1: \documentclass{emulateapj}

2: 

3: \lefthead{van der Wel et al.}

4: \righthead{Structural Evolution of Early-Type Galaxies}

5: \slugcomment{The Astrophysical Journal, Accepted}

6: 

7: \begin{document}

8: \newcommand{\clll}{CL~1358+62~}

9: \newcommand{\msl}{MS~2053-04~}

10: \newcommand{\cll}{RX~J0152.7-1357~}

11: \newcommand{\ms}{MS~1054-0321~}

12: \newcommand{\lynx}{RDCS~0848+4453~}

13: \newcommand{\clh}{RDCS~1252-2927~}

14: \newcommand{\sige}{\sigma_{\rm{eff}}}

15: \newcommand{\reff}{R_{\rm{eff}}}

16: \newcommand{\kms}{\>{\rm km}\,{\rm s}^{-1}~}

17: \newcommand{\mvir}{M_{\rm{dyn}}}

18: \newcommand{\dens}{\Sigma_{\rm{eff}}}

19: \newcommand{\msol}{M_{\odot}}

20: 

21: \title{Recent Structural Evolution of Early-Type Galaxies: Size Growth

22:   from $z=1$ to $z=0$\altaffilmark{1}}

23: 

24: \author{Arjen van der Wel\altaffilmark{2}, Bradford

25:   P. Holden\altaffilmark{3}, Andrew W. Zirm\altaffilmark{2}, Marijn

26:   Franx\altaffilmark{4}, Alessandro Rettura{2}, Garth

27:   D. Illingworth\altaffilmark{3} \& Holland C. Ford\altaffilmark{2}}

28: 

29: \altaffiltext{1}{Based on observations with the \textit{Hubble Space

30: Telescope}, obtained at the Space Telescope Science Institute, which

31: is operated by AURA, Inc., under NASA contract NAS5-26555, and

32: observations made with the \textit{Spitzer Space Telescope}, which is

33: operated by the Jet Propulsion Laboratory, California Institute of

34: Technology, under NASA contract 1407. Based on observations collected

35: at the European Southern Observatory, Chile (169.A-0458). Some of the

36: data presented herein were obtained at the W.M. Keck Observatory,

37: which is operated as a scientific partnership among the California

38: Institute of Technology, the University of California and the National

39: Aeronautics and Space Administration. The Observatory was made

40: possible by the generous financial support of the W.M. Keck

41: Foundation. }

42: 

43: \altaffiltext{2}{Department of Physics and Astronomy, Johns Hopkins

44:  University, 3400 North Charles Street, Baltimore, MD 21218; e-mail:

45:  wel@pha.jhu.edu}

46: 

47: \altaffiltext{3}{University of California Observatories/Lick

48:   Observatory, University of California, Santa Cruz, CA 95064}

49: 

50: \altaffiltext{4}{Leiden Observatory, Leiden University, P.O.Box 9513,

51:   NL-2300 AA Leiden, Netherlands}

52: 

53: \begin{abstract}

54:   Strong size and internal density evolution of early-type galaxies

55:   between $z\sim 2$ and the present has been reported by several

56:   authors.  Here we analyze samples of nearby and distant ($z\sim 1$)

57:   galaxies with dynamically measured masses in order to confirm the

58:   previous, model-dependent results and constrain the uncertainties

59:   that may play a role.  Velocity dispersion ($\sigma$) measurements

60:   are taken from the literature for 50 morphologically selected

61:   $0.8<z<1.2$ field and cluster early-type galaxies with typical

62:   masses $\mvir = 2\times10^{11}~\msol$.  Sizes ($\reff$) are

63:   determined with Advanced Camera for Surveys imaging.  We compare the

64:   distant sample with a large sample of nearby ($0.04<z<0.08$)

65:   early-type galaxies extracted from the Sloan Digital Sky Survey for

66:   which we determine sizes, masses, and densities in a consistent

67:   manner, using simulations to quantify systematic differences between

68:   the size measurements of nearby and distant galaxies.  We find a

69:   highly significant difference between the $\sigma$-$\reff$

70:   distributions of the nearby and distant samples, regardless of

71:   sample selection effects.  The implied evolution in $\reff$ at fixed

72:   mass between $z=1$ and the present is a factor of $1.97 \pm 0.15$.

73:   This is in qualitative agreement with semianalytic models; however,

74:   the observed evolution is much faster than the predicted evolution.

75:   Our results reinforce and are quantitatively consistent with

76:   previous, photometric studies that found size evolution of up to a

77:   factor of 5 since $z\sim 2$.  A combination of structural evolution

78:   of individual galaxies through the accretion of companions and the

79:   continuous formation of early-type galaxies through increasingly

80:   gas-poor mergers is one plausible explanation of the observations.

81: \end{abstract}

82: 

83: \keywords{galaxies: clusters: general---galaxies: elliptical and

84: lenticular, cD---galaxies: evolution---galaxies: formation---galaxies:

85: fundamental parameters---galaxies: general---galaxies: photometry}

86: 

87: \section{INTRODUCTION}\label{intro} 

88: 

89: Hierarchical galaxy formation models embedded in a $\Lambda$CDM

90: cosmology predict strong size evolution for massive galaxies.  A

91: higher gas fraction in high-redshift galaxies leads to more

92: dissipation and hence compact galaxies \citep[e.g.,][]{robertson06,

93:   khochfar06a}, and subsequent evolution such as dry merging or

94: accretion of smaller systems can increase the size of a galaxy

95: \citep[e.g.][]{loeb03, naab07}.  The models predict the strongest

96: sample-averaged size evolution for the most massive galaxies

97: \citep{khochfar06b} because of large differences in the gas fraction

98: at different redshifts and because the assembly of massive galaxies

99: continues until very late epochs in a hierarchical

100: framework\citep[e.g.,][]{delucia06}.

101: 

102: Evidence for significant size evolution between $z\sim 2$ and the

103: present has been building up quickly over the past few years

104: \citep[e.g.][]{trujillo04, trujillo06a, franx08}. In particular,

105: galaxies with low star formation rates and high stellar masses

106: ($\gtrsim 10^{11}~\msol$) appear to be extremely compact from $z\sim

107: 1.5$ \citep{daddi05, trujillo06b, trujillo07, longhetti07, cimatti08,

108:   rettura08} to $z\sim 2.5$ \citep{zirm07, toft07, vandokkum08b,

109:   buitrago08}.  Given the similarity between many of their observed

110: properties there is likely to be an evolutionary connection between

111: these distant compact galaxies and the present-day early-type galaxies

112: despite the measured large difference of 2 orders of magnitude in

113: surface mass density \citep[e.g.,][]{zirm07}.

114: 

115: The measurements of sizes and densities of high-redshift galaxies are

116: hampered by many systematic uncertainties (e.g., morphological

117: \textit{K}-corrections, surface brightness dimming, errors in

118: photometric redshifts, and mass measurements). Most of these errors,

119: however, are unlikely to fully account for the observed strong size

120: evolution.  The uncertainty in the mass estimates may be the

121: exception.  For the work in the literature, these mass estimates are

122: always based on the photometric properties of the galaxies.  For a

123: reasonable set of assumptions the photometric stellar mass estimates

124: are not uncertain by more than a factor of 2 or 3 and would not change

125: the inferred evolution significantly.  However, since we infer that

126: $z\sim 2$ galaxies must have physical central densities that are 3

127: orders of magnitude higher than those of local galaxies

128: \citep{vandokkum08b}, further verification of those apparently

129: reasonable assumptions is warranted.  For example, a stellar initial

130: mass function (IMF) that is radically different

131: \citep[e.g.,][]{larson05, fardal07, vandokkum08a, dave08} from a

132: Salpeter-like IMF \citep{salpeter55, scalo86, kroupa01, chabrier03,

133:   hoversten08} could reduce the stellar mass estimates by an order of

134: magnitude, producing perfectly normal galaxies by today's standards.

135: 

136: The spectacular nature of these compact galaxies at $z\sim 2$ could be

137: confirmed by direct, kinematical mass measurements.  However, the

138: quiescent nature of these objects and their consequent lack of

139: emission lines \citep{kriek06b} require absorption-line measurements

140: of their stellar velocity dispersions, which should be as high as

141: $400-500~\rm{km~s}^{-1}$ \citep{toft07, vandokkum08b}.  Unfortunately,

142: with the currently available instrumentation this is not

143: feasible. These $z \sim 2$ galaxies are prohibitively faint at

144: observed optical wavelengths \citep[see, e.g.,][]{cimatti08}, and

145: near-infrared spectroscopy is still maturing as a technique.

146: Continuum detections in the observed NIR have only recently become

147: possible \citep{kriek06b} for the brightest sources, and no detection

148: of absorption lines has been made.

149: 

150: At lower redshifts ($z\sim 1$) absorption-line spectroscopy has for

151: years been a powerful tool to study the evolution of distant

152: early-type galaxies \citep{vandokkum98, vandokkumstanford03,

153:   vandokkumellis03, wuyts04, vanderwel04, treu05a, holden05,

154:   vanderwel05, treu05b, diserego05, jorgensen05}. Size evolution is a

155: gradual process \citep[see, e.g.,][]{trujillo06a}; therefore,

156: intermediate changes in sizes and densities should be observable at

157: these redshifts.

158: 

159: In this paper we compile a sample of galaxies at redshifts $0.8<z<1.2$

160: with measured absorption line velocity dispersions from the literature

161: and that are visually classified as early-type galaxies with the aid

162: of \textit{Hubble Space Telescope} ({\it HST}) imaging from the

163: Advanced Camera for Surveys \citep[ACS;][]{ford98}. We measure the

164: galaxies' sizes from these ACS data.  We then compare this distant

165: sample with nearby early-type galaxies extracted from the Sloan

166: Digital Sky Survey \citep[SDSS;][]{york00}.  This comparison, with

167: careful control of systematic uncertainties, allows us to verify that

168: distant early-type galaxies are indeed significantly more compact than

169: their local counterparts.

170: 

171: The advantage of this approach is that the size and density

172: measurements are independent of the photometric properties of the

173: galaxies apart from the surface brightness profile.  The absence of

174: luminosity and other photometric properties from our analysis assures

175: us that our study does not suffer from the strong possible biases in

176: previous photometric work.  Moreover, deep, high-resolution ACS

177: imaging allows us to determine sizes of $z\sim 1$ galaxies to a

178: precision comparable to that nearby galaxies and in a consistent

179: manner.  Most previous studies verify for biases in the size

180: determinations within their distant samples

181: \citep[e.g.,][]{trujillo04, trujillo06a, cimatti08} but do not extend

182: this analysis to verify the consistency with size measurements of

183: nearby galaxies.

184: 

185: In \S~\ref{nearby} we describe the sample of nearby early-type

186: galaxies and derive the dynamical mass-size relation.  In

187: \S~\ref{distant} we construct the sample of $z\sim 1$ early-type

188: galaxies, determining their masses and sizes in a manner that is

189: consistent with the nearby sample.  In \S~\ref{sim} we quantify

190: systematic effects in our size measurements through simulations.  In

191: \S~\ref{results} we derive the evolution in the dynamical mass-size

192: relation.  In \S~\ref{dis} we compare our results with previous

193: measurements based on photometric mass estimates and semianalytic

194: model predictions, and we discuss size evolution in the broader

195: context of the evolving early-type galaxy population.  Finally, in

196: \S~\ref{sum} we summarize our results and conclusions.  We adopt the

197: following cosmological parameters:

198: $(\Omega_M,~\Omega_{\Lambda},~h) = (0.3,~0.7,~0.7)$.

199: 

200: 

201: %Our study is also complementary in the sense that it probes total

202: %galaxy masses instead of stellar masses. The redshift-dependent

203: %difference between the two may have a second-order effect on size and

204: %density evolution as \citet{robertson06} have suggested that more

205: %efficient cooling of gas in low-mass halos than in high-mass halos

206: %results in the tilt of the fundamental plane \citep[FP]{dressler87,

207: %djorgovskidavis87} and the corresponding mass-dependence of the

208: %central total-to-stellar mass density.

209: 

210: 

211: \section{NEARBY EARLY-TYPE GALAXIES}\label{nearby}

212: 

213: \subsection{Velocity Dispersions and Sizes}\label{nearby:sigrad}

214: 

215: We have extracted a large sample of early-type galaxies at redshifts

216: $0.04<z<0.08$ from the SDSS database \citep[DR6;][]{adelman08} based

217: on the criteria as outlined by \citet{graves07}.\footnote{The IDs,

218:   positions, redshifts, and dispersions for this sample were kindly

219:   provided by G. Graves} Briefly, galaxies on the red sequence and

220: either without emission lines or with high

221: $[\rm{OII}]$~to~$\rm{H}\alpha$ ratios are included in the sample.

222: These criteria effectively exclude star-forming galaxies, but include

223: genuine early-type galaxies with nuclear activity

224: \citep[see][]{yan06}.

225: 

226: The dispersion as measured within the spectroscopic aperture

227: ($\sigma_{\rm{ap}}$) is corrected to match the average dispersion

228: within the effective radius $\reff$ (measured as described below)

229: following \citet{jorgensen95spec}:

230: %

231: \begin{equation}

232:   \sige = \sigma_{\rm{ap}} \left(\frac{\reff}{R_{\rm{ap}}(z)}

233:   \right)^{-0.04},

234: \label{eq:apcor}

235: \end{equation}

236: %

237: where $R_{\rm{ap}}(z)$ is the radius of the SDSS spectroscopic fiber

238: ($1.5''$) in kpc at the distance of the galaxy.  We use the correction

239: from \citet{jorgensen95spec} for consistency with previously published

240: results.  We note that \citet{cappellari06} used better data to

241: improve the aperture correction, but the resulting difference in

242: $\sige$ is only a few percent.

243: 

244: We use GALFIT \citep{peng02} to determine effective radii from

245: the SDSS $g$-band imaging assuming an $R^{1/4}$ profile, leaving the

246: effective radius, the integrated magnitude, the position angle, the

247: axial ratio, and the position of the center as free parameters. The

248: point-spread function (PSF), which is used to deconvolve the image, is

249: constructed for each galaxy separately by co-adding the stars in the

250: frames after drizzling the cutouts to a common center. A more general

251: $R^{1/n}$ profile \citep{sersic68} may provide a more realistic

252: description of the surface brightness distribution of individual

253: early-type galaxies, especially in the presence of a significant

254: disk. However, $n=4$ provides a good description of the average

255: profile of early-type galaxies both nearby \citep{devaucouleurs48} and

256: at $z\sim 1$ (see \S~\ref{distant:profile}).  Moreover, introducing

257: $n$ as an additional free parameter results in unnecessarily large,

258: redshift-dependent systematic uncertainties in the size measurements

259: (see \S~\ref{sim}).

260: 

261: The size parameter that we use in this paper is the circularized

262: effective radius $\sqrt{ab} \equiv a \sqrt{q}$, where $a$ is the

263: effective radius along the major axis (the output parameter of GALFIT,

264: $b$ is the effective radius along the minor axis, and $q$ the axis

265: ratio (as calculated by GALFIT); $\sqrt{ab}$ is a good approximation

266: for optically thin luminosity distributions such as the generally

267: dust-poor early-type galaxies in our samples.  The systematic and

268: random errors of our size determinations are inferred from extensive

269: simulations described in \S~\ref{sim}.

270: 

271: The SDSS spectroscopic catalog suffers from several biases that may

272: mitigate size evolution measurements.  First, compact sources are not

273: targeted for spectroscopy as they may be mistaken for stars or because

274: their central surface brightnesses, i.e., their fiber magnitudes, are

275: too bright.  Second, almost all galaxies that in the literature

276: \citep[see the HyperLEDA database compiled by][]{paturel03} have been

277: claimed to have high, $>300\kms$ velocity dispersions have dispersions

278: of $<300\kms$ in the SDSS \citep[see][Appendix A]{bernardi07a}.  The

279: source of this discrepancy is unknown.  While Bernardi convincingly

280: argues that the SDSS dispersions are more reliable, there are a number

281: of galaxies with large, mutually consistent dispersion measurements

282: from multiple, independent observers, and for which the SDSS

283: dispersion is significantly smaller.  These potential caveats may

284: cause our size evolution measurements to be biased.  We refer to these

285: issues when we present our results in \S~\ref{results}.

286: 

287: \subsection{The Mass-Radius Relation and the Mass-Density Relation}\label{nearby:mr}

288: 

289: From $\sige$ and $\reff$ we derive the total dynamical mass and the

290: corresponding average surface mass density within $\reff$:

291: %

292: \begin{equation}

293:   \mvir = \frac{\beta \reff \sige^2}{G} ,

294: \label{eq:mass}

295: \end{equation}

296: %

297: \begin{equation}

298:   \dens = \frac{\beta \sige^2}{2 \pi G \reff} ,

299: \label{eq:dens}

300: \end{equation}

301: %

302: with $\beta=5$, which has been shown to hold for local galaxies

303: \citep{cappellari06}.  Following \citet{shen03} we adopt the following

304: characterization of the $\mvir$-$\reff$ relation:

305: %

306: \begin{equation} 

307:   \reff=R_c\left(\frac{\mvir}{M_c}\right)^b.

308: \label{eq:mr}

309: \end{equation}

310: %

311: With a least-squares linear fit to all galaxies with mass $\mvir >

312: 3\times 10^{10}~\msol$ we find that the slope is $b=0.56$ and the zero

313: point normalized to a characteristic mass $M_c=2\times 10^{11}~\msol$

314: is $R_c=4.80$~kpc.  We find statistically the same relation if we

315: perform a linear fit to the values of the median $\reff$ in 0.1 dex

316: wide mass bins in the range $10.5 < \log(\mvir) < 12.1$.  The scatter

317: around the best-fit relation is 0.14 dex.

318: 

319: Using stellar masses, $M_*$, \citet{shen03} find the same slope

320: $b=0.56$ for the $M_*$-$\reff$ relation.  Their zeropoint, however, is

321: larger ($R_c=6.14$~kpc).  This is likely due to the difference between

322: $\mvir$ and $M_*$ as \citet{cappellari06} show for a \citet{kroupa01}

323: IMF (which is also used by Shen et al.~2003) that $\mvir\sim 1.4M_*$.

324: This translates into a difference in $R_c$ of $\sim$20\%, close to the

325: observed difference.  Furthermore, \citet{shen03} analyze SDSS

326: $r$-band imaging whereas we use $g$-band imaging.

327: 

328: 

329: \section{DISTANT EARLY-TYPE GALAXIES}\label{distant}

330: 

331: \subsection{Velocity Dispersions and Sizes}\label{distant:sigrad}

332: 

333: Several authors have published velocity dispersion measurements of

334: early-type galaxies at $z\sim 1$.  We compile the data from three

335: different data sets for which the selection criteria are well

336: understood so that systematic effects introduced through selection

337: effects can be properly modeled.  Our compiled sample contains

338: galaxies in the redshift range $0.8<z<1.2$ in the Chandra Deep

339: Field-South \citep[CDF-S,][]{vanderwel04, vanderwel05} and the Hubble

340: Deep Field-North \citep[HDF-N,][]{treu05a, treu05b}. In addition, we

341: include galaxies in the massive, X-ray selected cluster \ms at

342: $z=0.831$ \citep{wuyts04}.  The seven cluster galaxies at $z>1$ for

343: which dispersions have been measured \citep{vandokkumstanford03,

344:   holden05} are not included because of the paucity of this sample,

345: which prevents us to accurately model selection effects.  The final

346: sample only contains galaxies with $S/N>10~\rm{\AA}^{-1}$ since

347: dispersions derived from spectra with lesser quality can suffer from

348: large ($>10\%$) systematic uncertainties.  The same aperture

349: corrections are included as for the nearby galaxies

350: (Eq.~\ref{eq:apcor}), with in this case $R_{\rm{ap}}(z)$ the radius of

351: a circle with area $1''\times 1.25''$ (the width of the slits and the

352: typical length of the extracted region) in kiloparsec at the redshift

353: of the galaxy. The data are given in Table \ref{tab}.

354: 

355: For all galaxies ACS imaging is available. In order to produce an

356: internally consistent data set, we re-measure the sizes of all

357: galaxies.  GOODS\footnote{See http://www.stsci.edu/science/goods/}

358: provides deep, publicly available ACS imaging of the CDF-S and the

359: HDF-N \citep{giavalisco04} in four filters. We use the F850LP

360: ($z_{850}$-band) images in order to match the rest-frame wavelength at

361: which the sizes of the nearby comparison sample are measured (the SDSS

362: $g$-band; see \S~\ref{nearby}).  For the \ms cluster ACS imaging has

363: been taken as part of the guaranteed time observation program

364: \citep{blakeslee06}.  We use the F775W ($i_{775}$-band) imaging as the

365: available $z$-band imaging is of lesser quality.  At this redshift

366: rest frame $g$ falls in between $i_{775}$ and $z_{850}$ such that the

367: morphological \textit{K}-correction is not a problem; the $z\sim 0.8$

368: galaxies in the sample of \citet{treu05b} are only $3\% \pm 4\%$

369: smaller in the $i_{775}$ band than in the $z_{850}$ band.

370: 

371: With GALFIT we determine effective radii in the same manner as for the

372: nearby galaxy sample (see \S~\ref{nearby:sigrad}). The PSF is

373: constructed with Tiny Tim \citep{krist95}, even though using stars

374: results in negligible differences \citep[see,~e.g.,][]{vanderwel05,

375:   treu05b}.  Errors are discussed in \S~\ref{sim}, and the data are

376: given in Table \ref{tab}.

377: 

378: We use Eqs.~\ref{eq:mass} and \ref{eq:dens} to compute masses and

379: surface densities. Again, we adopt $\beta=5$, which has been shown to

380: hold for distant nonrotating elliptical galaxies

381: \citep{vandermarel07b, vanderwel08b}.  For rotating early-type

382: galaxies the situation appears to be more complex \citep{vanderwel08b}

383: in the sense that $\beta$ is possibly $\sim$20\% larger than 5.  We

384: comment on the impact of this possible complication on our size

385: evolution measurement in \S~\ref{results:mrevol}.  A low-mass cutoff

386: of $3\times 10^{10}~\msol$ is applied since below this limit no useful

387: samples are available due to severe incompleteness of the surveys

388: \citep[see, e.g.,][]{vanderwel05}.

389: 

390: 

391: \subsection{The Average Surface Brightness Profile}\label{distant:profile}

392: 

393: In determining the sizes of the nearby and distant galaxies in the

394: previous sections we assumed that an $R^{1/4}$ profile provides an

395: accurate description of early-type galaxies.  We know this to be true

396: for nearby galaxies, but not for $z\sim 1$ early-type galaxies.

397: %Simulations assuming such an $R^{1/4}$ profile are useful but do not,

398: %of course, address the validity of that assumption.  

399: If a more general $R^{1/n}$ profile is adopted, measured values tend

400: to cluster around $n=4$ \citep[see, e.g.,][]{blakeslee06, rettura06}.

401: However, there is a possibility that the true values of $n$ are

402: different; at large radii the ``wings'' of the profile become quickly

403: overwhelmed by background noise, even in the deepest \textit{HST}

404: imaging.  Because $n$ and the measured $\reff$ are correlated,

405: assuming $n=4$ for all redshifts introduces systematic errors in case

406: $n$ evolves with redshift.

407: 

408: \begin{figure}[t]

409: \epsscale{1.1} 

410: \plotone{stack.eps}

411: \caption{ Stacked $z_{850}$-band image of 29 $z\sim 1$ elliptical

412:   galaxies in the CDF-S and the HDF-N and the residuals after

413:   subtracting $R^{1/n}$ model profiles, with $n=$ 3, 4, 5, 6, 7.  Each

414:   panel is 7.68" on a side, which corresponds to 62 kpc at $z=1$.  The

415:   circle indicates the effective radius as measured with the $R^{1/4}$

416:   model profile (0.31", or 2.5 kpc at $z=1$).  The model with $n=4$

417:   provides the best fit.  The model with $n=3$ produces positive

418:   residuals at large radii; models with high $n$ produce negative

419:   residuals.  This justifies our choice to adopt the $R^{1/4}$ law to

420:   model the surface brightness profiles of both local and distant

421:   early-type galaxies. }

422: \label{stack}

423: \end{figure}

424: 

425: In order to examine the profiles of the $z\sim 1$ galaxies at large

426: radii, we median-stack the $z_{850}$-band images of all elliptical

427: galaxies (S0s are excluded) without bright neighbors in our CDF-S and

428: HDF-N samples (see Fig.~\ref{stack}).  The images of the individual

429: galaxies are drizzled onto a common central position. Due to

430: imperfections in this procedure, the stacked PSF may not be an

431: accurate description of the PSF of the stacked image.  However, this

432: does not play a role since the deviations we are interested in have

433: scales that are an order of magnitude larger than the PSF.

434: 

435: With GALFIT we subtract $R^{1/n}$ profiles with integer values $n=3-7$

436: (see Fig.~\ref{stack}).  The negative residuals outside $\reff$~for

437: models with large $n$ and the positive residuals for models with small

438: $n$ indicate that these limiting cases provide poor fits of the outer

439: regions of elliptical galaxies at $z\sim 1$.  The $R^{1/4}$ and

440: $R^{1/5}$ profiles provide the best description of their average

441: surface brightness distributions. This visual impression is confirmed

442: by the $\chi^2$-values of the respective fits: $\chi^2=0.5$ for both

443: $n=4$ and for $n=5$, whereas $\chi^2>0.7$ for other values of $n$.

444: Interestingly, $n$ does not evolve significantly with redshift, and we

445: conclude that it is safe to assume that choosing $n=4$ for both nearby

446: and distant early-type galaxies does not introduce significant

447: systematic errors.

448: 

449: 

450: \section{SIMULATIONS OF SIZE MEASUREMENTS}\label{sim}

451: 

452: To test the robustness of our size determinations of local and distant

453: early-type galaxies in \S\S~\ref{nearby} and \ref{distant} we simulate

454: size measurements by using SDSS $g$-band imaging of 45 early-type

455: galaxies in the Virgo Cluster \citep{mei07}.  The pixels of the

456: mosaics of the Virgo Cluster galaxies are re-binned to account for the

457: different pixel scales of the various instruments, and different

458: cosmological distances of the galaxies at higher redshifts. The

459: redshift range is $z=0.04-0.08$ for the nearby sample and $0.8-1.2$

460: for the distant sample. The physical sizes of the simulated galaxies

461: are thus conserved.  For each redshift ($z=0.04, 0.06, 0.08, 0.8,

462: 1.2$) we run $\sim$200 simulations with different values for the flux

463: density of the simulated galaxies, which are chosen such that the

464: simulated galaxies have the same range in apparent magnitude as the

465: observed galaxies in our samples.  After convolution with the

466: appropriate PSF and the addition of Poisson noise the seed galaxies

467: are inserted into empty parts of real images.  Their sizes are

468: measured with GALFIT in the same manner, by fitting a $R^{1/4}$ law,

469: as for the real galaxies.  Because we are mainly interested in the

470: systematic differences in the size determinations within and between

471: our nearby and distant samples, we assume the size determinations

472: based on the $z=0.04$ simulated SDSS images of the Virgo Cluster

473: galaxies as the baseline against which we compare the other simulated

474: size measurements.

475: 

476: \begin{figure}[t]

477: \epsscale{1.2} 

478: \plotone{simula.eps}

479: \caption{ Random and systematic errors in the size determinations of

480:   early-type galaxies with SDSS imaging at ({\it a}) $z=0.06$ and

481:   ({\it b}) $z=0.08$ and with \textit{HST} imaging at ({\it c})

482:   $z=0.8$ and ({\it d}) $z=1.2$, all with respect to the size

483:   measurements at $=0.04$, which are used as the benchmark in our

484:   analysis. These are the results of simulations with 45 early-type

485:   galaxies in the Virgo Cluster. The systematic offsets are indicated

486:   by the dashed lines. The scatter in the offsets, considered to be

487:   the random error in the size determinations, are also listed.}

488: \label{simul}

489: \end{figure}

490: 

491: \begin{figure*}[t]

492: \epsscale{1.1}

493: \plottwo{sig_rad.eps}{sig_rads.eps}

494: \caption{ The $\sige$-$\reff$ distributions of the nearby sample of

495:   early-type galaxies (\textit{gray scale}) and the distant sample

496:   (data points). The red data points are cluster galaxies, and the

497:   blue data points are field galaxies.  The error bars at the top

498:   right indicate the typical values of the errors for the distant

499:   sample.  The solid line indicates $\log(\mvir)=10.5$.  Dotted lines

500:   indicate lines of constant $\mvir$ (\textit{parallel to the solid

501:     line}), spaced by 0.5 dex, and lines of constant surface density

502:   $\dens$ (\textit{perpendicular to the solid line}), also spaced by

503:   0.5 dex.  The left-hand panel shows the entire nearby sample; the

504:   right-hand panel only shows those galaxies in the nearby sample that

505:   would be included in the distant sample considering the selection

506:   effects, that apply to the surveys (see \S~\ref{results:sigrad}).

507:   The highly significant offset ($>99.9\%$ significance) between the

508:   local and distant samples implies significant structural evolution

509:   in the early-type galaxy population between $z\sim 1$ and the

510:   present. }

511: \label{sig_rad}

512: \end{figure*}

513: 

514: The results of the simulations are shown in Fig.~\ref{simul}.  Within

515: the nearby sample we find systematic, redshift-dependent differences,

516: of up to $\sim$10\%.  Random errors, derived from the scatter in the

517: sizes inferred from the simulated images, are typically less than 5\%.

518: Systematic differences between the nearby and distant samples can be

519: as large as 20\% for small galaxies, where at high redshift the sizes

520: are overestimated.  Random errors are typically 10--15\%. We find no

521: systematic trends with magnitude. The reason for this is that all

522: galaxies are relatively bright compared to the depth of the data sets,

523: such that the limiting factor in the size measurements is spatial

524: resolution.

525: 

526: Adopting a $R^{1/n}$ law with $n$ as a free parameter may improve the

527: quality of the fits to individual galaxies.  However, our simulations

528: reveal that the random errors increase to $\sim$20--25\%, without much

529: change in the systematic errors.  Together with the analysis of

530: stacked images (\S~\ref{distant:profile}), this test justifies our

531: choice to use the $R^{1/4}$ law to describe the surface brightness

532: profiles of all galaxies in both the nearby and the distant samples.

533: 

534: The sizes we derive in the \S\S~\ref{nearby} and \ref{distant}, and

535: the derived quantities $\mvir$ (Eq.~\ref{eq:mass}) and $\dens$

536: (Eq.~\ref{eq:dens}), are corrected to account for systematic

537: measurement errors.  Those corrections depend on redshift and are

538: interpolated between the values listed in Fig.~\ref{simul}.  For

539: simplicity the dependence on size is not taken into account, such that

540: the remaining systematic uncertainty is $\sim$5\%.

541: 

542: \section{STRUCTURAL EVOLUTION OF EARLY-TYPE GALAXIES}\label{results}

543: 

544: \subsection{Evolution of the $\sigma$-Radius Distribution}\label{results:sigrad}

545: 

546: In Figure \ref{sig_rad} we compare the $\sige$-$\reff$ distributions

547: of the nearby and distant early-type galaxy samples.  This unusual

548: projection of the fundamental plane \citep[FP;][]{dressler87,

549:   djorgovskidavis87} has a very large scatter.  However, the advantage

550: is that changes with redshift are independent of luminosity evolution.

551: Despite the large scatter, it is clear that the distant galaxies are

552: offset from the nearby galaxies.  Galaxies with the properties of

553: typical galaxies in the distant sample ($\sige\sim 250~\kms$;

554: $\reff\sim 3$ kpc) are rare in the local universe.  In the nearby

555: sample, galaxies with $\sige\sim 250~\kms$ have much larger sizes, and

556: galaxies with sizes $\reff\sim 3$ kpc have dispersions of $\sige\sim

557: 150~\kms$. These numbers are only intended to guide the eye. A

558: quantitative analysis of the differences between nearby and distant

559: galaxies is presented below.

560: 

561: The distant sample is not directly comparable with the nearby sample

562: in its entirety (\textit{left}), as the nearby sample reaches to much

563: lower masses.  In order to assess the question whether the true,

564: underlying $\sige$-$\reff$ distribution of distant galaxies is

565: different from the $\sige$-$\reff$ distribution of nearby galaxies, we

566: need to remove the galaxies in the nearby sample that would not be

567: included at $z\sim 1$ due to sample selection effects.  The sub-sample

568: of nearby galaxies that is observable at $z=1$ is shown in the right

569: panel of Fig.~\ref{sig_rad}.  The two criteria that the galaxies in

570: the observable sub-sample satisfy are $L>L_{\rm{min}}$ and $\reff <

571: R_{\rm{eff,max}}$.  $L_{\rm{min}} \sim 10^{10} L_{\odot,\rm{B}}$ is

572: the luminosity limit for the field $z\sim 1$ surveys \citep[see,

573: e.g.,][]{vanderwel05} after correcting for luminosity evolution

574: between $z=1$ and the present \citep[0.555 dex;][]{vandokkum07}.  For

575: the \ms cluster sample from \citet{wuyts04} this is $1.8\times

576: 10^{10}L_{\odot,\rm{B}}$.  The second criterion $\reff <

577: R_{\rm{eff,max}}$ takes into account that high signal-to-noise ratio

578: ($S/N$) spectra are harder to obtain for low surface brightness

579: galaxies than for high surface brightness galaxies with the same

580: luminosity; i.e., the distant sample is biased in favor of small

581: galaxies.  The $S/N$ of the spectra of \citet{vanderwel05} and

582: \citet{treu05b} do not precisely scale linearly with luminosity

583: $L=I\reff^2$, where $I$ is the surface brightness, but as $S/N \propto

584: I\reff^{1.6}$. This implies that, at fixed luminosity $L$, $S/N

585: \propto R^{-0.4}$. Since a dispersion measurement requires a minimum

586: $S/N$ ($\sim 12 \AA^{-1}$), a galaxy with luminosity $L$ has a maximum

587: radius $R_{\rm{eff,max}} \propto L^{2.5}$ for which its dispersion can

588: be determined. We use the luminosity limits of the surveys discussed

589: above to normalize the dependence between luminosity and maximum size;

590: we simply assume that for the smallest galaxies ($\reff = 1$~kpc) the

591: luminosity limit coincides with the size limit such that we have

592: %

593: \begin{equation}

594:   R_{\rm{eff,max}}~(\rm{kpc})=\left(\frac{L}{L_{\rm{min}}}\right)^{2.5}.

595: \label{eq:sel}

596: \end{equation}

597: 

598: One would expect that for galaxies smaller than 3 kpc the

599: signal-to-noise ratio of the spectra would not depend on size any

600: longer since seeing generally dominates the apparent sizes of such

601: small galaxies at $z\sim 1$.  Because of the variety of telescopes,

602: weather conditions, and data reduction techniques, this, however, is

603: washed out and not apparent in the data.  We note that the

604: introduction of, effectively, a rudimentary surface brightness

605: criterion is a step forward in modeling the selection effects with

606: respect to earlier attempts that only take total luminosity into

607: account.

608: 

609: The difference between the $\sige$-$\reff$ distributions at low and

610: high redshift is highly significant, even after taking selection

611: effects into account (Fig.~\ref{sig_rad}, \textit{right}).  The

612: two-dimensional Kolgomorov-Smirnov statistic has a high value

613: ($D=3.71$), which implies that it is extremely unlikely that the nearby

614: and distant samples are drawn from the same distribution.  By

615: repeatedly drawing samples from the nearby sample with the same size

616: as the distant sample we confirm this: less than 0.001\% of the

617: simulated samples have $D=3.71$ or higher.

618: 

619: 

620: \begin{figure}[t]

621: \epsscale{1.2}

622: \plotone{M_rad.eps}

623: \caption{Mass-size relation for the nearby sample (\textit{solid

624:     line}) and at $z\sim 1$ (\textit{dashed line}); the symbols are

625:   the same as in Fig.~\ref{sig_rad}.  For the derivation of the

626:   $\mvir$-$\reff$ relation for the nearby sample see

627:   \S~\ref{nearby:mr}; for the derivation of the $\mvir$-$\reff$

628:   relation for the distant sample see \S~\ref{results:mrevol}.  The

629:   smaller, inset panel shows the distribution of the two samples

630:   around the $\mvir$-$\reff$ relation of the nearby sample

631:   (\textit{the solid line in the large panel}).  The distant galaxies

632:   are $1.8\pm0.1$ times smaller than the nearby galaxies. It appears

633:   that the most massive galaxies do not show as large an offset. This

634:   indicates that size evolution may be slower for the highest mass

635:   galaxies than for low-mass galaxies, but it has to be kept in mind

636:   that these very massive galaxies are brightest cluster galaxies and

637:   may therefore have developed differently from other galaxies.}

638: \label{M_rad}

639: \end{figure}

640: 

641: \subsection{Evolution of the Mass-Radius Relation}\label{results:mrevol}

642: 

643: The structural difference between the nearby and distant samples

644: described in the previous section implies that the $\mvir$-$\reff$ and

645: $\mvir$-$\dens$ relations evolve with redshift.  In Fig.~\ref{M_rad}

646: we show the $\mvir$-$\reff$ relation for the distant sample, and

647: compare this with the equivalent relation for nearby galaxies derived

648: in \S~\ref{nearby:mr}.  Clearly, the relation shifts to smaller

649: radii from low to high redshift.

650: 

651: Parametrized as in Eq.~\ref{eq:mr} we find $R_c=2.58\pm0.17$

652: and $b=0.65\pm0.06$ with a scatter of $0.117\pm0.013$ dex (after

653: subtracting the observational uncertainties in quadrature).  The

654: errors are estimated with a bootstrap/Monte-Carlo simulation in which

655: the data points are randomly sampled and varied according to the

656: (correlated) measurement errors which are assumed to be Gaussian.  The

657: systematic error in $\reff$ of 5\% (see \S~\ref{sim}) is also taken

658: into account.

659: 

660: The treatment of the selection effects described in the previous

661: section shows that the observed size evolution seen in

662: Fig.~\ref{M_rad} is not an artifact.  However, given the nature of the

663: selection effects, which favor small galaxies over large galaxies, the

664: intrinsic amount of size evolution and possible evolution in the slope

665: and scatter of the $\mvir$-$\reff$ relation must be inferred through

666: careful modeling.  The goal is to derive the intrinsic $\mvir$-$\reff$

667: relation at $z\sim 1$ that reproduces the observed $\mvir$-$\reff$

668: distribution after applying the selection criteria.  We take an

669: iterative approach due to the interdependence of the selection

670: criteria and the amount of evolution in zeropoint, slope, and scatter

671: of the $\mvir$-$\reff$ relation.  In the following we de-evolve the

672: properties of the nearby sample to constrain the form of the true,

673: underlying $z\sim 1$~$\mvir$-$\reff$ relation.

674: 

675: %We assume that the mass function of early-type galaxies, other than

676: %its normalization, does not evolve between $z\sim 1$ and the present.

677: %This allows us to use the $\mvir$ distribution of the nearby sample.

678: The simplest evolutionary scenario is a change in the zero point $R_c$

679: (see Eq.~\ref{eq:mr}).  For each object in the nearby sample the size

680: is reduced by the same amount $\Delta\log(\reff)$, and those that do

681: not satisfy the selection criteria described in the previous section

682: are removed. From the remaining sub-sample the ``observed''

683: $\mvir$-$\reff$ relation is determined.  The different selection

684: criteria and sample sizes for field and cluster galaxies are taken

685: into account in this process, which is repeated for many different

686: values of $\Delta\log(\reff)$.  We find that an intrinsic value of

687: $R_c=2.64\pm0.18$ reproduces the observed value of $R_c=2.58\pm0.17$.

688: Hence, it appears that selection effects do not strongly affect the

689: inferred size evolution.

690: 

691: However, the scatter of the assumed intrinsic distribution (0.14 dex)

692: is higher than the observed scatter ($0.117\pm0.013$ dex).  This

693: cannot be explained by selection effects in the simple scenario

694: described above.  It is therefore required that the scatter, as well

695: as the zero point, is also treated as an evolving parameter.  This is

696: implemented in our analysis by reducing or increasing the offset of

697: each galaxy in the nearby sample from the best-fit $\mvir$-$\reff$

698: relation by a given fraction.  Doing so, we find that the best-fitting

699: zero point $R_c$ is not different from the earlier estimate based on a

700: nonevolving scatter.  We also find that the evidence for evolution in

701: the scatter is weak ($\sim 1.5~\sigma$).  This exercise mainly serves

702: to show that our size-evolution result in not sensitive to the amount

703: of evolution in the scatter allowed by the observations.

704: 

705: A similar verification must be carried out for evolution in the slope

706: of the $\mvir$-$\reff$ relation.  Allowing only the scatter and the

707: size to evolve, as described above, the inferred slope of the

708: ``observed'' $\mvir$-$\reff$ relation is 0.59, marginally consistent

709: with the true observed slope of $b=0.65\pm0.06$.  If we treat the

710: slope as an additional, third free parameter we confirm that evolution

711: in the slope, as constrained by our measurements, does not affect our

712: size-evolution measurement.  An intrinsic slope of $b=0.61$ provides a

713: better fit than the original slope of the $\mvir$-$\reff$ relation of

714: the nearby sample ($b=0.56$), but the difference is marginal ($\sim

715: 1~\sigma$).

716: 

717: We conclude that, despite (weak) evidence for evolution in the slope

718: and the scatter of the $\mvir$-$\reff$ relation with redshift, there

719: is no significant improvement in modeling the observations by adopting

720: slope and scatter as free parameters.  Modeling the evolution by a

721: fractional change in size, regardless of mass and offset from the

722: local $\mvir$-$\reff$ relation, provides an equally good fit.  Most

723: importantly, changing the slope and scatter within the range allowed

724: by the observations does not affect the inferred size evolution.  We

725: find that $R_c=2.64\pm0.18$ kpc at $z=0.90$, a factor of

726: $1.8\pm 0.1$ times smaller than at $z\sim 0.06$.

727: 

728: The weak evidence for a change in slope of the $\mvir$-$\reff$

729: relation may also be interpreted as a difference between field and

730: cluster galaxies, as the more massive galaxies in our sample tend to

731: be cluster galaxies.  Assuming that slope and scatter remain constant

732: but that the zero point of the $\mvir$-$\reff$ relation evolves

733: differently for field and cluster galaxies, we find that $R_c=2.49$

734: kpc for field galaxies and $R_c=3.06$ kpc for cluster galaxies.  The

735: 1~$\sigma$ error on this difference of 0.57 kpc is 0.32 kpc.  The true

736: error may be larger since in this estimate it is assumed that scatter

737: and slope behave the same in the different environments and that there

738: are no relative systematic errors in the size determinations of field

739: and cluster galaxies.  The evidence for a difference between the size

740: evolution of field and cluster early-type galaxies is therefore weak

741: \citep[see also][]{rettura08}. However, we have to keep in mind that

742: so far only a very small number of clusters is considered. Future

743: studies will need to extend the existing analyses to a larger number

744: of clusters to verify the general validity of the results.

745: 

746: So far, we have assumed that the masses of the galaxies do not change.

747: Our justification is that the scatter hardly depends on mass; the

748: effect of a changing mass function on modeling selection effects is

749: expected to be small.  However, physically speaking, it is unnatural

750: to propose size evolution without changes in the masses of galaxies.

751: Moreover, if the characteristic mass above which the number density of

752: galaxies drops off exponentially evolves with redshift, selection

753: effects will change as well.  The simplest way to implement mass

754: evolution is to assume that $M\propto \reff$ (Eq.~\ref{eq:mass}).

755: Including this in our modeling procedure shows that the effect on the

756: inferred size evolution is less than 5\%, and we therefore adopt the

757: results with no mass evolution.

758: 

759: We recall that the nearby sample is biased against compact early-type

760: galaxies (\S~\ref{nearby:sigrad}).  The potentially underestimated

761: number of galaxies with dispersions $\sigma>300~\kms$ is unlikely to

762: drastically affect the size-evolution determination for the sample as

763: a whole as the average dispersion of the galaxies in the distant

764: sample is smaller than that.  However, the slope of the local

765: $\mvir$-$\reff$ relation is possibly overestimated, which would lead

766: to an underestimate of the slope evolution.  More important is the

767: problem that small galaxies are missed because of their photometric

768: misclassification as stars in the SDSS.  To fully address this issue a

769: complete analysis of the SDSS photometric catalog is required, which

770: is clearly beyond the scope of this paper.  However, we can say that

771: it is highly unlikely that the average size of nearby early-type

772: galaxies is underestimated by a factor of 2 because of this bias.

773: On the other hand, for the interpretation of our results and

774: identifying the mechanisms responsible for size evolution (see

775: \S~\ref{dis:3}) this bias could prove to be important.

776: 

777: In \S~\ref{distant} we noted that the dynamical mass estimate as

778: adopted in this paper (Eq.~\ref{eq:mass}, with $\beta=5$) may be too

779: low for rotating early-type galaxies.  If this is the case, then size

780: evolution for these galaxies will be underestimated by $\sim$10\%.

781: Since this is within the uncertainties of our measurements we do not

782: take this further into account.

783: 

784: \begin{figure}[t]

785: \epsscale{1.2}

786: \plotone{M_dens.eps}

787: \caption{Mass-density relation at $z\sim 1$. The symbols and lines are

788:   the same as in Figs.~\ref{sig_rad} and \ref{M_rad}.  The $z\sim 1$

789:   early-type galaxies are $\sim 4$ times more dense than their nearby

790:   counterparts. The prediction of the semianalytic size-evolution

791:   model for elliptical galaxies from \citet{khochfar06b} is shown as

792:   the dotted line.  The error bars indicate the predicted size

793:   evolution between $z=0.8$ and $z=1.2$, the redshift range of our

794:   distant sample.  Despite qualitative agreement, there are

795:   significant quantitative differences between the predicted and

796:   observed evolution.}

797: \label{M_dens}

798: \end{figure}

799: 

800: Obviously, size evolution at fixed mass translates into density

801: evolution. This is illustrated in Fig.~\ref{M_dens} where we compare

802: the density distribution of $z\sim 1$ early-type galaxies with the

803: $\mvir$-$\dens$ relation for nearby galaxies.  Because $\dens$ does

804: not strongly depend on $\mvir$, evolutionary trends are readily

805: visible; $z\sim 1$ early-type galaxies are $\sim$4 times more dense

806: than their local counterparts.  The apparent change in slope can

807: possibly be explained by selection effects, completely analogous to

808: our conclusion that this is the case with the $\mvir$-$\reff$

809: relation.  Note that compared to the increase in projected density,

810: the increase in physical density will be even larger.

811: 

812: Up until recently, early-type galaxies were thought to evolve more or

813: less passively.  This appears to be an over-simplification and may

814: apply more to their stellar populations than to their structural

815: properties.  In the following section we discuss possible explanations

816: in the context of theoretical predictions and the comparison with

817: results from studies with different observational strategies.

818: 

819: 

820: \section{DISCUSSION}\label{dis}

821: 

822: \subsection{Comparison with Photometric Size-Evolution Measurements}\label{dis:1}

823: 

824: The main goal of this paper is to use dynamical measurements to

825: investigate whether early-type galaxies were smaller and denser in the

826: past.  Previous work has shown that the stellar mass surface density

827: is higher, but there are a number of issues with such studies as they

828: rely on stellar population models and they ignore possible changes in

829: the underlying dark matter profile.

830: 

831: In Fig.~\ref{lit_z} we compare the size-evolution results presented in

832: the previous section with size-evolution results for early-type

833: galaxies based on photometric mass estimates.  For all the literature

834: samples we take the mean redshift and the mean stellar mass

835: (normalized to the Kroupa IMF), and compute the mean offset from the

836: local mass-size relation from \citet{shen03}.  We include four

837: intermediate redshift cluster galaxy samples with photometrically

838: measured masses and sizes from WFPC2 or ACS imaging.  The data are

839: described by \citet{holden07} and the sizes are measured as described

840: in this paper (\S~\ref{distant}).  These four clusters are \clll at

841: $z=0.33$, \msl at $z=0.59$, and \ms and \cll, both at $z=0.83$.  Note

842: that we also include \ms in the present study with dynamical mass

843: measurements.  The agreement between the independent measurements

844: confirms that at least out to $z\sim 1$ dynamical and photometric mass

845: estimates based on optical colors and spectral energy distributions

846: agree within the statistical errors as was previously shown by

847: \citet{vanderwel06b}, \citet{rettura06}, and \citet{holden06}.

848: 

849: The literature samples have all been selected in different ways, and

850: so a direct comparison with our work may not be straightforward. Not

851: all samples are morphologically selected; many are selected by their

852: spectral or photometric properties. In the local universe there is

853: substantial overlap between samples of early-type galaxies that are

854: selected by different criteria; therefore, it is a reasonable

855: assumption to suppose this to also be the case at high redshift, where

856: different indicators (low star formation rates, red colors, smooth

857: visual appearance) also reflect a common nature. Recently, several

858: studies have shown hints that this is indeed the case

859: \citep[e.g.,][]{vandokkum08b, kriek08b}, but these issues need to be

860: further addressed in the future.

861: 

862: Even with this cautionary proviso, the broad agreement between the

863: results presented in this paper and the photometric results at higher

864: redshifts is striking. All studies included in Fig.~\ref{lit_z} are

865: consistent with significant size evolution of several factors between

866: $z\sim 1-2$ and the present for galaxies with a given mass.  A linear

867: fit in log-log space to our two data points at $z\sim 0.06$ and $z\sim

868: 1$ gives $\reff(z)\propto (1+z)^{-0.98\pm0.11}$.  With a linear fit to

869: the photometric data the inferred rate of evolution is

870: $\reff(z)\propto (1+z)^{-1.20\pm0.12}$, where the error is obtained

871: via a bootstrap/Monte Carlo simulation.

872: 

873: The broad agreement of our measurement of the size evolution of

874: early-type galaxies with the photometric studies is encouraging and

875: alleviates concerns about serious systematic effects that potentially

876: could have compromised previous work.  Most notably, uncertainties in

877: the photometric mass estimates used in all other previous work appear

878: to have a limited impact, at least compared to factors of $\gtrsim$5

879: which would mimic the strong, observed size and density evolution.

880: Uncertainties in photometric mass estimates on the level of a factor

881: of $\sim$2 due to differences among the various stellar population

882: models \citep[e.g.,][]{bruzual03, maraston05} remain an issue, but to

883: invoke, for example, an unconventional stellar IMF as an alternative

884: to radically different structural properties of high-redshift

885: early-type galaxies is no longer necessary.

886: 

887: Other systematic uncertainties cannot explain the observed evolution

888: either.  In our size measurements, systematic effects have been taken

889: into account (see Secs.~\ref{distant:profile} and \ref{sim}).  We are

890: confident, for example, that we would detect low-surface brightness

891: envelopes around distant galaxies.  Furthermore, we know that only a

892: minority of morphologically selected $z\sim 1$ early-type galaxies

893: ($\sim$10\%) show signs of nuclear activity

894: \citep[e.g.,][]{rodighiero07, vanderwel07a} such that it is unlikely

895: that central point sources affect our size measurements.  This is also

896: clear from the fact that the residuals of our $R^{1/4}$ profile fits

897: generally do not show central point sources and that none of the deep

898: spectra used to measure dispersions show evidence for nuclear

899: activity.  Furthermore, the good correspondence between the rest-frame

900: wavelength of the imaging data sets used at different redshifts assures

901: us that morphological $K$-corrections do not play a significant role.

902: 

903: Despite the broad consistency between our results and those previously

904: published, the agreement is not perfect.  There is a marginal

905: inconsistency at the $1.5~\sigma$ level between the size-evolution

906: measurement from kinematic data and the size-evolution measurement

907: from photometric data shown in Fig.~\ref{lit_z}.  This could point to

908: the presence of some systematic effects within the $z>1.5$ results.

909: Alternatively, the different studies sample galaxies with a wide range

910: in masses, and therefore mass-dependent size evolution could lead to

911: apparent discrepancies among the samples.  This is explored in the

912: following section.

913: 

914: Our robust results strengthen the results from previous studies.  We

915: conclude that early-type galaxies at $z=1$ are $\sim 2$ times smaller

916: than local early types with the same mass, and that at $z=2-2.5$ this

917: size difference is likely increased to a factor of $\sim$4, as

918: previously observed by \citet{zirm07}, \citet{toft07},

919: \citet{vandokkum08b}, and \citet{buitrago08}.

920: 

921: 

922: \subsection{Comparison with Model Predictions}\label{dis:2}

923: 

924: The fact that we see considerable evolution in galaxy size with

925: redshift is not surprising from a theoretical perspective.  Most

926: semianalytic models of galaxy formation in a $\Lambda$CDM universe

927: predict substantial size evolution over the past several billion

928: years.  A comparison between the observed and model-predicted amount

929: of size evolution will help to identify the mechanism(s) that are

930: responsible.  In Fig.~\ref{M_dens} we directly compare the observed

931: evolution in surface density with the predictions from the

932: semianalytic work by \citet{khochfar06b}.  For galaxies with a given

933: mass the model significantly under-predicts the evolution in size and

934: density, except, perhaps, for the most massive galaxies.  In our data

935: set we see no indication that the magnitude of size and density

936: evolution increases with galaxy mass, as predicted by the models.  In

937: fact, the most massive galaxies in our sample are precisely the only

938: ones that are not different from local massive galaxies.  Note,

939: however, that statistically speaking the evidence for mass-dependent

940: evolution is weak (see \S~\ref{results:mrevol}).  Moreover, the most

941: massive galaxies in our distant sample are a special subset, brightest

942: cluster galaxies.  Such galaxies have been shown to have properties

943: that deviate from those of other massive galaxies \citep[see,

944: e.g.,][]{vonderlinden07, bernardi07b}.

945: 

946: By including the $z=1.5-2.5$ photometric samples discussed in

947: \S~\ref{dis:1} we can place further constraints on the models.  In

948: Fig.~\ref{lit_M} we compare the observed size evolution of the

949: available samples, normalized to $z=1$, with the model predictions

950: from \citet{khochfar06b}.  Representing the model predictions by a

951: single line is justified by the fact that the predicted evolution of

952: $\log(\reff)$ with $\log(1+z)$ is very close to linear.  Again, the

953: observed size evolution is stronger than that predicted by the model.

954: 

955: It is interesting to note that, in qualitative agreement with the

956: model prediction, we see a hint that size evolution depends on mass in

957: the compilation presented in Fig.~\ref{lit_M}.  The samples containing

958: on average the lowest-mass galaxies display marginally less evolution.

959: It has to be kept in mind, however, that small sample sizes and

960: systematic effects are more important for determining second-order

961: effects such as mass dependence (see \S~\ref{results:mrevol}).  A

962: clue that systematic uncertainties may play a role is the remaining

963: difference between the kinematic and photometric samples.

964: Alternatively, it may signify non-linear evolution of $\log(\reff)$

965: with $\log(1+z)$.

966: 

967: \begin{figure}[t]

968: \epsscale{1.2}

969: \plotone{lit_z.eps}

970: \caption{Size evolution with redshift as derived in this paper with

971:   dynamically determined masses (\textit{large filled circles})

972:   compared with previous results based on photometric masses

973:   (\textit{small filled circles}).  The solid line connects our

974:   samples at $z\sim 0.06$ and $z\sim 1$, the dashed line is a linear

975:   least-squares fit to the small filled data points.  The open circles

976:   are samples of cluster galaxies with photometrically measured masses

977:   and serve as an illustration that size evolution shows a continuous

978:   trend between $z=2.5$ and the present.  The broad agreement in size

979:   evolution as derived from galaxies with dynamically and

980:   photometrically determines masses reinforces the conclusions of

981:   previous, photometric studies whose results were potentially

982:   mitigated by considerable systematic effects that do not affect our

983:   analysis. }

984: \label{lit_z}

985: \end{figure}

986: 

987: \begin{figure}[t]

988: \epsscale{1.2}

989: \plotone{lit_M.eps}

990: \caption{Size evolution per unit redshift vs. mean galaxy mass of our

991:   sample (\textit{large circle}) and samples taken from the literature

992:   (small squares; see Fig.~\ref{lit_z} for references).  Samples

993:   consisting of high-mass galaxies show somewhat stronger size

994:   evolution than samples consisting of low-mass galaxies, which is

995:   qualitatively, but not quantitatively, consistent with the

996:   predictions from the semianalytic model from \citet{khochfar06b}

997:   (\textit{solid line}).  This conclusion should be considered highly

998:   tentative, however, as this interpretation is hampered by systematic

999:   uncertainties and small sample sizes. }

1000: \label{lit_M}

1001: \end{figure}

1002: 

1003: \subsection{Size Evolution of Individual Galaxies}\label{dis:3}

1004: 

1005: It appears that the observed size evolution of a factor of $\sim$2

1006: between $z=1$ and the present for early-type galaxies with masses

1007: $\sim10^{11}~\msol$ is similar to the predicted evolution for

1008: early-type galaxies that are an order of magnitude more massive (see

1009: Figs.~\ref{M_dens} and \ref{lit_M}).  This suggests that the mechanism

1010: responsible for increasing the average size of early-type galaxies

1011: with time may be well understood, but that it is not implemented

1012: correctly in the current semianalytic model from \citet{khochfar06b}.

1013: The process of size evolution may occur at different times and under

1014: different circumstances than is now assumed.  This may be related to

1015: the late assembly of very massive galaxies in models of this kind

1016: \citep[see also, e.g.,][]{delucia06}, a prediction that is challenged

1017: by various observations \citep[e.g.,][]{cimatti06, scarlata07,

1018:   cool08}.

1019: 

1020: It is beyond the scope of this paper to fully discuss these possible

1021: discrepancies.  Instead we will explore the question whether the

1022: proposed physical processes responsible for size evolution are

1023: consistent with the observed trends.  In the semi-analytic models it

1024: is assumed (and this is confirmed by numerical simulations) that

1025: mergers drive size evolution.  The gas content of merging galaxies

1026: largely determines the relative size of the merger remnant compared to

1027: its ancestors.  Because gas fractions were higher in the past,

1028: galaxies that form early will be smaller than galaxies that form late.

1029: In the framework of cosmological simulations this means that galaxies

1030: at high redshift will be smaller because they were formed through

1031: gas-rich mergers and that those merger remnants can grow over time

1032: through subsequent mergers with other galaxies that are progressively

1033: more devoid of gas.

1034: 

1035: The question is whether the observed size evolution is dominated by

1036: size evolution of individual galaxies or simply by the addition of

1037: larger galaxies over time.  At $z=1$ only about 30--50\% of the

1038: present-day early-type galaxy evolution had formed \citep{bell04b,

1039:   brown07, scarlata07, faber07}.  If we assume that these galaxies

1040: will make up the 30--50\% most dense early-type galaxies in the

1041: present-day universe, then the scatter in the local $\mvir$-$\reff$

1042: relation implies that the sample-averaged size increases by a factor

1043: of 1.3--1.4 between $z=1$ and the present.  Such evolution is thus

1044: expected in the absence of size evolution of individual galaxies and

1045: this is less than the observed evolution of a factor of 2.  To explain

1046: the observed evolution by growth of the early-type galaxy population

1047: without changes in the sizes of individual galaxies, the number

1048: density of early-type galaxies is required to increase by an order of

1049: magnitude between $z=1$ and the present.  Such strong evolution is

1050: clearly ruled out by the above-mentioned determinations of the number

1051: density of red galaxies at $z\sim 1$.

1052: 

1053: Similarly, at $z=2$ only $\sim$10\% of the galaxies with masses

1054: $\gtrsim~10^{11}~\msol$ had been assembled \citep{kriek08b}; if those

1055: galaxies, evolve into the 10\% most dense present-day early-type

1056: galaxies then an increase in average size by a factor of $\sim$2 can

1057: be accounted for, less than the observed amount of evolution.  These

1058: arguments are in agreement with the conclusions from

1059: \citet{cimatti08}, who show that local galaxies with the same sizes

1060: and masses as galaxies in the $z=1-2$ samples are so rare in the local

1061: universe that it can be confidently ruled out that their structure

1062: remains unchanged up until the present day.  Note, however, that these

1063: arguments may be affected by the aforementioned biases in the SDSS

1064: (\S~\ref{nearby:sigrad}).

1065: 

1066: %An independent indication that the average size of early-type galaxies

1067: %increases over time due to the formation larger galaxies at later

1068: %times would be a decrease with redshift in scatter in the

1069: %$\mvir$-$\reff$ relation.  In our sample of galaxies we see tentative

1070: %evidence that for our sample of $z\sim 1$ galaxies with dynamically

1071: %measured masses the scatter is indeed marginally smaller ($\delta

1072: %\reff = 0.117\pm 0.013$ dex) than for the local sample ($\delta \reff

1073: %= 0.146\pm 0.002$ dex).

1074: 

1075: We conclude that size evolution due to the addition of larger galaxies

1076: over time contributes at most half of the observed evolution in the

1077: $\mvir$-$\reff$ relation.  The remainder must be due to size evolution

1078: of individual galaxies.  Numerical simulations have demonstrated that

1079: when early-type galaxies accrete neighbors without significant

1080: dissipational processes $\sige$ does not change by much and that, to

1081: first order, $\reff$ increases linearly with mass.  This does not

1082: depend strongly on the mass of the accreted object, i.e., the mass

1083: ratio of the merger \citep{boylan05, robertson06, boylan06}.

1084: 

1085: Simulations in a cosmological context show that an increase in size by

1086: a factor of 2 between $z\sim 1$ and the present is certainly possible

1087: \citep{naab07}.  The strong observed size evolution thus argues in

1088: favor of a scenario in which significant mass from low-mass companions

1089: is accreted onto existing early-type galaxies over the past $\sim$7

1090: Gyr, which also explains the broad tidal features that are frequently

1091: observed around early-type galaxies \citep{vandokkum05}.  As shown by

1092: \citet{feldmann08} such features are not necessarily, and are even

1093: quite unlikely to be, the result of major merger events and are most

1094: likely due to the accretion of low-mass, gas-poor satellites.

1095: 

1096: We note that the size evolution of individual galaxies and the

1097: evolution of the sample average are inseparable because galaxies

1098: evolve in mass as well as in size.  Nonetheless, it is important to

1099: distinguish this complex scenario from the simple picture in which

1100: early-type galaxies that form at different redshifts have different

1101: sizes but do not structurally evolve at later times.  The strong

1102: observed size evolution clearly rules out the latter, indicating that

1103: the build-up of the early-type galaxy population is a complex and

1104: ongoing process.

1105: 

1106: Finally, it is remarkable that the change in the sizes of early-type

1107: galaxies is consistent with and differs by less than 15\% from the

1108: change in the scale factor of the universe, $1+z$.  Within the

1109: standard cold dark matter scenario this is likely a coincidence since

1110: dissipational, strongly non-linear processes that are decoupled from

1111: cosmic expansion dominate at the kiloparsec scale of forming galaxies.

1112: Nonetheless, we cannot exclude the possibility that there is an

1113: underlying, fundamental reason that galaxies are scale-invariant with

1114: respect to a co-moving coordinate system.  In an alternative

1115: description of dark matter, i.e., Bose-Einstein condensed, ultra-light

1116: particles with a $\sim$10 kpc-sized wave function \citep[fuzzy dark

1117: matter or FDM,][]{sin94,hu00}, sizes of halos and their occupying

1118: galaxies possibly follow the cosmic expansion rate \citep{lee08}.

1119: 

1120: 

1121: \section{SUMMARY}\label{sum}

1122: 

1123: In \S~\ref{nearby} we construct a large sample of nearby

1124: ($0.04<z<0.08$) early-type galaxies extracted from the SDSS (DR6).  We

1125: use the pipeline velocity dispersion measurements and obtain our own

1126: size measurements in order to construct the local dynamical mass-size

1127: relation (\S~\ref{nearby:mr}).  In addition, we construct a sample of

1128: 50 morphologically selected early-type galaxies in the redshift range

1129: $0.8<z<1.2$ with measured velocity dispersions (\S~\ref{distant}).

1130: Sizes are determined from ACS imaging in the same manner as for the

1131: galaxies in the nearby sample, and systematic effects are quantified

1132: through simulations (\S~\ref{sim}).  The distant sample contains

1133: galaxies in the mass range $3\times 10^{10}~\msol < M \lesssim

1134: 10^{12}~\msol$, with a typical mass of $2\times 10^{11}~\msol$.

1135: 

1136: The main result is that the $\sige$-$\reff$ distributions of the

1137: nearby and distant samples are significantly different, even after we

1138: correct for the incompleteness of the distant sample at low masses

1139: (\S~\ref{results}).  The implied size evolution is $\reff \propto

1140: (1+z)^{-0.98 \pm 0.11}$, or a factor of $1.97 \pm 0.15$ between $z=1$

1141: and the present.  Similarly, the projected surface densities of the

1142: distant early-type galaxies are a factor of $\sim$4 higher than those

1143: of their local counterparts.  The stellar populations of early-type

1144: galaxies that already existed at $z=1$ may, for the most part, be

1145: passively evolving over the past 7--8 Gyr, however, their structural

1146: properties undergo substantial changes over that period.

1147: 

1148: Our results are in broad agreement (see \S~\ref{dis:1}) with

1149: previously published size-evolution measurements that are based on

1150: samples without dynamical mass measurements and, in some cases,

1151: without spectroscopic redshifts, high-resolution \textit{HST} imaging,

1152: and/or consistently determined sizes.  We therefore conclude that

1153: systematic effects, most notably those in the mass estimates, which

1154: potentially could have hampered previous studies are small relative to

1155: the observed amount of evolution.

1156: 

1157: The observed size evolution is in qualitative agreement with

1158: predictions from recent semianalytic models.  However, the predicted

1159: evolution is much slower than the observed evolution.  The observed

1160: size evolution of early-type galaxies can be understood within the

1161: context of the cold dark matter scenario in which galaxies that form

1162: late have larger sizes than galaxies that form early, due to lower gas

1163: fractions at late times, and the growth of individual galaxies through

1164: the mostly dissipationless accretion of satellites at later

1165: evolutionary stages.

1166: 

1167: \acknowledgements{We thank the referee for his helpful

1168:   review. A.~v.~d.~W. would like to thank Jenny Graves for sharing her

1169:   SDSS catalog of early-type galaxies, Eric Bell, Hans-Walter Rix, and

1170:   Pieter van Dokkum for interesting discussions, and Sadegh Khochfar

1171:   for providing his model predictions.  Support from NASA grant

1172:   NAG5-7697 is also gratefully acknowledged.  The authors wish to

1173:   recognize and acknowledge the very significant cultural role and

1174:   reverence that the summit of Mauna Kea has always had within the

1175:   indigenous Hawaiian community.  We are most fortunate to have the

1176:   opportunity to conduct observations from this mountain.  Funding for

1177:   the SDSS and SDSS-II has been provided by the Alfred P. Sloan

1178:   Foundation, the Participating Institutions, the National Science

1179:   Foundation, the U.S. Department of Energy, the National Aeronautics

1180:   and Space Administration, the Japanese Monbukagakusho, the Max

1181:   Planck Society, and the Higher Education Funding Council for

1182:   England. The SDSS Web Site is http://www.sdss.org/.  The SDSS is

1183:   managed by the Astrophysical Research Consortium for the

1184:   Participating Institutions. The Participating Institutions are the

1185:   American Museum of Natural History, Astrophysical Institute Potsdam,

1186:   University of Basel, University of Cambridge, Case Western Reserve

1187:   University, University of Chicago, Drexel University, Fermilab, the

1188:   Institute for Advanced Study, the Japan Participation Group, Johns

1189:   Hopkins University, the Joint Institute for Nuclear Astrophysics,

1190:   the Kavli Institute for Particle Astrophysics and Cosmology, the

1191:   Korean Scientist Group, the Chinese Academy of Sciences (LAMOST),

1192:   Los Alamos National Laboratory, the Max-Planck-Institute for

1193:   Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA),

1194:   New Mexico State University, Ohio State University, University of

1195:   Pittsburgh, University of Portsmouth, Princeton University, the

1196:   United States Naval Observatory, and the University of Washington. }

1197: 

1198: %\bibliographystyle{apj} 

1199: %\bibliography{mybib}

1200: \begin{thebibliography}{}

1201: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

1202: \bibitem[{{Adelman-McCarthy} {et~al.}(2008){Adelman-McCarthy}, {Ag{\"u}eros},

1203:   {Allam}, {Allende Prieto}, {Anderson}, {Anderson}, {Annis}, {Bahcall},

1204:   {Bailer-Jones}, {Baldry}, {Barentine}, {Bassett}, {Becker}, {Beers}, {Bell},

1205:   {Berlind}, {Bernardi}, {Blanton}, {Bochanski}, {Boroski}, {Brinchmann},

1206:   {Brinkmann}, {Brunner}, {Budav{\'a}ri}, {Carliles}, {Carr}, {Castander},

1207:   {Cinabro}, {Cool}, {Covey}, {Csabai}, {Cunha}, {Davenport}, {Dilday}, {Doi},

1208:   {Eisenstein}, {Evans}, {Fan}, {Finkbeiner}, {Friedman}, {Frieman},

1209:   {Fukugita}, {G{\"a}nsicke}, {Gates}, {Gillespie}, {Glazebrook}, {Gray},

1210:   {Grebel}, {Gunn}, {Gurbani}, {Hall}, {Harding}, {Harvanek}, {Hawley},

1211:   {Hayes}, {Heckman}, {Hendry}, {Hindsley}, {Hirata}, {Hogan}, {Hogg}, {Hyde},

1212:   {Ichikawa}, {Ivezi{\'c}}, {Jester}, {Johnson}, {Jorgensen}, {Juri{\'c}},

1213:   {Kent}, {Kessler}, {Kleinman}, {Knapp}, {Kron}, {Krzesinski}, {Kuropatkin},

1214:   {Lamb}, {Lampeitl}, {Lebedeva}, {Lee}, {Leger}, {L{\'e}pine}, {Lima}, {Lin},

1215:   {Long}, {Loomis}, {Loveday}, {Lupton}, {Malanushenko}, {Malanushenko},

1216:   {Mandelbaum}, {Margon}, {Marriner}, {Mart{\'{\i}}nez-Delgado}, {Matsubara},

1217:   {McGehee}, {McKay}, {Meiksin}, {Morrison}, {Munn}, {Nakajima}, {Neilsen},

1218:   {Newberg}, {Nichol}, {Nicinski}, {Nieto-Santisteban}, {Nitta}, {Okamura},

1219:   {Owen}, {Oyaizu}, {Padmanabhan}, {Pan}, {Park}, {Peoples}, {Pier}, {Pope},

1220:   {Purger}, {Raddick}, {Re Fiorentin}, {Richards}, {Richmond}, {Riess}, {Rix},

1221:   {Rockosi}, {Sako}, {Schlegel}, {Schneider}, {Schreiber}, {Schwope}, {Seljak},

1222:   {Sesar}, {Sheldon}, {Shimasaku}, {Sivarani}, {Smith}, {Snedden}, {Steinmetz},

1223:   {Strauss}, {SubbaRao}, {Suto}, {Szalay}, {Szapudi}, {Szkody}, {Tegmark},

1224:   {Thakar}, {Tremonti}, {Tucker}, {Uomoto}, {Vanden Berk}, {Vandenberg},

1225:   {Vidrih}, {Vogeley}, {Voges}, {Vogt}, {Wadadekar}, {Weinberg}, {West},

1226:   {White}, {Wilhite}, {Yanny}, {Yocum}, {York}, {Zehavi}, \&

1227:   {Zucker}}]{adelman08}

1228: {Adelman-McCarthy}, J.~K. et al. 2008, \apjs, 175, 297

1229: 

1230: \bibitem[{{Bell} {et~al.}(2004){Bell}, {Wolf}, {Meisenheimer}, {Rix}, {Borch},

1231:   {Dye}, {Kleinheinrich}, {Wisotzki}, \& {McIntosh}}]{bell04b}

1232: {Bell}, E.~F. et al. 2004, \apj, 608, 752

1233: 

1234: \bibitem[{{Bernardi}(2007)}]{bernardi07a}

1235: {Bernardi}, M. 2007, \aj, 133, 1954

1236: 

1237: \bibitem[{{Bernardi} {et~al.}(2007){Bernardi}, {Hyde}, {Sheth}, {Miller}, \&

1238:   {Nichol}}]{bernardi07b}

1239: {Bernardi}, M., {Hyde}, J.~B., {Sheth}, R.~K., {Miller}, C.~J., \& {Nichol},

1240:   R.~C. 2007, \aj, 133, 1741

1241: 

1242: %\bibitem[{{Blakeslee} {et~al.}(2003){Blakeslee}, {Franx}, {Postman}, {Rosati},

1243: %  {Holden}, {Illingworth}, {Ford}, {Cross}, {Gronwall}, {Ben{\'{\i}}tez},

1244: %  {Bouwens}, {Broadhurst}, {Clampin}, {Demarco}, {Golimowski}, {Hartig},

1245: %  {Infante}, {Martel}, {Miley}, {Menanteau}, {Meurer}, {Sirianni}, \&

1246: %  {White}}]{blakeslee03}

1247: %{Blakeslee}, J.~P. et al. 2003, \apjl, 596, L143

1248: 

1249: \bibitem[{{Blakeslee} {et~al.}(2006){Blakeslee}, {Holden}, {Franx}, {Rosati},

1250:   {Bouwens}, {Demarco}, {Ford}, {Homeier}, {Illingworth}, {Jee}, {Mei},

1251:   {Menanteau}, {Meurer}, {Postman}, \& {Tran}}]{blakeslee06}

1252: {Blakeslee}, J.~P. et al. 2006, \apj, 644, 30

1253: 

1254: \bibitem[{{Boylan-Kolchin} {et~al.}(2005){Boylan-Kolchin}, {Ma}, \&

1255:   {Quataert}}]{boylan05}

1256: {Boylan-Kolchin}, M., {Ma}, C.-P., \& {Quataert}, E. 2005, \mnras, 362, 184

1257: 

1258: \bibitem[{{Boylan-Kolchin} {et~al.}(2006){Boylan-Kolchin}, {Ma}, \&

1259:   {Quataert}}]{boylan06}

1260: {Boylan-Kolchin}, M., {Ma}, C.-P., \& {Quataert}, E. 2006, \mnras, 369, 1081

1261: 

1262: \bibitem[{{Brown} {et~al.}(2007){Brown}, {Dey}, {Jannuzi}, {Brand}, {Benson},

1263:   {Brodwin}, {Croton}, \& {Eisenhardt}}]{brown07}

1264: {Brown}, M.~J.~I. et al. 2007, \apj, 654, 858

1265: 

1266: \bibitem[{{Bruzual} \& {Charlot}(2003)}]{bruzual03}

1267: {Bruzual}, G., \& {Charlot}, S. 2003, \mnras, 344, 1000

1268: 

1269: \bibitem[{{Buitrago} {et~al.}(2008){Buitrago}, {Trujillo}, {Conselice}, 

1270:   {Bouwens}, {Dickinson}, \& {Yan}}]{buitrago08}

1271: {Buitrago}, F. et al. 2008, \apj, submitted, arXiv:0807.4141

1272: 

1273: \bibitem[{{Cappellari} {et~al.}(2006){Cappellari}, {Bacon}, {Bureau}, {Damen},

1274:   {Davies}, {de Zeeuw}, {Emsellem}, {Falc{\'o}n-Barroso}, {Krajnovi{\'c}},

1275:   {Kuntschner}, {McDermid}, {Peletier}, {Sarzi}, {van den Bosch}, \& {van de

1276:   Ven}}]{cappellari06}

1277: {Cappellari}, M. et al. 2006, \mnras, 366, 1126

1278: 

1279: \bibitem[{{Chabrier}(2003)}]{chabrier03}

1280: {Chabrier}, G. 2003, \pasp, 115, 763

1281: 

1282: \bibitem[{{Cimatti} {et~al.}(2008){Cimatti}, {Cassata}, {Pozzetti}, {Kurk},

1283:   {Mignoli}, {Renzini}, {Daddi}, {Bolzonella}, {Brusa}, {Rodighiero},

1284:   {Dickinson}, {Franceschini}, {Zamorani}, {Berta}, {Rosati}, \&

1285:   {Halliday}}]{cimatti08}

1286: {Cimatti}, A. et al. 2008, \aap, 482, 21

1287: 

1288: \bibitem[{{Cimatti} {et~al.}(2006){Cimatti}, {Daddi}, \& {Renzini}}]{cimatti06}

1289: {Cimatti}, A., {Daddi}, E., \& {Renzini}, A. 2006, \aap, 453, L29

1290: 

1291: \bibitem[{{Cool} {et~al.}(2008){Cool}, {Eisenstein}, {Fan}, {Fukugita},

1292:   {Jiang}, {Maraston}, {Meiksin}, {Schneider}, \& {Wake}}]{cool08}

1293: {Cool}, R.~J. et al. 2008, \apj, in press, arXiv:0804.4516

1294: 

1295: \bibitem[{{Daddi} {et~al.}(2005){Daddi}, {Renzini}, {Pirzkal}, {Cimatti},

1296:   {Malhotra}, {Stiavelli}, {Xu}, {Pasquali}, {Rhoads}, {Brusa}, {di Serego

1297:   Alighieri}, {Ferguson}, {Koekemoer}, {Moustakas}, {Panagia}, \&

1298:   {Windhorst}}]{daddi05}

1299: {Daddi}, E. et al. 2005, \apj, 626, 680

1300: 

1301: \bibitem[{{Dav{\'e}}(2008)}]{dave08}

1302: {Dav{\'e}}, R. 2008, \mnras, 152

1303: 

1304: \bibitem[{{De Lucia} {et~al.}(2006){De Lucia}, {Springel}, {White}, {Croton},

1305:   \& {Kauffmann}}]{delucia06}

1306: {De Lucia}, G., {Springel}, V., {White}, S.~D.~M., {Croton}, D., \&

1307:   {Kauffmann}, G. 2006, \mnras, 366, 499

1308: 

1309: \bibitem[{{de Vaucouleurs}(1948)}]{devaucouleurs48}

1310: {de Vaucouleurs}, G. 1948, Annales d'Astrophysique, 11, 247

1311: 

1312: \bibitem[{{di Serego Alighieri} {et~al.}(2005){di Serego Alighieri}, {Vernet},

1313:   {Cimatti}, {Lanzoni}, {Cassata}, {Ciotti}, {Daddi}, {Mignoli}, {Pignatelli},

1314:   {Pozzetti}, {Renzini}, {Rettura}, \& {Zamorani}}]{diserego05}

1315: {di Serego Alighieri}, S. et al. 2005, \aap,

1316:   442, 125

1317: 

1318: \bibitem[{{Djorgovski} \& {Davis}(1987)}]{djorgovskidavis87}

1319: {Djorgovski}, S., \& {Davis}, M. 1987, \apj, 313, 59

1320: 

1321: \bibitem[{{Dressler} {et~al.}(1987){Dressler}, {Lynden-Bell}, {Burstein},

1322:   {Davies}, {Faber}, {Terlevich}, \& {Wegner}}]{dressler87}

1323: {Dressler}, A., {Lynden-Bell}, D., {Burstein}, D., {Davies}, R.~L., {Faber},

1324:   S.~M., {Terlevich}, R., \& {Wegner}, G. 1987, \apj, 313, 42

1325: 

1326: \bibitem[{{Faber} {et~al.}(2007){Faber}, {Willmer}, {Wolf}, {Koo}, {Weiner},

1327:   {Newman}, {Im}, {Coil}, {Conroy}, {Cooper}, {Davis}, {Finkbeiner}, {Gerke},

1328:   {Gebhardt}, {Groth}, {Guhathakurta}, {Harker}, {Kaiser}, {Kassin},

1329:   {Kleinheinrich}, {Konidaris}, {Kron}, {Lin}, {Luppino}, {Madgwick},

1330:   {Meisenheimer}, {Noeske}, {Phillips}, {Sarajedini}, {Schiavon}, {Simard},

1331:   {Szalay}, {Vogt}, \& {Yan}}]{faber07}

1332: {Faber}, S.~M. et al. 2007, \apj, 665, 265

1333: 

1334: \bibitem[{{Fardal} {et~al.}(2007){Fardal}, {Katz}, {Weinberg}, \&

1335:   {Dav{\'e}}}]{fardal07}

1336: {Fardal}, M.~A., {Katz}, N., {Weinberg}, D.~H., \& {Dav{\'e}}, R. 2007, \mnras,

1337:   379, 985

1338: 

1339: \bibitem[{{Feldmann} {et~al.}(2008){Feldmann}, {Mayer}, \&

1340:   {Carollo}}]{feldmann08}

1341: {Feldmann}, R., {Mayer}, L., \& {Carollo}, C.~M. 2008, \apj, submitted, arXiv:0801.4764

1342: 

1343: \bibitem[{{Ford} {et~al.}(1998){Ford}, {Bartko}, {Bely}, {Broadhurst},

1344:   {Burrows}, {Cheng}, {Clampin}, {Crocker}, {Feldman}, {Golimowski}, {Hartig},

1345:   {Illingworth}, {Kimble}, {Lesser}, {Miley}, {Neff}, {Postman}, {Sparks},

1346:   {Tsvetanov}, {White}, {Sullivan}, {Krebs}, {Leviton}, {La Jeunesse},

1347:   {Burmester}, {Fike}, {Johnson}, {Slusher}, {Volmer}, \& {Woodruff}}]{ford98}

1348: {Ford}, H.~C. et al. 1998, in Presented at

1349:   the Society of Photo-Optical Instrumentation Engineers (SPIE) Conference,

1350:   Vol. 3356, Proc. SPIE Vol. 3356, p. 234-248, Space Telescopes and Instruments

1351:   V, Pierre Y. Bely; James B. Breckinridge; Eds., ed. P.~Y. {Bely} \& J.~B.

1352:   {Breckinridge}, 234--248

1353: 

1354: \bibitem[{{Franx} {et~al.}(2008){Franx}}]{franx08}

1355: {Franx}, M. et al. 2008, \apj, submitted

1356: 

1357: \bibitem[{{Giavalisco} {et~al.}(2004){Giavalisco}, {Ferguson}, {Koekemoer},

1358:   {Dickinson}, {Alexander}, {Bauer}, {Bergeron}, {Biagetti}, {Brandt},

1359:   {Casertano}, {Cesarsky}, {Chatzichristou}, {Conselice}, {Cristiani}, {Da

1360:   Costa}, {Dahlen}, {de Mello}, {Eisenhardt}, {Erben}, {Fall}, {Fassnacht},

1361:   {Fosbury}, {Fruchter}, {Gardner}, {Grogin}, {Hook}, {Hornschemeier}, {Idzi},

1362:   {Jogee}, {Kretchmer}, {Laidler}, {Lee}, {Livio}, {Lucas}, {Madau},

1363:   {Mobasher}, {Moustakas}, {Nonino}, {Padovani}, {Papovich}, {Park},

1364:   {Ravindranath}, {Renzini}, {Richardson}, {Riess}, {Rosati}, {Schirmer},

1365:   {Schreier}, {Somerville}, {Spinrad}, {Stern}, {Stiavelli}, {Strolger},

1366:   {Urry}, {Vandame}, {Williams}, \& {Wolf}}]{giavalisco04}

1367: {Giavalisco}, M. et al. 2004, \apjl, 600, L93

1368: 

1369: \bibitem[{{Graves} {et~al.}(2007){Graves}, {Faber}, {Schiavon}, \&

1370:   {Yan}}]{graves07}

1371: {Graves}, G.~J., {Faber}, S.~M., {Schiavon}, R.~P., \& {Yan}, R. 2007, \apj,

1372:   671, 243

1373: 

1374: \bibitem[{{Holden} {et~al.}(2007){Holden}, {Illingworth}, {Franx},

1375:   {Blakeslee}, {Postman}, {Kelson}, {van der Wel}, {Demarco}, {Magee},

1376:   {Tran}, {Zirm}, {Ford}, {Rosati}, \& {Homeier}}]{holden07} 

1377: {Holden}, B.~P. et al. 2007, \apj, 670, 190

1378: 

1379: \bibitem[{{Holden} {et~al.}(2006){Holden}, {Franx}, {Illingworth}, {Postman},

1380:   {Blakeslee}, {Homeier}, {Demarco}, {Ford}, {Rosati}, {Kelson}, \&

1381:   {Tran}}]{holden06}

1382: {Holden}, B.~P. et al. 2006, \apjl, 642, L123

1383: 

1384: \bibitem[{{Holden} {et~al.}(2005){Holden}, {van der Wel}, {Franx},

1385:   {Illingworth}, {Blakeslee}, {van Dokkum}, {Ford}, {Magee}, {Postman}, {Rix},

1386:   \& {Rosati}}]{holden05}

1387: {Holden}, B.~P. et al. 2005, \apjl, 620, L83

1388: 

1389: \bibitem[{{Hoversten} \& {Glazebrook}(2008)}]{hoversten08}

1390: {Hoversten}, E.~A., \& {Glazebrook}, K. 2008, \apj, 675, 163

1391: 

1392: \bibitem[{{Hu} {et~al.}(2000){Hu}, {Barkana}, \& {Gruzinov}}]{hu00}

1393: {Hu}, W., {Barkana}, R., \& {Gruzinov}, A. 2000, Physical Review Letters, 85,

1394:   1158

1395: 

1396: \bibitem[{{J\o rgensen} {et~al.}(1995){J\o rgensen}, {Franx}, \&

1397:   {Kjaergaard}}]{jorgensen95spec}

1398: {J\o rgensen}, I., {Franx}, M., \& {Kjaergaard}, P. 1995, \mnras, 276, 1341

1399: 

1400: \bibitem[{{J{\o}rgensen} {et~al.}(2005){J{\o}rgensen}, {Bergmann}, {Davies},

1401:   {Barr}, {Takamiya}, \& {Crampton}}]{jorgensen05}

1402: {J{\o}rgensen}, I., {Bergmann}, M., {Davies}, R., {Barr}, J., {Takamiya}, M.,

1403:   \& {Crampton}, D. 2005, \aj, 129, 1249

1404: 

1405: \bibitem[{{Khochfar} \& {Silk}(2006{\natexlab{a}})}]{khochfar06b}

1406: {Khochfar}, S., \& {Silk}, J. 2006{\natexlab{a}}, \apjl, 648, L21

1407: 

1408: \bibitem[{{Khochfar} \& {Silk}(2006{\natexlab{b}})}]{khochfar06a}

1409: ---. 2006{\natexlab{b}}, \mnras, 370, 902

1410: 

1411: \bibitem[{{Kriek} {et~al.}(2008){Kriek}, {van der Wel}, {van Dokkum}, {Franx},

1412:   \& {Illingworth}}]{kriek08b}

1413: {Kriek}, M., {van der Wel}, A., {van Dokkum}, P.~G., {Franx}, M., \&

1414:   {Illingworth}, G.~D. 2008, \apj, in press, arXiv:0804.4175

1415: 

1416: \bibitem[{{Kriek} {et~al.}(2006){Kriek}, {van Dokkum}, {Franx}, {Quadri},

1417:   {Gawiser}, {Herrera}, {Illingworth}, {Labb{\'e}}, {Lira}, {Marchesini},

1418:   {Rix}, {Rudnick}, {Taylor}, {Toft}, {Urry}, \& {Wuyts}}]{kriek06b}

1419: {Kriek}, M. et al. 2006, \apjl, 649, L71

1420: 

1421: \bibitem[{{Krist}(1995)}]{krist95}

1422: {Krist}, J. 1995, in ASP Conf. Ser. 77: Astronomical Data Analysis Software and

1423:   Systems IV, 349--+

1424: 

1425: \bibitem[{{Kroupa}(2001)}]{kroupa01}

1426: {Kroupa}, P. 2001, \mnras, 322, 231

1427: 

1428: \bibitem[{{Larson}(2005)}]{larson05}

1429: {Larson}, R.~B. 2005, \mnras, 359, 211

1430: 

1431: \bibitem[{{Loeb} \& {Peebles}(2003)}]{loeb03}

1432: {Loeb}, A., \& {Peebles}, P.~J.~E. 2003, \apj, 589, 29

1433: 

1434: \bibitem[{{Lee}(2008)}]{lee08}

1435: {Lee}, J.-W. 2008, arXiv:0805.2877

1436: 

1437: \bibitem[{{Longhetti} {et~al.}(2007){Longhetti}, {Saracco}, {Severgnini},

1438:   {Della Ceca}, {Mannucci}, {Bender}, {Drory}, {Feulner}, \&

1439:   {Hopp}}]{longhetti07}

1440: {Longhetti}, M. et al. 2007, \mnras,

1441:   374, 614

1442: 

1443: \bibitem[{{Maraston}(2005)}]{maraston05}

1444: {Maraston}, C. 2005, \mnras, 362, 799

1445: 

1446: \bibitem[{{Mei} {et~al.}(2007){Mei}, {Blakeslee}, {C{\^o}t{\'e}}, {Tonry},

1447:   {West}, {Ferrarese}, {Jord{\'a}n}, {Peng}, {Anthony}, \& {Merritt}}]{mei07}

1448: {Mei}, S. et al. 2007, \apj, 655, 144

1449: 

1450: %\bibitem[{{Mei} {et~al.}(2006){Mei}, {Holden}, {Blakeslee}, {Rosati},

1451: %  {Postman}, {Jee}, {Rettura}, {Sirianni}, {Demarco}, {Ford}, {Franx},

1452: %  {Homeier}, \& {Illingworth}}]{mei06}

1453: %{Mei}, S. et al. 2006, \apj, 644, 759

1454: 

1455: \bibitem[{{Naab} {et~al.}(2007){Naab}, {Johansson}, {Ostriker}, \&

1456:   {Efstathiou}}]{naab07}

1457: {Naab}, T., {Johansson}, P.~H., {Ostriker}, J.~P., \& {Efstathiou}, G. 2007,

1458:   \apj, 658, 710

1459: 

1460: \bibitem[{{Paturel} {et~al.}(2003){Paturel}, {Petit}, {Prugniel}, {Theureau},

1461:   {Rousseau}, {Brouty}, {Dubois}, \& {Cambr{\'e}sy}}]{paturel03}

1462: {Paturel}, G. et al. 2003, \aap, 412, 45

1463: 

1464: \bibitem[{{Peng} {et~al.}(2002){Peng}, {Ho}, {Impey}, \& {Rix}}]{peng02}

1465: {Peng}, C.~Y., {Ho}, L.~C., {Impey}, C.~D., \& {Rix}, H.-W. 2002, \aj, 124, 266

1466: 

1467: \bibitem[{{Rettura} {et~al.}(2006){Rettura}, {Rosati}, {Strazzullo},

1468:   {Dickinson}, {Fosbury}, {Rocca-Volmerange}, {Cimatti}, {di Serego Alighieri},

1469:   {Kuntschner}, {Lanzoni}, {Nonino}, {Popesso}, {Stern}, {Eisenhardt},

1470:   {Lidman}, \& {Stanford}}]{rettura06}

1471: {Rettura}, A. et al. 2006, \aap, 458, 717

1472: 

1473: \bibitem[{{Rettura} {et~al.}(2008){Rettura}}]{rettura08}

1474: {Rettura}, A. et al. 2008, \aap, in press, arXiv:0806.4604

1475: 

1476: \bibitem[{{Robertson} {et~al.}(2006){Robertson}, {Cox}, {Hernquist}, {Franx},

1477:   {Hopkins}, {Martini}, \& {Springel}}]{robertson06}

1478: {Robertson}, B., {Cox}, T.~J., {Hernquist}, L., {Franx}, M., {Hopkins}, P.~F.,

1479:   {Martini}, P., \& {Springel}, V. 2006, \apj, 641, 21

1480: 

1481: \bibitem[{{Rodighiero} {et~al.}(2007){Rodighiero}, {Gruppioni}, {Civano},

1482:   {Comastri}, {Franceschini}, {Mignoli}, {Fritz}, {Vignali}, \&

1483:   {Treu}}]{rodighiero07}

1484: {Rodighiero}, G. et al. 2007, \mnras, 376, 416

1485: 

1486: \bibitem[{{Salpeter}(1955)}]{salpeter55}

1487: {Salpeter}, E.~E. 1955, \apj, 121, 161

1488: 

1489: \bibitem[{{Scalo}(1986)}]{scalo86}

1490: {Scalo}, J.~M. 1986, Fundamentals of Cosmic Physics, 11, 1

1491: 

1492: \bibitem[{{Scarlata} {et~al.}(2007){Scarlata}, {Carollo}, {Lilly}, {Feldmann},

1493:   {Kampczyk}, {Renzini}, {Cimatti}, {Halliday}, {Daddi}, {Sargent},

1494:   {Koekemoer}, {Scoville}, {Kneib}, {Leauthaud}, {Massey}, {Rhodes}, {Tasca},

1495:   {Capak}, {McCracken}, {Mobasher}, {Taniguchi}, {Thompson}, {Ajiki}, {Aussel},

1496:   {Murayama}, {Sanders}, {Sasaki}, {Shioya}, \& {Takahashi}}]{scarlata07}

1497: {Scarlata}, C. et al. 2007, \apjs, 172, 494

1498: 

1499: \bibitem[{{S\'ersic}(1968)}]{sersic68}

1500: {S\'ersic}, J.~L. 1968, {Atlas de galaxias australes} (Cordoba, Argentina:

1501:   Observatorio Astronomico, 1968)

1502: 

1503: \bibitem[{{Shen} {et~al.}(2003){Shen}, {Mo}, {White}, {Blanton}, {Kauffmann},

1504:   {Voges}, {Brinkmann}, \& {Csabai}}]{shen03}

1505: {Shen}, S. et al. 2003, \mnras, 343, 978

1506: 

1507: \bibitem[{{Sin}(1994)}]{sin94}

1508: {Sin}, S.-J. 1994, \prd, 50, 3650

1509: 

1510: \bibitem[{{Toft} {et~al.}(2007){Toft}, {van Dokkum}, {Franx}, {Labbe},

1511:   {F{\"o}rster Schreiber}, {Wuyts}, {Webb}, {Rudnick}, {Zirm}, {Kriek}, {van

1512:   der Werf}, {Blakeslee}, {Illingworth}, {Rix}, {Papovich}, \&

1513:   {Moorwood}}]{toft07}

1514: {Toft}, S. et al. 2007, \apj, 671, 285

1515: 

1516: \bibitem[{{Treu} {et~al.}(2005{\natexlab{a}}){Treu}, {Ellis}, {Liao}, \& {van

1517:   Dokkum}}]{treu05a}

1518: {Treu}, T., {Ellis}, R.~S., {Liao}, T.~X., \& {van Dokkum}, P.~G.

1519:   2005{\natexlab{a}}, \apjl, 622, L5

1520: 

1521: \bibitem[{{Treu} {et~al.}(2005{\natexlab{b}}){Treu}, {Ellis}, {Liao}, {van

1522:   Dokkum}, {Tozzi}, {Coil}, {Newman}, {Cooper}, \& {Davis}}]{treu05b}

1523: {Treu}, T. et al. 2005{\natexlab{b}},

1524:   \apj, 633, 174

1525: 

1526: \bibitem[{{Trujillo} {et~al.}(2007){Trujillo}, {Conselice}, {Bundy}, {Cooper},

1527:   {Eisenhardt}, \& {Ellis}}]{trujillo07}

1528: {Trujillo}, I., {Conselice}, C.~J., {Bundy}, K., {Cooper}, M.~C., {Eisenhardt},

1529:   P., \& {Ellis}, R.~S. 2007, \mnras, 382, 109

1530: 

1531: \bibitem[{{Trujillo} {et~al.}(2006{\natexlab{a}}){Trujillo}, {Feulner},

1532:   {Goranova}, {Hopp}, {Longhetti}, {Saracco}, {Bender}, {Braito}, {Della Ceca},

1533:   {Drory}, {Mannucci}, \& {Severgnini}}]{trujillo06b}

1534: {Trujillo}, I. et al. 2006{\natexlab{a}}, \mnras, 373, L36

1535: 

1536: \bibitem[{{Trujillo} {et~al.}(2006{\natexlab{b}}){Trujillo}, {F{\"o}rster

1537:   Schreiber}, {Rudnick}, {Barden}, {Franx}, {Rix}, {Caldwell}, {McIntosh},

1538:   {Toft}, {H{\"a}ussler}, {Zirm}, {van Dokkum}, {Labb{\'e}}, {Moorwood},

1539:   {R{\"o}ttgering}, {van der Wel}, {van der Werf}, \& {van

1540:   Starkenburg}}]{trujillo06a}

1541: {Trujillo}, I. et al. 2006{\natexlab{b}}, \apj, 650, 18

1542: 

1543: \bibitem[{{Trujillo} {et~al.}(2004){Trujillo}, {Rudnick}, {Rix}, {Labb{\'e}},

1544:   {Franx}, {Daddi}, {van Dokkum}, {F{\"o}rster Schreiber}, {Kuijken},

1545:   {Moorwood}, {R{\"o}ttgering}, {van de Wel}, {van der Werf}, \& {van

1546:   Starkenburg}}]{trujillo04}

1547: {Trujillo}, I. et al. 2004, \apj, 604, 521

1548: 

1549: \bibitem[{{van der Marel} \& {van Dokkum}(2007)}]{vandermarel07b}

1550: {van der Marel}, R.~P., \& {van Dokkum}, P.~G. 2007, \apj, 668, 756

1551: 

1552: \bibitem[{{van der Wel} {et~al.}(2007){van der Wel}, {Franx}, {Illingworth}, \&

1553:   {van Dokkum}}]{vanderwel07a}

1554: {van der Wel}, A., {Franx}, M., {Illingworth}, G.~D., \& {van Dokkum}, P.~G.

1555:   2007, \apj, 666, 863

1556: 

1557: \bibitem[{{van der Wel} {et~al.}(2004){van der Wel}, {Franx}, {van Dokkum}, \&

1558:   {Rix}}]{vanderwel04}

1559: {van der Wel}, A., {Franx}, M., {van Dokkum}, P.~G., \& {Rix}, H.-W. 2004,

1560:   \apjl, 601, L5

1561: 

1562: \bibitem[{{van der Wel} {et~al.}(2005){van der Wel}, {Franx}, {van Dokkum},

1563:   {Rix}, {Illingworth}, \& {Rosati}}]{vanderwel05}

1564: {van der Wel}, A., {Franx}, M., {van Dokkum}, P.~G., {Rix}, H.-W.,

1565:   {Illingworth}, G.~D., \& {Rosati}, P. 2005, \apj, 631, 145

1566: 

1567: \bibitem[{{van der Wel} {et~al.}(2006){van der Wel}, {Franx}, {Wuyts}, {van

1568:   Dokkum}, {Huang}, {Rix}, \& {Illingworth}}]{vanderwel06b}

1569: {van der Wel}, A., {Franx}, M., {Wuyts}, S., {van Dokkum}, P.~G., {Huang}, J.,

1570:   {Rix}, H.-W., \& {Illingworth}, G. 2006, \apj, 652, 97

1571: 

1572: \bibitem[{{van der Wel} \& {van der Marel}(2008)}]{vanderwel08b}

1573: {van der Wel}, A., \& {van der Marel}, P.~M. 2008, \apj, in press, arXiv:0804.4228

1574: 

1575: \bibitem[{{van Dokkum} {et~al.}(2008){van Dokkum}, {Franx}, {Kriek}, {Holden},

1576:   {Illingworth}, {Magee}, {Bouwens}, {Marchesini}, {Quadri}, {Rudnick},

1577:   {Taylor}, \& {Toft}}]{vandokkum08b}

1578: {van Dokkum}, P. et al. 2008, \apjl, 677, L5

1579: 

1580: \bibitem[{{van Dokkum}(2005)}]{vandokkum05}

1581: {van Dokkum}, P.~G. 2005, \aj, 130, 2647

1582: 

1583: \bibitem[{{van Dokkum}(2008)}]{vandokkum08a}

1584: ---. 2008, \apj, 674, 29

1585: 

1586: \bibitem[{{van Dokkum} \& {Ellis}(2003)}]{vandokkumellis03}

1587: {van Dokkum}, P.~G., \& {Ellis}, R.~S. 2003, \apjl, 592, L53

1588: 

1589: \bibitem[{{van Dokkum} {et~al.}(1998){van Dokkum}, {Franx}, {Kelson}, \&

1590:   {Illingworth}}]{vandokkum98}

1591: {van Dokkum}, P.~G., {Franx}, M., {Kelson}, D.~D., \& {Illingworth}, G.~D.

1592:   1998, \apjl, 504, L17

1593: 

1594: \bibitem[{{van Dokkum} \& {Stanford}(2003)}]{vandokkumstanford03}

1595: {van Dokkum}, P.~G., \& {Stanford}, S.~A. 2003, \apj, 585, 78

1596: 

1597: \bibitem[{{van Dokkum} \& {van der Marel}(2007)}]{vandokkum07}

1598: {van Dokkum}, P.~G., \& {van der Marel}, R.~P. 2007, \apj, 655, 30

1599: 

1600: \bibitem[{{von der Linden} {et~al.}(2007){von der Linden}, {Best}, {Kauffmann},

1601:   \& {White}}]{vonderlinden07}

1602: {von der Linden}, A., {Best}, P.~N., {Kauffmann}, G., \& {White}, S.~D.~M.

1603:   2007, \mnras, 379, 867

1604: 

1605: \bibitem[{{Wuyts} {et~al.}(2004){Wuyts}, {van Dokkum}, {Kelson}, {Franx}, \&

1606:   {Illingworth}}]{wuyts04}

1607: {Wuyts}, S., {van Dokkum}, P.~G., {Kelson}, D.~D., {Franx}, M., \&

1608:   {Illingworth}, G.~D. 2004, \apj, 605, 677

1609: 

1610: \bibitem[{{Yan} {et~al.}(2006){Yan}, {Newman}, {Faber}, {Konidaris}, {Koo}, \&

1611:   {Davis}}]{yan06}

1612: {Yan}, R., {Newman}, J.~A., {Faber}, S.~M., {Konidaris}, N., {Koo}, D., \&

1613:   {Davis}, M. 2006, \apj, 648, 281

1614: 

1615: \bibitem[{{York} {et~al.}(2000){York}, {Adelman}, {Anderson}, {Anderson},

1616:   {Annis}, {Bahcall}, {Bakken}, {Barkhouser}, {Bastian}, {Berman}, {Boroski},

1617:   {Bracker}, {Briegel}, {Briggs}, {Brinkmann}, {Brunner}, {Burles}, {Carey},

1618:   {Carr}, {Castander}, {Chen}, {Colestock}, {Connolly}, {Crocker}, {Csabai},

1619:   {Czarapata}, {Davis}, {Doi}, {Dombeck}, {Eisenstein}, {Ellman}, {Elms},

1620:   {Evans}, {Fan}, {Federwitz}, {Fiscelli}, {Friedman}, {Frieman}, {Fukugita},

1621:   {Gillespie}, {Gunn}, {Gurbani}, {de Haas}, {Haldeman}, {Harris}, {Hayes},

1622:   {Heckman}, {Hennessy}, {Hindsley}, {Holm}, {Holmgren}, {Huang}, {Hull},

1623:   {Husby}, {Ichikawa}, {Ichikawa}, {Ivezi{\'c}}, {Kent}, {Kim}, {Kinney},

1624:   {Klaene}, {Kleinman}, {Kleinman}, {Knapp}, {Korienek}, {Kron}, {Kunszt},

1625:   {Lamb}, {Lee}, {Leger}, {Limmongkol}, {Lindenmeyer}, {Long}, {Loomis},

1626:   {Loveday}, {Lucinio}, {Lupton}, {MacKinnon}, {Mannery}, {Mantsch}, {Margon},

1627:   {McGehee}, {McKay}, {Meiksin}, {Merelli}, {Monet}, {Munn}, {Narayanan},

1628:   {Nash}, {Neilsen}, {Neswold}, {Newberg}, {Nichol}, {Nicinski}, {Nonino},

1629:   {Okada}, {Okamura}, {Ostriker}, {Owen}, {Pauls}, {Peoples}, {Peterson},

1630:   {Petravick}, {Pier}, {Pope}, {Pordes}, {Prosapio}, {Rechenmacher}, {Quinn},

1631:   {Richards}, {Richmond}, {Rivetta}, {Rockosi}, {Ruthmansdorfer}, {Sandford},

1632:   {Schlegel}, {Schneider}, {Sekiguchi}, {Sergey}, {Shimasaku}, {Siegmund},

1633:   {Smee}, {Smith}, {Snedden}, {Stone}, {Stoughton}, {Strauss}, {Stubbs},

1634:   {SubbaRao}, {Szalay}, {Szapudi}, {Szokoly}, {Thakar}, {Tremonti}, {Tucker},

1635:   {Uomoto}, {Vanden Berk}, {Vogeley}, {Waddell}, {Wang}, {Watanabe},

1636:   {Weinberg}, {Yanny}, \& {Yasuda}}]{york00}

1637: {York}, D.~G. et al. 2000, \aj, 120, 1579

1638: 

1639: \bibitem[{{Zirm} {et~al.}(2007){Zirm}, {van der Wel}, {Franx}, {Labb{\'e}},

1640:   {Trujillo}, {van Dokkum}, {Toft}, {Daddi}, {Rudnick}, {Rix},

1641:   {R{\"o}ttgering}, \& {van der Werf}}]{zirm07}

1642: {Zirm}, A.~W. et al. 2007, \apj, 656, 66

1643: 

1644: \end{thebibliography}

1645: 

1646: 

1647: \input{tab.tex}

1648: 

1649: 

1650: \end{document}

1651: