The Role of Galaxy Interactions and Mergers in Star Formation at z<1.3: Mid-Infrared Properties in the Spitzer First Look Survey

0.99 MB·English

by  C. R. Bridge

Accepted by ApJ
PreprinttypesetusingLATEXstyleemulateapjv.10/09/06
THE ROLE OF GALAXY INTERACTIONS AND MERGERS IN STAR FORMATION AT Z ≤1.3:
MID-INFRARED PROPERTIES IN THE SPITZER FIRST LOOK SURVEY 1
C.R. Bridge,2 P.N. Appleton,3 C.J. Conselice,4 P.I Choi,5 L. Armus,3 D. Fadda,3 S. Laine,3 F.R. Marleau,3 R.G.
Carlberg,2 G. Helou,3 L. Yan3
Accepted by ApJ
ABSTRACT
Bycombiningthe0.12squaredegreeF814WHubbleSpaceTelescope (HST)andSpitzer MIPS24µm
imagingintheFirstLookSurvey(FLS),weinvestigatethepropertiesofinteractingandmergingMid-
Infraredbright and faint sources at 0.2≤z≤1.3. We find a marginallysignificant increase in the pair
7 fraction for MIPS 24µm detected, optically selected close pairs, pair fraction= 0.25±0.10 at z∼1,
0 in contrast to 0.11±0.08 at z ∼0.4, while galaxies below our 24µm MIPS detection limit show a
0 pair fraction consistent with zero at all redshifts. Additionally, 24µm detected galaxies with fluxes
2 ≥0.1mJy areonaveragefive times more likely to be in a close galaxypairbetween 0.2≤z≤1.3 than

n galaxies below this flux limit. Using the 24µm flux to derive the total Far-IR luminosity we find that
a pairedgalaxies(earlystagemergers)areresponsiblefor27%±9%oftheIRluminositydensityresulting
J fromstarformationatz ∼1whilemorphologicallyclassified(latestage)mergersmakeup34%±11%.

3 Thisimpliesthat61%±14%oftheinfraredluminositydensityandinturn∼40%ofthestarformation
ratedensityatz ∼1canbeattributedtogalaxiesatsomestageofamajormergerorinteraction. We

1 arguethat, close pairs/mergersin a LIRG/ULIRG phase become increasinglyimportant contributers
v to the IR luminosity and star formation rate density of the Universe at z >0.7.
0
Subjectheadings:galaxies: evolution–galaxies: formation–galaxies: interactions–galaxies: starburst
4
0
1 1. INTRODUCTION closepairstudieshavesuggestedthatthemergerratehas
0
remained fairly constant from z ∼1 (Bundy et al. 2004;
7 Hierarchical models and observations suggest that
Lin et al.2004),andatz ≥0.7theIRpopulationisdom-
0 galaxymergersandinteractionsplayakeyroleingalaxy
inated by morphologically normal galaxies (Bell et al.
/ assembly and star formation, but to what extent is still
h 2005;Melbourne et al.2005;Lotz et al.2006). Thecom-
unclear. Studies of gas-rich mergers in the local uni-
p binationofthesetworesultssuggestthatthebulkofstar
verse (e.g., Antennae; see Schweizer 1982) and N-body
- formation at z ∼1 is not driven by major mergers.
o simulations(Mihos & Hernquist1996;Barnes2004)have
However it must be noted that different merger selec-
r revealed fundamental signatures of the galaxy merger
t tion criteria probe different stages of the merger pro-
s process, including tidal tails, multiple nuclei, and vio-
a lent bursts of star formation. While interaction-induced cess. Quantitative measurements of galaxy asymmetry
v: star formation is thought to be primarily responsible (Abraham et al. (1996a,b); Conselice et al. (2003)) are
morelikelytoprobelaterstages,while earlystagemerg-
i for ultra luminous infrared galaxies (ULIRGs, which
X have L ≥ 1012L ) both locally and at high redshift erscanbeidentifiedbycarefullysearchingforclosecom-
IR ⊙
panions. There should be some overlap between these
r (Sanders et al. 1988; Dasyra et al. 2006), luminous in-
a fraredgalaxies(LIRGs,L ∼1011−1012L )appearto techniques if galaxy pairs are close enough to have in-
IR ⊙
duced strong tidal interactions, but galaxies in pairs
have multiple driving mechanisms, merger-induced star
could also have normal morphologies, hence if early
formation being only one.
stage mergers are not considered, the impact interac-
Luminous infrared (IR) galaxiesare thought to be the
tions/merging have will be underestimated.
dominant producers of the cosmic infrared background
Traditionally, close pair studies have been carried out
(CIRB), and major contributors to the evolution of the
in the optical/near-IR (Patton et al. 1997, 2000, 2002;
cosmicstarformationrate(CSFR)ofgalaxies,especially
Carlberg et al. 2000; Le F`evre et al. 2000; Lin et al.
atz ≥0.7(Elbaz et al.2002;Le Floc’h et al.2005). The
2004; Bundy et al. 2004). However recent investigations
rapid decline from z ∼ 1 of the CSFR density has been
have begun to explore the Mid-IR properties (star for-
linked to a decline in the merger rate. However, recent
mation) of galaxy pairs, finding a Mid-IR enhancement
Electronicaddress: [email protected] in pairs separated by less then ten’s of kpc’s (Lin et al.
1SomeofthedatapresentedhereinwereobtainedattheW.M. 2006). The amount of IR luminosity stemming from in-
Keck Observatory, which is operated as a scientific partnership dividualprocesses(starformationorfueling anAGN)in
among the California Institute of Technology, the University of
interacting pairs and mergers still remains open. To in-
California and the National Aeronautics and Space Administra-
tion. TheObservatorywasmadepossiblebythegenerousfinancial vestigatethis questionwe haveconducted a study of the
supportoftheW.M.KeckFoundation. frequency of MIPS24µmdetected, andundetected close
2University of Toronto, 50 St. George Street, Toronto, ON, opticalgalaxypairsandmorphologicallydefinedmergers
Canada,M5S3H4
3Spitzer Science Center, California Institute of Technology, intheSpitzer FirstLookSurvey(FLS)6.Wefindthatthe
Pasadena,CA91125
4UniversityofNottingham,UniversityPark,NG92RD,United 6 For details of the FLS observation plan and the data release,
Kingdom seehttp://ssc.spitzer.caltech.edu/fls.
