1: \documentclass{emulateapj}
2: \usepackage{graphicx}
3: %\documentclass[12pt,preprint]{aastex}
4: \usepackage{times}
5:
6: \begin{document}
7:
8:
9: \markboth{Cheng-Jiun Ma} {Spectroscopic diagnostics in MACSJ0717}
10: %\shortauthors{Ma and Ebeling}
11: \title{The spatial distribution of galaxies of different spectral types in the massive intermediate-redshift
12: cluster MACSJ0717.5+3745\footnote{Based in part on data collected at Subaru
13: Telescope, which is operated by the National Astronomical Observatory of
14: Japan. Based also, in part, on observations obtained with MegaPrime/MegaCam,
15: a joint project of CFHT and CEA/DAPNIA, at the Canada-France-Hawaii
16: Telescope (CFHT) which is operated by the National Research Council (NRC) of
17: Canada, the Institute National des Sciences de l'Univers of the Centre
18: National de la Recherche Scientifique of France, and the University of
19: Hawaii. The spectroscopic data presented herein were obtained at the
20: W.M.\ Keck Observatory, which is operated as a scientific partnership among
21: the California Institute of Technology, the University of California and the
22: National Aeronautics and Space Administration. The Observatory was made
23: possible by the generous financial support of the W.M. Keck Foundation.}}
24: \author{Cheng-Jiun Ma, Harald Ebeling, David Donovan, and Elizabeth Barrett}
25: \affil{Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822, USA}
26:
27: \begin{abstract}
28:
29: We present the results of a wide-field spectroscopic analysis of the galaxy
30: population of the massive cluster MACSJ0717.5+3745 and the surrounding
31: filamentary structure (z$=$0.55), as part of our systematic study of the 12 most
32: distant clusters in the MACS sample. Of 1368 galaxies spectroscopically observed
33: in this field, 563 are identified as cluster members; of those, 203 are
34: classified as emission-line galaxies, 260 as absorption-line galaxies, and 17 as
35: E+A galaxies (defined by $\frac{H_{\delta}+H_{\gamma}}{2}>6$\AA\ and no
36: detection of [OII] and $H_{\beta}$ in emission). The variation of the fraction
37: of emission- and absorption-line galaxies as a function of local projected
38: galaxy density confirms the well-known morphology-density relation, and becomes
39: flat at projected galaxy densities less than $\sim 20$Mpc$^{-2}$. Interestingly,
40: 16 out of 17 E+A galaxies lie (in projection) within the ram-pressure stripping
41: radius around the cluster core, which we take to be direct evidence of
42: ram-pressure stripping being the primary mechanism that terminates
43: star-formation in the E+A population of galaxy clusters. This conclusion is
44: supported by the rarity of E+A galaxies in the filament which rules out galaxy
45: mergers as the dominant driver of evolution for E+A galaxies in clusters. In
46: addition, we find the 42 e(a) and 27 e(b) member galaxies, i.e., the
47: dusty-starburst and starburst galaxies respectively, to be spread out across
48: almost the entire study area. Their spatial distribution, which shows a strong
49: preference for the filament region, suggests that starbursts are triggered in
50: relatively low-density environments as galaxies are accreted from the field
51: population.
52:
53: \end{abstract}
54:
55: \keywords{Galaxy: evolution, Galaxy:formation, galaxies:clusters:general, techniques:spectroscopic}
56:
57: \section{Introduction}
58:
59:
60: It is widely accepted that the effect of environment is one of the most
61: important factors for galaxy evolution. Unlike field galaxies for which the
62: dominant physical mechanism driving evolution is galaxy-galaxy mergers, as has
63: been revealed in many large-scale surveys \citep{gome03,lewi03,coop07},
64: cluster galaxies evolve in a more complex fashion (see, e.g., the review by
65: \citet{bowe04}). As a result, cluster galaxies behave differently from field
66: galaxies at the same epoch. One of the best-studied examples of this difference
67: between the properties of galaxies in clusters and in the field is the
68: morphology-density relation\citep{oeml74,dres80}, i.e., the fact that the
69: fraction of late-type galaxies drops precipitously toward the center of a
70: cluster, whereas the fraction of early-type galaxies increases dramatically. In
71: addition, the Butcher-Oemler effect\citep{bo78}, which states that galaxies in
72: rich clusters tend to be bluer and more active (i.e.\ feature emission lines)
73: with increasing cluster redshift out to at least $z\sim$ 0.8, implies that the
74: build-up of the morphology-density relation is still ongoing at redshifts less
75: than unity. Therefore, much effort \citep{couc87,dres97, pogg99, smit05,
76: post05,mora06,tran07} has gone into establishing whether and, if so, how blue
77: late-type galaxies are transformed into early-type galaxies, specifically S0
78: galaxies.
79:
80: Clusters at intermediate redshift ($z\sim 0.5$) provide us with a unique
81: opportunity to examine this hypothesized transformation as well as the dominant
82: physical mechanisms driving it. Recent work in this field includes a study of
83: the morphological distribution of galaxies in the cluster Cl0024+16 at $z=0.39$
84: \citep{treu03}, and a comparison of the spatial distribution and spectroscopic
85: properties of spiral galaxies with quenched star formation (passive spirals) and
86: newly formed S0 galaxies in, again Cl0024+16, as well as in the more massive
87: system MS0451.6--0305 at $z=0.55$ \citep{mora07}. The authors of these studies
88: suggest that ram-pressure stripping \citep{gunn72} is not the only mechanism
89: involved in building up the population of S0 galaxies from late-type galaxies,
90: although it does accelerate the process. The transformation is actually
91: initiated much farther away from the cluster center, and at a slower pace, by
92: the processes of strangulation \citep{lars80} or galaxy-galaxy harassment
93: \citep{moor96}. Additional evidence of this evolution from blue late-type
94: galaxies to S0 galaxies is presented by \citet{tran07} who, using the
95: spectroscopic diagnostics H$_{\delta}$ and $D_n(4000)$ \citep{kauf03} in their
96: study of the cluster MS1054.4--0321 ($z=0.823$), confirms that S0 galaxies in
97: the cluster are, in general, younger than the elliptical
98: galaxies. \citet{tran07} suggest that these S0 galaxies have recently been
99: converted from late-type galaxies falling in from the field (see also
100: \citet{tran05}).
101:
102: To improve our understanding of the different physical effects governing galaxy
103: evolution from the cluster outskirts to the cluster core, we have embarked on an
104: extensive magnitude-limited wide-field spectroscopic survey of the galaxy
105: population of 12 clusters with redshifts above $0.5$, selected from the MAssive
106: Cluster Survey (MACS) \citep{ebel04,ebel07}. In this paper we present results
107: from our pilot study of MACSJ0717.5+3745 ($z=0.545$). MACSJ0717.5+3745, which
108: has the most extensive spectroscopic data set of these 12 at present, features a
109: complex X-ray morphology in the cluster core as well as a giant filament
110: described by \citet{ebel04}. We here focus on the spectroscopic analysis and
111: classification of galaxies in MACSJ0717.5+3745, with special emphasis on E+A
112: galaxies, a galaxy type that, as shall be explained in the following section,
113: holds great promise as a diagnostic for cluster-related mechanisms of galaxy
114: evolution.
115:
116: This paper is organized as follows. In \S2, we review the properties of E+A
117: galaxies. In \S3, we describe the data used for this study and give an overview
118: of the data reduction procedure. In \S4 we discuss the selection of the cluster
119: members used in our analysis, and describe the equivalent-width measurement and
120: spectral classification. In \S5, we present results based on the spatial
121: distribution of each spectral type of galaxies. In \S6 we interpret our
122: findings; a conclusion is given in \S7. Throughout this paper, we adopt the
123: concordance $\Lambda$CDM cosmology with $h_0=0.7$, $\Omega_{\lambda}=0.7$,
124: $\Omega_m = 0.3$. Magnitudes are quoted in the AB system.
125:
126: \section{E+A Galaxies}
127: One of the most promising approaches to understanding how blue and active
128: late-type galaxies are transformed into red and passive early-type systems
129: \citep{bowe04} is through studies of galaxies caught in transition between the
130: two types, for example E+A galaxies \citep{dres83}, also called k+a, a+k, or
131: post-starburst galaxies. Although there is no unique definition of this
132: transition type, its representatives are characterized by the presence of strong
133: Balmer absorption lines in their spectra, indicating a large fraction of A-type
134: stars, and weak or no ongoing star formation, often quantified by the strength
135: of the [OII] emission line. Since the duration of the E+A phase is relatively
136: short (less than 1 Gyr) E+A galaxies should be good tracers of the environment
137: that is host to a critical part of the evolution of galaxies in clusters.
138:
139: Although E+A galaxies were first found in clusters at intermediate redshift
140: (e.g., Dressler \& Gunn 1983, Couch \& Sharples 1987), their origin and habitat
141: was soon proven to be more varied by the discovery that the majority of E+A
142: galaxies in the local universe are in fact not located in dense environments
143: \citep{zabl96,goto03,goto07, hogg06, quin04}. As for the possible physical
144: origin of this type of galaxies, \citet{zabl96} and \citet{yang04} report
145: that many of the E+A galaxies in the field show evidence of galaxy mergers. At
146: higher redshift, however, E+A galaxies are predominantly found in the much
147: denser environment of clusters \citep{tran03,tran04}, which have thus been
148: suspected of being instrumental in the termination of star-formation activity in
149: these systems. In addition, \citet{pogg04} show that, at M$_v > -18.5$, the
150: post-starburst galaxies in Coma are fainter than the typical E+A galaxies found
151: in clusters at higher redshift, which suggests that the evolution of
152: post-starburst galaxies follows the "downsizing" trend proposed by
153: \citet{cowie96}. Any discrepancies between the properties of E+A galaxies in the
154: local field and in intermediate-redshift clusters may thus be explained by the
155: combination of evolution and selection biases, in the sense that the local
156: surveys sample predominantly luminous galaxies and hence do not probe the
157: population of fainter E+A galaxies in clusters (the faintest E+A galaxy in
158: \citet{zabl96} features M$_R \sim -18.5$, whereas the magnitude limit in
159: \citet{goto03} is M$_r < -21.84$).
160:
161: Unlike previous studies \citep{tran03, tran04, goto03}, the goal of this work is
162: not the issue of whether E+A galaxies are more likely to be found in clusters or
163: in the field; rather, we focus on the spatial distribution of E+A galaxies
164: around the massive cluster MACSJ0717.5+3745 to isolate the most probable
165: physical mechanism responsible for their creation in a cluster environment. In
166: this context we also attempt to address the question of the origin of the E+A
167: population by investigating the spatial distribution of possible star-burst
168: galaxies, the so-called e(a) and e(b) galaxies \citep{dres99,pogg00}.
169:
170:
171: \section{Data and Analysis}
172:
173: Using optical images obtained with the Suprime-Cam wide-field camera on the
174: Subaru 8m telescope on Mauna Kea \citep{miya02}, we identify galaxies within the
175: $34\times 27$ arcmin$^2$ Suprime-Cam field of view. We then employ color
176: selection to obtain spectra of a subset of presumed cluster members, primarily
177: using the DEep Imaging Multi-Object Spectrograph (DEIMOS) on the Keck-II 10m
178: telescope. In order to study the spectroscopic properties of galaxies in
179: different environments, the DEIMOS masks are designed to cover the entire
180: cluster including the filament and a satellite cluster. We use the spectroscopic
181: data to determine the redshift of galaxies, as well as their spectral types. We
182: follow the spectral classification scheme of \citet{tran03} (with some
183: modifications) to categorize the cluster galaxies into emission-line,
184: absorption-line, and E+A galaxies. The emission-line galaxies can be further
185: subdivided into e(a), e(b), and e(c) galaxies according to the equivalent width
186: of the H$_{\delta}$ absorption and [OII] emission lines
187: \citep{dres99}. Complementing our optical analysis of the galaxy population of
188: this system, X-ray data obtained with the imaging array of the Chandra Advanced
189: CCD Imaging Spectrometer (ACIS-I) are analyzed to provide a basic model of the
190: gaseous intracluster medium.
191:
192:
193: \subsection{Optical Photometry}
194: We use images in the B, V, R$_c$, I$_c$, and z$^\prime$ bands obtained with
195: Suprime-Cam, supplemented by images in the u$^*$ band obtained with the
196: MegaPrime camera on the CFHT 3.6m telescope. The observations were performed
197: from December 2000 to February 2004. All data are reduced employing standard
198: techniques, which have, however, been adapted to deal with the special
199: characteristics of the Suprime-Cam and MegaPrime data; details are given by
200: \citet{dono07}.
201:
202: In order to allow a robust estimate of the spectral energy distribution (SED) to
203: be obtained for all objects within the field of view, the imaging data from the
204: various passbands are seeing-matched using the technique described in
205: \citet{jeyh08}; this ensures that the images for all bands have the same
206: effective spatial resolution of 1$\arcsec$.
207:
208: Object catalogues are then created with SExtractor version 2.4.3\citep{bertin96}
209: in ``dual image'' mode with the R-band image as the reference detection
210: image. We separate stars from galaxies by fitting the stellar sequence in the
211: distributions of magnitude vs peak surface-brightness, and magnitude vs
212: half-light radius. Isophotal galaxy magnitudes calculated for elliptical
213: apertures are returned by SExtractor in the MAG\_AUTO parameter. The photometric
214: calibration of the Suprime-Cam imaging data is performed by means of a snapshot
215: observation of a nearby SDSS field\footnote{The SDSS photometry is transformed
216: into the Johnson-Cousin system using the equation of Lupton (2005) provided on
217: http://www.sdss.org/dr5/algorithms/sdssUBVRITransform.html.} and uses hundreds
218: of stars with magnitudes ranging from 16 mag to 18 mag. The calibration of our
219: MegaCam data is supplied by CFHT's Elixir
220: project\footnote{http://www.cfht.hawaii.edu/Instruments/Elixir/}.
