2 COMPETITION INDEXES FOR COMPARING THE HEIGHT AND THE DIAMETER GROWTH OF TREES CHARACTERIZED BY VARIOUS AND UNKNOWN SILVICULTURAL HISTORIES

For the moment, the present study is mainly exploratory. The indices proposed are based on the "postulate": the height/diameter ratio (H/D) of a tree depends on its average past competition status but is independent of the site conditions. The mean (H/D) ratio for trees subjected to the same silvicultural treatment is related to age too, in accordance with the following model: Log (H/D) = a-b.Age (1). The age, the height H and the diameter D of a given tree are measured. Then two competition indices Cd and Ch are created as Cd.D = Dr and Ch.H = Hr, where Hr and Dr are the dimensions of a reference tree that would be the same age and characterized by an average competition status. Both Hr and Dr are unknown, but the Hr/Dr ratio can be calculated according to (1). Thus, Cd/Ch is well-defined, and called alpha. For the moment, a very simple model is proposed to produce the competition indices: Cd = (alpha) (1-k) et Ch = (alpha) -k (2) Dendrometrical data available for 505 pedunculate oaks (Quercus robur L) from the Lorraine plateau were used to determine empirically k:k = 0.3. Initial attempts have been achieved in order to test the relevance of the indices proposed, using the same data and those of 529 sessile oaks (Q petraea (Matt) Liebl) from the same forests. Without using the competition indices, the radial growth of both species does not exhibit any significant long term trend, which contrasts with previous results from Silver fir (Abies alba Mill) and Beech (Fagus silvatica L) in the Vosges and the Jura mountains. After compensating ring widths with Cd, a clear increase since the year 1830 becomes evident: +60% for sessile oak, +32% for pedunculate oak. Furthermore, stepwise multiple regressions have been carried out in order to predict tree height using some ecological data. The variance explained is higher when the competition index Ch is used: 29% vs 22% without this use, and even 54% vs 37% in well-drained sites. Further validation tests are desirable, especially in order to confirm the reliability of the basic postulate. Moreover, it is likely that the model (2) could be improved. The indices Cd and Ch could be of great interest for studying site-yield relationships in the forests characterized by a perturbed silvicultural past, and, in dendroecological studies, for analysing the ecophysiological determinism of the radial growth of the main forest species.