1:
2: \documentclass[usenatbib]{mn2e}
3: \usepackage{graphics,amsmath}
4: \usepackage{rotating}
5: \bibliographystyle{mn2e}
6:
7:
8: \title[The TF relation at $z\sim0.85$]{The Tully-Fisher relation of galaxies at $z\sim 0.85$ in the DEEP2 survey}
9: \author[K.~Chiu, S.P.~Bamford, A.~Bunker]{Kuenley Chiu$^{1,2}$, Steven P. Bamford$^3$, Andrew Bunker$^{1}$ \\
10: $^{1}$\,School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL\\
11: $^{2}$\,chiu@astro.ex.ac.uk\\
12: $^{3}$\,Institute of Cosmology, University of Portsmouth, Portsmouth, PO1 2EG
13: }
14: \begin{document}
15:
16: \date{Accepted 20 Feb 2007 for publication in MNRAS}
17:
18: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2006}
19:
20: \maketitle
21:
22: \label{firstpage}
23:
24: \begin{abstract}
25:
26: Local and
27: intermediate redshift ($z\sim0.5$) galaxy samples obey well correlated
28: relations between the stellar population luminosity and maximal
29: galaxy rotation that define the Tully-Fisher (TF) relation.
30: Consensus is starting to be reached on the TF
31: relation at $z\sim0.5$, but work
32: at significantly higher
33: redshifts is even more challenging, and has been limited by small
34: galaxy sample sizes, the intrinsic scatter of galaxy properties, and
35: increasing observational uncertainties.
36: We present here the TF measurements of 41 galaxies at relatively high redshift,
37: spectroscopically observed with the Keck/DEIMOS
38: instrument by the DEEP2
39: project, a survey which will eventually offer a large galaxy sample
40: of the greatest depth and number yet achieved towards this purpose.
41: The 'first-look' sample analyzed here has a redshift range of
42: $0.75<z<1.3$ with $\langle z \rangle = 0.85$ and an intrinsic
43: magnitude range from $M_B$ of $-22.66$ to $-20.57$ (Vega). We find that
44: compared to local fiducial samples, a brightening of $1.5$
45: magnitudes is observed, and consistent with passive evolutionary models.
46:
47:
48:
49: \end{abstract}
50:
51: \begin{keywords}
52: galaxies: evolution --
53: galaxies: formation --
54: galaxies: kinematics and dynamics --
55: galaxies: spiral
56: \end{keywords}
57:
58: \section{Introduction}
59: \label{sec:intro}
60:
61: The observed evolution of the Tully-Fisher relation \citep[TF;][]{tully77}
62: has the potential to provide considerable insight into
63: how the properties of today's galaxy population were established. In
64: particular, this relation between the luminosity and rotation velocity
65: of disk galaxies -- probed through the total mass-to-light ratio of
66: galaxies -- reveals links
67: between dark matter haloes and the stellar
68: populations which form within them. This should ideally allow us to
69: distinguish between the various proposed theories of galaxy formation
70: and evolution.
71:
72: In the hierarchical structure formation paradigm of cold dark matter
73: (CDM), small galactic haloes ($10^{9-10}\,M_{\odot}$) first condensed
74: out at peaks in the near-uniform dark matter distribution.
75: Eventually, through sufficient numbers of mergers, these sub-units
76: accumulated to produce the massive haloes inhabited by the $\sim L*$
77: galaxies that dominate the stellar mass density from $z\sim1$ to the
78: present. One possibility is that these first, small haloes were the sites of
79: significant early star formation \citep{mo98,sommer03},
80: and the resulting stellar mass subsequently accumulated through mergers
81: (along with their dark matter haloes) into more
82: massive galaxies without the need for substantial later
83: star-formation. In this case, the growth of stellar mass in
84: individual galaxies was driven primarily by mergers of already
85: luminous sub-units, and hence the observed mass-to-light ratio should
86: have been relatively steady over time, evolving only slowly due to the
87: opposing actions of passive aging of the stellar populations and
88: residual star-formation.
89:
90: In contrast, there is some growing evidence that star formation was
91: suppressed in smaller haloes at early times \citep{nelan05}.
92: The suggested causes of this range from the increasing
93: ultraviolet background \citep{wyithe06}, the
94: effects of star formation feedback \citep{springel05} (particularly from
95: the theoretically expected initial population of very-massive,
96: zero-metallicity stars), or even AGN preferentially
97: enhancing star formation in more massive haloes \citep{silk05}.
98: Whichever mechanism is responsible, this implies that galaxies must
99: have formed a substantial amount of their stellar mass afterwards, at later times,
100: following the many mergers that assembled most of their dark matter haloes.
101: This is somewhat reminiscent of the old
102: monolithic collapse model for massive galaxy formation
103: \citep{eggen62}, and implies substantial evolution in the
104: mass-to-light ratio of galaxies -- as stars form in the already
105: assembled massive halo. This would also require more star formation per unit mass in smaller
106: haloes than in more massive ones at later times, the so called
107: `downsizing' effect.
108:
109: The relative order of mass assembly and star formation is cleary of
110: significant interest. Therefore, examining evolution in the
111: mass-to-light ratio of galaxies has been pursued as an important
112: indicator of the galaxy formation mechanism. Such evolution is often
113: probed through observed changes in fundamental scaling laws versus
114: redshift, such as the Faber-Jackson and Kormendy relations for
115: ellipticals \citep[believed to be projections of a fundamental
116: plane,][]{djorgovski87}. For spiral galaxies, the Tully-Fisher relation (TF)
117: \citep{tully77} is the primary scaling law.
118: Statistically well-populated TF diagrams over a range of redshifts
119: should be able to test the various galaxy formation options through
120: changes in the slope, intercept, and scatter of the fitted TF
121: relation.
122:
123: Tully-Fisher relations have been relatively well characterized for
124: galaxies nearby, originally for distance indicator purposes
125: \citep{pierce92,giovanelli97,dale01}, but recently in a more
126: inclusive manner for assessing galaxy evolution, and to compare with
127: higher redshift work \citep[e.g.,][]{kannappan02}. A number of studies
128: have measured the TF relation at intermediate redshifts ($z\sim
129: 0.4$--$1$) and attempted to infer its evolution with cosmic time.
130: However, a consensus on the TF relation evolution over the past Hubble
131: time (out to $z\sim 1$) has not yet been achieved, although very recent work
132: such as \citet{weiner06} have greatly improved upon the few existing high redshift studies.
133: Difficulty arises
134: due to the relatively low numbers of galaxies typically used;
135: the intrinsic scatter in the TF relation means that many tens of
136: galaxies are needed per redshift bin for a reliable measurement,
137: particularly to constrain the slope of the relation. The primary
138: cause of discrepancies between studies appears to be differing
139: selection criteria, and the manner in which these have been corrected
140: for, if at all. The applied corrections can vary widely
141: depending on the percieved aim of each particular study. In addition,
142: the use of different local relations, TF fitting methods, and internal
143: extinction corrections has complicated comparisons between studies.
144:
145: As a result, various groups have identified significant evolution in
146: the $B$-band TF relation out to $z\sim1$ \citep[e.g.,][]{vogt96,vogt97,rix97,
147: simard98,barden03,bohm04,bohm06,nakamura06},
148: or state an upper limit \citep{bamford06} on the evolution, while
149: others find no evolution once selection effects have been accounted for
150: \citep{simard99, vogt01} or kinematically disturbed galaxies discarded
151: \citep{flores06}. However, a consistent picture is begining to
152: emerge, as these studies become more mature. One fairly certain
153: aspect is that there has been rather little evolution in the intercept
154: of the TF, ranging from no evolution to a modest brightening of
155: $\sim 1$ mag at fixed rotation velocity by $z \sim 1$. This is much
156: less than would be expected if the star formation rates of massive
157: disk galaxies have evolved as strongly as is measured for the global
158: star formation rate density of the universe \citep{bamford06}.
159:
160: Constraining evolution in the TF slope is more difficult, though again
161: any evolution is fairly modest. Any overall brightening of the TF
162: that is observed appears to be mainly driven by low mass galaxies, as
163: noticed by \citet{bamford06}, identified as an additional population
164: by \citet{vogt01}, and reflected in the TF slope change measured by
165: \citet{bohm04,bohm06}. This is
166: particularly the case when considered along with the observed lack of
167: evolution in the $K$-band and stellar mass TF relations found by
168: \citet{conselice05}, implying that while the stellar to total mass of
169: galaxies stays roughly constant, the colour, and hence
170: luminosity-weighted age or fraction of stars that have recently
171: formed, has evolved, possibly with a dependence on galaxy mass.
172: However, it is still unclear how much of this effect is actually due
173: to differential luminosity evolution at a given rotation velocity, or
174: a symptom of underestimating the rotation velocities for an increasing
175: population of kinematically disturbed galaxies at higher redshifts.
176:
177:
178: \begin{figure}
179: \includegraphics[scale=0.35,angle=0]{zdist.ps}
180: \caption{Redshift distribution of galaxies measured in this work (solid line). Dotted line indicates distribution of full DEEP2 spectroscopic sample, prior to TF analysis selection, discussed in text. The full sample distribution has been scaled by 0.05 for comparison purposes. \label{zdistfig}}
181: \end{figure}
182:
183:
184:
185:
186:
187: \begin{figure*}
188: \resizebox{0.49\textwidth}{!}{\includegraphics*[125,165][489,739]{makematrix1.ps}}
189: \resizebox{0.49\textwidth}{!}{\includegraphics*[125,165][489,739]{makematrix2.ps}}
190: \caption{ Image and spectral model fitting results for
191: the galaxies in the present TF sample are shown, with
192: DEEP2 galaxy identification names. The first 3 columns illustrate a)
193: the original galaxy postage stamp image, b) the fitted model using
194: an exponential disk profile in GALFIT, and c) the model-subtracted
195: image. The last two columns show d) the original spectral flux at
196: the 3727\AA~ [OII] line, followed by e) the model generated by the
197: ELFIT2PY procedure. Galaxy images are $60\times60$ pixels
198: ($12''\times12''$), while the 2d spectra are 80 pixels wide
199: ($9''$, or 25\AA). \label{galfig} }
200: \end{figure*}
201:
202:
203: \begin{figure*}
204: \resizebox{0.49\textwidth}{!}{\includegraphics*[125,135][489,763]{makematrix3.ps}}
205: \resizebox{0.49\textwidth}{!}{\includegraphics*[125,135][489,763]{makematrix4.ps}}
206: \caption{ -- continued.}
207: \end{figure*}
208:
209:
210:
211:
212: \citet{flores06} use integral field spectroscopy to investigate this
213: question, and also find that the $K$-band, or stellar mass, TF
214: relation has not evolved for a small sample of undisturbed disks out
215: to $z \sim 0.7$, while galaxies with signs of kinematical disturbances
216: are generally measured at low rotation velocities for their
217: $K$-band luminosity. In the $B$-band, however, they show some
218: evidence for a brightening of $0.5$ mag, and a tentative
219: change of slope consistent with the studies of \citet{bohm04}, even for
220: undisturbed disks. \citet{smith04} find a consistent result from near-infrared
221: integral field spectroscopy for a galaxy at $z=0.82$.
222:
223: A novel method in TF studies is the use of strong lensing by galaxy
224: clusters to magnify distant galaxies \citep{swinbank03}. This enables
225: the examination of $z \sim 1$ galaxy rotation curves with resolution
226: comparable to $z \sim 0.1$ studies. \citet{swinbank06} use this
227: technique to study six $z \sim 1$ galaxies, four of which are found to
228: have regular rotation curves. These are found to define an $I$-band
229: TF relation very similar to that found locally, and a $B$-band TF with
230: similar slope to the local relation, but offset to brighter magnitudes
231: by $\sim 0.4$--$0.5$ mag.
232:
233: While such integral field work is important in providing a complete
234: map of the galaxy velocity field (an improvement on the spatial
235: losses accompanying the usual technique of long-slit spectroscopy) the
236: number of galaxies observable with this method is rather limited by
237: present instrumental capabilities. To measure rotation curves for a large sample
238: of galaxies requires spatially-resolved multi-object spectroscopy,
239: and currently the most efficient method is the use
240: of slit-masks with individual slits tilted to align with the
241: galaxy major axes. The DEIMOS optical spectrograph on Keck II
242: \citep{faber03} provides this capability, and a major project
243: exploiting this instrument has been the DEEP2 project \citep{davis04}. This
244: survey has undertaken spatially-resolved multi-slit spectroscopy of a large
245: sample of galaxies colour selected to lie at $0.7<z<1.4$. The red
246: sensitivity of the instrument, coupled with its stability and lack of
247: fringing, means that [OII]\,3727\,\AA\ rotation curves can be traced
248: cleanly throughout this redshift range.
249:
250: As well as spectroscopy, in order to construct a TF relation at
251: intermediate redshift one requires high quality imaging, to provide
252: magnitudes and importantly for measuring the inclination angles of the
253: targetted galaxies, needed to correct the rotation velocities for
254: projection effects.
255:
256: In this paper, we measure the rotation curves of $41$ galaxies at
257: $\langle z \rangle \sim 0.85$ from the DR1 release of the DEEP2 spectroscopy, and use
258: them to determine the evolution of the $B$-band TF relation. In Section~2 we
259: discuss the DEEP2 spectroscopy from Keck/DEIMOS, and the imaging from
260: CFHT. In Section~3 we detail the fitting of galaxies in
261: the imaging in order to derive inclinations, and the emission-line modelling to
262: measure the rotation velocities. Section~4 presents our resulting
263: Tully-Fisher relation and a comparison with previous work at low and
264: high redshift. Our conclusions are in Section~5. In this paper all magnitudes
265: are expressed based on Vega normalization, and throughout we adopt
266: the standard `concordance' cosmology of $\Omega_M=0.3$,
267: $\Omega_\Lambda=0.7$, $H_0=70$ km s$^{-1}$ Mpc$^{-1}$.
268:
269: \section{Data sources and processing}
270: \label{sec:data}
271:
272:
273:
274:
275: In this work, we aim to compare the fitted TF parameters for our
276: present sample with those for similar samples at intermediate and
277: lower redshifts. The main uncertainties in constructing a TF relation
278: usually arise from the galaxy rotation velocities. These are
279: determined from the rotation curves, typically using the shapes of
280: emission lines measured in the spatially-resolved two-dimensional
281: spectroscopy. These velocities are then corrected for projection
282: effects due to the inclination of the galactic disk, and misalignment
283: of the slit from the major axis of the galaxy, both determined from
284: morphological fits to the imaging data. Therefore, while the
285: intrinsic scatter of the galaxies themselves place limits on the
286: uncertainties of the fitted TF relation, galaxy imaging and
287: spectroscopic data of high signal-to-noise and resolution should
288: minimize the remaining fundamental sources of uncertainty
289: \citep{kannappan02}. In particular, in this work we take advantage of
290: the high spatial ($0\farcs12$/pixel) and wavelength resolution
291: ($68\,{\rm km\,s}^{-1}$) of the DEEP2 survey, combined with the wide
292: area imaging archive over the DEEP2 survey region from the {\it CFHT} CFH12K
293: mosaic imager in the $I$-band (corresponding to the rest-frame
294: $B$-band at $z\sim 0.9$).
295:
296: \subsection{DEEP2 spectroscopy}
297:
298: We began our sample selection by examining the DEEP2 dataset for
299: suitable galaxies for TF measurement. The DEEP2 spectroscopic
300: observations were based on an optical imaging survey taken from
301: 1999--2000 using the CFH12K mosaic camera \citep{cuillandre01} in the
302: $BRI$ bands to a depth of $R_{AB}<24.1$. Galaxy target selection and
303: slitmask design were conducted using this imaging (which covered four
304: fields composing a total of 3 deg$^2$),
305: with color cuts in $(B-R)$ \& $(R-I)$ to optimize the selection of
306: galxies in the redshift range $z=0.75$--$1.4$ \citep{coil04}.
307:
308: Although the DEEP2 project will eventually obtain DEIMOS spectroscopy
309: for some 40,000 galaxies over the total area described above, the
310: current data release contains $\sim 7500$ galaxies over $\sim 1.2$
311: deg$^2$. Of these, approximately 4300 have confident redshifts as
312: measured by the DEEP2 pipeline software ($Q=4$), and we inspected
313: these galaxies for strong emission lines at the expected position of
314: {[OII]3727\AA} in each spectrum. In this first (``1HS'') phase of the
315: DEEP2 project, one hour spectroscopic integrations were obtained on
316: each $16'\times4'$ slitmask, containing about 85 objects per mask.
317: Later phases of DEEP2 (``3HS'') will integrate on a smaller galaxy
318: sample for $>3$ hours. After being dispersed by the 1200/mm grating,
319: the wavelength coverage is $6500-9100$\,\AA , so {[OII]3727\AA} is
320: detectable over $0.77<z<1.42$, nearly the entire redshift range
321: targeted by the $BRI$ colour-cuts. The spectra have a typical
322: spectral resolution of $R=\lambda/\Delta\lambda_{\rm FWHM}\approx 4000
323: \approx 68\,{\rm km\,s}^{-1}$ determined by the $1''$ slit size,
324: sampled by {$\sim0.3$\AA} pixels. The spatial pixel size is
325: $0\farcs119$. Of the 4300 spectra, 388 were found to have strong
326: [OII] emission in the aforementioned wavelength range, and
327: suitable for measuring rotation velocities using the
328: line fitting task with integrated single-line signal of $S/N>20$.
329:
330: The spectra were provided by the DEEP2 team in the DR1 public
331: release\footnote{http://deep.berkeley.edu/DR1} as a fully reduced dataset,
332: with catalogs and wavelength-calibrated multi-extension FITS
333: 2d-spectra and 1d spectral extractions. We did little additional
334: processing except that prior to the model fitting procedures described
335: below, we took the additional step of rectifying the 2d spectra to
336: have orthogonally aligned spatial and spectral axes, with constant
337: pixel scale. This was carried out to correct spatial distortions
338: and because the slits of the DEIMOS
339: instrument can be tilted to accomodate the position angles of the
340: selected galaxies (within $\sim30 ^\circ$, constrained by the CCD
341: columns and desired spatial resolution). This results in a generally
342: diagonal (or even curved) wavelength solution across the spatial rows.
343: Using the DEEP2 wavelength solution at the given spatial center of
344: each galaxy in the slit mask, the solution was reproduced in all rows,
345: and the flux interpolated and rebinned to a constant dispersion of
346: $0.32$ \AA/pixel.
347:
348:
349: \subsection{CFH12K imaging}
350:
351: With the sample of suitable spectra identified, we extracted images of
352: the galaxies for measurement of the necessary shape parameters.
353: Searching the CFHT archive for the three currently released DEEP2
354: spectroscopic fields (out of four total to be made public), we
355: identified useful imaging exposures taken by the CFH12K instrument.
356: The CFH12K camera has a platescale of $0\farcs206$/pixel, and covers a
357: $42'\times28'$ rectangular field of view using a $2\times6$ array of
358: 4K$ \times $2K CCDs, for a total of $0.326$ square degrees per pointing.
359: Two fields, ``0230+0000'' and ``2330+0000'', had $I$-band imaging
360: which overlapped with most of the DEEP2 spectra where we identified
361: strong [OII] emission. With a median redshift of $\langle z \rangle =0.85$ for the
362: galaxy sample used here (Figure \ref{zdistfig}), the choice of the
363: $I$-band allows the photometry to be easily converted to the $B$-band
364: rest-frame absolute magnitude commonly used for TF studies, with
365: minimal $K$-correction uncertainties.
