The Determination of Normalizing Transformations for Nematode Count Data From Soil Samples and of Efficient Sampling Schemes

Abstract

Soil sampling experiments were conducted on 0.01 hectare (ha) plots of Fox loamy sand in areas of low, medium, or high population densities of Pratylenchus penetrans (Cobb). Nematode count data from 1, 20, and 40 core samples were statistically analysed to find normalizing transformations and to compare the efficiency of different sampling schemes. Although negative binomial distributions fitted the observed frequency distributions, normalizing transformations of the form log (x + c), that stabilized variance for all population densities, were obtained empirically. It was estimated that at least seven hours of field sampling and laboratory analysis are required to obtain an estimate of mean population density of a plot within 20 % of the true mean with 95 % confidence. An estimate of nematode population density in a 0.01 ha plot based on a subsample from one 20 or 40 core sample is shown to have very low precision.