The method perfected by G. Matherson allows the study of f(.vector.x) variables, the values of which depend on the coordinates (.vector.x) of the points which support them. The variogram enables description of a spatial structure expressing the relation between 2 points separated by the vector h with the quantity: {f(.vector.x + .vector.h) .sbd. f(.vector.x)}2. The mean of this quantity is used. If the connection is only dependent on h we have a function of h, which is the semi-variogram: .gamma.(.vector.h) = 1/2 E {f(.vector.x + .vector.h .sbd. f(.vector.x)}2. Competition between trees brings about spatial structures which can be characterized by the shape of the corresponding variograms, made for the values of the circumferences at 1.30 m. This phenomenon was compared for plantations of various species, of different age or spacing. In particular, one can notice that clones and populations of different genotypes have very different behaviors. In a non-clonal population, the apparent influence of one tree upon its neighbors passes a maximum when the spacing grows smaller and smaller, although one can suppose that the real influence is increasing. This subdued effect of the apparent competition bears some consequences on genetic selection and experimental design.