1: \documentclass[apj]{emulateapj}
2:
3: \newcommand{\vdag}{(v)^\dagger}
4: \newcommand{\myemail}{puziat@nrc.ca}
5:
6: \slugcomment{Accepted for publication in the Astronomical Journal}
7:
8: \shorttitle{Evidence for the Evolution of Young Early-Type Galaxies in
9: the GOODS/CDF-S Field}
10: \shortauthors{Puzia, Mobasher, \& Goudfrooij}
11:
12:
13: \begin{document}
14:
15: \title{Evidence for the Evolution of Young Early-Type Galaxies in the
16: GOODS/CDF-S Field}
17:
18: \author{Thomas H. Puzia\altaffilmark{1,2,3}, Bahram
19: Mobasher\altaffilmark{3,4}, and Paul Goudfrooij\altaffilmark{3}}
20: \altaffiltext{1}{Herzberg Institute of Astrophysics, 5071 West Saanich
21: Road, Victoria, BC, V9E 2E7, Canada}
22: \altaffiltext{2}{Plaskett Fellow}
23: \altaffiltext{3}{Space Telescope Science Institute, 3700 San Martin Drive,
24: Baltimore, MD 21218}
25: \altaffiltext{4}{Affiliated with the Space Sciences Department
26: of the European Space Agency}
27:
28:
29: \begin{abstract}
30: We have developed an efficient photometric technique for identifying young
31: early-type galaxy candidates using a combination of photometric redshifts,
32: spectral-type classification, and optical/near-infrared colors.~Applying
33: our technique to the GOODS HST/ACS and VLT/ISAAC data we have selected a
34: complete and homogeneous sample of young elliptical candidates among
35: early-type field galaxies. The distribution of structural parameters for
36: these candidates shows that their selection, which is based on early
37: spectral types, is fully consistent with early morphological types. We
38: investigate the evolution of their luminosities and colors as a function
39: of redshift and galaxy mass and find evidence for an increasing starburst
40: mass fraction in these young early-type galaxy candidates at higher
41: redshifts, which we interpret in terms of massive field galaxies
42: experiencing more massive/intense starbursts at higher
43: redshifts.~Moreover, we find indications for a systematically larger young
44: elliptical fraction among sub-$L^{\ast}/2$ early-type galaxies compared to
45: their brighter counterparts.~The total fraction among the field early-type
46: galaxies increases with redshift, irrespective of galaxy luminosity.~Our
47: results are most consistent with galaxy formation scenarios in which stars
48: in massive early-type field galaxies are assembled earlier than in their
49: low-mass counterparts.
50: \end{abstract}
51:
52: \keywords{galaxies: K+A galaxies --- galaxies: formation and evolution}
53:
54:
55: \section{Introduction}
56: The last star-formation burst in galaxies defines their photometric
57: appearance and is an important diagnostic of galaxy formation and
58: evolution. The class of rejuvenated early-type galaxies includes so-called
59: `K+A' galaxies\footnote{Historically, these galaxies were named ''E+As''
60: because of their early-type morphology appearance on the very first
61: images. However, subsequent studies have shown that many of these galaxies
62: contain significant disk components, so that the ''E'' becomes unjustified
63: and is substituted by ''K'', owing to the spectral type of the underlying
64: old stellar population \citep{franx93}.} which are objects that show
65: spectroscopic signatures of old (K-type) and young (A-type) stellar
66: populations \citep{dressler83}.~The lack of emission lines in their
67: spectra indicates that star-formation processes abruptly ceased $\sim\!1$
68: Gyr ago, followed by a quiescent evolution into normal early-type galaxies
69: \citep[e.g.][]{couch87}.
70:
71: Several scenarios have been suggested to explain this observed
72: rejuvenation phenomenon:~(1) galaxy-galaxy mergers
73: \citep[e.g.][]{zabludoff96}, (2) interactions of infalling galaxies with
74: the intracluster medium \citep[e.g.][]{gunn72, bothun86}, (3) galaxy
75: harassment \citep[e.g.][]{moore96, moore98}, (4) tidally induced star
76: formation \citep[e.g.][]{byrd90}, and (5) dusty starbursts
77: \citep[e.g.][]{poggianti99, poggianti00}. While tidal interaction and
78: galaxy harassment are important events in the overall context of galaxy
79: formation and evolution \citep[e.g,][]{bekki05}, they do not provide an
80: obvious explanation for the sudden halt of star formation without invoking
81: additional processes. Ram-pressure stripping and mergers are currently
82: seen as the most viable scenarios to describe the ignition and sudden
83: cessation of star formation in young early-type galaxies
84: \citep[e.g.][]{rose01, bekki03, shioya02, shioya04, quitnero04}.~Moreover,
85: recent radio observations appear to exclude a dusty starburst scenario in
86: general, as the measured radio fluxes are inconsistent with high star
87: formation rates (i.e. $\ga\!10-100\ M_{\odot}\ a^{-1}$) for virtually all
88: local post-starburst early-type galaxies \citep{miller01, goto04}.
89: However, \cite{smail99} does find evidence for faint radio fluxes in
90: spectroscopically confirmed post-starburst galaxies in distant dense
91: clusters at $z\approx0.4$.
92:
93: K+A fractions are known to vary significantly with both redshift and
94: environmental density.~Several studies have revealed that the brightest
95: K+A's in nearby clusters are sub-$L^{*}$ systems \citep{caldwell99,
96: poggianti04}, while at intermediate redshifts ($z\ga0.8$) K+As are found
97: with $L\approx3 L^{*}$ \citep{dressler99, tran03}, suggesting an increase
98: of K+A cluster galaxy mass with redshift.~So far, the perhaps most
99: unbiased comparison between the K+A fractions in field and cluster
100: environments was done by \cite{tran03, tran04} who found a field K+A
101: fraction of $\sim\!3$\% and a factor $\sim\!4$ higher fraction in cluster
102: environment for the redshift range $0.3\!<\!z\!<\!1$.~\citeauthor{tran04}
103: further noted that the field K+A fraction shows strong fluctuations and is
104: sensitive to selection criteria. Furthermore, the relatively short
105: lifetime of the K+A signature \citep[$\sim\!1$ Gyr,][]{couch87, belloni95,
106: barger96} implies that these galaxies are rare.~This indicates the need
107: for wide-area surveys to identify statistically significant samples of
108: young early-type galaxies and determine their number fractions in both
109: field and cluster environments. If their selection is performed
110: consistently between datasets and galaxy formation models, their abundance
111: may be a sensitive diagnostic for star-formation activity in field and
112: cluster galaxies as a function of redshift, making these rejuvenated
113: objects useful tools to constrain hierarchical galaxy formation scenarios.
114:
115: Keeping in mind that medium-resolution spectroscopy of all galaxies in
116: large-area surveys is very expensive in terms of telescope time, the aim
117: of this paper is to develop a photometric technique to identify young
118: early-type galaxy {\it candidates} (hereafter yE candidates) for
119: spectroscopic follow-up using a combination of photometric redshifts,
120: spectral-type classification, and optical/near-infrared color-color
121: diagrams.~Applying this technique to the GOODS/CDF-S dataset
122: \citep{giavalisco04}, we select a complete and homogeneous sample of yE
123: candidates in the redshift range $0\!<\!z\!\la\!1$.~Finally, we
124: investigate correlations between the colors of yE candidate galaxies with
125: redshift and galaxy mass, which, if confirmed by spectroscopy, may put
126: strong constraints on galaxy formation models.~We use the standard
127: $\Lambda$CDM cosmology parameters $\Omega_{M}=0.3$,
128: $\Omega_{\Lambda}=0.7$, and $H_{0}=70$ km s$^{-1}$ Mpc$^{-1}$ throughout
129: this work.
130:
131: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
132: \section{GOODS/CDF-S Data}
133:
134: One of the two fields covered by the Great Observatories Origins Deep
135: Survey (GOODS) Treasury Program is the {\it Chandra Deep Field South}
136: (CDF-S) with optical HST/ACS and ground-based VLT/ISAAC near-infrared
137: photometry \citep{giavalisco04}.~These data span the wavelength range from
138: near-UV to the near-infrared and provide ACS imaging with very high
139: spatial resolution (0.03\arcsec) for $\sim\!86000$ galaxies down to
140: $V\approx27.5$.~Object detection and photometry was performed with
141: SExtractor \citep{bertin96} after the images in all passbands were
142: convolved to a common FWHM$\approx\!0.45$\arcsec.~The final magnitudes are
143: the MAG\_AUTO SExtractor magnitudes, which measure the total light.
144:
145: Using the multi-waveband data available for galaxies in this field,
146: photometric redshifts were measured following the technique presented in
147: \cite{mobasher07}. Briefly, this approach compares the template Spectral
148: Energy Distributions (SEDs) for different spectral types of galaxies,
149: shifted in redshift space, with the observed SED for galaxies detected in
150: GOODS (see also Sect.~\ref{spectypes}). $\chi^2$ fits are performed and
151: the redshift and spectral type corresponding to the minimum $\chi^2$ value
152: are associated to those for the galaxy in question. The template spectral
153: types consist of elliptical, Sbc, Scd, and Im-type galaxies from
154: \cite{coleman80} and starburst templates from \cite{kinney96}.
155: Bayesian priors based on observed galaxy luminosity functions are used
156: \citep[as detailed in][]{mobasher07}. Comparison with galaxies with
157: available spectroscopic redshifts yields a photometric redshift accuracy
158: of $\sigma(\Delta z)\!=\!0.03$ where $\Delta z=(z_{\rm phot}-z_{\rm
159: spec})/(1+z_{\rm spec})$.
