How to start an ice age

Glaciations are neat – we’re currently in one, and the subjects of my undergrad research lived in the previous one, the Late Paleozoic Icehouse. The Phanerozoic has seen three glaciations so far: the Ordovician Glaciation, the Late Paleozoic Icehouse, and the Cenozoic Ice Ages (glaciation, icehouse, and ice age are often used interchangeably). There were even more in the Precambrian, including the famous “Snowball Earths” of the Proterozoic. Today I’m gonna try to summarize how this happens.

But first: what is an ice age? Put shortly, it’s a period in earth’s geologic history with lots of ice. This is primarily expressed via the presence of large permanent ice caps at one or both of the earth’s poles. In some cases (so far exclusively in the Precambrian), these ice caps may cover most – or all! – of the earth’s surface, making a “Snowball Earth”.a Other characteristics of these periods include extensive mountain glaciations, lower sea levels, and tropical biomes being limited to, well, the tropics. In an ice age, there are glacial and interglacial periods. Glacial periods are stretches of time with very extensive ice sheets, and interglacials are periods with more reduced ice sheets. These are primarily modulated by changes in the earth’s orbit and rotation (Milankovitch cycles), and glacial-interglacial cycles well-documented for both the recent Quaternary glaciation and the Late Paleozoic Icehouse.

The glacial cycle of temperatures and greenhouse gases – University of  Copenhagen
Fluctuations in atmospheric CO2, CH4, and temperatures across the last 800,000 years. Courtesy of the Niels Bohr Institute Centre for Ice and Climate.

a. Some have proposed that the Precambrian snowball earths may have been “Slushball Earths” instead, with the ice caps giving way to an open belt of ocean around the equator. Different studies disagree as to whether a “slushball” is more likely than a hard snowball.

The earth oscillates between “greenhouse” and “icehouse” worlds; the “icehouse” periods are these ice ages, and “greenhouse” periods are the more ice-free periods in between. It should be noted that the presence of glaciers does not an ice age make. There is evidence for glaciers being present in Early Cretaceous Australia (Alley et al. 2020), which would have been in the Antarctic Circle at the time. These glaciers were probably limited in extent and may have been seasonal, and the rest of the world was more consistent with the greenhouse world of the rest of the Mesozoic.

So how do you enter an icehouse world? The easiest way to do that is to remove greenhouse gases from the atmosphere. Greenhouse gases help trap heat in the earth’s atmosphere; the more there are, the hotter the earth will be. Carbon dioxide is a relatively easy greenhouse gas to modulate; it can be added or removed in large amounts by natural processes, and it persists for longer than, say, methane.b So removing CO2 can cool the planet enough for ice caps to form; indeed, drops in CO2 concentration in the past correlate with ice ages and glacial periods within them (e.g. Somelar et al. 2020; Ganopolski et al. 2016). Once ice caps start to form, they can induce a positive feedback loop: more ice increases the earth’s albedo, by providing more light-colored surface for sunlight to reflect off. If this light isn’t absorbed, it doesn’t heat the planet. This drives further cooling, which causes more ice sheet growth, and so on. The logical conclusion of this was expressed in Proterozoic “Snowball Earths” (Hoffman and Schrag 2002).

b. Although removal of methane is also a very effective way to cool the planet. Methane traps more heat than carbon dioxide. This would have been more of a factor for Proterozoic glaciations (e.g. Warke et al. 2020), as methane makes up less of the atmosphere in the Phanerozoic than it did in the Precambrian.

CO2 is primarily removed from the atmosphere via two processes: weathering and photosynthesis. Basically, exposed silicate and carbonate rocks can participate in a series of reactions that remove CO2 from the atmosphere and the oceans. Normally, this process keeps carbon dioxide levels from getting too high over geologic timescales. A rapid increase in surface area of weatherable rocks can increase the rates of weathering and CO2 drawdown, however. Uplift of mountain ranges or plateaus can expose a large amount of rocks to erode, and the chemical weathering of these rocks uses up atmospheric CO2. This is the most commonly proposed mechanism for initiating ice ages, from the Proterozoic Snowball Earths to the Cenozoic.

A diagram of the silicate weathering cycle that probably explains it better than I could in words. From Cockell et al. 2016.

Photosynthesis is also an important factor in starting icehouse worlds. The process of photosynthesis removes carbon dioxide from the atmosphere, and this carbon is incorporated into the growing photosynthesizers. If they burn or die and rot, then the carbon can go back into the atmosphere. But if they’re buried beforehand, the carbon gets trapped in the ground. If you do that a lot, you can remove a lot of CO2 from the atmosphere and cool the planet. Land plants can also increase rates of erosion or weatherability, and this may have influenced the Paleozoic glaciations, when life was still getting a foothold on land and the carbon cycle wasn’t used to them yet (Lenton et al. 2012). At least two, and maybe more, of the major glaciation events correlate to upticks in photosynthesis and/or growth and burial of plants.

The very first “Snowball Earth”, the Huronian Glaciation, happened immediately after the Great Oxygenation Event (2.5-2.45 Ga) – the appearance of photosynthesizing organisms. These photosynthesizing organisms removed a lot of carbon dioxide from the atmosphere (for the first time in earth’s history) and released a lot of oxygen. The oxygen reacted with methane in the atmosphere, producing the less effective greenhouse gas CO2 (which would have then been used by the photosynthesizing organisms). The drawdown of CO2 and elimination of methane sent the earth into an icehouse (Somelar et al. 2020; Warke et al. 2020).
Artist’s rendition of the Huronian Snowball Earth. By Oleg Kuznetsov, via Wikimedia Commons.

The Neoproterozoic Snowball Earths (yes, Snowball Earths, there were more than one) may have been influenced by the breakup of the continent Rodinia, which was centered around the equator. The large amount of continental land around the equator reduced the albedo at the equator, and breaking it up may have allowed land at the interior of the continents to weather more easily, by bringing them closer to water (Hoffman 1999). These glaciations may also have been influenced by diversification of marine eukaryotes. Phytoplankton diversifying in the upper ocean would have drawn down CO2. As they died and sank to anoxic deeper waters, they underwent remineralization that prevented the carbon from getting back to the atmosphere (Tziperman et al. 2011). Ultimately, though, it still isn’t certain what caused the Neoproterozoic Snowball Earths, or why there are non-glacial gaps in between the snowball events.

The Ordovician Glaciation is really weird in that a.) it’s short (only 2 million years of intense glaciation in a 20-million-year cooling period from 460-~440 Ma), and b.) it’s in the middle of what is otherwise a greenhouse world (McGhee 2018). Nobody can agree on what caused it. Proposed hypotheses include weathering of or carbon cycle disturbance from a large igneous province from the earlier Ordovician; we haven’t actually found this large igneous province, and probably never will (it could have eroded in the intervening 460 million years) but it isn’t inconsistent with the data and modeling (Lefebvre et al. 2010). A more interesting hypothesis ties into the origin of land plants. The first land plants appeared in around the mid-Ordovician (Salamon et al. 2018), and these plants probably rapidly spread in this new habitat. Although these plants were small and wouldn’t have taken down too much CO2 on their own, they would have increased weathering rates on land, which had never happened before due to lack of land plants. This could have happened over a very short timeframe, consistent with the rapid cooling of the late Ordovician (Lenton et al. 2012; Retallack 2015).

The Late Paleozoic Icehouse originated in the latest Devonian. The Devonian had a lot of tectonic activity, which formed a lot of mountain ranges and weatherable surface rock (McGhee 2018). It’s also known for being the time when true plants really start taking off on land; although land plants were around since the Ordovician, the first trees and forests show up around this time. These trees took in a lot of carbon dioxide in the atmosphere during photosynthesis, and as they died and were eventually buried they took this carbon into the ground with them. Less CO2 in the atmosphere from weathering and photosynthesis meant things started to got colder from the end of the Devonian. The incipient glaciation caused sea levels to drop, exposing large expanses of tropical lowlands. These lowlands were quickly covered by rich swampy forests filled with plants. These plants sucked up more CO2, and as they died and were buried, they locked the carbon into massive peat formations (McGhee 2018), which was eventually turned into coal.c Uplift of mountains in the Carboniferous produced even more surface area for weathering, driving cooling even further, and over the threshold required for large ice cap formation (Goddéris et al. 2017). This led to the formation of a large ice cap at the south pole, covering the southern part of Pangaea from the end of the Carboniferous well into the Permian (McGhee 2018).
The Pennsylvanian globe. Much of Gondwana was covered by a large ice cap. Map by Ron Blakey.

c. If you burn that coal today, you’re releasing carbon into the atmosphere that hasn’t been there for over 300 million years.

It’s a bit less certain what kicked off the Cenozoic Glaciation, but there’s evidence for a really interesting event happening that may have set it in motion. There’s evidence (in the form of lots and lots of spores) for massive amounts of the freshwater fern Azolla being deposited in the Arctic Ocean in the early-mid Eocene, 50-48 million years ago (Brinkhuis et al. 2006). Azolla is a fast-growing plant that, if it grows large enough, can self-sustain a large floating biomass (van Kempen et al. 2012). This suggests that the surface of the Arctic Ocean may have been freshwater, or at least fresher than it is now, and lots of Azolla was living and breeding there (Brinkhuis et al. 2006); alternatively, the Azolla grew en masse in nearby freshwater and was washed into the Arctic Ocean (Neville et al. 2019). Either way, individual Azolla eventually died, and so the massive mats of floating freshwater ferns were buried in large quantities in the Arctic, drawing down more and more CO2 – between 900 and 3,500 gigatons of carbon (Speelman et al. 2009). At this point Earth was well in a greenhouse period (this was shortly after the Paleocene-Eocene Thermal Maximum), and this event may have helped the climate cool and head in the other direction, eventually leading to the intense glaciation of the later Cenozoic. This was then exacerbated by weathering from several recently-uplifted mountain ranges, most prominently the Himalayas and Tibetan Plateau, which provided lots of weatherable rock starting in around the Oligocene (Zachos et al. 1999; Garzione 2008).
Harbingers of the cold. Photo by Harry Rose, via Wikimedia Commons.

Another potentially important factor present in glacial periods is the presence of land over the poles. It’s easier to form a massive ice sheet if there’s a surface to do it on. All the Phanerozoic glaciations have one thing in common: a continent over the South Pole (Scotese 2018). The Andean-Saharan glaciation had Gondwana covering the South Pole. By the time the Late Paleozoic Icehouse started, Gondwana had been welded into Pangaea, which was still centered in the southern hemisphere. Pangaea broke up in the Mesozoic, but Antarctica remained at the South Pole, where it sits to this day. Right now we also have a bunch of land around the north pole, but the exact pole isn’t covered. The northern ice cap of the Cenozoic is a first for the Phanerozoic (Scotese 2018). The northern ice cap is thinner and less extensive than the southern one, perhaps as a result of there being less land for ice caps to grow on. Continents covering the poles don’t necessarily mean there was a glaciation (Gondwana and later Antarctica have been in roughly the same spot since the Ordovician, and there’s a lot of greenhouse world in that time), but they may promote ice cap formation.

Estimated extents of ice sheets during the Cenozoic, Late Paleozoic, and Ordovician glaciations. Notice there’s a bunch of land near the poles they form at. Modified from Scotese (2018).

So that’s how ice ages formed in the past. What about the future? There are still ice caps, so we aren’t out of the ice age yet, we’re just in an interglacial period. Thanks to anthropogenic climate change, the Quaternary glaciation cycle is certainly gonna be disrupted; it’s estimated that the onset of the next glacial period won’t hit for at least 100,000 years from now (Ganopolski et al. 2016); basically, this interglacial will continue through to the next one. I’m personally optimistic; I don’t think the ice caps will completely disappear before we’re done here, but it would take time for them to grow back to the extent they had in the Pleistocene. Either way, though, the Cenozoic glaciation will eventually come to an end; if not by humans, then perhaps volcanism or continental drift (Antarctica probably can’t stay at the South Pole forever). The next ice age may appear with the next supercontinent, which is estimated to form in around 250 million years. If this supercontinent covers one of the poles (probably the north pole), conditions will be right for a huge ice cap to potentially form again (Way et al. 2020).


Alley, N.F.; Hore, S.B.; Frakes, L.A. (2020). “Glaciations at high-latitude Southern Australia during the Early Cretaceous”. Australian Journal of Earth Sciences 67(8): 1045-95.
Brinkhuis, H.; Schouten, S.; Collinson, M.E.; Sluijs, A.; Sinninghe Damaste, J.S.; Dickens, G.R.; Huber, M.; Cronin, T.M.; Onodera, J.; Takahashi, K.; Bujak, J.P.; Stein, R.; van der Burgh, J.; Eldrett, J.S.; Harding, I.C.; Lotter, A.F.; Sangiorgi, F.; van Konijnenburg-van Cittert, H.; de Leeuw, J.W.; Matthiessen, J.; Backman, J.; Moran, K.; the Expedition 302 Scientists (2006). “Episodic fresh surface waters in the Eocene Arctic Ocean”. Nature 441: 606-9.
Ganopolski, A.; Winkelmann, R.; Schellnhuber, H.J. (2016). “Critical insolation-CO2 relation for diagnosing past and future glacial inception”. Nature 529: 200-3.
Garzione, C.N. (2008). “Surface uplift of Tibet and Cenozoic global cooling”. Geology 36(12): 1003-4.
Goddéris, Y.; Donnadieu, Y.; Carretier, S.; Aretz, M.; Dera, G.; Macouin, M.; Regard, V. (2017). “Onset and ending of the late Paleozoic ice age triggered by tectonically paced rock weathering”. Nature Geoscience 10: 382-6.
Hoffman, P.F. (1999). “The break-up of Rodinia, birth of Gondwana, true polar wander and the snowball Earth”. Journal of African Earth Sciences 28(1): 17-33.
Hoffman, P.F.; Schrag, D.P. (2002). “The snowball Earth hypothesis: testing the limits of global change”. Terra Nova 14(3): 129-55.
van Kempen, M.M.L.; Smolders, A.J.P.; Lamers, L.P.M.; Roelofs, J.G.M. (2012). “Micro-Halocline Enabled Nutrient Recycling May Explain Extreme Azolla Event in the Eocene Arctic Ocean”. PLoS One 7(11): e50159.
Lefebvre, V.; Servais, T.; Francois, L.; Averbuch, O. (2010). “Did a Katian large igneous province trigger the Late Ordovician glaciation? A hypothesis tested with a carbon cycle model”. Palaeogeography, Palaeoclimatology, Palaeoecology 296: 310-9.
Lenton, T.M.; Coruch, M.; Johnson, M.; Pires, N.; Dolan, L. (2012). “First plants cooled the Ordovician”. Nature Geoscience 5: 86-9.
McGhee, G.R. (2018). Carboniferous Giants and Mass Extinction: The Late Paleozoic cCe Age World. Columbia University Press.
Neville, L.A.; Grasby, S.E.; McNeil, D.J. (2019). “Limited freshwater cap in the Eocene Arctic Ocean”. Scientific Reports 9: 4226.
Retallack, G.J. (2015). “Late Ordovician Glaciation Initiated by Early Land Plant Evolution and Punctuated by Grenhouse Mass Extinctions”. The Journal of Geology 123(6).
Salamon, M.A.; Gerrienne, P.; Steemans, P.; Gorzelak, P.; Filipiak, P.; Le Herisse, A.; Paris, F.; Cascales-Minana, B.; Brachaniec, T.; Misz-Kennan, M.; Niedzwiedzki, R.; Trela, W. (2018). “Putative Late Ordovician land plants”. New Phytologist 218(4): 1305-9.
Scotese, C.R. (2018). “An Estimate of the Volume of Phanerozoic Ice”. American Geophysical Union Fall Meeting 2018, Washington D.C.
Somelar, P.; Soomer, S.; Driese, S.G.; Lepland, A.; Stinchcomb, G.E.; Kirsmae, K. (2020). “CO2 drawdown and cooling at the onset of the Great Oxygenation Event recorded in 2.45 Ga paleoweathering crust”. Chemical Geology 548: 119678.
Speelman, E.N.; van Kempen, M.M.L.; Barke, J.; Brinkhuis, H.; Reichart, G.J.; Smolders, A.J.P.; Roelofs, J.G.M.; Sangiorgi, F.; de Leeuw, J.W.; Lotter, A.F.; Sinninghe Damste, J.S. (2009). “The Eocene Arctic Azolla bloom: environmental conditions, productivity and carbon drawdown”. Geobiology 7(2): 155-70.
Warke, M.R.; Di Rocco, T.; Zerkle, A.L.; Lepland, A.; Prave, A.R.; Martin, A.P.; Ueno, Y.; Condon, D.J.; Claire, M.W. (2020). “The Great Oxidation Event preceded a Paleoproterozic “snowball Earth””. Proceedings of the National Academy of Sciences 117(24): 13314-20.
Way, M.J.; Davies, H.S.; Duarte, J.C.; Green, J.A.M. (2020). “The Climates of Earth’s Next Supercontinent: Effects of Tectonics, Rotation Rate & Insolation”. European Geosciences Union General Assembly 2020, online.
Zachos, J.C.; Opdyke, B.N.; Quinn, T.M.; Jones, C.E.; Halliday, A.N. (1999). “Early Cenozoic glaciation, antarctic weathering, and seawater 87Sr/86Sr: is there a link?”. Chemical Geology 161: 165-80.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s