Iron occurrence in soils and sediments of a coastal catchment: a multivariate approach using self organising maps

The processes controlling the distribution and phases of Fe in soils and sediments of a forested coastal catchment in southeast Queensland, Australia are identified. The physicochemical attributes of 120 spatially distributed soil and sediment samples, along with geomorphic, geological and land-use characteristics were analysed using Kohonen's self organising maps (SOM) methodology. The SOM-based data analysis approach permits the analysis of complex multivariate datasets that are not well suited to traditional statistical methods. Across the catchment readily extractable Fe concentrations are low overall, despite a high proportion of Fe-concretions in many soils. However, two processes associated with elevated Fe occurrences are identified. Firstly, seasonal water-logging leads to the accumulation of Fe in clay-rich soils on lower slope positions. Mottling and the presence of lepidocrocite indicate cyclical redox conditions. Secondly, a high proportion of total readily extractable Fe is present in organically complexed form in the stream sediments. It is suggested that an accumulation of Fe-rich sediment and organic detritus in streams permits the release of Fe under anoxic conditions, which is then bound by organic material. Contrary to the findings of previous studies, vegetation type does not affect Fe; there is no major difference between the concentrations of Fe of soils under pine plantation and native vegetation. Landform, however, does play an important role. Local depressions on gentle-sloped, low-lying terrain and associated with high topographic wetness indices were found to enhance seasonal redox processes. The groupings obtained from the SOM analyses were internally consistent and can be regarded as 'process groups'. The results show that SOM is a tool that can aid in the interpretation of complex datasets and help identify geochemical processes operating on a catchment scale.