ENVIRONMENTAL EFFECTS ON GALAXY EVOLUTION IN …astronomy.swin.edu.au/~lcortese/PhD_Thesis_files/main.pdf · (b) Comprendere le propriet a delle polveri interstellari respon- sabili

UNIVERSITA’ DEGLI STUDI DI MILANO-BICOCCAScuola di Dottorato di Scienze

Corso di Dottorato di Ricerca in Fisica e AstronomiaXVIII ciclo

UNIVERSITE DE PROVENCE AIX-MARSEILLE IEcole Doctorale ”Physique et Sciences de la Matiere”

Doctorat en Rayonnement et PlasmasA.A.2004-2005

ENVIRONMENTAL EFFECTS ON GALAXY

EVOLUTION IN NEARBY CLUSTERS

Coordinatore del Dottorato: Prof. Claudio DestriDirecteur de l’Ecole Doctorale: Prof. Jean-Jacques Aubert

Tutore: Prof. Giuseppe GavazziDirecteur de these: Dott. Alessandro Boselli

Commissione-Jury:Dott. A. Boselli (Laboratoire d’Astrophysique de Marseille)Prof. V. Buat (Universite de Provence)Prof. G. Gavazzi (Universita di Milano - Bicocca)Prof. F. Haardt (Universita dell’Insubria)Rapporteurs:Prof. C. Balkowski (Observatoire Astronomique Paris-Meudon)Dott. B. Poggianti (Osservatorio Astronomico di Padova)

Tesi di Dottorato di:

Luca CorteseMatricola R00280

”Objectivity cannot be equated with mental blankness; rather,objectivity resides in recognizing your preferences and

then subjecting them to especially harsh scrutiny...and also in a willingness to revise or

abandon your theories when thetests fail (as they usually do).”

Stephen Jay Gould

Acknowledgments

This work represents the end point of my student career. After approximately twentyone years from my first entrance in a class room (it was September 1984 in Phoenix,AZ), I’m finally going to attend my last ”school” examination. Therefore I want toseize this opportunity in order to briefly remember and to thank some of the friendsmet during this journey.First of all I must thank my advisor Peppo Gavazzi, my scientific father, for hisprecious guidance and his teachings especially at the beginning of my research carrier.Special thanks to Alessandro Boselli, my co-advisor, first of all for the the last yearspent in Marseille: a splendid experience. Thanks also for all his helpful advices,comments and supports on this and other works during the last three years.Many people contributed, directly or indirectly, to this work, and I am grateful to allof them. Merci beaucoup to Samuel Boissier for all the interesting discussions and,above all, for his precious lessons of French. Thanks to Veronique Buat for his helpduring the year spent in Marseille and for having initiated me in the obscure secrets ofdust. Thanks to Barry Madore for his hospitality at the Carnegie Observatories, forhis kindness, support and, especially, for his help in improving my written English.Muchas gracias to Armando Gil de Paz for his precious help on making the GALEXdata available to me: without his contribution a great part of this work would not havebeen possible. Many thanks to Bianca Poggianti for a careful reading of my thesisand for her useful comments and suggestions. I would like also to thank Monica Colpifor her scientific and, especially, financial support during these three years. Arigatoto Tsutomu Takeuchi and Akio Inoue for useful discussions about dust and galaxyevolution, for their kindness and help during my stay in Marseille and for havingintroduced me to Japanese cuisine.Many friends made the last three years unique. At Milano University life wouldn’thave been as much fun without all Peppo’s students. In particular thanks to Ilaria,Lea and Paolo for their unique support and thanks also to Chri for having installedLinux on my laptop, making me able to write this work.In Marseille thanks a lot to all the ”Cafe du Coin”: Helene, Claude, Kassem, Peter,Fabrice and the others. Thanks for all the coffees and cakes, and for having receivedme with open arms even if I wasn’t able to speak French. Thanks to Alexie, Jean-

v

vi

Baptiste, Hector who shared the office with me, and a special thanks to Celine forhaving borne my never ending phone calls with my advisors, for her kindness and forher precious help in understanding french bureaucracy. Life in Marseille would havebeen completely different without the volley matches with Raph, Patrick, Seb, Mika,Fabrice and all the others.Finally, nothing of this would have been possible without the constant support of myparents and my brother Claudio, who have always encouraged me to continue thisbeautiful adventure.

This research was partly supported the Universita Italo-Francese through theVinci Programme and by the CNES through GALEX-Marseille.

Abstract

The environmental effects on galaxy evolution in nearby clusters are investigatedusing a multiwavelength dataset. The present analysis is focused on the propertiesof three (Abell 1367, Virgo and Coma) among the best studied clusters in the localUniverse. Due to the variety of their environmental conditions (e.g. spiral fraction,X-ray luminosity, evolutionary stage) they represent the most suitable ”laboratory”for comparative studies. By combining for the first time GALEX UV observationswith optical, near and far infrared data, the evolutionary history of cluster galaxiesis studied. The main goals of this thesis are: (a) The study of the dependence of theUV emission of galaxies from their morphological type, mass and the environmentthey inhabit, through the study of UV luminosity functions and color magnituderelations. (b) The study of UV dust extinction properties of local cluster galaxiesand investigation of possible empirical relations useful to estimate the amount of UVattenuation in local and high redshift galaxies. (c) Investigation of the effect of largescale structures assembling on galaxy evolution through the dynamical analysis ofAbell 1367, one of the best examples of a dynamically young local cluster of galaxies.(d) The characterization of the effects of different environmental mechanisms (i.e.gravitation interactions, ram pressure, preprocessing) on the evolutionary history ofcluster galaxies in order to gain more insight on the origin of the morphology-densityand star-formation-density relations.The observational evidences presented in this work suggest that: (I) Giant ellipticalsare an old, homogeneous population showing no or little evolution at least in thepast 8 Gyr; unlike dwarf ellipticals which still contain young stellar populations. (II)The importance of different environmental mechanisms has changed during the ageof the Universe. Tidal interactions and preprocessing probably dominated the pastUniverse and shaped part of the morphology-density relation during the phase ofcluster accretion of small groups. Ram pressure dominates in today clusters and issurely affecting the star formation history of galaxies but with less influence on theirmorphology. (III) The heterogeneous class of S0s galaxies, from bulge dominated tothe disky S0s, is not the result of a single transformation mechanism: if ram pressureis able to produce disk dominated S0s, tidal interactions (and thus preprocessing) arerequired to account for bulge dominated S0s. (VI) Different observational evidences

vii

viii

confirm the presence of a correlation between the mean age of stellar populationsand galaxy mass (downsizing effect). In the framework of the hierarchical model ofgalaxy formation, the origin of the downsizing effect remains unsolved. This clearobservational evidences represents one of today’s main challenge for models of galaxyevolution.

Riassunto

In questo lavoro vengono analizzati gli effetti dell’ambiente sull’evoluzione delle galassie,utilizzando una base di dati multi-lunghezza d’onda. In particolare tutta quest’analisie focalizzata sullo studio di tre differenti ammassi di galassie dell’Universo Locale:Abell1367, Virgo, Coma. Questi tre ammassi sono tra i piu studiati nell’Universolocale e, date le loro differenti proprieta (e.g. frazione di galassie a spirale, luminositaX, stadio evolutivo), rappresentano dei laboratori ideali per quantificare l’influenzadell’ambiente sull’evoluzione delle galassie. Combinando per la prima volta osser-vazioni ultraviolette del satellite GALEX a dati ottici, in vicino e lontano infrarossoviene ricostruita l’evoluzione delle galassie d’ammasso.I principali obiettivi di questa tesi sono: (a) Studiare il legame tra le proprietadell’emissione UV delle galassie, il loro tipo morfologico, la loro massa e l’ambiente incui esse si trovano, attraverso l’analisi delle funzioni di luminosita UV e delle relazionicolore-magnitudine. (b) Comprendere le proprieta delle polveri interstellari respon-sabili dell’assorbimento della radiazione ultravioletta e ricavare relazioni empiricheutili per poter quantificare l’assorbimento della radiazione ultravioletta in assenzadi osservazioni in lontano infrarosso. (c) Analizzare se e come lo stato dinamico diun ammasso e in grado di influenzare la storia evolutiva delle galassie, attraverso lostudio dell’ammasso di Abell1367: uno dei migliori esempi di ammasso locale, dinami-camente ancora giovane. (d) Quantificare l’influenza di diversi effetti d’ambiente (i.e.interazioni gravitazionali, ram-pressure, galaxy preprocessing) sull’evoluzione dellegalassie d’ammasso, in modo da comprendere le origini del fenomeno di segregazionemorfologica.Tutte le evidenze osservative presentate e analizzate in questo lavoro suggerisconoche: (I) Le ellittiche giganti rappresentano una popolazione vecchia, omogenea chenon ha subito una significativa evoluzione negli ultimi 8 Gyr; al contrario dell’ellittichenane che sono ancora oggi dominate da popolazioni stellari giovani. (II) L’influenzadell’ambiente sull’evoluzione delle galassie cambia sensibilmente con l’eta dell’Universo.Le interazioni gravitazionali ed il galaxy preprocessing sono stati gli effetti dominantinell’Universo passato e sembrano essere i responsabili, almeno in parte, del fenomenodi segregazione morfologica. La ram pressure sembra essere dominante negli ammassidi oggi. Questo meccanismo e sicuramente in grado di influenzare la storia di for-

ix

x

mazione stellare delle galassie, ma ha pochi effetti sulla loro morfologia. (III) Legalassie lenticolari (S0) risultano essere cosı il prodotto di processi completamentedifferenti: se oggi la ram pressure e in grado di trasformare una galassia a spirale inuna lenticolare con piccolo bulge, sono necessarie interazioni gravitazionali per pro-durre i grandi bulge osservati in molte lenticolari nell’Universo locale. (IV) Diverse, eindipendenti, evidenze osservative confermano l’esistenza di una forte correlazione trala massa degli oggetti e l’eta media della loro popolazione stellare (downsizing effect).Nel quadro dei modelli gerarchici di formazione delle strutture, l’origine del downsiz-ing effect e tutt’ora sconosciuta. La comprensione di questo fenomeno rappresentadunque una delle maggiori sfide per l’astronomia extragalattica.

Resume

Ce travaille est dedie a l’etude des effets d’environnement sur l’evolution des galaxiesdans l’Univers voisin, en utilisant un echantillon multi-longueur d’onde. En partic-ulier toute cette analyse est focalisee sur les proprietes des trois differents amas desgalaxies: Abell1367, Virgo et Coma. Ces trois amas des galaxies sont parmi les mieuxetudies dans l’Univers local et, en raison de la variete des leurs proprietes (par ex-emple fraction des galaxies a spirale, luminosite X, etat dynamique), ils represententdes laboratoires, les plus appropries, pour des etudes comparatives. En combinantpour la premiere fois des observations UV de GALEX a des donnees en optique, envoisin et en lointain infrarouge j’ai determine l’histoire evolutive des galaxies dans lesamas. Les buts principales de cette these sont: (a) Etudier la variation des proprietesUV des galaxies en fonction des proprietes de l’environnement ou elles se trouvent,de leur masse et type morphologique, en analysant les fonctions de luminosite en UVet les relations couleur-magnitude. (b) L’etude du taux d’absorption des photons UVpar les poussieres interstellaires, pour obtenir des relations empiriques tres utils pourquantifier l’attenuation par poussieres quand les donnees en infrarouge lointain sontabsentees. (c) Analyser l’effet de la formation des amas sur l’evolution des galaxies enetudiant l’amas d’Abell1367, un des meilleurs exemples d’amas voisin et dynamique-ment jeune. (d) Comprendre l’influence des differents effets d’environnement surl’histoire evolutive des galaxies d’amas, pour comprendre l’origine de la segregationmorphologique dans les amas.Touts les resultats obtenus dans ce travaille montrent que: (I) La population desgalaxies elliptiques geants est vieille et homogene. Elle ne montre pas d’evolutionau moins dans les dernieres 8 Gyr; au contraire des elliptiques naines qui contien-nent toujours populations stellaires jeunes. (II) L’importance relative des differentsmecanismes d’environment varie avec l’age de l’Univers. Les interactions de mareeet le prepocessing ont probablement dominees dans l’Univers passe et ont contribuees(au moins en partie) a la segregation morphologique, pendant la formation des amaspar des petits groupes des galaxies. La pression dynamique domine dans les amasd’aujourd’hui et elle affecte surement l’histoire de formation des etoiles des galaxiesavec moins d’influence sur leur morphologie. (III) La classe heterogene des galaxiesS0s (lenticuliers), n’est pas le resultat d’un seul mecanisme de transformation: si la

xi

xii

pression dynamique peut produire S0s dominees par le disque, les interactions demaree (et le preprocessing) sont exigees pour expliquer les S0s dominees par le bulbe.(IV) Differentes evidences suggerent la presence d’une correlation entre l’age moyendes populations stellaires et la masse des galaxies (downsizing effect). Dans le cadredu modele hierarchique de formation des galaxies, l’origine de cet effet n’est pas en-core resolue. Il represente aujourd’hui une des defis pour les modeles d’evolution desgalaxies.

Contents

1 Introduction 5

2 GALEX & GOLDMINE: A multiwavelength window on the LocalUniverse 152.1 The Galaxy Evolution Explorer . . . . . . . . . . . . . . . . . . . . . 15

2.1.1 The Prime Mission . . . . . . . . . . . . . . . . . . . . . . . . 162.1.2 Data collection mode . . . . . . . . . . . . . . . . . . . . . . . 172.1.3 Counts vs. magnitudes and fluxes conversions . . . . . . . . . 18

2.2 The Galaxy On Line Database Milano Network . . . . . . . . . . . . 20

3 The FAUST-FOCA UV luminosity function of nearby clusters 233.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.2 The Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.3 The UV luminosity functions . . . . . . . . . . . . . . . . . . . . . . 25

3.3.1 The composite cluster luminosity function . . . . . . . . . . . 263.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4 GALEX UV luminosity function of Abell1367 334.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334.2 UV data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334.3 The luminosity function . . . . . . . . . . . . . . . . . . . . . . . . . 374.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

5 Multiple merging in Abell1367 435.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435.2 Observations and data reduction . . . . . . . . . . . . . . . . . . . . . 445.3 The global velocity distribution . . . . . . . . . . . . . . . . . . . . . 465.4 Localized velocity structures . . . . . . . . . . . . . . . . . . . . . . . 485.5 The cluster dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.5.1 The North-West subcluster . . . . . . . . . . . . . . . . . . . . 535.5.2 The South-East subcluster . . . . . . . . . . . . . . . . . . . . 56

1

2 CONTENTS

5.6 Star formation activity in the infalling groups . . . . . . . . . . . . . 59

5.7 Cluster mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.8 Two-Body Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6 Unveiling the evolution of early type galaxies with GALEX. 69

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

6.3 The UV properties of early-type galaxies . . . . . . . . . . . . . . . . 71

6.4 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . 75

7 UV dust attenuation in normal star forming galaxies 81

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

7.2 The Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

7.2.1 The optically-selected sample . . . . . . . . . . . . . . . . . . 84

7.2.2 The starburst sample . . . . . . . . . . . . . . . . . . . . . . . 86

7.3 The LTIR/LFUV − β relation for normal star-forming galaxies . . . . 86

7.3.1 The dependence on the birthrate parameter . . . . . . . . . . 89

7.4 A(Hα) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

7.4.1 Estimate of A(Hα) . . . . . . . . . . . . . . . . . . . . . . . . 90

7.4.2 The β-A(Hα) relation . . . . . . . . . . . . . . . . . . . . . . 92

7.5 Relations between dust attenuation and global properties. . . . . . . 94

7.5.1 Metallicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

7.5.2 Luminosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7.5.3 Surface brightness . . . . . . . . . . . . . . . . . . . . . . . . . 99

7.5.4 LHα/LFUV ratio . . . . . . . . . . . . . . . . . . . . . . . . . . 101

7.6 A cookbook for determining LTIR/LFUV ratio . . . . . . . . . . . . . 103

8 High velocity interaction: NGC4438 in the Virgo cluster 107

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

8.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

8.3 The UV emission and the star formation history of NGC 4438 . . . . 110


9 Ram Pressure stripping: NGC4569 in the Virgo cluster 117

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

9.2 Data and models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

9.3 The star formation history of NGC 4569: model predictions . . . . . 120


CONTENTS 3

10 Galaxy Pre-processing: the blue group infalling in Abell1367 12710.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12710.2 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

10.2.1 HI observations . . . . . . . . . . . . . . . . . . . . . . . . . . 12810.2.2 UV to near-IR imaging . . . . . . . . . . . . . . . . . . . . . . 13110.2.3 Hα imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13110.2.4 MOS spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . 13110.2.5 High Resolution spectroscopy . . . . . . . . . . . . . . . . . . 134

10.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13510.3.1 Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13510.3.2 Hα properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 13810.3.3 HI properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 14310.3.4 The fate of the stripped gas . . . . . . . . . . . . . . . . . . . 14810.3.5 The metal content . . . . . . . . . . . . . . . . . . . . . . . . 14910.3.6 Dating the starburst. . . . . . . . . . . . . . . . . . . . . . . . 15110.3.7 CGCG97-120: simply a foreground galaxy, or an high velocity

intruder? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15610.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

10.4.1 The evolutionary history of the Blue Infalling Group . . . . . 15710.4.2 The contribution of preprocessing to cluster galaxies evolution. 158

11 Discussion & Conclusions 16511.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16511.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

A The extinction correction 179

B Estimate of the < 912A flux from Hα + [NII] 181

Bibliography 183

Chapter 1

Introduction

Eighty-five years are as short as a jiffy compared to the whole history of humanityand science, but this is the brief time men needed to upset their view of the Universethey inhabit. Let us return for a moment at the beginning of this story: July 26,1920, Harlow Shapley and Herber Curtis confront their positions on the size of theUniverse and the nature of the spiral nebulae in talks later called the Great Debate(see Trimble 1995, for a review). Curtis argued that the Universe is composed ofmany galaxies like our own, which had been identified by astronomers of his time asspiral nebulae. Shapley argued that these spiral nebulae were just nearby gas clouds,and that the Universe was composed of only one big Galaxy: our Milky Way.The resolution of the debate came in the mid 1920’s. Using the 100 inch telescopeat Mount Wilson, Edwin Hubble identified Cepheid variable stars in the AndromedaGalaxy (M31). These stars resulted far beyond the most distant stars known in ourgalaxy and allowed Hubble (1925) to show that M31 was a galaxy much like our own.With this discovery, the known universe expanded immensely and, in the same time,a new research area was born: extragalactic astronomy.Thanks to overwhelming technological progress, during its first ∼85 years of life,extragalactic astronomy has provided us with a detailed description of the Universefrom our neighbours (the Local Group) to its observable edges (the Cosmic MicrowaveBackground). We know that most of the visible matter in the Universe, in the form ofstars, gas, and dust grains, is organized in galaxies. Galaxies come in many differentforms and sizes (as clearly shown in Fig.1.1), but they can be broadly divided intotwo main species. Spirals, with a flattened, disk-like shape, blue colors, much gas anddust, and a widespread star formation activity that results in the presence withinthem of many young stars. Ellipticals, with a spheroidal shape, red colors, little or nogas and dust, and no star formation activity, thus containing exclusively old stars. Wealso know that the density of galaxies in the local Universe is not at all constant, butit spans from ∼ 0.2 ρ0 in voids to ∼ 5 ρ0 in superclusters and filaments, ∼ 100 ρ0 inthe cores of rich clusters, up to ∼ 1000 ρ0 in compact groups, where ρ0 is the average

5

6 1. Introduction

Figure 1.1: An example of the heterogeneous population of galaxies that inhabit ourUniverse. Mosaic of RGB (g,r,i) images adapted from Frei et al. (1996)

7

“field” density (Geller & Huchra 1989). It is well established that morphologicaltype and local density are not independent quantities. In their analysis of 55 nearbyclusters, Dressler (1980) and Whitmore et al. (1993) demonstrated that the fractionof spiral galaxies decreases from 60% in the “field” to virtually zero in the coresof rich clusters, compensated by an opposite increase of elliptical and S0 galaxies.This phenomenon, known as morphology segregation, is considered as the clearestobservational signature of significant environmental dependences of the processes thatgovern the formation and the evolution of galaxies. Understanding the origin of thisphenomenon (”Nature or Nurture?”) probably represents one of the major challengesof extragalactic astronomy. One possible way to overcome this problem is to takeadvantage of the effect provided by the finite speed of light. Observing today galaxiesat different distances means observing them at different epochs in the history of theUniverse, and thus with different ages. This investigative method is providing uswith a sort of evolutionary sequence for galaxies: starting from the pioneering workby Butcher & Oemler (1978, 1984) we know that distant (and thus young) cluster ofgalaxies contain a much higher fraction of blue galaxies than nearby clusters. RecentlyDressler et al. (1997) used high-resolution imaging with the Hubble Space Telescope(HST) to measure the morphology-density relation in the core regions of a sample ofrich clusters at z ∼0.5. They found that the fraction of lenticular galaxies (S0s) inclusters declined by a factor of 2-3 between z = 0 and z = 0.5, and this evolutionwas accompanied by a corresponding increase in the fraction of star-forming spirals(see also Couch et al. 1998; Treu et al. 2003). Many research groups have suggestedthat the predominance of early type galaxies in local clusters is the result of physicalprocesses that suppress star formation and eventually alter galaxy morphology. andseveral mechanisms have been proposed (see Boselli & Gavazzi 2006, for a detailedreview):

• Galaxy interaction with the intra-cluster medium (ICM).Ram pressure stripping (Gunn & Gott 1972). As a galaxy orbits through acluster, it experiences ram pressure from the ICM. When the ram pressure isgreater than the binding force, the cold gas will be stripped (Abadi et al. 1999;Quilis et al. 2000; Vollmer et al. 2001). Before leading to complete gas abla-tion, ram-pressure could produces significant compression ahead of the galaxytemporally increasing its star formation activity (Bekki & Couch 2003). Evenif it is well established that this phenomenon would finally lead to a gradualdecrease in galaxy star formation activity, its effects on galaxy morphology arenot yet completely understood (Fujita & Nagashima 1999; Mihos 2004a). Ram-pressure stripping is likely to be effective in the central region of clusters wherethe density of intra-cluster medium (ICM) is high.Viscous stripping (Nulsen 1982). In a galaxy travelling into the ICM the outerlayers of the interstellar medium (ISM) experience a viscosity momentum trans-

8 1. Introduction

fer that could be sufficient for dragging out part of its gas.Thermal evaporation (Cowie & Songaila 1977). If the ICM temperature is highcompared to the galaxy velocity dispersion, at the interface between the hotICM and the cold ISM the temperature of the ISM rises rapidly, thus the gasevaporate and is not retained by the galaxy gravitational field.Starvation (or strangulation) (Larson et al. 1980a). This mechanism consistsin the removal of the diffuse hot gas reservoir that is confined in the galaxyhalo. Since this tenuous halo is less bound than the cold gas in the disk, itsstripping is considerably easier (Bekki et al. 2002). A galaxy whose hot gasreservoir is removed slowly, exhausts its cold gas in more than one gigayear,because there is no supply of fresh gas from the surrounding hot gas. Note thatwhile stripping gas from disks induces a truncation of star formation activityon a short timescale (∼ 107 yr), strangulation is expected to affect a galaxy starformation history on a long time scale (> 1 Gyr) provoking a slowly decliningactivity which consumes the disk gas after the supply of cooling gas has beenremoved. All of the above mechanisms but starvation need relatively high den-sity of hot intra-cluster gas, and thus likely to happen in the central region ofclusters. However Fujita (2004) has pointed out that ram pressure and thermalevaporation could not be negligible in cluster sub-clump regions (small groupsaround a cluster).

• Galaxy-galaxy gravitational interaction. Collisions or close encounters betweengalaxies can have a strong effect on their morphology and star formation rates.Various simulations have shown that major mergers between disk galaxies canproduce galaxies resembling ellipticals as merger remnants (e.g.,Toomre & Toomre1972; Barnes & Hernquist 1996) and that accretion of small satellites onto spi-rals can transform the host spiral to S0 type (Walker et al. 1996). The tidalforces generated during the interaction tend to funnel gas toward the galaxycenter. It is likely that this will fuel a central starburst, ejecting a large frac-tion of material. Gas in the outer part of the disk, on the other hand, will bedrawn out of the galaxy by the encounter (Mihos 2004a). Although individualcollisions are expected to be most effective in groups because the velocity ofthe encounters is too high for such mergers to be frequent (Ghigna et al. 1998;Okamoto & Habe 1999), Moore et al. (1996) showed that the cumulative effectof many weak high velocity interactions (i.e. galaxy harassment) can also beimportant in cluster of galaxies. However its influence is largely limited to lowluminosity galaxies, while in bright spirals its effects are considerably milder(Mihos 2004a; Moore et al. 1996).

• Galaxy-Cluster gravitational interaction. Tidal compression of galactic gas viainteraction with the whole cluster potential can effectively perturb cluster galax-

9

ies, inducing gas inflow, bar formation, nuclear and perhaps disk star formation(Merritt 1984; Miller 1986; Byrd & Valtonen 1990). On the other hand, gas canbe hardly removed directly by the interaction (Boselli & Gavazzi 2006).

Although we have collected a plethora of observational evidences that at least someof these processes are playing a significant role on galaxy evolution we have not shedlight on the origin of the morphology density relation. This is in part due to the factthat we do not yet know their detailed physics and the relative importance of eachmechanism during the different phases of galaxy evolution.

Moreover the arduous effort of reconstructing the evolutionary history of galaxieswould turn out to be completely useless if we did not take into account that the wholeUniverse is evolving, changing the physical condition of the environments populatedby galaxies. In fact different and mostly independent observational evidences, as theCosmic Microwave Background radiation (Kogut et al. 2003), the large scale structure(Hawkins et al. 2003) and supernovae observations (Tonry et al. 2003), are telling usthat the Universe in not only expanding (Hubble & Humason 1931), but it is alsoaccelerating. If theorists are right, this implies that the Universe is dominated by itsenergetic and a matter dark components, whose nature is still completely unknown.The dark energy term (usually indicated with the cosmological constant Λ) allows forthe current accelerating expansion of the universe. Currently, ∼70% of the energydensity of the Universe is supposed to be in this form. The dark matter component ofthe Universe is supposed to be cold (i.e. not thermalized), non-baryonic, collisionless”material”. This component makes up ∼26% of the energy density of the presentUniverse and only the remaining ∼4% is the matter and energy we directly observe.The only way to shade light on the properties of our, mostly obscure, Universe is thusthrough numerical simulations (e.g. Kauffmann et al. 1993; Springel et al. 2005).In particular hierarchical galaxy formation (White & Rees 1978) models within a Λcold dark matter (ΛCDM) cosmogony are currently considered the most successfulparadigm for understanding the evolution of matter in the Universe. In this scenario,structures grow hierarchically via gravitational instability from small perturbationsseeded in the early epoch. The density of dark matter its component is a proxy for theepoch of initial collapse of a given structure: the most massive structures at any epochrepresent the earliest that collapsed (Springel & Hernquist 2003). After their collapse,structures grow up through infall of smaller groups (Kauffmann 1995). However thetypical size of the infalling groups increases with the age of the Universe but theirinfall rate considerably decreases (Ghigna et al. 1998; Okamoto & Habe 1999; Gnedin2003). This means that clusters have accreted great part of their galaxy populationin the past, through infalling of small groups. Today the accretion of new membersis supposed to be rare and to happen mainly through the merging of big subclusters.Adding the well known observational evidence that star formation rapidly decreases

10 1. Introduction

with the age of the Universe (Lilly et al. 1996; Madau et al. 1998), we are facing ascenario that, at a first look, seems to suggest that studying star formation in richclusters today is a melancholy affair. The Universe we inhabit today is old, and mostof its star formation activity has gone out. In addition (and this is the worse part ofthe story) the Universe dramatically evolved itself, altering continuously the physi-cal conditions of the environments populated by galaxies. This implies that galaxiescould have experienced different environmental effects during their history and thatthe dominant process in the local Universe could have been completely negligible inthe early stages of its evolution, while the process shaping galaxy evolution could beless important in today clusters. Let us imagine, as predicted by models, that a greatfraction of today cluster galaxies have infalled, within a compact group, into a cluster∼5 Gyr ago. While in the group environment tidal interactions were very strongand influenced significantly star formation activity and galaxy morphology; today, incluster environment, gravitational interactions are less probable due the large relativevelocities of cluster members (Ostriker 1980). Thus great part of galaxy evolutiontook place before the infalling into the cluster core.The discourage felt by a young student facing this music increases reading the re-cent review by Dressler (2004) on ”Star forming galaxies in clusters”: What we seein clusters today is only a faint echo of what once was... looking for star formationin today’s clusters is a little bit like searching for the last cashew in a picked-overnut-cup.[...] Star formation in rich clusters today is a pretty sad affair. Spirals are”running down compared to half-a-Hubble time ago. The spirals that will be drawninto rich clusters in the future will die the death of a thousand cuts: in the rich groupenvironment into which they have for so long been entrained, they are likely alreadyto have had their fates sealed long ago.Thus, what has he to do? Give up and concentrate all his efforts on the study of thehigh redshift, still young, Universe? Obviously the answer is no; and not because thiswork would be useless.High redshift and local observations are complementary to give more insights ongalaxy evolution and, until we will be able to understand all the physical mechanisminfluencing the present evolution of nearby galaxies (and we are still far from reachingthis goal), it would be an error to concentrate all our efforts only at high redshift.Observations of the high redshift Universe approach us to the mechanisms that maybeshaped part of the morphology density relation; however today there is still insuffi-cient high-quality data to put strong constraint on different models (Dressler 2004).On the contrary in the local Universe, maybe we are missing most of the action, butwe have the unique possibility to observe in detail galaxy properties over the wholerange of sizes and masses, and study in detail the effects of different environmentalmechanisms.In particular, what makes the local Universe still exciting? What can we learn aboutgalaxy evolution that would still be impossible if we moved to higher redshifts? Owing

11

to the high quality images we can obtain for local galaxies, an extremely accurate andhomogeneous morphological classification is possible down to MB ≤ −13, allowing adetailed discrimination among different subclasses of early-type galaxies (ellipticals,lenticulars, dwarfs) and among early-type galaxies and quiescent spirals (see the VirgoCluster catalogue a sort of ”milestone” of the morphological classification, Binggeliet al. 1985). Accurate morphological classification becomes a difficult task just atthe Coma cluster distance (z ∼0.025) and more or less impossible at higher redshift(Abraham et al. 1996a). The objects in the images are very small, thus it is veryhard to detect the fine structure elements needed to distinguish different classes. Inorder to solve this problem alternative classifications based on structural parameters(Abraham et al. 1996b) or on spectral type (Madgwick et al. 2002) have been proposedbut they are only useful to discriminate between a star forming disk and a quiescentbulge dominated galaxy, completely failing to distinguish between an elliptical anda lenticular or between an early-type galaxy and a bulge-dominated Sa spiral disks(Scodeggio et al. 2002; Gavazzi et al. 2002a). Thus at high redshift we can observethe evolution of the star formation-density relation (the Butcher-Oemler effect) butwe cannot investigate morphological transformations that eventually affected galaxyevolution (Smail et al. 1997; Fabricant et al. 2000; Smith et al. 2005).Moreover in the local Universe we can study galaxies spanning all ranges of mass andluminosity, reaching very faint (MB ∼ −13) low surface brightness (∼30 mag arcsec2)dwarf galaxies. This is crucial to study the (strong) dependence of galaxy evolution(Gavazzi et al. 2002a) and environmental effects with mass since the anti-hierarchicalrelation between star formation history and galaxy mass is one of the great challengefor models of galaxy evolution. Moreover dwarf galaxies today represent probablythe major failure of hierarchical galaxy formation models: cold dark matter theorypredicts that the groups and clusters of galaxies should contain many more dwarfobjects than the observed number of dwarf galaxies (Klypin et al. 1999; Moore et al.1999). Several explanation has been proposed (Somerville 2002), and even if no solu-tion has been found so far, it is indisputable that the only way to solve this problemis to understand the formation and evolution of dwarf galaxies, a task possible onlyin the local Universe.Another serious limit of high redshift observations is the quantification of star forma-tion activity in galaxies. The easiest and common way to estimate star formation ratefor distant galaxies is through rest-frame ultraviolet (UV) observations. However ul-traviolet emission is strongly affected by dust attenuation: absorption by dust grainsreddens the spectra at short wavelengths and modifies altogether the spectral energydistribution of galaxies. Since the UV radiation is emitted by young stars (t < 108 yr)that are deeply embedded in dust clouds than older stellar populations, rest-frameUV observations can lead to incomplete and/or biased reconstructions of the starformation activity and star formation history of galaxies. Moreover we have not yeta good characterization of the dust attenuation properties in galaxies and of their

12 1. Introduction

dependences with galaxy type (i.e. normal star forming galaxies vs. starburst) andno proper corrections have been achieved, having no possibility to correctly quantifythe star formation rate at high redshift. As extensively discussed in Chapter 7 of thiswork, understanding dust properties and looking for empirical relations suitable forderiving dust attenuation corrections is today possible only for low redshift galaxies:in this case the study of the local Universe is mandatory to correctly interpret whatwe observe in distant galaxies.Finally, as remarked by Poggianti (2004a), in order to understand what happens togalaxies in clusters, two crucial pieces of information are 1) the gas content of clustergalaxies (i.e. the fuel for future star formation) and 2) the spatial distribution ofthe gas and of the star formation activity within each galaxy (i.e. differences fromfield galaxies are good indicators of environmental effects); and both can be achievedonly in the local Universe. Neutral hydrogen (HI) and Hα observations observations1

are still a prerogative of nearby galaxies. In the near future, thanks to the adventof the Arecibo L-band Feed Array, it should be possible to detect an hydrogen massof ∼ 109M at z ∼0.15, but only with a very high integration time (∼ 70 hoursper beam). The few examples shown above represent only the tip of the iceberg ofthe unique capability of local Universe observations to disclose the secret of galaxyevolution. We would lose too much, without any significant improvement, if we aban-doned observations of nearby galaxies in order to move our attention at high redshiftgalaxies.

The aim of this work Firmly convinced of the great significance of nearby Uni-verse observations, I have concentrated all my PhD work on the study of environmen-tal effects on the evolution of nearby clusters. In particular this thesis will focus onthree different clusters: Abell1367, Virgo and Coma. These three clusters are amongthe best studied in the local universe and, due to the variety of their environmentalconditions (e.g. spiral fraction, X-ray luminosity, evolutionary stage) they representthe most suitable ”laboratory” for comparative studies. The novelty of this workis that in addition to the optical and near infrared observations carried out duringthe last fifteen years by G.Gavazzi and A.Boselli (available through the GOLDMinedatabase, Gavazzi et al. 2003a: http://goldmine.mib.infn.it) I will take for the firsttime advantage of recent UV observations by the Galaxy Evolution Explorer (GALEX,Martin et al. 2005). The use of a multiwavelength dataset is crucial to understandgalaxy evolution since different galactic components such as old, new or evolved stars;active galactic nuclei; the interstellar medium contribute in different amounts to theobserved emission at different wavelengths, from the radio to X-rays. Therefore, thecomparison of global emission properties at a wide range of wavelengths can give

1The Hα Balmer emission (λ=6562.8 A) is the most direct indicator of the current (< 4 106 yrs),massive (> 8 M) star formation activity in galaxies (Kennicutt 1998)

13

us precious insight on the relative importance of these components, as well as onthe origin of some parts of the emission spectrum. Since different emission bandshave different sensitivities to absorption, their comparison may also give us insightinto the dust content of the emitting regions. Moreover, comparison of global multi-wavelength emission properties of galaxies of different morphology can give us insighton the relative presence of different galactic components throughout the Hubble se-quence. While most of the studies of galaxies make use of individual energy bands,mainly the optical but also the radio and, more recently, the X-ray and infrared, it israrer to find work comparing data from two or more emission windows. In particularthe rest frame UV emission provides a powerful tool for measuring and understandingstar formation in galaxies at all epochs. Ironically, the interpretation of high-redshiftgalaxies in the rest UV is most limited by the lack of large, systematic surveys of low-redshift UV galaxies serving as a benchmark. However, before the launch of GALEX,only a few experiments had observed the nearby Universe at ultraviolet wavelengths(Smith & Cornett 1982; Lampton et al. 1990; Kodaira et al. 1990). Among them, theFOCA experiment (Milliard et al. 1991) allowed the first determinations of the UVLF of local field galaxies (Treyer et al. 1998; Sullivan et al. 2000) and and the localrest-UV anchor point for the star formation history plot. However its low sensitivityand resolution and the small sky area covered. With its large field of view (diameter∼1.2 degrees), high sensitivity and two ultraviolet filters, GALEX has opened a newera in the UV astronomy, providing us for the first time with a large, complete andhomogeneous dataset to study star formation activity in galaxies.Using this unique mine of data I will investigate the properties of galaxies from differ-ent points of view: one statistical, analyzing the global properties of the whole clustersample; and another much more focused on the study of particular objects consideredas prototypes of the different ways in which the environment could influence galaxyevolution. The comparison of all the observational results with models will be used tobuild up an evolutionary scenario for galaxies, linking the information I obtain in thiswork to what we know (or think to know) about the evolution of galaxies at higherredshifts.

The organization of the thesis In Chapter 2 I briefly describe the differentdatasets used in this work: the GALEX satellite and its mission, and the GOLDMinedatabase.In Chapter 3 and 4 I start the statistical analysis of cluster galaxies, computing theUV luminosity function for nearby clusters. The analysis presented in Chapter 3 wasperformed before the launch of GALEX, thus I used data from the FOCA (Milliardet al. 1991) experiment of the three nearby clusters studied in this work. WhenGALEX was launched I had the possibility to extend my analysis two magnitudesdeeper with higher quality data. First of all, this double estimate allow me to directly

14 1. Introduction

compare two different and independent datasets. Then the comparison between thecluster luminosity function and the field one is used to determine whether the envi-ronment affects the shape of the cluster luminosity function.In Chapter 5 I study the influence of the dynamical state of a cluster to the evolutionof galaxies, performing a detailed dynamical study of the Abell cluster 1367. Thiscluster is considered as the prototype of a dynamically young local cluster, thus rep-resenting a good place to study the effects of a cluster’s assembly on galaxy evolution.Although X-ray, radio and optical observations suggest that Abell 1367 is dynami-cally young and it is still undergoing the process of formation, detailed spatial anddynamical analysis of this cluster has not been attempted so far. Since the dynamicalstate of a cluster is directly linked with its evolution this work will allow us to havea clear picture of the past, current and future assembly history of this structure andits galaxies.In Chapter 6 I focus my attention on the population of early-type galaxies in clusters,in particular studying the UV properties of giant and dwarf ellipticals and lenticu-lars in the Virgo cluster, in order to determine whether these different morphologicaltypes had the same evolutionary history or not.On the contrary from Chapter 7 till the end of this work I move my attention tothe star forming cluster population. As discussed above if we want to use ultravioletradiation to correctly estimate star formation we need to correct for dust attenuation.Thus in Chapter 7 I present an analysis of dust attenuation properties in nearby clus-ter star forming galaxies, obtaining a cookbook in order to estimate dust attenuationwithout far infrared observations. This analysis represents the tip of the iceberg andonly a future comparison with different dust models will allow us to understand dustattenuation and to know how to correct UV observations of local and high redshiftsgalaxies. Thus, a statistical analysis of star formation activity in cluster galaxies us-ing UV data is still impossible. For this reason in Chapter 8, 9 and 10 I will focalizemy attention on the study of three particular cluster galaxies considered as the proto-types of the three main environmental effects observed in clusters: tidal interaction,ram pressure stripping and preprocessing, respectively. These unique astrophysicallaboratories will help me to understand the effects of different physical mechanismson galaxy evolution in more depth.Finally in Chapter 11 I will summarize the evolutionary scenario for cluster galaxieswhich emerged from this work.

Great part of this thesis is published or submitted for publication on major astro-nomical refereed journals: Gavazzi et al. (2003b, 2006); Cortese et al. (2003a, 2004,2005, 2006); Boselli et al. (2005a,b).

Chapter 2

GALEX & GOLDMINE: Amultiwavelength window on theLocal Universe

2.1 The Galaxy Evolution Explorer

The Galaxy Evolution Explorer (GALEX) is a NASA Small Explorer class mission.It consists of a 50 cm-diameter, modified Ritchey-Chretien telescope with four op-erating modes: Far-UV (FUV) and Near-UV (NUV) imaging, and FUV and NUVspectroscopy. The telescope has a 3-m focal and the field of view is 1.2 circular (seeFig 2.1 and Table 2.1). Spectroscopic observations are obtained at multiple grism-skydispersion angles, so as to mitigate spectral overlap effects. The FUV (1528A: 1344-1786A) and NUV (2271A: 1771-2831A) imagers (see Fig.2.2) can be operated one ata time or simultaneously using a dichroic beam splitter. The FUV detector is pre-ceded by a blue-edge filter that blocks the night-side airglow lines of [OI]1304, 1356,and Lyα. The NUV detector is preceded by a red blocking filter/fold mirror, whichreduces both zodiacal light background and optical contamination. The peak quan-tum efficiency of the detector is 12% (FUV) and 8% (NUV). The detectors are linearup to a local (stellar) count-rate of 100 (FUV), 400 (NUV) cps, which correspondsto mAB ∼ 14 − 15. The resolution of the system is typically 4.5/6.0 (FUV/NUV)arcseconds (FWHM), and varies by ∼20% over the field of view. Further detailabout the mission, in general, and the performance of the satellite, in specific, can befound in Martin et al. (2005) and Morrissey et al. (2005), respectively. The missionis nominally expected to last 38 months; GALEX was launched into a 700 km, 29

inclination, circular orbit on 28 April 2003.

15

162. GALEX & GOLDMINE: A multiwavelength window on the Local

Universe

Figure 2.1: Cross section of the instrument portion of GALEX. The optical path isoutlined in blue. Overall dimensions of the view shown are 1.5 m 1 m (adapted fromMorrissey et al. 2005).

2.1.1 The Prime Mission

GALEX is currently undertaking the first space UV sky-survey, including both imag-ing and grism surveys. The prime mission includes an all-sky imaging survey (AIS:75-95% of the observable sky, subject to bright-star and diffuse Galactic backgroundlight limits) (mAB ' 20.5), a medium imaging survey (MIS) of 1000 deg2 (mAB ' 23),a deep imaging survey (DIS) of 100 square degrees (mAB ' 25), and a nearby galaxysurvey (NGS). Spectroscopic (slit-less) grism surveys (R=100-200) are also being un-dertaken with various depths and sky coverage. Many of the GALEX fields overlapexisting and/or planned ground–based and space-based surveys being undertaken inother bands.All-sky Imaging Survey (AIS): The goal of the AIS is to survey the entire sky subjectto a sensitivity of mAB ' 20.5, comparable to the POSS II (mAB=21 mag) and SDSSspectroscopic (mAB=17.6 mag) limits. Several hundreds to 1,000 objects are in each1 deg2 field. The AIS is performed in roughly ten 100-second pointed exposures pereclipse (∼10 deg2 per eclipse).Medium Imaging Survey (MIS): The MIS covers 1000 deg2, with extensive overlap ofthe Sloan Digital Sky Survey. MIS exposures are a single eclipse, typically 1500 sec-onds, with sensitivity mAB ' 23, net several thousand objects, and are well-matchedto SDSS photometric limits.

2.1. The Galaxy Evolution Explorer 17

Item FUV Band NUV Band

Bandwidth: 1344 – 1786 A 1771 – 2831 AEffective wavelength (λeff): 1528 A 2271 AField of view: 1.28 1.24

Zero point (m0): 18.82 mag 20.08 magImage resolution (FWHM): 4.5 arcsec 6.0 arcsecSpectral resolution (λ/∆λ): 200 90Detector background (typical):

Total: 78 cnt sec−1 193 cnt sec−1

Diffuse: 0.66 cnt sec−1-cm−2 1.82 cnt sec−1-cm−2

Hotspots: 47 cnt sec−1 107 cnt sec−1

Sky background (typical): 2000 c-sec−1 20000 cnt sec−1

Limiting magnitude (5σ):AIS (100 sec): 19.9 mag 20.8 magMIS (1500 sec): 22.6 mag 22.7 magDIS (30000 sec): 24.8 mag 24.4 mag

Table 2.1: Selected Performance Parameters (Morrissey et al. 2005)

Deep Imaging Survey (DIS): The DIS consists of 20 orbit (30 ksec, mAB ' 25) ex-posures, over 80 deg2, located in regions where major multiwavelength efforts arealready underway. DIS regions have low extinction, low zodiacal and diffuse galacticbackgrounds, contiguous pointings of 10 deg2 to obtain large cosmic volumes, andminimal bright stars. An Ultra DIS of 200 ksec, mAB ∼ 26 mag is also in progress infour fields.Nearby Galaxies Survey (NGS): The NGS targets well-resolved nearby galaxies for1-2 eclipses. Surface brightness limits are mAB ∼27.5 mag arcsec−2. The 200 targetsare a diverse selection of galaxy types and environments (see Fig.2.3).Spectroscopic Surveys. The suite of spectroscopic surveys includes: the Wide-fieldSpectroscopic Survey (WSS), which covers the full 80 deg2 DIS footprint with com-parable exposure time (30 ksec), and reaches mAB ∼ 20 mag for S/N∼10 spectra; theMedium Spectroscopic Survey (MSS), which covers the high priority central field ineach DIS survey region (total 8 deg2) to mAB=21.5-23.0 mag, using 300 ksec expo-sures; and the Deep Spectroscopic Survey (DSS) covering 2 deg2 with 1,000 eclipses,to a depth o f mAB=23-24 mag.

2.1.2 Data collection mode

GALEX performs its surveys with plans that employ a simple operational schemerequiring only two observational modes and two instrument configurations. Eachorbit GALEX collects data during night segments (eclipses) and visits to a single


Universe

pre-programmed target. Each target consists either of a single pointing (single visitobservation) or multiple adjacent pointings (sub-visit observations). Currently sub-visits are only used for all-sky imaging survey (AIS) and in-flight calibration ob-servations. After removing instrument overhead, each eclipse typically yields up to1700 seconds of usable science data. During any visit or sub-visit observation thespacecraft attitude is controlled in a tight, spiraled dither. A spiral dither is usedto prevent ”burn-in” of the detector active area by bright objects and to averageover high spatial frequency response variations. For each sub-visit the spiral ditherpattern is restarted. Since celestial sources will move on the detector, the pipelinesoftware will reposition the time-tagged photons to common sky coordinates basedon the satellite aspect solution. As many as 12 sub-visits are allowed per eclipseperiod (typical for AIS), with all-sky survey sub-visits obtaining 100-110 s exposuretime per leg. For plans with sub-visit targets, a 20 second slew time is required tomove between each leg of the observation. For some survey plans (e.g. deep imaging,spectroscopy), a single visit is insufficient to build up the requisite signal-to-noise, soa series of visits are needed in order to obtain the minimum required exposure time.

