is small (0.26 ${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$). However, the small geometrical coverage samples (R/R₂₀₀ < 0.6 or R/R₂₀₀ < 0.8) show a 2σ smaller R_p (worse correlation) than that from the random selection ($\overline{{R}_{{\rm{p}}}}$). Therefore, we can conclude that small geometrical coverage indeed weakens the morphology–σ_cl correlation.

Table 9.  The R_p Distribution from the Random Selection (E+S0)

Geometrical Selection			Random Selectiona
Coverage	Ngb	Rpc	$\overline{{R}_{{\rm{p}}}}$ d	${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$ e	(${R}_{{\rm{p}}}-\overline{{R}_{{\rm{p}}}}$)
R/R200 < 0.6	688	0.494	0.696	0.106	−1.91 ${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$
R/R200 < 0.8	861	0.557	0.716	0.070	−2.27 ${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$
R/R200 < 1.0	978	0.660	0.728	0.056	−1.21 ${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$
R/R200 < 1.5	1219	0.746	0.737	0.034	0.26 ${\sigma }_{{{\rm{R}}}_{{\rm{p}}}}$

Notes.

^aWe randomly selected N_g galaxies at R < 2 R₂₀₀ and repeated 1000 times. ^bNumber of galaxies within the geometrical coverage. ^cThe R_p from the N_g geometrically selected galaxies. ^dThe median R_p from the random selection. ^eThe standard deviation of R_p from the random selection.

Download table as:  ASCIITypeset image

Observational studies have reported the dependence of morphological contents on the cluster velocity dispersion at high redshifts (Postman et al. 2005; Desai et al. 2007; Poggianti et al. 2009); however, it was not confirmed in the local universe. Poggianti et al. (2009) investigated the morphology of galaxies in clusters at 0.04 < z < 0.07 based on the WIde-field Nearby Galaxy-cluster Survey (WINGS; Fasano et al. 2006) and could not find a clear correlation between the morphological fraction and the cluster velocity dispersion. Likewise, Simard et al. (2009) reported that there is no trend between the fraction of early-type galaxies and the cluster velocity dispersion based on local clusters from the SDSS data.

These earlier findings on the morphological dependence on the cluster velocity dispersion conflict with our results. We suspect that different sampling strategies caused the discrepancy on this issue. First, both the Poggianti et al. and Simard et al. samples are limited to the galaxies located inside 0.6 R₂₀₀, whereas we extended our sampling criterion to 2 R₂₀₀. Besides, the Poggianti et al. sample is limited to the clusters whose cluster velocity dispersion is greater than 500 km s⁻¹. In Figure 11, it is difficult to find a correlation with only σ_cl > 500 km s⁻¹ clusters, even for the R < 2 R₂₀₀ sample.

Low levels of spectroscopic completeness of previous studies might also have weakened the morphology–σ_cl correlation. Following Cava et al. (2009), the spectroscopic completeness of WINGS clusters is around 0.5, which is considerably smaller than that of this study (Table 4). We presume that the spectroscopic completeness of Simard et al. (2009) is also significantly lower than that of this study. For the sample galaxies within the SDSS footprint, the spectroscopic completeness based only on the SDSS data is 36% (1276/3577), which is much smaller than ours (71%; 2547/3577). Spectroscopically incomplete samples have more chance to be biased toward specific types of galaxies (usually toward bright galaxies), which can generate a bias to the morphological fraction of clusters. Also, spectroscopic incompleteness can bring a significant error on the estimation of cluster velocity dispersion.

