Potential and risks of water reuse in Brandenburg (Germany) – an interdisciplinary case study

Exploiting wastewater as irrigation water to counteract water stress is a widely applied option for water reuse in arid regions of the world (Shoushtarian & Negahban-Azar 2020). This is already being implemented extensively, for example in Israel (Reznik et al. 2017) or California (Cooley & Phurisamban 2016). The use of reclaimed water (RW) counteracts competition among agriculture irrigation, drinking water supply, and industrial use. Hence, stressed fresh water resources are thought to be protected (Jaramillo & Restrepo 2017; Lavrnić et al. 2017; Schwaller et al. 2021), and benefits in terms of ecological costs can be accounted for (Passarini et al. 2014).

Water reuse comes along with potential hazards to human health through pathogen contamination of the irrigation water (Amorós et al. 2010; Amahmid et al. 2023). However, decades of experience and the development of quantitative risk assessments demonstrated water reuse to be a secure practice (ISO 2020; Wencki et al. 2020). If multibarrier principles are applied, which are composed of natural (e.g. reservoirs, soil passage), technical (e.g. filtration, disinfection), and administrative systems (e.g. risk assessment in the catchment, monitoring of connections to the sewage network), microbiological risks can be minimized effectively (Mohr et al. 2020).

Anthropogenic trace substances such as pharmaceuticals, pesticides, or industrial chemicals, posing long-term chemical risks, are less regulated (Helmecke et al. 2020; Shoushtarian & Negahban-Azar 2020). Most of these organic micro-pollutants (OMPs) are only insufficiently eliminated by wastewater treatment plants (WWTPs) with secondary treatment using activated sludge for reductions of chemical oxygen demand, nitrogen, and phosphorous (Margot et al. 2015; Alygizakis et al. 2020). Hence, persistent OMPs remain in secondary treated wastewater or irrigation water, respectively, and can accumulate in soil and plant material or contaminate groundwater, posing risks to ecosystems and human health. Dependent on OMPs' intrinsic molecule properties, composition of the soil, cultivated plants, and environmental and climatic conditions of a region, the fate of individual OMPs in environmental compartments is determined (Wu et al. 2015). The pathways are mainly divided in transfer of OMPs from soil to groundwater and from soil to plant, thus potentially entering drinking water sources and the food chain.

Ben Mordechay et al. (2021) analyzed the concentrations in RW for irrigation, soil, and commercially grown crops for 65 OMPs at 400 fields in Israel and reported that primarily OMPs' concentrations in RW determine the levels detected in soil and agricultural products. They found that green leaves of RW-irrigated plants obtain higher OMP concentrations than harvest organs of root crops (e.g. carrot and potato) or harvested fruits (e.g. tomato and banana), as also reported in previous studies (Wu et al. 2015; Christou et al. 2019).

The antiepileptic carbamazepine (CBZ) is a wastewater indicator substance (Jekel et al. 2015) and is well studied for plant uptake (Shenker et al. 2011; Marsoni et al. 2014; Chuang et al. 2019). The bioconcentration factor (BCF) and the root concentration factor (RCF) are defined as ratios between OMP concentrations (ng g⁻¹) in plants or roots, respectively, and in soil (McKone & Maddalena 2007) and are reported for CBZ between one and several hundreds. CBZ is, therefore, recommended as a reference for every study on this subject (Wu et al. 2015). Furthermore, CBZ can be transformed by soil micro-organisms and plants (Christou et al. 2019). The transformation product (TP) 10,11-epoxycarbamazepine accumulates in leaves and is a potentially genotoxic compound with an acceptable daily intake factor of 1,000 lower than that of the parent compound CBZ (Malchi et al. 2014; Paz et al. 2016). Non-ionic substances have been reported to transfer into plants more easily (Malchi et al. 2014), while ionic substances are probably retarded in the phloem (Goldstein et al. 2014).

