Galaxy And Mass Assembly (GAMA): The Life and Times of L* Galaxies

1.2 MB·English

by  A. S. G. Robotham

Mon.Not.R.Astron.Soc.000,1–30(2013) Printed31January2013 (MNLATEXstylefilev2.2)
Galaxy And Mass Assembly (GAMA): The Life and Times
L∗
of Galaxies
3
1 A.S.G. Robotham1,2⋆, J. Liske3, S.P. Driver1,2, A.E. Sansom4, I.K. Baldry5,
0
A.E. Bauer6, J. Bland-Hawthorn7, S. Brough6, M.J.I. Brown8, M. Colless6,
2
L. Christodoulou9, M.J. Drinkwater10, M.W. Grootes11, A.M. Hopkins6, L.S. Kelvin,1,2,12,
n
a P. Norberg13, J. Loveday9, S. Phillipps14, R. Sharp15, E.N. Taylor,7,16, R.J. Tuffs,11
J

0
1SUPA†, School of Physics& Astronomy, Universityof St Andrews, North Haugh, St Andrews, KY16 9SS, UK
3
2ICRAR‡, The Universityof WesternAustralia, 35 Stirling Highway, Crawley, WA 6009, Australia

3European Southern Observatory, Karl-Schwarzschild-Str. 2, 85748 Garching, Germany
]
O 4Jeremiah Horrocks Institute, Universityof Central Lancashire, Preston PR1 2HE, UK
5Astrophysics Research Institute, Liverpool John Moores University,Egerton Wharf, Birkenhead, CH41 1LD, UK
C
6Australian Astronomical Observatory, PO Box 296, Epping, NSW 1710, Australia
h. 7Sydney Institute for Astronomy, School of Physics, University of Sydney, NSW 2006, Australia
p 8School of Physics, Monash University,Clayton, Victoria3800, Australia
- 9Astronomy Centre, Universityof Sussex, Falmer, Brighton BN1 9QH, UK
o 10Department of Physics, The Universityof Queensland, Brisbane, QLD 4072, Australia
r 11Max Planck Institute for Nuclear Physics(MPIK), Saupfercheckweg 1, 69117 Heidelberg, Germany
st 12Institut fu¨rAstro- und Teilchenphysik, UniversittInnsbruck, Technikerstrae 25, 6020 Innsbruck, Austria
a 13Institute forComputational Cosmology, Department of Physics, Durham University,South Road, Durham DH1 3LE, UK
[ 14HH Wills PhysicsLaboratory, University of Bristol, Tyndall Avenue, Bristol, BS8 1TL, UK

15Research School of Astronomy & Astrophysics, The Australian National University,CotterRoad, Weston Creek, ACT 2611, Australia
1
16School of Physics, the Universityof Melbourne, Parkville, VIC 3010, Australia
v
9 07/01/2013 ABSTRACT
2
1 In this work we investigate in detail the effects local environment (groups and
7 pairs) has on galaxies with stellar mass similar to the Milky-Way (L∗ galaxies). A
.
1 volume limited sample of 6,150 galaxies is visually classified to determine emission
0 features, morphological type and presence of a disk. This large sample allows for the
3 significant characteristics of galaxies to be isolated (e.g. stellar mass and group halo
1 mass), and their codependencies determined.
v: We observethat galaxy-galaxyinteractionsplaythe mostimportantrole inshap-
i ing the evolution within a group halo, the main role of halo mass is in gathering the
X
galaxies together to encourage such interactions. Dominant pair galaxies find their
r overallstar formation enhanced when the pair’s mass ratio is close to 1, otherwise we
a
observe the same galaxies as we would in an unpaired system. The minor galaxy in a
pair is greatly affected by its companion galaxy, and whilst the star forming fraction
is always suppressed relative to equivalent stellar mass unpaired galaxies, it becomes
lower still when the mass ratio of a pair system increases.
We find that, in general,the close galaxy-galaxyinteractionrate drops as a func-
tionofhalomassforagivenamountofstellarmass.Wefindevidenceofalocalpeakof
interactionsfor Milky-WaystellarmassgalaxiesinMilky-Wayhalomassgroups.Low
mass halos,and in particular Local Group mass halos,are an important environment
for understanding the typical evolutionary path of a unit of stellar mass.
Wefindcompellingevidenceforgalaxyconformityinbothgroupsandpairs,where
morphological type conformity is dominant in groups, and emission class conformity
is dominant in pairs. This suggests that group scale conformity is the result of many
galaxy encounters over an extended period of time, whilst pair conformity is a fairly
instantaneous response to a transitory interaction.
Key words: cosmology – galaxies: environment – large scale structure
⋆ E-mail:[email protected]
2 A.S.G. Robotham et al.
1 INTRODUCTION variation in star formation, morphology and colour. This
indicates that there is no typical evolutionary history for
L∗galaxiesexistatthekneeofthepopularSchechter(1976)
these‘typical’galaxies. Wecan theoriseaboutalargenum-
parameterisation of the galaxy luminosity function. The
ber of factors that might directly alter the evolution of any
galaxy stellar mass function has a similar form, and as galaxy,includingL∗ones.Anon-comprehensivelistincludes
such the knee of the stellar mass function (M∗) represents
galaxy-galaxy interactions, local scale environment (e.g. its
broadlythesamegalaxypopulation.Duetotheshapeofthe
dark matter halo), and the large scale environment of the
galaxy luminosity function, L∗ galaxies dominate the inte-
cosmic web.
gratedmassandluminositydensitiesintheUniverseatlow
Many recent studies have investigated the role of
redshift (e.g. Baldry et al. 2012;Driver et al. 2012).
environment in driving the evolution of galaxies, with
To understand the integrated properties of all galax- much contradiction in the literature. Studies on SDSS by
ies we can, to the first order, just consider the prop- Weinmann et al. (2006) suggest strong evidence of ‘con-
erties that affect the evolution of ∼ L∗ galaxies. Sim- formity’ between central and satellite galaxies in halos
ply doing this offers insight into the fate of the median for a given amount of stellar mass, the suggestion from
and modal unit of stellar mass in a cosmological vol- Wang& White(2012) being that redder central galaxies of
ume. Further to this, apparent magnitude limited sur- a given stellar mass tend to occupy larger halos. This con-
veys at moderate redshifts (e.g. 2dFGRS, SDSS, MGC, clusion suggeststhatthehaloenvironmenthasasignificant
GAMA, see Colless et al. 2001; Adelman-McCarthy et al. roleindeterminingthepropertiesofgalaxies,ontopofthose
2008; Liske et al. 2003; Driveret al. 2011, respectively) are drivenpurelybystellarmass.Countertothisisrecentwork
dominated by L∗ galaxies in terms of raw number counts, by Peng et al. (2010) using z-COSMOS and Thomas et al.
eventhoughtheyarenotthemostpopuloustypeofgalaxy. (2010),whoboth broadlyconcludethatonce‘environment’
Thustheyare bothextremelyimportant in determiningin- has influenced the stellar mass of a galaxy (particularly a
tegrated properties of mass and light in cosmological vol- centralgalaxy),theremainingpropertiesarepredominantly
umes of the Universe, and the most easily studied variety drivenbythestellar mass. This is in alignment with earlier
of galaxy by virtue of the double power-law shape of the work by Baldry et al. (2006), which finds thecolour–stellar
luminosity function (see Cooray & Milosavljevi´c 2005, for mass and colour–concentration locus of galaxies is uncorre-
a discussion on the origin of L∗). By a happy coincidence lated with environment (which in this case is density de-
our own galaxy, the Milky-Way,is also an L∗ galaxy. This fined).
observational bias is not surprising, since it just means our This work putsthe investigation of L∗ galaxies intoan
sunresidesinthemost typicaltypeofgalaxytofindstellar
observational cosmological context by using data from the
mass.
Galaxy and Mass Assembly project (GAMA: Driver et al.
The combination of these factors makes a detailed in- 2011). GAMA is a multi-wavelength photometric and spec-
vestigation of L∗ galaxies an instructive exercise, informing troscopic survey,andis designed toanswer questionsabout
us both about the integrated properties of galaxies in cos- howmatterhasassembledonahugevarietyofscales:galax-
mological volumes, and the typicality of our own galaxy. It ies,groups,clustersandfilaments(i.e.alloftheenvironmen-
is important that we fully understand any statistical biases tal scales outlined above). The first phase of the redshift
present in the Milky-Way since, by virtue of proximity, it surveywasconductedontheAAT(knownasGAMA-I)and
will always be the galaxy we can investigate in the most thesedata are used in thiswork (Driveret al. 2011).
detail, and from which we can derive the most information GAMA aims to describe the physics occurring in dark
about its formation history. matter haloes across the largest mass range ever probed
However, question marks remain over how typical within a single survey,spanning at least 3 orders of magni-
the MW halo is in the context of the Universe and tude in mass: ∼ 1012h−1M⊙ < Mhalo < 1015h−1M⊙. The
how unusual its galaxy occupation statistics are (e.g. GAMA galaxy group catalogue (G3Cv1) was recently cre-
Boylan-Kolchin et al.2009,2010,2011;James & Ivory2011; ated using thefirst 3 years of GAMA-I redshift data and is
Lovell et al. 2012; Tollerud et al. 2011; Weisz et al. 2011; describedindetailinRobotham et al.(2011,R11hereafter).
Robotham et al. 2012). We are set to learn a vast amount This catalogue was constructed using an adaptive Friends-
abouttheMilky-Wayinthecomingdecades.Inthenearfu- of-Friends (FoF) grouping algorithm that was extensively
ture GAIA (Wilkinson et al. 2005) will measure space mo- tested on mock simulations. The G3Cv1 contains LG mass
tions and properties for 2 billion stars in the Local Group haloesouttoz∼0.1, allowing ustobetterunderstandhow
(LG) which includes all known member galaxies as part typicalMilky-Waytypegalaxiesarewithinacosmologically
of the new era of Galactic archeology studies. Amongst significant volumeof theUniverse.
likely discoveries, we will learn about dynamical equilib- This work has been divided into multiple papers. The
rium, or lack of it, for the first time. Building up to these firstpaperinvestigatedthespecificoccurrenceratesofMag-
hugely detailed surveys it is important we discover where ellanic Cloud mass galaxies around Milky-Way mass (L∗)
the Milky-Way fits into the bigger picture. Only then can galaxies—thisisanovelquestioninitsownrightsincemuch
weapply what weknowabout theMilky-Wayto cosmolog- cosmology depends upon the typicality of the Magellanic
ical galaxy formation models. Combining near-fieldcosmol- Cloud satellites of the Milky-Way (Robotham et al. 2012).
ogy (LG scale) and far-field cosmology (redshift surveys) Thissecondpaperinvestigateshowthedefiningcharacteris-
is key to completing the full picture of galaxy formation tics of L∗ galaxies (type, morphology, star formation rates,
(Freeman & Bland-Hawthorn 2002). halomassandstellarmassamongstothers)varywithin the
L∗ galaxies are observed to have extremely wide rang- totalgalaxypopulation,groupedgalaxiesandclosepairsys-
ing properties, inhabiting a regime where we see the most tems.Thiswillanswerquestionsonwhatphysicsaredriving
GAMA: The Life and Times of L∗ Galaxies 3
thecharacteristicsofL∗galaxieslikeourownMW.Thefinal mation of thick disks and pseudo bulges (e.g. Haywood
planned paperwill measuretheluminosity function of MW 2008; Kormendy & Kennicutt 2004, respectively). Orbital
mass halos, revealing how typical the dwarf galaxy popula- properties are also able to be changed more dramati-
tion of the MW is. cally and violently during encounters with other galaxies
There has been much written in the literature about (Barnes & Hernquist 1992a). Minor merger eventscan pro-
the different physical processes that have a measurable ef- duce subtle changes to stellar orbits, providing another
fect on the properties of a galaxy. Recent key observa- mechanism for the creation of a thick disk (Read et al.
tional discussions regarding the influence of environment 2008).Majormergerscanbeincoherentenoughtotransform
on the properties of galaxies include Baldry et al. (2006); a galaxy with regular rotational disk-like stellar structure
Weinmann et al. (2006); Peng et al. (2010); Thomas et al. (thecanonical late-typegalaxy) intoasystem with random
(2010); Kauffmann et al. (2012). The complicating factor orbits (thecanonical early-typegalaxy).
hasalwaysbeeninseparatingouttheclearprimaryrolethat Self-evidently‘secular’processesarenotadirectconse-
localenvironmenthasinproducingaparticulardistribution quence of environment since they describe processes that
ofgalaxies (e.g.luminosityorstellarmass:Robotham et al. should occur regardless of outside influence / triggering,
2006; Weinmann et al. 2006) and second order processes therefore they might be expected to be entirely predictable
thataffectpropertiesofthesegalaxies(e.g.colourandmor- from theinstantaneouspropertiesofagalaxy.However,en-
phology: Driveret al. 2006; Kelvin et al. 2012). vironment can play a role in disrupting secular processes,
Themajorrolesthatagroup(orhalo)hasinassembling e.g. a major merger would destroy any bar formation pro-
stellarmassare:feedinggasintogalaxiestoenhancestarfor- cess. Clearly thereis scope for environment to play at least
mation(Keres et al.2005;Dekel & Birnboim2006);heating a secondary role in almost all aspects of galaxy evolution,
and stripping gas out of galaxies (e.g. Gunn & Gott 1972; if only by virtue of disruption, delay and prevention mech-
Abadiet al. 1999; Balogh et al. 2000; Dekel& Birnboim anisms. Thepurposeof thispaperis tosimultaneously sep-
2006) or stripping of stars directly (Bekkiet al. 2001) and arate out the stellar mass, galaxy-galaxy interaction, and
providing a location for rapid galaxy-galaxy interactions halo mass terms in order to quantify the codependencies of
(Barnes & Hernquist 1992a), be they harassment, minor or secularlyandenvironmentallydrivenstellarmassevolution.
major merger processes (e.g. Moore et al. 1996; Cole et al. Secondarypropertiesofgalaxiesclearlydisplaydiffering
2000, respectively). All of these processes provide mecha- sensitivitytomodifyingexternaldrivers.E.g.Hαequivalent
nismsforstellarmassbuild-upingalaxiestobeeitheraccel- width (EW) is arelatively instantaneous response toan in-
eratedordecelerated,andwillhavenoisyaverageneteffects crease in star formation, accompanied by a change in inte-
as a function of stellar mass, halo mass and epoch. Except grated bandpasscolour to bluercolours. In the longer term
for the tidal dwarf variety of galaxies (Barnes & Hernquist thein-situgasandlocked-upstellarmetallictywillincrease,
1992b) allgalaxies areassumed tobelocated within adark which will create a redder SED. If this burst of star forma-
matterhaloofsomedescription,thoughtodatenorotation tion occurred due to a minor merger event then we might
curve has been measured for a ‘tidal dwarf’ that does not also expect orbitstobecome disrupted,leadingto achange
require some ‘dark’ component (Bournaud et al. 2007). We in the velocity dispersion and light profile of the galaxy. If
can therefore say that galaxy stellar mass is entirely cou- thechangeisdramaticenough(e.g.amajormerger)itmight
pled to environment, which is not to say the coupling is alsoinvokeachangeinthe‘type’classificationofthegalaxy.
trivial. The fact it is not trivial has invited a large variety Thus we have galaxy properties that are highly sensitive
offeedbackmodelsintheliterature(e.g.Benson et al.2003; to current star formation invocations (Hα-EW) and ones
Bower et al.2006;Croton et al.2006).Beyondthis,itisless that betray more about the integrated history of a galaxy
clear-cut what additional role the environment has on the (some component of colour and all aspects of morphology).
observed properties of galaxies. Depending on the physical processes that caused star for-
The second order properties of galaxies (those things mation we might also observe longer term changes to the
that are not stellar mass) can be largely divided into two morphology of the galaxy (in the case of minor or major
categories: luminous energy characteristics (any aspect of a merger events). In this work we aim to separate out these
galaxy defined by light output as a function of wavelength: differentsecondarypropertiesintothosemostdependenton
starformationrate(SFR)andgalaxycolouretc)andorbital different likely primary drivers of galaxy properties: stellar
energy characteristics (aspects concerning the position and mass, halo mass and galaxy-galaxy interactions.
velocityofstars:morphologyandcolourgradientsetc).Pro- In Section 2 we discuss the data used in this work in
cessesthatcanaffecttheluminousenergycharacteristicsare detail.InSection3wediscussthesampleselectionsapplied
mostly those that directly enhance or suppress star forma- to the data. In Section 4 we describe the visual classifica-
tion (which is then necessarily coupled to the stellar mass tion process that was used to categorise the morphological
build-up discussed above) and more contentiously any pro- type,emission class and presence of a disk. In Section 5 we
cess that affects the initial mass function (IMF) that stars discuss the different varieties of galaxy categorisation used
are forming with (see Gunawardhana et al. 2011, for recent for constructing later plots. In Section 6 we analyse how
work on variable IMFs). properties of MW mass galaxies are affected both by the
Orbital properties within galaxies can evolve natu- group-scaleandpair-scaleenvironment,wherethecodepen-
rally through ‘secular’ processes that allow for orbital en- dence of galaxy properties is discussed in detail. In Section
ergy redistribution, which we observe as brief and vary- 7wepulltogetherthemainresultsofSection6intoacoher-
ing manifestations such as pattern bars and spiral arms ent narrative and put the findings into context. In Section
(e.g. Sellwood & Wilkinson 1993; Bertin et al. 1989, re- 8 we briefly summarise themajor findingsof thework.
spectively), or permanent transitions such as the for- In this work we use ΩM = 0.25, ΩΛ = 0.75 and H0 =
4 A.S.G. Robotham et al.
70kms−1.Thisisconsistentwiththefirstpaperinthisseries correction for the majority of our work. The fidelity of the
that dealt specifically with the commonality of Magellanic spectral classifications is investigated below.
Cloud galaxies around L∗ systems (Robotham et al. 2012).
2.3 G3Cv1
2 DATA One of the principal science goals of GAMA is to make a
statistically significant analysis of low mass groups (M 6
2.1 GAMA
1013M⊙), helping to constrain the low mass regime of the
The Galaxy and Mass Assembly project (GAMA) is a ma- dark matter halo mass function and galaxy formation ef-
jor new multi-wavelength photometric and spectroscopic ficiency in Local Group like haloes. This low mass regime
galaxy survey (Driver et al. 2011). The final redshift sur- contains groups that have halo masses similar to the LG,
vey will contain ∼300,000 redshifts to rAB = 19.8 over allowingustoinvestigatehowcommontheLGenvironment
∼280 deg2,withasurveydesignaimedatprovidinganex- is, and puttingLGscience into a cosmological context.
ceptionally uniform spatial completeness (Robotham et al. The GAMA-I spectroscopic data was used to con-
2010a; Baldry et al. 2010; Driveret al. 2011). struct the first GAMA galaxy group catalogue (G3Cv1:
Extensive details of the GAMA survey characteristics Robotham et al. 2011). This group catalogue was con-
are given in Driveret al. (2011), with thesurvey inputcat- structed using an adaptive Friends-of-Friends (FoF) algo-
alogue described in Baldry et al. (2010) and the spectro- rithmthatwasoptimisedtorecovergroupsinthemagnitude
scopictilingalgorithminRobotham et al.(2010a).Thefirst limited GAMA-I spectroscopic data.
3 years of data obtained has been frozen for internal team Intotal G3Cv1contains14,388 groups(with multiplic-
use and is referred toas GAMA-I. ity>2),including44,186 galaxies out of apossible 110,192
Briefly, the GAMA-I survey covers three regions each galaxies, implying ∼40% of all galaxies are assigned to a
12×4degreescentredat09h,12hand14h30m(respectively group.SinceGAMA-Ihasanunevenr-bandselection func-
G09,G12andG15fromhere).Thesurveydepthsandareas tionbetweenGAMAregionsweuseanr<19.4magsubset
relevant for this study are: ∼ 96 deg2 to rAB = 19.4 mag of the full G3Cv1 for this work. This catalogue has 37,576
(G09 and G15) and ∼ 47 deg2 to rAB = 19.8 (G12)1. All galaxies in 12,200 groups.
regionsaremorethan98%complete(seeDriveret al.2011, The G3Cv1 contains a wealth of ancillary data regard-
for precise completeness details), with special emphasis on ingthepropertiesofthegroups,includingdynamicallyesti-
a high close pair completeness, which is greater than 95% mated halo masses. These halo masses are used extensively
for all galaxies with up to 5 neighbours within 40′′ of them in this work. For details regarding the reliability and bi-
(see Fig. 19 of Driveret al. 2011). Stellar masses used for ases the reader should refer to Robotham et al. (2011) and
this work are taken from Taylor et al. (2011). Alpaslan et al. (2012).
2.2 Spectral Analysis 2.4 Pair Catalogue
The spectra collected during the GAMA observations have This paper is investigating the effect that various kinds of
been flux calibrated, Balmer decrement corrected, aperture environmenthaveonthecharacteristicsofL∗galaxies.Close
corrected,andtheemissionlineequivalentwidthsmeasured. galaxy interactions have a significant role on the evolution
The spectral pipeline process is discussed in detail in Hop- of galaxies, so these systems are selected for close analy-
kins et al. 2013 (in prep). The line measurements havealso sis in this work. Part of the process of creating the G3Cv1
been converted into star formation rates, a process that is catalogue involves the construction of all galaxy pairs (see
described in Gunawardhana et al. 2013 (in prep). These Robotham et al. 2011, for details). This catalogue contains
SFRsare used directly in this work. galaxies with potentially quite large radial (velocity) and
Theseemission linemeasurementsarealsousedforau- tangential (spatial distance) separations between galaxies.
tomatic spectral classifications bycomparing the[OIII]/Hβ Usingthefullpaircataloguewouldincludegalaxypairswith
as a function of [NII]/Hα in the popular manner outlined very large dynamical times, and such pairs will have weak
in Baldwin et al. (1981), widely known as BPT classifica- evolutionary effects on the component galaxies. Instead we
tion. This process, and the classifications selections used, select a narrow window of interaction phase space in order
are discussed in detail in Hopkins et al. 2013 (in prep) and to preferentially extract galaxies that will be most affected
Gunawardhana et al. 2013 (in prep). These automatic clas- by close galaxy-galaxy interactions. The ‘close pair’ sample
sifications are supplemented by extensive visual checks to selected hereisbased on thatpresentedin Robotham et al.
minimise classification errors (discussed below). (2012), where we aimed to recover galaxy pairs that are
For this work we keep the classifications as simple as similar to the MW Magellanic Clouds system. This ensures
possible: Hα emission that is dominated by star formation wecaninvestigateavarietyofinteractionthatourownMW
irradiation, Hα emission that is dominated by AGN irradi- L∗galaxyhasexperienced,helpingtoplaceitinthebroader
ation and no Hα emission. This binary definitions of star family of L∗ galaxies.
formation means we are highly insensitive to the details of The SMC is ∼ 60kpc from the MW and travelling
flux calibration, Balmer decrement correction and aperture radially away at ∼ 17kms−1 and 301kms−1 tangentially
with respect to the MW: a net velocity of 302kms−1
(Nichols et al. 2011). The LMC is ∼ 50kpc from the MW
1 SeeBaldryetal.(2010)foradditional GAMA-Iselections. and travelling radially away at ∼ 89kms−1 radially and
GAMA: The Life and Times of L∗ Galaxies 5
Galaxiesinpairedsystems Frequency Totalgalaxies etiesof‘environment’investigatedinthiswork:galaxypairs
and galaxy groups. These are just two of an infinite num-
2 4,550 9,100
ber of possible definitions of environment, but they probe
3 370 1,110
4 45 180 the smallest physical scale environments possible, and as
5 4 20 such should possess the cleanest possible relationship be-
6 1 6 tween environment and galaxy properties due to the recent
dynamical events. To link this work closely with our own
Table1.Numberofpairsystemswithdifferentnumbersofgalax- MW environment (important since so much effort is being
ies. invested in surveys of the MW and local galaxies), we are
careful to ensure that our definitions of ‘pair’ fully capture
367kms−1 tangentially with respect to the MW: a net ve- interactions like the MW is undergoing with the LMC (its
locity of 378kms−1 (Nichols et al. 2011). To conservatively dominant pairwise interaction). We also select all groups
thataresimilar masstotheMilky-Wayhalo,allowing usto
recover all systems where the galaxies are in such spatial–
investigate how common the MW system is, given its halo
velocity proximity, we create a catalogue of pairs for this
mass as a prior.
work where the projected separation is rsep−proj < 70kpc
and the radial velocity separation is vsep−rad <400kms−1. First we add our pair selection criteria. We include all
pairs within a redshift limit of z =0.089 that have at least
SeeRobotham et al.(2012)fordetaileddiscussionongalaxy
one member selected in the mass sample described above.
pair selections.
Thisallows ustoobserveLMCequivalentgalaxiesthrough-
Applying these selection limits to recover all systems
out our sample (see Robotham et al. 2012, for details of
thathavesimilar pairwise dynamicalpropertiestotheMW
theselectionlimits).Togetherwiththelowerredshiftlimitof
and the Magellanic Clouds creates a catalogue containing
z=0.01thisgivesasampleof3,503galaxies.Inthisworkwe
5,964 galaxy pairs and 10,416 unique galaxies. These num-
use stellar mass rather than r-band magnitude to describe
bers are for the whole GAMA-I dataset, so include stellar
the galaxies, and the redshift limit z =0.089 is specifically
massesthatmightbehighlydissimilartotherealMWMag-
the implied r-band magnitude limit to observe LMC type
ellanic system.
galaxies.However,sincetheLMCisblue,late-typeandstar
Some galaxies will be members of more than one pair
formingthiswillactasaveryconservativeLMCstellarmass
(theMWispairedwithtwogalaxies—theSMCandLMC).
Wecreate groupingsthat contain all galaxy-galaxy associa- limit.IndeedtheMs >MS,LMC numbercountsdropoffat
a z ∼ 0.1. For consistency with Robotham et al. (2012) we
tions and count this as a single ‘pair’ system, e.g. the MW
use the more conservative limit of z =0.089 to select down
is in a pair system of 3, as are the SMC and LMC. In the
to LMC mass galaxies.
same way that simple galaxy pairs are only counted once,
In order to determine what role different environments
higherorderpairsystems arealso only countedonce. Table
may have on the evolution of L∗ systems, we then select
1 shows the frequency of different pair systems of galaxies.
all galaxies that are in any G3Cv1 group or galaxy-galaxy
Whilstwerecover5,964galaxypairs,oncecommonlinksare
pair containing any of the 3,503 galaxies in the above sub-
explored we are left with 5,446 systems.
set.Thisgivesusanexpandedsampleof5,054galaxies. We
add to this all galaxies in all groups that have a G3Cv1
halo mass between 7×1011M⊙ <MFoF < 1013M⊙ where
3 SAMPLE SELECTION
MFoF is defined using thefunctional form of thehalo mass
Alargecomponentoftheanalysisrequiredforthispaperre- usedinRobotham et al.(2011).Thishalo massrangecom-
quired comprehensive visual classification (described in de- fortably covers both sides of the MW halo mass used in
tail below). The task of visually classifying the data in this Robotham et al. (2012, 2.5 × 1012M⊙) and the total LG
mannerwasextremelylabourintensive—eachgalaxycould mass(MW+M31haloes=∼5×1012M⊙:Li & White2008).
take a minute to analyse and GAMA-I contains 129,765 Including this halo mass range guarantees that all systems
galaxies with redshifts above 0.01. Doing this for 8 hours withpropertiessimilartotheMWbyvirtueofstellarmass,
a day would take 270 days. Instead we selected a subset of and halo mass or close pair interactions, are certain to be
galaxies concentrated on L∗ systems (the focus of this pa- includedinoursample.Thisgivesusafinalsampleof6,150
per), and systems that share common properties with the galaxies to visually classify in detail. As long as the stel-
MW and its local environment (allowing us to identifying lar mass and halo mass ranges are kept within the limits
theevolutionary drivers for theMW). described above,this sample is volumelimited.
Sincethisworkispredominantlyinterestedintheprop- The result is our sample has two major components:
erties of MW mass (L∗) galaxies we had to comfortably in- • AllGAMAgalaxieswithMS >1010M⊙(masssample,
clude all GAMA galaxies with stellar mass similar to the
with 3,550 galaxies).
MW(5.0×1010M⊙:Flynn et al.2006).Aconservativelower • Within 0.01 < z < 0.089, all pairs and groups are in-
stellarmasslimitofMS =1010M⊙ waschosen.Duetothe cluded that have at least one member in the above mass
rapiddecreaseinthenumberofobjectsaswemoveintothe
selection. These two varieties of environment are therefore
higherstellarmassregimetherewasnoneedtoimplementa completedowntoLMCmassgalaxies.InadditionallG3Cv1
higherstellarmasscut-off.Intotalthisstellarmassselected groupswithahalomassintherange7×1011M⊙ <MFoF <
sampSlienccoenwtaeianrse3a,l5s5o0ingvaelasxtiigeas.tingtheenvironmentthatL∗ 1013M⊙ and 0.01 < z < 0.089 (volume sample, with 6,150
galaxies).
galaxies reside in, we add all galaxies that are associated
with the L∗ selection to the sample. There are two vari- Plots that only require a mass selection are defined by
6 A.S.G. Robotham et al.
the mass sample. Any plot the makes use of pair or group but distinct disks. Whilst late-type galaxies, by their def-
information is defined by the second volume limited selec- inition, always possess a disk like structure, there exists a
tion. By using this combination of stellar mass, group and large fraction of early-type galaxies where this might be a
pairselectionsweareabletoinvestigateallofthekeychar- keydistinguishingvariable(weincludeS0andlenticularsin
acteristics that definetheMW and theLG environment. thisclass).
To determine classes for these three categories a large
subset of the GAMA-I survey was studied in detail by
ASGR.Usingthesingleobjectviewer(SOV)tooldeveloped
4 VISUAL INSPECTION OF THE DATA
by JL it was possible to view the following information si-
Inthisworkweareprimarilyinterestedinhowthepresence multaneously for each galaxy:
ofagroupenvironment,oranothernearbygalaxy,canaffect • colour image;
the characteristics of a given stellar mass galaxy; particu- • basic catalogue information;
larly in theregime of MW mass galaxies (i.e. L∗ galaxies). • preferred spectrum;
Toundertakesuchanalysis it isnecessary toknowvar- • 2D modelling information, including difference image;
ious characteristics about L∗ mass galaxies (i.e. thekneeof • 1D surface brightness profile.
the stellar mass function, M∗). Some of these characteris-
ticscanbedetermineddirectlyfromthedata:colour,mass. Figure 1 is a screen grab of the SOV tool for GAMA
Othersaremoredifficulttoextractfromthedataduetothe galaxy 6926. Using the information available in the SOV
subjectivenatureofthevariableortheimperfect quality of ASGRassessed the following 3 galaxy characteristics:
thedata.
• whetherthegalaxyislate-type(visiblydiskdominated)
Galaxy type,bywhich wemean morphologically early-
or early-type(visibly bulge dominated);
typeorlate-type,isdifficulttodeterminefromthecatalogue
• whetherthegalaxy hasanyevidencefor adisk (binary
data directly. When analysing a fiducial sample of galax-
trueor false);
iestakenfrom Kelvin et al.(2012)therelationship between
• whether the galaxy exhibits star forming Hα emission,
S´ersic index and morphological type is highly imperfect: it
no Hα emission or is an obvious AGN (broad emission fea-
isnottruetosaythathighS´ersicindex(n>3)galaxiesare
tures).
early-type and low S´ersic index (n < 3) galaxies are late-
type. Inclination in particular can distort catalogue values, Thethreeclassificationsusedtoquantifyemissionclass
and the reliability of colour as a galaxy ‘type’ discriminant are shown in Figure 2. As well as assessing galaxies for
is heavily compromised by thepresence of dust. these different classes, data quality was simultaneously
Starformationrate(orspecificstarformationrate)ap- checked.Galaxieswereassessedastowhethertheyweresub-
pearstobeacharacteristicthatiswellsuitedtocatalogued structureoflargergalaxiesormisclassifiedstarsthathavea
data. However, analysis of a fiducial sample revealed that redshift virtue of background galaxy flux. 16 galaxies were
the current emission line based SFRs (Hopkins et al. in re-classifiedassubstructureand6galaxieswerere-classified
prep) fail to detect SFRs for ∼20% of galaxies where the as stars, i.e. 99.64% of galaxies in our original sample were
emission line is very weak and the spectrum has low S/N. correctlyidentifiedasbeinggalaxies(seeBaldry et al.2010,
The lack of a measurement is not an error, rather a prag- for details of the original GAMA star galaxy separation).
matic decision to improvethereliability of quoted SFRsby Finer resolution grading of properties was not possi-
ignoring marginal lines. A particular problem is that these bleovertheredshiftrangeprobed,i.e.itwasnotpossibleto
undetectedweaklystarforminggalaxiesaretypicallyfainter reliablyidentifytheHubbleclassorsimilarlowerorderclas-
lowermasssystems,creatingastellarmassbiasinouranal- sifications.Thepropertiesassessedweresubjectivelyconsid-
ysisofstarformation. Also,at presentGAMAincorporates ered to be the most information that could be robustly ex-
spectra from non-AAOmega sources (e.g. SDSS, 2dFGRS) tractedgiventhequalityofdata.Toguidelaterdiscussions,
andemissionlineanalysishasnotyetbeenrunontheentire lenticulargalaxiesandearly-typespiralgalaxieswouldboth
sample. This has a larger effect at lower redshifts, which is beclassified as early-typegalaxies possessing disks.
the regime of interest in this work, since a larger fraction
ofspectraarenon-AAOmegaforbrightermagnitudes.Ulti-
mately,independentlyidentifyingspectralemissionfeatures
5 GALAXY CATEGORISATION
visually was deemedtobeanappropriate routeforward for
the science investigated in this paper, where we are mostly For multi-dependent data mosaic plots are a very power-
interested in the binary divide of star formation: on or off. ful and information rich means of visualisation. They differ
Thisrouteofinvestigationisinformedbyrecentstudies(e.g. fromprincipalcomponentanalysis(PCA)inthattheyread-
Wijesinghe et al. 2012) showing the star forming fraction, ily handle discrete data classes, which PCA does not. For
not the rate of star formation, is the property sensitive to readers unfamiliar with how mosaic plots are constructed,
‘environment’. This requires us to push the usable S/N and best interpreted, AppendixA describes them in detail.
limits of the catalogues, which is discussed in more detail Itisstronglyrecommendedthatnovicestothemosaicplot-
below. ting technique start there. To ease the comprehension of
The presence, or otherwise, of a disk is also a key thelargenumberofplotspresented,themajorobservations
characteristic that is difficult to extract from catalogued drawnfromeachfigureareincludedintherelevantcaptions.
data. Using the data−model residual images produced by Thisshouldbeparticularlyhelpfulinexplainingthemosaic
Kelvin et al.(2012)itwasclearthatalargenumberofearly- plots to those unfamiliar with them.
type galaxies, even with high S´ersic indices, possess weak In order to present information in the form of mosaic
GAMA: The Life and Times of L∗ Galaxies 7
Figure2. Examplesofthethreeemissionclassesusedforvisual
classification. Top panel shows broad line features due to high
velocity emission from an AGN (class A). Middle panel shows
highequivalent width Hα emissionrelative to all other emission
lines, so a star formingspectrum (class S). Bottom panel shows
nosignsofanyHαemissionontopofthecontinuum (X).
plots all information must be discretised. Some categories
are naturally discrete: morphological type and presence of
Figure 1. Screen grab of a typical display using the SOV tool disk.Othershavetobecoercedintodiscretecategories:SFR
available for the GAMA database. Top-left is an image taken (as outlined above). Here we describe all of the galaxy cat-
directlyfromtheSDSSimageserver.Top-rightaresomekeycat- egorisations that are used to construct later mosaic plots,
aloguevaluesforthecurrentgalaxy.Inthemiddleisthepreferred alongwiththeabbreviationinparenthesesusedforlabelling.
GAMAspectrumwithpositionsofemissionandabsorptionlines
over-plotted. Directlybelowthespectrumisalistofother avail-
ablespectraintheGAMAdatabase (usefulwhentherearedata 5.1 Naturally Discrete Galaxy Categories
quality issues in key regions of the preferred spectrum). In this
example the only available spectrum is from 2dFGRS. Bottom- Belowarelistedallofthenaturallydiscreteclassesofgalaxy
leftistheimageoutputofKelvinetal.(2012).Withinthispanel propertiesinvestigated.Thedeterminationofthesedivisions
thetop-leftistheutoK-bandimage(r-bandisthedefault,but requirelesssubjectivity thanforcontinuousdata.However,
others can be selected by clicking on the links above), top-right not all categories are fully objective— opinion is still re-
is the final Galfit model, bottom-left shows the ellipses used to quired to determine whether a galaxy is early-type or late-
construct the 1D surface brightness profile, bottom-right is the type.
data−modelresidualimagewiththelocalPSFoverplotted.The
bottom-rightofthewholefigureisthe1Dsurfacebrightnesspro-
fileestimated usingIRAFEllipse,withkey values regardingthe 5.1.1 Galaxy Type
fitprintedinsidetheplotwindow.
Early-Type(e) / Late-Type(l)
As part of the visual classification process outlined in
Section 4, galaxies are categorised by simple Hubble type.
The divide used was whether the galaxy appeared to be
bulge dominated (Re inside the bulge regime, which is
determinedbya strongvisual breakin the1D light profile)
8 A.S.G. Robotham et al.
ordiskdominated (eithernobulge,orRe outsidethebulge 5.2 Naturally Continuous Galaxy Categories
regime). This estimation is informed by both multi-band
Below are listed all of the naturally continuous classes of
and r-band SDSS images, and the 1D profile of 2D single
galaxy properties investigated. The determination of these
S´ersic fitsgenerated byLSK (seeKelvin et al. 2012, for de-
divisionscontainsomesubjectivityonourpart.Makingcon-
tails on thesingle S´ersic fittingprocess). Furtherresolution
tinuousdatadiscreteobviouslyreducestheinformationcon-
beyond early-type and late-type is not attempted due to
tent available to us, but in practice it is often theonly way
the quality of the data. Lenticulars (S0) galaxies would be
to visualise behaviour of noisy relationships: it is common
classified as early-type in this scheme, as would early-type
practice to plot the running median in scatter plots, which
spiral galaxies.
isadiscreteseparationofthedata.Evenbinningdatawhen
plotting a luminosity function is a discretising process. Ob-
viously any further numerical analysis should make use of
5.1.2 Galaxy Disk
theunderlyingcontinuousdata.
False (F) / True(T)
As part of the visual classification process outlined in 5.2.1 Emission Strength
Section 4, galaxies are assessed to either possess some form
AGN (A) / StarForming (S) / No Emission (X)
of disk, or not. This subjective observation is based on the
r-band S´ersic profile subtracted residual images created by
As part of the visual classification process outlined in
LSK (see Kelvin et al. 2012, for details on the single S´ersic
Section 4 galaxies are visually assessed into these three
fitting process). Further detail beyond this binary status is
classes of emission. AGN classes (A in plots) are those
not attempted due to the quality of the data. Lenticulars
with clear Hα emission, but where the spectral signature
(S0)galaxies would beclassified aspossessing adisk inthis
is dominated by AGN irradiation. Star forming classes
scheme.
(S in plots) are those with clear Hα emission, but where
the spectral signature is dominated by star formation
irradiation. No Emission classes (X in plots) are those
5.1.3 Galaxy in Group without any Hα emission, though other lines may still be
present (typically weak [NII]).
False (F) / True(T)
Figure 2 gives an example of each. Galaxies were
both spectrally classified through a classic BPT separation
Logic class defining whether a galaxy is found within
(Baldwin et al.1981)andbyeye.TheGAMAapplicationof
aG3Cv1groupthatmakestheselection criteriaoutlinedin
theBPTseparation isdescribedindetailinGunawardhana
Sec3.
etal.2013(inprep)andHopkinset al.(2013),andincludes
all GAMA and SDSS origin Spectra (which covers 92.2%
of the sample investigated here). In the automatic classifi-
5.1.4 Galaxy is Central or Satellite cation scheme a galaxy is classed as ‘X’ when it cannot be
placedonaBPT,i.e.itismissingatleastoneoftheHα,Hβ,
Central Galaxy (C) / Satellite Galaxy (S)
[OIII] or [NII] lines. This differs to the definition desired in
thiswork, that ‘X’ refers only to systems with no Hαemis-
Class defining whether a galaxy is the largest (in terms of
sion.Byusingthisdefinitiontheseparationbetweenclasses
stellarmass) in itsgroup,otherwisereferred tointhiswork
is much clearer.
asthebiggestgroupgalaxy(BGG).Ifitisthenitisdefined
Separatelytothispurelyautomaticprocess,allspectra
as the ‘central’ galaxy (C), otherwise it is a ‘satellite’ (S).
were classified by eye. It is easy to identify AGN spectra
All ungrouped galaxies are defined as central galaxies in
when there are broad emission line features, but these are
their group of multiplicity 1 (but these 1 member groups
do not count as part of theG3Cv1). a minority of cases. In other cases the decision was made
by dividing the height of the [NII] and Hα features, and
classifying the galaxy as AGN when this ratio was clearly
above 0.5. This uses part of the information available in
5.1.5 Galaxy in Pair System
a formal BPT analysis, and will be approximate at best.
False (F) / True(T) As such, a clear weakness of visually classifying spectra is
separating marginal BPT classifications. All disagreements
Logic class defining whether a galaxy found within a were followed up at least once and a discernment made be-
G3Cv1 pair system that makes the selection criteria tweencaseswheretheautomaticprocesswasdistinguishing
outlined in Sec 3. between subtle emission ratios that cannot be accurately
determined byeye,and those wherean fittingerrorhas oc-
curred,creating a poor automated fit.
In cases where the BPT was able to select marginal
5.1.6 Galaxy is Largest in Pair System
AGN systems and Hα, Hβ, [NII] and [OIII] were all in
False (F) / True(T) emission, the BPT was preferred. However, the majority of
BPTselectedAGNclassificationsweretenuouslowpowered
Logic class defining whether a galaxy is the largest LINER-like systems with no significant Hα, Hβ or [OIII]
(in terms of stellar mass) in its pair system. flux. In these situations only a weak [NII] line was present
GAMA: The Life and Times of L∗ Galaxies 9
Visual Spectral Class 10-3 10-2 10-1 100 101 102 103
A S X
AGN (A)
A 154 4 644 1 Star Forming (SF) 1
Non Star Forming (X)
8 8
0. 0.
6 6
F 0. 0.
D
P
4 4
0. 0.
T ClassS8 3298 415 0.2 0.2
P
B
o
Aut 0 0
10-3 10-2 10-1 100 101 102 103
Ha −EW
Figure 4. Comparison of the Hα-EW PDFs for the AGN (A),
star forming (S) and non star forming (X) classifications used
X 68 229 818
in this work, where the Hα emission has been measured during
the GAMA line measuring process (see Hopkinsetal. 2013, for
details). As expected, where Hα has been measured the EW is
NA 8 351 119 significantlyhigherforstarformingcomparedtononstarforming
galaxies, with AGN classed galaxies sitting broadly between the
two.Toaidcomparisonwithother workwesuggestdividingthe
Hα-EW at 1 (vertical dashed black line) and classing galaxies
Figure 3. Mosaic plot comparing the automated BPT classifi- abovethisthresholdasAGN/starformingandclassinggalaxies
cations (horizontal bands) versus the visual classifications (ver- belowthisthresholdasnonstarforming.
tical coloured bands). For both classifications A=AGN, S= star
formingand X= non star forming.For the visual classifications,
green= AGN,blue= star formingand lightred= non star form-
ing. The most disagreement comes for AGN classes (A) where
Figure4demonstratestheHα-EWdistributionsforthe
verylowpoweredLINER-likesystemswithnovisibleHαarebe-
ingclassedasAGN.SincewearemostinterestedinHαemission, final classifications used in this work. Star forming galaxies
and only wish to exclude cases where a large component of the have a typical Hα-EW at least a factor 10 and on aver-
HαfluxisduetoAGNactivity,theseobjectsaremostlymovedto age a factor 100 larger than the visually identified non star
thenoemissionclass(X).Spectratakenfromsurveysotherthan forminggalaxies. Themajority ofnonstarforminggalaxies
GAMA and SDSS do not have an auto classification (shown as donothaveanymeasuredHαflux(83.8%).Thisplotshows
NA). For adetailed descriptionof how to interpret mosaicplots thePDFsforthesmallfractionthathavesomefluxdetected
thereadershouldrefertotheAppendix.
duringtheautomatedlinemeasuringconductedforGAMA
Observations:
(see Hopkinset al. 2013, for details). An approximate ver-
-Strongagreement for‘S’classemissions.
sion of ourclassifications can bemadeusingHα-EW alone:
-LargefractionofautoAGNclasseshavenoHαemission.
97.6% of galaxies that have an Hα-EW above 1 are either
AGN or star forming, and 93.3% of galaxies with Hα-EW
(as in the bottom panel of Figure 2), and the classifica- below 1 (including non measurable) are non star forming.
tion was changed to ‘X’, since there is no strong Hα flux. Using Hα-EW alone does not allow for a clean separation
This is the main source of disagreement revealed duringvi- of AGN and star forming galaxies (hence the popularity of
sualclassificationofspectra.Itisimportanttochangethese BPTdiagramsandsimilarschemes),butthesedivisionswill
classificationssinceasignificantnumberofAGNare,bythe allow some of the more dramatic results to becompared to
definitions we wish to use, erroneous. otherwork.
Figure 3 summarises the BPT versus visual classifica- The AGN classifications contain within them complex
tion. Overall 75.4% of automated BPT classifications agree subclasses that we do not attempt to resolve in this work:
with the visual classifications. For AGNs the figure drops i.e. QSO,Seyferttype1and 2,and strongLINERgalaxies.
to 19% for the reasons described above (only exceptionally Since the physical origin of these different classes are quite
obvious AGN can be reliably picked out by eye). For star different and some of the variation is due to observational
forming systemstheagreement ishigh,at88%. Finally,the bias(e.g.distinctionbetweenSeyfertclassesislargelyobser-
non emission classes agree at the 73% level. In addition, a vationalorientation),wewillfocusthisworkonthestarfor-
non-negligible (7.8%) fraction of sources do not have SDSS mation (S) and noemission (X) classes. However,theAGN
orGAMAspectra,thereforetheydonothaveaBPTclassi- classes will be listed in all mosaics in order to accurately
fication in the GAMA database (see bottom panels of Fig- representthefullpopulationfractions.Inourfinalcategori-
ure3).Thesesystemshadtobeclassed entirelybyeye,but sations,3.9% ofgalaxiesareAGNwith Hαemission, 63.1%
broadlyreflecttheoverallclassificationdistributionsseenfor havestarformingHαemission,and32.5%havenoHαemis-
the full sample. The main survey source for these galaxies sion. The final 0.5% of objects are not real galaxies, either
without an automated BPT classification is 2dFGRS. stars or substructureof extendedgalaxies.
10 A.S.G. Robotham et al.
5.2.2 Colour
0.4 0.6 0.8 1
Blue (b) / Green (g) / Red (r) Ms» Ms-MW
0 0
0 0
Colour is an important characteristic of a galaxy. Here 0 0
3 3
we treat colour as a purely relative quantity, and we use
g − r because we have detections in both bands for all 0 0
galaxies. It is known that the bimodal nature of galaxy 0.1 200 200
d
colour is enhanced by using a bluer band than g since N/
this only partly samples blue-wards of the 4000 ˚A break 0 0
0 0
(Baldry et al. 2004), but since colour does not form the 10 10
main focus of future analysis, completeness was preferred.
Also, since we dividecolour intorelative quantiles, thelack blue green red
ofg−r bimodalityshouldnotdramatically alterlaterplots 0 0
since bluer galaxies do tend to be bluer in both X−r and 0.4 0.6 0.8 1
g−r,whereX isanybandbluerthang.TheGalactic dust g−r
corrected and k-corrected (using k-correct and GAMA’s
Figure 5. The distribution of g−r colours for galaxies with
ugrizYJHKphotometry,Blanton & Roweis2007;Hill et al.
2011, respectively) g − r colour is separated into three stellarmasseswithin0.15dexofMS,MW.Theblue/green/red
divisionsusedareshownbythedifferentcolouredshading.Selec-
equal sized quantiles of data for galaxies within 0.15 dex
tionshownisthemasssample(Sec3).
of MS,MW. The limits of the quantiles are used to define Observations:
galaxies as blue,green or red in a relative sense: - There is no clear bimodality in the g−r colour for the MW
stellarmassselectionusedtodefinethecolourquantiles.
-Thebluepopulationselectsthemajorityoftheoutlyingwingin
g−r<0.77 = Blue(b) theg−rdistribution,whilstthegreenandredpopulationsample
0.77<g−r<0.86 = Green(g) different sides of the main population peaks. This suggests that
the green and red selections are likely to be selecting galaxies
g−r>0.86 = Red(r)
drivenbysimilarphysicalprocesses.
Figure 5 shows thedistribution of g−r colours for the
The discrete halo mass bins are defined in a similar vein
massselectionused,andhighlightsthedivisionsused.There
to the stellar mass bins: adjacent bins vary in mass by a
isnoclearbimodalityforthismassselectionandcolourcom-
factor two, and each bin therefore contains a mass range
bination,althoughthebluegalaxiesexistinadistinctwing.
of ±0.15 dex. Halo masses are taken from Robotham et al.
The green and red galaxies are largely selecting different
(2011),andarethevariablyscaled(withredshiftandgroup
sides of thedominant colour distribution peak.
multiplicity) dynamical masses that are discussed in that
work. MH,MW uses the halo mass of the MW calculated in
5.2.3 Stellar Mass Li & White (2008): MH,MW = 2.5×1012M⊙. The Y = 1
bin corresponds to haloes with MW and M31 masses. The
X×MS,MW (1) Y = 2 bin corresponds to groups that have the same halo
massasthewholeLGsystem(5.0×1012M⊙).Halomasses
We split GAMA galaxy stellar masses (taken from are 100% complete for groups where 1 6Y 62.
Taylor et al. 2011) into bins separated by factors of two, 2
and each bin therefore contains a mass range of ±0.15 dex.
MS,MW isthereferenceMWstellarmass.Sincewearecom-
plete down to MS = 1010M⊙ we define the bin centre of 6 GALAXY PROPERTY DEPENDENCIES
the X = 1 bin containing the MW as 5.67×1010M⊙ (so
it includes stellar mass of the MW, which we take to be 6.1 The Imperfect Relationship Between Colour,
MS,MW = 5.0×1010M⊙; Flynn et al. 2006), meaning we Emission, Type & Disk
haveacompletesamplewith2.5binwidthsonthelowmass
Having established a set of contingencies that describe
side exactly. The stellar mass range selected is 100% com-
galaxies in adiscretemanner,wecan nowconstruct mosaic
plete for galaxies where X > 1.
4 plotsthatrevealthemostsignificantdependencies.Combin-
For reference the MW is in the X =1 bin (MS,MW = ing information in this manner will give us an insight into
5.0×1010M⊙; Flynn et al. 2006); the LMC is in the X = whatdrivesgalaxyformationingalaxieswithastellarmass
312 bin (MS,LMC = 2.3 × 109M⊙; James & Ivory 2011); content similar to the MW. We can also determine which
the SMC is in the X = 1218 bin (MS,MW = 5.3×108M⊙; galaxy parameters are degenerate.
James & Ivory 2011); M31 is in the X =2 bin (MS,M31 = Galaxypopulationsareknowntobebimodalinanum-
1.0×1011M⊙;Hammer et al.2007)andM33isintheX = 81 berofproperties(seeBlanton et al.2003;Baldry et al.2004,
(MS,M33 =5.7×109M⊙; Verley et al. 2009). 2006;Driveret al.2006,andTayloretal.inprep).Themain
ones that we can investigate here are colour, disk presence,
morphological type and line emission. A simplistic descrip-
5.2.4 Halo Mass
tionofextra-galacticastronomymightbracketgalaxiesinto
Y ×MH,MW (2) two main classes (or dependencies): blue, disky, star form-