Application of Burnaby's and Goodall's similarity indexes for local soil classification

Land use assessment is among the practical purposes of soil classification. Several researches has been specifically focused on the use of conventional surveys in the evaluation of soil suitability for agriculture products whose quality is influenced by the interactions between soil, plants and the biological stock of the rhizosphere. Our aim was to expand the application of the soil suitability protocol by relating soil types with biodiversity and the ecological equilibria of biological communities. This goal can be pursued by combining the approaches to soil type delineation with the techniques of quantitative ecology that are based on the similarity theory. Given the qualitative scale of several field-recorded attributes make their numerical processing difficult, we focused on numerical techniques for multivariate sets of data to combine with geostatistics. We thought these techniques would originate local soil classes meaningful in terms of both soil processes and soil suitability evaluation. Since auger boring data are formally comparable to vegetation data, we tested Goodall's and Burnaby's pairwise similarity indexes: the former assumes that pairs of observations sharing an infrequent value are more similar than pairs which share more frequent values; the latter considers associations among attributes, giving higher weights to independent ones.We did an intensive soil survey in a 1200-ha flood plain of the Istria region, Croatia. The morphological characteristics of soil cores were recorded and analysed to produce pairwise similarities that were partitioned by hierarchical clustering into similarity vectors. Such vectors were in the end submitted to geostatistical analysis for the drawing up of similarity maps. Both similarity measures originated five partially overlapping clusters that were consistent with the main soil forming processes present in the investigated area. Goodall's index gave the most meaningful results, fulfilling three compulsory requirements for soil mapping: i) similarity vectors were meaningful in terms of fluvial dynamics; ii) similarities displayed a structured spatial variability; and, iii) similarity maps were consistent with the soil forming factors acting in the investigated area. The results obtained indicate that Goodall's similarity index could be currently used in soil suitability evaluation, allowing to better exploit field-recorded data and to extend its application to the relations existing between soil types and the ecological equilibria of biological communities.