The ASSETS method is derived from the National Estuarine Eutrophication Assessment (NEEA) conducted by the U.S. National Oceanic and Atmospheric Administration (NOAA) for evaluating both current eutrophic conditions and the effectiveness of management actions aimed at reducing eutrophic conditions (Bricker et al., 2003; Scavia and Bricker, 2006; Bricker et al., 2008). It has been developed and applied in coastal waters of the US, Europe, Australia, and China. This study adapted the pressure-state-response framework derived from the original ASSETS application (Bricker et al., 2003), and learned from the recent modifications described in the eutrophication assessment of the southwest Bohai Sea (Wu et al., 2013). The overall eutrophication condition (OEC) was determined by three component parts: water quality (WQ, pressure), ecological symptoms (ES, state), and future outlook of nutrient load (FO, response). The indicators and thresholds were modified to accommodate the relative sensitivity of local areas to nutrient-related degradation.

The NOWPAP CP method was developed to assess the eutrophication status in the NOWPAP region (NOWPAP CEARAC, 2011). It was separated into two steps, the same as OSPAR CP: the screening procedure was used to preliminarily assess eutrophication by detecting symptoms of eutrophication within minimum parameters (nutrients inputs, red tide event frequencies, and Chl-a), and the comprehensive procedure was used to assess the status and possible causes of eutrophication. The comprehensive procedure required the collection of information and data in line with the following four categories of parameters considering the degree and effects of nutrient enrichment: (I) degree, including DIN and DIP concentrations and their molar mass proportion; (II) direct effects, including Chl-a concentration and HABs, except for Noctiluca species blooms; (III) indirect effects, including DO and COD concentrations and fish kill events; and (IV) other possible effects, including Noctiluca species blooms and shellfish poisoning incidents. The indicators were assessed by classification (high or low) and trends (increasing, decreasing, or no change). The eutrophication status was classified into one of six categories: High-Increase (HI); High-No Trend (HN); High-Decrease (HD); Low-Increase (LI); Low-No Trend (LN); and Low-Decrease (LD). The eutrophication level of DO was determined in reverse to other parameters. For example, it was rated as “LD” when the DO values were above the reference value and showed an increasing trend and “HI” when they were below the reference value and had a decreasing trend.

In this study, most indicators and reference values conformed to those used in the Jiaozhou Bay assessment (Wu et al., 2019). In addition, slight modifications were made to accommodate local environmental characteristics and facilitate comparison with the ASSETS method as follows: (1) the riverine nutrient load based on the major drivers in the watershed was used instead of the riverine DIN concentration; (2) the surface DO concentration was replaced by the mean DO concentration; and (3) the threshold of the mean Chl-a concentration was increased from 5 μg/L to 10 μg/L. Details are shown in Table 2 . The three-year moving average method was used to calculate algal bloom incidences. In addition, changes in typical estuaries and bays were used to represent the trends of coastal zones due to insufficient long-term monitoring data for all sampling sites. The long-term trends of HABs were identified using the nonparametric Mann–Kendall test (Salmi et al., 2002) with long-term observation from 2001 to 2020. Changes in the other indicators were analyzed by comparison with reference conditions in the 1990s obtained from related reports and previous studies (Editorial Board of China Bay Survey, 1998a; Editorial Board of China Bay Survey, 1998b; Editorial Board of China Bay Survey, 1999; Du et al., 2003; He and Yuan, 2007; Huang et al., 2010; EPD (Environmental Protection Department of the Hong Kong Special Administrative Region), 2020; Hu et al., 2021; Zhang et al., 2022). A deviation greater than 50% from the reference value was considered an upward or downward trend. The assessment rating for each category was determined by selecting the most representative status of indicators.

