Résumé : Modeling inter-individual variability in plant populations is a key issue to enhance the predictive capacity of plant growth models at field level. In sugar beet, this variability is well illustrated by the phyllochron (thermal time elapsing between two successive leaf appearances): even if the mean phyllochron remains stable within a given variety, there is a high heterogeneity between individuals. When considering the dynamics of leaf appearance as a function of thermal time in sugar beet, two linear phases can be observed, leading to the definition of a hierarchical segmented model with four random parameters varying from one individual to another: thermal time of initiation, first phyllochron, rupture thermal time and second phyllochron. The SAEM-MCMC algorithm is used to estimate the model parameters.