On the interpretation of surprisingly high variation of soil map diversity in country-wide study of flood-affected agroecosystems using the legacy data in the Czech Republic

Abstract Legacy soil data have been produced worldwide to provide maps of soil classes or properties. In this study, we joined the legacy results of conventional soil mapping with the application of the pedodiversity concept and compositional data analysis in a comprehensive spatial analysis of the mapped soil diversity within flood-prone agroecosystems in the Czech Republic. After the statistical differences in Shannon’s entropy (conventional and adjusted with taxonomic distances) over different regional inundation areas had been statistically examined, their interpretation was based on Aitchison’s geometry of simplex, rendering an appropriate statistical framework to avoid methodological bias through the use of raw proportions in soil taxa compositions. Working with tailored orthonormal coordinates enabled us to verify the dependency of the soil taxa composition on the multi-category factors of agroecosystem type and flood-periodicity through a conventional linear modelling. The supervised compositional approach improved interpretability of soil information within the Czech legacy soil maps, and hence general patterns of pedodiversity in floodplains might be effectively described in an informative and quantitative manner. The most obvious differences of the soil map diversity proved the regional inundation areas within the catchments with the enhanced complexity of drainage network that might arise from multiple sources (substrate diversity, climatic zonality, man-induced landscape alteration).