Having spectroscopic information for all the galaxies in the field allows one to accurately measure local densities. In this study, however, the spectroscopic completeness reaches 1 in only two clusters (Table 4). The measurement based on projected distance highly overestimates local density due to foreground/background contamination. For a more realistic measurement of local density, we followed the method described in Fasano et al. (2015). First, we estimated the circular area per Mpc (a_i) that includes the ith nearest neighbor by the equation

$$\begin{eqnarray}&&{N}_{i}=i/{f}_{c}-{N}_{F}\,{a}_{i},\end{eqnarray} \tag{ 3 }$$

where f_c is the imaging coverage fraction for a_i, and N_F is the number of contaminations per Mpc. In most cases, a_i is fully covered by our image, and therefore f_c = 1. However, the local density of galaxies on the edge of the image can be overestimated because we cannot count galaxies beyond the FOV. In such cases, we calculated f_c by dividing the covered area by a_i. Out of 1409 total samples, 51 galaxies have f_c smaller than 1. We adopted N_F from Berta et al. (2006), who counted the number of field galaxies per square degree and per V-band magnitude based on the ESO-Spitzer wide-area Imaging Survey (ESIS) data. The V-band magnitude in their study was transformed into the $r^{\prime}$ magnitude by utilizing the transformation formula in Fukugita et al. (2007). We calculated a_i by increasing i until we got N_i greater than 10. Then, we calculated the area including the 10th nearest neighbor (A₁₀) by interpolating a_i and a_i − 1. Finally, we computed the local density (Σ₁₀) using the 10th nearest neighbor as follows:

$$\begin{eqnarray}&&\mathrm{log}\,{{\rm{\Sigma }}}_{10}=10/{A}_{10}+0.178({f}_{c}-1).\end{eqnarray} \tag{ 4 }$$

The term 0.178(f_c – 1) comes from Fasano et al. (2015) by comparing the local density with the result, including the SDSS galaxies that locate outside their FOV.

There is a correlation between local density and clustercentric distance (Figure 12). The cluster center shows the highest local density, and local density decreases as the distance from the cluster center increases. Within the virial radius (R/R₂₀₀ < 1), local density shows a tight correlation with R/R₂₀₀. Although the outskirts region of clusters (R/R₂₀₀ > 1) tends to show a smaller value of local density, the large scatter in Σ₁₀ indicates that significant numbers of galaxies in R/R₂₀₀ > 1 are located in locally dense regions.

Zoom In Zoom Out Reset image size

We present the morphology–density relation based on cluster galaxies in Figure 13. Highly dense regions preferentially have E and S0 galaxies. The lower the local density, the lower the E fraction and the higher the S_E and S_L fractions. The correlation between local density and clustercentric distance generates similar results on the morphology–density and morphology–radius relations. Smaller values of the clustercentric distance increase the fraction of early-type galaxies. As a result, we barely find S_E or S_L galaxies in the central region of clusters. The changes in the S0 fraction against local density (or clustercentric distance) are not significant. The overall trend of our cluster morphology–density relation is consistent with previous studies. Dressler et al. (1997), including a revised version of the Dressler (1980) morphology–density relation, showed that elliptical and spiral fractions are well correlated to local density, whereas S0 fractions have relatively little dependence on local density within the cluster environment. Fasano et al. (2015) also reported a relatively constant fraction of S0s to local density based on WINGS clusters.

Zoom In Zoom Out Reset image size

We found around 40% of S0 galaxies at the cluster outskirts region (R/R₂₀₀ > 1) before they were gravitationally bound to the cluster potential. Considering the low fraction of S0s in field environments (∼24%; e.g., Calvi et al. 2012), we presume that the major morphological transformations from spirals to S0s happened in the intermediate environments, somewhere between clusters and fields.

The group environment can be a possible place where this transformation happens (preprocessing; e.g., Mihos 2004). Previous studies suggest that group environments also hold similar environmental effects as clusters, such as strangulation, tidal interactions, and ram-pressure stripping, which is considered one of the main mechanisms to trigger morphological transformations by removing gas in the outer disk (Hester 2006; Rasmussen et al. 2006). However, the effect of ram-pressure stripping in the group environment is much milder than that in massive clusters Rasmussen et al. 2008); therefore, it is considered inefficient for triggering morphological transformations of large spirals. Freeland et al. (2010) also suggested that it is hard to explain the H i deficiency of group galaxies solely with ram-pressure stripping. Therefore, it is speculated that ram-pressure stripping is less likely to be the main mechanism for the morphological transformation that occurred before the galaxies came to cluster environments.

