In vitro effect‑based monitoring of water, sediment and soil from a floodplain restoration site in Central Europe

Background Floodplains are biodiversity hotspots and provide numerous ecosystem services. In recent decades, however, 70–90% of Europe’s floodplains have been structurally degraded. Accordingly, many (inter‑)national pro‑ grams aim to restore and protect floodplain ecosystems. The success of such measures also depends on the chemical contamination, especially of floodplain soils and sediments, which serve as sinks and sources for a variety of pol‑ lutants. In this study, we assess the current ecotoxicological status of a floodplain restoration site along the Main River (Frankfurt am Main, Germany) and estimate its development potential with respect to the influence of a local industrial plant and potential legacy contaminations. We therefore use in vitro effect‑based methods (EBMs) testing for baseline toxicity, mutagenicity, dioxin‑like and estrogenic activities, coupled with chemical analysis. Results Of all water bodies analyzed, the overall toxicity was highest in two flood depressions. In the respective water phase, estrogenic activities exceeded the environmental quality standard and sediment samples were positive for all tested endpoints. Chemical analysis of these sediments revealed high concentrations of polycyclic aromatic hydro‑ carbons. Soil samples from frequently flooded areas showed the highest mutagenic potential for both frameshift and point mutations with and without metabolic activation. The industrial effluent showed baseline toxic, mutagenic, and dioxin‑like activities, that were highly diluted in the Main River. In turn, most of the sediment samples down‑ stream of the industrial discharge showed significantly elevated baseline toxic, estrogenic and dioxin‑like activities as well as increased chemical contamination. Conclusion Based on the results of this study, we rate the overall ecotoxicological status of a recently established tributary and groundwater‑fed ponds as good, and identified two flood depressions near the Main River as hot spots of contamination. We assume that the observed mutagenicity in the floodplain soils is related to legacy contamina‑ tions from former aniline and azo dye production. In terms of the development potential of the floodplain restoration site, we emphasize considering the remobilization of pollutants from these soils and suppose that, in the long term, pollution of the Main River and the local industrial plant may negatively impact sediment quality in its tributaries. With this study, we confirmed the utility of in vitro EBMs for identifying chemically and ecotoxicologically relevant sites.


Floodplain restoration
Floodplains are among the most species-rich ecosystems in Central Europe and are therefore of great importance for biodiversity conservation [1].In addition, they fulfill a variety of ecological services, such as water and air filtration, carbon fixation, or urban climate improvement and contribute considerably towards achieving environmental policy objectives [2,3].Over the past decades, however, rivers have been channelized and structurally degraded to allow navigation, gain land, and protect against flooding.The corresponding loss of habitat structures has severe negative consequences for the environment.Accordingly, several international programs aim to protect and restore floodplain habitats, such as the European Water Framework Directive (EU-WFD) [4], the European Biodiversity Strategy [5], the Floods Directive [6], the Habitats and Birds Directives [7], as well as many national programs in Germany [8][9][10].In particular, the implementation of the EU-WFD in 2000 led to an increase in the number of respective restoration projects across Europe [11,12].Yet it is estimated that to date 70-90% of Europe's floodplains are ecologically degraded [3].In Germany, over 90% of the floodplains are still classified as clearly to very strongly altered [12].Chemical pollution, e.g., from agriculture and industry, has adverse and far-reaching effects on aquatic ecosystems as well.Floodplains play an important role in this context, as they can act as a sink and source for pollutants.During flood events, pollutants can, depending on their properties (e.g., persistence and polarity), bind and accumulate to the floodplain sediments and soils [13][14][15].During floods and construction work in floodplains, pollutants can be remobilized and become bioavailable again [16].One example of such source-sink dynamics of sedimentbound pollutants is the flooding of the Elbe River in 2002, when highly contaminated sediments from Elbe tributaries were resuspended in the water, transported, and deposited on fields and grazing lands [15,17].However, previous studies have shown that in addition to hydromorphological restoration measures, good water and sediment quality is crucial to improve, for example, the ecological status of water bodies according to the EU-WFD [18][19][20].This underlines the importance of a holistic approach when planning and evaluating floodplain restoration.

Study site
The investigated restoration site is a 90 ha floodplain in the east of the City of Frankfurt am Main (Hesse, Germany), which to date is the most extensive restoration measure on the Hessian section of the Main River (Fig. 1).Originating from headwaters of the Fichtel Mountains and the Franconian Alb in eastern Germany, the Main River flows from east to west through several Franconian low mountain ranges, large parts of the Franconian wine-growing region, and major cities such as Würzburg and Frankfurt am Main.Near the City of Mainz, after 527 km, it flows into the Rhine River.As federal waterway, it has been deepened, channelized, and has several dams so that long-distance migratory species are lacking completely [21].Moreover, the adjacent riparian areas are used for urban, industrial, and agricultural purposes.In 2021, the Main River is still considered structurally very strongly to completely altered, the ecological status is classified as moderate to poor, and the overall assessment of the ecological potential is rated as unsatisfactory [22].Although highly frequented federal waterways such as the Main River will remain significantly modified water bodies, they do provide valuable habitats and play an important role in biotope connectivity.
As the largest undeveloped floodplain in Frankfurt, the restoration site was mainly used for agriculture in the past.However, some valuable structures have developed, such as small softwood floodplain forests and riparian copses rich in deadwood (Fig. 1a).The restoration planning includes two artificial tributaries, which are 675 m and 1.7 km long, several oxbow ponds, flood depressions, bank flattening, meadows, and floodplain forests.About one-third of the area remains available for conventional agriculture (Fig. 1b).In the western part of the plan area, bank flattening, oxbow ponds and flood depressions were already established in 2014, although some will disappear in the large tributary when it is built around 2030.The smaller tributary near the bank and another large oxbow pond in the eastern part of the plan area were established in 2019 [23].The restoration of the floodplain plays an important role in the local species and biotope protection, as habitats for several endangered species are being created [24].
The restoration project is particularly interesting from an ecotoxicological perspective because the plan area is located next to an industrial plant with its wastewater discharge located 1.5 km upstream of the future tributary.The plant started producing aniline and azo dyes (also known as tar dyes) in 1870.During the twentieth century, the product range expanded by pharmaceuticals and cosmetics as well as specialty and laboratory chemicals.In 1981, after more than 100 years of discharging untreated wastewater into the Main River, it put a wastewater treatment plant (WWTP) into operation.While in 1970 the Main River around Frankfurt was so heavily polluted that only four of the former 30-35 fish species survived, the introduction of WWTPs led to a recovery of fish populations [21].Regarding the major flood events of the last decades, however, pollutants transported from  [23] further upstream the Main River and the effluents of the industrial plant are hypothesized to have been deposited on the floodplain.For example, in 1942For example, in , 1970For example, in , 1988For example, in , 1995For example, in , 2003, and 2011 major floods, rated as 10-and 20-year floods, affected the area of the future tributary (Fig. S1) [25].

In vitro effect-based methods for ecotoxicological monitoring of water bodies
According to the EU-WFD (Annex II and V) [4], the state of water bodies and their development in the course of restoration measures is assessed based on their chemical and ecological status.The ecological status is defined by monitoring a range of biotic communities which has the strength of potentially addressing complex mixtures of stressors (e.g., water and sediment pollution, habitat degradation, catchment land use, flow modification, impact of alien species).However, uncertainty remains about the cause of the observed status.The chemical status is presently assessed based on compliance with legally binding Environmental Quality Standards (EQSs) for 45 Priority Substances [26], and a set of nationally defined catchment-specific substances [4].The assessment gets even more complex when including additional chemical data from other matrices such as suspended particulate matter (SPM) and sediments that lack appropriate EQSs.Even when extended to a broad range of target chemicals using screening approaches, component-based mixture risk assessment is restricted to these targets.Non-targeted chemicals including metabolites and transformation products may be overlooked and complex chemical mixture effects might be underestimated.Effect-based methods (EBMs) are a suitable tool to face this challenge [27][28][29][30].They detect and quantify the effects of complex environmental samples with unknown chemicals on toxicologically relevant endpoints by studying the response of whole organisms (in vivo) or cellular bioassays (in vitro).This can also be helpful to link the chemical and the ecological status [18,20].A recently published EU Commission's proposal for amending the Water Framework Directive included EBMs in the legal text of the WFD [31,32].However, the Commission proposal limits EBMs to explorative studies and does not include the setting of EQS values based on EBM-methods.Further, EBMs are necessary for an effect-directed analysis that can help identify those chemicals that might cause adverse effects [27,33].Additionally, in vitro EBMs have the advantage of being suitable for high-throughput approaches, as they are cost-effective, quick and easy to implement, and require only small sample volumes.Their strong predictive power also allows for the selection of relevant sites for more extensive chemical analysis and subsequent in vivo testing, which is not feasible on a large scale.Accordingly, in vitro EBMs are also beneficial from an ethical perspective as they help reduce animal testing.

Study design
In the present study, we aim to assess the current ecotoxicological status of the existing waterbodies of the floodplain and to estimate the development potential of the restoration site.In this context, we hypothesize that i. the current ecotoxicological status of the floodplain water bodies and soils is adversely affected by legacy contamination.ii.pollutants from the current effluent of the local industrial plant adversely affect the water and SPM quality of the Main River.iii.pollutants from the effluents of the local industrial plant have adversely affected the sediment quality of the Main River over the past years.
To test these hypotheses, we analyzed samples of different matrices from the Main River and its floodplain (water, SPM, sediment, soil) by using a battery of in vitro EBMs for ecotoxicologically relevant endpoints: (1) Baseline toxicity is determined using the Microtox assay with the bioluminescent bacterium Aliivibrio fischeri.Results of the Microtox assay show a good correlation to more complex bioassays, which gives it a strong predictive power [34].Further, an interlaboratory comparison exercise verifies its reproducibility [35].(2) Mutagenicity is investigated with the Ames fluctuation assay, which detects the mutagenic potential of a sample based on the ability of Salmonella typhimurium strains to revert histidine auxotrophy.The strains used in this work, YG1041 for frameshift mutations and YG1042 for point mutations, are especially sensitive to mutagens from the group of nitrated aromatic hydrocarbons [36].The Ames fluctuation test [37], a microplate version of the classic plate-incorporation method, was evaluated for environmental samples in an international round-robin study [38].(3) Estrogenic activity via receptor activation is investigated using the Yeast Estrogen Screen (YES), a reporter gene assay with Saccharomyces cerevisiae expressing the estrogen receptor alpha (ERα).The assay is widely used for the analysis of wastewater and environmental samples.(4) Dioxin-like activity is analyzed with the Yeast Dioxin Screen (YDS), another reporter gene assay with Saccharomyces cerevisiae expressing the aryl hydrocarbon receptor (AhR, also known as dioxin receptor).The AhR functions as a transcription factor for various target genes commonly involved in biotransformation processes (e.g., cytochrome P450 A1), whereas corresponding ligands themselves usually represent substrates of the enzymes whose expression is induced via the AhR.Its activation can therefore be used to indicate xenobiotic metabolism.Synthetic ligands are mostly characterized by a planar molecular structure and are considered persistent organic pollutants.With 2,3,7,8-tetrachlorodibenzo-1,4-dioxin as the most potent known compound, they are grouped under the term dioxin-like compounds (DLCs) and induce a wide range of biological responses such as reproductive and developmental disorders, immunotoxicity, or cancer [39,40].(5) To complement the foundation for an ecotoxicological assessment and to test the applicability of in vitro EBMs to identify chemically relevant sites, the samples were analyzed for organic pollutants using a target screening methods based on liquid and gas chromatography coupled to high-resolution mass spectrometry (LC-HRMS and GC-HRMS).

Sampling sites
Within the floodplain, we sampled three groundwaterfed ponds (P1, P2, P3) and two small flood depressions near the bank (F1, F2).Three sampling sites are located in the small tributary which has two basin-like structures (T1, T3) that are connected by a narrow and deep part with a higher flow velocity (T2).In addition, we chose one sampling site in the area of the future tributary (T4) that is planned to be built along a depression in the restoration area (affected by 5-year floods).One further sampling site is located in the riparian area (R1) that is regularly flooded (at least by 5-year floods).The area is planned to develop from a wet meadow into a floodplain forest in the coming years.To assess the influence of the Main River and the current industrial effluent on the development potential of the floodplain, we sampled six sites on the right river bank in the direction of flow at regular intervals (Fig. 2, Table 1).This includes a reference site upstream of the industrial discharge (M1), one site directly at the discharge (M2) and four transect sites downstream of the discharge (M3, M4, M5, M6).Site M4 represents the inflow level of the future tributary.Site M6 represents a restored riverbank, where boulders were removed and backfilled with sand in 2014 to create a natural bank flattening.A detailed overview of the sampling sites including characteristics, coordinates, and the matrices sampled at each site, is provided in Table 1.
Sampling sites along the Main River (M), the tributary (T), the ponds (P), the flood depressions (F) and the riparian area (R).WWTP: wastewater treatment plant; us: upstream; ds: downstream; w: water; p: suspended particulate matter; ss: surface sediment; sc: core sediment; h: soil horizon

Sample collection and preparation
Water samples 2 L of water samples were taken from the tributary, the ponds and the Main River from 30 to 50 cm water depth, and in the flood depressions from 10 cm water depth.We further received a 24-h composite sample of the corresponding sampling day from the effluent of the biological wastewater treatment plant.All samples were stored at 10 °C overnight and filtered through glass microfiber filters (VWR International GmbH, No. 696, Cat No. 515-0879, 125 mm, particle retention: 1.0 µm, Darmstadt, Germany).Water samples were solid phase-extracted according to Giebner et al., but eluates from 2 L of filtered water were concentrated to a final volume of 400 µL [41].This resulted in 5.000-fold Fig. 2 Sampling sites along the Main River (M), the tributary (T), the ponds (P) and the riparian area (R).An additional sample was taken from the effluent of the wastewater treatment plant (E).White arrows indicate the direction of flow concentrated water extracts in DMSO.For the process control, 1 L ultrapure water was treated the same way.Detailed information on the water sampling is provided in the supporting information.

Suspended particulate matter, sediment and soil samples
SPM was sampled in the Main river for 35 days using self-made sedimentation traps that were placed between boulders in proximity to the bank with the inflow at about 10 cm water depth.Surface sediments from the tributary and the ponds were collected with a shovel.Sediments from the Main river were taken with pipes from up to 5 m water depth.We thereby obtained different sediment types.At the reference site M1 and the downstream sites M4 and M6 we found sandy river sediment (Fig. S3a).In contrast, at the discharge site M2 and the downstream sites M3 and M5 we found gray, clayey sediment (Fig. S3b).Soil samples were taken from the upper two horizons.Detailed information on the sampling is provided in the supporting information.
All samples were stored at − 21 °C until freeze-drying, sieved to < 2 mm, and stored at room temperature in the dark.For the pressurized liquid extraction, extraction cells were filled with 20 g of dry sediment, SPM or soil mixed with 5 g (25%) of analytical grade quartz sand (BÜCHI Quartz Sand; 0.3-0.9mm; Nr. 037689) as well as a layer of pure analytical grade quartz sand at the bottom and top up to the permissible filling level of the cells to prevent sediment clumping and clogging of the metal fits.For the process control, columns were filled with pure analytical grade quartz sand up to the permissible filling level.Extraction was performed using a BÜCHI SpeedExtractor E-916 with ethyl acetate (CAS: 141-78-6; LC-MS CHROMASOLV ™ , Honeywell Riedel-de Haën ™ , Germany) and acetone (CAS: 67-64-1; ROTISOLV ® ≥ 99,9%, LC-MS Grade, Carl Roth, Germany) as solvents (1:1, v/v) in two cycles at 100 °C and 103 bar (Preheat: 5 min; Static time: 5 min; Flush volume: 60% of cell volume; Purge time: 60 s) in a method optimized for multitarget screening of neutral compounds by GC-HRMS and LC-HRMS [42].The resulting eluates were concentrated to roughly 1 mL by rotary evaporation at 50 °C and 500 to 350 mbar (BÜCHI Multivapor P-6/Rotavapor R-300, BÜCHI Vacuum Pump V-300, BÜCHI Interface I-300 Pro, BÜCHI Recirculating Chiller F-305), transferred to glass vials, evaporated to about 100 µL under a nitrogen stream, and finally stored at − 21 °C.
We performed a subsequent column clean-up to remove co-eluted natural organic matter that would interfere with GC-HRMS and LC-HRMS analysis.This From the methanol fraction, 10% (vol.) were discarded to ensure an even removal from the total extract.The remaining 90% of the DCM and methanol fractions were combined and concentrated to 1.5 mL by rotary evaporation at 40 °C and 600 mbar (BÜCHI Syncore, BÜCHI Vacuum Pump V-700, BÜCHI Vacuum Controller V-855, BÜCHI Recirculation Chiller B-740).Afterwards, the solvent was exchanged for methanol under a nitrogen stream.In case of precipitation after storage at − 21 °C for 1 h, samples were passed through PTFE syringe filters, as described before, blown to dryness and reconstituted in exactly 1.5 mL methanol.From the resulting extracts, 10% (vol.) were taken for LC-HRMS analysis.
To use the remaining extract for biotesting, we replaced methanol with 810 µL of DMSO under a nitrogen stream, resulting in a final sample concentration of 20 g sediment, soil or SPM equivalents (SEQ)/mL.A process control was included for the entire clean-up process.

Microtox assay
Baseline toxicity was determined using the Microtox assay with Aliivibrio fischeri according to the International Standard Operation (ISO) guideline 11348-3 [44] but transferred to 96-well as previously described [45,46].In brief, extracts were tested in a 1:2 dilution series based on a 50-fold enrichment of the native water samples or 30 mg SEQ.The luminescence of A. fischeri was measured before and after 30 min of incubation with the respective sample.According to the ISO guideline, the measurements were corrected for the luminescence of the blank as well as the ratio of the average luminescence in the negative controls (x̅ (t30/t0)), resulting in a relative luminescence inhibition (%).Samples with inhibition of more than 20% were considered toxic.

Ames fluctuation assay
Mutagenicity was investigated using the Ames fluctuation assay with strains of Salmonella typhimurium.The assay was performed according to the ISO guideline 11350 [37] but with different Salmonella strains and respective modifications.Accordingly, the used cell densities were adjusted to 150 (YG1041 strain) and 180 (YG1042 strain) formazin attenuation units (FAU) when including metabolic activation (S9) and 170 (YG1041 strain) and 80 (YG1042 strain) FAU without S9.The extracts were diluted 300-fold in the test, resulting in a final concentration of 16.7 REF for water samples and 66.7 mg SEQ/mL for sediment, SPM, and soil extracts.In the case of cytotoxicity, extracts were further diluted so that the effluent sample (E) was tested at a final concentration of 6.7 REF and sample M6p with 26.7 mg SEQ/mL.2-Nitrofluorene (2NF; CAS 607-57-8) was used as a positive control when testing for mutagenicity without S9, 2-aminoanthracene (2-AA; CAS 613-13-8) when using S9.Reverse mutations were determined after incubation for 72 h at 37 °C in the dark.Wells with an optical density OD 420 < 0.4 (YG1042 without S9: 0.29) were considered relevant.For colored extracts that affected the OD 420 , the threshold was adjusted accordingly (YG1041 strain without S9: M1p, M5p, M6p: 0.55).Results were corrected for background mutations in the negative controls.Samples exceeding 20.8% revertants in two independent replicates were classified as mutagenic.

Yeast reporter gene assays
Estrogenic and dioxin-like activities were determined using reporter gene assays with recombinant strains of Saccharomyces cerevisiae.The operation procedure used in this work is based on Routledge and Sumpter [47] for the YES and Sohoni and Sumpter [48] for the YDS with several modifications.In brief, a 480-fold dilution of the extracts (resulting in a 10.4-fold final concentration of water samples and 41.6 mg SEQ/mL for sediment, soil and SPM samples), a solvent control (DMSO), and the positive control substances (YES: 17β-estradiol (CAS: 50-28-2; > 99%; Merck, Darmstadt), YDS: β-naphthoflavone (CAS: 6051-87-2; purum; Fluka/70415)) were prepared on a 96-well plate with eight technical replicates each.Extracts were further diluted when the cytotoxicity exceeded 20% compared to the negative control.The respective yeast solution was adjusted to 250 (YES) and 1000 (YDS) FAU.Samples were then incubated for 20 h at 1200 rpm and 30 °C.The plates were covered with Breathe-Easy membranes (Sigma-Aldrich Z380059) and scored between rows for additional oxygen supply.After cell growth was determined photometrically at 595 nm to investigate potential cytotoxicity, lacZ-buffer containing 4-methylumbelliferylβ-D-galactopyranoside (MUG; CAS 6160-78-7, Merck, Darmstadt) and L-1 dithiothreitol (DTT; CAS: 3483-12-3, Sigma-Aldrich) was added to each well.DTT was used to improve the permeability of the cell wall and thus the diffusion of MUG.Fluorescence (excitation = 360 nm, emission = 465 nm) was recorded after 1 h of incubation at 1200 rpm and 30 °C.The fluorescence data were corrected for blank, cell density, dilution and enrichment of the samples and converted to EQs of 17β-estradiol (YES) or β-naphthoflavone (YDS) using the respective concentration-response relationship.The limit of quantification (LOQ) was calculated using the mean activity of the negative/process controls and adding the threefold standard deviation.Final results were corrected for background activities in the process controls and are based on three independent tests with eight technical replicates each in the presence, two in the absence of an effect potential.
As most studies on DLCs and AhR-related effects refer to the potent ligand 2,3,7,8-tetrachlordibenzo-1,4-dioxin (TCDD), a conversion formula is useful to compare the β-NF-EQ measured in this study with TCDD literature values from other studies.For the calculation, 39 sediment samples with strongly varying dioxin-like activities were examined in parallel in the YDS with β-NF as positive control and in the µEROD with Danio rerio larvae using TCDD as a positive control.The resulting β-NFand TCDD-EQs made it possible to derive a conversion formula, based on a highly significant correlation (Spearman r = 0.927, p < 0.0001): TCDD − EQ pg g SEQ = 10 −0,04701+0,5108 * log β−NF−EQ ng g SEQ .

Statistical analysis
The statistical analyses were performed using Microsoft ® Excel ® (version 16.0, Microsoft Corporation, Redmond, USA) and GraphPad Prism ® (versions 5.03 and 9, Graph-Pad Software Inc., San Diego, California, USA).For the yeast reporter gene assays, datasets were analyzed for normal distribution using the D' Agostino and Pearson omnibus normality test.In the case of a normal distribution, significant differences between datasets were determined using a one-way-ANOVA with Bonferroni post hoc test (α = 0.05).When datasets were not normally distributed, a Kruskal-Wallis test with Dunn's post hoc test (α = 0.05) was used.Concentration-response relationship curves were derived using a four-parameter variable slope model with the bottom value constrained to the mean activity of all negative controls (0% receptor inhibition).For the Microtox assay, bottom and top values were constrained to zero and 100%, respectively.Statistical significance was calculated for samples with activities > LOQ using an unpaired t-test (α = 0.05).The Main River samples M2-M6 were each compared to the reference site M1.For datasets with significantly different variances, an unpaired t-test with Welch's correction was used.No statistical significance was calculated for floodplain samples due to the lack of an appropriate reference site.

Chemical analysis
The chemical analysis was performed by LC-HRMS and GC-HRMS and included 507 chemicals from 12 categories for water samples (LC-HRMS only) and 556 chemicals from 14 categories for SPM, soil, and sediment samples.The selection of contaminants was based on known or anticipated occurrence in the aquatic environment based on previous studies and literature data.A list of contaminants and their classification is provided in the supplementary data.Instrumental conditions and detection limits for grab water samples are based on the method described by Beckers et al. [49] and for SPM, soil and sediment samples as described by Machate et al. [50].Sample concentrations were blank corrected prior to further analysis.

Ecotoxicological status of the floodplain water bodies
To assess the current ecotoxicological status of the floodplain waterbodies, we tested water and surface sediment from the tributary (T), the ponds (P), and the flood depressions (F).
In the water phase, we detected very weak baseline toxicity in only three of eight samples (T2w, P3w, F1w), all other samples were not active in this assay (Fig. 3a).Mutagenic potential (> 20.8% revertants) was observed exclusively in one sample from the tributary (T2w) and only in YG1041 strains after metabolic activation with 36% revertants (Fig. 3b).The strongest estrogenic activities were found in the flood depressions with values of 1.14 ng E 2 -EQ/L in F1 and 0.56 ng E 2 -EQ/L in F2, which exceeded the EQS for E 2 of 0.4 ng/L by a factor 2.85 and 1.4, respectively (Fig. 3c).The YDS showed weak effects above the LOQ (0.04 µg β-NF-EQ/L) in all water samples with the strongest activity of 0.29 µg β-NF-EQ/L in sample T2w (Fig. 3d).In this sample taken from the junction of both basins of the tributary, we found both, elevated dioxin-like activity and mutagenic potential in the YG1041 strain after metabolic activation.These results may indicate the remobilization of sedimentbound contaminants due to the higher flow velocity.Although the sediment sample at this site was not active in the respective assays, there may be contamination at other depths or areas of the water body.It is Fig. 3 In vitro activities in water samples (w) from the tributary (T), the ponds (P), and the flood depressions (F) in the floodplain restoration site (see Fig. 2 and Table 1 for sampling site details).a Baseline toxicity in the Microtox assay is expressed as the mean and standard error of the mean (SEM) of 50% effect concentration (EC 50 ) based on the relative enrichment factor (REF) of the water samples.Non-toxic samples were set to 100 REF.b Mutagenicity in the Ames fluctuation assay is expressed as % revertants with > 20.8% revertants being considered mutagenic.Sample concentration in the test was 6.7 REF.The triangles represent the results for point mutations in the YG1042 strain, and the diamonds those for frameshift mutations in the YG1041 strain.Filled symbols indicates mutagenicity with metabolic activation (S9), and open symbols mutagenicity without S9.c Estrogenic activity in the Yeast Estrogen Screen (YES) expressed as mean and SEM in equivalents (EQ) of the positive substance 17β-estradiol (E 2 ).d Activity in the Yeast Dioxin Screen (YDS) expressed as mean and SEM in EQ of the positive substance β-naphthoflavone (β-NF).c, d The dotted line represents the limit of quantification (LOQ) not known, whether and how the soil was redistributed during the construction work.
In the respective surface sediment, we found moderate baseline toxicity with EC 50 values below 21.4 mg SEQ in all water bodies except one groundwater-fed pond (P3) that showed only low activity of 54.8 mg SEQ (Fig. 4a).This is comparable with measured activities in other river sediments [18,51].The very low to absent activity in the associated water samples suggests that there is no remobilization of respective compounds.However, to assess potential harmful effects in organisms it may be useful to test parameters such as benthic community structure.In all other in vitro EMBs with sediments from the floodplain, we only found distinct activities in both flood depressions F1 and F2 (Fig. 4b-d).Including metabolic activation of the sample, F1ss and F2ss induced mutations in the YG1041 (35% and 26.6% revertants, respectively), F2ss also mutations in the YG1042 strain (33.6%).Also without metabolic activation, F2ss induced mutations in the YG1041 strain (29.8%) (Fig. 4b).The mutagenicity of both samples could be caused by legacy contaminants (see also 3.2).Also, polycyclic aromatic hydrocarbons (PAHs) that have accumulated in flood depressions (Fig. 5) are among the most well-known environmental mutagens [52,53].
Estrogenic activities above the LOQ (0.7 ng E 2 -EQ/g SEQ) were exclusively found in the flood depressions with equivalent concentrations of 4.15 ng E 2 /g SEQ in F1 and 4.13 ng E 2 -EQ/g SEQ in F2 (Fig. 4c), which can be compared with the effect potentials found in Main River Fig. 4 In vitro activities in surface sediment (ss) and soil (h) samples from the tributary (T), the ponds (P), the flood depressions (F), and the riparian area (R) in the floodplain restoration site (see Fig. 2 and Table 1 for sampling site details).a Baseline toxicity in the Microtox assay is expressed as the mean and SEM of EC 50 based on mg/sediment or soil equivalents (SEQ).Non-toxic samples were set to 100 mg SEQ.b Mutagenicity in the Ames fluctuation assay is expressed as % revertants with > 20.8% revertants being considered mutagenic.Sample concentration in the test was 66.7 mg SEQ/mL.The triangles represent the results for point mutations in the YG1042 strain, and the diamonds those for frameshift mutations in the YG1041 strain.Filled symbols indicate mutagenicity with S9, open symbols mutagenicity without S9.c Estrogenic activity in the YES is expressed as mean and SEM in EQ of the positive substance E 2 .d Activity in the YDS is expressed as mean and SEM in EQ of the positive substance β-NF.c, d The dotted line represents the LOQ.For abbreviations see Fig. 3 sediments and SPMs (Fig. 8c).We therefore assume that the activities in both flood depressions originate from the nearby Main River.Also possible, due to the proximity to the pedestrian path, would be another anthropogenic input, such as dog urine and feces.
Dioxin-like activities in the YDS were highest in both flood depressions with 78.2 µg β-NF/g SEQ in F1 and 164 µg β-NF-EQ/g SEQ in F2, corresponding to 283 and 414 pg TCDD-EQ/g SEQ, respectively.These are by far the highest dioxin-like activities found in this study.All other sediment samples showed weak activities below 5 µg β-NF-EQ/g SEQ.
As shown by the chemical analysis of the floodplain sediment samples (Fig. 5), both flood depressions are heavily contaminated with PAHs, with cumulative concentrations of 6.45 and 59.6 µg PAHs/g in F1ss and F2ss, respectively.These concentrations are considerably higher than in the tributary, the area of the future tributary and the ponds (< 0.5 µg/g) and the riparian area (< 0.9 µg/g) and correlate with activities found in the YDS.Particularly high are the concentrations of benzo[a]pyrene (28.3 µg/g) and benzo[e] pyrene (21.4 µg/g) in sample F2ss.Both PAHs are products of incomplete combustion of organic matter and are introduced into the aquatic environment inter alia by atmospheric deposition.However, the significant concentration differences in both flood depressions might also indicate additional sources.Since the flood depression F2 is located next to a popular fishing spot, the comparably high concentrations could also result from the improper disposal of barbecue charcoal and cigarettes.We also found persistent organic pollutants (POPs) such as polychlorinated benzenes or biphenyls (PCBs) in the flood depressions, the soil samples from the floodplains, and also in the older Main River sediments (Fig. 9), suggesting that they are legacy contaminants.
Overall, sediment and water samples from the new tributary and the groundwater-fed ponds were unremarkable in the in vitro EBMs and showed a low chemical load compared to all other samples.In contrast, sediments from both flood depressions are active in all in vitro EMBs, and the EQS for E 2 was exceeded in the water phase.The artificial flood depressions represent temporary water bodies that regularly fall dry and have no access to the Main River.Accordingly, they do not provide habitats for fish, but for insect larvae and amphibians, for example, which might be suitable organisms for an ERA that accounts for the bioavailability of detected contaminants.

Influence of potential legacy contamination
As a consequence of the major flood events in the last decades, pollutants from the Main River and the effluent of the former Cassella AG are hypothesized to have been deposited on the floodplain.We analyzed soil samples from upper horizons in the riparian zone (R) and the area of the future tributary (T4).
None of the soil samples showed estrogenic activities (Fig. 4c).Also, baseline toxicity and dioxin-like activities were low with EC 50 values of 50.6 to 76.6 mg SEQ and 2.13 to 4.72 µg β-NF-EQ/g SEQ, respectively (Fig. 4a, d).However, we found strong mutagenic potential in all samples (T4h1, T4h2, R1h1, R1h2) in both strains, as well as with and without metabolic activation (Fig. 4b).In the riparian zone, we found up to 91.2% revertants and in the area of the future tributary up to 75.6%.Since both sampling sites are affected by at least 5-year floods, we assume that the observed mutagenicity is related to the hypothesized legacy contamination, which is further supported by the sensitivity of the used Ames strains to nitrated aromatic hydrocarbons.For over 150 years, the local industry has been producing and discharging mutagenic tar dyes without an operating WWTP [54,55].Concerning the development potential of the restoration site and remobilization of potential legacy contamination, the decisive factor will be whether the contaminated Fig. 5 Contamination in surface sediment (ss) and soil samples (h) from the tributary (T), the ponds (P), the flood depressions (F) and the riparian area (R) in the floodplain restoration site (see Fig. 2 and Table 1 for sampling site details) soil has direct contact with the water body.Fortunately, apart from the two flood depressions F1 and F2, none of the existing water bodies showed mutagenic activities, which can possibly be explained by their water depth.Soil samples were only taken from the upper two horizons with a depth of up to 44.5 cm, which is similar to the depth of the flood depressions.The sediment samples from the tributary and the ponds were taken from much deeper layers.This may indicate that the potential legacy contamination affects only the upper soil layers.Apart from this, the ponds are located in less heavily flooded areas and are therefore less contaminated with potential legacy contamination.However, we recommend considering the high mutagenicity in the tested soil layers when planning and implementing the future tributary.

Influence of the current industrial effluent on Main water and SPM quality
To assess the current impact of the on-site industrial plant, we analyzed a 24-h composite sample from the WWTP effluent (E), water samples from six sites along the Main river (M) (Fig. 6), and SPM samples collected over 35 days at three sampling sites (Fig. 8).
In the effluent sample, we found high baseline toxicity with an EC 50 of 2.88 REF.This activity was still measurable directly at the discharge point (M2) with an EC 50 value of 42.1 REF, which equals a 14.3-fold dilution.In the following transects, along with the reference site M1, we found no baseline toxicity (Fig. 6a).The effect potentials found in this study are common for industrial effluents, which are often more potent in the Microtox assay than municipal effluents [56,57].This might be relevant for the functionality of the biological treatment stage, as high toxicity to bacteria may impact the purification process [58].All SPM samples showed high baseline toxicity, both upstream and downstream of the WWTP, with EC 50 values of 3.34, 3.42, and 6.02 mg SEQ, respectively (Fig. 8a).We further found mutagenic potential in the effluent sample in three of the four tested Ames assays: 47.8% revertants in the YG1042 and 50.6% in the YG1041 strain without metabolic activation, as well as 49.5% in the YG1041 strain after metabolical activation of the sample.Again, the effect potential was still measurable at the discharge site M2 with 44.7%, 62.1%, and 80.8% revertants, respectively.Even 700 m downstream the WWTP discharge at site M3, we found 35% revertants in the YG1041 strain after metabolic activation.All other transect samples were not active in the Ames tests (Fig. 6b).Due to cytotoxicity, the effluent sample E was tested at a lower concentration (6.67 REF) than the other water extracts (16.7 REF), which could explain the comparably high activity in the Main River samples.The SPM samples showed no mutagenic potential (Fig. 8b).In all water samples from the Main River, estrogenic activities were below the LOQ (0.41 ng E 2 -EQ/L).In the effluent sample, we found no measurable estrogenic activity (Fig. 6c).In contrast, all SPM samples were active in the YES, with activities decreasing in the direction of flow from 5.61 ng E 2 -EQ/g SEQ at the reference site to 1.81 and 1.16 ng E 2 -EQ/g SEQ at sites M5 and M6, respectively (Fig. 8c).Other studies on SPM from the Rhine (Germany) and Meuse River (Netherlands) reported similar or lower E 2 -EQs of up to 2.35 [59] and 0.7 ng/g SEQ [60], respectively.The YDS showed significant to highly significant activities above the LOQ (0.01 µg β-NF-EQ/L) in all Main water samples compared to the reference site M1 (0.03 µg β-NF-EQ/L) (Fig. 6d; statistical data: Tab.S1).The highest activity was 6.22 µg β-NF-EQ/L in the effluent sample, followed by 0.35 µg β-NF-EQ/L at the discharge point, which equals an 18-fold dilution.The remaining downstream transects (M3-M6), showed weak dioxin-like activities of 0.04 to 0.06 µg β-NF-EQ/L, corresponding to a 1.3-to 2-fold increase compared to the reference site (Fig. 6d).We also found dioxin-like activities in all SPM samples.As observed in the YES and contrary to our expectation that the effluent causes to a higher dioxin-like activity in SPM samples downstream of the discharge, the activities decreased in the direction of flow from 14.8 µg β-NF-EQ/g SEQ at the reference site to 4.72 µg β-NF-EQ/g SEQ at site M5, and to 2.9 µg β-NF-EQ/g SEQ at site M6 (Fig. 8d; statistical data: Tab.S1), corresponding to 121, 67.5 and 52.7 pg TCDD-EQs/g SEQ, respectively.As the activities refer to dry weight, these observations could be explained by varying levels of organic matter, which would affect the pollutant load.However, the activities were lower than reported for SPM samples from the Rhine and the Neckar River with TCDD-EQ values ranging from 1160 pg/g to peak concentrations of 6640 pg/g during flood events [61][62][63][64].
Chemical analyses of the water samples from the Main river and the effluent sample show the same tendency as the in vitro EBMs (Fig. 7).The cumulative concentration of all contaminants is about 4.2 mg/L in the effluent and 0.6 mg/L at the discharge point M2, which corresponds to a sevenfold dilution.It then decreases greatly, so that the chemical profile of the Main transects downstream of the discharge is similar to the one at the reference site.Particularly striking is the main contaminant hexamethoxymethylmelamine (HMMM; category: polymer additives), which accounts for approximately 95% of the total chemical load in the effluent sample E (3.9 mg/L).It is then strongly diluted downstream, but stabilizes at about 6 µg/L at sites M5 and M6.HMMM is a cross-linker of melamine resins that are mainly used in the automotive industry.Not much is known about the toxic potential of HMMM, but over 21 persistent and mobile transformation products (TPs) have been found recently [65].We assume that the TPs of HMMM also occur in high concentrations in the Main River.However, the TPs were not analyzed in the target screening due to the lack of corresponding standards.After cleavage of all side chains, melamine remains.Its high persistence, mobility and toxicity make it important on a global scale.Lütjens et al. (2023) [66] detected melamine in about 90% of the European surface waters studied, with a concentration of 1.4 µg/L measured in the Main River in Frankfurt.They further identified the production of melamine-containing products as a major pathway for the presence of melamine in surface waters.The second most common pollutant in the effluent sample E is benzotriazole (BT; category: personal care and household) with a concentration of 134 µg/L.BT is completely diluted in the Main River, so that background concentrations of about 1 µg/L at the reference site are immediately found again downstream of the discharge.Including 5-methyl-1-H-benzotriazole (5M-1H-BT), the background levels in the Main River were about 1.5 µg/L.BTs are corrosion inhibitors and are widely used in industry and households, e.g., in dishwashing agents.They are high-volume production chemicals that are regularly detected in rivers around the world, with reported concentrations up to 7 µg/L [67].They are highly soluble in water and can be toxic to aquatic organisms [68].Excluding the main contaminants HMMM and BT with 5 M-1H-BT, the effluent sample consisted mainly of components of the categories personal care and household as well as biocides.Besides BT and 5 M-1H-BT, the most common contaminant in the Main River (M1, M3-M6) was the sweetener sucralose with constant concentrations of around 1 µg/L.Regardless of the industrial effluent, all SPM samples revealed a cumulative concentration of up to 8 µg/g, which are among the highest compared to the other sediment and soil samples (Fig. 9).The detected compounds in the SPM samples were mainly PAHs (up to 2.7 µg/g) and polymer additives (up to 1.8 µg/g).A decrease in intermediates is accompanied by an increase in POPs (mainly 1,2,4-trichlorobenzene).The main contaminant HMMM was found in the SPM samples in comparably low but increasing concentrations after the WWTP reference site (M1p: 0.23 µg/L, M5p: 1.5 µg/L, and M6p: 1.4 µg/L).
Detailed information on the results of the chemical analysis is provided in the supplementary material.
In conclusion, based on the results of the water analysis, we found that the local industrial plant represents a point source for pollutants that, at the time of sampling, caused baseline toxicity, mutagenic and dioxinlike activities.However, the dilution effect in the Main River is much greater than in smaller rivers [56], so environmentally relevant activities in the water phase can only be observed at the discharge point and, in case of mutagenicity, additionally 700 m downstream.We found no critical in vitro activities in the water samples at the inflow level of the future tributary.Further, the effluent does not seem to have an enhancing effect on the measured in vitro activities in the SPM samples.In particular, estrogenic and dioxin-like activities do not increase but decrease downstream of the treatment plant.This could be due to exchange processes, as the chemical analysis indicates a shifting contamination profile as well.Assuming that SPM pollution levels are similarly high over time, their sedimentation may have a negative impact on floodplain development.This was shown, for example, by Schulze et al. [61] who chemically and biologically analyzed SPMs and compared frequently versus infrequently inundated floodplain soils.However, measurements of the present study depend on single grab samples and SPM samples covering 35 days, so that a more comprehensive approach with Fig. 7 Contamination in water samples (w) from the Main River (M) and the WWTP effluent sample (E).M1 represents the reference site (see Fig. 2 and Table 1 for sampling site details).a Cumulative concentrations of all categories without the main contaminants hexamethoxymethylmelamine (HMMM) (b), benzotriazole (BT) and 5-methyl-1-H-benzo-triazole (5M-1H-BT; hatched part) (c) repeated sampling campaigns would be necessary for general statements.

Influence of former industrial effluents on Main River sediments
To also reflect the influence of the local industrial plant over the past years and to identify potential legacy contamination in the Main River (M), we tested surface sediment and additional core sediment from the restored bank flattening (M6) (Fig. 8).
In the transect samples downstream of the reference site, we found varying baseline toxicities that correlate with the respective sediment types we found at each side (Fig. S2).The sandy sediments M4ss and M6ss showed high baseline toxicities with EC 50 of 7.89 and 7.20 mg SEQ, respectively.The core sediment from the restored site was less active than the surface sediment (EC 50 of 24.7 mg SEQ).Since the reference site with comparable sediment characteristics showed low baseline toxicity (EC 50 of 63.6 mg SEQ), we suspect an influence of the WWTP effluent.Clayey sediments showed no (M2ss) or weak (M3ss and M5ss) activities (Fig. 8a).All samples showed dioxin-like activities above the LOQ (0.17 µg β-NF/g SEQ).Compared to the reference site, it was increased at all sites downstream of the WWTP discharge.We observed the highest activities in the Fig. 8 In vitro activities in suspended particulate matter (SPM; p), surface (ss) and core sediment (sc) samples from the Main River (M) (see Fig. 2 and Table 1 for sampling site details).a Baseline toxicity in the Microtox assay is expressed as mean and SEM of the EC 50 based on mg sediment or SPM equivalents (SEQ).Non-toxic samples were set to 100 mg SEQ.b Mutagenicity in the Ames fluctuation assay is expressed as % revertants with > 20.8% is considered mutagenic.Sample concentration in the test was 67.7 mg SEQ/mL.Due to cytotoxicity, sample M6p was tested at 26.7 mg SEQ/mL (*).The triangles represent the results for point mutations in the YG1042 strain, and the diamonds those for frameshift mutations in the YG1041 strain.Filled symbols indicate mutagenicity with S9, open symbols mutagenicity without S9.c Estrogenic activity in the YES is expressed as mean and SEM in EQ of the positive substance E 2 .d Activity in the YDS is expressed as mean and SEM in EQ of the positive substance β-NF.c, d The dotted line represents the LOQ.Statistical significances are provided in Table S1.For abbreviations see Fig. 3 sandy river sediments at sites M4 (8.23 µg β-NF-EQ/g SEQ, corresponding to 89.7 pg TCDD-EQ/g SEQ) and M6, with a decreasing activity from the surface (18.3 µg β-NF-EQ/g SEQ, corresponding to 135 pg TCDD-EQ/g SEQ) towards the core sediment (4.26 µg β-NF-EQ/g SEQ, corresponding to 64.1 pg TCDD-EQ/g SEQ) of the restored site (Fig. 8d).These results are consistent with the expected higher sedimentation rate in a bay located on a sliding slope.The dioxin-like activities found in this work are similar to those reported by Otte et al. for the Elbe river [69], and about 10-to 50-fold lower than measured at "highly contaminated" sites along the Elbe and Danube river [70,71].However, they are orders of magnitude higher than results from less polluted rivers such as the Nidda or the Horloff [18].Mutagenic potential was found only in the YG1041 strain (31.8% revertants) after metabolic activation in clayey sediments directly at the discharge site (M2) (Fig. 8b).Elevated estrogenic activities above the LOQ (0.70 ng E 2 -EQ/g SEQ) were detected in almost all samples downstream of the reference site (0.06 ng E 2 -EQ/g SEQ), with some being statistically significant to very significant (Fig. 8c; statistical data: Tab.S1).The highest activity with 4.91 ng E 2 -EQ/g SEQ was detected in clayey sediment from site M5.In general, also with respect to the mutagenicity at site M2, it can be assumed that the remobilization rate from clayey sediments is low, which minimizes their influence on the development of the floodplain.However, the restored site M6 showed comparably high estrogenic activities as well, with 2.9 ng E 2 -EQ/g SEQ in the surface and 2.31 ng E 2 -EQ/g SEQ in the core sediment (Fig. 8c).These values are slightly higher than reported for other European rivers using similar assays and extraction methods [72,73].
Sediments act as a sink for hydrophobic and persistent compounds and can therefore reflect a long period of pollution.Accordingly, the contamination of the Main River sediments (Fig. 9) indicated that the effluent of the local industrial plant was an important point source for such pollutants in the past.For example, PAH contamination is higher in the sandy sediments downstream of the WWTP.Compared to sediment samples that showed similar dioxin-like activities, the cumulative PAH concentrations found in this work are significantly higher with up to 5.1 µg/g [69].In addition to atmospheric deposition, which is the typical origin, PAHs can also enter the environment through wastewater discharges, as they also have industrial uses [74,75].Phenanthrene, for example, shows the highest concentration of all PAHs in Main River sediment with 2.5 µg/g at site M4.It is used for the synthesis of dyes [76].The second most common PAH in the river sediments is fluoranthene with up to 0.4 µg/g, which is used for the synthesis of pharmaceuticals [77].However, combustion and atmospheric deposition or unburned fossil fuels are typically the main source of PAHs in the environment.Like most PAHs, both phenanthrene and fluoranthene are rated as very persistent and very bioaccumulative and as Substances of Very High Concern.
Ultimately, sediments from the restored bank were particularly conspicuous in the in vitro EBMs and the chemical analysis.We detected decreasing effects from the surface to the core sediment for baseline toxicity, as well as estrogenic and dioxin-like activity.As the boulders were replaced with sand in 2014, there has been considerable contamination of the new river sediments within the last six years.We therefore hypothesize longterm accumulation of contaminants from the Main River in the tributaries as well.The good chemical status of sediments from the tributary may consequently reflect its short existence.

Conclusion
Based on a series of in vitro EBMs and chemical target screening, we aimed to assess the current ecotoxicological status of newly established waterbodies in a floodplain restoration site along the Main River (Frankfurt am Main, Germany) and to estimate its development potential with respect to the influence of the local industrial plant and potential legacy contamination.
We found high mutagenic potential in the upper soil horizons of frequently inundated areas of the floodplain and suspect that these are due to legacy contaminants from aniline and azo dye production in the past.To further confirm this hypothesis, chemical analysis of respective residues would be useful, as well as comparative studies of soil samples from less frequently flooded areas.We emphasize that remobilization of mutagenic contaminants should be considered in future construction work.Contrary to our hypothesis, we found that both water and sediment of the tributary and the groundwater-fed ponds showed negligible activities in the in vitro EBMs and low total chemical contamination.Based on our studies, we classify their ecotoxicological status as good.However, consistent with our hypothesis, we identified two flood depressions near the Main River as hot spots of contamination.Chemical analysis revealed high PAH concentrations as potential driver for dioxin-like activities.We conclude that legacy contamination from past flooding exclusively affects the upper soil layers so that only the shallow flood depressions are impacted.The analysis of a recent WWTP effluent sample showed distinct activities in most in vitro EBMs, which identifies the local industry as a point source of contaminants.However, the effluents were strongly diluted in the Main River, so that at the level of the restored floodplain, activities remained below ecotoxicologically relevant thresholds.Contrary to our hypothesis, the recent industrial discharge had no adverse effect on in vitro activities in SPM.Respective chemical analysis showed consistently high total contamination profiles, both upstream and downstream of the industrial discharge.We recommend long-term sampling approaches to assess the impact of the current industrial discharge.As we hypothesized, historical activities of the local industrial plant are reflected in Main River sediments.Even within the last six years, contaminants have accumulated in the sediments of a restored bank flattening, suggesting that pollution of the Main River may also adversely affect sediment quality in its tributaries in the long term.Although this work is a case study, it has a significance that goes beyond the specific region.The pollution situation we found is typical for other rivers and their floodplains in industrialized Central Europe and is therefore relevant for upcoming restoration projects there.
We have further confirmed the suitability of in vitro EBMs for the identification of both chemically and ecotoxicologically relevant sites.The exhaustive extraction method we have chosen in this study is well suited to identify hot spots of contamination.In addition, from a protective point of view, it is important to consider the "worst case scenario" when dealing with remobilization of pollutants.Such scenarios are becoming increasingly important, especially in times of climate change, as extreme weather events are becoming more frequent [78].A recent example from Central Europe is certainly the flood in summer 2021, where lowland river floods transported enormous amounts of contaminated sediments [79].To assess adverse effects on local species, however, designated relevant sites should be investigated for contaminant exposure and bioavailability (e.g., passive sampling, field studies, sediment contact testing) and linked to ecological monitoring data.

Fig. 1
Fig. 1 The floodplain restoration site in Frankfurt am Main. a The utilization structures until 2012 included meadows (light green), forest structures (dark green), sport fields (turquoise green), conventionally farmed fields (orange) and a few trails for visitors.b The restoration plan of the floodplain includes two artificial tributaries (675 m and 1700 m long), oxbow ponds, bank flattening (western part of the area), meadows (light green), floodplain forests (green), conventionally farmed fields (brown) and several trails with bridges for visitors.Grey arrows indicate the direction of flow.Figure modified from Beuerlein and Baumgartner Landscape Architects[23]

Fig. 6
Fig. 6 In vitro activities in water samples (w) from the Main river (M) and a 24 h composite sample of the wastewater treatment plant effluent (E) of the on-site industrial facility (see Fig. 2 and Table 1 for sampling site details).a Baseline toxicity in the Microtox assay is expressed as mean and SEM of EC 50 based on the REF of the samples.Non-toxic samples were set to 100 mg REF.b Mutagenicity in the Ames fluctuation test is expressed as % revertants with > 20.8% revertants being considered mutagenic.Sample concentration in the test was 6.7 REF.Due to cytotoxicity, sample E was tested at 16.7 REF (*).The triangles represent the results for point mutations in the YG1042 strain, and the diamonds those for frameshift mutations in the YG1041 strain.Filled symbols indicate mutagenicity with S9, open symbols mutagenicity without S9.Samples were tested in a 16.7-fold, sample E in 6.67-fold (*) enrichment.c Estrogenic activity in the YES is expressed as mean and SEM in EQ of the positive substance E 2 .d Activity in the YDS is expressed as mean and SEM in EQ of the positive substance β-NF.c, d The dotted line represents the LOQ.Statistical significances are provided in TableS1.For abbreviations see Fig.3

Fig. 9
Fig.9 Contamination in suspended particulate matter (p; SPM), surface (ss), and core sediment (sc) from the Main (M) River.M1 represents the reference.For sampling site details see Fig.2and Table1

Table 1
[43]ling sites additional purification step was previously shown to have no substantial impact on the effects and cytotoxicity in selected bioassays[43].To ensure comparability of the activities with chemical analyses, we used the purified extracts for bioassays as well.For the clean-up step, extracts were diluted in 1 mL of dichloromethane (DCM, SupraSolv ® , Cat No. 1.00668.2500,GC-MS grade) and passed through chromatography columns (Chromabond ® Flash RS 4 SiOH, PP, 40-63 µm, 4 g, REF: 732,800, Macherey-Nagel) that were previously conditioned with about 9 mL DCM using an Agilent 1200 LC pump.If necessary, samples were treated in an ultra-sonic bath to completely dissolve or suspend the extract in DCM.Purification was performed with 15 mL each of DCM and methanol at a flow rate of 8 mL/min.For particularly contaminated samples, the flow rate was adjusted to 2 mL/min to reduce the backpressure of the column.For GC-HRMS F2One of two flood depressions(1-year flood) close to the riverbank (built in 2014) 50.073667°N, 8.766500°E w, ss T1 Small tributary (built in 2019); shallow spot in 1st basin close to the inflow 50.107417°N, 8.769450°E w, ss T2 Small tributary (built in 2019); narrow spot with higher flow velocity between the basins 50.058667°N, 8.772233°E w, ss T3 Small tributary (built in 2019); shallow spot in 2nd basin close to the confluence into Main 50.053667°N, 8.770283°E w, ss T4 Site within a flood depression (5-year flood) along which the future tributary will be built 50.076333°N, 8.769150°E h R1 Wet meadow in the riparian area (5-year flood); area of a future floodplain forest 50.054167°N, 8.768750°E h analysis, 10% (vol.)from the DCM fraction was blown down under a nitrogen stream before 500 µL of ethyl acetate was added for solvent exchange.In case of precipitation after storage at − 21 °C for 1 h, samples were passed through 0.2 µm polytetrafluoroethylene (PTFE) syringe filters (Phenex ™ -PTFE 15 mm Syringe Filters, particle retention: 0.2 µm, Part No. AF0-2202-52, Phenomenex).
The luminescence inhibition of water samples is expressed as a 50% effect concentration (EC 50 ) based on the relative enrichment factor (REF) of the native water samples.For example, an EC 50 value of 10 would indicate that a tenfold enrichment of the sample is required to achieve 50% luminescence inhibition.EC 50 values for sediment, soil, and SPM samples refer to mg SEQ of the respective extract.For non-toxic samples, EC 50 values were set to 100 to allow mapping of these samples on the graph.The final results are based on three independent tests (n = 3) with two technical replicates each.