Simplifying the prediction of phenology with the DSSAT-CROPGRO-soybean model based on relative maturity group and determinacy

The use of crop models can be limited by the need to calibrate cultivar coefficients across a sufficiently wide range of environments. The DSSAT-CROPGRO-Soybean crop simulation model considers different temperature and photoperiod sensitivities during different crop developmental stages and/or for different cultivars. The use of generic phenology coefficients specific for a range of maturity groups (MGs) could allow accurate predictions of main developmental stages in soybean without requiring calibration. Phenology data collected in 2012 and 2013 from an irrigated regional planting-date experiment with maturity group (MG) 3 to 6 cultivars and latitudes from 30.6 to 38.9°N, were used to calibrate cultivar coefficients across all the environments. A set of generic coefficients were generated based on relative maturity group (rMG) and plant growth habit. Predictions of main developmental stages in the subsequent growing season (2014) using generic coefficients were similar to predictions based on calibrated coefficients, with a RMSE across all cultivars < 8 days. Several calibrations of cultivar coefficients were conducted testing different hypotheses of sensitivity to temperature and photoperiod in the model. Surprisingly, after the calibration, the model predicted with similar RMSEs the day of R1, first R5 seed, and R7 under the different hypothesis of model sensitivity to photoperiod and temperature. Therefore, the use of an optimization tool for calibration across several site x year x planting dates was efficient to obtain cultivar coefficients that minimized error in prediction, but did not provide meaningful insight regarding the mechanistic function of temperature and photoperiod coefficients describing phenology prediction.