Search

BACKGROUND: A likely increasing demand for varieties mixtures, landraces and genetic diversity in cropping systems will underpin calls for models able to generalize phenological development at the species level, at the same time as providing the expected range of phenological variability. In the present article, we aimed to obtain a generalized phenological model of durum wheat (Triticum durum, Desf.). RESULTS: Using a large phenological dataset embracing field data collected under different sowing dates, varieties and locations over the Italian peninsula, we searched for the phenophases enabling the best linear approximations between developmental rates and air temperature, aiming to minimize the residual variability from drivers other than temperature, as genetic and environmental diversity. The developmental rates of the resulting phases were then examined with respect to the mean daylength to determine possible additional relations with photoperiod. If a correlation with daylength was also present, the developmental rate is calibrated by multiple linear regression, or otherwise by simple linear regression of temperature. The resulting calibration, tested on an independent data subset, confirms that the model is able to generalize wheat development over the Italian peninsula with high accuracy (mean absolute error =3–8 days; r2 = 0.75–0.98), regardless of the wheat variety. CONCLUSION: The generalized phenological model is potentially suitable for many agro‐ecological and large‐scale applications. It is hoped that the model will aid in situations where phenological observations to parameterize a model are still lacking, as is probably the case for landraces and underutilized crop varieties. © 2019 Society of Chemical Industry [ABSTRACT FROM AUTHOR]

The scientific community offers numerous crop models with different levels of sophistication. In such a wide range of crop models, users should have the possibility to choose the most suitable, in terms of detail, scale and representativeness, to their objectives. However, even when an appropriate choice is made, model limitations should be clarified such that modelling studies are put in the proper perspective and robust applications are achieved. This work is an overview of available models to simulate crop growth and yield. A summary matrix with more than 70 crop models is provided, storing the main model characteristics that can help users to choose the proper tool according to their purposes. Overall, we found that two main aspects of models, despite their importance, are not always clear from the published references, i.e. the versatility of the models, in terms of reliable transferability to different conditions, and the degree of complexity. Hence, the developers of models should be encouraged to pay more attention to clarifying the model limitations and limits of applicability, and users should make an effort in proper model selection, to save time often devoted to iteration of tuning steps to force an inappropriate model to be adapted to their own purpose. © 2015 Society of Chemical Industry [ABSTRACT FROM AUTHOR]