Calibration, uncertainties and use of soybean crop simulation models for evaluating strategies to mitigate the effects of climate change in Southern Brazil

• Ano de defesa: 2016
• Autor(a) principal: Battisti, Rafael
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Glycine max L.; Adaptação à seca; Adaptation to drought; Cenários de clima futuro; Comparação de modelos; Crop management; Future climate scenarios; Manejo de cultura; Média de modelos; Models comparison; Models ensemble
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/11/11152/tde-03102016-162340/

The water deficit is a major factor responsible for the soybean yield gap in Southern Brazil and tends to increase under climate change. Crop models are a tool that differ on levels of complexity and performance and can be used to evaluate strategies to manage crops, according the climate conditions. Based on that, the aims of this study were: to assess five soybean crop models and their ensemble; to evaluate the sensitivity of these models to systematic changes in climate; to assess soybean adaptive traits to water deficit for current and future climate; and to evaluate how the crop management contribute to soybean yields under current and future climates. The crop models FAO - Agroecological Zone, AQUACROP, DSSAT CSM-CROPGRO-Soybean, APSIM Soybean, and MONICA were assessed. These crop models were calibrated using experimental data obtained during 2013/2014 growing season in different sites, sowing dates and crop conditions (rainfed and irrigated). For the sensitivity analysis was considered climate changes on air temperature, [CO2], rainfall and solar radiation. For adapting traits to drought, the soybean traits manipulated only in DSSAT CSM-CROPGRO-Soybean were deeper root depth, maximum fraction of shoot dry matter diverted to root growth under water stress, early reduction of transpiration, transpiration limited as a function of vapor pressure deficit, N2 fixation drought tolerance and reduced acceleration of grain filling period in response to water deficit. The crop management options strategies evaluated were irrigation, sowing date, cultivar maturity group and planting density. The estimated yield had root mean square error (RMSE) varying between 553 kg ha-1 and 650 kg ha-1, with d indices always higher than 0.90 for all models. The best performance was obtained when an ensemble of all models was considered, reducing yield RMSE to 262 kg ha-1. The crop models had different sensitivity level for climate scenario, reduction yield with temperature increase, higher rate of reduction of yield with lower rainfall than increase of yield with higher rainfall amount, different yields response with solar radiation changes due to baseline climate and model, and an asymptotic soybean response to increase of [CO2]. Combining the climate scenarios, the yield was affected mainly by reduction of rainfall (increase of solar radiation), while temperature and [CO2] interaction showed compensation effect on yield losses and gains. The trait deeper rooting profile had greater improvement in total production for the Southern Brazil, with increase of 3.3 % and 4.0 %, respectively, for the current and future climates. For soybean management, in most cases, the models showed that no crop management strategy has a clear tendency to result in better yields in the future if shift from the best management of current climate. This way, the crop models showed different performance against observed data, where the model parametrization and structure affected the response to alternatives managements to climate change. Although these uncertainties, crop models and their ensemble are an important tool to evaluate impact of climate change and alternatives to mitigation.