How to kill a galaxyHow to kill a galaxy
(A review of galaxy properties as a (A review of galaxy properties as a function of environment)function of environment)
Michael BaloghUniversity of Waterloo, Canada(Look for new job postings on AAS)
CollaboratorsRichard Bower , Simon Morris, Dave WilmanNo picture: Vince Eke, Cedric Lacey, Fumiaki Nakata
Durham
Ivan Baldry &Karl
GlazebrookJohns Hopkins
Baugh, Cole, Frenk (Durham)
Bob Nichol, Chris Miller & Alex Gray
Carnegie Mellon
John Mulchaey& Gus OemlerOCIW
Ray CarlbergToronto
Ian Lewis (Oxford)
and the 2dFGRS team
No picture: Taddy Kodama
Outline
1. Background and motivation
2. Low redshift: SDSS and 2dFGRS
3. Groups and clusters at z~0.5
4. GALFORM predictions
5. Conclusions
Outline
1. Background and motivation
2. Low redshift: SDSS and 2dFGRS
3. Groups and clusters at z~0.5
4. GALFORM predictions
5. Conclusions
B) External? Hierarchical build-up of structure inhibits star formation
A) Internal? i.e. gas consumption and “normal” aging
Why Does Star Formation Stop?
(Hopkins et al 2004)
Galaxy clusters: Galaxy clusters: the end of star the end of star
formation?formation?• “Dead” galaxies (i.e. little gas or star formation) found in rich clusters• Hierarchical formation models predict number of clusters increases with time.• So perhaps dense environments are responsible for terminating star formation?
Nature or Nurture?
• Nature? Elliptical galaxies only form in protoclusters at high redshift. Rest of population is due to infall.
• or Nurture? Galaxy evolution proceeds along a different path within dense environments.– If this is true in groups and clusters, then
environment could be the driving force of recent galaxy evolution…
Early type galaxiesBower, Lucey & Ellis 1992
Tight colour-magnitude relation (Faber 1973; Visvanathan & Sandage 1977; Terlevich et al. 2001)
• van Dokkum & Franx 1996:• M/L evolution consistent with high formation redshift
E
Morphology-Density Relation
Dressler 1980
Clusters
Fie
ld
S0Spirals
Morphology-density: evolution
Dressler et al. 1997; Couch et al. 1994; 1998Fasano et al. 2000Wide field HST: Treu et al. 2003
Log surface density
Nu
mb
er
of
gala
xies
RedshiftN
S0/N
E
Low redshift
Z~0.5
HI deficiency
Bravo-Alfaro et al. 2000
Davies & Lewis 1973
VLA imaging of Coma spirals
Mark I and II imaging of Virgo galaxies
18 nearby clusters: Solanes et al. 2001
Emission line fraction in SDSS and 2dFGRS (Balogh et al. 2004)
A901/902 supercluster (Gray et al. 2004) correlation with dark matter density
• Fraction of emission-line galaxies depends strongly on environment, on all scales
• Trend holds in groups, field, cluster outskirts (Lewis et al. 2002; Gomez et al. 2003)
• Fraction never reaches 100%, even at lowest densities
Star formation
Cluster infall regions
Emission lines
Dressler, Thompson & Shectman 1985; Also Gisler 1978
• Cluster galaxies of given morphological type show less nebular emission than field galaxies
• suggests star formation is suppressed in cluster galaxies
Em
issi
on
lin
e f
ract
ion
H distribution
Koopmann & Kenney 2004also: Vogt et al. 2004
• Cluster galaxies often show peculiar distribution of H emission: usually truncated, or globally suppressed
• In some cases, star formation is centrally enhanced (Moss & Whittle 1993; 2000)
Virgo spirals
H for Virgo galaxy
H for normal galaxy
Additional physics?• Ram-pressure stripping (Gunn & Gott 1972)
• Collisions / harassment (Moore et al. 1995)
• “Strangulation” (Larson et al. 1980; Balogh et al. 2000)
Additional physics?• Ram-pressure stripping (Gunn & Gott
1972)
• Collisions / harassment (Moore et al. 1995)
• “Strangulation” (Larson et al. 1980; Balogh et al. 2000)
Quilis, Moore & Bower 2000
short timescale
Kenney et al. 2003
Additional physics?• Ram-pressure stripping (Gunn & Gott 1972)
• Collisions / harassment (Moore et al. 1995)
• “Strangulation” (Larson et al. 1980; Balogh et al. 2000)important ingroups?
Also tidal effects from LSS? (Gnedin 2003)
Additional physics?• Ram-pressure stripping (Gunn & Gott 1972)
• Collisions / harassment (Moore et al. 1995)
• “Strangulation” (Larson et al. 1980; Balogh et al. 2000)
– Either through tidal disruption, or shock-heating to level at which it can’t cool (e.g. Springel & Hernquist 2001)
long timescale
Additional physics?• Ram-pressure stripping (Gunn & Gott 1972)
• Collisions / harassment (Moore et al. 1995)
• “Strangulation” (Larson et al. 1980; Balogh et al. 2000)
– Either through tidal disruption, or shock-heating to level at which it can’t cool (e.g. Springel & Hernquist 2001)
long timescale
• Ram pressure stripping of the disk could transform a spiral into a S0 (Gunn & Gott 1972; Solanes & Salvador-Solé 2001)
• Strangulation may lead to anemic or passive spiral galaxies (Shiyoa et al. 2002)
S to S0 transformation?Kenney et al. 2003Vollmer et al. 2004
Non-SF spiral galaxies from SDSS (Goto et al. 2003)First noted by Poggianti et al. (1999) in z~0.5 clusters
S to S0 transformation?• But bulges of S0 galaxies
larger than those of spirals (Dressler 1980; Christlein & Zabludoff 2004)
• Requires S0 formation preferentially from spirals with large bulges (Larson, Tinsley & Caldwell 1980) perhaps due to extended merger history in dense regions (Balogh et al. 2002)
Dressler 1980
Bulge size
1. S0 galaxies found far from the cluster core
– Galaxies well beyond Rvirial may have already been through cluster core (e.g. Balogh et al. 2000; Mamon et al. 2004; Gill et al. 2004)
2. Morphology-density relation holds equally well for irregular clusters, centrally-concentrated clusters, and groups
- but may be able to induce bursts strong enough to consume the gas
Gill et al. 2004
Groups (Postman & Geller 1984)
Local galaxy density (3d)
Sp
iral
fract
ion
Arguments against ram pressure stripping:
Observations: z~0.3
• Strangulation model:– infall rate +
assumed decay rate of star formation => radial gradient in SFR
• Radial gradients in CNOC clusters suggest ~2 Gyr
Balogh, Navarro & Morris (2000)
Outline
1. Background and motivation
2. Low redshift: SDSS and 2dFGRS
3. Groups and clusters at z~0.5
4. GALFORM predictions
5. Conclusions
Colour-magnitude relation
CMR for spiral galaxies also observed (e.g. Chester & Roberts 1964; Visvanathan 1981; Tully, Mould & Aaronson 1982)
SDSS allows full distribution to be quantified with high precision ( Baldry et al. 2003; Hogg et al. 2003;Blanton et al. 2003)
Sloan DSS data
Baldry et al. 2003(u-r)
Analysis of colours in SDSS data:
• Colour distribution in 0.5 mag bins can be fit with two Gaussians
• Mean and dispersion of each distribution depends strongly on luminosity
• Dispersion includes variation in dust, metallicity, SF history, and photometric errors
• Bimodality exists out to z~1 (Bell et al. 2004)
Bright
Faint
• 24346 galaxies from SDSS DR1. magnitude limited with z<0.08
• density estimates based on Mr<-20
Balogh et al. 2004
• Fraction of red galaxies depends strongly on density. This is the primary influence of environment on the colour distribution.
• Mean colours depend weakly on environment: transitions between two populations must be rapid (or rare at the present day)
• How rapid must the bluered transition be?
• colour evolves rapidly if timescale for star formation to stop is short
• if transformations occur uniformly in time:
• need <0.5 Gyr
• if transformations are more common in the past, longer timescales permitted
Blue Peak
Red Peak
H distribution• H distribution shows a
bimodality: mean/median of whole distribution can be misleading
Balogh et al. 2004
The star-forming population
• Amongst the star-forming population, there is no trend in H distribution with density
• Hard to explain with simple, slow-decay models (e.g. Balogh et al. 2000)
Isolated Galaxies
• Selection of isolated galaxies:– non-group
members, with low densities on 1 and 5.5 Mpc scales
• ~30% of isolated galaxies show negligible SF– environment must
not be only driver of evolution.
All galaxiesBright galaxies
Summary: SDSS & 2dFGRS
• SFH depends on environment and galaxy luminosity (mass) in a separable way.
• Colour and H distributions suggest any transformations must have a short timescale, or have occurred preferentially in the past– but how do you reconcile this with large
fraction of Virgo spirals with unusual H distributions? hmmm…
Outline
1. Background
2. Low redshift: SDSS and 2dFGRS
3. Groups and clusters at z~0.5
4. GALFORM predictions
5. Conclusions
Evolution in clusters and groups
• Results from low redshift surveys suggests we focus on two separate effects:
1. Evolution in the fraction of active galaxies2. Evolution in the SFR distribution of those
active galaxies
Orientation: if environment drives evolution, expect to see weaker evolution in clusters and groups than in isolated galaxies…
Butcher-Oemler Effect
Andreon, Lobo & Iovino 2004
• Concentrated clusters at high redshift may have more blue galaxies than concentrated clusters at low redshift• But blue fraction depends strongly on luminosity and radius so care needs to be taken to evaluate blue fraction at same luminosity limit, and within same (appropriate) radius.
Margoniner et al. 2001
Redshift
Blu
e f
ract
ion
Blu
e f
ract
ion
MMVV < -20 < -20
High densityHigh density
Low densityLow density
All galaxiesAll galaxies
RedshiftRedshift
Red
gal
axy
frac
tion
Red
gal
axy
frac
tion
Evolution of the red sequence
(Bell et al 2004)
• “Butcher-Oemler effect” also seen in the general field
Clusters
Field
2dF
Nakata et al., MNRAS, submitted
Postman, Lubin & Oke 2001van Dokkum et al. 2000
Fisher et al. 1998
Czoske et al. 2001
Cluster SFR evolution
• Based on sparsely-sampled [OII] spectroscopy
• Suggests fraction of star-forming galaxies evolves only weakly in clusters
• Different from colour evolution?
Cluster SFR evolution
Kodama et al. 2004
Couch et al. 2001Balogh et al. 2002Fujita et al. 2003
Tresse et al. 2002
Complete H studies:Even at z=0.5, total SFR in clusters lower than in surrounding field
FieldField
z~0.3 z~0.5
Cluster SFR evolution• Complete H based SFR
estimates
• Evolution in total SFR per cluster not well constrained
• considerable scatter of unknown origin
• systematic uncertainties in mass estimates make scaling uncertain
Kodama et al. 2004
Finn et al. 2003Finn et al. 2003
Cluster SFR evolution• Complete H based SFR
estimates
• Evolution in total SFR per cluster not well constrained
• considerable scatter of unknown origin
• systematic uncertainties in mass estimates make scaling uncertain
Kodama et al. 2004Finn et al. in prep
Finn et al. 2003
Evolution in groups
z~0.05: 2dFGRS (Eke et al. 2004)– Based on friends-of-friends linking
algorithm– calibrated with simulations. Reproduces
mean characteristics (e.g. velocity dispersion) of parent dark matter haloes
z~0.45: CNOC2 (Carlberg et al. 2001)– selected from redshift survey,
0.3<z<0.55– Cycle 12 HST imaging + deeper
spectroscopy with LDSS2-Magellan
Group comparison
Wilman et al. in 2004
Fra
ctio
n o
f n
on
-SF
g
ala
xies
• Use [OII] equivalent width to find fraction of galaxies without significant star formation
• most galaxies in groups at z~0.4 have significant star formation – in contrast with local groups
Wilman et al. 2004
• Fraction of non-SF galaxies increases with redshift
• for both groups and field
Fra
ctio
n o
f n
on
-SF
g
ala
xies
Groups
Group SFR evolutionF
ract
ion
of
non
-SF
g
ala
xies
Field
Group SFR evolution
Wilman et al. 2004
• shape of [OII] distribution evolves with redshift but does not depend on environment
• Result sensitive to aperture effects
Outline
1. Background
2. Low redshift: SDSS and 2dFGRS
3. Groups and clusters at z~0.5
4. GALFORM predictions
5. Conclusions
WIP: GALFORM model• GALFORM is Durham model of galaxy
formation (Cole et al. 2000)– parameters fixed to reproduce global properties of
galaxies at z=0 (e.g. luminosity function) and abundance of SCUBA galaxies at high redshift
• Use mock catalogues of 2dFGRS which include all selection biasses
• Predict H from Lyman continuum photons, choose dust model to match observed H distribution. This is the weak point at the moment.
• Assume hot gas is stripped from galaxies when they merge with larger halo (i.e. groups and clusters) which leads to strangulation of SFR (gradual decline)
GALFORM predictions• Fraction of SF galaxies
declines with increasing density as in data
• Similar results found by Diaferio et al. (2001; z=0.3 CNOC clusters) and Okamoto et al. (2003; morphology-density relation)
• Normalisation depends on SFR-H transformation, but trend is robust
GALFORM predictions• Over most of the density
range, correlation between stellar mass and SFR fraction is invariant
Therefore SFR-density correlation is due to mass-density correlation
• At highest densities, models predict fewer SF galaxies at fixed mass due to strangulation
• Trend with mass driven by selection effects which make analysis difficult
GALFORM predictions
1. Fraction of SF galaxies declines with increasing density as in data
2. At low densities, H distribution independent of environment
GALFORM predictions
1. Fraction of SF galaxies declines with increasing density as in data
2. At low densities, H distribution independent of environment
GALFORM predictions
1. Fraction of SF galaxies declines with increasing density as in data
2. At low densities, H distribution independent of environment
3. In densest environments, H distribution skewed toward low values * This is sensitive to SFR-Htransformation
however
Conclusions
• On average, galaxies in groups have less star formation than field galaxies
• Presence of non-star forming galaxies in the lowest densities means environment cannot be the only driver of galaxy evolution
• Galaxy interactions and mergers:– Build larger bulges in dense environments– Consume available gas in rapid starburst– Present in all environments, but more so at higher
densities– Establish red sequence in clusters at early times
• Strangulation, ram-pressure add additional suppression in dense regions at late times