Echantillonnage des populations aphidiennes en vergers de pechers. Comparaison de la precision de deux methodes a l'aide de modeles de simulation d'echantillonnage

English. Two methods of aphid sampling in peach orchards were compared. The first, a branch-cutting method, is based on a one meter length of branch cut per tree. The other consists of a visual observation of the whole tree for 2-4 min according to foliage density, using an infestation scale. When the classes in the scale were calibrated by comparison with the results from the first method a geometrical progression of aphid numbers was noted. The use of models to simulate sampling allowed comparison of the accuracy of the 2 methods. The number of sampled trees in every class was fitted to the POLYA-AEPPLI distribution (ANSCOMBE 1949). In the branchcutting method, we simulated the distribution of aphid population on a tree, using different distribution models. In the visual method, we constructed 2 probability matrices assigning an infested tree to every visual class (table 5). The coefficients of variation of the 2 sets of results were then fitted to the variation of every parameter (table 4 and 6). For low densities of inconspicuous species, which develop rather small colonies, well distributed in the whole tree, the branch-cutting method was slightly more precise. However, for high densities of conspicuous species whose colonies can reach some size, the visual method clearly showed greater efficiency. It also had the advantage of being paster and non-destructive, and allowing the dynamics of aphid populations on every tree to be followed.

French. Deux méthodes d’échantillonnage des pucerons sont comparées en vergers de pêchers : l’une est basée sur la coupe d’un mètre de rameaux par arbre échantillonné, l’autre sur l’examen individuel de l’arbre entier et l’utilisation de classes d’infestation. L’étalonnage de ces classes met en évidence une progression géométrique des effectifs aphidiens. L’utilisation de modèles de simulation d’échantillonnage permet de comparer la précision des 2 méthodes. Elle dépend essentiellement du degré d’infestation du verger et de l’espèce considérée. La méthode visuelle est généralement la plus précise, sauf lorsque la densité est très faible et les espèces peu visibles.