Sediment fingerprinting in agricultural catchments: A critical re-examination of source discrimination and data corrections

Abstract

Fine sediment source fingerprinting techniques have been widely applied in agricultural river catchments. Successful source discrimination in agricultural environments depends on the key assumption that land-use source signatures imprinted on catchment soils are decipherable from those due to other landscape factors affecting soil and sediment properties. In this study, we re-examine this critical assumption by investigating (i) the physical and chemical basis for source discrimination and (ii) potential factors that may confound source un-mixing in agricultural catchments, including particle size and organic matter effects on tracer properties. The study is situated in the River Tamar, a predominantly agricultural catchment (920km2) in south-west England that has also been affected by mining. Source discrimination focused on pasture and cultivated land uses and channel banks. Monthly, time-integrated suspended sediment samples were collected across seven catchments for a 12-month period. Physical and chemical properties measured in source soils and sediment included fallout radionuclides (137Cs, excess 210Pb), major and minor element geochemical constituents, total organic carbon and particle size. Source discrimination was entirely dependent on differences in tracer property concentrations between surface and sub-surface soils. This is based on fallout radionuclide concentrations that are surface-elevated, while many geochemical properties are surface-depleted due to weathering and pedogenetic effects, although surface soil contamination can reverse this trend. However, source discrimination in the study catchments was limited by (i) rotation of cultivated and pasture fields resulting in reduced differences between these two sources, and (ii) the cultivated source signature resembling a mix of the pasture and channel bank sources for many tracer properties. Furthermore, a combination of metal pollution from abandoned historic mines and organic enrichment of sediment from upland areas of peaty soils resulted in the non-conservative behaviour of some tracer properties in several catchments. Differences in the particle size and organic carbon content of source soils could explain much of the variation in these properties in downstream sediment, rather than selective transport effects. Inconsistent relationships between particle size, organic carbon and tracer property concentrations further undermined the basis for the use of widely applied corrections to tracer datasets. Sensitivity analysis showed that correcting source tracer data for differences in organic matter can produce large changes to source contribution estimates that cannot be justified, and such corrections should not be used. Confounding factors related to poor source discrimination and non-conservative behaviour are highly likely to affect sediment fingerprinting studies in many agricultural catchments. As a result, estimates of source contributions in many fingerprinting studies may contain significant unquantified errors. © 2013 Elsevier B.V.