Megacity development and the demise of coastal coral communities: Evidence from coral skeleton δ15N records in the Pearl River estuary

Historical coral skeleton (CS) δ18O and δ15N records were produced from samples recovered from sedimentary deposits, held in natural history museum collections, and cored into modern coral heads. These records were used to assess the influence of global warming and regional eutrophication, respectively, on the decline of coastal coral communities following the development of the Pearl River Delta (PRD) megacity, China. We find that, until 2007, ocean warming was not a major threat to coral communities in the Pearl River estuary; instead, nitrogen (N) inputs dominated impacts. The high but stable CS‐δ15N values (9‰–12‰ vs. air) observed from the mid‐Holocene until 1980 indicate that soil and stream denitrification reduced and modulated the hydrologic inputs of N, blunting the rise in coastal N sources during the early phase of the Pearl River estuary urbanization. However, an unprecedented CS‐δ15N peak was observed from 1987 to 1993 (>13‰ vs. air), concomitant to an increase of NH4+ concentration, consistent with the rapid Pearl River estuary urbanization as the main cause for this eutrophication event. We suggest that widespread discharge of domestic sewage entered directly into the estuary, preventing removal by natural denitrification hotspots. We argue that this event caused the dramatic decline of the Pearl River estuary coral communities reported from 1980 to 2000. Subsequently, the coral record shows that the implementation of improved wastewater management policies succeeded in bringing down both CS‐δ15N and NH4+ concentrations in the early 2000s. This study points to the potential importance of eutrophication over ocean warming in coral decline along urbanized coastlines and in particular in the vicinity of megacities.


