|
Résumé :
|
Knots correspond to the parts of branches included in the wood. They are classified into different categories according to the status of the branch at the time of their inclusion. They have various sizes and shapes. Knots are considered as defects lowering wood quality. They are one of the major criteria for wood classification. Local environment, through resource access and competition, governs tree morphology, in particular crown shape and size. Therefore, it is clear that the forestry choices regarding the stand influence branching as well as the associated formation of knots. Decision-support tools developed for forest managers or industry are only of interest if they make it possible to evaluate wood quality. They are designed in order to simulate various forestry strategies. Therefore, predictive models of wood quality criteria have to be able to reproduce the effects of these choices or strategies correctly. This is all the more important as since a few decades foresters tend to reduce plantation density and to intensify stand management. In this paper, we report on Scots pine branch diameter profiles in an experimental design of a 35 years old plantation spacing. Four plantation densities were tested in this trial, ranging from 1 320 stems/ha to 10 000 stems/ha. Several thinning schedules were applied for each initial density. The stands were divided in plots, so as to diversify the growth conditions. Branch measurements were performed on six trees in each of ten plots, representative of the range of forestry scenarios tested in the experimental design. At this stage, living and dead branches were treated together. The data were used to built a model of branch diameter profile along tree stem. Only individual tree variables were introduced into the model. Stand variables were not meaningful. We show that this model is not relevant as it poorly accounts for the effects of the silviculture on the branching in the dead branch area. The branch diameters were over- or underestimated depending on whether the initial density was high or low, which all together led to a minimization of the real impact of density on branchiness. Differences between thinning treatments were also noticed, although less significant. Therefore, such a model should not be used to compare forestry management strategies. However, in the upper part of the crown, model predictions are correct whatever the scenario. At this level in the crown, tree characteristics and branch age are sufficient to explain a great part of branch diameter variability, without bias. We conclude that it is necessary to calibrate a branchiness model on the basis of living branches only and, in addition, to estimate crown recession accurately in order to characterize dead branches. Living branch diameters depend on tree characteristics whereas natural crown recession depends on stand variables essentially. Consequently, we developed a new model for the prediction of living branch diameter profile along the stem. We lumped together the first sample with data previously collected in other trials to cover a greater range of stand ages, from 24 years to 35 years. Then we calibrated a provisional crown recession model at stand level. Complementary studies are needed to generalize and validate this first approach but this yet allows to make the system operational. Recently we added a module named Sylvestris to the forest modelling platform Capsis (Coligny). This module is aimed at simulating Scots pine growth and yield in pure and even-aged stands in France. The integration of the present branchiness equations into the module will allow to make a first assessment of the relevance of the forestry management methods regarding wood products quality.
|
