Statistical Approaches Link Sources of Sediment Contamination in Subtropical Reservoirs to Land Use: an Example from the Itupararanga Reservoir (Brazil)

The identification and characterisation of diffuse pollutant sources in reservoirs remain a challenge due to the complexity of catchments with their variety of land use types. A sediment fingerprinting approach was used in this investigation for determining the sources of contaminants in sediments. By using this approach, we demonstrated how the effects of land use on pollution load of the subtropical Itupararanga Reservoir in Brazil can be de-convoluted. Sediments were collected at seven sampling sites (S1–S7) over the length of the reservoir. This was matched by eight sampling sites (P1–P8) of soils from different land use types (agriculture, urban and forest). Investigated parameters included grain size, total nitrogen (TN), total phosphorus (TP), total carbon (TC), organic matter by loss on ignition (OM), total sulphur (TS), and major ions and metals (Ca 2+ , Mg 2+ , Na + , K + , NH 4 + , SO 4 2− , Cl − , NO 3 − , As, Al, Ba, Cr, Cu, Fe, Mn, Ni and Zn). Our fingerprinting approach helped to outline horizontal spatial heterogeneities (categorised as riverine, transitional and lacustrine areas) that were attributed mainly to sand (> 26.7%), Si (569 g kg −1 ) and Cr (336 mg kg −1 ) at S1 (riverine area). Moreover, fine particles of silt and clay leached from agricultural activities were enriched with OM, TP, TN, TC, As and Cr. These types of sediments were deposited into transitional and lacustrine areas. Furthermore, urban soils were a source of sand and phosphorus to sediments. The fingerprinting method reduced the number of relevant parameters for source identification and helped to identify non-point sources of sediments.