I'm happy to report that the refereeing process for my latest paper was one of those instances where I couldn't honestly find anything to complain about. In fact the reviewer provided several papers to cite which were, unusually, all genuinely relevant and interesting. And they weren't even an obvious case of "please cite me !" as so often happens with suggested citations ! Goodness, a reviewer actually providing interesting things from a desire to be helpful, whatever next...
Anyway, if you don't want to read my paper or the linked blog post you don't have to. It's basically a very simple one : we found an Ultra Diffuse Galaxy apparently losing gas in the Virgo Cluster. You might remember another such claim (which I've covered a few times previously, of which this post summarises the current state of affairs) which has proven controversial. This one, I think, should be relatively unambiguous as far as the main claim goes. What will likely cause more of a discussion – if anyone notices it at all – is the assertion that this might, like other UDGs, have a depleted dark matter content.
But what I want to do here is give a bit of a speed summary of the other papers I ended up reading as part of this. The referee seems to be of a background from the world of globular clusters, which is far outside my area of expertise. I tend to not pay enough attention to the studies of globular clusters in UDGs, which is a bit silly because that's a whole other area in which UDGs have proven controversial.
One of the claims in my latest paper was that if our UDG was typical of other UDGs in clusters, this implies that many of those also lack dark matter. The referee pointed out that there are different claims regarding the globular cluster content of field and (galaxy) cluster UDGs, meaning that they likely don't have the same origin... which rather limits the impact of what we can state regarding this one object. This led me on a paper chase down a rabbit hole, and for the sake of not having to write out everyone's names the while time, here are the links and abbreviations :
Forbes+2020 : F20; Lim+2020 : L20; Benavides+2021 : B21; Grishin+2021 : G21; Junais+2022 : JA22; Jones+2023 : JO23; Hartke+2025 : H25; Sandoval Ascencio+2025 : SA25.
Right then, with no less than eight different papers to cover, let's do this one thematically. There are a number of issues which most of the authors addressed in different ways, so I'll try and unify what they were all getting at.
Where do they become diffuse ?
That is, are they born big and fluffy or do they have big fluffiness thrust upon them ? Opinions differ, but here the consensus is reasonably clear. B21 argue that some field UDGs are too isolated for environment to have had much of an effect so they must be intrinsically large objects, saying that infall into a cluster will change other aspects but not size. F20, L20, and JO23 all imply that UDGs must have been large before cluster infall, but only on the circumstantial grounds that the differences between field and cluster UDGs are too stark to allow the one to evolve into the other. Only G21 argues directly against this, saying that modelling shows the removal of gas from UDGs significantly changes their gravitational potential, thus allowing stellar orbits to considerably expand.
I'll go through some of the more detailed arguments below, but I generally agree with the majority here. There are more than enough UDGs detected in the field that there simply doesn't seem to be any real need to invoke environment to explain the size of UDGs. It's entirely possible that some UDGs which now seem isolated lonely hermits were once denizens of more hedonistic environments, but it's not credible that this is true for all or even most of them.
I do think both B21 and G21 might be overstating things though. Removal of gas certainly does have the capability of making some galaxies more diffuse, but this alone can't possibly explain the majority of isolated UDGs. In short, environment is likely only a contributing factor, and not the dominant one.
Where are they quenched ?
Closely related to where they become diffuse is the question of where they stop forming stars. Broadly, most cluster UDGs are red and dead, whereas many field UDGs are blue, gas rich, and star forming. But there's a big selection effect here : cluster UDGs are easy to spot whereas field ones are much harder. The issue is distance. It's perfectly reasonable to assume that a large population of galaxies which is aligned on the sky with a cluster, and not found in control fields, is at the same distance as that cluster. In the field you don't have any such corroboration. Gas measurements can provide distances, but that leads to the selection effect : if you select by gas, of course you'll find that field UDGs are blue, gas-rich party animals.
Still, overall this picture holds up very well. Hardly any cluster members have been detected with gas, which is not unexpected : it makes complete sense than an infalling UDG will lose its gas and stop forming stars – in fact, it can hardly avoid it. The issue is more whether this can also happen in the field rather than whether it happens in clusters at all. While B21 say that even apparently very isolated, quenched UDGs are probably "backsplash" galaxies (that is, cluster members that escaped), SA25 say they've found at least two UDGs where this is unlikely.
It isn't clear to me why SA25's field-quenched galaxies can't be the same "used to be cool" UDGs that once hung around in rowdy galaxy clusters, as B21 prefer for similar objects. But it's crucial here that B21's conclusions are from simulations, not observations. How, then, could we tell them apart observationally ? How could we say that an isolated UDG was once quenched in a cluster versus in situ ? This is far from obvious.
Clearly, clusters can and do quench UDGs, but whether they can also quench in the field is not at all settled. The potential mechanism by which this would happen turns out to have very interesting consequences, but we need to examine a few other points before we get back to this.
How do field and cluster globular cluster populations in UDGs differ ?
Globular clusters are a component of galaxies I probably don't know enough about. They're dense little starballs that orbit throughout a galaxy's halo, and as JO23 describe, they're thought to form at the very first stages when a galaxy assembles. Establishing how many GCs a UDG has, though, seems a matter of some difficulty. Most studies appear to do this statistically : they use some selection criteria to see how many GC-like objects they find in a control field, then count how many similar objects they find in close proximity to a UDG. This means that in some cases, the estimated GC count for a galaxy can actually be negative since it might have less GCs than the control field predicts.
The situation is not at all clear. F20 found significant variations in the UDG-GCs within the Coma cluster, with some being especially rich compared to comparable non-cluster galaxies, while others were more typical of the populations seen in dwarf galaxies. Fornax dwarfs appear to have fewer GCs per galaxy, while L20 find that Virgo UDGs have more GCs than Fornax but less than in Coma.
JO23 go on to make a bold assertion that since gas-rich field UDGs have less GCs than in clusters, this definitively rules out field infall as the major origin of cluster UDGs. To be fair, it's very difficult to see how an infalling UDG could gain globular clusters, as long as the standard paradigm that GCs only form early in the universe is indeed correct. This is a pretty strong argument, but I'd add two major caveats : first, we know star formation can be triggered by the onset of ram pressure stripping, so maybe GC formation is also possible; secondly, less speculatively, I'd like to have much more secure data on the cluster GC population. Pretty much everyone seems to agree on a bimodality of cluster GC populations, with "multiple pathways to UDG formation" being an almost ubiquitous phrase in all of these papers.
What is their total mass ?
This bimodality is strongly evident when it comes to their mass estimates as well. F20, L20, J22, JO23, and SA25 all suggest that at least some UDGs are basically normal galaxies : the extreme tail-end of a population that happens to have a very low surface brightness, but with the same formation mechanism at work. Some of these authors (more than I was expecting) quite strongly support the notion that some UDGs are likely "failed giants", with a total mass comparable to the Milky Way, while allowing other UDGs to be essentially just faint dwarfs.
JO23 expresses this last option most bullishly, dismissing the interesting claims – and I think here very unfairly – of UDGs lacking in dark matter as being problems of data quality. In this view there's nothing much to explain, since most UDGs would be entirely normal objects, just the extreme end of those with typical characteristics. This is an especially weird and surprising perspective to me, since Jones has been at the forefront of research into so-called "Blue Blobs" : stellar structures which resemble galaxies but appear to lack dark matter !
F20 and others express in my view a more reasonable opinion. Some UDGs are quite likely just extreme dwarfs while others are failed giants, and everyone accepts that multiple formation mechanisms are at work to some extent. It was very surprising to me that only JO23 mentioned the possibility of a dark matter deficit, and then only in critique – but this may reflect the GC-centric bias of these studies. Because they're relatively easy to detect, GCs have become popular as a proxy to estimate the total mass of UDGs, and this is far easier than trying to measure the dynamics directly. It feels like the overall mood of the community here is geared towards measuring dark matter excess through GCs rather than deficit, which it itself quite interesting. But as L20 point out, this requires extrapolations across four orders of magnitude so maybe this isn't such a good idea.
What effect does environment have ?
J22 run some detailed modelling to see if ram pressure can reproduce the observed properties of cluster UDGs. They find that it can, which isn't that surprising : as you'd expect, it makes them red and dead. Furthermore they find that without this gas-loss based quenching, they can't get results which agree with observations. It seems like ram pressure stripping is very much a key aspect of the evolution of cluster-member UDGs. But JO23, again I think rather unfairly*, criticises this on the grounds of the GC population. The thing is that J22's study was based in Virgo, where from L20 we know that the GC population isn't much like those of other clusters, and we also know that GCs vary considerably even in the same regions.
* Apologies to Jones, but while I've often told anyone who'll listen about the importance of BBs, I can't say I agree with this particular work all that much.
It's also worth noting that the L20 UDGs agree with J22's scenario quite nicely, being generally redder and more numerous towards the cluster centre, but L20 themselves note that this is in contrast to previous findings. So there's a lot of controversy and complications here, and I wonder if there's a degree of mixing apples with oranges. Sure, not all GC changes can be explained as the result of field infall, but that doesn't invalidate the result that ram pressure is needed to explain other UDG changes – and those GC differences might not apply in Virgo anyway. Oh, and the L20 sample is rather small, just for good measure.
On another front, H25 present the case of a very interesting individual object which might just poke JO23 gently in the eye. They describe a UDG aligned with a stripped ionised gas filament seen from a larger cluster member galaxy, speculating that we're seeing a UDG actually in the process of formation. This object has no GCs of its own, which would definitely be at odds with F20/JO23's claims that cluster UDGs have more GCs than field UDGs.
Now that's potentially a really cool and weird result, but unfortunately this object is (to my mind)... totally unconvincing. It's perfectly smooth and structureless, with a considerable offset from the gas filament rather than a neat alignment. It's also red, implying it's already been stripped and quenched, but this makes very little sense if it's still in formation or was formed very recently. They suggest dust could explain this, but dust is seldom detected in low surface brightness objects (if it was external dust, this should be directly measurable by how much other objects in the vicinity appear reddened). The age estimate of several gigayears is also at odds with a recent formation, which ought to suggest an age of < 1 Gyr.
In my view this is no more than a projection effect, a perfectly innocent gas filament happening to appear near the UDG on the sky but without any true association between the two.
What I mainly learned from all of this is that UDGs are complicated.
The overwhelming consensus is that there are many different ways to skin a cat : almost certainly, some UDGs form by different mechanisms than others. The issue is where the balance lies, in establishing if any one scenario is more prevalent than the alternatives, and trying to find some common factors to give us a unified picture of what's going on.
Certainly I think the argument that field and cluster UDGs have different GC populations is a very powerful and important one. Were it the other way around, the situation would be different. It's relatively easy to imagine how a GC-rich field UDG could lose its GCs in a cluster, but gaining them ? Maybe not impossible, but it's hard to see that happening much.
The final interesting point is that both SA25 and JO23 propose a mechanism by which UDGs could be fluffed up even in isolation. They say that since they're extremely diffuse objects, they might experience bursty star formation rather than the more usual stable, continuous variety that occurs in brighter objects. This would periodically redistribute the gas, keeping it at a relatively low density. Every so often it would recollapse to a denser level and trigger another burst of star formation, and the cycle would repeat, eventually blowing out all their gas permanently and quenching them completely. UDGs might even then be a normal phase of evolution in dwarf galaxies : a temporary, transient state rather than a genuine type of object.
Intriguing... but I'm skeptical. If this was the case then I'd have expected simulations to have predicted this well ahead of time, along with the results of an apparent deficit of dark matter. We also don't see especially extended gas discs in field UDGs, nor (to my knowledge) any particular offset between the gas and stars. What few resolved gas observations we do have look like normal, stable, circular-ish discs.
All this underlines just how far we have to go. We have a great deal of data in some areas, but in others we have a woeful lack of even the basics. We don't have a common framework to decide when a UDG might be a failed giant or a dark matter-deficient object; weirdly, these two very different circumstances appear to result in objects which are optically indistinguishable. The old cliché is in this case emphatically true : more research is needed.
