Validação de análise radiográfica para exame de sementes infestadas com Sitophilus zeamais (Coleoptera: Curculionidae) em milho

• Autor(a) principal: Silva, Fabiano França da
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/29604
• topic: Sementes - Análise de Imagem; Gorgulho-do-milho - Detecção; Seeds - Image Analysis; Corn weevil - Detection; Fitotecnia

The examination of seeds infested or damaged by insects is one of the requirements established by the Brazilian standards for corn seed in order to release batches. The traditional methodology prescribed in the "Rules for Seed Analysis”, consisting of imbibition and individual cutting of seeds, is extremely time-consuming, leads to visual fatigue and jeopardizes the safety of the analyst. The technique of radiographic image analysis presents the potential for use in the detection of infested seeds and can be a viable alternative in relation to the traditional methodology. The objective of this work was to validate the methodology of radiographic analysis for the examination of infestations by the weevil Sitophilus zeamais (Coleoptera: Curculionidae) in corn seeds. Radiographic analyses were carried out in 10 laboratories, accredited with the MAPA (Ministry of Agriculture, Livestock and Supply) for this type of analysis, which received radiographs of seeds referring to four samples with different levels of infestation (0, 1, 3 and 5%). Data were analyzed according to ISTA (International Seed Testing Association) statistical procedures and complementary techniques, used to identify the presence of discrepant values (Hampel method) and outliers in the variances (Levene test for mean), to evaluate the effects of laboratories and levels (analysis of variance) and verify the repeatability, reproducibility, accuracy and robustness (critical limits of repeatability, reproducibility, Mandel h and k statistics). It was concluded that the methodology proposed for examination of weevil infestation in maize seeds, using radiographic analysis, is validated, since it shows exactness, robustness and precision within the critical limits of repeatability and reproducibility