1: \documentclass{emulateapj}
2:
3: \lefthead{A. van der Wel \& R.P. van der Marel}
4: \righthead{Spatially Resolved Kinematics at $z=1$}
5: \slugcomment{The Astrophysical Journal, Accepted}
6:
7: \begin{document}
8:
9: \title{Spatially Resolved Stellar Kinematics of Field Early-Type
10: Galaxies at $z=1$: Evolution of the Rotation Rate \altaffilmark{1}}
11:
12: \author{Arjen van der Wel\altaffilmark{2} \& Roeland P.~van der
13: Marel\altaffilmark{3}}
14:
15: \altaffiltext{1}{Based on observations collected at the European
16: Southern Observatory, Chile (169.A-0458), and on observations with
17: the \textit{Hubble Space Telescope}, obtained at the Space Telescope
18: Science Institute, which is operated by AURA, Inc., under NASA
19: contract NAS 5-26555.}
20:
21: \altaffiltext{2}{Department of Physics and Astronomy,
22: Johns Hopkins University, 3400 North Charles Street, Baltimore, MD
23: 21218; e-mail: wel@pha.jhu.edu}
24:
25: \altaffiltext{3}{Space Telescope Science Institute, 3700 San Martin
26: Drive, Baltimore 21218}
27:
28: \newcommand{\lta}{\lesssim}
29: \newcommand{\gta}{\gtrsim}
30: \newcommand{\kms}{\>{\rm km}\,{\rm s}^{-1}}
31:
32:
33: \begin{abstract}
34: We use the spatial information of our previously published VLT/FORS2
35: absorption line spectroscopy to measure mean stellar velocity and
36: velocity dispersion profiles of 25 field early-type galaxies at a
37: median redshift $z=0.97$ (full range $0.6<z<1.2$). This provides the
38: first detailed study of early-type galaxy rotation at these
39: redshifts. From surface brightness profiles from \textit{HST}
40: imaging we calculate two-integral oblate axisymmetric Jeans equation
41: models for the observed kinematics. Fits to the data yield for each
42: galaxy the degree of rotational support and the mass-to-light ratio
43: $M/L_{\rm Jeans}$. S0 and Sa galaxies are generally rotationally
44: supported, whereas elliptical galaxies rotate less rapidly or not at
45: all. Down to $M_B=-19.5$ (corrected for luminosity
46: evolution), we find no evidence for evolution in the fraction of
47: rotating early-type (E+S0) galaxies between $z\sim 1$ ($63\%\pm
48: 11\%$) and the present ($61\%\pm 5\%$). We interpret this as
49: evidence for little or no change in the field S0 fraction with
50: redshift. We compare $M/L_{\rm{Jeans}}$ with $M/L_{\rm{vir}}$
51: inferred from the virial theorem and globally averaged quantities,
52: and assuming homologous evolution. There is good agreement for
53: nonrotating (mostly E) galaxies. However, for rotationally
54: supported galaxies (mostly S0) $M/L_{\rm{Jeans}}$ is on average
55: $\sim 40\%$ higher than $M/L_{\rm{vir}}$. We discuss possible
56: explanations and the implications for the evolution of $M/L$ between
57: $z=1$ and the present and its dependence on mass.
58: \end{abstract}
59:
60: \keywords{galaxies: elliptical and lenticular, cD---galaxies:
61: kinematics and dynamics---galaxies: evolution---galaxies: fundamental
62: parameters}
63:
64: \section{INTRODUCTION}
65: \label{intro}
66:
67: Detailed studies of nearby early-type galaxies provide insight into
68: their stellar populations, dark matter content and kinematic and
69: spatial structure. In particular, resolved kinematic data reveal the
70: underlying gravitational potential \citep[e.g.,][]{vandermarel91,
71: cappellari06}, the presence of super-massive black holes
72: \citep[e.g.,][]{richstone90, verolme02} and dark halos
73: \citep[e.g.,][]{kronawitter00}, and the relative contributions of
74: pressure and rotation to its orbital energy \citep[e.g.,][]{binney78,
75: davies83, emsellem07}.
76:
77: Recently, the first, probing steps have been made toward spatially
78: resolving the stellar motions in more distant early-type galaxies
79: using absorption-line spectroscopy. The past few years have seen
80: studies aimed at constraining the dark halos of lensing galaxies
81: \citep{koopmans06}, dynamically distinguishing cluster E and S0
82: galaxies \citep{moran07b}, constraining the evolution in rotation rate
83: \citep{vandermarel07a}, and constraining the evolution of the
84: mass-to-light ratio ($M/L$) using the $M/L$ versus $\sigma$ relation at
85: intermediate redshift \citep{vandermarel07b}. Obviously, the spatial
86: resolution and errors in the kinematic quantities are much larger than
87: for local galaxies, and consequently, the level of detail is far
88: lower. Still, the spatially resolved information that is obtained by
89: going beyond studies of global quantities alone, combined with the
90: addition of cosmic time to the equation, has already revealed several
91: interesting results.
92:
93: \citet{moran07b} used rotation curves to address the question whether
94: E and S0 galaxies are coeval or not. They found that the relative
95: number of rotationally supported early-type galaxies at $z\sim 0.5$ is
96: lower than today, especially in clusters. Combined with extensive
97: studies of visually classified samples of galaxies out to $z\sim 1$
98: \citep{dressler97,postman05,smith05}, this is the best evidence to
99: date for relatively recent S0 formation, presumably through the
100: transformation of in-falling, star-forming Sa-like galaxies into
101: quiescent cluster S0 galaxies \citep[e.g.,][]{gunn72, larson80}.
102:
103: This classical picture does not address the high fraction of S0
104: galaxies at low densities. Approximately 60\% of all early-type
105: galaxies in the field are S0 galaxies. Moreover, recently it was shown
106: that the total fraction of early-type galaxies does not significantly
107: change between $z\sim 0.8$ and the present either in clusters
108: \citep{holden07} or in the field \citep{vanderwel07b} if only galaxies
109: with masses $M\gtrsim 0.5M^*$ are considered. This calls into question
110: the claim that S0 galaxies evolve differently from E galaxies,
111: although we note that this has so far not been explicitly addressed.
112: Because of these reasons it is important to not lose sight of
113: alternative scenarios for the formation of S0 galaxies. Merging that
114: does not completely destroy the disk is another proposed mechanism
115: \citep[e.g.,][]{bekki97}. This is more similar to the favored
116: formation mechanism for E galaxies and would therefore imply that E
117: and S0 are in fact one class of objects with a large range in
118: bulge-to-disk ratios. This ratio may be related to the mass ratio and
119: the dissipation of the mergers \citep[e.g.,][]{naab06b}. In this
120: scenario E and S0 galaxies are expected to be roughly coeval.
121:
122: Another interesting observation in these contexts is that
123: \citet{vandermarel07a} found some evidence that the relative number of
124: rotationally supported cluster elliptical galaxies decreases slightly
125: between $z\sim 0.5$ and the present. This points toward a scenario in
126: which elliptical galaxies gradually lose angular momentum through
127: interactions and mergers \citep[e.g.,][]{naab06}. This result is not
128: necessarily inconsistent with the conclusions from \citet{moran07b}.
129: The latter authors used rotation to trace the number of S0 galaxies,
130: whereas the sample of \citet{vandermarel07a} contained (almost) no S0
131: galaxies by construction. Therefore, these studies probed different
132: aspects of the early-type (E+S0) galaxy population.
133:
134: Besides constraining rotation, resolved kinematic data also allow for
135: a measurement of $M/L$. The evolution of $M/L$ with redshift is
136: typically studied using estimates based on the virial theorem or the
137: fundamental plane \citep[e.g.,][]{vandokkum07}. This uses only
138: globally averaged quantities (luminosity, radius, integrated velocity
139: dispersion), combined with assumptions about the dynamical structure
140: of the galaxies and the homology of their evolution. Spatially
141: resolved kinematics have the advantage that they allow the $M/L$ to be
142: inferred through detailed dynamical modeling, which removes some of
143: the assumptions inherent in these studies. Systematic effects may
144: affect not only the measured, average evolution of $M/L$ but also the
145: evolution of the tilt of the fundamental plane, i.e., the dependence
146: of $M/L$ evolution on galaxy mass \citep[e.g.,][]{vanderwel05,
147: treu05b}. Van der Marel \& van Dokkum (2007b) addressed this for a
148: sample of clusters elliptical galaxies at $z \sim 0.5$. They found
149: that bright, nonrotating galaxies are consistent with homologous
150: evolution, but that the situation may be more complicated for less
151: luminous, rotating galaxies. However, the small size of their sample
152: and the limited number of S0 galaxies left room for different
153: interpretations of this result.
154:
155: In this paper we study the resolved kinematics of a sample of field
156: early-type galaxies at $z \sim 1$, using spatially resolved spectra
157: from the Very Large Telescope (VLT) and imaging from the
158: \textit{Hubble Space Telescope} (\textit{HST}). This dataset was
159: presented earlier by \citet{vanderwel05}. Our analysis methodology is
160: largely similar to that of van der Marel \& van Dokkum (2007a,
161: 2007b). However, our sample is different from theirs in three
162: important areas. First, we focus on field galaxies instead of cluster
163: galaxies. Second, our sample includes significant numbers of S0 and
164: Sa galaxies, instead of just E galaxies. And third, we work at $z \sim
165: 1$ instead of $z \sim 0.5$, thus providing a much extended baseline
166: for studying evolution. In particular, we quantify the evolution of
167: the relative number of rotationally supported field early-type
168: galaxies from $z\sim 1$ to the present, and we examine the difference
169: between the $M/L$ inferred from spatially resolved and unresolved
170: kinematic data.
171:
172: We have organized the paper as follows. In \S\ref{secdata} we
173: introduce the data. In \S\ref{secmod} we briefly describe the
174: dynamical modeling. In \S\ref{secrot} we present our results
175: regarding rotational support. In \S\ref{secml} we present our
176: results regarding the $M/L$. In \S\ref{secsum} we summarize the
177: main conclusions. We adopt the following cosmological parameters:
178: $(\Omega_M,~\Omega_{\Lambda},~h) = (0.3,~0.7,~0.7)$.
179:
180: \section{DATA}
181: \label{secdata}
182:
183: \subsection{Spectroscopy}
184: \label{secspec}
185:
186: Very deep VLT/FORS2 spectroscopy of a sample of magnitude-limited,
187: morphologically selected field early-type galaxies at redshifts
188: $0.6<z<1.2$ in the Chandra Deep Field-South (CDF-S) and the foreground
189: of the $z=1.24$ cluster RDCS 1252.9-2927 (CL1252) were presented by
190: \citet{vanderwel05}. That paper provides a full description of the
191: data reduction, including the technique used to measure the global
192: velocity dispersion of each galaxy from its absorption features. In
193: this paper we examine a subsample of 25 early-type galaxies (10 E, 9
194: S0, and 6 Sa galaxies) that have such high-quality spectra that
195: variations of mean velocity $v$ and velocity dispersion $\sigma$ along
196: the slit can be measured. The median redshift of this subsample is
197: $z=0.97$. Since this is a magnitude-limited subsample it is not biased
198: against rotating or nonrotating galaxies, and it is representative for
199: the entire population of E+S0 galaxies at this redshift.
200:
201: For high-redshift galaxies, $\sigma$ is usually determined from an
202: extracted spectrum that is averaged over multiple pixels in the
203: spatial direction in order to optimize for signal-to-noise ratio
204: ($S/N$). Here, instead, we measure $v$ and $\sigma$ for each
205: individual row of pixels (0.25\" long) for which the $S/N$ is
206: sufficiently high for a robust measurement. We use the same wavelength
207: range and template stars as used by \citet{vanderwel05} such that
208: differences between the results from averaged and resolved spectra are
209: solely due to differences in the observed spectra themselves. The
210: mean velocity profiles, typically measured for 5--7 pixel rows, are
211: shown in Figure~\ref{V}. It is readily apparent that many galaxies
212: show signs of significant rotation. Measuring $\sigma$ requires better
213: $S/N$, therefore $\sigma$ can usually only be measured for 3--5 pixel
214: rows. The $S/N$ is typically 15 $\rm{\AA}^{-1}$ for the outermost
215: pixel rows for which we measure $\sigma$ (the central pixel rows
216: obviously have the highest $S/N$). For two galaxies (CDFS-18 and
217: CDFS-23) we have $S/N<8~\rm{\AA}^{-1}$ only for the outermost pixel
218: rows, which is not ideal for velocity dispersion
219: measurements. However, for consistency we retained these rows in our
220: analysis because the large resulting error bars are explicitly
221: accounted for in our modeling described in \S\ref{secmod}. In
222: Figure~\ref{sig} we show the $\sigma$ profiles. In both
223: Figures~\ref{V} and~\ref{sig} we over-plot the best-fitting models
224: that we describe in \S\ref{secmod} below.
225:
226: \begin{figure}[t]
227: \epsscale{1.2}
228: \plotone{f1.eps}
229: \caption{Profiles of mean velocity $v$ for 25 field early-type
230: galaxies at redshifts $0.6<z<1.2$. The points indicate the measured
231: velocity at each spatial pixel in the spectroscopic slits. The solid
232: curves show the predictions of the best-fitting models. For
233: comparison, dotted curves are the corresponding predictions when
234: seeing convolution and pixel/slit binning are not taken into
235: account. The visual morphology, the redshift, and the inferred
236: rotation parameter $k$ are given for each object. An asterisk
237: indicates that the maximum value for $k$ that is physically possible
238: is adopted instead of the best-fitting value for $k$. The ID numbers
239: correspond to those given in \citet{vanderwel05}, where more
240: information regarding the photometric and kinematic properties can
241: be found.}
242: \label{V}
243: \end{figure}
244:
245: \begin{figure}[t]
246: \epsscale{1.2}
247: \plotone{f2.eps}
248: \caption{Profiles of velocity dispersion $\sigma$ for the same 25
249: field early-type galaxies at redshifts $0.6<z<1.2$ as in
250: Fig.~\ref{V}. The points indicate the measured dispersion at each
251: spatial pixel in the spectroscopic slits. The solid curves show the
252: predictions of the best-fitting models. For comparison, dotted
253: curves are the corresponding predictions when seeing convolution and
254: pixel/slit binning are not taken into account. The visual
255: morphology, the redshift, and the rest-frame $B$-band model $M/L$ in
256: solar units are given for each object.}
257: \label{sig}
258: \end{figure}
259:
260: \subsection{Photometry}
261: \label{secphot}
262:
263: GOODS\footnote{http://www.stsci.edu/science/goods/} provides deep,
264: publicly available \textit{HST}/ACS imaging of the CDF-S
265: \citep{giavalisco04}. ACS imaging for the field of CL1252 is also
266: available \citep{blakeslee03}. We used these data to assign
267: morphological classifications (E, S0, Sa) based on visual inspection
268: of the images, following the strategy outlined by \citet{postman05}.
269: These classifications are used throughout this paper.
270:
271: Van der Wel et al. (2005) used the F850LP images to fit de Vaucouleurs
272: profiles to the objects in their sample in order to determine their
273: effective radii $R_{\rm{eff}}$ and surface brightnesses
274: $\mu_{\rm{eff}}$. For the present analysis we use the ACS data to
275: measure the full surface brightness profile for each galaxy. We follow
276: the same procedure described by \citet{vandermarel07a}. First, the
277: images are deconvolved with the point-spread function (PSF; for which
278: we use stars in the field) using the Lucy-Robertson
279: algorithm. Subsequently, we fitted elliptical isophotes to the
280: two-dimensional images with the IRAF task ELLIPSE. This yields
281: one-dimensional profiles of major-axis surface brightness, position
282: angle, and ellipticity. For galaxies at redshifts $z>0.85$ we use the
283: ACS F850LP images, and for galaxies at redshifts $z<0.85$ we use the
284: ACS F775W images. With these choices, the observed wavelength
285: corresponds as closely as possible to the rest-frame $B$ band. The
286: observed surface brightness profiles are transformed into the
287: rest-frame $B$ band using the method presented in
288: \citet{vandokkum07}. This facilitates the comparison with local galaxy
289: samples.
290:
291: \section{MODELS}
292: \label{secmod}
293:
294: For a full description of the modeling procedure we refer to
295: \citet{vandermarel07a}. In short, the rest-frame $B$-band photometry
296: described in \S\ref{secphot} is fitted with the projection of a
297: parameterized, oblate axisymmetric, constant axial-ratio luminosity
298: distribution. The results of the modeling are fairly insensitive to
299: variations in the unknown inclination (van der Marel \& van Dokkum
300: 2007a, 2007b). In the discussion below we adopt for each galaxy the
301: inclination angle that is most likely, given the probability
302: distribution of intrinsic axial ratios derived from large galaxy
303: catalogs of the local universe.
304:
305: Given the three-dimensional luminosity density, the Jeans equations
306: are solved under the assumptions of a constant $M/L$ and a
307: two-integral distribution function $f=f(E,L_z)$, where $E$ is the
308: energy and $L_z$ the angular momentum around the symmetry axis. The
309: models have $\overline{v_R^2} \equiv \overline{v_z^2}$, so their
310: velocity distribution is isotropic in the meridional plane. The ratio
311: of $\overline{v_\phi^2}$ to $\overline{v_R^2}$ is determined by the
312: requirement of hydrostatic equilibrium. The second azimuthal velocity
313: moment is split into mean and random components according to the
314: convenient parameterization
315: %
316: \begin{equation}
317: \label{satohk}
318: \overline{v_{\phi}} = k \left(\overline{v_{\phi}^2} - \overline{v_R^2}\right)^{1/2} .
319: \end{equation}
320: %
321: A value $k=0$ yields a galaxy that is nonrotating and fully pressure
322: supported. A value $|k|=1$ yields an oblate isotropic rotator, i.e., a
323: galaxy with sufficient rotation to fully explain its flattening. The
324: value of $|k|$ can be larger than unity, with a physical maximum that
325: depends on the shape of the galaxy, and set by the requirement that
326: $\sigma_\phi^2 \equiv \overline{v_{\phi}^2} - (\overline{v_{\phi}})^2$
327: is everywhere positive. In the following we always choose the major
328: axis position angle (which is indeterminate module $180^{\circ}$) so
329: that $k$ is positive.
330:
331: The solutions of the Jeans equations are integrated along the line of
332: sight, and then convolved in luminosity-weighted sense with the seeing
333: (typically 0.6\"-0.8\") and with the size of the pixels (0.25\" long)
334: in the slit (1.0\" wide) through which the galaxies' spectra were
335: taken, taking the angle between the orientation of the slit and the
336: major axis into account. The match of the predictions to the
337: kinematic data from \S\ref{secspec} is optimized by finding the
338: best-fitting value of the two available model parameters. The
339: mass-to-light ratio $M/L$ determines the total amount of rms motion
340: ($\sqrt{v^2+\sigma^2}$) whereas the parameter $k$ determines the
341: amount of rotational support (which is related to $v/\sigma$). Solid
342: curves in Figures~\ref{V} and~\ref{sig} show the best model fits. For
343: comparison, dotted curves show the model predictions if no account is
344: taken of seeing and pixel/slit binning. Clearly, the latter makes a
345: considerable difference and it is therefore important that this is
346: properly modeled. Best-fit values of $k$ are listed in the individual
347: panels of Figure~\ref{V}.
348:
349:
350: \begin{figure}
351: \epsscale{1.2}[t]
352: \plotone{f3.eps}
353: %\plottwo{L_k.eps}{L_k0.eps}
354: \caption {$B$-band luminosity vs.~rotation rate. The large symbols are
355: the 25 field early-type galaxies at redshifts $0.6<z<1.2$, for which
356: we show the rotation parameter $k$. The luminosities are corrected
357: for 0.58 dex of evolution between $z=1$ and the present
358: \citep{vandokkum07}. Filled symbols are E galaxies, open circles
359: are S0 galaxies, and open, dotted circles without error bars are Sa
360: galaxies, as determined by visual classification. The local
361: comparison sample is shown as small symbols, with the different
362: morphologies distinguished by the same symbols as used for the
363: distant sample. $(V/\sigma)^*$, which is comparable to $k$, is used
364: to quantify the rotation rate for the local galaxy sample. The
365: distributions of the local and distant samples are similar.
366: \label{L_k}}
367: \end{figure}
368:
369: \begin{figure}
370: \epsscale{1.2}[t]
371: \plotone{f4.eps}
372: \caption {Ellipticity vs.~rotation rate [$k$ for the distant sample,
373: $(V/\sigma)^*$ for the local sample]. The symbols that indicate
374: different visual classifications are the same as in Fig.~\ref{L_k}.
375: The ellipticity distribution, like the rotation rate distribution,
376: is similar for the local and distant samples.
377: \label{e_k}}
378: \end{figure}
379:
380: \section{ROTATION AT $z\sim 1$}
381: \label{secrot}
382:
383: \subsection{Evolution in Rotation Rate}
384: \label{rotevol}
385:
386: As is readily apparent from Figure~\ref{V}, many early-type galaxies
387: at $z\sim 1$ show signs of rotation. In Figure~\ref{L_k} we show using
388: large symbols the rotation parameter $k$ as a function of absolute
389: $B$-band magnitude, distinguishing between galaxies with different
390: visual morphologies. The absolute magnitudes are corrected for
391: luminosity evolution using the results of
392: \citet{vandokkum07}. Slightly different values would have been
393: reasonable as well (see, e.g., \S\ref{MLimplic}), but none of our
394: results depend sensitively on this.
395:
396: Only five out of the 25 galaxies show little or no rotation. These are
397: all elliptical galaxies. By contrast, all S0 and Sa galaxies show
398: significant rotation, with many consistent with being rotationally
399: supported. Figure~\ref{e_k} shows the rotation parameter $k$ as a
400: function of the apparent ellipticity $\epsilon \equiv 1-(b/a)$, using
401: similar symbols as in Figure~\ref{L_k}. This shows that the
402: nonrotating galaxies tend to be rounder than most of the rotating
403: galaxies.
404:
405: As noted by \citet{moran07b}, a single-component, spheroidal model may
406: not be representative for S0 galaxies, some of which clearly show
407: disks. Our models can in principle be extended to include disks
408: \citep{cinzano94}, but we have not explored that here. Therefore, the
409: inferred $k$ may not always necessarily correspond to the physical
410: property it represents in the model. Nonetheless, as a fit parameter,
411: $k$ still provides an effective measure of the relative importance of
412: rotation in the galaxy. In fact, Figures~\ref{L_k} and~\ref{e_k} show
413: that E and S0 galaxies can be distinguished fairly successfully based
414: on their kinematics, as quantified by the parameter $k$. Additional
415: use of luminosity and axial ratio information can further increase the
416: accuracy of such a kinematical classification scheme, as advocated by
417: \citet{moran07b}. The only galaxy that seems somewhat out of place in
418: Figure~\ref{L_k} is the most rapidly rotating elliptical
419: galaxy. However, this is also the least luminous galaxy in the sample,
420: and it may well have been visually misclassified.
421:
422: The clean kinematic separation between different morphological types
423: at $z\sim 1$ is surprising given the complex situation for local
424: early-type galaxies \citep{emsellem07}. Both E and S0 galaxies are
425: often fast rotators with overlapping kinematic properties
426: \citep{cappellari07}. The clean separation at higher redshifts is
427: therefore quite likely partially artificial. Morphological
428: classifications are difficult at high redshift, and recent work has
429: shown that the relative number of S0 galaxies at high redshift is
430: systematically underestimated as evidenced by the lack of S0 galaxies
431: with $\epsilon<0.3$ (B.~P.~Holden et al, in preparation). In addition,
432: elongated, rotating elliptical galaxies can be misclassified as S0
433: galaxies. Nonetheless, these concerns are not relevant for the main
434: goal of this paper, the evolution of the rotation of early-type
435: galaxies as a single class of objects.
436:
437: \begin{figure}[t]
438: \epsscale{1.2}
439: \plotone{f5.eps}
440: \caption{Histograms of early-type (E+S0) galaxy rotation rates. The
441: parameter $k$ is used for the distant sample (solid line with
442: Poisson error bars) and the quantity $(V/\sigma)^*$ is used for the
443: local sample (dashed line). Only galaxies brighter than
444: $M_{B}=-19.5$ are included (corrected for luminosity evolution,
445: in the case of the distant sample). There is no significant
446: difference between the two samples in terms of relative numbers of
447: slow-rotating and fast-rotating galaxies.}
448: \label{khist}
449: \end{figure}
450:
451: To address the issue of evolution in the rotation rate, we compare the
452: $z\sim 1$ sample with a local sample extracted from the Lyon
453: Extragalactic DataBase \citep[LEDA;][]{paturel97}, which is
454: distributed and made available as HyperLeda
455: \citep{paturel03}\footnote{http://leda.univ-lyon1.fr}. We restrict
456: the local sample to galaxies within $\sim 40$ Mpc (distance modulus
457: $m-M < 33$). In order to construct a field/group E+S0 sample we
458: included only galaxies with fewer than 80 neighbors brighter than
459: $M_{B}=-19.5$ within a cylinder with a 5 Mpc diameter (at the distance
460: of the galaxy) and with a difference in radial velocity with respect
461: to the cosmic microwave background of less than $1000\kms$. This
462: effectively selects a complete sample of 179 galaxies (70 Es, 109 S0s)
463: outside overdense structures (most notably, the Virgo Cluster). The
464: median projected surface density is $\Sigma = 0.4~\rm{Mpc}^{-2}$,
465: which is derived from the distance to the 7th nearest neighbor down to
466: the luminosity limit of the magnitude-limited sample with the same
467: radial velocity within $1000~\kms$. This is the same value as for a
468: large, volume-limited sample of galaxies at $z\sim 0.03$
469: \citep[see][]{vanderwel07b}. It is also comparable with the estimated
470: surface density (in co-moving units) for a sample of distant field
471: galaxies \citep[see][]{vanderwel07b} of which our distant sample is a
472: sub-sample. Maximum stellar velocities $V$ and velocity dispersions
473: $\sigma$ are known for 55 Es and 43 S0s \citep{prugniel96}. There are
474: 37 galaxies in this sample for which multiple, independent
475: measurements of the rotation velocity $V$ are available. For 31 of
476: these the measurements are consistent within the errors. The
477: measurements are inconsistent only for the remaining six \citep[see
478: Table A2 of][]{prugniel96}. Because especially galaxies with peculiar
479: properties tend to be targeted multiple times, the measurements for
480: the remainder of the sample may actually be even better. Therefore,
481: the kinematic measurements for the local sample are generally reliable
482: and reproduceable.
483:
484: We quantify the rotation rate of the local sample using the quantity
485: $(V/\sigma)^* = (V/\sigma) \sqrt{(1-\epsilon) / \epsilon}$. This
486: approximates the ratio of the actual rotation rate of the galaxy to
487: the ratio expected for an oblate isotropic rotator of the given
488: observed axial ratio \citep{davies83}. This quantity is therefore
489: similar to our model parameter $k$. The only difference is that $k$ is
490: defined locally in the meridional plane by equation~(1), whereas
491: $(V/\sigma)^*$ is defined in terms of the globally defined projected
492: quantities $V$, $\sigma$ and $\epsilon$. Van der Marel (1991; see his
493: Fig.~3\textit{a}) made a direct comparison of $k$ and $(V/\sigma)^*$
494: for a sample of local galaxies and found a good agreement to
495: $\sim$10\%. For the present paper we therefore compare $(V/\sigma)^*$
496: for the local sample directly to $k$ for the distant sample.
497:
498: The local sample is shown in Figures~\ref{L_k} and~\ref{e_k} using
499: small symbols. As compared to the distant sample, there are some
500: differences in the types of galaxies that show high rotation rates. In
501: the distant sample these are almost all flattened, visually classified
502: S0 galaxies. By contrast, in the local sample these include quite a
503: few roundish, visually classified E galaxies. It has been known for
504: some time \citep[e.g.,][]{rixwhite90} that local galaxy catalogs are
505: deficient in such galaxies. Of course, distant samples may be plagued
506: by similar, or worse, problems. However, our sample size is too small
507: to make definite statements about this issue.
508:
509: As noted earlier, possible uncertainties in the relative
510: classifications of E and S0 galaxies do not affect the assessment of
511: the rotation rate distribution of early-type (E+S0) galaxies in
512: general. Figure~\ref{khist} shows that the rotation rate histograms
513: are statistically indistinguishable for the local and distant samples:
514: $63\%\pm11\%$ (12 out of 19) of the E+S0 galaxies in the distant
515: sample have high rotation rates ($k>0.6$), and $59\%\pm5\%$ (58 out of
516: 98) of those in the local sample have high rotation rates
517: [$(V/\sigma)^*>0.6$].
518:
519: The similarity between these two numbers is striking, however, they
520: may not be directly comparable as the kinematic data for the local
521: sample are not complete, nor representative for the morphological
522: composition of the population: an E galaxy is twice as likely to have
523: kinematic data as an S0 galaxy (see the values above). If we, very
524: crudely, assume that all S0 galaxies rotate and all E galaxies do not,
525: then the fraction of rotating E+S0 galaxies would be 69\%. However,
526: in reality the relative fractions of rotating/nonrotating E and S0
527: galaxies are not very different (55\% and 65\%, respectively, see the
528: small data points in Fig.~\ref{L_k}). Based on these relative numbers
529: of rotating galaxies for E and S0 galaxies as separate classes we
530: estimate that the true relative fraction of rotating early-type
531: galaxies is 61\%, only slightly higher than the measured value of
532: 59\%.
533:
534: Concluding, we find no evidence for evolution in the rotation rate of
535: field early-type galaxies between $z=1$ and the present down to the
536: luminosity limit of the distant sample, $M_{B}=-19.5$. The distant
537: sample is small (as evidenced by the large error bars in
538: Fig.~\ref{khist}), but we would have detected an increase or decrease
539: by more than 25\% in the relative number of rotating galaxies if there
540: were such evolution.
541:
542: \subsection{Comparison with Previous Results}
543: \label{comprevious}
544:
545: The only previous measurement of the evolution of the relative number
546: of rotationally supported early-type (E+S0) galaxies is that of
547: \citet{moran07b}. Those authors found substantial evolution for
548: cluster early-type galaxies in the sense that fewer early-type
549: galaxies are rotationally supported at $z \sim 0.5$ than locally,
550: which is interpreted as evidence for a lower S0 fraction in the
551: distant clusters. This result for cluster galaxies is not directly
552: comparable to ours. However, the \textit{field} early-type galaxy
553: sample from \citet{moran07b} contains only $43\% \pm 14\%$ rotating
554: galaxies, which is less than the value for local galaxies inferred
555: from the LEDA sample described above. If this (marginally
556: significant) decrease between $z\sim 0.5$ and the present is real,
557: more evolution between $z\sim 1$ and the present can be expected, but
558: in our sample we do not observe this. Besides the low significance,
559: this comparison is furthermore hampered by differences in
560: methodology. Therefore it is interesting to compare our results and
561: methods with those from \citet{moran07b} in more detail.
562:
563: Both we and \citet{moran07b} measured spatially resolved kinematical
564: profiles of galaxies, but beyond that our methods are very different.
565: \citet{moran07b} infer global quantities (the rotation $v$, velocity
566: dispersion $\sigma$, and ellipticity), and they do so directly from
567: the data. Observational effects are not taken into account, and the
568: internal dynamics of the galaxies are not modeled. Our method aims to
569: model the internal dynamics, to take into account known differences
570: between galaxies (e.g., in their surface brightness profile), and to
571: account explicitly for all known observational influences on the
572: measured quantities. The approach of \citet{moran07b} has the
573: advantage of being simple and model-independent. However, ignoring the
574: effects of seeing convolution and pixel/slit binning can bias the
575: results. Figure~\ref{V} shows that these effects have a significant
576: impact on both the measured rotation value at the outermost radius and
577: the maximum value of the rotation curve. As a result, the relative
578: number of rotating galaxies can be underestimated.
579:
580: To quantitatively illustrate the effect of seeing convolution and
581: pixel/slit binning, we apply one of the kinematic classifiers of
582: \citet{moran07b}, namely $v/(1-\epsilon)$, to our datasets. Here, $v$
583: is defined as half the velocity range of a fitted straight line from
584: end to end of the measured velocity profile, approximating the maximum
585: rotation velocity. From the LEDA sample we derive that $65\%\pm5\%$ of
586: the local population satisfies the criterion from \citet{moran07b},
587: virtually the same fraction as galaxies with $(V/\sigma)^*>0.6$ (see
588: \S\ref{rotevol}). This demonstrates that for the purpose of
589: distinguishing rotating and nonrotating galaxies, the two methods are
590: in principle comparable.
591:
592: In our $z\sim 1$ sample only 7 out of 19 E+S0 galaxies ($37\%\pm11\%$)
593: have $v/(1-\epsilon)>85 \kms$, consistent with the results from
594: \citet{moran07b} for $z\sim 0.5$ field galaxies, but substantially
595: less than the number of galaxies with $k>0.6$ ($63\%\pm11\%$) in our
596: $z\sim 1$ field sample. However, if we use the maximum model
597: line-of-sight velocity, with observational effects taken into account
598: (i.e., the dotted curves instead of the solid curves in Fig.~\ref{V},
599: then we find a higher fraction with $v/(1-\epsilon)>85\kms$: $63\%\pm
600: 11\%$, the same as the fraction of galaxies with $k>0.6$. This
601: implies that omission of the effects of seeing convolution and
602: pixel/slit binning can lead to significant underestimates of the
603: number of rotating galaxies at high redshift. This may well explain
604: the relatively low number of kinematically classified field S0
605: galaxies found by \citet{moran07b} at $z\sim 0.5$. However, we cannot
606: assess the question as to what extent the above-described bias
607: contributes to the evolution that \citet{moran07b} found for cluster
608: galaxies.
609:
610:
611: %We have found in Figures~\ref{L_k} and~\ref{e_k} that $(V/\sigma)^*$
612: %(or its proxy $k$) is a good indicator of morphological
613: %classification. The quantity $(V/\sigma)^*$ is defined locally in
614: %terms of the maximum velocity $V$ along the rotation
615: %curve. Figure~\ref{V} shows that to make this quantity observationally
616: %accessible at high redshift it is important to properly take into
617: %account observational influences on the measured rotation curve, as we
618: %have done here. By contrast, \citet{moran07b} found that
619: %$(V/\sigma)^*$ did not work well for them because in general no
620: %maximum in the rotation curve is readily apparent in the data.
621:
622: Interestingly, the kinematic classifications by \citet{moran07b}
623: agree, statistically speaking, with visual morphological
624: classifications, at both low and high redshifts. This argues either
625: against a bias in their results or that visual classifications can
626: suffer from a similar bias against rotating/S0 galaxies. An
627: indication that high-$z$ classifications of S0 galaxies are indeed
628: hampered by systematic problems is that the ellipticity distribution
629: for the cluster early-type galaxy population does not change
630: significantly with redshift, suggesting that high-$z$ S0 galaxies are
631: systematically misclassified as E galaxies, more so than the other way
632: around (B.~P.~Holden et al., in preparation). Still, to what extent biases
633: contribute to an apparent decline in the S0 fraction with redshift
634: will probably continue to be a matter of debate for some time to
635: come. Also, the answer may well be different for cluster and field
636: galaxies. However, it is certainly intriguing that our detailed
637: observations and modeling of field galaxies at $z \sim 1$ show no
638: evolution in the fraction of rotating early-type galaxies.
639:
640: Note that our result does not necessarily imply that there is no
641: evolution in the early-type galaxy population. On the one hand,
642: rotationally supported early-type galaxies may merge and produce
643: pressure-supported elliptical galaxies \citep[e.g.,][]{naab06}. This
644: process would produce an increase in the fraction of
645: elongated/rotating galaxies with redshift, for which some tentative
646: evidence exists in clusters from the results of
647: \citet{vandermarel07a}. On the other hand, rotating, quiescent
648: galaxies may be formed out of Sa galaxies that cease to form stars,
649: contributing to the increase in the stellar mass density of red
650: galaxies \citep[e.g.,][]{bell04b, brown07} and the decrease of the
651: cosmic average star formation rate
652: \citep[e.g.,][]{lefloch05,noeske07a}. As long as the mechanisms that
653: increase and decrease the relative numbers of rotating and nonrotating
654: galaxies do not change their ratio, continued growth of both the E and
655: the S0 population is possible up to the present day.
656:
657: Neither the results of our own study nor that of \citet{moran07b} are
658: directly comparable to those obtained by \citet{vandermarel07a} using
659: the same technique as was in this paper. This is because their study
660: at $z \sim 0.5$ dealt with a cluster sample that was preselected to
661: contain almost no S0 galaxies. Therefore, it can shed little light on
662: the evolution of the S0 fraction with redshift.
663:
664:
665: \section{MASS-TO-LIGHT RATIOS}\label{secml}
666: \subsection{Comparison between Model and Virial $M/L$ Estimates}
667:
668: According to the virial theorem, the mass of a stellar system can be
669: written as
670: %
671: \begin{equation}
672: M_{\rm{vir}} = \frac{\beta R_{\rm eff} \sigma_{\rm eff}^2}{G} ,
673: \label{virial}
674: \end{equation}
675: %
676: where $R_{\rm eff}$ is the effective radius and $\sigma_{\rm eff}$ is
677: the luminosity-weighted velocity dispersion measured through an
678: aperture of size $R_{\rm eff}$. The homology parameter $\beta$
679: generally depends on the detailed density structure and velocity
680: dispersion anisotropy of the galaxy. Plausible models can span a wide
681: range of $\beta$ values. However, \citet{cappellari06} found that
682: early-type galaxies in the local universe all follow $\beta = 5.0 \pm
683: 0.1$, with surprisingly low scatter. This calibration was obtained by
684: calculating galaxy masses $M$ from detailed dynamical models for
685: integral-field kinematical data, including higher-order velocity
686: moments. The quantity $\sigma_{\rm eff}$ was measured by integration
687: over the spatially resolved two-dimensional velocity field. The virial
688: mass-to-light ratio $M/L_{\rm vir}$ can be calculated upon division by
689: the luminosity $L$. The latter can be measured through aperture
690: photometry or it can be estimated as $L = 2 \pi R_{\rm eff}^2 I_{\rm
691: eff}$, where $I_{\rm eff}$ is the average intensity inside $R_{\rm
692: eff}$.
693:
694: Equation~(\ref{virial}) is commonly used to estimate the mass of
695: distant galaxies. In doing so, one generally measures the velocity
696: dispersion $\sigma_c$ through a spectroscopic aperture. The velocity
697: dispersion $\sigma_{\rm eff}$ is then estimated by applying a
698: correction formula that is based on typical observations of local
699: galaxies \citep{jorgensen95}. However, the findings of
700: \citet{cappellari06} do not guarantee that equation~(\ref{virial})
701: with $\beta = 5$ is as accurate as it is in the local universe, for at
702: least two different reasons. First, galaxies may evolve so that the
703: homology parameter $\beta$ could be a function of redshift. Second,
704: the accuracy and applicability of the corrections from $\sigma_c$ to
705: $\sigma_{\rm eff}$ are not guaranteed at high redshift. The
706: spectroscopic apertures and the seeing in use there often exceed the
707: galaxy size $R_{\rm eff}$. Observations that adequately mimic this are
708: not available in the local universe. Moreover, even in the local
709: universe $\sigma_{\rm eff}/\sigma_c$ shows strong variations from
710: galaxy to galaxy \citep{cappellari06}. It is therefore necessary to
711: calibrate equation~(\ref{virial}), and the appropriate value of
712: $\beta$, at high redshift in the same way as was done locally. This is
713: possible for our sample by comparing the $M/L_{\rm{Jeans}}$ from our
714: dynamical models (\S\ref{secmod}) with the values of $M/L_{\rm
715: vir}$. Our data quality is obviously not comparable to the
716: two-dimensional velocity fields and higher-order moments that are
717: available locally. Nonetheless, this calibration is unique and has not
718: previously been done at these redshifts.
719:
720: In Figure~\ref{ML_ML} we compare $M/L_{\rm{vir}}$ and
721: $M/L_{\rm{Jeans}}$ for our sample at $z \sim 1$. We computed
722: $M/L_{\rm{vir}}$ using $\beta = 5$, with $\sigma_c$ and $R_{\rm{eff}}$
723: as measured by \citet{vanderwel05}. The spectroscopic aperture was
724: defined by the width of the spectroscopic slit ($1"$) and the height
725: of the extracted spectrum (usually, $1.25"$). The ratio
726: $\sigma_{\rm{eff}}/\sigma_c$ was estimated using the formula of
727: \citet{cappellari06}, which is consistent with those of
728: \citet{jorgensen95}. As for local galaxies, the slope of the
729: best-fitting linear relation between $M/L_{\rm{vir}}$ and
730: $M/L_{\rm{Jeans}}$ is consistent with unity. However, there is a
731: systematic offset that appears to depends on the galaxy type. This is
732: seen in the residuals $\log [(M/L)_{\rm{Jeans}} / (M/L)_{\rm{vir}}]$
733: shown in Figure~\ref{var_MLML}. For E galaxies, the residuals are
734: consistent with zero. Therefore, $\beta = 5$ is appropriate for E
735: galaxies at high redshift, and as a class E galaxies are consistent
736: with homologous evolution. For S0 (and Sa) galaxies, on the other
737: hand, $M/L_{\rm{Jeans}}$ is on average almost 40\% higher than
738: $M/L_{\rm{vir}}$. As shown in Figures~\ref{var_MLML}\textit{a} and
739: \ref{var_MLML}\textit{b}, these galaxies tend to be flatter and more
740: rapidly rotating than the E galaxies. These results are consistent
741: with the earlier findings of \citet{vandermarel07b} in clusters at $z
742: \sim 0.5$. However, the trends are much clearer in the present sample
743: because of its higher fraction of rapidly rotating S0 and Sa galaxies.
744:
745: \begin{figure}[t]
746: \epsscale{1.2}
747: \plotone{f6.eps}
748: \caption {Jeans $M/L$ vs.~virial $M/L$ (calculated using $\beta = 5$
749: in eq.~[\ref{virial}]) for the $z\sim 1$ galaxy sample. The symbols
750: that indicate different visual classifications are the same as in
751: Fig.~\ref{L_k}. E galaxies follow the relation $M/L_{\rm Jeans}
752: \approx M/L_{\rm vir}$ (dotted line). However, S0 galaxies on
753: average have $M/L_{\rm{Jeans}}$ systematically higher than
754: $M/L_{\rm{vir}}$ by $\sim 40\%$. The solid line is the least-squares
755: fit for the full sample, which has $M/L_{\rm{Jeans}} \propto
756: M/L_{\rm vir}^{1.02}$ (and has $M/L_{\rm{Jeans}}$ systematically
757: higher than $M/L_{\rm{vir}}$ by $\sim 20\%$).
758: \label{ML_ML}}
759: \end{figure}
760:
761: \begin{figure*}[t]
762: \epsscale{1.2}
763: \plotone{f7.eps}
764: \caption {The logarithm of the ratio $(M/L)_{\rm{Jeans}} /
765: (M/L)_{\rm{vir}}$ vs.~ellipticity (\textit{left panel}), $v/\sigma$
766: with $v$ as defined by \citet{moran07b} (\textit{middle panel}), and
767: effective radius (\textit{right panel}). The symbols that indicate
768: different visual classifications are the same as in
769: Fig.~\ref{L_k}. The residuals tend to be positive for galaxies of
770: S0 or Sa type, for galaxies with significant ellipticity or rotation
771: rate, or for galaxies of small size. Possible causes for this, and
772: implications for studies of fundamental plane evolution, are
773: discussed in the text.
774: \label{var_MLML}}
775: \end{figure*}
776:
777: There are several possible explanations for the trends observed in
778: Figure~\ref{var_MLML}. First, S0 and Sa galaxies at $z \sim 1$ may
779: have $\beta > 5$. However, this would imply that these galaxies evolve
780: nonhomologously, while E galaxies do not. This seems to us unlikely as
781: in that case it would be an odd coincidence that $\beta=5$ for all
782: types of local galaxies. Second, it is possible that our models have
783: overestimated the $M/L_{\rm{Jeans}}$ for rapidly rotating
784: galaxies. Such galaxies may have rapidly rotating disks that are not
785: well represented by our constant axial ratio oblate models. However,
786: \citet{cinzano94} addressed this issue explicitly for the well-studied
787: nearby early-type disk galaxy NGC~2974 and found that a
788: single-component model overestimated the $M/L$ by only 12\%. This is
789: not sufficient to explain the trends seen in
790: Figure~\ref{var_MLML}. Third, it is possible that the virial
791: equation~(\ref{virial}) for $M/L_{\rm{vir}}$ underestimates the true
792: mass of rapidly rotating galaxies at high redshift. Rotation
793: contributes to the hydrostatic support of the galaxy, so if it is
794: ignored then the $M/L$ will be underestimated. While rotation does
795: increase $\sigma_{\rm eff}$ through the effect of line broadening,
796: this may not fully capture the true hydrostatic importance of the
797: rotation component. This may be especially important if the seeing is
798: of order of or larger than the scale at which rotation is manifest.
799: Such problems are avoided for local galaxies for which much higher
800: quality data are available. This may explain the apparent differences
801: between the $M/L$ of rotating galaxies at low and high redshifts. For
802: our distant galaxies, correction formulae for $\sigma_{\rm
803: eff}/\sigma_c$ are used that are not themselves calibrated at high
804: redshift.
805:
806:
807: \subsection{Implications for the Evolution of $M/L$}
808: \label{MLimplic}
809:
810: In the previous section we demonstrated that $M/L_{\rm{vir}}$ may be
811: too low for rapidly rotating galaxies at higher redshifts, but not for
812: local early-type galaxies. Since the fundamental plane is in essence a
813: correlation between $M/L_{\rm{vir}}$ and other global galaxy
814: parameters, this will bias the amount of $M/L$ evolution inferred from
815: fundamental plane studies \citep[see][and references
816: therein]{vandokkum07}. Van der Wel (2005) found that the rate of
817: luminosity evolution for the sample of field galaxies that we also use
818: in this paper is $\Delta \log(M/L) = (-0.76 \pm 0.07)
819: z$. Figures~\ref{ML_ML} and~\ref{var_MLML} suggest that, averaged over
820: the rotating and nonrotating galaxies in our sample, $M/L_{\rm vir}$
821: may be too low by $\sim 20$\%. If so, then the evolution in $M/L$ is
822: reduced to $\log(M/L) = (-0.69 \pm 0.07) z$. This is steeper than the
823: evolution found for cluster galaxies, but not by much
824: \citep{vandokkum07}.
825:
826: The difference between $M/L_{\rm{Jeans}}$ and $M/L_{\rm{vir}}$ depends
827: on the galaxy rotation rate. Moreover, rotating galaxies tend to have
828: smaller $R_{\rm{eff}}$ than nonrotating galaxies (see
829: Fig.~\ref{var_MLML}\textit{c}). Therefore, our results also have
830: consequences for the measured evolution of the tilt of the fundamental
831: plane and hence for the slope in the relation between $M/L$ and either
832: $M$ or $\sigma$ (as discussed previously by van der Marel \& van
833: Dokkum 2007b). Evolution in the tilt of the fundamental plane has been
834: observed by many authors \citep{vanderwel04, treu05b, vanderwel05,
835: diserego05, jorgensen05}. However, it is a matter of debate to what
836: extent sample selection effects contribute to the observed evolution
837: \citep{vanderwel05, treu05b}. To illustrate the consequences of the
838: results presented in this paper we revisit the analysis from
839: \citet{vanderwel05}. They showed that the probability that the
840: observed distribution of $M_{\rm{vir}}$ and $M/L_{\rm{vir}}$ is drawn
841: from a parent population with the same distribution as the local
842: galaxy population, apart from a constant amount of luminosity
843: evolution inferred from the most massive galaxies, is as low as
844: $0.14\%$. We repeat the analysis but now using the distribution of
845: $M_{\rm{Jeans}}$ and $M/L_{\rm{Jeans}}$ and find that this increases
846: the probability to 3.0\%. In other words, the evidence for
847: mass-dependent evolution of $M/L$ remains strong, but it does become
848: weaker.
849:
850: These arguments illustrate that the often observed evolution of the
851: tilt of the fundamental plane is not necessarily entirely due to
852: mass-dependent evolution of the $M/L$, i.e., downsizing. In order to
853: determine to what extent the slope and the scatter of the relation
854: between $M/L$ and $M$ evolve, still deeper observations are required
855: to overcome the biases caused by sample selection effects in the
856: surveys conducted over the past few years. In addition, it may be
857: necessary to use dynamical modeling, as we do in this paper, to
858: overcome the shortcomings of virial and fundamental plane mass
859: estimates.
860:
861: \section{CONCLUSIONS}
862: \label{secsum}
863:
864: We use the spatial information of our previously published VLT/FORS2
865: absorption-line spectroscopy to measure mean stellar velocity and
866: velocity dispersion profiles of 25 field early-type galaxies in the
867: redshift range $0.6<z<1.2$, with median redshift $z=0.97$. The
868: kinematical profiles can be reliably measured even in the most distant
869: galaxies. Rotation is detected in the majority of the sample. Surface
870: brightness profiles are determined from \textit{HST}
871: imaging. Two-integral solutions of the Jeans equations for oblate
872: axisymmetric, constant axial-ratio models are calculated to interpret
873: the data, taking into account line-of-sight projection, seeing
874: convolution, and pixel/slit binning. This yields for each galaxy the
875: degree of rotational support, as quantified by the parameter $k$, and
876: the mass-to-light ratio $M/L_{\rm{Jeans}}$ \citep{vandermarel07a}.
877:
878: The rapidly rotating ($k \gta 0.6$) and slow-rotating ($k \lta 0.6$)
879: galaxies in the distant sample tend to have different global
880: properties. The rapidly rotating galaxies tend to have later
881: morphological types (S0 or Sa) and tend to be less luminous and more
882: elongated. The slow-rotating galaxies tend to have earlier
883: morphological types (E) and tend to be more luminous and rounder. The
884: systematic variations with luminosity and axial ratio suggest that the
885: correlation with morphological type is real, in agreement with the
886: findings of \citet{moran07b}. Local E and S0 galaxies show a more
887: complex and overlapping set of kinematic properties
888: \citep[e.g.,][]{cappellari07}. The contrasting clean separation of
889: rotating and nonrotating galaxies according morphological type
890: observed at $z\sim 1$ is most likely partially artificial and must be
891: related to the fact that round/face-on S0 galaxies are often
892: misclassified as E galaxies, at high redshift even more so than at low
893: redshift.
894:
895: We have compiled a local comparison sample to study evolution of the
896: field early-type (E+S0) galaxy rotation rate. The distribution of $k$
897: in our $z\sim 1$ sample is statistically indistinguishable from the
898: distribution of $(V/\sigma)^*$, a comparable measure of rotational
899: support, in the local sample. The relative fraction of rotating
900: galaxies does not change significantly between $z\sim 1$
901: ($63\%\pm11\%$) and the present ($61\%\pm5\%$). If rotation is taken
902: to be a reliable indicator of morphological type, then this provides
903: evidence for an unchanging fraction of S0 galaxies in the early-type
904: {\it field} galaxy population with redshift. This conflicts with the
905: findings of \citet{moran07b} who did not correct for the effects of
906: seeing convolution and pixel/slit binning on the measured rotation
907: curves. It is possible that this may have led to an underestimate of
908: the number of rapidly rotating field galaxies at $z\sim 0.5$ in their
909: study.
910:
911: We have compared the mass-to-light ratio $M/L_{\rm{Jeans}}$ from our
912: spatially resolved models to the values $M/L_{\rm{vir}}$ inferred by
913: applying the virial theorem to globally averaged quantities. For E
914: galaxies, which are generally slow-rotating, we find good agreement
915: using a homology parameter $\beta = 5$, which is consistent with the
916: value calibrated locally. So there is no evidence for non-homologous
917: evolution of E galaxies out to $z \sim 1$. On the other hand, for
918: elongated, rotating galaxies (which are often S0s) we find that
919: $M/L_{\rm{Jeans}}$ is on average $\sim 40\%$ higher than
920: $M/L_{\rm{vir}}$ (see Figs.~\ref{ML_ML} and \ref{var_MLML}). For
921: local galaxies this trend is not observed. This may hint at
922: non-homologous evolution of this galaxy population ($\beta$ increasing
923: with redshift); it may suggest that our dynamical models produce
924: biased results in these galaxies because of the neglect of cold disks;
925: or it may suggest that the virial formula for $M/L$ produces biased
926: results when applied to poorly resolved, rapidly rotating galaxies at
927: high redshift. We cannot unambiguously identify the true cause, but to
928: us the latter seems to be the most straightforward explanation. If so,
929: then studies of fundamental plane evolution overestimate both the
930: amount of evolution and the evolution in the tilt of the fundamental
931: plane (and the relation between $M$ and $M/L$), which is generally
932: interpreted as evidence of down-sizing.
933:
934: \acknowledgements{The authors would like to thank Pieter van Dokkum,
935: Marijn Franx, Michele Cappellari, Dan Kelson, and Stijn Wuyts for
936: helpful discussions and suggestions. A.~v.~d.~W.~acknowledges
937: support from NASA grant NAG5-7697.}
938:
939: \begin{thebibliography}{}
940: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
941:
942: \bibitem[{{Bekki}(1997)}]{bekki97}
943: {Bekki}, K. 1997, \apjl, 490, L37
944:
945: \bibitem[{{Bell} {et~al.}(2004){Bell}, {Wolf}, {Meisenheimer}, {Rix}, {Borch},
946: {Dye}, {Kleinheinrich}, {Wisotzki}, \& {McIntosh}}]{bell04b}
947: {Bell}, E.~F. et al. 2004, \apj,
948: 608, 752
949:
950: \bibitem[{{Binney}(1978)}]{binney78}
951: {Binney}, J. 1978, \mnras, 183, 501
952:
953: \bibitem[{{Blakeslee} {et~al.}(2003){Blakeslee}, {Franx}, {Postman}, {Rosati},
954: {Holden}, {Illingworth}, {Ford}, {Cross}, {Gronwall}, {Ben{\'{\i}}tez},
955: {Bouwens}, {Broadhurst}, {Clampin}, {Demarco}, {Golimowski}, {Hartig},
956: {Infante}, {Martel}, {Miley}, {Menanteau}, {Meurer}, {Sirianni}, \&
957: {White}}]{blakeslee03}
958: {Blakeslee}, J.~P. et al. 2003, \apjl, 596, L143
959:
960: \bibitem[{{Brown} {et~al.}(2007){Brown}, {Dey}, {Jannuzi}, {Brand}, {Benson},
961: {Brodwin}, {Croton}, \& {Eisenhardt}}]{brown07}
962: {Brown}, M.~J.~I. et al. 2007, \apj, 654, 858
963:
964: \bibitem[{{Cappellari} {et~al.}(2006){Cappellari}, {Bacon}, {Bureau}, {Damen},
965: {Davies}, {de Zeeuw}, {Emsellem}, {Falc{\'o}n-Barroso}, {Krajnovi{\'c}},
966: {Kuntschner}, {McDermid}, {Peletier}, {Sarzi}, {van den Bosch}, \& {van de
967: Ven}}]{cappellari06}
968: {Cappellari}, M. et al. 2006, \mnras, 366, 1126
969:
970: \bibitem[{{Cappellari} {et~al.}(2007){Cappellari}, {Emsellem}, {Bacon},
971: {Bureau}, {Davies}, {de Zeeuw}, {Falc{\'o}n-Barroso}, {Krajnovi{\'c}},
972: {Kuntschner}, {McDermid}, {Peletier}, {Sarzi}, {van den Bosch}, \& {van de
973: Ven}}]{cappellari07}
974: {Cappellari}, M. et al. 2007, \mnras, 379, 418
975:
976: \bibitem[{{Cinzano} \& {van der Marel}(1994)}]{cinzano94}
977: {Cinzano}, P., \& {van der Marel}, R.~P. 1994, \mnras, 270, 325
978:
979: \bibitem[{{Davies} {et~al.}(1983){Davies}, {Efstathiou}, {Fall}, {Illingworth},
980: \& {Schechter}}]{davies83}
981: {Davies}, R.~L., {Efstathiou}, G., {Fall}, S.~M., {Illingworth}, G., \&
982: {Schechter}, P.~L. 1983, \apj, 266, 41
983:
984: \bibitem[{{di Serego Alighieri} {et~al.}(2005){di Serego Alighieri}, {Vernet},
985: {Cimatti}, {Lanzoni}, {Cassata}, {Ciotti}, {Daddi}, {Mignoli}, {Pignatelli},
986: {Pozzetti}, {Renzini}, {Rettura}, \& {Zamorani}}]{diserego05}
987: {di Serego Alighieri}, S. et al. 2005, \aap, 442, 125
988:
989: \bibitem[{{Dressler} {et~al.}(1997){Dressler}, {Oemler}, {Couch}, {Smail},
990: {Ellis}, {Barger}, {Butcher}, {Poggianti}, \& {Sharples}}]{dressler97}
991: {Dressler}, A. et al. 1997, \apj, 490, 577
992:
993: \bibitem[{{Emsellem} {et~al.}(2007){Emsellem}, {Cappellari}, {Krajnovi{\'c}},
994: {van de Ven}, {Bacon}, {Bureau}, {Davies}, {de Zeeuw}, {Falc{\'o}n-Barroso},
995: {Kuntschner}, {McDermid}, {Peletier}, \& {Sarzi}}]{emsellem07}
996: {Emsellem}, E. et al. 2007, \mnras, 379, 401
997:
998: %\bibitem[{{Franx}(1993)}]{franx93}
999: %{Franx}, M. 1993, \pasp, 105, 1058
1000:
1001: \bibitem[{{Giavalisco} {et~al.}(2004){Giavalisco}, {Ferguson}, {Koekemoer},
1002: {Dickinson}, {Alexander}, {Bauer}, {Bergeron}, {Biagetti}, {Brandt},
1003: {Casertano}, {Cesarsky}, {Chatzichristou}, {Conselice}, {Cristiani}, {Da
1004: Costa}, {Dahlen}, {de Mello}, {Eisenhardt}, {Erben}, {Fall}, {Fassnacht},
1005: {Fosbury}, {Fruchter}, {Gardner}, {Grogin}, {Hook}, {Hornschemeier}, {Idzi},
1006: {Jogee}, {Kretchmer}, {Laidler}, {Lee}, {Livio}, {Lucas}, {Madau},
1007: {Mobasher}, {Moustakas}, {Nonino}, {Padovani}, {Papovich}, {Park},
1008: {Ravindranath}, {Renzini}, {Richardson}, {Riess}, {Rosati}, {Schirmer},
1009: {Schreier}, {Somerville}, {Spinrad}, {Stern}, {Stiavelli}, {Strolger},
1010: {Urry}, {Vandame}, {Williams}, \& {Wolf}}]{giavalisco04}
1011: {Giavalisco}, M. et al. 2004, \apjl, 600, L93
1012:
1013: \bibitem[{{Gunn} \& {Gott}(1972)}]{gunn72}
1014: {Gunn}, J.~E., \& {Gott}, J.~R. 1972, \apj, 176, 1
1015:
1016: \bibitem[{{Holden} {et~al.}(2007){Holden}, {Illingworth}, {Franx},
1017: {Blakeslee}, {Postman}, {Kelson}, {van der Wel}, {Demarco}, {Magee},
1018: {Tran}, {Zirm}, {Ford}, {Rosati}, \& {Homeier}}]{holden07} {Holden},
1019: B.~P. et al. 2007, \apj, 670, 195
1020:
1021: %\bibitem[{{Holden} {et~al.}(2008)}]{holden08}
1022: %{Holden}, B.~P. et al. 2008, \apj, in preparation
1023:
1024: \bibitem[{{J{\o}rgensen} {et~al.}(1995)}]{jorgensen95}
1025: Jorgensen, I., Franx, M., \& Kjaergaard, P. 1995, MNRAS, 276, 1341
1026:
1027: \bibitem[{{J{\o}rgensen} {et~al.}(2005){J{\o}rgensen}, {Bergmann}, {Davies},
1028: {Barr}, {Takamiya}, \& {Crampton}}]{jorgensen05}
1029: {J{\o}rgensen}, I., {Bergmann}, M., {Davies}, R., {Barr}, J., {Takamiya}, M.,
1030: \& {Crampton}, D. 2005, \aj, 129, 1249
1031:
1032: %\bibitem[{{Kelson} {et~al.}(2000){Kelson}, {Illingworth}, {van Dokkum}, \&
1033: % {Franx}}]{kelson00}
1034: %{Kelson}, D.~D., {Illingworth}, G.~D., {van Dokkum}, P.~G., \& {Franx}, M.
1035: % 2000, \apj, 531, 184
1036:
1037: \bibitem[{{Kronawitter} {et~al.}(2000){Kronawitter}, {Saglia}, {Gerhard}, \&
1038: {Bender}}]{kronawitter00}
1039: {Kronawitter}, A., {Saglia}, R.~P., {Gerhard}, O., \& {Bender}, R. 2000, \aaps,
1040: 144, 53
1041:
1042: \bibitem[{{Koopmans} {et~al.}(2006)}]{koopmans06}
1043: Koopmans, L. V. E., Treu, T., Bolton, A. S., Burles, S., \& Moustakas, L. A.
1044: 2006, ApJ, 649, 599
1045:
1046: \bibitem[{{Larson} {et~al.}(1980){Larson}, {Tinsley}, \& {Caldwell}}]{larson80}
1047: {Larson}, R.~B., {Tinsley}, B.~M., \& {Caldwell}, C.~N. 1980, \apj, 237, 692
1048:
1049: \bibitem[{{Le Floc'h} {et~al.}(2005){Le Floc'h}, {Papovich}, {Dole}, {Bell},
1050: {Lagache}, {Rieke}, {Egami}, {P{\'e}rez-Gonz{\'a}lez}, {Alonso-Herrero},
1051: {Rieke}, {Blaylock}, {Engelbracht}, {Gordon}, {Hines}, {Misselt}, {Morrison},
1052: \& {Mould}}]{lefloch05}
1053: {Le Floc'h}, E. et al. 2005, \apj, 632, 169
1054:
1055: \bibitem[{{Moran} {et~al.}(2007){Moran}, {Loh}, {Ellis}, {Treu}, {Bundy}, \&
1056: {MacArthur}}]{moran07b}
1057: {Moran}, S.~M., {Loh}, B.~L., {Ellis}, R.~S., {Treu}, T., {Bundy}, K., \&
1058: {MacArthur}, L.~A. 2007, \apj, 665, 1067
1059:
1060: \bibitem[{{Naab} {et~al.}(2006{\natexlab{a}}){Naab}, {Jesseit}, \&
1061: {Burkert}}]{naab06b}
1062: {Naab}, T., {Jesseit}, R., \& {Burkert}, A. 2006{\natexlab{a}}, \mnras, 372,
1063: 839
1064:
1065: \bibitem[{{Naab} {et~al.}(2006{\natexlab{b}}){Naab}, {Khochfar}, \&
1066: {Burkert}}]{naab06}
1067: {Naab}, T., {Khochfar}, S., \& {Burkert}, A. 2006{\natexlab{b}}, \apjl, 636,
1068: L81
1069:
1070: \bibitem[{{Noeske} {et~al.}(2007){Noeske}, {Weiner}, {Faber}, {Papovich},
1071: {Koo}, {Somerville}, {Bundy}, {Conselice}, {Newman}, {Schiminovich}, {Le
1072: Floc'h}, {Coil}, {Rieke}, {Lotz}, {Primack}, {Barmby}, {Cooper}, {Davis},
1073: {Ellis}, {Fazio}, {Guhathakurta}, {Huang}, {Kassin}, {Martin}, {Phillips},
1074: {Rich}, {Small}, {Willmer}, \& {Wilson}}]{noeske07a}
1075: {Noeske}, K.~G. et al. 2007, \apjl, 660, L43
1076:
1077: \bibitem[{{Paturel} {et~al.}(1997){Paturel}, {Andernach}, {Bottinelli}, {di
1078: Nella}, {Durand}, {Garnier}, {Gouguenheim}, {Lanoix}, {Marthinet}, {Petit},
1079: {Rousseau}, {Theureau}, \& {Vauglin}}]{paturel97}
1080: {Paturel}, G. et al. 1997, \aaps, 124, 109
1081:
1082: \bibitem[{{Paturel} {et~al.}(2003){Paturel}, {Petit}, {Prugnier},
1083: {Theureau}, {Rousseau}, {Brouty}, {Dubois}, \&
1084: {Cambr\'esy}}]{paturel03}
1085: {Paturel}. G. et al. 2003, \aap, 412, 45
1086:
1087: \bibitem[{{Postman} {et~al.}(2005){Postman}, {Franx}, {Cross}, {Holden},
1088: {Ford}, {Illingworth}, {Goto}, {Demarco}, {Rosati}, {Blakeslee}, {Tran},
1089: {Ben{\'{\i}}tez}, {Clampin}, {Hartig}, {Homeier}, {Ardila}, {Bartko},
1090: {Bouwens}, {Bradley}, {Broadhurst}, {Brown}, {Burrows}, {Cheng}, {Feldman},
1091: {Golimowski}, {Gronwall}, {Infante}, {Kimble}, {Krist}, {Lesser}, {Martel},
1092: {Mei}, {Menanteau}, {Meurer}, {Miley}, {Motta}, {Sirianni}, {Sparks}, {Tran},
1093: {Tsvetanov}, {White}, \& {Zheng}}]{postman05}
1094: {Postman}, M. et al. 2005, \apj, 623, 721
1095:
1096: \bibitem[{{Prugniel} \& {Simien}(1996)}]{prugniel96}
1097: {Prugniel}, P., \& {Simien}, F. 1996, \aap, 309, 749
1098:
1099: \bibitem[{{Richstone} {et~al.}(1990)}]{richstone90}
1100: {Richstone}, D.~O., {Bower}, G., \& {Dressler}, A. 1990, \apj, 353, 118
1101:
1102: \bibitem[{{Rix} \& {White}(1990)}]{rixwhite90}
1103: {Rix}, H.-W., \& {White}, S.~D.~M. 1990, \apj, 362, 52
1104:
1105: % \bibitem[{{Schwarzschild}(1979)}]{schwarzschild79}
1106: % {Schwarzschild}, M. 1979, \apj, 232, 236
1107:
1108: \bibitem[{{Smith} {et~al.}(2005){Smith}, {Treu}, {Ellis}, {Moran}, \&
1109: {Dressler}}]{smith05}
1110: {Smith}, G.~P., {Treu}, T., {Ellis}, R.~S., {Moran}, S.~M., \& {Dressler}, A.
1111: 2005, \apj, 620, 78
1112:
1113: \bibitem[{{Treu} {et~al.}(2005){Treu}, {Ellis}, {Liao}, {van Dokkum}, {Tozzi},
1114: {Coil}, {Newman}, {Cooper}, \& {Davis}}]{treu05b}
1115: {Treu}, T. et al. 2005, \apj, 633, 174
1116:
1117: \bibitem[{{van der Marel}(1991)}]{vandermarel91}
1118: {van der Marel}, R.~P. 1991, \mnras, 253, 710
1119:
1120: %\bibitem[{{van der Marel} {et~al.}(1998){van der Marel}, {Cretton}, {de Zeeuw},
1121: % \& {Rix}}]{vandermarel98}
1122: %{van der Marel}, R.~P., {Cretton}, N., {de Zeeuw}, P.~T., \& {Rix}, H.-W. 1998,
1123: % \apj, 493, 613
1124:
1125: \bibitem[{{van der Marel} \& {van Dokkum}(2007{\natexlab{a}})}]{vandermarel07a}
1126: {van der Marel}, R.~P., \& {van Dokkum}, P.~G. 2007{\natexlab{a}}, \apj, 668,
1127: 738
1128:
1129: \bibitem[{{van der Marel} \& {van Dokkum}(2007{\natexlab{b}})}]{vandermarel07b}
1130: ---. 2007{\natexlab{b}}, \apj, 668, 756
1131:
1132: \bibitem[{{van der Wel} {et~al.}(2004){van der Wel}, {Franx}, {van Dokkum}, \&
1133: {Rix}}]{vanderwel04}
1134: {van der Wel}, A., {Franx}, M., {van Dokkum}, P.~G., \& {Rix}, H.-W. 2004,
1135: \apjl, 601, L5
1136:
1137: \bibitem[{{van der Wel} {et~al.}(2005){van der Wel}, {Franx}, {van Dokkum},
1138: {Rix}, {Illingworth}, \& {Rosati}}]{vanderwel05}
1139: {van der Wel}, A., {Franx}, M., {van Dokkum}, P.~G., {Rix}, H.-W.,
1140: {Illingworth}, G.~D., \& {Rosati}, P. 2005, \apj, 631, 145
1141:
1142: \bibitem[{{van der Wel} {et~al.}(2007){van der Wel}, {Holden}, {Franx},
1143: {Illingworth}, {Postman}, {Kelson}, {Labbe}, {Blakeslee}, \&
1144: {Ford}}]{vanderwel07b}
1145: {van der Wel}, A. et al. 2007, \apj, 670, 206
1146:
1147: %\bibitem[{{van Dokkum} {et~al.}(1998){van Dokkum}, {Franx}, {Kelson}, \&
1148: % {Illingworth}}]{vandokkum98}
1149: %{van Dokkum}, P.~G., {Franx}, M., {Kelson}, D.~D., \& {Illingworth}, G.~D.
1150: % 1998, \apjl, 504, L17
1151:
1152: \bibitem[{{van Dokkum} \& {van der Marel}(2007)}]{vandokkum07}
1153: {van Dokkum}, P.~G., \& {van der Marel}, R.~P. 2007, \apj, 655, 30
1154:
1155: \bibitem[{{Verolme} {et~al.}(2002){Verolme}, {Cappellari}, {Copin}, {van der
1156: Marel}, {Bacon}, {Bureau}, {Davies}, {Miller}, \& {de Zeeuw}}]{verolme02}
1157: {Verolme}, E.~K. et al. 2002, \mnras, 335, 517
1158:
1159: %\bibitem[{{Wuyts} {et~al.}(2004){Wuyts}, {van Dokkum}, {Kelson}, {Franx}, \&
1160: % {Illingworth}}]{wuyts04}
1161: %{Wuyts}, S., {van Dokkum}, P.~G., {Kelson}, D.~D., {Franx}, M., \&
1162: % {Illingworth}, G.~D. 2004, \apj, 605, 677
1163:
1164: \end{thebibliography}
1165:
1166: \end{document}
1167: