The Build-Up of the Red Sequence in High Redshift Galaxy Clusters

The Build-up of the Red Sequence in High Redshift Galaxy Clusters

Pierluigi Cerulo

Presented in fulﬁllment of the requirements of the degree of Doctor of Philosophy

12 February 2015

Faculty of Science, Engineering and Technology Swinburne University

i Abstract

The primary scientific goal of this thesis is to study of the evolution of galaxies in high redshift clusters through the investigation of the build-up of the red sequence as a function of redshift. We address this problem from four complementary points of view, namely the study of the build-up of the faint end of the red sequence, the analysis of the morphology of red sequence galaxies, the investigation of the relationships between galaxy stellar populations, luminosity, and morphology, and the comparison between the properties of galaxies in low and high mass clusters. For this purpose we use a sample of 9 galaxy clusters at 0.8 < z < 1.5 from the HAWK- I Cluster Survey (HCS, Lidman et al. 2013), a program conducted with the infrared High Acuity Wide-field K-band Imager (HAWK-I) at the ESO Very Large Telescope (VLT), and aimed at the study of high-redshift clusters. Deep optical imaging from the Hubble Space Telescope (HST) and VLT, and up to 100 redshifts from various observing programs are also available for each of the clusters, making the HCS one of the richest and deepest samples of high redshift clusters currently available. Throughout the thesis we use the WIde-field Nearby Galaxy-cluster Survey (WINGS, Fasano et al. 2006) as a comparison sample of clusters at low redshifts. Given the diversity and heterogeneity of the HCS dataset, we set out, at the beginning of this project, to develop a method for the analysis of the clusters with the aim of using all the available photometric and spectroscopic information to produce unbiased multi-wavelength samples. We tested our method on the cluster XMMU J1229+0151, at z = 0.98, which is one of the lowest-redshift HCS clusters and has one of the richest datasets. We implemented two methods for the estimation of cluster membership, based on photometric redshifts and on statistical background subtraction, showing that they produced reliable and consistent results. We also adopted a coupled visual and automatic approach for the morphological classification of red sequence galaxies, showing that it results in a reliable separation between elliptical and S0 galaxies, as well as disc-dominated galaxies. The analysis of the HCS clusters shows that the red sequence was already in place at z = 1.5. In fact, we find that the red sequence slope and intrinsic scatter do not change significantly with redshift, in agreement with most results in the recent literature. We also find that the cluster red sequence was already developed at low luminosities, showing that the luminous-to-faint-ratio and the luminosity distributions are consistent with the z 0.05 clusters of the WINGS survey. Interestingly, we detect a population of ∼ ii luminous (VAB < 22.0 mag) and massive (log(M∗/M ) > 11.0) red sequence galaxies − at the centres of the clusters with high dark matter halo masses which are not present in low-mass clusters. In order to explain the properties of the HCS red sequence, we propose an evolutionary scenario based on the accretion of low-mass groups on to a central massive protocluster at z > 2. According to this scenario, the red sequence is populated both by the galaxies that ceased their star formation in the protocluster and by the galaxies that had their star formation quenched in the satellite groups. The latter populate the red sequence at all stellar masses, resulting in a fast build-up also at the faint end, as we observe in the HCS clusters. In this scenario, more massive clusters are formed at earlier times, with respect to their low-mass counterparts, and accrete satellite groups over longer timescales. Therefore, the central galaxies in these systems accrete low-mass satellite galaxies over longer timescales than those involved in lower-mass clusters, experiencing a higher mass growth. This can explain the presence of the high mass central galaxies in the most massive clusters.

As well as analyse the evolution of the cluster red sequence, we investigate the morphological properties of red sequence members. We find that the HCS red sequence is dominated by early-type galaxies as in lower-redshift clusters. However , the comparison with WINGS shows that there is a significant growth in the fraction of S0 galaxies, which become dominant on the WINGS red sequence at magnitudes VAB > 21.0 mag. Unlike − WINGS, the HCS red sequence is dominated by elliptical galaxies at all luminosities and stellar masses. We conclude that elliptical and S0 galaxies follow different evolutionary paths, and that the low-redshift S0s were formed as the result of the morphological transformation of quiescent spiral galaxies. We also find that, at both high and low redshifts, late-type galaxies make up 10% of the red sequence, and that they have slightly bluer colours than early-type galaxies. Our analysis also shows that the bright end of the red sequence is dominated by elliptical galaxies at low and high redshifts.

In order to study the relationships between galaxy luminosity, morphology, and stellar populations, we analysed a sample of spectra of red sequence galaxies in 3 HCS clusters at z 1 observed at the 8m Gemini-North telescope and at the 10m Keck I telescope. ∼ The preliminary results of this study, which are based on a purely qualitative analysis of the co-added spectra, suggest that S0 galaxies host younger stellar populations than those of elliptical galaxies. Our results also suggest that the age of the stellar populations increases with galaxy luminosity. We stress that these results are preliminary and that only after estimating stellar age and metallicity from the available spectra, we will be able to quantitatively investigate any relation between stellar populations, morphology, and iii luminosity of red sequence galaxies. iv v Acknowledgements

I would like to thank my supervisors Warrick Couch and Chris Lidman who gave me the opportunity to work on an exciting research project for my PhD. I would also like to thank the Australian Astronomical Observatory (AAO) for granting me a top-up scholarship and for hosting me during my visits. I thank the AAO staff for their help and assistance during my visits at the observatory. The research developed during my PhD would not have been possible without the help and support of the HCS collaboration who gave me useful feedback and comments for my paper “The morphological transformation of red sequence galaxies in the distant cluster XMMU J1229+0151”. In particular, I thank Simona Mei and Marc-Huertas-Company who hosted me at the Paris Observatory, allowing me to actively participate in their research activities, and giving me valuable suggestions for my work. I would also like to thank Ricardo Demarco and Julie Nantais for providing me with the most up to date redshift catalogues for the cluster RDCS J1252.9-2927, which have improved the quality and reliability of my results. I thank the members of external collaborations who provided me with additional data, catalogues, and software. The SofI images of the cluster XMMU J1229+0151 were provided by Joana Santos. The ISAAC images of the clusters RCS 2319.8+0038, RCS 0220.9- 0333, and RCS 2345-3633 were provided by Roberto Muñoz.The WINGS catalogues and additional related information were provided by Bianca Poggianti and Alessia Moretti. The software for stellar population measurements was provided by Max Spolaor and Pe- ter Jensen. The software for simulating galaxy images was provided by Boris Häußler. I thank Evelyn Caris for her help with the CarPy software and Keck data reduction. I thank Joshua Meyers for kindly providing his PSF fitting software. Finally I would like to thank the Centre for Astrophysics and Supercomputing for giving me the opportunity to work in a high quality research centre with an excellent work environment. In particular, I thank my review panel for their helpful suggestions and the PhD student coordinators Chris Blake and Virginia Kilborn for the useful meetings organised in these four years. I also thank Chris Blake, Darren Croton, Alister Graham, and Luca Cortese for their suggestions on the estimation of completeness limits, red sequence number counts, aperture photometry, and red sequence selection effects, and I thank Matt Owers and Max Spolaor for introducing me to MOS spectroscopy and stellar populations. The work presented in this thesis has been carried out with the Green and gSTAR vi supercomputers, and I thank the Swinburne supercomputing team for their support. vii

ix Statement of originality

The work presented in this thesis was carried out at the Centre for Astrophysics & Supercomputing of the Swinburne University of Technology and at the Australian Astro- nomical Observatory during the period 2010-2015. This thesis contains no material that has been accepted for the award of any other degree or diploma. To the best of my knowledge, this thesis contains no material previously published or written by another author, except where due reference is made in the text of the thesis. All ﬁgures were created by the author, except where due reference is made in the text of the thesis. All work presented is that of the author, except where due reference is made in the text of the thesis and I remain solely responsible for this work. All my principal and associate supervisors were co-authors of the relevant publications during which I received their guidance and input. The content of the Chapters listed below has appeared in refereed journals. Minor al- terations have been made to the published papers in order to maintain argument continuity and consistency of spelling and style:

Chapter 3, Appendix A, and part of Appendix B have been published in Monthly • Notices of the Royal Astronomical Society:

Cerulo, P., Couch, W. J., Lidman, C., Delaye, L., Demarco, R., Huertas-Company, M., Mei, S., S´anchez-Janssen, R.

The morphological transformation of red sequence galaxies in the distant cluster XMMU J1229+0151, 2014, MNRAS, 439, 2790

I acknowledge helpful discussions and critical feedback that were provided by my co-authors and the anonymous referee during the preparation of this publication.

Pierluigi Cerulo Melbourne, Australia 12/02/2015 x xi Acronyms, Abbreviations and Conventions

P. C: Pierluigi Cerulo • W. J. C.: Warrick J. Couch • C. L.: Chris Lidman • ESO: European Southern Observatory • ETG: Early-Type Galaxy • FORS2: FOcal Reducer and low dispersion Spectrograph 2 • GMOS: Gemini Multi-Object Spectrograph • GOODS: Great Observatories Origins Deep Survey • HAWK-I: High Acuity Wide-ﬁeld K-band Imager • HST: Hubble Space Telescope • IMF: Initial Mass Function • ISAAC: Infrared Spectrometer And Array Camera • LRIS: Low-Resolution Imager Spectrometer • LTG: Late-Type Galaxy • SED: Spectral Energy Distribution • SVM: Support Vector Machine • VLT: Very Large Telescope • WINGS: WIde-ﬁeld Nearby Galaxy-cluster Survey • xii Contents

Abstract i

Acknowledgements v

Declaration ix

Acronymus, Abbreviations and Conventios xi

List of Figures xvii

List of Tables xxxi

1 Introduction 1 1.1 Galaxy Clusters as Observational Laboratories ...... 2 1.2 The Cluster Red Sequence ...... 6 1.3 The Morphology of Galaxies and its Physical Implications ...... 10 1.4 Outline of the Thesis ...... 13

2 Data and Observations 15 2.1 The HAWK-I Cluster Survey ...... 15 2.2 Observations and Data Reduction: Imaging ...... 18 2.2.1 Advanced Camera for Surveys ...... 18 2.2.2 Wide Field Camera 3 ...... 19 2.2.3 Infrared Spectrometer And Array Camera (ISAAC) ...... 19 2.3 Observations and Data Reduction: Spectroscopy ...... 21 2.3.1 The Gemini North Observations of XMM J1229+0151 and RCS 0220.9-0333 ...... 23 2.3.2 Keck/LRIS Observations of the cluster RCS 2319.8+0038 ...... 25 2.3.3 Redshift Measurements ...... 27

3 Analysis Method 31 3.1 Introduction ...... 32 3.2 Observations and data reduction ...... 36 3.2.1 HST imaging ...... 37 3.2.2 Ground Based Imaging ...... 38 3.2.3 Ground-based Spectroscopy ...... 39 3.3 Data Analysis and Measurements ...... 41

xiii xiv Contents

3.3.1 Object Detection and PSF modelling ...... 41 3.3.2 Photometric Catalogue ...... 42 3.4 Results ...... 44 3.4.1 Cluster Membership: Photometric Redshifts ...... 44 3.4.2 Contamination from Field Interlopers ...... 46 3.4.3 Colour-Magnitude Diagram and Red Sequence ...... 47 3.4.4 Galaxy Morphology and Structure ...... 54 3.4.5 The low-redshift Cluster Samples ...... 60 3.5 Discussion ...... 62 3.5.1 The Faint End of the Red Sequence ...... 62 3.5.2 Morphological Evolution ...... 66 3.5.3 Structural Properties ...... 70 3.5.4 The Red Sequence Slope and Scatter ...... 71 3.6 Summary and Conclusions ...... 73

4 The Build-up of the Red Sequence 75 4.1 Photometry and Cluster Membership ...... 75 4.1.1 Object Detection and PSF Modelling ...... 75 4.1.2 Background Contamination ...... 76 4.2 The Cluster Red Sequence at 0.8 < z < 1.5 ...... 81 4.2.1 The Fitting Procedure ...... 81 4.2.2 The Individual Red Sequences ...... 89 4.3 Red Sequence Properties ...... 98 4.3.1 The Red Sequence Parameters ...... 98 4.3.2 The Luminous-to-Faint Ratio ...... 101 4.3.3 Red Sequence Luminosity Distribution ...... 103 4.4 Discussion ...... 105 4.4.1 The Evolution of the Red Sequence Parameters ...... 105 4.4.2 The Luminous-to-Faint Ratio and the Luminosity Distribution of the Cluster Red Sequence ...... 108 4.4.3 A Possible Scenario for the Build-up of the Cluster Red Sequence . . 115

5 Morphological Evolution 119 5.1 Morphology of the HCS Galaxies ...... 119 5.1.1 Classiﬁcation Procedure ...... 119 5.1.2 Testing Morphology I: Internal Comparison ...... 123 Contents xv

5.1.3 Testing Morphology II: The Elliptical vs S0 Separation ...... 124 5.1.4 Testing Morphology III: Comparison with the Literature ...... 129 5.2 Stellar Mass Estimate of Red Sequence Galaxies ...... 131 5.3 The low-redshift Comparison Sample ...... 134 5.4 Results ...... 135 5.4.1 Statistical Background Subtraction ...... 135 5.4.2 Morphological Fractions ...... 135 5.5 Morphological Transformations in Galaxy Clusters ...... 136

6 Spectral Properties 145 6.1 The Stellar Populations of Galaxies ...... 146 6.2 The Average Red Sequence Spectra of HCS Galaxies at z 1 ...... 150 ∼ 6.3 Discussion: The Properties of the Red Sequence Spectra ...... 151 6.3.1 Spectral Properties as a Function of Galaxy Morphology ...... 151 6.3.2 Spectral Properties as a Function of Galaxy Luminosity ...... 155

7 Summary and Future Plans 159 7.1 Summary and Conclusions ...... 159 7.2 Future Plans ...... 164

Bibliography 167

Appendices

A Morphological Classiﬁcation 195

B Automatic Morphology 199

List of Figures

1.1 The core of the cluster of galaxies Abell 370 at z = 0.375 as seen with the HST/ACS camera. Elliptical and S0 galaxies dominate the galaxy population of this cluster. The colour image was obtained by combining ACS images taken with the F475W, F625W, and F814W ﬁlters ...... 2

1.2 Colour-composite image of the spiral-rich Hercules cluster (Abell 2151, z = 0.036) obtained with the ESO/VLT Survey Telescope (VST). Unlike Abell 370 in Figure 1.1, dominated by elliptical and S0 galaxies, the core of this cluster is also populated by spiral galaxies. (Credit: ESO/INAF- VST/OmegaCAM. Acknowledgement: OmegaCen/Astro-WISE/Kapteyn Institute)...... 5

1.3 The (r Ks) vs Ks colour-magnitude diagram of the cluster RX J0152-1357 − at z = 0.84 from Demarco et al. (2010). The upper part of the diagram highlights the red sequence. The solid black line is the best-ﬁt straight line to the red sequence. See Figure 2 of Demarco et al. (2010) for the description of the symbols and colours. (Credit: Demarco, R. et al., 2010, ApJ, 725, 1252) ...... 7

1.4 The Hubble tuning fork diagram. From left to right: elliptical galaxies, S0 galaxies, normal spiral galaxies (top), barred spiral galaxies (bottom), and irregular galaxies. S0 galaxies are situated at the junction between the sequences of ellipticals and spirals. The images were obtained by combining HST/ACS exposures in the three ﬁlters F475W, F625W, and F814W. . . . 11

3.1 Colour image of XMM1229 obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks image. White circles are photometrically selected red sequence members and yellow squares are spectroscopically conﬁrmed members (see 3.4.1 and Table 3.3 for the determination § of the cluster membership and a complete list of red sequence galaxies, respectively)...... 35

xvii xviii List of Figures

3.2 Photometric coverage of the XMM1229 ﬁeld. From left to right: R SPECIAL (R), F775W (i), F850LP (z), F105W (Y), F125W (J), F160W (H), Ks. The F110W transmission curve is not plotted, as it covers the same spectral range of the F105W and F125W bands. The SofI J band transmission curve is not plotted because it overlaps with F125W. The F775W and F850LP bands used for the colour-magnitude diagram are highlighted by the arrows. The solid black line represents the template SED of an elliptical galaxy from Coleman et al. (1980) at the redshift of XMM1229...... 36

3.3 (left): Calibration of the zpeg photometric redshift estimate for the clus-

ter centre. On the x-axis it is plotted the value zspec of the spectroscopic redshifts measured with FORS2 in the cluster centre, while on the y-axis it

is plotted the discrepancy ∆z = (zspec zphot)/(1 + zspec). (right): Photo- − metric redshift distribution in the central region of XMM1229. The vertical dashed lines are the two limits used to deﬁne the cluster membership. The redshift distributions of the cluster members and of the cluster red sequence members (hatched histogram) are shown in the inset plot, where we also report the corresponding values of the galaxy peculiar velocities along the top horizontal axis...... 42 List of Figures xix

3.4 (left): Colour-magnitude diagram of the central region of XMM1229. The colours are measured on the F775W and F850LP PSF cross-convolved images, adopting fixed circular apertures with 100 radius ( 8 kpc at z = 0.98). ∼ Black dots represent all the photometrically selected cluster members and green dots are the red sequence members. Red dots are spectroscopically confirmed cluster members. The vertical dashed line represents the flux limit of visual morphology (see 3.4.4) and the sloping dotted line is the § 90% completeness limit in the F775W and F850LP images. The black dashed line is the linear fit to the red sequence and the two dotted lines

represent the -4σc and +7σc envelopes delimiting the red sequence. The error bars represent the median colour errors along the red sequence in bins of 0.5 magnitudes. (right): Observed colour-magnitude diagram in the outskirts of XMM1229. No cut in photometric redshifts is applied for this sample. Grey dots are all the galaxies observed in the cluster outskirts, red dots are spectroscopically conﬁrmed cluster members and green dots are red sequence galaxies selected as described in 3.4.3. Error bars represent § the median colour errors along the red sequence in bins of 0.5 magnitudes. The meaning of the lines is the same of the left panel...... 49 xx List of Figures

3.5 Morphological evolution of red sequence galaxies in clusters at 0.04 < z < 0.98. Left: magnitudes are normalised to the brightest bin in each sample. (Top left panel): morphological fractions along the red sequence of XMM1229. (middle left panel): fraction of morphological types along the red sequence of the spectroscopically conﬁrmed MORPHS cluster members (0.3 < z < 0.6). (bottom left panel): morphological fractions along the red sequence of the WINGS spectroscopically conﬁrmed cluster members. The analysis in WINGS is restricted to objects with V < 18.0 mag, corresponding to at least 50% spectroscopic completeness. The bright end of the red sequence is dominated by elliptical galaxies as in MORPHS and XMM1229. At intermediate and low luminosities, S0 galaxies are the most frequent

morphological class. On the top we report the apparent z850 magnitude

scale (down to the limit z850 = 24.0 mag considered in this paper). Right: morphological fractions as a function of V absolute magnitude. (Top right panel): XMM1229; (middle right panel): MORPHS; (bottom right panel): WINGS. In order to match the scales of the three samples, the magnitudes were passively evolved to z = 0. The absolute magnitude axis is reproduced on the top of the plot. For clarity, in all the plots, the points for each morphological type are shifted along the x-axis by 0.04 mag...... 56

3.6 Performance of GALFIT on the F850LP image. Grey points represent the results for each retrieved simulated galaxy image, while black points and error bars are the median and half of the 68% width of the distribution of

each quantity in each bin on the x-axis. (Top left): galaxy magnitude: ∆z850

as a function of input magnitude; (top right): half light radius: ∆R50 as a function of input magnitude; (bottom left): Sérsicindex: ∆n as a function of input magnitude; (bottom right): Sérsicindex: ∆n as a function of input Sérsicindex. The width of the input magnitude bins is 0.5 mag, while the width of the Sérsicindex bins is 1. The vertical dashed line represents the z = 24.0 mag limit for visual morphological classification (see 3.4.4). . . 59 850 § List of Figures xxi

3.7 (Left panel): Sérsicindex versus z850 magnitude along the red sequence. Galaxies with later morphological types tend to have lower Sérsicindices. The error bars represent the median GALFIT errors on the Sérsicindex in

bins of 0.5 magnitudes. On the top we also report the Vega MV passively evolved magnitudes (see 3.5.2 for details). (Central panel): Sérsicindex § vs stellar mass along the red sequence. There is weak correlation between Sérsicindex and stellar mass and disc-dominated galaxies tend to have lower masses. (Right panel): distribution of the values of the Sérsicindex for red sequence galaxies...... 65

3.8 Co-added FORS2 spectra of morphologically classiﬁed red sequence galaxies. The spectra are plotted in the observer frame and are shifted along the vertical axis for clarity. From top to bottom: elliptical, bulge-dominated, early disc-dominated galaxies. We highlight the Hδ, Hγ and G4300 features, and mask the O atmospheric absorption (A band, λ 7604 A).˚ 2 ∼ It can be seen that the Hδ absorption is stronger in early disc-dominated galaxies. This suggests that these galaxies probably just joined the red sequence after cessation of star formation. It can be noted a remarkable [OII] emission feature in the elliptical spectrum, which is contributed by the galaxies XMM1229 145 and XMM1229 73. As reported by Santos et al. (2009), these two spectra also show [OIII] in emission, which may indicate ongoing star formation...... 72

4.1 Filter combinations adopted for the study of the red sequence in the HCS clusters. Clusters are ordered by increasing redshift. The solid black line is a template spectral energy distribution of an elliptical galaxy from the library of Coleman et al. (1980). The blue and red solid lines are the blue and red filters adopted for the cluster red sequence, respectively. The blue and red dashed lines are the blue and red filters adopted in the estimation of field contamination in the GOODS-N/S fields. The names of the cluster (top row) and field (bottom row) filters are written in each plot together with the names and redshifts of the clusters. The vertical dashed lines represent the positions of the 4000 A˚ break at the redshifts of the clusters. . 77 xxii List of Figures

4.2 Eﬀect of the ISAAC J(Js)-band incompleteness on the cluster multiband datasets. The plot shows the (i z ) vs z colour-magnitude diagram 775 − 850 850 of the cluster RCS0220 (z = 1.03). Red diamonds are all the objects with J > 22.4 mag, the 90% completeness limit in this band. The black solid line is the best-ﬁt straight line to the observed red sequence and the dotted diagonal lines represent the red sequence boundaries. The diagonal dashed

line is the 90% magnitude completeness limit in the i775 band. As it can be seen, in the multiband sample the red sequence becomes incomplete already

at z850 = 23.0 mag, which is brighter than the 90% completeness limit in this band (26.7 mag). This eﬀect can result in artiﬁcially high luminous- to-faint ratios (See 4.3.2)...... 79 §

4.3 Left: Observed colour-magnitude diagrams of the HCS clusters within 0.54 × R200 from the cluster centroid (cluster central region, or cluster centre). Grey points are all the galaxies observed in the cluster centre, green diamonds are the members of the observed red sequence and red squares are the spectroscopically conﬁrmed cluster members. The black dashed line is the best-ﬁt straight line to the observed red sequence and the dotted parallel lines represent the red sequence envelope determined as discussed in 4.2. § The diagonal solid lines correspond to the 90% completeness limit while the dot-dashed diagonal lines represent the boundaries of the red sequence with the alternative selection ∆C < 3σ ...... 82 | | 22

4.3 Continued. Right: Observed colour-magnitude diagrams in the GOODS- N/S control fields used for field subtraction. Galaxies within a projected spatial region with radius 0.54 R are plotted in each figure. The fit and × 200 boundaries to the observed cluster red sequences, and the 90% magnitude completeness limit, are also plotted. The vertical dotted lines represent the apparent magnitude of the brightest red sequence galaxy in each cluster. Galaxies falling within the magnitude and colour ranges of the observed cluster red sequences are plotted as green diamonds. It can be seen that there are few field galaxies with magnitudes and colours in the ranges of the observed cluster red sequences. As a result, the field contamination of the cluster red sequence is low...... 83

4.3 Continued...... 84 List of Figures xxiii

4.4 Distribution of the red sequence parameters after 200 Monte Carlo simulations for the statistical estimate of cluster membership. The vertical solid line is the median value of each parameter while the vertical dashed lines represent the boundaries of the 68% conﬁdence intervals of the distributions. It can be seen that the red sequence slope and zero-point are highly correlated: steeper slopes correspond to redder zero-points (bottom-right panel)...... 87

4.5 Colour image of the cluster RX J0152.7-1357 (z = 0.84) obtained by combining the ACS F625W, F775W and F850LP images. In this and all the other colour images of the HCS clusters, the North is at the top and the East at the left of each ﬁgure. The scales of this and the other HCS colour images refer to the physical projected distance, at the redshift of the cluster, corresponding to 0.50...... 90

4.6 Colour image of the cluster RCS 2319.8+0038 (z = 0.91) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image...... 91

4.7 Colour image of the cluster RCS 0220.9-0333 (z = 1.03) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image. The centre of the cluster is eclipsed by a nearby face-on spiral galaxy which prevents the location of a BCG of this system from being identiﬁed...... 92

4.8 Colour image of the cluster RCS 2345-3633 (z = 1.04) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image...... 93

4.9 Colour image of the cluster XMM J0223-0436 (z = 1.22) obtained by combining the ACS F775W and F850LP images, and the HAWK-I J band image. 94

4.10 Colour image of the cluster RDCS J1252.9-2927 (z = 1.24) obtained by combining the WFC3 F105W, F125W and F160W images...... 95

4.11 Colour image of the cluster XMMU J2235.3-2557 (z = 1.39) obtained by combining the WFC3 F105W, F125W and F160W images...... 96

4.12 Colour image of the cluster XMMXCS J2215-1738 (z = 1.46) obtained by combining the ACS F775W and F850LP images, and the HAWK-I J band image...... 97 xxiv List of Figures

4.13 Evolution of the parameters of the rest-frame red sequence and comparison with the literature. From left to right: (B V ) vs V (left column), (U V ) − − vs V (middle column), (U B) vs B (right column). From top to bottom: − (B V ) and (U V ) colours at V = 20.5 mag, and (U B) colour − − − − at B = 21.4 mag (top row), rest-frame slope (middle row), rest-frame − intrinsic scatter (bottom row). The hydrodynamic simulations of Romeo et al. (2008) (cyan octagons) predict a strong evolution of the slope of the red sequence which becomes positive at z > 1. This is not in agreement with the observational results of the present work and other works on clusters at z > 0.8 ( e.g. Meyers et al. 2012 and Mei et al. 2009). The intrinsic scatter exhibits a wide range in (U V ) colour at z > 0.8 (0.01 < (U V ) < 0.25). − − The semi-analytic models of Menci et al. (2008) predict a large intrinsic scatter, although still consistent with the results of the present work and Mei et al. (2009). The (U B) colours at B = 21.4 mag in Mei et al. − − (2009) are systematically redder ( 0.1 mag) than those estimated in the ∼ HCS (top-right panel). All magnitudes are in the Vega system...... 100

4.14 The luminous-to-faint ratio (L/F ) of the cluster red sequence in the HCS. Top-left panel: redshift evolution of L/F . Top-right panel: L/F as a func-

tion of cluster halo mass MDM . The diamonds correspond to L/F of the

composite WINGS spectroscopic sample. The MDM estimate for WINGS corresponds to the median halo mass while the error bars correspond to the

width of the 68% conﬁdence interval of the MDM distribution. The bottom panels show L/F as a function of cluster redshift (left) and cluster halo mass (right) for the alternative selection of galaxies on the red sequence

with ∆C < 3σ22...... 109

4.15 Red sequence number counts in the HCS and WINGS. HCS clusters are ordered by increasing halo mass. Black points and solid connecting lines are for HCS, red circles and dotted connecting lines are for WINGS. The solid black line represents the median Schechter (1976) function of clusters in

the same redshift and MDM ranges of WINGS from Crawford et al. (2009).

Number counts are shown as a function of VAB-band absolute magnitude passively evolved to z = 0. The grey crosses and the solid connecting grey lines are the HCS red sequence number counts with the alternative selection

of red sequence galaxies with ∆C < 3σ22 ...... 111 List of Figures xxv

4.16 Red sequence number counts at low and high redshifts and low and high cluster halo mass. Colours and symbols are as in Fig. 4.15. Top panels: red sequence number counts in clusters at 0.8 < z < 1.1 (left) and 1.1 < z < 1.5 (right). Central panels: red sequence number counts in clusters with 14 14 MDM < 5 10 M (low-mass sample, left) and MDM 5 10 M (high- × ≥ × mass sample, right). Bottom panels: the same as in the middle panels but excluding RX0152 and RDCS1252, the two clusters with large diﬀerences between X-ray and weak-lensing halo masses (see discussion in 4.4.2). . . . 116 §

5.1 Observed colour-magnitude diagrams of the individual HCS clusters with the morphologically classiﬁed red sequence galaxies highlighted by diﬀerent

colours and symbols. All the red sequence galaxies with z850 < 24.0 mag were selected for morphological classiﬁcation. Grey points are all galaxies observed in the ﬁeld of each cluster within 0.54 R from the cluster × 200 centroid. Red crosses are elliptical galaxies, orange diamonds are bulge- dominated/S0 galaxies (BD), green triangles are early-type disc-dominated galaxies (EDD), and blue squares are late-type disc-dominated and irregular galaxies (LDD+Irr)...... 120

5.1 Continued. The lines are the same as in Figure 4.3. The solid lines represent the ﬁt to the observed red sequence. The dotted lines mark the boundaries of the red sequence. The diagonal dashed lines represent the 90% colour

completeness limits. As it can be seen, due to the magnitude cut z850 = 24.0, the red sequence is not covered down to the faint end in all the clusters and in some cases, as in XMMXCS2215 (bottom-left panel) only galaxies which are bluer than the best-ﬁt line fall in the sample...... 121

5.2 (a): Distribution of the ellipticity for galaxies classified as ellipticals, bulge- dominated and early disc-dominated in the HCS (top) and WINGS (bottom) samples. (b): the same as in (a) but with the alternative classification discussed in 5.1.3. Galaxies for which 50% of the classifiers agreed on one § type and the other 50% on a different type are now assigned to the latest- type class of the two. If 2 classifiers classified the galaxy as elliptical and 2 classified it as S0, the galaxy was assigned to the class of the S0s. In this way part of the population of S0 galaxies with low ellipticity (nearly face-on, e < 0.2) is recovered...... 125 xxvi List of Figures

5.3 Morphological parameters of HCS red sequence galaxies as estimated by galSVM. Top-left: asymmetry vs concentration, top-right: Gini coeﬃcient vs

concentration, bottom-left: M20 vs concentration, bottom-right: asymmetry vs Gini coeﬃcient. A combination of concentration, Gini coeﬃcient and

M20 is eﬀective in dividing the sample between early- and late-type galaxies. Disc-dominated and irregular galaxies are less concentrated, have lower Gini

coeﬃcients, and higher M20 values, indicating the presence of a higher level of substructure with respect to elliptical and S0 galaxies...... 130

5.4 Distribution of the Sérsicindex values, n, of elliptical (red histogram), bulge-dominated (orange histogram) and early-type disc-dominated (green filled histogram) galaxies in common with the sample of Delaye et al. (2014). The distributions for elliptical and S0 galaxies are statistically different

(PKS = 0.002) although the mean values of n are consistent within 1σ (see 5.1.3)...... 132 §

5.5 Left: Background corrected morphological fractions as a function of VAB absolute magnitude along the cluster red sequence in HCS (top panel) and WINGS (bottom panel). Right: Background corrected morphological fractions as a function of stellar mass along the cluster red sequence in HCS (top panel) and WINGS (bottom panel). The vertical dotted lines

at log(M∗/M ) = 10.7 and log(M∗/M ) = 11.5 represent the stellar mass limit of the HCS morphological sample and the maximum stellar mass of HCS red sequence galaxies, respectively. The plots show that elliptical galaxies are the dominant morphological class in the HCS clusters at all luminosities and masses while the red sequence of the WINGS clusters is

dominated by ellipticals at VAB < 21.0 mag (log(M∗/M ) > 11.5) and by − S0s at VAB > 21.0 mag (log(M∗/M ) < 11.5) ...... 137 −

5.6 Distributions of galaxy colour measured with respect to the red sequence,

(C CRS), for galaxies of diﬀerent morphological types in HCS. Early-type − disc-dominated, late-type disc-dominated and irregular galaxies are grouped into one class of late-type galaxies (blue histogram). Number counts are corrected for ﬁeld contamination using Equation 5.2...... 142 List of Figures xxvii

6.1 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with LRIS in the cluster RCS 2319.8+0038 (z = 0.91). Each image is 81 pixels on each side, corresponding to 4.0500. The ID, coordinates,

z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.1 below. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table...... 146

6.2 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with GMOS-N in the cluster XMMU J1229+0151 (z = 0.98). Each image is 81 pixels on each side, corresponding to 4.0500. The ID, coordinates,

z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.2. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table...... 147

6.3 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with GMOS-N in the cluster RCS 0220.9-0333 (z = 1.03). Each image is 81 pixels on each side, corresponding to 4.0500. The ID, coordi-

nates, z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.3. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table...... 148 xxviii List of Figures

6.4 Observed colour-magnitude diagrams of the clusters RCS 2319.8+0038 (z = 0.91, top-left), XMMU J1229+0151 (z = 0.98, bottom-left) and RCS 0220.9- 0333 (z = 1.03, top-right). Magenta stars represent the new spectroscopically confirmed cluster members, black pentagons are spectroscopic interlopers found from the GMOS-N and LRIS spectra, red crosses are spectroscopically confirmed elliptical galaxies, orange diamonds are spectroscopically confirmed S0 galaxies, green triangles are spectroscopically confirmed early-type disc-dominated galaxies. The diagonal solid lines are the best-fit straight lines to the observed red sequence of each cluster, the diagonal dotted lines mark the boundaries of the red sequence estimated as in Section 4.2.1, the diagonal dashed lines are the 90% magnitude completeness limits of the three samples. The vertical dashed lines mark the magnitude limit of

the morphological sample (z850 = 24 mag) and the boundary between bright

and faint spectroscopic targets (z850 = 22.0 mag). Morphologically classi- ﬁed galaxies that were detected in the Gemini or Keck spectra are marked by one of the morphological symbols (cross, diamond or triangle) and a star or pentagon. Error bars on the objects observed with GMOS-N and LRIS correspond to the photometric uncertainties estimated as discussed in Section 4.1.1...... 152

6.5 Composite spectra of morphologically classiﬁed red sequence galaxies in HCS clusters at z 1. From top to bottom: elliptical galaxies in RCS ∼ 2319.8+0038 (z = 0.91), elliptical galaxies in XMMU J1229+0151 (z = 0.98), elliptical galaxies in RCS 0220.9-0333 (z = 1.03), S0 galaxies in RCS 0220.9-0333 (z = 1.03). Galaxies were morphologically classiﬁed as discussed in Chapter 5. Spectra are in the observer reference frame. Age- and metal-sensitive absorption features are marked with vertical dashed lines drawn at the central wavelength of each feature at the redshifts of the clusters. Unlike elliptical galaxies, the CaII H line in the S0 composite spectrum of RCS 0220.9-0333 is more prominent than the CaII K line, suggesting a larger contribution from H in the observed CaII H feature. The Hδ line also appears more prominent than in elliptical galaxies in the same cluster. These two observations suggest that S0 galaxies host younger stellar populations than elliptical galaxies...... 153 List of Figures xxix

6.6 Composite spectra of red sequence galaxies in RCS 0220.9-0333 (z = 1.03) at diﬀerent apparent magnitudes. Top panel: red sequence galaxies with z < 22.0 mag; bottom panel: red sequence galaxies with z 22.0 mag. 850 850 ≥ The Hδ line appears more prominent in fainter galaxies than in brighter galaxies, suggesting that star formation was quenched more recently in less massive galaxies. The H6 index (rest-frame λ 3900.0 A)˚ seems also ∼ detectable in faint galaxies...... 156

6.7 Spectra of the S0 galaxy RCS2319 201 (top panel) and of the early-type disc-dominated galaxy RCS2319 174 (bottom panel) observed with LRIS in the cluster RCS2319 (z = 0.91). While the disc-dominated galaxy has a strong Hδ absorption line, the S0 galaxy reproduces the trend in the Ca H and K lines observed in the composite S0 spectrum of RCS0220...... 157

A.1 Morphologically classified red sequence galaxies in the central region (i.e. within 0.6 Mpc from the cluster centre) . Position and redshift (photometric and/or spectroscopic) of each object are listed in Table 3.3. The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field used in morphological classification (see Section 3.4.4)...... 196

A.2 Morphologically classified red sequence galaxies in the cluster outskirts (i.e. between 0.6 Mpc and 1.04 Mpc from the cluster centre). Only spectroscopically confirmed cluster members are shown as in this work we do not determine the membership of individual galaxies for this subsample. Posi- tion and redshift of each object are listed in Table 3.3. The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field used in morphological classification (see Section 3.4.4). The morphological classification of the objects in this subsample corresponds to the outcome of galSVM...... 197

A.3 Spectroscopically confirmed XMM1229 members excluded from the analysis in this paper. XMM1229 183, XMM1229 308, XMM1229 487 are blue cloud galaxies, while XMM1229 316 was excluded because SExtractor was unable to return a reliable estimate of the aperture magnitude (SExtractor FLAGS=16). The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field. For this subsample the visual morphological classification was performed by P. C...... 197 xxx List of Figures

A.4 Spectroscopically confirmed XMM1229 members excluded from the analysis in this paper because falling outside the ACS field. The classification was performed visually by P. C. on image cutouts taken from the HAWK-I Ks band image of the XMM1229 field and shown in this figure...... 198

B.1 Morphological parameters for red sequence galaxies in the centre of XMM1229

(Rcluster < 0.54 R ). Symbols and colours are the same used in Fig. 3.5. × 200 As expected, disc-dominated galaxies tend to have smaller values of Con- centration and Gini coeﬃcient. All morphological types show comparable

values of M20...... 203 List of Tables

2.1 The Hawk-I Cluster Survey (HCS) sample used in the present thesis with the clusters listed in order of increasing redshift. The dark matter halo masses

MDM quoted in the ﬁfth column from the left are taken from Jee et al. (2011). We do not consider in this table the other two clusters SpARCS J003550-431224 (z = 1.34) and ClG J0218.3-0510 (z = 1.62), which are not used in the analysis of the HCS red sequence in this thesis...... 17

2.2 Summary of the HCS imaging observations. The 90% magnitude completeness limit is estimated as described in 3.3.2 . The image quality FWHM § quoted for WFC3 corresponds to the intrinsic PSF FWHM obtained by de- convolving the observed PSF by the detector pixel response function (see 2.2.2 for a discussion). Each line refers to an instrument; e.g. the third § line of the XMM1229 entry refers to the WFC3 observations of this cluster. 22

2.3 Summary of the GMOS-N and LRIS observations of XMM1229, RCS0220 and RCS2319...... 25

2.4 Spectroscopically conﬁrmed members in the cluster XMM1229 from the GMOS-N observations. All spectroscopic cluster members are included in the table regardless of their colour...... 28

2.5 Spectroscopically conﬁrmed members in the cluster RCS0220 from the GMOS- N observations. All spectroscopic cluster members are included in the table regardless of their colour...... 29

2.6 Spectroscopically conﬁrmed members in the cluster RCS2319 from the Keck/LRIS observations. All spectroscopic cluster members are included in the table regardless of their colours...... 29

3.1 Summary of the HST observations of XMM1229...... 40

3.2 Summary of the ground based observations of XMM1229...... 40

xxxi xxxii List of Tables

3.3 Red sequence members of XMMU J1229+0151 (XMM1229). Red sequence members are photometrically selected as explained in 3.4.1 in the clus- § ter centre. Spectroscopically confirmed members are indicated with their spectroscopic redshift and the estimated photometric redshift. The table is divided into the following sections: (1) red sequence members within 0.6 projected Mpc from the cluster centre (centre sub-sample); (2) spectroscopically confirmed red sequence members between 0.6 and 1.04 projected Mpc from the cluster centre (outskirts); (3) spectroscopically confirmed cluster members in the blue cloud or with corrupted photometry; (4) spectroscopically confirmed cluster members detected only in the HAWK-I Ks image. Objects in Sections (3) and (4) of this table are excluded from the analysis in this paper...... 50 3.4 MORPHS visual classification and type conversion (See 3.4.4 for details § on the morphological scheme adopted for XMM1229)...... 62 3.5 WINGS morphological classification and type conversion (See 3.4.4 for § details on the morphological scheme adopted for XMM1229)...... 63 3.6 Median Sérsicindices and stellar masses, per morphological type, on the red sequence of XMM1229...... 70

4.1 Photometric set-up of red sequence and Pfield estimation in each HCS cluster. 78 4.2 Parameters of the ﬁts to the observed (top row) and ﬁeld-corrected (bottom row) red sequences with their respective 1σ uncertainties. Also shown are

the κl and κh factors adopted in the selection of the red sequence and the

values of the probability Pfield used in the estimation of field contamination. The uncertainties are discussed in 4.2.1...... 88 § 4.3 Parameters of the fits to the rest-frame red sequence with their respective 1σ uncertainties. In each entry: (B V ) vs V (top row), (U V ) vs V − − (middle row), (U B) vs B (bottom row). Magnitudes and colours are in − the Vega system...... 102 4.4 Red sequence luminous-to-faint ratio (L/F ) of the HCS and WINGS clusters. The values of the cluster halo mass from Jee et al. (2011) are shown in the fourth column from the left. The halo mass and uncertainties quoted for WINGS refer to the median and 68% confidence interval of the halo mass distribution of the subsample of WINGS clusters used in the comparison with HCS. The second row in each entry refers to the red sequence selection

with ∆C < 3σ22...... 104 List of Tables xxxiii

5.1 Mean values and standard errors of concentration, asymmetry, Gini coeﬃ-

cient and M20. The estimates for each morphological type and for early- and late-type galaxies are all shown...... 126

5.2 Median ellipticities, colours relative to the red sequence (C CRS), and total − fractions of red sequence galaxies FT in HCS and WINGS. The uncertainties

on the ellipticity and (C CRS) correspond to the boundaries of the 68% − conﬁdence interval of the distribution of each quantity. The uncertainties on the total morphological fractions correspond to the boundaries of the binomial 68% conﬁdence intervals estimated as in D’Agostini (2004) and

Cameron (2011). The median values of (C CRS) and ellipticity of WINGS − take into account the incompleteness of the spectroscopic sample. The

median (C CRS) of HCS is corrected for background contamination. . . . 131 −

6.1 Table 6.4. Details of the composite HCS red sequence spectra. Ng is the

number of galaxies in each composite spectrum. texp is the sum of the exposure times of all the spectra. The signal-to-noise ratio S/N is estimated at wavelengths longer than the position of the 4000 A˚ break, where most absorption features fall. Bright and faint composite spectra in RCS0220 correspond to all red sequence galaxies with z < 22.0 mag and z 850 850 ≥ 22.0 mag, respectively. Neff is the number of co-added spectra in each subsample taking into account multiple exposures...... 155

1 Introduction: The Environmental Drivers of Galaxy Evolution

Galaxies are complex systems. Understanding their evolution has been an active area of research since their discovery. Modern ground- and space-based observatories, working in diﬀerent regions of the electromagnetic spectrum, from the X-ray to the radio, have revealed a wide variety of phenomena. Modern supercomputing facilities, on the other hand, have allowed astrophysicists to develop theoretical models and simulate the evolution of galaxies across cosmic time. Putting together observations and simulations to build a coherent theoretical structure that consistently explains the formation and evolution of galaxies is one of the main challenges of today’s astrophysics.

In this thesis I will focus on the environmental drivers of galaxy evolution using data from space and ground-based observatories. In particular, I will focus on the evolution of galaxies in rich clusters, comparing a sample of 9 distant galaxy clusters at redshifts 0.8 < z < 1.5 with other publicly available samples of clusters at lower redshifts. The sample used in this thesis is part of the HAWK-I Cluster Survey (HCS, Lidman et al. 2013), a program aimed at the observational study of galaxy populations in distant clusters with the High Acuity Wide-ﬁeld K-band Imager (HAWK-I, Kissler-Patig et al. 2008) at the European Southern Observatory Very Large Telescope (ESO/VLT).

This chapter sets the context for this thesis by outlining the key topics and issues relevant to the study, reviewing the literature for the most pertinent work done previously in this ﬁeld, and distilling this into a summary of what motivates this thesis scientiﬁcally.

1 2 Chapter 1. Introduction

Figure 1.1 The core of the cluster of galaxies Abell 370 at z = 0.375 as seen with the HST/ACS camera. Elliptical and S0 galaxies dominate the galaxy population of this cluster. The colour image was obtained by combining ACS images taken with the F475W, F625W, and F814W ﬁlters

1.1 Galaxy Clusters as Observational Laboratories

The evolution of galaxies is regulated by internal and external processes. The first are related to galaxy stellar mass, while the second are related to the environment in which galaxies reside. In fact, galaxies are not isolated but are clustered over a range of scales from diffuse groups to massive and dense clusters and superclusters containing thousands of members and extending over several megaparsecs (Figure 1.1). One of the key problems in modern astrophysics is to understand how the evolution of galaxies is affected by the environment in which they reside.

Dense environments, such as clusters and groups, where galaxies undergo frequent and close gravitational encounters, can increase the probability of interactions that promote or quench star formation. While feedback processes from star formation, supernova ex- plosions and active galactic nuclei are proposed as the main internal drivers of galaxy evolution (Croton et al., 2006; Cantalupo, 2010), acting regardless of the environment, a large number of environment-related physical mechanisms have been proposed as external drivers of galaxy evolution. Gravitational interactions between galaxies such as mergers 1.1. Galaxy Clusters as Observational Laboratories 3 and harassment (Lavery & Henry, 1988; Moore et al., 1996), or ram-pressure stripping as a galaxy falls into the cluster and travels through the intracluster medium (Gunn & Gott, 1972; Gavazzi et al., 2013) are among the most extreme examples. The intracluster medium can also trigger star formation by compressing the interstellar gas, as shown by Bekki & Couch (2003). Strangulation (Larson et al., 1980) -that is, the removal of hot gas from the galaxy haloes due to tidal interactions with neighbouring galaxies- results in the gradual quenching of star formation after the entire cold gas reservoir present in the galaxy disc is consumed. Bekki & Couch (2011) show that tidal interactions with neighbouring galaxies can result in the bulge growth of a spiral galaxy due to star formation triggered by gas flowing towards the centre. As a consequence, the galaxy consumes its entire gas reservoir and undergoes a morphological transformation into a S0 (see Section 1.3). Although all these physical processes can lead to the accelerated consumption or loss of cold gas in galaxies, it is not clear what is the relative importance of each mechanism with respect to the other. There is evidence that the evolution of the brightest cluster galaxies (BCG) is driven, at least at z 1, by major (i.e. with stellar mass ratios between ∼ companions close to 1) gas poor (dry) mergers (Lidman et al., 2013), while ram pressure stripping is the main mechanism responsible for star formation quenching in low-mass cluster members (Gavazzi et al., 2013). It is also not well understood at which epoch particular mechanisms first come into play. In order to understand the effect of environment on galaxy evolution it is therefore necessary to study galaxies in different environments and at different redshifts. Galaxy clusters are the most massive virialised large-scale structures in the universe and for this reason they are very important in both astrophysics and cosmology. In particular, the broad range of environments from the dense cores to the diffuse infall regions, spanned by clusters, make them interesting for studying the effects of environment on galaxy evolution (De Lucia et al., 2007a; Lemaux et al., 2012; Fritz et al., 2014). According to the hierarchical scenario of structure formation, galaxy clusters formed after the collapse of the highest density peaks in the primordial matter distribution, sub- sequently accreting other smaller dark matter haloes (Fakhouri et al., 2010; Chiang et al., 14 15 2013). Their mass ranges from 10 to 10 M and they can extend for up to 10 Mpc. ∼ However, galaxies constitute only 5% of the total cluster mass, the rest consisting of ∼ the hot intracluster gas ( 15%), detectable at X-ray and radio wavelengths (Reiprich & ∼ Böhringer,2002; Reichert et al., 2011; Planck Collaboration et al., 2013), and dark matter (e.g. Clowe et al. 2006, Jee et al. 2011.) The study of the gaseous and dark matter components allows the overall structure and 4 Chapter 1. Introduction mass distribution of clusters to be investigated, which turns out to be very complex and heterogeneous. In fact, galaxy clusters can be broadly divided into two regions: the dense and virialised core, and the sparser and less virialised outskirts. This split is well observed +0.92 15 in local massive clusters, such as Coma, which has a mass of 2.65 10 M (Kubo ∼ −0.79 × et al., 2007)1, but is less evident at lower cluster masses (e.g. in the Hercules cluster, Figure 1.2). Furthermore, galaxy clusters are not isolated systems. They interact with the surrounding environment accreting isolated galaxies or groups (Lemaux et al., 2012; Fogarty et al., 2014). In the extreme case of mergers with other clusters, the core-outskirts division is broken and a new large scale structure is formed with different dynamical and physical properties with respect to the progenitor clusters (Roettiger et al., 1997). The structural complexity of galaxy clusters, as well as their evolving dynamical state, is reflected in the properties of their member galaxies. In rich and dynamically relaxed clusters, up to z 1.5, a correlation is observed between galaxy morphology and local ∼ density, whereby massive elliptical and S0 galaxies are more frequent in higher-density regions (Dressler, 1980; Postman et al., 2005; Hilton et al., 2009; Mei et al., 2012) (see Figure 1.1). In most cases clusters can have one or two giant ellipticals in their dense cores 12 which can reach stellar masses of up to 10 M . Spiral and irregular galaxies, instead, are more frequent towards the cluster outskirts. However, less relaxed clusters present an irregular and asymmetric matter distribution and the separation between core and outskirts becomes less sharp or absent (Longair, 2008). In these cases spiral and irregular galaxies can also be found at the centre of the cluster as it is observed in the nearby Hercules cluster (Abell 2151, Figure 1.2). The interplay between galaxy morphology and cluster dynamical state has been at the basis of morphological classifications of clusters. A popular scheme is that introduced by Bautz & Morgan (1970), which divides clusters in three types according to the brightness and morphology of their central galaxies. Thus Type I and Type II clusters contain giant ellipticals in their cores (like the Coma cluster), whereas Type III clusters do not show the presence of any prominent bright elliptical (like the Hercules cluster in Figure 1.2). Spiral-rich clusters are normally found among Type III systems. There have been several other morphological schemes proposed for the classification of galaxy clusters and detailed reviews and comparisons are given in Bahcall (1977) and Longair (2008). Since old and non star-forming (quiescent), galaxies have generally an elliptical or lenticular morphology, the morphology-density relation translates into a star-formation vs

1We converted the result of Kubo et al. (2007) to our adopted cosmology. A stated in the following −1 −1 chapters, we adopt a ΛCDM ﬂat cosmology with ΩΛ = 0.73, ΩM = 0.27, and H0 = 71 km · s · Mpc . +0.65 15 −1 The result obtained by Kubo et al. (2007) and quoted in the paper is: MDM = 1.88−0.56 × 10 h M . 1.1. Galaxy Clusters as Observational Laboratories 5

Figure 1.2 Colour-composite image of the spiral-rich Hercules cluster (Abell 2151, z = 0.036) obtained with the ESO/VLT Survey Telescope (VST). Unlike Abell 370 in Fig- ure 1.1, dominated by elliptical and S0 galaxies, the core of this cluster is also populated by spiral galaxies. (Credit: ESO/INAF-VST/OmegaCAM. Acknowledgement: OmegaCen/Astro-WISE/Kapteyn Institute). density relation, whereby the fraction of actively star-forming galaxies decreases with local galaxy density. This is what is observed in low-redshift clusters (e.g. Fritz et al. 2014). However, the number density of star forming galaxies in clusters increases with redshift, a property known as the Butcher-Oemler Effect (Butcher & Oemler, 1978). Hilton et al. (2010) detected actively star-forming galaxies in the core of the cluster XMMXCS J2215- 1738 at z = 1.46 and Tran et al. (2010) and Santos et al. (2014) showed that the fraction of star-forming galaxies in the cluster ClG J0218.3-0510, at z = 1.62, is even increasing towards regions of higher galaxy density. Most studies of the star-formation vs density relation in clusters at z > 1.5 have focused so far only on 1 or 2 systems and differences 6 Chapter 1. Introduction between the analysis techniques adopted by different authors have led to disagreements between their results. Thus the conclusions of Quadri et al. (2012) on ClG J0218.3-0510 are not in agreement with those of Tran et al. (2010), while Fassbender et al. (2014) show that in the cluster XDCP J0044.0-2033, at z = 1.58, the fraction of recently quenched galaxies increases towards the dense cluster centre with the fraction of star-forming galaxies having an approximately constant radial trend. Interestingly, Andreon et al. (2014) find that the cluster JKCS041, at z = 1.8, is an unusually massive structure with an already well developed red sequence similar to that of low-redshift clusters. The discovery of mature clusters at z 2 with prominent red sequences has also been reported by Tanaka et al. ∼ (2010) and Gobat et al. (2011). Although a comprehensive picture of the star formation vs density relation at high redshift has not been reached yet, there is clear evidence of more intense star-forming activity in clusters at z > 1.5 with respect to their local counterparts (Hilton et al., 2010; Hatch et al., 2011).

1.2 The Cluster Red Sequence

Galaxies have a bimodal colour distribution with a red sequence and a blue cloud spanning about three orders of magnitude in luminosity and stellar mass. This colour bimodality is observed in all environments (see e.g. Strateva et al. 2001, Baldry et al. 2004, Valentinuzzi et al. 2011) and is the result of physical diﬀerences between galaxies, with red objects generally being old, quiescent, and with elliptical or S0 (early-type) morphologies, whereas blue galaxies are mostly younger, star-forming, and disc shaped or irregular (late-type). Galaxies lying between the blue cloud and the red sequence have intermediate properties. Often referred to as green valley or transition galaxies (Cortese & Hughes, 2009), they may be redder, gas-poor spirals with little ongoing star formation or bluer, early-type galaxies with traces of recent merger-induced star formation (McIntosh et al., 2014). In principle the evolution of a galaxy population in a particular environment can be tracked by studying the migration of galaxies from the blue cloud to the red sequence as star formation ceases. However, this simple scenario is complicated if, for example, a passive red sequence galaxy undergoes a merger with a gas-rich companion. In this case bursts of star formation are prompted in the merger remnant, and the galaxy moves from the red sequence towards the blue cloud (Faber et al., 2007). Although such episodes are relatively uncommon, they need to be taken into account (Cortese & Hughes, 2009). The colour-magnitude relation of red sequence galaxies can be modelled with a straight line (Figure 1.3) and is studied in terms of its zero point, slope, and intrinsic scatter 1.2. The Cluster Red Sequence 7

Figure 1.3 The (r Ks) vs Ks colour-magnitude diagram of the cluster RX J0152-1357 at z = 0.84 from Demarco− et al. (2010). The upper part of the diagram highlights the red sequence. The solid black line is the best-ﬁt straight line to the red sequence. See Figure 2 of Demarco et al. (2010) for the description of the symbols and colours. (Credit: Demarco, R. et al., 2010, ApJ, 725, 1252)

(Sandage & Visvanathan 1978, Bower et al. 1992). The main physical interpretation of the red sequence is that of a mass vs metallicity relation, with the intrinsic scatter about the best ﬁt straight line being attributed to age diﬀerences between galaxies (Kodama & Arimoto, 1997). Galaxy clusters exhibit tight red sequences, with intrinsic scatter

σi < 0.15 mag (Ellis et al. 1997, Jafféet al. 2011, Bower et al. 1992), underlining negligible age differences between cluster members. With the advent of large sky surveys, such as the Sloan Digital Sky Survey (SDSS, York et al. 2000), with samples of up to 106 galaxies, it has been possible to study the colour-magnitude relation in great detail, and Gallazzi et al. (2006) have shown that age and metallicity both contribute to the slope and scatter.

Interestingly, the bright end of the red sequence in local galaxy clusters is observed 8 Chapter 1. Introduction to deviate from the linear fit, bending towards bluer colours (e.g. Janz & Lisker 2009, Valentinuzzi et al. 2011, Jiménezet al. (2011)). This bending is theoretically explained in terms of the accretion of low-mass, low-metallicity galaxies on to the brightest cluster galaxies, resulting in an increase in the total mass and a decrease in the total metallicity of the merger remnants (Jiménezet al., 2011). Although clearly explained by numerical simulations, this scenario has not been verified yet by observations. There is evidence, however, that the structural properties of the brightest galaxies in local clusters are different from those of normal elliptical galaxies. For example, Graham et al. (1996) showed that BCGs in local clusters may have Sérsicindices up to n = 10.0, which is higher than the value expected in normal elliptical galaxies (n = 4, de Vaucouleurs law, de Vaucouleurs 1948). Such light distributions may be either the result of extremely high central light and mass concentrations, or they may be caused by stellar haloes surrounding these galaxies and producing low surface brightness envelopes. This latter class of elliptical galaxies is also known as the central dominant or cD type. The stellar halo of cD galaxies is thought to be the result of mergers with low-mass companions, as it is observed in the Virgo cluster (Rudick et al., 2010), a conclusion that seems to find support from observations of the intracluster light at z 1 (Burke et al., 2012). These results, together with the simulations ∼ of Jiménezet al. (2011) and the results of Lidman et al. (2013) and Ascaso et al. (2014) on the mass growth of BGCs, support the notion of a merger-driven evolution of the bright end of the red sequence in galaxy clusters (see also Faber et al. 2007).

Hydrodynamical and N-body simulations (Romeo et al., 2008) and semi-analytical models (Menci et al., 2008), based on hierarchical merging, show that the slope of the cluster red sequence should gradually evolve with redshift, ﬂattening at z 0.7 1.0 and ∼ − then having a positive slope at higher redshifts. Observations of high redshift clusters show that there has been little or no evolution in the slope of the red sequence since z 1.5. The red sequence at z > 0.8 is found to have a negative slope (e.g. Lidman et al., ∼ 2008; Mei et al., 2009; Demarco et al., 2010) and is dominated by old and passive galaxies with early-type morphologies (Mei et al., 2009). This highlights a major deﬁciency in the models.

The interplay between baryonic and non-baryonic matter in the most massive galaxies, which can result in the suppression of star formation once a critical halo mass is reached 12.5 ( 10 M , Dekel & Birnboim 2006), should also be included in the simulations, in order ∼ to reproduce the observed negative slopes. This has been done, for instance, by Gabor & Davé(2012). In their hydrodynamical simulations these authors proposed an empirical quenching mechanism which is based on the regulation of gas cooling and inflow towards 1.2. The Cluster Red Sequence 9 the galaxy. In summary, in all the haloes containing at least 60% of their gas above the critical temperature 2.5 105 K, the gas is prevented to cool and flow into the galaxy × thus starving it of new fuel for star formation. Although the authors do not attribute the origin of the heating that keeps the gas hot to a particular physical mechanism (AGN jets, cosmic rays or gravitational heating due to infalling gas clumps), they show that a red sequence with a negative slope can be reproduced already at z = 1 regardless of the local environment of the galaxy. They show that the red sequence builds up rapidly between 10.5 10 10 z=1 and z-2 at stellar masses M∗ > 10 M and M∗ < 10 M , while at M∗ 10 the ∼ red sequence is underpopulated, exhibiting a deficit of galaxies.

Observations of galaxy clusters at z < 1.6 show that the faint end of the red sequence becomes gradually less populated as the redshift of the cluster increases, supporting the notion that less massive galaxies have their star formation quenched at later times with respect to more massive galaxies. This scenario, initially proposed in Tinsley (1968), is also known as downsizing (Cowie et al., 1996) and has been mainly observed in the field (e.g. Pozzetti et al. 2010 and references therein). Although most authors in the recent literature also find evidence for downsizing in clusters (Capozzi et al., 2010; De Lucia et al., 2007b; Huertas-Company et al., 2009a; Lemaux et al., 2012; Rudnick et al., 2012; Fassbender et al., 2014), this conclusion is still debated. Thus, in the works of Andreon (2008), Andreon et al. (2014), Crawford et al. (2009), Lidman et al. (2008), and De Propris et al. (2013), the authors show that there is no deficit of galaxies at the faint end of the red sequence and that, if a deficit exists, then it should be either considered peculiar to some clusters or attributed to selection effects caused by sample incompleteness. These results suggest that the cluster red sequence was already assembled at z = 1.5 with no significant build-up at low masses at later epochs.

The separate study of the properties of cluster members as a function of mass and environment is an effective way to investigate the relative importance of internal and external quenching mechanisms. In fact, Peng et al. (2010b), using data from the SDSS and the zCOSMOS surveys (Lilly et al., 2007), showed that, up to at least z = 1, stellar mass and environment independently affect the quenching of star formation in galaxies. These authors did not consider, however, the highest galaxy densities of clusters. Nantais et al. (2013a) showed that in the cluster RX J0152-1357, at z = 0.84, low-mass galaxies at the highest local densities have the same morphological composition of high mass galaxies in the lowest density regions, suggesting that mass and environment act independently also in clusters of galaxies. Muzzin et al. (2012) studied the separate effects of stellar mass and environment on the evolution of galaxies in the 0.8 < z < 1.4 clusters of the 10 Chapter 1. Introduction

Gemini Cluster Astrophysics Spectroscopic Survey (GCLASS). They found that in all environments, which they characterised according to local galaxy density, more massive galaxies were less star-forming and older. However, when they considered galaxies at fixed stellar mass, they found no trend of star formation rate (SFR), specific star formation rate (SSFR = SF R/Mstar) and 4000 A˚ break with local environment. Their conclusion was that while stellar mass is the primary driver of star-formation quenching in galaxies, the effect of the environment is to set the overall timescales for the quenching of star formation. The denser the environment, the faster is the quenching of star formation. This conclusion is also in agreement with that of Rettura et al. (2011) at similar redshifts. Studies of galaxy evolution in clusters should also take into account the global properties of clusters: their total mass, velocity dispersion, gas temperature and level of substructures. It is well established that there exist scaling relations connecting together the total cluster mass with its gas temperature or velocity dispersion (Reiprich & Böhringer, 2002; Jee et al., 2011), so that the velocity dispersion or X-ray luminosity can be used as a proxy for cluster mass. However, it is not completely understood how these scaling relations evolve with redshift (see Jee et al. 2011 for a discussion). It is important to consider the cluster members in the general context of the cluster properties and study the cluster luminosity function or the morphology-density relation as a function of cluster mass. Valentinuzzi et al. (2011) found little dependence of the ratio between the numbers of luminous and faint galaxies on the red sequence (L/F ) with velocity dispersion in the z 0.05 clusters of the Wide-field Imaging Nearby-cluster Survey (WINGS, Fasano et al. ∼ 2006), while De Propris et al. (2013) found no difference between the luminosity functions of relaxed and merging clusters at 0.2 < z < 0.6. On the other hand, Lemaux et al. (2012), who dissected the Cl1604 supercluster at z 0.9 in its constituent groups and clusters, ∼ found that the red sequences of the more virialised structures have a more populated faint end. These structures are also the most massive in the whole supercluster, suggesting that cluster total mass may play a role in setting the timescales of star formation quenching. Interestingly, Gabor & Davé(2014) show that the most massive haloes correspond to the environments with the highest galaxy number density, suggesting that correlations with total mass may be another way of studying the evolution of the red sequence in clusters.

1.3 The Morphology of Galaxies and its Physical Implications

Since the work of E. P. Hubble in 1926 (Hubble, 1926), galaxies have been qualitatively categorised in two main classes of elliptical and spiral, with a transition lenticular or S0 class characterised by the coupled presence of a bulge and a disc, as observed in 1.3. The Morphology of Galaxies and its Physical Implications 11

Figure 1.4 The Hubble tuning fork diagram. From left to right: elliptical galaxies, S0 galaxies, normal spiral galaxies (top), barred spiral galaxies (bottom), and irregular galaxies. S0 galaxies are situated at the junction between the sequences of ellipticals and spirals. The images were obtained by combining HST/ACS exposures in the three filters F475W, F625W, and F814W. spiral galaxies, but without spiral arms, as observed in elliptical galaxies. The presence or absence of a stellar bar was used as the criterion to split the classes of spirals and S0s into two groups of barred and normal galaxies, respectively. Galaxies not showing any elliptical or spiral feature (e.g. the Magellanic Clouds) were classified as irregular. Hubble’s classification is well represented by a tuning fork shaped diagram, the Hubble Tuning Fork (Figure 1.4). Other authors have since added more levels of complexity to the Hubble tuning fork, introducing intermediate classes based on the bulge-to-total light ratio and the presence of rings (e.g. de Vaucouleurs 1959). However, the elliptical vs spiral separation has remained the basic criterion for any morphological classification. The evolutionary links between galaxies of different morphologies have been a puzzle in astrophysics since the publication of the Hubble tuning fork. The observations have shown that elliptical and S0 galaxies have older stellar populations than spiral and irregular galaxies. Furthermore, elliptical and S0 galaxies host little or no ongoing star formation making them the main constituent of the red sequence. These similarities between stellar populations have motivated astronomers to study together, on one hand, ellipticals and S0s as one class of early-type galaxies and, on the other hand, spirals and irregulars as one class of late-type galaxies. The differences between stellar populations suggest that galaxies were born spirals and then evolved through subsequent phases into S0s and ellipticals. 12 Chapter 1. Introduction

As mentioned in Section 1.1, galaxies in clusters follow a morphology-density relation which translates into a star-formation vs density relation (Dressler, 1980; Tran et al., 2009), according to which early type and quiescent galaxies are more frequent at high local densities. These two correlations underline important connections between morphological and physical properties of galaxies and, in particular, they underline the fact that the star formation quenching, chemical evolution, and dynamical evolution must all be taken into account simultaneously in order to understand the processes that drive the evolution of galaxies in clusters. The two correlations also show the importance of environment in the evolution of galaxies and the fact that interactions between galaxies, or between galaxies and the intracluster gas, are fundamental to understand the formation and evolution of elliptical galaxies.

Interestingly, the studies of the morphology-density relation in clusters at diﬀerent redshifts show that this correlation has been in place since at least z = 1.5 (Dressler et al., 1997; Postman et al., 2005; Hilton et al., 2009; Mei et al., 2012). Such studies also show that the fraction of elliptical galaxies follows the same trend with local density in clusters at all redshifts. However, the fraction of spiral galaxies increases with redshift, while that of S0s decreases. This conclusion is in agreement with the results of Fasano et al. (2001), who showed that the global fraction of elliptical galaxies in clusters at 0.0 < z < 0.6 is constant with redshift. These results suggest that elliptical and S0 galaxies follow diﬀerent evolutionary paths, and that spiral galaxies are somehow transformed into S0 galaxies.

There are many physical processes which result in the disruption of spiral arms in galaxy disk and the subsequent transformation from spiral to S0 galaxies. For instance, bulge growth and the subsequent depletion of gas reservoirs can be responsible for the disruption of spiral arms (e.g. Somerville et al. 2008). Tidal forces experienced by galaxies, as proposed in Bekki & Couch (2011) and discussed in Section 1.1, may be another candidate, especially in high-density environments, where S0 galaxies are more frequent. The removal of interstellar gas via ram pressure stripping or strangulation is a further trigger of the transformation from spiral to S0 as, once the gas is removed from the disc, the random motion of stars in the disc can no longer be suppressed (Binney & Merriﬁeld, 1998). As a result, spiral arms are disrupted.

It is worth noting that all these mechanisms presuppose that galaxies quench their star formation before being transformed into S0s, thus first migrating to the red sequence as red spirals, and then transforming their morphologies (Wolf et al., 2009; Masters et al., 2010; Kovaˇcet al., 2010). This implies that the timescales of star formation quenching must be shorter than those of morphological transformation. Interestingly, the cosmo- 1.4. Outline of the Thesis 13 logical hydrodynamical simulations of Taranu et al. (2014) and Bahé& McCarthy (2015) show that, typically, a galaxy which falls into a cluster and is stripped of its interstellar gas would require 3-3.5 Gyr to quench its star formation and land on the red sequence, while Bekki & Couch (2011) (but see also Moore et al. 1998) predict a time t 5 Gyr for the ∼ transformation from spiral to S0. Although these timescales are related to the particular processes and simulation set-up considered by each author, they are indicative of the fact that star-formation quenching of cluster galaxies is faster than morphological transformation. Sánchez-Blázquezet al. (2009) found that the fraction of late-type galaxies on the red sequence of clusters at 0.4 < z < 0.8 from the ESO Distant Cluster Survey (EDisCS, White et al. 2005) increases with redshift, suggesting that galaxies reach the red sequence as gas-poor spirals and then are transformed into S0s. This observational result supports the notion of a causal link between quenching and morphological transformation in which the cessation of star formation in spiral galaxies is the condition for their transformation into S0s. Poggianti et al. (2006) proposed an evolutionary scenario in which cluster early-type galaxies are made up of two populations, a pristine population coeval to the cluster, and a quenched population probably resulting from the morphological transformation of red spiral galaxies. Such a scenario is supported by the findings of Sánchez-Blázquezet al. (2009) and is in agreement with studies of the stellar populations in galaxies of different morphologies. In fact, Poggianti et al. (2001) showed that the average stellar age of low- mass S0 galaxies in the Coma cluster is lower than both elliptical and higher mass S0s. Tran et al. (2007) showed that red sequence S0s in the cluster MS 1054-03, at z = 0.83, are younger than ellipticals but older than late-type galaxies. Mei et al. (2009) showed that S0 galaxies on the red sequence of the 8 clusters of the Advanced Camera for Surveys (ACS) Intermediate Redshift Cluster Survey (Ford et al., 2004), at 0.8 < z < 1.3, are younger than ellipticals at the same luminosities, while at lower luminosities the ages of elliptical and S0 galaxies are similar. These results all suggest that elliptical and S0 galaxies have different evolutionary histories in clusters, and that the evolution of spiral and S0 galaxies may be related, with spirals being the progenitors of S0 galaxies.

1.4 Outline of the Thesis

Motivated by the theoretical and observational results discussed above, this thesis involves a detailed study of the properties of the red sequence in a sample of 9 galaxy clusters at 14 Chapter 1. Introduction

0.8 < z < 1.5 (corresponding to the range 6.8-9.3 Gyr in lookback time2) drawn from the HAWK-I Cluster Survey (HCS, Lidman et al. 2013) with the speciﬁc goals of determining:

the build-up of the red sequence as a function of stellar mass; • the morphological composition of red sequence galaxies at these early epochs; • the relation between morphology and spectral properties in red sequence galaxies; • the relationship between galaxy properties and cluster global properties. • The range spanned in the HCS is especially interesting because it is just after the probable reversal of the star-formation vs density relation (Tran et al., 2010), where most of the physical processes described in Section 1.1, responsible for star-formation quenching in clusters, were probably very active. Furthermore, as it will be shown in the following chapters, the HCS offers one of the best datasets of images and one of the richest redshift catalogues available to date for high-redshift clusters. Another important component of this thesis study is a program of deep multi-object spectroscopy of 3 of the HCS clusters conducted with the Keck 10m and Gemini North 8m telescopes. The aim of these observations was to acquire spectra with a sufficient signal-to-noise ratio to study the stellar populations of red sequence galaxies with different morphologies. These observations have also significantly increased the number of redshifts available in the sample. The present thesis is divided into 7 chapters and structured has follows: Chapter 2 discusses the observations and reduction of the data used in the thesis. Chapter 3 presents the analysis strategy devised for the study of galaxies in the HCS and tested on the cluster XMMU J1229+0151. The content of this chapter has also been published as a separate article on Monthly Notices of the Royal Astronomical Society (MNRAS) (Cerulo et al., 2014). Chapter 4 describes the study of the build-up of the red sequence in the HCS clusters, while Chapter 5 discusses the morphological evolution of red sequence galaxies. Chapter 6 reports on the current status of the analysis of the stellar populations in HCS red sequence galaxies and shows the results obtained so far from the data taken at the Keck and Gemini telescopes. Chapter 7 summarises the main results of the thesis and outlines the future directions and perspectives of cluster research with HCS and other samples. Strictly technical aspects related to morphological classification are discussed in the Appendix.

2 −1 Lookback time is give here in the ﬂat ΛCDM cosmology adopted in this thesis: H0 = 71.0 km · s −1 · Mpc ,ΩM = 0.27, ΩΛ = 0.73. 2 Data And Observations

The present chapters is intended to give a description of the sample of galaxy clusters of the Hawk-I Cluster Survey (HCS), together with the observations and data reduction involved. The low-redshift cluster control sample of the WINGS survey, and the optical data of the GOODS survey, used as control ﬁeld for statistical background subtraction, are presented in Chapter 3. The chapter is organised as follows: 2.1 introduces the HCS § sample, 2.2 gives a detailed overview of the imaging data used in the thesis, while 2.3 § § discusses the spectroscopic sample and the observations conducted as part of the project.

Throughout this chapter we adopt a ΛCDM cosmology with ΩΛ = 0.73, Ωm = 0.27 and H = 71.0 km s−1 Mpc−1. Unless otherwise stated, all magnitudes are quoted in 0 · · the AB system (Oke, 1974).

2.1 The HAWK-I Cluster Survey

The Hawk-I Cluster Survey (HCS) is a near infrared (NIR) observing programme carried out with the High Acuity Wide-ﬁeld K-band Imager (HAWK-I, Pirard et al. 2004) on the European Southern Observatory (ESO) 8.2 m Very Large Telescope (VLT) with the aim of studying galaxy populations in clusters at redshift 0.8 < z < 1.6. The HCS sample consists of 9 clusters, seven taken from the Hubble Space Telescope (HST) Cluster Supernova Survey (Dawson et al., 2009), the cluster RXJ0152.7-135 (RX0152), which is part of the Advanced Camera for Surveys intermediate cluster survey (Ford et al., 2004; Postman et al., 2005; Mei et al., 2009), and one cluster from the Spitzer Adaptation of the Red Sequence Cluster Survey (SpARCS J003550-431224, z = 1.34, Muzzin et al. 2012; Lidman et al. 2012). All the clusters were observed at least in the HAWK-I Ks band, while only clusters at z > 1.1 were observed in both the J and Ks bands. The observations and data reduction

15 16 Chapter 2. Data and Observations of these images are discussed in Lidman et al. (2013), while the Ks-band observations are summarised in Section 3.2.2. The J band images were taken and processed similarly to the Ks band. For all the clusters the HAWK-I images encompass a 100 100 field of view with ∼ × a final image quality, parametrised by the full width at half maximum (FWHM) of the point spread function (PSF), FWHM = 0.300 0.400 in both the J and Ks bands. − The cluster RDCS J1252.9-2927 (RDCS1252), observed in the Js and Ks bands of the Infrared Spectrometer And Array Camera (ISAAC, Moorwood et al. 1998a), previously mounted on the ESO/VLT and now decommissioned, was also added to the HCS sample and studied together with the other clusters. This brings the final sample to a total of 10 clusters. The NIR observations and data reduction for RDCS1252 are discussed in Lidman et al. (2004), while we summarise the main properties in Table 2.2. The final mosaicked images, in both the Js and Ks bands, have 4.70 4.70 fields of views with image qualities × FWHM 0.400. ∼ More recently, the cluster ClG J0218.3-0510, at z = 1.62, (Tran et al., 2010; Bassett et al., 2013), which was observed with the HAWK-I camera in the J, Ks and Y bands (Rudnick et al., 2012) was also added to the sample, bringing the total number of clusters to 11. However, due to its peculiarities (see Section 1.2), this cluster has not been included in the analysis presented in the following chapters and will be considered in future studies. The sample of galaxy clusters analysed in the present thesis is composed of the nine HCS clusters for which HST Advanced Camera for Surveys (ACS) data are available in at least the F775W (i775) and F850LP (z850) bands. This subsample, constituted by the 7 clusters in common with the HST Cluster Supernova Survey and the two clusters RX0152 and RDCS1252, spans the redshift range 0.8 < z < 1.5 (2.5 Gyr) and is one of the deepest and richest currently available samples of galaxy clusters at z 1. Table 2.1 shows the ∼ global properties of these clusters, together with the number of spectroscopically confirmed members, while Table 2.2 summarises the optical and NIR observations. 2.1. The HAWK-I Cluster Survey 17 bers 30 18 21 29 20 42 25 26 109 Mem Confirmed Spectroscopically .62), which are not used in the ) = 1

M +0.7 −0.5 +2.3 −1.6 +1.7 −1.2 +1.8 −1.3 +1.1 −0.7 +2.5 −1.8 +1.2 −1.0 +1.7 −1.4 +3.0 −1.7 DM 14 M 4.4 5.8 5.3 4.8 2.4 7.4 6.8 7.3 4.3 (10 1.45 0.84 0.91 0.98 1.03 1.04 1.22 1.24 1.39 (J2000) Redshift -3:33:19 -4:36:18 .34) and ClG J0218.3-0510 (z +0:38:13 +1:51:34 -17:38:02 -13:57:00 -36:32:50 -29:27:00 -25:57:00 δ = 1 z (J2000) 12:52:00 22:35:00 α 22:15:58.5 01:52.8:00 23:19:53.9 12:29:28.8 02:20:55.7 23:45:27.3 02:23:03.7 quoted in the ﬁfth column from the left are taken from Jee et al. (2011). We do not consider in Name DM M Cluster J0152.7-135 (RX0152) J1229+0151 (XMM1229) J0223-0436 (XMMU0223) J1252.9-2927 (RDCS1252) J2235.3-2557 (XMMU2235) J2215-1738 (XMMXCS2215) CS 2345-3633 (RCS2345) CS 0220.9-0333 (RCS0220) RX CS 2319.8+0038 (RCS2319) R R R XMM XMM RDCS XMM XMMU able 2.1 The Hawk-I Cluster Survey (HCS) sample used in the present thesis with the clusters listed in order of increasing redshift. analysis of the HCS red sequence in this thesis. T this table the other two clusters SpARCS J003550-431224 ( The dark matter halo masses 18 Chapter 2. Data and Observations

2.2 Observations and Data Reduction: Imaging

This section summarises the observations and data reduction of the optical and NIR imaging data of the HCS clusters taken with space- and ground-based telescopes.

2.2.1 Advanced Camera for Surveys

The ACS data for six of the HCS clusters are taken from the sample of the HST Cluster Supernova Survey (Dawson et al. 2009). The survey was aimed at measuring cosmological parameters with type Ia supernovae targeting 23 clusters at redshifts 0.9 < z < 1.5. The two clusters RDCS J1252.9-2927 (RDCS1252, Postman et al. 2005, z = 1.24) and XMMU J2235.3-2557 (XMMU2235, Jee et al. 2009, z = 1.39), observed before the HST Cluster Supernova programme, were also added to the sample and processed in the same way as the other 23 clusters, bringing the sample to a total of 25 clusters. The HCS targeted 8 of the 25 clusters all observable from the Southern Hemisphere.

The HST Cluster Supernova Survey consists of deep images taken in the F775W (i775) and F850LP (z850) bands collected over multiple HST visits on each cluster. In particular, each visit was composed of 3 or 4 exposures in the z850 band and at least 1 in the i775 band.

This resulted into an average exposure time of 3,000 s in the i775 band and 10,000 s in the z850 band for each cluster. The images were processed with the standard Space Telescope Science Institute (STScI) reduction pipeline with the most up to date calibration files and were combined using Multidrizzle (Fruchter & Hook, 2002; Koekemoer et al., 2003) to the final pixel scale 0.0500/pixel. The field covered by these images is on average 50 50 with × a resulting image quality FWHM 0.0900 in both bands. The i and z images of the ∼ 775 850 clusters XMMU2235 and RDCS1252, which cover similar fields of view were re-processed adopting the same procedures.

The i775 and z850 images for the clusters RX0152 and RDCS1252 were taken over multiple orbits of HST. Four pointings, arranged in a 2 by 2 pattern, were used in order to increase the depth of the final image within 10 from the cluster centre. This resulted in deeper exposures with respect to the average of the HST Cluster Supernova sample (see Table 2.2). In addition to the F775W and F850LP filters, RX0152 was also observed in the F625W (r625) filter following the same strategy of the i775 and z850 observations. The RX0152 images were reduced and combined with the APSIS pipeline (Blakeslee et al., 2003) to a final pixel scale of 0.03500/pixel. The total field covered by the images is 60 60 × in all three bands with PSF FWHM 0.0700. ∼ 2.2. Observations and Data Reduction: Imaging 19

2.2.2 Wide Field Camera 3

The clusters XMMU J1229+0151 (XMM1229, see Chapter 3), RDCS1252, and XMMU2235 were also observed in the F105W (Y), F110W, F125W (J), and F160W (H) bands of the IR channel of the HST Wide Field Camera 3 (WFC3) during Program 12051 (P. I. S. Perlmutter), aimed at the calibration of the sensitivities of the HST Near Infrared Cam- era and Multi-Object Spectrometer (NICMOS) and WFC3 for faint objects (Rubin et al., 2015). The reduction of the XMM1229 data is described in Chapter 3 ( 3.2.1), and a § similar procedure was followed for RDCS1252 and XMMU2235, although for the latter two clusters we used the latest version of the DrizzlePac1 released by STScI in June 2012, after the reduction of the XMM1229 data. This package contains an updated and revised implementation of the Multidrizzle algorithm. The combined WFC3 images of the HCS clusters, drizzled to the ﬁnal pixel scale 0.0600/pixel, cover an area of 30 30, The observed image quality IQ, deﬁned as the convo- × lution between the PSF of the instrument and the pixel response function PRF of the IR detector, is expressed by the equation:

IQ = PSF PRF. (2.1) ⊗

00 The observed IQ in all the WFC3 images is (FWHM)obs 0.2 . If we approximate the ∼ PRF with a 3-dimensional rectangular function with width equal to the pixel scale of the IR detector (0.12800), and the instrumental PSF as a Gaussian, the FWHM of the intrinsic image quality can be approximated by the equation

q 2 2 (FWHM)int = (FWHM) 0.128 (2.2) obs − where (FWHM)int is the FWHM of the instrumental PSF. Table 2.2 shows the values of the intrinsic image qualities for all the WFC3 observations of XMM1229, RDCS1252, and 00 00 XMMU2235, which are all in the range 0.12 < (FWHM)int < 0.18 .

2.2.3 Infrared Spectrometer And Array Camera (ISAAC)

In this section we summarise the Infrared Spectrometer And Array Camera (ISAAC) observations of the three clusters RCS 2319.8+0038 (RCS2319, 0.91), RCS 0220.9-0333 (RCS0220, z = 1.03), and RCS 2345-3633 (RCS2345, z = 1.04).

1The latest version of the DrizzlePac can be downloaded from: http://www.stsci.edu/hst/HST_ overview/drizzlepac 20 Chapter 2. Data and Observations

The ISAAC observations of RCS2319, RCS0220, and RCS2345 (Mu˜nozet al. in preparation) are part of a NIR observing programme aimed at the study of the build-up of the red sequence in a sub-sample of clusters of the Red Sequence Cluster Survey (RCS, Glad- ders & Yee 2005) . The programme was distributed among four observing runs that took place between 2001 and 2003 and targeted 15 clusters at z 1. The three clusters ∼ contained in the HCS sample were observed during the ESO programmes 70.A-0378 and 71.A-0345.

ISAAC (Moorwood et al., 1998a) was an IR imager and spectrograph optimised for observations in the range 1 µm < λ < 5 µm, previously mounted on the ESO/VLT and decommissioned in 2013. In imaging mode the instrument had two arms, one optimised for observations in the range 1-3 µm (short wavelength or SW arm), and the other covering the entire 1-5 µm range (long wavelength or LW arm). The three clusters were all observed with the SW arm, which was equipped with a 1024 1024 HgCdTe Hawaii detector with × a pixel scale 0.148400/pixel. RCS0220 was observed with the J filter, while RCS2319 and RCS2345 were both observed with the Js filter. The wavelength range covered by the J filter is broader than that covered by Js. However, we found that the observed AB (J Js) − colour at 0.90 < z < 1.05 predicted for a model spectral energy distribution (SED), taken from the Bruzual & Charlot (2003) library, with formation redshift zf = 4.75, Salpeter (1955) initial mass function (IMF), exponentially declining star-formation history with τ = 1 Gyr, and solar metallicity, is (J Js) = 0.001. Given this small difference between − the apparent magnitudes in the two filters, in the following chapters we will refer to the ISAAC J-band observations of the clusters regardless of the fact that the filter involved is J or Js.

RCS0220 was observed for a total of 45 minutes, while both RCS2319 and RCS2345 were observed for 54 minutes. The observations consisted of a series of dithered exposures of 1.5 minutes each with the telescope randomly offset within a square region 3800 wide. Each exposure was dark and sky subtracted and it was finally flat-field corrected. The photometric calibration was performed on stars of the NICMOS (Persson et al., 1998) and UKIRT2 (Hawarden et al., 2001) photometric standard star catalogues, which were observed at low airmasses during the same observing night in which the science data were taken. Once the magnitude zero-point for each cluster was estimated, all the images were re-calibrated to a common magnitude zero-point JV ega = 28.0 mag. The ISAAC observations of the HCS clusters are summarised in Table 2.2.

2The United Kingdom Infrared Telescope 2.3. Observations and Data Reduction: Spectroscopy 21

2.3 Observations and Data Reduction: Spectroscopy

The HCS clusters have been targeted in various spectroscopic follow-up programmes of the HST Cluster Supernova Survey conducted at the Keck and VLT telescopes. In this work we use all the redshifts obtained from those observations. The clusters XMM1229, RCS2319, and RCS0220 were the target of deep spectroscopic observations that we conducted at the Keck and Gemini North telescopes with the aim of acquiring high signal-to-noise (S/N) spectra to study stellar populations and increase the cluster spectroscopic sampling towards faint magnitudes. The cluster RCS2319, which belongs to the RCS2319+00 supercluster, was also part of an extensive survey of the supercluster conducted at the Magellan, VLT, Subaru, and Gemini-North telescopes. This data-set and the result of the spectroscopic survey of the RCS2319 supercluster are extensively discussed in Faloon et al. (2013), and we refer to that paper for a description of the observations and data reduction. We included the redshift catalogue from this large spectroscopic data-set in the HCS sample. The cluster XMMU J0233-0436 (XMMU0223, z = 1.22) falls in the ﬁeld of view of the VIMOS Public Extragalactic Redshift Survey (Garilli et al., 2014; Guzzo et al., 2014), and the redshifts coming from this sample were added to the HCS sample. The cluster RX0152 was targeted in multiple observing runs conducted at various telescopes between 2005 and 2009 (see Demarco et al. 2010 for a summary of these observations), while RDCS1252 was observed in various runs in 2003, 2011, and 2012 with FORS23 at the VLT (Demarco et al., 2007; Nantais et al., 2013b). In the present work we include all the redshifts available for RX0152 and RDCS1252 and used in the works of Demarco et al. (2010) and Nantais et al. (2013b).

3FOcal Reducer and low dispersion Spectrograph 22 Chapter 2. Data and Observations §3.3.2 26.2 25.0 26.7 26.4 25.3 24.2 26.2 22.9 24.5 26.1, 24.8, , 23.44 26.2, 25.7 22.4 24.6 22.4 24.2 21.9 24.2 23.4 22.3 24.9 25.3 26.2, 25.8, 25.0 23.2, 23.3, 23.5 26.6, 25.9, 25.5 26.2, 26.6, 25.5, 24.6, 26.1, 22.6, 24.1, 26.8, 25.0, 27.3, 24.4, 24.9 26.7, 23.0, 26.1, 25.04, 0.10 0.10 0.10 0.34 0.38 0.36 0.36 0.09 0.10, 0.10, 0.097, 0.097, 0.07, 0.08 0.94 0.34 0.45 0.31 0.57 0.31 0.34 0.58 0.39 0.63 0.15, 0.14, 0.14 0.12, 0.11, 0.14 0.11, 0.13, 0.14 0.43, 0.48, 0.47, 0.08, 0.10, 0.36, 0.10, 0.10, 0.094, 0.094, 0.093, 0.091, (FWHM) (”) completeness limit Image Quality 90% magnitude 0.07, 0.13, 0.11, 0.11, (1.2) (1.2) (1.1) c c c (19.0) b (82.0) (1.2), F160W (1.2), F160W (1.2), F160W 850 c c c z (57.1) (6.8) (9.7) (9.6) (14.4) (16.9) (10.9) (14.4) (9.6) (14.02) (10.7) b b b d b b b b d d d d a b d d ISAAC d (3.2) (3.2) (2.7) 850 850 850 s) Ks 850 850 850 850 (2.3) §2.2.2 for a discussion). Each line refers to an instrument; e.g. the third line of the z z z Ks 850 z z z z Ks 3 Ks (19.2), z (9.6) (9.6) (9.6) (11.3) b ofI (10 S Ks(9.6) R(1.14) SAAC ISAAC ISAAC Ks Ks Ks 775 Ks I (1.1), F125W (1.1), F125W (1.1), F125W J (2.4), (4.5), i s s (8.2), (3.3), (4.2), (3.0), J (3.4), (29.9), (14.4), c c c J J b b (10.6), (11.04), (86.6), b b b b b b d d d J 775 775 J J 775 775 775 775 i i 775 775 Filter(Exposure Time) i i i i i i (19.0), SAAC F110W F110W F110W I b J 625 r (1.2), (1.2), (1.3), c c c F105W F105W F105W CS2215 CS0220 CS2345 CS2319 Cluster RX0152 ORS2 R R R XMM1229 ACS WFC3 HAWK-I F RDCS1252 b c XMMU2235 XMMU0223 d a able 2.2 Summary of the HCS imaging observations. The 90% magnitude completeness limit is estimated as described in XMMX T . The imagePSF quality by FWHM the quoted detector pixel for response WFC3 function corresponds (see to the intrinsic PSF FWHM obtained by de-convolving the observed XMM1229 entry refers to the WFC3 observations of this cluster. 2.3. Observations and Data Reduction: Spectroscopy 23

2.3.1 The Gemini North Observations of XMM J1229+0151 and RCS 0220.9-0333

Two HCS clusters, XMM1229 (z = 0.98) and RCS0220 (z = 1.03), were observed in 2012 with the Gemini North Multi-Object Spectrograph (GMOS-N), at the Gemini North Telescope on Mauna Kea, Hawaii. XMM1229 was observed during the queue programmes GN-2012A-Q45 and GN-2012A-Q80 (P.I.: W. J. Couch), between February and June 2012, while RCS0220 was observed during the queue programme GN-2012B-Q65 (P.I.: W. J. Couch) between August and October 2012. These two clusters were chosen because, at their redshifts, the age-sensitive absorption features Hδ, and Hγ fall in the range 8000.0 A˚< λ < 9500.0 A,˚ where the GMOS-N detector array is highly sensitive. Furthermore, several metal-sensitive absorption indices, such as F e4531 and C24668 are also observed in the same wavelength range. The observations of the clusters were all conducted in Nod & Shuffle mode (N&S, Glazebrook & Bland-Hawthorn 2001), allowing for higher spatial sampling with improved sky subtraction compared to conventional Multi-Object Spectroscopy (MOS). In fact, during a N&S observation the telescope is moved between two positions (nodding) and, according to the size of the nod, either two observations of the object at different positions in the slit, or an observation of the object and an observation of the empty sky can be acquired. The result is that the sky-subtraction is improved and the spatial sampling is increased because slitlets with 300 400 lengths can be used instead of the longer slits − normally required by conventional MOS observations. We observed the two clusters adopting the “band-shuffling” mode. With this configuration the GMOS-N detector, consisting in an array of three 2048 4608 pixel CCDs, was × divided in three regions along the major axis. The central region was used to observe the targets while the other two were kept unexposed and used to store the charge (shuffling) between two subsequent telescope nods. A typical observation consisted of 15 cycles composed of a sequence of four 60 s pointings with the target kept within the slit at two different offset positions. The total exposure time of each observation was, therefore, 1800 s. We used the R400 grating (central wavelength λc = 7640A)˚ and the OG515 order-blocking filter, resulting in the spectral resolution R = 1918 and the wavelength coverage 5000A˚ < λ < 10000A.˚ In order to increase the signal-to-noise ratio of the spectra, a 2 2 detector binning was adopted. × We took observations at three different central wavelengths (λc = 7500, 7640, 7800 A)˚ to correct for the 0.5 mm gaps between the CCDs, and we also applied the DTX dither in the spatial direction to reduce the noise resulting from charge traps produced by pixels 24 Chapter 2. Data and Observations which are inefficient in transferring charge. The total exposure time for each mask was thus 3 hours. We oriented the masks at different position angles to increase the spatial sampling of the clusters. Targets of particular interest, such as the brightest cluster galaxy (BCG) and other bright red sequence galaxies, were observed in more than one mask. Since the main scientific goal of the observing programme was the study of stellar populations in passive cluster members, priority was given to red sequence galaxies with z850 < 23.0 mag. Blue cloud galaxies were also targeted in the masks to verify their membership to the clusters, and other slits were placed on random galaxies selected in the HAWK-I Ks images of the two clusters. With this strategy, we were able to arrange 15-20 slits in each mask. We used 100 wide and 400 long slitlets for the observations of XMM1229, and in each of the two masks designed for this program we fitted 15 slits. Since the data were taken in poor weather conditions (see Table 2.3), we decided to give the highest priority to spectroscopically confirmed cluster red sequence members and take high-S/N spectra suitable for the study of stellar populations. For each of the two programmes, GN-2012A-Q45 and GN-2012A-Q80 in which the initial proposal was split, we prepared only one mask and observed the targets twice. Since most of the galaxies were observed in both programmes this resulted in a total of 16 objects observed in the XMM1229 field 14 of which were targeted in four 3 hours exposures. The observations of RCS0220 were taken in better weather conditions (see Table 2.3) and this allowed us to design four masks with 18-20 300 100 slitlets. A total of 75 galaxies × were observed in the RCS0220 field, 10 of which were targeted in more than one mask. Long-slit spectra of the spectrophotometric standard stars Feige 34 (Massey et al. 1990, GN-2012A-Q45) and BD+28 4211 (Massey et al. 1990, GN-2012A-Q80 and GN-2012B- Q65) were used for flux calibration of the science spectra while N&S darks, taken with the shutter closed and with the same shuffling pattern of the science data, were used to correct for dark current and charge traps formed during the shuffling phase in some pixels. The GMOS spectra of XMM1229 and RCS0220 were reduced with the Gemini Im- age Reduction and Analysis Facility (IRAF4) Data Reduction Package Version 1.115. The package consists of a series of wrappers to standard IRAF tasks (e.g. imcombine, ccdred, dispcor, identify, standard, sensfunc) which allow users to easily handle the complex multi-extension FITS architecture of the raw GMOS MOS spectra. It also contains

4IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Asso- ciation of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation. 5The latest version of the Gemini IRAF data reduction Package can be found at: http://www.gemini. edu/node/10795 2.3. Observations and Data Reduction: Spectroscopy 25 all the additional steps required for the reduction of N&S data. The raw exposures were divided in three groups according to the central wavelength of the observations and were separately reduced. The raw spectra were dark and overscan- subtracted and then divided by an average master flat-field image obtained by combining the different flat-fields taken during the observations. Cosmic ray rejection was performed using the LA-cosmic algorithm described by van Dokkum (2001). The sky subtraction was performed on the flat-fielded and cosmic ray cleaned images before wavelength calibration. For the latter, a series of spectra of a CuAr lamp were used. The one-dimensional spectra, extracted for each sky-subtracted and wavelength-calibrated 2D spectrum were combined with the IRAF task odcombine, which produced a final average spectrum corrected for the inter-detector gaps. The Gemini IRAF task gsextract, which we used for the extraction of the 1-D spectra, outputs also the 1-D variance spectra. We combined these spectra with odcombine to correct for the inter-detector gaps, obtaining a final spectrum corresponding to their sum. The square root of this spectrum corresponds to the RMS of the combined final spectrum. The combined science and RMS 1-D spectra were flux-calibrated using the IRAF task calibrate with the sensitivity functions derived from the standard stars observed in each programme.

Table 2.3 Summary of the GMOS-N and LRIS observations of XMM1229, RCS0220 and RCS2319.

Cluster Instrument Observing Program Seeing Number of Number of (FWHM) (00) Masks Targets XMM1229 GMOS-N GN-2012A-Q45 0.75-1.05 1 15 GMOS-N GN-2012A-Q80 0.75-1.05 1 15 RCS0220 GMOS-N GN-2012B-Q65 0.40-0.75 4 75 RCS2319 LRIS 2011B W116LA 0.50-0.60 2 70

2.3.2 Keck/LRIS Observations of the cluster RCS 2319.8+0038

The cluster RCS2319 was observed in September 2011 with the Low Resolution Imager Spectrometer (LRIS, Oke et al. 1995) at the Keck I telescope (Mauna Kea, Hawaii) as part of the observing programme 2011B W116LA (P. I. W. J. Couch). A total of 70 objects were observed in one night in excellent seeing conditions (FWHM 0.500 0.600). LRIS ∼ − is provided with a blue and a red arms, allowing for a broad wavelength coverage from the near ultraviolet (NUV) to the NIR. We used the 600/10000 grating for the red arm and the 300/500 grism for the blue arm, which allowed us to cover the range 3500A˚ < λ < 11000A˚ 26 Chapter 2. Data and Observations with a 300A˚ gap between the blue and red spectra due to the presence of the dichroic filter used as beam splitter. Since most of the age- and metal-sensitive absorption features fall in the wavelength range of the red detector, we only reduced the red spectra, leaving the blue spectra for future projects. We designed two masks with slits of variable lengths, assigned by the autoslit software according to the brightness of the targets, and 100 width. Red sequence galaxies with magnitudes z850 < 24.0 were given the highest priority in the mask design, while galaxies in the blue cloud and random targets selected in the HAWK-I Ks field were also placed in the masks. The LRIS red arm camera consists of a detector array composed of two LBNL 2k 4k CCDs with 15 µm pixels and 0.13500/pixel resolution. The two CCDs are arranged × in a row to form a 4k 4k system with the dispersion direction running parallel to the gap × between the CCDs. The core of RCS2319 fitted into one of the detectors, allowing us to use the second detector to sample the cluster outskirts. The two masks were placed at two different orientations with one CCD targeting the cluster core and the other CCD centred on adjacent fields. During the observations we adopted the strategy of offsetting the targets within the slits to improve sky subtraction. This resulted, for each mask, in a sequence of four 2250s exposures for each offset position and a total exposure time for each mask ∼ of 2.5 hours. We reduced one mask with the Carnegie Python Distribution (CarPy) 6, which implements the method for sky subtraction discussed in Kelson (2003). This method consists in fitting the spectrum of the sky before correcting for the spatial distortion of the slits and is shown to produce spectra of high quality for distant targets such as the members of RCS2319. The raw spectra were first corrected for overscan and flat-field, and then the sky was fitted and subtracted. The wavelength calibration was performed using the spectra produced by a HgNeArCdZn lamp and using the IRAF tasks identify and dispcor. Once a wavelength solution was found, it was applied to the sky-subtracted science spectra, and the four exposures were combined using the IRAF task imcombine, which also ran the cosmic ray rejection. The second mask was reduced with a different software pipeline, written by C.L., which implemented all the steps from the overscan subtraction to the flat-fielding and performed the sky subtraction with a method based on the same principles of Kelson (2003). This change was motivated by the fact that the supercomputer where CarPy was installed at

6CarPy is available at: http://code.obs.carnegiescience.edu/Code/ carnegie-python-distribution/building-carpy-from-source 2.3. Observations and Data Reduction: Spectroscopy 27 the Swinburne University was decommissioned and the installation of this package was not possible on a different computer. The quality of the final spectra, however, does not change. Long-slit observations of the spectrophotometric standard star Wolf 1346 (Massey et al., 1990) were used for the flux calibration of the spectra. To this purpose, the spectrum of the standard star was reduced in the same way as the science data using CarPy to correct for overscan, flat-field and to subtract the sky. The identify and dispcor IRAF tasks were used for wavelength calibration, while the sensitivity function was derived using the IRAF tasks standard and sensfunc. The latter was applied to the one-dimensional science spectra previously extracted with the IRAF task apall to obtain the final flux- calibrated spectra. The Keck/LRIS observations of the cluster RCS2319 are summarised in Table 2.3.

2.3.3 Redshift Measurements

We used the code RUNZ7 to measure the redshifts of the spectra observed in the Gemini and Keck programs. RUNZ allows the user to interactively measure redshifts by displaying the spectrum with a set of relevant absorption and emission features overlaid on it. The software determines two automatic estimates of the redshift, one based on the cross- correlation with a set of template spectra of various types (galaxies, stars, QSO), the other resulting from the fitting of emission lines with Gaussian functions. The two estimates are independent because emission lines are clipped during the cross-correlation process. The spectrum, with a set of absorption and emission lines identified during the automatic estimation, is shown in a graphic window, and the user can verify the measurement. If the automatic estimate of the redshift is consistent with the features detectable by eye in the spectrum, it will be kept as the final redshift estimate However, if the automatic redshift is not consistent with the lines visible in the spectra, the user can manually select one or more lines and mark them accordingly. The software will, at this point, re-compute the redshift, based on the names of the features inserted by the user. This will be the final estimate of the redshift for the object in consideration. We used a scale consisting of the following four flags to assess the quality of the redshifts:

Q = 4: secure redshift; •

7 RUNZ can be downloaded from: http://www.physics.usyd.edu.au/~scroom/runz/ 28 Chapter 2. Data and Observations

Q = 3: probable redshift; • Q = 2: possible (not secure for science); • Q = 1: unknown. •

With this ranking, only redshifts with Q 3 were retained for science. ≥ The redshifts for the RCS0220 and RCS2319 fields were independently estimated by P.C. and C.L. and were compared. The spectra with discrepant measurements for which at least one of the redshifters had assigned Q 3 were run through RUNZ a second ≥ time and the redshift was determined again. The cases with discrepant redshifts after the second RUNZ run were discussed between the two redshifters until a common agreement was reached. The redshift catalogues of RCS2319 and RCS0220 with the Q 3 objects ≥ contain 33 and 35 objects, respectively. We found 17 new cluster members in the field of RCS0220 (14 within 0.54 R from the cluster centre) and 11 new cluster members in × 200 the field of RCS2319 (6 within 0.54 R from the cluster centre). The confirmed cluster × 200 members from the Gemini and Keck observations are shown in Tables 2.4, 2.5, and 2.6. Since the GMOS-N observations in the field of XMM1229 targeted already spectroscopically confirmed members, the redshifts for these spectra were just estimated by P.C. and verified against the redshifts previously measured by the HST Cluster Supernova Team.

Table 2.4 Spectroscopically conﬁrmed members in the cluster XMM1229 from the GMOS- N observations. All spectroscopic cluster members are included in the table regardless of their colour.

ID α δ z Qz (J2000) J2000 XMM1229 145 12:29:29.9 +01:50:46.3 0.985 4 XMM1229 237 12:29:29.3 +01:51:21.8 0.974 3 XMM1229 265 12:29:28.9 +01:51:24.9 0.974 4 XMM1229 312 12:29:28.7 +01:51:37.0 0.975 4 XMM1229 349 12:29:33.6 +01:51:46.4 0.973 3 2.3. Observations and Data Reduction: Spectroscopy 29

Table 2.5 Spectroscopically conﬁrmed members in the cluster RCS0220 from the GMOS- N observations. All spectroscopic cluster members are included in the table regardless of their colour.

ID α δ z Qz (J2000) J2000 RCS0220 353 02:20:56.6 -03:32:57.5 1.028 3 RCS0220 365 02:20:54.9 -03:32:52.4 1.028 3 RCS0220 191 02:20:55.6 -03:33:47.9 1.021 3 RCS0220 230 02:20:55.5 -03:33:31.0 1.026 3 RCS0220 282 02:20:55.7 -03:33:19.3 1.028 3 RCS0220 182 02:20:56.2 -03:33:56.0 1.017 3 RCS0220 181 02:20:53.9 -03:33:54.5 1.019 3 RCS0220 306 02:20:51.9 -03:33:09.7 1.021 3 RCS0220 218 02:20:56.4 -03:33:32.4 1.028 3 RCS0220 377 02:20:53.5 -03:32:49.2 1.034 3 RCS0220 307 02:20:55.6 -03:33:10.7 1.029 3 RCS0220 319 02:20:55.1 -03:33:05.3 1.021 3 RCS0220 568 02:20:59.2 -03:31:45.4 1.016 3 RCS0220 585 02:21:01.1 -03:31:48.6 1.026 3 RCS0220 244 02:20:57.4 -03:33:30.5 1.017 4 RCS0220 311 02:20:55.9 -03:33:08.2 1.028 4 RCS0220 456 02:20:52.7 -03:32:21.4 1.033 4

Table 2.6 Spectroscopically conﬁrmed members in the cluster RCS2319 from the Keck/LRIS observations. All spectroscopic cluster members are included in the table regardless of their colours.

ID α δ z Qz (J2000) J2000 RCS2319 240 23:19:47.1 +00:38:17.7 0.893 3 HAWK I Ks 881 23:20:03.2 +00:37:59.5 0.899 4 RCS2319 1090 23:19:53.5 +00:38:16.6 0.905 4 RCS2319 106 23:19:56.0 +00:37:35.3 0.904 4 RCS2319 132 23:19:57.5 +00:37:46.8 0.895 3 RCS2319 174 23:19:54.4 +00:38:04.3 0.905 4 RCS2319 177 23:19:55.1 +00:38:03.1 0.899 4 RCS2319 201 23:19:55.5 +00:38:08.3 0.892 4 RCS2319 205 23:19:53.4 +00:38:13.4 0.901 4 RCS2319 219 23:19:53.3 +00:38:14.0 0.897 4 RCS2319 210 23:19:52.5 +00:38:10.7 0.901 4

3 Analysis Method: The Morphological Transformation of Red Sequence Galaxies in the Distant Cluster XMMU J1229+0151

This chapter contains the paper published in Monthly Notices of the Royal Astronomical Society, The morphological transformation of red sequence galaxies in the distant cluster XMMU J1229+0151, Cerulo, P., Couch, W. J., Lidman, C., Delaye, L., Demarco, R., Huertas-Company, M., Mei, S., S’anchez-Janssen, R., 2014, MNRAS, 439, 2790, noting that apart from the numbering of tables and ﬁgures, it has been reproduced in its exact form.

Abstract We present the results of a detailed analysis of galaxy properties along the red sequence in XMMU J1229+0151, an X-ray selected cluster at z = 0.98 drawn from the HAWK-I Clus- ter Survey (HCS). Taking advantage of the broad photometric coverage and the availability of 77 spectra in the cluster field, we fit synthetic spectral energy distributions, and estimate stellar masses and photometric redshifts, which we use to determine the cluster membership. We investigate morphological and structural properties of red sequence galaxies and find that elliptical galaxies populate the bright end, while S0 galaxies represent the predominant population at intermediate luminosities, with their fraction decreasing at fainter magnitudes. A comparison with the low-redshift sample of the WINGS cluster survey reveals that at z 1 the bright end of the red sequence of XMMU J1229+0151 ∼ is richer in S0 galaxies. The faint end of the red sequence in XMMUJ1229+0151 appears rich in disc-dominated galaxies, which are rarer in the low redshift comparison sample at the same luminosities. Despite these differences between the morphological composition of the red sequence in XMMUJ1229+0151 and in low redshift samples, we find that to

31 32 Chapter 3. Analysis Method within the uncertainties, no such diﬀerence exists in the ratio of luminous to faint galaxies along the red sequence.

3.1 Introduction

Clusters of galaxies are the most massive virialised large scale structures in the universe and, thanks to the broad range of densities available in these systems, they can be used as laboratories for the study of the environmental drivers of galaxy evolution (De Lucia et al., 2007b). According to the hierarchical scenario of structure formation, galaxy clusters form after the collapse of the highest density peaks in the primordial matter distribution, accreting other smaller dark matter haloes at later times. By the present day, they have 15 built up into systems whose total mass can reach up to 10 M , with characteristic sizes of a couple of Mpc. Up to z 1.5, red and passive elliptical and S0 galaxies are the predominant population ∼ in cluster cores, whereas blue and star-forming spiral and irregular galaxies dominate the outskirts (see Dressler 1980; Dressler et al. 1997; Postman et al. 2005; Hilton et al. 2009; Muzzin et al. 2012; Mei et al. 2012). However, the fraction of blue star forming galaxies in clusters increases with redshift (Butcher & Oemler, 1978) and above z 1.5 there is ∼ evidence of star formation in cluster cores (Hilton et al., 2010; Tran et al., 2010; Hatch et al., 2011; Hayashi et al., 2012) These results suggest that most of the processes that led to the establishment of the star-formation- and morphology-local density relations, as they are observed in local clusters, were active in the interval 1.0 < z < 1.5. A large number of mechanisms have been proposed to both trigger star formation and then quench it. Galaxy-galaxy merging within groups that are falling into the cluster for the first time, galaxy-galaxy harassment within the clusters as galaxies pass each other at high velocities, and collisional compression and ram pressure stripping of gas in galaxies by the hot intracluster medium are mechanisms that are all thought to be at play in the dense cluster environment (Gunn & Gott, 1972; Lavery & Henry, 1988; Moore et al., 1996; Bekki, 1999; Bekki & Couch, 2003). However, from the current observations, it is not clear which of these processes is driving galaxy evolution in clusters. Furthermore, clusters are highly heterogeneous systems, and core and outskirts constitute different environments, with different global physical properties. The interactions of the galaxies with their surroundings may therefore change substantially according to their location within the cluster. Regardless of the environment in which they reside, galaxies have a bimodal colour distribution, such that the colour-magnitude diagram of a galaxy population is characterised 3.1. Introduction 33 by a narrow sequence of red quiescent objects and a diffuse cloud of blue star-forming galaxies. As galaxies finish to form stars, depleting their gas reservoirs, they migrate from the blue cloud to the red sequence. Thus, in principle, the evolution of a galaxy population can be investigated by looking at the gradual build-up of the red sequence with redshift. However, bursts of star formation resulting from merger events between a quiescent and a star-forming galaxy may push the galaxy back to the blue cloud (see e. g.: Faber et al. 2007).

Kodama & Arimoto (1997) explained the red sequence as a mass-metallicity relation, the scatter about the best-fit straight line being driven by age differences among galaxies. However, Gallazzi et al. (2006) demonstrated that metallicity contributes to the scatter too and that the age-induced scatter is anti-correlated with stellar mass. As the local galaxy density increases, the red sequence becomes more pronounced with respect to the blue cloud (see e. g.: Hogg et al. 2004) and galaxy clusters are characterised by a tight and well defined red sequence, which can be used to estimate the cluster redshift when no spectroscopic information is available (Gladders & Yee, 2005; Andreon & Huertas- Company, 2011; Wilson, 2003).

Between z = 1 and z = 0, the number of red sequence galaxies fainter than MV = 20.0 is found to increase, approximately halving the relative ratio between luminous (i.e. − MV < 20.0) and faint galaxies (MV 20.0) (see e. g. Tanaka et al. 2005; De Lucia − ≥ − et al. 2007b; Gilbank & Balogh 2008; Capozzi et al. 2010; Bildfell et al. 2012; Lemaux et al. 2012). This deficit of galaxies at the faint end of the red sequence has been detected in clusters up to redshift z = 1.62 (Rudnick et al., 2012). However, Andreon (2008) studied the trend of the luminous to faint ratio in a sample of 28 galaxy clusters at 0.0 < z < 1.3, finding no deficit (see also Brown et al. 2008 for a similar conclusion in lower-density environments). The existence of a deficit of galaxies at the faint end of the red sequence supports the notion of a build-up at low masses. Low-mass galaxies quench their star formation at later times, with respect to higher-mass systems, and therefore they join the red sequence at later times. This property, which is known as downsizing (Cowie et al., 1996), is observed also in the field up to z 2 (see e.g.: Tanaka et al. 2005; Ilbert et al. ∼ 2010).

Muzzin et al. (2012) investigated the separate effects of mass and environment on the evolution of galaxies in a sample of clusters at 0.8 < z < 1.4 from the Gemini Cluster Astrophysics Spectroscopic Survey (GCLASS). They observed a decrease of the fraction of star-forming galaxies at increasing galaxy stellar masses and local densities. They found that, at fixed environment (i.e. cluster core and outskirts, and field), the star formation 34 Chapter 3. Analysis Method rate (SFR), specific star formation rate (SSFR) and the amplitude of the 4000 A˚ break

(Dn(4000)) were all correlated with stellar mass. However, at fixed stellar mass, the same quantities were not found to correlate with the environment. They concluded that stellar mass is the primary driver for the quenching of star formation, regardless of the environment, while the environment acts on the global galaxy population quenching the star formation rate over a relatively short period of time, regardless of galaxy stellar mass. In other words, the effect of the environment is to accelerate star-formation quenching leaving the correlations with mass of SFR, SSFR and Dn(4000) unchanged. The conclusions of Muzzin et al. (2012) extended to higher galaxy densities what had previously been observed in the field up to z = 2 by Peng et al. (2010b, 2012) and Quadri et al. (2012).

While Peng et al. (2010b) and Muzzin et al. (2012) suggest that environment plays a role in accelerating the shutting down of star-formation, neither explore the mechanisms by which environment does this. Demarco et al. (2010) investigated the properties of red sequence members in the cluster RX J0152.7-1357, at z = 0.84, ﬁnding that early-type galaxies at the faint end of the red sequence are systematically younger than galaxies at the bright end of the red sequence and are preferentially located in the cluster outskirts. They concluded that star formation in these galaxies had recently been quenched either through ram pressure stripping of gas by the hot intracluster medium, or rapid exhaustion of gas reservoirs from an earlier epoch of star formation caused by galaxy-galaxy merging. These results support once again the notion of a build-up of the red sequence at low masses, suggesting that the evolution of cluster galaxies ﬁts in the downsizing scenario.

We extend the work of Demarco et al. (2010) by making a comprehensive study of the morphological and structural properties of red sequence galaxies in the cluster XMMU J1229+0151 (hereafter XMM1229, Santos et al., 2009, Fig. 3.1), at z = 0.98. The cluster is part of the HAWK-I Cluster Survey (HCS, Lidman et al. 2013), comprising a sample of 9 galaxy clusters at 0.8 < z < 1.5. At the low redshift end in the HCS sample there is the cluster RX J0152.7-1357, whose galaxy population was studied in detail by Demarco et al. (2010). XMM1229 has imaging and spectroscopic data from several space and ground based facilities and its X-ray and dark matter properties were studied by Santos et al. (2009) and Jee et al. (2011), respectively. Furthermore, Santos et al. (2009) also analysed the properties of the spectroscopically conﬁrmed members of this cluster. The present work extends the analysis of Santos et al. (2009) to other cluster members selected with photometric redshifts, whose estimation has been possible thanks to the additional imaging data become available after the publication of that work and described in 3.2. § The position of the cluster in the redshift range of the HCS and the availability of 3.1. Introduction 35

Figure 3.1 Colour image of XMM1229 obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks image. White circles are photometrically selected red sequence members and yellow squares are spectroscopically conﬁrmed members (see 3.4.1 and Table 3.3 for the determination of the cluster membership and a complete list of§ red sequence galaxies, respectively). a considerable quantity of data make XMM1229 suitable to develop a method for the analysis of the red sequence in the HCS clusters. The method developed in this paper will be extended to the other clusters of the HCS sample in order to study the build- up of the red sequence in galaxy clusters at 0.8 < z < 1.5. In particular, our interest will be focused on three points: the characteristics of the red sequence itself (its shape, slope, and scatter), the ratio between luminous and faint galaxies, and the morphological properties of galaxies along the red sequence. The availability of the Wide Field Nearby Galaxy-clusters Survey (WINGS, Fasano et al. 2006) and the MORPHS survey (Smail et al., 1997) allow us to build comparison samples of clusters at z 0 and 0.3 < z < 0.6, ∼ respectively, which are used to compare the properties of the red sequence members in XMM1229 and, in a forthcoming paper, of the entire HCS sample.

The paper is organised as follows: we describe the observations and data analysis in 3.2. and 3.3. 3.4 presents the results of our study which are discussed in 3.5, while §§ § § we summarise our work and draw our conclusions in 3.6. Throughout the paper we use § −1 −1 a ΛCDM cosmology with H = 71 km s Mpc ,ΩM = 0.27, and Ω = 0.73. All 0 · · Λ magnitudes are quoted in the AB system (Oke, 1974), unless stated otherwise. 36 Chapter 3. Analysis Method

3.2 Observations and data reduction

XMM1229 was first discovered as an X-ray overdensity in the XMM-Newton Distant Cluster Project (XDCP) (Fassbender et al., 2011), a survey that used XMM-Newton telescope data to detect distant galaxy clusters. The cluster has been later targeted by several space- and ground-based telescopes (HST, VLT, NTT), providing us with a rich dataset covering the spectral region 0.65 < λ < 2.2µm (Fig. 3.2), as well as spectra for 26 cluster members. With the data having been acquired at different times and on different telescopes, with strategies varying with each program, the resulting data set is very diverse. This section describes this rich multi-wavelength data set detailing in particular the reduction methods we have employed to be able to utilise it as a complete ensemble.

Figure 3.2 Photometric coverage of the XMM1229 ﬁeld. From left to right: R SPECIAL (R), F775W (i), F850LP (z), F105W (Y), F125W (J), F160W (H), Ks. The F110W transmission curve is not plotted, as it covers the same spectral range of the F105W and F125W bands. The SofI J band transmission curve is not plotted because it overlaps with F125W. The F775W and F850LP bands used for the colour-magnitude diagram are highlighted by the arrows. The solid black line represents the template SED of an elliptical galaxy from Coleman et al. (1980) at the redshift of XMM1229. 3.2. Observations and data reduction 37

3.2.1 HST imaging

Advanced Camera for Surveys (ACS)

XMM1229 was imaged in November 2006 in the F775W (i775) and F850LP (z850) bands of the Wide Field Channel (WFC) of the Advanced Camera for Surveys (ACS), on board the Hubble Space Telescope (HST). The observations were part of the Hubble Space Telescope Cluster Supernova Survey (Dawson et al., 2009), aimed at the search of Type Ia supernovae (SNeIa) in cluster galaxies at 0.9 < z < 1.5. We summarise here the survey strategy and the data reduction process for the ACS observations of XMM1229, referring to Dawson et al. (2009) and Suzuki et al. (2012) for a more detailed description. The HST Cluster Supernova Survey collected i775 and z850 observations of 25 X-ray, optically or infrared (IR) detected galaxy clusters over the redshift range 0.9 < z < 1.5. The clusters were observed in multiple visits and, in each visit, at least one exposure in i775 and three or four exposures in z850 were collected. The images were calibrated using the standard calibration pipeline provided by the Space Telescope Science Institute (STScI) and were registered and stacked using Multidrizzle (Fruchter & Hook, 2002; Koekemoer et al., 2003), with a ﬁnal pixel scale of 0.05 00/pixel for all the clusters. XMM1229 was observed during 8 visits, with 9 and 30 exposures respectively collected for the i and z bands. The resulting image covers 50.1 50.1, 775 850 × with an image quality 1 of 0.0900 in both bands. The details of the ACS observations ∼ are reported in Table 3.1.

Wide Field Camera 3 (WFC3)

XMM1229 was imaged in the IR channel of Wide Field Camera 3 (WFC3), on board HST, as part of the program 12051 (P. I.: S. Perlmutter), aimed at the calibration of the sensitivities of NICMOS and WFC3 for faint objects. The observations were taken in 2010 May 24 in the F105W (Y), F110W, F125W (J) and F160W (H) ﬁlter bands, following a BOX-MIN dithering pattern. The images were reduced with calwf3, using the most recent versions of the calibration frames available for WFC3. We combined the reduced images ( flt ﬁles) with Multidrizzle, setting the drop size (pixfrac parameter) to 0.8 and the pixel size to 0.0600. The resulting images cover an area of 30 30, with image qualities2 that × vary between 0.1100 and 0.1400. The WFC3 observations of XMM1229 are summarised in

1Throughout this paper we use the FWHM of stars as a measure of the image quality. 2The image quality reported in Table 3.1 corresponds to the FWHM of the intrinsic PSF resulting from the deconvolution of the observed PSF and the pixel response function of the WFC3 IR detector (Koekemoer et al., 2011). 38 Chapter 3. Analysis Method

Table 3.1.

3.2.2 Ground Based Imaging

FORS2

XMM1229 was observed with the FOcal Reducer/low dispersion Spectrograph 2 (FORS2, Appenzeller et al., 1998), mounted on Yepun, the fourth unit of the 8 m ESO/VLT, on Cerro Paranal (Chile). The observations were taken during program 073.A-0737(A) (P.I. A. Schwope), an optical follow-up of the XDCP, carried out with the R SPECIAL and

Z GUNN filters. We only used the R SPECIAL (R) data, as the quality of the ACS z850 image is significantly higher. The FORS2 camera comprises a detector array of two 2k 4k × MIT CCDs, separated by a 7 pixel gap and, in order to correct for that, the XMM1229 field was observed in three dithered exposures of 380 s each, that were reduced using the FORS data reduction pipeline3. Each chip was separately bias subtracted and flat-field corrected, and a standard star field was used to estimate the zero point. The Software for Calibrating AstroMetry and Photometry (SCAMP, Bertin 2006) was then used to find an appropriate astrometric solution for the images, that were finally co-added using SWarp v2.19.1 (Bertin et al., 2002)4. The field of view of the final R band image of XMM1229 is 80 90 with an image quality of 0.600 and a resolution of 0.25200/pixel. The FORS2 × ∼ observations are summarised in Table 3.2.

SofI

XMM1229 was observed in the J and Ks filter bands of SofI (Moorwood et al., 1998b), mounted on the ESO New Technology Telescope (NTT), at the La Silla Observatory (Chile). The data were taken in March 2007 as part of a near infrared (NIR) follow-up of the XDCP. In this paper we only use the J band, as the Ks data from HAWK-I are considerably deeper (see 2.2.3 and Table 3.2). We summarise here the SofI observations § of the XMM1229 field, referring to Santos et al. (2009) for a more detailed description. XMM1229 was observed for a total of 40 minutes with the instrument operating in Large Field Mode, which has a 5 0 5 0 field of view, with a resolution of 0.29000/pixel. In × order to take into account the high variability of the NIR background, dithered exposures of the field were taken and reduced with the ESO/MVM software5. The resulting image quality is 0.900. The SofI J band observations are summarised in Table 3.2. ∼ 3http://www.eso.org/sci/software/pipelines/. 4The SCAMP and SWarp binary and source files can be found at http://www.astromatic.net. 5http://archive.eso.org/cms/eso-data/data-packages/eso-mvm-software-package 3.2. Observations and data reduction 39

HAWK-I

A subsample of the HST Cluster Supernova Survey, one cluster from SpARCS (Muzzin et al., 2012; Lidman et al., 2012) and the cluster RX J0152.7-1357, from the ACS In- termediate Redshift Cluster Survey (Ford et al., 2004), all observable from the southern hemisphere, were observed with the High Acuity Wide field K-band Imager (HAWK-I, Pirard et al. 2004), mounted at the Nasmith A focus of Yepun, the fourth unit of the 8 m ESO/VLT. These clusters are part of the HAWK-I Cluster Survey (HCS) (Lidman et al., 2013), a near-infrared program providing deep imaging data necessary for the study of the properties of old stellar populations in cluster early-type galaxies at high redshift. We refer to Lidman et al. (2013) for a detailed description of the HAWK-I observations and data reduction of the XMM1229 field. The HAWK-I camera consists of an array of 4 detectors, each defining the quadrant of a square surface, with a total area corresponding to a 7.05 7.05 field of view. The final × co-added mosaic of XMM1229 has a field of view of 100 100 with an image quality of × 0.3400 and a resolution of 0.100/pixel. The HAWK-I observations are summarised in Table 3.2.

3.2.3 Ground-based Spectroscopy

XMM1229 was observed with FORS2 in 2006 during the spectroscopic follow-up of SNe Ia in the HST Cluster Supernova Survey. We summarise here the observations and data reduction, referring the reader to Santos et al. (2009) and Suzuki et al. (2012) for more detailed descriptions. XMM1229 was observed five times with FORS2 using the 300I grism and the OG590 order sorting filter. This configuration produces a spectral resolution of 2.3 A/pixel˚ and a wavelength coverage extending from 5900 to 10,000 A.˚ Several SNe were detected in this cluster, and for this reason, and to allow the supernovae to be observed both near their maximum light and when their luminosity had significantly faded, XMM1229 was targeted multiple times. A total of 77 objects were observed, 26 of which were found to lie within 3σ from the average cluster redshift (z 0.98). The spectroscopically confirmed members ∼ of XMM1229 are represented as red dots in Fig. 3.4 and are listed in Table 3.3. Six galaxies were only detected in the HAWK-I image and we exclude them from the analysis of the red sequence, as no colour information is available. The galaxy XMM1229 316 was also not considered, as SExtractor (Bertin & Arnouts, 1996) failed to return a reliable estimate of the i775 and z850 aperture magnitudes (SExtractor FLAGS = 16: data within the aperture incomplete or corrupted, see also Table 3.3). 40 Chapter 3. Analysis Method C3 0 (H) 3 00 × 0.06 23.5 0 1112 0.14 3 F160W HST/WF C3 0 (J) 3 00 × 0.06 23.3 0 1212 0.13 3 F125W HST/WF C3 0 3 00 × 0.06 23.2 0 1112 0.11 3 F110W HST/WF 0 C3 0 (Y) 10 00 3 00 × × Ks 24.6 0.10 0 0.06 23.0 0.34 0 1312 11310 0.11 3 T/HAWK-I 10 F105W HST/WF VL ) 0 0 850 5 00 z CS 5.1 ( 00 J × 22.4 0 × 2280 0.94 0.290 5 0.05 25.0 0 0.09 10940 NTT/SofI 5.1 HST/A F850LP (R) ) §3.3.1 §3.3.1 0 0 9 00 775 CS 5.1 00 (i × 25.3 0 1140 0.63 0.252 × 8 T/FORS2 0.05 25.0 0 4160 0.08 §2.1.2 SPECIAL VL 5.1 HST/A R F775W a b §2.1.1 and /pixel) /pixel) 00 00 ( field field a e/instrument e/instrument pixel scale, see of the PSF modelled by PSFex, see scale ( osure time (s) osure time (s) scale imaged imaged quality (FWHM) quality (FWHM) completeness (mag) completeness (mag) exp exp pixel telescop FWHM FWHM of the PSF modelled by PSFex, see Drizzled telescop pixel b a a 90% 90% able 3.1 Summary of the HST observations of XMM1229. image image Table 3.2 Summary of the ground based observations of XMM1229. T 3.3. Data Analysis and Measurements 41

3.3 Data Analysis and Measurements

3.3.1 Object Detection and PSF modelling

We group the images according to the observing program and the instrument with which they were observed. This subdivision produces five groups: ACS, WFC3, FORS2, HAWK- I and SofI. We use a modified version of the GALAPAGOS IDL pipeline6 (Häussleret al., 2007) to run SExtractor in high dynamic range mode (HDR) on each set of observations and optimise the number of detections. In fact, when run with low detection thresholds, small detection areas and aggressive deblending, SExtractor can improperly split a bright object into a number of sub-components. On the other hand, when run with high detection thresholds and large detection areas, the software might fail in deblending two closely separated objects that are therefore classified as a single source. The HDR technique consists of two distinct runs of SExtractor: one to detect only the brightest sources (COLD run) and the other to detect the faintest sources and deblend objects with very close neighbours (HOT run). This is achieved by varying the parameters DETECT MINAREA, DETECT THRESH and DEBLEND MINCONT, and setting smaller detection areas, detection thresholds and deblending contrasts in the HOT run, when faint sources and close neighbouring objects are detected. This procedure produces two separate catalogues: one for the bright and the other for the faint sources, that are eventually merged into a single catalogue. GALAPA- GOS implements an algorithm in which the HOT sources that are within a certain distance from each COLD source are rejected. In fact, these sources are likely to be the result of improper deblending (e.g. substructures in bright nearby galaxies or star formation clumps in distant late-type galaxies) and would contaminate the final catalogue as spurious detections. We used the PSF Extractor (PSFex) software (version 3.9), written by the Terapix group (Bertin, 2011) to model the PSF in each image7. PSFex selects point sources in the half light radius vs flux plane using the SExtractor parameters FLUX RADIUS and FLUX APER as diagnostics of those two quantities. The PSF is modelled as a linear combination of basis vectors that can be chosen either using each pixel as a free parameter (pixel basis), as done in this work, or using actual point source images (Gauss-Laguerre and Karhunen- Loève bases). The FWHM of the PSFs for each photometric band are reported in the fifth row of Tables 3.1 and 3.2.

6 http://astro-staff.uibk.ac.at/~m.barden/galapagos/ 7PSFex can be downloaded at: http://www.astromatic.net/software/psfex 42 Chapter 3. Analysis Method

Figure 3.3 (left): Calibration of the zpeg photometric redshift estimate for the cluster centre. On the x-axis it is plotted the value zspec of the spectroscopic redshifts measured with FORS2 in the cluster centre, while on the y-axis it is plotted the discrepancy ∆z = (zspec zphot)/(1 + zspec). (right): Photometric redshift distribution in the central region of XMM1229.− The vertical dashed lines are the two limits used to deﬁne the cluster membership. The redshift distributions of the cluster members and of the cluster red sequence members (hatched histogram) are shown in the inset plot, where we also report the corresponding values of the galaxy peculiar velocities along the top horizontal axis.

3.3.2 Photometric Catalogue

In order to study the properties of the members of XMM1229 across the whole spatial extension of the cluster and given the different areas covered in each band pass, we decided to split our imaging database into two samples: the centre and the outskirts of the cluster. We produced a multiband catalogue for each sample. In this process we defined the central region as the one delimited by the WFC3 observed field, which has the smallest extent. This provided us with 8 photometric bands in the cluster centre (R, i775, z850, F105W, F110W, F125W, F160W, Ks) and 5 in the cluster outskirts (R, i775, z850, J, Ks). We used the SofI J band only for the analysis of the cluster outskirts because its spectral coverage overlaps with the one of the WFC3 F125W which, being deeper, was used as the J band of choice in the study of the cluster centre. The central region extends up to 600 kpc at the redshift of the cluster, which corresponds to 0.54 R (Jee et al., 2011), while the outskirts region approximately corresponds to the × 200 region between 0.6 Mpc and 1.04 Mpc, the upper limit being imposed by the width of the ACS field. The images of the cluster centre were degraded to the PSF of the R-band image, 3.3. Data Analysis and Measurements 43 while in the outskirts they were degraded to the PSF of the SofI J band, which had the broadest FWHM. We then ran GALAPAGOS on each of the PSF matched images, using the unconvolved images for detection. With the HST observations having been taken in more than one filter band, we used the z850 and F110W images for detection in the ACS and WFC3 groups, respectively. In order to remove any bias induced by intrinsic colour gradients, we measured aperture magnitudes within fixed circular apertures of 00 2 , corresponding to a physical radius Rap 8kpc at z = 0.98. With this choice the ∼ colour gradients for bright galaxies become negligible, while at low luminosities galaxies are almost entirely contained within the aperture radius. Magnitudes were corrected for galactic extinction using the dust maps of Schlegel et al. (1998) and applying the Cardelli et al. (1989) equations.

In order to consistently compare galaxy colours in the cluster centre and outskirts, following Meyers et al. (2012), we performed a cross-convolution of the (unconvolved) F775W and F850LP images, in which each image was convolved by the PSF of the other image. These are in fact the two photometric bands that were used for colour measurements in this work, as they almost bracket the 4000 A˚ break at the redshift of XMM1229 and are also the deepest images of the sample (see Fig. 3.2). The choice of this strategy for colour measurement allowed us to match the image qualities of the ACS images without degrad- ing them to the ground-based level. In fact, after cross-convolution, the resulting image quality is 0.1400, which is considerably narrower than in the ground-based images (see ∼ Table 3.2).

The resampling and co-addition steps of the data reduction, as well as the PSF matching process introduce correlations between pixels, which are not taken into account by SExtractor (see e.g. Casertano et al., 2000; Lidman et al., 2008; Trenti et al., 2011). Fol- lowing the method outlined in Labbéet al. (2003), we took random regions of sky on each image and measured the flux within different apertures. This provided us with a direct measure of the variation of the sky flux with the aperture radius and therefore allowed us to quantify the deviations from a purely poissonian noise. The advantage of this method, with respect to the application of analytical relations, as those outlined in Casertano et al. (2000), is that the noise is estimated directly on the images and no assumptions are made about the properties of the instrument or the co-addition and resampling algorithms used in data reduction.

In order to quantify the depth of the images, we inserted simulated galaxy images, generated as described in 3.4.4, in empty regions of each science image. We ran SExtractor § on each single image in single image mode with the same conﬁguration used for the original 44 Chapter 3. Analysis Method images and looked at the fractions of recovered simulated objects as a function of input magnitude. This fraction is a direct measurement of the incompleteness of the photometric catalogues extracted on each image. The 90% magnitude completeness limit, quoted in the sixth row of Tables 3.1 and 3.2, is used in this work to parametrise completeness in the XMM1229 sample.

3.4 Results

3.4.1 Cluster Membership: Photometric Redshifts

With the available FORS2 spectra, it is possible to study the red sequence only down to z850 = 23.0 (see Fig. 3.4). In order to estimate the distance of fainter galaxies and assess their membership to XMM1229, we could rely either on statistical background subtraction or photometric redshifts. These two methods are shown to produce comparable luminosity functions for red sequence cluster members (Rudnick et al., 2009). With nine available photometric bands in the XMM1229 field, spanning the range 0.65 < λ < 2.2 µm, corresponding to the rest-frame wavelength range 0.33 < λ < 1.1 µm, covering from the near-ultraviolet to the near-infrared regions of the spectrum, and given the availability of 77 spectra, we decided to use photometric redshifts to determine the membership of the cluster. We were able to determine photo-z’s for objects as faint as z850 = 24.0 on the red sequence, going one magnitude fainter than the limit imposed by the FORS2 spectroscopic observations (see also Santos et al. 2009) and remaining within the magnitude limit for a reliable morphological classification at z 1 (Postman et al., 2005). ∼ The program zpeg (Le Borgne & Rocca-Volmerange, 2002) was used to fit synthetic spectral energy distributions (SED), grouped in seven galaxy types, and built with the PEGASE´ spectral evolution code (Fioc & Rocca-Volmerange, 1997) assuming a Kroupa (2001) initial mass function (IMF). The template types cover a wide range of spectral classes, going from passive to actively star-forming galaxies. zpeg implements a χ2 minimisation procedure, in which the best fitting SED is the one which minimises the χ2 in a three-dimensional parameter space of age, redshift and template type. The metallicity of the synthetic SEDs is assumed to evolve with time according to the star formation history of each template and with the stars forming at the metallicity of the interstellar medium. No dust extinction is assumed. We used the available spectroscopic redshifts to calibrate the photo-z’s obtaining a median ∆z = (zspec zphot)/(1+zspec) = 0.0 0.05 in the cluster centre and ∆z = 0.02 0.11 − in the outskirts, where the uncertainty is computed as the normalised median absolute 3.4. Results 45 deviation (NMAD, see also Fig. 3.3, left panel). In both samples we fixed zphot = zspec for the spectroscopically confirmed cluster members and let zpeg perform the SED fitting with only age and template type as free parameters.

Cluster member candidates were defined as those galaxies in the range 0.8 < zphot < 1.2, the cut being chosen from the width of the photo-z distribution (see Fig. 3.3). Al- though the median fractional photo-z error for all galaxies in the range 0.8 < zphot < 1.2 is 12 %, we decided to focus only on the red sequence, as the photo-z estimate of blue galaxies is expected to be significantly uncertain with the available spectral coverage. In fact, the R band, which is the bluest for the XMM1229 samples, does not cover the blue side of the 4000 A˚ break at z < 0.6 and foreground galaxies may be misclassified as blue cluster members. A second quantity produced by zpeg is the stellar mass, which is defined as the mass locked into stars and is obtained with a median fractional uncertainty of ∼ 24% for red sequence cluster members in the central region8. The analysis of the stellar masses of red sequence galaxies in the cluster outskirts will be presented in a forthcoming paper.

Delaye et al. (2014) studied the stellar mass vs size relation in the HCS clusters. They estimated stellar masses using the lephare software (Arnouts et al., 1999; Ilbert et al., 2006) on a set of synthetic SEDs from the Bruzual & Charlot (2003) library with three different metallicities (0.2Z , 0.4Z , Z ), exponentially declining star formation histories and a Chabrier (2003) IMF. As a consistency check, we estimated the stellar masses of red sequence galaxies in XMM1229 using lephare on the same set of templates of Delaye et al. (2014) and adopting their cosmology9. As in Delaye et al. (2014), we also fixed the redshift of red sequence galaxies at z = 0.98 and we found that the so obtained stellar masses were a factor of 1.2 smaller. This small difference can be attributed to both the different sets of photometric bands used in the two works10 and the different apertures used for galaxy photometry, as Delaye et al. (2014) used MAG AUTO magnitudes instead of fixed aperture magnitudes. When comparing with our zpeg estimates, we found a median ratio of 1.35 between Delaye et al. (2014) and this work. Since this difference did not affect the conclusions of this paper and in order to be consistent with the cosmology chosen for this work, we kept our stellar mass estimates.

8The reader can refer to Bernardi et al. (2010) for conversions to stellar masses obtained with the most commonly used Chabrier and Salpeter IMFs. 9 −1 −1 Delaye et al. (2014) use a ΛCDM cosmology with H0 = 70 km · s · Mpc ,ΩM = 0.30, and ΩΛ = 0.70 10 Delaye et al. (2014) use i775, z850, J (from SofI) and Ks. 46 Chapter 3. Analysis Method

3.4.2 Contamination from Field Interlopers

In order to estimate the contamination of the XMM1229 red sequence, we used the HST/ACS F775W and F850LP images of the two fields of the Great Observatories Origins Deep Survey (GOODS, Giavalisco et al. 2004, version 2.0). These images are very similar to those of XMM1229, the only remarkable difference being their resolution (0.0300/pixel for GOODS vs 0.0500/pixel for XMM1229). GOODS is a deep astronomical survey centred on two fields: the Hubble Deep Field North (GOODS North) and the Chandra Deep Field South (GOODS South). The project was aimed at collecting deep multiband photometry from various space- and ground-based facilities (e.g. HST, Spitzer, Chandra and XMM-Newton, VLT, KPNO, Subaru), in order to accurately study the properties of distant galaxies. Spectroscopic follow-up observations were also carried out at Keck and VLT (Wirth et al., 2004; Vanzella et al., 2005, 2006, 2008; Popesso et al., 2009; Balestra et al., 2010), resulting in an extensive ensemble of imaging and spectroscopic data covering 300 square arcminutes. ∼ As for XMM1229, we followed the method outlined in 3.3 to process the GOODS § ACS images and model the PSF. The 90% magnitude completeness limits, estimated as described in 3.3.2, are i ,lim = 27.3 mag and z ,lim = 26.7 mag, about two magnitudes § 775 850 deeper than in the XMM1229 field. The i z colours were again measured on the 775 − 850 cross-convolved images within 200 fixed apertures. We adopted the method outlined in Pimbblet et al. (2002) to estimate the fraction of galaxies contaminating the XMM1229 red sequence, modifying the equation in Appendix A of their paper to take into account the galaxies with assigned spectroscopic redshift in the XMM1229 field. This method assigns to each galaxy within a certain range of magnitude and colour a probability of belonging to the field. As a result, this probability quantifies the amount of contamination of the observed colour-magnitude diagram in the cluster field. We divided the colour-magnitude planes of the GOODS and XMM1229 fields into cells of equal width in colour and magnitude and in each cell we computed the probability for each galaxy of being in the field:

Nfield A NXMM1229,cont Pfield = × − (3.1) NXMM NXMM ,cont 1229 − 1229 where Nfield is the number of galaxies in each cell of the GOODS colour-magnitude diagram, NXMM1229 is the number of galaxies in each cell of the XMM1229 observed colour- magnitude diagram, NXMM1229,cont is the number of galaxies in each cell of the XMM1229 observed colour-magnitude diagram which are known not to be in the cluster from their 3.4. Results 47 spectroscopic redshift, and A is the ratio between the areas of the XMM1229 ﬁeld and the GOODS ﬁeld. The main drawback of this method is that one can have Pfiled > 1 or

Pfiled < 0. As suggested by Pimbblet et al. (2002), in these cases the width of the cell is adjusted until 0.0 < Pfield < 1.0. For the purpose of this paper we only focused on the red sequence and we split it into two cells at 21.0 z < 22.5 and 22.5 z < 24.0 with the same colour width ≤ 850 ≤ 850 0.7 (i z ) < 1.1. This choice was motivated by the fact that the cells sample the ≤ 775 − 850 observed red sequence with sufficient resolution without falling into a regime of excessive low-number statistics (see Fig. 3.4). Furthermore, the magnitude limits are the same adopted for the estimation of the luminous to faint ratio in 3.5. The average Pfield for § the observed XMM1229 red sequence is Pfield 6% in the cluster centre and Pfield 32% ∼ ∼ in the outskirts. With the definition of field contamination given in Equation (1), the number of cluster members Ncluster in each colour-magnitude cell is defined as:

Ncluster = (1 Pfield) (NXMM NXMM ,cont). (3.2) − ∗ 1229 − 1229

We use this equation to correct for outliers in the estimation of the luminous-to-faint ratio (see 3.5). §

3.4.3 Colour-Magnitude Diagram and Red Sequence

The colour-magnitude diagrams for the cluster centre and outskirts are presented in Figure 4.4. In order to model the red sequence, we applied a robust linear fit, implementing the Tukey’s bi-square weight function, and restricting the fit to the photometrically confirmed members in the colour range 0.75 < (i z ) < 1.5 at z < 24.0. The model red 775 − 850 850 sequence was defined as:

(i z )RS = a + b (z 21.0) (3.3) 775 − 850 × 850 − where b is the slope and a represents the colour of a galaxy on the red sequence at z = 21.0. We obtained b = 0.044 0.017 and a = 0.94 0.03 for the cluster centre, 850 − where the uncertainties on a and b were estimated by generating 1000 bootstrap samples from the photometrically conﬁrmed members used to ﬁt the red sequence. Following

Lidman et al. (2004) and Mei et al. (2009), we estimated the intrinsic scatter σc of the 2 red sequence as the scatter that needed to be added to the colour error to have χe = 1.0, 2 2 where χ is the reduced χ . We found σc = 0.026 0.012, where the uncertainty was e 48 Chapter 3. Analysis Method again estimated by creating 1000 bootstrap samples from the sample of photometrically confirmed members used in the red sequence fit. We discuss the implications of these results in 3.5.4. § As shown in Table 3.2, the SofI J band 90% completeness limit is J = 22.4. This results into a loss of red sequence objects at magnitudes z850 > 22.5 in the photo-z selected outskirts sample. Therefore, in order to study the cluster outskirts. we first modelled the observed red sequence and then used Equation (2) to statistically subtract field interloper galaxies. We obtained the following result: a = 0.88 0.05, b = 0.01 0.03 and − σc = 0.052 0.015. It can be seen that the red sequence is shallower than in the cluster centre and has a larger intrinsic scatter, although the slopes and the intrinsic scatters for the two samples are still consistent. In the cluster centre the red sequence population was defined as those galaxies with

4σc < (i z ) (i z )RS < +7σc, where the limits were chosen after visually − 775 − 850 − 775 − 850 inspecting the colour-magnitude diagram, as the most suitable to bracket the red sequence.

For the cluster outskirts we used the boundaries 3σc and +4σc. − We defined the total galaxy magnitude as the extinction-corrected SExtractor MAG AUTO, although we are aware that this quantity underestimates the total flux, especially for elliptical and lenticular galaxies. Graham & Driver (2005) proposed general aperture corrections based on galaxy light profiles. However, in order to apply them to systems composed by a bulge and a disc, like S0 galaxies, one should first perform a bulge-disc decomposition which at z 1 becomes highly uncertain due to the high sky contamination at low fluxes. ∼ The photometric selection of the red sequence in the cluster centre produced a sample of 45 galaxies to which we added the bright spectroscopically confirmed member

XMM1229 145 (z850 = 21.9), falling just below the red sequence. This brought the final number of objects analysed in the cluster centre to 46. The red sequence members of the central region are listed in Table 3.3. We did not assign cluster membership to the single galaxies in the outskirts, as a detailed study of the cluster outskirts in the HCS will be presented in a forthcoming paper. For this reason in Table 3.3 we only report the 3 spectroscopically confirmed red sequence members in this subsample. Our sample almost doubles the number of red sequence galaxies analysed in Santos et al. (2009), who limited themselves to the spectroscopically confirmed cluster members. This allows us to study galaxy properties along the red sequence down to z850 = 24.0 mag, i.e. 1 mag fainter than in that work. 3.4. Results 49

Figure 3.4 (left): Colour-magnitude diagram of the central region of XMM1229. The colours are measured on the F775W and F850LP PSF cross-convolved images, adopting fixed circular apertures with 100 radius ( 8 kpc at z = 0.98). Black dots represent all the photometrically selected cluster members∼ and green dots are the red sequence members. Red dots are spectroscopically confirmed cluster members. The vertical dashed line represents the flux limit of visual morphology (see 3.4.4) and the sloping dotted line is the 90% completeness limit in the F775W and F850LP§ images. The black dashed line is the linear fit to the red sequence and the two dotted lines represent the -4σc and +7σc envelopes delimiting the red sequence. The error bars represent the median colour errors along the red sequence in bins of 0.5 magnitudes. (right): Observed colour-magnitude diagram in the outskirts of XMM1229. No cut in photometric redshifts is applied for this sample. Grey dots are all the galaxies observed in the cluster outskirts, red dots are spectroscopically confirmed cluster members and green dots are red sequence galaxies selected as described in 3.4.3. Error bars represent the median colour errors along the red sequence in bins of 0.5§ magnitudes. The meaning of the lines is the same of the left panel. 50 Chapter 3. Analysis Method (early) (early) (early) Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Morphology Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Disc-dominated Disc-dominated Disc-dominated spec tral region z 0.979 0.984 0.976 0.984 0.969 0.974 Cen tinued on next page phot 0.99 0.94 0.98 0.95 0.87 0.92 1.07 0.94 0.92 1.04 1.02 0.93 0.84 0.97 Con sub-sample); (2) spectroscopically confirmed red sequence members between 0.6 and outskirts); (3) spectroscopically confirmed cluster members in the blue cloud or with δ z centre +1:51:1.23 +1:51:5.37 +1:51:21.29 +1:51:22.78 +1:51:25.77 +1:50:35.57 +1:50:46.29 +1:50:54.82 +1:51:16.37 +1:51:21.82 +1:51:21.32 +1:51:20.90 +1:51:20.26 +1:51:19.77 .centre Spectroscopically confirmed members are indicated with their spectroscopic redshift α 12:29:29.297 12:29:29.420 12:29:24.819 12:29:25.972 12:29:29.028 12:29:28.428 12:29:31.043 12:29:29.199 12:29:28.315 12:29:29.944 12:29:27.718 12:29:31.099 12:29:32.062 12:29:30.160 237 240 241 243 244 248 255 260 128 145 172 190 200 229 §3.4.1 in the cluster ID XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 able 3.3: Red sequence members of XMMU J1229+0151 (XMM1229). Red sequence members are photometrically selected as T explained in and the estimatedprojected photometric Mpc redshift. from the The cluster table centre ( iscorrupted divided photometry; into (4) spectroscopically the confirmed(3) following cluster and members sections: (4) detected only of (1) in this red the table HAWK-I Ks sequence are image. excluded members Objects from within in the Sections 0.6 analysis in this paper. 1.04 projected Mpc from the cluster centre ( 3.4. Results 51 (late) (early) (early) (early) Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Elliptical Morphology Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Disc-dominated Disc-dominated Disc-dominated Disc-dominated spec z 0.969 0.977 0.976 tinued on next page phot 0.93 0.84 0.92 1.07 0.97 1.11 0.85 1.14 0.90 0.83 1.07 0.95 1.05 0.85 0.98 1.16 0.94 0.96 0.98 0.89 0.90 Con δ z able 3.3 – continued from previous page T +1:52:0.98 +1:52:3.62 +1:51:52.51 +1:51:54.91 +1:51:56.43 +1:51:24.20 +1:51:23.84 +1:51:24.92 +1:51:30.16 +1:51:28.46 +1:51:28.81 +1:51:31.27 +1:51:32.61 +1:51:33.33 +1:51:37.39 +1:51:36.97 +1:51:39.75 +1:51:38.12 +1:51:39.12 +1:51:40.51 +1:51:46.74 α 12:29:32.176 12:29:28.686 12:29:29.059 12:29:27.637 12:29:32.850 12:29:29.704 12:29:27.774 12:29:26.871 12:29:31.375 12:29:28.258 12:29:27.127 12:29:29.448 12:29:33.193 12:29:28.803 12:29:29.665 12:29:27.661 12:29:28.714 12:29:30.026 12:29:29.507 12:29:26.924 12:29:28.929 320 322 331 353 380 392 394 414 415 283 286 288 291 306 309 310 312 319 262 263 265 ID XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 52 Chapter 3. Analysis Method (early) (early) (early) (blue cloud) (early) (blue cloud) Elliptical Elliptical Elliptical Elliptical Elliptical Morphology Morphology Morphology Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Bulge-dominated Disc-dominated Disc-dominated Disc-dominated Disc-dominated region spec spec spec z z z 0.969 0.980 0.973 0.977 0.974 0.980 0.979 0.973 0.969 tinued on next page Outskirts phot phot phot 0.96 1.09 0.87 0.96 0.99 0.94 0.98 2.01 0.96 0.97 0.95 1.01 0.92 0.95 1.02 1.01 Con δ z δ z δ z able 3.3 – continued from previous page T +1:52:4.92 +1:52:6.73 +1:52:6.90 +1:51:00.95 +1:51:36.54 +1:52:13.31 +1:52:12.78 +1:52:16.55 +1:52:15.97 +1:52:18.41 +1:52:19.17 +1:52:29.49 +1:51:29.60 +1:50:11.02 +1:51:46.39 +1:51:52.19 ectroscopic members within ACS field excluded from present analysis Sp α α α 12:29:30.520 12:29:33.615 12:29:33.260 12:29:23.202 12:29:32.951 12:29:32.598 12:29:31.313 12:29:29.264 12:29:28.867 12:29:30.523 12:29:29.187 12:29:29.116 12:29:28.341 12:29:32.279 12:29:27.302 12:29:28.330 73 349 373 183 308 463 470 475 477 502 287 429 437 441 456 457 ID ID ID XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 XMM1229 3.4. Results 53 FLAGS=16, not used) (early) (blue cloud) Elliptical Elliptical Elliptical Elliptical Morphology Morphology Bulge-dominated Irregular Disc-dominated (late) (SExtractor Disc-dominated spec spec z z 0.973 0.968 0.975 0.974 0.976 0.976 0.978 0.973 ...... phot phot 0.72 δ z δ z ectroscopic members outside ACS field of view able 3.3 – continued from previous page T Sp +1:52:21.73 +1:51:36.42 +01:53:51.79 +01:53:33.84 +01:53:41.62 +01:53:33.19 +01:53:20.62 +01:53:13.54 α α 12:29:27.151 12:29:20.220 12:29:16.481 12:29:17.113 12:29:20.813 12:29:24.002 12:29:29.643 12:29:25.778 487 316 4661 4794 4800 4910 4956 5001 ID ID ORS2 ORS2 ORS2 ORS2 ORS2 F F F F F FORS2 XMM1229 XMM1229 54 Chapter 3. Analysis Method

3.4.4 Galaxy Morphology and Structure

We have shown in 3.4.2 that the outskirts of XMM1229 have a considerably large outlier § contamination and therefore in this paper we restrict the morphological and structural analyses of red sequence galaxies to the cluster centre. The morphological and structural analyses of the cluster outskirts in the entire HCS sample will be the subject of a forthcoming paper.

Morphological Classiﬁcation

We classified red sequence galaxies in the centre of XMM1229 using the ACS F850LP image, on which we were able to detect morphological features down to z850 = 24.0. The F850LP filter corresponds approximately to a rest frame SDSS g band, allowing us to compare directly with lower-redshift classifications performed either in the B or V bands (e.g. Fasano et al. 2012). Galaxies were classified by three of the authors independently (P. C., W. J. C. and C. L.) on image cutouts whose size varied according to the SExtractor Kron radius of each object. We also ran the galSVM software (see Huertas-Company et al. 2008, 2009b, 2011 and Appendix B) on the entire F850LP image, which provided us with a fourth independent and quantitatively-based classification. Because at z = 1 many higher-order morphological and structural features, such as spiral arms, bars and lenses are not resolved, we classified galaxies according to their observed bulge-to-total ratio (B/T) and split the red sequence sample into five broad morphological families: elliptical (E), bulge-dominated (BD), early disc-dominated (EDD), late disc-dominated (LDD) and irregular galaxies (Irr). A similar classification scheme was adopted by Postman et al. (2005) and Mei et al. (2012) to classify galaxies in ACS images of z 1 clusters and it allows us to investigate the main structural features of cluster ∼ members. Most S0s fall into the class of bulge-dominated galaxies, while the early and late disc classes comprise Hubble types going from Sa to Sbc and Sc to Scd, respectively. In the following we will use the terms ‘bulge-dominated’ and S0 interchangeably, although we are aware that with our scheme the bulge-dominated sample may be contaminated by Sa galaxies (see e. g. Mei et al. 2012) and the disc-dominated samples may be contaminated by low (B/T) S0 galaxies (S0c galaxies, Laurikainen et al. 2011). In fact, spiral arms become fainter in the gas-poor galaxies which are typical of the red sequence population. The four morphological classifications of the cluster centre agreed only for 4 galaxies, while 11 galaxies had the same type assigned in the three visual classifications. In particular, we note that while W. J. C. and C. L. agreed on 33 of the 46 red sequence 3.4. Results 55 galaxies, P. C. agreed with each author only on 16 and 17 galaxies, respectively. The main points of disagreement were the E/S0 distinction and the fact that faint elliptical galaxies tended to be classified by P. C. as disc-dominated systems with a very compact bulge and a faint disc. This underlines the challenges in morphological classifications of high redshift passive galaxies. For this reason we decided to adopt a majority rule for the assignment of the morphological types and thus the final morphological type was defined as the mode of the four independent classifications. There were four galaxies for which two of the classifiers assigned the type E and the other two the type BD, while in two other cases two of the classifiers assigned the type BD and the other two the type EDD. In these six cases we assigned type E to the 4 galaxies with no majority between type E and type BD and type BD to those with no majority between type BD and type EDD. For one galaxy (XMM1229 322) two classifiers assigned type E and the other two type EDD. In this case we decided to classify the galaxy as bulge-dominated. Given the very low number of objects with late-type disc and irregular morphologies in both the XMM1229 and the two lower-redshift comparison samples (see 3.4.5), we decided to merge the two classes § into one “late disc-dominated / irregular” morphological class. However, for completeness, in Tables 3.3, 3.4 and 3.5 we report the original morphological scheme with five classes.

The top panels of Fig. B.1 show that, as expected, disc-dominated galaxies tend to have lower values of concentration and Gini coeﬃcient with respect to elliptical and S0 galaxies (see e.g. Lotz et al. 2004). We also note that the values of M20 (lower left panel of Fig. B.1) for disc-dominated galaxies are comparable with those of early-type galaxies. We interpret this result as a consequence of the fading of spiral arms in gas-poor spiral galaxies.

The morphological classiﬁcations for the cluster centre and outskirts are reported in Table 3.3 (column 6), while thumbnail images for the galaxies listed in that table can be found in Fig. A.1-A.4. The morphology quoted for the cluster outskirts corresponds only to the output of galSVM.

Santos et al. (2009) classified visually red sequence galaxies in XMM1229 using a scheme similar to ours: of the 15 galaxies in common with our cluster centre sample, there are 9 galaxies with the same assigned morphological type. When comparing only early- and late-type galaxies, we find that 13 of the 15 galaxies have the same assigned type. Delaye et al. (2014) used galSVM to classify galaxies in the HCS clusters, dividing them into early- and late-type. The comparison with our classification shows that of the 45 galaxies in common to both samples, 38 (i.e. 84%) were classified as early or late-type in both works. 56 Chapter 3. Analysis Method

Figure 3.5 Morphological evolution of red sequence galaxies in clusters at 0.04 < z < 0.98. Left: magnitudes are normalised to the brightest bin in each sample. (Top left panel): morphological fractions along the red sequence of XMM1229. (middle left panel): fraction of morphological types along the red sequence of the spectroscopically conﬁrmed MORPHS cluster members (0.3 < z < 0.6). (bottom left panel): morphological fractions along the red sequence of the WINGS spectroscopically conﬁrmed cluster members. The analysis in WINGS is restricted to objects with V < 18.0 mag, corresponding to at least 50% spectroscopic completeness. The bright end of the red sequence is dominated by elliptical galaxies as in MORPHS and XMM1229. At intermediate and low luminosities, S0 galaxies are the most frequent morphological class. On the top we report the apparent z850 magnitude scale (down to the limit z850 = 24.0 mag considered in this paper). Right: morphological fractions as a function of V absolute magnitude. (Top right panel): XMM1229; (middle right panel): MORPHS; (bottom right panel): WINGS. In order to match the scales of the three samples, the magnitudes were passively evolved to z = 0. The absolute magnitude axis is reproduced on the top of the plot. For clarity, in all the plots, the points for each morphological type are shifted along the x-axis by 0.04 mag. 3.4. Results 57

The Morphological Composition of the Red Sequence.

We divided the red sequence of the central region into bins of 0.5 magnitudes in the range

21.0 < z850 < 24.0 and, in each bin, we computed the fraction of the galaxy population of a certain morphological type, FT , as:

NT,i FT,i = (3.4) Ntot,i

th where FT,i, is the fraction of galaxies of type T in the i bin, NT,i is the number of galaxies th th of type T in the i bin, and Ntot,i is the total number of galaxies in the i bin. The error bars were computed following the method outlined in Cameron (2011) to estimate the confidence intervals for a binomial probability distribution. More precisely, in each magnitude bin, the confidence interval was estimated using a Bayesian approach in which the binomial probability mass function was treated as the a posteriori probability distribution, given a uniform prior over the expected number of successes. The errors on the morphological fractions were therefore estimated as the difference between the measured value of the fraction and the upper and lower bounds of the confidence interval. This method is shown to give reliable confidence intervals even with small samples, as in the case of this paper. Furthermore, it allows us to treat without ambiguity the extreme cases FT = 1 and FT = 0. In fact, a gaussian approximation of the binomial distribution, suitable for large samples, would produce null errors, meaning certain estimates of the true value of the morphological fraction. We dealt with these two extreme cases defining as best estimate of FT the median of the a posteriori probability distribution and we estimated the errors as the difference between this value and the upper and lower bounds of the binomial confidence interval. The morphological fractions as a function of magnitude along the red sequence in XMM1229 are illustrated in the top panels of Fig. 3.5 and will be discussed in 3.5. §

Light Proﬁle Fitting and Structural Parameters

In order to investigate the connection between morphological and structural properties in red sequence galaxies, we fit a Sérsicfunction to the light distribution of the red sequence members. We used GALFIT (Peng et al., 2002, 2010a), implemented as part of the GALAPAGOS pipeline, on the whole F850LP image. The SérsicLaw is parametrised by the equation: (1/n) −κ[(r/re) −1] Σ(r) = Σee (3.5) 58 Chapter 3. Analysis Method where Σ(r) is the galaxy surface brightness as a function of projected radius, re is the half light radius, Σe is the surface brightness at the half light radius, n is the Sérsicindex and κ is a parameter which is coupled with n in such a way that half of the total light is always enclosed within re. The Sérsicindex is related to galaxy light concentration: higher values of n correspond to more concentrated light profiles. Therefore, spheroidal galaxies are expected to have higher Sérsicindices, while disc-dominated galaxies are characterised by low values of n. When n = 1, eq. (4) assumes the form of the Exponential Law, which is typical of spiral galaxies, while when n = 4, it assumes the form of the De Vaucouleurs Law (de Vaucouleurs, 1948), typical of elliptical galaxies.

Graham & Guzmán(2003) showed that the Sérsicindex correlates with the luminosity and stellar mass of elliptical galaxies, so that bright elliptical galaxies in cluster cores may reach up to n 10, while dwarf ellipticals may also have n 2 (see also Graham et al. ∼ ∼ 1996). Furthermore, Kormendy & Djorgovski (1989) found that the De Vaucouleurs Law holds only for elliptical galaxies in a narrow magnitude range centred at MB = 21. For − these reasons, unlike Santos et al. (2009), who constrained the value of the Sérsicindex between 1 and 4, we kept n as a free parameter when running GALFIT.

In order to test the reliability of GALFIT, we inserted 580 simulated galaxy images in random spots of the original F850LP image. The galaxies were built using the latest version of the simulation script developed by the MEGAMORPH collaboration (kindly provided by Boris Häußler;see also Häußleret al. 2013) and based on the method outlined in Häussleret al. (2007). The advantage of this method, with respect to other more traditional software (e.g. IRAF mkobject), is that the inner regions of the simulated galaxies are oversampled to take account of the higher curvature of the light profile towards the centre. In order to fully investigate the performance of GALFIT, no correlation was assumed between n and the luminosity of the mock galaxies. The results, after running GALAPAGOS with the same settings used for the original F850LP image, are shown in

Fig. 3.6. It can be seen that, at z850=24.0, GALFIT can still produce a reliable fit. From the bottom right panel of Fig. 3.6 it can also be seen that the output Sérsicindex appears slightly underestimated, although still consistent with the input value, at n > 3.5. However, we note here that there are less simulated objects with these values of the Sérsic index. The results of the Sérsicfit for the red sequence members of XMM1229 are shown in Fig. 3.7 and are discussed in 3.5.3. § 3.4. Results 59

Figure 3.6 Performance of GALFIT on the F850LP image. Grey points represent the results for each retrieved simulated galaxy image, while black points and error bars are the median and half of the 68% width of the distribution of each quantity in each bin on the x-axis. (Top left): galaxy magnitude: ∆z850 as a function of input magnitude; (top right): half light radius: ∆R50 as a function of input magnitude; (bottom left): Sérsic index: ∆n as a function of input magnitude; (bottom right): Sérsic index: ∆n as a function of input Sérsicindex. The width of the input magnitude bins is 0.5 mag, while the width of the Sérsicindex bins is 1. The vertical dashed line represents the z850 = 24.0 mag limit for visual morphological classification (see 3.4.4). § 60 Chapter 3. Analysis Method

3.4.5 The low-redshift Cluster Samples

MORPHS

In order to compare with galaxy clusters at lower redshift, we built two comparison samples using the spectroscopic catalogues of the MORPHS and WINGS surveys. The MORPHS sample (Smail et al., 1997) comprises ten clusters in the redshift range 0.3 < z < 0.6, observed with the Prime Focus Universal Extragalactic Instrument (PFUEI) on the 200 inch Palomar Hale telescope, and followed up with the Wide Field and Planetary Cam- era 2 (WFPC2) on HST. The clusters were observed in the g and r bands from the ground, while the WFPC2 images were taken in the F450W, F555W, F702W and F814W bands. Spectroscopic follow-up observations were conducted with the Carnegie Observa- tories Spectroscopic Multislit and Imaging Camera (COSMIC), on the Palomar 200-inch, the Low Dispersion Survey Spectrograph 2 (LDSS-2), on the William Herschel Telescope (WHT) at the Roque de los Muchachos Observatory (Spain), and the ESO Multi-Mode Instrument (EMMI) on NTT. The MORPHS spectroscopic observations and data reduction are described in Dressler et al. (1999). The sample comprises in total 424 cluster members. The aim of the MORPHS project was to study the morphological and spectral properties of galaxies in clusters at intermediate redshifts; therefore an accurate visual morphological classification was conducted on the WFPC2 images of each cluster field (see Smail et al., 1997). The subsample of morphologically classified galaxies in the MORPHS spectroscopic sample comprises 122 cluster members. Since the redshift range of the MORPHS clusters allowed morphological features such as spiral arms or bars to be distinguished on the WFPC2 images, the classification scheme consisted of 9 types going from -7 to 10 and corresponding to the De Vaucouleurs T types (de Vaucouleurs et al., 1976). In order to convert to the morphological scheme used for XMM1229, we grouped the galaxy morphological types and built broad classes similar to those used at z 1. The type conversion ∼ was performed following Poggianti et al. (1999) and is summarised in Table 3.4. In order to identify red sequence galaxies in these clusters, we transformed all the photometry on to the rest-frame B and V system. In fact, only a part of the MORPHS WFPC2 observations were taken with a blue (F450W or F555W) and a red filter (F814W). The clusters observed with the F702W filter did not have any blue image and for them we had to resort to the ground-based g and r observations (Dressler & Gunn, 1992). We determined the k-correction using the software zebra (Feldmann et al., 2006) and a passive template SED from Coleman et al. (1980). To facilitate the comparison with 3.4. Results 61

XMM1229, we also converted the MORPHS magnitudes to AB magnitudes. We ﬁtted the red sequence in the (B-V) vs V rest-frame colour-magnitude diagram using the same procedure adopted for XMM1229 and outlined in 3.4.3. We divided the red sequence § into bins of 0.5 magnitudes each, in the range 23.0 < V < 20.0 and, in each bin, − − we measured the fractions of the single morphological types. The depth of the MORPHS spectroscopic observations does not allow us to investigate the faint end of the red sequence, however it constitutes a good sample to study the behaviour of the morphological fractions at bright luminosities. Since the spectroscopic follow-up of MORPHS was mainly focused on the study of the Butcher-Oemler eﬀect, the target selection was biased towards bluer galaxies. Hence we corrected the morphological fractions using the normalisation factors given in Table 2 of Poggianti et al. (1999). The results are shown in the middle panels of Fig. 3.5 and are discussed in 3.5.2. Given the smaller population of the MORPHS § spectroscopic sample, we did not restrict the morphological analysis of the red sequence to Rcluster < 0.54 R as done for XMM1229 and WINGS (see 3.4.5). × 200 §

WINGS

The WINGS cluster survey consists of 77 X-ray selected galaxy clusters in the range 0.04 < z < 0.07, observed in the B and V bands with the Isaac Newton Telescope (Roque de los Muchachos Observatory) and MPG/ESO-2.2 m Telescope (La Silla). Spectroscopy from observations conducted at the WHT and the 3.9 m Anglo-Australian Telescope (AAT) is available for a subsample of 48 clusters with a total of 6000 redshifts (Cava et al., 2009). Given the diverse spectroscopic coverage of each cluster and the consequent impossibility of studying the clusters individually, we built a composite sample taking all the spectroscopically conﬁrmed members in each WINGS cluster with apparent magnitude V < 18.0. In fact, according to Fig. 3.5 of Cava et al. (2009), the spectroscopic success rate in this ﬂux region is greater than 50%. We took into account the incompleteness of the sample along the red sequence weighing each galaxy by the inverse of the spectroscopic success rate at its magnitude. A near-infrared (NIR) follow-up for 28 WINGS clusters was carried out by Valentinuzzi et al. (2009) in the J and K band of the Wide Field Camera (WFCAM) at the UK Infrared Telescope (UKIRT) in Hawaii.

In order to compare the same regions in units of R200, we restricted the analysis of the WINGS clusters to 0.54 R , corresponding to the same physical area covered by the × 200 XMM1229 centre sample. After converting on to the AB system and deriving B and V rest-frame absolute magnitudes, we ﬁtted the red sequence applying the same method used for XMM1229 and 62 Chapter 3. Analysis Method

Table 3.4 MORPHS visual classiﬁcation and type conversion (See 3.4.4 for details on the morphological scheme adopted for XMM1229). § T type Morphology XMM1229 morphology -7 D/cD Ellipticals -5 E -2 S0 Bulge-dominated 1 Sa Disc-dominated (early) 3 Sb 5 Sc 7 Sd Disc-dominated (late) 9 Sm 10 Irr Irregulars

MORPHS. The WINGS morphological classiﬁcation was taken from Fasano et al. (2012), who used an approach based on neural networks. Their morphological scheme consists of 18 types that we grouped into broad classes to match the classiﬁcation used for XMM1229 (see Table 3.5). The morphological composition of the red sequence in the WINGS cluster sample is shown in the bottom panels of Figure 3.5 and will be discussed in 3.5.2. §

3.5 Discussion

3.5.1 The Faint End of the Red Sequence

Most authors quantify the galaxy deﬁcit at the faint end of the red sequence with the ratio between the numbers of luminous and faint galaxies: Nlum/Nfaint, also known as the luminous-to-faint ratio. This is equivalent to constructing the red sequence luminosity function dividing the sample into only two magnitude bins. The most used convention is to consider as luminous all the red sequence members with MV < 20.0 and as faint all − the red sequence members with 20.0 MV < 18.2, where the absolute magnitudes − ≤ − are measured in the Vega system and are passively evolved to z = 0. This particular subdivision was ﬁrst proposed by De Lucia et al. (2007b) for galaxy clusters at 0.4 < z <

0.8 and was motivated by the fact that their data reached 5σ completeness at MV = 18.2, − while MV = 20.0 corresponded to about the midpoint of the red sequence. These limits − allowed them to study the red sequence down to four magnitudes fainter than the BCG in each cluster, and could easily be applied to lower-redshift samples (see e. g. Gilbank & Balogh 2008; Capozzi et al. 2010; Bildfell et al. 2012).

In order to estimate Nlum/Nfaint in XMM1229, we had to convert our AB z850 apparent magnitude to Vega V-band absolute magnitudes. For this purpose we used a Bruzual & 3.5. Discussion 63

Table 3.5 WINGS morphological classiﬁcation and type conversion (See 3.4.4 for details on the morphological scheme adopted for XMM1229). § type Morphology XMM1229 morphology -6 cD Ellipticals -5 E -4 E/S0 -3 S0− Bulge-dominated -2 S0 -1 S0+ 0 S0/a 1 Sa Disc-dominated (early) 2 Sab 3 Sb 4 Sbc 5 Sc Disc-dominated (late) 6 Scd 7 Sd 8 Sdm 9 Sm 10 Im Irregulars 11 compact Im

Charlot (2003) simple stellar population (SSP) model with solar metallicity, formation redshift zf = 4.75 and exponentially declining star formation with e-folding time τ = 1 Gyr. We used the k-correction estimated for this model by the Bruzual & Charlot (2003) colour and magnitude evolution software and converted to V-band magnitudes. The obtained V-band absolute magnitudes were ﬁnally passively evolved to z = 0. We also tried exponentially declining star formation models with zf = 3 and zf = 4, as well as SSP models with an initial burst of star formation and zf = 3, 4, 4.75, but we found that the model with exponentially declining star formation and zf = 4.75 was the best in reproducing the average observed i z colour at z = 0.98. After the conversion to Vega − V-band magnitudes, the 21.0 < z850 < 24.0 range over which we study the properties of the red sequence in XMM1229 was mapped on to the 22.0 < MV < 19.0 absolute − − magnitude range. We note that our V-band limit is about one magnitude brighter than that adopted by De Lucia et al. (2007b) and other authors at lower redshifts. However, if we considered magnitudes fainter than z850 = 24.0, we would fall into the incompleteness region of the colour-magnitude diagram (see Fig. 3.4). Thus, we ﬁxed the boundary between bright and faint galaxies at MV = 20.5 (z = 22.5), which corresponds to − 850 the midpoint of the XMM1229 red sequence and measured Nlum/Nfaint over the 22.0 < − 64 Chapter 3. Analysis Method

MV < 19.0 magnitude range. With this deﬁnition, we obtain Nlum/Nfaint = 0.70 0.15. − We measured the luminous-to-faint ratio on the composite spectroscopic WINGS sample applying the same V-band absolute magnitude cuts and obtaining: Nlum/Nfaint =

0.364 0.006. The errors on Nlum/Nfaint were estimated again following Cameron (2011). Although Nlum/Nfaint is higher in XMM1229, as described below, the expected cluster-to- cluster scatter in WINGS is large, so we cannot conclude that there is evidence of evolution in Nlum/Nfaint based on this initial sample. If we use statistical background subtraction to correct for contamination instead of photometric redshifts, we obtain Nlum/Nfaint = 0.68 0.13, which does not change our conclusions. The existence of a deficit at the faint end of the red sequence supports the notion that less massive galaxies settle on to the red sequence at later epochs. As found by Demarco et al. (2010) at z = 0.84, the faint end of the cluster red sequence is populated by young low-mass early-type galaxies that had recently ceased to form stars. However, this scenario is still questioned by the studies of Andreon (2008) and Crawford et al. (2009), who found little or no decrease in the number of faint galaxies in clusters up to z = 1.3. These authors attributed the observed deficit to measurement errors or cluster-to-cluster variations, claiming that the deficit may be a phenomenon that affects only some clusters. From Fig. 3.4 in Andreon (2008), we find that at z < 1, there is mild correlation between log (Nlum/Nfaint) and log (1 + z) (Spearman coefficient ρ = 0.45). At z > 1.0, Nlum/Nfaint appears constant. We note that for these clusters the red sequence was selected using the ACS i775 and z850 bands, which do not bracket the 4000 A˚ break at those redshifts. Therefore, the measurements of Andreon (2008) may be affected by fore- and background contamination of the red sequence sample. On the other hand, Capozzi et al. (2010), collecting different samples of clusters at z < 0.8, found a significantly stronger correlation (ρ = 0.89). Cluster-to cluster variations play an important role in the study of the build-up of the red sequence and they need to be taken into account especially when dealing with small samples. Valentinuzzi et al. (2011) estimated Nlum/Nfaint for 72 clusters in the WINGS photometric sample individually, and using statistical background subtraction to correct for interlopers. They used the same magnitude limits defined in De Lucia et al. (2007b) and from the analysis of the distribution of their results, presented in their Fig. 3.4, we

ﬁnd (Nlum/Nfaint) = 0.5 0.2, where the uncertainty corresponds to one half of the median 68% width of the Nlum/Nfaint distribution. This places XMM1229 in the upper tail of the Nlum/Nfaint distribution of WINGS.

Since the measurements of Nlum/Nfaint for individual clusters are characterised by 3.5. Discussion 65

Figure 3.7 (Left panel): Sérsicindex versus z850 magnitude along the red sequence. Galax- ies with later morphological types tend to have lower Sérsicindices. The error bars represent the median GALFIT errors on the Sérsicindex in bins of 0.5 magnitudes. On the top we also report the Vega MV passively evolved magnitudes (see 3.5.2 for details). (Central panel): Sérsicindex vs stellar mass along the red sequence. There§ is weak correlation between Sérsicindex and stellar mass and disc-dominated galaxies tend to have lower masses. (Right panel): distribution of the values of the Sérsicindex for red sequence galaxies. large uncertainties, particularly at high redshift (see also De Lucia et al. 2007b), comparisons using a single cluster need to be treated with caution. In order to use the luminous-to-faint ratio to detect any build-up of the red sequence at low masses in galaxy clusters and to measure any trends with redshift, as done by Capozzi et al. (2010), Bildfell et al. (2012) and Andreon (2008), one needs to study a statistically significant number of clusters. In a forthcoming paper, we will use the complete HCS sample to study the existence of any correlation of Nlum/Nfaint with redshift extending in this way to higher redshifts the works of Capozzi et al. (2010) and Bildfell et al. (2012), and looking at those epochs where Andreon (2008) found no correlation.

We also estimated Nlum/Nfaint in the cluster outskirts ﬁnding: Nlum/Nfaint = 0.35 0.11, which is consistent with no deﬁcit of faint galaxies at large distances from the cluster centre. However, we note that this result is still consistent to within 2σ with Nlum/Nfaint in the cluster centre. 66 Chapter 3. Analysis Method

3.5.2 Morphological Evolution

The left panel of Fig. 3.5 illustrates the comparison between the trends of the morphological fractions along the red sequence in XMM1229 and in the MORPHS and WINGS composite samples. In order to match the magnitude scales of the different samples, we offset the magnitudes to the brightest bin in each sample. In this way, we suppressed second-order effects such as the mismatches introduced by luminosity evolution and residual differences in photometry and k-correction, leaving only the magnitude differences between galaxies along the red sequence. This allowed us to compare more easily the trends of morphology along the red sequence, providing at the same time a clearer picture of the luminosity distribution for each morphological type. However, for completeness, in the right-hand panel of the figure, we also plot the morphological fractions as a function of absolute V-band magnitude passively evolved to z = 0. The top left panel of Fig. 3.5 shows that in XMM1229 the red sequence is mostly populated by elliptical and S0 galaxies. At the bright end, the red sequence is dominated by ellipticals, while going towards lower luminosities, the fraction of S0s increases and this class becomes predominant. At 22.5 < z < 23.5 ( 20.9 < MV < 19.9) the fractions of 850 − − ellipticals and S0s are comparable. Interestingly, we also see that at the same magnitudes there is a slight increase in the fraction of disc-dominated galaxies. At z850 > 23.5, the fraction of disc-dominated galaxies becomes dominant. The comparison with the WINGS sample shows that, at the relative magnitudes corresponding to the faint end in XMM1229, the red sequence is dominated by elliptical and S0 galaxies. We note a dearth of S0 galaxies in WINGS at the relative magnitudes in which S0s dominate the red sequence in XMM1229. However, it should be noted that this region of the red sequence is up to two magnitudes brighter in WINGS with respect to XMM1229 (see right-hand panel of Fig. 3.5). At similar absolute magnitudes the fractions of elliptical and S0 galaxies in the two samples look similar. This suggests that the bright end of the red sequence underwent significant evolution in the last 8 Gyr, becoming almost only populated by elliptical galaxies probably formed as a result of subsequent dry mergers (see Faber et al. 2007; Lidman et al. 2013). Vulcani et al. (2011a) studied the morphological fractions as a function of stellar mass in WINGS and in the clusters of the ESO Distant Cluster Survey (EDisCS, White et al. 2005), at 0.4 < z < 0.8. These authors excluded the BCGs from their analysis and still found a mild increase in the fraction of 10.9 elliptical galaxies at low redshift, at stellar masses above 10 M . The large difference in V-band luminosity between WINGS and the two higher-redshift samples overpredicts the mass growth of the brightest cluster galaxies by a factor of 2 3.5. Discussion 67

(Lidman et al., 2012). In order to test the reliability of our estimate of the WINGS V- band absolute magnitudes, we looked at the (V K) vs K colour-magnitude diagram for − all the WINGS clusters with V and K band data. In fact, the K band is a good tracer of old stellar populations and therefore it is a good proxy of stellar mass, while the V band is more sensitive to young stars. The adopted zf = 4.75 model with exponentially declining star formation history predicts (V K)V ega 3.3 at z = 0.055, the median redshift of − ∼ the WINGS clusters. We found that at VV ega < 22.0, for some of the clusters, the slope − of the (V K)V ega red sequence flattened or turned positive, making the average colour − of red sequence galaxies bluer. A flattening of the WINGS red sequence at VV ega < 21.5 − in the (B V ) vs V colour-magnitude diagram was also mentioned by Valentinuzzi et al. − (2011), although the authors did not discuss it. Interestingly, Bower et al. (1992) studied the (V-K) colour of galaxies in the Virgo and Coma clusters finding (V K)V ega 3.3 at − ∼ the bright end of the red sequence, consistent with model predictions. The change in slope of the red sequence in these clusters and the resulting high V-band luminosities can be the consequence of residual colour gradients or systematics in the WINGS photometry such as PSF variations or differences in the processing of the B, V and K band data. However, some authors found that cooling flows can induce star formation in BCGs (Crawford et al., 1999; Rafferty et al., 2008), while simulations by Jiménezet al. (2011) show that dry mergers between luminous red sequence galaxies and lower mass companions can result in more massive galaxies with lower total metallicity. These processes would all result in bluer colours at the bright end of the red sequence.

However, we point out that there are still big uncertainties in models of galaxy evolution. As outlined in detail by Skelton et al. (2012), the purely passive evolution of the red sequence, used in this work to compare clusters at different redshifts, underpredicts the brightness of the galaxies at low redshift, especially at the bright end of the red sequence. In particular, those authors point out that the concomitant effects of merger and star formation can explain both the slower evolution of the red sequence over the redshift range 0.0 < z < 1.0 and the flattening of its bright end. Therefore, in order to rigorously compare red sequence galaxies in clusters at different redshifts, both merger and star formation must be taken into account and the passively-only evolved absolute magnitudes reported in the right-hand panel of Fig. 3.5 should be considered as upper limits to the z = 0 analogues of the three cluster samples.

The faint end of the red sequence in XMM1229 is characterised by the increase in the fraction of disc-dominated galaxies. At similar absolute magnitudes (i.e. 20 < MV < − 19.0), the WINGS red sequence is dominated by S0 and elliptical galaxies. This suggests − 68 Chapter 3. Analysis Method that the disc-dominated galaxies observed in XMM1229 may be the progenitors of the S0 galaxies observed at similar absolute magnitudes in WINGS. Such a transformation could be the result of the stripping of gas from spiral galaxies that are falling into the cluster and are depleted of their gas reservoirs (strangulation). However, Bekki & Couch (2011) found that tidal interactions, due to slow encounters with other cluster members, can trigger bursts of star formation in the bulge of spiral galaxies, increasing the bulge fraction and transforming gas-rich spiral galaxies into gas-poor S0 galaxies. Owing to the recently formed stars, these galaxies should also have younger stellar ages. According to Bekki & Couch (2011), such a process should be more important for lower-mass spiral galaxies 10 ( 1.2 10 M ), and for objects residing near the cluster core. Interestingly, Poggianti ∼ × et al. (2001) showed that faint red sequence S0 galaxies in the Coma cluster are on average younger than their bright counterparts and with evidence of recent star formation. Fig. 3.8 shows the co-added FORS2 spectra of XMM1229 red sequence galaxies grouped by morphological type. It can be noted that the Hδ absorption line, a proxy for recent star formation, becomes stronger as one moves from elliptical to disc-dominated galaxies, suggesting that star formation terminated at more recent times in the latter and that these could be galaxies that just joined the red sequence at z = 0.98.11 A similar trend of the strength of the Hδ with galaxy morphology was found by van Dokkum et al. (1998) and Tran et al. (2007) in clusters at z 0.3 and z 0.8, respectively. Interestingly, ∼ ∼ Demarco et al. (2010) showed similar trends between Balmer features and morphology for all cluster galaxies (not only red sequence members) in the galaxy cluster RX J0152.7-1357 at z = 0.84 (see their Fig. 3.5).

The MORPHS sample shows a higher fraction Fdisc of disc-dominated galaxies, compared to both XMM1229 and WINGS. We also note that, as in XMM1229, disc-dominated galaxies become more frequent at fainter magnitudes. Sánchez-Blázquezet al. (2009) found that the fraction of red sequence early-type galaxies FE+S0 decreases with cosmic time in the range 0.4 < z < 0.8 in the EDisCS clusters, with a corresponding increase in the fraction of late-type galaxies. The depth of the MORPHS spectroscopic sample does not allow us to study the faint end of the red sequence and therefore to derive a reliable measurement of the disc fraction. However, Sánchez-Blázquezet al. (2009) measured FE S = (57 9)% + 0 in MORPHS, which is consistent with the decreasing trend observed in their sample from

75% to 55% at 0.4 < z < 0.8. These authors attributed the increase in Fdisc with cosmic time to the fact that, at the particular epochs spanned by EDisCS, many spiral galaxies joined the red sequence as they ceased to form stars. They also predicted a decrease of

11The S0 co-added spectrum shows a rather prominent absorption feature in the Hδ region. We refrain from any conclusion from measurements performed on this feature, as the line appears specially narrow. 3.5. Discussion 69

Fdisc at z < 0.4 due to morphological transformation of spiral galaxies into S0 galaxies. This agrees with what is observed in WINGS, where the fraction of disc-dominated galaxies is overall lower than the fraction of ellipticals and S0s and approximately constant with luminosity (see also Valentinuzzi et al. 2011 for analogous considerations). However, we stress here that the aim of the spectroscopic follow-up of MORPHS was the study of the Butcher-Oemler eﬀect and therefore the target selection was biased towards blue and spiral galaxies. Furthermore, we did not restrict the analysis of the spectroscopic MORPHS sample to within 0.54 R , as this sample was already small. For this reason, although × 200 we used the correction factors of Poggianti et al. (1999), the Fdisc reported in Fig. 3.5 should be considered as an upper limit

Studies of the morphology-density relation in samples of galaxy clusters up to z = 1 have found that the overall fraction of early-type galaxies decreases with redshift (Dressler et al. 1997; Postman et al. 2005). However, Dressler et al. (1997) and Postman et al. (2005) conducted their studies on luminosity selected samples, while Holden et al. (2007) found no significant decrease in the early-type fraction in a mass limited sample of cluster galaxies at 0.023 < z < 0.83, suggesting that the evolution detected in magnitude limited samples 10.6 must be significantly contributed by galaxies with stellar mass lower than 10 M , their adopted mass limit. Interestingly, Vulcani et al. (2011a) detected a decrease in FE+S0 with 10.2 redshift in their mass limited sample at M > 10 M and confirming it also with the Holden et al. (2007) mass limit. A corollary to these results is that the increase in the late- type fraction with redshift corresponds to a decrease in the S0 fraction, with the fraction of elliptical galaxies not changing significantly. We note in Fig. 3.5 that the fraction of elliptical galaxies along the red sequence follows similar trends with magnitude in all the three samples considered. Vulcani et al. (2011a) also found that the fraction of elliptical galaxies follows similar trends with stellar mass in the range 10.3 < log(M/M ) < 10.9 in clusters up to z 0.8 This suggests that elliptical and S0 galaxies in clusters follow ∼ different evolutionary paths, with the latter likely originating from the morphological transformation of passive spiral galaxies.

Mei et al. (2009) investigated the morphological fractions along the red sequence of a composite sample consisting of the 8 clusters of the ACS Intermediate Redshift Cluster Survey. They found flat trends with magnitude for both the early and late-type morphological fractions in the inner regions of the clusters (Rcluster < 0.6R200). Unlike this work, they split the red sequence into bins of 1 magnitude each. We repeated the measurement of the morphological fractions using bins of one magnitude and we found that the trends observed in Fig. 3.5 became less pronounced and almost flat. In order to effectively com- 70 Chapter 3. Analysis Method

Table 3.6 Median S´ersicindices and stellar masses, per morphological type, on the red sequence of XMM1229. Ellipticals Bulge-dominated Disc-dominated S´ersicindex (n) 4 2 2.6 1.1 1.9 0.8 log M? 10.5 0.5 10.6 0.3 10.2 0.2 M pare with Mei et al. (2009), we will need to build a composite red sequence sample with all the HCS clusters, as the trends observed in XMM1229 could be peculiar only to this cluster.

3.5.3 Structural Properties

Figure 3.7 (right panel) shows the distribution of the S´ersicindex for red sequence galaxies in XMM1229, grouped according to their morphological type, while the median values of S´ersicindex and stellar mass are summarised in Table 3.6. The uncertainties quoted in this table correspond to the 68% (1σ) width of the distribution of each quantity.12 We do not consider the behaviour of galaxy size in this paper as it is extensively discussed by Delaye et al. (2014) for the entire HCS sample.

In Fig. 3.7 we plot the Sérsicindex n, as a function of the z850 magnitude (left panel) and stellar mass (central panel). From the central panel, it can be seen that there is some correlation between Sérsicindex and stellar mass, as it would be expected by Graham & Guzmán(2003), although we find that it is rather weak (ρ = 0.3). The comparison between the left and central panels of Fig. 3.7 shows that the stellar mass gives a clearer separation between early- and late-type galaxies, suggesting that even those disc-dominated galaxies that appear to be bright in the F850LP band are actually lower-mass objects. However, it is important to stress that the median fractional uncertainty on galaxy stellar mass is 24 %, which is relatively high, and therefore this conclusion should be interpreted as an indication rather than an evidence of the build-up of the red sequence at low masses through quenching of star formation in spiral galaxies. We note that the Sérsicindex of the elliptical galaxies spans a broad range of values. In particular, we note that some ellipticals have n 6. These are mostly bright objects with ∼ z < 22.0, even though we obtain n = 6.6 0.6 for the galaxy XMM1229 353, which is a 850 10.1 relatively faint object (z850 = 23.2, M? = 10 M ). From Fig. 3.6 (bottom-left panel) it can be seen that at the magnitudes of XMM1229 353 ∆n = (nGALF IT ninput)/ninput can − be as large as 0.5, meaning that GALFIT can fit with a n > 4 profile a galaxy that would

12 There is one only irregular galaxy on the red sequence, for which we find n = 0.90 0.09 and M? = 9.8 10 M . 3.5. Discussion 71 be fitted by a de Vaucouleurs profile in an image with less sky noise than our F850LP image. Furthermore, as it is shown in Fig. 3.1 of Häussleret al. (2007), the differences between n = 4 and n > 4 profiles are mainly in the outskirts of the galaxies, where the contamination from the sky is higher. Interestingly, the value of n for the BCG (XMM1229 237) is: n = 6.4 0.3, which is consistent with results from local BCGs (see e.g. Graham et al. 1996, although Gonzalez et al. 2005 found that a double de Vaucouleurs profile produces a better fit). We point out that the BCG resides in a crowded field (see Fig. 3.1) and therefore this result can be driven by contamination from intracluster light and close neighbours rather than from material that is associated to the BCG itself.

3.5.4 The Red Sequence Slope and Scatter

Prior to this work, the red sequence of XMM1229 was studied by Santos et al. (2009) and Meyers et al. (2012), using the same F775W and F850LP images. In this section we describe the comparison between the results of the present and those works. The linear fit to the (i z ) versus z red sequence produces a negative slope 775 − 850 850 b = 0.044 0.017, consistent within the errors with Santos et al. (2009) and Meyers et al. − (2012) (b = 0.039 0.013 and b = 0.028, respectively). Unlike Santos et al. (2009), we − − did not restrict our fit only to the spectroscopically confirmed members, as our sample of photometrically selected cluster members was 1 mag deeper. Delaye et al. (2014) fitted the observed XMM1229 red sequence without any photometric or spectroscopic selection finding b = 0.03, a value closer to the Meyers et al. (2012) result. Santos et al. (2009) and − +0.003 Meyers et al. (2012) found intrinsic scatters σc = 0.039 and σc = 0.066−0.014, respectively, both consistent with our estimate σc = 0.026 0.012 within 1.1σ and 2.2σ, respectively. The shallower slopes found by Meyers et al. (2012) and Delaye et al. (2014) are a consequence of the fact that galaxy colours were estimated within one half-light radius in each galaxy in those works. As shown by Scodeggio (2001) and Bernardi et al. (2003), the steeper slope obtained with fixed apertures is to be attributed to the effect of residual colour gradients, which our large physical fixed aperture was not able to remove completely. However, as pointed out in both works, colours estimated within the galaxy half-light radius are also noisier and can produce large scatters, as demonstrated by the result of Meyers et al. (2012). Therefore, we think that the choice of a large fixed physical aperture, albeit unable to completely remove internal colour gradients, produces less noisy red sequences more suitable for evolutionary studies. In agreement with most works in the recent literature on high redshift clusters, we find 72 Chapter 3. Analysis Method

Figure 3.8 Co-added FORS2 spectra of morphologically classified red sequence galaxies. The spectra are plotted in the observer frame and are shifted along the vertical axis for clarity. From top to bottom: elliptical, bulge-dominated, early disc-dominated galaxies. We highlight the Hδ, Hγ and G4300 features, and mask the O2 atmospheric absorption (A band, λ 7604 A).˚ It can be seen that the Hδ absorption is stronger in early disc- dominated galaxies.∼ This suggests that these galaxies probably just joined the red sequence after cessation of star formation. It can be noted a remarkable [OII] emission feature in the elliptical spectrum, which is contributed by the galaxies XMM1229 145 and XMM1229 73. As reported by Santos et al. (2009), these two spectra also show [OIII] in emission, which may indicate ongoing star formation. 3.6. Summary and Conclusions 73 that the red sequence has a negative slope (Lidman et al. 2008; Mei et al. 2009; Demarco et al. 2010; Lemaux et al. 2012; Snyder et al. 2012). Using the conversions from the observed i and z photometries to rest-frame B and V magnitudes derived in 3.5.1, 775 850 § we find that our result is consistent with the median red sequence slope from Valentinuzzi et al. (2011) b = 0.042 0.007. This supports the notion of little or no evolution of the − slope of the red sequence since z = 1.5 (see e.g. Lidman et al. 2008). However, according to both hydrodynamical and N-body simulations (Romeo et al., 2008) and semianalytic models (Menci et al., 2008), only based on hierarchical merging, the slope of the cluster red sequence should gradually evolve with redshift, flattening at z 0.7 1.0, and then ∼ − turning positive at higher redshifts. The results of the observations clearly indicate that these models were incomplete and other processes need to be taken into account to explain the evolution and the build-up of the red sequence. The simulations of Jiménezet al. (2011), based on a hybrid model constituted by a N-body cosmological simulation and a semi analytic model, produce red sequences with negative slopes for clusters at z 1, in ∼ better agreement with the observations. In these simulations, massive elliptical galaxies at the bright end of the red sequence constitute a primordial population which formed in the clusters and then evolved by accreting other galaxies at later epochs, thus shaping the red sequence as it is observed in the local universe.

3.6 Summary and Conclusions

We have presented a detailed analysis of the properties of red sequence members in the cluster XMMU J1229+0151 (XMM1229), at z 0.98. A study of the X-ray properties, as ∼ well as the analysis of the properties of the spectroscopically confirmed cluster members were presented by Santos et al. (2009). The availability of deep WFC3 images in four IR filters and HAWK-I Ks data for this field allowed us to fit synthetic spectral energy distributions and determine reliable photometric redshifts. We used the latter to estimate cluster membership, extending the analysis of the red sequence to about twice the galaxies considered in Santos et al. (2009), and down to one magnitude fainter than the limit of the FORS2 spectroscopic observations. Our estimates of the red sequence slope and scatter are consistent with the results of Santos et al. (2009), Meyers et al. (2012) and Delaye et al. (2014), even though the latter two authors found a shallower red sequence and a larger scatter due to differences in the adopted strategy for aperture photometry. The luminous-to-faint ratio measured on the red sequence of XMM1229 is higher than that measured in the WINGS composite sample, although the two estimates are consistent within the uncertainties. The luminous-to-faint 74 Chapter 3. Analysis Method ratio in the cluster outskirts is consistent with no deficit of galaxies at the faint end of the red sequence. After splitting into morphological classes, we found that the red sequence is predomi- nantly populated by elliptical and S0 galaxies, whose fractions follow different trends with magnitude. As it is observed at lower redshifts, the bright end of the red sequence appears to be dominated by elliptical galaxies. Therefore, elliptical galaxies have been constituting the dominant class at the bright end of the red sequence since at least z = 1. At intermediate luminosities, elliptical galaxies follow trends with magnitude which are similar to those observed in the MORPHS and WINGS composite samples. The faint end of the red sequence of XMM1229 is characterised by the increase in the fraction of disc-dominated galaxies. At similar absolute magnitudes, the WINGS red sequence is dominated by elliptical and S0 galaxies, suggesting that the latter may be the descendants of z = 1 red spirals. We note that there is significant evolution of the bright end of the red sequence, resulting in a population of galaxies that are not observed at z=1. However, we caution against the fact that the WINGS red sequence extends to up to 2 mag brighter in the V band with respect to XMM1229 and 1.5 mag brighter with respect to MORPHS. We attribute this difference to the effect of colour gradients or biases in the WINGS optical and IR photometry, although the presence of young and/or metal-poor stellar populations in bright red sequence galaxies can result in bluer colours at the bright end of the red sequence, as observed in some of the WINGS clusters.. The method presented in this paper and devised to deal with archival data from different telescopes allows us to maximise the number of candidate cluster members on the red sequence using photometric redshifts or statistical background subtraction. It will be applied to the entire HCS sample to investigate the build-up of the red sequence at 0.8 < z < 1.5 and to study the morphological evolution of its members. 4 The Build-up of the Red Sequence in High Redshift Galaxy Clusters

The present chapter describes the study of the build-up of the red sequence in the clusters of the HCS sample at redshift 0.8 < z < 1.5. The analysis method outlined in Chapter 3 and published in Cerulo et al. (2014) is applied to the entire sample, and the implications from the comparison with the nearby clusters of the spectroscopic WINGS survey are discussed. The chapter is organised as follows: Section 4.1 describes the photometry and the estimation of the membership for each cluster. The properties of the red sequence in the individual clusters are discussed in Section 4.2, while Section 4.3 presents the results of the measurements of the red sequence ﬁtting parameters, luminous-to-faint ratio, and luminosity distributions. In Section 4.4 we discuss our results comparing them with the literature and proposing a possible scenario for the build-up of the red sequence in galaxy clusters.

Throughout this chapter we adopt a ΛCDM cosmology with ΩΛ = 0.73, Ωm = 0.27, and H = 71.0 km s−1 Mpc−1, as done in Chapters 2 and 3. Unless otherwise stated, 0 · · all magnitudes are quoted in the AB system (Oke, 1974).

4.1 Photometry and Cluster Membership

4.1.1 Object Detection and PSF Modelling

Tables 2.1 and 2.2 summarise the global properties and observations of the HCS sample. We followed the procedures described in 3.3.1 to detect objects in each image and model § the PSF. In summary, we used a modiﬁed version of the GALAPAGOS code (H¨aussler et al., 2007) to run SExtractor (Bertin & Arnouts, 1996) in high dynamic range mode. This allowed us to detect the faintest objects in the images with a reliable deblending of

75 76 Chapter 4. The Build-up of the Red Sequence the sources in the cores of the clusters. GALAPAGOS runs SExtractor twice, the first time using a configuration setting optimised for the detection of bright objects (COLD run) and the second time adopting a configuration setting optimised for the detection of faint objects (HOT run). When the two individual runs are completed, the software merges the catalogues rejecting the double detections from the sample. The ACS and WFC3 observations targeted the clusters in more than one band and hence, for these data sets, we could run SExtractor in dual image mode. We performed the detection on the F850LP images for the ACS fields and on the F110W images for the WFC3 fields and used the other images for measurement. Although four of the clusters have HAWK-I data in both the J and Ks bands, we did not run SExtractor in dual image mode for these images because the sizes of the fields are different. The PSF was modelled with PSFEx Version 3.9 (Bertin, 2011) in all the images, and the PSF FWHMs are reported for each band in Table 2.2. We built the multiband photometric catalogues for each cluster by matching the PSF of the single images to the broadest PSF. These catalogues were used to estimate the stellar masses of red sequence galaxies with the lephare code (Arnouts et al., 1999; Ilbert et al., 2006), which are discussed in Chapter 5. Galaxy colours were measured on the images obtained by convolving each image by the PSF of the other image in the filter pair used for the study of the colour-magnitude diagram (cross-convolution, see also 4.1.2 and Figure 4.1). Thus, for example, in the § cluster RX0152 the F775W image was convolved by the PSF of the F625W image, and the F625W image was convolved by the PSF of the F775W image. This allowed us to correct for PSF differences between the images, avoiding the PSF matching with the broadest PSF and the consequent reduction in image quality and depth.

4.1.2 Background Contamination

Unlike XMM1229, we could not estimate cluster membership for all the HCS clusters with the photometric redshifts because some of the merged multiband datasets were not complete down to the faint end of the red sequence. This problem arose especially with the clusters observed in the ISAAC J and Js bands, as is shown in Figure 4.2 in the case of the cluster RCS0220 (z = 1.03). The red diamonds in the ﬁgure are all galaxies with J > 22.4 mag, which is the 90% completeness limit in the ISAAC J band for this cluster ﬁeld. The blue squares are red sequence galaxies with magnitudes brighter than the F775W 90% completeness limit (diagonal dashed line) not detected in the ISAAC image. It is evident from the plot that the red sequence becomes already incomplete at z850 = 22.0 mag in 4.1. Photometry and Cluster Membership 77

1.2 RX0152 (z =0.84) RCS2319 (z =0.91) 1.2

max 0.8 0.8 max F625W F775W F775W F850LP F/F F/F 0.4 F606W F775W F775W F850LP 0.4

0.0 0.0 10000 20000 10000 20000 1.2 XMM1229λ(A)◦ (z =0.98) RCS0220λ(A)◦ (z =1.03) 1.2

max 0.8 0.8 max F775W F850LP F775W F850LP F/F F/F 0.4 F775W F850LP F775W F850LP 0.4

0.0 0.0 10000 20000 10000 20000 1.2 RCS2345λ(A)◦ (z =1.04) XMMU0223λ(A)◦ (z =1.22) 1.2

max 0.8 0.8 max F775W F850LP F775W J (HAWK-I) F/F F/F 0.4 F775W F850LP F775W F125W 0.4

0.0 0.0 10000 20000 10000 20000 1.2 RDCS1252λ(A)◦ (z =1.24) XMMU2235λ(A)◦ (z =1.39) 1.2

max 0.8 0.8 max F775W F125W F850LP F125W F/F F/F 0.4 F775W F125W F850LP F125W 0.4

0.0 0.0 10000 20000 10000 20000 1.2 XMMXCS2215λ(A)◦ (z =1.46) λ(A)◦

max 0.8 F850LP Ks (HAWK-I) F/F 0.4 F850LP Ks (ISAAC)

0.0 10000 20000 λ(A)◦

Figure 4.1 Filter combinations adopted for the study of the red sequence in the HCS clusters. Clusters are ordered by increasing redshift. The solid black line is a template spectral energy distribution of an elliptical galaxy from the library of Coleman et al. (1980). The blue and red solid lines are the blue and red filters adopted for the cluster red sequence, respectively. The blue and red dashed lines are the blue and red filters adopted in the estimation of field contamination in the GOODS-N/S fields. The names of the cluster (top row) and field (bottom row) filters are written in each plot together with the names and redshifts of the clusters. The vertical dashed lines represent the positions of the 4000 A˚ break at the redshifts of the clusters. 78 Chapter 4. The Build-up of the Red Sequence

Table 4.1 Photometric set-up of red sequence and Pfield estimation in each HCS cluster.

Cluster Name redshift (z) Filter Bands (cluster) Filter Bands (ﬁeld)

RX0152 0.84 r625, i775 V606, i775 RCS2319 0.91 i775, z850 i775, z850 XMM1229 0.98 i775, z850 i775, z850 RCS0220 1.03 i775, z850 i775, z850 RCS2345 1.04 i775, z850 i775, z850 XMMU0223 1.22 i775, JHAW K−I i775, F125W RDCS1252 1.24 i775, F125W i775, F125W XMMU2235 1.39 z850, F125W z850, F125W XMMXCS2215 1.46 z850, Ks (HAWK-I) z850, Ks (ISAAC)

the multiband catalogue, and that there are objects as bright as z850 = 22.6 mag not detected in the J band. This incompleteness can affect the study of the evolution of the red sequence by artificially increasing the values of the luminous-to-faint ratio ( 4.3). § The clusters RX0152 (z = 0.84), XMMU0223 (z = 1.22), and XMMXCS2215 (z = 1.46) were observed in only 4 passbands and, despite the high quality and depth of the images, the limited photometric coverage resulted in poor agreement with the spectroscopic redshifts. For instance, for the cluster XMMU0223 we obtained the normalised redshift difference ∆z = (zspec zphot)/(1 + zspec) = 0.08 0.14, where the values are − − the median and half width of the 68% confidence interval of the ∆z distribution. Such a dispersion corresponds to a ∆z distribution three times broader than that found for XMM1229, for which we obtained ∆z = 0.00 0.05 and, consequently, to a significantly more uncertain estimate of the cluster membership with respect to that cluster. Thus we decided to estimate the membership of all the HCS clusters with the statistical background subtraction technique presented in 3.4.2, and in order to consistently study all § the clusters, we re-analysed the XMM1229 red sequence adopting the same membership criteria used for the other clusters and explained in the following paragraphs. We find that the conclusions of Cerulo et al. (2014) regarding the colour and morphological properties of the cluster members remain unchanged.

The ﬁlter pairs for the study of the individual red sequences were chosen so that they bracketed the 4000 A˚ break at the redshift of the clusters. This allowed us to minimise the contamination from galaxies in the blue cloud. We achieved this goal in all the HCS 4.1. Photometry and Cluster Membership 79

1.4

1.2

AB 1.0

0.8

0.6

0.4

( F 775 W − 850 LP ) 0.2

all galaxies within 0.54 R200 from cluster centre) 0.0 · galaxies with JISAAC >22.4 mag red sequence galaxies not detected in JISAAC 19.2 20.0 20.8 21.6 22.4 23.2 24.0 24.8 25.6 F850LPAB(MAGAUTO)

Figure 4.2 Effect of the ISAAC J(Js)-band incompleteness on the cluster multiband datasets. The plot shows the (i775 z850) vs z850 colour-magnitude diagram of the cluster RCS0220 (z = 1.03). Red diamonds− are all the objects with J > 22.4 mag, the 90% completeness limit in this band. The black solid line is the best-fit straight line to the observed red sequence and the dotted diagonal lines represent the red sequence boundaries. The diagonal dashed line is the 90% magnitude completeness limit in the i775 band. As it can be seen, in the multiband sample the red sequence becomes incomplete already at z850 = 23.0 mag, which is brighter than the 90% completeness limit in this band (26.7 mag). This effect can result in artificially high luminous-to-faint ratios (See 4.3.2). § clusters except RCS2319 (z = 0.91), XMM1229 (z = 0.98), RCS0220 (z = 1.03), and RCS2345 (z = 1.04), for which we used the ACS F775W and F850LP filters. To varying degrees, part of the F775W filter lies redward of the 4000 A˚ break. However, while for the last 3 clusters more than half of the F775W transmission curve covers wavelengths λ < 4000 A˚ at the redshifts of the clusters, in the case of RCS2319 more than 50% of the transmission curve of the F775W filter falls at wavelengths λ > 4000 A˚ at the redshift of this cluster. As a result, the red sequence is flatter than in other clusters at similar redshifts (Figure 4.3, Table 4.2). The photometric bands used for the study of the cluster red sequences are summarised in Table 4.1 and plotted in Figure 4.1 as solid blue and red lines.

We built control fields for each cluster using the data taken in the same bands in the GOODS North and/or South fields (hereafter GOODS-N and GOODS-S). This was possible for all the clusters except RX0152, XMMU0223, and XMMXCS2215. For the first cluster we had to resort to the GOODS-N and S ACS F606W (V606) images, which overlap with the spectral region sampled by the F625W band, while we used the Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS, Grogin et al. 2011; Koeke- 80 Chapter 4. The Build-up of the Red Sequence moer et al. 2011) WFC3 F125W images of the GOODS-S field for XMMU0223. We used the ISAAC Ks-band images of the GOODS-S field (Retzlaff et al., 2010) for background subtraction in the XMMXCS2215 field since, to our knowledge, no observations in the HAWK-I Ks band with fields of view of the same order of magnitude of the GOODS fields are available. The filter pairs used for background estimation in each cluster are shown in Table 4.1, and their transmission curves are plotted in Figure 4.1 as dashed blue and red lines.

The GOODS images were downloaded from the dedicated survey repositories1 2, and the source detection and photometry were performed following the same procedure adopted for the cluster ﬁelds.

The PSF of the ISAAC Ks band images varies considerably across the ﬁeld, with a FWHM spanning the range 0.3800 < F W HM < 0.58”. Hence, prior to analyse these images, we matched the PSFs of each image section to that of the image section with the broadest PSF. For this purpose, since the range across which the PSF varies is of the order of the ISAAC pixel scale (0.1500/pixel), we oversampled all the image sections to the ﬁner pixel scale 0.0300/pixel and modelled the PSF with PSFex as done with the other images. Then we convolved each oversampled image by a Gaussian kernel with

q FWHM = FWHM 2 FWHM 2, (4.1) max − i where FWHMmax is the FWHM of the broadest PSF, and FWHMi is the FWHM of the PSF of the image to convolve. In this procedure we assumed that the PSF of the image could be approximated by a Gaussian function. This is not always the case, and a Moffat (1969) function is shown to produce a better fit to the distribution of stellar profiles, particularly in the lower surface brightness tails. Nevertheless, for the purpose of this work, which is the measurement of integrated magnitudes of galaxies within large (100 radius) circular apertures, we expect that the Gaussian approximation does not affect our results.

Following the approach adopted for the analysis of XMM1229, we divided the region of the colour-magnitude diagram where the red sequence is situated in two rectangular cells, respectively corresponding to the regions brighter and fainter than the red sequence magnitude mid point. Then we computed the ﬁeld contamination probability, that is, the

1CANDELS: http://candels.ucolick.org/data_access/Latest_Release.html 2ISAAC ESO/GOODS: http://www.eso.org/sci/activities/garching/projects/goods.html 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 81 probability for a galaxy to belong to the ﬁeld as:

Nfield A Ncluster,cont Pfield = × − (4.2) Ncluster Ncluster,cont − in each colour-magnitude cell, where Nfield is the number of galaxies in each GOODS-N/S colour-magnitude cell, Ncluster and Ncluster,cont are respectively the number of galaxies and the number of spectroscopic interlopers in each colour-magnitude cell in the cluster field, and A is the ratio between the areas of the cluster and GOODS-N/S fields. The right-hand panels of Figure 4.3 show the plots of the colour-magnitude diagrams over regions of the GOODS fields covering the same areas of each cluster central region (i.e. R < 0.54 R ). Also represented in each plot are the best-fit lines and the boundaries × 200 of the cluster red sequences with the bright and faint magnitude limits. A qualitative comparison with the cluster colour-magnitude diagrams plotted in the left-hand panels of Figure 4.3 shows that the field contamination in the cluster central region is globally low and mainly affects the faint end of the red sequence. In particular, we find that the field contamination probability spans the range 0.01 < Pfield < 0.20 (Table 4.2).

4.2 The Cluster Red Sequence at 0.8 < z < 1.5

4.2.1 The Fitting Procedure

We studied the red sequence of candidate cluster members within 0.54 R of the × 200 cluster centre, which allowed us to reduce the contamination by interloper galaxies and at the same time to directly compare our results with XMM1229 and with the low-redshift cluster sample of the WINGS spectroscopic survey. In order to quantify the effect of field contamination on the parameters of the red sequence, namely zero-point, slope and intrinsic scatter, we studied the red sequence in two subsequent steps: (1) we fitted the observed red sequence, and obtained first guesses of the parameters with the uncertainties due to photometric error; (2) we fitted the field-corrected red sequence and evaluated the contribution of the interloping objects on the estimates of the fit parameters. In the first step we fitted a straight line to the observed red sequence by applying a robust line fitting technique based on the Tukey’s bi-square weight function (Press et al., 2002). We considered all the galaxies down to the 90% magnitude completeness limit estimated as in Section 3.3.2 by measuring the fraction of recovered simulated galaxies inserted in random empty regions of the images. The functional form of the red sequence 82 Chapter 4. The Build-up of the Red Sequence

RX0152 (z=0.84) RX0152 (z =0.84) All galaxies with R <0.54 R × 200 Red Sequence members 2.0 Spectroscopic cluster members 2.0

1.5 1.5

1.0 1.0 (F625W-F775W) (F606W-F775W) 0.5 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F775W (MAG AUTO) F775W (MAG AUTO) RCS2319 (z=0.91) RCS2319 (z =0.91) All galaxies with R <0.54 R × 200 Red Sequence members Spectroscopic cluster members 1.5 1.5

1.0 1.0

(F775W-F850LP) 0.5 (F775W-F850LP) 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F850LP (MAG AUTO) F850LP (MAG AUTO) XMM1229 (z=0.98) XMM1229 (z =0.98) 2.0 2.0 All galaxies with R <0.54 R × 200 Red Sequence members Spectroscopic cluster members

1.5 1.5

1.0 1.0

(F775W-F850LP) 0.5 (F775W-F850LP) 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F850LP (MAG AUTO) F850LP (MAG AUTO)

(a) Figure 4.3 Left: Observed colour-magnitude diagrams of the HCS clusters within 0.54 × R200 from the cluster centroid (cluster central region, or cluster centre). Grey points are all the galaxies observed in the cluster centre, green diamonds are the members of the observed red sequence and red squares are the spectroscopically conﬁrmed cluster members. The black dashed line is the best-ﬁt straight line to the observed red sequence and the dotted parallel lines represent the red sequence envelope determined as discussed in 4.2. The diagonal solid lines correspond to the 90% completeness limit while the dot- dashed§ diagonal lines represent the boundaries of the red sequence with the alternative selection ∆C < 3σ . | | 22 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 83

RCS0220 (z=1.03) RCS0220 (z =1.03) All galaxies with R <0.54 R 2.0 × 200 2.0 Red Sequence members Spectroscopic cluster members

1.5 1.5

1.0 1.0 (F775W-F850LP) (F775W-F850LP) 0.5 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F850LP (MAG AUTO) F850LP (MAG AUTO) RCS2345 (z=1.04) RCS2345 (z =1.04) 2.0 All galaxies with R <0.54 R 2.0 × 200 Red Sequence members Spectroscopic cluster members

1.5 1.5

1.0 1.0

(F775W-F850LP) 0.5 (F775W-F850LP) 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F850LP (MAG AUTO) F850LP (MAG AUTO) XMMU0223 (z=1.22) XMMU0223 (z =1.22) All galaxies with R <0.54 R × 200 3.0 Red Sequence members 3.0 Spectroscopic cluster members

2.5 2.5

2.0 2.0

1.5 1.5

1.0 1.0 (F775W - F125W) (F775W - F125W)

0.5 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F125W (MAG AUTO) F125W (MAG AUTO)

(b)

Figure 4.3 Continued. Right: Observed colour-magnitude diagrams in the GOODS-N/S control fields used for field subtraction. Galaxies within a projected spatial region with radius 0.54 R200 are plotted in each figure. The fit and boundaries to the observed cluster red sequences,× and the 90% magnitude completeness limit, are also plotted. The vertical dotted lines represent the apparent magnitude of the brightest red sequence galaxy in each cluster. Galaxies falling within the magnitude and colour ranges of the observed cluster red sequences are plotted as green diamonds. It can be seen that there are few field galaxies with magnitudes and colours in the ranges of the observed cluster red sequences. As a result, the field contamination of the cluster red sequence is low. 84 Chapter 4. The Build-up of the Red Sequence

RDCS1252 (z=1.24) RDCS1252 (z =1.24) All galaxies with R <0.54 R × 200 3.0 Red Sequence members 3.0 Spectroscopic cluster members 2.5 2.5

2.0 2.0

1.5 1.5

1.0 1.0 (F775W - F125W) (F775W - F125W)

0.5 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F125W (MAG AUTO) F125W (MAG AUTO) XMMU2235 (z=1.39) XMMU2235 (z =1.39) All galaxies with R <0.54 R × 200 2.5 Red Sequence members 2.5 Spectroscopic cluster members

2.0 2.0

1.5 1.5

1.0 1.0 (F850LP - F125W) (F850LP - F125W) 0.5 0.5

0.0 0.0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 F125W (MAG AUTO) F125W (MAG AUTO) XMMXCS2215 (z=1.46) XMMXCS2215 (z =1.46) All galaxies with R <0.54 R 4 × 200 4 Red Sequence members Spectroscopic cluster members

3 3 HAWK − I HAWK − I

2 2

(F850LP - Ks 1 (F850LP - Ks 1

0 0 19 20 21 22 23 24 25 26 19 20 21 22 23 24 25 26 Ks(HAWK I) (MAG AUTO) Ks(ISAAC) (MAG AUTO) −

CRS = a + b(m 21.0) (4.3) − where m is the apparent magnitude, a is the zero-point, that is the observer-frame colour at m = 21.0 mag, and b is the slope. CRS is the galaxy colour in the observer-frame system. The uncertainties on the zero-point and slope were defined as half the width of the 68% confidence intervals of the distributions of the two parameters after 1000 bootstrap iterations. The intrinsic scatter of the red sequence σc was estimated, as in Cerulo et al. (2014), by computing the amount of scatter added to the photometric colour error in order 2 to get χe = 1.0. The uncertainty δσc was defined as half the width of the 68% confidence interval of the distribution of σc after 1000 bootstrap iterations. We defined the red sequence as the locus:

κlσc < ∆C < +κhσc, (4.4) − where ∆C = (C CRS) is the difference between the observed (C) and best-fit (CRS) − galaxy colours, σc is the intrinsic scatter of the red sequence, and κl and κh are factors that were estimated by visually inspecting the colour-magnitude diagram as the most suitable to bracket the red sequence (see also 3.4.3 and Cerulo et al. 2014). The choice § of the values of κl and κh, which are shown in Table 4.2, was not based on quantitative considerations on the shape of the colour distribution along the red sequence and, in order to test the effect of this selection against a selection based on the photometric scatter, as that adopted in Delaye et al. (2014), we also applied an alternative selection consisting in assigning to the observed red sequence all the galaxies within 3σ22 from the best-fit straight line. σ22 is, in this case, the colour uncertainty on the red sequence at m = 22.0 mag, which is (except in XMMXCS2215) the typical magnitude of a bright red sequence member. The selection based on the intrinsic scatter is represented by the dotted diagonal lines in Figure 4.3, while the selection based on σ22 is plotted as dot-dashed diagonal lines in the same figure. The effects of this alternative selection on the luminous-to-faint ratio and luminosity distributions will be discussed in Section 4.4. The observed colour-magnitude diagrams in the central regions of the HCS clusters are plotted in the left-hand panels of Figure 4.3, where the members of the observed red sequences are highlighted by green diamonds, and the spectroscopically confirmed cluster members are represented by red squares. Clusters are ordered by increasing redshift. We note that the photometric errors in some cluster red sequences become considerably large at m > 23.0 mag, and when this effect is particularly severe (e.g. RCS2319, middle- 86 Chapter 4. The Build-up of the Red Sequence left panel in Figure 4.3a), we exclude those galaxies from the fit to the red sequence. In the second step we followed the method outlined by Valentinuzzi et al. (2011) to statistically estimate the cluster membership of red sequence galaxies. We ran 200 Monte

Carlo simulations in which, at each iteration, a random number 0 < Psim < 1 was assigned to each galaxy. This number was compared with the field contamination probability Pfield and all the galaxies with Psim > Pfield were retained as cluster members. We fitted a straight line to the selected cluster members using the same procedure adopted in the fit to the observed red sequence, and we estimated the zero-point, slope and intrinsic scatter of the contamination-free red sequence. The median and half of the 68% confidence interval of the distributions of the red sequence parameters after 200 iterations were respectively taken as the estimate and uncertainty of each quantity. Figure 4.4 shows these distributions in the case of the cluster RCS0220. The results for all the other clusters are summarised in Table 4.2 where, for each cluster, we show the fit parameters of the observed and field-corrected red sequences (top and bottom row in each entry, respectively). It can be seen (bottom-right panel in Figure 4.4) that the values of the slope and zero-point of the red sequence, obtained with our fitting procedure, are highly correlated and that steeper slopes correspond to redder zero-points. The parameters of the fit to the observed and field-corrected red sequences are all consistent to within the uncertainties, and we note that the uncertainty due to field contamination is a factor of 0.2 lower than the uncertainty due to photometric errors. In ∼ the following analysis, we will consider the zero-point, slope, and scatter estimated from the fit to the field-corrected red sequence with the uncertainties defined as the sum in quadrature of the bootstrap and Monte Carlo errors. This will allow us to simultaneously take into account the contributions of photometric errors and field contamination in each cluster sample. We note that the field contamination uncertainties on the parameters of the red sequence in the cluster XMMXCS2215 are 0. This apparently absurd result is a consequence of the low Pfield in this cluster. In fact, Pfield 0.01 translates into a fraction of field ∼ interlopers < 0.5 on the observed red sequence. In practice more than 90% of the values of the red sequence parameters after the 200 Monte Carlo simulations are contained in a single peak corresponding to the observed and not field-corrected red sequence. 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 87

Figure 4.4 Distribution of the red sequence parameters after 200 Monte Carlo simulations for the statistical estimate of cluster membership. The vertical solid line is the median value of each parameter while the vertical dashed lines represent the boundaries of the 68% confidence intervals of the distributions. It can be seen that the red sequence slope and zero-point are highly correlated: steeper slopes correspond to redder zero-points (bottom- right panel). 88 Chapter 4. The Build-up of the Red Sequence σ field P 0.11,0.12 0.07,0.16 0.09,0.09 0.03,0.12 0.10,0.17 0.09,0.22 0.09,0.22 0.02,0.12 0.012,0.009 6,9 3,3 3,3 4,4 2,4 6,6 3,4 5,7 10,8 min,max κ c 0.009 0.007 0.002 0.012 0.001 0.009 0.004 0.02 0.003 0.02 0.016 0.003 0.019 0.009 0.04 0.00 δσ 0.0015 0.004 §4.2.1. c σ 0.24 0.02 0.12 0.24 0.12 0.034 0.059 0.042 0.033 0.041 0.032 0.025 0.025 0.116 0.097 0.086 0.109 0.018 δb) 0.00 0.012 0.004 0.003 0.015 0.005 0.016 0.004 0.019 0.003 0.04 0.009 0.03 0.004 0.03 0.009 0.15 0.013 −0.10 −0.01 −0.05 −0.01 −0.10 0.001 −0.033 −0.034 −0.012 −0.030 −0.052 −0.031 −0.032 −0.010 −0.048 −0.028 −0.076 −0.036 δa) slope (b a 0.00 0.02 0.007 0.02 0.006 0.03 0.005 0.03 0.008 0.03 0.003 0.05 0.004 0.04 0.006 0.06 0.01 0.04 2.67 1.23 0.75 0.93 0.98 0.94 2.11 2.10 1.46 1.47 2.67 factors adopted in the selection of the red sequence and the values of the probability 1.234 0.765 0.942 0.989 0.967 2.113 2.114 h κ ) zero-point ( and l κ 1.24 1.39 1.46 0.84 0.91 0.98 1.03 1.04 1.22 Name redshift (z CS2215 CS2345 CS2319 CS0220 RX0152 R R R XMM1229 RDCS1252 XMMU2235 XMMU0223 Cluster XMMX used in the estimation of field contamination. The uncertainties are discussed in field able 4.2 Parameters of the fits to the observed (top row) and field-corrected (bottom row) red sequences with their respective 1 P T uncertainties. Also shown are the 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 89

4.2.2 The Individual Red Sequences

We now discuss the properties of each HCS cluster and of its red sequence. XMM1229 (z = 0.98) is extensively discussed in Chapter 3 and Cerulo et al. (2014), and therefore it is not repeated here. We note that the HCS sample spans a relatively narrow range of cosmic ages (τ 2.5 Gyr in our cosmology) which is shorter than the average timescales ∼ of star formation quenching in galaxy clusters predicted by simulations (3.0-3.5 Gyr, see Taranu et al. 2014). Therefore, we do not expect a significant build-up of the red sequence over this redshift range, although this does not exclude that some galaxies may experience processes which result in shorter quenching timescales (e.g. ram-pressure stripping) and in transitions from the blue cloud to the red sequence in less than 2.5 Gyr. The global properties of the HCS clusters are summarised in Table 2.1 where the values of the dark matter halo masses quoted for each cluster are taken from Jee et al. (2011). The cluster masses were estimated by performing a weak lensing analysis on the same ACS images used here and were defined as the total mass enclosed within 1 R from the cluster × 200 centroid. The red sequence parameters quoted in this section correspond to the field- corrected values in the observer frame with the uncertainties estimated as the sum in quadrature of the bootstrap and Monte Carlo errors (see Section 4.2.1).

RX J0152.7-1357 (RXJ0152, z = 0.84)

RX J0152.7-1357 (RXJ0152, Figure 4.5), at z = 0.84, is a system consisting of three merging galaxy clusters detected as clumps in the ROSAT Deep Cluster Survey (RDCS, +0.7 14 Rosati et al. 1998). The cluster has a dark matter halo mass MDM = 4.4 10 M −0.5 × as estimated from the weak lensing analysis of Jee et al. (2011). We note that Delaye +1.8 14 et al. (2014) report the value M = 7.3 10 M for the total mass of this cluster 200 −1.7 × estimated from its X-ray luminosity within 1 R from the cluster centre, which is × 200 higher, although still consistent to within the uncertainties, with the Jee et al. (2011) estimate. We will discuss the effect of this difference on the red sequence properties in 4.4.1 Detailed studies of the morphological and physical properties of the members of § RX0152 are presented in Demarco et al. (2005, 2010) and Nantais et al. (2013a,b). In the present work we use the HST/ACS observations in the F625W, F775W and F850LP bands, the ESO/VLT Ks HAWK-I observations, and all the redshifts available in the cluster field. The F625W and F775W bands were chosen to study the colour-magnitude diagram, and the fit to the red sequence produces the following result: zero-point a = 1.23 0.02, slope b = 0.034 0.013, intrinsic scatter σc = 0.018 0.009. The colour-magnitude diagram − is shown in Figure 4.3a (top-left panel). 90 Chapter 4. The Build-up of the Red Sequence

Figure 4.5 Colour image of the cluster RX J0152.7-1357 (z = 0.84) obtained by combining the ACS F625W, F775W and F850LP images. In this and all the other colour images of the HCS clusters, the North is at the top and the East at the left of each ﬁgure. The scales of this and the other HCS colour images refer to the physical projected distance, at the redshift of the cluster, corresponding to 0.50.

Interestingly, as noted by Lidman et al. (2013), the brightest cluster galaxy, a massive elliptical with i = 20.80 0.06 mag (log(M∗/M ) = 11.36 0.03), is located 268 775 kpc to the north-east of the centre of the northern clump. The centres of the two main sub-clusters (northern and southern clump) are dominated by two systems consisting of close pairs of elliptical galaxies. Given the complexity of the structure of RX0152, we only consider the region within 0.54 R from the centre of the northern clump which × 200 approximately corresponds to the centres of the ACS images.

RCS 2319.8+0038 (RCS2319, z = 0.91)

RCS 2319.8+0038 (RCS2319, Figure 4.6), at z = 0.91, is part of the RCS2319+00 supercluster, composed of three interacting galaxy clusters, of which RCS 2319.8+0038 was the ﬁrst discovered in the Red Sequence Cluster Survey (Gladders & Yee, 2005). In addition to RCS 2319.8+0038 the RCS2319+00 supercluster also comprises the two smaller clusters RCS 232003+0033.5 and RCS 231946+0030.6 which are not considered in the present study and for which we refer to Faloon et al. (2013) for a detailed analysis. The cluster +2.3 14 has a dark matter halo mass MDM = 5.8 10 M and is one of the most massive −1.6 × 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 91

Figure 4.6 Colour image of the cluster RCS 2319.8+0038 (z = 0.91) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image. clusters in the HCS sample. For the study of the red sequence of RCS2319 we used the HST/ACS observations in the F775W and F850LP bands, the VLT/ISAAC Js images and the HAWK-I Ks data. We also used all the available redshifts including those obtained from our observations with LRIS at the Keck I telescope (Chapter 2). Since at z = 0.91 more than 50% of the transmission curve of the F775W filter lies redward of the 4000 A˚ break (Figure 4.1), the red sequence is flatter than in the other clusters (Figure 4.3a middle-left panel). The linear fit to the red sequence yields the zero point a = 0.77 0.02 and the slope b = 0.012 0.013. The intrinsic scatter is σc = 0.042 0.008. We note − that the central and brightest galaxy of the cluster (RCS2319 205) is 1 mag brighter compared to the rest of the red sequence (z = 19.91 0.09 mag) and also massive 850 (log(M∗/M ) = 11.5 1.5). This galaxy has a smaller elliptical companion, RCS2319 219 (z = 21.13 0.08, log(M∗/M ) = 11.20 0.03), located 10.5 kpc to the NW. 850

RCS 0220.9-0333 (RCS0220, z = 1.03)

RCS 0220.9-0333 (RCS0220, Figure 4.7), at z = 1.03, is an optically selected cluster with a +1.8 14 dark matter halo mass MDM = 4.8 10 M . Unlike RCS2319 and the northern and −1.3 × southern clumps of RX0152, this cluster does not show a dominant bright elliptical galaxy at its centre. However, the cluster centre is eclipsed by a nearby face-on spiral galaxy which prevents galaxies behind it, including any possible giant elliptical situated at the cluster 92 Chapter 4. The Build-up of the Red Sequence

Figure 4.7 Colour image of the cluster RCS 0220.9-0333 (z = 1.03) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image. The centre of the cluster is eclipsed by a nearby face-on spiral galaxy which prevents the location of a BCG of this system from being identified. centre, to be observed (see also Lidman et al. 2013 for a discussion on the properties of the BCG)3. For the study of the red sequence we used the HST/ACS F775W and F850LP data, the VLT/ISAAC J-band data and the HAWK-I Ks-band images, as well as all the available spectra in this field including those obtained from our GMOS-N observations (Chapter 2). The F775W and F850LP bands, which at z = 1.03 bracket the 4000 A˚ break, were used to study the cluster colour-magnitude diagram (Figure 4.3b, top-left panel) and the linear fit to the red sequence yields the zero-point and slope a = 0.99 0.03 and b = 0.032 0.016. The intrinsic scatter is σc = 0.025 0.010. −

RCS 2345-3633 (RCS2345, z = 1.04)

RCS 2345-3633 (RCS2345, Figure 4.8), at z = 1.04, is an optically selected cluster which was discovered in the Red Sequence Cluster Survey (Gladders & Yee, 2005). With a dark +1.1 14 matter halo mass MDM = 2.4 10 M , it is the least massive cluster in the HCS −0.7 × sample and it is characterised by a sparse morphology with no signiﬁcantly dominant galaxy in its core. Interestingly, the brightest cluster galaxy, as identiﬁed by Lidman

3Figure 4.7 shows that a red galaxy can be seen behind the foreground spiral between the two upper spiral arms. However, we do not have spectral information on this object. 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 93

Figure 4.8 Colour image of the cluster RCS 2345-3633 (z = 1.04) obtained by combining the ACS F775W and F850LP images, and the HAWK-I Ks band image. et al. (2013), is a high inclination disc-dominated galaxy; however, since it falls beyond 0.54 R from the cluster centre, it is not considered in the present work and we refer to × 200 Lidman et al. (2013) for a description of its properties. In order to study the red sequence in RCS2345, we used the HST/ACS F775W and F850LP band images, the VLT/ISAAC J band image, and the HAWK-I Ks data, as well as all the redshifts available in the ﬁeld. The F775W and F850LP bands, which at z = 1.04 bracket the 4000 A˚ break, were used to study the colour-magnitude diagram. The linear ﬁt to the red sequence yields the zero- point a = 0.98 0.03 and the slope b = 0.048 0.019. The intrinsic scatter of the red − sequence is σc = 0.03 0.02. The colour-magnitude diagram of this cluster is plotted in the middle-left panel of Figure 4.3b

XMM J0223-0436 (XMMU0223, z = 1.22)

XMM J0223-0436 (XMMU0223, Figure 4.9), at z = 1.22, is a massive X-ray detected +2.5 14 cluster which, with MDM = 7.4 10 M , is the most massive system in the HCS −1.8 × sample. The cluster was discovered as an overdensity in the XMM-Newton Large-scale Structure Survey (XMM-LSS, Pierre et al. 2004; see Andreon et al. 2005 and Bremer et al. 2006 for this particular cluster) and it was followed up with other telescopes and instruments at optical and infrared wavelengths. The cluster was also targeted in the 94 Chapter 4. The Build-up of the Red Sequence

Figure 4.9 Colour image of the cluster XMM J0223-0436 (z = 1.22) obtained by combining the ACS F775W and F850LP images, and the HAWK-I J band image. spectroscopic follow-up of the HST Cluster Supernova Survey and, since it falls in the field of the VIMOS4 Public Extragalactic Redshift Survey (VIPERS, Garilli et al. 2014; Guzzo et al. 2014), additional redshifts were available from that database. In the present work we use the HST/ACS F775W and F850LP images, and the HAWK-I J- and Ks- band data, together with all the redshifts available in the cluster field. The F775W and J bands, which at z = 1.22 bracket the 4000 A˚ break, were used to study the cluster colour-magnitude diagram which is plotted in Figure 4.3b (bottom-left panel). XMMU0223 is characterised by a prominent red sequence with three dominant galaxies (J 20.2 mag) about 1 mag brighter than the rest of the red sequence population. Two of ∼ these galaxies, namely XMMU0223 228 and XMMU0223 252, are situated near the cluster centroid, while XMMU0223 308 is an elliptical galaxy offset 5800 (485 kpc) to the NW ∼ of the cluster centre. It appears to be at the centre of a group of 4 galaxies one of which, i.e. XMMU0223 312, is a spectroscopically confirmed elliptical with z = 21.9 0.2 mag. 850 XMMU0223 228, which was also identified and studied as the cluster BCG in Lidman et al. (2013), is a nearly edge-on disc-dominated galaxy with two fainter neighbouring galaxies at 1-1.500 ( 10 kpc) to the south. XMMU0223 252 is an elliptical galaxy with a ∼ close neighbouring S0 galaxy, XMMU0223 260, situated 10 kpc to the south-west. ∼

4The ESO/VLT Visible MultiObject Spectrograph (Le F`evreet al., 2003) 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 95

The linear ﬁt to the red sequence yields zero-point and slope a = 2.11 0.05 and b = 0.03 0.04, and intrinsic scatter σc = 0.12 0.02. −

RDCS J1252.9-2927 (RDCS1252, z = 1.24)

Figure 4.10 Colour image of the cluster RDCS J1252.9-2927 (z = 1.24) obtained by combining the WFC3 F105W, F125W and F160W images.

RDCS J1252.9-2927 (RDCS1252, Figure 4.10), at z = 1.24, is a massive cluster +1.2 14 (MDM = 6.8 10 M ) detected in the ROSAT Deep Cluster Survey and already −1.0 × studied by several authors in the past (Rosati et al. 2004; Lidman et al. 2004; Demarco et al. 2007; Mei et al. 2009; Nantais et al. 2013b; Delaye et al. 2014). In the present work we use the HST/ACS F775W and F850LP images, the HST/WFC3 F105W, F110W, F125W and F160W images, the ISAAC J- and Ks-band images, and all the redshifts available in the ﬁeld. The F775W and F125W (J) bands, which at z = 1.24 bracket the 4000 A˚ break, were used to study the colour-magnitude diagram. The linear ﬁt to the red sequence yields zero-point and slope a = 2.11 0.04 and b = 0.08 0.03, and intrinsic scatter − σc = 0.097 0.017. The colour-magnitude diagram of RDCS1252 is shown in Figure 4.3c (top-left panel) RDCS1252 is characterised by a prominent red sequence with the bright end dominated by a group of three elliptical galaxies 0.5 1.0 mag brighter than the rest of the red sequence − population. The two brightest galaxies (RDCS1252 859 and RDCS1252 856) form a close 96 Chapter 4. The Build-up of the Red Sequence pair of massive elliptical galaxies (log(M∗/M ) 11.2) near the cluster centroid separated ∼ 00 by a projected distance dp = 1.7 (14 kpc). The third brightest galaxy is an elliptical 00 located at a projected distance dp = 40 (335 kpc) to the south-west of the cluster centroid and forming a pair with the elliptical galaxy RDCS1252 1173 (log(M∗/M ) = 10.54 0.03). The value of the total mass of RDCS1252, estimated from its X-ray luminosity, and +1.1 14 quoted in Table 2 of Delaye et al. (2014), is M = 4.4 10 M , which is lower, 200 −1.0 × although still consistent to within the errors, with respect to the estimate of Jee et al. (2011) reported at the beginning of this section.

XMMU J2235.3-2557 (XMMU2235, z = 1.39)

Figure 4.11 Colour image of the cluster XMMU J2235.3-2557 (z = 1.39) obtained by combining the WFC3 F105W, F125W and F160W images.

XMMU J2235.3-2557 (XMMU2235, Figure 4.11), at z = 1.39, is a massive galaxy +1.7 14 cluster (MDM = 7.3 10 M ) serendipitously discovered in XMM-Newton obser- −1.4 × vations of the galaxy NGC 7314 (Mullis et al., 2005). For the study of this cluster we used the HST/ACS F775W and F850LP data, the HST/WFC3 F105W, F110W, F125W and F160W images, and the HAWK-I J- and Ks-band data. We also used all the redshifts available in the ﬁeld. The F850LP and F125W bands, which at z = 1.39 bracket the 4000 A˚ break, were used to study the colour-magnitude diagram of the cluster. The 4.2. The Cluster Red Sequence at 0.8 < z < 1.5 97 cluster exhibits a prominent red sequence with a bright central galaxy which, similarly to RCS2319, RDCS1252 and XMMU0223, is 1 mag brighter than the rest of the red ∼ sequence population. This galaxy is a giant elliptical situated near the cluster centroid with log(M∗/M ) = 11.05 0.05 and with two faint neighbouring galaxies at projected distances 10-13 kpc from its centre. The linear ﬁt to the red sequence yields zero-point ∼ and slope a = 1.47 0.06 and b = 0.04 0.03 and intrinsic scatter σc = 0.11 0.02. The − colour-magnitude diagram of XMMU2235 is shown in Figure 4.3c (middle-left panel)

XMMXCS J2215-1738 (XMMXCS2215, z = 1.46)

Figure 4.12 Colour image of the cluster XMMXCS J2215-1738 (z = 1.46) obtained by combining the ACS F775W and F850LP images, and the HAWK-I J band image.

XMMXCS J2215-1738 (XMMXCS2215, Figure 4.12), at z = 1.46, is the most distant +3.0 cluster in the HCS sample and it is also one of the least massive (MDM = 4.3 −1.7 × 14 10 M ). The cluster was discovered as an extended source in the XMM-Newton Cluster Survey (XCS, Sahlénet al. 2009) and it was later observed with different telescopes and instruments at optical and infrared wavelengths (HST, Spitzer, VLT). For the present work we use the HST/ACS F775W and F850LP data, and the HAWK-I J- and Ks-band images, together with all the redshifts available in the field. The F850LP and Ks bands were used to study the colour-magnitude diagram and the red sequence because they bracket the 4000 A˚ break at z = 1.46. The linear fit to the red sequence of this cluster yields the zero-point and slope a = 2.67 0.05 and b = 0.10 0.14, while the intrinsic scatter is − 98 Chapter 4. The Build-up of the Red Sequence

σc = 0.24 0.04. The colour-magnitude diagram of XMMXCS2215 is shown in Figure 4.3c (bottom-left panel). The core of XMMXCS2215 also hosts a population of star-forming galaxies (Hilton et al., 2010) and the cluster lacks a dominant luminous galaxy at its centre (Lidman et al., 2013). The value of the halo mass estimated from the cluster X-ray luminosity, quoted in +0.5 14 Table 2 of Delaye et al. (2014), is M = 2.0 10 M , which is lower, although still 200 −0.6 × consistent to within the uncertainties, than the weak lensing estimate of Jee et al. (2011). As seen in Figure 4.3c, the red sequence is sparser than in the other clusters, suggesting that XMMXCS2215 is a young system.

4.3 Red Sequence Properties

The aim of this chapter is the study of the build-up of the red sequence in galaxy clusters at 0.8 < z < 1.5. We address the problem from two diﬀerent points of view, namely the study of the evolution of the red sequence zero-point, slope and scatter, and the analysis of the build-up of the red sequence as a function of galaxy luminosity. While the ﬁrst aspect is discussed in Section 4.3.1, the second is developed in Sections 4.3.2 and 4.3.3.

4.3.1 The Red Sequence Parameters

In order to study the rest-frame red sequence of the HCS clusters and compare with results in the literature, we converted the observer-frame photometries to rest-frame Vega U, B and V bands. For this purpose, we followed the approach discussed in Appendix B of Mei et al. (2009). We used a set of synthetic stellar population models from the Bruzual & Charlot (2003) spectral library spanning the range in formation redshift 2.0 < zf < 5.0, with three metallicity values (i.e.: 0.4Z , Z , 2.5Z ), two laws of star formation history (instantaneous burst and exponentially decaying with e-folding time τ = 1Gyr), and Salpeter (1955) IMF. For each of the 43 generated models we extracted the observed and rest-frame colours at the redshift of each cluster and ﬁtted the linear relation

Crf = A + B Cobs (4.5) ∗ for the conversion from observed to rest-frame system. Crf is the rest-frame colour which can be (B V ), (U V ), or (U B), while Cobs is the colour in the observer-frame system − − − adopted for the study of the red sequence in each cluster (Table 4.1). We also extracted the observed and rest frame magnitudes at the redshift of each cluster for each model and 4.3. Red Sequence Properties 99 used the equation:

Mrf = mobs + α + β Cobs (4.6) ∗ for the conversion to rest-frame magnitude. Mrf is the rest-frame absolute magnitude at the redshift of the cluster, which can be either B or V , and mobs and Cobs are respectively the apparent magnitude and colour used in the study of each individual red sequence and shown in Table 4.1. From Equation 4.5, it follows that the slope and intrinsic scatter of the rest-frame red sequence can be approximated by the relations:

(Slope)rf = B (Slope)obs (4.7) ∗ σrf = B σobs (4.8) ∗ where (Slope)rf and σrf are respectively the rest-frame red sequence slope and intrinsic scatter, and (Slope)obs and σobs are respectively the observed red sequence slope and intrinsic scatter.

We estimated the colours at VV ega = 20.5 mag and BV ega = 21.4 mag to compare − − the evolution of the red sequence zero-point with literature results. The reason for this particular combination of choices is that these are the magnitudes at which the zero-points were measured in the works of Romeo et al. (2008) and Mei et al. (2009) which are shown in Figure 4.13. Valentinuzzi et al. (2011) estimated the red sequence zero-point in the rest-frame (B V ) vs V red sequences of the WINGS clusters at VV ega = 0.0 mag. To − make the comparison easier and consider the evolution of the zero-point at a magnitude more typical of cluster red sequence galaxies, we evaluated the (B V ) red sequence colour − 5 of the WINGS clusters at VV ega = 20.5 mag . − Figure 4.13 shows the redshift evolution of the zero-point, slope and scatter of the

(B V )V ega vs VV ega (left column), (U V )V ega vs VV ega (middle column), and (U B)V ega − − − vs BV ega (right column) red sequences. The results of the present work are plotted as black ﬁlled circles with error bars which correspond to the 1σ total uncertainty on the observed red sequence parameters propagated according to the equations:

r h 2 2 2 i δCZP = B δ(ZP ) + (mZP 21.0) δ(Slope) (4.9) ∗ obs | − | obs δ(Slope)rf = B δ(Slope)obs (4.10) ∗ δσrf = B δσobs (4.11) ∗ 5The parameters of the ﬁt to the WINGS red sequences were kindly provided by B.M. Poggianti (private communication). 100 Chapter 4. The Build-up of the Red Sequence

(B V) vs V (U V) vs V (U B) vs B − − 0.50 − 0.95 1.4 0.45

0.90 0.40 1.2 0.85 0.35

0.80 0.30 1.0 0.75 0.25

zero-point 0.20 0.70 0.8 0.15 0.65 0.6 0.10 0.60 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6

0.10 0.10 0.10

0.05 0.05 0.05

0.00 0.00 0.00

slope 0.05 0.05 0.05

0.10 0.10 0.10

0.15 0.15 0.15

0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6

0.25 0.25 0.25

0.20 0.20 0.20

0.15 0.15 0.15 c σ 0.10 0.10 0.10

0.05 0.05 0.05

0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8z 1.0 1.2 1.4 1.6 This work Meyers et al. (2012) Romeo et al. (2008) Valentinuzzi et al. (2011) Jaffe et al. (2011) Mei et al. (2009) Snyder et al. (2012) Ellis et al. (1997) Menci et al. (2008)

Figure 4.13 Evolution of the parameters of the rest-frame red sequence and comparison with the literature. From left to right: (B V ) vs V (left column), (U V ) vs V (middle column), (U B) vs B (right column). From− top to bottom: (B V ) and− (U V ) colours at V = 20.−5 mag, and (U B) colour at B = 21.4 mag (top− row), rest-frame− slope (middle− row), rest-frame intrinsic− scatter (bottom− row). The hydrodynamic simulations of Romeo et al. (2008) (cyan octagons) predict a strong evolution of the slope of the red sequence which becomes positive at z > 1. This is not in agreement with the observational results of the present work and other works on clusters at z > 0.8 ( e.g. Meyers et al. 2012 and Mei et al. 2009). The intrinsic scatter exhibits a wide range in (U V ) colour at z > 0.8 (0.01 < (U V ) < 0.25). The semi-analytic models of Menci et al.− (2008) predict a large intrinsic scatter,− although still consistent with the results of the present work and Mei et al. (2009). The (U B) colours at B = 21.4 mag in Mei et al. (2009) are systematically redder ( 0.1− mag) than those estimated− in the HCS (top-right panel). All magnitudes are in the Vega∼ system. 4.3. Red Sequence Properties 101 where δ(ZP )obs, δ(Slope)obs and δσobs are the uncertainties on the observed red sequence zero-point, slope and intrinsic scatter, respectively, and mZP is the apparent magnitude in the observer-frame system corresponding to VV ega = 20.5 mag or BV ega = 21.4 mag. − − The parameters of the rest-frame red sequences are summarised in Table 4.3 and their evolution will be discussed in 4.4. §

4.3.2 The Luminous-to-Faint Ratio

We analyse the build-up of the red sequence as a function of galaxy luminosity by adopting two complementary approaches, namely the study of the evolution of the ratio between luminous and faint galaxies (luminous-to-faint ratio, L/F ), and the study of the luminosity distribution of galaxies along the red sequence. While the latter will be considered in Section 4.3.3, we focus here on the measurement of L/F . We have selected from the WINGS spectroscopic data set all those clusters with total 14 masses MDM 5 10 M or velocity dispersions σ > 670 km/s. This mass cut assured ≥ × us that only the clusters with total masses greater than, or equal to, the mass predicted in cosmological simulations of structure formation (e.g. Fakhouri et al. 2010; Chiang et al. 2013) for the descendant of the least massive HCS cluster (i.e. RCS2345) were considered in the analysis. This selection produced a sub-sample of 29 galaxy clusters. In order to compare between different HCS clusters, and between HCS and WINGS, we converted our observer-frame photometries to rest-frame VAB absolute magnitudes as discussed in 4.3.1 using Equations 4.4 and 4.5 and we passively evolved the rest-frame § magnitudes to z = 0. We considered all red sequence galaxies down to VAB = 19.5 mag, − which is the 90% VAB completeness limit of XMMXCS2215, the shallowest sample in our data set. We defined as luminous all the galaxies with VAB < 20.5 mag and as faint all − those with 20.5 VAB < 19.5; this subdivision corresponds to the mid point of the − ≤ − red sequence magnitude range in XMMXCS2215. The same ranges were adopted in the WINGS sample and we defined the luminous-to-faint ratio as the ratio L/F between the numbers of luminous and faint galaxies. Figure 4.15 shows that this cut results in the loss of part of the faint galaxy population in the deepest data sets, such as RDCS1252 and RX0152; furthermore, it does not allow a direct comparison with most literature results, which were obtained adopting the ranges VV ega < 20.0 mag and 20.0 < VV ega < 18.2 − − − for luminous and faint galaxies, respectively (see e.g.: De Lucia et al. 2007b; Andreon 2008; Capozzi et al. 2010; Valentinuzzi et al. 2011). The numbers of luminous and faint galaxies were corrected for field contamination as discussed in Section 4.3.3. The estimates of L/F are summarised in Table 4.4, while Figure 4.14 shows the trends 102 Chapter 4. The Build-up of the Red Sequence

Table 4.3 Parameters of the ﬁts to the rest-frame red sequence with their respective 1σ uncertainties. In each entry: (B V ) vs V (top row), (U V ) vs V (middle row), (U B) vs B (bottom row). Magnitudes− and colours are in the Vega− system. −

Cluster Name redshift (z) zero-point (a δa) slope (b δb) σc δσc

RX0152 0.84 0.83 0.03 0.026 0.010 0.014 0.007 1.05 0.08 − 0.06 0.02 0.034 0.017 0.28 0.03 −0.036 0.013 0.019 0.010 −

RCS2319 0.91 0.76 0.03 0.009 0.010 0.034 0.006 0.89 0.07 − 0.02 0.02 0.078 0.014 0.15 0.03 −0.012 0.013 0.044 0.008 −

XMM1229 0.98 0.75 0.04 0.04 0.012 0.032 0.009 0.88 0.08 − 0.09 0.03 0.07 0.02 0.21 0.03 −0.051 0.016 0.041 0.012 −

RCS0220 1.03 0.78 0.04 0.026 0.013 0.020 0.008 0.95 0.09 − 0.06 0.03 0.047 0.019 0.22 0.03 −0.033 0.016 0.026 0.010 −

RCS2345 1.04 0.73 0.05 0.038 0.015 0.020 0.017 0.81 0.11 − 0.09 0.03 0.05 0.04 0.17 0.04 −0.049 0.019 0.026 0.02 −

XMMU0223 1.22 0.77 0.04 0.010 0.014 0.044 0.009 0.94 0.08 − 0.02 0.03 0.096 0.019 0.18 0.02 −0.012 0.017 0.052 0.010 −

RDCS1252 1.24 0.74 0.03 0.029 0.010 0.037 0.006 0.86 0.06 − 0.06 0.02 0.081 0.014 0.179 0.019 −0.034 0.012 0.043 0.007 −

XMMU2235 1.39 0.75 0.05 0.019 0.016 0.059 0.012 0.89 0.11 − 0.04 0.03 0.12 0.02 0.18 0.04 −0.021 0.017 0.064 0.013 −

XMMXCS2215 1.46 0.79 0.09 0.03 0.05 0.079 0.012 0.99 0.19 −0.07 0.10 0.16 0.02 0.255 0.017 −0.03 0.05 0.083 0.013 − 4.3. Red Sequence Properties 103 of the luminous-to-faint ratio with cluster redshift (left-hand panels) and total mass (right- hand panels). The error bars reported in the plots correspond to the 68% Poissonian conﬁdence intervals which were estimated adopting the approximations of Ebeling (2003). The error bars on the halo mass of the HCS clusters are taken from Table 2 of Jee et al. (2011). The value of the mass reported for WINGS is the median halo mass of the sample and the error bars correspond to the 68% width of the distribution. The bottom panels of Figure 4.14 show the plots of L/F as a function of redshift and cluster halo mass for the alternative selection of galaxies on the red sequence with ∆C < 3σ . These results | | 22 will be discussed in 4.4.2. §

4.3.3 Red Sequence Luminosity Distribution

We divided the red sequence of each cluster into 0.5 mag bins in the photometric band used for the colour-magnitude diagram and we used the GOODS control ﬁelds to estimate

Pfield in each bin as deﬁned in Equation 4.1. We converted to passively evolved z = 0 absolute VAB magnitudes and estimated the red sequence number counts as:

m X Nbin = Ngal,i (1 Pfield,i) (4.12) × − i=1

th where Ngal,i is the i galaxy in the bin and Pfield,i is the expected fraction of field interlopers associated with the galaxy; m is the total number of galaxies in each magnitude bin. The number counts in the individual HCS clusters are plotted in Figure 4.15 while in Figure 4.16 are plotted the number counts for the composite sub-samples of low and high redshift clusters (top panels) and low and high mass clusters (central and bottom panels). The error bars in the plots correspond to the 68% Poissonian confidence intervals evaluated with the approximations of Ebeling (2003). The red points connected by red dashed lines in Figures 4.15 and 4.16 are the red sequence number counts in the composite WINGS spectroscopic sample, which were obtained after converting the WINGS photometry to passively evolved VAB absolute magnitudes. We divided the WINGS red sequence, defined as explained in 3.4.5, into bins of § 0.5 magnitudes down to Vobs = 18.0 mag, corresponding to the 50% completeness of the spectroscopic sample (Cava et al., 2009), and counted the galaxies in each bin. The errors were estimated as for the HCS clusters assuming a Poissonian distribution of the number counts and adopting the approximations of Ebeling (2003). The error bars shown in the plots correspond to the 68% Poissonian confidence intervals. 104 Chapter 4. The Build-up of the Red Sequence

Table 4.4 Red sequence luminous-to-faint ratio (L/F ) of the HCS and WINGS clusters. The values of the cluster halo mass from Jee et al. (2011) are shown in the fourth column from the left. The halo mass and uncertainties quoted for WINGS refer to the median and 68% conﬁdence interval of the halo mass distribution of the subsample of WINGS clusters used in the comparison with HCS. The second row in each entry refers to the red sequence selection with ∆C < 3σ22.

14 Cluster Name redshift (z) MDM δMDM ) 10 M L/F δL/F ) ×

+5.0 +0.016 WINGS 0.05 6.0−3.0 0.705−0.016 +0.013 0.439−0.013 +0.7 +0.16 RX0152 0.84 4.4−0.5 0.42−0.12 +0.4 0.6−0.3 +2.3 +0.4 RCS2319 0.91 5.8−1.6 0.9−0.3 +0.3 0.4−0.2 +1.7 +0.20 XMM1229 0.98 5.3−1.2 0.60−0.19 +0.20 0.28−0.13 +1.8 +0.3 RCS0220 1.03 4.8−1.3 0.8−0.2 +0.3 0.5−0.2 +1.1 +0.5 RCS2345 1.04 2.4−0.7 1.0−0.4 +1.1 1.5−0.8 +2.5 +0.4 XMMU0223 1.22 7.4−1.8 0.8−0.3 +0.6 0.8−0.4 +1.2 +0.3 RDCS1252 1.24 6.8−1.0 0.6−0.2 +0.30 0.31−0.18 +1.7 +0.5 XMMU2235 1.39 7.3−1.4 0.8−0.4 +0.8 1.0−0.6 +3.0 +0.5 XMMXCS2215 1.46 4.3−1.7 1.4−0.4 +0.3 0.7−0.3 4.4. Discussion 105

The black solid line in Figures 4.15 and 4.16 is the best-ﬁt Schechter (1976) function to the red sequence luminosity distribution of a subsample of galaxy clusters selected from the sample of Crawford et al. (2009) with redshift and velocity dispersions in the same range of the 29 WINGS clusters. The Schechter (1976) law is expressed by the equation:

2 h ∗ i(α+1) 0.4(M∗−M) Φ(M) = φ∗ ln 10 100.4(M −M) e10 (4.13) 5 × where M is the absolute magnitude in a given band, M ∗ is the magnitude at the turn-over point of the luminosity distribution, α is the slope of the faint end of the distribution and φ∗ is the value of the number counts or number count density at M ∗. The median values and 1σ errors which we find for the Schechter parameters of the low-redshift Crawford et al. (2009) sample are: M ∗ = 21.6+0.4, α = 0.88+0.30 and φ∗ = 100+50 10−3 Mpc−3. In V − −0.6 − −0.17 −40 × ∗ ∗ order to obtain MV we converted the B-band Vega M values listed in Table 3 of Crawford et al. (2009) to our VAB photometry taking into account the different cosmology adopted in that work. In Figures 4.15 and 4.16 the WINGS number counts and the Crawford et al. (2009) luminosity function are appropriately normalised to match the values of the HCS number counts at M ∗. The grey points connected by grey solid lines in the two figures are the HCS number counts measured on the red sequence defined with the alternative selection ∆C < 3σ . The red sequence number counts will be discussed in 4.4.2. | | 22 §

4.4 Discussion

The present section discusses the results presented in Section 4.3 comparing them with other works published in the recent literature and then proposing an evolutionary scenario for the build-up of the cluster red sequence.

4.4.1 The Evolution of the Red Sequence Parameters

Figure 4.13 shows the redshift evolution of the parameters of the fit to the rest-frame cluster red sequence. In the figure are also plotted the results from Valentinuzzi et al. (2011) for the WINGS clusters and other results from works studying the cluster red sequence at redshifts 0.4 < z < 1.5. A comparison of the red sequence parameters with literature results is a highly delicate task because the definitions of quantities, such as the intrinsic scatter, and the fitting techniques adopted in each work may be different. Differences in the photometric techniques (e.g. fixed vs variable aperture photometry), 106 Chapter 4. The Build-up of the Red Sequence the spectral libraries used in the estimate of the absolute magnitudes, and the adopted cosmologies may also contribute to systematic differences between the results of different works. Furthermore, some authors prefer to apply a morphological selection to their red sequence samples retaining only elliptical and S0 galaxies. With these caveats in mind we now proceed to the comparison of the HCS red sequences with those of other clusters studied in the literature.

The top panels of Figures 4.13 show the evolution of the rest-frame zero-points in the (B V ), (U V ) and (U B) colours, respectively. The (B V ) red sequence − − − − colours at VV ega = 20.5 mag are all consistent, within the uncertainties, with the median − value calculated on the red sequence of the WINGS clusters from the Valentinuzzi et al. (2011) measurements (green triangle). The latter estimates show a broad range of values ( 0.3 mag) and a slightly redder median colour ( 0.1 mag) than those observed in HCS. ∼ ∼ The (U V ) red sequence colours at VV ega = 20.5 mag are consistent with the values − − predicted by the simulations of Romeo et al. (2008) at z 1 (cyan octagons), while the ∼ (U B) colours at BV ega = 21.4 mag are bluer than those measured in Mei et al. (2009) − − for clusters at similar redshifts (red triangles). Two of the HCS clusters, namely RX0152 (z = 0.84) and RDCS1252 (z = 1.24), are also in the sample analysed by Mei et al. (2009) and, while the (U B) colours are still consistent within the uncertainties for RX0152, − the results are discrepant in the case of RDCS1252. We attribute this difference to the different formation redshift ranges and filter sensitivity curves used in the conversion to rest-frame magnitudes, and to the different photometric techniques adopted in Mei et al. (2009), who measured colours within apertures of variable sizes. However, as a further element of distinction, we also point out that Mei et al. (2009) analysed only early-type galaxies on the red sequence and considered only the spectroscopically confirmed cluster members. Although, as shown by Mei et al. (2009), the zero-point is stable to changes in the limiting magnitude of the sample and to its morphological mixing, we note that the (U B) colour, is more sensitive to star formation rate than (B V ) and, therefore, the − − inclusion of late-type galaxies in our red sequence sample may result in bluer zero-points.

The middle panels in Figure 4.13 show the evolution of the rest-frame red sequence slope in all the filter pairs together with the results of other studies and the predictions of the hydrodynamical simulations of Romeo et al. (2008). It can be seen that our results are consistent with the results of Meyers et al. (2012) (red squares) and Mei et al. (2009) in the same redshift range. The sample of Meyers et al. (2012) comprises also all the HCS clusters except RX0152 and, interestingly, these authors find shallower slopes than in the present analysis. We attribute this difference to the fact that Meyers et al. (2012) 4.4. Discussion 107 measured galaxy colours within variable apertures and, as shown by Scodeggio (2001), such a technique produces shallower red sequence slopes. Interestingly, despite the offset in the zero-point, the HCS (U B) vs B red sequence slopes are consistent with those − measured by Mei et al. (2009). Romeo et al. (2008) predicted positive slopes at z > 1 (middle panel), which does clearly not agree with any of the observational results shown in the plots. We rather note that the measured rest-frame slopes at z > 1 are all consistent with those measured at lower redshift by Ellis et al. (1997) (magenta pentagons) and Valentinuzzi et al. (2011).

The bottom panels in Figure 4.13 show the evolution of the rest-frame scatter. We see that our results are consistent with Valentinuzzi et al. (2011) (left-hand panel) although clusters at z > 1.1 in HCS tend to have slightly higher values. Our results are also consistent with the observations of Meyers et al. (2012), Snyder et al. (2012) (green crosses) and Mei et al. (2009) at similar redshifts and with the results of Ellis et al. (1997) and Jaﬀ´eet al. (2011) (blue squares) at 0.4 < z < 0.8. Our measurements also agree with the predictions of the hydrodynamical simulations of Romeo et al. (2008) (middle panel) and of the semi-analytic models of Menci et al. (2008) (right-hand panel, blue stars), although the latter work predicts higher average intrinsic scatters (σc 0.14 mag) in the (U B) ∼ − vs B red sequence, which are higher than those measured in HCS and in Mei et al. (2009).

The results of this study, and the comparison with previous works, support the notion of an early assembly of the cluster red sequence, which at z = 1 has already a negative slope and an intrinsic scatter consistent with those measured at lower redshift. This is in agreement with the recent cosmological simulations of Gabor & Davé(2012) but not with the earlier theoretical works of Menci et al. (2008) and Romeo et al. (2008). The latter authors, indeed, predicted either a flat non-evolving red sequence (Menci et al., 2008), or a strongly evolving red sequence which, at z = 0.8, is expected to have a positive slope and a large intrinsic scatter (Romeo et al. 2008, cyan octagons in Figure 4.13). Romeo et al. (2008) propose two scenarios to explain the build-up of the cluster red sequence and the resulting positive slope at high redshifts. In the first one, massive galaxies accrete star- forming neighbouring galaxies and then land on the red sequence after star formation has ceased. In the second scenario galaxies accrete other galaxies through dry mergers gaining mass and luminosity (see also Faber et al. 2007) but without additional star formation. In both cases the bright end of the red sequence is assembled after z 1 and this results in ∼ the cluster red sequence having a positive slope above this redshift. This late build-up may be the result of an improper treatment of internal processes, such as supernova and AGN feedback, which are shown to be crucial in driving star formation quenching in galaxies 108 Chapter 4. The Build-up of the Red Sequence

(Croton et al., 2006). Gabor & Davé(2012) implemented a feedback mechanism based on the interplay between the galaxy and its host halo regardless of the environment of the galaxy. When 12.5 the halo mass crosses a critical value (10 M ), virial shocks are triggered and the halo gas is consequently heated. This prevents its infall towards the galaxy and the fuelling of star formation. As it is shown in Figure 1 of Gabor & Davé(2012), these simulations are successful in reproducing a red sequence with a negative slope at z = 1. The critical 10.5 halo mass corresponds to a stellar mass M? 10 M at 1.0 < z < 2.0 (Moster et al., ∼ 2010), which is typical of bright red sequence galaxies in the HCS, as it will be shown in Chapter 5. This suggests that the feedback mechanism proposed by Gabor & Davé (2012) can accelerate quenching in massive galaxies inhabiting cluster cores at z > 1.5. These galaxies become quiescent and build up the bright end of the red sequence at z > 1. Nonetheless, it is interesting to note that Fassbender et al. (2014) find that the brightest galaxy in the core of the cluster XDCP J0044.0-2033, at z = 1.58, shows signs of recent star formation in its spectrum (post-starburst) and is bluer than the red sequence. Thus this cluster seems consistent with a scenario in which the BCG has recently accreted star-forming galaxies which moved it towards the blue cloud. The simulations of Romeo et al. (2008) and Menci et al. (2008) also predict high red sequence scatters at z > 0.8, higher than those observed in clusters at z > 1. According to Romeo et al. (2008), at 0.75 < z < 1.5, the intrinsic scatter varies in the range 0.1 <

σc,UV < 0.3 while Menci et al. (2008) predict 0.05 < σc,UB < 0.30 at the same redshifts. Although the predictions of these two simulations are similar, the redshift trends are diﬀerent in the two works. In fact, while according to Romeo et al. (2008) the intrinsic scatter of the red sequence should monotonically increase at 0.75 < z < 1.5, Menci et al. (2008) predict constant intrinsic scatter with redshift. The observational results presented in Figure 4.13 suggest a ﬂat trend of the red sequence scatter with redshift, although the range spanned by σc is broad ( 0.3 mag) and measurements can have large ∼ uncertainties. As already pointed out, the HCS clusters at z > 1.1 have slightly higher intrinsic scatter than those at lower redshifts, but the measurements are still consistent within the uncertainties.

4.4.2 The Luminous-to-Faint Ratio and the Luminosity Distribution of the Cluster Red Sequence

Recent works on galaxy clusters at 0.8 < z < 1.5, such as those of Rettura et al. (2011) and Muzzin et al. (2012), ﬁnd that the evolution of cluster galaxies is primarily driven 4.4. Discussion 109

τ (Gyr) 13.7 11.3 9.4 8.0 6.8 5.9 5.2 4.6 4.1 2.5

2.0

1.5

1.0 ( L/F ) ( L/F ) 1.0

0.5 0.5

2.5 1 2 3 4 5 6 7 8 9 10 z 14 MDM( 10 M ) · ¯ 2.0

1.5

1.0 ( L/F ) ( L/F ) 1.0

0.5 0.5

0.2 0.4 0.6 0.8 1.0 1.2 1.4 2 3 4 5 6 7 8 9 z 14 MDM( 10 M ) · ¯

Figure 4.14 The luminous-to-faint ratio (L/F ) of the cluster red sequence in the HCS. Top- left panel: redshift evolution of L/F . Top-right panel: L/F as a function of cluster halo mass MDM . The diamonds correspond to L/F of the composite WINGS spectroscopic sample. The MDM estimate for WINGS corresponds to the median halo mass while the error bars correspond to the width of the 68% confidence interval of the MDM distribution. The bottom panels show L/F as a function of cluster redshift (left) and cluster halo mass (right) for the alternative selection of galaxies on the red sequence with ∆C < 3σ22. by stellar mass and that the effect of the environment is to accelerate star formation quenching. In this scenario one should expect a significant build-up of the red sequence with redshift. In particular, since massive galaxies are observed to be more efficient in quenching star formation (downsizing, Cowie et al. 1996), one should expect a deficit of 110 Chapter 4. The Build-up of the Red Sequence galaxies at the faint end of the red sequence in clusters at higher redshift. Figure 4.14 (top panels) shows the trend of the luminous-to-faint ratio with cluster redshift and halo mass. L/F appears constant with both redshift and halo mass, although the distributions are broad in both cases. This result agrees with the conclusions of Andreon (2008), Crawford et al. (2009) and De Propris et al. (2013), suggesting that no deficit is observed at the faint end of the red sequence in clusters at z < 1.5 in the magnitude range VAB < 19.5 mag. This conclusion does not change if we select galaxies − on the red sequence by considering the objects with C CRS < 3σ , where σ is the | − | 22 22 colour uncertainty on the red sequence at apparent magnitude m = 22.0 mag. We note that the errors on L/F are large, underlining the high uncertainties related to this quantity and also observed in other works (e.g.: De Lucia et al. 2007b). Our trends with halo mass, with both selections, agree with De Lucia et al. (2007b) who found no significant difference in L/F between clusters with high (σ > 600 km/s) and low (σ < 600 km/s) velocity dispersions at 0.4 < z < 0.8, although we note that only 14 three of the clusters analysed in De Lucia et al. (2007b) have masses MDM 5 10 M , ≥ × our adopted boundary between low and high mass clusters. Figure 4.15 shows the luminosity distributions of red sequence galaxies in each individual cluster ordered by increasing halo mass. It can be seen that the limiting magnitude

VAB = 19.5 mag adopted in the definition of the luminous-to-faint ratio imposes a bright − limit on the deepest datasets which does not allow us to investigate the properties of the faint end of the red sequence in some of the clusters. However, all the clusters appear consistent with no truncation at the faint end of the red sequence. The number counts remain indeed consistent both with the WINGS spectroscopic red sequence and with the low-z Schechter fit of the Crawford et al. (2009) sample. Interestingly, in the specific case of the cluster XMMU2235 (z = 1.39), our results agree with Lidman et al. (2008) who found no truncation in the red sequence of this cluster. The grey crosses connected by solid lines are the HCS red sequence number counts estimated by adopting the alternative selection C CRS < 3σ . It can be seen that the effect of this selection is a loss of | − | 22 galaxies at faint luminosities, where photometric errors are larger (Figure 4.2) and galaxies may be easily scattered off the red sequence. Despite the loss of objects at the faint end, the number counts are still consistent with those obtained with the selection based on the intrinsic scatter. We note that the WINGS number counts (red circles and dashed lines) decrease at

VAB > 20.0 mag although they are still consistent with the Crawford et al. (2009) − luminosity function. This trend suggests that the sample may be incomplete at these 4.4. Discussion 111

102 102 HCS RCS2345 (z =1.04) HCS XMMXCS2215 (z =1.46) HCS (3σ cut) HCS (3σ cut) WINGS WINGS 101 101 gal gal

0 0 N 10 10 N

10-1 10-1

102 102 HCS RX0152 (z =0.84) HCS RCS0220 (z =1.03) HCS (3σ cut) VAB HCS (3σ cut) VAB WINGS WINGS 101 101 gal gal

0 0 N 10 10 N

10-1 10-1

102 102 HCS XMM1229 (z =0.98) HCS RCS2319 (z =0.91) HCS (3σ cut) VAB HCS (3σ cut) VAB WINGS WINGS 101 101 gal gal

0 0 N 10 10 N

10-1 10-1

102 102 HCS RDCS1252 (z =1.24) HCS XMMU2235 (z =1.39) HCS (3σ cut) VAB HCS (3σ cut) VAB WINGS WINGS 101 101 gal gal

0 0 N 10 10 N

10-1 10-1 2 10 -24 -23 -22 -21 -20 -19 -18 HCS XMMU0223V (z =1.22) HCS (3σ cut) AB V WINGS AB 101 gal

0 N 10

10-1 -24 -23 -22 -21 -20 -19 -18 VAB

Figure 4.15 Red sequence number counts in the HCS and WINGS. HCS clusters are ordered by increasing halo mass. Black points and solid connecting lines are for HCS, red circles and dotted connecting lines are for WINGS. The solid black line represents the median Schechter (1976) function of clusters in the same redshift and MDM ranges of WINGS from Crawford et al. (2009). Number counts are shown as a function of VAB-band absolute magnitude passively evolved to z = 0. The grey crosses and the solid connecting grey lines are the HCS red sequence number counts with the alternative selection of red sequence galaxies with ∆C < 3σ22 . 112 Chapter 4. The Build-up of the Red Sequence magnitudes. In fact this region corresponds to the apparent magnitude range 17.0 <

Vobs < 18.0, where the WINGS spectroscopic sample approaches the 50% completeness limit (Cava et al., 2009). However, as shown in De Lucia et al. (2004) and Barkhouse et al. (2007) in clusters at z < 0.2, the red sequence luminosity function is characterised by a downturn in the range 19.0 < V < 17.0 which is then followed by an upturn. Several − − works show that the luminosity and mass functions of both ﬁeld and cluster galaxies is well ﬁtted by a double Schechter function, one for luminous/massive galaxies and the other for faint/dwarf galaxies (e.g. Driver & Phillipps 1996; Barkhouse et al. 2007). The WINGS number counts were corrected for spectroscopic incompleteness across the entire range

Vobs < 18.0 mag and therefore we expect that the decrease at V > 20.0 mag is due to a − real lack of galaxies at these luminosities. This deficit would also agree with the median +0.16 L/F = 0.66 estimated by Valentinuzzi et al. (2011) across the range VV ega < 18.2 −0.18 − mag using the same magnitude ranges of De Lucia et al. (2007b) to divide into bright and faint galaxies 6. In the top panels of Figure 4.16 we show the red sequence number counts for the composite samples of clusters at 0.8 < z < 1.1 (low-z, left-hand panel) and 1.1 < z < 1.5 (high-z, right-hand panel). In these two cases the magnitude limit is set by the shallowest data-set in each composite sample. We note that there is no significant difference between low- and high-z clusters in the HCS, and no evident truncation of the red sequence with respect to WINGS at VAB < 20.0 mag. This conclusion further supports the scenario − of an early build-up of the cluster red sequence, although we note that the magnitude cut is 2 mag brighter than those considered in most works up to z = 0.8 (e.g. De Lucia et al. 2007b). The effect of the alternative selection with C CRS < 3σ is also in these | − | 22 two cases the loss of galaxies at faint luminosities. The number counts are, however, still consistent with WINGS and the Schechter function. The central panels in Figure 4.16 show the red sequence number counts in the two low (left-hand panel) and high (right-hand panel) halo mass samples. The low and high 14 mass samples are defined as those containing all clusters with MDM < 5 10 M and × 14 MDM 5 10 M , respectively. This value approximately corresponds to the turn-over ≥ × point of the halo mass function (see e.g. Figures 1 and 2 in Jenkins et al. 2001). The low- mass sample contains the clusters RX0152, RCS0220, RCS2345 and XMMXCS2215, while the high mass sample contains the clusters RCS2319, XMM1229, XMMU0223, RDCS1252

6This value is 0.16 higher than the value reported by Cerulo et al. (2014). This correction has been made possible after the original Valentinuzzi et al. (2011) catalogues were made available by the WINGS collaboration. Although higher the current value is still consistent to within the uncertainties with L/F quoted by Cerulo et al. (2014) and read from Figure 4 of Valentinuzzi et al. (2011). 4.4. Discussion 113 and XMMU2235. The magnitude limit for the number counts is set by the shallowest data set in each of the sub-samples. We note again that there is no difference between the two distributions, at least down to VAB = 20.0 mag. − The bottom panels of Fig. 4.16 show the red sequence number counts of the low- and high-mass samples after the removal of RX0152 and RDCS1252. In fact, as shown in 4.2.2, if we adopted the total masses estimated from the X-ray luminosities, RX0152 § would be assigned to the high-mass sample and RDCS1252 would be assigned to the low- mass sample. We note that now the red sequence of the massive sample appears populated at magnitudes VAB < 22.0 mag where we detect no object in the low-mass sample. This − result is in agreement with Lemaux et al. (2012) who found that the red sequence in the more massive and virialised clusters of the Cl1604 supercluster, at z = 0.9, is populated at luminosities log(LB/L ) > 10.9 where the less massive systems show a lack of objects. This result suggests that the bright end of the red sequence evolves faster in more massive clusters and agrees with the positive correlation existing between cluster mass and stellar mass of the brightest cluster galaxy (see e.g. Lidman et al. 2012). We will interpret this result within our proposed scenario for the build-up of the cluster red sequence in 4.4.2. § Here we point out that the difference at the bright end of the red sequence is driven by only 8 galaxies and that the subsamples of low- and high-mass clusters are made of 3 and 4 clusters, respectively, thus not ruling out the hypothesis that the difference between the two luminosity distributions is due to statistical fluctuations.

We also note that the bright end of the WINGS red sequence is populated at VAB < 22.5 mag where there is no galaxy detected in the HCS. This range is mainly, but not − only, populated by the WINGS BCGs and our results suggest that these massive galaxies formed at redshifts z < 0.8, probably via subsequent dry mergers, as suggested in (Faber et al., 2007). Interestingly, Lidman et al. (2013) find that BCG growth at z 1 is driven ∼ by major mergers, although accretion of small companions may play an important role in in the evolution of these galaxies at lower redshifts (Jiménezet al., 2011). Rudnick et al. (2012) also found a deficit at the bright end of the red sequence in the cluster ClG J0218.3-0510 at z = 1.6, supporting our conclusions.

The alternative selection of red sequence galaxies with C CRS < 3σ results also | − | 22 in this case in a loss of galaxies at faint magnitude, although the number counts are still consistent with WINGS and the Schechter function. The faint end of the red sequence does not show any deficit of galaxies in all the subsamples, confirming the result of the luminous-to-faint ratio, and suggesting a scenario in which the cluster red sequence is already assembled at z = 1.5 at the faint end. We 114 Chapter 4. The Build-up of the Red Sequence note that our conclusion is not in agreement with most works in the recent literature. De Lucia et al. (2007b), Capozzi et al. (2010) and Bildfell et al. (2012) all detected a truncation of the cluster red sequence at z < 0.8 which was then observed at higher redshifts (z = 1.6 1.8) by Rudnick et al. (2012) and Fassbender et al. (2014) (but see − Andreon et al. 2014 for different conclusions). Thus, in the remaining of this section, we discuss the possible causes of disagreement between our results and those in which a truncation of the red sequence is seen.

As already mentioned, a direct comparison with the results of De Lucia et al. (2007b), Capozzi et al. (2010) and Bildfell et al. (2012) is not possible with our data because we are limited to magnitudes V < 19.5 mag, while those authors were able to study the red − sequence down to VV ega = 18.2 mag. The different magnitude ranges adopted for the − definition of the bright and faint samples can influence the results of the measurements. One other source of disagreement is related to the selection of the sample, which can be populated only by clusters that have a deficit of galaxies at the faint end of the red sequence. For example, Valentinuzzi et al. (2011), adopting the same magnitude ranges for the definition of the bright and faint samples, obtained a median L/F consistent with the value found by De Lucia et al. (2007b) at z = 0.57 in the EDisCS survey. If one only compares these two samples, no truncation of the red sequence would be detected. On the other hand, Figure 9 of De Lucia et al. (2007b) shows that all measurements of L/F in EDisCS are consistent within the errors, and that L/F at z = 0.4 is consistent with the values of L/F found in their z = 0 comparison samples. The uncertainties in L/F are generally large because of the scatter between different clusters, as it can also be seen from Capozzi et al. (2010) and Bildfell et al. (2012) and, therefore, any conclusion regarding the existence of a truncation of the red sequence from such measurements should be treated with caution.

An artificial deficit of galaxies at the faint end of the red sequence can be the result of surface brightness incompleteness, which prevents diffuse objects brighter than the magnitude completeness limit to be detected. In our sample, surface brightness incompleteness is responsible for the loss of 3-5 % of the objects with VAB > 20.0 mag in all clusters − and, therefore, its effect on our results is negligible. One other artificial deficit, as shown throughout this section, can also be the result of the boundaries chosen for the selection of galaxies on the red sequence.

Although the measurements of L/F are highly uncertain and the current samples of clusters at z > 0.4 are small compared to those available at lower redshifts, our results do not exclude that the truncation of the red sequence is a characteristic of some individual 4.4. Discussion 115 clusters. Studies of the stellar populations of cluster galaxies at z 1 suggest that low mass ∼ galaxies were quenched at later epochs with respect to more massive galaxies (Muzzin et al. 2012; Nantais et al. 2013b), hence supporting an evolutionary scenario in which stellar mass is the primary driver of the build-up of the red sequence and the effect of the environment is to accelerate star formation quenching in galaxies. This scenario is in agreement with a truncation of the red sequence and does not exclude that a deficit can be detected at magnitudes fainter than those allowed by current data. However, larger and more homogeneous samples are needed in order to carry out such analyses with higher accuracy and statistical significance.

4.4.3 A Possible Scenario for the Build-up of the Cluster Red Sequence

Gabor & Davé(2014) showed that the independence of internal and environmental star formation quenching Peng et al. (2010b) can be explained in the framework of structure formation by considering the relation between halo and galaxy properties in hydrodynamical simulations. In particular, they showed that dark matter haloes with masses 14 MDM 10 M host only environments with high galaxy densities. The scenario pro- ∼ posed by these authors for the formation and evolution of galaxies in clusters consists in the initial formation of the central galaxy, the progenitor of the z = 0 BCG, in a high- density region of the universe at z > 2. This galaxy will have an initial phase of intense star formation fuelled by the inflow of cold gas. Massive and highly star-forming galaxies at z = 2 4 are indeed observed at IR and submillimetre wavelengths and the observations − of Marchesini et al. (2014) suggest that these galaxies are the progenitors of z = 0 galaxies with log(M∗/M ) > 11.8. The initial dark matter halo accretes matter from the surrounding environment and, 12.5 at some time, the critical mass MDM = 10 M (Dekel & Birnboim, 2006) for the onset of virial shock heating is reached. At this point the gas becomes sufficiently hot to stop its inflow to the galaxy. The galaxy will therefore land on the red sequence by z 2 ∼ becoming the central galaxy of a protocluster of galaxies and continuing to accrete matter through mergers with surrounding low-mass satellite galaxies. It is important to stress here that feedback processes driven by AGN and supernovae can also be responsible for the cessation of star formation in massive star-forming galaxies (Croton et al., 2006) The protocluster will continue to accrete smaller haloes (groups and individual galaxies) which will host a fraction of already quiescent galaxies (preprocessing). The accreted galaxies will populate the red sequence at intermediate and low masses. The quenching of 116 Chapter 4. The Build-up of the Red Sequence

HCS HCS HCS (3σ cut) HCS (3σ cut) 2 0 8 1 1 1 1 1 5 2 10 WINGS .

101 101 gal gal N N

100 100

10-1 10-1

HCS HCS HCS (3σ cut) VAB 14 HCS (3σ cut) VAB 14 2 2 WINGS MDM <5 10 M WINGS MDM >5 10 M 10 · ¯ · ¯ 10

101 101 gal gal N N

100 100

10-1 10-1

HCS HCS HCS (3σ cut) VAB 14 HCS (3σ cut) VAB 14 2 2 WINGS MDM <5 10 M WINGS MDM >5 10 M 10 · ¯ · ¯ 10

101 101 gal gal N N

100 100

10-1 10-1

-24 -23 -22 -21 -20 -19 -18 -24 -23 -22 -21 -20 -19 -18 VAB VAB

Figure 4.16 Red sequence number counts at low and high redshifts and low and high cluster halo mass. Colours and symbols are as in Fig. 4.15. Top panels: red sequence number counts in clusters at 0.8 < z < 1.1 (left) and 1.1 < z < 1.5 (right). Central panels: 14 red sequence number counts in clusters with MDM < 5 10 M (low-mass sample, 14 × left) and MDM 5 10 M (high-mass sample, right). Bottom panels: the same as in the middle panels≥ × but excluding RX0152 and RDCS1252, the two clusters with large diﬀerences between X-ray and weak-lensing halo masses (see discussion in 4.4.2). § star formation is accelerated in the dense environments of the groups (Rettura et al., 2011; Muzzin et al., 2012) and therefore the faint end of the red sequence will be composed for 4.4. Discussion 117 a substantial fraction by passive accreted satellites. This will result in a fast build-up of the cluster red sequence which, at z 1, will be also populated at log(M∗/M ) < 10.5. ∼ We note that Gabor & Dav´e(2014) show that 30% of red satellite galaxies at z = 0 are preprocessed galaxies, which therefore constitute a relevant fraction of the total red population. In a more mathematical form, this accretion scenario can be expressed, at a given epoch t, by the following equation:

fRS(t) = fRS,c(t) + fRS,pp(t) (4.14) where fRS is the total fraction of red sequence galaxies in the cluster, fRS,c is the red sequence fraction originated in the central protocluster, and fRS,pp(t) is the fraction of preprocessed galaxies accreted with the satellite groups. In this scenario we can also explain the presence of massive central galaxies in more massive clusters. In fact, if galaxies formed earlier in the highest overdensities, they also accreted satellites over a longer time span, thus becoming more massive than their counterparts in lower-mass haloes. In Chapter 5 we will discuss the morphological evolution of red sequence galaxies in the framework of the present accretion scenario, while in Chapter 6 we will show how galaxy spectra can be used to investigate the evolution of the red sequence as a function of luminosity and morphology.

5 The Morphological Evolution of Red Sequence Galaxies in Clusters

The present chapter discusses the morphological transformations of red sequence galaxies in clusters via the comparison of the morphological properties of galaxies in the HCS and WINGS samples. The morphological fractions along the red sequence in the two samples, as a function of luminosity and stellar mass, are compared in order to deduce the evolutionary paths galaxies of diﬀerent morphological type might have taken. The understanding of how and where the morphological fractions change allows one to gain an understanding of the physical processes that may be at play in the cluster environment. This chapter is organised as follows: Section 5.1 describes the morphological classiﬁ- cation of red sequence galaxies in the HCS and Section 5.2 the measurements of stellar masses. Section 5.3 introduces the comparison sample of low-redshift clusters while Sec- tion 5.4 presents the results of the morphological analysis. Section 5.5 eventually discusses possible scenarios for the morphological evolution in clusters of galaxies.

Throughout this chapter we adopt a ΛCDM cosmology with ΩΛ = 0.73, Ωm = 0.27 and H = 71.0 km s−1 Mpc−1 as done in Chapter2, Chapter 3 and Chapter 4. Unless 0 · · otherwise stated, all magnitudes are quoted in the AB system (Oke, 1974).

5.1 Morphology of the HCS Galaxies

5.1.1 Classiﬁcation Procedure

The procedure adopted for the morphological classiﬁcation of the HCS red sequence galaxies is analogous to that described in 3.4.4 for the XMM1229 cluster. The HST/ACS § F850LP (z850) images, which are the deepest available in the HCS sample, were used to perform the morphological classiﬁcation in all the clusters. The z850 band corresponds to

119 120 Chapter 5. Morphological Evolution

RX0152 (z=0.84) RCS2319 (z=0.91) 1.8 1.4 1.6 1.4 1.2 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 F625W - F775W

0.4 F775W - F850LP 0.2 0.2 0.0 0.0

20 22 24 26 20 22 24 26 F775W (AUTO) F850LP (AUTO) XMM1229 (z=0.98) RCS0220 (z=1.03) 1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4 F775W - F850LP F775W - F850LP 0.2 0.2

0.0 0.0

19 20 21 22 23 24 25 19 20 21 22 23 24 25 F850LP (AUTO) F850LP (AUTO) RCS2345 (z=1.04) XMMU0223 (z=1.22) 1.4 2.8

1.2 2.4

1.0 2.0

0.8 1.6 0.6 1.2 0.4 0.8 F775W - F850LP

0.2 F775W - J (HAWK-I) 0.4 0.0 0.0 20 22 24 26 19 20 21 22 23 24 25 F850LP (AUTO) J (HAWK-I) (AUTO)

(a) Figure 5.1 Observed colour-magnitude diagrams of the individual HCS clusters with the morphologically classified red sequence galaxies highlighted by different colours and symbols. All the red sequence galaxies with z850 < 24.0 mag were selected for morphological classification. Grey points are all galaxies observed in the field of each cluster within 0.54 R200 from the cluster centroid. Red crosses are elliptical galaxies, orange diamonds× are bulge-dominated/S0 galaxies (BD), green triangles are early-type disc- dominated galaxies (EDD), and blue squares are late-type disc-dominated and irregular galaxies (LDD+Irr). 5.1. Morphology of the HCS Galaxies 121

RDCS1252 (z=1.24) XMMU2235 (z=1.39) 2.8 2.8

2.4 2.4

2.0 2.0

1.6 1.6

1.2 1.2

0.8 0.8 F775W - F125W F850LP - F125W

0.4 0.4

0.0 0.0 20 22 24 26 19 20 21 22 23 24 25 F125W (AUTO) F125W (AUTO) XMMXCS2215 (z=1.46) 4.8

4.2 all galaxies 3.6

3.0 Elliptical 2.4 S0 1.8 Disc-domin. (I) 1.2 F850LP - Ks (HAWK-I) Disc-domin. (II) / Irregulars 0.6

0.0 19 20 21 22 23 Ks (HAWK-I) (AUTO)

(b) Figure 5.1 Continued. The lines are the same as in Figure 4.3. The solid lines represent the fit to the observed red sequence. The dotted lines mark the boundaries of the red sequence. The diagonal dashed lines represent the 90% colour completeness limits. As it can be seen, due to the magnitude cut z850 = 24.0, the red sequence is not covered down to the faint end in all the clusters and in some cases, as in XMMXCS2215 (bottom-left panel) only galaxies which are bluer than the best-fit line fall in the sample. 122 Chapter 5. Morphological Evolution the SDSS g band in the redshift range 0.8 < z < 1.24 and to the SDSS u band at redshift 1.3 < z < 1.5. This allowed us to make a direct comparison with the morphologies of red sequence galaxies in the low-z clusters of the WIde-field Nearby Galaxy-cluster Sur- vey (WINGS, Fasano et al. 2012) without the need to apply morphological k-corrections (Windhorst et al., 2002).

We selected all red sequence galaxies within 0.54 R from the cluster centre with × 200 total (SExtractor MAG AUTO) magnitudes z850 < 24.0 mag, the limit down to which we were able to distinguish morphological features (Cerulo et al. 2014; Delaye et al. 2014; Postman et al. 2005). The selection resulted in a subsample of 428 galaxies that we will refer to as the morphological sample in the following sections. The morphological sample also includes the XMM1229 members classiﬁed in Cerulo et al. (2014). As shown in Figure

5.1, the limiting magnitude z850 = 24.0 mag of the morphological sample does not allow us to study the morphology of all red sequence galaxies in all the clusters. For example, we cannot study the morphology of galaxies at the faint end of the red sequence in the clusters RCS2319, RDCS1252 and XMMU2235, while in the cluster XMMXCS2215 only galaxies which are bluer than the best-ﬁt line to the observed red sequence fall in the morphological sample. We will discuss the implications of this limit in Section 5.5.

We divided the morphological sample into ﬁve types which were based on the apparent bulge-to-total (B/T ) light ratio, namely, ellipticals (pure bulges, E), bulge-dominated (BD), which correspond to S0/S0a galaxies, early-type disc-dominated (EDD), corresponding to Hubble types in the range Sa-Sbc, late-type disc-dominated (LDD), corresponding to Hubble types in the range Sc-Scd, and irregular galaxies (Irr). Since only 30 galaxies (7%) of the morphological sample have late-type disc or irregular morphologies, we decided to merge these two classes into one family of LDD+Irr galaxies (blue squares in Figure 5.1). Therefore, as previously done in the study of the morphological properties of XMM1229, in the following we will present the results of the investigation of the morphological evolution of the four classes of E, BD, EDD and LDD+Irr galaxies. Since most of the S0 galaxies fall into the class of bulge-dominated, we will use these two terms interchangeably.

We adopted a coupled visual and automated classification procedure whereby each galaxy was visually inspected by three independent classifiers (i.e. P.C., W.J.C and C.L.) and a fourth independent classification was obtained by running the galSVM software (Huertas-Company et al., 2008, 2011) on the F850LP images. galSVM is an IDL package based on Support Vector Machines (SVM), which are machine learning algorithms particularly suited for the solution of problems of classification in large samples. The software 5.1. Morphology of the HCS Galaxies 123 also implements an a posteriori estimate of the probability for each galaxy to be of a certain morphological type. This allows the uncertainty in the classification to be quantified (see Appendix B). The morphological type was defined as the mode of the four classifications. In cases in which two classifiers agreed on one type and the other two on a different type, the type corresponding to the earliest morphological class was assigned to the galaxy. Thus, if two classifiers classified a galaxy as elliptical and the other two as S0, the galaxy was assigned to the morphological class of the ellipticals. In fact, with the cosmological surface brightness dimming, the disc of the faintest galaxies falls below the level of the sky noise and, therefore, the chances of incorrectly detecting a disc in the visual classification are high. In this way the halo of an elliptical galaxy can be confused with a low-inclination disc. We will discuss this in more detail, together with the overall reliability of the morphological classification, in the next section. Galaxies for which each classifier assigned a different type were not assigned to any class. Only 7 objects (2%) are unclassified in the HCS morphological sample. Example postage stamp images of the galaxies classified on the red sequence of the cluster XMM1229 (z = 0.98) are shown in Appendix A.

5.1.2 Testing Morphology I: Internal Comparison

In order to assess the reliability of the morphological classification of red sequence galaxies in the HCS, we investigate the agreement between different individual classifiers and between the three visual classifiers and galSVM, and discuss all the possible sources of uncertainties. There was full agreement between the four classifiers on only 77 galaxies, which correspond to 18% of the morphological sample, while the three human classifiers agreed on assigning the same type to 173 galaxies, corresponding to 40% of the morphological sample. These fractions are an improvement on the agreement found in the classification of red sequence galaxies in XMM1229 (8% for the agreement of the four classifiers and 14% for the agreement of the human classifiers), but are still low and underline the difficulties in the morphological classification of distant galaxies. The agreement between classifiers improves if the four morphological classes are merged into broad classes of early- (i.e. elliptical+S0) and late-type (i.e. disc-dominated and irregulars) galaxies. In this case the four classifiers agree in assigning the same early or late type to 57% of the sample while the three human classifiers agree in assigning the same type to 70% of the sample. Al- though there is still a considerable disagreement in the classification of early- and late-type 124 Chapter 5. Morphological Evolution galaxies (40% between all classifiers and 30% between human classifiers), we see that the agreement in the classification, when the distinction between elliptical and S0 galaxies is removed, improves by a factor of 3 (2 if only visual classification is taken into account). This suggests that the main source of error in our classification lies in the separation between elliptical and S0 galaxies. Such a conclusion is not surprising since, as it is well known in the literature (e.g. Mei et al. 2009; Huertas-Company et al. 2008, 2011), face-on S0 galaxies can be easily misclassified as elliptical galaxies. This difficulty is exacerbated in distant galaxies, where the cosmological surface brightness dimming, which has a (1 + z)4 dependence, causes the lowest-surface-brightness features, such as discs and spiral arms, to become fainter than the sky background. As a result, in the faintest S0 or disc-dominated galaxies with low inclination, only the bulge is visible, and they are classified as ellipticals. Interestingly, we note that when comparing the visual classification, defined as the mode of the three individual visual classifications, and the galSVM classification, 19% of the objects with early-type disc-dominated visual morphology are classified as elliptical galaxies by galSVM. This mismatch is explained by the fact that the automated classification, which relies upon the measurement of morphological coefficients based on the light distribution of the galaxies, is even less efficient in detecting low-surface-brightness features in the images. We also note that 33% of the galaxies with visual S0 type were classified by galSVM as early-type disc-dominated. This mismatch certainly underlines the difficulties inherent in the distinction between two galaxy types based only on B/T ratio.

5.1.3 Testing Morphology II: The Elliptical vs S0 Separation

In order to test the reliability of the separation between elliptical and S0 galaxies we study the distributions of 6 morphological parameters, namely ellipticity, concentration, asymmetry, Gini coefficient, second order moment of the light distribution, and Sérsic index. The ellipticity e of galaxies is a proxy for both inclination and bulge-to-total light ratio (Binney & Merrifield, 1998) and, in particular, the ellipticity is directly proportional to the inclination of galaxies, so that face-on galaxies have e 0 and edge-on galaxies have ∼ e 1. Figure 5.2 shows the distribution of the ellipticity of galaxies classified as ellipticals ∼ (red), bulge-dominated (orange), and early-type disc-dominated, in both the HCS (upper panels) and WINGS samples (bottom panels). The top-left panel shows the distributions for the classification adopted in this work. In the top-right panel we change the criterion for the attribution of the morphological class of those HCS galaxies for which two classifiers 5.1. Morphology of the HCS Galaxies 125

60 ellipticity distributions 60 ellipticity distributions ellitpical ellitpical 50 S0 50 S0 disc-dominated (I) disc-dominated (I) 40 40

30 30

20 20

10 10 number of galaxies number of galaxies 0 0 ellipticity ellipticity (e) 500 500

400 400

300 300

200 200

100 100 number of galaxies number of galaxies 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 ellipticity (e) ellipticity (e) (a) (b)

Figure 5.2 (a): Distribution of the ellipticity for galaxies classified as ellipticals, bulge- dominated and early disc-dominated in the HCS (top) and WINGS (bottom) samples. (b): the same as in (a) but with the alternative classification discussed in 5.1.3. Galaxies for which 50% of the classifiers agreed on one type and the other 50% on§ a different type are now assigned to the latest-type class of the two. If 2 classifiers classified the galaxy as elliptical and 2 classified it as S0, the galaxy was assigned to the class of the S0s. In this way part of the population of S0 galaxies with low ellipticity (nearly face-on, e < 0.2) is recovered. 126 Chapter 5. Morphological Evolution

Table 5.1 Mean values and standard errors of concentration, asymmetry, Gini coeﬃcient and M20. The estimates for each morphological type and for early- and late-type galaxies are all shown.

Morphological Concentration Asymmetry Gini Coeﬃcient M δM 20 20 Type (CCon δCCon) A δA (G δG)

Elliptical 2.81 0.02 0.0729 0.005 0.511 0.003 1.620 0.016 − Bulge-Dominated 2.78 0.02 0.073 0.004 0.538 0.004 1.617 0.018 −

Early-type 2.47 0.04 0.076 0.008 0.473 0.007 1.56 0.03 Disc-Dominated −

Late-type 2.13 0.04 0.08 0.02 0.414 0.008 1.33 0.04 Disc-Dominated + Irr −

early-type galaxies 2.801 0.016 0.073 0.004 0.520 0.003 1.619 0.012 ( Elliptical + S0 ) −

late-type galaxies 2.35 0.03 0.078 0.009 0.452 0.006 1.48 0.02 ( disc-dominated + Irr ) −

agreed on a type and the other two agreed on a different type and, rather than assigning the earliest morphological type between the two, we assign the latest. Thus, if it happened that two classifiers assigned a galaxy to the class of the ellipticals and the other two to the class of the S0s, the galaxy was classified as S0. In this way it is possible to evaluate the contamination of the elliptical and S0 samples due to the arbitrary assignment of a galaxy, for which 50% of the classifiers agreed on a certain type and 50% on a different type, to a particular morphological class. In this test we only consider the first three morphological classes since late-type discs and irregular galaxies represent together only 7% of the morphological sample. The histograms of the WINGS galaxies are corrected for the incompleteness of the spectroscopic sample. As discussed in Section 3.4.5, galaxies in WINGS were classified, adopting a neural-network approach, in 18 morphological classes, which we converted to our morphological scheme as illustrated in Table 3.5. More details on the classification of the WINGS galaxies, and the software and procedures adopted, can be found in Fasano et al. (2012). 5.1. Morphology of the HCS Galaxies 127

The median values of the ellipticities, and the boundary of the 68% confidence intervals, are shown in Table 5.2. We note that the median values of the HCS ellipticities, for each morphological type, are consistent within 1σ with the median values measured in WINGS. However, the WINGS sample exhibits a population of low-ellipticity S0 and early-type disc- dominated galaxies, at e < 0.1, which we do not observe in the HCS sample even if we adopt the alternative classification scheme (top-right panel). This reflects the difficulty in detecting face-on discs in our morphological classification due to the cosmological surface brightness dimming. This effect is especially severe for passive galaxies where the absence of star-formation makes discs fainter in the U, B and V bands, which are the regions of the galaxy spectrum probed by the F850LP images at the redshifts of the HCS clusters. Vulcani et al. (2011b), using the visual morphological classification of Desai et al. (2007) of a subset of 10 EDisCS clusters (White et al., 2005)1, found that only 5% of the S0 galaxies at 0.4 < z < 0.8 have e < 0.1, suggesting that low-ellipticity lenticular galaxies were rare systems at those epochs. Interestingly, Vulcani et al. (2011b) also show that the number of low-ellipticity S0s is higher in nearby clusters. The reliability of the EDisCS classification is tested with galaxy light profiles and the chance of misclassifica- tions, particularly between elliptical and S0 galaxies, is shown to be low (Desai et al., 2007). Given the rarity of low-ellipticity S0 galaxies in distant clusters, we expect that the contamination of the elliptical subsample from face-on S0s in HCS is low and thus it should not influence the conclusions of our analysis. Interestingly, we note that the WINGS elliptical galaxies contain more low-ellipticity systems than the HCS. This difference becomes even more significant if one takes into account the contamination from face-on S0 and disc-dominated galaxies in the HCS. This property of the WINGS ellipticals is also mentioned in Vulcani et al. (2011b) who explained the existence of these galaxies as the result of galaxy mergers (see Holden et al. 2009 for a different conclusion). We will not address this problem in the present work and here we only state that a two-sample Kolmogorov-Smirnov (KS) test returns PKS 0.02 that the ∼ ellipticity distributions of WINGS and HCS elliptical galaxies were drawn from the same parent distribution. This implies that the distributions at z = 0 and z = 1 are different for cluster ellipticals and that elliptical galaxies tend to be rounder in WINGS with respect to HCS. We also note that WINGS hosts a population of high-ellipticity disc-dominated and S0 galaxies, at e > 0.6, which is not observed in HCS. This can be a result of the smaller angular sizes of distant galaxies, with respect to nearby galaxies, which approach the width

1ESO Distant Cluster Survey. 128 Chapter 5. Morphological Evolution of the PSF of the instrument. As a consequence, due to the convolution of their intrinsic light distributions with the PSF in the ACS F850LP ﬁlter, these objects appear rounder.

As a further test for the reliability of the elliptical vs S0 separation in HCS we compare four of the morphological parameters computed by galSVM, namely concentration (Equation B.2), asymmetry (Equation B.3), Gini coefficient (Equation B.4), and second order moment of the galaxy light distribution M20 (Equation B.5). These particular coefficients provide information on the amount of concentration and level of homogeneity of the light distribution in galaxies. The planes formed by concentration and asymmetry (top-left panel), concentration and Gini coefficient (top-right panel), concentration and

M20 (bottom-left panel), and Gini coefficient and asymmetry (bottom-right panel) are shown in Figure 5.3. The mean values and standard errors of the four coefficients are shown in Table 5.1. Huertas-Company et al. (2008) show that galSVM is highly efficient in separating early- and late-type galaxies using Concentration and Asymmetry; however, indices such as the Gini Coefficient and M20 allow galSVM to distinguish between galaxies with or without substructures such as spiral arms, star-forming regions, and bars.

We find that the Gini coefficient is particularly effective in separating elliptical from S0 galaxies. In fact, there is a 5σ difference between the mean values of the Gini coefficient for elliptical and S0 galaxies, the latter having a higher mean Gini coefficient. We also find that a two-sample KS test between the distributions of the Gini coefficient of elliptical and

S0 galaxies returns PKS < 0.1% rejecting the null-hypothesis that the two distributions are drawn from the same parent distribution. The discrepancy between the Gini coefficients of elliptical and S0 galaxies can be attributed to the presence of the disc in S0 galaxies, which results in a more dishomogenous light distribution. The mean values and the uncertainties shown in Table 5.1 confirm that the 4 morphological coefficients considered in this section are effective, with the only exception of asymmetry, in separating between early- and late-type galaxies. Interestingly, as shown in the asymmetry vs Gini coefficient plane, disc-dominated galaxies are found at values of Gini Coefficient less than 0.5, in broad agreement with Meyers et al. (2012), who used these two parameters to select galaxies with different morphologies in the HST Cluster Supernova Survey. In all four panels of Figure 5.3 it can be seen that early (red and orange symbols) and late-type galaxies (green and blue symbols) occupy two different regions of the planes formed by the morphological coefficients. Disc-dominated and irregular galaxies are less concentrated, have lower Gini coefficients, and higher M20 values, indicating the presence of a higher level of substructure with respect to elliptical and S0 galaxies.

Figure 5.4 shows the distributions of the Sérsicindex n of the HCS galaxies, from the 5.1. Morphology of the HCS Galaxies 129 measurements of Delaye et al. (2014)2, for each morphological type. The mean and standard errors for the Sérsicindices of elliptical and S0 galaxies (red and orange histograms, respectively) are: nE = 3.7 +1.1 and nS = 3.0 1.0. Although these two values are 0 consistent to within 1σ, we note that S0 galaxies tend to reside in the low-n region of the distribution of nE suggesting that S0 galaxies tend to have lower values of the Sérsic index compared to elliptical galaxies. A two-sample KS test returns PKS = 0.003 indicating that the two distributions are statistically different. This suggests that elliptical and S0 galaxies in our sample have different structural properties and that S0 galaxies, as expected, are less concentrated than elliptical galaxies. The comparisons discussed in this section, between morphological and structural parameters of galaxies with different morphologies, show that our morphological classification is reliable in separating elliptical from S0 galaxies, at least down to z850 = 23.0 mag. At fainter magnitudes the effects of the cosmological surface brightness dimming become severe and faint face-on S0s or disc-dominated galaxies can erroneously be classified as ellipticals. However, given the low number of these galaxies in distant clusters found by Vulcani et al. (2011b), the contamination of the sample of red sequence ellipticals in HCS should not influence the conclusions of our analysis.

5.1.4 Testing Morphology III: Comparison with the Literature

In Cerulo et al. (2014) we compared the morphological classification of XMM1229 with Santos et al. (2009) and Delaye et al. (2014) finding that our classification agreed with the visual classification of Santos et al. (2009) for 9 of the 15 galaxies in common (13/15 if only the early- vs late-type separation was taken into account), and that 38/46 galaxies were classified as early-type both by us and by Delaye et al. (2014). We can now repeat the same exercise with the entire morphological sample and compare with the morphological classifications published for other HCS clusters. Delaye et al. (2014) published a catalogue of red sequence early-type galaxies classified with galSVM. Of the galaxies in common with our morphological sample, only 5% were not classified as early-type (i.e. elliptical or S0) by us. Hilton et al. (2009) visually classified galaxies within 0.75 Mpc of the centre of the cluster XMMXCS2215, at z = 1.46, using ACS F850LP images as in this work. Of the 16 galaxies in common with our sample, 13 were assigned the same types by Hilton et al. (2009) and by us and, if we consider only the early- vs late-type separation, the two

2The catalogue with the structural parameters and masses of HCS early-type galaxies is available with the electronic version of the paper. 130 Chapter 5. Morphological Evolution

0.65 0.25 0.60 0.20 0.55 0.15 0.50

0.10 0.45 0.40 Asymmetry 0.05 Gini Coefficient 0.35 0.00 0.30

1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0 Concentration Concentration

0.25 0.5 0.20

1.0 0.15 20

M 1.5 0.10

Asymmetry 0.05 2.0 0.00

1.5 2.0 2.5 3.0 3.5 4.0 0.3 0.4 0.5 0.6 Concentration Gini Coefficient

Elliptical Disc-domin. (I) S0 Disc-domin. (II) / Irregulars

Figure 5.3 Morphological parameters of HCS red sequence galaxies as estimated by galSVM. Top-left: asymmetry vs concentration, top-right: Gini coefficient vs concentration, bottom- left: M20 vs concentration, bottom-right: asymmetry vs Gini coefficient. A combination of concentration, Gini coefficient and M20 is effective in dividing the sample between early- and late-type galaxies. Disc-dominated and irregular galaxies are less concentrated, have lower Gini coefficients, and higher M20 values, indicating the presence of a higher level of substructure with respect to elliptical and S0 galaxies. 5.2. Stellar Mass Estimate of Red Sequence Galaxies 131

Table 5.2 Median ellipticities, colours relative to the red sequence (C CRS), and total − fractions of red sequence galaxies FT in HCS and WINGS. The uncertainties on the ellipticity and (C CRS) correspond to the boundaries of the 68% conﬁdence interval of the distribution of− each quantity. The uncertainties on the total morphological fractions correspond to the boundaries of the binomial 68% conﬁdence intervals estimated as in D’Agostini (2004) and Cameron (2011). The median values of (C CRS) and ellipticity of WINGS take into account the incompleteness of the spectroscopic− sample. The median (C CRS) of HCS is corrected for background contamination. −

Morphological Ellipticity (C CRS) FT Ellipticity (C CRS) FT Type (HCS) (HCS)− (HCS) (WINGS) (WINGS)− (WINGS)

+0.09 +0.10 +0.03 +0.15 +0.07 +0.004 Elliptical 0.18−0.11 0.00−0.10 0.56−0.03 0.17−0.12 0.00−0.06 0.379−0.004

Bulge-Dominated 0.38+0.10 0.01+0.09 0.28+0.02 0.47+0.16 0.02+0.07 0.509+0.004 −0.12 −0.09 −0.03 −0.20 − −0.06 −0.004

+0.15 +0.13 +0.015 +0.2 +0.06 +0.003 Early-type 0.45−0.16 0.02−0.11 0.111−0.020 0.6−0.3 0.03−0.08 0.109−0.003 Disc-Dominated − −

+0.011 +0.0005 Late-type 0.044−0.012 0.0026−0.0004 Disc-Dominated + Irr

classifications agree in all the cases. Blakeslee et al. (2006) visually classified galaxies in the cluster RX0152 using a scheme based on the Hubble T-types. After converting to our morphological scheme, we find that, of the 28 galaxies in common to the two samples, 18 (64%) were assigned the same type in the two works. However, if we limit ourselves to the early- vs late-type subdivision, the classifications agree for 25 galaxies (i.e. 89% of the common sample). We conclude that our morphological classification of the HCS red sequence galaxies is robust also when compared to other classifications of the same galaxies published in the literature. Hence, it constitutes a reliable representation of the morphological content of the red sequence in clusters at redshift 0.8 < z < 1.5.

5.2 Stellar Mass Estimate of Red Sequence Galaxies

We use the software lephare (Arnouts et al., 1999; Ilbert et al., 2006) to estimate stellar masses of red sequence galaxies in the HCS. lephare is based on a χ2 minimisation

0.4 ) ¯

/M 0.2 star,Delaye

0.0 ) − log( M ¯ /M

star 0.2 log( M

0.4

9.5 10.0 10.5 11.0 11.5 12.0 132 log(Mstar/M ) Chapter 5. Morphological Evolution ¯

5 number of galaxies

0 0 1 2 3 n4 5 6 7 8

Figure 5.4 Distribution of the Sérsicindex values, n, of elliptical (red histogram), bulge- dominated (orange histogram) and early-type disc-dominated (green filled histogram) galaxies in common with the sample of Delaye et al. (2014). The distributions for elliptical and S0 galaxies are statistically different (PKS = 0.002) although the mean values of n are consistent within 1σ (see 5.1.3). § algorithm which fits the available photometry to a library of template spectral energy distributions (SED) that can be set in the input configuration file. We used a set of templates drawn from the Bruzual & Charlot (2003) library with three different metallicities (0.2Z , 0.4Z , Z ), exponentially declining star formation rates, and Chabrier (2003) initial mass function (IMF), and we fitted them to the multiwavelength samples obtained for each cluster as discussed in 4.1.1. This set of templates was the same § adopted by Delaye et al. (2014) allowing us to make a direct comparison with that work. We estimated again the stellar masses in XMM1229 using this set-up and the multiband sample of Cerulo et al. (2014). For each cluster we set the redshift of all galaxies in the catalogues to the redshift of the cluster reported in Table 2.1. We defined the stellar mass error as half of the width of the 68% confidence contour of the χ2 surface in the 2- dimensional parameter space of galaxy stellar mass and age. The median fractional error on the estimates of the stellar mass of the galaxies in the morphological sample is 0.13. For the galaxies in common with the Delaye et al. (2014) sample we find a median difference ∆M = log(M∗/M ) log(MDelaye/M ) = 0.04 0.14, where the error corre- − − sponds to the 68% width of the ∆M distribution. This difference translates into a median ratio (M∗/MDelaye) 0.9. We attribute this residual difference to the different strate- ∼ gies adopted for photometry, as Delaye et al. (2014) used unconvolved total (SExtractor MAG AUTO) magnitudes whereas we use PSF matched 100 radius aperture magnitudes, and 5.2. Stellar Mass Estimate of Red Sequence Galaxies 133 to the different cosmologies adopted in the two works3. Furthermore, the different numbers of photometric bands used for some of the clusters (i.e. RDCS1252, XMMU2235, XMM1229) are also likely to contribute to ∆M.

In order to compare clusters at diﬀerent redshifts we built a mass-complete sample and estimated the stellar mass limit following two methods. The ﬁrst method, described in Pozzetti et al. (2010), involves estimating the mass of a galaxy, at a certain redshift, under the assumption that it is observed at the magnitude limit of the survey and that the mass-to-light ratio of galaxies is constant at all luminosities. This leads, for a galaxy of stellar mass M∗, observed at magnitude z850 in the ACS F850LP band, to the equation:

log(Mlim/M ) = log(M∗/M ) 0.4(zlim z ) (5.1) − − 850 where Mlim is the limiting mass and zlim the limiting magnitude of the morphological sample, i.e. zlim = 24.0 mag. We compute Mlim for the 20% faintest galaxies in the morphological sample of each cluster and take the upper 95% boundary of the Mlim distribution as the mass completeness limit of our sample. By restricting the calculation of Mlim to the 20% faintest galaxies of the morphological sample, only galaxies with mass- to-light ratios typical of the faint end of the morphological sample are considered, and the inclusion of the most massive galaxies is avoided. We applied Equation 5.1 to the spectroscopically conﬁrmed members of each HCS cluster and obtained log(Mlim/M ) = 10.6 at z = 1.46, the redshift of XMMXCS2215, which is the most distant cluster in the HCS.

The second method consists in measuring the stellar mass of a hypothetical object having a spectral energy distribution equal to a model synthetic SED with solar metallicity, formation redshift zf = 4.75, Salpeter (1955) IMF, and exponentially declining star formation rate with e-folding time τ = 1 Gyr. Assuming that the object has z850 = 24.0 mag, and that the galaxy mass-to-light ratio is constant at all luminosities, we obtain from

Equation 5.1 log(Mlim/M ) = 10.8 at z = 1.45. This value is close, but not equal, to that obtained with the Pozzetti et al. (2010) method, and therefore we adopt the mean of the two values, log(Mlim/M ) = 10.7, as the mass limit for the morphological analysis in the following sections. This value, which is plotted in Figure 5.5 as a vertical dashed line, is close to the mass limit found by Delaye et al. (2014) (log(Mlim/M ) = 10.8).

3 −1 −1 Delaye et al. (2014) adopted a ΛCDM cosmology with H0 = 70 km · s · Mpc 134 Chapter 5. Morphological Evolution

5.3 The low-redshift Comparison Sample

In order to study the morphological evolution of red sequence galaxies, we built a comparison sample of morphologically classiﬁed galaxies from WINGS-SPE, the spectroscopic follow-up of the WINGS survey (Cava et al., 2009).

Morphologies in this sample are presented in Fasano et al. (2012). Galaxies were clas- siﬁed, using a neural-network approach, in 18 types approximately corresponding to the Hubble T-types. We converted the WINGS morphologies to our morphological scheme as 14 illustrated in Table 3.5 and selected only the clusters with masses MDM > 5 10 M . × In fact, as discussed in Section 4.3.2, this is the mass predicted by simulations of structure formation in a ΛCDM universe for the descendant of the least massive HCS cluster (Fakhouri et al., 2010; Chiang et al., 2013).

We ﬁtted a straight line to the rest-frame (B V )AB vs VAB red sequence of the − composite WINGS sample considering only galaxies within 0.54 R from the centre of × 200 each WINGS cluster. This ensured that the same physical projected region was considered at all redshifts. Then we selected all the galaxies with VAB,obs < 18.0 mag, which is the magnitude down to which the spectroscopic survey is 50% complete.

Stellar masses in WINGS-SPE were estimated by Fritz et al. (2011) who adopted a model fitting approach in which all the available optical and near-infrared photometry, and the spectra, were used to estimate stellar masses and stellar ages of cluster members. A Salpeter (1955) IMF was assumed to fit the simple stellar population (SSP) models to the WINGS photometry and spectra. The stellar mass of the WINGS galaxies is defined as the total mass of stars, both in their nuclear-burning phase and in the remnants (white dwarfs, neutron stars and black holes). These values of the stellar masses were corrected to account for radial colour gradients by adding an empirical correction term provided in the WINGS catalogues and defined in Equation 3 of Fritz et al. (2011). The average error on the estimates of the total stellar masses in WINGS is 0.2 dex (Moretti et al., 2014).

Unlike Cerulo et al. (2014), we do not use here the MORPHS sample to compare the HCS and WINGS with clusters at 0.3 < z < 0.6. The selection for the MORPHS spectroscopic follow-up was indeed aimed at the study of the Butcher-Oemler eﬀect and therefore it privileged star-forming galaxies (Poggianti et al., 1999). This may result in overestimated fractions of disc-dominated galaxies along the red sequence. 5.4. Results 135

5.4 Results

The present section discusses the measurements of the morphological fractions and the statistical estimate of ﬁeld contamination in each HCS cluster sample.

5.4.1 Statistical Background Subtraction

Field contamination was statistically estimated after building control fields of morphologically classified galaxies in GOODS. For each cluster we selected random GOODS galaxies spanning the same colour range of the (i z ) vs z cluster red sequence, and with 775 − 850 850 z850 < 24.0 mag. This yielded 9 control samples including 100-400 galaxies which we classified adopting the same procedure described in Section 5.1.1.

Then we divided the z850 magnitude range of the cluster and control morphological samples into 0.5 mag bins, and in each bin i we measured the expected fraction of ﬁeld interlopers as: Nfield,i A Rcorr Pfield,i = × × (5.2) NRS,i

th where Nfield,i is the number of classified GOODS galaxies in the i magnitude bin, NRS,i is the number of classified cluster red sequence galaxies in the ith magnitude bin, A is the ratio between the cluster and GOODS areas, and Rcorr is the ratio between the total number of GOODS galaxies in the ith magnitude bin (i.e. classified and not classified) and

Nfield,i. Spectroscopic interlopers were already rejected from the morphological sample, and not classiﬁed, therefore they are not considered in Equation (5.2)

The background corrected number counts Nbin of galaxies of each morphological type is: m X Nbin = Ngal,i (1 Pfield,i) (5.3) × − i=1 th where Ngal,i is the i galaxy in the bin and m is the total number of galaxies in the bin.

5.4.2 Morphological Fractions

The background-corrected morphological fractions of red sequence galaxies, as a function of absolute magnitude and stellar mass, were estimated as in Equation 3.4 by measuring the ratio in each magnitude (or stellar mass) bin i:

NT,i FT,i = (5.4) Ntot,i 136 Chapter 5. Morphological Evolution where FT,i, is the fraction of galaxies of type T , NT,i is the number of galaxies of type T , and Ntot,i is the total number of galaxies. We adopted magnitude bins of 0.5 mag and mass bins of 0.5 dex to study the trends of the red sequence morphological fractions as a function of absolute magnitude and stellar mass, respectively. Following Cerulo et al. (2014), we used V-band AB absolute magnitudes to study morphology as a function of luminosity. We used Equation 4.6 to convert galaxy observed magnitudes to rest-frame V-band absolute magnitudes and then we passively evolved the rest-frame magnitudes to z = 0. Rest-frame VAB magnitudes for the WINGS galaxies were obtained by using the distance moduli provided in the WINGS catalogues and the k- corrections of Poggianti (1997). The red sequence morphological fractions as a function of absolute magnitude and galaxy stellar mass are shown in Figure 5.5 and will be discussed in Section 5.5. The uncertainties on the morphological fractions plotted in Figure 5.5 correspond to the binomial errors estimated as the 68% Bayesian confidence intervals of the beta distribution. This method is the same adopted for XMM1229 in Chapter 3 and is extensively discussed in D’Agostini (2004), Andreon et al. (2006) and Cameron (2011). These authors show that such a posteriori approach is particularly accurate with small samples and it also allows a reliable treatment of the extreme cases FT,i = 0.0 and FT,i = 1.0. In fact, in order to account for the finite probability of observing a galaxy of a certain morphological type in a magnitude or mass bin, the median of the beta distribution is used as the best estimate of the morphological fraction. The uncertainties are then defined as the difference between the median and the upper and lower bounds of the 68% confidence interval of the distribution.

5.5 Discussion: Morphological Transformations in Galaxy Clusters

The left-hand panel of Fig. 5.5 shows the morphological fractions of red sequence galaxies as a function of VAB absolute magnitude in the HCS (top) and WINGS (bottom) composite red sequences. The right-hand panels of Figure 5.5 show the plots of the morphological fractions as a function of galaxy stellar mass in HCS (top) and WINGS (bottom). It can be seen that the HCS stellar mass completeness limit log(Mlim/M ) = 10.7 results in a shallow sample and allows the study of the trends of morphological fractions only within a narrow 0.8 dex mass range. In the present discussion we will consider the trends in both the mass selected and the magnitude selected samples with the caveat that the results at 5.5. Morphological Transformations in Galaxy Clusters 137

HAWK-I Cluster Survey (HCS) (0.8

0.8 0.8

T 0.6 0.6 T F F

0.4 0.4

0.2 0.2

0.0 0.0

V log(Mstar/M ) AB ¯ 1.0 WINGS (0.04

0.8 0.8

T 0.6 0.6 T F F

0.4 0.4

0.2 0.2

0.0 0.0 24 23 22 21 20 19 18 9.5 10.0 10.5 11.0 11.5 12.0 12.5 VAB log(M /M ) ∗ ¯ (a) (b)

Figure 5.5 Left: Background corrected morphological fractions as a function of VAB absolute magnitude along the cluster red sequence in HCS (top panel) and WINGS (bottom panel). Right: Background corrected morphological fractions as a function of stellar mass along the cluster red sequence in HCS (top panel) and WINGS (bottom panel). The vertical dotted lines at log(M∗/M ) = 10.7 and log(M∗/M ) = 11.5 represent the stellar mass limit of the HCS morphological sample and the maximum stellar mass of HCS red sequence galaxies, respectively. The plots show that elliptical galaxies are the dominant morphological class in the HCS clusters at all luminosities and masses while the red sequence of the WINGS clusters is dominated by ellipticals at VAB < 21.0 mag (log(M∗/M ) > 11.5) − and by S0s at VAB > 21.0 mag (log(M∗/M ) < 11.5) − 138 Chapter 5. Morphological Evolution masses log(M∗/M ) < 10.7 in the latter sample are valid only at z < 1.46. The comparison between HCS and WINGS shows, in agreement with Mei et al. (2009), that the cluster red sequence was dominated by early-type galaxies already at z 1. How- ∼ ever, while the HCS red sequence is dominated by elliptical galaxies at all stellar masses and luminosities, it can be seen that the fraction of S0 galaxies FS0 in WINGS becomes higher than the fraction of elliptical galaxies FE at magnitudes VAB > 21.0 mag, and − stellar masses log(M∗/M ) < 11.3. We also note an upturn of FE in HCS at magnitudes

VAB > 19.5 mag. Since we pointed out in Section 5.1.3 that the contamination from − misclassified face-on S0 galaxies should amount to 5% at these magnitudes, the upturn ∼ is likely to be real and not a selection effect. The upturn is also visible, although less pronounced, at stellar masses log(M∗/M ) < 10.5 in the magnitude limited morphological sample. Table 5.2 shows the values of the total morphological fractions, i.e. measured over the entire VAB absolute magnitude range covered by the red sequence in HCS and WINGS. Our results show that the increase in the fraction of S0 galaxies at lower redshifts, and the corresponding decrease in the fraction of elliptical galaxies, are statistically significant. In fact, there is a > 5σ difference between the total fractions of these two morphological types in HCS and WINGS. The fractions of early-type disc-dominated, and late-type disc-dominated+Irr galaxies,

FD and Flate, respectively, are low across the entire absolute magnitude and stellar mass ranges spanned by the HCS and WINGS. Late-type galaxies constitute approximately 10% of the red sequence population in both samples. The fraction of bulge-dominated galaxies decreases at faint magnitudes after reaching a maximum at VAB 20.0 mag. A similar trend can be observed at log(M∗/M ) < 10.8 ∼ − in the magnitude limited sample. This result, and the predominance of S0 galaxies in the WINGS red sequence, suggest that the evolutionary paths followed by elliptical and S0 galaxies are diﬀerent and that, in particular, the latter may be the result of morphological transformation of quenched spirals. These cluster members, which were in the blue cloud while they were forming stars, joined the red sequence after they had their star formation quenched. This hypothesis is supported by two observations, namely the quenching of star formation being more eﬃcient at higher stellar masses - commonly known as downsizing (Cowie et al., 1996) - regardless of local environment (Peng et al., 2010b), and the evidence for younger stellar populations (0.5-1.0 Gyr) in red sequence S0 galaxies, with respect to ellipticals, in clusters at redshifts 0.8 < z < 1.5 (Tran et al. 2007; Mei et al. 2009, see also Chapter 6). In Chapter 4 we argued that galaxy clusters are built up through accretions of groups 5.5. Morphological Transformations in Galaxy Clusters 139 on to a main massive protocluster. A fraction of the group galaxies is already quenched at the time of accretion on the protocluster because of the high local densities of groups. However, the galaxy densities in the satellite groups are lower than in the protocluster and, as a result, star formation quenching happens, on average, over longer time-scales (Rettura et al. 2011; Muzzin et al. 2012; Taranu et al. 2014; Bahe & McCarthy 2014, Kovaˇcet al. 2010). In particular, at the time of accretion, the population of red sequence galaxies in groups will consist mainly of group centrals, which can be either elliptical or S0 galaxies.

This scenario can explain the peak observed in the fraction of S0 galaxies at VAB =

20.0 mag and log(M∗/M ) < 10.8 because not all the central galaxies of the accreted − groups are ellipticals. Some groups, and particularly the least dense, may have a S0 or even a spiral galaxy in their centres and these galaxies may quench their star formation as a consequence of bulge growth induced by the accretion of dwarf galaxies (Somerville et al., 2008). This process, known as morphological quenching (Martig et al., 2009) is, however, slow (> 4 Gyr) and would account only for a small fraction of the massive S0 galaxies observed on the HCS red sequence (log(M∗/M ) > 10.7), namely only those accreted with the satellite groups. In order to test this scenario, observations of assembling protoclusters at redshift z > 2 are necessary (Cooke et al., 2014; Yuan et al., 2014).

Our proposed morphological evolution is consistent with the non-evolving fraction of elliptical galaxies reported in several works studying clusters at z < 1.2 (Dressler et al., 1997; Fasano et al., 2001; Postman et al., 2005; Desai et al., 2007; Mei et al., 2012). In fact, in our scenario, elliptical galaxies are made up of an in situ population which formed in the central protocluster, and of an accreted population which was formed in the centres of satellite groups. These galaxies underwent a rapid evolution driven by mergers whose occurrence was facilitated in the central regions of protoclusters and groups where relative velocities were low due to the low densities of these systems. These mergers induced rapid starbursts, as those observable in submillimetre galaxies at z = 3 4, which rapidly − exhausted the galaxy gas reservoirs, and at the same time caused the transformation of the initial spiral galaxies into ellipticals. These galaxies evolved passively with time or underwent dry merger events which increased their masses and sizes without causing the onset of new signiﬁcant star-formation episodes.

Mei et al. (2009) investigated the scatter of the red sequence for elliptical and S0 galaxies in clusters at 0.8 < z < 1.4 and found that, at magnitudes MB,V ega < 21.0 − mag, the stellar populations of S0 galaxies are 0.5 Gyr younger than those of elliptical ∼ galaxies. At MB,V ega 21.0 mag, stellar populations have similar ages in galaxies of ≥ − 140 Chapter 5. Morphological Evolution both morphological types. This result can be explained with the morphological evolution proposed here if one considers that a fraction of the massive S0s is the result of the transformation of quenched spiral galaxies in the central protocluster, and the remaining is accreted as the morphologically-quenched centrals of the lowest-density groups. In both cases stellar populations would be younger because the galaxies have been recently quenched. Low-luminosity ellipticals, on the other hand, would be accreted at later times as the centrals of the densest infalling groups. In this case the younger ages of these galaxies are the result of the longer timescales of star-formation quenching in the satellite groups with respect to the central protocluster. Finally, the population of low-luminosity S0s would be contributed by low-mass satellites in the protocluster and in the groups which, due to their mass and their residing in the outskirts, quenched their star formation at later epochs before being transformed into S0 galaxies.

Hence, our results suggest that elliptical galaxies in the HCS are made up of two populations, namely a primordial population of giant galaxies with stellar masses log(M∗/M ) > 10.8 formed in the main protocluster, and a population of dwarf galaxies at masses log(M∗/M ) < 10.8 which was partly accreted and partly formed in low-density regions of the central protocluster. A similar picture was proposed by Poggianti et al. (2006) to interpret the evolution of the fraction of star-forming galaxies in clusters at z < 0.8.

Our morphological evolution can also be connected with the observation that the fraction of spiral galaxies on the cluster red sequence reaches a maximum at z 0.6 and then ∼ decreases at z < 0.4, as shown in Sánchez-Blázquezet al. (2009). Vulcani et al. (2011a) found that the fraction of late-type galaxies in the EDisCS clusters is higher compared to WINGS and, interestingly, that the fractions of EDisCS late-type galaxies and WINGS S0 galaxies as a function of stellar mass are similar down to log(M∗/M ) = 10.3. The latter also supports the idea of an evolutionary link between S0 and spiral galaxies. Although a direct comparison with Vulcani et al. (2011a) is not possible because the authors investigated all cluster members without separating into blue cloud and red sequence, those results, and the findings of Sánchez-Blázquezet al. (2009), suggest that during the epochs corresponding to 0.4 < z < 0.8 most of the recently quenched late-type galaxies joined the red sequence to be transformed into S0 galaxies at z < 0.4. In this scenario the cessation of star formation is faster than the morphological transformation from spiral to S0, which would need at least 4 Gyr to occur (see e.g. Bekki & Couch 2011).

Quenching mechanisms aﬀecting galaxies in clusters include ram-pressure stripping, which has been shown to be the main responsible for star-formation quenching in spiral galaxies in the Virgo cluster (Gavazzi et al., 2013), strangulation, that is, the gradual 5.5. Morphological Transformations in Galaxy Clusters 141 removal of gas from the galaxy halo, which has been shown to have time-scales ( 3.0 ∼ Gyr) consistent with the build-up of the red sequence in simulations of galaxy clusters (Taranu et al., 2014), and bulge growth induced by tidal interactions with neighbouring galaxies (Bekki & Couch, 2011). All these mechanisms lead to the accelerated depletion of the gas reservoirs that are necessary to form new stars in galaxies. Spiral galaxies, therefore, join the red sequence keeping their morphology (red spirals, Wolf et al. 2009) and then, due to the lack of gas in the disc, which is necessary to suppress random motions and sustain the spiral structure, they undergo morphological transformations that turn them into S0 galaxies. Alternatively, spiral arms can be disrupted by tidal interactions with neighbouring galaxies, which perturb the disc during a close passage (Moore et al., 1998). We think that these two mechanisms are competing in the dense environment of galaxy clusters.

We note that the results presented in this section do not agree with those shown in Chapter 3 for the morphological fractions in XMM1229. In fact, in that cluster we found that, while the fraction of elliptical galaxies remained approximately constant along the red sequence, the fraction of S0 galaxies reached a maximum at VAB 20.8 mag and ∼ − then decreased. Correspondingly, the fraction of disc-dominated galaxies increased at

VAB > 20.3 mag. This suggested that the faint red disc-dominated galaxies could be the − progenitors of the S0 galaxies dominating the red sequence of the WINGS clusters at the same magnitudes.

The trends observed in XMM1229 may be peculiar of that cluster, although we stress that the magnitude limit of the morphological sample z850 = 24.0 mag produces a variable sampling of each cluster’s red sequence. For example, as shown in Figure 5.1, in the clusters RCS2319 and RDCS1252, the magnitude selection does not allow us to investigate the faint end of the red sequence down to the 90% completeness limit of the sample (diagonal dashed line), while in other clusters, such as XMM1229 or RCS2345, we are able to classify a larger fraction of the cluster red sequence. Therefore, we do not exclude that with deeper images, enabling us to sample the red sequence of each cluster at least one magnitude fainter than the current limit, we may detect a similar upturn in the fraction of late-type galaxies.

Table 5.2 shows that HCS and WINGS late-type galaxies, that is, the entire population of disc-dominated and irregular galaxies, tend to be located at the blue side of the red sequence. Their colours, C, measured with respect to the red sequence, CRS, are indeed +0.13 +0.06 (C CRS) = 0.02 and (C CRS) = 0.03 for HCS and WINGS, respectively. − − −0.11 − − −0.08 These values, which correspond to the median and the boundaries of the 68% conﬁdence intervals of the distributions, are consistent with the median values of (C CRS) of elliptical − 142 Chapter 5. Morphological Evolution

64 ellitpical

56 S0 Late-type 48

40 N 32

0.5 0.4 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 (C C ) − RS

Figure 5.6 Distributions of galaxy colour measured with respect to the red sequence, (C CRS), for galaxies of different morphological types in HCS. Early-type disc-dominated, late-type− disc-dominated and irregular galaxies are grouped into one class of late-type galaxies (blue histogram). Number counts are corrected for field contamination using Equation 5.2. and S0 galaxies in both samples4. Figure 5.6 also shows that the three distributions cover the same colour range in HCS. However, this result suggests that, in agreement with Tran et al. (2007) (see also Figure 3.8), late-type galaxies may have younger stellar ages, compared to early-type galaxies. The bright end of the red sequence in HCS and WINGS, where most of the brightest cluster galaxies (BCG) reside in both samples, is dominated by elliptical galaxies which 12.4 in the low-z sample can be as massive as 10 M . We note that the difference in stellar mass is more pronounced than that in absolute magnitude and, in particular, that the

HCS is not populated by galaxies with masses log(M∗/M ) > 11.5. This difference is significantly higher than the values expected from the mass-growth factors estimated for BCGs at z < 1 ( 2-2.5 Lidman et al. 2012; Ascaso et al. 2014), indicating that bright red sequence galaxies in WINGS are exceptionally massive. However, we point out that the range log(M∗/M ) > 11.5 is poorly populated in WINGS and corresponds to about 1% of the full WINGS-SPE sample. Interestingly, Cerulo et al. (2014) reported bluer (V K) − colours for bright red sequence galaxies in WINGS, compared to those implied by the spectral evolution models of Bruzual & Charlot (2003) and by the observations of Bower et al. (1992) in clusters at z = 0. In order to prove the morphological evolution scenario proposed in this section, a study of the stellar populations of red sequence galaxies with different morphologies and

4 The median values of (C − CRS ) correspond to the weighted medians of the distributions. This estimator allows one to take into account, on one hand, the ﬁeld contamination in the HCS sample, and on the other hand, the incompleteness of WINGS-SPE. 5.5. Morphological Transformations in Galaxy Clusters 143 masses is necessary. Deeper images, enabling us to probe the red sequence down to fainter magnitudes than those allowed by the ACS F850LP images, are also necessary to investigate the morphological content of the faint end of the red sequence. While the latter is part of the future plans of the HCS collaboration that are discussed in Chapter 7, the ﬁrst results from the ongoing analysis of the spectral properties in three HCS clusters at z 1 are shown in Chapter 6 of this thesis. ∼

6 The Spectral Properties of Red Sequence Cluster Galaxies

This chapter presents the early results from the spectroscopic follow-up program of HCS clusters at the Gemini and Keck telescopes. Spectra have been obtained for red sequence galaxies in the clusters RCS 2319.8+0038 (RCS2319, z = 0.91), XMMU J1229+0151 (XMM1229, z = 0.98), and RCS 0220.9-0333 (RCS0220, z = 1.03) These spectra have been co-added as a function of galaxy morphological type and luminosity, and the links between these two quantities and the stellar populations of cluster galaxies are discussed. The results presented in this chapter are preliminary due to the limited numbers of galaxies observed spectroscopically in each cluster and the analysis of the spectra is purely qualitative at this stage. Much more extensive spectroscopy of the HCS clusters will be required before more robust and deﬁnitive conclusions can be drawn. This is among the future plans of the collaboration.

The chapter is divided into three sections: Section 6.1 provides a brief introduction to spectroscopic-based stellar populations analysis and explains its importance in studies of galaxy evolution. Section 6.2 presents our ﬁrst such stellar population analysis results for the red sequence galaxies in the HCS. These are then discussed in Section 6.3 together with the future directions that will be undertaken in this study.

Throughout the chapter we adopt a ΛCDM cosmology with ΩΛ = 0.73, Ωm = 0.27, and H = 71.0 km s−1 Mpc−1 as done in all the previous chapters. Unless otherwise 0 · · stated, all magnitudes are quoted in the AB system (Oke, 1974).

145 146 Chapter 6. Spectral Properties

RCS2319_205 RCS2319_219 RCS2319_106 RCS2319_210

elliptical elliptical elliptical elliptical

RCS2319_177 RCS2319_201 RCS2319_174

disc-domin. elliptical bulge-domin. (early)

Figure 6.1 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with LRIS in the cluster RCS 2319.8+0038 (z = 0.91). Each image is 81 pixels 00 on each side, corresponding to 4.05 . The ID, coordinates, z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.1 below. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table.

Table 6.1 ID α δ (F 850LP )AB Morphology Redshift (J2000) (J2000) (MAG AUTO)

RCS2319 205 23 : 19 : 53.4 +00 : 38 : 13.5 19.91 0.09 Elliptical 0.901 RCS2319 219 23 : 19 : 53.4 +00 : 38 : 14.2 21.13 0.08 Elliptical 0.897 RCS2319 106 23 : 19 : 56.0 +00 : 37 : 35.4 21.43 0.07 Elliptical 0.904 RCS2319 210 23 : 19 : 52.5 +00 : 38 : 10.8 21.50 0.07 Elliptical 0.901 RCS2319 177 23 : 19 : 55.1 +00 : 38 : 03.1 21.64 0.06 Elliptical 0.899 RCS2319 201 23 : 19 : 55.5 +00 : 38 : 08.3 21.81 0.06 S0 0.892 RCS2319 174 23 : 19 : 54.4 +00 : 38 : 04.4 21.12 0.08 disc- 0.905 dominated (I)

6.1 The Stellar Populations of Galaxies

The study of stellar populations is important in understanding the processes that drive the evolution of galaxies in diﬀerent environments. Quantities such as stellar age and metal- 6.1. The Stellar Populations of Galaxies 147

XMM1229_237 XMM1229_312 XMM1229_145 XMM1229_265

elliptical elliptical elliptical elliptical

Figure 6.2 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with GMOS-N in the cluster XMMU J1229+0151 (z = 0.98). Each image is 81 pixels 00 on each side, corresponding to 4.05 . The ID, coordinates, z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.2. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table.

Table 6.2 ID α δ (F 850LP )AB Morphology Redshift (J2000) (J2000) (MAG AUTO)

XMM1229 237 12 : 29 : 29.3 +01 : 51 : 21.8 21.23 0.11 Elliptical 0.974 XMM1229 312 12 : 29 : 28.7 +01 : 51 : 37.0 21.48 0.06 Elliptical 0.975 XMM1229 145 12 : 29 : 29.9 +01 : 50 : 46.3 21.90 0.09 Elliptical 0.985 XMM1229 265 12 : 29 : 28.9 +01 : 51 : 24.9 22.16 0.10 Elliptical 0.974

licity can be estimated from the properties of the spectral continuum or from the strength of the absorption and emission lines with higher accuracy than those obtained from the ﬁtting of synthetic SEDs to broad band photometry (Fritz et al., 2011). Spectroscopy provides more detailed stellar population information, allowing us to better tackle the important question of what paths galaxies of diﬀerent morphology and stellar masses follow in their evolution.

In studies of the red sequence as a function of redshift, the analysis of stellar populations is crucial for understanding the processes that led to the quenching of star formation and the consequent build-up of the red sequence. In fact, it is well-known that colours are related to both stellar age and metallicity of galaxies, and that (Worthey, 1994):

∂C ∂C ∂t 3 = (6.1) ∂Z ∂t ∂Z ' −2 148 Chapter 6. Spectral Properties

RCS0220_191 RCS0220_218 RCS0220_282 RCS0220_244

elliptical elliptical elliptical elliptical

RCS0220_311 RCS0220_377 RCS0220_307 RCS0220_182

elliptical elliptical elliptical bulge-domin.

RCS0220_319 RCS0220_365 RCS0220_306 RCS0220_353

bulge-domin. bulge-domin. bulge-domin. bulge-domin.

Figure 6.3 HST/ACS F850LP postage stamp images of the red sequence galaxies observed with GMOS-N in the cluster RCS 0220.9-0333 (z = 1.03). Each image is 81 pixels 00 on each side, corresponding to 4.05 . The ID, coordinates, z850 apparent magnitudes, morphological types, and redshifts are shown in Table 6.3. Galaxies are ordered by decreasing luminosity, within each morphological class, going from elliptical to early-type disc-dominated galaxies. The order in which the objects appear is the same in the ﬁgure and in the table. where C is the colour, t is the stellar age, and Z is the metallicity. Equation 6.1 shows that age and metallicity are anti-correlated: a given colour can be the result of either young stellar populations with high metallicities or old stellar populations with low metallicities. This eﬀect is known as the age-metallicity degeneracy. In order to break the age-metallicity degeneracy, evolutionary models of stellar populations have been built from spectral libraries (Worthey, 1994; Thomas et al., 2003, 2011), allowing ages and metallicities of galaxies to be derived from particular sets of absorption lines measured in galaxy spectra. Faber et al. (1985) introduced a set of spectral indices known as the Lick System, which has been the system used in most studies of galaxy stellar populations (e.g. Jones et al. 6.1. The Stellar Populations of Galaxies 149

Table 6.3 ID α δ (F 850LP )AB Morphology Redshift (J2000) (J2000) (MAG AUTO)

RCS0220 191 02 : 20 : 55.6 03 : 33 : 47.8 21.06 0.10 Elliptical 1.021 RCS0220 218 02 : 20 : 56.4 −03 : 33 : 32.4 21.20 0.12 Elliptical 1.028 RCS0220 282 02 : 20 : 55.7 −03 : 33 : 19.3 21.33 0.06 Elliptical 1.028 RCS0220 244 02 : 20 : 57.4 −03 : 33 : 30.3 22.10 0.10 Elliptical 1.017 RCS0220 311 02 : 20 : 55.9 −03 : 33 : 08.1 22.23 0.08 Elliptical 1.028 RCS0220 377 02 : 20 : 53.5 −03 : 32 : 49.0 22.33 0.16 Elliptical 1.034 RCS0220 307 02 : 20 : 55.6 −03 : 33 : 10.6 23.46 0.11 Elliptical 1.029 RCS0220 182 02 : 20 : 56.3 −03 : 33 : 55.9 21.48 0.09 S0 1.017 RCS0220 319 02 : 20 : 55.1 −03 : 33 : 05.3 21.7 0.2 S0 1.021 RCS0220 365 02 : 20 : 54.9 −03 : 32 : 52.3 21.95 0.10 S0 1.028 RCS0220 306 02 : 20 : 51.9 −03 : 33 : 09.6 22.21 0.11 S0 1.021 RCS0220 353 02 : 20 : 56.6 −03 : 32 : 57.3 22.57 0.09 S0 1.028 −

2000; Proctor & Sansom 2002; Thomas et al. 2005; Spolaor et al. 2010; Onodera et al. 2014). The system is based on a series of age- and metal-sensitive absorption features resulting from atomic and molecular transitions at rest-frame wavelengths 4000A˚ < λ < 6000A.˚

Worthey & Ottaviani (1997) showed that the Balmer absorption lines Hδ (λ = 4102A)˚ and Hγ (λ = 4341A)˚ are highly sensitive to galaxy stellar age, with little dependence on metallicity. On the other hand, features such as the iron lines C 4668 (λ 4668.0 A)˚ and 2 ∼ Fe5015 (λ 5015 A)˚ have high metallicity sensitivity with little dependence on stellar age ∼ (Worthey, 1994). At 0.8 < z < 1.5, the redshift range of the HCS clusters, these lines can be observed with current optical spectrographs on 8-10 m class telescopes. Thus we have undertaken a spectroscopic follow-up campaign of the HCS clusters at the Gemini-North and Keck telescopes with the aim of acquiring quality spectra of red sequence galaxies to study their stellar populations and derive their stellar ages and metallicities. We have observed 3 clusters, namely RCS 2319.8+0038 (RCS2319, z = 0.91), XMMU J1229+0151 (XMM1229, z = 0.98), and RCS 0220.9-0333 (RCS0220, z = 1.03). The targeted galaxies are shown in Figures 6.1, 6.2, and 6.3, and their coordinates, F850LP apparent magnitude, morphology, and redshift are summarised in Tables 6.1, 6.2, and 6.3. The taking and reduction of the spectra presented in this chapter, together with the measurements of the redshifts, are discussed in Section 2.3. 150 Chapter 6. Spectral Properties

6.2 The Average Red Sequence Spectra of HCS Galaxies at z 1 ∼ The measurement of useful Lick indices requires spectra with signal-to-noise ratios S/N & 10. At z 1, where bright red sequence galaxies have apparent magnitudes in the range ∼ 20.0 < z < 22.0, this requires exposures of 105 s on an 8m class telescope (Table 6.4). 850 ∼ In our Keck and Gemini observations the exposure time for each target was 2.5 hrs < texp < 3.0 hrs (9000-10800 s), which resulted in 4.0 < S/N < 8.0 in each individual galaxy spectrum. Some galaxies were observed in multiple masks and in these cases we achieved 9.0 < S/N < 10.0 in the region of the spectrum where the Lick indices fall. With these values of S/N, the estimates of galaxy age and metallicity with the Lick analysis would have large uncertainties and the results would not be reliable. Therefore, in order to increase the signal-to-noise ratio of the red sequence spectra, we combined them to form composite spectra for galaxies of different morphological types and luminosities. For this purpose we shifted all the spectra, in each cluster, to a common redshift, and we used the IRAF task odcombine to combine the individual spectra and obtain an average final spectrum. We repeated this exercise for galaxies of different morphological types and luminosity obtaining the composite spectra shown in Figures 6.5 and 6.6, respectively. We have plotted in Figure 6.4 the colour-magnitude diagrams of the three clusters RCS2319, XMM1229, and RCS0220 observed in our spectroscopic campaign. Spectro- scopically confirmed members of the 3 clusters are represented as magenta stars. We note that there is only one S0 and one early-type disc-dominated galaxy among the LRIS red sequence spectroscopically confirmed members of RCS2319, while we have only elliptical galaxies in XMM1229. The only cluster for which we have more than one galaxy for each morphological type is RCS0220, although no disc-dominated galaxies fall among the spectroscopically confirmed members of this cluster. RCS0220 is also the only cluster for which it is possible to study the spectra of red sequence galaxies as a function of luminosity. Figure 6.4 shows that there are 6 galaxies at z850 < 22.0 mag and 6 galaxies at z 22.0 mag. In the following sections we will therefore discuss the average properties 850 ≥ of elliptical galaxies in RCS2319, XMM1229, and RCS0220, while we will study the average properties of S0 galaxies only in RCS0220, comparing them with elliptical galaxies in the same cluster and in the other two clusters. For completeness, we also show the spectra of the only S0 and early-type disc-dominated galaxies in RCS2319 in Figure 6.7. We will also discuss the properties of bright and faint red sequence galaxies in RCS0220 comparing the composite spectra of galaxies at z < 22.0 mag and z 22.0 mag. 850 850 ≥ All the Lick absorption indices, together with the CaII K and H lines at λ = 3933.0A˚ 6.3. Discussion: The Properties of the Red Sequence Spectra 151 and λ = 3968.0A,˚ the MgI line at λ = 3838.3A,˚ and the [OII] emission line doublet at λ = 3727A,˚ are plotted in Figures 6.5-6.7 as vertical dashed lines intersecting the expected position of their central wavelengths at the redshifts of the clusters. Each line is also labelled according to its name. The characteristics of the composite spectra are summarised in Table 6.4. Table 6.4 shows that the composite spectra of red sequence galaxies with different morphologies all have S/N > 9, which allows them to be used for Lick indices analysis. However, the table also shows that signal-to-noise ratios of this size can only be achieved by combining at least 6-7 spectra of galaxies with z850 < 24.0 mag. As shown in Figures 6.5 and 6.6, we are successful in identifying absorption features in the composite spectra up to λ = 9500A,˚ while at longer wavelengths the emission lines due to transitions in the atmospheric OH− radical contaminate the spectra. As a result, any absorption line at these wavelengths cannot be detected. The results presented in Figures 6.5-6.7 and their implications in the context of the evolution of the cluster red sequence are qualitatively discussed in Section 6.3.

6.3 Discussion: The Properties of the Red Sequence Spectra

6.3.1 Spectral Properties as a Function of Galaxy Morphology

Figure 6.5 shows, from top to bottom, the composite spectra of red sequence galaxies with diﬀerent morphological types in RCS2319, XMM1229, and RCS0220. The spectra of the S0 galaxy and early-type disc-dominated galaxy observed in RCS2319 are plotted in Figure 6.7. Our data allow us to study the behaviour of the Balmer Hδ and Hγ lines with morphological type. We are also able to study Hβ in the composite elliptical spectra of RCS2319 and XMM1229. Absorption features generated by iron and calcium atomic transitions at rest-frame wavelengths 4000A˚< λ < 5000A˚ are also visible in the composite spectra, allowing the average metallicities of galaxies with diﬀerent morphological types to be estimated. The Hδ and Hγ lines are highly sensitive to the age of the stellar populations (Worthey & Ottaviani, 1997), and, in particular, Hδ is strongest (equivalent width EW (Hδ) > 4) for stellar populations in the age range 0.1 Gyr < t < 1.5 Gyr (Poggianti & Barbaro, 1997). In the absence of emission lines sensitive to ongoing star formation, such as the [OII] doublet1, which falls in the wavelength range of our spectra, a strong Hδ indicates

1The [OII] doublet is also a proxy for ongoing AGN activity in galaxies. 152 Chapter 6. Spectral Properties

1.6 1.4 RCS2319 (z=0.91) RCS0220 (z=1.03)

AB 1.2 1.0 0.8 0.6 0.4 0.2 ( F 775 W − 850 LP ) 0.0 0.2 20 22 24 26 20 22 24 26 F850LPAB (MAG AUTO) F850LPAB (MAG AUTO) 1.6 XMM1229 (z=0.98) 1.4 field interlopers

AB 1.2 1.0 new spectroscopic members 0.8 Elliptical 0.6 S0 0.4 0.2 Disc-domin. (I) ( F 775 W − 850 LP ) 0.0 0.2 20 22 24 26 F850LPAB (MAG AUTO)

Figure 6.4 Observed colour-magnitude diagrams of the clusters RCS 2319.8+0038 (z = 0.91, top-left), XMMU J1229+0151 (z = 0.98, bottom-left) and RCS 0220.9-0333 (z = 1.03, top-right). Magenta stars represent the new spectroscopically confirmed cluster members, black pentagons are spectroscopic interlopers found from the GMOS-N and LRIS spectra, red crosses are spectroscopically confirmed elliptical galaxies, orange diamonds are spectroscopically confirmed S0 galaxies, green triangles are spectroscopically confirmed early-type disc-dominated galaxies. The diagonal solid lines are the best-fit straight lines to the observed red sequence of each cluster, the diagonal dotted lines mark the boundaries of the red sequence estimated as in Section 4.2.1, the diagonal dashed lines are the 90% magnitude completeness limits of the three samples. The vertical dashed lines mark the magnitude limit of the morphological sample (z850 = 24 mag) and the boundary between bright and faint spectroscopic targets (z850 = 22.0 mag). Morphologically classified galaxies that were detected in the Gemini or Keck spectra are marked by one of the morphological symbols (cross, diamond or triangle) and a star or pentagon. Error bars on the objects observed with GMOS-N and LRIS correspond to the photometric uncertainties estimated as discussed in Section 4.1.1. that star formation has been recently quenched. More specifically, the last episode of star formation happened 1.5-2.0 Gyr before the cosmic epoch corresponding to the redshift of the galaxy (Couch & Sharples 1987, Poggianti et al. 1999). 6.3. Discussion: The Properties of the Red Sequence Spectra 153

RCS2319 (z =0.91) Ellipticals 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5[OII] MgI Hδ Mg2

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A) XMM1229 (z =0.98) Ellipticals 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5 [OII] MgI Hδ

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A) RCS0220 (z =1.03) Ellipticals 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5 [OII] MgI Hδ

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A) RCS0220 (z =1.03) S0s 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5 [OII] MgI Hδ

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A)

Figure 6.5 Composite spectra of morphologically classiﬁed red sequence galaxies in HCS clusters at z 1. From top to bottom: elliptical galaxies in RCS 2319.8+0038 (z = 0.91), elliptical galaxies∼ in XMMU J1229+0151 (z = 0.98), elliptical galaxies in RCS 0220.9-0333 (z = 1.03), S0 galaxies in RCS 0220.9-0333 (z = 1.03). Galaxies were morphologically classiﬁed as discussed in Chapter 5. Spectra are in the observer reference frame. Age- and metal-sensitive absorption features are marked with vertical dashed lines drawn at the central wavelength of each feature at the redshifts of the clusters. Unlike elliptical galaxies, the CaII H line in the S0 composite spectrum of RCS 0220.9-0333 is more prominent than the CaII K line, suggesting a larger contribution from H in the observed CaII H feature. The Hδ line also appears more prominent than in elliptical galaxies in the same cluster. These two observations suggest that S0 galaxies host younger stellar populations than elliptical galaxies. 154 Chapter 6. Spectral Properties

Elliptical galaxies in all three clusters do not show any emission feature and the Hδ and Hγ are weak, as expected from old stellar populations. We also note that the CaII K absorption line at 3933.0 A˚ is stronger than the CaII H line at 3968.0 A.˚ The S0 galaxies in RCS0220 do not present any emission line and exhibit a slightly more prominent Hδ absorption line than the elliptical galaxies, whereas the Hγ line in the elliptical and S0 galaxies is similar. However, unlike the elliptical galaxies, the CaII H line is more prominent than the CaII K line. As discussed in Rose (1985), the observed CaII H feature is a combination of the CaII transition at λ = 3968.0 A˚ and the Balmer H absorption line at λ = 3970.0 A.˚ As discussed in Leonardi & Rose (1996), the ratio (CaII H + H)/(CaII K) increases in the presence of young stars because of the weakening of the Ca lines and the H becoming stronger. As a result, the observed CaII H line appears stronger than the CaII K line. This result suggests that red sequence S0 galaxies in RCS0220 host a higher fraction of young stars than elliptical galaxies and are therefore younger.

Figure 6.7 shows that the CaII H line in the RCS2319 S0 galaxy spectrum is stronger than the CaII K line, while in the spectrum of the disc-dominated galaxy this trend is similar to that of elliptical galaxies. However, we also note that the Hδ absorption line is stronger in the disc-dominated galaxy than in the S0 galaxy. Thus the spectra of these two galaxies suggest that in both cases there is the presence of young stellar populations, supporting the notion that star formation has recently ceased. While this is interesting, we cannot draw any general conclusion on the stellar population properties of red sequence S0 and late-type galaxies. In order to investigate the properties of these galaxies we need to acquire larger samples of spectra.

Our results on the diﬀerent qualitative trends of the CaII lines in elliptical and S0 galaxies are in agreement with Tran et al. (2007), who found that red sequence S0 galaxies in the cluster MS 1054-03 at z = 0.83 are younger than ellipticals. It is also in agreement with the conclusions of Demarco et al. (2010), who showed that S0 galaxies in the cluster RX J0152.7-1357 at z = 0.84 had their last episode of star formation 0.6 Gyr later than elliptical galaxies.

The results presented in this section, together with the increase in the fraction of red sequence S0 galaxies with redshift discussed in Chapter 5, support the scenario in which elliptical and S0 galaxies follow diﬀerent evolutionary paths with the latter being the descendants of quenched spiral galaxies. Interestingly, in Figure 3.8 we showed that the FORS2 composite spectrum of early-type disc-dominated galaxies on the XMM1229 red sequence has stronger Hδ than S0 and elliptical galaxies. In this spectrum the CaII H + H absorption line is also more prominent than the CaII K line. This suggests, 6.3. Discussion: The Properties of the Red Sequence Spectra 155 in agreement with Tran et al. (2007) and Demarco et al. (2010), that late-type galaxies represent the youngest component of the cluster red sequence. We stress that the results just discussed come from a purely qualitative analysis of the composite spectra of red sequence HCS galaxies with diﬀerent morphologies. Therefore, their interpretation in the context of the evolution of the red sequence is not based on any measurement or statistical analysis. The work presented here is ongoing and the next step will consist in the measurements of the absorption features of the composite spectra and in the estimates of stellar age and metallicity. This will be part of a forthcoming work while new observations will be proposed for the Keck and Gemini telescopes to acquire more spectra also in other HCS clusters.

Table 6.4. Details of the composite HCS red sequence spectra. Ng is the number of galaxies in each composite spectrum. texp is the sum of the exposure times of all the spectra. The signal-to-noise ratio S/N is estimated at wavelengths longer than the position of the 4000 A˚ break, where most absorption features fall. Bright and faint composite spectra in RCS0220 correspond to all red sequence galaxies with z850 < 22.0 mag and z850 22.0 mag, respectively. Neff is the number of co-added spectra in each subsample taking≥ into account multiple exposures.

ID Instrument Ng Neff texp (s) S/N

RCS2319 elliptical Keck/LRIS 6 6 45000.0 10.0 XMM1229 elliptical GMOS-N 4 14 151200.0 14.0 RCS0220 elliptical GMOS-N 7 10 108000.0 10.0 RCS0220 S0 GMOS-N 5 8 86400.0 9.0 RCS0220 bright GMOS-N 6 11 118800.0 10.0 RCS0220 faint GMOS-N 6 7 111600.0 9.0

6.3.2 Spectral Properties as a Function of Galaxy Luminosity

Figure 6.6 shows the composite spectra of red sequence galaxies with z850 < 22.0 mag (bright, top panel) and z 22.0 mag (faint, bottom panel) in RCS0220. The two 850 ≥ subsamples have similar morphological compositions. The bright subsample has 3 elliptical and 3 S0 galaxies, while the faint subsample has 4 elliptical and 2 S0 galaxies. The composite spectra of bright and faint red sequence galaxies are similar. Both composite spectra have weak Balmer absorption features and the CaII K line is stronger than the CaII H + H line, as observed in the composite spectra of elliptical galaxies. Interestingly, we note an absorption line that appears between MgI and the K line in 156 Chapter 6. Spectral Properties

RCS0220 (z =1.03) F850LP <22.0 2.0 AB H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5 [OII] MgI Hδ

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A) RCS0220 (z =1.03) F850LP 22.0 2.0 AB ≥ H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5 [OII] MgI Hδ

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A)

Figure 6.6 Composite spectra of red sequence galaxies in RCS 0220.9-0333 (z = 1.03) at diﬀerent apparent magnitudes. Top panel: red sequence galaxies with z850 < 22.0 mag; bottom panel: red sequence galaxies with z850 22.0 mag. The Hδ line appears more prominent in fainter galaxies than in brighter galaxies,≥ suggesting that star formation was quenched more recently in less massive galaxies. The H6 index (rest-frame λ 3900.0 A)˚ seems also detectable in faint galaxies. ∼ the composite spectrum of faint galaxies. The region of the spectrum at λ 3900 A˚ ' corresponds to the higher-order Balmer transitions Hη (3835.0 A)˚ and Hζ (3889.0 A).˚ These transitions cannot be individually resolved with the grating and ﬁlter combination (R400/OG515) used in the GMOS-N observations and, therefore, they are blended in one single absorption feature at λ 3900 A˚ . As discussed in Demarco et al. (2010) and Nantais ∼ et al. (2013b), this feature is sensitive to the presence of young stellar populations and can be studied as the absorption index H6. Its equivalent width, EW(H6), is correlated with the presence of young stellar populations in galaxies.

Thus the bottom panel of Figure 6.6 suggests that faint galaxies may host a higher fraction of young stellar populations with respect to bright galaxies. It ca also be seen that the Hδ line in faint galaxies is stronger than in bright galaxies. This would suggest that faint galaxies have a lower stellar age than bright galaxies, in agreement with Mei et al. (2009). However, only through the measurement of the absorption indices it can be assessed whether faint galaxies are younger than bright galaxies on the red sequence.

We note that the faint galaxies do not present the same shape of the CaII H and K lines of S0 galaxies, where a larger contribution of H to the observed CaII H line was 6.3. Discussion: The Properties of the Red Sequence Spectra 157 observed. This is the result of the fact that, in the faint subsample, the number of elliptical galaxies is larger than that of S0 galaxies and, therefore, this subsample is dominated by elliptical galaxies. As done in Section 6.3.1, we stress here that these results are merely qualitative, and their signiﬁcance can only be tested by measuring the age and metal-sensitive absorption indices. This work is ongoing and will be part of a forthcoming paper, while new observations are needed to investigate spectral properties at z 22.0 mag also in other 850 ≥ clusters. In this chapter we have shown that exposure times of 105 s, on 8-10 m class tele- ∼ scopes, are necessary to acquire spectra with S/N & 10.0 and perform the study of the stellar populations of z 1 red sequence galaxies with the Lick analysis. With such ex- ∼ tremely long exposure times, work is very time-consuming and challenging, particularly if stellar population analysis is to be conducted on an individual galaxy basis. Therefore, in order to study the stellar populations in individual red sequence galaxies at z 1, it ∼ will be necessary to wait until the next generations of 20-40 m class telescopes, such as the Giant Magellan Telescope (GMT), the European Extremely Large Telescope (E-ELT), and the Thirty Meter Telescope (TMT) will be operating.

RCS2319 (z =0.91) S0 galaxies 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5[OII] MgI Hδ Mg2

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A) RCS2319 (z =0.91) disc-dominated (I) galaxies 2.0 H6 K H Ca4227 G Hγ Ca4431 Hβ

1.5[OII] MgI Hδ Mg2

8000 1.0 F/F

0.5 Fe4383 Fe4531 C2 4668

0.0 7000 7500 8000 8500 9000 9500 10000 λ (A)

Figure 6.7 Spectra of the S0 galaxy RCS2319 201 (top panel) and of the early-type disc-dominated galaxy RCS2319 174 (bottom panel) observed with LRIS in the cluster RCS2319 (z = 0.91). While the disc-dominated galaxy has a strong Hδ absorption line, the S0 galaxy reproduces the trend in the Ca H and K lines observed in the composite S0 spectrum of RCS0220.

7 Summary and Future Plans

This concluding chapter summarises the main scientiﬁc results of this thesis and outlines the future directions that will be taken in the study of high redshift galaxy clusters.

7.1 Summary and Conclusions

The Hawk-I Cluster Survey (HCS) consists of a sample 9 galaxy clusters at redshifts 0.8 < z < 1.5. Multi-band imaging, from both space- and ground-based telescopes, and spectra are available for all the clusters. Further spectra have been acquired at the 8m Gemini North and 10m Keck telescopes to study the stellar populations within the red sequence galaxies of three of the HCS clusters. The observations and reduction of all the data used in the present thesis are discussed in Chapter 2. The primary scientiﬁc goal of this PhD thesis was to study the evolution of the red sequence in the HCS clusters focusing on four complementary aspects:

Chapter 3: The Analysis Method

Chapter 3 discussed the method developed for the analysis of the HCS clusters and the results obtained from the study of the cluster XMMU J1229+0151 (XMM1229), at z = 0.98. We chose this cluster as a testbed for developing the techniques necessary for the analysis

159 160 Chapter 7. Summary and Future Plans of the diverse HCS dataset and presented a comprehensive study of its red sequence in Cerulo et al. (2014).

We developed an effective photometric analysis based on PSF-matching, which allowed us to measure reliable colours minimising the effects of colour gradients. The suppression of these effects was also crucial for building multi-wavelength samples for the estimation of stellar masses and photometric redshifts by using SED fitting techniques.

In order to reliably estimate the membership of galaxy clusters, we used photometric redshifts and statistical background subtraction. We tested both methods on XMM1229 and showed that they produced reliable and consistent results.

We estimated morphologies of red sequence galaxies adopting an approach based on both visual and automatic classiﬁcation. The red sequence is populated by galaxies that have little or no ongoing star formation. As a result, features like spiral arms, which host most of the star formation, are attenuated in the images. This, together with the cosmological surface brightness dimming, complicates the detection of discs in distant galaxies. We were able to minimise the impact of this eﬀect by selecting galaxies that were brighter than z850 = 24.0 mag. We divided the red sequence into the following four classes in order of decreasing apparent bulge-to-total light ratio (B/T ): elliptical (i.e. pure bulges), bulge-dominated (S0), early-type disc-dominated (Sa-Sbc), and late-type disc-dominated (Sc-Sd). To these classes we added a class of irregular galaxies.

We found that the red sequence of XMM1229 was already assembled at z = 0.98. We estimated its zero-point, slope, and intrinsic scatter and found that they were consistent with the results of other works on the same cluster published in the recent literature. We also found that the rest-frame slope and intrinsic scatter were consistent with the values estimated in the z 0.05 WINGS clusters. The luminous-to-faint ratio (L/F ) of ∼ XMM1229 red sequence galaxies was higher but still consistent with the median value of the WINGS clusters.

Interestingly, the XMM1229 red sequence shows an upturn in the fraction of disc- dominated galaxies at faint magnitudes (VAB > 20.0 mag) with a corresponding de- − crease in the fraction of S0 galaxies. At the same luminosities, the WINGS red sequence is dominated by S0 galaxies. The fraction of elliptical galaxies remains approximately constant along the red sequence following a trend similar to WINGS. We argued that low-redshift S0 galaxies are likely to be the descendants of the disc-dominated galaxies seen in XMM1229. 7.1. Summary and Conclusions 161

Chapter 4: The Build-up of the Red Sequence

Chapter 4 presented the analysis of the red sequence in the HCS clusters. We fitted the red sequence in the observed colour-magnitude diagram and then derived its rest-frame zero- point, slope, and intrinsic scatter to compare with observational and theoretical works in the recent literature. We found that there is little or no evolution of the red sequence slope and zero-point, and that, unlike the predictions of the simulations of Romeo et al. (2008), the cluster red sequence already has a negative slope at z = 1.5. The intrinsic scatter in the red sequence that we measured for the HCS clusters is consistent with most works in the recent literature, although measurements of this quantity have large uncertainties. To investigate the effects of stellar mass on the build-up of the cluster red sequence, we studied both the luminous-to-faint ratio and the luminosity distribution of red sequence members. The comparison with WINGS showed that the red sequence was already assembled in most clusters in the HCS sample (viz at 0.8 < z < 1.5). This conclusion is in agreement with the works of Andreon (2008), Crawford et al. (2009), and De Propris et al. (2013) but is at odds with the results of most works in the recent literature (e.g. De Lucia et al. 2007b; Capozzi et al. 2010; Rudnick et al. 2012), who found a deficit of galaxies at the faint end of the red sequence. We also found that, after dividing the HCS sample in 14 14 clusters with dark matter halo masses MDM < 5 10 M and MDM 5 10 M , more × ≥ × massive clusters showed the presence of luminous red sequence galaxies that are not seen in the low-mass clusters. These galaxies, which are situated at the centres of the clusters, are all 0.5-1.0 mag brighter than the rest of the red sequence. The absence of a deficit at the faint end of the red sequence at z 1 suggests that ∼ the build-up of the cluster red sequence was fast. We proposed an evolutionary scenario in which galaxy clusters are built up by the accretion of low-mass groups on to a central massive protocluster at z > 2. Galaxies are quenched both in the main, central protocluster and in the satellite groups. Protocluster members are quenched over shorter timescales than those involved in the groups because the densities are higher. When groups are accreted, a fraction of their galaxies is already quenched and populates the red sequence of the cluster at all luminosities. In this scenario the cluster red sequence has an accelerated build-up because it is enriched by the galaxies that are preprocessed in the satellite groups. The presence of massive elliptical galaxies in the centres of the most massive HCS clusters is explained by the fact that the most massive haloes formed earlier than the least massive ones. Galaxies at the centres of those haloes, therefore, accreted more satellites, and their masses became larger than those of the centrals in less massive clusters. Although this scenario can explain the results of Chapter 4, more observations of pro- 162 Chapter 7. Summary and Future Plans toclusters at z > 1.5 are needed to test it. Observations of clusters and protoclusters at z 2 have shown that these systems are not virialised yet and have a clumpy structure ∼ (Cooke et al., 2014; Yuan et al., 2014), suggesting that these clumps will eventually merge to form a relaxed cluster. This is in agreement with our accretion scenario and with theoretical simulations of structure formation in a ΛCDM universe (Fakhouri et al., 2010; Chiang et al., 2013). However, the detection and spectroscopic follow-up of z > 2 protoclusters is difficult, and only a few of these systems have been identified and confirmed so far (Tanaka et al., 2013; Cooke et al., 2014; Yuan et al., 2014).

Chapter 5: Morphological Evolution

Chapter 5 presented the morphological analysis of the red sequence in the HCS clusters. We applied the same classification scheme used for XMM1229 and studied the behaviour of the morphological fractions with galaxy luminosity and stellar mass. We showed that the HCS red sequence is dominated by elliptical galaxies at all luminosities and stellar masses, whereas in WINGS the red sequence becomes dominated by S0 galaxies at VAB > 21.0 − mag and log(M∗/M ) < 11.5. The total fraction of S0 galaxies is 20% higher in WINGS than in the HCS. This result supports the notion that elliptical and S0 galaxies follow different evolutionary paths, and that the S0 galaxies observed in low-redshift clusters were formed as the result of morphological transformation of quiescent spiral galaxies. This conclusion is corroborated by previous works that showed that the fraction of late- type galaxies on the cluster red sequence peaks at z 0.6 (Sánchez-Blázquezet al., 2009), ∼ and by the fact that the timescales of star-formation quenching are shorter than those of morphological transformation (Moore et al., 1998; Bekki & Couch, 2011; Taranu et al., 2014). Thus galaxies have their star formation quenched and join the red sequence as gas- poor spirals. Due to the lack of gas, which is necessary to suppress random motion in the galactic disc, the spiral arms are dissolved. Alternatively, spiral arms can be disrupted, in the dense cluster environment, by close encounters with neighbouring galaxies. In Chapter 5 we stated that these two mechanisms are competing and that we will need to study the local environment of each red sequence spiral to assess which of the two mechanisms is prevailing. One further result of Chapter 5 is that late-type galaxies make up only 10% of the total red sequence population at both high and at low redshift. We did not find any upturn of early-type disc-dominated galaxies at faint magnitudes as observed in XMM1229, and we concluded that the trend in the fraction of early-type disc-dominated galaxies is peculiar to XMM1229. On the other hand, we noted that the magnitude selection of the HCS 7.1. Summary and Conclusions 163 morphological sample, z850 < 24.0 mag, results in a shallow sampling of the red sequence for some of the clusters. For example, in the cluster RDCS J1252.9-2927 (RDCS1252), at z = 1.24, the red sequence is covered only down to F 125W = 23.0 mag, which is 2 mag brighter than the 90% completeness limit in the F775W band (Figure 5.1). In other words, with our magnitude selection, we are not able to study morphological fractions at the faint end of the red sequence in this cluster. Therefore, we did not exclude that higher fractions of late-type galaxies could be present at magnitudes fainter than the limit of the morphological sample but, in order to test that, deeper images are needed. Interestingly, we found that late-type galaxies on both the WINGS and HCS red sequences are slightly bluer than ellitpicals and S0s.

We ﬁnally observed that the bright end of the red sequence is dominated by elliptical galaxies in both HCS and WINGS. We noted that the luminosities and stellar masses of the WINGS brightest red sequence galaxies are especially high. The ratios between the masses of these galaxies and those of the brightest red sequence galaxies in HCS are higher than the values implied by the mass growth factors of brightest cluster galaxies derived in recent works (e.g.: Lidman et al. 2012; Ascaso et al. 2014). However, as also extensively discussed in Chapter 3, the WINGS brightest red sequence galaxies represent a peculiar 12.4 class of object with extremely high stellar masses (up to 10 M ), and colours which are bluer than those predicted by the evolution of synthetic stellar populations or by other studies of clusters at similar redshifts.

The HCS red sequence presents an increase in the fraction of elliptical galaxies at faint luminosities (VAB > 19.5 mag) and low stellar masses (log(M∗/M ) < 10.7 in the − magnitude limited sample). We showed that face-on S0 galaxies misclassiﬁed as ellipticals contribute only a very small fraction to this upturn.

We proposed a scenario for the morphological evolution of red sequence galaxies in which elliptical galaxies are made up of two populations, a primordial population of giant galaxies with stellar masses log(M∗/M ) > 10.8, formed in the main protocluster, and a population of dwarf galaxies with stellar masses log(M∗/M ) < 10.8, which were partly accreted and partly formed in the lowest-density regions of the central protocluster. S0 galaxies, on the other hand, constitute a third population of quenched early-type galaxies originating from the morphological transformation of quiescent spiral galaxies on the red sequence. This scenario is similar to the one proposed by Poggianti et al. (2006) for clusters at 0.4 < z < 0.8, and it also explains the age diﬀerences between elliptical and S0 galaxies and between bright and faint galaxies observed on the cluster red sequence at z 1 (Tran ∼ et al., 2007; Mei et al., 2009; Muzzin et al., 2012; Nantais et al., 2013b). 164 Chapter 7. Summary and Future Plans

Chapter 6: Spectral Properties of Red Sequence Galaxies

Chapter 6 presented the early results of the analysis of the spectra of red sequence galaxies taken for 3 HCS clusters, namely RCS 2319.8+0038 (RCS2319, z = 0.91), XMMU J1229+0151 (XMM1229, z = 0.98), and RCS 0220.9-0333 (RCS0220, z = 1.03). These clusters were observed with the Keck and Gemini North telescopes with the aim of acquiring spectra for the measurements of stellar age and metallicity of red sequence galaxies. In order to measure these two quantities with the analysis of the absorption indices of the Lick system (Faber et al., 1985), it is necessary to have deep spectra with signal-to- noise ratios S/N > 10. We have shown that on 8-10 m class optical telescopes, such as Keck and Gemini North, such deep spectra can be obtained only with very long exposures ( 105 s). Thus measurements of age and metallicity in individual red sequence galaxies, ∼ at the redshifts of the HCS clusters, would require extremely time-consuming observations. While the exposure times for such observations will be considerably reduced when the next generations of 20-40 m class telescopes, such as the Giant Magellan Telescope and the European Extremely Large Telescope, will be operating, we showed that spectral stacking allows us to attain S/N > 9 and to estimate average stellar age and metallicities on the red sequence. Thus we built composite spectra for galaxies of diﬀerent morphological types and luminosities in each cluster. Our qualitative analysis of these spectra shows that S0 and early-type disc dominated galaxies present evidence for younger stellar populations with respect to elliptical galaxies. This result is in agreement with previous studies of the spectra of red sequence galaxies at z 1, which showed that elliptical galaxies have on ∼ average older stellar ages than S0 and late-type galaxies (Tran et al., 2007; Demarco et al., 2010). We also found evidence for the presence of younger stellar populations in galaxies with z 22.0 mag with respect to galaxies with z < 22.0 mag, in agreement with 850 ≥ 850 Muzzin et al. (2012) and Nantais et al. (2013b). We stress that our analysis is purely qualitative and based on the apparent strength of age-sensitive absorption indices. A quantitative analysis is necessary to test these conclusions and is part of the future plans of the HCS collaboration.

7.2 Future Plans

The research activities of the HAWK-I Cluster Survey are still ongoing and the results presented in this thesis lay the foundations for future studies aimed at the understanding of high redshift clusters. The next step to test our evolutionary scenario for the build-up 7.2. Future Plans 165 of the red sequence will be the quantitative study of the stellar populations as a function of galaxy luminosity and morphology. In this way we will have a clear picture of the links between stellar age, luminosity, and morphology of galaxies. We also plan to study the spatial trend of galaxy morphology and star-formation rate in the HCS clusters. With their redshift range (viz. 0.9 < z < 1.5), the HCS clusters cover an interval in cosmic time which is just subsequent to the changes in the star-formation vs density relation observed at z > 1.5 by Tran et al. (2007) and Fassbender et al. (2014). Therefore, it is suitable to search for any flattening with redshift of these two relations. Finally, we plan to study the differences between cluster and field galaxies taking advantage of the wealth of publicly available images and spectra in the fields covered by CANDELS. This will be especially important to test our scenarios for the build-up of the red sequence and the morphological evolution with galaxies that reside in lower-density environments. These three aspects of galaxy clusters that we plan to investigate will allow us to draw a comprehensive picture of the properties of galaxies in dense environments and to study the physical mechanisms responsible for the evolution of galaxies in these complex systems.

Bibliography

Abraham, R. G., van den Bergh, S., Glazebrook, K., Ellis, R. S., Santiago, B. X., Surma, P., & Griﬃths, R. E., ApJS, 107:1, November 1996.

Abraham, R. G., van den Bergh, S., & Nair, P., ApJ, 588:218–229, May 2003.

Andreon, S., MNRAS, 386:1045–1052, May 2008.

Andreon, S. & Huertas-Company, M., A&A, 526:A11, February 2011.

Andreon, S., Valtchanov, I., Jones, L. R., Altieri, B., Bremer, M., Willis, J., Pierre, M., & Quintana, H., MNRAS, 359:1250–1260, June 2005.

Andreon, S., Quintana, H., Tajer, M., Galaz, G., & Surdej, J., MNRAS, 365:915–928, January 2006.

Andreon, S., Newman, A. B., Trinchieri, G., Raichoor, A., Ellis, R. S., & Treu, T., A&A, 565:A120, May 2014.

Appenzeller, I., Fricke, K., Fürtig,W., Gässler,W., Häfner,R., Harke, R., Hess, H.-J., Hummel, W., Jürgens, P., Kudritzki, R.-P., Mantel, K.-H., Meisl, W., Muschielok, B., Nicklas, H., Rupprecht, G., Seifert, W., Stahl, O., Szeifert, T., & Tarantik, K., The Messenger, 94:1–6, December 1998.

Arnouts, S., Cristiani, S., Moscardini, L., Matarrese, S., Lucchin, F., Fontana, A., & Giallongo, E., MNRAS, 310:540–556, December 1999.

Ascaso, B., Lemaux, B. C., Lubin, L. M., Gal, R. R., Kocevski, D. D., Rumbaugh, N., & Squires, G., MNRAS, 442:589–615, July 2014.

Bahcall, N. A., ARA&A, 15:505–540, 1977.

Bahe, Y. M. & McCarthy, I. G., ArXiv e-prints, October 2014.

Bah´e,Y. M. & McCarthy, I. G., MNRAS, 447:969–992, February 2015.

Baldry, I. K., Glazebrook, K., Brinkmann, J., Ivezi´c, Z.,ˇ Lupton, R. H., Nichol, R. C., & Szalay, A. S., ApJ, 600:681–694, January 2004.

Balestra, I., Mainieri, V., Popesso, P., Dickinson, M., Nonino, M., Rosati, P., Teimoorinia, H., Vanzella, E., Cristiani, S., Cesarsky, C., Fosbury, R. A. E., Kuntschner, H., & Rettura, A., A&A, 512:A12, March 2010.

167 168 Bibliography

Barkhouse, W. A., Yee, H. K. C., & L´opez-Cruz, O., ApJ, 671:1471–1496, December 2007.

Bassett, R., Papovich, C., Lotz, J. M., Bell, E. F., Finkelstein, S. L., Newman, J. A., Tran, K.-V., Almaini, O., Lani, C., Cooper, M., Croton, D., Dekel, A., Ferguson, H. C., Kocevski, D. D., Koekemoer, A. M., Koo, D. C., McGrath, E. J., McIntosh, D. H., & Wechsler, R., ApJ, 770:58, June 2013.

Bautz, L. P. & Morgan, W. W., ApJ, 162:L149, December 1970.

Bekki, K., ApJ, 510:L15–L19, January 1999.

Bekki, K. & Couch, W. J., ApJ, 596:L13–L16, October 2003.

Bekki, K. & Couch, W. J., MNRAS, 415:1783–1796, August 2011.

Bernardi, M., Sheth, R. K., Annis, J., Burles, S., Finkbeiner, D. P., Lupton, R. H., Schlegel, D. J., SubbaRao, M., Bahcall, N. A., Blakeslee, J. P., Brinkmann, J., Cas- tander, F. J., Connolly, A. J., Csabai, I., Doi, M., Fukugita, M., Frieman, J., Heckman, T., Hennessy, G. S., Ivezi´c, Z.,ˇ Knapp, G. R., Lamb, D. Q., McKay, T., Munn, J. A., Nichol, R., Okamura, S., Schneider, D. P., Thakar, A. R., & York, D. G., AJ, 125: 1882–1896, April 2003.

Bernardi, M., Shankar, F., Hyde, J. B., Mei, S., Marulli, F., & Sheth, R. K., MNRAS, 404:2087–2122, June 2010.

Bertin, E. Automatic Astrometric and Photometric Calibration with SCAMP. In Gabriel, C., Arviset, C., Ponz, D., & Enrique, S., editors, Astronomical Data Analysis Software and Systems XV, volume 351 of Astronomical Society of the Paciﬁc Conference Series, page 112, July 2006.

Bertin, E. Automated Morphometry with SExtractor and PSFEx. In Evans, I. N., Acco- mazzi, A., Mink, D. J., & Rots, A. H., editors, Astronomical Data Analysis Software and Systems XX, volume 442 of Astronomical Society of the Paciﬁc Conference Series, page 435, July 2011.

Bertin, E. & Arnouts, S., A&AS, 117:393–404, June 1996.

Bertin, E., Mellier, Y., Radovich, M., Missonnier, G., Didelon, P., & Morin, B. The TERAPIX Pipeline. In Bohlender, D. A., Durand, D., & Handley, T. H., editors, Astro- nomical Data Analysis Software and Systems XI, volume 281 of Astronomical Society of the Paciﬁc Conference Series, page 228, 2002. Bibliography 169

Bildfell, C., Hoekstra, H., Babul, A., Sand, D., Graham, M., Willis, J., Urquhart, S., Mahdavi, A., Pritchet, C., Zaritsky, D., Franse, J., & Langelaan, P., MNRAS, 425: 204–221, September 2012.

Binney, J. & Merriﬁeld, M. Galactic Astronomy. 1998.

Blakeslee, J. P., Anderson, K. R., Meurer, G. R., Ben´ıtez,N., & Magee, D. An Auto- matic Image Reduction Pipeline for the Advanced Camera for Surveys. In Payne, H. E., Jedrzejewski, R. I., & Hook, R. N., editors, Astronomical Data Analysis Software and Systems XII, volume 295 of Astronomical Society of the Paciﬁc Conference Series, page 257, 2003.

Blakeslee, J. P., Holden, B. P., Franx, M., Rosati, P., Bouwens, R. J., Demarco, R., Ford, H. C., Homeier, N. L., Illingworth, G. D., Jee, M. J., Mei, S., Menanteau, F., Meurer, G. R., Postman, M., & Tran, K.-V. H., ApJ, 644:30–53, June 2006.

Bower, R. G., Lucey, J. R., & Ellis, R. S., MNRAS, 254:601, February 1992.

Bremer, M. N., Valtchanov, I., Willis, J., Altieri, B., Andreon, S., Duc, P. A., Fang, F., Jean, C., Lonsdale, C., Pacaud, F., Pierre, M., Shupe, D. L., Surace, J. A., & Waddington, I., MNRAS, 371:1427–1434, September 2006.

Brown, M. J. I., Zheng, Z., White, M., Dey, A., Jannuzi, B. T., Benson, A. J., Brand, K., Brodwin, M., & Croton, D. J., ApJ, 682:937–963, August 2008.

Bruzual, G. & Charlot, S., MNRAS, 344:1000–1028, October 2003.

Burke, C., Collins, C. A., Stott, J. P., & Hilton, M., MNRAS, 425:2058–2068, September 2012.

Butcher, H. & Oemler, Jr., A., ApJ, 226:559–565, December 1978.

Cameron, E., pasa, 28:128–139, June 2011.

Cantalupo, S., MNRAS, 403:L16–L20, March 2010.

Capozzi, D., Collins, C. A., & Stott, J. P., MNRAS, 403:1274–1282, April 2010.

Cardelli, J. A., Clayton, G. C., & Mathis, J. S., ApJ, 345:245–256, October 1989.

Casertano, S., de Mello, D., Dickinson, M., Ferguson, H. C., Fruchter, A. S., Gonzalez- Lopezlira, R. A., Heyer, I., Hook, R. N., Levay, Z., Lucas, R. A., Mack, J., Makidon, 170 Bibliography

R. B., Mutchler, M., Smith, T. E., Stiavelli, M., Wiggs, M. S., & Williams, R. E., AJ, 120:2747–2824, December 2000.

Cava, A., Bettoni, D., Poggianti, B. M., Couch, W. J., Moles, M., Varela, J., Biviano, A., D’Onofrio, M., Dressler, A., Fasano, G., Fritz, J., Kjærgaard, P., Ramella, M., & Valentinuzzi, T., A&A, 495:707–719, March 2009.

Cerulo, P., Couch, W. J., Lidman, C., Delaye, L., Demarco, R., Huertas-Company, M., Mei, S., & S´anchez-Janssen, R., MNRAS, 439:2790–2812, February 2014.

Chabrier, G., ApJ, 586:L133–L136, April 2003.

Chiang, Y.-K., Overzier, R., & Gebhardt, K., ApJ, 779:127, December 2013.

Clowe, D., Bradaˇc,M., Gonzalez, A. H., Markevitch, M., Randall, S. W., Jones, C., & Zaritsky, D., ApJ, 648:L109–L113, September 2006.

Coleman, G. D., Wu, C.-C., & Weedman, D. W., ApJS, 43:393–416, July 1980.

Conselice, C. J., Bershady, M. A., & Jangren, A., ApJ, 529:886–910, February 2000.

Conselice, C. J., Bershady, M. A., Dickinson, M., & Papovich, C., AJ, 126:1183–1207, September 2003.

Cooke, E. A., Hatch, N. A., Muldrew, S. I., Rigby, E. E., & Kurk, J. D., MNRAS, 440: 3262–3274, June 2014.

Cortese, L. & Hughes, T. M., MNRAS, 400:1225–1240, December 2009.

Couch, W. J. & Sharples, R. M., MNRAS, 229:423–456, December 1987.

Cowie, L. L., Songaila, A., Hu, E. M., & Cohen, J. G., AJ, 112:839, September 1996.

Crawford, C. S., Allen, S. W., Ebeling, H., Edge, A. C., & Fabian, A. C., MNRAS, 306: 857–896, July 1999.

Crawford, S. M., Bershady, M. A., & Hoessel, J. G., ApJ, 690:1158–1180, January 2009.

Croton, D. J., Springel, V., White, S. D. M., De Lucia, G., Frenk, C. S., Gao, L., Jenkins, A., Kauﬀmann, G., Navarro, J. F., & Yoshida, N., MNRAS, 365:11–28, January 2006.

D’Agostini, G., ArXiv Physics e-prints, December 2004. Bibliography 171

Dawson, K. S., Aldering, G., Amanullah, R., Barbary, K., Barrientos, L. F., Brodwin, M., Connolly, N., Dey, A., Doi, M., Donahue, M., Eisenhardt, P., Ellingson, E., Faccioli, L., Fadeyev, V., Fakhouri, H. K., Fruchter, A. S., Gilbank, D. G., Gladders, M. D., Goldhaber, G., Gonzalez, A. H., Goobar, A., Gude, A., Hattori, T., Hoekstra, H., Huang, X., Ihara, Y., Jannuzi, B. T., Johnston, D., Kashikawa, K., Koester, B., Konishi, K., Kowalski, M., Lidman, C., Linder, E. V., Lubin, L., Meyers, J., Morokuma, T., Munshi, F., Mullis, C., Oda, T., Panagia, N., Perlmutter, S., Postman, M., Pritchard, T., Rhodes, J., Rosati, P., Rubin, D., Schlegel, D. J., Spadafora, A., Stanford, S. A., Stanishev, V., Stern, D., Strovink, M., Suzuki, N., Takanashi, N., Tokita, K., Wagner, M., Wang, L., Yasuda, N., Yee, H. K. C., & Supernova Cosmology Project, T., AJ, 138: 1271–1283, November 2009.

De Lucia, G., Poggianti, B. M., Arag´on-Salamanca,A., Clowe, D., Halliday, C., Jablonka, P., Milvang-Jensen, B., Pell´o,R., Poirier, S., Rudnick, G., Saglia, R., Simard, L., & White, S. D. M., ApJ, 610:L77–L80, August 2004.

De Lucia, G., Poggianti, B. M., Arag´on-Salamanca, A., White, S. D. M., Zaritsky, D., Clowe, D., Halliday, C., Jablonka, P., von der Linden, A., Milvang-Jensen, B., Pell´o, R., Rudnick, G., Saglia, R. P., & Simard, L., MNRAS, 374:809–822, January 2007b.

De Propris, R., Phillipps, S., & Bremer, M. N., MNRAS, 434:3469–3486, October 2013. de Vaucouleurs, G., Annales d’Astrophysique, 11:247, January 1948. de Vaucouleurs, G., Handbuch der Physik, 53:275, 1959. de Vaucouleurs, G., de Vaucouleurs, A., & Corwin, J. R. Second reference catalogue of bright galaxies. In Second reference catalogue of bright galaxies, 1976, Austin: University of Texas Press., page 0, 1976.

Dekel, A. & Birnboim, Y., MNRAS, 368:2–20, May 2006.

Delaye, L., Huertas-Company, M., Mei, S., Lidman, C., Licitra, R., Newman, A., Raichoor, A., Shankar, F., Barrientos, F., Bernardi, M., Cerulo, P., Couch, W., Demarco, R., Mu˜noz,R., S´anchez-Janssen, R., & Tanaka, M., MNRAS, 441:203–223, April 2014. 172 Bibliography

Demarco, R., Rosati, P., Lidman, C., Homeier, N. L., Scannapieco, E., Ben´ıtez, N., Mainieri, V., Nonino, M., Girardi, M., Stanford, S. A., Tozzi, P., Borgani, S., Silk, J., Squires, G., & Broadhurst, T. J., A&A, 432:381–394, March 2005.

Demarco, R., Rosati, P., Lidman, C., Girardi, M., Nonino, M., Rettura, A., Strazzullo, V., van der Wel, A., Ford, H. C., Mainieri, V., Holden, B. P., Stanford, S. A., Blakeslee, J. P., Gobat, R., Postman, M., Tozzi, P., Overzier, R. A., Zirm, A. W., Ben´ıtez,N., Homeier, N. L., Illingworth, G. D., Infante, L., Jee, M. J., Mei, S., Menanteau, F., Motta, V., Zheng, W., Clampin, M., & Hartig, G., ApJ, 663:164–182, July 2007.

Demarco, R., Gobat, R., Rosati, P., Lidman, C., Rettura, A., Nonino, M., van der Wel, A., Jee, M. J., Blakeslee, J. P., Ford, H. C., & Postman, M., ApJ, 725:1252–1276, December 2010.

Desai, V., Dalcanton, J. J., Arag´on-Salamanca,A., Jablonka, P., Poggianti, B., Gogarten, S. M., Simard, L., Milvang-Jensen, B., Rudnick, G., Zaritsky, D., Clowe, D., Halliday, C., Pell´o,R., Saglia, R., & White, S., ApJ, 660:1151–1164, May 2007.

Dressler, A., ApJ, 236:351–365, March 1980.

Dressler, A. & Gunn, J. E., ApJS, 78:1–60, January 1992.

Dressler, A., Oemler, Jr., A., Couch, W. J., Smail, I., Ellis, R. S., Barger, A., Butcher, H., Poggianti, B. M., & Sharples, R. M., ApJ, 490:577, December 1997.

Dressler, A., Smail, I., Poggianti, B. M., Butcher, H., Couch, W. J., Ellis, R. S., & Oemler, Jr., A., ApJS, 122:51–80, May 1999.

Driver, S. P. & Phillipps, S., ApJ, 469:529, October 1996.

Ebeling, H., MNRAS, 340:1269–1278, April 2003.

Ellis, R. S., Smail, I., Dressler, A., Couch, W. J., Oemler, Jr., A., Butcher, H., & Sharples, R. M., ApJ, 483:582–596, July 1997.

Faber, S. M., Friel, E. D., Burstein, D., & Gaskell, C. M., ApJS, 57:711–741, April 1985.

Faber, S. M., Willmer, C. N. A., Wolf, C., Koo, D. C., Weiner, B. J., Newman, J. A., Im, M., Coil, A. L., Conroy, C., Cooper, M. C., Davis, M., Finkbeiner, D. P., Gerke, B. F., Gebhardt, K., Groth, E. J., Guhathakurta, P., Harker, J., Kaiser, N., Kassin, S., Kleinheinrich, M., Konidaris, N. P., Kron, R. G., Lin, L., Luppino, G., Madgwick, D. S., Bibliography 173

Meisenheimer, K., Noeske, K. G., Phillips, A. C., Sarajedini, V. L., Schiavon, R. P., Simard, L., Szalay, A. S., Vogt, N. P., & Yan, R., ApJ, 665:265–294, August 2007.

Fakhouri, O., Ma, C.-P., & Boylan-Kolchin, M., MNRAS, 406:2267–2278, August 2010.

Faloon, A. J., Webb, T. M. A., Ellingson, E., Yan, R., Gilbank, D. G., Geach, J. E., Noble, A. G., Barrientos, L. F., Yee, H. K. C., Gladders, M., & Richard, J., ApJ, 768:104, May 2013.

Fasano, G., Poggianti, B., Couch, W., Bettoni, D., Kjærgaard, P., & Moles, M., Astro- physics and Space Science Supplement, 277:417–420, 2001.

Fasano, G., Marmo, C., Varela, J., D’Onofrio, M., Poggianti, B. M., Moles, M., Pignatelli, E., Bettoni, D., Kjærgaard, P., Rizzi, L., Couch, W. J., & Dressler, A., A&A, 445: 805–817, January 2006.

Fasano, G., Vanzella, E., Dressler, A., Poggianti, B. M., Moles, M., Bettoni, D., Valentin- uzzi, T., Moretti, A., D’Onofrio, M., Varela, J., Couch, W. J., Kjærgaard, P., Fritz, J., Omizzolo, A., & Cava, A., MNRAS, 420:926–948, February 2012.

Fassbender, R., B¨ohringer, H., Nastasi, A., Suhada,ˇ R., M¨uhlegger,M., de Hoon, A., Kohnert, J., Lamer, G., Mohr, J. J., Pierini, D., Pratt, G. W., Quintana, H., Rosati, P., Santos, J. S., & Schwope, A. D., New Journal of Physics, 13(12):125014, December 2011.

Fassbender, R., Nastasi, A., Santos, J. S., Lidman, C., Verdugo, M., Koyama, Y., Rosati, P., Pierini, D., Padilla, N., Romeo, A. D., Menci, N., Bongiorno, A., Castellano, M., Cerulo, P., Fontana, A., Galametz, A., Grazian, A., Lamastra, A., Pentericci, L., Som- mariva, V., Strazzullo, V., Suhada,ˇ R., & Tozzi, P., A&A, 568:A5, August 2014.

Feldmann, R., Carollo, C. M., Porciani, C., Lilly, S. J., Capak, P., Taniguchi, Y., Le F`evre,O., Renzini, A., Scoville, N., Ajiki, M., Aussel, H., Contini, T., McCracken, H., Mobasher, B., Murayama, T., Sanders, D., Sasaki, S., Scarlata, C., Scodeggio, M., Shioya, Y., Silverman, J., Takahashi, M., Thompson, D., & Zamorani, G., MNRAS, 372:565–577, October 2006.

Fioc, M. & Rocca-Volmerange, B., A&A, 326:950–962, October 1997.

Fogarty, L. M. R., Scott, N., Owers, M. S., Brough, S., Croom, S. M., Pracy, M. B., Houghton, R. C. W., Bland-Hawthorn, J., Colless, M., Davies, R. L., Jones, D. H., Allen, 174 Bibliography

J. T., Bryant, J. J., Goodwin, M., Green, A. W., Konstantopoulos, I. S., Lawrence, J. S., Richards, S., Cortese, L., & Sharp, R., MNRAS, 443:485–503, September 2014.

Ford, H., Postman, M., Blakeslee, J. P., Demarco, R., Jee, M. J., Rosati, P., Holden, B. P., Homeier, N., Illingworth, G., & White, R. L. The Evolutionary Status of Clusters of Galaxies at z 1. In Block, D. L., Puerari, I., Freeman, K. C., Groess, R., & ∼ Block, E. K., editors, Penetrating Bars Through Masks of Cosmic Dust, volume 319 of Astrophysics and Space Science Library, page 459, January 2004.

Fritz, J., Poggianti, B. M., Cava, A., Valentinuzzi, T., Moretti, A., Bettoni, D., Bressan, A., Couch, W. J., D’Onofrio, M., Dressler, A., Fasano, G., Kjærgaard, P., Moles, M., Omizzolo, A., & Varela, J., A&A, 526:A45, February 2011.

Fritz, J., Poggianti, B. M., Cava, A., Moretti, A., Varela, J., Bettoni, D., Couch, W. J., D’Onofrio D’Onofrio, M., Dressler, A., Fasano, G., Kjærgaard, P., Marziani, P., Moles, M., & Omizzolo, A., A&A, 566:A32, June 2014.

Fruchter, A. S. & Hook, R. N., PASP, 114:144–152, February 2002.

Gabor, J. M. & Dav´e,R., MNRAS, 427:1816–1829, December 2012.

Gabor, J. M. & Dav´e,R., ArXiv e-prints, May 2014.

Gallazzi, A., Charlot, S., Brinchmann, J., & White, S. D. M., MNRAS, 370:1106–1124, August 2006.

Garilli, B., Guzzo, L., Scodeggio, M., Bolzonella, M., Abbas, U., Adami, C., Arnouts, S., Bel, J., Bottini, D., Branchini, E., Cappi, A., Coupon, J., Cucciati, O., Davidzon, I., De Lucia, G., de la Torre, S., Franzetti, P., Fritz, A., Fumana, M., Granett, B. R., Ilbert, O., Iovino, A., Krywult, J., Le Brun, V., Le F`evre,O., Maccagni, D., Malek, K., Marulli, F., McCracken, H. J., Paioro, L., Polletta, M., Pollo, A., Schlagenhaufer, H., Tasca, L. A. M., Tojeiro, R., Vergani, D., Zamorani, G., Zanichelli, A., Burden, A., Di Porto, C., Marchetti, A., Marinoni, C., Mellier, Y., Moscardini, L., Nichol, R. C., Peacock, J. A., Percival, W. J., Phleps, S., & Wolk, M., A&A, 562:A23, February 2014.

Gavazzi, G., Savorgnan, G., Fossati, M., Dotti, M., Fumagalli, M., Boselli, A., Guti´errez, L., Hern´andezToledo, H., Giovanelli, R., & Haynes, M. P., A&A, 553:A90, May 2013.

Giavalisco, M., Ferguson, H. C., Koekemoer, A. M., Dickinson, M., Alexander, D. M., Bauer, F. E., Bergeron, J., Biagetti, C., Brandt, W. N., Casertano, S., Cesarsky, C., Chatzichristou, E., Conselice, C., Cristiani, S., Da Costa, L., Dahlen, T., de Mello, D., Bibliography 175

Eisenhardt, P., Erben, T., Fall, S. M., Fassnacht, C., Fosbury, R., Fruchter, A., Gardner, J. P., Grogin, N., Hook, R. N., Hornschemeier, A. E., Idzi, R., Jogee, S., Kretchmer, C., Laidler, V., Lee, K. S., Livio, M., Lucas, R., Madau, P., Mobasher, B., Moustakas, L. A., Nonino, M., Padovani, P., Papovich, C., Park, Y., Ravindranath, S., Renzini, A., Richardson, M., Riess, A., Rosati, P., Schirmer, M., Schreier, E., Somerville, R. S., Spinrad, H., Stern, D., Stiavelli, M., Strolger, L., Urry, C. M., Vandame, B., Williams, R., & Wolf, C., ApJ, 600:L93–L98, January 2004.

Gilbank, D. G. & Balogh, M. L., MNRAS, 385:L116–L119, March 2008.

Gladders, M. D. & Yee, H. K. C., ApJS, 157:1–29, March 2005.

Glazebrook, K. & Bland-Hawthorn, J., PASP, 113:197–214, February 2001.

Gobat, R., Daddi, E., Onodera, M., Finoguenov, A., Renzini, A., Arimoto, N., Bouwens, R., Brusa, M., Chary, R.-R., Cimatti, A., Dickinson, M., Kong, X., & Mignoli, M., A&A, 526:A133, February 2011.

Gonzalez, A. H., Tran, K.-V. H., Conbere, M. N., & Zaritsky, D., ApJ, 624:L73–L76, May 2005.

Graham, A., Lauer, T. R., Colless, M., & Postman, M., ApJ, 465:534, July 1996.

Graham, A. W. & Driver, S. P., pasa, 22:118–127, 2005.

Graham, A. W. & Guzm´an,R., AJ, 125:2936–2950, June 2003.

Grogin, N. A., Kocevski, D. D., Faber, S. M., Ferguson, H. C., Koekemoer, A. M., Riess, A. G., Acquaviva, V., Alexander, D. M., Almaini, O., Ashby, M. L. N., Barden, M., Bell, E. F., Bournaud, F., Brown, T. M., Caputi, K. I., Casertano, S., Cassata, P., Castellano, M., Challis, P., Chary, R.-R., Cheung, E., Cirasuolo, M., Conselice, C. J., Roshan Cooray, A., Croton, D. J., Daddi, E., Dahlen, T., Dav´e,R., de Mello, D. F., Dekel, A., Dickinson, M., Dolch, T., Donley, J. L., Dunlop, J. S., Dutton, A. A., Elbaz, D., Fazio, G. G., Filippenko, A. V., Finkelstein, S. L., Fontana, A., Gardner, J. P., Garnavich, P. M., Gawiser, E., Giavalisco, M., Grazian, A., Guo, Y., Hathi, N. P., H¨aussler,B., Hopkins, P. F., Huang, J.-S., Huang, K.-H., Jha, S. W., Kartaltepe, J. S., Kirshner, R. P., Koo, D. C., Lai, K., Lee, K.-S., Li, W., Lotz, J. M., Lucas, R. A., Madau, P., McCarthy, P. J., McGrath, E. J., McIntosh, D. H., McLure, R. J., Mobasher, B., Moustakas, L. A., Mozena, M., Nandra, K., Newman, J. A., Niemi, S.- M., Noeske, K. G., Papovich, C. J., Pentericci, L., Pope, A., Primack, J. R., Rajan, 176 Bibliography

A., Ravindranath, S., Reddy, N. A., Renzini, A., Rix, H.-W., Robaina, A. R., Rodney, S. A., Rosario, D. J., Rosati, P., Salimbeni, S., Scarlata, C., Siana, B., Simard, L., Smidt, J., Somerville, R. S., Spinrad, H., Straughn, A. N., Strolger, L.-G., Telford, O., Teplitz, H. I., Trump, J. R., van der Wel, A., Villforth, C., Wechsler, R. H., Weiner, B. J., Wiklind, T., Wild, V., Wilson, G., Wuyts, S., Yan, H.-J., & Yun, M. S., ApJS, 197:35, December 2011.

Gunn, J. E. & Gott, III, J. R., ApJ, 176:1, August 1972.

Guzzo, L., Scodeggio, M., Garilli, B., Granett, B. R., Fritz, A., Abbas, U., Adami, C., Arnouts, S., Bel, J., Bolzonella, M., Bottini, D., Branchini, E., Cappi, A., Coupon, J., Cucciati, O., Davidzon, I., De Lucia, G., de la Torre, S., Franzetti, P., Fumana, M., Hudelot, P., Ilbert, O., Iovino, A., Krywult, J., Le Brun, V., Le F`evre,O., Maccagni, D., Malek, K., Marulli, F., McCracken, H. J., Paioro, L., Peacock, J. A., Polletta, M., Pollo, A., Schlagenhaufer, H., Tasca, L. A. M., Tojeiro, R., Vergani, D., Zamorani, G., Zanichelli, A., Burden, A., Di Porto, C., Marchetti, A., Marinoni, C., Mellier, Y., Moscardini, L., Nichol, R. C., Percival, W. J., Phleps, S., & Wolk, M., A&A, 566:A108, June 2014.

Hatch, N. A., Kurk, J. D., Pentericci, L., Venemans, B. P., Kuiper, E., Miley, G. K., & R¨ottgering,H. J. A., MNRAS, 415:2993–3005, August 2011.

H¨aussler,B., McIntosh, D. H., Barden, M., Bell, E. F., Rix, H.-W., Borch, A., Beckwith, S. V. W., Caldwell, J. A. R., Heymans, C., Jahnke, K., Jogee, S., Koposov, S. E., Meisenheimer, K., S´anchez, S. F., Somerville, R. S., Wisotzki, L., & Wolf, C., ApJS, 172:615–633, October 2007.

H¨außler,B., Bamford, S. P., Vika, M., Rojas, A. L., Barden, M., Kelvin, L. S., Alpaslan, M., Robotham, A. S. G., Driver, S. P., Baldry, I. K., Brough, S., Hopkins, A. M., Liske, J., Nichol, R. C., Popescu, C. C., & Tuﬀs, R. J., MNRAS, 430:330–369, March 2013.

Hawarden, T. G., Leggett, S. K., Letawsky, M. B., Ballantyne, D. R., & Casali, M. M., MNRAS, 325:563–574, August 2001.

Hayashi, M., Kodama, T., Tadaki, K.-i., Koyama, Y., & Tanaka, I., ApJ, 757:15, Septem- ber 2012.

Hilton, M., Stanford, S. A., Stott, J. P., Collins, C. A., Hoyle, B., Davidson, M., Hosmer, M., Kay, S. T., Liddle, A. R., Lloyd-Davies, E., Mann, R. G., Mehrtens, N., Miller, C. J., Nichol, R. C., Romer, A. K., Sabirli, K., Sahl´en,M., Viana, P. T. P., West, M. J., Bibliography 177

Barbary, K., Dawson, K. S., Meyers, J., Perlmutter, S., Rubin, D., & Suzuki, N., ApJ, 697:436–451, May 2009.

Hilton, M., Lloyd-Davies, E., Stanford, S. A., Stott, J. P., Collins, C. A., Romer, A. K., Hosmer, M., Hoyle, B., Kay, S. T., Liddle, A. R., Mehrtens, N., Miller, C. J., Sahl´en, M., & Viana, P. T. P., ApJ, 718:133–147, July 2010.

Hogg, D. W., Blanton, M. R., Brinchmann, J., Eisenstein, D. J., Schlegel, D. J., Gunn, J. E., McKay, T. A., Rix, H.-W., Bahcall, N. A., Brinkmann, J., & Meiksin, A., ApJ, 601:L29–L32, January 2004.

Holden, B. P., Illingworth, G. D., Franx, M., Blakeslee, J. P., Postman, M., Kelson, D. D., van der Wel, A., Demarco, R., Magee, D. K., Tran, K.-V., Zirm, A., Ford, H., Rosati, P., & Homeier, N., ApJ, 670:190–205, November 2007.

Holden, B. P., Franx, M., Illingworth, G. D., Postman, M., van der Wel, A., Kelson, D. D., Blakeslee, J. P., Ford, H., Demarco, R., & Mei, S., ApJ, 693:617–633, March 2009.

Hubble, E. P., ApJ, 64:321–369, December 1926.

Huertas-Company, M., Rouan, D., Tasca, L., Soucail, G., & Le F`evre,O., A&A, 478: 971–980, February 2008.

Huertas-Company, M., Foex, G., Soucail, G., & Pell´o,R., A&A, 505:83–96, October 2009a.

Huertas-Company, M., Tasca, L., Rouan, D., Pelat, D., Kneib, J. P., Le F`evre,O., Capak, P., Kartaltepe, J., Koekemoer, A., McCracken, H. J., Salvato, M., Sanders, D. B., & Willott, C., A&A, 497:743–753, April 2009b.

Huertas-Company, M., Aguerri, J. A. L., Bernardi, M., Mei, S., & S´anchez Almeida, J., A&A, 525:A157, January 2011.

Ilbert, O., Arnouts, S., McCracken, H. J., Bolzonella, M., Bertin, E., Le F`evre,O., Mellier, Y., Zamorani, G., Pell`o,R., Iovino, A., Tresse, L., Le Brun, V., Bottini, D., Garilli, B., Maccagni, D., Picat, J. P., Scaramella, R., Scodeggio, M., Vettolani, G., Zanichelli, A., Adami, C., Bardelli, S., Cappi, A., Charlot, S., Ciliegi, P., Contini, T., Cucciati, O., Foucaud, S., Franzetti, P., Gavignaud, I., Guzzo, L., Marano, B., Marinoni, C., Mazure, A., Meneux, B., Merighi, R., Paltani, S., Pollo, A., Pozzetti, L., Radovich, M., Zucca, E., Bondi, M., Bongiorno, A., Busarello, G., de La Torre, S., Gregorini, L., Lamareille, F., Mathez, G., Merluzzi, P., Ripepi, V., Rizzo, D., & Vergani, D., A&A, 457:841–856, October 2006. 178 Bibliography

Ilbert, O., Salvato, M., Le Floc’h, E., Aussel, H., Capak, P., McCracken, H. J., Mobasher, B., Kartaltepe, J., Scoville, N., Sanders, D. B., Arnouts, S., Bundy, K., Cassata, P., Kneib, J.-P., Koekemoer, A., Le F`evre,O., Lilly, S., Surace, J., Taniguchi, Y., Tasca, L., Thompson, D., Tresse, L., Zamojski, M., Zamorani, G., & Zucca, E., ApJ, 709: 644–663, February 2010.

Jaffé,Y. L., Aragón-Salamanca,A., De Lucia, G., Jablonka, P., Rudnick, G., Saglia, R., & Zaritsky, D., MNRAS, 410:280–292, January 2011.

Janz, J. & Lisker, T., ApJ, 696:L102–L105, May 2009.

Jee, M. J., Rosati, P., Ford, H. C., Dawson, K. S., Lidman, C., Perlmutter, S., Demarco, R., Strazzullo, V., Mullis, C., B¨ohringer,H., & Fassbender, R., ApJ, 704:672–686, October 2009.

Jee, M. J., Dawson, K. S., Hoekstra, H., Perlmutter, S., Rosati, P., Brodwin, M., Suzuki, N., Koester, B., Postman, M., Lubin, L., Meyers, J., Stanford, S. A., Barbary, K., Barrientos, F., Eisenhardt, P., Ford, H. C., Gilbank, D. G., Gladders, M. D., Gonzalez, A., Harris, D. W., Huang, X., Lidman, C., Rykoﬀ, E. S., Rubin, D., & Spadafora, A. L., ApJ, 737:59, August 2011.

Jenkins, A., Frenk, C. S., White, S. D. M., Colberg, J. M., Cole, S., Evrard, A. E., Couchman, H. M. P., & Yoshida, N., MNRAS, 321:372–384, February 2001.

Jim´enez,N., Cora, S. A., Bassino, L. P., Tecce, T. E., & Smith Castelli, A. V., MNRAS, 417:785–800, October 2011.

Jones, L., Smail, I., & Couch, W. J., ApJ, 528:118–122, January 2000.

Kelson, D. D., PASP, 115:688–699, June 2003.

Kissler-Patig, M., Pirard, J.-F., Casali, M., Moorwood, A., Ageorges, N., Alves de Oliveira, C., Baksai, P., Bedin, L. R., Bendek, E., Biereichel, P., Delabre, B., Dorn, R., Esteves, R., Finger, G., Gojak, D., Huster, G., Jung, Y., Kiekebush, M., Klein, B., Koch, F., Lizon, J.-L., Mehrgan, L., Petr-Gotzens, M., Pritchard, J., Selman, F., & Stegmeier, J., A&A, 491:941–950, December 2008.

Kodama, T. & Arimoto, N., A&A, 320:41–53, April 1997.

Koekemoer, A. M., Fruchter, A. S., Hook, R. N., & Hack, W. MultiDrizzle: An Inte- grated Pyraf Script for Registering, Cleaning and Combining Images. In Arribas, S., Bibliography 179

Koekemoer, A., & Whitmore, B., editors, HST Calibration Workshop : Hubble after the Installation of the ACS and the NICMOS Cooling System, page 337, 2003.

Koekemoer, A. M., Faber, S. M., Ferguson, H. C., Grogin, N. A., Kocevski, D. D., Koo, D. C., Lai, K., Lotz, J. M., Lucas, R. A., McGrath, E. J., Ogaz, S., Rajan, A., Riess, A. G., Rodney, S. A., Strolger, L., Casertano, S., Castellano, M., Dahlen, T., Dickin- son, M., Dolch, T., Fontana, A., Giavalisco, M., Grazian, A., Guo, Y., Hathi, N. P., Huang, K.-H., van der Wel, A., Yan, H.-J., Acquaviva, V., Alexander, D. M., Almaini, O., Ashby, M. L. N., Barden, M., Bell, E. F., Bournaud, F., Brown, T. M., Caputi, K. I., Cassata, P., Challis, P. J., Chary, R.-R., Cheung, E., Cirasuolo, M., Conselice, C. J., Roshan Cooray, A., Croton, D. J., Daddi, E., Dav´e,R., de Mello, D. F., de Ravel, L., Dekel, A., Donley, J. L., Dunlop, J. S., Dutton, A. A., Elbaz, D., Fazio, G. G., Filippenko, A. V., Finkelstein, S. L., Frazer, C., Gardner, J. P., Garnavich, P. M., Gawiser, E., Gruetzbauch, R., Hartley, W. G., H¨aussler,B., Herrington, J., Hopkins, P. F., Huang, J.-S., Jha, S. W., Johnson, A., Kartaltepe, J. S., Khostovan, A. A., Kirshner, R. P., Lani, C., Lee, K.-S., Li, W., Madau, P., McCarthy, P. J., McIn- tosh, D. H., McLure, R. J., McPartland, C., Mobasher, B., Moreira, H., Mortlock, A., Moustakas, L. A., Mozena, M., Nandra, K., Newman, J. A., Nielsen, J. L., Niemi, S., Noeske, K. G., Papovich, C. J., Pentericci, L., Pope, A., Primack, J. R., Ravindranath, S., Reddy, N. A., Renzini, A., Rix, H.-W., Robaina, A. R., Rosario, D. J., Rosati, P., Salimbeni, S., Scarlata, C., Siana, B., Simard, L., Smidt, J., Snyder, D., Somerville, R. S., Spinrad, H., Straughn, A. N., Telford, O., Teplitz, H. I., Trump, J. R., Vargas, C., Villforth, C., Wagner, C. R., Wandro, P., Wechsler, R. H., Weiner, B. J., Wiklind, T., Wild, V., Wilson, G., Wuyts, S., & Yun, M. S., ApJS, 197:36, December 2011.

Kormendy, J. & Djorgovski, S., ARA&A, 27:235–277, 1989.

Kovaˇc,K., Lilly, S. J., Knobel, C., Bolzonella, M., Iovino, A., Carollo, C. M., Scarlata, C., Sargent, M., Cucciati, O., Zamorani, G., Pozzetti, L., Tasca, L. A. M., Scodeggio, M., Kampczyk, P., Peng, Y., Oesch, P., Zucca, E., Finoguenov, A., Contini, T., Kneib, J.-P., Le F`evre,O., Mainieri, V., Renzini, A., Bardelli, S., Bongiorno, A., Caputi, K., Coppa, G., de la Torre, S., de Ravel, L., Franzetti, P., Garilli, B., Lamareille, F., Le Borgne, J.-F., Le Brun, V., Maier, C., Mignoli, M., Pello, R., Perez Montero, E., Ricciardelli, E., Silverman, J. D., Tanaka, M., Tresse, L., Vergani, D., Abbas, U., Bottini, D., Cappi, A., Cassata, P., Cimatti, A., Fumana, M., Guzzo, L., Koekemoer, A. M., Leauthaud, A., Maccagni, D., Marinoni, C., McCracken, H. J., Memeo, P., Meneux, B., Porciani, C., Scaramella, R., & Scoville, N. Z., ApJ, 718:86–104, July 2010. 180 Bibliography

Kroupa, P., MNRAS, 322:231–246, April 2001.

Kubo, J. M., Stebbins, A., Annis, J., Dell’Antonio, I. P., Lin, H., Khiabanian, H., & Frieman, J. A., ApJ, 671:1466–1470, December 2007.

Labb´e,I., Franx, M., Rudnick, G., Schreiber, N. M. F., Rix, H.-W., Moorwood, A., van Dokkum, P. G., van der Werf, P., R¨ottgering,H., van Starkenburg, L., van der Wel, A., Kuijken, K., & Daddi, E., AJ, 125:1107–1123, March 2003.

Larson, R. B., Tinsley, B. M., & Caldwell, C. N., ApJ, 237:692–707, May 1980.

Laurikainen, E., Salo, H., Buta, R., & Knapen, J. H., MNRAS, 418:1452–1490, December 2011.

Lavery, R. J. & Henry, J. P., ApJ, 330:596–600, July 1988.

Le Borgne, D. & Rocca-Volmerange, B., A&A, 386:446–455, May 2002.

Le F`evre,O., Saisse, M., Mancini, D., Brau-Nogue, S., Caputi, O., Castinel, L., D’Odorico, S., Garilli, B., Kissler-Patig, M., Lucuix, C., Mancini, G., Pauget, A., Sciarretta, G., Scodeggio, M., Tresse, L., & Vettolani, G. Commissioning and performances of the VLT- VIMOS instrument. In Iye, M. & Moorwood, A. F. M., editors, Instrument Design and Performance for Optical/Infrared Ground-based Telescopes, volume 4841 of Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, pages 1670–1681, March 2003. doi: 10.1117/12.460959.

Lemaux, B. C., Gal, R. R., Lubin, L. M., Kocevski, D. D., Fassnacht, C. D., McGrath, E. J., Squires, G. K., Surace, J. A., & Lacy, M., ApJ, 745:106, February 2012.

Leonardi, A. J. & Rose, J. A., AJ, 111:182, January 1996.

Lidman, C., Rosati, P., Demarco, R., Nonino, M., Mainieri, V., Stanford, S. A., & Toft, S., A&A, 416:829–837, March 2004.

Lidman, C., Rosati, P., Tanaka, M., Strazzullo, V., Demarco, R., Mullis, C., Ageorges, N., Kissler-Patig, M., Petr-Gotzens, M. G., & Selman, F., A&A, 489:981–988, October 2008.

Lidman, C., Suherli, J., Muzzin, A., Wilson, G., Demarco, R., Brough, S., Rettura, A., Cox, J., DeGroot, A., Yee, H. K. C., Gilbank, D., Hoekstra, H., Balogh, M., Ellingson, E., Hicks, A., Nantais, J., Noble, A., Lacy, M., Surace, J., & Webb, T., MNRAS, 427: 550–568, November 2012. Bibliography 181

Lidman, C., Iacobuta, G., Bauer, A. E., Barrientos, L. F., Cerulo, P., Couch, W. J., Delaye, L., Demarco, R., Ellingson, E., Faloon, A. J., Gilbank, D., Huertas-Company, M., Mei, S., Meyers, J., Muzzin, A., Noble, A., Nantais, J., Rettura, A., Rosati, P., S´anchez-Janssen, R., Strazzullo, V., Webb, T. M. A., Wilson, G., Yan, R., & Yee, H. K. C., MNRAS, 433:825–837, July 2013.

Lilly, S. J., Le F`evre,O., Renzini, A., Zamorani, G., Scodeggio, M., Contini, T., Carollo, C. M., Hasinger, G., Kneib, J.-P., Iovino, A., Le Brun, V., Maier, C., Mainieri, V., Mignoli, M., Silverman, J., Tasca, L. A. M., Bolzonella, M., Bongiorno, A., Bottini, D., Capak, P., Caputi, K., Cimatti, A., Cucciati, O., Daddi, E., Feldmann, R., Franzetti, P., Garilli, B., Guzzo, L., Ilbert, O., Kampczyk, P., Kovac, K., Lamareille, F., Leau- thaud, A., Borgne, J.-F. L., McCracken, H. J., Marinoni, C., Pello, R., Ricciardelli, E., Scarlata, C., Vergani, D., Sanders, D. B., Schinnerer, E., Scoville, N., Taniguchi, Y., Arnouts, S., Aussel, H., Bardelli, S., Brusa, M., Cappi, A., Ciliegi, P., Finoguenov, A., Foucaud, S., Franceschini, A., Halliday, C., Impey, C., Knobel, C., Koekemoer, A., Kurk, J., Maccagni, D., Maddox, S., Marano, B., Marconi, G., Meneux, B., Mobasher, B., Moreau, C., Peacock, J. A., Porciani, C., Pozzetti, L., Scaramella, R., Schiminovich, D., Shopbell, P., Smail, I., Thompson, D., Tresse, L., Vettolani, G., Zanichelli, A., & Zucca, E., ApJS, 172:70–85, September 2007.

Longair, M. S. Galaxy Formation. 2008.

Lotz, J. M., Primack, J., & Madau, P., AJ, 128:163–182, July 2004.

Marchesini, D., Muzzin, A., Stefanon, M., Franx, M., Brammer, G. G., Marsan, C. Z., Vulcani, B., Fynbo, J. P. U., Milvang-Jensen, B., Dunlop, J. S., & Buitrago, F., ApJ, 794:65, October 2014.

Martig, M., Bournaud, F., Teyssier, R., & Dekel, A., ApJ, 707:250–267, December 2009.

Massey, P., Gronwall, C., & Pilachowski, C. A., PASP, 102:1046–1051, August 1990.

Masters, K. L., Mosleh, M., Romer, A. K., Nichol, R. C., Bamford, S. P., Schawinski, K., Lintott, C. J., Andreescu, D., Campbell, H. C., Crowcroft, B., Doyle, I., Edmondson, E. M., Murray, P., Raddick, M. J., Slosar, A., Szalay, A. S., & Vandenberg, J., MNRAS, 405:783–799, June 2010.

McIntosh, D. H., Wagner, C., Cooper, A., Bell, E. F., Kereˇs,D., Bosch, F. C. v. d., Gallazzi, A., Haines, T., Mann, J., Pasquali, A., & Christian, A. M., MNRAS, 442: 533–557, July 2014. 182 Bibliography

Mei, S., Holden, B. P., Blakeslee, J. P., Ford, H. C., Franx, M., Homeier, N. L., Illingworth, G. D., Jee, M. J., Overzier, R., Postman, M., Rosati, P., Van der Wel, A., & Bartlett, J. G., ApJ, 690:42–68, January 2009.

Mei, S., Stanford, S. A., Holden, B. P., Raichoor, A., Postman, M., Nakata, F., Finoguenov, A., Ford, H. C., Illingworth, G. D., Kodama, T., Rosati, P., Tanaka, M., Huertas-Company, M., Rettura, A., Shankar, F., Carrasco, E. R., Demarco, R., Eisenhardt, P., Jee, M. J., Koyama, Y., & White, R. L., ApJ, 754:141, August 2012.

Menci, N., Rosati, P., Gobat, R., Strazzullo, V., Rettura, A., Mei, S., & Demarco, R., ApJ, 685:863–874, October 2008.

Meyers, J., Aldering, G., Barbary, K., Barrientos, L. F., Brodwin, M., Dawson, K. S., Deustua, S., Doi, M., Eisenhardt, P., Faccioli, L., Fakhouri, H. K., Fruchter, A. S., Gilbank, D. G., Gladders, M. D., Goldhaber, G., Gonzalez, A. H., Hattori, T., Hsiao, E., Ihara, Y., Kashikawa, N., Koester, B., Konishi, K., Lidman, C., Lubin, L., Morokuma, T., Oda, T., Perlmutter, S., Postman, M., Ripoche, P., Rosati, P., Rubin, D., Rykoﬀ, E., Spadafora, A., Stanford, S. A., Suzuki, N., Takanashi, N., Tokita, K., Yasuda, N., & Supernova Cosmology Project, T., ApJ, 750:1, May 2012.

Moﬀat, A. F. J., A&A, 3:455, December 1969.

Moore, B., Katz, N., Lake, G., Dressler, A., & Oemler, A., Nature, 379:613–616, February 1996.

Moore, B., Lake, G., & Katz, N., ApJ, 495:139–151, March 1998.

Moorwood, A., Cuby, J.-G., Biereichel, P., Brynnel, J., Delabre, B., Devillard, N., van Dijsseldonk, A., Finger, G., Gemperlein, H., Gilmozzi, R., Herlin, T., Huster, G., Knud- strup, J., Lidman, C., Lizon, J.-L., Mehrgan, H., Meyer, M., Nicolini, G., Petr, M., Spyromilio, J., & Stegmeier, J., The Messenger, 94:7–9, December 1998a.

Moorwood, A., Cuby, J.-G., & Lidman, C., The Messenger, 91:9–13, March 1998b.

Moretti, A., Poggianti, B. M., Fasano, G., Bettoni, D., D’Onofrio, M., Fritz, J., Cava, A., Varela, J., Vulcani, B., Gullieuszik, M., Couch, W. J., Omizzolo, A., Valentinuzzi, T., Dressler, A., Moles, M., Kjærgaard, P., Smareglia, R., & Molinaro, M., A&A, 564: A138, April 2014.

Moster, B. P., Somerville, R. S., Maulbetsch, C., van den Bosch, F. C., Macci`o,A. V., Naab, T., & Oser, L., ApJ, 710:903–923, February 2010. Bibliography 183

Mullis, C. R., Rosati, P., Lamer, G., B¨ohringer,H., Schwope, A., Schuecker, P., & Fass- bender, R., ApJ, 623:L85–L88, April 2005.

Muzzin, A., Wilson, G., Yee, H. K. C., Gilbank, D., Hoekstra, H., Demarco, R., Balogh, M., van Dokkum, P., Franx, M., Ellingson, E., Hicks, A., Nantais, J., Noble, A., Lacy, M., Lidman, C., Rettura, A., Surace, J., & Webb, T., ApJ, 746:188, February 2012.

Nair, P. B. & Abraham, R. G., ApJS, 186:427–456, February 2010.

Nantais, J. B., Flores, H., Demarco, R., Lidman, C., Rosati, P., & Jee, M. J., A&A, 556: C4, August 2013a.

Nantais, J. B., Rettura, A., Lidman, C., Demarco, R., Gobat, R., Rosati, P., & Jee, M. J., A&A, 556:A112, August 2013b.

Oke, J. B., ApJS, 27:21, February 1974.

Oke, J. B., Cohen, J. G., Carr, M., Cromer, J., Dingizian, A., Harris, F. H., Labrecque, S., Lucinio, R., Schaal, W., Epps, H., & Miller, J., PASP, 107:375, April 1995.

Onodera, M., Carollo, C. M., Renzini, A., Cappellari, M., Mancini, C., Arimoto, N., Daddi, E., Gobat, R., Strazzullo, V., Tacchella, S., & Yamada, Y., ArXiv e-prints, November 2014.

Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W., AJ, 124:266–293, July 2002.

Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W., AJ, 139:2097–2129, June 2010a.

Peng, Y.-j., Lilly, S. J., Kovaˇc,K., Bolzonella, M., Pozzetti, L., Renzini, A., Zamorani, G., Ilbert, O., Knobel, C., Iovino, A., Maier, C., Cucciati, O., Tasca, L., Carollo, C. M., Silverman, J., Kampczyk, P., de Ravel, L., Sanders, D., Scoville, N., Contini, T., Mainieri, V., Scodeggio, M., Kneib, J.-P., Le F`evre,O., Bardelli, S., Bongiorno, A., Caputi, K., Coppa, G., de la Torre, S., Franzetti, P., Garilli, B., Lamareille, F., Le Borgne, J.-F., Le Brun, V., Mignoli, M., Perez Montero, E., Pello, R., Ricciardelli, E., Tanaka, M., Tresse, L., Vergani, D., Welikala, N., Zucca, E., Oesch, P., Abbas, U., Barnes, L., Bordoloi, R., Bottini, D., Cappi, A., Cassata, P., Cimatti, A., Fumana, M., Hasinger, G., Koekemoer, A., Leauthaud, A., Maccagni, D., Marinoni, C., McCracken, H., Memeo, P., Meneux, B., Nair, P., Porciani, C., Presotto, V., & Scaramella, R., ApJ, 721:193–221, September 2010b.

Peng, Y.-j., Lilly, S. J., Renzini, A., & Carollo, M., ApJ, 757:4, September 2012. 184 Bibliography

Persson, S. E., Murphy, D. C., Krzeminski, W., Roth, M., & Rieke, M. J., AJ, 116: 2475–2488, November 1998.

Pierre, M., Valtchanov, I., Altieri, B., Andreon, S., Bolzonella, M., Bremer, M., Disseau, L., Dos Santos, S., Gandhi, P., Jean, C., Pacaud, F., Read, A., Refregier, A., Willis, J., Adami, C., Alloin, D., Birkinshaw, M., Chiappetti, L., Cohen, A., Detal, A., Duc, P.-A., Gosset, E., Hjorth, J., Jones, L., Le F`evre,O., Lonsdale, C., Maccagni, D., Mazure, A., McBreen, B., McCracken, H., Mellier, Y., Ponman, T., Quintana, H., Rottgering, H., Smette, A., Surdej, J., Starck, J.-L., Vigroux, L., & White, S., JCAP, 9:011, September 2004.

Pimbblet, K. A., Smail, I., Kodama, T., Couch, W. J., Edge, A. C., Zabludoﬀ, A. I., & O’Hely, E., MNRAS, 331:333–350, March 2002.

Pirard, J.-F., Kissler-Patig, M., Moorwood, A., Biereichel, P., Delabre, B., Dorn, R., Fin- ger, G., Gojak, D., Huster, G., Jung, Y., Koch, F., Le Louarn, M., Lizon, J.-L., Mehrgan, L., Pozna, E., Silber, A., Sokar, B., & Stegmeier, J. HAWK-I: A new wide-ﬁeld 1- to 2.5-µm imager for the VLT. In Moorwood, A. F. M. & Iye, M., editors, Ground-based Instrumentation for Astronomy, volume 5492 of Society of Photo-Optical Instrumen- tation Engineers (SPIE) Conference Series, pages 1763–1772, September 2004. doi: 10.1117/12.578293.

Planck Collaboration, Ade, P. A. R., Aghanim, N., Armitage-Caplan, C., Arnaud, M., Ashdown, M., Atrio-Barandela, F., Aumont, J., Aussel, H., Baccigalupi, C., & et al., ArXiv e-prints, March 2013.

Poggianti, B. M., A&AS, 122:399–407, May 1997.

Poggianti, B. M. & Barbaro, G., A&A, 325:1025–1030, September 1997.

Poggianti, B. M., Smail, I., Dressler, A., Couch, W. J., Barger, A. J., Butcher, H., Ellis, R. S., & Oemler, Jr., A., ApJ, 518:576–593, June 1999.

Poggianti, B. M., Bridges, T. J., Carter, D., Mobasher, B., Doi, M., Iye, M., Kashikawa, N., Komiyama, Y., Okamura, S., Sekiguchi, M., Shimasaku, K., Yagi, M., & Yasuda, N., ApJ, 563:118–123, December 2001.

Poggianti, B. M., von der Linden, A., De Lucia, G., Desai, V., Simard, L., Halliday, C., Arag´on-Salamanca,A., Bower, R., Varela, J., Best, P., Clowe, D. I., Dalcanton, J., Jablonka, P., Milvang-Jensen, B., Pello, R., Rudnick, G., Saglia, R., White, S. D. M., & Zaritsky, D., ApJ, 642:188–215, May 2006. Bibliography 185

Popesso, P., Dickinson, M., Nonino, M., Vanzella, E., Daddi, E., Fosbury, R. A. E., Kuntschner, H., Mainieri, V., Cristiani, S., Cesarsky, C., Giavalisco, M., Renzini, A., & GOODS Team, A&A, 494:443–460, February 2009.

Postman, M., Franx, M., Cross, N. J. G., Holden, B., Ford, H. C., Illingworth, G. D., Goto, T., Demarco, R., Rosati, P., Blakeslee, J. P., Tran, K.-V., Ben´ıtez, N., Clampin, M., Hartig, G. F., Homeier, N., Ardila, D. R., Bartko, F., Bouwens, R. J., Bradley, L. D., Broadhurst, T. J., Brown, R. A., Burrows, C. J., Cheng, E. S., Feldman, P. D., Golimowski, D. A., Gronwall, C., Infante, L., Kimble, R. A., Krist, J. E., Lesser, M. P., Martel, A. R., Mei, S., Menanteau, F., Meurer, G. R., Miley, G. K., Motta, V., Sirianni, M., Sparks, W. B., Tran, H. D., Tsvetanov, Z. I., White, R. L., & Zheng, W., ApJ, 623: 721–741, April 2005.

Pozzetti, L., Bolzonella, M., Zucca, E., Zamorani, G., Lilly, S., Renzini, A., Moresco, M., Mignoli, M., Cassata, P., Tasca, L., Lamareille, F., Maier, C., Meneux, B., Halliday, C., Oesch, P., Vergani, D., Caputi, K., Kovaˇc,K., Cimatti, A., Cucciati, O., Iovino, A., Peng, Y., Carollo, M., Contini, T., Kneib, J.-P., Le F´evre,O., Mainieri, V., Scodeggio, M., Bardelli, S., Bongiorno, A., Coppa, G., de la Torre, S., de Ravel, L., Franzetti, P., Garilli, B., Kampczyk, P., Knobel, C., Le Borgne, J.-F., Le Brun, V., Pell`o,R., Perez Montero, E., Ricciardelli, E., Silverman, J. D., Tanaka, M., Tresse, L., Abbas, U., Bottini, D., Cappi, A., Guzzo, L., Koekemoer, A. M., Leauthaud, A., Maccagni, D., Marinoni, C., McCracken, H. J., Memeo, P., Porciani, C., Scaramella, R., Scarlata, C., & Scoville, N., A&A, 523:A13, November 2010.

Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. Numerical recipes in C++ : the art of scientiﬁc computing. 2002.

Proctor, R. N. & Sansom, A. E., MNRAS, 333:517–543, July 2002.

Quadri, R. F., Williams, R. J., Franx, M., & Hildebrandt, H., ApJ, 744:88, January 2012.

Raﬀerty, D. A., McNamara, B. R., & Nulsen, P. E. J., ApJ, 687:899–918, November 2008.

Reichert, A., B¨ohringer,H., Fassbender, R., & M¨uhlegger, M., A&A, 535:A4, November 2011.

Reiprich, T. H. & B¨ohringer,H., ApJ, 567:716–740, March 2002.

Rettura, A., Mei, S., Stanford, S. A., Raichoor, A., Moran, S., Holden, B., Rosati, P., Ellis, R., Nakata, F., Nonino, M., Treu, T., Blakeslee, J. P., Demarco, R., Eisenhardt, 186 Bibliography

P., Ford, H. C., Fosbury, R. A. E., Illingworth, G., Huertas-Company, M., Jee, M. J., Kodama, T., Postman, M., Tanaka, M., & White, R. L., ApJ, 732:94, May 2011.

Retzlaﬀ, J., Rosati, P., Dickinson, M., Vandame, B., Rit´e,C., Nonino, M., Cesarsky, C., & GOODS Team, A&A, 511:A50, February 2010.

Roettiger, K., Loken, C., & Burns, J. O., ApJS, 109:307–332, April 1997.

Romeo, A. D., Napolitano, N. R., Covone, G., Sommer-Larsen, J., Antonuccio-Delogu, V., & Capaccioli, M., MNRAS, 389:13–26, September 2008.

Rosati, P., Della Ceca, R., Norman, C., & Giacconi, R., ApJ, 492:L21–L24, January 1998.

Rosati, P., Tozzi, P., Ettori, S., Mainieri, V., Demarco, R., Stanford, S. A., Lidman, C., Nonino, M., Borgani, S., Della Ceca, R., Eisenhardt, P., Holden, B. P., & Norman, C., AJ, 127:230–238, January 2004.

Rose, J. A., AJ, 90:1927–1956, October 1985.

Rubin, D., Aldering, G., Amanullah, R., Barbary, K., Dawson, K. S., Deustua, S., Faccioli, L., Fadeyev, V., Fakhouri, H. K., Fruchter, A. S., Gladders, M. D., de Jong, R. S., Koekemoer, A., Krechmer, E., Lidman, C., Meyers, J., Nordin, J., Perlmutter, S., Ripoche, P., Schlegel, D. J., Spadafora, A., Suzuki, N., & Supernova Cosmology Project, T., ArXiv e-prints, February 2015.

Rudick, C. S., Mihos, J. C., Harding, P., Feldmeier, J. J., Janowiecki, S., & Morrison, H. L., ApJ, 720:569–580, September 2010.

Rudnick, G., von der Linden, A., Pell´o,R., Arag´on-Salamanca,A., Marchesini, D., Clowe, D., De Lucia, G., Halliday, C., Jablonka, P., Milvang-Jensen, B., Poggianti, B., Saglia, R., Simard, L., White, S., & Zaritsky, D., ApJ, 700:1559–1588, August 2009.

Rudnick, G. H., Tran, K.-V., Papovich, C., Momcheva, I., & Willmer, C., ApJ, 755:14, August 2012.

Sahl´en,M., Viana, P. T. P., Liddle, A. R., Romer, A. K., Davidson, M., Hosmer, M., Lloyd-Davies, E., Sabirli, K., Collins, C. A., Freeman, P. E., Hilton, M., Hoyle, B., Kay, S. T., Mann, R. G., Mehrtens, N., Miller, C. J., Nichol, R. C., Stanford, S. A., & West, M. J., MNRAS, 397:577–607, August 2009.

Salpeter, E. E., ApJ, 121:161, January 1955. Bibliography 187

Sánchez-Blázquez,P., Jablonka, P., Noll, S., Poggianti, B. M., Moustakas, J., Milvang- Jensen, B., Halliday, C., Aragón-Salamanca,A., Saglia, R. P., Desai, V., De Lucia, G., Clowe, D. I., Pelló,R., Rudnick, G., Simard, L., White, S. D. M., & Zaritsky, D., A&A, 499:47–68, May 2009.

Sandage, A. & Visvanathan, N., ApJ, 223:707–729, August 1978.

Santos, J. S., Rosati, P., Gobat, R., Lidman, C., Dawson, K., Perlmutter, S., B¨ohringer, H., Balestra, I., Mullis, C. R., Fassbender, R., Kohnert, J., Lamer, G., Rettura, A., Rit´e,C., & Schwope, A., A&A, 501:49–60, July 2009.

Santos, J. S., Altieri, B., Tanaka, M., Valtchanov, I., Saintonge, A., Dickinson, M., Fou- caud, S., Kodama, T., Rawle, T. D., & Tadaki, K., MNRAS, 438:2565–2577, March 2014.

Schechter, P., ApJ, 203:297–306, January 1976.

Schlegel, D. J., Finkbeiner, D. P., & Davis, M., ApJ, 500:525, June 1998.

Scodeggio, M., AJ, 121:2413–2419, May 2001.

Skelton, R. E., Bell, E. F., & Somerville, R. S., ApJ, 753:44, July 2012.

Smail, I., Dressler, A., Couch, W. J., Ellis, R. S., Oemler, Jr., A., Butcher, H., & Sharples, R. M., ApJS, 110:213, June 1997.

Snyder, G. F., Brodwin, M., Mancone, C. M., Zeimann, G. R., Stanford, S. A., Gonzalez, A. H., Stern, D., Eisenhardt, P. R. M., Brown, M. J. I., Dey, A., Jannuzi, B., & Perlmutter, S., ApJ, 756:114, September 2012.

Somerville, R. S., Hopkins, P. F., Cox, T. J., Robertson, B. E., & Hernquist, L., MNRAS, 391:481–506, December 2008.

Spolaor, M., Kobayashi, C., Forbes, D. A., Couch, W. J., & Hau, G. K. T., MNRAS, 408: 272–292, October 2010.

Strateva, I., Ivezi´c, Z.,ˇ Knapp, G. R., Narayanan, V. K., Strauss, M. A., Gunn, J. E., Lupton, R. H., Schlegel, D., Bahcall, N. A., Brinkmann, J., Brunner, R. J., Budav´ari,T., Csabai, I., Castander, F. J., Doi, M., Fukugita, M., Gy˝ory,Z., Hamabe, M., Hennessy, G., Ichikawa, T., Kunszt, P. Z., Lamb, D. Q., McKay, T. A., Okamura, S., Racusin, J., Sekiguchi, M., Schneider, D. P., Shimasaku, K., & York, D., AJ, 122:1861–1874, October 2001. 188 Bibliography

Suzuki, N., Rubin, D., Lidman, C., Aldering, G., Amanullah, R., Barbary, K., Barrientos, L. F., Botyanszki, J., Brodwin, M., Connolly, N., Dawson, K. S., Dey, A., Doi, M., Donahue, M., Deustua, S., Eisenhardt, P., Ellingson, E., Faccioli, L., Fadeyev, V., Fakhouri, H. K., Fruchter, A. S., Gilbank, D. G., Gladders, M. D., Goldhaber, G., Gonzalez, A. H., Goobar, A., Gude, A., Hattori, T., Hoekstra, H., Hsiao, E., Huang, X., Ihara, Y., Jee, M. J., Johnston, D., Kashikawa, N., Koester, B., Konishi, K., Kowalski, M., Linder, E. V., Lubin, L., Melbourne, J., Meyers, J., Morokuma, T., Munshi, F., Mullis, C., Oda, T., Panagia, N., Perlmutter, S., Postman, M., Pritchard, T., Rhodes, J., Ripoche, P., Rosati, P., Schlegel, D. J., Spadafora, A., Stanford, S. A., Stanishev, V., Stern, D., Strovink, M., Takanashi, N., Tokita, K., Wagner, M., Wang, L., Yasuda, N., Yee, H. K. C., & Supernova Cosmology Project, T., ApJ, 746:85, February 2012.

Tanaka, M., Kodama, T., Arimoto, N., Okamura, S., Umetsu, K., Shimasaku, K., Tanaka, I., & Yamada, T., MNRAS, 362:268–288, September 2005.

Tanaka, M., De Breuck, C., Venemans, B., & Kurk, J., A&A, 518:A18, July 2010.

Tanaka, M., Toft, S., Marchesini, D., Zirm, A., De Breuck, C., Kodama, T., Koyama, Y., Kurk, J., & Tanaka, I., ApJ, 772:113, August 2013.

Taranu, D. S., Hudson, M. J., Balogh, M. L., Smith, R. J., Power, C., Oman, K. A., & Krane, B., MNRAS, 440:1934–1949, May 2014.

Thomas, D., Maraston, C., & Bender, R., MNRAS, 339:897–911, March 2003.

Thomas, D., Maraston, C., Bender, R., & Mendes de Oliveira, C., ApJ, 621:673–694, March 2005.

Thomas, D., Maraston, C., & Johansson, J., MNRAS, 412:2183–2198, April 2011.

Tinsley, B. M., ApJ, 151:547, February 1968.

Tran, K.-V. H., Franx, M., Illingworth, G. D., van Dokkum, P., Kelson, D. D., Blakeslee, J. P., & Postman, M., ApJ, 661:750–767, June 2007.

Tran, K.-V. H., Saintonge, A., Moustakas, J., Bai, L., Gonzalez, A. H., Holden, B. P., Zaritsky, D., & Kautsch, S. J., ApJ, 705:809–820, November 2009.

Tran, K.-V. H., Papovich, C., Saintonge, A., Brodwin, M., Dunlop, J. S., Farrah, D., Finkelstein, K. D., Finkelstein, S. L., Lotz, J., McLure, R. J., Momcheva, I., & Willmer, C. N. A., ApJ, 719:L126–L129, August 2010. Bibliography 189

Trenti, M., Bradley, L. D., Stiavelli, M., Oesch, P., Treu, T., Bouwens, R. J., Shull, J. M., MacKenty, J. W., Carollo, C. M., & Illingworth, G. D., ApJ, 727:L39, February 2011.

Valentinuzzi, T., Woods, D., Fasano, G., Riello, M., D’Onofrio, M., Varela, J., Bettoni, D., Cava, A., Couch, W. J., Dressler, A., Fritz, J., Moles, M., Omizzolo, A., Poggianti, B. M., & Kjærgaard, P., A&A, 501:851–864, July 2009.

Valentinuzzi, T., Poggianti, B. M., Fasano, G., D’Onofrio, M., Moretti, A., Ramella, M., Biviano, A., Fritz, J., Varela, J., Bettoni, D., Vulcani, B., Moles, M., Couch, W. J., Dressler, A., Kjærgaard, P., Omizzolo, A., & Cava, A., A&A, 536:A34, December 2011. van Dokkum, P. G., PASP, 113:1420–1427, November 2001. van Dokkum, P. G., Franx, M., Kelson, D. D., Illingworth, G. D., Fisher, D., & Fabricant, D., ApJ, 500:714, June 1998.

Vanzella, E., Cristiani, S., Dickinson, M., Kuntschner, H., Moustakas, L. A., Nonino, M., Rosati, P., Stern, D., Cesarsky, C., Ettori, S., Ferguson, H. C., Fosbury, R. A. E., Giavalisco, M., Haase, J., Renzini, A., Rettura, A., Serra, P., & GOODS Team, A&A, 434:53–65, April 2005.

Vanzella, E., Cristiani, S., Dickinson, M., Kuntschner, H., Nonino, M., Rettura, A., Rosati, P., Vernet, J., Cesarsky, C., Ferguson, H. C., Fosbury, R. A. E., Giavalisco, M., Grazian, A., Haase, J., Moustakas, L. A., Popesso, P., Renzini, A., Stern, D., & GOODS Team, A&A, 454:423–435, August 2006.

Vanzella, E., Cristiani, S., Dickinson, M., Giavalisco, M., Kuntschner, H., Haase, J., Nonino, M., Rosati, P., Cesarsky, C., Ferguson, H. C., Fosbury, R. A. E., Grazian, A., Moustakas, L. A., Rettura, A., Popesso, P., Renzini, A., Stern, D., & GOODS Team, A&A, 478:83–92, January 2008.

Vulcani, B., Poggianti, B. M., Arag´on-Salamanca,A., Fasano, G., Rudnick, G., Valentin- uzzi, T., Dressler, A., Bettoni, D., Cava, A., D’Onofrio, M., Fritz, J., Moretti, A., Omizzolo, A., & Varela, J., MNRAS, 412:246–268, March 2011a.

Vulcani, B., Poggianti, B. M., Dressler, A., Fasano, G., Valentinuzzi, T., Couch, W., Moretti, A., Simard, L., Desai, V., Bettoni, D., D’Onofrio, M., Cava, A., & Varela, J., MNRAS, 413:921–941, May 2011b.

White, S. D. M., Clowe, D. I., Simard, L., Rudnick, G., De Lucia, G., Arag´on-Salamanca, A., Bender, R., Best, P., Bremer, M., Charlot, S., Dalcanton, J., Dantel, M., Desai, V., 190 Bibliography

Fort, B., Halliday, C., Jablonka, P., Kauﬀmann, G., Mellier, Y., Milvang-Jensen, B., Pell´o,R., Poggianti, B., Poirier, S., Rottgering, H., Saglia, R., Schneider, P., & Zaritsky, D., A&A, 444:365–379, December 2005.

Wilson, G., ApJ, 585:191–204, March 2003.

Windhorst, R. A., Taylor, V. A., Jansen, R. A., Odewahn, S. C., Chiarenza, C. A. T., Conselice, C. J., de Grijs, R., de Jong, R. S., MacKenty, J. W., Eskridge, P. B., Frogel, J. A., Gallagher, III, J. S., Hibbard, J. E., Matthews, L. D., & O’Connell, R. W., ApJS, 143:113–158, November 2002.

Wirth, G. D., Willmer, C. N. A., Amico, P., Chaﬀee, F. H., Goodrich, R. W., Kwok, S., Lyke, J. E., Mader, J. A., Tran, H. D., Barger, A. J., Cowie, L. L., Capak, P., Coil, A. L., Cooper, M. C., Conrad, A., Davis, M., Faber, S. M., Hu, E. M., Koo, D. C., Le Mignant, D., Newman, J. A., & Songaila, A., AJ, 127:3121–3136, June 2004.

Wolf, C., Aragón-Salamanca,A., Balogh, M., Barden, M., Bell, E. F., Gray, M. E., Peng, C. Y., Bacon, D., Barazza, F. D., Böhm,A., Caldwell, J. A. R., Gallazzi, A., Häußler, B., Heymans, C., Jahnke, K., Jogee, S., van Kampen, E., Lane, K., McIntosh, D. H., Meisenheimer, K., Papovich, C., Sánchez, S. F., Taylor, A., Wisotzki, L., & Zheng, X., MNRAS, 393:1302–1323, March 2009.

Worthey, G., ApJS, 95:107–149, November 1994.

Worthey, G. & Ottaviani, D. L., ApJS, 111:377–386, August 1997.

York, D. G., Adelman, J., Anderson, Jr., J. E., Anderson, S. F., Annis, J., Bahcall, N. A., Bakken, J. A., Barkhouser, R., Bastian, S., Berman, E., Boroski, W. N., Bracker, S., Briegel, C., Briggs, J. W., Brinkmann, J., Brunner, R., Burles, S., Carey, L., Carr, M. A., Castander, F. J., Chen, B., Colestock, P. L., Connolly, A. J., Crocker, J. H., Csabai, I., Czarapata, P. C., Davis, J. E., Doi, M., Dombeck, T., Eisenstein, D., Ellman, N., Elms, B. R., Evans, M. L., Fan, X., Federwitz, G. R., Fiscelli, L., Friedman, S., Frieman, J. A., Fukugita, M., Gillespie, B., Gunn, J. E., Gurbani, V. K., de Haas, E., Haldeman, M., Harris, F. H., Hayes, J., Heckman, T. M., Hennessy, G. S., Hindsley, R. B., Holm, S., Holmgren, D. J., Huang, C.-h., Hull, C., Husby, D., Ichikawa, S.-I., Ichikawa, T., Ivezi´c, Z.,ˇ Kent, S., Kim, R. S. J., Kinney, E., Klaene, M., Kleinman, A. N., Kleinman, S., Knapp, G. R., Korienek, J., Kron, R. G., Kunszt, P. Z., Lamb, D. Q., Lee, B., Leger, R. F., Limmongkol, S., Lindenmeyer, C., Long, D. C., Loomis, C., Loveday, J., Lucinio, R., Lupton, R. H., MacKinnon, B., Mannery, E. J., Mantsch, Bibliography 191

P. M., Margon, B., McGehee, P., McKay, T. A., Meiksin, A., Merelli, A., Monet, D. G., Munn, J. A., Narayanan, V. K., Nash, T., Neilsen, E., Neswold, R., Newberg, H. J., Nichol, R. C., Nicinski, T., Nonino, M., Okada, N., Okamura, S., Ostriker, J. P., Owen, R., Pauls, A. G., Peoples, J., Peterson, R. L., Petravick, D., Pier, J. R., Pope, A., Pordes, R., Prosapio, A., Rechenmacher, R., Quinn, T. R., Richards, G. T., Richmond, M. W., Rivetta, C. H., Rockosi, C. M., Ruthmansdorfer, K., Sandford, D., Schlegel, D. J., Schneider, D. P., Sekiguchi, M., Sergey, G., Shimasaku, K., Siegmund, W. A., Smee, S., Smith, J. A., Snedden, S., Stone, R., Stoughton, C., Strauss, M. A., Stubbs, C., SubbaRao, M., Szalay, A. S., Szapudi, I., Szokoly, G. P., Thakar, A. R., Tremonti, C., Tucker, D. L., Uomoto, A., Vanden Berk, D., Vogeley, M. S., Waddell, P., Wang, S.-i., Watanabe, M., Weinberg, D. H., Yanny, B., Yasuda, N., & SDSS Collaboration, AJ, 120:1579–1587, September 2000.

Yuan, T., Nanayakkara, T., Kacprzak, G. G., Tran, K.-V. H., Glazebrook, K., Kewley, L. J., Spitler, L. R., Poole, G. B., Labb´e,I., Straatman, C. M. S., & Tomczak, A. R., ApJ, 795:L20, November 2014.

Appendices

193

A Morphological Classiﬁcation

The details of the morphological classification of red sequence members in XMM1229 are described in Section 3.4.4 and Appendix B. In this appendix we present thumbnail images of the morphologically classified galaxies in each of the subsamples presented in Table 3.3. The postage stamp images were taken from the HST/ACS F850LP image of the XMM1229 field, which was used for morphological classification. Table 3.3 also lists positions and photometric and/or spectroscopic redshifts of all the objects. The images presented in the following figures belong, in the succession, to the following subsets: cluster centre, cluster outskirts, spectroscopically confirmed members that are excluded because they are not in the red sequence or because the photometry is inaccurate, spectroscopically confirmed members excluded from the analysis because they lie outside the area covered by the ACS images. For the last subsample the cutout images were taken from the HAWK-I Ks image. In each sample the objects are ordered, according to their morphological type, in the following order: elliptical, bulge-dominated, disc-dominated (early), disc-dominated (late), irregular. The classification for objects in the cluster outskirts corresponds to the output of galSVM. No visual classification was performed on this sample in the present work. The size of the ACS F850LP cutout images is 4.0500 on each side, corresponding to a physical size of 32.5 kpc at z = 0.98. The size of the HAWK-I Ks cutout images is instead 8.100 on each side, corresponding to a physical size of 65 kpc at z = 0.98

195 196 Appendix A. Morphological Classiﬁcation

XMM1229 229 XMM1229 237 XMM1229 241 XMM1229 244 XMM1229 243 XMM1229 260

elliptical elliptical elliptical elliptical elliptical elliptical

XMM1229 262 XMM1229 265 XMM1229 287 XMM1229 312 XMM1229 320 XMM1229 331

elliptical elliptical elliptical elliptical elliptical elliptical

XMM1229 353 XMM1229 380 XMM1229 145 XMM1229 475 XMM1229 470 XMM1229 415

elliptical elliptical elliptical elliptical elliptical elliptical

XMM1229 394 XMM1229 172 XMM1229 200 XMM1229 240 XMM1229 248 XMM1229 288

elliptical bulge-domin. bulge-domin. bulge-domin. bulge-domin. bulge-domin.

XMM1229 286 XMM1229 283 XMM1229 291 XMM1229 309 XMM1229 310 XMM1229 319

bulge-domin. bulge-domin. bulge-domin. bulge-domin. bulge-domin. bulge-domin.

XMM1229 322 XMM1229 429 XMM1229 437 XMM1229 441 XMM1229 457 XMM1229 477

bulge-domin. bulge-domin. bulge-domin. bulge-domin. bulge-domin. bulge-domin.

XMM1229 502 XMM1229 128 XMM1229 190 XMM1229 255 XMM1229 263 XMM1229 306

disc-domin. disc-domin. disc-domin. disc-domin. disc-domin. bulge-domin. (early) (early) (early) (early) (early)

XMM1229 392 XMM1229 456 XMM1229 463 XMM1229 414

disc-domin. disc-domin. disc-domin. disc-domin. (early) (early) (early) (late)

Figure A.1 Morphologically classified red sequence galaxies in the central region (i.e. within 0.6 Mpc from the cluster centre) . Position and redshift (photometric and/or spectroscopic) of each object are listed in Table 3.3. The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field used in morphological classification (see Section 3.4.4). 197

XMM1229 73 XMM1229 373 XMM1229 349

elliptical elliptical bulge-domin.

Figure A.2 Morphologically classified red sequence galaxies in the cluster outskirts (i.e. between 0.6 Mpc and 1.04 Mpc from the cluster centre). Only spectroscopically confirmed cluster members are shown as in this work we do not determine the membership of individual galaxies for this subsample. Position and redshift of each object are listed in Table 3.3. The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field used in morphological classification (see Section 3.4.4). The morphological classification of the objects in this subsample corresponds to the outcome of galSVM.

XMM1229 308 XMM1229 183 XMM1229 316 XMM1229 487

disc-domin. disc-domin. disc-domin. (early) (early) (late) irregular

Figure A.3 Spectroscopically confirmed XMM1229 members excluded from the analysis in this paper. XMM1229 183, XMM1229 308, XMM1229 487 are blue cloud galaxies, while XMM1229 316 was excluded because SExtractor was unable to return a reliable estimate of the aperture magnitude (SExtractor FLAGS=16). The image cutouts were taken from the HST/ACS F850LP image of the XMM1229 field. For this subsample the visual morphological classification was performed by P. C. . 198 Appendix A. Morphological Classification

FORS2 4661 FORS2 4800 FORS2 4956 FORS2 5001 FORS2 4794 FORS2 4910

disc-domin. elliptical elliptical elliptical elliptical bulge-domin. (early)

Figure A.4 Spectroscopically confirmed XMM1229 members excluded from the analysis in this paper because falling outside the ACS field. The classification was performed visually by P. C. on image cutouts taken from the HAWK-I Ks band image of the XMM1229 field and shown in this figure. B Automatic Morphological Classification

galSVM is based on Support Vector Machines (SVM), a particular family of machine- learning algorithms that, given a training sample of objects with known classification, fit a hyperplane in the space of the measured parameters to separate between two particular classes. In the case of galaxy morphology, the parameter space is defined by a certain number of morphological coefficients measured on the images. For the XMM1229 field, we measured and used the following seven non-parametric morphological coefficients:

Concentration (C)

The concentration index defines the amount of light contained in the inner galaxy region relative to the total galaxy flux. We use two complementary definitions of this parameter, the first based on the flux ratio within an inner and an outer aperture (CAbraham, Abraham et al. 1996) and the other based on the ratio between an outer radius and an inner radius

(CConcelice, Conselice et al. 2000). The equations deﬁning the concentration index are therefore: P I(i, j)in CAbraham = P (B.1) I(i, j)out

rout CConselice = 5 log (B.2) rin where I(i, j)in and I(i, j)out are respectively the intensities at position (i, j) within the inner and outer radius, while rin and rout are respectively the inner and outer radii.

In our run of galSVM we used both CAbraham and CConselice. Huertas-Company et al. (2008) shows, in fact, that an increase in the number of parameters in a SVM based classiﬁcation produces higher precision in the ﬁnal result without introducing degeneracies.

199 200 Appendix B. Automatic Morphology

Asymmetry (A)

The asymmetry parameter measures the degree of rotational symmetry of an object and is deﬁned by the following equation:

1 P I(i, j) I (i, j) P B(i, j) B (i, j) A = | − 180 | | − 180 | (B.3) 2 P I(i, j) − P I(i, j)

◦ where I(i, j) and I180(i, j) are respectively the image and its 180 rotated counterpart, ◦ while B(i, j) and B180(i, j) are respectively the image background and its 180 rotated counterpart. Elliptical and S0 galaxies tend to have higher concentration and lower asymmetry than spiral and irregular galaxies.

Gini Coeﬃcient (G)

The Gini coefficient (Abraham et al., 2003) was first introduced in Economics to measure the distribution of wealth across a given population. Translated to galaxy images it quantifies the homogeneity of a galaxy’s light distribution and is expressed by the following equation: n 1 X G = (2i n 1)I(i, j) (B.4) In¯ (n 1) − − − i where I¯ is the average intensity across the galaxy image and n > 2 is the total number of pixels in the image. In general concentration and Gini coefficient are correlated; however G is independent from the spatial distribution of light and, therefore, an actively star forming disc galaxy may have low C and high G.

M20

The second order moment of the light distribution (M20, Lotz et al. 2004) quantiﬁes the spatial distribution of substructures within galaxies (e. g. bars, bright nuclei, star clusters), and is deﬁned by the equation:

P f [(x x )2 + (y y )2] M = log i i i c i c (B.5) 20 Pn − 2 − 2 fi[(xi xc) + (yi yc) ] i − − th In this equation fi represents the ﬂux of the i pixel of the galaxy, and (xi, yi) and (xc, yc) are respectively the coordinates of the ith pixel and the central pixel. The extent of each galaxy is estimated on the SExtractor segmentation image. The sum at the numerator is performed on the 20 % brightest pixels (from which the subscript 20 comes). 201

Smoothness (S)

The smoothness parameter (Conselice et al., 2003) quantiﬁes the degree of substructure within a galaxy and it is deﬁned by:

P P 1 I(i, j) IS(i, j) B(i, j) BS(i, j) S = | − | | − | (B.6) 2 P I(i, j) − P I(i, j) where IS is the galaxy image, smoothed by a boxcar function having width 0.25rp, with rp the galaxy petrosian radius. BS is the background estimated on the smoothed image.

Ellipticity

In addition to the previous six quantities, galSVM uses the ellipticity parameter measured by SExtractor and defined as: b e = 1 (B.7) − a where a and b are the ellipse parameters associated with the ellipse semi-major and semi- minor axes computed for each galaxy. We refer to Huertas-Company et al. (2008) for a more detailed explanation of these quantities. The training set used in this classification was generated from a sample of ∼ 14,000 visually classified galaxies from g-band images of the SDSS DR7 (Nair & Abraham, 2010). In order to reproduce as much as possible the conditions of high redshift galaxies, the images were degraded according to the zphot and F850LP MAG AUTO distributions in the XMM1229 field. The advantage of this technique is that it takes into account the effect of both redshift and flux on the detection of morphological features and at the end of the degradation process one has a sample of galaxies of known morphologies as they would appear if they were at the redshift of the galaxies in the XMM1229 F850LP image. SVM are binary classifiers and, in order to reproduce a morphological scheme similar to the one adopted in the visual classification, galSVM needed to be run more than once. In the first run, galSVM used the entire training sample to classify galaxies into early- and late-type. Then, two subsequent runs, using only early- or late-type galaxies in the training sample, allowed the software to split the early-type broad class into elliptical and S0 galaxies, and the late-type broad class into early discs and late discs plus irregular galaxies. In order to estimate the uncertainty on the classification, galSVM implements a scheme of posterior probabilities which quantify the likelihood of the classes assigned by the SVM classifier. The morphological type is the one that maximises the composite 202 Appendix B. Automatic Morphology conditional probability: P (T ) = P (BT )P (T BT ) (B.8) | where T is the morphological type and BT is the broad morphological class, which can be either early- (ETG) or late-type (LTG). We assigned morphological classes to the galaxies classified by galSVM according to the following criterion:

ellipticals: galaxies with P (ETG) > 0.5 and P (E ETG) > 0.5; • |

bulge dominated: galaxies with P (ETG) > 0.5 and P (E ETG) 0.5; • | ≤

early disc-dominated: galaxies with P (ETG) 0.5 and P (EDD LT G) > 0.5; • ≤ |

late disc-dominated plus irregulars: galaxies with P (ETG) 0.5 and P (EDD LT G) • ≤ | ≤ 0.5;

With this definition, each galaxy has still a finite but lower probability of belonging to a morphological class different from that assigned, and this quantifies the uncertainty on the classification. Figure B1 shows the distributions of concentration, asymmetry, Gini coefficient and

M20 for red sequence galaxies in the centre of XMM1229 classified as described in Section 3.4.4. Types were assigned to each galaxy following a majority rule in which the morphological class was defined as the mode of the four independent classifications. As expected by e.g. Lotz et al. (2004), disc-dominated galaxies have lower values of concentration and

Gini coeﬃcient. However, the values of M20 appear still comparable with those of elliptical and S0 galaxies. This can be attributed to the fading of the spiral arms in gas-poor spiral galaxies. 203

0.25 0.65

0.20 0.60

0.15 0.55

0.10 0.50

Asymmetry 0.05 0.45 Gini Coefﬁcient 0.00 0.40

0.05 0.35 − 1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0 Concentration Concentration 1.0 0.25 − 1.2 − 0.20 1.4 − 0.15 20 1.6 M − 0.10

1.8 Asymmetry − 2.0 0.05 − 2.2 0.00 − 1.5 2.0 2.5 3.0 3.5 4.0 0.4 0.5 0.6 0.7 Concentration Gini Coefﬁcient

Figure B.1 Morphological parameters for red sequence galaxies in the centre of XMM1229 (Rcluster < 0.54 R200). Symbols and colours are the same used in Fig. 3.5. As expected, disc-dominated× galaxies tend to have smaller values of Concentration and Gini coeﬃcient. All morphological types show comparable values of M20.