The development of airborne pollen and fungal spores models aims among others, to evaluate and predict their concentrations and/or seasonal dynamics; to establish, develop and/or validate atmospheric transport models; and/or to correlate concentrations with impacts or changes detected in environmental processes. The general objective of this thesis was to develop different statistical models to study the spatial-temporal variation of the main airborne pollen and fungal spores in Catalonia. In the first chapter we analyzed the predictive and dispersion models present in the literature, concluding that the models based on observations (OBM) have been frequently used to predict future concentrations, the models based on phenology (PHM) to characterize the flowering period, and a range of other models (Other models) to establish spatial estimates in unmonitored areas. PHM are more diverse, easier to apply and they use a smaller number of independent variables than OBM. The dispersion models (other models) are classified according to the direction of the modeling (Forward or Backward) and present a smaller number of applications due to their high technical-scientific requirements, being their greatest limitation the establishment of the flow and the emission source. In the second chapter, we validated a gamma distribution model to characterize annual series of pollen and spores data and to establish the few of them showing representative patterns. From the results we concluded that the α parameter of the model changed reasonably from year to year, depending on the meteorological conditions, with good inter-annual and spatial stability and allowing to establish a generic classification of the airborne particles studied in five categories for pollen and five for fungal spores. This classification showed a strong relation of α parameter with the ecological distribution of the plant (potential and/or ornamental) in the case of the pollen taxa and with the land use and/or the bioclimate of the zone in the case of the fungal spores. The classification established will allow reducing the number of taxa with which to develop new prediction models in the third chapter we selected 12 taxa (six pollen and six fungal) and developed predictive models based on concentration thresholds instead of on concentrations. We evaluated the predictive potential of of logistic regression models and regression tree models at four concentration thresholds (low, medium, high and very high). The two models showed similar results regarding the relationship and/or influence of the meteorological parameters in the different thresholds, presenting highly satisfactory values on sensitivity and specificity during the validation. However, we observed that the logistic regression models gave a greater precision (sensitivity) in establishing the exceedance of a concentration threshold and therefore that they are the most suitable for future estimates. Finally, in chapter 4 it was undertaken a trend analysis with the annual index of 20 fungal spore taxa of the period 1995-2013. The analysis showed t12 taxa with significant increasing trends and two with decreasing trends. These results were mainly attributed the diversity of geographic characteristics and land use of the studied localities. Also, the increasing trends of the temperature and the instability of the precipitation in Catalonia in the last 50 years could be stimulating the sporulation on mountain areas and limiting it on the coast, thus affecting the presence of spores.
|Date of Award||26 Sept 2017|
|Supervisor||Jordina Belmonte Soler (Director) & Maria Concepcion De Linares Fernandez (Director)|