Delayed Stellar Mass Assembly in the Low Surface Brightness Dwarf Galaxy KDG 215^∗

The interplay between the gaseous and stellar components of low-mass galaxies is extremely complex. Gas-rich dwarf irregular galaxies (i.e., systems with significant mass reservoirs of neutral hydrogen, H i) often host ongoing massive star formation (SF; as traced by Hα emission, with characteristic timescales of <10 Myr) and have significant ultraviolet (UV) emission (tracing SF on longer timescales of 100–200 Myr). The stochastic sampling of the upper portion of the initial mass function is known to be a complicating factor in low-mass galaxies (Lee et al. 2009). While significant progress has been made in understanding how the properties of the resolved stellar populations and the integrated UV luminosities are related (McQuinn et al. 2015), it remains very difficult to predict how a certain gas-rich dwarf irregular galaxy converts its available reservoir of gas into stars.

An especially important type of galaxy in this regard is the low surface brightness dwarf. Traditionally defined as having low central surface brightnesses (μ_B > 23.0 mag arcsecond⁻²), low surface brightness galaxies span all galaxy masses and morphological types (e.g., McGaugh & Bothun 1994; de Blok et al. 1995). Many (but not all) low surface brightness galaxies host ongoing massive SF as traced by Hα emission (e.g., Schombert et al. 2011). The dwarf members of this class have been extensively studied in the context of the threshold gas mass surface density required for SF (e.g., van der Hulst et al. 1993; van Zee et al. 1997), as well as in the derivation of high-resolution rotation curves and the resulting constraints on the dark matter distribution within galaxies (e.g., de Blok et al. 2001; de Blok & Bosma 2002).

Nearby low surface brightness dwarf galaxies offer a unique opportunity to study their recent SF on a spatially resolved basis. This allows for the reconstruction of the recent evolutionary pathways that have led to their current low surface brightness state. Gas-rich low surface brightness systems that are quiescent in terms of current massive SF (i.e., are Hα non-detections at meaningful sensitivity levels) are especially interesting, as they offer a glimpse of the conditions in which SF has ceased altogether. Is this "quenching" of SF due to the current conditions of the gas? Is it caused by feedback from previous SF? Is the efficiency of the SF process fundamentally different than in more massive systems? Answering these questions provides important empirical constraints on simulations of the evolution of low-mass galaxies (e.g., Hopkins et al. 2014; Oñorbe et al. 2015).

The subject of this work, KDG 215 (originally cataloged in Karachentseva 1968, also known as LEDA 44055, F575−3 from Schombert et al. 1992, or D575−5 from Schombert et al. 1997), is a galaxy that possesses unique characteristics. First, its optical surface brightness is extremely low. Of the more than 175 irregular galaxies studied in Hunter & Elmegreen (2006), KDG 215 has the second-lowest central surface brightness (μ_V = 24.69 ± 0.15 mag arcsecond⁻²). Schombert et al. (2011) found a somewhat higher central surface brightness (μ_V = 23.80 mag arcsecond⁻²). Second, KDG 215 has a current star formation rate (SFR) of zero. Of the more than 60 low surface brightness galaxies in Schombert et al. (2011), KDG 215 is one of only four Hα non-detections (see also Karachentsev & Kaisina 2013). Third, the source is nearby and has sufficiently deep Hubble Space Telescope ( HST) images to allow precision color–magnitude diagram (CMD) work. The distance measurement by Karachentsev et al. (2014) places the object securely in the Local Volume (D = 4.83 ± 0.34 Mpc). Fourth, the source is gas rich. The H i properties were first measured in Salzer et al. (1990) and Eder & Schombert (2000), in which the total H i flux integrals were measured to be S_{H i} = 4.48 Jy km s⁻¹ and 4.37 Jy km s⁻¹, respectively. The recently completed Arecibo Legacy Fast ALFA (ALFALFA) catalog (Haynes et al. 2018) revises the total H i flux integral up to S_{H i} = 5.51 ± 0.06 Jy km s⁻¹. Finally and most importantly, as we demonstrate in this manuscript, KDG 215 has a star formation history (SFH) that is extreme compared to that of any other known dwarf galaxy: a significant fraction of the stellar mass has been formed within the last 1.25 Gyr.

2.1. Very Large Array (VLA) H i Observations

2.2. HST Observations

Figure 1(a) shows the color HST image of KDG 215 (F606W as blue, F814W as red, and the linear average of the two filters as green). The low surface brightness nature of the stellar population of KDG 215 is readily apparent. The stellar population has a total physical extent of ∼2 kpc, and is dominated by the emission from stars with blue colors.

Zoom In Zoom Out Reset image size

The F814W versus (F606W−F814W) CMD is shown in panel (b) of Figure 1. The CMD is populated with more than 11,000 individual stars, distributed in four main regions: the blue plume (containing main-sequence and blue helium-burning stars), the red giant branch, the red supergiant region, and red clump stars. The density of stars at fainter magnitudes is higher, and so these regions are plotted as number density contours for ease of interpretation.

From this CMD, the TRGB distance is identified as ${5.11}_{-0.17}^{+0.33}\,\mathrm{Mpc}$ (shown by the thick red line and shaded region of the CMD). This value is slightly larger than the previous TRGB estimate in Karachentsev et al. (2014), although the measurements agree within errors. At this distance, the H i mass of KDG 215 is M_{H i} = (3.40 ± 0.34) × 10⁷ M_⊙. The current stellar mass (using the total stellar mass formed and a return fraction of 0.412 from Vincenzo et al. 2016; see the discussion in Section 5) is ${M}_{\star }=({8.0}_{-3.2}^{+1.2}$) × 10⁶ M_⊙.

H i gas is detected at the 3σ level or above over ∼35 km s⁻¹ in the final datacube. The collapse of the three-dimensional datacube into two-dimensional moment maps produces the images presented in Figure 2. Panel (a) shows the H i mass surface density image in units of 10²⁰ cm⁻². The H i gas is well resolved by the beam (roughly 10 full beam widths across the galaxy). The peak H i column density is 7.5 × 10²⁰ cm⁻², which corresponds to an H i mass surface density N_{H i} = 6 M_⊙ pc⁻². Figure 3 compares the H i column densities (shown as contours) with images from HST (analyzed in detail below), GALEX (tracing SF on 100–200 Myr timescales), and Spitzer (tracing the majority of the stellar mass, though significantly contaminated by foreground and background objects). The H i mass surface density maximum is roughly co-spatial with the region with the highest UV and optical flux.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 2(b) and (c) show the intensity weighted velocity field and velocity dispersion, respectively. The H i kinematics of KDG 215 are complex. There is a weak signature of projected rotation with magnitude ∼10 km s⁻¹, oriented more or less south-to-north. However, there are no single values for the kinematic major or minor axes. Further, the magnitude of the projected rotation (∼10 km s⁻¹) is only slightly larger than the average H i velocity dispersion throughout the disk (∼8–10 km s⁻¹). As discussed in detail in McNichols et al. (2016), this signifies the empirical transition between galaxies with obvious rotational support (where rotation velocity largely exceeds the H i velocity dispersion) and galaxies that are pressure supported. Attempts to fit tilted ring models to the H i velocity field were not successful. Fitting the three-dimensional cube with Gaussians to extract a velocity field produced no significant changes from the intensity weighted image shown in Figure 2(b). With the current H i data we are unable to measure the dynamical mass of KDG 215 with confidence.

Figure 4 presents the SFH of KDG 215 as derived from the HST images. These plots are created using the CMD-fitting algorithm MATCH (Dolphin 2002), to which we refer the reader for details. Very simply, MATCH creates simple stellar populations that are combined to create a best-fit CMD based on a Poisson likelihood statistic. The photometric depth of our data reaches ∼2 mag below the tip of the red giant branch (see Figure 1), allowing us to explore the most recent ∼1 Gyr with significant time resolution. The HST data do not have sufficient photometric depth to parse the SF episodes beyond ∼1 Gyr, and so everything older than 1.25 Gyr is grouped into one single time bin. For details on the temporal resolution of SFHs derived from HST data, see the discussion in McQuinn et al. (2010).

Zoom In Zoom Out Reset image size

Figure 4(a) shows the global SFR of KDG 215 as a function of (logarithmic) time. This demonstrates that KDG 215 had a low SFR for most of the Hubble time. Within the most recent Gyr, KDG 215 underwent a significant episode of strong SF. The peak intensity (∼0.01 M_⊙ yr⁻¹) occurred around 1 Gyr ago, and then fell steadily to the current rate (consistent with zero). This SF event produced more than half of the total stellar mass formed in KDG 215 (see discussion below).

Figure 4(b) shows the cumulative SFH of KDG 215 as a function of (linear) time. The fraction of the total stellar mass ever formed in the galaxy [${M}_{\star }=({1.38}_{-0.56}^{+0.20}$) × 10⁷ M_⊙] rises from zero to unity at z = 0. The first time bin is large, covering log(age) 9.1–10.15 (corresponding to look-back times 1.25–14.1 Gyr). At the end of this first time bin, the fraction of the total stellar mass formed is ${0.44}_{-0.10}^{+0.26}$. Interpreted at face value, the galaxy had formed only 44% of its total stellar mass. Due to the limitations imposed by the photometric depth, the uncertainties on this fraction (which include both random and systematic terms, with the latter dominating the error budget) allow for a range between 34% and 70%. From the best-fitting SFH, the median look-back time by which 50% of the mass of all stars ever formed had been created in KDG 215 is of the order of a Gyr. It is important to note that while the median age of the stellar population is young, ancient stellar populations (predominantly red giants; see Figure 1) are present.

The delayed stellar mass assembly of KDG 215 makes it an important test of models of structure formation. The significant burst of SF that occurred ∼1 Gyr ago is qualitatively similar in terms of relative strength and duration to the strongest bursts seen in the simulations of Governato et al. (2015). The primary difference is that this SF episode occurs at a much later epoch in KDG 215 than those seen in the simulations. Baryonic physics could allow for this delayed stellar mass assembly. As discussed in Christensen et al. (2016), the efficiency of outflows increases as galaxy mass decreases. If early SF (i.e., activity at z ≳ 0.5, in the oldest time bin shown in Figure 4) drove significant outflow episodes that were followed by eventual gas recycling, then the overall SFR in KDG 215 would be low over cosmological timescales.

The triggering mechanism of the starburst episode ∼1 Gyr ago may be related to the local environment of KDG 215. Figure 5 plots the position of KDG 215 along with all galaxies from the Tully et al. (2016) database located in a 6 Mpc cube centered roughly on KDG 215. Positions are plotted in Supergalactic coordinates. Panel (a) shows the SGX versus SGY plane (i.e., looking down on the Supergalactic plane), while panel (b) shows the SGY versus SGZ plane (in which the Supergalactic plane is clearly defined). The nearest neighbor galaxies of KDG 215 are the spiral M64 and the irregular galaxy IC 4107 (also known as KK 177; Karachentseva & Karachentsev 1998). For KDG 215 at 5.11 Mpc (this work), M64 at 5.30 Mpc (Tully et al. 2016), and KK 177 at 4.82 Mpc (Karachentsev et al. 2018), the physical separations are 295 kpc (KDG 215 and M64) and ∼400 kpc (KDG 215 and KK 177).

Zoom In Zoom Out Reset image size

If KDG 215 encountered the higher density environment of M64 a few Gyr ago, then this may have initiated the SF event. This scenario is similar to the reignition of SF seen in simulations of low-mass field dwarfs (Wright et al. 2018). SF is quelled by stellar feedback or reionization, but much of the gas heated by these mechanisms remains in the halos of the galaxies. Interactions with low-density streams of gas in the local environment of the galaxies compresses the gas in the halos, inducing SF. While the characteristics of such suppression and reignition are typically only discernable in SFHs with higher temporal resolution (created from a CMD that reaches below the old main-sequence turn-off), the period of quiescence in KDG 215 is so extended that the suppression and reignition is apparent in Figure 4.

KDG 215 is one of only a few known galaxies that demonstrate such a delayed SFH. The closest known analogs are the Local Group galaxies Leo A (Cole et al. 2007) and the Aquarius dIrr (Cole et al. 2014). Like KDG 215, both systems have a young median look-back time for forming 50% of their stellar mass (4.2 Gyr for Leo A and 6.8 Gyr for Aquarius) and low SFRs. Interestingly, KDG 215 has larger H i and stellar masses than both Leo A (M_⋆ = 3.3 × 10⁶ M_⊙, M_{H i} = 1.1 × 10⁷ M_⊙; Cole et al. 2007; Kirby et al. 2017) and Aquarius (M_⋆ = (1–2)× 10⁶ M_⊙, M_{H i} = 2.7 × 10⁶ M_⊙; Cole et al. 2014). Further, it is slightly more gas-rich (M_{H i}/M_⋆) than both systems.

The distance of KDG 215 does not allow us to parse the ancient SFH with the temporal precision that was achieved for Leo A and Aquarius. The full error budget on the fraction of total stellar mass formed at a look-back time of 1.25 Gyr implies that KDG 215 may have formed as much as 70% or as little as 34% of its total stellar mass at this point. However, it is important to stress that even if the fraction is 70%, KDG 215 still stands out as extreme in comparison to Leo A and Aquarius. At a look-back time of 1.25 Gyr, both Leo A and Aquarius have each formed more than 90% of their total stellar mass (see Figure 8 of Cole et al. 2014). The SFH of KDG 215 is both unique and extreme.

The authors thank the NRAO for making the "Observing for University Classes" program available to the astronomical community. The authors are grateful to Macalester College for supporting this project. I.K. acknowledges support by RFBR grant 18-02-00005.

Facility: VLA - Very Large Array, HST.