Collaborators

- S. Miller, K. Penner , GEMS collaboration (H. W Rix, R. Skelton, R. Somerville, E. Bell, C. Wolf, Z. Zheng) & C. Conselice

- S. Khochfar, R. Somerville, P. Hopkins, A. Benson, A. Maller- I. Marinova, T. Weinzirl, F. Barazza

Merger History and Impact on Star Formation at z<=1 &

[Constraints on bulge assembly via mergers since z<=4 ]

Shardha Jogee

University of Texas at Austin

Goals

What is relative importance of different galaxy assembly modes as f(z) : major mergers, minor mergers, cold gas accretion, secular mode

1) Provide empirical constraints on major + minor merger history out to z~1

2) Compare with predictions from LCDM-based models

3) By how much is <SFR> enhanced in normal vs visibly merging system? 4) What % of the SFR density comes from visibly merging galaxies ?

Jogee et al & GEMS team 2008, 2009 (arXiv:0810.5617)

Merger History & Impact on Star Formation at z<=1

Ingredients

- z ~ 0.2 to 0.8 (Tback~3 to 7 Gyr)

- High mass (M/M0 >= 2.5x1010): Complete for red seq and blue cloud : N~800 galaxies

- Interm. mass (M/M0 >= 1x109): Complete for blue cloud only N~3700 galaxies

- ACS F606W 0.1” images from GEMS (Rix et al 2004) - Spectro-photo z + stellar masses from COMBO-17 (Borch+06; Wolf+04)- UV and IR-based SFR from COMBO-17 & Spitzer (Bell et al 2007)

Methodology : identifying mergers

Method 1

• Visual classification of ~3700 galaxies by 3 classifiers

• Identify systems (of M>Mcut) which show evidence of having experienced a merger of mass ratio >1/10 in the last visibility timescale

Method 2

• Automated CAS criterion based on asymmetry A and clumpiness S• A> 0.35 and A >S

Visual classification of Interacting vs Non-Interacting systems

Non-interacting E-Sd Non-Interacting Irr1

Galaxies with small –scale asymmetries that can be internally triggered (e.g., via stochastic SF or low V/ without any galaxy-galaxy interactions.

Mergers

Systems with evidence (e.g.,morphological distortions) indicative of a merger of M1/M2>1/10 in last t_vis. e.g., single remnants with tidal tails, warps, strongly asymmetric arms, double nuclei, galaxies bounded by a bridge,

e.g. very close (d<< t_vis * v) pair of overlapping galaxies

Example of interacting systems

2 at similar z 2 at similar z

Separate mergers into major minor, major/minor

Mergers

Systems with evidence (e.g.,morphological distortions) indicative of a merger of M1/M2>1/10 in last t_vis.

Ambiguous: Major or MinorClear Major (M1/M2>1/4)

- Double nuclei same L - Contact pair w/ M1/M2>1/4 and z1~z2- Train wreck

Clear Minor (1/10 < M1/M2 <1/4)

- Contact pair with M1/M2 ~1/4 to 1/10 and z1~z2- Single system where disk has survived, but shows a warp or strong tidal signatures

% of clear majors % of clear minor % of minor or major

Test effect of bandpass shift and SB dimming on visual f

• In last bin z =0.6--0.8 - rest-frame of GEMS V image shifts to near-UV (3700-3290 A) - SB dimming by factor of 8

• Compare f from GEMS v vs deep, redder GOODS z

• Results changes by less than 1.07

Methodology : identifying mergers

Method 1

• Visual classification of ~3700 galaxies by 3 classifiers

• Identify systems (of M>Mcut) which show evidence of having experienced a merger of mass ratio >1/10 in the last visibility timescale

Method 2

• Automated CAS criterion based on asymmetry A and clumpiness S A> 0.35 and A >S

Results

What are the visual types of the M*>1e9 systems picked by the CAS criterion (A>0.35 and A>S) ?

1) 44% (z~0.3) to 80% (z~0.7) are visually-classified non-interacting (Irr1, E-Sd) galaxies high contamination from non-interacting systems especially at z>0.5

2) the remaining are visually-classified merger systems [50% to 70% of latter are picked]

Merger fraction from CAS vs visual classifications

Mergers/interactions missed by the CAS criterion (A>0.35,A>S)

Non-Interacting galaxies picked by

CAS criterion (A>0.35,A>S)

Merger history of massive galaxies since z~0.8 (last 7 Gyr)

Jogee et al. & GEMS collaboration 2009

For high mass (M>=2.5e10) galaxies Interaction fraction f (for mass ratio >1/10) ~ 8% to 9% fraction of clear major (M1/M2>=1/4) interactions ~1% to 3% fraction of clear minor (1:4 to 1/10) interactions ~ 4% to 8% fraction of ambiguous minor or major interactions ~ 1% to 2%

For an assumed visibility time of 0.5 Gyr, this implies that over Tb=3-7 Gyr (z=0.2-0.8) , every massive galaxy has undergone 0.7 interactions of mass ratio >1/10, of which 1/4 are major mergers, 2/3 are minor mergers, and rest are major/minor.

Compare merger rate of galaxies with LCDM models

• Data Rate= n f /Tvis for (major+minor) • Models solid line = f(major + minor) dotted lime = f_major

• Models - 3 SAMs w/ AGN feedback - HOD w/ AGN feedback - SPH cosmological

For high mass galaxies,the (major + minor) merger rate of models - show factor of 5 dispersion - bracket the observed rate & show qualitative agreement

Jogee et al. & GEMS collaboration 2009

Star formation over last 7 Gyr

SFRUV ~ 0.1--25 Mo yr-1

Median (SFRIR/SFRUV) ~ 4 for 900 galaxies with both Spitzer and UV data

significant obscured SF

Mean SFR of visible mergers is enhanced only by a modest factor (~1.6 to 2) w.r.t that of non-interacting galaxies

<SFR> in Mergers vs Non-Interacting Galaxies over last 7 Gyr

3 measures of SFR

1) SFRUV from LUV of COMBO-17 for full sample [N= 3698]

2) SFRUV + SFRIR from Spitzer 24 mu, detected in only 24% of sample [N=878]

3) SFRUV + SFRIR-stacked from stacking 24 mu frame (Zheng et al 2007) for 87% of sample

Similar results by Robaina et al. in prep

(Jogee et al. & GEMS collaboration 2009)

They find max SFR of most mergers is only enhanced by ~2 to3, compared to isolated case

(Di Matteo, P. et al. 2007)

Statistical study of several hundred TREE-SPH simulations of major mergers of different B/D, gas, orbital parameters, etc

SFR density from mergers over last 7 Gyr

For M*>=1e9 Mo & M*>=2.5e10 visible mergers account for less than 30% of the SFR density over z~0.2--0.8 (Tb=3 to 7 Gyr)

• Decline in SFR density driven by shutdown in SF of normal galaxies (Gas consumption by SF ? Decline in smooth gas accretion rate ? Transition of SF to lower masses )

(Jogee et al. & GEMS collaboration 2009)

Summary: Merger History & Impact on SF over 7 Gyr

1. Merger history for high mass (M>=2.5e10) galaxies

- Fraction of mergers (of mass ratio >1/10) ~ 8% to 9%

- For an assumed visibility time of 0.5 Gyr, this implies that over Tb=3-7 Gyr, every massive galaxy has undergone 0.7 interactions of mass ratio 1/10, of which 1/4 are major mergers, 2/3 are minor mergers, and rest are major/minor.

2. Visual vs automated CAS methods CAS merger criterion capture 50% to 70% of visually-classified mergers at z>0.5, has high contamination rate (80% at z~0.7) by non-interacting systems

3. Comparison with LCDM-based models For high mass galaxies, the (major + minor) merger rate of models show a factor of 5 dispersion and bracket the observed rate. Qualitative agreement

4. Impact on SF For both M*>=1e9 Mo and M*>=2.5e10 Mo, visible mergers - have their mean SFR enhanced by only ~1.6 to 2 wrt to non-interacting galaxies - account for less than 30% of the SFR density over z~0.2--0.8 (Tb=3 to 7 Gyr

Most (66%) high mass spirals have low bulge B/T < 0.2 (77% have n<2) Such bulges exist in barred & unbarred galaxies across different H-types (S0-Sc).

