Climate extremes are major constraints on agricultural productivity, especially in tropical regions experiencing rapid expansion and intensification of soybean agriculture. This study analyzes spatiotemporal changes in soybean yields in response to droughts and heatwaves across highly productive municipalities in Brazil’s five macroregions from 1989 to 2020. By combining high-resolution meteorological data, satellite-based evapotranspiration estimates, and municipal-level crop yield data, we used standardized drought indices (Standardized Precipitation Index SPI, Standardized Precipitation Evapotranspiration Index SPEI) and a heat index (Warm Spell Duration Index WSDI) with spatiotemporal linear regression analyses to explore the links between climate variability and soybean yields across Brazil’s diverse agroclimatic zones. The results show a clear rise in the frequency and severity of compound drought–heat events, especially in the Northeast and South frontiers, where yield sensitivity to hydroclimatic stress is highest. Municipal-level linear regression analyses and spatial patterns indicate that short-term dry events, rather than long-term climate trends, are the main drivers of recent yield variability, with notable spatial spillover effects observed across municipalities. Cristalina and Bom Jesus, for example, exhibit significant negative trends (p < 0.05) in both SPEI-6 (−0.04 and −0.03) and SPI-6 (0.04 and −0.03), indicating a consistent drying tendency over time. Over the 30-year period, municipalities accumulated total soybean yield losses of 3292.3 thousand tonnes (kt), corresponding to an average reduction of 3.7% relative to 5-year detrended yield. These findings highlight the increasing vulnerability of rainfed agriculture in Brazil and emphasize the critical role of seasonal timing, crop phenology, and regional climate patterns for effective climate risk management. This study provides empirical evidence linking combined extremes to agricultural performance and presents a scalable framework for early warning systems and for climate-resilient policy development.
Parisoto et al. (Wed,) studied this question.