Forests, through the regulation of regional water balances, provide a number of ecosystem services, including water for agriculture, hydroelectric power generation, navigation, industry, fisheries, and human consumption. Large-scale deforestation triggers complex non-linear interactions between the atmosphere and biosphere, which may impair such important ecosystem services. This is the case for the Southwestern Amazon, where three important river basins (Juruá, Purus, and Madeira) are undergoing significant land-use changes. Here, we investigate the potential impacts of deforestation throughout the Amazon on the seasonal and annual water balances of these river basins using coupled climatic and hydrologic models under several deforestation scenarios. Simulations without climate response to deforestation show an increase in river discharge proportional to the area deforested in each basin, whereas those with climate response produce progressive reductions in mean annual precipitation over all three basins. In this case, deforestation decreases the mean annual discharge of the Juruá and Purus rivers, but increases that of the Madeira, because the deforestation-induced reduction in evapotranspiration is large enough to increase runoff and thus offset the reduction in precipitation. The effects of Amazon deforestation on river discharge are scale-dependent and vary across and within river basins. Reduction in precipitation due to deforestation is most severe at the end of the dry season. As a result, deforestation increases the dry-season length and the seasonal amplitude of water flow. These effects may aggravate the economic losses from large droughts and floods, such as those experienced in recent years (2005, 2010 and 2009, 2012, respectively).

Bioenergy from sugarcane production is considered a key mitigation strategy for global warming. Improving its representation in land surface models is important to further understand the interactions between climate and bioenergy production, supporting accurate climate projections and decision‐making. This study aimed to calibrate and evaluate the Joint UK Land Environment Simulator (JULES) for climate impact assessments in sugarcane. A dataset composed of soil moisture, water and carbon fluxes and biomass measurements from field experiments across Brazil was used to calibrate and evaluate JULES‐crop and JULES‐BE parametrizations. The ability to predict the spatiotemporal variability of sugarcane yields and the impact of climate change was also tested at five Brazilian microregions. Parameters related to sugarcane allometry, crop growth and development were derived and tested for JULES‐crop and JULES‐BE, together with the response to atmospheric carbon dioxide, temperature and low‐water availability (CTW‐response). Both parametrizations showed comparable performance to other sugarcane dynamic models, with a root mean squared error of 6.75 and 6.05 t ha⁻¹ for stalk dry matter for JULES‐crop and JULES‐BE, respectively. The parametrizations were also able to replicate the average yield patterns observed in the five microregions over 30 years when the yield gap factors were taken into account, with the correlation (r) between simulated and observed interannual variability of yields ranging from 0.33 to 0.56. An overall small positive trend was found in future yield projections of sugarcane using the new calibrations, with exception of the Jataí mesoregion which showed a clear negative trend for the SSP5 scenario from the years 2070 to 2100. Our simulations showed that an abrupt negative impact on sugarcane yields is expected if daytime temperatures above 35°C become more frequent. The newly calibrated version of JULES can be applied regionally and globally to help understand the interactions between climate and bioenergy production.

Process-based models (PBM) are important tools for understanding the benefits of Integrated Crop-Livestock Systems (ICLS), such as increasing land productivity and improving environmental conditions. PBM can provide insights into the contribution of agricultural production to climate change and help identify potential greenhouse gas (GHG) mitigation and carbon sequestration options. Rehabilitation of degraded lands is a key strategy for achieving food security goals and can reduce the need for new agricultural land. This study focused on the calibration and validation of the DayCent PBM for a typical ICLS adopted in Brazil from 2018 to 2020. We also present the DayCent parametrization for two forage species (ruzigrass and millet) grown simultaneously, bringing some innovation in the modeling challenges. We used aboveground biomass to calibrate the model, randomly selecting data from 70% of the paddocks in the study area. The calibration obtained a coefficient of determination (R²) of 0.69 and a relative RMSE of 37.0%. During the validation, we used other variables (CO₂ flux, grain biomass, and soil water content) measured in the ICLS and performed a double validation for plant growth to evaluate the robustness of the model in terms of generalization. R² validations ranged from 0.61 to 0.73, and relative RMSE from 11.3 to 48.3%. Despite the complexity and diversity of ICLS results show that DayCent can be used to model ICLS, which is an important step for future regional analyses and large-scale evaluations of the impacts of ICLS.