5DepartmentofPhysicsandAstronomy,PomonaCollege,CA
2 Bridge et al.
fraction of 24µm detected, optically selected close pairs sample included sources with K <20.2, R >19.0 and a
s
and mergers increases with redshift, and are important g,R,i color selection that restricted the numbers of low
contributorstotheIRluminosityandstarformationrate redshift (z≤0.6) sources. For the 2004 post-launch cam-
density at z ∼1. paign,apurely24µmselectedsample(f >120µJy)was
24
In the discussion that follows, any calculation re- targeted for follow-up. The combined I distribution
AB
quiring cosmology assumes Ω =0.3, Ω =0.70, and of targeted and detected sources is shown in Figure 1
M Λ
H =70kms−1Mpc−1. (top) along with the cumulative redshift identification
0
efficiency (bottom). The overall spectroscopic complete-
2. PHOTOMETRICANDSPECTROSCOPIC ness(definedhereasthefractionoftargetedsourceswith
OBSERVATIONS high quality redshifts) is ∼70% for the full sample and
The Spitzer extragalatic component of the FLS is ∼80% for sources with I <25.0. For a more detailed
AB
a 3.7 deg2 region centered around R.A.=17h18m00s, description of the observing strategy, primary selection
decl.=59o30′00′′. Observations of this field were taken criteriaandtheoverallfluxandredshiftdistributionssee
using all four Infrared Array Camera (IRAC) channels Choi et al. (2006).
(Fazio et al. 2004) and three Multiband Imaging Pho- SincewewereexploringtheMid-IRpropertiesofgalax-
tometer (MIPS) bands (Rieke et al. 2004). Additional ies no optical limit was imposed, instead an IR lumi-
ground base images in u*,g’ from CFHT’s MegaCam nosity cut ( LIR ≥ 5.0 × 1010 or 1.0 × 1011 L⊙) was
(Shim et al. 2006), g’, i’ data from Palomar 200” LFC used,sothatafaircomparisoncouldbemadeatdifferent
andNOAO4-mRandK’band(Faddaetal. 2004;Glass- redshifts. The absolute B magnitude (MB) distribution
man et al. 2006 in prep) have also been obtained. This of the MIPS spectroscopic sample between 0.5<z<1.3
work focuses on the 0.12 deg2 ACS-HST F814W imag- probes -21≤MB≤-19 fairly uniformly and we are not
ingoftheverificationstrip,whichhas3σdepthsinMIPS strongly biased at higher redshifts.
24µmof0.1mJy. Object detection andphotometry were
performed using Sextractor (Bertin & Arnouts 1996).
Particularcarewastakentoensureaccuratede-blending
ofgalaxiesincloseproximitytooneanother,whileavoid-
ingdetectionsofsubstructurewithinasinglegalaxy,con-
sistent with other reductions of HST imaging with close
galaxy pairs in mind (Patton et al. 2005). There were
∼59,000 sources extracted within the F814W band
AB
(hereafterextractedmagnitudes referredto asI ). We
AB
compared our number counts to those from the Hubble
Deep Field (HDF) North and South and determined a
limiting magnitude of I ∼27.4.
AB
Using the full MIPS catalog from the FLS we selected
24µmsources within the areacoveredby the ACS imag-
ing (∼0.12 deg2). In order to correlate the MIPS ob-
jects with those identified in the optical we first cross-
identified sources from the MIPS 24µm sample to the
IRAC catalog using a tolerance radius of 2.0′′. This
choice was primarily motivated by the FWHM of the
MIPS 24µm (PSF∼6′′) and confirmed by visual inspec-
tion. We then cross-correlated the IRAC/MIPS catalog
totheACSsamplewhichwebandmergedwithu*,g’and
Rrequiringa positionalagreementof≤1′′. Whenmulti-
ple counterparts were identified, we selected the closest
object. Ultimately we found 1155 ACS sources also de-
tected by IRAC and MIPS at 24µm. Fig. 1.—(Upper)TheIAB magdistributionfortheFLSsample
Theredshiftsusedinthisstudyweredeterminedexclu- (solid),thesampletargetedforspectroscopy(long-dash),andthose
wherespectroscopicredshiftswereacquired(dotted). (Lower)The
sively from optical spectroscopy. They were obtainedby
cumulative redshift identification efficiency. The overall spectro-
cross-correlating the ACS sample, limited to IAB ≤26.5 scopic completeness is ∼ 70% for the full sample and ∼80% for
(N∼29,000) with various FLS spectroscopic datasets. sourceswithIAB <25.0.
Thevastmajorityofthe includedredshifts(≥97%)were
obtained with the Deep Imaging Multi-Object Spectro-
Cross-correlationoftheband-mergedphotometriccat-
graph (DEIMOS) on the W.M. Keck II 10-m telescope;
alogs with the redshift samples results in a data set of
however, the final sample also included a few redshifts
476 sources with I <26.5 and 0.2≤z≤1.3. Of those,
basedonSloanDigitizedSky Survey(SDSS) andWIYN AB
245 (51%) are MIPS 24µm-detected with a measured
Hydra/MOS (Marleau et al. 2006 in prep) spectra.
L ≥5.0×1010 L . Theremaining231(49%)arenon-
GalaxiesintheFLSVerificationregionweretargetedfor IR ⊙
detected at 24µm. Despite the fact that the MIPS and
spectroscopic follow-up during two DEIMOS campaigns
non-MIPS galaxies were selected slightly differently, the
that bracketed Spitzer’s launch. The selection criteria
resultant colors of objects with spectroscopic redshifts
for these campaigns are summarized below. For the
have uniform color properties (Figure 2).
2003 pre-launch campaign, targets were selected based
on NIR (K ) and optical (g,R,i) colors. The primary
s
Interacting & Merging Galaxies: Spitzer’s View 3
3.1. Pair Statistics
Weappliedtheclosepairstechniquetoidentifytheav-
erage number of close companions per galaxy, hereafter
N . This measurement is similar in nature to the pair
c
fractionwhenthereareinfrequenttriplesorhigherorder
N-tuples. Sincethisisthecasehere,N willbeoccasion-
c
ally referred to as the pair fraction. Companions were
selected using a standard operational close pair defini-
tion of 5h−1kpc≤ r ≤ 20h−1kpc, and an optical
projected
magnitude difference (∆m) ≤ 1.5 compared to the host
galaxy, to select nearly equal mass major mergers. The
term “host” or “primary” galaxy are both used to ref-
erence the pair member with a measured redshift. The
innerradiusisappliedtoavoiddetectionofsubstructure
within a galaxy, while the outer 20h−1kpc limit repre-
sents the radius within which satellites are expected to
strongly interact with the halo of the host and merge
Fig. 2.— Color-color plot where red squares depict the 24µm within 0.5-0.9 Gyrs (Patton et al. 1997; Conselice et al
detected objects with spectroscopic redshifts, while blue triangles
2003). We find 87 close pairs out of 476 galaxies which
show the undetected 24µm spectroscopic sample. The black dots
represent the full FLS-ACS catalog for comparison. The 24µm fulfill these criteria (see Table 1).
detected and undetected spectroscopic samples occupy the same To study the fraction of IR-bright galaxies in pairs,
colorspace. Someobjects werenotdetected inallfourbands due
we split the pair sample into two sub-sets: those which
tothefieldcoverage,depthsandseeingdifferencesbetweenfilters.
weredetectedandthose undetected with MIPSat24µm
down to the flux limits of our survey (0.1 mJy). Figure
3 shows a subset of close pairs (both detected and unde-
3. STATISTICSOFCLOSEPAIRS,ANDMERGERS tected at 24µm) with MIPS contours. Due to the small
Toproperlyconstraintheroleinteractionsandmergers separationsof close pairs (20h−1kpc correspondsto 3.6”
play in galaxy evolution, all stages of the process must at z ∼ 1) relative to the beam of the MIPS 24µm im-
be considered. Typically, merger history analyses utilize ages (FWHM ∼6”), there are a few instances (5) where
either pair or structural methods. Galaxies in pairs are only a single 24µm detection is found centered between
pre-mergers, or systems undergoing interactions, while the pair members (see middle left imagein Figure 3). In
morphological or structural methods find galaxies that these cases we assume all 24µm flux is coming from the
have already undergone a merger and are dynamically primary galaxy.
relaxing.