The relative proportions of WQ ( Figure 3 ) classification as good, moderate, poor, and bad were as follows: 4%, 35%, 59%, and 2%, respectively. High or moderately high levels of WQ were widespread along the Guangdong coast due to high concentrations of DIN and DIP, showing the highest score (0.83) in Z4. A moderately low level (graded good) of WQ appeared in Z6 with the lowest score of 0.25. The annual variation in WQ showed an improvement in the study area from 2015 to 2018 ( Figures 4 , 5 ) because the relative proportions of water bodies as excellent and good increased from 4% to 39%.

The ES ratings for the Guangdong coast are shown in Figure 3 , of which 53% had a moderately low ES, 9% were moderate, 32% were moderately high, and 6% had a high ES. The scores of the eight zones for the primary symptoms were between moderate and high, revealing high levels of Chl-a biomass. High levels of secondary symptoms were mostly confined to Z4 and Z6 due to the distinct low-oxygen conditions in the bottom waters. Z2 and Z5 had a moderate rating for secondary symptoms mainly because of the periodic occurrence of nuisance algal blooms. As a result, the final ES ratings were graded bad for Z4 and Z6, followed by Z2 and Z5, which had a poor grade. The annual variation in ES ( Figures 4 , 5 ) showed opposite trends in the study area during 2015-2018, as the relative proportions of excellent and good increased from 43% to 59%, while that of bad increased from 6% to 23%. Notable enhancements in ES were seen in the western part of Guangdong Province, such as Z1 and Z3, whereas degradation was seen in the eastern part.

The application of the percentile-based approach for ES assessment was illustrated for Chl-a and DO ( Figure 6 ) using data from the PRE between 2015 and 2018. The 90^th percentile Chl-a in the northern part of the PRE (Z4) was more than twice as high as that in the southern part of the PRE (Z5), both exceeding the threshold (10 μg/L). The 10^th percentile DO was lower than 3 mg/L with a coverage rate of 100% for Z4, while it was periodically close to 4 mg/L with a coverage rate of 55% for Z5, indicating more severe hypoxic conditions for Z4. HABs occurred twice a year in the PRE, with an affected area of less than 100 km² and durations between 4 and 13 days. These symptoms resulted in a poor ecological condition in the PRE, rating moderately high in the south and high in the north.

Spearman correlation coefficient analysis revealed that PO and IN increased significantly (p<0.01) in the study area, while CF showed a decreasing trend (p<0.01). Moreover, OM and AP showed no obvious change. Significant spatial variations in these indicators ( Figure S4 ) existed among different coastal areas due to the regional imbalance of development. The final FO grade based on the trend analysis of the five major drivers for each zone is shown in Table S5 . The results demonstrated that nutrient loads for Z4~Z6 might continue to increase due to the growing economy and population. The nutrient loads for Z8 might decrease due to the distinct declines in PO, CF and OM. The other coastal zones might have no obvious changes in nutrient discharge.

The OEC ratings for the eight coastal zones were between good and bad ( Figure 3 ), showing higher ratings in Z2, Z4, Z5 and Z6 (graded bad for Z4 and poor for the other zones). Lower ratings of OEC were shown in Z1 and Z7 (both graded good). The relative proportions of OEC classification as good, moderate, poor, and bad were as follows: 35%, 27%, 36%, and 2%, respectively. The results indicated the most severe eutrophication problems were in the PRE, showing high OEC in the north and moderately high OEC in the south. The eastern and western parts of Guangdong Province presented a similar proportion of OEC, placing most zones in the good and moderate categories.

The degree of nutrient enrichment (Category I) between 2015 and 2018 was at a high level, as determined by high concentrations of DIN for most coastal zones except Z1, Z6, and Z7 ( Table 4 ). Coastal zones with high ratings for Category I also presented a high level of direct effects (Category II), mainly because their annual maximum Chl-a concentrations exceeded the threshold. In addition, the direct effects were graded high for Z6 and Z7 due to the high level of dinoflagellate and other species blooms. The DO and COD concentrations were at low levels for the study area, while a high rating of Noctiluca species blooms was confined only in Z6. Fish kills and shellfish poisoning incidents have rarely been reported in the study area in recent years, thereby gaining a low rating in the assessment. As a result, the indirect effects (Category III) and other possible effects (Category IV) were graded low for all zones except Z6, which had a high Category IV rating. In summary, most coastal zones were considered eutrophication problem areas with a high rating for at least one category except Z1, which was the only zone graded low for all four categories.

Increasing DIN and DIP were observed in many estuaries and bays along the Guangdong coast since the 1990s Figures 7A, C and D . As a result, most zones were classified as increase in Category I except Z8, which had no apparent change in DIN concentration. Category II also demonstrated an increasing trend in the PRE and eastern areas (including Z6, Z7 and Z8) mainly due to the enhancement of Chl-a in these areas over the past 30 years. Low-oxygen problems in the PRE and Mirs Bay have become increasingly severe as identified through long-term observations, resulting in an increasing trend in Category III for Z4, Z5 and Z6. Nevertheless, changes in Chl-a and DO on the west coast of Guangdong Province remained unclear due to insufficient data. Only a few studies provided the evidence of increasing COD in some typical bays, contributing to a rise in Category III for the western zones (including Z1, Z2 and Z3).

The incidence and affected area of algal blooms had decreased significantly (p<0.05) during 2001-2020 based on the Mann–Kendall test ( Table S6 ). The decline was mainly due to reductions in the occurrence of diatom and chromophyta species blooms. Significant increases were observed in Z7 (diatom species) and Z5 (Noctiluca species), while noticeable decreases were identified in Z5 (diatom species), Z6 (dinoflagellate and other species) and Z8 (dinoflagellate and other species). In addition, noticeable increases were observed in the affected area and duration of HABs for Z3, despite the incidence of HABs had no statistically significant change. Besides, incidences of fish kills and shellfish poisoning were uncommon during the last 20 years. Therefore, there was no obvious change in Category IV for most zones in the long-term analysis, with the exception of Z5.

The results for the four categories are shown in Table 4 .

This study revealed high levels of water quality for most coastal areas ( Figure 3 ), such as the PRE, Zhanjiang Harbor, Leizhou Bay, Shantou Harbor, which agreed with previous studies (Yin and Harrison, 2008; Lu et al., 2009; Zhang et al., 2016; Lao et al., 2019). The poor water quality of the PRE was due to the large amount of riverine nutrient inputs and sewage discharge from dense population and intensive industry (Huang et al., 2003; Yin and Harrison, 2008; Lu et al., 2009). Intensive agricultural in watersheds ( Figure S4 ) was the major pressure on water quality for Z2 and Z3 (both classified as moderately high). In addition, the western coastal currents, which were greatly affected by the Pearl River plume, might also bring in abundant nutrients (Shu et al., 2018; Lao et al., 2019). Shantou Harbor and adjacent areas (Z8) was another hotspot with high nutrient loads, receiving lots of nutrients from large rivers including the Hanjiang, Rongjiang and Lianjiang. These rivers were all characterized by dense population and intensive agricultural activities in their catchments.

The annual comparison of water quality between 2015 and 2018 ( Figure 4 ) revealed an improvement in Z1 and Z7. The declines in population ( Table S5 ) and precipitation ( Figure 7A ), which had reduced nutrient inputs from sewage and agriculture non-point source pollution (Carpenter et al., 1998; Lao et al., 2019), might be the main causes. Particularly in Z7, despite not statistically significant, CF, OM and AP in this area had decreased during 2015-2018, contributing to the reduction of agricultural pollutants. In comparison, there were no apparent changes in areas with high or moderately high WQ ratings, indicating heavy nutrient stress from anthropogenic inputs in these areas, although in some areas (such as the PRE) a noticeable decreasing trend in CF was observed.

