ANÁLISE RADIOGRÁFICA E MULTIESPECTRAL NA AVALIAÇÃO DA QUALIDADE DE SEMENTES DE Handroanthus impetiginosus

Abstract

Image analysis techniques are non-destructive, rapid, and objective alternatives to evaluate the quality of forest seeds. The lack of fast methods that do not compromise the seed for assessing the physiological quality of forest seeds hinders the efficiency of reforestation and conservation programs. Thus, the objective of this work was to evaluate the quality of purple ipe seeds (Handroanthus impetiginosus) using X-ray and multispectral image analysis techniques, integrated with physiological and morphometric parameters, to validate these complementary methodologies. Initially, with the seeds identified and numbered, radiographic images were obtained to classify the seeds into three classes according to the degree of filling of the embryonic cavity. Subsequently, with the same seeds, multispectral images were acquired in the near-infrared bands (880, 940, and 970 nm). With the seedlings from the emergence test, gas exchange analysis and chlorophyll A and B measurement were performed. Tests such as germination, seedling length, and electrical conductivity were also carried out. Subsequently, multivariate analysis of the data was performed using Principal Component Analysis (PCA). It was concluded that 80% of the evaluated seeds presented embryonic cavity filled by more than 50% (Class 1), corresponding to seeds that originated normal seedlings, while seeds with partial or absent filling were associated with abnormal seedlings and non-viable seeds, respectively. Viable seeds showed lower reflectance in NIR bands, a pattern associated with higher lipid content in tissues, and PCA efficiently integrated spectral, germinative, and photosynthetic variables, explaining 70.98% of the total variance in the first two components.