Spatializing exposure related to the agricultural use of pesticides: the contribution of the Matphyto crop-exposure matrices

Background. In the literature, the study of the relationship between pesticide exposure and various diseases is hampered by the difficulty of assessing exposure, particularly in a retrospective manner. Crop-exposure matrices (CEM) have been developed to address this difficulty. Objective. Our method consists in combining CEM with land use data from agricultural censuses to determine geographic indices of environmental and occupational exposure. We illustrate this approach for two active substances. Methods. The CEM of four crops are used (straw cereals, corn, potatoes, vines). They each describe three groups of pesticides, their chemical families and their active substances using three indices per region and period since 1960. These data are combined with agricultural censuses (1979, 1988, 2000, 2010) to calculate indices of environmental and occupational exposure for each canton of the French metropolitan territory. Two active substances (chlortoluron, pyrimicarb) illustrate this methodology, and the results are compared with pesticide purchase data (BNV-D). Results. The two active substances show a temporal and spatial evolution. The proportion of areas/farms treated with pyrimicarb increased between 1979 and 2010, in contast to chlortoluron. The North has the highest values for pyrimicarb, this gradient is supported by a strong spatial autocorrelation. Areas are more heterogeneous for chlortoluron. Comparisons with purchase data show high correlations (>= 0.80). Significance. The combination of CEMs with agricultural censuses data allows to describe pesticide exposure at a fine geographical level (the canton). This approach allows to identify spatial evolutions, confirmed by comparison with pesticide purchase data, and temporal changes.