Advanced search
Start date

Chlorophyll fluorescence and multispectral imaging for carrot and tomato seed quality evaluation

Full text
Patrícia Aparecida Galletti
Total Authors: 1
Document type: Master's Dissertation
Press: Piracicaba.
Institution: Universidade de São Paulo (USP). Escola Superior de Agricultura Luiz de Queiroz (ESALA/BC)
Defense date:
Examining board members:
Clíssia Barboza da Silva; Haynna Fernandes Abud; Laércio Junio da Silva
Advisor: Clíssia Barboza da Silva

Optical methods have been extensively explored for assessing seed quality, in order to meet the growing demand from the agricultural and food industry. Chlorophyll fluorescence and multispectral imaging are rapid, non-destructive and accurate technologies that can generate consistent information on different seed quality traits. This research aimed to investigate a new approach using recent methods to analyze the physiological potential of tomato and carrot seeds. We used tomato seeds from Gaúcho and Tyna cultivars represented by three and four lots, respectively, and carrot seeds from Brasília and Francine cultivars, each represented by four lots. This research was divided into five stages: 1) physiological potential tests (germination and vigor); 2) chlorophyll fluorescence analysis at 620/730 nm, 630/700 nm, 645/700 nm and 660/700 nm excitation/emission combinations; 3) multispectral imaging at 19 wavelengths (365 to 970 nm); 4) X-ray imaging; 5) photosynthesis, chlorophyll fluorescence and chlorophyll index of seedlings. Chemometrics methods based on principal component analysis (PCA), random forest (RF) and quadratic discriminant analysis (QDA) were applied to the multispectral data. Chlorophyll fluorescence images at 620/730 nm (tomato) and 660/700 nm (carrot) allow the discrimination of seed lots with different physiological potential, in addition, the combinations of 620/730 nm (tomato) and 645/700 nm (tomato and carrot) were feasible to discriminate cultivars. Using a PCA method, the multispectral data discriminated carrot cultivars, but not tomato cultivars. The RF algorithm revealed the most meaningful wavelengths for seed vigor classification: 365, 570, 590, 660 and 690 nm in tomato seeds, and 365, 405, 430, 940 and 970 nm in carrot seeds. The QDA model based on the five most meaningful wavelengths assigned by the RF algorithm showed high precision in distinguishing seeds with higher-vigor from seeds with lower-vigor, with accuracies varying between 89 and 94% in tomato seed lots and between 86 and 97% in carrot seed lots. Immature tomato and carrot seeds showed a higher proportion of internal empty spaces observed by the radiographic images, higher chlorophyll fluorescence and higher multispectral reflectance. Tomato seeds with lower-vigor produced seedlings with reduced photosynthesis capacity, higher chlorophyll fluorescence and lower chlorophyll index. In conclusion, chlorophyll fluorescence and chemometric-based multispectral imaging are efficient tools for non-destructive and reliable prediction of the physiological potential of tomato and carrot seed lots. In view of the growing demand in the agricultural and food industry, these new approaches could be potentially used in quality assurance programs through different seed production stages, with rapid, objective and accurate ranking of seed lots. (AU)

FAPESP's process: 18/24777-8 - Chlorophyll fluorescence and multispectral image analysis to evaluate the quality of carrot and tomato seeds
Grantee:Patrícia Aparecida Galletti
Support type: Scholarships in Brazil - Master