Long-term seasonal changes of the Danube River eco-chemical status in the region of Serbia

Abstract

Seasonal spatial and temporal changes of selected eco-chemical parameters in section of the Danube River flowing through Serbia were analyzed. Data for electrical conductivity (EC), dry and suspended matter, residue on ignition, chemical oxygen demand (COD), biochemical oxygen demand (BOD-5), ultraviolet extinction, dissolved oxygen (DO), oxygen saturation, pH, nitrates, total phosphorus, and nitrogen were collected between 1992 and 2006. The use of monthly medians combined with linear regression and two-sided t test has been proven to be the best approach for resolving trends from natural variability of investigated parameters and for determining trend significance. Patterns of temporal changes between different months were examined. It was also determined that spatial trends of some parameters oscillate in predictable manner, increasing in one part of the year and declining in the other. Regression slope coefficients, an excellent indicator for determining when the water quality is changing the most along the course of the Danube, reach their maximum during summer for temperature (t), electric conductivity, nitrates, and total N, while in the same season suspended matter, COD, BOD-5, DO, and oxygen saturation coefficients reach their minimum. Correlations for used data sets of selected parameters were analyzed for better understanding of their behavior and mutual relations. It was observed that as Danube flows through Serbia, its general eco-chemical status either stagnates or improves, but the rate of river self-purification often depends on the season of the year.

Statistical analysis of the influence of major tributaries to the eco-chemical status of the Danube River

“…With a few exceptions, Kendall rank test did not report any statistically significant lack of randomness, which implicates that temporal linear trends do not significantly affect data independence. The test did not get statistically significant results at P=0.05 level even in some cases when temporal trend was detected in the previous research performed by Živadinović et al (2010) and Ilijević et al (2012). Low power of Kendall test is consequence of high variability which substantially exceeds the gradual changes observed during time.…”

“…This characteristic of the temperature leads to very inflated variance compared to the mean, which deteriorates the ability of ANOVA to determine any statistically significant difference among analyzed sampling locations. Nevertheless, in all other cases, the post hoc tests successfully divided average values of the analyzed sampling locations into different subgroups (Table 6), including parameters like DO, EC, nitrates, alkalinity etc., whose values also considerably oscillate depending on the season of the year (Ilijević et al 2012).…”

“…The same data has already been used to estimate eco-chemical status of the Danube River and to determine long-term trends of the eco-chemical parameters change (Ilijević et al 2012;Živadinović et al 2010).…”

“…The outliers were excluded to avoid their influence. We have already investigated existence of seasonal and long-term temporal trends in our previous research (Ilijević et al 2012;Živadinović et al 2010). Most of the parameters had negative temporal trends, especially BOD-5, COD, nitrates, and total P. Detection of temporal trends can be significantly facilitated when yearly and monthly medians are used instead of raw data.…”

“…Advantage of this approach is that the samples are collected by trained professionals and that laboratory analysis was performed according to the well-known and accepted standard methods which were already specified by Živadinović et al (2010) and Ilijević et al (2012; along with the sampling procedure and description of the sampling sites). Many scientific projects focused on the research of the rivers are limited to the shorter periods of time of just a few months or at best a few years.…”