Studying the spectrometric features of forest seeds to improve sowing qualities: a retrospective cluster analysis of the scientific landscape trends

Forest seeds spectral data in the visible and infrared regions of electromagnetic radiation lengths quite effectively differentiate the origin, viability, types of seeds, their infestation with pests and diseases, the ability to absorb and lose water. The search for a method of seed testing that is both experimentally simple, fast and effective for predicting germination is necessary to increase the energy efficiency of forest nurseries in the production of planting material. The retrospective references systematization (N = 55, 1998-2023, terms [Scholar Query = seeds* AND (spectr* OR optic*) (properties OR features) AND analysis]) into clusters was carried out on the basis of eight performance criteria represented by rank variables. The level of similarity and difference between clusters is determined by the method of the most distant neighbors with the grouping of data by the square of the Euclidean distance. The most distant criterion from other criteria is the level of invasiveness of testing (the square of the Euclidean distance is 25, p < 0.05). Correlation analysis of nonparametric criteria indicates a direct strong interaction between the level of financial and organizational costs (Spearman coefficient ρ = 0.77; p = 0.0008), time costs and low machine learning capability (ρ = 0.725; p = 0.0008). In the future, it is planned to periodically supplement the set of systematic data to obtain an objective assessment of seed testing methods, as well as using a seed passport to evaluate the relationship of RGB spectral data of more than 1 000 individual seeds with early growth of seedlings in a post-pyrogenic experimental site of the forest landscape of the Voronezh region by example (Pinus sylvestris L. var. Negorelskaya).