Explicative and marginal models for a marking strategy of a unevenaged stand. We describe the forester behavior selecting trees for thinning in a unevenaged stand. We have studied the binary decisions ( cut or keep the tree) sequence, recorded in the order the decisions are made. The two models presented correspond to two different situations. First, in the conditional model, the decision concerning a tree is made according to its own properties and the properties of the trees next to it, as well as the decisions made before. The total cut and its variability are estimated by simulations on the scale of the stand. Secondly, the marginal model quantifies the averaged effect of the tree properties on the probability of cut. It is associated with a transition model describing the influence of the previous decisions and having independent parameters. It allows to break off the 'history' and the path followed. By simulation, we have marked trees corresponding to a determined cut, in different cases of decisions correlations. The parameters of the models have been estimated using forester data and different behavior hypothesis have been tested. By simulations, we have studied the variability of the cut according to its aims, depending on the path followed and the parameter relative to the decision correlations.
