ALMA-JELLY I: High Resolution CO(2-1) Observations of Ongoing Ram Pressure Stripping in NGC 4858 Reveal Asymmetrical Gas Tail Formation and Fallback ¹¹1Released on XXXX

In this paper we present the first results from the ALMA-JELLY program. The goal of this program is to map CO(2-1) at 1” resolution in a selection of very strong RPS in nearby high-mass clusters. The ALMA JELLY galaxies reside in the Coma, Leo, and Norma clusters, with 18, 8, and 2 galaxies in each cluster respectively.

ALMA-JELLY comprises combined 12m and 7m data obtained during ALMA Cycle 8 ²²2PID 2021.1.01616.L for 27 of the galaxies, with 1 (ESO 137-001) obtained from archival data (See Jáchym et al., 2019). The galaxies were selected based on having a prominent one-sided H$\alpha$ ionised gas tail in three nearby rich clusters, Norma, Leo and Coma (Sun et al., 2010; Yagi et al., 2010; Fossati et al., 2012; Yagi et al., 2017; Gavazzi et al., 2018). The galaxy sample covers a broad range of stellar masses ($\sim 10^{7.5}-10^{10}M_{\odot}$). It also includes a wide range of evolutionary stage of stripping, from early stages with tails connected to star-forming disks, to late stages with detached tails, and passive or post-starburst disks. The complete sample and full description of the project will be presented in Jachym et al. (in prep.).

NGC 4858 is one of the best nearby examples of a galaxy with strong CO emission and a multi-component asymmetric molecular gas tail. This makes it an ideal galaxy to demonstrate the quality of the ALMA-JELLY data and the science made possible through this program.

The Coma cluster’s proximity ($\sim 100$ Mpc) and high mass $\left(M>10^{15}M_{\odot}\right)$ makes it an ideal environment for studying strong RPS events. The location of NGC 4858 in the cluster, and its tail direction (See Section 4.1 for more details) is indicated in Figure 1.

NGC 4858 has one of the smallest projected distances to the cluster centre out of the ALMA-JELLY sample, at 355 kpc ($\sim$0.14 $r_{vir}$) (Lokas & Mamon, 2003). The galaxy is highly redshifted with respect to the cluster mean velocity, with a relative velocity of $v_{rad}=+2421$ km s^-1. The combination of small projected distance and high line-of-sight velocity puts NGC 4858 at the upper envelope of galaxies in the Coma cluster phase-space diagram (See Figure 11 of Boselli et al., 2022), suggesting it has recently entered the cluster. Because the galaxy is redshifted with respect to the Coma cluster mean, the downstream direction of the tail is pointed towards the observer. We estimate the galaxy’s disk-wind angle (the angle between the vector normal to the galaxy disk plane and the wind vector) to be $\phi_{\text{DW}}=75_{-27}^{+10}$ in Section 7.1.

NGC 4858 has an HI gas mass of $2.5\times 10^{8}$M_⊙ (Molnár et al., 2022). This corresponds to an HI deficiency factor of Def${}_{HI}=0.94$, suggesting that NGC 4858 is missing more than 90% of its expected HI gas (Parkash et al., 2018). NGC 4858 is overall blue in color, indicating significant recent star formation. Gallego et al. (1996) identify NGC 4858 as a bright H$\alpha$ emitting galaxy. Yagi et al. (2010) discovered the presence of a large H$\alpha$ gas tail extending north from the galaxy disk, indicating some combination of shock-excited gas and/or star-formation in the tail. The distribution of H$\alpha$ is highly asymmetric, with more emission on the northwest side of the galaxy, which is the trailing side of the galaxy rotating into the wind.

NGC 4858’s disk and tail are bright in NUV/FUV emission (Smith et al., 2010). The alignment of UV and H$\alpha$ emission, combined with the presence of bright H$\alpha$ clumps in the tail, suggests that a significant fraction of the H$\alpha$ is coming from HII regions around newly-formed stars. This is further confirmed by the HST F350LP/F600LP image of the galaxy (see Figure 2) which shows clumps of blue stars throughout the tail. The presence of stars in the tail makes NGC 4858 a “jellyfish galaxy”.

