THE ACS LCID PROJECT: ON THE ORIGIN OF DWARF GALAXY TYPES—A MANIFESTATION OF THE HALO ASSEMBLY BIAS?^*

The Local Cosmology from Isolated Dwarfs (LCID) project¹⁶ obtained the first CMDs reaching the oMSTOs and precise SFHs for six relatively isolated Local Group dwarf galaxies. The observations were designed to achieve a 2 Gyr age resolution at old ages, and this has been achieved as shown by Hidalgo et al. (2013). Two dIrrs, IC1613 and Leo A (Cole et al. 2007; Skillman et al. 2014); two dTs, LGS3 and Phoenix (Hidalgo et al. 2009, 2011); and two isolated dSphs in the Local Group, Cetus and Tucana (Monelli et al. 2010a, 2010b) were studied. The reader is referred to the above papers for details on the SFH determination. Subsequently, another dIrr galaxy, DDO210, has been studied (Cole et al. 2014).

Figure 1 (upper panel) displays the SFHs for the two dSph and the two dT galaxies of the LCID sample. Tucana and Cetus share the common characteristic of having formed over 90% of their stars before 10 Gyr ago, and they host no stars younger than 8–9 Gyr. The SFHs of the two dT galaxies are remarkably similar to those of the dSphs: they formed over 80% of their stars before 9 Gyr ago in spite of having maintained residual star formation during the rest of their evolution. The lower panel of Figure 1 displays the SFHs of the two dIrrs in our sample, Leo A and IC1613. In contrast to the former SFHs, those of the dIrrs do not show a dominant early burst of star formation; instead, over 60% of their stars formed at intermediate and young ages.

Zoom In Zoom Out Reset image size

We now consider dwarfs that are, at present, found close to the large spirals. Precise SFHs are available for a number of dSph satellites of the Milky Way (MW; Gallart et al. 1999; Aparicio et al. 2001; Carrera et al. 2002; Dolphin 2002; Lee et al. 2009; de Boer et al. 2012, 2014; del Pino et al. 2013; Weisz et al. 2014a), for the Magellanic Clouds (e.g., Noël et al. 2009; Cignoni et al. 2013; Meschin et al. 2014), and for two M31 satellites: AndXVI and AndII (Weisz et al. 2014b; M. Monelli et al. 2015, in preparation). We have represented some of these in Figure 2, together with that of DDO210. Even though they are not homogeneous among themselves or with the LCID SFHs, these results generally agree that most MW satellites (UMi, Draco, Sextans, Scl, CnVI, plus the very faint dwarfs; Brown et al. 2014) formed most of their stars before ≃10 Gyr ago. The more distant dSph satellites, Fornax, Leo I, and Carina, show substantial intermediate-age populations: their SFHs peaked at ages younger than 10 Gyr, and most of their star formation occurred at intermediate ages. In fact, the SFHs of these dSphs are similar to those of dIrr galaxies for most of their lifetimes: they have low initial star formation rates (SFRs) and high SFRs at intermediate ages. The main difference occurs in the last ≃2 Gyr or less, when they stopped star formation. They are classified as dSphs for their current properties, but their history is similar to that of dIrr galaxies.

Zoom In Zoom Out Reset image size

The availability of precise SFHs reaching the earliest star formation events for a growing sample of dwarf galaxies enables us to take a fresh look into dwarf galaxy types based on evolutionary histories rather than current properties. Although this is currently possible only for a limited number of objects in the Local Group, it provides new insight on possible mechanisms producing these dwarf galaxy types.

Based on their full evolutionary histories, we propose that dwarf galaxies can be grouped in two classes.

• 1.  
  Fast dwarfs are those that started their evolution with a dominant star formation event, but their period of star formation activity was short (≲few Gyr; see the upper panel of Figures 1 and 2).
• 2.  
  Slow dwarfs formed a small fraction of their stellar mass at an early epoch, and continued forming stars until the present (or almost; see the lower panel of Figures 1 and 2).