If a galaxy–galaxy merger is the most effective way of morphological transformation (Barnes & Hernquist 1992), it could have preferentially occurred in group environments rather than in cluster environments. Just et al. (2010) reported that S0 fractions have rapidly increased in groups or poor clusters since z ∼ 0.5, but not in rich clusters, suggesting that galaxy–galaxy interactions are responsible for the present S0 galaxies. The merger-induced S0 formation scenario also explains a nonnegligible fraction of S0 galaxies in field environments.

Nearly constant S0 frequency over the cluster region does not necessarily rule out the morphological transformation in the cluster environment. Moreover, we still detect increasing/decreasing fractions of elliptical/spiral galaxies with respect to increasing local density or decreasing clustercentric distance. Although we cannot pinpoint the mechanism causing morphological transformations in the cluster environment, our results suggest that transformations from spirals to S0s and from S0s to ellipticals would happen at a similar rate, and the S0 fraction remains the same.

Ultraviolet fluxes are sensitive to young stars and often used to identify galaxies hosting recent episodes of star formation (RSFs; Yi et al. 2005; Kaviraj et al. 2007; Jeong et al. 2009; Sheen et al. 2016). We tried to get a hint of environmental quenching on cluster galaxies based on the optical and near-ultraviolet (NUV; λ_eff = 2267 Å) fluxes obtained from this study and the Galaxy Evolution Explorer (GALEX; Martin et al. 2005) GR 6 plus 7 Data Release.

In Figures 14 and 15, we present the median $g^{\prime}$ − $r^{\prime}$ and NUV − $r^{\prime}$ colors and the RSF galaxy fraction (f_RSF) of each morphological type as a function of local density and clustercentric distance, respectively. The RSF galaxies are conservatively defined by selecting galaxies whose NUV − $r^{\prime}$ is lower than 4, corresponding to the value 2σ below the red sequence of E types in the NUV − $r^{\prime}$ color–magnitude diagram.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The median colors of late-type galaxies are much bluer than those of early-type galaxies, even in the cluster environment. The S_L-type galaxies are dominated by RSF galaxies (f_RSF ∼ 0.9) maintaining their blue color regardless of local density, although we barely find S_L-type galaxies in the highly dense region (i.e., cluster center). On the other hand, S_E types show a significant change in color according to local density, especially at dense regions (e.g., log Σ₁₀ > 1.5 or R/R₂₀₀ < 0.5) where we found a sudden decrease in f_RSF. This finding is in agreement with that of Solanes et al. (2001), who reported that the H i–deficient fraction of early spirals significantly increases toward the cluster center but that of late spirals rather mildly increases. We could not find a statistically significant difference in the mean mass and B/T according to local density within the same morphological type. Therefore, the color change in S_E types is less likely to come from the difference in the mass or bulge fraction.

Late-type galaxies actively forming stars seem to lose their gas in the cluster potential. The cosmological hydrodynamic simulation by Cen et al. (2014) suggested that gas-rich galaxies lose most of their cold gas during the first infall into the cluster center. In that sense, a significant number of S_E types would have spent a long time in the cluster environment losing their gas. As a result, we found a significant change in colors and f_RSF of S_E at R/R₂₀₀ < 0.5. On the other hand, it may not have been long enough for S_L types to lose their cold gas since their infall into the cluster environment.

