Análise temporal da sinalização elétrica em plantas de soja submetidas a diferentes perturbações externas

• Ano de defesa: 2017
• Autor(a) principal: Saraiva, Gustavo Francisco Rosalin
• Orientador(a): Souza, Gustavo Maia
• Banca de defesa: Pereira, Danillo Roberto , Vieira, Luiz Gonzaga Esteves , Costa, Ernane José Xavier , Von Zuben, Fernando José
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade do Oeste Paulista
• Programa de Pós-Graduação: Doutorado em Agronomia
• Departamento: Doutorado em Agronomia
• País: Brasil
• Palavras-chave em Português: Algoritmos de classificação, Complexidade, Eletrofisiologia Vegetal, Entropia Aproximada.
• Palavras-chave em Inglês: Classification algorithms, Complexity, Plant Electrophysiology, Approximate Entropy.
• Área do conhecimento CNPq: CIENCIAS AGRARIAS::AGRONOMIA
• Link de acesso: http://bdtd.unoeste.br:8080/jspui/handle/jspui/1087

Plants are complex organisms with dynamic processes that, due to their sessile way of life, are influenced by environmental conditions at all times. Plants can accurately perceive and respond to different environmental stimuli intelligently, but this requires a complex and efficient signaling system. Electrical signaling in plants has been known for a long time, but has recently gained prominence with the understanding of the physiological processes of plants. The objective of this thesis was to test the following hypotheses: temporal series of data obtained from electrical signaling of plants have non-random information, with dynamic and oscillatory pattern, such dynamics being affected by environmental stimuli and that there are specific patterns in responses to stimuli. In a controlled environment, stressful environmental stimuli were applied in soybean plants, and the electrical signaling data were collected before and after the application of the stimulus. The time series obtained were analyzed using statistical and computational tools to determine Frequency Spectrum (FFT), Autocorrelation of Values and Approximate Entropy (ApEn). In order to verify the existence of patterns in the series, classification algorithms from the area of machine learning were used. The analysis of the time series showed that the electrical signals collected from plants presented oscillatory dynamics with frequency distribution pattern in power law. The results allow to differentiate with great efficiency series collected before and after the application of the stimuli. The PSD and autocorrelation analyzes showed a great difference in the dynamics of the electric signals before and after the application of the stimuli. The ApEn analysis showed that there was a decrease in the signal complexity after the application of the stimuli. The classification algorithms reached significant values in the accuracy of pattern detection and classification of the time series, showing that there are mathematical patterns in the different electrical responses of the plants. It is concluded that the time series of bioelectrical signals of plants contain discriminant information. The signals have oscillatory dynamics, having their properties altered by environmental stimuli. There are still mathematical patterns built into plant responses to specific stimuli.