Calibração dos modelos CSM CERES-Maize e CSM IXIM para quatro híbridos de milho cultivados no Brasil

Abstract

Maize (Zea mays L.) is one of the world's most important agricultural crops, playing a fundamental role in food security, animal production, and the bioeconomy. Given projections of yield reductions resulting from climate change, crop simulation models have been widely used to evaluate crop growth and performance under different environmental conditions. Among the models available in the DSSAT platform, CSM-CERES-Maize and CSM-IXIM stand out due to differences in the representation of physiological processes related to vegetative and reproductive plant growth. The objective of this study was to calibrate and evaluate the CSM-CERES-Maize and CSM-IXIM models for four commercial maize hybrids cultivated in Brazil and to compare model performance in simulating crop growth and yield under tropical conditions. For this purpose, data from a field experiment conducted in Lavras, Minas Gerais, Brazil, during the 2018/2019 growing season were used, involving four hybrids with different maturity classes. Phenological, vegetative growth, biomass, leaf area index, and grain yield data were used to calibrate the genotypic coefficients of both models. Simulations were performed using DSSAT version 4.8.5, and model performance was evaluated through the coefficient of determination (R²), root mean square error (RMSE), and Willmott’s index of agreement (d). Calibration resulted in specific sets of genotypic coefficients for each hybrid and model, providing substantial improvements compared with the default parameter values. Observed grain yield ranged from 7,816 to 14,042 kg ha⁻¹ among the evaluated hybrids and was adequately reproduced by both models. CSM-CERES-Maize and CSM-IXIM showed a high capacity to simulate phenological and yield-related variables, with high coefficients of determination and indices of agreement. The greatest discrepancies between the models were observed in the estimates of total aboveground biomass and stem biomass, reflecting structural differences in the representation of carbon assimilation and partitioning processes. The results demonstrate that both models are capable of satisfactorily representing the growth, development, and productivity of tropical maize hybrids under Brazilian conditions, constituting robust tools for management studies, genotype evaluation, and agricultural scenario analyses in tropical environments.