2.1.3 Counts vs. magnitudes and fluxes conversions

All GALEX data are normalized to their relative exposure time, thus each count (cnt)measured on a GALEX image is in reality a cnt per sec (CPS). Below are given someequations useful to convert galaxies counts into fluxes or magnitudes. To convertfrom GALEX counts per sec (cps) to flux (erg cm−2 s−1 A−1):

FUV : F lux [erg cm−2 s−1 A−1] = 1.40 × 10−15 × CPS (2.1)

NUV : F lux [erg cm−2 s−1 A−1] = 2.06 × 10−16 × CPS (2.2)

To convert from GALEX counts per sec to magnitudes in the AB system (Oke 1974):

FUV : m(AB) = −2.5 × log(CPS) + 18.82 (2.3)

NUV : m(AB) = −2.5 × log(CPS) + 20.08 (2.4)

Thus to convert from flux to AB magnitudes:

FUV : m(AB) = −2.5 × log(F lux [erg cm−2 s−1 A−1]

1.40 × 10−15

)

+ 18.82 (2.5)

NUV : m(AB) = −2.5 × log(F lux [erg cm−2 s−1 A−1]

2.06 × 10−16

)

+ 20.08 (2.6)

The current estimates are that the zero-points defined here are accurate to within+/- 10% (1 sigma).

2.1. The Galaxy Evolution Explorer 19

Figure 2.2: The transmittance profile for the NUV and FUV GALEX filters. Differentgalaxy spectral energy distributions are superposed.

Figure 2.3: Example of GALEX image. GALEX NGS observation of NGC4631. Inthe color table, red-green (gold) is used for NUV, and blue for FUV.


Universe

2.2 The Galaxy On Line Database Milano Net-

work

The Galaxy On Line Database Milano Network (http://goldmine.mib.infn.it) is de-signed to provide access to all the data collected by G.Gavazzi, A.Boselli (Tutor andCo-Tutor of this thesis) and collaborators during several observational campaigns,started in 1985 and still in progress, aimed at providing the phenomenology of localgalaxies in the widest possible frequency range. The creation of the World Wide Website and of the MySQL database has been performed by P. Franzetti and A. Donatiand a detailed description of the database architecture can be found in Donati (2004).GOLDmine is focused on 9 local clusters of galaxies: A262 (Perseus-Pisces), Cancer,A1367, A1656 (Coma), Virgo, A2147, A2151, A2197, A2199 (Hercules). In additionit contains a filament of nearly isolated galaxies, the so called “Great Wall”, thusproviding the ideal laboratory for comparative analyses of galaxies in different envi-ronments, spanning a factor of 20-100 in local galaxy density. Objects are selectedin the above regions with strictly optical completeness criteria. Galaxies brighterthan mp = 15.7 are taken from the Catalogue of Galaxies and of Clusters of Galaxies(CGCG) by Zwicky et al. (1961) in all clusters except Virgo where objects brighterthan mp = 20.0 are taken from the Virgo Cluster Catalogue (VCC) by Binggeli et al.(1985). Obviously, due to the factor of ∼ 5 difference in distance between Virgo andthe other clusters, this selection limit results in dwarf galaxies being included in ourdatabase only for the Virgo cluster. However globally GOLDmine covers the wholerange (4 orders of magnitude) of luminosities spanned by real galaxies. GOLDminecontains 3649 galaxies. Extensive campaigns were carried out to observe as many aspossible of the 3649 target galaxies through all possible observational windows, a taskthat we did not complete yet.The parameters listed in the GOLDmine database are divided into 5 categories: Gen-eral, Continuum and Line photometry, Dynamical and Structural.They can be obtained from GOLDmine by querying the database for an individualgalaxy name or “by parameters”, “by near name or position” or “by available im-ages”. In this case all galaxies in a given range of photographic magnitude, andmorphological type can be selected.General parameters include Catalogue designations, (J2000) celestial coordinates, op-tical diameters, photographic magnitude, redshift, distance, morphological type.Continuum parameters include: UV, U, B, V, J, H, K magnitudes computed at theoptical radius (25th mag arcsec−2) (see Gavazzi et al. 1996); IRAS 60 and 100 mi-cron fluxes; radio continuum fluxes densities at 0.6 and 1.5 GHz. Line photometryincludes: the atomic (HI) and molecular (H2) hydrogen mass; the Hα+[NII] lineequivalent width and flux. Dynamical parameters include: the width of the HI line,with a quality flag; the width of the Hα line and the central velocity dispersion.

2.2. The Galaxy On Line Database Milano Network 21

Structural parameters include: the light concentration index (C31); the effective ra-dius Re; the effective surface brightness µe; the total asymptotic magnitude. Thesequantities (see Scodeggio et al. 2002) are given separately for the H, V and B bands.The novelty of GOLDmine consists of its image section, where images can be down-loaded in JPG and FITS format. Images include:Finding Charts from the Digitized Palomar Sky Survey for all galaxies. Broad bandimages obtained in the B, V, H and K bands. Narrow band images in the light of Hαand a red image of the underlying stellar continuum near Hα. RGB images. For somegalaxies we combined several images to obtain “true” color pictures. Radial profilesof the light distribution as obtained on the available (B, V, H) images (see Gavazziet al. 2000). When at least two radial profiles are available the color radial profile isalso shown. Optical spectra integrated over the whole surface of the galaxy, obtainedin drift-scan mode, i.e. by drifting the spectrograph slit over the galaxy extension (seeGavazzi et al. 2002a, 2004). Spectral Energy Distributions (SEDs) from the UV tothe centimetric radio continuum obtained from broad-band photometry. The plotteddata are total fluxes (extrapolated to the optical radii), unlike the individual aper-ture data given by NED. However they are given as observed, i.e. uncorrected forextinction from our Galaxy and for internal extinction (see Boselli et al. 2003a). It isour goal to provide a homogeneous set of keywords in all FITS header to character-ize the data, including: effective integration time, filter, telescope, WCS parameters,photometric effective zero point. This homogenization is not yet complete. As alsoremarked in Chapter 7, the high quality of data available through GOLDMine, makethis datasample one of the most appropriate for studying the evolution of nearbygalaxies.

Chapter 3

The FAUST-FOCA UV luminosityfunction of nearby clusters

3.1 Introduction

The study of the galaxy luminosity function (hereafter LF) provides us with a fun-damental tool for testing theories of galaxy formation and for reconstructing theirevolution to the present. Recent determinations of the galaxy LF at various frequen-cies, in various environments (i.e.De Propris et al. 2003; Madgwick et al. 2002) andin a number of redshift intervals (i.e.Ilbert et al. 2004) have improved our knowledgeof galaxy evolution and the role played by the environment in regulating the starformation activity of galaxies. The optical cluster LF is significantly steeper thanthat in the field (Trentham et al. 2005). This steepening is due to quiescent galaxies,more frequent at low luminosities in clusters, while the LF of cluster star formingobjects is similar to that in the field (De Propris et al. 2003). The causes of thisdifference might reside in the density-morphology relation (Dressler 1980; Whitmoreet al. 1993) and in particular in the overabundance of dwarf ellipticals in rich clusters(Ferguson & Sandage 1991), whose origin is currently debated in the framework ofthe environmental effects on galaxy evolution.The ultraviolet emission UV( ∼ 2000 A), being dominated by young stars of interme-diate masses (2 < M < 5 M, Boselli et al. 2001) represents an appropriate tool toidentify and quantify star formation activity. Although before the launch of GALEX,the shape of local field UV LF (Sullivan et al. 2000) was supposed to be well deter-mined, there was still a fair amount of uncertainty on the UV luminosity function ofclusters. Its slope was undetermined due to the insufficient knowledge of the back-ground counts (Cortese et al. 2003b). Andreon (1999) proposed a very steep faintend (α ∼ −2.0,−2.2), significantly different from the field LF (α ∼ −1.5). HoweverCortese et al. (2003b) pointed out that this steep slope is likely caused by an un-

23

24 3. The FAUST-FOCA UV luminosity function of nearby clusters

derestimation of the density of background galaxies and proposed a flatter faint-endslope (α ∼ −1.35 ± 0.20). Unfortunately the statistical uncertainty was too high formaking reliable comparisons between the cluster and the field LFs. In this chapterI re-compute the cluster UV luminosity function with two major improvements overprevious determinations. We increase the redshift completeness of the UV selectedsample using new spectroscopic observations of Coma and Abell 1367 (see Chapter5 and Cortese et al. 2004), and compute for the first time the UV LF of the Virgocluster. These improvements are not sufficient to constrain the LF of each individualcluster, however the UV composite luminosity function, constructed for the first timein this paper can be compared with that of the field. Doing so I try anticipating oneof the main goals of the Galaxy Evolution Explorer (GALEX) which, as shown inthe next Chapter, will help us shade light on the UV properties of galaxies and theirenvironmental dependences.We assume a distance modulus µ= 31.15 for the Virgo cluster (Gavazzi et al. 1999a),µ=34.80 for Abell 1367 and µ=34.91 for the Coma cluster (Gavazzi et al. 1999b).

3.2 The Data

The sample analyzed in this chapter comprises the UV sources detected in Virgo,Coma and Abell 1367 clusters by the FOCA (Milliard et al. 1991) and FAUST(Lampton et al. 1990) experiments. The FOCA balloon-borne wide field UV camera(λ = 2000A; ∆λ = 150A) observed ∼ 3 square degrees (∼ 8 Mpc2) in the Abell 1367(unpublished data) and Coma clusters (Donas et al. 1995) and ∼ 12 square degrees(∼ 1 Mpc2) in the Virgo cluster (data are taken from the extragalactic databaseGOLDMine, Gavazzi et al. 2003a). The FOCA observations of Virgo are not suf-ficient to compile a complete catalog: no sources brighter than mUV ∼ 12.2 weredetected due to the small area covered. We thus complement the UV database withthe wide field observations performed by the FAUST space experiment (λ = 1650A;∆λ = 250A) in the Virgo direction (Deharveng et al. 1994), covering ∼ 100 squaredegrees (∼ 8.8 Mpc2). The FAUST completeness limit is mUV ∼ 12.2 (Cohen et al.1994), significantly lower than the FOCA magnitude limit: mUV ∼ 18.5. Howevercombining the two UV catalogs we hope to constrain the shape of the UV luminosityfunction across 7 magnitudes. We use the FAUST observations for mUV < 12.2 andthe FOCA observations for mUV ≥ 12.2. To account for the different response func-tion of FAUST and FOCA filters we transform the UV magnitudes taken by FAUSTat 1650A assuming a constant color index: UV(2000) = UV(1650) + 0.2 mag (Dehar-veng et al. 1994, 2002). We think however that this difference does not bias the galaxypopulations selected by the two experiments. The estimated error on the UV magni-tudes is 0.3 mag in general, but it ranges from 0.2 mag for bright galaxies, to 0.5 magfor faint sources observed in frames with larger than average calibration uncertain-

3.3. The UV luminosity functions 25

ties. The UV emission associated with bright galaxies is generally clumpy, thus it hasbeen obtained by integrating the flux over the galaxy optical extension, determinedat the surface brightness of 25 mag arcsec−2 in the B-band. The spatial resolution ofthe UV observations is 20 arcsec and 4 arcmin for FOCA and FAUST respectively.The astrometric accuracy is therefore insufficient for unambiguously discriminatingbetween stars and galaxies. To overcome this limitation, we cross-correlate the UVcatalogs with the deepest optical catalogs of galaxies available: the Virgo ClusterCatalog (VCC, Binggeli et al. 1985), complete to mB ∼ 18, for the Virgo clusterand the r′ band catalog by Iglesias-Paramo et al. (2003), complete to mr′ ∼ 20, forComa and Abell 1367. We used as matching radius the spatial resolution of eachobservation. In case of multiple identifications we select the galaxy closest to the UVposition. The resultant UV selected sample is composed of 156 galaxies in Virgo, 140galaxies in Coma and 133 galaxies in Abell 1367.

3.3 The UV luminosity functions

Unlike the VCC catalog, the Coma and A1367 r′ catalogs used for star/galaxy dis-crimination do not cover all the area observed by FOCA but only the cluster cores.This reduces our analysis to an area of ∼ 1 square degrees (∼ 2.6 Mpc2) in Comaand ∼ 0.7 square degrees (∼ 1.8 Mpc2) in Abell 1367.Including new spectroscopic observations (Cortese et al. 2004), the redshift complete-ness of the UV selected sample reaches the 65% in Abell 1367, the 79% in Coma andthe 83% in Virgo. The redshift completeness per bin of magnitude of each cluster islisted in Table 3.1. We remark that for MUV ≤ −16.5 (corresponding to the FOCAmagnitude limit in Coma and Abell1367), the redshift completeness of the Virgo clus-ter sample is 98%.As discussed by Cortese et al. (2003b), the general UV galaxy counts (Milliard et al.1992) are uncertain and cannot be used to obtain a reliable subtraction of the back-ground contribution from the cluster counts. Therefore, in order to compute thecluster LF, we use the statistical approach recently proposed by De Propris et al.(2003) and Mobasher et al. (2003). We assume that the UV spectroscopic sample is’representative’, in the sense that the fraction of galaxies that are cluster members isthe same in the (incomplete) spectroscopic sample as in the (complete) photometricsample. For each magnitude bin i we count the number of cluster members NM , thenumber of galaxies with a measured recessional velocity NZ and the total number ofgalaxies NT . The completeness-corrected number of cluster members in each bin is:

Ni =NMNT

NZ(3.1)


Table 3.1: Integral redshift completeness in bin of 0.5 magnitudes.

Redshift completenessMUV ≤ Virgo Coma Abell1367

−21.75 − − 100%−21.25 − 100% 100%−20.75 100% 100% 100%−20.25 100% 100% 100%−19.75 92% 100% 100%−19.25 95% 100% 100%−18.75 97% 100% 100%−18.25 97% 97% 100%−17.75 97% 95% 95%−17.25 98% 84% 80%−16.75 98% 79% 65%

NT is a Poisson variable, and NM is a binomial variable (the number of successesin NZ trials with probability NM/NZ). Therefore the errors associated with Ni aregiven by:

δ2Ni

N2i

=1

NT+

1

NM−

1

NZ(3.2)

The completeness-corrected number of cluster members obtained from (3.1) are givenin Table 3.2 and the luminosity functions for the four studied samples are shown inFig.3.1. The two different datasets used for the Virgo cluster have only one magnitudebin (MUV = −18.75) overlap. In this bin the two LFs are in agreement and thereis no indication that a change of slope occurs. We thus feel comfortable combiningthem into a composite Virgo UV luminosity function across 7 magnitudes.In order to determine whether the LFs of the three clusters are in agreement weperform a two-sample χ2 test. We obtain P (χ2 ≥ χ2

obs) ∼82% for the Virgo andAbell1367 LFs, P (χ2 ≥ χ2

obs) ∼87% for the Virgo and the Coma cluster LFs andP (χ2 ≥ χ2

obs) ∼98% for the Coma and Abell1367 LFs, pointing out that the threeLFs are in fair agreement within their completeness limits.

3.3.1 The composite cluster luminosity function

The uncertainties of each individual cluster luminosity function are too large to fita complete Schechter (Schechter 1976) function to the data and compare it withthe field UV LF. However combining the three data-sets analyzed in this paper we

3.3. The UV luminosity functions 27

Figure 3.1: The UV luminosity functions for the four analyzed data sets.

compute the UV composite luminosity function of 3 nearby clusters. The compositeLF is obtained following Colless (1989), by summing galaxies in absolute magnitudebins and scaling by the area covered in each cluster. The number of galaxies in thejth absolute magnitude bin of the composite LF (Ncj) is given by:

Ncj =1

mj

∑

i

Nij

Ai

(3.3)

where Nij is the completeness-corrected number of galaxies in the jth bin of the ithcluster, Ai is the area surveyed in the ith cluster and mj is the number of clusterscontributing to the jth bin. The errors in Nij are computed according to:

δNcj =1

mj

[

∑

i

(δNij

Ai

)2]1/2

(3.4)

where δNij is the error in the jth bin of the ith cluster determined in (3.2). Theweight associated to each cluster is computed according to the surveyed area, insteadof the number of galaxies brighter than a given magnitude, as used by Colless (1989).

The UV composite luminosity function is given in Fig.3.2 in the full magnituderange. However since for magnitudes fainter than MUV ∼ −16.5 the only availabledata are the Virgo FOCA observations, we fit the composite luminosity function with


Figure 3.2: The composite UV luminosity function of 3 nearby clusters. The solidline represents the best Schechter fit to the data for MUV ≤ −16.5.

the Schechter functional form (Schechter 1976):

φ(MUV) = 0.4 ln 10 φ∗ 100.4(M∗−MUV)(α+1) e−100.4(M∗−MUV)

only for MUV ≤ −16.5, that is the completeness limit in Coma and Abell 1367.The resulting Schechter parameters are M ∗

UV = −20.75 ± 0.40 and α = −1.50 ±0.10. The faint-end slope is consistent within 1 σ with the lower limit for Coma andA1367 recently proposed by Cortese et al. (2003b), but significantly flatter than theslope α ∼ −2.0,−2.2 found for Coma by Andreon (1999), suggesting that this verysteep luminosity function was due to an underestimate of the density of backgroundgalaxies.

3.4 Discussion

Although the UV(2000 A) radiation is dominated by young stars of intermediatemasses (2<M<5M, Boselli et al. 2001), it is frequently detected also in early-typegalaxies with no recent star formation episodes (Deharveng et al. 2002). Unfortu-nately we have no morphological (or spectral) classification for all the UV selectedgalaxies in order to separate the contribution of late and early type galaxies. How-

3.4. Discussion 29

Figure 3.3: The UV bi-variate composite luminosity functions of nearby clusters. Red(UV − B > 2) and blue (UV − B < 2) galaxies are indicated with empty and filledcircles respectively.

ever, based on the spectral energy distributions computed by Gavazzi et al. (2002a),we can use the total color UV − B, available for the 94% of galaxies in our sample,to discriminate between red elliptical (UV − B > 2) and blue spiral (UV − B < 2)galaxies. B magnitudes are taken from the VCC (Binggeli et al. 1985), the Godwinet al. (1983) catalog and the Godwin & Peach (1982) catalog for Virgo, Coma andAbell 1367 respectively.The bi-variate composite luminosity function derived for galaxies of known UV − Bcolor is shown in Fig.3.3. It shows that the star forming galaxies dominate the UVLF for MUV ≤ −18, as Donas et al. (1991) concluded for the first time. Conversely,for MUV ≥ −17.5, the number of red and blue galaxies is approximately the same,pointing out that, at low luminosities, the UV emission must be ascribed not only tostar formation episodes but also to Post-Asymptotic Giant Branch (PAGB) low massstars in early type galaxies (Deharveng et al. 2002). Similarly, if we restrict the anal-ysis to the fraction (∼ 50 %) of objects with known morphological type, we find thatlate-types (Sa or later) dominate at bright UV luminosities, while early-type objectscontribute at the faint UV levels. Since Virgo and Abell1367 are spiral-rich clusterswhile Coma is spiral-poor, one might expect that the LFs of the three clusters ob-tained combining all types should have different shapes, contrary to the observations.The point is that the combined LF of the two types is dominated, at high UV lumi-


Table 3.2: The completeness-corrected differential number of galaxies per bin of mag-nitude

MUV Ni

mag Virgo Virgo Coma Abell 1367(Faust) (Foca)

−21.75 0 0 0 1−21.25 0 0 1 0−20.75 2 0 0 1−20.25 1 0 5 1−19.75 7 0 3 4−19.25 9 0 3 4−18.75 13 2 5 3−18.25 0 2 8.6 6−17.75 0 3 7.7 6.7−17.25 0 3 15.8 10.1−16.75 0 4 18.6 12.7

nosity by the spiral component, while at low luminosity early- and late-type galaxiescontribute similarly. The UV LF of the spiral component are similar in the threeclusters. At faint UV luminosities also the number density of early-type galaxies isapproximately the same in the three clusters. Only at relatively high UV luminositythe number density of early-type galaxies in the Coma cluster exceeds significantlythat of the other two clusters, but it is still much lower than the one of the late-typecomponent. Therefore the LF obtained by combining early- with late-type galaxiesresults approximately the same in the three clusters.The cluster composite luminosity function has identical slope and similar M ∗ as theUV luminosity function computed by Sullivan et al. (2000) for the field: M ∗

UV =−21.21 ± 0.13, α = −1.51 ± 0.10, as shown in Fig. 3.4. This result is quite sur-prising since we have just shown that at low luminosity the contribution of ellipticalsis not negligible, and early-type galaxies are expected to be more frequent in highdensity environments. This result seems in contradiction with recent studies of clus-ter galaxies carried out in Hα (Iglesias-Paramo et al. 2002) and B-bands (De Propriset al. 2003). They find that the LFs of star forming galaxies in clusters and in the fieldhave the same shape, contrary to early type galaxies in clusters that have a brighterand steeper LF than their field counterparts (De Propris et al. 2003). In order tounderstand this apparent difference between optical and UV luminosity functions weneeded to wait the launch of GALEX and higher quality (and more homogeneous)

3.4. Discussion 31

Figure 3.4: The cluster and the field UV luminosity functions. The composite clusterLF is given with filled circles. The solid line indicates the best Schechter fit of thefield LF of Sullivan et al. (2000). The normalization is such that the two LFs matchat MUV ∼ −19.25.

UV observations.

Chapter 4

GALEX UV luminosity function ofAbell1367

4.1 Introduction

As I have shown in the previous Chapter, before the launch of the Galaxy EvolutionExplorer (GALEX), the FOCA experiment allowed the first determinations of the UVLF of local field galaxies (Treyer et al. 1998; Sullivan et al. 2000) and of nearby clusters(Donas et al. 1991; Andreon 1999; Cortese et al. 2003b). Combining the FOCA andFAUST data Cortese et al. (2003a) determined the first composite LF of nearbyclusters. They found no significant differences with the LF in the field. However thisearly determination was affected by large statistical errors due to the uncertainty inthe UV background counts (Cortese et al. 2003b). GALEX has opened a new era ofextragalactic UV astronomy. In particular it provides for the first time precise UVphotometry of galaxies over large stretches of the sky (Xu et al. 2005), thus makingthe background subtraction method more reliable than in the past. Moreover itshigher sensitiveness, higher resolution, large field of view make GALEX observationsa unique homogeneous sample for statistical analysis of galaxies UV properties.

4.2 UV data

GALEX provides far-ultraviolet (FUV; λeff = 1528A, ∆λ = 442A) and near-ultraviolet(NUV; λeff = 2271A, ∆λ = 1060A) images with a circular field of view of ∼ 0.6 de-grees radius. The spatial resolution is ∼5 arcsec. The data analyzed in this Chapterconsist of two GALEX pointings of the Abell cluster 1367, with a mean exposure timeof 1460s, , centered at R.A.(J2000)=11:43:41.34 Dec(J.2000)=+20:11:24.0 (e.g. offsetto the north of the cluster to avoid a star bright enough to threaten the detector, seeFig.4.1). Sources were detected and measured using SExtractor (Bertin & Arnouts

33

34 4. GALEX UV luminosity function of Abell1367

Figure 4.1: The GALEX observation of Abell1367. ROSAT X-ray contour are su-perposed in black. The tick rectangular region indicates the region covered by theoptical catalogues used for the star/galaxy discrimination.

1996). The 100% completeness limit is mAB ∼ 21.5 both in FUV and NUV (Xuet al. 2005). As the NUV images are significantly deeper than the FUV, sourceswere selected and their parameters determined in the NUV. FUV parameters wereextracted in the same apertures. We used a larger SExtractor deblending parame-ter compared to the standard GALEX pipeline, providing reliable MAGAUTO also forvery extended sources. The calibration uncertainty of the NUV and FUV magnitudesis ∼ 10% (Morrissey et al. 2005). Magnitudes are corrected for Galactic extinctionusing the Schlegel et al. (1998) reddening map and the Galactic extinction curve ofCardelli et al. (1989). The applied extinction corrections are of 0.18 and 0.17 mag forthe NUV and FUV bands respectively. To avoid artifacts present at the edge of the

4.2. UV data 35

Figure 4.2: Comparison between FOCA (upper image) and GALEX (lower image)observation of the center of Abell1367. It emerges clearly the strong improvement inresolution and sensitiveness of new GALEX data.


Figure 4.3: Left: The comparison between FOCA and GALEX NUV (left) and FUV(right) magnitudes of galaxies in Abell1367. The continuum line shows the best linearfit to the data.

field, we considered only the central 0.58 deg radius from the field center. A reliablestar/galaxy discrimination was achieved by matching the GALEX catalog againstthe deepest optical catalogs available for A1367 (B < 22.5 mag and r ′ < 21 mag;Iglesias-Paramo et al. 2003), using a search radius of 6 arcsec, as adopted by Wyderet al. (2005) for the estimate of the GALEX local field LF. The optical catalogs do notinclude a negligible part (∼ 0.09 square degrees) of the GALEX field. A total numberof 292 galaxies in the FUV and of 480 galaxies in NUV with mAB ≤ 21.5 are detectedin the ∼ 0.96 square degrees field (∼ 2.5Mpc2) analyzed in this Chapter. Great partof the field observed by GALEX covers the area studied in the previous Chapterwith FOCA observations. The two observations of the cluster center are presentedin Fig.4.2: emerges clearly the strong improvement in resolution and sensitivenessof new GALEX data. In Fig.4.3 (left) we compared the UV magnitudes determinedfrom FOCA and from GALEX NUV observations for the 96 galaxies detected by bothinstruments. The two sets of measurements are in satisfactory agreement. The linearregression between the two datasets is:

MGALEX(2310A) = (1.02 ± 0.03) × MFOCA(2000A) + (1.74 ± 0.51) (4.1)

MGALEX(1530A) = (1.04 ± 0.04) × MFOCA(2000A) + (1.71 ± 0.70) (4.2)

with a mean dispersion of 0.23 and 0.32 mag in NUV and FUV bands respectively,consistent with the error assumed in the previous Chapter for FOCA observations.

4.3. The luminosity function 37

Figure 4.4: The redshift completeness per bin of UV magnitude in Abell 1367.

Band Sample Schechter ParametersM∗ α

NUV A1367 −19.77 ± 0.42 −1.64 ± 0.21NUV Field −18.23 ± 0.11 −1.16 ± 0.07FUV A1367 −19.86 ± 0.50 −1.56 ± 0.19FUV Field −18.04 ± 0.11 −1.22 ± 0.07

UV(2000A) Composite cluster −18.79 ± 0.40 −1.50 ± 0.10

Table 4.1: Best Fitting Parameters.

4.3 The luminosity function

The determination of the cluster LF requires a reliable estimate of the contribu-tion from background/foreground objects to the UV counts. This can be accuratelyachieved for mAB ≤ 18.5, since at this limit our redshift completeness is ∼ 90 %(Cortese et al. 2003b, 2004; see Fig. 4.4). The redshift completeness drops rapidlyat magnitudes fainter than mAB ∼ 18.5, thus requiring the contamination to be esti-mated statistically. Two methods are usually applied for the computation of clusterLFs. The first one is based on the statistical subtraction of field galaxies, per binof UV magnitude, that are expected to be randomly projected onto the cluster area,as derived by Xu et al. (2005). Alternatively, the completeness corrected methodproposed by De Propris et al. (2003) is to be preferred when the field counts havelarge uncertainties. It is based on the assumption that the UV spectroscopic sample


Figure 4.5: The GALEX NUV (left) and FUV (right) LF for Abell 1367. Open dotsare obtained using the subtraction of field counts obtained by Xu et al. (2005); filleddots are obtained using the completeness corrected method. The solid line representsthe best Schechter fit. The dotted line shows the composite nearby clusters 2000 ALF by (Cortese et al. 2003a). The dashed line represents the GALEX local field LF(Wyder et al. 2005), normalized in order to match the cluster LF at MAB ∼ −17.80.

Figure 4.6: The NUV (left) and FUV (right) bi-variate LF of A1367. Star-forming andquiescent galaxies are indicated with empty triangles and filled squares respectively.The dashed line represents the GALEX local field LF (Wyder et al. 2005), normalizedas in Fig.4.5

4.3. The luminosity function 39

(e.g. membership confirmed spectroscopically) is ’representative’ of the entire cluster,i.e. the fraction of galaxies that are cluster members is the same in the (incomplete)spectroscopic sample as in the (complete) photometric one. For each magnitude bini we count the number of cluster members NM (i.e. galaxies with velocity in therange 4000<V<10000 km s−1; Cortese et al. 2004), the number of galaxies with ameasured recessional velocity NZ and the total number of galaxies NT . The ratioNZ/NT , corresponding to the redshift completeness in each magnitude bin is shownin Fig.4.4. The completeness-corrected number of cluster members in each bin isNi = (NM × NT )/NZ. NT is a Poisson variable, and NM is a binomial variable (thenumber of successes in NZ trials with probability NM/NZ). Therefore the errors asso-ciated with Ni are given by (δ2Ni/N

2i ) = (1/NT ) + (1/NM)− (1/NZ). The NUV and

FUV LFs using both methods (see Fig 4.5) are in good agreement for MAB ≥ −14.3.In the last bin the two methods are inconsistent as the completeness corrected methodpredicts a higher slope than the statistical background subtraction. This disagree-ment is likely due to the severe redshift incompleteness for MAB ≥ −14.3. In anycase we take the weighted mean of the two determinations.Due to the small number of galaxies populating the high luminosity bins (i.e. onlythree objects brighter than MAB ∼ −18.3), the LFs are not well fitted with a Schechterfunction (Schechter 1976): the best-fit M∗ turns out to be brighter than the brightestobserved galaxy.For this reason we first determine the faint-end (−18.3 ≤ MAB ≤−13.3) slope in each band, fitting the LFs with a power law (Φ(M) = c 10kM) byminimizing χ2. The α parameter of the Schechter function can be derived from kusing the relation α = −(k/0.4 + 1). Then we fit the LFs with a Schechter function,keeping α fixed to the value previously obtained. This is not the canonical Schechterfit, but it provides a more realistic set of parameters than using a three-free-parameterfit. The best fit parameters and their errors are listed in Table.4.3.In order to separate the contribution to the LF of star-forming from quiescent galax-ies, we divide the sample into two classes. Using Hα imaging data (Iglesias-Paramoet al. 2002; Gavazzi et al. 1998, 2002b, 2003a) and optical spectroscopy (Cortese et al.2003b, 2004) we can discriminate between star-forming (EW (Hα) > 0 A) and quies-cent (EW (Hα) = 0 A) objects. Unfortunately neither UV field counts for differentmorphological types nor a measure of EW (Hα) for all the UV selected galaxies areavailable. Thus we can only apply the completeness corrected method to determinethe bi-variate LFs. We assume that in each bin of magnitude the fraction of star-forming and quiescent cluster members is the same in the (incomplete) spectroscopicsample as in the (complete) photometric sample. The bi-variate LFs derived by thismethod are shown in Fig.4.6.


4.4 Discussion

As shown in Fig.4.5, the GALEX LFs have a shape consistent with the compositeLF of nearby clusters as constructed in the previous Chapter (see also Cortese et al.(2003a)). Conversely, whatever fitting procedure one adopts, they show a steeperfaint-end slope and a brighter M ∗ than the GALEX field LF recently determinedby Wyder et al. (2005). In fact the GALEX local field luminosity function shows afaintest bright-end and a flatter faint end than the previous determination by Sullivanet al. (2000), but the reason for this difference is not yet clearly understood. Wyderet al. (2005) argued that magnitudes estimated by FOCA are on average brighterthan the GALEX one, with the difference becoming larger for fainter sources; sug-gesting that these offsets and nonlinearities in the FOCA photometry could accountfor a major part of the observed difference between the two field luminosity func-tions. However we have shown that this seems not the case at least for Abell1367observations. On the contrary I think that part of the problem could be due not todifferent photometric estimates but to the different areas used by GALEX and FOCAto estimate the field LF. In the case of FOCA, Treyer et al. (1998) and Sullivan et al.(2000) used the pointing of Abell 1367 to estimate the field LF: thus a partial contam-ination of cluster galaxies could explain why the FOCA field and cluster LF resultsvery similar.The brighter M∗ observed in Abell1367 is probably to be ascribed to the particulargalaxy population of this cluster. In fact Abell 1367 is a young cluster of galaxiescomposed of at least four dynamical units at the early stage of a multiple mergingevent (see Chapter 5 and Cortese et al. 2004). Some galaxies have their star formationenhanced due to interaction with the cluster environment, and it is this populationthat is responsible for the bright M ∗ observed in this cluster.Conversely the high faint-end slope observed in this cluster is due to the significantcontribution of non star-forming systems at faint UV magnitudes. In fact, as shownin Fig.4.6, star-forming galaxies dominate the UV LF for MAB ≤ −17 mag, as Donaset al. (1991) concluded for the first time. For MAB ≥ −16 mag however, the numberof red galaxies increases very rapidly1. This result is consistent with an UV LF con-structed starting from the r′ LF computed by Iglesias-Paramo et al. (2003): assuminga mean color NUV − r′ ∼ 1 mag and NUV − r′ ∼ 5 for star-forming and quiescentgalaxies respectively, we are able to reproduce the contribution, at low UV luminosi-ties, of elliptical galaxies. Moreover the difference observed between NUV and FUVcluster LFs can be understood looking at the FUV-NUV color magnitude relation(computed only for confirmed cluster members) shown in Fig.4.7. The star-formingobjects dominate at high UV luminosities while the quiescent systems contribute more

1The bi-variate LFs cannot be compared with the ones computed by Treyer et al. (2005) for thefield, since their samples do not contain ellipticals but only spiral galaxies.

4.4. Discussion 41

Figure 4.7: The FUV-NUV color magnitude relation for confirmed members of A1367.Symbols are as in Fig.4.6

Figure 4.8: The optical (r′-band) distribution for star forming (blue histogram) andquiescent (red histogram) galaxies in our sample.


at faint magnitudes. Their mean FUV-NUV color is ∼ 1.5 mag thus they influencethe LF at higher luminosities in the NUV than in the FUV (see Fig.4.6). The opticalluminosity distribution of star forming and quiescent systems, presented in Fig. 4.8,points out clearly that early type galaxies contributing to the UV faint end slopeare the giant, optically bright, galaxies that dominate the bright end of the opticalluminosity functions. This means that, in UV, the steeper faint end slope observed inclusters is only due to the contribution of giant ellipticals and not of dwarf ellipticalgalaxies, as observed at optical wavelengths. We can thus conclude that, in clusters,a significant fraction of the low luminosity UV emission comes massive early typegalaxies. This result is expected since in the field the fraction of quiescent systems issignificantly lower than that of star forming objects (Dressler 1980; Whitmore et al.1993), thus their contribution to the LF is negligible. Moreover, the UV emission ofellipticals has a different nature from the one emitted by star forming systems. Infact it does not arise from newly born stars but from low mass old post asymptoticgiant branch stars (O’Connell 1999), as I will discuss in depth in Chapter 6.Finally, the LFs of cluster star-forming systems have a faint-end slope (α ∼ −1.25 ±0.2) consistent within the statistical uncertainties with the GALEX field LF. Thesimilar shape observed in the LF of star forming galaxies in different environmentsgoes in the same direction with recent studies of cluster galaxies carried out in Hα(Iglesias-Paramo et al. 2002) and B-bands (De Propris et al. 2003). They find thatthe LFs of star forming galaxies in clusters and in the field have the same shape,contrary to early type galaxies in clusters that have a brighter and steeper LF thantheir field counterparts (De Propris et al. 2003). This indicates that, whatever mech-anism (i.e. ram pressure, tidal interaction, galaxy harassment) quenches/enhancesthe star formation activity in late-type cluster galaxies, it influences similarly andwith a short time scale the giant and the dwarf components , so that the shape oftheir LF is unchanged and only the normalization is modified.

Chapter 5

Multiple merging in Abell1367

5.1 Introduction

Clusters of galaxies represent the most massive gravitationally bound systems in theUniverse. They provide us with valuable insights into the formation of large-scalestructures, as well as into the formation and evolution of galaxies. The hierarchi-cal model predicts that galaxy clusters are formed by accretion of units of smallermass at the nodes of large-scale filaments (West et al. 1991; Katz & White 1993).Statistical analyses of clusters have shown that even at low redshift a high fractionof clusters presents substructures, implying that clusters are still dynamically youngunits, undergoing the process of formation (Dressler & Shectman 1988).The Abell cluster 1367 (z ∼ 0.0216) lies at the intersection of two filaments, the firstextending roughly 100 Mpc from Abell 1367 toward Virgo (West & Blakeslee 2000),the second connecting Abell 1367 to Coma (as a part of the Great Wall, Zabludoffet al. 1993). With its irregular X-ray distribution (Jones et al. 1979; Bechtold et al.1983; Grebenev et al. 1995), high fraction of spiral galaxies and low central galaxydensity, Abell 1367 can be considered as the prototype of a nearby dynamically youngcluster.ASCA X-ray observations pointed out the existence of a strong localized shock inthe intra-cluster medium (ICM) suggesting that Abell 1367 is experiencing a mergingbetween two substructures (Donnelly et al. 1998). Moreover recent Chandra obser-vations (Sun & Murray 2002), and a preliminary analysis of the XMM data (Formanet al. 2003), indicate the presence of cool gas streaming into the cluster core, sup-porting a multiple merger scenario.Optical and radio observations also suggest that this cluster is currently experiencinggalaxy infall into its center. Gavazzi et al. (1995, 2001a) discovered two head-tailradio sources associated with disk galaxies with an excess of giant HII regions ontheir leading edges, in the direction of the cluster center. The observational scenario

43

44 5. Multiple merging in Abell1367

is consistent with the idea that ram-pressure (Gunn & Gott 1972) is, for a limitedamount of time, enhancing the star formation of galaxies that are entering the clustermedium. In addition Gavazzi et al. (2003b) pointed out the existence of a group ofstar bursting galaxies infalling into the cluster core.Although X-ray, radio and optical observations suggest that Abell 1367 is dynami-cally young and it is still undergoing the process of formation, detailed spatial anddynamical analysis of this cluster has not been attempted so far. Girardi et al. (1998)detected a secondary peak in the cluster velocity distribution, suggesting that Abell1367 is a binary cluster, but their analysis was based on ∼ 90 redshifts, insufficientfor drawing a detailed model of the cluster kinematics.Cortese et al. (2003b) carried out a deep (r′ < 20.5) spectroscopic survey of the cen-tral ∼ 1.3 square degrees of Abell 1367 adding 60 new spectra (33 members). Here Ipresent new measurements for 119 galaxies (adding another 33 cluster members). Intotal 273 redshifts were measured in the region, out of which 146 are cluster members,allowing the first detailed dynamical analysis of Abell 1367.

5.2 Observations and data reduction

The cluster region analyzed in this Chapter covers an area of ∼ 1.3 square degreescentered at α(J.2000) = 11h44m00s δ(J.2000) = 19d43m30s. r ′ imaging material wasused to extract a catalogue of galaxy candidates in Abell 1367 complete to r ′ ∼ 20.5mag, and to select the targets of the present spectroscopic survey. Spectroscopy ofAbell 1367 was obtained with the AF2-WYFFOS multi fiber spectrograph at the4.2m William Herschel Telescope (WHT) on La Palma (Spain) during 2003, March27-29. WYFFOS has 150 science fibers of 1.6 arcsec diameter coupled to a bench-mounted spectrograph which relies on a TEK 1024 × 1024 CCD. The 316R gratingwas used, giving a dispersion of ∼240 A/mm, a resolution of ∼ 6A FWHM, and atotal spectral coverage of ∼5600 A. The spectra were centered at ∼ 6500A, thuscovering from 3600 A to 9400 A. We allocated typically ∼ 70 objects to fibers in agiven configuration and, on average, 15 sky fibers. A total of 4 configurations wereexecuted, with an exposure time of 4x1800 sec for each configuration. Argon lampsfor wavelength calibration were obtained for each exposure.The reduction of the multi fiber spectra was performed in the IRAF1 environment,using the IMRED package. After bias subtraction, the apertures were defined ondome flat-field frames and used to trace the spectra on the CCD. The arc spectrawere extracted and matched with arc lines to determine the dispersion solution. The

1IRAF is distributed by the National Optical Astronomy Observatories, which is operated by theAssociation of Universities for Research in Astronomy, Inc., under the cooperative agreement withthe National Science Foundation

5.2. Observations and data reduction 45

Observatory Dates N. gal. Spectrograph Dispersion Coverage CCD pixA/mm A µm

WHT March 03 98 AF2-WYFFOS 240 3600-9400 1024 × 1024 TEK 24Cananea March 03 12 LFOSC 228 4000-7100 576 × 384 TH 23Loiano March 03 - Feb. 04 9 BFOSC 198 3600-8900 1340 × 1300 EEV 20

Table 5.1: The spectrograph characteristics

rms uncertainty of the wavelength calibration ranged between 0.1 and 0.3 A. Thelamps’ wavelength calibration was checked against known sky lines. These were foundwithin ∼ 0.5 A of their nominal position, providing an estimate of the systematic un-certainty on the derived velocity of ∼ 25 km s−1. The object spectra were extracted,wavelength calibrated and normalized to their intensity in the interval 5400-5600 A.A master sky spectrum, that was constructed by combining various sky spectra wasnormalized to each individual science spectrum and then subtracted from it. Unfor-tunately strong sky residuals were left after this procedure, limiting the number ofuseful spectra to 98 (as listed in Tab. 5.9).Nine additional long-slit, low dispersion spectra were obtained in March 2003 and inFebruary 2004 using the imaging spectrograph BFOSC attached to the Cassini 1.5mtelescope at Loiano (Italy). Another twelve spectra were taken with LFOSC at the2.1m telescope of the Guillermo Haro Observatory at Cananea (Mexico). These ob-servations were performed using a 2.0 arcsec slit and the wavelength calibration wassecured with exposures of HeAr and XeNe lamps at Loiano and Cananea respectively.The on-target exposure time ranged between 15 and 30 min according to the bright-ness of the targets. After bias subtraction, when 3 or more frames of the same targetwere obtained, these were combined (after spatial alignment) using a median filterto help cosmic rays removal. Otherwise the cosmic rays were removed using the taskCOSMICRAYS and/or under visual inspection. The lamps wavelength calibrationwas checked against known sky lines. These were found within ∼ 1 A from theirnominal position, providing an estimate of the systematic uncertainty on the derivedvelocity of ∼ 50 km s−1. After subtraction of sky background, one-dimensional spec-tra were extracted from the frames.The redshift were obtained using the IRAF FXCOR Fourier cross-correlation (Tonry& Davis 1979) task, excluding the regions of the spectra affected by night-sky lines.Moreover all the spectra and their best correlation function were visually examinedto check the redshift determination.Table 5.2 lists the characteristics of the instrumentation in the adopted set-up.The 119 new velocity measurements presented in this Chapter are listed in Table 5.9as follows:Column 1: Galaxy designation.


Figure 5.1: Cumulative redshift distribution for galaxies in the studied region.

Column 2, 3: (J2000) celestial coordinates, measured with few arcsec uncertainty.Column 4: r′ band magnitude.Column 5: observed recessional velocity.Column 6: telescope (WHT=William Herschel Telescope; LOI=Loiano; CAN=Cananea)Combining the new set of 119 redshifts (given in Tab. 5.9) with the ones available fromthe literature (NED; Cortese et al. 2003b; Rines et al. 2003), we have the redshift for273 galaxies of which 146 are cluster members (4000 km s−1 ≤ V ≤ 10000 km s−1).The cumulative redshift distribution, in the observed area, as a function of the r′

magnitude is shown in Fig.5.1. The completeness is ∼ 70% at r′ < 17.5, and it dropsto ∼ 45% at r′ < 18.5.

5.3 The global velocity distribution

The line of sight (LOS) velocity distribution for the 146 cluster members is shown inFig. 5.2. The mean and standard deviation are known to be efficient estimators ofthe central location and scale when the underlying population is gaussian. Unfortu-nately they are not minimum variance estimators when the nature of the observedpopulation is significantly non-Gaussian. The best location and scale estimators mustbe resistant to the presence of outliers and robust to a broad range of non-Gaussianunderlying populations. Thus, following Beers et al. (1990), we consider the biweightestimator as the best estimator of location (CBI) and scale (SBI) of the cluster ve-

5.3. The global velocity distribution 47

Figure 5.2: Velocity histogram and stripe density plot for the members of Abell1367. Arrows mark the location of the most significant weighted gaps in the velocitydistribution.

locity distribution.We find a location CBI = 6484 ± 81 km s−1 and a scale SBI = 891 ± 58 km s−1,in agreement with previous studies (e.g. Girardi et al. 1998; Struble & Rood 1999).Visual inspection of Fig. 5.2 suggests that the velocity distribution differs from aGaussian, a deviation that should be quantified using appropriate statistical tests.We analyze the higher moments of the distributions using the kurtosis and the skew-ness shape estimators. Kurtosis indicates a difference in the tails length comparedto a Gaussian (positive kurtosis is indicative of long tails). Skewness indicates asym-metry (positive skewness implies that the distribution is depleted from values lowerthan the mean location, conversely negative skewness denotes a depletion of valueshigher than the mean).In addition we calculate the asymmetry index (AI) and tail index (TI) introducedby Bird & Beers (1993) as alternatives to the distribution higher moments. Theseindicators measure the shape of a distribution but, contrary to skewness and kurtosis,which depend on the estimate of the location and the scale of the underlying distri-bution, they are based on the order statistics of the dataset. The AI measures thesymmetry in a population by comparing gaps in the data on the left and right sidesof the sample median. The TI compares the spread of the dataset at 90% level to thespread at the 75% level.The kurtosis, skewness and the TI reject a Gaussian distribution with a confidence


Test Value Rejection of a gaussian

AI -0.077 ≤ 80 %TI 1.240 >99 %

Skewness 0.269 >99 %Kurtosis 2.680 >99 %

W 0.963 98.7 %

Table 5.2: 1D substructure indicators for the whole cluster sample

level of ≥99%, suggesting that the cluster velocity distribution has longer tails thana Gaussian of the same dispersion. Moreover, in order to assess the normality of thevelocity distribution, we use the Wilk - Shapiro (W) test (Yahil & Vidal 1977). Con-trary to the χ2 and Kolmogorov Smirnov, this test does not require any hypothesison the mean and variance of the normal distribution. The W test rejects normalitywith a confidence level of 98.7%, in agreement with kurtosis, skewness and TI (seeTable 5.3).The departure from a normal distribution could result from a mixture of several ve-locity distributions with different location and smaller velocity dispersion than thewhole sample; thus, using the program ROSTAT (Beers et al. 1990), we investigatethe presence of significant gaps (Beers et al. 1991) in the velocity distribution, indi-cating subclustering. A weighted gap is defined by:

yi =(

i(N − i) ∗ (xi+1 − xi))1/2

where N is the number of values in the dataset. A weighted gap is significant if itsvalue, relative to the midmean (the mean of the central 50% of the dataset) of theother weighted gaps, is greater than 2.25. This value corresponds to a probabilityof occurrence in a normal distribution of less than 3%. We detected six significantweighted gaps in the Abell 1367 velocity distribution. The stripe density plot of radialvelocities and the position of each gap (indicated with an arrow) are shown in Fig.5.2. The velocity of the object preceding each gap, the normalized size of the gap andthe probability of finding a normalized gap of the same size and position in a normaldistribution are listed in Table 5.3.

5.4 Localized velocity structures

Given the non-Gaussian nature of the velocity distribution, we looked for spatiallylocalized variations in the LOS velocity and velocity dispersion distributions. First ofall we applied the three 3D tests commonly used to quantify the amount of substruc-

5.4. Localized velocity structures 49

Velocity Gap Significancekm s−1

5742 2.53 1.40%5835 2.66 1.40%6619 2.90 0.60%6880 2.64 1.40%7059 3.01 0.20%7542 2.33 3.00%

Table 5.3: The most significant weighted gaps detected in the velocity distribution ofthe whole cluster sample.

tures in galaxy clusters: the ∆ test (Dressler & Shectman 1988), the α test (West &Bothun 1990) and the ε test (Bird 1994).The ∆ test is based on the comparison of the local mean velocity, Vlocal, and thevelocity dispersion, σlocal, associated to each cluster member (computed using its 10nearest neighbors) with the mean velocity V , and dispersion σ, of the whole galaxysample. For each galaxy, the deviation is defined by:

δ2 =11

σ2[(Vlocal − V )2 + (σlocal − σ)2]

The observed cumulative deviation ∆, defined as the sum of the δ’s for the clustermembers, is used to quantify the presence of substructures. As shown by Pinkneyet al. (1996) for samples with no substructures, the value of ∆ is approximately equalto the total number of galaxies, while it is larger in the presence of substructures.The α test measures how much the centroid of the galaxy distribution shifts as a resultof correlations between the local kinematics and the projected galaxy distribution.The centroid of the whole galaxy distribution is defined as:

xc =1

N

N∑

i=1

xi yc =1

N

N∑

i=1

yi

For each galaxy i and its 10 nearest neighbors in the velocity space, the spatialcentroid is defined as:

xic =

∑11j=1 xj/σj

∑11j=1 1/σj

yic =

∑11j=1 yj/σj

∑11j=1 1/σj


Indicator Value Prob. of substructures

∆ 206.5 99.8 %α 0.161 Mpc 55.7 %ε 5.44 1013 M 68.4 %

Table 5.4: 3D substructure indicators for our sample

where σj is the velocity dispersion for galaxy j and its 10 nearest neighbors in projec-tion. Finally the presence of substructures in the cluster sample is quantified usingthe α statistic defined as:

α =1

N

N∑

i=1

[(xic − xc)

2 + (yic − yc)

2]1/2

which represents the mean centroid shift for the galaxy cluster. The higher the valueof α, the higher the probability of substructures.The ε test quantifies the correlations between the position and the projected massestimator (Heisler et al. 1985), defined as:

MPME =( 32

πGN

)

N∑

j=1

v2zjrj

where vzj is the radial peculiar velocity with respect to the nearest neighbors group(composed by a galaxy and its 10 nearest neighbors) and rj is the projected distancefrom the center of the nearest neighbor group. The substructure statistic is thendefined as:

ε =1

Ngal

N∑

i=1

MPME

which represents the average mass of the nearest neighbors groups in the cluster.Since galaxies in the nearest neighbors groups have small projected separations, ε isgenerally smaller than the global mass estimate. ε is lower for a cluster with sub-structures than for a relaxed system.The value and the significance of the above tests are listed in Table 5.4. These statis-tical tests are calibrated using 1000 Monte Carlo simulations that randomly shufflethe velocity of galaxies, keeping fixed their observed position, thereby destroying anyexisting correlation between velocity and position. The probability of subclusteringis then given as the fraction of simulated clusters for which the test value is lower(larger for the ε test) than the observed one. Assuming that these tests reject the null

5.5. The cluster dynamics 51

Figure 5.3: Local deviations from the global kinematics for galaxies in Abell 1367 asmeasured by the Dressler & Shectman (1988) test. Galaxies are marked with opencircles whose radius scales with their local deviation δ from the global kinematics.The ROSAT X-ray contours are shown with dotted lines.

hypothesis if the confidence level is greater than 90%, only the ∆ test finds evidenceof substructures (see Table 5.4). The local deviations from the global kinematics asmeasured by the ∆ test are shown in Fig 5.3. The positions of galaxies are markedwith open circles whose radius scales with their local deviation δ from the globalkinematics. The presence of a substructure with a high deviation from the globalcluster kinematic is evident projected near the cluster core.More insights on the cluster dynamical state can be achieved by comparing the resultsof the one and three dimensional statistical tests with the N-body simulations per-formed by Pinkney et al. (1996). These authors analyzed how the significance levelof statistical tests of substructure varies in different cluster merging scenarios. Thedeviation of the velocity distribution from a Gaussian and the detection of substruc-ture provided by the ∆ test suggest that Abell 1367 is in the early merging stage,∼ 0.2 Gyr before core crossing.

5.5 The cluster dynamics

The analysis of the galaxy distribution, of the local mean LOS velocity and of thevelocity dispersion give further insight onto the cluster structure. The iso-densitymap of the cluster members (computed using the 10 nearest neighbors to each point)


Figure 5.4: Palomar DSS image of the central region (∼1.3 square degrees) of Abell1367 studied in this Chapter. The iso-density contours for the 146 confirmed clustermembers are superposed. The lowest iso-density contour correspond to 3σ abovethe mean density in the field (left). The ROSAT X-ray contours are superposed inred (right). The straight line indicates the position of the abrupt gas temperaturegradient detected by ASCA (Donnelly et al. 1998), used to divide our sample intotwo subclusters: the North-West and the South-East.

is shown in Fig.5.4 (left). The galaxy distribution appears elongated from north-westto south-east with two major density peaks. The highest density region correspondsapproximately to the center of the NW X-ray substructure detected by ROSAT (Don-nelly et al. 1998), while the secondary density peak is slightly offset from the X-raycluster center (α(J.2000) = 11h44.8m δ(J.2000) = 19d42m, Donnelly et al. 1998).Moreover the south galaxy density peak roughly coincides with the substructure de-tected by the ∆ test (see Fig.5.3) and with the infalling group of star-forming galaxiesstudied by Gavazzi et al. (2003b).The iso-density contours superposed on the ROSAT X-ray contours are shown in

Fig.5.4 (right). The region between the two major density peaks coincides with thestrong gradient in the gas temperature (see the straight line in Fig.5.4, right) observedfor the first time by ASCA (Donnelly et al. 1998) and recently confirmed by Chandra(Sun & Murray 2002). This abrupt temperature change is strongly suggestive of ashock which has generated during a collision between two substructures, probably as-sociated with the SE and the NW galaxy density peaks. In fact N-body simulationsshow that temperature structures and X-ray morphology similar to the one observedin Abell 1367 are typical of clusters at an early merging phase (∼ 0.25 Gyr beforecore crossing) (Schindler & Mueller 1993; Gomez et al. 2002).The merging scenario is further supported by the LOS velocity and velocity disper-sion fields (computed using the 10 nearest neighbors to each point) shown in Fig.


Figure 5.5: The LOS velocity field (left) and the velocity dispersion field (right) for thewhole region studied in this Chapter. The LOS velocity and the velocity dispersionare computed using the 10 nearest neighbors to each pixel, whose size is 36 arcsec2.The iso-density contours for the 146 confirmed cluster members are superposed inblack.

5.5. The SE subcluster has higher LOS velocity and velocity dispersion than the NWsubstructure. The region with the highest LOS velocity and velocity dispersion lies∼ 6 arcmin N from the X-ray cluster center and it coincides with the substructuredetected by the ∆ test. This result points out the presence of a group of galaxiesinfalling in the SE cluster core (see Sec.5.5.2).Thus the NW subcluster appears as a relaxed system with the lowest velocity disper-sion among the whole sample; on the other hand the SE subcluster appears far fromrelaxation, and it is probably experiencing a multiple merging event.We use the position of the gas temperature gradient, shown by the straight line inFig.5.4 (right), to divide our sample into two regions and to study separately thedynamical properties of the two subclusters.A sketch of the cluster dynamical model discussed in the next section is given inFig.5.6.

5.5.1 The North-West subcluster

The NW subcluster is composed of 86 galaxies and includes two density peaks: thehighest and a secondary one located at the western periphery of the subcluster (la-beled as W subcluster in Fig.5.6), with a weak X-ray counterpart. It has a similarmean location (CBI = 6480 ± 87 km s−1) and a lower scale (SBI = 770 ± 60 km s−1)than the whole cluster.


Figure 5.6: A 3D sketch of Abell 1367 summarizing the various sub-componentsdescribed in Section 5.5. The cluster is viewed from its near side, as suggested by theeyeball indicating the observer’s position.

Fig.5.8 shows the LOS velocity distribution of this subcluster. The W test rejectsthe Gaussian hypothesis at a confidence level of 39%. Thus the LOS velocity dis-tribution is consistent with a Gaussian distribution, suggesting that this subclusteris a virialized system. Moreover its increasing velocity dispersion profile (see Fig.5.9) is consistent with a relaxed cluster undergoing two body relaxation in the densecentral region, with circular velocities in the center and more isotropic velocities inthe external regions (Girardi et al. 1998).However this subcluster also shows some evidences of merging (see Fig.5.7). Thebrightest galaxy of this cloud CGCG97-095 (NGC3842), located ∼2 arcmin SE fromthe NW density peak, is a radio galaxy classified as a narrow-angle tail (NAT) (Blitonet al. 1998). The tail orientation (indicated with an arrow in Fig. 5.7) suggests thatthis galaxy (and the associated substructure) is moving from north-west to south-east, toward the main cluster core.Moreover two CGCG (Zwicky et al. 1961) galaxies, 97-073 and 97-079, show fea-


Figure 5.7: Blow-up of the NW substructure of Abell 1367. The arrows indicate thedirection of radio head tails associated with 97-079 and 97-073 and the orientation ofthe NAT radio galaxy 97-095. The dashed region shows the distribution of the diffusecluster radio relic (Gavazzi 1978). The iso-density contours for the 146 confirmedcluster members are superposed.

tures consistent with the infall scenario. Gavazzi et al. (1995, 2001a) found thatboth galaxies have their present star formation enhanced along peripheral HII re-gions which developed at the side facing the direction of motion through the clusterIGM. Their neutral hydrogen is significantly displaced in the opposite side (Dickey& Gavazzi 1991), where 50 kpc long tails are detected both in the light of the syn-chrotron radiation (Gavazzi & Jaffe 1987) and in Hα (Gavazzi et al. 2001a). Theobservational scenario is consistent with the idea that ram-pressure (Gunn & Gott1972) is enhancing for a limited amount of time the star formation of galaxies thatare entering the cluster medium for the first time.However these two galaxies appear not directly associated with the center of the NWsubcluster since they lie at a projected distance of ∼0.34 Mpc from the main densitypeak (see Fig.5.7). Moreover their large distance (∼0.48 Mpc) from the shock frontobserved in X-ray between the NW and the SE substructure indicates that these ob-jects do not belong to the main galaxy density peak infalling into the cluster center.Conversely they are at a projected distance of only 0.08 Mpc from the center of theW subcluster, suggesting that they are associated with this subcloud.For these reasons we consider an alternative scenario in which these two galaxies be-long to a secondary substructure infalling into the NW substructure from the westernside (see Fig. 5.6). This picture is supported by the presence of the extended radiorelic detected both in X-ray and radio continuum in this region (Gavazzi 1978; Gavazzi& Trinchieri 1983). Cluster radio halos contain fossil radio plasma, the former outflowof a radio galaxy, that has been revived by shock compression during cluster merging


Figure 5.8: The LOS velocity distribution for galaxies in the NW (upper) and in theSE (lower) subclusters.

(Enßlin et al. 1998; Enßlin & Bruggen 2002). The radio relic observed in Abell 1367extends, south-west to north-east, from 97-073 to 127-040 with a projected extent of0.8 Mpc (see Fig.5.7). The age of its electrons is estimated to be ∼ 0.2 Gyr (Enßlinet al. 1998). The only plausible source of high energy electrons available in this re-gion is the NAT galaxy 97-095, presently at ∼0.25 Mpc from the relic and whose tailspoint exactly in the relic direction. Assuming that the fossil radio halo originatedfrom 97-095, we find that the infall velocity of this galaxy into the SE subcluster isV ∼ 1250 km s−1, consistent with the typical infall velocity of cluster galaxies. Thusthe presence of the radio relic results consistent with a merging scenario in whichthe W subcluster, containing 97-079 and 97-073, is infalling into the NW substruc-ture, compressing the plasma ejected from 97-095 and re-accelerating the electrons torelativistic energies.

5.5.2 The South-East subcluster

The SE cloud is composed of 60 galaxies associated with the X-ray cluster center. Ithas the highest LOS velocity and dispersion of the whole sample (see Fig.5.5) with alocation CBI = 6596 ± 137 km s−1 and a scale SBI = 1001 ± 70 km s−1. Its velocitydistribution, shown in Fig. 5.8, appears significantly non-Gaussian. The W test re-jects the Gaussian hypothesis at a confidence level of 96.8%, supporting the idea that


Figure 5.9: The velocity dispersion radial profile of the NW (upper) and the SE(lower) subclusters.

the cluster center is far from relaxation. This is in agreement with the decreasingvelocity dispersion profile of this region (see Fig.5.9), consistent with isotropic veloci-ties in the center and radial velocities in the external regions, as expected in the caseof galaxy infall onto the cluster (Girardi et al. 1998).The velocity distribution of Fig. 5.8 has three peaks at ∼ 5500 km s−1, ∼ 6500 km s−1

and ∼ 8200 km s−1 respectively, probably associated with three separate groups.Moreover we remark that the galaxy gaps between the three peaks are fairly consis-tent with two of the most significant weighted gaps detected in the global velocitydistribution (V ∼ 5800 km s−1 and V ∼ 7500 km s−1).In order to check for any position-velocity segregation, we divide the SE subcluster inthree groups according to their LOS velocity: galaxies with V < 5800 km s−1 belongto the low velocity group, galaxies with V > 7500 km s−1 belong to the high velocitygroup and galaxies with intermediate velocity belong to the SE subcluster.The projected distribution of the three groups is shown in Fig.5.10. The high-velocity

group (V ∼ 8200 km s−1, triangles) appears segregated in the northern part of the SEcloud, extending ∼20 arcmin in right ascension but only ∼7 arcmin in declination.It is associated with the substructure detected by the ∆ test (see Fig. 5.3) and withthe infalling group of star-forming galaxies recently discovered by Sakai et al. (2002)and by Gavazzi et al. (2003b). Its spatial segregation and high star formation activitysuggest that this group is a separate unit infalling into the cluster, probably from the


Figure 5.10: The distribution of galaxies belonging to the South-East subcluster.Triangles indicate galaxies with LOS velocity > 7500 km s−1, circles galaxies withLOS velocity < 5800 km s−1 and squares galaxies with LOS velocity comprises in therange 5800 km s−1 < V < 7500 km s−1. The ROSAT X-ray contours are shown.

near side (see Fig. 5.6). It is remarkable that Sun & Murray (2002), using Chandraobservations of the cluster center, discovered a ridge-like structure around the clustercenter, ∼6 arcmin south from the center of the high velocity group, probably asso-ciated with a compact merging subcluster (perhaps this group) penetrating the SEcluster core.The low-velocity group (V ∼ 5500 km s−1, circles in Fig.5.10) seems segregated inthe eastern part of the cloud, perhaps infalling from the eastern side into the clustercore (Fig. 5.6). This scenario is also supported by the detection of cool gas streaminginto the cluster core from the eastern side (Forman et al. 2003), probably associatedwith this low velocity group of galaxies.Galaxies with V ∼ 6500 km s−1 (squares in Fig.5.10) are homogeneously distributedover the SE subcluster, representing its virialized galaxy population. However thebrightest galaxy in this group 97-127 (NGC3862) is a NAT radio galaxy with veryextended radio tails pointing in the direction of the low velocity group (Gavazzi et al.1981), suggesting motion relative to the IGM.The velocity-space segregation observed in the SE subcluster suggests that the clustercenter is experiencing multiple merging of at least two separate groups, supportingthe idea that it is far from relaxation. This picture is consistent with the high gas

5.6. Star formation activity in the infalling groups 59

Figure 5.11: The LOS velocity distribution for emission line (upper) and non emissionline galaxies (lower) in the whole cluster sample.

entropy in this region, since in absence of a cool dense core the substructures infallinginto the major cluster can penetrate deep inside, disturbing the cluster core dynamics(Churazov et al. 2003).A sketch of the various substructures identified in Abell 1367 by the present study, isgiven in Fig. 5.6. Five substructures are detected. Two clouds, the NW and SE sub-clusters, are in the early merging phase, meanwhile three smaller groups are infallinginto Abell 1367. The W subcloud, associated with the head-tail systems 97-073/79,is probably infalling into the NW subcluster, exciting the radio relic observed in be-tween the two structures. The other two groups are infalling into the SE subcluster:the low velocity group from the eastern side, while the high velocity group from thenear side.

5.6 Star formation activity in the infalling groups

The dynamical study presented in the previous sections indicates that Abell 1367 is adynamically young cluster in the early stage of a multiple merging event involving atleast five substructures. Since merging is expected to trigger star formation in clustergalaxies (Bekki 1999), we study separately the spatial and velocity distribution ofthe star forming galaxies. Only 49 out of the 146 cluster members show recent star


Figure 5.12: Projected density map of non emission line (left) and emission line (right)galaxies in Abell 1367. The iso-density contours of the 146 confirmed cluster membersare superposed.

formation activity (e.g. Hα line in emission, Iglesias-Paramo et al. 2002; Gavazzi et al.2003a; Cortese et al., in preparation). Fig.5.11 shows the LOS velocity distribution ofgalaxies divided into emission line (upper panel) and non emission line (lower panel)galaxies. The star forming sample has higher location and scale (CBI = 6704 ±168 km s−1, SBI = 1076 ± 76 km s−1) than the quiescent sample (CBI = 6446 ±79 km s−1, SBI = 738±58 km s−1). According to a two-sample Kolmogorov-Smirnovtest the two velocity distributions have only ∼5% probability of being consistent,suggesting a different origin and/or evolution. We remark that, if the star forminggalaxies are infalling onto the cluster along radial orbits, their velocity dispersionshould be ∼

√2 times the velocity dispersion of the relaxed sample, as observed in

this case. This result suggests that star forming systems are an infalling populationwhile the non-star forming galaxies represent the virialized cluster population.The projected density distribution of star forming and non star forming is shownin Fig.5.12. The highest density of non emission line systems is observed near thecenter of the NW substructure. This morphological segregation further supports theidea that the NW cloud is a relaxed system merging for the first time into the SEsubcluster.The emission line galaxies have a different distribution. The highest density of starforming systems is in the infalling groups, i.e. in the high velocity group infalling intothe SE subcluster and in the W cloud infalling into the NW substructure, suggestingthat their interaction with the cluster environment is triggering some star formationactivity. Indeed in these systems the fraction of star forming galaxies lies between64% and 36%, decreasing to 31% in the NW substructure and to 20% in the SEsubcluster.

5.7. Cluster mass 61

Sample RH MV MPM

Mpc 1014M 1014M

A1367 all types 0.41 7.04 ± 0.90 7.82 ± 2.50A1367 non-star forming 0.37 4.35 ± 0.70 5.11 ± 0.90A1367 NW all types 0.30 3.87 ± 0.62 6.12 ± 1.52A1367 NW non-star forming 0.24 2.47 ± 0.46 3.29 ± 0.59A1367 SE all types 0.27 5.80 ± 0.88 6.87 ± 1.20A1367 SE non-star forming 0.26 3.90 ± 0.83 5.58 ± 0.74

Table 5.5: Mass estimate for Abell 1367

5.7 Cluster mass

The virial theorem is the standard tool used to estimate the dynamical mass of galaxyclusters. Under the assumptions of spherical symmetry and hydrostatic equilibriumand if the mass distribution follows the distribution of the observed galaxies indepen-dent of their luminosity, the total gravitational mass of a cluster is given by

MV =3π

Gσ2RH

where σ is the galaxy velocity dispersion and RH is the cluster mean harmonic radius:

RH =N(N − 1)∑

i>j R−1ij

where N is the total number of galaxies.An alternative approach is to use the projected mass estimator (Heisler et al. 1985),defined as

MPM =32

πGN

∑

i

V 2i Ri

where Vi is the observed radial component of the velocity of the i galaxy with respectto the systemic cluster velocity, and Ri is its projected separation from the clustercenter. The numerical factor 32 assumes that galaxy orbits are isotropic. In case ofpurely radial or purely circular orbits this factor becomes 64 or 16 respectively.Mass estimates obtained using the two above methods and their uncertainties arelisted in Table 5.7. We remark that these mass estimates are probably biased by thedynamical state of Abell 1367, which appears far from virialization. In particularthe presence of substructures leads to an overestimate of the cluster mean harmonicradius and velocity dispersion, and thus of the virial mass (Pinkney et al. 1996). For


this reason the mass derived for the whole cluster and for the SE and NW subclustersseparately is probably overestimated. Assuming that the early type sample representsthe virialized cluster population (see previous section), we also derive mass estimatesfor the three dynamical units using the non star forming systems only.For all the studied samples the virial mass estimates are affected by smaller uncer-tainties and yield smaller values than the projected mass estimates. This can be dueto the contamination by interlopers (Heisler et al. 1985) or, more probably, to theassumption of isotropic orbits. Indeed assuming purely radial or circular orbits themass estimate varies by a factor of 2, becoming consistent with the virial mass.The mass inferred from the non-star forming population are, as expected, systemati-cally lower than the ones obtained from all types. The value obtained for the wholesample is consistent with the mass estimates available in the literature (MV = 7.26±1.40 1014M Girardi et al. 1998; MV = 6.07±0.93 1014M, MPM = 6.28±0.80 1014M

Rines et al. 2003).

5.8 Two-Body Analysis

In this section we investigate whether the two clouds A1367NW, A1367SE and thethree groups infalling into the SE and NW subclusters form gravitationally boundsystems. For each system we apply the two-body analysis described by Beers et al.(1991). The two subclumps are treated as point masses moving on radial orbits. Theyare assumed to start their evolution at time t=0 with zero separation, and are movingapart or coming together for the first time in their history. For bound radial orbits,the parametric solutions to the equations of motion are:

R =Rm

2(1 − cos χ)

t =( R3

m

8GM

)1/2

(χ − sin χ)

V =(2GM

Rm

)1/2 sin χ

(1 − cos χ)

where R is the components separation at time t, and V is their relative velocity. Rm

is the separation of the subclusters at maximum expansion and M is the total massof the system. Similarly, the parametric solutions for the unbound case are:

R =GM

V 2∞

(cosh χ − 1)

t =GM

V 3∞

(sinh χ − χ)

5.8. Two-Body Analysis 63

Figure 5.13: The bound and unbound orbit regions in the (Vrel, α) plane. Thebound-incoming solutions (BIa and BIb), the bound-outgoing solutions (BO) andthe unbound-outgoing (UO) solutions are indicated with solid lines. The dotted linesshow the dividing line between bound and unbound regions. The vertical solid linesrepresent the observed Vrel and the dashed regions their associated 1σ uncertainty.


V = V∞

sinh χ

(cosh χ − 1)

where V∞ is the asymptotic expansion velocity.The system parameters V and R are related to the observables Vrel (the LOS relativevelocity) and Rp (the projected separation) by:

Vrel = V sin α, Rp = R cos α

where α is the angle between the plane of the sky and the line joining the centers ofthe two components. The two systems are thus closed by setting the present time tot0 = 13 Gyr (the age of the Universe in a Ωm=0.3 and Ωλ=0.7 cosmology) and solvediteratively to determine the projection angle as a function of Vrel.We determine two solutions for each two-body model, assuming two extreme valuesfor the total mass of each system ranging from the virial mass of the non-star formingpopulation to the virial mass of the whole cluster. Table 5.8 summarizes the adoptedparameters of the two-body analysis, and Fig. 5.13 shows the computed solutions inthe (α, Vrel) plane. The vertical lines represent the observed values of Vrel and thedashed regions their associated 1σ uncertainties.The solutions have three different regimes: an unbound-outgoing regime (UO), abound-outgoing regime (BO) and a bound-ingoing regime (BI). It is easy to showthat the unbound solutions will lie in the region of the (α, Vrel) plane where:

V 2relRp ≤ 2GMtot sin2 α cos α.

The dotted lines in Fig. 5.13 show the dividing line between bound and unboundregions.In the BO regime, the two subclumps are still separating and have not yet reachedthe maximum expansion.The BI regime describes the system after maximum expansion. For each Vrel, thereare two corresponding values of α, a large and a small one. The large value assumesthat the substructures are far apart, with low relative velocity, while the small valueimplies that the subclusters are close together near the plane of the sky (see Fig. 7in Beers et al. 1991). Thus we split the BI regime into two branches, called BIa andBIb.The probability of each solution, computed following the procedure described byBeers et al. (1991), is given in Table 5.8. Our result is that the A1367NW/SE andthe A1367SE/High Velocity group systems are bound with 100% probability andpresently infalling with 96% and 100% probability respectively. The A1367NW/Wand the A1367SE/Low Velocity group systems are bound at 99% and 96% probabilityrespectively. We conclude that all systems constituting Abell 1367 are gravitationallybound at ≥ 96% probability.

5.9. Conclusions 65

System Mtot Vrel ± ∆Vrel Rp Solution ProbabilityBIa BIb BO UO

1014 M km s−1 Mpc % % % %

A1367NW/SE 7.04 84 ± 162 0.45 57 40 3 04.35 84 ± 162 0.45 55 41 4 0

A1367NW/W 7.04 500 ± 200 0.37 57 40 2 12.47 500 ± 200 0.37 56 41 2 1

A1367SE/Low Vel. gr. 7.04 1000 ± 200 0.38 58 40 0 23.90 1000 ± 200 0.38 57 39 0 4

A1367SE/High Vel. gr. 7.04 1500 ± 200 0.08 56 44 0 03.90 1500 ± 200 0.08 58 42 0 0

Table 5.6: Two-body model parameters

5.9 Conclusions

I have presented a dynamical analysis of the central ∼ 1.3 square degrees of the galaxycluster Abell 1367, based on 273 redshift of which 119 are new measurements. TheLOS velocity distribution of the 146 cluster members is significantly non Gaussian,suggesting that the cluster is dynamically young. The member galaxies show anelongated distribution along the NW-SE direction with two major density peaks,consistent with the X-ray morphology. The strong difference in the LOS velocity andvelocity dispersion of the two density peaks, the abrupt gas temperature gradientdetected in X-rays and the 3D statistical tests support a merging scenario involvingat least two subclusters. Moreover the dynamical properties of the NW and SE cloudssuggest an even more complex picture, summarized in Fig. 5.6. At least anothergroup of star forming galaxies (the high velocity group) infalling into the cluster coreis detected, suggesting a multiple merging event. Furthermore our analysis suggeststhe presence of two other groups infalling into the cluster center. In the North-Westpart of Abell 1367 a group of galaxies (W subcluster), associated with the infallinggalaxies 97-073/79 and with the radio relic observed in this region, is probably mergingwith the relaxed core of the NW subcluster. In the South part another group (thelow velocity group) is infalling from the eastern side into the disturbed core of theSE subcluster. These three subgroups have a higher fraction of star forming galaxiesthan the cluster core, as expected during the early phase of merging events.The multiple merging scenario is consistent with the location of Abell 1367 being atthe intersection of two filaments, the first extending roughly 100 Mpc from Abell 1367toward Virgo (West & Blakeslee 2000) and the second extending between Abell 1367


and Coma (as a part of the Great Wall, Zabludoff et al. 1993). As predicted by Katz& White (1993) this is the natural place for Abell 1367 to evolve into a rich relaxedcluster.

5.9. Conclusions 67

Name R.A. Dec. r’ V Tel.(J.2000) (J.2000) mag km s−1

114000+195426 114000.62 195426.7 15.98 10883 CAN114159+193227 114159.52 193227.3 15.63 21228 CAN114200+195846 114200.83 195846.0 17.09 6420 WHT114208+191905 114208.01 191905.0 19.04 23456 WHT114212+195650 114212.47 195650.3 17.73 20278 WHT114213+193001 114213.87 193001.6 16.92 23641 WHT114215+200427 114215.59 200427.0 19.20 6100 WHT114219+200548 114219.15 200548.0 16.45 6841 CAN114224+195329 114224.39 195329.8 18.29 31440 WHT114224+191157 114224.48 191157.0 16.39 28546 CAN114226+194317 114226.24 194317.1 17.50 23416 WHT114230+191447 114230.62 191447.5 18.60 27304 WHT114230+192553 114230.95 192553.8 17.80 45683 WHT114238+194718 114238.24 194718.6 17.04 25610 WHT114239+195145 114239.78 195145.9 19.06 53710 WHT114240+195627 114240.26 195627.5 18.63 19946 WHT114243+191615 114243.81 191615.8 18.72 5312 WHT114249+193935 114249.85 193935.1 19.14 72429 WHT114250+193955 114250.47 193955.7 19.22 13759 WHT114252+195656 114252.17 195656.4 16.69 5936 WHT114254+193851 114254.40 193851.3 17.17 6406 WHT114254+194033 114254.93 194033.6 18.87 71389 WHT114258+194321 114258.13 194321.1 18.98 6523 WHT114258+194053 114258.37 194053.9 19.25 71436 WHT114258+194644 114258.53 194644.2 19.00 88274 WHT114258+195612 114258.94 195612.7 18.41 7059 WHT114259+194801 114259.71 194801.1 18.92 71600 WHT114300+192515 114300.65 192515.2 18.42 53145 WHT114301+194758 114301.24 194758.9 18.67 72572 WHT114301+195313 114301.97 195313.5 18.61 46935 WHT114307+192807 114307.13 192807.3 17.37 32298 WHT114307+193029 114307.16 193029.8 17.93 23763 WHT114310+192526 114310.09 192526.4 16.62 19188 WHT114310+191519 114310.29 191519.2 17.41 23578 WHT114313+200747 114313.18 200747.9 16.40 5383 CAN114314+194821 114314.49 194821.7 19.29 71433 WHT114314+192534 114314.99 192534.3 15.76 23867 CAN114317+195525 114317.25 195525.1 18.88 30273 WHT114317+194658 114317.61 194658.2 15.69 6295 CAN114318+201523 114318.05 201523.3 17.81 46170 WHT114319+192520 114319.68 192520.9 15.99 6757 WHT114320+193637 114320.44 193637.1 19.71 44171 WHT114320+195206 114320.66 195206.2 18.15 52416 WHT114322+195704 114322.06 195704.7 16.94 7909 WHT114324+194121 114324.66 194121.4 18.33 35778 WHT114332+201326 114332.24 201326.1 16.46 33438 CAN114332+195108 114332.72 195108.2 19.07 14313 WHT114335+200005 114335.47 200005.6 16.38 20600 CAN114336+193930 114336.07 193930.8 19.24 44616 WHT114337+193835 114337.17 193835.8 17.31 12502 WHT114337+201533 114337.82 201533.5 20.19 11464 WHT114339+193446 114339.09 193446.2 16.01 7477 WHT114342+193636 114342.18 193636.3 19.26 71296 WHT114343+195607 114343.12 195607.8 18.56 19711 WHT114345+201252 114345.50 201252.2 19.27 20476 WHT114350+195702 114350.16 195702.0 17.98 6848 WHT114350+194138 114350.83 194138.0 19.13 72744 WHT114353+195004 114353.42 195004.6 19.23 27946 WHT114353+194422 114353.45 194422.2 15.66 6141 WHT114353+194315 114353.61 194315.8 17.17 23578 WHT

Table 5.7: The 119 new redshift measurements


Name R.A. Dec. r’ V Tel.(J.2000) (J.2000) mag km s−1

114356+201404 114356.80 201404.9 18.42 72058 WHT114357+201122 114357.69 201122.7 17.06 5348 WHT114358+195330 114358.86 195330.2 19.22 6200 WHT114359+195630 114359.51 195630.8 20.37 6992 WHT114402+194742 114402.65 194742.7 17.52 43665 WHT114403+200552 114403.70 200552.6 15.80 5698 WHT114404+192922 114404.17 192922.8 18.59 53335 WHT114404+195956 114404.65 195956.6 17.33 33830 WHT114407+193850 114407.21 193850.9 17.10 20877 WHT114407+193143 114407.71 193143.1 18.44 53424 WHT114412+195503 114412.22 195503.9 17.65 20916 WHT114412+195633 114412.27 195633.4 17.02 6244 WHT114412+201119 114412.92 201119.7 19.25 74731 WHT114415+193037 114415.25 193037.5 16.58 6502 WHT114415+193012 114415.33 193012.3 18.27 35227 WHT114417+194543 114417.28 194543.9 18.14 66264 WHT114422+194628 114422.16 194628.2 15.70 6527 CAN114426+195951 114426.10 195951.5 16.98 30102 WHT114430+194258 114430.30 194258.3 18.78 40347 WHT114432+195341 114432.19 195341.6 18.89 42649 WHT114432+194734 114432.98 194734.6 18.82 71100 WHT114447+201248 114447.20 201248.5 18.17 6699 WHT114449+195628 114449.72 195628.9 16.70 5539 WHT114501+195504 114501.97 195504.5 18.79 45708 WHT114503+193831 114503.00 193831.2 16.76 6193 LOI114503+194743 114503.14 194743.9 17.91 23374 WHT114504+201412 114504.25 201412.2 18.31 5477 WHT114505+194057 114504.83 194056.9 15.67 6506 LOI114506+200849 114506.38 200849.9 19.23 3822 WHT114509+194845 114509.38 194845.4 17.49 19831 WHT114509+193316 114509.40 193316.2 15.80 7409 LOI114509+194526 114509.65 194526.9 16.90 19834 WHT114516+193245 114516.18 193245.1 16.80 19669 WHT114517+200120 114517.10 200120.7 15.32 14745 LOI114517+201108 114517.29 201108.8 18.21 79253 WHT114517+200110 114517.64 200110.0 15.46 14713 LOI114520+194220 114520.33 194220.3 20.44 54544 WHT114520+193259 114520.49 193259.4 17.48 4653 WHT114522+195146 114522.62 195146.5 21.14 18012 WHT114524+201239 114524.33 201239.3 18.73 44376 WHT114526+201056 114526.27 201056.8 16.40 20134 CAN114529+195658 114529.39 195658.2 16.29 24000 WHT114530+193639 114530.37 193639.4 17.20 40000 LOI114531+200217 114531.31 200217.5 19.78 45691 WHT114533+194505 114533.88 194505.9 18.11 31440 WHT114533+200028 114533.97 200028.7 17.56 35830 WHT114536+194253 114536.19 194253.7 18.60 48966 WHT114540+194302 114540.32 194302.8 17.74 5545 WHT114543+193854 114543.65 193854.9 16.93 7828 LOI114543+193905 114543.77 193905.9 16.30 7301 LOI114544+194013 114544.86 194013.3 17.18 19487 WHT114545+193151 114545.66 193151.4 18.57 6880 LOI114545+201200 114545.78 201200.3 19.11 27431 WHT114548+192708 114548.13 192708.4 16.72 30193 WHT114549+195915 114549.88 195915.3 15.87 20035 CAN114550+194824 114550.61 194824.6 18.85 41484 WHT114602+194754 114602.12 194754.3 19.62 73746 WHT114605+195151 114605.35 195151.0 18.86 46635 WHT114620+194518 114620.85 194518.0 17.58 45683 WHT

Table 5.7: Continue

Chapter 6

Unveiling the evolution of earlytype galaxies with GALEX.

6.1 Introduction

In Chapters 3 and 4 I have shown that at low UV luminosities the contribution ofearly-type quiescent galaxies is not negligible. This represents the first evidence of amorphology/star formation - density relation at ultraviolet wavelengths and demon-strates that we cannot blindly assume all UV selected galaxies are star-forming sys-tems, especially at low UV luminosities and in high density environments. This alsopoints out the strong potential of ultraviolet observations for studying all clustergalaxies: not only star-forming systems in which UV emission traces the presenceof newly born stars, but also early type galaxies whose emission is usually ascribedto low mass old post asymptotic giant branch stars. The excess ultraviolet radia-tion from giant early-type galaxies is in fact supposed to arise from hot low massstars in late stages of stellar evolution (O’Connell 1999). All theoretical, spectral andimaging evidences have recently converged towards the view that the UV emissionoriginates from He-burning, extreme horizontal branch stars, their post-HB progenyand post-AGB stars in the dominant, metal rich stellar population of elliptical galax-ies. However it is still unknown whether the UV emission of all early type galaxiesis dominated by the contribution of old stellar populations independently from thegalaxy morphology (i.e. ellipticals vs. lenticulars) and luminosity (i.e. dEs vs. gi-ant Es). In particular it would be interesting to know if the UV properties of dwarfelliptical galaxies differ from those of giants, as much as other structural (Gavazziet al. 2005) and kinematic (van Zee et al. 2004) properties depend on luminosity,due to their different star formation histories (single episodic vs. burst) (Ferguson &Binggeli 1994; Grebel 2000). In fact, a recent burst of star formation would stronglycontribute to the UV emission of an elliptical galaxy, even if its stellar population is

69

70 6. Unveiling the evolution of early type galaxies with GALEX.

dominated by old low mass stars.Due to morphological segregation (Whitmore et al. 1993), nearby clusters are theideal targets for assembling complete, volume limited samples of early-type objects.As part of a study aimed at analyzing the environmental dependence of galaxy evo-lution, we observed large portions of the Virgo cluster with GALEX (Boselli et al.2005a). Owing to the superior quality of the photographic material obtained bySandage and collaborators, an extremely accurate and homogeneous morphologicalclassification exists for Virgo galaxies, down to mB ≤ 18 mag (MB ≤-13 assuming adistance of 17 Mpc), allowing a detailed discrimination among different subclasses ofearly-type galaxies (ellipticals, lenticulars, dwarfs) and from quiescent spirals. Fur-thermore a wealth of ancillary data for many Virgo members, covering a large portionof the electromagnetic spectrum from the visible to the infrared is available from theGOLDMine database (Gavazzi et al. 2003a).

6.2 Data

The analysis presented in this Chapter is based on an optically selected sample ofearly-type galaxies including giant and dwarf systems (E, S0, S0a, dE and dS0) ex-tracted from the Virgo Cluster Catalogue of Binggeli et al. (1985), which is completeto mB ≤18 mag (MB ≤ -13). The Virgo cluster region was observed in spring 2004as part of the All Imaging Survey (AIS) and of the Nearby Galaxy Survey (NGS)carried out by the Galaxy Evolution Explorer (GALEX) in two UV bands: FUV(λeff = 1528A, ∆λ = 442A) and NUV (λeff = 2271A, ∆λ = 1060A), covering 427 ob-jects. See Chapter 2 and Martin et al. (2005) and Morrissey et al. (2005) for detailson the GALEX instrument and data characteristics.The present sample includes all Virgo cluster early-type systems detected in the NUVGALEX band (264 objects, 194 from the NGS); of these, 126 (of which 74 from theNGS) have been also detected in the FUV. The resulting sample is thus ideal for theproposed analysis as it provides us with the first large volume-limited sample of ellip-tical, lenticular and dwarf galaxies spanning 4 dex in luminosity with homogeneousdata. Whenever available, we extracted fluxes from the deep NGS images, obtainedwith an average integration time of ∼ 1500 sec, complete to mAB ∼ 21.5 in the NUVand FUV. Elsewhere UV fluxes have been extracted from the less deep AIS images(∼ 70 sq. degrees), obtained with an average integration time of ∼ 100 sec, completeto mAB ∼ 20 in both the FUV and NUV bands. The resulting sample, although notcomplete in both UV bands, includes giants and dwarf systems: at a limiting magni-tude of MB ≤ -15, 71 % of the observed galaxies have been detected in the NUV, 46%in the FUV. All UV images come from the GALEX IR1.0 release. UV fluxes wereobtained by integrating GALEX images within elliptical annuli of increasing diameterup to the optical B band 25 mag arcsec−2 isophotal radii consistently with the optical

6.3. The UV properties of early-type galaxies 71

and near-IR images. Independent measurements of the same galaxies obtained indifferent exposures give consistent photometric results within 10% in the NUV and15% in the FUV in the AIS, and about a factor of two better for bright (NUV ≤16)galaxies. The statistical uncertainty in the UV photometry is on average a factor of∼ 2 better in the NGS than in the AIS especially for fainter objects.UV data have been combined with multifrequency data taken from the GOLDMinedatabase (http://goldmine.mib.infn.it; Gavazzi et al. 2003a). These are B and Vimaging data, mostly from Gavazzi et al. (2005) and Boselli et al. (2003a), and near-IR H imaging from Gavazzi et al. (2000, 2001c). Optical and near-IR data have onaverage a photometric precision of ∼ 10%. Spectroscopic metallicity index Mg2 andvelocity dispersion data come from GOLDMine or from Golev & Prugniel (1998) andBernardi et al. (2002).Galaxies analyzed in this Chapter are all bona-fide Virgo cluster members: given the3-D structure of the cluster, distances have been assigned following the subclustermembership criteria of Gavazzi et al. (1999a). Owing to the high galactic latitude ofVirgo, no galactic extinction correction was applied (AB ≤ 0.05).

6.3 The UV properties of early-type galaxies

Despite the complex 3-D structure of Virgo (Gavazzi et al. 1999a), the uncertaintyon the distance (hence on the luminosity) of the target galaxies, does not constitutea major source of dispersion in the determination of the color-magnitude (CMR) re-lation. Figure 6.1 shows various UV to optical and near-IR CMRs. Similar resultsare obtained if, instead of the mass-tracer H band luminosity (Zibetti et al. 2002),we use the B band absolute magnitude. The NUV to optical (Fig. 6.1b) and near-IR(Fig. 6.1a) CMRs are well defined and are similar to optical or near-IR CMRs, withbrighter galaxies having redder colors, independent of their morphological type: thecolor index (NUV −V ) increases by ∼ 2 magnitudes from dwarfs (LH ∼ 108 LH) togiants (LH ∼ 1011.5 LH), while (NUV −H) changes by ∼ 3 mag. A weak flatteningof the relation appears for LH ≥ 1010 LH. This behavior confirms the one reportedby Ferguson (1994) in the (B − V ) vs. MB CMR.On the contrary, the FUV to optical (Fig. 6.1d) and near-IR (Fig. 6.1c) CMRs differsystematically for dwarfs and giant systems: galaxies brighter than LH ∼ 109.5 LH

have similar red colors, while for LH ≤ 109.5 LH colors become progressively bluer.Even if this trend can be due to a selection effect, (reddest dwarfs being undetectablein the FUV), it is indisputable that there exists a significant population of dEs withbluer colors than Es and S0s. The dichotomy between giants and dwarfs is even moreapparent in the UV color index (FUV − NUV ) (see Fig. 6.2). The (FUV − NUV )becomes redder with increasing luminosity for dwarf ellipticals while, on the contrary,it becomes bluer for giant ellipticals (Fig. 6.2a). The blueing relation is tight among


Figure 6.1: The near-UV (left column) and far-UV (right column) to optical and near-IR color magnitude relations. Colors are in the AB magnitude system. Open circlesare for ellipticals, filled circles for dwarfs, crosses for lenticulars (S0-S0a). Galaxiesredder than the dashed line are undetectable by the present survey (at the NGS limit).Largest 1σ errors for luminous and dwarf systems are given.

6.3. The UV properties of early-type galaxies 73

Table 6.1: Main relations for early type galaxies

x y a b R rms

Ellipticals1

LH FUV − NUV −0.30 ± 0.14 +4.52 ± 1.52 −0.47 0.31LH FUV − H −0.22 ± 0.19 +10.55 ± 2.10 −0.28 0.43LH NUV − H 0.17 ± 0.18 +4.85 ± 1.85 0.22 0.47LH FUV − V −0.15 ± 0.18 +8.38 ± 1.88 −0.21 0.38LH NUV − V 0.26 ± 0.12 +2.55 ± 1.30 0.45 0.31B − H FUV − NUV −0.84 ± 0.45 +3.22 ± 0.98 −0.43 0.32σ FUV − NUV −1.35 ± 0.37 +4.39 ± 0.89 −0.69 0.26

Lenticulars

LH FUV − NUV −0.28 ± 0.15 +4.40 ± 1.62 −0.31 0.45LH FUV − H 0.31 ± 0.21 +0.75 ± 2.00 0.27 0.58LH NUV − H 0.61 ± 0.11 +0.51 ± 1.17 0.65 0.36LH FUV − V 0.03 ± 0.23 +6.62 ± 2.38 0.03 0.59LH NUV − V 0.49 ± 0.09 +0.26 ± 1.00 0.68 0.25B − H FUV − NUV −1.00 ± 0.32 +3.70 ± 0.70 −0.49 0.42σ FUV − NUV −1.29 ± 0.39 +4.28 ± 0.84 −0.58 0.39

Dwarfs

LH FUV − NUV 1.73 ± 0.41 −13.90 ± 2.16 0.52 0.59LH FUV − H 2.55∗ ± 0.55 −15.97∗ ± 4.96 0.68∗ 0.91∗

LH NUV − H 0.91 ± 0.19 −2.72 ± 1.68 0.56 0.57LH FUV − V 1.91∗ ± 0.55 +11.35∗ ± 4.93 0.60∗ 0.87∗

LH NUV − V 0.63 ± 0.17 +1.28 ± 1.05 0.49 0.47B − H FUV − NUV 0.95 ± 0.45 +0.12 ± 0.73 0.40 0.60σ FUV − NUV − − −

Notes to Table:Col. 1 and 2: x and y variablesCol. 3 and 4: slope a and intercept b of the bisector linear fit with weighted variablesCol. 5: Pearson correlation coefficientCol. 6: mean dispersion around the best fit1: excluding VCC 1499*: uncertain values because of the UV detection limit


Figure 6.2: The relationship between the UV color index (FUV − NUV ) and a)the total H band luminosity, b) the B-H color index, c) the logarithm of the centralvelocity dispersion and d) the Mg2 index. Symbols are as in fig. 6.1. Labeled pointsindicate objects having unusual radio or optical properties (see Sect. 3).

6.4. Discussion and conclusion 75

ellipticals (see Table 1) and barely observed in lenticulars because of their higherdispersion 1.A similar behavior between ellipticals and lenticulars is observed in the (FUV −NUV )color relation (Fig. 6.2b): this mixed giant population becomes bluer in the UV withincreasing reddening in the (B − H) color index.The behavior of dwarf ellipticals is different: although with a huge dispersion, the(FUV −NUV ) color index reddens as the (B−H) and the other optical color indexes.The dichotomy between dwarf and giant systems cannot be observed in the run of(FUV −NUV ) vs. central velocity dispersion (which is directly related to the systemtotal dynamical mass; Fig. 6.2c) nor as a function of the metallicity sensitive (Pog-gianti et al. 2001) Mg2 Lick index (Fig. 6.2d) because these two parameters are notavailable for dwarfs. In ellipticals and lenticulars the UV color index (FUV −NUV )depends on both the metallicity index Mg2 and σ in a way opposite to the behavior atoptical wavelengths, where galaxies are redder when having higher Mg2 and velocitydispersions.

6.4 Discussion and conclusion

For the first time the UV properties of early-type galaxies have been studied down toMB ∼ -15 mag. The comparison with previous studies is thus limited to the brightestobjects. Our CMR can be compared with the one obtained by Yi et al. (2005) basedon a complete sample of bright early-type objects (Mr ≤ -20 mag) extracted fromthe Sloan Digital Sky Survey (SDSS) by Bernardi et al. (2003). The CMR presentedby Yi et al. (2005) (NUV − r vs. Mr) shows a significantly larger dispersion (σ ≥1.5 mag) than the one found in Virgo (see Table 1). As discussed in Yi et al. (2005),the large dispersion in their CMR can be ascribed to galaxies with a mild or residualstar formation activity included in the Bernardi et al. (2003) sample. If restricted tothe ”UV weak” sample, the dispersion in the Yi et al. relation drops to 0.58 mag,i.e. still larger than the one seen in the Virgo cluster in the same luminosity range.Despite possible larger distance uncertainties in the SDSS, the difference in the scat-ter between our and the Yi et al. (2005) CMR might arise from the classification inthe SDSS that uses concentration indices and luminosity profiles in discriminatinghot from rotating systems. It is in fact conceivable that the larger dispersion in theCMR of ”UV weak” galaxies of Yi et al. (2005) comes from the contamination of qui-

1The scatter in the blueing relation among ellipticals decreases significantly (from 0.31 to 0.10) ifwe exclude the misclassified post-starburst dwarf VCC 1499 (Gavazzi et al. 2001c; Deharveng et al.2002), the radio galaxy M87, VCC 1297 (the highest surface brightness galaxy in the sample ofGavazzi et al. (2005)) and VCC 1146. Beside its extremely high surface brightness, making VCC1297 a non standard object, we do not have any evidence indicating a peculiar star formation historyor present nuclear activity in VCC 1297 and VCC 1146 that could justify their exclusion.


escent, bulge-dominated Sa spiral disks, that have structural (concentration indicesand light profiles) or population properties (colors and spectra) similar to ellipticalsand lenticulars (Scodeggio et al. 2002).The monotonic increase of the (NUV − V ) and (NUV − H) colors with luminosity,similar to the one observed in the visible bands by Ferguson (1994) strongly suggeststhat both in dwarfs and giant systems the NUV 2310 A flux is dominated by the samestellar population (main sequence low mass stars) emitting at longer wavelengths. Onthe contrary the different behaviour of the (FUV − V ) and (FUV − H) colors withluminosity, and the clear dichotomy observed in the (FUV − NUV ) vs LH CMRstrongly support a different origin for the FUV emission in dwarf and giant systems.The reddening of the UV color index with luminosity observed in dwarf ellipticals,similar to the one observed in late type galaxies, indicates that the UV spectral energydistribution of low mass early type galaxies is shaped by the contribution of youngstellar populations. This is shown in Fig.6.3 where the available optical spectra forour sample of dEs are shown. It clearly emerges that UV bluer systems have emissionlines or strong Balmer line in absorptions witnessing present or recent star forma-tion activity. Moreover at increasing luminosity their (FUV − NUV ) color indexreddens as the optical colors confirming that in these systems the FUV emission isdominated by the contribution of young main sequence stars. This is not the casefor giant early type systems: the plateau observed in the FUV-optical CMRs and theblueing of the (FUV − NUV ) color with luminosity (i.e. the UV upturn) suggestthat far ultraviolet emission comes from low mass old post asymptotic giant branchstars. This is also confirmed in Figs.6.5 and 6.4 where the optical spectra availablefor ellipticals and S0s in our sample are presented: as expected, all the spectra aredominated by the contribution of the old stellar populations. Moreover the observedtrend between (FUV −NUV ) and the metallicity sensitive Mg2 index, reproduced bymodels (Bressan et al. 1994; Yi et al. 1998), confirms the early IUE result of Bursteinet al. (1988). Conversely Rich et al. (2005) did not find any correlation between thecolor index (FUV −r) and Mg2 nor with the velocity dispersion σ in a large sample ofSDSS early-type galaxies observed by GALEX. Their lack of correlation might derivefrom insufficient dynamic range in Log σ (2.1-2.4 km s−1) and Mg2 (0.18-0.30). Theblueing of the UV color index with luminosity, metallicity and velocity dispersionindicates that the UV upturn is more important in massive, metal rich systems. Thisis consistent with stellar population models which predict that the strength of theUV upturn is mainly driven by stellar metallicity.The accurate morphological classification in our sample allow us to discriminate be-tween E and S0s and to study separately the two populations. The higher dispersionin the (FUV −NUV ) vs. LH relation observed for the lenticulars compared to the ex-tremely tight one for ellipticals (see Table 1), bears witness to a different evolutionaryhistory for the two Hubble types: while cluster ellipticals represent an homogeneouspopulation, S0s are a heterogeneous class probably formed by different independent


Figure 6.3: The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectra available fordwarf ellipticals are presented.


Figure 6.4: The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectra available forellipticals are presented.


Figure 6.5: The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectra available forlenticulars are presented.


physical mechanisms (see also Chapters 9 and 11), and with various star formationhistories as also determined from kinematic and spectroscopic observations (Dressler& Sandage 1983; Neistein et al. 1999; Hinz et al. 2003).Using the available optical spectra we investigate the presence of residual star for-mation still present in our sample of giant early type galaxies. Only one S0 galaxy,VCC1003, shows a mild residual star formation activity (Hα in emission), while threeellipticals (VCC881,M87,VCC1619) and three S0s (VCC1030,VCC1062,VCC1253)have [NII] in emission and Hα in absorption, a typical feature of low ionization activegalactic nuclei. This suggest that the difference observed between ellipticals and S0scannot be ascribed to recent episodes of star formation but probably resides on theirdifferent past star formation history. Combining this result with the one obtained inChapter 4, we can conclude that, at low UV luminosities, the significant contributionof giant early type systems to the ultraviolet luminosity function must be ascribednot to young stellar populations but to old low mass post-AGB stars.The newest result of this Chapter, shown in Fig. 6.2, addresses the question raisedby O’Connell (1999) concerning the dependence of the UV properties on galaxy mor-phology. We have shown that a dichotomy exists between giant and dwarf ellipticalsand, to a lesser extent, between ellipticals and lenticulars. The opposite behavior(reddening of the UV color index with luminosity) of dwarfs with respect to giants,similar to that observed for spirals, indicates that the UV spectra of low luminosityobjects are shaped by the contribution of young stars, thus are more sensitive to thegalaxy’s star formation history than to the metallicity. This implies that the stellarpopulation of dwarfs has been formed in discrete and relatively recent episodes, asobserved in other nearby objects (Grebel 2000).More evidences are building up that mass drives the star formation history in hotsystems (Trager et al. 2000; Gavazzi et al. 2002a; Caldwell et al. 2003; Poggianti2004b) as in rotating ones (Gavazzi et al. 1996; Boselli et al. 2001) and that thestellar population of massive ellipticals is on average older than that of dwarfs.

Chapter 7

UV dust attenuation in normalstar forming galaxies

7.1 Introduction

The use of ultraviolet emission in order to study the properties of star forming galax-ies is not an easy a rapid task. The presence of dust in galaxies represents one of themajor obstacles complicating a direct quantification of the star formation activity inlocal and high redshift galaxies. Absorption by dust grains reddens the spectra atshort wavelengths completely modifying the spectral energy distribution of galaxies.Since the UV radiation is emitted by young stars (t < 108 yr) that are generally moreaffected by attenuation from surrounding dust clouds than older stellar populations,rest-frame UV observations can lead to incomplete and/or biased reconstructions ofthe star formation activity and star formation history of galaxies affected by dustabsorption, unless proper corrections are applied.In recent years our understanding of dust attenuation received a tremendous impulsefrom studies of local starburst galaxies (i.e.Calzetti et al. 1994; Heckman et al. 1998;Meurer et al. 1999; Calzetti 2001; Charlot & Fall 2000), that were based on threeindicators: the ratio of the total infrared to far-ultraviolet emission (LTIR/LFUV ),the ultraviolet spectral slope β (determined from a power-law fit of the form f ∼ λβ

to the UV continuum spectrum in the range 1300 and 2600 A, Calzetti et al. 1994)and the Balmer decrement. The total-IR (TIR) to UV luminosity ratio method (i.e.Buat 1992; Xu & Buat 1995; Meurer et al. 1995, 1999) is based on the assumptionthat a fraction of photons emitted by stars and gas are absorbed by the dust. Thedust heats up and subsequently re-emits the energy in the mid- and far-infrared.The amount of UV attenuation can thus be quantified by means of an energy bal-ance. This method is considered the most reliable estimator of the dust attenuationin star-forming galaxies because it is almost completely independent of the assumed

81

82 7. UV dust attenuation in normal star forming galaxies

extinction mechanisms (i.e. dust/star geometry, extinction law, see Buat & Xu 1996;Meurer et al. 1999; Gordon et al. 2000; Witt & Gordon 2000). When the spectrum isdominated by a young stellar population the ultraviolet spectral slope β, is found tohave a weak dependence on metallicity, IMF, and star formation history (Leitherer &Heckman 1995). Thus the difference between the observed β and the one predicted bymodels can be entirely ascribed to dust attenuation (Meurer et al. 1999). However insystems with no or mild star formation activity the UV spectral slope can be stronglycontaminated by the old stellar populations, whose contribution increases β (flattensthe UV continuum, Boissier et al. 2005). Thus the spectral slope of mildly star form-ing systems could be intrinsically different from the one of starburst galaxies, even inthe absence of dust attenuation (Kong et al. 2004).Meurer et al. (1999) have shown that in starburst galaxies the total far-infrared to ul-traviolet luminosity ratio correlates with the ultraviolet spectral slope, β (commonlyreferred to as the IRX-UV relation). They pointed out that this relation allows reli-able estimates of the attenuation by dust at ultraviolet wavelengths based on β.The Balmer decrement gives an estimate of the attenuation of ionized gas and not ofthe stellar continuum as in the previous two methods. It is based on the comparisonof the observed Hα/Hβ ratio with its predicted value (2.86 for case B recombina-tion, assuming an electronic density ne ≤ 104 cm−3 and temperature ∼ 104 K; e.g.,Osterbrock 1989). Calzetti et al. (1994) found a significant correlation between theultraviolet spectral slope β and the Balmer decrement Hα/Hβ. Starting from thisempirical relation they obtained an attenuation law (known as the Calzetti attenua-tion law) often adopted to correct UV observations for dust attenuation in absence ofboth far-infrared observations and estimates of the ultraviolet spectral slope (Steidelet al. 1999; Glazebrook et al. 1999).Unfortunately the above empirical relations have been established only for starburstgalaxies and they seem not to hold for normal star forming galaxies. Recently, Bell(2002) suggested that quiescent galaxies deviate from the IRX-UV relation of star-burst galaxies, because they tend to have redder ultraviolet spectra at fixed totalfar-infrared to ultraviolet luminosity ratio. Kong et al. (2004) confirmed this resultand interpreted the different behaviour of starbursts and normal galaxies as due toa difference in the star formation histories. They proposed that the offset from thestarburst IRX-UV relation can be predicted using the birthrate parameter b (e.g. theratio of the current to the mean past star formation activity). However an inde-pendent observational confirmation of the correlation between the distance from thestarburst IRX-UV relation and the birthrate parameter has not been obtained so far(Seibert et al. 2005). Even the Calzetti law does not seem to be universal. Buat et al.(2002) showed that for normal star forming galaxies the attenuation derived from theCalzetti law is ∼0.6 mag larger than the one computed from the FFIR/FUV ratio andtheir result has been recently confirmed by Laird et al. (2005).Why do normal star-forming galaxies behave differently from starbursts? Do normal

7.1. Introduction 83

galaxies follow different empirical relations that can be exploited to correct for dustattenuation in absence of far infrared observations? If this is the case, is there atransition between starburst and normal galaxies? Which physical parameters driveit? Answering these questions will be important for a better understanding of theinteraction of dust and radiation specifically in nearby dusty star forming galaxies,but it also has direct consequences for our understanding and interpretation of galaxyevolution in a general context. Firstly it seems mandatory to characterize the dust at-tenuation properties of normal galaxies, to compare them with the ones of starburstsand to derive new recipes for the UV dust attenuation correction. This topic cameonce again to the fore with the launch of the Galaxy Evolution Explorer (GALEX).This satellite is delivering to the community an unprecedented amount of UV dataon local and high redshift galaxies that require corrections for dust attenuation butcurrently lack far-infrared rest-frame data. The time is ripe to explore new methodsfor correction of these data, that might provide new insights on galaxy evolution.Whenever they can be combined with other data, GALEX observations provide thebest available ultraviolet data for studying the dust attenuation properties of galaxies.Multiwavelength photometric and spectroscopic observations are in fact mandatoryin order to: determine metallicity, ionized gas attenuation (A(Hα)), luminosity andmass, test the validity of the relations followed by starbursts (Heckman et al. 1998),explore relations that might prove useful to correct ultraviolet magnitudes and tocompare them with various models of dust attenuation. Recent extensive spectro-scopic and photometric surveys, like the Sloan Digital Sky Survey (SDSS, Abazajianet al. 2005) and the Two Degree Field Galaxy Redshift Survey (2dF, Colless et al.2001) have opened the path to studies of fundamental physical parameters basedon enormous datasets. However, spectroscopic observations of nearby galaxies sufferfrom strong aperture effects, making these datasets not ideal for the purpose of thepresent investigation. In fact, Kewley et al. (2005) have recently shown that apertureeffects produce both systematic and random errors on the estimate of star-formation,metallicity and attenuation. To reduce at least the systematic effects they suggestselecting only samples with fibres that capture > 20% of the light. This requiresz > 0.04 and z > 0.06 for SDSS and 2dF respectively: too distant to detect bothgiant and dwarf star forming systems with GALEX and IRAS.Although significantly smaller than the SDSS, the dataset we have been building upover the last 10 years with data taken over a large stretch of the electromagnetic spec-trum for few thousand galaxies in the local universe (worldwide available from thesite GOLDMine; Gavazzi et al. 2003a) turns out to be appropriate for the purposesof the present investigation. It includes drift-scan mode integrated spectra, narrowband Hα and broad band optical and near-infrared imaging for a volume limited sam-ple of nearby galaxies in and outside rich clusters. The combination of GALEX andIRAS observations with these ancillary data allows us to study the dust attenuationproperties in a sizable sample of normal star forming galaxies not suffering from the


aperture bias and to compare observations with model predictions.

In this chapter I investigate the relations between dust attenuation and globalgalaxy properties and compare them with the ones observed in starburst galaxies. Theaim of this work is to provide some empirical relations based on observable quantities(thus model independent) suitable for deriving dust attenuation corrections when farinfrared data are not available. For this reason all relations obtained throughout thischapter will be given as a function of LTIR/LFUV , the observable that we considerthe best dust attenuation indicator. We choose not to transform LTIR/LFUV intoa (model dependent) estimate of the far ultraviolet extinction A(FUV ), leaving thereader free to choose his/her preferred dust model (i.e. Meurer et al. 1999; Buatet al. 1999, 2002, 2005; Gordon et al. 2000; Panuzzo et al. 2003; Burgarella et al.2005, Inoue et al. in preparation). We assume that quantities are related linearly andresidual plots are presented in order to test the validity of this hypothesis. Moreover,since we are looking for new recipies to estimate the LTIR/LFUV ratio, this quantityhas to be considered as the dependent variable, implying the use of an unweightedsimple linear fit to estimate the best fitting parameters (Isobe et al. 1990).

7.2 The Data

7.2.1 The optically-selected sample

The analysis presented in this work is based on an optically selected sample of late-type galaxies (later than S0a) including giant and dwarf systems extracted from theVirgo Cluster Catalogue (VCC, Binggeli et al. 1985) and from the CGCG catalogue(Zwicky et al. 1961). The data include ∼ 300 square degrees covering most of theVirgo, Abell1367 and Abell262 clusters, the southwest part of the Coma cluster andpart of the Coma-A1367 supercluster (11h30m < R.A. < 13h30m; 18 < decl. < 32)observed in spring 2004 as part of the All-sky Imaging Survey (AIS) and of theNearby Galaxy Survey (NGS) carried out by GALEX in two UV bands: FUV(λeff = 1530A, ∆λ = 400A) and NUV (λeff = 2310A, ∆λ = 1000A). Details ofthe GALEX instrument and characteristics can be found in Martin et al. (2005) andMorrissey et al. (2005). Our sample has the quality of being selected with the criterionof optical completeness. All galaxies brighter than a threshold magnitude are selectedin all areas. In Coma-A1367 supercluster and A262 cluster all galaxies brighter thanmp=15.7 were selected from the CGCG catalogue (Zwicky et al. 1961). The Virgoregion contains all galaxies brighter than mp=18 from the VCC catalogue (Binggeliet al. 1985). We thus consider our sample an optically selected, volume limited sam-ple.We include in our analysis all late-type galaxies, detected in both NUV and FUVGALEX bands and in both 60 µm and 100 µm IRAS bands (157 objects). When-

7.2. The Data 85

ever available, we extracted UV fluxes from the deep NGS images, obtained with amean integration time of ∼ 1500 sec, complete to mAB ∼ 21.5 in the NUV and FUV.Elsewhere UV fluxes have been extracted from the shallower AIS images (∼ 70 sq. de-grees), obtained with a mean integration time of ∼ 100 sec, complete to mAB ∼ 20 inboth the FUV and NUV bands. All UV images come from the Internal Data Releasev1 (IR1.0). UV fluxes were obtained by integrating GALEX images within ellipticalannuli of increasing diameter up to the optical B band 25 mag arcsec−2 isophotalradii, consistently with the optical and near-IR images. Independent measurementsof the same galaxies obtained in different exposures give consistent photometric re-sults within 10 % in the NUV and 15% in the FUV in the AIS, and a factor of ∼two better for bright (NUV ≤16) galaxies. The uncertainty in the UV photometry ison average a factor of ∼ 2 better in the NGS than in the AIS, particularly for faintobjects. The typical uncertainty in the IRAS data is 15% (Boselli et al. 2003a).UV and far-infrared data have been combined to multifrequency data. These are opti-cal and near-IR H imaging (mostly from Gavazzi et al. 2000, 2005; Boselli et al. 2003a),optical drift-scan spectra (Gavazzi et al. 2004; Gavazzi et al. in prep.) and Hα imag-ing (Boselli & Gavazzi 2002; Boselli et al. 2002a; Gavazzi et al. 1998, 2002b; Iglesias-Paramo et al. 2002; Gavazzi et al. in prep.), great part of which are available fromthe GOLDMine galaxy database (Gavazzi et al. 2003a) (http://goldmine.mib.infn.it).From the 157 galaxies selected we exclude Active Galactic Nuclei (AGN). AGNs havebeen selected using either the classification provided by NED, if available, or by in-spection to the integrated spectra of Gavazzi et al. (2004): we exclude galaxies withlog([OIII]/Hβ) > 0.61/(log([NII]/Hα)−0.05)+1.3 (Kauffmann et al. 2003a). Thiscriterion reduces the sample to 128 galaxies, spanning a range of six magnitudes in Bband (-22< MB <-16) and of three orders of magnitude in mass1 (9 < M < 12 M).Unfortunately ancillary data are not available for all galaxies observed by GALEX,we thus further divided the data in two subsamples. Sixty six galaxies in the primarysample have all the necessary complementary data (e.g. Hα photometry, Hα/Hβratio, metallicity, H-band photometry; see Gavazzi et al. 2000, 2002a,b, 2004 for theselection criteria adopted in each survey). The remaining 62 galaxies form the sec-ondary sample. We cannot exclude a possible contamination of AGN in the secondarysample, since no spectra are available for these objects. In all figures objects belongingto the primary sample will be indicated with filled circles while the secondary sampleas empty circles. Since only galaxies belonging to the primary sample are present inall the plots analyzed in the presented work, all correlations will be quantified usingonly the primary sample. Data from UV to near-IR have been corrected for Galacticextinction according to Burstein & Heiles (1982).We assume a distance of 17 Mpc for the members of Virgo Cluster A, 22 Mpc forVirgo Cluster B, and 32 Mpc for objects in the M and W clouds (Gavazzi et al. 1999a).

1Computed using the relation between LH and M by Gavazzi et al. (1996)


Members of the Cancer, A1367, and Coma clusters are assumed to lie at distances of65.2, 91.3, and 96 Mpc, respectively. Isolated galaxies in the Coma supercluster areassumed at their redshift distance, adopting H0 = 75 km s−1 Mpc−1.

7.2.2 The starburst sample

In order to compare the properties of our sample with starbursts, we compile a datasetof starburst galaxies observed by IUE from the sample of Calzetti et al. (1994). Weconsider 29 galaxies, excluding AGNs and galaxies that have not been observed byIRAS at 60 or 100 µm. Complementary data such as FIR, Hα fluxes and Balmerdecrements are taken from Calzetti et al. (1995), metallicities come from Heckmanet al. (1998) and H-band photometry (available only for 18 galaxies) from (Calzetti1997). Excluding the far infrared fluxes, all these quantities are obtained withinan apertures of ∼ 20 × 10arcsec2, consistent with IUE observations Calzetti et al.(1994). Thus we stress that aperture effects could strongly affect any comparisonwith normal galaxies for which all data are homogeneously integrated values. Firstof all, if the UV emission is more extended than IUE field of view the LTIR/LFUV

ratio is overestimated2. In addition, even when physical quantities are obtained inthe same IUE apertures, the presence of age and metallicity gradients in galaxiesmakes not trivial any comparison with the integrated values obtained for normal starforming galaxies (Kewley et al. 2005). All the observables, but the ultraviolet spectraslope β, are calibrated in a consistent way with our sample of normal galaxy. Theultraviolet spectral slope of starbursts is obtained by fitting IUE spectra (Calzettiet al. 1994), while for GALEX observations it comes from the FUV-NUV color index(see next Section). However, as shown by Kong et al. (2004), these two calibrationsare consistent each other and do not introduce any systematic difference between thetwo samples.

7.3 The LTIR/LFUV − β relation for normal star-

forming galaxies

Meurer et al. (1999) have shown that the ratio of far infrared to far ultraviolet lu-minosity tightly correlates with the UV colors of starburst galaxies. This relation,known as the infrared excess-ultraviolet (IRX-UV) relation, is often presented as β vs.LTIR/LFUV relation. As discussed in the introduction, we will refer to the LTIR/LFUV

ratio as the best indicator of UV dust attenuation and we will calibrate on it all thefollowing relations. In order to determine the dust emission, we compute the total

2However Meurer et al. (1999) argued that the majority of UV flux for their starburst samplelies within the IUE aperture

7.3. The LTIR/LFUV − β relation for normal star-forming galaxies 87

infrared flux emitted in the range 1-1000 µm, following Dale et al. (2001):

log(fTIR) = log(fFIR) + 0.2738 − 0.0282 × log(f60

f100) +

+0.7281 × log(f60

f100)2 + 0.6208 × log(

f60

f100)3 +

+0.9118 × log(f60

f100)4 (7.1)

where fFIR is the far-infrared flux, defined as the flux between 42 and 122 µm (Helouet al. 1988):

fFIR = 1.26 × (2.58 × f60 + f100) × 10−14 [Wm−2] (7.2)

and f60 and f100 are the IRAS fluxes measured at 60 and 100 µm (in Jansky). Thetotal infrared luminosity is thus:

LTIR = 4πD2fTIR (7.3)

The β parameter as determined from GALEX colors is very sensitive to the galaxystar formation history (see for example Calzetti et al. 2005). For this reason weassume β as defined by Kong et al. (2004):

β =log(fFUV ) − log(fNUV )

−0.182=

= 2.201 × (FUV − NUV ) − 1.804 (7.4)

where fFUV and fNUV are the near and far ultraviolet observed fluxes respectively(in erg cm2 s−1 A−1), and FUV and NUV are the observed magnitudes.The relationship between the ratio of total infrared luminosity (LTIR) obtained from(7.1) to the far-ultraviolet fluxes and the UV spectral slope β (or the FUV-NUV color)for our sample of nearby star forming galaxies is given in Fig.7.1. Several functionalforms of the LTIR/LFUV −β relation can be found in the literature (i.e. Meurer et al.1999; Kong et al. 2004); we simply adopt a linear fit: log(LTIR/LFUV ) = a × β + b.This functional form is consistent with other previously proposed for β > −2, whileit diverges for β < −2. Since the majority of normal and starbursts galaxies haveβ > −2 our choice is justified. This represents the simplest and less parameterdependent way to study the relation between two quantities.3 We find a strongcorrelation (Spearman correlation coefficient rs ∼0.76 for the primary sample andrs ∼0.65 for the secondary sample, both corresponding to a probability P (rs) >99.9%

3We tested this hypothesis fitting our data with functional forms similar to the ones proposedby Meurer et al. (1999) and Kong et al. (2004): no significative improvement in the scatter of thisrelation is obtained.


Figure 7.1: Ratio of the total infrared to far ultraviolet luminosity as a function ofthe ultraviolet spectral slope (lower x-axis) and the FUV-NUV color (upper x-axis).Open circles indicates our secondary sample while filled circles represent the primarysample. The dashed line represents the best linear fit to starburst IRX-UV relation.The solid line indicates the best bisector linear fit for our primary sample. The starsindicate the sample of IUE starbursts. Mean error bars for the plotted data are shownin the lower right corner, in this and subsequent figures. The residuals from the bestlinear fit for normal galaxies are shown in the bottom panel.

7.3. The LTIR/LFUV − β relation for normal star-forming galaxies 89

that the two variables are correlated) between the total infrared to far ultraviolet ratioand the spectral slope, but significantly different from the one observed for starburstgalaxies (dashed line in Fig.7.1; Meurer et al. 1999). A χ2 test rejects at a confidencelevel higher than 99.9%, the hypothesis that the two samples follow the same relation.The best linear fit for our primary sample (solid line in Fig.7.1) is:

log(LTIR

LFUV) = (0.70 ± 0.06) × β + (1.30 ± 0.06) (7.5)

The uncertainty in the estimate of the LTIR/LFUV using equation (7.5) is ∼ 0.26±0.02dex for the primary sample but it increases to ∼ 0.35 ± 0.03 dex, if we considerthe whole sample (e.g. primary and secondary samples), consistent with the meanuncertainty observed for starburst galaxies (Meurer et al. 1999). A large contribution(∼ 0.21±0.02 dex) to the observed scatter in Eq.(7.5) is due to the uncertainty on theestimate of LTIR/LFUV and β. This result confirms once more that the LTIR/LFUV −βrelation for normal galaxies deviates from the one observed for starbursts, as pointedout by previous studies of nearby galaxies (i.e. Bell 2002; Kong et al. 2004; Boissieret al. 2005; Buat et al. 2005; Seibert et al. 2005; Burgarella et al. 2005, Boissier et al.in prep.) and individual HII regions in nearby galaxies (Calzetti et al. 2005).

7.3.1 The dependence on the birthrate parameter

What physical mechanisms drive the difference observed in the LTIR/LFUV − β be-tween normal star forming galaxies and starbursts? Recently Kong et al. (2004)interpreted the offset as an effect of the different star formation history experiencedby galaxies and proposed that the distance from the starburst IRX-UV can be pre-dicted using the birthrate parameter b (e.g. the ratio of the current to the mean paststar formation activity, Kennicutt et al. 1994). In order to test if the perpendiculardistance dS from the LTIR/LFUV − β relation for starbursts correlates with the starformation history of normal galaxies, we compute the birthrate parameter followingBoselli et al. (2001):

b =SFRt0(1 − R)

LH(Mtot/LH)(1 − DMcont)(7.6)

where R is the fraction of gas that stellar winds re-injected into the interstellarmedium during their lifetime (∼ 0.3, Kennicutt et al. 1994), t0 is the age of thegalaxy (that we assume ∼12 Gyr), DMcont is the dark matter contribution to theMtot/LH ratio at the optical radius (assumed to be 0.5; Boselli et al. 2001). Wecompute the H-band luminosity following Gavazzi et al. (2002a):

log LH = 11.36 − 0.4 × H + 2 × log(D) [L]


Figure 7.2: Relation between the birthrate parameter computed from the Hα emis-sion, and the distance from the LTIR/LFUV −β relation for starbursts. The solid linerepresents the best linear fit.

where D is the distance to the source (in Mpc), and the SFR from the Hα lumi-nosity (corrected for [NII] contamination and for dust extinction using the Balmerdecrement) following Boselli et al. (2001):

SFR =LHα

1.6 × 1041[M/yr] (7.7)

Fig.7.2 shows the relation between the birthrate parameter (eq.7.6) and the distancefrom the LTIR/LFUV − β relation for starburst galaxies. The two quantities arecorrelated (rs ∼0.40, corresponding to a correlation probability P (rs) ∼99.8%) butwith a large scatter. Given the value of observational uncertainties, it is not worthtrying to use the observed trend to reduce the dispersion in the LTIR/LFUV − βrelation for normal galaxies. This result confirms that part of the dispersion in theLTIR/LFUV − β relation for normal star forming galaxies appears an effect of thedifferent star formation history experienced by galaxies, as proposed by Kong et al.(2004).

7.4 A(Hα)

7.4.1 Estimate of A(Hα)

The attenuation in the Balmer lines can be deduced from the comparison of theobserved ratio LHα/LHβ with the theoretical value of 2.86 obtained for the recombi-nation case B, an electronic density ne ≤ 104 cm−3 and temperature ∼ 104 K. The

7.4. A(Hα) 91

variation of this value with density its negligible and with temperature is ≤5% (inthe range between 5000 K and 20000 K, Caplan & Deharveng 1986). The underly-ing absorption was deblended from the Hβ emission line using a multiple componentfitting procedure. To do this the emission line is measured and subtracted from thespectra. The resulting absorption line is also measured with respect to a referencecontinuum. These two measurements are used as first guess in a fitting algorithmwhich fits jointly the emission and absorption lines to the reference continuum. Forobjects whose Hβ is detected in emission but the deblending procedure is not ap-plied (no absorption feature is evident) a mean additive correction for underlyingabsorption equal to -1.8 in flux and -1.4 Ain EW is used. These values correspondto the fraction of the (broader) absorption feature that lies under the emission line.We adopt a dust screen geometry and the Milky Way extinction curve (e.g. Kenni-cutt 1983; Calzetti et al. 1994). Whereas varying the extinction curves has negligibleeffects in the visible, the dust screen assumption seems to under-estimate the extinc-tion by ∼0.2 mag compared with the amount deduced from the measurements of thethermal radio continuum (Caplan & Deharveng 1986; Bell & Kennicutt 2001). We donot apply any correction for Hα underlying absorption (Charlot & Longhetti 2001).However, since all the objects have EW (Hα + [NII]) > 3A, the underestimate inthe value of A(Hα) is negligible. In fact no change (at a 99% significance level) isobserved comparing the best fits obtained in this work and the ones obtained addingto the Hα the same fixed underlying absorption used for Hβ when the underlying isnot detected. We assume that the errors on A(Hα) are mainly due to the uncertaintyon the Hβ flux. These errors represent in fact the lower limits because we do not ac-count for the uncertainty introduced by the fitting of the lines. They range from 0.01to 0.43 mag and are found strongly anti-correlated with EW(Hβ) (see Gavazzi et al.2004). Adopting the definition of the Balmer decrement as in Gavazzi et al. (2004):

C1(Hβ) =log( 1

2.86× LHα

LHβ)

0.33(7.8)

Since the A(Hα) attenuation is:

A(Hα) = 1.0861

eβα − 1ln(

1

2.86×

LHα

LHβ

) (7.9)

From (7.8) and (7.9) we obtain:

A(Hα) = 1.0861

eβα − 1× 0.33 × C1(Hβ) ln(10) (7.10)


and assuming a galactic extinction law (eβα = 1.47) we derive:

A(Hα) = 1.756 × C1(Hβ) (7.11)

A(Hα) = 0.85 mag is obtained on average, consistent with previous studies (e.g. Ken-nicutt 1983, 1992; Thuan & Sauvage 1992; Kewley et al. 2002). Eleven galaxies haveHβ undetected in emission but the underlying stellar absorption is clearly detected.For them we derive a 3×σlower limit to the Hβ flux (fHβ) using (Gavazzi et al. 2004):

fHβ < 3 × rms(4500−4800) × Hα(HWHM) (7.12)

assuming that Hα and Hβ emission lines have similar HWHM (Half Width HalfMaximum). As shown in Eq.(7.8) a change in the theoretical value of the LHα/LHβ

ratio would only produce a small (≤5%) constant over (or under) estimate of theionized gas attenuation, thus leaving unchanged the shape and dispersions of theobserved relations, only affecting the values of the best fitting parameters.

7.4.2 The β-A(Hα) relation

Calzetti et al. (1994) found a strong relationship between the ultraviolet spectral slopeβ and the Balmer decrement Hα/Hβ. For our starburst sample these two quantitiesare correlated (rs ∼0.81) as follows (see also blue stars in Fig.7.3):

β = (0.75 ± 0.10) × A(Hα) − (1.80 ± 0.13) (7.13)

This empirical relation was used by Calzetti et al. (1994) to deduce an attenuationlaw (the Calzetti law), often applied to high redshift galaxies (i.e. Steidel et al. 1999;Glazebrook et al. 1999). Contrary to the LTIR/LFUV − β relation the Calzetti lawhas not yet been tested for a sample of normal star forming galaxies. Buat et al.(2002) showed that for normal star forming galaxies the attenuation derived fromthe Calzetti law is ∼0.6 larger than the one computed from FIR/UV ratio. Thisresult has been recently confirmed by Laird et al. (2005) on star forming galaxies atz ∼1. In order to check the Calzetti law on our sample we use the measure of theHα/Hβ described in the previous subsection. Fig. 7.3 shows the relation betweenβ and A(Hα) for our sample (empty and filled circles). For the primary sample weobtain rs ∼0.58 (P (rs) >99.9%) and:

β = (0.37 ± 0.07) × A(Hα) − (1.15 ± 0.08) (7.14)

flatter than for starburst galaxies (see Fig.7.3). At low A(Hα) normal galaxies showon average a less steep ultraviolet spectral slope than starbursts. In addition normalgalaxies with the same value of β span a range of ∼ 1 mag in A(Hα). At higher

7.4. A(Hα) 93

Figure 7.3: The relation between the ultraviolet spectral slope β and the Hα atten-uation obtained from the Balmer decrement. Symbols are as in Fig.7.1. Solid linerepresents the best linear fit to our primary sample (equation 7.14) while the dashedline indicate the best-fit for starburst galaxies obtained by Calzetti et al. (1994) (equa-tion 7.13). Arrows indicate galaxies for which the value of A(Hα) is a lower limit ofthe real value (i.e. Hβ undetected). The residuals from the best linear fit for normalgalaxies are shown in the bottom panel.


attenuation the two samples appear consistent. Our result suggest that the Calzettilaw cannot be applied to normal galaxies. On the contrary, the relation between βand A(Hα) for normal galaxies, could be used to obtain a new attenuation law.

7.5 Relations between dust attenuation and global

properties.

7.5.1 Metallicity

Heckman et al. (1998) have shown that the ultraviolet spectral slope and metallic-ity of starbursts are well correlated. To determine the metal content of our galaxieswe average five different empirical determinations based on the following line ratios:R23 ≡ ([OII]λ3727 + [OIII]λ4959, 5007)/Hβ (Zaritsky et al. 1994; McGaugh 1991),[NII]λ6583/[OII]λ3727 (Kewley & Dopita 2002), [NII]λ6583/Hα (van Zee et al. 1998)and [OIII]λ5007/[NII]λ6583 (Dutil & Roy 1999). The mean uncertainty in the abun-dances is 0.10dex. In Fig. 7.4 we study the relationship between the gas metallicitiesand the LTIR/LFUV ratio (left-panel) and β (right-panel) for normal star forming andstarburst galaxies. For normal galaxies the LTIR/LFUV ratio correlates (rs ∼0.59,P (rs) >99.9%) with the gas abundance:

log(LTIR

LFUV) = (1.37 ± 0.24) × 12 + log(O/H) − (11.36 ± 2.11) (7.15)

with a dispersion of ∼ 0.35 ± 0.03 in log(LTIR/LFUV ). As for the LTIR/LFUV − βrelation normal galaxies differ from starbursts. At comparable metallicity normalgalaxies show a lower LTIR/LFUV (lower attenuation) than starbursts, in agreementwith the recent result by Boissier et al. (2004) who studied radial extinction profilesof nearby late-type galaxies using FOCA and IRAS observations. Unexpectedlywe find however that normal star forming galaxies follow exactly the same (signif-icant, rs ∼ 0.58, P (rs) >99.9%) relationship between metallicity and ultravioletspectral slope β determined for starbursts by Heckman et al. (1998) (see right panelof Fig.7.4). This might indicate that even though a normal and a starburst galaxywith similar gas metallicity have similar UV spectral slopes, they suffer from a sig-nificantly different dust attenuation, perhaps suggesting a different dust geometry(Witt & Gordon 2000). However we stress that this effect might occur due to aper-ture effects in the IUE data: while β is not significantly contaminated by apertureeffects, the LTIR/LFUV ratio could be overestimated producing the observed trend(the total infrared luminosity is obtained by integrating the IRAS counts over the fullgalaxy extension, while the ultraviolet one is taken from IUE’s significantly smalleraperture 20×10 arcsec2). This idea could be supported by the correlation (rs ∼ 0.49,

7.5. Relations between dust attenuation and global properties. 95

Figure 7.4: Relation between gas metallicity and the LTIR/LFUV ratio (left) or β(right). Symbols are as in Fig.7.1. The solid lines show the best linear fit for ourprimary sample. The residuals from the best linear fits for normal galaxies are shownin the upper panels.

Figure 7.5: Relation between the galaxy size and the LTIR/LFUV ratio for starburst(left panel) and normal galaxies (right panel). Symbols are as in Fig. 7.1. Meanvalues and uncertainties in bins of 0.30 log(Diameter) are given.


P (rs) >99.9% see Fig.7.5) observed between the starbursts’ optical diameters and theLTIR/LFUV ratio, completely absent in our sample of normal galaxies (rs ∼ 0.006,P (rs) ∼25%). GALEX observations of starburst galaxies will rapidly solve this riddle.

Dust to Gas ratio

The correlation between attenuation and metallicity can be interpreted assuming thatthe ultraviolet radiation produced by star forming regions suffers a dust attenuationincreasing with the dust to gas ratio, which correlates with metallicity. (e.g. Issa et al.1990; Inoue 2003). In order to check this hypothesis we compute the dust to gas ratiofollowing Boselli et al. (2002b). In normal galaxies the dust mass is dominated by thecold dust emitting above ∼200 µm. The total dust mass can be estimated providedthat the 100-1000 µm far-IR flux and the cold dust temperature are known. Fittingthe SEDs of normal galaxies with a modified Planck law νβ Bν(Td), with β = 2 (Altonet al. 2000), the total dust mass can be determined from the relation (Devereux &Young 1990):

Mdust = CSλD2(ea/Tdust − 1) M (7.16)

where C depends on the grain opacity, Sλis the far-IR flux at a given wavelength (inJy), D is the distance of the galaxy (in Mpc), Tdust is the dust temperature, and adepends on λ. Only IRAS data at 60 and 100 µm are available for our sample and,given the strong contamination of the emission at 60 µm by very small grains, the 60 to100 µm ratio does not provide a reliable measure of Tdust (Contursi et al. 2001). Tdust

determined by Alton et al. (1998) consistently with Contursi et al. (2001), seems tobe independent of the UV radiation field, of the metallicity or of the total luminosity(Boselli et al. 2002b). Therefore we will adopt the average value Tdust = 20.8 ± 3.2K for all our galaxies introducing an uncertainty of ∼50% on the estimate of Mdust

(equation (7.16)). We then estimate the dust mass of the sample galaxies using (7.16)with C = 1.27 MJy−1Mpc−2, consistent with Contursi et al. (2001), and a=144 Kfor Sλ = S100 µm (Devereux & Young 1990). The determination of the dust to gasratio, in a way consistent with that obtained in the solar neighbourhood, requiresthe estimate of the gas and dust surface densities, thus of the spatial distribution ofdust and gas over the discs. Unfortunately only integrated HI and H2 masses areavailable for our spatially unresolved galaxies. It is however reasonable to assumethat the cold dust and the molecular hydrogen are as extended as the optical disc(Alton et al. 1998; Boselli et al. 2002b). To determine the mean HI surface densitywe adopt (Boselli et al. (2002b)):

log ΣHI = 20.92(±0.17) − 0.65(±0.11) × (def(HI)) cm−2

where def(HI) is the galaxy HI deficiency. Thus the dust to gas ratio is obtained from


Figure 7.6: Relation between the gas to dust ratio and the LTIR/LFUV ratio (left) orβ (right). Symbols are as in Fig. 7.1. The solid line shows the best linear fit for ourprimary sample.

the ratio of the dust surface density to the sum of molecular and neutral hydrogensurface densities. In Fig. 7.6 we compare the relation between the LTIR/LFUV ratio(left panel) and β (right panel) with the dust to gas ratio. The gas to dust ratiobarely correlates with the LTIR/LFUV ratio (R∼0.38). Contrary to metallicity, we donot find a significant correlation (R∼0.11) with the ultraviolet spectral slope. This isprobably due to the high uncertainty in our estimate of Mdust consequent to assumingthe same temperature for all our galaxies (Mdust ∝ ea/Tdust , thus small errors (∼15%)on Tdust propagate onto ∼50% errors on Mdust).

7.5.2 Luminosity

Since it is well known that the metallicity of normal galaxies strongly correlates withgalaxy luminosity (e.g. Skillman et al. 1989; Zaritsky et al. 1994) and mass (e.g.Tremonti et al. 2004), it is worth considering the correlation between attenuation andgalaxy luminosity. Fig.7.7 shows the relationships between the dust attenuation indi-cators LTIR/LFUV and β and the H-band luminosity. The infrared to far ultravioletratio correlates (rs ∼0.49, P (rs) >99.9%) with the total H-band luminosity:

log(LTIR

LFUV

) = (0.34 ± 0.10) × log(LH

L

) − (2.66 ± 0.88) (7.17)

The dispersion of this relation is ∼ 0.39±0.03 in log(LTIR/LFUV ). Since the H-bandluminosity is proportional to the dynamical mass (Gavazzi et al. 1996), this impliesa relationship between dust attenuation and dynamical mass. Also in starbursts the


Figure 7.7: Relation between the H-band luminosity and the LTIR/LFUV ratio (left)or β (right). Symbols are as in Fig. 7.1. The solid line shows the best linear fit forour primary sample. The residuals from the best linear fit for normal galaxies areshown in the upper panel.

Figure 7.8: Relation between the TIR+FUV luminosity and the LTIR/LFUV ratio(left) or β (right). Symbols are as in Fig. 7.1.


total H-band luminosity is correlated (rs ∼0.37, P (rs) ∼99.5%) with the LTIR/LFUV

ratio and the great part of starbursts appear offset (to 99% confidence level) from therelation of normal galaxies. On the contrary, no difference is observed between thetwo samples in the β-LH plot, in agreement with what observed for metallicity. Fi-nally Fig.7.8 shows the relation between the bolometric luminosity (LTIR +LFUV) andthe dust attenuation, computed assuming that the UV emission is absorbed by dustand emitted in the far infrared. The correlation coefficient (rs ∼0.31, P (rs) ∼98%)indicates that the two quantities correlate, as for starburst galaxies (Heckman et al.1998). This is not the case if we examine the relation between the ultraviolet spectralslope β and the bolometric luminosity (Fig.7.8 right panel): while there is no corre-lation (rs ∼ 0.002, P (rs) ∼20%) for our sample of normal galaxies, a clear relation(rs ∼ 0.68, P (rs) >99.9%) holds for starbursts. Starbursts with higher bolometricluminosity (high TIR emission) show lower ultraviolet slope, consistent with the ideathat high TIR emission corresponds to high attenuation (low β).

