Format: ORAL

Giuseppe Frenguelli

Department of Agriculture Science, University of Perugia, Perugia, Italy

The release of pollen from a plant-source is characterized by a time distribution which is the consequence of the various flowering stages. The beginning of the release of pollen and the quantity of pollen that each plant produces are in relation to a series of biotic factors such as the genotype, the age, the size of the plants, and environmental factors such as the climate, other than edafic and phytosanitary factors. In the aerobiological studies phenological models are used to obtain information useful for forecasting the beginning of pollination of many species, expecially of agricoltural and allergological interst, and to know what will be the severity of the pollination or the quantity of pollen which will be released into the atmosphere day by day. Aerobiological historical data bases on the pollen content in the air, evidence how the climate change is affecting distinct species in different geographical areas.The aerobiology-floral phneology binomial can contribute to the improvement of bio-geographical and ecological information of vegetation, the forest health and vitality being important for the protection of the plant biodiversity. When a predictive model on beginning of pollination is being developed, it is necessary to take into consideration the different responses to environmental factors of the species considered. Temperature-based models are commonly used to predict floral bud-burst as universal models disregarding that they are site- and species-specific. Models working well in some regions might not respond so well in other areas due to genetic characteristics that enable plants to show different tolerance to particular environmental conditions. Modeling to predict the performance of the pollen season with the intention of giving a meaningful result as early as possible has led to the development of many analytical mathematical models, increasingly complex, like time series models, multivariate regression models, Artificial Neural Networks and more.