Model-based assessment of maize cropping under conventional and conservation agriculture in highland Mexico

The Cropsyst crop-soil-simulation model was used to assess the performance of conservation tillage in comparison to conventional tillage during 13 years of continuous maize cropping in highland Mexico. We tested if the calibration and validation requirements for CropSyst could be met using data sets, which were routinely collected by agronomists. Highest maize yield was observed under zero-tillage with retained residues. Simulation results indicated that this was due to more favorable moisture conditions, attenuating water stress in adverse years. Soil mineral N concentration measured in 1998 indicated the likelihood for N-stress under zero-tillage with residues retained. CropSyst additionally predicted N-stress as a yield limiting factor in other years, despite a seemingly optimal N supply by mineral fertilizer. CropSyst could predict yield under conventional tillage with residues retained and under zero-tillage with residues removed reasonably well, indicated by a modified Nash-Sutcliffe coefficient of efficiency (E-1) of 0.32 and 0.48. Yield predictions for conventional tillage with residues removed were poor (E-1 = -0.05) and those for zero-tillage with residues retained insufficient (E-1 = -0.20). Nonetheless, simulation results highlighted systematic differences between treatments with regard to water and N-dynamics. CropSyst lacks routines to account for soil crusting, the temporal impact of tillage on soil hydraulic conditions and the effect of surface residues physically restraining surface water runoff. These model shortcomings and the lack of detailed and continuous field measurement constrained detailed analyses and discussion of quantities produced by the model. (c) 2006 Elsevier B.V. All rights reserved.