1: %&latex

2: \documentclass[12pt,preprint]{aastex}

3: %\linespread{1.6}

4: 

5: \shorttitle{LMC}

6: \shortauthors{Carrera et al.}

7: 

8: %% This is the end of the preamble.  Indicate the beginning of the

9: %% paper itself with \begin{document}.

10: %\usepackage{color}

11: %\usepackage[normalem]{ulem}

12: 

13: \begin{document}

14: 

15: %% LaTeX will automatically break titles if they run longer than

16: %% one line. However, you may use \\ to force a line break if

17: %% you desire.

18: 

19: \title{The Chemical Enrichment History of the Large Magellanic Cloud}

20: 

21: \author{R. Carrera\altaffilmark{1} and C. Gallart}

22: \affil{Instituto de Astrof\'{\i}sica de Canarias, Spain}

23: \email{rcarrera@iac.es}

24: \email{carme@iac.es}

25: 

26: \author{Eduardo Hardy\altaffilmark{1}}

27: \affil{National Radio Astronomy Observatory \altaffilmark{2}, Chile}

28: \affil{Departamento de Astronom\'{\i}a, Universidad de Chile\altaffilmark{3}, Chile}

29: \email{ehardy@nrao.edu}

30: 

31: \author{A. Aparicio}

32: \affil{Instituto de Astrof\'{\i}sica de Canarias, Spain}

33: \affil{Departamento de Astrof\'{\i}sica, Universidad de La Laguna, Spain}

34: \email{antapaj@iac.es}

35: 

36: \and

37: 

38: \author{R. Zinn}

39: \affil{Department of Astronomy, Yale University, USA}

40: 

41: \altaffiltext{1}{Visiting Astronomer, Cerro Tololo Inter-American Observatory.

42: CTIO is operated by AURA, Inc.\ under contract to the National Science

43: Foundation.}

44: \altaffiltext{2}{The National Radio Astronomy Observatory is a facility of the National Science 

45: Foundation operated under cooperative agreement by Associated Universities, Inc.} 

46: \altaffiltext{3}{Adjoint Professor} 

47: %% Mark off your abstract in the ``abstract'' environment. In the manuscript

48: %% style, abstract will output a Received/Accepted line after the

49: %% title and affiliation information. No date will appear since the author

50: %% does not have this information. The dates will be filled in by the

51: %% editorial office after submission.

52: 

53: \begin{abstract}

54: Ca II triplet spectroscopy has been used to derive stellar metallicities for individual stars in four LMC fields

55: situated at galactocentric distances of 3\arcdeg, 5\arcdeg, 6\arcdeg\@ and 8\arcdeg\@ to the north of the Bar. Observed metallicity

56: distributions show a well defined peak, with a tail toward 

57: low metallicities. The mean

58: metallicity remains constant until 6\arcdeg\@ ([Fe/H]$\sim$-0.5 dex), while for the outermost field, at

59: 8\arcdeg, the mean metallicity is substantially lower than in the rest of the disk ([Fe/H]$\sim$-0.8 dex). The combination

60: of spectroscopy with deep CCD photometry has allowed us to break the RGB age--metallicity degeneracy and

61: compute the ages for the objects observed spectroscopically. The obtained age--metallicity relationships for our four fields are statistically indistinguishable. We conclude that the lower mean metallicity in the outermost field is a consequence of it having a lower fraction of intermediate-age stars, which are more metal-rich than the older stars. The disk age--metallicity relationship is similar to that for clusters. However, the lack of objects with ages between 3 and 10 Gyr is not observed

62: in the field population. Finally, we used data from the literature to derive consistently the age--metallicity relationship of the bar. Simple chemical evolution models have been used to reproduce the observed age--metallicity

63: relationships with the purpose of investigating which mechanism has participated in the evolution of the disk and bar. We find that while the disk age--metallicity relationship is well reproduced by close-box models or models with a small degree of outflow, that of the bar is only reproduced by models with combination of infall and outflow. 

64: \end{abstract}

65: 

66: %% Keywords should appear after the \end{abstract} command. The uncommented

67: %% example has been keyed in ApJ style. See the instructions to authors

68: %% for the journal to which you are submitting your paper to determine

69: %% what keyword punctuation is appropriate.

70: 

71: %% Authors who wish to have the most important objects in their paper

72: %% linked in the electronic edition to a data center may do so in the

73: %% subject header.  Objects should be in the appropriate "individual"

74: %% headers (e.g. quasars: individual, stars: individual, etc.) with the

75: %% additional provision that the total number of headers, including each

76: %% individual object, not exceed six.  The \objectname{} macro, and its

77: %% alias \object{}, is used to mark each object.  The macro takes the object

78: %% name as its primary argument.  This name will appear in the paper

79: %% and serve as the link's anchor in the electronic edition if the name

80: %% is recognized by the data centers.  The macro also takes an optional

81: %% argument in parentheses in cases where the data center identification

82: %% differs from what is to be printed in the paper.

83: 

84: \keywords{galaxies: evolution ---

85: galaxies: stellar content --- Magellanic clouds}

86: 

87: %% From the front matter, we move on to the body of the paper.

88: %% In the first two sections, notice the use of the natbib \citep

89: %% and \citet commands to identify citations.  The citations are

90: %% tied to the reference list via symbolic KEYs. The KEY corresponds

91: %% to the KEY in the \bibitem in the reference list below. We have

92: %% chosen the first three characters of the first author's name plus

93: %% the last two numeral of the year of publication as our KEY for

94: %% each reference.

95: 

96: \section{Introduction}

97: 

98: Despite decades of work, there are still significant gaps in our

99: knowledge of the LMC's star formation and chemical enrichment

100: histories \citep{ol96}. This is motivated in

101: part by the vastness of its stellar populations and to our limitations

102: in observing sizable samples of stars. The age distribution of star

103: clusters is relatively well known \citep[e.g.][]{geisler97}: there

104: is an age interval between 10 and 3 Gyr  with almost no clusters. This so-called age-gap may

105: also correspond to an abundance gap \citep{ol91}, since the old

106: clusters are metal-poor while the young ones are relatively metal-rich. The star formation history (SFH)

107: of field stars is much less precisely known. Studies using {\itshape HST} data in small fields,

108: suggest that the SFH of the LMC disk has been more or less continuous,

109: with some increase in the star formation rate (SFR) in the last few Gyr

110: \citep[e.g.][]{Smecker-Hane02,castro01,holtzman99}. This contrasts with previous results (based on much

111: shallower data), which found a relatively young age (a few Gyr) for the

112: dominant LMC population \citep{hardy84,bertelli92,westerlund95,vallenari96}. The SFH of the

113: LMC can now be studied using sufficiently deep ground-based data (e.g.\ reaching the oldest main-sequence turn-off with good

114: photometric precision) in large areas and in different positions of the galaxy

115: \citep[e.g.][hereafter Paper I]{gshpz04,gshpz05}.

116: 

117: Less is known

118: about the LMC chemical enrichment history. The age--metallicity relation (AMR)

119: normally used for the LMC is defined by star clusters 

120: \citep[e.g.][]{ol91,geisler97,bica98,dirsch00}. All works based

121: on clusters obtain similar results.  The mean metallicity jumps from [Fe/H] $\sim -1.5$ for the oldest

122: clusters to [Fe/H] $\sim -0.5$ for the youngest ones. There were no studies on the field population AMR until the

123: last decade. \citet{dopita97}, from a study of $\alpha$-elements in planetary nebulae, obtained a

124: result qualitatively similar to that  found in the clusters although only ten objects were used. Later,

125: \citet{bica98}, using Washington photometry, and \citet{c00} and \citet{dirsch00}, using Str\"ongrem photometry,

126: obtained the metallicity of RGB stars in different positions of the LMC. \citet{c00} also observed 20 stars

127: with the infrared CaII triplet (CaT). All of them found that the age-gap observed in the clusters is not found in

128: the field population. More recently, \citet{c05} have obtained

129: stellar metallicities for almost 400 stars in the bar of the LMC, also using CaT lines. The metallicity for each star has been combined with its position in the color--magnitude diagram (CMD) to

130: estimate its age. The obtained AMR shows a similar behavior to that of the clusters for the oldest

131: population. However, while for the clusters the metallicity has increased over the last 2 Gyr, this has not happened in

132: the bar. Finally \citet{gratton04} measured the metallicity of about 100 bar RR Lyrae, obtaining an average

133: metallicity of [Fe/H] $=-1.48$.

134: 

135: Another point that it is necessary to investigate is the presence, or not, of an abundance gradient in the LMC.

136: From observations of clusters, \citet{ol91} and \citet{santos99} found no evidence for the presence of a radial

137: metallicity gradient in the LMC. The only evidence

138: for a radial metallicity gradient in the LMC cluster system was reported by \citet{kon93} 

139: based on six outer LMC clusters

140: ($\geq$ 8 kpc). \citet{hill95} were the first to report evidences of a gradient in the field 

141: population from high-resolution spectroscopy, using a sample of nine 

142: stars in the bar and in the

143: disk. They found that the bar is on average 0.3 dex more metal-rich than the 

144: disk population, but the disk

145: stars studied are located within a radius of 2\arcdeg. Subsequent work by \citet{cioni03}, 

146: detected that the C/M ratio between number of asymptotic giant branch stars of spectral 

147: types C and M increased when moving away from the bar and within

148: a radius  of 6$\fdg$7. As the C/M ratio is anticorrelated with metallicity, this 

149: increment implies a decrease in metallicity. Finally, an outward radial

150: gradient of decreasing metallicity was also found by \citet{alves04} from infrared 

151: CMD using the 2MASS survey.

152: 

153: We have obtained deep photometry with the Mosaic II CCD Imager on the CTIO 4m telescope in four disk fields at

154: different distances from the center of the LMC (RA=5$^h$23$^m$34$\fs$5, $\delta$=-69$\arcdeg$45'22``) with a quality

155: similar to that obtained by {\itshape HST} in more crowded areas (see Paper I). 

156: The position of these fields, together with a

157: description of their CMDs, are presented in Paper I, 

158: and detailed SFHs will be published in forthcoming papers. In the present investigation we 

159:  focus on obtaining stellar metallicities for a significant number 

160: of individual RGB stars in these four fields using spectra obtained with the HYDRA spectrograph at the CTIO

161: 4m telescope. In Section \ref{targetselection} we present our target selection. The observations and data reduction are presented in Section \ref{datareduction}. The radial velocities of the stars in our sample are obtained in Section \ref{radialvelocities}. In Section \ref{cat} we discuss the calculation of the CaT equivalent widths and the determination of metallicities. Section \ref{agedetermination} presents the method used to derive ages for each star by combining the information on their metallicity and position on the CMD. The analysis of the data is presented in Section \ref{analysis}, where the possible presence of a kinematically hot halo is discussed and the AMR and the SFH in each of our fields are described and compared. In the final section, the AMRs are compared to theoretical models and conclusions are drawn about the chemical evolution of the LMC.

162:  

163: \section{Target Selection\label{targetselection}}

164: 

165: Our starting point are deep CMDs of four $36'\times 36'$ fields located $\sim$3$\arcdeg$,

166: 5$\arcdeg$, 6$\arcdeg$ and 8$\arcdeg$ from the LMC center, which are presented and discussed in Paper I. All fields are located north of the

167: LMC bar, and their

168: CMDs reach the

169: oldest main-sequence turn-offs. A clear gradient exists in the amount of intermediate-age

170: and young populations from the inner to the outermost fields, in the sense that the 

171: young population is more prominent in the inner parts

172: (see Paper I for a more detailed discussion). 

173: 

174: In each field we have selected the stars to be observed with HYDRA in two windows in the CMD, which are plotted in Figure \ref{dcmbox}. The bluest color limit

175: has been selected to avoid upper red-clump and red supergiant stars and the lower magnitude limit to avoid including red-clump stars in the spectroscopic sample. The blue edge of the window is sufficiently blue to ensure that no metal 

176: poor

177: stars would be excluded (the position of the blue ridge corresponds

178: approximately with the position of $\sim$2 Gyr old stars with [Fe/H]=-3).

179: Similarly, the opposite edge is sufficiently red so that even old stars with

180: solar metalicities, or even higher, fit within the box. These selected stars

181: have been ordered from the brightest to the faintest ones, with no color restriction. This list has been

182: used as input for the configuration task of the instrument, which tries to optimize the number of allocated fibers. The first stars in the list, the brightest ones, have priority over the others.

183: 

184: \section{Observations and Data Reduction\label{datareduction}}

185: 

186: The observations were carried out in two different runs, in December 2002 and in

187: January 2005, at the CTIO 4 m telescope with the HYDRA multifiber spectrograph. We used

188: the KPGLD grating, which provides a central wavelength of

189: 8500 \AA, and a OG590 order-blocking filter. The physical pixels were binned 2 $\times$ 1 in the spectral direction yielding a

190: dispersion of 0.9 \AA /pix. HYDRA has a field of view of about 40 arcmin,

191: which on account of the high density of stars in our fields, allowed us to observe more than

192: 100 stars in each configuration, with the 140 available fibers.  Because of technical problems and bad weather,

193: on our first run we were able to observe only two

194: configurations located at 5$\arcdeg$ and 8$\arcdeg$ respectively. In the more successful second run,

195: one fiber configuration in each field was observed except for the closest one, where

196: we observed two configurations. Stars in two globular clusters, NGC 4590 and NGC 3201, were also observed. Those in the first were used as radial velocity standards, but also to 

197: compare the equivalent widths determined through the present set-up with those setups

198: used for the cluster observations reported in 

199: \citet[][hereafter Paper II]{carrera06}. Since the differences were 

200: negligible (see Figure 1 of paper II), the second cluster was included 

201: in the calibration of the CaT as metallicity indicator. We concluded that the LMC equivalent widths obtained with the present observational setup could be directly 

202: transformed into [Fe/H] using the calibration in paper II. Equivalent widths, magnitudes and radial velocities for the 702 stars observed in the LMC are listed in Table \ref{starobs}.

203: 

204: The data reduction was done following the procedure described in {\sl Knut Hydra notes}\footnote{http://www.ctio.noao.edu/spectrographs/hydra/hydra-knutnotes.html}.

205: First, cosmic rays were removed from the images using the 

206: IRAF\footnote{IRAF is distributed by the National Optical Astronomy

207: Observatory, which is operated by the Association of Universities for 

208: Research in Astronomy, Inc., under cooperative agreement with the

209: National Science Foundation.} Laplacian edge-detection

210: routine \citep{dokkun01}. All images were then bias- and 

211: overscan-subtracted, and trimmed. Flat exposures were taken during the day with a diffusing screen installed in the spectrograph to obtain

212: the so-called "milk-flat", which is used to correct the effect of the different sensitivities among pixels. It was applied to all images

213: using \textsl{ccdproc}. The spectra were extracted, flat-field

214: corrected (see bellow), and calibrated in wavelength using \textsl{dohydra}, a program developed

215: specifically to reduce the data obtained with this spectrograph. The lamp flats acquired

216: at the beginning of each configuration were used to define and trace each aperture. The sky flats were used to

217: eliminate the residuals after the flat-field correction with the milk-flat. Arcs, obtained before and after each

218: configuration, were used for

219: wavelength calibration.

220: 

221: \textsl{Dohydra} can also perform sky subtraction. However, we noticed that after sky

222: subtraction, there remained  significant sky line residues.  We therefore

223:  developed our own procedure to remove the contribution

224: of the sky lines from the object spectra. For a given configuration, we obtain an

225: average high signal--to--noise ratio (S/N) sky spectrum from all fibers placed on the sky. 

226: Before subtracting this averaged sky from each

227: star spectrum, we need to know the relation between the intensity of the sky in 

228: each fiber (which varies from fiber to fiber owing

229: to the different fiber responses) and the average sky. This relation is used as 

230: a weight (which may depend on wavelength) that

231: multiplies the average sky spectrum before subtracting it from each star. Our task 

232: minimizes the sky line residuals over the whole spectral region considered and 

233: allows very accurate removal of the sky emission lines. An example of raw, sky and final object

234:  is shown in Figure \ref{sky}.

235: 

236: We obtained between three and four exposures of 2700 s in each configuration. 

237: To minimize the contribution of

238: residuals arising from cosmic rays and bad pixels, all spectra of the same 

239: object were combined. Finally, each object spectrum was normalized with the

240: \textsl{continuum} task by fitting a polynomial, making sure to 

241: exclude all absorption lines in our wavelength range, such as the CaT lines themselves.

242:  

243: \section{Radial Velocities\label{radialvelocities}}

244: 

245: Our main purpose in obtaining radial velocities for our program stars is to 

246: reject possible foreground objects. However, radial velocities

247: can provide more useful information, as discussed in depth in Section \ref{kinematics}. 

248: The radial velocities were calculated with the \textsl{fxcor} task

249: in IRAF, which performs the cross-correlation described by \citet{td79} 

250: between the spectra of the target and selected templates of known radial

251: velocities. As templates, we used three stars in the globular cluster NGC 4590 (M68) 

252: with radial velocities obtained by \citet{geisler95}.

253: The final radial velocities for all the stars in our sample were obtained as the average 

254: of the velocities obtained with the three templates, weighted by the width of the corresponding correlation

255: peaks. Radial velocity histograms for each field are plotted in Figure \ref{vel_dist}. 

256: A Gaussian function was fitted in order to obtain

257: the central peak and the $\sigma$ of each

258: distribution (values listed in Table \ref{radialvelocity}). The velocity

259: distribution of the field  located 3\arcdeg\ from the center shows two peaks, which are marked in Figure

260: \ref{vel_dist}. Stars with radial velocity in the range

261: 170 $\leq V_r\leq$ 380 km s$^{-1}$ \citep{zhao03} are considered  LMC members. Only about 20 stars in each

262: configuration of the instrument have been excluded based on their radial velocity. These results are discussed in depth in Section \ref{kinematics}.

263: 

264: \section{CaT Equivalent Widths and Metallicity Determination\label{cat}}

265: 

266: The metallicity of the RGB stars is obtained following the procedure described in Paper II.

267: In short, the equivalent width is the line area normalized to the continuum. The continuum is

268: calculated from the linear fit to the mean value of each bandpass, defined to obtain the continuum position. We

269: have used the continuum and line bandpasses defined by \citet{cen01}, which are listed in Table \ref{bandastable}. The

270: equivalent widths are calculated from profile fitting using a Gaussian plus

271: a Lorentzian. This combination provides the best fit to line core and wings as discussed in Paper II. The CaT index is defined as the sum of the equivalent widths of the three CaT lines,

272: denoted as $\Sigma Ca$. The $\Sigma Ca$ calculated for each observed star and its uncertainty are given in

273: Table \ref{starobs}, together with the star magnitude and radial velocity. The reduced equivalent width,

274: $W'_I$, for each star has been calculated as the value of $\Sigma Ca$ at M$_I$=0, using the slope obtained in Paper II for the calibration

275: clusters in the M$_I$--$\Sigma Ca$ plane ($\beta_I$ = $-0.611$ \AA mag$^{-1}$).

276: 

277: In Paper II we obtained relationships between the reduced equivalent widths, $W'_V$ and $W'_I$, and

278: metallicity, on the

279: \citet{zw84}, \citet[hereafter CG97]{cg97} and \citet{ki03} scales. In this case, we will only use the

280: relationships obtained on

281: the CG97 metallicity scale, because it is the only one in which the metallicities of our open and globular calibration clusters have been obtained in a homogeneous way from high-resolution spectroscopy.

282: 

283: In order to verify that the $V$ and $I$ relationships give similar results, we have applied both to obtain the

284: metallicities of the stars observed in the field located at 5\arcdeg. The differences are on average $-$0.01

285: dex, and never larger than 0.15 dex for stars with $(V-I)\leq$ 2.5. For stars with colors larger than this, the

286: differences are on average 0.2 dex and never get

287: above 0.35 dex. The calibration stars used in Paper II have $(V-I)<2.5$, so the

288: metallicities calculated for the reddest stars are necessarily extrapolations of the 

289: relationships. On average, the relationship

290: based on M$_I$ yields metallicities slightly lower than those obtained from M$_V$, 

291: but always within the uncertainties. In what follows we will use

292: the relationship based on M$_I$, because: i) the photometric accuracy in $I$

293: is slightly better than in $V$, especially for the reddest

294: stars on the RGB; ii) the RGB is better resolved in the $I$ band; and iii) from a theoretical

295: point of view \citep{pont04},

296: it is likely that the relationship based on M$_I$ is less sensitive to age. In short, then, the metallicity for each star is given by:

297: 

298: \begin{equation}\label{metaleq}

299: [Fe/H]_{CG97}=-2.95+0.38\Sigma Ca+0.23M_I

300: \end{equation}

301: 

302: Binarity could affect the derived metallicity only in the unlikely case of the two stars in the system being RGB stars of a very similar mass. In this case the system would be $\sim$0.7 mag brighter than a single star. As the binarity might not affect the equivalent width of the CaT lines, the derived metallicity would be lower by $\sim$0.4 dex than the actual metallicity.

303: 

304: Figure \ref{misigmaca} shows the position of LMC radial velocity member stars in the 

305: M$_I$--$\Sigma$Ca plane for our

306: four fields, from the innermost one (top left) to the outermost field (bottom right). The star with $\Sigma Ca$ = 11 and M$_I$ = $-$2 in the field

307: situated at 5$\arcdeg$\@ has a well measured $\Sigma Ca$ and radial velocity 

308: V$_r$ = 264 km s$^{-1}$ (the mean V$_r$ of the sample is 278 $\pm$ 20 km

309: s$^{-1}$), so it seems to be an LMC RGB star with unusually large metallicity. However, 

310: a reliable metallicity cannot be obtained with the

311: present calibration, which is only valid for [Fe/H] $\leq$ +0.47. The extrapolation of our 

312: calibration gives [Fe/H] = +0.72 $\pm$ 0.4 for this

313: star.

314: 

315: The metallicity distribution of each field is shown in Figure

316: \ref{histo_fields} and will be discussed \emph{in extenso} in Section \ref{analisismetallicitydistribution}. In

317: Figure \ref{dcm_feh} the observed stars have been plotted  in

318: the color--magnitude diagram  employing different symbols as a function of their metallicity. 

319: It is important to highlight that there is no correlation between the

320: metallicity of the stars and their position in the color--magnitude diagram,

321: as was also found in other galaxies \citep[e.g.][]{pont04,c05,koch05}. This shows once more that the metallicities derived

322: from the position of the stars in the RGB are unreliable in systems with complex SFHs.

323: 

324: \section{Determination of stellar ages\label{agedetermination}}

325: 

326: For a given age, the position of the RGB on the color--magnitude diagram depends mainly on metallicity

327: in the sense that more metal-rich stars are redder than more metal-poor ones. At the same time, for a fixed

328: metallicity, older stars are also

329: redder than younger ones. If we take into account the logical assumption that younger stars 

330: are also more metal-rich, age may partly counteract the effect of

331: metallicity. The combination of both effects is the well known RGB age--metallicity 

332: degeneracy, which sets a limit on the amount of information that can be retrieved from the

333: RGB using only photometric data. However, if stellar metallicities are obtained independently from another source, 

334: such as spectroscopy, we should in principle be able to break the age--metallicity

335: degeneracy. \citet{pont04} and \citet{c05} have derived stellar ages from isochrones 

336: for stars whose metallicity had been previously

337: obtained from low-resolution spectroscopy. In our case, instead of

338: directly comparing  the position of the star in the CMDs with 

339: isochrones, we have obtained a relation for the age as a

340: function of color, magnitude and metallicity from a synthetic CMD. This allows to easily obtain ages for the large number of stars in our sample. 

341: 

342: The reference synthetic CMD has been computed using IAC-STAR\footnote{available on the Web

343: at http://iac-star.iac.es} \citep{aparicio04}. Its main input parameters are the SFR and

344: the chemical enrichment law, as a function of time. Since we are interested in a general relation, we have chosen a

345: constant SFR for the whole range of ages (13 $\geq$ (Age/Gyr) $\geq$ 0) and a chemical enrichment law such that a star of any age

346: can have any metallicity in the range $-2.5\leq$ [Fe/H] $\leq$ +0.5. Both the constant SFR and the chemical enrichment

347: law ensure that there is no age or metallicity bias. We have computed two

348: synthetic CMDs, one using the BaSTI

349: stellar evolution library \citep{pie04} and another  using the Padova library \citep{girardi02}

350:  respectively. The ranges of ages and metallicities

351: covered, as well as other important features of each library are listed in Table 1 of \citet{gallartza05}.

352: 

353: Since the uncertainty of the age assigned by stellar evolution models to bright AGB stars is large, we will only compute ages for RGB and AGB stars fainter than the RGB tip (M$_I\sim$-4 in Figure \ref{dcmbox})

354: . We have chosen the synthetic stars in the box below the tip of the RGB shown in

355: Figure \ref{dcmbox}. We have computed a polynomial relationship which gives

356: the age as a function of [Fe/H] $\equiv$\   log(Z/Z$_\odot$), (V-I) and M$_V$. In order to 

357: minimize the $\sigma$ and to improve the correlation coefficient of the relation, 

358: different linear, quadratic and cubic terms of each observed magnitude have been added. When the addition of

359: a term did not improve the relation we rejected it. The best combination of these

360: parameters is in the form of Equation \ref{rela} and the values obtained for each stellar evolution

361: library are shown in Table \ref{tableage}. Note that each stellar evolution library has

362:  its own age scale, and some differences might exist among them.

363: 

364: \begin{equation}\label{rela}

365: log(age)=a+b(V-I)+cM_V+d[Fe/H]+f(V-I)^2+g[Fe/H]^2+h(V-I)^3

366: \end{equation}

367:  

368: The uncertainties of each term on Equation \ref{rela} are given in Table \ref{tableage}. However, the way in which they

369: propagate into the error of the age is complex. For this reason a Monte Carlo test has been performed to

370: estimate age errors. The synthetic stellar population computed from BaSTI stellar evolution models has been

371: used to this purpose. That based on the Padua library would have produced similar results. The test consists

372: in computing, for each synthetic star, several age values for stochastically varying $[Fe/H]$, $(V-I)$ and $M_{V}$ according  to a gaussian probability

373: distribution of the corresponding $\sigma$ ($\sigma_{[Fe/H]}\sim$0.15 dex; $\sigma_{(V-I)}\sim$0.001 and

374: $\sigma_{M_V}\sim$ 0.001). The $\sigma$ value of the obtained ages provide an estimation of the age error when

375: Equation \ref{rela} is used. The values obtained for different age intervals are shown in Figure \ref{erroredad}. The error increases for older ages.

376: 

377: We have also investigated the accuracy with which the age of a real stellar population is measured. The

378: stars in open and globular clusters in Paper II, which ages are independently

379: known, are used for this purpose. We adopt

380:  the ages estimated by \citet{sw02} and \citet{swp04}, which are in a common scale, for a globular and open clusters, respectively. They are listed in Table 1

381: of Paper II. 

382: 

383: The age of each

384: cluster star has been estimated from Equation \ref{rela} assuming the distance modulus and reddening listed in Paper II. The metallicity of each

385: star has been calculated following the same procedure as for the LMC stars 

386: (see Section \ref{cat}). The age of each cluster has been obtained as the mean age derived from 

387: all the stars in each cluster and the quoted uncertainty is the standard deviation of this mean. Ages derived for each cluster from the BaSTI and Padova

388: relationships in Table \ref{tableage}, versus the reference values for each

389: cluster, have been plotted in Figure \ref{cluster}. The relationships saturate for ages older than 10 Gyr

390: because differences of 1 or 2 Gyr

391: produce only negligible variations of position in the 

392: CMD. In both cases, the 

393: age of clusters younger than 10 Gyr are well reproduced. Differences between the age derived with each model 

394: and the reference value are within the error bars, with the exception of cluster NGC 6819 for the BaSTI relationship. The large errorbar of the youngest cluster, NGC 6705, is due to the fact that the stars observed in this cluster are fainter that the synthetic stars used to derive the relationships. Therefore, we are extrapolating the relations to obtain the age of this cluster. As both models produce similar values, we adopted for simplicity the relationships derived from the BaSTI stellar library \citep{pie04}.

395: 

396: \section{Analysis\label{analysis}}

397: 

398: \subsection{Stellar kinematics of the LMC\label{kinematics}} 

399: 

400: In Figure \ref{kunkel} we have plotted the mean of the velocity distribution for each field 

401: \citep[including the][ bar field]{c05} as a function of

402: its Galactic longitude. The mean velocity changes from field to field due to the disk rotation

403: of the LMC. The radial velocities of carbon stars

404: derived by \citet{kunkel97} have also been plotted. Our data are in good agreement with their result,

405:  which is consistent with the presence of 

406: a rotational disk.

407: 

408: If a classical Milky Way-like halo existed in the LMC (i.e. old, metal-poor and with high velocity dispersion), a dependency between the velocity dispersion and metallicity/age of the stars in our sample might be observed. The first evidence of the

409: presence of a kinetically hot spheroidal population was reported by \citet{hughes91}. They found that the LMC

410: long period variables, related to an old stellar population, have a high velocity dispersion (33 km

411: s$^{-1}$),

412: with a low rotational component. \citet{miniti03}

413:  measured the kinematics of 43 RR Lyrae stars in the inner regions of the LMC and 

414:  found that the velocity dispersion of these stars

415: is

416: 53 $\pm$ 10 km s$^{-1}$, which they associated with the presence of a kinematically hot halo populated by old 

417: metal-poor stars. Some studies of the intermediate-age and old populations

418: have found that the velocity dispersion increases with age 

419: \citep[e.g.][]{hughes91, schommer92, graff00}. In fact \citet{graff00}, using C stars, have found the stars of the disk

420: belong to two populations: a young disk population containing 20\% of stars with a velocity dispersion of 8 km

421: s$^{-1}$, and an old disk population containing the remaining stars with a velocity dispersion of 22 km

422: s$^{-1}$.

423: 

424: We can check the presence of a hot halo with our sample stars. Assuming that stars with similar metallicities are of similar ages (see below), a possible dependency of velocity dispersion with metallicity could indicate whether different stellar populations have different kinematics.

425: 

426: The procedure has been the follow up. The mean velocity in each field has been subtracted from the radial velocity of each star to eliminate the rotational component of the disk. The total metallicity range has been divided into three bins. The velocity dispersion of stars in each metallicity bin is listed

427: in Table \ref{vrmetaltable} and plotted in Figure \ref{vrmetal}. The velocity dispersion for old and

428: metal-poor stars, 26.4 km s$^{-1}$, is slightly smaller than the value found by \citet{hughes91} for the old long period variables

429: ($\sigma$=33 km s$^{-1}$) and significantly smaller than the dispersion found in RR Lyrae stars

430: \citep[$\sigma\sim$53 km s$^{-1}$,][]{miniti03}. \citet{c05} found a velocity dispersion of $\sigma$ = 40.8 km

431: s$^{-1}$ for the most metal-poor stars in the bar, which is also higher than the value found here. The metal-poor stars in our sample seem to be members of the thick disk instead of the halo. In

432: short, from our data we have found no clear evidence for the presence of a kinematically hot halo populated by old stars.

433: 

434: \subsection{Metallicity distribution\label{analisismetallicitydistribution}}

435: 

436: The metallicity distribution for each field is shown in Figure \ref{histo_fields}. We have fitted a Gaussian to

437: each one in order to obtain its mean value and  dispersion (Table \ref{metallicitybin}). The mean value is constant at [Fe/H]$\sim$-0.5 dex up to

438: 8$\arcdeg$, where the mean metallicity decreases by about a factor of two ([Fe/H]=-0.8 dex). All

439: metallicity distributions are similar. They show a clear peak with a tail toward  low metallicities. We 

440: complement our sample with the results by \citet{c05}, who derived 

441: stellar metallicities in the bar in a similar way as here. 

442: As demonstrated in Paper II, their CaT index is equivalent to ours. Using their

443: measurements of the $\Sigma Ca$ we can calculate the metallicities from Equation \ref{metaleq}.

444: We have also fitted a Gaussian to the resulting

445: bar metallicity

446:  distribution to obtain its mean value. On average, the bar is slightly more 

447: metal-rich than the inne disk [Fe/H]$\sim$-0.4 dex).

448:   

449: \subsection{Age--metallicity relationships\label{agemetallicity}}

450: 

451: To understand the nature of the observed metallicity distributions, and to join insight on SFR

452: and the chemical enrichment history of the LMC fields, we have estimated the age of each star in our sample

453: following the procedure described in Section \ref{agedetermination}. These individual age determinations have a much 

454: larger uncertainty than the metallicity calculations. However, they are still 

455: useful because we are interested in the general trend rather

456: than in obtaining precise values for individual stars.

457: 

458: We assumed that the oldest stars 

459: have the same age as the oldest cluster in our

460: galaxy, for which we adopt 12.9 Gyr \citep[NGC 6426,][]{sw02}. This agrees with a 13.7$\pm$0.2 Gyr

461: old Universe \citep{spergel03} where the

462: first stars formed about 1 Gyr after the Big Bang. Another important point is the youngest age that we can find in our sample. According to stellar evolution models \citep{pie04}, we do not expect

463: to find stars younger than 0.8 Gyr in the

464: region of the RGB where we selected our objects. However, application of Equation \ref{rela} may result in ages younger

465: than this value. To avoid this contradiction, and taking into account that the 

466: age determination uncertainty for the younger stars is about 1 Gyr, we assigned an age of 0.8 Gyr to those

467: stars for which Equation \ref{rela} gives younger values. Finally, as the relationships used to estimate

468: the age have been calculated for stars below the tip of the RGB, only stars fainter than this point have been

469: used. 

470: 

471: The age of each star versus its

472: metallicity (i.e. the AMR) has been plotted in Figure \ref{amrfields} for each field. The many stars with the same age that appear both at the old (12.9 Gyr) and young (0.8 Gyr) limits are the consequence of the boundary conditions imposed on the ages, and the exact values should not be taken at face value. The age error in each age interval, computed in Section \ref{agedetermination}, is indicated in the top panel. The age distribution for each field has been plotted in inset

473: panels. The histogram is the age distribution without taking into account 

474: the age uncertainty. The solid line is the

475: same age distribution computed by taking into account the age error. To do this, a Gaussian probability

476: distribution is used to represent the age of each star. The mean and $\sigma$ of the Gaussian are the age obtained for the

477: star and its error, respectively. The area of each Gaussian is unity. In

478: stars near the edges, the wings of the distributions may extend further than the age physical limits. In such

479: cases, we truncated the wing and rescaled the rest of the distribution such that the area remains unity.

480: 

481: Stars brighter than M$_I$=-3.5 were not used in Paper II to obtain the metallicity calibration of $\Sigma$Ca (Equation \ref{metaleq}). However, we observed stars brighter than M$_I$=-3.5

482: in the LMC. Therefore, extrapolation of Equation \ref{metaleq} has been used to obtain the

483: metallicity of these stars, but only for those below the tip of the RGB. It is necessary to check whether this

484: has introduced any bias in the obtained AMR. Paper\ II demonstrated that

485: the sequences described by cluster stars used for the calibration in the M$_I$--$\Sigma Ca$ plane are not exactly

486: linear and  have a quadratic component. In the interval M$_I$=0 -- -3.5 the difference between the

487: quadratic and linear behaviors are negligible. However, for brighter stars, the deviation from the linear

488: behavior might be important, resulting in a possible underestimation of the metallicity of these stars. To check this point, in Figure \ref{testamr} we have plotted

489: as filled squares stars with magnitudes within the range of the calibration.

490: Open circles are stars whose metallicities have been obtained by extrapolating  

491: the relationship (-3.5$\geq M_I\geq$-4). The mean metallicity and its dispersion for several age intervals have been also plotted for both cases. The differences between both groups

492: are smaller than the observed dispersion and therefore we conclude that no strong bias is produced as a

493: consequence of extrapolating Equation \ref{metaleq} for bright stars. 

494: 

495: Figure \ref{amrfields} shows that the AMR is, within the

496: uncertainties, very similar for all fields. As expected, the most metal-poor stars in each field are also the oldest 

497: ones. A rapid chemical enrichment at a very early epoch is followed by a period of very slow metallicity

498: evolution until around 3 Gyr ago, when the galaxy started another period of chemical enrichment that is still

499: ongoing. Furthermore, the age histograms for the three innermost fields are similar, although

500: the total number of stars decreases when we move away from the centre. The outermost field has a lower

501: fraction of ''young'' (1--4 Gyr) intermediate-age stars. This indicates that its lower mean metallicity is

502: related to the lower fraction of intermediate-age, more metal-rich stars rather than to a different chemical

503: enrichment history (for example, a slower metal enrichment).

504:  

505: With the aim of obtaining a global AMR for the disk, in the following we will quantitatively address the

506: question of whether the AMRs of all our disk fields are

507: statistically the same. We have calculated the mean metallicity of the stars and its dispersion in six age intervals, for each field and for the combination of the four.

508: The values obtained are listed in Table \ref{testchi2}. To compare the values obtained 

509: in each field with those for the combined sample, we

510: performed a $\chi^2$ test such that $\chi^2=\sum_{i=1}^6\frac{(Z_i^{field}-Z_i^{comb})^2}{\sigma_i^2}$,

511: where $\sigma^2_i$ is the sum of the uncertainties squared in the age bin $i$ of the field and the combined

512: AMR. The last column in Table \ref{testchi2} shows the values $\chi^2_\nu=\chi^2/\sqrt{5}$. From these values we may conclude that the AMR of each field is equivalent to that of the combined sample to a probability of more than 99 per cent. 

513: 

514: Now we will compare the global AMR obtained for the disk with that observed in the bar.

515: The bar stellar ages have been obtained using the relation calculated in Section

516: \ref{agedetermination}. The age versus metallicity for

517: each star in the bar (\textsl{left}) and disk (\textsl{right}) have been plotted in Figure

518: \ref{amrbardisk}. Notice that with a different method and different stellar evolution models, the AMR obtained

519: here for the bar is quite similar to that obtained by \citet{c05}. We performed a $\chi^2$ test as before between the galaxy disk and bar AMRs. They are the same to within a probability of 90 per cent. However, a visual comparison would indicate that, while in the disk the metallicity has

520:   risen monotonically over the last few Gyr, this is not the case for

521: the bar. However, from the comparison of the mean metallicities and the errorbars for stars in this youngest bin (1--3 Gyr) it appears  that this difference is statistically significant.

522: 

523: The global disk AMR is qualitatively similar to that of the LMC cluster system \citep[e.g.][]{ol91,dirsch00},

524: except for the lack of intermediate-age clusters (Figure \ref{cumulos}). Clusters show a rapid

525: enrichment phase around 10 Gyr ago which is also observed in the disk and 

526: the bar. The second period of chemical enrichment in the last 3 Gyr is also observed both in the field and in the clusters. The field intermediate-age stars may have contributed to

527: the chemical enrichment at recent times. Note that the \citet{grocholski06} sample has a maximum 

528: metallicity [Fe/H]=-0.6, i.e. it does not seem to participate on the high 

529: metallicity tail at young ages that we and the other authors analysing clusters samples find. In 

530: the three works, there are a number of young clusters with low 

531: metallicity for their age, which \citet{bekki07} associate with metal-poor gas stripped from the SMC 

532: due to tidal interaction between the SMC, LMC and the Galaxy over the 

533: last 2 Gyr. A few stars also have low metallicity for their age in our 

534: sample, but they represent a minor contribution to the total.

535: 

536: \subsection{The Star Formation History}\label{sfhcal}

537: 

538: More information can be retrieved from the AMRs. In particular

539: we can also derive an approximate SFH

540: of the LMC if we can establish that the observed stars are representative of the total population. The number of stars in each age bin can be transformed into

541:  stellar total mass, after accounting for the

542:  number of stars with a given age that have

543: already died. To evaluate this correction 

544: we have computed a synthetic CMD using IAC-star 

545: \citep{aparicio04}. We have

546: assumed a constant SFR and used the relations derived for the chemical enrichment. As we have only observed a fraction of the total stars in the

547: region of the RGB, we have rescaled the result to the total

548: number of stars in this region.

549: 

550: In order to check whether the procedure used to select the stars observed

551: spectroscopically introduces any bias in the computed SFH, we have calculated a

552: synthetic CMD with a known chemical enrichment history and SFR. It was computed such that the number of stars

553: in the selection region was the same as in the field at 3\arcdeg. The input SFR as a function of age,

554: $\psi(t)$, is plotted as a

555: solid line in Figure \ref{pruebaconf}. To check that the

556: selection criteria did not introduce biases, we have selected objects from this synthetic CMD in the same way

557: as from the LMC fields (see Figure \ref{dcmbox}). We assigned random positions to each synthetic star and, using the HYDRA

558: configuration software, we computed 20 test configurations. The number of objects selected in each

559: configuration, about 115, is similar to the observed stars in the LMC fields. For each test configuration, we

560: calculated $\psi(t)$ following the procedure described above. The mean of the 20 total tests is the dashed line in

561: Figure \ref{pruebaconf}. The error bars are the standard deviation of the 20 tests. We repeated the same test, but obtaining the ages of each synthetic star using Equation \ref{rela}. The result is the dotted line in Figure \ref{pruebaconf}.

562: As we can see in this Figure, the recovered $\psi(t)$

563: are consistent with the real one. In half of the bins the differences are within the error bars and in the rest

564: they are within 3$\sigma$. 

565: 

566: The $\psi(t)$ derived from our four fields are shown in Figure \ref{hfecampos} (\textsl{solid line}), which has been computed from the age

567: distribution, taking into account the age error (see Figure \ref{erroredad}). The uncertainty  in the

568: computed $\psi(t)$ has been estimated as the square root of the SFR value for each age (\textsl{dotted lines}).

569: However, values of the SFR for different ages are not independent from each other because the integral for the full

570: age interval is a boundary condition of the solution. In other words, fluctuations above the best solution should be compensated by others below it, and a solution close to, for example, the upper (or the lower) dashed line is very unlike. As a comparison, the $\psi(t)$ computed

571: without taking into account the age error has also been plotted in 

572: Figure \ref{hfecampos}. All fields have a first episode of

573: star formation more than 10 Gyrs ago. Then, $\psi(t)$ decreases until $\sim$5 Gyr ago, when it rises again in the inner regions of the LMC (r$\leq$6\arcdeg). This enhancement of $\psi(t)$ is not observed in the outermost field.

574: 

575: We followed the same procedure to calculate $\psi(t)$ for the bar stars. The bar $\psi(t)$ is shown in Figure \ref{hfe}. In this case we do not

576: know the total number of objects in the region where the observed stars were selected. For this reason, we have

577: plotted the percentage, over the total, that the SFR represents in each age bin. In the right panel of the same

578: Figure, the $\psi(t)$ of the disk, obtained as a combination of the four fields in our sample, is shown. The bar $\psi(t)$ is 

579: similar to that derived in small fields from {\itshape HST} deep photometry 

580: \citep{holtzman99,Smecker-Hane02}, as indeed is expected since the {\itshape HST} fields

581: are in the same region in which the RGB stars were observed. Both in the bar and in the

582: disk, the galaxy started forming stars more than 10 Gyr ago, although

583: in the bar the initial enhancement of $\psi(t)$ is not observed. For intermediate ages, the SFR was low until about 3 Gyr ago, when another increase in the SFR, which is particularly intense in 

584: the bar, is

585: observed. This is consistent with previous investigations

586: \citep[e.g.][]{hardy84,Smecker-Hane02} and with the SFR derived from the clusters. The cluster age gap

587: coincides with the age interval in which star formation has been less efficient.

588:   

589: \subsection{A Chemical Evolution Model for the LMC\label{modelsec}}

590: 

591: In what follows we use the AMR to obtain information on the physical

592: parameters governing the chemical evolution of the LMC.  As a first

593: approximation, we will try to reproduce the derived relations with

594: simple models.

595: 

596: Under the assumption of instantaneous recycling approximation,

597: following \citet{tinsley80} and \citet{peimbert94}, the

598: heavy-element mass fraction in the ISM, $Z$, evolves via:

599: 

600: \begin{equation}\label{variaz}

601: \mu\frac{dZ}{d\mu}=\frac{y(1-R)\psi+(Z_f-Z)f_I}{-(1-R)\psi+(f_I-f_O)(1-\mu)}

602: \end{equation}

603: 

604: where $f_I$ and $f_O$ are the inflow and outflow rates respectively,

605: $Z_f$ is the metallicity of the infall gas, $y$ denotes the yield,

606: $R$ is the mass fraction returned to the interstellar medium by a

607: generation of stars, relative to the mass locked in stars and stellar

608: remnants in that generation, $\psi$ is the SFR and $\mu$ is defined

609: as $\mu=M_g/M_b$, where M$_g$ is the gas mass and M$_b$ is the total

610: baryonic mass, i.e., the mass participating in the chemical evolution

611: process. In general, $Z, \mu, \psi, f_I$ and $f_O$ are explicit

612: functions of time, while $R$ and $y$ may change according to

613: characteristics of the stellar population like metallicity and initial

614: mass function, but are usually assumed to be constant for a given

615: population.

616: 

617: When there are no gas flows in and out of the system ($f_I=f_O=0$),

618: the model is called a closed-box model and the solution to Equation

619: \ref{variaz} provides the variation of $Z(t)$ as a function of

620: $\mu(t)$ via

621: 

622: \begin{equation}

623: Z(t)=Z_i+y\ln\mu(t)^{-1}

624: \end{equation}

625: 

626: In an infall scenario, in which gas flows into the system at a rate

627: proportional to the amount of star formation, $f_I=\alpha(1-R)\psi$,

628: $\alpha$ being a free parameter, we can integrate Equation

629: \ref{variaz} for $\alpha\neq1$, assuming that $f_O=0$, and obtaining:

630: 

631: \begin{equation}

632: Z(t)=Z_i+\frac{y+Z_f\alpha}{\alpha}\left\lbrack 1-

633: \left(\alpha-\frac{1-\alpha}{\mu(t)}\right)^{-\alpha/(1-\alpha)}\right\rbrack 

634: \end{equation}

635: 

636: For $\alpha$=1, the solution of Equation \ref{variaz} is

637: 

638: \begin{equation}

639: Z=Z_i+y\left\lbrack 1-e^{\left(1-\mu^{-1}\right)}\right\rbrack

640: \end{equation}

641: 

642: In an outflow scenario, in which gas scapes from the system at a rate

643: $f_O=\lambda(1-R)\psi$, $\lambda$ being a free parameter, we obtain

644: from Equation \ref{variaz} assuming $f_I=0$, and for ($\lambda\neq1$): 

645: 

646:  \begin{equation}

647: Z(t)=Z_i+\frac{y}{\lambda-1}\ln\left\lbrack

648: \frac{\lambda-1}{\mu(t)}-\lambda\right\rbrack 

649: \end{equation}

650: 

651: Finally, we will consider a model combination of inflow and outflow,

652: and assume that the system gas flow is proportional to the mass that

653: has taken part in the star formation: $f=(\alpha-\lambda)(1-R)\psi$,

654: where we can define $\beta=\alpha-\lambda$. In this case, we can

655: integrate Equation \ref{variaz} for $\beta\neq$1, to obtain

656: 

657: \begin{equation}

658: Z(t)=Z_i+\frac{y+Z_f\alpha}{\alpha}\left\lbrack

659: 1-\left(\beta-\frac{1-\beta}{\mu(t)}\right)^{-\beta/(1-\beta)}\right\rbrack 

660: \end{equation}

661: 

662: In summary, the chemical evolution of the galaxy is given by the yield $y$, the initial metallicity of the gas Z$_i$, the fraction of

663: gas mass $\mu(t)$, which is a function of $\psi(t)$, and $R$, the mass

664: fraction returned to the interstellar medium. In the case of

665: inflow and/or outflow, also by the additional parameters $\alpha$ and

666: $\lambda$.

