astro-ph0308541/ms.tex
1: %%%%%%%%%%%%%%%
2: % Define a switch for submission mode
3: %%%%%%%%%%%%%%%
4: 
5: \newif\ifsubmode
6: %\submodefalse
7: \submodetrue
8: 
9: %%%%%%%%%%%%%%%
10: % Preamble
11: %%%%%%%%%%%%%%%
12: 
13: \ifsubmode
14:   \documentclass[12pt,preprint]{aastex}
15: %\received{}
16:  %\revised{}
17:  %\accepted{July}
18:   \citestyle{aa}
19: 
20: \else
21:   %\documentclass[12pt,preprint2]{emulateapj}
22:   \font\sevenrm=cmr7 
23:   \def\kmm{{\sevenrm KMM}~}
24:   \citestyle{aa} 
25: \fi
26: 
27: \usepackage{graphicx}
28: %\usepackage{doublespace}
29: 
30: %%%%%%%%%%%%%%%                                                           
31: % User defined macros                                                     
32: %%%%%%%%%%%%%%%                                                           
33:                                                                           
34: \newcommand{\etal}{{et al.~}}                                             
35: \newcommand{\lta}{\lesssim}                                               
36: \newcommand{\gta}{\gtrsim}                                                
37: \newcommand{\kms}{\,\rm km\,s^{-1}}                                       
38: 
39: \newenvironment{inlinefigure}{
40: \def\@captype{figure}
41: \noindent\begin{minipage}{0.999\linewidth}\begin{center}}
42: {\end{center}\end{minipage}\smallskip}
43: 
44: \shorttitle{The $z \sim 4$ Lyman Break Galaxies: Colors and Theoretical Predictions}
45: \shortauthors{Idzi et al.}
46: 
47: %\received{2003 May 24}
48: \received{2003 May 24}
49: 
50: \begin{document}
51: 
52: \title{The $z \sim 4$ Lyman Break Galaxies: Colors and Theoretical Predictions\altaffilmark{5,6}}
53: 
54: \author{
55: Rafal Idzi\altaffilmark{2},
56: Rachel Somerville\altaffilmark{1},
57: Casey Papovich\altaffilmark{4}, 
58: Henry C. Ferguson\altaffilmark{1,2}, 
59: Mauro Giavalisco\altaffilmark{1},
60: Claudia Kretchmer\altaffilmark{2}, 
61: Jennifer Lotz\altaffilmark{3}
62: }
63: 
64: \altaffiltext{1}{Space Telescope Science Institute, 3700 San Martin Drive, 
65: Baltimore, MD  21218, USA; somerville@stsci.edu, ferguson@stsci.edu, mauro@stsci.edu}
66: 
67: \altaffiltext{2}{Department of Physics and Astronomy, The Johns Hopkins
68: University, 3400 N. Charles St., Baltimore, MD 21218; idzi@stsci.edu, claudia@pha.jhu.edu}
69: 
70: \altaffiltext{3}{Department of Astronomy and Astrophysics, University
71: of California, Santa Cruz, CA 95064; jlotz@scipp.ucsc.edu}
72: 
73: \altaffiltext{4}{Steward Observatory, The University of Arizona, 933 North
74:  Charry Avenue, Tuscon, AZ 85721; papovich@as.arizona.edu}
75:  
76: \altaffiltext{5}{Based on observations taken with the NASA/ESA Hubble
77: Space Telescope, which is operated by the Association of Universities
78: for Research in Astronomy, Inc.\ (AURA) under NASA contract
79: NAS5--26555}
80: 
81: \altaffiltext{6}{Based on observations collected at the European Southern Observatory,
82:  Chile (ESO Programmes 168.A-0485, 64.0-0643, 66.A-0572, 68.A-0544)}
83:  
84:  \begin{abstract}
85: We investigate several fundamental properties of $z \sim 4$ Lyman-break 
86: galaxies by comparing observations with the predictions of a
87: semi-analytic model based on the Cold Dark Matter theory of
88: hierarchical structure formation. We use a sample of $B_{435}$-dropouts 
89: from the Great Observatories Origins Deep Survey, and complement the ACS 
90: optical $B_{435}$, $V_{606}$, $i_{775}$, and $z_{850}$ data with the 
91: VLT ISAAC $J$, $H$, and $K_{s}$ observations. We extract $B_{435}$-dropouts 
92: from our semi-analytic mock catalog using the same color criteria and magnitude limits
93: that were applied to the observed sample. We find that the $i_{775} - K_{s}$ 
94: colors of the model-derived and observed $B_{435}$-dropouts are in good agreement. 
95: However, we find that the $i_{775}-z_{850}$ colors differ significantly,
96: indicating perhaps that either too little dust or an incorrect extinction curve have been used. 
97: Motivated by the reasonably good agreement between the model and observed data
98: we present predictions for the stellar masses, star formation rates, and ages for the $z \sim 4$ 
99: Lyman-break sample. We find that according to our model the color selection criteria used to 
100: select our $z \sim 4$ sample surveys $67 \%$ of all galaxies at this epoch down to $z_{850} < 26.5$. 
101: We find that our model predicts a $\sim 40 \%$ mass build-up between the $z \sim 4$ and $z \sim 3$ 
102: epochs for the UV rest-frame $L^{*}$ galaxies. Furthermore, according to our model, at least $50 \%$ 
103: of the total stellar mass resides in relatively massive UV-faint objects that fall below our observational 
104: detection limit. 
105: \end{abstract}
106: \keywords{galaxies: evolution --- galaxies: fundamental parameters --- galaxies: high-redshift --- galaxies: photometry --- cosmology: observations --- cosmology: theory}
107:  
108:  \section{Introduction}
109:  The "Lyman-break'' color selection technique has been shown to be a
110:  highly effective means of selecting galaxies at high redshift $z \gta 2$ \citep{SAG1999,MFD1996}.
111:  Fairly large samples of Lyman-break galaxies (LBGs) in the redshift
112:  range $2 \lta z \lta 3.5$ have been compiled, and the characteristics
113:  of these objects have been studied with much vigor over the past
114:  several years. Optical and NIR photometric colors have been used to
115:  ascertain various attributes of these high-redshift objects. For
116:  example, colors spanning the rest-frame Balmer break ($\lambda \sim
117:  4000$ \AA) have been used to constrain the range of possible stellar
118:  ages and masses of these galaxies through application of maximum-likelihood 
119:  techniques to synthesized stellar population models \citep{PDF2001,SSA2001}. 
120: 
121:  As well, many researchers have compared theoretical models of galaxy
122:  evolution with various aspects of the observations \citep{WSB2001,BCF1998}.
123:  \citet[][hereafter SPF2001]{SPF2001} have used semi-analytic models to
124:  perform detailed comparative studies on the $z \sim 3$ sample. In SPF2001
125:  it was shown that the luminosity function and rest-frame UV and UV-optical colors of
126:  $z \sim 3$ LBGs could be reproduced well by the models, implying that
127:  the ages and dust contents of the model galaxies were consistent with
128:  the observations. 
129:  
130:  In this letter, we compare the colors of observed $B_{435}$-dropouts, selected from ACS imaging 
131:  from the Great Observatories Origins Deep Survey (GOODS) and supplemented with 
132:  deep NIR ($JHK_{s}$) data from ISAAC on the VLT, with the colors of model-derived $B_{435}$-dropouts. 
133:  We then present model predictions  for physical properties of these objects, such as stellar masses,
134:  ages, and star formation rates. We also investigate selection effects on the stellar mass content 
135:  for the redshift that we sample, and we look at the stellar mass build-up between the $z \sim 4$ 
136:  and $z \sim 3$ epochs.
137: 
138:  All quoted magnitudes throughout the letter are in the AB magnitude system \citep{O1974}.
139: 
140:  \section{Observations and Measurements}\label{observations}
141: 
142:  We use three epochs of HST ACS observations of the Chandra Deep Field
143:  South (CDF-S) obtained as part of the GOODS program. The data consist
144:  of image mosaics in all four ACS bands $B_{435}$, $V_{606}$,
145:  $i_{775}$, and $z_{850}$ (the ACS F435W, F606W, F775W, F850LP filters,
146:  respectively) spanning wavelength range of $0.4$ to 1 $\micron$, and
147:  covering approximately a field of view of $10 \arcmin$ x $16 \arcmin$. 