221:
222: \subsection{X-ray Data}
223: The X-ray data are used to derive global cluster properties, most importantly
224: the virial radius and the ram-pressure stripping radius. To this end, we
225: analyze the data taken with the Chandra ACIS-I instrument in January 2003 with a
226: total exposure time of 60 ks (ObsID 4200). The data reduction is performed
227: following the procedure described by \citet{rude05}. Fig.~\ref{xrayim} shows the
228: adaptively smoothed X-ray emission in the 0.5--7 keV band as observed by ACIS-I.
229: %%%
230: The X-ray luminosity and temperature are listed in Table~\ref{xrayprop}; a
231: dynamical analysis based on the X-ray properties and the velocity dispersion
232: will be presented in \S5.3.
233:
234:
235: \begin{figure}[h]
236: \epsscale{1.0}
237: \plotone{f1.eps}
238: \figcaption[paper_xray.ps]{X-ray
239: emission in the field of MACSJ0717.5+3745 as observed with Chandra's ACIS-I
240: detector in the 0.5--7 keV band. The shown image has been adaptively smoothed
241: using the \textit{asmooth} algorithm \citep{ebel06} requiring a minimal
242: significance with respect to the local background of 99\%. The shaded cross
243: marks the regions falling onto the chip gaps of the ACIS-I detector. A
244: close-up view of the cluster core based on the same X-ray data has been
245: previously published by \citet{ebel07}. [The full resolution figure will be published in ApJ] \label{xrayim}}
246: \end{figure}
247:
248: \subsection{Optical Spectroscopy}
249:
250: MACSJ0717.5+3745 has the most extensive spectroscopic data in our on-going
251: follow-up study of the dynamical structure and galaxy properties in 12 clusters
252: with $z>0.5$ in the MACS sample \citep{ebel07,jeyh08}. The spectroscopic data
253: set used here is compiled primarily from the observation of 18 masks with the
254: DEIMOS spectrograph on the Keck-II telescope. In addition, some spectra obtained
255: using the Low Resolution Image Spectrometer (LRIS) on the Keck-I telescope, and
256: the Gemini Multi-Object Spectrographs (GMOS) on the Gemini telescope are
257: included. The instrumental configuration of these observations is summarized by
258: \citet{barrett05}, from where we have extracted the essential information listed
259: in Table~\ref{spectdata}. The chosen setup is a compromise between the
260: requirements of wavelength coverage from the [OII] $\lambda\lambda 3727$ line to
261: the H$_{\beta}$ line at a redshift of about 0.5, and moderately high spectral
262: resolution. After reducing the spectra using the standard DEIMOS pipeline
263: developed by the DEEP2 team, the redshifts are determined and verified manually
264: using at least two prominent spectral features, such as (in absorption) Calcium
265: H and K, H$_{\delta}$, or the G band, and (in emission) [OII] $\lambda\lambda
266: 3727$, H$_{\beta}$, and [OIII] $\lambda\lambda 4959,5007$. Based on repeated
267: observations of individual objects, we estimate the error in the resulting
268: redshifts to be about 70 km/s, which is consistent with the spectral resolution
269: of the observations.
270:
271: The final spectroscopic catalogue comprises 1103 galaxies, of which 563 are
272: cluster members (Table~\ref{spectsum}). Fig.~\ref{plain} shows the location of
273: the spectroscopically observed objects within our study region which is
274: highlighted by the outline of the DEIMOS masks. The full redshift distribution
275: is shown in Fig.~\ref{redshiftplt}. Cluster membership is assigned to all
276: galaxies with $0.522<z<0.566$, the redshift range given by $z=z_{cl}\pm
277: %%%
278: 2.5\sigma$, where the cluster redshift z$_{cl}=0.5446$ and velocity dispersion
279: $\sigma=1612$ km s$^{-1}$ (Table~\ref{xrayprop}) are determined from the
280: redshift distribution of galaxies within the virial radius of the
281: cluster\citep{ebel07}. The final catalogue, including redshifts and equivalent
282: widths of the spectral features ([OII], H$_{\delta}$, H$_{\gamma}$, D$_n$(4000),
283: H$_{\beta}$, and [OIII]), will be published in a separate data paper.
284:
285:
286: %Table 2 : spectroscopic log
287: \begin{deluxetable}{lccccc}
288: \tabletypesize{\scriptsize}
289: \tablewidth{0pc}
290: \tablecolumns{6}
291: \tablecaption{Summary of spectroscopic observations \label{spectdata}}
292: \tablehead{
293: \colhead{Instrument} & \colhead{Date} & \colhead{Grating} & \colhead{$\lambda_c$} & \colhead{Filter} &\colhead{Total Exposure Time} \\
294: \colhead{} & \colhead{} & \colhead{} & \colhead{(\AA)} & \colhead{} &\colhead{(sec)}
295: }
296: \startdata
297: LRIS\tablenotemark{a} & 2000-11-20 & 600/7500 & 6200 & GC495 & 3600 \\
298: LRIS\tablenotemark{a} & 2000-11-20 & 600/7500 & 6200 & GC495 & 4800 \\
299: LRIS\tablenotemark{a} & 2000-11-21 & 600/7500 & 6200 & GC495 & 3600 \\
300: LRIS\tablenotemark{a} & 2002-11-29 & 600/7500 & 6500 & GC495 & 5400 \\
301: LRIS\tablenotemark{a} & 2002-11-29 & 600/7500 & 6500 & GC495 & 6000 \\
302: GMOS-N\tablenotemark{a} & 2004-03-12 & B600 & 6700 & GC455 & 5400 \\
303: DEIMOS\tablenotemark{a} & 2003-12-23 & 600ZD & 6500 & GC455 & 7200 \\
304: DEIMOS\tablenotemark{a} & 2004-12-16 & 600ZD & 7000 & GC455 & 5400\\
305: DEIMOS\tablenotemark{a} & 2004-12-16 & 600ZD & 7000 & GC455 & 5400\\
306: DEIMOS\tablenotemark{a} & 2004-12-16 & 600ZD & 7000 & GC455 & 5400\\
307: DEIMOS\tablenotemark{a} & 2005-02-12 & 600ZD & 7000 & GC455 & 3239 \\
308: DEIMOS\tablenotemark{a} & 2005-02-12 & 600ZD & 7000 & GC455 & 6900 \\
309: \cline{1-6} \vspace*{-2mm} \\
310: DEIMOS & 2006-12-22 & 600ZD & 6300 & GC455 & 5400 \\
311: DEIMOS & 2006-01-31 & 600ZD & 6300 & GC455 & 6047 \\
312: DEIMOS & 2006-01-31 & 600ZD & 6300 & GC455 & 5400 \\
313: DEIMOS & 2006-01-31 & 600ZD & 6300 & GC455 & 5400 \\
314: DEIMOS & 2006-01-31 & 600ZD & 6300 & GC455 & 4991 \\
315: DEIMOS & 2006-02-01 & 600ZD & 6300 & GC455 & 5400 \\
316: DEIMOS & 2006-02-01 & 600ZD & 6300 & GC455 & 3600 \\
317: DEIMOS & 2006-02-01 & 600ZD & 6300 & GC455 & 5400 \\
318: DEIMOS & 2006-02-01 & 600ZD & 6300 & GC455 & 3600 \\
319: DEIMOS & 2008-01-05 & 600ZD & 6300 & GC455 & 5400 \\
320: DEIMOS & 2008-01-06 & 600ZD & 6300 & GC455 & 5400 \\
321: DEIMOS & 2008-01-07 & 600ZD & 6300 & GC455 & 5400
322: \enddata
323:
324: \tablenotetext{a} {Reference: \citet{barrett05}}
325:
326: \end{deluxetable}
327:
328:
329:
330:
331: %\onecolumn
332: \begin{figure*}[ht!]
333: \plotone{f2.eps}
334: \figcaption[f2.eps]{Spatial distribution of galaxies
335: observed spectroscopically: small dots mark the location of all sources with
336: measured spectrum, large dots mark confirmed cluster members. The overlaid
337: contours show the projected galaxy density as computed by
338: \textit{asmooth}\citep{ebel06}. An image of the distribution of all galaxies
339: in the photometric redshift catalogue (\S5.1) is adaptively smoothed such that
340: all features in the smoothed image are $3\sigma$ significant with respect to
341: the local background, where feasible. The plotted contours are located at
342: $17$, $35$, $71$, $142$, and $285$ Mpc$^{-2}$, respectively; they are truncated
343: where the significance of the signal is less than $3\sigma$. The blue boundary
344: indicates the outline of all DEIMOS masks and defines the study region for
345: this paper. All areas observed with GMOS and LRIS are covered by the DEIMOS
346: masks. The two red circles (1.6 Mpc radius) indicate two regions of particular
347: interest which will be discussed in more detail in \S5.4 and \S5.6: (a) the
348: center of the cluster (dashed), and (b) the endpoint of the filament
349: (dash-dotted). Both of these regions contain a local peak of the galaxy
350: density. The green line crudely describes the large-scale orientation of the
351: overall system and will be used in \S5.3 to assess how galaxy properties vary
352: as a function of environment.[The full resolution figure will be published in ApJ]\label{plain}}
353: \end{figure*}
354: %\twocolumn
355:
356:
357: \begin{figure}[ht]
358: \epsscale{1.0} \plotone{f3.eps} \figcaption[f3.eps]{Redshift
359: distribution for all spectroscopically observed galaxies. The systemic
360: redshift of MACSJ0717.5+3745 is found to be $z_{\rm cl}=0.545$. The dashed
361: lines at $z=0.522$ and $z=0.566$ ($z=z_{\rm cl}\pm3\sigma$) mark the
362: spectroscopic extent of the cluster; the black area highlights the cluster
363: members. A zoomed-in view of the redshift range near the cluster redshift is
364: shown in the sub-panel.\label{redshiftplt}}
365: \end{figure}
366:
367:
368:
369: \section{Spectroscopic Analysis}
370: \subsection{Object Selection and Completeness}
371:
372: Since it is impractical to cover the entire sample of galaxies spectroscopically
373: from the cluster core to the outskirt, we had to select targets in a way that
374: increases the efficiency of cluster member detections while minimizing any
375: potential bias. Since, at the time this project started, we did not have imaging
376: data in a sufficient number of bandpasses to determine credible photometric
377: redshifts, the target list for our spectroscopic observations is selected using
378: the color magnitude diagram for the V and R$_c$ filters (Fig.~\ref{cmd}), making
379: use of the cluster red sequence. To maximize completeness, we limit our survey
380: to relatively bright objects ($m_{\rm R_c}\leq 22.25$, which is, equivalently,
381: M$_{\rm Rc} \leq $M$^*$+2 at z=$0.55$) while adopting a generous color cut
382: toward the blue end of the distribution. The latter was adjusted in the course
383: of the survey such that its locus coincides with the V--R$_c$ color (relative to
384: the red sequence) at which the fraction of galaxies found to be cluster members
385: falls below 20\% (Fig. ~\ref{membership}). However, there will be unavoidably a
386: few extremely blue galaxies that are in fact cluster members but do not satisfy
387: our selection criteria. We estimate the resulting incompleteness by fitting a
388: heuristic Gaussian model to the data shown in Fig.~\ref{membership}. If we
389: assume that the probability derived from the observed sample is representative,
390: and that the Gaussian fit can be extrapolated to yet bluer colors, we would
391: expect our survey to be incomplete at the $\sim 2\%$ level. We note, however,
392: that this incompleteness is severely color-dependent which, as we will show in
393: \S5.2, could potentially cause the fraction of emission-line galaxies, but not
394: the fraction of E+A galaxies, to be significantly underestimated.
395:
396: We define the completeness of our spectroscopic survey as the ratio of the
397: number of sources observed to the number of sources in the photometric catalogue
398: (within the color band marked in Fig.~\ref{cmd}), including spectra from which we
399: failed to measure a redshift\footnote{We include in this statistic a small
400: number of stars which failed to be eliminated by the star-galaxy separation
401: discussed in \S3.1}. In addition we define the efficiency of our survey as the
402: ratio of the number of high-quality spectra, from which both the redshift and
403: the equivalent width of spectral features can be measured accurately (again
404: within the color limits marked in Fig.~\ref{cmd}). The completeness and
405: efficiency as a function of galaxy magnitude are shown in the top panel of
406: Fig.~\ref{completeness}. Inside the entire study region, the completeness is
407: roughly constant at about $70\%$ at R$_c< 21.75$, with the curve only dropping
408: within the very last bin before the magnitude limit at R$_c=22.25$. Also shown
409: in the top panel of Fig.~\ref{completeness} is the completeness within a radius
410: of 2 Mpc from the center of the cluster: again
411: the completeness is roughly independent of magnitude (except for the final bin
412: before our global magnitude limit) but now at a much higher level near unity.
413: Although the difference between these two curves shows that the completeness of
414: our survey is not spatially uniform, the relative spatial distribution of
415: galaxies of different types should be unaffected.
416:
417: The bottom panel in Fig.~\ref{completeness} shows the efficiency in determining
418: the galaxy redshift and the efficiency for spectroscopic classification. The
419: former is almost perfect over the full magnitude range. The criteria used to
420: classify galaxies according to spectral type (defined in the following two
421: sections) are much stricter; still, the efficiency for spectroscopic
422: classification remains better than $0.8$ at all magnitudes, reflecting the good
423: signal-to-noise ratio of most of our spectra. Only a handful of spectra feature
424: very low signal-to-noise ratios which are caused either by poor slit positioning
425: or by erroneously high magnitudes reported by SExtractor. If the redshift can be
426: measured but no spectral type can be determined, the reasons are usually that
427: the key spectral features fall onto the chip gap or coincide with telluric
428: lines.
