Plate Tectonics and Climate during the Cenozoic – Watts Up With That?

By Andy May

In this post I examine the proxies used to compare CO₂ to temperature from 66 million years ago (Ma) until today and comment on the quality of the comparison. In addition, we look at the Cenozoic plate tectonic events that affected global climate. Figure 1 compares Westerhold et al.’s deep-sea d18O (the Oxygen-18 isotope anomaly, a temperature proxy) to the d13C (the Carbon-13 isotope anomaly), both measurements are from the same fossils, so they can be directly compared. One of the problems with many temperature/CO₂ plots is often they are from different sources and locations and due to dating errors and differing temporal resolutions, they are not directly comparable. While d13C is not a direct CO₂ estimate, it is related to the CO₂ concentration in the deep ocean. Atmospheric and ocean CO₂ concentration estimates are compared to d13C in figure 2.

Major plate tectonic events are noted in figure 1 and a conversion from d18O to deep sea temperature is given in blue on the left. The highest temperatures in the Cenozoic are from the early Eocene (~56-48 Ma) when deep sea temperature exceeded 12°C higher than today. This was accompanied by a dramatic drop in deep-sea CO₂. As already mentioned, d13C is not an estimate of CO₂ concentration, but related to it. Proxy estimates of CO₂ from Rae, et al. are compared to Westerhold’s d13C estimates in figure 2.

The match in figure 2 is not great and both datasets have problems, but the similarities in trends are obvious. The estimates of CO₂ concentration reported by Rae, et al. are discontinuous and from a variety of proxies that are dated by many different authors with many different techniques. It is clear from the scatter that the assumption that CO₂ is evenly distributed globally is not applicable at this compressed time scale. Notice the PETM (Paleocene-Eocene Thermal Maximum) carbon isotope excursion (CIE) event at ~56 Ma shows up dramatically in both records. This large divergence in the ratio of carbon-13 to carbon-12 is a prominent global rock-record phenomenon and a reliable geological time marker that occurred between 55.6 and 55.4 Ma. Possible reasons for the CIE and the PETM are discussed here. This geological event and the following warm period comprise the most dramatic climatic event in the Cenozoic.

One important event at the beginning of the PETM, between 56 and 55.6 Ma, was the North Atlantic Igneous Province or “NAIP” volcanism. This was a huge series of volcanic eruptions that accompanied the opening of the North Atlantic and placed over 5 km of lava between Greenland and northern Europe (Stokke, et al., 2020). It nearly turned the North Sea into a lake. But, regardless of the reasons for the PETM and the Early Eocene Climatic Optimum (EECO, ~56-48 Ma) period, they are very noticeable in the rock record and easily identifiable in geological sections all over the world.

After the EECO, deep-sea temperatures begin a long decline. At first CO₂ increases, but at the beginning of the Oligocene it begins to decline, with the decline becoming more dramatic in the Middle Miocene.

During the PETM and in the warm period that followed it, the continents were configured as shown in figure 3.

Notice that all the oceans are connected by a seaway in the lower and middle northern latitudes, it is marked with a white line. India is moving through the Indian Ocean and on its way to collide with Asia. The Arctic is probably isolated by land and the Southern Ocean is blocked by land masses connecting South America and Australia to Antarctica. This is the warmest planetary configuration and the EECO, is classified as a “hothouse” climate by both Christopher Scotese (Scotese, Song, Mills, & Meer, 2021) and Westerhold, et al. Hothouse climates have global average temperatures (land and ocean) above 20°C and there is no year-round ice on either pole.

Arctic sea surface temperatures (SSTs) during the EECO may have reached 24°C. Estimates of the global average SST today vary a bit, but HadSST4 estimates a global average of about 20.5°C and NOAA estimates about 19.7°C, so the EECO Arctic SST temperature was probably 4°C warmer than the global average today.

The next major event is when India collides with Asia, this occurs between 46 and 44 Ma as shown in figure 4. The collision began as early as 59 Ma, but marine fossils in Himalayan sediments don’t disappear until 45 Ma (Hu, et al., 2016).

Coincident with this collision is some modest cooling and an increase in CO₂. As the Himalayas grow after this collision, they begin to cause planetary waves (more specifically orographic gravity waves) that can dramatically affect Northern Hemispheric weather (Trenberth & Chen, 1988) and (Kuchar, et al., 2022). Planetary waves impact the northern polar vortex, which is a major determinant of Northern Hemispheric winter weather.

The next major event is the opening of the Drake Passage which connects the Southern Ocean all around Antarctica. This event is very gradual, but appears to be complete by 34 Ma, as shown in figure 5. Like most ocean passage openings or closings it is hard to pin down and estimates of when it opened vary from 49 to 17 Ma. Antarctic ice began to grow about 44 Ma, and by 34 Ma the ice cap is complete. This coincides with a dramatic decline in global temperature and a drop in CO₂.

The next major event occurs around 31 Ma when the eastern Mediterranean is cut off from the Indian Ocean as shown in figure 6. The timing of the separation of the Mediterranean and the Indian Ocean is often debated and could have happened as late as 14 Ma, we prefer an earlier closing, sometime between 31 and 24 Ma. Sedimentology suggests the latest possible closure date was 24 Ma.

Next, around 17 Ma, the North Atlantic fully opens and connects to the Arctic. Panama probably begins to close at this time restricting the connection between the Atlantic and the Pacific, and the western Mediterranean closes at Spain. The western Mediterranean might have closed as late as 6 Ma, but certainly it was severely restricted by 17 Ma. These events coincide with a dramatic drop in global temperatures and deep-sea CO₂. The events are circled in figure 7. The North Atlantic opening is completed by about 13 Ma, the full and permanent closing of Panama is not complete until around 3 Ma.

These very dramatic events coincide with a steep drop in temperature and CO₂ that ends the Middle-Miocene Climatic Optimum. The closing of the Isthmus of Panama takes a long time, and exactly when it finally closed is the subject of much debate (Coates & Stallard, 2013), but the closure is certainly complete by 3 Ma as shown in figure 8.

Long-term climate changes have many causes, but one of the major factors is plate tectonics and continental drift. When the continents and oceans are oriented north-south as they are today, which restricts west-east (zonal) air flow and encourages north-south (meridional) air flow, the world is colder. The opposite is the case when west-east flow is encouraged by open ocean connections in the mid- to low-latitudes as shown in figure 3.

Another major influence on long-term climate change are the Milankovitch cycles (also see here). The influence of plate tectonics on climate change is very long-term, on the order of tens of millions of years, whereas the Milankovitch cycles work on the order of hundreds of thousands of years. Shorter periods of change are normally related to changes in the Sun itself, these work on periods shorter than a few thousand years.

In the Westerhold study where the excellent data plotted in figure 1 came from, the authors noticed a strong correlation between the astronomical Milankovitch cycles of 21, 41, 100, and 405 thousand years (kyr) length and patterns in their global deep-sea d18O and d13C data. Because the repeating Milankovitch astronomical cycles are computable and more reliable and accurate than any other dating technique, they used them to sequence and date the data plotted in figure 1. Their description of how this worked is in section 5 (“Astrochronology”) of their supplementary materials. For records older than 20 Ma only the longer eccentricity cycles could be used. The most prominent and stable cycle was the 405 kyr eccentricity cycle.

Westerhold, et al. conclude that their chronology is accurate to ±100 kyrs for the Pleistocene and Eocene, ±50 kyrs for the Oligocene, ±10 kyrs for the Miocene and Pleistocene. This sort of accuracy is remarkable if true, and it seems reasonable given their technique.

Comparing the known Cenozoic climate changes with Scotese’s plate tectonic reconstruction shows a coincidence of major climate changes with large geological events on a scale of many millions of years. Thus, it is easy, and logical, to conclude that the geological events caused the longer-term changes. I found it very encouraging that Westerhold, et al. could “see” the Milankovitch astronomical cycles in their deep-sea fossil records so clearly that they could be used for dating them. One of the biggest problems with comparing CO₂ records to temperature records is that the CO₂ records are made using different samples that must be dated separately from the temperature proxy samples. This gives me a lot more confidence in the d13C data in figures 1 and 2 than in the Rae, et al. data shown in figure 2. Further the prominent gaps in the Rae et al. CO₂ data leaves too much to the imagination.

Westerhold’s deep-sea d13C proxy is not a direct CO₂ proxy, but it can be paired directly with the d18O temperature proxy, and it is continuous. These characteristics make it superior to other CO₂ proxies in my opinion.

I should note that the exact timing of the major plate tectonic events discussed in this post is the subject of furious debate in the geological community (Hu, et al., 2016; Torfstein & Steinberg, 2020, Coates & Stallard, 2013). The precise dates when India collided with Asia, the Isthmus of Panama closed, or the North Atlantic opened to the Arctic are not known. They occured over millions of years and different geological studies can reasonably provide different dates depending upon the data used. Thus, the dates given in this study are just based on my best judgement and are open for debate.

Download the bibliography here.

Related