astro-ph0511423/ms.tex
1: \documentclass{emulateapj}
2: 
3: \newcommand{\myemail}{amber.straughn@asu.edu}
4: \shortauthors{Straughn et al.}
5: \shorttitle{Tadpole Galaxies in HUDF}
6: 
7: 
8: 
9: \newcommand{\etal}	{\mbox{et al.\,}}
10: \newcommand{\ie}	{\mbox{i.e.}}
11: \newcommand{\eg}	{\mbox{e.g.}}
12: \newcommand{\Mo}	{\mbox{M$_{\odot}$}}
13: \newcommand{\cle}	{\ensuremath{\lesssim}}
14: \newcommand{\cge}	{\ensuremath{\gtrsim}}
15: \newcommand{\absPA}	{\ensuremath{|{\rm PA}|}}
16: 
17: \newcommand{\fig}[1]{Figure~\ref{#1}}
18: \newcommand{\tab}[1]{Table~\ref{#1}}
19: 
20: 
21: \begin{document}
22: 
23: \title{Tracing Galaxy Assembly: Tadpole Galaxies in the Hubble Ultra-Deep Field}
24: 
25: \author{Amber N. Straughn, Seth H. Cohen, Russell E. Ryan Jr, Nimish P. Hathi,
26: Rogier A. Windhorst, \& Rolf A. Jansen} 
27: 
28: \affil{Department of Physics and Astronomy, Arizona State
29: University, Tempe, AZ 85281; \myemail}
30: 
31: \begin{abstract}
32: In the Hubble Ultra Deep Field (HUDF) an abundance of galaxies is seen
33: with  a  knot  at  one   end  plus  an  extended  tail,  resembling  a
34: tadpole. These  ``tadpole galaxies'' appear  dynamically unrelaxed ---
35: presumably  in an  early merging  stage ---  where  tidal interactions
36: likely  created  the  distorted  knot-plus-tail morphology.   Here  we
37: systematically select  tadpole galaxies from the HUDF  and study their
38: properties  as  a  function  of  their photometric  redshifts.   In  a
39: companion HUDF variability study  presented in this issue, Cohen \etal
40: (2005) revealed a total  of 45 variable objects believed  to be Active
41: Galactic Nuclei (AGN).  Here we show that this faint AGN sample has no
42: overlap  with  the  tadpole  galaxy  sample, as  predicted  by  recent
43: theoretical work.   The tadpole morphology --- combined  with the lack
44: of overlap with the variable  objects --- supports the idea that these
45: galaxies are in the process of an early-stage merger event, i.e., at a
46: stage that  likely precedes the  ``turn-on'' of any AGN  component and
47: the onset of any point-source  variability.  We show that the redshift
48: distribution  of tadpole galaxies  follows that  of the  general field
49: galaxy population, indicating that --- if most of the tadpole galaxies
50: are  indeed  dynamically young  ---  the  process  of galaxy  assembly
51: generally kept up with the  reservoir of available field galaxies as a
52: function of  cosmic epoch.  These new  observational results highlight
53: the importance  of merger-driven processes  throughout cosmic history,
54: and  are  consistent  with  a  variety of  theoretical  and  numerical
55: predictions.
56: 
57: \end{abstract}
58: 
59: \keywords{galaxies: formation --- galaxies: mergers --- galaxies: active
60: galactic nuclei --- cosmology}
61: 
62: 
63: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
64: 
65: \section{Introduction} \label{introduction}
66: The origin of disk galaxies has long been thought to occur through the
67: process of dissipational collapse in a Cold Dark Matter (CDM) universe
68: (White \& Rees, 1978).   Within this paradigm, hierarchical clustering
69: (Navarro, Frenk,  \& White 1997)  produces dark matter halos  in which
70: dissipational  collapse of  the  residual gas  occurs.  The  resulting
71: disks  retain the  kinematic  information of  their  host dark  matter
72: potential wells (Blumenthal  \etal 1986). Recent numerical simulations
73: have  resolved  some   long-standing  discrepancies  in  the  standard
74: dissipational  collapse  scenario  by  including  previously-neglected
75: energetic feedback from central supermassive black holes during galaxy
76: merging  events  (e.g. Robertson  \etal  2005).   In particular,  they
77: emphasize the relationship between the central black hole mass and the
78: stellar  velocity  dispersion, which  confirms  the  link between  the
79: growth of black holes and their host galaxies (di Matteo, Springel, \&
80: Hernquist  2005; Springel,  di  Matteo, \&  Hernquist 2005ab).   These
81: theoretical   predictions  place   merger-driven   scenarios  on   the
82: forefront, suggesting that galaxy merger activity is a crucial element
83: in  a cosmological  description of  the Universe.   The  present study
84: provides   observational  support  for   many  of   these  theoretical
85: predictions.
86: 
87: A large  abundance of galaxies in  the Hubble Ultra  Deep Field (HUDF;
88: Beckwith \etal 2005) appear dynamically unrelaxed, which suggests they
89: must  play  an  important  role  in  the  overall  picture  of  galaxy
90: evolution.    In  particular,   we   notice  many   galaxies  with   a
91: knot-plus-tail morphology.   This particular morphology  constitutes a
92: large, well-defined  subset of the  irregular and peculiar  objects in
93: the HUDF that  is uniquely measurable as described  in Section 3.  The
94: selection  of  this  specific  morphology  also ties  closely  to  the
95: numerical  simulations  described   above  (di  Matteo,  Springel,  \&
96: Hernquist  2005;  Springel, di  Matteo,  \&  Hernquist 2005ab),  which
97: predict  a  stage  of  merger-driven  galaxy  evolution  that  closely
98: resembles these tadpole galaxies in a distinct phase that does not yet
99: show AGN activity (as discussed further in Section 7).  In particular,
100: this  morphology appears  to represent  an \emph{early}  stage  in the
101: merging of  2 nearly-equal mass galaxies.   We systematically selected
102: galaxies displaying  this knot-plus-tail  morphology from the  HUDF; a
103: representative sample  of tadpoles is shown  in \fig{fig:tp1} (details
104: of sample selection are given in Section 3).  All the selected tadpole
105: galaxies contain the asymmetric,  pointlike source with a diffuse tail
106: morphology,  some with  multiple knots;  all of  which we  believe are
107: undergoing  recent interactions.  They  are mostly  linear structures,
108: some resembling the ``chain'' galaxies first reported by Cowie, Hu, \&
109: Songaila  (1995).  When  more  than two  clumps  come together,  these
110: objects may  be more  akin to the  luminous diffuse objects  and clump
111: clusters (Conselice \etal 2004;  Elmegreen, Elmegreen, \& Sheets 2004;
112: Elemegreen,  Elmegreen, \& Hirst  2004), or  other types  of irregular
113: objects  (van  den  Bergh  2002).   Elmegreen  \etal  (2005)  visually
114: classify 97 HUDF galaxies (down  to 10 pixels in size) as ``tadpoles''
115: and  126 as  ``double-clump.''   Some of  the  galaxies classified  by
116: Elmegreen \etal as ``double-clump'' were identified as tadpoles by our
117: code, due  either to the unresolved  nature of one  clump (which would
118: have been  detected as a ``tail''  in our analysis) or  to the diffuse
119: nature of  one end of the  object.  Since our goal  in selecting these
120: tadpoles  was to  sample galaxies  that had  \emph{recently} undergone
121: interaction,   inclusion  of  some   of  these   ``double-clumps''  is
122: warranted.  One high redshift object in our sample has been studied in
123: detail by Rhoads \etal (2005).   A few objects with multiple knots are
124: detected by  our selection  software, but the  majority have  a single
125: prominant knot with an extended tail.
126: 
127: In  this paper, we  present the  photometric redshift  distribution of
128: tadpole galaxies, and compare it with the redshift distribution of the
129: general  field   galaxy  population.   This  paper  is   organized  as
130: follows. In \S 2,  we describe the HUDF data, and in  \S 3 the tadpole
131: sample  selection. In \S  4, we  discuss why  the majority  of tadpole
132: galaxies are  likely \emph{not} chance alignments,  but instead mostly
133: dynamically  young  objects.  In  \S  5,  we  discuss  their  redshift
134: distribution, in \S  6 their relation to galaxy assembly,  and in \S 7
135: their possible relation to AGN growth.
136: 
137: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
138: 
139: \section{Hubble Ultra Deep Field Data} \label{data}
140: 
141: The Hubble  Ultra Deep  Field is  a 400 orbit  survey in  four filters
142: carried out  using the  Advanced Camera for  Surveys (ACS)  aboard the
143: \emph{Hubble Space Telescope} (\emph{HST})  of a single field centered
144: on    RA(J2000)=$03^{\rm   h}    32^{\rm    m}   39.\!\!^{\rm    s}0$,
145: Dec(J2000)=$-27^{\circ}  47' 29\farcs1$.   The 144-orbit  F775W ($i'$)
146: image is deepest,  followed by F850LP ($z'$; 144  orbits), F606W ($V$;
147: 56 orbits), and F435W ($B$; 56 orbits). The HUDF reaches $\sim$1.0 mag
148: deeper  in $B$  and $V$  and  $\sim$1.5 mag  deeper in  $i'$ than  the
149: equivalent  filters   in  the   Hubble  Deep  Field   (Williams  \etal
150: 1996). From  the $\sim$10,000 objects detected in  the HUDF (Koekemoer
151: 2004), we will select the sample of tadpole galaxies and analyze their
152: properties using the $i'$-band  image, because it provides the highest
153: sensitivity of the four filters.  Yan \& Windhorst (2004b) discuss how
154: this results in a bias  against objects at z$\,\gtrsim5.5$.  This bias
155: is small,  and only  concerns the high  redshift tail of  the redshift
156: distribution. Note,  however, that tadpole  galaxies at z$\,\simeq5.5$
157: \emph{do} exist (\eg, Rhoads \etal 2005).
158: 
159: 
160: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
161: 
162: \section{Tadpole Sample Selection} \label{sampleselection}
163: 
164: The  first step  in  this  analysis is  to  systematically select  the
165: galaxies  that have  the  characteristic tadpole  shape.  We  selected
166: sources  in  the  F775W   ($i'$)  band  to  $i'_{AB}=28.0$  mag  using
167: \texttt{SExtractor} (Bertin \& Arnouts 1996).  The objects of interest
168: all have a bright ``knot'' at one end with an extended ``tail'' at the
169: other.  \texttt{SExtractor} selects  objects  from an  image based  on
170: different input parameters, and  adjusting them results in the desired
171: selection  of  sources.  The   crucial  parameter  at  this  stage  is
172: \textsf{\small  DEBLEND\_MINCONT}, which governs  the manner  in which
173: nearby peaks in  flux are considered part of a  single object and thus
174: are counted as one source. With the deblending parameter set to a high
175: value,  \texttt{SExtractor}  will  separate  nearby flux  maxima  into
176: separate sources.  In contrast,  when the deblending  is set to  a low
177: value,  the  program will  count  the  nearby  maxima largely  as  one
178: source. Two  different source catalogs are thus  generated: the highly
179: deblended catalog will contain  many point-like sources, including the
180: knots of potential tadpole  galaxies.  The catalog with low-deblending
181: will  contain  extended   sources,  including  the  tadpole  galaxies'
182: tails. The catalogs  contain many more sources than  the desired ones,
183: and  the correctly  shaped objects  must  be selected  from these  two
184: initial  catalogs.   The  desired  tadpole  galaxies   have  a  nearly
185: unresolved knot or concentration, and an extended tail, so these types
186: of  sources must  be selected  from the  initial catalog,  and related
187: spatially such that they represent  real objects. All of the following
188: procedures were performed using IDL.
189: 
190: Both input \texttt{SExtractor}  catalogs, described above, contain the
191: following information  about the selected  sources: $x$ and  $y$ pixel
192: locations, length of the semi-major  and semi-minor axis ($a$ and $b$)
193: of  \texttt{SExtractor}  ellipses, and  the  angle  ($\theta$) of  the
194: semi-major  axis   from  north  through  east.   The  following  input
195: parameters were  calibrated using a  training set of  tadpole galaxies
196: that were  manually selected by  visual inspection of portions  of the
197: HUDF.   First, the  knots of  the  tadpole galaxies  were selected  by
198: setting an  axis-ratio limit.  A ``knot'' was  defined to be  a source
199: from the  highly deblended  catalog with an  axis ratio  greater (\ie,
200: rounder) than  some critical value  (in our case, $b/a>0.70$).  In the
201: same way, the  ``tails'' needed to be elongated  objects, so a similar
202: procedure  was  performed  on   the  objects  from  the  catalog  with
203: low-deblending, but with the criterion that their $b/a< 0.43$. The two
204: new lists of correctly shaped  objects had to be related physically on
205: the image,  thus, a new  set of objects  was defined where a  knot was
206: within a certain  distance of the geometrical center  of a tail.  This
207: distance  was taken  to  be $<4a$  (in  semi-major axis  units of  the
208: tail). We  also required that  the knot be  at least $>0.1a$  from the
209: tail's  geometrical  center, since  we  are  searching for  asymmetric
210: objects, and  want to  eliminate upfront as  many of the  true edge-on
211: mid-type spiral disks as possible. The objects also must have the knot
212: near one end of the tail,  and this was accomplished by selecting only
213: those tails and knots that  had a relative angle $\theta$ --- measured
214: with  respect  to  the  semi-major  axis  of the  tail  ---  that  was
215: $\leq20^{\circ}$. This  step prevented including knots  and tails that
216: were close together on the image,  but not physically part of the same
217: galaxy.
218: 
219: 
220: \begin{table}
221: \caption{Galaxy Selector Input Parameters\tablenotemark{\dag}}
222: \label{table:param}
223: \begin{tabular*}{0.48\textwidth}
224:    {@{\extracolsep{\fill}}lr}
225: \hline
226: \hline
227: \multicolumn{1}{l}{Parameter} & \multicolumn{1}{r}{Value}\\
228: $ $&$ $\\
229: \hline
230: $b/a$ limit: knots & $>$0.70 \\
231: $b/a$ limit: tails & $<$0.43 \\
232: Distance to center (in a-axis units) & $<$4 \\
233: Angle difference $\theta$ (tail-knot) & $\leq20^{\circ}$ \\
234: Total number of tadpoles automatically selected & 154 \\
235: DEBLEND\_MINCONT (knots) & 0.000005 \\
236: DEBLEND\_MINCONT (tails) & 0.1 \\
237: \hline
238: \tablenotetext{\dag}{Table 1 lists \texttt{SExtractor} parameters used
239: in  the tadpole  sample  selection.   $b$ \&  $a$  are semi-major  and
240: semi-minor  axis lengths  of the  ellipses used  in  selecting tadpole
241: knots    and   tails.     DEBLEND\_MINCONT   (knots/tails)    is   the
242: \texttt{SExtractor} parameter that governs in what manner nearby peaks
243: in flux are considered part of a single object and are thus counted as
244: one source.}
245: \end{tabular*}
246: \end{table}
247: 
248: 
249: These  selection  criteria  provided  a  list  of  the  tadpole-shaped
250: galaxies.  The  final number  of  tadpoles  selected  depended on  the
251: selection program's input parameters for the limiting axis-ratios, the
252: distance at which the knots and tails were considered related, and the
253: angle  difference between  the knot  and  the semi-major  axis of  the
254: tail. The values of these parameters  are given in Table 1. With these
255: values,  the tadpole  galaxy  selection program  detected 154  sources
256: total. This sample was then refined as follows.
257: 
258: A large majority of the 154 tadpole galaxy candidates selected had the
259: characteristic  elongated  knot-plus-tail  morphology, although  there
260: were some  anomalies.  In total, 14 (9\%)  obvious mis-detections were
261: visually rejected  because they  were very faint,  on the edge  of the
262: image, or  in the  outskirts of large  face-on spiral  galaxies, where
263: both  knotty  and  diffuse  regions  are common  and  spatially  close
264: together.  A  visual examination  of the field  also produced  25 more
265: tadpole  galaxies  not  found  by  the selector  program  due  to  the
266: inability  of  \texttt{SExtractor}  to correctly  separate  particular
267: point-like  sources  within  these  galaxies.   These  extra  selected
268: tadpole objects  visually obeyed the morphological  criteria that were
269: used to define  the main sample. Our total  final sample thus contains
270: 165 tadpole galaxies, a subset of which is shown in \fig{fig:tp1}.  In
271: our final  sample, less than 10\%  of the selected  tadpoles appear as
272: normal  edge-on  disk galaxies;  the  vast  majority  have the  highly
273: asymmetric morphology.   In terms of visual  vs.\ automatic selection,
274: we find our sample to be  about 91\% (140/154) reliable and about 86\%
275: (154/179) complete.  The  final set of 165 tadpoles  galaxies will now
276: be studied  as a separate  class of dynamically unrelaxed  objects and
277: compared to the general field galaxy population in the HUDF.
278: 
279: \begin{figure*}
280: \centerline{\epsfxsize=\hsize{\epsfbox{f1.ps}}}
281: \caption{F775W ($i'$)  band mosaic of  a subset of the  tadpole galaxy
282: sample in the HUDF.  Stamps  retain the orientation of the HUDF; north
283: is toward the top of the page  and east is to the left.  Index numbers
284: are  displayed in  the upper-left  corner of  the  stamps; photometric
285: redshifts  are  given in  upper-right  corners.  Parentheses  indicate
286: photometric redshifts  with errors $>$1 based  on HyperZ calculations.
287: Stamps are 3 arcsec on a  side.  A table of coordinates for the entire
288: tadpole  sample  is given  in  Section 3;  the  entire  sample of  165
289: tadpoles appears in color in the online supplement.  The vast majority
290: of  our   tadpole  sample  contains   the  distinctive  knot-plus-tail
291: morphology,  while sample  contamination  by normal  (non-interacting)
292: edge-on disk galaxies is less than 10\%.  }\label{fig:tp1}
293: \end{figure*}
294: 
295: 
296: 
297: \section{Why Tadpole Galaxies are Not Chance Alignments}
298: \label{whyyoung}
299: 
300: In this section we demonstrate  that these tadpole galaxies are likely
301: not chance alignnments of tails  and unrelated knots.  We first select
302: \emph{all} elongated  diffuse structures (``tails'') in  the HUDF, and
303: then  measure the  angle  $\theta$ of  the  nearest off-centered  knot
304: within  a  radius  $r\leq  4a$  ($\leq 2''$).   Chance  alignments  of
305: unrelated tails and knots would  show a random distribution of angles;
306: however,  \fig{fig:theta} shows  that there  clearly is  an  excess of
307: knots  near $|\theta|\simeq  0^\circ$.  The  excess peak  contains 154
308: knots, while  the average number of knots  with $|\theta|\ge 10^\circ$
309: is $\sim$15  per $5^\circ  \theta$-bin.  \fig{fig:theta} thus  shows a
310: significant overabundance  of knots near the end  of elongated diffuse
311: structures  as compared  to randomly  distributed knots.   Hence, this
312: physically  meaningful result  suggests that  the majority  of tadpole
313: galaxies are  not just chance alignments of  unrelated knots. Instead,
314: we  believe they  are mostly  linear structures  which  are undergoing
315: interactions.  When  compared to models of galaxy  mergers (di Matteo,
316: Springel, \&  Hernquist 2005; Springel, di Matteo,  \& Hernquist 2005;
317: Robertson  \etal 2005,  Hopkins  \etal 2005),  these objects  strongly
318: resemble dynamically young objects in the early stages of merging.
319: \begin{figure} 
320: \epsscale{1.2}
321: \plotone{f2.ps}
322: \caption{Distribution of (the absolute  value of) angles ($\theta$) of
323: all off-centered knots  found within a radius $r\leq  4a$ ($\leq 2''$)
324: from  the center  of each  elongated  diffuse structure  in the  HUDF,
325: showing  a  clear  excess  of knots  near  $|\theta|\simeq  0^\circ$.}
326: \label{fig:theta}
327: \end{figure}
328: 
329: 
330: %%%%%%%%%%%%%%%%%%%%%%%
331: 
332: \section{The Redshift Distribution of Tadpole Galaxies} 
333: \label{redshiftdistribution}
334: 
335: To  investigate  the occurrence  of  tadpole  galaxies throughout  the
336: history  of the universe,  we calculate  photometric redshifts  of all
337: HUDF galaxies  to $i'_{AB}=28.0$ mag.  All  photometric redshifts were
338: calculated  from  the   HUDF  $BViz$(+$JH$)  photometry  using  HyperZ
339: (Bolzonella \etal  2000). In order to  investigate associated redshift
340: errors,   we   compared  our   photometric   redshifts  to   published
341: spectroscopic redshifts for CDFS 70 objects. We find an rms scatter of
342: 0.15    for     the    fractional    photometric     redshift    error
343: $\delta$=(photoz-specz)/(1+specz) if all  70 objects are included, and
344: 0.10 when we  reject a few of the most  obvious outliers.  This result
345: is fully consistent with prior claims of photometric redshift accuracy
346: in  the literature (Lanzetta  \etal 1997,  Mobasher \etal  2004).  The
347: accuracy of our photometric redshift estimates depends on the accuracy
348: of the measured magnitudes in  each of the available filters.  It also
349: is largely independent  of the shape of an  object (although magnitude
350: errors for more extended, lower  surface brightness objects tend to be
351: somewhat  larger  than those  for  more  concentrated, higher  surface
352: brightness objects of the same total magnitude).
353: 
354: The redshift distribution  of all galaxies in the  HUDF (solid line in
355: \fig{fig:hist})   is   as  expected,   with   the   primary  peak   at
356: $0.5\leq\,$z$\,\leq1.0$   and   a    generally   declining   tail   at
357: z$\,\simeq4$--5.   These trends  were  also seen  in  the general  HDF
358: redshift  distribution of  faint field  galaxies (Driver  \etal 1998).
359: Also  apparent  is  a  lack  of  objects  at  z$\,\simeq1$--2  due  to
360: unavailable   UV   spectral   features  crossing   the   $BViz$(+$JH$)
361: filters. This occurs because the  HUDF does not have deep enough F300W
362: or  $U$-band (ultra-violet)  data,  unlike the  situation  in the  HDF
363: (Williams \etal 1996). This redshift bias, however, is the \emph{same}
364: for  both the  tadpole and  the general  field galaxy  populations. In
365: \fig{fig:hist},  the tadpole  galaxy distribution  is multiplied  by a
366: factor  of 16 for  best comparison  with that  of the  field galaxies.
367: Within the  available statistics,  the redshift distribution  shape of
368: the tadpole galaxies follows that  of the general field galaxies quite
369: closely. This suggests that if tadpole galaxies are indeed dynamically
370: young objects  related to early-stage  mergers, they may occur  in the
371: same   proportion   to   the    field   galaxy   population   at   all
372: redshifts.  Tadpole galaxies  may  therefore be  good  tracers of  the
373: galaxy assembly process.  The  ratio of the two redshift distributions
374: N(z) was calculated  as well, and the resulting  percentage of tadpole
375: galaxies  is  plotted  in  \fig{fig:per}  as a  function  of  redshift
376: together  with  the statistical  errors.  Overall,  the percentage  of
377: tadpole  galaxies is roughly  constant at  $\sim$6\% with  redshift to
378: within the  statistical errors  for the redshift  range probed  in our
379: study ($0.1\leq$z$\leq4.5$).
380: 
381: 
382: 
383: 
384: \begin{figure}
385: \epsscale{1.2}
386: \plotone{f3.ps}
387: \caption{Photometric   redshift  distribution   of  galaxies   in  the
388: HUDF. The solid black histogram shows the redshift distribution of all
389: HUDF  field galaxies  to  $i'_{AB}=28.0$, while  the dashed  histogram
390: shows the  redshift distribution of  the tadpole galaxies.  The latter
391: was multiplied by 16$\times$ for best comparison of its shape with the
392: redshift distribution of the field galaxies.}\label{fig:hist} 
393: \end{figure}
394: 
395: \begin{figure} 
396: \epsscale{1.2}
397: \plotone{f4.ps} 
398: \caption{Percentage of total galaxies  that are tadpoles is plotted as
399: a function of photometric  redshift. Within the statistical errors, on
400: average  about  6\%  of all  galaxies  are  seen  as tadpoles  at  all
401: redshifts.}\label{fig:per}
402: \end{figure}
403: 
404: 
405: \section{Tadpole Galaxies as Tracers of Galaxy Assembly} \label{tracers}
406: 
407: The fact that about 6\% of  all field galaxies are seen in the tadpole
408: stage is  a measurement  with potentially important  consequences.  In
409: light  of  simulations  by   Springel  \etal  (2005)  that  predict  a
410: tadpole-like  stage $\simeq$0.7 Gyr  after a  major merger  begins, we
411: suggest  that   this  particular  tadpole   morphology  represents  an
412: early-merger stage of two galaxies  with comparable mass.  If this 6\%
413: indicates  the fraction of  time that  an average  galaxy in  the HUDF
414: spends in  an early-merger stage during  its lifetime, and  if most of
415: these low-luminosity  objects started forming the bulk  of their stars
416: at  the end of  the reionization  epoch at  z$\simeq6-7$ (e.g.  Yan \&
417: Windhorst 2004a,  2004b), then  each galaxy would  spend about  6\% of
418: 12.9 Gyr (i.e. 0.8 Gyr)  since z$\simeq7$ in a distinctly recognizable
419: merger or tadpole stage. At  the median redshift at which the tadpoles
420: are seen  (z$_{med}$ $\simeq1.6$; see \fig{fig:hist}),  each object is
421: then seen at an age of about 4 Gyr if born at z$\simeq$7. Each tadpole
422: is $\simeq1$" (or $\simeq8$ kpc) across (\fig{fig:tp1}), and given the
423: fluxes  measured,  each clump  in  a  tadpole  has roughly  M$\,\simeq
424: 10^8$--$10^9\,\Mo$  in  stars  (see,  e.g.,  Papovich,  Dickinson,  \&
425: Ferguson 2001  who estimated stellar masses of  lyman break galaxies).
426: For these  rough estimates of their physical  parameters, the freefall
427: timescale   for  each  tadpole   is  roughly   $\tau\cle$  (few$\times
428: 10^7$)--$10^8$ years,  or $\simeq$6\% of  the galaxy lifetime  at that
429: redshift.  Hence, if  every  galaxy is  seen  in a  tadpole stage  for
430: $\simeq$0.8  Gyr   of  its  lifetime,  then  it   may  have  undergone
431: $\sim$10-30 mergers  during its lifetime.   During the early  stage of
432: each merger, it  would be temporarily seen as  a tadpole. More complex
433: mergers may  lead to  irregular/peculiar and train-wreck  type objects
434: and the  luminous diffuse objects  or clump clusters, which  are among
435: the  type  of  objects  that  dominate  the  galaxy  counts  at  faint
436: magnitudes (Driver  \etal 1998).  In  this paper, we limit  the sample
437: selection to two clumps passing  by each other, which we believe leads
438: to  the more  uniquely  classifiable tadpole  morphology.  Given  that
439: tadpoles only trace a certain  type and stage of merging galaxies, the
440: above  statistics  are likely  a  lower limit  on  the  number of  all
441: mergers.  In  summary, each  galaxy seen today  may have had  of order
442: one- to two- dozen mergers since  most of its Population II stars were
443: born  at z$\simeq$7,  and  given  the small  masses  and short  merger
444: timescales involved, would then be  seen as tadpole galaxies for about
445: 6\% of their life-time.
446: 
447: \fig{fig:per} suggests that tadpole galaxies --- if indeed dynamically
448: young objects --- appear to occur  in the same proportion to the field
449: galaxy  population at  all redshifts  probed in  this  study.  Tadpole
450: galaxies  may therefore  be  good  tracers of  the  process of  galaxy
451: assembly.  This  implies that  the process of  galaxy assembly  --- as
452: traced  by  tadpole  galaxies  ---  keeps up  with  the  reservoir  of
453: available    galaxies    as     a    function    of    redshift    for
454: $0.1\!\leq\!z\!\leq\!4.5$.   Our result  is  in excellent  agreement
455: with  the  predictions  of  Robertson  \etal (2005)  that  describe  a
456: merger-driven scenario  to build up  disk galaxies, and  is consistent
457: with Rhoads \etal (2005) who conclude that their z=5.4 galaxy (tadpole
458: \# 165 in Table 2) is strongly indicative of a galaxy in assembly.
459: 
460: 
461: \section{Relation between Tadpole Galaxies and Active Galactic Nuclei}
462: 
463: 
464: In a companion paper in this issue, Cohen \etal (2005) present a study
465: of  the  variable  objects in  the  HUDF  which  have a  point  source
466: component with  a measurably variable  flux on timescales  of 0.4--3.5
467: months, which roughly corresponds  to 0.5--5.5 weeks in the rest-frame
468: at  the median redshift  of the  sample (z$_{med}$  $\simeq1.5-2$). In
469: particular,  they  found  45  plausibly variable  objects  among  4644
470: galaxies  to $i'_{AB}=28.0$ mag.   They argue  that these  objects are
471: most likely variable  because they host weak AGN.  Sometimes these AGN
472: are  in  the galaxy  center,  but often  they  occur  off-center in  a
473: dynamically unrelaxed system. This prompts the question: What fraction
474: of tadpole galaxies contains a variable weak AGN in its knots? This is
475: a critical  issue, because it is  widely believed that  the process of
476: merging in galaxies  can also disturb the inner  accretion disk around
477: the supermassive black  hole (SMBH) and switch on  the AGN.  Among our
478: 165 tadpole  galaxies, none  coincide with the  sample of  45 variable
479: objects  or with the  x-ray sources  in the  Chandra Deep  Field South
480: (Alexander, D.M. \etal  2005).  Recent state-of-the-art hydrodynamical
481: models (di  Matteo, Springel, \& Hernquist 2005;  Springel, di Matteo,
482: \&  Hernquist 2005; Hopkins  \etal 2005)  suggest that  during (major)
483: mergers, the  black hole accretion  rate peaks considerably  after the
484: merger  started,   and  after  the  star  formation   rate  (SFR)  has
485: peaked. Specifically, their models suggest that, for massive galaxies,
486: a  tadpole stage  is seen  typically about  0.7 Gyr  after  the merger
487: started, but $\sim$0.9 Gyr before  the SMBH accretes most of its mass,
488: which  is when  the galaxy  displays  strong AGN  activity. Since  the
489: lifetimes   of   QSO's   and    radio-galaxies   are   known   to   be
490: $\cle\,$(few$\times  10^7$)--$10^8$ years  (Martini, P.  2004, Grazian
491: \etal 2004), these hydrodynamical models thus imply that the AGN stage
492: is expected to occur  considerably \emph{after} (i.e., $\ge$1 Gyr) the
493: early-merger  event during  which the  galaxy is  seen in  the tadpole
494: stage.  The observed lack of  overlap between the tadpole galaxies and
495: the AGN  sample in the  HUDF provide direct observational  support for
496: this prediction.  Recent studies by Grogin \etal (2005) find asymmetry
497: (A)  values that  are similar  between AGN  and non-AGN  samples; this
498: result is consistent with our study in light of the theoretical models
499: mentioned above,  which indicate that  AGN activity is seen  only well
500: \emph{after} the  merger has  taken place and  the galaxy  has settled
501: into  a more dynamically  relaxed state.   In addition,  Hopkins \etal
502: (2005) have  quantified the  timescales that  quasars will  be visible
503: during  merging  events, noting  that  for  a  large fraction  of  the
504: accretion time, the quasar  is heavily obscured.  In particular, their
505: simulations show that  during an early merging phase  --- our observed
506: tadpole  phase  --- the  intrinsic  quasar  luminosity  peaks, but  is
507: completely  obscured.  Only  after  feedback from  the central  quasar
508: clears out  the gas will  the object become  visible as an  AGN.  This
509: should  be   observable  by  the   Spitzer  Space  Telescope   in  the
510: mid-infrared (IR) as a  correspondingly larger fraction of IR-selected
511: obscured faint QSO's.
512: 
513: In conclusion,  tadpole galaxies  are a class  of easily-identifiable,
514: dynamically  young objects that  exist throughout  the history  of the
515: Universe and are good tracers of galaxy assembly.  They provide strong
516: observational   support   for  the   validity   of  recent   numerical
517: simulations, and highlight the importance of mergers to the process of
518: galaxy assembly and AGN growth.
519: 
520: 
521: 
522: \section{Acknowledgements}\label{Acknowledgements}
523: 
524: We thank the STScI staff, and in particular Steve Beckwith, Ray Lucas,
525: and Massimo  Stiavelli, for their persistent efforts  to implement the
526: Hubble Ultra  Deep Field in the  best possible way.  This research was
527: partially  funded  by  NASA  grants  GO-9793.08-A  and  AR-10298.01-A,
528: awarded by  STScI, which is operated  by AURA for  NASA under contract
529: NAS 5-26555;  as well as by the  NASA Space Grant program  at ASU, and
530: the Harriet G. Jenkins Predoctoral Fellowship Program.
531: 
532: 
533: \begin{thebibliography}{}
534: \bibitem[Alexander et al.(2003)]{ax03} Alexander, D.M. et al. 2003, AJ 126, 539
535: \bibitem[Bolzonella, M., Miralles, J., Pello, R. (2000)]{hyper} Bolzonella, M., Miralles, J., Pello, R. 2000, A\&A, 363, 476
536: \bibitem[Barger et al. (2003)]{bar03} Barger, A.J. et al. 2003 ApJL, 584, L61
537: \bibitem[Beckwith et al. (2005)]{beck05} Beckwith, S., et al. 2005, AJ, in preparation
538: \bibitem[Bertin \& Arnouts(1996)]{sex} Bertin, E. \& Arnouts, S. 1996, A\&AS, 117, 363
539: \bibitem[]{414} Blumenthal, G., Faber, S., Flores, R., Primack, J. 1986, ApJ 301, 27
540: \bibitem[]{415} Cohen, S. \etal 2005, ApJ (this volume)
541: \bibitem[Conselice, Bershady \& Jangren 2000]{con00} Conselice, Bershady \& Jangren 2000, ApJ 529, 886
542: \bibitem[Conselice et al. 2003]{con03} Conselice, C. et al. 2003, AJ 126, 1183
543: \bibitem[Conselice et al. 2004]{con04} Conselice, C. et al 2004, ApJL 600, L139
544: \bibitem[Cowie et al. 1996]{cow96} Cowie, L., Songaila, A., Hu, E., \& Cohen, J.G. 1996, AJ 112, 839
545: \bibitem[Cowie, Hu, \& Songaila 1995]{cow95} Cowie, L., Hu, E., \& Songaila, A. 1995, AJ 110, 1576
546: \bibitem[di Matteo, Springel, \& Hernquist 2005]{dimatteo05} Di Matteo, T., Springel, V., \& Hernquist, L. 2005, Nature, 433, 604
547: \bibitem[Driver et al. (1998)]{dri98} Driver, S., Windhorst, R., Griffiths, R. 1998 ApJ 453, 48
548: \bibitem[Elmegreen, Elmegreen, \& Hirst 2004]{elm04a} Elmegreen, D., Elmegreen, B., \& Hirst, A. 2004, ApJL 604, L21
549: \bibitem[Elmegreen, Elmegreen, \& Sheets 2004]{elm04b} Elmegreen, D., Elmegreen, B., \& Sheets, C. 2004, ApJ 603, 74
550: \bibitem[Elmegreen et al. 2005]{elm05} Elmegreen, D., Elmegreen, B., Rubin, D., Schaffer, M. 2005, ApJ 631, 85
551: \bibitem[]{425} Grazian, A. \etal 2004, AJ 127, 592
552: \bibitem[]{426} Grogin, N. \etal 2005, ApJL 627, 97
553: \bibitem[]{427} Hopkins, P., Hernquist, L., Martini, P., Cox, T., Robertson, B., Di Matteo, T., Springel, V. 2005, ApJL, 625, L71
554: \bibitem[Hopkins, P., Hernquist, L., Cox, T., di Matteo, T., Robertson, B., Springel, V. 2005] {}Hopkins, P., Hernquist, L., Cox, T., di Matteo, T., Robertson, B., Springel, V. 2005, ApJ, submitted (astro-ph/0506398)
555: \bibitem[Koekemoer (2004)]{koe04} Koekemoer, A. 2004, ``The Hubble Ultra Deep Field ACS/WFC Combined Images--V1.0'' http://cadcwww.dao.nrc.ca/udf/acs-wfc/h\_udf\_wfc\_Readme.txt 
556: \bibitem[]{lan} Lanzetta, K.M. \etal 1997, Proceedings of the STScI 1997 May Symposium (astro-ph/9709166)
557: \bibitem[]{430} Martini, P. 2004, Carnegie Obs. Astrophys. Ser. 1, Coevolution of Black Holes and Galaxies (Cambridge: Cambridge University Press), 170
558: \bibitem[]{mobasher} Mobasher, B. \etal 2004, ApJL 600, 167
559: \bibitem[]{431} Navarro, J., Frenk, C., White, S. 1997, ApJ 490, 493
560: \bibitem[Odewahn et al. 1996]{odewahn96} Odewahn, S. et al. 1996, ApJL 472, L13
561: \bibitem[Odewahn et al. 2002]{odewahn02} Odewahn, S. et al. 2002, ApJ 568, 539
562: \bibitem[]{papovich} Papovich, C., Dickinson, M, \& Ferguson, H. 2001, ApJ 559, 620
563: \bibitem[Rhoads et al. 2005]{rhoads05} Rhoads et al. 2005, ApJ 621, 582
564: \bibitem[]{435} Robertson, B. \etal 2005, ApJL, submitted (astro-ph/0503369)
565: \bibitem[Springel, Di Matteo \& Hernquist 2005a]{springel05} Springel, V., di Matteo, T., \& Hernquist, L. 2005a, ApJ, 620, 79
566: \bibitem[Springel et al. 2005]{sp2} Springel, V., di Matteo, T., Hernquist, L. 2005b, MNRAS 361, 776
567: \bibitem[van den Bergh 2002]{vdBergh02} van den Bergh, S., 2002, PASP 114, 797
568: \bibitem[]{438} White, S.D.M. \& Rees, M.J. 1978, MNRAS 183, 341
569: \bibitem[Williams, R. et al 1996]{williams96} Williams, R. et al. 1996, AJ 112, 1335 
570: \bibitem[Windhorst, R.A. et al. 2002]{windhorst02} Windhorst, R. et al. 2002, ApJS 143, 113
571: \bibitem[Yan, H., \& Windhorst, R.A. 2004a]{441}  Yan, H., \& Windhorst, R. 2004, ApJL 600, L01
572: \bibitem[Yan, H., \& Windhorst, R.A. 2004b]{442} Yan, H., \& Windhorst, R. 2004, ApJL 612, L93
573: \end{thebibliography}
574: 
575: \input{tab2.tex}
576: 
577: 
578: \end{document}
579: