Are sponges the closest relatives of the rest of the animals?

March 21, 2021 • 9:30 am

A new paper in Nature Communications highlights an ongoing controversy in the evolution of animals: what are the closest relatives of living multicellular animals?

First, though, we need to refresh ourselves on what “animals” are. Merriam-Webster defines them adequately:

Any of a kingdom (Animalia) of living things including many-celled organisms and often many of the single-celled ones(such as protozoans) that typically differ from plants in having cells without cellulose walls, in lacking chlorophyll and the capacity for photosynthesis, in requiring more complex food materials (such as proteins), in being organized to a greater degree of complexity, and in having the capacity for spontaneous movement and rapid motor responses to stimulation.

We’re leaving out the single-celled “animals” here (under “outgroups” in the figure below) and concentrating on multicellular animals.

The multicellular animals include (as the phylogenies show below), Ctenophores, or comb jellies, Porifera (sponges), Placozoans (free living but small multicellular organisms), Cnidaria (corals, jellyfish, sea anemones and their relatives), and Bilateria (everything else; all animals with a head and tail end, as well as a belly and a back at some stage of their life, including echinoderms which have these features as larvae).  Over the years, a combination of developmental, morphological, and molecular analysis has given rise to the two conflicting family trees shown below.

Both trees are the same except for a dispute about the “animal outgroup” (the “breakaway group” or “sister group”), the closest living relative to the vast bulk of the animals, and the first group to branch off from the rest. One school, shown on the left, adheres to the ctenophores, or comb jellies, as this sister group. The other, shown on the right, maintains that sponges occupy this position, and ctenophores branched off later.


Here’s an example of a ctenophore (photos from Wikipedia):

And a bunch of sponges:

Now the case for sponges as the sister group is based on the observation that ctenophores share unique features with the other animals, including elements of nervous systems, and (except for Placozoans) muscles and a tubelike digestive system (“gut”). But sponges have none of these. Moreover, sponges are made up of collared cells, or choanocytes, which are similar to “choanoflagellates“, singled-celled protozoans thought to be the closest relative to all the animals from sponges on down. This similarity implies that the common ancestor of multicellular animals might have been something spongelike, supporting the second phylogeny above. That implies that sponges changed relatively little after multicellular animals evolved, while everything else changed a lot more.

But some molecular phylogenies have suggested that the more complex ctenophores might be the outgroup instead of sponges.  This is a bit more problematic to both me and Matthew (see his BBC broadcast below), for if sponges are really more closely related to other animals than are ctenophores, why do ctenophores have muscles, nerves, and an in—>out digestive system like most other animals, but sponges lack these things? To hold that ctenophores are the sister group instead of sponges requires that you posit one of two possibilities:

A.) The common ancestor of all animals had nervous systems and muscles and a gut, which persist in all groups but the sponges, and the sponges lost these features. That seems unlikely, but it’s possible.


B.) The common ancestor of all animals lacked these features, but they evolved independently in the choanoflagellates and all other animals save sponges. This seems even more unlikely since it requires the independent evolution of three complex traits in two separate groups (ctenophores and [other animals minus sponges]).

This principle of “parsimony” alone suggests that sponges are the sister group, didn’t lose any of those features, and muscles, nerves, and a gut evolved only once.

The new article in Nature Communications supports the “sponges first” scenario. Click on the screenshot below to read the article, see the pdf here, and find the reference at the bottom of this post. The authors used a new way of making phylogenies using DNA data, dubbed “partition site-heterogeneous models” to eliminate artifacts that may have erroneously shown ctenophores as the sister group of other animals. I’m not going to explain that method and, to be sure, I don’t understand it. In fact, the main results of the paper for the layperson can be described very simply: the new method shows that sponges are the sister group to all animals, a result that makes sense.

I just gave you the punch line, but have to add that the controversy isn’t settled. It is settling, however, as more and more biologists come around to the “sponges split off first” scenario. (I won’t even mention the controversy about the placozoans and ctenophores, and where they fit with relationship to Cnidaria.) Let me just put in the authors’ paragraph where they say that their finding of sponges as the sister group of all other animals is definitive: (my emphasis):

Several studies have already shown that gene family and unpartitioned phylogenomic analyses using more sophisticated substitution models reject Ctenophora sister in favour of Porifera sister. Here, we have consolidated these findings by directly showing that the primary remaining lines of evidence supporting Ctenophora sister, partitioned phylogenomics and measures of underlying support (such as ΔPSlnl values), do not do so when better-fitting site-heterogeneous models are incorporated into the analysis. Thus, the Ctenophora-sister hypothesis can now be wholly rejected in favour of the traditional Porifera-sister scenario of animal evolution, wherein the animal ancestor did not possess key traits such as a nervous system, muscles or a mouth and gut.

Ctenophores as the sister group is now “wholly rejected”! I suspect that not all animal systematists would accept this hypothesis. I do, tentatively, but I don’t fully understand the complex methods of analyzing DNA data (they used 60 animal groups, 406 genes, and 88,384 DNA sites).  My view of these complex methods is the same one that my academic grandfather, Theodosius Dobzhansky, held towards the experts in mathematical population genetics (Dobzhansky was innumerate): “Papa knows best.”

For a fuller explication of the conflict, as well as an overview of animal evolution in general, you can’t do better than Matthew’s 2018 Discovery PROGRAM on the BBC. The controversy about sponges-first versus ctenophores-first starts at 17:45. This program is very good, involves interviews with a lot of different biologists, and should be very clear to the sentient layperson. Plus it’s only half an hour long. Spend this Sunday learning a bit about animal evolution!

Click on the screenshot to hear the show:


Redmond, A.K., and A. McLysaght 2021. Evidence for sponges as sister to all other animals from partitioned phylogenomics with mixture models and recoding. Nat Commun 12, 1783 (2021).

63 thoughts on “Are sponges the closest relatives of the rest of the animals?

  1. Are sponges the closest relatives of the rest of the animals?

    Glancing at my 22 year old son on the sofa watching TV…yeah…I buy that.

      1. Most Porifera I’ve seen go through sequential inhalant and exhalant phases, with the net absorption (or exorption(?)) rate being very low. Like a city, it’s commuting population and it’s resident population.
        Is 22 years long enough to tell the average? I ask, thinking of Zhou Enlai and the dubious comment about the French Revolution (of 1789, 1830, or 1848; let alone the other contenders for the capitalisation)?

  2. All interesting. I’m gonna study this.
    I always review The Situation in my Evolution class, with a note that it is a bit odd that of all things, taxonomists have substantially revised the animal evolutionary tree several times and cannot seem to agree on this very basal detail. Several of the other revisions chafe like hell. ((Really?? Flatworms are Protostomes? Insects are… Crustaceans??)). But I do accept those revisions now, albeit with a feeling that evolution toys with my heart.
    I also discuss this issue about Ctenophores vs sponges. For now I will continue to teach it as an as yet unresolved controversy; the mother of all polytomies, and wait for the dust to settle.
    And Placozoans! Don’t get me started. Simpler than sponges and lacking in real tissues, and yet genetically they show signs of once having muscle and nerve tissue. Were they some chunk of an Ediacaran animal skin that just did not die?

      1. But we can still have fun if we know the right answers, but cannot distinguish them from the wrong ones. Or, we don’t agree about which answers are wrong and which are right.
        If they have nothing else to lose sleep over, palaeontologists can wake up at god-awful o’clock and worry that the necessary evidence to decide (a palaeontological question) really has not been found because it has never been preserved. That goes through my mind as I stare, through rain-spotted spectacles, at the water running over billion-year old mudrocks seeking macroscopic fossils to accompany their (very rare) microfossils.

          1. “If we knew what we were doing, we couldn’t call it research.”
            Not sure whose that is. It sounds Haldanish. Or maybe Wheeler? Feynman?

      1. Agreed on both points. Flatworms as coelomate sisters was a better story (fortunately a few are still there iirc…acoels I think?). Similarly, I am finally getting around to accepting turtles as diapsids.

        1. Yes, that’s correct: the Acoelomorpha (without kidney-like structures called nephridia) are the sister group to the “coelomate” animals (which got a new name, the Nephrozoa), including a bunch of lineages like the rest of the flatworms that lack coeloms.

          1. Yeah we get set in our views and they are hard to change. The ctenophores-first idea seemed to have a lot of good genetic evidence over the last 10 years, but most of the people I know who teach invertebrate zoology refused to teach the idea. Or at best they would, ahem, teach the controversy. So I guess sometimes that conservatism is a good thing if it turns out this new analysis is correct and it was sponges first all along.

        2. Well, I’ve got my line (“We need another group of diapsids like we need another hole in our heads”).
          All I need to do is find a circumstance to use it.

  3. Great post!
    And I’ll definitely check out Matthew’s Discovery program.
    This discourse is close to a “concern” I have. Is it true that multicellular animal forms (as defined here) evolved only once in earth’s history? (Apparently this is not true for plants.). If so, this would seem to be a potential bottleneck in the path to “intelligent” life, and could impact estimates of the likelihood of intelligent life evolving elsewhere. Related to this are conjectures of how multicellular animal life could get started, which usually cite low-probably events like cells “swallowing” cells in order to create the needed internal structures. I’d appreciate any thoughts or corrections related to this question.

    1. All living multicellular animals share a multicellular ancestor, so evolved just once as far as we know. The same is true for plants (because fungi and the various algaes are not ‘plants’). Cells swallowing other cells is a hypothesis for the evolution of eukaryotic (nucleated) cells, not (afaik) multicellularity.

      1. This is an interesting and informative thread. Also, fun to see familiar ‘faces’ – or in this case, turtle avatars afoot.

      2. Although all animals are thought to have evolved from one multicellular ancestor, multicellularity itself is thought to have evolved many times . Mesomycetozoans, for example are not thought to have a multicellular ancestor with metazoans or fungi. Or, another example, just think of slime moulds, or plants for that matter.

        The symbiosis of archaea and bacteria forming the eukaryotic cell is more than a hypothesis. It is supported by such vast amounts of evidence we may call it a theory.

  4. So is this anything like the chicken/egg argument. If it gets down to which one do you want in the family photo I would go with the sponges. Simple is better I guess.

    1. The chicken-egg argument is a …. (searches for a simile – “straw man” isn’t really right) misleading way of thinking about things. Eggs go back at least as far as the origin of the amniotes (“reptiles” and all their descendents) about 250-270 million years ago, whereas chickens are modern birds (origin about 90 million years ago), birds (origin about 130 million years ago) or dinosaurs (origin about 200 million years ago). That creationists think it’s an effective challenge says more about their mathematics than their palaeontology.

        1. Oh, we can play the “every missing link creates a need for two new transitional forms” game all the way to the limit as baraminology tends to genealogy.

  5. I thought genetics settled this problem in favour of the Ctenophores (as being more closely related). How wrong can one be, not the systematics, but it being settled.

    It was always ‘obvious’ that sponges did not lose it as the sister group, the traditional view. It never made much sense, as our host mentions, sponges losing a nervous system, muscles or trans- gut, or worse, multiple origins of the complex set of nervous system, muscles and a trans-gut. I guess it is settled now.

    1. This is where I really regret not having my Psion palmtop any more. I spent months stunning myself into unconsciousness making notes from Margulis’ Five Kingdoms (largely, an attempt at describing every phylum of life on Earth), including some detailed notes on a couple of what the 1980s reported as very obscure tiny-tiny-tiny metazoa of just a few dozen cells, which between version 1 of the book and my reading (self-stunning) were identified as extremely degenerated arthropods (IIRC) which had lost

      a nervous system, muscles or trans- gut

      because they weren’t needed to live as parasites in the gills of cephalopods.
      If I’d still got a working Psion, I’d be able to tell you phylum name, origin, and references. But … dead Psion, lost work in dead formats, lesson learned. I’m actually running a backup of my data pile at this moment, for posting to the other end of the country.
      Metazoan phyla losing whole tissue systems is not a new idea. Change the environment, change the minimal systems needed to survive.

      1. Yes indeed, but most are parasites, bathing in their food as it were, not free-living animals like sponges.
        Trichoplax/placozoa is a different proposition, we do not even know how they live in the wild.
        Indeed DNA may give them an unexpected place, just like the coeloma-lacking plathelminthes.

        1. Sorry, but I was under the impression that the benthonic substrate-cruiser ecological position of Trichoplax was well established. I’m not sure about the new additions to the taxon.

  6. A very interesting read. Thank you. The terminology confuses me—“closest” and “sister group”. I am interpreting the meaning as sponges are the most ancestral group. Is that right?

  7. By a strange coincidence (?) WEIT reader Dom sent me the link to Matthew Cobb’s BBC radio programme yesterday and I’ve just finished listening to it. Fascinating, though well above my paygrade to express an opinion beyond the obvious appeal of the parsimonious “sponges first” scenario.

              1. A family in-joke. When they were little, our two oldest were always bickering over toys and so I nicknamed them “Grizzle and Snatch” after the protagonists of their favourite CBeebies show, Nuzzle and Scratch ( “We’re alpacas from the Andes/with hooves instead of handies”).

  8. I listened to the programme yesterday – Matthew shared it on Twitter- & looked at the paper but I am about as numerate as Dobzhansky though not as smart, so hoped you would have a butchers & go through it for us!

    PS readers may be interested in this Observer article on Tardigrades & how studying their ability to survive as tuns – dried husks – may help delivery & storage of viral vaccines

    1. I’m glad you credited that article to The Observer, as we’ve just had an animated family discussion about the importance of the distinction between it and The Guardian. I seem to have lost 1 against 5, (assuming that Marcus Clawrelius wasn’t on my side).

      1. (assuming that Marcus Clawrelius wasn’t on my side)

        Were you the one with the food can opener? The externally-adapted opposable thumbs of Felis domesticus

      2. I am not a Guardian reader. I AM an Observer reader & I deplore the way the Guardian has eviscerated the great paper & Guardianized it…

  9. I have a question about whether or not the retinas of cephalopods and mammals are reversed because of the developmental consequences involved in being protostomes and deuterostomes. Is it possible that their common ancestor did not depend on having a retina facing one way or the other? If so, how did the two evolutionary paths constrain the subsequently developed anatomy? I’m only a biology undergrad. Answering these questions myself has been difficult, and none of my professors have been able to provide an answer.

    1. There doesn’t seem to be a protostome-like and a deuterostome-like eye, and it’s likely that those animal lineages are not constrained to develop one specific type of eye.

      Some scallops (molluscs, close to cephalopods) have image-forming eyes with a two-layered retina. The light-focusing layer in that eye is a mirror at the back of the eye behind the retina. Although there is a lens in front of the retina, it isn’t the light-focusing structure.

      This is a good reference
      DOI: 10.1126/science.aam9506

      The money quote:

      “The upper (distal) retina is composed of ciliated photoreceptor bodies and resembles a vertebrate photoreceptor, whereas the lower (proximal) retina is composed of microvilli and is similar to the photoreceptor cells of other invertebrates (7 ). Calculations show that a simple spherical mirror would have its focal point in the distal retina (3), suggesting that only the distal retina is involved in image formation.”

      So the upper mammal-like layer is the image-forming part of the scallop eye. Presumably the lower or proximal cephalopod-like layer made of microvilli does something else.

      Some jellyfish also have image-forming eyes with a lens, retina, and the ability to point the eye at different parts of the visual field.

      I don’t work on this kind of comparative developmental biology, but my understanding is that the common ancestor of all those lineages probably had eyes that could detect light direction and intensity (all of those animal clades have a similar suite of genes that are expressed in eye development), but probably not form an image. The oldest fossils that would be old enough to be the common ancestors of animals with image-forming eyes (the sister group to the sponges) don’t seem to have had image-forming eyes. But my knowledge of the developmental genetics and the palaeontology is out of date and I could be wrong.

      1. I’m reminded of Chapter 5 in Dawkins’ “Climbing moMount Improbable”:

        There is another kind of animal, however, that definitely uses a bona fide curved mirror to form an image, albeit it has a lens to help. Once again, it was discovered by that King Midas of animal eye research, Michael Land. The animal is the scallop. The photograph in Figure 5.18c is an enlargement of.a small piece (two shell-corrugations in width) of the gap of one of these bivalves. Between the shell and the tentacles is a row of dozens of little eyes. Each eye forms an image, using a curved mirror which lies well behind the retina. It is this mirror that causes each eye to glow like a tiny blue or green pearl. In section, the eye looks like Figure 5.18d. As I mentioned, there is a lens as well as a mirror, and Ill come back to this. The retina is the whole grayish area lying between the lens and the curved mirror. The part of the retina which sees the sharp image projected by the mirror is the portion tightly abutting the back of the lens. That image is upside-down and it is formed by rays reflected backwards by the mirror.

        So, why is there a lens at all? Spherical mirrors like this one are subject to a particular kind of distortion called spherical aberration. A famous design of reflecting telescope, the Schmidt, overcomes the problem by a cunning combination of lens and mirror. Scallop eyes seem to solve the problem in a slightly different way. Spherical aberration can theoretically be overcome by a special kind of lens whose shape is called a Cartesian oval. Figure 5.18e is a diagram of a theoretically ideal Cartesian oval. Now look again at. the profile of the actual lens of the scallop eye (Figure 5.18d). On the basis of the striking resemblance, Professor Land suggests that the lens is there as a corrector for the spherical aberration of the mirror which is the main image-forming device.

        You’ll see the images when you click in the text itself

    2. Eyes themselves originated at least 40 times in metazoa, although they have some commonalities in the basics, which gives us the idea that the ultimate origin of light perception may have originated only once in metazoa:
      – all use a Retinol-like pigment (chromophore)
      – this pigment is bound to opsin (protein) in all (AFAIK)
      – the activated opsin binds to G-protein starting a signal cascade
      – in all the Pax genes appears to be involved (among others)

      There are 2 types of photoreceptors though: rhabdomeric (microvilli) (using Phospholipase C (PLC) and Inositol phosphate (IP3) in their cascade, and ciliary (Phosphodiesterase (PDE) and C-Guanosyl monophosphate), so photoreceptors originated at least twice, probably in the same ancestor.
      The proterstomes (predominantly) use rhabdomeric photoreceptors as photoreceptors, while the deuterostomes use the ciliary photoreceptors as the actual photoreceptor. So there might be a kind of proterostome-deuterostome dichotomy after all.
      Note that -as Mike pointed out- the vertebrates use their rhabdomeric PRcells higher up in the chain as horizontal, amracine and retinal ganglion cells.
      you can download the pdf for free.

      A brilliant lecture by Dan Eric Nilsson on YouTube:

      Also recommended: “Animal Eyes” by Michael F. Land and Dan-Eric Nilsson. It describes how the different eyes actually work physically.

  10. I wish researchers soon rich a rock-solid consensus opinion about this, so that I stop feeling guilty when browsing these slides of my teaching presentations.

  11. That was a heated debate that came to a head 2018ish, for the reasons this paper mention of site-heterogeneous models as well as better methods to compare models. (Site-heterogeneous models account for differences in evolution rates.)

    I can’t remember the papers right now.

    [But FWIW here is one of them. though I can’t vouch for it:

    We show that data-recoding methods [13, 14, 15] reduce compositional heterogeneity in these datasets and that models accommodating site-specific amino acid preferences can better describe the recoded datasets. Increased model adequacy is associated with significant topological changes in support of Porifera-sister. ]

    But one paper was sufficiently good that it prompted a short seminar during a weekly meeting at the lab I was at. The conclusion was that Porifera-sister was the better model.

    And I haven’t seen anything that convincingly changed that since then.

    I will have a look and listen, thanks!

  12. As Mike says, eye structure does not seem to follow the proterostome-deuterostome dichotomy.
    Another useful way to consider things is to examine the variety of eye structures within one phylum. Yesterday (two days ago? “-ish”) I was thinking about the eyes of trilobites. One phylum of arthropods (super-phylum Ecdysozoa), the tremendously important Palaeozoic trilobites had three very different types of element in their (mostly) compound eyes, while there are nearly as many detailed types of eyes (each composed of only one of the three types of element) as there were genera of trilobites. “Nearly”, because a significant number of genera lost their eyes, at many points within the trilobite family tree, and another significant number of trilobites had simple eyes composed of only one element (of whichever type) on each side. Within the trilobites (arthropods, including insects), the details of eye structure seem quite labile, without paying attention to the question of protostome-deuterostome bauplan.
    Another example : Nautilus (an extant class (order? fairly high level taxon anyway) of the cephalopod molluscs has a very simple “pinhole camera” eye, while other members of the cephalopod molluscs, the decapoda (squids, cuttlefish) and octopoda (octopodes, octopussies) have a much more complex eye with, as you say, externally-wired retinae. And with lenses – which are absent in the Nautiloidae.
    Closer to home – all vertebrates share very similar eyes. But I’ll bet a pint of good beer that the detailed structure of the pineal gland and it’s correlates are also quite variable. In some taxa of the amniote terrestrial tetrapods (that’s us, dino-chickens, but not frogs or sharks) the pineal opens to the surface of the skull and is described as a “third eye”. The extant member of this group, New Zealand’s tuatara would be an interesting species to look at whenever testing evolutionary ideas – it’s a taxon with no significant fossil record this side of the late Triassic (~190 Myr ago).

  13. A.) The common ancestor of all animals had nervous systems and muscles and a gut, which persist in all groups but the sponges,

    I don’t think the Placozoa have guts. Well, not the one I studied in textbooks ages ago. The more recently recognised two genera – I feel the need to study. And to watch Matthew’s program.

    1. Oh, that’s nice – it’s coming down as an MP3. Very user-friendly. No need to re-download it to listen a seventh time over a particular point. Just 23MB – it would have fitted onto my first hard drive. Just. Off to listen.

  14. For those who are into the palaeontology of the Ediacaran – and let’s face it, who in their right mind isn’t – my occasional bug-watching cow-orker Dave Marshall has several hours more about it on his Palaeocast podcasts (produced for the Palaeontological Association), specifically episodes 5, 40-odd (he’s less than consistent about naming conventions), 50 (IIRC, that includes some of the work discussed by one of Matthew’s interviewees), 70-odd, 84, 102(that is more about the “small shelly fossils” interval at the very base of the Cambrian, when the hard-to-comprehend life forms of the Ediacaran was incrementally replaced by forms more easily recognised as relatives of modern (or at least, Palaeozoic) life forms), and 104.
    Ediacaran faunas make appearances in several other episodes – particularly Dave’s “audio postcards from palaeontological conference” type ensemble episodes, and I’m sure that you’ll find a good few hours of educational entertainment there.
    Ohh, palaeo pussy cats! “Episode 121/122: Dietary ecology of Smilodon fatalis” – a double-length interview/ discussion – 2 hours or so instead of one – on what sabre-tooth kitties ate.
    Someone mentioned running a formal course in evolution, and an offshoot of Palaeocast is the V-NHM – a Vitrual Natural History Museum of online material for educational use. Might be useful resources. More PalAss funding.

Leave a Reply