Use of the APSIM model in long term simulation to support decision making regarding nitrogen management for pearl millet in the Sahel

Soil fertility and climate risks are hampering crop production in the Sahelian region. Because experiments with only a few fertility management options on a limited number of sites and years cannot fully capture the complex and highly non-linear soil-climate-crop interactions, crop growth simulation models may suitably complement experimental research to support decision making regarding soil fertility and water management. By means of a long term (23 years) scenario analysis using the Agricultural Production Systems Simulator (APSIM) model, this study investigates millet response to N in view of establishing N recommendations better adapted to subsistence small-holder millet farming in the Sahel. Prior to this, the APSIM model was tested on a rainfed randomized complete block experiment carried out during the 1994 and 1995 cropping seasons, having contrasting rainfall conditions. The experiment combined, at three levels each, the application of cattle manure (300, 900 and 2700kgha⁻¹), millet residue (300, 900 and 2700kgha⁻¹) and mineral fertilizer (unfertilized control, 15kgNha⁻¹ +4.4kgPha⁻¹ and 45kgNha⁻¹ +13.1kgPha⁻¹) at ICRISAT Sahelian Center, Niger. The model suitably predicted plant available water PAW and the simulated water and nitrogen stress were in agreement with measurement (water) and expectation (N) regarding the fertilizer and rainfall conditions of the experiment. APSIM simulations were in satisfactory agreement with the observed crop growth except for the highest crop residue application rates (>900kgha⁻¹). For biomass and grain yield, the model performance was relatively good in 1994 but biomass yields were slightly overpredicted in 1995. The model was able to adequately reproduce the average trend of millet grain yield response to N inputs from manure and fertilizer, and to predict the overall observed higher grain yield in 1995 compared to 1994, despite the better rainfall in 1994. The 23-year, long term scenario analysis combining different application rates of cattle manure, millet residue and mineral fertilizer, showed that moderate N application (15kgNha⁻¹) improves both the long term average and the minimum yearly guaranteed yield without increasing inter-annual variability compared to no N input. Although it does imply a lower average yield than at 30kgNha⁻¹, the application of 15kgNha⁻¹ appears more appropriate for small-holder, subsistence farmers than the usual 30kgNha⁻¹ recommendation as it guarantees higher minimum yield in worst years, thereby reducing their vulnerability.