How Do Ice Ages Happen? Exploring Paleosalinity and Thermohaline Circulation

Fill a glass of water from the sea and try to drink it. You gag and your lips pucker. After all, dissolved in that liter of the ocean are around 35 grams of salt. Now, imagine you tried to do this same thing 1 million years ago, 10 million years ago, 100 million years ago, even 500 million years ago. Would you ever be able to drink the water — or would it ever be as salty as the Dead Sea today? These are the questions that investigating paleosalinity helps us answer. We can use a variety of methods — from rough estimates based on our knowledge of the earth system to direct evidence from water droplets preserved in old rocks — to determine how salty the ocean was throughout Earth’s history.

Figuring out how ocean salinity has changed is important not simply because it helps us fill pages in our history book. Changing salinity affects ocean circulation, which in turn has a huge impact on climate. For instance, recent studies have suggested that changes in thermohaline circulation are part of how the Earth cycles between glacial and interglacial periods. In other words, ocean circulation may be the missing link between orbital variations that we know are linked to the cycle of ice ages and the huge swings in CO2 and temperature that are directly responsible for plunging us into glacial periods. In short, tracing paleosalinity helps us understand just how the Earth’s temperature can change so drastically.

This paper is therefore driven by the question: How can paleosalinity help us understand climatic variation, especially that caused by changes in thermohaline circulation? To begin to answer this question, I will present two methods of determining paleosalinity: first, by making an inventory of evaporites; and second, by looking at fluid inclusions. I then turn to the effects that these changes in salinity have on climate, especially by looking at models of how thermohaline circulation would differ and past cases where similar things occurred (for instance, in the Younger Dryas).

Today, we calculate the salinity of seawater by measuring the mass of dried salts in a given volume — usually stated in grams per liter or (equivalently) parts per thousand. For instance, today’s seawater is roughly 34–35‰ (or 3.4–3.5%) saline (i.e. 34–35 grams of salt per liter of seawater). Usually, when we think of salt we refer to sodium chloride (NaCl). But salts in chemical nomenclature include a larger variety of compounds. Four such salts, or eight ions (Cl, SO42−, HCO3, Br, and Na+, Mg2+, Ca2+, K+), make up 99.8% by weight of the solids dissolved in seawater. NaCl comprises 85.1%, with chloride alone making up 55%. Many researchers therefore use the concentration of chloride ions as a proxy for salinity — this despite the problem that an increase in the concentration of chloride ions could just show that chloride dominated more than other ions, not that the overall salinity itself increased. Still, chloride does not easily enter into minerals and instead resides almost entirely in seawater, ocean sediments, and evaporites. The concentration of chloride ions therefore remains a decent proxy for ocean salinity.

This is the background against which a 2006 paper produced a history of salinity (Hay et al. 2006). The authors considered sources and sinks of chloride ions throughout Earth’s history. Analogously, they also considered whether the volume of water on Earth has stayed steady, concluding that it has overall been unchanged, though the proportion in the ocean depends on how much is locked up as ice. What is really important for Hay et al., though, is the evaporite inventory. Evaporites are large amounts of solids that represent the extraction of salt from the oceans. For instance, drilling in the Gulf of Mexico in 1967 revealed a vast layer of salt from the Jurassic beneath the oceanic sediments. The total global amount of halite (salt deposits) discovered was astonishing: the minimum estimate is 19.6 quintillionkilograms (19.6 × 1018 kg), or more than half of the total salt in the oceans today — the maximum estimate is that an unbelievable 95% of the present total is trapped in halite deposits. If all of this had been dissolved at one time, the salinity of seawater would have been much higher: 57‰ to 73‰ instead of today’s 34‰ to 35‰. This extreme scenario is unlikely, but the discovery of such large amounts of evaporites raised questions about just how salinity had been different in the past.

Hay et al. meticulously track the inventory of these evaporites over the past 540 million years (the Phanerozoic eon). They combine these data with their estimates of water volume to calculate salinity at different points in the Earth’s history, which they summarize in the following graph (figure 1):

Figure 1. The mean salinity of the ocean during the Phanerozoic eon (with today at the left). Source: Hay et al. 2006.

We see a secular (long-term) decline of salinity from 540 Ma to today. This decline is significant but not huge; the maximum salinity of the ocean was around 50‰, while salt lakes today range from 140‰ (Utah’s Great Salt Lake) to 340‰ (the Dead Sea) — at minimum double and at most seven to ten times higher than even the saltiest period Hay et al. reconstruct. Even in these hypersaline environments today, we find life. Red Sea corals today, in fact, thrive in salinities of 41‰ to 43‰ — conditions that were widespread in the Mesozoic and Triassic. Indeed, Hay et al. suggest that “at present many marine animals and plants are living nearer their low salinity tolerance limit rather than their high limit.” So, then, what is the real significance of finding elevated paleosalinity levels?

Before answering this question, we should consider a second way of estimating paleosalinity. Some of the most exciting work of the past few decades has been on fluid inclusions: little drops of seawater trapped in evaporite crystals. These drops give us a window into the composition of oceans that we otherwise have only indirect access to. Of course, the trapped water is not perfectly preserved. Most significantly, these drops were trapped as salt was formed through evaporation. But we have experimental evidence of these “evaporation pathways” that allows us to reconstruct with some confidence the chemical composition of past oceans (see in particular Horita, Zimmermann, and Holland 2002; a succinct overview is provided in section 8.2 of Tyrrell 2013). The other difficulty when working with fluid inclusions is the extremely small size, which necessitates unconventional laboratory methods. These techniques — including ion chromatography, inductively coupled plasma mass spectrometry, and x-ray dispersive systems — allow analysis of inclusions as small as 20 micrometers in diameter (or about the width of a human hair). Overcoming these technical difficulties led to a discovery that overturned the long-held assumption that ocean composition has been more or less steady (see e.g. Rubey 1951). By contrast, fluid inclusions tell us that ancient seawater was clearly very different than contemporary oceans.

So what, in particular, does this tell us about paleosalinity? One especially promising area of study are Precambrian oceans. While Hay et al. could only estimate paleosalinity to 542 million years ago — since older evaporites are much scarcer on today’s surface — fluid inclusions are present from much, much older rocks. This is why many studies have focused on samples from South African, Canadian, and Australian cratons (areas of exposed rocks that are billions of years old). A 1997 study of South Africa’s Ironstone Pools indicated that Archean seawater had about 165% of today’s chloride — in other words, 3.2 billion years ago the Earth’s oceans were significantly saltier (de Ronde et al. 1997). In 1998, Paul Knauth observed that such salinity levels pose problems for most non-microbial life (Knauth 1998); by contrast, cyanobacteria are quite tolerant of salinity variations. Knauth therefore suggests that the dominance of cyanobacteria in the Archean “may be, in part, because higher salinities were an impediment to the evolution of more complicated life forms.” Knauth later expanded on this claim in a 2005 article, where he argued that the Archean Ocean was 1.2 to 2 times as saline as today and hence that “marine life was limited to microbes (including cyanobacteria) that could tolerate the hot, saline early ocean” (Knauth 2005).

Almost at the same time, another group of scientists conducted a statistical analysis of fluid inclusions trapped in Archean quartz deposits around the world to show that salinity may have been around double today’s value (Weiershäuser and Spooner 2005). A 2004 paper similarly showed salinity from Western Australia around four times today’s value (Foriel et al. 2004). Such studies supported Knauth’s assertion that the evolution of continents and the desalination of oceans is a prerequisite for the evolution of complex life. However, a more recent article casts doubt on this line of argument. The authors analyzed Archean quartz crystals from Australia and South Africa with the extended argon–argon method to show that the salinity of oceans 3 billion years ago is “comparable to, or even lower than, the modern one” (Marty et al. 2018). The authors acknowledged that there may have been localized saltier environments, as indicated by De Ronde et al.; but overall, their finding of globally lower salinity in the Archean undermines Knauth’s assertion that salinity was crucial to the evolution of life. While fluid inclusions continue to be an exciting method, the issues I discussed above clearly continue to pose problems in determining paleosalinity.

More promising work, in my opinion, is possible if we restrict our analysis of fluid inclusions to the Phanerozoic, or the last 540 million years. This is where we begin to see the importance of paleosalinity not just on life but also on thermohaline circulation and hence short-term climate fluctuations. In essence, the argument is simple. Ocean circulation depends on the fact that cold water and more saline water both sink, because they are more dense. This is what thermohaline circulation means: seawater mixes and moves around both because of differences in temperature (thermo-) and salinity (-haline). (Although wind and tidal forces are inextricable from the broader circulatory patterns, it is useful to focus on temperature and salinity for our purposes.) If the overall salinity of the Earth’s oceans was different in the past, this pattern of thermohaline circulation would also have changed. In particular, Hay et al. argued that at higher salinities less energy is needed to make water more dense, and hence ocean water mixed more actively in the past. In other words, at higher salinities “temperature becomes more and more dominant in controlling the density of seawater” — as opposed to the density increases through salinization that dominate in today’s thermohaline circulation. Salinization involves phase changes, and phase changes (from liquid water to solid ice, for instance) require a lot of energy. As Hay et al. conclude:

The phase changes required to increase seawater density today act as a flywheel on the ocean–atmosphere system, consuming energy and thereby slowing down the rate at which deep water formation can take place. At higher ocean salinities these phase changes are not required, and much less energy is needed to drive the thermohaline circulation. A more saline ocean could convect much more readily than it does today. (p. 42)

If Hay et al. are right (that is, the Phanerozoic has seen a decline in ocean salinity), then ocean waters were more well-mixed in the past. Is this true? And, if so, why does it matter?

Unfortunately, the first question is difficult to answer with certainty. Too many technical challenges remain with fluid inclusion analysis to make feasible the large-scale data collection and processing that would be needed to definitively prove a secular decline in ocean salinity in the Phanerozoic. Still, a consensus has developed that ocean chemistry, including salinity, has changed over the past 542 million years — we’re just not quite sure how. For instance, a 2001 article argued that “oscillations in global seawater chemistry” parallel “long-term changes in seafloor spreading rates, global sea level, and ‘greenhouse’ versus ‘icehouse’ … during the Phanerozoic because they are all driven by plate tectonics” (Lowenstein et al. 2001, 2003). A 2013 article confirmed this observation and provided stronger, higher-resolution data by restricting its analysis to the last 36 million years (Brennan, Lowenstein, and Cendón 2013). Yet more evidence for secular changes in ocean composition was provided by a fascinating 2015 article on fossilcorals as an archive of seawater chemistry (Gothmann et al. 2015, 2017). These are just a few of the more relevant papers that have used fluid inclusions to show long-term changes in ocean chemistry in the Phanerozoic.

Fluid inclusions have also proven interesting on shorter time scales — and this is a way to get at the second question about why Hay et al.’s conclusion matters. A 2002 article used data from benthic foraminifera and pore fluids from four Ocean Drilling Program cores to track both salinity (as the concentration of chloride ions) and temperature (as δ18O values) around the last glacial maximum 20 000 years ago (Adkins, McIntyre, and Schrag 2002). The two measures were strongly correlated. In other words, the study confirmed what we expected: at the last glacial maximum, ice volume was high. δ18O in benthic foraminifera are heavier when the water temperature is colder, and when global ice volume is greater (because ice locks up “light” H216O preferentially to “heavy” H218O) — so at the last glacial maximum δ18O values in benthic foraminifera were heavy. At the same time, the large volume of fresh water locked up in glaciers and ice sheets meant that the remaining oceans were more saline than at present — hence the correlation between a peak in the concentration of chloride ions and δ18O values. In short, fluid inclusions can help us better understand the changes in ocean chemistry at glacials and interglacials. In fact, the relationship is bidirectional. Not only does climate affect ocean chemistry and hence fluid inclusions, but ocean circulation also affects climate. This is the fundamental point Hay et al. are trying to make: we need to know what past ocean chemistry was like (including paleosalinity) to understand how thermohaline circulation was different than today.

Thermohaline circulation in turn is integral to the climate cycle. We have known for centuries that the Earth has seen repeated cycles of glaciation. For the past 2.58 million years, the Earth has been in its most recent ice age, the Pleistocene glaciation. The ice volume fluctuates cyclically alongside the climate (see figure 2). Since a classic article in the 70s (Hays, Imbrie, and Shackleton 1976), these cycles are accepted as correlated with Milankovitch cycles, three orbital variations that affect how much solar radiation reaches the Earth’s surface. These variations correlate well with the glacial–interglacial cycles. However, the Milankovitch cycles only partly explain the cycles, since the variation in solar radiation is far too small to account for the large swings in temperature observed on the earth. Some of the amplification can be explained with reference to standard climate feedbacks — for instance, the well-known greenhouse effect (including amplification by water vapor). Indeed, we are quite sure that CO2  rises alongside the Milankovitch cycles and temperature fluctuations responsible for glacial periods (Shakun et al. 2012). But this still isn’t getting us closer to an actual mechanism by which we could connect Milankovitch cycles and the swings in CO2 and temperature.

Figure 2. Source: Wikimedia Commons; adapted from Lisiecki and Raymo 2005.

In recent years, the field seems to be converging on an explanation that has something to do with ocean circulation. Two examples will illustrate my point. The first is a classic issue with Milankovitch cycles. What Milankovitch himself considered the most important orbital variation (obliquity) occurs on a 41 000–year cycle, not every 100 000 years (Milanković 1998). So what caused the Mid-Pleistocene Transition (MPT) between these two periods, and an accompanying increase in the amplitude of glacials (see figure 2)? One hypothesis is that the removal of regolith (soil, gravel, etc.) by glacial erosion made ice sheets less mobile and increased their albedo (reflectivity), reducing the ability of ice to quickly respond to orbital forcing. A related hypothesis is that CO2 gradually declined in a world covered by ice. A paper published just a few weeks ago used a climate model to show that these are feasible mechanisms to explain the Mid-Pleistocene Transition (Willeit et al. 2019). An alternative hypothesis, though, is that the MPT is caused by a much weaker glacial thermohaline circulation (Pena and Goldstein 2014). This article reports neodymium values — which function as a proxy for mixing of ocean waters — from cores from the southeast Atlantic. Before the MPT, the difference between glacial and interglacial neodymium values is small — so there is sustained and vigorous mixing because of thermohaline circulation, even during glacial periods. By contrast, the neodymium values during post-MPT glacials is much more positive, indicating much weaker thermohaline circulation (while interglacials stay the same). A weak thermohaline circulation means less chance for surface and deep waters to mix, which facilitates reduced CO2 levels and hence further cooling.

Is this a hypothesis the literature supports? Generally, yes (see the overview provided by Lohmann, Zhang, and Knorr 2016). A 2014 article used δ13C values from benthic foraminifera in the North Atlantic to show the intensification of deep-water formation (in other words, weaker mixing) during the MPT (Poirier and Billups 2014). A 2016 paper, though focused on providing a high-resolution Mg/Ca record (a direct proxy for temperature) from benthic foraminifera for the past 1.5 million years, also concluded that it was “thermohaline reorganization [that] created a new climate regime with enhanced sensitivity to the 100 kyr orbital cycle” (Ford et al. 2016). Most recently, a reconstruction of foraminiferal trace element (B/Ca, Cd/Ca) and Nd isotope proxies confirmed that “weaker overturning circulation and Southern Ocean biogeochemical feedbacks facilitated deep ocean carbon storage, which lowered the atmospheric partial pressure of CO2 and thereby enabled expanded terrestrial ice volume at the mid-Pleistocene transition” (Farmer et al. 2019). There is at least one cautionary voice, though: another record from neodymium data suggests that the strength of mixing during interglacials has not varied significantly, and hence can be decoupled from the concentration of CO2 that others have suggested controls the MPT (Howe and Piotrowski 2017). In any case, this question is worth pursuing in detail: current melting in Greenland will affect thermohaline circulation in the North Atlantic, and it behooves us to know how.

Figure 3. Source: the United States Geological Survey.

The effects of freshwater released to the North Atlantic are shown perhaps best by the example of the Younger Dryas. After the last glacial maximum, the earth began warming steadily. For 1500 years, though, the Earth went through an abnormally cold (and dry) period (see figure 3). A small flower, the Arctic dryas, expanded much farther south than it is found today. What could have caused such a sudden drop in temperature? One of the most popular theories, championed by Wallace Broecker, is that the Younger Dryas was triggered by a flood of fresh water from proglacial lakes in North America (Johnson and McClure 1976; Rooth 1982; Broecker, Peteet, and Rind 1985; Broecker 2006). As the Laurentide ice sheet retreated, it left behind vast lakes in what is now the northern US. At some point, proponents of this hypothesis argue, the waters of these lakes spilled over and poured massive volumes of cold, fresh water into the North Atlantic. This in effect shut down the local thermohaline circulation. The fresh water (which floated because it was less saline) interrupted deep ocean circulation and in turn caused widespread cooling (Tarasov and Peltier 2005; Alley 2007).

Figure 4. Three possible pathways for flooding from proglacial lakes. Source: Murton et al. 2010.

Two issues remain with this theory. First, it is not entirely clear just how this interruption of thermohaline circulation played out on climate, especially farther afield and on different time scales. Second, and most significantly, we don’t have a clear understanding of the pathway the fresh water took from the lakes to the ocean. Such massive floods so recently in the geologic record should leave huge scars on the Earth. Indeed, many scientists are looking for evidence of these floods in sediments and canyons all over likely locations in North America (see figure 4). For instance, a 2010 paper claimed to identify the path (or one of the paths), through the Mackenzie River to the Arctic Ocean (Murton et al. 2010). This hypothesis was lent credence by recent δ18O data from the Arctic Ocean (Keigwin et al. 2018). Fresh water is much lighter (isotopically) than ocean water, so a flood of this magnitude should leave an unmistakable period of locally light δ18O values that correspond with the flood that initiated the Younger Dryas — just what Keigwin et al. found in benthic foraminifera near the mouth of the Mackenzie. Just as interesting as this debate, but less often discussed, is the incredibly abrupt termination of the Younger Dryas. One 1998 paper claimed the warming into the current stable Holocene might have taken only “as little as 1 to 3 years” — with seven degrees Celsius of warming in that short time (Fawcett et al. 1997). By comparison, the past two hundred years have seen at most 1.5 degrees of warming. We should note, though, that local change, especially at the poles, may be significantly greater than the global mean (Partin et al. 2015) — which never exceeded 2℃ over the Younger Dryas (Shakun et al. 2012). It remains striking, though, how little attention is paid to the termination of the Younger Dryas, relative to its onset (an issue noted by Pearce et al. 2013).

As these two examples illustrate, ocean circulation is extremely important for climate changes throughout the Pleistocene (for more examples see Lohmann, Zhang, and Knorr 2016; Thiagarajan et al. 2014; Matsumoto 2017; Sigman, Hain, and Haug 2010). In turn, thermohaline circulation is closely linked with ocean chemistry and in particular paleosalinity. We need to understand how the mixing of waters was different in the past in order to understand the mechanisms for both abrupt (Younger Dryas) and long-term (MPT) changes; in order to understand the changes in thermohaline circulation, we need a better grasp of how the chemical composition of the oceans (including its salts) was different in the past.

At the end of the day, we really don’t know how the Earth has swung so drastically from glacial to interglacial periods over the past 2.58 million years. How and why do we go from kilometers of ice covering large parts of major landmasses to tropical plants extending over the globe? These changes can be quite frightening. After all, the abrupt onset and termination of the Younger Dryas both occurred when there were plenty of humans around — and it is surely no coincidence that agriculture began not long after its end. What was it like to live through such changes? To answer such questions, we must come face-to-face with a central question preoccupying earth sciences today: What is the cause and mechanism of glacial–interglacial transitions? We have a good sense that the way the Earth orbits around the Sun has something to do with it. We’re pretty sure that CO2 plays a large role, as do many of the feedbacks we still see at play today (water vapor, albedo, plant biomass, and more). And we’re more and more certain that changes in the ocean’s thermohaline circulation are an important link in this mechanism. But a full picture — something we need quite urgently today — remains elusive. I do not suggest that this paper has helped color in this picture. But I have demonstrated clearly that part of this understanding needs to be an account of secular changes in ocean chemistry, particularly in the form of paleosalinity.


Adkins, Jess F., Katherine McIntyre, and Daniel P. Schrag. 2002. “The Salinity, Temperature, and δ18O of the Glacial Deep Ocean.” Science 298 (5599): 1769–73.

Alley, Richard B. 2007. “Wally Was Right: Predictive Ability of the North Atlantic ‘Conveyor Belt’ Hypothesis for Abrupt Climate Change.” Annual Review of Earth and Planetary Sciences 35 (1): 241–72.

Brennan, Sean T., Tim K. Lowenstein, and Dioni I. Cendón. 2013. “The Major-Ion Composition of Cenozoic Seawater: The Past 36 Million Years from Fluid Inclusions in Marine Halite.” American Journal of Science 313 (8): 713–75.

Broecker, Wallace S. 2006. “Was the Younger Dryas Triggered by a Flood?” Science 312 (5777): 1146–48.

Broecker, Wallace S., Dorothy M. Peteet, and David Rind. 1985. “Does the Ocean–Atmosphere System Have More than One Stable Mode of Operation?” Nature 315 (6014): 21.

Farmer, J. R., B. Hönisch, L. L. Haynes, D. Kroon, S. Jung, H. L. Ford, M. E. Raymo, et al. 2019. “Deep Atlantic Ocean Carbon Storage and the Rise of 100,000-Year Glacial Cycles.” Nature Geoscience 12 (5): 355.

Fawcett, Peter J., Anna Maria Ágústsdóttir, Richard B. Alley, and Christopher A. Shuman. 1997. “The Younger Dryas Termination and North Atlantic Deep Water Formation: Insights from Climate Model Simulations and Greenland Ice Cores.” Paleoceanography 12 (1): 23–38.

Ford, Heather L., Sindia M. Sosdian, Yair Rosenthal, and Maureen E. Raymo. 2016. “Gradual and Abrupt Changes during the Mid-Pleistocene Transition.” Quaternary Science Reviews 148 (September): 222–33.

Foriel, Julien, Pascal Philippot, Patrice Rey, Andrea Somogyi, David Banks, and Bénédicte Ménez. 2004. “Biological Control of Cl/Br and Low Sulfate Concentration in a 3.5-Gyr-Old Seawater from North Pole, Western Australia.” Earth and Planetary Science Letters 228 (3): 451–63.

Gothmann, Anne M., Jarosław Stolarski, Jess F. Adkins, and John A. Higgins. 2017. “A Cenozoic Record of Seawater Mg Isotopes in Well-Preserved Fossil Corals.” Geology 45 (11): 1039–42.

Gothmann, Anne M., Jarosław Stolarski, Jess F. Adkins, Blair Schoene, Kate J. Dennis, Daniel P. Schrag, Maciej Mazur, and Michael L. Bender. 2015. “Fossil Corals as an Archive of Secular Variations in Seawater Chemistry since the Mesozoic.” Geochimica et Cosmochimica Acta 160 (July): 188–208.

Hay, William W., Areg Migdisov, Alexander N. Balukhovsky, Christopher N. Wold, Sascha Flögel, and Emanuel Söding. 2006. “Evaporites and the Salinity of the Ocean during the Phanerozoic: Implications for Climate, Ocean Circulation and Life.” Palaeogeography, Palaeoclimatology, Palaeoecology 240 (1): 3–46.

Hays, J. D., John Imbrie, and N. J. Shackleton. 1976. “Variations in the Earth’s Orbit: Pacemaker of the Ice Ages.” Science 194 (4270): 1121–32.

Horita, Juske, Heide Zimmermann, and Heinrich D. Holland. 2002. “Chemical Evolution of Seawater during the Phanerozoic: Implications from the Record of Marine Evaporites.” Geochimica et Cosmochimica Acta 66 (21): 3733–56.

Howe, Jacob N. W., and Alexander M. Piotrowski. 2017. “Atlantic Deep Water Provenance Decoupled from Atmospheric CO 2 Concentration during the Lukewarm Interglacials.” Nature Communications 8 (1): 2003.

Johnson, R. G., and B. T. McClure. 1976. “A Model for Northern Hemisphere Continental Ice Sheet Variation.” Quaternary Research 6 (3): 325–353.

Keigwin, L. D., S. Klotsko, N. Zhao, B. Reilly, L. Giosan, and N. W. Driscoll. 2018. “Deglacial Floods in the Beaufort Sea Preceded Younger Dryas Cooling.” Nature Geoscience 11 (8): 599.

Knauth, L. Paul. 1998. “Salinity History of the Earth’s Early Ocean.” Nature 395 (6702): 554–55.

———. 2005. “Temperature and Salinity History of the Precambrian Ocean: Implications for the Course of Microbial Evolution.” Palaeogeography, Palaeoclimatology, Palaeoecology 219 (1–2): 53–69.

Lisiecki, Lorraine E., and Maureen E. Raymo. 2005. “A Pliocene-Pleistocene Stack of 57 Globally Distributed Benthic δ18O Records.” Paleoceanography 20 (1): n/a-n/a.

Lohmann, Gerrit, Xu Zhang, and Gregor Knorr. 2016. “Abrupt Climate Change Experiments: The Role of Freshwater, Ice Sheets and Deglacial Warming for the Atlantic Meridional Overturning Circulation.” Polarforschung; 85.

Lowenstein, Tim K., Lawrence A. Hardie, Michael N. Timofeeff, and Robert V. Demicco. 2003. “Secular Variation in Seawater Chemistry and the Origin of Calcium Chloride Basinal Brines.” Geology 31 (10): 857–60.

Lowenstein, Tim K., Michael N. Timofeeff, Sean T. Brennan, Lawrence A. Hardie, and Robert V. Demicco. 2001. “Oscillations in Phanerozoic Seawater Chemistry: Evidence from Fluid Inclusions.” Science 294 (5544): 1086–88.

Marty, Bernard, Guillaume Avice, David V. Bekaert, and Michael W. Broadley. 2018. “Salinity of the Archaean Oceans from Analysis of Fluid Inclusions in Quartz.” Comptes Rendus Geoscience 350 (4): 154–63.

Matsumoto, Katsumi. 2017. “Tantalizing Evidence for the Glacial North Atlantic Bottom Water.” Proceedings of the National Academy of Sciences 114 (11): 2794–96.

Milanković, Milutin. 1998. Canon of Insolation and the Ice-Age Problem. Belgrade: Zavod za Udz̆benike i Nastavna Sredstva.

Murton, Julian B., Mark D. Bateman, Scott R. Dallimore, James T. Teller, and Zhirong Yang. 2010. “Identification of Younger Dryas Outburst Flood Path from Lake Agassiz to the Arctic Ocean.” Nature 464 (7289): 740–43.

Partin, J.W., T.M. Quinn, C.-C. Shen, Y. Okumura, M.B. Cardenas, F.P. Siringan, J.L. Banner, K. Lin, H.-M. Hu, and F.W. Taylor. 2015. “Gradual Onset and Recovery of the Younger Dryas Abrupt Climate Event in the Tropics.” Nature Communications 6 (September).

Pearce, Christof, Marit-Solveig Seidenkrantz, Antoon Kuijpers, Guillaume Massé, Njáll F. Reynisson, and Søren M. Kristiansen. 2013. “Ocean Lead at the Termination of the Younger Dryas Cold Spell.” Nature Communications 4 (April): 1664.

Pena, Leopoldo D., and Steven L. Goldstein. 2014. “Thermohaline Circulation Crisis and Impacts during the Mid-Pleistocene Transition.” Science 345 (6194): 318–22.

Poirier, Robert K., and Katharina Billups. 2014. “The Intensification of Northern Component Deepwater Formation during the Mid-Pleistocene Climate Transition: Mid-Pleistocene Deep Water Circulation.” Paleoceanography 29 (11): 1046–61.

Ronde, Cornel E. J. de, Dominic M. de R. Channer, Kevin Faure, Colin J. Bray, and Edward T. C. Spooner. 1997. “Fluid Chemistry of Archean Seafloor Hydrothermal Vents: Implications for the Composition of circa 3.2 Ga Seawater.” Geochimica et Cosmochimica Acta 61 (19): 4025–42.

Rooth, Claes. 1982. “Hydrology and Ocean Circulation.” Progress in Oceanography 11 (2): 131–149.

Rubey, William W. 1951. “Geologic History of Sea Water: An Attempt to State the Problem.” GSA Bulletin 62 (9): 1111–48.

Shakun, Jeremy D., Peter U. Clark, Feng He, Shaun A. Marcott, Alan C. Mix, Zhengyu Liu, Bette Otto-Bliesner, Andreas Schmittner, and Edouard Bard. 2012. “Global Warming Preceded by Increasing Carbon Dioxide Concentrations during the Last Deglaciation.” Nature 484 (7392): 49–54.

Sigman, Daniel M., Mathis P. Hain, and Gerald H. Haug. 2010. “The Polar Ocean and Glacial Cycles in Atmospheric CO2 Concentration.” Nature 466 (7302): 47–55.

Tarasov, Lev, and W.R. Peltier. 2005. “Arctic Freshwater Forcing of the Younger Dryas Cold Reversal.” Nature 435 (7042): 662–65.

Thiagarajan, Nivedita, Adam V. Subhas, John R. Southon, John M. Eiler, and Jess F. Adkins. 2014. “Abrupt Pre-Bølling–Allerød Warming and Circulation Changes in the Deep Ocean.” Nature 511 (7507): 75–78.

Tyrrell, Toby. 2013. On Gaia: A Critical Investigation of the Relationship between Life and Earth. Princeton: Princeton University Press.

Weiershäuser, L., and E. T. C. Spooner. 2005. “Seafloor Hydrothermal Fluids, Ben Nevis Area, Abitibi Greenstone Belt: Implications for Archean (∼2.7 Ga) Seawater Properties.” Precambrian Research 138 (1): 89–123.

Willeit, M., A. Ganopolski, R. Calov, and V. Brovkin. 2019. “Mid-Pleistocene Transition in Glacial Cycles Explained by Declining CO2 and Regolith Removal.” Science Advances 5 (4): eaav7337.

Let me know what you think!