Identificação aprimorada de seringais infestados por fitonematoides no município de Prata, MG, por sensoriamento remoto

• Ano de defesa: 2020
• Autor(a) principal: Domingos Júnior, Fausto Antônio
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Agronomia
• Palavras-chave em Português: Hevea brasiliensis; Nematoide de galha; Sentinel-2; Google Earth Engine; Geoprocessamento; Root knot nematode; Geoprocessing; Agronomia; CNPQ::CIENCIAS AGRARIAS::AGRONOMIA::FITOTECNIA
• Link de acesso: https://repositorio.ufu.br/handle/123456789/32236; http://doi.org/10.14393/ufu.te.2021.5007

The rubber tree has great socio-economic importance being the main source of natural rubber in the world. Due to the expansion of heveiculture, some concern has risen due to phyto-sanitary problems, among which are the nematodes. Meloidogyne exigua has been the most important species in rubber tree plantations due to its high proliferation, leading to nutrition deficiency symptoms, and causing plant death. Thus, this study developed a large scale diagnosis method for orchards infested by the root-knot nematode, improving NDVI with a time series of images available by the satellites Sentinel-2A and 2B. Also, field samples were taken to determine nematode population density and diversity in the orchards. The study area was the county of Prata, MG, and all the rubber tree orchards with closed canopies were mapped. Soil and roots geo-referenced samples were collected in two areas with confirmed infestation by M. exigua where the presence of the pathogen was actually observed. A script was generated in the platform Google Earth Engine to confirm the behavior of NDVI values in the orchards in Prata, MG, to select the best time series for classification. Subsequently, the classification script was created. The classification was based on the upper and lower limits of pixel values, established by the average and standard deviation of NDVI values observe in the training points. Field truth points were sampled to validate the classification map and confirmation of nematode density in the orchards. The methodology presented an accuracy of 87.5% and Kappa index of 0.75. The image time series considered as most adequate for this methodology was from January to June, and the month of October. Pratylenchus sp. was found in every sample, representing 77.58% of the population density. In contrast, Meloidogyne sp. was observed in 18 samples, representing 14.69% of the total population density. All other plant parasite nematodes found in the area represented only 7.72% of the population. It was estimated that 1,983.44 hectares have infestation signs by M. exigua, corresponding to 72.57% of the 2,733.06 hectares of rubber tree orchards evaluated in the county.