Apparent nitrogen mineralization and recovery of nitrogen supply in field trials with vegetable crops

A major cause of uncertainty in predicting the demand of nitrogen (N) fertilizer by means of mathematical models is the treatment of apparent net N mineralization (ANM) and apparent recovery rate (REC) of mineral N supply (N fertilizer plus soil mineral N at planting). REC and ANM were analysed in 29 multi-level N fertilizer trials conducted with a range of vegetable crops over a period of 11 years. REC differed substantially between experiments (0.29 to 1.24). ANM ranged from -110 to 140 kg N ha(-1); it decreased with an increasing N supply, and increased with the time between planting and harvest. A simple regression model was used to predict ANM. The model consisted of both a N supply-dependent and supply-independent term, and used N supply and growing time as input parameters. Whereas the model significantly reduced the variance in measured ANM (r(2) = 0.45, n = 129), the residuals showed a considerable variation (standard deviation (SD) = 37 kg N ha(-1)). It is not feasible to predict N fertilizer demand without predicting net N mineralization and fertilizer recovery. Therefore, the regression model is recommended despite its inaccuracy, which is taken into account by adding a security margin to the fertilizer recommendation.