429:
430: %%%
431: \begin{figure}[ht]
432: \epsscale{1.0} \plotone{f4.eps} \figcaption[f4.eps]{Color magnitude diagram
433: for galaxies inside our study region as defined in Fig~\ref{plain}. Blue
434: filled circles mark objects with observed spectra; red filled circles denote
435: spectroscopically confirmed cluster members (cf.\
436: Fig.~\ref{redshiftplt}). Black dots are all remaining objects in our
437: SExtractor galaxy catalogue. The dashed lines indicate the selection
438: criteria used for our spectroscopic survey. The red sequence marked by the
439: solid line is obtained from a linear fit to the data points redder than
440: V-R$_{c}\sim 0.9$ and brighter than R$_{c}\sim 22.5$. [The full resolution figure will be published in ApJ]\label{cmd}}
441: \end{figure}
442:
443: \begin{figure}[h]
444: \epsscale{1.0} \plotone{f5.eps}
445: \figcaption[f5.eps]{The probability of being a cluster member for
446: all galaxies within our study region and brighter than our magnitude cut as
447: estimated by the ratio between the number of cluster members and the number of
448: galaxies with spectroscopic redshift. The x-axis is the relative V-R$_{c}$
449: color with respect to the color of the red sequence, which is magnitude
450: dependent (Fig.~\ref{cmd}). The fitted function (we use an ad-hoc Gaussian
451: model) allows us to estimate the number of cluster members missed by our
452: survey because of their extremely blue color. \label{membership}}
453: \end{figure}
454:
455:
456: \begin{figure}[h]
457: \epsscale{1.0} \plotone{f6.eps}
458: \figcaption[f6.eps]{Completeness (top) and efficiency (bottom)
459: of our spectroscopic survey as a function of galaxy magnitude. The color
460: criterion of Fig.~\ref{cmd} has been applied. Top: survey completeness for the
461: entire study region (filled symbols) and for Region A (open symbols). Bottom:
462: survey efficiency, i.e. success rate for redshift determination (open symbols)
463: and spectral classification (filled symbols). In both panels, the dashed line
464: indicates the magnitude limit of the spectroscopic
465: sample. \label{completeness}}
466: \end{figure}
467:
468:
469: \subsection{Equivalent Width Measurement:}
470:
471: In order to allow a comparison of our results with those from previous
472: spectroscopic studies of cluster galaxies we follow the approach of
473: \citet{tran03} to measure equivalent widths (EW), i.e.\ the inverse-variance
474: weighted integration of the flux over a spectral line, using integration windows
475: as defined in \citet{fish98}. EW of three spectral features are measured this
476: way: [OII] $\lambda\lambda 3727$, H$_{\delta}$, and H$_{\gamma}$. For those
477: galaxies with strong emission features but weak or non-detectable continuum, a
478: lower limit to the EW is calculated based on the variance of the spectrum near
479: the spectral feature in question. An exception is the case of the H$_{\beta}$
480: emission line for which the EW could be underestimated if absorption is present
481: too. We account for this potential bias by subtracting the EW of the
482: H$_{\beta}$ absorption as provided by stellar population models \citep{bc03}
483: fitted to the data from 4060\AA\ to 5360\AA\ (rest frame) using the
484: template-fitting code \textit{pPXF} \citep{capp04}. We adopt the convention that
485: negative (positive) EW values indicate emission (absorption). Hereafter, we will
486: use the convention that $[OII]$ refers to the EW of [OII] in emission, whereas
487: $H_{\delta}$ and $H_{\gamma}$ refer to the EW of the respective lines in
488: absorption. Since we have fitted the emission and absorption components of
489: H$_{\beta}$ separately, the EW of H$_{\beta}$ in absorption and emission will be
490: specified as $H_{\beta,ab}$ and $H_{\beta,em}$.
491:
492: \subsection{Spectral Type Classification:}
493:
494: We classify cluster galaxies as emission-line galaxies, absorption-line
495: galaxies, and E+A galaxies, following the spirit of the classification used by
496: \citet{tran03} with a few modifications\footnote{Our goal is to classify
497: galaxies according to activity, as reflected in their spectroscopic
498: features. Unfortunately, a universally accepted system for such a
499: classification has yet to be established (see the review in
500: \citet{goto06}). This is especially true for E+A galaxies: since the exact
501: locus of the dividing line between E+A galaxies and both emission- and
502: absorption-line galaxies is ill-defined, the final classification in this grey
503: area remains somewhat uncertain.}. The classification scheme of \citet{tran03}
504: only requires the relatively blue features [OII], H$_{\delta}$, and
505: H$_{\gamma}$, which is convenient for spectroscopic surveys of clusters out to
506: redshift $z\approx 1$ for which redder spectral lines are observationally
507: inaccessible in the optical passband. [OII] emission is an indicator of current
508: activity (be it star formation or the presence of an active nucleus), whereas
509: the two Balmer line features\footnote{It is worth pointing out in this context
510: that historically the classification of E+A galaxies is based on the
511: absorption features H$_{\delta}$ and H$_{\gamma}$, without any attempts at
512: quantifying, or accounting for, emission components in these lines.} are
513: indicators of star formation within the last few hundred million years. The
514: Balmer emission lines, however, are more robust diagnostics of star-formation
515: activity (e.g. \citet{mous06}) than [OII]. Since the latter is easily
516: underestimated because of extinction, the absence of [OII] emission does not
517: necessarily preclude the presence of star formation or other galaxy activity.
518: Although H$_{\alpha}$ is not accessible to optical surveys for galaxies at
519: $z>0.5$, the H$_{\beta}$ line of most of the galaxies in our target cluster
520: falls within the observable spectral range (although it comes dangerously close
521: to the broad atmospheric H$_2$O absorption band at 7600\AA). We therefore use
522: both $[OII]$ and $H_{\beta,em}$ to define emission-line galaxies, the specific
523: criteria being: $[OII] < -5 $\AA\ (the same cutoff as \citet{tran03}) or
524: $H_{\beta,em} < -5 $\AA. Galaxies with $[OII] > -5 $\AA\ and $H_{\beta,em} > -5
525: $\AA\ will be classified as absorption-line galaxies if $\frac{(H_{\delta} +
526: H_{\gamma})}{2} < 4$\AA\ (the latter criterion being identical to the one used
527: by \citet{tran03}).
528:
529: As for E+A galaxies, we deviate yet further from \citet{tran03} who simply
530: define E+A galaxies to be all galaxies that are neither emission- or
531: absorption-line galaxies. The definition adopted for this work is more
532: conservative in that we require an E+A galaxy to have {\em no} detectable
533: $[OII]$ or $H_{{\beta},em}$ and $\frac{( H_{\delta}+ H_{\gamma})}{2}>$6\AA\ (as
534: compared to the cutoff value of 4\AA\ of \citet{tran03}). This definition leaves
535: a small fraction of our galaxy sample unclassified, in recognition of the
536: aforementioned gray areas between E+A galaxies and both emission- and
537: absorption-line galaxies.
538:
539: The motivation for our choice of more stringent criteria is twofold. As shown by
540: \citet{goto03} and \citet{goto07}, local samples of E+A galaxies compiled
541: without information about H$_{\alpha}$ may suffer from as much as $52\%$
542: contamination by H$_{\alpha}$ emission-line galaxies. \citet{blak04}, however,
543: found that a linear combination of the strengths of three Balmer absorption
544: lines (H$_{\beta}$, H$_{\delta}$, and H$_{\gamma}$) can be efficiently used to
545: identify this contaminating population. Following \citet{zabl96} we thus
546: calculate the average of $H_{\delta}$, $H_{\gamma}$, and $H_{\beta,ab}$ for all
547: galaxies in our sample that meet the more generous definition of E+A galaxies of
548: \citet{tran03}, and for which the H$_{\beta}$ line is observable. We find
549: that only $45\%$ of them meet the criterion that $\frac{(H_{{\beta},ab}+
550: H_{\gamma}+ H_{\delta})}{3}>5.5$\AA\ \citep{blak04}, implying that more
551: than half of the population thus selected may in fact feature H$_{\alpha}$ in
552: emission. If, however, we raise the threshold value for $\frac{(
553: H_{\gamma}+H_{\delta})}{2}$ from 4 to 6\AA, then $100\%$ of the galaxies
554: thus selected, and for which the H$_{\beta}$ line is available, fulfill the
555: criteria of \citet{blak04}. The second motivation for our more stringent
556: definition of E+A galaxies is that the detection of [OII] in a galaxy is not an
557: unambiguous sign of star formation but could also be indicative of non-stellar
558: radiation. Indeed, \citet{yan06} find that a large fraction of post-starburst
559: galaxies may be misidentified as undergoing current star formation, although the
560: [OII] emission actually originates from AGN. Since our project has currently no
561: means of efficiently distinguishing between star-bursting galaxies and
562: post-starburst galaxies containing AGN, and since the compilation of a
563: statistically complete sample of post-starburst galaxies is not the main goal of
564: this work, we exclude from our E+A sample all galaxies featuring {\em any}\/
565: [OII] emission, in order to keep the resulting sample as clean as possible and
566: avoid complications caused by the presence of AGN for the interpretation of the
567: mechanisms responsible for the termination of star formation activity. Although
568: the sample of E+A galaxies selected by our more conservative criteria thus
569: misses a possibly sizeable and interesting fraction of the entire post-starburst
570: population as suggested in \citet{yan06}, it is better suited to study the
571: interactions between galaxies and the intra-cluster medium, and it is also less
572: contaminated by weakly star-forming galaxies.
573:
574: Following \citet{dres99}, we can classify the emission-line galaxies further
575: into e(a), e(b), and e(c) sub-types based on the strength of the $H_{\delta}$
576: absorption and $[OII]$ emission lines. Both of these features are indicative of
577: starbursts within a few million years to one billion years before the time of
578: observation. Specifically, e(a) galaxies are defined as systems with $H_{\delta}
579: > 4$\AA\ and $-40$\AA$<[OII]<-5$\AA, whereas e(b) galaxies feature
580: $[OII]<-40$\AA, regardless of $H_{\delta}$; the spectral evidence of young and
581: massive stars in either of these subtypes suggests recent starbursts. All other
582: emission-line galaxies, i.e.\ systems lacking signs of explosive star formation
583: in the recent past are classified as e(c). The difference between e(b), e(a),
584: and E+A galaxies is thus the decreasing strength of the [OII] emission line.
585: The absence of [OII] in E+A galaxies suggests that the star formation activity
586: is halted, whereas, at the opposite extreme, strong [OII] emission makes the
587: e(b) sub-type promising candidates of starburst galaxies. The, in terms of [OII]
588: strength, intermediate e(a) galaxies, on the other hand, are not easily placed
589: in this sequence since, as mentioned before, the strength of [OII] emission is
590: highly sensitive to dust extinction. \citet{pogg00}, for instance, find about
591: half of a sample of very luminous infrared galaxies
592: [log(L$_{IR}$/L$_{\sun}$)$>$11.5] to exhibit e(a) spectra. The distinction
593: between these three populations is thus ambiguous; some of the E+A galaxies may
594: actually be e(a) galaxies whose [OII] emission is completely extinguished by
595: dust. Conversely, some of the e(a) galaxies could in fact be post-starburst
596: galaxies with some residual star formation. In spite of these ambiguities, we
597: emphasize that e(a) and E+A galaxies are in general distinct populations, which
598: is -- as we shall demonstrate -- reflected in their different distribution
599: within the cluster environment. The relevance of e(a) and e(b) galaxies in the
600: context of this study is the potential for both galaxy types to trace the E+A
601: progenitor population.
602:
603: For clarity, the definition of the spectral types is summarized in
604: Table~\ref{spectdef}. The number of galaxies of each type is given in
605: Table~\ref{spectsum}. Note that, because of our conservative definition of E+A
606: galaxies, about 8\% of all cluster members with spectroscopic information are
607: missing in Table~\ref{spectsum} as well as in the analysis that follows. They
608: are the population with weak [OII], weak H$_{\beta}$, and Balmer absorption
609: lines whose strength falls between the absorption-line and E+A criteria
610: (4\AA$<\frac{(H_{\delta}+H_{\gamma})}{2}<$6\AA). They are probably a mixture of
611: dusty star-forming galaxies and post-star-forming galaxies. In \citet{tran03}
612: and \citet{tran04}, galaxies of this type are also classified as E+A galaxies.