160:
161: In order to check the accuracy of our spectral types, we compare these
162: with the morphology of galaxies measured from the GOODS-S ACS images. The
163: morphologies are estimated using the concentration and asymmetry
164: parameters \citep{conselice97, abraham96}, measured for individual
165: galaxies (B. Mobasher- private communication). We found a very good
166: correlation between the morphology parameters and spectral types,
167: particularly for elliptical galaxies. Therefore, we have reliable
168: identification of elliptical galaxies, based on their spectral types,
169: which constitutes the main selection criterion for the present study.
170:
171: We use the information on redshifts and spectral types to compute
172: $K$-corrections for all galaxies in our sample. Throughout the
173: rest of this work we use AB magnitudes \citep{oke74}\footnote{AB
174: magnitudes are defined by $m_{\rm AB}=-2.5\log f_{\nu}-48.6$, where
175: $f_{\nu}$ is in ergs s$^{-1}$ cm$^{-2}$ Hz$^{-1}$. Alternatively, one can
176: write $m_{\rm AB}=-2.5\log f_{\nu}+8.9$, where $f_{\nu}$ is the flux in
177: Jy.}, and correct for Galactic foreground reddening $E_{B-V}=0.008$
178: derived from the \cite{schlegel98} maps. We select galaxies brighter than
179: $M_{K}\!=\!-19.0$ mag from the initial GOODS/CDF-S sample which have a
180: mean photometric redshift accuracy of $\Delta z\leq0.15$ corresponding to
181: a 95\% confidence interval. This photometric redshift uncertainty limit is
182: introduced in order to keep the average color uncertainty below 0.15 mag
183: due to k-corrections (see also Sect.~\ref{ln:ssp}).~The final sample is
184: volume-selected and complete to $z\approx1$. We exclude objects where the
185: photometry was compromised by saturated pixels, corrupted or truncated
186: isophotal apertures, and where the extraction algorithm experienced
187: problems during deblending or the photometric errors were larger than
188: $\Delta m=0.15$ mag in any filter. The final catalog includes objects
189: with HST/ACS $BViz$ and VLT/ISAAC $JHK_s$ magnitudes\footnote{In the
190: following we shall refer to the $K_s$ filter as the $K$ band.}.
191:
192: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
193:
194: \section{Stellar Population Models}
195: \label{ln:ssp}
196:
197: \subsection{Diagnostic Colors}
198: Based on the \cite{bruzual03} simple stellar population (SSP) models and
199: their 2003 version of the GALAXEV stellar population synthesis code, we
200: compute optical/near-infrared photometric colors for composite stellar
201: populations (CSP).~We use the Salpeter IMF with limits at $0.1$ and $100
202: M_{\odot}$.~By comparison with the control sample of spectroscopically
203: confirmed K+As, we identify the best diagnostic combination
204: optical/near-infrared colors to select yE candidates.
205:
206: Given the availability of many filters in our GOODS dataset
207: ($BVizJHK_{s}$) we investigate several filter combinations that may most
208: efficiently reduce the age-metal\-licity-extinction degeneracy in the
209: color-color plane. Due to the presence of a hotter $\sim1$ Gyr young
210: stellar population, the $B$-band flux of yE galaxies is expected to be
211: enhanced relative to older ($\gg1$ Gyr) stellar populations.~For slightly
212: younger stellar populations with ages $\sim\!0.4\!-\!0.8$ Gyr the
213: Asymptotic Giant Branch (AGB) is densely populated with relatively cool
214: stars \citep{renzini81} causing a stronger output in the near-infrared
215: $K$-band \citep{persson83}.~Since a small population of luminous
216: thermally-pulsing AGB stars (TP-AGB) can significantly contribute to the
217: integrated light of galaxies, the consideration of such short-lived
218: evolutionary phases in SSP models is essential \citep{maraston98,
219: mouhcine03}.~Past this so-called AGB-phase transition, near-infrared light
220: is primarily sensitive to the mean temperature of the red giant branch,
221: which is mainly driven by the luminosity-weighted mean metallicity of
222: composite stellar populations.
223:
224: \begin{figure*}[!t]
225: \centering
226: \includegraphics[bb=0 100 400 400, width=12cm]{fig1.pdf}
227: \caption{Comparison of restframe solar-metallicity SEDs for
228: single-burst stellar populations with ages $t=290$ Myr, 640 Myr, 1.4 Gyr,
229: 2.5 Gyr, and 11 Gyr.~A dotted line illustrates the SED of a
230: solar-metallicity stellar population with a constant star formation rate
231: after 6 Gyr.~All SEDs were taken from \cite{bruzual03}.~Total filter
232: throughputs for ACS and the ISAAC camera, including mirror reflectivity,
233: window transmission, and detector quantum efficiency, are shown at
234: the top of the panel.}
235: \label{ps:sedsolar}
236: \end{figure*}
237:
238: In Figure~\ref{ps:sedsolar} we show the effective filter throughputs for
239: HST/ACS and the VLT/ISAAC instrument together with spectral energy distributions
240: (SEDs) for single-burst stellar populations for ages $t=290$ Myr, 640
241: Myr, 1.4 Gyr, 2.5 Gyr, and 11 Gyr at solar metallicity. For illustration
242: purposes all SEDs were normalized to the flux at $1.3\ \mu m$. It is clear
243: from the figure that the $B$ band contains the most age-sensitive flux and
244: that the $B\!-\!J$ color is a very good age indicator.~It is instructive
245: to note that the Lyman break is a sensitive indicator to identify
246: constantly star-forming or very young post-starburst galaxies
247: (i.e.~$t\la0.5$ Gyr, see dotted SED in Fig.~\ref{ps:sedsolar}).~However,
248: the Balmer break is the more sensitive proxy to identify post-starburst
249: galaxies with intermediate ages (i.e.~$t\ga0.5$ Gyr), such as K+A
250: galaxies.~To find a good metallicity indicator we need to move to the
251: near-infrared, past the $z$-band, where the continuum slope shows little
252: impact of varying age.~Here, the $H$-band continuum suffers from
253: variations due to strong water vapor absorption, which is sensitive to the
254: mean temperature of the stellar population
255: (i.e.~age-sensitivity).~However, the relative continuum fluxes in $J$ and
256: $K$ show little variation with age, and we define $J\!-\!K$ as our best
257: metallicity indicator which exhibits only modest age sensitivity during
258: the AGB-phase transition \citep[see also][]{ferraro00}.
259:
260: The best choice of the most age-sensitive and the most
261: metallicity-sensitive color depends also on data quality.~For reference,
262: the average color uncertainty introduced by the photometric redshift error
263: is $\la0.15$ mag in both $B\!-\!J$ and $J\!-\!K$ for all galaxy types at
264: $z\la1$. Given the mean uncertainty of all color combinations of our data,
265: the most age-sensitive color is $B\!-\!J$ and the most
266: metallicity-sensitive color is $J\!-\!K$, which is virtually independent
267: of dust reddening.~This $BJK$ combination optimizes the selection of
268: post-starburst intermediate-age stellar populations at redshifts
269: $z\!\la\!1$.~For higher redshifts, the age-metallicity degeneracy dilutes
270: the age/metallicity sensitivity of this color combination as the restframe
271: flux at $\lambda\!<\!1\ \mu m$ moves into the $J$ band, and a
272: different filter combination is required.
273:
274: A comparable photometric selection technique, although specifically
275: targeting star-forming galaxies at higher redshifts, was described by
276: \cite{daddi04}.~Similar techniques to detect intermediate-age globular
277: clusters (i.e.~single-burst stellar populations) in nearby galaxies have
278: been recently used by \cite{goudfrooij01}, \cite{puzia02}, and
279: \cite{hempel04}.
280:
281: \begin{figure}[!b]
282: \centering
283: \includegraphics[bb=125 100 400 400, width=5cm]{fig2.pdf}
284: \caption{Comparison of $B$ magnitudes for spectroscopically confirmed K+A galaxies
285: taken from \cite{galaz00} with those taken from the APM catalog.~A vertical solid line
286: shows the mean formal uncertainty.}
287: \label{ps:compphot}
288: \end{figure}
289:
290: \begin{figure*}[!ht]
291: \centering
292: \includegraphics[bb=0 0 400 400, width=8cm]{fig3a.pdf}
293: \includegraphics[bb=0 0 400 400, width=8cm]{fig3b.pdf}
294: \caption{{\it Left panel}: Comparison of $(B\!-\!J)_{\rm AB}$
295: vs.~$(J\!-\!K)_{\rm AB}$ restframe colors of spectroscopically confirmed
296: K+A galaxies ({\it open symbols}) with SSP models of \cite{bruzual03}. Open
297: squares mark cluster K+A galaxies, open circles are field K+As.~Isochrones
298: ({\it solid lines}) are plotted for ages $t=200$,
299: 500, 700 Myr, and 1, 1.3, 2, 5, and 15 Gyr.~Iso-metallicity lines ({\it dotted
300: lines}) are shown for the metallicities [Z/H]~$=-2.3, -1.7, -0.7, -0.4,
301: 0.0$, and $+0.4$ dex.~{\it Right panel}:~Same data, but this time with SSP
302: models of \cite{anders03}. Here isochrones ({\it solid lines}) were plotted for
303: the ages $t=200$, 500, and 700 Myr, and 1, 1.3, 2, 5, and 14 Gyr.