148:  The specifics of data acquisition and reduction, as well as object detection and
149:  photometry can be found in \citet{G2003a}.
150: 
151:  We supplement our ACS data set with the VLT ISAAC near-infrared imaging. 
152:  The NIR data consists of the J, H, and $K_{s}$ bands that extend our wavelength coverage 
153:  out to $\lambda \sim 4400$ \AA (rest-frame $z \sim 4$). We combine the ACS and ISAAC data 
154:  using a method that optimally matches the relative photometry between these bands 
155:  \citep{PapPhD,P2003}.  
156:  
157:  \section{Theoretical Model}\label{theoretical model}
158: 
159:  We make use of a semi-analytic model, based on the hierarchical
160:  structure formation paradigm in a $\Lambda$CDM cosmology
161:  ($\Omega_{\Lambda} = 0.7$, $\Omega_{m} = 0.3$, $\rm h = H_{0}/100$ 
162:  $\kms$$\rm Mpc^{-1} = 0.7$). The model treats the formation of structure
163:  via a hierarchical `merger tree', and includes a treatment of gas
164:  cooling, star formation, supernova feedback and chemical enrichment,
165:  galaxy mergers, stellar populations and dust (see \citet{SP1999} and
166:  SPF2001 for details). We follow halo merger histories down to a
167:  circular velocity of $\rm V_c =30 \kms$, as gas collapse and star
168:  formation in smaller halos is assumed to be suppressed by the presence
169:  of a photoionizing background \citep[see][and references
170:  therein]{S2002}. We use the multi-metallicity stellar SEDs from the
171:  STARDUST models \citep{DGS1999}, with a Kennicutt IMF. Dust extinction
172:  is modeled in a similar manner as in SPF2001. We use this model to
173:  produce a `mock-GOODS' catalog with the same geometry, sky area,
174:  filter passbands, etc. as the real GOODS.
175: 
176:  \section{Galaxy Sample}\label{galaxy sample}
177: 
178:  The $B_{435}$-dropout sample was chosen using the color selection criteria described in detail by \citet{G2003b}. 
179:  The sample was refined by applying detection limits at (S/N)$_{z,i} > 5$ for the ACS data set, and another one at
180:  (S/N)$_{K_{s}} > 1$ for the matched NIR set. These signal-to-noise restrictions limited our sample 
181:  to galaxies with $z_{850} < 26.5$, which helped curtail the presence of spurious detections. The relatively faint 
182:  $K_{s}$ cut-off was necessitated by the desire for a large enough sample for our study and did not significantly
183:  affect our conclusions. After further visual inspection, we chose 136 color-selected galaxies for analysis.
184: 
185:  Using the same color selection criteria that were applied to the observed sample we produced a model-derived
186:  $B_{435}$-dropout sample. We then applied a magnitude cut-off of $z < 26.5$ in order to comply with the
187:  observational detection limit. Before applying the color selection criteria we incorporated simulated observational scatter 
188:  into our model-derived photometric catalogs. The observational scatter was drawn from a Gaussian distribution with
189:  the typical signal-to-noise values found in the observed CDF-S data set. 
190: 
191:  Due to the dearth of spectroscopic data, empirical Monte Carlo simulations were performed to estimate 
192:  the redshift distribution of the observed $B_{435}$-dropout sample. These simulations
193:  were based on artificial LBGs distributed over a wide redshift range ($2.5 < z < 8$) with assumed distribution
194:  functions of UV luminosity, SED, morphology, and size, adjusted to match the colors of observed $B_{435}$-dropouts
195:  observed at $z \sim 4$. Reader is urged to see \citet{G2003b} for a detailed discussion on these simulations. 
196:  The redshift distribution of the simulated color-selected sample was found to have a mean value of  $z \sim 3.78$ with 
197:  a standard deviation of $\pm 0.34$. 
198:  
199:  One of the interesting tests that we can perform is to test for the incompleteness of our color-selected sample. 
200:  According to the Monte Carlo simulations, $73 \%$ of all the simulated $B_{435}$-dropout galaxies down to $z_{850} < 26.5$ 
201:  and in the interval $3.44 < z < 4.12$ are recovered using our color selection criteria.  When we apply the color criteria to 
202:  our model, we select $67 \%$ of all model galaxies in the same $3.44 < z < 4.12$ redshift range, down to the same 
203:  limiting magnitude of $z_{850} < 26.5$. This implies that the simulations performed by \citet{G2003b} and the 
204:  semi-analytic model show concordant incompleteness estimates with respect to the $B_{435}$-dropout selection
205:  technique in the above redshift range and down to our detection limit. 
206: 
207:  \section{Galaxy Properties}\label{galaxy properties} 
208: 
209:  The $i_{775} - K_{s}$ vs. $i_{775} - z_{850}$ (1550 - 4400 vs. 1550 - 1700 \AA, rest-frame $z \sim 4$)
210:  colors for the observed and model-derived $B_{435}$-dropout samples are presented in Figures 1a and 1b, 
211:  respectively. A Kolmogorov-Smirnov test reveals a $19 \%$ likelihood that the $i_{775} - K_{s}$ colors are 
212:  drawn from the same underlying distribution for the two $B_{435}$-dropout samples. A probability of 
213:  2 X $10^{-8}$ is obtained for the corresponding $i_{775} - z_{850}$ colors. This indicates a relatively good 
214:  agreement in the $i_{775} - K_{s}$ color distribution between the two samples, but a poor correlation in the 
215:  $i_{775} - z_{850}$ colors. 
216: 
217:  The $i_{775} - z_{850}$ colors probe the slope of the UV continuum, which is believed to be
218:  primarily an indicator of internal dust content  in young stellar populations \citep[e.g.][]{MHC1999}.
219:  In our model we used dust normalized against the $z = 0$ data and a Galactic extinction curve.
220:  The apparent disparity could be fixed by employing a different extinction curve \citep{C1997}, 
221:  or including the expected dependence of the extinction on the age of the stellar populaton \citep{CF2000}.
222: 
223:  Aside from the photometric analysis, we compared the luminosity function from our model with the one 
224:  for G-dropouts from \citet{SAG1999} and B-dropouts from HDF-N to ensure proper number counts. 
225:  We found good agreement, which is not surprising since our model is very similar to the one that was used 
226:  (and optimized) in SPF2001. In fact, the reader is urged to refer to SPF2001 for a detailed discussion on this 
227:  topic.
228:  
229:  Several researchers \citep[e.g.][]{PDF2001,SSA2001} have used simple
230:  parameterized star formation histories and stellar population
231:  synthesis models to estimate physical properties of LBG's at $z \sim 3$
232:  using the optical-NIR photometry as a constraint on the star formation
233:  history. Here, as we have shown that our model reproduces (modulo dust) 
234:  the observed color distribution for objects selected via their $B_{435} - V_{606}$ 
235:  and $V_{606} - i_{775}$ colors and corresponding magnitude limits, 
236:  we argue on similar grounds that our model should reproduce the statistical distribution 
237:  of underlying stellar ages and masses of the observed $B_{435}$-dropout population. 
238:  Thus we can use our models to obtain estimates for some of these quantities. 
239: 
240:  In Figure 2 we show the stellar masses of our color-selected model galaxies. 
241:  We note that the stellar masses range from $10^{8}$ to $10^{10}$ $\rm h^{-2}$ 
242:  $\rm M_{\sun}$, which is roughly two orders of magnitude less than the stellar masses of
243:  the present day $L_{*}$ spirals and ellipticals --- this indicates that, as with the $z \sim 3$ population, 
244:  the $z \sim 4$ Lyman-break galaxies are not the fully assembled progenitors of the present-day $L
245:  > L_{*}$ galaxies \citep{GSM96,S96a}, and that several generations of merging events must take 
246:  place between $z \sim4 $ and the present epoch. The median mass of $\rm Log[M_{star}] \sim 9.26$
247:  ($\rm M_{\sun}$ $\rm h^{-2}$) is 0.5 dex less than for the $z \sim 3$ galaxies studied 
248:  in SPF2001. This implies a stellar mass build-up between the two epochs. 
249:  To further explore the last point we have looked at the mean stellar masses of all $L^{*}$ 
250:  galaxies measured in rest-frame UV and predicted by our model at $z \sim 3$ and $z \sim 4$. 
251:  We find that at $z \sim 3$, with $m_{*} = 24.358$ (UV rest-frame), we get a mean value of
252:  $\rm Log[M_{star}] = 9.74$ ($\rm M_{\sun}$ $\rm h^{-2}$) in a $m_{*} \pm 0.5$ magnitude interval, 
253:  and at $z \sim 4$, with $m_{*} = 24.998$ (UV rest-frame), we get a mean value of 
254:  $\rm Log[M_{star}] = 9.58$ ($\rm M_{\sun}$ $\rm h^{-2}$), again in a $m_{*} \pm 0.5$ magnitude interval.
255:  This corresponds to a mass build-up of approximately $\sim 40 \%$. This type of mass build-up 
256:  between $z \sim 4$ and $z \sim 3$ is similar to that inferred from the \citet{P2003}
257:  comparison of volume-averaged SEDs of observed Lyman-break galaxies. 
258: 
259:  Our model also provides us with the star formation rates (SFR) and ages of our
260:  color-selected model galaxies. Our model results tell us that the values of these two quantities for the 
261:  $z \sim 4$ sample are very similar to the ones found for the $z \sim 3$ sample studied by SPF2001. 
262:  We observe a branch of actively star-forming galaxies, and a broader branch of
263:  galaxies with lower SFR, corresponding to bursting and fading galaxies. The star formation rates for 
264:  the brightest galaxies in the model approach 100 $\rm M_{\sun}$ $\rm yr^{-1}$, similar to what has
265:  been found for $z \sim 3$ Lyman-break galaxies studied by SPF2001. 
266: 
267:  We have also looked at the distribution of \emph{stellar-mass-weighted} mean
268:  stellar ages. We found that the distribution is very broad and skewed toward ages of
269:  less than 300 Myr, with a median age of 200 Myr, and a peak at about
270:  100 Myr. There is, however, a tail of objects with older stellar populations, reaching a mean 
271:  age of $\sim 1$ Gyr --- close to the age of the universe at that redshift in our adopted cosmology 
272:  ($\sim 1.5$ Gyr). 
273: 
274:  Drawing on the stellar mass distribution results (Figure 2) we plot in Figure 3 the mass distribution 
275:  of all (i.e. not just color-selected) galaxies from our mock catalog, limited to galaxies with
276:  $\rm Log[M_{star}] > 9.26$ ($\rm M_{\sun}$ $\rm h^{-2}$) (the median value from Figure 2) 
277:  and spanning the redshift range of  $3.44 < z < 4.12$. As was mentioned in \S\ref{galaxy sample}, we select 
278:  $67 \%$ of model galaxies in this redshift range with our color selection criteria, down to a limiting magnitude of 
279:  $z_{850} < 26.5$. This corresponds to $62 \%$ of the total stellar mass available in that redshift range, down
280:  to that magnitude limit. Hence, with our color criteria, we 'observe' the majority of the stellar mass that resides in galaxies 
281:  brighter than $z_{850} < 26.5$ and spanning $3.44 < z < 4.12$. Figure 2 indicates that $50 \%$ of our color-selected 
282:  objects lie above $\rm Log[M_{star}] > 9.26$ ($\rm M_{\sun}$ $\rm h^{-2}$), we find though that only $50 \%$ of all model 
283:  galaxies more massive than $\rm Log[M_{star}] > 9.26$ ($\rm M_{\sun}$ $\rm h^{-2}$) are brighter than $z_{850} < 26.5$. 
284:  So for model-derived objects residing in $3.44 < z < 4.12$ we only sample $50 \%$ of the available stellar mass down to
285:  $z_{850} < 26.5$. The rest of the stellar mass resides in relatively massive UV-faint galaxies. The color selection 
286:  incompleteness down to $z_{850} < 26.5$ and the mass contained in the UV-faint objects with magnitudes $z_{850} > 26.5$ 
287:  are two effects that conspire to severely limit the amount of mass selected with optical surveys \citep{F2003}. 
288:  Given how well our model predicts the observed colors (modulo dust), number counts, and other properties, this points to 
289:  a substantial deficit in optically-selected galaxies, and by extension, the total stellar mass. Since our model catalog was limited 
290:  to galaxies with $z_{850} < 28.0$ values, the 'unseen' mass fraction estimate should be taken as a lower limit.
291: 
292:  \section{Summary and Conclusions}\label{summary} 
293: 
294:  We performed a comparative analysis of the color-selected $B_{435}$-dropout samples taken from GOODS 
295:  observations and a variant of a $\Lambda$CDM-based semi-analytic model. We found that the color
296:  selection technique used to obtain the $B_{435}$-dropout sample does a relatively good job in selecting a 
297:  complete census of galaxies spanning $z \sim 3.78 \pm 0.34$ and down to $z_{850} < 26.5$. The color
298:  selection incompleteness limits are roughly the same when applied to our model and the empirical Monte Carlo
299:  simulations preformed by \citet{G2003b}. We select $67 \%$ of all galaxies and $62 \%$ of stellar mass in that redshift range,
300:  down to $z_{850} < 26.5$. This lends further credence to applicability of the $B_{435}$-dropout color selection 
301:  technique outlined in \citet{G2003b}, which was designed based on simpler, more empirical models of galaxy SEDs at 
302:  this epoch, and has not, as yet, been extensively verified with spectroscopic data.
303: 
304:  We found a relatively good agreement between the model-derived and observed CDF-S $B_{435}$-dropout 
305:  galaxy $i_{775} - K_{s}$ colors. The $i_{775} - z_{850}$ colors showed much less agreement, however, we attribute 
306:  the discrepancy to potentially inadequate dust recipe employed in the model. In general, we found that the stellar age, 
307:  mass, and the star formation properties of the $z \sim 4$ sample were similar to the $z \sim 3$ sample studied by 
308:  SPF2001. We found a model-derived $\sim 40 \%$ stellar mass build-up for the UV rest-frame $L^{*}$ galaxies 
309:  between the two epochs. In addition we found that if we look at all galaxies predicted by our model, spanning 
310:  $3.44 < z < 4.12$ redshift range, then at least $50 \%$ of the stellar mass contained in objects with 
311:  $\rm Log[M_{star}] > 9.26$ ($\rm M_{\sun}$ $\rm h^{-2}$) is missed due to the $z_{850} < 26.5$ observational limit. 
312:  If our model is correct, and it does show reasonable agreement with respect to colors (modulo dust), number counts, 
313:  and other physical quantities, this result has significant implications for the completeness of the optically-selected 
314:  surveys at this epoch. 
315: 
316:  \acknowledgments
317: 
318:  Support for this work was provided by NASA through grant GO09583.01-96A from the 
319:  Space Telescope Science Institute, which is operated by the Association of Universities
320:  for Research in Astronomy, Inc.\ (AURA), under NASA contract NAS5-26555.
321: 
322:  Support for this work, part of the {\it Space Infrared Telescope 
323:  Facility (SIRTF)} Legacy Science Program, was provided by NASA through 
324:  Contract Number 1224666 issued by the Jet Propulsion Laboratory, 
325:  California Institute of Technology under NASA contract 1407.
