The knowledge of size-distribution and lability of metals and nutrients in freshwater systems is important for estimation of the ecological effects of mining. However, it is still limited in several mining areas such as the Quadrilátero Ferrífero (Brazil) which was severely polluted by the collapse of the Fundão tailings dam in November 2015. In this study, results of an investigation from 2014 using a neural network named self-organising map (SO-Map) into the conditions of selected trace metals that are of particular importance to mining areas (Cr, Cu, Co, Mn, Ni, Pb, Zn) are presented. Additionally, P was considered by its high importance as a nutrient and sites later affected by the dam burst were also included by chance. Water samples were collected at six sites in dry and rainy seasons and filtered and ultrafiltered for determination of total dissolved (<0.45 μm) and truly dissolved (<1 kDa) fractions. Diffusive gradients in thin films (DGT) devices were deployed in situ for determination of the DGT-labile fraction. All data were analysed using SO-Map and Spearman's rank correlation. Phosphorus in the Carmo River occurred mainly in the truly dissolved and DGT-labile fractions. The higher amounts of this element in the river water (up to 263 μg L-1 of total P) might be related to untreated sewage discharge. Moreover, the concentrations of other trace metals (Mn, Cu, Co, Ni, Zn) were high, even under the "natural" conditions (before the dam failure) due to natural and anthropogenic factors such as local lithology and mining.
Keywords: Diffusive gradients in thin films; Kohonen neural network; Mining impacts; Size fractionation; Ultrafiltration.
Copyright © 2022 Elsevier Ltd. All rights reserved.