366:
367:
368: The individual $6\times10$ minute dithered imaging exposures in each field were
369: combined using standard image reduction techniques, including the use
370: of bad pixel masks and calibration frames (``master detrend''
371: bias/dark/flat-field frames) identified for each camera run in the
372: CFH12K archive. Cosmic ray rejection was implemented in addition to
373: bad pixel masking and local sky subtraction during the median
374: combining of the separate exposures. Finally, each reduced image was
375: astrometrically corrected to the USNO-B1.0 catalog \citep{monet03},
376: using the WCSTools routines of D.~Mink, which automatically update
377: the appropriate header WCS information given a standard star catalog
378: and chosen reference stars in the image of interest. This allowed
379: automated extraction of the image of each galaxy selected this work,
380: based on the J2000 RA/Dec positions given in the DEEP2 spectroscopic
381: catalogs. At this point, 290 (out of 388) galaxies with strong [OII] flux
382: were also covered by these two CFH12K imaging mosaics, and
383: remained in the sample for further analysis.
384:
385:
386:
387:
388: \section{Data Analysis}
389:
390: For the $290$ objects selected to have strong [OII] line emission
391: in the DEIMOS spectroscopy, and with CFH12K $I$-band images, we used
392: the imaging to model the morphological parameters (giving the disk
393: inclincation angle, $i$, and position angle), and the spectroscopy to
394: model the emission-line rotation curve (yielding $V_{\rm rot}\sin
395: i$). We describe both of these below.
396:
397: The TF relation also requires absolute magnitudes ($M_B$ in this
398: instance), and we determined these from the $I$-band imaging,
399: introducing a redshift-dependent $K$-correction to match the
400: restframe-$B$ with the observed-$I$ filter. Finally we corrected the
401: absolute magnitude for intrinsic extinction in the disk galaxy,
402: depending on inclination and luminosity.
403:
404: \subsection{Galaxy image and spectral profile fitting procedures}
405: \label{sec:fitting}
406:
407: First, image shape modeling was carried out on each galaxy using the
408: GALFIT program of \citet{peng02}, in order to calculate the position
409: angle (PA) and axial ratio values required in subsequent steps. For
410: each galaxy, a $100\times100$ pixel cutout ($12''\times 12''$) was
411: made from the CFH12K image, preserving astrometric and flux
412: information. A point spread function (PSF) was produced by stacking
413: $\sim 10$ good point sources within each CFH12K array chip, resulting in a
414: profile with a FWHM of 2.5 pixels ($0.52''$).
415:
416: GALFIT convolves this PSF with a model galaxy image based on input
417: initial parameter estimates ($x$, $y$ position, effective radius,
418: magnitude, axial ratio, position angle) and perturbs these parameters
419: until the residuals of the model-subtracted image are minimized. The
420: initial input parameters were determined by the SExtractor program
421: \citep{bertin96} on the galaxy images. An exponential disk profile was chosen as the
422: desired galaxy model, and the program iterated to completion,
423: returning each of the minimized input parameters with errors. Of the
424: $290$ galaxies input to GALFIT, half returned acceptable
425: fits, while the remaining galaxies generally failed to converge due to
426: unsalvagable flux profiles (i.e., galaxies not well described by
427: exponential disks, or unresolved in the image). Examples of galaxy
428: profile fits, and the raw/subtracted CFH12K images are shown in Figure
429: \ref{galfig}.
430:
431:
432:
433:
434:
435: \input{galtable4}
436:
437:
438:
439: The fitting of faint emission lines in order to extract reliable
440: rotation velocities is a challenging task, and highlights the
441: desirability of high resolution and high signal-to-noise 2-dimensional
442: spectra in order to determine confident galaxy rotation curves. In
443: order to extract as much information as possible from the emission
444: lines, and to consistently model the effects of inclination, seeing,
445: finite slit width and instrumental resolution, we employed an enhanced
446: version of the ELFIT2PY routine developed by SPB \citep{bamford05},
447: which is based partly on the synthetic emission line fitting software
448: of \citet{simard98}.
449:
450: The software is given the galaxy inclination, seeing, instrumental
451: spectral resolution, slit position angle misalignment with respect to
452: the galaxy major axis, slit tilt in the mask, and produces synthetic
453: emission lines by constructing model 2D galaxy flux and velocity
454: profiles, assuming a set of parameters, and then mimicking the
455: observations. ELFIT2PY uses the Metropolis searching algorithm
456: to find the set of parameters which minimize the
457: residuals between the synthetic and observed emission lines. The
458: searched parameters are the rotation velocity, the emission flux
459: scalelength (assuming an exponential disk), the turnover radius of the
460: rotation curve, and the spatial and wavelength centres of the emission
461: line. The use of more complicated flux profiles was attempted, but it
462: was found that these generally could not be constrained. An intrinsic rotation
463: curve of the form $V_{\rm rot}/(r^a + r_t^a)^{1/a}$ is assumed
464: \citep{courteau97}, with $a = 5$ (found to best reproduce the observed
465: rotation curve shapes, though ill-constrained, and in agreement with
466: \citealt{bohm04}), and where $r$ is the radius and $r_t$ is the turnover
467: radius. Such an intrinsic rotation curve is necessary in this case,
468: rather than the discontinuous form used previously with ELFIT2PY, in
469: order to adquately fit the high resolution of the
470: DEEP2 spectra. Note that the emission flux scalelength and the
471: turnover radius of the intrinsic rotation curve were allowed to vary
472: independently, which was frequently found to be necessary. The total
473: flux in the line was measured by integrating the background- and
474: continuum-subtracted image of the line, and held fixed for the fit,
475: with a constant [OII] doublet ratio of $f([{\rm OII} ] ~3727.092{\rm~ \AA})/f([{\rm OII}] ~3729.875~{\rm \AA}) =0.8$. The background level
476: was also fixed at zero.
477:
478:
479:
480: In previous versions of ELFIT2PY, the chi-squared comparison between
481: the synthetic and observed lines was done on a pixel by pixel basis in
482: the spectra. However, some other groups \citep[e.g.,][]{bohm04} first
483: trace the synthetic and observed emission lines, by fitting the
484: position of the line in each spatial row independently, and then
485: compare these traces to produce a chi-squared value used to judge the
486: goodness of fit. This has the advantage of removing one level of
487: dependence on the assumed galaxy flux profile, though it still of
488: course plays a role in the construction of the synthetic line. It also
489: weights the fit to preferentially reproduce the observed rotation
490: curve rather than the flux profile, which may be affected by discrete
491: star-formation regions and asymmetries. However, this is at the cost
492: of leaving the emission scalelength unconstrained, which must
493: therefore by fixed relative to the rotation curve turnover radius.
494: Therefore a hybrid approch was adopted to take advantage of the
495: high resolution of the DEEP2 spectra (which
496: allow a high quality trace of the emission line to be measured), while
497: somewhat reducing the influence of asymmetries in the observed flux
498: profile which are apparently more prevalent at the high redshifts of
499: the current sample.
500:
501: This hybrid method combined the chi-squared values from pixel-to-pixel
502: comparisons of the synthetic and observed spectra, and those from
503: comparisons of both the trace centres and FWHM of the synthetic and
504: observed lines. The variation of the observed FWHM in each spatial
505: row helps constrain the rotation curve as it is a measure of the
506: gradient of the rotation curve at each spatial position. If the
507: rotation velocity is increasing quickly with radius, the observed FWHM
508: in that spatial row is higher. This combined chi-squared value was
509: the quantity used by the Metropolis algorithm to determine the
510: best-fitting parameters and their confidence intervals.
511:
512: The traces were performed by fitting a double Gaussian function to
513: each spatial row independently. The Gaussians had a fixed separation
514: appropriate for the [OII] doublet at the redshift of the galaxy, and
515: the same FWHM. Only the centre of the doublet, single line FWHM and
516: amplitudes of the two components were allowed to vary. Fits were
517: judged to be reliable if the Levenberg-Marquardt least-squares
518: minimisation converged from the appropriately chosen initial parameters,
519: if the errors on the best-fitting centre and amplitudes were sensible,
520: and if the best-fitting FWHM was larger than the spectral resolution.
521:
522: Note that the velocity convention in this work is to use the $V_{\rm
523: rot}$ ``half-velocity'' term, which specifies the maximum absolute
524: rotation velocity, of the asymptotic assumed intrinsic rotation curve,
525: with respect to the center of the galaxy, rather than the
526: velocity-width, $2V_{\rm rot}$, traditionally used to express the
527: total velocity gradient measured across the entire galaxy. Examples
528: of synthetic emission-line images, created using the best-fit
529: parameters, appear in Figure \ref{galfig}.
530:
531:
532: The emission lines of the 290 galaxies selected thus far were
533: inspected visually in more detail before fitting, and those with no
534: sign of being spatially extended were rejected. Of the remaining galaxies,
535: those with unsuccessful GALFIT
536: measurements of their inclination and position angle were also
537: rejected. Then, only galaxies with inclinations more than $30\degr$ from
538: face-on, and hence a reasonable component of the rotation velocity
539: along the line-of-sight, were fit with ELFIT2PY.
540: In addition, galaxies with a large misalignment angle
541: ($>30\degr$) between their major axes and the slit were rejected,
542: to avoid large and uncertain corrections for this effect.
543: $78$ galaxies were thus remaining to be fit with ELFIT2PY.
544:
545: In a number of cases, the best fitting parameters from ELFIT2PY
546: indicated that the rotation velocity was unreliable, for example due
547: to the emission flux profile being consistent with a point source, or
548: the end of the reliably measurable emission line occuring at smaller
549: radii than the best-fitting intrinsic rotation curve turnover radius.
550: In order to remove such uncertain galaxies from the sample we imposed
551: the following quality criteria.
552:
553: If the scalelength of the fitted emission flux exponential disk
554: profile is too small, then the fit is unlikely to be useful, and thus
555: is rejected. The limit adopted is $0.14\arcsec$ -- the angle subtended
556: by one pixel for a galaxy observed with a slit tilt of $30\degr$, the
557: worst case we admit. This corresponds to $1$--$1.2$ kpc for the
558: redshifts of our objects. In order to also reject galaxies with
559: uncertain emission scalelengths which are close this limit, we reject
560: any galaxy for which the fit scalelength is consistent with being
561: below $0.14\arcsec$ at the $1\sigma$ level.
562:
563: In order to consider how far out in each galaxy we measure the
564: emission line to, we define the \emph{extent} of the line as half of
565: the spatial distance over which the line is reliably traced, after
566: removing single `good' points with no `good' neighbours as spurious.
567: If the extent is less than that one would expect for a point source given
568: the seeing, which we take as $0.5''$ then we reject that galaxy from
569: further consideration.
570:
571: Finally, we also consider whether the transition from the inner
572: rotation curve slope to the flat region is observed with sufficient
573: signal-to-noise. This is done by requiring that the radius at which
574: the fit intrinsic rotation curve turns over is within the reliable
575: extent of the line. Furthermore, in the case where the turnover is not
576: observed, the turnover radius will be highly uncertain. To exclude
577: such galaxies we require that the turnover radius is inconsistent with
578: being beyond the line extent with at least $1\sigma$ confidence.
579:
580: Following the application of these quality criteria, $41$ galaxies
581: remained with reliable rotation velocity measurements.
582:
583:
584:
585: \subsection{Galaxy photometry, $K$-correction, and extinction correction}
586:
587: In addition to the rotation velocity, the second input required for
588: the TF relation is the galaxy absolute magnitude. In order to compare
589: with other samples, while avoiding large $K$-corrections to the
590: photometry, we use the absolute magnitude in the
591: rest-frame $B$-band, $M_B$, in the AB zero-point system. For our
592: galaxies, with median a redshift of $\langle z \rangle =0.85$, this roughly corresponds
593: to the observed-frame $I$ filter (centered at 8100 \AA). For each
594: galaxy of interest, the $I$-band photometry was extracted from the
595: DEEP2 photometric catalogs accompanying the spectroscopic dataset \citep{coil04}.
596: These catalogs were produced using the IMCAT imaging and photometric
597: reduction software of \citet{kaiser95}, having been finally calibrated
598: to the SDSS standard system. As an independent check on the $I$-band
599: photometry, we also verified magnitudes during our SExtractor
600: and GALFIT proceedures.
601:
602:
603:
604: \input{fitstable}
605:
606:
607: Using synthetic
608: template spectra of Sbc, Scd, and Im galaxy types \citep{bruzual03}, appropriate for the range of
609: disk galaxies with star formation and hence [OII] emission, we
610: simulated flux through the rest-frame $B(z=0)$ and redshifted $I(z)$ filters.
611: We applied the average resulting
612: $K$-correction and error to each galaxy according to its redshift.
613:
614: Finally, we applied an inclination-dependent extinction correction to
615: each galaxy. Historically, the corrections of \citet{tully85} and
616: \citet{tully98} have been used most commonly, the latter incorporating
617: an additional dependence on galaxy luminosity (or rotation
618: velocity). We adopt the \citet{tully98} method, for comparison of our
619: work to recent TF samples at similar and lower redshifts. The
620: extinction is defined in \citet{tully98} to be:
621: \[A = \gamma_B \log (a/b), \]
622: with
623: \[\gamma_B = -0.35 (15.31 + M_B ),\]
624: where $a/b$ is the galaxy major-to-minor axial ratio.
625: Table \ref{galtable} displays all of the
626: measured parameters of the galaxies in our
627: sample.
628:
629:
630:
631: \section{The Tully-Fisher Relation}
632: \label{sec:TFdiscuss}
633:
634: From the original catalog of $4300$ galaxies with high quality redshifts
635: in the DEEP2 DR1 data release, the aforementioned steps have reduced
636: the sample significantly, specifically through inspection of the 2d
637: spectra for [OII] emission ($N=400$), imaging data overlap matching
638: ($N=290$), galaxy image profile modeling ($N=150$), and suitable
639: inclinations and slit misalignments for emission line modeling ($N=78$).
640: Following rejection of those galaxies for which we judge the measured
641: rotation velocity to be unreliable, as described in section
642: \ref{sec:fitting}, we now have a useful sample of $41$ galaxies with
643: full photometry, morphology and rotation information. The sample
644: spans a redshift range of $0.75<z<1.3$ with $\langle z \rangle = 0.85$ and an intrinsic
645: $M_B$ magnitude range of $-22.66$ to $-20.57$.
646:
647:
648: \begin{figure}
649: \includegraphics[scale=0.5,angle=-90]{TF_final.ps}
650:
651: \caption{
652: The TF relation of the 41 galaxies in the present sample are plotted
653: in absolute $M_B$ magnitude versus maximal rotation velocity, and
654: overlaid with the resulting linear TF fits discussed in text (in order of
655: decreasing steepness, solid red line: free fit, dotted blue line: slope of
656: V01, dashed green line: slope of TP00).
657: Light dotted blue and dashed green lines indicate
658: the original local V01 and TP00 relations, again in order of decreasing
659: steepness.
660: Object 32001370 (extreme upper left) was excluded
661: from fits as a $>3\sigma$ outlier. \label{newtfplot}}
662: \end{figure}
663:
664:
665: From this sample, we carried out several least-squares fittings
666: to a straight line function: $M_B= a + b\log V_{rot} $, where $a$ and
667: $b$ are the intercept and slope of the TF relation, respectively.
668: The slope and/or intercept value, and scatter width were minimized
669: as the free parameters, and as in \citet{bamford06}, we use weights of
670: $1/\sigma^2$ for each point, where $\sigma^2 = \sigma_{\log V_{rot}}^2
671: + b^2 \sigma_{M_B}^2 + \sigma_{int}^2$, and $\sigma_{int}$ is the
672: intrinsic scatter of the TF relation. One data point was rejected by sigma-clipping,
673: specifically object 32001370, due to its unusual position more than $3\sigma$ from
674: the main sample. Three fits were then carried out -- first,
675: a free-fit of all parameters was allowed. Because this yielded a
676: relatively poorly constrained slope, two subsequent fits were made with
677: slope fixed to local reference samples, specifically those of \citet{tully00} or \citet{verheijen01}.
678: Each of these comparison samples employed a version of the \citet{tully98} extinction correction,
679: as is used for our sample here.
680: Table 2 lists these fitted parameters, and Figure \ref{newtfplot} displays the
681: resulting TF distribution and fits.
682:
683:
684:
685:
686:
687:
688:
689:
690:
691: These fitted parameters present several interesting aspects in comparison to the
692: local fiducial TF relation. First, an overall brightening of $1.5\pm 0.2$ mag
693: is seen in this sample at $\langle z \rangle \sim0.8$ compared to $z\sim0$ galaxies at present.
694: This brightening is consistent with \citet{bohm04} who find an offset of $-(1.22\pm0.56~ {\rm mag})z$, as
695: well as \citet{bamford06}, who find an offset of $ (-1.0\pm0.5~ {\rm mag})z$.
696: The strength of the effect may be explained in part by a colour selection effect
697: in the DEEP2 galaxy sample, which leads to observation of relatively younger stellar population galaxies
698: (and in turn, decreased mass-to-light ratios). A larger sample from the full
699: DEEP2 data may allow this colour-TF relation to be evaluated, as in \citet{kannappan04}.
700: Also of interest is that
701: the brightening of this sample
702: compared to the local fiducial samples seems to be strongest at the bright end of the
703: TF relation ($M_B<-20.5$). This is in contrast with the result of \citet{bohm04},
704: in which little brightening ($\sim0.5$ mag) is found the relatively bright galaxies,
705: but significant brightening found in the faint end galaxies ($\sim1.5$ mag, $M_B\sim-18$).
706:
707:
708: Does this magnitude offset
709: accord with the hierarchical picture, in which small galaxies with
710: pre-formed stellar populations merge into progressively more massive
711: systems, whose mass-to-light ratios change only slowly with the
712: dimming of the passively evolving galaxies? We have calculated the possible
713: contribution of such passive evolution to observed changes in high
714: redshift galaxy luminosities and find that such dimming over time is
715: consistent. Using Bruzual \& Charlot exponential decay
716: models, we find that young
717: galaxies at $z=0.85$ (with luminosity-weighted ages of 3--5 Gyr)
718: will have dimmed by $\Delta M_B \sim1.0$ mag over the 6.5 Gyr to $z=0$.
719: Hence we conclude that the observed
720: brightening of the TF relation in this sample is consistent with pure
721: luminosity evolution.
722:
723:
724: We additionally find that the amount of the brightening offset is within the intrinsic
725: scatter of the sample, and comparable to the scatter found in the
726: local reference populations discussed above. Because intrinsic
727: scatter in the TF relation reflects variations in the mass-to-light
728: ratio of individual galaxies, stellar mass fraction, and deviations
729: from ordered rotation, these effects may be useful to separate in
730: larger similar samples.
731:
732: A few points are worth consideration in the future. First, it
733: will be useful to examine whether TF samples are really dominated by
734: disk galaxies, and the extent to which unidentified irregulars may be
735: influencing and scattering the relation. This may be partially influenced
736: by the selection of the present sample, which is based on the color cut producing the
737: DEEP2 galaxies as well as the use of relatively bright galaxies in order
738: to accomplish the spectral and spatial fitting tasks here. Also, while this and other
739: TF studies examine the blue, star-formation dominated region of the
740: galaxy spectrum, the redder portion of galaxy flux displays tighter
741: correlation and may be more useful for detection of parameter
742: evolution. The effects of momentarily bright star formation
743: episodes (mostly influencing the restframe blue bands) could be mitigated.
744: Such work (at $z=1$ for example) would benefit from $H$ or $K$-band
745: imaging, corresponding to $I$ band in the restframe, as has been demonstrated
746: by \citet{conselice05}. Finally, with future
747: improvements in spectroscopic and imaging spatial
748: resolution (such as with {\it HST}), it may be possible to more thoroughly separate the disk
749: and bulge flux components of examined galaxies, and even address
750: non-uniform star-forming mass regions.
751:
752: \section{Conclusions}
753: \label{sec:conclusions}
754:
755: The population of 41 reliable galaxies examined here is a significant
756: addition to the existing sample at $\langle z \rangle \sim 0.85$, with spectral
757: resolution of $R\sim4000$ allowing the determination of the TF parameters
758: and their evolution. Our detection of a 1.5 mag brightening offset in
759: the TF intercept compared to the present $z=0$ relation is consistent with
760: the previous work of \citet{bohm04}, \citet{bamford06}, and the recent results
761: of \citet{weiner06}. We suggest that this is consistent with
762: passive evolution of galactic sub-units that have already undergone
763: star-formation prior to having assembled into the large galaxies we
764: see today. This is in contrast to the scenario where significant new
765: star formation is proposed to occur within the dark halos that
766: assembled first.
767:
768: Even if galaxies experience only short episodes of star formation
769: activity, and then evolve passively without further mergers, we have
770: demonstrated that changes in correlated parameters, such as in the TF
771: relation, should be observable. As stellar populations age, their
772: decreasing UV luminosities ought to be reflected through changes in
773: galactic mass-to-light ratios; conversely, activity through mergers
774: and rekindled stellar activity leaves similarly observable signatures.
775:
776: In the future, it may be possible to separate larger samples of
777: similar galaxies into redshift bins spanning the entire high redshift
778: range of interest, in order to further quantify the evolution of the
779: TF intercept. The DEEP2 survey provides a promising dataset to carry
780: out such work, and to extend such studies towards examining the role
781: of individual galaxy variations in the overall scatter of TF and
782: similar relations. And while our and other concurrent studies are
783: capable of revealing evolution of the TF linear fit intercept versus
784: redshift, in principle, changes in the TF fit slope itself could also
785: be probed more accurately than has yet been possible -- perhaps
786: revealing differences in useful observables such as star formation
787: efficiency versus time.
788:
789:
790:
791: \section*{Acknowledgments}
792: We thank the DEEP2 collaboration for making available the high-quality
793: data gathered over several years at the CFHT and Keck observatories. We also
794: appreciate the comments and suggestions of the anonymous referee. KC acknowledges funding from a PPARC rolling grant, and AJB
795: acknowledges a Philip Leverhulme Prize. The analysis pipeline used to
796: reduce the DEIMOS data was developed at UC Berkeley with support from
797: NSF grant AST-0071048.
798:
799: \begin{thebibliography}{}
800: \bibitem[Barden et al.(2003)]{barden03} Barden M., Lehnert M. D., Tacconi L., Genzel R., White S., Franceschini A., 2003, astro-ph/0302392
801: \bibitem[Bamford et al.(2005)]{bamford05} Bamford, S.~P., Milvang-Jensen, B., Arag{\'o}n-Salamanca, A., \& Simard, L.\ 2005, MNRAS, 361, 109
802: \bibitem[Bamford et al.(2006)]{bamford06} Bamford, S.~P., Arag{\'o}n-Salamanca, A., \& Milvang-Jensen, B. 2006, MNRAS, 366, 308
803: \bibitem[Bertin \& Arnouts(1996)]{bertin96} Bertin, E., \& Arnouts, S. 1996, AAPS, 117, 393
804: \bibitem[B{\"o}hm et al.(2003)]{bohm03} B{\"o}hm, A., Ziegler, B.~L., Fricke, K.~J., \& FDF Team, t.\ 2003, Ap\& SS, 284, 689
805: \bibitem[B{\"o}hm et al.(2004)]{bohm04} B{\"o}hm, A., et al. 2004, AAP, 420, 97
806: \bibitem[B{\"o}hm \& Ziegler(2006)]{bohm06} B{\"o}hm, A., \& Ziegler, Bodo L. 2006, astro-ph/0611326
807: \bibitem[Bruzual \& Charlot(2003)]{bruzual03} Bruzual, G., \& Charlot, S.\ 2003, MNRAS, 344, 1000
808: \bibitem[Bundy et al.(2005)]{bundy05} Bundy, K., Ellis, R.~S.,
809: \& Conselice, C.~J.\ 2005, ApJ, 625, 621
810: \bibitem[Coil et al.(2004)]{coil04} Coil, A.~L., Newman, J.~A., Kaiser, N., Davis, M., Ma, C.-P., Kocevski, D.~D., \& Koo, D.~C.\ 2004, ApJ, 617, 765
811: \bibitem[Conselice et al.(2005)]{conselice05} Conselice, C.~J., Bundy, K., Ellis, R.~S., Brichmann, J., Vogt, N.~P., \& Phillips, A.~C.\ 2005, ApJ, 628, 160
812: \bibitem[Courteau(1997)]{courteau97} Courteau, S.\ 1997, AJ, 114, 2402
813: \bibitem[Cowie et al.(2003)]{cowie03} Cowie, L.~L., Barger, A.~J., Bautz, M.~W., Brandt, W.~N., \& Garmire, G.~P.\ 2003, ApJL, 584, L57
814: \bibitem[Cowley et al.(1997)]{cowley97} Cowley, D.~J., Faber, S., Hilyard, D.~F., James, E., \& Osborne, J. 1997, Proc. SPIE, 2871, 1107
815:
816: \bibitem[Cuillandre et al.(2001)]{cuillandre01} Cuillandre, J.-C., \& et al. 2001, ASP Conf.~Ser.~232: The New Era of Wide Field Astronomy, 232, 398
817: \bibitem[Dale et al.(2001)]{dale01} Dale, D.~A., Giovanelli, R., Haynes, M.~P., Hardy, E., \& Campusano, L.~E.\ 2001, AJ, 121, 1886
818: \bibitem[Davis et al.(2004)]{davis04} Davis, M., Gerke, B.F., Newman, J.A. 2004, astro-ph/0408344
819: \bibitem[Djorgovski \& Davis(1987)]{djorgovski87} Djorgovski, S., \& Davis, M.\ 1987, ApJ, 313, 59
820: \bibitem[Eggen et al.(1962)]{eggen62} Eggen, O.~J., Lynden-Bell, D., \& Sandage, A.~R.\ 1962, ApJ, 136, 748
821: \bibitem[Eyles et al.(2005)]{eyles05} Eyles, L.~P., Bunker, A.~J., Stanway, E.~R., Lacy, M., Ellis, R.~S., \& Doherty, M.\ 2005, MNRAS, 364, 443
822:
823: \bibitem[Faber et al.(2003)]{faber03} Faber, S.~M., et al.\ 2003, Proc. SPIE, 4841, 1657
824:
825: \bibitem[Flores et al.(2006)]{flores06} Flores, H., Hammer, F., \& Puech, M.\ 2006, New Astronomy Review, 50, 430
826:
827: \bibitem[Ford et al.(2002)]{ford02} Ford, H.C., et al. 2002, Proc. SPIE, 4854, 81
828: \bibitem[Giovanelli et al.(1997)]{giovanelli97} Giovanelli, R., Haynes, M.~P., da Costa, L.~N., Freudling, W., Salzer, J.~J., \& Wegner, G.\ 1997, ApJl, 477, L1
829: \bibitem[Heckman et al.(2004)]{heckman04} Heckman, T.~M., Kauffmann, G., Brinchmann,J., Charlot, S., Tremonti, C., \& White, S.~D.~M.\ 2004, ApJ, 613, 109
830: \bibitem[Kaiser et al.(1995)]{kaiser95} Kaiser, N., Squires, G., \& Broadhurst, T.\ 1995, ApJ, 449, 460
831:
832: \bibitem[Kannappan et al.(2002)]{kannappan02} Kannappan, S.~J., Fabricant, D.~G., \& Franx, M.\ 2002, AJ, 123, 2358
833: \bibitem[Kannappan \& Barton(2004)]{kannappan04} Kannappan, S.~J., \& Barton, E.~J.\ 2004, AJ, 127, 2694
834:
835: \bibitem[Krist \& Hook(2004)]{krist04} Krist, J. \& Hook, R. 2004, Tiny Tim User's Guide, http://www.stsci.edu/software/tinytim
836: \bibitem[Merloni(2004)]{merloni04} Merloni, A.\ 2004, MNRAS, 353, 1035
837: \bibitem[Metevier et al.(2006)]{metevier06} Metevier, A., Kannappan, S., Phillips, D., et al. 2006, in preparation
838: \bibitem[Milvang-Jensen et al.(2003)]{milvang03} Milvang-Jensen, B., Arag{\'o}n-Salamanca, A., Hau, G.~K.~T., J{\o}rgensen, I., \& Hjorth, J.\ 2003, MNRAS, 339, L1
839: \bibitem[Mo et al.(1998)]{mo98} Mo, H.~J., Mao, S., \& White, S.~D.~M.\ 1998, MNRAS, 295, 319
840:
841: \bibitem[Monet et al.(2003)]{monet03} Monet, D.~G., et al. 2003, AJ, 125, 984
842: \bibitem[Nakamura et al.(2006)]{nakamura06} Nakamura, O., Arag{\'o}n-Salamanca, A., Milvang-Jensen, B., Arimoto, N., Ikuta, C., \& Bamford, S.~P.\ 2006, MNRAS, 366, 144
843: \bibitem[Nelan et al.(2005)]{nelan05} Nelan, J.~E., Smith, R.~J., Hudson, M.~J., Wegner, G.~A., Lucey, J.~R., Moore, S.~A.~W., Quinney, S.~J., \& Suntzeff, N.~B.\ 2005, ApJ, 632, 137
844:
845: \bibitem[Peng et al.(2002)]{peng02} Peng, C.~Y., Ho, L.~C., Impey, C.~D., \& Rix, H.-W.\ 2002, AJ, 124, 266
846: \bibitem[Phillips et al.(2002)]{phillips02} Phillips, A.~C., Faber, S., Kibrick, R., Wallace, V., \& DEIMOS Team 2002, Bulletin of the American Astronomical Society, 34, 1320
847: \bibitem[Pierce \& Tully(1992)]{pierce92} Pierce, M.~J., \& Tully, R.~B.\ 1992, ApJ, 387, 47
848: \bibitem[Tully \& Pierce(2000)]{tully00} Tully, R.~B., \& Pierce, M.~J.\ 2000, ApJ, 533, 744
849: \bibitem[Rix et al.(1997)]{rix97} Rix, H.-W., Guhathakurta, P., Colless, M., \& Ing, K.\ 1997, MNRAS, 285, 779
850: \bibitem[Silk(2005)]{silk05} Silk, J.\ 2005, MNRAS, 364, 1337
851: \bibitem[Simard \& Pritchet(1998)]{simard98} Simard, L., \& Pritchet, C.~J.\ 1998, ApJ, 505, 96
852: \bibitem[Simard \& Pritchet(1999)]{simard99} Simard, L., \& Pritchet, C.~J.\ 1999, PASP, 111, 453
853: \bibitem[Sommer-Larsen et al.(2003)]{sommer03} Sommer-Larsen, J., G{\"o}tz, M., \& Portinari, L.\ 2003, ApJ, 596, 47
854:
855: \bibitem[Smith et al.(2004)]{smith04} Smith, J.~K., et al.\ 2004, MNRAS, 354, L19
856: \bibitem[Springel et al.(2005)]{springel05} Springel, V., Di Matteo, T., \& Hernquist, L.\ 2005, MNRAS, 361, 776
857:
858: \bibitem[Swinbank et al.(2003)]{swinbank03} Swinbank, A.~M., et al.\ 2003, ApJ, 598, 162
859: \bibitem[Swinbank et al.(2006)]{swinbank06} Swinbank, A.~M., Bower, R.~G., Smith, G.~P., Smail, I., Kneib, J.-P., Ellis, R.~S., Stark, D.~P., \& Bunker, A.~J.\ 2006, MNRAS, 368, 1631
860: \bibitem[Tully \& Fisher(1977)]{tully77} Tully, R.~B., \& Fisher, J.~R.\ 1977, AAP, 54, 661
861: \bibitem[Tully \& Fouque(1985)]{tully85} Tully, R.~B., \& Fouque, P.\ 1985, ApJs, 58, 67
862: \bibitem[Tully et al.(1998)]{tully98} Tully, R.~B., Pierce, M.~J., Huang, J.-S., Saunders, W., Verheijen, M.~A.~W., \& Witchalls, P.~L.\ 1998, AJ, 115, 2264
863: \bibitem[Tully \& Pierce(2000)]{tully00} Tully, R.~B., \& Pierce, M.~J.\ 2000, ApJ, 533, 744
864: \bibitem[Verheijen(2001)]{verheijen01} Verheijen, M.~A.~W.\ 2001, ApJ, 563, 694
865: \bibitem[Vogt et al.(1997)]{vogt97} Vogt, N.~P., et al.\ 1997, ApJL, 479, L121
866: \bibitem[Vogt et al.(1996)]{vogt96} Vogt, N.~P., Forbes, D.~A., Phillips, A.~C., Gronwall, C., Faber, S.~M., Illingworth, G.~D., \& Koo, D.~C.\ 1996, ApJL, 465, L15
867: \bibitem[Vogt(2001)]{vogt01} Vogt, N.~P.\ 2001, ASP Conf.~Ser.~240: Gas and Galaxy Evolution, 240, 89
868: \bibitem[Weiner et al.(2006)]{weiner06} Weiner, B.J., et al. 2006, astro-ph/0609091
869: \bibitem[Wyithe \& Loeb(2006)]{wyithe06} Wyithe, J.~S.~B., \& Loeb, A.\ 2006, Nature, 441, 322
870: \bibitem[Ziegler et al.(2003)]{ziegler03} Ziegler, B.~L., B{\"o}hm, A., J{\"a}ger, K., Heidt, J., M{\"o}llenhoff, C. 2003, ApJl, 598, L87
871:
872: \end{thebibliography}
873:
874: \appendix
875:
876:
877: \label{lastpage}
878:
879: \end{document}
880:
881: