Thesis (Ph. D.)--University of Wisconsin--Madison, 1965. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references.

The problems that arise in developing a sampling design for the various life stages of the larch casebearer, Coleophora laricella (Hbn), are treated in relation to the changes in population density and distribution throughout the life cycle of the insect. The basic principles of population sampling are followed in respect to stratification of the sampling universe into its logical components. Replications and successive samplings are satisfied. The peculiarities of branch structure, with short shoots, long shoots and fascicles of needles provide criteria for the ultimate sampling units. Sampling was conducted with relation to: position of trees in the stand (interior, edge, or 'open grown1 trees); different crown levels; different branches at the same level based on exposure to sky-light; different 6-inch (15cm) segments of a branch throughout its length and different stages in the insect1s life cycle. The main purpose of this investigation was to test the hypothesis that shifts in population concentrations influence the accuracy of sampling at fixed points in the crown of a tree. If this be so, refinement of techniques may become possible. Of the theoretical distributions tested the negative binomial gave the best fit to the data for all life stages except the egg stage which approached the normal distribution. Analyses of two sample units, the needle fascicle and the 15~cm branch section respectively, revealed similar statistical distributions. The variance of the number of insects per needle fascicle calculated for each tree sampled was highly related to the mean. Therefore, approximate normality of the data was achieved by application of Taylor's power transformation, but with the modification of adding a "C" constant to the variable before raising it to the power of P,in the equation: Y. = (X. + C)'[sup P], where Y[sub i] = transformed observation, X[sub 1] = original observation, C = 1 and p = (1 - 1/2b) where b is a constant derived by the method of least squares. The analyses of variance showed that tree-to-tree, vertical and horizontal position of the samples in the tree crown were the most consistent variables influencing the distribution of eggs, larvae and pupae, while exposure of the tree in the stand had little effect. The distributions of eggs with relation to the quantity of needles, type of shoots, and condition of needles (oviposit ion sites) on the branch were examined. The determinant factors in insect distributions were also recognized. A practical three-stage sampling design was developed by considering variations between trees, and vertical and horizontal strata within the tree crown. The first stage is concerned with selection of the tree(s), the second stage would be the crown level within a tree, and the third stage the branches within each crown level. The variable to be estimated should be the number of insects per needle fascicle or short shoots (spurs) in winter. Such a sampling design would provide estimates of population trend and mortality within a generation. / Forestry, Faculty of / Graduate

Thesis (M.W.M.)--University of Michigan, 1953.