The collision that changed the world

Fifty million years ago India collided with an island arc which rimmed Eurasia (Bouilhol et al., 2013; see Figure 1). This collision appears to have launched a ramp-like change in our planet’s environment which continues today (Figure 2). The oxygen isotope record stored in the shells of benthic foraminifera tells us that since that time the Earth’s polar regions have steadily cooled (Zachos et al., 2001). This cooling culminated in the permanent glaciation of Antarctica followed by the cyclic glaciation of North America and Europe. Based on measurements of the isotopic composition of lithium contained in the shells of planktonic foraminifera, a case is made that chemical weathering began a steady increase close to the time of the collision (Froelich and Misra, 2014). It continues today. In addition, evidence from halite–hosted fluid inclusions (Brennan et al., 2013) and from the magnesium content of ridge crest calcites (Coggon et al., 2010) suggest that Mg to Ca ratio in seawater steadily increased over the last several tens of millions of years (Figure 3). Further, the switch from calcitic to aragonitic reefs (Stanley and Hardie, 1999) and the increase in the magnesium content of reef aragonites (Anne Olery-Gothman, personal communication) adds weight to this conclusion. Although this large increase in Mg to Ca ratio remains loosely defined, it may have begun around the time of the collision.

Two other records undergo abrupt changes at the time of the collision (Figure 1). One is the sulfur isotope record for marine barite (Paytan et al., 2004) and the other is the carbon isotope record for amber (Tappert et al., 2013). Both undergo several per mil abrupt upward jumps centered close to 50 million years ago (see Figure 2). Although it is tempting to conclude that these jumps herald a sharp drop in atmospheric O₂ content, it is not at all clear that this is the correct interpretation. The abruptness of these jumps is puzzling. As shown in Figure 1, although the encounter with the Tethys subduction zone took place 50.2 ± 1.5 million years ago, that with the Asia landmass itself did not occur until 40.4 ± 1.3 million years ago. (Bouilhol et al., 2013). Hence, rather than being abrupt, the collision itself was spread over several million years.

The physical consequences of this collision were global. Not only did it lead to the creation of the Himalaya Mountains and the Tibetan Plateau, but it forced a global reorganization of the movement of the crustal plates. The sharp bend in the trace of the Emperor Seamounts occurred close to 50 million years ago (Sharp and Clague, 2006). Australia’s rate of northward drift increased dramatically and the opening of the Mariana back-arc basins commenced at that time (Reagan et al., 2013). Taken together, this reorganization of plate motions set the global environment on a new course.

One might ask why 50 million years ago and not 40 million years ago when the ⁸⁷Sr to ⁸⁶Sr ratio began its climb for it is widely accepted that this climb was a result of the India-Asia collision. It is my opinion that the late John Edmond had the correct explanation for this rise (Edmond, 1992). He called on the metamorphism of granites driven to great depth as a result of the collision to redistribute radiogenic strontium from chemical-erosion-resistant Rb-rich micas and amphiboles into chemical-erosion-prone Rb-poor plagioclase and calcite. If so, then it is reasonable that millions of years would have elapsed before these rocks were unroofed and exposed to weathering.

The most important record has yet to be mentioned. What did the atmosphere’s CO₂ content do during this time interval? The ¹³C record kept by alkenones is touted to be our most reliable CO₂ proxy (Zhang et al., 2013). Although it extends back only 38 million years, it tells us that between then and now the CO₂ content of the atmosphere dropped from a high of about 1500 µatm to somewhere between 200 and 400 µatm (Zhang et al., 2013). However, rather than undergoing a ramp-like decline, much of the drop appears to have occurred in a single step centered at about 25 million years ago. Since then this record suggests that the atmosphere’s CO₂ content has remained quite flat. It is interesting to note that the record based on the difference between the δ¹³C for marine CaCO₃ and that for marine bulk organic material has a quite different shape (Falkowski et al., 2004). It was flat before 60 million years ago and began a continuing decline somewhere between 60 and 40 million years ago (Figure 4). The steepest part of the Δδ¹³C decline occurs during the last 20 million years (Figure 4). As the alkenones used as a ¹³C proxy are produced only by the surface dwelling coccolithophorid Emiliania huxleyi, they are the favored proxy. By contrast, the precursors of the tiny residue of organic matter found in deep sea sediments have yet to be identified. Because of this, were it not for the fact that they yield a quite different time history for the increase in δ¹³C, I would not mention them. As much of the polar cooling has occurred during the last 20 million years, a decrease in CO₂ would be expected. Although the alkenone record shows no significant change, that for marine organic residue undergoes its steepest change during this time interval.

In this connection, a novel approach to the use of leaf stomata as an atmospheric CO₂ proxy was recently published (Franks et al., 2014). Combining the number and area of stomata with the ¹³C to ¹²C ratio in the leaf material, the authors calculate a stomatal resistance. This approach allows them to use leaves or needles from any C-3 plant rather than only those from a single long-lived species such as ginkgo. Using this approach, Franks et al., (2014) adequately reproduce the Mauna Loa CO₂ record and also the Antarctic ice core CO₂ record for the last 25,000 years. Preliminary application of their approach to fossil leaves confirm the alkenone-based finding that CO₂ has remained constant within the measurement error (±50 µatm) during the last 10 million years. However, the failure of the stomata resistance proxy to document a decrease in CO₂ centered at 8 million years ago is puzzling for it reinforces a long-standing problem. Based on the δ¹³C in horse teeth enamel, it has been demonstrated that, over one or two million years, horses on four continents went from eating C-3 to eating C-4 grass (Wang et al., 1994). The most reasonable explanation for this globe-encompassing transition is that it was brought about by a decrease in atmospheric CO₂ content. One explanation for the drop in the δ¹³C in marine CaCO₃ over the last 20 million years is a continuing increase in the extent of C-4 grasslands. Again, this could be the result of a steady decrease in the atmosphere’s CO₂ content.

As first pointed out by Walker et al. (1981), the CO₂ content of the atmosphere must be driven to that level where the supply of calcium to the ocean from the weathering of silicate rocks balances the release of CO₂ from the planet’s interior. This is necessary because, as indicated by the ¹³C to ¹²C record in marine CaCO₃, 80 or so percent of the carbon added to the ocean is removed as CaCO₃ (Broecker, 1970). If for some reason balance between the calcium and CO₂ delivery rates were to be disrupted, it would result in a change in the atmosphere’s CO₂ content. The consequent change in global temperature would serve as a feedback driving back into balance the rate of Ca delivery. So, if the cooling of the last 50 million years was the result of a drawdown of atmospheric CO_2, then there must have been a continuing tendency toward an oversupply of Ca which was compensated by a steady CO₂-driven cooling. Two explanations for this oversupply come to mind.

1) The rate of planetary outgassing of CO₂ has steadily declined. One possibility is that plate motions have slowed. However, reconstructions based on sea floor magnetics make this unlikely (Rowley, 2002). Another possibility might be that the amount of CaCO₃ contained in subducted marine sediments decreased as subduction shifted from the CaCO₃-rich sediments of the Tethys Sea to the CaCO₃-poor sediments of the Pacific Ocean (Edmond and Huh, 2003).

2) There has been a tendency for the release of calcium from igneous rocks to accelerate. One reason could be that some combination of mountain building and mountain glaciation has steadily increased the rate of mechanical erosion thereby increasing the supply of surface area available for chemical weathering (Froelich and Misra, 2014; Pogge von Strandmann and Henderson, 2015). This tendency has been counter balanced by a decrease in atmospheric CO₂ content.

The steady increase in the ocean’s ⁷Li to ⁶Li ratio is thought to reflect an increase in the extent of fractionation which occurs when the lithium released by chemical weathering of continental rocks is incorporated into clay minerals. As clay mineral formation occurs in marine basalts as well as in continental soils, one possibility is that this trend reflects the increase in the isotope fractionation associated with the long term cooling of the deep ocean. Another is that it reflects an increase in the split between lithium mineralized at ambient temperature and that mineralized at a high temperature (i.e., in mid-ocean-ridge hydrothermal systems). I would like to think that it is the latter and that it is a stand-in for the magnitude of calcium release from continental igneous rocks. However, much more will have to be learned about lithium isotope geochemistry before a meaningful interpretation can be made. Field studies such as that carried out by Pogge von Strandmann and Henderson (2015) are essential in this regard. However, the importance to this paper is that the rise in ⁷Li to ⁶Li ratio in the ocean began close to 50 million years ago.

The simultaneous upward jumps in the δ³⁴S in marine barite (Paytan et al., 2004) and in the δ¹³C of amber (Tappert et al., 2013) are difficult to explain. Were it not for the fact that both jumps occurred close to 50 million years ago, I would be tempted to downplay their importance. But they do suggest that something dramatic occurred at the time of the collision. Making the situation more complicated is the observation that during the 50 million years following the jumps, the δ³⁴S of marine barite has remained nearly the same while δ¹³C in amber has become ever more negative. Were the amber shift the result of changing atmospheric CO₂ content, then δ¹³C should have become ever more positive as is the case for alkenones. Tappert et al. (2013) suggest instead that this record records changes in the atmosphere’s O₂ content (i.e., a sharp drop at 50 million years ago followed by a continuous rise since that time). Their case is based on the observation that the rubisco enzyme reacts with O₂ as well as with CO₂. If the CO₂ produced by the reaction with O₂ were recycled, it would be doubly fractionated, lowering the δ¹³C of the organic matter. The problem with this explanation is that the laboratory experiments conducted to determine the impact of O₂ utilization have yielded carbon shifts far too small to account for the large amber shift (for example, see Berry et al., 1972). These experiments must be repeated.

It is tempting to opt for the O₂ explanation for it offers a way to explain the abrupt 4 per mil rise in δ³⁴S. A drop in O₂ would lead to an increase in the fraction of sulfur removed from the ocean as sulfide and hence the increase in the ³⁴S to ³²S ratio in marine SO₄. But, if so, it is difficult to explain the absence of a ramp-like change during the last 50 million years. Of course, the rise could also be explained by a change in the fractionation factor between oxidized and reduced sulfur. In either case, a one-time large jump in the isotope ratio is not easily explained. In this connection, using a technique developed by Burdett et al. (1989), an independent sulfur isotope record that has been produced in Jess Adkins’ laboratory at CalTech confirms that for barite (personal communication). It is based on the analysis of the trace amounts of SO₄ contained in foraminifer shells. However, the 4 per mil rise, rather than being close to instantaneous, is spread over a few million years.

Despite our inability to adequately interpret the records kept by these proxies, clearly the collision between India and Asia triggered environmental changes which continue today. Were this trend to be operative for another 30 or so million years, the Earth would once again become a “snowball”. Interesting in this regard is that the cooling has been latitude dependent. Hence, the portion of the planet’s surface which is habitable would likely gradually shrink toward the equator. The ¹⁸O record kept by planktic foraminifera suggests that the 10°C cooling experienced by the deep ocean was accompanied by only a 5°C or so cooling at the equator (Figure 5).

However, were humans to survive, they would certainly have the where-with-all to compensate for such a cooling. But, of course, if one considers the many ways in which we could do ourselves in, it is more likely that we won’t be there. In which case, the Earth will remain on its natural course.

© 2015 Broecker. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Discussions with Mike Bender, John Higgins, Kate Freeman, Steve Cande, Dennis Kent, Tim Lowenstein, and Dan Schrag have helped me to frame my thinking. Technical assistance by Elizabeth Clark, Patricia Catanzaro and Joan Totton is much appreciated.