• Bulge + Disk+ Bar decomposition of H image of high mass (M*/M0 ~1010 -1011 ) spirals

Constraints on bulge assembly and merger history since z<4

Weinzirl, Jogee, Khochfar, Burkert & Kormendy 2009, ApJ, in press (arXiv:0807.0040)

Major mergers at z<=2 lead to bulges with present- day B/T > 0.2

Also true for most major mergers at z<=4

Comparing with Hierarchical Models

• Models from Khochfar & Burkert (2005) Khochfar & Silk 2006) - DM halo merger trees from the extended Press-Schechter formalism - Semi-analytic prescriptions for star formation (Cox+08) , cooling, SN feedback

For Galaxies with a past major merger

(Weinzirl, Jogee, Khochfar, Burkert & Kormendy 2009)

Major mergers since z<=4 fail seriously (by a factor of ~30) to account for the low B/T<0.2 bulges present in 2/3 of high mass spirals

It is likely that such bulges are primarily built by minor mergers and secular processes at z<=4

Present-day Data Model galaxies Model galaxies Model Galaxies B/T w/ major merger w/o major merger All

since z<=4 since z <=4------------------------------------------------------------------------------------------------------------------ B/T <=0.2 66% ~1.6% ~66% ~68%0.2 <B/T<= 0.4 26% ~5% ~13% ~18% B/T > 0.4 8% ~13% ~1 % ~14%

Model vs data

(Weinzirl, Jogee, Khochfar, Burkert & Kormendy 2009)

Comparing with Hierarchical Models

Make a quantitative comparison with the predicted

B/T distribution from CDM-based models

(Khochfar & Burkert 2005; Khochfar & Silk 2006)

DM halo merger trees from the extended Press-

Schechter formalism (Somerville & Kolatt 1999)

Baryonic physics from semi-analytic prescriptions

for SF, cooling, supernovae feedback

Major merger (M1/M2 1/4) dynamics:

Major mergers set B/T to 1; B/T declines after

major mergers due to disk buildup by cold

accretion

A galaxy with a past major merger can have

B/T 0.2 at z=0 only if z last 2

Courtesy of Khochfar & Burkert

Galaxies with a past major merger

Distribution of B/T and Bulge Index

Mean B/T, bulge index are consistent with other

work (e.g., Laurikainen et al. 2007; Graham &

Worley 2008).

68% of bulges have B/T0.2; 77% have n2 .

Such bulges exist in barred and unbarred

galaxies across a wide range in Hubble type!

Weinzirl et al. 2008

Weinzirl et al. 2008

Galaxies with a past major merger

Distribution of B/T: Data vs Model

The fraction of model galaxies with a past

major merger and B/T≤0.2 is 3%, more than 20

times smaller than the observed fraction (66%).

B/T≤0.2 bulges cannot have been built by

major mergers!

Weinzirl et al. 2008

Data Model(Major + minor)

Model(Minor only)

Model(All mergers)

B/T 0.2 65.5% 3.09% 64.1% 67.2%

B/T > 0.2 34.5% 18.6% 14.3% 32 .9%

Constraining Galaxy Evolution…..

Need empirical constraints on baryonic ‘physics’

Merger history and its impact on SF part 1 of this talk Bulge properties as f (M) part 2 … Mechanisms redistributing angular momentum: mergers, bars Feedback (SF and AGN)

(Springel et al. 2005)

CDM models: good paradigm for DM + structure on large scales

• Predictions for galaxy evolution

= f (baryonic physics)

• Possible areas of discord

Substructure or missing satellite problem

Angular momentum problem

Problem of bulgeless and low B/T galaxies

Merger fraction from visual classifications versus CAS

Jogee et al 2008, 2009

• For high M/Mo>=2.5e10

CAS-based f agrees within a factor of less than two with visual f

• For interm M/Mo>=1e9

CAS method overestimates f by a factor of 3 at z>0.5… as it picks up a large number of non-interacting galaxies (E-Sd and Irr1)

Properties and Origin of Bulges

in High Mass Spirals

Motivation

Bulges provide important clues about galaxy formation, but the absence of bulges is

likewise interesting!

