AGC 226178 and NGVS 3543: Two Deceptive Dwarfs toward Virgo

As part of program 15183 (PI: D. Sand), AGC 226178 was targeted with ACS in the F606W and F814W filters. The total exposure times were 2120 and 2180 s for the two filters, respectively. A combined F606W and F814W false-color image of the system is shown in Figure 1 (left). The standard ACS tools in DOLPHOT (Dolphin 2000, 2016) were used to align the individual exposures and perform photometry on all pointlike sources. ²⁵ To select stars from the resulting DOLPHOT catalog, we selected all type 1 and 2 (pointlike) objects with magnitude uncertainties of less than 0.3 (in both filters) and no photometry flags. We set a crowding limit of 1 mag in the two filters combined and enforced a combined absolute sharpness value of less than $\sqrt{0.075}$ and a roundness threshold of less than 1 in both filters. Galactic extinction corrections (Schlafly & Finkbeiner 2011) were made based on the dust maps of Schlegel et al. (1998) at the position of each star. The typical E(B – V) value was 0.024 mag for both objects.

The completeness limit of each combined field was estimated using artificial star tests in DOLPHOT. We generated 2 × 10⁵ artificial stars with F606W magnitudes ranging from 21 to 30 and F606W – F814W colors in the range −1 < (F606W – F814W) <2. These were randomly placed over the image and extracted as if real stars. They were then split into color bins, and the recovery fraction as a function of F814W magnitude was fit with an error function. The measured limits (from the error function shape parameters) for 50% and 90% recovery fractions were then fit with the combination of a horizontal line and a one-sided parabola (e.g., Figure 1, middle panels). At (F606W − F814W) = 1 mag, the F814W 90% and 50% completeness limits are 26.4 and 26.9 mag, respectively.

Object AGC 226178 was observed previously as part of the ALFALFA "Almost Darks" sample (Cannon et al. 2015) and rereduced for this work. Standard calibration and reduction methods were applied using the Common Astronomy Software Applications (CASA) package (McMullin et al. 2007). However, this rereduction used an improved automatic masking of sources during the tclean task. A full description of the data reduction pipeline will be presented in Inoue et al. (in preparation). These data have a channel width of 7.81 kHz (∼1.65 km s⁻¹) and a total bandwidth of 8 MHz. The final imaging used Briggs's robust = 0.5 weighting to provide a compromise between sensitivity and angular resolution for the detected H i emission. The resulting beam size was 56'' × 45''. During imaging, the channels were averaged and rebinned to a velocity resolution of 5 km s⁻¹. The resulting rms noise in 5 km s⁻¹ channels is 1.2 mJy beam⁻¹. The source mask for AGC 226178 was generated with the Source Finding Application (SoFiA; Serra et al. 2014, 2015). SoFiA's main algorithm adds pixels to a mask based on a signal-to-noise ratio (S/N) threshold after smoothing to various resolutions both spatially and in velocity. We applied a 3.5σ threshold after smoothing with spatial Gaussian kernels with widths of approximately one and two times the beam diameter and a box kernel over 20 and 40 km s⁻¹ (four and eight channels). A 90% reliability threshold was applied, which SoFiA estimates based on the apparent flux of negative (presumably spurious) sources. The resulting moment zero map is shown overlaid on a Dark Energy Camera Legacy Survey (DECaLS; Dey et al. 2019) g-band image in Figure 2 (top left).

