Predicting dissolved inorganic nitrogen leaching in European forests using two independent databases

Regional-scale databases can be particularly useful for identifying relationships between dissolved inorganic nitrogen (N) leaching in forests and environmental drivers, which in turn allow an assessment of the risk of ecosystem damage, such as forest acidification and eutrophication of downstream water bodies. However, detecting the 'signal' of a significant correlate to N leaching against a background of wide variability in other factors requires a large number of sites, and the validation of models developed requires a similarly large number of independent sites. Here we use two large and fully independent databases of forest ecosystems across Europe to develop and validate indicators of N saturation and leaching. One database was used for model development and the other for validating these models. Among 35 variables considered, the most significant indicators of N leaching in the model development database were: the flux of dissolved inorganic N in deposition, mean annual temperature, mean altitude, the site drainage (plot vs catchment), needle- and litter-N concentration, organic horizon C:N ratio, and subsoil pH. Altitude was not a consistent predictor (it was significant in the development database but not in the validation database), and needle and litter N concentration, plot vs catchment, and subsoil pH all showed high intercorrelation with N deposition and so were not significant in models already including N deposition. The most consistent and useful indicators of N leaching were throughfall N deposition, organic horizon C:N ratio and mean annual temperature. Sites receiving low levels of N deposition (< 8 kg N ha^{- 1} y^{- 1}) showed very low output fluxes of N and were simulated separately from more polluted forests. In general, the models successfully predicted N leaching (mean of ± 5 kg N ha^{- 1} y^{- 1} between observed and predicted) from forests at early to intermediate stages of nitrogen saturation but not from nitrogen-saturated sites. Thus, simple relationships developed from combining (1) external drivers (deposition, temperature) and (2) site conditions (nitrogen status of soils) can successfully estimate nitrogen leaching from forests that have not yet been highly damaged by N deposition.