304: Iso-metallicity tracks ({\it dotted lines}) are indicated for [Z/H]~$=-1.7,
305: -0.7, -0.4, 0.0$, and $+0.4$ dex. Each panel has the extinction vector
306: for E$_{B-V}=0.2$ mag indicated in the upper left corner. Mean uncertainties
307: of the \cite{galaz00} data are shown in the upper right corner.}
308: \label{ps:sspcomp}
309: \end{figure*}
310:
311:
312: \subsection{Spectroscopically confirmed K+As}
313: To test our photometric selection technique we use the K+A galaxy sample
314: of \cite{galaz00}.~This data set contains spectroscopically confirmed K+As
315: in nearby ($z\approx0.05$) and distant ($z\approx0.3$) clusters and in the
316: nearby field ($z\approx0.1$).~All galaxies have been classified as K+As
317: \citep{couch87, franx93, caldwell97, zabludoff96}, without any indications
318: of emission lines.~The near-infrared photometry of the
319: \citeauthor{galaz00} study was performed with the Las Campanas NICMOS3
320: Hg:Cd:Te detector in the CIT filter system \citep{persson98}. We
321: compute the transformations to the \citep[][hereafter BB]{bb88} standard system
322: using \citet{bruzual03} SEDs with ages from 0.2
323: Gyr to 15 Gyr and [Fe/H] values from $-2.25$ to $+0.4$ dex, and find the
324: following (BB)--(CIT) transformations
325: \begin{eqnarray*}
326: J_{\rm BB} &=& J_{\rm CIT} - 0.0045 + 0.0102\, (J-K)_{\rm CIT} \\
327: (J-K)_{\rm BB} &=& 0.0147 + 0.9835\, (J-K)_{\rm CIT} \\
328: \end{eqnarray*}
329: with RMS scatter of 0.0009 mag and 0.007 mag, respectively.~Galaz'
330: Table~2 lists Johnson B magnitudes as reported in NED\footnote{the
331: NASA/IPAC Extragalactic Database, http://nedwww.ipac.caltech.edu} at the
332: time, likely coming from various original sources. We compare these $B$
333: magnitudes with those in the APM
334: catalog\footnote{http://www.ast.cam.ac.uk/\~{}apmcat} \citep{maddox90} and
335: show the comparison plot in Figure~\ref{ps:compphot}.~We find significant
336: offsets for the optical photometry between the \citeauthor{galaz00} and
337: the APM catalog.~Since the magnitudes in the APM catalog were derived in
338: an internally consistent way, we adopt the APM $B$ magnitudes for further
339: analysis in this paper.~The mean uncertainties in $BJK$ colors of the
340: comparison sample are 0.12, 0.10, and 0.18 mag, respectively.~All optical
341: and near-infrared magnitudes from the GOODS catalog and the comparison
342: sample represent total magnitudes.
343:
344:
345: \subsection{Confirmed K+As in the $BJK$ color-color plane}
346: In Figure~\ref{ps:sspcomp} we compare $(B\!-\!J)_{\rm AB}$
347: vs.~$(J\!-\!K)_{\rm AB}$ restframe colors of spectroscopically confirmed
348: K+A galaxies with SSP model predictions from two different groups:
349: \citeauthor{bruzual03} (\citeyear{bruzual03}, hereafter BC03) and
350: \citeauthor{anders03} (\citeyear{anders03}, hereafter AF03).~The BC03
351: predictions were calculated in the AB system with the same filter
352: transmission functions as the GOODS/CDF-S data. For the AF03 models, which
353: are tabulated in Vega-based magnitudes using the BB filter system,
354: we follow the Vega-AB transformations: $K_{\rm AB}=1.891+K_{\rm Vega}$, $J_{\rm
355: AB}=0.91+J_{\rm Vega}$, and $B_{\rm AB}=-0.105+B_{\rm Vega}$.~As expected,
356: the location of confirmed K+As in the color-color plane corresponds to old
357: to intermediate ages. We find that the absolute age calibration differs
358: between the two models.~The BC03 models predict (luminosity-weighted) ages
359: between $\sim\!15$ and $\sim\!2$ Gyr for most K+As, while the models of
360: AF03 predict younger ages in the range $\sim\!1\!-\!5$ Gyr.~The
361: metallicity scales on the other hand are very similar in both models. The
362: majority of confirmed K+A galaxies have luminosity-weighted metallicities
363: between [Z/H]~$\approx-0.7$ and $0.0$ dex, with few outliers at
364: super-solar metallicities, which might be reddened objects.
365:
366:
367: \subsection{Composite Stellar Populations in the BJK Color-Color Plane}
368: Several groups compared the mean colors of K+A galaxies using optical and
369: near-infrared photometry with population synthesis models
370: \citep[e.g.][]{newberry90, belloni95, barger96, shioya02, balogh05}.~In
371: this work we combine spectral-type fitting with model predictions and
372: investigate the influence of different metallicities and mass fractions of
373: the starburst population on colors and luminosities of post-starburst
374: galaxies in the optical/near-infrared color-color plane and in
375: color-magnitude diagrams.
376:
377: \begin{figure*}[!ht]
378: \centering
379: \includegraphics[bb=0 0 400 400, width=8cm]{fig4a.pdf}
380: \includegraphics[bb=0 0 400 400, width=8cm]{fig4b.pdf}
381: \caption{{\it Left panel}: Evolution of $(B\!-\!J)_{\rm AB}$
382: vs.~$(J\!-\!K)_{\rm AB}$ restframe colors for an SSP with a starburst that
383: ignites after 10 Gyr of passive evolution and decays for another 5
384: Gyr.~Two set of isochrones with ages $t=10.1, 10.2, 10.5, 11, 11.5$, and
385: 14.9 Gyr are shown for the metallicities [Z/H]~$=0.4$ ({\it solid lines})
386: and $-2.3$ dex ({\it dot-dashed lines}) of the starburst stellar
387: population.~The starburst strength amounts to 10\% of the total final stellar
388: mass.~{\it Right panel}:
389: Evolution of colors for different starburst strengths/masses for a solar
390: metallicity burst.~Ischrones for ages $t=10.2, 10.5, 11.5$, and 14.9 Gyr
391: are plotted with different line types ({\it solid, dashed,} and {\it
392: dotted lines}) indicating the dependence on starburst mass $M_{\rm
393: burst}=0.05, 0.1$, and 0.5 $M_{\rm total}$. For reference we overplotted
394: in both panels the SSP model grid from the left panel of
395: Figure~\ref{ps:sspcomp}.~All predictions were computed using the SSP
396: models of \cite{bruzual03}. Spectroscopically confirmed K+A galaxies are
397: as in Figure~\ref{ps:sspcomp}.}
398: \label{ps:sspburst}
399: \end{figure*}
400:
401: Figure~\ref{ps:sspburst} shows the evolution of BC03 isochrones where a
402: starburst is ignited after 10 Gyr of passive SSP evolution.~Assuming a
403: minor merger, we assign 10\% of the final stellar mass to the starburst
404: population. We adopt two metallicities Z~$=0.04$ (twice solar metallicity)
405: and Z~$=10^{-4}$ ([Z/H]~$=-2.3$ dex) for the starburst stellar population,
406: which are superposed on top of the full range of metallicities
407: ([Z/H]~$=-2.3$ to $+0.4$ dex.) of the underlying, older stellar
408: population.~Note the significant difference in $B\!-\!J$ colors during
409: the first $\sim\!1.5$ Gyr after the starburst.~If determined relative to
410: passively evolving galaxies, this color offset is a sensitive
411: post-starburst indicator and will be discussed below in more detail.
412:
413: In the right panel of Figure~\ref{ps:sspburst} we show the influence of
414: changing starburst mass fraction on the post-starburst galaxy colors.~This
415: time we choose solar metallicity for the starburst with mass
416: fractions are 5\%, 10\% (minor merger), and 50\% (major merger) of the
417: final total stellar mass. The Figure illustrates that the higher the
418: starburst mass fraction the bluer the $B\!-\!J$ color is during the
419: post-starburst phase, in agreement with previous studies
420: \citep[e.g.][]{barger96}.~In addition, there is a mass
421: fraction-metallicity degeneracy of post-starburst near-infrared colors.
422: Metal-rich major starbursts can exhibit the same $J\!-\!K$ colors as
423: metal-poor minor starbursts.~This degeneracy is most prominent for
424: intermediate metallicities in the range $-0.7\la$~[Z/H]~$\la0$.~For more
425: extreme metallicities the burst strength can be constrained with
426: increasing confidence.~In other words, yE galaxies with starburst mass
427: fractions $\la 10$\% and $J\!-\!K\ga0$ are likely to have experienced a
428: recent starburst with super-solar metallicity.~The same mass-fraction
429: limit and $J\!-\!K\la-0.3$ mag indicate sub-solar starburst metallicities.
430:
431: In summary, $B\!-\!J$ is mainly degenerate in age and starburst mass
432: fraction, $J\!-\!K$ is mainly degenerate in metallicity and starburst mass
433: fraction.~We conclude that the $B\!-\!J$ vs. $J\!-\!K$ color-color plane
434: can be used to identify post-starburst galaxies.~But only extreme
435: metallicities of the post-starburst stellar population can be narrowed
436: down robustly.~The color-color plane is not suitable to derive starburst
437: mass fractions for intermediate metallicities without any other
438: information.~However, these degeneracies can be partly lifted by including
439: restframe $K$-band luminosities.~Our findings are relatively independent
440: of the choice of SSP models, as the systematics in the AF03 models give
441: the same results.
