"The work lends support to the notion that climate change will challenge agriculture in Africa," David Lobell of Stanford University, US, told environmentalresearchweb. "The results are not as disastrous as some have claimed, but they are definitely big enough to suggest that major adaptations are needed in this region."

One in three people in Sub-Saharan Africa is chronically hungry and national economies in the region are strongly dependent on agriculture.

The projections showed that the mean aggregate production changes for maize, sorghum, millet, groundnut and cassava were –22%, –17%, –17%, –18% and –8% respectively. Countries that currently produce the most crops tended to be projected to have the largest losses as climate changes. The researchers say this suggests that well-fertilized modern seed varieties are more susceptible to heat-related losses.

Rice and wheat are also a key source of calories in the region but the researchers did not include them in the study because they are often irrigated and data on irrigation is sparse.

Together with Wolfram Schlenker of Columbia University, Lobell started the analysis by looking at data from individual countries. "But [we] weren't having much success in seeing a clear effect of weather on crop production," he said. "By combining all countries into a panel dataset, we were able to see a much clearer signal of weather."

The pair, who used 16 climate-change models to predict future temperatures and precipitation in the region, believe that their analysis is relatively robust. Previous studies of how climate change could affect crops in Africa "generally made several assumptions that weren't well tested, and provided only a best guess of what would happen". The researchers say this meant that existing estimates did not inspire a lot of confidence.

"Our findings show a surprisingly robust picture of how weather affects yields in Africa," said Lobell. "By robust I mean it doesn't depend very much on whether we use different weather datasets, or different assumptions about how weather could affect yields. In all cases we project a negative impact of climate change without adaptation."

Possible climate adaptation measures include crop varieties with greater drought and heat tolerance, improved and expanded irrigation systems, rainwater-harvesting technologies, disaster-relief efforts and insurance programmes.

Now Lobell and Schlenker plan to look at impacts in other regions and to examine the potential for trade to buffer local production changes. "We will also try to evaluate new data as it becomes available, in order to refine our estimates," said Lobell.

The researchers reported their work in ERL.