An analysis of variability in the activity of nitrifiers in a soil under pasture .II. Some problems in the geostatistical analysis of biological soil properties

Abstract

An investigation of spatially dependent variability in the activity of soil nitrifiers indicated a number of shortcomings in the application of geostatistical methods to the analysis of spatial variation in biological soil properties which inherently have high variability. The relationship between the sill variance (Co+C) and sample variance (s2)), problems in the identification of anisotropic variation, and the effect of the scale of sampling on s2 are considered in relation to the sampling design used for collection of sample data. It is concluded that for minimum complication, sampling designs should be symmetrical in at least four principal directions, and that some sampling sites should be positioned at points between the nodes of the grid to ensure as wide a spread as possible in the distances between pairs of observations.