Field-specific simulations of net N mineralization based on digitally available soil and weather data : II. Pedotransfer functions for the pool sizes

Avoiding surplus N fertilization without reducing crop yields could be accomplished by accounting for current net N mineralization in N fertilizer recommendations. N simulation models would allow a quantitative consideration of important factors and could be based upon digitally mapped data. Soil-specific temperature and water functions that were derived in part I of the paper needed a differentiation between only three soil groups and the two allocating criteria were taken from digital soil maps. Here, the objectives were to experimentally determine pedotransfer functions (PTFs) for the pool sizes of two organic N pools (Nfast, Nslow) that could be calculated via digitally available data and need a minimum set of easily accessible management data. Interestingly, most important input data for the PTFs of both pool sizes were mean clay contents of the texture class (German soil classification system). However, the underlying mechanisms might be different, as Nslow could be positively influenced by clay-associated mineralizable SOM, whereas Nfast could be positively related to clay content due to higher yield potential and thus more residues on finertextured soils. For Nslow including the humus class improved the accuracy of the PTF (r2 = 0.60; P\0.050). For Nfast it was important to include a negative influence of the mean fall temperature of the preceding year (r2 = 0.42; P\0.010), probably due to its influence on residue degradation before winter. Surprisingly, easily accessible management data, e.g. previous crop, did not improve the predictions in this study. Field studies with plant cover will have to further prove the applicability of the derived PTFs.