N leaching across European forests: Derivation and validation of empirical relationships using data from intensive monitoring plots

Empirical relationships to predict the leaching flux of dissolved inorganic nitrogen in forested ecosystems as a function of N-deposition and stand and site characteristics have been derived using an updated version of the UN-ECE/EC intensive monitoring database, including data for the period up to the year 2000. These relationships were validated on an independent database with literature data. For the model development, reliable N budgets were available for 57 forest sites in Europe for a median period of 5 years. Multiple regression analysis showed that the measured N leaching fluxes could be well explained by a number of different empirical relationships. The simplest model included only the N throughfall flux and explained 30% of the variance in observed leaching fluxes. The most complex model showed a positive relationship between N leaching fluxes and N throughfall flux, temperature and the pH of the mineral topsoil and a negative relationship with the C/N ratio of the organic layer. This relationship explained 42% of the observed variance in leaching fluxes. The empirical equations explained 47-64% of the observed variation in leaching flux in an independent validation database. The best relationships were the one that included only N throughfall as a predictor, and a relationship that also included the C/N ratio of the organic layer as a factor to express differences in reaction above or below a critical C/N ratio. The median error was 211 mol/(ha year) for the relationship with N throughfall and 240 mol/(ha year) for the relationship that also included the C/N ratio. The median relative errors were 70 and 50%, respectively, for the two relationships. These large errors are mainly due to a general overestimation of N leaching fluxes at sites with a nitrogen-leaching fraction below 0.3. These are primarily Nordic sites with low total N-deposition levels and mid-latitude sites with relatively high C/N ratios. (c) 2006 Elsevier B.V. All rights reserved.