442:
443: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
444: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
445:
446: \section{Selection of {\footnotesize y}E Candidates}
447: \subsection{Spectral Types}
448: \label{spectypes}
449: In the following we develop a robust method to identify yE galaxy
450: candidates from the GOODS/CDF-S photometric catalog using a combination of
451: photometric redshifts, spectral-type classification, and
452: optical/near-infrared colors.~The photometric redshifts and spectral types
453: are measured by using template SEDs, shifting them in redshift steps and
454: performing $\chi^2$ fits to the observed SEDs at each step. The templates
455: corresponding to the minimum $\chi^2$ values are selected, with its
456: spectral type and redshift assigned to the galaxy in question. We use the
457: observed templates for elliptical, Sbc, Scd and Im-type galaxies from
458: \cite{coleman80} and the starburst SEDs from \cite{kinney96}. In order to
459: increase the resolution in spectral types, we also divide the interval
460: between adjacent spectral classes into two intermediate bins. For
461: instance, between the two discrete spectral types for ellipticals
462: ($T\!=\!1$) and Sbc-type spectra ($T\!=\!2$) we include two discrete
463: sub-categories ($T=1\onethird, 1 \frac{2}{3}$) that are linear mixtures
464: between the two boundary types. This allows a more continuous spectral
465: classification across the Hubble types while not causing degeneracy. The
466: choice of steps of 1/3 unit was made as a compromise between the goal to
467: increase the resolution of spectral types on one side and the goal to
468: avoid creating degeneracies in the phot-z routine on the other. Details of
469: the photometric redshift and spectral-type measurement are presented in
470: \cite{mobasher04, mobasher07}.~We plot the distribution of the discrete
471: spectral types in Figure~\ref{ps:spectype} and define all objects that
472: were assigned slightly earlier spectral types ($T\!=\!1\onethird$) than
473: passively evolving galaxies ($T\!=\!1$) as yE candidates, i.e.~objects
474: which are bluer than normal ellipticals but redder than normal early-type
475: spirals.~Their location in Figure~\ref{ps:spectype} is marked by a hatched
476: histogram.~Because later galaxy types can exhibit similar $BJK$ colors as
477: genuine yE galaxies, we use the spectral-type information to exclude
478: late-type star-forming galaxies and to include only galaxy candidates with
479: early spectral types in the following analysis.~The distribution of
480: photometric redshifts of all objects is shown in Figure~\ref{ps:redshift}.
481: Most of our yE candidates have redshifts between $z\approx0.2$ and $0.8$,
482: with few objects at slightly higher redshifts $z\approx1.0$.
483:
484: \begin{figure}[!t]
485: \centering
486: \includegraphics[bb=0 0 400 400, width=8cm]{fig5.pdf}
487: \caption{Distribution of spectral types $T$ that were assigned by the photometric
488: redshift routine for all objects in our selected GOODS/CDF-S sample
489: \citep[for details on the photometric redshift fitting routine we refer the reader
490: to][]{mobasher04, mobasher07}.~The locations of individual spectral types are labeled.~A
491: sub-population of several objects have a slightly earlier spectral type ($T\!=\!1\onethird$)
492: than passively evolving galaxies ($T\!=\!1$).~These are our yE candidates
493: and are marked in the plot with a hatched bin.}
494: \label{ps:spectype}
495: \end{figure}
496:
497: \begin{figure}[!th]
498: \centering
499: \includegraphics[bb=0 0 400 400, width=8cm]{fig6.pdf}
500: \caption{Distribution of photometric redshifts for all selected objects
501: in our GOODS/CDF-S photometric dataset \citep[for details
502: on the photometric redshift routine see][]{mobasher04}. The parametrization
503: by spectral-type is identical to the one in Figure~\ref{ps:spectype}.}
504: \label{ps:redshift}
505: \end{figure}
506:
507: \begin{figure*}[!t]
508: \centering
509: \includegraphics[bb=0 0 400 400, width=5.69cm]{fig7a.pdf}
510: \includegraphics[bb=0 0 400 400, width=5.69cm]{fig7b.pdf}
511: \includegraphics[bb=0 0 400 400, width=5.69cm]{fig7c.pdf}
512: \caption{Distribution of restframe $B$-band CAS parameters \citep[concentration, asymmetry, and smoothness, see][]{conselice03} taken from \cite{ravi06} for all selected objects in our GOODS/CDF-S photometric dataset parameterized by their spectral type as in Figure~\ref{ps:spectype}. Note the strong separation of yE candidates in the concentration parameter ({\it left panel}) from galaxies with later spectral types.}
513: \label{ps:morph}
514: \end{figure*}
515:
516:
517: \subsection{Morphologies}
518: To investigate the yE candidate morphologies we cross-correlate the
519: GOODS/CDF-S sample with the structural parameter classification of
520: \cite{ravi06}.~Based on the $BViz$ HST/ACS images the authors computed CAS
521: parameters ('asymmetry', 'concentration', and 'smoothness' defined in
522: \citealt{conselice03}) for all our sample sources.~Because of the redshift
523: distribution of our sources each ACS filter probes a different wavelength
524: and therefore a different morphological (stellar) component in a
525: galaxy.~To allow a comparison of structural parameters that is independent
526: of redshift we compute the redshift-corrected restframe $B$-band
527: morphologies by linearly interpolating the CAS parameters between the
528: pivot wavelengths of each ACS filter \citep[see][]{sirianni05} according
529: to the redshift of each source.~In Figure~\ref{ps:morph} we show the
530: distribution of restframe $B$-band CAS parameters for all sample galaxies
531: sub-divided by spectral-type as in Figure~\ref{ps:spectype}.~We find that
532: the overall spectral-type classification of our yE candidates
533: is consistent with early-type morphologies, which confirms the robustness
534: of our procedure to select early-type post-starburst galaxies. In
535: particular, the distribution of the CAS 'concentration' parameter
536: indicates that very few yE candidate galaxies with later-type
537: morphologies were selected by our photometric selection.
538:
539: \subsection{$BJK$ Color-Color Diagram}
540: The left panel of Figure~\ref{ps:bjk} shows the GOODS/CDF-S data in the $BJK$
541: color-color plane. Different symbols depict different spectral types as
542: assigned to each object by the photometric redshift fitting routine.~Most
543: objects with SEDs consistent with passive early-type galaxies clump around
544: restframe $(B\!-\!J)_{\rm AB}\approx1.9$ and $(J\!-\!K)_{\rm AB}\approx0$.
545: Objects with slightly later SED types (our yE candidates) scatter towards
546: bluer $B\!-\!J$ colors and have intermediate $J\!-\!K$ colors that place
547: their mean locus between spiral and star-forming galaxies.~{\it We find
548: that the spectral-type selection is consistent with the $BJK$ color
549: selection and morphological classification.} Moreover, most yE candidates
550: are consistent with the previously computed composite stellar population
551: models for post-starburst galaxies.~Only one yE candidate at
552: $(B\!-\!J)_{\rm AB}\!\approx\!1.6$ and $(J\!-\!K)_{\rm AB}\!\approx\!0.35$
553: deviates significantly from the model predictions and might be a dusty
554: starburst.~For reference, we plot the control sample of spectroscopically
555: confirmed K+As (open symbols).
556:
557: To provide a comparison independent of SSP model predictions we show in
558: the right panel of Figure~\ref{ps:bjk} nearby early-type \citep[taken
559: from][]{michard05} and late-type galaxies \citep[taken from][]{perez03a}
560: in the same color-color plane as the GOODS/CDF-S data.~We find a good
561: agreement between the early-type galaxies in GOODS/CDF-S and the local
562: galaxy sample. However, the smaller photometric errors of the local
563: galaxy sample allow a more clearly detected offset of the
564: spectroscopically confirmed K+A galaxies towards bluer $B\!-\!J$ colors
565: (i.e., younger ages) compared to the local early-type galaxies, which line
566: up in a tight sequence. We also notice that fewer spiral (squares) and Irr
567: galaxies (triangles) lie outside of the SSP grid which is likely again,
568: partly due to smaller photometric errors of the nearby sample.~We
569: investigate whether this disagreement may be due to systematics in the
570: photometric redshift determination, but find no differences in the error
571: distributions between spiral and irregular galaxies, in particular in
572: $J\!-\!K$ that appears to be primarily responsible for the scatter.~A
573: matching between the photometric redshifts of our sample and currently
574: available spectroscopic redshifts from the literature \citep{lefevre04}
575: results in good agreement (see Table~\ref{photspecz}).~Although we compare
576: restframe colors in both samples, the galaxies in our GOODS/CDF-S data are
577: more distant than the nearby comparison sample. Evolutionary factors, such
578: as possibly enhanced dust fractions, may be responsible for the enhanced
579: scatter in our data. Deep mid-IR imaging and optical to infrared
580: spectroscopy should be able to resolve this issue.
581:
582:
583: \begin{deluxetable}{lrrr}
584: \tabletypesize{\scriptsize}
585: \tablecaption{Comparison between photometric and spectroscopic
586: redshifts. \label{photspecz}}
587: \tablewidth{0pt}
588: \centering
589: \tablehead{
590: \colhead{subsample} &
591: \colhead{$\langle\Delta z\rangle$\tablenotemark{a}} &
592: \colhead{$\sigma$} & \colhead{$N$}
593: }
594: \startdata
595: E &$-0.109\pm 0.004$ & 0.061 & 14\\
596: yE &$-0.119\pm 0.012$ & 0.083 & 7\\
597: Sab&$ 0.028\pm 0.004$ & 0.134 & 37\\
598: Scd&$-0.061\pm 0.046$ & 0.185 & 4\\
599: Starburst&$ 0.224\pm0.000$ & \dots & 1\\
600: \enddata
601: \tablenotetext{a}{$\langle\Delta z\rangle=\langle z_{\rm phot} - z_{\rm spec}\rangle$.}
602:
603: \end{deluxetable}
604:
605: \begin{figure*}[!th]
606: \centering
607: \includegraphics[bb=0 0 400 400, width=8.9cm]{fig8a.pdf}
608: \includegraphics[bb=0 0 400 400, width=8.9cm]{fig8b.pdf}
609: \caption{{\it Left panel}: $BJK$ color-color diagram (using de-reddened
610: rest-frame AB magnitudes) of galaxies in the GOODS/CDF-S Field. The data
611: symbols are parameterized by the galaxy spectral types, and are early-type
612: ({\it filled circles}), spiral ({\it filled squares}), irregular ({\it
613: filled triangles}), and starbursts ({\it stars}).~yE galaxy
614: candidates are marked by {\it small dark dots}. Spectroscopically
615: confirmed K+A galaxies taken from \cite{galaz00} are indicated by open
616: symbols, where open squares mark cluster K+A galaxies and open circles are
617: field K+As. Average total photometric errors, including uncertainties of the photometry and the photometric redshift determination, are indicated in the upper right
618: corner for yE candidates ({\it solid cross}) and the mean photometric uncertainties for confirmed K+A galaxies ({\it dotted cross}). {\it Right panel}: $BJK$ color-color
619: diagram for local galaxies. Open circles are nearby ellipticals from
620: \cite{michard05}, open squares and triangles are Sab and Scd spirals,
621: while stars show local starburst galaxies, both taken from \cite{perez03a}. The
622: same spectroscopically confirmed K+A sample as in the left panel is
623: overplotted for comparison. The solid error bar shows average
624: uncertainties for the elliptical sample, while the dotted cross is for the
625: later-type galaxies. SSP models in both panels are identical to those in
626: the left panel of Figure~\ref{ps:sspburst}.
627: }
628: \label{ps:bjk}
629: \end{figure*}
630:
631: We define the population of yE galaxy candidates as those objects that
632: have slightly later spectral types ($T\!=\!1\onethird$, see
633: Fig.~\ref{ps:spectype}) and $BJK$ colors that are inconsistent with
634: passively evolving (reddest) early-type galaxies.~The selected yE
635: candidates are marked in Figure~\ref{ps:bjk} by red circles with central
636: black dots.~This definition might exclude some genuine K+As since the
637: control sample contains some spectroscopically confirmed K+A galaxies (in
638: galaxy clusters) with $BJK$ colors consistent with passively evolving
639: early-type galaxies.~Taking into account the average photometric error,
640: this fraction is expected to be less than $\sim\!15$\%. We point out that
641: even with the color, spectral-type, and morphology selection, our yE
642: candidate sample may be contaminated by early-type spiral galaxies that
643: can mimic the morphologies of elliptical galaxies (e.g., when in a largely
644: face-on configuration) and the SEDs of yE candidates because of their low,
645: but continuous star formation rates. Based on the local comparison sample
646: we estimate this contamination fraction between $\sim\!10\!-\!30\%$,
647: depending on the exact color selection. Again, deep follow-up spectroscopy
648: should be able to address this contamination fraction.
649:
650: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
651: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
652:
653: \section{Discussion}
654:
655: \subsection{Color-Magnitude Diagrams}
656: Post-starburst galaxies tend to be brighter in the optical than their
657: passive counterparts \citep[e.g.][]{tran03}, which is likely due to their
658: additional $\sim\!1$ Gyr old stellar population. However, optical
659: photometry is not able to provide strong constraints on how the stellar
660: masses of these rejuvenated galaxies compare with those of passively
661: evolving galaxies.~The $K$-band stellar mass-to-light ratio changes by a
662: factor of $\sim\!3$ from $\sim\!1.5$ Gyr to 13 Gyr \citep[see
663: also][]{drory04}, which is about an order of magnitude smaller a change
664: than in the optical.~$M/L_{K}$ changes by about a factor 1.5 between 500
665: Myr and 1.5 Gyr (the K+A phase).~Hence, relative to optical colors a
666: $M_{K}$ selection defines the best approximation to a stellar-mass
667: selected dataset.
668:
669: In Figure~\ref{ps:cmd} we show color-magnitude (CM) diagrams for all
670: early-type galaxies ($T\!\leq\!1.5$) and the selected yE candidates as
671: well as the K+A control sample and nearby ellipticals.~Overplotted are
672: color-magnitude relations for early-type galaxies taken from the large
673: galaxy survey of \cite{mobasher86}, in good agreement with the GOODS data
674: and the nearby elliptical galaxy sample.~We find no significant difference
675: between red-sequence galaxies and yE candidates in the near-infrared
676: $J\!-\!K$ vs.~$M_{K}$ diagram, but we detect a significant offset in the
677: $B\!-\!J$ vs.~$M_{K}$ diagram where the yE candidates have significantly
678: bluer colors at the same $M_{K}$ (stellar mass) compared to the red
679: sequence.
680:
681: We show the influence of changing luminosity-weighted age ($\Delta t$),
682: metallicity ($\Delta Z$), total galaxy mass ($\Delta M_{\rm gal.}$), and
683: starburst-mass fraction ($\Delta M_\ast$) as vectors in both CM diagrams.
684: $\Delta t$ and $\Delta M_{\rm gal.}$, and $\Delta Z$ and $\Delta M_\ast$
685: are highly degenerate in the $J\!-\!K$ vs.~$M_{K}$ diagram. In fact, the
686: $\Delta t$ and $\Delta M_{\rm gal.}$ vectors are almost parallel to the
687: red sequence (see upper panel in Fig.~\ref{ps:cmd}) and have, therefore,
688: negligible influence on the scatter in the $J\!-\!K$ color, which is
689: mainly driven by metallicity and the starburst-mass fraction.~The
690: relatively small scatter in this diagram hints at a small scatter in
691: metallicity and starburst-mass fraction and/or is a signature of a
692: correlation between these parameters.~This degeneracy can be partly lifted
693: with the $B\!-\!J$ vs. $M_{K}$ diagram (see below).~Four yE
694: candidates fall significantly off the red sequence in the $J\!-\!K$ vs.
695: $M_{K}$ diagram.~They are likely to have experienced significantly
696: different starburst events than the rest of the sample.~One possible
697: explanation is that the metallicity of their starburst population is
698: significantly lower.
699:
700: The most influential parameters in the $B\!-\!J$ vs.~$M_{K}$ diagram are
701: age and the starburst mass fraction (besides the total galaxy mass).~Both
702: vectors are highly inclined relative to the red sequence, which is a
703: graphic explanation of the scatter in this plot.~Metallicity plays a
704: negligible role. We use the $B\!-\!J$ vs.~$M_{K}$ diagram to discuss the
705: redshift evolution of our yE candidates in the next section.
706:
707: \begin{figure}[!ht]
708: \centering
709: \includegraphics[bb=0 0 400 400, width=8.5cm]{fig9.pdf}
710: \caption{Optical/near-infrared color-magnitude diagrams of early-type galaxies ({\it
711: filled circles}) and yE candidates ({\it dotted circles}).~Large open (magenta) symbols mark spectroscopically confirmed K+As \citep{galaz00} in the field ({\it circles}) and
712: cluster environment ({\it squares}).~Small open (red) circles are nearby elliptical galaxies from \cite{michard05}, as in the right panel of Figure~\ref{ps:bjk}. The nearby data were corrected for passive evolution, according to \cite{bruzual03} models, to the reference redshift $z=0.5$ of our GOODS data, which includes only a correction of $\Delta(B-J)=0.2$ mag.~Arrows indicate the relative changes in $\Delta t$ between 500 Myr and 1.5 Gyr (short arrow), and 500
713: Myr and 13 Gyr (long arrow), $\Delta Z$ between $0.2Z$ and solar
714: metallicity, $\Delta M_{\rm gal.}$~between $10^{10}$ and
715: $10^{11}M_{\odot}$, and $\Delta M_{\ast}$ between 5\% and 50\% of the
716: starburst mass fraction.~Color-magnitude relations for early-type galaxies from
717: the survey of \cite{mobasher86} are overplotted as dashed lines.~Average
718: photometric errors of yE candidates ({\it solid cross}) and
719: spectroscopically confirmed K+As ({\it dotted cross}) are plotted in the
720: upper right corner of each panel.}
721: \label{ps:cmd}
722: \end{figure}
723:
724:
725: \subsection{Starburst Mass Evolution}
726: In the hierarchical merging scenario of galaxy formation, massive
727: structures are expected to form on extended timescales from smaller
728: sub-units with a considerable fraction of stars forming relatively
729: recently \citep[e.g.][]{springel05}.~If, in contrast, more massive
730: structures form first, as seen in the monolithic collapse scenario, the
731: majority of stars form at high redshifts ($z\ga2$) in intense starbursts
732: \citep[e.g.][]{larson75}.~The fraction of yE signatures among massive
733: early-type galaxies is a measure of dissipative merging activity (recently
734: often named ``wet merging''), which is expected to be a function of
735: redshift in the hierarchical merging picture. In contrast, no such
736: evolution is expected in the early monolithic collapse scenario out to
737: $z\!\sim\!2$.~This difference is expected to depend on environment and to
738: be more pronounced in the field \citep{benson02}, which we probe here with
739: the GOODS dataset.~The fraction of yE galaxies and the strength of the
740: yE signatures among massive early-type field galaxies as a function of
741: redshift therefore puts strong constraints on galaxy formation models.