These two evolutionary paths do not map directly onto the current, commonly adopted dwarf galaxy classification (in dIrrs, dTs and dSphs). Most notably, some dSphs have important intermediate-age and young populations, and thus SFHs that resemble those of dIrrs: in our sample, all dIrrs are slow dwarfs, while some dSphs are fast and others are slow.

What else, besides similar SFHs, do slow dwarfs share? The distant dIrr galaxies with known full SFHs, IC1613, Leo A, and DDO210, are all located over 400 kpc away from the MW or M31, and have negative (or small for DDO210) radial velocities with respect to MW, M31, and/or the Local Group center of mass (Figure 3). Among the closest slow dwarfs, the Magellanic Clouds and Leo I have been found to have first entered the MW virial radius just a few Gyr ago (Sohn et al. 2013; Kallivayalil et al. 2013). This might indicate that all these slow dwarfs assembled in a low-density environment and are only recently entering the Local Group.

Zoom In Zoom Out Reset image size

In contrast, most fast dwarfs are close MW satellites, or isolated dSphs (Cetus and Tucana). The latter break the morphology–density relation in the Local Group. However, they have radial velocities (Lewis et al. 2007; Fraternali et al. 2009; see Figure 3) compatible with their having been close to the Local Group barycenter at early times, presumably when their assembly was taking place.

Table 1.  Basic Properties of LCID Galaxies

Galaxy	MV a	Rh a	MTg	MHIa	${(m-M)}_{0}$	RMW, VMW a	${{\rm{R}}}_{M31}$, ${{\rm{V}}}_{M31}$ a	RLG, VLG a
	mag	(')	$(\times {10}^{6}{M}_{\odot }$)	$(\times {10}^{6}{M}_{\odot }$)	mag	Kpc, Km s−1	Kpc, Km s−1	Kpc, Km s−1
Phoenix	−9.9	3.76	3.2 ± 0.3	0.12	23.09 ± 0.1b	415, −103	868, −104	556, −106
LGS3	−10.1	2.10	1.9 ± 0.1	0.38	24.07 ± 0.15c	773, −155	269, −43	422, −74
IC1613	−15.2	6.81	100a	65	24.44 ± 0.10d	758, −154	520, −64	517, −90
LeoA	−11.7	2.15	6.0a	11	24.50 ± 0.10e	803, −19	1200, −46	941, −41
Cetus	−11.2	3.20	7.0 ± 0.3	⋯	24.46 ± 0.12f	756, −27	681, 46	603, 26
Tucana	−9.5	1.10	3.2 ± 0.1	⋯	24.74 ± 0.12f	882, 99	1355, 62	1076, 73

Notes.

^aFrom McConnachie (2012). ^bHidalgo et al. (2009). ^cHidalgo et al. (2011). ^dBernard et al. (2010). ^eBernard et al. (2013). ^fBernard et al. (2009). ^gHidalgo et al. (2013).

Download table as:  ASCIITypeset image

Most (mostly theoretical) research into the transformation from a gas-rich to a gas-poor dwarf galaxy has focused on determining how dSph galaxies lost their gas. Dekel & Silk (1986), Mac Low & Ferrara (1999), Salvadori et al. (2008), and Sawala et al. (2010), among others, have explored internal feedback and the efficiency of gas ejection through supernova-driven outflows. In general, models indicate that feedback should be able to totally remove the gas only in extremely low-mass (few 10⁶ ${M}_{\odot }$ of baryonic mass) dwarf galaxies. The inability of feedback alone to totally remove the gas from currently gas-free galaxies, together with the existence of a striking morphology–density relation, has led to the current consensus that environmental processes such as ram-pressure or tidal stripping by a massive central halo must play an important role in stripping the gas from dwarf galaxies and halting star formation (see Mayer 2010 for a review). Finally, including the effects of an ionizing UV background has been shown to increase the efficiency of gas loss due to both internal feedback (Salvadori & Ferrara 2009; Sawala et al. 2010) and ram-pressure stripping (Mayer 2010).

However, it is important to note that most MW dSph satellites stopped forming stars at a very similar (within 2–3 Gyr) early time (≃10 Gyr ago). If environmental effects were crucial in removing their gas, one would expect that they entered the virial radius of the MW or other massive companion, also at very similar and early times, for their SFHs to share a common pattern. Cosmological simulations are not conclusive on this point. Several studies show that infall times around $z\sim 1$ are most typical inside MW-sized halos (e.g., Rocha et al. 2012; Wetzel et al. 2015), which would be too late to explain the early star formation truncation in many fast satellites. However, other studies of halos that assemble earlier than average, and that lead to a realistic replica of the MW (Guedes et al. 2011), find that a fraction of dSphs could be hosted in subhalos accreting at $z\sim 2$–2.5 (Tomozeiu et al. 2015).

We now examine whether or not there are theoretical indications that progenitors of slow and fast dwarfs may be different, and whether or not this difference can be linked to the location of the progenitor's halo when the majority of its mass assembly took place.

As discussed in Section 2.3, the current positions and space motions seem to indicate that slow dwarfs formed preferentially in lower-density environments than fast dwarfs. This could imply a delayed assembly history for the dwarf dark matter halos, as predicted for lower peaks in the field of density fluctuations (e.g., Lagos et al. 2009) and reflected in a number of theoretical investigations that will be discussed next.

Sawala et al. (2014) present mass-assembly histories for present-day satellites and isolated dwarfs in 12 cosmological volumes resembling the Local Group. They find that for dark matter subhalos in the mass range $M={10}^{9-10}{M}_{\odot }$, which are expected to host the majority of Local Group dwarfs, the redshift at which a halo progenitor reached 1/2 of its peak mass was ${z}_{1/2}\;\simeq$ 1.5–2 for field dwarfs and ${z}_{1/2}\;\simeq$ 4 for satellites. Brooks & Zolotov (2014) also find that satellites accrete most of their mass earlier than field dwarfs and show an earlier peak in their SFHs than the simulated field dwarfs. While in our scenario there is no one-to-one correspondence between satellites and fast dwarfs, or between field galaxies and slow dwarfs, this correspondence is expected to happen in most cases. Thus, we expect that the characteristics of fast and slow dwarfs would emerge in theoretical studies if grouped according to their early, rather than current, locations.

The scenario proposed by Benítez-Llambay et al. (2015) also fits well within the slow versus fast dwarf hypothesis. They use a high-resolution cosmological simulation of the Local Group to discuss the dramatic effects of reionization on the baryonic component of halos with virial temperatures at z_reion similar to or below ${10}^{4}\;{\rm{K}}$. This characteristic temperature defines a "threshold" mass at z_reion that strongly influences the future SFH of the system. (i) Systems collapsing early with mass at reionization just above the threshold are able to form stars before reionization, but their star formation is abruptly truncated by the combined effects of reionization and feedback from the early stellar population. They are usually characterized by a population of old stars. (ii) In halos with masses below the threshold for star formation at z_reion and gas densities high enough to survive photoevaporation due to self-shielding, star formation can instead be delayed and only (re-)started at later times (e.g., Ricotti 2009) when the host halo becomes massive enough to allow some of the gas to cool and fragment. These two types of systems can be associated with fast and slow dwarfs, respectively. Finally, in their simulation of a sample of seven dwarf galaxies in a low-density environment, Shen et al. (2014) find that although above a virial mass of ${10}^{9}{M}_{\odot }$ star formation commences quite early ($z\gt 4$), all of their dwarfs would be classified as slow.

These scenarios for the formation and evolution of slow and fast dwarfs are still compatible with final star formation shutdown when a slow dwarf enters the host halo area of influence. In fact, tidal effects could play a role in the final removal of gas at late times in some slow MW satellites, such as Leo I or Fornax, which stopped their star formation only recently, as opposed to isolated slow dwarfs, like Leo A or IC1613, that are still forming stars and retain sizable amounts of gas.