The Survey of Centaurus A's Baryonic Structures (SCABS). II. The Extended Globular Cluster System of NGC5128 and its Nearby Environment
Matthew A. Taylor, Thomas H. Puzia, Roberto P. Muñoz, Steffen Mieske, Ariane Lançon, Hongxin Zhang, Paul Eigenthaler, Mia Sauda Bovill
MMNRAS , 1–26 (2016) Preprint 24 April 2018 Compiled using MNRAS L A TEX style file v3.0
The Survey of Centaurus A’s Baryonic Structures(SCABS). II. The Extended Globular Cluster System ofNGC 5128 and its Nearby Environment
Matthew A. Taylor , (cid:63) , Thomas H. Puzia , Roberto P. Mu˜noz , Steffen Mieske ,Ariane Lan¸con , Hongxin Zhang , Paul Eigenthaler , and Mia Sauda Bovill Institute of Astrophysics, Pontificia Universidad Cat´olica de Chile, Av. Vicu˜na Mackenna 4860, 7820436 Macul, Santiago, Chile European Southern Observatory, Alonso de Cordova 3107, Vitacura, Santiago, Chile Observatoire astronomique de Strasbourg, Universit´e de Strasbourg, CNRS, UMR 7550, 11 rue de l’Universite, F-67000 Strasbourg, France Space Telescope Science Institute, 3700 San Martin Drive, 21218, Baltimore, Maryland, USA
Accepted XXX. Received YYY; in original form 24 April 2018
ABSTRACT
New wide-field u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) Dark Energy Camera observations centred on the nearby giantelliptical galaxy NGC 5128 covering ∼
21 deg are used to compile a new catalogue of ∼ ∼
140 kpc of NGC 5128. We find evidence for a transitionat a galactocentric radius of R gc ≈
55 kpc from GCs “intrinsic” to NGC 5128 to thoselikely to have been accreted from dwarf galaxies or that may transition to the intra-group medium of the Centaurus A galaxy group. We fit power-law surface numberdensity profiles of the form Σ
N,R gc ∝ R Γgc and find that inside the transition radius,the red GCs are more centrally concentrated than the blue, with Γ inner , red ≈ − . inner , blue ≈ − .
40, respectively. Outside this region both profiles flatten, moredramatically for the red GCs (Γ outer , red ≈ − .
33) compared to the blue (Γ outer , blue ≈− . g (cid:48) − z (cid:48) ) = 1 .
27 mag colour of the inner red population is consistentwith arising from the amalgamation of two giant galaxies each less luminous thanpresent-day NGC 5128. Both in- and out-ward of the transition radius, we find thefraction of blue GCs to dominate over the red GCs, indicating a lively history of minor-mergers. Assuming the blue GCs to originate primarily in dwarf galaxies, we modelthe population required to explain them, while remaining consistent with NGC 5128’spresent-day spheroid luminosity. We find that that several dozen dwarfs of luminosities L dw ,V (cid:39) − . L V, (cid:12) , following a Schechter luminosity function with a faint-end slopeof − . (cid:46) α (cid:46) − .
25 is favoured, many of which may have already been disrupted inNGC 5128’s tidal field.
Key words: galaxies: star clusters: general – galaxies: elliptical and lenticular, cD –galaxies: formation – galaxies: individual: NGC 5128
Resolved stellar population studies are the best direct probesof a galaxy’s evolutionary past, but today’s instrumentationlimits these techniques to galaxies just outside the LocalGroup. Luckily, the formation environment of a galaxy isforever encoded in the ubiquitous globular cluster (GC) sys-tems that surround them. Stars generally form in a clus-tered fashion governed by their environmental conditions(e.g. Lada & Lada 2003; Mac Low & Klessen 2004; Porte- (cid:63)
E-mail: [email protected] (MAT) gies Zwart, McMillan & Gieles 2010), which dissolve onGyr timescales for low-mass clusters ( M (cid:63) < ∼ ), but thecompact configurations of massive clusters ( M (cid:63) > ∼ ) en-able their survival, while their relatively high luminosities( M V (cid:46) − . c (cid:13) a r X i v : . [ a s t r o - ph . GA ] A p r M. A. Taylor et al.
Figure 1.
The spatial footprint of the SCABS observations, with the fields studied in this work marked as blue tiles. The position ofNGC 5128 is shown by the green star, while the surrounding cloud of black dots indicates the population of confirmed GCs. The reddashed ellipse shows the ∼
300 kpc extent of NGC 5128’s virial radius, and the tiles considered in this work cover ∼ −
150 kpc ingalactocentric radius. Different tiles are indicated by the numbers shown, with tile 1 centred on NGC 5128 itself.
Harris 1991; Ashman & Zepf 1998; Brodie & Strader 2006;Ashman & Zepf 2008; Peng et al. 2008; Georgiev et al. 2010).The well-known bimodal colour/metallicity distribu-tions of GCs (e.g. Searle & Zinn 1978; Zepf & Ashman 1993;Ostrov et al. 1993; Whitmore et al. 1995; Elson & Santi-ago 1996; Gebhardt & Kissler-Patig 1999; Puzia et al. 1999,2004, 2005a, 2006; Kundu & Whitmore 2001; Larsen et al.2001; Peng et al. 2006, 2011; Spitler et al. 2006; Goudfrooijet al. 2007; Brodie et al. 2012; Usher et al. 2012, 2015) act asprobes of galactic formation histories (e.g. Brodie & Strader2006, for a review). This bimodality has been interpretedin two ways. The merger scenario sees the blue/metal-poorGCs representing those that formed in-situ from pristine gasalong with their giant hosts, while the red/metal-rich peakcorresponds to GCs that formed later in gas-rich merger in-duced starbursts (e.g. Ashman & Zepf 1992; Kissler-Patig1997; Forbes et al. 1997b; Beasley et al. 2002; Spitler et al.2006). Conversely, in the hierarchical build-up scenario, itmay be that the red GCs are the primordial GCs that formand are rapidly enriched in the dense star-forming environ-ments of their giant hosts, while blue GCs are accreted fromdwarf galaxies during minor-merger events (e.g. Cˆot´e et al. 1998, 2000; D’Abrusco et al. 2014a,b, 2015). The latter in-terpretation seems to be supported by the observation thatred GCs often coincide with the underlying bulge light oftheir giant hosts, while the blue GCs tend to be more preva-lent in the outer galaxy halos (e.g. Geisler et al. 1996; Forbeset al. 1997b; Ashman & Zepf 1998; Cˆot´e et al. 1998, 2000;Forbes et al. 2001; Bassino et al. 2006; Brodie & Strader2006; Spitler et al. 2006; Goudfrooij et al. 2007; Faifer etal. 2011; Forbes et al. 2012; D’Abrusco et al. 2014a,b, 2015;Kartha et al. 2016), as well as the fact that dwarf galaxies areknown to host primarily metal-poor GCs (Peng et al. 2006;Puzia & Sharina 2008) that would tend to deposit their GCsin the giants host’s halo. In any case, knowledge of the dis-tributions in both colour and space of both GCs and dwarfgalaxies around a giant galaxy provides useful leverage onunderstanding the ancient environment that gave rise to it.NGC 5128 is the dominant giant elliptical (gE) galaxy inthe nearby (3 . ± . MNRAS000
Harris 1991; Ashman & Zepf 1998; Brodie & Strader 2006;Ashman & Zepf 2008; Peng et al. 2008; Georgiev et al. 2010).The well-known bimodal colour/metallicity distribu-tions of GCs (e.g. Searle & Zinn 1978; Zepf & Ashman 1993;Ostrov et al. 1993; Whitmore et al. 1995; Elson & Santi-ago 1996; Gebhardt & Kissler-Patig 1999; Puzia et al. 1999,2004, 2005a, 2006; Kundu & Whitmore 2001; Larsen et al.2001; Peng et al. 2006, 2011; Spitler et al. 2006; Goudfrooijet al. 2007; Brodie et al. 2012; Usher et al. 2012, 2015) act asprobes of galactic formation histories (e.g. Brodie & Strader2006, for a review). This bimodality has been interpretedin two ways. The merger scenario sees the blue/metal-poorGCs representing those that formed in-situ from pristine gasalong with their giant hosts, while the red/metal-rich peakcorresponds to GCs that formed later in gas-rich merger in-duced starbursts (e.g. Ashman & Zepf 1992; Kissler-Patig1997; Forbes et al. 1997b; Beasley et al. 2002; Spitler et al.2006). Conversely, in the hierarchical build-up scenario, itmay be that the red GCs are the primordial GCs that formand are rapidly enriched in the dense star-forming environ-ments of their giant hosts, while blue GCs are accreted fromdwarf galaxies during minor-merger events (e.g. Cˆot´e et al. 1998, 2000; D’Abrusco et al. 2014a,b, 2015). The latter in-terpretation seems to be supported by the observation thatred GCs often coincide with the underlying bulge light oftheir giant hosts, while the blue GCs tend to be more preva-lent in the outer galaxy halos (e.g. Geisler et al. 1996; Forbeset al. 1997b; Ashman & Zepf 1998; Cˆot´e et al. 1998, 2000;Forbes et al. 2001; Bassino et al. 2006; Brodie & Strader2006; Spitler et al. 2006; Goudfrooij et al. 2007; Faifer etal. 2011; Forbes et al. 2012; D’Abrusco et al. 2014a,b, 2015;Kartha et al. 2016), as well as the fact that dwarf galaxies areknown to host primarily metal-poor GCs (Peng et al. 2006;Puzia & Sharina 2008) that would tend to deposit their GCsin the giants host’s halo. In any case, knowledge of the dis-tributions in both colour and space of both GCs and dwarfgalaxies around a giant galaxy provides useful leverage onunderstanding the ancient environment that gave rise to it.NGC 5128 is the dominant giant elliptical (gE) galaxy inthe nearby (3 . ± . MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 trivial and the proximity of NGC 5128 acts as both an ad-vantage and disadvantage. On one hand, even the faintestGCs are observationally accessible; however, NGC 5128’shalo spreads across a large portion of the sky, requiring verywide-field imagers to probe into the extreme halo. For thisreason, the identification of the GC system has come in fitsand starts, with the first confirmed GC identified by Gra-ham & Phillips (1980), and handfuls of GCs found through-out the balance of the 1980s and into the 1990s (e.g. van denBergh et al. 1981; Harris et al. 1984a,b; Hesser et al. 1984,1986; Sharples, R. 1988; Harris et al. 1988; Minniti et al.1996; Holland et al. 1999). With the turn of the millennium,GCs finally began to be discovered en masse with dozensidentified through deep imaging with multiple broad-bandoptical filters (e.g. Rejkuba 2001; Peng 2003; Harris et al.2004a; Peng et al. 2004a; G´omez et al. 2006). More recently,Harris et al. (2012) (hereafter H12) made use of exceptional( ∼ . (cid:48)(cid:48) ) seeing to image 1.55 deg ( ∼ ×
90 kpc ) centredon the galaxy and identified (cid:38)
800 GC candidates based on B - and R -band optical photometry. These candidates are oftremendous use for large-scale spectroscopic follow-up, butthe uncertain level of foreground and background contam-ination and restricted spectral energy distribution (SED)coverage limits its utility to characterize the overall GC sys-tem of NGC 5128 with a large degree of confidence.Regardless, much has already been learned aboutNGC 5128 from the ∼
600 confirmed GCs (van den Bergh etal. 1981; Hesser et al. 1984, 1986; Harris et al. 1992; Jablonkaet al. 1996; Held et al. 1997; Peng et al. 2004c; Woodley etal. 2005; Rejkuba et al. 2007; Beasley et al. 2008; Wood-ley et al. 2010a). For example, in accords with other giantgalaxies, NGC 5128’s GCs show a bi- and possibly tri-modalcolour/metallicity distribution (Minniti et al. 1996; Held etal. 1997; Harris et al. 2002b; Peng et al. 2004c; Beasley etal. 2008; Spitler et al. 2008; Sinnott et al 2010; Woodley etal. 2010b), corresponding to at least two distinct GC pop-ulations. There is significant evidence supporting a recentmajor merger (Baade & Minkowski 1954; Graham 1979; In-nanen 1979; Tubbs 1980; Malin et al. 1983; Hesser et al.1986; Quillen et al. 1993; Minniti et al. 1996; Stickel et al.2004), which may have given rise to at least one new gen-eration of GCs being produced since the earliest years ofNGC 5128’s past (e.g. van den Bergh et al. 1981; Hesser etal. 1984, 1986; Harris et al. 1992; Peng et al. 2004c; Woodleyet al. 2005; Rejkuba et al. 2007; Beasley et al. 2008; Woodleyet al. 2010b). Despite the large amount of knowledge gainedfrom the known GCs around NGC 5128, the seemingly sim-ple question of the total population of GCs is still somewhatuncertain, with total estimates ranging from ∼ − (cid:48) (Harris et al. 1984a, 2002b, 2004b, 2006, 2010b). In fact,the spatial distribution shows significant variation (Wood-ley et al. 2007; Harris et al. 2012), indicating that the totalpopulation extends beyond 45 (cid:48) , possibly along a certain pref-erential axis which may trace unknown tidal features frompast mergers.In this work, we use new CTIO/DECam wide-field,five-band ( u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) ) photometry to identify GC candidateswithin ∼
140 kpc of NGC 5128. The data cover ∼
21 deg ,and are a subset of the ∼
72 deg “ Survey of Centaurus A’sBaryonic Structures ” (SCABS) imaging campaign presented in the first paper of this series (Taylor et al. 2016, hereafterPaper I.). In addition to presenting a new list of likely candi-dates, we also subject the candidate list of H12 to the sameselection criteria, showing that many are likely to be stellarin nature. We show via a preliminary analysis of the over-all GC system characteristics that our u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) photometryis sufficient to detect the majority of GC candidates within ∼
140 kpc with a high degree of confidence. With these can-didates, recent near-infrared (NIR) wide-field imaging andextra spectroscopic confirmations in the near future, a com-plete census of NGC 5128’s GC system is nearly at hand,along with the myriad lines of scientific inquiry that willcome with it.The paper is organized as follows. § § § §
6. Through-out this work we adopt a distance modulus for NGC 5128 of( m − M ) = 27 . ± .
05 mag, corresponding to a distance of3 . ± . The analysis presented in this contribution, as well as theresults and discussion based on it, are formed upon theSCABS dataset presented and described in Paper I. Webegin with the catalogue of sources with near-ultraviolet(NUV) and optical u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) photometry and morphologicalproperties produced by the SCABS data reduction proce-dures. These data consist of >
500 000 sources with com-plete sets of u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) photometry, all with on-sky coordi-nates corresponding to the inner ∼
21 deg of the Centau-rus A galaxy group, reaching out to ∼
150 kpc in galacto-centric radius from NGC 5128 (see Fig. 1). We note herethat, while Tables 2 and 4 list photometric data that are un-corrected for foreground extinction, in practice we de-reddenall sources on a point-by-point basis using the extinctionmaps of Schlafly & Finkbeiner (2011) and use these valuesfor the analysis.Altogether, this rich dataset is exploited to identifypoint-like and marginally resolved sources, the numbers ofwhich are reduced by two orders of magnitude to reveal alarge catalogue of likely GCs as described in the following.
This section describes the process of converting the cal-ibrated SCABS images into catalogues of point-like andmarginally resolved sources from which a GC candidate
MNRAS , 1–26 (2016)
M. A. Taylor et al. list is selected. The distance to NGC 5128 of 3.8 Mpc cor-responds to a physical scale of 18 pc/ (cid:48)(cid:48) . Given GC sizes of1 −
20 pc (0 . − . (cid:48)(cid:48) ) with a typical average half-light ra-dius of ∼ ∼ . (cid:48)(cid:48) ), and typical PSF FWHM of 1 − . (cid:48)(cid:48) in all bands during our observations, we expect most GCsto be point-like or marginally resolved with sizes accountingfor 4 −
111 percent of their PSF FWHM, or ∼
10 percentfor an average GC at 1 . (cid:48)(cid:48) FWHM.
A common strategy for identifying GCs in galaxy systemsoutside of the Local Group is to exploit their relatively lim-ited colour dispersions via colour-magnitude diagrams. Forexample, H12 used this technique to discover (cid:38)
800 newGC candidates around NGC 5128 using B and R photome-try. While this strategy is useful as a first guess, Mu˜noz et al.(2014) recently showed the dramatically increased effective-ness of identifying GCs in colour-colour diagnostic space cov-ering the NUV to near-infrared (NIR) SED range. Presently,there is no matching wide-field NIR photometry availablefor NGC 5128, so here we combine the u (cid:48) , r (cid:48) , and z (cid:48) filtersfrom our survey which maximize the separation of GCs frombackground objects and, in particular, foreground stars.The upper panel of Fig. 2 illustrates the utility of lever-aging the full optical SED for GC selection. The ( u (cid:48) − r (cid:48) ) –( r (cid:48) − z (cid:48) ) plane is shown for all 77 336 sources detected in allfive bands in tile 1, with the logarithmic number density indi-cated by the grey-scaled hexagonal bins. We recover 691/833GC candidates from the H12 catalogue, and over-plot themas red dots, while the 548/643 radial-velocity confirmed GCs(Woodley et al. 2010b, and references therein; Peng, E. W., private communication ) recovered in all five filters are plot-ted as green triangles. Finally, the blue stars in Fig. 2 showthe 369/373 confirmed foreground stars that were recovered(Peng et al. 2004a, Peng, E. W., private communication ).The key to this selection technique lies in the inclusionof the NUV u (cid:48) -band fluxes. While the effective temperaturesof cool stars and red giant branch (RGB) populations areprobed by the ( r (cid:48) − z (cid:48) ) colour, the ( u (cid:48) − r (cid:48) ) index strad-dles the 4000 ˚A break and is thus sensitive to hot stellarpopulations in star-forming galaxies and/or hot horizontalbranch (HHB) stars in GCs. The u (cid:48) -band flux contributedby HHB stars in metal-poor GCs thus makes them appearbluer in ( u (cid:48) − r (cid:48) ) than foreground stars at a given ( r (cid:48) − z (cid:48) ) ,while GCs lacking an extended horizontal branch will notexhibit bluer colours and fall along the redder areas of theGC sequence that is clearly separate from the stellar locus.The larger coloured symbols in the upper panel of Fig. 2show the redshift evolutionary tracks of the SED of a galaxyformed at z ≈
3, based on the
P´egase population syn-thesis models (Fioc & Rocca-Volmerange 1997). Four starformation histories are assumed for the simulated galaxy,including an initial starburst followed by passive evolution(burst+PE), a constant star formation rate (cSFR), an ex-ponentially declining SFR ( e SFR), and a galaxy formed viaan initial starburst followed by constant, low-level star for-mation (burst+lSFR). The pathways followed by the galax-ies through the u (cid:48) r (cid:48) z (cid:48) -plane differ significantly among them-selves, and give rise to the diffuse background populationtowards redder ( r (cid:48) − z (cid:48) ) colours at a given ( u (cid:48) − r (cid:48) ) . While the burst+PE model seen at low redshift closely follows theGCs, galaxies that experience virtually any residual star for-mation will remain satisfactorily distinct from GCs at all z .Since a canonical stellar initial mass function (e.g.Salpeter IMF or Chabrier IMF; Salpeter 1955; Chabrier2003) dictates that the stellar mass of a GC is dominated bycool, low-mass stars, their contribution to the overall lightis weak compared to those near the main sequence turnoff(MSTO) and the RGB, leading to the clear separation ofGCs from the plume of stars seen at redder ( r (cid:48) − z (cid:48) ) , whilebeing more similar at the MSTO. Altogether, the ( u (cid:48) − r (cid:48) ) –( r (cid:48) − z (cid:48) ) is the most effective optical two-colour diagnosticplane for GC selection. In practice, the upper panel of Fig. 2shows that the confirmed GCs populate a well-defined se-quence that falls redward of the ( r (cid:48) − z (cid:48) ) foreground stellarlocus, and are distinct from the MSTO of Galactic starsat redder colours. The reported H12 candidate sample iscontaminated by a large number of likely false-positives, asmany of them fall directly on the stellar locus (i.e. bluewardof the GC sequence), or towards redder ( r (cid:48) − z (cid:48) ) colourswhere the population of confirmed GCs is lower, and thelikelihood of background galaxy contamination increases.We augment our photometric colour selection techniquewith morphological information as shown in the bottompanel of Fig. 2, and discussed in more detail in § Source Extractor (SE) pa-rameter spread model against the apparent i (cid:48) -band mag-nitude for the same sources and symbol definitions as above.A relatively clean separation can be seen between fore-ground stellar sources that show a median spread model = 1 . ± . × − and the confirmed GCs which showmodestly larger median spread model = 5 . ± . × − .Interestingly, the H12 sample shows indications of heavyforeground star and background source contamination, asmany fall directly on the stellar spread model (cid:39) i (cid:48) – spread model combinations inconsistent withthe radial-velocity confirmed GC sample. The large numberof potential H12 GC imposters may be due to the use ofa single ( B − R ) colour, and we defer a quantification ofthis sample to § The first step in cleaning the source catalogues of nonGC-like objects is the removal of extended backgroundsources. To this effect, we use the SE / PSFEx parameters spread model and spreaderr model following a proce-dure similar to previous works (e.g. Desai et al. 2012; An-nunziatella et al. 2013; Koposov et al. 2015). spread model is an improvement on the older class star star–galaxy sep-aration parameter. It uses the local PSF model, Φ, to dis-criminate between star-like and more extended sources byconvolving it with a circular exponential disk based on thePSF model FWHM and calculating, spread model = Φ T x Φ T Φ − G T xG T Φ (1)where G is the more extended model and x is the image vec-tor centred on the source. In this way, spread model forms MNRAS000
P´egase population syn-thesis models (Fioc & Rocca-Volmerange 1997). Four starformation histories are assumed for the simulated galaxy,including an initial starburst followed by passive evolution(burst+PE), a constant star formation rate (cSFR), an ex-ponentially declining SFR ( e SFR), and a galaxy formed viaan initial starburst followed by constant, low-level star for-mation (burst+lSFR). The pathways followed by the galax-ies through the u (cid:48) r (cid:48) z (cid:48) -plane differ significantly among them-selves, and give rise to the diffuse background populationtowards redder ( r (cid:48) − z (cid:48) ) colours at a given ( u (cid:48) − r (cid:48) ) . While the burst+PE model seen at low redshift closely follows theGCs, galaxies that experience virtually any residual star for-mation will remain satisfactorily distinct from GCs at all z .Since a canonical stellar initial mass function (e.g.Salpeter IMF or Chabrier IMF; Salpeter 1955; Chabrier2003) dictates that the stellar mass of a GC is dominated bycool, low-mass stars, their contribution to the overall lightis weak compared to those near the main sequence turnoff(MSTO) and the RGB, leading to the clear separation ofGCs from the plume of stars seen at redder ( r (cid:48) − z (cid:48) ) , whilebeing more similar at the MSTO. Altogether, the ( u (cid:48) − r (cid:48) ) –( r (cid:48) − z (cid:48) ) is the most effective optical two-colour diagnosticplane for GC selection. In practice, the upper panel of Fig. 2shows that the confirmed GCs populate a well-defined se-quence that falls redward of the ( r (cid:48) − z (cid:48) ) foreground stellarlocus, and are distinct from the MSTO of Galactic starsat redder colours. The reported H12 candidate sample iscontaminated by a large number of likely false-positives, asmany of them fall directly on the stellar locus (i.e. bluewardof the GC sequence), or towards redder ( r (cid:48) − z (cid:48) ) colourswhere the population of confirmed GCs is lower, and thelikelihood of background galaxy contamination increases.We augment our photometric colour selection techniquewith morphological information as shown in the bottompanel of Fig. 2, and discussed in more detail in § Source Extractor (SE) pa-rameter spread model against the apparent i (cid:48) -band mag-nitude for the same sources and symbol definitions as above.A relatively clean separation can be seen between fore-ground stellar sources that show a median spread model = 1 . ± . × − and the confirmed GCs which showmodestly larger median spread model = 5 . ± . × − .Interestingly, the H12 sample shows indications of heavyforeground star and background source contamination, asmany fall directly on the stellar spread model (cid:39) i (cid:48) – spread model combinations inconsistent withthe radial-velocity confirmed GC sample. The large numberof potential H12 GC imposters may be due to the use ofa single ( B − R ) colour, and we defer a quantification ofthis sample to § The first step in cleaning the source catalogues of nonGC-like objects is the removal of extended backgroundsources. To this effect, we use the SE / PSFEx parameters spread model and spreaderr model following a proce-dure similar to previous works (e.g. Desai et al. 2012; An-nunziatella et al. 2013; Koposov et al. 2015). spread model is an improvement on the older class star star–galaxy sep-aration parameter. It uses the local PSF model, Φ, to dis-criminate between star-like and more extended sources byconvolving it with a circular exponential disk based on thePSF model FWHM and calculating, spread model = Φ T x Φ T Φ − G T xG T Φ (1)where G is the more extended model and x is the image vec-tor centred on the source. In this way, spread model forms MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 Figure 2.
The colour-colour and morphological GC selection diagrams. (
Top ): The ( r (cid:48) − z (cid:48) ) vs. ( u (cid:48) − r (cid:48) ) colours for all sources withcomplete u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) photometry in tile 1 is shown (grey-scale number density map), with the H12 GC candidates (red dots), the confirmedpopulation of GCs (green triangles), and known foreground stars (blue stars) over-plotted. The solid black curve represents a spline fitto the confirmed GCs and is meant to guide the eye. The large symbols mapped to the upper colour-bar show evolutionary tracks for theobserved colours of a galaxy formed at z ≈
3, assuming four different star formation histories (see § Bottom ): The morphology-luminosity plane for the same populations (see § , 1–26 (2016) M. A. Taylor et al.
Table 1.
Source catalogue summary. Columns list tiles 1–7 with the final column showing the totals of each type of source indicated bythe rows, with all sources shown in the first row, point and GC-like sources in the second, and GC candidates in the bottom row.tile 1 tile 2 tile 3 tile 4 tile 5 tile 6 tile 7 Total N src
77 336 67 349 60 627 64 316 77 528 87 345 82 348 516 849 N pt
68 434 58 167 51 336 51 876 66 780 74 861 67 495 438 949 N GC
761 362 300 256 344 327 326 2 676
Figure 3.
The galaxy–point source classification diagram. As inthe lower panel of Fig. 2, the logarithmic number density of allsources detected in tile 1 is indicated by the grey-scaled hexago-nal binning, and green triangles denote the confirmed GCs. Notethat many of the brighter GCs are marginally resolved at thedistance of NGC 5128, indicated by the slight departures from spread model (cid:39)
0. The red curves indicate the fine-tuning crite-ria of separating point and GC-like sources from the extendedbackground sources (see text). an effective discriminant between the modelled PSF and amore extended source at bright magnitudes, with blendingoccurring at fainter magnitudes that is quantified by therelated spreaderr model parameter.Fig. 3 shows this technique in practice, which shows thesame data as the lower panel of Fig. 2, but without knownstars or H12 candidates. Red curves show extra fine-tuningto account for the marginally resolved nature of GCs at thedistance of NGC 5128. True point sources are expected toshow spread model = 0 . ± (0 . spreaderr model )(Desai et al. 2012), and we combine these criteria with an ex-tra selection strategy to create our point and GC-like sourcecatalogue. We note here that the term “GC-like” is used torefer to all marginally resolved sources, which are further re-fined in § spread model = 0 . ± . µ ( m i (cid:48) ) = .
003 + (cid:15) ( m i (cid:48) ) m i (cid:48) ≤ . .
003 + (cid:15) ( m i (cid:48) ) + δ ( m i (cid:48) ) 16 . < m i (cid:48) < . .
003 + (cid:15) ( m i (cid:48) ) m i (cid:48) ≥ . m i (cid:48) is the apparent i (cid:48) -band PSF magnitude, (cid:15) ( m i (cid:48) ) is the running mean spreaderr model along bins in m i (cid:48) , and δ ( m i (cid:48) ) is the running median of spread model for the known GCs within the same bins. We calculate δ ( m i (cid:48) )for 16 . < m i (cid:48) < . δ ( m i (cid:48) ), we first 2 . σ -clip the binned GC data, and findthat restricting the range to a relatively conservative m i (cid:48) < . ± .
003 bounds on spread model , andthe dotted lines show the spread model ± (0 .
003 + (cid:15) ( m i (cid:48) ))relation, to which we fit a third degree polynomial and add δ ( m i (cid:48) ) for spread model > . ∼
90 percent) of confirmed GCs, and repeat the process fortiles 2–7 (see Fig. 1) to construct our final point and GC-like source catalogues, including marginally resolved sourcesthat fall within magnitudes typical of GCs. Using a list of157 confirmed background galaxies in tile 1 compiled usingthe
Nasa Extragalactic Database (NED ), we find that thisprocedure successfully culls 140 objects ( ∼ spread model consistent with beingunresolved. In general, we find that ∼ −
85% of the sourcesin the SCABS imaging are point or GC-like, and we list theirphotometric measurements in Table 2 with their associatedstatistical and systematic error estimates listed in Table 3.
Having effectively culled extended background sources fromthe catalogues, the next step is to differentiate between star-and GC-like objects. Similar to previous work (e.g. Peng etal. 2004a), we use the full optical SED coverage in colour-colour space to accomplish this task. While this colour-spaceis exploited to construct our final list of GC candidates,there is still some ambiguity between the stellar and GCloci toward bluer colours. The following describes a novelprobabilistic approach to separate GCs from stars includingefforts to quantify the level of confidence for each source wehave in our final GC candidate catalogue.
We first generate a catalogue of 369 spectroscopically con-firmed foreground stars from Peng et al. (2004a), supple-mented by newly confirmed foreground stars (Peng et al., Python package:
NumPy/poly1d http://ned.ipac.caltech.edu MNRAS000
NumPy/poly1d http://ned.ipac.caltech.edu MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 Table 2.
Catalogue of point and GC-like sources. Cols. (1) and (2) list on-sky coordinates, followed by their u (cid:48) , g (cid:48) , r (cid:48) , i (cid:48) , and z (cid:48) PSFmagnitudes in cols. (3)–(7). Listed magnitudes are not corrected for foreground reddening. This table is available in its entirety inmachine-readable form. α ( J δ ( J u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) ( hh : mm : ss ) ( ◦ : (cid:48) : (cid:48)(cid:48) ) (mag) (mag) (mag) (mag) (mag)13:10:38.71 − − − − − − − − − − Table 3.
Catalogue of point and GC-like source statistical and systematic photometric error budgets derived and described in detail inPaper I. The first two columns show coordinates corresponding to the source list in Table 2, followed by sets of two columns that listtheir statistical and systematic error estimates for each filter. This table is available in its entirety in machine-readable form. α ( J δ ( J δu (cid:48) stat δu (cid:48) sys δg (cid:48) stat δg (cid:48) sys δr (cid:48) stat δr (cid:48) sys δi (cid:48) stat δi (cid:48) sys δz (cid:48) stat δz (cid:48) sys ( hh : mm : ss ) ( ◦ : (cid:48) : (cid:48)(cid:48) ) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag)13:10:38.71 − − − − − − − − − − private communication ). From this catalogue, we model apopulation of foreground stars using the Besan¸con modelof stellar population synthesis of the Galaxy (Robin etal. 2003). For each SCABS tile, we query the Besan¸conmodel for differential counts of foreground stars in the range15 . ≤ m u (cid:48) ≤ . u (cid:48) − g (cid:48) ) , ( g (cid:48) − r (cid:48) ) ,( r (cid:48) − i (cid:48) ) , and ( i (cid:48) − z (cid:48) ) colours in the range of the knownstellar and GC populations. The queries are made overareas equivalent to the DECam footprint, centred on thecoordinates of tiles 1–7. The catalogue of known stars isthen randomly drawn from until the expected population isreached, and assigned their corresponding ( u (cid:48) − r (cid:48) ) ± δ u (cid:48) r (cid:48) and ( r (cid:48) − z (cid:48) ) ± δ r (cid:48) z (cid:48) colours where δ is drawn from a uniform U (0 , ∆) distribution and ∆ is the SE photometric error in agiven band. In this way, for tiles 1–7, the colour–colour spaceis populated by an expected population of foreground stars,based on photometry of known stellar sources.Similarly, a model of GCs is constructed based on the548 confirmed GCs with complete u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) photometry. Aswith the modelled foreground stars, we randomly draw fromthe sample of known GCs until the modelled populationreaches an assumed total population of GCs. With no priorknowledge of the true population, we assume a maximumexpected number of N pred = 2 000 based on previous popu-lation estimates for tile 1 (Harris et al. 1984a, 2002b, 2006,2010b), where the surface number density of GCs is pre-sumed to be highest, and a correspondingly lower number http://model.obs-besancon.fr for tiles 2–7. For the latter regions, we assume N pred = 200 touniformly sample the ( u (cid:48) − r (cid:48) ) –( r (cid:48) − z (cid:48) ) parameter spaceand minimize stochastic under-sampling of the u (cid:48) r (cid:48) z (cid:48) dia-gram. Having populated the u (cid:48) r (cid:48) z (cid:48) -diagram with modelled starsand GCs, the next step is to assign a probability for eachobject in the point source catalogue. For each source, a sur-rounding box is drawn based on its photometric errors. Thenumber of modelled GCs within the selection box, n GC , aswell as the number of modelled stars, n ∗ , are counted andused to calculate the probability of being a GC, P (GC) as, P (GC) = n GC n GC + n ∗ (2)and an initial rejection of any candidates with P (GC) < . u (cid:48) r (cid:48) z (cid:48) -diagramfor the sources with colour parameterizing P (GC). The samesolid black relation as in Fig. 2 follows the GC locus which ismade visible by the cloud of P (GC) (cid:39) . P (GC) toward bluer colours wherethe stellar sequence begins to dominate. The lower panelof Fig. 4 shows the distribution of P (GC) where the sametransition from foreground stars to GC candidates is seen toflatten out for P (GC) (cid:38) .
25, and sharply transitions to the P (GC) (cid:39) . MNRAS , 1–26 (2016)
M. A. Taylor et al.
Figure 4.
GC candidate probabilities for all tile 1 point- andGC-like sources. (
Top ): The u (cid:48) r (cid:48) z (cid:48) -diagram with symbol colourparameterizing P (GC). P (GC) ≈ . Bottom ): Thedistribution of P (GC) for the whole sample (grey) and final GCcandidates (blue). The black dashed line denotes the initial cuton P (GC) before further refinement is performed. The final listis dominated by high P (GC) candidates, with ∼ P (GC) ≤ . panel, the cloud of P (GC) ≈ . spread model ≤ . P (GC) selection procedure serves to cull (cid:38)
90 percent of point-likesources, with an additional ∼
80 percent removed due totruly point source morphologies. For tiles 2-7, where the sur-face number density of GCs is expected to be dramaticallylower, we find a correspondingly, and consistently, highercolour/ P (GC) cull rate of ∼
98 percent, with an additional ∼ −
80 percent removed on the second spread model filtering.
We investigate the effects of changing the assumed total pop-ulation, N pred , and show the results in Fig. 5. The GC selec-tion procedure is run for the central tile (see Fig. 1) and oneof the outer tiles, but with the assumed true population ofGCs, N pred , drawn from U (10 , U (10 , N GC for the individualiterations are shown by the small dots, which give rise tothe binned averages (connected dots) braced by the ± σ curves. We find that for tile 1, N GC climbs sharply for in-creasing N pred (cid:46) N pred = 4 000.For the Outer Ring, where the GC density is presumed to bedramatically lower, N GC shows a relatively stable plateau for200 (cid:46) N pred (cid:46) N GC – N pred correlation arises fromthe stochastic sampling of the GC locus. For example, if N pred is assumed to be too small, then the modelled GC lo-cus does not adequately sample the full u (cid:48) r (cid:48) z (cid:48) colour space,resulting in smaller N GC . On the other hand, particularlyfor tiles 2-7, if one assumes too high N pred , then the mod-elled population risks oversampling regions that cut into thestellar locus, with N GC larger than one might expect forthe extreme halo of a gE. Given this result, we move for-ward with N pred = 2000 and 200, and account for potentialstochastic sampling using an iterative strategy before build-ing our final candidate catalogue as described below. Star clusters of stellar masses, M ∗ (cid:46) M (cid:12) typically dis-perse on Gyr timescales (see e.g., Portegies Zwart, McMil-lan & Gieles 2010, for a review), whereas M ∗ (cid:38) × M (cid:12) compact stellar systems enter the realm of ultra-compactdwarf galaxies (UCDs), where chemical and photometricproperties begin to diverge from typical GCs (e.g. Ha¸se-gan et al. 2005; Mieske et al. 2006, 2008a; Taylor et al.2010, 2015). Given typical stellar V -band mass-to-light ra-tios for Local Group GCs of Υ V ∗ (cid:39) . M (cid:12) L − (cid:12) (McLaugh-lin 2000; McLaughlin & Fall 2008; Strader et al. 2011),this range in M ∗ translates to V -band luminosities of0 . × (cid:46) L V /L (cid:12) (cid:46) . At a distance modulus of( m − M ) = 27 .
88 mag, these luminosities translate to ap-parent V -band magnitudes of 17 . (cid:46) m V / mag (cid:46) .
5. Withthis in mind, we remove candidates outside of this range in m V , assuming a transformation to V -band from g (cid:48) and r (cid:48) ofJester et al. (2005): V = g (cid:48) − . g (cid:48) − r (cid:48) ) − .
01 (3)Despite the robust GC selection procedure, there issome stochasticity remaining from the random nature of the
MNRAS000
MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 N pred N G C Central Tile ± σ Outer Ring ± σ Figure 5.
Variation in predicted GC candidates with a priori assumed populations. The GC selection technique is applied tothe data with several dozen realizations, each assuming a dif-ferent true population of GCs. Assumed populations are drawnfrom U (10 , U (10 , N GC ± σ results. N GC is seen to generally stabilize for N pred (cid:38) (cid:46) N pred (cid:46) stellar and GC modelling procedure. In other words, in agiven pass through the algorithm, sub-regions in u (cid:48) r (cid:48) z (cid:48) -spacemay be over- or under-sampled by the modelled GC andstellar populations, resulting in a handful of GC candidatesthat may be identified in one pass, but fail to be selected inanother. To account for this effect, we iterate the algorithmover each tile, and keep unique candidates in each iterationuntil we reach a pass that fails to identify any new GCs.Fig. 6 illustrates this procedure for tiles 1–7, with the growthof the final catalogues shown for each iteration. The resultsfor each tile are indicated by the different symbols/colours,which rapidly flatten as the final N GC are approached intypically (cid:46)
20 iterations. While a strict asymptotic limitis technically not reached, the results show that few newGC candidates would be identified with further iterations.We note that testing different assumed N pred (Fig. 5) showsthat the asymptotic behaviour persists regardless of assumedunderlying populations, but with faster convergence accom-panying larger assumed GC populations.Finally, we find some sources with large photometric er-rors relative to their magnitudes, often spatially correspond-ing to the DECam chip-gaps, are found to contaminate thefinal GC candidate catalogues. These sources are κσ -clippedout of the final samples, with κ tuned for each tile until thescatter in magnitude error versus magnitude relations areminimized in all filters, resulting in the slightly smaller totalcandidate counts shown in Table 1 compared to those indi-cated on Fig. 6. This last cut serves to typically reduce thefinal candidate lists by up to 25 percent. Iteration N G C Tile 1Tile 2Tile 3Tile 4Tile 5Tile 6Tile 7
Figure 6.
An illustration of the iterative procedure undertakento compile the final GC candidate catalogues. The ordinate showsthe number of candidates identified in a given tile for each itera-tion indicated along the abscissa, with individual tiles shown bythe different symbols/colours. Asymptotic behaviours are seen forall tiles, which all required several iterations before converging ontheir final values. ( u − r ) [mag] ( r − z ) [ m a g ]
17 18 19 20 21 i [mag] S P R E A D _ M O D E L _ i New CandidatesConfirmed GCsH12 Candidates 0.320.400.480.560.640.720.800.880.96 G C C a n d i d a t e P r o b a b ili t y Figure 7.
The colour-colour and morphological diagrams for thenew and previously reported GCs and candidates. As in Fig. 2,green triangles and red circles show confirmed GCs and H12 can-didates, respectively, that survive our colour-selection criteria,while dots represent the new GC candidates, which are colourcoded to indicate P (GC). The total GC candidate numbers ( N GC ) in Table 1 representall survivors of the selection procedure described above, in-cluding those already known in the literature. Running thecatalogue of Woodley et al. (2010b) through our selectionprocedure results in 230/548 surviving GCs (a ∼
60 percent
MNRAS , 1–26 (2016) M. A. Taylor et al.
Table 4.
Catalogue of new GC candidates. Col. (1) lists new, homogenized identifications, while col. (2) lists former IDs and/or thosereported in H12 (arbitrarily numbered). Cols. (3) and (4) list on-sky coordinates, with GC “probabilities”, galactocentric radii, andazimuthal angles in cols. (5)–(7). Cols. (8)–(17) list u (cid:48) , g (cid:48) , r (cid:48) , i (cid:48) , and z (cid:48) PSF magnitudes without accounting for foreground extinction.ID ID α ( J δ ( J P GC R gc Φ u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) (T17) (W10/H12 ) ( hh : mm : ss ) ( ◦ : (cid:48) : (cid:48)(cid:48) ) (arcmin) ( ◦ EoN) (mag) (mag) (mag) (mag) (mag)T17-0001 ... 13:10:45.60 − − − − − − − − − − Table 5.
GC candidate photometric errors. The first three columns show IDs and coordinates corresponding to the source list in Table 4,followed by sets of two columns that list the statistical and systematic error estimates for each filter, as described in § α ( J δ ( J δu (cid:48) stat δu (cid:48) sys δg (cid:48) stat δg (cid:48) sys δr (cid:48) stat δr (cid:48) sys δi (cid:48) stat δi (cid:48) sys δz (cid:48) stat δz (cid:48) sys (T17) ( hh : mm : ss ) ( ◦ : (cid:48) : (cid:48)(cid:48) ) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag)T17-GC0001 13:10:45.60 − − − − − − − − − − culling fraction), while we can only say with confidence that232/691 of the H12 candidates are truly GCs (a ∼
67 percentcull). Cross matching the 761 GC candidates in tile 1 withthe 643 confirmed GCs (Woodley et al. 2010b, Peng et al., private communication ) reveals 251 recovered clusters, withan additional 21 overlapping the surviving H12 candidatelist. Subtracting these 272 candidates from our catalogueleaves a total of 2 404 new GC candidates, a subsample ofwhich is listed in Tables 4 and 5 along with spatial informa-tion and photometric measurements.While it is beyond the scope of this work to determinethe true nature of those H12 candidates that did not sur-vive our selection procedure , it is worth mentioning two“clumps” of GC candidates reported by H12 between az-imuthal angles Φ ≈ − ◦ , and at galactocentric radii R gc = 20 and 40 kpc. Our final catalogue of likely GCs showsa slight enhancement of GC candidates corresponding tothe “clump” nearest to NGC 5128, but no compelling over-density is seen at R gc ≈
40 kpc. As noted by the authors,these “clumps” are likely to be background galaxy clustersthat appear in their B + R imaging as GCs, and further il-lustrates the importance of wide SED sampling when usingcolours to search for GCs.Cross-matching the 232 likely H12 GCs with the con-firmed GCs leaves 199 unique candidates. Considering these,the 643 previously confirmed GCs, and our 2 404 new can- This analyze the nature of these object will be part of subse-quent spectroscopic studies. didates implies that a total approaching N GC,t ≈ ∼
140 kpc of NGC 5128. This numberis higher than the total population of 1000–2000 predicted inprevious works (e.g. Harris et al. 1984a, 2004b, 2006, 2010b),but we note that these works consider a markedly smallerspatial scale for NGC 5128’s halo than we examine in thiswork. In fact, if only the GCs and candidates within a pro-jected 50 (cid:48) (55 kpc) of NGC 5128 are considered, outside ofwhich there are indications of a transition to a different GCpopulation (see § N GC,t drops to ∼ § u (cid:48) r (cid:48) z (cid:48) colour-colour diagram with colour illustrating P (GC)for the new GC candidates, while the confirmed GCs andsurviving H12 candidates shown again as green triangles,and red dots. The bottom panel shows the same samples inthe morphological classification diagram. The transition to-ward lower P (GC) can be seen toward bluer colours, andas constructed, the swarm of new GC candidates closelyapproximates the sequence of confirmed GCs. We note inthe top panel that there are a handful of GCs that exhibit spread model parameters < . MNRAS000
140 kpc of NGC 5128. This numberis higher than the total population of 1000–2000 predicted inprevious works (e.g. Harris et al. 1984a, 2004b, 2006, 2010b),but we note that these works consider a markedly smallerspatial scale for NGC 5128’s halo than we examine in thiswork. In fact, if only the GCs and candidates within a pro-jected 50 (cid:48) (55 kpc) of NGC 5128 are considered, outside ofwhich there are indications of a transition to a different GCpopulation (see § N GC,t drops to ∼ § u (cid:48) r (cid:48) z (cid:48) colour-colour diagram with colour illustrating P (GC)for the new GC candidates, while the confirmed GCs andsurviving H12 candidates shown again as green triangles,and red dots. The bottom panel shows the same samples inthe morphological classification diagram. The transition to-ward lower P (GC) can be seen toward bluer colours, andas constructed, the swarm of new GC candidates closelyapproximates the sequence of confirmed GCs. We note inthe top panel that there are a handful of GCs that exhibit spread model parameters < . MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 not losing significant numbers from our final foreground starculling morphological cut. For the first time, a catalogue of confirmed GCs bolsteredby a near-complete list of GC candidates reaching into theextreme halo of NGC 5128 is in-hand. Noting the homoge-neous areal coverage of tiles 1–7, attention now turns to thecolour and spatial characteristics of the catalogue. The wellestablished bimodal colour distributions of GC systems (e.g.Searle & Zinn 1978; Zepf & Ashman 1993; Ostrov et al. 1993;Whitmore et al. 1995; Elson & Santiago 1996; Gebhardt &Kissler-Patig 1999; Kundu & Whitmore 2001; Larsen et al.2001; Brodie & Strader 2006; Peng et al. 2006; Spitler etal. 2006; Goudfrooij et al. 2007; Richtler et al. 2012) areoften used to broadly infer the metallicities of the underly-ing GC stellar populations in that red GCs represent moremetal-rich populations, while bluer GCs imply lower metal-licities (e.g. Puzia et al. 2005b). In one paradigm, red GCspresumably form and are enriched during the initial giantstarbursts that lead to the buildup of the majority of theirhost galaxy’s stellar content. On the other hand, blue GCspreferentially form in more quiescent environments and/orshallower host potential wells (i.e. dwarf halos, see Cˆot´e etal. 1998, 2000; D’Abrusco et al. 2014a,b, 2015). However,this notion is under dispute, with lines of evidence suggest-ing that blue GCs form from pristine gas in situ with theirgiant hosts (e.g. Ashman & Zepf 1992; Forbes et al. 1997b;Beasley et al. 2002; Spitler et al. 2006) with the red peakcorresponding to later merger-induced starbursts.With the above in mind, understanding GC luminosityand colour distributions, together with spatial variations, isuseful to infer overall properties of their environment. Thefollowing sections discuss these distributions, beginning withcolours/luminosities of the GC candidate catalogue (includ-ing the confirmed GCs and H12 survivors), followed by aspatial analysis. Finally, the two are combined in an effortto search for potential colour- and space-dependent patternsand/or features, as well as their implications on the assemblyhistory of NGC 5128 and its surroundings.
The GC luminosity function (GCLF) describes the luminos-ity distribution of GCs, and is typically of log-normal formwith a near-universal peak, or turnover, near M V ≈ − . M ∗ ≈ M (cid:12) ; e.g. Jacoby et al. 1992; Richtler 1995; Harris2001; McLaughlin & van der Marel 2005; Brodie & Strader2006). The five panels of Fig. 8 illustrate the GCLFs for the u (cid:48) , g (cid:48) , r (cid:48) , i (cid:48) , and z (cid:48) filters. In all panels, the grey histogramsshow the distributions for GCs/candidates, which we esti-mate with non-parametric Epanechnikov-kernel probabilitydensity estimates (KDEs; solid black curves) to the 90 per-cent completeness limited data (light red shading). We in-dicate the KDE peaks by the dashed black lines, which weuse to fit N ( µ, σ ) relations to the bright/high-mass and Python package: scikit-learn Python package:
SciPy/optimize faint/low-mass candidates. First, we fit the bright GC candi-dates by mirroring them about the KDE peaks, and use thepeaks and sample σ ’s as initial guesses in the optimization.The solid blue curves show the resulting N ( µ, σ ) curves fit tothe bright candidates, with the fit parameters (ˆ µ , ˆ σ ) listedbelow the sample values in each panel. The dashed green andblue lines indicate the sample medians, ˜ µ , and means, ˆ µ , re-spectively. We then fit the faint samples by mirroring themabout the KDE peaks, with the same initial guesses. Theresulting dotted blue lines fail to represent the bright candi-dates well, and may be indicative of incompleteness towardfainter magnitudes, despite the 90 percent completeness es-timates. This incompleteness may be a result of the final cuton spread model , which may remove some faint, point-likeGC candidates along with the foreground stars (see Fig. 3).In any case, this effect cannot be improved upon with theoptical data without introducing significant stellar contam-ination to the catalogues, and will be discussed in greaterdetail in subsequent papers of this series, once our NIR dataof the central SCABS field is fully analyzed. The combina-tion of optical and NIR data will allow us to select GCs withthe uiK technique (see Mu˜noz et al. 2014) and reduce thecontamination by foreground stars and background galaxiesto a few percent.We convert the ( g (cid:48) − r (cid:48) ) to the V -band using Eq. 3and find that the V -band sample median of ˜ µ = − . ± .
02 mag is ∼ .
36 mag brighter than the typical peak of ∼ − .
09 mag (after correcting for A V = 0 .
315 mag of fore-ground extinction; Schlafly & Finkbeiner 2011), which maybe affected by the high-luminosity tail shown by the KDEfits. Conversely, a N ( − . , .
13) fit to the bright GCs showsa mean M V that concurs with the peak KDE to the data,and recovers the expected V -band GCLF peak of − .
09 mag.With that said, the non-parametric KDEs clearly follow thedata more closely than the normal distributions, particularlyin the case of the high-luminosity tails that are seen in eachband.The apparent bright tails might indicate that NGC 5128is overabundant in massive GCs and/or UCDs, or that thecandidate list still suffers some contamination from stellarsources. Given the dominance of P (GC) = 1 . P (GC) ≤ . (cid:46) ∼
60 percent of the confirmed GC population mightintroduce a selection bias that could give rise to the high-luminosity tail. To check against this effect, the grey dottedhistograms in Fig. 8 show the ∼
300 bona-fide GCs thatdid not survive the selection algorithm, but were nonethe-less recovered in our imaging. It is important to note thatthis sample is already represented in the grey shaded dis-tributions, and the dotted histograms are re-normalized tobetter compare to the overall sample. In each filter, the un-recovered, but bona-fide GCs show the high-luminosity tail,suggesting that the overall sample statistically resembles thebright end of NGC 5128’s GCLF, and that it is likely to beoverabundant in massive GCs/UCDs. In any case, furtherrefinement of the GC candidate catalogue will await either
MNRAS , 1–26 (2016) M. A. Taylor et al. P ( M u ) ˜ µ u = -5.56 ± σ u = 0.94 ± ˆ µ u = -5.34 ± ˆ σ u = 1.19 ± µ KDE = -5.22 u -band
90% 50%15.9 16.9 17.9 18.9 19.9 20.9 21.9 22.9 23.9 24.9
Apparent Magnitude P ( M g ) ˜ µ g = -7.11 ± σ g = 0.94 ± ˆ µ g = -6.86 ± ˆ σ g = 1.15 ± µ KDE = -6.77 g -band P ( M r ) ˜ µ r = -7.69 ± σ r = 0.91 ± ˆ µ r = -7.36 ± ˆ σ r = 1.09 ± µ KDE = -7.36 r -band P ( M i ) ˜ µ i = -8.08 ± σ i = 0.90 ± ˆ µ i = -7.83 ± ˆ σ i = 1.03 ± µ KDE = -7.81 i -band
12 11 10 9 8 7 6 5 4 3
Absolute Magnitude P ( M z ) ˜ µ z = -8.18 ± σ z = 0.89 ± ˆ µ z = -7.85 ± ˆ σ z = 1.07 ± µ KDE = -7.91 z -band Sample Median, ˜ µ Fit Mean, ˆ µ Bright GCLFFaint GCLFEpanechnikov KDEKDE Peak
Figure 8.
Globular cluster candidate luminosity functions in the u (cid:48) , g (cid:48) , r (cid:48) , i (cid:48) , and z (cid:48) filters. Grey shading shows sample distributionsusing an optimal binning technique (Knuth 2006), and non-parametric Epanechnikov-kernel probability density estimates are shown bythe solid black curves, limited to the 90% completeness-limited data for the GCLFs (light red shading). Listed in the upper left cornersare sample medians (dashed green lines) and standard deviations, alongside their 1 σ bootstrapped error estimates. Alternatively, thesolid blue curves show N ( µ, σ ) fits to the brightest/most massive candidates mirrored about the peak of the KDE (dashed black lines),with the corresponding ˆ µ shown by the dashed blue lines and listed with the fit ˆ σ . Similarly, the dotted blue curve shows a normal fit tothe faintest candidates mirrored about the peak of the KDE, which does not represent the overall sample well. Lastly, we show by thedotted gray histograms the re-normalised distributions of previously confirmed GCs that failed to satisfy our selection algorithm, notingthat these GCs are included in the solid shaded histograms. Point-source completeness levels corresponding to ≤
50% is indicated by thedarker red shading. large-scale spectroscopic campaigns and/or the extension ofthe photometric SED coverage.Dynamical friction arguments imply that the GCLFturnover for dwarf galaxies should become slightly fainterover time with respect to giants (e.g. Lotz et al. 2001; Bekki 2010). Observational evidence suggests that dwarfgalaxy GCLF peaks can be as much as M V ∼ . M V ≈ − . MNRAS000
50% is indicated by thedarker red shading. large-scale spectroscopic campaigns and/or the extension ofthe photometric SED coverage.Dynamical friction arguments imply that the GCLFturnover for dwarf galaxies should become slightly fainterover time with respect to giants (e.g. Lotz et al. 2001; Bekki 2010). Observational evidence suggests that dwarfgalaxy GCLF peaks can be as much as M V ∼ . M V ≈ − . MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 M p e a k u u -band M p e a k g g -band M p e a k r r -band M p e a k i i -band R gc [arcmin] M p e a k z z -band Figure 9.
Evolution of the GCLF peaks with galactocentricradius. The peak magnitudes of the u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) GCLFs are de-rived from KDE fits to GCs binned in rolling R gc ± (cid:48) probingNGC 5128’s halo in 1 (cid:48) steps. that there also exists evidence to the contrary, where dwarfGCLF peaks–at least for dwarf irregulars–are indistinguish-able from giants (e.g. Seth et al. 2004; Strader et al. 2006;Georgiev et al. 2009). Fig. 9 investigates this idea by look-ing at the behaviour of the NGC 5128 GCLF shape as afunction of R gc . The GC candidates are binned in rollingwindows of ∆ R = 10 (cid:48) in steps of 1 (cid:48) , and the peak of aKDE fit is determined for each bin individually. The peaksare shown for all five GCLFs against R gc where a notice-able ∼ . − . g (cid:48) r (cid:48) i (cid:48) z (cid:48) filters inthe region 15 (cid:48) (cid:46) R gc (cid:46) (cid:48) , which becomes less pronouncedtoward the bluer bands. Beyond R gc ≈ (cid:48) , each declinesby ∼ . − . ∼ . − . u (cid:48) -band GCLF, but with a less dramatic∆ M u (cid:48) ≈ . − . § (cid:48) (cid:46) R gc (cid:46) (cid:48) region to a futurework. Distributions of each colour permutation are shown inFig. 10 with bluer filters corresponding to upper rows, andcolour indices probing wider SED coverage from left to right.Clear bi-modalities are visible from the binned data (greyhistograms), which are prominent when the NUV u (cid:48) filteris used (top row), and become less significant with smallerSED sampling width. Single, double, and triple component Gaussian mixture models (GMMs) are fit to each colourcombination, with the best model chosen based on the min-imization of the Bayesian Information Criterion, which isa penalized method of model selection (Schwarz 1978). Inall cases a two-component GMM is preferred (solid blacklines), which is decomposed into the dashed black lines andcorrespondingly coloured shading. The GMMs are used toclassify GCs as either blue or red, and ( r (cid:48) − i (cid:48) ) is the soleindex where the number ratio of blue to red GCs, ξ b / r , isbelow unity. It is clear that NUV photometry strongly aidsthe separation of blue and red GCs, as shown by the con-sistent ξ b / r = 1 . − .
21 when using NUV colour indices,compared to the widely varying ξ b / r predicted by non-NUVbased colours. We thus exploit the wide SED sampling ofthe ( u (cid:48) − z (cid:48) ) colour index to classify the GC catalogue andadopt ξ b / r = 1 .
16 for the total sample.We note that, like the GCLFs, the significant culling ofbona-fide GCs via the strict selection criteria risks introduc-ing biases into the overall colour distributions. To investi-gate this, we show by the dashed histograms in the upperpanels of Fig. 10 the u (cid:48) -based colour distributions for thoseconfirmed GCs that did not survive the selection algorithm.By comparison to the overall samples, it is clear that thebi-modality and relative blue/red peak heights are broadlyconsistent, indicating that it is unlikely that the selectioncriteria introduces strong biases in the colour distributionsand classification of the new GCs as blue or red.It is well established that the mean metallicity of GCsystems increases with the metallicity of their hosts, withcorrespondingly redder mean colours accompanying higherhost mass/luminosity/metallicity (e.g. van den Bergh 1975;Brodie & Huchra 1991; Forbes et al. 1997b; Larsen et al.2001; Burgarella et al. 2001; Lotz et al. 2004; Peng et al.2006). At the same time, Peng et al. (2006) showed that ξ b / r for gE galaxies in the Virgo cluster decreases with luminosityfor galaxies in the magnitude range − < M B < −
15 mag.Given NGC 5128’s M B = − . ξ b / r slightly less than what is presentlyobserved. With that said, a smaller ξ b / r ≈ .
11 has alreadybeen measured for a sample of 194 NGC 5128 GCs by Harriset al. (2004b); however, their results were based on metal-licity sensitive Washington photometry, and were limited to R gc = 45 (cid:48) , so a direct comparison to the present results mightnot be warranted. Even so, limiting our sample to only thoseGC candidates within 45 (cid:48) results in ξ b / r = 1 . ξ b / r shownby NGC 5128 then provides a strong clue to the dominantmechanism behind its mass assembly. Table 6 lists some sim-ple statistics of the blue and red components of NGC 5128’sGC sample, with information shown for the population asa whole, and separated based on proximity to the host. For Python package: scikit-learn
MNRAS , 1–26 (2016) M. A. Taylor et al. ( u − g ) [mag] p ( x ) ξ b / r = 1 . ( u − r ) [mag] ξ b / r = 1 . ( u − i ) [mag] ξ b / r = 1 . ( u − z ) [mag] ξ b / r = 1 . ( g − r ) [mag] p ( x ) ξ b / r = 2 . ( g − i ) [mag] ξ b / r = 1 . ( g − z ) [mag] ξ b / r = 1 . ( r − i ) [mag] p ( x ) ξ b / r = 0 . ( r − z ) [mag] ξ b / r = 1 . -0.25 0.00 0.25 0.50 0.75 ( i − z ) [mag] p ( x ) ξ b / r = 4 . Figure 10.
Globular cluster colour distributions. Optimally binned GC candidate distributions are shown by the grey shading for eachpossible colour combination with filters running through u (cid:48) , g (cid:48) , r (cid:48) , i (cid:48) , and z (cid:48) from top to bottom, and left to right. Over-plotted arebest-fit GMMs, which sum to the black curves shown. The ratios of blue GC candidates to red are indicated on each panel, which arebased on the automatic classification from the respective GMMs. e shows a normal fit to the faintest candidates mirrored about the peakof the KDE, which does not represent the overall sample well. The dashed gray histograms in the upper panels show the u (cid:48) -based colourdistributions of previously confirmed GCs that failed to satisfy our selection algorithm, noting that these GCs are included in the solidshaded histograms. Final blue/red classification is based on the ( u (cid:48) − z (cid:48) ) colour. comparison with the results of Peng et al. (2006), the medianand mean ( g (cid:48) − z (cid:48) ) colours are listed for each component andwe find a median blue ( g (cid:48) − z (cid:48) ) colour of ˜ µ blue ≈ .
90 magto be consistent with the expectation for a gE of NGC 5128’sluminosity in Virgo. The same cannot be said of the red GCs,in that at all radii ˜ µ red ≈ .
26 is ∼ . − .
06 mag bluerand more consistent with a Virgo gE fainter than NGC 5128.Taken together, this result might be suggestive of the bluepopulation having been assembled through long-term accre- tion processes and/or tidal stripping of dwarfs, whereas thered population was assembled by a major merger of galaxiesboth similar in mass.Further study of the GC colour distributions is deferredto § MNRAS000
06 mag bluerand more consistent with a Virgo gE fainter than NGC 5128.Taken together, this result might be suggestive of the bluepopulation having been assembled through long-term accre- tion processes and/or tidal stripping of dwarfs, whereas thered population was assembled by a major merger of galaxiesboth similar in mass.Further study of the GC colour distributions is deferredto § MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 α [J2000] δ [ J ]
17 kpc 55 kpc 132 kpc =
33 kpc NE -8 -7 -6 -5 -4 -3 -2 -1 Probability Density
Figure 11.
The spatial distribution of all globular cluster candidates. The coordinates of all GC candidates are shown by the small blackpoints, which strongly cluster around NGC 5128’s location in the centre of the figure. The colour map represents a 2D exponential-kernelprobability density estimate to the data, where the darker shading indicates regions of projected GC spatial over-densities. Over-plottedon the figure is the confirmed dwarf galaxy population (dark green dots; Cˆot´e et al. 1997; van den Bergh 2000; Karachentsev et al. 2007;Crnojevi´c et al. 2014, 2016), and newly identified dwarf candidates (green squares; M¨uller et al. 2015, 2017). To guide the eye, ellipsesare drawn at R gc = 15 (cid:48) , 50 (cid:48) , and 120 (cid:48) and labelled with the corresponding physical scales. We note the sharp decline in the probabilitydensity near the edge of the tile 1–7 fields-of-view.MNRAS , 1–26 (2016) M. A. Taylor et al.
Table 6.
Colour statistics for the GC catalogue. Col. (1) lists thesample being considered, followed by median and mean ( g (cid:48) − z (cid:48) ) colours for the blue and red components, and the ratio of blueGCs to red based on the ( u (cid:48) − z (cid:48) ) classification. All parametershave 0.01 mag 1 σ bootstrapped errors.Sample ˜ µ blue ˜ µ red µ blue µ red ξ b / r (mag) (mag) (mag) (mag)Total 0.90 1.26 0.92 1.28 1.16 R gc < (cid:48) R gc ≥ (cid:48) Table 7.
Results of the linear regression analysis for the GCcatalogue. Col. (1) lists the sample under consideration, whilecols. (2)–(7) show the best fit power slope and the associatedvariance score for the total sample within R gc ≤ (cid:48) , the inner( R gc (cid:46) (cid:48) ) sample, and the outer (70 (cid:48) (cid:46) R gc (cid:46) (cid:48) ) candidates.Sample Γ all r Γ inner r Γ outer r Total − − − − − − − − − An alternative but equally important property of a GC sys-tem is its spatial distribution. An intriguing trend of giantgalaxy GC systems is that redder GCs tend to cluster to-ward smaller R gc than their bluer counterparts that mayhave been deposited at larger radii during satellite accretionevents (e.g. Geisler et al. 1996; Forbes et al. 1997b; Ash-man & Zepf 1998; Cˆot´e et al. 1998, 2000; Forbes et al. 2001;Puzia et al. 2004; Bassino et al. 2006; Brodie & Strader 2006;Spitler et al. 2006; Goudfrooij et al. 2007; Faifer et al. 2011;Forbes et al. 2012; D’Abrusco et al. 2014a,b, 2015; Karthaet al. 2016).As a first look at the global distribution of GCs andcandidates around NGC 5128, Fig. 11 shows a scatter plot ofthe GC catalogue (black dots), with the colour indicating anon-parametric exponential-KDE with a 0 . ◦ bandwidth.While the distribution is generally unstructured, the GCdistribution closest to NGC 5128 appears slightly elongatedalong the major isophotal axis along azimuthal angles Φ =35 ◦ / ◦ (E leading N) of the host, as noted in previousworks (e.g. Harris et al. 2004b; Peng et al. 2004b; Woodleyet al. 2007, 2010a). Recognizing the absence of spatial biasin the present data, this effect appears to be a real feature ofNGC 5128’s GC system; however, the bias against the minoraxis disappears outside of ∼ − (cid:48) .Beyond R gc ≈ (cid:48) ( ∼
17 kpc), Fig. 11 shows a faint,broad over-density centred at ( α, δ ) ≈ (202 . ◦ , − . ◦ ), ex-tending in a counterclockwise arc to coordinates ( α, δ ) ≈ (200 . ◦ , − . ◦ ). An equivalent broad over density is notseen between R gc ≈ −
80 kpc NE of NGC 5128, but atenuous connection to another ∆ R gc ≈
10 kpc-thick arc canbe seen spiraling clockwise from ( α, δ ) ≈ (203 . ◦ , − . ◦ ) to( α, δ ) ≈ (200 . ◦ , − . ◦ ), N of the host. Interestingly, theNE arc projects directly across at least two known dwarfs(green dots) in the system, and could be associated with a Python package: scikit-learn disrupting dwarf in the system (Crnojevi´c et al. 2016, theirDw3), although the lack of 3D information precludes draw-ing conclusions on possible physical associations. While in-triguing, a more detailed analysis of the significance of thesefeatures is deferred to a future work.
Fig. 12 shows alternative visualizations of the spatial distri-butions of the GC candidates, considering only those within R gc ≤ (cid:48) to avoid artificial biases arising from inhomo-geneities at the edge of the SCABS tiles (1-7, see Figs. 1and 11). The top panel shows the projected radial surfacenumber density profile, Σ N ( R gc ), calculated by individuallybinning the total, blue, and red subsamples, and determin-ingΣ N ( R gc ) = N GC π (cid:0) R ,o − R ,in (cid:1) (4)in radial annuli, where R gc ,o and R gc ,in are the outer andinner radii of each annulus. A linear regression analysis isapplied to fit a linear relation in logarithmic space to deter-mine the power law shapes that best explain the data, withresults listed in Table 7. We note that in what follows weexclude the bins centred at R gc (cid:39) (cid:48) to avoid artificially bi-asing the results with the ring-type over-density mentionedabove, which is represented in the top panel of Fig. 12 bythe notable uptick shown by the open circles. First, a singlepower-law of the form,Σ N ( R gc ) ∝ R Γgc (5)is fit to all of the data (solid curves) whose r scores indicatethat a power-law slope of Γ = − .
22 explains 95 percentof the total sample variance, while Γ = − .
25 and − . N ( R gc ) profiles show power-laws with similar slopes. Apartfrom the sample as a whole, these relations do not appearto adequately explain the data. Splitting the GC candidatesample into “inner” ( R gc < (cid:48) ) and “outer” (60 (cid:48) ≤ R gc ≤ (cid:48) ) populations, and applying the linear regression analy-sis to each individually shows a different picture. Here thefits explain the data better, finding that Γ = − . − . − .
78 accounts for (cid:38)
98 percent of the total, blue,and red variances, respectively, for the inner sample. Mean-while, the fits for the outer subsamples all show shallowerΓ = − . − .
61, and − .
33 slopes, but with much morepenalized r scores that are likely to arise from the substruc-ture noted above.In accords with previous works on NGC 5128’s andother GC systems, we find that the inner red populationshows a steeper relations than the blue. Conversely, theouter samples show the opposite behaviour. All profiles flat-ten, but more sharply for the red GCs, its poor r = 0 . Python package: scikit-learn
MNRAS000
MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 R gc [arcmin] -2 -1 Σ N ( R g c ) [ a r c m i n − ] All GCsBlue GCsRed GCs Σ N,R gc < ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc ≥ ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc ≥ ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc ≥ ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc < ∝ R − . Σ N,R gc ≥ ∝ R − . Φ [deg E of N] P r o b a b ili t y D e n s i t y All GC Candidates Φ [deg E of N] P r o b a b ili t y D e n s i t y Inner GC Candidates Φ [deg E of N] P r o b a b ili t y D e n s i t y Outer GC Candidates
Figure 12.
Radial and azimuthal distributions of GC candidates. (
Top) : The Σ N ( R gc ) profile calculated along radial bins, with solidlines indicating power-law fits to all GCs with R gc ≤ (cid:48) , and dashed and dot-dashed curves showing fits to data inside and outside of ∼ (cid:48) . Black curves represent all GC candidates, while blue and red are coded to the respective GC subsamples in all panels. In all fits weexclude the anomalously high Σ N ( R gc ) points centred at R gc (cid:39) (cid:48) (open circles) to avoid a potential bias arising from the over densitydiscussed in the text. ( Second from top ): Distribution of Φ in units of degrees E from N with the grey shading representing the totalpopulation. Solid lines outline the binned data, while Epanechnikov-KDEs are shown by the dashed curves. (
Second from bottom ): Sameas the previous panel, but with blue/red histograms indicating the blue/red samples with R GC < (cid:48) . The coloured histograms are shownwith lowered opacities so that regions of overlap (darker shading) are highlighted, and solid curves show Epanechnikov-KDEs fit to thedata. ( Bottom ): Same as the previous panel, but showing only the blue/red samples with R GC ≥ (cid:48) . finding red giant branch stars to populate the halo out toat least 140 kpc (e.g. Crnojevi´c et al. 2013; Rejkuba et al.2014). Unfortunately, a robust quantitative comparison be-tween GC and stellar radial surface number density pro-files is complicated by differing tidal stripping timescales be-tween GCs and galaxy spheroid stars during merger events(e.g. Smith et al 2013) combined with the–possibly spatiallybiased–likely underestimate of the total GC population aris-ing from our conservative selection procedure. Nonetheless,we point out qualitatively that the flattening of Σ N beyond R gc ≈ − (cid:48) compares nicely to a flattening of the stellarprofile beyond this radius (Crnojevi´c et al. 2013). Mean-while, the difficulty in fitting a smooth power-law to GCsin the outer halo intriguingly hints at the strong spatialvariability of RGB metallicities out to R gc ≈ (cid:48) (Re-jkuba et al. 2014), especially if such features arise fromdisturbed substructure from prior merger events. Regard-less, a robust statistical comparison between GC and stellar spatial/metallicity distributions in NGC 5128’s extreme halodeserves its own dedicated effort, and so we must defer it toa future work. The bottom three panels of Fig. 12 show distributions ofazimuthal angle, Φ, in degrees East of North. The secondpanel from the top shows the distributions of the total pop-ulation (grey shading), with the blue and red samples over-plotted by respectively coloured solid lines. Epanechnikov-KDEs with 15 ◦ bandwidths are shown by the dashed lines.The “inner” and “outer” (based on a R gc = 50 (cid:48) cut) popu-lations are shown separately in the bottom two panels re-spectively, where the blue and red populations are indicatedby the respective colours with opacity increased to show re-gions of overlap. The data are again smoothed by 15 ◦ KDEs,which are represented by the solid blue and red relations.
MNRAS , 1–26 (2016) M. A. Taylor et al. ˆ w ( θ ) R gc Total 2-Point Correlation Function ( ± σ )Blue GC 2-Point Correlation Function ( ± σ )Red GC 2-Point Correlation Function ( ± σ ) 0.0 11.1 22.1 33.2 44.2 55.3 66.3 77.4 88.4 99.5 110.6 121.6 132.7 θ [kpc] ˆ w ( θ ) R gc < θ [arcmin] ˆ w ( θ ) R gc ≥ Figure 13.
Two-point angular correlation function analysis ofthe GC candidates. (
Top ): Two-point correlation functions for thetotal, blue, and red populations are indicated by the correspond-ing curves, with shaded ± σ bootstrapped errors. ( Middle ): Sameas above, but with clustering only considered for the GC sampleswithin R gc < (cid:48) of NGC 5128. Note that only clustering on scalesof < (cid:48) is considered. ( Bottom ): The analysis applied to the outer( R gc ≥ (cid:48) ) subsamples. In all panels, the lower abscissa indicates R gc in angular units, while the upper shows physical projecteddistance. The overall population shows a mildly bimodal struc-ture corresponding to the major axis along the Φ ≈ ◦ –Φ ≈ ◦ line, with indications for another peak near Φ ≈ ◦ .Considering the inner and outer samples separately showsthat the bimodal structure is more prevalent for the innerGCs, with only a mild over density seen near Φ ≈ ◦ forboth red and blue samples, and only the blue GCs showingany indication for an elevated surface number density nearΦ ≈ − ◦ . While the GC system seems to only showmild evidence for bi-modality along NGC 5128’s major axisout to R gc (cid:38) (cid:48) , it does appear consistent with the elliptic-ity already reported for its resolved red giant star populationat similar scales (Crnojevi´c et al. 2013; Rejkuba et al. 2014).One other feature of note is that seen at Φ ≈ ◦ , which isnotable in the inner sample for only the red subsample, andis more significant in the outer sample, where both blue andred candidates show signs of an over density. In general, de-spite the steeper radial surface number density profile shownby the red GCs for the inner sample, the azimuthal distri-butions show relatively similar behaviours at all radii, withonly mild indications for azimuthal sub-peaks that wouldindicate clustering on different length scales. The central clustering and potential arc(s) of GCs aroundNGC 5128 are the most obvious coherent GC structures inFig. 11, and here we attempt to quantify potential levels ofclustering. To this effect, the results of a two-point angular correlation function ( w ( θ ); Landy & Szalay 1993) analysis are shown in Fig. 13. Briefly, given a set of points on thesky, w ( θ ) describes the level and scale of clustering by con-sidering the probability of finding two points separated byan angle θ compared to what would be expected from a ran-dom distribution (i.e. w ( θ ) ≈ θ represents the angular distance from each pointunder consideration, and not from the centre of NGC 5128,i.e. R gc = 0 (cid:48) .The top panel of Fig. 13 shows w ( θ ) calculated for thetotal, blue, and red populations. For all samples, there ex-ists a particularly strong probability of clustering on scales (cid:46)
20 kpc, which is likely dominated by the concentration ofGCs directly around NGC 5128. Interestingly, all three sam-ples show very similar behaviour, with evidence for cluster-ing on all scales (cid:46)
40 kpc, above which there is a marginallikelihood of clustering. This result can be visually verifiedby Fig. 11, where there is concentrated clustering in the cen-tre of the field, with multiple smaller clumps throughout theregion, and a few larger structures that often correspond tosections of the “arc”.As implied by Fig. 12, if the “inner” and “outer” GCpopulations are of different natures, then they might showdifferent indications of clustering. To investigate this, themiddle and bottom panels of Fig. 13 show w ( θ ) correspond-ing to the “inner” and “outer” populations, respectively. Themiddle panel shows that for the R gc < (cid:48) GCs, the strongclustering of the core population dominates w ( θ ) such thatclustering is most significant on scales (cid:46)
20 kpc. Moreover,the inner, red population shows a higher probability of clus-tering on small scales, consistent with their steeper Σ N ( R gc )profile. Conversely, concurrent with the shallower blue GCslope, there is less evidence for clustering at the smallestscales, with w blue ( θ ) declining less sharply than the redsample out to R gc ≈
20 kpc, which represents the extentof NGC 5128’s inner halo. As a side-note, while outside of R gc ≈
20 kpc there is no evidence whatsoever for GC clus-tering, all samples show w ( θ ) (cid:38) ≈ (cid:48) . Thisfeature is a censoring artifact, and is to be expected sinceoutside of the largest scale of interest (i.e. R gc (cid:38) (cid:48) ; darkgrey shading), there exists a growing probability of findingpoints separated by (cid:46) (cid:48) compared to what is expected ofa uniform distribution within the larger region where thereare artificially no points.The bottom panel of Fig. 13 paints a somewhat differ-ent picture. Here the “outer” GCs show significant clusteringon scales (cid:46) (cid:48) . This result can be seen visually on Fig. 11,where the KDE shows higher density regions in the outskirtsof the field covering a wide range of spatial scales. Inter-estingly, in this case, it is the blue population that hintsat a slightly higher probability of clustering at the small-est scales together with a steeper decline. This feature isconsistent with the steeper outer Σ N ( R gc ) profile shown inFig. 12 (top panel). Altogether, the evidence of sub-10 kpcscale clustering of blue GCs supports the model in whichthey primarily come from dwarf galaxies and may hint at alarge reservoir of undiscovered or previously disrupted dwarfsatellites in the extreme halo of NGC 5128, similar to whatis predicted by the Λ Cold Dark Matter cosmological frame- Python package: astroML
MNRAS000
MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 ( u − z ) [mag] p ( x ) R gc < ξ b/r =1 . ( u − z ) [mag] p ( x ) R gc ≥ ξ b/r =1 . Φ [deg E of N] ξ b / r All GCs R gc < R gc ≥ Figure 14.
The inner and outer blue-to-red GC candidate frac-tions. (
Upper panels ): Two-component GMM fits (solid blackcurves) to the inner ( R gc < (cid:48) ; left panel) GC candidates, andthe outer ( R gc ≥ (cid:48) ; right), with grey shading representing theunderlying data. Blue/red shading represents the correspondingblue and red gaussian components. ( Lower panel ): ξ b / r is shownas functions of Φ in 36 ◦ -wide bins. The total population within120 (cid:48) of NGC 5128 is shown by the black relation, while the greenand brown relations show results for the inner and outer samples,respectively. work (Klypin et al. 1999; Moore et al. 1999) and hinted at inrecent surveys of nearby galaxy clusters (Mu˜noz et al. 2015;S´anchez-Janssen et al. 2016). Figs. 14 and 15 further probe the spatial distributions ofthe “inner” and “outer” samples of blue and red GCs. Theupper panels of Fig. 14 show two-component GMM fits tothe “inner” GCs (left), and the “outer” candidates (right).Both samples show larger than unity ξ b / r , marginally forthe outer candidates, and more dramatically for the inner.Meanwhile, the lower panel of Fig. 14 shows the Φ depen-dence of ξ b / r . The total (black relation), inner (green), andouter (brown) candidate samples are binned by ∆Φ = 36 ◦ ,and ξ b / r , shown along the ordinate, is calculated in each bin.The outer sample shows on average lower ξ b / r , especially inthe range 0 ◦ (cid:46) Φ (cid:46) ◦ , and particularly at Φ ≈ ◦ − ◦ , ∼ ◦ , and ∼ ◦ .Complementary to Fig. 14, Fig. 15 shows the ratio ofexponential-KDEs applied as in Fig. 11, but to each of theblue and red subsamples. The colour shading represents ar-eas where the GCs of each colour dominate the surface num-ber density distribution. Of particular note are the high ξ b / r features mentioned above. While it can be seen that ξ b / r is indeed generally higher inwards of R gc ≈ (cid:48) , the blue GCs tend toward the outskirts of the region, with a partic-ularly high concentration in the 130 ◦ (cid:46) Φ (cid:46) ◦ wedge S ofNGC 5128, but overall the central regions show a strongerrepresentation by the red GC sample.As seen previously, the region within R gc ≈ (cid:48) is rep-resented well by each colour, and is surrounded by clumpyblue over densities corresponding to the ξ b / r spikes seen inFigs. 12 and 14. At larger R gc , both populations cluster atthe scales indicated by Fig. 13, with half of the dwarf galax-ies showing some indications of projected associations withregions of blue over-densities. In the following, we explore the implications that the respec-tive “inner” and “outer” GC candidate populations have forthe past and present dwarf galaxy populations of NGC 5128.We note that, while the apparent break in the power-law fitsshown in Fig. 12 might imply a transition to the Centaurus Aintra-group medium, this may not reflect the true natureof these objects. For example, observational and theoreticalevidence suggests that at least some giant galaxies residewithin dual-halos, including a diffuse metal-poor stellar com-ponent extending dozens of kpc from their hosts that arisefrom either two-component star formation/chemical enrich-ment mechanisms (e.g. Rejkuba et al. 2005; Harris et al.2007; Rejkuba et al. 2011; Lee & Jang 2016), and/or fromthe accretion of multiple satellites throughout a galaxy’s for-mation history (e.g. Bullock & Johnston 2005; Abadi et al.2006; Johnston et al. 2008; Cooper et al. 2010; Deason etal. 2013; Park & Lee 2013; Ibata et al. 2014). With this inmind we move forward to consider separately the inner, in-trinsic, GC population of NGC 5128, followed by the outergroup that likely reflects GCs accreted onto the host’s ex-treme outer halo and/or a transition to the Centaurus Aintra-group GC population.
If the inner GCs at R gc < (cid:48) , corresponding to ∼
55 kpcor ∼
20 percent of NGC 5128’s virial radius, are intrinsicto NGC 5128, they have strong implications for the assem-bly of the host. The median ( g (cid:48) − z (cid:48) ) = 1 . ± .
01 magcolour of the inner red GCs is 0 .
03 mag bluer than the1 . ± .
02 mag median ( g (cid:48) − z (cid:48) ) colour expected of a gi-ant galaxy sharing NGC 5128’s M B = − . M B in the range − .
25 to − .
75 mag and( g (cid:48) − z (cid:48) ) = (1 . − . ± .
02 mag (Peng et al. 2006).A merger of such galaxies would produce a combined lu-minosity of M B ≈ − . − . R gc ≈ (cid:48) requires an explanation even ifthey were brought in during the merger event.The new and H12 candidates within 50 (cid:48) together with MNRAS , 1–26 (2016) M. A. Taylor et al. α [J2000] δ [ J ] ABC =
33 kpc NE -2 -1 P blue /P red Figure 15.
The projected spatial probability density of blue GCs relative to red. Relative probability densities are highlighted by thedarker red and blue regions, with white areas indicating equal contributions of blue and red GCs. Small points show locations of the GCcandidates, while confirmed and candidate dwarf galaxies are indicated by small X’s and circles, respectively. Dashed ellipses are drawnat R gc = 50 (cid:48) and 120 (cid:48) , while lightly shaded ellipses highlight regions in NGC 5128’s outer halo with unexpectedly high densities of redGC candidates.. the confirmed GCs implies an intrinsic total population ofat least 1 066. We can then calculate the specific frequencyas defined by (Harris & van den Bergh 1981), S N = N GC · . M V +15) = (cid:0) . × (cid:1) N GC L V /L V, (cid:12) (6)Adopting a de-reddened V -band luminosity of M V = − . S N ≈ .
9, a value not unreasonable for agiant E/S0 galaxy (Harris & van den Bergh 1981; Harriset al. 2013). We note that the distribution of the red GCcomponent clusters more closely to NGC 5128 than the blue GCs, which extend well out into the R gc < (cid:48) halo witha shallower Σ N ( R gc ) profile. This is consistent with otherhigh S N galaxies, in that they tend to have higher ξ b / r , withmetal-poor (blue) GCs preferentially following more radiallyextended, shallow distributions and extending further outthan the host’s underlying starlight (Forbes et al. 1997b). While the consistency with previous works is encouraging,the relatively high ξ b / r calls for an explanation of its origins. MNRAS000
9, a value not unreasonable for agiant E/S0 galaxy (Harris & van den Bergh 1981; Harriset al. 2013). We note that the distribution of the red GCcomponent clusters more closely to NGC 5128 than the blue GCs, which extend well out into the R gc < (cid:48) halo witha shallower Σ N ( R gc ) profile. This is consistent with otherhigh S N galaxies, in that they tend to have higher ξ b / r , withmetal-poor (blue) GCs preferentially following more radiallyextended, shallow distributions and extending further outthan the host’s underlying starlight (Forbes et al. 1997b). While the consistency with previous works is encouraging,the relatively high ξ b / r calls for an explanation of its origins. MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 The number of blue GCs that need to be accounted for issimply, N blue = ξ b / r N GC,t ξ b / r (7)If minor mergers with NGC 5128’s giant progenitors repre-sent the sole sources of the blue GCs, then the number offormer dwarfs, N dw , of a luminosity, L V, dw , needed to buildup such population is approximated by the number of GCsthat each galaxy can be expected to contribute, N GC , dw .From the definition of S N , then trivially, N dw = N blue N GC , dw = ξ b / r N GC , t ξ b / r · . × S N, dw L V, dw (8)which can be simplified given ξ b / r = 1 .
33 (c.f. Table 6) forthe 1 066 inner NGC 5128 GCs to, N dw = 5 . × S N, dw L V, dw (9)Taking approximate S N, dw for a dwarf at a given L V, dw from the rich compilation of S N derived for dwarf galaxies inthe local universe by Georgiev et al. (2010), Table 8 showsthe approximate numbers of dwarfs of a given L V, dw thatwould be required to fully account for the inner blue GCsdetected. Also listed are the magnitudes of the combineddwarf progenitors, and the fraction of NGC 5128’s currentluminosity that such a reservoir of dwarfs would contribute.While the numbers are rough approximations, it is illustra-tive that a reservoir of >
250 10 − L V, (cid:12) -class dwarfs is suf-ficient to provide the blue GCs, but insufficient to build up asignificant fraction of NGC 5128’s current luminosity. Mean-while, ∼
100 10 L V, (cid:12) -class dwarfs are required to providesuch blue GC population, and could have contributed up to ∼
33 percent of NGC 5128’s light. We note that if a smallernumber of 10 L V, (cid:12) -class dwarfs is responsible, this wouldbe sufficient to build up NGC 5128’s total luminosity, butwould not provide a means of explaining the origins of thered GC population. This last fact suggests then, that at leastdozens of minor mergers of 10 − L V, (cid:12) dwarf galaxies withgiant galaxy progenitors are required to explain NGC 5128’scurrent state.We further explore the purported dwarf galaxy pop-ulation needed to give rise to the blue GC candidates bymodelling the galaxy luminosity function (LF) assuming aSchechter function of the form (Schechter 1976),Φ( L, α ) dL = φ ∗ (cid:18) LL ∗ (cid:19) α · e − ( L/L ∗ ) dL (10)or in terms of magnitudes,Φ( M, α ) dM = 0 . ×× φ ∗ (cid:104) . ( M ∗ − M ) (cid:105) ( α +1) · e − . ( M ∗− M ) dM (11)where M ∗ is the characteristic galaxy magnitude, whichfollowing Smith et al. (2009) and Ferrarese et al. (2016)we adopt as M ∗ V = − .
84 mag. While the φ ∗ normaliza-tion is somewhat dependent on the filter and magnituderange of a given dataset, we assume a typical value of φ ∗ =1 . × − h Mpc (Schechter 1976; Smith et al. 2009) with h = 68 (km / s) Mpc − (Planck Collaboration et al. 2014). We Table 8.
Estimates of the dwarf galaxy reservoirs required toprovide the blue GC population of NGC 5128. Col. (1) lists the ex-ample luminosity classes of the purported dwarfs, followed by theapproximate S N in col. (2). Col. (3) shows the number of dwarfsrequired to provide the observed blue GCs, while the last twocolumns list approximations of the combined luminosities of eachdwarf reservoir and the fractions of NGC 5128’s current light thatthey would represent. L V, dw S N, dw N dw M V, dw f L V ( L (cid:12) ) (mag) R gc < (cid:48) ( inner GC system )5 ×
39 266 − . ×
21 247 − . × − . × − . R gc ≥ (cid:48) ( outer GC system )5 ×
39 468 − . ×
21 435 − . × − . × − . employ this distribution for − . ≤ M V ≤ − .
17 mag(corresponding to 2 × ≥ L V, (cid:12) ≥ ). Testing α in therange ( − . , . (cid:48) of NGC 5128. We repeat this process until N GC , blue are contributed, and record the full dwarf sample. We iter-ate this procedure 1 000 times for a given α and list themean numbers of 10 − , 10 − , 10 − , and 10 − . L V, (cid:12) -class dwarfs required to build the blue GC population inTable 9 at varying α . We also show the total numbers ofdwarfs predicted, alongside their combined magnitudes andfractions of NGC 5128’s current day luminosity.Based on the results for the inner GC candidates listedin Table 9, we find that a top-heavy LF ( α (cid:39) .
0) is dis-favoured. Such a form predicts that as many as 11 dwarfs of (cid:38) L V, (cid:12) may have contributed to the observed blue GCcandidates, but with little to no contribution by lower massdwarfs. More importantly, the combined luminosity of sucha dwarf population represents (cid:38)
73 percent of NGC 5128’sspheroid light, but fails to provide an avenue for the cre-ation of the red GC population, assuming them to have beenformed during the monolithic collapse of NGC 5128’s giantprogenitors. Distributions more top-heavy than α > − . f L V (cid:38) .
65 in all cases. For − . (cid:46) α (cid:46) − .
0, we find that the increased fractions of blue GCsprovided by lower-mass dwarfs sufficiently explain the GCpopulation, while providing a fraction of NGC 5128’s currentday light on the order of 25–60 percent. At more bottom-heavy dwarf LFs ( α (cid:46) − . (cid:38)
200 10 − L V, (cid:12) -classdwarfs being required to provide the blue GCs, while pro-viding only a small fraction of NGC 5128’s current spheroidlight. We choose 10 . M (cid:12) as the upper sampling limit as it marksthe highest luminosity of known dwarfs around NGC 5128.MNRAS , 1–26 (2016) M. A. Taylor et al.
Table 9.
Results of dwarf galaxy population modelling for the inner ( R gc < (cid:48) ) and outer ( R gc ≥ (cid:48) ) regions around NGC 5128.Col. (1) lists the assumed Schechter function slopes, followed by the required dwarf galaxy populations in luminosity bins increasingfrom cols. (2)–(5). The total numbers of dwarf galaxies are listed in col. (6), followed by their combined luminosities and fractions ofNGC 5128’s current day light in cols. (7)–(8). α N − L V, (cid:12) N − L V, (cid:12) N − L V, (cid:12) N − . L V, (cid:12) N dw M V, dw f L V (mag) R gc < (cid:48) ( inner GC system )0 .
00 0 0 9 11 21 − .
06 0.73 − .
25 0 2 12 10 25 − .
04 0.72 − .
50 1 5 15 9 31 − .
00 0.69 − .
75 6 10 18 8 42 − .
93 0.65 − .
00 21 20 20 6 68 − .
79 0.57 − .
25 63 33 18 3 119 − .
49 0.44 − .
50 136 40 12 1 191 − .
91 0.26 − .
75 209 34 6 0 249 − .
12 0.13 − .
00 255 23 2 0 280 − .
35 0.06 R gc ≥ (cid:48) ( outer GC system )0 .
00 0 1 17 19 38 − .
66 1.27 − .
25 0 3 21 18 43 − .
64 1.25 − .
50 2 8 26 16 54 − .
61 1.21 − .
75 10 18 32 13 75 − .
54 1.14 − .
00 37 36 35 10 120 − .
40 1.00 − .
25 111 59 32 6 210 − .
10 0.76 − .
50 240 70 22 2 336 − .
54 0.46 − .
75 368 60 10 0 440 − .
74 0.22 − .
00 450 40 3 0 494 − .
97 0.11
Altogether, these results suggest that a faint-end LFslope of − . (cid:46) α (cid:46) − .
50 is consistent with the build-up of NGC 5128’s intrinsic population of blue GCs. Thisoutcome is in good agreement with the recent work on theinner ∼
300 kpc of the Virgo galaxy cluster by Ferrarese etal. (2016), who found α = − . ± .
02 to best representthe LF in the dwarf regime, and is only marginally higherthan their α = − . ± .
05 found for the Local Groupdwarf galaxy population. Interestingly, Ferrarese et al. findthat if the Virgo UCDs are assumed to be the remnantsof nucleated dwarf galaxies stripped of their outer stellarhalos (and thus GCs), they find a much steeper slope of − . ± .
06, which appears inconsistent with our results.With that said, our GC selection procedure intentionallyfiltered out objects encroaching upon UCD luminosities, andso a future dedicated search for UCDs in the outer halo ofNGC 5128 may result in a promising reservoir of additionalblue GC progenitors.
Most of the dwarfs contributing to the inner population ofGCs are likely to have merged with NGC 5128 and/or itsgiant progenitors in the past, but the same cannot neces-sarily be assumed for the extreme outer halo of NGC 5128.In this way, carrying out the exercise above with the outerpopulation serves as a rough prediction on the populationof dwarfs that have either already been disrupted, or maystill be present in the extreme halo. The bottom halves ofTables 8 and 9 list these results, which predict an even richerdwarf population than the inner halo. If the LF of dwarfs inthe extreme halo of NGC 5128 follows that within R gc < (cid:48) , this result requires that >
100 10 − L V, (cid:12) and dozens of10 − L V, (cid:12) -class dwarfs with projected R gc (cid:38) (cid:48) , manyof which might have already been disrupted in NGC 5128’stidal field, while their GC populations survived. The mostmassive of these dwarfs must still be or have until the recentpast been present.The Centaurus A/M83 galaxy complex has been shownrecently to be potentially rich in low-mass dwarf galaxies,with 57 promising candidates reported based on wide fieldDECam imaging (M¨uller et al. 2015, 2017) and 15 confirmeddwarfs within the region studied in this work (van den Bergh2000; Karachentsev et al. 2007; Crnojevi´c et al. 2014, 2016);however, very little wide-field imaging has been done at suffi-cient depths to robustly detect dwarfs of L V (cid:46) L V, (cid:12) . Ex-cellent work was recently done to detect several new dwarfs(Crnojevi´c et al. 2016) via resolved stellar over densities, butthe faintest of this sample barely reaches L V ≈ L V, (cid:12) .Any dwarfs of similar luminosities, but more diffuse mor-phologies and/or projected along less-fortunate axes (as-suming triaxial structures), would therefore still remain un-detected, as would more massive satellites with yet morediffuse stellar distributions. With that said, Crnojevi´c et al.detected clear signals of loops and streams that indicate thatdwarfs are still being actively disrupted to the present dayout to at least R gc ≈ (cid:48) ( ∼
132 kpc), thus it is likelythat past dwarf galaxy interactions may be behind the pop-ulation of blue GCs at such distance from NGC 5128. Suchinteractions have recently been shown to produce preferen-tial stripping of GC systems (Smith et al 2013; Smith et al.2015).Even so, given the emerging evidence for at leastone, and possibly two, planes of dwarf satellites around
MNRAS000
MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 NGC 5128 (Tully et al. 2015; M¨uller et al. 2016), one mightexpect that if the outer GCs arise from accreted dwarf GCsystems, then they should reflect this origin by assuming acoherent structure aligned with the plane associated withNGC 5128 itself (see M¨uller et al., their Fig. 1). A planaralignment of outer GCs is clearly not seen in Fig. 11 nor, inparticular, the lower panel of Fig. 12, where the azimuthaldistribution indicates large-scale near homogeneity. Whileit is tempting to attribute this as evidence against dwarfgalaxy origins, the NGC 5128-centric spatial sampling ofonly ∼
120 kpc precludes this interpretation, considering theMpc-scale distribution of the satellite planes.An interesting feature of Fig. 15 is the existence of sev-eral (cid:38) (cid:48) -scale structures dominated by red GCs, high-lighted by shaded ellipses and labelled ‘A’, ‘B’, and ‘C’. Themost significant (‘A’) lies > (cid:48) to the NW of NGC 5128and the existence of these GCs, along with those thatmake up the two red over-densities at (90 kpc , ◦ ) and( R gc , Φ) ≈ (90 kpc , ◦ ) (labelled ‘B’, and ‘C’, respectively)is puzzling and merits follow-up studies. These GC over-densities may have origins in giant background galaxies ordistant galaxy clusters, but visual inspection of the imagesshows no obvious evidence for possible background hosts.Evidence that the kinematics of the inner halo GCsshares similarities with the overall kinematics of the Cen-taurus A group suggests that the outer halo of the groupis dynamically connected to the rest of the group (Wood-ley 2006). In this scenario, the notion that NGC 5128 hasprimarily been built up by minor mergers, with only a fewmajor mergers contributing seems to be consistent with thepresent findings. While the results of the previous sectionsupport a recent merger of two equal-mass giants, it does notrule out a more ancient merger whose evidence is no longerdetectable. In this case, these clumps could represent thelast coherent structures resulting from these violent events.On the other hand, the recent discovery of “ultra-diffuse”galaxies (UDGs) in large galaxy clusters (van Dokkum et al.2015; Koda et al. 2015; Mihos et al. 2015; Mu˜noz et al. 2015;Mart´ınez-Delgado et al. 2016) have been proposed to repre-sent failed L ∗ galaxies with deep potentials relative to theirstellar masses. Little is known about their respective GCsystems, but recent work has indicated that they may havehigh S N (cid:38)
30 with as many as dozens of GCs (Beasley &Trujillo 2016; Peng & Lim 2016; van Dokkum et al. 2016).These works find that UDGs are likely to host primarilyblue GC populations more similar to dwarfs than giants, al-though with only one UDG GC system studied so far, fewgeneral conclusions can be drawn. In any case, the large GCsubstructures seen in Figs. 11 and 15 could represent very in-teresting targets for follow-up deep imaging campaigns thatwould further build upon this work.
In this work, new wide-field CTIO/DECam imaging in theoptical u (cid:48) g (cid:48) r (cid:48) i (cid:48) z (cid:48) filters of the central ∼
21 deg of the Cen-taurus A galaxy group is analyzed. Two-colour diagnostic di-agrams are combined with source morphologies to constructa near-complete catalogue of GC candidates as far out as ∼
140 kpc from the centrally dominant galaxy NGC 5128.We find a total of 2 676 GC candidates, of which 2 404 are newly identified, and provide new measurements of manypreviously radial-velocity confirmed GCs.We use Gaussian mixture models to classify the GC can-didates as either blue or red, and find the well establishedbimodal distribution in GC colours for giant galaxies is welldefined for NGC 5128’s GC system. The GC system as awhole shows a larger number ratio of blue GCs with respectto red, i.e. ξ b / r = 1 .
16. Evidence is presented for distinctpopulations inside and outside of R gc ≈ (cid:48) and we suggestthat the inner population is likely intrinsic to NGC 5128 it-self, while the outer population may begin to sample thoseGCs associated with the intra-group environment, or at leasthave been deposited there by accreted satellites. The innersample shows a slightly higher ξ b / r = 1 .
33, compared to 1 . N ( R gc ) ∝ R Γgc for NGC 5128’s GC sys-tem and find that the inner population is consistent withother gE galaxies both in colour-dependent slopes of − . (cid:46) Γ (cid:46) − .
40 (e.g. Puzia et al. 2004), and that the red GCsare more centrally concentrated than the blue, which ex-tend into NGC 5128’s outer halo. This trend reverses outsideof ∼ (cid:48) , as the blue GCs show a slightly steeper power-law relation. Overall, the spatial distribution of NGC 5128’sGC system is not uniform. A two-point angular correla-tion function analysis provides evidence for clustering onall scales below ∼
20 kpc for the inner population, with redGCs showing stronger clustering toward smaller scales. Mildevidence for a reversal of this trend is seen for the outer pop-ulation, as the blue population shows slightly stronger evi-dence for small-scale clustering, consistent with the notionthat they may be hinting at an as-yet unknown populationof low surface-brightness dwarf galaxy hosts. Alternatively,they might be the last remaining coherent structures frompreviously disrupted dwarf satellites, as observed in the Lo-cal Group (Mackey et al. 2014). Both red and blue outerGC samples show a much more shallow decline of w ( θ ), sug-gestive of clustering at larger scales than found in the innerpopulation. Finally, we find mild evidence for a coherent overdensity, or stream, of GCs outside of R gc ≈ (cid:48) , which willbe statistically quantified in a future work.The median ( g (cid:48) − z (cid:48) ) ≈ .
90 mag colour of the bluecomponent is consistent with what is expected for a gEgalaxy of similar luminosity in the Virgo cluster (Peng etal. 2006), while the ∼ .
27 mag colour of the red componentis consistent with a build up from two giant galaxies eachon the order of M V ≈ − . − . ∼ (cid:48) of NGC 5128, ifthe assumption is made that they all have origins in dwarfhaloes, this would require dozens of minor-mergers with10 − L V, (cid:12) -class dwarfs during the assembly of NGC 5128and its giant progenitors. Likewise, if the outer populationof blue GCs is to be explained by dwarf halo hosts, thenyet more low-mass dwarfs are either possibly lying unde-tected, or have already been disrupted within ∼
140 kpc ofNGC 5128.The unexpected presence of large numbers of both redand blue GCs in the extreme halo of NGC 5128 provides rich
MNRAS , 1–26 (2016) M. A. Taylor et al. opportunities for follow up studies, which will be conductedin future contributions. Our SCABS imaging is undergo-ing a careful background subtraction and future work willcharacterize the as-yet relatively unknown dwarf populationin the Centaurus A galaxy group. Additionally, the inclu-sion of near-infrared imaging will refine our new GC candi-date catalogue further, with very low contamination by fore-ground stars and background galaxies (Mu˜noz et al. 2014).Nonetheless, with the complete census of true GCs soon tobe in-hand, secure spectroscopic follow-up targets will beparamount to unveiling the global velocity map of CentaurusA and its environment, and place unprecedented constraintson the mass assembly history of this iconic galaxy.
ACKNOWLEDGEMENTS
We extend our gratitude to Marina Rejkuba for providingextensive and constructive comments that significantly im-proved the manuscript. We also wish to warmly thank Si-mon ´Angel, and Yasna Ordenes-Brice˜no for fruitful discus-sions, and especially Eric Peng for additionally providingus with catalogues of new confirmed foreground stars andGCs prior to publication. M.A.T. acknowledges the finan-cial support through an excellence grant from the “Vicer-rector´ıa de Investigaci´on” and the Institute of AstrophysicsGraduate School Fund at Pontificia Universidad Cat´olicade Chile and the European Southern Observatory GraduateStudent Fellowship program. T.H.P. acknowledges supportby a FONDECYT Regular Project Grant (No. 1161817) andthe BASAL Center for Astrophysics and Associated Tech-nologies (PFB-06). H.Z. was supported in part by FONDE-CYT Postdoctoral Fellowship Grant (No. 3160538). P.E. ac-knowledges support from of a FONDECYT PostdoctoralFellowship Grant (No. 3130485) and the CHINA-CONICYTFellowship Project (No. CAS150023). M.S.B. was supportedin part by a FONDECYT Postdoctoral Fellowship Grant(No. 3130549).This research has made use of the NASA AstrophysicsData System Bibliographic Services, the NASA Extragalac-tic Database, and the SIMBAD database, operated at CDS,Strasbourg, France (Wenger et al. 2000).. Software usedin the analysis includes the
Python/NumPy v.1.11.0 and
Python/Scipy v0.17.0 (Jones et al. 2001; Van Der Waltet al. 2011, ), Python/astropy (v1.1.1; Astropy Collaboration et al. 2013, ), Python/matplotlib (v1.5.1; Hunter2007, http://matplotlib.org/ ), Python/scikit-learn (v0.16.1; Pedregosa et al. 2012, http://scikit-learn.org/stable/ ), and
Python/astroML (v0.3; VanderPlas et al.2012, ) packages.This work is based on observations at Cerro Tololo Inter-American Observatory, National Optical Astronomy Obser-vatory (CNTAC Prop. ID: 2014A-0610; PI: Matthew Tay-lor), which is operated by the Association of Universities forResearch in Astronomy (AURA) under a cooperative agree-ment with the National Science Foundation. This projectused data obtained with the Dark Energy Camera (DE-Cam), which was constructed by the Dark Energy Survey(DES) collaboration.
REFERENCES
Abadi, M. G., Navarro, J. F., & Steinmetz, M. 2006, MNRAS,365, 747Annunziatella, M., Mercurio, A., Brescia, M., Cavuoti, S., &Longo, G. 2013, PASP, 125, 68Arnold, J. A., Romanowsky, A. J., Brodie, J. P., et al. 2011, ApJ,736, L26Ashman, K. M., & Zepf, S. E. 1992, ApJ, 384, 50Ashman, K. M., & Zepf, S. E. 1998, Globular Cluster Systems(Cambridge, UK; New York : Cambridge University Press)Ashman, K. M., & Zepf, S. E. 2008, Globular Cluster Systems,(Cambridge, UK: Cambridge University Press)Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al.2013, A&A, 558, A33Baade, W., & Minkowski, R. 1954, ApJ, 119, 215Bassino, L. P., Richtler, T., & Dirsch, B. 2006, MNRAS, 367, 156Beasley, M. A., Baugh, C. M., Forbes, D. A., Sharples, R. M., &Frenk, C. S. 2002, MNRAS, 333, 383Beasley, M. A., Bridges, T., Peng, E., et al. 2008, MNRAS, 386,1443Beasley, M. A., & Trujillo, I. 2016, arXiv:1604.08024Bekki, K. 2010, MNRAS, 401, 2753Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393Bertin, E. 2011, Astronomical Data Analysis Software and Sys-tems XX, 442, 435Brodie, J. P., & Huchra, J. P. 1991, ApJ, 379, 157Brodie, J.P., & Strader J. 2006, ARA&A, 44, 193Brodie, J. P., Usher, C., Conroy, C., et al. 2012, ApJ, 759, L33Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931Burgarella, D., Kissler-Patig, M., & Buat, V. 2001, AJ, 121, 2647Chabrier, G. 2003, PASP, 115, 763Cooper, A. P., Cole, S., Frenk, C. S., et al. 2010, MNRAS, 406,744Cˆot´e, S., Freeman, K. C., Carignan, C., & Quinn, P. J. 1997, AJ,114, 1313Cˆot´e, P., Marzke, R. O., & West, M. J. 1998, ApJ, 501, 554Cˆot´e, P., Marzke, R. O., West, M. J., & Minniti, D. 2000, ApJ,533, 869Crnojevi´c, D., Ferguson, A. M. N., Irwin, M. J., et al. 2013, MN-RAS, 432, 832Crnojevi´c, D., Sand, D. J., Caldwell, N., et al. 2014, ApJ, 795, 35Crnojevi´c, D., Sand, D. J., Spekkens, K., et al. 2016, ApJ, 823,19D’Abrusco, R., Fabbiano, G., Mineo, S., et al. 2014, ApJ, 783, 18D’Abrusco, R., Fabbiano, G., & Brassington, N. J. 2014, ApJ,783, 19D’Abrusco, R., Fabbiano, G., & Zezas, A. 2015, ApJ, 805, 26Deason, A. J., Belokurov, V., Evans, N. W., & Johnston, K. V.2013, ApJ, 763, 113Desai, S., Armstrong, R., Mohr, J. J., et al. 2012, ApJ, 757, 83Durrell, P. R., Harris, W. E., Geisler, D., & Pudritz, R. E. 1996,AJ, 112, 972Elson, R. A. W., & Santiago, B. X. 1996, MNRAS, 278, 617Faifer, F. R., Forte, J. C., Norris, M. A., et al. 2011, MNRAS,416, 155Ferrarese, L., Cˆot´e, P., S´anchez-Janssen, R., et al. 2016, ApJ, 824,10Fioc, M., & Rocca-Volmerange, B. 1997, A&A, 326, 950Forbes, D. A., Grillmair, C. J., & Smith, R. C. 1997, AJ, 113,1648Forbes, D. A., Brodie, J. P., & Grillmair, C. J. 1997, AJ, 113,1652Forbes, D. A., Georgakakis, A. E., & Brodie, J. P. 2001, MNRAS,325, 1431Forbes, D. A., Ponman, T., & O’Sullivan, E. 2012, MNRAS, 425,66 MNRAS000
Abadi, M. G., Navarro, J. F., & Steinmetz, M. 2006, MNRAS,365, 747Annunziatella, M., Mercurio, A., Brescia, M., Cavuoti, S., &Longo, G. 2013, PASP, 125, 68Arnold, J. A., Romanowsky, A. J., Brodie, J. P., et al. 2011, ApJ,736, L26Ashman, K. M., & Zepf, S. E. 1992, ApJ, 384, 50Ashman, K. M., & Zepf, S. E. 1998, Globular Cluster Systems(Cambridge, UK; New York : Cambridge University Press)Ashman, K. M., & Zepf, S. E. 2008, Globular Cluster Systems,(Cambridge, UK: Cambridge University Press)Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al.2013, A&A, 558, A33Baade, W., & Minkowski, R. 1954, ApJ, 119, 215Bassino, L. P., Richtler, T., & Dirsch, B. 2006, MNRAS, 367, 156Beasley, M. A., Baugh, C. M., Forbes, D. A., Sharples, R. M., &Frenk, C. S. 2002, MNRAS, 333, 383Beasley, M. A., Bridges, T., Peng, E., et al. 2008, MNRAS, 386,1443Beasley, M. A., & Trujillo, I. 2016, arXiv:1604.08024Bekki, K. 2010, MNRAS, 401, 2753Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393Bertin, E. 2011, Astronomical Data Analysis Software and Sys-tems XX, 442, 435Brodie, J. P., & Huchra, J. P. 1991, ApJ, 379, 157Brodie, J.P., & Strader J. 2006, ARA&A, 44, 193Brodie, J. P., Usher, C., Conroy, C., et al. 2012, ApJ, 759, L33Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931Burgarella, D., Kissler-Patig, M., & Buat, V. 2001, AJ, 121, 2647Chabrier, G. 2003, PASP, 115, 763Cooper, A. P., Cole, S., Frenk, C. S., et al. 2010, MNRAS, 406,744Cˆot´e, S., Freeman, K. C., Carignan, C., & Quinn, P. J. 1997, AJ,114, 1313Cˆot´e, P., Marzke, R. O., & West, M. J. 1998, ApJ, 501, 554Cˆot´e, P., Marzke, R. O., West, M. J., & Minniti, D. 2000, ApJ,533, 869Crnojevi´c, D., Ferguson, A. M. N., Irwin, M. J., et al. 2013, MN-RAS, 432, 832Crnojevi´c, D., Sand, D. J., Caldwell, N., et al. 2014, ApJ, 795, 35Crnojevi´c, D., Sand, D. J., Spekkens, K., et al. 2016, ApJ, 823,19D’Abrusco, R., Fabbiano, G., Mineo, S., et al. 2014, ApJ, 783, 18D’Abrusco, R., Fabbiano, G., & Brassington, N. J. 2014, ApJ,783, 19D’Abrusco, R., Fabbiano, G., & Zezas, A. 2015, ApJ, 805, 26Deason, A. J., Belokurov, V., Evans, N. W., & Johnston, K. V.2013, ApJ, 763, 113Desai, S., Armstrong, R., Mohr, J. J., et al. 2012, ApJ, 757, 83Durrell, P. R., Harris, W. E., Geisler, D., & Pudritz, R. E. 1996,AJ, 112, 972Elson, R. A. W., & Santiago, B. X. 1996, MNRAS, 278, 617Faifer, F. R., Forte, J. C., Norris, M. A., et al. 2011, MNRAS,416, 155Ferrarese, L., Cˆot´e, P., S´anchez-Janssen, R., et al. 2016, ApJ, 824,10Fioc, M., & Rocca-Volmerange, B. 1997, A&A, 326, 950Forbes, D. A., Grillmair, C. J., & Smith, R. C. 1997, AJ, 113,1648Forbes, D. A., Brodie, J. P., & Grillmair, C. J. 1997, AJ, 113,1652Forbes, D. A., Georgakakis, A. E., & Brodie, J. P. 2001, MNRAS,325, 1431Forbes, D. A., Ponman, T., & O’Sullivan, E. 2012, MNRAS, 425,66 MNRAS000 , 1–26 (2016) he Extended Globular Cluster System of NGC 5128 Gebhardt, K., & Kissler-Patig, M. 1999, AJ, 118, 1526Geisler, D., Lee, M. G., & Kim, E. 1996, AJ, 111, 1529Georgiev, I. Y., Puzia, T. H., Hilker, M., & Goudfrooij, P. 2009,MNRAS, 392, 879Georgiev, I. Y., Puzia, T. H., Goudfrooij, P., & Hilker, M. 2010,MNRAS, 406, 1967G´omez, M., Geisler, D., Harris, W. E., et al. 2006, A&A, 447, 877Goudfrooij, P., Schweizer, F., Gilmore, D., & Whitmore, B. C.2007, AJ, 133, 2737Graham, J.A. 1979, ApJ, 232, 60Graham, J.A., & Phillips, M.M. 1980, ApJ, 239, L97Harris, W. E., & van den Bergh, S. 1981, AJ, 86, 1627Harris, G. L. H., Hesser, J. E., Harris, H. C. & Curry, P. J. 1984,ApJ, 287, 175Harris, H. C., Harris, G. L. H., Hesser, J. E., & MacGillivray,H. T. 1984, ApJ, 287, 185Harris, H. C., Harris, G. L. H., & Hesser, J. E. 1988, in IAU Symp.126, Globular Cluster Systems in Galaxies, ed. J. E. Grindlay& A. G. D. Philip (Dordrecht: Kluwer), 205Harris, W. E. 1991, ARA&A, 29, 543Harris, G. L. H., Geisler, D., Harris, H. C., & Hesser, J. E. 1992,AJ, 104, 613Harris, W. E. 2001, Saas-Fee Advanced Course 28: Star Clusters,223Harris, G. L. H., Geisler, D., Harris, W. E., & Hesser, J. E. 2002, inIAU Symp. 207, Extragalactic Star Clusters, ed. D. P. Geisler,E. K. Grebel, & D. Minniti (San Francisco, CA: ASP), 309Harris, G. L. H., Geisler, D., Harris, W. E., et al. 2004, AJ, 128,712Harris, G. L. H., Harris, W. E., & Geisler, D. 2004, AJ, 128, 723Harris, W. E., Harris, G. L. H., Barmby, P., McLaughlin, D. E.,& Forbes, D. A. 2006, AJ, 132, 2187Harris, W. E., Harris, G. L. H., Layden, A. C., & Wehner, E. M. H.2007, ApJ, 666, 903Harris, G. L. H. 2010, PASA, 27, 457Harris, G. L. H., Rejkuba, M., & Harris, W. E. 2010, PASP, 27,475Harris, G. L. H., G´omez, M., Harris, W. E., et al. 2012, AJ, 143,84Harris, W. E., Harris, G. L. H., & Alessi, M. 2013, ApJ, 772, 82Ha¸segan, M., Jord´an, A., Cˆot´e, P., et al. 2005, ApJ, 627, 203Held, E. V., Federici, L., Testa, V., & Cacciari, C. 1997, TheNature of Elliptical Galaxies; 2nd Stromlo Symposium, 116,500Hesser, J. E., Harris, H. C., van den Bergh, S., & Harris, G. L. H.1984, AJ, 276, 491Hesser, J. E., Harris, H. C., & Harris, G. L. H. 1986, ApJ, 303,L51Holland, S., Cˆot´e, & Hesser, J. E. 1999, A&A, 348, 418Hunter, J. D. 2007, CISE, 9, 90Ibata, R. A., Lewis, G. F., McConnachie, A. W., et al. 2014, ApJ,780, 128Innanen, K. A. 1979, AJ, 84, 7Jablonka, P., Bica, E., Pelat, D., & Alloin, D. 1996, A&A, 307,385Jacoby, G. H., Branch, D., Ciardullo, R., et al. 1992, PASP, 104,599Jester, S., Schneider, D. P., Richards, G. T., et al. 2005, AJ, 130,873Johnston, K. V., Bullock, J. S., Sharma, S., et al. 2008, ApJ, 689,936-957Jones E., Oliphant E., Peterson P., et al. 2001-, SciPy: OpenSource Scientific Tools for Python, [Online; accessed 2016-05-10]Karachentsev, I. D., Tully, B. R., Dolphin, A., et al. 2007, AJ,133, 504Kartha, S. S., Forbes, D. A., Alabi, A. B., et al. 2016, MNRAS,458, 105 Kissler-Patig, M. 1997, A&A, 319, 83Klypin, A., Kravtsov, A. V., Valenzuela, O., & Prada, F. 1999,ApJ, 522, 82Knuth, K. H. 2006, arXiv:physics/0605197Koda, J., Yagi, M., Yamanoi, H., & Komiyama, Y. 2015, ApJ,807, L2Koposov, S. E., Belokurov, V., Torrealba, G., & Evans, N. W.2015, ApJ, 805, 130Kundu A., & Whitmore B. C. 2001, AJ, 121, 2950Lada, C. J., & Lada, E. A. 2003, ARA&A, 41, 57Landy, S. D., & Szalay, A. S. 1993, ApJ, 412, 64Larsen S. S., Brodie J. P., Huchra J. P., Forbes D. A., GrillmairC. J. 2001, AJ, 121, 2974Lauberts, A., & Valentijn, E. A. 1989, Garching: European South-ern Observatory, | c1989Lee, M. G., & Jang, I. S. 2016, ApJ, 822, 70Lotz, J. M., Telford, R., Ferguson, H. C., et al. 2001, ApJ, 552,572Lotz, J. M., Miller, B. W., & Ferguson, H. C. 2004, ApJ, 613, 262Mac Low, M.-M., & Klessen, R. S. 2004, Reviews of ModernPhysics, 76, 125Mackey, A. D., Lewis, G. F., Collins, M. L. M., et al. 2014, MN-RAS, 445, L89Malin, D. R., Quinn, P. J., & Graham, J. A. 1983, ApJ, 272, L5Mart´ınez-Delgado, D., L¨asker, R., Sharina, M., et al. 2016, AJ,151, 96Masters, C. E., & Ashman, K. M. 2010, ApJ, 725, 868McLaughlin, D. E. 2000, ApJ, 539, 618McLaughlin, D. E., & van der Marel, R. P. 2005, ApJS, 161, 304McLaughlin, D. E., & Fall, S. M. 2008, ApJ, 679, 1272Mieske, S., Hilker, M., Infante, L., & Jord´an, A. 2006, AJ, 131,2442Mieske, S., Hilker, M., Jord´an, A., et al. 2008, A&A, 487, 921Mihos, J. C., Durrell, P. R., Ferrarese, L., et al. 2015, ApJ, 809,L21Miller, B. W., & Lotz, J. M. 2007, ApJ, 670, 1074Minniti, D., Alonso, M. V., Goudfrooij, P., Jablonka, P., & Mey-lan, G. 1996, ApJ, 467, 221Moore, B., Ghigna, S., Governato, F., et al. 1999, ApJ, 524, L19M¨uller, O., Jerjen, H., & Binggeli, B. 2015, A&A, 583, A79M¨uller, O., Jerjen, H., Pawlowski, M. S., & Binggeli, B. 2016,A&A, 595, A119M¨uller, O., Jerjen, H., & Binggeli, B. 2017, A&A, 597, A7Mu˜noz, R. P., Puzia, T. H., Lan¸con, A., et al. 2014, ApJS, 210, 4Mu˜noz, R. P., Eigenthaler, P., Puzia, T. H., et al. 2015, ApJ, 813,L15Ostrov, P., Geisler, D., & Forte, J. C. 1993, AJ, 105, 1762Park, H. S., & Lee, M. G. 2013, ApJ, 773, L27Pedregosa, F., Varoquaux, G., Gramfort, A., et al. 2012, JMLR,12, 2825Peng, E. 2003, Ph.D. thesis, Johns Hopkins Univ.Peng, E. W., Ford, H. C., & Freeman, K. C. 2004a, ApJS, 150,367Peng, E. W., Ford, H. C., & Freeman, K. C. 2004b, ApJ, 602, 685Peng, E. W., Ford, H. C., & Freeman, K. C. 2004c, ApJ, 602, 705Peng, E. W., Jord´an, A., Cˆot´e, P., et al. 2006, ApJ, 639, 95Peng, E. W., Jord´an, A., Cˆot´e, P., et al. 2008, ApJ, 681, 197Peng, E. W., Ferguson, H. C., Goudfrooij, P., et al. 2011, ApJ,730, 23Peng, E. W., & Lim, S. 2016, ApJ, 822, L31Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2014,A&A, 571, A1Portegies Zwart, S. F., McMillan, S. L. W., & Gieles, M. 2010,ARA&A, 48, 431Puzia, T. H., Kissler-Patig, M., Brodie, J. P., & Huchra, J. P.1999, AJ, 118, 2734Puzia, T. H., Kissler-Patig, M., Thomas, D., et al. 2004, A&A,415, 123MNRAS , 1–26 (2016) M. A. Taylor et al.
Puzia, T. H., Perrett, K. M., & Bridges, T. J. 2005a, A&A, 434,909Puzia, T. H., Kissler-Patig, M., Thomas, D., et al. 2005b, A&A,439, 997Puzia, T. H., Kissler-Patig, M., & Goudfrooij, P. 2006, ApJ, 648,383Puzia, T. H., & Sharina, M. E. 2008, ApJ, 674, 909-926Quillen, A. C., Graham, J. R., & Frogel, J. A. 1993, ApJ, 412,550Rejkuba, M. 2001, A&A, 369, 812Rejkuba, M., Greggio, L., Harris, W. E., Harris, G. L. H., & Peng,E. W. 2005, ApJ, 631, 262Rejkuba, M., Dubath, P., Minniti, D., & Meylan, G. 2007, A&A,469, 147Rejkuba, M., Harris, W. E., Greggio, L., & Harris, G. L. H. 2011,A&A, 526, A123Rejkuba, M., Harris, W. E., Greggio, L., et al. 2014, ApJ, 791,L2Richtler, T. 1995, Reviews in Modern Astronomy, 8, 163Richtler, T., Bassino, L. P., Dirsch, B., & Kumar, B. 2012, A&A,543, A131Robin, A. C., Reyl´e, C., Derri`ere, S., & Picaud, S. 2003, A&A,409, 523Salpeter, E. E. 1955, ApJ, 121, 161S´anchez-Janssen, R., Ferrarese, L., MacArthur, L. A., et al. 2016,ApJ, 820, 69Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103Schwarz, G. E. 1978, Ann. Stat., 6, 461Schechter, P. 1976, ApJ, 203, 297Searle, L., & Zinn, R. 1978, ApJ, 225, 357Seth, A., Olsen, K., Miller, B., Lotz, J., & Telford, R. 2004, AJ,127, 798Sharples, R. 1988, in IAU Symp. 126, Globular Cluster Systemsin Galaxies, ed. J. E. Grindlay & A. G. D. Philip (Dordrecht:Kluwer), 545Sinnott, B., Hou, A., Anderson, R., Harris, W. E., & Woodley,K. A. 2010, AJ, 140, 2101-2108Smith, A. J., Loveday, J., & Cross, N. J. G. 2009, MNRAS, 397,868Smith, R., S´anchez-Janssen, R., Fellhauer, M., et al. 2013, MN-RAS, 429, 1066Smith, R., S´anchez-Janssen, R., Beasley, M. A., et al. 2015, MN-RAS, 454, 2502Spitler, L. R., Larsen, S. S., Strader, J., et al. 2006, AJ, 132, 1593Spitler, L. R., Forbes, D. A., & Beasley, M. A. 2008, MNRAS,389, 1150Stickel, M., van der Hulst, J. M., van Gorkom, J. H., Schimi-novich, D., & Carilli, C. L. 2004, A&A, 415, 95Strader, J., Brodie, J. P., Spitler, L., & Beasley, M. A. 2006, AJ,132, 2333Strader, J., Caldwell, N., & Seth, A. 2011, AJ, 142, 8Taylor, M. A., Puzia, T. H., Harris, G. L.H, et al. 2010, ApJ, 712,119Taylor, M. A., Puzia, T. H., Gomez, M., & Woodley, K. A. 2015,ApJ, 805, 65Taylor, M. A., Mu˜noz, R. P., Puzia, T. H., et al. 2016, MNRAS, submitted
Tonini, C. 2013, ApJ, 762, 39Tubbs, A.D. 1980, ApJ, 241, 969Tully, B.R., Libeskind, N.I., Karachentsev, I.D., et al. 2015, ApJ,802, 25Usher C., Forbes, D. A., Brodie, J. P. et al. 2012 MNRAS, 426,1475Usher, C., Forbes, D. A., Brodie, J. P., et al. 2015, MNRAS, 446,369van den Bergh, S. 1975, ARA&A, 13, 217van den Bergh, S., Hesser, J.E., & Harris, G.L.H 1981, AJ, 86, 24van den Bergh, S. 2000, AJ, 119, 609 Van Der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Comp.in Sci. and Eng., 13, 22van Dokkum, P. G., Abraham, R., Merritt, A., et al. 2015, ApJ,798, L45van Dokkum, P., Abraham, R., Brodie, J., et al. 2016,arXiv:1606.06291VanderPlas, J., Connolly, A. J., Ivezic, Z., & Gray, A. 2012,Proceedings of Conference on Intelligent Data Understand-ing (CIDU), pp. 47-54, 2012., 47Villegas, D., Jord´an, A., Peng, E. W., et al. 2010, ApJ, 717, 603Wenger, M., Ochsenbein, F., Egret, D., et al. 2000, A&AS, 143,9Whitmore, B. C., Sparks, W. B., Lucas, R. A., Macchetto, F. D.,& Biretta, J. A. 1995, ApJ, 454, L73Woodley, K. A., Harris, W E., & Harris, G. L. H. 2005, AJ, 129,2654Woodley, K. A. 2006, AJ, 132, 2424Woodley, K. A., Harris, W. E., Beasley, M. A., et al. 2007, AJ,134, 494Woodley, K. A., Harris, W. E., Puzia, T. H., et al. 2010, ApJ,708, 1335Woodley, K. A., G´omez, M., Harris, W. E., Geisler, D., & Harris,G. L. H. 2010, AJ, 139, 1871Zepf, S. E., & Ashman, K. M. 1993, MNRAS, 264, 611This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000