Fe-Mn concentrations in upland loess soils in mid-continental north America: A step towards dynamic soil survey

Abstract

Iron-manganese (Fe-Mn) concentrations are common morphological indicators of soil oxidation-reduction reactions and are widely used to infer soil hydrology. The factors determining occurrence and distribution of Fe-Mn concentrations at the continental scale have been rarely studied. The objective of this study was to explore the distribution of Fe-Mn concentrations in upland loess soils and identify their favorable environment conditions in the mid-continental North America. A total of 3966 soil profiles were collected from the National Soil Information System database (NASIS). A decision tree (DT) model was used to analyze the spatial relationships between Fe-Mn concentration formation and its relevant environmental factors (climate, soil properties, and land use/land cover), and to identify key factors of Fe-Mn concentrations distribution and occurrence in upland loess soils. The DT model was evaluated with 25% validation dataset with an accuracy of 79%. Results showed that Fe-Mn concentrations were more likely to occur in mesic areas with large precipitation and low internal soil drainage rates, compared to frigid and thermic areas with low precipitation and large internal soil drainage rates. Annual precipitation and temperature were identified as the most important climatic factors influencing the occurrence of Fe-Mn concentrations and showed a complex interaction with soil properties at multiple scales. Soil properties related to potential wet to saturated soil conditions were also important for the occurrence of Fe-Mn concentrations. Results demonstrated the importance of using knowledge of pedogenesis in the selection of environmental covariates and the interpretation of DT model results. Understanding where Fe-Mn concentrations tend to form in the landscape will contribute to improving decisions about proper use of soil that may be limited by wetness and possible inferences about climate influences.