02427naa a2200241 a 450000100080000000500110000800800410001902400510006010000260011124501090013726000090024652017010025565000160195665300140197265300270198665300170201365300140203065300200204470000180206470000180208270000290210077300560212921780742025-08-19       2025    bl uuuu     u00u1 u    #d7 ahttps://doi.org/10.1016/j.eja.2025.1278052DOI1 aOLIVEIRA, M. P. G. de  aA role for regression trees in the calibration of a new process-based crop model.h[electronic resource]  c2025  aWhile some researchers aim to standardize crop model calibration, there is still room for enhancing current practices. In this study, we explore the use of regression trees to enhance model understanding within the commonly used trial-and-error method. As an approximation of this approach, we sampled multiple values and combined them into different parameter sets. We used regression trees to model simulation errors from these sets and investigated the relationships they made explicit. We also used the trees to aid in the choice of parameter values. From the multiple sets, we selected the best among those that followed the rule leading to the lowest average simulation error. We compared these results to those of the set that directly minimizes simulation error and to a recently developed calibration protocol, used as a performance reference. Different subsets of the dataset were used for calibration and evaluation to assess robustness. The trees consistently identified relationships between parameter values and low errors across all subsets and approaches, including how parameter choices compensate for one another. As a calibration approach, all methods assessed performed similarly well for aboveground biomass and evapotranspiration, as well as for yield in one of the three calibrated cultivars. The most marked differences appeared in the variability of parameter values obtained across subsets, for which none of the approaches matched the protocol. While regression trees contributed to improved model understanding, particularly in support of manual calibration, a combined approach using the protocol to determine values and trees to visualize outcomes appears promising.  aCalibration  aAgS model  aÁrvores de regressão  aCalibração  aMetamodel  aSurrogate model1 aCUADRA, S. V.1 aBENDER, F. D.1 aMONTEIRO, J. E. B. de A.  tEuropean Journal of Agronomygv. 171, 127805, 2025.