Lenticulars hardly change their red colors and show low f_RSF within the virialized region (i.e., log Σ₁₀ > 0.5 or R/R₂₀₀ < 1). However, we can find bluer NUV − $r^{\prime}$ color and higher f_RSF from S0s at cluster outskirts regions. Present cluster S0s would include both populations that already had S0 morphology even before they came into clusters (hereafter preformed S0s) and populations that were originally spirals when they arrived at the cluster outskirts and transformed into S0s within the cluster environments (hereafter transformed S0s). Most S0s at R/R₂₀₀ > 1 would be preformed populations, but S0s within the virialized region (R/R₂₀₀ < 1) would be a mixture of both preformed and transformed populations, which makes it difficult to find the unique origin of (cluster) S0s.

For transformed S0s, we suspect that the star formation was quenched in a cluster before they changed their morphology from spirals into S0s, and therefore we do not detect bluer colors and higher f_RSF than Es within the virialized region. The changes in colors and f_RSF of S_E types also support this scenario; therefore, we suspect that red S_E galaxies will eventually transform into S0s. The blue NUV − $r^{\prime}$ color and higher f_RSF of preformed S0s at R/R₂₀₀ > 1 imply that their morphology transformation preceded star formation quenching. That is, a significant fraction of preformed S0s may not be generated by a star formation quenching derived by gas stripping. This explanation accords closely with our discussion of the constant S0 fraction at cluster outskirts in the previous section. Moreover, we found a link between galaxies with merger features and RSF galaxies, which will be discussed in the following series of this study based on the phase-space analysis by Rhee et al. (2017).

Elliptical galaxies show red colors and low f_RSF (∼0.1) no matter where they are located within a cluster. The star formation quenching of transformed Es would precede morphological transformation because S0s at R/R₂₀₀ < 1, possible progenitors of transformed Es, have already been quenched. Although preformed Es seem to have been quenched before they arrived in the cluster environment, it is difficult to speculate on the quenching mechanism with our data. Observational and theoretical studies for the intermediate environment (i.e., groups) would give a clue to this question.

Galaxy clusters are considered a hostile environment for galaxy mergers, and we indeed found a low frequency (∼4%) of ongoing mergers. Note that the ongoing merger fraction is exactly matched with that in Sheen et al. (2012). However, around 20% of cluster galaxies within R/R₂₀₀ < 2 show signatures of post-mergers. In earlier work, Sheen et al. (2012) reported that 24% of 167 red-sequence galaxies show signatures of post-mergers in four rich Abell clusters, whereas Adams et al. (2012) detected 3% of post-merger galaxies among their 3551 early-type galaxies in the cluster environment. Even considering the subjectivity of the visual inspection, these two studies show contradicting results of the frequency of recent mergers in the cluster environment. The limiting surface brightness of Sheen et al. (2012) reaches μ_r' ∼ 28 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$, and Adams et al. (2012) and this study have used images with similar depth (μ_r' ∼ 26.5 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$ in 3σ). We can understand more detection of faint features in Sheen et al. (2012), because faint features can survive extremely long in low surface brightness (Ji et al. 2014). However, there is more than 15% difference in the post-merger frequency between Adams et al. (2012) and this study.

Sheen et al. (2012) and this study conducted a visual inspection using both original and model subtracted images; Adams et al. (2012), however, inspected only residual images after subtracting a model from the ellipse procedure. We experienced that the ellipse task sometimes fits faint features and diminishes the light related to disturbed features, making it hard to be detected. Moreover, Adams et al. (2012) examined only the outer regions of galaxies to deal with extended light by masking out regions brighter than 1% of the peak luminosity of a galaxy. In this study, we detect merger-related features (e.g., multiple cores, asymmetric light distribution) even near the galactic center. Besides, both Sheen et al. (2012) and this study utilized spectroscopically confirmed cluster members, whereas Adams et al. (2012) selected cluster galaxies based on the red sequence, which can contain foreground/background contamination. These are possible explanations for the discrepancy in the detection of merger features, but we are still puzzled by such a big difference in the frequency of recent mergers in the cluster environment.

