Tidal destruction in a low mass galaxy environment: the discovery of tidal tails around DDO 44

TL;DR

This paper reports the discovery of a large stellar tidal stream around the dwarf galaxy DDO 44, indicating tidal disruption likely caused by interaction with the nearby galaxy NGC 2403, and discusses the rarity of such systems in simulations.

Contribution

First detection of a stellar tidal stream around a low-mass galaxy DDO 44, providing insights into tidal interactions in low-mass galaxy environments.

Findings

Tidal streams extend over 50 kpc from DDO 44.

Approximately 25-30% of the total luminosity is in the streams.

Analog systems in simulations are rare, especially at large separations.

Abstract

We report the discovery of a $>1^\circ$ ($\sim50$ kpc) long stellar tidal stream emanating from the dwarf galaxy DDO 44, a likely satellite of Local Volume galaxy NGC 2403 located $\sim70$ kpc in projection from its companion. NGC 2403 is a roughly Large Magellanic Cloud stellar-mass galaxy 3 Mpc away, residing at the outer limits of the M 81 group. We are mapping a large region around NGC 2403 as part of our MADCASH (Magellanic Analogs' Dwarf Companions and Stellar Halos) survey, reaching point source depths (90% completeness) of ($g, i$) = (26.5, 26.2). Density maps of old, metal-poor RGB stars reveal tidal streams extending on two sides of DDO 44, with the streams directed toward NGC 2403. We estimate total luminosities of the original DDO 44 system (dwarf and streams combined) to be $M_{i, \rm{tot}} = -13.4$ and $M_{g, \rm{tot}} = -12.6$, with $\sim25-30\%$ of the luminosity in the streams. Analogs of $\sim$LMC-mass hosts with massive tidally disrupting satellites are rare in the Illustris simulations, especially at large separations such as that of DDO 44. The few analogs that are present in the models suggest that even low-mass hosts can efficiently quench their massive satellites.

Tables

Table 1. Table 1: Properties of DDO 44 and its stream

Parameter	Value	Reference
RA	07:34:11.50	NED
Dec	+66:52:47.0	NED
$m-M$	27.36$\pm$0.07	this work
$D$ (Mpc)	2.96$\pm$0.10	this work
$M_V$	$-12.9$	this work
$M_{\ast }$ ($M_{\odot }$)	$2\times {10}^7$	this work
$\left[\mathrm{Fe}/H\right]$	$-1.6\pm 0.3$	this work
Ellipticity	0.6	this work
$R_e$ (kpc)	0.74$\pm$0.02	Jerjen et al. (2001)
(B)	26.00	Jerjen et al. (2001)
Hi mass ($M_{\odot }$)		KK07aaKarachentsev & Kaisin (2007)
stream extent ($°$)	$\sim 1°$	this work
stream extent (kpc)	$\sim 50$	this work
$L_{\mathrm{stream}}/L_{\mathrm{tot}}$bbFraction of luminosity in the stream.	$\sim 20-30\%$	this work

Full text

galaxies: dwarf, galaxies: halos, galaxies: individual (NGC 2403, DDO 44), galaxies: photometry

††journal: ApJ††facilities: Subaru+HSC, PS1††software: astropy (Robitaille et al., 2013; Price-Whelan et al., 2018), gala (Price-Whelan, 2017), Matplotlib (Hunter, 2007), NumPy (van der Walt et al., 2011), Topcat (Taylor, 2005).

1 Introduction

Deep surveys covering large sky areas have in recent years greatly expanded the number of dwarf galaxy satellites known around the Milky Way (MW; e.g., Kim & Jerjen, 2015; Drlica-Wagner et al., 2015, 2016; Laevens et al., 2015; Homma et al., 2018; Torrealba et al., 2016, 2018) and M31 (e.g., Martin et al., 2016; McConnachie et al., 2018). The newly discovered diminutive galaxies include extremely low-luminosity “ultra-faint dwarfs” (UFDs; e.g., Willman et al. 2005; Belokurov et al. 2007 — see the recent review by Simon 2019), as well as many relic streams from tidally disrupted satellites criss-crossing the Galactic halo (e.g., Belokurov et al. 2006; Grillmair & Carlin 2016; Shipp et al. 2018). In parallel to these discoveries, models of structure formation and evolution within the $\Lambda$-Cold Dark Matter ($\Lambda$CDM) framework have generated predictions of the number of satellites expected, as well as properties such as their luminosity functions and metallicities (Benson et al., 2002; Zolotov et al., 2012; Wetzel et al., 2016; Jethwa et al., 2018; Bose et al., 2018; Kim et al., 2018; Nadler et al., 2019). Although the match with the MW and M31 — the only systems with robust samples of satellites — is remarkably good, it is unclear whether our Galaxy and its nearest massive neighbor are representative of massive galaxies more generally, or if theoretical models are over-tuned to the Local Group. Resolved stellar maps of nearby massive galaxies are now being painstakingly assembled, revealing the satellite systems (down to the scale of ultrafaint dwarf galaxies) of Cen A (Crnojević et al., 2019), NGC 253 (Sand et al., 2014; Romanowsky et al., 2016; Toloba et al., 2016), M81 (Chiboucas et al., 2013), M101 (Merritt et al., 2014; Bennet et al., 2017; Danieli et al., 2017; Müller et al., 2017; Bennet et al., 2019) and M94 (Smercina et al., 2018), among others. We are thus entering an era in which we may explore the stochasticity of satellite populations around a variety of hosts, as well as their dependence on environment and host properties (e.g. Bennet et al., 2019). These can be used to make more precise tests of galaxy formation and the $\Lambda$CDM cosmological model.

Of particular interest are satellites in less dense environments than the ones highlighted above. Satellites of the MW in particular show signs of experiencing many types of environmental quenching and disruption simultaneously (Barkana & Loeb, 1999; Mayer et al., 2006; Grcevich & Putman, 2009; Nichols & Bland-Hawthorn, 2011; Brown et al., 2014; Slater & Bell, 2014; Fillingham et al., 2015; Wetzel et al., 2015). Because so many processes are likely to affect the satellites, it is often difficult to assess their relative importance, and how that importance scales with properties of the host (not necessarily limited to halo mass). One way to disentangle these processes, and to highlight the scales at which each mechanism kicks in, is to consider low-mass hosts. Hosts inhabiting halos smaller than the Milky Way’s ought not to have hot-gas halos (Birnboim & Dekel, 2003), so ram-pressure stripping and possibly starvation may be significantly reduced as compared to the MW (although they are likely to have cool circumgalactic media (CGM); Bordoloi et al., 2014). Moreover, they should have gentler tidal fields, reducing the effects of tidal heating and stripping. Thus, we expect satellite galaxies of low-mass hosts to be more like field galaxies, and less influenced by their environment. The environmental processes that are important for less massive hosts are likely to be different than those most relevant to MW-sized galaxies. These hypotheses remain to be tested.

We have an additional motivation to study satellites of low-mass galaxies, in that many of the recently discovered dwarf galaxies within the MW halo are thought to have originated as satellites of the Large and Small Magellanic Clouds (LMC, SMC), and only recently fell into the MW (e.g., Jethwa et al., 2016; Sales et al., 2017; Dooley et al., 2017; Kallivayalil et al., 2018). The luminosity function of the new discoveries is unexpected, though — there are no massive ($M_{*}>10^{4}M_{\odot}$) MC satellite candidates (though Pardy et al. 2019 suggest that the Carina and Fornax dSphs may be associated with the MCs), but many that are much smaller, at odds with typical stellar-mass–halo-mass relations (Dooley et al., 2017).222A similar result is found for the MW as a whole (Kim et al., 2018). It is unknown whether the LMC and SMC had a more typical luminosity function at infall and many satellites have been stripped from them by the MW, or if this luminosity function is typical and is telling us something new about galaxy formation in small halos.

To extend the mass range of hosts for which satellite searches have been carried out to lower-mass (Magellanic Cloud-mass) systems, without the difficulty of interpreting the interplay of the LMC and its satellites with the Galactic halo, we are conducting a census of nearby LMC stellar-mass analogs. With this survey — Magellanic Analogs’ Dwarf Companions and Stellar Halos (MADCASH) — we are searching for the satellite populations of MC-mass galaxies within $\sim 4$ Mpc of the MW. Some early results from this ongoing survey include the discovery of the $M_{\rm V}\sim-9.7$ dwarf galaxy Antlia B near NGC 3109 (Sand et al., 2015; Hargis et al., 2019), the detection of extended stellar populations around nearby galaxy IC 1613 (Pucha et al., 2019), and our discovery of a faint ($M_{\rm V}\sim-7.7$) satellite of NGC 2403 (Carlin et al., 2016). In this work, we highlight the discovery of a dwarf satellite being tidally disrupted around nearby ($D\sim 3.2$ Mpc; Karachentsev et al. 2013) low-mass (stellar mass $M_{*}\sim 7\times 10^{9}~{}M_{\odot}$; roughly $2\times$ LMC stellar mass) spiral galaxy NGC 2403, a relatively isolated system at the outskirts of the M81 group.

The dwarf spheroidal DDO 44 is a relatively massive dwarf ($M_{\rm R}\sim-13.1$, similar to the Fornax satellite of the MW; Karachentsev et al. 1999) that is at a distance and velocity consistent with orbiting as a satellite of NGC 2403. Here we report evidence that the dwarf spheroidal DDO 44 has stellar tidal tails extending at least $\sim 0.5\arcdeg$ ($\sim 25$ kpc) from its center. This discovery is based on data from our deep, wide-area imaging survey to a projected radius of $\sim 100$ kpc around NGC 2403.

In Sec. 2, we introduce our discovery data set and analysis procedure. We show the key characteristics of the stream in Sec. 3. In Sec. 4, we discuss what the stream implies for the relationship between the orbital and star-formation histories of DDO 44, and the frequency of small galaxy disruption by low-mass hosts. We highlight our key results in Sec. 5.

2 Data and Analysis

Deep imaging data were obtained with Hyper Suprime-Cam (HSC; Furusawa et al. 2018; Kawanomoto et al. 2018; Komiyama et al. 2018; Miyazaki et al. 2018) on the Subaru 8.2m telescope. The $1.5^{\circ}$ diameter field of view of HSC corresponds to $\sim 80$ kpc at the $D\sim 3.0$ Mpc distance of NGC 2403, enabling a relatively efficient survey to beyond a projected radius of $d>100$ kpc around NGC 2403 (a large fraction of the $\sim 120-180$ kpc virial radius of an isolated LMC-mass analog; see, e.g., estimates of $R_{\rm vir}$ in Dooley et al. 2017). Our data consist of seven HSC pointings (see map in Figure 3): the CENTER, EAST, and WEST fields were observed on 2016 February 9–10, while the four additional HSC fields (NW, NE, SW, and SE) were observed 23–24 December 2017. All observations consist of $10\times 300$s exposures in $g$-band (known as “HSC-G” at Subaru) and $10\times 120$s in $i$ (“HSC-I2”). We also observed short, $5\times 30$s sequences of exposures to improve photometry at the bright end. All observations from both runs were obtained in seeing between $\sim 0.5-0.9^{\prime\prime}$, under clear skies.

The data were processed with the LSST pipeline, a version of which was forked to create the reduction pipeline for the HSC-SSP survey (Aihara et al., 2018a, b). Details of the reduction steps can be found in Bosch et al. (2018). In short, we performed forced PSF photometry on co-added frames in each filter, and calibrated both astrometrically and photometrically to PanSTARRS-1 (PS1; Schlafly et al. 2012; Tonry et al. 2012; Magnier et al. 2013). We applied extinction corrections based on the Schlafly & Finkbeiner (2011) coefficients derived from the Schlegel et al. (1998) dust maps. All results presented in this work are based on extinction-corrected PSF magnitudes.

For separation of point sources from unresolved background galaxies, we compare the ratio of PSF to cmodel fluxes for all sources, where the cmodel is a composite bulge/exponential plus de Vaucouleurs profile fit to each source. Point sources should have flux ratios $f_{\rm PSF}/f_{\rm cmodel}\sim 1$, while extended sources will contain additional flux in the model measurement that is not captured by the PSF. Figure 1 shows the $i$-band flux ratio as a function of $i$-band PSF magnitude. A large number of sources (especially at the bright end) are concentrated around unity in this figure. We allow for an intrinsic width of $\pm 0.03$ in the flux ratio, and select sources whose $1\sigma$ uncertainties in flux ratio place them within this $\pm 0.03$ window. The point source candidates selected in this way are shown as red points in Figure 1, with extended sources (i.e., “not point sources”) as gray points. Note that some background galaxies will contaminate the point source sample below $i\sim 23$, where background galaxies far outnumber stars. Through the rest of this work we will analyze only this point source sample calibrated to the PS1 photometric system.

To characterize the completeness of our photometric catalog, we injected artificial stars into the images using Synpipe (Huang et al., 2018), which was written for HSC-SSP, and has been incorporated into the LSST pipeline. The resulting completeness (i.e., the fraction of artificial stars recovered by the photometric pipeline) in a $20\arcmin$ region centered on DDO 44 (excluding the central $2\arcmin$ where crowding is too extreme to resolve stars; a total of 18975 artificial stars were injected in this region, or $\sim 15~{}{\rm arcmin}^{-2}$) is given in Figure 2. We fit a function of the form given in Martin et al. (2016) to these curves, and estimate 50% (90%) completeness limits of $i=26.7~{}(26.2)$ and $g=27.5~{}(26.5)$. In the lower panels of the figure, we compare the input and recovered magnitudes for the artificial stars in both bands. These are centered on zero, so we are confident that no systematic offsets are present in our photometry.

3 A Stream around DDO 44

One of the primary goals of our large-area imaging campaign around NGC 2403 is to search for its dwarf galaxy companions and/or the remnants of destroyed satellites. Thus one of the first things we did upon finishing the data reduction was to select stars with color-magnitude diagram (CMD) positions consistent with metal-poor RGB stars at the distance of NGC 2403, and plot their density on the sky. This RGB density map centered on NGC 2403 is shown in Figure 3. For the most part, the map shows a fairly uniform distribution over the entire region surveyed. This is most likely predominantly fore-/back-ground contamination, with little or no “halo” population in the outer regions around NGC 2403. We note that the $0.75^{\prime}$ bins in this map are about the size of a typical faint dwarf galaxy at the distance of NGC 2403 — $1^{\prime}\approx 0.9$ kpc at $D=3.0$ Mpc. Thus the faint dwarf galaxy MADCASH J074238+652501-dw found by Carlin et al. (2016) is almost completely contained in a single pixel of this map (approximately at $(\alpha-\alpha_{0})\sim 0.5$ and $(\delta-\delta_{0})\sim-0.2$), and thus not visible as an obvious overdensity. The most striking feature in Figure 3 is the prominent blob corresponding to the known dwarf spheroidal galaxy DDO 44 to the north (and slightly west) of NGC 2403. Our deep Subaru+HSC data enable us to see for the first time that DDO 44 has streams of stars emanating from it, oriented along the direction toward (and away from) NGC 2403; i.e, DDO 44 is tidally disrupting beyond doubt.

3.1 TRGB distance, isochrone fit

In Figure 4 (left panel) we show a CMD of all stars between 2–4 arcmin of the center of DDO 44. There is a prominent metal-poor RGB (highlighted by the solid black box) as well as a significant number of AGB stars (dashed gray box) signaling the presence of intermediate-age ($\sim 2-8$ Gyr old) stellar populations (as was found in HST imaging by Karachentsev et al. 1999 and Alonso-García et al. 2006). The dotted blue box in each panel denotes the location where young main-sequence stars would appear, if present. By comparing the DDO 44 field with an equal-area background region (right panel), the number of objects in this box is consistent with being background (likely unresolved galaxies/QSOs, given their blue colors). In our later analysis, we show (e.g., the lower panel of Fig. 5) that there is no concentration of blue sources at the position of DDO 44, confirming that these are background objects rather than young stars in DDO 44.

To refine the distance of DDO 44, we first selected stars with colors consistent with metal-poor RGB stars ($0.8<(g-i)_{0}<2.1$), and within $6^{\prime}$ of the center of DDO 44. We binned these stars as a function of magnitude, then convolved the luminosity function with a zero-sum Sobel edge-detection filter (in particular, one with values [$-1.0,-2.0,-1.0,0.0,1.0,2.0,1.0$], as in Jang & Lee 2017). We identify a narrow peak in the convolved luminosity function, corresponding to the tip of the RGB (TRGB), at $i_{\rm TRGB}=23.95\pm 0.05$. From 10-Gyr Dartmouth isochrones (Dotter et al., 2008) at [Fe/H]$=-1.6\pm 0.3$, we estimate an $i$-band TRGB absolute magnitude333Note that the $i$-band TRGB magnitude is virtually independent of metallicity for metal-poor populations (as we confirmed with isochrones of various metallicities), so that our choice of metallicity has no bearing on the derived distance modulus. (in the PS1 system) of $-3.41^{+0.02}_{-0.01}$. We thus derive a distance to DDO 44 of $2.96\pm 0.10$ Mpc (i.e., distance modulus $(m-M)_{0}=27.36\pm 0.07$). We note that the same TRGB code applied to the main body of NGC 2403 yields an identical distance modulus of $(m-M)_{\rm N2403}=27.36$. Our derived distance modulus is in agreement with other recent determinations for DDO 44 ($(m-M)_{0}=27.36\pm 0.09,27.39\pm 0.13$, and 27.45; Jacobs et al. 2009; Alonso-García et al. 2006; Dalcanton et al. 2009), though somewhat at odds with the value of $(m-M)_{0}=27.50$ given in the COSMICFLOWS-3 database (Tully et al., 2016).

At a distance of 2.96 Mpc, the on-sky separation between DDO 44 and NGC 2403 of $78.3\arcmin$ corresponds to a projected separation of $\sim 67$ kpc.

After determining the TRGB magnitude (and thus distance) of DDO 44, we wish to estimate the system’s metallicity. To do so, we create a set of old (10 Gyr) Dartmouth isochrones in 0.1-dex metallicity intervals, and use a least-squares minimization based on differences between the DDO 44 stellar sample and the isochrones to find a best-fitting metallicity of [Fe/H]$=-1.6\pm 0.3$. This mean metallicity is consistent with those measured by Karachentsev et al. (1999, $-1.7\pm 0.4$), Alonso-García et al. (2006, $-1.54\pm 0.14$), and Lianou et al. (2010, $-1.67\pm 0.19$) via HST imaging. Figure 4 shows a CMD of the central $2-4\arcmin$ field around DDO 44, with the best fit isochrone at [Fe/H] = $-1.6$ overlaid, along with isochrones at $\pm 0.5$ dex in metallicity.

The RGB of DDO 44 is wider than expected solely based on photometric errors. This could be due in part to photometric scatter induced by the significant unresolved emission in the body of DDO 44. However, Alonso-García et al. (2006) estimated that as much as $20\%$ of the total stellar content of DDO 44 is contributed by the intermediate-age population (likewise, Lianou et al. 2010 found the fraction of AGB stars relative to RGB number to be $f_{\rm AGB}=0.11$). Thus, a single 10-Gyr population should not be expected to reproduce the width of the RGB. Estimating the relative contributions of the different age populations (i.e., a star formation history) is beyond the scope of the current study (and is typically best achieved with data reaching the oldest MSTO). Finally, we note that we calculated a metallicity distribution under the assumption that only 10 Gyr populations were present (assigning stellar metallicities based on isochrones), and found a metallicity spread of $\sim 0.49$ dex (determined by fitting a Gaussian to the distribution). The mean metallicity and metallicity spread (with the caveat that we have assumed a single age) is similar to those of Milky Way dSphs with similar luminosities (e.g., the Sculptor dSph; Simon 2019), suggesting that some of the RGB width is contributed by a metallicity spread in DDO 44, while the presence of intermediate-age populations may account for some additional broadening of the RGB.

Based on the lack of blue stars in HST images of DDO 44, Karachentsev et al. (1999) estimated that the most recent star formation in DDO 44 was at least 300 Myr ago. We do not see evidence of this young population beyond $2\arcmin$. However, a significant population of bright AGB stars above the TRGB (also seen in HST data by Karachentsev et al. 1999 and Alonso-García et al. 2006) suggests that intermediate-age populations are present in the outer regions of DDO 44. Indeed, as noted previously, Alonso-García et al. (2006) found that $\sim 20\%$ of the stellar population of DDO 44 consists of intermediate-age (between $\sim 5-8$ Gyr, and at least older than 2 Gyr) populations.

3.2 Stellar populations in the stream

To facilitate analysis of the stream, we first derived the transformation to a coordinate system aligned with the stream. We determined the central position at points along the stream by fitting Gaussians to the stellar density in slices of $0.1^{\circ}$ in declination. Using the two points immediately adjacent to DDO 44, but on the north and south side, we derived the transformation to a great circle coordinate frame using the gala444http://gala.adrian.pw/en/latest/ software. This transformation places DDO 44 at the origin, with angle $\phi_{1}$ along the stream, and $\phi_{2}$ perpendicular to the stream. A map of RGB stars in the transformed coordinates is shown in the left panel of Figure 5.

We then selected narrow strips of $\left|\phi_{2}\right|<0.05^{\circ}$, and extracted an RGB star density profile as a function of $\phi_{1}$ (i.e., along the stream). This profile is shown in the right panel of Figure 5, where we have subtracted the mean density in background regions that do not contain bright star holes (seen as white voids in Fig. 5). The three background regions are at $-0.7\arcdeg<\phi_{1}<-0.1\arcdeg$, $0.15\arcdeg<\phi_{2}<0.4\arcdeg$; $-0.2\arcdeg<\phi_{1}<0.35\arcdeg$, $-0.4\arcdeg<\phi_{2}<-0.15\arcdeg$; and $0.2\arcdeg<\phi_{1}<0.7\arcdeg$, $0.15\arcdeg<\phi_{2}<0.4\arcdeg$. This density profile shows RGB overdensities extending to at least $0.3^{\circ}$ from DDO 44 on either side, or $\sim 15$ kpc at the distance of DDO 44.

Fig. 6 highlights CMDs of stars extracted in bins along the stream. The central panel contains the core of DDO 44 (within $0.1^{\circ}$), with panels to the left (south) and right (north) showing similar stellar populations extending into the stream. On both sides of DDO 44, the stream is barely noticeable (if at all) at $\left|\phi_{1}\right|\gtrsim 0.4^{\circ}$. The best-fit isochrone with [Fe/H] = $-1.6$ is a good match to the RGB stellar populations in both the core and stream. The bright AGB stars visible in the central regions of DDO 44 are seen in small numbers at all radii (and indeed, the density profile seen in Fig. 5 suggests that the AGB stars extend as far as the RGB stars in the stream). The fact that the oldest RGB stars and the intermediate-age AGB populations are both extended suggests that DDO 44 had little to no population gradient in its core before being tidally disrupted (however, Lianou et al. 2010 found that metal-rich populations are more centrally concentrated in DDO 44 than metal-poor stars).

3.3 Total luminosity

To estimate the total luminosity of DDO 44, including stars in its streams, we summed the flux of all RGB stars brighter than $i_{0}<26.5$, applying a completeness correction to each star’s flux based on the fits in Figure 2. We then corrected for the unmeasured luminosity below the cutoff magnitude using a Dartmouth isochrone (Dotter et al., 2008) with [Fe/H]$=-1.6$, 10 Gyr age, and power-law luminosity function slope of $-1.5$.555We note that if instead we use a Salpeter IMF, our derived total luminosity of DDO 44 changes by $<0.03$ mags. From this luminosity function, we determine that $\sim 22\%$ of the flux is in stars brighter than $i=26.5$; we thus apply a correction to the total flux to account for the remaining 78% of the light. Finally, we account for the “missing” data due to stellar crowding near the center of DDO 44 by excluding the inner $2\arcmin$ from our calculations. Adopting $\mu_{\rm R,0}=24.1$ mag arcsec*-2* and a scale length of $39\arcsec$ (Karachentsev et al., 1999), we estimate that $\sim 60\%$ of the light is contained within our excluded $2\arcmin$ region.

We find $M_{i,{\rm tot}}=-13.4$ and $M_{g,{\rm tot}}=-12.6$ for the total luminosity of DDO 44 and the stars in its streams, where we take the region between $|\phi_{1}|<0.1\arcdeg$ to be the main body of DDO 44. Of the total, $\sim 17\%$ and $\sim 11\%$ of the flux is contained in the northern ($0.1\arcdeg<\phi_{1}<0.7\arcdeg$) and southern ($-0.7\arcdeg<\phi_{1}<-0.1\arcdeg$) portions of the stream, respectively. Given the many large corrections detailed in the previous paragraph, it is difficult to place uncertainties on these estimates. To facilitate comparison to the measurement by Karachentsev et al. (1999) of $M_{R}=-13.1$, we transform these absolute magnitudes in the PanSTARRS system to Johnson-Cousins $R$-band using the relations from Table 6 of Tonry et al. (2012). This yields a total $M_{R}\sim-13.3$ based on our HSC measurements. Removing the $\sim 20-30\%$ of the resolved stars’ flux that we find beyond $6\arcmin$ of the DDO 44 center would reduce our derived luminosity of DDO 44 by $\sim 0.2-0.3$ mag, placing our estimate for the central body of the galaxy in excellent agreement with that of Karachentsev et al. (1999).

To make connection with theory, it is useful to translate from the object’s absolute magnitude to stellar mass. Our measured luminosity transforms to $M_{V}=-12.9$ (it is Local Group convention to report $V$-band absolute magnitudes), which corresponds to $M_{*}=2.0\times 10^{7}~{}M_{\odot}$ (assuming $V$-band stellar $(M/L)_{\rm V}$ of 1.6; Woo et al. 2008). Note that an estimate based on the $K$-band luminosity from Karachentsev et al. 2013, assuming $(M/L)_{\rm K}=1$, gives $M_{*}=6.0\times 10^{7}~{}M_{\odot}$ for DDO 44.

By fitting Gaussians to the resolved stellar surface density along the major and minor axes, we find an ellipticity $\epsilon\equiv 1-b/a\approx 0.6$. It is unsurprising to find that DDO 44 is rather extended, and that its ellipticity is similar to that of the tidally disrupting Sagittarius dSph ($\epsilon=0.64$; McConnachie 2012). Also like the Sagittarius dSph, DDO 44’s surface brightness (as measured by Jerjen et al. 2001) lies below typical dwarfs at its luminosity (see, e.g., Figure 7 from McConnachie 2012), as expected for a system undergoing tidal disruption. DDO 44’s closest analogs in luminosity and stellar mass are, according to McConnachie (2012), Sagittarius ($M_{V}=-13.5$), Fornax ($M_{V}=-13.4$), and And VII ($M_{V}=-12.6$). Of these three, only Sagittarius is clearly disrupting — deep imaging data show no hints of tidal features for Fornax (Wang et al., 2019), and the ellipticities of Fornax and And VII are far lower. Our derived metallicity for DDO 44 of [Fe/H] $=-1.6$ is near, but slightly on the low metallicity side of, the luminosity-metallicity relation for Local Group dwarf galaxies (e.g., Kirby et al. 2013, McConnachie 2012). The metallicities of Sagittarius, Fornax, and And VII are all significantly higher (Kalirai et al., 2010; Kirby et al., 2011; Carlin et al., 2012; McConnachie, 2012; Hasselquist et al., 2019).

We summarize the properties of DDO 44 and its stream in Table 4.1, including some relevant data from the literature.

4 DDO 44 and its streams in context

DDO 44 is clearly a disrupting dwarf, but questions remain about its history and association with a larger host. In this section, we argue that NGC 2403 is the most likely host for DDO 44. This conclusion allows us to use simulations to estimate how rare (or not) it is for a large dwarf to be disrupting around a low-mass host, and consider how the orbit of DDO 44 explains various features of its star-formation history (SFH). We may also place DDO 44 in the context of the NGC 2403 satellite system, and consider whether NGC 2403’s satellite luminosity function is in line with expectations from the $\Lambda$CDM paradigm.

4.1 DDO 44 is a satellite of NGC 2403

We consider whether DDO 44 is in fact a satellite of NGC 2403 or of the neighboring galaxy NGC 2366.

DDO 44 has a heliocentric radial velocity of $213~{}{\rm km~{}s}^{-1}$ (Karachentsev et al., 2011; Tully et al., 2016)666Note that this velocity is apparently based solely on an HII region offset from the center of DDO 44, but likely associated with it. We could not locate any extant velocity measurements based on the stellar body of DDO 44., while the Tully et al. “COSMICFLOWS-3” catalog gives $v_{\rm hel}=141~{}{\rm km~{}s}^{-1}$ for NGC 2403. This small difference in their relative velocities (for context, this velocity difference of $\sim 70$ km s*-1* is much less than the escape velocity from NGC 2403 of $\gtrsim 200$ km s*-1*; see Fig. 7), in addition to the nearly identical distance moduli of DDO 44 and NGC 2403 (Sec. 3.1) is suggestive of an association between the two galaxies.