Reply to Meerburg et al.: Growing areas in Brazil and the United States with similar exposure to extreme heat have similar yields

In response to the letter from Meerburg et al. (1), we examine whether farmers in Brazil have been more successful adapting to extremely hot temperatures than farmers in the United States. We find that soybean yields in the state of Mato Grosso (Brazil) are comparable to yields in Southern Illinois, and so are temperatures >30 °C. Thus, the results from Brazil are consistent with our earlier findings from the United States (2).

There are some important differences between Brazil and the United States. For example, solar radiation is less seasonal in Brazil due to the closer location to the equator, which allows for an extended growing season and double cropping. Despite this, we find comparable yields for regions in Brazil and the United States with similar exposure to extremely hot temperatures, contrary to the assertion by (1).

A key point of our original study was that any statistical analysis of the link between weather (or climate) and crop yields must account not only for the entire distribution of temperature outcomes, but also for the specific locations where crops are grown. Weather records should be representative of the growing area, and yields should be representative of average growing conditions. Meerburg et al. compare yields in a particular year with favorable weather conditions (2008) in Mato Grosso, the Brazilian state with the highest average soybeans yields, to average yields in the entire United States. We find this comparison misleading. Some areas of the United States have much greater exposure to extreme heat than Mato Grosso and vice versa.

We contrast Mato Grosso to Illinois, which is one of the most productive states in the United States. Average maximum daily summer temperatures in the southern part of Mato Grosso are indeed higher than in Illinois, but a satellite scan reveals that no soybeans are grown in that area. Once maximum temperatures are averaged over the area where soybeans are actually grown, the average exposure to degree days >30 °C in Mato Grosso in 1961–1990 is similar to the southern half of Illinois, as are average yields between 1990–2005. Thus, there is no contradiction between Mato Grosso and our earlier findings for the United States. Additional documentation is available from the authors.

Meerburg et al. argue that yields in the United States have been trending upward and are expected to do so in the future. We agree with this point but emphasize that we controlled for it in our earlier analysis. Increasing state-specific yield trends are highly significant in all of our regression specifications. At the same time, the relative influence of extreme heat was found to be the same in the earlier and later years of our sample. This indicates that breeders have been successful at raising average yields but not, so far, at significantly improving heat tolerance. To clarify: yields are predicted to decline 30–82% compared to what they would have been without climate change. These predictions hold growing locations fixed and exclude CO₂ fertilization and possible future adaptations.