The use of covariance analysis to improve a forestry experimentation precision

The effect of a thinning method on a tropical forest dynamics was examined using an experimental approach. This method is closed to forestry economy and it consists in eliminating non commercial species trees in order to increase the ingrowth rate of the whole commercial ones. This paper shows how a statistical technique, covariance analysis, can be used to cure an insufficiency of the experimental material. This insufficiency results here in inequality between initial total basal area of the treated quadrats and that of the untreated quadrats. The profit of precision is 1.054. It provided the following estimates of the increase rate in basal area of the commercial species: 0.73 and 0.92 m²/ha/year respectively for the treated quadrats and the untreated quadrats. When initial total basal area of the quadrats is not regarded as an auxiliary variable, the equivalent values are 0.53 and 1.05 m²/ha/year.