By Andy May
18O is a rare isotope of oxygen. The ratio of 18O to the normal 16O in foraminifera fossils (“forams”) can be used to estimate paleo-ocean temperatures. Higher values mean lower temperatures. A recent article on geologypage.com (here) led me to Bernard, et al., 2017, which has experimental data that suggest 18O concentrations can be altered in fossils by solid-state diffusion after fossilization. This can corrupt the measurement and the resulting calculated temperature. According to Bernard and colleagues, the 18O concentration alteration is visually imperceptible, so one cannot tell the fossil has been altered by visual inspection. If their results are valid, how will this impact our view of climate history?
Fractionation of 18O and 16O takes place during evaporation and precipitation, so another problem for this paleothermometer is glacier and polar ice, which delay the return of excess 16O in water vapor to the ocean. Correlations of δ18O to temperature are for an ice-free Earth, the formation of polar ice caps and mountain glaciers have an additional effect on δ18O. In figure 1, the δ18O record and the δ13C record are shown for the Cenozoic (65 million years ago to the present day).
Figure 1 (source: Zachos, et al., 2001)
The red temperature scale at the bottom of figure 1 is only applicable when there is no significant permanent polar ice, Zachos, et al. call this the ice-free temperature. It is applicable prior to about 33 million years ago when Antarctica and the North Pole were ice-free, at least in their respective summer months. Once permanent ice caps appear, they remove 16O from the oceans indefinitely and lead to a large positive δ18O in the oceans and in forams. Both records shown in figure 1 are for global deep-sea pelagic fossils, water depths greater than 1,000 meters.
In the experiments, Bernard, et al. exposed foraminifera tests (shells) to elevated pressures and temperatures when submersed in pure H218O. The forams were then monitored with a micro-mass spectrometer and they observed oxygen-isotope changes. They then attempted to model the isotope diffusion process during sediment burial. Their models suggest that on a time scale of tens of millions of years, the isotopic diffusion could cause ice-free δ18O calculated ocean temperature estimates to be higher than reality. For this reason, estimated temperatures in the Paleogene and Cretaceous may have been lower than we estimate today. The error due to diffusion is sensitive to the initial foram temperature, so this is particularly significant for forams that grew in the deep ocean or at the poles.
Thus, the older paleotemperatures in figure 1 may be too high, if Bernard, et al. are correct. How much and how does this effect our view of ancient climate? First, let’s discuss how ancient climates are currently reconstructed. There are many indicators of ancient climate. A brief list of common paleotemperature indicators, other than δ18O concentration, follows. Remember that over time the continents have moved, relative to the equator and the poles. The following list of indicators assumes that this has been considered, as it is in figure 2 for the Upper Cretaceous.
Mg/Ca ratio: The proportion of magnesium in foram shells is proportional to ocean temperatures, this is considered an accurate thermometer in unaltered Holocene sediments.
Alkenone Proxies: UK’37 is an alkenone index that is temperature sensitive. TEX86 is a very widespread and easily identified alkenone that is commonly used as a temperature proxy as far back as the Jurassic (200 to 145 million years ago). Finally, there is a newer alkenone temperature proxy called LDI or the long chain diol index. These three proxies do not always agree and can reflect temperatures at different water depths and different seasons (Smith, et al., 2013)
Plant leaf analysis: Some plant leaves change their shape and size depending upon the air temperature. With some species these characteristics can be used to infer temperature. This proxy can be used for both Cenozoic and Cretaceous temperatures. See Wilf, 1997.
Pollen: Certain plants only occur in narrow latitude bands, by measuring the amount of their pollen in sediments, they can be used as a temperature proxy. See here.
Lithology suite: For very old rocks this is the most reliable temperature proxy. Certain rocks tend to occur only in warm climates and others only in cold climates. The map and legend in figure 1 are from Christopher Scotese’s monograph “Some thoughts on global climate change: The transition from Icehouse to hothouse,” see here. In figure 1 the continents are shown in their Upper Cretaceous (100 to 66 million years ago) locations. To see the location of the continents anytime in the last 1.5 billion years, see this Youtube video by Prof. Christopher Scotese. This is an age discussed in Bernard, et al. The Upper Cretaceous rocks, suggest that the warm temperate climate, which currently is seen in the southeastern United States and in Italy and Spain, occurs much farther north and south than it does today.
Figure 2 (source here)
Many environmental factors have been found to affect oxygen isotope ratios, some are discussed by Spero, et al., 1997. The alkalinity, carbon dioxide and carbonate problems discussed by Spero, et al. will also affect the accuracy of very old (pre-Cenozoic) δ18O and δ13C paleotemperature reconstructions.
Bernard, et al. discovered that elevated temperatures and pressure cause a slow process of solid state diffusion of oxygen isotopes that result in the calculation of higher temperatures than were present when the foram test (shell) was formed. The problem is much worse in forams that formed in colder water, either on the ocean floor (benthic) or in high latitudes. It is much less of a problem in surface dwelling (planktonic) forams or in the tropics. The result is that the use of oxygen isotope temperature proxies can make the ocean temperatures appear more divergent, from pole to equator and from sea floor to ocean surface, in the Cretaceous and early Cenozoic than they were. For sediments less than 10 million years old or from the tropics, this process will have little effect. To see the impact that Bernard, et al. computed, see figure 3.
Figure 3, source Bernard, et al., 2017
The left side of figure 3 shows the modeled effect of the temperature difference (center scale is the δ18O estimated temperature for each line) by age and typical sediment burial conditions. The temperature labels, for each line, are the actual foram formation temperature. The difference between the δ18O estimate temperature and the original foram temperature is large in colder polar waters and for benthic (bottom dwelling) forams than for sea-surface (planktonic) forams in the tropics. This can also be seen in the model results shown in figure 3b. Further, younger sediments (less than 10 million years old) are not significantly affected, especially if they are of planktonic origin.
If the study is accurate and stands the test of time some paleo-climatological maps of the Cretaceous and early and middle Cenozoic may need to be revised. However, because many other temperature proxies are also used over these periods of time, it is unlikely to have a large effect on the overall picture. It will affect deep water ocean temperature estimates and estimates of the polar to equator temperature gradient. This work affects ocean paleotemperature estimates, it should not affect δ18O temperature estimates from ice cores. The ice core temperatures are too low for this sort of diffusion to take place. Figure 4 is a compilation of paleotemperatures from various sources over time (original author Goswami, 2011, PhD thesis):
Figure 4, average Pole to equator temperature gradient for the Cretaceous, source Christopher Scotese here.
There are many sources of paleotemperature data plotted in figure 4 and you can find them in the references in Scotese’s monograph. Some are oxygen isotope data, but not all. If what Bernard, et al. have discovered is true, the Cretaceous curve shown is probably closer to the faint gray line that shows the current temperature distribution. It cannot be the same, since we know there were no polar ice caps in the Cretaceous.
The Cretaceous and the Cenozoic up to 33 million years ago were much warmer than today, we can tell that because polar ice caps did not exist then and tropical ocean surface temperatures were higher. But, the paper by Bernard et al. will affect deep-ocean temperature estimates and polar-ocean temperature estimates. Bernard, et al. conclude:
“In conclusion, accounting for the diffusion-controlled burial-induced O isotope re-equilibration of fossilised benthic foraminifera removes the requirement for a strong, continuous and global cooling of the deep-ocean (on the order of 15 °C) during the late Cretaceous and the Paleogene. Furthermore, the present study suggests that the vertical and latitudinal temperature gradients of the late Cretaceous and Paleogene oceans were likely not very different from the current ones. Importantly, because O isotope re-equilibration through solid-state diffusion is a slow process, it likely had little impact on recent (<10 Ma) high frequency signals, such as the glacial to interglacial ﬂuctuations (driven by oscillations of Earth’s orbit and mainly related to ﬂuctuations of the seawater O isotope composition). However, these processes have potentially attenuated the relative amplitudes of older, transient signals, such as the Eocene-Oligocene transition or the Palaeocene-Eocene thermal maximum.”
I am certainly no expert in chemistry and have no way of knowing if this study and their conclusions are correct or not. The procedures and analysis seem reasonable to me, but that is as far as I can go. Certainly, the work has implications in paleoclimatology. I would hope to see others, with access to the very expensive equipment needed for studies like this, would attempt to duplicate this work and study it further. It is an important study in a critical area.