Globular Cluster Populations in Four Early-Type Poststarburst Galaxies
Aparna Maybhate, Paul Goudfrooij, Francois Schweizer, Thomas H. Puzia, David Carter
aa r X i v : . [ a s t r o - ph ] O c t Draft October 29, 2018
Globular Cluster Populations in Four Early-Type PoststarburstGalaxies Aparna Maybhate , Paul Goudfrooij , Fran¸cois Schweizer , Thomas H. Puzia , DavidCarter ABSTRACT
We present a study of the globular cluster systems of four early-type post-starburst galaxies using deep g - and I -band images from the ACS camera aboardthe
Hubble Space Telescope (HST) . All the galaxies feature shells distributedaround their main bodies and are thus likely merger remnants. The color distri-bution of the globular clusters in all four galaxies shows a broad peak centeredon g − I ≈ .
4, while PGC 6240 and PGC 42871 show a significant number ofglobular clusters with g − I ≈ .
0. The latter globular clusters are interpreted asbeing of age ∼
500 Myr and likely having been formed in the merger. The colorof the redder peak is consistent with that expected for an old metal-poor popula-tion that is very commonly found around normal galaxies. However, all galaxiesexcept PGC 10922 contain several globular clusters that are significantly brighterthan the maximum luminosity expected of a single old metal-poor population.To test for multiple-age populations of overlapping g − I color, we model theluminosity functions of the globular clusters as composites of an old metal-poorsubpopulation with a range of plausible specific frequencies and an intermediate-age subpopulation of solar metallicity. We find that three of the four sample Based on observations with the NASA/ESA
Hubble Space Telescope , obtained at the Space TelescopeScience Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., underNASA contract NAS5-26555 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218; [email protected],[email protected] Carnegie Observatories, 813 Santa Barbara Street, Pasadena, CA 91101; [email protected] Plaskett Fellow, Herzberg Institute of Astrophysics, 5071 West Saanich Road, Victoria, BC V9E 2E7,Canada; [email protected] Astrophysics Research Institute, Liverpool John Moores University, 12 Quays House, Egerton Wharf,Birkenhead, CH41 1 LD, United Kingdom; [email protected] ∼ Subject headings: galaxies: elliptical and lenticular — galaxies: individual(PGC6510, PGC 10922, PGC 42871, PGC 6240) — galaxies: interactions — galaxies:star clusters
1. Introduction
Mergers seem to have played a major role in determining the shapes and dynamics ofelliptical galaxies. A few galactic mergers still occur and offer valuable clues to past evolu-tionary processes. Young globular clusters formed during mergers hold strong promise forage-dating such events, besides helping shed light on the cluster-formation process itself.Globular clusters are very useful probes of the dynamical and chemical assembly historyof galaxies. Many globular cluster (GC) systems in normal giant elliptical galaxies show abimodal color distribution, indicating the occurrence of a second event/mechanism of clusterformation. The “merger” model suggests that metal-rich (“red”) clusters are formed duringmajor mergers of gas-rich galaxies (Schweizer 1987; Ashman & Zepf 1992). Globular clus-ter systems in young merger remnants such as NGC 7252 (e.g., Miller et al. 1997) reveal abimodal color distribution featuring not only the well-known “universal” halo population ofold metal-poor GCs, but also hundreds of second-generation GCs that are young, metal-rich,and have most probably formed from the metal-enriched gas associated with the progenitorspirals. The many structural properties that merger remnants such as NGC 1275, NGC3921, and NGC 7252 share with normal ellipticals suggest not only that these merger rem-nants are proto-ellipticals (e.g., Schweizer 1998) but also that many normal ellipticals withbimodal color distributions may have formed their metal-rich GCs in a similar manner. Ifthis is indeed the case, it should be possible to find ellipticals with second-generation GC sys- 3 –tems of intermediate age ( ∼ ∼ years. Approximately 10% of these early-type shell galax-ies have colors, absolute magnitudes, and spectra characteristic of “poststarburst” galaxies(featuring strong Balmer lines Carter et al. 1988).A poststarburst spectrum is characterized by the presence of strong Balmer absorptionlines (indicative of A-type stars), but without strong [O ii ] or H α emission lines (Dressler & Gunn1983). The existence of strong Balmer absorption lines indicates that these galaxies haveexperienced star formation in the past ∼ global star formation. Itis interesting to investigate the global properties of the poststarburst class of galaxies andexplore the connection (if any) between properties of the host galaxy and its star formationhistory. We will address these issues through a study of the GC systems of these galaxiessince they provide a good handle on the impact of major star formation episodes in a galaxy.If numerous intermediate-age GCs with ages consistent with those of the poststarburst pop-ulation in the nucleus were to be found, we could be fairly confident that the A stars wereformed in a relatively vigorous star formation event, likely associated with a recent dissipa-tive galaxy merger. Conversely, an absence of such GCs would indicate that no strong starformation was associated with the event that caused star formation to end ∼ −
2. Observations and Data Reduction
The four galaxies were observed as part of HST General Observer program 10227 (PI:Goudfrooij), using the Wide Field Channel (WFC) of ACS with the filters F475W andF814W. The observations consisted of several long exposures plus a few short ones to guar-antee getting an unsaturated galaxy center. The total exposure times were 7874s, 9350s,8369s, and 21440s in F475W and 2712s, 3970s, 5908s, and 7300s in F814W for PGC 6510,PGC 10922, PGC 42871, and PGC 6240, respectively. The data were processed with theACS on-the-fly pipeline, which included dark and bias subtraction and flat fielding. Indi-vidual flat-fielded images were carefully checked for satellite trails and saturated pixels inthe central region of each galaxy. These pixels were flagged and masked out. The individualimages of a galaxy in each band were co-added using the PyRAF task MULTIDRIZZLE(Koekemoer et al. 2002). This resulted in images cleaned of cosmic rays and corrected forgeometric distortion. A combination of the short and long-exposure images enabled us toget good resultant multidrizzled images with unsaturated centers. The isophotal contours ofthe four galaxies are shown in Fig. 1. It is obvious from the figure that the sample galaxiesshow significant differences in global morphology, even though they all share the propertiesof a poststarburst spectrum and shell structure. PGC 6510 appears to be a disky elliptical,PGC 10922 shows rounder isophotes but has a distorted center (due to dust absorption),while the isophotes of both PGC 6240 and PGC 42871 appear irregular in the outer regions.
3. Cluster Candidate Selection
To select cluster candidates, the F475W and F814W images obtained from MUL-TIDRIZZLE were first added together to get a high signal-to-noise g+I image for eachgalaxy. Elliptical isophotes were then fit to this coadded image using the task ELLIPSE PyRAF is a product of the Space Telescope Science Institute, which is operated by AURA for NASA , and allowing the center, ellipticity, and position angle of the isophotesto vary. This yielded a smooth model of each galaxy’s count distribution. The search forsources was performed on an image created by dividing the g+I image by the square rootof the model image. This ensures uniform shot noise characteristics over the whole image.The detection threshold for selecting sources was set at 4 σ above the background. Typicalhalf-light radii of globular clusters fall into the range 1 – 20 pc (Kundu & Whitmore 2001;van den Bergh & Mackey 2004; Jord´an et al. 2005). At the distance of our sample galaxies,the spatial scale is 14 – 25 pc per ACS / WFC pixel. Thus, we expect the globular clusters toappear as nearly unresolved point sources. There are two advantages in using the g+I im-ages, rather than the individual g - or I -band images, for cluster-candidate selection. Firstly,the coadded images reach a greater depth than the individual images, and secondly, thephotometric zero-point of a g+I image is significantly less color dependent than that of theindividual images. A detailed illustration and discussion of these advantages can be foundin Goudfrooij et al. (2007).To perform cluster photometry, a smooth elliptical model was constructed for eachgalaxy in each filter in a manner similar to the one described above. This model was sub-tracted from the corresponding drizzled image to get a residual image. The cluster candidatesare clearly visible once the underlying galaxy light has been subtracted. Aperture photom-etry was then performed through an aperture of 3 pixel radius for all the detected sources.The photometry lists thus obtained for both filters were then matched for further analysis.Aperture corrections from 3 pixel to 10 pixel radius were determined using a few brightpoint sources in each band. The corrections from 10 pixel radius to infinity were taken fromSirianni et al. (2005). Finally, the F475W and F814W magnitudes were converted from theinstrumental system (STMAG) to SDSS g and Cousins I magnitudes in the VEGA systemvia the SYNPHOT package in STSDAS; for details, see Paper I. The g and I magnitudesand the g − I color were corrected for Milky Way foreground reddening using A V given inTable 1 and the relations A g = 1 . × A V and E ( g − I ) = 0 . × A V .After discarding clusters with photometric errors > g − I . Using the population-synthesis models GALEV of Anders & Fritze-v. Alvensleben (2003), we determined that therange 0 . ≤ g − I ≤ . to 10 years andmetallicities Z from 0.0004 to 0.05 (where Z = 0 .
02 corresponds to solar metallicity), asshown in Fig. 2. The GALEV models are calibrated to the ACS filters and were hence chosento avoid introducing systematic errors while converting to a standard system. However, the STSDAS is a product of the Space Telescope Science Institute, which is operated by AURA for NASA V − I . We also verified that the results obtainedusing the Maraston (2005) models are consistent with those from GALEV as shown in Fig.2. An extensive comparison between the various Simple Stellar Population (SSP) models canbe found in Pessev et al. (2008). Sources with colors outside the range 0 . ≤ g − I ≤ . . < FWHM < . g+I image). Artificial clusters were added to the image in batches of 100for five different background levels and in 0.25 mag intervals. Using the background levelsin the parts of the image farthest away from, and nearest to, the galaxy center where clus-ters are still detected as the two extremes, the five background levels were determined bydividing this interval into five logarithmically equal intervals. The radial-intensity profile ofthe artificial clusters was determined by fitting PSFs to the real clusters in the g+I image.A smooth elliptical model of each host galaxy was obtained, and the g+I image was dividedby the square root of this model image. Source detection was then performed on this imagein the same manner as done previously for the actual sources. Other criteria (permissibleerror in photometric magnitude, FWHM, and compactness) were also applied in exactly thesame manner as before. A typical sample of completeness curves, obtained for PGC 6510 isshown in Fig. 3. To determine the radial extent of the cluster candidates that are physically associatedwith the target galaxy versus compact background galaxies that have similar colors andapparent magnitudes, we examined the surface number density of the GC candidates as afunction of galactocentric radius. This was achieved by dividing each galaxy image intoannular rings centered on the galaxy center and computing the number of sources per unitarea in each ring. The bright background due to the presence of the galaxy would prevent thedetection of faint sources in the inner regions, while sources of the same magnitude would be 7 –easily detected in the outer regions of low background. To account for this effect, we applieda magnitude cut-off to the sources that were considered for the surface density measurements.This cut-off was chosen to be at the magnitude associated with a completeness value of 80%at a galactocentric radius of 6 ′′ . Only sources outside that radius were taken into account.Figure 4 shows the results. We find that the surface-density profile flattens off beyond acertain radius for each galaxy. This radius was taken to be the limiting radius for thesources associated with that galaxy and was found to be 60 ′′ , 94 ′′ , 104 ′′ , and 78 ′′ for PGC6510, PGC 10922, PGC 42871, and PGC 6240, respectively. All sources detected beyondthis radius were considered background sources.The radial GC surface-number-density profiles are in most cases similar to the sur-face brightness profile of the parent galaxies, which is consistent with the situation for themetal-rich subpopulation of GCs in normal early-type galaxies (Harris & Racine 1979; Harris1986; Ashman & Zepf 1998; Goudfrooij et al. 2007). However, the GCs in PGC 10922 seemto have a more extended distribution, similar to that seen for metal-poor GCs in normalearly-type galaxies. Quantitatively, simple power-law fits to the outer data points yieldexponents ranging from − .
09 to − .
94, whereas surface-number-density profiles for GCsin normal elliptical galaxies show power-law exponents in the range between − − I -band in PGC 6510, PGC 10922, andPGC 42871 (with their positions, I -band magnitudes and distance from the galaxy center)is given in Table 3. Paper I already contains a list of the brightest clusters in PGC 6240 andit is not repeated here.
4. Globular Cluster Color Distributions
Color–magnitude diagrams for the clusters associated with each galaxy within the lim-iting galactocentric radius as found in the previous section are shown for equal areas ofincreasing galactocentric distance in Fig. 5. Note that the magnitudes and colors have beencorrected for foreground reddening as described in Sect. 3. For comparison, we indicatethe area which would be populated by old metal-poor GCs similar to those in the halo ofour Galaxy, scaled to the distance of each galaxy. Typically a large number of GCs with g − I ≈ . − ∼ g − I ≈ .
4, we therefore divide up the clusters into four equal-area bins of increasinggalactocentric radii (see different symbols in Fig. 5). All color distributions appear broaderthan the halo GC distributions in our Galaxy as outlined by the rectangles. The innermostGCs in most of the galaxies have mean colors that are somewhat redder ( g − I ≈ .
5) thanthose of old metal-poor GCs. While this could in principle be a result of reddening by dustin the inner regions that shifts the clusters to fainter magnitudes and redder colors, thereare very few clusters found in the inner dusty regions. Alternatively, it could indicate thepresence of a second-generation population of clusters with ages slightly larger than 1.0 Gyr.A comparison of the dimming vector for a 1.0 Gyr old population aging to 1.6 Gyr (shown inFig. 5) with the distribution of the inner clusters in the CMD does suggest the possibility ofthese clusters being slightly older than 1.0 Gyr. Apart from that, there is also a population of(inner) clusters with colors significantly bluer than old metal-poor GCs in PGC 6240. Thus,it appears likely that the GC systems of these poststarburst galaxies are made up of morethan a single-age population. The question of how to disentangle the different populationsin a quantitative way will be discussed further in the next Section.The color distribution of the GCs was analyzed by first computing the non-parametricEpanechnikov-kernel probability-density function (Silverman 1986) of all objects within thelimiting radius. The probability-density function of all objects outside this radius was thencomputed similarly to provide an estimate of the contamination by compact backgroundgalaxies. Finally, the background density estimate was scaled to that of the cluster candi-dates by area, and the true color distribution of the GCs was derived by statistical subtrac-tion. Figure 6 shows the color distributions of the sources within the limiting radius, thebackground sources, and the inner sources after statistical background subtraction. The finalcolor distribution appears mostly unimodal in PGC 6510, while that of PGC 42871 shows thepossible presence of a second, bluer population similar to that found earlier for PGC 6240(Maybhate et al. 2007). According to the SSP models of Anders & Fritze-v. Alvensleben(2003) shown in Fig. 2, these can be explained by a metal-rich population with ages of afew hundred Myr. In case of PGC 10922, the color distribution shows some indications ofclusters with colors ≈ ≈ − . . [Fe/H] . .
0) GCs in any of the samplegalaxies (they should show up at 1 . . g − I . . is generallypresent in normal early-type galaxies (e.g., Peng et al. 2006). This is discussed further inSection 6.
5. Luminosity Functions
To derive cluster luminosity functions (LFs), we used the completeness curves describedin Sect. 3 to assign a completeness value to each cluster. This value was computed via bilinearinterpolation in cluster magnitude and background value. Clusters with completeness valuesless than 25% were excluded from being counted. The remaining clusters were divided intotwo groups based on their distance from the center of the galaxy. Clusters within the limitingradius (determined in Sect. 3) were designated as actual GC candidates and those outsidethis radius as likely background contaminants. The LFs of the GC candidates were thencorrected for background contamination using the scaled LFs of the background. The finalcorrected LFs are shown in Fig. 7.If the LFs were made up entirely of clusters belonging to an old metal-poor populationsimilar to that found in our Galaxy, and hence had a gaussian form with a turn-over at M I = − . σ = 1 . M I ≈ − .
0. However, except for PGC 10922, thegalaxies in our sample show clusters significantly brighter than this value. While the GCcolor distribution alone may show no obvious indication of the presence of more than onesubpopulation, the fact that overluminous GCs are seen in the LF which are unlikely tobelong to an old population suggests that the observed LF is due to the superposition of anold population and a younger population with mean colors that are similar to one another.For example, the LF of PGC 42871 shows an excess of luminous clusters at magnitudesbrighter than M I = − .
0. Inspection of the CMD (Fig. 5) shows that these luminousclusters predominantly have g − I colors redder than 1.5 mag and, hence, are not associatedwith the bluer peak seen in Fig. 6. SSP models show that between the ages of 1 and 2 Gyr,the colors of metal-rich clusters of solar metallicity are indistinguishable from those of typicalold metal-poor clusters. However, these intermediate-age clusters are brighter than the oldclusters by ∼ As the above example shows, the color distribution of clusters in poststarburst galaxiesof age ∼ g − I > .
15 to avoid contamination by clusters associated with the bluer peak seen inthe color distribution.We estimate the contributions of the old metal-poor GCs and the intermediate-age GCsin the following manner. First, we compute the total g and I magnitudes and g − I of thegalaxies using the ellipse fitting task and elapert within STSDAS. PGC 42871 and PGC6240 do not show evidence for their integrated colors being significantly reddened by dust.However, PGC 6510 has redder colors and the color-index map (Fig. 12) also shows strongevidence for patchy dust distributed asymmetrically around the central region. Since thedust distribution is asymmetric about the central region, we use the unreddened regionsto compute the reddening in the dusty regions and correct the total magnitudes and colorof PGC 6510. The corrected values are given in Table 4. To estimate the gaussian LFsassociated with the old metal-poor GC population, we need to apply luminosity fadingto the galaxies’ light. Treating the diffuse light of the galaxies as SSPs, we estimate theluminosity-weighted age for a solar-metallicity population from the total integrated g − I color index of each galaxy and look up the expected fading in the B and V bands when thegalaxy ages to 14 Gyr using the GALEV SSP models (see Table 4). A comparison with the 11 –Maraston models is also shown in the table. The following calculations are done using valuesfrom the GALEV models.Using the galaxy’s faded M V value in the equation for the specific frequency: S N = N GC . M V +15) (1)(i.e., the number of star clusters per galaxy luminosity normalized to an absolute V magni-tude of −
15 (Harris & van den Bergh 1981)), we calculate the number of old GCs expectedfor four typical values of S N = 3.0, 1.0, 0.5, 0.25. The specific frequency is known to increasesystematically along the Hubble sequence, from 0.5 ± ± M I = − . σ = 1 . S N . The contributions of the gaussians for each specific frequencyare plotted in the left panel of Fig. 8 for each galaxy. The residuals obtained by subtractingthese estimated gaussians from the observed LF are shown in the right panel. We find ineach case that the residuals are fit well by power laws with exponents α between − .
54 and − .
17. These exponents are consistent with values found for young and intermediate-agecluster systems (e.g., Whitmore et al. 1999, 2002; Goudfrooij et al. 2001b, 2004, 2007).Figure 8 illustrates that the observed luminosity functions can be well modeled as com-posites of an old metal-poor population, as is normally seen in early-type galaxies, and anintermediate-age population that is more metal-rich. However, note that the relative contri-butions of these two populations to the total LF vary significantly from galaxy to galaxy: Theclusters in PGC 6510 seem to be predominantly intermediate-age metal-rich ones, whereasalmost all clusters in PGC 10922 can be attributed to an old metal-poor population. Thisresult is best seen for S N values of 1.0 and 2.0 [cases (b) and (c) in the top right panel ofPGC 10922].
6. The Cluster System – Host Galaxy Connection
In this section we discuss the properties of the cluster systems in the four early-typepoststarburst galaxies of our sample in terms of implications related to the assembly of theirhost galaxies. 12 –
The GC color distribution of most normal early-type galaxies is bimodal, including a“blue” peak with a mean color that is similar to that of sentence the end of Section 6.1.metal-poor GCs in the halo of our Galaxy and a “red” peak with a mean color similarto that of the underlying diffuse galaxy light. The mean color of the red peak has beenshown to strongly correlate with the luminosity of the parent galaxy (e.g., Larsen et al.2001; Peng et al. 2006). According to the GALEV models, this peak should be seen at g − I = 1.74, 1.83 and 2.02 for Z = 0.2 Z ⊙ , 0.4 Z ⊙ , and 1.0 Z ⊙ respectively. However, we seeno clear indication of the presence of this “red peak” in any of the galaxies in our sample.This indicates that the progenitor galaxies lacked any significant number of old, metal-richGCs typically seen in normal early-type galaxies. This in turn seems to point to late-typespirals as being the progenitors of these poststarburst early-type shell galaxies since old,metal-rich clusters are typically associated with spheroidal components of galaxies (see alsoForbes et al. 2001; Goudfrooij et al. 2003). Thus, it seems unlikely that the progenitors ofthese poststarburst galaxies were ellipticals as also indicated by the presence of shells andother sharp features indicative of former disks.However, it is interesting to ask whether the sample galaxies will eventually evolve intoelliptical galaxies similar to present-day normal ellipticals. A comparison of the GCs ofpresent-day ellipticals with those of our sample galaxies can shed light on this. For anysuch comparison, we need to take into account the changes in the GC systems of our samplegalaxies as they evolve to older ages. One important such change is due to cluster disruption.Star clusters are vulnerable to disruption by a variety of processes operating on different timescales. Since the intermediate-age clusters in our sample are ∼ × M ⊙ . The number of intermediate-age clusters with masses greater than this value isnot expected to be affected by disruption to within 10% (Fall & Zhang 2001; Goudfrooij et al.2007). These relatively massive clusters are thus assumed to survive and become “old” metal-rich red GCs as the galaxy ages. This allows us to calculate the number of old metal-poorGCs and the number of expected metal-rich GCs using the results from Fig. 8 for each ofthe four values of S N . We derive the expected number of metal-rich GCs in each galaxy bycalculating the number of intermediate-age clusters with masses ≥ × M ⊙ given by thepower-law fit, and by applying passive evolution of the galaxies’ V -band luminosity to an ageof 14 Gyr using the GALEV SSP models. The resulting ratio of metal-poor to metal-rich (i.e., 13 –blue to red) clusters thus obtained is compared with that in early-type galaxies of the Virgocluster (Peng et al. 2006) in the top panel of Fig. 9. We do this for all galaxies except PGC10922, which does not show evidence for the presence of any significant intermediate-age GCpopulation (see Sect. 5.1 and the Appendix). As the figure shows, the values for PGC 6240and PGC 42871 are consistent with those for early-type Virgo cluster galaxies if 0 . . S N .
1. Under the assumption that these early-type poststarburst galaxies will indeed evolve tobecome “normal” early-type galaxies at old age, this finding not only suggests that the mostlikely progenitors of these poststarburst galaxies were spiral galaxies with Hubble types Sb orlater (Harris 1991; Ashman & Zepf 1998; Goudfrooij et al. 2003; Chandar, Whitmore, & Lee2004), but also provides new evidence based on GC system properties to support the viewthat mergers of such spiral galaxies can indeed produce “normal” early-type galaxies at oldage. The presence of strong tidal features in the sample galaxies (and many poststarburstgalaxies in general) also suggests disk-dominated progenitors (e.g., Zabludoff et al. 1996;Mihos & Hernquist 1996).Note that the above analysis was performed by taking into account the luminosity-weighted age of the galaxies. Table 4 shows that the integrated colors of PGC 6240 andPGC 42871 are indeed consistent with those of their intermediate-age GC populations.Hence, this assumption seems fair for those two galaxies. However, for PGC 6510, theintegrated color of the galaxy is somewhat redder than that of the intermediate-age GCpopulation. To evaluate whether this difference may account for its low values of the ratio ofthe number of metal-poor to metal-rich clusters seen in the top panel of Fig. 9, we considerthe alternative scenario where the diffuse light of the galaxies is produced by a combinationof an old and a young population (i.e., that the galaxy in question is an “E+A” galaxy ).In this context, we assign the young population to have an age as indicated by the Balmerline equivalent widths in the central spectra of Carter et al. (1988, see Appendix A), and tohave solar metallicity. For the old population, we estimate a metallicity by evaluating thegalaxies’ absolute K -band magnitudes relative to that of NGC 4472, the brightest ellipti-cal galaxy in the Virgo cluster, and then utilizing the [Fe/H] versus galaxy M B relation ofPeng et al. (2006, transformed to M K using K galaxy magnitudes from 2MASS). Using theGALEV SSP models and the Maraston models, we then model the integrated g − I color ofthe galaxies (listed in Table 4) as a linear combination of an old component (with an age of14 Gyr) plus a young component (with an age determined as mentioned above). Finally, theluminosity fading of this composite galaxy to an effective age of 14 Gyr is derived by fadingonly the young component. Table 5 lists all relevant fading values for the sample galaxies. Many papers now use the more general term k+a, where the k stands for the spectral type of an oldstellar population (e.g., Franx 1993).
14 –Using the faded magnitudes thus obtained, we repeat the procedure in Section 5.1and model the GC luminosity function as a superposition of an old metal-poor and anintermediate-age metal-rich population, and re-derive the predicted ratio of the number ofblue to red GCs at an age of 14 Gyr. For PGC 6240 and PGC 42871, this exercise resultsin a predicted number of old metal-poor GCs that far exceeds the number of GCs observed,even at the brightest GC magnitudes. This indicates that the use of luminosity-weightedages to model the contributions of the metal-poor and metal-rich GCs gives more realisticresults than using a combination of ages for those two galaxies. This is consistent with thefact that the integrated colors of those two galaxies are very similar to the mean color oftheir intermediate-age GCs. On the other hand, for PGC 6510, we find that the scenarioinvolving a combination of young and old ages yields a prediction for the number ratio ofthe blue to red GCs that is fully consistent with those of normal early-type galaxies if theprogenitor galaxies had S N ≈ . To address the spatial extent of the poststarburst population in the sample galaxies, weuse the radial color distribution of the host galaxy as a proxy. Bartholomew et al. (2001) findthat poststarburst galaxies in distant clusters tend to have slightly bluer gradients towardsthe center than “normal” early-type galaxies. Recent numerical simulations by Bekki et al.(2005) suggest that elliptical E+A galaxies formed by major mergers should have positiveradial color gradients (i.e., bluer color in the inner regions). Examining the radial g − I color-index profile of the underlying galaxies obtained from the ellipse fits made in the twopassbands, we find that PGC 6240 and PGC 42871 get redder outwards, whereas PGC 6510and PGC 10922 show redder colors in the inner regions (Fig. 11). This behavior is also seenin the two-dimensional g − I color-index maps shown in Fig. 12. It may well be relevantthat the two galaxies that show evidence for hosting the youngest GC populations (i.e.,PGC 6240 and PGC 42871) also show the bluest color in their central regions. These twogalaxies also have the strongest H δ equivalent widths in their nuclear spectrum among thesample galaxies (Carter et al. 1988, see also Appendix A). The other two galaxies (PGC6510 and PGC 10922) show dusty central regions (especially PGC 10922), and it seemslikely that the central reddening in their radial color distributions can at least partly beattributed to dust (cf. Fig. 12). Note that the inner color profiles of the latter two galaxiesshow a dip to bluer colors towards their very center (Fig. 11), which may well represent a 15 –signature of the poststarburst population in these galaxies. However, Figs. 11 and 12 seemto argue that the spatial extent of the poststarburst population in PGC 10922 and (to alesser extent) PGC 6510 is smaller than that in PGC 6240 and PGC 42871. We note thatthis finding is reinforced by the properties of the GC systems of these galaxies. We suggestthat deep optical spectra of the target galaxies be obtained to verify the spatial extent ofthe poststarburst population. This would also provide quantitative information on the issueas to how well the integrated color of the target galaxies can reliably be interpreted as asimple stellar population, which is relevant to analyses related to the evolution of GC specificfrequencies in intermediate-age galaxies (cf. Sect. 6.1). As mentioned briefly in Sect. 5.1, the GC system of PGC 6510 shows some interestingproperties. Figures 8 and 9 show that if the current integrated g − I color of PGC 6510 isinterpreted in terms of an SSP (i.e., a single age and metallicity), the modeling in Sect. 5results in the old metal-poor GCs making up an unexpectedly small fraction of the totalGC system unless the progenitor galaxies had an unusually high specific frequency of oldmetal-poor GCs ( S N & S N values are unknown among gas-rich galaxies, whereas a significant amount of gas wasneeded to trigger the starburst that led to the poststarburst spectrum and the formation ofthe significant number of intermediate-age GCs. However, in case of the alternative scenarioin which the integrated color of PGC 6510 is due to a superposition of a young and an oldcomponent (see Sect. 6.1 above for details), our modeling shows that progenitor galaxies with S N values of 0.5 – 1 (typical for late-type galaxies) can account for a more significant fractionof GCs in PGC 6510 being of the old metal-poor kind, especially at the faint end of the LF(compare Fig. 10 with the top left panel of Fig. 8). To further constrain the fractions of GCsin this galaxy that are of intermediate age versus old, we utilize the large size of the GCpopulation to compare the radial surface number density profiles of the bright vs. the faintGCs in this galaxy. To avoid introducing a bias due to the high background in the innermostregions of the galaxy, we first determine the background level at which a GC of magnitude I = 25.5 is detected at 80% completeness and discard all GCs at higher background levels forthis exercise. We then calculate the completeness-corrected surface-number densities of thebrightest 33% and the faintest 33% of the GCs selected this way. We find that the surface-number-density profile of the bright GCs follows the surface brightness profile of the galaxyvery closely (see Fig. 13, left panel), whereas the faint GCs show a more extended radialdistribution. This reinforces the idea that the brightest GCs are primarily intermediate-age ones since the radial profile of intermediate-age GCs is expected to follow the surface 16 –brightness profile of the parent galaxy (e.g., Schweizer et al. 1996; Whitmore et al. 1997;Goudfrooij et al. 2007). We also plot the completeness- and background-corrected LFs ofthe inner half and the outer half of the GC system (within the outer radius beyond whichobjects were assigned to the background) and find that the inner half of the GC systemhosts relatively more bright GCs (see Fig. 13, right panel). Both of the above results lendcredence to the idea that PGC 6510 hosts a significant number of intermediate-age GCs andthat old GCs populate the LF mainly at fainter magnitudes, as predicted by our modelingin Sect. 6 for the case of a composite age structure for the diffuse light of PGC 6510.
7. Summary and Conclusions
We have analyzed
HST/ACS images of four early-type shell galaxies, for which spec-troscopy of the central region revealed a poststarburst spectrum, in order to study theproperties of their globular cluster systems. Our results are summarized as follows. • The color distributions of the globular clusters in all four galaxies show a broad peakcentered on g − I ≈ .
4, while PGC 6240 and PGC 42871 also have a significant numberof GCs with g − I ≈ .
0. The mean color of the former peak is consistent with SSPmodel predictions for both an old ( ∼
14 Gyr), metal-poor ([Z/H] ∼ − .
5) populationand an intermediate-age (1 – 2 Gyr) population of roughly solar metallicity. The GCswith g − I ≈ . ∼
500 Myr and likely having formed inthe merger. Except for PGC 10922, the galaxies host several GCs in the redder peakthat are brighter than the maximum luminosity expected of a single old, metal-poor“halo” GC of the kind commonly found around normal galaxies. To test for multiplepopulations of overlapping color around g − I ≈ .
4, we fit the observed GCLFs ascomposites of a Gaussian (as seen in “old” GC systems) and a power law (seen inyoung and intermediate-age GC systems). By scaling the Gaussian component usingplausible values of the specific frequencies ( S N values) of old metal-poor GCs seen inpresent-day normal galaxies, we find the following:1. We deduce the presence of a substantial population of intermediate-age GCs inthree out of four galaxies in our sample. These GCs have ages between 1 – 1.5Gyr which is comparable to typical lifetimes of the shells, providing evidence thatthese GCs likely formed during the same merger event that formed the shells.2. The integrated colors of PGC 6240 and PGC 42871 are consistent with thoseof the intermediate-age GCs, suggesting that the bulk of the field stars wereformed in the same star formation event that formed the intermediate-age GCs. 17 –Interpreting the integrated light of PGC 6240 and PGC 42871 as a single-agestellar population of solar metallicity, the ratio of the numbers of metal-poor tometal-rich GCs are consistent with those of present-day (giant) early-type galaxiesif the S N values of the merger progenitor galaxies were in the range 0 . . S N . S N values is consistent with that of late-type spiral galaxies. Underthe assumption that these early-type poststarburst galaxies will indeed evolve tobecome “normal” early-type galaxies at old age, this finding not only suggeststhat the most likely progenitors of the poststarburst galaxies in our sample werelate-type spiral galaxies, but also provides new evidence (based on GC systemproperties) to support the view that mergers of such spiral galaxies can indeedproduce “normal” early-type galaxies at old age.3. The integrated color of PGC 6510 is redder than that of its bright GCs that arevery likely of intermediate age, suggesting that a significant fraction of the fieldstars are older than those GCs. Interpreting the integrated light of PGC 6510 asa superposition of a young population (with solar metallicity) and a 14-Gyr oldpopulation (with a metallicity estimated from the galaxy’s K -band luminosity),the ratio of the numbers of metal-poor to metal-rich GCs is consistent with thoseof present-day (giant) early-type galaxies if the S N values of the merger progenitorgalaxies were in the range 0 . . S N .
1, just like the situation for PGC 6240and PGC 42871. GCs at the bright end of the GCLF in PGC 6510 follow thesurface brightness profile of the parent galaxy closely, supporting the notion thatthey are indeed intermediate-age clusters. In contrast, GCs at the faint end of theLF (but still at 80% completeness) show a flatter surface number density profile,consistent with the presence of a significant fraction of old GCs among the fainterGCs.4. We find no evidence for the presence of intermediate-age GCs in PGC 10922, thegalaxy with the smallest H δ equivalent width in its nuclear spectrum among oursample. This may partly be due to the presence of a very dusty inner region (of ∼ • The color distributions of GCs in all four galaxies appear devoid of any old metal-richclusters, which are generally associated with the spheroidal component of (early-type)galaxies. This indicates that the progenitor galaxies must have lacked any substantialbulge and were most likely of late Hubble type. 18 –In closing, we note that our analysis is based on g – and I –band imaging and, hence, onmagnitudes and one color index only. Given the age-metallicity degeneracy among opticalcolors (e.g., Fig. 2), additional spectra and/or near-infrared icolors will clearly be needed tocheck the ages and metallicities deduced for the second-generation GCs.Early-type poststarburst galaxies are thought to represent an intermediate phase in theformation of (at least) some elliptical galaxies from mergers. Our results, based on propertiesof GC systems of such galaxies, generally support the idea that mergers of spiral galaxies areaccompanied by the formation of clusters which evolve and form the old, metal-rich peak ofthe GC color distribution as the merger remnant evolves into an elliptical galaxy. A. Notes on individual galaxiesPGC 6510 : The equivalent width (EW) of H δ in the central spectrum for this galaxy is5 . ± . ∼ g − I color-index map of PGC 6510 shows an asymmetricaldust distribution (Fig. 12). The properties of the GC system of this galaxy are somewhatpeculiar and do not seem to be consistent with those of the other galaxies in the sample orwith normal present-day ellipticals, unless significant disruption of the newly formed clusterswill take place over the next ∼
10 Gyr. However, modeling the diffuse light of the galaxy asa combination of an old and a young population gives plausible results for the properties ofthe GC systems.
PGC 10922 : The H δ EW for this galaxy is 3 . ± . g − I map, where the centralregion appears surrounded by a circular dust ring with spurs. It seems plausible that thepoststarburst population is localized near the center and the dusty structure in this regionprevents us from detecting any clusters there. 19 – PGC 42871 : The H δ EW is 12 . ± . ∼ g − I ≈ .
7, which is interpreted as representing a population of younger-age ( ∼
80 – 200 Myr) clusters based on a comparison with SSP models of solar metallicity.The redder peak at g − I = 1.35 is likely made up of a composite population of old metal-poorclusters and intermediate-age solar-metallicity clusters, as discussed in Section 5. Thus, wefind evidence for the presence of three separate subpopulations of GCs in this galaxy. PGC 6240 : The H δ EW for this galaxy is 13 . ± . ∼ g − I = 0.85 consistent with a ∼ HST
Program GO-10227 was provided by NASA through a grant to PG fromthe Space Telescope Science Institute, which is operated by the Association of Universitiesfor Research in Astronomy, Inc., under NASA contract NAS5–26555.
Facilities:
HST (ACS).
REFERENCES
Anders, P., & Fritze-v. Alvensleben, U. 2003, A&A, 401, 1063Ashman, K.M., & Zepf, S.E. 1998, Globular Cluster Systems (Cambridge Univ. Press)Ashman, K.M., & Zepf, S.E. 1992, ApJ, 384, 50Barmby, P., Huchra, J.P., & Brodie, J, P. 2001, AJ, 121, 1482Bartholomew, L.J., Rose, J.A., & Gaba, A.E. 2001, AJ, 122, 2913Bekki, K., Couch, W.J., Shioya, Y., & Vazdekis, A. 2005, MNRAS, 359, 949Brown, R.J.N., Forbes, D.A., Kissler-Patig, M., & Brodie, J.P. 2000, MNRAS, 317, 406Bruzual, G.A., & Charlot, S. 2003, MNRAS, 344, 1000 20 –Burstein, D., & Heiles, C. 1982, AJ, 87, 1165Buyle, P., Michielsen, D., De Rijcke, S., Pisano, D.J., Dejonghe, H., & Freeman, K. 2006,ApJ, 649, 163Carter, D., Prieur, J.L., Wilkinson, A., Sparks, W.B., & Malin, D.F. 1988, MNRAS, 235,813Chandar, R., Whitmore, B. C., & Lee, M. G. 2004, ApJ, 611, 220di Matteo, P., Combes, F., Melchior, A.-L., & Semelin, B. 2007, A&A, 468, 61Dressler, A., & Gunn, J.E. 1983, ApJ, 270, 7Dupraz, C., & Combes, F. 1986, A&A, 166, 53Fall, S.M., & Zhang, Q. 2001, ApJ, 561, 751Forbes, D.A., Beasley, M.A., Brodie, J.P., & Kissler-Patig, M. 2001, ApJ, 563, L143Franx, M. 1993, PASP, 105, 1058Gnedin, O.Y., & Ostriker, J.P. 1998, ASPC, 136, 56Goudfrooij, P., Mack, J., Kissler-Patig, M., Meylan, G., & Minniti, D. 2001a, MNRAS, 322,643Goudfrooij, P., Alonso, M.V., Maraston, C., & Minniti, D. 2001b, MNRAS, 328, 237Goudfrooij, P., Strader, J., Brenneman, L., Kissler-Patig, M., Minniti, D., & Huizinga, E.J.2003, MNRAS, 343, 665Goudfrooij, P., Gilmore, D., Whitmore, B.C., Schweizer, F. 2004, ApJ, 613, L121Goudfrooij, P., Schweizer, F., Gilmore, D., & Whitmore, B.C. 2007, AJ, 133, 273Harris, W.E., & Racine, R. 1979, ARA&A, 17, 241Harris, W.E., & van den Bergh, S. 1981, AJ, 86, 1627Harris, W.E. 1986, AJ, 91, 822Harris, W.E. 1991, ARA&A, 29, 543Harris, W.E. 1996, AJ, 112, 1487 21 –Hau, G.K.T., Carter, D., & Balcells, M. 1999, MNRAS, 306, 437Hernquist, L., & Quinn, P. 1988, ApJ, 331, 682Hernquist, L., & Quinn, P. 1989, ApJ, 342, 1Jord´an, A., Cˆote, P., Blakeslee, J. P., et al. 2005, ApJ, 634, 1002Kaviraj, S., Kirkby, L.A., Silk, J., & Sarzi, M. 2007, astro-ph/0707.3570
Koekemoer, A.M., Fruchter, A.S., Hook, R.N., & Hack, W. 2002, in
The 2002 HST Calibra-tion Workshop: Hubble after the Installation of the ACS and the NICMOS CoolingSystem , ed. S. Arribas, A. Koekemoer, & B. Whitmore (Baltimore: Space TelescopeScience Institute), 337Kundu, A., et al. 2005, ApJ, 634L, 41Kundu, A., & Whitmore, B. C. 2001, AJ, 121, 2950Larsen, S. S., Brodie, J. P., Huchra, J. P., Forbes, D. A., & Grillmair, C. J., 2001, AJ, 121,2974Larsen, S.S., Brodie, J.P., Beasley, M.A., Forbes, D.A., Kissler-Patig, M., Kuntschner, H.,& Puzia, T.H. 2003, ApJ, 585, 767Malin, D.F., & Carter, D. 1983, ApJ, 274, 534Maraston, C. 2005, MNRAS, 362, 799Maybhate, A., Goudfrooij, P., Schweizer, F., Puzia, T., & Carter, D. 2007, AJ, 134, 1729(Paper I)Mihos, J. C., & Hernquist, L. 1996, ApJ, 464, 641Miller, B.W., Whitmore, B.C., Schweizer, F., & Fall, S.M. 1997, AJ, 114, 2381Minniti, D., Alonso, M.V., Goudfrooij, P., Jablonka, P., & Meylan, G. 1996, ApJ, 467, 221Peng, E.W., Ford, H.C., & Freeman, K.C. 2004, ApJ, 602, 705Peng, E., et al. 2006, ApJ, 639, 95Pessev, P.M., Goudfrooij, P., Puzia, T.H., & Chandar, R. 2008, arXiv:0801.2375
Puzia, T.H., Zepf, S.E., Kissler-Patig, M., Hilker, M., Minniti, D., & Goudfrooij, P. 2002,A&A, 391, 453 22 –Puzia, T.H., et al. 2004, A&A, 123, 143Quinn, P.J. 1984, ApJ, 279, 596Salpeter, E.E. 1955, ApJ, 121, 161Schweizer, F. 1987, in
Nearly Normal Galaxies , ed. S.M. Faber (New York: Springer), 18Schweizer, F., Miller, B.W., Whitmore, B.C., & Fall, S.M. 1996, AJ, 112, 1839Schweizer, F. 1998, in
Galaxies: Interactions and Induced Star Formation , ed. D. Friedli, L.Martinet, & D. Pfenniger (New York: Springer), 105Schweizer, F., & Seitzer, P. 1998, AJ, 116, 2206Schweizer, F., Seitzer, P., & Brodie, J, P. 2004, AJ, 128, 202Silverman, B.W. 1986, in
Density Estimation for Statistics and Data Analysis , Chap andHall/CRC Press, Inc.Sirianni, M., et al. 2005, PASP, 117,1409Skrutskie, M.F., et al. 2006, AJ, 131, 1163Thomson, R.C. 1991, MNRAS, 253, 256Thomson, R.C., & Wright, A.E. 1990, MNRAS, 247, 122van den Bergh, S., & Mackey, A. D. 2004, MNRAS, 354, 713Vesperini, E., & Zepf, S. E., ApJ, 587, L97Vesperini, E., Zepf, S. E., Kundu, A., & Ashman, K. M. 2003, ApJ, 593, 760Whitmore, B.C., Miller, B.W., Schweizer, F., & Fall, S.M. 1997, AJ, 114, 1797Whitmore, B. C., Zhang, Q., Leitherer, C., Fall, S. M., Schweizer, F., & Miller, B. W. 1999,AJ, 118, 155Whitmore, B. C., Schweizer, F., Kundu, A., & Miller, B. W. 2002, AJ, 124, 147Zabludoff, A.I., Zaritsky, D., Lin, H., Tucker, D., Hashimoto, Y., Schectman, S.A., Oemler,A., & Kirshner, R.P. 1996, ApJ, 466, 104
This preprint was prepared with the AAS L A TEX macros v5.2.
23 –Table 1. General properties of the sample.Parameter PGC 6510 PGC 10922 PGC 42871 PGC 6240Malin & Carter (1983) MC 0148 −
836 MC 0247 −
833 MC 1241 −
339 MC 0140 − − G 013 AM 1241 −
335 AM 0139 − ∗ (J2000) 1h 46m 21.9s 2h 53m 35.9s 12h 44m 05.2s 1h 41m 30.98sDec ∗ (J2000) − ◦ ′ ′′ − ◦ ′ ′′ − ◦ ′ ′′ − ◦ ′ ′′ Type E-S0 S0 S0-a S0 v hel (km s − ) 4652 ±
15 4819 ±
15 6074 ±
15 8216 ± v LG (km s − ) 4365 4529 5944 7936Velocity disp. (km s − ) 181.9 ± ± ± ± ∗∗ m − M M B − − − − M K s − − − − A ∗∗∗ V ∗ From ACS images, this work. ∗∗ Using H = 75 km s − M pc − ∗∗∗ Burstein & Heiles (1982)Note. — All other parameters are taken from LEDA (http://leda.univ-lyon1.fr/ or com-puted using values from LEDA and H = 75 km s − M pc − )Table 2. Results of power-law fits to the GC surface density.Galaxy I -band cut-off Radial range( ′′ ) Power-law exponentPGC 6510 26.0 14 – 137 − . ± . − . ± . − . ± . − . ± .
28 24 –Table 3. Positions and photometry of the 30 brightest GCCs in PGC 6510, PGC 10922,and PGC 42871RA (2000) Dec (2000) I (mag) g − I (mag) r ∗ ( ′′ )PGC 65101 46 22.42 −
83 23 58.74 20.38 ± ± −
83 23 58.71 21.15 ± ± −
83 24 13.69 21.62 ± ± −
83 24 30.83 21.73 ± ± −
83 24 01.83 21.80 ± ± −
83 23 48.45 21.91 ± ± −
83 23 56.99 22.05 ± ± −
83 24 01.64 22.05 ± ± −
83 23 55.09 22.13 ± ± −
83 23 47.14 22.15 ± ± −
83 24 22.08 22.22 ± ± −
83 23 52.50 22.41 ± ± −
83 23 53.05 22.48 ± ± −
83 24 16.34 22.54 ± ± −
83 23 54.19 22.55 ± ± −
83 24 02.90 22.56 ± ± −
83 23 02.92 22.57 ± ± −
83 23 56.47 22.61 ± ± −
83 23 56.69 22.73 ± ± −
83 24 08.91 22.81 ± ± −
83 23 46.24 22.97 ± ± −
83 23 18.06 22.99 ± ± −
83 24 02.48 23.05 ± ± −
83 24 05.53 23.05 ± ± −
83 23 53.34 23.06 ± ± −
83 23 58.87 23.13 ± ± −
83 23 59.59 23.18 ± ± −
83 23 37.82 23.20 ± ± −
83 23 55.94 23.24 ± ± −
83 23 53.04 23.25 ± ± −
83 08 46.14 20.26 ± ± −
83 09 30.32 20.34 ± ± −
83 09 01.99 20.49 ± ± −
83 08 19.50 20.82 ± ± −
83 09 27.39 20.84 ± ± −
83 08 11.53 20.94 ± ± I (mag) g − I (mag) r ∗ ( ′′ )2 52 31.78 −
83 08 46.46 21.63 ± ± −
83 08 24.44 21.74 ± ± −
83 07 07.20 21.78 ± ± −
83 10 15.81 21.80 ± ± −
83 09 04.06 21.87 ± ± −
83 07 37.37 21.99 ± ± −
83 06 52.96 22.24 ± ± −
83 08 00.83 22.41 ± ± −
83 08 47.67 22.58 ± ± −
83 07 18.51 22.71 ± ± −
83 08 33.09 22.74 ± ± −
83 09 12.94 22.80 ± ± −
83 09 14.10 22.81 ± ± −
83 08 58.09 23.18 ± ± −
83 07 16.66 23.24 ± ± −
83 07 51.23 23.29 ± ± −
83 08 31.17 23.38 ± ± −
83 08 21.60 23.62 ± ± −
83 07 35.29 23.66 ± ± −
83 08 04.21 23.83 ± ± −
83 09 36.50 24.10 ± ± −
83 07 11.32 24.12 ± ± −
83 07 23.31 24.20 ± ± −
83 08 44.55 24.24 ± ± −
34 10 29.57 20.25 ± ± −
34 12 16.48 20.40 ± ± −
34 11 47.39 20.72 ± ± −
34 10 31.77 20.94 ± ± −
34 13 11.03 21.27 ± ± −
34 12 02.16 21.50 ± ± −
34 12 35.41 21.65 ± ± −
34 12 23.02 22.23 ± ± −
34 13 16.33 22.27 ± ± −
34 11 53.42 22.33 ± ± −
34 12 30.94 22.52 ± ± −
34 12 09.38 22.63 ± ± −
34 12 53.29 22.66 ± ± I (mag) g − I (mag) r ∗ ( ′′ )12 44 04.56 −
34 12 04.23 22.67 ± ± −
34 12 10.99 22.80 ± ± −
34 11 17.92 23.04 ± ± −
34 11 01.01 23.14 ± ± −
34 12 12.38 23.16 ± ± −
34 11 41.88 23.16 ± ± −
34 12 24.86 23.21 ± ± −
34 12 23.44 23.24 ± ± −
34 12 25.57 23.28 ± ± −
34 12 22.47 23.35 ± ± −
34 11 54.34 23.41 ± ± −
34 12 15.98 23.43 ± ± −
34 12 16.41 23.47 ± ± −
34 10 32.79 23.47 ± ± −
34 11 52.03 23.67 ± ± −
34 12 20.27 23.78 ± ± −
34 10 36.40 23.82 ± ± Galaxy g ( g − I ) SSP Age V est Fading in
V M V, faded (mag) (mag) (Gyr) (mag) (mag) (mag)(1) (2) (3) (4) (5) (6) (7)Anders & FvAPGC6510 14.17 1.49 1.4 13.68 2.27 -17.86PGC 42871 13.04 1.30 1.0 12.61 2.54 − . − . − . − . g magnitude corrected for Galactic foreground extinc-tion. Column (3): Total g − I color corrected for Galactic foreground extinction. Column (4): Luminosity-weighted age at solar metallicity. Column (5): V magnitude of the galaxy at the age in the previous column.Column (6): Fading in V to an age of 14 Gyr. Column (7): Faded absolute V magnitude. Table 5: Fading results using a combination of young and old components
Galaxy Young age L frac , young Fading (mag)(Gyr)
B V B V (1) (2) (3) (4) (5) (6)Anders & FvAPGC 6510 1.0 0.61 0.53 0.93 0.72PGC 42871 0.3 0.58 0.41 0.92 0.55PGC 6240 0.3 0.58 0.41 0.92 0.55MarastonPGC 6510 1.0 0.46 0.40 0.61 0.50PGC 42871 0.3 0.47 0.34 0.69 0.43PGC 6240 0.3 0.47 0.34 0.69 0.43Note. — Column (1): Object ID. Column (2): Age of young component as derived from Balmer linestrengths in central spectrum (cf. Appendix A). Column (3): Current luminosity fraction of young componentin the B filter. Column (4): Same as column (3), but in the V filter. Column (5): Fading of compositepopulation (due to young component) in B to an age of 14 Gyr. Column (6): Same as column (5), but in V .
29 –Fig. 1.— Contour plots of the sample galaxies. Contours represent I -band surface-brightnessvalues from 13.5 mag arcsec − to 21.5 mag arcsec − for PGC 6240 ( bottom right ) and 13.5mag arcsec − to 20.5 mag arcsec − for the remaining galaxies ( top left: PGC 6510, topright:
PGC 10922, and bottom left:
PGC 42871). In spite of several common properties likemorphological type, poststarburst spectrum, and presence of shells, the isophotes show asubstantial variety among the four galaxies. 30 –Fig. 2.—
Top panel : Time evolution of the g − I color index using GALEV SSP models(Anders & Fritze-v. Alvensleben 2003). Model curves are plotted for a Salpeter (1955) IMFand metallicities as indicated in the figure. g − I = 1.35 is consistent with both the universalmetal-poor population (Z ∼ ⊙ ) as well as a 1 Gyr population with solar metallicity. Bottom panel : Time evolution of the g − I color index for the blue horizontal branch usingMaraston models (Maraston 2005). The ages obtained for g − I = 1.35 is consistent withthat obtained using the GALEV models. 31 –Fig. 3.— A set of typical completeness curves determined for the combined g+I image forfive different values of the background. The curves shown are for PGC 6510 and representfrom left to right: 1000, 530, 276, 150, 80 counts per pixel for an effective g+I exposuretime of 705 s. 32 –Fig. 4.— The surface number density of the GC candidates is compared to the surfacebrightness profile of the underlying galaxy. The crosses denote the surface brightness of theunderlying galaxy light with an arbitrary zeropoint. The squares represent the logarithmicnumber density of GC candidates per square arcsec. The solid lines represent power-law fitsto the GC surface number density in the outer regions. 33 –Fig. 5.— g versus g − I color–magnitude diagrams of the GC candidates in each galaxy withinthe limiting radius given in Sect. 3.1. The red, green, blue, and black circles represent GCcandidates within equal areas of increasing galactocentric radius with the red ones foundat the smallest radii. The rectangle enclosed by dashed lines represents the magnitude andcolor range expected for old metal-poor GCs, and the arrow is the vector for dimming byage from 1 Gyr to 1.6 Gyr as determined from the GALEV models. 34 – NNNN
Fig. 6.— GC color distributions for each of the sample galaxies. The open histogram in eachpanel represents the observed color distribution corrected for foreground reddening, whilethe hatched histogram represents the background-subtracted distribution. The distributionof the background sources is shown in the lower panel for each case. The solid black lineis the probability-density estimate of the uncorrected distribution. The solid red line is anon-parametric probability-density estimate using an adaptive Epanechnikov kernel of thebackground-corrected GC color distribution. The dashed red lines mark the bootstrapped90% confidence limits. 35 –Fig. 7.— Luminosity functions of the GCs in the I -band corrected for background con-tamination and completeness. The I -band magnitudes have been corrected for foregroundreddening. Note the presence of several luminous clusters brighter than M I = − . PGC 42871 PGC 6240PGC 10922PGC 6510 I
Fig. 8.— I -band luminosity functions (LFs) of the GCs in the sample galaxies. The his-togram in the left panel of each figure shows the LF of the galaxy corrected for backgroundcontamination and completeness. The dashed curves represent the estimated gaussians rep-resenting the old metal-poor GCs with S N values of 3.0, 1.0, 0.5, and 0.25 (from the top curveto the bottom curve using the GALEV model), for PGC 6510, PGC 42871, and PGC 6240.In the case of PGC 10922, the dashed curves represent the estimated gaussians representingthe old metal-poor GCs with S N values of 3.0, 2.0, 1.0, 0.5, and 0.25, respectively. The his-tograms in the right panel for each galaxy show the residual LF obtained after subtractingthe contribution of each gaussian from the total LF. The solid lines in the right panels arepower-law fits to these residual cluster LFs. The best-fit exponents α are indicated in eachpanel. 37 –Fig. 9.— Top panel : The number ratio of metal-poor blue GCs to metal-rich red GCs ofthe sample galaxies (using GALEV) for S N (= 3.0, 1.0, 0.5, and 0.25 from the top to thebottom) of the old metal-poor GCs is plotted versus galaxy M B and compared with valuesfor normal early-type galaxies (black filled circles) from Peng et al. (2006). The M B values ofthe sample galaxies have been faded from their current luminosity-weighted ages to an age of14 Gyr. PGC 10922 is not plotted because we do not detect any significant intermediate-ageGC population in it, whereas we do so in the other galaxies. Bottom panel : The number ratioof metal-poor blue GCs to metal-rich red GCs (using GALEV) computed by considering theintegrated light of the galaxies to be due to a superposition of old and young components.See Sect. 6.1 for more detailed information. 38 –
PGC 6510 I
Fig. 10.—
Left panel: I -band LF of the GCs in PGC 6510. The histogram in the leftpanel shows the LF of the galaxy corrected for background contamination and completeness(similar to that shown in Fig. 8. The dashed curves represent the estimated gaussiansrepresenting the old metal-poor GCs with S N values of 3.0, 1.0, 0.5, and 0.25 (from the topto the bottom curve). In this case, the diffuse light of the galaxy is modeled as a combinationof an old and a young population.(See Sect. 6.1 for details). Right panel:
The histogramsshow the residual LF obtained after subtracting the contributions to each gaussian fromthe total LF. The solid lines are power-law fits to these residual cluster LFs. The best-fitexponents α are indicated in each panel. 39 –Fig. 11.— The radial distribution of the g − I color of the diffuse galaxy light. The centralregions of PGC 6510 and PGC 10922 are redder than their outer regions, whereas PGC 42871and PGC 6240 have blue central regions. In case of PGC 6510, the open squares representthe observed colors and the open triangles represent the colors obtained after correction fordust. 40 –Fig. 12.— g − I color-index maps of the central regions of the sample galaxies. Darker regionsrepresent bluer colors and lighter regions represent redder colors. The images are 5 kpc oneach side. Clockwise from top left: PGC 6510, PGC 10922, PGC 6240 and PGC 42871. 41 –Fig. 13.— Left panel : The surface number density of the brightest 33% (open red squares)and the faintest 33% of the clusters with completeness ≥
80% in the innermost radial bin(filled blue triangles) in PGC 6510 are plotted as a function of galactocentric radius. Thesurface brightness profile of PGC 6510 is also plotted on an arbitrary scale. Note that theprofile of the bright clusters follows the galaxy surface-brightness profile closely.
Right panel :The difference in the number of clusters per magnitude bin between the inner and the outerhalves of the system of cluster candidates of PGC 6510 plotted for clusters with I ≤ . ≥ ′′ , which ensures a completeness of ≥≥