Sediment apportionment in paired, small watersheds with runoff convergence from no-tillage forage and grain production in the subtropics

Abstract

Areas with diversified livestock farming and intense agricultural use can lead to soil degradation, particularly in environmentally fragile regions. We collected samples of suspended sediments and potential sediment sources for five years (between 2016 and 2020), and determined geochemical tracer properties using inductively coupled plasma optical emission and X-ray fluorescence. We applied a combined sampling strategy based on land uses, being agricultural areas and unpaved roads, and erosion forms of drainage channels. In paired watersheds occupied with dairy and grain production, draining into one of the largest artificial lakes in southern Brazil. The geochemical tracers (Mn, Ti, and Fe) were chosen based on the conservativeness test, Kruskal-Wallis, and LDA correctly classified 77% of the sediment sources in both watersheds. The average RMSE and MAE for virtual mixtures were, respectively, 17% and 16% in the north watershed and 18% and 17% in the south watershed. The MixSIAR model showed unpaved roads contributed to most sediments (38% and 55%), followed by channel banks (37% and 30%) and crops (25% and 15% for the north and south watersheds, respectively). Unpaved roads are allocated perpendicular to the main contour lines, flow accumulation channels, surface flow direction, and landscape slopes. Thus, roads concentrate runoff and sediment from agricultural fields. The sediment apportionment reveals the need to focus on runoff control and road maintenance to reduce the risk of sedimentation and contamination of reservoirs, avoiding the loss of storage capacity for the supply of water and production of hydroelectric power.