Predictors for assessing the flowering duration and dynamics of the complex umbellate of fennel in seed production

International audience

英语. Flowering is a critical stage of fennel seed production, which strongly depends on climate factors. In crops such as fennel, the complexity of the umbellate structure and the lack of knowledge about its flowering dynamics make the prediction of the duration of this phenological stage uneasy. In the context of climate change, a reliable predictive tool for the duration of fennel flowering is needed. Our aim was to increase knowledge on fennel flowering dynamics and to propose a model to be used by seed producers. In 2018 and 2019, we studied the development of secondary and tertiary umbels during flowering stage which was determined from the Bundesanstalt Bundessortenamt und Chemische Industrie scale, in plants of four genotypes grown in tunnels. Multiple linear regression (MLR) models were used to select the more accurate equations which consisted of one or more variables as predictors of flowering duration and its dynamics across secondary and tertiary umbels. In all model subsets, the criterion-based procedure was used, followed by a criterion-based hybrid procedure to take advantage of the Mallows' Cp, adjusted R-2 and Bayesian information criterion, for equations and predictor selection. Two variables related to temperature, namely the temperature summation expressed in degree days and the number of days with temperature lower than 15 degrees C, were identified as relevant parameters for predicting flowering duration expressed in days, in secondary as well as tertiary umbels. In addition, the equations and accurate predictors identified for modelling differences in time between flowering starts and ends of secondary and tertiary umbels also highlighted the importance of the number of days when the temperature amplitude is higher than 20 degrees C, the number of days when the mean temperature is higher than 27 degrees C, and the number of days when the relative humidity is higher than 80%.