Mining and evaluation of stressed soils: Stochastic aspects of empirical modelling approach

摘要

In this paper attempts are made to present a new form of multivariate empirical model for bioindication of topsoil fertility apart from site-specific conditions. In presented simplified version, the model is based on comparison of measured values with model-predicted minirmim for microbial biomass carbon (MB-C) and with model-predicted maximum for extracellular organic carbon, extractable in 0.5 M K_2SO_4 (EX-C). Model estimates were derived from correlations between estimated microbial parameters and important soil abiotic factors (C_(ox), pH_(water), pH_(KCl), soil texture, Q_(4/6)). Relatively small amount of required parameters makes the model suitable even for screening evaluations of relatively high number of sites. The efficiency of model prognoses to discriminate among different anthropogenic and agricultural sites was proved using standard stochastic methods. Ecotoxicological relevance of involved microbial parameters, practical advances and also limitations of this approach are discussed.