7.5.3 Surface brightness

Wang & Heckman (1996) interpreted the increase of dust attenuation with rotationalvelocity (or mass) as due to the variations in both the metallicity and surface densityof galactic disk with galactic size. Fig.7.9 shows the variation of the effective H-bandsurface brightness (defined as the mean surface brightness within the radius thatcontains half of the total galaxy light) and the dust attenuation. The two quantitiesare strongly anti-correlated (rs ∼-0.63, P (rs) >99.9%):

log(LTIR

LFUV

) = (−0.28 ± 0.04) × µe(H) + (5.92 ± 0.81) (7.18)

with a scatter of ∼ 0.34 ± 0.03 in log(LTIR/LFUV ): ∼1.2σ lower than the valueobtained for H-band luminosity and consistent with the one obtained for the gasmetallicity. Unfortunately in this case we cannot compare the behaviour of normalgalaxies with the one of starbursts due to the lack of an estimate of µe for the star-bursts. Does this relation indicate that UV dust extinction depends on the thicknessof stellar disk, or does it follows from the correlation between attenuation and starformation surface density? To attack this question we determine the SFR density (de-fined as the ratio between the SFR determined from Hα (eq.7.7) and optical galaxyarea). Fig.7.10 shows the relation between the SFR density and log(LTIR/LFUV ).

The two quantities are correlated (rs ∼0.44, P (rs) >99.9%) with a dispersion of∼ 0.39 ± 0.03 in log(LTIR/LFUV ), ∼1.2σ larger than the one observed for the meanH-band surface brightness4. Since the contribution of observational uncertainties to

4The difference between the two relations does not change if instead of the half-light radius, weuse the total radius to estimate µe(H)


Figure 7.9: Relation between the mean H-band surface brightness (µe) and theLTIR/LFUV ratio (left) or β (right). Symbols are as in Fig. 7.1. The solid lineshows the best linear fit for our primary sample. The residuals from the best linearfit for normal galaxies are shown in the upper panel.

Figure 7.10: Relation between the star formation rate density and the LTIR/LFUV

ratio (left) or β (right). Symbols are as in Fig. 7.1. The solid line shows the bestlinear fit for our primary sample.The residuals from the best linear fit for normalgalaxies are shown in the upper panel.


the scatter in the two relations is ∼ the same (0.18± 0.02), our result might suggestthat the UV attenuation is primarily correlated with the thickness of stellar disk,supporting the hypothesis of Wang & Heckman (1996) that both gas metallicity andstar surface density are directly connected with the physical properties of dust (i.e.quantity and spatial distribution).

7.5.4 LHα/LFUV ratio

Buat et al. (2002) suggested that the LHα/LFUV ratio could be another potentialattenuation indicator but they found a scattered correlation between LHα/LFUV andA(FUV ), confirmed by Bell (2002). This correlation is expected since both Hα andUV emission are star formation indicators. The Hα luminosity comes from starsmore massive than 10 M and it traces the SFR in the last ≤ 107 yr while the UVluminosity comes from stars of lower mass (M≥ 1.5 M) and it can be used as anindicator of the SFR in the last ≈ 108 yr. This means that under the conditionthat the star formation is approximately constant in the last ≈ 108 yr the ratioLHα/LFUV (corrected for attenuation) should be fixed. Thus the ratio between theextinction corrected LHα and the observed LFUV should be a potential attenuationindicator. In Fig.7.11 we analyze the relationship between the dust attenuation andthe LHα/LFUV ratio, where LHα is the Hα luminosity corrected for dust attenuationusing the Balmer decrement and for the contamination of [NII]. The two quantitiesturn out to be strongly correlated (rs ∼0.76, P (rs) >99.9%):

log(LTIR

LFUV) = (0.84 ± 0.07) × log(

LHα

LFUV) − (0.59 ± 0.12) (7.19)

The dispersion around this relation is ∼ 0.24 ± 0.02 in log(LTIR/LFUV ), consistentwith the one observed for the log(LTIR/LFUV ) − β relation. The high correlationand low scatter between the two quantities is expected since the two variables aremutually related: the FUV luminosity appears in both axes and LTIR and LHα areknown to be correlated (Kewley et al. 2002), explaining why in the left panel ofFig.7.11 starbursts and normal galaxies show the same trend. The right-panel ofFig.7.11 shows the relation between the ultraviolet slope and the LHα/LFUV ratio. Inthis case starbursts and normal galaxies behave differently: at any given β starburstshave an higher LHα/LFUV than normal galaxies, consistent with what expected forgalaxies experiencing a burst of star formation (Iglesias-Paramo et al. 2004).

A secure determination of the Balmer decrement for large samples is still a hardtask, especially at high redshift, thus we look for a relation similar to Eq.(7.19) usingthe observed Hα luminosity (Lobs

Hα). The LobsHα/LFUV and log(LTIR/LFUV ) ratios are

yet correlated (see Fig.7.12) but the correlation coefficient is lower than the previous


Figure 7.11: Relation between the Hα and far ultraviolet luminosity and theLTIR/LFUV ratio (left) or β (right). Symbols are as in Fig. 7.1. Hα luminosityis corrected for dust attenuation using the Balmer decrement, while the FUV flux isuncorrected. The solid lines show the best linear fit for our primary sample. Theresiduals from the best linear fit for normal galaxies are shown in the upper panels.

Figure 7.12: Relation between the observed Hα and far ultraviolet luminosity andthe LTIR/LFUV ratio (left) or β (right). Symbols are as in Fig. 7.1. Hα luminosity isthe observed value not corrected for dust attenuation. The solid lines show the bestlinear fit for our primary sample.The residuals from the best linear fit for normalgalaxies are shown in the upper panels.

7.6. A cookbook for determining LTIR/LFUV ratio 103

case (rs ∼0.49, P (rs) >99.9%). The best linear fit gives:

log(LTIR

LFUV

) = (1.10 ± 0.17) × log(Lobs

Hα

LFUV

) − (0.59 ± 0.21) (7.20)

with a mean absolute deviation of ∼ 0.34 ± 0.03 (∼3.3σ higher than for Eq.7.20).

7.6 A cookbook for determining LTIR/LFUV ratio

In this chapter we investigated the relations between dust attenuation, traced by theLTIR/LFUV ratio, and other global properties of normal star forming galaxies. Fur-thermore we compared the dust attenuation in normal and starbursts galaxies usingmultiwavelength datasets. The amount of dust attenuation is found to correlate withthe UV colors, gas metallicity, mass and mean surface density but, generally speaking,differently for normal and starburst galaxies. Determine whether this difference is realor is due to aperture effects requires the analysis of GALEX observations for a sampleof starburst galaxies. The dispersion in the LTIR/LFUV − β relation correlates withthe birthrate parameter b, suggesting that the observed scatter is, at least partly, dueto differences in the star formation history. These results stress that estimating theUV dust attenuation, and consequently the star formation rate of normal galaxies (athigh redshift in particular) is highly uncertain (≥50%) when rest-frame far infraredobservations are not available. Moreover the sample selection criteria could stronglyaffect its properties, as recently pointed out by Buat et al. (2005) and Burgarellaet al. (2005). They studied the dust attenuation properties and star formation activ-ity in a UV and in a FIR selected sample, showing that the former shows correlationswith global galaxy properties, such as mass and bolometric luminosity, that the FIRselected sample does not. Their results stress that the dust attenuation propertiesare very heterogeneous and that LTIR/LFUV cannot be derived in a robust mannerwhen far infrared observations are not available.However the present investigation has shown that among optically-selected samplesof normal galaxies with no nuclear activity a number of empirical relations exists,allowing to derive the LTIR/LFUV ratio (and its uncertainty). Once the attenuationat UV is determined it can be transformed to any other λ, only knowing the shape ofthe attenuation law and dust geometry (i.e. Calzetti et al. 1994; Gavazzi et al. 2002a;Boselli et al. 2003a).In Table 7.1 we list all the relations, their associated r.m.s., mean absolute deviationfrom the best fit (m.a.d.)5 and the Spearman correlation coefficient.

5The mean absolute deviation is less sensitive to the contribution of outliers than the standarddeviation. For a Gaussian distribution the mean absolute deviation (m.a.d.) is ∼

√

2/π × (r.m.s.),while it is lower (higher) for a heavier (lighter) tailed distribution. As shown in Table 1 the values


Before we proceed describing our recipes, we have to investigate whether the scatterin these relations is physical or is only driven by observational uncertainties. In thelatter case, in fact, our cookbook would not be very useful, since it would be valid onlyfor observations with the same uncertainties as our datasets. For H-band luminosity,H-band surface brightness, Lobs

Hα/LFUV ratio and metallicity the contribution of obser-vational uncertainties to the observed scatter varies from ∼ 18% (r.m.s.∼ 0.17±0.02)for LH to ∼40% (r.m.s.∼ 0.21 ± 0.02) for 12 + log(O/H) and Lobs

Hα/LFUV : even ac-counting for the contribution of measurements errors, the relative difference in thescatter of these relations does not change. On the contrary this confirms that therelation involving LH is the one with the highest ”physical” dispersion, while for theother three relations the scatter is similar.The situation is worse for the relations involving β and the LHα/LFUV ratio: thecontribution of observational errors is ∼70-76% (∼ 0.21 ± 0.02). Thus it is impossi-ble to determine which of these two relations has the lowest scatter and representsthe best way to estimate dust attenuation without far infrared observations. We canconclude that observational errors could account for the difference scatter observed inthe relations involving β and the LHα/LFUV ratio, but not for the difference observedin all the other relations. Our results can thus be used to suggest different ways tocorrect for UV dust attenuation.Ia) The LTIR/LFUV − β relation still represents one of the best way to quantify dustattenuation. The uncertainty in the value of log(LTIR/LFUV ) is ∼ 0.26 ± 0.03. Ib)If the UV spectral slope β is unknown but we know LHα (corrected for attenuation)we can obtain the ultraviolet attenuation using equation (7.19), with a r.m.s. of0.24 ± 0.02. This relation is valid under the assumption that the star formation rateis approximately constant in the last ≈ 108 yr.IIa) If we know Lobs

Hα, but no estimate of A(Hα) is available, we can use Eq.(7.20)(rms∼ 0.34 ± 0.03).IIb) If neither β nor Hα luminosity are available we are left with the relations withH-band surface brightness6 (r.m.s.∼ 0.34 ± 0.03) and, in the worse case,III) H-band luminosity (rms∼ 0.39 ± 0.03 ).Summarizing, these relations allow us to estimate the value of the LTIR/LFUV ratiowith an average uncertainties of∼0.32 dex. This value corresponds approximately toσ(A(FUV )) ∼0.5 mag, assuming log(LTIR/LFUV ) = 1 (the mean value for our sam-ple) and using the model of Buat et al. (2005). This is the lowest uncertainty on theestimate of the LTIR/LFUV ratio in absence of far infrared observations. However wecaution the reader that this value holds only for an optically-selected sample and thatsamples selected according to different criteria, especially FIR-selected, could containhigher dispersions. The cookbook presented in this chapter is obviously insufficient

obtained for r.m.s. and m.a.d. are consistent with the ones expected for a Gaussian distribution6Since we need Hα flux to estimate metallicity, Eq.(7.15) cannot be used in this case.

7.6. A cookbook for determining LTIR/LFUV ratio 105

x a b m.a.d.a rmsb rs

β 0.70 ± 0.06 1.30 ± 0.06 0.20 ± 0.02 0.26 ± 0.02 0.7612+log(O/H) 1.37 ± 0.24 −11.36 ± 2.11 0.26 ± 0.02 0.35 ± 0.03 0.59LH/L 0.34 ± 0.10 −2.66 ± 0.88 0.29 ± 0.03 0.39 ± 0.03 0.49µe(H) −0.28 ± 0.04 5.92 ± 0.81 0.25 ± 0.02 0.34 ± 0.03 −0.63LHα/LFUV 0.84 ± 0.07 −0.59 ± 0.12 0.19 ± 0.02 0.24 ± 0.02 0.76Lobs

Hα/LFUV 1.10 ± 0.17 −0.59 ± 0.21 0.27 ± 0.02 0.34 ± 0.03 0.49

a: Mean absolute deviation from the best fit.b: Standard deviation from the best fit.

Table 7.1: Linear realtions useful to estimate the LTIR/LFUV ratio(log(LTIR/LFUV ) = a × x + b).

to understand dust attenuation and know how to correct UV observations of localand high redshifts galaxies, but it represents only the tip of the iceberg. The nextsteps should be the folowings: a) compare all the relations obtained in this work withdifferent models in order to try to determine the physical properties of dust b) usemodels and data in order to estimate a new attenuation law from the far-ultravioletto the near-infrared valid for normal star forming galaxies, as the one obtained forstarbursts by Calzetti et al. (1994). Only knowing the dust attenuation law we will beable to correct for dust extinction all our observations and thus to correctly estimatethe star formation rate in galaxies.

Chapter 8

High velocity interaction:NGC4438 in the Virgo cluster

This analysis represents the tip of the iceberg and only a future comparison withdifferent dust models will allow us to understand dust attenuation and to know howto correct UV observations of local and high redshifts galaxies. A statistical analysisof star formation activity in cluster galaxies using UV data is therefore still impossible.For this reason, in the last three chapter of this work, I will focalize my attentionon the study of three particular cluster galaxies considered as the prototypes of thethree main environmental effects observed in clusters: tidal interaction, ram pressurestripping and preprocessing, respectively. These unique astrophysical laboratorieswill be used to deeply understand the effects of different physical mechanisms ongalaxy evolution..

8.1 Introduction

NGC 4438 (Arp 120) is the clearest example of an ongoing tidal interaction in anearby cluster of galaxies. Apparently located close to the Virgo cluster center (∼300 kpc from M87), NGC 4438 is a bulge-dominated late-type spiral showing longtidal tails (30 kpc) thought to be induced by a recent dynamical interaction withthe nearby SB0 galaxy NGC 4435. Multifrequency observations covering the elec-tromagnetic spectrum from X-rays (Kotanyi et al. 1983; Machacek et al. 2004) toradio continuum (Hummel & Saikia 1991), including both spectro-photometric andkinematical (Kenney et al. 1995; Chemin et al. 2005) data, have been carried out inthe past to study the nature of this peculiar system. These observations have shownthat the violent interaction between the two galaxies perturbed the atomic (Cayatteet al. 1990) and molecular (Combes et al. 1988) gas distribution, causing both gasinfall toward the center which might have induced nuclear activity (Kenney et al.

107

108 8. High velocity interaction: NGC4438 in the Virgo cluster

1995; Kenney & Yale 2002; Machacek et al. 2004), and gas removal in the externalparts displacing part of the gas in the ridge in between the two galaxies (Combeset al. 1988). Both multifrequency observational data (Kenney et al. 1995; Machaceket al. 2004) and model predictions (Combes et al. 1988; Vollmer et al. 2005) favor arecent (∼ 100 Myr) high-velocity, off-center collision between NGC 4435 and NGC4438.Except for mild nuclear activity, it is still unclear whether the dynamical interac-tion between the two galaxies induced extra-nuclear star formation events: the lowHα/[NII] ratio and the similar X-ray and Hα morphology of NGC 4438 indicate thatthe Hα emission is in this case not due to the ionizing radiation but is probably dueto gas cooling phenomena (Machacek et al. 2004).The UV emission is dominated by young stars of intermediate masses (2 < M < 5M)and provides us with an alternative star formation tracer. As part of the NearbyGalaxy Survey (NGS), we have observed the central 12 deg2 of the Virgo clusterusing the Galaxy Evolution Explorer (GALEX). A distance of 17 Mpc for Virgo isadopted.

8.2 Data

The GALEX data used in this work include far-ultraviolet (FUV; λeff = 1530A, ∆λ =400A) and near-ultraviolet (NUV; λeff = 2310A, ∆λ = 1000A) images. The dataconsist of 2 independent GALEX pointings centered at R.A.(J2000)= 12h29m01.2s,Dec(J2000)= +1310’29.6” (819 sec) and R.A.(J2000)= 12h25m25.2s, Dec(J2000)=1310’29.6” (1511 sec), for a total of 2330 sec of integration time.To study the star formation history of NGC 4438, the UV data have been combinedwith visible and near-IR images of the galaxy taken from the GOLDmine database(Gavazzi et al. 2003a), from the SDSS Data release 3 (Abazajian et al. 2005) fromthe 2MASS survey (Jarrett et al. 2003) and from the CFHT and SUBARU archives.These are Hα+[NII] (Boselli & Gavazzi 2002), B (Boselli et al. 2003a), K’ (Boselliet al. 1997b), u, g, r, i, z SDSS, R CFHT and SUBARU and H 2MASS images. Forthe main body of the galaxy (region 4 in Fig. 8.1, see next sect.) we added theintegrated spectrum (3500-7000 A ; Gavazzi et al. 2004). The current calibrationerrors of the NUV and FUV magnitudes are on the order of ∼ 10% (Morrissey et al.2005), comparable to that at other frequencies.

8.2. Data 109

Figure 8.1: The combined NUV and FUV image of NGC 4438. The regions describedin sect. 3 of the text are labeled 1 to 7. The horizontal line is 10 kpc long (assuminga distance of 17 Mpc).


8.3 The UV emission and the star formation his-

tory of NGC 4438

Figure 8.1 shows the UV image of NGC 4438, obtained by combining together theNUV and FUV frames in order to increase the S/N. The UV emission of the galaxyis mostly due to compact, bright regions in the central part of the galaxy (marked asregion 4 in Fig. 8.1), in the northern tidal tail (region 2) and in the section of thesouthern tail closest to the main body of the galaxy (region 5). The UV emission ismostly diffuse in the extended western part of the galaxy (region 3) and at the edgeof the southern tidal tail (region 6). In addition Figure 8.1 shows the presence of twoextended and patchy emission to the north-west of the galaxy (∼ 15-25 kpc from thenucleus, marked as region 1 and region 7). These features, previously undetected inother visible and/or near-IR bands, are similar to a tidal tail: region 1 is ∼ 20 kpclong and ∼ 2 kpc wide, while region 7 is considerably smaller (∼ 2 kpc). The RGBimage of the galaxy obtained by combining the FUV, NUV and B frames (see Fig.8.4)shows the color of the different regions: while the edge of both the northern and thesouthern tidal tails (region 3 and 6) are red (thus dominated by relatively old stars),regions 2 and 5 as well as the newly discovered regions 1 and 7, have blue colors andseem therefore to be dominated by a younger population.The Hα+[NII] emission map, given in Fig. 8.2 as a contour plot superposed on theNUV image of NGC 4438, shows a lack of massive, ionizing young O-B stars (Kenni-cutt 1998). The Hα+[NII] emission observed in region 5 has a different morphologythan the UV one; on the contrary its distribution is the same observed in X-ray asstated by Machacek et al. (2004) (see Fig.8.3). This evidence confirms the conclu-sions of Machacek et al. (2004) that the Hα+[NII] emission is not due to the ionizingradiation but is probably associated with the cooling gas.What is the nature of the newly discovered extragalactic UV emitting regions? Theaverage NUV surface brightness of these features is ∼ 28.5 ABmag arcsec−2, whilethey are undetected both in the SUBARU R band (360 sec) image down to a sur-face brightness limit of ∼ 27.8 mag arcsec−2 and in Hα down to a surface brightnesslimit of ∼ 5 10−17 erg s−1 cm−2 arcsec−2 (see Fig.8.2), implying a log(NUV/Hα) ≥0.3. Extra-planar diffuse regions with an excess of UV over Hα flux ratio (i.e.log(NUV/Hα) ≥ 1, as that observed at 11 kpc from the disk of M82, are often in-terpreted as due to the UV radiation produced by the central starburst and locallyscattered by diffuse dust (Hoopes et al. 2005). It is unlikely that scattered light isresponsible for the UV emission since the steep slope of the UV spectrum (β=-2.32and -2.05, as defined by Kong et al. 2004) is typical of a recent unreddened star-burst (Calzetti 2001) and is unexpected in a scattering scenario since the dust albedois greater in the NUV than in the FUV (Draine 2003) (i.e.β ≤ −1). Furthermorethe lack of a powerful central starburst (as in M82) and the large distance of these

8.3. The UV emission and the star formation history of NGC 4438 111

Figure 8.2: The Hα+[NII] contours (red, in arbitrary scale, in between 8 10−17 and 610−16 erg cm−2 s−2 arcsec−2, with σ= 5 10−17 erg cm−2 s−2 arcsec−2, from Boselli &Gavazzi (2002)) are superposed to the NUV gray-level image of NGC 4438.

Figure 8.3: Chandra image of NGC 4438 in the 0.3-2 keV energy band with Hαcontours superposed. Adapted from Machacek et al. (2004)


relatively patchy regions from the nucleus seem to exclude the scattering scenario.These data suggest that regions 1 and 7 are post starbursts, induced by the violentinteraction with NGC 4435. In addition the absence of Hα emission associated withall the UV emitting regions suggests that the starburst lasted for a relatively shorttime, since it is not producing young, massive O-B stars any more. This is probablybecause the atomic and molecular gases, needed to feed star formation, have beenremoved during the interaction (Combes et al. 1988; Vollmer et al. 2005)1 .In order to date the starburst and reconstruct the star formation history of the galaxy,we have determined the spectral energy distribution (SED) of each region (see Fig.8.4)and then fitted it with a simple model of galaxy evolution. To this end we make theassumption that dust attenuating the SED is present only in region 4, where we cor-rect the UV to near-IR data using the far-IR to UV flux ratio as done in Boselli et al.(2003a) and described in Appendix A. This restricted application is reasonable sinceno dust emission has been observed in the tidal tails with ISOCAM (Boselli et al.2003b); furthermore, in regions 1 and 7, dust is unexpected since it has not yet beenproduced by the young stellar population, as confirmed by the steep β parameter(see also Chapter 7). Assuming that NGC 4438 was a normal late-type object beforeinteracting with NGC 4435, we use the models of Boissier & Prantzos (2000) in orderto reconstruct its SED before the interaction. The two parameters of the model (spinλ and rotational velocity VC) are constrained by the observed total H-band luminos-ity and velocity rotation of NGC4438, leading to λ=0.01 and VC=290 km s −1. InFig.8.4 we compare the model with the SED of the main body of the galaxy (region4), composed by an old population with no significant contribution from the recentstarburst. Both the total SED and the optical spectrum produced by the model arein good agreement, confirming that the adopted technique is able to reproduce thegalaxy SED before the interaction. We then assume that the evolved stellar popula-tion of each region, if present, is the one given by the model and removed from themain body of the galaxy by the tidal interaction, while the younger population isproduced by the induced starburst. For each region, we thus combine the SED of anevolved stellar population with the one produced by an instantaneous burst of starformation obtained using Starburst 99 (Leitherer et al. 1999) for a solar metallicityand a Salpeter IMF between 1 and 100 M. For each age and intensity of the burst,we determine the best combination of evolved population+ burst by fitting the FUVto K band SED and rejecting solutions in disagreement (i.e. too bright) with theupper limits. We then adopt the age corresponding to the lowest reduced χ2 2. The

1The upper limit of the HI surface density for these regions is ∼ 1 M pc−2 (Cayatte et al. 1990)2All ages with χ2 < 1 are acceptable solutions. Given the small number of photometric points

available for regions 1 and 7 (2 GALEX bands), the fitted solution for a combination of a burst andan old population (two parameters) can be almost perfect (resulting in very low χ2, ≤ 10−2), as longas the obtained fit is in agreement with the limits at other wavelengths. Whenever the fit producesa SED not satisfying a detection limit, this solution is rejected.


results of our fitting procedure are presented in Fig.8.4. For each region (excludingregion 4) two panels are given. The lower panels show the observed SED of eachregion (crosses, or arrows if are upper limits) and the best SED obtained from thefitting procedure (black line). The relative contribution of the evolved and youngstellar populations to the observed SEDs are indicated in red and blue respectively.The burst luminosity contribution (for the age corresponding to the minimum χ2) inthe band FUV, B and K is also given. In the upper panels the variation of the reducedχ2 parameter (black continuum line) and of the burst mass fraction (red dotted line)as a function of the age of the burst are given. This exercise gives an interestingresult: the strong UV emission of regions 1 and 7 is due to a coeval starburst ∼ 6-20Myr old. The age and the duration of the starburst are strongly constrained both bythe lack of Hα emission and by the blue UV slope of the spectrum (lower limit to theage) and by the lack of an old stellar population (upper limit to the duration). Theburst age for the other region cannot be determined with the same precision, but wecan only put a lower limit to their age. Regions 2 and 5 are consistent with an olderstarburst (≥ 100 Myr, as suggested by their redder UV slope: β=-0.33 and -0.67 inregions 2 and 5 respectively) which probably ended ∼ 10 Myr ago as indicated bythe lack of any Hα emission. Conversely the stellar population in regions 3 and 6appear not significantly affected by the recent burst. Moreover it is interesting tonote that, while the fraction of stars produced by this burst is dominant in regions 1and 7, the sum of the stars produced by the burst in all regions (including the innerpart) contributes to the total galaxy stellar mass by less than 0.1 %, an extremelylow value for such a violent interaction.


These observations have major consequences in constraining the evolution of clustergalaxies. A high-velocity off-center collision between two galaxies of relatively similarmass, whose violence is able to perturb the stellar distribution producing importanttidal tails, is insufficient to induce a significant instantaneous starburst. This resultmight be representative only of the nearby Universe where encounters of gas-richgalaxies are probably rare since clusters are dominated by gas-poor early-type galax-ies such as the companion galaxy NGC 4435. It is conceivable, however, that at higherredshifts, where clusters are forming, stellar masses produced by a starburst inducedby interactions predicted by the models of Moore et al. (1996) (galaxy harassment)might be more important given the higher fraction of gas-rich galaxies.The other interesting result is the long time differential between the age of the in-teraction (∼ 100 Myr as determined by dynamical simulations, (Combes et al. 1988;Vollmer et al. 2005) and the beginning of the starburst (∼ 10 Myr in regions 1 and7, ∼ 100 Myr in regions 2 and 5). This result is totally consistent with the models of


Mihos et al. (1991) that predict for close-by encounters an enhancement of the starformation activity in the inner disk during some 100 Myr, stopping once the gas reser-voir is exhausted as in NGC 4438. In the tidal tails, on the contrary, star formationis expected to increase after ∼ 100 Myr, the time needed by the gas to re-collapse,but then ceasing after a few Myr because the expansion of the tidal tail brings thegas surface density to subcritical values (no HI and CO has been detected in theseregions). If these systems are dynamically stable and survive the interaction, theymight be at the origin of some dwarf galaxies in the cluster similar to those observedin the Stephan’s Quintet by Mendes de Oliveira et al. (2004) or in other interactingsystems (Neff et al. 2005; Hibbard et al. 2005; Saviane et al. 2004) Being produced bya single starburst, these gas poor systems might evolve into dwarf ellipticals, typicalof rich clusters. Otherwise they will simply increase the fraction of unbound stars,contributing to the Virgo intracluster light (Willman et al. 2004).


Figure 8.4: The RGB (FUV=blue, NUV=green, B=red) color map of NGC 4438 andNGC 4435. The SED of each region defined in Fig. 1 are given in the lower plot ofeach frame. Crosses indicate the observed data, arrows upper limits (in mJy), thered dashed line the evolved population fit as determined by the model of Boissier &Prantzos (2000), the dotted blue line the starburst SED (from Starburst 99) and thedashed green line the combined fitting model. The burst luminosity contribution (forthe age corresponding to the minimum χ2) in the band FUV, B and K is also given.The upper panel gives the variation of the reduced χ2 parameter (black continuumline, in logarithmic scale) and of the burst mass fraction (red dotted line) as a functionof the age of the burst (in Myr). The lower panel of region 4 gives the integrated3500 to 7000 A, R=1000 spectrum of the main body of the galaxy (black continuumline) compared to the fitted model (red dashed line).

Chapter 9

Ram Pressure stripping: NGC4569in the Virgo cluster

9.1 Introduction

Spiral disks can lose their atomic gas content during dynamical interactions with thehot and dense intergalactic medium (IGM) (Gunn & Gott 1972) and/or in directinteractions with nearby objects (Merritt 1983). These interactions can quench theirstar formation activity (Gavazzi et al. 2002c) leaving the objects to become anemic(van den Bergh 1976). To explain the well known morphological segregation effectsDressler (1980) it has been suggested that these quiescent spirals could evolve intolenticulars; however, observations and model predictions give still contradictory re-sults (see Boselli & Gavazzi 2005 for a review).Despite the on-going physical processes (tidal interactions were probably dominantat early epochs, while galaxies-IGM interactions are more important at present), itis clear that the fuel supply needed to feed star formation is more efficiently removedwhere the host-galaxy potential well is weakest, i.e., in the outer disk. Given thestrong relation between the gas surface density and the star formation activity inspiral disks, commonly known as the Schmidt law (Kennicutt 1998; Boissier et al.2003), it is expected that star formation will be quenched in the outer (lower density)portions of the disk. While interferometric observations of galaxies in Virgo haveclearly shown that HI disks are less extended in those objects located close to thecluster center (Cayatte et al. 1990), the observational evidence for a truncation of thestar forming disks has been proven by Hα imaging (Koopmann & Kenney 2004b,a).Although a truncation of the disk profile has been predicted (Larson et al. 1980b), westill do not know what the passive evolution of a stellar disk is once its gas is removed.In particular, it is unclear whether the progressive radial suppression of star formationis able to reproduce the structural properties of lenticulars, generally characterized as

117

118 9. Ram Pressure stripping: NGC4569 in the Virgo cluster

having higher surface brightness of their stellar disks and higher bulge-to-disk ratiosthan spirals (Dressler 1980).We have been collecting multi-frequency data for a large sample of late-type galax-ies in nearby clusters and in the field in order to undertake comparative statisticalanalyses of any systematic differences between cluster and field objects. Combinedwith multi-zone models for the chemical and spectrophotometric evolution of galax-ies (Boissier & Prantzos 2000), this unique database is helping us understanding theevolution of cluster spirals. As a first step during my thesis I studied the radial pro-files of the Virgo cluster galaxy NGC 4569 (M90). NGC 4569, the prototype anemicgalaxy as defined by van den Bergh (1976), is extremely deficient in HI, having just∼ one tenth of the atomic gas of a comparable field galaxy of similar type and di-mensions. The galaxy has a truncated Hα and HI radial profile (at a radius of ∼ 5kpc; see Fig.9.3) as firstly noticed by Cayatte et al. (1994) and Koopmann & Kenney(2004a), witnessing a recent interaction with the cluster environment. NGC 4569 islocated close (∼ 1 degree) to the cluster center. Being one of the largest galaxies (∼10 arcmin) in the Virgo cluster, NGC 4569 is the ideal candidate for our study sinceit can be spatially resolved at almost all wavelengths considered here. The combi-nation of the multi-frequency 2-D data with our spectrophotometric models allow usto study, for the first time, the radial evolution of the different stellar populations inthis prototype, gas-stripped cluster galaxy with the aim of understanding whether itsstructural properties can evolve into those of a typical cluster lenticular (S0) galaxy.

9.2 Data and models

The large amount of spectrophotometric data available for NGC 4569, collected inthe GOLDMine database (Gavazzi et al. 2003a), allow us to reconstruct its radialprofile at different wavelengths: from the new GALEX UV bands (at FUV=1530 andNUV=2310 A), to the visible B and V (Boselli et al. 2003a), Sloan u, g, r, i, z (Abaza-jian et al. 2005) and near-IR J, H and K bands (Boselli et al. 1997b; 2MASS Jarrettet al. 2003). Hα+[NII] narrow band imaging, used to trace the recent star formationactivity, is available from Boselli & Gavazzi (2002). HI profiles are from Cayatte et al.(1994), while H2 profiles, determined from CO data using a luminosity dependent COto H2 conversion factor (from Boselli et al. 2002b) are taken from the BIMA surveyof Helfer et al. (2003) for the inner disk, and from Kenney & Young (1988) for theouter disk. The accuracy of the photometric imaging data is, on average ∼ 10 %.The galaxy rotation curve has been taken from Rubin et al. (1999). Unfortunatelyno metallicity gradient information is available for NGC 4569.The radial profiles have been constructed by integrating the available images withinelliptical, concentric annuli. The ellipticity and position angles have been determinedand then fixed using the deepest B band image following the procedure of Gavazzi

9.2. Data and models 119

et al. (2000) (see Fig.9.3). To avoid any possible contamination by the NW arm,whose kinematical properties indicate that it is not probably associated with the stel-lar disk but rather formed during the interaction with the ICM (Chung et al. 2005),the arm was masked in the construction of the radial profiles. If included, its con-tribution would be perceptible only in the FUV filter at radii > 8 kpc, increasingthe surface brightness by < 0.5 mag. The UV to near-IR broadband images of thegalaxy have been corrected for internal extinction using the recipe of Boissier et al.(2004), assuming a typical UV extinction gradient for a galaxy of the luminosity andscalelength of NGC 4569. In this case, in fact, the far-IR (IRAS) to UV flux ratio can-not be used to estimate the extinction because both fluxes are contaminated by thenuclear activity of the galaxy. The extinction in the other visible and near-IR bandshas been determined using the prescription of Boselli et al. (2003a) and described inAppendix A. Hα+[NII] narrow-band imaging has been corrected for [NII] contami-nation and dust extinction (Balmer decrement) using the integrated spectroscopy ofGavazzi et al. (2004). To study the evolution of the disk of NGC 4569 at variousradii, we have used the multi-zone chemo-spectrophotometric models of Boissier &Prantzos (2000), updated with an empirically-determined star formation law Boissieret al. (2003). These models have a resolution of ∼1 kpc, significantly lower than theone of our multiwavelength datasets (0.08-0.4 kpc). The errors in the surface bright-ness and color profiles have been computed following Gil de Paz & Madore (2005).For this reason we degraded all our images at the model resolution, and we extractthe smoothed profiles used for the comparison between models and data. The nuclearemission due to the central AGN has been masked since the model is not able toreproduce the AGN activity (see Fig.9.4). The two model parameters (spin λ androtational velocity VC) are constrained by the H-band luminosity profile (determinedassuming a distance of 17 Mpc) and the rotation curve of the galaxy, making thereasonable assumption that both of these observables are unperturbed during the in-teraction. This gives λ=0.04 and VC = 270 km s−1 (see Fig.9.1). To compute the Hαprofile, the number of ionizing photons predicted by Version 5 of STARBURST 99Vazquez & Leitherer (2005) for a single generation of stars distributed on the Kroupaet al. (1993) initial mass function (as used in our models) is convolved with our starformation history, and converted into a Hα flux as described in Appendix B.In addition to this model (valid for an unperturbed galaxy) we add an episode of rampressure gas stripping. For simplicity, we adopt the plausible scenario of Vollmer et al.(2001): the galaxy has crossed the dense IGM only once, on an elliptical orbit. Theram pressure exerted by the IGM on the galaxy ISM varies with time (t) following aLorentzian profile (see Fig.9.2):

ε = ε0(∆t)2

((∆t)2 + (t − t0)2)(9.1)


Figure 9.1: The radial profile of observed (open symbols) and extinction-corrected(filled symbols) H-band surface brightness (left) and of the rotational velocity (center)used to constrain the model without interaction (represented by the black solid line).The total gas radial profile (right) predicted by the unperturbed model (solid blackline) is compared to the observed one (green filled circles), obtained by summing theHI component (red line) to the molecular one (blue and light blue) and correctingfor Helium contribution (× 1.4), and to the model including the interaction (blackdashed line).

where t0 is when the galaxy crosses the dense cluster core at high velocity and ε0 isthe value of ram pressure at t0. Following Vollmer et al. 2001 we assume a widthprofile ∆t = 9 × 107 years. In order to determine the amount of stripped gas wemake the hypothesis that the gas is removed at a rate that is directly proportionalto the galaxy gas column density Σgas and inversely proportional to the potential ofthe galaxy, measured by the total (baryonic) local density Σpotential (provided by the

model). The gas-loss rate adopted is then finally equal to ε Σgas

Σpotential. The two free

parameters in our model are then t0 and ε0. We make the further assumption thatno extra star formation is induced during the interaction.

9.3 The star formation history of NGC 4569: model

predictions

Once the width of the interaction event, ∆t, is fixed, it is possible to choose simul-taneously t0 and ε0 because the amount of gas left and its radial distribution dependstrongly on ε0 while the resulting stellar light profiles depend mainly on t0 (see Fig.9.3for some examples). If the cluster core crossing time is recent only the youngest stel-lar populations (emitting in Hα, whose age is ≤ 4 106 yrs, or far-UV, ≤ 108 yrs)have had time to feel the progressive radial suppression of the star formation activity.


Figure 9.2: Ram pressure stripping intensity (in arbitrary units) as a function of time(Eq.9.1). Adapted from Vollmer et al. (2001).

Comparing model predictions with the spectrophotometric radial profiles of clustergalaxies can thus be used to date the dynamical interaction with the IGM. We thusfitted the data with models for different values of t0 and ε0. An important modi-fication applied to the usual χ2 test is that its value was artificially put to 100 forany model predicting surface brightnesses in disagreement with observational limits(non detections at relatively large radii) in order to reject these solutions. The modelbest matching the properties of NGC 4569 (Fig.9.4) is characterized by ε0 = 1.2 M

kpc−2 yr−1 and t0 = 100 Myr. This is largely consistent with the dynamical modelsof Vollmer et al. (2004a), who obtained t0 ∼ 300 Myr. Although not reproducingperfectly the surface brightness profile, this model is able to qualitatively reproducethe truncation of the total gas disk profile (see Fig.9.1) and of the Hα and UV radialprofiles (Fig.9.3) as well as the milder truncation observed at longer wavelengths.

It is interesting to note that although older cluster core crossing epochs give moretruncated disk profiles in the old stellar populations (B and i bands, blue dashedline), this is not the case in the gas profile which is modified by contributions fromthe recycled gas.


The present work gives the first quantitative estimate of the structural evolution ofstellar disks in cluster galaxies due to gas removal caused by a dynamical interaction


Figure 9.3: The radial profile of the observed (empty green circles) and extinction-corrected (filled green circles) total gas, Hα, FUV (1530 A), NUV (2310 A), B and isurface brightness. The yellow shaded area marks the range in between the observed(bottom side) and extinction-corrected (top side) surface brightness profiles. Surfacebrightnesses are compared to the model predictions without interaction (black solidline) or with interaction for several ε0 and t0 parameters. Equal maximum efficiency(ε0=1.2 M kpc−2 yr−1) and different age: t0=100 Myr, red continuum line (theadopted model); t0=500 Myr, grey long dashed line, t0=1.5 Gyr, dashed magentaline. Equal age (t0=100 Myr) and different maximum efficiency: ε0=3 M kpc−2

yr−1, blue dotted line; ε0=1/3 M kpc−2 yr−1, orange dotted line.


of the galaxy with the IGM. Although the model only qualitatively reproduces theobserved multi-wavelength radial profiles (the mismatch being attributed to resolu-tion effects) it delivers a strong message concerning the passive stellar evolution ofstripped disks. First of all it is clear that the truncation of the total gas disk profileis soon reflected in the young population stellar disk, confirming the predictions ofLarson et al. (1980b). As observed in NGC 4569, this gas-stripped galaxy has a colorgradient opposite to that of normal, isolated spirals, which generally have bluer colorsin their outer disks (see Fig.9.5 and Fig. 9.4). NGC 4569 is bluest towards the center.The trend is especially true for colors tracing the relatively young populations (<∼108 yr); colors tracing populations older than the interaction event present the usualgradient (i.e., redder towards the center). The inversion of the color gradient, hereobserved for the first time in a cluster galaxy, is well reproduced by our model.The consequence of these findings in the interpretation of the evolution of clusterspiral galaxies is significant. One of the most intriguing and still open question re-garding the effects of the environment on the evolution of galaxies is that of the originof lenticulars, and their overabundance in the centers of rich clusters. Are lenticularsan independent population of galaxies formed in the primordial high-density environ-ments, or were they spiral disks whose star formation activity has been quenched oncetheir gas reservoir was removed by the unfavorable cluster environment? Althoughthe second interpretation seems logical, simple statistical considerations in the semi-nal work of Dressler (1980) show that this idea is not supported by the observations:spirals have lower surface brightnesses and bulge-to-disk ratios than lenticulars, andspirals are rotationally supported while lenticulars are dynamically hotter systems.Structural (and kinematical) modifications must thus be invoked if spirals are to betransformed into lenticulars.The present work has shown for the first time how a galaxy-cluster IGM interactionis able to remove most of the gas reservoir inducing important structural modifica-tions in the disk properties. We have in fact shown that, because of the differentialradial stellar evolution of spiral disks, we expect that cluster spirals have (at leastat short wavelengths) more truncated disk profiles, inverting the outer color gradientwith respect to similar but unperturbed objects. The surface brightness of the disk,however, mildly decreases in Hα and in the UV bands while remains mostly constantat longer wavelengths even 5 Gyr after the interaction (Fig. 9.4d). The differentialevolution of the stellar disk due to gas stripping alone is thus not able to reproducethe structural properties of present-day lenticulars. Gravitational perturbations, suchas tidal interactions with other galaxies (Merritt 1983), interactions with the clusterpotential well (Byrd & Valtonen 1990) or a mixture of both (called ‘galaxy harass-ment’ by Moore et al. 1996) must be invoked to reproduce the observed properties ofnearby lenticulars.This new study and analysis is consistent with the idea that the present evolution oflate-type galaxies in clusters differs from that at earlier epochs, where late-type galax-


ies were mostly perturbed by dynamical interactions (pre-processing and/or galaxyharassment; Dressler 2004, Moore et al. 1996) which were able to thicken the stel-lar disks thereby producing the present-day cluster lenticulars. We hope to confirmthis original result in the near future once multi-frequency data come available for astatistical significant sample of late-type cluster galaxies.


(d)

(a) (b)

(c)

Figure 9.4: The observed and model surface brightness (a), color (b) radial profiles ofNGC 4569. In the model profiles the continuum lines are for models with gas removal,dashed lines for unperturbed models. c) the reduced χ2 as a function of the look-backtime of the ram-pressure event for a few efficiencies ε0 (M kpc−2 yr−1). Models werecomputed each 100 Myr for 0 < t0 < 500 Myr, 200 Myr for 1.5 < t0 < 0.5 Gyr and1 Gyr for 6.5 < t0 < 1.5 Gyr; and each 0.2 M kpc−2 yr−1 efficiencies between 0.4and 1.6 (only the more relevant are shown here). d) the variation of the effectivesurface brightness (mean surface brightness within Re, the radius containing half ofthe total light) and radius due to differential variation of the star formation historyof NGC 4569. Open triangles are for the unperturbed model, the other symbols fordifferent ages of the interaction (100 Myr, 1.5 and 5.5 Gyr).


Figure 9.5: The RGB (continuum subtracted Hα =blue, FUV=green, NUV=red)color map of NGC 4569

Chapter 10

Galaxy Pre-processing: the bluegroup infalling in Abell1367

10.1 Introduction

In the previous two chapters we have investigated the effects of the environment onthe properties of galaxies inhabiting the core of the Virgo cluster. However galaxiesinteract with the harsh environment well before having reached the center of a cluster.In particular, if we believe that structures grow hierarchically, galaxy clusters formnot by accreting individual galaxies randomly from the field, but rather through theinfall of less massive groups falling in, along large scale filaments. Galaxy groupsmay therefore represent a natural site for a preprocessing stage in the evolution ofcluster galaxies. These infalling groups have velocity dispersions that are significantlysmaller than that of cluster, permitting the slow gravitational interaction typicallyobserved in field galaxies. Moreover even in compact groups ram pressure seems tobe able to displace the gas from the disk of galaxies (Fujita 2004; Roediger & Hensler2005). This means that probably at least part of the morphological and star for-mation properties of cluster galaxies derives from earlier epochs and very differentconditions than the ones observed in today clusters (Dressler 2004). Environmentalinteractions in the infalling groups may thus represent a preprocessing step in theevolution of cluster galaxies (Mihos 2004a). Unfortunately, witnessing preprocessingin local Universe is a real challenge since we live in a Λ-dominated Universe wherethe infall rate is significantly lower than in the past (Gottlober et al. 2001). Today,we observe a plethora of clusters experiencing multiple merging (Gavazzi et al. 1999a;Donnelly et al. 2001; Cortese et al. 2004), but the structures involved are subclusterswith a mass ∼ 5 × 1014 M, considerably higher than the typical mass of a compactgroup ∼ 1013 M (Mulchaey 2000), as the North and South subclusters in Abell1367studied in Chapter 5 (see Table 5.7). However Abell1367 represents a unique excep-

127

128 10. Galaxy Pre-processing: the blue group infalling in Abell1367

tion among local, dynamically young, clusters since in addition to massive evolvedsubstructures it is also experiencing the merging of a compact group infalling directlyinto the cluster core. This group has a velocity dispersion of only ∼ 170km s−1, andit is infalling into the cluster core at a very high speed (∼ 1700km s−1). The rarity ofthis phenomenon could probably explain the unique properties observed in this group.In fact it was independently discovered by Iglesias-Paramo et al. (2002) and Sakaiet al. (2002) during two deep Hα surveys of nearby clusters, representing the regionwith the highest density of star forming systems ever observed in the local Universe.Sakai et al. (2002) argued that this group lies in the cluster background, having nointeraction with the cluster environment. On the contrary the dynamical analysispresented in Chapter 5, is consistent with an infalling scenario, as also proposed byGavazzi et al. (2003b). Moreover this picture is supported by X-ray observations: Sun& Murray (2002) (using Chandra observations) discovered extended gas features anda ridge near the SE cluster center. They proposed that these features are associatedwith a new merging component penetrating the SE subcluster. XMM clearly detectsa cold front near the center of the SE subcluster, probably associated with a groupinfalling into the cluster core (A. Finoguenov, private comm.). All these observationalevidences suggest that we are witnessing, for the first time in the local Universe, acompact group infalling into a core of a dynamically young cluster. It thus representsa unique laboratory to study with the great detail possible only in the local Universe,a physical process typically expected in clusters at high redshift. The study of thisgroup could therefore help us shading light on the possible influence that preprocess-ing might have and have had on the past evolution of galaxies now populating highdensity environments.During the last few years we thus collected a great amount of multiwavelength spec-troscopic and imaging observations in order to try to reconstruct the history of thisrare group of galaxies, which represents the only compact group infalling into thecenter of a galaxy cluster ever observed in the nearby Universe. Throughout thischapter I will refer to this group as the Blue Infalling Group (BIG), as defined byGavazzi et al. (2003b)

10.2 Observations

10.2.1 HI observations

Using the refurbished 305-m Arecibo Gregorian radio telescope we observed the BIGregion in March 2005. We obtained observations for 4 different positions covering thegroup center and its NW outskirt (see Fig.10.1). Data were taken with the L-BandWide receiver, using nine-level sampling with two of the 2048 lag subcorrelators set toeach polarization channel. All observations were taken using the position-switching

10.2. Observations 129

Figure 10.1: The four Arecibo HI pointings obtained in the region of the BIG group,superposed to the r′ band image. The size of each circle correspond to the telescopebeam.


Figure 10.2: GALEX NUV image of the Blue Infalling group (BIG).

technique, with each blank sky (or OFF) position observed for the same duration, andover the same portion of the telescope dish as the on-source (ON) observation. Each5min+5min ON+OFF pair was followed by a 10s ON+OFF observation of a well-calibrated noise diode. The velocity resolution was 2.6 km s−1, the instrument’s beamat 21 cm is 3′.5×3′.1 and the pointing accuracy is about 15′′. Flux density calibrationcorrections are good to within 10% (and often much better), see the discussion of theerrors given in O’Neil (2004).Using standard IDL data reduction software available at Arecibo, corrections wereapplied for the variations in the gain and system temperature with zenith angleand azimuth. A baseline of order one to three was fitted to the data, excludingthose velocity ranges with HI line emission or radio frequency interference (RFI).The velocities were corrected to the heliocentric system, using the optical convention,and the polarizations were averaged. All data were boxcar smoothed to a velocityresolution of 12.9 km s−1 for further analysis.


10.2.2 UV to near-IR imaging

The Blue Infalling Group has been observed by GALEX in April 2004, within the twopointings of the Abell cluster 1367. The observations are centered at R.A.(J2000)=11:43:41.34 Dec(J.2000)=+20:11:24.0 (e.g. offset to the north of the cluster to avoid astar bright enough to threaten the detector), with a mean exposure time of 1460s, asdescribed in Chapter 4. Fig.10.2 shows the GALEX NUV image of the Blue InfallingGroup. UBVRH photometry for CGCG (Zwicky et al. 1961) galaxies is taken fromGavazzi et al. (2003a).

10.2.3 Hα imaging

We observed BIG using the Device Optimized for the LOw RESolution (DOLORES)attached at the Nasmyth B focus of the 3.6m TNG in the photometric nights of17th May and 18th June, 2004. The observations were taken through a [SII] narrowband filter centered at ∼ 6724A and a width of ∼ 57A covering the redshifted Hαand [NII] lines. The underlying continuum was taken through a broadband (Gunn)r′ filter. Images, split in 6 exposures of 1200 sec in the narrow band filter and 5exposures of 300 sec in the r′ broadband filter, for a total of 2 hours and 30 minutesexposure respectively, were taken with a seeing of ∼ 1.2 arcsec. The photometriccalibration was achieved by exposing the spectrophotometric star Feige 34. Afterbias subtraction and flat-fielding, the images were combined. The intensity in thecombined OFF-band frame was normalized to that of the combined ON-band one bythe flux ratio of several field star. The NET image was obtained by subtracting thenormalized OFF-band frame to the ON-band one. The resulting OFF and NET-bandframes are shown in Figs. 10.6 and 10.7 respectively. Hα+[NII] fluxes and EWs areobtained as described in Boselli et al. (2002a).

10.2.4 MOS spectroscopy

We observed the BIG region in MOS mode with the ESO/3.6m and with the TNGtelescope. The ESO/3.6m observations were taken in the photometric nights of May5th and 6th 2003 with the ESO Faint Object Spectrograph and Camera (EFOSC). Weused the MOS mode of EFOSC to obtain the spectra of 9 of the emitting line knots.The EFOSC spectrograph was used with a 300 gr/mm grating and 2048×2048 thinnedLoral CCD detector, which provided coverage of the spectral region 3860 − 8070A.Slits width of 1.75” yielded a resolution of ∼ 19A. We obtained eleven exposure of1530 sec, for a total exposure time of ∼ 4.65 hours.The TNG observations were taken in the photometric nights of 26th March and 22ndApril 2004 with DOLORES. We used the MOS mode of DOLORES to obtain thespectra of 8 of the emitting line knots, of the nuclear region of CGCG97-125 and of 14


Figure 10.3: High-contrast Hα+[NII] band frame of the BIG group.


Name R.A. Dec Velocity(J2000) (J2000) (km s−1)

TNG ESO − MOS Sakai02 Gavazzi03

K1 114444.18 194816.0 8422 ± 153 8265 ± 117 − 8098DW3 d 114445.97 194744.4 − 8564 ± 151 − −DW3 e 114445.97 194741.1 − 8072 ± 124 − −DW3 a 114446.43 194741.2 8490 ± 180 − 8266 −97-114b 114446.56 194640.3 − 8656 ± 132 8504 838397-114a 114447.41 194649.8 − 8763 ± 124 − 8425K2 a 114450.61 194605.1 − 8080 ± 140 8070 8089K2 b 114449.71 194604.7 8309 ± 165 − − −DW2 c 114451.12 194718.7 8380 ± 188 − − −DW2 b 114451.17 194717.5 − 8221 ± 146 − 8077DW2 a 114451.67 194713.5 8253 ± 292 − − −K5 114451.76 194752.7 − 8241 ± 112 − 7995DW1 b 114453.78 194731.5 − − 8070 −DW1 c 114454.29 194728.6 8343 ± 223 − − −DW1 a 114454.64 194732.9 − 8265 ± 136 8161 806797-125b 114454.89 194611.3 8261 ± 191 8396 ± 132 8170 −K3 114455.28 194803.3 8020 ± 212 − − −97-125a 114455.99 194628.0 − − − 8330

Table 10.1: Redshifts of the galaxies in the BIG group.

galaxies in the region. The DOLORES spectrograph was used with a grating whichprovided coverage of the spectral region 3200 − 8000A. Slits width of 1.6” yielded aresolution of ∼ 17A. We obtained six exposure of 1800 sec, for a total exposure timeof 3 hours. All the emitting line regions observed in MOS spectroscopy are shown inFig.10.3In addition we took spectra of the bright galaxy CGCG97-114 using the Loiano/1.52m telescope. The BFOSC spectrograph attached at the Loiano telescope was usedwith a 300 gr/mm grating and 1300 × 1340 thinned EEV CCD detector, which pro-vided a spectral coverage 3600− 8900A. A slit width of 2.00” yielded a resolution of∼ 20A. The observations were taken in the ”drift-scan” mode, with the slit parallelto the galaxy major axis, drifting over the optical surface of the galaxy. The totalexposure time was 2400 sec.The reduction of the spectra was carried out using standard tasks in the IRAF pack-age. Bias subtraction and flat-field normalization was applied using median of severalbias frames and flat-field exposures. The various exposures were combined using a


median filter, thus removing the cosmic rays. The λ calibration was carried out usingIDENTIFY −REIDENTIFY −FITCOOR on exposures of He/Ar lamps for eachslit, and the calibration was transferred to the science frames using TRANSFORM .Typical errors on the dispersion solution are of ∼ 0.5− 1A, as confirmed by the mea-surements of the sky lines. However, since the resolution of our spectra is ≥ 13A weassume an rms of 3A on our wavelengths calibration. The two-dimensional frameswere sky subtracted using BACKGROUND. One-dimensional spectra were ob-tained integrating the signal along the slit using APSUM . The apertures were lim-ited to regions where the signal intensity was above 1 σ of the sky noise.Spectra were flux-calibrated using the spectrophotometric standard star: ltt 3864 forthe ESO, Feige 67 for the TNG and Feige 34 for the Loiano observations.The redshift of each knot was derived as the mean of the individual redshift obtainedfrom each emission line. Our results are shown in Tab. 10.1 and compared with theprevious measurements by Sakai et al. (2002) and Gavazzi et al. (2003b).

Line measurements

All spectra were shifted to the rest frame wavelength and normalized to their intensityin the interval 5400-5600 A. The flux-calibrated, normalized spectra are presented inFig. 10.18. Under visual inspection of the spectra we carried out the measurementof the emission lines using SPLOT . This provided a list of fluxes and EWs withrespect to a user defined continuum level. Hα (λ6563) is bracketed by the weaker[NII] doublet ([NII1] λ6548 and [NII2] λ6584). The three lines are not well resolved,thus using the task SPLOT we performed a two Gaussian fit to the blended emissionsproviding an estimate of the line ratio [NII]λ6584/(Hα+[NII]λ6548). The two brightgalaxies CGCG97-125 and CGCG97-114 show evidence for underlying absorption incorrespondence to emission lines. We de-blended the underlying absorption from theemission lines as discussed in Chapter 7.In order to compare our observations with the ones presented by Sakai et al. (2002)we re-measured, using the method described above, the spectra taken at the StewartObservatory 2.3m Bok telescope and at the 6.5m MMT by these authors. The twosets of measurements presented in Tab. 10.2 are found in fair agreement.

10.2.5 High Resolution spectroscopy

We obtained high dispersion long-slit spectra of CGCG97-125 and CGCG97-120 withthe 1.93 mtelescope of the Observatoire de Haute Provence (OHP), equipped with theCARELEC spectrograph coupled with a 2048×512 TK CCD, giving a spatial scale of0.54 arcsec per pixel. The observations were carried out in the night of April 20, 2004in approximately 2 arcsec seeing conditions through a slit of 5 arcmin×2 arcsec. Theselected grism gives a spectral resolution of 33 A/mm or 0.45 A/pix and a spectral

10.3. Results 135

Object Tel. C1 [OII] Hβ [OIII] [OIII2] Hα [NII2]

K1 ESO (0.00) 3.68 1.00 0.94 2.53 − −K1 TNG 0.00 3.49 1.00 0.95 2.57 2.86 0.58DW3 a TNG 0.02 8.26 1.00 0.31 1.06 2.86 0.59DW3 d ESO 0.00 2.01 1.00 0.60 1.34 2.86 0.6997-114b ESO 0.33 4.25 1.00 0.57 1.83 2.86 0.2697-114a ESO 0.24 3.37 1.00 0.26 0.70 2.86 0.4597-114 LOI 0.75 2.61 1.00 0.21 0.36 2.86 0.64K2 a ESO 0.17 4.06 1.00 0.77 2.09 2.86 0.50K2 b TNG 0.23 8.43 1.00 0.64 0.82 2.86 0.32DW2 a TNG 0.16 5.45 1.00 0.73 0.99 2.86 0.60DW2 b ESO > 0.1 > 5.18 1.00 < 1.00 < 0.99 2.86 < 0.61K5 ESO 0.56 − 1.00 0.47 0.65 2.86 0.29DW1 b MMT 0.33 3.62 1.00 0.74 2.50 2.86 0.35DW1 c TNG 0.00 2.76 1.00 0.47 1.53 2.86 0.66DW1 a ESO 0.30 − 1.00 0.82 2.39 2.86 0.35DW1 a MMT 0.20 3.75 1.00 0.80 2.41 2.86 0.4997-125b ESO 0.55 − 1.00 0.51 1.25 2.86 0.3797-125b TNG 0.04 3.60 1.00 0.35 1.28 2.86 0.6097-125 OHP 0.88 9.23 1.00 1.06 1.97 2.86 0.8497-125 TNG 0.90 6.58 1.00 1.15 1.89 2.86 1.21K3 TNG 0.00 2.87 1.00 0.25 0.97 2.86 0.89

Table 10.2: Line fluxes, corrected for internal extinction, of the galaxies in the BIGgroup.

coverage in the region 6080-6990 A containing the redshifted Hα ( λ 6562.8 A), the[NII] doublet (λ 6548.1, 6583.4 A) and the [SII] doublet (λ 6717.0, 6731.3 A).

10.3 Results

10.3.1 Kinematics

Table 10.1 lists the positions and radial velocities of the objects that were measuredspectroscopically. Our observations confirm the physical association of all the emit-ting line objects with the bright galaxies CGCG97-114 and CGCG97-125. On thecontrary the brightest galaxy in this region CGCG97-120 seems not associated withthis group, having a recessional velocity of 5635 km s−1 (see also Section 10.3.7). Thevelocity of galaxies in BIG (< V >= 8230 km s−1 and σV = 170 km s−1) exceedssignificantly the mean cluster velocity of < V >= 6484 km s−1 (σV = 891 km s−1),


Figure 10.4: Upper panel: The position and the width (rectangular areas on theright) of the three slits obtained for CGCG97-125. The slits are superposed to theHα + [NII] net image. Lower Panel: The three different rotations curves obtained forCGCG97-125. Letters indicate the different regions as labeled in the upper panel.

10.3. Results 137

Figure 10.5: The low resolution 2D spectrum obtained at ESO/3.6 for the knotsDW3d (left) and DW3e (right), shows a significant difference (∼ 500km s−1) in thevelocity of the two knots.

suggesting that it is infalling at ∼ 1700 km s−1 into the cluster core.The high resolution spectra obtained at the OHP telescope give us more insights onthe dynamical state of CGCG97-125. Velocity plots of CGCG97-125 were extractedfrom each spectrum by measuring the wavelength of the Hα line in each pixel alongthe slits. The three rotation curves so obtained are given in Fig.10.4. In each diagramthe recessional velocity is plotted as a function of position along the slit (the spatialaxis runs from E (left) to W (right)). All the three spectra show regions with mul-tiple velocity components, especially in correspondence to the galaxy center wheretwo sudden velocity jumps of ∼100-150km s−1 are clearly present. It is interestingto note that the velocity of these jumps decrease from ∼ 8400 km s−1 to ∼ 8150km s−1 and their position moves to east, passing from the north to the south partof the galaxy. Even if several examples of kinematic disturbances has been observedin normal galaxies (Rubin et al. 1999; Haynes et al. 2000) and interacting systems(Jore et al. 1996; Duc & Mirabel 1998), the features observed in CGCG97-125 areextremely rare. To our knowledge, the only other galaxy with the same character-istics is UGC6697 (Gavazzi et al. 2001b), the merging systems in the NW part ofAbell1367 (see Chapter 5). The velocity jumps observed in the rotation curve ofCGCG97-125 are consistent with the idea that this galaxy has experienced a mergingin the past; however its properties are unusual if compared with what expected froma similar phenomenon. During the accretion of a satellite, the gas falling into thegalaxy center is expected to relax before the gas at the outskirts of the galaxy. Therelaxation time is in fact ∝ R/V, where R is the radial distance from the center andV is the rotational velocity. On the contrary in this case, the major anomalies areobserved near the galaxy center while in the outer part the rotation curve presents a


Name r′ Hα flux EW (Hα + [NII]) SFRa

mag erg cm−2 s−1 A M yr−1

97125 13.99 (1.33 ± 0.29) × 10−13 27 ± 3 1.4997114 15.06 (6.59 ± 0.71) × 10−14 34 ± 5 0.74DW1 17.93 (1.60 ± 0.17) × 10−14 128 ± 15 0.09DW2 18.96 (3.67 ± 0.86) × 10−15 25 ± 7 0.02DW3 19.11 (4.47 ± 0.96) × 10−15 56 ± 15 0.02

a: obtained using equation 7.7. L(Hα) corrected for [NII] contribution and extinction using values obtained from spectroscopy (see

Table 10.2).

Table 10.3: Properties of galaxies in BIG.

typical S shape. Detailed dynamical simulations of a minor merger experienced by anS0 galaxy are thus mandatory to try to understand the particular features observedin this galaxy.The MOS spectroscopy collected at the ESO/3.6m and at the TNG telescopes givesus some information regarding the internal dynamic of DW3. The emitting line knotscomposing this system have considerably different recessional velocities, ranging be-tween ∼ 8000 km s−1 and ∼ 8600 km s−1. The western (DW3-b) and the eastern(DW3-a) knots have a recessional velocity of ∼ 8250−8300 km s−1 significantly lowerthan the one observed in the northern knot DW3-d (∼ 8564 km s−1) and ∼ 250 km s−1

higher than the redshift of the southern knot DW3-e. This great difference is clearlyvisible in Fig.10.5 where the two emitting line knots DW3-d and DW3-e are observedwithin the same slit (thus the relative offset is not affected by any uncertainty inthe wavelength calibration). The observed high velocity gradient (∼ 500 km s−1)suggests that these five knots are probably not gravitationally bound, and thus thatDW3 does not represents a dwarf virialized system.

10.3.2 Hα properties

When observed in optical broad band images this group does not show any unex-pected feature if compared with other group of galaxies. The r′-band luminosityfunction of BIG in the interval -19.2 < Mr′ < -12.2 has a slope α ∼ -1, consistentwith the r′-band luminosity function of Hickson compact groups (Hunsberger et al.1998), suggesting that originally BIG was a normal compact group. On the opposite,BIG represents a true exception as far as its Hα properties. At least ten out of 12star-forming regions are associated with dwarf systems (or extragalactic HII regions)with E.W.(Hα + [NII]) often exceeding 100 A(Gavazzi et al. 2003b). These galaxies,in spite of being ∼1000 times smaller than typical giant galaxies, are currently form-ing stars at a 10 times higher rate (per unit mass) than normal galaxies of similar

10.3. Results 139

Figure 10.6: Stellar shells are seen around galaxy 97-125 in the r′ band image of BIG.No continuum emission is detected from the low brightness trails (except K2).


Figure 10.7: Extended low brightness trails appear in the Hα+[NII] NET frame ofBIG.

10.3. Results 141

luminosity, as derived from their L(Hα) (see Table 10.3). As remarked by Sakai et al.(2002), it is the first time that such a high density of star-forming galaxies has beenseen in a nearby cluster, in spite of having collected data over an area of A1367,Coma, and the Virgo Cluster approximately 500 times larger than the group size.Moreover new Hα images of BIG obtained last year reinforce the uniqueness of thisgroup revealing a spectacular Hα filamentary structure on top of which the star form-ing knots observed by Sakai et al. (2002) and Gavazzi et al. (2003b) represent the tipof the iceberg. Multiple loops of ionized gas appear with a projected length exceeding150 kpc, a typical transverse size of 5 kpc, among the most extended low-brightnessHα emission features ever detected (see Fig. 10.7). One stream (labeled NW in Fig.10.7) extends from the northern edge of the frame to the dwarf galaxy DW3, withan extension of ∼ 100kpc. The second and brightest one (labeled W in Fig. 10.7)traces a loop around galaxy 97-120 and seems connected to the bridge (labeled K2in Fig.10.7) between 97-114 and 97-125, that was known from previous studies. Ifthis is the case, the total projected extension of the NW and W trails would result∼ 150 kpc. In addition to the filamentary features, at least two other diffuse Hαregions (labeled S and E in Fig.10.7) are detected. The total diffuse (Hα + [NII])emission (e.g. excluding the contribution of the three bright galaxies and of the tendwarfs/HII regions previously discovered) results ∼ 1.2×10−13 erg cm2 s−1 i.e. similarto the flux collected from one of the bright galaxies, and the typical surface brightnessis 10−17.6 − 10−18.3 erg cm−2 s−1 arcsec−2. Along the filaments we detect typically anE.W.(Hα + [NII]) ≥ 100 − 150 A.The loop around 97-120 alone contributes with ∼ 2.4 × 10−14 erg cm2 s−1, as ob-tained integrating the Hα + [NII] emission in a circular corona of 10 kpc radius andan annulus of 5 kpc, centered on 97-120. The derived line intensity is 2.05 Rayleigh(1 Rayleigh = 106/4π photons cm−2 s−1 sr−1), corresponding to an emission measure(EM) of 5.7 cm−6 pc. Assuming a torus geometry with a circular section of radius∼ 5 kpc and a filling factor of 1, the plasma density results ne∼ 3.3 × 10−2cm−3 andthe ionized column density Ne ∼ 5×1020 cm−2 (the inferred densities would be higherif the gas is in clumps or filaments, which is likely). The emission measure in the NWtrails results lower (∼ 1.3 cm−6 pc) than in the loop around 97-120 and the plasmadensity is ne ∼ 1.1 cm−3.The trails geometry is strongly suggestive of a rosetta orbit typical of tidal disruptionof a satellite galaxy. However contrary to other known examples of tidal streams thefeatures here observed show strong Hα emission and no continuum emission aboveΣr′ = 26.8 mag arcsec−2 (even though this limit is insufficient to rule out the pres-ence of stellar streams of brightness as low as observed for example in the M31 stream(Ibata et al. 2001)). The case offered by BIG seems therefore unique as it combinesthe, eventually present, faint stellar brightness of tidal streams with strong line emis-sions of tidal tails. What mechanism have produced a such unusual feature? In orderto test the tidal disruption scenario we use the formalism of Johnston et al. (2001),


assuming: 1) that at least one of the gas trails is from a dwarf intruder merged into97-125 and 2) that the geometry of the undetected stellar streams is the same of theobserved gaseous trails. The intrinsic geometry of a streamer from a totally disruptedsatellite can be used to estimate the mass m and age t of a young streamer:

m ∼ 1011(w

R

)3(

Rp

10 kpc

) (

vcirc

200 km/s

)2

M , (10.1)

and the time since its disruption

t ∼ 0.01 Ψ

(

R

w

) (

Rcirc

10 kpc

) (

200 km/s

vcirc

)

Gyr , (10.2)

where w is the width of the streamer at radius R, Ψ is its angular length, Rp is thepericentric distance of the orbit and Rcirc is the radius of the circular orbit with thesame energy as the true orbit. Of course, we cannot measure Rcirc directly, but wecan approximate it as being halfway between the adopted apocenter and pericenter.Thus adopting a projected ratio of the loop width w, to the radius R, of ∼ 0.15, apericentric distance of Rp ∼ 15 kpc, an orbit with the same energy of a circular orbitof radius Rcirc ∼ 30 kpc and a rotation velocity Vcirc ∼ 298km s−1 (Vogt et al. 2004),we obtain a satellite mass ∼ 1 × 109M and an age of the interaction ∼ 1.3 Gyr.The mean surface brightness of the tidal debris is then obtained using the followingequation:

µr′(t) = −2.5 log

[

f

(

10M/L,ν

Υ

) (

1 Gyr

t

)]

−2.5

3log

[(

vcirc

200 km/s

)(

m

108M

)(

10 kpc

R

)]

+ 23.9 + M,r′ ,(10.3)

where M,r′ is the r′ absolute magnitude of the Sun, Υ is the r′ mass to light ratioof the satellite and f is the mass fraction loses by the satellite. The mean surfacebrightness of the tidal debris in BIG is

µr′ = 26.8 − 2.5 log

[(

3

Υ

) (

f

1

)]

(10.4)

consistent with the undetection of continuum emission above Σr′ = 26.8 mag arcsec−2.However I stress the reader that this simulation is based on the interaction betweentwo field galaxies and not infalling into the cluster center as in this case. The mostdramatic difference between field mergers and those in a cluster is in the evolutionof tidal debris. In the field, most of the material stripped into tidal tails remainsloosely bound to the host galaxy, forming a clear tracer of the gravitational interac-

10.3. Results 143

tion. In the cluster encounter, the cluster tidal field quickly strips the material fromthe galaxy, dispersing it throughout the cluster and making these tidal tracers veryshort-lived (Mihos 2004a). Thus we can assume the obtained value of µr as a lowerlimit for the surface brightness in the stellar trails.Although the undetection of stellar emission in the trails does not help us ruling out atidal stream nature for these trails, their strong Hα emission makes BIG a unique ex-ample among interacting systems and compact groups. Conversely other known tidaltails have E.W.(Hα) ranging from zero (i.e. the Stephan’s quintet) to ∼ 20 A (i.e.the Mice (NGC 4676) and some Hickson compact groups). Moreover tidal streamsdiscovered in interacting systems (e.g. Shang et al. 1998; Forbes et al. 2003) are de-tected only in continuum with no Hα emission, even if associated to strong starburstmerging systems (Wehner & Gallagher 2005). For these reasons, the unique featuresobserved in BIG make us suppose that not only tidal interaction can produce the Hαtrails but that probably the mutual influence of tidal and non-gravitational forces(e.g. ram pressure) can explain the physical properties of this group. In order toexplain the properties of these trails we need a mechanism able to strip gas fromgalaxies with little or no influence on the stellar component: a condition respectedonly by galaxy interaction with the hot intracluster medium. This scenario is alsosupported by the discovery in the NW part of Abell1367 of two low surface brightnessHα cometary tails, with a total length of 75 kpc, associated with two star formingsystems: CGCG 97-073 and 97-079 (Gavazzi et al. 2001a). In fact, the morphologyand properties of the tails (which typical size and gas densities are similar to the trailsobserved in BIG) suggests that galaxies in the NW group are experiencing ram pres-sure due to their high velocity motion through the IGM. The only difference betweenthese two cases is that CGCG97-079 and CGCG97-073 are infalling into the clusteras isolated systems, while galaxies in BIG are infalling within a compact group wheregravitational interactions are not negligible.

10.3.3 HI properties

HI observations give us additional hints on the properties of this unique group.CGCG97-125, the brightest member of BIG, has a normal hydrogen content: M(HI)= 3.9×109M (Sakai et al. 2002), implying an HI deficiency1= -0.21. Its HI columndensity distribution appears asymmetric, with the highest signal in the western sideof the galaxy, as seen in the HI map obtained by Sakai et al. (2002) and reproducedin Fig.10.8, suggesting that this galaxy is strongly perturbed by an external agent.On the contrary CGCG97-114 has an HI mass of only M(HI) = 3.0 × 108M (Sakaiet al. 2002) with a resulting HI deficiency of 0.7. This low content is surprising if

1The HI deficiency is defined as the difference, in logarithmic units, between the observed HI massand the value expected from an isolated galaxy with the same morphological type T and opticallinear diameter D: HI DEF = < log MHI(T

obs, Dobsopt) > −logMobs

HI (Haynes & Giovanelli 1984)


Figure 10.8: HI column density distribution in BIG. Contours are 0.5, 1.0, 2.0, 3.0,4.0, 5.0, and 6.020cm−2. Adapted from Sakai et al. (2002)

Figure 10.9: HI position-velocity diagram centered on CGCG 97-125. Adapted fromSakai et al. (2002)

10.3. Results 145

compared with the high star formation activity (0.74 Myr−1, see Table 10.3). Atthis current SFR, the total HI mass of CGCG 97-114 would be depleted in 2.5 × 108

yr or in 9.4× 108 if we add the total molecular gas mass (4× 108) detected by Boselliet al. (1997a) in this galaxy. This suggests that the galaxy is currently experiencingan intense, transient burst of star formation. In addition to the two detected CGCGgalaxies, in Fig.10.8 there appears to be extended HI, mostly around CGCG 97-125.The HI extension appears to be a continuation of the Hα structure to the west ofCGCG 97-125. This extended structure is typical of galactic merger remnants (Hib-bard & van Gorkom 1996) and suggests that a recent merger has affected this galaxy.The HI distribution around CGCG 97-125 is extended not only in the plane of the skybut in the velocity dimension. The position-velocity diagram presented by Sakai et al.(2002) centered on CGCG 97-125 is shown in Figure 10.9. The velocity distributionshows a regular gradient across the galaxy (the optical major axis of CGCG 97-125is very close to east-west) ranging from 8090 up to 8490 kms−1, corresponding to arotation speed of 298 km s−1 when corrected for inclination (Vogt et al. 2004). Thisvalue is exceptionally high for a galaxy of the same luminosity, which usually has arotation speed of ∼200 km s−1. Thus, both the HI distribution and the HI kinematicsyet available suggest that CGCG 97-125 is a quite peculiar object.Addition information concerning the HI properties of this group can be obtained fromArecibo observations. In Fig.10.10 are shown the four spectra obtained for the dif-ferent pointings of BIG. Unfortunately three of the four pointings (97-125, 97-120,97-114) are surely not independent due to the large overlap in the observed fields.The only, if any, independent observation is represented by the NW field, since it is faraway from all bright galaxies and has a relatively small overlap with the field centeredon 97-120. Since the side-lobes are located ∼5 arcmin from the field center, we canexclude a strong contamination of the NW pointing from the bright galaxies in BIG. Inaddition to the strong HI emission in the velocity range 8000−8500km s−1, associatedto the star forming galaxies, a new component in the velocity range 7500−8000km s−1

is clearly present in all the four spectra. This emission is not associated with any ofthe Hα emitting regions since no one of the star forming objects has a recessionalvelocity below ∼ 8000km s−1 (see Table 10.1). This fact strongly emerges in Fig.10.11where we compared this spectrum with the mean spectrum obtained from the fourpointings. A great fraction of the NW Hα trail described in the previous sectionlies exactly in the region observed by the BIG-NW pointing, suggesting that the HIemission is probably associated with this feature. This could mean not only thatthere is neutral hydrogen associated with these structures, but also that their re-cessional velocity is significantly lower than the mean group velocity (assuming thatthe low velocity component is associated the Hα trails also in the other three point-ings), strongly supporting a ram pressure stripping scenario. Infalling at 1700 km s−1

through the ICM, whose density is ρ ∼ 6 × 10−4atoms cm−3 at their present periph-eral location (A. Finoguenov, private comm.), galaxies in this group will experience


Figure 10.10: The HI spectra obtained for each pointing.

10.3. Results 147

Figure 10.11: Comparison between the combined HI spectrum obtained from thefour different Arecibo pointings, and the single pointing on the NW trail. It appearsclearly the presence of a low velocity component not associated to the bright galaxiesin BIG.


a ram pressure:P = ρv2 ∼ 3 10−11[dyn cm−2] (10.5)

to an order of magnitude higher. Assuming a stellar surface densities σS ∼ 3×10−2

g cm−2 and an interstellar gas surface densities σg ∼ 10−3 g cm−2, the restoringgravitational force of galaxies is:

F = 2πGσSσg ∼ 1.3 10−11[dyn cm−2]

Thus the restoring gravitational forces (pressure) at their interiors, are significantlysmaller than the ram pressure. In the long run, the increasing ram pressure willfully strip their gaseous material leading to the complete ablation of their interstellargas, thus suppressing the star formation because of fuel exhaustion. A stripped blobof typical radius R of 2.5 kpc and mass M = 108 M; might even experience adeceleration,

a =(P − F )πR2

M= 1.6 10−8[cm s−2]

with a consequent measurable velocity decrease of ∆V = 500 km s−1 in a time asshort as 108 yrs, as observed in this case.

10.3.4 The fate of the stripped gas

Different predictions are made in the literature for what happens to the gas onceit has been stripped. The large extent of the Hα trails and its associated HI gasindicates that it can survive for some 108 yr or even 1 Gyr. This may suggest thatevaporation by the ICM is slow, e.g. because the heat flow is saturated and/or thata tangled magnetic field slows down the heat flow into the trail (Vollmer et al. 2001),as observed in the extended HI plume recently discovered in Virgo by Oosterloo &van Gorkom (2005). In spiral galaxies, if the HI column density is above a few times1020 cm−2 , star formation almost invariably occurs (Boissier et al. 2003). The meancolumn densities in the trail is ∼ this value, suggesting that it could locally exceedthis threshold. Hence, star formation could occur locally in the trails, provided theprocesses that regulate star formation for a cloud in the ICM are similar to thosefor gas clouds in spiral galaxies. The Hα emission in the trails could be signature ofstar formation in act, representing the most extended example of extragalactic starformation ever observed. However we have no evidence of stellar emission from thetrails and the dynamical picture of BIG is consistent with the idea that at least partof the gas that has been stripped is just ionized by ram pressure. In this case theplasma density derived in section 10.3.2 implies an exceedingly short recombinationtime in the ionized trails τr = 1/Neαa ∼ 2-7 Myr, where αa = 4.2 × 10−13 cm3 s−1

(Osterbrock 1989). Can their exceedingly short recombination time of few Myr be

10.3. Results 149

reconciled with an age between some 108 yr and 1.5 Gyr? We need a mechanism tosustain the ionization along the tail and the presence of the cluster IGM comes tohelp. The clouds stripped from a galaxy infalling onto the IGM might be kept ionizedby X-ray bremsstrahlung emission of the IGM. Following Vollmer et al. (2001) andMaloney et al. (1996) the X-ray ionizing photon flux (φi) is:

φi =ln(0.1/0.0136)

1.6 × 10−9

FX

1.5= 8.3 × 108 FX photons cm−2 s−1

where FX is the X-ray flux. Assuming a total cluster X-ray luminosity of 4 ×1043erg cm−2 (Donnelly et al. 1998) the X-ray flux at a projected distance of ∼ 125kpc from the X-ray center (where BIG is observed) is ∼ 2.5 × 10−6erg cm−2s−1 andφi results ∼ 2.1 × 103photons cm−2 s−1. In equilibrium this gives rise to an ionizedcolumn density Ne = φi/αane. Using ne = 10−2cm−3 we obtain Ne ∼ 5 × 1020cm−2,consistent with value measured in the ionized tails. This simple calculation showsthat the stripped gas can survive in the hostile IGM, being kept ionized by the X-rayphotons.

10.3.5 The metal content

In order to determine the metal content of the observed emission line knots we followedthe same procedure described in Chapter 7. The metallicities obtained from thedifferent methods are shown in Tab.10.4. The uncertainty in the abundances is up to±0.2dex.All the star-forming regions in BIG are surprisingly metal-rich. Their metallicitylies in the range 8.5 < 12 + log(O/H) < 8.9. It is well known that irregular andspiral galaxies follow a ”metallicity - luminosity relation” (Skillman et al. 1989).Fig.10.12 shows the ”metallicity - luminosity relation” for galaxies in the Virgo cluster(empty circles, taken from Gavazzi et al. in prep., and obtained using the samemethods and calibrations) and for the star-forming systems in BIG (triangles). Thetwo bright galaxies CGCG97-114 and CGCG97-125 have a normal metal contentfor their luminosity. Conversely the star-forming knots show higher abundances fortheir intrinsic luminosities. If the faintest systems (K1, K5, K2, 114a and 114b)were isolated independently-evolved dwarf galaxies we would measure a metallicity0.6-1.2 dex lower than the one observed in this case. Moreover their abundancesare consistent with the values measured for tidal dwarf systems showing a metalcontent of 12 + log(O/H) ∼ 8.60 independent from their absolute magnitude (e.g.Duc & Mirabel 1999; Duc et al. 2000). The HII regions DW1, DW2 and DW3 havea high-metal content but consistent, within the calibration uncertainties, with theabundances observed in dwarf galaxies of the same luminosity. However (as discussedin Section10.3.1) the star-forming knots in DW3 are probably not gravitationally


Object Tel. Ra23 Rb

23 NII/OIIc NII/Hαd OIII/NIIe Mean Stdev

K1 ESO 8.53 8.51 − − − 8.52 0.02K1 TNG 8.55 8.52 8.75 8.66 8.55 8.61 0.09DW3 a TNG − 8.62 − 8.66 8.70 8.66 0.04DW3 d ESO 8.91 8.79 8.92 8.73 8.68 8.81 0.1197-114b ESO 8.59 8.52 − 8.30 8.48 8.47 0.1397-114a ESO 8.86 8.71 8.68 8.54 8.72 8.70 0.1097-114 LOI 9.00 8.83 8.85 8.70 8.92 8.86 0.11K2 a ESO 8.56 8.51 8.65 8.59 8.56 8.57 0.05K2 b TNG − 8.64 − 8.39 8.64 8.56 0.14DW2 a TNG 8.53 8.44 8.61 8.67 8.71 8.59 0.11K5 ESO − − − 8.35 8.66 8.50 0.22DW1 b MMT 8.56 8.53 − 8.43 8.48 8.50 0.06DW1 c MMT − 8.42 − 8.64 8.58 8.55 0.12DW1 c TNG 8.81 8.70 8.84 8.71 8.66 8.75 0.08DW1 a ESO − − − 8.43 8.49 8.46 0.04DW1 a MMT 8.55 8.52 8.67 8.58 8.53 8.57 0.0697-125 OHP − 8.73 − 8.82 8.65 8.73 0.0897-125 TNG − 8.49 8.77 8.98 8.72 8.74 0.2097-125b ESO − − − 8.45 8.59 8.52 0.1097-125b TNG 8.76 8.64 8.75 8.67 8.67 8.69 0.05K3 TNG 8.89 8.75 8.90 8.84 8.78 8.83 0.07

a: Zaritsky et al. 1994b: McGaugh 1991c: Kewley & Dopita 2002d: Van Zee et al. 1998e: Dutil & Roy 1999

Table 10.4: Metallicities of the galaxies in the BIG group.

10.3. Results 151

Figure 10.12: The relation between Metallicity and B-band Luminosity (with linearbest-fit) for galaxies in nearby clusters (empty circles, adapted from Gavazzi et al.2004). The triangles mark the mean metallicity obtained for the individual knots ofBIG.

bound, thus each knot should be considered as a single faint extragalactic HII regionwith a metallicity ∼ 0.8 dex higher than the one obtained from the metallicity-luminosity relation.These results rule out an evolutionary scenario in which the faint HII region discoveredin BIG are normal independently evolved dwarf galaxies, reinforcing the scenario ofSakai et al. (2002) who proposed that these systems formed from enriched materialstripped by tidal interactions from the two brightest galaxies in BIG.

10.3.6 Dating the starburst.

Contrary to the gaseous filaments, current star formation is clearly observed in allcompact HII regions, dwarf and giant galaxies composing BIG, suggesting that burstsof star formation are presently taking place in this group. Do we have any hinton when the inset of the star bursting phase took place? The dwarf galaxy DW2,and in particular knots DW2b and DW2c show clear Post-Star-Burst signatures intheir spectra, with low residual current star-formation. They have an extremely bluecontinuum (B-R ∼ 0.16), strong Balmer absorption (EW(Hδ)∼ 8A) and [OII] andHα in emission. In particular a clear age gradient is observable passing from DW2a,where star formation is still in place, to DW2c, that shows strong Balmer lines inabsorption with some evidences of residual star formation (see Fig.10.18). These


features indicates that the starburst ended already ∼ 108 years ago (e.g. Poggianti& Barbaro 1997; Poggianti et al. 1999; Kauffmann et al. 2003b).

The star formation history of CGCG97125

The best piece of information for dating the interaction is provided by the brightestgroup galaxy: CGCG97-125. This galaxy is classified as S0a in the CGCG cata-logue (Zwicky et al. 1961), consistent with its red B-R color index (∼ 1.34, see alsoFig.10.17) and with the shape of the continuum optical spectrum (Fig.10.13). How-ever this system is far from being a normal early type galaxy. The presence of stellarshells around CGCG97-125 (see Fig.10.6) clearly indicates a past interaction/mergerevent, as also supported by the disturbed rotation curve analyzed in the previoussections. Numerical simulations predict that the stars from a satellite make a systemof shells several 108 yr after the end of the merging event and then they last for morethan 1 Gyr (Kojima & Noguchi 1997). The spectrum of 97-125 shows a continuum andabsorption features typical of elliptical galaxies; however superimposed to it there arestrong emission lines (see Fig.10.13) indicating that this galaxy is still experiencing astrong burst of star formation: a kind of rejuvenated early type galaxy. Using the blueline-strength indices to determine the age of the last star forming event (Longhettiet al. 1999) (Hδ/FeI ∼1.00 , H+K(CaII)∼ 0.91 and ∆4000∼ 1.78) we estimate thatthe age of the last starburst is ∼1 Gyr, in agreement with the prediction derived fromthe presence of the stellar shells. However these models assume an instantaneousburst (SSP), that is clearly not the case of CGCG97125, the obtained age thus repre-sents only a lower limit of the real burst age. In Fig.10.13 we compare the drift-scanintegrated spectrum and the nuclear spectrum of CGCG97125 obtained at the OHPand TNG telescope respectively: the integrated spectrum appears considerably bluerthan the nuclear one. We can use this difference in order to try to reconstruct the re-cent star formation history of this galaxy. Therefore, assuming that 1)the continuumof the nuclear spectrum is dominated by the old stellar population with no significantcontribution from the recent starburst while 2) the integrated one is strongly con-taminated by new stars produced during the burst, we can try to estimate the ageof the interaction and the stellar mass produced during the burst. Tidal interactionsand merging usually produce a sinking of the gas to the galaxy center triggering aburst of star formation, in contrast with our first assumption. Thus in order to testthe validity of our hypothesis we used the SED fitting procedure proposed by Gavazziet al. (2002a) and developed by Franzetti (2005). We assume a ”a la Sandage” starformation history (SFH):

SFH(t, τ) =t

τ 2× exp(−

t2

2τ 2) (10.6)

and the Bruzual & Charlot (2003) (BC03) population synthesis models. We fitted

10.3. Results 153

Figure 10.13: Comparison between the drift-scan integrated (blue) and nuclear (red)spectrum of CGCG97-125.

Spectrum Z Mass τ tZ log(M/M) Gyr Gyr

Nuclear 0.04 11.01 1.00 13Starburst 0.04 9.27 0.80 1.4

Table 10.5: Best-fitting parameters for the nuclear and starburst component ofCGCG97125.


the nuclear spectrum of CGCG97125, corrected for extinction2 assuming t=13 Gyr, aSalpeter IMF (α = 2.35 from 0.1 to 100 M; Salpeter 1955) and exploring a parametergrid in metallicity (Z) and τ . Z is let free to vary from 1/50 to 2.5 Z in five steps:0.0004, 0.004, 0.008, 0.02, and 0.05. τ varies from 0.1 to 25 Gyr in 45 approximatelylogarithmic steps. The best-fitting parameters obtained using the BC03 models aresummarized in Table 10.5. The best value of τ is consistent with the one (τ ≤ 3.1Gyr)obtained by Gavazzi et al. (2002a) fitting a template of S0 galaxies. This result val-idates our assumption that the continuum of the nuclear spectrum is dominated byan old stellar population of the same age expected for an unperturbed S0. By nor-malizing the obtained model to the observed H-band magnitude and subtracting itto the integrated UV to near-IR SED of CGCG97125, we have the possibility to esti-mate the starburst contribution to the galaxy emission, the burst age and the stellarmass produced during the star formation. The best-fitting parameters obtained forthe starburst SED are summarized in Table 10.5. The burst age results ∼1.4 Gyrand the stellar mass produced during the burst is ∼ 2× 109 M, consistent with thevalues previously obtained from independent estimates (i.e. dynamical models). Theresulting best fitting SED for CGCG97-125 is shown in Fig.10.14 (black model). Themodel well reproduces the observations from the far-ultraviolet to the near infrared,with the exception of the near-ultraviolet. This disagreement does not depends onthe model assumption but on the attenuation law used to correct for internal dustattenuation. In fact, as shown in Appendix II, we assume a Milky Way attenuationlaw (thus with a bump at ∼ 2175 A) that seems not to be valid for normal star form-ing galaxies (see Chap. 7), producing an overestimate of the real galaxy emission innear-ultraviolet. However this does not influence our results as shown in Fig.10.14.We can thus conclude that the star burst in CGCG97-125 initiated ∼1-1.5 Gyr ago,probably produced by a minor merging of a ∼ 2×109 M satellite, and is still takingplace. Our result points out that a minor merging able to disturb the morphologyand the dynamics of a giant galaxy as CGCG97-125, seems not able to strongly mod-ify the mean age of the stellar population, producing only a small fraction (∼2%) ofnew stars (see also Fig.10.15). As in the case of NGC4438 (see Chapter 8), this resultmight probably be representative only for a minor merging into, today gas poor, earlytype galaxies.

2We corrected all the spectrophotometric data using the ultraviolet spectral slope β as suggestedin Chapter 7 for FUV data and the method described in Appendix A for NUV and optical obser-vations. We assume that the nuclear and integrated spectrum are affected by the same amount ofdust extinction, as supported by the similar value for the Hα/Hβ ratio obtained in the two spectra(see Table 10.2).

10.3. Results 155

Figure 10.14: The SED of CGCG97-125, corrected for internal extinction. Nuclearand drift-scan integrated spectra are shown in green. Black circles indicate photomet-ric observations and their relative uncertainties. Best fitting models for the nuclearspectrum (red) and for the starburst component (blue) are given. The resulting bestfitting SED for CGCG97-125 is presented in black.

Figure 10.15: The star formation history of CGCG97-125 as obtained from the SEDfitting procedure.


Figure 10.16: The 2D high resolution spectrum (left) and the optical rotation curve(right) of CGCG97-120

10.3.7 CGCG97-120: simply a foreground galaxy, or an high

velocity intruder?

In the previous sections I never mentioned the brightest galaxy present in the BIGregion: the spiral galaxy CGCG97-120. This massive system has a recessional veloc-ity of ∼5635 km s−1, thus blueshifted with respect to A1367 by approximately 800km s−1. Observations of the neutral hydrogen line show that CGCG97-120 has lostapproximately 90% of its original hydrogen content (HI deficiency = 0.9), suggestingthat the galaxy has crossed the cluster core and that the ram pressure exerted bythe dense intergalactic medium might have caused its hydrogen deficiency. The greatvelocity difference between this galaxy and BIG (∼ 2500km s−1) seems to rule out anyassociation between the two systems, as argued by Sakai et al. (2002) and Gavazziet al. (2003b). However the deep Hα images obtained at the TNG telescope reproposethe question: one of the Hα trail traces in fact a perfect loop around CGCG97-120.Only a blind chance? As shown in Fig.10.6 and Fig.10.16 the galaxy morphology andkinematic are completely unperturbed, showing no signs of interaction. Moreoverthe scattering angle of an interaction between a satellite galaxy and CGCG97-120would be of only ∼4-10 degrees, assuming a classical scattering model and an impactparameter of ∼10 kpc: too small to produce the observed loop. Thus for the momentwe have to suppose that the association between CGCG97-120 and the Hα trails isonly a blind chance.

10.4. Discussion 157

10.4 Discussion

10.4.1 The evolutionary history of the Blue Infalling Group

The amount of information provided by the multiwavelength observations presentedin this paper allow us to reconstruct the evolutionary history of BIG, during thelast 1-2 Gyr. At the beginning of the story BIG was a normal compact group ofgalaxies with a typical dispersion velocity of ∼150-200 km s−1, composed of at leastthree galaxies: a massive evolved early type spiral (CGCG97-125), a massive latetype spiral (CGCG97-114) and a gas rich dwarf galaxy (the satellite that has feedCGCG97-125) with a stellar mass ∼ 109M. Lying in the outskirts of Abell1367 ithas been attracted by the cluster potential starting its infall into the cluster core at amean velocity of ∼ 1700 km s−1. During their journey, all galaxies are perturbed bygravitational interaction with members, as observed in all compact groups. Starsand gas are stripped, forming tidal tails, bridges (as K2), extragalactic compactHII regions (as K5 and K1) and tidal dwarfs (as DW1, DW2 and DW3). Tidalinteractions lowered the restoring force by loosening the potential well of all galaxiesin the group making easier stripping gas from the infalling galaxies by ram pressureand producing the unique Hα trails observed in BIG. In particular the gas rich satelliteis partially dismantled by the combined action of tidal forces and ram pressure, andfinally merged into CGCG97-125 producing stellar shells and a burst of star formation.The combination of gravitational forces and ram pressure is not only consistent withthe evidence that BIG is a compact group that is infalling at ∼ 1700 km s−1 intothe core of Abell1367, but is also necessary to try to explain all the aspects thatmake BIG so unique among other known interacting systems and merger remnants:i.e. the unexpectedly high star formation observed in this group, the presence ofextended Hα trails and its associated neutral hydrogen, the lack of large-scale tidaltails and, as pointed out by Gavazzi et al. (2003b), the colors of dwarf objects DW1,DW2 and DW3 that are significantly bluer than tidal dwarfs observed in interactingsystems (Weilbacher et al. 2000). The IGM compression is in fact able to triggersome star formation in the gas clouds contained within tidal structures (Bekki &Couch 2003), while ram pressure may push some of these clouds free of their parentgalaxies, explaining the absence of tidal features and the extremely blue colors of thedwarf objects in BIG. Recently Mayer et al. (2005) have shown that gravitationaltides can aid ram pressure stripping by diminishing the overall galaxy potential. Thegas stripped along tails fragments into dense clouds and sheet-like structures pressureconfined by the ambient medium with the approximately the same column densityobserved in our case. However their simulations are focused on the evolution of dwarfs(Vrot ∼ 40 km s−1) systems orbiting around a Milk Way like galaxy, and it is not clearwhat would be the effects of the same mechanisms on a massive galaxy infalling intoa cluster.


10.4.2 The contribution of preprocessing to cluster galaxiesevolution.

Galaxy clusters formed not by accreting individual galaxies randomly from the field,but rather through the infall of small groups, falling in along large scale filaments;thus this group represents an unique laboratory reproducing the physical conditionexpected in a cluster still in formation. What can we learn about galaxy cluster evo-lution studying BIG? First of all, we are witnessing the first clear example of a wellformed S0 galaxy infalling into the core of a cluster of galaxies. This observationalevidence suggests that S0 galaxies can form outside clusters and subsequently fallinto them: groups environment is in fact considered as the best place where gravi-tational interactions should operate efficiently and transform a normal spiral into anS0. Moreover gravitational interactions among group’s members are still in act, andCGCG97-125 has recently (∼1.5 Gyr) experienced a minor merging event. The burstof star formation, however, is not able to strongly affect its global optical properties,since the mass of new stars produced is only ∼ 2% of the whole galaxy mass, con-sistent with the recent results obtained by Boselli et al. (2005a) in the Virgo cluster.This suggests that the mechanism responsible of the transformation of CGCG97-125into an S0 is older than 2 Gyr, corresponding to a redshift z ≥ 0.2. At its currentSFR (1.49 M yr−1) the total HI mass of CGCG 97-125 would be depleted in 2.6×109

yr, implying a total burst duration of ∼4 Gyr: consistent with the typical time-scaleof the Butcher-Oemler effect (Butcher & Oemler 1978, 1984).Tidal interactions within group members are not only able to produce morpholog-ical transformation in galaxies, but also to create new systems formed by gas andstars stripped from group’s members. This is the case of the extremely high numberof metal rich star forming dwarfs/extragalactic HII region detected in the infallinggroup. What will be the future of these stripped systems? It is improbable that allthe stripped clusters will infall into the main galaxies, rebuilding the gaseous disk asobserved in field mergers: in fact cluster tides and ram pressure stripping act mutu-ally to strip off the material and to inhibit the disk resettling process. If they aredynamically bounded, they could be the progenitors of dwarf cluster galaxies as de-scribed in models by Kroupa (1998) and Duc et al. (2004). Simulations predict thatyoung compact massive star clusters formed during the merger of gaseous disk galax-ies coalesce within a few 100 Myr forming objects with masses of order 107−109 M,as observed in this group, with negligible dark-matter content. However, till todaymajor mergers were supposed not to have produced a significant fraction of the dwarfpopulation, since each merger are expected to spawn only one or two tidal dwarfgalaxies. Thus the discovery of a great number of extragalactic star forming knotsin BIG (∼10 for one merging event) seems indicate that a tidal formation scenariofor part of the dwarf population in cluster could be reasonable, especially at higherredshift, where groups like BIG are expected to infall at higher rate into the core of


young clusters. Being produced by a single starburst, these systems might also evolveinto dwarf ellipticals, typical of rich clusters.Otherwise, if they will disperse they stars and gas into the cluster their will simply in-crease the fraction of unbound stars, contributing to the Abell1367 intracluster light,supporting the idea that preprocessing could have had a strong contribution in theamount and distribution of intracluster light (Mihos 2004b). A strong contributionto the intracluster light in Abell1367 would also be provided by the Hα trails if someresidual star formation is taking place. In this case these features would represent themost extended example of extragalactic star formation ever observed in the Universe.Surely the Hα trails are strongly contributing to the ICM enrichment, suggesting thata considerable amount of the cluster enrichment might derive from these late-typeintruders, as opposed to winds from elliptical galaxies, commonly accepted as themajor sources of pollution (e.g. Madau et al. 2001; Mori et al. 2002). This idea isstrongly supported by the presence in the NW part of Abell1367 of other two galaxieswith Hα trails (Gavazzi et al. 2001a), pointing out that this may not be a rare phe-nomenon in young clusters. Moreover in the last years an increasing number of X-ray(Hayakawa et al. 2004) and optical (Gavazzi et al. 2001a; Oosterloo & van Gorkom2005) observations has shown that ram pressure stripping could have an importantrole on IGM enrichment and recent combined N-body and hydrodynamic simulationshave pointed out that more of the 10% of the intracluster medium originated fromgas stripped by ram pressure (Domainko et al. 2005). Thus, combined X-ray andoptical studies of infalling groups should help us to shed more light on the effect ofpreprocessing not only on the evolution of cluster galaxies, but also on process of IGMenrichment, an issue which remains unsettled (Tornatore et al. 2004). The evolution-ary scenario here presented points out the great importance of groups like BIG notonly for galaxy evolution but also for the evolution of clusters itself. Since the infallrate of these groups is considerably higher at high redshift, this analysis points outthe strong contribution that small compact groups have probably had in shaping theproperties of both galaxies and clusters of galaxies. Thus BIG represents a RosettaStone group, giving us the chance to shed light, with the great details possible onlyin the local Universe, on physical processes typically expected in young, far away,clusters.


Figure 10.17: B-R color map of BIG (Blue = B; Red = R).


Figure 10.18: The observed smoothed (step 3) one dimensional spectra. The objectidentification and telescope are labeled on each panel.


Figure 10.18: Continue.


Figure 10.18: Continue.

Chapter 11

Discussion & Conclusions

11.1 Discussion

In the Introduction to this thesis I have argued that try to recover galaxy evolutionduring the last 13 Gyr only from observations of today’s Universe represents a real,but fundamental, challenge. The Universe we inhabit is old, and most of the fun isover. In addition (and this is the worst part of the story) the Universe dramaticallyevolved itself, continuously altering the physical conditions of the environments pop-ulated by galaxies. However, as shown in this work, we can still achieve importantpieces of information from the study of local galaxies and, combining this informationwith that obtained at higher redshift, we can try to paint a picture of our knowledgeabout the evolution of galaxies in clusters; exactly the effort that I’m going to attempin this conclusion. What evolutionary scenario emerges from this work?The first estimates of the UV cluster luminosity functions from FOCA, FAUST andGALEX observations, here presented, point out that at these wavelengths the clusterLF is considerably steeper than the field one. The steepening of the UV LF from lowto high density environment is due to the increasing contribution of early-type, nonstar forming galaxies, passing from the field to the cluster core. This represents thefirst evidence of a morphology/star formation - density relation at ultraviolet wave-lengths and demonstrates that we cannot blindly consider UV selected galaxies asstar-forming systems, especially at low UV luminosities and in high density environ-ments. However this also point out the strong potential of ultraviolet observation instudying all cluster galaxies: not only star-forming systems which UV emission tracesthe presence of newly born stars, but also early type galaxies in which such emissionmust be ascribed to low mass old post asymptotic giant branch stars. So let mesummarize what I have learned about the evolution of these different morphologicaltypes.

165

166 11. Discussion & Conclusions

The evolution of elliptical galaxies For the first time, in this work the UV prop-erties of early-type galaxies have been studied down to MB ≤ -15 mag. The newestresult addresses the question raised by O’Connell (1999) concerning the dependenceof the UV properties on galaxy morphology. We have shown that a dichotomy ex-ists between giant and dwarf ellipticals and, to a lesser extent, between ellipticalsand lenticulars. The blueing of the UV color index with luminosity, metallicity, andvelocity dispersion indicates that the UV upturn is more important in massive, metal-rich systems. Since the UV upturn originates from a minority population of old hothelium-burning horizontal-branch (HB) stars, which emission becoms detectable afterat least 10 Gyr (e.g. O’Connell 1999; Brown et al. 2000; Greggio & Renzini 1990;Tantalo et al. 1996), the relation found for giant ellipticals and its small dispersionsuggest that clusters ellipticals represent an old, homogeneous population. This isalso consistent with the dynamical analysis of Abell 1367 where we found evidencethat elliptical galaxies have a Gaussian velocity distribution with a smaller velocitydispersions than the whole cluster sample, representing the virialized, old, cluster pop-ulation. This picture is supported by both higher redshift observations and N-bodysimulations. The population of elliptical galaxies in clusters show little evolution intheir colors and no structural evolution since at least redshift of z ∼1 (Treu et al.2005; Smith et al. 2005). The attempt to reproduce this observational evidence withN-body simulations (Springel & Hernquist 2003) results in the invalidation of theparadigm of elliptical formation by mergers of spiral galaxies. At the time the clusterellipticals were formed in rich clusters, there were simply few if any spiral to merge.It appears clear from simulations, as well as observations of the high-z Universe, thatlarge spiral galaxies as we know them today were very rare at z>2 (Driver et al. 1998;Dickinson et al. 2003; Trujillo et al. 2004; Conselice et al. 2005). It is specially truein dense environment where galaxies had too little time to form large disk from theaccretion of high-angular momentum material. This is also supported by HST Deepfields: beyond z=1 the number density of large, well-formed spirals begins to rapidlydiminish in favor of a smaller, chaotically arranged systems (Labbe et al. 2003). Inaddition the recent Millennium simulation (Springel et al. 2005) has shown that thegiant ellipticals observed in nearby clusters today, were already formed and massiveat very high redshifts (z ∼ 16) and harboured in the center of regions where thefirst structures developed: the progenitors of rich clusters. We can thus concludethat both observations (at all redshifts) and simulations are consistent with the ideathat clusters giant ellipticals are an old, homogeneous population showing no or littleevolution at least in the past 8 Gyr.

This is not the case for the cluster population of dwarf elliptical galaxies. The op-posite behavior in the UV color magnitude relations (reddening of the UV color indexwith luminosity) of dwarfs with respect to giant ellipticals, similar to that observedfor spirals, indicates that the UV spectra of low luminosity objects are shaped by thecontribution of young stars, thus presenting a very different star formation history.


This implies that the stellar population of dwarfs has been formed in discrete and rel-atively recent episodes, as observed in other nearby objects (Grebel 2000). Howeverthis result is not sufficient to discriminate between different theoretical models for dEformation: primordial objects that lost their gas in a supernova-driven galactic wind(Yoshii & Arimoto 1987; Nagashima & Yoshii 2004), dwarfs irregular infalling intocluster and transformed by ram pressure (van Zee et al. 2004) and/or harassment(Moore et al. 1998), or tidal dwarfs (Kroupa 1998; Duc et al. 2004). The higher fre-quency of dwarf ellipticals in high density environments supports the idea that theyare objects transformed by the harsh cluster environments. However the presence ofobservational evidence supporting at least the first two scenarios, and the very highdispersions observed in the UV color magnitudes and in structural and kinematicparameters (de Rijcke et al. 2005) seem to suggest that dwarf ellipticals are mostlikely a mixed population with primordial, and more recently transformed objectsco-existing in the present day Universe. Moreover the reddening of the UV colorindex with luminosity is new evidence that mass drives the star formation history inhot systems (Trager et al. 2000; Gavazzi et al. 2002a; Caldwell et al. 2003; Poggianti2004b) as in rotating ones (Gavazzi et al. 1996; Boselli et al. 2001, see also below).This phenomenon, today refereed as downsizing effect, is observed in both cluster andfield and at least till z ∼0.8 indicating the presence of an ”anti-hierarchical” historyfor star formation in galaxies. The presence of a downsizing effects in all galaxies,independent from their morphological type, represents today the major challenge forCDM models.

The evolution of lenticular galaxies Unlike the rather passive evolution ob-served in cluster ellipticals, much stronger evolution seems present in the populationof cluster S0s. The dispersion observed in the UV color magnitude relation, consider-able higher than ellipticals’, bears witness to recent, minor episodes of star formationcombined with an old stellar population, as determined also from kinematic andspectroscopic observations (Dressler & Sandage 1983; Neistein et al. 1999; Hinz et al.2003). This result is consistent with recent studies of stellar population in early typegalaxies which found significant differences between the ages of the stellar popula-tions of ellipticals and of the S0 galaxies, supporting the scenario of spirals evolvinginto S0s (Kuntschner & Davies 1998; van Dokkum et al. 1998; Terlevich et al. 1999;Poggianti et al. 2001; Smail et al. 2001). All these results are supported by the factthat the fraction of S0s in rich clusters has increased significantly since a redshift ofz ∼ 1 (Smith et al. 2005), with a corresponding decrease of spiral fraction (Dressleret al. 1997).What is the mechanism responsible for the transformation of a gas rich spiral into alenticular galaxy? The toy model presented in Chapter 9 has shown that ram pres-sure alone cannot account for all of the S0 population observed in nearby clusters.


Galaxy-cluster IGM interaction is able to remove most of the gas reservoir inducingimportant structural modifications in the disk properties. We expect that cluster spi-rals have (at least at short wavelengths) more truncated disk profiles, inverting theouter color gradient with respect to similar but unperturbed objects, and then pro-ducing anemic spirals, similar to disk dominated S0s. The surface brightness of thedisk, however, mildly decreases in Hα and in the UV bands while remaining mostlyconstant at longer wavelengths even 5 Gyr after the interaction. Excluding the inter-action with the ICM, the only mechanism able to produce a structural modificationin spiral galaxies are gravitational interaction. Tidal interactions between galaxiesaffect both stars and gas. Stars respond by forming arms and bars, while the gas flowsdirectly toward the central regions within about 108 yr after the initial collision. Thesinking of the gas towards the galaxy center could trigger a burst of star formationand, on longer timescales, a truncation of the stellar disk (Iono et al. 2004), thus al-tering galaxy morphology. On the other hand we have to exclude harassment since itsinfluence is largely limited to low luminosity galaxies, while in bright spirals its effectsare much more limited (Mihos 2004a; Moore et al. 1996). Thus merger-driven S0 for-mation mechanisms appear not to work inside the cluster potential, since low velocityinteractions are extremely rare. On the other hand, these processes should operateefficiently in the group environment, where the encounter velocities are smaller andcluster tides and the hot ICM are not important. The group environment can createS0s and feed them into the accreting cluster. Although the accretion into the clustercore is expected to happen at higher redshift, we have shown in Chapter 10 a clearexample of this phenomenon observed in the local Universe. The starbursting groupinfalling into the core of Abell1367, represents probably the best example of galaxypreprocessing ever observed. The brightest member of this group is an S0 galaxy instrong gravitational interaction with the other group members. It is thus likely thatmany of these S0s were processed via mergers in the group environment before beingincorporated into clusters; especially in the past where the groups’ infall rate was con-siderably higher than today. Moreover the discovery of other groups of S0 galaxies instrong interaction such as the one in the outskirts of the Ursa Major cluster stronglysupport this formation scenario (van Gorkom 2004). This formation scenario is sup-ported by the observational evidence that the bulk of S0 population in clusters wasformed between z∼0.2 and z∼1, when the rate of infall of small group was the highestexperienced by clusters of galaxies (Mihos 2004a). Finally, S0s are a heterogeneousclass, from the bulge dominated to the disky S0s, and it should not be surprisingthat a single mechanism cannot fully account for the range of S0s types (Hinz et al.2001, 2003): if ram pressure is able to produce disk dominated S0s (objects similarto the anemic galaxies of Van der Berg), tidal interaction (and thus preprocessing)are required to account for bulge dominated S0s.


The evolution of the star formation activity in cluster spiral galaxies Wecan conclude that the bulk of the bulge dominated S0 cluster population was formedat higher redshifts, and in environments where tidal interactions were more proba-ble. However the morphology density relation, and in particular the star formation -density relation, as we observe it today is not fully established at high redshift, be-cause we observe how it evolves in clusters, with star forming late type spirals beingtransformed into anemic galaxies with quenched star formation of the same morpho-logical type. In order to have an idea of this phenomenon, look at Fig.11.1. It showsthe distribution of median value of EW(Hα) as a function of the morphological typefor Virgo galaxies with normal gas content (HI-def< 0.4; filled circles) and deficientgalaxies (HI-def> 0.4; empty circles). The figure emphasize that, within each Hub-ble class, galaxies with normal HI content have EW(Hα) systematically higher by afactor two then their deficient counterpart. What is the major mechanism (if any)responsible of this reduction is still unknown: observational results are not alwaysconsistent each other, and their interpretation results not straightforward at all.Let me start from the results presented in Chapter 3 and 4. I compared the UV lu-minosity function of nearby clusters and local field showing that the shape of the LFfor star-forming galaxies does not change significantly in different environments. Theeasiest interpretation of this result is that the dwarf to giant star forming galaxiesratio is independent from the environment; the only thing that changes is the absolutefraction of star forming galaxies (i.e. the normalization of the luminosity function).This is a very simple picture but consistent with the recent work of Balogh et al.(2004) who have shown that the distribution of Hα equivalent widths in star form-ing galaxies does not depend strongly on the local density, while the fraction of starforming galaxies is a steep function of the local density, in all environments. Under-standing the origin of these observed trends is one of the most interesting questionsto be answered, since it probably include the key to shed light on the environmentalinfluence on today’s galaxy evolution.First of all, these results seem to suggest that the mechanism that affects the starformation when a galaxy enters a dense environment, must work on a short time scale(≤ 107 − 108 yr), and must affect bright and faint galaxies in the same way, in orderto preserve the shape of the luminosity function and of the EW(Hα) distribution. Al-though this excludes strangulation from playing a major role in galaxy evolution, dueto its high time scale (≥1 Gyr), it is not clear which mechanism dominates betweenram pressure and tidal interactions. A lot of research groups (i.e. Dressler 2004;Balogh et al. 2004; Goto 2005) have proposed tidal interactions as the major mech-anism responsible of galaxy transformation. This idea is mainly supported by thefact that the decreasing of EW(Hα) with local density has approximately the sameshape in all environments, from cluster to groups (see Fig. 11.2). Since in groups rampressure stripping is supposed to be absent (even if Fujita 2004 has shown that rampressure could be also important in groups), the only mechanisms available to quench


Figure 11.1: The distribution of the individual HαE.W. measurements in the Virgocluster along the Hubble sequence (small dots) and of the median EW(Hα) in binsof Hubble type. Error bars are drawn at the 25th and 75th percentile of the distribu-tion.Filled symbols represent HI-def< 0.4 (unperturbed) objects and open symbolsHI-def> 0.4 (HI deficient) galaxies.


10 1

0.1 1 100

10

20

30

10 1

0.1 1 100

10

20

30

Figure 11.2: The star formation rate as a function of density, comparing groups ofgalaxies with clusters. The upper and lower horizontal dashed lines show the 75%percentile and the median of the equivalent widths. The hashed region shows therelation for the complete sample, while the solid line shows the relation for systemswith 500 km s−1 < σ < 1000 km s−1 (left) and σ < 500 km s−1 (right). Thedependence on local density is identical irrespective of the velocity dispersion of thewhole system. Figure taken from Bower & Balogh (2004).

star formation are low velocity encounters: we return once again to the idea that themain mechanism responsible of the evolution of spiral galaxies is preprocessing in lit-tle groups. It is indisputable that this simple interpretation rules out the galaxy-ICMinteraction; however as shown in this work, in local clusters the role of ram pressureseems significant. First of all in Chapter 5, we have shown that star forming galaxiesin Abell1367 have an higher velocity dispersion than the quiescent population. Thisis observed also in other clusters (Sodre et al. 1989; Stein 1997; Biviano et al. 1997;Adami et al. 1998), and reflects the fact that spirals have an higher velocity disper-sion than ellipticals, and a velocity distribution hardly Gaussian (Boselli & Gavazzi2006). By itself it provides evidence for infall of star forming galaxies into clusters. Ifa consistent fraction of star forming galaxies are still today passing from low to highdensity environments, their star formation activity will be soon quenched in orderto reproduce the observed trends in luminosity function and EW(Hα) distributions.Since in today’s clusters tidal interaction are less probable, this result supports aram pressure scenario. In addition van Gorkom (2004) has shown that the velocitydispersion of gas rich galaxies is far from Gaussian contrary to the one of HI deficientgalaxies, suggesting that gas rich galaxies that enter the cluster center are likely to


Figure 11.3: The ratio of the isophotal Hα and r′ radii as a function of the HIdeficiency for galaxies in the Virgo cluster.

be serious affected by interaction with the ICM. This is only the first, and if possibleless strong evidence, of the role played by galaxy-intracluster medium interaction.In Chapter 9, we have argued that the population of anemic spirals in clusters, withtruncated star forming disks, is produced by ram pressure stripping and that the timescale of the interaction is short (∼100 Myr). In addition a growing number of spiralgalaxies are found with unusual morphology in HI, Hα and radio continuum, suchas the head tail galaxies CGCG97-073 and CGCG97-079 in Abell1367 (see Chapter5, Gavazzi et al. 1995, 2001a), NGC4522 (Vollmer et al. 2004b), NGC4388 (Yoshidaet al. 2004), NGC4569 (see Chapter 9) in Virgo and CGCG160-055 and CGCG160-095 (Bravo-Alfaro et al. 2000, 2001) in Coma. These are prime candidates for ongoingram pressure stripping.In order to try to determine how important are ICM-ISM interactions for galaxy evo-lution, as a part of the undergraduate thesis of I.Arosio (Arosio 2005), we analyzedthe morphological distribution of galaxies in Virgo and Coma showing that the ratioof the Hα to optical radius correlates with the HI deficiency (see Fig.11.3). This resultis consistent with the increase of the fraction of galaxies with truncated star formingtoward the center of the Virgo cluster, observed by Koopmann & Kenney (2004a) and


Figure 11.4: The clustercentric radial distribution of the individual EW(Hα) mea-surements in the Virgo cluster. High and low (B-band) luminosity galaxies are givenwith open and filled dots respectively. Median in bins of 0.5 R/RV ir are given. Errorbars mark the 25th and 75th percentile of the distribution.

with the prediction of the ram pressure model presented in Chapter 9. In additionthe strong correlation between EW(Hα) and HI deficiency observed in nearby clusters(Gavazzi et al. 2002c) completes the ram pressure supporting scenario. To summa-rize, in nearby clusters we observe galaxies that experience ram pressure stripping:the dominant effect on cluster disk galaxies is a reduction of the star formation rate,which goes hand in hand with the HI deficiency, and for most of the galaxies thisseems due to ram pressure.How can we conciliate these results with the universal shape of the EW(Hα) vs den-sity relation presented by Balogh et al. (2004)? This apparent contradiction awaitsan explanation.What I think emerges from this work is that the truth lies probably in the middle:it is indisputable that galaxy preprocessing (and in particular tidal interaction) hasplayed a significant role, especially in shaping the properties of bright (giant) galaxiesat higher redshift; but at the same time it is unquestionable that we observe todaynormal ”field-like” galaxies, not affected by any preprocessing, infalling alone (or in


very loose groups), for the first time into clusters and on which ram pressure’s ef-fects are clearly evident, as shown in the dynamical study of Abell1367. In part, itmight thus be correct to affirm that while tidal interactions have dominated in thepast and have probably shaped the morphology-density relation for giant galaxies,ram pressure dominates in today clusters and is surely affecting the star formationhistory of galaxies but with less influence on their morphology. What is still far frombeing understood is the downsizing effect (i.e. the correlation of the mean age ofstellar populations with the mass). In fact this effect is clearly present in clusters(Gavazzi et al. 2002a; Kodama et al. 2004; Poggianti et al. 2004) where, on the con-trary, environmental effects (whichever you prefer) are expected to be more efficientin quenching star formation in dwarfs than in giant galaxies, since gas and stars areless bounded to the galaxy. This effect is clearly evident in Fig.11.4 where is shownthe dependence of the EW(Hα) on the cluster-centric distance in the Virgo cluster.While the decline in the star formation rate is clear for giant galaxies (passing fromEW(Hα)∼35A at 2 virial radii to EW(Hα)∼6A in the cluster center), we do notidentify any significant trend for dwarf galaxies. This result could be explained ifwe assume that a significant replenishment of dwarf galaxies is occurring into richclusters at the present cosmological epoch. An high infall of dwarf systems is alsosupported by the fact that the velocity dispersion of dwarf star forming galaxies isconsiderably higher than the one of high luminosity spiral systems (Adami et al.1998). Thus understanding how and at which rate galaxies infall and have infalledinto cluster represents another important key to shade light on the evolution of starformation activity with clusters. If confirmed, a high infall rate for today’s dwarfgalaxies will represent a new interesting challenge for hierarchical models of galaxyevolution, in their unfinished attempt of reproduce the Universe we inhabit.

11.2. Conclusions 175

11.2 Conclusions

In this thesis I have investigated the environmental effects on galaxy evolution innearby clusters using a multiwavelength dataset. In particular this analysis has beenfocused on the properties of three different local clusters: Abell1367, Virgo and Coma.These three clusters are among the best studied in the local Universe and, due tothe variety of their environmental conditions (e.g. spiral fraction, X-ray luminosity,evolutionary stage), they represent the most suitable ”laboratory” for comparativestudies. By combining for the first time GALEX UV observations with optical, nearand far infrared data we derived extensive observational evidence of cluster galaxyevolution.

• I determined the first far-UV and near-UV luminosity functions for nearbyclusters finding that in clusters the faint end slope is steeper than in the field.This difference is entirely due to the contribution at low UV luminosities ofnon-star-forming, massive early-type galaxies that are significantly overdensein clusters; while the luminosity function of cluster star-forming galaxies isconsistent with the field one. This indicates that, whatever mechanism affectsthe star formation activity in late-type cluster galaxies, it influences similarlyand with a short time scale the giant and the dwarf components.

• I investigated the dynamical state of Abell1367 showing that this cluster is stilla young cluster forming at the intersection of large scale filaments. At leasttwo subgroups are currently infalling into the main cluster. They show a higherfraction of star forming galaxies than the cluster core, as expected during theearly phase of merging events, confirming that the building up of large scalestructures can strongly affects the evolutionary history of galaxies.

• I studied for the first time the UV properties of a volume-limited sample ofearly-type galaxies showing the presence of a clear dichotomy in the FUV-opticalcolor magnitude relations between giant and dwarf ellipticals. For elliptical andlenticular galaxies, the (FUV-NUV) color becomes bluer with increasing lumi-nosity and with increasing reddening of the optical or near-IR color indices. Forthe dwarfs, the opposite trend is observed. These results are consistent with theidea that the UV emission is dominated by hot, evolved stars in giant systems,while in dwarf ellipticals residual star formation activity is more common.

• While investigating the star formation history of galaxies in nearby clustersusing ultraviolet observations, it has been mandatory to study UV dust atten-uation properties of nearby galaxies in order to look for new recipe’s in oder tocorrect GALEX data. I confirmed that normal galaxies follow a LTIR/LFUV −βrelation offset from the one observed for starburst galaxies. The dispersion of


this relation is found to weakly correlate with the galaxy star formation history.I studied the correlation of dust attenuation with other global properties, suchas the metallicity, dynamical mass, ionized gas attenuation, Hα emission andmass surface density providing some empirical relations from which the totalinfrared to far ultraviolet ratio (LTIR/LFUV ) can be estimated when far infrareddata are absent. This result represents only the tip of the iceberg of a study ofdust properties in normal galaxies. Only comparing data with models we willbe able to properly correct data for dust extinction and thus to estimate thestar formation rate in galaxies.

Finally I studied in great details the star formation history of three different systemsconsidered as the prototypes of the three main environmental mechanisms able toperturb galaxy evolution, namely: high velocity interactions, ram pressure strippingand galaxy preprocessing.

• We showed that in today’s cluster galaxies high-velocity tidal encounters be-tween two galaxies of similar mass are able to perturb the stellar distributionand thus produce important tidal tails, but are not sufficient to significantlyincrease the star formation activity of cluster galaxies.

• Moreover we demonstrated that ram pressure stripping alone is not able totransform a spiral galaxy into an S0, reproducing the structural properties ofpresent-day lenticulars.

• Strong transformations in both morphology and star formation activity can beproduced by the mutual effects of low velocity encounters and ram pressurestripping in small groups infalling into the cluster core (preprocessing), as ob-served in Abell1367. Studying this unique example of preprocessing in the localUniverse we showed that infalling groups could have a strong influence not onlyon galaxy evolution but also on the evolution of cluster galaxies, significantlycontributing to the enrichment of the intracluster medium and to the intraclus-ter light.

Considering all these observational results I conclude that

• Giant ellipticals are an old, homogeneous population showing no or little evolu-tion at least in the past 8 Gyr unlike dwarf ellipticals which still contains youngstellar populations.

• The importance of different environmental mechanism is directly linked withthe age of the Universe.

• Tidal interactions and prepocessing have probably dominated in the past Uni-verse and has shaped part of the morphology-density relation during the clusteraccretion of small groups.

11.2. Conclusions 177

• Ram pressure dominates in today clusters and is surely affecting the star for-mation history of galaxies with less influence on their morphology.

• The heterogeneous class of S0s galaxies, from bulge dominated to the disky S0s,is not the result of a single transformation mechanism: if ram pressure is ableto produce disk dominated S0s, tidal interaction (and thus preprocessing) arerequired to account for bulge dominated S0s.

• Different observational clues confirm the presence of a correlation between themean age of stellar populations and the mass of their parent galaxies (downsizingeffect). In the framework of the hierarchical model of galaxy formation, theorigin of the downsizing effect remains unsolved and represents one of the mainchallenges for models of galaxy evolution.

Appendix A

The extinction correction

Here we present the method used to correct for dust attenuation multiwavelengthobservations used in this work, and proposed by Boselli et al. (2003a) As discussed inChapter 7 the observed stellar radiation of galaxies, from UV to near-IR wavelengths,is subject to internal extinction (absorption plus scattering) by the interstellar dust.Estimating the dust extinction at different λ in external galaxies is very difficult(it has been done only for the Magellanic clouds). This difficulty is mainly dueto two reasons: a) the extinction strongly depends on the relative geometry of theemitting stars and of the absorbing dust within the disc of galaxies. The young stellarpopulation are mostly located along the disc in a thin layer, while the old populationsforms a thicker layer. This point is further complicated by the fact that different dustcomponents (very small grains, big grains etc.), which have different opacities to theUV, visible or near-IR light, have themselves different geometrical distributions bothon the large and small scales. b) it is still uncertain whether the Galactic extinctionlaw is universal, or if it changes with metallicity and/or with the UV radiation field.Detailed observations of resolved stars in the Small Magellanic Cloud by Bouchetet al. (1985) indicate that the extinction law in the optical domain is not significantlydifferent from the Galactic one in galaxies with a UV field ∼10 times higher and ametallicity ∼10 times lower than those of the Milky Way. A steeper UV rise and aweaker 2200 Abump than in the Galactic extinction law have been however observedin the LMC and SMC (Mathis 1990). After the results of Chapter 7 the adoption ofthe Galactic extinction law for external galaxies could seem not reasonable, however inthis moment we have not yet a good alternative, moreover no simple analytic functionsdescribing the geometrical distribution of emitting stars and absorbing dust, both onsmall and large scales, are yet available. The radiative transfer models of Witt &Gordon (2000) have however shown that the FIR to UV flux ratio, being mostlyindependent of the geometry, of the star formation history (the two radiations areproduced by similar stellar populations) and of the adopted extinction law, is a robustestimator of the dust extinction at UV wavelengths. From the value of the TIR/FUV

179

180 A. The extinction correction

(measured or obtained with the recipes presented in Chapter 7 we can thus estimateA(FUV), following Buat et al. (2005):

A(FUV ) = −0.0333 ∗ y3 + 0.3522 ∗ y2 + 1.1960 ∗ y + 0.4967 [mag] (A.1)

where y is log(TIR/FUV ). A(λ) can be derived from A(FUV) once an extinctionlaw and a geometry for the dust and star distribution are assumed. We adopt thesandwitch model, where a thin layer of dust of thickness ζis embedded in a thick layerof stars:

A(λ) = −2.5 · log

( [

1 − ζ(λ)

2

]

(

1 + e−τ(λ)·sec(i))

+

+

[

ζ(λ)

τ(λ) · sec(i)

]

·(

1 − e−τ(λ)·sec(i))

)

[mag] (A.2)

where the dust to stars scale height ratio ζ(λ) depends on λ (in units of A) as:

ζ(λ) = 1.0867−5.501 × 10−5 · λ. (A.3)

This has been calibrated adopting the average between the optically thin and opticallythick cases with λ dependent dust to star scale height ratios given by Boselli &Gavazzi (1994). Observations of some edge-on nearby galaxies show that it is stillunclear whether ζ depends or not on λ (Xilouris et al. 1999). As shown in Gavazziet al. (2002a), however, similar values of Ai(λ)are obtained in the case of a sandwitchmodel and of the extreme case of a slab model (ζ = 1), meaning that the highuncertainty on ζ is not reflected on A(λ). In the case of the FUV band ( λ ∼ 1530A), ζ = 1, and Eq. A.2 reduces to a simple slab model. In this case τ(UV) can bederived by inverting Eq. A.2:

τ(UV) = [1/sec(i)] ·(

0.0259 + 1.2002 × Ai(FUV) + 1.5543 × Ai(FUV)2 +

− 0.7409 × Ai(FUV)3 + 0.2246 × Ai(FUV)4)

(A.4)

using the galactic extinction law k(λ) (Savage & Mathis 1979), we than derive:

τ(λ) = τ(UV) · k(λ)/k(UV) (A.5)

and we compute the complete set of Ai(λ) using Eq. A.2.

Appendix B

Estimate of the < 912A flux fromHα + [NII]

The stellar radiation field with λ <912 Aionizes the gas, which re-emits, via recom-bination lines. If the gas is optically thick in the Lyman continuum, the number ofphotons in a specific recombination line is directly proportional to the number of starphotons in the Lyman continuum. In the case of Hβ this number is given by equation(5.23) in Osterbrock (1989). For Hα we have:

∞∫

ν0

Lν

hνdν = LHα · C (B.1)

where:

1/C = hνHβ

αeffHβ (Ho, T )

αB(Ho, T )

FHα

FHβ

(B.2)

Assuming T=10000K and the Osterbrock case B:αeff

Hβ (Ho, T ) = 3 × 10−14(cm3 sec−1)αB(Ho, T ) = 2.59 × 10−13(cm3 sec−1)and FHα

FHβ= 2.87

From the Hα luminosity it is thus possible to recover the number of ionizing photons,which can be compared with the similar quantity derived from the integral on themodel spectrum.A conservative estimate of the uncertainty on the derived < 912 A flux is 1 mag.

181

Bibliography

Abadi, M. G., Moore, B., & Bower, R. G. 1999, MNRAS, 308, 947

Abazajian, K., Adelman-McCarthy, J. K., Agueros, M. A., et al. 2005, AJ, 129, 1755

Abraham, R. G., Tanvir, N. R., Santiago, B. X., et al. 1996a, MNRAS, 279, L47

Abraham, R. G., van den Bergh, S., Glazebrook, K., et al. 1996b, ApJS, 107, 1

Adami, C., Biviano, A., & Mazure, A. 1998, A&A, 331, 439

Alton, P. B., Trewhella, M., Davies, J. I., et al. 1998, A&A, 335, 807

Alton, P. B., Xilouris, E. M., Bianchi, S., Davies, J., & Kylafis, N. 2000, A&A, 356,795

Andreon, S. 1999, A&A, 351, 65

Arosio, I. 2005, Undergraduate Thesis, Universita degli Studi di Milano-Bicocca

Balogh, M., Eke, V., Miller, C., et al. 2004, MNRAS, 348, 1355

Barnes, J. E. & Hernquist, L. 1996, ApJ, 471, 115

Bechtold, J., Forman, W., Jones, C., et al. 1983, ApJ, 265, 26

Beers, T. C., Flynn, K., & Gebhardt, K. 1990, AJ, 100, 32

Beers, T. C., Gebhardt, K., Forman, W., Huchra, J. P., & Jones, C. 1991, AJ, 102,1581

Bekki, K. 1999, ApJ, 510, L15

Bekki, K. & Couch, W. J. 2003, ApJ, 596, L13

Bekki, K., Couch, W. J., & Shioya, Y. 2002, ApJ, 577, 651

Bell, E. F. 2002, ApJ, 577, 150

183

184 BIBLIOGRAPHY

Bell, E. F. & Kennicutt, R. C. 2001, ApJ, 548, 681

Bernardi, M., Alonso, M. V., da Costa, L. N., et al. 2002, AJ, 123, 2990

Bernardi, M., Sheth, R. K., Annis, J., et al. 2003, AJ, 125, 1817

Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393

Binggeli, B., Sandage, A., & Tammann, G. A. 1985, AJ, 90, 1681

Bird, C. M. 1994, AJ, 107, 1637

Bird, C. M. & Beers, T. C. 1993, AJ, 105, 1596

Biviano, A., Katgert, P., Mazure, A., et al. 1997, A&A, 321, 84

Bliton, M., Rizza, E., Burns, J. O., Owen, F. N., & Ledlow, M. J. 1998, MNRAS,301, 609

Boissier, S., Boselli, A., Buat, V., Donas, J., & Milliard, B. 2004, A&A, 424, 465

Boissier, S., Gil de Paz, A., Madore, B. F., et al. 2005, ApJ, 619, L83

Boissier, S. & Prantzos, N. 2000, MNRAS, 312, 398

Boissier, S., Prantzos, N., Boselli, A., & Gavazzi, G. 2003, MNRAS, 346, 1215

Boselli, A., Boissier, S., Cortese, L., et al. 2005a, ApJ, 623, L13

Boselli, A., Cortese, L., Deharveng, J. M., et al. 2005b, ApJ, 629, L29

Boselli, A. & Gavazzi, G. 1994, A&A, 283, 12

Boselli, A. & Gavazzi, G. 2002, A&A, 386, 124

Boselli, A. & Gavazzi, G. 2006, PASP, submitted

Boselli, A., Gavazzi, G., Donas, J., & Scodeggio, M. 2001, AJ, 121, 753

Boselli, A., Gavazzi, G., Lequeux, J., et al. 1997a, A&A, 327, 522

Boselli, A., Gavazzi, G., & Sanvito, G. 2003a, A&A, 402, 37

Boselli, A., Iglesias-Paramo, J., Vılchez, J. M., & Gavazzi, G. 2002a, A&A, 386, 134

Boselli, A., Lequeux, J., & Gavazzi, G. 2002b, A&A, 384, 33

BIBLIOGRAPHY 185

Boselli, A., Sauvage, M., Lequeux, J., Donati, A., & Gavazzi, G. 2003b, A&A, 406,867

Boselli, A., Tuffs, R. J., Gavazzi, G., Hippelein, H., & Pierini, D. 1997b, A&AS, 121,507

Bouchet, P., Lequeux, J., Maurice, E., Prevot, L., & Prevot-Burnichon, M. L. 1985,A&A, 149, 330

Bower, R. G. & Balogh, M. L. 2004, in Clusters of Galaxies: Probes of CosmologicalStructure and Galaxy Evolution, 326

Bravo-Alfaro, H., Cayatte, V., van Gorkom, J. H., & Balkowski, C. 2000, AJ, 119,580

Bravo-Alfaro, H., Cayatte, V., van Gorkom, J. H., & Balkowski, C. 2001, A&A, 379,347

Bressan, A., Chiosi, C., & Fagotto, F. 1994, ApJS, 94, 63

Brown, T. M., Bowers, C. W., Kimble, R. A., Sweigart, A. V., & Ferguson, H. C.2000, ApJ, 532, 308

Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000

Buat, V. 1992, A&A, 264, 444

Buat, V., Boselli, A., Gavazzi, G., & Bonfanti, C. 2002, A&A, 383, 801

Buat, V., Donas, J., Milliard, B., & Xu, C. 1999, A&A, 352, 371

Buat, V., Iglesias-Paramo, J., Seibert, M., et al. 2005, ApJ, 619, L51

Buat, V. & Xu, C. 1996, A&A, 306, 61

Burgarella, D., Buat, V., & Iglesias-Paramo, J. 2005, MNRAS, 360, 1413

Burstein, D., Bertola, F., Buson, L. M., Faber, S. M., & Lauer, T. R. 1988, ApJ, 328,440

Burstein, D. & Heiles, C. 1982, AJ, 87, 1165

Butcher, H. & Oemler, A. 1978, ApJ, 226, 559

Butcher, H. & Oemler, A. 1984, ApJ, 285, 426

Byrd, G. & Valtonen, M. 1990, ApJ, 350, 89

186 BIBLIOGRAPHY

Caldwell, N., Rose, J. A., & Concannon, K. D. 2003, AJ, 125, 2891

Calzetti, D. 1997, AJ, 113, 162

Calzetti, D. 2001, PASP, 113, 1449

Calzetti, D., Bohlin, R. C., Kinney, A. L., Storchi-Bergmann, T., & Heckman, T. M.1995, ApJ, 443, 136

Calzetti, D., Kennicutt, R. C., Bianchi, L., et al. 2005, ApJ, in press, astro-ph/0507427

Calzetti, D., Kinney, A. L., & Storchi-Bergmann, T. 1994, ApJ, 429, 582

Caplan, J. & Deharveng, L. 1986, A&A, 155, 297

Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245

Cayatte, V., Kotanyi, C., Balkowski, C., & van Gorkom, J. H. 1994, AJ, 107, 1003

Cayatte, V., van Gorkom, J. H., Balkowski, C., & Kotanyi, C. 1990, AJ, 100, 604

Charlot, S. & Fall, S. M. 2000, ApJ, 539, 718

Charlot, S. & Longhetti, M. 2001, MNRAS, 323, 887

Chemin, L., Cayatte, V., Balkowski, C., et al. 2005, A&A, 436, 469

Chung, A., van Gorkom, J. H. Kenney, J. D. P., & Vollmer, B. 2005, astro-ph/0507592

Churazov, E., Forman, W., Jones, C., & Bohringer, H. 2003, ApJ, 590, 225

Cohen, M., Sasseen, T. P., & Bowyer, S. 1994, ApJ, 427, 848

Colless, M. 1989, MNRAS, 237, 799

Colless, M., Dalton, G., Maddox, S., et al. 2001, MNRAS, 328, 1039

Combes, F., Dupraz, C., Casoli, F., & Pagani, L. 1988, A&A, 203, L9

Conselice, C. J., Blackburne, J. A., & Papovich, C. 2005, ApJ, 620, 564

Contursi, A., Boselli, A., Gavazzi, G., et al. 2001, A&A, 365, 11

Cortese, L., Boselli, A., Buat, V., et al. 2006, ApJ, in press

Cortese, L., Boselli, A., Gavazzi, G., et al. 2005, ApJ, 623, L17

BIBLIOGRAPHY 187

Cortese, L., Gavazzi, G., Boselli, A., Iglesias-Paramo, J., & Carrasco, L. 2004, A&A,425, 429

Cortese, L., Gavazzi, G., Boselli, A., et al. 2003a, A&A, 410, L25

Cortese, L., Gavazzi, G., Iglesias-Paramo, J., Boselli, A., & Carrasco, L. 2003b, A&A,401, 471

Couch, W. J., Barger, A. J., Smail, I., Ellis, R. S., & Sharples, R. M. 1998, ApJ, 497,188

Cowie, L. L. & Songaila, A. 1977, Nature, 266, 501

Dale, D. A., Helou, G., Contursi, A., Silbermann, N. A., & Kolhatkar, S. 2001, ApJ,549, 215

De Propris, R., Colless, M., Driver, S. P., et al. 2003, MNRAS, 342, 725

de Rijcke, S., Michielsen, D., Dejonghe, H., Zeilinger, W. W., & Hau, G. K. T. 2005,A&A, 438, 491

Deharveng, J.-M., Boselli, A., & Donas, J. 2002, A&A, 393, 843

Deharveng, J.-M., Sasseen, T. P., Buat, V., et al. 1994, A&A, 289, 715

Devereux, N. A. & Young, J. S. 1990, ApJ, 359, 42

Dickey, J. M. & Gavazzi, G. 1991, ApJ, 373, 347

Dickinson, M., Papovich, C., Ferguson, H. C., & Budavari, T. 2003, ApJ, 587, 25

Domainko, W., Mair, M., Kapferer, W., et al. 2005, astro-ph/0507605

Donas, J., Milliard, B., & Laget, M. 1991, A&A, 252, 487

Donas, J., Milliard, B., & Laget, M. 1995, A&A, 303, 661

Donati, A. 2004, Undergraduate Thesis, Universita degli Studi di Milano

Donnelly, R. H., Forman, W., Jones, C., et al. 2001, ApJ, 562, 254

Donnelly, R. H., Markevitch, M., Forman, W., et al. 1998, ApJ, 500, 138

Draine, B. T. 2003, ApJ, 598, 1017

Dressler, A. 1980, ApJ, 236, 351

188 BIBLIOGRAPHY

Dressler, A. 2004, in Clusters of Galaxies: Probes of Cosmological Structure andGalaxy Evolution, 207–+

Dressler, A., Oemler, A. J., Couch, W. J., et al. 1997, ApJ, 490, 577

Dressler, A. & Sandage, A. 1983, ApJ, 265, 664

Dressler, A. & Shectman, S. A. 1988, AJ, 95, 985

Driver, S. P., Fernandez-Soto, A., Couch, W. J., et al. 1998, ApJ, 496, L93+

Duc, P.-A., Bournaud, F., & Masset, F. 2004, A&A, 427, 803

Duc, P.-A., Brinks, E., Springel, V., et al. 2000, AJ, 120, 1238

Duc, P.-A. & Mirabel, I. F. 1998, A&A, 333, 813

Duc, P.-A. & Mirabel, I. F. 1999, in IAU Symp. 186: Galaxy Interactions at Low andHigh Redshift, 61

Dutil, Y. & Roy, J. 1999, ApJ, 516, 62

Enßlin, T. A., Biermann, P. L., Klein, U., & Kohle, S. 1998, A&A, 332, 395

Enßlin, T. A. & Bruggen, M. 2002, MNRAS, 331, 1011

Fabricant, D., Franx, M., & van Dokkum, P. 2000, ApJ, 539, 577

Ferguson, H. C. 1994, in Dwarf Galaxies, 475

Ferguson, H. C. & Binggeli, B. 1994, A&A Rev., 6, 67

Ferguson, H. C. & Sandage, A. 1991, AJ, 101, 765

Forbes, D. A., Beasley, M. A., Bekki, K., Brodie, J. P., & Strader, J. 2003, Science,301, 1217

Forman, W., Churazov, E., David, L., et al. 2003, astro-ph/0301476

Franzetti, P. 2005, Ph.D. Thesis, Universita degli Studi di Milano-Bicocca

Frei, Z., Guhathakurta, P., Gunn, J. E., & Tyson, J. A. 1996, AJ, 111, 174

Fujita, Y. 2004, PASJ, 56, 29

Fujita, Y. & Nagashima, M. 1999, ApJ, 516, 619

Gavazzi, G. 1978, A&A, 69, 355

BIBLIOGRAPHY 189

Gavazzi, G., Bonfanti, C., Sanvito, G., Boselli, A., & Scodeggio, M. 2002a, ApJ, 576,135

Gavazzi, G., Boselli, A., Cortese, L., et al. 2006, A&A, in press

Gavazzi, G., Boselli, A., Donati, A., Franzetti, P., & Scodeggio, M. 2003a, A&A, 400,451

Gavazzi, G., Boselli, A., Mayer, L., et al. 2001a, ApJ, 563, L23

Gavazzi, G., Boselli, A., Pedotti, P., Gallazzi, A., & Carrasco, L. 2002b, A&A, 386,114

Gavazzi, G., Boselli, A., Pedotti, P., Gallazzi, A., & Carrasco, L. 2002c, A&A, 396,449

Gavazzi, G., Boselli, A., Scodeggio, M., Pierini, D., & Belsole, E. 1999a, MNRAS,304, 595

Gavazzi, G., Carrasco, L., & Galli, R. 1999b, A&AS, 136, 227

Gavazzi, G., Catinella, B., Carrasco, L., Boselli, A., & Contursi, A. 1998, AJ, 115,1745

Gavazzi, G., Contursi, A., Carrasco, L., et al. 1995, A&A, 304, 325

Gavazzi, G., Cortese, L., Boselli, A., et al. 2003b, ApJ, 597, 210

Gavazzi, G., Donati, A., Cucciati, O., et al. 2005, A&A, 430, 411

Gavazzi, G., Franzetti, P., Scodeggio, M., Boselli, A., & Pierini, D. 2000, A&A, 361,863

Gavazzi, G. & Jaffe, W. 1987, A&A, 186, L1

Gavazzi, G., Marcelin, M., Boselli, A., et al. 2001b, A&A, 377, 745

Gavazzi, G., Perola, G. C., & Jaffe, W. 1981, A&A, 103, 35

Gavazzi, G., Pierini, D., & Boselli, A. 1996, A&A, 312, 397

Gavazzi, G. & Trinchieri, G. 1983, ApJ, 270, 410

Gavazzi, G., Zaccardo, A., Sanvito, G., Boselli, A., & Bonfanti, C. 2004, A&A, 417,499

Gavazzi, G., Zibetti, S., Boselli, A., et al. 2001c, A&A, 372, 29

190 BIBLIOGRAPHY

Geller, M. J. & Huchra, J. P. 1989, Science, 246, 897

Ghigna, S., Moore, B., Governato, F., et al. 1998, MNRAS, 300, 146

Gil de Paz, A. & Madore, B. F. 2005, ApJS, 156, 345

Girardi, M., Giuricin, G., Mardirossian, F., Mezzetti, M., & Boschin, W. 1998, ApJ,505, 74

Glazebrook, K., Blake, C., Economou, F., Lilly, S., & Colless, M. 1999, MNRAS, 306,843

Gnedin, O. Y. 2003, ApJ, 582, 141

Godwin, J. G., Metcalfe, N., & Peach, J. V. 1983, MNRAS, 202, 113

Godwin, J. G. & Peach, J. V. 1982, MNRAS, 200, 733

Golev, V. & Prugniel, P. 1998, A&AS, 132, 255

Gomez, P. L., Loken, C., Roettiger, K., & Burns, J. O. 2002, ApJ, 569, 122

Gordon, K. D., Clayton, G. C., Witt, A. N., & Misselt, K. A. 2000, ApJ, 533, 236

Goto, T. 2005, MNRAS, 359, 1415

Gottlober, S., Klypin, A., & Kravtsov, A. V. 2001, ApJ, 546, 223

Grebel, E. K. 2000, in ESA SP-445: Star Formation from the Small to the LargeScale, 87

Grebenev, S. A., Forman, W., Jones, C., & Murray, S. 1995, ApJ, 445, 607

Greggio, L. & Renzini, A. 1990, ApJ, 364, 35

Gunn, J. E. & Gott, J. R. I. 1972, ApJ, 176, 1

Hawkins, E., Maddox, S., Cole, S., et al. 2003, MNRAS, 346, 78

Hayakawa, A., Furusho, T., Yamasaki, N. Y., Ishida, M., & Ohashi, T. 2004, PASJ,56, 743

Haynes, M. P. & Giovanelli, R. 1984, AJ, 89, 758

Haynes, M. P., Jore, K. P., Barrett, E. A., Broeils, A. H., & Murray, B. M. 2000, AJ,120, 703

BIBLIOGRAPHY 191

Heckman, T. M., Robert, C., Leitherer, C., Garnett, D. R., & van der Rydt, F. 1998,ApJ, 503, 646

Heisler, J., Tremaine, S., & Bahcall, J. N. 1985, ApJ, 298, 8

Helfer, T. T., Thornley, M. D., Regan, M. W., et al. 2003, ApJS, 145, 259

Helou, G., Khan, I. R., Malek, L., & Boehmer, L. 1988, ApJS, 68, 151

Hibbard, J. E., Bianchi, L., Thilker, D. A., et al. 2005, ApJ, 619, L87

Hibbard, J. E. & van Gorkom, J. H. 1996, AJ, 111, 655

Hinz, J. L., Rieke, G. H., & Caldwell, N. 2003, AJ, 126, 2622

Hinz, J. L., Rix, H.-W., & Bernstein, G. M. 2001, AJ, 121, 683

Hoopes, C. G., Heckman, T. M., Strickland, D. K., et al. 2005, ApJ, 619, L99

Hubble, E. & Humason, M. L. 1931, ApJ, 74, 43

Hubble, E. P. 1925, The Observatory, 48, 139

Hummel, E. & Saikia, D. J. 1991, A&A, 249, 43

Hunsberger, S. D., Charlton, J. C., & Zaritsky, D. 1998, ApJ, 505, 536

Ibata, R., Irwin, M., Lewis, G., Ferguson, A. M. N., & Tanvir, N. 2001, Nature, 412,49

Iglesias-Paramo, J., Boselli, A., Cortese, L., Vılchez, J. M., & Gavazzi, G. 2002, A&A,384, 383

Iglesias-Paramo, J., Boselli, A., Gavazzi, G., Cortese, L., & Vılchez, J. M. 2003, A&A,397, 421

Iglesias-Paramo, J., Boselli, A., Gavazzi, G., & Zaccardo, A. 2004, A&A, 421, 887

Ilbert, O., Tresse, L., Zucca, E., et al. 2004, astro-ph/0409134

Inoue, A. K. 2003, PASJ, 55, 901

Iono, D., Yun, M. S., & Mihos, J. C. 2004, ApJ, 616, 199

Isobe, T., Feigelson, E. D., Akritas, M. G., & Babu, G. J. 1990, ApJ, 364, 104

Issa, M. R., MacLaren, I., & Wolfendale, A. W. 1990, A&A, 236, 237

192 BIBLIOGRAPHY

Jarrett, T. H., Chester, T., Cutri, R., Schneider, S. E., & Huchra, J. P. 2003, AJ,125, 525

Johnston, K. V., Sackett, P. D., & Bullock, J. S. 2001, ApJ, 557, 137

Jones, C., Mandel, E., Schwarz, J., et al. 1979, ApJ, 234, L21

Jore, K. P., Broeils, A. H., & Haynes, M. P. 1996, AJ, 112, 438

Katz, N. & White, S. D. M. 1993, ApJ, 412, 455

Kauffmann, G. 1995, MNRAS, 274, 153

Kauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003a, MNRAS, 346, 1055

Kauffmann, G., Heckman, T. M., White, S. D. M., et al. 2003b, MNRAS, 341, 33

Kauffmann, G., White, S. D. M., & Guiderdoni, B. 1993, MNRAS, 264, 201

Kenney, J. D. & Young, J. S. 1988, ApJS, 66, 261

Kenney, J. D. P., Rubin, V. C., Planesas, P., & Young, J. S. 1995, ApJ, 438, 135

Kenney, J. D. P. & Yale, E. E. 2002, ApJ, 567, 865

Kennicutt, R. C. 1983, ApJ, 272, 54

Kennicutt, R. C. 1992, ApJ, 388, 310

Kennicutt, R. C. 1998, ARA&A, 36, 189

Kennicutt, R. C., Tamblyn, P., & Congdon, C. E. 1994, ApJ, 435, 22

Kewley, L. J. & Dopita, M. A. 2002, ApJS, 142, 35

Kewley, L. J., Geller, M. J., Jansen, R. A., & Dopita, M. A. 2002, AJ, 124, 3135

Kewley, L. J., Jansen, R. A., & Geller, M. J. 2005, PASP, 117, 227

Klypin, A., Kravtsov, A. V., Valenzuela, O., & Prada, F. 1999, ApJ, 522, 82

Kodaira, K., Watanabe, T., Onaka, T., & Tanaka, W. 1990, ApJ, 363, 422

Kodama, T., Yamada, T., Akiyama, M., et al. 2004, MNRAS, 350, 1005

Kogut, A., Spergel, D. N., Barnes, C., et al. 2003, ApJS, 148, 161

Kojima, M. & Noguchi, M. 1997, ApJ, 481, 132

BIBLIOGRAPHY 193

Kong, X., Charlot, S., Brinchmann, J., & Fall, S. M. 2004, MNRAS, 349, 769

Koopmann, R. A. & Kenney, J. D. P. 2004a, ApJ, 613, 866

Koopmann, R. A. & Kenney, J. D. P. 2004b, ApJ, 613, 851

Kotanyi, C., van Gorkom, J. H., & Ekers, R. D. 1983, ApJ, 273, L7

Kroupa, P. 1998, MNRAS, 300, 200

Kroupa, P., Tout, C. A., & Gilmore, G. 1993, MNRAS, 262, 545

Kuntschner, H. & Davies, R. L. 1998, MNRAS, 295, L29

Labbe, I., Rudnick, G., Franx, M., et al. 2003, ApJ, 591, L95

Laird, E. S., Nandra, K., Adelberger, K. L., Steidel, C. C., & Reddy, N. A. 2005,MNRAS, 359, 47

Lampton, M., Deharveng, J. M., & Bowyer, S. 1990, in IAU Symp. 139, 449

Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980a, ApJ, 237, 692

Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980b, ApJ, 237, 692

Leitherer, C. & Heckman, T. M. 1995, ApJS, 96, 9

Leitherer, C., Schaerer, D., Goldader, J. D., et al. 1999, ApJS, 123, 3

Lilly, S. J., Le Fevre, O., Hammer, F., & Crampton, D. 1996, ApJ, 460, L1+

Longhetti, M., Bressan, A., Chiosi, C., & Rampazzo, R. 1999, A&A, 345, 419

Machacek, M. E., Jones, C., & Forman, W. R. 2004, ApJ, 610, 183

Madau, P., Ferrara, A., & Rees, M. J. 2001, ApJ, 555, 92

Madau, P., Pozzetti, L., & Dickinson, M. 1998, ApJ, 498, 106

Madgwick, D. S., Lahav, O., Baldry, I. K., et al. 2002, MNRAS, 333, 133

Maloney, P. R., Hollenbach, D. J., & Tielens, A. G. G. M. 1996, ApJ, 466, 561

Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005, ApJ, 619, L1

Mathis, J. S. 1990, ARA&A, 28, 37

194 BIBLIOGRAPHY

Mayer, L., Mastropietro, C., Wadsley, J., Stadel, J., & Moore, B. 2005, MNRAS sub-mitted, astro-ph/0504277

McGaugh, S. S. 1991, ApJ, 380, 140

Mendes de Oliveira, C., Cypriano, E. S., Sodre, L., & Balkowski, C. 2004, ApJ, 605,L17

Merritt, D. 1983, ApJ, 264, 24

Merritt, D. 1984, ApJ, 276, 26

Meurer, G. R., Heckman, T. M., & Calzetti, D. 1999, ApJ, 521, 64

Meurer, G. R., Heckman, T. M., Leitherer, C., et al. 1995, AJ, 110, 2665

Mihos, J. C. 2004a, in Clusters of Galaxies: Probes of Cosmological Structure andGalaxy Evolution, 278

Mihos, J. C. 2004b, in IAU Symposium, 390

Mihos, J. C., Richstone, D. O., & Bothun, G. D. 1991, ApJ, 377, 72

Miller, R. H. 1986, A&A, 167, 41

Milliard, B., Donas, J., & Laget, M. 1991, Advances in Space Research, 11, 135

Milliard, B., Donas, J., Laget, M., Armand, C., & Vuillemin, A. 1992, A&A, 257, 24

Mobasher, B., Colless, M., Carter, D., et al. 2003, ApJ, 587, 605

Moore, B., Ghigna, S., Governato, F., et al. 1999, ApJ, 524, L19

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

Moore, B., Lake, G., & Katz, N. 1998, ApJ, 495, 139

Mori, M., Ferrara, A., & Madau, P. 2002, ApJ, 571, 40

Morrissey, P., Schiminovich, D., Barlow, T. A., et al. 2005, ApJ, 619, L7

Mulchaey, J. S. 2000, ARA&A, 38, 289

Nagashima, M. & Yoshii, Y. 2004, ApJ, 610, 23

Neff, S. G., Thilker, D. A., Seibert, M., et al. 2005, ApJ, 619, L91

Neistein, E., Maoz, D., Rix, H., & Tonry, J. L. 1999, AJ, 117, 2666

BIBLIOGRAPHY 195

Nulsen, P. E. J. 1982, MNRAS, 198, 1007

O’Connell, R. W. 1999, ARA&A, 37, 603

Okamoto, T. & Habe, A. 1999, ApJ, 516, 591

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

O’Neil, K. 2004, AJ, 128, 2080

Oosterloo, T. & van Gorkom, J. 2005, A&A, 437, L19

Osterbrock, D. E. 1989, Astrophysics of gaseous nebulae and active galactic nu-clei (Research supported by the University of California, John Simon GuggenheimMemorial Foundation, University of Minnesota, et al. Mill Valley, CA, UniversityScience Books, 1989, 422 p.)

Ostriker, J. P. 1980, Comments on Astrophysics, 8, 177

Panuzzo, P., Bressan, A., Granato, G. L., Silva, L., & Danese, L. 2003, A&A, 409, 99

Pinkney, J., Roettiger, K., Burns, J. O., & Bird, C. M. 1996, ApJS, 104, 1

Poggianti, B. 2004a, Baryons in Dark Matter Halos

Poggianti, B. M. 2004b, in Clusters of Galaxies: Probes of Cosmological Structureand Galaxy Evolution, 246

Poggianti, B. M. & Barbaro, G. 1997, A&A, 325, 1025

Poggianti, B. M., Bridges, T. J., Komiyama, Y., et al. 2004, ApJ, 601, 197

Poggianti, B. M., Bridges, T. J., Mobasher, B., et al. 2001, ApJ, 562, 689

Poggianti, B. M., Smail, I., Dressler, A., et al. 1999, ApJ, 518, 576

Quilis, V., Moore, B., & Bower, R. 2000, Science, 288, 1617

Rich, R. M., Salim, S., Brinchmann, J., et al. 2005, ApJ, 619, L107

Rines, K., Geller, M. J., Kurtz, M. J., & Diaferio, A. 2003, AJ, 126, 2152

Roediger, E. & Hensler, G. 2005, A&A, 433, 875

Rubin, V. C., Waterman, A. H., & Kenney, J. D. P. 1999, AJ, 118, 236

Sakai, S., Kennicutt, R. C., van der Hulst, J. M., & Moss, C. 2002, ApJ, 578, 842

196 BIBLIOGRAPHY

Salpeter, E. E. 1955, ApJ, 121, 161

Savage, B. D. & Mathis, J. S. 1979, ARA&A, 17, 73

Saviane, I., Hibbard, J. E., & Rich, R. M. 2004, AJ, 127, 660

Schechter, P. 1976, ApJ, 203, 297

Schindler, S. & Mueller, E. 1993, A&A, 272, 137

Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525

Scodeggio, M., Gavazzi, G., Franzetti, P., et al. 2002, A&A, 384, 812

Seibert, M., Martin, D. C., Heckman, T. M., et al. 2005, ApJ, 619, L55

Shang, Z., Brinks, E., Zheng, Z., et al. 1998, ApJ, 504, L23

Skillman, E. D., Kennicutt, R. C., & Hodge, P. W. 1989, ApJ, 347, 875

Smail, I., Ellis, R. S., Dressler, A., et al. 1997, ApJ, 479, 70

Smail, I., Kuntschner, H., Kodama, T., et al. 2001, MNRAS, 323, 839

Smith, A. M. & Cornett, R. H. 1982, ApJ, 261, 1

Smith, G. P., Treu, T., Ellis, R. S., Moran, S. M., & Dressler, A. 2005, ApJ, 620, 78

Sodre, L. J., Capelato, H. V., Steiner, J. E., & Mazure, A. 1989, AJ, 97, 1279

Somerville, R. S. 2002, ApJ, 572, L23

Springel, V. & Hernquist, L. 2003, MNRAS, 339, 312

Springel, V., White, S. D. M., Jenkins, A., et al. 2005, Nature, 435, 629

Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickinson, M., & Pettini, M. 1999,ApJ, 519, 1

Stein, P. 1997, A&A, 317, 670

Struble, M. F. & Rood, H. J. 1999, ApJS, 125, 35

Sullivan, M., Treyer, M. A., Ellis, R. S., et al. 2000, MNRAS, 312, 442

Sun, M. & Murray, S. S. 2002, ApJ, 576, 708

Tantalo, R., Chiosi, C., Bressan, A., & Fagotto, F. 1996, A&A, 311, 361

BIBLIOGRAPHY 197

Terlevich, A. I., Kuntschner, H., Bower, R. G., Caldwell, N., & Sharples, R. M. 1999,MNRAS, 310, 445

Thuan, T. X. & Sauvage, M. 1992, A&AS, 92, 749

Tonry, J. & Davis, M. 1979, AJ, 84, 1511

Tonry, J. L., Schmidt, B. P., Barris, B., et al. 2003, ApJ, 594, 1

Toomre, A. & Toomre, J. 1972, ApJ, 178, 623

Tornatore, L., Borgani, S., Matteucci, F., Recchi, S., & Tozzi, P. 2004, MNRAS, 349,L19

Trager, S. C., Faber, S. M., Worthey, G., & Gonzalez, J. J. 2000, AJ, 120, 165

Tremonti, C. A., Heckman, T. M., Kauffmann, G., et al. 2004, ApJ, 613, 898

Trentham, N., Sampson, L., & Banerji, M. 2005, MNRAS, 357, 783

Treu, T., Ellis, R. S., Kneib, J.-P., et al. 2003, ApJ, 591, 53

Treu, T., Ellis, R. S., Liao, T. X., & van Dokkum, P. G. 2005, ApJ, 622, L5

Treyer, M., Wyder, T. K., Schiminovich, D., et al. 2005, ApJ, 619, L19

Treyer, M. A., Ellis, R. S., Milliard, B., Donas, J., & Bridges, T. J. 1998, MNRAS,300, 303

Trimble, V. 1995, PASP, 107, 1133

Trujillo, I., Rudnick, G., Rix, H.-W., et al. 2004, ApJ, 604, 521

Vazquez, G. A. & Leitherer, C. 2005, ApJ, 621, 695

van den Bergh, S. 1976, ApJ, 206, 883

van Dokkum, P. G., Franx, M., Kelson, D. D., et al. 1998, ApJ, 500, 714

van Gorkom, J. H. 2004, in Clusters of Galaxies: Probes of Cosmological Structureand Galaxy Evolution, 306

van Zee, L., Salzer, J. J., Haynes, M. P., O’Donoghue, A. A., & Balonek, T. J. 1998,AJ, 116, 2805

van Zee, L., Skillman, E. D., & Haynes, M. P. 2004, AJ, 128, 121

Vogt, N. P., Haynes, M. P., Herter, T., & Giovanelli, R. 2004, AJ, 127, 3273

198 BIBLIOGRAPHY

Vollmer, B., Balkowski, C., Cayatte, V., van Driel, W., & Huchtmeier, W. 2004a,A&A, 419, 35

Vollmer, B., Beck, R., Kenney, J. D. P., & van Gorkom, J. H. 2004b, AJ, 127, 3375

Vollmer, B., Braine, J., Combes, F., & Sofue, Y. 2005, A&A in press, astro-ph/0507252

Vollmer, B., Cayatte, V., Balkowski, C., & Duschl, W. J. 2001, ApJ, 561, 708

Walker, I. R., Mihos, J. C., & Hernquist, L. 1996, ApJ, 460, 121

Wang, B. & Heckman, T. M. 1996, ApJ, 457, 645

Wehner, E. H. & Gallagher, J. S. 2005, ApJ, 618, L21

Weilbacher, P. M., Duc, P.-A., Fritze v. Alvensleben, U., Martin, P., & Fricke, K. J.2000, A&A, 358, 819

West, M. J. & Blakeslee, J. P. 2000, ApJ, 543, L27

West, M. J. & Bothun, G. D. 1990, ApJ, 350, 36

West, M. J., Villumsen, J. V., & Dekel, A. 1991, ApJ, 369, 287

White, S. D. M. & Rees, M. J. 1978, MNRAS, 183, 341

Whitmore, B. C., Gilmore, D. M., & Jones, C. 1993, ApJ, 407, 489

Willman, B., Governato, F., Wadsley, J., & Quinn, T. 2004, MNRAS, 355, 159

Witt, A. N. & Gordon, K. D. 2000, ApJ, 528, 799

Wyder, T. K., Treyer, M. A., Milliard, B., et al. 2005, ApJ, 619, L15

Xilouris, E. M., Byun, Y. I., Kylafis, N. D., Paleologou, E. V., & Papamastorakis, J.1999, A&A, 344, 868

Xu, C. & Buat, V. 1995, A&A, 293, L65

Xu, C. K., Donas, J., Arnouts, S., et al. 2005, ApJ, 619, L11

Yahil, A. & Vidal, N. V. 1977, ApJ, 214, 347

Yi, S., Demarque, P., & Oemler, A. J. 1998, ApJ, 492, 480

Yi, S. K., Yoon, S.-J., Kaviraj, S., et al. 2005, ApJ, 619, L111

BIBLIOGRAPHY 199

Yoshida, M., Ohyama, Y., Iye, M., et al. 2004, AJ, 127, 90

Yoshii, Y. & Arimoto, N. 1987, A&A, 188, 13

Zabludoff, A. I., Geller, M. J., Huchra, J. P., & Ramella, M. 1993, AJ, 106, 1301

Zaritsky, D., Kennicutt, R. C., & Huchra, J. P. 1994, ApJ, 420, 87

Zibetti, S., Gavazzi, G., Scodeggio, M., Franzetti, P., & Boselli, A. 2002, ApJ, 579,261

Zwicky, F., Herzog, E., & Wild, P. 1961, Catalogue of galaxies and of clusters ofgalaxies (Pasadena: California Institute of Technology (CIT))

List of Figures

1.1 An example of the heterogeneous population of galaxies that inhabitour Universe. Mosaic of RGB (g,r,i) images adapted from Frei et al.(1996) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1 Cross section of the instrument portion of GALEX. The optical pathis outlined in blue. Overall dimensions of the view shown are 1.5 m 1m (adapted from Morrissey et al. 2005). . . . . . . . . . . . . . . . . 16

2.2 The transmittance profile for the NUV and FUV GALEX filters. Dif-ferent galaxy spectral energy distributions are superposed. . . . . . . 19

2.3 Example of GALEX image. GALEX NGS observation of NGC4631.In the color table, red-green (gold) is used for NUV, and blue for FUV. 19

3.1 The UV luminosity functions for the four analyzed data sets. . . . . . 273.2 The composite UV luminosity function of 3 nearby clusters. The solid

line represents the best Schechter fit to the data for MUV ≤ −16.5. . . 283.3 The UV bi-variate composite luminosity functions of nearby clusters.

Red (UV −B > 2) and blue (UV − B < 2) galaxies are indicated withempty and filled circles respectively. . . . . . . . . . . . . . . . . . . . 29

3.4 The cluster and the field UV luminosity functions. The compositecluster LF is given with filled circles. The solid line indicates the bestSchechter fit of the field LF of Sullivan et al. (2000). The normalizationis such that the two LFs match at MUV ∼ −19.25. . . . . . . . . . . . 31

4.1 The GALEX observation of Abell1367. ROSAT X-ray contour aresuperposed in black. The tick rectangular region indicates the regioncovered by the optical catalogues used for the star/galaxy discrimination. 34

4.2 Comparison between FOCA (upper image) and GALEX (lower image)observation of the center of Abell1367. It emerges clearly the strongimprovement in resolution and sensitiveness of new GALEX data. . . 35

4.3 Left: The comparison between FOCA and GALEX NUV (left) andFUV (right) magnitudes of galaxies in Abell1367. The continuum lineshows the best linear fit to the data. . . . . . . . . . . . . . . . . . . 36

201

202 LIST OF FIGURES

4.4 The redshift completeness per bin of UV magnitude in Abell 1367. . . 37

4.5 The GALEX NUV (left) and FUV (right) LF for Abell 1367. Opendots are obtained using the subtraction of field counts obtained by Xuet al. (2005); filled dots are obtained using the completeness correctedmethod. The solid line represents the best Schechter fit. The dottedline shows the composite nearby clusters 2000 A LF by (Cortese et al.2003a). The dashed line represents the GALEX local field LF (Wyderet al. 2005), normalized in order to match the cluster LF at MAB ∼−17.80. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.6 The NUV (left) and FUV (right) bi-variate LF of A1367. Star-formingand quiescent galaxies are indicated with empty triangles and filledsquares respectively. The dashed line represents the GALEX localfield LF (Wyder et al. 2005), normalized as in Fig.4.5 . . . . . . . . . 38

4.7 The FUV-NUV color magnitude relation for confirmed members ofA1367. Symbols are as in Fig.4.6 . . . . . . . . . . . . . . . . . . . . 41

4.8 The optical (r′-band) distribution for star forming (blue histogram)and quiescent (red histogram) galaxies in our sample. . . . . . . . . . 41

5.1 Cumulative redshift distribution for galaxies in the studied region. . . 46

5.2 Velocity histogram and stripe density plot for the members of Abell1367. Arrows mark the location of the most significant weighted gapsin the velocity distribution. . . . . . . . . . . . . . . . . . . . . . . . 47

5.3 Local deviations from the global kinematics for galaxies in Abell 1367as measured by the Dressler & Shectman (1988) test. Galaxies aremarked with open circles whose radius scales with their local deviationδ from the global kinematics. The ROSAT X-ray contours are shownwith dotted lines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.4 Palomar DSS image of the central region (∼1.3 square degrees) ofAbell 1367 studied in this Chapter. The iso-density contours for the146 confirmed cluster members are superposed. The lowest iso-densitycontour correspond to 3σ above the mean density in the field (left). TheROSAT X-ray contours are superposed in red (right). The straight lineindicates the position of the abrupt gas temperature gradient detectedby ASCA (Donnelly et al. 1998), used to divide our sample into twosubclusters: the North-West and the South-East. . . . . . . . . . . . 52

5.5 The LOS velocity field (left) and the velocity dispersion field (right)for the whole region studied in this Chapter. The LOS velocity andthe velocity dispersion are computed using the 10 nearest neighbors toeach pixel, whose size is 36 arcsec2. The iso-density contours for the146 confirmed cluster members are superposed in black. . . . . . . . 53

LIST OF FIGURES 203

5.6 A 3D sketch of Abell 1367 summarizing the various sub-componentsdescribed in Section 5.5. The cluster is viewed from its near side, assuggested by the eyeball indicating the observer’s position. . . . . . . 54

5.7 Blow-up of the NW substructure of Abell 1367. The arrows indicatethe direction of radio head tails associated with 97-079 and 97-073 andthe orientation of the NAT radio galaxy 97-095. The dashed regionshows the distribution of the diffuse cluster radio relic (Gavazzi 1978).The iso-density contours for the 146 confirmed cluster members aresuperposed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5.8 The LOS velocity distribution for galaxies in the NW (upper) and inthe SE (lower) subclusters. . . . . . . . . . . . . . . . . . . . . . . . . 56

5.9 The velocity dispersion radial profile of the NW (upper) and the SE(lower) subclusters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.10 The distribution of galaxies belonging to the South-East subcluster.Triangles indicate galaxies with LOS velocity > 7500 km s−1, circlesgalaxies with LOS velocity < 5800 km s−1 and squares galaxies withLOS velocity comprises in the range 5800 km s−1 < V < 7500 km s−1.The ROSAT X-ray contours are shown. . . . . . . . . . . . . . . . . . 58

5.11 The LOS velocity distribution for emission line (upper) and non emis-sion line galaxies (lower) in the whole cluster sample. . . . . . . . . . 59

5.12 Projected density map of non emission line (left) and emission line(right) galaxies in Abell 1367. The iso-density contours of the 146confirmed cluster members are superposed. . . . . . . . . . . . . . . . 60

5.13 The bound and unbound orbit regions in the (Vrel, α) plane. Thebound-incoming solutions (BIa and BIb), the bound-outgoing solu-tions (BO) and the unbound-outgoing (UO) solutions are indicatedwith solid lines. The dotted lines show the dividing line between boundand unbound regions. The vertical solid lines represent the observedVrel and the dashed regions their associated 1σ uncertainty. . . . . . 63

6.1 The near-UV (left column) and far-UV (right column) to optical andnear-IR color magnitude relations. Colors are in the AB magnitudesystem. Open circles are for ellipticals, filled circles for dwarfs, crossesfor lenticulars (S0-S0a). Galaxies redder than the dashed line are un-detectable by the present survey (at the NGS limit). Largest 1σ errorsfor luminous and dwarf systems are given. . . . . . . . . . . . . . . . 72

6.2 The relationship between the UV color index (FUV − NUV ) and a)the total H band luminosity, b) the B-H color index, c) the logarithmof the central velocity dispersion and d) the Mg2 index. Symbols areas in fig. 6.1. Labeled points indicate objects having unusual radio oroptical properties (see Sect. 3). . . . . . . . . . . . . . . . . . . . . . 74

204 LIST OF FIGURES

6.3 The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectraavailable for dwarf ellipticals are presented. . . . . . . . . . . . . . . . 77

6.4 The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectraavailable for ellipticals are presented. . . . . . . . . . . . . . . . . . . 78

6.5 The relationship between the UV color index (FUV − NUV ) and thetotal H band luminosity. Symbols are as in fig. 6.1. The optical spectraavailable for lenticulars are presented. . . . . . . . . . . . . . . . . . . 79

7.1 Ratio of the total infrared to far ultraviolet luminosity as a function ofthe ultraviolet spectral slope (lower x-axis) and the FUV-NUV color(upper x-axis). Open circles indicates our secondary sample while filledcircles represent the primary sample. The dashed line represents thebest linear fit to starburst IRX-UV relation. The solid line indicatesthe best bisector linear fit for our primary sample. The stars indicatethe sample of IUE starbursts. Mean error bars for the plotted data areshown in the lower right corner, in this and subsequent figures. Theresiduals from the best linear fit for normal galaxies are shown in thebottom panel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

7.2 Relation between the birthrate parameter computed from the Hα emis-sion, and the distance from the LTIR/LFUV −β relation for starbursts.The solid line represents the best linear fit. . . . . . . . . . . . . . . . 90

7.3 The relation between the ultraviolet spectral slope β and the Hα at-tenuation obtained from the Balmer decrement. Symbols are as inFig.7.1. Solid line represents the best linear fit to our primary sample(equation 7.14) while the dashed line indicate the best-fit for starburstgalaxies obtained by Calzetti et al. (1994) (equation 7.13). Arrows in-dicate galaxies for which the value of A(Hα) is a lower limit of the realvalue (i.e. Hβ undetected). The residuals from the best linear fit fornormal galaxies are shown in the bottom panel. . . . . . . . . . . . . 93

7.4 Relation between gas metallicity and the LTIR/LFUV ratio (left) or β(right). Symbols are as in Fig.7.1. The solid lines show the best linearfit for our primary sample. The residuals from the best linear fits fornormal galaxies are shown in the upper panels. . . . . . . . . . . . . . 95

7.5 Relation between the galaxy size and the LTIR/LFUV ratio for starburst(left panel) and normal galaxies (right panel). Symbols are as in Fig.7.1. Mean values and uncertainties in bins of 0.30 log(Diameter) aregiven. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

LIST OF FIGURES 205

7.6 Relation between the gas to dust ratio and the LTIR/LFUV ratio (left)or β (right). Symbols are as in Fig. 7.1. The solid line shows the bestlinear fit for our primary sample. . . . . . . . . . . . . . . . . . . . . 97

7.7 Relation between the H-band luminosity and the LTIR/LFUV ratio(left) or β (right). Symbols are as in Fig. 7.1. The solid line showsthe best linear fit for our primary sample. The residuals from the bestlinear fit for normal galaxies are shown in the upper panel. . . . . . . 98

7.8 Relation between the TIR+FUV luminosity and the LTIR/LFUV ratio(left) or β (right). Symbols are as in Fig. 7.1. . . . . . . . . . . . . . 98

7.9 Relation between the mean H-band surface brightness (µe) and theLTIR/LFUV ratio (left) or β (right). Symbols are as in Fig. 7.1. Thesolid line shows the best linear fit for our primary sample. The residualsfrom the best linear fit for normal galaxies are shown in the upper panel.100

7.10 Relation between the star formation rate density and the LTIR/LFUV

ratio (left) or β (right). Symbols are as in Fig. 7.1. The solid lineshows the best linear fit for our primary sample.The residuals fromthe best linear fit for normal galaxies are shown in the upper panel. . 100

7.11 Relation between the Hα and far ultraviolet luminosity and the LTIR/LFUV

ratio (left) or β (right). Symbols are as in Fig. 7.1. Hα luminosity iscorrected for dust attenuation using the Balmer decrement, while theFUV flux is uncorrected. The solid lines show the best linear fit forour primary sample. The residuals from the best linear fit for normalgalaxies are shown in the upper panels. . . . . . . . . . . . . . . . . . 102

7.12 Relation between the observed Hα and far ultraviolet luminosity andthe LTIR/LFUV ratio (left) or β (right). Symbols are as in Fig. 7.1.Hα luminosity is the observed value not corrected for dust attenuation.The solid lines show the best linear fit for our primary sample.Theresiduals from the best linear fit for normal galaxies are shown in theupper panels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

8.1 The combined NUV and FUV image of NGC 4438. The regions de-scribed in sect. 3 of the text are labeled 1 to 7. The horizontal line is10 kpc long (assuming a distance of 17 Mpc). . . . . . . . . . . . . . 109

8.2 The Hα+[NII] contours (red, in arbitrary scale, in between 8 10−17 and6 10−16 erg cm−2 s−2 arcsec−2, with σ= 5 10−17 erg cm−2 s−2 arcsec−2,from Boselli & Gavazzi (2002)) are superposed to the NUV gray-levelimage of NGC 4438. . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

8.3 Chandra image of NGC 4438 in the 0.3-2 keV energy band with Hαcontours superposed. Adapted from Machacek et al. (2004) . . . . . . 111

206 LIST OF FIGURES

8.4 The RGB (FUV=blue, NUV=green, B=red) color map of NGC 4438and NGC 4435. The SED of each region defined in Fig. 1 are given inthe lower plot of each frame. Crosses indicate the observed data, arrowsupper limits (in mJy), the red dashed line the evolved population fitas determined by the model of Boissier & Prantzos (2000), the dottedblue line the starburst SED (from Starburst 99) and the dashed greenline the combined fitting model. The burst luminosity contribution(for the age corresponding to the minimum χ2) in the band FUV, Band K is also given. The upper panel gives the variation of the reducedχ2 parameter (black continuum line, in logarithmic scale) and of theburst mass fraction (red dotted line) as a function of the age of theburst (in Myr). The lower panel of region 4 gives the integrated 3500to 7000 A, R=1000 spectrum of the main body of the galaxy (blackcontinuum line) compared to the fitted model (red dashed line). . . . 115

9.1 The radial profile of observed (open symbols) and extinction-corrected(filled symbols) H-band surface brightness (left) and of the rotationalvelocity (center) used to constrain the model without interaction (rep-resented by the black solid line). The total gas radial profile (right)predicted by the unperturbed model (solid black line) is compared tothe observed one (green filled circles), obtained by summing the HIcomponent (red line) to the molecular one (blue and light blue) andcorrecting for Helium contribution (× 1.4), and to the model includingthe interaction (black dashed line). . . . . . . . . . . . . . . . . . . . 120

9.2 Ram pressure stripping intensity (in arbitrary units) as a function oftime (Eq.9.1). Adapted from Vollmer et al. (2001). . . . . . . . . . . 121

9.3 The radial profile of the observed (empty green circles) and extinction-corrected (filled green circles) total gas, Hα, FUV (1530 A), NUV (2310A), B and i surface brightness. The yellow shaded area marks the rangein between the observed (bottom side) and extinction-corrected (topside) surface brightness profiles. Surface brightnesses are compared tothe model predictions without interaction (black solid line) or with in-teraction for several ε0 and t0 parameters. Equal maximum efficiency(ε0=1.2 M kpc−2 yr−1) and different age: t0=100 Myr, red continuumline (the adopted model); t0=500 Myr, grey long dashed line, t0=1.5Gyr, dashed magenta line. Equal age (t0=100 Myr) and different max-imum efficiency: ε0=3 M kpc−2 yr−1, blue dotted line; ε0=1/3 M

kpc−2 yr−1, orange dotted line. . . . . . . . . . . . . . . . . . . . . . 122

LIST OF FIGURES 207

9.4 The observed and model surface brightness (a), color (b) radial profilesof NGC 4569. In the model profiles the continuum lines are for modelswith gas removal, dashed lines for unperturbed models. c) the reducedχ2 as a function of the look-back time of the ram-pressure event for afew efficiencies ε0 (M kpc−2 yr−1). Models were computed each 100Myr for 0 < t0 < 500 Myr, 200 Myr for 1.5 < t0 < 0.5 Gyr and 1 Gyrfor 6.5 < t0 < 1.5 Gyr; and each 0.2 M kpc−2 yr−1 efficiencies between0.4 and 1.6 (only the more relevant are shown here). d) the variation ofthe effective surface brightness (mean surface brightness within Re, theradius containing half of the total light) and radius due to differentialvariation of the star formation history of NGC 4569. Open trianglesare for the unperturbed model, the other symbols for different ages ofthe interaction (100 Myr, 1.5 and 5.5 Gyr). . . . . . . . . . . . . . . . 125

9.5 The RGB (continuum subtracted Hα =blue, FUV=green, NUV=red)color map of NGC 4569 . . . . . . . . . . . . . . . . . . . . . . . . . 126

10.1 The four Arecibo HI pointings obtained in the region of the BIG group,superposed to the r′ band image. The size of each circle correspond tothe telescope beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

10.2 GALEX NUV image of the Blue Infalling group (BIG). . . . . . . . 130

10.3 High-contrast Hα+[NII] band frame of the BIG group. . . . . . . . . 132

10.4 Upper panel: The position and the width (rectangular areas on theright) of the three slits obtained for CGCG97-125. The slits are super-posed to the Hα + [NII] net image. Lower Panel: The three differentrotations curves obtained for CGCG97-125. Letters indicate the dif-ferent regions as labeled in the upper panel. . . . . . . . . . . . . . . 136

10.5 The low resolution 2D spectrum obtained at ESO/3.6 for the knotsDW3d (left) and DW3e (right), shows a significant difference (∼ 500km s−1)in the velocity of the two knots. . . . . . . . . . . . . . . . . . . . . . 137

10.6 Stellar shells are seen around galaxy 97-125 in the r′ band image ofBIG. No continuum emission is detected from the low brightness trails(except K2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

10.7 Extended low brightness trails appear in the Hα+[NII] NET frame ofBIG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

10.8 HI column density distribution in BIG. Contours are 0.5, 1.0, 2.0, 3.0,4.0, 5.0, and 6.020cm−2. Adapted from Sakai et al. (2002) . . . . . . 144

10.9 HI position-velocity diagram centered on CGCG 97-125. Adapted fromSakai et al. (2002) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

10.10The HI spectra obtained for each pointing. . . . . . . . . . . . . . . 146

208 LIST OF FIGURES

10.11Comparison between the combined HI spectrum obtained from the fourdifferent Arecibo pointings, and the single pointing on the NW trail. Itappears clearly the presence of a low velocity component not associatedto the bright galaxies in BIG. . . . . . . . . . . . . . . . . . . . . . . 147

10.12The relation between Metallicity and B-band Luminosity (with linearbest-fit) for galaxies in nearby clusters (empty circles, adapted fromGavazzi et al. 2004). The triangles mark the mean metallicity obtainedfor the individual knots of BIG. . . . . . . . . . . . . . . . . . . . . . 151

10.13Comparison between the drift-scan integrated (blue) and nuclear (red)spectrum of CGCG97-125. . . . . . . . . . . . . . . . . . . . . . . . . 153

10.14The SED of CGCG97-125, corrected for internal extinction. Nuclearand drift-scan integrated spectra are shown in green. Black circles in-dicate photometric observations and their relative uncertainties. Bestfitting models for the nuclear spectrum (red) and for the starburst com-ponent (blue) are given. The resulting best fitting SED for CGCG97-125 is presented in black. . . . . . . . . . . . . . . . . . . . . . . . . . 155

10.15The star formation history of CGCG97-125 as obtained from the SEDfitting procedure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

10.16The 2D high resolution spectrum (left) and the optical rotation curve(right) of CGCG97-120 . . . . . . . . . . . . . . . . . . . . . . . . . . 156

10.17B-R color map of BIG (Blue = B; Red = R). . . . . . . . . . . . . . . 160

10.18The observed smoothed (step 3) one dimensional spectra. The objectidentification and telescope are labeled on each panel. . . . . . . . . . 161

10.18Continue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

10.18Continue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

11.1 The distribution of the individual HαE.W. measurements in the Virgocluster along the Hubble sequence (small dots) and of the medianEW(Hα) in bins of Hubble type. Error bars are drawn at the 25th and75th percentile of the distribution.Filled symbols represent HI-def< 0.4(unperturbed) objects and open symbols HI-def> 0.4 (HI deficient)galaxies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

11.2 The star formation rate as a function of density, comparing groups ofgalaxies with clusters. The upper and lower horizontal dashed linesshow the 75% percentile and the median of the equivalent widths.The hashed region shows the relation for the complete sample, whilethe solid line shows the relation for systems with 500 km s−1 < σ <1000 km s−1 (left) and σ < 500 km s−1 (right). The dependence onlocal density is identical irrespective of the velocity dispersion of thewhole system. Figure taken from Bower & Balogh (2004). . . . . . . . 171

LIST OF FIGURES 209

11.3 The ratio of the isophotal Hα and r′ radii as a function of the HIdeficiency for galaxies in the Virgo cluster. . . . . . . . . . . . . . . . 172

11.4 The clustercentric radial distribution of the individual EW(Hα) mea-surements in the Virgo cluster. High and low (B-band) luminositygalaxies are given with open and filled dots respectively. Median inbins of 0.5 R/RV ir are given. Error bars mark the 25th and 75th per-centile of the distribution. . . . . . . . . . . . . . . . . . . . . . . . . 173

List of Tables

2.1 Selected Performance Parameters (Morrissey et al. 2005) . . . . . . . 17

3.1 Integral redshift completeness in bin of 0.5 magnitudes. . . . . . . . . 263.2 The completeness-corrected differential number of galaxies per bin of

magnitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.1 Best Fitting Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . 37

5.1 The spectrograph characteristics . . . . . . . . . . . . . . . . . . . . . 455.2 1D substructure indicators for the whole cluster sample . . . . . . . . 485.3 The most significant weighted gaps detected in the velocity distribution

of the whole cluster sample. . . . . . . . . . . . . . . . . . . . . . . . 495.4 3D substructure indicators for our sample . . . . . . . . . . . . . . . 505.5 Mass estimate for Abell 1367 . . . . . . . . . . . . . . . . . . . . . . . 615.6 Two-body model parameters . . . . . . . . . . . . . . . . . . . . . . . 655.7 The 119 new redshift measurements . . . . . . . . . . . . . . . . . . . 675.7 Continue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

6.1 Main relations for early type galaxies . . . . . . . . . . . . . . . . . . 73

7.1 Linear realtions useful to estimate the LTIR/LFUV ratio (log(LTIR/LFUV ) =a × x + b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

10.1 Redshifts of the galaxies in the BIG group. . . . . . . . . . . . . . . 13310.2 Line fluxes, corrected for internal extinction, of the galaxies in the BIG

group. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13510.3 Properties of galaxies in BIG. . . . . . . . . . . . . . . . . . . . . . . 13810.4 Metallicities of the galaxies in the BIG group. . . . . . . . . . . . . . 15010.5 Best-fitting parameters for the nuclear and starburst component of

CGCG97125. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

211

Documents

ENVIRONMENTAL EFFECTS ON GALAXY EVOLUTION IN …astronomy.swin.edu.au/~lcortese/PhD_Thesis_files/main.pdf · (b) Comprendere le propriet a delle polveri interstellari respon- sabili