Relative influence of natural watershed properties and human disturbance on stream solute concentrations in the southwestern Brazilian Amazon basin

[1] We use a synoptic sampling of stream water to quantify the effect of soil type, rock type, deforestation extent determined by Landsat TM imagery, and urban population density on stream solute concentrations for 60 different watersheds in the dry season and 49 in the wet season in the southwestern Brazilian Amazon basin. Catchment areas range between 18 and 12,500 km. Soil exchangeable cation content explains most of the variance in stream concentrations of cations, dissolved silicon (Si), and acid neutralizing capacity (ANC) in both forested and deforested basins based on regression analysis (R range 0.61–0.81), though the mechanism underlying the relationship is unknown. We use the relationship between soil exchangeable cation content and stream solute concentrations in forested catchments to estimate the predisturbance and postdisturbance signal concentrations of stream solutes for deforested watersheds. Signal concentrations of potassium (K), sodium (Na), and chloride (Cl) increase with deforestation extent for streams on gneiss, granite, and sedimentary rock in both the dry and wet seasons after the effect of soil type has been statistically removed. Sulfate (SO4) signal concentrations increase with deforestation extent in the dry season only. For catchments >40% deforested but with no urban populations, the ratios of disturbed to predisturbance stream concentrations range from 1.2 to 3.1 (K), 0.8 to 2.2 (Na), 0.7 to 2.7 (SO4), and 2.4 to 14.6 (Cl) in the dry season and 1.2 to 2.2 (K) and 0.6 to 3.4 (Cl) in the wet season. Urban population density strongly affects Cl and SO4 concentrations in both seasons and Na concentrations in the dry season; the urban signal comprises 40–58% of the dry season Cl signal and 83–89% of the wet season Cl signal for watersheds with >250 persons per km. Streams on mica-schist, mafic rock and carbonate shale have higher concentrations of Ca,Mg, Si, andANC thanwatersheds on gneiss, granite, tertiary sediments or sandstone for a given soil exchangeable cation content in both seasons. Simple mass balance calculations suggest that supplemental cattle salts could comprise a significant fraction of the Na and Cl signals in the wet season but would not significantly impact wet season concentrations of Ca, Mg, or K.