Early-type galaxies in different environments: an HI view
Tom Oosterloo, Raffaella Morganti, Alison Crocker, Eva Juette, Michele Cappellari, Tim de Zeeuw, Davor Krajnovic, Richard McDermid, Harald Kuntschner, Marc Sarzi, Anne-Marie Weijmans
aa r X i v : . [ a s t r o - ph . C O ] J u l Mon. Not. R. Astron. Soc. , 1–15 (2010) Printed 10 April 2018 (MN L A TEX style file v2.2)
Early-type galaxies in different environments: an H I view Tom Oosterloo , ⋆ , Raffaella Morganti , , Alison Crocker , † , Eva J ¨utte , ,Michele Cappellari , Tim de Zeeuw , , Davor Krajnovi´c , , Richard McDermid ,Harald Kuntschner , Marc Sarzi , Anne-Marie Weijmans , ASTRON - Netherlands Institute for Radio Astronomy, Postbus 2, 7990 AA Dwingeloo, The Netherlands Kapteyn Astronomical Institute, University of Groningen Postbus 800, 9700 AV Groningen, The Netherlands Sub-department of Astrophysics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford OX1 3RH Department of Astronomy, University of Massachusetts, 710 North Pleasant Street Amherst, MA 01003-9305 USA Astronomisches Institut, Ruhr-Universit¨at Bochum, Universit¨atsstrasse 150, D-44801 Bochum, Germany European Southern Observatory, Karl-Schwarzschild Strasse 2, 85748 Garching bei M¨unchen, Germany Sterrewacht Leiden, Universiteit Leiden, Niels Bohrweg 2, 2333 CA Leiden, The Netherlands Gemini Observatory, 670 N. A’ohoku Place, Hilo, Hawaii 96720 USA Space Telescope European Coordinating Facility, Karl-Schwarzschild-Str. 2, D-85748 Garching bei M¨unchen, Germany Centre for Astrophysics Research, Science & Technology Research Institute, University of Hertfordshire, Hatfield, United Kingdom Dunlap Institute for Astronomy & Astrophysics, University of Toronto, 50 St. George Street, Toronto, ON M5S 3H4, Canada
Accepted 2010 July 12. Received 2010 July 7 ; in original form 2010 March 5
ABSTRACT
We present an analysis of deep Westerbork Synthesis Radio Telescope observations of theneutral hydrogen in 33 nearby early-type galaxies selected from a representative sample stud-ied earlier at optical wavelengths with the
SAURON integral-field spectrograph. This is thedeepest homogeneous set of H I imaging data available for this class of objects. The samplecovers both field environments and the Virgo cluster. Our analysis shows that gas accretionplays a role in the evolution of field early-type galaxies, but less so for those in clusters.The H I properties of SAURON early-type galaxies strongly depend on environment. Fordetection limits of a few times M ⊙ , H I is detected in about 2/3 of the field galaxies,while <
10% of the Virgo objects are detected. In about half of the detections, the H I formsa regularly rotating disc or ring. In many galaxies unsettled tails and clouds are seen. AllH I discs have counterparts of ionised gas and inner H I discs are also detected in moleculargas. The cold ISM in the central regions is dominated by molecular gas ( M H /M HI ≃ ).Assuming our sample is representative, we conclude that accretion of H I is very commonfor field early-type galaxies, but the amount of material involved is usually small and theeffects on the host galaxy are, at most, subtle. Cluster galaxies appear not to accrete H I , orthe accreted material gets removed quickly by environmental effects. The relation between H I and stellar population is complex. The few galaxies with a significant young sub-population allhave inner gas discs, but for the remaining galaxies there is no trend between stellar populationand H I properties. A number of early-type galaxies are very gas rich, but only have an oldpopulation. The stellar populations of field galaxies are typically younger than those in Virgo.This is likely related to differences in accretion history. There is no obvious overall relationbetween gas H I content and global dynamical characteristics except that the fastest rotatorsall have an H I disk. This confirms that if fast and slow rotators are the result of differentevolution paths, this is not strongly reflected in the current H I content. In about 50% of thegalaxies we detect a central radio continuum source. In many objects this emission is froma low-luminosity AGN, in some it is consistent with the observed star formation. Galaxieswith H I in the central regions are more likely detected in continuum. This is due to a higherprobability for star formation to occur in such galaxies and not to H I -related AGN fuelling. Key words: galaxies: elliptical and lenticular, cD — galaxies: evolution ⋆ E-mail:[email protected] † ASTRON/JIVE Summer Student 2007c (cid:13)
Tom Oosterloo et al.
One of the central topics in current extra-galactic astronomy is howearly-type galaxies form and how their properties change throughcosmic times. This is a particularly difficult task as this class ofobjects shows a diversity in its properties that goes beyond present-day simulations. The generally accepted framework is that early-type galaxies form in a hierarchical way through the accretion andmerging of smaller systems. Although hierarchical growth in it-self is a relatively simple premise, the details of early-type galaxyformation and evolution are very complex and depends on a largenumber of parameters which is likely the reason for the observedvariety of the final galaxies.One of the main issues is how much gas is involved in theaccretion/merging process and to what extent it involves approxi-mately equal-mass systems or the accretion of small companions.Theoretical work has shown that the amount of gas involved inthe growth of early-type galaxies can be a major factor, in partic-ular in determining the morphological and dynamical structure ofearly-type galaxies. For example, more anisotropic and slowly ro-tating galaxies would result from predominantly collisionless ma-jor mergers, while faster rotating galaxies are produced by moregas-rich mergers and accretions. The inner structure (i.e. cores vscusps) is also likely related to the type of merger/accretion (e.g.Bender, Burstein, & Faber 1992; Jesseit, Naab, & Burkert 2005;Naab, Jesseit, & Burkert 2006; Hopkins et al. 2009; Jesseit et al.2009).Some arguments suggest that the evolution of early-typegalaxies is ’dry’, i.e. gas does not play an important role. The ba-sic argument is that the amount of stars in red galaxies has at leastdoubled since z = 1 , while the red colours of early-type galaxiesindicate that they are dominated by old stars. This indicates that thegrowth since z = 1 is dry, i.e. it is not accompanied with much starformation (e.g. Bell et al. 2004; van Dokkum 2005; Tal et al. 2009).While this may suggest that globally the amount of gas in-volved since z = 1 is at most modest, several observational studies,touching on several topics, show that gas does play at least somerole. For example, stellar-population studies show that many sys-tems do contain a (often small) sub-population of relatively youngstars that may have formed from accreted gas (e.g. Trager et al.2000; Tadhunter et al. 2005; Yi et al. 2005; Serra et al. 2006; Serraet al. 2008; Kaviraj et al. 2010). Similarly, early work by, e.g., Ma-lin & Carter (1983) and Schweizer & Seitzer (1992), has shownthat direct morphological signs of accretion are observed in a largefraction of early-type galaxies and that such signs correlate withthe presence of a young stellar sub-population, indicating the pres-ence of gas in these accretions. More recent work has shown thatthis is also the case for samples of early-type galaxies that origi-nally seemed to support the dry-merging hypothesis (Donovan, Hi-bbard, & van Gorkom 2007; S´anchez-Bl´azquez et al. 2009; Serra& Oosterloo 2010). Dynamically distinct stellar and gaseous sub-components are often found in early-type galaxies. In many cases,such sub-components are both chemically and kinematically dis-tinct (McDermid et al. 2006), strongly suggesting that external gashas entered the system. The orbital structure of fast-rotating early-type galaxies also seems to indicate that gas was involved in theirevolution (Emsellem et al. 2007; Cappellari et al. 2007). Finally, thetight scaling relations found for early-type galaxies place a ratherconservative upper limit on the fraction of stellar mass assembledvia dissipationless merging (e.g. Nipoti, Londrillo, & Ciotti 2003;Nipoti, Treu, & Bolton 2009).Recently, also direct evidence for the importance of gas has been found. Early-type galaxies in the nearby Universe used to begenerally perceived to be gas poor. Although indeed they typicallyhave less cold gas than spiral galaxies, it is now becoming clearthat cold gas is present perhaps most of them, in particular thosein the field. For example, for a detection limit of a few × M ⊙ ,molecular gas is detected in at least a quarter of early-type galax-ies (Welch & Sage 2003; Sage, Welch, & Young 2007; Combeset al. 2007; Young et al. 2010). Recent work on the neutral hydro-gen in early-type galaxies suggests that, in terms of detection limitsof a few times M ⊙ , about half the field early-type galaxiesare detected (Morganti et al. 2006; Grossi et al. 2009; Serra et al.2009). The H I datacubes obtained by Morganti et al. (2006) alsoshow, in terms of the characteristics of the neutral hydrogen de-tected, the class of field early-type galaxies appears to be rich andvaried, much more so than spirals.An interesting aspect is that early-type galaxies in clusters ap-pear to have different gas properties from those in the field (diSerego Alighieri et al. 2007; Serra et al. 2009). Therefore, if gasplays a role in the evolution of early-type galaxies, this should be-come visible by comparing properties of cluster early-type galaxieswith those of objects in the field.In this paper we expand on the results obtained in Morgantiet al. (2006) with particular focus on the effect of environment. Wepresent deep H I imaging observations for 22 galaxies selected fromthe SAURON sample (de Zeeuw et al. 2002) where we have, in con-trast to Morganti et al. (2006), also selected galaxies from the Virgocluster. Combining these new data with those of Morganti et al.(2006) resulted in deep H I data for 33 SAURON galaxies north ofdeclination +10 ◦ . This is the largest and deepest collection of H I imaging data available for early-type galaxies. The SAURON sam-ple is well suited for a detailed comparison between field and clus-ter galaxies: for all
SAURON galaxies a wealth of information isavailable, including 3-D spectroscopy of the stars and of the ionisedgas as well as data obtained in many other wavebands. A potentialconcern is that in principle the H I properties were part of the selec-tion of the SAURON sample. However, in practise this affected theselection of only a few objects of the
SAURON sample of 48 early-type galaxies. Importantly, the selection was done independent ofenvironment. Therefore, results based on a comparison of the H I properties of SAURON galaxies in different environments shouldbe robust.The paper is organised in the following way. We describe thesample selection and the observations in Section 2. In Section 3 wepresent the results for the new H I detected galaxies. In Section 4we discuss the observed H I properties of early-type galaxies in thecontext of their evolution. I OBSERVATIONS
We observed 22 galaxies from the
SAURON sample (de Zeeuwet al. 2002) with the Westerbork Synthesis Radio Telescope(WSRT). For one of these galaxies (NGC 4486/M87), the verystrong radio continuum emission prevented us to obtain a data cubeof good enough quality. This object is, therefore, excluded fromthe analysis. As mentioned above, the newly observed galaxies ex-pand the sample presented in Morganti et al. (2006) with objectsdown to the declination limit of +10 ◦ and including galaxies thatare member of the Virgo cluster. Lowering the declination limit isa compromise between increasing the sample size while maintain-ing good image quality. For galaxies close to the declination limit,due to the east-west layout of the WSRT, the beam elongation be- c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view NGC V centr D pc/ ′′ Date Int.Time Beam Noise H I Noise Cont. H I contourskm s − Mpc h ′′ × ′′ ( ◦ ) mJy beam − mJy beam − cm − (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)524 2379 23.3 113 ×
12 85 × ×
12 72 × ×
12 43 × × × × ×
12 76 × ×
12 62 × × × × × × × × × × × × × × Table 1.
Summary of the observations of the galaxies in the sample. (1) Galaxy identifier. (2) Systemic velocity at which we centred the H I observation band.(3) Galaxy distance (Tonry et al. 2001, corrected by subtracting 0.06 mag, see Mei et al. (2005)), Tully (1988) or from the LEDA database assuming a Hubbleflow with H ◦ = 75 km s − Mpc − . (4) Linear scale. (5) Date of observation. (6) Integration time in hours. (7) Beam size. (8) Noise level in the H I cube(natural). (9) Noise level of the continuum image. (10) Contour levels of the total intensity images shown in Fig. 1. comes significant, and spatially resolved information becomes lim-ited. However, given that most of the H I structures detected so farare very extended and of low surface brightness, the large WSRTbeam is not a major disadvantage. For the observations, a singleobserving band of 20 MHz (corresponding to ∼ − ), cen-tred on the systemic velocity of the target, and 1024 channels forboth polarisations was used. All observations were done with themaxi-short antenna configuration. In most cases, we observed thetargets for 12 h. For five objects that turned out to have interestingbut faint H I detections, or for which the single 12-h observationgave a tentative detection, follow up observations ( × h) wereobtained. The details of the observations are given in Table 1.The calibration and analysis were done using the MIRIAD package (Sault, Teuben, & Wright 1995). The data cubes wereconstructed with a robust weighting equal to 0 (Briggs 1995).In the cases where faint, extended structures were detected (e.g.NGC 3489), additional cubes were constructed using naturalweighting. The cubes were made by averaging channels in groupsof two, followed by Hanning smoothing, resulting in a velocity res-olution of 16 km s − . This was done to match the spectral reso-lution to the expected line widths. The r.m.s. noise and restoringbeam sizes of each cube are given in Table 1. For reference: a beamsize of 1 arcminute corresponds to about 5 kpc for a galaxy at adistance of 15 Mpc.The line-free channels were used to obtain an image of the ra-dio continuum of each galaxy. The continuum images were madewith uniform weighting. The r.m.s. noise and beam of these im-ages are also given in Table 1. Radio continuum emission was notdetected in 14 of the objects. All detected continuum sources areunresolved. Neutral hydrogen is detected in our new observations in, or near,six galaxies. Some details about the individual detections are givenat the end of this section. Five of the detections are of H I in fieldearly-type galaxies (out of 8 field galaxies observed), one is in agalaxy that is a member of the Virgo cluster (out of 13 Virgo galax-ies observed). Figure 1 shows the H I total intensity images of thedetections. In the strongest detections - NGC 3032, NGC 3489,NGC 4262 - the H I is located in a regular disc/ring-like structure,which in the case of NGC 3489 connects to a long low-surfacebrightness tail of H I .In NGC 3608, the H I is detected about ∼ arcmin (cor-responding to ∼ kpc) from the galaxy, at a velocity close tothe systemic velocity of NGC 3608 and without any obvious opti-cal counterpart. A few other (also early-type) galaxies are nearbyso it is difficult to assign the H I to NGC 3608 unambiguously.A fifth H I detection is NGC 3384. Together with NGC 3377 andNGC 3389, this galaxy is surrounded by the well-known Leo Ring(Schneider et al. 1983; Schneider 1989) of which we conclude thatat least one H I cloud is likely associated with NGC 3384. Finally,in NGC 524 we detect a small H I cloud near the edge of the op-tical body. The results for the field galaxies, with regard to bothmorphology and detection rate, are in line with those we obtainedin our previous study of SAURON galaxies (Morganti et al. 2006).A clear result is that Virgo early-type galaxies clearly havedifferent H I properties than field galaxies. This is discussed inmore detail in Secs 4.1 and 4.6. For galaxies where no H I wasdetected, the upper limits on the H I mass were calculated as threetimes the statistical error of a signal with a width of 200 km s − over one synthesised beam. The upper limits of the H I mass rangefrom a few times M ⊙ to × M ⊙ . The quantitative results c (cid:13) , 1–15 Tom Oosterloo et al.
Figure 1.
Total H I intensity images (contours) superimposed onto Digital Sky Survey optical images of the newly detected objects - NGC 524, NGC 3032,NGC 3384 (part of the Leo Ring), NGC 3489, NGC 3608 and NGC 4262. The contour levels are given in Table 1. The horizontal bar in each panel indicates10 kpc. Total H I intensity image for NGC 3384 has been made excluding the high-velocity system belonging to ¡NGC 3389 (see text)c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view NGC Type M HI M HI /L B H I morph Env S . log P . M ⊙ mJy W/Hz(1) (2) (3) (4) (5) (6) (7) (8)524 S0 . × < . × < . N F < . < . × < . × < . N F < . < < . × < . N F 0.87 19.023384 S0 . × < . < . × . × . × < . × < . N C 1500 22.84382 S0 < . × < . N C < . < < . × < . N C < . < < . × < . N C < . < < . × < . N C 1.52 19.654473 E < . × < . N C < . < < . × < . N C 1.16 19.554550 S0 < . × < . N C < . < < . × < . N C 84.1 21.384564 E < . × < . N C < . < < . × < . N C < . < < . × < . N C < . < . × < . × < . N F < . < . . × . × . × . × . × . × < . × < N F < . < . . × . × < . < . < . × < . N F < . < . Table 2.
Measurements based on our radio observations. The top part of this table is based on the observations presented here. For completeness, we includethe parameters from Morganti et al. (2006). (1) Galaxy identifier. (2) Hubble type (NED). (3) Total H I mass. (4) Ratio of total H I mass and the absolute B -band luminosity L B . (5) Code describing H I morphology: N: not detected, C: isolated cloud, A: accretion, D: disc (6) Environment code: F: field, C: Virgocluster (7) Continuum flux (or σ upper limits) at 1.4 GHz. (8) Total radio power at 1.4 GHz. are given in Table 2. Since we will discuss the H I results for all SAURON galaxies observed, we also include the parameters of thegalaxies observed by Morganti et al. (2006).The H I masses of the newly detected objects range be-tween M ⊙ to a few times M ⊙ . The relative gas content( M HI /L B ) ranges from < . M ⊙ /L ⊙ to . M ⊙ /L ⊙ .Typical values for M HI /L B for spiral galaxies, depending on typeand luminosity, range from 0.1 M ⊙ /L ⊙ to above 1.0 M ⊙ /L ⊙ (Roberts & Haynes 1994). These values underline the well-knownfact that early-type galaxies are H I poor relative to spiral galaxies.The sizes of the H I structures observed in this sample vary betweena few kpc up to ∼
40 kpc. Figure 1 shows that the peak column den-sity istypically at most a few times 10 cm − . As already found inmany earlier studies (e.g. van Driel & van Woerden 1991; Morgantiet al. 1997, 2006; Serra et al. 2006; Oosterloo et al. 2007), thesecolumn densities are lower than the critical surface density for starformation (Kennicutt 1989; Schaye 2004; Bigiel et al. 2008). Al-though this result implies the absence of widespread star formationactivity, given the relatively low spatial resolution of our data, the column density will be above the star formation threshold in local,small regions and some star formation can be expected.Before we further discuss the possible implications of our H I observations, we summarise the H I characteristics for the individ-ual objects detected.NGC524 - We detect a small H I cloud in the outer regions ofthis galaxy. Possibly this corresponds to a small gas-rich compan-ion, although no direct optical counterpart is visible on fairly deepoptical images (Jeong et al. 2009). A small galaxy, of which theredshift is not known, is seen about 1 arcmin from the H I cloud.Possibly a small companion is stripped from its H I by NGC 524,in a similar way as is occurring in, e.g., NGC 4472 (McNamara etal. 1994). NGC 524 has a weak disc of ionised gas which rotateswith the same sense as the stars (Sarzi et al. 2006). This disc is alsodetected in CO with an implied molecular gas mass of . × M ⊙ (Crocker et al. unpublished). No counterpart to this gas disc isdetected in H I , implying that most of the cold ISM in the centralregions of NGC 524 is in the form of molecular gas.NGC3032 - In this galaxy, a small regularly rotating H I discis found, with a total H I mass of . × M ⊙ . The H I disc c (cid:13)000
40 kpc. Figure 1 shows that the peak column den-sity istypically at most a few times 10 cm − . As already found inmany earlier studies (e.g. van Driel & van Woerden 1991; Morgantiet al. 1997, 2006; Serra et al. 2006; Oosterloo et al. 2007), thesecolumn densities are lower than the critical surface density for starformation (Kennicutt 1989; Schaye 2004; Bigiel et al. 2008). Al-though this result implies the absence of widespread star formationactivity, given the relatively low spatial resolution of our data, the column density will be above the star formation threshold in local,small regions and some star formation can be expected.Before we further discuss the possible implications of our H I observations, we summarise the H I characteristics for the individ-ual objects detected.NGC524 - We detect a small H I cloud in the outer regions ofthis galaxy. Possibly this corresponds to a small gas-rich compan-ion, although no direct optical counterpart is visible on fairly deepoptical images (Jeong et al. 2009). A small galaxy, of which theredshift is not known, is seen about 1 arcmin from the H I cloud.Possibly a small companion is stripped from its H I by NGC 524,in a similar way as is occurring in, e.g., NGC 4472 (McNamara etal. 1994). NGC 524 has a weak disc of ionised gas which rotateswith the same sense as the stars (Sarzi et al. 2006). This disc is alsodetected in CO with an implied molecular gas mass of . × M ⊙ (Crocker et al. unpublished). No counterpart to this gas disc isdetected in H I , implying that most of the cold ISM in the centralregions of NGC 524 is in the form of molecular gas.NGC3032 - In this galaxy, a small regularly rotating H I discis found, with a total H I mass of . × M ⊙ . The H I disc c (cid:13)000 , 1–15 Tom Oosterloo et al. co-rotates with the ionised gas observed (Sarzi et al. 2006), al-though the detail in the observed velocity field of the ionised gasis limited by binning effects. The molecular gas is found in a cen-trally concentrated rotating structure (Young, Bureau, & Cappel-lari 2008). The ionised, molecular and neutral gas are co-rotating.Interestingly, all these gaseous components are counter-rotating with respect to the bulk of the stars in this galaxy, strongly sug-gesting an external origin of the gas observed. Some stars haveformed from this gas disc because McDermid et al. (2007) havefound the presence of a small stellar core that is counter-rotatingto the bulk of the stellar body (and hence co-rotating with the gasdisc). The CO observations reveal a molecular gas reservoir of . - . × M ⊙ (Combes et al. 2007; Young, Bureau, & Cappellari2008). Thus, also in NGC 3032 most of the cold ISM is in themolecular phase.NGC3377/3379/3384–TheLeoGroup- The situation in NGC3384 is very complex. This galaxy is member of the M96 groupthat is famous for its large H I ring encircling several galaxies ofthis group (the Leo Ring; Schneider et al. 1983). Subsequent to theobservations reported here, we have imaged the H I over the entireregion and a full study of the Leo Ring will be published elsewhere.One result of this work is that the Leo Ring appears to form a largespiral-like structure that appears to end very close, both in spaceand in velocity, to NGC 3384. At the endpoint of this spiral-likestructure a fairly bright H I cloud is observed and this is the cloudwe identify here with NGC 3384. However, it is conceivable thatmost of the H I of the Leo Ring originated from NGC 3384 (seeMichel-Dansac et al. 2010). A complication is that the H I spiralappears to be interrupted by an interaction of the spiral galaxy NGC3389 with the Leo Ring, modifying its structure. NGC 3389 has along tail of H I observed at velocities about 600 km s − redshiftedwith respect to the H I shown in Fig. 1, but shows no optical signsof a tidal interaction.NGC3489 - In NGC 3489 we find an inner rotating H I struc-ture aligned with the galaxy, as well as a low-column density tail.The total H I mass of this barred galaxy is . × M ⊙ . Thisgalaxy might be in the process of accreting a small gas cloud orcompanion, and forming an inner disc from this material. The cen-tral H I structure shows regular rotation with the same sense of ro-tation as the ionised gas and the stellar component. Molecular gaswith a mass of . × M ⊙ has been detected in this galaxy byCombes et al. (2007).NGC3608- In the vicinity of NGC 3608 we detect a large H I structure, about 40 kpc in size and about 70 kpc from the galaxy. NoH I was detected on NGC 3608 itself. A smaller H I cloud is alsodetected, closer to NGC 3608. Since the galaxy is a member of aloose group, the intergalactic H I clouds might reflect past interac-tions between the group members, but none of the nearby groupgalaxies hosts H I today. This field shows similarities with the el-liptical galaxy NGC 1490 (Oosterloo et al. 2004) where a numberof large H I clouds (with a total H I mass of almost M ⊙ ) areobserved that are lying along an arc 500 kpc in length and at a dis-tance of 100 kpc from NGC 1490. The stars and the ionised gasin NGC 3608 are kinematically decoupled, at least in the centre,where the stars show a regular rotation pattern. No CO was foundin this galaxy in the survey of Combes et al. (2007).NGC4262 - This strongly barred galaxy is the only object ina dense environment in which we detect in H I . The cold gas isdistributed in a large ring. This ring shows regular rotation, albeitwith signs of non-circular orbits. The observed structure is mostlikely a resonance ring due to the bar. The ionised gas rotates in Figure 2. H I velocity field of NGC 4262. Iso-velocity contours run from1160 km s − (top left) to 1510 km s − (bottom right) in steps of 25 kms − the same sense as the H I whereas the stellar rotation is decoupledfrom that. No CO was detected by Combes et al. (2007). Adding together the data from Morganti et al. (2006) and thepresent work, we have data cubes of 33 early-type galaxies fromthe representative
SAURON sample of 48. Using this sample, weexpand the analyses done in Morganti et al. (2006). In particular,we investigate here also the relation with the CO observations thathave become recently available as well as the effect of environmenton the presence of neutral hydrogen. We discuss the effect of theenvironment in Sec. 4.1 while in the rest of the discussion we fo-cus on field galaxies, where the majority of the H I detections arefound. It is well established that the H I properties of spirals in clusters arestrongly affected by the dense environment. Overall, spiral galax-ies in clusters are deficient in H I compared to field spirals (e.g.Giovanardi, Krumm, & Salpeter 1983; Solanes et al. 2001, and refstherein), while imaging studies show that the H I discs in spiralsin clusters are clearly affected by the dense environment (Cayatteet al. 1990; Oosterloo & van Gorkom 2005; Chung et al. 2009).The above studies have shown that dynamical interactions betweengalaxies, as well as stripping by the hot intracluster medium, areresponsible for removing a large fraction of the neutral gas fromcluster spirals. These mechanisms affect mainly the outer disc re-gions, because the inner gas discs in cluster galaxies have similarproperties as those in field galaxies (Kenney & Young 1988; Younget al. 2010). Another important fact is that in the Virgo cluster thepopulation of small, gas-rich galaxies that can be accreted by largergalaxies in smaller than in the field (Kent 2010). Therefore, galax-ies entering a cluster not only lose gas, but they are also not ableto replenish their gas supply by accreting companions. Thereforecluster galaxies become gas poor and remain so and will evolveinto early-type galaxies. c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view Galaxy density parameter N u m b e r o f g a l a x i e s DetectionsNon-detections
Figure 3.
Distribution of the estimate of the local galaxy density as given inthe NCG for galaxies detected and not detected in H I . Galaxies with galaxydensity above 2 are member of the Virgo cluster Given the above, one might expected that the H I propertiesof early-type galaxies strongly depend on the local galaxy density,even more so than for spirals. Observational evidence that this isindeed the case comes from studies based on the Arecibo LegacyFast ALFA survey (ALFALFA Giovanelli et al. 2005). di SeregoAlighieri et al. (2007) and Grossi et al. (2009) have used this sur-vey to select early-type galaxies located in the Virgo cluster andin low density environments respectively. di Serego Alighieri et al.(2007) found a detection rate of only 2% for early-type galaxies thatare member of the Virgo cluster. On the other hand, the detectionrate for the early-type galaxies in low density environments (Grossiet al. 2009) is about ten times higher (25%). A similar contrast inH I properties has been observed by Serra et al. (2009). Althougha simple division into cluster and field galaxies does not do justiceto the wide range of H I properties observed in field galaxies (seebelow), our data, that moreover have a noise level about a factor 4better than the ALFALFA data, strongly confirm the large differ-ence in detection rates between galaxies in low and in high densityenvironments. To illustrate this, we have divided our sample intocluster and field sub-samples and for the moment we ignore thewide range of the H I properties observed. To quantify the environ-ment of our sample galaxies, we have used the estimates of theirlocal galaxy densities given in the Nearby Galaxy Catalog (NGC,Tully 1988). This catalog gives, on a spatial grid of 0.5 Mpc, thedensity of galaxies brighter than –16 mag. Applying this informa-tion to our sample clearly separates it into low- and high-densitysub-samples (see Fig. 3). The former sub-sample we will refer to asthe field sample and the latter corresponds to galaxies in the Virgocluster.Figure 3 shows the distribution of the local galaxy density forgalaxies not detected in H I and for galaxies where at least someH I was detected in or near the galaxy. This figure shows that onlyone object (NGC 4262) out of 13 cluster galaxies (8%) has beendetected in H I , while in, or near, 14 of the 20 non-cluster galaxies(70%) at least some H I is detected, albeit with a large variety of H I properties ranging from small, off-centre clouds to large, regularH I discs. It is clear that, even with more sensitive observations, itis much more likely to detect H I in or near field galaxies than in ornear Virgo early-type galaxies.An important observation is that, as for Virgo spirals, the de-tection rate of molecular gas in early-type galaxies shows a much weaker environmental dependence (Combes et al. 2007; Young etal. 2010). Since these CO observations mostly refer to the centralregions, this suggest that in early-type galaxies, gas removal, as incluster spirals, occurs predominantly from their outer regions. Thesimilarity of the effects of stripping suggest that gas removal is dueto the same mechanism. The observation that the H I detection rateof early-type galaxies seems to be more affected than that of spi-rals is consistent with the fact that early-type galaxies form a morerelaxed Virgo population than Virgo spirals (Binggeli, Tammann,& Sandage 1987). The different dynamical properties of the twopopulations suggest that early-type galaxies have been member ofthe Virgo cluster for a longer period and therefore may have suf-fered more from the environmental effects that remove gas fromgalaxies. Another effect can be that, compared to spirals, the H I in early-type galaxies is more often found in the outer regions ofthe galaxies so that it is more easily affected by interactions in thecluster.Nevertheless, some of our detections in the field, e.g. NGC3032 and NGC 3489, have their H I in an inner disc that could sur-vive in the dense environment of Virgo. Still, such H I discs are notseen in our observations of Virgo galaxies. This could be explainedif such inner discs in field galaxies are the remnants of relativelyrecent accretions (see also next section). The flat slope of the faint-end of the H I mass function observed for the Virgo cluster (Kent2010) indicates that the population of small gas-rich objects, i.e.those objects that appear to supply field early-type galaxies withfresh gas, is smaller in the Virgo cluster than in the field. There-fore, once a Virgo early-type galaxy loses its gas through environ-mental effects, there is much less chance that a new supply of gasis accreted and once a Virgo early-type galaxy is H I poor, it willlikely remain so, unlike galaxies in the field. Even if a gas-rich ob-ject accreted, interactions with the high density environment willmore likely remove such gas during the accretion while it is stillonly loosely bound to the galaxy. That accretions in clusters areless gas rich is also observed for shell galaxies by Hibbard & San-som (2003). The low gas accretion rate for cluster galaxies is alsosuggested by the observation that in several field galaxies we ob-serve ongoing accretion, in stark contrast with the cluster galaxies,although the dense cluster environment implies a shorter time overwhich direct signs of interaction are visible. A lower detection rateof direct signs of ongoing interaction in clusters is also observedin other wave bands (e.g., Tal et al. 2009). Both gas removal andthe lack of new gas supplies drive the morphological evolution ofgalaxies in the Virgo cluster from late type to early type. I in field early-type galaxies The results of the previous section show that there is a large differ-ence in overall H I properties between cluster and field early-typegalaxies. However, as many earlier studies have shown, within thegroup of field early-type galaxies there is a very large range in H I properties and the situation with respect to H I in early-type galax-ies is much more complex than just a simple division into clusterand field. Here we discuss this in some more detail.Combining the results from Morganti et al. (2006) and thepresent work, we have deep H I images for 20 field early-typegalaxies selected from the SAURON sample of 48. For 13 of theseobjects, we have detected H I in or near them. Figure 1, togetherwith Fig. 1 from Morganti et al. (2006), shows that there is a verylarge range in properties of the H I structures in early-type galax-ies, ranging from a single small cloud, to large, regular gas discs.Nevertheless, some overall trends can be seen and the morphology c (cid:13) , 1–15 Tom Oosterloo et al. and kinematics of the detected H I structures can be divided into 3broad categories.The first category (denoted C, for cloud) contains those objectswhere the H I is found in small clouds, where in some cases it iseven not obvious whether the H I clouds are likely to be boundto the galaxy, or whether they are, e.g., ”free-floating” remnantsof a past interaction or even a low surface brightness companion.The galaxies that fall in this category are NGC 524, NGC 3384,NGC 3608, NGC 5982 and NGC 7332. The H I masses involvedare small and are all less than a few times M ⊙ .The second category (denoted A, for accretion) containsgalaxies where the H I is found in unsettled structures that areclearly connected to a recent or an ongoing gas-rich accretion. Thegalaxies that fall in this category are NGC 1023, NGC 2768 andNGC 5198, while to some extent NGC 3489 could also fall in thiscategory. The H I in NGC 1023 shows overall rotation, but the kine-matics shows many irregularities and the H I is clearly not settled.The H I masses involved range from several times M ⊙ to over M ⊙ .The final category (denoted D, for disc/ring) refers to galaxieswhere most of the H I is found in a fairly regularly rotating disc orring. This category contains the galaxies NGC 2685, NGC 3032,NGC 3414, NGC 3489, NGC 4150 and NGC 4278, while also theonly cluster galaxy detected (NGC 4262) falls in this category. InNGC 2685 and in NGC 4278 the H I disc is large, i.e. extendingbeyond the bright optical body, while in NGC 3032, NGC 3489and in NGC 4150 the H I forms a small, inner gas disc. In NGC3414, the H I appears to form a polar ring or disc. For galaxies withdiscs, the H I masses involved range from several times M ⊙ toover M ⊙ .The first conclusion that can be drawn is that in about half thegalaxies where H I is detected, it is found in a disc-like structure.Based on the first set of observations of the SAURON sample, Mor-ganti et al. (2006) had found that gas discs appear to be common inearly-type galaxies, while a similar result was found for early-typegalaxies detected in the HIPASS survey (Oosterloo et al. 2007).The extended set of observations of the
SAURON sample clearlyconfirms this.The second conclusion is that accretion of gas is common inearly-type field galaxies. For all galaxies detected the H I data re-veal that accretion is on-going, or has occurred in the recent past.Galaxies with a gas disc also show clear signs of accretion: in NGC3489, a faint, extended gas tail is detected which connects to theinner gas disc. Similarly, near NGC 4150 a small H I cloud is de-tected, while even the outer regions of the large, and presumablyolder, H I disc in NGC 4278 are connected to two large gas tails.The disc in NGC 2685 is heavily warped while the disc in NGC3414 is polar, hence it is likely that also in these galaxies the gas hasbeen accreted. The gas disc in NGC 3032 is counter-rotating to thestars. The conclusion is that gas discs in early-type galaxies formthrough accretion and that this is an ongoing process. Some of thevariation we see in H I properties in our sample may be reflectingdifferent stages of such accretion events. For example, NGC 2768 isaccreting gas and a small inner polar disc is forming (Crocker et al.2008). It is quite possible that this galaxy will evolve into a systemsimilar to NGC 3414. The accretion and inner discs in NGC 3489and NGC 4150 appear to have a similar history, where NGC 4150is probably at a slightly more evolved stage. NGC 3032 may be atan even more evolved stage. Another example is NGC 1023 wherea large and fairly massive H I structure is observed that shows anoverall rotation pattern, but that clearly has not settled into a disc. This system may evolve into a galaxy with a large, regular gas disc,similar to the one seen in NGC 4278.In a few galaxies, the amount of H I detected is above M ⊙ , i.e. similar to the amount of H I in the Milky Way. This sug-gests that in those cases the object accreted must have been fairlymassive. Moreover, the large extent and regular disc kinematics in-dicate that some must have formed several Gyr ago. However, inmost galaxies the H I masses involved are smaller and correspondto that of galaxies like the Magellanic Clouds or smaller. Assuming yr for the timescale of a typical accretion (e.g. Sancisi et al.2008; Tal et al. 2009), the detection rate and observed H I massesimply that the accretion rate for cold gas is smaller than 0.1 M ⊙ yr − for most field early-type galaxies. This suggests that, evenallowing for the fact that the H I is only a fraction of the mass ac-creted, currently early-type galaxies grow only by a modest amountthrough accretion. This has to be the case because, although mostgalaxies are small in size, most of the mass in galaxies is already inlarge galaxies (Renzini 2006). Therefore there is no large enoughreservoir of small galaxies available for large galaxies to grow sub-stantially by accretion of companions. We note that only one galaxy(NGC 2685) shows clear optical peculiarities that can be associatedwith accretion (in this case polar dust lanes). This underlines that,although accretion often occurs, the mass of the accreted object is,in most cases, small compared to that of the host galaxy and theeffects on the host galaxy usually are at most subtle. This is in linewith optical imaging studies of other samples of early-type galax-ies (e.g. Schweizer & Seitzer 1992; van Dokkum 2005; Tal et al.2009) which have shown that most early-type galaxies in the fieldand in small groups show signs of small accretion events. The newaspect from our results is that small amounts of gas are involved inthis continuing assembly of field early-type galaxies.In a recent review, Sancisi et al. (2008) concluded that for atleast 25% of field spiral galaxies there is direct evidence that a smallgas-rich companion or gas cloud is accreting, or has been accretedin the recent past. Also for many of these objects the optical imagedoes not show clear signs of interaction and the accretion is onlyvisible in H I observations. It appears that, as far as the character ofaccretion is concerned, there is not much difference between fieldspiral galaxies and field early-type galaxies and to some extent, theH I properties of early-type galaxies bear a resemblance with thoseof the outer regions of spiral galaxies. Sancisi et al. (2008) estimatethat, for field spiral galaxies, the accretion rate for cold gas is about0.1 to 0.2 M ⊙ yr − , somewhat higher than we estimate for ourearly-type galaxies. Several theoretical studies have shown that the dynamical struc-ture of a merger remnant critically depends on the amount of gas,and hence dissipation, present in (one of) the progenitors (e.g. Ben-der, Burstein, & Faber 1992; Jesseit, Naab, & Burkert 2005; Naab,Jesseit, & Burkert 2006; Hopkins et al. 2009; Jesseit et al. 2009).Moreover, a systematic change of the importance of dissipative ef-fects as function of mass may explain the different dynamical prop-erties of high-mass and low-mass early-type galaxies (Davies et al.1983).In Morganti et al. (2006) we had found that the H I detectionsare uniformly spread through the ( ǫ, V /σ ) diagram and we con-cluded, although the sample used was small, that if fast and slowrotators represent the relics of different formation paths, this didnot appear to be reflected in the current characteristics of the H I .The discussion of the previous section showed that, except in a few c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view λ R f N+CD
Figure 4.
Cumulative distribution of λ R for galaxies classified as havingan H I disc (class D) compared with that for galaxies undetected in H I orwhere only a small H I cloud was detected (classes N and C). cases, the H I currently detected is mainly due to recent small ac-cretions. Because the dynamical structure of a galaxy is the resultof the evolution over a Hubble time, a clear observable link withthese recently accreted small amounts of gas may not be expected.For our extended dataset there is, similar to the result of Mor-ganti et al. (2006), not much evidence that, for most galaxies, thecurrent H I content is connected to the dynamical characteristics ofthe galaxy. Different from Morganti et al. (2006), we use the param-eter λ R , introduced by Emsellem et al. (2007) to describe the im-portance of rotation. This parameter involves luminosity weightedaverages over the full two-dimensional stellar kinematic field as aproxy to quantify the observed projected stellar angular momentumper unit mass. It can have values between 0 and 1, and apart fromprojection effects, higher values of λ R suggest that rotation is moreimportant for the dynamics of the galaxy.In Fig. 4 we show the cumulative distributions of λ R for thegroup of galaxies that one could see as fast gas rotators (i.e. classD) and for the group of galaxies that are H I non- or slow gas rota-tors (classes N and C). No clear dichotomy emerges. For the discyH I detections, the distribution of λ R appears to go to higher values,which would suggest that for some galaxies with an H I disc rota-tion is also important for the stellar component. On the other hand,there is good overlap for smaller values of λ R suggesting that thepresence of an H I disc is not a good discriminator. This is alsosuggested by, for example, the fact of the sixteen fast rotating fieldgalaxies, only six have the H I distributed in a disk.The SAURON galaxies were selected to sample uniformly the projected axial ratio and the E/S0 morphology. Both quantities arerelated to the galaxy inclination as is λ R . For this reason the dis-tribution of λ R is related in a complex way to the selection criteriaand it is difficult to interpret quantitatively. An alternative way tolook at the relation between dynamics and H I morphology, whileincluding possible inclination effects, is by using the ( V /σ, ε ) dia-gram (Binney 2005). The study of the distribution of the SAURON galaxies on the diagram was discussed in Cappellari et al. (2007).It was shown that fast-rotators ETGs tend to lie in a restricted re-gion of that diagram, defined at the lower boundary by a lineartrend between intrinsic flattening and orbital anisotropy for edge-on systems (the magenta line in Fig. 5). Lowering the inclinationmoves galaxies to the left of that line on the ( V /σ, ε ) diagram asillustrated in Fig. 5 (for a detailed explanation see Cappellari et ε ( V / σ ) e ° ° ° ° ° ° ° ° ° ° Figure 5. ( V/σ, ε ) diagram for galaxies in our HI sample. Crosses are nondetections, Ellipses with axis are are HI disks, Empty circles are isolatedHI clouds and filled squares are HI accretion. Red/blue colours indicateslow/fast rotators. The NGC numbers of the HI detected galaxies are indi-cated. The magenta line corresponds to the relation δ = 0 . ε intr betweenanisotropy and intrinsic ellipticity. The dotted lines indicate how the relationis transformed when the inclination (indicated at the top) is decreased. Thevalues are taken from (Cappellari et al. 2007), where a detailed explanationof the diagram is given. al. (2007)). By plotting our different H I morphologies on the di-agram we conclude that: (i) The galaxies with the fastest intrin-sic (edge-on) V /σ in our sample all happen to posses H I disks.To these we should add NGC 2974, which has an intrinsic V /σ larger than NGC 3489 and also posses an H I disk (Weijmans etal. 2008). Within our limited number statistics this would suggeststhat an accretion/merger involving a large amount of gas is requiredto produce the galaxies most dominated by rotation; (ii) Howeverthe reverse is not true as H I disks can be present also at interme-diate V /σ , which implies that H I disks do not necessarily producea rotation-dominated object. A regular disk is seen in fact even inthe slow rotators NGC 3414. Larger samples are needed to conclu-sively understand the link between H I and galaxy dynamics, butthis work illustrates that there is not a simple connection. I and molecular gas Atomic hydrogen is not the only tracer of the cold ISM in galax-ies. A CO survey of the
SAURON sample (Combes et al. 2007) hasshown that a significant fraction of the
SAURON galaxies containsmolecular gas. It is therefore interesting to compare the H I prop-erties of the SAURON galaxies with those of the CO. More recentobservations (see Tab. 3) have somewhat modified the list of COdetections from Combes et al. (2007). The earlier CO detections inNGC 4278 and NGC 7457 have not been confirmed, while CO hasnow been detected in NGC 524. Of the sample we discuss in thispaper, six galaxies are detected both in CO and in H I (NGC 524,NGC 2685, NGC 2768, NGC 3032, NGC 3489 and NGC 4150),out of a total of nine CO detections and fifteen H I detections. In-terestingly, all these galaxies show small amounts of star formation(Shapiro et al. 2010, see also Sect. 4.6). In five of the CO detec- c (cid:13)000
SAURON galaxies containsmolecular gas. It is therefore interesting to compare the H I prop-erties of the SAURON galaxies with those of the CO. More recentobservations (see Tab. 3) have somewhat modified the list of COdetections from Combes et al. (2007). The earlier CO detections inNGC 4278 and NGC 7457 have not been confirmed, while CO hasnow been detected in NGC 524. Of the sample we discuss in thispaper, six galaxies are detected both in CO and in H I (NGC 524,NGC 2685, NGC 2768, NGC 3032, NGC 3489 and NGC 4150),out of a total of nine CO detections and fifteen H I detections. In-terestingly, all these galaxies show small amounts of star formation(Shapiro et al. 2010, see also Sect. 4.6). In five of the CO detec- c (cid:13)000 , 1–15 Tom Oosterloo et al.
NGC H I type M H M H / M HI CO references M ⊙ central(1) (2) (3) (4) (5)524 none . × > < . × < . × . × . × < . × < . × . × < . × < < . × < . × > . × > . × > Table 3.
Molecular mass to H I mass ratios for the galaxies with CO de-tections and/or central H I . (1) Galaxy identifier. (2) Classification of thecentral H I . (3) Total molecular gas mass. (4) Ratio of total molecular gasmass to the H I mass observed in the central interferometric beam. (5) Ref-erences: (a) Crocker et al. unpublished; (b) Combes et al. (2007); (c) Schin-nerer & Scoville (2002); (d) Crocker et al. (2008); (e) Young, Bureau, &Cappellari (2008) (f) Crocker et al. (in preparation). tions H I is also seen in the central region of the galaxy (NGC 2685,NGC 2768, NGC 3032, NGC 3489, NGC 4150) and the CO and theH I show very similar kinematics, indicating that the same compo-nent is detected both in CO and H I . The exception is NGC 524where no atomic counterpart is detected of the central moleculardisc. Similarly, in the Virgo galaxies NGC 4459, NGC 4477 andNGC 4550 CO was found, but no H I .For a proper comparison of the H I and the H properties, it isimportant to keep in mind that the H I observations typically probea region of several tens of kpc in size, i.e. an entire galaxy and itsimmediate environment, while the CO observations refer only tothe central region of a galaxy on kpc scales. A comparison basedon total H I content and a central CO measurement is not necessar-ily meaningful. The spatial resolution of our WSRT observationsis well matched to the field-of-view of the IRAM 30-m dish usedby Combes et al. (2007). We therefore compare the CO measure-ments with the H I detected in the single WSRT beam centred onthe galaxy while taking into account the central morphology of theH I of our galaxies. We divided the central H I morphology in threecategories: no central H I , central dip in the H I (within a large-scaleH I structure), or central H I peak. The category for each galaxy islisted in Table 3 and the relation with the CO properties is givenin Fig. 6. This figure clearly shows that a statistical relation existsbetween the presence of H I and CO in the central regions. Galax-ies with centrally peaked H I are more likely to have CO than theother two types. Of the six galaxies with a centrally peaked H I distribution, five have a central CO component. For comparison, ifone would use the total H I content, the statistics clearly gets di-luted: of all fifteen H I detections, only six are detected in CO. Weconclude that centrally peaked distributions of H I tend to harbourcorresponding central molecular gas distributions. The only excep-tion to this rule seems to be NGC 3414. However, the kinematicsof the H I in this galaxy suggests that this H I likely forms an edge-on polar ring and that the observed central H I peak may be due toprojection.A few more things can be learned from this comparison. Inthe five galaxies where both H I and CO are detected in the cen- HI central peak HI central dip HI offset/non-detection N u m b e r o f g a l a x i e s CO not detectedCO detected
Figure 6.
CO detection histogram as a function of central H I morphology. tral regions, interferometric CO observations (Crocker et al. 2008;Young et al. 2010; Crocker et al. 2010) show that the kinematicsand the morphology of both components are clearly connected andthe same physical structure is detected. A very nice example is theinner gas disc of NGC 3032 for which the extent and kinematicsas seen in CO and in H I match exactly. A very interesting aspectis that the combination of the CO and H I observations of NGC2768, NGC 3032, NGC 3489 and NGC 4150 gives clear evidencethat these inner gas structures form by the accretion of small-gas-rich objects. The inner gas discs that are seen in NGC 2768, NGC3489 and NGC 4150 connect to large, faint H I plumes that areseen at larger radii, clearly showing that these gas structures formedfrom accreted gas. The fact that the inner gas disc in NGC 3032 iscounter-rotating to the stars also shows an external origin. Our re-sults also show that the cold ISM of the inner gas discs detectedboth in H I and CO is mainly in the form of molecular gas, withthe molecular gas mass being about 10 times higher than that of theatomic gas. Interestingly, this mass ratio is very similar to that seenin the centres of nearby spiral galaxies (e.g. Leroy et al. 2008),despite the very different state of the ISM in these two types ofgalaxies. I and ionised gas With the new H I observations presented in this paper, we furtherconfirm the result of Morganti et al. (2006) that galaxies with reg-ular H I discs tend to have strong, extended emission from ionisedgas that has the same kinematics as the H I while galaxies withunsettled H I structures have less ionised gas. For the galaxies forwhich we present data for the first time, regular discs of ionisedgas are detected with SAURON in NGC 3032, NGC 3489 andNGC 4262 (Sarzi et al. 2006), exactly the three galaxies in the newobservations in which regular H I discs are found. In the remain-ing H I detections, where the H I is found in small clouds offsetfrom the centre (NGC 524, NGC 3384 and NGC 3608), only smallamounts of ionised gas are detected. We therefore reiterate that, inearly-type galaxies, a regular H I disc always has an ionised coun-terpart. Several earlier studies (e.g. Serra et al. 2006; Morganti et al. 2006)have concluded that the relation between H I content and stellarpopulation for field early-type galaxies is complex. Our data show c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view Stellar age (Gyr) f central H I no central H I Stellar age (Gyr) f fieldVirgo Figure 7.
Left: cumulative distributions of stellar ages inside 1 R e from single-population models from Kuntschner et al. (2010) where the sample is split inthose galaxies with and those without H I detected in their central region. Right: Same distributions, but now the sample is split in field and Virgo galaxies that, on the scale of the spatial resolution of our data, the columndensities of the H I are below the threshold above which star for-mation is generally occurring. However, the highest column den-sities observed are not very much below this threshold, so giventhe low spatial resolution of our data, some star formation couldoccur in smaller regions. Low-level star formation (typically belowlevels of 0.1 M ⊙ yr − ) is indeed observed in some of the gas-rich SAURON galaxies (Temi, Brighenti, & Mathews 2009; Jeong et al.2009; Shapiro et al. 2010, see also below).First we discuss to what extent the H I properties are con-nected to the ongoing star formation. As expected, some relationseems to hold. This is best illustrated by comparing our data withthose from Shapiro et al. (2010). Based on Spitzer data, Shapiroet al. (2010), similarly to Temi, Brighenti, & Mathews (2009), ob-serve small amounts of star formation in a subset of the
SAURON galaxies. The star formation they observe only occurs in galaxiescharacterised by fast rotation. The morphology of the star formingregions is that of a thin disc or ring. Moreover, they were able todistinguish two modes of star formation. In one mode, star forma-tion is a diffuse process and corresponds to widespread young stel-lar populations and high specific molecular gas content. Shapiro etal. (2010) associate this star formation with small accretion events.In the second mode, the star formation is occurring more centrallyconcentrated, while outside the region of star formation only oldstellar populations are present. Smaller amounts of molecular gasare connected to these central star formation events. Shapiro et al.(2010) speculate that in at least some of these objects, the star areforming from gas resulting from internal mass loss.It is interesting to see that in 4 of the 5 galaxies classified byShapiro et al. (2010) to have widespread star formation and that areobserved by us, both CO and H I is detected in the central regions(Table 3). In contrast, in none of the galaxies with centrally concen-trated star formation H I is detected in the centre. This confirms thesuggestion by Shapiro et al. (2010) that widespread star formationin early-type galaxies is connected with higher gas content. We donote that for some early-type galaxies GALEX data show that thisrelation between gas content and star formation also exists at largeradius. Examples are NGC 404 (Thilker et al. 2010), NGC 2974(Weijmans et al. 2008) and ESO 381–47 (Donovan et al. 2009). Al-most all galaxies with widespread star formation and observed byus, have H I discs (NGC 2685, NGC 3032, NGC 3489, NGC 4150),quite distinct from galaxies with central star formation (no detec- tion of NGC 4459 and NGC 4477, while only a small, offset H I cloud is detected in NGC 524). The only exception is NGC 4550,a galaxy in the Virgo cluster with widespread star formation andnot detected in H I . This is a very unusual galaxy with two counter-rotating stellar discs. We also note that in all four galaxies withwidespread star formation and detected by us, the H I is likely tohave been accreted (see Sect. 4.2). This confirms the conclusion ofShapiro et al. (2010) that widespread star formation is associatedwith accretion.The above results show that some star formation can occurin the gas reservoirs of early-type galaxies, although not all gas-rich galaxies show star formation. Overall, the connection with theproperties of the stellar population is poor. Some galaxies indeedbehave according to the expectation that the presence of a rela-tively large amount of H I indeed is connected to the propertiesof the stellar population, but the rule seems to be that for everyrule there are exceptions. For example, NGC 1023, NGC 3414 andNGC 4278 have extensive reservoirs of neutral hydrogen, but donot show any evidence for the presence of a young stellar subpopu-lation. To characterise the overall stellar populations, we have usedthe single-population-equivalent stellar ages inside 1 R eff as de-rived for the SAURON galaxies by Kuntschner et al. (2010). Oneway to investigate the connection between H I and stellar popula-tion is to compare these stellar ages of galaxies with H I to thoseof galaxies without central H I . In Fig. 7 we give the cumulativedistributions of this stellar age for galaxies where H I is detectedin the central regions and for those with no central H I . This figuresuggests that some difference may exist between the two groups ofgalaxies, with some gas-rich galaxies having younger stellar ages.However, the difference between the distributions is mainly causedby only a few gas-rich galaxies that have quite young stellar ages(i.e. NGC 3032, NGC 3489 and NGC 4150 which are known tohave ongoing star formation). Many galaxies with central H I havesimilar stellar populations as gas-free galaxies. Some galaxies withfairly young or intermediate ages are in fact gas free (e.g. NGC3377 and NGC 7457) and some gas-rich galaxies have large stellarages. A similar trend is seen when the total H I content is used in-stead of the central one. This is further illustrated in Fig. 8 where weplot the relative global gas content versus stellar age for the differ-ent H I morphologies. No clear overall trend is visible in this figure,except (again) that the three youngest galaxies all have a central H I disc that is also detected in CO. The at most weak connection be- c (cid:13)000
SAURON galaxies. The star formation they observe only occurs in galaxiescharacterised by fast rotation. The morphology of the star formingregions is that of a thin disc or ring. Moreover, they were able todistinguish two modes of star formation. In one mode, star forma-tion is a diffuse process and corresponds to widespread young stel-lar populations and high specific molecular gas content. Shapiro etal. (2010) associate this star formation with small accretion events.In the second mode, the star formation is occurring more centrallyconcentrated, while outside the region of star formation only oldstellar populations are present. Smaller amounts of molecular gasare connected to these central star formation events. Shapiro et al.(2010) speculate that in at least some of these objects, the star areforming from gas resulting from internal mass loss.It is interesting to see that in 4 of the 5 galaxies classified byShapiro et al. (2010) to have widespread star formation and that areobserved by us, both CO and H I is detected in the central regions(Table 3). In contrast, in none of the galaxies with centrally concen-trated star formation H I is detected in the centre. This confirms thesuggestion by Shapiro et al. (2010) that widespread star formationin early-type galaxies is connected with higher gas content. We donote that for some early-type galaxies GALEX data show that thisrelation between gas content and star formation also exists at largeradius. Examples are NGC 404 (Thilker et al. 2010), NGC 2974(Weijmans et al. 2008) and ESO 381–47 (Donovan et al. 2009). Al-most all galaxies with widespread star formation and observed byus, have H I discs (NGC 2685, NGC 3032, NGC 3489, NGC 4150),quite distinct from galaxies with central star formation (no detec- tion of NGC 4459 and NGC 4477, while only a small, offset H I cloud is detected in NGC 524). The only exception is NGC 4550,a galaxy in the Virgo cluster with widespread star formation andnot detected in H I . This is a very unusual galaxy with two counter-rotating stellar discs. We also note that in all four galaxies withwidespread star formation and detected by us, the H I is likely tohave been accreted (see Sect. 4.2). This confirms the conclusion ofShapiro et al. (2010) that widespread star formation is associatedwith accretion.The above results show that some star formation can occurin the gas reservoirs of early-type galaxies, although not all gas-rich galaxies show star formation. Overall, the connection with theproperties of the stellar population is poor. Some galaxies indeedbehave according to the expectation that the presence of a rela-tively large amount of H I indeed is connected to the propertiesof the stellar population, but the rule seems to be that for everyrule there are exceptions. For example, NGC 1023, NGC 3414 andNGC 4278 have extensive reservoirs of neutral hydrogen, but donot show any evidence for the presence of a young stellar subpopu-lation. To characterise the overall stellar populations, we have usedthe single-population-equivalent stellar ages inside 1 R eff as de-rived for the SAURON galaxies by Kuntschner et al. (2010). Oneway to investigate the connection between H I and stellar popula-tion is to compare these stellar ages of galaxies with H I to thoseof galaxies without central H I . In Fig. 7 we give the cumulativedistributions of this stellar age for galaxies where H I is detectedin the central regions and for those with no central H I . This figuresuggests that some difference may exist between the two groups ofgalaxies, with some gas-rich galaxies having younger stellar ages.However, the difference between the distributions is mainly causedby only a few gas-rich galaxies that have quite young stellar ages(i.e. NGC 3032, NGC 3489 and NGC 4150 which are known tohave ongoing star formation). Many galaxies with central H I havesimilar stellar populations as gas-free galaxies. Some galaxies withfairly young or intermediate ages are in fact gas free (e.g. NGC3377 and NGC 7457) and some gas-rich galaxies have large stellarages. A similar trend is seen when the total H I content is used in-stead of the central one. This is further illustrated in Fig. 8 where weplot the relative global gas content versus stellar age for the differ-ent H I morphologies. No clear overall trend is visible in this figure,except (again) that the three youngest galaxies all have a central H I disc that is also detected in CO. The at most weak connection be- c (cid:13)000 , 1–15 Tom Oosterloo et al. M HI /L B (M sun / L B, sun ) S t e ll a r a g e ( G y r ) NGC 3032NGC 3489NGC 4150
Figure 8.
Stellar age inside 1 R eff plotted as function of relative H I con-tent. The different symbols refer to H I morphology: filled circles: regulardisc/ring (class D); stars: unsettled H I structures (class A); open circles:small or offset H I clouds (class C); triangles: upper limits tween H I and stellar age is to a large extent similar to that seen forthe ionised gas. Emsellem et al. (2007) show that those SAURON galaxies with relatively much ionised gas also have young stellarages, but on the other hand several galaxies with young or interme-diate age populations are free of ionised gas.The results discussed above clearly show that the relation be-tween H I and stellar populations is complex. This is not unex-pected, since the stellar population is the result of the evolution ofa galaxy over its entire life, while the current H I content only re-flects the present state. In Sect. 4.2 we showed that some amountsof gas are accreted by field early-type galaxies at irregular intervals.Often the amounts of gas involved are small and, even if entirelyconverted into stars, will only leave subtle signatures in the stellarpopulation that are not always easy to detect (see also e.g. Serra &Oosterloo 2010). Moreover, the efficiency with which accreted gasis turned into stars depends on many factors. For example, accre-tions characterised by loss of gas angular momentum (e.g., retro-grade encounters) result in efficient gas infall that triggers centralstar formation. This may be observed as a young population in arelatively gas-rich galaxy, but after a while, the gas reservoir willbe exhausted and the remnant is observed as H I -poor and centrallyrejuvenated (Serra et al. 2006). On the other hand, interactions inwhich gas retains its angular momentum (e.g., prograde encoun-ters) result in large H I tidal tails that can later be re-accreted toform large H I discs. Because of their large extent, the column den-sities in these discs is low and at most some star formation willoccur in localised regions at large radius (see e.g. Oosterloo et al.2007). Recent work also suggest that bulges can have a stabilisingeffect on discs, preventing star formation, even when significantamounts of gas is present (Martig et al. 2009).It is instructive to consider the effect of the environment. Theresults discussed so far show that the relation between current gascontent and stellar population is complex. However, by comparingthe Virgo early-type galaxies with those in the field, one may getan idea of the effects of gas accretion over longer periods of time.Our data indicate that for field early-type galaxies, gas accretiondoes play a significant role in determining the stellar population.This can be seen by comparing the cumulative distributions of thestellar ages of Virgo and field galaxies (Fig. 7). These distributionsshow that there is a trend of field early-type galaxies, as a popu- HI central HI offset HI non-detection N u m b e r o f g a l a x i e s Continuum not detectedContinuum detected
Figure 9.
Histogram of the distribution of radio continuum detection asfunction of H I detection and morphology. lation, having younger stellar ages than Virgo early-type galaxies.This difference likely reflects the different long-term accretion his-tory of the two groups of galaxies. Field early-type galaxies regu-larly accrete small amounts of gas from their environment and, overtime, this leaves an observable signature in the stellar population. Incontrast, galaxies in Virgo, being in a gas poor environment, growmuch less by accretion of gas-rich companions and even if somegas is accreted, it is removed on a short timescale by the clusterenvironment. The stellar population of cluster early-type galaxiesis much less rejuvenated during their evolution compared to fieldearly-type galaxies. I and the radio continuum As described in Sec. 2, our observations have also allowed to ex-tract images of the radio continuum emission. As detailed below,these images are in many cases much deeper than those availableso far. This resulted in a significant number of new detections ofthe radio continuum associated with our target galaxies. We detectradio continuum in 13 of the 20 field galaxies and in 5 of the 13cluster galaxies. With the exception of a few well-known objectswith strong radio continuum (e.g. NGC 4278, NGC 4374/M84),the majority of the detected sources have a radio flux of at most afew mJy. In all detected sources, the radio continuum comes fromthe central region of the galaxy.The interesting result is that there appears to be a trend be-tween detection of radio continuum and detection of H I in oraround the galaxy. Figure 9 shows the distribution of radio con-tinuum detections as function of H I presence and H I morphol-ogy. The histogram shows that the radio continuum detection rateis higher for objects where also H I is detected, with a suggestion ofan additional trend that galaxies with H I but not in their central re-gions are less likely to be detected in radio continuum than galaxieswith H I in the centre, but more likely than galaxies with no H I atall. This suggests that the cold gas somehow contributes in feedingthe processes that produce the radio continuum emission in somegalaxies. It is well known that both star formation and AGN activ-ity can contribute to the radio continuum emission from early-typegalaxies (Wrobel & Heeschen 1991). Here we address the questionis to what extent the observed trend with HI properties is connectedto radio emission from star formation or from radio-loud AGN.To study this question, we estimated the radio continuum fluxexpected to be observed from the amounts of star formation seen by c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view Shapiro et al. (2010) and compared these with the observed fluxes.Of the 8 star forming galaxies overlapping between Shapiro et al.(2010) and us, 7 are detected in radio continuum. We used the rela-tion given by Bell (2003) to convert star formation rates to radio lu-minosities. For NGC 2685, NGC 3032, NGC 3489, NGC 4150 andNGC 4459, the observed radio continuum flux matches, within theerrors, what is expected from the observed star formation rates andwe conclude that the observed radio continuum is associated withthe star formation. NGC 524 is known to have a faint radio AGN(Nagar, Falcke, & Wilson 2005), and when correcting the observedradio flux for this AGN, also for this galaxy the observed radio con-tinuum matches that what is expected based on the observed starformation rate. Our upper limit to the continuum flux of NGC 4550is consistent with the observed star formation in this galaxy. Be-cause Shapiro et al. (2010) derive the star formation rates from FIRdata, the good match between predicted and observed radio con-tinuum in these galaxies implies that they follow the well-knownradio-FIR correlation. Only in one star forming galaxy (NGC 4477)is the observed radio flux much larger than expected (by about afactor 7), which most likely means that this galaxy harbours a radioloud AGN.In the previous section, we showed that galaxies with H I intheir central regions are more likely to have star formation, whilehere we find that in many of these star forming galaxies, the radiocontinuum detected is due to this star formation. For about half thegalaxies represented in the first bin of Fig. 9 this is the case. On theother hand, for both other H I classes shown in this figure, only oneof the objects is forming stars. These results show that the trendseen in Fig. 9 is at least partly due to enhanced star formation ingalaxies with central H I .The spatial resolution of our observations is relatively low(corresponding to a linear scale of the order of a kpc for the mostdistant galaxies). This makes it difficult to decide, based on theradio data alone, whether the observed radio continuum is due toa radio-loud AGN. We therefore have searched the literature forhigher resolution radio continuum data.One possibility is to look into the FIRST survey (Faint Imagesof the Radio Sky at Twenty-Centimeters Becker, White, & Helfand1995). FIRST data have better spatial resolution ( ∼ ′′ ), but highernoise level (about 1 mJy beam − ) compared to our WSRT obser-vations. A search for detections in this survey has been also doneby Sarzi et al. (2010). Of the 9 field galaxies detected by us at thelevel of a few mJy (i.e. such that the sensitivity of FIRST wouldallow to detect these sources) 7 are detected by FIRST. In only oneof these detections - NGC 3032 - the radio emission is resolvedby FIRST, suggesting the presence of star formation, consistent theresults presented here on this galaxy.More instructive is, however, to consider the work of Nagar,Falcke, & Wilson (2005, and references therein). Their work fo-cuses on detecting radio nuclei with high brightness temperature( > K) and/or jet-like structures as unambiguous indicationsfor the presence of an AGN. They concentrated on low-luminosityactive galactic nuclei (LLAGN) and AGN selected from the Palo-mar Spectroscopic sample of northern galaxies that they have stud-ied at high frequency (15 G Hz) with the VLA (0.15 ′′ resolution)and with VLBI at intermediate and high radio frequencies. At thesehigh resolutions (corresponding to a linear size of a few tens of pc),they detect radio emission in almost 50 % of the sources, suggest-ing a high incidence of radio cores.Interestingly, their sample includes a fair number of our ob-jects, although, unfortunately, the sensitivity of their observationsis not as good as of our WSRT observations (flux limit of 1-1.5 mJy for the VLA observations and 2.7 mJy for the VLBI obser-vations). Of the galaxies that are in common, for 11 is our WSRTflux above their detection limit and all are detected in the high-resolution observations of Nagar et al., suggesting the presence ofa compact (core?) structure in these galaxies. One further object(NGC 5198) is not included in the Nagar et al. list but is classi-fied as AGN by Sarzi et al. (2010) and also this object is detectedin radio (both by WSRT and by FIRST). Almost all these objectsare observed not to have star formation (e.g. Shapiro et al. 2010).We conclude that many of our early-type galaxies harbour a radio-load AGN. Interestingly, these AGN are evenly distributed over thethree H I classes used in Fig. 9 suggesting that AGN activity is notconnected to whether there is cold gas in the central region of anearly-type galaxy or not. The observed relation between central gascontent and radio continuum seems therefore to be due to a higherprobability for radio emission from star formation if a galaxy has arelatively large amount of gas in the central regions. We have presented new, deep Westerbork Synthesis Radio Tele-scope observations of the neutral hydrogen in 22 nearby early-type galaxies selected from a representative sample of early-typegalaxies studied earlier at optical wavelengths with the
SAURON integral-field spectrograph. Combined with our earlier observa-tions, this resulted in deep H I data on 33 nearby early-type galax-ies. This is the largest homogeneous dataset of H I imaging dataavailable for this class of objects.In contrast to our earlier study (Morganti et al. 2006), thissample both covers field environments and the Virgo cluster. Wefind that the H I properties strongly depend on environment. Fordetection limits of a few times M ⊙ , we detect H I in about 2/3of the field galaxies, while for Virgo galaxies the detection rate is < I , this atomic gas is in a regularly rotating H I discor ring. In many objects, unsettled H I structures are detected, suchas tails and clouds, sometimes connecting to a regular H I disc. Thissuggests that the gas discs form through accretion. We concludethat accretion of H I commonly occurs in field early-type galaxies,but typically with very modest accretion rates. In contrast, clustergalaxies do not accrete cold gas. All the H I discs that we observedhave counterparts of ionised gas. Moreover, galaxies with an in-ner H I disc also have an inner molecular gas disc. The strikinglysimilar kinematics of these different tracers shows that they are allpart of the same structure. The combination of H I and CO imag-ing clearly shows, through the detection of gas tails, that the innerdiscs are the result of accretion events. The cold ISM in the centralregions is dominated by molecular gas ( M H /M HI ≃ ).There is no obvious overall relation between current gas con-tent and internal dynamics. This is not very surprising given that inmost galaxies with H I it is due to the recent accretion of smallamounts of gas while the dynamical characteristics are set overmuch longer timescales. Within our limited number statistics, thefastest rotating galaxies all posses H I disks. This would suggestthat an accretion/merger involving a large amount of gas is requiredto produce the galaxies most dominated by rotation. However, thereverse is not true as H I disks can be present also at intermediate V /σ . The similarity between the kinematics of the H I and that ofthe stars seen in some of the galaxies with large, regular H I disks, c (cid:13) , 1–15 Tom Oosterloo et al. suggests that an accretion/merger involving a large amount of gaswas part of the evolution of those systems.We observe a close relationship between gas content and thesmall amounts of star formation which occurs in some of the
SAURON galaxies. Galaxies with widespread star formation aregas rich and are galaxies that have experienced a recent accretionevent. The radio continuum emission detected in these galaxies isconsistent with the star formation observed. However, as alreadynoticed by Morganti et al. (e.g. 2006) and Serra et al. (2006), therelation between H I and the overall properties of the stellar popu-lation is very complex. The few galaxies with a significant youngsub-population and/or star formation, have inner gas discs. For theremaining galaxies there is no trend between stellar populationand H I properties. Very interestingly, we find a number early-typegalaxies that are very gas rich but are not forming stars and have notdone so for a while as they only have an old stellar population. Oneexample of such a galaxy is NGC 4278 where the fact that the largeH I disc shows very regular kinematics implies that this galaxy hasbeen gas rich for at least a few Gyr. Despite this, there is no evi-dence for a young stellar population in this galaxy. In addition, wefind that the stellar populations of our field galaxies are typicallyyounger than those in Virgo. This is expected because field galax-ies are likely to have accreted some gas in the last few Gyr, while inthe Virgo cluster this is not the case. The difference in stellar pop-ulation is reflecting of the differences in accretion history of coldgas between the two groups of galaxies.In about 50% of our sample, we detect a central radio contin-uum source. In many galaxies, the continuum emission is due toa radio-loud AGN, but continuum emission from star formation isalso detected in some galaxies. Galaxies with star formation followthe radio-FIR correlation. The presence of radio continuum emis-sion correlates with the H I properties, in the sense that galaxieswith H I in the central region are more likely to be detected in con-tinuum compared to galaxies without H I . Galaxies with H I but inoff-centre structures behave in between. This trend is mainly due tothe star formation observed in galaxies with central gas reservoirs,and is not related to AGN fuelling.In this paper, we have presented a number of interesting trendsthat suggest that gas and gas accretion plays a role in the evolutionof early-type galaxies, in particular those found in the field. Al-though our collection of H I images is the largest and deepest avail-able for early-type galaxies, the number is still fairly small and thestatistical basis for most of the trends we find is not strong. Similardata on larger samples will be needed to put the results presentedhere on a more solid basis. We are in the process of collecting deepH I imaging data for a much larger sample of early-type galaxies(the ATLAS sample, see http://purl.org/atlas3d Cappellari et al.2010; Serra et al. 2009) that will allow us to further investigate thetrends described here. ACKNOWLEDGMENTS
We thank the anonymous referee for helpful suggestions which im-proved the paper. The Westerbork Synthesis Radio Telescope is op-erated by the Netherlands Foundation for Research in AstronomyASTRON, with support of NWO. This research has made use ofthe NASA/IPAC Extragalactic Database (NED) which is operatedby the Jet Propulsion Laboratory, California Institute of Technol-ogy, under contract with the National Aeronautics and Space Ad-ministration. The Digitized Sky Survey was produced at the SpaceTelescope Science Institute under US Government grant NAG W- 2166. MC acknowledge support from a STFC Advanced Fellow-ship (PP/D005574/1).
REFERENCES
Allen S. W., Dunn R. J. H., Fabian A. C., Taylor G. B., ReynoldsC. S., 2006, MNRAS, 372, 21Balmaverde B., Baldi R. D., Capetti A., 2008, A&A, 486, 119Becker R. H., White R. L., Helfand D. J., 1995, ApJ, 450, 559Bell E. F., 2003, ApJ, 586, 794Bell E. F., et al., 2004, ApJ, 608, 752Bender R., Burstein D., Faber S. M., 1992, ApJ, 399, 462Bigiel F., Leroy A., Walter F., Brinks E., de Blok W. J. G., MadoreB., Thornley M. D., 2008, AJ, 136, 2846Binggeli B., Tammann G. A., Sandage A., 1987, AJ, 94, 251Binney J., 2005, MNRAS, 363, 937Briggs D. S., 1995, in Bulletin of the American AstronomicalSociety Vol. 27, High Fidelity Interferometric Imaging: RobustWeighting and NNLS Deconvolution. pp 1444Cappellari M., et al., 2007, MNRAS, 379, 418Cappellari M., et al., 2010, in prepCayatte V., van Gorkom J. H., Balkowski C., Kotanyi C., 1990,AJ, 100, 604Chung A., van Gorkom J. H., Kenney J. D. P., Crowl H., VollmerB., 2009, arXiv, arXiv:0909.0781Combes F., Young L. M., Bureau M., 2007, MNRAS, 377, 1795Crocker A. F., Bureau M., Young L. M., Combes F., 2008, MN-RAS, 386, 1811Crocker A. F. et al. in prep.Davies R. L., Efstathiou G., Fall S. M., Illingworth G., SchechterP. L., 1983, ApJ, 266, 41De Lucia G., Springel V., White S. D. M., Croton D., KauffmannG., 2006, MNRAS, 366, 499de Zeeuw P. T., Bureau M., Emsellem E., Bacon R., Carollo C. M.,Copin Y., Davies R. L., Kuntschner H., Miller B. W., Monnet G.,Peletier R. F., Verolme E. K., 2002, MNRAS, 329, 513di Serego Alighieri S., et al., 2007, A&A, 474, 851Donovan J. L., Hibbard J. E., van Gorkom J. H., 2007, AJ, 134,1118Donovan J. L., et al., 2009, AJ, 137, 5037Emsellem E., Cappellari M., Peletier R. F., McDermid R. M.,Bacon R., Bureau M., Copin Y., Davies R. L., Krajnovi´c D.,Kuntschner H., Miller B. W., de Zeeuw P. T., 2004, MNRAS,352, 721Emsellem E., et al., 2007, MNRAS, 379, 401Giovanardi C., Krumm N., Salpeter E. E., 1983, AJ, 88, 1719Giovanelli R., et al., 2005, AJ, 130, 2598Grossi M., et al., 2009, A&A, 498, 407Hibbard J. E., Sansom A. E., 2003, AJ, 125, 667Hopkins P. F., Lauer T. R., Cox T. J., Hernquist L., Kormendy J.,2009, ApJS, 181, 486Jeong H., et al., 2009, MNRAS, 398, 2028Jesseit R., Naab T., Burkert A., 2005, MNRAS, 360, 1185Jesseit R., Cappellari M., Naab T., Emsellem E., Burkert A., 2009,MNRAS, 397, 1202Kaviraj S., Tan K.-M., Ellis R. S., Silk J., 2010, arXiv,arXiv:1001.2141Kenney J. D., Young J. S., 1988, ApJS, 66, 261Kennicutt Jr. R. C., 1989, ApJ, 344, 685Kent B. R., 2010, AAS, 41, 386 c (cid:13) , 1–15 arly-type galaxies in different environments, an H I view Kereˇs D., Katz N., Weinberg D. H., Dav´e R., 2005, MNRAS, 363,2Kuntschner H., Emsellem E., Bacon R., Bureau M., CappellariM., Davies R. L., de Zeeuw P. T., Falc´on-Barroso J., Krajnovi´cD., McDermid R. M., Peletier R. F., Sarzi M., 2006, MNRAS,369, 497Kuntschner H., 2010, MNRAS, submittedLeroy A. K., Walter F., Brinks E., Bigiel F., de Blok W. J. G.,Madore B., Thornley M. D., 2008, AJ, 136, 2782Malin D. F., Carter D., 1983, ApJ, 274, 534Martig M., Bournaud F., Teyssier R., Dekel A., 2009, arXiv,arXiv:0905.4669McDermid R. M., et al., 2006, MNRAS, 373, 906McDermid R. M., et al., 2007, NewAR, 51, 13McNamara B. R., Sancisi R., Henning P. A., Junor W., 1994, AJ,108, 844Mei S., et al., 2005, ApJ, 625, 121Michel-Dansac L., et al., 2010, ApJ, 717, L143Morganti R., Sadler E. M., Oosterloo T., Pizzella A., Bertola F.,1997, AJ, 113, 937Morganti R., de Zeeuw P. T., Oosterloo T. A., McDermid R. M.,Krajnovi´c D., Cappellari M., Kenn F., Weijmans A., Sarzi M.,2006, MNRAS, 371, 157Naab T., Jesseit R., Burkert A., 2006, MNRAS, 372, 839Nagar N. M., Falcke H., Wilson A. S., 2005, A&A, 435, 521Nipoti C., Londrillo P., Ciotti L., 2003, MNRAS, 342, 501Nipoti C., Treu T., Bolton A. S., 2009, ApJ, 703, 1531Oosterloo T., Morganti R., Sadler E. M., Ferguson A., van derHulst T., Jerjen H., 2004, IAUS, 217, 486Oosterloo T., van Gorkom J., 2005, A&A, 437, L19Oosterloo T. A., Morganti R., Sadler E. M., van der Hulst T., SerraP., 2007, A&A, 465, 787Renzini A., 2006, ARA&A, 44, 141Roberts M. S., Haynes M. P., 1994, ARA&A, 32, 115Sage L. J., Welch G. A., 2006, ApJ, 644, 850Sage L. J., Welch G. A., Young L. M., 2007, ApJ, 657, 232S´anchez-Bl´azquez P., Gibson B. K., Kawata D., Cardiel N., Bal-cells M., 2009, MNRAS, 400, 1264Sancisi R., Fraternali F., Oosterloo T., van der Hulst T., 2008,A&ARv, 15, 189Sarzi M., Falc´on-Barroso J., Davies R. L., Bacon R., Bureau M.,Cappellari M., de Zeeuw P. T., Emsellem E., Fathi K., Krajnovi´cD., Kuntschner H., McDermid R. M., Peletier R. F., 2006, MN-RAS, 366, 1151Sarzi M., et al., 2010, MNRAS, 402, 2187Sault R. J., Teuben P. J., Wright M. C. H., 1995, ASPC, 77, 433Schaye J., 2004, ApJ, 609, 667Schinnerer E., Scoville N., 2002, ApJ, 577, L103Schneider S. E., Helou G., Salpeter E. E., Terzian Y., 1983, ApJ,273, L1Schneider S. E., 1989, ApJ, 343, 94Schweizer F., Seitzer P., 1992, AJ, 104, 1039Serra P., Trager S. C., van der Hulst J. M., Oosterloo T. A., Mor-ganti R., 2006, A&A, 453, 493Serra P., Trager S. C., Oosterloo T. A., Morganti R., 2008, A&A,483, 57Serra P., et al., 2009, in: Panoramic Radio Astronomy: Wide-field 1-2 GHz research on galaxy evolution. Published online athttp://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=89, id.56Serra P., Oosterloo T. A., 2010, MNRAS, 401, L29Shapiro K. L., et al., 2010, MNRAS, 402, 2140 Solanes J. M., Manrique A., Garc´ıa-G´omez C., Gonz´alez-CasadoG., Giovanelli R., Haynes M. P., 2001, ApJ, 548, 97Tadhunter C., Robinson T. G., Gonz´alez Delgado R. M., Wills K.,Morganti R., 2005, MNRAS, 356, 480Tal T., van Dokkum P. G., Nelan J., Bezanson R., 2009, arXiv,arXiv:0908.1382Temi P., Brighenti F., Mathews W. G., 2009, ApJ, 707, 890Thilker D. A., et al., 2010, ApJ, 714, L171Trager S. C., Faber S. M., Worthey G., Gonz´alez J. J., 2000, AJ,119, 1645Tonry J. L., Dressler A., Blakeslee J. P., Ajhar E. A., FletcherA. B., Luppino G. A., Metzger M. R., Moore C. B., 2001, ApJ,546, 681Tully R. B, 1988, Nearby Galaxy Catalog (NGC), CambridgeUniversity Press, Cambridgevan Dokkum P. G., 2005, AJ, 130, 2647van Driel W., van Woerden H., 1991, A&A, 243, 71Weijmans A.-M., Krajnovi´c D., van de Ven G., Oosterloo T. A.,Morganti R., de Zeeuw P. T., 2008, MNRAS, 383, 1343Welch G. A., Sage L. J., 2003, ApJ, 584, 260Wrobel J. M., Heeschen D. S., 1991, AJ, 101, 148Yi S. K., et al., 2005, ApJ, 619, L111Young L. M., Bureau M., Cappellari M., 2008, ApJ, 676, 317Young L. M. et al. in prep. c (cid:13)000