613:
614:
615: %Table 3: summary of definition
616: \begin{deluxetable*}{cl}
617: \tabletypesize{\scriptsize}
618: \tablewidth{0pc}
619: tablecolumns{2}
620: \tablecaption{Definition of spectral types\label{spectdef}}
621: \tablehead{
622: \colhead{Type} & \colhead{Criteria} }
623: \startdata
624: Emission-line & $[OII] < -5 $\AA\ or $H_{\beta,em} < -5 $\AA\ \\
625: Absorption-line & $[OII] > -5$\AA\, $H_{\beta,em} > -5 $\AA\, and $\frac{(H_{\delta} +H_{\gamma})}{2} < 4$\AA\ \\
626: E+A & no detection of $[OII]$ and $H_{\beta,em}$, and $\frac{(H_{\delta} + H_{\gamma})}{2} > 6$\AA\ \\
627: \cline{1-2}\vspace*{-2mm} \\
628: e(a) & Emission-line galaxies with $H_{\delta}> 4$\AA\, and $-40$\AA$<[OII]<-5$\AA \\
629: e(b) & Emission-line galaxies with $[OII]<-40$\AA \\
630: e(c) & Emission-line galaxies with $H_{\delta} < 4$\AA\, and $-40$\AA$<[OII]<-5$\AA
631: \enddata
632: \end{deluxetable*}
633:
634:
635: %Table 4 : spectral type summary
636: \begin{deluxetable}{ccc}
637: \tabletypesize{\scriptsize}
638: \tablewidth{0pc}
639: \tablecolumns{3}
640: \tablecaption{Spectral types summary\label{spectsum}}
641: \tablehead{
642: \colhead{Type} & \colhead{All} & \colhead{cluster members} }
643: \startdata
644:
645: all spectra\tablenotemark{a} & 1368 & \nodata \\
646: sample spectra & 1147 & \nodata \\
647: redshift measurements & 1103 & 563 \\
648: spectra with spectral diagnostics & 1023 & 530 \\
649: \cline{1-3} \vspace*{-2mm} \\
650: Emission-line galaxies & 507 & 203 \\
651: Absorption-line galaxies & 426 & 260 \\
652: E+A galaxies & 21 & 17 \\
653: Unclassified & 69 & 50 \\
654: \cline{1-3} \vspace*{-2mm} \\
655: e(a) & 76 & 42 \\
656: e(b) & 66 & 27 \\
657: e(c) & 365& 134
658: \enddata
659: \tablenotetext{a}{No magnitude or color cuts are applied. Spectra obtained serendipitously are included.}
660: \end{deluxetable}
661:
662:
663: \section{Result:}
664:
665: \subsection{Photometric Redshift}
666: From our photometry in the u*, B, V, R$_c$, I$_c$, and z$^\prime$ bands,
667: photometric redshifts are derived for galaxies with R$_c<$24.0 ($\sim$~M$^*$+4
668: at z$=$0.55) using the adaptive SED-fitting code \textit{Le Phare}
669: \citep{arno99,ilbe06}. \textit{Le Phare} employs ${\chi}^2$ optimization,
670: comparing the observed magnitudes with those predicted from an SED library. An
671: adaptive method is applied to adjust the photometric zero-points for all
672: passbands by using the sample of spectroscopic redshifts as a training
673: set\footnote{The resulting correction of typically 0.05 mag is consistent with
674: the statistical and systematic uncertainty of our photometric
675: calibration.}. No Bayesian prior for the galaxy redshift distribution is
676: assumed. We use the library of empirical templates of \citet{ilbe06} and the
677: Calzetti extinction law \citep{calz01} for reddening corrections applied to
678: templates later than Sbc. The training set is obtained from the spectroscopic
679: redshift database by selecting all galaxies with an R$_c$-band magnitude
680: brighter than 21.5 and a minimal separation from the closest SExtractor-detected
681: neighbor of $3.5 \arcsec$ to avoid photometric errors from close pairs or
682: blended sources. The resulting training set comprises 273 galaxies at redshifts
683: ranging from 0.15 to 0.9. The comparison of spectroscopic and photometric
684: redshifts is shown in Fig.~\ref{pzVSspz}. The statistical error of the
685: photometric redshifts, as derived from the Gaussian distribution of the
686: residuals shown in the insert of Fig.~\ref{pzVSspz}, is found to be $\Delta z =
687: 0.024$.
688:
689: The main purpose of the photometric redshift catalogue is to allow us to
690: estimate the local projected galaxy density with better statistics than would be
691: feasible from the spectroscopic data set alone. On the basis of photometric
692: redshifts, we define cluster members to be the galaxies with $| z_{ph} - z_{cl}
693: | < \sigma_{ph-z}$, where $\sigma_{ph-z} = (1+z_{cl}) \Delta z$. Note that
694: this redshift cut ($\pm \sigma_{ph-z} = \pm0.036$) is much more generous than
695: that used before for the spectroscopic redshifts ($\pm3\sigma = \pm 0.022$); the
696: chosen cutoff value represents a good compromise between completeness and
697: contamination. The resulting catalogue of ``photometric'' cluster members
698: includes 2218 galaxies and allows us to estimate the local galaxy density in the
699: entire Suprime-Cam field of view. The much higher accuracy of our spectroscopic
700: redshifts, on the other hand, provide us with an opportunity to study the
701: dynamical structure of the cluster-filament system in detail.
702:
703:
704: Using the photometrically selected set of cluster members, we map the surface
705: density of cluster galaxies. A smoothed version of this distribution (shown in
706: Fig.~\ref{plain}) is obtained with the adaptive-smoothing algorithm
707: \textit{asmooth} \citep{ebel06} which preserves the signal-to-noise ratio of all
708: significant features on all scales. The comparison of the result with the maps
709: derived by \citet{ebel04} and \citet{jeyh08} using only galaxies on the cluster
710: red sequence proves instructive in as much as it highlights the dominance of
711: blue galaxies in low-density, filamentary environments. We show in
712: Fig.~\ref{galdenmap}, side by side, the galaxy surface density maps obtained by
713: us using photometric redshifts as described above, and the equivalent map from
714: \citet{jeyh08} which is based entirely on galaxies with V--R$_c$ colors
715: consistent with the cluster red sequence. Note the dramatic difference in the
716: relative prominence of the main cluster and the filament.
717:
718: \begin{figure}[h]
719: \plotone{f7.eps}
720: \figcaption[f7a.eps]{Surface density of cluster members using
721: either photometric redshifts (top) or selection by red-sequence color (bottom)
722: \citep{jeyh08}. The contour levels are given by $1.2^n \Sigma_{bkg}$, where
723: $\Sigma_{bkg}$ is the background surface density of each plot. The bold,
724: ragged contour marks the boundary of regions outside of which the significance
725: of any structure in these adaptively smoothed images falls below the 3$\sigma$
726: confidence level. [The full resolution figure will be published in ApJ] \label{galdenmap}}
727: \end{figure}
728:
729:
730: In order to allow a direct comparison with the results from earlier work on the
731: impact of cluster environment on galaxy activity \citep{koda01, balogh04, sato06},
732: we also calculate for all ``photometric'' cluster members the projected local
733: density $\Sigma_{10}$, defined as the galaxy number density within a circle
734: whose radius is given by the separation to the $10^{th}$ closest neighbor. The
735: resulting values of $\Sigma_{10}$ are consistent with the surface density
736: distribution shown in the top panel of Fig.~\ref{galdenmap}.
737:
738:
739:
740: \begin{figure}[h]
741: %\epsscale{1.0}
742: \epsscale{0.8}
743: \plotone{f8.eps}
744: \figcaption[f8.eps]{Photometric versus spectroscopic redshifts for
745: galaxies in the field of MACSJ0717.5+3745; we only show galaxies from our
746: \textit{Le Phare} training set (see text for details). The lower panel shows
747: the residuals d$z$ as a function of the spectroscopic redshift, while the
748: insert in the top left corner shows the histogram of the residuals as well as
749: a Gaussian model used to estimate their standard deviation $\Delta
750: z$. [The full resolution figure will be published in ApJ]\label{pzVSspz}}
751: \end{figure}
752:
753: \subsection{Broad-Band Colors of Cluster Members of Different Spectral Types}
754: The comparison of Fig.~\ref{cmd} with the color-magnitude diagram for different
755: spectral types of galaxies (left panel of Fig.~\ref{cc}) illustrates the
756: potential bias caused by the blue color cut discussed in \S4.1. Not
757: surprisingly, all of the bluest cluster members found in our study are
758: emission-line galaxies, the fraction of which is thus likely to be
759: underestimated. This is particularly true for the e(b) subtype which we find to be
760: heavily concentrated at the blue, faint end of our selection in color-magnitude
761: space. If all cluster members bluer than our color cut are in fact e(b) galaxies
762: -- clearly the worst-case scenario --, the number of e(b) galaxies in
763: MACSJ0717.5+3745 could be underestimated by as much as $\sim$ 30\%. We shall keep
764: this in mind when, later, discussing the fraction of e(b), e(a), and
765: emission-line galaxies in general. Note, however, that the E+A and
766: absorption-line galaxies lie well away from the color cut, so that their numbers
767: can be expected to be unaffected by color-dependent incompleteness.
768:
769: Figure~\ref{cc} compares the magnitude and color distributions of cluster
770: members of different spectral type. As has been noted before by other authors,
771: for example \citet{dres83}, \citet{tran03}, and \citet{tran07}, E+A galaxies are
772: found to exhibit broad-band colors that are intermediate between those of
773: absorption- and emission-line galaxies, as is to be expected from their spectral
774: definition as a
775: %%%add some comparison of the mean and scale
776: composite of old and new stellar populations. Differences between the various
777: spectroscopic types can be characterized by measuring the mean and scatter of
778: the relative V-R$_{c}$ color of a given galaxy class with respect to the red
779: sequence (Table~\ref{colorstats}). The mean relative color of zero found for
780: absorption-line galaxies confirms our linear fit to the cluster red sequence
781: calculated without selection of any specific spectroscopic types
782: (Fig.~\ref{cmd}). The small scatter in the relative V-R$_{c}$ color of
783: absorption-line galaxies is consistent with the result of, e.g.,
784: \citet{blakeslee06} and \citet{tran07} that early-type cluster members are
785: formed at $z\sim 2$. The mean relative colors of the other spectroscopically
786: defined galaxy types discussed in this study are, in order of increasing
787: blueward distance from the red sequence, E+A galaxies, e(a), and e(b) galaxies,
788: a ranking consistent with the hypothesis that E+A galaxies represent an
789: intermediate stage in the evolutionary sequence from active emission-line
790: galaxies to passively evolving early-type galaxies. Interestingly, this scenario
791: is also supported by the different scatter around the mean relative color for
792: the different populations: the color scatter of the E+A galaxies is
793: significantly smaller than the one of e(a) and e(b) galaxies, their potential
794: precursors \citep{pogg99}, but still larger than the one of the evolved
795: population of absorption-line galaxies forming the cluster red sequence.
796: Although their spectral properties imply that E+A galaxies evolved from the e(a)
797: and/or e(b) population less than one billion years ago, their color
798: characteristics have thus already become remarkably similar to those of
799: absorption-line galaxies.
800:
801: Another interesting fact that emerges from Fig.~\ref{cc} relates to the mean
802: color of e(b) galaxies. Their locus at the extreme blue end of the distribution
803: shown in the left bottom panel of Fig.~\ref{cc} is not solely the result of the
804: strong [OII] emission to which they owe their definition. Since [OII] falls into
805: the V band at $z=0.55$, the fact that the majority of the e(b) galaxies found in
806: our study also exhibits very blue U-V colors (right bottom panel of
807: Fig.~\ref{cc}) indicates significant continuum emission in the UV rest frame.
808: Follow-up observations are needed of the few e(b) galaxies that are found to be
809: unusually red to establish whether nuclear activity is responsible for the
810: strong [OII] emission observed.
811:
812:
813: %%%new table
814: \begin{deluxetable}{lcc}
815: \tabletypesize{\scriptsize}
816: \tablewidth{0pc}
817: \tablecolumns{3}
818: \tablecaption{Relative color statistics of spectroscopic types\label{colorstats}}
819: \tablehead{
820: \colhead{} & \colhead{mean\tablenotemark{a}}& \colhead{scale\tablenotemark{a}}
821: }
822: \startdata
823:
824: Absorption-line galaxies & $0.00^{+0.005}_{-0.004}$ & $0.0068^{+0.0004}_{-0.0004}$ \\
825: E+A galaxies & $-0.05^{+0.03}_{-0.03} $ & $0.08^{+0.01}_{-0.03} $\\
826: Emission-line galaxies\tablenotemark{b} & $-0.26^{+0.01}_{-0.02} $ & $0.176^{+0.009}_{-0.006} $\\
827: \cline{1-3}
828: e(a) galaxies\tablenotemark{b} & $-0.30^{+0.02}_{-0.03}$ & $0.12^{+0.01}_{-0.02}$ \\
829: e(b) galaxies\tablenotemark{b} & $-0.42^{+0.04}_{-0.04}$ & $0.18^{+0.02}_{-0.06}$ \\
830: e(c) galaxies\tablenotemark{b} & $-0.23^{+0.02}_{-0.02}$ & $0.175^{+0.01}_{-0.009}$
831: \enddata
832: \tablenotetext{a}{Mean and scale are calculated using the biweight estimator in \citet{beers90}}
833: \tablenotetext{b}{Because of the blue color cut mentioned in \S4.1, the mean and scattering of the emission-line galaxies and the subtypes, in particular the e(b) galaxies, may be biased.}
834: \end{deluxetable}
835:
836:
837: %%% separate the figures into two panels.
838: \begin{figure}[h]
839: \epsscale{1.0} \plotone{f9.eps}
840: \figcaption[f9.eps]{Top: (left) Color-magnitude diagram for galaxies of
841: different spectral type. Open red diamonds mark absorption-line galaxies,
842: open blue squares emission-line galaxies, and filled green circles E+A
843: galaxies. The dashed lines are the color and magnitude cuts from
844: Fig.~\ref{cmd}. (right) U-V v.s. V-R$_c$ of galaxies of different spectral
845: type. Bottom: same as top panel, but showing E+A galaxies in comparison with
846: their hypothesized progenitor population, e(a) and e(b) galaxies (magenta
847: and cyan symbols respectively). [The full resolution figure will be published in ApJ] \label{cc}}
848: \end{figure}
849:
850:
851: \subsection{Dynamical Scales}
852:
853:
854: %Table 5 : X-ray property
855: \begin{deluxetable}{lcl}
856: \tabletypesize{\scriptsize}
857: \tablewidth{0pc}
858: %\tablenum{3}
859: \tablecolumns{3}
860: \tablecaption{Cluster property\label{xrayprop}}
861: \tablehead{
862: \colhead{} & \colhead{value}& \colhead{unit}
863: }
864: \startdata
865:
866: $f_x$\tablenotemark{a} & $2.74\pm 0.03$ & $10^{-12}$~ergs~s$^{-1}$~cm$^{-2}$\\
867: $L_x$\tablenotemark{a} & $24.6\pm 0.3 $ & $10^{44}$~ergs~s$^{-1} $ \\
868: $kT$\tablenotemark{a} & $11.6\pm 0.5$ & kev \\
869: $\beta$ & $1.1\pm 0.1$ & \\
870: $\rho_{0}$ & $1.71\pm 0.05$ & $10^{14}~M_{\sun}$~Mpc$^{-3}$ \\
871: $r_c$ & $92 \pm 6 $ & arcsec \\
872: \cline{1-3}
873: %%%
874: z$_{cl}$\tablenotemark{b} & $0.5446\pm 0.0005$ & \\
875: $\sigma$\tablenotemark{b} & $1612 \pm 70$ & km~s$^{-1}$ \\
876: $R_{\rm vir,gal}$\tablenotemark{c} & $3.0\pm 0.2$ & Mpc \\
877: $R_{\rm vir,x-ray}$ \tablenotemark{c} & $2.9\pm 0.1$ & Mpc \\
878: $M_{x}$\tablenotemark{c} & $28\pm 5$ & $10^{14}$~M$_{\sun}$ \\
879: $R_{rs}$ & $1.9$ & Mpc
880: \enddata
881: %%%
882: \tablenotetext{a}{Ebeling et al. (2007)} \tablenotetext{b}{The values for the
883: mean redshift z$_{cl}$ and velocity dispersion $\sigma$ are calculated
884: iteratively using the redshifts of cluster members within the virial radius (2.9
885: Mpc) and thus differ slightly from the values published by \citet{ebel07}
886: where a radius of 1 Mpc was used.} \tablenotetext{c}{The uncertainty is
887: estimated by error propagation.}
888: \end{deluxetable}
889:
890: A fundamental problem of many forms of dynamical analysis is the inherent
891: assumption of virial equilibrium. Models appropriate for the description of
892: clusters in the process of active assembly, such as MACSJ0717.5+3745 or --
893: albeit to a lesser degree -- MS0451.6-0305 \citep{mora07}, would have to take
894: into account the effects of mergers and complex substructure on a wide range of
895: scales. With the exception of the simplest cases this is infeasible, in part
896: owing to our insufficient knowledge of the true three-dimensional geometry of the
897: system. With this caveat in mind, we note that the simplistic dynamical scales
898: derived in this section should only be considered crude estimates.
899:
900: The most relevant dynamical scales for this work are the virial radius $R_{vir}$
901: and the ram-pressure stripping radius $R_{rs}$. To obtain estimates of these
902: quantities, we need a description of the cluster's gas density profile as well
903: as a global X-ray temperature. The former is obtained by fitting a $\beta$-model
904: \citep{cava76}
905: \begin{equation}
906: \rho_{gas} = \rho_{0}\left[1+\left(\frac{r}{r_c}\right)^2\right]^{-\frac{3}{2}\beta} \label{beta}
907: \end{equation}
908: to the observed X-ray surface brightness. Specifically, the center of the X-ray
909: emission is determined by fitting a two-dimensional elliptical $\beta$-model to
910: the exposure-weighted image after point sources have been removed. The radial
911: profile is then extracted using this center position. The subsequent fit of a
912: one-dimensional $\beta$-model is limited to radii larger than $11\arcsec$ (70
913: kpc) to minimize the impact of the complex merging features in the heavily
914: disturbed cluster core. A global X-ray temperature is measured by extracting the
915: X-ray spectrum from $r=$70 kpc to $r=$0.94 Mpc (r$_{1000}$) and using
916: \textit{Sherpa} to fit a MEKAL plasma model \citep{mewe85} with the absorption term
917: frozen at the Galactic value. From the X-ray temperature and the $\beta$-model
918: profile, we estimate the virial radius $R_{vir}$ using the formula of
919: \citet{arna02}
920: \begin{eqnarray*}
921: R_{vir} & = & 3.80~\beta_T^{1/2}\Delta_z^{-1/2}~(1+z)^{-3/2} \nonumber \\
922: & & \times(\frac{kT}{10{\rm keV}})^{1/2}h_{50}^{-1}~{\rm Mpc} \nonumber
923: \end{eqnarray*}
924: with
925: \begin{eqnarray*}
926: \Delta_z = (200\Omega_0)/(18\pi^2\Omega_z). \nonumber
927: \end{eqnarray*}
928: Here $\beta_T$ is the normalization of the virial relation, i.e.,
929: $GM_v/(2R_{vir})= \beta_T~{\rm k}T$. The virial radius thus derived is about
930: 2.9 Mpc, which is consistent with the virial radius (3 Mpc) estimated from the velocity
931: %%% add a formula from Treu et al.
932: dispersion using the relation \citep{gunn72}
933: \begin{eqnarray*}
934: R_{vir} = 1.7{\rm h}^{-1}{\rm Mpc} \frac{\sigma}{1000\,{\rm km~s}^{-1}} [\Omega_m(1+z)^{3}+\Omega_{\Lambda}]^{-0.5}.
935: \end{eqnarray*}
936: The ram-pressure stripping radius is estimated by equating the gas pressure
937: required to strip a Milky-Way-like galaxy from all its gas to the gas density
938: obtained previously in our X-ray analysis (Eqn.~\ref{beta}). The former can be
939: obtained from the stripping requirement \citep{gunn72, treu03}
940: \begin{eqnarray*}
941: \rho_{gas}v_i^2 & > & 2.1\cdot 10^{-12} {\rm N m}^{-2} \left(\frac{v_{rot}}{220\, {\rm km~s}^{-1}}\right)^2\left(\frac{r_h}{10\,{\rm kpc}}\right)^{-1} \nonumber\\
942: & & \times \left(\frac{\Sigma_{HI}}{8\cdot10^{20} m_H\,{\rm cm}^{-2}}\right)
943: \end{eqnarray*}
944: Here $v_i$ is the velocity of the infalling galaxy which, in all other respects
945: (rotation velocity $v_{rot}$, scale length $r_h$, and HI surface density
946: $\Sigma_{HI}$), is assumed to be similar to the Milky Way, as indicated by the
947: units used above. The infall velocity $v_i$ is estimated as
948: \begin{eqnarray*}
949: v_{\rm i} (r)= \sqrt{\frac{2GM_{\rm x}}{r}-\frac{GM_{\rm x}}{R_{vir}}},
950: \end{eqnarray*}
951: where $M_{\rm x}$ is the total mass within $R_{vir}$ derived from the X-ray
952: properties of the cluster (see, e.g., \citet{maug03}. A summary of the results
953: as well as of related X-ray cluster properties is given in Table~\ref{xrayprop}.
954:
955: Additional dynamical scales, that are of less interest to us here, can be
956: computed to assess the efficiency of tidally triggered star formation, tidal
957: stripping, starvation, harassment, and galaxy mergers. As discussed by
958: \citet{treu03}, the first two effects are only important within a few hundred
959: kpc of the cluster core and are thus difficult to study statistically for any
960: single cluster. In addition, the complex X-ray morphology in the core of
961: MACSJ0717.5+3745 makes this system particularly ill-suited for quantitative
962: modeling. Starvation and harassment, on the other hand, are efficient at larger
963: %%% add references
964: radii, with the starvation radius being equivalent to the virial radius \citep{balogh00}, and
965: harassment being almost independent of cluster-centric distance \citep{moor96,moor98}. Since
966: probably all cluster galaxies under study here are thus subject to harassment,
967: it would not be easy to isolate the impact of this particular effect. The last
968: mechanism, galaxy merging, is largely irrelevant in the densest parts of the
969: cluster environment, owing to the large velocity dispersion, but peaks near the
970: %%% add reference
971: virial radius (see e.g. \citep{ghigna98}).
972:
973:
974:
975: \subsection{Redshift Distribution of Cluster Members}
976:
977: The dynamical scales derived in the previous section refer to the primary
978: cluster, but exclude the filament. As a first look into the properties of the
979: large-scale structure observed, all the way from the cluster core to the
980: extended peak in the galaxy density in Region B, we show in Fig.~\ref{zonaline}
981: the distribution of galaxy redshifts projected onto a line starting northwest of
982: the cluster center and ending southeast of the filament, as indicated in
983: Fig.~\ref{plain}. We find the average redshift of the galaxies in Region B,
984: $\bar{z}_B = 0.5468\pm0.0004$, to be significantly higher than that of the
985: galaxies around the main cluster, $\bar{z}_A = 0.544\pm0.0006$. The velocity
986: difference between these two regions is thus 630 km/s. In addition, the
987: velocity dispersion of the galaxies in Region B is $\sigma_{z,B} = (1050\pm65)$
988: km/s, significantly less than the value of $\sigma_{z,A} = (1630^{+75}_{-93})$ km/s
989: found in Region A. Although the Chandra ACIS-I image of MACSJ0717.5+3745 covers
990: a fair fraction of Region B, it just misses the galaxy density peak at its
991: center. Since, in addition, the exposure of this observation is only 60 ks, we
992: are currently unable to obtain an estimate of the gas temperature and mass in
993: Region B. Deriving a mass estimate from $\sigma_{z,B}$ is not meaningful because
994: of the clearly unvirialized nature of this structure.
995:
996:
997: \begin{figure}[ht]
998: \plotone{f10.eps}
999: \figcaption[paper_zonaline2.ps]{
1000: Top: redshift distribution of cluster members as a function of their projected
1001: distance from the cluster center, measured along the line shown in
1002: Fig.~\ref{plain}. Bottom: redshift histogram for galaxies in the two regions A
1003: and B of Fig.~\ref{plain}; the red lines refer to Region A (cluster center),
1004: the blue lines to Region B (filament). [The full resolution figure will be published in ApJ] \label{zonaline}}
1005: \end{figure}
1006:
1007:
1008: \subsection{Spatial Distribution of Cluster Members}
1009:
1010: We show in Fig.~\ref{galxradial} the distribution of different galaxy types as a
1011: function of (radial) distance to the cluster center. The fraction of
1012: absorption-line galaxies drops from $\sim$80\% in the cluster core to $\sim$30\%
1013: at the virial radius, where the fraction of emission-line galaxies shows the
1014: opposite behavior, increasing from about 5\% at the cluster center to almost
1015: 70\% at the virial radius. Beyond 3.5 Mpc, however, both trends are reversed as
1016: we approach the galaxy density peak of Region B at 4.5 Mpc. At the largest radii
1017: probed by our study, the ratio of the numbers of absorption- and emission-line
1018: galaxies begins to approach the field value. An interpretation of our data in
1019: the context of the well known morphology-density relation\citep{dres80} will be
1020: given in the next section.
1021:
1022: Interestingly, the distribution of E+A galaxies follows neither of the above
1023: trends. Rather, almost all E+A galaxies (16 out of 17) are found to reside in
1024: Region A. No E+A galaxy is detected at cluster-centric distances exceeding 2
1025: Mpc, with the sole exception of a single specimen near the southern edge of
1026: Region B. Remarkably, the radial locus of the abrupt decline in the number of
1027: E+A galaxies thus coincides almost perfectly with the ram-pressure stripping
1028: radius, $R_{rs}=1.9$ Mpc.
1029:
1030: The spatial distribution of starburst galaxies is very different again: the
1031: distribution of both e(a) and e(b) galaxies follows, qualitatively, the
1032: distribution of their parent type, emission-line galaxies in general. Comparing
1033: the e(a) and e(b) distributions, the fraction of e(b) galaxies appears to show a
1034: more gradual and monotonic rise with cluster-centric distance than that of e(a)
1035: galaxies. The difference between the e(a) and e(b) distributions is, however,
1036: not significant: a Kolmogorov-Smirnov test finds the probability of them being
1037: drawn from the same parent distribution to be 0.69. We note though, that the
1038: aforementioned incompleteness of up to 30\% in our e(b) sample would likely
1039: increase the fraction of e(b) galaxies at all cluster-centric radii.
1040:
1041:
1042: \begin{figure}[h]
1043: \epsscale{1.0} \plotone{f11.eps}
1044: \figcaption[f11.eps]{Distribution of galaxies of different
1045: spectral type with cluster-centric distance. The symbols the same as in
1046: Fig.~\ref{cc}. In each bin the fraction is calculated by dividing the
1047: number of member galaxies of a given spectral type by the total number of
1048: member galaxies observed spectroscopically. The plotted error bars assume
1049: Poisson statistics. Some symbols are shifted slightly by $\pm0.02$ Mpc for
1050: clarity. \label{galxradial}}
1051: \end{figure}
1052:
1053: Although more difficult to interpret quantitatively, the two-dimensional spatial
1054: distribution of galaxies of different spectral types provides additional
1055: insight. To facilite a visual analysis, we show, in Fig.~\ref{galxdistr}, the
1056: respective distributions separately for emission-line galaxies, absorption-line
1057: galaxies, and -- in a third panel -- E+A, e(a), and e(b) galaxies. The contours
1058: of the projected galaxy density of Fig.~\ref{galdenmap} (top panel) are
1059: overlaid. We compare the distributions shown in these three panels in the
1060: following.
1061:
1062: The absorption-line galaxies are strongly concentrated in the double core of
1063: MACSJ0717.5+3745 proper, a region which is almost entirely devoid of
1064: emission-line galaxies. High concentrations of absorption-line galaxies are also
1065: found at the locations of two galaxy groups embedded in the filament, both of
1066: which are also detected in X-rays (Fig.~\ref{xrayim}). The string of three
1067: high-density regions near the nominal cluster center thus causes an excess of
1068: absorption-line galaxies in Region A, which extends linearly over almost 3 Mpc
1069: northwest to southeast of the cluster core. By contrast, the observed
1070: overdensity of cluster members in Region B has a much less pronounced effect on
1071: the distribution of absorption- and emission-line galaxies: overall, both galaxy
1072: types exhibit similar concentrations. Interestingly, absorption-line galaxies in
1073: this region are, however, found preferentially south of the center of Region B,
1074: whereas emission-line galaxies dominate the northern part of this area, roughly
1075: 0.5 Mpc away in projection. It is this large-scale segregation between
1076: absorption- and emission-line galaxies that is responsible for the shift of the
1077: overall distribution of cluster members in Region B shown in the top and center
1078: panel of Fig.~\ref{galdenmap}; the apparent, not well defined center of this
1079: region is inhabited by a concentration of e(a) and e(b) galaxies (bottom panel
1080: of Fig.~\ref{galdenmap}). Relating these observations to the true
1081: three-dimensional distribution of galaxy types in Region B is, however, greatly
1082: complicated by both projection effects and the likely presence of significant
1083: peculiar velocities \citep{barrett05}. Looking at yet larger scales, it is clear
1084: from Fig.~\ref{galxdistr} that the distribution of emission-line galaxies is
1085: vastly more extended over our study region than is the distribution of
1086: absorption-line galaxies. The lack of pronounced overdensities in the
1087: distribution of emission-line galaxies suggests strongly that a significant
1088: fraction of emission-line galaxies observed apparently near high-density regions
1089: of the MACSJ0717.5+3745 system are in fact projected there from much larger
1090: distances along the line of sight. This point is worth keeping in mind for the
1091: interpretation of the true spatial distribution of all types of emission-line
1092: galaxies.
1093:
1094: E+A galaxies, finally, are found to be distributed almost evenly within the
1095: ram-pressure stripping radius $R_{rs}$; we again note, however, the likely
1096: presence of projection effects. Since the small number of E+A galaxies in our
1097: sample does not allow us to derive a deprojected radial density profile, we can,
1098: at present, not distinguish between a spatial model in which E+A galaxies
1099: inhabit a shell around the cluster core, and one in which their distribution
1100: increases toward the cluster core as is expected for the absorption-line
1101: galaxies. Although a comparison of the respective projected radial distributions
1102: of E+A and absorption-line galaxies suggests the former, a Kolmogorov-Smirnov
1103: test finds them to be consistent with each other at the $1\sigma$ confidence
1104: level. From our study of this cluster alone we are thus unable to address the
1105: issue of whether E+A galaxies avoid the very cores of galaxy clusters as has
1106: been suggested by other authors \citep{tran03,dres99}.
1107:
1108: \begin{figure*}[h]
1109:
1110: %\epsscale{0.9}
1111: \epsscale{0.48}
1112: \plotone{f12a.eps}
1113: \plotone{f12b.eps}
1114: \plotone{f12c.eps}
1115:
1116: \figcaption[paper_galxtype_distr1.ps]{Projected spatial distribution of
1117: different galaxy types in the MACSJ0717.5+3745 system. Symbols are as in
1118: Fig.~\ref{cc}, with the addition of small black dots which mark the locations
1119: of all galaxies within our study region that have been observed
1120: spectroscopically. Also shown are the galaxy-density contours from
1121: Fig.~\ref{plain}.
1122: %%
1123: The green circles have radii of $R_{vir}$ (dashed) and $R_{rs}$ (solid),
1124: respectively, as estimated in \S3.3 [The full resolution figure will be published in ApJ]\label{galxdistr}}
1125: \end{figure*}
1126:
1127:
1128: \subsection{Projected Density Distribution of Cluster Galaxies}
1129:
1130: One of the most important parameters for studies of galaxy evolution is the
1131: density of the local environment. In Fig.~\ref{galxdensity} we plot the fraction
1132: of galaxies of different types as a function of the projected local galaxy
1133: density, $\Sigma_{10}$, as estimated from the sample of photometrically selected
1134: cluster members (i.e., $z_{ph}=$0.507 to 0.580). The observed monotonic rise and
1135: fall of the fractions of emission- and absorption-line galaxies (top panel) with
1136: decreasing galaxy density, $\Sigma_{10}$, confirms our interpretation of the
1137: trends already seen in Fig.~\ref{galxradial} and suggests strongly that the
1138: projected density is a more fundamental physical parameter for the distribution
1139: of the emission- and absorption-line galaxies than the distance to the cluster
1140: center. This result is consistent with the relation between star-formation rate
1141: and local density, as well as with the morphology-density relation, from studies
1142: of other clusters at intermediate redshift \citep{dres80,pogg99,koda01}. Note,
1143: however, how the fraction of either galaxy type is almost insensitive to the
1144: density of the environment until the latter reaches a value typical of galaxy
1145: groups ($\sim$ 20 Mpc$^{-2}$), at which point the divide between absorption- and
1146: emission-line galaxies begins to open up dramatically \citep{koda01}.
1147:
1148: The bottom panel of Fig.~\ref{galxdensity} emphasizes the stark contrast between
1149: the environmental dependence of E+A galaxies and that of e(a) and e(b)
1150: galaxies. The latter two galaxy types appear to, qualitatively, follow the trend
1151: already seen for emission-line galaxies in general (which is not all that
1152: surprising given that the e(a) and e(b) classification denotes subtypes of
1153: emission-line galaxies), but with a possibly higher sensitivity to local density
1154: already at values well below those characteristic of galaxy groups. E+A
1155: galaxies, on the other hand, appear to follow the opposite trend of favoring
1156: environments of intermediate to high density.
1157:
1158: The histogram of the $\Sigma_{10}$ values observed in Region A and Region B
1159: shown in Fig.~\ref{galdens2pks} illustrates that the range of projected local
1160: galaxy densities in Region B overlaps significantly with the one encountered in
1161: the primary cluster region. Indeed, both emission- and absorption-line galaxies
1162: show the same trends in either region (Figs.~\ref{galxradial},\ref{galxdensity}):
1163: their fractions are mainly a function of $\Sigma_{10}$. This is not true for
1164: E+A galaxies: although the majority of E+A galaxies found in MACSJ0717.5+3745
1165: are associated with $\Sigma_{10}$ values that are most commonly encountered in
1166: Region B, only one out of our 17 E+A galaxies is in fact detected there. If
1167: density were the determining factor in the creation of E+A galaxies, their
1168: distribution with $\Sigma_{10}$ in Region A would lead to a prediction of 9
1169: E+A galaxies in Region B -- a number that is inconsistent with the observed
1170: count of one at the 3$\sigma$ significance level. We conclude that, at least
1171: for the system under investigation here, the physical mechanism responsible for
1172: the termination of star formation in this galaxy type must be a different one.
1173:
1174:
1175: \begin{figure}[h]
1176: \epsscale{1.0} \plotone{f13.eps}
1177: \figcaption[f13.eps]{Fraction of galaxies of different spectral
1178: types as a function of projected local density. Symbols are as in
1179: Fig.~\ref{galxdistr}. Error bars assume Poisson statistics. The symbols for
1180: E+A galaxies and absorption-line galaxies are slightly shifted horizontally
1181: for clarity.\label{galxdensity}}
1182: \end{figure}
1183:
1184:
1185: \begin{figure}[h]
1186: \epsscale{1.0} \plotone{f14.eps}
1187: \figcaption[paper_galdensity2pks.ps]{Histogram of projected local galaxy
1188: densities for Region A (solid line) and Region B (dashed line). See
1189: Fig.~\ref{plain} for a definition of these regions. The filled (open) green
1190: circles mark the densities at which we detect E+A galaxies within Region A
1191: (B).\label{galdens2pks}}
1192: \end{figure}
1193:
1194:
1195: \section{Discussion}
1196:
1197: The arguably most interesting result of our spectral analysis of the galaxy
1198: population of the cluster/filament system MACSJ0717.5+3745 is the spatial
1199: distribution of E+A galaxies, which may provide important clues about the
1200: physical mechanism triggering the E+A transition phase between actively star
1201: forming and passively evolving galaxies. Three aspects need to be considered in
1202: this context: what is the origin of the galaxies that evolve into E+A galaxies,
1203: what are the mechanisms that trigger the starburst, and what are the
1204: mechanism(s) responsible for terminating it?
1205:
1206:
1207: \subsection{E+A Progenitors}
1208:
1209: A first clue about the origin of the E+A galaxies in MACSJ0717.5+3745 can be
1210: obtained from their spatial distribution. If the galaxies that develop the E+A
1211: spectral characteristics originated from the large-scale filament or the galaxy
1212: groups embedded therein, one would expect E+A galaxies to be found
1213: preferentially near the filament-cluster interface. According to
1214: Fig.~\ref{galxdistr} this is not the case. Although the spatial distribution of
1215: E+A galaxies exhibits a slight elongation in the direction of the (projected)
1216: direction of infall along the filament, their location with respect to the
1217: cluster core is consistent with a roughly spherical distribution which, in turn,
1218: suggests an isotropic distribution of the progenitor population.
1219:
1220: A different approach would be to attempt to directly identify the most likely
1221: progenitors. The strong H$_{\delta}$ absorption in the spectra of E+A galaxies
1222: indicates that not only the starburst stopped both abruptly and recently (within
1223: a few hundred millions years), but also that a very significant population of
1224: young stars was thus generated (e.g.\ \citet{barger96}, and the review in
1225: \citet{pogg04}). This is particularly true for this study since we have
1226: intentionally selected the E+A galaxies with extreme Balmer absorption
1227: features. Galaxies currently undergoing strong star formation i.e., the e(a) and
1228: e(b) subtypes of emission-line galaxies, are thus natural candidates for the
1229: role of progenitors of E+A galaxies, and should, to some extent, trace the
1230: spatial distribution of the progenitor population.
1231: Examining, again, Fig.~\ref{galxdistr} we see that both e(a) and e(b) galaxies
1232: show a rather flat distribution which, however, prefers the general direction of
1233: the filament and is, globally, distinctly un-isotropic with respect to the
1234: cluster core. Although the distribution of e(a) and e(b) galaxies thus does not
1235: rule out the hypothesis that the filament represents an important reservoir from
1236: which E+A galaxies originate, the combined evidence appears to favor a picture
1237: in which E+A galaxies evolve from a population of galaxies residing
1238: predominantly in a halo surrounding the entire cluster+filament complex. The
1239: characteristic density of this typical e(a)/e(b) environment is higher than that
1240: of the infall region but lower than the one found in galaxy groups --- as a
1241: result, these potential E+A progenitors are missing from the denser regions of
1242: our study area as well as from the field.
1243:
1244: Results from our complete survey of the galaxy population of the most distant
1245: MACS clusters should allow us to address these issues with a much larger degree
1246: of certainty.
1247:
1248: \subsection{Source of Starburst Galaxies}
1249:
1250: The next question is how the starbursts are triggered in cluster galaxies. The
1251: spatially fairly even distribution of e(a) and e(b) galaxies
1252: Fig.~\ref{galxdistr}, which nonetheless shows a preference for the general area
1253: of the filament, represents evidence that starbursts are triggered in
1254: regions well outside the cluster, and in fact well outside the central regions
1255: of the filament too. The densities typical of the environment in which e(a) and
1256: e(b) galaxies predominantly reside can be estimated from both the spatial
1257: distribution in the filament region (Fig.~\ref{galxdistr}) and our direct
1258: measure of the projected galaxy density (Fig.~\ref{galxdensity}). Taking into
1259: account the certainty of projection effects we arrive at \S5.5. This result that
1260: star formation is triggered at radii approaching the virial radius and
1261: environmental densities well below that of galaxy groups is qualitatively
1262: consistent with the findings of \citet{koda01} and \citet{marc07}.
1263:
1264: \subsection{Star Formation Termination}
1265:
1266: Although it is generally accepted that environmental effects other than galaxy
1267: mergers are important to quench star formation in E+A galaxies in clusters
1268: (\citet{tran03} and references therein), there has, so far, been little direct
1269: observational evidence that would favor any particular physical mechanism. From
1270: our wide-field study of the spatial distribution of E+A and other types of
1271: spectrally classified galaxies in MACSJ0717.5+3745, we believe to have found the
1272: strongest evidence to date for ram-pressure stripping being the most effective,
1273: and in fact possibly only, mechanism driving the rapid evolution of E+A galaxies
1274: in clusters\footnote{Note that we refer here to the extreme E+A galaxies as
1275: defined by our strict criteria of Balmer line absorption. We do not rule out
1276: the possibility that other physical mechanisms play a significant role in the
1277: evolution of post-starburst galaxies with weaker Balmer absorption
1278: lines.}. This is because almost all of the E+A galaxies in the
1279: MACSJ0717.5+3745 system are found to lie, in projection, near the cluster
1280: core\footnote{Although spurious trends could conceivably be introduced into the
1281: observed spatial distribution if all or most of our E+A galaxies were in fact
1282: dust-enshrouded emission-line galaxies \citep{smail99,sato06}, it is difficult
1283: to imagine reasons for which any dust-extinction patterns should correlate
1284: this closely with cluster radius.} and within the ram-pressure stripping
1285: radius $R_{rs}$.
1286: %%%
1287: Simulations (e.g.\ \citep{tonn07}) show that ram-pressure stripping is the most
1288: efficient mechanism to remove the gas in the galaxies, and, therefore, to
1289: terminate star formation. Furthermore, the complex substructure in both X-ray
1290: emission (Fig.~\ref{xrayim}) and galaxy density (Fig.~\ref{galdenmap}) makes
1291: MACSJ0717.5+3745 a perfect representative of a the kind of cluster merger
1292: explored in the numerical simulation of \citet{fujita99}. The latter authors
1293: find that the ram-pressure on galaxies near the cluster core increases
1294: dramatically as clusters collide, and that cluster mergers also bring many blue
1295: galaxies near the cluster center -- the combination of these two effects leads
1296: to an increase in the fraction of post-starburst galaxies increases in the
1297: cluster core.
1298: %%%
1299: The results of our study firmly rule out galaxy mergers as the primary driver
1300: for the creation or evolution of E+A galaxies. If galaxy mergers were
1301: responsible for the termination of star formation in E+A galaxies, we would
1302: expect to see their distribution peak in regions where the galaxy densities and
1303: relative velocities are most conducive to mergers, i.e., group-like environments
1304: and the cluster infall region itself. This is clearly inconsistent with the
1305: results presented in \S5.6 (Figs.~\ref{galxdistr} and \ref{galdens2pks}). The
1306: less violent galaxy-galaxy interaction mechanisms harassment and starvation are
1307: ruled out too, not only because they cannot explain the observed concentration
1308: of E+A galaxies near the regions of steeply increasing intra-cluster gas
1309: density, but also because they act over effective time scales \citep{treu03} of
1310: several billion years and are thus too slow to sharply terminate the star
1311: formation in E+A galaxies. By contrast, the time \citep{treu03} needed for a
1312: galaxy to fall from the virial radius, where star formation would be triggered,
1313: to the ram-pressure stripping radius, where the star formation would be
1314: terminated again essentially instantaneously, and then on to the cluster core
1315: ($\sim$ 1Gyr) is well matched to the lifetime of the E+A phase ($\sim$ 1Gyr). We
1316: note though that galaxy mergers are not ruled out in terms of effective time
1317: scale (a few hundreds million years) as a mechanism for the creation of E+A
1318: galaxies in general. In the field mergers appear indeed to be the dominant
1319: mechanism \citep{zabl96,yang04,goto03, tran04}. This dichotomy between the
1320: origin of E+A galaxies in low- and high-density environments has previously been
1321: suggested by \citet{tran03} and \citet{tran04}.
1322:
1323:
1324: \subsection{Comparison to Other Clusters}
1325:
1326: \begin{figure}[t]
1327: %\epsscale{0.9}
1328: \epsscale{1.0}
1329: \plotone{f15.eps}
1330: \figcaption[paper_MS0451.eps]{Spatial distribution of different galaxy types in
1331: MS0451.6--0305 and Cl0024+16 based on data from \citet{mora07}. The symbols
1332: are the same as in Fig.~\ref{galxdistr}. The solid circle has radius
1333: $R_{vir}$, whereas the dashed circle marks $R_{rs}$ as shown in Figure 1 of
1334: \citet{mora07}. [The full resolution figure will be published in ApJ]
1335: \label{Moran}}
1336: \end{figure}
1337:
1338:
1339: %%% add reasons why these two system,
1340: To test whether our conclusions extend beyond MACSJ0717.5+3745, we investigate
1341: the distribution of E+A galaxies in the only two other clusters at intermediate
1342: redshift for which spectroscopic data of comparable quality are available over a
1343: similarly large field of view, namely MS0451.6--0305 and Cl0024+16\footnote{EW
1344: measurements for these systems were obtained from
1345: http://www.astro.caltech.edu/$\sim$smm/clusters/.}\citep{treu03,mora07}. Although
1346: the details of their survey in \citet{mora07}, such as the selection criteria,
1347: the magnitude limit, the instrumental setup, and the technical details of the EW
1348: measurements\footnote{\citet{mora07} adopt the Lick system\citep{worthey94} for
1349: their EW measurements; we perform our spectral classification using their
1350: $H{\delta_A}$ and $H{\gamma_A}$ values. Although use of the Lick-system
1351: results will affect our classification to some extent, we expect the
1352: differences to be insignificant comparing to other systematic uncertainties.},
1353: are different from those of our work, a qualitative comparison should still
1354: be permitted. In addition, we attempt a comparison between MACSJ0717.5+3745 and
1355: a similarly massive cluster in the local universe, to test whether the spatial
1356: distribution of E+A galaxies is consistent. This would imply that, regardless of
1357: any redshift evolution in the relative prevalence of galaxies of a given
1358: spectroscopic type, ram-pressure stripping is the sole physical mechanism
1359: driving the evolution of E+A galaxies in clusters at all epochs. Since E+A
1360: galaxies are rare in local clusters we are left with a single system featuring
1361: enough E+A galaxies for a statistically meaningful comparison, the Coma
1362: cluster\citep{pogg04}.
1363:
1364: %%% a new paragraph to introduce MS0451 and CL0024
1365: MS0451.6--0305 ($z=0.54$, L$_{x} = 17~10^{44}$ergs~s$^{-1}$, T$_x = 7$~keV,
1366: Gioia \& Luppino 1994, Ebeling et al.\ 2007) is comparable to MACSJ0717.5+3745
1367: in that it is one of the most X-ray luminous and best-studied clusters at $z\sim
1368: 0.5$. Unlike MS0451.6--0305 and MACSJ0717.5+3745, Cl0024+16 ($z=0.39$)
1369: is an optically selected cluster and much less X-ray luminous (L$_{x} =
1370: 2.9~10^{44}$ergs~s$^{-1}$, T$_{x}=3.5$~keV, Zhang et al.\ 2005). We nonetheless
1371: include here for comparison purposes to investigate the relevance of cluster
1372: properties such as total mass, dynamical state, and core density. The Coma
1373: cluster, our local reference, is the most massive cluster at $z<0.05$ (T$_{x}=9.0$~keV (Donnelly et al. 1999).
1374:
1375: As shown in Fig.~\ref{Moran}, E+A galaxies are rare in both MS0451.6--0305
1376: ($1^{+1.1}_{-0.6}$\%) and Cl0024+16 ($0^{+0.7}_{-0}$\%) compared to
1377: MACSJ0717.5+3745 ($3.2^{+1.0}_{-0.8}$\%), although the uncertainties are
1378: large. However, their spatial distribution (all within $R_{rs}$) is consistent
1379: with what we observe in MACSJ0717.5+3742. In addition, the fact that the E+A
1380: galaxies in MS0451.6--0305 avoid the dense cluster core lends support to the
1381: hypothesis (see \S5.5) that the true three-dimensional spatial distribution of
1382: this type of galaxies resembles a shell inside the ram-pressure stripping
1383: radius. No E+A galaxy at all is found in Cl0024+16. However, as reflected in its
1384: much smaller ram-pressure stripping radius, this cluster is much less massive
1385: than either of MS0451.6--0305 and MACSJ0717.5+3745. In fact, the density of this
1386: cluster's core region is still likely to be overestimated because of the unusual
1387: geometric and dynamical properties of Cl0024+16, which is known to be a
1388: high-velocity merger along our line of sight \citep{czoske02}.
1389:
1390:
1391: Our results for MACSJ0717.5+3742 are
1392: consistent with the distribution of e(a)/e(b) galaxies in MS0451.6--0305 and
1393: Cl0024+16, too. In either cluster, these starburst candidates are distributed much
1394: more widely than the E+A or absorption-line population.
1395:
1396: The spatial distribution of E+A galaxies in Coma \citep{pogg04} is also
1397: consistent with our result\footnote{The study of \citet{pogg04} addresses
1398: properties of k+a galaxies selected based on $H_{\delta}$ and $[OII]$;
1399: although this definition is different from ours, the selected galaxy type is
1400: similar to our E+A galaxies.}. \citet{pogg04} separate E+A galaxies into two
1401: categories according to their color. Based on the relative position of the red
1402: sequence and the E+A galaxies in Fig.~\ref{cc}, the E+A galaxies in
1403: MACSJ0717.7+3745 mostly belong to the class of "blue E+A" galaxies, which
1404: \citet{pogg04} find to be concentrated near the core of Coma. A similar
1405: sub-population is selected by focusing on the "extreme" E+A galaxies with
1406: $H_{\delta} > 6$\AA (see also \S4.3). Applying this criterion to the k+a sample
1407: of \citet{pogg04} we again find most of them (7 out of 9) to be located near the
1408: cluster center but avoiding the very core. Finally, \citet{pogg04} speculate
1409: that the distribution of these youngest E+A galaxies in Coma traces the X-ray
1410: substructure, suggesting that the origin of these extreme E+A galaxies may be
1411: related to merger-induced shocks in the cluster gas. While, qualitatively, this
1412: picture is not in conflict with our findings for MACSJ0717.7+3745, a known major
1413: merger (Fig.~\ref{xrayim} shows the complex X-ray morphology of the cluster
1414: core), this hypothesis would make the absence of E+A galaxies in Cl0024+16 even
1415: harder to explain.
1416:
1417: \section{Conclusion}
1418:
1419: %%% four new paragraphs
1420: MACSJ0717.5+3745 is a very X-ray luminous cluster at $z=0.55$ which shows
1421: complex sub-cluster and filamentary structure in both X-ray surface brightness
1422: and galaxy density. Along the filament, we detect a significant offset in the
1423: average redshift of galaxies, corresponding to $\sim 630$km~s$^{-1}$ in
1424: velocity, as well as a decrease in velocity dispersion from the core of the
1425: primary cluster to the end of filament. These variations are likely a
1426: combination of spatial (Hubble flow) and kinematic effects (peculiar velocities)
1427: in this highly disturbed, merging system; a detailed analysis and discussion of
1428: the three-dimensional geometry and dynamics of MACSJ0717.5+3745 will be
1429: presented in a forthcoming paper.
1430:
1431: Using the spectroscopic data obtained by us, primarily with the DEIMOS
1432: spectrograph on the KECK-II telescope, we perform an extensive analysis of the
1433: spatial distribution of galaxies of different spectroscopic types, from the
1434: cluster core to the cluster outskirts and the connected linear filament. The
1435: spectroscopic sample is selected using blue and red color cuts defined relative
1436: to the empirical cluster red sequence in the V and R$_{c}$ bands. The final
1437: spectroscopic catalog is more than $70\%$ complete at R$_{c} < 21.75$. The applied
1438: blue color cut causes the number of emission-line galaxies in this structure, in
1439: particular the starburst galaxies, to be underestimated -- our analysis of the
1440: spatial distribution of the cluster members of various spectroscopic types should,
1441: however, not be affected.
1442:
1443: We classify cluster members into emission-line, absorption-line, and E+A
1444: galaxies using the strength of the [OII] and H$_{\beta}$ emission lines, as well
1445: as the absorption features H$_{\delta}$ and H$_{\gamma}$. The emission-line
1446: galaxies are further categorized into the subtypes e(a), e(b), and e(c),
1447: depending on the strengths of the [OII] emission- and H$_{\delta}$
1448: absorption-line features. We emphasize that in order to eliminate contamination
1449: from absorption- or emission-line galaxies, we adopt a very strict criterion for
1450: our definition of E+A galaxies, namely $\frac{(H_{\delta}+H_{\gamma})}{2} >
1451: 6$\AA, and also require the absence of [OII] and H$_{\beta}$ emission. The
1452: location of these E+A galaxies in color-magnitude and color-color diagrams
1453: confirms that they represent the transition phase between emission-line
1454: galaxies, in particular the e(a) and e(b) sub-types, and absorption-line
1455: galaxies.
1456:
1457: The main result of this study is that E+A galaxies are found to reside almost
1458: exclusively within the ram-pressure radius of the cluster. Since E+A galaxies
1459: are not observed in significant numbers in other regions of this system where
1460: the local galaxy density is similarly high, and since the relatively high galaxy
1461: velocities near the ram-pressure radius are unfavourable for galaxy mergers, we
1462: conclude that galaxy-gas interactions, rather than galaxy mergers, are likely
1463: the dominant mechanism for the evolution of the E+A galaxies, at least in
1464: clusters at intermediate redshift. Our interpretation that the star formation in
1465: these galaxies is terminated because the gas reservoir is being stripped away by
1466: the intra-cluster medium as galaxies fall into the cluster core is consistent
1467: with the results obtained for other two clusters, MS0451.6--0305 and
1468: Cl0024+16. E+A galaxies are only found within the ram-pressure radius of
1469: MS0451.6--0305, and no E+A galaxies at all are found in Cl0024+16, a low-mass
1470: system featuring a very small ram-pressure stripping radius. The large number
1471: of E+A galaxies around the core of MACSJ0717.7+3745, compared to the similarly
1472: X-ray luminous cluster MS0451.6--0305, can be naturally explained by the
1473: significant merger activity in the latter system, an effect that is consistent
1474: with the results of numerical simulations.
1475:
1476: Although the spatial distribution of E+A galaxies in MACSJ0717.7+3745 shows
1477: clear evidence of environmental effects on galaxy evolution in clusters, many
1478: issues still need further investigation. Specifically, we need to better
1479: understand the role of cluster mergers, and what exactly triggers starbursts as
1480: galaxies are falling into the cluster. Both of these questions can be addressed
1481: by extending this work to the full sample of the 12 most distant MACS clusters,
1482: thereby covering systems of very different dynamical states and a wide range of
1483: large-scale environments. For MACSJ0717.7+3745, accurate star-formation rates
1484: measured from infrared data obtained with Spitzer will help greatly in this
1485: regard. In addition we aim to use galaxy morphologies and deep X-ray
1486: observations to study in detail the gas-galaxy interactions which we suspect to
1487: be responsible for the termination of star formation activity and thus the
1488: creation of E+A galaxies in clusters.\\ \\
1489:
1490: We thank Leif Wilden and Glenn Morris of Stanford University for assistance with
1491: the photometric calibration of our SuprimeCam data for MACSJ0717.7+3745 and the
1492: anonymous referee for advice and suggestions that helped to improve this
1493: paper. Financial support for this work was provided by the National Aeronautics
1494: and Space Administration through Chandra Award Number GO3-4168X issued by the
1495: Chandra X-ray Observatory Center, which is operated by the Smithsonian
1496: Astrophysical Observatory for and on behalf of the National Aeronautics Space
1497: Administration under contract NAS8-03060. This research has made use of software
1498: provided by the Chandra X-ray Center (CXC) in the application packages CIAO,
1499: ChIPS, and Sherpa. The analysis pipeline used to reduce the DEIMOS data was
1500: developed at UC Berkeley with support from NSF grant AST-0071048.
