The role of crop simulation modeling in managing fertilizer use in maize production systems in Northern Ghana

The agricultural landscape in Sub-Sahara Africa is dominated by smallholders and characterised by high heterogeneity due to high variation in soil, crop varieties and land management practices. As a consequence, crop yields vary considerably in space and time. This study aimed at developing a production domain for different maize varieties in northern Ghana in response to the conditions, using a simulation modelling approach. Data from a two-year multi-location nitrogen (N) response trial in northern Ghana were used to calibrate and evaluate the performance of the Decision Support System for Agro-technological Transfer (DSSAT) crop model, which was then used to simulate maize yields across the region. The model performed adequately in simulating maize phenology, with the relative root mean square error (RRMSE) below 15%. The response of grain and biomass yield of different varieties of maize to nitrogen fertilizer application was also adequately simulated, with model efficiency coefficients ranging between .68 and .88 across the varieties and planting dates. In general, grain yields were lower in the Sudan Savannah than in the Guinea Savanna zones of northern Ghana. Also, mid planting produced higher yields in the Sudan Savanna zone whereas for the Guinean zone, it was the early and late planting that produced higher maize yields. The simulations showed that the application of 60 kg N ha^-1 was optimal for the early and the extra-early varieties, but higher N application rate of 90 kg N ha^-1 was optimal for the intermediate variety. It is concluded that the DSSAT crop model enabled an effective generation of crop yield data suitable for decision support, agricultural planning and policy formulation in northern Ghana.