| INTRODUC TI ON
In the second half of the 20th century, humanity entered a phase of unprecedented urbanization. Today, more than 50% of the global population lives in a city (Grimm et al., 2008), and urban population is expected to rise to 80% by 2050 (Bettencourt & West, 2010). The world's urbanization does not spread homogeneously over the globe, but is concentrated mainly within coastal areas (Tibbetts, 2002). Coastlines host 45% of the global population, with an average population density over a 100 ind.km −2 , which is >2.5 times the mean global density. The world's highest population densities are found in immense urbanized areas hosting >20 million people, which result from the clustering of adjacent cities. These urban complexes are often referred to as "megacities" (Tibbetts, 2002). The world currently hosts 12 megacities, 10 of them located in coastal areas ( Figure 1a). As such, the environmental footprint of humankind is disproportionately high within coastal areas, leading to concerns over the fate of coastal marine habitats.
The ocean is a common sink for anthropogenic N sources (sewage, artificial fertilizers, NOx), with the transfer occurring either indirectly (through atmospheric deposition, groundwater, or river discharge) or directly (by sewage outfall). Excess N is detrimental to the marine ecosystems and it has been linked to a myriad of cascading effects: anoxic zones formation (Rabalais, Turner, & Wiseman, 2002), harmful algal blooms (NRC, 2000), and reduction in biodiversity (Fabricius, De'ath, McCook, Turak, & Williams, 2005).
The world's coastlines host sensitive and diverse ecosystems, such as salt marshes, mangroves, forests, and coral reefs, that provide F I G U R E 1 The urban development on the world's coastlines is a rising threat for coral reefs. (a) The map shows the population density (source: CIESIN, 2017), the world's megacities (source: UN, 2018), coral reefs (source: UNEP-WCMC, WorldFish Centre, WRI, TNC, 2010), and the world's hotspot of coral biodiversity, the Coral Triangle (Veron et al., 2011). (b) Detail map of the Pearl River Delta region showing the eight major cities composing the megacity and the Hong Kong Special Administrative Region (HKSAR) and their populations. The four sites sampled for this study are indicated with outlined symbols. The coral communities having experienced a major decline are highlighted in color. Maps were made with the geographic information software QGIS (www.qgis.com) (a) (b) critical services. Coral reefs in particular, provide a wealth of environmental services that sustain millions of people throughout the world: food (Hughes et al., 2012), income (Spalding et al., 2017), coastal protection (Ferrario et al., 2014), biodiversity (Knowlton et al., 2010), and medicinal potential (Bruckner, 2002). Coral reefs are vulnerable to rising temperatures (Hughes et al., 2018), overfishing (Maire et al., 2016;Newman, Paredes, Sala, & Jackson, 2006), and eutrophication (Duprey, Yasuhara, & Baker, 2016). In particular, coastal urbanization is now exerting a strong pressure on coral reefs near two megacities located at the edge (Jakarta, the Pearl River Delta [PRD]) and within (Manila) the Coral Triangle, the world's coral biodiversity hotspot ( Figure 1a). Information is desperately needed on the role of coastal urbanization on the N budget of coral reef ecosystems.
Evaluating the respective impacts of global change (i.e., ocean warming and/or acidification) and regional processes (i.e., eutrophication) on coral reef ecosystems can be challenging. Satellites and underwater loggers are continuously monitoring the world ocean's temperature while coral geochemical records (e.g., stable oxygen isotopes δ 18 O) extend these records over the last 300 years (Hennekam et al., 2018;Zinke, Loveday, Reason, Dullo, & Kroon, 2014). Yet, the historical and current pressure of eutrophication on coral reefs is much more difficult to assess. Indeed, the lack of reliable environmental baselines and the challenge of identifying the anthropogenic N sources responsible for coastal eutrophication hamper our efforts to reduce anthropogenic impacts on marine ecosystems by designing and implementing effective N mitigation policies. Coastal areas with a long history of urbanization provide key case studies on the history of anthropogenic N footprint on the marine environment.
The PRD (Guangdong province, China) has been experiencing intense urbanization for at least the last 200 years, with a population of 1.5-2 million reported in the city of Guangzhou at the turn of the 19th century (Benedict, 1988;Gauducheau, 1911). In the 1980s, the start of the PRD Economic Zone resulted in the clustering of eight major cities in the delta (Guangzhou, Foshan, Dongguan, Jiangmen, Zhongshan, Zhuhai, Shenzhen, and Huizhou) into one of the most populated megacities, gathering ~120 million inhabitants ( Figure 1b).
Today, the Pearl River estuary is notorious for its severely eutrophic waters with DIN concentrations comprised between 2 µM and above 50 µM toward the rivermouth (Ye, Ni, Xie, Wei, & Jia, 2015), whereas typical DIN concentrations on coral reefs are generally less than 1 µM (Bell, Elmetri, & Lapointe, 2014). This eutrophy is responsible for coral biodiversity loss (Duprey et al., 2016) and harmful algal blooms that have severely affected the fish aquaculture economy in the area (Yin, Harrison, Chen, Huang, & Qian, 1999). Paradoxically, the coastlines surrounding the Pearl River estuary are still home to coral communities hosting about 90 scleractinian coral species (Duprey et al., 2016;Veron et al., 2011), including the long-lived genus Porites , offering the opportunity to investigate the link between the evolution of the N footprint of the PRD megacity and the decline of coral communities over time.
The nitrogen isotopic composition of hard coral's skeletal organic matrix where δ 15 N = ([( 15 N/ 14 N) sample / ( 15 N/ 14 N) air ] − 1) × 1,000) has been shown to reflect the isotopic composition of the nitrogen supply to coral reef (Wang et al., 2016). Long-lived hard coral, such as the genus Porites, can thus be used to track changes in natural and anthropogenic N sources over time (Duprey, Wang, et al., 2017;Wang et al., 2018) . Here, we report CS-δ 15 N records spanning the last 5,000 years from coral fragments recovered from sedimentary deposits, museum-held coral specimens, and coral cores. The objectives of this study are (a) to produce an environmental baseline by identifying the natural sources/processes regulating the N cycle in the Pearl River estuary before the PRD megacity started to develop, (b) to document the evolution of the N footprint of the Pearl River over the last 5,000 years, and (c) to compare this evolution to the decline of coral communities located in the Pearl River estuary.

| Hard coral material
All the material used in this study was collected on the eastern edge of the Pearl River estuary, within the Hong Kong Special Administrative Region-HKSAR ( Figure 1b). Local coral communities are typically found at depths less than 10 m (Thompson & Cope, 1980). For the present study, only scleractinian (i.e., photosymbiotic) genera were selected, to ensure that the skeletal geochemistry would record the upper part of the water column (i.e., <10 m) and that isotopic data from different corals were comparable (Wang et al., 2014(Wang et al., , 2016. Two specimens of the massive coral Porites (references HK001 and HK006-The University of Hong Kong coral cores repository) were selected for this study. Museum-held and sedimentary coral samples used in this study are listed in Tables S1-S4.

