Modelling future coastal water pollution: impacts of point sources, socio-economic developments & multiple pollutants

Urbanisation is happening worldwide. In 2100, over 70% of the population is projected to live in highly urbanised areas. As a result, urban wastewater discharge may increase. This may add multiple pollutants to rivers and coastal waters. However, current knowledge on how urbanisation-related socio-economic developments affect coastal water pollution is limited. In this study, we analysed individual and combined impacts of wastewater treatment improvements, economic growth and city expansion on future coastal water pollution from point sources (sewage and open defecation) by sub-basin taking a multi-pollutant approach. We improved the existing MARINA-Multi model (version Global-1.0) by integrating hydrology and pollutant retentions in order to quantify river exports of total dissolved nitrogen, total dissolved phosphorus, microplastics and triclosan to coastal waters for 2010 and 2100 using scenario analysis. Globally, river exports from point sources are projected to more than double by 2100 for all pollutants, especially in Africa and Asia. Wastewater treatment improvements, economic growth and city expansion can have a positive (less pollution) or negative (more pollution) impact on future coastal water pollution. These impacts differ among pollutants and sub-basins. Wastewater treatment improvements may globally reduce multi-pollutant issues (−30% to −38% change on average) compared to the reference scenario (positive impact). Economic growth and city expansion may globally enhance multi-pollutant issues (+15% to +25% and +28% to +33% change on average, respectively) compared to the reference scenario (negative impact). A combined scenario, accounting for all three socio-economic developments simultaneously, may globally reduce or enhance pollutant issues (−21% to +50% change on average) compared to the reference scenario. In the combined scenario, the reinforcements of positive and negative impacts are pollutant- and region-dependent. Our study gives insights into future coastal water pollution, which aids in identifying management strategies for urban areas, hence contributing to reaching Sustainable Development Goal 14.


Introduction
Coastal waters often offer a wide array of crucial ecosystem services [1].Hence, healthy coastal waters are vital for nature and society, resulting in their protection being part of Sustainable Development Goal (SDG) 14: Life Below Water.Yet, coastal waters are often polluted by a cocktail of pollutants exported by rivers as a result of human activities on land.Examples of pollutants are nutrients, plastics and chemicals.Impacts of coastal water pollution, amongst others, include the toxic effects of algae blooms resulting from excessive nutrient loadings [2,3], potential toxic effects for marine food webs due to ingestion of microplastics (MP) [4][5][6], and endocrine disruption effects of antimicrobial agents (chemicals) such as triclosan (TCS) [7,8].
Our world has changed rapidly in the last decades, and even more changes are expected.Examples of these changes are socio-economic developments such as population growth, technological developments for wastewater treatment, economic growth and city expansion.Seminal work reported that socio-economic developments affect pollutant emissions into waters [9][10][11][12].Additionally, urban areas are expected to grow by 200% by 2100 compared to the baseline of Li et al [13] for 2013.This implies increasing sewage connections [9], as urban areas generally have a higher level of sewage connections compared to rural areas [14].Subsequently, urban wastewater discharges are expected to increase (i.e. point source), especially when treatment is not sufficient.However, current knowledge on how urbanisationrelated socio-economic developments affect coastal water pollution is limited.Global water quality models can help to generate this knowledge, especially for data-poor areas [15,16].Examples of existing largescale water quality models are the Models to Assess River Inputs of pollutaNts to seAs (MARINA) [9,17,[18][19][20], Global Nutrient Export from WaterSheds (Global NEWS) model [21], Integrated Model to Assess the Global Environment-Global Nutrient Model (IMAGE-GNM) [10], dynamical surface water quality (DynQual) model [22], and the Soil and Water Assessment Tool (SWAT) [23][24][25].While there is a wide variety of large-scale water quality models, there is also a large variety of pollutants to study, and each model has their strengths and limitations [26].This is also shown in the Inter-Sectoral Impact Model Intercomparison Project for the Water Quality Sector (www.isimip.org/).Yet, existing global models often focus on single pollutants [15-17, 27, 28], inputs to rivers [9,19] or recent past [20].This calls for a more integrated modelling approach to better understand the impacts of socio-economic changes on multiple pollutants in coastal waters.
