Date of Award
Doctor of Philosophy
The Lake Erie Fisheries Program of the Ontario Ministry of Natural Resources used a stratified sampling design with partial replacement of catch stations from 1987 to 1990 to monitor the fish community. Estimation methodology associated with partial replacement sampling design based on double sampling regression estimates was not suitable for the small sample sizes encountered and for multivariate descriptive study.;An unbiased estimation methodology for detecting changes and for descriptive study has been developed for two or more sampling occasions in either a simple or a stratified statistical population. The methodology for two occasions links well known cases in statistical inference theory, such as the Student t-test, the Satterthwaite t-test and the paired t-test, in a single inference procedure where the previous tests are specific cases. For more than two occasions, the methodology links well known oneway anova between years (independent samples) and the two-way anova with one replicate per cell of a "year by unit" design (paired samples) in a single inference procedure where the preceding tests are specific cases. Rules for the computation of degrees of freedom for inference are suggested.;A Principal Component Analysis is also suggested to determine principal components with related patterns of change, habitat influences as well as other correlated patterns in the observations. The technique is based on the standardization of the covariance matrix with the sampling variances associated with the estimation methodology developed. This technique is compared with PCA on covariances and on correlations.;The suggested estimation methodology and PCA are applied to Lake Erie fish populations monitored in the area of the lake under study over the 1987-90 period.
Lemaire, Jacques, "The Lake Erie Program: Monitoring Fish Stock Changes Over Years" (1993). Digitized Theses. 2299.