Résumé
Many models already exist through literature dealing with root system representation, among which pure structure models such as Root Typ (Pagès 2004), SimRoot (Lynch 1997), AmapSim (Jourdan 1997); diffusion PDE models (Bastian 2008; Bonneu 2009) and structure/function that are rather scarce and recent (Dupuy 2010)may be aroused. Nevertheless in these studies, root architecture modeling was not carried out at organ level including environmental influence and not designed for integration into a whole plant characterization. We propose here a multidisciplinary study on root system from field observations, architectural analysis, formal and mathematical modeling and finally software simulation. Each speciality is individually investigated through an integrative and coherent approach that leads to a generic model (DigR) and its software simulator that is designed for further integration into a global structure/function plant model. DigR model is based on three main key points: (i) independent root type identification (ii) architectural analysis and modeling of root system at plant level; (iii) root architecture setup indexed on root length. Architecture analysis (Barthelemy 2007) applied to root system (Atger 1994) leads to root type organisation for each species. Roots belonging to a particular type share dynamical and morphological characteristics. Root architectural setup consists in topological features as apical growth, lateral branching, senescence and death, and geometrical features as secondary growth and axes spatial positioning. These features are modeled in DigR through 23 parameters whose values can evolve as a function of length position along the root axes for each root type. Topology rules apical growth speed, delayed growth, death and self pruning probabilities. Branching is characterized by spacing and mixture of lateral root types. Geometry rules root diameter increase, branching and growth directions (including local deviations and global reorientation). DigR simulator provides a user interface to input parameter values specific to each species. It is integrated into the Xplo environment (Taugourdeau 2010). Its internal multi-scale memory representation is ready for dynamical 3D visualization, statistical analysis and saving to standard formats (MTG(Godin 2007), Obj,). DigR is simulated in a quasiparallel computing algorithm and may be used either as a standalone application or integrated in other simulation platforms. This will allow further implementation of functional - structural interactions during growth simulation. The software is distributed under free LGPL license and is dedicated both to biologists and modelers. Shown applications (fig. 1) mimic the diversity of root systems and emphasize the genericity of the model according to different sets of parameter values. Examples (fig. 2) prove that additional knowledge may be plugged to DigR to simulate root plasticity facing environmental constraints. Further work will be carried out to apply DigR to various species and to connect DigR to biophysical soil models (Gérard 2008; Zhang et al. 2002); to aerial part models (Barczi 2008); to ecophysiological models (Mathieu 2009, Bornhoffen 2007); and finally to mix this pure descriptive model to a PDE model that handles fine root diffuse modelling (Bonneu 2009). (Texte intégral)