Zoom In Zoom Out Reset image size

Panoramic, integral-field, intermediate-resolution (R = 2000–4000) spectroscopy in the wavelength range 4650–9300 Å of an $\simeq 1\buildrel{\,\prime}\over{.} 0\times 1\buildrel{\,\prime}\over{.} 0$ field centered on AGC 226178 was acquired with MUSE@VLT (Bacon et al. 2014) as part of observing program 0101.B-0376A (PI: R. Muñoz). Here we provide only an essential description of the process of data reduction and analysis following the procedures of Beccari et al. (2017); a detailed description of these procedures will be provided in a dedicated paper (M. Bellazzini et al. 2022, in preparation). Six ${t}_{\exp }=966\,{\rm{s}}$ exposures were acquired with a dithering scheme based on regular derotator offsets to improve flat-fielding and homogeneity of the image quality across the field. The raw data were wavelength and flux calibrated and combined into a single stacked data cube. Then we searched for individual sources with SExtractor (Bertin & Arnouts 1996) as peaks standing ≥3.0σ above the background in an Hα and white-light image ²⁶ obtained from the stacked cube by integration in wavelength (over 4650–9300 Å). Next, we measured the aperture fluxes of the detected sources (with SExtractor, aperture radius = 15) in each slice of the cube with a wavelength step of 1.25 Å. Ultimately, the fluxes of the individual slices were recombined into a 1D spectrum for each source. Visual inspection of all of the extracted spectra lead us to identify 15 sources (most of which are resolved or marginally resolved) with at least Hα in emission in the range of velocity spanned by galaxies in Virgo (−500 < cz_⊙ km s⁻¹ < 3000; e.g., Mei et al. 2007). These sources are all coincident with the various components of AGC 226178 shown in Figure 1 (left), and all have a very similar recession velocity, with a mean 〈cz_⊙〉 = 1584 km s⁻¹ and a standard deviation of σ = 4 km s⁻¹.

The ALFALFA (Giovanelli et al. 2005) observations were taken with the 305 m Arecibo telescope in Puerto Rico between 2005 and 2011. The survey adopted a two-pass drift-scan strategy, giving a total effective integration time of 48 s at any given point within the survey footprint. Radio frequency interference was semimanually identified and flagged before the drift scans were combined into cubes spanning 24 × 24 on the sky, with 1' pixels and an angular resolution of approximately 4'. The cube containing AGC 226178 spans approximately −2000 < cz_⊙ km s⁻¹ < 3000 with a channel width of ∼5 km s⁻¹ and velocity resolution of 10 km s⁻¹ (the full survey redshift range of −2000 < cz_⊙ km s⁻¹ < 18,000 was split into four overlapping subranges). The cube has an rms noise of 2.4 mJy channel⁻¹ and a beam size of 38 × 35. Further details regarding the ALFALFA survey and its data products can be found in Haynes et al. (2011, 2018). The original detection of the H i emission of AGC 226178 was at an S/N of 10.3 and cz_⊙ = 1581 km s⁻¹.

Figure 1 shows the HST image of the AGC 226178 and NGVS 3543 system and the CMDs of the two sources. The AGC 226178 CMD contains almost exclusively blue stars, while that of NGVS 3543 is dominated by red stars. Object NGVS 3543 has a fairly smooth elliptical morphology, partially resolved into stars in the HST image, whereas AGC 226178 is decidedly clumpy and appears to be broken into multiple components. The regions (marked by green dashed lines) used to produce the CMD of AGC 226178 were manually constructed after comparison of the HST images and Hα detections in MUSE (Section 3.4). In the case of NGVS 3543 (cyan dashed–dotted ellipse), the half-light radius and axial ratio from Junais et al. (2021) were used (r_e = 2238, i = 301, PA = 617) to produce the CMD for all stars within 2r_e (note that the ellipse plotted only extends to 1r_e).

In addition to the AGC 226178 CMD, in the bottom right panel of Figure 1, we overplot PARSEC (Bressan et al. 2012) stellar isochrones for a range of ages. ²⁷ All isochrones have a metallicity of [M/H] = −0.39 (Section 3.4) and the assumed distance to Virgo (16.5 Mpc). We note that, as isochrones do not depend on the initial mass function (IMF), comparing the CMD to isochrones is robust against an atypical shape of the IMF, which is a possibility for an unusual object such as AGC 226178.

The CMD is complex but reveals a young stellar population that is similar to the HST CMD of SECCO 1 (Sand et al. 2017; Beccari et al. 2017). There is a population of faint blue stars (F814W ≳ 24.5, F606W – F814W ≲ 0) that are likely a combination of young main-sequence stars and slightly older blue helium-burning stars. There is also a clear sequence of stars with 23.5 ≲ F814W ≲ 26.5 and F606W – F814W ≳ 0.6 mag that are likely red helium-burning (RHeB) stars. There are no evident red giant branch (RGB) stars, which is to be expected if AGC 226178 is at the distance of the Virgo cluster, as the tip of the RGB (TRGB) occurs at F814W ≈ 27 at 16.5 Mpc (e.g., Rizzi et al. 2007; Jang & Lee 2017). The brightness of the RHeB stars can be used to estimate the stellar population age (e.g., McQuinn et al. 2011), with the brightest star corresponding to an age of ∼10 Myr (at F814W ≈ 22.5 mag), while the faintest corresponds to an age of ∼50 Myr. Older ages are difficult to confirm because the helium-burning branch becomes incomplete at fainter magnitudes. The age spread seen in the CMD is thus similar to that inferred from the GALEX imaging and H ii region spectroscopy in the MUSE data (Sections 3.2 and 3.4).

The CMD of NGVS 3543 is discussed in detail in Section 4.3.

Object AGC 226178 is strongly detected in both the near-UV (NUV) and far-UV (FUV) bands (Figure 2, top right) in GALEX. The total flux in each band was measured from background-subtracted GALEX tiles (GI2_125026_AGESstrip2_04 in both NUV and FUV). They were converted to magnitudes following the standard GALEX conversions (Morrissey et al. 2007) and a Galactic extinction correction based on the E(B – V) values of Schlafly & Finkbeiner (2011) at the location of each source and R_NUV = 8.20 and R_FUV = 8.24 (Wyder et al. 2007). Assuming a distance of 16.5 Mpc for all candidates and a bolometric solar absolute magnitude of 4.74, the apparent magnitudes were converted to luminosities and, finally, to star formation rates (SFRs) following Iglesias-Páramo et al. (2006). Uncertainties in the UV fluxes and SFRs were estimated by first masking the brightest 1% of pixels in the background-subtracted GALEX tiles (to remove bright sources) and then randomly placing 10⁴ circular apertures (of equal area to the aperture in question) across the entire tile in order to estimate the rms noise.

This gives the NUV flux in the combined regions of AGC 226178 as (1.74 ± 0.07) × 10⁻¹⁶ erg s⁻¹ cm⁻² Å⁻¹ and the FUV flux as (4.02 ± 0.09) × 10⁻¹⁶ erg s⁻¹ cm⁻² Å⁻¹. These equate to SFR estimates of $\mathrm{log}\tfrac{{\mathrm{SFR}}_{\mathrm{NUV}}}{{{\rm{M}}}_{\odot }\,{{\rm{yr}}}^{-1}}=-3.03\pm 0.06$ and $\mathrm{log}\tfrac{{\mathrm{SFR}}_{\mathrm{FUV}}}{{{\rm{M}}}_{\odot }\,{{\rm{yr}}}^{-1}}=-3.18\pm 0.06$, respectively. The fact that these two estimates are so similar suggests that the SFR in AGC 226178 has been relatively constant over the past 100 Myr.

As pointed out by Junais et al. (2021), NGVS 3543 is weakly detected in GALEX (both NUV and FUV). Making equivalent measurements (to those above) for NGVS 3543, we find an only slightly lower UV flux ((1.18 ± 0.19) × 10⁻¹⁶ erg s⁻¹ cm⁻² Å⁻¹ in NUV and (1.47 ± 0.24) × 10⁻¹⁶ erg s⁻¹ cm⁻² Å⁻¹ in FUV), though spread over a much larger area. This UV detection (particularly in FUV) would normally imply that SF has not completely ceased, as was suggested by Junais et al. (2021). However, the lack of young blue stars in the CMD (Figure 1, top middle) does not appear to agree with this interpretation. Another possibility is that blue horizontal branch stars below the completeness limit are contributing to the UV flux of this object (analogous to Yoon et al. 2004; Goudfrooij 2018). Such stars are known to exist in the old stellar populations of Local Group dwarf spheroidals (e.g., Monaco et al. 2003; Martin et al. 2017) and globular clusters (e.g., Perina et al. 2012; Dalessandro et al. 2012). However, for NGVS 3543, the color FUV − NUV = 0.58 ± 0.25, which is slightly bluer than for any of the examples above.

Figure 2 (top left) shows the VLA H i emission contours of AGC 226178 overlaid on a DECaLS g-band image. This map was constructed with SoFiA and extends to lower gas column densities than the equivalent map (based on the same observations) in Cannon et al. (2015). This deeper map reveals an extension that appears to point toward the pair of VCC galaxies, VCC 2037 (cz_⊙ = 1142 km s⁻¹) and 2034 (cz_⊙ = 1507 km s⁻¹), to the SW. However, no connection is evident in the VLA data (Figure 2, bottom left), even though both of these galaxies were detected in the VLA observations of AGC 226178 (about 14' from the pointing center, where the primary beam response is at about 55%).

The total H i flux of AGC 226178 in the VLA data is 0.3 dex lower than in ALFALFA (0.31 Jy km s⁻¹ compared to 0.62 Jy km s⁻¹), strongly suggesting that the VLA has missed extended or low column density emission (this was also noted as a possibility by Cannon et al. 2015). The ALFALFA H i emission is centered at cz_⊙ = 1581 km s⁻¹, which in this direction is consistent with the membership of Virgo (e.g., Masters 2005). Assuming a distance of 16.5 Mpc, the larger of these fluxes equates to a total H i mass of $\mathrm{log}{M}_{{\rm{H\; I}}}/{{\rm{M}}}_{\odot }=7.6$. Junais et al. (2021) estimated the stellar mass of AGC 266178 as ∼5 × 10⁴ M_⊙, which means the ratio of gas to stellar mass is ∼1000 (after multiplying the H i mass by 1.4 to account for helium). This value is exceptional, even for low-mass, gas-rich galaxies (Huang et al. 2012), indicating that AGC 226178 is not a normal low-mass galaxy.

Motivated in part by the gas inferred to be missing from the VLA observations, we returned to the original ALFALFA cube and constructed a moment zero map with SoFiA using smoothing kernels of one and two times the beam size, 0, 15, and 30 km s⁻¹, and a threshold of 3.5σ (a reliable threshold for studying extended H i streams; e.g., Taylor et al. 2020). The ALFALFA map (Figure 2, bottom right) shows an apparent H i connection between VCC 2034 and AGC 226178, resolving the question of where the latter's gas likely originated. We note here that the moment map excludes the velocity range over which there is emission from VCC 2037, so although some contours overlap with it in projection, all of the emission is actually associated with VCC 2034 and AGC 226178.

The ALFALFA data have much worse angular resolution than the VLA data; however, they also have significantly better column density sensitivity to extended emission. The majority of the additional emission in the bottom right versus bottom left panel of Figure 2 is simply below the column density sensitivity of the VLA data and would require a prohibitively long integration time to detect with the VLA (at this resolution).

To attempt to recover some of this emission in the VLA data, the data were reimaged after applying a uv taper of 750λ, thereby degrading the resolution of the data to about 25 but improving the 3σ column density sensitivity to 3.2 × 10¹⁸ cm⁻² (0.025 M_⊙ pc⁻²) over 20 km s⁻¹, approaching the level of the ALFALFA data. In Figure 3, we show position–velocity slices for both the ALFALFA data and the tapered VLA data, covering the space between AGC 226178 and VCC 2034.

Zoom In Zoom Out Reset image size

The H i emission of AGC 226178 is seen as a clump in the upper left corner of the middle and right panels of Figure 3, and VCC 2034 is the clump in the lower right corner (of either panel); VCC 2037 is well outside the velocity range plotted. The additional emission shown in the ALFALFA moment zero map (Figure 2, bottom right) appears to be made up of very low S/N features that extend from both objects toward the other (Figure 3, middle). However, when summed together over several consecutive channels to make the moment zero map, these combine to form relatively high-confidence features. These features may have formed a continuous bridge in the past or may still do so below the sensitivity of the data. The brightest part of the extension from AGC 226178 is also visible in the VLA position–velocity slice (Figure 3, right), and there may be a feature extending to lower velocities, but it is at the level of the noise. Although the quoted column density sensitivities of the two data sets now only differ by ∼25%, the ALFA beam still covers a 2.5 times larger area. Therefore, if the emission were physically spread over an area larger than the VLA synthesized beam (∼25), then the effective sensitivity of the VLA data would drop significantly relative to the ALFALFA data. It is also worth noting that the two slices have different widths (see caption of Figure 3), again meaning that more extended emission could be missed by the VLA slice. Finally, the quoted column density sensitivity of the VLA data is for the pointing center (approximately the position of AGC 226178). At the location of VCC 2034, the sensitivity is approximately a factor of 2 worse.

As most of the features seen in the ALFALFA data cannot be conclusively corroborated or refuted, even with the tapered VLA data, we will primarily consider the ALFALFA data when discussing our interpretation of the system in Section 4. ²⁸

The MUSE observations of AGC 226178 find an abundance of clumps of Hα emission. Based on line ratio diagnostics (O iii/Hβ, N ii/Hα, and S ii/Hα), this emission can be confidently classified as SF regions (thresholds from Kewley et al. 2001; Kniazev et al. 2008). The total Hα flux measured by MUSE is 3.13 × 10⁻¹⁶ erg cm⁻¹ s⁻¹. This equates to a total SFR estimate of $\mathrm{log}\tfrac{{\mathrm{SFR}}_{{\rm{H}}\alpha }}{{{\rm{M}}}_{\odot }\,{{\rm{yr}}}^{-1}}=-3.09$ (converted as in Equation (2) of Kennicutt 1998), in close agreement with the UV SFR estimates. The mean heliocentric velocity of these H ii regions (cz_⊙ = 1584 km s⁻¹) also reconfirms that they are indeed associated with the H i emission.

Finally, the mean oxygen abundance was estimated by averaging over two different indicators (N2 and O3N2, adopting the calibration by Pettini & Pagel 2004) for the five sources for which it was possible to measure both indicators (as done by Bellazzini et al. 2018). These were corrected for extinction based on Hα/Hβ. The average value, 〈12 + logO/H〉 = 8.3 ± 0.1, supports the finding that the gas likely came from a more massive galaxy that pre-enriched it. ²⁹ A complete analysis of the abundance measurements will be presented in a separate paper focusing on the MUSE observations (Bellazzini et al. in preparation).

Based on the stellar mass–metallicity relation (MZR) of Andrews & Martini (2013), this metallicity should correspond to a galaxy of $\mathrm{log}{M}_{* }/{M}_{\odot }=8.4\pm 0.4$. Object VCC 2034 has an i-band magnitude of 14.77 ± 0.02 and g − i = 0.76 ± 0.03 (Kim et al. 2014). Using the scaling relation of Taylor et al. (2011) and assuming a distance of 16.5 Mpc gives $\mathrm{log}{M}_{* }/{M}_{\odot }=8.2\pm 0.1$ for VCC 2034, making it consistent with being the source of AGC 226178's gas.

Given its position and velocity, the preferred distance to AGC 226178 is 16.5 Mpc, our assumed distance to objects associated with the Virgo cluster. However, without a direct distance measurement to either AGC 226178 or VCC 2034 (the apparent source of its gas), there remains the possibility that neither is genuinely in Virgo, with the second most likely distance being ∼10 Mpc (the distance to VCC 2037, based on a TRGB measurement; Karachentsev et al. 2014).

Given the morphology of the H i emission, it is reasonable to assume that AGC 226178 and VCC 2034 are at the same distance, meaning that a distance estimate to either object would constrain both. We considered estimating a Tully–Fisher relation (TFR) distance to VCC 2034; however, given its low mass, apparent low inclination, and ongoing interaction, this approach would not be reliable. In the absence of a robust distance estimate, here we note several points in favor of assuming 16.5 Mpc as the distance to this system.

• 1.  
  The single strongest argument for D ≈ 16.5 Mpc is that the numerous H ii regions detected in the MUSE observations of AGC 226178 mean that it must contain very young stars (<10 Myr). When isochrones (assuming D = 16.5 Mpc) of various ages are overplotted on its CMD (Figure 1, bottom right), the brightest stars overlap with the 10 Myr isochrone. Moving AGC 226178 to 10 Mpc would shift this isochrone (and all others) approximately 1 mag brighter. Therefore, if AGC 226178 were at 10 Mpc, then stars of F814W ≈ 21.5 should be present in the CMD. However, there are none. As the completeness is essentially 100% at this magnitude (and the photometric errors are small), it would be difficult to explain the absence of these stars, whereas a distance of 16.5 Mpc naturally explains the observed population.
• 2.  
  Object VCC 2037 is known to be at ∼10 Mpc, and it is separated from VCC 2034 by approximately 350 km s⁻¹, which would be a large velocity offset if the two were part of the same foreground structure. If that were the case, then most likely, they would not be gravitationally bound to each other (as both are dwarf galaxies with M* ∼ 10⁸ M_⊙), and we would be seeing the system at a special time, right as they pass by each other.
• 3.  
  If VCC 2034 and VCC 2037 were at the same distance, it would be likely that they would be interacting and that this interaction would be visible in H i emission, especially given the abundance of loosely bound gas in the vicinity of both galaxies. However, in both the ALFALFA and VLA data cubes, there is no clear sign of a bridge or tails extending between the two galaxies.

In light of the points above, for the remainder of the discussion, we will assume that AGC 226178 is at 16.5 Mpc, in the Virgo cluster. However, at the relevant points, we will indicate how our interpretation might change if it were actually at 10 Mpc.

The morphology of the integrated H i emission seen in the ALFALFA data (Figure 2, bottom right) indicates that the gas in AGC 226178 originated in VCC 2034, about 70 kpc to the SW (at the distance of Virgo). There are two main mechanisms by which gas is stripped in this way: tidal stripping and ram pressure stripping. The former can occur in almost any scenario where two or more galaxies (at least one of which contains neutral gas) strongly interact; the latter is only active in a region with a sufficiently dense intergalactic medium, such as a galaxy cluster. If AGC 226178 and VCC 2034 are in the Virgo cluster, then both of these mechanisms must be considered.

The one-sided morphology of the H i tail initially suggests ram pressure stripping, as these tails trail in the wake of a galaxy as it falls through the ICM and are thus one-sided. However, these tails also typically (though not exclusively; e.g., Cramer et al. 2019) point approximately radially away from the cluster center (e.g., Chung et al. 2009), as galaxies usually fall toward the center. In this case, the center of the Virgo cluster is approximately NW, whereas the tail extends to the NE, almost perpendicular.

The tail is approximately 70 kpc long (in projection), and if we assume it is ∼200 Myr old (i.e., at least twice as old as the oldest stars we identified), then VCC 2034 would have to have a transverse velocity of ∼350 km s⁻¹ for the tail to have been formed by ram pressure stripping. The mean radial velocity of Virgo cluster galaxies is 1138 km s⁻¹ (Mei et al. 2007), and the radial velocity of VCC 2034 is 1507 km s⁻¹, which, in this scenario, would make its total velocity relative to the cluster center about ∼500 km s⁻¹. With this velocity, it would still be comfortably bound to the cluster, and the transverse velocity is small enough that it could have acquired this in the past, e.g., if it fell toward the cluster as part of a group. Therefore, ram pressure stripping is still a plausible scenario, despite the direction of the tail.

We also note that a visual inspection of the morphology of VCC 2034 in DECaLS and GALEX imaging does not appear to show a trail of stars accompanying the H i tail. This is the expected behavior of ram pressure stripping, as it acts only on the interstellar medium, whereas tidal stripping is gravitational and affects all matter equally. However, this is not conclusive, as H i gas is typically much more spatially extended (and therefore more loosely bound) than a galaxy's stellar component, meaning that H i can become heavily disturbed before any disruption of the stars is evident.

