Bisphenol A in German watersheds: Part II. FlowEQ model‐based characterization of sources and current and future conditions

Increasing scientific and regulatory concern regarding environmental concentrations of bisphenol A (BPA) increases the need to understand the sources and sinks of this chemical. We developed a coupled flow network/fugacity‐based fate and transport model to assess the contribution of different emissions sources to the concentration of BPA in surface water in Germany. The model utilizes BPA loadings and sinks, BPA physicochemical properties, a water flow network, environmental characteristics, and fugacity equations. The model considers industrial emissions, leaching from BPA‐containing articles, wastewater treatment and bypass events, and emissions from landfills. The model also considers different scenarios that account for changes in the usage profile of BPA. Model predictions compare favorably to measured surface water concentrations, with the modeled concentrations generally falling within the range of measured values. Model scenarios that consider reductions in BPA usage due to government‐mandated restrictions and voluntary reductions in usage predict falling BPA concentrations that are consistent with the most recent monitoring data. Model predictions of the contributions from different usage scenarios and wastewater treatment methods can be used to assess the efficacy of different restrictions and waste handling strategies to support efforts to evaluate the costs and benefits associated with actions aimed at reducing BPA levels in the environment. This feature of the model is of particular importance, given current efforts to update the regulations regarding BPA usage in the EU. The model indicates that as the current restriction on BPA in thermal paper works through the paper recycling process, BPA concentrations will continue to decrease. Other actions, such as upgrades to the stormwater and wastewater infrastructure to minimize the frequency of storm‐related bypasses, are predicted to provide more meaningful reductions than additional restrictions on usage. Integr Environ Assess Manag 2024;20:226–238. © 2023 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).


INTRODUCTION
Scientific and regulatory attention has been paid to bisphenol A (BPA) as a substance of potential interest.Therefore, understanding the sources and fate and transport of BPA is critical for understanding the contribution of various sources to the concentrations of BPA in surface water.
Bisphenol A may enter the environment through various industrial and consumer-related pathways (Cousins et al., 2002;Tappert et al., 2023).For the industrial pathways, BPA may enter the environment during the manufacturing of BPA or through secondary industrial uses such as the manufacturing of polymers and other substances based on or containing BPA or the application of BPA-containing coatings.Consumer sources and emissions to the environment, such as those related to the use of consumer products and/or articles containing BPA, may be more important than the industrial sources and emissions.The application of fugacity-based modeling to triclosan highlights the potential significance of consumer products as a substantial source of chemicals (Bock et al., 2010;Lyndall et al., 2010).Sources and pathways may include, but are not limited to, leaching from consumer goods during use, as well as BPA releases occurring during waste management practices like landfilling or incineration.Moreover, consumer-related uses of BPA-containing articles can lead to the release of BPA into the environment through domestic wastewater entering wastewater treatment plants (WWTPs), surface water runoff, or landfills.Industrial and consumer-related sources may discharge BPA to surface water, air, and to a lesser extent, soil.
Multimedia fate and transport models provide a mechanism by which known sources of BPA can be used to predict concentrations in the environment and these predictions can be compared with measured concentrations (Bock et al., 2010;Citra, 2004;Cousins et al., 2002;Mackay, Guiardo, Paterson, & Cowan, 1996;Mackay, Guiardo, Paterson, Kicks, et al., 1996;Mackay et al., 1992).Traditional fugacity-based models consider (1) loadings to various environmental media; (2) partitioning between phases such as water, suspended particulates, air, soil, sediment, and porewater; (3) advective and diffusive transport processes; and (4) removal due to degradation processes and burial in deep sediments.If the model characterizes the important sources, sinks (e.g., removal during treatment, degradation, and burial in deep sediments), and transport processes with sufficient accuracy, then the predictions will compare favorably to measured concentrations (Bock et al., 2010).The objective of this study was to develop a model capable of estimating BPA concentrations in geographical areas where measurements are lacking.Furthermore, the model aimed to predict how alterations in manufacturing, usage, and disposal practices could influence future environmental concentrations of BPA.

