Brazil’s leadership in soybean and maize production depends on predictable rainfall in the Amazon-Cerrado agricultural frontier. Here we assess whether agricultural expansion and intensification in the region are approaching a climatic limit to rainfed production. We show that yields decline in years with unusually low rainfall or high aridity during the early stages of crop development—a pattern observed in rainfed and irrigated areas alike. Although agricultural expansion and intensification have increased over time, dry–hot weather during drought events has slowed their rate of growth. Recent regional warming and drying already have pushed 28% of current agricultural lands out of their optimum climate space. We project that 51% of the region’s agriculture will move out of that climate space by 2030 and 74% by 2060. Although agronomic adaptation strategies may relieve some of these impacts, maintaining native vegetation is a critical part of the solution for stabilizing the regional climate.

巴西在大豆和玉米生产方面的领先地位取决于亚马逊-塞拉多农业前沿地区可预测的降雨量。在这里，我们评估该地区的农业扩张和集约化是否正在接近雨养生产的气候限制。我们表明，在作物发育的早期阶段，降雨量异常低或干旱程度高的年份产量会下降——这种模式在雨养和灌溉地区都观察到。尽管随着时间的推移农业扩张和集约化有所增加，但干旱事件期间的干热天气减缓了它们的增长速度。最近的区域变暖和干燥已经使 28% 的当前农业用地脱离了其最佳气候空间。我们预计，到 2030 年，该地区 51% 的农业将脱离气候空间，到 2060 年将达到 74%。