What might explain the cooling of the Antarctic region at the end of the Eocene 34 Ma ago? Sage considers a recent attempt at an explanation from Nature, and remains unconvinced. There are still questions to be answered. (A bit technical in places)
Nature (30 July 2014) includes a letter entitled ‘Antarctic Glaciation caused Ocean Circulation Changes at the Eocene –Oligocene Transition’ (Goldner, Herold, Huber, 2014: Nature. doi:10.1038/nature13597.). The authors use ocean-atmosphere modelling to explore possible causes of the cooling of the world at this time – about 34 million years ago.
The authors begin by identifying two main hypotheses for ‘global cooling’ (as stated in the abstract) and/or for the ‘pattern of cooling’ (as per the main text) at and around the boundary between the Eocene and Oligocene geological periods – roughly 34.1 to 33.6 million years ago.
The first hypothesis, the Ocean Gateway hypothesis, concerns the opening of both the Drake Passage, located between the southernmost tip of South America and the Antarctic continent, and the Tasman Gateway, between Australia and Antarctica. The idea being that once these seaways became open, a circumpolar current could form, isolating the Antarctic from warmer ocean current from the tropics. This isolation being considered the cause of the first glaciation of the Antarctic which had up to that point had a temperate climate. The Antarctic climate of the Eocene was relatively warm despite the continent occupying the southern polar region, more or less as it does today. This hypothesis has received much support, and is often mentioned in popular reference works, such as Wikipedia, as well as in peer-reviewed journals.
CO2 – The Wonder Gas
The second hypothesis involves a change in atmospheric CO2. This assumes that a significant reduction in atmospheric CO2, and hence a reduced ‘greenhouse effect’, cooled the oceans and the globe. There is good evidence of a correlation between significant (although not unprecedented) falls in atmospheric CO2 and temperature falls at the time of the transition between the Eocene and the Oligocene.
These Won’t Do It
Neither of these hypotheses are fully satisfactory to the authors. Using the δ18O record for benthic foraminifers, obtained from cores of ocean sediments, which is used as a proxy for ocean temperatures, Goldner et al establish evidence for thermal stratification patterns for the South and (proto) North Atlantic Ocean of the time. They find that cooling caused by reduced atmospheric CO2 alone could not explain the patterns they find. Secondly, they consider that the magnitude of the changes in CO2 levels required by the CO2 hypothesis are unlikely, given the data available from CO2 proxies for the period.
They also argue that the tectonic processes that opened the two gateways would have been too slow, and in any case don’t really match in time, the accepted dates for the cooling of the Antarctic and that suggested by the foraminifer record.
The authors suggest that the formation of the ice sheet itself, when included in models of climate and ocean circulation, produces results which better match the observed data. Basically, an Antarctic ice sheet provides feedback processes from increasing albedo and raised Antarctic topography.
Modelling the Past
They perform four ‘experiments’ with models, adjusting variables as follows:
- Adding an ice sheet to the Antarctic land mass of roughly the same size as observed today. For this model, atmospheric CO2 is held constant at 1120 ppm, which they believe is representative for the later (but pre-cooling) Eocene.
- A model with no Antarctic ice sheet, but with atmospheric CO2 decreasing to 560 ppm. Again, they consider this value as representing the post-cooling early Oligocene.
- A model with both an Antarctic ice sheet and the reduced CO2
- A model with the two gateways open and then closed, all other variables constant (although they don’t clearly state at what levels CO2 are, or if an ice sheet is included or not).
Put simply, the authors modelling with the Antarctic ice sheet in place results in conditions most closely matching those derived from the benthic foraminifer data. These results are also similar to that achieved by other workers who used purely CO2 driven models, but the current authors achieved the result with lower assumed levels of atmospheric CO2. However, the modelled effects of opening the ocean gateways showed only weak changes in ocean circulation and heat transport, by comparison to the other models.
As an amateur – albeit one with a degree in Earth Science – I found the paper somewhat difficult to follow. The effects of just varying the CO2 levels from 1120 ppm to 560 ppm were not obvious from the figures, nor clearly described in the text. I had to look at the methods and figures, which are only available in the on-line version (£) to get the answers I was seeking.
According to the modelling, the effect of including an Antarctic ice sheet accounts for 70% of the observed δ18O at high southern latitudes, by implication the fall in CO2 levels represents the remaining 30%. Given the modelling of ocean currents, the reduction in CO2 explains 100% of the δ18O at high northern latitudes. The authors also say that the inclusion of the reduction in atmospheric CO2 also makes the modelled change in temperatures for both the surface and the deep ocean “more homogenous”.
What should be Made of This?
Having read the paper, I am reasonably convinced that the gateway hypothesis is unlikely to have been the cause of the Antarctic glaciation by itself. However, there are many other workers who might disagree and they must have time to respond.
However, I am unconvinced that this means, as the authors imply and as Dan Lunt – also writing in Nature (£) – more strongly states, that this means that the CO2 hypothesis is therefore entirely supported. The authors state that cooling has two main hypotheses, not that there are only two. Indeed, by putting in an Antarctic ice sheet as a given, Goldner et al are almost implying that something else is going on, even if one totally accepts the efficacy of changes in CO2 on global temperatures. Indeed, Lunt does accept this point. He also rightly says that combining the work of Goldner et al with models of ice sheet growth might produce more interesting results.
One of the feedbacks suggested by the authors whereby the change in circulation they modelled might exacerbate global cooling is by increasing upwelling in Antarctic regions. This improves productivity of the ocean, allowing the growth of sea plants and animals and ventilating the ocean better, thus drawing down more atmospheric carbon. (This effect was not actually in their model – indeed, as they were holding CO2 levels constant, it couldn’t have been). At no point, however, do they mention the physical effect of the increased solubility of CO2 in colder water, which in itself might explain the fall in atmospheric CO2 rather than the other way about.
The issue is that, while the model does produce cooling broadly in line with the data from the foraminifers, the reason for the ice sheet formation in the first place is still not answered. Without the assumed glaciation, but with the CO2 reduction, the modelled temperature anomaly at the ocean surface is about 2-3oC (extended fig 1b). Is this in itself enough to initiate a glaciation? Of course, this must depend on the actual starting temperatures given the topographic elevation of the Antarctic. Given that Mount Vinson is currently about 4900 m high, and would presumably be higher still without isostatic sinking due to the modern ice overburden, it’s almost certain that mountain glaciers were present in Antarctica during the warm Eocene, just as they are present today in Africa. It seems to me unlikely that these would be large enough to form the basis for the massive ice sheet required by Goldner et al’s model.
There is a recognised mass extinction event at the Eocene – Oligocene boundary. The causes of this are disputed. In general, global cooling by an unspecified initial cause, seems to be the preferred explanation, but it has been suggested that meteorite strikes may have had an influence. Perhaps this brought about a temporary ‘nuclear winter’ that allowed Antarctic mountain glaciers to grow to a point where the processes modelled by Goldner and his co-workers could operate? This is speculative, and I offer no opinion here, but it is not impossible.
Some workers have suggested that ice sheet growth might occur at CO2 levels higher than the minimum assumed by Goldner et al, although Goldner et al do say (in the on-line material) that their modelled results provide enough snowfall to maintain their initial ice sheet, they don’t say that it could actually form – because they assume it’s either there or it’s not. Also interesting is that the lowest CO2 levels they model are 560 ppm. This is much higher than today’s ~398 ppm, so their argument suggests that the Antarctic ice sheet should be stable now. Many climate change ‘alarmists’ would surely disagree.