Whendiscussingthepairfraction,andinferredmerger 3.2. Field Correction
rates(asdefinedinSection4),itmustbenotedthatthese
Sincewehaveredshiftinformationforonlytheprimary
measurements are highly dependent on the techniques
galaxyandnotthecompanionsweneedtoconsiderwhat
and selection criteria used to identify ongoing mergers,
fraction of these close pairs are a result of random pro-
especially for galaxies at high redshifts.
jection effects. A field correction was determined using
The first method, which we will call the “close pair
two separate methods to account for these close pairs.
method”, is to count the number of galaxy pairs within
The first assumed the same optical magnitude and red-
someprojectedseparation,r ,andmagnitudedif-
projected
shift distributions independently for both the detected
ference (∆m). If we assume that these systems will
and undetected 24µm samples, while the positions were
merge within a given time-scale due to dynamical fric-
randomized. The close pair algorithm was applied to
tion, we can determine the merger rate. Although not
50 realizations of these mock catalogs and the average
allpairswillmerge,they canpotentiallytriggerstarfor-
N for each redshift bin was taken to be the pair frac-
mation through gravitational interactions. An alterna- c
tion expected from random. This assumes the absence
tive is to selectmergingsystems basedonmorphological
of clustering.
indicators either by overall appearance (Le F`evre et al.
We investigated the environments of 24µm detected
2000) or computational measurements such as asymme-
andundetectedobjectsonscalesofr <20h−1kpc,
try (A), andclumpiness (S) of asystem(Conselice et al. projected∼
and found them to be comparable, confirming that the
2000, 2002; Conselice 2003), or Gini coefficient (G), and
increaseinpair-fractionofthe24µmdetectedpairsisnot
M parameters(Abraham et al.2003;Lotz et al.2006).
20
because they preferentially lie in clusters. On the other
Due to the comparatively limited spatial resolution of
Spitzer (compared with optical imaging), seaching for hand,thereisaweakindicationthatgalaxiesdetectedat
24µm are more likely to lie in small groups. Since such
closegalaxypairsormorphologicalsignaturesofinterac-
groups may, in some cases, be physical associations, we
tion atMid-IR wavelengthsis currently restrictedto the
countsuchcasesasseparatepairs. However,thenumber
nearby universe. However, we can correlate optically-
ofthese casesis small,anddoesnotinfluence ourresults
selectedpairs/mergerswithglobalMid-IRpropertiesand
in any significant way. The second method utilizes the
investigate the IR activity in these systems out to high
I magnitudedistributionofthefullphotometriccata-
redshifts. AB
log (∼59,000sources), and determines the averagenum-
ber of companions, within 1.5 mags (I ), normalized
AB
to the area covered by 5h−1kpc≤r ≤20h−1kpc.
projected
4 Bridge et al.
Fig. 4.— The field corrected pair fraction Nc as a function of
redshiftasmeasuredintheopticalandIR.Thestarsrepresentthe
measurementfromour24µmdetectedsample,trianglesdepictthe
field corrected pair fraction of the undetected 24µm sub-set, and
Fig. 3.—Asubsetofpairedgalaxiesinoursample. Eachimage filled circles show the combined pair fraction of the two samples.
is 60h−1kpc on a side with axes in arcseconds, centered on the Other optically-determined pair fractions appear as open squares
pairmemberwiththespectroscopicredshiftalsoreferredtoasthe (DEEP2 Field 1) and triangles (Field 2) (Linetal. 2004), cross
primaryor host pair member (white circle), while the companion forCNOC2(Patton etal.2002),whilethedashedlineshowstheir
ishighlightedinablackcircle. Thetwoupperrowsareclosepairs bestpowerlawfitof(1+z)m (m=1.08±0.40). Thenear-IRpair
whichweredetectedat24µm,whilethelowersetarefromtheun- fractiondeterminedbyBundyetal.(2004)isshownbydiamonds.
detected24µmpairedsample. The3−10σ24µmfluxcontoursare Errorsforthisworkarederivedusingjacknifestatistics,whilethe
overlaid. Thelabelsarespectroscopicredshift,andIABmagnitude IRpairfractionerrorsimplementedcountingstatisticsandDEEP2,
oftheprimarygalaxy. CNOC2errorsaredeterminedviabootstrap. ALIR≥5.0×1010
limit was imposed on the 24µm detected close pairs. Note that
The results obtainedfromthe two field correctionmeth- eachworkimposes aslightlydifferent luminosityand stellarmass
odsagreedwithin∼2%,whichisnegligiblecomparedto limit.
the uncertainly in N . The average of the two methods fraction of MIPS bright sources is marginally significant
c
was taken to be the final field correction. Both the pair due to the small number of sources in the highest red-
catalog and randomly generated catalogs were visually shift bin, more MIR selected samples between z =1-1.5
inspected forfalse pairsdue to single galaxiesbeing bro- are required to strengthen our findings.
kenupintomultiplecomponentsinthesourceextraction We would like to be able to rule out the possibility
phase,orcontaminatingstarsinthephotometriccatalog, that N is biased by the brightest IR sources at z ≥ 1,
c
and were removed. since merger fractions change as a function of lumi-
nosity and mass (Conselice et al. 2003; Xu et al. 2004).
3.3. Pair Fractions To address this we placed a higher IR luminosity limit
(L ≥ 7.0×1011) on the sample, so that at z ≥ 0.7
The field-correctedoptical pair fractions for the 24µm IR
the same populations were being probed (optically we
detected and undetected sub-samples are presented in
are probing -22<M <-19). We still find an increase in
Figure 4 and Table 1. Errors are computed using ∼ B∼
the jackknife technique (Efron & Tibshirani 1986), e.g. Nc from the lower (0.8 ≤ z ≤ 1.0) to the higher (z ≥ 1)
given a sample of N galaxies the variance is given by redshiftbinsofsimilarmagnitudecomparedtowhenthe
[(N −1)/NP δ2]1/2. The partial standard deviations, lower IR limit (LIR ≥ 5.0×1010) was used. Therefore
δi, are computiedi for each object by taking the differ- the increase in Nc found at z ≥1 is likely not a result of
merely probing brighter IR systems but rather due to a
ence between N , the quantity being measured and the
c
physicalincreaseinthe mergerateforthe 24µmpopula-
same quantity with the ith galaxy removed, N , such
ci
that δi=N −N . tion,howeverdeeper24µmimagingandspectroscopyare
c ci
requiredtoconfirmthis. Whenweconsidertheaveraged
Toallowamoredirectcomparisontobemadebetween
pair fraction over 0.2 ≤ z ≤ 1.3 for the 24µm detected
the generally lower-luminosity low-z pairs, and those at
samplewefindthatgalaxiesaboveafluxlimitof0.1mJy
higher redshift, we derived pair fractions for MIPS de-
tectedgalaxieswith anL ≥5.0×1010 (approximately arefivetimesmorelikelytobeinaclosegalaxypair,than
IR
those below this limit.
the IR luminosity of the famous Antenna Galaxies). In
thiswayweensurethatthesub-luminousgalaxiesdonot
3.4. Morphological Mergers
strongly influence the pair fractions in the lowest red-
shift bin. The derived N for 24µm detected close pairs Toexplorethestructuralcomponentsofgalaxiesinour
c
is ∼ 11%±8% at z ∼ 0.4 and increases to 25%±10% sample we used the CAS (Concentration, Asymmetry,
at z∼1. In contrast, close pairs with no 24µm detection Clumpiness)quantitativeclassificationsystem(Conselice
show no increase with redshift and have pair fractions 1997; Conselice et al. 2000, 2002; Conselice 2003), and
consistent with zero at all redshifts. The higher pair visualclassifications. Tomeasurethemergerfractionus-
Interacting & Merging Galaxies: Spitzer’s View 5
TABLE 1
FLS Close PairStatistics
z Ngal NcD NcR Nc κ
24µm Detected
0.2-0.5 32 0.188(6) 0.078(2.5) 0.110±0.083 0.83
0.5-0.80 82 0.171(14) 0.057(4.7) 0.114±0.040 0.93
0.80-1.0 82 0.122(10) 0.029(2.4) 0.093±0.038 0.90
1.0-1.3 49 0.429(21) 0.182(8.9) 0.247±0.086 0.67
24µmUndetected
0.2-0.5 44 0.136(6) 0.102(4.5) 0.034±0.052 1.00
0.5-0.80 76 0.132(10) 0.134(10.2) 0±0.039 1.00
0.80-1.0 56 0.214(12) 0.193(10.8) 0.021±0.064 0.83
1.0-1.3 55 0.145(8) 0.180(9.9) 0±0.065 0.75
Note. —Ngal isthenumberofgalaxieswith aspectroscopicredshift,NcD
is the numberof companionsperhost fulfillingour pair criteria, while NR is
c
the number of projected companions per host from the field. The corrected
fractionofcompanionsperhostisgivenas,(NcD-NcR)/Ngal,witherrorsbeing
determinedusingajacknifetechnique. Numbersappearinginparenthesesrefers
tothenumberofclosepairsintherespectiveredshiftbins. Undetectedat24µm
refers to sources below the limits of our survey (0.1mJy). The constant κ is
the fractional number of mergersper host galaxy. A LIR ≥ 5.0×1010 limit
wasimposed.
ingstructuralclassificationswevisuallyinspectedthefull
24µm detected spectroscopic catalog with the following
groupings: early type (E, S0), mid-types (Sa-Sb), late-
types (Sc-irr), compact systems, disturbed disks, and
mergers. The methodology for carrying out this clas-
sificationisdescribedindetailinConselice et al.(2005).
Basically, each galaxy was viewed on a computer screen
andclassifiedinto one of ourtypes. Overallwe find that
55%± 5% of 24µm detected galaxies are disks, which
is consistent with Bell et al. (2005); Lotz et al. (2006),
while26%±5%aremergingsystemsand∼6%wereclas-
sified as disturbed disks and are possible minor mergers.
A fraction of the disk-dominated objects do show some
visual signs of a morphological disturbance, or are in a
pair, as we will discuss later in this paper.
Galaxies undergoing a major merger event can also
generally be identified by their large asymmetries in the
rest frame optical (Conselice et al. 2000, 2003). We de-
fined a major merger as a galaxy having an asymmetry
(A) ≥ 0.35 and I ≤ 26.5 (see Figure 5 for exam-
AB
ples). This limit has been shown to be a clean way to
find galaxy mergers, without significant contamination
from non-merging galaxies (Conselice 2003). Figure 6 Fig. 5.— A subset of 24µm detected galaxies in our sample
shows how the merger fraction for CAS defined mergers classifiedasamerger. Eachimageis30h−1kpconasidewithaxes
inarcseconds. The3−10σ 24µm fluxcontours areoverlaid. The
evolves as a function of redshift for both 24µm detected
objects (top panel) and LIRG/ULIRG galaxies (bottom labelsincludespectroscopicredshift,andF814WAB magnitude.
panel). As with the 24µm detected close pair sample inspected and classified by four of the authors to be ei-
there is an elevated merger fraction compared to other ther disk or bulge-dominated. We find that 81% of the
works(Cassata et al.2005;Lotz et al.2006)inwhichno 24µmpairs havedisk morphologieswhile 74%ofthe un-
24µm flux limit was imposed, and a slight indication of detected24µmhostswerealsodiskdominated,hencethe
evolution with redshift, but it is statistically consistent discrepancy between the asymmetries of the two groups
with m ∼ 1.0 (dashed line), where m is the slope of a is not causedby classificationdifferences, but rather is a
power-law of form (1+z)m later discussed in §5. physical effect.
We also performed a CAS analysis of our close pairs
sample which revealed that 24µm detected pairs are no- 4. MERGERRATES
tably more asymmetric than the undetected-MIPS close
One of the goals of studying mergers and interactions
pairs (Figure 7), suggesting that interactions and colli-
is to determine how the galaxy merger rate evolves with
sions may play a role in their IR activity. If the 24µm
redshift. Most studies of galaxy mergers involve deter-
detected close pairs were generally of a different mor-
miningthemergerfraction,yetthemergerrate,whichis
phological classification than those pairs undetected at
defined as the number of galaxies merging per unit time
24µm the discrepancy in the asymmetries could be ex-
per unit volume, is a more physical quantity that can
plained. Toaddressthisissueeachclosepairwasvisually
be used to determine the full merger history. The rate
6 Bridge et al.
numberofmergerstakingplaceperunittimeperunitco-
movingvolume(ℜ ). Sinceweareprimarilyinterested
mgv
in mergers which are also Mid-IR bright systems we will
have to restrict ourselves to measuring ℜ because the
mg
evolution of the 24µm luminosity function with redshift
is currently not well constrained, and our redshifts are
not complete enough to reconstruct this evolution.
In order to determine ℜ we need to identify sys-
mg
tems which are destined to merge. We have approached
thismeasurementfromthreedifferentperspectives,close
pairsto selectpre-mergersor interactions,visualinspec-
tion to select interactions after the first passage, and
late stage mergers, as well as CAS criteria which quan-
titatively selects for later stage mergers. By combin-
ing information about the number of ongoing mergers
(N ) and the time-scales, (T ) on which they will un-
m mg
dergo said merger, one can estimate an overall merger
rate ℜ =N /T . Each method of identifying merg-
mg m mg
ers/interactions is capturing a different snapshot of the
Fig. 6.— Merger fraction as a function of redshift, usingquan-
titative morphological criteria. The filled squares represent mea- merger process, each with different merger timescales.
surements from FLS (this work) of sources with a 24µm detec- The value of N is directly proportional to the num-
c
tion. The“x’s”marktheresultsofConseliceetal.(2003),triangles
ber of mergers per galaxy (N ), such that N = κN
depict Cassataetal. (2005), and diamonds represent Lotzetal. m m c
(κ is a constant relating to the number of mergers per
(2006). Toppanelshowsthemergerfractionofother studieswith
no24µmcriteriaimposed,whilethemeasurementsfromthiswork galaxy). Hence, the merger rate detemined using close
bareestmfietrogfer(s1w+izt)hmLIuRsin≥g5a.l0lp×o1in0t1s0L(f⊙ro.mTthheedoatshheerdstliundeisess)hwowitshtnhoe galaxy pairs is given by ℜmg = κNc/Tmg. The value
of κ depends on the nature of the merging systems un-
MIPSlimitimposed(m=1.08). Bottom panel shows the merger
fraction for LIRG/ULIRG galaxies (LIR ≥ 1.0×1011L⊙). Error der consideration. If one were to identify a pure set of
barsderivedusingPoissonstatistics. galaxy pairs each consisting of one companion undergo-
ing a merger, then κ = 1.0. In our case it exclusively
accounts for close pairs which are in doubles and per-
haps higher order N-tuples. Our definition of κ differs
by a factor two from Patton et al. (2000) which in this
instance would have κ = 0.5 since they have redshifts
for both pair members and one merger is made up of
two companions. We have redshift information for only
one pair member, therefore one merger is made up of a
primary and one companion. The merger rate equation
for merging galaxies selected by visual classification and
CAS parameters is simply ℜ = f /T , where f
mg mg mg gm
is the galaxy merger fraction.
Before the merger fraction can be used to calculate
the merger rate we need to understand the time-scale
in which a merger occurs. Each technique of identifying
mergers has a different time-scale since each is sensitive
to a different interval of the merger process. There are
two main methods that have been used to estimate the
Fig. 7.—Asymmetry-Concentrationdiagramforthe24µmde- time-scale of a merger: dynamical friction arguments,
tected close pairs(squares) and undetected pairs(triangles). The andN-bodymodels. Thedetailsofthesemethodsarebe-
longdashed(vertical)lineseparatesmergingandnon-mergingsys- yondthe scopeofthispaperbut seePatton et al.(2000)
tems(A≥0.35isconsideredamerger),whilethedash-dottedlines
and Conselice (2006) for a review. We take the aver-
separateearly(upper)tomidtolate(lower)typegalaxiesdefined
byConselice(2003). Generallyclosepairsaremid/latetypegalax- age merger time-scale for a set of close companions of
ies,andsomewouldalsobemorphologicallyclassifiedasamerger. roughly equal mass to merge as ∼0.5 Gyr±0.25,derived
fromdynamicalarguments(Patton et al.2000;Conselice
in which galaxiesmergealso affects the mass function of 2006). Conselice (2006) showedthrough N-body simula-
galaxies, and is likely linked to the cosmic star forma- tions that visual classification selects on-going mergers
tionrate. Sinceweareconsideringaverybroadrangein over a longer time-scale (1.0 Gyr±0.25) since the hu-
the merger process, from early-stage or pre-mergers se- maneyedetectsbothearlyandlaterstagemergers,while
lectedviaclosegalaxypairs,andlater-stagemergerscho- the asymmetry of a galaxyis sensitive to 0.41 Gyr±0.17
sen based on morphological criteria, we must be careful (Conselice 2006) of the merger sequence.
when determining their respective merger rates, as the
time-scales for these processes are all different.
There are two variations of the merger rate definition.
The first is the number of mergers that a galaxy will
undergoperunittime(ℜ ),andthe secondisthetotal
mg
Interacting & Merging Galaxies: Spitzer’s View 7
5. THEEVOLUTIONOFTHEGALAXYMERGERRATE
0.2≤Z≤1.3
Within the past two decades numerous studies have
been performed to estimate the evolution of the galaxy
merger fraction, using both the close pair technique
(Zepf & Koo 1989; Burkey et al. 1994; Carlberg et al.
1994; Yee & Ellingson 1995; Patton et al. 1997, 2000;
Le F`evre et al. 2000; Lin et al. 2004; Bundy et al. 2004)
and morphological parameters (Le F`evre et al. 2000;
Conselice et al. 2003; Lavery et al. 2004; Lotz et al.
2006). Evolution in the galaxy merger fraction is of-
ten parameterizedby a power-law of form (1+z)m, and
has yielded a wide range of results, spanning 0.m.5.
The large spread in values is in part due to the different
selection criteria used to identify merging systems and
biases from optical contamination or redshift complete-
ness. Patton et al. (1997) considered these biases and
demonstrated that most results to that date were con- Fig. 8.—Thenumberofmergerspergalaxy(LIR ≥5×1010)asa
sistent with their estimate of m = 2.9±0.9. Recently, functionofredshift. Threemerger/interactionselectiontechniques
areapplied, closepairs (squares), CAScriteria(stars), and visual
optical and near-IR close pair studies (Lin et al. 2004;
classification(triangles), whilemerger rates forthe combined (w/
Bundy et al. 2004) have derived merger fractions with andw/o24µmdetections)usingtheclosepairsmethodareshown
littleredshiftevolution(m∼1),ashavesomemorpholog- with circles. The long dashed curve is the best fit of the form
(1+z)m, using the FLS data for the three techniques; the dot-
ical studies using (G), and M (Lotz et al. 2006).
20 dashed curve represents the best fit for the combined total close
When we consider all the close pairs identified in our pairs(MIPSandnon-MIPSpairs).
sample, both those detected at 24µm and not, we find a
merger fraction and rate consistent with recent studies
ure9). MIPSpairsandmergerssharea similarluminos-
showinglittle redshift evolution. However,when we sep-
ity distribution to 24µm bright field galaxies, although
aratethepairsampleintosystemswitha24µmdetection
red-AGNseem generally more luminous which is in part
above 0.1 mJy, and those below it, we do see a stronger
due to template mismatches (Chary & Elbaz 2001).
evolutionofN withredshift,(recallthatN ∝ℜ )and
c c mg
The L of a galaxy is a combined measure of the
thereforealsointhe mergerfractionandrate(Figure8). IR
reprocessed UV photons intercepted by dust from mas-
Similarly, visually classifiedmergers and those identified
sive young stars and AGN. Therefore to investigate the
via asymmetry levels (A ≥ 0.35) using the CAS param-
contribution an interacting or merging galaxy makes to-
eters,also showredshift evolutionin the mergerfraction
wards the total L density from star formation alone
andrate. Themergerfractioncomputedusingthediffer- IR
we must first remove AGN from our sample. Due to
ent methods are in good agreement when normalizedby
the nonuniform rest-frame spectral coverageof our sam-
their respective time-scales, reinforcing the idea that we
ple we rely on the four-band IRAC color selection used
are probing different phases of the merger process. Con-
by (Lacy et al. 2004) to identify and remove AGN can-
sideringallthree mergerselectiontechniqueswefindthe
bestfitofthemergerrateparameterizedbyℜ(0)(1+z)m didates (Figure 10). Over the modest redshift range of
tobe 0.077±0.045,2.12±0.93,withareducedχ2=0.39. our sample, this method is still effective at separating
IR-warmAGNfromstarburstsystems. WefindanAGN
Thisresultsuggeststhatwhenone considersasampleof
contamination rate of ∼ 12% for the full 24µm sample,
close galaxy pairs solely on their optical fluxes, brighter
while ∼14% of the hosts in a pair or merger were char-
than M =-19, little evolution of the merger rate with
B
acterized as AGN.
redshift is found. However close pairs emitting 24µm
WithAGNcandidateobjectsremovedwecaninferthe
flux exhibit an increase in the mergerrate with redshift.
The infrared luminosity limit (L ≥ 1011L ) imposed contributiontotheLIRdensityfromstarformationcom-
IR ⊙ ingfrom24µmgalaxiesinainteraction/mergerasafunc-
on the close pairs and mergers allows us to primarily
tion of redshift. We derive the number of statistically
probe systems in a LIRG/ULIRG phase at z ≥ 0.4 (see
“real” galaxy pairs from our pair fraction result at each
nextsectionfordetails). Theincreaseofthemergerfrac-
redshift interval and determine the total L density
tion and rate of this population of galaxies coupled with IR
fromclosepairswhichisinturndividedbytheL den-
the fact that LIRG/ULIRG galaxies dominate the SFR IR
sity from the whole sample. We find that paired galax-
density at z ≥ 0.7 (Le Floc’h et al. 2005) suggests that
merging does in fact play an increasingly important role ies (LIR ≥ 1011L⊙) are responsible for 27%+−98%% of the
in star formation out to z ∼1. IR background stemming from star formation at z ∼ 1.
Since we only know the redshift of the host galaxy we
6. TOTALINFRAREDLUMINOSITIESOFMERGERS select “real” close pairs in a statistical sense, and derive
One way to quantify the role merging galaxies play in errorbarsfortheclosepairscontributionbythespreadof
triggeringstarformationis to investigatetheir contribu- 50realizationsoftheL densityfromdifferentcombina-
IR
tion to IR luminosity densities. Infraredluminosities (8- tionsof24µmgalaxypairs. Wealsoappliedthisanalysis
1000µm ) were calculated utilizing the 24µm fluxes and toCASandvisuallyclassifiedmergers,whichmakeupan
two different template methods: Chary & Elbaz (2001); additional∼12%,and∼22%oftheIRluminositydensity
Dale et al.(2001)inasimilarmannerasLe Floc’h et al. respectively. Naturallythereisasomeoverlapinmergers
(2005)forthefullMIPS24µmspectroscopicsample(Fig- identified through close pair criteria and morphological
8 Bridge et al.
Fig. 9.— Infrared luminosity LIR[8−1000µm] vs. redshift for Fig. 11.—ThefractionofLIRdensity(asaresultofstarforma-
theMIPS24µmsamplewithspectroscopicredshifts,brokendown tion)asafunctionofredshiftcomingfromLIRG/ULIRGgalaxies
into galaxies in a close pair (red triangle), CAS mergers (open inaclosepair(pre-mergerphase)shownbytriangles,ormoread-
green diamonds), AGN candidates (yellow star), AGN candidates vanced stage mergers defined morphologically (circles). The star
in a close pair (blue square), AGN mergers (open black square), symbolindicates the total combined contribution fromclosepairs
and field galaxies (black circles). LIRGs lie above the horizontal andCASmergers,whilesquares depictthe total fromclosepairs,
dashedlineatLIR≥1011L⊙ CAS, and visually classified mergers. Note an infrared limit of
LIR≥1011L⊙ wasimposed.
parameters,sinceinteractingpairscanexhibittidaltails
andasymmetricstructures,causingthemtoalsobeiden-
The combination of these three merger selection tech-
tifiedmorphologicallyasmergers. We foundthat37%of
niquesidentifiesalargerangeinthemergerprocess,from
CAS defined mergers were also in a close pair, and 31%
pre-merger to late stage mergers, implying that ∼60%
ofvisuallyidentifiedmergerswerealsoclassifiedby CAS
of the infrared luminosity density at z∼1 can be at-
asmerging. Incaseswhereamergingsystemwasidenti-
tributed to galaxies involved in some stage of a major
fied using multiple techniques it’s contribution was only
merger(Figure11). Theremaining∼40%oftheIRback-
counted once. For example, if a merger identified mor-
ground from LIRG/ULIRGs is likely to predominately
phologically(either throughCASorvisualinspection)is
come from active, isolated gas-rich star-forming spirals,
also in a close pair it is removed from the morphologi-
with some contribution from minor mergers.
cal merger catalog, or if a CAS merger is also identified
If we exclude visually classified mergers the close
visually the merger is removed from the visual merger
pair/merger contribution to the IR density is ∼38%, in
catalog. This insures that no close pair or merger is
good agreement with Lin et al. (2006) who estimate a
counted more than once when deriving the contribution
moderatecontributionfrominteractingandmergingsys-
from interactions and mergers to the IR luminosity den-
tems of <36%. It must be noted however that neither
sity. ∼
Lin et al.(2006)orthisworkhaveconsideredthecontri-
butionfromminormergersandarethereforelowerlimits.
6.1. Star Formation in Mergers & Interactions
Animportantandhighlydebatedquestionis: howim-
portant are galaxy mergers in understanding the dra-
matic decline of the cosmic SFR density from z ∼
1 to the present day? It has been well established
that mergers and interactions can induce violent bursts
of star formation (Schweizer 1982; Barton et al. 2000;
Mihos & Hernquist1996;Cox et al.2006). Sotoinvesti-
gate this contribution we derived the SFR for our 24µm
detected close pairs and mergers, using their L . The
IR
infrared luminosity of a galaxy is a star formation rate
tracerwhichisunaffectedbythe extinctionofdust. The
dominant heat sources of most dusty, high-opacity sys-
tems such as LIRGs and starbursts is stellar radiation
Fig. 10.— IRACcolor-colorplotusingtheregionintheFLScov- from young stars. In these types of systems the LIR
ered by the ACS imaging. Circles represent objects withspectro- can be converted into a SFR using the calibration of
scopicredshiftsanddetectionsinallfourIRACchannels. Thered Kennicutt (1998), SFR = 4.5×10−44L (ergss−1),
trianglesindicateobjectswhichhavemetthecolorcriteria(shown IR IR
where L is the integrated luminosity from 8-1000µm
bythe dashed line) ofan AGN candidate (Lacyetal.2004). The IR
green stars and blue squares depict objects in a close galaxy pair as determined in section 6.0.
ormergerwhosehostwasalsoflaggedasanAGNcandidate. We estimated the contribution mergers and interac-
Interacting & Merging Galaxies: Spitzer’s View 9
tions above L ≥ 1011L make to the SFR density at must also be mentioned that we are probing to higher
IR ⊙
z ∼1 in two ways. The first is simply to consider their redshiftsthanBell et al.(2005),whichfoundthatmajor
contributiontotheL densitywhichisastarformation galaxy mergers account for ≤30% of the IR luminosity
IR
tracer. Section6.0determinedthatmergersandinterac- density at z∼0.7, consistent with our findings of 35%
tions at z∼1 (above L ≥ 1011L ) are responsible for at that redshift. Our results also agree that at z ∼0.7
IR ⊙
40-60%oftheIRluminositydensity. Usingtheresultsof isolated undisturbed spiral galaxies are a primary con-
Le Floc’h et al.(2005) whichshowedthat z ≥0.7 LIRGs tributor,however,theinfluenceshiftstointeractionsand
produce ∼ 70% of the star formation rate density, we mergers at z>0.7.
caninfer thatmergersandinteractionsinLIRG/ULIRG OurfindingspointtoanincreasedimportanceofMIPS
phasewouldberesponsiblefor∼30−40%(0.6×70%)of brightinteractionsandmergerstotheIRluminosityden-
the SFR density at z ∼1, since IR activity traces dusty sity and SFR density at z≥0.7. This conclusion is not
star formation. hampered by the small statistics of the z >1 bin. Fig-
The second more detailed approach utilizes the SFR ure11showstheIRluminositydensitycontributionfrom
density directly arising from our sample of mergers and interactions/mergers at z∼0.7 to be ∼37% and 52% at
interactions. At 1.0≤z≤1.3 we find that 59% (12 close z∼0.9,reinforcing this increasing trend.
pairs, and 17 later stage mergers) of galaxies detected
at 24µm are involved in some stage of an interaction or
merger. Inthissameredshiftrangeoursampleisinsensi- 7. DISCUSSION
tive to galaxies with IR luminosities ≤1011.5. To correct Usinga spectroscopicsampleoffieldgalaxiesfromthe
for this we derived a scaling factor (∼7) simply by com- ACScomponentoftheFLSanddividingitintotwosub-
paring the number of observed objects of a given LIR sets, those with a 24µm detection (above 0.1mJy) and
in a specific redshift range to the number expected from those without (or below) we identified optically merg-
models(Lagache et al.2004). Howevertogoanyfurther ing/interactingsystemsviaclosepairstatisticsandmor-
we must assume that our spectroscopic sample is repre- phologicalmethods. We find that roughly 25% of galax-
sentative of this population at ∼1, and by all accounts ies emitting at 24µm have a close companion at z ∼1
this appears to be true. Using the derived pair fraction while at z ∼0.5 only ∼ 11% are in pairs. In contrast,
wecantheninferthetotalnumberofmajormergersand those undetected at MIPS 24µm showed a pair fraction
interactions occurring (fulfilling our criteria) in a given consistent with zero at all redshifts (0.2≤z≤1.3). On
volume and LIR limit. The lower limit of the SFR den- average MIPS 24µm galaxies are five times more likely
sity at z ∼1 from merging and interacting galaxies is than non-MIPS sources to have a close companion over
found to be 0.066 M yr−1Mpc−1. Using the extinction 0.2≤z≤1.3. When the samples are combined (regardless
⊙
corrected “Lilly-Madau” plot (0.1585 M yr−1Mpc−1 at of24µmflux)wefindpairfractionsconsistentwithprevi-
⊙
z =1) (Thompson et al. 2001) we find that mergers and ous studies (Lin et al. 2004; Bundy et al. 2004) showing
interactions are responsible for at least 42% of the SFR little evolution with redshift.
densityatz ∼1(assumingmergerscontribute60%ofthe An important and open question is the cause of star
IRdensity). Bothapproachesareingoodagreement,and formation in LIRG galaxies at high-z. Some morpho-
areonlyalowerlimit,sinceobjectsflaggedasAGNwere logical studies have suggested that since at least half
not consideredeven thoughsome of their L is a result of the LIRG galaxies exhibit disk dominated morpholo-
IR
of star formation, and minor mergers which have been gies (Bell et al. 2005; Lotz et al. 2006) at z ≥0.7 and
shown to also induce bursts of star formation were not lownon-evolvingmergerfractions(Lotz et al.2006)that
included. the driver of IR activity in high-z LIRGs is from on-
These results have interesting implications for the going star-formation from isolated gas-rich spirals and
physical mechanisms that drive the decline in the cos- notmergerorinteractioninduced. OnebiasofHSTmor-
mic SFR (CSFR) density from z ∼ 1 to present day. phological studies involving the identification of merg-
They suggest that when all stages of the merger process ing/intereactingsystemsisthelimitationofdetectinglow
are considered (pre-merger to later stage merger) ma- surface brightness features such as tidal tails caused by
jor interactions and mergers contribute close to half of closeinteractions,whichcanleadto anunderestimateof
the z ∼ 1 SFR density, and the decline in the num- the importance of mergers in the evolution of galaxies
ber of 24µm detected mergers/inteactions is a signifi- at z <1. Ultimately both close pair and morphological
cant, but perhaps not the primary driver for the decline techniques must be applied and considered, to obtain a
in the cosmic SFR. This conclusion differs in interpre- completemajormergertimeline. Ouranalysisisthefirst
tation from Bell et al. (2005); Melbourne et al. (2005); toprobemergerrateevolutioncombiningclosepairsand
Wolf et al. (2005); Lin et al. (2006); Lotz et al. (2006), later stage mergers while considering the IR activity of
whichgenerally suggestthat the evolutionof the merger these systems.
rate is not a significant underlying cause of the decline We find that close pair statistics, visually classified
in the cosmic SFR, but rather a strong decrease in the mergers, and those identified via quantitative CAS pa-
SFRofmorphologicallyundisturbedspiralgalaxiesisthe rameters all showed similar evolution in their merger
dominant mechanism. Their results do not preclude the rates. Fitting the mergerrate evolutionfunction ℜ(z)∝
possibility that their “star forming (undisturbed) disks” (1+z)m for 24µm detected mergers above 0.1mJy, we
could be in widely separated pairs, and when we only find m = 2.12±0.93. This result agrees with previous
consider quantitatively defined morphological mergers claims of an increase (m ≥ 2) of the merger rate out
ourresultsareconsistentwiththeirsstressingtheimpor- to z ∼1 (Patton et al. 1997, 2000; Le F`evre et al. 2000;
tanceofconsideringthemergerprocessinitsentirety. It Conselice et al. 2003; Cassata et al. 2005). Howeverthis
evolution is not seen when IR faint (< 0.1mJy) merg-
10 Bridge et al.
ers are included, suggesting that it is the LIRG-merger standingoftheevolutionandmassassemblyofluminous
population that is evolving with redshift. IR galaxies.
TheMid-IRemissionofLIRGsisindicativeofdusten-
shrouded star formation (and some AGN activity), and
at z ≥0.7 they dominate the IR luminosity density and WewouldliketothankH.Shim,M.Im,R.Chary,and
in turn the volume-averagedstar formation rate density C. Borys for their contributions to this work, V. Char-
atz ∼1. Weestimatethatclosegalaxypairsarerespon- mandaris for useful suggestions,and the anonymousref-
sible for ∼ 27% of the IR luminosity density resulting eree for valuable comments that improved the clarity of
from star formation at z ∼ 1, while later stage mergers thepaper. Thisworkisbasedonobservationsmadewith
contribute ∼ 35%. This implies that 40-60% of the in- the Spitzer Observatory, which is operated by the Jet
frared luminosity density at z ∼ 1 can be attributed to Propulsion Laboratory, California Institute of Technol-
galaxies involved in some stage of a major merger, indi- ogy, under NASA contract 107. Support for this work
cating that merger-driven star formation is responsible was provided in part by the Spitzer Graduate Student
for 30-40% of the star formation density at z ∼ 1. This Fellowship program and an Ontario Graduate Scholar-
value is a lower limit since minor mergers and interac- ship in Science and Technology. The authors wish to
tions/mergers with an AGN were not considered. recognize and acknowledge the very significant cultural
Ultimately, our findings suggest that interactions and role and reverence that the summit of Mauna Kea has
mergersof LIRGphase galaxiesplay anincreasinglyim- always had within the indigenous Hawaiian community.
portant role in both the IR luminosity and SFR density We are most fortunate to have the opportunity to con-
from z ≥ 0.7 out to z ∼1.3, and are vital to our under- duct observations from this mountain.
REFERENCES
Abraham, R. G., van den Bergh, S., Glazebrook, K., Ellis, R. S., Fadda,D.,etal.2006,AJ,131,2859
Santiago, B.X., Surma,P., & Griffiths,R. E. 1996, ApJS, 107, Fazio,G.G.,etal.2004, ApJS,154,10
1 Kennicutt,R.C.,Jr.1998,ARA&A,36,189
Abraham, R. G., Tanvir, N. R., Santiago, B. X., Ellis, R. S., Lacy,M.,etal.2004, ApJS,154,166
Glazebrook, K.,&vandenBergh,S.1996, MNRAS,279,L47 Lacy,M.,etal.2005, ApJS,161,41
Abraham,R.G.,vandenBergh,S.,&Nair,P.2003,ApJ,588,218 Lagache, G.,etal.2004,ApJS,154,112
Barnes,J.E.2004,MNRAS,350,798 Lavery, R. J., Remijan, A., Charmandaris, V., Hayes, R. D., &
Barton,E.J.,Geller,M.J.,&Kenyon,S.J.2000,ApJ,530,660 Ring,A.A.2004, ApJ,612,679
Bell,E.F.,etal.2005,ApJ,625,23 LeF`evre,O.,etal.2000,MNRAS,311,565
Bertin,E.,&Arnouts,S.1996,A&AS,117,393 LeFloc’h,E.,etal.2005,ApJ,632,169
Bundy, K., Fukugita, M., Ellis, R. S., Kodama, T., & Conselice, Lilly,S.J.,LeFevre,O.,Hammer,F.,&Crampton,D.1996,ApJ,
C.J.2004,ApJ,601,L123 460,L1
Burkey, J. M., Keel, W. C., Windhorst, R. A., & Franklin, B. E. Lin,L.,etal.2004,ApJ,617,L9
1994, ApJ,429,L13 Lin,L.,etal.2006,ApJ,accepted, astro-ph/0607272
Carlberg,R.G.,Pritchet,C.J.,&Infante,L.1994,ApJ,435,540 Lotz,J,M.,Davis,M.,Faber,S.M.,Guhathakurta, P.,Gwyn,S.,
Carlberg,R.G.,etal.2000,ApJ,532,L1 Huang,J.,Koo,D.C.,LeFloc’h,E.etal.2006,ApJ,submitted.
Cassata,P.,etal.2005,MNRAS,357,903 Madau,P.,Pozzetti, L.,&Dickinson,M.1998,ApJ,498,106
Chary,R.,&Elbaz,D.2001,ApJ,556,562 Melbourne,J.,Koo,D.C.,&LeFloc’h,E.2005,ApJ,632,L65
Choi,P.I.,etal.2006,ApJ,637,227 Mihos,J.C.,&Hernquist,L.1996, ApJ,464,641
Conselice,C.J.1997,PASP,109,1251 Patton, D. R., Pritchet, C. J., Yee, H. K. C., Ellingson, E., &
Conselice, C. J., Bershady, M.A., & Jangren, A. 2000, ApJ, 529, Carlberg,R.G.1997, ApJ,475,29
886 Patton, D.R.,Carlberg,R.G.,Marzke,R.O.,Pritchet, C.J.,da
Conselice,C.J.,Gallagher,J.S.,III,&Wyse,R.F.G.2002,AJ, Costa,L.N.,&Pellegrini,P.S.2000,ApJ,536,153
123,2246 Patton,D.R.,etal.2002, ApJ,565,208
Conselice,C.J.2003,ApJS,147,1 Patton, D.R.,Grant,J.K.,Simard,L.,Pritchet, C.J.,Carlberg,
Conselice, C. J., Bershady, M.A., Dickinson, M., & Papovich, C. R.G.,&Borne,K.D.2005,AJ,130,2043
2003, AJ,126,1183 Rieke,G.H.,etal.2004,ApJS,154,25
Conselice,C.J.,Chapman,S.C.,&Windhorst,R.A.2003,ApJ, Sanders,D.B.,Soifer,B.T.,Elias,J.H.,Madore,B.F.,Matthews,
596,L5 K.,Neugebauer,G.,&Scoville,N.Z.1988,ApJ,325,74
Conselice,C.J.,Blackburne,J.A.,&Papovich,C.2005,ApJ,620, Schweizer,F.1982,ApJ,252,455
564 Thompson,R.I.,Weymann,R.J.,&Storrie-Lombardi,L.J.2001,
Conselice,C.J.2006,ApJ,638,686 ApJ,546,694
Cox,T.J.,etal.2006, MNRAS,submitted,astro-ph/0503201 Shim,H.,Im,M.,Pak,S.,Choi,P.,Fadda,D.,Helou,G.,&Storrie-
Dale, D. A., Helou, G., Contursi, A., Silbermann, N. A., & Lombardi,L.2006, ApJS,164,435
Kolhatkar,S.2001,ApJ,549,215 Wolf,C.,etal.2005,ApJ,630,771
Dasyra,K.M.,etal.2006,ApJ,638,745 Xu,C.K.,Sun,Y.C.,&He,X.T.2004,ApJ,603,L73
Efron,B.1981,Biometrika,68,589 Yee,H.K.C.,&Ellingson,E.1995, ApJ,445,37
Efron,B.,&Tibshirani,R.1986, Stat.Sci.,1,54 Zepf,S.E.,&Koo,D.C.1989,ApJ,337,34
Elbaz, D., Cesarsky, C. J., Chanial, P., Aussel, H., Franceschini,
A.,Fadda,D.,&Chary,R.R.2002,A&A,384,848
Fadda, D., Jannuzi, B. T., Ford, A., & Storrie-Lombardi, L. J.
2004, AJ,128,1