OSTRICH-CROPGRO multi-objective optimization methodology for calibration of the growing dynamics of a second-generation transgenic soybean tolerant to high temperatures and dry growing conditions

Abstract

CONTEXT: Current estimates show that the impact of climate change on agriculture will result in yield losses in major crops of 8–43%, mainly due to a combination of drought and heat. Second-generation transgenic crops are expected to mitigate these constraints. Soybean HaHB4 is a second-generation transgenic crop tolerant to high temperatures and dry growing conditions created in Argentina carrying the sunflower HaHB4 gene. Soybean HaHB4 has been approved in Argentina, the United States, Brazil, Paraguay, Canada and China. OBJECTIVE: This study presents a robust methodology to calibrate the CROPGRO-soybean model for the growth and development of soybean HaHB4. The approach consists of a holistic treatment of calibration parameters, objective functions, model responses and measured data. Based on the differences between transgenic and controls obtained in field trials, the proposed methodology includes species parameters related to those physiological traits that present the most significant differences, i.e. heat and drought tolerance with no yield penalties, increased light interception and photosynthetic rate, increased crop biomass and crop yield and improved water use efficiency. METHODS: We define multiple objective functions as a way of handling multiple simulated responses in the calibration procedure. For this, we connect CROPGRO with the OSTRICH software toolkit. Adjustments for initial water content in the soil profile, soil root growth factor and root depth progression were made using soft data procedures. The basic procedure for automatic calibration was modified by considering a semi-automated calibration process. The calibration sequence considers phenological development, water balance, biomass and yield parameters. RESULTS AND CONCLUSIONS: We observed good accuracy in the calibrated simulation of soybean HaHB4 development at all phenological stages, with RMSE = 1.62 days. The soil water balance reached RMSE = 27.4 mm. An acceptable biomass simulation at maturity was reached, with RMSE = 34% of average observed values. The grain yield was well predicted, with RMSE = 636 kg/ha. To verify the robustness of the calibrated model we evaluated it for grain yield prediction in fourteen field experiments for different growing seasons, water conditions and locations across the Argentine Pampas. The model accurately simulated grain yield with RMSE = 408 kg/ha, d-index = 0.93 and R2=0.77. SIGNIFICANCE: This study provides a calibration procedure for climate-resilient cultivars that are still missing for long-term studies on climate change impacts. The importance of modelling a climate-resilient crop in the framework of the soil–plant-atmosphere system is a step towards ensuring food security.