Assessing the impact of two conventional wastewater treatment plants on small streams with effect-based methods

Characterization of the wastewater treatment plants Mörfelden-Walldorf and Bickenbach

The medium-sized WWTPs Mörfelden-Walldorf (MW) and Bickenbach (B) are located in the southwestern region of Hessen in Germany and have approximately 48,000 and 32,000 population equivalents (PE), respectively. The WW is currently processed with conventional treatment (mechanical, biological treatment, nitrification, denitrification, phosphate precipitation) in both WWTPs. The receiving watercourses for the treated WW are the Geräthsbach (waterbody code 239818) for the WWTP MW, and the Landbach (waterbody code 239628) for the WWTP B. They are classified as small streams in riverine floodplains (type 19) and exhibit poor chemical status, strongly or completely altered morphological structure, as well as moderate and poor ecological status, respectively (Hessisches Landesamt für Naturschutz, Umwelt und Geologie (HLNUC), 2022). Additionally, the mean flow (MQ) in these streams is 238 L/s (Geräthsbach) and 125 L/s (Landbach) (Hessisches Landesamt für Naturschutz, Umwelt und Geologie (HLNUC), 2022). Ten km upstream WWTP MW, there is another WWTP in Langen (Fig. 1).

Collection and storage of samples

For the study, four sampling sites were chosen for each of the two WWTPs (Table 1).

In two sampling campaigns in April/May 2022 (spring) and October/November 2022 (fall), 2.1 L surface water samples were taken upstream (US) and downstream (DS) the WWTPs in the receiving rivers, and WW samples were taken after the mechanical (MT) and biological treatment (BT) at each WWTP at the beginning and at the end of the four weeks active biomonitoring period. Accordingly, two water samples per site were obtained and analyzed per season. During the sampling campaign in fall 2022, the outflow of the WWTP in Langen, which is located 10 km upstream of WWTP MW at the Geräthsbach, was also considered (sample code BT-L), because the first collection campaign had shown that the Geräthsbach already showed high activities in the EBM upstream the WWTP MW. At each sampling site, physicochemical parameters including temperature, pH, dissolved oxygen, and conductivity were measured in the field with a portable multi-meter (HQ40d, Hach, Germany). The water samples were collected in glass vessels precleaned with deionized water, acetone, and ethanol and heated to 200 °C before use. The samples were transported and stored at 7 °C in darkness. Phosphate, ammonium, nitrite, and nitrate concentrations were determined using spectroquant test kits (Merck, Darmstadt, Germany) immediately after bringing the samples to the laboratory.

Pretreatment of samples

Water samples were filtered through glass microfiber filters (VWR International GmbH, No. 692, European Cat. No. 516-0885, 90 mm, particle retention: 1.5 µm, Darmstadt, Germany). Then, 2 L of each water sample was solid-phase extracted (SPE) with OASIS HLB cartridges (6cc, 200 mg; Waters, Milford, MA, USA) according to Giebner et al. (2018) and transferred to a final volume of 400 µL DMSO so that the final extracts were 5,000-fold concentrated compared to the aqueous samples. Extracts were stored in glass vials at −25 °C until further processing. To check a potential contamination of the samples during the sample preparation, 2 L of ultrapure water was extracted in parallel (SPE blank) (Giebner et al., 2018).

Analysis of samples with effect-based methods

Recombinant yeast reporter gene assays.

The Yeast Estrogen Screen (YES), Yeast Androgen Screen (YAS), Yeast Anti-Estrogen Screen (YAES), Yeast Anti-Androgen Screen (YAAS), and Yeast Dioxin Screen (YDS) were used to assess the receptor-mediated agonistic and antagonistic endocrine activity of the extracted water samples as well as the activation of the aryl hydrocarbon receptor assay according to Brettschneider et al. (2019a); Brettschneider et al. (2019b), Giebner et al. (2018) and Schneider et al. (2020). These assays use genetically modified strains of Saccharomyces cerevisiae, which contain the human estrogen receptor α, androgen and aryl hydrocarbon receptor, respectively. The YAES and YAAS were performed with native samples (filtrated surface water or WW samples without SPE processing) because SPE enrichment of water samples from surface waters and WWTPs leads to a strong loss of anti-estrogenic and anti-androgenic activities (Brettschneider et al., 2019a; Giebner et al., 2018; Schneider et al., 2020). The YES, YAS and YDS were performed in three independent experiments with every sample, the YAES and YAAS in two independent experiments, with eight technical replicates per sample. The measured activities are expressed as equivalent concentrations (EQ) for the positive substances 17β-estradiol (YES), testosterone (YAS), 4-hydroxytamoxifen (YAES), flutamide (YAAS) and β-naphthoflavone (YDS) and were corrected regarding dilution and enrichment so that equivalent concentrations relate back to the native water sample.

Ames fluctuation test.

The Ames fluctuation assay (ISO, 2012) was performed to assess the mutagenic potential of the SPE samples, using Salmonella typhimurium strains YG1041 and YG1042 (Hagiwara et al., 1993) with and without metabolic activation by rodent liver enzymes (S9-mix; Envigo CRS, Roßdorf, Germany).

The test is based on the inability of genetically modified S. typhimurium strains to synthesize the essential amino acid histidine and, therefore, not being able to survive in a histidine-free medium (Maron & Ames, 1983). The bacteria are exposed to the SPE extracts in a 16.7-fold final sample concentration (0.2% v/v solvent) and incubated for 72 h at 37 °C in a histidine-free medium. After this time, a color change of the medium indicates the survival of the bacteria by regaining the ability to synthesize histidine because of a back mutation.

YG1041 and YG1042 strains without the S9 mix were tested with 2-nitrofluorene as a positive control. Conversely, these strains with S9-mix were tested with 2-amino anthracene as a positive control. The mutagenic potential of the samples was assessed via the percentage of reverted wells, subtracted by the mutants of negative historical controls. Samples exceeding the threshold of 20.8% revertant wells were considered as mutagenic. Every sample was tested with both strains, with and without the S9-mix. If an effect was observed, the test was consequently conducted two times.

In vivo assays

The in vivo assays were applied as an active in-situ monitoring at the upstream and downstream samplings sites of both WWTPs as described by Brettschneider et al. (2019a) and Harth et al. (2018).

Active monitoring with Gammarus fossarum.

The gammarids for the active monitoring were collected at the upper reach of the river Urselbach (Hessen, Germany, 50°13′30.1″N 8°31′06.1″E) 5 days before the start of the spring and fall campaigns. This site was selected because the upper reach of the Urselbach is entirely natural and free of anthropogenic inputs. Also, a genetic characterization of the G. fossarum population has shown that only G. fossarum type B (or clade 11; Weigand et al., 2020) occurs there. After sampling, gammarids were transported to the laboratory and kept in Urselbach stream water in a temperature-controlled room at 12 °C with oxygen supply until the start of the active monitoring campaign. At the US and DS sampling sites of both WWTPs, two stainless steel cages each containing three enclosures (12.5 cm × 6.0 cm) were fixed on the riverbed in the direction of flow. The enclosures were capped with a net (mesh size 1.0 mm) at the ends to ensure a water flow. Each enclosure, representing a replicate, contained ten gammarids, resulting in a total of 60 gammarids per site which were exposed in situ for four weeks. Black alder leaves (Alnus glutinosa) were provided as food in the enclosures ad libitum and two small nets (polytetrafluorethylene, 8.2 cm × 3.3 cm) served as hiding places to prevent cannibalism.

After four weeks of exposure, the cages were taken to the laboratory, and the gammarids were fixed in 70% ethanol and stored separately in Eppendorf tubes. As biological endpoints the mortality and fecundity index (number of eggs divided by the body length) were assessed.

For all measurements, a stereomicroscope (Olympus SZ61 R, Olympus Corporation, Tokio, Japan) with a digital camera (JVC Digital Camera KY-F75U, Victor Company of Japan Ldt., Yokohama, Japan) and the Software Diskus (Version 4.50.1458, Carl H. Hilgers, Königswinter, Germany) was used.

Active monitoring with Potamopyrgus antipodarum.

The mudsnails were collected at the rivers Lumda and Horloff (Hessen, Germany) in spring and at the rivers Gambach and Waschbach (Hessen, Germany) in fall. These sites were selected because the upper reaches of these streams are free of anthropogenic inputs. Snails with a shell length between 3.5 and 4.5 mm were used, following the protocol by Brettschneider et al. (2019a). At the US and DS sampling sites of both WWTPs, two stainless steel cages containing three smaller steel cages (4.5 cm × 3.5 cm, mesh size 0.7 mm) were fixed on the riverbed in the direction of flow. Ten snails were placed in each small cage, representing the six replicates and resulting in a total of 60 snails which were exposed in situ for four weeks. Organic carrot cubes served as food supply and were refilled after two weeks.

As biological endpoints the mortality of the snails, the shell length and the number of embryos of the surviving snails were determined using a stereomicroscope with an ocular micrometer (SMZ-168; Motic, Xiamen, China). The active monitoring with P. antipodarum followed the principle of the OECD guideline 242 (OECD, 2016) to assess the impact of reproductive toxicants, including EDCs, on this species.

Statistical analyses

Statistical analyses were performed using GraphPad Prism version 5.03 for Windows (GraphPad Software, San Diego, California, USA). Normal distribution was tested using the D’Agostino and Pearson test. Continuous data were analyzed with t-tests for normally distributed data and Welch’s correction for unequal variances. All data were tested as unpaired data and with 95% confidence interval. Quantal data (mortality) were analyzed with Fisher’s exact test. The level of significance was defined as α = 0.05 and indicated in the graphs with asterisks as follows: * = p < 0.05; ** = p < 0.01 and *** = p < 0.001.

Physicochemical parameters

The conventional treatment process (BT vs. MT) in both WWTPs removed more than 90% of the ammonium and phosphate and around 33% of the conductivity, independent of the season. Moreover, the oxygen saturation increased significantly during the fall campaign at the WWTP Mörfelden-Walldorf (MW) by 79% and in spring at the WWTP Bickenbach (B) by 89%. In the BT sample the ammonium concentrations were significantly reduced compared to MT, but the nitrate concentrations increased significantly by factor 9.6 (spring) to 19.6 (fall) in MW and by factor 2.5 (spring) to 2.9 times (fall) in B. Considering the sum concentrations of ammonium-N, nitrite-N and nitrate-N, inorganic N was removed by 90% and 94% in MW and 97 and 94% in B in spring and fall, respectively.

In vitro assays

Baseline toxicity test

The baseline toxicity, as determined with the Microtox assay, was significantly higher DS compared to US, except at WWTP MW in fall (Fig. 2). Furthermore, all sample sites, except US-B in both seasons showed a significant baseline toxicity. The baseline toxicity was removed by 90% in the WWTPs MW and B in spring and by 96% and 97% in fall. After including the effluent from the WWTP Langen (BT-L in Fig. 2A; located upstream of the WWTP MW) in the analyses in fall, it became apparent that the baseline toxicity, which was already significantly increased at US-MW, can be attributed to the discharges from the WWTP Langen.

Estrogenic and anti-estrogenic activity

All samples exhibited a significant estrogenic activity above the limit of quantification (LOD = 0.148 ng E-EQ/L). The estrogenic activity increased significantly at the DS sites compared to US at both WWTPs, with the exception of MW WWTP in fall which showed a 6% decrease (Fig. 3). Despite over 90% removal by WWTPs, estrogenic activity was 22% higher at the WWTP MW in spring, and 60% at the WWTP B in both seasons compared to their respective US sites.

The WWTPs did not cause significant increase in anti-estrogenic activity in the receiving streams (Fig. 4). Both WWTPs had a high removal efficiency of anti-estrogenic activities of over 80% in MW and 90% in Bickenbach in both seasons.

Androgenic and anti-androgenic activity

Androgenic and anti-androgenic activities were analyzed in 5,000-fold concentrated SPE extracts and native water samples, respectively. While no cytotoxic effects occurred in the native water samples, the observed cytotoxicity in the SPE extracts necessitated the analysis of diluted extracts. With the exception of the MT samples, no other sample from the two WWTPs and in the receiving streams showed an androgenic and anti-androgenic activity with a LOD of 25.6 ng testosterone-EQ/L and 771 µg flutamide-EQ/L, respectively during the spring and fall campaigns.

Dioxin-like activity

The MT samples were also cytotoxic for the yeast cells in the YDS, so that the dioxin-like activity had to be assessed in serial dilutions (Fig. 5). However, in the case of Bickenbach, even with a dilution series, it was not possible to calculate the dioxin-like activity of the MT samples (Fig. 5B).

All sample sites showed dioxin-like activity above the limit of detection (LOD = 33.8 ng β-naphthoflavone-EQ/L). Dioxin-like activity increased significantly in the DS compared to the US samples, except for the fall campaign at the WWTP MW (Fig. 5A). The increases were 21% and 11% in MW and 85% and 72% in B during the spring and fall campaigns, respectively. The calculation of the removal efficiency of dioxin-like activity was only possible for the WWTP MW because of the cytotoxicity even with highly diluted MT samples of the WWTP B. In MW, elimination rates of 60% and 53% were observed.

Ames fluctuation test

The tests with both strains without the S9-mix resulted in negative results for all samples (less than 20.8% revertant colonies). Due to their high cytotoxicity, most MT samples from the WWTPs could not be analyzed with the Ames fluctuation test. All analyzed BT samples from the WWTPs MW, B and Langen proved to be mutagenic in the tests with the strains YG1041 and YG1042 with S9 mix, as did the DS samples (Table 2). The US samples of the WWTP MW were also mutagenic with both strains, while the US sample at WWTP B was not mutagenic in the spring campaign, but mutagenic with the YG1042 strain with S9-mix in the fall campaign.

Active monitoring with Gammarus fossarum

In both rivers the mortality in US was higher in fall than in spring (Fig. 6). There was also a significantly higher mortality in DS (p < 0.05) compared to US (except WWTP B in fall) with an increase of 66% and 77% at WWTP MW and B, respectively.

The fecundity index was assessed for the spring campaign but did not show any significant difference between DS and corresponding US. For the fall campaign, the fecundity index could not be calculated because the gammarid populations were outside their reproductive phase.

Active monitoring with Potamopyrgus antipodarum

The mortality increased significantly at DS at both WWTPs in fall, while there was no effect on mortality during the spring campaign (Fig. 7).

The embryo numbers as the parameter for reproduction in P. antipodarum increased at DS of both WWTPs in the spring campaign, although this increase was not statistically significant (Fig. 8). In contrast, the embryo numbers at DS of the WWTP MW were on the same level as at US in the fall campaign but increased significantly (p < 0.01) at the WWTP B in the same period.

Physicochemical parameters

Regardless of the effectiveness of both WWTPs in reducing the concentrations of phosphate and inorganic nitrogen, they still had a strong impact on the physicochemical parameters in the receiving streams, which carried between 78% and 92% treated WW downstream of the discharger. The oxygen saturation decreased in DS-MW and DS-B during both seasons. The conductivity and the nitrate concentrations exceeded the reference value of 800 µS/cm and 0.05 mg/L, respectively for river type 19 in all DS samples during the spring and fall campaign (Halle & Müller, 2014). Despite the efficient ammonium removal in both WWTPs, the ammonium concentrations in the DS samples exceeded the reference value of 0.4 mg/L by up to factor 2.8 (MW in spring). Conversely, the concentrations of nitrate and phosphate were below the reference value during spring and fall in the DS samples (10 and 0.05 mg/L, respectively) (German Environment Agency, 2017; Halle & Müller, 2014). Finally, the mean values of the pH (not shown in Table 3) have met the requirements for good chemical status.

Dioxin-like activity

The raw WW at the WWTP MW contained compounds with aryl-hydrocarbon receptor agonistic activity, resulting in a high dioxin-like activity, which was reduced by more than 50% during biological treatment. This elimination rate is lower than in the study of Magdeburg et al. (2014) reporting an elimination of dioxin-like activity of more than 81%. For WWTP B, the elimination of dioxin-like activity through conventional WW treatment could not be determined because the raw WW samples were cytotoxic after mechanical treatment or the enzymes responsible for luminescence were directly inhibited by the samples (Reifferscheid et al., 2011). Völker et al. (2016) found a similar result for the raw WW of a WWTP in their study.

Despite the significant reduction of dioxin-like activity in both WWTPs, the remaining activity was sufficient to significantly increase the activity in the Geräthsbach and Landbach, except for the fall sample from the Geräthsbach, where the activity increased by 12.6% which was not significantly different. The measured activities in both rivers are comparable to the rivers Horloff (up to 1150 ng β-NF-EQ/L) and Nidda (up to 1,180 ng β-NF-EQ/L) in the north of Frankfurt, which also receive effluents form WWTPs but higher compared to rivers not impacted by WWTP effluents with activities less than 200 ng β-NF-EQ/L (Brettschneider et al., 2019b; Brettschneider et al., 2023).

The seasonal comparison shows significantly higher dioxin-like activities (p < 0.001) at all sampling points in MW in spring compared to fall, while for the WWTP B there is only a corresponding trend without significant differences. Although both WWTPs contribute to the high dioxin-like activity in the rivers, other sources are relevant. The dioxin-like activity in the water phase and in the sediments of rivers is primarily caused by polycyclic aromatic hydrocarbons (PAHs), which enter the waters also via atmospheric deposition and surface runoff from traffic areas (Boxall & Maltby, 1997; Maltby et al., 1995).

Mutagenicity (Ames fluctuation test)

The findings of high mutagenic activities in samples treated with S9 mix show that the conventionally treated WW contains substances that are only converted into mutagenic compounds through metabolic activation, as typically occurs by enzymes in the liver of vertebrates or the midgut gland of molluscs, crustaceans and other invertebrate taxa. Furthermore, it shows that conventional WW treatment is not sufficient to eliminate these compounds. The high mutagenic activity at the sampling points in both rivers influenced by the discharge of treated WW demonstrates the urgent need for action to reduce this undesirable effect through advanced WW treatment. Giebner et al. (2018) were able to show that activated carbon filtration in particular is suitable for removing mutagenic substances from WW, while oxidative processes such as ozonation can lead to the formation of alkylating agents, nitrosamines and other mutagenic substances in the treated WW (Mestankova et al., 2014; Schmidt & Brauch, 2008; Tsutsumi, Inami & Mochizuki, 2010).

Active monitoring with Gammarus fossarum and Potamopyrgus antipodarum

To investigate potential pollution effects on the invertebrate community, we conducted an active monitoring study with the crustacean Gammarus fossarum and the gastropod Potamopyrgus antipodarum. Both species have widely been used for ecotoxicological studies, including the assessment of WWTP effluents and their impacts on the biocenosis in receiving waters (Brettschneider et al., 2019a; Brettschneider et al., 2023; Harth et al., 2018; Schneider et al., 2020). G. fossarum is an indicator species of a low-mountain range biocenosis (Landesamt fuer Natur Umwelt und Verbraucherschutz (LANUV), 2015), is particularly sensitive to anthropogenic pressures on water bodies and plays an important role in the food web as a shredder and food for macroinvertebrates, fish, or birds (Besse et al., 2013). This dioecious and oviparous species has a breeding season of up to 10 months per year, with reproduction peaking in late winter and spring. Female G. fossarum retain the fertilized eggs and developing embryos until hatching in an open brood chamber, the marsupium, which is formed by oostegites, long appendages of the gill-carrying legs. Although changes in sex ratio and feminization effects have been described in freshwater amphipods following exposure to synthetic estrogens (Watts, Pascoe & Carroll, 2002) and estrogenically active WW (Schneider, Oehlmann & Oetken, 2015), the typical response of G. fossarum to exposure to chemicals is a decrease of the number of eggs/embryos in the marsupium (Brettschneider et al., 2019a; Brettschneider et al., 2023; Harth et al., 2018). P. antipodarum is an euryoecious snail inhabiting small creeks, streams, lakes, and estuaries. Males are extremely rare in European populations and were never found in our long-term laboratory culture and their source populations with more than 200,000 specimens analyzed. Females reproduce parthenogenetically and are ovoviviparous: Eggs develop in the anterior part of the pallial oviduct section, which is transformed into a brood pouch. Older embryos are situated in the anterior and younger embryos in the posterior part of the brood pouch. The embryos are released through the female aperture when the eggshell tears open (Organisation for Economic Co-operation and Development (OECD), 2016). As with G. fossarum, this allows to assess the reproductive output of each female by evaluating the numbers of eggs and embryos in the marsupium or brood pouch. P. antipodarum is particularly sensitive to endocrine active chemicals and responds to exposure to estrogenic substances by increasing embryo production (Duft et al., 2003; Duft et al., 2007; Jobling et al., 2004).

Given the prominent ecological function of G. fossarum in the aquatic community of low-mountain range rivers, the significant increased mortality downstream of the WWTP MW to 59% in spring and 81% in fall is worrying. Although the same tendency was observed downstream of the WWTP B in both seasons, the mortality levels were lower and increased only significantly during the spring campaign. These finding are in line with results of other active monitoring studies such as Brettschneider et al. (2019a), who found increased mortality rates of 42% and 68% in G. fossarum downstream of WWTPs with 43,800 PE and 78,000 PE, respectively.

Also, the mortality of P. antipodarum increased significantly downstream of both WWTPs in fall, while there was no significant change in mortality in spring. This finding corresponds to previous studies on other WW-influenced rivers, in which a significant increase in the mortality of P. antipodarum occurred downstream of WWTPs, especially in fall, while the effects were significantly smaller in spring (Brettschneider et al., 2019a). In addition, in previous studies in which both species were used together for active monitoring, mortality was generally lower for P. antipodarum than for G. fossarum, possibly because the snail can avoid exposure peaks by temporarily closing its shell with the operculum (Brettschneider et al., 2023).

While no influence on reproduction could be determined for G. fossarum, the number of embryos for P. antipodarum increased downstream of the two WWTPs, except for the fall campaign in the Landbach. Since it is known that P. antipodarum responds to exposure to substances with estrogenic activity by increasing embryo production (Duft et al., 2003; Duft et al., 2007; Jobling et al., 2004) and the estrogenic activity determined by the YES was significantly increased at the study sites located downstream the WWTPs, this indicates that the estrogenic exposure is the responsible factor for the higher embryo numbers. The very high toxicity below the WWTP MW in fall, which is also reflected in the highest mortalities in G. fossarum and P. antipodarum, possibly masks the estrogenic effect on the number of embryos in the snail. Also in other studies, the number of embryos in P. antipodarum increased below WWTP effluents or when exposed to conventionally treated WW (Brettschneider et al., 2019a; Schneider et al., 2020; Stalter, Magdeburg & Oehlmann, 2010).

Together, these findings demonstrate that the applied methods are suited to assess the negative effects of effluents from conventional WWTPs on the receiving water bodies and the aquatic biocenosis in these rivers. For the next project phase after the implementation of the 4th treatment step with a combination of ozonation, followed by activated carbon filtration in the two WWTPs, the same physicochemical parameters and EBMs should therefore be used to enable an effective comparison between the status quo before and after implementation. However, the Salmonella typhimurium strain YG7108, which has proven to be particularly sensitive for detecting mutagenic compounds formed during ozonation (Giebner et al., 2018; Magdeburg et al., 2014; Schneider et al., 2020), will also be considered for the Ames fluctuation test in the future. This makes it possible to check whether these compounds are effectively removed by activated carbon filtration after ozonation. The use of the same battery of EBMs allows to assess the efficiency of the upgrading measure and to record potential pollution reduction in the receiving waters.

The findings of our study can be extrapolated to other smaller rivers that receive conventionally treated WW from relatively large WWTPs. The current study demonstrates that the effluents originating from the two WWTPs negatively affect key life cycle of representative macroinvertebrate species in the water bodies and underlines the necessity to upgrade WWTPs with a 4th treatment step to achieve the target of a good chemical and ecological status of surface waters according to the EU Water Framework Directive (European Commission, 2019).