Sampling design and precision of basal area growth and stand structure in uneven-aged northern hardwoods

Abstract

Sampling design (distribution of plots, plot sizes, and number of sample plots) is an important consideration in planning a field study since it determines the bias and precision of information obtained. In this study, we evaluated the effects of two commonly used sampling designs on the precision of estimates of means related to stand basal area, considered as a whole or by diameter class, and 10-year growth components, in six 2-ha areas located in uneven-aged northern hardwood stands of Quebec managed under selection cutting. The study compares random and systematic sampling designs of one to 10 plots, with plots varying in size from 0.05 ha to 1.25 ha. Increasing the total area sampled (i.e., sampling area) in a stand from 0.05 ha to about 0.25 ha resulted in large precision gains; further increases in sampling area had more limited effects on precision. A sampling area of at least 0.5 ha would be required to obtain a minimally acceptable precision in estimating means of total basal area and 10-year growth components. A much larger sampling area would be required to obtain sufficient precision in estimating mean basal area by diameter class, often required to study stand structure. Precision can also be increased by dividing the sampling area into several smaller plots rather than using a single large plot. We found no clear difference in the results between random and systematic selection approaches.