Spatial variation of airborne pollen over south-east France: characterization and implications for monitoring networks management

The French airborne pollen monitoring network (RNSA) is currently regrouping 70 Hirst-type pollen traps covering the whole French territory. The aim of this paper is to introduce a simple statistical methodology that can be used to characterize pollen spatial variation. This pilot study is restricted to a limited portion of the RNSA network (18 monitoring stations), eight taxa of allergenic relevance, and a 3-year period (2003-2005). The first step of the approach consisted in quantifying the trap-to-trap pollen similarities on the basis of an original index, called mean Pollinic Distance (mPD), that relies on the comparison of pollen concentration time series. Regression analyses were next conducted with different spatial variables. Distance, latitude and altitude differences were identified as significant predictors of pollen variations, as measured by mPD. In order to further characterize pollen spatial properties, cluster analysis was performed with mPD as the distance estimate. Interestingly, the clusters of sites identified on the basis of the similarity of their pollen profiles, correspond to distinct geographic areas that might be interpreted as homogeneous air masses. The results have major implications for monitoring networks management since they provide an objective basis (1) for choosing the relevant scale to elaborate and supply pollen-related information, and (2) for optimizing networks configuration.