326: 
327:  %\bibliographystyle{apj}
328:  %\bibliography{apjmnemonic,bib}
329: 
330: \begin{thebibliography}{}
331: 
332: \bibitem[\protect\citeauthoryear{{Baugh} et~al.}{{Baugh} et~al.}{1998}]{BCF1998}
333: {Baugh}, C.~M., {Cole}, S., {Frenk}, C.~S., \& {Lacey}, C.~G. 1998, ApJ, 498, 504
334: 
335: \bibitem[\protect\citeauthoryear{{Calzetti}}{{Calzetti}}{1997}]{C1997}
336: {Calzetti}, D. 1997, AJ, 113, 162
337: 
338: \bibitem[\protect\citeauthoryear{{Charlot} \& {Fall}}{{Charlot} \& {Fall}}{2000}]{CF2000}
339: {Charlot}, S., {Fall}, S.~M. 2000, ApJ, 539, 718C
340: 
341: \bibitem[\protect\citeauthoryear{{Devriendt}, {Guiderdoni}, \& {Sadat}}{{Devriendt}, {Guiderdoni}, \& 
342: {Sadat}}{1999}]{DGS1999}
343: {Devriendt}, J.~E.~G., {Guiderdoni}, B., \& {Sadat}, R. 1999, A\&A, 350, 381D
344: 
345: \bibitem[\protect\citeauthoryear{{Franx et~al.}}{{Franx} et~al.}{2003}]{F2003}
346: {Franx}, M. et~al. 2003, ApJ, 587, L79
347: 
348: \bibitem[\protect\citeauthoryear{{Giavalisco et~al.}}{{Giavalisco} et~al.}{2003a}]{G2003a}
349: {Giavalisco}, M. et~al. 2003a, this ApJL issue
350: 
351: \bibitem[\protect\citeauthoryear{{Giavalisco et~al.}}{{Giavalisco} et~al.}{2003b}]{G2003b}
352: {Giavalisco}, M. et~al. 2003b, this ApJL issue
353: 
354: \bibitem[\protect\citeauthoryear{{Giavalisco}, {Steidel}, \& {Macchetto}}{{Giavalisco}, {Steidel}, \&
355:  {Macchetto}}{1996}]{GSM96}
356: {Giavalisco}, M., {Steidel}, C., \& {Macchetto}, D. 1996, ApJ, 470, 189
357: 
358: \bibitem[\protect\citeauthoryear{{Madau} et~al.}{{Madau} et~al.}{1996}]{MFD1996}
359: {Madau}, P., {Ferguson}, H., {Dickinson}, M., {Giavalisco}, M., {Steidel}, C., \& {Fruchter}, A. 1996, 
360: MNRAS, 283, 1388
361: 
362: \bibitem[\protect\citeauthoryear{{Meurer}, {Heckman}, \& {Calzetti}}{{Meurer}, {Heckman}, 
363: \& {Calzetti}}{1999}]{MHC1999}
364: {Meurer}, G., {Heckman}, T., \& {Calzetti}, D. 1999, ApJ, 521, 64
365: 
366: \bibitem[\protect\citeauthoryear{{Oke}}{{Oke}}{1974}]{O1974}
367: {Oke}, J.~B. 1974, ApJS, 27, 21
368: 
369: \bibitem[\protect\citeauthoryear{{Papovich} et~al.}{{Papovich} et~al.}{2003}]{P2003}
370: {Papovich}, C. et~al. 2003, this ApJL issue
371: 
372: \bibitem[\protect\citeauthoryear{{Papovich}}{{Papovich}}{2002}]{PapPhD}
373: {Papovich}, C. 2002, Ph. D. thesis, The Johns Hopkins University
374: 
375: \bibitem[\protect\citeauthoryear{{Papovich}, {Dickinson}, \& {Ferguson}}{{Papovich}, {Dickinson}, \& 
376: {Ferguson}}{2001}]{PDF2001}
377: {Papovich}, C., {Dickinson}, M., \& {Ferguson}, H.~C. 2001, ApJ, 559, 620
378: 
379: \bibitem[\protect\citeauthoryear{{Shapley} et~al.}{{Shapley} et~al.}{2001}]{SSA2001}
380: {Shapley}, A.~E., {Steidel}. C.~C., {Adelberger}, K. L., {Dickinson}, M.,
381:  {Giavalisco}, M., \& {Pettini}, M. 2001, ApJ, 562, 95
382: 
383: \bibitem[\protect\citeauthoryear{{Somerville}}{{Somerville}}{2002}]{S2002}
384: {Somerville}, R.~S. 2002, ApJ, 572, L23
385: 
386: \bibitem[\protect\citeauthoryear{{Somerville}, {Primack}, \& {Faber}}{{Somerville}, {Primack}, \& 
387: {Faber}}{2001}]{SPF2001} 
388:  {Somerville}, R.~S., {Primack}, J.~R., \& {Faber} 2001, ApJ, 310, 1087
389: 
390: \bibitem[\protect\citeauthoryear{{Somerville} \& {Primack}}{{Somerville} \&
391:   {Primack}}{1999}]{SP1999} 
392:  {Somerville}, R.~S.,  \& {Primack}, J.~R. 1999, MNRAS, 310, 1087
393:  
