9.1. Ultra-Faint Dwarfs Around M31
The natural first step in studying UFDs beyond the Milky Way is exploring the vicinity of M31. The Pan-Andromeda Archaeological Survey (PAndAS) has now imaged the M31 halo out to a projected radius of ∼ 150 kpc using the Canada-France-Hawaii Telescope (McConnachie et al., 2009), discovering 17 new dwarf galaxies (Martin et al., 2006, 2009, 2016c, Ibata et al., 2007, Irwin et al., 2008, McConnachie et al., 2008, Richardson et al., 2011). An additional 8 dwarfs in the vicinity of Andromeda have also been discovered since 2004, mainly in SDSS and Pan-STARRS (Zucker et al., 2004, Zucker et al., 2007, Majewski et al., 2007, Slater, Bell & Martin, 2011, Bell, Slater & Martin, 2011, Martin et al., 2013b, a). While a handful of these dwarf galaxies may not be true satellites of M31, all of them except Andromeda XXVII (Conn et al., 2012) are likely located within the M31 virial radius. The currently known M31 satellite population reaches as faint as MV ≈ −6 (Martin et al., 2016c), including 8 UFDs according to our definition. The sizes and luminosities of the ultra-faint M31 satellites are in excellent agreement with the locus established by Milky Way dwarfs in Fig. 2.
9.2. Surveys Outside the Local Group
Detecting UFDs at even larger distances is difficult because of their low surface brightnesses and small sizes. In the nearest galaxy groups at distances of 3−4 Mpc, the most luminous red giants have apparent magnitudes of r ∼ 24.5−25. Since faint dwarfs contain few stars near the tip of the red giant branch, imaging to fainter than 26th magnitude is necessary to identify an ultra-faint dwarf at these distances as an overdensity of resolved stars. In dedicated deep surveys and HST imaging of nearby galaxy clusters, several objects near or below our magnitude limit separating UFDs from dSphs have recently been discovered, including d0944+69 (MV = −6.4; Chiboucas, Karachentsev & Tully, 2009, Chiboucas et al., 2013) in the M81 group, Virgo UFD1 (MV = −6.5; Jang & Lee, 2014) in the Virgo cluster, CenA-MM-Dw7 (MV = −7.2; Crnojević et al., 2016a) in the Centaurus A group, MADCASH J074238+652501-dw (MV = −7.7 Carlin et al., 2016) around NGC 2403, and Fornax UFD1 (MV = −7.6; Lee et al., 2017) in the Fornax cluster. Low-surface brightness dwarfs in the Local Volume with luminosities in the UFD regime can also be identified via their diffuse light (e.g., Bennet et al., 2017, Danieli, van Dokkum & Conroy, 2018). The sample of UFDs in other environments is still too small and heterogeneous for comparative studies, but the luminosities and radii of these dwarfs seem to be consistent with the properties of the Milky Way satellites shown in Fig. 2.
The first significant sample of UFDs beyond the Local Group will likely be revealed by LSST. The stacked end-of-survey LSST images will reach fainter than 27th magnitude in g and r bands, up to ∼ 1 mag beyond the depth of the current state-of-the-art PISCeS (e.g., Sand et al., 2014, Crnojević et al., 2016a) and MADCASH (Carlin et al., 2016) surveys. Extrapolating from current results, LSST should be sensitive to galaxies as faint as MV ≈ −6 in galaxy groups at 3−4 Mpc, and even lower luminosity systems in the local field at 1−2 Mpc (e.g., Tollerud et al., 2008, LSST Science Collaboration et al., 2009) via resolved stars. Systematic searches for UFDs throughout this volume will enable the galaxy luminosity function to be probed down to extremely faint absolute magnitudes across a wide range of environments.
In more massive dwarf galaxies (M* > 107 M⊙), population studies demonstrate that star formation is shut off only by environmental effects (Geha et al., 2012). The lack of gas or ongoing star formation among satellites of the Milky Way and M31 suggests that starvation and ram-pressure stripping are the primary mechanisms for environmental quenching down to masses as small as M* ≈ 105.5 M⊙ (Wetzel, Tollerud & Weisz, 2015, Fillingham et al., 2015, 2016, 2018). In the ultra-faint regime, however, the available star formation histories show that star formation ended ∼ 12 Gyr ago (Brown et al., 2014) even though at least some of the galaxies were likely accreted by the Milky Way more recently (Rocha, Peter & Bullock, 2012, Simon, 2018, Fritz et al., 2018a). At lower stellar masses, the timing of quenching, N-body-based models, and hydrodynamic simulations all suggest that reionization is responsible for shutting off star formation (Brown et al., 2014, Jeon, Besla & Bromm, 2017, Fitts et al., 2017, Rodriguez Wimberly et al., 2019). If this hypothesis is correct, then UFDs can form anywhere and need not be in close proximity to massive galaxies. LSST would therefore be expected to find large numbers of such systems beyond the boundary of the Local Group (Rodriguez Wimberly et al., 2019).
9.3. Connection to Observations of the High-Redshift Universe
In a recent series of important papers, Boylan-Kolchin, Weisz, and collaborators have quantified the correspondence between dwarf galaxies observed today in the Local Group and faint galaxies at high redshift. Boylan-Kolchin et al. (2015) used the observed star formation histories of nearby dwarfs to calculate their ultraviolet (UV) luminosities as a function of time. 10 They showed that reionizing the universe require a significant contribution of UV photons from galaxies at least as faint as the Fornax dSph. Even with the James Webb Space Telescope such galaxies will not be detectable at z ∼ 7 (Boylan-Kolchin et al., 2015). Moreover, Boylan-Kolchin et al. (2016) demonstrated via comparison to N-body simulations that the Local Group is comparable in size to the Hubble Ultra Deep Field, and is a cosmologically representative volume at dwarf galaxy masses. Weisz & Boylan-Kolchin (2017) then examined the UV LF in the reionization era. Given that the observed properties of UFDs today demonstrate that galaxies as faint as MUV ∼ −3 existed at high redshift, they showed that if the currently measured faint-end slope of the UV LF (α ∼ −2; e.g., Livermore, Finkelstein & Lotz (2017)) is extrapolated to MUV = −3 then UFDs dominate the ionizing photon production of the universe. However, this assumption substantially overpredicts the observed dwarf galaxy population of the Local Group. If the faint-end slope is shallower (α = −1.25), as estimated by Koposov et al. (2008) from SDSS data, then only bright dwarfs contribute to reionization.
This analysis highlights the complementarity between direct observations of the epoch of reionization and studies of the ancient stars in the closest galaxies. Local Group observations can probe the population of typical galaxies orders of magnitude fainter than will be possible at high redshift in the foreseeable future. As described in the preceding sections, these galaxies can also be dissected star by star, with detailed kinematic, chemical, mass, age, and spatial information. On the other hand, the distant universe provides much better statistics, access to a variety of environments, and the opportunity to compare galaxy populations across cosmic time, none of which can be done nearby. At the intersection between the two we can learn about the sources the reionized the universe, the halo masses associated with faint galaxies, and stellar populations and nucleosynthesis in the first galaxies.
10 At z = 7, MUV = 0.71 MV (z = 0) − 2.71, such that classical dSphs had UV magnitudes in the reionization era similar to their V-band magnitudes today, while UFDs had high-z UV magnitudes ∼ 1−2 mag brighter than their present-day optical magnitudes. Back.