Pattems and models of flowering of some Gampanulaceae Juss. species

The present work is devoted to the phenology of individual flowering and the construction of structure-dynamic models of this process on its basis. The results of the study of the flowering phenology of Campanula bononiensis, C. sarmatica and Platycodon grandiflorus are presented. The data obtained characterize both the phenological (time and duration of flowering, lifespan of individual flowers) and structural features (degree of branching of the inflorescence, length of floral axes, number of flowers, order of their blooming) that describe the flowering of a monocarpic shoot. Inflorescences of the species are elongated and multiflorous, of the compound type inherent for Campanulaceae, and characterized by a high variability of all structural features. Observation data were processed by standard statistical methods and used to construct stochastic computer models of flowering shoots, while omissions in data were restored by using the maximum likelihood method. Flowering patterns of the species, due to differences in phenological and structural features, have been revealed. It has been shown that flowering curves depend on the synchrony in the flowers blooming on the main (first-order) axis and lateral (second-order) axes. C. bononiensis has one asymmetrical peak with a broadening on the left, achieved with the simultaneous blooming of flowers in the upper and lower parts of the main axis and on lateral axes in the middle part of the inflorescence, where the first-order flowers have already finished blooming (they provided the broadening). Flowering curves for C. sarmatica and P grandiflorus are bimodal, with the first peak being due to the flowers blooming on the main axis and the second one on lateral axes. The constructed models reproduce the patterns of individual flowering well, with natural variability, and can be used to simulate the flowering of a group of individuals (population), for example, in landscape design. In combination with visualization tools, they can be used for augmenting plant phenotyping datasets with rendered images of synthetic plants for the purpose of training neural networks in this field