394: \bibitem[\protect\citeauthoryear{{Steidel} et~al.}{{Steidel} et~al}{1999}]{SAG1999} 
395:  {Steidel}, C., {Adelberger}, K.~A., {Giavalisco}, M., {Dickinson}, M., \& {Pettini}, M. 1999, 
396:  ApJ, 519, 1
397: 
398: \bibitem[\protect\citeauthoryear{{Steidel} et~al.}{{Steidel} et~al.}{1996a}]{S96a}
399: {Steidel}, C., {Giavalisco}, M., {Dickinson}, M., \& {Adelberger}, K. 1996a, AJ, 112, 352 
400:   
401: \bibitem[\protect\citeauthoryear{{Wechsler} et~al}{{Wechsler} et~al.}{2001}]{WSB2001}
402: {Wechsler}, R.~H., {Somerville}, S.~S., {Bullock}, J.~S., {Kolatt}, T.~S., {Primack}, J.~R., 
403: {Blumenthal}, G.~R., \& {Dekel}, A. 2001, ApJ, 554, 85
404: 
405: \end{thebibliography}
406: 
407: \clearpage
408: 
409: \begin{figure*}
410: \resizebox{0.5\textwidth}{!}{\includegraphics{f1a}}
411: \resizebox{0.5\textwidth}{!}{\includegraphics{f1b}}
412: \figcaption{The $i_{775} - K_{s}$ versus $i_{775} - z_{850}$ colors of the $z \sim 4$ color-selected galaxies. 
413: Panel (a) shows the 136 observed $B_{435}$-dropout galaxies {\it filled-diamond} from the CDF-S. Panel (b) 
414: shows a Hess diagram of the `Collisional-Starburst' model galaxies with artificial observational scatter folded
415: in, and selected with the same color criteria and magnitude 
416: limits that were applied to the observed sample. Contours have been superimposed to help 
417: guide the eye; they represent the $30^{th}$, $50^{th}$, $70^{th}$, and $90^{th}$ percentiles. In addition, 
418: histograms are included for both the model ({\it gray solid-line}) and observed ({\it black dash-line}) galaxy colors. 
419: This figure illustrates the relative agreement in the $i_{775} -K_{s}$ color distributions for the model-derived and 
420: observed $B_{435}$-dropout galaxies, while at the same time illustrating an appreciable mismatch in the 
421: $i_{775} - z_{850}$ colors.\label{fig1}}
422: \end{figure*}
423: \clearpage
424: 
425: \begin{inlinefigure}
426: \resizebox{\textwidth}{!}{\includegraphics{f2}}
427: \figcaption{The stellar masses of the color-selected model galaxies. The {\it filled-diamond} symbols show stellar 
428: masses of the individual color-selected model galaxies vs. their corresponding $z_{850}$ magnitudes. 
429: The histogram shows the projected distributions for the same color-selected model galaxies 
430: with an imposed $z_{850} < 26.5$ magnitude limit (shown as a {\it gray dash-line}). The predicted masses are 
431: two magnitudes lower than the stellar masses of the present day $L_{*}$ spirals and ellipticals.\label{fig2}}
432: \end{inlinefigure}
433: \clearpage
434: 
435: \begin{inlinefigure}
436: \resizebox{\textwidth}{!}{\includegraphics{f3}}
437: \figcaption{This figure illustrates the stellar mass distribution of all model-derived galaxies spanning the redshift 
438: range of $3.44 < z < 4.12$ and with $\rm Log[M_{star}] > 9.26$ ($\rm M_{\sun}$ $\rm h^{-2}$)
439: (the median value from Figure 2). Galaxies were binned into 0.25 magnitude intervals and weighted by their 
440: corresponding stellar mass. A {\it gray dash-line} is included to delineate our observational magnitude limit. This 
441: figure shows the amount of model-predicted mass potentially missed by the current optical surveys.\label{fig2}}
442: \end{inlinefigure}
443: 
444: \end{document}
445: 
446: 
447: