A site-specific prediction model for nitrogen leaching in conventional and organic farming

Food production has a profound eutrophication impact on waterbodies via nutrient leaching. To provide reliable life cycle assessments of the eutrophication potential of agricultural products, accurate nitrogen leaching models are needed. Although many dynamic nitrogen leaching models are in use, their suitability for farm-level assessments remains limited when their requirements for site specific data or numerous parameters are not met. In Finland, less data intensive leaching models for life cycle assessments have been developed using data from conventional farming, however, the suitability of these models for organic farming remains unknown. In this work, we developed new nitrogen leaching models that are applicable to both conventional and organic pro-duction. While this paper does not aim to argue in favor of organic or conventional farming it provides tools that can be used to inform decisions about management practices from the environmental perspective. We utilized up to 16 years of field measurements from two leaching fields in Finland. We developed prediction equations for nitrogen leaching for two soil types: sand soil and clay soil. According to our statistical analysis based on the data, the relevant factors for explaining nitrogen leaching included soil type, rainfall, whether the farming is done organically, and the availability of nitrogen for leaching. Computed nitrogen balance as such was found to be a poor proxy for nitrogen available for leaching, while nitrate nitrogen concentration measurement of the soil carried out in the fall was found to be a valuable predictor. Organic farming, with a crop rotation resembling that of conventional farming, resulted on average in 20% less nitrogen leached per hectare as compared to conventional farming with 95% C.I. [-34%,-3%]. The developed models are suitable for integration into a life cycle assessment framework, and especially the models utilizing nitrate nitrogen were shown to be applicable to a wide range of different crop types, making the model well-suited for plots with diverse crop rotations.