摘要
Phase development of the olive tree is important for many purposes (e.g. adaptability, management, crop modelling). Many studies on the prediction of flowering used only data from one location and/or use a simple thermal time approach, what impairs the model ability to be used under different conditions. In this study, three models were evaluated and compared. Model 1 is a chill–heating model, that is a generalisation and simplification of the Utah model, with a thermal time approach in the forcing phase; Model 2 has a below 7°C chill-hours model, followed by the same approach in the forcing phase; and Model 3 has no chilling description, and relies on a thermal time approach after 1 February. All models were calibrated using a data set of dates of flowering of five olive varieties that were grown in at least three locations, and the total chilling units accumulated (TU) until bud dormancy release, in the first two models, and the thermal time (TT) from this phase until flowering occurrence were determined. Validation followed on pooled data from 10 varieties grown in Cordoba, using the parameters from the calibration process (i.e. species level parameters) and the variety-specific TUs and TTs. The modelling efficiency was 0.92, 0.90 and 0.85, and the root mean square error of the predictions was 2.2, 2.5 and 2.8 days for Models 1–3, respectively. Although all three models depicted a good performance, Model 1 is more appropriate because it is physiologically meaningful. It should be preferred in all cases that the satisfaction of the chilling requirements of the species or variety is in doubt, and under different climate conditions. Three global warming scenarios A–C (daily maximum–minimum temperature increases of 1–3°C) were analysed, using the three models. All models and scenarios predict that there is a substantial advancement of the date of flowering. Only Models 1 and 2 show that the warmer scenarios indicate no normal flowering in some varieties/years. Models 1 and 2 further show the possibility that some compensation occurs in the warmer scenarios. Scenario A predicts that flowering is 10.0 and 9.3 days earlier than normal using Models 1 and 2, respectively. Scenario C shows that the advance of flowering for 1°C average temperature increase, in relation to Scenario B, is 7.4 and 5.2 days for Models 1 and 2, respectively. Model 1 and the algorithm that accompanies it might be useful to model the flowering occurrence of other woody species.