Temperate rainforests near the South Pole during peak Cretaceous warmth.

The mid-Cretaceous period was one of the warmest intervals of the past 140 million years1-5, driven by atmospheric carbon dioxide levels of around 1,000 parts per million by volume6. In the near absence of proximal geological records from south of the Antarctic Circle, it is disputed whether polar ice could exist under such environmental conditions. Here we use a sedimentary sequence recovered from the West Antarctic shelf-the southernmost Cretaceous record reported so far-and show that a temperate lowland rainforest environment existed at a palaeolatitude of about 82° S during the Turonian-Santonian age (92 to 83 million years ago). This record contains an intact 3-metre-long network of in situ fossil roots embedded in a mudstone matrix containing diverse pollen and spores. A climate model simulation shows that the reconstructed temperate climate at this high latitude requires a combination of both atmospheric carbon dioxide concentrations of 1,120-1,680 parts per million by volume and a vegetated land surface without major Antarctic glaciation, highlighting the important cooling effect exerted by ice albedo under high levels of atmospheric carbon dioxide.

Decoupling of greenhouse gas emissions from global agricultural production: 1970-2050.

Since 1970 global agricultural production has more than doubled; contributing ~1/4 of total anthropogenic greenhouse gas (GHG) burden in 2010. Food production must increase to feed our growing demands, but to address climate change, GHG emissions must decrease. Using an identity approach, we estimate and analyse past trends in GHG emission intensities from global agricultural production and land-use change and project potential future emissions. The novel Kaya-Porter identity framework deconstructs the entity of emissions from a mix of multiple sources of GHGs into attributable elements allowing not only a combined analysis of the total level of all emissions jointly with emissions per unit area and emissions per unit product. It also allows us to examine how a change in emissions from a given source contributes to the change in total emissions over time. We show that agricultural production and GHGs have been steadily decoupled over recent decades. Emissions peaked in 1991 at ~12 Pg CO2 -eq. yr(-1) and have not exceeded this since. Since 1970 GHG emissions per unit product have declined by 39% and 44% for crop- and livestock-production, respectively. Except for the energy-use component of farming, emissions from all sources have increased less than agricultural production. Our projected business-as-usual range suggests that emissions may be further decoupled by 20-55% giving absolute agricultural emissions of 8.2-14.5 Pg CO2 -eq. yr(-1) by 2050, significantly lower than many previous estimates that do not allow for decoupling. Beyond this, several additional costcompetitive mitigation measures could reduce emissions further. However, agricultural GHG emissions can only be reduced to a certain level and a simultaneous focus on other parts of the food-system is necessary to increase food security whilst reducing emissions. The identity approach presented here could be used as a methodological framework for more holistic food systems analysis.