742:
743: To investigate the redshift evolution of our yE candidates, we subdivide
744: the sample into four redshift bins between $z=0$ and 1. For each galaxy
745: we determine the $B\!-\!J$ offset with respect to the red sequence and
746: plot the residuals versus their absolute $K$-band luminosity in
747: Figure~\ref{ps:bjresid}.~Our analysis is limited in redshift space by the
748: completeness of our GOODS data in the $K$ band. A $10\sigma$ point source
749: detection in the VLT/ISAAC data is feasible down to $m_{K}({\rm AB})=25.1$
750: mag \citep{giavalisco04}, which translates into $M_{K, {\rm AB}}=-19.0$
751: mag at a redshift of $z=1$ (indicated by a vertical line in
752: Figure~\ref{ps:bjresid}).~Hence, completeness is not an issue for the
753: following analysis.
754:
755: \begin{figure}[!t]
756: \centering
757: \includegraphics[bb=0 0 400 400,width=8.5cm]{fig10.pdf}
758: \caption{Residual $B\!-\!J$ colors with respect to the red sequence (see
759: Fig.~\ref{ps:cmd}) as a function of absolute $K$ band magnitude for
760: early-type galaxies.~Dark dots mark yE candidates.~Open symbols indicate
761: spectroscopically confirmed K+A as in Fig.~\ref{ps:cmd}; taken from
762: \cite{galaz00}.~The symbols size
763: is parametrized by redshift.~All confirmed K+A galaxies are located at
764: redshifts $z\leq0.12$.~Vectors indicate the influence of varying age,
765: metallicity, and starburst mass fraction as in Fig.~\ref{ps:cmd}.~Average
766: photometric errors are indicated in the upper left corner of the panel,
767: where the dotted cross indicates typical errors of the confirmed K+As and
768: the solid cross shows errors for yE candidates.~A long vertical line
769: shows the $10\sigma$ points source detection limit at $z\!=\!1$.~Small
770: vertical lines indicate $L^{\ast}$ at redshifts 0.2 and 1.0 for field
771: galaxies \citep{drory03}.~Note the paucity of yE candidates at magnitudes
772: fainter than $M_{K, {\rm AB}}\approx-21.2$ and
773: $\delta(B\!-\!J)_{\rm AB}\la-0.8$ mag.}
774: \label{ps:bjresid}
775: \end{figure}
776:
777: We observe that at any given $M_{K}$ the $B\!-\!J$ residuals are
778: systematically larger for higher redshift galaxies.~Spectroscopically
779: confirmed K+As show a similar trend, but their redshift range is
780: significantly smaller ($z\leq0.12$) than that of our sample ($z\leq1$, see
781: Fig.~\ref{ps:redshift}).~Note that confirmed field K+As show larger
782: $\delta(B\!-\!J)$ than confirmed cluster K+As, which lie closer to the red
783: sequence.~We also find tentative evidence for a systematic difference in
784: $\langle\delta(J\!-\!K)\rangle$ between confirmed {\it field} K+As (open
785: circles) and the combination of our yE candidates and confirmed {\it
786: cluster} K+As. It appears that a metallicity difference is the most
787: obvious way to explain this offset, given that in the more age-sensitive
788: $B\!-\!J$ panel the yE candidates lie among, or are bluer than, the nearby
789: confirmed field and cluster K+As.~However, the photometric errors of the
790: spectroscopically confirmed K+A sample require that this intriguing result
791: be confirmed with deep spectroscopy.
792:
793: Parameters likely responsible for the evolution in $\delta(B\!-\!J)$ with
794: redshift are {\it (i)\/} a younger age of the post-starburst
795: population at higher redshift, or {\it (ii)\/} a larger mass fraction of
796: the post-starburst population at higher redshifts (we established in
797: Section~\ref{ln:ssp} that metallicity has a smaller effect on
798: $\delta(B\!-\!J)$). First we consider age effects. Under the assumption
799: that the yE candidates indeed host $0.5\!-\!1.5$ Gyr old post-starburst
800: populations with strong Balmer absorption-line spectra, the vectors in
801: Figure~\ref{ps:bjresid} show that the difference in $\delta(B\!-\!J)$
802: between yE candidates with $z\la0.4$ and those with $z\!>\!0.6$ is in
803: principle similar to the effect of age fading from 0.5 to 1.5 Gyr.
804: However, it is hard to imagine that post-starburst populations in yE
805: galaxies at higher redshift (be it induced by galaxy interactions or by
806: ram-pressure stripping) would be systematically younger than in
807: low-redshift ones, given the short duration of the yE phenomenon relative
808: to the difference in look-back times within this redshift range.~In the
809: adopted WMAP cosmology, the difference in look-back times between $z =
810: 0.2$ and 1.0 is $7.8 - 2.4 = 5.4$ Gyr, significantly longer than the
811: duration of the yE phenomenon.~Instead, one would expect the ages of such
812: post-starburst populations to be randomly distributed (between 0.5 and 1.5
813: Gyr) at any redshift.~We, therefore, suggest that it is more likely that
814: the increase of yE candidates' $\delta(B\!-\!J)$ with redshift is
815: primarily due to an {\it increasing mass fraction of the post-starburst
816: population}.~If indeed the yE phenomenon identifies an important era in
817: the assembly history of early-type galaxies, the lack of low-luminosity
818: ($M_K\!\ga\!-21$) yE candidates at $z > 0.6$ in our GOODS sample would
819: constitute strong evidence in favor of galaxy formation scenarios in which
820: more massive early-type galaxies in the field are assembled earlier than
821: their low-mass counterparts, which is in line with the ''downsizing''
822: picture \citep[e.g.][]{cowie96}.
823:
824: On the other hand, we currently cannot exclude the possibility that the
825: higher-redshift yE candidates do not actually host a (one)
826: post-starburst population, but instead have systematically younger overall
827: (luminosity-weighted) ages than the lower-redshift yE
828: candidates.~More frequent bursts at higher redshifts, due to more frequent
829: mergers, might be a possible explanation, which would lead to
830: systematically younger observed ages.~Given the relevance of this issue in
831: terms of its power to constrain early-type galaxy formation scenarios, we
832: suggest a spectroscopic followup to test whether our yE candidates
833: really do contain K+A features in their spectra.
834:
835:
836: \subsection{Evolution of the yE Fraction}
837: The fraction of post-starburst galaxies in nearby clusters was reported to
838: increase with decreasing galaxy luminosity \citep[e.g.][]{poggianti01,
839: smail01}.~The morphology bias prevents a direct comparison between the
840: yE fractions in the cluster and field environment.~We, therefore,
841: select only early-type galaxies from our sample and investigate the
842: yE fraction in the field.~A similar trend of an increasing post-starburst
843: fraction with decreasing galaxy luminosity is present in our GOODS sample
844: of field galaxies.~We quantify this trend by splitting the data at
845: $L^{\ast}/2$ along the pure luminosity evolution vector and counting the
846: number of passive Es and yE candidates.~The corresponding values for
847: the division for each redshift bin into super-$L^{\ast}/2$ and
848: sub-$L^{\ast}/2$ systems are $M^{\ast}_{K}=-22.27, -22.43, -22.64$, and
849: $-22.75$ mag for $z=0.3, 0.5, 0.7$, and $0.9$, respectively.~Note that the
850: samples are complete down to $M_B=-16.3$ mag and $M_{K}=-19.0$ mag at
851: $z\!=\!1$, which leaves our sample sensitive to objects with
852: restframe colors $B-K\la2.7$ mag, but excludes extremely reddened objects,
853: such as EROs \citep[e.g.][]{daddi00}. However, these galaxies are most
854: frequent beyond redshift unity \citep[e.g.][]{georgakakis06, simpson06}
855: and are expected not to be a significant part of our initial sample.~To
856: improve sample statistics we merge the number counts of the two upper and
857: two lower redshift bins, which results in a high and low-redshift sample
858: divided at $z=0.6$.~Galaxy number counts are given in
859: Table~\ref{tab:counts}.
860:
861: We confirm the trend of a higher fraction of post-starbursts among
862: early-type galaxies toward less luminous objects.~In the high-redshift
863: sample we find a yE candidate fraction of 86\% among all sub-$L^{\ast}/2$
864: early-type galaxies, whereas only 22\% of the more luminous (massive)
865: objects are yE candidates. The yE fractions shrink towards lower
866: redshifts to 51\% for sub-$L^{\ast}/2$ systems and 11\% for their brighter
867: counterparts.~We suggest that these high yE fractions be confirmed
868: (or refuted) using follow-up spectroscopy.
