Application of Ceres-Millet model of DSSAT for simulating millet varieties under different sowing windows in Niger

ABSTRACT Choosing an appropriate variety and planting date could help farmers increase the recurrent low pearl millet (Pennisetum glaucum (L.) R. Br.) yields recorded in Niger. Field experiments were conducted at the National Agricultural Research Institute of Niger (N’Dounga) during the 2016 and 2017 rainy seasons to evaluate the performance of the CERES-Millet model in simulating the growth and yield of pearl millet varieties. Treatments consisted of two sowing windows (late June and mid-July) and four varieties (HKP, ZATIB, CIVT, and H80-10 GR) arranged in a split-plot design with three replications. Sowing dates were allocated to the main plots, whereas varieties were assigned to the subplots. The data collected on days to 50% flowering, days to physiological maturity, grain yield, and above-ground biomass during 2016 and 2017 were used to calibrate and evaluate the CERES-Millet model for simulating sowing windows. The model was used to simulate the long-term effects of different sowing windows using long-term historical weather data from 1983 to 2017 in Kollo. The results for model calibration showed that simulated growth and yield of millet were in good agreement with their corresponding observed values. The d-index was 0.99 for days to anthesis, 0.97 for days to physiological maturity, 0.93 for grain yield, and 0.79 for above-ground biomass. The results for both calibration and evaluation showed that normalized root mean square errors were less than 10%. The values of the d-index were also within the acceptable range for all the parameters. Therefore, the CERES-Millet model was robust enough to successfully simulate millet growth and yield in Niger. Seasonal analysis revealed that sowing should be done from early June to mid-June for ZATIB and H80-10 GR varieties. However, it should be done from early June to late June for CIVT and from mid-June to early July for HKP.