Numerical Study of the Three Gorges Dam Influences on Chlorophyll-a in the Changjiang Estuary and the Adjacent East China Sea

CHEN Dongxing , LIU Qianqian, and YIN Kedong ,

1) School of Marine Sciences/Guangdong Key Laboratory of Marine Resources and Coastal Engineering, Sun Yat-sen University, Guangzhou 510275, China

2) Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 510275, China

3) Department of Physics and Physical Oceanography, University of North Carolina Wilmington, Wilmington, NC 28403, USA

Abstract The impoundment of the Three Gorges Dam (TGD) has altered downstream hydrological characteristics and sediment discharge, and it has caused ecological impacts, such as changes in chlorophyll-a (Chl-a) in estuaries and coastal oceans. To investigate the TGD’s influence on Chl-a’s interannual and seasonal variations in the Changjiang Estuary and the adjacent coastal East China Sea, a physical-biogeochemical model was developed with numerical experiments covering a decade, including TGD’s preperiod (pre-TGD, 1999–2003) and post-period (post-TGD, 2004–2008). The modeling results demonstrate an annual increase in the regional average Chl-a from pre- to post-TGD, with the largest increase reaching up to 20.8% in spring in the nearshore region beyond the Changjiang mouth. The spatial variations in Chl-a are high, with the largest variation being observed around the 20 –40 m isobaths. The increase in Chl-a is influenced by improved light and nutrient conditions, which is attributed to dam construction and fertilization by human activities. A decline in nitrogen-phosphorus fertilizer usage ratio along the Changjiang River watershed after the TGD’s impoundment is another factor that influences the Chl-a’s variation. The modeling results also show severe NO3 and PO4 imbalances with a larger N/P ratio during the post-TGD period, especially in regions with large Chl-a increases.Moreover, the greater increase in the usage of phosphorus fertilizer than nitrogen fertilizer influences Chl-a’s variation and has potential promotion effects on the outbreak of harmful algal bloom events.

Key words physical-biogeochemical model; Chl-a; coastal ecosystem; East China Sea

1 Introduction

The East China Sea (ECS) is a region with high primary productivity. As one of the most important factors of plankton growth, a large amount of nutrients along rivers pour in the coastal ocean from upstream watersheds. The Changjiang River dominates the nutrient input to the coastal region of the ECS. With abundant nutrients loading from the rivers, the Changjiang Estuary and its adjacent coastal region have been reported to have large chlorophyll-a(Chl-a) and seasonal variations in phytoplankton biomass (Zhuet al., 2009). During spring, surface Chl-ais high, ranging from 0.2 to 5.9 mg m−3in the Changjiang Estuary (Puet al., 2001). As a seasonal phenomenon, red tides ofProcentrum donghaienseusually occur in May at the inner front of the Changjiang Estuarine plume, becoming a seasonal phenomenon (Luet al.,2005). During summer, a stratified water column caused by the large estuarine plume due to high river discharge* Corresponding author. E-mail: yinkd@mail.sysu.edu.cn results in a large area with Chl-aranging from 0.11–3.38 mg m−3, which is correlated with the formation of hypoxia(Zhuet al., 2011). During autumn, the average Chl-aover the whole Changjiang Estuary is 1.86 mg m−3, with nitrate being the limiting factor (Songet al., 2008). During winter,the strong mixing caused by northwest monsoon reduces the phosphate limitation, yet the low temperature and high turbidity inhibit the increase of Chl-a(Zhouet al.,2008). The variation in Chl-ais not isolated but interacts with environmental factors and events.

The Three Gorges Dam (TGD) in the Changjiang River can deteriorate the ecosystem of the Changjiang Estuary and the adjacent coastal region through its regulation on the Changjiang diluted water (CDW) and sediment discharge (Daiet al., 2011). The weakened pollutant load by backwater effect in the reservoir area and increased use of fertilizers make the Changjiang River one of the eutrophic rivers in the world (Liet al., 2018). TGD’s regulation increases nutrients, decreases turbidity, and weakens the estuarine plume in the ECS, thereby resulting in an increase in phytoplankton biomass and a decrease in the diatom-dinoflagellate ratio, although diatoms are still the dominant functional group (Zhouet al., 2017). Limited riverine dissolved silicon exports to the estuary and coastal ocean because of abundant silicon retention in the reservoir area after the closure of the TGD is predicted to change the phytoplankton community (Jianget al., 2010).Aside from ECS, many other estuaries are exposed to environmental changes, such as saltwater intrusion in an estuary at the southern Gulf of Mexico, particular organic carbon source and flux change in Geum estuary, and transformation of phytoplankton groups in Guadiana estuary(Domingueset al., 2012; Alcerrecahuertaet al., 2019;Kanget al., 2019). These evidence indicate that ecological issues caused by dam constrictions are common all over the world.

Various research methods, such as lab experiments,in-situcruise investigations and ocean color remote sensing, have been used to study Chl-adue to its significance in marine ecosystems. However, these methods have limited spatial and temporal coverages and resolutions and are susceptible to inclement weather conditions. To overcome the disadvantages, numerical models have been widely used to investigate oceanic biogeochemical processes, including Chl-adistribution and primary production in the South China Sea, front variations in Chl-ain the North Pacific, and responses of Chl-ato upwelling in the Changjiang Estuary. Despite the extensive studies about the influences from the regulation of the TGD on environmental conditions, works that model its effect on long-term changes in Chl-aand its influential factors in the Changjiang Estuary and the adjacent coastal ECS is still lacking. In this study, we use a three-dimensional physical-biogeochemical model to study the biogeochemical processes in the Changjiang Estuary and the adjacent coastal ECS during different seasons before and after TGD impoundment. Specifically, the aims of this work are 1) to evaluate the variability in Chl-aunder the influence of the TGD; 2) to analyze how the salinity and correlated nutrients’ change and their effect on Chl-a; and 3) to examine the relationship between fertilization in the Changjiang watershed and nutrient ratio in the downstream coastal ocean, and their possible connection with the occurrence of harmful algal bloom events (HABs).

2 Model Description

2.1 Physical-Biogeochemical Model

The biogeochemical model used in the study is the Carbon, Silicate, and Nitrogen Ecosystem model (CoSiNE),coupled with a hydrodynamic model, namely, the Regional Ocean Modeling System (ROMS), which represents a hydrostatic, primitive equation, and the Boussinesq ocean general circulation model. The computational domain is a rectangle area, ranging from 23.5˚N to 41˚N and 117.5˚E to 132˚E, covering the Bohai, Yellow, and ECS and parts of the Pacific Ocean and Japan seas in the northeast (Fig.1).The model has 30 terrain layers following vertical levels and a 1/24˚ homogeneous horizontal resolution. Solar radiations are derived from National Centers for Environmental Prediction reanalysis data, and heat flux is calculated from the prescribed shortwave/longwave radiations. Air-sea momentum flux sources from the Blended Sea Winds dataset and provides 0.25˚ gridded ocean vector winds. Monthly averaged freshwater and sediment discharges for the Changjiang River were obtained from the records of the Datong station, which is the closest large hydrological station to the Changjiang Estuary (Yangtze Statistical Yearbook, 1999–2008; River Sediment Bulletin of China, 1999–2008). Monthly discharge for the model computation in other regions contained six major rivers,namely, the Huanghe, Liaohe, Yalvjiang, Hanjiang, Qiantangjiang, and Minjiang Rivers, which were referred to by Chang and Isobe (2003) and Zhouet al.(2015). More details on the model setting and description can be found in Chenet al.(2019).

Fig.1 Computational domain of the ROMS-CoSiNE model. Contours represent isobaths, and the white box indicates our research area. Colored dots represent in-situ sampling stations from available studies (Zhou et al., 2003; Chai et al., 2006; Li et al., 2009; Sha et al., 2009; Wang et al., 2009). The red diamond and green inverted triangle represent locations of the TGD and the Datong hydrological station, respectively.

