RESUMEN
Nitrogen fertilisation is an essential part of modern agriculture, providing food for a growing human population, but also causing environmental impacts when reactive nitrogen (N) is released to the environment. The amount and impact of these emissions are difficult to quantify in life cycle assessment (LCA), due to their site-dependent nature. This study compared seven models for direct soil nitrous oxide (N2O) emissions, seven models for N leaching and five characterisation models for marine eutrophication impact assessment, selected to represent medium-effort options for accounting for spatial variation in emissions and impact assessment. In a case study, the models were applied to wheat cultivation at two Swedish sites to estimate climate and marine eutrophication impact. Direct N2O emissions estimated by the models varied by up to five-fold at one of the sites and contributed 21-56% of the total climate impact. Site-dependent models gave both lower and higher N2O emissions estimates than the site-generic Tier 1 model from the Intergovernmental Panel on Climate Change (IPCC). Estimated N leaching also varied by up to fivefold at one of the sites and contributed 47-93% of the total eutrophication potential, depending on model choice. All site-dependent models estimated lower N leaching than the site-generic IPCC Tier 1 model. Marine eutrophication impact estimates varied by almost an order of magnitude depending on characterisation model choice. The large variation between models found in this study highlights the importance of model choice for N emissions and marine eutrophication impact assessment in LCA of crop cultivation. Due to the divergence of model outcomes and different limitations of some of the models, no general recommendations on choosing soil N2O emissions model, N leaching model or characterisation model for marine eutrophication could be given.