The TRGB distance to the second galaxy "missing dark matter". Evidence for two groups of galaxies at 13.5 and 19 Mpc in the line of sight of NGC1052

TL;DR

This study uses the Tip of the Red Giant Branch method to determine galaxy distances, revealing two galaxy groups at 13.5 and 19 Mpc and clarifying the distance to NGC1052-DF4, impacting dark matter studies.

Contribution

It provides a precise TRGB-based distance measurement to NGC1052-DF4, resolving previous discrepancies and identifying two galaxy groups along the line of sight.

Findings

NGC1052-DF4 is at 14.2 Mpc, closer than previously thought.

The field contains two galaxy groups at 13.5 Mpc and 19 Mpc.

Globular clusters in NGC1052-DF4 are similar to those in the Milky Way.

Abstract

A second galaxy ``missing dark matter" (NGC1052-DF4) has been recently reported. Here we show, using the location of the Tip of the Red Giant Branch (TRGB), that the distance to this galaxy is 14.2+-0.7 Mpc. This locates the galaxy 6 Mpc closer than previously determined. We also analyse the distances to the brightest galaxies in the field-of-view (FOV) of NGC1052. We find this field is populated by two groups of galaxies in projection: one dominated by NGC1052 and NGC1047 at ~19 Mpc and another group containing NGC1042 and NGC1035 (as well as [KKS2000]04 and NGC1052-DF4) at ~13.5 Mpc. At a distance of 13.5 Mpc the globular clusters of NGC1052-DF4 have the same properties than globular clusters in the Milky Way and other dwarf galaxies.

Figures

Tables

Table 1. Table 1: Distances to the galaxies in the FOV of NGC1052. The foreground extinction in the HST bands F606W and F814W are indicated for reference. The velocity v h ​ e ​ l subscript 𝑣 ℎ 𝑒 𝑙 v_{hel} corresponds to the mean heliocentric radial velocity. In order to make a sensible comparison of the velocities of these galaxies with the Hubble Flow, we include also the velocities with respect to the Cosmic Microwave Background (CMB) v C ​ M ​ B subscript 𝑣 𝐶 𝑀 𝐵 v_{CMB} which have been estimated subtracting 217 km/s to v h ​ e ​ l subscript 𝑣 ℎ 𝑒 𝑙 v_{hel} taking into account the location on the sky of these galaxies and the solar motion of 370.06 km/s towards the direction defined by Galactic coordinates (263.914°, +48.2646°). See HYPERLEDA (Makarov et al., 2014 ) for further details.

Name	RA	Dec	Distance	AF606W	AF814W	$v_{hel}$	$v_{CMB}$	Method
	(J2000)	(J2000)	(Mpc)	(mag)	(mag)	(km/s)	(km/s)
			Diffuse galaxies
[KKS2000]04	02:41:46.8	-08:24:09.3	13.0$\pm$0.4	0.060	0.037	1793$\pm$2	1576$\pm$2	TRGB
NGC1052-DF4	02:39:15.1	-08:06:58.6	14.2$\pm$0.7	0.062	0.038	1445$\pm$4	1228$\pm$4	TRGB
			Bright galaxies
NGC1035	02:39:29.1	-08:07:58.6	14$\pm$3	0.062	0.038	1262$\pm$16	1045$\pm$16	Tully-Fisher
NGC1042	02:40:24.0	-08:26:01.0	12.6$\pm$2.3	0.071	0.044	1371$\pm$1	1155$\pm$1	Tully-Fisher
NGC1047	02:40:32.8	-08:08:51.4	19$\pm$2	0.063	0.039	1415$\pm$15	1198$\pm$15	Physical association with NGC1052
NGC1052	02:41:04.8	-08:15:20.8	19$\pm$2	0.066	0.041	1484$\pm$6	1268$\pm$6	SBF

Equations

$$M_{F814W}^{ACS}=-4.06+0.20\times [(F606W-F814W)-1.23]$$

$${\gamma }_I=0.92+1.63(\log (V_{max}-V_{min})-2.5)$$

$$\mu =I-M_I=I+21.39+8.81(\log (V_{max}-V_{min})-2.5)$$

Full text

galaxies: evolution — galaxies: structure — galaxies: kinematics and dynamics – galaxies: formation

††journal: ApJ

1 Introduction

The determination of the physical properties of the astronomical objects relies on our ability to have an accurate determination of their distances. However, measuring reliable distances to most of the astronomical sources remains as one of the most difficult enterprise in our science. Recently, van Dokkum et al. (2018a) and van Dokkum et al. (2019a) have reported the presence of two extraordinary low surface brightness galaxies ([KKS2000]04 and NGC1052-DF4) in the field-of-view of the bright galaxy NGC1052. These two galaxies are proposed to have a dark matter content compatible with zero. Such claim is based on the assumption that both galaxies are at a distance of 20 Mpc.

The distance to both dwarf galaxies was measured using the surface brightness fluctuation (SBF) technique. In particular, van Dokkum et al. (2019a) used the calibration provided by Blakeslee et al. (2010) to derive a distance of $\sim$20 Mpc. Unfortunately, the reliability of using a calibration which is established for red and massive galaxies to determine the distance to bluer and less massive galaxies is not at all obvious. In fact, Blakeslee & Cantiello (2018) warn about the use of such a calibration in a range where it has not been explored: “The SBF method is not well-tested at these colors and low stellar densities…”. A warning that has been vindicated by recent analysis of the SBF methodology exploring low mass galaxies (see e.g. Cantiello et al., 2018; Carlsten et al., 2019) which directly shows that the extrapolation of Blakeslee et al. (2010) is inaccurate111The extrapolation of the relation provided by Blakeslee et al. (2010) is offset by around 0.55 mag for these bluer and low mass galaxies. This corresponds to an artificial increase in the distance of $\sim$4.5 Mpc. to measure the distance to these types of galaxies. Fortunately, there are alternative ways to measure the distance to these diffuse systems.

When galaxies are nearby enough so their stellar populations can be resolved into stars, one of the most precise determination of their distances can be obtained by the location of the TRGB in the colour-magnitude diagram (CMD) of their stars. In a recent paper (Trujillo et al., 2019), we show that the location of the TRGB of the CMD of the galaxy [KKS2000]04 (popularised as NGC1052-DF2) gives a distance of 13.4$\pm$1.1 Mpc222Together with another four independent distance indicators the final distance was established on 13.0$\pm$0.4 Mpc.. This result was shown to be robust against crowding effects as demonstrated by the fact that the distance to the object was independent to the use of stars in the innermost central region or in the periphery of the object. In this Letter, we analyze the CMD of the other dwarf “lacking dark matter” galaxy: NGC1052-DF4. We will show that the location of the TRGB of this galaxy puts this object at a distance of 14.2$\pm$0.7 Mpc. Finally, we also make an analysis of the distances to the brightest galaxies in the FOV of NGC1052. We find tantalising evidence that there are two groups of objects located in the same FOV: one dominated by NGC1052 at $\sim$19 Mpc and another one centred on NGC1042 at $\sim$13.5 Mpc.

2 Data

Hubble Space Telescope (HST) imaging of the NGC1052-DF4 galaxy explored in this work was taken as part of the programme GO-14644 (PI: van Dokkum) and consists of two Advanced Camera for Survey (ACS) orbits for each object: one in F606W (V606; 2180s) and one in F814W (I814; 2320s). A first analysis of this data was conducted by Cohen et al. (2018). Data were retrieved from the MAST archive webpage333https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html. Each orbit is split into four different images whose individual exposures are of 545s (F606W) and 580s (F814W). The pixel size is 0.05″.

The ACS exposures we used were the ones produced by the standard STScI pipeline: i.e. bias- and dark-current-subtracted, flat-fielded and with the CTE correction applied. This produces the calibrated ”flc” images. To create the CMDs of all the diffuse galaxies we follow the prescriptions given by Monelli et al. (2010). The photometry was made on the individual $flc$ images using the DAOPHOT/ALLFRAME set of programs (Stetson, 1987, 1994). We extensively masked the background galaxies in our images to remove potential contamination in our catalogue by artificial deblending of these sources. Our code performs a simultaneous data reduction of the images of a given target, assuming individual Point Spread Functions (PSFs) once an input list of stellar objects is provided. The final list of point-sources was generated on the stacked median image obtained by registering and co-adding the eight individual available frames (four per filter) of each galaxy. The source detection algorithm was iterated twice. The photometry of the stars is given in the Vega system, adopting the zero-points appropriate for the observations date ($zp_{F606W}$ = 26.404, $zp_{F814W}$ = 25.516)444https://acszeropoints.stsci.edu/. The photometric catalogue was finally cleaned using quality cuts. We used the sharpness parameter provided by DAOPHOT, rejecting sources with $|sha|>$ 0.1. The use of the sharpness parameter for effectively cleaning the sample of false detections (i.e. galaxies, unrecognised doubles, bad pixels or cosmic rays) has been routinely adopted in the literature (see e.g. Aloisi et al., 2007; Turri et al., 2017).

3 The TRGB distance to NGC1052-DF4

An accurate estimate of the distance to the galaxies can be obtained using the TRGB when the stellar populations are resolved (Lee et al., 1993). The TRGB marks the end of the red giant branch phase, producing a well-defined discontinuity in the luminosity function of the stars, which can be easily identified from photometric data. This method has two strong advantages: first, the TRGB is an intrinsically bright feature (MI$\sim$ -4) and, second, the luminosity of the TRGB in the I/F814W filters has a very mild dependence on the age and the metallicity (Salaris & Cassisi, 1997). For such reasons, the method can be reliably used beyond 10 Mpc. We took the calibration by Rizzi et al. (2007)

[TABLE]

which accounts for the mild dependency on the metallicity by taking into account a color term. We applied a color correction to the F814W photometry as suggested by McQuinn et al. (2017). This results in a steeper RGB and therefore a better defined TRGB.

The resulting modified and de-reddened CMDs as a function of radius are shown in the left panels of Fig. 1. The location of the areas used to explore the CMDs are displayed in Fig. 2. Fig. 1 shows, from top to bottom, the CMD of sources within 1 Re (Re=16.5 arcsec; Cohen et al., 2018, panel a)), between 1 and 2 Re (panel c)), and within 2 Re (panel e)). The central position of the galaxy was centred at coordinates (R.A.=02:39:15.1 ; Dec=-08:06:58.6) which corresponds to the pixel position (2868.7, 299.5) of chip 2 of the adopted reference image. The horizontal coloured lines mark the position of the TRGB, identified from the luminosity function (LF) displayed as a black histogram in the corresponding right panels. The three LFs present a sudden jump at magnitude F814W0,Vega $\approx$ 26.7 mag. The statistical significance of this feature (assuming a Poissonian distribution) is 3.2$\sigma$ (99.86%), 2.2$\sigma$ (97.2%) and 4.0$\sigma$ (99.99%) (in panel a), c) and e) respectively). The LF function was convolved with a Sobel kernel (Sakai et al., 1996, 1997) K=[-2,-1,0,1,2]. The position of the apparent magnitude of the TRGB (labelled in the right panels) was derived identifying the peak corresponding to the largest jump in the LF.

The distance modulus was derived with the Rizzi et al. (2007) zero points, obtaining $(m-M)_{0}$ = 30.76 $\pm$ 0.10 mag, corresponding to a distance D = 14.2 $\pm$ 0.7 Mpc555Note that the distance to DF4 is compatible with the lower limit TRGB distance ($>$9.7 Mpc) estimated in Cohen et al. (2018). It is worth mentioning that we used a different culling criteria than Cohen et al. (2018) to extract our photometry, as ours is based on DAOPHOT while theirs was based on DOLPHOT.. The final error budget includes the uncertainty of the calibration relation by Rizzi et al. (2007) and the error on the determination of the TRGB position. We also checked how the location of the TRGB would change, if instead of using $|sha|>$ 0.1, we would have used the less conservative $|sha|>$ 0.5. In that case, the location of the TRGB would be brighter by 0.1 mag, making the galaxy closer by 0.7 Mpc (i.e. D=13.5 $\pm$ 0.7 Mpc).

The excellent agreement between the distance determination in the innermost region (R $<$ Re, panel a)) and in a more external annulus (1Re $<$ R $<$ 2Re, panel c)), strongly suggests that the contamination by blends (if any) is marginally affecting our distance estimation. Panel g) of Fig. 1 presents the CMD of a Control Field with an area of 50$\times$50 arcsec2 located at the coordinates ($R.A.$,$Dec$) = (02:39:19.2; -08:08:46.7). These coordinates correspond to the pixel position (500,1500) of chip 2 of the adopted reference image. The center of the Control Field is about 7.5 Re away from the center of NGC1052-DF4. The area of the Control Field (2500 arcsec2) is chosen to represent an area similar to the one used for exploring the CMD of the galaxy (855 arcsec2 within 1 Re and 2565 arcsec2 between 1 and 2 Re). The negligible, if any, presence of stars belonging to NGC1052-DF4 in the location of the Control Field results in a completely different distribution of sources in its CMD, as well as in the LF peaking at significantly fainter magnitude (panel h), red line).

To further support that the detection of the TRGB is not being significantly affected by noise peaks or possible blends, we have conducted the following diagnostic. We have explored the number of detected stars above and below the TRGB as a function of the radial distance. This is done in Fig. 3. As we are using only chip 2 in our analysis, to fill the distribution of stars below -0.9Re we have mirrored the stars above +0.9Re. We have divided the detected stars in two groups: AGBs (stars that are brighter than the TRGB) and RGBs (stars that are fainter than the TRGB). Both groups expand 0.5 mag above and below the magnitude of the TRGB. The upper panels of Fig. 3 show the spatial location of the AGBs and RGBs around the central position of the galaxy. The lower panels of such figure show the number density of the detected stars as a function of radius. The figure clearly indicates that both AGBs and RGBs tend to cluster around the central part of the galaxy. More importantly, the number density of the stars declines following exactly the same trend as the surface brightness distribution of the object. In Fig. 3 we draw both the number density of stars as observed directly (open symbols) and corrected by the incompleteness of the data (filled symbols). Incompleteness simulations were done by injecting a large number of artificial stars (200000) of different magnitude across the image. For comparison, we overplotted the surface brightness distribution obtained on the F814W using the IRAF ELLIPSE routine (green dotted points) and the Sérsic model fitting solution to the surface brightness distribution obtained by Cohen et al. (2018) (red line). As it can be seen the agreement is excellent.

4 Distances to the brightest galaxies in the field-of-view of NGC1052

Unfortunately, none of the brightest galaxies in the field-of-view of NGC1052 (i.e. R$<$30 arcmin centered on this object) has the distance estimated using the TRGB method. This implies that their distances have been measured using less reliable techniques as the SBF or the Tully-Fisher relation.

4.0.1 The distance to NGC1042

The distance to NGC1042 has been measured using different methods. On one hand, there have been attempts to measure the distance using the reconstruction of the Large Scale Structure along the line of sight of the object. The most accurate measurement, which made a full model of the peculiar velocity to the Galaxy within $80\;\mathrm{Mpc}$ and also corrects the peculiar velocity of NGC 1042 provides a distance of 13.2 Mpc (parameter d${}_{k}^{c}$ in Table 5 of Theureau et al., 2007). On the other hand, the distance to the galaxy has been explored by mean of the Tully-Fisher relation. This last approach is the one producing a larger scatter on the values to the system, ranging from $\sim$4 (Tully et al., 2008) to $\sim$8 (Tully et al., 1992) Mpc. The main reason for this uncertainty is the lack of an accurate estimation of the inclination of the galaxy. In fact, values as low as 37$\degree$ have been measured (Luo et al., 2016, based on 2MASS photometry) but also larger values have been also provided (57$\degree$; Tully et al., 2008). The reason for such discrepancies is that the axis ratio and position angle of the galaxy change along the radial distance due to the presence of a bar and the spiral arms of the object (see Fig. 4). To overcome this difficulty and have an accurate estimation of the inclination of NGC1042, it is necessary to measure the ellipticity of the isophotes of the galaxy in its outer parts, a region where the internal structure of the galaxy is not playing any role.

In order to provide an accurate estimation of the Tully-Fisher distance to this object we have measured the inclination of the galaxy at its periphery (155\arcsec$$<R$<$195$\arcsec$). At such a radial distance the galaxy is not affected by the internal complex structure. In fact, both the axis ratio and the position angle remain pretty stable at such location (see Fig. 4). We have followed the recipe provided by Tully & Courtois (2012) to get the distance to the object. This recipe requires three parameters: a) a measure of the rotation of the galaxy by using the 21 cm HI spectral line. For that, we use the value provided by van Gorkom et al. (1986) in their Fig. 14 (i.e. Vmax-Vmin$\sim$145 ($\sin i$)-1 km s*-1*); b) the apparent luminosity of the galaxy given in the I band; and c) the inclination of the galaxy obtained from the ellipticity of the photometric image.

As a proxy for the I band, we use the SDSS i-band. We have retrieved the g, r and i-band from the SDSS survey. This data is deep enough to provide accurate photometry down to 26.5 mag/arcsec2 (r-band; Pohlen & Trujillo, 2006). We use the axis ratio of the isophotes in the i-band to estimate the axis ratio of the galaxy. In particular, we use the average value of the axis ratio of the isophotes in the radial range 155\arcsec$$<R$<$195$\arcsec$. The mean axis ratio is 0.832$\pm$0.004. In Fig. 4, we plot the shape of an elliptical contour with such axis ratio at a radial distance of 190$\arcsec$. The contour represents very well the outer part of the object. Once we have the axis ratio, we estimate the inclination of the galaxy using the following expression $\cos i=[((b/a)^{2}-q_{0}^{2})/(1-q_{0}^{2})]^{1/2}$ with q0=0.2 (Tully & Courtois, 2012). This results in $\cos i$=0.824 (i.e. i=34.4$\degree$).

In order to measure the I band magnitude of the galaxy, we apply the following transformation from the SDSS photometry to the Cousin system: I=r-1.2444$\times$(r-i)-0.3820 (http://classic.sdss.org/dr5/algorithms/sdssUBVRITransform.html#vega_sun_colors; Lupton 2005). The SDSS apparent (AB) magnitudes we measure correspond to the integrated magnitudes down to the i-band 26 mag/arcsec2 isophote. These are: 11.24$\pm$0.05 mag (g-band), 10.80$\pm$0.05 mag (r-band), 10.48$\pm$0.05 mag (i-band). This results in I=10.02$\pm$0.07 mag. We can compare this value with the one provided for this galaxy in the Extragalactic Distance Database (EDD) catalogue (I=10.19 mag; column 23 in Table 1 of Tully et al., 2008). Using the following Galactic extinction correction for the SDSS filters (0.065 and 0.049 mag; r and i-bands respectively), the observed Vega I band magnitude after correcting the Galactic extinction is 9.97$\pm$0.07 mag. To correct for the effect of the internal extinction of NGC1042 we use the following expression: A${}_{i}^{{}^{\prime}}$=\gamma_{I}$$\log(a/b) where a/b is the major to minor axis ratio and $\gamma_{I}$ is given by Tully & Courtois (2012):

[TABLE]

obtaining $\gamma_{I}$=0.725, so A${}_{i}^{{}^{\prime}}$=0.068 mag. The result of applying the internal and Galactic extinction correction is I=9.91 mag for NGC1042.

Finally, we measure the distance modulus to NGC1042 using the following expression (Tully & Courtois, 2012):

[TABLE]

which results in $\mu$=30.24$\pm$0.41 and a distance of 12.6$\pm$2.3 Mpc. Using the I magnitude provided in the EDD catalogue would give 13.5$\pm$2.6 Mpc. In order to have the galaxy located at 19 Mpc, the inclination of the object would need to be as low as 26.7 deg (b/a=0.898)666Using the EDD I-band magnitude, I=10.19 mag, the inclination would need to be 28.0 deg (b/a=0.888), again far away from the observations., which is incompatible with the shape of its external isophotes (see Fig. 4). Moreover, at all radial distances the inclination of the galaxies is not compatible with being close to a face-on projection as suggested by van Dokkum et al. (2019b). It is finally worth noting that the Tully-Fisher distance of NGC1042 is compatible, within the error bars, with the distance obtained from the Large Scale Structure reconstruction (13.2 Mpc; Theureau et al., 2007).

4.0.2 The distance to NGC1052 and NGC1047

The distance to NGC1052 has been estimated using the surface brightness fluctuations (SBF) technique both in the I band (19.4$\pm$2 Mpc; Tonry et al., 2001) and in the F160W band (18$\pm$2 Mpc; Jensen et al., 2003). The two results agree on a distance compatible with $\sim$19 Mpc777The Large Scale Structure reconstruction done by Theureau et al. (2007) gives also a compatible distance of 17.4 Mpc for this object.. Interestingly, NGC1052 is interacting with another galaxy in the field-of-view: NGC1047. This is clearly shown in the Dragonfly deep image of the group shown in Fig. 4 of van Dokkum et al. (2019a) and the deep image by the Heron telescope provided by Fig. 1 of Müller et al. (2019). This interaction is also very likely responsible for the perturbed distribution of HI around NGC1052 found in van Gorkom et al. (1986). In fact, this idea fits very well with the tail of HI gas of NGC1052 directly pointing to NGC1047 (see Fig. 5 of van Gorkom et al., 1986). It is worth noting that distance quoted in the literature for NGC1047 of 4.6 Mpc (Bottinelli et al., 1984) is clearly incompatible with the ongoing interaction between this object and NGC1052.

.

4.0.3 The distance to NGC1035

Another bright galaxy in the FOV of NGC1052 is the spiral NGC1035. This galaxy is particularly interesting due to its vicinity to NGC1052-DF4. The most recent Tully-Fisher distance determination for this galaxy has been obtained by Sorce et al. (2014). These authors quote a selection-bias-corrected distance estimate of 14$\pm$2.9 Mpc for NGC1035. This makes the galaxy distance compatible with NGC1042.

Finally, there is another bright galaxy located at 28.1$\arcmin$ from NGC1052: NGC1069. This object, however, has a heliocentric radial velocity of $\sim$9400 km s*-1* which makes it incompatible with being physically associated either to NGC1042 or NGC1052. The distance estimations for the galaxies in the field-of-view of NGC1052 are summarised in Table 1 and Fig. 5.

5 Discussion

The results shown in this work reinforce the idea that the galaxies in the line-of-sight of NGC1052 are grouped into two structures. We argue that the closest structure is formed by NGC1042 and NGC1035 and contains the dwarf galaxies: [KKS2000 04] and NGC1052-DF4. This group of galaxies is located at $\sim$13.5 Mpc. In projection, we then have a second group, whose principal galaxy is NGC1052 and also contains NGC1047. This group is located at $\sim$19 Mpc. Despite being placed at different distances, the two groups share similar heliocentric velocities. When these velocities are transformed to the CMB rest-frame and therefore a comparison with the Hubble Flow can be done, we find the following. The average velocity of the two most massive galaxies located at $\sim$13.5 Mpc (NGC1042 and NGC1035) is 1100 km/s (i.e. the velocity of this group deviates from the expected Hubble Flow at that distance by $\sim$+150 km/s), while the average velocity of the two massive galaxies located at 19 Mpc (NGC1047 and NGC1052) is 1233 km/s (i.e. the departure from the Hubble Flow is $\sim$-100 km/s). In both cases, the absolute deviation from the Hubble Flow is very similar and not very large. We refer the reader to Trujillo et al. (2019) for an in-depth analysis of the largest discrepant galaxy [KKS2000 04]. In short, the values of the velocities of the galaxies in these two groups are not at all unexpected considering the typical peculiar velocities, in comparison to the Hubble flow, observed among the nearby galaxies (see a discussion about this in Section 5 of Trujillo et al., 2019). A distance of 13.5 Mpc would also make the globular clusters of DF4 to have a less extreme nature. Their intrinsic brightness would be fainter and their sizes would be smaller, going from a median size of 4.1 pc (van Dokkum et al., 2019a) to 2.8 pc, which is in better agreement with the median size of both globular clusters in the Milky Way (2.9${}^{+0.1}_{-0.2}$ pc) and dwarf galaxies (3.2${}^{+0.1}_{-0.2}$ pc) (see e.g. Trujillo et al., 2019).

Finally, it is worth mentioning how the effect of the distance impact on the claim of the dark matter content of NGC1052-DF4. van Dokkum et al. (2019a) infer a distance for this galaxy of 20 Mpc. With the TRGB distance measured here, 14.2$\pm$0.7 Mpc, the dynamical mass of the galaxy (assuming an isothermal dark matter halo) decreases by a factor of 20/14.2 (i.e. $\sim$1.4) while its stellar mass decreases by a larger factor of 100.4×(31.505-30.761) (i.e. $\sim$2). van Dokkum et al. (2019a) also assumes a (M/L)stars,V=2.0$\pm$0.5$\Upsilon$⊙, while we find, for these types of diffuse galaxies, based on their Spectral Energy Distribution (from UV to IR) and assuming a Chabrier Initial Mass Function (IMF), a value of (M/L)stars,V=1.07${}^{+0.80}_{-0.54}$ $\Upsilon$⊙ (Trujillo et al., 2019), which decreases by another factor of two the stellar mass content888Ruiz-Lara et al. (2019) have recently measured spectroscopically the age and metallicity of [KKS2000 04]: 8.7 Gyr and [M/H]=-1.18. With these estimations and assuming a Chabrier IMF (M/L)stars,V=1.56 $\Upsilon_{\odot}$.. The combination of these two factors would reduce the total stellar mass of NGC1052-DF4 by a factor of $\sim$3, alleviating significantly the absence of dark matter previously reported. To end this discussion, it is also worth mentioning the strong relevance the assumption on the shape of the dark matter halo has for measuring the dark matter content of these galaxies. In Trujillo et al. (2019), we showed that the total dark matter halo that can accommodate the dynamics of KKS2000 04 changes by a factor of $\sim$3 from 109.1 M*⊙* (using a NFW halo) to 109.6 M*⊙* (using a Burkert cored halo with core radius of 4 kpc). In short, the larger the core of the dark matter halo the larger is the amount of dark matter compatible with the observed dynamics. There is growing evidence (both theoretical and observational) that the shape of the dark matter halo of the ultra diffuse galaxies is probably dominated by an extended dark matter core (see for instance the works by Di Cintio et al., 2014; van Dokkum et al., 2019c). All in all, the proposition that both KKS2000 04 and NGC1052-DF4 galaxies are “missing dark matter” is still far from being placed on sure footing.

We thank the referee for his/her careful reading of the manuscript which help to improve the quality of the analysis presented in this work. We thank Nushkia Chamba for her help on the analysis of the properties of NGC1042. Mireia Montes, Javier Román and Mike Beasley provided a lot of interesting comments. I.T. acknowledges financial support from the European Union’s Horizon 2020 research and innovation programme under Marie Sklodowska-Curie grant agreement No 721463 to the SUNDIAL ITN network. We also acknowledge support from the Fundación BBVA under its 2017 programme of assistance to scientific research groups, for the project ”Using machine-learning techniques to drag galaxies from the noise in deep imaging”. This research has been partly supported by the Spanish Ministry of Economy and Competitiveness (MINECO) under grants AYA2016-77237-C3-1-P and AYA2017-89076-P.