1: \documentclass[12pt,preprint]{aastex}
2: %% manuscript produces a one-column, double-spaced document:
3: %%\documentclass[manuscript]{aastex}
4: %% preprint2 produces a double-column, single-spaced document:
5: %%\documentclass[preprint2]{aastex}
6: %%\documentclass[12pt,preprint]{aastex}
7: %%\documentstyle[11pt,aasms4]{article}
8: %% Sometimes a paper's abstract is too long to fit on the
9: %% title page in preprint2 mode. When that is the case,
10: %% use the longabstract style option.
11: %% \documentclass[preprint2,longabstract]{aastex}
12:
13: %\documentclass{emulateapj}
14: \newcommand{\lta}{{\>\rlap{\raise2pt\hbox{$<$}}\lower3pt\hbox{$\sim$}\>}}
15: \newcommand{\gta}{{\>\rlap{\raise2pt\hbox{$>$}}\lower3pt\hbox{$\sim$}\>}}
16:
17: %\newcommand{\vdag}{(v)^\dagger}
18: %\newcommand{\myemail}{skywalker@galaxy.far.far.away}
19: %\slugcomment{Draft Version 5/5/2008}
20: %\shorttitle{The Environment of High-z QSOs}
21: %\shortauthors{Kim et al.}
22:
23:
24: \usepackage{graphicx}
25: \usepackage{amssymb}
26: %\usepackage{epstopdf}
27:
28: \begin{document}
29:
30: \title{The Environments of High Redshift QSOs}
31: \author{Soyoung Kim\altaffilmark{2}}
32: \email{sykim@pha.jhu.edu}
33: \author{Massimo Stiavelli\altaffilmark{2,3}}
34: \email{mstiavel@stsci.edu}
35: \author{M. Trenti\altaffilmark{2}}
36: \author{C.M. Pavlovsky\altaffilmark{2}}
37: \author{S.G. Djorgovski\altaffilmark{4}}
38: \author{C. Scarlata\altaffilmark{5}}
39: \author{D. Stern\altaffilmark{6}}
40: \author{A. Mahabal\altaffilmark{3}}
41: \author{D. Thompson\altaffilmark{3}}
42: \author{M. Dickinson\altaffilmark{7}}
43: \author{N. Panagia\altaffilmark{8}}
44: \author{G. Meylan\altaffilmark{9}}
45:
46: \altaffiltext{1}{Based on observations with the NASA/ESA {\it Hubble Space Telescope},
47: obtained at the Space Telescope Science Institute, which is operated by the Association of
48: Universities of Research in Astronomy, Inc., under NASA contract NAS5-26555}
49:
50: \altaffiltext{2}{The Johns Hopkins University, 3400 N. Charles St.,Baltimore, MD 21218}
51: \altaffiltext{3}{Space Telescope Science Institute, Baltimore, MD 21218}
52: \altaffiltext{4}{California Institute of Technology,
53: MS 105-24, Pasadena, CA 91125}
54: \altaffiltext{5}{Spitzer Science Center, Pasadena, CA }
55: \altaffiltext{6}{Jet Propulsion Laboratory, California Institute of Technology,
56: Mail Stop 169-506, Pasadena, CA 91109}
57: \altaffiltext{7}{National Optical Astronomical Observatories,
58: P.O. Box 26732, Tucson, AZ 85726}
59: \altaffiltext{8}{Space Telescope Science Institute, Baltimore, MD 21218; Supernova Ltd., Olde Yard Village 131, Northsound Road, Virgin Gorda, British Virgin Islands}
60: \altaffiltext{9}{Laboratoire d'Astrophysique, Ecole Polytechnique F\'ed\'erale de Lausanne (EPFL)
61: Observatoire de Sauverny, CH-1290 Versoix, Switzerland }
62:
63: \begin{abstract}
64: We present a sample of $i_{775}$-dropout candidates identified in five
65: Hubble Advanced Camera for Surveys fields centered on Sloan Digital
66: Sky Survey QSOs at redshift $z\sim 6$. Our fields are as deep as
67: the Great Observatory Origins Deep Survey (GOODS) ACS images which
68: are used as a reference field sample. We find them to be overdense in two fields,
69: underdense in two fields, and as dense as the average density of
70: GOODS in one field. The two excess fields show significantly
71: different color distributions from that of GOODS at the 99\%
72: confidence level, strengthening the idea that the excess objects are
73: indeed associated with the QSO. The distribution of $i_{775}$-dropout
74: counts in the five fields is broader than that derived from GOODS at
75: the 80\% to 96\% confidence level, depending on which selection criteria
76: were adopted to identify $i_{775}$-dropouts; its width cannot be explained by cosmic
77: variance alone.
78: Thus, QSOs seem to affect their environments in complex ways. We suggest the picture
79: where the highest redshift QSOs are located in very massive overdensities and are
80: therefore surrounded by an overdensity of lower mass halos.
81: Radiative feedback by the QSO can in some cases prevent halos from
82: becoming galaxies, thereby generating in extreme cases an
83: underdensity of galaxies. The presence of both enhancement and
84: suppression is compatible with the expected differences between lines
85: of sight at the end of reionization as the presence of residual
86: diffuse neutral hydrogen would provide young galaxies with shielding
87: from the radiative effects of the QSO.
88: \end{abstract}
89:
90: \keywords{galaxies : high-redshift --- early universe : galaxy formation
91: --- quasars}
92:
93: \section{INTRODUCTION}
94:
95: Observational astronomy has finally reached the point of beginning
96: to probe the era of reionization of hydrogen. The long search for
97: Gunn-Peterson \citep{Gunn65} troughs in the spectra of increasingly
98: higher redshift QSOs has finally become fruitful with the Sloan
99: Digital Sky Survey (SDSS). A dramatic increase in the intergalactic
100: hydrogen absorption at $z\simeq6$ was detected in the spectra of
101: high-redshift SDSS QSOs \citep[e.g.,][]{Becker01,Djorgovski01, White03}.
102: This was followed by the possible detection of a Gunn-Peterson
103: trough in the spectrum of QSO SDSS J1030+0524 at $z=6.28$
104: \citep{Fan01}. The case for termination of the reionization epoch at
105: $z\sim6$ is now relatively solid \citep[e.g.,][]{Fan06}, even if
106: not universally agreed upon \citep[e.g.,][]{Lidz06, Bolton07}. At
107: the same time, the Compton optical depth $\tau=0.084\pm0.016$ from
108: the five year WMAP data \citep{Komatsu08} is compatible with a
109: somewhat extended reionization process terminating at $z\simeq6$
110: \citep[e.g.,][]{Shull07}.
111:
112: Despite the growing consensus that reionization may have terminated
113: at $z\simeq6$, it is extremely unlikely that it occurred in a
114: universally synchronized fashion. Fluctuations from line of sight to
115: line of sight are generally expected due to clumpiness of the IGM,
116: and the gradual development and clumpy distribution of the first
117: ionizing sources, either proto-galaxies or early AGN
118: \citep[e.g.,][]{Miralda00}. Thus, reionization is expected to occur
119: gradually as the UV emissivity increases \citep[cf.][]{McDonald01},
120: with the lowest density regions becoming fully reionized first. This
121: is also suggested by modern numerical simulations
122: \citep[e.g.,][]{Ciardi03,Gnedin97,Gnedin04} which predict an
123: extended period of reionization, starting at $z \sim 15$ or even
124: higher and ending at $z\sim 6$ \citep[see
125: also][]{Cen03,Haiman03,Somerville03,Wyithe03}.
126:
127: If reionization is completed at $z\simeq6$, it is reasonable to
128: attempt to identify the galaxies responsible for it. The combined
129: Great Observatory Origins Deep Survey \citep[GOODS;][]{Giavalisco04}
130: and Hubble Ultra Deep Field \citep[HUDF;][]{Beckwith06} have provided a
131: large sample of $i_{775}$-dropout galaxies. Unfortunately, their estimated
132: ionizing flux is insufficient to reionize the universe under
133: standard assumptions \citep{Bunker04,Dickinson04,Bouwens07}; one
134: would have to assume top heavy, very metal-poor stellar populations
135: \citep{Stiavelli04}, or rely on a burgeoning population of dwarf
136: galaxies brought about by a steep faint end slope of the luminosity
137: function \citep{Yan04}. A last alternative is that reionization was
138: very gradual \citep[e.g.,][]{Bouwens07}. Unfortunately, testing
139: these ideas is observationally challenging. At the same time, given
140: the predominance of the HUDF data on the derivation of the faint end
141: luminosity function of $i_{775}$-dropouts, one is led to wonder how much
142: these results are affected by cosmic variance given the small volume
143: probed by the HUDF. On this issue, conflicting claims regarding the
144: density of HUDF $i_{775}$-dropouts can be found in the literature, with
145: \citet{Bouwens07} arguing in favor of an underdensity \citep[see
146: also][]{Oesch07} while \citet{Malhotra05} argued in favor of an
147: overdensity \citep[however, see][]{TS08}.
148:
149: In general, one would expect very high redshift galaxies to be highly clustered, especially
150: if purely gravitational clustering effects were amplified by positive feedback.
151: Thus, in order to address the importance and sign of feedback in the environments where
152: they should be easiest to detect, we were led to focus on fields centered on $z\gta6$ QSOs
153: as they should be the most clustered environments at
154: these very high redshifts and the strongest cases of feedback available for study.
155:
156: Indeed, a generic expectation in most models of galaxy formation is
157: that the most massive density peaks in the early universe are likely
158: to be strongly clustered \citep{Kaiser84, Efstathiou88}. The
159: evidence for such bias is already seen with large samples of
160: Lyman-break galaxies at $z\sim3 - 3.5$ \citep{Steidel03}, and in
161: Lyman $\alpha$ selected galaxy samples \citep[e.g.,][]{Venemans03, Ouchi05},
162: and it should be even stronger at higher redshifts. An excess in the
163: number of galaxies and in the density of star formation was also
164: discovered in a systematic Keck survey of fields centered on known
165: $z>4$ quasars \citep[e.g.,][]{Djorgo99, Djorgovski99,Djorgovski03}.
166: The high metallicity associated with QSOs \citep{Barth03} -- even at $z \gta 6$ --
167: is often interpreted as evidence that they are located at the center
168: of massive (proto--)galaxies, thereby corroborating the overall
169: picture. These arguments justify the expectation that QSOs at $z
170: \simeq 6$ most likely highlight some of the first perturbations that
171: become non--linear in the density distribution of matter \citep[see
172: e.g.,][]{TS07}.
173:
174: However, QSOs are not ``quiet neighbors''. The intense emission of
175: ionizing radiation associated with QSOs ionizes the surrounding IGM
176: and may even photo-evaporate gas in neighboring dark halos before
177: this has an opportunity to cool and form stars \citep{Shapiro01}. In
178: this context, QSOs would suppress galaxy formation in their
179: vicinities. One would then observe a paucity of galaxies near a QSO
180: despite the underlying excess of dark halos. Moreover, near the reionization epoch
181: the fraction of neutral hydrogen in the IGM may change rapidly, possibly shifting the
182: balance of the two effects. It would be exciting to see a change
183: from source enhancement to suppression around reionization by
184: observing a sample of $z=6$ QSOs.
185:
186: It is with this goal in mind that we started a study of the
187: environment of the five then known QSOs at $z \gtrsim 6$ using the
188: Advanced Camera for Surveys (ACS) on board the Hubble
189: Space Telescope (HST) to obtain images in the F775W
190: ($i_{775}$) and in the F850LP ($z_{850}$) filters so as to identify
191: candidate objects at $z=6$ as $i_{775}$-dropout galaxies. All five fields
192: were observed to the same depth as GOODS in the $i_{775}$ and
193: $z_{850}$ bands so that GOODS can be used as a reference field
194: sample.
195:
196: In a previous paper \citep{Stiavelli05}, we analyzed the number of
197: $i_{775}$-dropout galaxies identified in a HST/ACS field centered on the SDSS QSO J1030+0524
198: at $z=6.28$. In this field we found a very significant excess of sources compared
199: to the density of $i_{775}$-dropouts seen in GOODS, thus suggesting that clustering
200: wins over negative feedback. \citet{Zheng06} also observed a radio-loud QSO at $z\sim6$, SDSS J0836+0054, using ACS and detected a significant overdensity of i-dropout galaxies in its vicinity.
201: In this paper, we analyze four additional QSO fields in order to test and expand this result.
202:
203: Section 2 is a description of the observations and data analysis.
204: Section 3 describes our $i_{775}$-dropout objects and their properties.
205: Section 4 contains discussion of our results and Section 5 summarizes our conclusions.
206: In this paper we use AB magnitudes and assume the cosmological parameters,
207: $H_{0}= 70\, {\rm km}\, {\rm s}^{-1}\, {\rm Mpc}^{-1}$, $\Omega_{m}= 0.26$, and $\Omega_{\Lambda}= 0.74$.
208:
209:
210: \section{DATA REDUCTION AND ANALYSIS}
211:
212: We observed five fields centered on five SDSS QSOs at redshift $z\gta6$
213: with the ACS/WFC on board HST. The QSOs were the most distant quasars known at the
214: time of our original Cycle 12 proposal. All are radio-quiet. Our targets were SDSS
215: J1148+5251 at $z=6.40\pm0.01$ \citep{Barth03}, SDSS J1030+0524 at $z=6.28\pm0.03$, SDSS J1306+0356 at $z=5.99\pm 0.03$, SDSS J1048+4637 at $z=6.23\pm0.03$, and SDSS J1630+4012 at $z=6.05\pm0.03$ \citep{Fan01, Fan03}.
216:
217: Table 1 summarizes the observations. Our observations in the F775W
218: ($i_{775}$) and the F850LP ($z_{850}$) filters were designed to have similar exposure times to those used for the original (version 1.0) GOODS data products. The data were processed by the ACS pipeline CALACS that carries out bias and dark current removal and flat-fielding. The
219: individual calibrated images ({\it flt} files) were combined into a
220: single image for each filter using Multidrizzle, a pyraf application
221: based upon the drizzle algorithm \citep{Fruchter02}. Drizzle also
222: requires weight maps which we computed following the same procedure as was used for the GOODS data reduction:
223: %
224: \begin{equation}
225: Variance = \frac{\left[ (Dt+fB) + \sigma_{read} ^{2}
226: \right]}{(ft)^{2}}
227: \end{equation}
228: %
229: \begin{equation}
230: Weight = \frac{1}{(Variance)}
231: \end{equation}
232: %
233: where $D$ is the dark current (electron/sec/pixel), $f$ is the pixel
234: value of the reference flat field, $B$ is the background (electron/pixel) measured
235: in flat-fielded images, $t$ is the exposure time (second), and
236: $\sigma_{read}$ is the read-out noise (electron/pixel). We ran
237: MultiDrizzle (Koekemoer et al. 2002) with parameters {\it pixfrac}=
238: 1.0, {\it final\_scale}= 0.03 and {\it final\_wht\_type}= {\it ivm} (individual weight map).
239: The area of the final images is approximately 11.3 arcmin$^{2}$.
240: We measured the actual background noise in the drizzled ACS images, measuring
241: and correcting for the correlation between pixels introduced by the drizzling and
242: resampling process, and compared this to the variance predicted by the noise model
243: used to generate the weight maps (equations 1 and 2). This correction was also verified
244: by block averaging the images and measuring the resulting noise directly on scales
245: larger than the inter-pixel correlation lengths. The variance maps were adjusted
246: using this correction, and converted to rms maps which were provided to SExtractor \citep{Bertin96}
247: to modulate the source detection thresholds and to compute photometric uncertainties.
248:
249: The catalogs were obtained using SExtractor, run on
250: the drizzled science images and with the same input parameters as
251: those for the GOODS catalogs (for both the HDFN and the CDFS). We applied the same procedures to all five
252: fields. The $z_{850}$ band images were used as the detection images
253: when running SExtractor in dual-image mode. We required objects to be
254: detected at a signal-to-noise (S/N)$>5$ in the $z_{850}$ band.
255: For the total magnitude of a source, we adopted SExtractor's
256: MAG\_AUTO values. The adopted magnitude zero points were 25.6405 and
257: 24.8432 in $i_{775}$ and $z_{850}$, respectively. We computed
258: $i_{775}-z_{850}$ colors using the MAG\_ISO values to compare the
259: same isophotes in the two bands.
260: For $i_{775}$ band sources detected
261: at less than the two sigma level in isophotal apertures, we computed
262: lower limits for the colors using the 2$\sigma$ upper limit to the $i_{775}$ band isophotal flux.
263: The Galactic extinction estimate of E(B-V) was obtained from \citet{Schlegel98}
264: for GOODS and each QSO field. We determined the corrections for the $i_{775}$ and $z_{850}$ magnitudes using SYNPHOT. The actual corrections in the
265: two bands were as follows: 0.024 and 0.018 for HDFN, 0.016 and 0.012 for CDFS,
266: 0.048 and 0.036 for J1030+0524; 0.022 and
267: 0.016 for J1630+4012; 0.036 and 0.027 for J1048+4637; 0.044 and
268: 0.033 for J1148+5251; and 0.060 and 0.045 for J1306+0356. The
269: limiting magnitudes and completeness levels were comparable to those
270: of GOODS catalogs.
271:
272: \section{CANDIDATE OBJECTS}
273:
274: The selection criteria are based on the $i_{775}-z_{850}$ color,
275: a magnitude limit $z\leq26.5$, limits on S/N ratios, and the
276: SExtractor extraction $flag=0$ which identifies non-saturated and isolated sources outside the masked zones. We have considered two different
277: values of S/N$=5$ and $8,$ and the color limits of $i_{775}-z_{850}=
278: 1.3$ and $1.5$. Objects selected with S/N$>5$ and
279: $i_{775}-z_{850} > 1.3$ will constitute our least restrictive sample S1; objects with
280: S/N$>5$ and $i_{775}-z_{850} > 1.5$ are our sample S2; and those
281: with S/N$>8$ and $i_{775}-z_{850} > 1.3$ are our sample S3. We eliminate objects that reside near the edges and on the star
282: diffraction spikes, as well as objects that appear to be artifacts
283: during visual inspection. GOODS candidates were selected by the same
284: selection criteria using the GOODS catalogs (version 1.1), including visual inspection.
285: However, as the QSO fields only have ACS imaging in two bands, we do not require non-detections ($<2 \sigma$) in the $B_{435}$ and $V_{606}$ as was implemented in the \citet{Dickinson04} selection of $i_{775}$-dropouts in the GOODS fields. Therefore, our GOODS $i_{775}$-dropout sample is different from the one used in \citet{Dickinson04}. Table~2 shows the number of $i_{775}$-dropouts selected in QSO fields and GOODS for different S/N ratios and color limits.
286: In Table~2, the number of $i_{775}$-dropouts in GOODS is normalized
287: to the area of a single ACS/WFC field ($\sim$ 11.3 arcmin$^{2}$).
288: The measurements of all quasar field candidates with $i_{775}-z_{850}
289: > 1.3$ and S/N$>5$ are listed in Table~3.
290:
291: Contamination by stars is a potential concern. We estimated a priori
292: the possible contamination from stars by using as a proxy the number
293: density of stars brighter than visual magnitude $m_{v}=21$ at the
294: Galactic latitude of the five QSO fields. All fields have lower star
295: density than the mean star density at the galactic latitude of each QSO \citep{Zombeck90} at the galactic latitude of each QSO. In particular, the J1030+0524
296: field has a lower star density than GOODS, while the other
297: overdense field, J1630+4012, has a star density 4.8 times higher than GOODS.
298: This suggests a degree of caution is necessary in excluding stars. We have
299: identified stars using the SExtractor star-galaxy index, S/G, half-light radius, $r_{hl}$ and $z_{850}$ mag. The criteria for stars were S/G$\geq0.85$, $r_{hl}\leq0.1$
300: arcsecond, and $z_{850}<25.5$ applied to the S1 samples. We found no stellar $i_{775}$-dropout candidates in our five fields but found 16 stellar $i_{775}$-dropout candidates (0.55 stars per ACS field) in GOODS.
301:
302: Our target QSOs are not all flagged as stars because of the long wavelength point source halo effect seen with ACS. The point spread function in the F850LP filter is characterized by a long
303: wavelength halo which is due to light traveling through the CCD,
304: bouncing off the front side at a large angle, going once again
305: through the CCD and being detected. This effect is very
306: wavelength-dependent (and thus, for high-redshift QSOs, redshift-dependent).
307: Well-exposed images of a QSO will show this extended halo and the
308: QSO will fail to be identified as a star. The same would be true for
309: very red stars. However, if we artificially dim the QSOs to have similar apparent magnitudes as the other $i_{775}$-dropouts, the halos drop below the noise level and the fainter versions of our QSOs are identified as stars.
310:
311: We also estimated the possible contamination by stars fainter than 25.5 by considering the
312: candidates with S/G$\geq0.85$, and half light radius $r_{hl}\leq0.1$
313: arcsecond. In Table 3, we have two objects (A8 and B2) in J1030+0524
314: and J1630+4012 that satisfy this relaxed criteria. When applied to GOODS, we found 11 (very red) objects (0.38 objects per ACS field) out of 235 objects selected using the S1 criteria.
315:
316: For S/N$> 5$ and $i_{775}-z_{850}>1.3$ (our selection S1) we
317: see that two fields, J1030+0524 and J1630+4012, show an overdensity;
318: J1048+4637 has approximately the same number density of $i_{775}$-dropouts as GOODS; and the
319: J1148+5251 and J1306+0356 fields appear underdense compared to GOODS.
320:
321: We have verified whether the variations in the number of candidates could be due to
322: field-to-field background noise variations. We find these variations to be generally
323: small and that the background noise is highest in the field of J1030+0524, i.e., the one
324: with the largest excess. Thus, we conclude that background noise variations are not
325: affecting our results.
326:
327: Figures 1 through 5 show for each field the number counts as a
328: function of the $z_{850}$ magnitude (panel a) and as a function of
329: $i_{775}-z_{850}$ color (panel b). Panel c shows the count
330: distribution as a function of magnitude for objects redder
331: than $i_{775}-z_{850}=1.5$ and panel d shows the number of objects
332: redder than a given $i_{775}-z_{850}$ color in 0.1-magnitude bins for $i_{775}-z_{850}>0.9$.
333: The solid line shows the data for galaxies in the QSO fields.
334: The dotted line shows the distributions for the GOODS fields.
335:
336: Figure~1-(d) shows the color distribution of galaxies in the
337: J1030+0524 field (excluding the QSO) and GOODS. Their
338: distributions appear to be different, especially around $i_{775}-z_{850}\sim 2$.
339: In Figure~2-(d), the color distributions of J1630+4012 and GOODS appear
340: to be different for $i_{775}-z_{850}>1.7$.
341: We applied the Chi-square ($\chi^2$) test on the binned color distributions to determine the significance of the differences between the color
342: distributions of the QSO fields compared to GOODS. We
343: focused on sources with S/N $>5$ that fall in the color interval
344: $1.3<i_{775}-z_{850}< 2.6$. For J1030+0524, the chi-square test yielded
345: a $\chi^2$ statistic of 30 and a probability of $P=0.3$\%
346: where P is the one-tailed probability that obtains a value of $\chi^2$ or greater --- e.g., there is less than a 0.3\% chance that both the GOODS and J1030+0524 $i_{775}$-dropout samples were drawn from the same distribution over the color range considered.
347: For the other overdense field, J1630+4012, we found $\chi^2=52$ and $P<0.1\%$.
348: For the other three fields, $\chi^2= 11$ and $P=41$\% for J1048+4637, $\chi^2=7$ and $P=83$\% for J1148+5251, and $\chi^2=7$ and $P=83$\% for J1306+0356.
349: For two overdense fields, the probability is not more than 0.3\% regardless of the specific criterion we use (S2 and S3 samples).
350: Thus, our candidates in both overdense fields
351: have significantly different color distributions compared to GOODS.
352:
353: Figure~6 shows substantial
354: spatial clustering of the $i_{775}$-dropout candidates in the J1030+0524 field:
355: when the field is divided in half across the diagonal, almost all of the
356: sources are in the south-west half of the field. This makes the
357: excess in J1030+0524 even more significant. The color magnitude
358: diagram of candidates listed in Table 3 is presented in Figure~7, showing that
359: the overdense fields have fainter $i_{775}$-dropouts than GOODS. It is
360: notable that \citet{Willott05} in their less sensitive survey for $i_{775}$-dropouts around high-redshift SDSS QSOs, including J1030+0524, found no overdensities. The upper panel of Figure~8 shows
361: half-light radius versus $z_{850}$ for the $i_{775}$-dropout candidates from GOODS and the QSO fields. There is an upper envelope to the size-magnitude
362: relation, and the bottom panel of Figure~8 shows a histogram comparing the size distribution of GOODS and QSO field $i_{775}$-dropout half-light radii. It appears that the candidates in the overdense fields are more compact than those in GOODS, but this is not statistically significant.
363:
364: \section{DISCUSSION}
365:
366: Despite a complete reanalysis and a change in the type of SExtractor
367: magnitudes used to compute the $i_{775}-z_{850}$ color for dropout selection (from AUTO
368: to ISO mags), we confirm the overdensity in the J1030+0524 field
369: reported in \citet{Stiavelli05}.
370: The overdensity is significant not only in the counts by
371: themselves but also in the color distribution.
372: Indeed, the departure of the color distribution of J1030+0524 and J1630+4012 is in the
373: sense of having an excess of red dropouts with precisely the colors
374: that one would expect from objects at the redshift of the two QSOs.
375: This makes the excess even more convincing.
376:
377: One uncertain component of the comparison with GOODS is the
378: possible contamination by low redshift and Galactic interlopers.
379: Figure~9 shows the fraction of GOODS $i_{775}$-dropout objects selected by us but rejected when using the full GOODS $i_{775}$-dropout criteria including the $V_{606}$ data \citep{Beckwith06} to the number of GOODS $i_{775}$-dropouts selected by our criteria vs. the $i_{775}-z_{850}$ color.
380: At $i_{775}-z_{850} > 1.7$, where the excess of $i_{775}$-dropouts is large in the J1030+0524 and J1630+4012 fields, there is less than 15\% contamination from potential foreground objects. Statistically, the full GOODS criteria would remove more objects from GOODS than from the J1030+0524 or J1630+4012 field because the latter have a redder color distribution. Thus, we do not think that the detected excess is due to interloper contamination.
381:
382: In order to understand how unusual it is to identify this distribution
383: of over- and underdensities, we consider the number of $i_{775}$-dropouts
384: identified in 30 distinct and non-overlapping ACS fields in GOODS.
385: Figure~10 presents the resulting histogram of the number of
386: $i_{775}$-dropouts identified per unique GOODS ACS field using the
387: S1 and the S2 selection criteria. These distributions are reasonably
388: well fit by Poisson distributions with a mean of 6.5 (3.13)
389: $i_{775}$-dropouts per ACS field for the S1 (S2) selection criteria.
390: Using these distributions from GOODS, we create 10,000 Monte Carlo
391: (MC) quintuplets, where each MC quintuplet is generated by randomly
392: selecting five independent numbers of $i_{775}$-dropouts, each corresponding
393: to a single ACS field. We then test how many MC quintuplets have
394: the counts we have observed. For the S1, we find that only $0.06 \pm 0.02$\% of the MC
395: quintuplets have exactly two overdense and two underdense fields.
396: For the S2, this probability is only $0.03 \pm 0.09$\%.
397: For the S3, six $i_{775}$-dropouts in one ACS field is the maximum number among the 30 ACS fields in GOODS so any MC quintuplets cannot be generated to have more than six $i_{775}$-dropouts. However, since one QSO field has 10 $i_{775}$-dropouts, we have zero probability for S3.
398: The error bars on these probabilities
399: are calculated by considering variations between 10 independent
400: subsets of 1,000 MC quintuplets. This comparison to GOODS empirical
401: dropout statistics suggests that the QSOs are indeed affecting their
402: environments.
403:
404: Estimating the likelihood of the counts observed in our fields on
405: the basis of the $i_{775}$-dropout count distribution in GOODS is not
406: entirely appropriate as even GOODS is affected by cosmic
407: variance because within both the CDFS and the HDFN, the ACS fields are all
408: adjacent. We can use the conservative model of cosmic variance of \citet{TS08} to estimate the
409: likelihood of our detected counts. This model is based on extended Press-Schechter theory
410: as well as synthetic catalogs extracted from N-body simulations of structure formation.
411: In this case, we establish the probability with $10^{6}$ MC quintuplets. We find that the likelihood of a MC quintuplet matching our observed distribution of over- and underdense fields
412: using the S1 criteria is $0.9\pm0.08\%$. S2 has a likelihood of $0.3\pm0.05\%$,
413: and S3 has a likelihood of $0.8\pm0.09\%$.
414: This result is less significant than that derived from the
415: GOODS distribution, but it is comforting that the
416: significance does not decrease when using samples with more
417: stringent color or S/N selections. Thus, while we cannot claim for
418: our overall sample a very significant detection of a discrepancy
419: from a distribution dominated by cosmic variance alone, our
420: distribution remains unlikely at the 99\% level.
421:
422: A criticism to this type of analysis is that these are not a-priori
423: probabilities as we knew the outcome of the experiment before
424: carrying out the statistical tests. This is only partly correct
425: because the main idea of the HST proposal was indeed to look for
426: overdensities or underdensities compared to the field even though
427: the statistical test was not specified. Moreover, it is
428: possible to design an experiment that does not depend as much on the
429: observed counts, namely to evaluate the probability that out of the
430: five fields only one is within one (Poissonian) $\sigma$ of the
431: mean, i.e. within $8.08\pm2.84$ for selection S1, within $3.95\pm1.99$
432: for selection S2, or within $2.96\pm1.72$ for selection S3.
433: Here the formal Poisson $\sigma$ is used only to define an inner
434: interval and has no attached probability significance. Probabilities
435: are estimated by comparing how our observed object count
436: distribution compares to that expected from cosmic variance. We find
437: that the probability of finding no more than one out of five fields in the inner
438: interval is of 20\% for S1, 4\% for S2, and 5.8\% for S3. The same a priori test based on the observed counts distribution in GOODS would give a probability of finding no more than one object in the inner interval of 1.5\% for S1, 0.4\% for S2, and 1.5\% for S3.
439: This reinforces the view that the QSO fields have a distribution of $i_{775}$-dropout counts
440: broader than what is expected by cosmic variance alone.
441:
442: \section{CONCLUSIONS}
443:
444: Summarizing our results, we find two fields where the numbers of $i_{775}$-dropout galaxies and their $i_{775}-z_{850}$ color distributions are significantly different (at 99\% confidence) than the averages for galaxies selected in the same way from GOODS fields.
445: When we look at the distribution of all five fields, we find that it is likely (at $80 - 96$\% confidence level, depending on selection and specific statistical test) that the
446: distribution of counts in the QSO fields is broader than that of
447: GOODS and cannot be explained by cosmic variance alone.
448:
449: We now discuss the possible implications of our results assuming
450: that the departure from the expected distribution of
451: field $i_{775}$-dropouts is indeed real. The fact that we observe both overdensities
452: and underdensities is somewhat puzzling.
453: We know that QSOs at $z=6$ are very rare objects and are most likely associated with overdensities on large scales. Tracing a pencil beam with the area of an ACS field through a cold dark matter (CDM) simulation box with the method of Trenti \& Stiavelli (2008), we do not find correlations over $\Delta z \geq 0.3$. This is not surprising as $\Delta z=0.3$ corresponds to about 90 Mpc $h^{-1}$ at $z\sim6$ and on those scale the CDM power spectrum predicts a value of the mass fluctuation $\sigma_M$ many orders of magnitude lower than the value that can be associated with the QSO itself. From this point of view, the redshift range probed by $i_{775}$-dropouts spans at least three uncorrelated volumes.
454:
455: A QSO at $z \sim 6$ is expected to live in the most massive halos within
456: $\approx$ Gpc$^3$ comoving volumes, with masses of the order
457: of $\approx 4 \times 10^{12} h^{-1} M_{\sun}$ \citep[e.g.,][]{Springel05}. Thus the dark matter halo mass function in the vicinity of
458: the QSO halo will be biased by the presence of a rare overdensity
459: \citep[e.g.,][]{Barkana04}. To quantify the impact of the QSO on
460: the expected number counts in its immediate neighborhood we use the
461: model of Munoz \& Loeb (2007). From their Fig. 4 we derive that
462: around the QSO there should be between 6 and 7 $i_{775}$-dropouts living in
463: dark halos of mass $>5 \times 10^{10} h^{-1} M_{\sun}$ taking
464: into account an assumed duty cycle of 0.25 for LBGs.
465: The duty cycle is used to establish a halo mass scale for the observed galaxies
466: by requiring that the number of halos of the required mass be equal to
467: the number of objects divided by the duty cycle. Adopting a duty cycle allows us to determine a mass scale from the number of objects and to avoid using the ill-measured M/L of galaxies at z $\gta 6$. However, the results do not
468: depend critically on the choice of duty cycle for range between 1 and 0.1.
469: Our fields do not probably reach a depth that allows us to probe these halo masses with high
470: completeness, but still we would expect to detect 2-3 of such LBGs
471: or more if the ``duty cycle'' were higher.
472:
473: In this light, deficits in the number of $i_{775}$-dropout candidates are surprising. Indeed
474: 2/3 of the expected objects are in uncorrelated volumes and
475: should not be affected by the presence of the QSO. The one third
476: affected by the QSO now becomes a very small number and detecting a
477: deficit in any single field is generally going to be statistically
478: insignificant. It is interesting to note that at the time this
479: project was planned the expected number of $i_{775}$-dropouts in GOODS was
480: thought to be higher \citep[e.g.,][]{Dickinson04} so that a deficit
481: would have been better quantifiable.
482: Despite these considerations, the fact remains that we do seem to detect fields that have a deficit
483: of $i_{775}$-dropout counts compared to the field. If we really had physical
484: overdensities and physical underdensities near the QSO, what would
485: be the origin of this effect? One possible explanation is that two
486: physical mechanisms are simultaneously at play: the density of halos
487: near the QSO is indeed higher but feedback by the QSO prevents many
488: of these halos from becoming galaxies. The \ion{H}{2} regions generated by
489: luminous quasars can affect the formation and clustering of galaxies.
490: \citet{Wyithe05} derived \ion{H}{2} size from displacement of quasar host galaxy redshift and the Gunn-Peterson trough redshift. The \ion{H}{2} region size of the QSO J1030+0524 is the largest of the five quasars but the second overdense field, J1630+4012 and the most underdense field,
491: J1306+0356, have very similar \ion{H}{2} region sizes.
492: The field with density comparable to GOODS,
493: J1048+4637, has the smallest \ion{H}{2} region size.
494:
495: Thus, we find no evident correlation between density of $i_{775}$-dropouts and
496: \ion{H}{2} region size. This may or may not be significant as \ion{H}{2} region
497: sizes are roughly correlated with the luminosity of the quasars and
498: their lifetimes; the latter measurements are not very accurate.
499: We see a weak trend between counts and QSO luminosity as the two faintest
500: QSOs are the two overdense ones and the most luminous QSO is one of the
501: underdense ones. However, the most underdense QSO field(J1306+0356) is
502: the third luminous QSO and within 0.04 mag from that of the most
503: overdense (J1030+0524).
504: On the basis of these considerations we
505: conjecture that the suppression of galaxy formation which we may be
506: witnessing could be the result of percolation of ionized Hydrogen
507: bubbles. This would make it dependent, but not uniquely driven, by
508: the QSO properties. Clearly it would be desirable to study these
509: effects with better statistics.
510:
511: Interestingly, \citet{Maselli08}, with an entirely different method, find conclusion similar to ours: namely, that J1630+4012 is overdense while J1148+5251 and J1306+0356 are underdense. They also find an overdensity around SDSS J0836+0054, also found to be overdense by \citet{Zheng06}. \citet{Maselli08} predict that ionizing radiation from clustered galaxies for J1630+4012 exceeds the one from the quasar by a factor of five. We estimate the ionizing flux of our candidates.
512: The total UV flux observed by summing the $z_{850}$ photometry from all of our $i_{775}$-dropout candidates is 7.0\% and 8.5\% of the quasar flux in $z_{850}$ for J1030+0524 and J1630+4012, respectively.
513: For any reasonable spectral energy distribution, the excess is too small to affect the ionizing contributions. In order to have an influence
514: at this level of overdensity, the excess should span a much larger area than that provided here.
515: Clearly to further clarify these findings, we would need a larger sample as well as more extended data over overdense fields.
516:
517: \acknowledgments
518: We thank the referee for careful reading and valuable comments. This work was partially supported by HST GO grant of 01087 and 01168.
519: SGD and AAM acknowledge a partial support from the NSF grant AST-0407448,
520: and the Ajax Foundation. The work of DS was carried
521: out at the Jet Propulsion Laboratory, California Institute of Technology,
522: under a contract with NASA.
523:
524: \begin{thebibliography}{}
525:
526: \bibitem[Barkana \& Loeb(2004)]{Barkana04} Barkana, R., \& Loeb,
527: A.\ 2004, \apj, 609, 474
528:
529: \bibitem[Barth et al.(2003)]{Barth03} Barth, A.~J., Martini,
530: P., Nelson, C.~H., \& Ho, L.~C.\ 2003, \apjl, 594, L95
531:
532: \bibitem[Becker et al.(2001)]{Becker01} Becker, R.~H., et al.\
533: 2001, \aj, 122, 2850
534:
535:
536: \bibitem[Beckwith et al.(2006)]{Beckwith06} Beckwith, S.~V.~W., et
537: al.\ 2006, \aj, 132, 1729
538:
539:
540: \bibitem[Bertin \& Arnouts(1996)]{Bertin96} Bertin, E., \&
541: Arnouts, S.\ 1996, \aaps, 117, 393
542:
543: \bibitem[Bolton \& Haehnelt(2007)]{Bolton07} Bolton, J.~S., \& Haehnelt, M.~G.\ 2007, \mnras, 374, 493
544:
545:
546: \bibitem[Bouwens et al.(2007)]{Bouwens07} Bouwens, R.~J.,
547: Illingworth, G.~D., Franx, M., \& Ford, H.\ 2007, \apj, 670, 928
548:
549: \bibitem[Bunker et al.(2004)]{Bunker04} Bunker, A.~J., Stanway,
550: E.~R., Ellis, R.~S., \& McMahon, R.~G.\ 2004, \mnras, 355, 374
551:
552: \bibitem[Cen(2003)]{Cen03} Cen, R.\ 2003, \apj, 591, 12
553:
554: \bibitem[Ciardi et al.(2003)]{Ciardi03} Ciardi, B., Ferrara, A.,
555: \& White, S.~D.~M.\ 2003, \mnras, 344, L7
556:
557: \bibitem[Dickinson et al.(2004)]{Dickinson04} Dickinson, M., et
558: al.\ 2004, \apjl, 600, L99
559:
560:
561: \bibitem[Djorgovski(1999)]{Djorgo99} Djorgovski, S.~G.\ 1999,
562: The Hy-Redshift Universe: Galaxy Formation and Evolution at High Redshift,
563: 193, 397
564:
565:
566: \bibitem[Djorgovski et al.(2001)]{Djorgovski01} Djorgovski, S.~G.,
567: Castro, S., Stern, D., \& Mahabal, A.~A.\ 2001, \apjl, 560, L5
568:
569:
570: \bibitem[Djorgovski et al.(1999)]{Djorgovski99} Djorgovski, S.~G.,
571: Odewahn, S.~C., Gal, R.~R., Brunner, R.~J., \& de Carvalho, R.~R.\ 1999,
572: Photometric Redshifts and the Detection of High Redshift Galaxies, 191, 179
573:
574:
575: \bibitem[Djorgovski et al.(2003)]{Djorgovski03} Djorgovski, S.~G.,
576: Stern, D., Mahabal, A.~A., \& Brunner, R.\ 2003, \apj, 596, 67
577:
578:
579: \bibitem[Efstathiou \& Rees(1988)]{Efstathiou88} Efstathiou, G., \&
580: Rees, M.~J.\ 1988, \mnras, 230, 5P
581:
582:
583: \bibitem[Fan et al.(2001)]{Fan01} Fan, X., et al.\ 2001, \aj,
584: 122, 2833
585:
586: \bibitem[Fan et al.(2003)]{Fan03} Fan, X., et al.\ 2003, \aj,
587: 125, 1649
588:
589: \bibitem[Fan et al.(2006)]{Fan06} Fan, X., et al.\ 2006, \aj,
590: 132, 117
591:
592:
593: \bibitem[Fruchter \& Hook(2002)]{Fruchter02} Fruchter, A.~S., \&
594: Hook, R.~N.\ 2002, \pasp, 114, 144
595:
596:
597: \bibitem[Giavalisco et al.(2004)]{Giavalisco04} Giavalisco, M., et
598: al.\ 2004, \apjl, 600, L93
599:
600: \bibitem[Gnedin(2004)]{Gnedin04} Gnedin, N.~Y.\ 2004, \apj, 610,
601: 9
602:
603: \bibitem[Gnedin \& Ostriker(1997)]{Gnedin97} Gnedin, N.~Y., \&
604: Ostriker, J.~P.\ 1997, \apj, 486, 581
605:
606:
607: \bibitem[Gunn \& Peterson(1965)]{Gunn65} Gunn, J.~E., \&
608: Peterson, B.~A.\ 1965, \apj, 142, 1633
609:
610:
611: \bibitem[Haiman \& Holder(2003)]{Haiman03} Haiman, Z., \&
612: Holder, G.~P.\ 2003, \apj, 595, 1
613:
614: \bibitem[Kaiser(1984)]{Kaiser84} Kaiser, N.\ 1984, \apjl, 284, L9
615:
616: \bibitem[Koekemoer et al.(2002)]{Koekemoer02} Koekemoer, A.~M.,
617: Fruchter, A.~S., Hook, R.~N., \& Hack, W.\ 2002, The 2002 HST Calibration
618: Workshop : Hubble after the Installation of the ACS and the NICMOS Cooling
619: System, Proceedings of a Workshop held at the Space Telescope Science
620: Institute, Baltimore, Maryland, October 17 and 18, 2002.~ Edited by
621: Santiago Arribas, Anton Koekemoer, and Brad Whitmore.~Baltimore, MD: Space
622: Telescope Science Institute, 2002., p.337, 337
623:
624: \bibitem[Komatsu et al.(2008)]{Komatsu08} Komatsu, E., et al.\
625: 2008, ArXiv e-prints, 803, arXiv:0803.0547
626:
627: \bibitem[Lidz et al.(2006)]{Lidz06} Lidz, A., Oh, S.~P.,
628: \& Furlanetto, S.~R.\ 2006, \apjl, 639, L47
629:
630: \bibitem[Malhotra et al.(2005)]{Malhotra05} Malhotra, S., et al.\
631: 2005, \apj, 626, 666
632:
633: \bibitem[Maselli et al.(2008)]{Maselli08} Maselli, A., et al.\
634: 2008, \mnras, submitted
635:
636: \bibitem[McDonald \& Miralda-Escud{\'e}(2001)]{McDonald01}
637: McDonald, P., \& Miralda-Escud{\'e}, J.\ 2001, \apjl, 549, L11
638:
639:
640: \bibitem[Miralda-Escud{\'e} et al.(2000)]{Miralda00}
641: Miralda-Escud{\'e}, J., Haehnelt, M., \& Rees, M.~J.\ 2000, \apj, 530, 1
642:
643:
644: \bibitem[Mu{\~n}oz \& Loeb(2008)]{Mu07} Mu{\~n}oz, J.~A., \& Loeb, A.\ 2008, \mnras, 385, 2175
645:
646: \bibitem[Oesch et al.(2007)]{Oesch07} Oesch, P.~A., et al.\
647: 2007, \apj, 671, 1212
648:
649: \bibitem[Ouchi et al.(2005)]{Ouchi05} Ouchi, M., et al.\ 2005,
650: \apjl, 620, L1
651:
652: \bibitem[Schlegel et al.(1998)]{Schlegel98} Schlegel, D.~J.,
653: Finkbeiner, D.~P., \& Davis, M.\ 1998, \apj, 500, 525
654:
655: \bibitem[Shapiro \& Raga(2001)]{Shapiro01} Shapiro, P.~R., \&
656: Raga, A.~C.\ 2001, Revista Mexicana de Astronomia y Astrofisica Conference
657: Series, 10, 109
658:
659: \bibitem[Shull \& Venkatesan(2007)]{Shull07} Shull, M., \&
660: Venkatesan, A.\ 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0702323
661:
662: \bibitem[Somerville et al.(2003)]{Somerville03} Somerville, R.~S.,
663: Bullock, J.~S., \& Livio, M.\ 2003, \apj, 593, 616
664:
665: \bibitem[Springel et al.(2005)]{Springel05} Springel, V., et al.\
666: 2005, \nat, 435, 629
667:
668: \bibitem[Steidel et al.(2003)]{Steidel03} Steidel, C.~C.,
669: Adelberger, K.~L., Shapley, A.~E., Pettini, M., Dickinson, M., \&
670: Giavalisco, M.\ 2003, \apj, 592, 728
671:
672: \bibitem[Stiavelli et al.(2005)]{Stiavelli05} Stiavelli, M., et
673: al.\ 2005, \apjl, 622, L1
674:
675: \bibitem[Stiavelli et al.(2004)]{Stiavelli04} Stiavelli, M., Fall,
676: S.~M., \& Panagia, N.\ 2004, \apjl, 610, L1
677:
678: \bibitem[Trenti \& Stiavelli(2007)]{TS07} Trenti, M., \&
679: Stiavelli, M.\ 2007, \apj, 667, 38
680:
681: \bibitem[Trenti \& Stiavelli(2008)]{TS08} Trenti, M., \&
682: Stiavelli, M.\ 2008, \apj, 676, 767
683:
684: \bibitem[Venemans et al.(2003)]{Venemans03} Venemans, B.~P., Kurk,
685: J.~D., Miley, G.~K., \& R{\"o}ttgering, H.~J.~A.\ 2003, New Astronomy
686: Review, 47, 353
687:
688:
689: \bibitem[White et al.(2003)]{White03} White, R.~L., Becker,
690: R.~H., Fan, X., \& Strauss, M.~A.\ 2003, \aj, 126, 1
691:
692: \bibitem[Willott et al.(2005)]{Willott05} Willott, C.J., et al.,
693: 2005, in ``Growing Black Holes'', ESO (Garching), Merloni et
694: al. eds, (also astro-ph/0410306)
695:
696: \bibitem[Wyithe \& Loeb(2003)]{Wyithe03} Wyithe, J.~S.~B., \&
697: Loeb, A.\ 2003, \apjl, 588, L69
698:
699:
700: \bibitem[Wyithe et al.(2005)]{Wyithe05} Wyithe, J.~S.~B., Loeb,
701: A., \& Carilli, C.\ 2005, \apj, 628, 575
702:
703:
704: \bibitem[Yan \& Windhorst(2004)]{Yan04} Yan, H., \&
705: Windhorst, R.~A.\ 2004, \apjl, 600, L1
706:
707: \bibitem[Zheng et al.(2006)]{Zheng06} Zheng, W., et al.\ 2006,
708: \apj, 640, 574
709:
710: \bibitem[Zombeck(1990)]{Zombeck90} Zombeck, M.V. 1990, Handbook of Space Astronomy and Astrophysics
711: (Cambridge University Press: Cambridge), p.77
712:
713:
714: \end{thebibliography}
715:
716: \clearpage
717:
718: \input{tab1}
719: \input{tab2}
720: \clearpage
721: \input{tab3}
722: \clearpage
723:
724: \begin{figure}
725: \epsscale{.80}
726: \plotone{f1.eps}
727: \caption{ Distributions of the number of objects
728: versus magnitude and color, comparing objects in the SDSS J1030+5032
729: field (solid histograms) to those in the GOODS field (dotted
730: histograms). The GOODS counts are normalized to the area of the
731: quasar field ($\sim$11.3 arcmin$^{2}$). Panel (a) shows the total
732: counts vs. $z_{850}(AUTO)$ of all objects with no
733: selection criteria applied. Panel (b) shows the total counts vs.
734: $i_{775}$-$z_{850}$ of all objects with no selection
735: criteria applied. Panel (c) shows the candidates with
736: $i_{775}$-$z_{850}>1.5$; the brightest object is the target QSO.
737: Panel (d) shows the number of objects redder than a given
738: $i_{775}$-$z_{850}$, excluding the target QSO, with the GOODS counts renormalized to the QSO counts at $0.9<i_{775}$-$z_{850}<1.0$.}
739: \end{figure}
740: \clearpage
741:
742: \begin{figure}
743: \epsscale{.80}
744: \plotone{f2.eps}
745: \caption{ Same as Fig 1. for J1630+4012 }
746: \end{figure}
747: \clearpage
748:
749: \begin{figure}
750: \epsscale{.80}
751: \plotone{f3.eps}
752: \caption{ Same as Fig 1. for J1048+4637 }
753: \end{figure}
754: \clearpage
755:
756: \begin{figure}
757: \epsscale{.80}
758: \plotone{f4.eps}
759: \caption{ Same as Fig 1. for J1148+5251 }
760: \end{figure}
761: \clearpage
762:
763: \begin{figure}
764: \epsscale{.80}
765: \plotone{f5.eps}
766: \caption{ Same as Fig 1. for J1306+0356 }
767: \end{figure}
768: \clearpage
769:
770: \begin{figure}
771: \epsscale{0.8}
772: \plotone{f6.eps}
773: \caption{ The spatial distribution of $i_{775}$-dropouts redder than 1.3 (circles) and
774: redder than 1.5 (open squares) for S/N$>5$ in the J1030+0524 field.
775: The star represents the QSO SDSS J1030+0524. The axes of x and y are in arcseconds and they are relative to the QSO position. East is to the left and north is up.}
776: \end{figure}
777: \clearpage
778:
779: \begin{figure}
780: \epsscale{0.8}
781: \plotone{f7.eps}
782: \caption{ Color-magnitude distribution of $i_{775}$-dropouts, selected using S1 criteria, in GOODS (small black dots) and the five QSO fields. The overdense QSO fields are indicated with red diamonds, the underdense QSO fields are indicated with blue asterisks, and the intermediate density QSO field is indicated with green crosses.
783: The brightest objects, with $z_{850} \lta 21$, are the QSOs.}
784: \end{figure}
785: \clearpage
786:
787: \begin{figure}
788: \epsscale{0.8}
789: \plotone{f8.eps}
790: \caption{ The {\it upper panel} shows half-light radii
791: of $i_{775}$-dropout candidates (S1 sample) with respect to $z_{850}$ magnitude for excess (red
792: diamonds), deficit (blue asterisks), and intermediate (green crosses) density fields, as well as for GOODS (black dots). The five bright objects at $z_{850} \lta 21$ are the QSOs. The solid line is fitted to the upper limit of the logarithmic radii of the objects in GOODS. The {\it lower
793: panel} shows the histogram of the distances from the
794: upper envelope on the upper panel to the logarithmic radii data of the
795: all five QSO fields (orange line) and the GOODS field (black line) along the axis of ordinates. The histogram of GOODS is normalized to the area of the QSO fields.}
796: \end{figure}
797: \clearpage
798:
799: \begin{figure}
800: \epsscale{0.8}
801: \plotone{f9.eps}
802: \caption{The ratio of the number of GOODS $i_{775}$-dropouts selected by us (S1) but rejected by the full GOODS selection criteria, including S/N$(V_{606})<2.0$ or $V_{606}-z_{850}<2.8$ \citep{Beckwith06} to the number of GOODS $i_{775}$-dropouts selected by our criteria. }
803: \end{figure}
804: \clearpage
805:
806: \begin{figure}
807: \epsscale{0.8}
808: \plotone{f10.eps}
809: \caption{ Histograms ({\it upper panels})
810: of the number of $i_{775}$-dropouts per ACS field in the GOODS North and
811: South (Solid lines) for S1 sample (on the left) and S2 sample (on the
812: right). Poisson distributions (dashed lines) with a mean of 6.5 for $i_{775}-z_{850}>1.3$ and 3.13 for $i_{775}-z_{850}> 1.5$ are fitted. The 30 ACS fields are overlaid in the whole GOODS field. The {\it bottom panels} show cumulative distributions from the histograms}
813: \end{figure}
814: \clearpage
815:
816:
817: \end{document}
818: