Scholarly article on topic 'Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River'

Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River Academic research paper on "Environmental engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Water Research
OECD Field of science
Keywords
{"Standard faecal indicator bacteria" / "Microbial source tracking" / "Host associated genetic faecal markers" / Bacteroidetes / qPCR / "Large river" / "Joint Danube survey"}

Abstract of research paper on Environmental engineering, author of scientific article — A.K.T. Kirschner, G.H. Reischer, S. Jakwerth, D. Savio, S. Ixenmaier, et al.

Abstract The microbial faecal pollution of rivers has wide-ranging impacts on a variety of human activities that rely on appropriate river water quality. Thus, detailed knowledge of the extent and origin of microbial faecal pollution is crucial for watershed management activities to maintain safe water use. In this study, the microbial faecal pollution levels were monitored by standard faecal indicator bacteria (SFIB) along a 2580 km stretch of the Danube, the world's most international river, as well as the Danube's most important tributaries. To track the origin of faecal pollution, host-associated Bacteroidetes genetic faecal marker qPCR assays for different host groups were applied in concert with SFIB. The spatial resolution analysis was followed by a time resolution analysis of faecal pollution patterns over 1 year at three selected sites. In this way, a comprehensive faecal pollution map of the total length of the Danube was created, combining substantiated information on both the extent and origin of microbial faecal pollution. Within the environmental data matrix for the river, microbial faecal pollution constituted an independent component and did not cluster with any other measured environmental parameters. Generally, midstream samples representatively depicted the microbial pollution levels at the respective river sites. However, at a few, somewhat unexpected sites, high pollution levels occurred in the lateral zones of the river while the midstream zone had good water quality. Human faecal pollution was demonstrated as the primary pollution source along the whole river, while animal faecal pollution was of minor importance. This study demonstrates that the application of host-associated genetic microbial source tracking markers in concert with the traditional concept of microbial faecal pollution monitoring based on SFIB significantly enhances the knowledge of the extent and origin of microbial faecal pollution patterns in large rivers. It constitutes a powerful tool to guide target-oriented water quality management in large river basins.

Academic research paper on topic "Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River"

Accepted Manuscript

WATER RESEARCH

Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River

A.K.T. Kirschner, G.H. Reischer, S. Jakwerth, D. Savio, S. Ixenmaier, E. Toth, R. Sommer, R.L. Mach, R. Linke, A. Eiler, S. Kolarevic, A.H. Farnleitner

PII: S0043-1354(17)30625-5

DOI: 10.1016/j.watres.2017.07.052

Reference: WR 13095

To appear in: Water Research

Received Date: 27 February 2017

Revised Date: 19 July 2017

Accepted Date: 20 July 2017

Please cite this article as: Kirschner, A.K.T., Reischer, G.H., Jakwerth, S., Savio, D., Ixenmaier, S., Toth, E., Sommer, R., Mach, R.L., Linke, R., Eiler, A., Kolarevic, S., Farnleitner, A.H., Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River, Water-Research (2017), doi: 10.1016/j.watres.2017.07.052.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

2 (rtiL

La rtLii Ll. li rul dev elo p rtie rli of t. Jhd L rtti rOcacci C 0 ncertti Ationi jl aftg I he i r>.i e js i rtg diithi (gf ey =hided IT«) of tha □ jf lL e n ■ the midsti rim, ji the Itfft (uppii line) i^dthe i ight i l^er = ide (lam eMlrtfi) Arid irt the tnbuti ifij Sirrpled dunrtg JD5 2C_3. M flt-sirrpled ti ibtirICS ■!4 jliu irdiiclcd art t he V" IXUi 5 ymbaI =lie tf i i ei 4ccai 'Jrtg Iu Iag-ti j rt=f□ i(Yie d£. e<?ti ild EftterflCHCl £artcei\i A larts. Colour« depict the fArc ¿J pulkLiui- levels; blue - little, gi eert - madei its, yellam - Crt.tt.lc jJ, 01 MgC - sti artg tfid Fed WCHllvf pa II L±ia ft. Q i i s hided 11 lirtgle t i i ui cite d id hi gi c artti i L="_ I. larti f i am t he mi ft li ib Lti i ii.

Multiparametric monitoring of microbial faecal pollution reveals the dominance of human contamination along the whole Danube River

Kirschner AKT1,2*, Reischer GH2,3, Jakwerth S1,2, Savio D2,3, Ixenmaier S3, Toth E4, Sommer R1,2,

Mach RL3, Linke R2,3, Eiler A5,6, Kolarevic S7 & Farnleitner AH2,3,8 1

Medical University of Vienna, Institute for Hygiene and Applied Immunology, Water Hygiene, Kinderspitalgasse 15, A-1090 Vienna, Austria

2Interuniversity Cooperation Centre for Water & Health; www.waterandhealth.at 3Technische Universität Wien, Institute for Chemical, Biological and Environmental Engineering, Research Group of Environmental Microbiology and Molecular Diagnostics 166/5/4, Gumpendorferstrasse 1A, A-1060 Vienna, Austria

4Eötvös Lorand University of Sciences, Department of Microbiology, Pazmany Postny 1/c, H-1117 Budapest, Hungary

5eDNA solutions Ltd, Björkasgatan 16, 43131 Mölndal

6University of Gothenburg, Department of Chemistry and Molecular Biology, Medicinaregatan 9C, 40530 Göteborg

7University of Belgrade, Chair of Microbiology, Studentski trg 16, SRB-11000 Belgrade, Serbia Karl Landsteiner University of Health Sciences, Dr.-Karl-Dorrek-Straße 30, 3500 Krems, Austria

Correspondence to: Alexander K.T. Kirschner, Medical University of Vienna, Institute for Hygiene and Applied Immunology, Water Hygiene, Kinderspitalgasse 15, 1090 Vienna, Austria; alexander.kirschner@meduniwien.ac.at

Keywords: standard faecal indicator bacteria, microbial source tracking, host associated genetic faecal markers, Bacteroidetes, qPCR, large river, Joint Danube Survey

ABSTRACT

The microbial faecal pollution of rivers has wide-ranging impacts on a variety of human activities that rely on appropriate river water quality. Thus, detailed knowledge of the extent and origin of microbial faecal pollution is crucial for watershed management activities to maintain safe water use. In this study, the microbial faecal pollution levels were monitored by standard faecal indicator bacteria (SFIB) along a 2,580 km stretch of the Danube, the world's most international river, as well as the Danube's most important tributaries. To track the origin of faecal pollution, host-associated Bacteroidetes genetic faecal marker qPCR assays for different host groups were applied in concert with SFIB. The spatial resolution analysis was followed by a time resolution analysis of faecal pollution patterns over one year at three selected sites. In this way, a comprehensive faecal pollution map of the total length of the Danube was created, combining substantiated information on both the extent and origin of microbial faecal pollution. Within the environmental data matrix for the river, microbial faecal pollution constituted an independent component and did not cluster with any other measured environmental parameters. Generally, midstream samples representatively depicted the microbial pollution levels at the respective river sites. However, at a few, somewhat unexpected sites, high pollution levels occurred in the lateral zones of the river while the midstream zone had good water quality. Human faecal pollution was demonstrated as the primary pollution source along the whole river, while animal faecal pollution was of minor importance. This study demonstrates that the application of host-associated genetic microbial source tracking markers in concert with the traditional concept

of microbial faecal pollution monitoring based on SFIB significantly enhances the knowledge of the extent and origin of microbial faecal pollution patterns in large rivers. It constitutes a powerful tool to guide target-oriented water quality management in large river basins.

1 INTRODUCTION

The microbial faecal pollution of rivers negatively impacts a variety of human activities that rely on good river water quality. Among other reasons, the production of drinking water from river bank filtration, water used for crop irrigation and the watering of animals, and river stretches used for recreation are the most relevant river ecosystem services that directly affect the health of humans and animals. Because rivers are widely used as receiving waters for wastewater treatment plant (WWTP) effluents, the direct inflow of untreated wastewater and diffuse pollution, the microbial river water quality is often severely impaired (Kirschner et al. 2009, Servais et al. 2007). Even in regions with state-of-the-art wastewater treatment, high levels of microbial faecal pollution may occur (Mayer et al. 2016), unless WWTPs are equipped with a final disinfection step such as UV irradiation (Das 2001). In Europe, the environmental water quality is generally regulated by the EU Water Framework Directive (European Parliament & Council 2000) that defines environmental targets for all surface and ground waters in the European Union. However, the microbial water quality targets are only indirectly addressed and mentioned in the Urban Wastewater Treatment Directive (European Council 1991), the Bathing Water Directive (European Parliament & Council 2006) and the Drinking Water Directive (European Council 1998). There are no target values or regulatory values regarding microbial faecal pollution for river water at the European level, which may potentially lead to conflicts between upstream and downstream countries if a transboundary river is highly polluted.

The river Danube is the second-longest river in Europe with a total length of 2,780 km, and it is the most international river in the world. Its catchment area covers 801,500 km2, with approximately 81 million inhabitants in 19 countries (Schmedtje et al. 2005). Drinking water production from river bank filtrates and the supply of water for domestic, agricultural and industrial use are of major importance in all of these countries. Moreover, the Danube is an important international transportation route and recreation area (Kirschner et al. 2009). Microbiological pollution along the Danube may originate from point sources such as discharges of treated or untreated sewage from human sources or livestock enterprises and from non-point sources such as urban and agricultural runoff or wildlife. Microbiological contamination from faecal pollution by anthropogenic sources is considered to be a crucial problem throughout the Danube River basin, posing a threat to all types of water uses (Kirschner et al. 2009). Thus, detailed knowledge on the extent and origin of microbial faecal pollution is crucial for watershed management activities to maintain safe water use. For example, in the case of drinking water production from river bank filtrates, this knowledge can help water companies to derive performance targets for water safety plans, based on quantitative microbial risk assessment (Derx et al. 2016).

The extent of recent microbial faecal pollution in surface waters is usually determined by the use of standard faecal indicator bacteria (SFIB) such as E. coli and intestinal enterococci. According to the EU Bathing Water Directive (European Parliament & Council 2006), the assessment of the microbiological bathing water quality is exclusively based on these two parameters. However, the determination of the SFIB does not provide information about the pollution source. In the past decade, microbial source tracking (MST) has become an indispensable tool for uncovering the origin of microbial faecal pollution in different environmental water systems (Hagedorn et al. 2011). Among the several available

97 approaches (Caldwell et al. 2011, Hagedorn and Weisberg 2011, McQuaig and Noble 2011,

98 Wuertz et al. 2011), host-associated genetic Bacteroidetes markers have been shown to be

99 one of the most reliable targets for MST (Kildare et al. 2007, Reischer et al. 2013). Based on

100 qPCR detection, this approach is highly sensitive, specific, accurate and precise for MST in

101 combination with SFIB determination (Mayer et al. 2016), and it has been successfully

102 applied in a wide variety of aquatic ecosystems (Farnleitner et al. 2011, Liu et al. 2015,

103 Reischer et al. 2007, Stapleton et al. 2009).

104 In 2001 and 2007, the microbial faecal pollution of the river Danube was mapped along a

105 stretch of approximately 2,600 km during two Joint Danube Surveys organized by the

106 International Commission for the Protection of the Danube River (ICPDR) (Kavka and Poetsch

107 2002, Kirschner et al. 2009, Kirschner et al. 2008). However, in those investigations, only

108 midstream samples were analysed without considering the site-specific cross-sectional

109 pollution pattern. It has been hypothesized that in large rivers such as the Danube, the

110 mixing of polluted water masses from wastewater influents or tributaries only occurs slowly

111 and may take several dozen kilometres before being detectable midstream (Velimirov et al.

112 2011). Thus, potential pollution hot-spots may have been overlooked. In the current study,

113 microbial faecal pollution levels were therefore monitored in the midstream and at both

114 riversides as well as in the Danube's most important tributaries. To track the origin of faecal

115 pollution, host-associated genetic faecal markers for different host groups were determined

116 simultaneously at all the sampling sites along the river and the tributaries. The spatial

117 resolution analysis during the Joint Danube Survey 2013 was followed by a time resolution

118 analysis of faecal pollution patterns over one year in 2014 at three selected sites along the

119 river. In this way, a comprehensive, updated and extended faecal pollution map of

the Danube was created, combining substantiated information on both the extent and origin of microbial faecal pollution over the total length of a river that extends over a continental scale.

2 MATERIALS AND METHODS

2.1 Joint Danube Survey

The Joint Danube Survey (JDS) was organized by the ICPDR in 2001, 2007 and 2013. It is the world's biggest and most international river expedition with dozens of scientists and laboratories involved. In 2013, samples were taken by boat in the summer between Aug 12 and Sep 26 at 70 sampling stations along the longitudinal profile of the Danube in consecutive order from the upper section (river km 2,581) to the Delta (river km 18), including the major tributaries and branches at their confluence with the Danube. The sampling sites were positioned within regions that cover the whole range of different land-use classes such as large municipalities, areas with intensive agriculture, forests, grasslands or wetlands. A detailed map of the river and the sampling stations can be found in the Suppl. Information (Fig. S1). In addition to the 68 official JDS stations (Fig. S1), samples were taken at two additional stations for microbiological analysis only; at the Inn tributary (rkm 2,225) and downstream from the city of Vienna (rkm 1,930). During the JDS, dry weather conditions and baseflow conditions prevailed, with a single precipitation event (25 mm on August 27), leading to an increase in the discharge from approx. 1,200 m3/s to 2,000 m3/s between sampling sites 10 (rkm 1,895) and 13 (rkm 1,869). Details on the discharge measurement and data can be found in the Supplemental information, Table S1. During the JDS, the Danube water temperatures ranged from 18 to 23°C.

2.2 Sampling

144 Water samples were collected by hand from small boats (anchored at the primary ship) at a

145 water depth of approx. 30 cm, according to ISO 19458:2006 (International Organization for

146 Standardization 2006), in two sterile 1-L glass flasks (flask A and B) from 54 sampling stations

147 along the Danube, and from 16 tributaries or branches. Because the vertical mixing of water

148 masses in the Danube is most likely a rapid process (Velimirov et al. 2011), samples from a

149 30 cm depth are regarded as representative for the whole water column. At all the Danube

150 stations (except station JDS1) and at the Inn, Drava, Tisza, and Sava tributaries, samples were

151 taken from the middle and the left and right sides of the river. The other tributaries and

152 branches were sampled only from the middle of the river. Tributaries were sampled at a

153 maximum of 500 m upstream of the confluence with the Danube. Samples from the left and

154 right sides were taken approx. 15 m away from the shoreline, and samples from the middle

155 were taken from the middle of the cross-sectional profile. All the samples were immediately

156 processed in the on-board laboratory of the primary ship.

157 2.3 Microbiological analysis

158 Escherichia coli and intestinal enterococci were used as general indicators of faecal pollution

159 (World Health Organization 2003), and they were determined using standard cultivation-

160 based methods. Genetic faecal Bacteroidetes markers associated with human faecal

161 pollution (BacHum, HF183II), ruminant faecal pollution (BacR) and porcine faecal pollution

162 (Pig2Bac) were determined via quantitative PCR (qPCR). For quality control purposes, the

163 AllBac marker was included in the analysis.

164 Escherichia coli - E. coli concentrations were determined according to ISO 9308-2:2012

165 (International Organization for Standardization 2012) with Colilert 18 (IDEXX, Ludwigsburg,

166 Germany), using two sample volumes (100 ml, 1 ml). The samples were incubated at 36 ±

2°C for 18 - 22 hours and analysed in a UV cabinet. Quantitative values were obtained by comparing the results with the MPN table provided by the manufacturer. Intestinal enterococci - Enterococci concentrations were determined by the standard method according to ISO 7899-1:1998 (International Organization for Standardization 1998) with commercially available MUD/SF microtitre plates (BIORAD, Vienna, Austria) using two dilutions (1:2 and 1:20). The microtitre plates were incubated at 43 ± 2°C for 32 - 40 hours and analysed in a UV cabinet. Quantitative values were obtained by comparing the results with the MPN table given in ISO 7899-1:1998.

Classification system for faecal pollution - To enable the assessment of the faecal pollution levels, the faecal indicators were classified by a system of 5 microbiological water quality categories according to Kavka et al. (Kavka et al. 2006) and Kirschner et al. (Kirschner et al. 2009) (Supplemental Information Table S2). To set up this scheme, two concentrations derived from the EU Bathing Water Quality Directive 2006 (European Parliament & Council 2006) were used as anchor points (1,000 CFU/MPN for E. coli and 400 CFU/MPN for enterococci). Faecal pollution levels in quality classes I and II are below these values, and quality classes III, IV, and V exceed these values. The EU Bathing Water Directive and the assessment of the bathing water quality could not be applied to the JDS data set since the bacterial indicator data for faecal pollution generated during the JDS are single measurements. According to the EU Bathing Water Directive, the assessment of the bathing water quality should always comprise at least 16 samples that are compiled in relation to the current bathing season and the three preceding bathing seasons, based upon 95-percentile and 90-percentile evaluations, respectively.

DNA filtration and extraction - From each sample, duplicate subsamples (one from flask A and one from flask B) were filtered through 0.2 ^m polycarbonate filters with a diameter of

47 mm. Depending on the turbidity of the samples, the filtration volume was between 100 and 300 ml. Clean control filters were frozen and stored alongside the sample filters at each sampling station. The filters were immediately frozen at -20°C, and after no more than 3 weeks, all the filters were transferred to a -80°C freezer in the home laboratory. DNA was extracted from the filters by phenol-chloroform extraction combined with bead-beating (Reischer et al. 2008). The extracted DNA was dissolved in 100 ^l of Tris buffer (10 mM, pH 8.0) and stored at -80°C until qPCR analysis. Extraction controls were routinely run alongside each extraction batch.

Microbial faecal source tracking assays - All the flask A samples were analysed using microbial faecal source tracking assays. In parallel, an additional 40 flask B samples were analysed to check for sample variability. For qPCR quality assurance and inhibition control, all the DNA extracts were measured by AllBac qPCR assay (Layton et al. 2006), which targeted the abundant Bacteroidetes phylum to ensure the presence of amplifiable bacterial DNA in duplex with an internal amplification control (IAC) (Anderson et al. 2011), targeting the ntb2 gene from tobacco to rule out PCR inhibition. Furthermore, all the samples were quantified in 1:4 and 1:16 dilutions for the additional assessment of PCR inhibition. For microbial source tracking, the human-associated faecal markers BacHum (Kildare et al. 2007) and a recently modified version of HF183II (Green et al. 2014) were determined by qPCR, indicating human-associated faecal pollution. The ruminant-associated BacR qPCR assay (Reischer et al. 2007) and the pig-associated Pig2Bac qPCR assay (Mieszkin et al. 2010) were included as methods for detecting these specific animal faecal pollution sources. All the qPCR assays were adapted to run on the Rotor-Gene Q thermocycler by using the RotorGene Multiplex PCR mastermix (Qiagen Inc.). Quantification was performed by running plasmid standard dilution series of known concentration. No-template controls were applied

to all the instrument runs. For the source-associated assays, all the samples (1:4 dilutions) were run in duplicate. None of the samples showed evidence of PCR inhibition, and all the filtration/extraction controls and PCR no-template controls were negative. PCR efficiencies (eff.) and coefficients of determination (r2) as calculated from linear regression of the standard dilution series for the applied assays were: AllBac (eff. 100-105%, r2 > 0.96), BacHum (eff. 98-105%, r2 > 0.97), HF183II (eff. 100-105%, r2 > 0.99), BacR (eff. 96-103%, r2 > 0.99), Pig2Bac (eff. 98-101%, r2 > 0.99).

2.4 Annual Cycle

To assess the temporal variability in the faecal pollution, monthly samples were taken at three critical sampling stations along the Danube in 2014, downstream from Vienna, downstream from Budapest and downstream from Belgrade. Vienna (1.7 million inhabitants) has been equipped with a secondary WWTP since 1980; it was upgraded with a second biological treatment step in 2005, and it cleans all the city's municipal wastewater (2 MPE). Budapest (1.7 million inhabitants) has had a central WWTP since 2010 that cleans the majority of the municipal wastewater (1.6 MPE). Belgrade (1.35 million inhabitants) does not have a WWTP.

In Vienna and Budapest, monthly samples were taken from bridges (left, middle, and right) downstream from the treated wastewater inflow. In Belgrade, samples were taken from small boats downstream from the city. The samples were immediately brought to the home laboratories in insulated cooling boxes and processed as described above. In Belgrade and Budapest, DNA samples were stored at -20°C and brought to Vienna in July 2014 and February 2015. MST marker determinations were performed by qPCR as described above.

2.5 Statistical analysis

All the statistical analyses were performed with SPSS 22 (IBM, New York, USA). All the microbiological data were log10-transformed after adding 1 to each value. A Spearman rank correlation and a simple linear regression were used to study the relationships among the different faecal indicators and source tracking markers. Significance was accepted at p < 0.05. To study the relationship between the microbial faecal parameters and the background environmental data, a principal component analysis (Varimax rotation with Kaiser Normalization) was performed. All the JDS 2013 background data were retrieved from the ICDPR Danubis database (https://www.icpdr.org/wq-db/) with the kind permission of the ICPDR.

3 RESULTS

3.1 Longitudinal faecal pollution patterns reveal critical pollution in the middle section of the Danube, several hot-spots along the whole river and in specific tributaries

Along the upper section of the Danube (Germany, Austria and Slovakia, rkm 2,600 to 1,800), all the midstream samples exhibited low to moderate E. coli pollution levels (Figure 1A; the same graph including all WWTPs > 40.000 population equivalents can be found in the Supplemental Information, Fig. S2). However, at three sampling stations, the left and right near-shore sites showed critical and strong pollution. Surprisingly, at Kelheim (Germany), the left riverbank sample exhibited the highest E. coli value of all the Danube samples (34,500 MPN/100 ml). Elevated pollution levels were also observed in Austria at two near-shore sites (Oberloiben left, 2,420 MPN/100 ml and downstream from Vienna right, 2,430 MPN/100 ml). The higher E. coli numbers from the right riverside downstream from Vienna can be explained by the inflow of the Danube channel a few hundred metres upstream of the sampling station. This channel holds the recipient water from the Vienna WWTP effluent. No

satisfactory explanation could be found for Kelheim and Oberloiben, except that both sites are situated next to a pier. All the other near-shore sites showed similar E. coli levels as the respective midstream sites. The three major investigated tributaries joining the Danube in this river section had 0.5 log (Inn, Morava) to 1.2 log (Moson Danube) higher E. coli concentrations than the Danube, with the Moson Danube receiving wastewater from the Hungarian city of Gyor.

In the middle section (Hungary, Croatia, and Serbia, rkm 1,800 to 900), several midstream stations showed critical E. coli pollution levels (Figure 1A). At Dunafoldvar, which is situated approximately 60 km downstream from Budapest, the midstream sample showed the highest E. coli values of all the midstream Danube samples (12,500 MPN/100 ml). From downstream from Budapest (rkm 1,632) until Banatska Palanka (rkm 1,071), not a single Danube sample was classified as having low E. coli pollution. Specifically, in the stretch from Novi Sad to Velika Morava that includes the cities of Belgrade and Pancevo, all the midstream samples were critically polluted. Downstream from Novi Sad, the left riverside sample was markedly more polluted (6,100 MPN/100 ml) than the midstream sample (980 MPN/100 ml), and downstream from Belgrade, even samples from both the left and right riversides showed strong E. coli pollution levels (11,600 and 12,500 MPN/100 ml, respectively). At the end of the middle Danube section, the large reservoirs for the two Iron Gate hydropower plants significantly reduced the E. coli pollution levels to low, most likely due to sedimentation and longer water residence times. At Vrbica (Serbia)/Simijan (Romania) (rkm 926), the lowest E. coli value of all the Danube samples was recorded for the midstream sample (2 MPN/100 ml), while both the left (2,400 MPN/100 ml) and right-side samples (980 MPN/100 ml) showed critical and moderate pollution levels. The tributaries and branches joining in this Danube section had contrasting levels of pollution such as the

critically polluted Drava and Velika Morava and the little polluted Rackeve-Soroksar arm and Sava (midstream sample). The left-side Sava sample was moderately polluted with E. coli, such as the Tisza and Vah tributaries.

In the lowest Danube section (Romania, Bulgaria, Ukraine; rkm 900 to 18), all but one midstream sample showed low to moderate E. coli pollution levels. Three right and two left riverside samples exhibited critical pollution levels. Amongst those, the sites downstream from Russenski Lom (rkm 488, right) and downstream from Arges (rkm 429, left), the two most polluted tributaries/branches of the whole river basin, were found. The other three sites were downstream from Zimnicea/Svistov (rkm 550, right) and at Chiciu/Silistra (rkm 378, both sides). The Timok tributary (860 MPN/100 ml) was much more polluted than the river Danube in terms of E. coli, leading to a higher downstream value at the respective riverside (rkm 834, left). The two tributaries Iskar and Jantra showed little E. coli pollution, while Siret and Prut were critically and moderately polluted.

In general, the faecal pollution pattern derived from intestinal enterococci resembled the E. coli pattern (Figure 1B). Both parameters were highly significantly inter-correlated in the Danube (rho = 0.685; p < 0.001) and its tributaries (rho = 0.596; p < 0.001) (Table 1). In the upper section, Kelheim (left; > 34,600 MPN/100 ml), Oberloiben (left; 480 MPN/100 ml) and downstream from Vienna (right; 390 MPN/100 ml) showed the highest enterococci concentrations. In the middle section, downstream from Budapest (right; 940 MPN/100 ml), Dunafoldvar (middle; 2,500 MPN/100 ml), the Drava tributary (460 MPN/100 ml) and the section between Novi Sad and Velika Morava (mostly left and right riverside samples; 110 -2,300 MPN/100 ml) showed critical pollution values. Additionally, at Iza/Szony (rkm 1,761), the left riverside was critically polluted. After a significant drop along the Iron Gates, the enterococci concentrations were elevated for the left and right riversides at Vrbica

(Serbia)/Simijan (Romania) (rkm 926). In the lowest section, Rusenski Lom (46,900 MPN/100 ml) and Arges (3,100 MPN/100 ml) showed the highest concentrations of all the investigated tributaries and branches, with both leading to significantly elevated downstream concentration levels at their respective Danube riversides. In addition, the sampling station downstream from Zimnicea/Svistov (rkm 550; left) exhibited strong faecal pollution (27,700 MPN/100 ml).

