Ian Harding (left) and James Eldrett.

Some 33.5 million years ago, a warm, humid greenhouse Earth gave way to a cooler planet that would eventually have ice blanketing its poles. The change, one of Earth's most profound climate shifts, marks the boundary between the Eocene and Oligocene epochs. However, during the past few years there has been some debate about when the first polar ice actually formed.

Two years ago, Ian Harding, James Eldrett and their colleagues reported physical evidence that ice was present during the Eocene–Oligocene transition in the region of modern-day Greenland, and that the ice may have formed as early as 38 million years ago — earlier than expected (J. Eldrett et al. Nature 446, 176–179; 2007). The finding caused a stir because atmospheric carbon dioxide concentrations for the period had been estimated at double pre-industrial levels, meaning that it was unlikely that they could have supported the development of large ice sheets.

The climate shift coincides exactly with the first appearance of ice on Greenland.

Harding, Eldrett and two new co-authors have now gathered sufficient evidence to reconstruct the climatic conditions in Greenland during the Eocene–Oligocene transition. They find that temperatures would have allowed the build-up of ice to begin during that time, putting the argument to rest.

The project first started in 2003, when Eldrett was finishing up his doctoral work in Harding's lab at the National Oceanography Centre in Southampton, UK. He was analysing a sediment core from a drilling site deep below the Norwegian–Greenland Sea, which turned out to be one of the most complete geological records of the Eocene–Oligocene boundary. An analysis of pebble striations in one sample led to the 2007 study that revealed the earliest evidence of ice on Greenland. For Eldrett, it was also the beginning of a stint of moonlighting, following up on the finding at nights and over weekends while pursuing a day job as a stratigrapher for Shell Exploration and Production UK in Aberdeen.

Eldrett and Harding had not previously been able to obtain climatic data from samples from the Norwegian-Greenland Sea site, which would support their 2007 finding of ice formation during the Eocene–Oligocene transition. This was because the samples were completely devoid of biogenic calcium carbonate, a chemical signature typically used to reconstruct past climates.

The samples were, however, incredibly rich in well-preserved microfossils of marine plankton, and of pollen and spores that had been washed into the sea from terrestrial plants. “Looking down the microscope for the first time, there were pristine examples of the diverse, organic-walled fossils characteristic of that time period,” recalls Eldrett.

The duo figured that they could capitalize on the pollen fossil record to track changes in vegetation and, by proxy, climate changes. “We knew which plant species were present, but we didn't know how to translate that into climate changes,” says Eldrett. “So our first port of call was David Greenwood, to try to get him on board.” Greenwood, a palaeobotanist at Brandon University in Manitoba, Canada, had previously reconstructed past climates in the Arctic Circle on the basis of plant fossils.

To link the fossils to climate, the team opted for a method that uses the nearest living relative of an ancient plant species to define the temperature and precipitation parameters likely to support that plant's growth. Their analysis reveals that during the early Eocene, Greenland's flora was similar to that of modern-day Florida, and included ferns, palms and cycads, as well as trees such as hickory and willow. Later Eocene sediments also contained fir, spruce and pine pollens from coniferous forests. In the Eocene–Oligocene boundary sediments, coniferous trees dominate, with frost-sensitive plants such as palms disappearing.

The team's climate reconstruction indicates that mean winter temperatures in Eocene Greenland were above 5°C, whereas during the Eocene–Oligocene transition they dropped to 0–2°C. As the Oligocene approached, Greenland began experiencing colder winters and greater annual temperature ranges (see page 969).

The climate shift coincides exactly with the first appearance of ice on Greenland, indicating that, even with higher CO2 levels, the initiation of continental ice formation in polar regions may at least have been possible. “This work is another page in the story of unravelling the driving mechanisms of profound climate change in the geological past,” says Harding.

With this study completed, Eldrett says he'll probably slow his moonlighting down. This summer, he plans to spend his weekends walking in the Scottish Highlands and cycling.

See also Nature 446, xiii; 2007.