The results of the ASSETS assessment revealed significant temporal and spatial variations in the ecological response to nutrient enrichment for different zones ( Figure 3 , Table 4 ). For Z2, Z4 and Z5, the ES ratings were as high as the WQ ratings, indicating that excess nutrients were the key causative factors of eutrophication. Z1, Z3, Z7, and Z8 showed lower ES ratings compared to their WQ ratings, revealing their lower sensitivity to nutrient enrichment. A sharp contrast between the low rating of WQ and the high rating of ES was observed in Z6, indicating that the hydrological conditions and biochemical processes might play a more important role in determining the health of coastal ecosystems (Cloern, 2001; Ferreira et al., 2005). In the case of Z6, the combination of a high Chl-a concentration, low bottom DO concentration, and periodic occurrence of HABs led to a high ES rating. Previous studies have demonstrated that the high primary production and frequent algal blooms in this area were probably due to the suitable temperature conditions, highly efficient nutrient circulation, and nutrient supplementation from upwelling (Sun et al., 2011; Wu et al., 2016; Wu et al., 2017). In addition, seasonal hypoxia observed in Mirs Bay and its adjacent waters was driven by coupled physical and biochemical processes (Li et al., 2014; Zhang et al., 2018). Therefore, it is needed for further tests to gain insight into the magnitude of ecological sensitivity to nutrient enrichment for different coastal systems.

The results revealed that more severe ecological symptoms occurred in the northern part of the PRE than in the southern part. The high rating in the north was determined by the high Chl-a and severe hypoxic conditions ( Figure 6 ), which were rarely noticed in previous studies (Zhang and Li, 2010; Lu and Gan, 2015). The research areas in previous studies were often confined to the middle and lower parts of the estuary with a salinity higher than 10 PSU; however, those areas were recognized as the southern part of the PRE in this study. Lu and Gan (2015) found that the high Chl-a in the upper reach was likely induced by the inactive phytoplankton from the river discharge rather than by cells generated locally, and this result was supported by the low DO concentration at the same location. In addition, the persistent low-oxygen waters in the north observed in this study confirmed that the hypoxic conditions in the upper part were mainly due to the large amounts of pollutants and low-oxygen waters from upstream reaches (Qian et al., 2018; Hu et al., 2021). Unlike the upper reach (northern part), the high biomass and the seasonal low-oxygen condition in the southern part resulted from the combination of physical and biochemical effects associated with variable riverine discharge and stratification during summer (Zhang and Li, 2010; Lu and Gan, 2015; Hu et al., 2021).

The annual analysis of the ecological response to nutrient enrichment revealed a remarkable shift in the southern part of the PRE, showing a moderate ES rating in 2017 and dropping to a low rating in 2018. The 10^th percentile DO concentration decreased from 4.9 mg/L to 3.1 mg/L, and the number of algal bloom events increased from zero to three during 2017-2018. Various factors may have contributed to this abrupt change due to the complex physical and biochemical interactions in the PRE (Yin et al., 2000; Huang et al., 2003; Zhang and Li, 2010). Researchers have noticed that changes in the riverine sediment loads might be essential in controlling biological processes by regulating the turbidity distribution, which significantly influenced phytoplankton growth in estuaries (Cloern, 1987; Tian et al., 2009; Horemans et al., 2020). The reduction in sediment supply and subsequent decline in turbidity for large estuaries, such as the Yangtze River estuary and the PRE, could strengthen oxygen depletion and increase the risk of HABs, as the light-shading effect of suspended sediments on primary production was greatly weakened (Wang et al., 2016; Hu et al., 2021). In this study, a remarkable decline in sediment loads of the PRE (measured at the Gaoyao, Shijiao, and Boluo stations) was observed between 2017 and 2018 ( Figure 8B ), decreasing from 34.52 Mt/a to 10.11 Mt/a. This might be the critical factor leading to the dramatic deterioration of ecological conditions in the southern part of the PRE. It is required for future work to investigate the roles and contributions of the changes with long-term observations.