Reframing water-related ecosystem services flows

To sustain water-dependent economic and socio-ecological systems, natural capital and its interactions with other capitals is gaining attention, but a clear understanding of how to manage natural capital sustainably and how to make decisions relevant to water-related ecosystem services is yet to be achieved. In this study, we extended the framing of water-related ecosystem service flows as a cycle, integrating water quantity and quality and capturing the flows of ecosystem services (i.e., green phase) and ecosystem disservices (i.e., red phase), and connecting natural capital, built capital, and beneficiaries. We applied this framework to the Jiulong River watershed in China, using hydrological models to model water quantity and quality based on historical observations and experimental data. Our results showed that, during the green phase, the interactions of natural capital and built capital significantly improved water quality in downstream areas with higher flows. During the red phase, built capital reduced ecosystem disservices by ~10% while natural capital further reduced it by over one half. Our framework can provide information for natural capital management, eco-compensation, and pollutant management relevant to water-related ecosystem services. Synopsis: We extended the framing of ecosystem service flows as a cycle to inform the management of water-related ecosystem services. beneficiaries, and their complex interactions. Our perspective


Introduction
The concept of natural capital is gaining worldwide attention for its ability to inform sustainable management of social-ecological systems. Natural capital refers to the living and non-living components of ecosystems that directly or indirectly provide services for human beings (Guerry et al., 2015). Yet, natural capital cannot provide services to beneficiaries without the presence of people (i.e., human capital), communities (i.e., social capital), and the built environment (i.e., built capital; Costanza et al., 2014). Natural capital must therefore interact with other forms of capital to realize its benefits (Costanza et al., 2017).
Recent studies have measured the provision of ecosystem services (ES) by accounting for not only stocks of natural capital but also other forms of capital. For example, the effects of human-derived products, technology, and infrastructure (i.e., built or manufactured capital) as well as trust, community involvement, and social cohesion (i.e., social capital) are increasingly being considered in combination with natural capital (Bateman and Mace, 2020;Barnes-Mauthe et al., 2015;Chen and Graedel, 2015;Jone et al., 2016).
A broader consideration of the influence of multiple types of capital is essential to meeting the challenge of sustainability of natural capital and ES (Schröter et al., 2017).
Despite much progress in recent decades, a clear understanding of how natural capital interacts with other capitals to produce ES flows to beneficiaries remains elusive (Jones et al., 2016). Several elements have been proposed to improve this understanding, including measuring ES supply, demand, and beneficiaries (Lin et al., 2020); characterizing the whole ES chain (Mandle et al., 2020); and identifying mediating factors in ES delivery (Mandle et al., 2020).
However, ES flow was initially defined as a single flow from suppliers to beneficiaries (Bagstad et al., 2014), and current thinking has not expanded on the ES chain from this one-way flow (Jone et al., 2016;Koellner et al., 2018;Palmer and Rubi, 2019), leading to only a partial understanding of ES dynamics. For some ESs, conceptualizing ES flows as a cycle that is affected by different types of capital, instead of a one-way flow from suppliers to beneficiaries, can enable a more comprehensive and sophisticated understanding of ES dynamics. In addition, few studies have considered what becomes of ESs once they have been used by beneficiaries. For at least some ecosystem goods and provisioning services, such as freshwater and food, byproducts are generated following use (Martin-Ortega et al., 2015). This often takes the form of degradation of the resource or stock in terms of quality or quantity (Maseyk et al., 2017).
Water is one component of natural capital and water-related ESs such as freshwater supply (i.e., water quantity) and water purification (i.e., water quality) show complex cyclical dynamics (Keeler et al., 2012). Hydrological processes determine ecohydrological functions through impacts on water quantity, quality, location, and timing and their effects on water availability to beneficiaries (Brauman, 2015). This waterrelated ES is provided to different beneficiaries such as industry, agriculture, and human drinking, generating water-related ES flows (Bagstad et al., 2014). ES flows interact with different forms of capital, where, for example, industry and domestic sectors obtain water from treatment plants and pipes (i.e., built capital) (Meehan et al., 2020).
This built capital thus provides and mediates essential services for human well-being (Weisz et al., 2015).
To illustrate and analyze the complex interactions between capital types in waterrelated ES flows, here, we focus on nitrogen, an essential nutrient required for the maintenance of Earth's biological systems (Palmer et al., 2017). Substantial amounts of nitrogen are released into ecosystems via run-off, leading to increases in riverine nitrogen concentrations and eutrophication (Clark et al., 2017;Guo et al., 2020;Motallebi et al., 2018). This phenomenon degrades the eco-hydrological function of aquatic ecosystems, thereby converting water-related ESs into an ecosystem disservice (ED;Krkoška et al., 2016). EDs are negative environmental effects resulting from anthropogenic impacts, including the release of substances or side effects of the deliberate manipulation of ecosystems (Lyytimäki et al., 2009). EDs interact with ESs (von Döhren and Haase, 2015) where, for example, applying more nitrogen enhances vegetation growth and water interception by plants, thereby affecting water availability Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 12 January 2021 doi:10.20944/preprints202101.0210.v1 6 for people (Jones et al., 2014).
Biophysical models provide a useful framework to capture interactions between different forms of capital and ES flows (Kim et al., 2017). Water-quality models can estimate how changing capital and ES management affect nitrogen loading to downstream endpoints of a watershed (Keeler et al., 2012). They can also be used to identify nitrogen sources and retention processes in terrestrial and aquatic ecosystems by accounting for nitrogen discharge originating from effluents (e.g., drained from built capital) or agricultural activities (i.e., beneficiaries; Alexander and Gorman Sanisaca, 2019;Zhou et al., 2018). Basin-scale runoff models could be applied to simulate water quantity from various land uses by computing hydrological cycle components such as evapotranspiration, infiltration, interflow, and base flow (Lin et al., 2020). As water quality is sensitive to changes in water quantity, it is important to integrate these two aspects when quantifying interactions between different forms of capital and waterrelated ESs (Murphy and Sprague, 2019).
Here, we propose a new framework to account for the complex interactions among natural capital, built capital, beneficiaries, and the site-specific flows of waterrelated ESs. To evaluate and demonstrate its effectiveness, we apply the framework in the Jiulong River watershed in China. The study area supports extensive crop and livestock production, has experienced rapid urbanization, and provides drinking water to over 12 million people. We first assessed water quantity using the Hydrological

Study framework
We reframed the conceptualization of water-related ES flows as a cycle comprising a green phase and a red phase with three components: natural capital, built capital, and beneficiaries ( Figure 1). Of these components, built capital works as a mediating factor.
Natural capital serves as a sending system while beneficiaries act as a receiving system.
We considered water and its environment as natural capital following the framework of Alternatively, EDs may be delivered via wastewater treatment plants, which partly limit pollutant loads and ecological impacts. Additionally, ecological restoration and rehabilitation programs promote green infrastructure to reduce EDs from wastewater treatment plants. The effluent discharged from built capital is then purified by natural capital and stored as natural capital assets.  (2018)). The green arrow indicates improvements in water quality, while the red arrow indicates decreases in water quality.

Study area
The Jiulong River watershed is located in the southeastern area of Fujian Province, China. The Jiulong River includes two major tributaries, the North River and West River, with a drainage area of ~1.4 million ha converging in Zhangzhou and Longyan Cities 9 ( Figure 2). The watershed has an annual discharge flow of ~1.2 × 10 6 GL drained into estuarine and coastal waters. As an essential water source for Longyan, Zhangzhou, and Xiamen Cities, the Jiulong River supports over 12 million people as of 2019 according to the Statistics Yearbook in Longyan (2020), Zhangzhou (2020), and Xiamen (2020).
Agriculture has intensified in the Zhangzhou plain since the 1990s with orchards of bananas, longans, red grapefruits, and lychees as well as livestock breeding, leading to the eutrophication of water bodies (Zhou et al., 2018). Significant amounts of effluent and manure from piggeries have been discharged into the North River, leading to a large dinoflagellate bloom in the watershed in January 2009 (Du et al., 2013). Pollutant management, particularly nitrogen, is therefore critical to water resources and sustainability in the study area.

Natural capital
To characterize the ecosystem services cycle, we first quantified natural capital in the Sub-catchments were merged into 11 basins as indicated in Figure  We assessed water quality based on total nitrogen (TN) concentration using the RSPARROW model. Three types of variables are necessary to model water quality in RSPARROW: pollutant sources, land-water delivery factors, and other fixed variables.
All data sources are detailed in Table S1. As it is difficult to separate nitrogen sources into different species, such as sources of ammonia nitrogen (NH4 + ) and nitrate (NO3 -) from fertilizers, we modeled RSPARROW-TN as a proxy for water quality. However, according to the environmental quality standards for surface water in China ( GB3838-2002), there is no benchmark for riverine water quality regarding TN. Therefore, the concentration of TN was converted to NH4 + by multiplying TN by 0.3 following the analysis of historical monitoring data (2015-2018; Figure S4).

Beneficiaries
Following Lin et al. (2020), we identified water-related ES beneficiaries as the actual anthropogenic use (i.e., consumption) of water-related ES, including by agricultural, domestic, public, and industrial sectors. Annual water demand in each grid cell was calculated as: Where , , , and represent public, domestic, industrial, and agricultural demand for water, respectively. indicates the average daily residential water use per capita in each city (L person -1 day -1 ), while indicates the average daily public water use (e.g., by government) per capita in each city (L person -1 day -1 ).
Factors 365 and 1,000 are used to convert domestic water demand from L person -1 day -1 to m 3 person -1 year -1 . represents the average annual industrial water use for each RMB 10,000 (~1,422 USD) of industrial value added (i.e., m 3 per RMB 10,000 industrial value added). V is the industrial value added per hectare (RMB 10,000 ha −1 ).
represents the average annual water use for irrigation per hectare (m 3 ha -1 ). We assumed that agriculture always received water nearest to its location with no effect of water quality.

Water-related ecosystem services cycle
The assessment of the ES cycle included ES flow and ED. For ES flow, differences between the 30-m resolution spatial layers of water-related ES supply and beneficiaries were calculated. We summed up the differences in each basin as water surplus. When ES flow satisfied the demand of a local area, the remaining water flowed from basin to basin, accumulating from upstream to downstream following the hydrography of the basin.
EDs equaled the total amount of nitrogen load drained from beneficiaries, such as sewage from domestic use, industry, deposition from precipitation, and agriculture to the environment. There are three phases of nitrogen loads: 1) loads generated by beneficiaries, 2) loads discharged after removal from built capital, and 3) riverine China Pollution Census (SCIES, 2008). The production coefficient was replaced with the effluent coefficient to compute nitrogen loads discharged from livestock. Detailed information on production coefficients and effluent coefficients can be found in Table   S4. Fertilizer application data were derived from the Statistical Yearbook (2018)

Natural capital
During calibration, the HSPF model predicted water quantity with a monthly ENS of 0.79 and R 2 of 0.80 ( Figure S1). During validation, we computed an ENS of 0.65 and R 2 of 0.61 in the North River. For the West River, monthly ENS was 0.65 and R 2 was 0.78 during calibration, while monthly ENS was 0.65 and R 2 was 0.66 during validation.
Those indicators indicated good model performance. A scatterplot of observed versus predicted water quality (i.e., nitrogen load) is shown in Figure S2. Model fit was generally unbiased with efficient performance in the study area ( Figure S2). A closer look at the overall statistical fit of the RSPARROW model indicated that model skill was strong for predicted load (RMSE = 0.39-0.40, R 2 = 0.93, R 2 Yield = 0.92; Table S2).
Sub-catchments with small areas had higher uncertainty due to limitations in the RSPARROW model. Based on environmental quality standards in China, most water quality classes in the study area were in classes I to III, while some sub-catchments exhibited poor water quality (classes IV to >V; Figure S2).
Seven parameters were selected from an initial 17 parameters to achieve robust model performance with 95% confidence intervals (Table S3). Sewage discharge (including domestic and industrial discharge), fertilizer application, and livestock manure were identified as key nitrogen sources in the study area. Drainage density and slope were identified as major land delivery factors, while stream decay variables were characterized as key water delivery factors. All variance inflation factors were low (<10), indicating that the selected parameters were statistically independent from one another.

Beneficiaries
Hotspots of water withdrawn by industry were mostly distributed in the central urban areas of Longyan City, Yanshi Basin ( Figure S3). Water consumed by the public and domestic sectors was related to population density. The Yanshi Basin, Huashan Basin, and the outlet of the Punan Basin were hotspot areas of water withdrawn for human drinking and government use, showing similar trends as industrial water consumption.

Ecosystem services flow
Water-related ecosystem services supply, beneficiaries, and net change in 11 basins are shown in Figure 3. We found high spatial heterogeneity in water ES supply (quantity and quality) among basins. Basins that provided larger amounts of water showed higher water quality (i.e., lower environmental standards classes), for example as observed in the Zhangping downstream Basin. Some basins contributed less water, and those were typically characterized by a single classification indicating poor water quality (e.g., class >V for the Chuangchang Basin). The large difference (approaching 1,200 GL) between the largest and smallest basin indicated high heterogeneity in water quantity.
Compared to water supply, water consumed by various beneficiaries was more homogeneous, and industry and agriculture consumed the largest amounts of water ( Figure 3b). High water supply and small consumption by beneficiaries contributed to large net surpluses in all but one basin (Figure 3c). ES net surplus was mostly driven by high ES supply, except for the Yanshi and Punan Basins, which experienced smaller surpluses. Zhangzhou Basin showed an ES deficiency, as it consumed more water than was supplied (Figure 3b). Overall, water quality improved from upstream areas to downstream areas, with high water quantity in the study area. In the North River, water-related ES flow accumulated from the mid-stream area to the downstream area, accompanied by water quality improvement (Figure 4). Compared to the North River, the West River provided less water, which was of poorer quality in both upstream and downstream regions ( Figure 4).

Ecosystem disservices
The interactions of built capital and natural capital contributed to a decrease in EDs (i.e., nitrogen load; Table 1). Nitrogen transport in natural capital diminished remarkably compared to its delivery from built capital in the whole watershed. In comparison to the North River, the West River showed small reductions in nitrogen from built capital but higher reductions from natural capital. Proportions of nitrogen removal from built and natural capitals were higher in downstream areas.

Ecosystem service cycle
The interactions of natural capital, built capital, and beneficiaries contributed to spatial heterogeneity in ES flow and EDs in the study area ( Figure 5). With a large amount of water, the North River served as a water ES provider for the whole watershed, with good water quality. This was attributed to the installation of drinking water treatment plants, wastewater treatment plants, and their interactions with natural capital, especially in the Hua'an and Punan Basins. Some sub-catchments suffered from waterrelated ES flow deficiencies, such as ES flows from sub-catchments 42 to 41 and 41 to 37 ( Figure 5), which were determined by dynamic interactions between natural capital and beneficiaries. These interactions also resulted in the spatial heterogeneity of EDs.
Agricultural beneficiaries widely distributed in the West River (e.g., orchards and livestock breeding) generated more EDs, leading to poor water quality and lower water quantity in the West River. Similar trends occurred in the Yanshi Basin of the North River, as most livestock were bred in Longyan City.

Discussion
In this study, we extended the framing of ES flows as a cycle to illustrate the interactions among natural capital, built capital, and beneficiaries within water-related ES. We applied the framework in the Jiulong River watershed, an important water source for the southern Fujian province, which suffers from serious nitrogen pollution (Du et al., 2013). We show that the water-related ES cycle is connected by a green phase (ES flow) and a red phase (EDs; Figure 1). In the green phase, the interactions of natural capital and built capital significantly improve water quality in areas with large amounts of water. In the red phase, built capital reduces ED by ~10% while natural capital reduces this further by over one half. Our new framework can be used to inform the sustainable management of natural capital via eco-compensation schemes, nitrogen management strategies, and built capital interventions.
Water quantity and quality both strongly affected water-related ES supply, with water quantity determining the availability of sufficient ES supply (Grizzetti et al., 2016). Water yield in the study area was influenced by the spatial heterogeneity of climate conditions (including precipitation, evaporation, and temperature), basin topology (including river density and catchment size), and land-use change (Kindu et al., 2016;Lin et al., 2020). This generated large differences in water quantity among basins. With limited water quantity, ES deficiency occurred in the Zhangzhou Basin, making it a recipient of ES from other basins.
The reliability of water ES supply depended on its quality (Ma et al., 2020a). Water quality was affected by the quantity of pollutant sources, eco-hydrological processes, and pollutant control policies (Alexander and Gorman Sanisaca, 2019;Ma et al., 2020b).
Regions with lower ES supply (e.g., the Huashan, Chuangchang, and Yongfeng Basins) exhibited poor water quality, which exacerbated water ES scarcity (Mortazavi-Naeini et al., 2019). Changing water quantity and quality also affect many aspects of benefits and costs incurred by beneficiaries (Keeler et al., 2012). Compared to ES beneficiaries in the domestic and public sectors, ES beneficiaries in the industrial and agricultural sectors consumed large amounts of ES, which may indicate that beneficiaries from agriculture and industry seize water ES surplus to satisfy their needs (Wang et al., 2015). via chemical treatments might be necessary (Ma et al., 2020b). Beneficiaries from the Zhangzhou Basin need to invest in water quality or pay extra for those services, or else water quality will exacerbate regional water resource inequality (Ma et al., 2020a).
In summary, the green phase of the ES cycle included water ES flows from upstream regions to downstream regions, with improving water quality due to increasing water quantity in downstream regions and dilution (Rice and Westerhoff, 2017). The interaction of natural capital and built capital also affected water-related ES supply. As a mediating factor, drinking water treatment plants contributed to improvements in water quality (Aristi et al., 2015). Built capital (e.g., drinking water treatment plants and piped water connections) served as a connector between natural capital and beneficiaries (Meehan et al., 2020). Drinking water treatment plants managed by the local government can mediate water-related ES demand by setting water prices, with a direct influence on people's willingness to pay (Mu et al., 2019).
For example, the urban water pricing reform that adopted increasing block rate tariffs in 2002 in China has reduced annual residential water demand by 3-5% during 2002-2009, which help address growing water scarcity (Zhang et al., 2017).
The red phase included the flows of EDs transported from beneficiaries to natural capital, thereby degrading the ecosystem. In this study, EDs were derived from effluents (Lyytimäki et al., 2009), with the quantity of generated EDs determined by pollutant sources such as point sources (i.e., sewage discharge) and non-point sources (i.e., livestock manure, wet deposition, and fertilizer application; Zhou et al., 2018).
Intensive agricultural activities such as livestock breeding always led to poor water quality in host basins, for example as in the Chuangchang Basin ( Figure 5) (Harrison et al., 2014;Hallouin et al., 2018;Pommier et al., 2018). It is important to note that although nitrogen is lost when transported in the socio-ecological system, the effluents of wastewater treatment plants and most nonpoint sources in watercourses alter many features of freshwater ecosystems and unavoidably degrade water quality (Aristi et al., 2015). Some nature-based solutions, such as green infrastructure (e.g., ecological ditches, buffer strips, and vegetated ponds) can influence pollutant retention in agricultural fields (Berg et al., 2016;Wu et al., 2017).
Our result also indicated that basins with lower ecosystem function and built capital tended to provide more EDs than ES. For example, the downstream areas of the North River, such as the Punan Basin, with a high built capital and natural environment capacity, provided large amounts of cleaner drinking water, while basins with less built capital supplied more EDs (e.g., the Chuangchang Basin). This finding extends existing work that suggests a positive correlation between ecological function and the production of ES and EDs to include both natural environment and built capital (Campagne et al., 2018).
Our study was affected by a number of limitations. In both Punan and Zhengdian stations, HSPF model validation performed poorly due to data accessibility and availability issues ( Figure S1). on historical records. There was a standard deviation for the fixed coefficient of ±10% between the North River and the West River, and a ±3% standard deviation among seasons, and 9.85% among the 18 sampling sites. Finally, we only considered waterrelated ES in this study, hence, our framework remains to be extended to further types of ES (i.e. food supply, recreational benefits).
An ES cycle perspective could provide guidelines for eco-compensation. In China, eco-compensation combines ecological compensation and payments for ES, a combination of punitive mechanisms following the 'polluter-pays' principle and incentive-based mechanisms following the 'beneficiary-pays and provider-gets' principle (Shang et al., 2018). The ES cycle characterized ES flows from providers to beneficiaries, providing information for incentive-based mechanisms for ES payments.
The ES cycle also illustrated how pollutants led to EDs, potentially informing 'polluterpays' mechanisms. Interactions of different capitals and their impacts on water-related ES can provide a more detailed information for eco-compensation. For example, beneficiaries from areas with lower pollutant retention from ecosystems and built capital may need to pay more for ecological restoration (La Notte et al., 2015).
Nitrogen management strategies, such as improving nitrogen-use efficiency, source reduction, pollutant retention, and restoration or nitrogen resources reallocation, are all practical ways to reduce nitrogen carried into the ecosystem to maintain ecosystem function (Mueller et al., 2017;Wu et al., 2017;Gerber et al., 2016). However, links between such proposed strategies and different capitals are poorly defined. The ES cycle perspective can provide this missing link between pollutant management and different capitals. For example, the contributions of natural capital and built capital to EDs, such as nitrogen removal, nitrogen retention, and waste decomposition, can provide guidelines for pollutant control (Wu et al., 2017). For example, policymakers should balance water-related ES and EDs to decrease the cost of drinking water treatment and wastewater treatment by assessing gross ecosystem products (Ango et al., 2014;Price et al., 2018;Ouyang et al., 2020).
Our study characterized a water-related ES cycle consisting of natural capital, built capital, beneficiaries, and their complex interactions. Our perspective highlighted that 23 EDs can be reduced by investing in both built capital and natural capital. More effective strategies are needed to support the sustainable management of natural capital, which aims to ensure the flow of benefits by maintaining the condition of natural capital stocks (Bateman et al., 2020). Built capital, such as gray and green infrastructure, also plays an important role in providing benefits through well-designed interventions (Chen and Graedel, 2015). Yet, we should also be aware that built capital consumes natural resources and induces impacts on the environment, such as climate change and habitat loss (Weisz et al., 2015).