Bioelectrical pattern discrimination of Miconia plants by spectral analysis and machine learning

Abstract

AbstractWe have conducted an in loco investigation into the species Miconia albicans (SW.) Triana and Miconia chamissois Naudin (Melastomataceae), distributed in different phytophysiognomies of three Cerrado fragments in the State of São Paulo, Brazil, to characterize their oscillatory bioelectrical signals and to find out whether these signals have distinct spectral density. The experiments provided a sample bank of bioelectrical amplitudes, which were analyzed in the time and frequency domain. On the basis of the power spectral density (PSD) and machine learning techniques, analyses in the frequency domain suggested that each species has a characteristic biological pattern. Comparison between the oscillatory behavior of the species clearly showed that they have bioelectrical features, that collecting data is feasible, that Miconia display a bioelectrical pattern, and that environmental factors influence this pattern. From the point of view of experimental Botany, new questions and concepts must be formulated to advance understanding of the interactions between the communicative nature of plants and the environment. The results of this on-site technique represent a new methodology to acquire non-invasive information that might be associated with physiological, chemical, and ecological aspects of plants.HighlightIn loco characterization of the bioelectrical signals of two Miconia species in the time and frequency domain suggests that the species have distinct biological patterns.