Clean and accurate soil quality monitoring in mining areas under environmental rehabilitation in the Eastern Brazilian Amazon.

Soil quality monitoring in mining rehabilitation areas is a crucial step to validate the effectiveness of the adopted recovery strategy, especially in critical areas for environmental conservation, such as the Brazilian Amazon. The use of portable X-ray fluorescence (pXRF) spectrometry allows a rapid quantification of several soil chemical elements, with low cost and without residue generation, being an alternative for clean and accurate environmental monitoring. Thus, this work aimed to assess soil quality in mining areas with different stages of environmental rehabilitation based on predictions of soil fertility properties through pXRF along with four machine learning algorithms (projection pursuit regression, PPR; support vector machine, SVM; cubist regression, CR; and random forest, RF) in the Eastern Brazilian Amazon. Sandstone and iron mines in different chronological stages of rehabilitation (initial, intermediate, and advanced) were evaluated, in addition to non-rehabilitated and native forest areas. A total of 81 soil samples (26 from sandstone mine and 55 from iron mine) were analyzed by both traditional wet-chemistry methods and pXRF. The available/exchangeable contents of K, Ca, B, Fe, and Al, in addition to H+Al, cation exchange capacity at pH = 7, Al saturation, soil organic matter, pH, sum of bases, base saturation, clay, and sand were accurately predicted (R2 > 0.70) using pXRF data, with emphasis on the prediction of Fe (R2 = 0.93), clay content (R2 = 0.81), H+Al (R2 = 0.81), and K+ (R2 = 0.85). The best predictive models were developed by RF and CR (86%) and when considering pXRF data + mining area + stage of rehabilitation (73%). The results highlight the potential of pXRF to accurately assess soil properties in environmental rehabilitation areas in the Amazon region (yet scarcely evaluated under this approach), promoting a more agile and cheaper preliminary diagnosis compared to traditional methods.

Phosphorus lability increases with the rehabilitation advance of iron mine land in the eastern Amazon.

Impacted areas by iron mining may face challenges in the management of phosphate fertilization and reduced efficiency of rehabilitation practices, thus extending the time required for the rehabilitation of these areas. The objective of this study was to evaluate phosphorus (P) lability in soils of native forest and ferriferous canga areas (savanna vegetation above ironstone outcrops covering iron ore deposits) and in iron mine waste piles undergoing rehabilitation. Benches of the analysed waste pile differ in age of rehabilitation: as the initial rehabilitation stage (INI), we consider benches with fewer than 3 years of rehabilitation; the intermediate stage (INT) were benches with up to 5 years of rehabilitation; and the advanced rehabilitation stage (ADV) corresponds to benches with more than 8 years of rehabilitation activities. Organic and inorganic P fractions were analysed in these areas by chemical fractionation and were classified according to the degree of soil lability. The results show that in the canga environment, there was a predominance of inorganic fractions of moderate lability and moderate stability, with a strong dependency of the soil organic matter (SOM) on the P fractions, whereas there was a greater participation of the moderately labile organic fractions in the forest than in the canga. On the other hand, in the rehabilitation areas, there was an increase in the labile organic and inorganic fractions as the rehabilitation process advanced. The distribution of P in areas undergoing rehabilitation indicates that there is a tendency for P levels to resemble those of native environments, such as the forests.