Statistical power of commonly used point methods in grassland monitoring

Abstract

It has become almost universally accepted that the management of conservation areas and game ranches be subject to pre-determined sets of objectives and goals. Key to management decision making are the results of vegetation monitoring protocols that are designed to indicate changes in vegetation status. These changes are frequently measured as a proportional species composition and measure of basal cover and based on samples of 100-200 points in specific fixed sites. Minor changes of the magnitude of the order of single percentage points often form the basis of a management decision relating to herbivore stocking rates. These values are often quoted sans confidence limits or any other measure of dispersion. This paper investigates the statistical power of within stand replicates of three commonly encountered point methods, replicated within the same stand on Suikerbosrand Nature Reserve, Gauteng Province, and shows unacceptably high Type II error probabilities.