Calibration and Evaluation of Ceres-Sorghum Model for Use in Sorghum (Sorghum bicolor L. Moench) Production in Sudan Savanna of Nigeria

Abstract

A study was conducted at Bayero University Teaching and Research Farm (BUK), Kano, and the International Institute of Tropical Agriculture (IITA) Research Farm, Minjibir. The experiment employed a Randomized Complete Block Design (RCBD) with four replications, using eight sorghum varieties. Data from both locations were used to calibrate and evaluate the CERES-Sorghum model within the DSSAT framework. Calibrated genetic coefficients for all eight cultivars were incorporated into the DSSAT cultivar file. Model performance was assessed using model efficiency (EF), index of agreement (D-index), and root mean square error (RMSE). Calibration results showed strong agreement between observed and simulated values for days to flowering, physiological maturity, leaf number per plant, grain yield, and biomass (D-index: 0.97–0.95; EF: 0.87–0.80; RMSE: 125.68–1.37). Similar accuracy was observed during evaluation. CSR-01, SAMSORG-44, and 2192-2N had the highest panicle partitioning coefficients (G2), while SAMSORG-14, SAMSORG-17, and 12KN ICSV-188 recorded the lowest. CSR-02 and ICSV-400 had intermediate values. For the juvenile phase thermal time (P1), SAMSORG-14, 12KN ICSV-188, and CSR-02 had the highest values, while SAMSORG-17 and SAMSORG-44 had the lowest. ICSV-400 and CSR-01 were intermediate. Photoperiod sensitivity coefficients (P2R) ranged from 1 to 300 GDD, with SAMSORG-17 highest and ICSV-400 lowest. Overall, the model demonstrated satisfactory performance and proved reliable in simulating sorghum growth and yield across the Sudan Savanna agro-ecological zone of Nigeria.