Li, HaoInsua-Costa, DamiánKoppa, AkashGeirinhas, João LucasKeune, JessicaHolgate, Chiara M.Dominguez, FrancinaDeman, Victoria M.H.Teuling, Adriaan J.Miralles, Diego G.2026-02-112026-02-11ORCID:/0000-0002-3645-2856/work/205113899https://hdl.handle.net/1885/733805413Stable and predictable wet-season rainfall is crucial for soybean production in Brazil. However, climate and land-use changes, particularly Amazon deforestation, have increased rainfall variability in the region in recent decades. Here, we investigate long-term growing-season rainfall changes over two major soybean breadbaskets in Brazil from the perspective of atmospheric moisture transport. Utilising a novel moisture tracking framework based on a Lagrangian model guided by observations, we identify moisture source regions where evaporation contributed to rainfall over these breadbaskets. Furthermore, we quantify the relative contributions of source evaporation versus atmospheric (thermo)dynamics changes to downwind rainfall variability. Our results indicate that deforestation-induced evaporation declines have negatively impacted downwind rainfall in the breadbasket regions. However, strengthened circulation, evidenced by increased water vapour transport and low-level wind speeds consistent with decreased tree cover, has enhanced moisture transport from upwind regions (including Amazonia and the Atlantic Ocean) to the Brazilian soybean breadbaskets. This highlights the compensatory effects of deforestation on rainfall through decreased evaporation and altered atmospheric (thermo)dynamics, and how these effects may influence downwind soybean productivity in South America. Further understanding these interactions is critical for developing land management strategies to mitigate the agricultural impacts of climate change in the region.The computational resources and services used in this work were provided by the VSC (Flemish Supercomputer Center), funded by the FWO and the Flemish Government, Department of Economy, Science and Innovation (EWI). H.L. thanks the China Scholarship Council (CSC) for financial support under grant agreement 202006350051. D.I.C. acknowledges funding from the Bijzonder Onderzoeksfonds (BOF) from Ghent University (BOF24/PDO/027). C.M.H. acknowledges funding from the Australian Research Council (ARC) Centre of Excellence for the Weather of the 21st Century (CE230100012). D.G.M. acknowledges support from the European Research Council (ERC) Consolidator grant HEAT (101088405).10en© The Author(s) 2025.202510.1038/s41612-025-01152-3105011277110