We inspected whether galaxy mergers are preferentially found in more or less massive clusters (Figure 16). We could not find a clear correlation between the frequency of galaxy mergers and the cluster velocity dispersion. We also detect a small fraction of ongoing mergers regardless of the cluster velocity dispersion. The large p-values from the Pearson correlation also support that there is no relation between the frequency of mergers and cluster velocity dispersion. This result implies that the frequency of galaxy mergers is less correlated with the cluster potential.

Zoom In Zoom Out Reset image size

We examined the frequency of galaxy mergers according to local density and the clustercentric distance (Figure 17). The PM fraction gradually decreases toward the dense region or the cluster center, whereas we cannot find clear tendencies of the detection rate of OM with local density. Our findings generate speculation that the merger events that made the features we detect now preceded the galaxy accretion into the cluster environment. The more frequent detection of PM in the cluster outskirts also supports our scenario.

Zoom In Zoom Out Reset image size

The negative correlation between clustercentric distance and the time since galaxies infall into the cluster environment explains the decreasing PM fraction toward the cluster center (Bahé et al. 2012; Rhee et al. 2017). Following Figure 3 in Bahé et al. (2012), most of the galaxies in the cluster core (i.e., R/R₂₀₀ < 0.3) have spent 4 Gyr or more in cluster environments. Yi et al. (2013) numerically estimated the lifetime of merger features and suggested that merger features can survive 3.9 Gyr in μ_r' ∼ 28 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$ depth. Although it is difficult to directly adopt the lifetime because it can vary depending on the type of merger (e.g., mass ratio, merging orbit, and gas contents), we can expect a smaller PM fraction in the central region than in the outskirts region of clusters.

In our result, we found around a five times higher fraction of PM galaxies (∼20%) than of OM galaxies (∼4%). Following the hydrodynamic simulation by Ji et al. (2014), merger features can live 2.56 times longer than the merging timescale in a cluster potential assuming μ_r' ∼ 28 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$ (1.79 times for μ_r' ∼ 25 $\mathrm{mag}\,{\mathrm{arcsec}}^{-2}$). The evolution of the merger rate is another factor to be considered. Observational and theoretical studies indicated that the merger rate in the past was higher than it is today (e.g., Bundy et al. 2009; Bridge et al. 2010; Lotz et al. 2011; Lee & Yi 2013; Rodriguez-Gomez et al. 2015). Following the prediction by Rodriguez-Gomez et al. (2015), the evolution of the merger rate is proportional to (1 + z)^2.4−2.8, which suggests that the merger rate declined by roughly a factor of 1.5 within the last 3 Gyr.

The low OM fraction (4.7%) does not account for the observed PM fraction (25.3%) at the cluster outskirts (1.5 < R/R₂₀₀ < 2), even though we consider the evolution of the merger rate and the timescale of merger features. The mergers that made PM galaxies probably happened beyond 2 R₂₀₀ because of the time delay between OM and PM galaxies. We investigated PM and OM fractions beyond 2 R₂₀₀ using 207 galaxies that are brighter than −19.8 in the $r^{\prime}$ band and have a similar redshift range to target clusters but are located beyond 2 R₂₀₀ and found 16 OM (7.7%) and 56 PM (27.1%) galaxies.

The significant difference in the detection of PM and OM in the cluster environment can be marginally explained by adopting all explanations regarding the evolution of merger rate, timescale of merger features, and higher OM fraction beyond 2 R₂₀₀. However, our spectroscopic observations do not fully cover those fields beyond 2 R₂₀₀, and the spectroscopic completeness of the R > 2 R₂₀₀ region is around 0.56, which is somewhat lower than that of the R < 2 R₂₀₀ region (0.72). Besides, there are a lot of uncertainties in estimating timescales of mergers and features according to the details of mergers such as mass ratio, progenitor type, merging orbit, and environment. The PM and OM galaxies in the cluster environment can be fully explained through idealized observations targeting galaxies in neighboring cluster regions (i.e., R > 2 R₂₀₀) and predictions of merging timescales from simulations for various types of mergers.