RESUMO
The transfer of radionuclides from soil to the food chain often begins with uptake by plant root system. The roots of most angiosperms showed symbiosis with arbuscular mycorrhizal fungi (AMF) and to understand the transfer process of these toxic elements it is important to consider different physical, chemical and biological factors in soils. In the present study, three grass species (Poaceae), Zea mays, Chrysopogon zizanioides and Aristida setifolia were cultivated with and without organic fertilization in experimental blocks on natural soils, at Fazenda Vargem Formosa (VF) with low uranium (U) contents in the soil, and in the leached ore deposit at the Uranium Concentrate Unit Mine (URA) in Caetité (Uraniferous Province of Lagoa Real - Brazil). In the present study, the biomass production of plants, their rate of root colonization by AMF, the levels of U in soils, roots and leaves, as well as different physico-chemical parameters related to soil fertility were evaluated. The data analysis was performed using Artificial Neural Networks (ANNs), specifically Self-Organizing Maps (SOMs). The levels of available uranium in the soil ranges from 0.33 to 1.11 mg kg-1 in VF and from 177.5 to 475.8 mg kg-1 in URA. The results revealed high percentage of root AMF colonization, even in soils with high U contents. There was an inverse relationship between soil U content and its transfer to the plant organs, with U transfer rates being influenced by plant species and not by soil parameters. C. zizanioides had the lowest transfer factor to the shoot and the highest mass productivity under conditions of high U content in soil. The results indicate that C. zizanioides is an important species for use in the recovery of U mining areas.