869:
870: \begin{deluxetable}{crrrrcc}
871: \tabletypesize{\scriptsize}
872: \tablecaption{Number counts of E galaxies and yE candidates. \label{tab:counts}}
873: \tablewidth{0pt}
874: \centering
875: \tablehead{
876: \colhead{} &
877: \multicolumn{2}{c}{sub-$L^{\ast}/2$} &
878: \multicolumn{2}{c}{$>L^{\ast}/2$} & \colhead{} & \colhead{}
879: \\
880: \colhead{redshift} &
881: \colhead{$N_{\rm E}$} &
882: \colhead{$N_{\rm yE}$} &
883: \colhead{$N_{\rm E}$} &
884: \colhead{$N_{\rm yE}$} &
885: \colhead{$f_{<L^{\ast}/2}$\tablenotemark{a}} &
886: \colhead{$f_{>L^{\ast}/2}$\tablenotemark{b}}
887: }
888: \startdata
889: $z>0.6$ & 2 &12& 7 & 2 & 0.86 & 0.22 \\
890: $z<0.6$ &18 &19& 16 & 2 & 0.51 & 0.11 \\
891: \enddata
892: \tablenotetext{a}{Fraction $f_{<L^{\ast}/2}$ of yE galaxy candidates
893: among all sub-$L^{\ast}/2$ early-type galaxies.}
894: \tablenotetext{b}{Fraction $f_{>L^{\ast}/2}$ of yE galaxy candidates
895: among all $>\!L^{\ast}/2$ early-type galaxies.}
896: \end{deluxetable}
897:
898: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
899:
900: \section{Conclusions}
901: Based on the combination of photometric redshifts, spectral-type
902: classification, and optical/near-infrared colors, we have selected a
903: sample of young early-type galaxy candidates (yE candidates) from
904: the GOODS/CDF-S dataset.~Our technique relies on spectral-type fitting and
905: color-color/color-magnitude diagnostic diagrams and provides an efficient
906: method for selecting yE candidates using {\it only} photometric
907: information.~An analysis of CAS parameters shows that the selected
908: yE candidates have early morphological types, which confirms that
909: spectral-type and morphological-type selection are fully consistent with
910: each other.~The comparison of population synthesis models and observed
911: properties of a sample of spectroscopically confirmed K+A galaxies
912: provides strong circumstantial evidence that the selected candidates are
913: genuine field early-type post-starburst galaxies.
914:
915: We study the systematics that drive colors and magnitudes of
916: post-starburst galaxies in diagnostic diagrams using current population
917: synthesis models.~Our analysis reveals evidence for a changing starburst
918: mass fraction with increasing redshift in the sense that more
919: massive/intense starbursts may be responsible for the yE signatures
920: of massive field early-type galaxies at higher redshifts.~Furthermore, we
921: find a higher yE candidate fraction in sub-$L^{\ast}/2$ early-type
922: galaxies compared to their more luminous counterparts.~Within the redshift
923: range of our data ($z\!\la\!1$) we may be witnessing evidence for enhanced
924: merging in the field towards higher redshifts, especially for
925: high-luminosity galaxies.~Similar results are obtained by studies of
926: luminosity-weighted mean ages of nearby early-type galaxies which indicate
927: that low-$L$ galaxies show a much greater scatter to younger
928: luminosity-weighted mean ages than high-$L$ galaxies
929: \citep[e.g.][]{caldwell03, perez03b, sanchez06}.~If follow-up spectroscopy
930: of the yE candidates identified in this paper confirms their K+A
931: nature, this suggests that the assembly of stars in high-$L$ early-type
932: field galaxies occurred earlier than in their lower-$L$ counterparts.
933:
934: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
935:
936: \acknowledgments
937:
938: We thank Peter M. Pessev for his help in providing relevant near-IR filter
939: passband information. We are grateful to the referee, James Rose, for his
940: insightful comments that improved the content of this paper.~THP
941: gratefully acknowledges support in form of a Plaskett Fellowship at the
942: Herzberg Institute of Astrophysics of the National Research Council of
943: Canada. He also acknowledges support through an ESA Research Fellowship
944: and partial financial support through grant GO-10129 from the Space
945: Telescope Science Institute, which is operated by AURA, Inc.,~under NASA
946: Contract NAS5-26555.~This publication makes use of data products from the
947: Two Micron All Sky Survey, which is a joint project of the University of
948: Massachusetts and the Infrared Processing and Analysis Center/California
949: Institute of Technology, funded by the National Aeronautics and Space
950: Administration and the National Science Foundation.
951:
952: \begin{thebibliography}{}
953:
954: \bibitem[Abraham et al.(1996)]{abraham96} Abraham, R.~G., van den
955: Bergh, S., Glazebrook, K., Ellis, R.~S., Santiago, B.~X., Surma, P., \&
956: Griffiths, R.~E.\ 1996, \apjs, 107, 1
957: \bibitem[Anders \& Fritze-v.~Alvensleben(2003)]{anders03}
958: Anders, P., \& Fritze-v.~Alvensleben, U.\ 2003, \aap, 401, 1063
959: \bibitem[Balogh et al.(2005)]{balogh05} Balogh, M.~L., Miller,
960: C., Nichol, R., Zabludoff, A., \& Goto, T.\ 2005, \mnras, 360, 587
961: \bibitem[Barger et al.(1996)]{barger96} Barger, A.~J.,
962: Aragon-Salamanca, A., Ellis, R.~S., Couch, W.~J., Smail, I., \& Sharples,
963: R.~M.\ 1996, \mnras, 279, 1
964: \bibitem[Bekki \& Couch(2003)]{bekki03} Bekki, K., \& Couch,
965: W.~J.\ 2003, \apjl, 596, L13
966: \bibitem[Bekki et al.(2005)]{bekki05} Bekki, K., Couch, W.~J.,
967: Shioya, Y., \& Vazdekis, A.\ 2005, \mnras, 359, 949
968: \bibitem[Belloni et al.(1995)]{belloni95} Belloni, P., Bruzual,
969: A.~G., Thimm, G.~J., \& Roser, H.-J.\ 1995, \aap, 297, 61
970: \bibitem[Benson et al.(2002)]{benson02} Benson, A.~J., Ellis,
971: R.~S., \& Menanteau, F.\ 2002, \mnras, 336, 564
972: \bibitem[Bessell \& Brett(1988)]{bb88} Bessell, M.~S., \&
973: Brett, J.~M.\ 1988, \pasp, 100, 1134
974: \bibitem[Bertin \& Arnouts(1996)]{bertin96} Bertin, E., \&
975: Arnouts, S.\ 1996, \aaps, 117, 393
976: \bibitem[Bruzual \& Charlot(2003)]{bruzual03} Bruzual, G., \&
977: Charlot, S.\ 2003, \mnras, 344, 1000
978: \bibitem[Bothun \& Dressler(1986)]{bothun86} Bothun, G.~D.~\&
979: Dressler, A.\ 1986, \apj, 301, 57
980: \bibitem[Bundy et al.(2005)]{bundy05} Bundy, K., Ellis, R.~S.,
981: \& Conselice, C.~J.\ 2005, \apj, 625, 621
982: \bibitem[Byrd \& Valtonen(1990)]{byrd90} Byrd, G., \&
983: Valtonen, M.\ 1990, \apj, 350, 89
984: \bibitem[Caldwell \& Rose(1997)]{caldwell97} Caldwell, N., \&
985: Rose, J.~A.\ 1997, \aj, 113, 492
986: \bibitem[Caldwell et al.(1999)]{caldwell99} Caldwell, N., Rose,
987: J.~A., \& Dendy, K.\ 1999, \aj, 117, 140
988: \bibitem[Caldwell et al.(2003)]{caldwell03} Caldwell, N., Rose,
989: J.~A., \& Concannon, K.~D.\ 2003, \aj, 125, 2891
990: \bibitem[Coleman et al.(1980)]{coleman80} Coleman, G.~D., Wu,
991: C.-C., \& Weedman, D.~W.\ 1980, \apjs, 43, 393
992: \bibitem[Conselice(1997)]{conselice97} Conselice, C.~J.\ 1997,
993: \pasp, 109, 1251
994: \bibitem[Conselice(2003)]{conselice03} Conselice, C.~J.\ 2003,
995: \apjs, 147, 1
996: \bibitem[Couch \& Sharples(1987)]{couch87} Couch, W.~J.~\&
997: Sharples, R.~M.\ 1987, \mnras, 229, 423
998: \bibitem[Cowie et al.(1996)]{cowie96} Cowie, L.~L., Songaila,
999: A., Hu, E.~M., \& Cohen, J.~G.\ 1996, \aj, 112, 839
1000: \bibitem[Daddi et al.(2000)]{daddi00} Daddi, E., Cimatti, A.,
1001: \& Renzini, A.\ 2000, \aap, 362, L45
1002: \bibitem[Daddi et al.(2004)]{daddi04} Daddi, E., Cimatti, A.,