Classical bulges are often absent locally:

15% of edge-on galaxies are bulgeless (Kautsch et al. 2006)

20% of i<60○ low-mass disks are quasi-bulgeless (Barazza, Jogee, &

Marinova, 2008)

11/19 galaxies with D<8 Mpc and Vc>150 km/s have pseudobulges

(Kormendy & Fisher 2008)

Must compare distribution of bulge-to-total mass (B/T) and bulge index to CDM-based

models for high and low masses Barazza, Jogee, Marinova (2008)

Kautsch et al. 2006

Weinzirl et al. 2008 (ApJ submitted; arXiv:0807.0040) has two goals:

Quantify B/T and bulge index for nearby high mass galaxies

Make a detailed quantitative comparison with CDM-based models

Sample

Drawn from OSU Bright Spiral Galaxy Survey:

Bright local field galaxies with mB≤12

Reference sample for bars in the local Universe

(Eskridge 2000; Marinova & Jogee 2007)

Use H-band light to trace stellar mass

Main sample is 146 i < 70○ galaxies, complete for

M* > 1010 M⊙ and M

B < -19.3

Luminosity Decomposition

Galaxy light is emitted from physically and dynamically distinct components:

Most previous 2D decompositions have used only bulge-disk models (e.g., Allen et al. 2006)

Inclusion of the bar in 2D bulge-disk-bar decomposition is important:

B/T and bulge index are overstated in 2D bulge-disk decomposition of barred

galaxies (Laurikainen et al. 2005)

60% of galaxies are barred in H-band (Marinova & Jogee 2007)

Optical bar fraction is higher in galaxies without prominent bulges (Odewahn

1996; Barazza, Jogee, Marinova 2008; Marinova et al. 2008; Aguerri et al. 2008)

I r ,= I Bulge r , I Disk r , I Bar r , I Spiral r ,...

We perform 2D bulge-disk and bulge-disk-bar decomposition with

GALFIT (Peng et al. 2002)

Decomposition With GALFIT

For 78% of galaxies, nuclear point sources were added to the best model

(to account for AGN, HII nuclei, nuclear star clusters)

Input guesses for

single Sersic component

Stage 1:

One Component

Stage 2:

Two Components

(bulge+disk or bar+disk)

Stage 3:

Three

Components

(bulge+disk+bar)

Stage 1 outputs are

input guesses for bulge

(disk b/a, PA fixed to pre-determined values)

Stage 2 outputs are

input guesses for bulge and disk

Include guesses for bar parameters.

(disk b/a, PA fixed to pre-determined values)

Fit single Sersic profile

Fit Sersic profile + exponential disk

Fit Sersic bulge +

exponential disk + Sersic

bar

Choose the best fit from Stage 2 and Stage 3 based on:

2

Residuals

Model parameters

Data image

Sample Decomposition For NGC 4643

Light redistributed

from bulge & disk

to bar✓

Weinzirl et al. 2008

Fitr

e or h('')

n b/a PALuminosity

(%)

Stage 1 Sersic 27.9 4.44 0.80 -51.0 100

Stage 2BulgeDisk

23.9338.9

4.161.00

0.900.84

-51.166.9

34.665.4

Stage 3BulgeDiskBar

5.4348.221.3

2.531.000.62

0.900.840.37

60.566.9-45.8

25.054.120.9

Stage 1:

One Component

Stage 2:

Two Components

(bulge+disk)

Stage 3:

Three

Components

(bulge+disk+bar)

B/T usually changes between

Stage 2 and Stage 3 by a factor of

<4

Residual bar light

Sample and MethodResults

Distribution of B/T and Bulge Index

Mean B/T, bulge index are consistent with other

work (e.g., Laurikainen et al. 2007; Graham &

Worley 2008).

68% of bulges have B/T0.2; 77% have n2 .

Such bulges exist in barred and unbarred

galaxies across a wide range in Hubble type!

Weinzirl et al. 2008

Weinzirl et al. 2008

Bar Fraction vs B/T and Bulge Index

H-band bar fraction is 58% (84/146), in agreement with other studies on the same data

(Marinova & Jogee 2007; Laurikainen et al. 2004; Eskridge et. al 2000)

Is H-band bar fraction sensitive to B/T and bulge index?

Is there a relationship between bulges and bars?

Secular evolution may build low-B/T, disky bulges

Or, low-B/T galaxies with no ILR are more susceptible bars induced by swing amplification

with a feedback loop (Julian & Toomre 1966; Toomre 1981; Binney & Tremaine 1987)

Bar fraction with bulge n2 64.3% 4.53%

Bar fraction with bulge n>2 35.3% 8.20%

Bar fraction with bulge B/T0 .2 67.6% 5.44%

Bar fraction with bulge B/T>0.2 35.9% 7.68%

H-band bar

fraction is greater

by a factor of two

for low B/T and

low bulge index

galaxies!

Comparing with Hierarchical Models

Make a quantitative comparison with the predicted

B/T distribution from CDM-based models

(Khochfar & Burkert 2005; Khochfar & Silk 2006)

DM halo merger trees from the extended Press-

Schechter formalism (Somerville & Kolatt 1999)

Baryonic physics from semi-analytic prescriptions

for SF, cooling, supernovae feedback

Major merger (M1/M2 1/4) dynamics:

Major mergers set B/T to 1; B/T declines after

major mergers due to disk buildup by cold

accretion

A galaxy with a past major merger can have

B/T 0.2 at z=0 only if z last 2

Courtesy of Khochfar & Burkert

Galaxies with a past major merger

Bulge formation mechanisms include major mergers (M1/M2 1/4), minor mergers (1/10<M 1/M2<1/4), and secular

evolution

Minor Mergers and Secular Evolution

Minor mergers add all stellar mass in satellite to bulge of

primary

Secular processes are neglected

Contribution of minor mergers:

Satellite deposits stars in central region of the primary

Gas inflow from tidally induced bars and tidal torques

Contribution of secular evolution:

Bar-driven inflow between mergers

Boxy/peanut bulges from bar bending/buckling

Included in model

Neglected in model

Distribution of B/T: Data vs Model

The fraction of model galaxies with a past

major merger and B/T≤0.2 is 3%, more than 20

times smaller than the observed fraction (66%).

B/T≤0.2 bulges cannot have been built by

major mergers!

Weinzirl et al. 2008

Data Model(Major + minor)

Model(Minor only)

Model(All mergers)

B/T 0.2 65.5% 3.09% 64.1% 67.2%

B/T > 0.2 34.5% 18.6% 14.3% 32 .9%

Sample: 146 i < 70○ galaxies; complete for M* > 1010 M⊙ and MB < -19.3

Model: Hierarchical CDM-based models from Khochfar, Burkert, & Silk

Results:

Low B/T0.2 bulges are found in 68% of spirals; n2 bulges are found in 77%

Fraction of model galaxies with past major mergers at z<=4 and B/T<0.2 is more than 20

times smaller than the observed fraction

Future observational work:

Measure ages of bulges relative to bars and disks with IFU spectroscopy

Ongoing decomposition of the dense Coma cluster (ACS Coma Cluster Treasury Survey; Carter et al. 2008)

Study properties of massive disks at 1.5<z<3 from the GOODS NICMOS survey (Conselice et al. 2008)

Summary & Future Work

M/L Ratio

Backup slide of Schneider plot for M/L ratio and stellar populations

Bell equation for stellar mass+photometric vs dynamical mass + definitions

Swing Amplifier

Diagram of swing amplifier loop. Destruction of ILR

Stellar Masses

Bell equation for stellar mass+photometric vs dynamical mass + definitions

Extra slides :

Specific SFR vs Mass : the fractional growth of galaxies

Only modest enhancement in SSFR of Dist/Int vs normal galaxies

Larger SSFR (fractional growth) in low mass than high mass galaxies at z<1 (“downsizing”)

See Noeske et al al (2007) : SFR as f(M) and proposal of staged SF model

4) Assumed baryonic physics

- Model of ISM

- Recipes for star formation and feedback,

- Mechanisms to redistribute angular momentum (mergers, bars, dynamical friction)

1) Limited dynamic range + spatial resolution [N=1010, D=500Mpc/h, Resolution~5kpc/h]

31Mpc/h

DM

Light

2) Halo occupation statistics

3) DM halo merger galaxy merger history

Challenges for predicting how galaxies evolve

Model predictions not unique/robust

CDM models = good paradigm for how structure and DM evolves on large scales

Millenium Run : 1010 particles Follows DM in region D=15 Mpc/h Resolution = 5 kpc/h

(Springel et al. 2005)

Challenges for CDM models of galaxy evolution

4) Assumed baryonic physics

Model of ISM, recipes for star formation and

feedback, mechanisms to redistribute angular

momentum (mergers, bars, dynamical friction)

1) Limited dynamic range + spatial resolution Cannot simultaneously model large-scale

environment and resolve galaxy

components (bulge, bar, disk)

[N=1010, D=500Mpc/h, Resolution~5kpc/h]

31Mpc/h

DM

Light2) Halo occupation statistics

3) DM halo merger galaxy merger history

Challenges for predicting how galaxies evolve

Model predictions not unique/robust

1) Status of challenges to LCDM models of galaxy evolution? - Angular momentum problem

- Challenge of galaxies with no bulges or bulges of (low B/T, n)

- Substructure or missing satellite problem

- Cusp–core controversy

Latest empirical constraints on the history of (mergers, SF, and structural assembly)

Are problems alleviated by improvement in resolution + baryonic physics (feedback)

2. New challenges ?

- massive disks at z~1.5 to 3 with high SFR/bulges but no signs of major mergers

- mass function of very massive galaxies

3. Relative importance of different galaxy assembly modes as f(z) major mergers, minor mergers, cold gas accretion, secular modes

4. SF and AGN activity: triggers and feedback

Broad Questions For This Workshop

(A) Challenge of galaxies with no bulge or low (B/T, n) bulges

1) Major mergers build classical bulges

• Violent relaxation of stars spheroid of low v/ n=4 (or 2<n <6)• B/T at z~0 depends on epoch of last major merger & subsequent disk buildup

(A) Challenge of galaxies with no bulge or low (B/T, n) bulges

3) Minor mergers build …. bulges• Gas inflow driven by induced bar and tidal torques SF builds disky component?• Satellite accretion in central region builds/enhances bulge. Structure?

2) Secular processes build disky pseudobulges and boxy bulges • Gas inflow driven by a bar in non-interacting galaxy SF builds disky, high v/low n<2.5 stellar component = disky/pseudobulge (Kormendy 93)• Buckling instability + vertical ILRs make edge-on bars look peanut/boxy (Combes, Shlosman)

Every galaxy that had a major merger at an epoch when its mass was a significant fraction of its present-day mass should harbor a classical bulge with a significant bulge-to-total (B/T) ratio.

1) In low mass/late type galaxies: bulgeless galaxies are frequent

- late type galaxies are often bulgeless (Boker et al. 2002)

- 15% of edge-on SDSS galaxies are thin bulgeless disks (Kautsch et al. 2006)

- 20% of i<60 SDSS galaxies at z<0.03 appear bulgeless (Barazza, Jogee, Marinova 08)

(A) Challenge of galaxies with no bulge or low (B/T n) bulges

(Kautsch et al. 2006) (Barazza, Jogee, Marinova 08)

2) Even high mass spirals show a high frequency of low (B/T, n) bulges

- Most S0 -S0/Sa have bulges with Sersic n < 2 (Balcells et al 03; Laurikainen et al 07)

- 11/19 galaxies with D<8 Mpc& Vc>150 km/s have pseudobulges (Kormendy & Fisher 08)

(A) Challenge of galaxies with no bulge or low (B/T n) bulges

- Most of 400 spirals along Hubble sequence have B/T<0.25 (Graham & Worley 2008)

- For a sample of 140 M*>1e10 spirals:

66% have B/T < 0.2 & 77% have n< 2.

SAM models predict that galaxies with a

past major merger can only account for

3% of spirals with such low B/T.

Are remaining bulges built via

minor mergers and secular modes?

(Weinzirl et al. 08; See talks by Khochfar,

Weinzirl, Balcells)

(Weinzirl et al. 08)

QUESTIONS/OPEN ISSUES

Theory

1) Can cosmological simulations produce enough bulgeless/low B/T galaxies ?

2) Do main processes for removing low J gas differ in high vs low mass systems?

3) Models have focused primarily on major mergers. How do we better incorporate

bulge building via secular evolution, minor mergers, and cold gas accretion?

[Talks: Burkert, Navarro, Governato, Dekel, Khochfar, Combes,,Shlosman, Cox, Stewart, Hopkins]

Observations

4) Fold in Ages + Kinematics+ Metallicity w/ structure of (B/T, n) of bulges

5) How do bulge, bar, disks vary in field vs cluster enviroments ?

[e.g.,Talks by Barroso, Juric, Brown, Balcells, Fisher, Weinzirl, Marinova, Graves]

6) Direct empirical constraints on minor and merger history out to z~2

[see talks by Balcells, Sanjuan, Robaina, Sketlon, Stewart, Conselice, Jogee]