Influência de variáveis macroclimáticas sobre as principais doenças do arroz

• Autor(a) principal: Aguiar, Jordene Teixeira de
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• dARK ID: ark:/38995/0013000000h49
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/7405
• topic: Oryza sativa L.; Mudanças climáticas; Modelagem; Planejamento agrícola; Sensoriamento remoto; Oryza sativa L.; Climate change; Modelling; Crop management; Remote sensing; AGRONOMIA::FITOSSANIDADE

The influence of climatic variables on rice diseases was assessed in Brazil. Firstly, it was necessary to validate climate data from remote sensoring, retrieved from NASA’s database Prediction of Worldwide Energy Resource (POWER). POWER data were compared to climate records of surface stations from the National Institute of Meteorology (INMET). Climate data consisted of a time series (2004-2014) of monthly average temperatures and rainfall. Validation tests were carried out with Pearson’s coefficient of correlation and adjustment of linear regression models between the satellite data and surface stations. Further, data accuracy was checked according to average absolute error, root mean square deviation and concordance index. Monthly rainfall from most regions satisfactory correlated with Pearson coefficients between 0.75 and 0.95 (P<0.05). In contrast, maximum and minimum temperatures recorded by satellites showed irregular results that vary by region. In these cases, remote sensing did not detect extreme weather events, such as heavy rainfall or drought. Monthly rainfall comparisons also showed the most consistent results for all regions in accuracy tests. The endorsed data supported the next stage of this work, regarding the effects of climate variables on rice diseases. This investigation counted on a historical series of disease severities recorded in field tests, carried out between 1983 and 2014. Climatic data from INMET, EMBRAPA and NASA/POWER was arranged in a matrix of environmental variables, and tested for correspondence with disease severities recorded in 15 sites for at least eight years. Redundant climate variables were eliminated by principal component analysis. With structured data disposed in two datasets of climate (explanatory variables) and disease severity and productivity (response variables), canonical correlation analysis was performed (CCA) by location and by regions. The influence of climate on disease severity was demonstrated in only five sites, according to CCA models significant at 5%, with their first two axes explaining over than 50% of explanatory variables. In such sites, the total variation in disease severity was partially explained by climate variables. In the regional approach, climate variables did not significantly influence rice diseases in the North Region. Nevertheless, significant models demonstrated the correlation between climatic variables and disease in the Center-West and Northeast, despite the small percentage of explanation by the first two axes. In general, higher disease severity was related to rainfall and lower minimum temperatures during the reproductive stage of rice plots. In all cases, yield was not related to environmental variables.