The CoSiNE model was developed by Chaiet al.(2002),and consists of 13 state variables: four nutrients (silicate,nitrate, phosphate, and ammonium), two phytoplankton groups (small phytoplankton and diatom), two zooplankton groups (microzooplankton and mesozooplankton),two detritus pools (detritus-nitrogen and detritus-silicon),alkalinity, oxygen, and carbon dioxide. CoSiNE is a nitrogen-based model that uses the Redfield stoichiometric ratios (Redfield, 1963) with a C/N molar ratio of 7.3(Anderson and Sarmiento, 1994) to convert nitrogen into a carbon unit. The C/Chl-aratio is 50/1 as suggested by Xiu and Chai (2012) in a study on phytoplankton dynamics in the North Pacific.

Table 1 lists the major referenced parameters used in the CoSiNE model, which are related to the biological processes of phytoplankton. In CoSiNE, phytoplankton are classified into two groups (P1-small phytoplankton; P2-diatom) based on the different diameters. Chl-aconcentration is the sum of the two groups and converts to a standard unit of μg L−1for convenient analysis. Their biomass is regulated by two different grazers (Z1- microzooplankton;Z2-mesozooplankton) and four different nutrients (NO3-nitrate; NH4-ammonium; PO4-phosphate; SiOH4- sillicate)(Eqs. (1) and (2)). Details of more specific processes, such as nutrients uptake, grazing and mortality, were described in Chaiet al.(2002) and Zhouet al.(2017).

Table 1 Parts of the parametric settings related to the biological process

Data on nutrients were obtained from Jianget al.(2010)for the Changjiang River and interpolated into monthly averages as the model input. The other rivers were based on Liuet al.(2003, 2009). The model was initialized with a climatology run by using open boundary conditions from the Pacific-CoSiNE model and spun up for 10 years to reach a quasi-equilibrium state (Zhouet al., 2015). The model was then restarted from the last step of the tenth year and continued running from January 1990 to December 2008. More details, such as modules, general process, and model setting, can be referred to in Chaiet al.(2002), Zhouet al.(2015), and Chenet al.(2019). Our model focuses on the area from the Changjiang Estuary to 50 m isobaths in the ECS (28˚–33˚N, 120˚–125˚E) (Fig.1). Seasons are divided as spring: March – May; summer: June – August;autumn: September – November; and winter: December–February, and seasonal averages are based on sea surface data from 1999 to 2008.

Application data and usage ratio of essential fertilizers– namely, nitrogen and phosphorus – along the 11 Changjiang River provinces (11 provinces in total) have been collected (China Statistical Yearbook, 1999–2008) for an analysis of their relationships with the occurrence of HABs (China Marine Disaster Bulletin, 1999–2008) and with our modeled Chl-aannual concentration variation in the Changjiang Estuary and the adjacent coastal ECS.

2.2 Model Evaluation

Zhouet al.(2015) and Chenet al.(2019) validated the model’s performance through physical processes, including variation in particulate organic carbon. The model results were compared with monthly and regionally averaged surface Chl-afrom Moderate Resolution Imaging Spectroradiometer (MODIS) data (Fig.2) to evaluate the model’s performance in capturing temporal variations of phytoplankton biomass. The MODIS data from 2000 to 2008 with a 4 km spatial resolution are used. They show strong seasonal variability in Chl-a,with maximum concentrations in spring and low concentrations in winter(Figs.2a, 2b). Our model results capture the seasonal variability in both range and phase with a root mean squared error of 0.19 for the climatological Chl-acomparisons(Fig.2b). The model is consistent with MODIS interannual variability, and captures some abnormal fluctuations,such as anomalous high values in the years 2001 and 2006.In-situChl-adata from different studies (Fig.1; Zhouet al.,2003; Chaiet al., 2006; Liet al., 2009; Shaet al., 2009;Wanget al., 2009) were compared with our modeled results (Fig.2c). The modeled Chl-avalues show a highly significant correlation within-situobservations (r2= 0.83,P＜ 0.05). In general, our model efficiently captures the temporal pattern and spatial variations of Chl-ain our research area.

Fig.2 a) Comparison of the time series of monthly averaged Chl-a between the model and MODIS; b) comparison of climatological Chl-a between the model and MODIS; c) comparison of Chl-a between the model and in-situ observations in the research area (Fig.1).

3 Results

3.1 Water Flow and Sediment Discharge

The impoundment of the TGD began in 2003. Therefore,we divide the years 1999–2008 into two periods (Fig.3): 1)pre-TGD from 1999 to 2003 and 2) post-TGD from 2004 to 2008. Hydrological characteristics downstream change significantly due to the impoundment of the dam. From pre- to post-TGD, the range of seasonal water flow decreases. During the pre-TGD period, the monthly average water flow reached a maximum of 73.29×103m3s−1in July 1999 and a minimum of 8.10×103m3s−1in February 1999, with a range of about 65.2×103m3s−1. During the post-TGD period, the maximum monthly average water flow was 48.99×103m3s−1in September 2005, and the minimum was 8.58×103m3s−1in February 2004, with the range reduced to about 40.4×103m3s−1. A comparison of the difference between the two periods shows that the maximum water flow and range have varying degrees of decline after the impoundment. Meanwhile, sediment loading shows a dramatic decline during the post-TGD periods. At the Datong hydrological station, the sediment discharge decreases by 49.7% from pre- to post-TGD,with an annual mean of 1.25 ± 0.27 Gt yr−1in post-TGD.The TGD has been closed since 2003; this closure resulted in large-scale sediment trapping behind the dam,which is responsible for almost 50% of the sediment decrease at the estuary (Daiet al., 2018).

Fig.3 Time series of the model input CDW flow and sediment discharge. The red area represents the water flow of the pre-TGD period, while the blue area represents the post-TGD period, and green stars indicate sediment discharge. Data were collected from the Datong hydrological station.

3.2 Chl-a Variation

Strong seasonal and spatial variability of Chl-ais observed in the research area (Fig.4). During the pre-TGD period, in spring, high Chl-a(＞ 4 μg L−1) appears near the estuary and Hangzhou Bay and gradually decreases toward the offshore area to ＜ 1 μg L−1. Under the influence of the southwest monsoon and CDW, the high Chl-aarea extends northeastward from spring to summer. In autumn and winter, the high Chl-aarea is smaller near the Changjiang Estuary, reducing by about 30% compared with spring and summer. A substantial increase of Chl-aoccurs in the post-TGD period compared with the pre-TGD period. The largest increase (1–3 μg L−1) is concentrated on the nearshore region beyond the estuary, where the high Chl-aarea expands to 122.5˚E. In the offshore sea region,Chl-a’s average range of increase is 0–0.25 μg L−1due to the insufficient nutrients support.

Our model shows that the Chl-aregional average concentrations and distribution changes significantly from pre- to post-TGD throughout the seasons (Table 2). More than 17000 data points are calculated, showing that the regional average Chl-aconcentrations exhibit a widespread increase in all seasons from pre- to post-TGD. The seasonal differences indicate the largest change of 20.8%during spring and the smallest change of 14.8% during autumn. Summer and winter have average increases of 17.9% and 16.1%, respectively.

Table 2 Average seasonal Chl-a (unit: μg L−1) during preand post-TGD in the Changjiang Estuary and the adjacent coastal ECS

A sectorial coordinate diagram with correlation coefficients between salinity and Chl-achanges from pre- to post-TGD around different isobaths is designed to explore TGD’s most affected area (Fig.5). Different isobaths affect gradients in salinity, which are related to the extent of freshwater influences in the study area. The correlation coefficients around different isobaths are calculated; the maximum correlation coefficient (0.77) occurs on 30 m isobaths (data collected from 20–30 m) during spring, with a salinity change of 5.1% and a Chl-achange of 31.5%,as shown by the green triangle in Fig.5. The second largest increase of Chl-a, which is marked by the red triangle,appears at the 30 m isobath in summer, with a Chl-achange rate of 23.7% and a salinity change rate of 4.3%. Seasonal variation indicates that high Chl-aseasons – spring and summer – usually have relatively larger fluctuations in both Chl-aand salinity. For example, during spring and summer, the 20 m isobath exhibits Chl-achanges of 19.2%and 14.6%, while salinity changes reach 3.9% and 3.1%,respectively. In contrast, Chl-aand salinity changes during autumn and winter have fluctuations with relative narrow ranges, that is 11.7% and 8.8% for Chl-aand 2.3%and 1.4% for salinity, respectively. The above correlation analysis demonstrates that the Chl-aaround the 20 to 30 m isobaths is most affected by TGD’s regulation, and the result varies seasonally.

4 Discussion

4.1 Effects of TGD on Chl-a in the Changjiang Estuary and the Adjacent ECS

After the impoundment, available studies from 2003 to 2006 show that about 60% of silt from the mainstream was trapped in the reservoir area and about 6.65×108m3sediment accumulated in the reservoir (Xu and Milliman,2009; Huet al., 2013). This condition results in the continuously decreased sediment discharge to the coastal ocean (Fig.3). The change in hydrological characteristics and sediment discharge has a strong impact on the downstream ecological environment. After the impoundment,the reservoir’s water level rose to 175 m, which indicates a gradual increase of up to 3.93×1010m3in volume (Changet al., 2018). These changes influence the ecosystem of the Changjiang Estuary and the adjacent coastal ocean, resulting in phenomena such as shrinkage of the delta intertidal wetland area and an increase in primary production.Previous studies on the TGD’s ecological influences focused on short-term variations in small regional ranges.Studies on a larger area with long-term variations are lacking. After the impoundment, the variation of water flow and sediment discharge became the critical factors that produce chain effects on ecological changes in the Changjiang Estuary and the adjacent coastal ocean of ECS.

Fig.4 Seasonal distribution of Chl-a (unit: μg L−1) during pre-TGD and the difference between post- and pre-TGD(post- minus pre-) near the Changjiang Estuary and the adjacent coastal ECS.

Fig.5 Sectorial coordinate diagram to quantify the salinity change influences on Chl-a changes between preand post-TGD periods. The crosses, circles, triangles,squares, and diamonds represent different isobaths (10:0–10 m; 20: 10–20 m; 30: 20–30 m; 40: 30–40 m; 50: 40–50 m, respectively), while the green, red, blue, and black colors represent spring (Spr.), summer (Sum.),autumn (Aut.), and winter (Win.), respectively.

Influenced by the variation of water and sediment discharge, Chl-apresents a significant change. Spatial variations are found in the increase in Chl-afrom pre- to post-TGD periods, especially around the Changjiang Estuary area (Fig.4). The high Chl-aarea expands toward the offshore region, and the rapid-increase regions locate around the Changjiang Estuary with concentration change rates greater than 50% during spring. During summer, the high Chl-aarea has a smaller increase rate. A large increase is concentrated further offshore and further north than during spring due to the strong CDW and southwest monsoon in this season. Autumn and winter have relatively smaller change rates, and most of the increase is concentrated on the coastal area offshore of the Changjiang Estuary and inside the estuary during winter. Reduced CDW flow and lower incoming solar radiation could be the dominant limiting factors during these two seasons.

The widespread increase in Chl-aduring the post-TGD period demonstrates that the environmental conditions become more favorable for phytoplankton growth in the Changjiang Estuary and the adjacent coastal ECS. Chl-ashows the largest increase in a nearshore region beyond the Changjiang Mouth (around 31˚N, 122.5˚E), which is a high-nutrient place (Shenet al., 2008; Jianget al., 2015).This region expands after the impoundment. Light penetration was one of the limiting factors and was improved by the reduced sediment load (Zhuet al., 2009). Increased Chl-aexplains this phenomenon, especially in the nearshore region beyond the Changjiang mouth where light is limited.

The variation in coastal Chl-ais associated with the estuarine plume or diluted water mass, which is represented by a strong correlation between Chl-aand salinity (Jianget al., 2015). Chenet al.(2019) demonstrated the TGD’s effect on the freshwater-influenced area by examining the isohaline changes. The results of the sectorial coordinate diagram validate our finding described in the previous paragraph, that is, that diluted water could have strong influences on the Chl-a’s variation (Fig.5). Riverine sediment often sinks quickly after diluted water flows out of the river mouth, thereby resulting in the formation of the turbidity maximum zone where abundant nutrients are released (Shenet al., 2008). Higher Chl-aand algal blooms often appear near this zone as a result of abundant nutrients and sufficient light. In general, large changes in Chl-a(＞10%) are concentrated in the isobath range of 20 – 40 m,especially during spring and summer; and small changes(＜10%) appear at offshore regions, occurring more often during autumn and winter. The region with Chl-achanges coincides with the area that exhibits a large salinity change.These variables have high correlation coefficients, indicating that the variation of Chl-ais largely influenced by the TGD. The impoundment of TGD affects the spatial extent of the CDW, thereby causing a variation in the turbidity maximum zone between the isobaths of 20 and 40 m. As for the inner estuary and offshore region, turbidity and oligotrophic conditions are disadvantageous for the growth of phytoplankton, respectively. Therefore, in these regions, Chl-aconcentrations are less influenced by the interaction of different water masses, and changes become more conservative. Seasonal fluctuations of water outflow regulated by the impoundment and drainage of TGD cause the shrinkage and expansion of the CDWinfluenced area (Guoet al., 2012). The modeling result is consistent with the general findings: the phytoplankton growth in the nearshore region is limited by light due to the high sediment resuspension and by nutrients in the offshore area, which is a region that is less influenced by CDW.

4.2 Relationships Among Chl-a, Nutrients,Fertilizers and HABs

Nitrogen (nitrate, NO3) and phosphorus (phosphate, PO4)are considered the most limiting nutrients for phytoplankton growing in coastal oceans. In the ECS, high NO3and PO4are associated with CDW because the Changjiang River brings most of the nutrients. Correlation analyses of NO3versusChl-aand PO4versusChl-ain different seasons and TGD periods were conducted (Fig.6). The correlation coefficients between Chl-aand NO3(rChl-a,NO3)and Chl-aand PO4(rChl-a,PO4) increased for all four seasons from pre- to post-TGD. Coefficients are highest during winter (0.79 and 0.82) and lowest during summer(0.35 and 0.37) for pre- and post-TGD periods, respectively. When compared withrChl-a,NO3,rChl-a,PO4is higher,and the maximum coefficients are 0.83 and 0.87 during winter while the minimum coefficients are 0.38 and 0.41 during summer for the two periods, respectively.

During winter and spring, the Changjiang River has a low outflow, the CDW-influenced area shrinks, and reduced sediment discharge provides sufficient light, thereby making nutrients one of the primary influencing factors(Zhuet al., 2009). This finding is indicated by the maximum coefficient ofrChl-a,PO4, which can reach 0.87 during winter. During summer and autumn, a high river discharge from the Changjiang River brings abundant nutrients. With abundant nutrients, the primary limiting factor in the studied region shifts to light intensity (Zhuet al., 2009; Jianget al., 2015), and the correlation coefficients between Chl- aand nutrients decrease. Furthermore,rChl-a,PO4is larger t hat theChangjiangEstuary and the adjacentcoastal ECS thanrChl-a,NO3, whichisconsistentwith studiesthat found are phosphate-limited (P-limitation) (Liuet al., 2016).Overall, bothrChl-a,NO3andrChl-a,PO4increase from preto post-TGD; this finding indicates that, during post-TGD,the change of biomass is more connected with the change of nutrients than that during pre-TGD. The change in the correlation coefficient could result from the more dominant roles played by nutrients in the phytoplankton dynamics during post-TGD, which is related to the shift in nutrient ratio induced by human activities.

Fig.6 Correlation analysis of seasonal Chl-a with NO3 and PO4 during the pre-TGD and post-TGD period, respectively. All correlations are significant at P ＜ 0.05.

Usually, phytoplankton uptakes nutrients from seawater at a specific proportion (nitrate [N]/phosphate [P]≈16:1; Redfield (1963)). The variations in N/P ratio during pre- and post-TGD periods (Fig.7) show that most of the regions in our research area have an N/P that far exceeds 16 and that the N/P ratio decreases gradually from the coast to offshore. Average ratios during different seasons range from 17:1 to 22:1, being generally higher than 30:1 in the coastal region and lower than 15:1 in the offshore area. Spring and summer, which are the seasons that have a relatively high Chl-aconcentration, simultaneously have a large area of a high N/P ratio. The highest N/P ratio appears in the region outside the Changjiang Mouth (around 31˚N, 122.5˚E), which is near the 30 m isobath, coinciding with Chl-a’s largest change in area.The N/P ratio from pre- to post-TGD continuously increases, thereby indicating an aggravation of nutrient imbalance. High N/P ratio more often appears at the inner estuary during winter due to the shrinkage of the influenced area and reduced water flow/denitrification, thereby causing phytoplankton to concentrate and consume nutrients in that region (Liuet al., 2016). During P- limitation, massive nutrient consumption by phytoplankton causes a severe N/P ratio imbalance. In general, Figs.3–7 indicate that TGD’s regulation on water flow and sediment discharge affects salinity, light, and the proportion structure of nutrients, eventually influencing the variation of Chl-a, especially for the nearshore region beyond the estuary.

Fig.7 Seasonal distribution of the N/P ratio in the Changjiang Estuary and the adjacent ECS during the pre- and post-TGD period.

HABs, an excessive growth of oceanic algae, increased annually from pre- to post-TGD in the ECS, particularly after the TGD’s impoundment in June 2003. Human agricultural fertilization-induced algal blooms are often reported and analyzed around the world (Beusenet al.,2016). Along the Changjiang River watershed, from preto post-TGD, nitrogen fertilizer increased by 2.4% from 793.1 to 812.3 Mt yr−1(1 Mt = 106t), while phosphorus fertilizer increases even more quickly, reaching 10.3% from 248.1 to 273.6 Mt yr−1; HABs annual occurrence frequency increases from 33.6 times to 54.8 times with an average Chl-aconcentration increase of 1.40 to 1.86 μg L−1(Fig.8a).The annual N-P fertilizer usage ratio is (2.99 ± 0.06):1 with the maximum of 3.1:1 in 2000 and the minimum of 2.9:1 in 2003, however, the average N/P ratio in our research area reaches (24.2 ± 0.98):1, with the maximum of 26.0:1 in 1999 and the minimum of 22.8:1 in 2003(Fig.8b). By comparing the deviations from each year to the mean value of the whole TGD period, we found that the trend of N-P fertilizer usage ratio reverses from above average during pre-TGD to below average after the TGD’s impoundment, while the N/P concentration ratio has the opposite trend.

Fig.8 (a) Application of artificial chemical fertilizer during the TGD periods, the Chl-a annual concentration and their relation with HABs occurrence frequency in the Changjiang Estuary and coastal ECS. (b) differences and deviations between N-P fertilizer usage ratio (blank histograms) in Changjiang and N/P concentration (cross histograms) in the research area and their connection with HABs occurrence frequency.

During the post-TGD period, the growth of phytoplankton depends more on NO3and PO4, and the imbalance of N/P ratio and HABs occurrence frequency simultaneously intensifies. Generally, nutrients are directly correlated with terrestrial input, and chemical fertilizer is one of the most important impact factors. The continuous growth in the use of chemical fertilizers in the Changjiang River watershed increases nutrients in the downstream estuary and the adjacent coastal region through surface runoff. With a policy being implemented around 2003 by the Chinese government to address agriculture accession to the World Trade Organization, a balanced fertilization technique to be applied to prompt an increase in the crop yield. This condition causes an increase in the use of phosphorus fertilizer (compared to nitrogen fertilizer) and results in the decrease of N-P fertilizer usage ratio (Yanget al., 2001). The N-P fertilizer usage ratio has now dropped below 3:1. The increase in phosphorus fertilizer increases phosphate flow into the Changjiang Estuary,which is the P-limited area, thereby enhancing the growth of phytoplankton in that region.

Overall, the variation of Chl-aaround the Changjiang Estuary and the adjacent coastal ECS is influenced by a series of environmental changes. After the impoundment of TGD in 2003, roughly 60% of the silt from the main stream is trapped in the reservoir region (Xu and Milliman,2009), resulting in the decrease in suspended sediment in the Changjiang Estuary from 0.54 g L−1in the 1960s to 0.28 g L−1in 2008 (Liet al., 2012). Sediment that accumulated in the reservoir area improves the light conditions in the downstream estuary. Meanwhile, enhanced biological activity and phosphorus fertilization accelerate the phosphate release process, thereby further increasing phosphates in the Changjiang Estuary. The increase in phosphate mitigates the P-limitation, enhances phytoplankton growth, and potentially accelerates HABs occurrence, frequency, and severity. Since the impoundment of the dam,the extensive nutrient utilization by the increased phytoplankton exacerbates the unbalanced N/P ratio in coastal ECS, especially in the nearshore area beyond the estuary where a maximum Chl-aconcentration is found.

5 Summary and Conclusions

In this study, a 3D physical-biogeochemical model (ROMS-CoSiNE) is developed to evaluate the Chl-a’s seasonal and interannual variation in the Changjiang Estuary and the adjacent coastal ECS during two TGD periods. Comparisons between the results and observations of the proposed model, including satellite andin-situdata, demonstrate that the model’s performance is well suited for our research area. Historical observations and studies reveal that TGD’s regulations in river flow into the ECS exhibit a slight but stable decrease. Sediment discharge presents a dramatic decrease of 49.7% from pre- to post- TGD.

The modeled Chl-avariation indicates a widespread increase during all seasons; the largest change appears during spring with a more than 20% increase, and the largest spatial increase occurs in the nearshore region beyond the Changjiang mouth (around 31˚N, 122.5˚E). A sectorial coordinate diagram has been used to conduct a correlation analysis, including different seasons, salinity, and Chl-achange during pre- and post-TGD periods. The result indicates that Chl-a’s large change (＞10%) concentrates in the water depth range of 20 – 40 m, which corresponds to the region that is influenced the most by the TGD.

Two essential nutrients, NO3and PO4, are analyzed with Chl-a. Both nutrients have correlations and a continuous increase from pre- to post-TGD. The NO3and PO4ratio shows that our research area belongs to a P-limited region.The growing N/P ratio affects the variation of Chl-a, particularly in the region with a high Chl-aconcentration.The variation of fertilization throughout the Changjiang River watershed, coupled with the effect from the impoundment of TGD, leads to nutrient variation in the Changjiang Estuary and the adjacent coastal region of ECS, thereby influencing the increase in Chl-aconcentration and HABs occurrence.

Our model provides intriguing insights into the effect of TGD on the phytoplankton ecology in the coastal regions of ECS. The model’s performance in the interaction of plankton and ecological variables needs further evaluation through additional physical and biogeochemical observations. Higher model resolution, improved biologicalchemical algorithm, and more precise terrestrial input data are required to reveal ecological changes between rivers and oceans under anthropogenic influence in the future.

Acknowledgements

This study is supported by grants from the Guangdong-NSFC Joint Theme Project (No. U1701247), the National Natural Science Foundation of China (No. 91328203), the Southern Marine Science and Engineering – Guangdong Laboratory (Zhuhai) (No. 311019006), and the Sun Yat-sen University Supercomputing Funding (No. 42000-526037 00). We thank Drs. Fei Chai, Feng Zhou, and Peng Xiu for providing essential information. Lastly, we also acknowledge NASA for providing the MODIS data (https://modis.gsfc.nasa.gov/).