Human biomonitoring studies in Israel (Schapira et al. 2020; Ben Mordechay et al. 2022) revealed elevated CBZ concentrations in urine of men, women, and children, most of them being vegetarians consuming vegetables irrigated with RW. Schapira et al. (2020) pointed out the need to resolve the contradiction between recommended daily vegetable consumption and the chemical risks of RW irrigation.

Risk assessments often determine the thresholds of toxicological concern (TTC) with differing results depending on OMP concentrations and irrigated crops. On the one hand, negligible health risks are reported for CBZ (BCF > 100) in maize, rice, ryegrass, and wheat (Delli Compagni et al. 2020) as well as for diclofenac (BCF > 100 after 3 years) in tomatoes through uptake from RW after tertiary treatment and disinfection (Christou et al. 2017). On the other hand, comparable low CBZ concentrations (0.5 μg L⁻¹) lead to elevated 10,11-epoxycarbamazepine concentrations in the leaves of carrots and sweet potatoes. The TTC value of 62.5 ng kg⁻¹ for children of 25 kg was found to be exceeded by eating 25 g of carrot leaves or 90 g of sweet potato leaves per day, which is common in parts of Asia and Africa. This does not imply direct toxic effects but indicates the need for detailed toxicity analysis on this OMP. Moreover, there are other OMPs, e.g. lamotrigine with a TTC value of 2.5 ng kg⁻¹ for children that is exceeded by eating half a carrot (60 g) per day (Malchi et al. 2014).

Previous studies conclude that more research is highly recommended, regarding different agricultural products, substances, soils, and climatic conditions to properly manage related risks (Helmecke Fries & Schulte 2020). The European regulation on minimum requirements for water reuse is applied since June 2023 (EU 2020) and is in line with worldwide regulations and guidelines on water reuse providing no restriction for OMPs in spite of recommending their consideration in the site-specific risk management (Shoushtarian & Negahban-Azar 2020).

The obtained interdisciplinary perspectives are interconnected through the same sandy soil and RW used for irrigation, which allows for common evaluation of separately analyzed pathways soil to groundwater and soil to plant. Since the focus is first on potential water sources and second on the risks of OMP transfers, agricultural yields are not assessed and no nutrient deficiency experiments are conducted.

Wastewater generated in Brandenburg is classified in centralized (sewage network and WWTP) and decentralized disposal, subdivided in cesspit tanks and septic tanks (small-scale WWTP). Respective amounts are officially reported for 2019 (MLUK 2021) and a mean water consumption of 106 L d⁻¹ and person (Destatis 2018) was used for calculations. The actual use of irrigation water was reported lastly for 2009 (Berlin-Brandenburg 2012) and, therefore, needed to be projected based on the agriculturally used area in 2019 (Destatis 2021). Calculations were done for the growing season (April–October) and for the whole reference year 2019.

Furthermore, fertilizer application was analyzed for potential substitution by remaining nitrogen (N) and phosphorous (P) in RW. Due to lack of available data, the mean value of domestic sales volume in 2019–2021 was used to estimate applied fertilizer amounts (Destatis 2018). Human excreta are calculated with 11 g N and 1.8 g P contents per day and person.

RW was taken from the effluent of a large WWTP in Berlin with secondary treatment and UV-disinfection between May 1 and September 30. Each batch was analyzed for OMP concentrations immediately after collection from the WWTP, and the stored RW was reanalyzed to evaluate potential OMP losses. Accompanying chemical parameters for RW are summarized in Table 1.

Table 1

Characteristic parameters for RW used for irrigation

.	Base parameters .		Anions .				Cations .
	pH .	EC .	.	.	Cl− .	.	K+ .	Na+ .	Ca2+ .	Mg2+ .
	.	μS cm−1 .	mg L−1 .				mg L−1 .
n	5	5	11	15	3	6	15	15	15	15
Mean	7.5	1,212	37.8	0.59	155.4	110.6	84.9	28.5	11.3	112.7
SD	0.2	66	2.9	0.21	7.7	7.9	16.0	1.9	1.2	7.3

.	Base parameters .		Anions .				Cations .
	pH .	EC .	.	.	Cl− .	.	K+ .	Na+ .	Ca2+ .	Mg2+ .
	.	μS cm−1 .	mg L−1 .				mg L−1 .
n	5	5	11	15	3	6	15	15	15	15
Mean	7.5	1,212	37.8	0.59	155.4	110.6	84.9	28.5	11.3	112.7
SD	0.2	66	2.9	0.21	7.7	7.9	16.0	1.9	1.2	7.3

Notes: n is the number of measurements of RW batches, EC is the electrical conductivity representing the salinity of RW, and SD is the according standard deviation.

View Large

The sandy soil was taken from an agricultural field in Brandenburg (north east Germany, 52°16′34.3″N 13°04′39.7″E), which is located in a region that was formed in the last ice age as a terminal moraine. The soil material was taken from four different layers (Table 2), with layer A representing the top soil of horizon A and all subsequent layers representing the mineral horizon.

Analyzed arugula was harvested after 65 days at BBCH¹ stage 59. Harvested plant material was dried at 30 °C and homogenized afterward with a ceramic ball mill for further analyses. Soil samples were taken from each vegetated pot immediately after harvest and were air dried.

The water samples were stored at −20 °C until sample preparation, then centrifuged for 10 min at 5,000 rpm at room temperature and for an additional 10 min at 13,000 rpm at 4 °C, and subsequently analyzed by direct injection.

The solid–liquid extraction for soil and plant samples was adapted from a method by Riemenschneider et al. (2017). The roots and leaves of the arugula plant were processed individually by weighing 0.5 g of soil or plant material in a 15 mL centrifuge glass. Extraction was performed twice using a mixture of methanol and Milli-Q water (1:1, v:v). Following the protocol of Riemenschneider et al. (2017), an aliquot of the two combined supernatants of both extractions were centrifuged for 10 min at 13,000 rpm and 4 °C.

In lysimeter studies, OMPs were analyzed by high-performance liquid chromatography (HPLC) coupled to a triple-quadrupole mass spectrometer (MS/MS) as described in more detail by Zeeshan et al. (2023). The analyzed OMPs were quantified using an internal calibration with isotopic labeled standards.

OMP analyses for the pot experiments were performed by using supercritical fluid chromatography (SFC) (Waters Acquity UPC2 system) coupled to a triple-quadrupole mass spectrometer (Waters TQXS) using a Ethylene Bridged Hybrid (BEH) column based on a method by Schulze et al. (2020). For quantification of CBZ and DZA in RW samples, external calibration was used. The apparent recoveries were included by spiking the samples with the reference standards. For soil and crop samples, matrix-matched calibrations using non-treated soil and arugula material were used. In addition, for soil samples, apparent recoveries were included analogous to the water samples.

About 11% of Brandenburg's inhabitants are not connected to sewage networks. A total of 75,000 use septic tanks for wastewater disposal and 205,000 use cesspit tanks for wastewater storage. Vacuum trucks are regularly collecting the stored wastewater and transport it several kilometers to the next sewage network connection. These decentralized sources for RW alone could theoretically supply one-fourth (28%) of irrigation water needs in 2019 (Figure 4). This may offer potential for innovative point-of-use reuse concepts fitting to local socio-economic structures (UBA 2021).

Furthermore, RW add loads of nutrients to the field and can therefore substitute parts of fertilization. We estimate that about 68,500 t N (as ammonia) and 3,400 t P (as phosphorus pentoxide) are applied to the fields in Brandenburg each year. With regard to the untreated wastewater in cesspit tanks, amounts of 4,300 t N and 1,000 t P accrue during a growing season, which could substitute theoretically a maximum of 6% N fertilizer and 28% P fertilizer. With respect to much lower concentrations in the effluent of WWTPs (exemplarily calculated with data from Table 1), only 145 t N and 150 t P with substitution potential of 0.2% N fertilizer and 4.4% P fertilizer, respectively, would have been reached by replacement of the irrigation water with RW in 2019.

The calculated (maximum) potentials for irrigation water substitution appear to be promising. The concomitant fertilizer substitution potentials by dissolved nutrients in RW are a positive side effect and worth considering in fertilization planning. At the same time, negative effects on plants could arise from salts contained in RW (Ofori et al. 2021). An electrical conductivity (EC) of 1,212 μS cm⁻¹ found in the analyzed RW (cf. Table 1) indicates a slight to moderate restriction for salt-sensitive plants in arid or semi-arid regions (Ayers & Westcot 1985). However, this does not apply for our case study region and, therefore, EC is not restricted. The perspective of irrigation water demand and available RW sources indicates water reuse as a viable option for Brandenburg to counteract water scarcity and competition. Putting this into practice would require further consideration of infrastructure, transportation, and water treatment, especially for the decentralized sources. However, risks by dissolved contaminants (OMPs) need to be assessed with another perspective.

DZA and CBZ concentrations in the used RW batches for irrigation ranged between 0.6 and 3.3 μg L⁻¹ and 0.3 and 1.4 μg L⁻¹, respectively (Figure 5, top ‘RW’). After the vegetation period of 2021, the seepage rates increased and DZA was successively transported down the soil profile, clearly visible by the temporally shifted concentration peaks along the lysimeter depths (Figure 5, z = 10–90 cm). The concentration decreases with increasing soil depth due to diffusion and hydrodynamic dispersion. Five months after the start of RW irrigation, DZA reaches the lower boundary (i.e. the effluent) of the lysimeters (Figure 5, bottom). The applied DZA mass, from July to August 2021, has left the lysimeters at the lower boundary until April 2022. These observations suggest that neither sorption nor degradation play an important role for the fate of DZA in sandy soils. These findings are supported by our observations in laboratory batch experiments, where neither degradation nor sorption was observed (data not shown). Moreover, a similar behavior of DZA was found in other studies. Ternes et al. (2007), for example, observed a very persistent and mobile behavior for DZA in lysimeters also filled with a sandy soil and suggested that DZA could be used as conservative wastewater tracer in soils and groundwater. Kalsch (1999) and Haiß & Kümmerer (2006) investigated biodegradability of DZA in activated sludge and also found that DZA is persistent.

Interestingly, the DZA concentration peaks in the upper layers (up to 40 cm) during the vegetation period in 2022 are roughly two times higher than that in RW concentrations. This phenomenon is explained by high evapotranspiration in the vegetation period (Figure 5, green area in the top subplot ‘RW’), when water leaves the system due to vaporization of root water uptake and DZA remains in the soil. In particular, the increased concentration in 40 cm depth suggests that DZA is not or only partly taken up by the barley plants.

Generally, CBZ concentrations were very low in all depths (Figure 5, z = 10–90 cm). Solely in 10 cm depth, values above the limit of quantification (LOQ) of 5 ng L⁻¹ were detected during the irrigation periods, due to macropore transport. These findings suggest that CBZ is either strongly sorbed to the mineral or organic surfaces and/or microbially transformed within the first few centimeters of the soil. Hence, none of the CBZ applied with the irrigation water reached the effluent. This assumption is consistent with conducted laboratory batch experiments under aerobic conditions, in which especially sorption reduced the concentration of CBZ in the soil solution (data not shown). Several other studies reported minor biodegradability for CBZ under aerobic condition classifying it as a persistent substance in the vadose soil zone (Li et al. 2013; Grossberger et al. 2014; Thelusmond et al. 2018); however, König et al. (2016) reported anaerobic transformation of CBZ during bank filtration. Ternes et al. (2007) observed that sorption and/or biodegradation took place to a certain extent during the passage of CBZ through lysimeters. However, in their experiments, degradation/retardation of CBZ during the soil passage was relatively low and CBZ was detected in all sampled depths of their lysimeters. Similar to the findings of Ternes et al. (2007), Paz et al. (2016) also detected CBZ in all depths of their lysimeters (1 m depth), one filled with a sandy loam and another with loamy sand soil. CBZ sorption is known to be mainly governed by soil organic matter (Paz et al. 2016), whose quality might change at different pH values (e.g. due to protonation and deprotonation of functional groups). Thus, the contradictive findings may be explained by soil pH, since Ternes et al. (2007) and Paz et al. (2016) reported neutral pH values and our examinations show slightly acidic soil pH (Table 2). Accordingly, the environmental fate of CBZ can differ strongly due to site-specific effects related to soil properties.

To this state, no threshold concentrations or maximum loadings exist for OMPs regarding the soil to groundwater transfer pathways in German legislation². However, the regulatory aspects of RW for irrigation are frequently discussed, and environmental and health quality criteria with regard to OMP ought to be developed (Helmecke et al. 2020).

DZA was found in highest concentrations in arugula roots between 2,400 and 6,700 ng g⁻¹ dry weight (d.w.), whereas much lower concentrations of around 70 ng g⁻¹ d.w. were quantified in arugula leaves. This is also shown by the calculated root and BCFs, demonstrating that DZA is taken up by the roots but not transported to the leaves (Table 4). However, as discussed elsewhere (Castan et al. 2023), anionic chemicals like DZA are only partially absorbed due to the negatively charged cell membranes of the roots and the resulting repulsive forces. This implies that DZA may not pass through the root epidermis and is only attached to the root surface. In contrast, concentrations in the soil material were significantly lower around 2 ng g⁻¹ d.w. This demonstrates the poor sorption potential to soil surfaces as well as the mobile character of DZA transporting the compound with the water toward the plant. These data are in line with the results from the lysimeter studies above, showing that DZA tends to be strongly transferred vertically with the water and can potentially reach groundwater (cf. Figure 5).

Another OMP analyzed was CBZ, which is well studied and can thus act as an indicator substance. CBZ was identified in low concentrations in the soil material (<1 ng g⁻¹ d.w.). As reported elsewhere (Scheytt et al. 2006; Martínez-Hernández et al. 2016), sorption of CBZ is low. These results stand in contrast to those shown in lysimeter studies above since a removal of CBZ in the very upper soil layers was shown. This can result from a removal of CBZ by biodegradation in soil and plant material, e.g. TPs of CBZ were reported in several crops (Malchi et al. 2014; Martínez-Hernández et al. 2016; Riemenschneider et al. 2016). In addition, it is shown that through the uptake of RW, increased concentrations of CBZ were found in both plant materials in similar levels at around 71 (arugula leaves) and 32 ng g⁻¹ d.w. (arugula roots). Goldstein et al. (2014) reported comparable results for plant uptake of CBZ irrigated with treated wastewater containing similar concentrations of CBZ. Also, the RCF (70) and BCF (155) for CBZ show an uptake by arugula.

In comparison, these two chemicals showed different uptake capacities by the plant. While for DZA a BCF of 18 (less uptake) was calculated, CBZ has a much higher potential for plant uptake (BCF = 155). This can possibly result from its ionic state, since non-charged molecules, like CBZ (pH 7.5), may more readily cross cell membranes.

In addition, the maximum amounts of fresh arugula that would still be tolerated from a health perspective for the determined concentrations of DZA and CBZ were calculated. Since both DZA and CBZ are classified as Cramer class III (Malchi et al. 2014; Neumann & Schliebner 2019), a TTC value of 1.5 μg/kg bodyweight and per day was used (Kroes et al. 2004). For DZA (c_f.w. = 6.68 ng g⁻¹), an adult with 70 kg bodyweight could eat 16 kg of arugula per day, while for CBZ (c_f.w. = 11.80 ng g⁻¹), 9 kg arugula per day could be eaten to remain below the threshold. These values are comparable to the ones from Malchi et al. (2014). The amounts stand in no relation to the daily amount of fresh arugula a person normally consumes and thus are negligible for an adult from a single substance perspective.

In this study, we showed interdisciplinary examinations on water reuse providing perspectives that are necessary to elucidate potentials and risks. Our case study region, Brandenburg (Germany), possesses centralized and decentralized sources for RW. Effluent from centralized WWTPs could substitute 245% while decentralized cesspit and septic tanks could substitute 28% of water used for agricultural irrigation in 2019. Furthermore, fertilizers could partly be substituted by nutrients in RW from cesspit tanks, and storage options during winter time may increase the potential. Water reuse is, therefore, a viable option to mitigate competition for water in Brandenburg. However, implementation requires further consideration of infrastructure, transportation, and water treatment, especially for decentralized sources. As a limitation, it must also be considered that minimum discharge from WWTPs may be a requirement for streams.

The perspectives of OMP transfer pathways from local sandy soil to groundwater and from the soil to plant were examined by lysimeter studies and greenhouse pot experiments. The fate of the comparable large and heavy X-ray contrast media DZA and the antiepileptic CBZ as common indicators for OMP was analyzed. This should allow for a first risk estimation based on realistic OMP concentrations in the RW.

DZA showed tracer-like transport in the lysimeters without retardation posing a high risk for groundwater contamination. Also, DZA was found to accumulate in or on the roots of arugula with a remarkable RCF of 1,925 ± 34% in pot experiments. Probably it is transported with the water flux due to its high mobility but is not able to pass through the epidermis due to its size and negative charge. However, a small part of DZA was quantified in arugula leaves with a BCF of 18 ± 22%.

CBZ seems to be removed in the very upper soil layers in the lysimeter studies. It, therefore, poses no risk for groundwater contamination for the sandy soil of our case study. However, other studies showed vertical CBZ transfer in sandy soil that potentially indicates a long-term risk also for groundwater in Brandenburg. Contradicting, CBZ could not be found in greater concentrations in the soil material of the pot experiments demonstrating (a) no or barely any sorption to it or (b) probably transformation. However, CBZ showed an uptake by arugula with 70 ± 7% RCF and 155 ± 12% BCF indicating a potential transfer toward plants.

While bringing together these perspectives, water reuse with RW can be a promising option for agricultural irrigation in Brandenburg, if water quality is managed with regard to OMP and potential transfer pathways. The two examples, DZA and CBZ, showed a wide range of possible environmental fates and related risks due to substance-specific properties. Comparison with literature data showed that soil properties (inorganic, organic, and microbiological) together with cultivated plants alter OMPs' fate.

For a sound risk management, fate predicting tools for wastewater treatment, the pathway soil to groundwater, and the pathway soil to plant would be very helpful. As this has to be based on intrinsic molecule properties, local soil properties, and agricultural product, this is a major challenge, especially since there is a lack of general parameters (e.g. D_OW or D_OC) to predict sorption of potentially ionic OMPs (Sigmund et al. 2022).

However, the load of OMPs in RW is the driving parameter for concentrations found in the environment. It is, therefore, highly recommended to implement advanced wastewater treatment (tertiary treatment). In addition, treatment should be selected based on the case-specific OMPs, which are considered most likely to contaminate groundwater and plants in the area where the water is to be reused.

As shown for DZA and CBZ, polar and mobile as well as non-polar and non-mobile OMPs are transferred into agricultural products. Hence, to address all these aspects, monitoring and extended analyses of OMPs are needed to minimize potential risks.

We thank Silke Pabst for HPLC-MS/MS analyses and Pia Schünemann for sample preparation and peak integration for SFC-MS/MS analyses. We are very grateful to Manuela Helmecke for her critical proofreading of the final manuscript. This investigation was supported by the German Federal Ministry of Education and Research (BMBF) within the project PU₂R (UBA 2021), contract 02WV1564.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.