MATERIALS AND METHODS
For chemicals discharged to surface water, the processes that control the concentration in surface water are (1) transport and dilution, (2) partitioning between different environmental media (e.g., sediment and surface water), (3) eventual burial in deep sediments, and (4) degradation.We developed the FlowEQ model to combine these processes into a single watershed-scale model.FlowEQ combines the transport processes that are the defining characteristic of the MoRE/ MONERIS model (Modeling of Regionalized Emissions/ MOdelling Nutrient Emissions in RIver Systems) (Umweltbundesamt [UBA], 2010) with the multimedia partitioning and media-specific fate and transport characterization of a fugacity-based model (Bock et al., 2010;Citra, 2004;Cousins et al., 2002;Mackay, Guiardo, Paterson, & Cowan, 1996;Mackay, Guiardo, Paterson, Kicks, et al., 1996;Mackay et al., 1992) into a single framework.The MONERIS model and its successor, the MoRE model, are empirical models that quantify nutrient emissions via point source and diffuse pathways into river systems (Behrendt, Bach, et al., 2002;Behrendt, Dannowski, et al., 2002;Behrendt et al., 2000).In FlowEQ, loadings to the model units serve as the primary input into level III fugacity model equations from Mackay, Guiardo, Paterson, and Cowan (1996) and Mackay et al. (1992).Fugacity models are based on the partitioning of a chemical into air, water, sediment, and soil (Citra, 2004;Mackay, Guiardo, Paterson, & Cowan, 1996;Mackay, Guiardo, Paterson, Kicks, et al., 1996;Mackay et al., 1992).Although equilibrium is not assumed, steady-state conditions are, that is, concentrations do not change because the rate of removal processes such as degradation and advection are equal to the rate of source inputs (Mackay, Guiardo, Paterson, & Cowan, 1996).The fugacity portion of FlowEQ uses the equations from the multimedia fugacity model Equilibrium Criterion (EQC) by Mackay, Guiardo, Paterson, and Cowan (1996).In FlowEQ, three factors determine the partitioning of the chemical in the environment: (1) the media into which it is discharged, (2) the physicochemical properties of the chemical, and (3) the physical properties of the environment.FlowEQ assumes homogeneous mixing throughout each model unit; in reality, conditions can significantly depart from steady-state conditions in and near source areas.In these source areas and at small spatial scales, dilution is the primary process influencing concentrations.The formulation of a level III fugacity model is well defined in previous work (Bock et al., 2010;Cousins et al., 2002;Mackay, Guiardo, Paterson, & Cowan, 1996).FlowEQ is implemented as a series of R programs (R Core Team, 2023) that model watersheds upstream to downstream, tracking and compiling the contributions to surface water associated with different emissions categories.
Following release, the partitioning of BPA between different environmental media is regulated by its physicochemical properties.Table 1 provides a summary of these characteristics.Cousins et al. (2002) surveyed the available data on environmental degradation and reported the distribution of half-lives in air, water, soil, and sediment used in an EU System for the Evaluation of Substances (EUSES) modeling exercise (RIVM, 2004).The half-lives indicate that BPA is biodegradable in environmental media, with half-lives of days to months for water, soil, and sediment.Bisphenol A is rapidly degraded in air by hydroxyl radicals, with a half-life of hours (Cousins et al., 2002;Registration, Evaluation, Authorization, and Restriction of Chemical Substances [REACH], 2010; Staples et al., 1998) (Table 1).

Environmental characteristics
The model units define the smallest semi-independent building blocks of a fugacity model (Citra, 2004;Mackay, Guiardo, Paterson, & Cowan, 1996;Mackay, Guiardo, Paterson, Kicks, et al., 1996;Mackay et al., 1992).Each of the model units requires specific inputs for various environmental parameters (e.g., land area, water area, organic carbon content of the abiotic media, etc.) and discharges (to air, water, and soil).In FlowEQ, the basic model units, or more simply units, are the coordination areas defined in the MoRE/ MONERIS model.The FlowEQ model domain is populated by the coordination areas and is defined as the watersheds in Germany and adjoining countries as characterized in the MoRE/MONERIS model (UBA, 2010) (Figure 1).
In general, the environmental characteristics for each FlowEQ unit were defined based on the values used for the EUSES model (RIVM, 2004) (Supporting Information: Table SI-1).The EUSES model is another fugacity-based model used to evaluate chemical usage and predicts environmental concentrations for REACH registration and authorization purposes.Other characteristics, such as land cover, population, rainfall, and so forth, were based on the MoRE/MONERIS model and data provided by UBA and Statistisches Bundesamt.The derivations of the specific parameters are described in detail in the Supporting Information: Environmental Characteristics of the Model Domain.

Bisphenol A loadings
In addition to defining environmental characteristics, fugacity-based models require detailed information regarding the emissions of BPA to each coordination area for the different emissions pathways (Citra, 2004;Mackay, Guiardo, Paterson, & Cowan, 1996;Mackay, Guiardo, Paterson, Kicks, et al., 1996;Mackay et al., 1992).Bisphenol A emissions sources and loadings to FlowEQ are described in detail in Tappert et al. (2023) and Supporting Information: Emission Sources.For the purpose of this assessment, emissions were classified into the following categories: Primary manufacturing.The EU has no specific reporting requirements for industrial BPA emissions; therefore, this assessment relies on voluntary reporting by the manufacturers.Manufacturing activities that either produced or used BPA in the synthesis of other substances include BPA, polycarbonates, epoxy resins, polyvinyl chloride (PVC), and phenoplastics.Estimates of industrial emissions from these processes were provided by the manufacturers, and account for the period of 2015-2020.The use of 2015 emissions estimates is expected to overestimate current and future emissions due to the general trend of fewer BPA applications in recent years.The individual processes are discussed in detail in Supporting Information: Emission Sources-Primary Manufacturing and the reported or estimated emissions to air and surface water by coordination area, in kg/a, are compiled in Supporting Information: Table SI-4.These emissions differ across the coordination areas and across the various substance pathways.
Bisphenol A was used in the production of thermal paper, which is commonly used for point-of-sale receipts.As of 2 January 2020, BPA cannot be legally placed on the market in the EU in thermal paper at concentrations greater than or equal to 0.02% by weight, compared with an estimated concentration of 1.39% prior to this regulation (Tappert et al., 2023).Thus, thermal paper manufacturing emissions to air and water are assumed to be negligible.However, BPA from thermal paper entering the waste-paper recycling loop is assessed below.Downstream manufacturing.Downstream manufacturing of articles that are composed of BPA or BPA-containing substances represents another pathway through which BPA can be released into the environment.These processes are included in this evaluation and are further discussed in detail in Supporting Information: Emission Sources-Downstream Manufacturing.Supporting Information: Table SI-4 presents total loadings to the environment from the primary and downstream manufacturing emissions pathway, summed across different uses.
Consumer use.The use of BPA-containing articles by consumers may result in the release of BPA into the environment.
There are two main usage patterns considered in this analysis: (1) indoor usage and (2) outdoor usage.The indoor use of BPA-containing articles can result in the release of BPA into wastewater, which may subsequently find its way into surface water.Bisphenol A releases originating from indoor usage enter consumer wastewater and are subsequently transported to diverse wastewater treatment systems, each with varying removal efficiencies.Outdoor usage of BPAcontaining articles may result in the direct release of BPA into surface water.For example, recycled PVC liners and foils are often installed outdoors, and the BPA could leach from the article and enter surface water via stormwater runoff.The derivation of the BPA leached from consumer products and per capita loadings is described in detail in Tappert et al. (2023) and in Supporting Information: Emission Sources-Consumer Use.Briefly, BPA loadings to surface water and wastewater from the consumer use of articles were calculated using concentrations in articles, annual article usage, fraction of BPA leached to water, and the population of Germany.
The per capita loadings calculated in Tappert et al. (2023) were translated to coordination area-specific loadings using coordination area-specific populations presented in Supporting Information: Table SI-2.It is conservatively assumed that BPA leached from outdoor usage of BPA-containing articles directly enters the environment without wastewater treatment and/or BPA removal.For discharges to wastewater, emissions were converted into loadings (kg/year) based on pathway-specific removal efficiencies calculated from WWTP removal efficiencies and bypass events, as discussed below.
Wastewater treatment plant removal efficiencies and bypass events The coordinates of major municipal WWTPs in Germany were provided by UBA (2015) and demonstrate that WWTPs are evenly distributed throughout the country.This finding confirms that consumer wastewater emissions can be modeled as a per capita source for the purpose of this evaluation.An additional advantage of this approach is that the specific locations of WWTPs outside of Germany, but within the model domain, are not required by the model.Wastewater treatment is modeled based on (1) measured BPA removal efficiencies determined in Tappert et al. (2023), in addition to literature values; (2) information provided by UBA on the number of people connected to sewers versus those not connected for each coordination area from 2011, adjusted to account for population growth between 2011 and 2020; and (3) estimates of the average frequency of bypass events based on information from UBA combined with data on the status of stormwater handling infrastructure in each federal state, which was used to estimate the frequency of bypass events for each coordination area.These calculations are summarized below and are described in detail in Supporting Information: Wastewater Treatment.
Removal efficiencies.The WWTP concentrations measured in Tappert et al. (2023) were supplemented with recent (post-2005) data from the peer-reviewed literature (Musolff et al., 2009) and are compiled in Supporting Information: Table SI-5.It should be noted that only data from the 2018 measurement campaign from Tappert et al. (2023) were used for modeling efforts.The WWTPs measured in 2020 and 2022 were a subset of those measured in 2018, and therefore, data from 2018 were used to maximize the comparability across the different facilities.Bisphenol A removal efficiencies were calculated by comparing the average BPA influent and effluent concentrations: BPA removal rate (%) = (1 − effluent/influent) × 100%.The average value of duplicate influent and effluent samples was used to calculate removal efficiencies.
The average removal efficiency including data from all WWTPs, except for anomalous results for No. 9, was 88.9% (Supporting Information: Table SI-5).This result is discussed in more detail in Tappert et al. (2023).The data were further divided based on the plants' dominant influent source(s).Two WWTPs, No. 4 and No. 8, had substantially greater concentrations of BPA in influents compared to other facilities, which can be explained by their influents being dominated by paper industry wastewater (>50%).The removal efficiency of these two paper-industry dominated WWTPs was 98.7%, which was used to model BPA removal associated with treatment of wastewater from paper industries.Influent for WWTP No 13 was from 100% industrial sources and was also excluded from estimation of municipal WWTP removal efficiencies.The remaining 20 facilities (19 values from this study and one from Musolff et al., 2009) were considered representative of municipal WWTPs, and the average removal efficiency of 87.6% was used in the model for treatment of wastewater influents not arising from paper industries.Thus, the municipal-and paper industrydominated removal efficiencies were applied separately to their respective influent sources as discussed below.
A sewage treatment plant model (Bock et al., 2010;Clark et al., 1995), modified to utilize the parameters from the EUSES model to describe European wastewater treatment practices, was used to predict the removal of BPA during primary treatment, prior to aeration, settling, and clarification.The STP model predicts a removal efficiency of 11.6% during primary treatment (5.2% due to settling and 6.4% due to degradation); this value was rounded to 12% and selected as a representative value for the removal efficiency during minimal wastewater treatment (i.e., bypass and minimal treatment, discussed below).
Additional details are provided in Supporting Information: Wastewater Treatment-Removal Efficiencies.
Bisphenol A treatment by coordination area.Umweltbundesamt reported the population connected to a sewer, the population connected to a sewer that connects to a municipal WWTP, and the total population for each analytical unit based on 2011 population estimates.The 2011 values were used to derive coordination area-specific estimates of the population served by various wastewater treatment technologies.For areas outside of Germany, the proportions used in the downstream analytical units from the same coordination area were used as a surrogate for the proportions in the analytical areas outside of Germany.The population statistics for each analytical area were summed to estimate the population served by each treatment technology in each coordination area.
Bypass frequencies were estimated as follows: UBA reported that approximately 15% of the time wastewater supplied to WWTPs is diverted to surface water in the region of Niederrhein (personal communication from UBA staff, December 2020).This value (15%) was used as the "medium" bypass frequency with low (10%) and high (20%) bypass frequencies to other regions of Germany based on infrastructure.Brombach and Dettmar (2016) provide an assessment of the stormwater infrastructure for each federal state, which was used to classify each coordination area (Supporting Information: Wastewater Treatment, BPA Treatment by Coordination Area and Supporting Information: Table SI-6).We assumed that this diverted wastewater was discharged to surface water with minimal treatment (12% BPA removal efficiency).
We assumed that the wastewater that is not directed to a sewer directly connected to a traditional WWTP with high removal is directed to one of the following treatment options: (1) a biological treatment system with an estimated removal efficiency of 60% or (2) discharged to surface water with minimal treatment (12% removal efficiency).Option (2) includes wastewater that is directed to a sewer but is not sent to a WWTP via historical and/or former channel systems (so-called Bürgermeisterkanäle) and is discharged directly to surface water.Data regarding the relative proportions directed to each of these three options have not been identified; therefore, it was assumed that wastewater is evenly divided between these treatment options.Additional details regarding the derivation of the fraction of wastewater subjected to each treatment type and removal efficiency are presented in Supporting Information: Wastewater Treatment-BPA Treatment by Coordination Area and are tabulated in Supporting Information: Table SI-6.The emissions to surface water (discussed below) were estimated based on these treatment pathways and removal efficiencies combined with estimates of the loadings to wastewater.

Loadings from landfills
Landfills.As described in Tappert et al. (2023) and in Supporting Information: Landfills, leachates from 21 participating landfills with different treatment technologies and wastewater streams were sampled and analyzed.The measured data were used as a basis for calculating average BPA loadings of wastewater stream categories, shown in Supporting Information: Table SI-7, and presented below: 1) Onsite treatment + WWTP (before (a.) and after (b.) onsite treatment) 2) No onsite treatment + WWTP 3) Onsite treatment + direct discharge 4) No onsite treatment + direct discharge 5) External (unknown).
Leachate from categories 1, 2, and 5 is discharged to WWTPs, while categories 3 and 4 discharge leachates directly to surface water.It should be noted that leachate from category 4 is not treated onsite or at a WWTP (i.e., 0% BPA removal).
A data set provided by Landesamt für Natur, Umwelt, und Verbraucherschutz Nordrhein-Westfalen (LANUV) reported the number of wastewater streams per category that are present in North Rhine-Westphalia.By combining the information from LANUV and Tappert et al. (2023), it was possible to estimate total BPA loadings from landfills (active and inactive) and the share of each category in North Rhine-Westphalia, as shown in Supporting Information: Table SI-8.For categories that discharge to WWTPs, respective loadings to surface water were calculated based on the removal efficiencies during treatment, as described below.
Even though no precise information, such as location or type of treatment, was available for all of Germany, it was possible to obtain the overall number of active landfills per Federal State from destatis.de (2016).There were 1082 landfills across 15 Federal States, with 0 identified in Berlin.The number of active landfills in each Federal State was used to estimate the number of landfills per coordination area based on surface area (Supporting Information: Table SI-9).The available data were not sufficiently detailed to allow the landfills to be individually classified into the various categories above.Detailed data were only available for North Rhine-Westphalia.Therefore, it was assumed that the share of each category in North Rhine-Westphalia was equivalent to Germany as a whole.
The loadings to WWTPs and surface water by coordination area for each category were calculated using the information provided in Supporting Information: Landfills.The municipal WWTP removal efficiency (87.6%), discussed above, was applied to landfill leachates that were sent for offsite WWTP treatment (i.e., categories 1, 2, and 5), and the same bypass frequencies and lowered removal efficiencies during bypass applied to consumer wastewater were used for landfill leachate.We assumed that landfill leachate was only sent to high removal efficiency WWTPs (87.6% removal).
The number of old waste deposits, or wild landfills, in Germany is not precisely understood, but estimates range from 50 000 to 100 000 (BMU, 2006;Burmeier, 2014;Umweltbundesamt, 2017aUmweltbundesamt, , 2017b)).It was assumed that not all of these areas contained waste categories that would result in emissions of BPA; therefore, 37 500 wild landfills were assumed for this evaluation.Wild landfill loadings were calculated based on population fractions in each coordination area.For these wild landfills, direct discharge to surface water was assumed (0% removal) and the loadings were assumed to be equivalent to the active landfills with no treatment (category 4, 0.00775 kg/a).

Paper recycling model scenarios
The most significant change in recent BPA usage is the restriction of the use of BPA in thermal paper introduced in 2020.As described below and in Supporting Information: Paper Recycling, we modeled three scenarios to evaluate the effects of removal of BPA from thermal paper and subsequent removal from the paper recycling loop and recycled paper.The three scenarios include (1) Current A, (2) Current B, and (3) Future.The two current scenarios are based on the understanding that BPA is no longer being used in the manufacture of thermal paper in Germany but BPAcontaining paper articles continue to be present in wastepaper to be recycled.The Current A and Current B Scenarios differ in how emissions from the paper recycling pathway are estimated.In the Current A Scenario, the contributions from paper recycling are based on measured BPA concentrations in paper articles (Ramboll, 2019), the proportion of wastepaper being recycled, and the amount of BPA extracted to wastewater during paper recycling.The Current B Scenario differs from the Current A Scenario in that contributions from paper recycling are based on measured effluent loadings to surface water from WWTPs serving paper recyclers (see Supporting Information: Paper Recycling for additional details).
The Future Scenario assumes that BPA has completely dissipated from thermal paper and recycled paper.It thus assumes zero BPA emissions from paper recycling and recycled paper-based articles as the ultimate consequence of the European restriction of the use of BPA in thermal paper.It further considers future reductions of BPA content in recycled PVC.As it is unclear as to how long it will take before BPA completely abates from the paper recycling cycle, the Future Scenario is not linked to a definite point in time.

RESULTS AND DISCUSSION
The emissions associated with each of the pathways described above were used as primary loadings to the FlowEQ model and are listed in Supporting Information: Table SI-3.Detailed emissions for each pathway are shown in detail in Supporting Information: Table SI-4.Table 2 summarizes BPA emissions to surface water.Figure 2 presents the loadings to surface water for each coordination areas for the three scenarios.In the Current A and Current B Scenarios, emissions associated with paper are the dominant source of BPA discharges to surface water, accounting for 85% and 77% of the total, respectively.All the remaining sources individually account for less than 10% of the total in the current scenarios.The second highest source of emission in the current scenarios are landfills, followed by articles made from recycled material (PVC and tires).Consumer use of articles containing epoxy resin and polycarbonate account for 3.5%-5.5% and 0.1%-0.2% of the total in Current A and Current B Scenarios, respectively.
In the Future Scenario, landfills represent the single largest source of BPA loadings to surface water (53%).The next largest source is consumer uses (33%), with epoxy resin-containing articles accounting for the majority (32%) and polycarbonate-containing articles contributing marginally (1%).Further sources of loadings include articles containing recycled material (9.5%) and manufacturing (4.5%).These loadings estimates also demonstrate the predicted dramatic reductions in BPA loadings to surface water in the Future Scenario, representing almost an order of magnitude reduction as compared to the Current A Scenario (Table 2).This reduction is mainly attributable to the removal of BPA from thermal paper and the paper recycling chain, with further recycled materials (e.g., PVC) also contributing to the reduction as compared to the current scenarios.
The model predicts surface water concentrations for each coordination area under each scenario.The surface water BPA concentrations in each coordination area calculated for each of the three scenarios are shown in Supporting Information: Table SI-11 and summarized in Table 3.The results indicate that the highest concentrations are predicted for the Current A Scenario (assessment of BPA releases associated with paper recycling based on measured BPA content in recycled paper articles), followed by the Current B Scenario (assessment of BPA releases associated with paper recycling based on measured effluent loadings to surface water from WWTPs serving paper recyclers).The Future Scenario is based on the assumption of zero BPA releases from paper recycling and articles made from recycled paper, as well as lower BPA concentrations in recycled PVC.The predicted BPA concentrations in the Future Scenario are much lower than those calculated in the two current scenarios, with a reduction of 89% from Scenario A and an 83% reduction from Scenario B.
The model results are presented by emissions type for each coordination area in Figure 3, which demonstrates that emissions associated with paper (consumer use of paper and paper recycling) account for the majority of BPA in surface water for the scenarios Current A and Current B (65%-98%).Landfills, and to a lesser extent the use of recycled articles and the consumer use of polycarbonate and epoxy resin-containing articles, are minor but discernible contributors.The results for the Future Scenario show that across the different coordination areas, landfills account for, on average, just over 50% of the BPA in surface water.The next most important contributors are consumer articles made from epoxy resins (32%) and polycarbonate (1.2%), articles made from recycled materials such as PVC and tire shred (9.7%), industrial emissions (3.1%), and advection from upstream areas (1.2%).It should be noted that while the relative contributions from these articles are projected to be larger in the Future Scenario compared with the current scenarios, the mean surface water concentrations are an order of magnitude lower.Supporting Information: Figure S-1 shows the same results in Figure 3, but the bars show the contribution by discharge pathway.These results confirm the prediction that emissions to surface water account for the majority of the BPA in surface water, with advection from upstream areas accounting for a small proportion and emissions to air accounting for a de-minimis fraction.
As discussed in Tappert et al. (2023) and the Supporting Information, many of the emission sources discharge to wastewater, which is then subjected to wastewater treatment prior to entering surface water.These sources include industrial effluent, indoor consumer use, some landfill leachate, and the effluent from paper recyclers.The model assumes that wastewater is processed by treatment plants that may also process stormwater and therefore are subject to bypass events.The specific assumptions associated with modeling bypass events are documented in the Supporting Information.Supporting Information: Figure S-2 shows the model estimates for surface water concentrations, with the bars shaded to show the contributions associated with direct discharges to surface water and air, wastewater treated by modern plants with a high removal efficiency, water treated by plants with a low removal efficiency, and discharges associated with bypass events.The results demonstrate that bypass events contribute a large proportion of the BPA in surface water predicted by the model.In fact, for many coordination areas, it represents the largest share of the BPA predicted in surface water.For the Current A Scenario, the maximum contribution from direct discharge was 13.8%,  22% from high removal, 11.3% from low removal, and 53% from bypass.These results indicate that improvements to stormwater infrastructure design to minimize bypass events are expected to result in dramatic reductions of BPA.Note that for Scenario Current B, discharges from WWTPs serving paper recyclers were based on direct measurements of effluent loadings after treatment and were classified as direct discharges in the final output.This method did not directly consider treatment efficiency and therefore did not consider high removal versus bypass removal for the Scenario Current B.

Comparison with measured surface water
Bisphenol A concentrations in surface water were evaluated using surface water monitoring data collected from 1998 to 2020 and provided by the sources listed in Supporting Information: Table SI-12.The data included measurements from 15 out of 16 German federal states and 37 out of 46 coordination areas.Nondetected results were included as one-half of the reported detection limit for this analysis.The average detection limit for nondetected results prior to 2015 was 0.024 µg/L (range: 0.005-0.1 µg/L), decreased to 0.014 µg/L (range: 0.005-0.05µg/L) for data from 2015 to 2018, and remained consistent at 0.015 µg/L (range: 0.005-0.03µg/L) for post-2018.Although the detection limit for BPA has been decreasing over time, the observed decrease in surface water concentrations is independent of these changes in detection limits.Summary statistics were generated by time period and coordination area (Supporting Information: Table SI-13) and federal state (Supporting Information: Table SI-14).Supporting Information: Table SI-14 includes samples that had insufficient location information to allow the measurements to be assigned to a specific coordination area but could be assigned to a federal state.The results are summarized in Table 4 and show

Source assessment
The results provide valuable information regarding the relative influence of various BPA uses and discharge pathways on the concentration in surface water.The results demonstrate that certain uses and pathways cause negligible contributions, while others are more significant.The model predicts that the concentration of BPA in surface water will decrease significantly after the restriction of the use of BPA in thermal paper has had sufficient time to work through the article lifecycle and the article recycling cycle.The trends of decreasing surface water concentrations with time that are predicted by the model are already being observed in the surface water measurements, demonstrating the power of the model and its usefulness in understanding how changes in usage can influence surface water concentrations of BPA.The strong concurrence of the modeled and measured surface water concentrations also indicates that the significant sources have been identified and that the underlying assessments for the various source contributions are plausible.
In the two current usage scenarios (Current A and Current B), emissions are based on the conditions present at the time of the removal of BPA from thermal paper manufacturing.Bisphenol A emissions from thermal paper manufacturing were assumed to be zero, but BPA was still assumed to be fully present in the paper recycling process chain.The total loadings to surface water for these scenarios are 18 503 and 11 920 kg/a for scenarios Current A and Current B, respectively.In these scenarios, emissions associated with paper recycling and the use of paper articles containing recycled content dominate the loadings to surface water and therefore also are the single largest contributor to the BPA in surface water.The model concludes that this pathway accounts for between 77% and 85% of the BPA currently measured in surface water.The next most important source is landfills (5.8%-9.1%),followed by recycled articles (e.g., PVC and tire shred) (4.8%-7.5%).The model indicates that the use of consumer articles made from virgin materials are very minor contributors, such as articles made from epoxy resins (3.5%-5.5%)and polycarbonate (0.1%-0.2%).Manufacturing of BPA and the respective polymers as well as their processing (0.5%-0.7%) are also very minor contributors.
In the Future Scenario, the model emissions are based on the assumption of a complete removal of BPA from paper articles, including recycled paper, and the paper recycling process.This scenario also considers a reduction in the BPA content in other recycled articles, that is, in recycled PVC, as the BPA content in the feedstock is expected to decline due to reductions in the use of BPA in the parent materials (see the Supporting Information).The total emissions in the Future Scenario are calculated to be 2036 kg/a, nearly an order of magnitude less than the Current A Scenario.Emissions associated with paper were assumed to be zero and the loadings from the other recycled articles were assumed to decline by more than a factor of 4. Consequently, the model predicts that landfills will become the dominating source of loadings to surface water (53%) emissions in the Future Scenario.The consumer use of BPA-containing articles, namely, epoxy resins and polycarbonate, accounts for 33% of the loadings to surface water, with epoxy resins contributing 32% and polycarbonate the remaining 1% in the Future Scenario.However, the model is based on conservative assumptions that most likely significantly overestimate the loadings from consumer uses of materials made from epoxy resins and polycarbonate.Plastics Europe and its members are conducting additional research together with the German authority BAM to further understand and refine the loadings estimates from this pathway.Articles made from the other recycled materials, tire shred, and recycled PVC are calculated to contribute 9.5% to the Future Scenario loadings.Loadings associated with manufacturing account for 4.1% of the BPA in Future Scenario surface water according to the model.
The Future Scenario predicts that the BPA concentration in surface water will further decrease significantly from current levels by as much as an order of magnitude.The average modeled BPA concentration in surface water for the Current A Scenario is 0.053 µg/L, compared with the modeled future concentration of 0.0037 µg/L, which represents a 93% reduction.This predicted average surface water concentration is even lower than the lowest limit of detection applicable to all recent monitoring data (LOD = 0.005 µg/L).

CONCLUSION AND OUTLOOK
In the current scenarios, BPA associated with the use and recycling of paper articles is the most important source.It is important to note that both the Current A and Current B Scenarios yield very similar conclusions, confirming the robustness of the model and the input data and assumptions used.The model predicts that as the European restriction on the use of BPA in thermal paper promulgates through thermal paper manufacturing and eventually the entire paper recycling system, BPA concentrations in surface water will further decrease.This prediction is consistent with the currently already observed decline in BPA in surface water.It should be noted that the model cannot predict the rate of the disappearance of BPA in the paper cycle; thus, it is unknown when the paper cycle will cease being an important source.In the Future Scenario, landfills are predicted to be the largest source of BPA in surface water.However, the current restrictions on landfilling in Germany mean that currently, there are only de minimis new loadings of BPAcontaining materials to landfills.As the residual inventory of BPA in legacy landfills gradually degrades or is leached from the waste, loadings of BPA are also expected to decline, eventually reaching de minimis levels.However, modeling these reductions is beyond the capabilities of the current model and the time required for these reductions cannot be predicted.
When assessing the emissions of BPA due to consumer use, it is important to consider wastewater treatment processes.The model predicts that although bypass events account for only 15% of the wastewater discharges to surface water, bypass events account for approximately 50% of the BPA emissions to surface water from the wastewater pathway.This emphasizes the importance of proper stormwater management practices with the goal of minimizing the frequency of bypass events and the discharge of minimally treated wastewater.
The model results indicate that the restrictions currently in place are expected to result in continuing decreases in BPA concentrations in surface water.Improved management of landfills and landfill leachate and improvements in stormwater infrastructure can also provide significant reductions in loadings and therefore reductions in surface water BPA concentrations.Improvements to this infrastructure are also expected to reduce loadings of other substances such as other chemicals of emerging concern.Because consumer use does not represent the majority of BPA loadings to surface water, additional restrictions on BPA in polycarbonate and epoxy resins in consumer articles are not predicted to provide large reductions in loadings, while improvements to landfill leachate and stormwater management are expected to provide more meaningful reductions in emissions to surface water.
The analysis of historical and recent surface water BPA concentrations indicates that the model predictions of falling surface water concentrations are consistent with the measured concentrations.The complexity of the BPA cycle and the longevity of many BPA-based applications mean that it may take many years for changes in BPA usage to fully work though the environment and for surface waters to reach new stable concentrations.The Future Scenario predicts continued reductions in BPA loadings to surface water.However, the Future Scenario does not include several of the potential further reductions in loadings, for example, the expected decline of BPA loading from landfills due to the recent restrictions on landfilling most waste.Therefore, even in the absence of further restrictive measures, future BPA concentrations in surface water are predicted to be much lower than current concentrations, at least by an order of magnitude.

LIMITATIONS
As with any model, the results are limited by the accuracy and completeness of the model inputs and model formulation.The model predicts a stable average concentration within each unit and does not account for small-scale geographic or temporal variations in emissions and other processes within the units of the model.Small-scale concentrations are expected to be much more variable than predicted by the model.

FIGURE 2
FIGURE 2 Loadings (kg/a) to surface water for each coordination area by emission type and model scenario.CU, consumer use a clear decrease in the BPA concentrations over time, as values declined from pre-2015 over 2015-2018 to post-2018.The average post-2018 concentration is 69% lower than the pre-2015 value.Additionally, the frequencies of detection for pre-2015, 2015-2018, and post-2018 are 59%, 42%, and 29%, respectively, and are thus declining over time as well.Multiple factors likely contributed to the post-2015 reductions in BPA concentrations in surface water, including (1) decreasing usage of BPA as an additive in manufacturing (i.e., PVC); (2) potential improvements in WWTP processes, including handling of bypass events; and (3) potential improvements in the handling of legacy landfill leachates.Note that the maximum measured and maximum modeled values in Table 4 are not directly comparable.The modeled values represent an estimated average concentration for each of the model units, and the maximum value therefore represents the highest average across the model units.The maximum measured value is the maximum of individual grab samples.We combined the model results and the surface water monitoring data to allow a more direct comparison of the results.Figure 4 provides a key for interpreting the model and/or surface water summary plot in Figure 5.The surface water data for each time period and coordinate area are presented using box plots and the model results are presented as points.The model results show that the highest modeled concentrations are associated with the Current A and Current B Scenarios, with the Future Scenario results being much lower.The corresponding box plots of concentrations measured in surface water show a consistent trend of decreasing concentrations in the more recent time periods.The results indicate a strong correspondence between the measured and modeled results.With the modeled results generally falling within the measured results, the coordination areas with the highest model concentrations also tend to have the highest measured ones.This comparison highlights the ability of the model to predict the concentrations in surface water with reasonable accuracy.It indicates that the significant sources have been identified and allows the assessment of the various source contributions.The results also demonstrate that the model prediction of decreasing surface water concentrations after the restriction on BPA in thermal paper took effect is consistent with the measured data.

TABLE 2
Summary of bisphenol A (BPA) loadings (kg/a) by scenario, source, and pathway Abbreviations: CU, consumer use; PVC, polyvinyl chloride.a Loadings are summed for the entire model domain, not just Germany.b Consumer use = articles made from polycarbonate and epoxy resins (service life).

TABLE 3
Summary of modeled bisphenol A surface water concentrations in the coordination areas by model scenario a,b a All results are in µg/L.bCoordination area-specific concentrations presented in Supporting Information: TableSI-11.