ROBUST REGRESSION-ANALYSIS OF GROWTH IN BASAL AREA OF NATURAL PINE STANDS IN GEORGIA AND ALABAMA, 1962-72 AND 1972-82

Net growth and gross growth in basal area of selected plots of natural pine stands in Georgia and Alabama are examined under previously used models. Since large scale survey data may contain outlying observatio ns, we use a procedure based on a linear model that is resistant to th e influence of outliers. Our objective is to determine if the results of a previously used model hold when a linear model is fit to the data using our robust procedure. The data are drawn from forest inventory analysis measurements over two periods (cycle 4 and cycle 5). The anal ysis includes a bootstrap testing procedure. This testing procedure do es not depend on parametric assumptions. Growth of the th me species s tudied in Georgia consistently showed a significant decline from the f irst period to the second period. A similar but less consistent declin e in growth was observed in Alabama.