The galaxy has an X-ray tail observed in 0.5-2 keV with Chandra, indicating the presence of hot gas in the tail (See Sun et al., 2021, supplementary Figure 11). The galaxy also has a radio continuum tail that extends approximately 30-40 kpc north of the galaxy disk, first revealed in Chen et al. (2020) and in Roberts et al. (2021, 2024). The large physical extent of this tail provides us with the most reliable estimate for the on-sky tail angle for the galaxy, which we estimate as $\theta_{\text{tail}}=-15\pm 20\degree$ measured counterclockwise from north (See Section 7.1).

The elliptical galaxy NGC 4860 (GMP 3792) is observed nearby. Although these galaxies appear close in projection ($\sim 40"$), their high relative velocities (approximately 1500 km s^-1) along the line of sight, combined with no evidence for tidal features in either galaxy, suggests that they are not interacting.

The Coma cluster has a redshift of $z=0.0235$ and a velocity dispersion of $\sim 950$ km s^-1(Sohn et al., 2017). We will use the standard cosmological parameters from Hinshaw et al. (2013) ($H_{0}=69.7$ km s^-1 Mpc^-1). We adopt a luminosity distance for Coma of 100 Mpc, corresponding to a scale of 1” = 0.463 kpc. NGC 4858 has an estimated stellar mass of $4.9\times 10^{9}M_{\odot}$, making it an intermediate mass spiral galaxy (Molnár et al., 2022).

We describe in Section 2 the new ALMA CO data and ancillary datasets. After analyzing the stellar body (Section 3), we explore the CO distribution and compare to other tracers (Section 4). We then explore the global CO kinematics (Section 5) and describe individual regions with highly non-circular motions (Section 6). In Section 7 we discuss the theory behind the development of asymmetries in the inner RPS tail. and compare the observational data to simulations. We summarize the results in Section 8. We include a detailed discussion of deriving the disk-wind angle in Appendix A, and describe in detail the possible origins for observed velocity residuals in Appendix B.

NGC 4858 was observed by ALMA with 5 mosaic pointings using the 7m array, and 13 mosaic pointings with the 12m array. ALMA observed the 230.5 GHz CO(2-1)line, and any subsequent mention of CO emission implies the CO(2-1) transition unless stated otherwise.

The ALMA data were cleaned using the PHANGS-ALMA data reduction pipeline (Leroy et al., 2021), which we will briefly describe. Instead of calling tclean once to clean a dataset, as is the standard procedure, the PHANGS-ALMA pipeline differs by calling tclean multiple times over multiple scales, while supplying masks to guide deconvolution. The data is also corrected for the primary beam, and then automatically trimmed to an appropriate image size. We choose to use a natural weighting scheme (Briggs robust=2) to give us the best surface brightness sensitivity.

The datacube has 480 x 303 pixels (corresponding to $72\times 45$ arcseconds, or $34\times 21$ kpc, respectively), and channel widths of 3.8 km s^-1. The synthesized primary beam has major and minor axes of 1.35 and 0.87 arcseconds, respectively, and a position angle of $30\degree$. The RMS of the cube is measured at $\sim 2.3$ mJy Beam^-1.

Moment maps (CO emission, mean velocity, and velocity dispersion) from the masked datacube are shown in Figure 4. We generate masks for the cleaned datacube using the Python maskmoment package, developed by Tony Wong ⁴⁴4See https://github.com/tonywong94/maskmoment. This maximizes the amount of geniune emission included in the masked cube. The weakest features are required to be detected at $>3.0\sigma$ in at least 2 contiguous channels, and must cover a minimum spatial area as large as the synthesized beam.

From the moment-0 map, (Figure 4), we measure an integrated CO flux (disk + tail) of $S_{CO}\Delta v=46.2\pm 2.7$ Jy km s^-1. We can then follow the procedure in Solomon et al. (1997) to obtain a CO line luminosity (in units of brightness temperature) using

This ratio is high, given that field galaxies at the stellar mass of NGC 4858 have typical ratios of $M_{\text{H}_{2}}/M_{*}\approx 0.08$ (Popping et al., 2014). The large H₂ mass fraction is likely due to a combination of an intrinsically high mass fraction, plus the adopted Milky Way value for CO/$M_{H_{2}}$ possibly being too high for RPS galaxies. Since the L_CO/M_∗ is high, the large HI deficiency factor in NGC 4858 might be caused in part by HI being converted to H₂ (Moretti et al., 2020).

Ground-based R-band and narrow-band (H$\alpha$ at the Coma cluster) images of NGC 4858 come from Yagi et al. (2010), to which we refer the reader for a more in-depth explanation of the data acquisition and processing (also Yagi et al., 2017). These images were obtained using the Suprime-Cam, located at the prime focus of the Subaru telescope (Miyazaki et al., 2002).

In the northeast region near the elliptical galaxy, continuum subtraction for the H$\alpha$ image is imperfect and not all continuum starlight is removed from the narrow-band filter. We indicate this region in Figure 3 (panels a and c). This is unlikely to affect analysis of the H$\alpha$ tail of NGC 4858, where the continuum is low and the distribution is not smooth.

For this project we also use publicly-available HST Wide-Field Camera 3 (WFC3) F350LP and F600LP images from the HST Legacy Archive (PI Gregg, Proposal ID: 13777), shown in Figure 2. The wavelength bands correspond to $3400$Å and $5800$Å rest-frame emission for NGC4858, respectively. The 350LP image best reveals both the distributions of young stars in the galaxy, as well as dust extinction. The data described here may be obtained from the MAST archive at https://dx.doi.org/10.17909/ts8f-7f96 (catalog doi:10.17909/ts8f-7f96).

The LOFAR 144 MHz radio data used in this work come from the Lofar Two-Meter Sky Survey (LoTSS) (See Shimwell et al., 2017). A cutout surrounding emission from NGC 4858 was created from the larger Coma cluster mosaic image (For a more detailed analysis of this region, see Roberts et al., 2021). From this cutout we estimate a total 144 MHz flux density of $5\times 10^{-4}$ Jy for NGC 4858.

We wish to measure asymmetries relative to the projected orbital path. We think that the outer tail is the best tracer of the orbital path, and that the inner tail is asymmetric due to rotational effects which are largest in the inner tail. For the tail angle we adopt the centroid of the LOFAR radio continuum tail, since that is the tail component measured furthest from the galaxy. Figure 6 measures the flux in slices across the tail, increasing in distance along our estimated wind direction. We find a position angle of of $\theta_{\text{wind}}=-15\degree$ equally divides the radio continuum flux at distances 18-27 kpc, and adopt this as the tail angle.

Furthermore, small dust filaments seen in the northwest region of the HST color image likely trace the local flow direction, which approximately aligns with the tail position angle measured from the radio continuum tail. There are some elongated dust and stellar features closer to the inner disk with different angles. However, such features do not always trace the large-scale wind angle, since the local flow direction close to the disk can be different due to various effects including rotation and shielding (Abramson & Kenney, 2014).

The combined effects of ram pressure and rotation are expected to produce gas distributions that are more complex than simple head-tail morphologies which are predicted for face-on ram pressure interactions. In particular, large asymmetries may occur in the inner tail.

To better understand the combined effects of ram pressure and rotation on NGC 4858, we analyze the CO spatial distribution in the galaxy outskirts as a function of azimuthal angle. We achieve this by placing elliptical annuli of increasing radii, and with the same position angle and ellipticity as the stellar disk, as measured in Section 3. To explore the possible effects of compression, the outer radius of the innermost annulus is set to the radius of the truncated CO disk. We then take the median flux in regions of $\theta\pm 5\degree$. We estimate the uncertainty using the 2D uncertainty map generated by Maskmoment which combines the datacube’s uncertainty level and the number of unmasked channels per pixel. We then multiply this by the square root of the area of each azimuthal slice.

We bisect the flux distribution twice, first along the leading/trailing sides, and then along the approximate wind direction. This leaves us with 4 quadrants, which are shown in Figure 7. The quadrants are ordered to follow the galaxy’s rotational direction, which is clockwise in the case of NGC 4858. Finally, the first two quadrants are the leading-side quadrants. The strength of local ram pressure varies between different quadrants, being strongest on the unshielded (leading) side rotating into the wind, and being weakest on the shielded (trailing) side rotating with the wind.

Figure 8 shows the results of this analysis, which offers a few key insights. The innermost annulus at $R=1.8$kpc shows more flux in quadrants 1 and 2 (the leading side) than quadrants 3 and 4 (the trailing side). The ratio of fluxes in the innermost annulus for these two sides is $F_{\text{lead}}/F_{\text{trail}}=1.25$. Correspondingly, the three brightest CO peaks are near the truncated southern edge, and the brightest peak is near the southern leading point. This agrees with the expected gas compression from a highly-inclined ram pressure event, where a large portion of gas found on the leading side will be pushed further into the disk, whereas gas on the trailing side has less impeding gas along the wind direction and can therefore be accelerated more efficiently.

Flux in the galaxy disk outskirts ($R=4.6$kpc and beyond) is almost entirely concentrated in quadrant 4, or the trailing-side quadrant rotating into the wind. Indeed, almost $90\%$ of the CO flux in the outskirts is found in this quadrant. Quadrant 3 also has some flux in the innermost annulus, which is evidence of the expected elongation of gas along the trailing side overall, and also in the outermost annuli as some distant tail clouds are found there.

Also seen is a trend in the azimuthal location of the tail peaks, which shift counterclockwise as a function of increasing radius. The gas tail also becomes more laterally symmetric at larger distances, changing from residing completely in Q4 near the base of the tail, to $\sim 45\%$ of the CO flux residing in Q3 at the outermost annulus. Given that this galaxy has spiral arms, this is likely caused by a combination of inherent spiral structure and rotation, which is expected to be less important in the outer RPS tail. These peaks would not change location for a more uniform, straight gas tail. The peaks also appear to become more separated as the annulus radius increases.

In the following section we describe and discuss notable morphological features in the CO distribution. We also discuss how these features appear at different wavelengths, including Subaru H$\alpha$ and HST broadband optical.

To best reveal non-circular motions which may be caused by RPS, we first subtract circular motions using a simple circular-velocity model and explore features revealed in the residual map. The location of the features relative to the major/minor axes of the galaxy provides information on the overall kinematics of these gas clumps.

Residual velocities may correspond to vertical, azimuthal, or radial motions depending on their location relative to the kinematic major or minor axis. Assuming that the motions lie within the disk plane (which is a reasonable approximation given a highly-inclined disk-wind angle which will keep gas near the disk plane), these residuals will correspond to motions outlined in Table 2. More details are discussed in Appendix B.

To obtain a residual velocity map for NGC 4858, we choose to fit a simple circular velocity model to the moment-1 CO velocity map directly. We are concerned with how the CO gas velocities compare to the simplest model of the pre-stripped galaxy (i.e. only circular motions with a typical rotation curve). To obtain the rotation curve we use the ${}^{\mathrm{3D}}\textsc{Barolo}$ fitting package (Di Teodoro & Fraternali, 2015). We use the ellipticity and position angle of the stellar disk obtained by isophotal analysis as our inputs (Section 3). We fix these parameters, as well as $x_{0}$, $y_{0}$, and $V_{\text{sys}}$ and assume no radial velocities. This provides us with estimates for the rotation curve within the CO truncation radius, but we seek to estimate the rotation curve out to the outskirts of the stellar disk. To do this, we fit a simple universal rotation curve (URC) fitting function given by

Since the rotation curve beyond the inner disk is somewhat uncertain, we have assessed the robustness of our derived velocity residuals by testing a range of outer galaxy rotation curves. In addition to our preferred rotation curve based on the Tully-Fisher relation, we also tried a rotation curve that is flat beyond the inner CO disk, and one with a maximum rotation velocity twice as high as expected from the Tully-Fisher relation. The corresponding velocity residual maps are shown in Appendix C. None of the features we discuss with significant velocity residuals can be explained away by any plausible rotation curve, although the amplitude of the residuals does vary for some of the features. Additionally, features residing along the kinematic minor axis (such as the inner-tail redshifted feature) are insensitive to any changes in the rotation curve.

We generate our velocity map using this rotation curve with a simple disk rotation model that expresses line-of-sight velocities in polar coordinates ($r,\theta$) as

The southern side of the CO gas disk close to the truncation radius of $\sim 3$kpc and between $210-280\degree$ is blueshifted to between -25 and -50 km s^-1(ID 1 in Figure 10). This region extends approximately 1kpc radially into the gas disk. Based on its location and the ram pressure wind vector, it is easy to conclude that this is gas being accelerated directly by ram pressure, in a region of the galaxy not well-shielded from the wind. The highest level of blue-shifting in this region, at 40-50 km s^-1, is towards the southernmost edge of the disk.

On the northern side of the disk is a cohesive strongly blueshifted feature (ID 2 in Figure 10). This appears in the azimuthal analysis at the 3.2 kpc annulus, reaching a residual velocity of approximately -40 km s^-1. Assuming that any stripped material is on the near side of the disk, this might correspond to vertical motions moving away from the disk mid-plane. It does not reside on either axis, so both azimuthal (faster than rotation velocity) and radial (outward) motions can contribute. Regardless of a what type is contributing, this would correspond to gas clouds being stripped away from the gas disk.

The one western blob with CO (B2, discussed in Section 4.3.2), has a very strong velocity residual of approximately -100 km s^-1. Based on its location, the residual velocity can be a combination of all three possible motions. Assuming this cloud is in front of the disk, this may correspond to outward radial motions, azimuthal motions slower than the expected circular velocity, or vertical motions moving away from the disk mid-plane. These motions would be that expected from a dense cloud experiencing strong ram pressure acceleration.

Both tail components (IDs 4 and 5 in Figure 10) have an approximately linear velocity gradient. Figure 13 compares the PVDs to the velocity gradient expected from rotation. The velocity gradient in both tail components is approximately -20km s^-1 kpc^-1 with greater blueshifts away from the disk. These gradients are about 2 times larger than that predicted from rotation alone. The region around the bases of both arms has a larger velocity dispersion compared to the gas disk, which suggests either high turbulence or multiple velocity components.

The eastern arm has velocities at its base consistent with rotation, whereas the outer arm has a gradually more blueshifted residual moving away from the disk. This is consistent with a relatively undisturbed base, but ram pressure acceleration of the outer arm, suggesting that the arm is being stretched apart by ram pressure. Towards the end of the tail component there appear to be two distinct clouds, contributing to higher velocity dispersion in this area.

The western arm has a similar velocity gradient, but has an overall redshift, relative to both the eastern arm and the expected rotation velocities. The outer arm has a modest blueshifted residual velocity, but the base of the tail, which is near the minor axis, has redshifted residual velocities which we discuss in greater detail in the following section. The western arm leads the eastern arm, and was previously at the location of the eastern arm. This might suggest a similar history to the gas in the eastern arm, in which ram pressure acceleration of the outer arm created a larger velocity gradient, with higher blueshifts for the outer arm. But now as this arm has rotated into the quadrant rotating into the wind, the entire arm is falling back toward the galaxy, giving it an overall redshift.

Contrary to the other features with high velocity residuals in NGC 4858, there is one major component that instead has redshifted residuals (ID 6 in Figure 10). This feature is located around the base of the ”bunny-ears” where the tail bifurcates, and extends down the western side of the disk. The residual velocities of this region range from $\sim 20$ km s^-1, to upwards of $50-60$ km s^-1 closer to the bifurcation.

Most of the redshifted material lies close to the minor axis. Since dust extinction implies that the tail is in front of the stellar disk, the redshifted gas would correspond to vertical motions moving towards the disk plane, or inward radial motions. In the intermediate regions where redshifted gas is seen between the major and minor axis, this might also correspond to gas moving faster than the circular speed. Given that off-axis regions are connected to the regions on the minor axis, it is reasonable to assume that the minor-axis motions (vertically and radially inwards) are the primary motions throughout. Though we cannot determine whether vertical or radial motions are more important, these motions all correspond to material falling into the disk potential.

The disk-wind angle is the angle between the wind vector and the normal vector to the disk plane. The details of this calculation can be found in Appendix A.

The galaxy’s gas tail has a plane-of-sky component extending northwards from the disk, as well as dust extinction and reddening that suggests the tail extends in front of the stellar disk (see Figure 12). This sets constraints on allowable tail angles and therefore total velocity of the galaxy. Under these simple constraints, the maximum range of disk-wind angles is $38\degree$ (largest allowed face-on component) to $90\degree$ (completely edge-on).

To get an estimate for the total velocity of NGC 4858, we can use the galaxy dynamics package Gala to model the mass distribution of the Coma cluster (Price-Whelan et al., 2024). We model the Coma cluster mass distribution as an NFW profile with a total mass of $1.2\times 10^{15}M_{\odot}$, with a concentration ($c=R_{\text{vir}}/R_{s}=9$) (Lokas & Mamon, 2003). Assuming that the projected distance from the cluster center is the true distance to the cluster center, the expected total velocity at NGC 4858’s current location is 3410 km s^-1. We can also take the uncertainty on this estimate to be equivalent to the velocity dispersion of Coma. This corresponds to a velocity along the plane of the sky at $v_{\text{POS}}=2402_{-1842}^{+1196}$ km s^-1, and a disk-wind angle of $\phi_{\text{DW}}=75_{-27}^{+10}$ degrees, or highly inclined. The uncertainty on this estimate is large, due to the many assumptions required of the galaxy’s total velocity. However, the generous uncertainty estimate based on galaxy velocity dispersion should cover all such uncertainties.

While RPS is typically described using the Gunn and Gott criterion (Gunn & Gott, 1972), this treatment is insufficient to fully describe how material is accelerated, and the subsequent observed gas distribution. Ram pressure affects the linear momentum of a gas cloud principally via mixing with ICM gas, where the velocity of a gas cloud can be described as the fraction of ICM gas present in the cloud (Tonnesen & Bryan, 2021). However, despite ram pressure being weaker on the side rotating with the wind, it is evident from simulations that this is the region where ram pressure is more efficient at stripping material (Roediger & Brüggen, 2006; Jáchym et al., 2009). The effects on angular momentum due to torques imposed on gas by the ram pressure wind need to also be considered.

Figure 14 visualizes the different effects on two gas clouds located on opposite sides of the galaxy: one rotating into the wind, and one rotating against the wind. The strength of these torques will not be equal, as the local ram pressure acceleration on the side rotating into the wind is stronger than for the side rotating with the wind. For the cloud rotating with the wind, the torque vector aligns with the angular momentum vector, increasing angular momentum. This process helps to accelerate the gas into divergent, more extended orbits. This is an important mechanism for helping to push out gas in highly inclined stripping events where rotation cannot be neglected. This will be more efficient in regions where this torque is highest, where there is less gas downstream from the clouds, and where the gas is already moving in a similar direction to the wind (as that gas will require smaller changes in momentum in order to be stripped).

For gas rotating into the wind, the torque vector is opposite to the angular momentum vector, reducing the angular momentum and decreasing the kinetic energy of the gas particles. This promotes a torque that drives gas deeper into the gravitational potential (Akerman et al., 2023). This may be an important process for fueling AGN, which might help explain enhanced AGN activity observed in RPS galaxies (Radovich et al., 2019).

The importance of ram pressure torque-induced changes in angular momentum can be evaluated by considering the differences in local ram pressure on either side of the galaxy disk. Analytically, ram pressure is stronger on the side of the galaxy rotating into the wind. The strength of asymmetries in ram pressure strengths is given by

Figure 15 identifies in the HST F600LP image a collection of stellar complexes, coincident but close to the edge of the CO emission in the disk, with considerably different morphologies. Three of the complexes on the side rotating against the wind (W1-W3) have distinct head-tail morphologies. We do not identify any distinct features in the CO disk that relate to the stellar complexes, but this may be due to resolution and/or sensitivity limits in the ALMA data. All of the tails point northwards ($\theta_{\text{tail}}=-45\degree,-25\degree$ and $10\degree$ respectively). These structures appear to be extended along the same direction as the ram pressure wind. The tails have lengths of approximately 0.5 kpc. While W1 and W3 have mostly diffuse tails, W2 exhibits signs of clumpy structure within the tail, with multiple bright knots present. One other identified complex on this side of the galaxy, W4, does not show strong signs of a tail, but may be slightly extended towards the west. On the side rotating with the wind, complexes are also identified (E1-E3) but lack any clear tail structure.

It may be initially surprising that these differences in morphology appear in stellar distributions. The complexes can be seen as ”mini jellyfish galaxies”, where the same processes that create stellar tails in RPS galaxies are also occurring on a smaller scale in gas clouds near the stripping radius. On the side of the galaxy rotating into the wind, the local ram pressure is strong enough to ablate gas from the clouds. On the side rotating with the wind, the local ram pressure is sufficiently reduced by rotation to not create such an asymmetrical structure in accelerated gas clouds. The difference in local ram pressure, described in Equation 6, appears sufficient in the case of NGC 4858 to induce such morphological differences.

While ram pressure can accelerate and ultimately strip gas, it does not always accelerate gas to the velocity required to escape from the galaxy’s gravitational potential well. In a scenario where gas is stripped from the disk but not accelerated to the escape velocity, this material has instead been accelerated to a highly elongated orbit, and may fall back into the disk.

Sticky N-body and hydrodynamical simulations show the development of an asymmetrical tail in the snapshots of a galaxy’s evolution (Vollmer et al., 2001; Lee et al., 2020; Akerman et al., 2023). This confirms that an asymmetrical inner tail does not require hydrodynamical effects to develop. Fallback has been predicted by many simulations (Vollmer et al., 2001; Tonnesen & Bryan, 2009; Akerman et al., 2023; Zhu et al., 2023; Sparre et al., 2023), and has been observed in molecular gas observations of other RPS galaxies (e.g. Cramer et al., 2021).

Figure 16 presents a schematic for the overall phases of inner-tail gas evolution and compares these phases to snapshots of two galaxy-scale simulations (one hydrodynamical, one sticky N-body) under highly-inclined ram pressure. Both of these simulations show the development of an asymmetric tail on a timescale of approximately 400-500 Myr, which is likely a similar timescale for NGC 4858’s ongoing ram pressure event. The phases can be described broadly as:

1. 1.

   Early Phase: Ram pressure initially blows out gas preferentially on the side of the galaxy rotating with the wind.

2. 2.

   Intermediate Phase: Angular momentum moves the material from the side rotating with the wind, to the side rotating into the wind.

3. 3.

   Late Phase: The tail persists and concentrates on the side rotating against the wind. It is in this region where fallback is most likely to occur.

In the case of NGC 4858 and the two ”bunny ear” tail components, it is likely that observations are showing pre-existing spiral arms that were stripped at different times and are now at different phases of inner-tail evolution. Comparing their locations with the phases in Figure 16, it appears that the central tail component (ID 4) is in the intermediate phase and the other component (ID 5) is in the late phase. This implies that as evolution in this galaxy proceeds, the easternmost component might evolve to have similar morphological and kinematic characteristics to its counterpart.

While simulations tend to develop an arm in Q4, it should be noted that these simulations create such features with no pre-existing spiral structure in the initial, un-stripped galaxy disk. NGC 4858 on the other hand clearly has prominent spiral structure, so its evolution deviates somewhat from the simulation predictions. The same forces that act in Q4 to develop an inner tail arm will also act to strengthen any pre-existing spiral arm structure accelerated into the inner tail region.