Stop stripping the dwarves, they don't like it !

Today's paper revisits a very minor but interesting storm in a teacup.

Back in 2021, Junais et al. reported on a possible Ultra Diffuse Galaxy losing gas in the Virgo Cluster. At face value it all looked very convincing. The HI gas detection was very clear, nicely offset from the UDG-candidate (basically an especially faint, fluffy sort of galaxy if you aren't keeping up with things – shame on you !) but still overlapping it. At the very centre of the gas detection was a sort of ragged line of blue starlight, plus there were some patches of stars scattered about as well. It's all very much as you'd expect if this was star formation occurring in the stripped material.

Okay, ram pressure is old hat. But to find evidence of this occurring in a UDG would be especially interesting : it would allow us to start investigating whether UDGs in clusters (which seem to be pretty common) are the same as those in the general field (which are known to exist, but we don't know how many there are). In particular, there's this whole controversy over whether they lack dark matter or not, in which case the effects of stripping might be quite different since the gravitational forces involved would be much less. And also it would show whether both cluster and field UDGs form by the same process, or whether there are multiple ways to form the same sort of objects.

All this was strongly challenged by Jones et al. 2021  They said, no, hang on, the distances are all wrong. Using high-resolution Hubble data, they were able to show that the distance to the UDG-candidate is actually much closer than the Virgo Cluster, and it also seemed to be linked to another, much brighter galaxy (VCC 2034) by a giant bridge of HI, so presumably that would also be at the same distance. The patchy starlight, however, could well be in the Cluster, in which case it would require a different explanation because it doesn't look anything like a galaxy.

You may or may not remember that I'm moderately skeptical about all this. It's not that I don't believe the distance estimate... it's that I'm wary about them after that whole "ping-pong" series of papers concerning some other UDGs – a debate which apparently still isn't fully settled. That suggests we shouldn't take any single value as definitive but should wait for multiple analyses. 

And the HI stream... although I did send Jones some of our deeper (WAVES) data, and he was able to find the stream by taking a slice through at the right angle... it feels very off to me. Given that our HI data is about 3-4x deeper than the original ALFALFA observations, I'd expect it to be immediately obvious in our data. It isn't. My suspicion is that the analysis and source-finding package SoFiA (which is hella powerful) is oversmoothing here, creating the appearance of a bridge because of the degraded resolution, as is used for increasing sensitivity.

I don't know for sure though. I'm moderately skeptical, but no more than that.

Enter today's paper by Yu-Zhu Sun and friends. They use a combination of new deep data from FAST and high resolution data from the VLA. And this paints a pretty convincing picture that the patchy starlight is not the result of gas stripping from VCC 2034, even if they agree that it's nothing to do with the UDG candidate. This is my favourite sort of paper in that it doesn't actually solve the mystery but just demonstrates that things were even weirder than initially thought.

This is all quite complicated : there are several different galaxies in this region, plus the fuzzy stars, plus the possible gas bridge, and conflicting distance claims for all of them. Let's try a few simple diagrams to illustrate. I'll start with the main hypotheses proposed by Junais and Jones. But, since both groups quite rightly caveat their conclusions and aren't definitive, and don't all deal with the same objects, I'm going to try and standardise and simplify things a little bit. This should be enough to get a general sense of what's going on, but this is very much a limited guide. Needless to say, these are not to scale !

Hypothesis 1

The most straightforward interpretation is the original : that the fuzzy starlight ("The Fuzz", a.k.a. AGC 226178) close to the UDG candidate (here UDG-X, official designation NGVS 3543) and is the result of star formation in its stripped gas. The nearby pair of galaxies VCC 2034/2037 are seemingly unrelated. 

All galaxies, in this scenario, are in Virgo at about 17 Mpc distance from us. The gas cloud associated with UDG-X and the Fuzz align well and VCC 2034/2037 is rather far away, so an association isn't at all natural. VCC 2034 has its own gas, showing clear signs of removal. In fact this extends in the direction of UDG-X but doesn't reach nearly far enough, so the orientation doesn't appear to indicate anything interesting. It's also aligned with VCC 2037, but that too is imperfect (not covering the whole of VCC 2037 and the local maxima of the gas is not aligned with the galaxy's centre) and the velocities of the two galaxies don't match well. So this too may just be a coincidence – the two objects might both be in the cluster, but at sufficiently different distances that they aren't actually related. Regardless, they really don't seem to have anything to do with UDG-X at all.

Hypothesis 2

The second scenario relies on a number of additional observations. Direct distance estimates suggest that both UDG-X and VCC 2037 are at 10 Mpc, much closer than the Virgo Cluster (17 Mpc, estimated elsewhere to be be 1-2 Mpc deep). However the Fuzz seems still to be at the cluster distance, and there's a much larger bridge of HI apparently connecting it to VCC 2034. So essentially, the Fuzz results from gas stripping of the cluster member VCC 2034, whereas UDG-X is so close to us that's actually not a cluster member at all : it may or may not relate to VCC 2037 instead. This would make UDG-X an uninteresting normal dwarf galaxy, but the Fuzz becomes very interesting as a rare example of star formation in a gas tail.

Note again that that the existence of the large HI envelope is uncertain, and that it's probably not a great idea to trust the distance estimates overmuch. Furthermore, as we're about to see, even the high resolution HI data can't be treated as gospel.

Hypothesis 3

Stressing that the latest paper is even more cautious, here's their essential idea : there's no big HI envelope and both VCC 2034/2037 show independent HI tails (in the new VLA data) that don't align with the Fuzz or UDG-X at all. UDG-X may well be foreground (again making it a normal dwarf galaxy), but neither it nor the Fuzz are directly related to any of the major galaxies in the general vicinity. What, then, is the origin of the Fuzz in this scenario ?

A tricky question indeed, one which they understandably don't commit to answering. Their main conclusion is that the Fuzz is likely not stable and in the process of disintegration, but as to what formed it in the first place, they don't (can't) say.

Disclaimer : I know a few of the co-authors very well, have published with them, and certainly hope to do so again ! They raise many excellent points, but there are a few with which I disagree. For example, they say that the HI cloud around the Fuzz has a "well-defined" velocity gradient of 10 km/s, but that's the width of the HI line itself so I'm very skeptical that this can be in any sense meaningful.

They do, however, have both new, extremely sensitive FAST data (even slightly deeper than WAVES), and new VLA data which should be of even higher resolution than the earlier observations. The FAST data fails to show the large HI envelope, as does WAVES – and taken together this seems to quite reasonably disprove its existence. I had in mind a simple project to see if this could really result from how the data was processed... maybe one day I'll have the time to try it, as it would be nice to know exactly how this happened if indeed it doesn't exist.

What about UDG-X ? The FAST data is highly sensitive but low resolution, and can't distinguish gas associated with the Fuzz (which definitely does exist) from UDG-X. The Sun et al. VLA data, however, shows much less of a head-tail morphology than the earlier data, now appearing to only be associated with the Fuzz. That makes it unlikely that Fuzz is the result of gas stripping from UDG-X, though it can't be said with too much confidence. There could still be diffuse gas in UDG-X which the VLA wouldn't detect, or the entire gas of the Fuzz might have been displaced wholesale from UDG-X.

And when they say they detect a velocity gradient in this case, it looks a lot more like a very sudden change to me. Their dynamical mass estimates – how much mass is needed to keep the system stable – are, I think, stretching things beyond the quality that the data can sustain, given how narrow the velocity width of the object is. That said, they say the total amount of dark matter that would be present is so low that this is unlikely be a dark/dim galaxy candidate : more likely it's some form of debris. That seems entirely reasonable from the low line width, even if I'd be skeptical about the exact dark matter mass estimate.

But is the debris stable ? That's much harder to answer. A lot of recent work has found candidates for so-called "blue blobs", which are interpreted as gravitationally-bound clumps of gas and stars that formed by the removal of gas from ordinary galaxies by ram pressure. In essence this would be a new class of stellar system, not really galaxies in the classical sense (since they'd have no dark matter) but not star clusters either (being very much larger and formed by a totally different mechanism).

Personally I rather like this idea, but here they place a few well-aimed holes in the scenario. The high metallicity of the clouds seemed in Jones like strong evidence that the clouds originated from within galaxies, as otherwise their chemistry should be basically hydrogen and bugger all else – you need prolonged star formation to cause significant enrichment, which isn't going to happen at their current pathetic levels of star formation activity. But here they say it could happen through mixing with the gas in the cluster itself. On the other hand, the paper they cite in support of this says that metallicity should drop with distance from the parent galaxy, whereas all the blue blobs have essentially the same high metallicity value. So this is an interesting critique, but not a fully convincing one.

Similarly, they're rather skeptical of the whole pressure confinement scenario for blue blobs – the idea here being that the gas within the cluster helps prevent them from disintegration. Now when we simulated this for dark clouds with very high velocity dispersions, we found it flat-out didn't work. But we were investigating rather exceptional systems, and simulations of low velocity dispersion systems have found very much more favourable results (as you'd expect anyway : with a low dispersion, things can only expand more slowly by definition). So I think their toy model is overthinking things. In any case, given the extremely low dispersion of the Fuzz's gas cloud, it would only expand by 10 kpc in a billion years... even if it is technically disintegrating, it's doing so so slowly that it might as well not be.

Finally, I don't agree at all with their interpretation regarding the location of the blue blobs within the cluster. The previous paper by Dey suggested that they're found in regions of modest cluster gas density because this is where they can both form and survive for a while; they avoid the denser core because this would rapidly destroy them. But Sun et al. claim that a "more natural" suggestion is that actually these objects are all outside the cluster in 3D space and only appear projected against it. Surely, though, if that were the case, we'd be equally likely to see such objects projected against the core ! To me, that the distribution of the objects relates to the geometry of the cluster feels like extremely compelling evidence that they are indeed within the cluster.

The long and short of it is that this is a very complex system, and it all serves to underscore that even observations don't always get the last word. It's particularly interesting that the new VLA data looks markedly different to the earlier findings, showing distinctly different structures. Likewise, I have to wonder why everyone is treating the distance estimates with such high confidence, given recent prominent debacles about how damn difficult it is to get these right.

As it stands, it now looks a lot less likely that the origin of the Fuzz can be explained by a giant gas stream from VCC 2034. But I, for one, am by no means convinced that we can rule out the original suggestion of stripping from a UDG, and I downright disagree that we can be so confident that it's a disintegrating gas cloud rather than a ram pressure dwarf. It's likely not a dark galaxy, however. 

Which leaves the usual question of : what would it take to resolve all this ? This is very tricky. Well, the question of the long gas stream could be easily answered by running SoFiA over data sets with artificial signals injected of similar configurations to the current system; if the long stream results from oversmoothing, this ought to be reproducible. Distance measurements are much harder to resolve unambiguously, but at a minimum, another team need to try this independently, preferably using different data. As to why the various VLA data of the same objects looks so different, however, I'm at a loss. It's definitely a weird system, but certainly an interesting weird.

The Bunny Rabbit of Death

Today's paper is a bit more technical than usual, but sometimes you've gotta tackle the hard stuff.

Ram pressure stripping is something we seem to understand pretty well on a large scale. When galaxies enter a massive cluster containing its own gas, pressure builds up that can push out the gas in the galaxy. If it's going fast enough, and/or the cluster gas is dense enough, then the galaxy can loose all of its gas pretty quickly. No ifs or buts, it just looses all its gas, stops forming stars, realises it's made incredibly poor life choices, and dies.

Yeah, literally, it dies. It's run out of fuel for star formation, which means all its remaining massive blue stars aren't replaced when they explode as supernovae in a few million years. Slowly it turns into a "red and dead" smooth, structureless, boring disc, and maybe eventually an elliptical. There's a wealth of evidence that ram pressure is the dominant mechanism of gas loss within clusters, and everything seems to just basically... work. Which is nice.

But, as ever, the details are where it gets interesting. In the extreme case, what you'll see is a galaxy with a big long tail of gas, one single plume stretching off until it's torn apart and dissolved in the chaos of the cluster. 

Even here things can be complicated though. Some tails seem to have multiple components : extremely hot X-ray emitting gas, cooler neutral atomic hydrogen detectable with radio telescopes, intermediate temperature ionising gas that emits over very narrow "Hα" optical wavelengths, and very cold gas indeed that emits in the sub-mm regime. They may or may not have stars forming within the plume, and all of these different components can have radically different structures. Or they might all line up quite neatly. Sometimes all of these phases are present, sometimes just one or two.

And then, if a galaxy isn't in the extreme case, it can be even more complicated. If the ram pressure isn't enough to accelerate the gas to escape velocity, it can still be pushed out only to fall back in somewhere else in the disc. In short, it gets messy.

This paper attempts to understand one of those messy cases. It's part of the ALMA JELLY program, a large ALMA observing program run by my officemate Pavel Jachym (conflict of interest : declared ! BOX TICKED). Here they introduce the first analysis of one of their 28 target galaxies and tackle the important question (though they would never dare state it thus) : 

Why does it look like the Playboy bunny rabbit ?

Wait, wait... why is it called ALMA JELLY ? It's not an acronym as far as I know. Instead, "jellyfish" galaxies have become a popular name for galaxies experiencing ram pressure stripping as some of them have distinct, narrow tails that look very much like the tentacles of a jellyfish. The term has become somewhat abused lately, often used for any ram-pressure stripping galaxy regardless of what its tail looks like. Here they attempt to take back control of the term and define it as galaxies which have stars forming it their stripped material. This often occurs in narrow tendrils so it's a pretty good proxy for jellyfish-like structures, and highlights the unusual physics at work in these cases.

And, why ALMA ? ALMA observes the cold molecular gas, which is generally agreed to be the main component of star formation. The target here already has many observations at other wavelengths, but the molecular gas has been traditionally tough to observe. Now they can fill in the gap, and with extreme resolution too. 

So, the bunny rabbit. The first target for ALMA JELLY is NGC 4858. It's certainly a prime example of a jellyfish galaxy, with clear, bright tendrils of stars extending in one direction directly away from the centre of the Coma cluster in which it resides. It's also close to the cluster centre, where ram pressure ought to be very strong. Its got observations at a bunch of different wavelengths and it is, in short, a right proper mess. Really, it's the kind of thing I might be minded to throw up my hands and say, "hahahah no, I'm not touching that with a barge pole". Or, failing that, I might wave my hands furiously and say, "something something HYDRODYNAMICS !".

Hydrodynamic effects, the complicated interactions between two or more different fluids, are an easy get-out. Mixing of fluids causing extremely complex structures, so if something's a mess, it's a safe bet that hydrodynamics can explain it. Though, in that case you ought to run simulations to test if that really works or not.

Here they don't. Instead they try the much braver task of explaining it without any dedicated simulations, and even those simulations they do use don't have full hydrodynamic effects – just some very basic approximations of the major forces at work from the external gas. And yet they seem to have come up with a pretty convincing explanation.

It works like this. First, NGC 4858 is a grand design spiral, with two prominent spiral arms. As it rotates, each arm moves through a region where its subjected to varying ram pressure forces, which are greatest on the side rotating away from the cluster centre (where the gas is moving fastest away from the cluster, making it easiest to remove). A single, dense arm thus gives rise to a single, dense plume of gas – a tail. But this tail gas preserves some of the rotation it had around the galaxy's centre, so it doesn't just get blasted out into space – it keeps moving around the galaxy. This brings it into the shadow of the galaxy, protecting it from the wind of the cluster. Some of the gas is lucky enough that the greatly reduced ram pressure is now essentially impotent, and it falls back onto the galaxy.

Not all of it though. Some keeps going. If any makes it right around to the other side of the galaxy, it moves back into the zone of death and gets finally stripped away by the cluster gas once and for all. The key is that before it reaches this point, the gas gets compressed as it starts to hit the wind again. In the simulations they use as a reference, the galaxy doesn't have prominent spiral arms and shows a single prominent tail; they surmise that because NGC 4858 has two arms, this could naturally give rise to two tails (or ears).

Their observations also show direct evidence of gas returning to the galaxy. The ALMA observations allow them to make a velocity map of the gas, and there's one big feature which is discontinuous with the rest of the velocity structure. And again, that fits with the basic model of how they expect rotating gas to behave.

I've simplified and shortened this one quite a lot, missing out on any number of interesting details. And there's an awful lot more they could still do with this data. But to me, the first thing I wondered when I first saw the ALMA image was "why is it a bunny rabbit ?". I was expecting this to have a much more complex non-answer, featuring hand-waving and invocations to hydrodynamics galore, possibly involving a chicken sacrifice. As it is, they managed to come up with a decent explanation without any of that, which is no mean feat. Both the bunnies and the chickens can rest easy.

Now all they have to do is convince Playboy to give them a sponsorship deal...

The Miniscule Candidate

Following on from those couple of papers on possible dark galaxies, comes... another paper on dark galaxies !

This one is a completely different sort of beast. While identifying optically dark galaxies is normally done by looking for their gas instead of their stars, here they use good old-fashioned optical telescopes instead. Even weirder, having found something which is optically faint but not dark, they then go on to infer its dark matter content without measuring its dynamics at all !

If this all sounds very strange, that's because it is. It's by no means crazy, but it must be said that some of the claims here should be taken with a very large pinch of salt.

Let's go right back to basics. A good working definition of a galaxy is a system of gas and/or stars bound together by dark matter. True, there are some notable exceptions like so-called tidal dwarf galaxies, but it's questionable whether we shouldn't drop the "galaxy" for those objects altogether (maybe replace it with "system" or something instead). Clearly they're physically very different from most galaxies, which are heavily mass-dominated by their dark matter.

A dark galaxy, then, is just a dark matter halo with maybe some gas but definitely no stars. Or is it ? For sure, if it really has literally zero stars, then such an object would definitely count as a dark galaxy. But what if it had just one star and billions of solar masses worth of dark matter ? Would it really be worth getting hung up on that point ? Presumably the physics involved in its formation would be basically the same as a truly dark object.

Generally speaking, most people would allow an object to qualify as a dark galaxy even if it had some small mass in stars. At present there's no strict definition, however, and so few candidate objects are known that setting a quantitative limit wouldn't really help. Right now, we don't know nearly enough about the physics of the formation of such objects, and indeed the jury's still out on when any of them exist at all. 

(Some people prefer the term "almost dark", which annoys me intensely. I prefer to call them dim when they have some detectable stars, but it hasn't caught on).

Anyway, you can see how this explains using an optical telescope to search for dark galaxies. But actually, here they go a step further. Rather than looking for the ordinary stellar emission from galaxies, which are normally in diffuse discs, they look only for the light emitted by the compact, relatively bright globular clusters. Most galaxies have these dense starballs which orbit around in their halos quite separately from their main stellar disc. What these authors are looking for are cases where they find groups of globular clusters without an accompanying disc : essentially, star clusters orbiting all by themselves in their dark matter halos. 

This is an interesting grey area in terms of calling something a dark galaxy, but I'd be inclined to say such objects would qualify. The physics at work in forming dense globular clusters and the diffuse stellar disc is quite different, so at the very least, these would certainly be extremely interesting.

Here they present the imaginatively named "Candidate Dark Galaxy 2". Really ? Yes, really. That's the name they're going with. Bravo, team.

(Actually, snarkcasm aside, this is a wee bit insulting, considering that there have been many candidate dark galaxies over the years, but I'll let that pass).

It turns out they had a previous candidate (you can guess the name) which is even more extreme than this one. CDG-1* consists of four globular clusters in close proximity to each other with no detectable diffuse emission between them at all. I won't attempt to discuss the complicated statistical methods they use to identify globular clusters without parent galaxies; at the words "trans-dimensional Markov chain" my eyes glazed over anyway. I can safely mention a few points though : 1) They don't have spectroscopic measurements of the globular clusters so they can't robustly estimate their distances*; 2) Their initial catalogues of globular cluster candidates are surely incomplete, but 3) Since they do careful inspection of the candidate cluster groups they do find, we can be confident that the associations they identify are real.

* I honestly can't remember if I heard about this at the time or not. I may have missed it or just forgotten about it.

* Spectroscopy gives you velocity, which is a very powerful constraint on (though not quite a direct measure of) distance.

CDG-2 initially consisted of three globular clusters, but here, using new data from Hubble and Euclid, they identify a fourth. While they still don't have spectroscopy, the new data confirms that the candidates are all unresolved. That means they cannot possibly be close objects, and in fact their colours and other parameters are consistent with their being in the Perseus galaxy cluster* at 75 Mpc distance. So it seems very unlikely that they're either significantly closer or further away. And while their might be a few free-floating globular clusters in Perseus (ripped off their parent galaxies by tidal encounters and the like), it's not very likely that they'd happen to be so close together.

* This can sometimes get very confusing. A globular cluster is a cluster of stars that orbits around a parent galaxy; a giant galaxy might host, say, several dozen such objects. A galaxy cluster is a whole bunch of galaxies, each with their own population of globular clusters, all swarming around together.

The killer argument that this is highly likely to be an actual galaxy, though, is that here they detect diffuse stellar mission between the globular clusters. The thing just looks like a galaxy, albeit an extremely faint one. The chance of a tidal encounter creating something like this isn't worth considering.

Ahh, but is it a dark galaxy ? That's where things get a lot more speculative. While we can be pretty sure about the distance of the object and their physical association, only spectroscopic measurements would really give a good handle on the total mass. Measuring how fast things are moving lets you infer how much mass you need to hold them together. Without this, they rely on scaling relations, extrapolating based on the globular clusters to infer a massive amount of dark matter : probably there are a few million solar masses of stars present in total, but it could easily have a hundred billion solar masses of dark matter based on the scaling relations. 

These are, however, truly enormous extrapolations. Given that Ultra Diffuse Galaxies are now known which have significantly lower dark matter contents than typical galaxies, but these too have globular clusters, I'd be wary about digging any deeper into this one until they get some spectroscopy.

 Even so, it's clearly a very interesting object indeed. Arguably even more interesting, however, is CDG-1, which still has no diffuse emission detected at all. Even if the extreme dark matter content turns out to be a wrong estimate, if either of them have any at all, they're still super weird objects. Hopefully when they find CDG-3 I won't be caught quite so unawares.