Agroforestry Systems 30: 175-197, 1995. 9 1995 Kluwer Academic Publishers. Printed in the Netherlands. A model simulating above- and below-ground tree architecture with agroforestry applications P. DE REFFYE 1, E HOULLIER 2, E BLAISE 1, D. BARTHELEMY l, J. DAUZAT 1 and D. AUCLAIR 1 Unitd de Mod~lisation des Plantes, CIRAD/GERDAT (Centre de Coopdration Internationale en Recherche Agronomique pour le D~veloppement), B.P. 5035, 34032 Montpellier cedex 1, France; 2 ENGREF, 14 rue Girardet, 54042 Nancy, France Key words: plant architecture, plant growth, plant modelling, stochastic processes Abstract. Modelling plant growth and architecture requires two consecutive and complemen- tary approaches. The first is a qualitative botanical analysis, in which the development sequence of a tree is studied by the identification of various levels of organisation and of homogeneous subunits. All of these - architectural unit, axis, growth unit - follow particular growth processes which can be described by using the second approach, the quantitative analysis. Modelling of the functioning of meristems based upon stochastic processes has been carried out since 1980, in combination with a large amount of experimental work on temperate and tropical plants. Calculations involved in tree simulations from field data are based upon the probabilistic Monte Carlo method for the topological part and on analytical geometry for the morphological part. Computer graphics methods are then used to visualise the computed plant. Several sectors in agroforestry are concerned with application of such plant architecture modelling: tree growth and yield, radiative transfers, timber quality and mechanics, simulation of competition, interaction between plant morphology and physiology. Introduction Agroforestry involves several components within an agricultural system, of which trees are one main element. These components interact with one another: trees influence growth (quantity and quality) of the crop or fodder species, they influence the microclimate and thus the energy balance and the production of animals [Sibbald et al., 1994], and conversely herbaceous plants and animals influence the growth of trees [Sibbald and Sinclair, 1990]. All the interactions are related to the spatial layout of the system. In designing agroforestry combinations, knowledge of the spatial devel- opment of the various components is extremely important in order to optimise the crop-tree and/or the tree-tree interactions. The occupation of space by a tree and its evolution with time, as a result of its architecture [Hall6 et al., 1978], influence physiological functions and interactions, such as light interception and transmission, water and nutrient uptake, and competition. Interactions between components are the main features of interest in agro- forestry research, and particularly in agroforestry modelling. The knowledge and understanding of tree architectural development is therefore of utmost importance.