Oldest Animals Discovered! Or are they..?
- By olderthanthehills
- On 19/10/2018
- Comments (0)
The media this week (e.g. above, New Scientist, and The Times) is all over a paper by Zumberge et al. in Nature Ecology and Evolution, describing what is claimed to be the oldest evidence for crown-group animals: specifically, chemical fossils (biomarkers) from very early sponges. Due to my interest in this area, I was asked to write an editorial about it in the same journal (with Benjamin Nettersheim, a biomarker specialist), which has sometimes had snippets quoted along with the news articles. The news stories, though, always need to condense and simplify, so here's a fuller version of the discussion, and my responses to the responses...
The paper: https://www.nature.com/articles/s41559-018-0702-4/metrics
Example news story: https://www.newsweek.com/oldest-animal-life-discovered-635-million-year-old-fossil-1171769
I'm going to give an outline recapping the main points, and then address a few of the things arising from paper and its reporting, just to make my position clear. It's potentially a really major discovery in its implications, but there are a lot of complications as well... which is why sound-bites just don't do the story justice. So, the story revolves around the discovery by Zumberge, Gordon Love and colleagues of a distinctive fossilised molecule called 26-methylstigmastane (26-mes). This is the stable remains of 26-methylstigmasterol, one of a class of compounds (sterols, a group of steroids) that includes cholesterol. Sterols include a wide range of subtle variations, such as 26-mes, but some groups of organisms are more diverse in their sterols than others. In particular, sponges appear to produce as many different types as are known from all other groups combined.
This is where it gets really interesting, because sponges are thought to be the earliest (or almost the earliest) group of living animals to branch from the tree, and therefore we can use distinctive sterane fossils to track when animals first appeared, even in the absence of recognisable body fossils. The work follows previous work by the same group on another sterol, 24-ipc, which has been claimed to be a biomarker for demosponges, the moost abundant and diverse living class of sponges. This has been debated, though, for several reasons. Firstly, it is known to occur also in one group of algae (albeit one that seems to have developed the ability much later), and secondly that the abundance of the biomarker declines dramatically just as sponges (and other obvious animals) enter the fossil record and diversify.
The new results seem to solve one problem (26-mes isn't known from algae, or indeed from any other group besides demosponges), but not the other: it's another chemical signal that is abundant in the late Precambrian (including the Snowball Earth interval), but which declines into the Cambrian. So what are we to make of the new data? Firstly, I want to say that I have no problem with the results themselves (it's not my area, but still...), and I agree it's an extremely provocative and interesting discovery. That's not really up for argument. What's less clear is the implications.
Is 26-mes definitely from sponges?
This entire argument revolves around the fact that among living organisms, 26-mes is currently known only from certain groups of demosponges. As Jonny Antcliffe (a vociferous foe of biomarkers!) has said, 99% of everything is extinct, so how can we really say anything about what other groups evolved it? This is a valid argument, but there are also some counter-arguments. For example, the ability to make complex sterols of any sort is quite rare. The ability to construct stigmasterols, or propylcholesterols, or whatever, is generally restricted to particular groups, and many types of organisms don't seem to make many interesting sterols at all. Extinct members of these groups are not likely to have been able to either, so we're looking at the level of major groups. Of course, there could well be completely extinict major groups of eukaryotes... and that's the real problem: we can never be 100% sure that our biomarkers are from what we think they are.
The key point with this story, though (and largely missed by most of the news articles), is that it's a second demosponge biomarker, from the same age and rocks. This sterol occurs in different living demosponges to 24-ipc, but it's all the same class. The argument goes that if the odds of something else evolving one such compound convergently (independently) are quite low, then the odds of evolving two of them must be minute. Ergo, these really are from sponges.
That sounds pretty compelling. But remember... complex sterol production seems to run together, so being able to create one may make it easier to create others. And worse than that, almost any sterol produced by anything has a counterpart in a demosponge. There's actually nothing all that special about 26-mes, other than that it turned up in this work. If we were looking at the only two fancy sterols produced by a group of animals (and uniquely by them), that would be a lot more convincing. In this case, though, almost any complex sterol that was evolved at that time (by whatever) would be matched up to sponges... because they do everything. It's interesting that a different group has previously reported another complex sterane, dubbed cryostane, from even older sediments in the Snowball Earth period, which has similarities to some produced by sponges but is not known from any living ones.
Added onto all that is the question of how well we know the sterols produced even by living eukaryotes. Animals, yes, we have a pretty good idea of (we think), but many other groups of eukaryotes have been studied only cursorily. Until we really know in detail which groups can produce what, and can also make a good argument that other groups could not have evolved the ability (for specific biochemical reasons), the sponge interpretation has to remain a hypothesis and not a fact.
A random bit of washed-up bath sponge: one of the types with a purely organic skeleton.
So where are the fossils?
Yes, I'm mainly a palaeontologist, so this preoccupies me rather a lot. I'll try to be moderately brief! Basically, if these steranes represent demosponges (the authors showed that these complex sterols are basically not found in their closest relatives or other sponge groups), then this is a group with a good fossil record, going all the way back to the early Cambrian... and then they vanish. As shown by a couple of recent reviews (one by myself and Lucy Muir, and another by Jonathan Antcliffe and colleagues), there are simply no viable sponges in the Precambrian fossil record. Almost all the claimed examples are clearly something else, and the only remaining possibilities are too vague to be able to say anything useful about their origins. So... where are they?
The first thing palaeontologists look for are chemical biases: something envionmental that could explain their absence. It is also true that many groups of demosponges have very small spicules (made of opal-A, a form of silica) or only organic skeletons. The organic skeletons preserve quite well in Cambrian deposits that also preserve other delicate organic remains like algae, and are well known from the Burgess Shale. The spicules are harder to fossilize, because silica saturation in seawater is quite low, and these minute spicules can therefore dissolve relatively quickly. Go back through the fossil record, though, and seawater silica levels were higher, leading to better preservation of spicules, so that there are numerous Cambrian and Ordovician deposits where limestones have been dissolved to yield beautifully preserved microscopic spicules... including distinctive demosponges forms. These spicules can also be preserved (and are, abundantly) in black mudstones, through replacement by pyrite (iron sulfide).
In the late Precambrian, the fossil record is admittedly rather weird... but what do we know about it? Well, there are actually quite a large number of Burgess Shale-type deposits: organic preservation in mudstones. These preserve exclusively algae and microbial structures, together with a few ambiguous fossils that no-one is quite sure of... but no organic skeletons of demosponges. There are also pyritised fossil assemblages, with organic remains replaced by pyrite in black mudstones. Not a single spicule has ever been found.
When we look at the shallow-water deposits, we see a lot of cherts together with the limestones and sandstones. This is because seawater silica concentration was even higher in the Precambrian than the Cambrian, a factor which helped the preservation of the soft-bodied (and deeply enigmatic) Ediacara Biota. Why were silica levels so high? Because silica-secreting organisms (read sponges, together with radiolarians) had not yet become abundant enough to draw much of it out of solution. Therefore, silica spicules should be preserved much more widely in these rocks than they are in the Cambrian. Instead, we find yet more absence. Limestones have been dissolved by the bucket-load, but no convincing spicules have fallen out. Cherts are etched and sectioned in vast quantities, and yet still the spicules are just not there... until somewhere around the base of the Cambrian, when the chances of their preservation drop off dramatically.
Skeleton-less sponge ancestors?
So, the obvious way out of this is to posit that the demosponges were there... but their skeletons weren't. They just hadn't evolved them yet, and therefore there is nothing to find. This ties in with traditional view of each class of sponges arising from an amorphous, blobby ancestor, and is also Gordon Love's suggestion - at least, as quoted in New Scientist! However, it also ignores the fossil record completely, and that's something that we really mustn't do.
The origin of modern demosponges in actually visible in the fossil record, as we've been describing in a series of papers over the past few years (with more to come). The modern class is split into two major groups, one with spicule (plus organic) skeletons and one with just organic (plus a few odd encrusting forms that have lost even that). Trace them back, and you come to Vauxia and Hazelia in the Cambrian: a complex of species where we can actually see the split happening. This includes, for example, the last traces of spicules in the otherwise organic skeleton of Vauxia bellula. Demosponges didn't originate from a blobby, skeleton-less ancestor; their ancestors were delicate, thin-walled vases, with a spicule and organic fibre skeleton.
What about the stem group of demosponges? Perhaps they were non-skeletal (which practically requires them to be encrusting, from a functional perspective), and the complex sterols evolved deeper in their history. For that, we need to look at their next-nearest relatives, the hexactinellids or 'glass sponges'. These also share silica spicules, secreted in the same way around an organic axial filament, but their spicules are mostly of a different shape three-dimensional crosses known as hexactins.
Pheronema sp.: a modern hexactinellid
Hexactinellids are generally much more conservative in their body form than demosponges, and usually have radial symmetry - like early fossil sponges generally, and the earliest hazeliids. One fossil sponge, Cyathophycus, that is traditionally assigned to hexactinellids, has hexagonal axial canals: a derived feature of demosponges. This genus also has an inner skeletal layer of minute monaxons, in a hazeliid-like array. A recently described fossil, Conciliospongia, takes this one step further with a full hazeliid skeleton that includes some small hexactins. In combination with the similarity of the secretion mechanism of the spicules in the two classes, the fossils leave no doubt that the stem lineage of demosponges did indeed secrete a siliceous spicule skeleton, and that the earliest stem lineage had larger and more obvious spicules than the crown group.
Worse than that, though, is that the earlier ancestors of demosponges and hexactinellids (crown-group Silicea) also had spciules, probably going back to the stem group of Porifera as a whole. There are a large number of problematic early fossils that cannot be included in the living classes, and these form the bulk of the Cambrian sponge record. In some cases their spicules are several (or even tens of) centimetres long, with not only silica, buut also a thick organic layer, and sometimes also calcite. These spicules not only make preservation easier due to their size, but offer multiple chemical pathways to preserve them through. Such spicules are found in Cambrian rocks in mudstones, limestones, siltstones, and sandstones. If demosponges, a derived class, were present in the Cryogenian, then all of these must have been there as well... and the fossil record has run out of excuses for hiding them.
The first thing to say is that I've got nothing against the work itself, which looks to be careful and sound (although I'd have very much liked to see some evidence that they'd looked for fossils in the same samples the biomarkers came from). No, my problem is interpretational. At the moment, we have two lines of evidence that are both powerful, self-consistent and compelling in isolation... and utterly contradictory. There's no middle ground here; either the fossils or the biomarkers (or both!) are fundamentally flawed and wrong. That's not a criticism of whoever is ultimately being misled, but a statement of fact that reflects a fascinating riddle of modern science.
What I don't like to see is certainty where none can truthfully be found. I've seen phrases like, "chemistry doesn't lie" (relating to a different recent biomarker story, as it happens) along with attempts to dismiss the fossil record problems by saying these were non-skeletal, soft-bodied stem-group demosponges (New Scientist article)... but they're just rhetoric. In this time of interdisciplinary research, we need to be taking a lot more care to avoid dismissing another field just because it disagrees with the evidence from our own. It's easy to dismiss a subject one doesn't know much about, but palaeontology has just as much of a voice as chemistry in this arena.
In the end, we're left with a fascinating, and even riveting problem. Molecular clocks broadly agree with the biomarkers in predicting deep (or even deeper) origins for sponges, but these models are also hit the same or even worse problems when comparing with the fossil record (a story for another time). It's time for us all to join forces to try to tackle the problem with all the tools at our disposal. Competition at the expense of realistic assessments of the evidence is not in anyone's interest, or that of the truth.