Tropical deforestation has global impacts, contributing to climate change and biodiversity loss. However, efforts to address deforestation rely on local land-use decisions, which are shaped by the regional socio-economic, environmental, and policy context. The project goals were to understand how deforestation in the Brazilian Amazon influences the regional climate; the consequences of climate variability for agricultural production and household livelihoods; and how individual and policy responses to economic and environmental change create feedbacks to land-use and water scarcity. We created a unique database integrating socio-economic and geophysical variables at the property scale for 1,300 properties in the state of Rondonia, and at the municipality scale for the Brazilian Legal Amazon. The data come from diverse sources, including a panel survey of farm households; field measurements of discharge for a range of stream sizes; calibrated satellite data on land cover and vegetation condition, burned area, rainfall, temperature, and surface water; existing geospatial databases on soil texture, fertility and hydrogeology; agricultural census data on milk yield; and qualitative interviews with farmers and government officials. We developed models to simulate the impacts of deforestation on regional climate, and the impacts of policy and geophysical characteristics on farm-level management decisions.

We find significant effects of potential deforestation on regional climate conditions. Clearing current protected areas in Rondonia would decrease dry season rainfall up to 30% in some agricultural regions of the state, while deforestation of unprotected areas throughout the Brazilian Amazon would reduce rainfall in Rondonia by around 20%. These changes, combined with projected increases in temperatures, would create considerable risks for agriculture. We use integrated policy-climate modeling to project outcomes under alternative scenarios for Federal forest legislation and environmental spending. We observe 3.5 times more deforestation under a scenario with low enforcement of conservation policies than a scenario with high enforcement, with effects concentrated on the active deforestation frontier and along roads. This results in warmer and drier meteorological conditions in the areas deforested under the low-enforcement scenario. In addition to their direct climate effects, forests also enhance resilience to global climate conditions by providing a greater share of atmospheric moisture in Rondonia during periods of continental drought. 

Individual farmers are affected by changes in the regional climate. For example, the exceptional drought experienced in our study site in 2024 reduced surface water flows in small streams by 20-100%, with the largest impacts on the smallest streams. Farmers perceive water stress in relation to both the absolute amount of water available and the amount relative to what is typical on their property, particularly when the length or intensity of the dry season changes. Based on year-to-year weather variation, we find that increases in temperature and decreases in rainfall reduce milk yields at the municipal scale, consistent with previous literature in other climatic regions. This implies that improvements in productivity will be necessary to sustain dairy production in the future. Farmers adapt to chronic seasonal water stress by constructing water infrastructure such as wells, dams and ponds; adjusting cattle numbers; and intensifying production, e.g. with supplemental feeding. Irrigation of pasture remains uncommon. Farmers adapt to variability of dry-season rainfall by diversifying production. Farm productivity and land management choices are also influenced by environmental and socio-economic factors other than climate. Remote-sensing estimates of vegetation condition suggest that pasture productivity varies strongly by rock type and soil type, which in turn affects fertilizer use and land-use trajectories. Regional infrastructure and access to information can accelerate the pace of intensification. For example, we find that connection to the electrical grid enabled farmers to increase their incomes by investing in technologies such as refrigeration, and social media and agricultural extension visits significantly boosted the adoption of more intensive pasture management practices. 

This project had significant broader impacts on students and international collaboration. It provided interdisciplinary and international research opportunities for 12 undergraduate students, 18 graduate students and two postdoctoral researchers based in the US. Five of the seven PhD students and postdocs funded through the project were female, and four of these were Latina, supporting retention of under-represented early-career researchers in STEM. We developed deep, sustained relationships between faculty in the US and Rondonia, Brazil through co-authoring papers, co-teaching classes, co-advising graduate students and extended visits to one another's institutions. We expanded capacity for quantitative research in the Brazilian Amazon with participation of over 100 Brazilian students in field-based data collection and training workshops that enhanced their theoretical and practical knowledge of statistical methods and replication practices, expanded their professional networks, and supported their career development with opportunities to publish replication reports. Project team members exchanged knowledge, shared data, and co-developed models of farmer responses to land-use policies with local government agencies in Rondonia, and disseminated findings to international academic audiences with presentations at professional meetings and publications in peer-reviewed journals.

Last Modified: 12/20/2024
Modified by: Katrina Mullan