1501:
1502:
1503: \begin{thebibliography}{}
1504: \bibitem[Arnaud et al.(2002)]{arna02} Arnaud, M., Aghnaim, N., \& Neumann, D. 2002, A\&A, 389, 1
1505: \bibitem[Arnouts et al.(1999)]{arno99} Arnouts, S., Cristiani, S., Moscardini, L. et al. 1999, MNRAS, 310, 540
1506: \bibitem[Balogh et al.(2000)]{balogh00} Balogh, M. L., Navarro, J. F., \& Morris, S. L. 2000, ApJ, 540, 113
1507: \bibitem[Balogh et al.(2004)]{balogh04} Balogh, M. L. et al. 2004, MNRAS, 348, 1355
1508: \bibitem[Barger et al.(1996)]{barger96} Barger, A. J., Maragon-Salamanca, A., Ellis, R. S., Couch, W. J., Smail, I., \& Sharples, R. M. 1996, MNRAS, 279, 1
1509: \bibitem[Barnes \& Hernquist(1992)]{barnes92} Barnes, J. E. \& Hernquist, L. 1992, ARAA, 30, 705
1510: \bibitem[Barrett (2005)]{barrett05} Barrett, E., 2005, PhD thesis
1511: \bibitem[Beers et al.(1990)]{beers90} Beers, T. C., Flynn, K., \& Gebhardt, K. 1990, AJ, 100, 32
1512: \bibitem[Bertin \& Arnouts(1996)]{bertin96} Bertin, E. \& Arnouts, S. 1996, A\&AS, 117, 393
1513: \bibitem[Blake et al.(2004)]{blak04} Blake, C., et al. 2004, MNRAS, 355, 713
1514: \bibitem[Blakeslee et al.(2006)]{blakeslee06} Blakeslee, J. P., Holden, B. P., Franx, M. et al. 2006, ApJ, 644, 30
1515: \bibitem[Bower \& Balogh(2004)]{bowe04} Bower, R. G. \& Balogh, M. L. 2004, Cluster if Galaxies: Probes of Cosmological Structure and Galaxy Evolution from the Carnegie Observatories Centennial Symposia, 325 % review
1516: \bibitem[Bruzual \& Charlot(2003)]{bc03} Bruzual, G. \& Charlot, S. 2003, MNRAS, 344, 1000
1517: \bibitem[Butcher \& Oemler(1978)]{bo78} Butcher, H \& Oemler, A. Jr. 1978, ApJ, 219, 18
1518: \bibitem[Calzetti(2001)]{calz01} Calzetti, D. 2001, NewAR, 45, 601
1519: \bibitem[Cappellari \& Emsellem(2004)]{capp04} Cappellari, M., \& Emsellem, E. 2004, PASP, 116, 816, 138
1520: \bibitem[Cavaliere \& Fusco-Femiano(1976)]{cava76} Cavaliere A.\ \& Fusco-Femiano R. 1976, A\&A, 49, 137
1521: \bibitem[Cooper et al.(2007)]{coop07} Cooper, M. C., Newman, J. A,, Weiner, B. J., et al. 2007, ApJ % DEEP
1522: \bibitem[Couch \& Sharples(1987)]{couc87} Couch, W. J., \& Sharples, R. M. 1987, MNRAS, 229, 423
1523: \bibitem[Cowie et al.(1996)]{cowie96} Cowie, L. L., Songaila. A., Hu, E. M., \& Cohen, J. G. 1996, AJ, 112, 839
1524: \bibitem[Czoske et al.(2002)]{czoske02} Czoske, O., Moore, B., Kneib, J.-P., Soucail, G. 2002, A\&A, 386, 31
1525: \bibitem[Donnelly et al.(1990)]{donn90} Donnelly, R. H., Markevitch, M., Forman, W., Jones, C., Churazov, E., \& Gilfanov, M. 1999, ApJ, 513, 690
1526: \bibitem[Donovan(2007)]{dono07} Donovan, D. 2007, PhD thesis
1527: \bibitem[Dressler(1980)]{dres80} Dressler, A. 1980, ApJ, 236, 351
1528: \bibitem[Dressler \& Gunn(1983)]{dres83} Dressler, A. \& Gun, J. E. 1983, ApJ, 270, 7
1529: \bibitem[Dressler et al.(1997)]{dres97} Dressler, A., et al. ApJ, 490, 577
1530: \bibitem[Dressler et al.(1999)]{dres99} Dressler, A., Smail, I., Poggianti, B. M., Butcher, H., Couch, W. J., Ellis, R. S., \& Oemler, A. J. 1999, ApJS, 122, 51
1531: \bibitem[Ebeling et al.(2001)]{ebel01} Ebeling, H., Edge, A., \& Henry, P. 2001, ApJ, 553, 668
1532: \bibitem[Ebeling et al.(2004)]{ebel04} Ebeling, H., Barrett, E., \& Donovan, D. 2004, ApJL, 609, 49 %M0717
1533: \bibitem[Ebeling et al.(2006)]{ebel06} Ebeling, H., White, D., \& Rangarajan, F. 2006, MNRAS, 368, 65 %asmooth
1534: \bibitem[Ebeling et al.(2007)]{ebel07} Ebeling, H., Donovan, D., Ma, C.-J., Edge, A., \& van Speybroeck, L. 2007, ApJL, 661, 33
1535: \bibitem[Fisher et al.(1998)]{fish98} Fisher, D., Fabricant, D., Franx, M., \& van Dokkum, P. 1998, ApJ, 498, 195
1536: \bibitem[Fujita et al.(1999)]{fujita99} Fujita,Y., Takizawa,M.,Nagashima,M.,\& Enoki, M. 1999,PASJ, 51,L1
1537: \bibitem[Ghigna et al.(1998)]{ghigna98} Ghigna, S., Moore,B., Governato, F., Lake, G., Quinn,T., \& Stadel, J. 1998, MNRAS, 300, 146
1538: \bibitem[Gioia \& Luppino(1994)]{gioia94} Gioia, I. M.,\& Luppino, G. A. 1994, ApJS, 94, 583
1539: \bibitem[Gomez et al.(2003)]{gome03} Gomez, P. L., Nichol, R. C., Miller, C. J., et al. 2003, ApJ, 584, 210 %SDSS
1540: \bibitem[Goto et al.(2003)]{goto03} Goto, T., Nichol, R., Miller, C., et al. 2003, PASJ, 55, 771
1541: \bibitem[Goto(2006)]{goto06} Goto, T. 2006, MNRAS, 369, 1765
1542: \bibitem[Goto(2007)]{goto07} Goto, T. 2007, MNRAS, 381, 187
1543: \bibitem[Gunn \& Gott(1972)]{gunn72} Gunn, J. E. \& Gott, J. R. III 1972, ApJ, 176, 1 % ram pressure stripping
1544: \bibitem[Hogg et al.(2006)]{hogg06} Hogg, D. W., Masjedi, M., Berlind, A. A., Blanton, M. R., Quintero, A., D., \& Brinkmann, J. 2006, ApJ, 650, 763
1545: \bibitem[Ilbert et al.(2006)]{ilbe06} Ilbert, O., Arnouts, S., \& McCracken, H., et al. 2006, A\&A, 457, 841
1546: \bibitem[Kartaltepe et al.(2008)]{jeyh08} Kartaltepe, J., Ebeling, H., Ma, C.-J. Donovan, D., Barrett, E. in preparation
1547: \bibitem[Kauffmann et al.(2003)]{kauf03} Kauffmann, G. et al. 2003, MNRAS, 341, 33 % D4000-age
1548: \bibitem[Kewley et al.(2006)]{kewl06} Kewley, L., Groves, B., Kauffmann, G., \& Heckman, T. 2006, MNRAS, 372, 961
1549: \bibitem[Kodama et al.(2001)]{koda01} Kodama, T, Smail, I, Nakata, F., Okamura, S., \& Bower, R. G. 2001, ApJL, 562, 9
1550: \bibitem[Larson et al.(1980)]{lars80} Larson, R. B., Tinsley, B. M., \& Caldwell, C. N. 1980, ApJ, 237, 692 % strangulation
1551: \bibitem[Lewis et al.(2003)]{lewi03} Lewis, I., Balogh, M., De Propris, R., et al. 2003, MNRAS, 334, 673 % 2dF
1552: \bibitem[Marcillac et al.(2007)]{marc07} Marcillac, D., Rigby, J., Rieke, G., \& Kelly, D. 2007, ApJ, 654 % spitzer cluster starburst
1553: \bibitem[Maughan et al.(2003)]{maug03} Maughan B.J., Jones L.R., Ebeling H., Perlman E., Rosati P., Frye C., Mullis C.R. 2003, ApJ, 587, 589
1554: \bibitem[Mewe et al.(1985)]{mewe85} Mewe, R., Gronenschild, E.H.B.M., \& van den Oord, G.H.J. 1985, A\&AS, 62, 197
1555: \bibitem[Miyazaki et al.(2002)]{miya02} Miyazaki S.\ et al.\ 2002, PASJ, 54, 833
1556: \bibitem[Moustakas et al.(2006)]{mous06} Moustakas, J., Kennicutt, R. C., \& Tremonti, C. A. 2006, ApJ, 642, 775
1557: \bibitem[Moore et al.(1996)]{moor96} Moore, B., Katz, N, Lake, G, Dressler, A, \& Oemler, A 1996, Nature, 379, 613 %harassment
1558: \bibitem[Moore et al.(1998)]{moor98} Moore, B., Lake, G., \& Katz, N. 1998, ApJ, 1007
1559: \bibitem[Moran et al.(2006)]{mora06} Moran, S., Ellis, R., Treu, T., Salim, S., Rich, R. M., Smith, G. P., \& Kenib, J.-P. 2006, ApJL, 641, 97
1560: \bibitem[Moran et al.(2007)]{mora07} Moran, S., Ellis, R., Treu, T., Smith, G., Rich, R., \& Smail, I. 2007, ApJ, 671, 1503 %MS0451, Cl0024
1561: \bibitem[Oemler(1974)]{oeml74} Oemler, A. Jr. 1974, ApJ, 194, 1
1562: %\bibitem[Owner et al.(2007)]{owne07} Owner et al. 2007 % 2dF XXXX
1563: \bibitem[Poggianti et al.(1999)]{pogg99} Poggianti, B. M., Smail, I, Dressler, A, Couch, W. J., Barger, A. J., Butcher, H, Ellis, R. S., \& Oemler, A., Jr. 1999, ApJ, 518, 576
1564: \bibitem[Poggianti \& Wu(2000)]{pogg00} Poggianti, B. M., \& Wu, H. 2000, ApJ, 529, 157
1565: \bibitem[Poggianti et al.(2004)]{pogg04} Poggianti, B., Bridges, T., Komiyama, Y., et al. 2004, ApJ, 601 %EpA in coma and high redshift clusters
1566: \bibitem[Poggianti et al.(2006)]{pogg06} Poggianti, B., von der Linden, A., De Luccia, G., et al. 2006, ApJ, 642 % evolution of the star formation activity in clusters
1567: \bibitem[Postman et al.(2005)]{post05} Postman, M. et al. 2005, ApJ, 623, 721
1568: \bibitem[Quintero et al.(2004)]{quin04} Quintero, A. D., Hogg, D., W., \& Blanton, M. R. 2004, ApJ, 602
1569: \bibitem[Ruderman et al.(2005)]{rude05} Ruderman, J., \& Ebeling, H. 2005, ApJL, 623, L81
1570: \bibitem[Sato \& Martin(2006)]{sato06} Sato, T., \& Martin, C. 2006, ApJ, 647
1571: \bibitem[Smail et al.(1999)]{smail99} Smail, I., Morrison, G., Gray, M. E., Owen, F. N., Ivison, R. J., Kneib, J.-P., \& Ellis, R. S. 1999, ApJ, 525, 609
1572: \bibitem[Smith et al.(2005)]{smit05} Smith, G. P., Treu, T., Ellis, R. S., Moran, S. M., \& Dressler, A. 2005, ApJ, 620, 78 %morphology-density -> S0
1573: \bibitem[Tonnesen et al.(2007)]{tonn07} Tonnesen, S., Bryan, G. L., \& van Gorkom, J. H. 2007, ApJ, 671, 1434
1574: \bibitem[Tran et al.(2003)]{tran03} Tran, K-V, Franx, M., Kelson, D., \& van Dokkum, P. 2003, ApJ, 599 %EpA (in 3 intermediate redshift cluster)
1575: \bibitem[Tran et al.(2004)]{tran04} Tran, K-V, Franx, M., Illingworth, G., van Dokkum, P., Kelson, D., \& Mages, D. 2004, ApJ, 609 %EpA field 0.3<z<1.0
1576: \bibitem[Tran et al.(2005)]{tran05} Tran, K-V., van Dokkum, P., Illingworth, G., Kelson, D., Gonzalez, A., \& Franx, M. 2005, ApJ, 619, 134 % MS2053 BO effect, did I use this one?
1577: \bibitem[Tran et al.(2007)]{tran07} Tran, K-V., Franx, M., Illingworth, G., van Dokkum, P., Kelson, D., Blakeslee, J., \& Postman, M. 2007, ApJ, 661 %MS1054 z=0.83, forming of redsequence
1578: \bibitem[Treu et al.(2003)]{treu03} Treu, T., Ellis, R., Kneib, J., Dressler, A., Smail, I., Czoske, O., Oemler, A., \& Nararajan, P. 2003, ApJ, 591
1579: \bibitem[Worthey et al.(1994)]{worthey94} Worthey, G, Faber, S. M., Gonzalez, J.J., \& Burstein, D. 1994, ApJS, 94, 687
1580: \bibitem[Yagi et al.(2006)]{yagi06} Yagi, M., Goto, T. \& Hattori, T 2006, ApJ, 642, 152
1581: \bibitem[Yan et al.(2006)]{yan06} Yan, R., Newman, J., Faber, S., Konidris, N., Koo, D., \& Davis, M. 2006, ApJ, 648, 281
1582: \bibitem[Yang et al.(2004)]{yang04} Yang, Y, Zabludoff, A. I., Zaritsky, D., Lauer, T. R., \& Mihos, J. C. 2004, ApJ, 607, 258 %E+A
1583: \bibitem[Zabludoff et al.(1996)]{zabl96} Zabludoff, A. I., et al. 1996, ApJ, 466, 104
1584: \bibitem[Zhang et al.(2005)]{zhang05} Zhang, Y. Y., Bohringer, H., Mellier, Y., Soucail, G., \& Forman, W. 2005, A\&A, 429, 85
1585:
1586: \end{thebibliography}
1587:
1588: \end{document}
1589:
1590:
1591:
1592:
1593:
1594: