Tidal dwarf galaxies in the nearby Universe
Sugata Kaviraj, Daniel Darg, Chris Lintott, Kevin Schawinski, Joseph Silk
aa r X i v : . [ a s t r o - ph . C O ] A ug Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 15 June 2018 (MN L A TEX style file v2.2)
Tidal dwarf galaxies in the nearby Universe
Sugata Kaviraj , ⋆ , Daniel Darg , Chris Lintott , Kevin Schawinski , andJoseph Silk Blackett Laboratory, Imperial College London, London SW7 2AZ, UK Department of Physics, University of Oxford, Keble Road, Oxford, OX1 3RH, UK Department of Physics, Yale University, New Haven, CT 06511, USA Einstein Fellow
15 June 2018
ABSTRACT
We present a statistical observational study of the tidal dwarf (TD) population inthe nearby Universe by exploiting a large, homogeneous catalogue of galaxy mergerscompiled from the
Sloan Digital Sky Survey . 95% of TD-producing mergers involve twospiral progenitors (typically both in the blue cloud), while most remaining systemshave at least one spiral progenitor. The fraction of TD-producing mergers where bothparents are early-type galaxies is less than 2%, suggesting that TDs are unlikely to formin such mergers. The bulk of the TD-producing systems inhabit a field environmentand have mass ratios greater than ∼ ∼ ∼ g − r ) colour and the TD colours are not affected by the presenceof AGN activity in their parents. An analysis of their star formation histories indicatesthat TDs contain both newly formed stars (with a median age of ∼
30 Myrs) and oldstars drawn from the parent disks, each component probably contributing roughlyequally to the stellar mass of the object. Thus TDs are not formed purely through gascondensation in tidal tails but host a significant component of old stars from the parentdisks. Finally, an analysis of the TD contribution to the observed dwarf to massivegalaxy ratio in the local Universe indicates that ∼
6% of dwarfs in nearby clusters mayhave a tidal origin, if TD production rates in nearby mergers are representative ofthose in the high-redshift Universe. Even if TD production rates at high redshift wereseveral factors higher, it seems unlikely that the entire dwarf galaxy population todayis a result of merger activity over the lifetime of the Universe.
Key words: galaxies: dwarf - galaxies: interactions - galaxies: starburst - galaxies:formation - galaxies: active
Galaxy mergers are a key driver of cosmological evo-lution, stimulating intense star formation episodes (e.g.Barnes & Hernquist 1992a), fuelling the growth of centralblack holes (e.g. Springel et al. 2005) and altering the mor-phological mix of the visible Universe (e.g. Toomre 1977;Steinmetz & Navarro 2002). While much of the literaturehas focussed on phenomena in the central regions of merg-ing systems, few studies have, until recently, studied the ⋆ E-mail: [email protected]; [email protected] impact of the merger process at large distances from theremnant. Up to a third of the pre-encounter material inthe merger progenitors is tidally ejected during the inter-action, into the tidal tails and bridges that form aroundthe remnant (e.g. Toomre 1977; Barnes & Hernquist 1992b;Duc & Mirabel 1999; Combes 1999; Springel & White 1999;Hibbard et al. 2005). This collisional debris, especially thataround gas-rich mergers, typically hosts star-forming re-gions, some of which may become progenitors of self-boundobjects with masses typical of dwarf galaxies (e.g. Zwicky1956; Schweizer 1978; Schombert et al. 1990; Mirabel et al.1991, 1992; Hibbard et al. 2005). In contrast to normal c (cid:13) Sugata Kaviraj dwarf galaxies, these ‘tidal dwarfs’ (TDs) are relativelymetal-rich, with metallicities typical of the outer regionsof spiral disks (e.g. Duc & Mirabel 1999; Duc et al. 2000;Weilbacher et al. 2000, 2003), free of (non-baryonic) darkmatter, since their potential wells are too shallow to cap-ture significant amounts of dark matter particles (e.g.Bournaud & Duc 2006; Duc et al. 2004, but see Gentile etal. 2007, Milgrom 2007) and may contribute a significantfraction of the nearby dwarf galaxy census (e.g. Kroupa1997; Hunsberger et al. 1996; Okazaki & Taniguchi 2000;Metz & Kroupa 2007).Two main mechanisms are postulated for TD for-mation. Jeans instabilities within the gas in the tidaltails can lead to gravitational collapse and the formationof self-bound objects (Elmegreen et al. 1993), akin toprocesses that produce giant molecular clouds. The Jeansmasses are typically high - as the gas is heated by themerger - enabling the formation of relatively massiveobjects, some of which share the properties of local dwarfgalaxies (e.g. Elmegreen et al. 1993; Struck et al. 2005;Bournaud & Duc 2006; Wetzstein et al. 2007; Smith et al.2008). Alternatively, a large fraction of the stellar materialin the progenitor disk may be ejected into the outerregions of the tidal tail, providing a local potential wellinto which gas condenses and fuels star formation (e.g.Barnes & Hernquist 1992b; Duc et al. 2004; Hancock et al.2009). In the first scenario the stellar component is likelyto be dominated by young stars, while in the second asubstantial fraction of the stellar material is expectedto be composed of old stars from the disks of the par-ent galaxies (see the recent review by Bournaud 2010).While a rich theoretical and observational literature hasdeveloped on the properties of nearby TDs (e.g. Wallin1990; Schombert et al. 1990; Hibbard et al. 1994; Duc et al.2000; Heithausen & Walter 2000; Braine et al. 2001;Hibbard et al. 2001; Temporin et al. 2003; Mundell et al.2004; Neff et al. 2005; Hancock et al. 2007; Recchi et al.2007; Bournaud et al. 2008; Werk et al. 2008; Boquien et al.2009; Sheen et al. 2009; Koribalski & L´opez-S´anchez 2009;Boquien et al. 2010; Wen et al. 2011), only relatively smallsamples of TDs have typically been exploited in any givenstudy. A large statistical study of TDs at low redshift isclearly desirable.An impediment to such a study is that a large, sta-tistically meaningful sample of galaxy mergers in the lo-cal Universe has so far been lacking. This is because,given the small merger fraction at low redshift (a few per-cent, see e.g. Abraham et al. 1996; Conselice et al. 2003;Lavery et al. 2004; De Propris et al. 2005; Conselice et al.2008; Darg et al. 2010a), a significant volume of the localUniverse must be observed in order to extract an adequatelylarge sample of merging systems. While the advent of mod-ern surveys, such as the
Sloan Digital Sky Survey (SDSS;York et al. 2000), has made such data available, the iden-tification of mergers remains a challenge, both due to theprodigious size of these datasets and the technical difficultyin identifying peculiar systems like galaxy mergers.Automated methods have often been employed to ex-tract mergers from survey data but most have some limi-tations. Galaxy ‘close pairs’ - which are likely progenitorsof mergers - can be identified in spectroscopic surveys (e.g.Patton et al. 2000; Le F`evre et al. 2000; Nikolic et al. 2004; Ellison et al. 2008; Rogers et al. 2009). However, close-pairstudies in the SDSS are likely to miss up to 80% of mergingsystems because fibre collisions prevent the SDSS from ob-taining spectra for two objects that are within 55 arcsecondsof each other in a single visit (see e.g. Darg et al. 2010a).Quantitative morphological parameters, e.g. Concentration,Asymmetry, Clumpiness, M and the Gini coefficient,have been extensively employed (often through the use ofneural networks) to classify galaxy morphologies in largesurveys (e.g. Abraham et al. 1996; Conselice et al. 2000;Abraham et al. 2003; Conselice et al. 2003; Ball et al. 2004;Lotz et al. 2004; Ferreras et al. 2005; Lahav et al. 1995;Lisker 2008; Andrae et al. 2011). However, it is difficult todefine a parameter space that is unique to mergers and theresults of such quantitative methods are typically checkedand calibrated against visual inspection (e.g. Abraham et al.1996; Ferreras et al. 2009; Jogee et al. 2009), which is ar-guably the most reliable method of morphological classifi-cation. The utility of visual inspection becomes particularlyimportant for identifying peculiar systems, such as ongoingmergers and post-mergers (e.g. Cassata et al. 2005; Kaviraj2010b). However, since it is prohibitively time-consumingfor large datasets, visual inspection of the SDSS has, untilthe advent of the Galaxy Zoo (GZ) project, been limited tovery small fractions (a few percent or less) of the full spec-troscopic galaxy sample in this survey (Fukugita et al. 2007;Schawinski et al. 2007; Nair & Abraham 2010).GZ is a citizen-science project which has used 250,000+volunteers from the general public to morphologically clas-sify the entire SDSS spectroscopic sample ( ∼ c (cid:13) , 000–000 idal dwarf galaxies in the nearby Universe Figure 1.
Examples of tidal dwarf candidates in the Galaxy Zoo mergers sample. Tidal dwarfs are selected as separate photometricobjects, identified by the SDSS pipeline, that are clearly associated with the tidal debris around the merger in question. The positionsof these objects are marked by the red crosses. This figure is available in colour in the online version of the journal. in Section 6, we explore whether the TD population couldmake a significant contribution to the dwarf galaxy censusin nearby clusters. We summarise our findings in Section 7.
TDs are identified through visual inspection of the co-added g, r, i
SDSS images of each merger. Separate photometricobjects, extracted by the SDSS pipeline, that are clearly as-sociated with the tidal debris in each merger are selected asTDs. In ∼
20% of the cases there are multiple photometricobjects associated with the same TD, with one object typi-cally containing more than 90% of the flux. In such cases wesum the fluxes of all the photometric objects to estimate theflux of the TD in question. We note that if we simply usedthe brightest photometric object in each of these cases thegeneral conclusions of our study would remain unaffected.This procedure yields 405 TDs. For each TD we also recordan approximate position in relation to their parent galaxy- at the tidal-tail tips, within the tail or at the base of thetidal tail.Figure 1 presents examples of TDs in our study. Theposition of the individual photometric objects, identified bythe SDSS pipeline, that are selected as TDs are indicatedon the images by red crosses. It should be noted that the identification of these objects relies on their association withthe tidal debris in the parent mergers. However, since theobjects are still ‘attached’ to their parents (which allowsus to identify them as potential TDs in the first place),we cannot be certain whether they will evolve into inde-pendent self-bound objects and eventually contribute to thedwarf galaxy population. Hence the objects identified hereare, strictly speaking,
TD candidates .The number of TDs per merger does not evolve acrossour redshift range (0 . < z < . c (cid:13) , 000–000 Sugata Kaviraj averaging procedure does not affect our results. The medianredshift of the TDs studied in this paper is z ∼ . KCORRECT code (Blanton et al.2003a; Blanton & Roweis 2007). The absolute magnitudesare used to estimate stellar masses, using the calibrations ofBell et al. (2003). The error on these masses can be up to0.3 dex. Figure 2 presents distributions of the basic proper-ties (redshift, absolute r -band magnitude and stellar mass)of the TD sample in this paper. Median values are indi-cated using the dashed vertical lines. Note that, in the bot-tom panel (stellar mass), three additional vertical lines areshown, which indicate median values for TDs at the tips oftidal tails (red), within the tails (green) and at the base oftails (blue). We return to these TD subsets in Section 3.3below. We begin by cataloguing the properties of TD-producingmergers and compare them to the general merger popula-tion. 95% of binary mergers that produce TDs involve twospiral progenitors, while most remaining systems have atleast one spiral progenitor. The fraction of TD-producingmergers where both parent galaxies have early-type mor-phology is less than 2% (at least in the sample studied here),strongly suggesting that TDs are unlikely to form in suchmergers. It is instructive to check whether the significantlack of TDs in early-type - early-type (E-E) mergers is a realeffect or whether they are not identified (partly) because thetidal tails in these mergers are too faint to be clearly detectedin the standard SDSS imaging. By virtue of being a large-area survey, the standard SDSS imaging is relatively shallow,with only ∼
54 second exposures in every filter (the r -banddetection limit is ∼
22 mag). To probe this issue further weexplore the images of E-E mergers in this sample that lie inthe SDSS Stripe 82 ( − ◦ < α < ◦ , − . ◦ < δ < . ◦ )that offers 2 mag deeper imaging than the standard SDSSsurvey. The Stripe 82 has been imaged multiple times aspart of the SDSS Supernova Survey (Frieman et al. 2008)and achieves limiting magnitudes of ∼
24 mag in r -band,sometimes revealing faint tidal debris in mergers that maybe invisible in the standard imaging (Kaviraj 2010a). Sinceit has an area of 270 deg , compared to the 9583 deg area ofthe DR6 from which the Darg et al. sample is constructed,only 9 E-E mergers in the Darg et al. study lie in this re-gion of the sky. However, visual inspection of these imagesdo not reveal any TDs not identified in the standard images,leaving our conclusions above unchanged.Taking ( g − r ) ∼ . The total number of mergers in the Darg et al. sample is 3373,with an E-E fraction of around 12%. The area of the Stripe 82is 3% of the DR6 (270 deg /9583 deg ). Thus we expect (270 × × . / ∼
11) E-E mergers in the Stripe 82. The actualnumber is 9 (consistent within counting errors).
Figure 2.
Distributions of redshift (top), r -band absolute mag-nitude (middle) and stellar mass (bottom) for the TD sample inthis study. The stellar masses are calculated using the calibrationsof Bell et al. (2003) and have errors of up to ∼ ∼ − ). Magnitude errors are taken from the SDSS DR6database. This figure is available in colour in the online versionof the journal. c (cid:13) , 000–000 idal dwarf galaxies in the nearby Universe Figure 3.
Parent mass ratios of TD-producing mergers (solidline) compared to the general merger population (dotted line).95% of TD-producing mergers have mass ratios greater than ∼ ∼ Blanton et al. 2003b), we find that in 85% of the parentmergers both progenitors are blue. In 12% at least one pro-genitor is blue, while in the remaining 3% of parent systemsboth progenitors are on the red sequence. Not unexpectedlyTD formation becomes significantly more likely when bothmerger progenitors are gas-rich (and therefore in the bluecloud).
Figure 3 indicates that 95% of TDs are produced by parentsystems whose constituent galaxies have mass ratios greaterthan ∼ ∼ ∼ In Figure 4 we show both the physical separation of TDsfrom their parents (left panel) and the separation normalisedby the half-light radii ( R / ) of the parent galaxies. 95% ofTDs are within ∼
20 kpc of their parent galaxies, correspond-ing to ∼ R / (the median normalised separation is ∼ ∼ R / ), generally consistent with the theoreticalsimulations of Bournaud & Duc (2006).As indicated in the bottom panel of Figure 2 above,TDs that lie further along the tidal tail appear to be moremassive. The offset in the median masses (indicated by thedashed lines) of TDs born at the tips of the tidal tails com-pared to those born at the base of the tails is ∼ We explore the local environment of TD-producing merg-ers by cross-matching with the SDSS environment catalogueof Yang et al. (2007, 2008), who use a halo-based groupfinder to separate the SDSS into 300,000+ structures, span-ning rich clusters to the field. The catalogue provides esti-mates for the masses of the host dark matter haloes of in-dividual SDSS galaxies, which are related to the traditionalclassifications of environment (field/group/cluster). Haloeswith masses greater than 10 M ⊙ represent clusters, whilethose with masses in the range 10 M ⊙ to 10 M ⊙ repre-sent groups. Smaller DM haloes represent the field. Figure5 indicates that TD-producing mergers favour lower-densityenvironments than the general merger population. ∼
90% ofTD-producing mergers reside in the field, with the remain-ing systems inhabiting groups. Almost none of the systemsreside in clusters. This result is consistent with the fact thatthe availability of cold gas is a strong function of local envi-ronment. A cluster environment, in particular, is expectedto be cold-gas-poor (e.g. Solanes et al. 2001) and thereforehostile to TD formation.
We proceed by comparing how TD properties compare tothose of their parent mergers. Figure 6 indicates that thestellar masses of 95% of TDs are less than 10% of the stel-lar mass of their parent mergers. The median TD-to-parentstellar mass ratio is around 0.6% (shown using the dashedline in Figure 6). Note that, since the masses are calcu-lated from the photometric data, they correspond only tothe stellar component of the galaxy. While the dynamical(total) masses of the TDs are expected to be similar totheir stellar masses (since they do not contain significantamounts of dark matter), this is not the case for the par-ent spiral galaxies, which may contain 3-5 times as muchdark as luminous matter (e.g. van Albada & Sancisi 1986;Ashman 1992; Salucci & Burkert 2000; Noordermeer et al.2007; Salucci & Frigerio Martins 2009) inside ∼
10 diskscalelengths (typically 20-40 kpc). The total
TD-to-parentmass ratios are therefore likely to be several factors smallerthan the stellar values derived here.We now compare the TD colours to those of their par-ent galaxies. Recent studies have suggested that the pres-ence of an AGN in a galaxy can affect the colours of ob-jects in their immediate vicinity (Shabala et al. 2011), plau-sibly due to interaction between AGN-driven outflows andthe gas reservoirs of these nearby galaxies. It is conceiv-able, therefore, that TD formation might also be affectedby outflows due to nuclear activity in their parents. Thismay either suppress star formation by removing gas fromthe star-forming regions, as is typically envisaged in nega-tive feedback scenarios (e.g. Silk & Rees 1998; Croton et al. c (cid:13)000
TD-to-parentmass ratios are therefore likely to be several factors smallerthan the stellar values derived here.We now compare the TD colours to those of their par-ent galaxies. Recent studies have suggested that the pres-ence of an AGN in a galaxy can affect the colours of ob-jects in their immediate vicinity (Shabala et al. 2011), plau-sibly due to interaction between AGN-driven outflows andthe gas reservoirs of these nearby galaxies. It is conceiv-able, therefore, that TD formation might also be affectedby outflows due to nuclear activity in their parents. Thismay either suppress star formation by removing gas fromthe star-forming regions, as is typically envisaged in nega-tive feedback scenarios (e.g. Silk & Rees 1998; Croton et al. c (cid:13)000 , 000–000 Sugata Kaviraj
Figure 4. LEFT:
Physical separation of TDs from their parent galaxies.
RIGHT:
Physical separation normalised by the half-light radius ( R / ) of the parents. Median values are indicated using dashed vertical lines. The median separation is ∼
17 kpc andthe median normalised separation is ∼ . R / . 95% of TDs are within ∼ R / of their parent galaxies. We assume that theuncertainties in RA and DEC values are negligible (hence no error bar is shown on the separations). The error in the normalisedseparation (right-hand panel) is driven by the error in the half-light radii of the parent galaxies. GANDALF code (Sarzi et al. 2006) . We assume that galaxies classifiedas composite, Seyfert or LINER host AGN.Figure 7 shows that TDs are typically bluer than theirparents. The median ( g − r ) colour offset between TDs andparents is ∼ We investigate the star formation histories (SFHs) of TDs,in particular the relative fraction of stellar mass that is com- ∼ sarzi/PaperV nutshell/PaperV nutshell.htmlfor more details. posed of new stars compared to the fraction that is consti-tuted by old stars from the parent disks. We estimate theSFH of each TD by comparing colours constructed from theSDSS ( u, g, r, i, z ) magnitudes to a library of synthetic pho-tometry that is based on model SFHs designed to approxi-mate the stellar content of each TD.Each model SFH is constructed using two instantaneousstarbursts. The first burst, which characterises the old, un-derlying stellar population in the parent disks, is assumedto have an age of 7 Gyr, which represents an average age forthe old disk stars. Recent studies that have decoupled therecent star formation from the old, underlying populationsin star-forming spirals suggest average ages for the old starsaround this value (see Kaviraj et al. 2009). We have checkedthat our conclusions remain unaffected if we change the ageof the old stars to 10 Gyrs.The second burst, which represents the young stellarcontent of the TDs, is allowed to vary in (i) age between1 Myr and 7 Gyr and (ii) mass fraction between 0 and1. We also include a range of values for the internal dustextinction, parametrised in terms of E B − V from 0 to 1.The dust extinction is applied using the empirical law ofCalzetti et al. (2000) to the SFH as a whole. We assume thatthe model SFHs have a metallicity of 0.3 Z ⊙ , which is typicalof the outer regions of spiral disks (e.g. Duc & Mirabel 1998;Weilbacher et al. 2000). The free parameters are, therefore,the age ( t ) and mass fraction ( f ) of the second burst andthe dust content ( E B − V ) of the TD. The model SFH librarycontains 1.5 million individual models. To build a libraryof synthetic photometry, each combination of free parame-ters is combined with the stellar models of Yi (2003) andconvolved with the correct SDSS filtercurves. Since our TDsample spans a range in redshift (0 . < z < . δz = 0 . u, g, r, i, z ) colours to everymodel in the synthetic library. In a Bayesian framework (see c (cid:13) , 000–000 idal dwarf galaxies in the nearby Universe Figure 5.
Local environments of TD-producing mergers (solidline) compared to the general merger populations (dotted line).TD-producing mergers typically inhabit field environments. Notethat the Yang et al. (2007) catalogue, from which the environ-ment measures are extracted, does not provide any error infor-mation on the host dark-matter halo masses.
Figure 6.
Ratio of TD stellar mass to parent stellar mass. 95%of TDs have stellar masses that are less than ∼
10% of the stellarmasses of their parent galaxies. The median stellar mass ratio is ∼ e.g. Sivia & Skilling 2006), for a vector X denoting param-eters in the model and a vector D denoting the measuredobservables (in this case the colours),prob( X | D ) ∝ prob( D | X ) × prob( X ) , (1)where prob( X | D ) is the probability of the model given thedata (which is the quantity of interest), prob( D | X ) is theprobability of the data given the model and prob( X ) is theprior probability distribution of the model parameters. Sincewe assume a flat prior in all our model parameters above,prob( X ) = constant so thatprob( X | D ) ∝ prob( D | X ) . (2) Figure 7.
TD ( g − r ) colour vs. parent ( g − r ) colour. We showparents with and without AGN using the vertical lines. AGN areidentified using an optical emission-line-ratio analysis (see text inSection 4 for details). Median values are indicated using verticallines. Assuming gaussian errors implies thatprob( D | X ) ∝ exp( − χ / , (3)where exp( − χ /
2) is the likelihood function, with χ de-fined in the standard way, as the sum of the normalizedresiduals between the model-predicted observables M i andthe observed values D i i.e. χ = N X i =1 (cid:16) M i − D i σ i (cid:17) (4)The error that enters into the χ equation ( σ i ) is com-puted by adding, in quadrature, the observational uncertain-ties with the errors adopted for the stellar models, whichwe assume to be 0.05 mag in each optical filter (Yi 2003).prob( X | D ) is a joint probability distribution, dependenton all the model parameters. From this joint distribution,each free parameter is marginalised to extract its one-dimensional probability distribution. We take the medianvalue of this distribution as the best estimate of the param-eter in question. The 25th and 75th quartile values (whichencompass 50% of the probability) provide an estimate ofthe uncertainty in the parameter. This yields, for every TD,a best estimate and error for each free parameter. It is worthnoting that the derived error represents the combined un-certainty in the parameter estimate due to the observationaland model errors and the various degeneracies between thefree parameters.Figure 8 presents the distribution of free parametersfor our TD population. Not unexpectedly, and in agreementwith the wider literature, we find that a substantial youngstellar component exists in the TDs, with ages less than To isolate the effect of a single parameter X1 in, for example, atwo-parameter model [prob( X | D ) ≡ prob( X , X | D )] we can in-tegrate out the effect of X2 to obtain the marginalized probabilitydistribution for X1: prob( X | D ) = R ∞ prob( X , X | D ) dX .c (cid:13) , 000–000 Sugata Kaviraj
Figure 8.
Estimated values of the free parameters that drive theTD star formation histories: age ( t ; top) and mass fraction ( f ;middle) of the young stars and internal extinction in the galaxy( E B − V ; bottom). Median values of the distributions are shownusing the dashed lines. While a substantial young stellar compo-nent exists in TDs, with ages less than ∼ ∼
30 Myr), an equally significant component, drawn fromold stars in the parent disks, also appears to be present in thesesystems. ∼ ∼
30 Myr). The de-rived mass fractions in young stars largely range between20 and 80% with a median value of ∼ E B − V ∼ . A V ∼ < . ∼ ± the bulk of the TDs are inconsistentwith a purely young stellar population . It is likely, therefore,that TDs are not formed purely through gas condensationsin the tidal tails but that their potential wells contain sig-nificant contributions from pre-existing stellar material fromthe parent disks. We conclude our analysis by investigating the potential con-tribution of TDs to the dwarf galaxy population observedin the Universe today. The analysis in Section 3 indicatesthat TDs typically form in gas-rich or wet major mergersthat involve two spiral galaxies. If the number of wet ma-jor mergers experienced by a massive galaxy over a Hubbletime is N wet , the average number of TDs produced per wetmajor merger is N TD and the fraction of TDs that survivefor a Hubble time is S, then the number of TDs expectedper massive galaxy today is estimated to be N wet × N TD × S. (5)Integration of the empirical major merger rate in mas-sive galaxies over time indicates that every massive galaxytypically experiences ∼ z ∼ z > ∼
50% of TDs with masses greater than M ⊙ are likely to survive for a Hub-ble time (Bournaud 2010). TD-producing mergers each cre-ate, on average, 1.2 TDs in this mass range. However, only ∼
18% of major gas-rich mergers produce such TDs in thefirst place. Hence, the average number of TDs with massesgreater than 10 M ⊙ per gas-rich major merger is ∼ × × . × . . observed galaxy mass function indicates thatdwarf galaxies are the dominant galaxy type in the lo-cal Universe (e.g. Sandage et al. 1985; van den Bergh 1992;Sabatini et al. 2003). The ratio of dwarf to massive galax-ies (D/M) in Coma (Secker & Harris 1996), restricted todwarfs with masses greater than ∼ M ⊙ ( M ( r ) < − . ∼ c (cid:13) , 000–000 idal dwarf galaxies in the nearby Universe ratio is ∼ ∼
6% of the dwarfpopulation in clusters could plausibly have a tidal origin.It should be noted that this estimate assumes that theTD production rate in high-redshift mergers is similar tothat in their nearby counterparts. Mergers at high redshifttypically involve higher gas masses (e.g. Daddi et al. 2010;Tacconi et al. 2010) and may yield more TDs than their lo-cal counterparts (e.g. Wetzstein et al. 2007). However, sim-ulations of high-redshift major mergers (Bournaud et al.2011), in which the interstellar medium is more clumpyand turbulent than in their nearby counterparts (e.g.Elmegreen et al. 2009), suggest that these interactions donot produce the long tidal tails seen in local mergers. Thismay have implications for the lifetime of tidal objects, sincethey are formed closer to their parent galaxies, making themmore vulnerable to disruption. Definitive studies of the TDproduction rate at high redshift requires both further simu-lation work and empirical studies of high-redshift mergers atthe peak epoch of stellar mass assembly (2 < z <
4, see e.g.Madau et al. 1998; Hopkins 2004; Hopkins & Beacom 2006)using high-resolution data e.g. from the Wide Field Camera3 (WFC3) or the Extremely Large Telescopes (ELTs). Nev-ertheless, it is worth noting that even if TD production rateswere several factors higher in the early Universe, it remainsunlikely that the entire local dwarf galaxy census has a tidalorigin.
We have performed a statistical observational study of theTD population in the local Universe, by exploiting a large,homogeneous sample of galaxy mergers compiled from theSDSS DR6 using the Galaxy Zoo project. The aim of thiswork has been to explore the statistical properties of lo-cal TDs, both to complement existing observational studies(which are typically based on relatively small samples ofmergers) and as a comparison to the wide body of theoreti-cal work that has recently been performed on the formationand evolution of TDs.Our results indicate that 95% of TD-producing mergersinvolve interactions between two spiral galaxies, both typi-cally residing in the blue cloud. The overwhelming majorityof these parent systems have mass ratios greater than ∼ ∼ g − r ). The presence of anAGN in the parent galaxies does not affect the TD colours.It is worth noting that only around a fifth of gas-rich majormergers produce massive TDs (with masses greater than 10 M ⊙ ).An analysis of their star formation histories indicatesthat TDs contain both newly formed stars and old stellarmaterial drawn from the disk of their parent galaxies. Theyoung stellar components have ages less than ∼ ∼
30 Myr in the TD population as awhole. The young components contribute stellar mass frac-tions between 20 and 80%, with a typical value of ∼ E B − V ∼ . A V ∼ < . ∼
6% of the dwarfsin local clusters may be of tidal origin, assuming that theTD production rate in the nearby Universe is representativeof that in high-redshift mergers. Observational studies ofTDs in high-redshift mergers, using forthcoming data fromthe WFC3 and the ELTs, are keenly anticipated to furtherexplore the role of mergers in the formation of the dwarfgalaxy population at the present day.
ACKNOWLEDGEMENTS c (cid:13) , 000–000 Sugata Kaviraj
University, the United States Naval Observatory, and theUniversity of Washington.
REFERENCES
Abraham R. G., Tanvir N. R., Santiago B. X., Ellis R. S.,Glazebrook K., van den Bergh S., 1996, MNRAS, 279, L47Abraham R. G., van den Bergh S., Nair P., 2003, ApJ, 588,218Adelman-McCarthy J. K., et al. 2008, ApJS, 175, 297Andrae R., Jahnke K., Melchior P., 2011, MNRAS, 411,385Ashman K. M., 1992, PASP, 104, 1109Baldwin J. A., Phillips M. M., Terlevich R., 1981, PASP,93, 5Ball N. M., Loveday J., Fukugita M., Nakamura O., Oka-mura S., Brinkmann J., Brunner R. J., 2004, MNRAS,348, 1038Barnes J. E., Hernquist L., 1992a, ARAA, 30, 705Barnes J. E., Hernquist L., 1992b, Nature, 360, 715Bell E. F., McIntosh D. H., Katz N., Weinberg M. D., 2003,ApJS, 149, 289Bell E. F., Naab T., McIntosh D. H., et al. 2006, ApJ, 640,241Blanton M. R., et al. 2003a, AJ, 125, 2348Blanton M. R., et al. 2003b, ApJ, 594, 186Blanton M. R., Roweis S., 2007, AJ, 133, 734Boquien M., Duc P., Galliano F., Braine J., Lisenfeld U.,Charmandaris V., Appleton P. N., 2010, AJ, 140, 2124Boquien M., Duc P., Wu Y., Charmandaris V., LisenfeldU., Braine J., Brinks E., Iglesias-P´aramo J., Xu C. K.,2009, AJ, 137, 4561Bournaud F., et al. 2011, ApJ, 730, 4Bournaud F., 2010, Advances in Astronomy, 2010Bournaud F., Bois M., Emsellem E., Duc P., 2008, As-tronomische Nachrichten, 329, 1025Bournaud F., Duc P., 2006, A&A, 456, 481Bournaud F., Duc P.-A., Amram P., Combes F., Gach J.-L., 2004, A&A, 425, 813Braine J., Duc P., Lisenfeld U., Charmandaris V., VallejoO., Leon S., Brinks E., 2001, A&A, 378, 51Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koorn-neef J., Storchi-Bergmann T., 2000, ApJ, 533, 682Cassata P., Cimatti A., Franceschini A., et al. 2005, MN-RAS, 357, 903Combes F., 1999, in J. E. Barnes & D. B. Sanders ed.,Galaxy Interactions at Low and High Redshift Vol. 186 ofIAU Symposium, Extended Gas in Interacting Systems.pp 89–+Conselice C. J., 2007, in Combes F., Palous J., eds, IAUSymposium Vol. 235 of IAU Symposium. pp 381–384Conselice C. J., Bershady M. A., Dickinson M., PapovichC., 2003, AJ, 126, 1183Conselice C. J., Bershady M. A., Jangren A., 2000, ApJ,529, 886Conselice C. J., Rajgor S., Myers R., 2008, MNRAS, 386,909Croton D. J., et al. 2006, MNRAS, 365, 11Daddi E., et al. 2010, ApJ, 713, 686Darg D. W., et al. 2010a, MNRAS, 401, 1043Darg D. W., et al. 2010b, MNRAS, 401, 1552 De Propris R., Liske J., Driver S. P., Allen P. D., CrossN. J. G., 2005, AJ, 130, 1516Duc P., Bournaud F., Masset F., 2004, A&A, 427, 803Duc P., Brinks E., Springel V., Pichardo B., Weilbacher P.,Mirabel I. F., 2000, AJ, 120, 1238Duc P., Mirabel I. F., 1999, in J. E. Barnes & D. B. Sandersed., Galaxy Interactions at Low and High RedshiftVol. 186 of IAU Symposium, Tidal Dwarf Galaxies. pp61–+Duc P.-A., Mirabel I. F., 1998, A&A, 333, 813Ellison S. L., Patton D. R., Simard L., McConnachie A. W.,2008, AJ, 135, 1877Elmegreen B. G., Elmegreen D. M., Fernandez M. X.,Lemonias J. J., 2009, ApJ, 692, 12Elmegreen B. G., Kaufman M., Thomasson M., 1993, ApJ,412, 90Ferguson H. C., Sandage A., 1991, AJ, 101, 765Ferreras I., Lisker T., Carollo C. M., Lilly S. J., MobasherB., 2005, ApJ, 635, 243Ferreras I., Lisker T., Pasquali A., Kaviraj S., 2009, MN-RAS, 395, 554Frieman J. A., et al. 2008, AJ, 135, 338Fukugita M., et al. 2007, AJ, 134, 579Gentile G., Famaey B., Combes F., Kroupa P., Zhao H. S.,Tiret O., 2007, A&A, 472, L25Hancock M., Smith B. J., Struck C., Giroux M. L., Apple-ton P. N., Charmandaris V., Reach W. T., 2007, AJ, 133,676Hancock M., Smith B. J., Struck C., Giroux M. L., HurlockS., 2009, AJ, 137, 4643Heithausen A., Walter F., 2000, A&A, 361, 500Hibbard J. E., et al. 2005, ApJL, 619, L87Hibbard J. E., Guhathakurta P., van Gorkom J. H.,Schweizer F., 1994, AJ, 107, 67Hibbard J. E., van der Hulst J. M., Barnes J. E., RichR. M., 2001, AJ, 122, 2969Hopkins A. M., 2004, ApJ, 615, 209Hopkins A. M., Beacom J. F., 2006, ApJ, 651, 142Hunsberger S. D., Charlton J. C., Zaritsky D., 1996, ApJ,462, 50Jogee S., et al. 2009, ApJ, 697, 1971Kauffmann G., Heckman T. M., White S. D. M., et al.2003, MNRAS, 346, 1055Kaviraj S., 2010a, MNRAS, 406, 382Kaviraj S., 2010b, MNRAS, 408, 170Kaviraj S., Peirani S., Khochfar S., Silk J., Kay S., 2009,MNRAS, 394, 1713Kaviraj S., Schawinski K., Silk J., Shabala S. S., 2010,arXiv:1008.1583Kewley L. J., Groves B., Kauffmann G., Heckman T., 2006,MNRAS, 372, 961Khochfar S., Burkert A., 2003, ApJ, 597, L117Koribalski B. S., L´opez-S´anchez ´A. R., 2009, MNRAS, 400,1749Kroupa P., 1997, New Astronomy, 2, 139Lahav O., et al. 1995, Science, 267, 859Lavery R. J., Remijan A., Charmandaris V., Hayes R. D.,Ring A. A., 2004, ApJ, 612, 679Le F`evre O., Abraham R., Lilly S. J., et al. 2000, MNRAS,311, 565Lin L., Koo D. C., Willmer C. N. A., et al. 2004, ApJ, 617,L9 c (cid:13) , 000–000 idal dwarf galaxies in the nearby Universe Lintott C., et al. 2011, MNRAS, 410, 166Lintott C. J., et al. 2008, MNRAS, 389, 1179Lisker T., 2008, ApJS, 179, 319Lotz J. M., Madau P., Giavalisco M., Primack J., FergusonH. C., 2006, ApJ, 636, 592Lotz J. M., Primack J., Madau P., 2004, AJ, 128, 163Madau P., Pozzetti L., Dickinson M., 1998, ApJ, 498, 106Metz M., Kroupa P., 2007, MNRAS, 376, 387Milgrom M., 2007, ApJL, 667, L45Mirabel I. F., Dottori H., Lutz D., 1992, A&A, 256, L19Mirabel I. F., Lutz D., Maza J., 1991, A&A, 243, 367Mould J. R., et al. 2000, ApJ, 536, 266Mundell C. G., James P. A., Loiseau N., Schinnerer E.,Forbes D. A., 2004, ApJ, 614, 648Nair P. B., Abraham R. G., 2010, ApJS, 186, 427Neff S. G., et al. 2005, ApJL, 619, L91Nikolic B., Cullen H., Alexander P., 2004, MNRAS, 355,874Noordermeer E., van der Hulst J. M., Sancisi R., SwatersR. S., van Albada T. S., 2007, MNRAS, 376, 1513Okazaki T., Taniguchi Y., 2000, ApJ, 543, 149Patton D. R., Carlberg R. G., Marzke R. O., Pritchet C. J.,da Costa L. N., Pellegrini P. S., 2000, ApJ, 536, 153Recchi S., Theis C., Kroupa P., Hensler G., 2007, A&A,470, L5Rogers B., Ferreras I., Kaviraj S., Pasquali A., Sarzi M.,2009, MNRAS, 399, 2172Sabatini S., Davies J., Scaramella R., Smith R., Baes M.,Linder S. M., Roberts S., Testa V., 2003, MNRAS, 341,981Salucci P., Burkert A., 2000, ApJL, 537, L9Salucci P., Frigerio Martins C., 2009, in E. P´econtal,T. Buchert, P. di Stefano, & Y. Copin ed., EAS Pub-lications Series Vol. 36 of EAS Publications Series, Themass distribution in Spiral galaxies. pp 133–140Sandage A., Binggeli B., Tammann G. A., 1985, AJ, 90,1759Sarzi M., Falc´on-Barroso J., Davies R. L., et al. 2006, MN-RAS, 366, 1151Schawinski K., Kaviraj S., Khochfar S., et al. 2007, ApJS,173, 512Schombert J. M., Wallin J. F., Struck-Marcell C., 1990,AJ, 99, 497Schweizer F., 1978, in E. M. Berkhuijsen & R. Wielebinskied., Structure and Properties of Nearby Galaxies Vol. 77of IAU Symposium, Galaxies with long tails. pp 279–284Secker J., Harris W. E., 1996, ApJ, 469, 623Shabala S. S., Kaviraj S., Silk J., 2011, MNRAS, pp 336–+Sheen Y., et al. 2009, AJ, 138, 1911Silk J., 2005, MNRAS, 364, 1337Silk J., Rees M. J., 1998, A&A, 331, L1Sivia D. S., Skilling J., 2006, Data Analysis–A BayesianTutorial, 2nd edn. Oxford Science PublicationsSmith B. J., Struck C., Hancock M., Giroux M. L., Ap-pleton P. N., Charmandaris V., Reach W., Hurlock S.,Hwang J., 2008, AJ, 135, 2406Solanes J. M., Manrique A., Garc´ıa-G´omez C., Gonz´alez-Casado G., Giovanelli R., Haynes M. P., 2001, ApJ, 548,97Springel V., Di Matteo T., Hernquist L., 2005, MNRAS,361, 776Springel V., White S. D. M., 1999, MNRAS, 307, 162 Steinmetz M., Navarro J. F., 2002, Nature, 7, 155Strateva I., et al. 2001, AJ, 122, 1861Struck C., Kaufman M., Brinks E., Thomasson M.,Elmegreen B. G., Elmegreen D. M., 2005, MNRAS, 364,69Tacconi L. J., et al. 2010, Nature, 463, 781Temporin S., Weinberger R., Galaz G., Kerber F., 2003,ApJ, 587, 660Toomre A., 1977, in Tinsley B. M., Larson R. B., eds, Evo-lution of Galaxies and Stellar Populations pp 401–+Tortora C., Antonuccio-Delogu V., Kaviraj S., Silk J.,Romeo A. D., Becciani U., 2009, MNRAS, 396, 61van Albada T. S., Sancisi R., 1986, Royal Society of LondonPhilosophical Transactions Series A, 320, 447van Breugel W., Filippenko A. V., Heckman T., Miley G.,1985, ApJ, 293, 83van den Bergh S., 1992, A&A, 264, 75Veilleux S., Osterbrock D. E., 1987, ApJS, 63, 295Wallin J. F., 1990, AJ, 100, 1477Weilbacher P. M., Duc P., Fritze-v. Alvensleben U., 2003,A&A, 397, 545Weilbacher P. M., Duc P., Fritze v. Alvensleben U., MartinP., Fricke K. J., 2000, A&A, 358, 819Wen Z., Zheng X., Zhao Y., Gao Y., 2011, ArXiv e-printsWerk J. K., Putman M. E., Meurer G. R., Oey M. S., Ryan-Weber E. V., Kennicutt Jr. R. C., Freeman K. C., 2008,ApJ, 678, 888Wetzstein M., Naab T., Burkert A., 2007, MNRAS, 375,805Yang X., Mo H. J., van den Bosch F. C., 2008, ApJ, 676,248Yang X., Mo H. J., van den Bosch F. C., Pasquali A., LiC., Barden M., 2007, ApJ, 671, 153Yi S. K., 2003, ApJ, 582, 202York D. G., et al. 2000, AJ, 120, 1579Zwicky F., 1956, Ergebnisse der exakten Naturwis-senschaften, 29, 344 c (cid:13)000