667: 

668: We can now apply the former relations to the LMC. In the previous

669: section we have derived $\psi(t)$ for the LMC bar and disk, the latter

670: being obtained from the combination of the four fields in our

671: sample. $\mu$ is computed explicitly as a function of time from $\psi(t)$, the balance of gas flowing to and from the system, and using as a boundary condition the current LMC gas fraction. In the case of the LMC disk, we have assumed the value derived by

672: \citet{kim98} from observations of H {\footnotesize I}. These authors

673: derived $M_g=5.2\times10^8 M_{\odot}$ and estimated that the total

674: mass of the disk is $M_b=2.5\times10^9 M_{\odot}$. With these values,

675: we find the current gas fraction in the disk, $\mu_f=0.21$. For the

676: bar we have assumed the value of $\mu=0.08$ given by

677: \citet{westerlund90}. For the yield, we have assumed the value

678: obtained by \citet{pagel95} for the solar neighborhood,

679: $y=0.014$. Assuming the \citet{scalo86} initial mass function,

680: \citet{maeder93} calculated a returned fraction associated to this

681: yield of $R=0.44$. Putting all these ingredients together, we will now

682: explore several scenarios in order to reproduce the AMR of the disk

683: and the bar.

684: 

685: In Figure \ref{modeldisk} we have plotted the age and metallicity of

686: each star in the disk, together with the mean metallicity and its

687: dispersion, for stars in each age bin. In this figure, chemical

688: evolution models have been superposed for different scenarios and

689: parameter assumptions (see the figure caption). In the case of the infall models (\textsl{green dashed lines}), we have assumed a zero-metallicity infalling gas, in order to obtain the current

690: metallicity. They do not reproduce

691: the general behavior of the observed AMR. The disk AMR is well reproduced by outflow models (\textsl{blue dot-dashed lines}) with a relatively large range of $\lambda$. Finally, the combination of

692: inflow and outflow (\textsl{pink and cyan long--short dashed lines}) models also reproduce the metallicity

693: tendency (in this case the model that best matches the observed data

694: has $\alpha$ = 0.2 and $\lambda$ = 0.05). The previous models have

695: been computed assuming the yield observed in the solar

696: neighborhood which is probably too large for the LMC. With a smaller yield, $y=0.008$ (\textsl{red thick solid line}), the observed

697: AMR can be reproduced by a closed-box model. This yield would correspond to

698: stars with $Z=0.001$ \citep[see][for details]{maeder93}, which is

699: slightly low compared with the mean LMC metallicity. Bursting and smooth models computed by \citet{pagel98} have been

700: plotted as comparison (\textsl{brown lines}). The bursting model was computed assuming two burst of star formation, one

701: about 12 Gyr ago and a strong one about 3 Gyr ago, and some fraction of infall and outflow. Note that their

702: predicted values agree with our mean metallicity for each age bin, within the errorbars, specially at old and

703: intermediate--age. It also qualitatively agrees with the episode of faster chemical evolution with started

704: around $\sim$2 Gyr ago. The smooth model was computed assuming a constant star formation rate and does not match with the observations.

705: 

706: In the case of the bar (Figure \ref{modelbar}), the very slowly rising 

707: metallicity, almost constant over the last few Gyr, is best

708: reproduced by the combination of models with inflow and outflow with

709: infall parameter $\alpha$ = 1.2 and $\lambda$ = 0.4-0.6 (\textsl{pink long--short dashed lines}). Infall

710: models (\textsl{green dashed lines})  with $\alpha$ = 0.4-0.6 marginally reproduce the observed trend, but they predict higher metallicities than observed in the last few Gyr, unless a smaller yield would be

711: assumed. Finally, outflow and closed-box models predict a rise of metallicity in the last few Gyr more steep than observed. The \citet{pagel98} bursting model agrees worse with the bar AMR than with that of the disk, specially in the youngest age bins.

712: 

713: \section{Conclusions} \label{conclusions}

714: 

715: Using infrared spectra in the CaT region, we have obtained metallicities and radial velocities for a sample of stars in four LMC fields.

716: Metallicities have been calculated using the relationships between the equivalent width of the CaT lines, $\Sigma Ca$, and metallicity derived in

717: Paper II. In addition, we have estimated the age of each star using a relationship derived from a synthetic CMD which,

718: from the color, magnitude and metallicity of a star, allow us to estimate its age. The main results of this paper are:

719: \begin{itemize}

720: \item The velocity distribution observed in each field agrees with the rotational thick disk kinematics of the

721: LMC. The velocity dispersion is slightly larger for the most metal-poor stars. However, the  values obtained do not indicate

722: the presence of a kinematically hot halo.

723: \item The metallicity distribution of each field has a well defined peak with a tail toward low

724: metallicities. The mean metallicity is constant until the field at 6\arcdeg\@ ([Fe/H]$\sim$-0.5 dex), and  is a factor two more

725: metal-poor for the outermost field ([Fe/H]$\sim$-0.8 dex).

726: \item The AMR observed in each disk field is compatible with a single global relationship for the disk. We conclude that the outermost field is more metal-poor on average because it contains a lower fraction of relatively young stars (age$\leq$5 Gyr), which are also more metal-rich.

727: \item The disk AMR shows a prompt initial chemical enrichment. Subsequently, the metallicity increased very

728: slowly until about 3 Gyr ago, when the rate of metal enrichment increased again. This AMR is

729: similar to that of the cluster system, except for the lack of clusters with ages between 3 and 10 Gyr. The

730: recent fast enrichment observed in the disk and  in the cluster system is not observed in the bar.

731: \item The $\psi(t)$ of the three innermost fields show a first episode of star formation until about 10 Gyr ago, followed by a

732: period with a low SFR until $\sim$5 Gyr ago, when the SFR increases, and reaches its highest values $\sim$2-3 Gyr ago. The outermost field does not show the recent increase of SFR. The second main episode is also observed, and is more prominent in the bar, where an increased SFR at old ages is not observed. The lower SFR between 5 and 10 Gyr ago is probably related

733: to the age-gap observed in the clusters.

734: \item Under the assumption of a solar yield, the disk AMR is well reproduced either by a chemical evolution model with outflow with

735: $\lambda$ between 1 and 2, so the disk losses the same amount of gas that has taken part in the star formation, or by models combining infall and outflow with $\alpha$=0.2 and

736: $\lambda$=0.05, which means that the galaxy is almost a closed-box system. With

737: a smaller yield, the AMR could also be reproduced with a closed-box

738: model. The bar AMR is well reproduced by models with a combination of inflow and outflow with

739: $\alpha$=1.2 and $\lambda$ between 0.4 and 0.5. This suggests that the amount of infalling gas was larger than the amount that participated in the star formation in the bar, and also that the amount of gas that escaped the bar was 50\% of the total that participated in star formation.

740: \end{itemize}

741: 

742: \section{Discussion} \label{discussion}

743: The main result in this paper is that all our fields, covering

744: galactocentric radius from 3\arcdeg\@ to 8\arcdeg\@ from the center (2.7 to 7.2 kpc) share a very similar AMR. The mean metallicity is very

745: similar in the three innermost fields ([Fe/H] $\simeq$-0.5 dex), and it

746: is a factor of $\simeq 2$ smaller in the outermost field ([Fe/H] $\simeq$-0.8 dex). Because the

747: AMRs are the same, this has to be related with the lower fraction of

748: relatively young stars (age $\le$ 5 Gyr), which are also more metal-rich, in the outer part. We find, therefore, a change in the age

749: composition of the disk population beyond a certain radius ($\simeq$ 6

750: kpc; note however that in this study we are only sensitive to

751: populations older than $\simeq$ 0.8 Gyr, and that differences among

752: fields for younger ages could also exist), while the chemical

753: enrichment history seems to be shared by all fields.

754: 

755: The comparison of the AMR observed in the disk with simple chemical evolution models suggests 

756: that the LMC is most likely losing some of its gas. This is in agreement with the work by 

757: \citet{nidever07} which suggests that the main contribution to the Magellanic Stream comes from the 

758: LMC instead of from the SMC as it was believed until now \citep[e.g.][]{putman03}. Alternatively, the LMC 

759: could be an almost closed-box system with small gas exchanges ($\alpha$=0.2, $\lambda$=0.05). This could be in

760: agreement with the models by \citet{bekki07} which suggest that the LMC has received metal-poor gas 

761: from the SMC in the last 2 Gyr. In our field we observed some stars with a low metallicity to which we assign an age younger than 2 Gyr. However, the uncertainty in the age determination is large.

762: 

763: We have also obtained the AMR of the LMC bar from the data by \citet{c05}. The

764: bar AMR differs from the one of the disk in the last 5 Gyr: while in

765: the disk the metallicity has increased in this time, in the bar it has

766: remained approximately constant. This feature is best reproduced by

767: models combination of outflow and a relatively large infall of pristine gas.  Models with a smaller yield,

768: but with an infall of previously enriched gas also reproduce the

769: observed bar AMR. This would be in agreement with the prediction that a typical bar instability pushes the gas of the disk towards the center \citep{sellwood93}, so the infall gas is expected to be previously pre-enrichment by the disk chemical evolution. This would be also the case in the scenario suggested by \citet{bekki05}, in which the bar would have formed from disk material as a consequence of tidal interactions between the LMC, the SMC and the Milky Way about 5 Gyr ago, the moment when the bar AMR differs from that of the disk.  Also, a bar is expected to destroy any metallicity gradient within a certain radius, as it is observed. 

770: 

771: \citet{gshpz04} derived the LMC surface brightness profile using

772: deep resolved star photometry of the four fields in the current

773: spectroscopic study, and found that it remains exponential to a radius

774: of 8\arcdeg\@ ($\simeq$ 7 kpc), with no evidence of disk

775: truncation. Combining this information with that on the deep CMD of

776: the outermost field, which contains a large fraction of

777: intermediate-age stars, they concluded that the LMC disk extends (and

778: dominates over a possible halo) at a distance of at least 7 kpc from

779: its center.  In the present study, the kinematics of the stars in

780: these four fields confirm this conclusion: the velocity dispersion of

781: all four fields is similar, around 20 km/sec. If the stars are binned

782: by metallicity, the velocity dispersion of the most metal-poor bin is

783: slightly larger than that of the metal-richer bins ($\sigma_V

784: \simeq 25$ km/sec as opposed to $\simeq 20$ km/sec) but still not

785: large enough to indicate the presence of a halo, even one formed as a

786: consequence of the interaction with the Galaxy. In this case, \cite{bekki04} predicted a velocity dispersion $\simeq$ 40 km/s at a

787: distance of 7.5 kpc from the LMC center (similar to the distance of

788: our outermost field). Other authors have also failed to find a

789: kinematically hot halo \citep{freeman83, schommer92, graff00,

790: zhao03}. The first evidence of the presence of a kinetically hot, old

791: spheroidal population was reported by \citet{hughes91} using a sample

792: of long period variables, which are related to an old stellar

793: population. Recently, \citet{miniti03} observed spectroscopically a

794: sample of 43 bar RR Lyrae stars, and obtained a large velocity dispersion

795: (53 $\pm$ 10 km s$^{-1}$) for them. It is possible that our failure

796: (and that of other authors) to find evidence of a hot stellar halo is

797: related with a low contrast of the halo population with respect to the

798: disk one, even at large galactocentric radius as our outermost field.

799: 

800: \acknowledgments

801: 

802: AA, CG, and RC acknowledge the support from the Spanish 

803: Ministry of Science and Technology (Plan Nacional de Investigaci\'on Cient\'{\i}fica,

804: Desarrollo, e Investigaci\'on Tecnol\'ogica, AYA2004-06343) and from 

805: the Instituto de Astrof\'{\i}sica de Canarias (grants P3/94 and

806: 3I1902). RZ acknowledges the support of National Science Foundation grant AST05-07362. This work has made use

807: of the IAC-STAR Synthetic CMD computation code. IAC-STAR is supported and maintained by the computer division

808: of the Instituto de Astrof\'{\i}sica de Canarias.

809: 

810: %% To help institutions obtain information on the effectiveness of their

811: %% telescopes, the AAS Journals has created a group of keywords for telescope

812: %% facilities. A common set of keywords will make these types of searches

813: %% significantly easier and more accurate. In addition, they will also be

814: %% useful in linking papers together which utilize the same telescopes

815: %% within the framework of the National Virtual Observatory.

816: %% See the AASTeX Web site at http://www.journals.uchicago.edu/AAS/AASTeX

817: %% for information on obtaining the facility keywords.

818: 

819: %% After the acknowledgments section, use the following syntax and the

820: %% \facility{} macro to list the keywords of facilities used in the research

821: %% for the paper.  Each keyword will be checked against the master list during

822: %% copy editing.  Individual instruments can be provided in parentheses,

823: %% after the keyword, but they will not be verified.

824: 

825: Facilities: \facility{CTIO(HYDRA)}.

826: 

827: \begin{thebibliography}{}

828: \bibitem[Alves(2004)]{alves04} Alves, D. R. 2004, \nar, 49, astro-ph0408336

829: %\bibitem[Aparicio, Carrera \& Mart\'{\i}nez-Delgado(2001)]{aparicio01} Aparicio, A., Carrera, R., \& Mart\'{\i}nez-Delgado, D. 2001, \aj, 122, 2524

830: \bibitem[Aparicio \& Gallart(2004)]{aparicio04} Aparicio, A., \& Gallart, C. 2004, \aap, 128, 1465

831: %\bibitem[Armandroff \& Zinn(1988)]{az88} Armandroff, T. E., \& Zinn, R. 1988, \aj, 96, 92

832: %\bibitem[Armandroff \& Da Costa (1991)]{adc91} Armandroff, T. E. \& Da Costa, G. S. 1991, \aj, 101, 1329

833: \bibitem[Bekki et al.(2004)]{bekki04}Bekki, K., Couch, W. J., Beasley, M. A., Forbes, M. A., Chiba, M., \& Da Costa, G. S. 2004, \apj, 610,

834: L93

835: \bibitem[Bekki \& Chiba(2005)]{bekki05}Bekki, K., \& Chiba, M., 2005, \mnras, 356, 680 

836: \bibitem[Bekki \& Chiba(2007)]{bekki07}Bekki, K., \& Chiba, M., 2007, \mnras, in press

837: %\bibitem[Benedict et al.(2002)]{benedict02}Benedict, G. F., McArthur, B. E., Fedrick, L. W., Harrison, T. E., Slesnick, C. L., Rhee, J.,

838: %Patterson, R. J., Skrutskie, M. F., Franz, O. G., Wasserman, L. H., Jefferys, W. H., Nelan, E., van Antena, W., Shelas, P. J., Hemenway,

839: %P. D., Duncombe, R. L., Story, D., Whipple, A. L., \& Bradley, A. J. 2002, \aj, 121, 1695

840: \bibitem[Bertelli et al.(1992)]{bertelli92}Bertelli, G., Mateo, M., Chiosi, C. \& Bressan, A. 1992, \apj, 388, 400

841: %\bibitem[Besla et al.(2007)]{besla07}Besla, C., Kallivayalil, N., Hernquist, L., Robertson, B., Coz, T.J., van

842: %der Marel, R. P. \& Alcock, C. 2007, \apj, submitted

843: \bibitem[Bica et al.(1998)]{bica98} Bica, E., Geisler, D., Dottori, H., Clari{\'a}, J.~J., Piatti, A.~E., \& Santos, J.~F.~C., Jr.\ 

844: 1998, \aj, 116, 723

845: \bibitem[Carrera et al.(2007)]{carrera06}Carrera, R., Gallart, C., Pancino, E. \& Zinn, R. 2007, \aj, 134, 1298

846: (Paper II)

847: \bibitem[Carretta \& Gratton (1997)]{cg97} Carretta, E., \& Gratton, R. G. 1997, \aaps, 121, 95 (CG97)

848: \bibitem[Castro et al.(2001)]{castro01}Castro, R., Santiago, B. X., Gilmore, G. F., Beaulieu, S. \& Johnson, R. A. 2001, \mnras, 326, 333

849: \bibitem[Cenarro et al.(2001)]{cen01} Cenarro, A. J., Cardiel, N., Gorgas, J., Peletier, R. F., Vazdekis, A., \& Prada, F. 2001, \mnras, 326,

850: 959

851: \bibitem[Cioni \& Habing(2003)]{cioni03} Cioni, M.-R. L., \& Habing, H. J. 2003, \aap, 402, 133 

852: \bibitem[Cole, Smecker-Hane, \& Gallagher (2000)]{c00} Cole, A. A., Smecker-Hane, T. A., \& Gallagher III, J. S. 2000, \aj, 120, 1808 

853: %\bibitem[Cole et al.(2004)]{c04} Cole, A. A., Smecker-Hane, T. A., Tolstoy, E., Bosler, T. L., \& Gallagher III, J. S. 2004, \mnras, 347, 367

854: \bibitem[Cole et al.(2005)]{c05}Cole, A. A., Tolstoy, E., Gallagher III, J. S. \& Smecker-Hane, T. A. 2005,

855: \aj, 129, 1465 

856: \bibitem[Dirsch et al.(2000)]{dirsch00}Dirsch, B., Richtler, T., Gieren, W.P. \& Hilker, M. 2000,

857: \aap, 360, 133

858: \bibitem[Dopita et al.(1997)]{dopita97} Dopita M. A., Vassiliadis, E., Wood, P.R., Meatheringhan, J. S., Harrington, J. P., Bohlin, R. C.,

859: Ford, H.C., Stecher, T. P., \& Maran, S. P. 1997, \apj, 474, 188

860: \bibitem[Freeman, Illingworth, \& Oemler(1983)]{freeman83}Freeman, K. C., Illingworth, G., \& Oemler Jr, A. 1983, \apj, 272, 488

861: %\bibitem[Gallart et al.(1999)]{gallart99}Gallart, C., Freedman, W. L., Aparicio, A., Bertelli, G., \& Chiosi, C. 1999, \aj, 118, 2245

862: \bibitem[Gallart et al.(2004)]{gshpz04}Gallart, C., Stetson, P. B., Hardy, E., Pont, F., \& Zinn, R. 2004, \apj, 614, L109

863: \bibitem[Gallart, Zoccali \& Aparicio (2005)]{gallartza05}Gallart, C., Zoccali, M., \& Aparicio, A. 2005, \araa, 43, 387 

864: \bibitem[Gallart et al.(2007)]{gshpz05}Gallart, C., Stetson, P. B., Pont, F., Hardy, E., \& Zinn, R. 2007, in

865: preparation, (Paper I)

866: \bibitem[Geisler et al.(1995)]{geisler95}Geisler, D., Piatti, A. E., Clari\'{a}, J. J., \& Minniti, D. 1995, \aj, 109, 605

867: \bibitem[Geisler et al.(1997)]{geisler97}Geisler, D., Bica, E., Dottori, H., Clari\'{a}, J. J., Piatti, A. E.,  \& Santos Jr., J. F. C.

868: 1997, \aj, 114, 1920

869: 

870: \bibitem[Girardi et al.(2002)]{girardi02}Girardi, L., Bertelli, G., Bressan. A., Chiosi, C., Groenewegen, M: A. T., Marigo, P., Salasnich, B.,

871: \& Weiss, A. 2002, \aap, 391, 195

872: \bibitem[Gratton et al.(2004)]{gratton04} Gratton, R. G., Bragaglia, A., Clementini, G., Carretta, E., Di Fabrizio, L., Maio, M., \&

873: Taribello, E. 2004, \aap, 421, 937

874: \bibitem[Graff et al.(2000)]{graff00}Graff, D. S., Gould, A. P., Suntzeff,

875: N. B., Schommer, R. A., \& Hardy, E. 2000, \apj, 540, 211

876: \bibitem[Grocholski et al.(2006)]{grocholski06}Grocholski, A. J., Cole, A. A., Sarajedini,

877: A., Geisler, D., \& Smith, V. V. 2006, \aj, 132, 1630

878: \bibitem[Hardy et al.(1984)]{hardy84}Hardy, E., Buonanno, R., Corsi, C. E., Janes, K. A. \& Schommer, R. A. 1984, \apj, 278, 592 

879: \bibitem[Hill, Andrievsky, \& Spite(1995)]{hill95} Hill, V., Andrievsky, S., \& Spite, M. 1995, \aap, 293, 347

880: \bibitem[Holtzman et al.(1999)]{holtzman99}Holtzman, J. A., Gallagher, J. S., III, Cole, A. A., Mould, J. R., Grillmair, C. J., Ballester, G. E., Burrows, C. J., Clarke,

881: J. T., Crisp, D., Evans, R. W., Griffiths, R. E., Hester, J. J., Hoessel, J. G., Scowen, P. A., Stapelfeldt, K. R., Trauger, J. T. \&

882: Watson, A. M., 1999, \aj, 118, 2262 

883: \bibitem[Hughes, Wood, \& Reid(1991)]{hughes91} Hughes, S. M. G., Wood, P. R., \& Reid, N. 1991, \aj, 101, 1304

884: %\bibitem[Irwin \& Tolstoy(2002)]{irwint02} Irwin, M., \& Tolstoy, E. 2002, \mnras, 336, 643

885: \bibitem[Kim et al.(1998)]{kim98}Kim, S., Staveley-Smith, L., Dopita, M. A., Freeman, K. C., Sault, R. J., Kesteven, M. J. \& McConnell, D.

886: 1998, \apj, 503, 674

887: \bibitem[Koch et al.(2006)]{koch05}Koch, A., Grebel, E. K., Wyse, R. F. G., Kleyna, J. T., Wilkinson, M. I., Harbeck, D. R., Gilmore, G. F.,

888: \& Evans, N. W.. 2006, \aj, 131, 895

889: \bibitem[Kontizas, Kontizas, \& Michalitsiano(1993)]{kon93}Kontizas, M., Kontizas, E., \& Michalitsiano, A. G. 1993, \aap, 269, 107

890: \bibitem[Kraft \& Ivans(2003)]{ki03}Kraft, R. P., \& Ivans, I. I. 2003, \pasp, 115, 143

891: \bibitem[Kunkel et al.(1997)]{kunkel97}Kunkel, W. E., Demers, S., Irwin, M. J., \& Albert, L. 1997, \apj, 488, L129

892: \bibitem[Maeder(1993)]{maeder93} Maeder, A. 1993, \aap, 268, 833 

893: \bibitem[Minniti et al.(2003)]{miniti03} Minniti, D., Borissova, J., Rejkuba, M., Alves, D. R., Cook, K. H. \&

894: Freeman, K. C., 2003, Science, 301, 5639

895: %\bibitem[Nikolaev \& Weinberg(2000)]{nikolaev00} Nikolaev, S., \& Weinberg, M. D. 2000, \aj, 542, 804 

896: \bibitem[Nidever et al.(2007)]{nidever07} Nidever, D. L., Majewski, S. R. \& Butler Burton, W. 2007, \apj,

897: submitted 

898: \bibitem[Olszewski et al.(1991)]{ol91} Olszewski, E. W., Schommer, R. A., Suntzeff, N. B., \& Harris, H., C. 1991, \aj, 101, 515

899: %\bibitem[Olszewski et al.(1993)]{ol93} Olszewski, E. W. 1993 in ASP Conf. Ser. 48, The Globular Cluster-Galaxy Connection, ed. G. H. Smith \&

900: %J. P. Brodie (San Francisco: ASP), 351

901: \bibitem[Olszewski et al.(1996)]{ol96} Olszewski, E. W., Suntzeff, N. B., \& Mateo, M. 1996, \araa, 34, 511

902: %\bibitem[Peterson \& Green(2002)]{pg02} Peterson \& Green

903: \bibitem[Pagel \& Tautvaisiene(1995)]{pagel95} Pagel, B.~E.~J., \& Tautvaisiene, G.\ 1995, \mnras, 276, 505

904: \bibitem[Pagel \& Tautvaisiene(1998)]{pagel98} Pagel, B.~E.~J., \& Tautvaisiene, G.\ 1998, \mnras, 299, 535

905: \bibitem[Peimbert, Colin \& Sarmiento(1994)]{peimbert94}Peimbert,\ M.,\ Colin,\ P.\ \&\ Sarmiento,\ A. 1994, in {\sl "Violent Star Formation From 20 Doradus to QSOs"} ed. G. Tenorio-Tagle. Cambridge University Press. 

906: \bibitem[Pietrinferni et al.(2004)]{pie04}Pietrinferni, A., Cassisi, S., Salaris, M., \& Castelli, F. 2004, \apj, 612, 168

907: \bibitem[Pont et al.(2004)]{pont04} Pont, F., Zinn, R., Gallart, C., Hardy, E., \& Winnick, R. 2004, \aj, 127, 840

908: \bibitem[Putman et al.(2003)]{putman03} Putman, M. E., Staveley-Smith, L., Freeman K. C., Gibson, B. K. \&

909: Barnes, D. G. 2003, \apj, 586, 170

910: %\bibitem[Rutledge et al.(1997a)]{r97a} Rutledge, G. A., Hesser, J. E., Stetson, P. B., Mateo, M., Simard, L., Bolte, M., Friel, E. D., \&

911: %Copin, Y. 1997a, \pasp, 109, 883

912: %\bibitem[Rutledge et al.(1997b)]{r97b} Rutledge, G. A., Hesser, J. E., \& Stetson, P. B. 1997b, \pasp, 109, 907

913: \bibitem[Salaris \& Weiss(2002)]{sw02}Salaris, M., \& Weiss, A. 2002, \aap, 388, 492

914: \bibitem[Salaris, Weiss, \& Percival(2004)]{swp04}Salaris, M., Weiss, A., \& Percival, S. M. 2004, \aap, 414, 163

915: \bibitem[Santos Jr. et al.(1999)]{santos99} Santos Jr., F. F. C., Piatti, A. E., Cari\'a, J. J., Geisler, D., \& Dottori, H. 1999, \aj, 117,

916: 2841

917: %\bibitem[Saviane et al.(2000)]{saviane00} Saviane, I., Rosenberg, A., Piotto, G., \& Aparicio, A. 2000, \aap, 355, 966

918: \bibitem[Scalo(1986)]{scalo86} Scalo, J. M. 1986, \fcp, 11, 1

919: \bibitem[Schommer et al.(1992)]{schommer92} Schommer, R. A., Suntzeff, N. B., Olszewski, E. W. \& Harris, H. C.

920: 1992, \aj, 103, 447

921: \bibitem[Sellwood \& Wilkinson(1993)]{sellwood93}Sellwood, J. A. \& Wilkinson, A. 1993, Reports on Progress in Physics, 56, 173

922: \bibitem[Smecker-Hane et al.(2002)]{Smecker-Hane02}Smecker-Hane, T. A., Cole, A. A., Gallagher, J. S., III \& Stetson, P. B. 2002, \apj, 566, 239

923: \bibitem[Spergel et al.(2003)]{spergel03}Spergel, D. N., Verde, L., Peiris, H. V., Komatsu, E., Nolta, M. R., Bennett, C. L., Halpern, M.,

924: Hinshaw, G., Jarosik, N., Kogut, A., Limon, M., Meyer, S. S., Page, L., Tucker, G. S., Weiland, J. L., Wollack, E., Wright, E. L. 2003,

925: \apjs, 148, 175 

926: %\bibitem[Staveley-Smith et al.(2003)]{staveleysmith03} Staveley-Smith, L., Kim, S., Calabretta, M. R., Haynes, S. R., \& Kesteven, M. J.

927: %2003, \mnras, 339, 87

928: %\bibitem[Suntzeff et al.(1993)]{sunt93} Suntzeff, N. B., Mateo, M., Terndrup, D. M., Olszewski, E. W., Geisler, D., \&

929: %Weller, W. 1993, \apj, 418, 208

930: \bibitem[Tinsley(1980)]{tinsley80}Tinsley, B. M.,  1980, \fcp, 5, 287

931: \bibitem[Tonry \& Davis(1979)]{td79}Tonry, J., \& Davis, M. 1979, \aj, 84, 1511

932: %\bibitem[van der Marel(2001)]{vandermarel01}  Van der Marel, R. P.  2001, \aj, 122, 1827

933: %\bibitem[van der Marel et al.(2002)]{vandermarel02}  Van der Marel, R. P., Alves, D. R., Hardy, E., \& Suntzeff, N. B. 2002, \aj, 124,

934: %2639

935: \bibitem[van Dokkun(2001)]{dokkun01}van Dokkun, P. G.  2001, \pasp, 113, 1420

936: \bibitem[Vallenari et al.(1996)]{vallenari96}Vallenari, A., Chiosi, C., Bertelli, G. \& Ortolani, S. 1996, \aap, 309, 358

937: \bibitem[Westerlund et al.(1990)]{westerlund90}Westerlund, B. E. 1990, \aapr, 2, 29

938: \bibitem[Westerlund, Linde, \& Lynga(1995)]{westerlund95}Westerlund, B. E., Linde, P. \& Lynga, G. 1995, \aap, 298, 39

939: %\bibitem[Zaritsky(2004)]{zaritsky04} Zaritsky, D. 2004 \aj, 614, L37

940: \bibitem[Zinn \& West(1984)]{zw84} Zinn, R., \& West, M. J. 1984 \apjs, 55, 45

941: %\bibitem[Zhao \& Evans(2003)]{zhao00} Zhao, H. S., \& Evans, N. W. 2000 \aj, 545, L35

942: \bibitem[Zhao et al.(2003)]{zhao03} Zhao, H. S., Ibata, R., Lewis, G. F., \& Irwin, M. J. 2003 \mnras, 339, 701

943: \end{thebibliography}

944: 

945: \clearpage

946: 

947: \begin{figure}

948: \epsscale{1}

949: \plotone{f01.eps}

950: \caption{CMD of the RGB region of the field situated at 6\arcdeg\@ from the LMC center. The regions used to

951: select stars to be observed spectroscopically are shown.\label{dcmbox}}

952: \end{figure}

953: 

954: \clearpage

955: \begin{figure}

956: \epsscale{1}

957: \plotone{f02.eps}

958: \caption{Example of sky subtraction. Top: raw spectrum. Middle: sky spectrum obtained

959:  from  a combination of fibers located on the

960: sky. Bottom: The final spectrum after  sky subtraction.\label{sky}}

961: \end{figure}

962: 

963: \clearpage

964: 

965: \begin{figure}

966: \epsscale{1}

967: \plotone{f03.eps}

968: \caption{Radial velocity distributions of our four LMC fields. The mean value of each distribution is shown in

969: each panel, and its position is marked with an arrow. Note that there are two peaks in the field situated at

970: 3\arcdeg.\label{vel_dist}}

971: \end{figure}

972: 

973: \clearpage

974: 

975: \begin{figure}

976: \epsscale{1}

977: \plotone{f04.eps}

978: \caption{Position in the M$_I$--$\Sigma$Ca plane of LMC stars for which membership has been confirmed from

979: their radial velocity. The typical $\Sigma Ca$ error is shown in the bottom right corner of each panel.

980: Isometallicity lines have been plotted

981: for reference. The

982: solid part of each line is the magnitude interval covered by the cluster stars used for the calibration

983: (see Paper II). The dashed part is the region in which the calibration is

984: extrapolated.\label{misigmaca}}

985: \end{figure}

986: 

987: \clearpage

988: 

989: \begin{figure}

990: \epsscale{1}

991: \plotone{f05.eps}

992: \caption{Metallicity distributions for the four fields in our sample. A Gaussian has been fitted to 

993: each distribution in order to obtain its

994: mean and  dispersion. The values obtained are shown in each panel.\label{histo_fields}}

995: \end{figure}

996: 

997: \clearpage

998: 

999: \begin{figure}

1000: \epsscale{1}

1001: \plotone{f06.eps}

1002: \caption{Position of the observed star in the

1003: color-magnitude diagram. Different metallicity intervals are represented by different symbols, as indicated in the label. Solid lines are

1004: the fiducial sequences of four globular clusters (from right to left: NGC 104: [Fe/H] = $-0.78$; NGC 5904:

1005: [Fe/H] =$-1.12$; NGC 5272: [Fe/H] = $-1.34$ and NGC 6341: [Fe/H] = $-2.16$).\label{dcm_feh}}

1006: \end{figure}

1007: 

1008: \clearpage

1009: 

1010: \begin{figure}

1011: \epsscale{1}

1012: \plotone{f07.eps}

1013: \caption{Uncertainty in the age calculated using Equation \ref{rela} obtained from the BaSTI stellar evolution

1014: models. See text for details.\label{erroredad}}

1015: \end{figure}

1016: 

1017: \clearpage

1018: 

1019: \begin{figure}

1020: \epsscale{0.6}

1021: \plotone{f08.eps}

1022: \caption{Ages derived for the cluster sample in Paper II, using Equation \ref{rela} and BaSTI and Padova

1023: models versus the reference ages. The solid lines are the one-to-one relation. 

1024: \label{cluster}}

1025: \end{figure}

1026: 

1027: \clearpage

1028: 

1029: \begin{figure}

1030: \epsscale{1}

1031: \plotone{f09.eps}

1032: \caption{Radial velocities for carbon stars from \citet{kunkel97} in 

1033: the band $-36\arcdeg \leq b \leq -33\arcdeg$. The triangles

1034: are the carbon stars in our fields at 

1035: 5\arcdeg~and 8\arcdeg. The filled square represents the

1036: data from \citet{c05} for the bar. The filled dots are our data listed in Table \ref{radialvelocity}. 

1037: The error bars represent the dispersions of the

1038: velocity distributions.\label{kunkel}}

1039: \end{figure}

1040: 

1041: \clearpage

1042: 

1043: \begin{figure}

1044: \epsscale{1}

1045: \plotone{f10.eps}

1046: \caption{Radial velocity dispersion of the disk stars as a function of metallicity. 

1047: Although the velocity dispersion increases from metal-rich to metal-poor stars, this increment is not

1048: clear enough to conclude that a metal-poor stellar halo is present.\label{vrmetal}}

1049: \end{figure}

1050: 

1051: \clearpage

1052: 

1053: \begin{figure}

1054: \epsscale{1}

1055: \plotone{f11.eps}

1056: \caption{Age--metallicity relationships for the four LMC fields in our sample. Inset panels show the age

1057: distribution computed taking into account (\textsl{solid line}) and not (\textsl{histogram}) the age determination uncertainties.

1058: The top panel show the age error in each interval. The left panels show the metallicity

1059: error.

1060: \label{amrfields}}

1061: \end{figure}

1062: 

1063: \clearpage

1064: 

1065: \begin{figure}

1066: \epsscale{1}

1067: \plotone{f12.eps}

1068: \caption{AMR of the field situated at 5\arcdeg\ for different star groups. Squares: objects with magnitudes in

1069: the range of the M$_I$--$\Sigma Ca$ calibration obtained in Paper\ II; open circles:

1070: objects in the same field with magnitudes brighter than the cluster calibration stars. The corresponding average metallicity and

1071: dispersion in different age bins are also plotted. \label{testamr}}

1072: \end{figure}

1073: 

1074: \clearpage

1075: 

1076: \begin{figure}

1077: \epsscale{1}

1078: \plotone{f13.eps}

1079: \caption{Same as Figure \ref{amrfields} but for the bar (\textsl{left}) and the disk fields combined (\textsl{right}).\label{amrbardisk}}

1080: \end{figure}

1081: 

1082: \clearpage

1083: 

1084: \begin{figure}

1085: \epsscale{1}

1086: \plotone{f14.eps}

1087: \caption{LMC clusters AMR by \citet[\textsl{squares}]{ol91}; \citet[\textsl{triangles}]{dirsch00} and \citet[\textsl{hexagons}]{grocholski06}. The mean metallicity in six age bins of our global disk has been plotted (\textsl{filled

1088: points}). Note that the metallicity scales of each work may not be exactly the same.\label{cumulos}}

1089: \end{figure}

1090: 

1091: %\clearpage

1092: 

1093: \begin{figure}

1094: \epsscale{1}

1095: \plotone{f15.eps}

1096: \caption{Results of the test on the reliability of the SFH computed using our spectroscopic RGB sample. Solid

1097: line: injected SFH; dashed line: mean and sigma of the recovered SFH after the Hydra filter assignment

1098: simulation; dotted line: the same as dashed one but with the age obtained from Equation \ref{rela}. See text for detail.\label{pruebaconf}}

1099: \end{figure}

1100: 

1101: %\clearpage

1102: \begin{figure}

1103: \epsscale{1}

1104: \plotone{f16.eps}

1105: \caption{SFR, as a function of time, for our four fields. Note that the $y$-axis scale is different in the top and

1106: bottom panels. Histogram is the $\psi(t)$ computed from the age distribution without taking into account the age uncertainty. Solid line is the same but computed by taken into account the age error, while dotted lines are its uncertainty.\label{hfecampos}}

1107: \end{figure}

1108: 

1109: %\clearpage

1110: \begin{figure}

1111: \epsscale{1}

1112: \plotone{f17.eps}

1113: \caption{SFR as a function of time for the bar (\textsl{left}) and disk (\textsl{right}) derived from the spectroscopic sample.\label{hfe}}

1114: \end{figure}

1115: 

1116: \clearpage

1117: 

1118: \begin{figure}

1119: \epsscale{1}

1120: \plotone{f18.eps}

1121: \caption{Chemical evolution models overplotted on the AMR for the disk. Red solid lines are the closed-box

1122: models with $y=0.008$,  (\textsl{thick line}) and $y=0.014$ (\textsl{thin line}), respectively.

1123: Blue dot-dashed lines are outflow models for $\lambda =$ 0.6, 1.3, 2.0 and 2.5. Green dashed lines are infall models with

1124: $\alpha=$ 1, 0.9, 0.5 and 0.3. Pink long--short dashed lines are models with infall

1125: and outflow for $\alpha=$ 0.2, and $\lambda=$ 0.1 (\textsl{upper}) and 0.05 (\textsl{lower}). Finally, cyan short-long dashed line is a model with inflow

1126: and outflow for $\alpha=$ 0.3 and $\lambda=$ 0.2. All infall and outflow models assume $y=0.014$. The brown solid and dashed lines represent the bursting and smooth models computed by \citet{pagel98}.\label{modeldisk}}

1127: \end{figure}

1128: 

1129: \clearpage

1130: 

1131: \begin{figure}

1132: \epsscale{1}

1133: \plotone{f19.eps}

1134: \caption{As Figure \ref{modeldisk} for the bar. Red solid line is a 

1135: closed-box model. Blue dot-dashed lines are outflow models for

1136: $\lambda=$ 0.8 and 0.3. Green short-dashed lines are infall models with $\alpha=$ 1, 0.6 and

1137: 0.4. Pink long--short dashed lines are models with inflow

1138: and outflow for $\alpha=$ 1.2 and $\lambda=$ 0.4, 0.5 and 0.6. All models assume y=0.014. As in the previous

1139: figure, brown lines represent the models computed by \citet{pagel98}.\label{modelbar}}

1140: \end{figure}

1141: 

1142: \clearpage

1143: 

1144: \begin{deluxetable}{ccccccccc}

1145: \tabletypesize{\scriptsize}

1146: %%\rotate

1147: \tablecaption{LMC RGB stars observed\label{starobs}}

1148: \tablewidth{0pt}

1149: \tablehead{

1150: \colhead{$\alpha_{2000}$} & \colhead{$\delta_{2000}$} & \colhead{$\Sigma$ Ca} & \colhead{$\sigma_{\Sigma Ca}$} & \colhead{V} & \colhead{I} & \colhead{$V_r (km s^{-1}$)} & \colhead{$\sigma_{V_r} (km s^{-1}$)} & \colhead{Comments} 

1151: }

1152: \startdata

1153: 05:08:53.92 & -66:49:55.1 & 7.8 & 0.8 & 17.67 & 16.36 & 287.3 & 0.7 &              \\

1154: 05:08:55.21 & -66:47:34.7 & 8.7 & 0.2 & 16.66 & 14.70 & 240.9 & 0.6 &              \\

1155: 05:08:53.89 & -67:02:50.0 & 2.4 & 0.2 & 17.19 & 14.25 &  23.0 & 0.6 & No member   \\

1156: 05:08:55.40 & -66:53:45.4 & 8.9 & 0.4 & 16.79 & 15.09 & 292.3 & 0.5 &              \\

1157: 05:08:58.58 & -66:45:49.2 & 9.4 & 0.2 & 16.00 & 13.99 & 291.9 & 0.5 &  \\

1158: \enddata

1159: \tablecomments{Table \ref{starobs} is published in its entirety in the electronic edition of The Astronomical 

1160: Journal. A portion is shown here for

1161: guidance regarding its form and content}

1162: 

1163: \end{deluxetable}

1164: 

1165: %\clearpage

1166: \begin{deluxetable}{ccc}

1167: \tabletypesize{\scriptsize}

1168: %%\rotate

1169: \tablecaption{Radial velocities and velocity dispersion for each field.\label{radialvelocity}}

1170: \tablewidth{0pt}

1171: \tablehead{

1172: \colhead{Field} & \colhead{$\langle V_r\rangle (km s^{-1})$} & \colhead{$\sigma_{V} (km s^{-1})$} 

1173: }

1174: \startdata

1175: Bar & 260 & 24 \\

1176: 3\arcdeg & 269 & 27 \\

1177: 5\arcdeg & 278 & 20 \\

1178: 6\arcdeg & 293 & 15 \\

1179: 8\arcdeg & 282 & 14 \\

1180: \enddata

1181: \end{deluxetable}

1182: 

1183: \begin{deluxetable}{cc}

1184: \tabletypesize{\scriptsize}

1185: \tablecaption{Line and continuum bandpasses

1186: \label{bandastable}}

1187: \tablewidth{0pt}

1188: \tablehead{

1189: \colhead{Line Bandpasses (\AA)} & \colhead{Continuum bandpasses (\AA)}}

1190: \startdata

1191: 8484-8513 & 8474-8484\\

1192: 8522-8562 & 8563-8577\\

1193: 8642-8682 & 8619-8642\\

1194: \nodata & 8799-8725\\

1195: \nodata & 8776-8792\\

1196: \enddata

1197: \end{deluxetable}

1198: 

1199: 

1200: \begin{deluxetable}{lcccccccc}

1201: \tabletypesize{\scriptsize}

1202: %%\rotate

1203: \tablecaption{Coefficients of Equation \ref{rela}.\label{tableage}}

1204: \tablewidth{0pt}

1205: \tablehead{

1206: \colhead{Model} & \colhead{a} & \colhead{b} & \colhead{c} & \colhead{d} & \colhead{f}  & \colhead{g} & \colhead{h} & \colhead{$\sigma$}

1207: }

1208: \startdata

1209: BaSTI & 2.57$\pm$0.10 & 9.72$\pm$0.15 & 0.70$\pm$0.003 & -1.51$\pm$0.007 & -3.86$\pm$0.08 &-0.19$\pm$0.007 &

1210: 0.49$\pm$0.01 & 0.38 \\

1211: Padova & 1.69$\pm$0.07 & 10.6$\pm$0.1 & 0.75$\pm$0.002 & -1.87$\pm$0.007 & -4.14$\pm$0.06 &-0.24$\pm$0.006 & 0.51$\pm$0.009 &0.37 \\

1212: \enddata

1213: \end{deluxetable}

1214: 

1215: \clearpage

1216: 

1217: \begin{deluxetable}{ccccc}

1218: \tabletypesize{\scriptsize}

1219: %%\rotate

1220: \tablecaption{Velocity dispersion for each metallicity bin.\label{vrmetaltable}}

1221: \tablewidth{0pt}

1222: \tablehead{

1223: \colhead{[Fe/H]} & \colhead{Age (Gyr)} & \colhead{N$_*$} & \colhead{$\langle V_r\rangle (km s^{-1})$} & \colhead{$\sigma_{V} (km s^{-1})$}

1224: }

1225: \startdata

1226: $\geq$-0.5 & $<$3 & 226 & 0.1 & 20.5\\

1227: -0.5 to -1 & 2-10 & 178 & 1.9 & 20.5\\

1228: $\leq$-1 & $>$10 & 100 & 1.8 & 26.4\\

1229: \enddata

1230: \end{deluxetable}

1231: 

1232: %\clearpage

1233: 

1234: \begin{deluxetable}{ccc}

1235: \tabletypesize{\scriptsize}

1236: %%\rotate

1237: \tablecaption{Mean values of metallicity distributions.\label{metallicitybin}}

1238: \tablewidth{0pt}

1239: \tablehead{

1240: \colhead{Field} & \colhead{$\langle$[Fe/H]$\rangle$} & \colhead{$\sigma_{[Fe/H]}$} 

1241: }

1242: \startdata

1243: Bar & -0.39 & 0.19 \\

1244: 3\arcdeg & -0.47 & 0.31 \\

1245: 5\arcdeg & -0.50 & 0.37 \\

1246: 6\arcdeg & -0.45 & 0.31 \\

1247: 8\arcdeg & -0.79 & 0.44 \\

1248: \enddata

1249: \end{deluxetable}

1250: 

1251: \begin{deluxetable}{cccccccc}

1252: \tabletypesize{\tiny}

1253: \tablecaption{Average metallicity in six age bins.\label{testchi2}}

1254: \tablewidth{0pt}

1255: \tablehead{

1256: \colhead{Field} & \colhead{$\langle[Fe/H]_{\leq1.5}\rangle$} &  \colhead{$\langle[Fe/H]_{1.5-3.5}\rangle$}

1257: & \colhead{$\langle[Fe/H]_{3.5-5.5}\rangle$} & \colhead{$\langle[Fe/H]_{5.5-8.5}\rangle$} & \colhead{$\langle[Fe/H]_{8.5-11}\rangle$} & 

1258: \colhead{$\langle[Fe/H]_{\geq11}\rangle$} & \colhead{$\chi^2_\nu$} \\ 

1259: }

1260: \startdata

1261: Bar & -0.27$\pm$0.13 & -0.32$\pm$0.16 & -0.53$\pm$0.26 & -0.65$\pm$0.30 & -1.02$\pm$0.41 & -1.25$\pm$0.56 &

1262: ... \\

1263: Disk\tablenotemark{a} & -0.17$\pm$0.21 & -0.39$\pm$0.15 & -0.60$\pm$0.16 & -0.71$\pm$0.18 & -0.80$\pm$0.27 & -1.25$\pm$0.41 &

1264: ... \\

1265: 3\arcdeg & -0.16$\pm$0.17 & -0.40$\pm$0.14 & -0.58$\pm$0.15 & -0.67$\pm$0.20 & -0.69$\pm$0.07 & -1.13$\pm$0.43 & 0.05\\

1266: 5\arcdeg & -0.16$\pm$0.28 & -0.39$\pm$0.15 & -0.63$\pm$0.18 & -0.72$\pm$0.16 & -0.83$\pm$0.35 & -1.39$\pm$0.42 & 0.02\\

1267: 6\arcdeg & -0.15$\pm$0.16 & -0.38$\pm$0.16 & -0.60$\pm$0.16 & -0.75$\pm$0.20 & -0.83$\pm$0.14 & -1.34$\pm$0.24 & 0.02\\

1268: 8\arcdeg & -0.16$\pm$0.21 & -0.45$\pm$0.13 & -0.62$\pm$0.20 & -0.89$\pm$0.16 & -0.92$\pm$0.18 & -1.25$\pm$0.32 & 0.16\\

1269: \enddata

1270: \tablenotetext{a}{Combination of the four fields.}

1271: \end{deluxetable}

1272: 

1273: \end{document}

1274: