The effect of temperature on oak pollen season was studied at ten sites located in different bioclimatic areas of the Iberian Peninsula. In this area there are two biogeographical regions with different forest characteristics. A mixed forest constituted by deciduous Quercus species and other trees represents the Eurosiberian Region, located in the North and North-western Iberian Peninsula. Mediterranean Region, located in the North-eastern, Central and Southern Iberian Peninsula, is characterised mainly by a forest of evergreen Quercus species. This ecosystem has a number of remarkable features that make it attractive not only from an ecological standpoint but also both economically and culturally. One research priority in these areas is the study of phenology and the effect of climate which are allow us to predict future consequences of changes in the environment, including the highly-likely forecast change in climate.

Aerobiological data were taken using volumetric sampler Hirst type. These data belong to the Spanish Aerobiological Network (REA). The National Wheather Institute supplied meteorological data. An historical database from 1992 up to now has been used. A Growing Degree Method (GDDº), which extrapolates the daily temperature curve to a sine wave, was used to calculate the cumulate temperature before flowering. A wide range of possible threshold was tested. For future daily climate data the HadCM3A2a scenario of the General Circulation Model (GCM) developed by the Hadley Centre (UK) were used.

Total pollen emission was different between sites and study areas, although in general they were very high, particularly in sites located in the Mediterranean Region. Regardless of the number of years studied, a bi-annual pattern was observed; this was less marked in Southern areas due to the drought affecting them in the early years of study. A forecast model displayed that fortnight rainfall and maximum temperature, two months prior to pollen season, were the most important parameters to predict annual pollen emission.

The start of the pollen season varied considerable between sites and study areas. Regardless of the number of years studied, regression trend equation presented a negative slope coefficient at all the sites, which indicates that the pollination season has tended to start earlier in resent years. A cumulative-heat model was evaluated for predicting the start of future season. Results displayed different threshold for different climatic zones. Average GDDº values were validated using the years 2000 and 2001 as independent variables, yielding a different of only a few days between predicted and observed data of first pollen release dates. This model was forced with replicated future temperature simulation from the GCM to estimate the pollen season start date for the year 2099. An average advance of a month is predicted for Quercus pollen season at the end of the century in the Iberian Peninsula. Significant differences in the forecasting dates for the year 2099 have been detected between coastal and inland zones, particularly in the North-eastern area, due to the less temperature increase expected to the coastal areas.