1. Gas-rich and gas-poor dwarf galaxies:
on the origin of the different dwarf galaxy types
Carme Gallart & LCID
Instituto de Astrofísica de Canarias
In this talk I will present the final conclusions of the LCID project, that has studied for the first time. These conclusions challenge the current, most accepted views of the formation and evolution of the different dwarf galaxy types. I will try to be brief in order to allow some time for discussion

2. IC 1613
LEO I
Dwarf spheroidal (dSph): No gas, no recent star formation, exponential or King profiles
Dwarf Irregular (dIrr): Gas, HII regions, recent star formation, exponential profiles.
µ0  21
M  109 M○
Z  0.1 Z○
MB > -15
M32
Phoenix
This characteristics allow us to classify dwarf galaxies in 4 main important types:
dSph: no gas, no recent star formation, exponetial or king profiles
Irregular: Gas, HII regions, recent star formation, exponential profiles
dE: no gas, no recent star formation, R^1/4 profiles
Transition (dIrr/dSph)
Transition (dT): Little or no gas, little recent star formation, exponential profiles.
Dwarf elliptical (dE): No gas, no recent star formation, R1/4 profiles

3. -similar relationships btw σ, rc, μC, MV
The different dwarf galaxy types (dSph, dIrr, dT) share some similarities, and show some differences. Some ‘textbook knowledge”, not fully accurate, is:
-similar relationships btw σ, rc, μC, MV
(Kormendy85);
-all can be fitted by exponential profiles
(Faber&Lin83)
- (most) dIrr rotate while (most) dSph do not
-morphology-density relation (with exceptions)
-differences in gas content

4. OR All dwarfs have old stars, consistent with a
common initial epoch of star formation.
They all must have looked like dIrr galaxies.
And then, at some point, some lost their gas
Morphological transformation “dIrr”-> dSph ?
(this is a quite common implicit assumption; there is even a ´transition class’ of dwarf galaxies: dT)
Were the precursors of dSph like today’s dIrr? (“different by nurture”) OR
Were dSph and dIrr born different? (“different by nature”)

5. How did dSph galaxies lose their gas?
DIFFERENT BY “NURTURE”?
Most (mostly theoretical) research on the transformation
“dIrr” -> dSph
has focused on answering the question:
How did dSph galaxies lose their gas?
Internal processes:
-Gas expulsion through star formation (Larson’74; Dekel&Silk’86, McLow & Ferrara1999, Ferrara & Tolstoy 2000, Salvadori et al. 2008, Sawala et al. 2010)
-Gas exhaustion by star formation (though difficult to remove 100% of gas)
External processes (supported by the morphology-density relation):
-Tidal stirring (Mayer et al. 01; 07; Klimentowski et al. 09): can transform infalling disky dwarfs into dSph-like objects, in ~several Gyr
-Tidal stirring + gas stripping (ram-pressure + tidal), to produce really low gas fraction (Mayer et al. 01, Passeto et al. 2003)
-What about isolated dSph? (Cetus & Tucana: D’Onghia‘09, Sales et al.‘07, Kazantzidis et al. ‘11)
-Internal/External + UV background: enhancing gas loss/stripping
(Sawala et al. 10, Mayer 10)

6. Key information: Early Star Formation Histories
OR, DIFFERENT BY “NATURE”?
In order to answer the question:
Were dSph and dIrr born different?
Key information: Early Star Formation Histories
If accretion onto a big halo, or interaction with another dwarf galaxy was responsible for stopping star formation in a gas-rich dwarf that otherwise would have evolved to a “normal” dIrr galaxy,
then one would expect to see, in the star formation histories:
- similar early SFHs for dSph and dIrr
- a variety of durations/epochs for the end of star formation in dSph

7. LOCAL COSMOLOGY FROM ISOLATED DWARFS (LCID) PROJECT:
Main goal:
To obtain, for the first time, color-magnitude diagrams (CMDs) reaching the oldest main-sequence turnoffs for a sample of isolated dwarf galaxies, i.e. beyond the Milky Way satellite system
FIRST EVER oMSTO CMDs FOR A dIRR

8. LCID CMDs sample Compl. > 90% MORE GAS CONTENT LEO-A
dSph
tran
dIrr
CETUS
LGS-3
IC1613
Compl.
> 90%
MORE GAS CONTENT
TUCAN A
PHOENIX
LEO-A
-These are the observed CMDs of the sample.
-Vertically are ordered for different morphological type
- Horizontally are ordered from less to more gas content.
- It's worth to mention now how similar are the two CMD of the dSph.
- Also as we go to more gas content galaxies, the main-sequence of the CMD is more populated.
- Note also that we are reaching 90% of the completeness at the turn-off
- The apparent magnitude reaches ~29 mag in the red filter.

9. LCID SFH results MORE GAS CONTENT dSph tran dIrr CETUS LGS-3 IC1613
Monelli et al (2010)
Hidalgo et al (2011)
Skillman et al (2013)
MORE GAS CONTENT
TUCANA
PHOENIX
LEO-A
These are the SFHs for the 6 galaxies.
- We can see that exist a relation between the current gas content and the number of intemediate and young stars formed.
- All the galaxies except LeoA present a high star formation rate
Before 10 Gyr.
- LeoA has a very star formation rate before 10 Gyr. Its main peak
Of the star formatin has been at 5 Gyr ago.
Monelli et al (2010)
Hidalgo et al (2009)
Hidalgo et al (2013)

10. Most likely Blue Straggler stars: Monelli et al. 2012
LCID SFH results
dSph
tran
dIrr
CETUS
LGS-3
IC1613
Most likely Blue Straggler stars:
Monelli et al. 2012
Monelli et al (2010)
Hidalgo et al (2010)
Skillman et al (2010)
MORE GAS CONTENT
TUCANA
PHOENIX
LEO-A
These are the SFHs for the 6 galaxies.
- We can see that exist a relation between the current gas content and the number of intemediate and young stars formed.
- All the galaxies except LeoA present a high star formation rate
Before 10 Gyr.
- LeoA has a very star formation rate before 10 Gyr. Its main peak
Of the star formatin has been at 5 Gyr ago.
Monelli et al (2010)
Hidalgo et al (2009)
Hidalgo et al (2010)

11. VERY DIFFERENT EARLY SFHs FOR dIRR & dSph/dT
TUC
CET
LGS3
PHO
I1613
LEOA
[M/H]
Look-back time
Ψ(t)
The dIrrs in the sample don’t show a dominant early burst of star formation. Also, slower metallicity enrichment.
Well, with this picture we can answer one of the main question of the project: the effect of the reionization.
If we palce the epoch of reionization around here, it seems to have not stopped the star formation
In any of these galaxies.

12. dIrr SFHs “similar” to MCs:
Noël et al. 2009
Meschin et al. 2013
LCID 2013

13. - similar early SFHs for dSph and dIrr
If accretion onto a big halo, or interaction with another dwarf galaxy was responsible for stopping star formation in a gas-rich dwarf that otherwise would have evolved to a “normal” dIrr galaxy,
then one would expect to see, in the star formation histories:
- similar early SFHs for dSph and dIrr
- a variety of durations/epochs for the end of star formation in dSph

14. VERY SIMILAR AGE OF END OF SF IN LCID dSph/dT
TUC
CET
LGS3
Look-back time
Ψ(t)
Look-back time
This cases are unclear because observational errors tends to broaden the measure features of the SFH.
For example, if we create a simple, very narrow burst and we try to recover it by simulating the observational errors using the same processes as for the real
Galaxies, the SFH resulting is a Gaussian with a sigma which depends on the input age of the narrow burst.
How reliable are the recovered
ages at early times?
Tests with ‘mock’ stellar populations:
Observational errors tend to broaden the measured features of the SFH.

15. VERY SIMILAR AGE OF END OF SF IN LCID dSph/dT
Re-ionization epoch
TUC
CET
LGS3
Look-back time
Ψ(t)
Reionization seems to have
not stopped star formation
in any of these galaxies.
Reionization was proposed
to stop star formation in
small galaxies (one way out
of the “missing satellite
problem”) .
Tucana
Look-back time
Cetus
LGS3
Tested several input mock
stellar populations of different
ages.
Mock stellar populations
recovered as if they were real.
Show a Gaussian mock
input/output here as an
example.
We have analyzed how observational errors affect to the recovered SFHs of the
Galaxies by testing several mocks SFHs.
We are showing here those mock SFHs which best match the observed one.
The dashed line shows the input, mock SFH. The solid line shows the recoverd mock SFH.
We also show overpploted the SFHs of each galaxy. The agreement is very good.
This confirm that the main bulk of the star formation in LGS3 and Cetus
Is placed after the reionization epoch.
And the SFH of Tucana is compatible with a star formation around the reionization epoch.
The question arrising now is: if reionization did not stop the star formation,
which mechanism is the responsable on the decrease on the star formation rate?
However, all three galaxies formed the bulk of their stars before z≈3, or 11 Gyr

16. VERY SIMILAR AGE OF END OF SF IN LCID dSph/dT
What about the Milky Way dSph satellites?
BaSTI isochrones (Pietrinferni et al. 2004) of 9 &13 Gyr, Z=0.0003, superimposed

17. VERY SIMILAR AGE OF END OF SF IN LCID dSph/dT
What about the Milky Way dSph satellites?
“Blue plumes” are most likely blue straggler stars sequences
(Monelli et al. 2012, Mapelli et al. 2007, 2009, Momany et al. 2008)
Figure 2: The vs. MV diagram for globular clusters (Piotto et al. 2004) open clusters (De Marchi et al. 2006) and dwarf spheroidal galaxies. The horizontal line shows the mean BSS frequency for Milky Way field stars (Preston & Sneden 2000).
Another interesting feature is the significant difference between the BSS frequency of Carina with that derived for dwarf galaxies with similar luminosity, i.e. Draco, Ursa Minor, Sextans, Sculptor and Leo II. Although it is only a lower limit, the "BSS'' frequency for Carina is of great help in suggesting a threshold near which a galaxy BSS frequency might hide some level of recent star formation. The aforementioned 5 galaxies however have a lower BSS frequency, a hint that these galaxies possess a normal BSS population rather than a young MS. This confirms previous conclusions for Sextans (Lee et al. 2003) and Ursa Minor (Carrera et al. 2002), but is in contradiction with that of Aparicio et al. (2001) for Draco. However, the Aparicio et al. conclusion was mainly based on the detection of the VC stars, a feature that, as we argued, remains an ambiguous indicator. Indeed, Fig. 2 shows that the BSS frequency in Draco is very close to that of Ursa Minor, a galaxy acceptably known to possess an old BSS population.
BaSTI isochrones (Pietrinferni et al. 2004) of 9 &13 Gyr, Z=0.0003, superimposed

18. The odd guys

19. Need a different scenario
If accretion onto a big halo, or interaction with another dwarf galaxy was responsible for stopping star formation in a gas-rich dwarf that otherwise would have evolved to a “normal” dIrr galaxy,
then one would expect to see, in the star formation histories:
- similar early SFHs for dSph and dIrr
- a variety of durations/epochs for the end of star formation in dSph
Need a different scenario

20. Note that the LCID galaxies are isolated!!
Re-ionization epoch
TUC
CET
LGS3
Ψ(t)
Look-back time
If re-ionization didn’t stop star formation, then what did it?
SNe Feedback?
Models by Mac Low & Ferrara (1999) for SNe feedback indicate that this mechanism may not remove completely the gas in these galaxies.
SNe Feedback + UV background?
Models by Sawala et al. (2010) indicate that UV background + feedback should have halted star formation at z ∼ 6, for M<8x108M. Above, self-shielding becomes effective, allowing SF to continue beyond z=6
Strigari et al (2008): M300≈107M ; MT≈109M
Walker et al. (2009): MT compatible with ≈109M
Mac Low & Ferrara (1999)
We have calculated the barionic mass of these galaxies and compare with the mechanical energy
Driven by supernova produced by them.
Using models by Mac Low and Ferrara, we can place
Tucana, Cetus and LGS3 on this figure of Mac Low and Ferrera wich show as a function of the barionic mass and mecanical energy produced by supernova, the regime of no mass loss,
blow-out/mass loss, and Blow away or complete gas removal.
The three galaxies fall in the region of partial mass loss meaning that they lost most of the gas but not all. It seems that supernova feedback could not be enough to stop the star formation.
On the other hand, models by Sawala et al. indicate that reionization together with supernova feedback should have halted the star formation at z~6
Sawala et al. (2010)

21. But seems a bit contrived to me: Occam’s razor?
Note that the LCID galaxies are isolated!!
Re-ionization epoch
TUC
CET
LGS3
Ψ(t)
But seems a bit contrived to me: Occam’s razor?
Look-back time
‘External effects’ not totally ruled out:
Radial velocities of Tucana and Cetus (Fraternali et al. 2009; Lewis et al. 2007) not incompatible with passage through the inner parts of the Local Group (possible gas stripping: Mayer et al 2001, 2006)
Resonant stripping (D’Onghia 2009 )
Cosmic ménage à trois (Sales et al. 2007): Tucana & Cetus could be the lighter member of a satellite pair ejected to a highly energetic orbit
Merger of disky dwarfs: (Kazantzidis et al. 2011)
Mac Low & Ferrara (1999)
We have calculated the barionic mass of these galaxies and compare with the mechanical energy
Driven by supernova produced by them.
Using models by Mac Low and Ferrara, we can place
Tucana, Cetus and LGS3 on this figure of Mac Low and Ferrera wich show as a function of the barionic mass and mecanical energy produced by supernova, the regime of no mass loss,
blow-out/mass loss, and Blow away or complete gas removal.
The three galaxies fall in the region of partial mass loss meaning that they lost most of the gas but not all. It seems that supernova feedback could not be enough to stop the star formation.
On the other hand, models by Sawala et al. indicate that reionization together with supernova feedback should have halted the star formation at z~6
Sawala et al. (2010)

22. Stellar populations in MW dSph possibly in place before tidal stirring and stripping could start acting! (Sawala et al. 2010)

23. dIrr and dSph different by NATURE
EVIDENCE FROM LCID PROJECT (+ literature):
1) Different early SFH of dIrr and dSph
2) Sincronicity in end of star formation time in dSph & dT galaxies
dIrr and dSph different by NATURE
PROPOSED SCENARIO
(Compatible with morphology-density relation)
-’Milder’ early star formation mode in dIrr galaxies as compared with dSph. Possibly because they are born in lower density environment, with no interactions triggering SF? They have an initially lower density? (Carraro et al. 2001) Other?
-‘Milder’ star formation activity  milder feedback  less gas loss?
(however, models have difficulty in producing mild SFR: Sawala et al. 2012)
-The high level early star formation activity in dSph (possibly triggered by environmental effects), aided with some extra energy from the UV reionization background, responsible for gas loss in dSph? (Sawala et al. 2010)
-Tidal effects (stirring, stripping) playing a role in the final removal of gas?
MAIN drivers of gas loss in dSph are internal effects

24. THE END

25. DIFFERENT BY “NURTURE”?
External processes (supported by the morphology-density relation):
----Tidal interaction with a large host galaxy----
Don’t explain the ‘isolated dSph’ such as Cetus or Tucana
Ways out:
Radial velocities of Tucana and Cetus (Fraternali et al. 2009; Lewis et al. 2007) not incompatible with passage through the inner parts of the Local Group (possible gas stripping)
Resonant stripping (D’Onghia )
Cosmic ménage à trois (Sales et al. 2007): Tucana & Cetus , lighter member of a satellite pair ejected to a highly energetic orbit
Merger of disky dwarfs: (Kazantzidis et al. 2011)

26. ? DIFFERENT BY “NURTURE”? External processes
-TAKE TIME: a best case scenario to produce an ‘old’ dSph:
“[with enhanced gas stripping due to the UV ionizing background] we can argue that if the progenitors of Draco & Umi fell into the MW at z>1.5-2 [10 Gyr], then ram-pressure combined with tides was able to remove their entire gas content in a couple of orbits [2-3 Gyr]. Mayer (2010)”
-expected variations of time of end of star formation depending on infall time and orbit (in principle supported by the “variety of star formation histories of MW dSph satellites”: but see later)
Ways out: ?

28. Conclusion SFH of LCID dIrr, (including MCs?)
FRACTION of Old pops (RR Lyrae diagram)
Compare early SFHs of one dSph & one dIrr of similar mass: Tucana & Leo A?
=>=> NOT THE SAME!!!!!
HYPOTHESIS: dIrr started SF in a ‘mild’ mode while dSph started violently.
Influence of environment?? This would still explain the morphology-density relation.
Probably this is an ingredient that is still lacking in galaxy evolution models: SFHs of dIrr by Sawala: high initial SFR.
Conclusion

32. The distribution of stars in the observed CMD is compared with that of a number of simple populations in a model CMD. I
observed CMD
model CMD
By comparison of the observed CMD with a model CMD where we know the
Position of simple populations, we can obtain the SFH using our algorithms
- I this figure we represent the resutl as the mass of stars formed as a function of age and metallicity.
- The projection over the metallicity axis give us the star formation rate as a funcion of age only
- The same in over the age axis to obtain the star formation rate as a function of metallicity only.
Look-back time (Gyr)
By using a merit function, we obtain the combination of simple stellar populations that best reproduces the observed CMD, i.e., the star formation rate and the chemical evolution law, as a function of time.
Aparicio & Gallart (2004)
Aparicio & Hidalgo (2009)

33. Stellar populations of dSph possibly in place before tidal stirring and stripping could start acting! (Sawala et al. 2010)