Galaxy and Mass Assembly (GAMA): Demonstrating the power of WISE in the study of Galaxy Groups to z<0.1
M.E. Cluver, T.H. Jarrett, E.N. Taylor, A.M. Hopkins, S. Brough, S. Casura, B.W. Holwerda, J. Liske, K.A. Pimbblet, A.H. Wright
DDraft version June 16, 2020
Typeset using L A TEX twocolumn style in AASTeX62
Galaxy and Mass Assembly (GAMA): Demonstrating the power of
WISE in the study of Galaxy Groups to z < . M.E. Cluver,
1, 2
T.H. Jarrett, E.N. Taylor, A.M. Hopkins, S. Brough, S. Casura, B.W. Holwerda, J. Liske, K.A. Pimbblet, and A.H. Wright Centre for Astrophysics and Supercomputing, Swinburne University of Technology, John Street, Hawthorn 3122, Victoria, Australia Department of Physics and Astronomy, University of the Western Cape, Robert Sobukwe Road, Bellville, South Africa Department of Astronomy, University of Cape Town, Rondebosch, South Africa Australian Astronomical Optics, Macquarie University, 105 Delhi Rd, North Ryde, NSW 2113, Australia School of Physics, University of New South Wales, NSW 2052, Australia Hamburger Sternwarte, Universit¨at Hamburg, Gojenbergsweg 112, 21029 Hamburg, Germany Department of Physics and Astronomy, 102 Natural Science Building, University of Louisville, Louisville KY 40292, USA E.A.Milne Centre for Astrophysics, Department of Physics, University of Hull, Cottingham Road, Kingston-upon-Hull, HU6 7RX, UK Ruhr-Universit¨at Bochum, Astronomisches Institut, German Centre for Cosmological Lensing (GCCL), Universit¨atsstr. 150, 44801Bochum, Germany (Accepted June 12, 2020)
Submitted to ApJABSTRACTCombining high-fidelity group characterisation from the Galaxy and Mass Assembly (GAMA) surveyand source-tailored z < . WISE survey, we present a comprehensive study of theproperties of ungrouped galaxies, compared to 497 galaxy groups (4 ≤ N FoF ≤
20) as a function of stellarand halo mass. Ungrouped galaxies are largely unimodal in
WISE color, the result of being dominatedby star-forming, late-type galaxies. Grouped galaxies, however, show a clear bimodality in
WISE color,which correlates strongly with stellar mass and morphology. We find evidence for an increasing early-type fraction, in stellar mass bins between 10 (cid:46) M stellar (cid:46) M (cid:12) , with increasing halo mass. Usingungrouped, late-type galaxies with star-forming colors (W2 − W3 > stellar > . M (cid:12) ), consistent with mass quenching. Group galaxies with massesM stellar < . M (cid:12) show evidence of quenching consistent with environmentally-driven processes.The stellar mass distribution of late-type, quenched galaxies suggests they may be an intermediatepopulation as systems transition from being star-forming and late-type to the “red sequence”. Finally,we use the projected area of groups on the sky to extract groups that are (relatively) compact for theirhalo mass. Although these show a marginal increase in their proportion of high mass and early-typegalaxies compared to nominal groups, a clear increase in quenched fraction is not evident. Keywords: catalogs — surveys — infrared:galaxies — galaxies:groups:general — galaxies:star formation INTRODUCTIONThe local ( z < .
1) Universe offers us the clearestand most complete view for studying the feeding, feed-
Corresponding author: Michelle [email protected] back, and quenching processes that drive and regulatestar formation within cosmic structures of varying den-sity. The formation of large-scale structure and sub-structure in the universe lies at the heart of the hier-archical paradigm of Λ-CDM (e.g. Davis et al. 1985),where groups of galaxies merge into clusters, filaments,walls, and superclusters, creating the cosmic web. a r X i v : . [ a s t r o - ph . GA ] J un Cluver et al.
In the nearby universe, we observe the so-called“morphology-density” relation (e.g. Dressler 1980; Post-man, & Geller 1984; Goto et al. 2003; Blanton & Mous-takas 2009), the suppression of star formation in highdensity environments (e.g. Balogh et al. 1998; Couch etal. 2001; Lewis et al. 2002), and the bimodality of the lo-cal galaxy population as a “blue cloud” of star-forminggalaxies and a “red sequence” of quenched, passivelyevolving systems (e.g. Strateva et al. 2001; Blanton etal. 2003; Baldry et al. 2004; Balogh et al. 2004; Tayloret al. 2015). However, the pathways that lead to theseobserved trends, and how they are connected, remainunclear. This is in part due to the challenge of disen-tangling a number of possible mechanisms, acting as afunction of either stellar mass (e.g. Baldry et al. 2006;Peng et al. 2010), environment (e.g. Peng et al. 2010;Peng et al. 2012; Bluck et al. 2016), or morphology (e.g.Martig et al. 2009; Bluck et al. 2014). The complicationin distinguishing between these is that it is clear that both mass and environment play a role, which meansthat we have to carefully measure and control for mass-dependent effects in order to isolate and characterisethe effects of environment.“Mass” or secular quenching (Driver et al. 2006; Penget al. 2010) translates to more massive galaxies quench-ing independent of environment, i.e. due to internalprocesses such as AGN feedback. Alternatively, “envi-ronmental quenching” applies to galaxies quenched dueto external processes, i.e. their environment, indepen-dent of stellar mass (Peng et al. 2010). It is worthnoting that “halo quenching” (e.g. Birnboim, & Dekel2003; Dekel, & Birnboim 2006), the virial shock heatingof intergalactic gas falling into a galactic dark matterhalo, has been proposed as driving both processes (e.g.Gabor, & Dav´e 2015).A further complication, however, is the timescales onwhich the physical quenching mechanisms operate, e.g.the starvation (strangulation) of the gas supply (e.g.Larson et al. 1980; Peng et al. 2015) due to dense envi-ronments (e.g. van de Voort et al. 2017), or the heatingof galactic halos by, e.g. large-scale AGN jets (Crotonet al. 2006) and shocks (e.g. Birnboim, & Dekel 2003;Dekel, & Birnboim 2006). Here the former would bean example of environmental quenching and the latter,mass quenching. The proposed framework of cosmic webdetachment suggests that the starvation process due tothe disruption of filaments that supply gas to galaxies,encompasses the role of external processes and is ableto reproduce observations such as the dependence of thered fraction on mass and local density (Aragon Calvo etal. 2019). In dense environments, particularly, processes such asgalaxy harassment (e.g. Farouki, & Shapiro 1981; Mooreet al. 1996), tidal encounters (e.g. Toomre, & Toomre1972; Barnes, & Hernquist 1992), and various mecha-nisms of gas stripping, such as ram-pressure stripping(e.g. Gunn, & Gott 1972), tidal stripping (e.g. Mihos2004), and viscous stripping (e.g. Nulsen 1982; Ras-mussen et al. 2006), can contribute to the evolutionarypathways (morphological transformation and quench-ing) of individual galaxies. It is therefore prudent, albeitchallenging, to control for environment when investigat-ing the pathways of galaxy evolution.Galaxy evolution in the group environment is of par-ticular interest given that 40-50% of galaxies in the lo-cal ( z ∼
0) Universe, reside in groups (Eke et al. 2004;Robotham et al. 2011). It is the most common environ-ment in which galaxies are found, especially compared toclusters which are rare by comparison ( ∼ i from the AL-FALFA survey (Giovanelli et al. 2005), the processingof galaxies in the group environment is observed as anincreasing deficiency of H i -rich galaxies at the centersof groups, with increasing optical group membership.It is therefore evident that galaxy groups play a keyrole in understanding galaxy transformations and iden-tifying the mechanisms that dominate in these envi-ronments is key. In this work, we aim to provide abenchmark view of the mid-infrared properties of galaxygroups, focusing on groups with membership between 4and 20 (halo mass between 10 . to 10 M (cid:12) /h), rep-resentative of the most typical overdensities in the local( z < .
1) universe, excluding pair and triple systems (see
AMA-
WISE z < . Groups
Wide-FieldInfrared Explorer ( WISE , Wright et al. 2010) to investi-gate the diversity of mass (halo and stellar), morphology,and star formation properties within this population.
WISE surveyed the entire sky at wavelengths of 3.4 µ m(W1), 4.6 µ m (W2), 12 µ m (W3), and 23 µ m (W4; Brownet al. 2014a) and hence traces both dust-free stellar massand dust-reprocessed star formation (e.g. Jarrett et al.2012, 2013; Cluver et al. 2014), particularly in the lo-cal universe where its sensitivity to both is the mostuniform. It is therefore a valuable resource for study-ing galaxy populations in wide-area surveys and partic-ularly useful for studying the global measurements ofinteracting systems potentially generating excess dustthrough triggered star formation (e.g. Marcillac et al.2007). The mid-infrared photometry used in this studyhas been carefully tailored to suit both galaxies thatare resolved and unresolved by WISE (see Section 2.2.for details), enabling a detailed mid-infrared study ofthe color, stellar mass and star formation properties ofgrouped and ungrouped galaxies, and exploring the rolesof morphology, halo mass, and compactness. This cantherefore be extended in a straightforward way to largerareas, with improved uniformity and statistics, providedhighly complete redshifts are available.In this study we exploit the robust identification ofgroup galaxies in Galaxy and Mass Assembly survey(Driver et al. 2011), which was constructed with an em-phasis on high completeness, making it an ideal datasetfor group galaxy science. In the equatorial regions ofG09, G12, and G15, covering ∼
180 degrees (Hopkins etal. 2013; Baldry et al. 2018), the survey achieves 98.48%completeness to a limiting magnitude of r AB = 19 . WISE colors (3.1), stellar mass and mor-phology (3.2), the star-forming main sequence (3.3), andcompactness (3.4). A summary of our main results andtheir significance is given in Section 4, and conclusionsin Section 5.The cosmology adopted throughout this paper is H =70 km s − Mpc − , h = H / M = 0 .
27 and Ω Λ =0 .
73. All magnitudes are in the Vega system, as adoptedby the
WISE survey (as described in Jarrett et al. 2011).All linear fits are performed using the Hyper-Fit package(Robotham & Obreschkow 2015). DATA AND SAMPLE SELECTIONOur primary dataset is drawn from the three equa-torial regions (G09, G12, and G15) of the GAMA IIspectroscopic survey which cover an area of 180 degree to a limiting magnitude of r AB = 19 . The GAMA Group Catalog
The GAMA Galaxy Group Catalog (G C) is con-structed using an iterative friends-of-friends algorithm,making use of mock GAMA light cones in order to refinethe group-finding algorithm (full details are provided inRobotham et al. 2011). The G C assigns ∼
40% ofgalaxies to groups with multiplicity N >
1, i.e. pairs and
Cluver et al. groups (Robotham et al. 2011). For this work, we makeuse of the most recent version of the G C, G3Cv10, con-structed after the completion of the equatorial fields.The catalog has been constructed from the main sur-vey catalog, TilingCatv46, extracting galaxies with anAUTOZ redshift quality flag of NQ ≥ . < Z CMB < .
6– this sample is included in the G3Cv10 DMU (DataManagement Unit) as G3CGalv10.The group catalog extracted from G3CGalv10 is pro-vided as G3CFoFGroupv10 and we impose the followingselection criteria:1. median redshift of all groups is z < . < .
1) – although this greatly reduces the size of oursample, it is most suited to the angular resolution(6 (cid:48)(cid:48) ) and sensitivity of
WISE
2. group membership (multiplicity) between 4 and20 galaxies, i.e. 4 ≤ Nfof ≤
20, corresponding toderived halo masses between 10 and 10 M (cid:12) /h(see Figure 21 of the Appendix)3. select only groups that are entirely containedwithin the survey volume (GroupEdge=1), i.e nopartial groups due to the edges of the survey areincludedThese criteria reflect our focus on the nearby universe(with higher sensitivity and fewer observational biases)and our aim of investigating the properties of systemsdominated by multi-member interactions in the absenceof virialised halos associated with a pervasive hot intra-cluster medium.We find 498 groups that satisfy our criteria, consist-ing of 3195 galaxies. Galaxies that are left ungrouped inG3CGalv10 are designated “non-G3C galaxies” and weconsider these to be the “least grouped” galaxies withinGAMA. In addition to isolated field galaxies, they aremost likely the central galaxies of groups where the othermember galaxies were too faint to be detected. In thestudy of Barsanti et al. (2018), they showed that manyof these galaxies can be associated with groups using analternative prescription, such as projected phase space.However, since we are interested in examining the im-pact of group environment specifically, we postulate anyeffects will be strongest within the FoF-defined groupsand weakest in the non-G3C galaxies. To this end weselect least-grouped, non-G3C galaxies with z < . ∼ R σ , where R is the radius contain-ing 50% of the group members and σ is the velocitydispersion) multiplied by a scaling factor, “A”, requiredto get a median-unbaised halo mass estimate (for Nfof ≥ WISE Photometry is the pri-mary source of WISE counterparts to GAMAsources.2. The WISECatv02 catalog covers the G09, G12 andG15 fields.3. Sources that are potentially resolved by
WISE (seeCluver et al. 2014), are measured on reconstructed‘drizzle’ images (Jarrett et al. 2012), with native
WISE resolution and 1 (cid:48)(cid:48) pixels, as described inCluver et al. (2014). In bands where the source isresolved, the isophotal (integrated flux) photome-try is reported, which captures better than 90% ofthe total flux for the source (Jarrett et al. 2019).4. Due to the sensitivity of the W1 and W2 bands,unresolved extended sources (i.e. galaxies re-solved by optical imaging) are not well-measuredby profile-fit photometry. Here the standard aper-ture photometry (w1mag, w2mag), correspondingto a circular aperture of 8.25 (cid:48)(cid:48) , is reported. https://irsa.ipac.caltech.edu/data/download/wise-allwise/ AMA-
WISE z < . Groups Redshift l o g M s t e ll a r ( o p t ) ( M ) No WISE measurementWISE Catalogue source C o un t s (b) non-G3C Galaxies (ungrouped) Redshift l o g M s t e ll a r ( o p t ) ( M ) No WISE measurementWISE Catalogue source C o un t s (a) G3C Group Catalog Galaxies Figure 1.
Optically-derived stellar masses for a) the G3C galaxies in our sample, and b) the non-G3C galaxies indicate theWISE cross-match sample to be highly complete when imposing a mass cut of 10 M (cid:12) .
5. For sources unresolved in the W3 and W4 bands,the profile-fit photometry (w3mpro, w4mpro) fromthe ALLWISE catalog is reported, providing thebest sensitivity for these cases (Cluver et al. 2014).We modify this catalog by replacing non-detection,upper limits (from ALLWISE) with forced photometryprovided by the LAMDAR code DMU (LamdarCatv01)as detailed in Wright et al. (2016). This provides use-ful constraints when shifting to restframe photometry,and enables the propagation of meaningful upper limits.The G3C and non-G3C galaxies of our sample are thencrossmatched to the modified
WISE photometry catalogusing the CATAID identifier; the statistics are listed inTable 1.The magnitude limit used for galaxy selection in theGAMA survey leads to a sample whose mass com-pleteness is a function of redshift. We make use ofthe optically-derived StellarMasses DMU (StellarMass-esv19) as detailed in Taylor et al. (2011), using h = 0 . WISE coun-terpart (and therefore do not have a
WISE stellar mass),as a function of redshift. The W1 (3.4 µ m) band of WISE is its most sensitive, but Figure 1 indicates that sourceswith low stellar mass can be missed by
WISE as they lack a substantial old stellar population. As shown, aM stellar ≥ . M (cid:12) selection creates an approximatelymass complete sample to z < . WISE is less suited to detect (see Table 1).2.3.
Derived Quantities
Multiwavelength optical and near-infrared photome-try from LamdarCatv01 (Wright et al. 2016), correctedfor Galactic foreground dust extinction using Schlegelet al. (1998) (GalacticExtinctionv03), is used in com-bination with the
WISE mid-infrared (mid-IR) catalogdescribed above to determine rest-frame photometry byfitting to the empirical template library of Brown et al.(2014b).The aim of our study is to use
WISE as our primarydata source; given its all-sky coverage, a study of thiskind can be extended in a uniform way. In a sense weare therefore simulating regions of the sky that do notshare the extensive multiwavelength coverage in GAMA,and hence we use
WISE to determine both the stellarmass and SFRs of the galaxies in our sample. The stel-lar masses are derived using equation (1) of Cluver etal. (2014). This relation was determined using
WISE -resolved galaxies calibrated to the GAMA stellar masses
Cluver et al.
Table 1.
Crossmatch statisticsNo. of Galaxies
WISE
Matches CompletenessG3C 3195 2871 90%G3C optical stellar mass cuta
WISE stellar mass cutb optical stellar mass cuta
WISE stellar mass cutb a Using optically-derived stellar masses where available b Using
WISE -derived stellar masses
Redshift G r o u p M e m b e r s h i p C o un t s Groups
Counts
Figure 2.
Group membership distribution, as a function ofredshift, of the G3C groups in the sample. of Taylor et al. (2011), derived from stellar populationsynthesis modelling, and assuming a Chabrier (2003)IMF. For convenience we reproduce it here:log M stellar /L W1 = − . − W2) − . , (1)with L W1 ( L (cid:12) ,W ) = 10 − . M − M (cid:12) , W1 ) , where M is theabsolute magnitude of the source in W1, M (cid:12) , W1 = 3 . . µm − W . µm is the rest-frame W1 − W2 color of the source(see Jarrett et al. 2013, for further details). This equa-tion is only applied within the limits of the calibrationi.e. W1 − W2 color from − /L W1 of 0.21 to 0.91); for galaxies with onlyW1 detections (e.g. dwarfs) a constant M /L W1 of 0.6 isused (see Jarrett et al. 2019). For consistency with our SFR relation (see below), we convert our stellar massesto a Kroupa (2001) IMF using the offsets from Zahid etal. (2012) corresponding to 0.03 dex.After calculating the WISE -derived stellar masses, weimpose a stellar mass cut of log M stellar ≥ . (cid:12) as dis-cussed above (see Table 1). After the cut, one group of4 galaxies (GroupID: 200857) is no longer represented inour sample, leaving 497 groups whose redshift distribu-tion is shown in Figure 2. Inspection of the properties ofthis excluded group (located at z = 0 . < M (cid:12) , reflecting broad consistency between the two mea-sures of stellar mass. This would be an interesting groupin its own right, but illustrates the rarity of such groupsin our volume.SFRs are determined using the W3 (12 µ m) band, afterremoving the contribution from the stellar continuum(see Cluver et al. 2017). Dust-reprocessed star forma-tion, as traced by the mid-infrared, probes star forma-tion on timescales (cid:38)
100 Myr, but compares favourablyto optically-derived values that require, sometimes large,dust corrections (see Cluver et al. 2017). We make useof equation (4) from Cluver et al. (2017), calibrated tothe total infrared luminosities of the SINGS/KINGFISHsample (Dale et al. 2017) and assuming a Kroupa IMF(Kroupa 2001), reproduced here:log SFR (M (cid:12) yr − ) =(0 . ± . L µ m( L (cid:12) ) − (7 . ± . , (2)where L µ m is the monochromatic 12 µ m luminosity, νL ν (12 µ m), determined from the restframe-correctedW3 band.We note that some dust-reprocessed emission withinthe W3 band is due to heating from the old stellar pop-ulation; this is particularly true of massive ellipticals.Although this is somewhat ameliorated when removing AMA-
WISE z < . Groups
Photometry Quality Cuts
For our primary analysis, we are particularly inter-ested in the stellar mass dependence of our sampleswhich requires controlling for stellar mass in a reliableway. We impose a restriction of signal to noise (S/N) inW1 − W2 color (S/N >
5) which corresponds to a stellarmass error less than 0.5 dex; this is necessary to limitcontamination across bins. We explore the consequencesof this selection further in section A of the Appendix,but note that it chiefly impacts the number of galax-ies in our sample at the low mass end (M stellar < . M (cid:12) ). The requirements we have imposed on our sam-ple mean that all numbers and figures presented in thiswork should be compared in relative and not absoluteterms.We further apply a S/N cut in W2 − W3 (S/N > − W3 color, we report a low S/N SFR.Upper limit SFRs to z < . − W3 color, but a large stellar continuumdominates the W3 band – such as is the case for highmass elliptical galaxies – and results in an upper limit af-ter correcting for this. In addition, an upper limit ariseswhen little to no star formation is detected; this occursat high mass due to systems having low or negligible starformation, but also at the low mass end (M stellar < M (cid:12) ), where the low surface brightness of these systemsimpacts the W3 detections of these systems. We notethat for the analysis presented in section 3.3, the WISE
W3 band sensitivity means we are complete to z < . stellar ≥ M (cid:12) . However, for the lowlevels of star formation of the M stellar < M (cid:12) pop-ulation, there is redshift dependence; we therefore testeach source against its distance to determine if a W3flux (and hence SFR) could be detected. This limitsthe number of sources we can count in our lowest stel-lar mass bins and therefore reduces our statistical powerthere. 2.5. Visual Morphological Classification
In order to investigate the morphological mix ingroup environments, we make use of the VisualMor-phologyv03 DMU which contains the visual identifi-cation (following Driver et al. 2012) for galaxies inthe GAMA II equatorial regions to z < . giH -bandimages from the SDSS (York et al. 2000), VIKING(Edge et al. 2013) or UKIDSS (Lawrence et al. 2007)large area survey data.This DMU additionally includes the Hubble Typeclassifications, following Kelvin et al. (2014), for theGAMA II equatorial regions, but which is only avail-able for galaxies to z < .
06. Of the galaxies with bothan ELLIPTICAL CODE and HUBBLE TYPE CODEclassification (i.e. z < . • ELLIPTICAL CODE=1: 73% are classified as Eand 22% are classified S0-Sa, with only 1% classi-fied as either Sab-cd or Sd-Irr. • ELLIPTICAL CODE=10: 79% are classified aseither Sab-cd or Sd-Irr, with 4% classified as E,and 10% classified as S0-Sa.For the purposes of this study, therefore, we make useof the ELLIPTICAL CODE classification, where avail-able, and assign galaxies with ELLIPTICAL CODE=1to be “Early-type”, i.e. bulge-dominated systems, andthose with ELLIPTICAL CODE=10 as “Late-type”,i.e. disk-dominated galaxies. After applying our stellarmass cut, we have only 2 galaxies in the non-G3C sam-ple without a classification (ELLIPTICAL CODE=0).Similarly for 11 galaxies in the G3C sample. ANALYSIS3.1.
WISE colours
We consider first the
WISE color-color diagrams of thenon-G3C and G3C galaxies, color-coded by 12 µ m SFR,as shown in Figure 3; we include the “color sequence”and delimited regions from Jarrett et al. (2019). In theupper panels we show the distribution of systems withreliable W2 − W3 colors (S/N > − W3 sources.As expected, star formation is generally highest atlarge W2 − W3 color, decreasing to the left where lowstar-forming, large stellar mass systems reside (Jarrettet al. 2019). It is apparent from the histogram in theupper panel of Figure 3a that the W2 − W3 color dis-tribution of the non-G3C galaxies is largely unimodal,dominated by mid-IR “red” colors corresponding to ac-tively star-forming systems; this is consistent with the
Cluver et al.
W2-W3 (mag) W - W ( m a g ) spheroids intermediate disksactive star-forming disksAGN and extrema W2-W3, SFRW2-W3, U/L SFRJarrett et al. (2019) C o un t s W2-W3Low S/N W2-W3U/L W2-W3 l o g S F R m ( M / y r ) (a) Non-G3C sample W2-W3 (mag) W - W ( m a g ) spheroids intermediate disksactive star-forming disksAGN and extrema W2-W3, SFRW2-W3, U/L SFRJarrett et al. (2019) C o un t s W2-W3Low S/N W2-W3U/L W2-W3 l o g S F R m ( M / y r ) (b) G3C sample Figure 3.
