Planktonic Paradise on Slushball Earth:
a scenario for the metazoan radiation

by Richard Cowen

First Web version written May 26, 2000.
Mini-essay updated February 25, 2001 to include work by Baum and Crowley, and refinement of my ideas.

In trying to reconstruct the origin and radiation of metazoans over the past forty years, biologists have relied mostly on analogy and homology with the anatomy, functional ecology, and development of living animals. Now we have new fossils from the Late Proterozoic and early Cambrian; we understand the time-frame much more precisely; we have rapidly increasing information on the genetic sequences of living metazoans; and our knowledge of the physical changes on Earth has grown enormously. Synthesizing current data, I present here a new scenario for the metazoan radiation.

We do not know when the first metazoans evolved: obviously, it was after the protists did. The fossil record is barren of unequivocal metazoans until the Ediacaran/Vendian organisms at perhaps 565 Ma. There is, however, tenuous evidence of a much earlier origin and radiation of metazoans. "Molecular clock" arguments based on genetic analysis of living metazoans suggest a divergence of basal metazoans perhaps as early as 900 Ma and perhaps as late as 650 Ma. These dates depend on the assumption of clock-like molecular change (demonstrably untrue), and on "divergence dates" between metazoan taxa that are imprecisely known. In any case, it is reasonable that metazoans should have existed before we have fossil evidence for them, especially if they were small and soft-bodied.

Anatomical and genetic evidence alike separates the earliest metazoans into three clades that represent the ancestors of sponges, cnidarians, and bilaterian metazoans, with ctenophores as the likely sister group of bilaterians. In turn, early bilaterians seem to have differentiated early into three clades whose principal pragmatic diagnostic characters are based on growth style of the adult. Ecdysozoans are characterized by growth that includes molting of the epidermis; lophotrochozoans are characterized by accretionary growth of epidermal structures; and deuterostomes are characterized by growth that includes resorption and remodelling of internal support structures. (Recent research is summarized by Adouette et al. [2000]). Representatives of all these major metazoan clades have been recognized among Ediacaran faunas, and many derived subclades occur in the early Cambrian fossil record after the so-called "Cambrian explosion". In addition, Ediacaran faunas include organisms that are extinct and/or of uncertain affinity. (Recent research is summarized by Conway Morris [2000].) Therefore, a convincing scenario would include a Proterozoic but pre-Ediacaran radiation, and an explanation for the lack of a fossil record of it.

Ediacaran and Cambrian faunas occur after major glaciations centered around 700-600 Ma. The glaciations seem to have been extraordinarily extensive compared with Phanerozoic "ice ages", so that the Late Proterozoic interval that includes them is coming to be known as the "Cryogenic" and the scenario that attempts to explain their wide latitudinal distribution is usually known as "Snowball Earth". Furthermore, it has been suggested that there is a causal link between Snowball Earth and the metazoan radiation: that the dramatic physical events on Earth promoted the radiation (Hoffman et al. 1998; Hoffman and Schrag 2000).

I argue here that the scenario of Snowball Earth is more extreme than the evidence suggests, and that the linkage suggested between Snowball Earth and the metazoan radiation is highly unlikely. Instead, we argue for a milder Slushball Earth that was nevertheless a significant influence on the history of life. Rather than squeezing metazoans almost to extinction, we argue that Slushball Earth in fact provided an environment that was uniquely benign for plankton. In that planktonic paradise, eukaryotic life passed through an evolutionary filter that promoted the radiation of diverse clades of planktonic micrometazoans. When Slushball Earth ended, those micrometazoans evolved rapidly, perhaps explosively, into the larger and more ecologically catholic macrometazoans that we see in the Ediacaran and Cambrian fossil record.

Snowball Earth

Recently, evidence has been building that Earth went through a series of dramatic cold periods late in the Proterozoic, from perhaps 750 Ma to 600 Ma. Many deposits from this period contain glacial debris, and many of them occur in regions that are reliably reconstructed near the Equator at the time. These deposits imply massive and widespread glaciation.

One scenario that seeks to explain these deposits is called "Snowball Earth". The summary that follows is based on explicit recent versions of the model by Hoffman et al. (1998) and Hoffman and Schrag (2000). Snowball Earth began because Earth's continents were arranged along the equator in the Late Proterozoic, either as isolated masses, or as the supercontinent Rodinia. In any case, this left Earth with very large oceans covering middle and high latitudes. In such oceans, as in the Southern Ocean today, the polar seas were isolated from warm tropical waters by the pattern of surface currents, and developed large floating ice sheets. As these ice sheets spread, they reflected more and more solar radiation into space in a runaway albedo cooling reaction that stopped only when the sea was covered with ice, with surface temperatures down around -40 C. The surface of the ocean was frozen all the way to the Equator (except where volcanoes existed).

As polar ice caps spread, photosynthesis was choked off, and most of the life in the oceans died off. In fact, the only surviving life would have been around seafloor hot vents, and (perhaps) in the surface ice. As glaciers grew on and ground into the equatorial continents, they left behind physical traces of glaciation that we can now see preserved in the sedimentary record.

While the ice was advancing, the equatorial continents continued to use up atmospheric carbon dioxide as their silicate rocks weathered, with the result that by the time the snowball was complete, the atmosphere had little carbon dioxide left to act as a greenhouse gas. That would, you would think, seal the fate of Earth, locked permanently into a snowball.

Volcanoes continued to erupt, however, putting carbon dioxide back into the atmosphere, until there was once again enough carbon dioxide to trap solar heat and melt the ice. However, calculations suggest it would have taken an enormous amount of carbon dioxide to break the grip of Snowball Earth. The ice did not melt until volcanoes had erupted around 350 times more carbon dioxide than there is in our present atmosphere.

The ice then melted quickly, but the enormous reservoir of carbon dioxide in the atmosphere rocketed the whole Earth directly into a "greenhouse" hot period, with temperatures averaging around 50 C (over 120 F). Tremendous (acid) rains then acted on the sterilized continents, the ocean was flooded with carbonate, and thick limestones were deposited very quickly on top of the glaciogenic rocks. Finally, weathering and photosynthesis brought down carbon dioxide levels, and the world recovered biologically. However, the geographic set-up that had begun the Snowball Earth cycle was still present, so the cycle then repeated itself. The Snowball Earth scenario envisages at least four closely spaced glaciations.

So, then, a Snowball Earth cycle was begun by geography (and oceanography), and ended by volcanic eruption.

Snowball Earth proponents suggests that the great climate swings might have favored the evolution of early metazoans. The link is not clear: there are vague references to small isolated patches of organisms challenged by environmental change, with rapid evolution among the few (single-celled) survivors. During Snowball Earth episodes, Earth's organisms would have gone through boom-and-bust cycles of enormous range. Extinctions must have been dramatic. The ice essentially cut off all photosynthesis on the Earth, yielding an anoxic ocean. Iron deposition during the glaciation supports this component of the scenario. Also, the carbon isotope fluctuations across Snowball Earth are large, beginning as normal (organic carbon:total carbon = 1:2), then moving to basically non-biological (ratio = 0) as the carbonates were deposited.

In the end, the break-up of the tropical continents moved major land masses out of the tropics, and the geography of the Earth evolved away from a Snowball state.

Problems With Snowball Earth.

I think there are problems with the extreme scenario of Snowball Earth.

1. Climatology.
The Snowball Earth scenario is based on a rather simple climatological model originally proposed by Budyko (reference?). It is this model which projects the extraordinarily low temperatures, the global ice cover over the oceans, and, by extension, the tremendous build-up of carbon dioxide required to break the glacial grip on Earth. Newer, better climatic models are beginning to be applied to the Late Proterozoic geography of Earth (for example, Hyde et al. 2000; Baum and Crowley 2001), and they suggest much milder climatic conditions. Obviously, this removes the requirement for desperately low global temperatures, and for heroic carbon dioxide levels in the atmosphere to break the grip of the glaciations. Baum and Crowley (2001), for example, have GCM runs which allow for stable equilibrium of Earth's climate, with the physical presence of low-latitude continental ice-sheets and floating sea-ice over much of the world's oceans. But these models also show stable open tropical waters (between ~25N and ~25S), representing about 40% of today's open ocean surface, at cool to mild temperatures (up to 10C at the Equator). Yet the atmosphere has only 2.5 times today's carbon dioxide, and the stability of the model implies that any small rise in those levels will easily revert conditions to "Earth normal".

2. Sedimentology.
The Proterozoic glacial sediments include evidence of dropstones, falling from melting icebergs into soft sediment. Such icebergs can only have come from erosive glaciers on land, and must be floating in the ocean. This is incompatible with the extreme Snowball Earth model in which the entire global ocean is frozen. This would have literally dried up the air masses that fed snowfall on the continents, and major mountain glaciers could not have formed. Today's big Antarctic ice shelves are formed from glacial flow, not formed in situ on the sea surface. Drop stones are therefore evidence for significant open water near the Proterozoic continents, consistent with Baum and Crowley's scenario, and this argues against a global sea-ice barrier separating oceans from atmosphere in the Snowball scenario.

3. Geochemistry.
If most or all of the ocean was covered with ice, photosynthesis would have been drastically reduced, but there still would have been potentially massive populations of anoxic bacteria, including methanogens thriving on organic matter in the seafloor sediments, and on cold- and hot-water vents. In and on the seafloor, these organisms would be impervious to Snowball Earth conditions unless the ocean froze all the way to the bottom.

So the Snowball Earth ocean (extreme version or not) would have had a potentially enormous amount of methane in the water and in the sediments. Katz et al. (1999) provided good evidence for methane venting on a very large scale late in the Paleocene, around 55 Ma, enough to raise world oceanic temperatures by several degrees. Surely this parallel is enough to think that methane can't be ignored in Snowball Earth models.

Suess et al. (1999) have provided a good analogy for a Snowball methane generator. The Sea of Okhotsk is a marginal basin on the eastern coast of Siberia. Slightly less saline than the main ocean and lying in a dramatically seasonal climate, it is covered by floating ice about 7 months a year. Methane is trapped under the winter ice and released through cracks the next summer. In addition, there are huge deposits of methane gel (clathrate) in the seafloor sediments in the Sea of Okhotsk. Suess et al. mention the recent discovery of methane plumes 500 meters high billowing out of methane seeps; and there are mineralized chimneys suggesting that these sites are not ephemeral.

If we extrapolate the Okhotsk methane flux to an area compatible with Proterozoic geography, it is clear that there would have been a huge methane build-up under the sea ice in the oceans. It remains to be guessed whether the methane would have been released in frequent small emissions, or in occasionally huge bursts. But it is likely that the methane acted as another mitigating factor to ameliorate the climate of Snowball Earth, as well as the extreme chemistry imputed to its atmosphere.

Carbon isotope fluctuations across Snowball Earth are large, beginning as normal (organic carbon: total carbon = 1:2), then moving to basically non-biological (ratio = 0) as the carbonates were deposited.

Could the Snowball Earth carbon isotope anomaly pattern be ascribed to methane release? It was the carbon isotope signature that led to the Snowball Earth scenario calling for a virtually complete cessation of biological production during and after the glaciation. If that isotope signature has a large methane component in it, then perhaps we do not need to call for a complete shut-off of photosynthesis.

Methane seeps are associated with carbonates. Massive global release of methane could set off massive global carbonate deposition. So at least some of the carbonate caps of Snowball Earth could have had a methane assist. Now, it's an oxidative reaction, as I understand it. What does this mean? Maybe that the shallow ocean around the continents wasn't as totally anoxic as the Snowball Earth scenario would suggest? And that, in turn, would mean that some eukaryotes would have been just fine? And that someone, somewhere, was churning out oxygen in photosynthesis?

4. Physical Conditions Around The Continents.
Jere Lipps (1999 AGU abstract) criticised the Snowball Earth concept, based on his research in Antarctica. He pointed out that there are always intermittent lanes between ice blocks as currents, winds, and other stresses move floating ice. Lipps and his students have SCUBA dived through holes blasted with dynamite through the ice, and found that a lot of life, including large active metazoans, flourishes under the ice shelves of Antarctica. It is too simplistic to argue that floating ice could seal off the atmosphere from the ocean. See a later popular article describing a dive under the ice shelf.

Vincent et al. (2000) argue much the same way as Jere Lipps has done, this time based on the life that flourished in the seasonally wet surface water on an ice shelf in the Canadian Arctic. Here is the abstract of that paper from the Springer Verlag site.

5. Biological Conditions in the Cryogenic Ocean
In the Snowball scenario, light does not penetrate the thick, continuous, permanent floating ice that covers the world ocean. Photosynthesis is shut off, and the oceanic waters become anoxic. Metazoans and eukaryotes, which require oxygen, are largely wiped out. The Snowball Earth scenario envisages a few eukaryotes surviving in shallow meltwater pools in the surface of the ice sheets, or in oases around volcanoes. Snowball Earth proponents suggests that the great climate swings might have favored the origin and evolution of early metazoans. The link is not clear: there are vague references to small isolated patches of organisms challenged by environmental change, with rapid evolution among the few (single-celled) survivors. It is not clear how this scenario could reasonably have led to rapid evolution among the few stressed survivors in such extreme conditions, though it is easy to envisage massive extinction.

Slushball Earth

Evidence thus suggests a Slushball Earth rather than a Snowball Earth. Floating ice would have covered 60% of the ocean, but this left a lot of open water in the tropics. Global surface temperatures may have equilibrated close to 0C rather than the -50C of the Snowball model. Slushball Earth could have been generated and ended by moderate changes in carbon dioxide (or methane) levels, and would have been followed by a short and moderate greenhouse period rather than an extraordinary one. Methane could have contributed to carbonate deposition and carbon isotope shifts after glaciation, rather than drawing only on volcanogenic carbon dioxide. Most important in terms of evolution and paleontology, Slushball Earth could have acted to nurture the origin and early evolution of metazoans.

1. Conditions in the Slushball Ocean.
Light would not penetrate thick, continuous, permanent floating ice, and the ocean water below the ice sheets would quickly have become anoxic. Eukaryotes and metazoans (if any existed at the time) would have been wiped out from ocean waters and from the ocean floor, which would revert to an Archaean biology of anaerobic bacteria and archaea. Any biological productivity there would have been based on chemosynthesis.

However, conditions in open water in the tropics would have been radically different. Here there would have been little or no seasonal fluctuation in climate. Active erosion by mountain glaciers on the equatorial continents would have provided a steady year-round supply of nutrients to ice shelves along the coastline and, via icebergs, to the surrounding waters. As in Pleistocene glaciations, iron enrichment from dust would have been an importsnt supplement to "normal" nutreit supply. Solar radiation would have been uniform and intense. Within ocean waters, there would have been active vertical transport and corresponding upwelling of nutrient-rich water in the open equatorial areas. As an analogy, I suggest the perennial upwelling in the Southern Convergence today, or along the Equator; but Slushball upwelling might have been more intense because the area of available open water would have been small in comparison to the area of putative downwelling.

2. The Ecology of Slushball Earth
The nutrient-rich surface equatorial waters of Slushball Earth would have supported dramatic year-round productivity, uninterrupted by the seasonal darkness that cuts down today's Antarctic productivity in the winter months. Fall-out from the surface productivity would have driven the equatorial seafloor to anoxia, no matter how oxygen-rich the surface layers were. This is compatible with evidence that iron-rich sediments were deposited on the sea floor during Proterozoic glacial periods.

The surface equatorial waters of Slushball Earth would have been a planktonic paradise, whose only analogy in today's world might be the upwelling area off the coast of Peru; however, there were no Proterozoic El Nio phenomena to bring occasional catastrophe. In particular, the extraordinary, and permanent, productivity along the Slushball Equator would have provided an ideal setting for the evolution of micrometazoans, small, effective metazoan predators on the surface plankton. So Slushball planktonic communities could have included a rich and varied component of the earliest metazoans. How would that work?

Snowball Earth worries me because of the argument that calls for catastrophe to trigger the metazoan revolution. That doesn't happen: major crises cause major extinctions. It's not clear to me that repeatedly wiping the ocean free of oxygen is the way to foster the evolution of metazoans. Sure, plenty of bacteria, especially anoxic ones, would thrive; but the essence of eukaryotes is that they need oxygen for respiration. The eukaryotic survivors of a snowball Earth would have been few indeed.

I can imagine that AFTER one or many snowballs, single-celled eukaryotes would have found plenty of food on the ocean floors: organic mud and bacteria. Then I could imagine these single-celled eukaryotes rapidly evolving into animals that loved such conditions: sponges (filtering bacteria), cnidarians (which are essentially built of thin sheets of tissue that absorb organic molecules), and worms (eating mud on the seafloor).

But you have to have the eukaryotes that were their ancestors survive Snowball Earth(s), preferably in some number and variety. Slushball Earth provides that opportunity, and, I argue, provides ideal conditions for metazoan evolution, not the catastrophically bleak conditions suggested for Snowball Earth.

I might argue also that the evolution of larger metazoans helped to prevent the recurrence of the extreme glaciations that characterize either Snowball Earth or Slushball Earth. Surface productivity could no longer draw down carbon dioxide to critically low levels because primary producers were eaten back by new planktonic predators (small or larval metazoans). Stronger metazoan burrowers (arthropods and segmented worms) dug up buried carbon from seafloor muds and recycled it into carbon dioxide. And higher populations in nearshore waters intercepted nutrients before they reached the oceanic sea surface.

A potential problem is that molecular folks are proposing that there were actual ancestral metazoans long before Snowball Earth kicked in. I don't believe that proposal at the moment, but if there were good evidence for it, I would argue that it jeopardizes the idea of an extreme Snowball Earth, based on the logic of the previous paragraph.

Evidence from the development of metazoan embryos suggests strongly that all the earliest metazoans were "micrometazoans" that would have looked like the larval stages of today's simple metazoans, and quite unlike the adult stages of their modern descendants (see summary by Peterson and Davidson 2000). They would all have been planktonic, feeding on bacteria and protists. This ecological reconstruction makes sense in terms of the oxygen levels in a "normal" Late Proterozoic ocean: most likely highest in the surface layers where photosynthesis occurred, and likely to have been rather low in the sediments of the ocean floor.

The advent of Slushball Earth would have removed eukaryotes and metazoans (if any) from most of the oceans. But the equatorial waters would have provided not only a refuge, but a virtual paradise for protists and micrometazoans specialized to feed on them.

After the last Slushball episode, protists and micrometazoans would have been able to exploit the seafloors that had been accumulating organic sediment for the duration of the glaciation(s). This would have provided an impetus for micrometazoans to have evolved adaptations for crawling and deposit-feeding: obviously this would have come about by evolving the Hox gene complexes that set up appropriate sensory and locomotory appendages in a metamorphosed, larger "adult" metazoan (see Peterson and Davidson 2000). Metazoans such as sponges could have adapted to seafloor life by specializing for bacterial capture. Cnidarians, perhaps already large planktonic feeders, may have evolved the sessile polyp configuration at this time.

As planktonic micrometazoans evolved into metazoans and occupied a great number of ecological niches, we see the Vendian/Cambrian radiation. Geneticists have been arguing that metazoan roots are deep in the Precambrian, and paleontologists have been arguing that if so, there is no fossil evidence of them. This controversy is explained by the Slushball Earth scenario outlined here. Conway Morris (2000) may have mixed metaphors when he wrote "the motor of the Cambrian explosion was largely ecological", but it is clear that he is right.

An Aside: the Early Proterozoic Snowball Earth

Around 2400 Ma, an earlier Snowball Earth began to form. Apparently, at the 1999 GSA meeting, James Kasting suggested an oxygen trigger for it. The Archean atmosphere would have had at least some methane in it, but as the Proterozoic began, the atmosphere and ocean came to have some free oxygen. One of the immediate effects of free oxygen would have been to remove much of the methane from the atmosphere, and as its greenhouse effect failed, the world dropped into a Snowball Earth.

As in the later Snowball Earth scenario, the lock of Early Proterozoic Snowball Earth was broken by the build-up of carbon dioxide from volcanic eruptions. In addition, according to Kirschvink et al (2000), Snowball Earth was followed by a tremendous oxygen surge as cyanobacteria bloomed in the oceans. This not only dropped a lot of iron out of the ocean, but laid down the world's largest manganese deposit, the Kalahari Manganese Field in southern Africa.

We do not have a decent paleogeography for Earth at this time, so we can't do the same sorts of global climatic modelling. However, I would be surprised if a version of Baum and Crowley's GCM did not provide an equally mild Slushball Earth for the Early Proterozoic as well. It is a delicious thought that the evolution of the first eukaryotes might have been nurtured in an early Slushball, just as the first metazoans were nurtured (in my opinion) in the later Slushball. Even I would resist further speculation on this theme at this time, though I may change my mind if we have more ecological and evolutionary speculation from the geophysicists and geochemists who are the prime movers in ideas about Snowballs.

New paper in 2005.
April 8, 2005. Snowball or Slushball Earth? This is a paper in Science, so it will soon be available on the Web. Bodiselitsch, B., et al. 2005. Estimating duration and intensity of Neoproterozoic snowball glaciations from Ir anomalies. Science 308: 239-242, and comment, p. 181. Will soon be on the Web at here and here, respectively. Here's what it says. A Snowball Earth would accumulate incoming space dust on the ice, loaded with iridium. When it melted, there would be an outwash of all the iridium all at once, which would give an iridium spike in the sediments. Indeed, such a spike occurs after the two main major glaciations linked with the Snowball hypothesis.

I don't see why a Slushball ice sheet on the tropical continents wouldn't also accumulate iridium that would also outwash to give an iridium spike: in other words, I don;t see how this study differentiates between Snowball and Slushball. The samples were taken at sites that were along the edge of the tropical continent. So let's do a thought experiment. Antarctica today has been accumulating space dust for 30 m.y. or so. If it melts, as it must some day, won't it produce an iridium spike? Would that imply that Antarctica had been part of a Snowball Earth? I think the answers to these two questions are Yes, and No, respectively. So I'm not going to alter this piece yet.





Web pages 1998-2003:

The Early Proterozoic Snowball Earth of 2300-2400 Ma:

Last updated April 7, 2005.

Links checked September 30, 2005

[Return to Chapter 4]