Our study aims to explore the individual and combined impacts of wastewater treatment improvements, economic growth and city expansion on future coastal water pollution from point sources (sewage and open defecation) by sub-basin taking a multipollutant approach.We define coastal water pollution as the river export of pollutants to the river mouth.In our study, we improved the Model to Assess River Inputs of pollutaNts to seAs (MARINA-Multi, version 1 of Strokal et al [9]) model by adding pollutant retention and hydrology.This enabled us to simulate past (2010) and future (2100) coastal water pollution associated with point sources from sub-basins for three pollutant groups simultaneously: nutrients (total dissolved nitrogen (TDN) and total dissolved phosphorus (TDP)), plastics (microplastics) and chemicals (triclosan as antibacterial agent).We focussed on sewage and open defecation as point sources.We used a reference scenario that combines Shared Socioeconomic Pathways 2 (SSP2) [29] and Representative Concentration Pathway 2.6 (RCP2.6)[30], referred to as SSP2-RCP2.6[31], to simulate projections for the year 2100.We used alternative scenarios to further study the impacts of individual and combined socio-economic developments, including improved wastewater treatment, economic growth and city expansion.

An improved MARINA-multi model
We improved the existing MARINA-Multi Global-1.0model [9], which focused on loadings of pollutants into rivers, by integrating hydrology and pollutant retentions.Our improved model version quantifies past (2010) and future (2100) river exports of TDN, TDP, MP, and TCS from point sources to coastal waters by sub-basin and source.The model runs for 10 226 sub-basins in the world (see figure S1 in SI 1).Yet, our analysis only encompassed the 9 482 sub-basins that are part of the drainage area discharging pollutants into the sea, thus reaching coastal waters (the other sub-basins are inland).We distinguished three sub-basin urbanisation classes: low (<30% urban population), moderate (30%-70% urban population), and high (>70% urban population) (see figure S8 in SI 6).The novelty of this model version is that we integrated geospatiallyexplicit hydrological input data and retention processes of multiple pollutants during the river export (i.e.retention in rivers, in reservoirs or lakes, and through water consumption).We did this for the world.This integration was important because (1) it gave insights into how multiple pollutants from sewage and open defecation reach coastal waters, from which sub-basins and under which scenarios, (2) it enhanced our understanding of upstreamdownstream dynamics in river export of multiple pollutants, and (3) it was an important step for potential future model developments (e.g.studying biogeochemical interactions between pollutants in rivers and impacts of climate change).The routing scheme for the river export of pollutants from sub-basin to sub-basin was developed and implemented for the studied pollutants in this model based on the subbasin scale modelling approach of Strokal et al [18].For the retention processes, we integrated existing modelling approaches for TDN [9,18], TDP [9,18], MP [9,28,32], and TCS [9,27] (see SI 2 for details).A glossary with the most frequently used abbreviations can be found in the SI.
We used three main equations to quantify river export of multiple pollutants from point sources to coastal waters by sub-basin (equations ( 1)-(3f )).The inputs of pollutants from point sources that are exported by rivers to the river mouth (coastal waters) were calculated as follows: Mpnt i.j = RSpnt i.j • FE riv.i.outlet.j• FE riv.i.mouth.j(1) where: Mpnt i.j is the annual river export of pollutant i from point sources by sub-basin j (kg yr −1 or g yr −1 ); RSpnt i.j is the input of pollutant i from point sources to rivers in sub-basin j (kg yr −1 or g yr −1 ); FE riv.i.outlet.j is the fraction of inputs of pollutant i to rivers that is exported to the outlet of sub-basin j (0-1); FE riv.i.mouth.j is the fraction of pollutant i at the outlet of sub-basin j that is exported to the river mouth (coastal waters) from sub-basin j (0-1).Here, we used RSpnt i.j data from Strokal et al [9].
The fraction of pollutants that is retained or lost during the river export to the sub-basin outlet is calculated as follows: where: L i.j is the fraction of pollutant i retained/lost due to biogeochemical processes in rivers of sub-basin j (0-1); D i.j is the fraction of pollutant i retained/lost in reservoirs or lakes of sub-basin j (0-1); FQrem j is the fraction of pollutants removed from sub-basin j through consumption of water (0-1).L i.j and D i.j are pollutant-specific retention factors and dependent on sub-basin characteristics (see tables S2-S3 and SI 2 for details).Sub-basin hydrology and population growth are used to quantify pollutant losses through water consumption (see box S1 in SI 2).The fraction of pollutants that is exported from the sub-basin outlet to the river mouth of the subbasin is calculated as follows (the routing scheme that is based on the sub-basin scale modelling approach of Strokal et al [18] where: FE riv.i.mouth.j is the fraction of pollutant i at the outlet of sub-basin j that is exported to the river mouth from sub-basin j.Retention of pollutant i in the water system depends on the sub-basin type of sub-basin j.Sub-basin types are indicated by three-letter codes: j for sub-basin, u, m, or d for upstream, middlestream, downstream, respectively, and T or C for tributaries or main channel, respectively.The equation's right-hand side, describes the fraction of pollutant i that is exported from the subbasin outlet (sub-basin type in superscript) to the next sub-basin outlet (sub-basin type in subscript) following river flow to, eventually, the river mouth (e.g.juT FE riv.i.outlet.juC ) (see table S5 for details).

Socio-economic development scenarios
Our reference scenario for the year 2100 accounted for moderate socio-economic developments and low climate change impacts (SSP2-RCP2.6,see SI 4 for details).In this scenario, total population is expected to increase by 31% globally compared to 2010 (table S7).However, urban population is expected to grow much more, namely 106% increase by 2100 (table S7).Wastewater treatment may not improve to a large extent in the future compared to 2010 (see more details in figure S3 and table S7).We used scenario data from Strokal et al [9] and included projections for hydrological data (see SI 4 for details).
We developed four alternative scenarios (AS WWTP , AS ECON , AS CITY , AS WWTP-ECON-CITY ) to quantify the individual impacts of wastewater treatment improvements (AS WWTP ), economic growth (AS ECON ) and city expansion (AS CITY ), and their combined effects (AS WWTP-ECON-CITY ) on future coastal water pollution from point sources.AS WWTP assumed an increase in pollutant removal efficiency of 80% relative to the reference scenario for all pollutants.AS ECON assumed an increase in gross domestic product (GDP, reflecting income levels) and human development index (HDI, reflecting the societal developments) of 80% relative to the reference scenario.AS CITY assumed an increase in people connected to sewage systems of 80% relative to the reference scenario.AS WWTP-ECON-CITY included all aforementioned assumptions to reflect their combined impacts relative to the reference scenario (see details in table S8).
We analysed the changes in river exports of pollutants between the alternative scenarios and the reference scenario per world region (figure S1).Here, decreases in river export of pollutants compared to the reference scenario were associated with positive impacts on coastal water pollution in 2100.Increases in river export of pollutants compared to the reference scenario were associated with negative impacts.The alternative scenarios are novel in that they improved our understanding of relations between socio-economic development and urbanisation, and how those relations affect river exports of multiple pollutants into the coastal waters among world regions.

Future coastal water pollution from point sources
In the past (2010), river exports of pollutants from point sources were high in many European, Asian and North American coastal waters (figure 1).African coastal waters were not polluted to that extent.However, this may change in the future.By 2100, river export of all pollutants from point sources globally is projected to more than double by 2100 under the reference scenario.Like in 2010, highly urbanised sub-basins still hold the biggest share in river export for each of the four pollutants (78%-83%) in 2100 (figure S9).However, the population living in highly urbanised sub-basins is projected to have more than tripled by 2100 (from 1.9 to 6.0 billion people, figure S8).On one hand, this implies that more people may live in sub-basins with multi-pollutant issues associated with point sources.On the other hand, this also implies that more people may have access to sewage connections and wastewater treatment plants.Additionally, we see a shift in most-polluted world regions (figure 1).In 2010, Europe had the highest yield (i.e.kg pollutant km −2 sub-basin area) in river export for each of the four pollutants.In 2100, the highest yield is projected for Asia, followed by Africa (except for MP).This may be explained by the projected increases in urbanisation levels (Asia: 327%, Africa: 769%), population growth (Asia: 5%, Africa: 155%) and rapid economic (Asia: 435%, Africa: 500%) and human (Asia: 20%, Africa: 21%) developments, while wastewater treatment efficiency, despite improvement efforts, is assumed to remain relatively low in these areas (average removal efficiencies Asia: 24%, Africa: 23%).Despite increased urbanisation levels (134%) and population growth (80%), Oceania will likely remain the least polluted world region in 2100.Relatively high wastewater treatment efficiencies may explain this.Future trends differ between regions and pollutants under the reference scenario (figure 1).
In 2100, river exports of TDN and TDP from point sources are projected to increase by 150% and 133%, respectively.This will result in 6 810 kton for TDN and 1 044 kton for TDP in coastal waters globally.The highest increases for TDN and TDP are projected for sub-basins in Africa, southern Asia, North America, and Central and South America (figure S10).This may be explained by increases in urbanisation levels, population growth, and rapid economic and human development (figures S4 and S9).Moreover, these areas are characterised by increased urban sewage connections (figures S5 and S6).Hence, the major efforts to improve wastewater treatment appear insufficient to reduce nutrient pollution to the levels of 2010.Decreases in river exports of TDN and TDP are projected for sub-basins located in Europe, northern Asia, and Oceania (only for TDP) (figure S10).Mainly because population growth will be limited, but urban sewage connections will increase with advanced treatment (figures S3 and S6).
In 2100, river exports of MP and TCS from point sources are projected to increase by 101% and 103%, respectively.This will result in 101 kton of MP and 0.4 kton of TCS in coastal waters globally.For MP, the highest increases are projected for sub-basins in Africa, southern Asia, Central and South America, Oceania and some sub-basins in North America and Europe (figure S10).As mentioned before, this may be explained by increases in urbanisation levels, (urban) population growth, and rapid economic and human development (except for Oceania) (figures S4 and S8).Moreover, these areas are characterised by increased urban sewage connections (except for Oceania) (figures S5 and S6).In these areas, the major efforts to improve wastewater treatment appear insufficient to reduce MP pollution.For TCS, the highest increases are projected for sub-basins in Africa, Asia, Oceania, and parts of North America (figure S10).In these areas, TCS removal during wastewater treatment is projected to stay the same or slightly increase (figure S3).Here, the increase may explained by either a growing (urban) population or an increased number of urban sewage connections (figure S6).Decreases in the river exports of MP are mainly projected for sub-basins in Europe, northern Asia, and specific areas in North America (figure S10).In these areas, the removal fraction of MP during wastewater treatment is projected to be high (except for Greenland).These areas are characterised by a declining population (except for the United States) and increased urban sewage connections (figure S6).Decreases in the river exports of TCS are mainly projected for sub-basins in Central and South America and parts of Europe, North America, and northern Asia (figure S10).In these areas, the removal fraction of TCS during wastewater treatment is projected to improve further (figure S3).Moreover, these areas have a history of very high levels of human development and relatively high GDP and are projected to remain at such high levels in 2100.

Impacts of socio-economic development
Alternative scenarios show that wastewater treatment improvements, economic growth and city expansion (i.e.socio-economic developments) can either have a positive (less pollution) or negative (more pollution) impact on future coastal water pollution from point sources compared to the reference scenario (2100, section 3.1, figure 2).The impacts differ among world regions and pollutants.Generally, the largest differences in the impact amongst world regions are projected for TDP.Overall, the pollution levels in Africa and Asia are projected to be most negatively impacted in all alternative scenarios.These world regions are rapidly urbanising under a growing population (Africa) and rapid economic and human development (Africa and Asia).Pollution levels in North America and Europe are projected to be most positively impacted.These world regions have very advanced wastewater treatment plants, very high human development and high GDPs.
AS WWTP (i.e.AS1 in figure 2) shows that wastewater treatment improvements globally may reduce multi-pollutant issues (−30% to −38% change on average) compared to the reference scenario (positive impacts).Coastal water pollution by TCS (except for Central and South America) and TDP is projected to be most positively impacted.Even though TDP removal increased notably in Central and South America, no improvements are projected for TCS removal in this world region.This is related to the poor TDP and TCS removal fractions in the reference scenario (see the Discussion section).The most positive impacts are projected for North America, Europe and Oceania.For these regions, advanced wastewater treatment plants are expected to remove nearly all pollutants originating from sewage systems.Hence, the pollutant loads to rivers from point sources remain relatively low.The least positive impacts are projected for Africa, where it is hard to compete with the spikes in population and urban sewage connections and reduce pollution in 2100 to the level of 2010.
AS ECON (i.e.AS2 in figure 2) shows that economic growth globally may enhance multi-pollutant issues for TDN, TDP and MP (+15 to +25% change on average) compared to the reference scenario (negative impacts).Coastal water pollution by TCS is projected not to be affected by economic growth.However, coastal water pollution by MP (except for Europe and North America) is projected to be most negatively impacted.This results from spikes in plastic consumption rates due to human development in these regions.In Europe and North America, plastic consumption is projected to be already very high in the reference scenario and, therefore, not affected by additional human development in AS ECON .The highest increases in pollutant loads are projected for Africa and Asia.This is mainly due to the high increases in MP pollutant loads in these areas due to major increases in plastic consumption rates resulting from human development.For TDN and TDP, the increases in pollutant loads are comparable for all world regions as excretion rates are expected to increase with economic growth gradually.
AS CITY (i.e.AS3 in figure 2) shows that city expansion globally may enhance multi-pollutant issues for all pollutants (+28 to +33% change on average) compared to the reference scenario (negative impacts).Increases in pollutant loads are comparable for all pollutants.The highest increases in pollutant loads are projected for Africa and Asia.These areas also have projections to majorly expand their connections to sewage systems in both urban (i.e.130% and 75% increase, respectively) and rural areas (i.e.144% and 187% increase, respectively).
AS WWTP-ECON-CITY (i.e.AS4 in figure 2) shows that a combined scenario, accounting for all three socio-economic developments simultaneously, globally may reduce or enhance pollutant issues (−21% to +50% change on average) compared to the reference scenario (positive and negative impacts).Here, reinforcements of positive and negative impacts are pollutant and region-dependent.For all world regions, TDN and TDP loads increase between 23 and 77% compared to the reference scenario and, thus, have negative impacts on coastal water pollution in 2100.This implies that improved wastewater treatments cannot outweigh the impacts of economic growth and city expansion on nutrient loads.Hence, limiting economic growth and city expansion while improving wastewater treatment is needed in all world regions to create positive impacts regarding nutrient pollution from point sources.
Contrarily, MP and TCS loads both decrease and increase (change between −71% and +135%) compared to the reference scenario and, thus, have both positive and negative impacts on coastal water pollution depending on the regional characteristics.Coastal water pollution by MP is positively impacted for Europe, North America, Central and South America, and Oceania, and negatively impacted for Africa and Asia.This implies that in Africa and Asia, improved wastewater treatment cannot outweigh the economic growth and city expansion for MP pollution.Hence, limiting economic growth and city expansion, while improving wastewater treatments is needed in these areas to reduce MP pollution from point sources.It is different for the other world regions, where focussing on improved wastewater treatment could help reduce MP pollution from point sources.Coastal water pollution by TCS is positively impacted for all world regions except for Central and South America.This implies that improved wastewater treatment outweighs the effects of city expansion on TCS pollution.However, an exception is when the TCS removal level is very low in the reference scenario (i.e.below 11%).Hence, this implies that wastewater treatment withstands positive impacts on TCS pollution from point sources as long as a specific TCS removal threshold has been crossed.

Discussion
We presented an improved multi-pollutant model version for projecting point source-associated coastal water pollution for sub-basins worldwide.Our improved model integrates existing and wellevaluated modelling approaches for TDN [9,18], TDP [9,18], MP [9,28,32] and TCS [9,27] and accounts for pollutant-specific retention processes in rivers, reservoirs and through water consumption at the sub-basin scale.Globally, on average, 28%-80% of the pollutants are retained during river export depending on the sub-basin characteristics (see SI 8 for details).To evaluate our model outputs, we compare them with other models and observation data (see tables S10-S12).Our river export estimates of 2 721 kton of TDN are much lower than 7 531 kton of TN from Beusen et al [10] and lower than 1 731 kton of TDN from Seitzinger et al [33] (table S10).For TDP, our river export estimate of 448 kton is lower than 727 kton from Seitzinger et al [33] and 958 kton from Beusen et al [10] (table S10).This is because we do not consider particulate nutrients and our approach is based on sub-basins, which is different from those studies (gridded or basin).Our future projections coincide with future projections of those existing studies showing increasing trends.Other studies (see table S10), show a wide variation in their global annual river export of MP (i.e.49-980 kton).Our estimate of 51 kton is at the lower end of this range.At the European scale, our estimates are generally lower than those of Siegfried et al [32] (table S10).Our estimates for river exports of TCS in 2010 for the Danube, Elbe and Zhujiang are similar to those of Van Wijnen et al [27] (table S10).Our estimate of 1.07 ton TCS for the Ganges is higher than 0.67 ton from Van Wijnen et al [27] (table S10).The general future trend of multi-pollutant issues is similar to those described in earlier studies [9,34], highlighting Africa as a hotspot for future surface water pollution (table S10).Although comparing our model results to observation data may be unfair for nutrients, as we only account for point sources, it builds trust in our model as highly urbanised rivers seem to have a higher share of point source pollution (tables S12(a) and (b)).This is in line with previous studies [33,35].
Our four alternative scenarios (AS WWTP , AS ECON , AS CITY , AS WWTP-ECON-CITY ) provided new insights into how individual and combined socio-economic developments affect future coastal water pollution from point sources.They aimed to show how the impact differs among world regions regarding directions and magnitude.Our scenario design was simple but transparent.However, such simplicity may not have captured the full complexity of the relationship between socio-economic development and coastal water pollution.We described the limitations of our scenario design, such as a lack of feedback loops between the impact of water pollution on economic developments [12,36,37], in SI 7. Nevertheless, we believe that our results presented valuable insights and highlight that different world regions call for different actions.This stresses the importance of accounting for region-dependent interactions between socio-economic developments (e.g.drivers) and pollutant groups when developing policies.This could aid in reaching SDG14: life below water, by identifying effective solutions for pollutants from sewage systems and open defecations.
We consider our model a useful tool for identifying sources, drivers and solutions to water quality issues originating from point sources [26], but it has limitations.First, we underestimated the actual coastal water pollution as we did not account for diffuse sources from land (e.g.agriculture, macroplastic fragmentation) and human activities on coastal waters (e.g.aquaculture).This may have bigger implications for some regions and pollutants than for others.For example, the study of Strokal et al [17] showed that, globally, sewage systems are the most important source of microplastic pollution, except for Africa where mismanaged plastic waste is the dominant (diffuse) source.Triclosan is predominantly a point source, but depending on sub-basin characteristics an unknown amount may enter rivers via diffuse sources through sewage sludge applications on land [38,39].However, as triclosan is rather immobile in soils we expect diffuse sources of triclosan to be minimal or local specific [40,41].For nutrients, on the other hand, the exclusion of diffuse sources likely has considerable impacts on the total pollution ending up in coastal waters (we may underestimate global river exports of nutrients by 60%-90% depending on the nutrient form (TDN or TDP) for the year 2010, see SI 3 for details) [10,20,21,42].This especially holds for the Southern Hemisphere, the Subarctic and Great Plains in North America, and parts of Asia [20].However, in 2010, a large share of the population already lived in point sourcedominated sub-basins (37% for TDN, 54% for TDP, 24% for TDN + TDP, see SI 3 for details) [20].In addition, the share of point sources is expected to become more important in the future under projected increasing urbanisation trends.This is the focus of our study.Second, we only modelled a limited cocktail of pollutants (i.e.TDN, TDP, MP and TCS), while coastal waters receive numerous pollutants.Third, we ignored potential biogeochemical interactions among pollutants.For example, chemical additives may leach from MP, causing hazardous solids to be a secondary source of chemicals in rivers [43].Fourth, we acknowledge limitations in our modelling approach for pollutant retention calculations (see SI 3).For example, microplastic retention in rivers depends, amongst others, on discharge [44] and extreme weather events [45,46], which are not considered in our model because our model provides annual average pollution levels at the sub-basin scale.Fifth, our scenario assumptions may have resulted in over-or underestimations of future water pollution.We believe that this did not affect the main messages of our manuscript.This is because we focussed on a large-scale analysis using annual averaged pollution levels.We focussed on point sources and their impacts.Our model results are not suitable for assessing local water quality issues.
We believe that insights of our study could contribute to society and science in three ways.First, identifying future multi-pollutant hotspots for coastal waters from point sources by sub-basin, which is relevant for SDG14.Such information could help to facilitate timely actions to mitigate impacts on coastal waters.This particularly holds for people living in upcoming hotspot regions and datapoor regions such as Africa.Moreover, local mitigation efforts may have global benefits [47].Second, we increase our understanding of the drivers of future point-source-associated coastal water pollution by sub-basin.Such information might be useful for mitigating regional impacts and identifying solutions [47].Solutions like plastic banning or recycling may vary among world regions depending on drivers and regional characteristics [48].Third, we contribute to supporting the design of reduction strategies for multi-pollutant issues in coastal waters under different socio-economic development scenarios.Using a multi-pollutant perspective when designing reduction strategies might be more effective than a single-pollutant perspective due to synergies and trade-offs in pollution control [9].Moreover, this study indicates that socio-economic developments can reinforce positive and negative impacts on future coastal water pollution depending on regional characteristics.This shows that water pollution is a complex issue, calling for science and policy interventions that account for linkages between pollutants and regions.

Conclusions
We developed an improved version of the MARINA-Multi model for projecting point-source-associated coastal water pollution for sub-basins worldwide.Our insights show that, globally, river exports of all pollutants from point sources are projected to more than double between 2010 and 2100 under a reference scenario, especially in Africa and Asia.Socioeconomic developments such as improved wastewater treatment, increased economic growth and city expansion can either have positive (less pollution) or negative (more pollution) impacts on future coastal water pollution.Combined, these socio-economic developments may reinforce regional differences in pollution, with average changes of −21% to +50% in coastal water pollution compared to a reference scenario.Reducing coastal water pollution is of utmost importance for conserving coastal ecosystems, as targeted by Sustainable Development Goal 14. Future research may focus on further improving the modelling approach by accounting for diffuse sources, a comprehensive pollutant cocktail including biogeochemical interactions, and linking the results to regional policy developments for reduction strategies accounting for pollutant-region linkages.Such research developments would support initiatives like the World Water Quality Alliance and the Inter-Sectoral Impact Model Intercomparison Project.

Figure 1 .
Figure 1.River exports of pollutants to coastal waters from point sources by sub-basin in 2010 (left panels) and 2100 (right panels).Pollutants include total dissolved nitrogen (TDN; upper panels), total dissolved phosphorus (TDP; upper middle panels), microplastics (MP; lower middle panels) and triclosan (TCS; lower panels).SSP2-RCP2.6 is a reference scenario (see Methods for details).SSP is short for Shared Socio-economic Pathway.RCP is short for representative concentration pathway.Source: the MARINA-Multi model (see section 2 for the model and scenario descriptions).

Figure 2 .
Figure 2. Alternative scenarios (AS1-4) to show the individual and combined impacts of wastewater improvements (AS1 & AS4), economic growth (AS2 & AS4), and city expansion (AS3 & AS4) on coastal water pollution from point sources in 2100 compared to a reference scenario (see Methods for details).(a) Reflects the impacts on coastal water pollution on a global scale.(b) Reflects the impacts on coastal water pollution among world regions.The impacts on coastal water pollution are reflected by changes in river exports of the pollutants to coastal waters compared to the reference scenario (i.e. the purple line).Changes in pollutant loads below or within the purple line reflect positive impacts on coastal water pollution (less pollution).Changes in pollutant loads above or outside the purple line reflect negative impacts on coastal waters (more pollution).The reference scenario accounts for moderate socio-economic developments and low climate impacts (SSP2-RCP2.6) by 2100.SSP is short for shared socio-economic pathway.RCP is short for representative concentration pathway.Pollutants include total dissolved nitrogen (TDN), total dissolved phosphorus (TDP), microplastics (MP), and triclosan (TCS).Source: the MARINA-Multi model (see section 2 for the model and scenario descriptions).