Varying temporal and spatial effects of climate on maize and soybean affect yield prediction

The relationship between climatic factors and crop yields for maize and soybeans in 3 major producing countries (the United States, Brazil, and China) was analyzed statistically. Temporal changes in the climate−yield relationship were considered, and the temporal and spatial variations were evaluated. County-level data from 1980 to 2006 were collected for each country and allocated to 1.125° × 1.125° grids. Data were analyzed for temporal changes in the effect of climate on yields using the particle filtering method for each grid. The ‘current’ effect of temperature on crop yields was shown to be geographically symmetric around the optimal temperatures (19.51°C for maize and 20.66°C for soybeans) in the Northern Hemisphere. In the Southern Hemisphere, the ‘current’ effect of precipitation was more influential than that of temperature. The effects of these climatic factors changed over time. Whereas the negative effect of high temperatures has been mitigated around the corn belt of the United States during the last 3 decades, it has intensified in Brazil, northern China, and the southern United States for maize. Consequently, predicted future yields differed marginally depending on the relationship (past or current effect) used for the prediction, even when they were summarized on country scales. This study suggests that temporal changes in the relationship between weather and crop yields should be considered for better future predictions.