1003: Renzini, A., Fontana, A., Mignoli, M., Pozzetti, L., Tozzi, P., \&
1004: Zamorani, G.\ 2004, \apj, 617, 746
1005: \bibitem[Dressler \& Gunn(1983)]{dressler83} Dressler, A.~\& Gunn,
1006: J.~E.\ 1983, \apj, 270, 7
1007: \bibitem[Dressler et al.(1999)]{dressler99} Dressler, A., Smail,
1008: I., Poggianti, B.~M., Butcher, H., Couch, W.~J., Ellis, R.~S., \& Oemler,
1009: A.~J.\ 1999, \apjs, 122, 51
1010: \bibitem[Drory et al.(2003)]{drory03} Drory, N., Bender, R.,
1011: Feulner, G., Hopp, U., Maraston, C., Snigula, J., \& Hill, G.~J.\ 2003,
1012: \apj, 595, 698
1013: \bibitem[Drory et al.(2004)]{drory04} Drory, N., Bender, R.,
1014: Feulner, G., Hopp, U., Maraston, C., Snigula, J., \& Hill, G.~J.\ 2004,
1015: \apj, 608, 742
1016: \bibitem[Ferraro et al.(2000)]{ferraro00} Ferraro, F.~R.,
1017: Montegriffo, P., Origlia, L., \& Fusi Pecci, F.\ 2000, \aj, 119, 1282
1018: \bibitem[Franx(1993)]{franx93} Franx, M.\ 1993, \apjl, 407, L5
1019: \bibitem[Galaz(2000)]{galaz00} Galaz, G.\ 2000, \aj, 119, 2118
1020: \bibitem[Georgakakis et al.(2006)]{georgakakis06} Georgakakis, A.,
1021: Hopkins, A.~M., Afonso, J., Sullivan, M., Mobasher, B., \& Cram, L.~E.\
1022: 2006, \mnras, 367, 331
1023: \bibitem[Giavalisco et al.(2004)]{giavalisco04} Giavalisco, M., et
1024: al.\ 2004, \apjl, 600, L93
1025: \bibitem[Goto(2004)]{goto04} Goto, T.\ 2004, \aap, 427, 125
1026: \bibitem[Goudfrooij et al.(2001)]{goudfrooij01} Goudfrooij, P.,
1027: Alonso, M.~V., Maraston, C., \& Minniti, D.\ 2001, \mnras, 328, 237
1028: \bibitem[Gunn \& Gott(1972)]{gunn72} Gunn, J.~E., \& Gott,
1029: J.~R.~I.\ 1972, \apj, 176, 1
1030: \bibitem[Hempel \& Kissler-Patig(2004)]{hempel04} Hempel, M., \&
1031: Kissler-Patig, M.\ 2004, \aap, 428, 459
1032: \bibitem[Kauffmann \& Charlot(1998)]{kauffmann98} Kauffmann, G.~\&
1033: Charlot, S.\ 1998, \mnras, 294, 705
1034: \bibitem[Kinney et al.(1996)]{kinney96} Kinney, A.~L., Calzetti,
1035: D., Bohlin, R.~C., McQuade, K., Storchi-Bergmann, T., \& Schmitt, H.~R.\
1036: 1996, \apj, 467, 38
1037: \bibitem[Larson(1975)]{larson75} Larson, R.~B.\ 1975, \mnras,
1038: 173, 671
1039: \bibitem[Le F{\`e}vre et al.(2004)]{lefevre04} Le F{\`e}vre, O.,
1040: et al.\ 2004, \aap, 428, 1043
1041: \bibitem[Maraston(1998)]{maraston98} Maraston, C.\ 1998, \mnras,
1042: 300, 872
1043: \bibitem[Maddox et al.(1990)]{maddox90} Maddox, S.~J.,
1044: Efstathiou, G., Sutherland, W.~J., \& Loveday, J.\ 1990, \mnras, 243, 692
1045: \bibitem[Michard(2005)]{michard05} Michard, R.\ 2005, \aap, 441,
1046: 451
1047: \bibitem[Miller \& Owen(2001)]{miller01} Miller, N.~A., \& Owen,
1048: F.~N.\ 2001, \apjl, 554, L25
1049: \bibitem[Mobasher et al.(1986)]{mobasher86} Mobasher, B., Ellis,
1050: R.~S., \& Sharples, R.~M.\ 1986, \mnras, 223, 11
1051: \bibitem[Mobasher et al.(2004)]{mobasher04} Mobasher, B., et al.\
1052: 2004, \apjl, 600, L167
1053: \bibitem[Mobasher et al.(2007)]{mobasher07} Mobasher, B., et al.\
1054: 2007, ApJS in press, astro-ph/0612344
1055: \bibitem[Moore et al.(1996)]{moore96} Moore, B., Katz, N.,
1056: Lake, G., Dressler, A., \& Oemler, A.\ 1996, \nat, 379, 613
1057: \bibitem[Moore et al.(1998)]{moore98} Moore, B., Lake,
1058: G., \& Katz, N.\ 1998, \apj, 495, 139
1059: \bibitem[Mouhcine \& Lan{\c c}on(2003)]{mouhcine03} Mouhcine, M.,
1060: \& Lan{\c c}on, A.\ 2003, \aap, 402, 425
1061: \bibitem[Newberry et al.(1990)]{newberry90} Newberry, M.~V.,
1062: Boroson, T.~A., \& Kirshner, R.~P.\ 1990, \apj, 350, 585
1063: \bibitem[Oke(1974)]{oke74} Oke, J.~B.\ 1974, \apjs, 27, 21
1064: \bibitem[P{\'e}rez-Gonz{\'a}lez et al.(2003a)]{perez03a}
1065: P{\'e}rez-Gonz{\'a}lez, P.~G., Gil de Paz, A., Zamorano, J., Gallego, J.,
1066: Alonso-Herrero, A., \& Arag{\'o}n-Salamanca, A.\ 2003a, \mnras, 338, 508
1067: \bibitem[P{\'e}rez-Gonz{\'a}lez et al.(2003b)]{perez03b}
1068: P{\'e}rez-Gonz{\'a}lez, P.~G., Gil de Paz, A., Zamorano, J., Gallego, J.,
1069: Alonso-Herrero, A., \& Arag{\'o}n-Salamanca, A.\ 2003b, \mnras, 338, 525
1070: \bibitem[Persson et al.(1983)]{persson83} Persson, S.~E.,
1071: Aaronson, M., Cohen, J.~G., Frogel, J.~A., \& Matthews, K.\ 1983, \apj,
1072: 266, 105
1073: \bibitem[Persson et al.(1998)]{persson98} Persson, S.~E., Murphy,
1074: D.~C., Krzeminski, W., Roth, M., \& Rieke, M.~J.\ 1998, \aj, 116, 2475
1075: \bibitem[Poggianti et al.(1999)]{poggianti99} Poggianti, B.~M.,
1076: Smail, I., Dressler, A., Couch, W.~J., Barger, A.~J., Butcher, H., Ellis,
1077: R.~S., \& Oemler, A.~J.\ 1999, \apj, 518, 576
1078: \bibitem[Poggianti \& Wu(2000)]{poggianti00} Poggianti, B.~M., \&
1079: Wu, H.\ 2000, \apj, 529, 157
1080: \bibitem[Poggianti et al.(2001)]{poggianti01} Poggianti, B.~M., et
1081: al.\ 2001, \apj, 563, 118
1082: \bibitem[Poggianti et al.(2004)]{poggianti04} Poggianti, B.~M.,
1083: Bridges, T.~J., Komiyama, Y., Yagi, M., Carter, D., Mobasher, B., Okamura,
1084: S., \& Kashikawa, N.\ 2004, \apj, 601, 197
1085: \bibitem[Puzia et al.(2002)]{puzia02} Puzia, T.~H., Zepf,
1086: S.~E., Kissler-Patig, M., Hilker, M., Minniti, D., \& Goudfrooij, P.\ 2002,
1087: \aap, 391, 453
1088: \bibitem[Quintero et al.(2004)]{quitnero04} Quintero, A.~D., et
1089: al.\ 2004, \apj, 602, 190
1090: \bibitem[Ravindranath et al.(2006)]{ravi06} Ravindranath, S.,
1091: et al.\ 2006, ArXiv Astrophysics e-prints, arXiv:astro-ph/0606696
1092: \bibitem[Renzini(1981)]{renzini81} Renzini, A.\ 1981, Ann. Phys., 6, 87
1093: \bibitem[Rose et al.(2001)]{rose01} Rose, J.~A., Gaba, A.~E.,
1094: Caldwell, N., \& Chaboyer, B.\ 2001, \aj, 121, 793
1095: \bibitem[S{\'a}nchez-Bl{\'a}zquez et al.(2006)]{sanchez06}
1096: S{\'a}nchez-Bl{\'a}zquez, P., Gorgas, J., Cardiel, N., \& Gonz{\'a}lez,
1097: J.~J.\ 2006, \aap, 457, 809
1098: \bibitem[Schlegel et al.(1998)]{schlegel98} Schlegel, D.~J.,
1099: Finkbeiner, D.~P., \& Davis, M.\ 1998, \apj, 500, 525
1100: \bibitem[Shioya et al.(2002)]{shioya02} Shioya, Y., Bekki, K.,
1101: Couch, W.~J., \& De Propris, R.\ 2002, \apj, 565, 223
1102: \bibitem[Shioya et al.(2004)]{shioya04} Shioya, Y., Bekki, K.,
1103: \& Couch, W.~J.\ 2004, \apj, 601, 654
1104: \bibitem[Simpson et al.(2006)]{simpson06} Simpson, C., et al.\
1105: 2006, \mnras, 373, L21
1106: \bibitem[Sirianni et al.(2005)]{sirianni05} Sirianni, M., et al.\
1107: 2005, \pasp, 117, 1049
1108: \bibitem[Smail et al.(1999)]{smail99} Smail, I., Morrison, G.,
1109: Gray, M.~E., Owen, F.~N., Ivison, R.~J., Kneib, J.-P., \& Ellis, R.~S.\
1110: 1999, \apj, 525, 609
1111: \bibitem[Smail et al.(2001)]{smail01} Smail, I., Kuntschner,
1112: H., Kodama, T., Smith, G.~P., Packham, C., Fruchter, A.~S., \& Hook, R.~N.\
1113: 2001, \mnras, 323, 839
1114: \bibitem[Springel et al.(2005)]{springel05} Springel, V., et al.\
1115: 2005, \nat, 435, 629
1116: \bibitem[Tran et al.(2003)]{tran03} Tran, K.~H., Franx, M.,
1117: Illingworth, G., Kelson, D.~D., \& van Dokkum, P.\ 2003, \apj, 599, 865
1118: \bibitem[Tran et al.(2004)]{tran04} Tran, K.~H., Franx, M.,
1119: Illingworth, G.~D., van Dokkum, P., Kelson, D.~D., \& Magee, D.\ 2004,
1120: \apj, 609, 683
1121: \bibitem[Zabludoff et al.(1996)]{zabludoff96} Zabludoff, A.~I.,
1122: Zaritsky, D., Lin, H., Tucker, D., Hashimoto, Y., Shectman, S.~A., Oemler,
1123: A., \& Kirshner, R.~P.\ 1996, \apj, 466, 104
1124:
1125: \end{thebibliography}
1126:
1127: \end{document}