The one-sided form of the tail also does not rule out tidal stripping, as real examples of tidal stripping of H i seldom follow a simple morphology and are frequently quite one-sided (e.g., Hibbard et al. 2001). The main issue with interpreting this H i tail as a tidal feature is the lack of an obvious candidate for the perturber, which we discuss in detail below.

4.2.1. Which Galaxy Could Have Stripped Gas from VCC 2034?

Upon first inspection, the close proximity of AGC 226178 and NGVS 3543 might imply that the two objects are physically associated and at approximately the same distance. Using the CMD of NGVS 3543 (Figure 1, top middle), we attempted to measure the TRGB to obtain an accurate distance estimate. However, these measurements failed to reliably converge, likely owing to the paucity of stars and the proximity of the TRGB to the completeness limit. Instead, we turn to the observed luminosity function shown in Figure 4. This indicates that the TRGB likely occurs at F814W ≈ 26 (very close to the 90% completeness limit), which corresponds to a distance of ∼10 Mpc. The stars brighter than this in the CMD likely belong to an asymptotic giant branch (AGB) population (analogous to the case in Sand et al. 2014). We also find that the distribution of these candidate AGB stars is considerably more compact than that of the candidate RGB stars, suggesting that they may be a younger population. The CMD of NGVS 3543 can also be compared to another dwarf spheroidal known to be in the Virgo cluster, Dw J122147+132853 (Bellazzini et al. 2018, their Figure 6). The majority of the detected stars in this dwarf are fainter than F814W = 27, and there are almost no stars brighter than F814W = 26. This comparison reaffirms that NGVS 3543 is significantly nearer than the Virgo cluster.

Zoom In Zoom Out Reset image size

Our distance estimate of ∼10 Mpc contradicts Junais et al. (2021), who assumed that AGC 226178 and NGVS 3543 were at the same distance (in the Virgo cluster) and physically interacting due to a faint bridge of UV emission between them (their Figure 1). However, with the HST imaging (and MUSE observations), it is clear that there is a clump of blue stars projected between the main body of AGC 226178 and NGVS 3543 (green dashed circle near the center of the left panel of Figure 1), which likely explains this apparent bridge of UV emission. Given that AGC 226178 is highly clumpy and irregular, this configuration could easily have occurred by chance.

Junais et al. (2021) also estimated the mass of NGVS 3543 (without a ram pressure stripping event) as $\mathrm{log}{M}_{* }/{M}_{\odot }=7.31\pm 0.27$ (if it were in Virgo), which, if it follows the MZR of Kirby et al. (2013), would mean it has 〈[Fe/H]〉 = −1.3 ± 0.2. ³⁰ Assuming a solar ratio of Fe/O (Asplund et al. 2009), this equates to 12 + logO/H ≈ 7.4. This value is almost an order of magnitude lower than that of AGC 226178 (Section 3.4), reinforcing that it is unlikely that the gas in AGC 226178 originated in NGVS 3543 (as hypothesized by Junais et al. 2021). Moving NGVS 3543 to 10 Mpc would reduce the stellar mass estimate of Junais et al. (2021) to $\mathrm{log}{M}_{* }/{M}_{\odot }=6.88\pm 0.27$, its absolute magnitude to M_g = −12.5, and its metallicity estimate to 〈[Fe/H]〉 = −1.4 ± 0.2, further exacerbating the discrepancy.

At a distance of ∼10 Mpc, it is likely that NGVS 3543 is associated with VCC 2037, approximately 12' (36 kpc) to the SW, which has a prior TRGB distance estimate of 9.6 Mpc (Karachentsev et al. 2014). Although at 10 Mpc, NGVS 3543 would not quite be classified as a UDG (as its half-light radius would be <1.3 kpc), it is similar in appearance. Tidal stripping is a promising mechanism for forming such galaxies (e.g., Bennet et al. 2018; Carleton et al. 2019; Jiang et al. 2019; Liao et al. 2019; Jones et al. 2021); thus, its association with VCC 2037 might explain its appearance. Furthermore, it is possible that an interaction with VCC 2037 led to the recent quenching of this object that Junais et al. (2021) proposed. Alternatively, NGVS 3543 may simply be a field dwarf at a slightly different distance to VCC 2037. However, if this were the case, it would be an exceptionally odd system, as it has a diffuse morphology and UV emission but no detectable H i, and its CMD is dominated by red stars.

The position of the stars in the CMD of AGC 226178 (Figure 1, bottom right) indicates that there is likely some spread in their ages in the approximate range 10–100 Myr. This would also be consistent with the close agreement in the SFR estimates from NUV and FUV, suggesting that the SFR has been somewhat constant on the order of 100 Myr. Furthermore, a sustained SFR of ∼10⁻³ M_⊙ yr⁻¹ (Section 3.2) over 100 Myr would produce a total stellar mass of ∼10⁵ M_⊙, similar to the stellar mass estimate of Junais et al. (2021), 5 × 10⁴ M_⊙. At its present SFR, AGC 226178 could go on producing stars for several Hubble times without running out of gas. However, it is unlikely that it will be able to retain its gas on a long (>1 Gyr) timescale.

As it appears to have formed from stripped material, AGC 226178 is unlikely to contain a significant quantity of dark matter, and as its total baryonic mass is less than 10⁸ M_⊙, it does not meet the threshold typically assumed for a TD to be long-lived (Bournaud & Duc 2006). This means that AGC 226178 is unlikely to be self-gravitating, and even if it is, it is unlikely to remain so on a gigayear timescale. Therefore, it may not be appropriate to refer to it as a galaxy (Willman & Strader 2012), and we have tried to avoid doing so. However, unlike in field and group environments, where TDs are typically studied, it has been suggested that the hot ICM may actually assist low-mass, gas-rich objects in remaining bound longer than would be expected from their internal gravity alone, as this can add an additional external compression to the gas (Burkhart & Loeb 2016).

The fate of the H i component of a very similar object (SECCO 1) was investigated in detail by Bellazzini et al. (2018) and Calura et al. (2020). These works performed a series of high-resolution hydrodynamical simulations of a 10⁷ M_⊙ gas cloud moving at up to 400 km s⁻¹ through a hot medium, finding that a significant fraction of its initial H i content could persist for on the order of 1 Gyr. However, instabilities in the gas cloud build up as it moves through the hot ICM, and on longer timescales, it will be broken up into smaller clouds and evaporate. As H i makes up the vast majority of the baryonic mass of AGC 226178 and provides the medium for external pressure equilibrium, once the H i is lost, it will almost certainly be unbound.

Given its low stellar mass, AGC 226178 is only visible at all in the optical because of its young, bright, blue stars. Once SF ceases, it will rapidly fade. After 100 Myr, only a handful of stars will be above the 90% completeness limit of our HST observations, and after 500 Myr, almost no individual stars will be detectable at all. Assuming its total stellar mass has not dramatically increased in the intervening time (such that its integrated light would be detectable), it will become, for all intents and purposes, invisible (cf. Román et al. 2021), and its aging stars will become part of the intracluster light of Virgo.

As alluded to by Sand et al. (2017), SECCO 1 and AGC 226178 are not the only objects of their kind. Both were originally identified by their H i emission in the ALFALFA survey and were noteworthy due to their near-invisible (in SDSS images) stellar counterparts. As part of this work, AGC 226178 was reidentified (independent of its original H i detection) during a visual search for faint, blue, and UV-emitting objects in the Virgo cluster using ∼100 deg² of publicly available NGVS images and GALEX tiles. A full analysis of this search will be presented in a subsequent paper (Jones et al. in preparation), but at least three additional objects were identified with very similar optical and UV properties to AGC 226178. Assuming that the other objects are also in Virgo and extremely young (as AGC 226178 appears to be), it is reasonable to assume that such objects are being continually created in the Virgo cluster. At present, these ∼five objects are known, and, based on the discussion above, each one may only be visible for ∼500 Myr. Therefore, these objects must be being produced in Virgo at a rate of ∼one per 100 Myr. However, it remains to be seen whether these form a uniform population with related formation mechanisms or are really a mixture of many different types of objects that happen to have similar appearances in the optical and UV.