STATISTICAL TOOL FOR SOIL BIOLOGY .10. GEOSTATISTICAL ANALYSIS

Soil living organisms currently exhibit complex spatial patterns at va rious scales. Conventional methods for studying spatial dispersion are based on various aggregation indices or probability distribution anal ysis. Since these methods do not take into account the actual location of the sampling sites, they provide no information on the spatial dis tribution at scales larger than the sampling unit size. The geostatist ical analysis is a way to analyse the spatial pattern of a variable at scales ranging from the minimum to the largest intersample distance. The variogram indicates whether the variable is spatially dependent or not. If a structure is present, the kriging local interpolation proce dure provides estimates of the variable and their estimation error. Co ntour mapping of these values gives accurate maps of both the variable and the reliability of the estimated values. Kriging is a local estim ation method that yields fine description of short and large-scale str uctures whereas traditional interpolation procedure by trend surface a nalysis only describes large-scale patterns. At a further stage, the r elationship between two spatially dependent variables can be examined by cross-variogram analysis. The latter procedure allows the study of the complex relationships that occur either between biological variabl es or biological and environmental variables.