3.2 Cross-sectional patterns reveal the strong representative status of midstream samples but also specific local lateral discontinuities

Over the total length of the Danube, midstream samples representatively depicted the microbial faecal pollution levels at the respective river sites (Figure 2). The midstream E. coli concentrations were significantly correlated with the left-side (rho= 0.55; p < 0.01) and right-side concentrations (rho= 0.56; p < 0.01). A similar pattern was observed for enterococci, but the correlation coefficients were lower (rho = 0.35 for the left and 0.28 for right sides, p < 0.05) due to the higher detection limit of the microtitre plate method. However, specific left and right riverside samples showed significantly increased pollution levels in comparison to the midstream sample. This trend was mostly observed downstream of large municipalities with and without wastewater treatment or polluted tributaries (Figure 2). With the exception of downstream Jantra (enterococci), only the results from Dunafoldvar (60 km downstream Budapest) showed higher SFIB values for the midstream sample than for both riverside samples.

3.3 Microbial faecal pollution constitutes an independent principal component

To analyse the potential relationships to the ancillary environmental data, chemophysical data (oxygen, pH, conductivity, temperature, and discharge), nutrient data (dissolved organic carbon, total phosphorus, and total nitrogen), biological data (chlorophyll a, total and large

334 bacterial cell numbers) and chemical tracers related to anthropogenic pollution (acesulfame,

335 carbamazepine, and caffeine) were included in the principal component analysis together

336 with the microbiological data. Within the multidimensional component matrix, the microbial

337 faecal pollution parameters constituted an independent principal component apart from the

338 other variables (Table 2, Figure 3). Even the chemical tracers of anthropogenic origin

339 (caffeine, carbamazepine, and acesulfame) did not cluster with the microbiological

340 parameters. When examining the specific left and right riverside samples with surprising

341 critical, strong or excessive pollution (e.g. Kelheim, Oberloiben), the ancillary data could not

342 satisfactorily explain the high observed pollution levels.

343 3.4 Several midstream stations showed lower pollution levels in JDS 3 than in JDS 2

344 A comparison between the midstream data from 2013 (JDS 3) and 2007 (JDS 2) revealed a

345 similar pollution pattern (see (Kirschner et al. 2009)) and a highly significant correlation

346 between the years (rho = 0.738; P < 0.001, n = 48). However, the median concentrations

347 were slightly lower in 2013 in comparison to those of 2007 (with a median difference of 0.4

348 log for both E. coli and enterococci) and at several stations, markedly (> 1 log) lower E. coli

349 concentrations occurred in 2013 (Figure 4, left panel). This trend towards lower values in

350 2013 was also valid when the data were normalized to the discharge, i.e., when the total

351 load was calculated (Figure 4, right panel). Among the stations with markedly lower E. coli

352 levels are the stations downstream from large municipalities with wastewater discharge

353 such as Budapest, Wildungsmauer (that is 35 km downstream Vienna), downstream from

354 Linz, Bratislava, and downstream from Svistov. In addition, the two branches (Moson

355 Danube and Rackeve-Soroksar arm) also showed markedly (approx. 2 log) lower

356 concentrations in 2013 (data not shown). Enterococci were not included in the comparison

357 because there were too many missing data and data below or at the detection limit (15

358 MPN/100 ml).

359 3.5 Microbial faecal pollution in the Danube and its major tributaries is dominated by

360 human faecal pollution

361 The human-associated faecal marker BacHum was detected in 92.4% of all the investigated

362 Danube samples (n = 159) and in 100% of all the tributary samples (n = 22) (Table 3). The

363 concentrations ranged from 250 to 1.3 x 106 marker equivalents (ME)/100 ml in the Danube

364 and from 260 to 4.6 x 106 ME/100 ml in the tributaries. The highest concentration in the

365 Danube was observed downstream from Arges (left). The tributary with the highest

366 concentration was Rusenski Lom, followed by Arges (4.5 x 106 ME/100 ml). The BacHum

367 concentrations in the midstream samples were 0.1 to 0.2 log (median values) and 0.3 to 0.6

368 log (maximum values) lower compared to the left and right riverside samples, but at all three

369 positions, > 90% positive results were obtained. The second human-associated faecal marker

370 HF183II was present in > 80% of all the investigated Danube samples, while in the

371 tributaries, it was detected in 59% of the samples (Table 3). The concentrations were

372 approximately 0.5 log lower than the BacHum concentrations, with a maximum value of 4.0

373 x 105 ME/100 ml in the Danube (downstream Arges) and 1.5 x 106 ME/100 ml in the

374 tributaries (Rusenski Lom). In both the Danube and its tributaries, BacHum and HF183II were

375 highly significantly inter-correlated (Table 1) with rho = 0.894 (p < 0.001) and rho = 0.903 (p

376 < 0.001), respectively. In general, the results from flask A and flask B matched very well for

377 both BacHum (rho = 0.910, p < 0.001, n= 40) and HF183II (rho = 0.938, p < 0.001, n = 40),

378 showing low sample variability and corroborating the high reliability of the chosen assays

379 (Supplemental information, Figure S3).

In contrast to human-associated markers, the ruminant (BacR) and pig (Pig2Bac)-associated markers were of minor importance along the whole Danube River and its major tributaries. Only between 4 and 9% of the samples were positive for both markers, with equal percentages for the Danube and its tributaries (Table 3). In addition, the maximum concentrations were 2 to 3 logs lower than the human-associated markers. The highest Pig2Bac concentration (6.9 x 103 ME/100 ml) was found in the Sulina arm of the Danube Delta (left, rkm 26), and the highest BacR concentration (2.9 x 103 ME/100 ml) was found in the Jantra tributary (merging at rkm 537).

3.6 Human-associated faecal markers showed a highly significant correlation with the standard faecal indicators in the Danube

Both E. coli and enterococci were significantly correlated with the human-associated markers of faecal pollution (Table 1). The highest and most significant (p < 0.001) correlation was found for E. coli in the Danube with rho = 0.776 and 0.848 for BacHum and HF183II, respectively. In the tributaries, the correlation was significant at a level of p < 0.05. The enterococci showed correlation coefficients with human faecal markers between 0.494 and 0.685 (p < 0.001) in the Danube. In the tributaries, no significant correlation to human faecal markers (0.338, p > 0.05 and rho = 0.326, p > 0.1 for BacHum and HF183II, respectively) was observed. Figure 5 shows that for the whole data set, 67% and 59% of the E. coli variability was explained by HF183II and BacHum, respectively. For the Danube, the coefficients of variation were 65% and 60%, respectively, and for the tributaries, they were 78% and 64%. Similar values were obtained for left-side, midstream and right-side data (Figure S4, Supplemental Information). Due to the low quantity of positive signals, no correlations were calculated for ruminant and pig-associated markers.

3.7 Annual patterns in faecal pollution

At the three stations downstream of the large capitals (Vienna, Budapest, and Belgrade), the samples were analysed monthly over one year in 2014. Despite some temporal variability, the data confirmed the basic findings from the JDS 3 "snapshot" analysis. In Vienna, only low and moderate pollution levels were recorded (Figure 6A, B). The right-side samples always showed significantly higher faecal pollution than the midstream samples (T-test, t = 2.91; p < 0.01 for E. coli and t = 3.38; p < 0.01 for enterococci) due to the inflow of the WWTP effluent on this side. The left-side samples showed no significant differences relative to the midstream samples (p > 0.1). In Budapest, the pollution was mostly found to be at moderate levels, with a few critical samples (midstream and right-side only) and two samples with strong pollution levels (enterococci only) (Figure 6C, D). The right-side concentrations of both E. coli and the enterococci were significantly higher than the midstream concentrations (t = 1.90; p < 0.05 and t = 2.71; p < 0.05, respectively), and the left-side concentrations were significantly lower (t = 2.90; p < 0.01 and t = 3.63; p < 0.01, respectively). In Belgrade, the majority of all the samples showed critical pollution levels. At several time-points, the midstream samples showed moderate pollution levels, while strong pollution was observed in the right riverside. Significantly higher pollution levels occurred in both river sides than in the midstream segments (p < 0.05 for the left riverside and p < 0.001 for the right riverside). The human-associated faecal markers BacHum and HF183II showed similar spatial distribution patterns (Fig. 7), with the highest marker concentrations on the right riversides for all three stations. In general, the marker concentrations downstream from Vienna and Budapest were of comparable magnitude, while downstream from Belgrade, they were approximately one log higher (Table S3, Supplemental information).

All the faecal-associated parameters were highly significantly inter-correlated (rho = 0.510 to 0.963, p < 0.001, and n = 111; see Table S5, Supplemental Information). BacHum and

HF183II explained 39% and 42% of the E. coli variability, respectively (Fig. S5, Supplemental Information).

4 DISCUSSION

The chosen approach of using standard faecal pollution indicators together with advanced microbial source tracking markers over a large spatial and temporal scale allowed us to comprehensively characterize the microbial faecal pollution patterns along the whole Danube River. This study was built on two previous whole-river surveys, in which the longitudinal development of faecal pollution was only based on the SFIB concentrations that were determined for the midstream area of this large river (Kirschner et al. 2009, Kirschner et al. 2015). Here, we extended this sampling concept by including left and right riverside samples, different host-associated source tracking markers and a one-year analysis at three specific sites to assess the temporal variability of faecal pollution. To the best of our knowledge, no study of comparable size and sampling depth characterizes the microbial faecal pollution patterns of such a large river.

In general, the different methods used to determine the microbial faecal pollution (standard faecal indicators and human genetic faecal markers) were highly and significantly inter-correlated, and they yielded congruent pictures of the pollution patterns along the Danube. Nevertheless, the E. coli measurements performed according to ISO 9308-2:2012 ((International Organization for Standardization 2012); Colilert) had a more satisfactory resolution than the enterococci measurements performed in accordance with ISO 78991:1998 ((International Organization for Standardization 1998); microtitre plate method) because for the latter, many values at and below the detection limit were recorded.

However, in the principal component analysis, all the microbial faecal pollution parameters constituted a separate component and did not cluster with any of the environmental background data, including chemical tracers of anthropogenic origin.

The general microbial faecal pollution pattern during JDS 2013 was similar to the one observed in 2007, with the highest pollution levels in the middle section of the Danube and faecal pollution hot-spots in the Rusenski Lom (BUL) and Arges (ROM) tributaries (compare to (Kirschner et al. 2009)). However, in the stretch downstream from Budapest (rkm 1,632,1434), lower pollution levels were observed in 2013, including the Rackeve-Soroksar branch, which was a hot-spot of faecal pollution in 2007. Clearly, the implementation of the Central Budapest WWTP with a capacity of 1.6 Mio population equivalents did have a positive effect on the faecal pollution levels of the Danube in this stretch, with the E. coli concentrations being 0.4 to 1 log lower in 2013 than in 2007. Only in the midstream at Dunafoldvar (rkm 1,560) were similarly high values observed during both years. Because the wastewater of the Budapest WWTP is dumped in the middle of the Danube, we hypothesize that the plume is still visible even 60 km downstream due to the inefficient mixing of the water masses (Velimirov et al. 2011). Apart from Dunafoldvar, the highest contamination levels in the midstream of the Danube occurred in Serbia, in the stretch between Novi Sad and the Velika Morava tributary. Here, two large municipalities, Novi Sad and the Serbian capital Belgrade, are without state-of-the-art wastewater treatment.

When interpreting the differences between the years, the specific river regime must be considered. In 2013, baseflow conditions prevailed along the whole Danube (Supplemental Information, Table S1). For 2007, an incomplete set of discharge data is available. During that year, the discharge levels during the JDS were 33% to 48% higher in the upper stretch (rkm 2,204 - 1,942) and 88% to 128% higher in the lower stretch (rkm 1,071 - 130) of the

Danube. In the middle stretch (rkm 1,560 - 1,200), the discharge during JDS 2007 was slightly (15% to 28%) lower than it was during JDS 2013 (Supplemental Information, Table S1). From Figure 4 it became obvious that the lower concentrations observed in 2013 in comparison to 2007 are not attributed to a dilution effect but to a general reduction of faecal pollution in the Danube. This reduction could be caused by more effective wastewater treatment (e.g., the implementation of the central WWTP at Budapest). Additionally, reduced diffuse pollution may have reached the river, because baseflow conditions were experienced during JDS 2013 throughout the whole expedition, and only a single precipitation event occurred downstream from Vienna.

In comparison to 2007, the additional sampling from both riversides in 2013 allowed for the identification of further and partly surprising hot-spots of faecal pollution. In general, the midstream samples appropriately reflected the faecal pollution status of the Danube at the specific river sites. It has been hypothesized that the horizontal mixing of the water masses along the Danube is a slow process, which may occur over several hundreds of kilometres (Velimirov et al. 2011). Therefore, local pollution at the riverside may not be reflected in the midstream samples. This trend was found not only for several stations in the middle and the lower sections of the Danube (downstream from Ruse or from Arges) but also for two stations in the upper section of the Danube that exhibited critical or strong pollution levels at the left riverside, Kelheim (GER) and Oberloiben (AUT). Because both countries perform state-of-the-art wastewater treatments and both stations are far downstream of large municipalities, local peculiarities (vicinity to a pier) are assumed as the reasons for the high observed levels of microbial faecal pollution. A third station with critical pollution levels was downstream from Vienna, which is obviously caused by the inflow of the Vienna WWTP. Other stations with a striking difference in faecal pollution levels between the midstream

and the riversides were observed primarily in the lower section at rkm 926 (Vrbica-SRB/Simian-ROM), where the E. coli and enterococci levels were 2 to 3 orders of magnitude higher at the riversides than in the middle of the Danube, and at rkm 550 (downstream Svistov-BUL/Zimnicea-ROM), where excessive enterococci pollution levels were observed. Large municipalities obviously have a tremendous effect on the microbiological water quality of the Danube. After the large cities Vienna, Budapest, Novi Sad, and Belgrade, the microbial faecal pollution levels were generally increased. In addition, the tributaries that are used as the receiving waters of municipal wastewater such as the rivers Arges (Bucharest-ROM) and Rusenski Lom (Ruse-BUL) showed strong or even excessive pollution levels, indicating that human faecal pollution is a major source of microbial faecal pollution. With the microbial source tracking markers used in this study, we could now convincingly demonstrate that human faecal pollution is dominant throughout the whole length of the river Danube. Human-associated genetic faecal Bacteroidetes markers were present in 78% (HF183II) to 96% (BacHum) of all the samples, and BacHum (Kildare et al. 2007) exhibited higher prevalence and concentrations in comparison to HF183II (Green et al. 2014). A recent study investigating the human-associated marker concentrations in wastewater and faecal samples has shown that HF183II is slightly less abundant and more source-specific than BacHum in the faecal sources (unpublished data). The human-associated markers were highly significantly inter-correlated and correlated with the E. coli and enterococci concentrations as well. For the Danube and its tributaries, the E. coli concentrations explained 59% to 78%, respectively, of their variability, and HF183II showed slightly higher coefficients of determination. By contrast, ruminant (BacR; (Reischer et al. 2007)) and pig-associated (Pig2Bac; (Mieszkin et al. 2010)) markers were only rarely detected (< 10% positive samples), and the median concentrations were one to three orders of magnitude

lower than the human-associated marker concentrations, despite the fact that animal farming and pastureland occur along the whole river. It must be considered that baseflow conditions prevailed during the whole river expedition, with only a single precipitation event downstream from Vienna. Low surface runoff may have led to reduced, diffuse pollution entering the river, leading to an underestimation of the non-point animal faecal pollution sources. Another potential source of animal faecal pollution that was not considered in this study may be faeces from wild and livestock birds. No general marker for all bird species is available so far, but markers for gulls (Lee et al. 2013), ducks (Kobayashi et al. 2013) and poultry (Ryu et al. 2014) exist. Moreover, horses and fish may also partly contribute to the faecal pollution in the Danube River, but no reliable genetic faecal markers associated with those two groups are known. However, because the ruminant and pig markers that also target wildlife such as deer and boar were relatively unimportant, we consider the role of the other animal groups to be of similarly little importance to the general faecal pollution of the Danube River.

During the annual cycle that covered a wide range of hydrological conditions at all three selected sites, the investigations revealed considerable seasonal variability in all the measured microbial faecal pollution parameters, but they corroborated the conclusions drawn from the Joint Danube Survey. In comparison between the three capitals, Vienna showed the lowest (little to moderate) pollution levels, followed by Budapest, while Belgrade (without wastewater treatment) exhibited critical to strong faecal pollution. In nearly all cases, the midstream values were lower than the values observed at the left and right riversides for all the measured parameters, including the genetic faecal source tracking markers. It is obvious that human faecal pollution is dominating at all three stations downstream from the capitals with approx. 1.5 to 2 Mio inhabitants. However, the

prevalence and abundance of ruminant and pig markers in the midstream, where faecal pollution should reflect the mixed picture of all upstream contamination sources, were not different from the ones observed at both riversides and not higher than those observed during the JDS. This finding clearly supports the results from the spatial analysis in which human faecal pollution is the dominant source along the whole Danube.

5 CONCLUSIONS

In addition to traditional cultivation, the future monitoring of microbial faecal pollution will increasingly rely on the implementation of advanced culture-independent techniques that simultaneously allow for the identification of the sources of contaminations. For the first time, standard indicators of faecal pollution were applied in combination with several host-associated genetic faecal markers for the whole length of a large international river at a high spatial resolution. In addition to identifying the hot-spots and general patterns of faecal pollution, we could show that the microbial faecal pollution constituted an independent component within the environmental matrix. We could also clearly demonstrate that human faecal pollution is the primary source of microbial faecal contamination along the whole Danube River. It turned out that for whole river monitoring purposes, midstream samples are largely representative of the faecal pollution levels at a specific river site and its longitudinal development along a large river. However, at a few somewhat unexpected sites, high pollution levels occurred along the lateral zones of the river, although the midstream zone had good microbial water quality. Therefore, if site-specific considerations have to be performed (e.g. a certain location shall be used as a source to provide water for irrigation) a specific assessment of the microbiological water quality at the site of interest is recommended, including a detailed temporal and spatial (cross-sectional) investigation, even

572 at stretches where good microbiological water quality is obvious from the midstream

573 monitoring data. We are convinced that this study provides a solid scientific basis for guiding

574 the authorities and decision makers towards optimally designed monitoring schemes for

575 state-of-the-art microbial faecal pollution management in large rivers.

577 6 ACKNOWLEDGEMENTS AND FUNDING INFORMATION

578 The authors thank the organizers of the Joint Danube Survey 2013 and the team of people

579 (scientists, engineers, and ship crew) who participated in this survey. The authors gratefully

580 acknowledge the ancillary set of environmental data that was provided by the International

581 Commission for the Protection of the Danube River (ICPDR, Vienna, Austria) for statistical

582 analysis. The survey was organized by the ICPDR and logistically supported by the Austrian

583 Federal Ministry of Agriculture, Forestry, Environment and Water Management (BMLFUW).

584 The majority of the funding for the microbiological analyses came from the Austrian Science

585 Fund FWF (project P25817-B22 and P23900-B22), additional financial support came from the

586 ICPDR and BMLFUW and the Vienna Doctoral Programme for Water Resources Systems

587 (FWF, DKplus W1219-N22).

588 REFERENCES

589 Anderson, A., Pietsch, K., Zucker, R., Mayr, A., Muller-Hohe, E., Messelhausser, U., Sing, A., Busch, U.

590 and Huber, I. (2011) Validation of a Duplex Real-Time PCR for the Detection of Salmonella spp. in

591 Different Food Products. Food Analytical Methods 4(3), 259-267.

592 Caldwell, J., Payment, P. and Villemur, R. (2011) Microbial Source Tracking: Methods, Applications

593 and Case Studies. Hagedorn, C., Blanch, A.R. and Harwood, V.J. (eds), pp. 229-250, Springer Science +

594 Business Media LLC New York.

595 Das, T.K. (2001) In: "Clean Technologies and Environmental Policy". , Springer Berlin/Heidelberg. 3

596 (2): 69-80.

597 Derx, J., Schijven, J., Sommer, R., Zoufal-Hruza, C.M., van Driezum, I.H., Reischer, G., Ixenmaier, S.,

598 Kirschner, A., Frick, C., de Roda Husman, A.M., Farnleitner, A.H. and Blaschke, A.P. (2016)

599 QMRAcatch: Human-Associated Fecal Pollution and Infection Risk Modeling for a River/Floodplain

600 Environment. J Environ Qual 45(4), 1205-1214.

601 European Council (1991) Council Directive 91/271/EEC of 21 May 1991 concerning urban waste-

602 water treatment. Official Journal L135, 40-52.

603 European Council (1998) Council Directive 98/83/EC of 3 November 1998 on the quality of water

604 intended for human consumption. Official Journal L330, 32-54.

605 European Parliament & Council (2000) Directive 2000/60/EC of the European Parliament and the

606 Council of 23 October 2000 establishing a framework for Community action in the field of water

607 policy. Official Journal L327, 1 - 72.

608 European Parliament & Council (2006) Directive 2006/7/EC of the European Parliament and of the

609 Council of 15 February 2006 concerning the management of bathing water quality and repealing

610 Directive 76/160/EEC. Official Journal L 064, 37 - 51.

611 Farnleitner, A.H., Reischer, G.H., Stadler, H., Kollanur, D., Sommer, R., Zerobin, W., Barrella, K.M.,

612 Truesdale, J.A., Casarez, E.A. and DiGiovanni, G.D. (2011) Microbial Source Tracking: Methods,

613 Applications and Case Studies. Hagedorn, C., Blanch, A.R. and Harwood, C.R. (eds), pp. 399-431,

614 Springer Science+Business Media LLC New York.

615 Green, H.C., Haugland, R.A., Varma, M., Millen, H.T., Borchardt, M.A., Field, K.G., Walters, W.A.,

616 Knight, R., Sivaganesan, M., Kelty, C.A. and Shanks, O.C. (2014) Improved HF183 quantitative real-

617 time PCR assay for characterization of human fecal pollution in ambient surface water samples. Appl

618 Environ Microbiol 80(10), 3086-3094.

619 Hagedorn, C., Blanch, A.R. and Harwood, V.J. (2011) Microbial Source Tracking: Methods,

620 Applications and Case Studies, Springer Science + Business Media LLC New York.

621 Hagedorn, C. and Weisberg, S.B. (2011) Microbial Source Tracking: Methods, Applications and Case

622 Studies. Hagedorn, C., Blanch, A.R. and Harwood, V.J. (eds), pp. 189-206, Springer Science + Business

623 Media LLC New York.

624 International Organization for Standardization (1998) ISO 7899-1 Water quality - Detection and

625 enumeration of intestinal enterococci in surface and waste water - Part 1: Miniaturized method

626 (Most Probable Number) for surface and waste water, p. 21, International Organization for

627 Standardization, Geneve, Switzerland.

628 International Organization for Standardization (2006) ISO19458: Water quality -- Sampling for

629 microbiological analysis, p. 18pp, International Organization for Standardization, Geneva,

630 Switzerland.

International Organization for Standardization (2012) ISO 9308-2:2012 Water quality -- Enumeration of Escherichia coli and coliform bacteria -- Part 2: Most probable number method., p. 45, International Orgnization for Standardization, Geneva. Switzerland.

Kavka, G.G., Kasimir, G.D. and Farnleitner, A.H. (2006) Microbiological water quality of the River Danube (km 2581 - km 15): Longitudinal variation of pollution as determined by standard parameters, pp. 415-421, Austrian Committee Danube Research / IAD, Vienna, Austria.

Kavka, G.G. and Poetsch, E. (2002) Joint Danube Survey 1, technical report. . Literathy, P., Koller-Kreimel, V. and Liska, I. (eds), pp. 86-95, ICPDR Vienna, Austria.

Kildare, B.J., Leutenegger, C.M., McSwain, B.S., Bambic, D.G., Rajal, V.B. and Wuertz, S. (2007) 16S rRNA-based assays for quantitative detection of universal, human-, cow-, and dog-specific fecal Bacteroidales: a Bayesian approach. Water Res 41(16), 3701-3715.

Kirschner, A.K., Kavka, G.G., Velimirov, B., Mach, R.L., Sommer, R. and Farnleitner, A.H. (2009) Microbiological water quality along the Danube River: integrating data from two whole-river surveys and a transnational monitoring network. Water Res 43(15), 3673-3684.

Kirschner, A.K.T., Kavka, G.G., Reischer, G.H., Sommer, R., Blaschke, A.P., Stevenson, M., Vierheilig, J., Mach, R.L. and Farnleitner, A.H. (2015) Handbook of Environmental Chemistry, special volume "The Danube River Basin";. J, L.I.S. (ed), pp. 439-468, Springer Verlag; Berlin.

Kirschner, A.K.T., Kavka, G.G., Velimirov, B., Reischer, G.H., Mach, R.L. and Farnleitner, A.H. (2008) Joint Danube Survey 2, Final scientific report. Liska, I., Wagner, F., Slobodnik, J. (ed), pp. 86-95, ICPDR Vienna, Austria.

Kobayashi, A., Sano, D., Hatori, J., Ishii, S. and Okabe, S. (2013) Chicken- and duck-associated Bacteroides-Prevotella genetic markers for detecting fecal contamination in environmental water. . Appl Microbiol Biotechnol 97, 7427-7437.

Layton, A., McKay, L., Williams, D., Garrett, V., Gentry, R. and Sayler, G. (2006) Development of Bacteroides 16S rRNA gene TaqMan-based real-time PCR assays for estimation of total, human, and bovine fecal pollution in water. Appl Environ Microbiol 72(6), 4214-4224.

Lee, C., Marion, J.W. and Lee, J. (2013) Development and application of a quantitative PCR assay targeting Catellicoccus marimammalium for assessing gull-associated fecal contamination at Lake Erie beaches. Science of the Total Environment 454, 1-8.

Liu, R., Cheng, K.H., Wong, K., Cheng, S.C. and Lau, S.C. (2015) Differential utility of the Bacteroidales DNA and RNA markers in the tiered approach for microbial source tracking in subtropical seawater. Appl Microbiol Biotechnol 99(13), 5669-5681.

Mayer, R.E., Bofill-Mas, S., Egle, L., Reischer, G.H., Schade, M., Fernandez-Cassi, X., Fuchs, W., Mach, R.L., Lindner, G., Kirschner, A., Gaisbauer, M., Piringer, H., Blaschke, A.P., Girones, R., Zessner, M., Sommer, R. and Farnleitner, A.H. (2016) Occurrence of human-associated Bacteroidetes genetic source tracking markers in raw and treated wastewater of municipal and domestic origin and comparison to standard and alternative indicators of faecal pollution. Water Research 90, 265-276.

McQuaig, S.M. and Noble, R.T. (2011) Microbial Source Tracking: Methods, Applications and Case Studies. Hagedorn, C., Blanch, A.R. and Harwood, V.J. (eds), pp. 113-135, Springer Science + Business Media LLC New York.

Mieszkin, S., Yala, J.F., Joubrel, R. and Gourmelon, M. (2010) Phylogenetic analysis of Bacteroidales 16S rRNA gene sequences from human and animal effluents and assessment of ruminant faecal pollution by real-time PCR. J Appl Microbiol 108(3), 974-984.

Reischer, G.H., Ebdon, J.E., Bauer, J.M., Schuster, N., Ahmed, W., Astrom, J., Blanch, A.R., Bloschl, G., Byamukama, D., Coakley, T., Ferguson, C., Goshu, G., Ko, G., de Roda Husman, A.M., Mushi, D., Poma, R., Pradhan, B., Rajal, V., Schade, M.A., Sommer, R., Taylor, H., Toth, E.M., Vrajmasu, V., Wuertz, S.,

677 Mach, R.L. and Farnleitner, A.H. (2013) Performance characteristics of qPCR assays targeting human-

678 and ruminant-associated bacteroidetes for microbial source tracking across sixteen countries on six

679 continents. Environ Sci Technol 47(15), 8548-8556.

680 Reischer, G.H., Haider, J.M., Sommer, R., Stadler, H., Keiblinger, K.M., Hornek, R., Zerobin, W., Mach,

681 R.L. and Farnleitner, A.H. (2008) Quantitative microbial faecal source tracking with sampling guided

682 by hydrological catchment dynamics. Environ Microbiol 10(10), 2598-2608.

683 Reischer, G.H., Kasper, D.C., Steinborn, R., Farnleitner, A.H. and Mach, R.L. (2007) A quantitative real-

684 time PCR assay for the highly sensitive and specific detection of human faecal influence in spring

685 water from a large alpine catchment area. Lett Appl Microbiol 44(4), 351-356.

686 Ryu, H., Elk, M., Khan, I.U., Harwood, V.J., Molina, M., Edge, T.A. and Domingo, J.S. (2014)

687 Comparison of two poultry litter qPCR assays targeting the 16S rRNA gene of Brevibacterium sp.

688 Water Res 48, 613-621.

689 Schmedtje, U., Bachmann, J., Behrendt, H., Birk, S., Biza, P., D'Eugenio, J., V. Gils, J., Grath, J.,

690 Hamchevici, C., Hansen, W., Interwies, E., Kampa, E., Lindinger, H., Liska, I., Popescu, L., Popovici, M.,

691 Pottgiesser, T., Sigmund, G., Sommerhäuser, M., Speck, S., Stubauer, I., Vogel, B., Weller, P.,

692 Winkelmann-Oei, G. and Zinke, A. (2005) Danube Basin Analysis (WFD Roof Report 2004). The

693 Danube River Basin District - River basin characteristics, impacts of human activities and economic

694 analysis required under Article 5, Annex II and Annex III, and inventory of protected areas

695 required under Article 6, Annex IV of the EU Water Framework Directive (2000/60/EC) - Part A

696 - Basin-wide overview., ICPDR, Vienna, Austria; 191 pp.

697 Servais, P., Garcia-Armisen, T., George, I. and Billen, G. (2007) Fecal bacteria in the rivers of the Seine

698 drainage network (France): sources, fate and modelling. Sci Total Environ 375(1-3), 152-167.

699 Stapleton, C.M., Kay, D., Wyer, M.D., Davies, C., Watkins, J., Kay, C., McDonald, A.T., Porter, J. and

700 Gawler, A. (2009) Evaluating the operational utility of a Bacteroidales quantitative PCR-based MST

701 approach in determining the source of faecal indicator organisms at a UK bathing water. Water Res

702 43(19), 4888-4899.

703 Velimirov, B., Milosevic, N., Kavka, G.G., Farnleitner, A.H. and Kirschner, A.K. (2011) Development of

704 the bacterial compartment along the Danube River: a continuum despite local influences. Microb

705 Ecol 61(4), 955-967.

706 World Health Organization (2003) Guidelines for safe recreational water environments. Volume 1:

707 Coastal and fresh waters, World health Organization, Geneva, Switzerland.

708 Wuertz, S., Wang, D., Reischer, G. and Farnleitner, A.H. (2011) Microbial Source Tracking: Methods,

709 Applications and Case Studies. Hagedorn, C., Blanch, A.R. and Harwood, V.J. (eds), pp. 61-112,

710 Springer Science + Business Media LLC New York.

FIGURE LEGENDS

Figure 1: Longitudinal development of E. coli (A) and Enterococci concentrations (B) along the increasing discharge of the Danube in the midstream, at the left (upper line) and the right river-side (lower line) and in the tributaries sampled during JDS 2013. Not-sampled tributaries are also indicated on the x-axes. Symbol size varies according to log-transformed E. coli and Enterococci concentrations [(MPN+1/100 ml)]. Colours depict the faecal pollution levels according to Table S2 (Suppl. Information): blue - little, green - moderately, yellow -critically, orange - strongly and red - excessively polluted. Danube stations with remarkable pollution levels are named. Total discharge of the Danube (grey-shaded area) can be calculated by adding left and right-side contributions. Grey-shaded triangles indicate discharge contributions of the main tributaries. RMD-canal: Rhine-Main-Danube canal; RSD-arm: Rackeve-Soroksar Danube arm. The position of Iron Gate I and II are shown in red.

Figure 2: Scatterplots of E.coli and Enterococci concentrations for comparing the results from the midstream with results from both river-sides. The stations where the concentrations were above 100 MPN/100 ml (log10 = 2) and differed by more than 1 log are named. The middle dashed lines indicate the 1:1 curve, the outer dashed lines indicate 1 log difference.

Figure 3: Two-dimensional component plot of the principal component analysis. The microbiological faecal parameters constitute an independent component (surrounded by square). ENT: Enterococci, ECOLI: E. coli; BacHum & HF183: human genetic faecal markers; BP: bacterial production; OXY: oxygen; CHLA: chlorophyll a; BNLRG: large bacterial cell numbers, BNTOT: total bacterial cell numbers; PTOT: total phosphorus; NTOT: total nitrogen; TEMP: water temperature; DOC: dissolved organic carbon; COND: electrical conductivity; DISCH: discharge: CARB: carbamazepine; CAF: caffeine; ACE: acesulfame; all data was

738 retrieved from the ICDPR Danubis database (https://www.icpdr.org/wq-db/) with kind

739 permission of the ICPDR

740 Figure 4: Scatterplots of the E.coli concentrations (left) and the load (right) for comparing

741 results from 2013 and 2007 (midstream only). Those stations where the concentrations

742 differed by more than 1 log are named (except Dunafoldvar). The middle dashed lines

743 indicate the 1:1 curve, the outer dashed lines indicate 1 log difference.

744 Figure 5: Regression analysis between the E. coli concentrations and the concentrations of

745 the human-associated genetic faecal markers BacHum and HF183II for all data, for the

746 Danube and for the tributary data set.

747 Figure 6: E. coli and Enterococci concentrations during the annual cycle (January 2014748 January 2015) at the three selected Danube stations downstream from Vienna, downstream

749 from Budapest and downstream from Belgrade. Colours depict the faecal pollution levels

750 according to Table S2 (Suppl. Information). Large circles: midstream, small circles: left-side

751 and small diamonds: right-side samples.

752 Figure 7: Box-whisker plots of the human-associated genetic faecal marker concentrations

753 (BacHum and HF183II) during the annual cycle (Jan 2014-Jan 2015) at the three selected

754 Danube stations downstream from Vienna, downstream from Budapest and downstream

755 from Belgrade. Significant differences between the left, middle and right side samples are

756 marked with asterisks: * p < 0.05; ** p < 0.01, *** p < 0.001.

1 Table 1: Spearman rank correlation coefficients and p-values for correlations between the

2 faecal pollution parameters in the Danube (upper panel) and the tributaries (lower panel).

Danube (n = 160) ENT BacHum HF183II

rho E. coli p-value 0.685 0.000 0.776 0.000 0.848 0.000

ENT rho p-value 0.492 0.000 0.561 0.000

BacHum rho p-value 0.894 0.000

Tributaries (n = 26) ENT BacHum HF183II

rho E. coli p-value 0.596 0.001 0.465 0.017 0.483 0.020

ENT rho p-value 0.338 0.091 0.326 0.129

BacHum rho p-value 0.903 0.000

1 Table 2: Component matrix of the principal component analysis.

Variables Component

1 2 3 4 5

HF183 [ME/100 ml] 0.91 -0.14 0.11 < 0.1 < 0.1

BacHum [ME/100 ml] 0.91 -0.16 0.13 -0.11 < 0.1

E. coli [MPN/100 ml] 0.84 -0.23 -0.23 -0.14 0.11

Enterococci [MPN/100 ml] 0.65 0.02 -0.45 < 0.1 < 0.1

Bacterial production [^gC/L/h] 0.54 0.30 < 0.1 < 0.1 0.42

Oxyen [mg/L] -0.22 0.85 < 0.1 -0.11 -0.20

pH -0.26 0.83 < 0.1 < 0.1 -0.19

Chlorophyll a [^g/L] 0.03 0.81 0.16 0.13 0.17

Total nitrogen [mg/L] -0.08 0.09 0.82 -0.18 0.25

Discharge [m3/sec] 0.11 -0.08 -0.74 < 0.1 < 0.1

Total phosphorus [^g/L] 0.14 0.14 0.68 0.24 -0.18

Large cell numbers [cells/ml] 0.10 0.25 < 0.1 0.81 < 0.1

Total cell numbers [cells/ml] -0.09 -0.07 -0.39 0.79 < 0.1

Acesulfame [^g/L] -0.16 -0.27 0.27 0.56 0.26

Conductivity [^S/cm] -0.04 -0.08 0.36 0.56 < 0.1

Carbamacepine [^g/L] -0.06 -0.17 0.20 0.17 0.78

Temperature [°C] 0.16 0.23 < 0.1 < 0.1 0.58

Caffeine [^g/L] 0.09 -0.38 -0.41 < 0.1 0.57

Dissolved organic carbon (mg/L] -0.02 0.08 < 0.1 0.15 < 0.1

% explained variance 22.4 12.7 11.8 10.6 7.0

1 Table 3: Prevalence and concentrations of the host associated genetic faecal markers

2 BacHum, HF183II, BacR and Pig2Bac [ME/100 ml] in the Danube and the tributaries during

3 Joint Danube Survey 2013.

Sample % Median Min Max 5% 95%

number positive percentile percentile

Danube BacHum 159 92.4 6.6x103 2.5x102 1.3x106 5.8x102 ) 1.5x105

total HF183II 164 82.3 1.3x103 5.3x101 4.0x105 1.3x102 4.3x104

BacR 164 7.3 3.3x102 5.3x101 2.8x103 5.3x101 1.6x101

Pig2Bac 164 8.5 9.6x102 1.8x102 6.9x103 2.1x102 6.9x103

Danube BacHum 55 90.9 5.8x103 5.6x102 1.3x106 7.9x102 2.0x105

left HF183II 55 89.1 1.3x103 7.5x101 4.0x105 1 1.1x102 3.7x104

BacR 56 8.9 2.7x102 7.5x101 2.8x103 1.0x102 2.4x103

Pig2Bac 56 8.9 1.9x103 2.3x102 6.9x103 3.7x102 6.0x103

Danube BacHum 54 90.7 4.4x103 2.5x102 3.1x105 4.3x102 7.0x104

middle HF183II 54 79.6 1.1x103 5.3x101 9.1x104 1.8x102 1.8x104

BacR 54 3.7 4.2x102 3.7x102 4.7x102 3.7x102 4.7x102

Pig2Bac 54 9.3 8.8x102 6.1x102 1.4x103 6.3x102 1.3x102

Danube BacHum 50 96.0 7.1x103 5.8x102 5.8x105 8.6x102 1.4x105

right HF183II 55 78.2 1.5x103 7.5x101 9.9x104 2.1x102 5.4x104

BacR 54 9.3 3.2x102 5.3x101 6.2x102 5.3x101 5.7x102

Pig2Bac 54 7.4 7.9x102 1.8x102 6.8x103 2.1x102 6.0x103

Tributaries BacHum 22 100 2.4x103 2.6x102 4.6x106 2.9x102 7.2x105

HF183II 22 59.1 2.8x103 5.8x101 1.5x106 1.2x102 6.5x105

BacR 25 8.0 1.6x103 3.4x102 2.9x103 4.7x102 2.8x103

Pig2Bac 25 4.0 9.5x102 -- -- -- --

.Я1 й

щ «Р Р

(Я £

TT о л.

о сэ о сч

■ ■.D

1 1 / \

— = JC

à i H Í !

Ш I 1 i

ZI ÍL

О) 2 О

С y Kelheim s y / ifc / » s

Vrbica О Ob ds. J о Brbiben ^rges о / / 9 » V r 1 и S r S 0 У

s, Svistov Szcny { OcofT y • /Dur V о mföldvar

4 f /' С / h S m / fé О У S s- / m У ÁJ / \ 1 ■f y 1-

0 12 3 4 5

Е. со//-m id [log((MPN+1) /100 ml)]

"о о о о о

Q Kelheim Q ds. Svisto ds. Arges V H* y / y s У H / fini /

Szony о У / * R- y » y » /

Oberloïben ^Vrbica о о / 4) y g/ / ' / J* s r У * s r

eis. Oit с/о * p- О D una fold1 far

/ m R- y CT / 0

Enterococci-mid [log((MPN+1) /100 ml)]

ds. Svlstiïv

фф ds. Vienna,'

ds. Ruse Ф

Vrblca

Dunafolduar

2 О) J О

1 "i О

Е. coli-mid [log((MPN+1) /100 ml)]

Chiciu us- ^

О О 7

ds. Svist.. <> ds.Ruse<> ds.V

ф Vrbica

Mora va

Duriaföldvar

Jartra

0 12 3 4

Enterococci-mid [log((MPN+1) /100 ml)]

1 "и 1 о

о о о

0 <й - -I—'

oxv CP PH O CHLA

NTOT Oc BNLRG TEMP BP O

O PTOT DOC r n "N

ENTO" OS o ACE 0 ONCt 0 DISCH CARB O CAF EMT V BecHum HF1S30 o ECO LI J

i i i i i i i

Component 1

- ds. Bu / y J Hapest Du o' ✓ + ■ nafoldtfar o/' ■i

Sil f Me ds. Svisl Wild uii| iny Giurgen o ds. Linz / dve / ¡smauer Oyö ^ ° / O / / J / V ■ y> y * / y

/ /C > i^O } .«p / ^ o / / ✓ r'

/ / * \ I I 1 I 1 00 r i> f r r f t / III 1 1 1 1 III 1 1 1 1 Iii Iii i II 1 1 1 1 1

0 1 2 3 4 5

£ coli 2013 [log((MPN+1) /100 ml)]

ds GTurgenf Dunaföldv^r Budapest OD

Wildu Oc ds Li Szony Q Js Svis ngsmauery Q Q < o O D O X : S /

8/ Medve ^ ,'0 o / m / ^ ...... {JJ o o p'o f s y <t f r / s s f f j II 1 1 1 1 1 * *> * * V r r — J> / III 1 1 1 1 1 1 1 1 1 1 III

8 9 10 11 12

E. coli 2013 [log((MPN+1) /sec)l

y = 0.77x + 1,91 Pr = 0.59

° û /

- • o

8° ""o0

v = Q.92x + 0.79 R= = 0.67

OBacHum OHF1S3II

2 3 4 5

F. coli flog ((MPN+1)/100 mljl

Danube

V = 0.7Sx+ 1.95 Ra = 0.60

P ° 8° °

y = 0.92x + 0.S2 R2 = 0.55

OBacHum OHF1S3II

E. coli [log((MPN+l)/100 ml)

Tributaries

y = 0.SSx+ 1.31 R~ = 0.64

û Os Os

O -O /

0<3pOQ

v = 1.05x + 0.13 R2 = 0.78

OBacHum OHF1S3II

E. coli [log ((MPN+1)/100 ml)]

_QJ -j

downstream Vienna

downstream Vienna

* * ö 0 ° to) ^

w w w ^ o

JFMAMJ J ASONDJ

_QJ -j

downstream Budapest

downstream Belgrade

U U O U O

downstream Belgrade

h u n CG

Vienna

Budapest

Belgrade

■J ^

left middle right left middle right left middle right

Vienna

Budapest

left middle right

left middle right

left middle right

Highlights

• Faecal pollution patterns along the whole Danube river are shown

• Combination of standard parameters and microbial source tracking markers used

• Dominance of anthropogenic pollution along the whole Danube River

• Midstream samples generally depict microbial faecal pollution levels

• High pollution levels of the lateral zones at specific, partly unexpected sites