Long-term trends of physical, chemical and biological variables in the River Danube 1957-1995: A statistical approach

A 39-year series (1957-1995) of data on fourteen physical, chemical and biological variables from the Austrian section of the River Danube west of Vienna-Nussdorf was analysed statistically to detect long-term trends of the variables in relation to human activities (represented by time), discharge and water temperature. Principal component analysis distinguished four main components explaining 72% of the total variance: PC1 contains total phosphorus, soluble reactive phosphorus, ammonium and potassium permanganate values, PC2 contains nitrate-N, chloride and oxygen, PC3 contains BOD5 and nitrite-N, and PC4 contains numbers of bacteria. Trends in time were most pronounced for variables in PC1, but also occurred in PC2; variables in PC3 and PC4 had no trends. Seasonal patterns were marked for variables in PC2, slightly less in PC1, and least in PC3 and PC4. Concentrations were minimal in summer and maximal in winter, inversely related to discharge (maximal in summer). Following reductions in point-source nutrient inputs to the Danube, in both Germany and Austria, mean concentrations in the river have fallen by at least half since the 1980s. Chloride and nitrate-N also show trends towards lower concentrations. The Danube is well-oxygenated, with concentrations near air saturation values. A mathematical relationship between concentrations of the variables and river discharge (Q), water temperature (T) and time (t), was established to determine mean trends and predictions against a background of considerable seasonal and stochastic variability; for the single variables Q explained 0-20% and T explained 2-58% of the variation in the concentrations. The relationship was highest for variables in PC2 and lowest for those in PC3 and PC4, where Q and T had little or no influence. Reasons for rising and falling long-term trends with time are discussed in detail. Austria now contributes only minor proportions to the nutrient load of the Danube, which is causing eutrophication of the Black Sea downstream, and water quality of the Austrian section of the river is good. The advantages and problems of statistical process analysis are discussed in relation to environmental monitoring programs and the different specific requirements of compliance monitoring.