| Age model and dating
The age model of coral HK006 was based on growth band counting. The age model of core HK001 is complicated by the two growth stops ( Figure 2). The age model was thus based on a reanalysis of the original material of Wang et al. (2011) and from additional slabs.
X-rays of all slabs were obtained at the Ocean Park Veterinary Hospital  (Ramsey, 2009(Ramsey, , 2017 using the updated Marine13 radiocarbon age calibration curve (Reimer et al., 2013). Additional information about the calibration is available in the Supporting Information section: Calibration of the radiocarbon dates.

| Coral skeleton stable nitrogen isotope ratio (CS-δ15N) analysis
HK001 and HK006 were sampled at an annual resolution by milling each growth band using a diamond burr mounted on a handheld drill. Coral fragments (museum and sedimentary) were ground using a mortar and pestle. Coral samples were then sieved and the fraction with a grain size of 63-250 µm was kept for analysis in combusted (500°C-5 hr) borosilicate vials. CS-δ 15 N analysis was by persulfate oxidation to nitrate followed by nitrate isotopic analysis with the denitrifier method (Sigman et al., 2001;Wang et al., 2014;Weigand, Foriel, Barnett, Oleynik, & Sigman, 2016). Samples were prepared at The School of Biological Sciences, The University of Hong Kong, HKSAR and analyzed at the Sigman Laboratory at Princeton University, NJ, USA and replicated at the Martínez-García Laboratory at the Max Planck Institute for Chemistry, Mainz, Germany. In both laboratories, international amino acid standards (USGS-40 and USGS-41) and nitrate standards (IAEA-NO 3 and USGS34) were used for calibration and blank correction, and an in-house coral standard was used to monitor reproducibility. The average analytical reproducibility was 0.17‰ (n = 22; Figure S1).
More information about the analysis can be found in the Supporting Information section: Coral skeleton stable nitrogen isotopes analysis.

| δ 18 O analysis
Coral skeleton δ 18 O is widely used as a proxy for sea surface temperature (SST) and/or sea surface salinity (SSS; Weber & Woodhead, 1972). To assess the impact of climate change on the coral communities of the Pearl River estuary, 100 µg aliquots of HK001 samples were analyzed for δ 18 O in the inorganic stable isotopes Laboratory at The Max Planck Institute for Chemistry, Mainz, Germany. Two inhouse carbonate standards, calibrated to international standards, were used to ensure the quality of the analysis. The analytical reproducibility was 0.12‰ (n = 192; 1SD).

| Environmental datasets
The Hong Kong Environmental Protection Department water quality monitoring program (www.epd.gov.hk) provided monthly SST  Figure S2). More information about the datasets used to construct this time series can be found in Table S5.

| PRD demographic data
Population time series over the period 1950-2010 were compiled for each of the eight major cities of the PRD megacity (Guangzhou, Foshan, Dongguan, Jiangmen, Zhongshan, Zhuhai, Shenzhen, Huizhou) and for the HKSAR using the United Nation Report: Revision of the World Urbanization Prospects (UN, 2018). Population data were available for years 1950, 1960, 1970, 1980, 1990, 2000, and 2010 and the population evolution was expressed as the average percentage of population change between each of these years, for each city ( Figure 3g; Table S6).
F I G U R E 2 δ 18 O and CS-δ 15 N records from Porites core HK001. The analytical error is 0.12‰ for δ 18 O and 0.17‰ for CS-δ 15 N analyses

| Environmental datasets
The Pearl River estuary has a strong seasonality marked by an average summer (June-July-August) SST of 28.2 ± 1.4°C and average winter (December-January-February) values of 18.0 ± 2.2°C.

Pearson Correlation).
Seasonal hydrological changes (rainfall, Pearl River outflow) results in a 2 psu drop in salinity (30.0 ± 2.5 psu) in summer relative to winter values (32.5 ± 0.9 psu). No trend is observed in the

| Population change in the PRD cities
The population of the major PRD cities was stable from 1950 to 1980 (Table S6). In 1950, aside from Guangzhou and Hong Kong, which had already populations >1 million, none of the seven other cities of the delta had populations above 100,000 inhabitants. The entire population of the delta remained close to 8 million until 1980.

| Coral core HK001
The age model reanalysis of core HK001 provided the following Although both δ 18 O and CS-δ 15 N records did present decadal fluctuation, no correlation was found between δ 18 O and CS-δ 15 N for each of the core sections (p > .05; Pearson correlation), indicating that, in an estuarine context, seawater temperature and salinity variation were not coupled to changes in N cycling.

| CS-δ 15 N baseline
A total of 34 coral samples from 12 genera was gathered from the museum collections, and 8 coral samples were recovered from the sediment core/ tombolo excavation (Tables S2-S4 Mean CS-δ 15 N values (genus Acropora only) were: pre-industrial group = 9.8 ± 0.8‰ (n = 6), 19th century group = 10.5 ± 1.0‰ (n = 5), and 1972-1983 group = 9.7 ± 0.6‰ (n = 14). Mann-Whitney pairwise comparison failed to detect any difference among the three periods, suggesting that the processes driving the N cycle in the PRD estuary were relatively stable over the last 5,000 years until the 1980s.

| DIN concentration in eastern HK
Coral communities of the HKSAR have been shown to present a dramatic reduction of coral-specific richness and coral cover at   1987-1993and 1994. Each water quality data point is the average of 25-156 measurements. The dataset used for this figure is available in Table S7 west

| Are coral communities in the Pearl River estuary affected by global warming and its associated hydrological changes?
The PRD megacity provides a unique opportunity to assess the respective roles of global and regional anthropogenic activities in coral reef decline. Although the impact of urbanization in the PRD on marine ecosystems has been widely documented over the last

| Controls of the N cycle in the Pearl River estuary from the mid-Holocene until the 1980s
The stable nitrogen isotope values from coral samples found in sedi- with South China Sea seawater, which has a shallow thermocline nitrate δ 15 N value of ~5‰ . N 2 fixation is an additional possible contributor to the eastward CS-δ 15 N decline.
Denitrification and related redox processes in the PRD, its tributaries, and the Pearl River estuary may lead to an excess of phosphorus F I G U R E 6 CS-δ 15 N difference (Δ) between the Pearl River estuary and waters to the east during (a) the 21st century (Porites data; Figure S5) and (b) the pre-industrial period (Acropora data). Direct comparison of Acropora and Porites CS-δ 15 N values was avoided due to potential inter-genera offsets caused by trophic level differences. Estuarine sites included Siu A Chau and Sham Wan, and eastern sites included Pak Lap Tsai and Bluff Island (Figure 1b) relative to N, which then encourages N 2 fixation as the Pearl River estuary waters mix with low-nutrient seawater. The substantial anthropogenic inputs do not appear to cause a large N 2 fixation response in the Pearl River estuary today. However, this may be due to elevated DIN concentration throughout the Pearl River estuary, with N 2 fixation occurring further offshore.
In the Pearl River estuary, coral growth is challenged by the low winter temperature (down to 13°C) causing frequent "low temperature bleaching," in particular in massive Porites (Ang, 2002;McCorry, 2002); as a result, in the Pearl River estuary, corals achieve most of their annual growth in summer (i.e., April-October). The N turnover in the tissues and symbionts of the genus Porites is close to 3 months (Rangel et al., 2019); as such, the N sources present in the summer are likely overrepresented in the CS-δ 15 N records.
Given that 80% of the Pearl River discharge occurs in summer (Lee, Harrison, Kuang, & Yin, 2006), when the Southwest monsoon pushes the river plume eastward (i.e., toward the sampling sites; Stable CS-δ 15 N values ranging from 9‰ to 12 ‰ until the 1980s were unexpected considering that the population in the PRD has experienced dramatic changes over this period, with a population increase from 1-2 million in mid-19th century to over 8 million in 1980 (Benedict, 1988;Gauducheau, 1911; Table S5), and the development of livestock agriculture (Morton, 1989) and aquaculture (Morton, 1975), together with the use of synthetic fertilizers (www. earth-policy.org/data_center). These changes should have strongly altered the pre-industrial isotopic baseline, with sewage, manure, livestock, and aquaculture being enriched in 15 N, whereas synthetic fertilizers tend to be depleted in 15 N (Kendall, Elliott, & Wankel, 2007 Accelerated N loss in response to increased anthropogenic N loadings would have tended to raise the δ 15 N of the residual DIN (Amundson et al., 2003;Kendall et al., 2007). As DIN is transferred through the terrestrial-estuarine continuum, it experiences various transformations, the most important being the microbially mediated removal of DIN through the oxidation of NH 4 + to NO 3 − (nitrification) and the subsequent reduction of NO 3 to N 2 O and N 2 (denitrification) (Peterson et al., 2001;Seitzinger et al., 2006). This N removal can have a strong influence on the 15 N/ 14 N ratio, preferentially removing 14 N and leaving the DIN pool enriched in 15 N (Granger et al., 2011;Martinelli et al., 1999;Seitzinger et al., 2006). Soils and streams play a key role in N removal by providing suboxic and organic matter-rich substrates suitable for coupled nitrification/denitrification. In the case of the Pearl River watershed, the higher degree of consumption would have maintained that DIN at adequately low concentrations that prevented its δ 15 N from overwhelming that of the other N sources to the Pearl River estuary. Moreover, because of the high degree of consumption in the sites of occurrence, the isotopic impact of denitrification in soils and marine sediments is often minimal (Brandes & Devol, 1997;Houlton, Sigman, & Hedin, 2006). Studies of heavily eutrophicated watersheds have revealed that the nitrification/denitrification process is particularly resilient to high anthropogenic loads and responds to increasing load by increasing rates of denitrification (Battaglin, Kendall, Chang, Silva, & Campbell, 2001;Panno, Hackley, Kelly, & Hwang, 2006;Peterson et al., 2001). These effects together would explain why the DIN input from the PRD was never so great as to completely overwhelm the low-δ 15 N signal of South China Sea-sourced nitrate in the outer Pearl River estuary watershed. This indicates that temperature and its control of soil/ stream denitrification prevailed over N source variation in setting the river δ 15 N DIN. The mean summer NO 3 δ 15 N was 11.8 ± 2.7‰, which is consistent with the CS-δ 15 N range of 9‰-12‰ recorded in the Pearl River estuary corals until the 1980s.
Efficient N removal upstream of the PRD would have minimized DIN supply to the Pearl River estuary. Moreover, unlike most eutrophied estuaries, water in the Pearl River estuary has a short residence time, 1-3 weeks (Dong, Su, Li, Xia, & Guan, 2006), and a strong upwelling regime develops during summer, diluting the Pearl River plume offshore (Lee et al., 2006). These factors would tend to prevent the development of sub/anoxic water layers required for direct denitrification in the water column (Naqvi, 1994) or the high flux of organic matter to the seabed required for coupled partial nitrification and denitrification in the sediments. As these processes are most prone to drive strong isotope fractionation during DIN loss (Granger et al., 2011;Sigman et al., 2003), this would explain the lack of a clear isotopic signal from N processing within the Pearl River estuary up to the 1980s.
Canonical sedimentary denitrification within the Pearl River estuary has likely always been active. However, as mentioned above, this process typically expresses minimal isotopic fractionation (Brandes & Devol, 1997;Lehmann, Sigman, & Berelson, 2004). In summary, we argue that DIN removal by upstream denitrification and the relatively active circulation in the Pearl River estuary were responsible for the elevated, yet stable, CS-δ 15 N values found in the Pearl River estuary before 1980, despite the progressively increasing anthropogenic N discharge to the Pearl River hydrologic system (Figure 7a).

| Pearl River delta megacity development: 1980-2000
The second phase of the PRD urbanization is characterized by ex-  (Figure 4a,b). The lack of correlation between NH 4 + concentration and salinity during 1987-1993 ( Figure 5c) suggests that NH 4 + did not enter via the Pearl River itself but instead that the NH 4 + discharge occurred nearby or directly into the estuary. In this scenario, the anthropogenic N load partially bypassed the soil and stream denitrification in the Pearl River and its upstream hydrologic system, vastly raising the DIN load to the estuary (Figure 7b). At the same time, this anthropogenic N did undergo some processing and loss, leading to a high δ 15 N for the DIN input, in the early 1980s, with minor to moderate coral mortality (Clark, 1998;Collinson, 1997;Cope, 1984;McCorry, 2002). The following decade was characterized by more frequent and intense coral degradation events (Figure 3). The high rainfall observed in the decade 1990-2000 (>800 mm/ month; Figure S2) and the subsequent salinity drop were proposed as the cause of the widespread coral mortality (Clark, 1998;Collinson, 1997;McCorry, 2002 does not seem to respond to extreme seasonal rainfall events in the Pearl River estuary, probably due to the short residence time of the water in the estuary (Dong et al., 2006). Although these salinity values were on the lower end associated with coral reefs and probably induced a significant stress to the coral communities, we argue that the salinity drop alone was not sufficient to trigger the collapse of entire coral communities as observed in Siu A Chau and Sham Wan and the severe mortality observed along the HKSAR coastlines ( Figure S3). Indeed, similarly high rainfall events occurred 13 times during the period 1853-1997, including the more recent years 1952years , 1957years , 1959years , and 1966.
No coral die-offs were reported prior to the decade 1990(McCorry, 2002, arguing that the high rainfall events alone cannot explain the coral demise in the Pearl River estuary. The progressive increase in the frequency and the magnitude of the coral communities' degradation events throughout the period 1980-2000 was synchronous with the increase in DIN concentration in coastal waters (Figure 3). This suggests that the coral decline in the Pearl River estuary was primarily driven by eutrophication caused by the development of the PRD megacity. The greater severity of mortality toward the Pearl River estuary ( Figure S3), where DIN concentrations are higher (Duprey et al., 2016), also supports this interpretation. Duprey et al. (2016) have suggested an annual average concentration of 2 µM DIN as a threshold for the coral communities of the Pearl River estuary: beyond this threshold, coral communities experience a severe decrease in coral cover and coral-specific richness. During the period 1990-1995, the DIN concentration in eastern HK was consistently above this threshold (Figure 5a), suggesting that the coral communities' decline was caused by an increase in the DIN concentration.
This long-term exposure to anthropogenic stress is likely to have weakened the resilience of the coral communities, making them more vulnerable to the severe eutrophication of 1980-2000.

| Anthropogenic N footprint in the Pearl River estuary in the 21st century
The implementation of waste management policies and the expan- Noticeable improvement in water quality over the last 20 years has also been reported by Duprey, McIlroy, et al. (2017) and Wong et al. (2018), indicating that adequate N mitigation policies can have a positive impact on the water quality without impacting economic growth. Yet, the level of anthropogenic N discharge, in particular from domestic waste, remains high and the development of N removal infrastructure is still behind the current waste production rate (Qin et al., 2014).
The inversion of the west-east CS-δ 15 N gradient along Hong Kong coastlines reveals that the N cycling in the Pearl River estuary remains strongly altered 20 years after the eutrophication event ( Figure 7). During the pre-industrial period, the CS-δ 15 N gradient in the Pearl River estuary, with CS-δ 15 N decreasing eastward, was set by the high-δ 15 N signal of watershed denitrification to the west and mixing with lower δ 15 N NO 3 from the South China Sea to the east (Figure 7a). The east-west isotopic gradient observed today is reversed, with CS-δ 15 N being ~2‰ lower toward the Pearl River estuary compared to eastern values (Figure 7c). The presence of anomalously low-δ 15 N DIN sources to the Pearl River estuary has been observed in NO 3 δ 15 N (Archana et al., 2018;Ye et al., 2015) and in Porites tissue δ 15 N . This indicates that low-δ 15 N N is being discharged directly into the estuary, bypassing the upstream isotopic homogenization described earlier (Figure 7c). The origin of the low-δ 15 N DIN sources may be related to increased synthetic fertilizers use in coastal areas, but it may also originate from industrial waste discharge, the isotopic composition of which is poorly constrained (Huang et al., 2018;Ye et al., 2015).
Low-δ 15 N N sources' discharge into the estuary cannot alone explain the reversed isotopic gradient observed today. A possible mechanism is an expansion eastward of the Pearl River's nutrient footprint due to the increased nutrient discharge in the estuary. Because high phytoplankton biomass is commonly found at the edge of the coastal plume (Xu et al., 2008;Yin, 2002), the expansion of the Pearl River plume has probably increased the supply of DIN with a high δ 15 N due to isotopic fractionation during assimilation by phytoplankton.
Indeed, Xu et al. (2008) found that, today, primary productivity is light limited due to the high turbidity found in the western part of the Pearl River estuary, whereas the summer peak chlorophyll-a concentration is found further east, in southern Hong Kong waters. As such, plankton assimilation is more likely to increase the δ 15 N of the N pool in the eastern part of the Pearl River estuary than in the western part, potentially explaining the CS-δ 15 N gradient (Figure 7c). This highlights that eutrophication is still a major issue in the Pearl River estuary, such that N mitigation should remain an important target of the PRC's and HKSAR's governments for the 21st century.

ACK N OWLED G EM ENTS
We thank SD Cairns, The Smithsonian National Museum of