WISE color-color diagram (W1 − W2 vs W2 − W3), color-coded by SFR, for the (a) non-G3C sample and (b) G3Cgalaxies, after applying the W2 − W3 S/N cut. Also shown are galaxies where the requirement of a robust SFR have beenremoved (black points). The upper panels additionally reflect sources with low S/N and upper limit W2 − W3 color that arenow shown in the main panel. The non-G3C sample is dominated by star-forming systems with large W2 − W3 color, while theG3C sample shows relatively more sources at low W2 − W3 color. The color-morphology divisions and color-color “sequence”are from Jarrett et al. (2019). optical color distribution ( u − r ) found for ungroupedgalaxies at z < . − W3 color. We note that the mid-IR-derived SFRs of galaxies in the “AGN and extrema”zone (see Jarrett et al. 2011; Stern et al. 2012) are likelycontaminated by hot dust from AGN heating, and areconsidered to be unreliable in their mass and SFR trac-ers (and therefore excluded from this analysis).Included in the main panels of Figure 3, as blackpoints, are galaxies with a well-determined W2 − W3color, but without a robustly determined SFR. Thereis an inherent challenge in studying systems that areceasing to form stars (or have ceased star formation, i.e.passive) using diagnostics that require a well-determinedSFR. In the case of the mid-IR, the galaxy must havedust-obscured SF in order to gauge the activity throughISM heating, wholly separate from the photosphericemission from evolved stars. In the mid-IR, bulge-dominated galaxies with little star formation will stillhave a reliable, if low, W2 − W3 color as their Rayleigh-Jeans continuum dominates the W3 band (and not warmdust from star formation). This continuum is removed when calculating the 12 µ m-derived SFR, which can re-sult in little to no (reliably) detectable warm dust. Fig-ure 3 shows these cases as black points and demonstratesthat the W2 − W3 color probes further down by almostan entire magnitude (W2 − W3 < . Stellar Mass and Morphology
We next explore a more physically-informed diagnos-tic, stellar mass (M stellar ) versus W2 − W3 color. In Fig-ure 4 we divide the non-G3C and G3C samples intoearly-type (i.e. bulge-dominated) and late-type (i.e.disk-dominated), making use of the visual morphologyclassification outlined previously. This clearly showsthat the dominance of star-forming galaxies seen in Fig-ure 3a is directly attributed to late-type systems, withlittle contribution from early-types (only 14% of non-G3C galaxies in our sample are classified as early-type),even at low W2 − W3 color/ high stellar mass (M stellar > M (cid:12) ).In contrast, the G3C sample (Figure 4b) shows aclearly bimodal distribution in stellar mass, with early-type systems (33% of our sample) dominating at the AMA-
WISE z < . Groups W2-W3 (mag) l o g M s t e ll a r ( M ) Early-type non-G3C galaxiesLate-type non-G3C galaxies C o un t s Early-typeLate-type
Counts (a) non-G3C sample
W2-W3 (mag) l o g M s t e ll a r ( M ) Early-type G3C galaxiesLate-type G3C galaxies C o un t s Early-typeLate-type
Counts (b) G3C sample
Figure 4.
Stellar mass as a function of W2 − W3 color, color-coded by morphological classification (early-type=red, late-type=blue). The (a) non-G3C sample and (b) G3C sample have very different distributions in this phase space. high stellar mass/low W2 − W3 end. This framing ofstellar mass vs W2 − W3 color also shows that low-masssystems do not have the largest W2 − W3 color, i.e. thelate-type distribution turns back to bluer mid-IR col-ors at low mass indicating lower dust content and SFRactivity.The difference in the grouped and ungrouped stellarmass distributions is consistent with what was foundby Alpaslan et al. (2015) in GAMA. We note that thelower mass and late-type nature of the non-G3C galaxies(when considered the central of an undetected group)are in agreement with Robotham et al. (2013) who findthat centrals of lower mass (M stellar < . M (cid:12) ) aremore likely to be late-type than centrals of higher mass.We see from Figure 4 that, not only do the sourcecounts of the ungrouped sample drop off rapidly forM stellar > . M (cid:12) , but that very few are classifiedas early-type compared to the same stellar mass rangein the grouped sample; this makes them very differentpopulations. In the mass range of M stellar > . M (cid:12) where galaxies are turning off the star-forming main se-quence, it shows that, at fixed mass, the grouped galax-ies are more likely to be early-type. i.e. one impor-tant consequence of being in a group is the associatedincrease in the early-type fraction; we investigate thisfurther in the next section. Considering the stellar mass distributions (right pan-els in Figure 4), the late-types (blue histograms) in theG3C sample show a tendency to higher mass comparedto the non-G3C sample. This implies that the changingstellar mass function also modifies the mass distributionsof different morphological types.Next we turn to the dynamical (halo) masses of thegroups in our sample. We divide the WISE -G3C sampleinto three halo mass bins, choosing limits that provideroughly equal numbers of galaxies. Three groups thathave MassAFunc= 0 are excluded; the remaining aredivided as shown in Table 2.
Table 2.
Dynamical Mass of Group SampleLog Mass Range Number of Number of(M (cid:12) /h) Groups GalaxiesGroup Mass 1 ≤ .
95 223 853Group Mass 2 12.95 – 13.4 159 891Group Mass 3 13.4 – 14.1 112 824
Splitting into group mass bins (Figure 5), we notethe increase in systems at high stellar mass, M stellar > . M (cid:12) , (shown in the right panels) with increasing0 Cluver et al.
W2-W3 (mag) l o g M s t e ll a r ( M ) Early-type Group Mass 1Late-type Group Mass 1 C o un t s Early-typeLate-type
Counts (a) Group Mass 1 ( < . M (cid:12) /h) W2-W3 (mag) l o g M s t e ll a r ( M ) Early-type Group Mass 2Late-type Group Mass 2 C o un t s Early-typeLate-type
Counts (b) Group Mass 2 (10 . – 10 . M (cid:12) /h) W2-W3 (mag) l o g M s t e ll a r ( M ) Early-type Group Mass 3Late-type Group Mass 3 C o un t s Early-typeLate-type
Counts (c) Group Mass 3 ( > . M (cid:12) /h) Figure 5.
As in Figure 4, but for the individual group mass bins listed in Table 2. With increasing group mass, the relativeproportion of early-type systems increases (corresponding to an increase of systems at high stellar masses): in Group Mass 1,27% are early-type, 31% in Group 2, and and 39% for the largest halos. halo mass. This is consistent with the variation of thegalaxy stellar mass function with halo mass (e.g. Yanget al. 2009; Alpaslan et al. 2015). In addition, the shiftto low W2 − W3 colors observed in Figure 3b appears tobe driven by an increased population of early-type sys-tems (with increasing halo mass), which dominates the high stellar mass population. In Figure 5, Group Mass1 is 26.7( ± ± ± − W3 cut. If we considerour entire sample (i.e. with only a W1 − W2 selection)and M stellar > M (cid:12) where our sample is most com- AMA-
WISE z < . Groups log M stellar (M ) E a r l y - T y p e F r a c t i o n non-G3CGroup Mass 1Group Mass 2Group Mass 3 Figure 6.
The Early-type fraction in bins of stellar mass forthe non-G3C (black), Group Mass 1 (green), Group Mass 2(brown), and Group Mass 3 (purple). Errors are determinedfrom bootstrap resampling. This indicates that in the binsbetween M stellar ∼ M (cid:12) and M stellar ∼ M (cid:12) we maybe seeing an increase in early-type fraction associated withincreased halo mass. plete, we find that Group Mass 1 is 29.7( ± . ± ± stellar ∼ M (cid:12) and M stellar ∼ M (cid:12) the early-type frac-tion appears to increase with group halo mass withinthe stellar mass bins.We can see that the transition from ungrouped to ourlowest halo mass has changed the stellar mass profileand fraction of early-types in a noticeable way. Galaxiesin groups have either built (high) stellar mass more effi-ciently compared to ungrouped systems, with more high-mass galaxies as halo mass increases, or their formation history means they have had more time to build mass.One may expect increased interactions and merging ingroup environments, which would be consistent with theaccompanied increase in early-types (i.e. bulge growth)with increasing halo mass. Gravitational torques dueto tidal interactions can cause gas to flow to the cen-tres of galaxies, leading to centrally-concentrated starformation and corresponding bulge growth. Schaefer etal. (2019) find that galaxies with M stellar > M (cid:12) inhigh-mass groups are more likely to experience centrally-concentrated star formation, whilst Bluck et al. (2014)find that bulge mass is most strongly correlated withpassive fraction, consistent with the inside-out growthparadigm. This observed increase in high stellar massgalaxies and early-types with increasing group halo masswill likely impact the observed fraction of quenchedgalaxies (e.g. Peng et al. 2010; Davies et al. 2019b),which we examine in the next section.To summarise, we find progressively more high-massand early-type systems with increasing group halo masscompared to the ungrouped sample which is dominatedby late-type galaxies and has relatively few galaxies withM stellar > . M (cid:12) .3.3. The Star-forming Main Sequence
The correlation between stellar mass and SFR forstar-forming galaxies (Noeske et al. 2007; Elbaz et al.2007; Daddi et al. 2007) has become an indispensabletool for identifying and studying the properties of typ-ical star-forming galaxies to high redshift (e.g. Brisbinet al. 2019). It also provides a natural means to sepa-rate samples into star-forming, transitioning and passivegalaxies (see, for example, Bluck et al. 2014; Renzini, &Peng 2015; Bluck et al. 2016; Davies et al. 2019b; Wang,B. et al. 2020).The color-coding used in Figure 3 showed the connec-tion between SFR and W2 − W3 color, with high star-formation broadly corresponding to large W2 − W3 color(due in part to SFR being derived from the W3 lumi-nosities after removing the stellar contribution). We ex-ploit this using the late-type, non-G3C sample and selectstar-forming galaxies based on their W2 − W3 color (us-ing W2 − W3 > stellar plane, as shown in Figure 7(corresponding to an assumed Kroupa IMF, Cluver etal. 2017) with a best-fit relation given by:log SFR(M (cid:12) yr − ) = 0 .
93 log M stellar (M (cid:12) ) − . , (3)with σ = 0 .
29 reflecting the intrinsic spread of the dis-tribution. The distribution is well-contained within ± σ of the relation, as shown in Figure 7. We find that2 Cluver et al. log M stellar (M ) l o g S F R ( M y r ) log SFR (M yr ) = 0.93 log M stellar (M ) -9.08 SFMS SelectionSFMS Fit2Quenching SeparatorElbaz et al. (2007) C o un t s SFMSLow Mass
Counts
W2-W3
Figure 7.
The SFMS as determined from the late-type non-G3C sample (colored points), selecting galaxies with W2 − W3 > z < .
1. Low mass galaxies with high S/N SFRs are shown as black points, with no morphology or color selection imposed.The dotted line shows the upper 2 σ envelope, while the dashed line represents the quenching separator, which is the lower 2 σ boundary from the SFMS fit for M stellar ≥ M (cid:12) and a modified selection for M stellar < M (cid:12) to accommodate lower SFRsat the low mass end. our relation closely matches the relation of Elbaz et al.(2007), derived using SDSS galaxies at z < . ∼
100 Myr) compared to, for ex-ample, H α sensitive to (cid:46)
20 Myr (e.g. Kennicutt 1998).We use the lower 2- σ (i.e. 0.6 dex) boundary (seeFigure 7) to separate systems on the SFMS and thosethat are below (where M stellar ≥ M (cid:12) ) – we refer tothese as “quenched”, but note that these include transi-tional (or “quenching”) systems, in the process of mov-ing off the SFMS. At the high mass end, our selectionof quenched systems, therefore, includes galaxies that would be considered in the “green valley” by some stud-ies (see e.g. Bluck et al. 2016; Janowiecki et al. 2020).Our SFMS selection, however, has not taken into ac-count the low mass population in the non-G3C sample,with correspondingly low SFRs; the W2 − W3 > stellar < . M (cid:12) and W2 − W3 ≤
3. To account for the increased scat-ter, we bin in stellar mass and determine the values at3 standard deviations below the SFMS and fit to thosepoints, intersecting the 2 σ line at 10 M (cid:12) . The equa-tion for the quenching separator at low mass is thereforegiven by:log SFR(M (cid:12) yr − ) = 1 .
635 log M stellar (M (cid:12) ) − . M stellar < M (cid:12) (4) AMA-
WISE z < . Groups log M stellar (M ) l o g S F R ( M y r ) log SFR (M yr ) = 1.08 log M stellar (M ) -10.5 Non-G3C SFMSG3C SFMS C o un t s ( n o r m . ) Non-G3C SFMSG3C SFMS
Counts (norm.)
W2-W3
Figure 8.
Replicating the selection used in Figure 7 (late-type galaxies with W2 − W3 > Considering now the equivalent SFMS for the G3Csample, we impose the same selection (late-type mor-phology with W2 − W3 >
3) and obtain Figure 8, wherewe see a slightly steeper relation, consistent with a smallshift to higher stellar mass and higher SFR (as seenin the normalised histogram comparisons). This wouldsuggest that in the group sample, galaxies on the star-forming sequence are experiencing a “feast before thefamine”, tending to higher star formation compared totheir ungrouped counterparts. This could be the resultof triggered elevated star formation due to increased in-teractions (e.g. Moreno et al. 2019), or minor mergers,occurring in the group environment. It is therefore clearthat the selection employed when determining a SFMSselection is important. For the analysis that follows, wewill use the quenching separator as determined from theSFMS and low mass population of the non-G3C sample,as a control to test for differences compared to the groupenvironment.3.3.1.
The SFMS and the Quenching of Star Formation
In Figure 9 we show the log SFR–log M stellar distri-bution, for the entire (a) non-G3C and (b) G3C sam- ples. We also show the low S/N and upper limit SFRvalues, which are included in the stellar mass distribu-tions (upper panels). Using the quenching separator de-fined in the previous section, the upper panels reflect thequenched and unquenched distributions for both sam-ples. The unquenched distributions of both appear sim-ilar as a function of stellar mass. This suggests that theoverall SFRs of galaxies within the SFMS is largely ag-nostic to being in a grouped or ungrouped environment,although we have shown in Figure 8 that the slope ofthe relation is somewhat steeper, suggesting a slightlydifferent mass dependence.In the G3C sample (Figure 9b), we clearly see the“turnover” of SFR at high stellar mass, such that forM stellar > . M (cid:12) the SFR of galaxies is broadly de-creasing, consistent with what is found in other studies(e.g. Kauffmann et al. 2003; Salim et al. 2007). Thisis less evident in the non-G3C sample due to the steepdrop-off in high-mass systems.Here we see why the slope of the SFMS is important;Figures 7 and 8 have shown that the shape of our star-forming selection in both samples is similar. The dif-ference is in Figure 9b where the greater proportion ofmassive galaxies with intermediate star formation ratesmeans that if we would fit a relation to the entire star-forming sample (i.e. with no WISE color cut), the slopewould be quite different. Any difference would thenreflect the proportion of star-forming versus quiescentgalaxies, rather than differences in the properties of star-forming galaxies.
Table 3.
Unquenched and Quenched Fractions for the Un-grouped (non-G3C) and Grouped (G3C) SamplesSample Unquenched Quenchednon-G3C 74.6 ( ± ± ± ± ± ± ± ± ± ± The overall fraction of quenched systems (see Table 3)in the G3C sample (51.5 ± ± > M (cid:12) , i.e. high-mass systems, inthe G3C sample. We note that the fractions of quenchedand unquenched galaxies presented in this section shouldbe interpreted relative to each other and not in absolute4 Cluver et al. log M stellar (M ) l o g S F R ( M y r ) Non-G3C Galaxieslow S/Nupper limitsQuenching Separator9.0 9.5 10.0 10.5 11.0 11.50200400 C o un t s UnquenchedQuenched (a) non-G3C sample log M stellar (M ) l o g S F R ( M y r ) G3C Galaxieslow S/Nupper limitsQuenching Separator9.0 9.5 10.0 10.5 11.0 11.5050100150 C o un t s UnquenchedQuenched (b) G3C sample
Figure 9.
The log SFR–log M stellar distribution for the (a) non-G3C and (b) G3C samples, respectively; the dividing lineseparates what we consider in this study to be unquenched (above the line) and quenched (below the line) systems. The upperpanels reflect the stellar mass distribution of the unquenched (unfilled histogram) and quenched (filled histogram) population.The overall fraction of quenched systems in the G3C sample is 51.5% compared to just 25% in the non-G3C sample; see Table3. terms, due to their dependence on choice of SFMS andquenching separator.In Figure 10a we consider the stellar mass distributionof quenched and unquenched systems for the non-G3C(black) and G3C (orange) samples. The steep drop-offof unquenched galaxies at M stellar > . M (cid:12) is mir-rored in both samples and can be explained in terms ofthe relative paucity of high SFR systems at high stellarmass. In the G3C population, the quenched and un-quenched populations are approximately equal, with atransition from one to the other at M stellar ∼ . M (cid:12) .The star-forming, or unquenched, population of the non-G3C sample dominates, while the G3C sample shows alarger fraction of quenched systems at stellar masses ofM stellar > . M (cid:12) .The quenching of galaxies with M stellar > . M (cid:12) (corresponding to a galaxy halo mass of M stellar > . M (cid:12) ) is expected from the break in the to stel-lar mass - halo mass (SMHM) relation (Behroozi et al.2013), and corresponds to the mass scale relevant to theviral shock heating of accreting gas. The quenched pop-ulation is clearly dominated by massive systems and thisconnection to stellar mass can be understood as beingdriven by a lack of incoming gas to replenish star for- mation leading to starvation (so-called mass quenchingor secular evolution). In Figure 10a, we see that therising quenched fraction in the G3C sample peaks at ∼ M stellar ∼ . , whereas it peaks in a lower mass binfor the non-G3C sample. This would indicate a veryrapid response to viral shock-heating, whereas starva-tion is expected to lead to a gradual decline. Therefore,although the connection to stellar mass is clear, addi-tional mechanisms may be at work.To investigate this further, we extract the number ofunquenched and quenched systems in each stellar massbin and calculate the quenched fraction (i.e. number ofquenched galaxies relative to the total number of galax-ies in that stellar mass bin). Figure 10b shows howthe quenched fraction of G3C galaxies compares to non-G3C galaxies in each stellar mass bin. In the highestmass bins, the quenched fractions of both samples con-verge to 1 due to a lack of galaxies on the SFMS at highmass. However, as can be seen in Figure 10a, the rela-tive dearth of high mass galaxies in the non-G3C sampleoverall drives up the quenched fraction somewhat artifi-cially. This highlights a limitation of using a comparisonto an ungrouped (or “field”) sample due to the underly-ing differences in their stellar mass functions. However, AMA-
WISE z < . Groups log M stellar (M ) P e r c e n t a g e o f P a r e n t S a m p l e ( % ) non-G3C Unquenchednon-G3C QuenchedG3C UnquenchedG3C Quenched (a) non-G3C and G3C comparison log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C GalaxiesG3C Galaxies (b) quenched fraction of the non-G3C (black) and G3C (orange)samples Figure 10. (a) The percentage of unquenched (circles) and quenched (squares) systems in each stellar mass bin for the non-G3C(black) and G3C galaxies (orange), respectively. The shape of the distribution of unquenched systems appears similar for bothnon-G3C and G3C samples, but the quenched distribution for the G3C sample indicates a larger population at high stellarmass. (b) The quenched fraction in each mass bin for the non-G3C (black) and G3C galaxies (orange). The numbers at thebottom of the plot reflect the galaxies found in each bin with non-G3C quenched in pink, non-G3C unquenched in black, G3Cquenched in red, and G3C unquenched in blue. Errors are calculated using bootstrap resampling in each bin. it is clear that the quenched fraction in the G3C sampleis higher in each mass bin for M stellar ≥ . M (cid:12) , re-flecting that star formation in the G3C sample is beingimpacted at both low and high mass. We note that thepoor statistics in the two lowest mass bins of our samplemeans we are insensitive to any differences in this massrange (see section A of the Appendix).The quenched fraction at low mass (M stellar < . M (cid:12) ) is expected to reflect mechanisms such asgas stripping, strangulation, harassment etc. acting onthe star-forming population i.e. galaxies at low massare likely more susceptible to environmental quenchingin more massive halos (see, for example, Peng et al.2010; Peng et al. 2012; Davies et al. 2019b; Liu et al.2019; Li et al. 2020). Our lowest mass bins do notshow a clear separation, but this is also where we havethe least statistical power (see section A). However, allother mass bins in this range reflect a larger quenchedfraction in the G3C sample, indicative of environmental(i.e. external) processes affecting star formation.Table 3 provides the corresponding quenched and un-quenched populations in each of the different group massbins; this shows that although the increase of quenched fraction with halo mass is clear, there is a large step fromGroup Mass 1 to Group Mass 2, and only a relativelysmall increase to Group Mass 3. This motivates for alarger sample that would allow for increased divisions inhalo mass to explore this transition further.In Figure 11 we provide the quenched fractions ineach stellar mass bin for the individual group mass sam-ples. Considering the range M stellar < . M (cid:12) , anelevated quenched fraction is seen at all group massesfor M stellar (cid:38) . M (cid:12) . We note that as we approachM stellar ∼ . , the quenched fractions appear to in-crease with Group Mass, which is broadly consistentwith Davies et al. (2019b). Above M stellar ∼ . M (cid:12) ,Group Mass 1 shows little difference compared to thenon-G3C sample, consistent with its relatively smallerproportion of high mass systems. Both Group Mass 2and Group Mass 3 show a large quenched fraction, rel-ative to the ungrouped and Group Mass 1 sample, be-tween M stellar ∼ . M (cid:12) and M stellar ∼ M (cid:12) , butwith no clear halo mass dependence.3.3.2. Quenching and Morphology Cluver et al. log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C GalaxiesGroup Mass 1 (a) Group Mass 1 log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C GalaxiesGroup Mass 2 (b) Group Mass 2 log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C GalaxiesGroup Mass 3 (c) Group Mass 3 Figure 11.
The quenched fraction in each mass bin for the three group mass bins. In each panel, the number of galaxies ineach bin is given at the bottom, with the non-G3C quenched galaxies in pink and the non-G3C unquenched in black. For thegroup mass bins, quenched galaxies are in red and unquenched in blue. As in Figure 10b, errors are calculated using bootstrapresampling in each bin. Notably for Group Mass 2 and 3 (panels (b) and (c), respectively) the quenching mechanisms operatingat high stellar mass are more effective i.e. in more massive group halos.
AMA-
WISE z < . Groups log M stellar (M ) l o g S F R ( M y r ) Non-G3C Early-typelow S/Nupper limitsNon-G3C Late-typelow S/Nupper limits C o un t s Non-G3C Early-typeNon-G3C Late-type
Counts (a) non-G3C sample log M stellar (M ) l o g S F R ( M y r ) G3C Early-typelow S/Nupper limitsG3C Late-typelow S/Nupper limitsMilky Way C o un t s G3C Early-typeG3C Late-type
Counts (b) G3C sample
Figure 12.
The log SFR–log M stellar distribution for the (a) non-G3C and (b) G3C sample, now color-coded by morphologicalclassification (see also Figure 22 in the Appendix for comparison). The G3C sample sees a shift to higher stellar mass drivenby an increase in early-types. In panel (b) we include the location of the Milky Way (from Mutch et al. 2011). Cluver et al.
