Abstract
To build climate-resilient livestock systems, public policies must be informed by bioclimatic zoning, enabling region-specific interventions and more efficient resource allocation. This study aimed to conduct bioclimatic zoning for dairy cattle farming in the state of Maranhão, Brazil. Big data analysis techniques and predictive geostatistical modeling were applied to historical (2012-2023) and future climate scenarios under intermediate (RCP4.5) and high-intensity (RCP8.5) greenhouse gas emissions. Kriging maps of THI revealed a decreasing north-south thermal gradient, with values exceeding 80 during critical years. Milk yield losses were more pronounced in high-producing cows, reaching up to 5 kg/cow/day under extreme heat. Areas identified as drought-prone exhibited spatial patterns consistent with THI distributions. The projections indicate that, under the RCP 4.5 scenario, over 60% of Maranhão will exhibit average THI values between 78 and 81 by the end of the century. Under the RCP 8.5 scenario, large areas of the state are expected to reach THI values above 86. Under these conditions, estimated milk production losses may exceed 4 kg/cow/day for moderate-yielding animals and 9 kg/cow/day for high-yielding ones, respectively. The results reinforce the importance of bioclimatic zoning to support informed policymaking in the context of climate change.