We next fold our visual morphology classifications intothe SFMS analysis by color-coding the non-G3C andG3C galaxies as early- or late-type (Figure 12). Forthe non-G3C sample we see the dominance of late-typesnoted previously, but also that the SFR and stellar massdistributions of the early-types show the greatest differ-ence between the non-G3C and G3C samples. For theG3C sample we observe a shift of the early-type dis-tribution to high stellar mass (also seen in Figure 4b).This is accompanied by the (expected) shift to lowerSFRs of the early-type distribution, which can be seenin the right side panel of Figure 12b. Considering thelate-type stellar mass distributions (blue histograms inthe upper panels of Figure 12), we observe a shift tohigher masses in the G3C sample (also seen in Figure4). To confirm the validity of the observed distributionof late- and early-types using our adopted classification,we use the detailed bulge-to-disk decomposition of Ca-sura et al. (in prep), available for a sub-sample of ourgalaxies, in Section C of the Appendix (Figure 22).In Figure 12b, we include the position of the MilkyWay in log SFR–log M stellar space, from Mutch et al.(2011), who found it to lie within the “green valley”.In our classification, the Milky Way is a late-type,quenched system i.e. below the SFMS. As noted earlier,we expect this class to include systems in the processof quenching and the Milky Way is a good example ofa galaxy with low, but not extinguished star formation.Mutch et al. (2011) suggest that both the Milky Wayand M31 represent a population of galaxies in the midstof a transitional process and that the limited availabilityof cold gas is the main cause for the observed decline ofstar formation. They posit that both galaxies will qui-escently evolve onto the red sequence before they mergein ∼ stellar > M (cid:12) ), but the stel-lar mass distribution of late, unquenched galaxies ap-pears quite similar to the non-G3C sample with the only difference being the relative proportion (41% vs.66%). As with the non-G3C sample, the fraction of un-quenched, early-type systems in the G3C sample is thesmallest contributor and comparable to that of the non-G3C distribution. The increase of high-mass, late-typesystems in the G3C sample seen in Figures 4 and 12 ap-pears to drive a corresponding increase in the quenchedlate-type population.In the context of this study, therefore, both late-typeand early-type systems are found below the star-formingmain-sequence, but the latter to a greater degree thanthe former. This is consistent with the fact that at highstellar mass (M stellar > . M (cid:12) ), where systems turnoff the star-forming sequence, early-types are more com-mon than late-types (see also Kelvin et al. 2014; Moffettet al. 2016).Bearing in mind that early-types were excluded fromthe SFMS fit through morphology and color, the fractionof early-types that are unquenched (i.e. within the star-forming sequence) only make up 6.8% of the total (seeTable 4. This is very slightly lower than what is foundfor the non-G3C sample (8.1%). There is therefore no in-dication that bulge-dominated galaxies are more preva-lent on the SFMS in grouped galaxies, despite their in-creased numbers at high mass.Using the quenching separator as before (Figure 13),we provide the breakdown of quenched and unquenchedsystems, separated by morphology in Figure 14, forthe different group mass bins. The relative fractionsof unquenched and quenched, late- and early-types(also for the non-G3C and parent G3C samples) isprovided in Table 4. We observe that the fractionof quenched early-types increases with increasing halomass, from 28( ± ± ± AMA-
WISE z < . Groups log M stellar (M ) C o un t s Late Unquenched (non-G3C)Late Quenched (non-G3C)Early Unquenched (non-G3C)Early Quenched (non-G3C) (a) non-G3C sample log M stellar (M ) C o un t s Late Unquenched (G3C)Late Quenched (G3C)Early Unquenched (G3C)Early Quenched (G3C) (b) G3C sample
Figure 13.
The stellar mass distributions of the unquenched (unshaded) and quenched (shaded) systems, sub-divided intoearly- and late-type. The non-G3C sample (a) is dominated by unquenched late-type systems, whereas the G3C sample (b)shows a large fraction of quenched early-type, and also a large fraction of quenched late-type systems.
Table 4.
Unquenched and Quenched, Early- and Late-types in the non-G3C and G3CSamplesSample Unquenched Quenched Unquenched QuenchedLate-type Late-type Early-type Early-typenon-G3C 65.8( ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± not change within the uncertainties to Group Mass 3)with the expected shift to more massive stellar massescompared to the late, unquenched population. In stellarmass distribution (Figure 14), the late-type, quenchedsystems appear to form an intermediate population be-tween the late, unquenched and early, quenched popula-tions. We expect several pathways of transformation tobe operating in groups (including mergers) and this mayindicate a pathway of transformation (late, unquenchedto early, quenched via late-type, quenched) that oper-ates more efficiently at some halo masses.Finally, the quenched fraction by morphology, perstellar mass bin, is presented for early-types (Figure15) and late-types (Figure 16) in each of our samples. Considering the early-types, Figure 15a shows that forM stellar > M (cid:12) , early-types in groups show a higherquenched fraction, driven largely by galaxies in GroupMass 2 (Figure 15c) and Group Mass 3 (Figure 15d).The statistics in the lowest mass bins are understand-ably poor given the low numbers of early-types at thesestellar masses.Considering the late-types (Figure 16), we see thatfor M stellar < . M (cid:12) , late-types in groups are pref-erentially quenched – this is particularly clear in GroupMass 2 and Group Mass 3. This points to external (en-vironmental) processes acting on late-types and impact-ing their star formation. Comparison of Figure 16c and0 Cluver et al. log M stellar (M ) C o un t s Late UnquenchedLate QuenchedEarly UnquenchedEarly Quenched (a) Group Mass 1 ( < . M (cid:12) /h) log M stellar (M ) C o un t s Late UnquenchedLate QuenchedEarly UnquenchedEarly Quenched (b) Group Mass 2 (10 . – 10 . M (cid:12) /h) log M stellar (M ) C o un t s Late UnquenchedLate QuenchedEarly UnquenchedEarly Quenched (c) Group Mass 3 ( > . M (cid:12) /h) Figure 14.
The stellar mass distributions of the unquenched (unshaded) and quenched (shaded) systems, sub-divided intoearly- and late-type, for the three group mass bins. The increasing quenched fraction, of both early- and late-type systems,with increasing group mass is evident, although with a less dramatic shift from Group Mass 2 to Group Mass 3. An increase inquenched late-types is observed, particularly from Group Mass 1 to Group Mass 2.
Figure 16d indicates that the effect is most noticeablein Group Mass 3 in the 10 . < M stellar < . range.For masses M stellar > . , we curiously observe thatin our lowest halo mass bin, the quenched fraction oflate-types is lower than what is found for the non-G3Csample. This implies that in these mass bins the late-type galaxies are actually forming more stars and notless. For Group Mass 2 and 3, however, we see increased quenching of late-types with values that are broadly con-sistent with each other. Once again, larger samples andfiner divisions of halo mass will allow for a more defini-tive investigation.In this section we have observed the effect of group en-vironment on the star formation properties of galaxies.Group galaxies are quenched (i.e. have moved below theSFMS) relative to the ungrouped sample, at both high AMA-
WISE z < . Groups log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Early-typeG3C Early-type (a) non-G3C and G3C log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Early-typeGroup Mass 1 Early-type (b) Group Mass 1 ( < . M (cid:12) /h) log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Early-typeGroup Mass 2 Early-type (c) Group Mass 2 (10 . – 10 . M (cid:12) /h) log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Early-typeGroup Mass 3 Early-type (d) Group Mass 3 ( > . M (cid:12) /h) Figure 15.
The quenched fraction of early-types per stellar mass bin, comparing the non-G3C sample to a) the G3C sample,b) Group Mass 1, c) Group Mass 2, and d) Group Mass 3 samples, respectively. In each panel, the number of galaxies in eachbin is given at the bottom, with the non-G3C quenched galaxies in pink and the non-G3C unquenched in black. For the G3Csamples, quenched galaxies are in red and unquenched in blue. As in Figure 11, errors are calculated using bootstrap resamplingin each bin. Cluver et al. log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Late-typeG3C Late-type (a) non-G3C and G3C log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Late-typeGroup Mass 1 Late-type (b) Group Mass 1 ( < . M (cid:12) /h) log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Late-typeGroup Mass 2 Late-type (c) Group Mass 2 (10 . – 10 . M (cid:12) /h) log M stellar (M ) Q u e n c h e d F r a c t i o n non-G3C Late-typeGroup Mass 3 Late-type (d) Group Mass 3 ( > . M (cid:12) /h) Figure 16.
The quenched fraction of late-types per stellar mass bin, comparing the non-G3C sample to a) the G3C sample,b) Group Mass 1, c) Group Mass 2, and d) Group Mass 3 samples, respectively. In each panel, the number of galaxies in eachbin is given at the bottom, with the non-G3C quenched galaxies in pink and the non-G3C unquenched in black. For the G3Csamples, quenched galaxies are in red and unquenched in blue. As in Figure 11, errors are calculated using bootstrap resamplingin each bin.
AMA-
WISE z < . Groups stellar > . ), particularlyin high mass halos. The quenched fraction at low stellarmass M stellar < . is dominated by late-types andis observed (to varying degrees) in all halo mass bins.We find an intermediate population (in stellar mass) oflate-type, quenched systems suggesting an evolutionarypathway via the quenching of disk-dominated galaxies.3.4. Compactness
Although relatively rare in the local universe (Mc-Connachie et al. 2009), compact groups are consideredlaboratories of extreme (and seemingly rapid) evolution(Johnson et al. 2007; Walker et al. 2010), where interac-tions and merging pathways dominate (e.g. Barnes 1989;Verdes-Montenegro et al. 2001). For example, the old-est known compact group, Stephan’s Quintet (Stephan1877), offers a unique and pristine perspective on shockcooling pathways (e.g. Appleton et al. 2006; Cluver etal. 2010) and the role of turbulence in the suppressionof star formation (e.g. Guillard et al. 2012; Appleton etal. 2017).Hickson (1982) identified 100 compact groups in thePalomar Observatory Sky Survey, introducing a set ofthree criteria: a) richness, b) isolation, and c) com-pactness. This methodology has been used to con-struct similar catalogues using surveys such as theCOSMOS-UKST Southern Galaxy Catalogue (Iovino2002), 2MASS (D´ıaz-Gim´enez et al. 2012; D´ıaz-Gim´enez& Zandivarez 2015), and SDSS (McConnachie et al.2009). Compact groups have also been identified usingthe FoF algorithm on surveys such as the CfA2 (Bartonet al. 1996), and SDSS (Sohn et al. 2016); however, in-completeness due to fibre collsions particularly disruptsdense, compact structures, such as compact groups. Re-cent work has advocated for combining the two selectionmethods, for example, D´ıaz-Gim´enez et al. (2018) usingSDSS, and Zheng & Shen (2020) who combined redshiftinformation from SDSS, LAMOST, and GAMA.In this work we take an alternative approach and con-sider the on-sky “compactness” of our entire G3C groupsample, making use of the convex hull parameters in-cluded as part of the G3C catalog. A convex hull de-scribes the minimum 2D area (or 3D volume) that en-closes a collection of points, such that the surface is onlypermitted to bend inwards. Therefore, for each galaxygroup, an independent smooth surface is constructed,in comoving space, that contains all the galaxies in thegroup. This hull, therefore, has a 2D surface area andcontains a 3D volume. It can also be projected ontothe RA-Dec plane to reflect the projected area of the
Table 5.
Division of G3C Sample according toCompactness CriterionDesignation No. of Groups No. of GalaxiesLoose 125 793Nominal 251 1299Compact 118 469 group (in units of (Mpc/h) ) – for the G3C sample thisis referred to as the “d2radec” parameter.We investigate the distribution of the d2radec param-eter as a function of group membership and group halomass in section D of the Appendix. Unsurprisingly,when comparing d2radec and dynamical mass (see Fig-ure 24), an increase in halo mass broadly correlates toan increase in d2radec; i.e., more massive groups tendto have a larger projected area on the sky (i.e. in theRA-Dec plane) than smaller groups. We apply a linearfit to this distribution, which is described by:log d2radec (Mpc / h) = 1 .
01 log Mass dyn . (M (cid:12) / h) − . , (5)with an intrinsic spread of σ = 0 .
59, as shown in Fig-ure 24 of section D of the Appendix. We use this todefine a compactness criterion where groups that lie inthe top quartile defined by this relation are designated“loose”, whereas those in the bottom quartile are des-ignated “compact”. We note that the largest point ofdifference to traditional compact groups is the richnesscriterion that requires that at least 4 member galaxiesare of similar brightness (within 3 magnitudes of thebrightest member), and therefore mass. Our definitionis purposefully more general and can therefore not beconsidered true compact groups (in the Hickson 1982,sense), and should not be compared as such. We referto groups that lie between the top and bottom quartileas “nominal”, i.e. they are as compact on the sky as wewould expect from their group halo mass. The numbersof groups and the corresponding number of galaxies ineach division is given in Table 5.In order to examine the SFR properties, we use thequenching separator as before and show in Figure 17the relative proportions of early- and late-types that areforming stars at an expected rate (unquenched), andthose that have fallen below that threshold (quenched).Given the low numbers in our loose and compact sam-ples, we increase our stellar mass bin size to 0.25 dexcompared to previous sections.4
Cluver et al. log M stellar (M ) C o un t s Late Unquenched (Loose)Late Quenched (Loose)Early Unquenched (Loose)Early Quenched (Loose) (a) Loose Groups log M stellar (M ) C o un t s Late Unquenched (Nominal)Late Quenched (Nominal)Early Unquenched (Nominal)Early Quenched (Nominal) (b) Nominal Groups log M stellar (M ) C o un t s Late Unquenched (Compact)Late Quenched (Compact)Early Unquenched (Compact)Early Quenched (Compact) (c) Compact Groups
Figure 17.
The stellar mass distributions of the unquenched (unshaded) and quenched (shaded) systems, sub-divided intoearly- and late-type, for the loose, nominal and compact group delineations described in the text. The loose and nominal stellarmass distributions of unquenched late-types and quenched early-types appear fairly similar, however, the compact systems arestrongly peaked at M stellar ∼ . M (cid:12) in both the late unquenched and early quenched populations. Comparing the loose and nominal groups, the stellarmass distributions of late-type, unquenched and early-type, quenched systems appear similar. The compactgroup galaxies, however, clearly have a larger propor-tion of early-type, quenched systems, relative to theirlate, unquenched population. This is reflected in Table6 where those populations have roughly equal numbers, compared to the loose and nominal groups, albeit withlarge uncertainties due to the small sample size.Considering only the high stellar mass population(M stellar > . M (cid:12) ), both early- and late-type, thecompact groups have relatively more galaxies at highstellar mass (45%), relative to the loose (37%) andnominal groups (39%). Considering the whole sample,they have an increased proportion of early-types, both AMA-
WISE z < . Groups log M stellar (M ) Q u e n c h e d F r a c t i o n Nominal GroupsLoose Groups (a) Loose vs Nominal Groups log M stellar (M ) Q u e n c h e d F r a c t i o n Nominal GroupsCompact Groups (b) Compact vs Nominal Groups
Figure 18.
The quenched fraction for (a) loose vs nominal and (b) compact vs nominal groups, as defined by the compactnesscriterion outlined in the text. In each panel, the number of galaxies in each bin is given at the bottom, with the nominal quenchedgalaxies in pink, and nominal unquenched galaxies in black. The red text indicates the number of quenched galaxies in eitherloose (panel a) or compact (panel b) groups. The numbers of unquenched loose galaxies (panel a) and unquenched compactgalaxies (panel b) are given in blue. As previously indicated, errors are calculated using bootstrap resampling in each bin. Fromthis it appears that quenching efficiency is less impacted by compactness, as compared to morphological transformation.
Table 6.
Unquenched and Quenched, Early- and Late-types in the Loose,Nominal and Compact SamplesSample Unquenched Quenched Unquenched QuenchedLate-type Late-type Early-type Early-typeLoose 42.8( ± ± ± ± ± ± ± ± ± ± ± ± quenched and unquenched (44%) compared to the loose(41%) and nominal (41%) groups. This could indicateincreased tidal interactions in the more compact sys-tems, promoting the growth of bulges (via inside-outgrowth) and overall stellar mass. This is consistent withsimulations (e.g. Brasseur et al. 2009) and observations(e.g. Deng et al. 2008; Walker et al. 2010) indicatingincreased fractions of early-type galaxies in traditionalcompact groups, compared to typical groups and un-grouped galaxies. However, within the errors, we do notsee any significant differences in the overall fractions ofquenched systems when comparing the loose, nominaland compact populations. We investigate this further in Figure 18 where we showthe quenched fractions for nominal, loose and compactgroups in each stellar mass bin. Although our samplesof loose and compact group galaxies are small, we seethat for M stellar < . M (cid:12) , neither loose nor compactgroups show evidence for increased quenching. However,for M stellar > . M (cid:12) , we see that there may be dif-ferences compared to nominal groups, but this requiresfurther investigation. DISCUSSIONOur aim in this study is to examine the mid-infrared(
WISE ) and morphological properties of aggregate sam-ples of galaxies: ungrouped, grouped, differing in halo6
Cluver et al. mass, and varying in compactness. We have shown that
WISE is well-suited to examine the mass and star forma-tion properties of galaxies as a function of environment,in the z < . WISE W2 − W3 color of the G3Csample (Figure 3) appears to be largely driven by themorphological composition of this population. In thenon-G3C sample, by comparison, we observe a dearth ofgalaxies at low W2 − W3 (the mid-IR blue end), whichcan be attributed to a lack of high-mass, early-typegalaxies. Within the G3C sample, increasing halo masscorresponds to a larger proportion of early-type galax-ies at high stellar mass (Figure 5). These trends can belargely understood in the context of the morphology-density relation, driven by the variation of the stel-lar mass function with halo mass (e.g. Yang et al.2009; Alpaslan et al. 2015) combined with the stellarmass function of morphological types (e.g. Moffett et al.2016), which produces a change in the mix of disk- andspheroid-dominated morphologies in the denser groupenvironment (see also Grootes et al. 2017), and is con-sistent with the findings of, for example, Bluck et al.(2014); Liu et al. (2019). In Figure 6, however, wesee tentative evidence for a changing early-type frac-tion in stellar mass bins 10 (cid:46) M stellar (cid:46) M (cid:12) , aswell as an overall increased fraction of early-types whenconsidering M stellar > M (cid:12) in increasing halo massbins. This suggests that stellar mass is not solely drivingthe morphology-density relation (see e.g. Bamford et al.2009).Compared to the SFMS determined for ungroupedgalaxies (Figure 7), we find that a similarly constructedSFMS in grouped galaxies (Figure 8) shows a slightlysteeper slope, suggesting increased star formation asa function of mass in this population. This may beconnected to the behaviour we observe in Figure 16bwhere the quenched fraction of late-types in our low-est halo mass bin, show a lower quenched fraction com-pared to the ungrouped sample in the stellar mass rangeM stellar ≥ . M (cid:12) .Considering systems that have moved below theSFMS, we find an increase in quenched fraction withincreasing halo mass (Figure 11 and Table 3), con-sistent with the increase in passive fraction observedin other studies (e.g. Brinchmann et al. 2004; Blucket al. 2014; Davies et al. 2019b). The quenched frac-tion is dominated by early-types at high mass (Figure13 and Table 4), in line with previous findings (e.g.Bluck et al. 2014; Liu et al. 2019) and consistent with avarying stellar mass function combined with the mass-morphology mapping. The overall increase in the frac-tion of quenched, early-types with increasing halo mass is clearly aligned to an increase in high stellar masssystems, in line with secular processes.However, the pathways that lead to this behaviourare not as clear. We find evidence of an increase inthe quenched population at M stellar < . M (cid:12) (Fig-ure 11), which appears to be driven by the quenching oflate-types at these stellar masses, particularly in moremassive halos (Figure 16), which we expect from envi-ronmental (i.e. external) quenching mechanisms oper-ating in this regime (see also Peng et al. 2010; Davieset al. 2019b; Liu et al. 2019; Li et al. 2020).Although we see a clear separation in the stellar massdistributions of unquenched, disk-dominated galaxiesand quenched, bulge-dominated galaxies across halomass, the population of quenched, late-types (Figure14) at intermediate stellar mass (where we expect stel-lar mass quenching to become important; Peng et al.2010) could be the product of environmental processesand may in the future transition to early-type morpholo-gies as a result of mergers (see also Mutch et al. 2011).This population increases from Group Mass 1 to GroupMass 2, but does not change significantly when movingto the largest group mass bin (see Table 4). This sug-gests that the processes forming this population do notnecessarily become more efficient in larger mass grouphalos, possibly due to other mechanisms becoming moredominant. Larger samples would be needed to track thebehavior of this population as a function of group halomass, with the inclusion of cold and hot gas measuresproviding a more definitive picture of their evolutionarystate.Turning to the SFMS itself, the lack of unquenchedearly-types i.e. early-types on the sequence (Figure 12and 13) is in agreement with recent work from Wang,B. et al. (2020), using SDSS-IV MaNGA, who find thatthe SFMS is dominated by spirals with small bulges.The lack of increase (in fact, decrease) of the early-type,unquenched population with increasing halo mass (Fig-ure 14, Table 4), when the overall number of early-typesis increasing, suggests that these galaxies are predom-inantly found below the SFMS, in line with previousstudies (e.g. Bluck et al. 2014; Cook et al. 2020). How-ever, this does not imply a physical causation betweenthe presence of a bulge and the quenching of star forma-tion (e.g. Bluck et al. 2014; Lilly & Carollo 2016; Wang,E. et al. 2018; Cook et al. 2019).When considering group compactness, our resultssuggest that relatively compact groups have a tendencyto host a larger fraction of high-mass and early-typesystems, but only reflect a small increase in over-all quenched fraction compared to loose and nominalgroups (Figure 17 and Table 6). They do, however, AMA-
WISE z < . Groups i masscontent; Hess & Wilcots 2013), as the observed changesin quenched fraction can be an expected consequence ofan increase in early-type fraction, itself a consequenceof the changing stellar mass distribution with increasinghalo mass.Although it is tempting to connect the changing prop-erties we observe across environments to mechanismsoperating as a function of density, such as interactionsand mergers, the differing formation histories of galax-ies makes direct comparisons infeasible. For example,galaxies in relatively large group halo masses are ex-pected to have experienced more mergers, leading to anincreased proportion of early-types at high stellar mass(e.g. Rodriguez-Gomez et al. 2016; Deeley et al. 2017).However, the observed merger fraction in the local uni-verse does not appear to be more than ∼
5% (e.g. Darget al. 2010; Robotham et al. 2014). In addition, ha-los in dense environments form earlier than halos of thesame mass in less dense environments, and are thereforesubject to “assembly bias” (e.g. Sheth & Tormen 2004;Croton et al. 2007; Wilman et al. 2013).Observationally, we have the added conundrum thatonly by identifying the real progenitors of galaxies inour samples can we study how galaxies have been trans-formed by environment. This “progenitor bias” meansthat changes in populations can be driven by changesin membership, rather than through changes in individ-ual members (e.g. van Dokkum & Franx 2001; Carolloet al. 2013; Cortese et al. 2019). In fact, it has beenshown that observed correlations between galaxy struc-ture and the quenching of galaxies can be explained asa consequence of the sizemass relation for star-forminggalaxies (Lilly & Carollo 2016). And we note that in thework of Cortese et al. (2019), current numerical simu-lations, such as EAGLE (Schaye et al. 2015), indicatethat pre-processing has a limited effect on the struc-tural properties of galaxies (but, see also De Lucia et al.2012).Observationally, recent studies have found growing ev-idence for a transitional or “characteristic” mass asso-ciated with environmental processes (Wang, E. et al.2018, 2020; Li et al. 2020). The characteristic stellarmass (M ∗ , ch ) (Li et al. 2020) is a function of the halomass of the group and the mass of the central; it implies that the quenching of galaxies is not driven by a simplecentral-satellite dichotomy, but rather by the interac-tions between internal and external processes (Wang, E.et al. 2020; Li et al. 2020). In this paradigm, galax-ies above M ∗ , ch quench by internal (secular) processesand not environmental processes. Conversely, galaxiesbelow M ∗ , ch are more likely to quench due to external,environmental processes. In addition, it is chiefly thegalaxies above the characteristic mass that are buildingstellar mass by mergers and tidal interactions (but seealso Joshi et al. 2020). Li et al. (2020) also investi-gates how M ∗ , ch relates to quenched fraction as a func-tion of bulge-to-total ratio and location within the halo;their findings lend support to the physically-motivateddichotomy that arises from M ∗ , ch .Forthcoming H i interferometric surveys will likelyplay a key role in progressing our understanding ofenvironmental processes; by linking the detailed spa-tial information of cold gas to environmental measuresand multiwavelength data, morphological transforma-tion and pre-processing can be identified and studiedas it takes place. Further to that, linking these studiesto the relative location of dense structures within thecosmic web will enable a more comprehensive view ofthe role of secular versus environmental processes. CONCLUSIONSMaking use of the high fidelity group properties andvisual morphologies of the GAMA survey, we show that
WISE can be used effectively to investigate the colors,star formation, and stellar mass of galaxies in differ-ent environments (measured as halo/dynamical mass)to z < .
1. We summarise our findings as: • The G3C population (i.e. galaxies within groups;4 ≤ Nfof ≤
20) clearly show different
WISE color,stellar mass, and morphological composition com-pared to the ungrouped (non-G3C) sample. Thechanging stellar mass function with increasedgroup halo mass is also evident. We find tentativeevidence of an overall increasing early-type frac-tion with increasing halo mass (when consideringsystems with M stellar > M (cid:12) ) and also in stel-lar mass bins between 10 (cid:46) M stellar (cid:46) M (cid:12) that may suggest that the morphology-densityrelation is not purely a consequence of the mass-morphology relation and a varying stellar massfunction. • We determine a SFMS (using late-type, ungroupedgalaxies, with
WISE colors of W2 − W3 >
3) of the8
Cluver et al. form:log SFR(M (cid:12) / yr − ) = 0 .
93 log M stellar (M (cid:12) ) − . , (6)and use this to define a quenching separator, delin-eating star-forming galaxies on the sequence, fromthose that are transitioning to being passive sys-tems. • Using the quenching separator we show that withincreasing halo mass, there is an accompanied in-crease in high mass (M stellar > . M (cid:12) ) early-type systems that have moved off the star-formingsequence, in line with mass quenching. • We also find evidence of an increase in thequenched fraction of galaxies in groups in themass range M stellar < . M (cid:12) , indicative ofenvironmental quenching processes acting on late-type systems. This effect appears more significantin our most massive halo, but requires a largersample for more robust statistics. • We observe quenched, late-type galaxies to forman intermediate population in stellar mass betweenthe late, unquenched and early, quenched samplesconsistent with an evolutionary pathway wheredisk galaxies experience declining star formationon the way to the red sequence; i.e. in addition tothe paradigm of gas-rich disk galaxies merging toform the red sequence. • Galaxies in groups that are compact, compared tothe aggregate relation of compactness versus halomass, have a higher proportion of early-type andhigh mass systems overall, but reflect a similarfraction of quenched galaxies compared to nom-inally compact groups.Expanding to a larger z < . i Pathfinders will be pioneers in this phasespace. For galaxy groups, measuring the baryon con-tent locked in hot gas (e.g. from eROSITA) will allowfor detailed studies of the baryon cycle as a function ofenvironment.However, the robust determination of environmentmetrics through highly complete redshift information will be the limiting factor to extending this kind of studyto larger volumes with the statistical fidelity needed tostudy the detailed pathways by which the cosmic webforms and galaxies are built.In the interim, our objective has been to show that
WISE is well-suited to galaxy evolution studies in thelocal ( z < .
1) universe. We have shown that environ-ment is correlated with mass. And also that at fixedmass, environment correlates with the relative numberof quenched vs. unquenched galaxies. And moreover,again at fixed mass, environment correlates with theproperties of star-forming galaxies. This demonstratesthe difficulty in disentangling the different ways in whichgalaxies are shaped by their environments, and showsthe need for very careful analysis of large galaxy sam-ples.
AMA-
WISE z < . Groups A. PHOTOMETRIC DATA PROPERTIESAs detailed in section 2.4, we have applied a signal to noise (S/N) photometric quality cut in W1 − W2 color (S/N > − W2 sources dominate the low mass end (M stellar < M (cid:12) )in both samples. This is understandable given that these systems lack an abundant old stellar population that givesrise to the near-infrared light we are using to trace stellar mass.Figure 19 additionally shows the distribution of low S/N and upper limit SFRs within the Primary Sample (dashedhistogram) and Excluded Systems (shaded region). This shows that the Excluded Systems are dominated by galaxieswith less well-determined SFRs – this is due to the low SFRs of these low mass galaxies, which rapidly drop beyond thesensitivity of the WISE
W3 band. Adding these sources to our Primary Sample would therefore further increase thenoise within each mass bin, in addition to the noise across mass bins introduced by relaxing the stellar mass robustnessrequirement. log M stellar (M ) C o un t s Primary SampleLow S/N & U/L SFRsExcluded SystemsLow S/N & U/L SFRs (a) non-G3C Sample log M stellar (M ) C o un t s Primary SampleLow S/N & U/L SFRsExcluded SystemsLow S/N & U/L SFRs (b) G3C Sample
Figure 19.
The distribution in
WISE stellar mass of the Primary Sample (dark line) and those systems excluded on the basisof the error in stellar mass being larger than 0.5 dex (Excluded Systems; light line) as discussed in the text. We note that bydefinition the stellar masses of the Excluded Systems, and therefore their true distribution, are highly uncertain. The dashedlines reflects the distribution of sources within the Primary Sample that have low S/N and upper limit SFRs. Similarly, theshaded histogram shows that the low S/N and upper limit SFRs dominate the Excluded Systems.
Although this reduces the number of galaxies at the lowest stellar masses (particularly for the non-G3C sample) andwill limit our statistical power in these mass bins, excluding them does not influence our analysis as we expect little tono evolution to z = 0 . − W2); this is shown inFigure 19. We find that our results are highly consistent with what is found in Figure 10b, examining in particularthe low mass end where any differences would manifest.0
Cluver et al. log M stellar (M ) Q u e n c h e d F r a c t i o n Figure 20.
Replicating Figure 10b, we calculate quenched fraction in each mass bin for the non-G3C (black) and G3C galaxies(orange), but including the Excluded Systems with stellar mass being larger than 0.5 dex. The numbers at the bottom of theplot reflect the galaxies found in each bin with non-G3C quenched in pink, non-G3C unquenched in black, G3C quenched inred, and G3C unquenched in blue. Errors are calculated using bootstrap resampling in each bin.B.
MASS PROPERTIES OF THE NON-G3C AND G3C SAMPLESWe include the distribution of dynamical (halo) mass and group membership of our G3C sample in Figure 21. It isevident that groups of membership 4 dominate our sample and show the broadest range in dynamical mass, consistentwith the G3C analysis presented in Robotham et al. (2011). Groups with membership 10 < Nfof ≤
20 are largelylimited to having halo masses of log Dynamical Mass > . (cid:12) /h. As shown in Robotham et al. (2011), the accuracyof the halo mass derived from dynamical mass estimates after applying a scaling factor, are median unbiased for Nfof ≥
4. However, the standard deviation of the distribution increases strongly as a function of decreasing multiplicity,as given by Equation (20) in Robotham et al. (2011). For our sample, this ranges from 0.45 – 0.74 dex and is largelydriven by uncertainties in the derived velocity dispersions for low group membership. It should therefore be borne inmind that the halo masses derived from the dynamical estimates are susceptible to scatter, particularly for the lowgroup memberships that dominate our sample. However, our halo mass bins are chosen to be purposefully large tolessen the impact of this scatter. C. THE STAR FORMATION – STELLAR MASS DIAGRAM IN HALO MASS AND BULGE-TO-DISKDECOMPOSITIONIn Figure 14 we presented the log SFR– log M stellar distribution for the (a) non-G3C and (b) G3C samples, makinguse of our visual morphological classifications to separate early- and late-type. Detailed bulge-to-disk comparisonshave been performed on the KiDS g, r, and i -band imaging (de Jong et al. 2017) for z < .
08 galaxies in the GAMA IIequatorial survey regions using ProFit (Robotham et al. 2017) and are provided as part of the BDDecompv03 DMU(Casura et al., in prep.). We make use of these measurements to investigate the consistency of our results, albeit witha redshift subset of our sample. In Figure 22 we have grouped galaxies with a single-fit S´ersic index of > . > . ≤ . ≤ .
5. Compared to Figure 12, we findgood agreement with the morphological composition and distribution of the non-G3C and G3C samples, even withthe smaller subset of systems.
AMA-
WISE z < . Groups log Dynamical Mass (M /h) G r o u p M e m b e r s h i p C o un t s Group Sample
Counts
Figure 21.
The distribution of dynamical (halo) mass and group membership in our z < . log M stellar (M ) l o g S F R ( M y r ) Non-G3C Early-typelow S/Nupper limitsNon-G3C Late-typelow S/Nupper limits C o un t s Non-G3C Early-typeNon-G3C Late-type 0 100
Counts (a) non-G3C Sample log M stellar (M ) l o g S F R ( M y r ) Early-type G3Clow S/Nupper limitsLate-type G3Clow S/Nupper limits C o un t s G3C Early-typeG3C Late-type 0 20 40
Counts (b) G3C Sample
Figure 22.
The log SFR–log M stellar distribution for the (a) non-G3C and (b) G3C sample using the bulge-to-total decompo-sitions of Casura et al. (in prep.), available for galaxies in our sample to z < .
08. To make this comparison we have groupedgalaxies with a single-fit S´ersic index of > . > . ≤ . ≤ .
5. Although this isa subset of our full sample, we see a similar distribution to that of Figure 12, indicating that the visual morphology classificationsand detailed decompositions broadly agree. Cluver et al. D. COMPACTNESS log d2radec (Mpc/h) G r o u p M e m b e r s h i p C o un t s Group Galaxies
Counts
Figure 23.
Group Membership of the G3C sample as a function of the convex hull parameter, d2radec. As group membershipincreases, the d2radec parameter tends to increase, reflecting a larger projection of the group on the sky.
In this section we investigate the convex hull parameter, d2radec, introduced in Section 3.4 of the paper. In Figure23 we plot group membership as a function of d2radec, which shows that as group membership increases, the projectedarea of the group on the sky tends to be larger. We note that groups of membership 4 show a broad range in thed2radec parameter, which is consistent with the large spread in halo mass observed in Figure 21.
10 11 12 13 14 log Halo Mass (M /h) l o g d r a d e c ( M p c / h ) Group Mass 1Group Mass 2Group Mass 3
Figure 24.
The d2radec parameter as a function of dynamical (halo) mass, color-coded by group mass bin (Table 2). The solidline in indicates the best fit, with dashed lines indicating the division into top and bottom quartile.
In Figure 24 we examine the d2radec parameter as a function of dynamical mass (that goes as the square of thevelocity dispersion), which shows a clear correlation. We determine a best-fit to the distribution, given by:log d2radec (Mpc / h) = 1 .
01 log Mass dyn . (M (cid:12) / h) − . , (D1) AMA-
WISE z < . Groups σ = 0 . WISE and NEOWISE are funded by theNational Aeronautics and Space Administration. This research has made use of and python packages: astropy (Astropy Collaboration et al. 2018, 2013), matplotlib http://matplotlib.org/ (Hunter2007),
NumPy
SciPy
Facilities:
WISE, AAT, VST REFERENCES
Alpaslan, M., Driver, S., Robotham, A. S. G., et al. 2015,MNRAS, 451, 3249Appleton, P. N., Guillard, P., Togi, A., et al. 2017, ApJ,836, 76Appleton, P. N., Xu, K. C., Reach, W., et al. 2006, ApJL,639, L51Aragon Calvo, M. A., Neyrinck, M. C., & Silk, J. 2019, TheOpen Journal of Astrophysics, 2, 7Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz, B. M.,et al. 2018, AJ, 156, 123Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., etal. 2013, A&A, 558, A33Baldry, I. K., Liske, J., Brown, M. J. I., et al. 2018,MNRAS, 474, 3875Balogh, M. L., McGee, S. L., Mok, A., et al. 2016, MNRAS,456, 4364Baldry, I. K., Alpaslan, M., Bauer, A. E., et al. 2014,MNRAS, 441, 2440Baldry, I. K., Balogh, M. L., Bower, R. G., et al. 2006,MNRAS, 373, 469Baldry, I. K., Glazebrook, K., Brinkmann, J., et al. 2004,ApJ, 600, 681 Balogh, M. L., Baldry, I. K., Nichol, R., et al. 2004, ApJL,615, L101Balogh, M. L., Schade, D., Morris, S. L., et al. 1998, ApJL,504, L75Barnes, J. E. 1989, Nature, 338, 123Barnes, J. E., & Hernquist, L. 1992, ARA&A, 30, 705Barton, E., Geller, M., Ramella, M., et al. 1996, AJ, 112,871Barsanti, S., Owers, M. S., Brough, S., et al. 2018, ApJ,857, 71Bamford, S. P., Nichol, R. C., Baldry, I. K., et al. 2009,MNRAS, 393, 1324Behroozi, P. S., Wechsler, R. H., & Conroy, C. 2013, ApJL,762, L31Bianconi, M., Smith, G. P., Haines, C. P., et al. 2018,MNRAS, 473, L79Birnboim, Y., & Dekel, A. 2003, MNRAS, 345, 349Blanton, M.R. & Moustakas, J., 2009,
ARA& A , 47, 159Blanton, M. R., Hogg, D. W., Bahcall, N. A., et al. 2003,ApJ, 594, 186Bluck, A. F. L., Mendel, J. T., Ellison, S. L., et al. 2016,MNRAS, 462, 2559 Cluver et al.
Bluck, A. F. L., Mendel, J. T., Ellison, S. L., et al. 2014,MNRAS, 441, 599Brasseur, C. M., McConnachie, A. W., Ellison, S. L., et al.2009, MNRAS, 392, 1141Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004,MNRAS, 351, 1151Brisbin, D., Aravena, M., Daddi, E., et al. 2019, A&A, 628,A104Brough, S., Croom, S., Sharp, R., et al. 2013, MNRAS, 435,2903Brown, M. J. I., Jarrett, T. H., & Cluver, M. E. 2014,PASA, 31, e049Brown, M. J. I., Moustakas, J., Smith, J.-D. T., et al. 2014,ApJS, 212, 18Carollo, C. M., Bschorr, T. J., Renzini, A., et al. 2013,ApJ, 773, 112Chabrier, G. 2003, PASP, 115, 763Cluver, M. E., Jarrett, T. H., Dale, D. A., et al. 2017, ApJ,850, 68Cluver, M. E., Jarrett, T. H., Hopkins, A. M., et al. 2014,ApJ, 782, 90Cluver, M. E., Appleton, P. N., Boulanger, F., et al. 2010,ApJ, 710, 248Contini, E., Gu, Q., Ge, X., et al. 2020, arXiv e-prints,arXiv:2001.01369Cook, R. H. W., Cortese, L., Catinella, B., et al. 2020,MNRAS, 493, 5596Cook, R. H. W., Cortese, L., Catinella, B., et al. 2019,MNRAS, 490, 4060Cortese, L., van de Sande, J., Lagos, C. P., et al. 2019,MNRAS, 485, 2656Cortese, L., Gavazzi, G., Boselli, A., et al. 2006, A&A, 453,847Couch, W. J., Balogh, M. L., Bower, R. G., et al. 2001,ApJ, 549, 820Croton, D. J., Springel, V., White, S. D. M., et al. 2006,MNRAS, 365, 11Croton, D. J., Gao, L., & White, S. D. M. 2007, MNRAS,374, 1303Daddi, E., Dickinson, M., Morrison, G., et al. 2007, ApJ,670, 156Dale, D. A., Cook, D. O., Roussel, H., et al. 2017, ApJ,837, 90Darg, D. W., Kaviraj, S., Lintott, C. J., et al. 2010,MNRAS, 401, 1552Davies, L. J. M., Lagos, C. del P., Katsianis, A., et al. 2019,MNRAS, 483, 1881Davies, L. J. M., Robotham, A. S. G., Lagos, C. del P., etal. 2019, MNRAS, 483, 5444 Davies, L. J. M., Robotham, A. S. G., Driver, S. P., et al.2016, MNRAS, 455, 4013Davies, L. J. M., Robotham, A. S. G., Driver, S. P., et al.2015, MNRAS, 452, 616Davis, M., Efstathiou, G., Frenk, C. S., et al. 1985, ApJ,292, 371De Lucia, G., Weinmann, S., Poggianti, B. M., et al. 2012,MNRAS, 423, 1277Deeley, S., Drinkwater, M. J., Cunnama, D., et al. 2017,MNRAS, 467, 3934Dekel, A., & Birnboim, Y. 2006, MNRAS, 368, 2Deng, X.-F., He, J.-Z., Ma, X.-S., et al. 2008, CentralEuropean Journal of Physics, 6, 185De Propris, R., Baldry, I. K., Bland-Hawthorn, J., et al.2014, MNRAS, 444, 2200D´ıaz-Gim´enez, E., Zandivarez, A., & Taverna, A. 2018,A&A, 618, A157D´ıaz-Gim´enez, E., & Zandivarez, A. 2015, A&A, 578, A61D´ıaz-Gim´enez, E., Mamon, G. A., Pacheco, M., et al. 2012,MNRAS, 426, 296Dressler, A. 1980, ApJ, 236, 351Driver, S. P., et al. 2019, The Messenger, 175, 56Driver, S. P., Wright, A. H., Andrews, S. K., et al. 2016,MNRAS, 455, 3911Driver, S. P., Robotham, A. S. G., Kelvin, L., et al. 2012,MNRAS, 427, 3244Driver, S. P., Hill, D. T., Kelvin, L. S., et al. 2011,MNRAS, 413, 971Driver, S. P., Allen, P. D., Graham, A. W., et al. 2006,MNRAS, 368, 414Dˇzudˇzar, R., Kilborn, V., Murugeshan, C., et al. 2019,MNRAS, 490, L6Edge, A., Sutherland, W., Kuijken, K., et al. 2013, TheMessenger, 154, 32Eke, V. R., Frenk, C. S., Baugh, C. M., et al. 2004,MNRAS, 355, 769Elbaz, D., Daddi, E., Le Borgne, D., et al. 2007, A&A, 468,33Farouki, R., & Shapiro, S. L. 1981, ApJ, 243, 32Fujita, Y. 2004, PASJ, 56, 29Gabor, J. M., & Dav´e, R. 2015, MNRAS, 447, 374Giovanelli, R., Haynes, M. P., Kent, B. R., et al. 2005, AJ,130, 2598Goto, T., Yamauchi, C., Fujita, Y., et al. 2003, MNRAS,346, 601Grootes, M. W., Dvornik, A., Laureijs, R. J., et al. 2018,MNRAS, 477, 1015Grootes, M. W., Tuffs, R. J., Popescu, C. C., et al. 2017,AJ, 153, 111
AMA-
WISE z < . Groups Guillard, P., Boulanger, F., Pineau des Forˆets, G., et al.2012, ApJ, 749, 158Gunn, J. E., & Gott, J. R. 1972, ApJ, 176, 1Haines, C. P., Pereira, M. J., Smith, G. P., et al. 2015, ApJ,806, 101Hess, K. M., & Wilcots, E. M. 2013, AJ, 146, 124Hickson, P. 1982, ApJ, 255, 382Hunter, J. D. 2007, Computing in Science and Engineering,9, 90Hopkins, A. M., Driver, S. P., Brough, S., et al. 2013,MNRAS, 430, 2047Janowiecki, S., Catinella, B., Cortese, L., et al. 2020,MNRAS, 493, 1982Jarrett, T. H., Cluver, M. E., Brown, M. J. I., et al. 2019,ApJS, 245, 25Jarrett, T. H., Cluver, M. E., Magoulas, C., et al. 2017,ApJ, 836, 182Jarrett, T. H., Masci, F., Tsai, C. W., et al. 2013, AJ, 145, 6Jarrett, T. H., Masci, F., Tsai, C. W., et al. 2012, AJ, 144,68Jarrett, T. H., Cohen, M., Masci, F., et al. 2011, ApJ, 735,112Johnson, K. E., Hibbard, J. E., Gallagher, S. C., et al.2007, AJ, 134, 1522Joshi, G. D., Pillepich, A., Nelson, D., et al. 2020, arXive-prints, arXiv:2004.01191de Jong, J. T. A., Verdoes Kleijn, G. A., Erben, T., et al.2017, A&A, 604, A134Just, D. W., Kirby, M., Zaritsky, D., et al. 2019, ApJ, 885, 6Iovino, A. 2002, AJ, 124, 2471Kauffmann, G., Heckman, T. M., White, S. D. M., et al.2003, MNRAS, 341, 54Kelvin, L. S., Driver, S. P., Robotham, A. S. G., et al.2014, MNRAS, 444, 1647Kelvin, L. S., Driver, S. P., Robotham, A. S. G., et al.2014, MNRAS, 439, 1245Kennicutt, R. C. 1998, ARA&A, 36, 189Kraljic, K., Arnouts, S., Pichon, C., et al. 2018, MNRAS,474, 547Kroupa, P. 2001, MNRAS, 322, 231Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980, ApJ,237, 692Lawrence, A., Warren, S. J., Almaini, O., et al. 2007,MNRAS, 379, 1599Li, P., Wang, H., Mo, H. J., et al. 2020, arXiv e-prints,arXiv:2003.09776Lilly, S. J., & Carollo, C. M. 2016, ApJ, 833, 1Liske, J., Baldry, I. K., Driver, S. P., et al. 2015, MNRAS,452, 2087Liu, C., Hao, L., Wang, H., et al. 2019, ApJ, 878, 69 Lewis, I., Balogh, M., De Propris, R., et al. 2002, MNRAS,334, 673McConnachie, A. W., Patton, D. R., Ellison, S. L., et al.2009, MNRAS, 395, 255Marcillac, D., Rigby, J. R., Rieke, G. H., et al. 2007, ApJ,654, 825Martig, M., Bournaud, F., Teyssier, R., et al. 2009, ApJ,707, 250Mihos, J. C. 2004, Recycling Intergalactic and InterstellarMatter, 390Moffett, A. J., Ingarfield, S. A., Driver, S. P., et al. 2016,MNRAS, 457, 1308Moore, B., Katz, N., Lake, G., et al. 1996, Nature, 379, 613Moreno, J., Torrey, P., Ellison, S. L., et al. 2019, MNRAS,485, 1320Mutch, S. J., Croton, D. J., & Poole, G. B. 2011, ApJ, 736,84Noeske, K. G., Weiner, B. J., Faber, S. M., et al. 2007,ApJL, 660, L43Nulsen, P. E. J. 1982, MNRAS, 198, 1007Peng, Y., Maiolino, R., & Cochrane, R. 2015, Nature, 521,192Peng, Y.-. jie ., Lilly, S. J., Renzini, A., et al. 2012, ApJ,757, 4Peng, Y.-j., Lilly, S. J., Kovaˇc, K., et al. 2010,
ApJ , 721, 193Porter, S. C., Raychaudhury, S., Pimbblet, K. A., et al.2008, MNRAS, 388, 1152Posti, L., Fraternali, F., & Marasco, A. 2019, A&A, 626,A56Postman, M., & Geller, M. J. 1984, ApJ, 281, 95Rasmussen, J., Ponman, T. J., & Mulchaey, J. S. 2006,MNRAS, 370, 453Renzini, A., & Peng, Y.-. jie . 2015, ApJL, 801, L29Roberts, I. D., & Parker, L. C. 2017, MNRAS, 467, 3268Robotham, A. S. G., Taranu, D. S., Tobar, R., et al. 2017,MNRAS, 466, 1513Robotham, A. S. G., Liske, J., Driver, S. P., et al. 2013,MNRAS, 431, 167Robotham, A. S. G., & Obreschkow, D. 2015, PASA, 32,e033Robotham, A. S. G., Driver, S. P., Davies, L. J. M., et al.2014, MNRAS, 444, 3986Robotham, A. S. G., Norberg, P., Driver, S. P., et al. 2011,MNRAS, 416, 2640Robotham, A., Phillipps, S., & de Propris, R. 2010,MNRAS, 403, 1812Rodriguez-Gomez, V., Pillepich, A., Sales, L. V., et al.2016, MNRAS, 458, 2371Salim, S., Rich, R. M., Charlot, S., et al. 2007, ApJS, 173,267 Cluver et al.
Sarron, F., Adami, C., Durret, F., et al. 2019, A&A, 632,A49Schaye, J., Crain, R. A., Bower, R. G., et al. 2015,MNRAS, 446, 521Schechter, P. 1976, ApJ, 203, 297Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ,500, 525Schaefer, A. L., Croom, S. M., Scott, N., et al. 2019,MNRAS, 483, 2851Schaefer, A. L., Croom, S. M., Allen, J. T., et al. 2017,MNRAS, 464, 121Sheth, R. K., & Tormen, G. 2004, MNRAS, 350, 1385Sohn, J., Geller, M. J., Hwang, H. S., et al. 2016, ApJS,225, 23Speagle, J. S., Steinhardt, C. L., Capak, P. L., et al. 2014,ApJS, 214, 15Strateva, I., Ivezi´c, ˇZ., Knapp, G. R., et al. 2001, AJ, 122,1861Stephan, M. 1877, MNRAS, 37, 334Stern, D., Assef, R. J., Benford, D. J., et al. 2012, ApJ,753, 30Taylor, E. N., Hopkins, A. M., Baldry, I. K., et al. 2015,MNRAS, 446, 2144Taylor, E. N., Hopkins, A. M., Baldry, I. K., et al. 2011,MNRAS, 418, 1587Toomre, A., & Toomre, J. 1972, ApJ, 178, 623van de Voort, F., Bah´e, Y. M., Bower, R. G., et al. 2017,MNRAS, 466, 3460Vazquez-Mata, J.A., Loveday, J., Baldry, I.K. et al.(submitted) Verdes-Montenegro, L., Yun, M. S., Williams, B. A., et al.2001, A&A, 377, 812Virtanen, P., Gommers, R., Oliphant, T.E. et al. 2020,