The NGM (continental shelf off the states of Louisiana, and Texas) has an area of ∼60,000 km² with depths < 200 m. The Mississippi-Atchafalaya River system is the primary source of fresh water (∼80%) and new¹⁴ nutrients (∼90%) to the NGM. Volume transports of the river system range between 220 and 630 km³ yr^–1 with a mean of 530 km³ yr^–1 (Aulenbach et al., 2007). Depending on river discharge, the spatial extent of the resulting buoyant, nutrient-rich coastal plume varies from 10,000 to 35,000 km² with a maximum seasonal extent typically during May-June5¹⁵. On average, the river-system delivers ∼1.3 × 10⁹ kg N yr^–1 (Dunn, 1996; Turner et al., 2007). Discharge is lowest on average during September-October and highest during March-April (Dunn, 1996; Dagg and Breed, 2003; Turner et al., 2007). Of this input, it is estimated that denitrification and anammox¹⁶ remove 40–50% of nitrate inputs annually (Seitzinger and Giblin, 1996). Direct atmospheric deposition of DIN to the NGM is small (<1%) in comparison (Goolsby et al., 1999).

Riverine inputs of nitrate¹⁷ have increased by at least threefold since WW II, with most of this increase occurring between 1970 and 1983 due to increased use of nitrogen fertilizers in the river’s large watershed (Goolsby et al., 1999). Since 1983, annual riverine inputs of DIN have fluctuated between ∼ 0.6 × 10⁹ kg yr^–1 in 2000 and ∼ 1.8 × 10⁹ kg yr^–1 in 1993. In contrast to N, there has been relatively little change in P discharge by the Mississippi over a similar period so that the N:P ratio of the dissolved nutrient pool is consistently above the Redfield molar ratio of 16 (Redfield, 1958).

Phytoplankton biomass is correlated with the riverine supply of N (Fennel et al., 2011), and the annual cycle of phytoplankton biomass is characterized by a spring maximum and an autumn minimum (Bode and Dortch, 1996; Gomez et al., 2018). Spatially, phytoplankton biomass and NPP vary with salinity as follows (Lohrenz et al., 1997, 1999; Dagg and Breed, 2003; Gomez et al., 2018):

The spring bloom in the mesohaline fuels the development of an extensive hypoxic zone that is generally confined to a relatively thin (1–4 m) bottom boundary layer over the inner shelf from March to October (Dagg et al., 2007; Justić et al., 2007; Fennel and Testa, 2019). Seasonal hypoxia began expanding in the 1950s with rapid increases in the time-space extent of hypoxia during the 1970s (Rabalais and Turner, 2019). The spatial extent of hypoxia over the shelf has varied between < 5,000 km² in 2000 and 22,720 km² in 2017 with a mean of 13,700 km² making it one of the largest open shelf hypoxic zone in the world (Rabalais and Turner, 2019; Figure 4). Seasonal hypoxia has resulted in losses of benthic biodiversity and biomass as longer-lived species are eliminated (Committee on Environment and Natural Resources [CENR], 2000; Rabalais and Turner, 2019).

In addition to bottom water hypoxia, increases in nutrient loading appears to be promoting the growth of potentially toxic phytoplankton. The dinoflagellate Karenia brevis is the primary toxic species in the NGM where it is ubiquitous at background levels of < 1 cell ml^–1 (Steidinger et al., 1998). Blooms of K. brevis (>10 cells ml^–1) occur almost annually off the west coast of Florida, but historically have occurred less frequently along the Texas coast (Hetland and Campbell, 2007). The abundance of Pseudo-nitzschia spp. has increased over the shelf since the 1950s, a trend that may be related to the long-term increase in nutrient loading (Dortch et al., 1997; Parsons and Dortch, 2002; Baustian et al., 2016). Peaks in the abundance of potentially toxin-producing dinoflagellates (Dinophysis spp. and Prorocentrum spp.) have been observed to develop in concert with the spring peak in river flow (Bargu et al., 2016).

The ECS has an area of ∼750,000 km², 75% of which is < 200 m deep. With a mean volume transport of ∼ 900 km³ yr^–1 (Liu et al., 2003), the Changjiang (Yangtze) River accounts for > 90% of river runoff and is the largest source of nutrient loadings to the ECS (Yan et al., 2010; Wu et al., 2011; Tong et al., 2015). The annual cycle of flow is characterized by a summer maximum and a winter minimum (Chen et al., 2016). Under high summer flows, the River’s coastal plume spreads eastward over an area that can cover as much as 30% of the ECS in contrast to the winter low flow period when the plume is confined to a narrow band along the coast south of the River’s mouth (Dong et al., 2010; Tong et al., 2015).

Riverine inputs of N have been increasing since the 1960s (Zhou et al., 2008; Chen et al., 2019). Despite the Three Gorges Dam, which began storing water in 2013, inputs of DIN from the Changjiang increased by nearly an order of magnitude from 0.220 × 10⁹ kg N yr^–1 in 1970 to 2.0 × 10⁹ kg N yr^–1in 2012 (Tong et al., 2015; Lin et al., 2017; Strokal et al., 2020). N-fixation contributes only ∼ 0.013 × 10⁹ kg N yr^–1or < 1% of riverine inputs (Zhang R. et al., 2012). Losses of N via denitrification and anammox are estimated to be equivalent to riverine inputs of anthropogenic DIN (Lin et al., 2017). At the same time, the dam has reduced the mass transport of suspended matter and, therefore, the input of P to the ECS (Xu et al., 2015). As a consequence, N:P molar ratios have increased to >100 (Huang et al., 2019).

Dissolved nutrients are also delivered to the ECS by the Taiwan Warm Current (TWC), the Kuroshio, and atmospheric deposition (Chen, 1996, 2008; Zhang et al., 2019). Atmospheric deposition of N directly to the ECS in 2012 is estimated to have been ∼80% of riverine inputs. Eutrophication caused by nutrient inputs from the Changjiang are most pronounced in the near field plume (salinity < 30) over the inner shelf (0–50 m) while inputs from the TWC dominate the mid-field (salinity 31–32) over the mid-shelf (50–100 m) and inputs from the Kuroshio dominate the far field (salinity > 32) over the outer shelf (100–200 m) (Wang et al., 2014). Atmospheric deposition is distributed over the entire ECS (Zhang et al., 2019).

The annual cycle of NPP is characterized by a summer maximum and a winter minimum. Interannual variations (1998–2007) in the spatial extent of high Chl are driven by variations in river discharge (Xiuren et al., 1988; Kim et al., 2009). Increases in discharge have led to increases in NPP, increases in the abundance of small phytoplankton (<20 μm) and dinoflagellates, and decreases in diatoms in the coastal plume (Zhou et al., 2008; Li et al., 2019). While an interannual trend in Chl for the ECS as a whole (1996–2014) is not apparent (O’Reilly, 2017), Chl levels in the coastal plume have increased as anthropogenic nutrient inputs increased while Chl in the far field decreased due to increases in vertical stratification as the upper ocean warms (Kong et al., 2019). As the plume spreads and mixes with open shelf water, Chl concentrations decline from high levels (>10 mg m^–3) in the plume to low levels (<0.5 mg m^–3) in oligotrophic open shelf waters, a pattern that is driven by the riverine supply of nutrients (Wang and Wang, 2007; Chen, 2008; Mackey et al., 2017).

NPP in the plume is the primary source of particulate organic matter to the ECS (Zuo et al., 2016). As the Changjiang plume flows over ambient ocean water and the surface layer warms, the water column becomes increasingly stratified and this supply of organic matter fuels bacterial oxygen demand causing bottom waters to become hypoxic (Rabouille et al., 2008; Liu et al., 2015; Wang et al., 2016; Zuo et al., 2016; Qian et al., 2107). Persistent summer hypoxia causes high mortality rates among sessile benthic species and reduces recruitment to economically important fish populations (Levin et al., 2009).

Hypoxia was first reported in 1959 (Zhu et al., 2011). Interannual increases in NPP has led to increases in the spatial extent of bottom water hypoxia over the inner shelf from ∼1,800 km² in 1959 to ∼ 13,700 km² in 1999 and > 15,400 km² in 2006 (Levin et al., 2009; Li et al., 2011; Wang et al., 2015; Qian et al., 2107), a trend that has been attributed to elevated nutrient inputs due to fertilizer use in the Changjiang watershed (Wu et al., 2019). Bottom water hypoxia now covers > 15% of the ECS making it one of the largest coastal hypoxic zones in the world (Chen et al., 2007; Wang et al., 2016; Zhu et al., 2016).

Intrusions of nutrient-rich oceanic water from the Kuroshio also contribute to the development of hypoxia. The hypoxic region north of 30° N is dominated by Changjiang inputs, with its N loads supporting 74% of oxygen consumption; south of 30°N, oceanic nitrogen sources become more important, supporting 39% of oxygen consumption during the hypoxic season, but the Changjiang remains the main control on hypoxia formation also in this region (Große et al., 2020). The importance of oceanic nutrient supply distinguishes hypoxia in the ECS from the otherwise comparable situation in the NGM, where a similar spatial extent of hypoxia is fueled by riverine inputs of anthropogenic nutrients from the Mississippi-Atchafalaya River system (Fennel and Testa, 2019).

In addition to the impacts of hypoxia, increases in the N:P molar ratio promote blooms of toxic dinoflagellates (Glibert et al., 2018; Huang et al., 2019). Reported toxic algal events along the coast increased from undetected in the 1950s and 60s to 10 in the 70s, 25 in the 80s, and > 100 in the 90s (Yan et al., 2002). During 2000–2006, the trend continued with most events occurring during summer (Yan et al., 2010) in Zhejiang coastal waters, particularly in the Zhoushan Archipelago which is home to the largest marine fishery in China (Wang and Wu, 2009). Particularly, large scale blooms (covering an area > 1,000 km²) have been recorded every year since 1998, and the dinoflagellate Prorocentrum donghaiense has become the recurrent bloom species for more than 10 years (Li et al., 2014; Lu et al., 2014). Blooms of other potentially toxic dinoflagellates (Karodinium veneficum, Karenia mikimotoi, K. veneficum, Alexandrium tamarense, A. catenella, and Heterosigma akashiwo) have also been observed (Lu et al., 2014; Zhou et al., 2015; Wang Y.-F. et al., 2018). Toxic dinoflagellate blooms have resulted in millions of dollars of lost fish landings (Tang et al., 2006; Zhou et al., 2008; Li et al., 2014; Mackey et al., 2017; Glibert et al., 2018; Wang R. et al., 2018; Chen et al., 2019).

Eutrophication may also be an important factor contributing to increases in jellyfish abundance and blooms (Mills, 2001; Purcell et al., 2007; Diaz and Rosenberg, 2008; Richardson et al., 2009; Dong et al., 2010; Condon et al., 2011; Brotz et al., 2012; Boero et al., 2016). Three jellyfish species (Aurelia aurita, Cyanea nozakii, and Nemopilema nomurai) form large blooms, the frequency of which has been increasing since the 1950s (Dong et al., 2010; Brotz et al., 2012). N. nomurai, the giant jellyfish, is arguably the most serious threat to fisheries since it is most abundant in the Changjiang plume where it preys on juvenile fish (Sun et al., 2015).

With a surface area of 344,000 km² and depths of < 50 m, Australia’s GBR is the Earth’s largest reef system. As the reef developed over the last 20,000 years on the continental shelf, it formed a large coastal lagoon (<40 m deep) into which rivers discharge from 35 watersheds (total mean discharge of 70 km³ yr^–1). Today, rivers are the largest single source of new nutrients to the lagoon (Furnas et al., 1997; Brodie et al., 2011, 2012; Devlin et al., 2015), and elevated nutrient concentrations are measurable at distances of hundreds of kilometers from river mouths (Devlin and Brodie, 2005). Riverine inputs are characterized by episodic peaks in flow with most volume transport occurring during November–April when large flow events are most frequent and > 90% of land-based nutrient inputs occurs (Brodie et al., 2011).

Since European settlement, annual riverine inputs of N and P to the lagoon have increased from ∼ 0.014 × 10⁹ kg N yr^–1 to 0.080 × 10⁹ kg N yr^–1 and from 1.8 × 10⁶ kg P yr^–1 to 16 × 10⁶ kg P yr^–1 (Brodie et al., 2009; Kroon et al., 2012). The form of N delivered has also changed from predominantly dissolved organic nitrogen (DON) to predominantly DIN (Harris, 2001; Brodie and Mitchell, 2005). In addition to riverine inputs, N-fixation supplies 0.01–0.21 × 10⁹ kg N y^–1, rainfall 2.7 × 10⁶ kg N y^–1, and upwelling 0.001–0.004 × 10⁹ kg N y^–1 (Furnas et al., 1997, 2011; Benthuysen et al., 2016). Export via denitrification is estimated to be in the range of 0.016–0.024 × 10⁹ kg N y^–1 (Furnas et al., 2011) or about 25% of total inputs.

Long-term monitoring data show that hard coral cover on the GBR has reduced by > 70% over the past century Bell et al., 2014). Waters of the lagoon become progressively more productive as they flow through the reef system, an increase that is attributed largely to benthic N-fixation (Bell et al., 2014). Current levels of nutrient enrichment support Chl concentrations that range from annual means of < 0.3 μg liter^–1 to 0.7 μg liter^–1 (Brodie et al., 2007). Under flood conditions during April–November, peaks in river discharge support phytoplankton blooms in the central and southern lagoon that yield Chl concentrations of 1–20 μg liter^–1 (Devlin and Brodie, 2005; Devlin and Schaffelke, 2009; Brodie et al., 2009, 2010; McKinnon et al., 2013). Nutrient-driven increases in phytoplankton biomass and macroalgal cover have been shown to be associated with long term coral reef decline (De’ath and Fabricius, 2010; D’Angelo and Wiedenmann, 2014). Clear water¹⁸ and low macroalgal cover promote high coral species richness (De’ath and Fabricius, 2010). There is growing evidence that Acanthaster planci¹⁹ predation events and coral bleaching are exacerbated by eutrophication and that the lack of recovery of the reefs is primarily a consequence of high phytoplankton biomass (Bell et al., 2014; Allen et al., 2019). Thus, 23% of the reef system has been degraded in areas with Chl concentrations > 0.2 μg liter^–1, concentrations that are considered to be indicative of eutrophication in the GBR lagoon (Bell, 1992; Bell et al., 2014). De’ath and Fabricius (2010) predict that reducing agricultural runoff could reduce macroalgal cover by 39% and increase the species richness of hard corals and phototrophic octocorals by 16 and 33%, respectively.

N:P ratios are consistently < 16 and many processes in coral reefs are nitrogen limited (Furnas et al., 2005), and increases in river-borne inputs of P have promoted the growth of Trichodesmium spp. and other N-fixing organisms that introduce new N into the lagoon-reef system at rates that are far higher now than in the past (Bell and Elmetri, 1995; Bell et al., 1999; Messer et al., 2017; Blondeau-Patissier et al., 2018). This additional input of new N appears to be enhancing increases in phytoplankton biomass above that expected based on riverine inputs of N alone, and there is evidence that this is a significant factor in the demise of fringing reefs in the inner GBR lagoon (Bell and Elmetri, 1995; Brown et al., 2018). In addition, blooms of Trichodesmium spp. are known to be a source of ciguatera in the ciguatera food chain²⁰ (Kerbrat et al., 2011), and several genera of potentially toxic dinoflagellates have also become abundant in the lagoon (Gambierdiscus, Prorocentrum and Ostreopsis), a trend that may be driven by the ongoing eutrophication of the lagoon (Skinner et al., 2013).

The BS has an area of 420,000 km² with a basin that has a mean depth of ∼54 m, a number of sub-basins (>150 m deep), and a shallow sill (<20 m) in the Danish Straits separating it from the North Sea and Atlantic Ocean. Of the 250 rivers flowing into the Baltic, seven account for ∼ 45% of the total flow (∼ 455 km³ yr^–1) (HELCOM, 2011), most of which occurs in the eastern BS. As a consequence, surface salinity gradually increases from 2 to 4 in the northeastern Gulf of Bothnia between Sweden and Finland to 18–26 in the southwestern Kattegat between Denmark and Sweden (Gustafsson and Westman, 2002; Elmgren et al., 2015). In winter, the water column is mixed down to a permanent halocline (40–80 m). In summer, a seasonal thermocline forms above the permanent halocline and a three-layered structure develops with a warm and low salinity surface layer, a higher salinity intermediate layer of cold water, and an oxygen depleted bottom layer of warmer and saltier water (Liblik and Lips, 2019).

Rivers account for ∼70% of anthropogenic nutrient inputs to the Baltic, and concentrations of DIN are highest in coastal waters from the Belt Sea in the southwest to the Gulfs of Finland and Bothnia to the northeast (HELCOM, 2018a; Sonesten et al., 2018). Flow generally peaks during April–June and is relatively low during August–January (Hordoir and Meier, 2010). Nutrient inputs were high during 1995–2002 (0.65–0.90 × 10⁹ kg N yr^–1 and 33–43 × 10⁶ kg P yr^–1) compared to 2003–2015 (0.50–0.78 × 10⁹ kg N yr^–1 and 22–35 × 10⁶ kg P yr^–1). N input via N-fixation by cyanobacteria is estimated to be ∼ 0.40 × 10⁹ kg N yr^–1 (Olofsson et al., 2020), nearly equivalent to riverine inputs. Atmospheric deposition also declined during this period from ∼0.30 × 10⁹ kg N yr^–1 in 1995 to 0.21 × 10⁹ kg N yr^–1 in 2011. Relatively low inputs via river discharge and wet precipitation during 2003–2015 were due to dry periods with low river runoff (2003, 2014, 2015). Denitrification removes an estimated 42–96% of riverine nitrate inputs annually (Dalsgaard et al., 2013), and, over the long term, anthropogenic nutrient enrichment has led to accumulations of phosphorus (P) in benthic sediments to an extent that internal releases of P to the water column under anoxic conditions off-sets reductions in land-based inputs of anthropogenic P (Gustafsson et al., 2012).

Mean Chl concentrations remained < 1 μg liter^–1 from 1880 to 1950 and increased to 2–4 μg liter^–1 during 1990–2009 (Hieronymus et al., 2018). Two seasonal bloom periods characterize most of the BS, spring blooms of diatoms and dinoflagellates and cyanobacterial blooms in late summer (Spilling et al., 2018). N assimilation during the spring bloom leads to a low N:P ratio during summer which favors N-fixing cyanobacteria during warm, calm summer months (Wasmund et al., 2001, 2005). Summer cyanobacterial blooms have been intensifying since 1982, a trend that is correlated with increases in anthropogenic P loading and the magnitude of hypoxia (Pliñski et al., 2007; Funkey et al., 2014; Savchuk, 2018).

Hypoxic bottom water in the Baltic proper (Gotland Sub-Basin) has been present for at least the last 100 years, but increased in spatial and temporal extent over the last 3 decades (Duncombe, 2018) as the buoyancy of the surface layer above the permanent halocline increased due to increasing temperature and decreasing salinity (Liblik and Lips, 2019). During 1993–2016, the spatial extent of hypoxic bottom water in the main basins of the Baltic (Figure 5) increased from ∼5,000 km^–2 (1.3% of the Baltic) to ∼ 70,000 km^–2 (19% of the Baltic) (Jokinen et al., 2018; Limburg and Casini, 2019). At its maximum extent, the spatial extent of hypoxia is the largest anthropogenically enhanced hypoxic zone in the world (Schmale et al., 2016).

Benthic communities have been severely impacted by increases in phytoplankton biomass due to the rapid attenuation of sunlight with depth in the surface layer and hypoxia in the bottom layer (HELCOM, 2009). Impacts include the following:

With an area of 18,900 km², the continental shelf of the NA is a relatively flat platform that extends from the coastline to a water depth of 100 m (mean depth = 33.5 m). A progressive increase in eutrophication began in the early 20th C, continued to ∼ 1978, and subsequently began to decrease (Sangiorgi and Donders, 2004). The Po River accounts for most inputs of freshwater and nutrients with minor contributions from the Soča (Isonzo) and Adige Rivers (Pettine et al., 1998; Cozzi and Giani, 2011). The annual cycle of Po River discharge is characterized by seasonal peaks during April-June and September–December with minimum flows during August (Marini et al., 2008). During May–September when the water column is well stratified, the plume of the Po River spreads across much of the northern basin and maintains a wide cross-shelf profile as it flows southward. The volume transport of the Po during 1970–2002 fluctuated around 47 km³ yr^–1 with a minimum discharge of ∼30 km³ yr^–1 and a maximum discharge of ∼80 km³ yr^–1 (Giani et al., 2012; Montanari, 2012; Ninfo et al., 2018). A marked decrease in the flow of the Po River occurred during 2003–2007 due to a prolonged drought (Cozzi and Giani, 2011; Djakovac et al., 2015).

Diffuse sources account for 80% of N input and 93% of P input (European Topic Center [ETC], 2016). Prior to the drought years, river borne transports of N and P to the NA reached 0.262 × 10⁹ kg TN yr^–1 and 11.1 × 10⁶ kg TP yr^–1. These inputs decreased by 50–70% during the drought years, but returned to their pre-drought levels in subsequent years (Cozzi and Giani, 2011; Giani et al., 2012). N-fixation does not contribute significantly to the N budget, and denitrification accounts for ∼ 42% of N losses (∼ 0.10 × 10⁹ kg yr^–1) or ∼40% of the riverine N inputs (Degobbis and Gilmartin, 1990).

Phytoplankton NPP and biomass are highest in the Po River plume (Figure 7; Degobbis et al., 2000; Fiori et al., 2016), and annual cycles mirror the annual cycle of river flow while annual cycles in oligotrophic waters beyond the plume have a relatively small amplitude with small peaks during October-November (Revelante and Gilmartin, 1992; Harding et al., 1999; Mozetič et al., 2010). Chl concentrations ranged from 4 to 35 μg liter^–1 in the plume to 1–4 μg liter^–1 in offshore waters of the central NA and were < 1 μg liter^–1 in coastal waters of the eastern NA (Polimene et al., 2006; Solidoro et al., 2009; Mozetič et al., 2010). Phytoplankton blooms in the coastal plume of the Po are dominated by large diatoms (>20 μm) interspersed by diverse assemblages of nanoflagellates (<20 μm) with small dinoflagellate blooms during spring-summer. Small phytoplankton dominate in more oligotrphic waters outside the plume where microbial food webs predominate (Polimene et al., 2006). Coccolithophores are minor but regular inhabitants of these phytoplankton communities (Bernardi-Aubry et al., 2004; Totti et al., 2005, 2019).

The 2003–2007 drought resulted in marked decreases in phytoplankton NPP and Chl. Concentrations of nutrients and Chl remained low after higher river discharges returned during 2008–2016, possibly due to improved control of nutrient loads to the watershed (Totti et al., 2019). However, greater intrusions of high salinity, oligotrophic waters from the eastern Mediterranean Sea also played a role (Solidoro et al., 2009). More recent observations suggest that phytoplankton biomass may be increasing in the southern reaches of the western NA where both DIN and DIP concentrations have been increasing (Totti et al., 2019).

Sedimentation of phytoplankton biomass to the benthos, estimated to be 25–30% of phytoplankton NPP, fuels high biological oxygen demand in bottom waters from late spring to early autumn (Staresinié et al., 1982; Alvisi and Cozzi, 2016). During this period, dissolved oxygen levels in bottom water are low and episodes of bottom water hypoxia occur frequently. The timing, duration and spatial extent of these episodes varies from year-to-year. During 1972–2012, the frequency of hypoxic events was high each year from 1988 to 1991, especially in the western NA (Djakovac et al., 2015). While hypoxia is related to phytoplankton blooms induced by riverine inputs of nutrients during spring and autumn, summer hypoxia occurs during periods of low runoff suggesting that strong vertical stratification and weak circulation promoted hypoxic events during summer (Alvisi and Cozzi, 2016). Hypoxic zones range in spatial extent from several km² to 4,000 km² and ultimately affect most of the NA (Riedel et al., 2008). Thus, repeated episodes of hypoxia caused mass mortalities of benthic macrofauna during the 1970s and 1980s, the largest of which were caused by anoxic episodes during the summers of 1974, 1977, 1983, 1988, and 1989 (Fedra et al., 1976; Stachowitsch, 1984; Ott, 1992).

In addition to phytoplankton sedimentation, the formation and deposition of macro-aggregates (“mucilage” or mare sporco) of organic matter rich in bacteria also fuels high biological oxygen demand, anoxia and mass mortalities during May-September (Degobbis, 1989). Mucilage episodes are unique to the NA in terms of the size of macro aggregates (up to 3 m across) and their abundance and spatial coverage (hundreds of km² in both coastal and offshore waters). Episodes have occurred sporadically since the seventeenth century (Fonda Umani et al., 1989; Sellner and Fonda Umani, 1999; Giani et al., 2005), but increased in frequency during 1980–2004 (Danovaro et al., 2009). Major episodes (2–6 weeks in duration) occurred annually during 1988–1991, 1996–1998, and 2000–2004. Following a prolonged episode from December, 2006 through April, 2007, episodes of relatively short duration were recorded in 2014 (August) and 2018 (July–August). Although these episodes tend to occur following increases in river runoff, and mucilage production occurs during diatom blooms (McKinney, 2007), anthropogenic nutrient enrichment may not be directly responsible (Degobbis et al., 2005).

Since the 1990s, many toxic or potentially toxic phytoplankton species have inhabited the NA including dinoflagellates (e.g., Dinophysis spp., Alexandrium minutum, Noctiluca scintillans, Protoceratium reticulatum, Gonyaulax spinifera, and Lingulodinium polyedrum), diatoms (Psedo-nitzschia spp.), and a raphidophyte (Fibrocapsa japonica) (Sellner and Fonda Umani, 1999; Fonda Umani et al., 2004; Mikaelyan et al., 2014; Corriero et al., 2016). The first occurrence of human intoxication due to the consumption of bivalves contaminated with diarrhetic shellfish poisoning (DSP) toxins was documented in 1989 (Pistocchi et al., 2012). The first detection of bivalves contaminated by toxins produced by a bloom of Alexandrium minutum that cause paralytic shellfish poisoning (PSP) was in 1994 (Honsell et al., 1996). Domoic acid, produced by the diatom Pseudo-nitzschia spp., is a potentially lethal phycotoxin that causes amnesic shellfish poisoning (ASP), and was first detected in bivalve-tissue in 2006 (Ujević et al., 2010). Bivalves farmed in coastal waters of Emilia Romagna and in Slovenian mariculture areas have also been contaminated frequently by Dinophysis spp. toxins (okadaic acids and dinophysitoxin) that cause DSP (Francé and Mozetič, 2006; Pistocchi et al., 2012). Alerts due to the presence of biotoxins in the Gulf of Trieste increased steadily from 10 yr^–1 in 2005 to 60 yr^–1 in 2010 and then declined to 1 yr^–1 in 2013 and 2014 (Lipizer et al., 2017).

Given the pattern of interannual variations in river flow and associated nutrient inputs described above, it is unlikely that the frequency of toxic phytoplankton episodes is related directly to the magnitude of nutrient enrichment per se. However, relative inputs of N and P can affect the abundance of potentially toxic phytoplankton as well as toxin production (Glibert et al., 2014; Glibert and Burford, 2017). Many marine dinoflagellate species are more abundant and toxic when N is in stoichiometric excess over P (N:P > 16), as is the case in the Northern Adriatic where N:P ratios have increased from a mean of 64 during the 1960s to > 100 in more recent years.

CB is a long, narrow system (315 km long and 5.6–56 km wide) with a surface area of 11,600 km² (Xiong and Berger, 2010) and three salinity zones (Harding et al., 2019): oligohaline < 5–18, mesohaline 5–18, and euryhaline > 18. A relatively deep (10–50 m) central channel runs the length of the mesohaline zone. The Susquehanna River discharges directly into the oligohaline zone and, with a mean flow of 36 km³ yr^–1 (range 20–60 km³ yr^–1), accounts for 60–90% of the freshwater input to CB (Schubel and Pritchard, 1986; Zhang et al., 2013). Except for the effects of tropical storms, the volume transport of the Susquehanna has a well-defined annual cycle with maximum flow during late winter-early spring and low flow during summer. Large interannual variations occur (20–60 km³ yr^–1) in both the magnitude and timing of seasonal maximum and minimum flows.

The Susquehanna drives an estuarine circulation characterized by seaward transport of low salinity surface water and landward (counter) transport of higher salinity bottom water, a circulation pattern that increases the residence time of nutrients in the Bay (Pritchard, 1967; Shen and Wang, 2007; Du and Shen, 2016). Peak flow during winter-spring sets up buoyancy-induced vertical stratification that persists into early fall, with high flow years having a stronger halocline separating upper and bottom layers (Boynton and Kemp, 2000; Kemp et al., 2005).

Dissolved nitrate accounts for most input of new N while most P input is associated with the input of suspended sediments (Boynton et al., 1995; Zhang et al., 2013). It is estimated that N and P loads increased by factors of 7 (N) and 13 (P) between the beginning of the Anthropocene²¹ and 1985–86 when loads reached 0.081 × 10⁹ kg N yr^–1 and 3.75 × 10⁶ kg P yr^–1 (Boynton et al., 1995; Boynton and Kemp, 2000). N loading during 1945–2011 increased from 1945 to around 1990, followed by a steady decline (Zhang et al., 2013). Mean inputs during 1978–2011 were 0.071 × 10⁹ kg N yr^–1 and 3.3 × 10⁶ kg P yr^–1 (Hirsch, 2012). Relative to riverine inputs, inputs from direct precipitation to the Bay are small (7% of TN and 15% of TP). N-fixation is assumed to be negligible (Marino et al., 2002), and benthic denitrification that is estimated to remove ∼ 0.034 × 10⁹ kg N yr^–1 or ∼40% of annual riverine N inputs (Feng et al., 2015).

The annual cycle of phytoplankton biomass in the Bay features a large spring bloom which is most pronounced in the mesohaline Bay where integrated water column Chl can exceed 1,000 mg m^–2 (Malone et al., 1988, 1996) due to phytoplankton growth, the concentrating effect of the two-layered estuarine circulation on phytoplankton biomass, and low grazing pressure (Malone et al., 1988, 1996; White and Roman, 1992). The spring bloom is dominated by large diatoms that deplete nitrate and silicate in surface waters and have high sinking rates (Conley and Malone, 1992) resulting in the deposition of phytoplankton biomass to bottom waters and benthic sediments (Malone et al., 1988; Harding et al., 2002). Remineralization of this organic matter during summer results in oxygen depletion and a large efflux of dissolved phosphate and ammonium into bottom waters which gradually mix into surface waters (Boynton et al., 1995; Boynton and Kemp, 2000) where they fuel local, episodic summer blooms of small diatoms, chlorophytes, cyanobacteria, dinoflagellates, and picophytoplankton (Malone, 1991; Adolf et al., 2006; Marshall et al., 2009). For the most part, summer blooms are confined to the euphotic zone and there is relatively little sedimentation to the benthos. Consequently, the positive feedback that characterizes the Baltic (Figure 5) is not well developed in CB.

Annual cycles (1982–2004) of Chl and phytoplankton NPP for the oligohaline, mesohaline, and polyhaline zones of the Bay have been characterized as follows (Harding et al., 2002, 2020):

Seasonal to interannual variations in riverine N input regulate the distribution of phytoplankton biomass in both time and space (Harding et al., 2020), consistent with the conclusion that P limitation occurs predominantly in the oligohaline reach of the Bay, while N limits phytoplankton biomass and NPP on the scale of the CB ecosystem (Malone et al., 1996). Anthropogenic nutrient loading led to a 5- to 10-fold increase of surface Chl in the polyhaline salinity zone and a 1.5- to 2-fold increase in the oligohaline and mesohaline zones from the 1950s to the 1990s (Harding et al., 2002, 2020). However, while euphotic zone Chl continued to increase in the oligohaline and mesohaline zones during 1985–2015, no trend was observed in the polyhaline zone (Harding et al., 2020).

Dissolved oxygen in bottom waters declines rapidly during April-May leading to hypoxic bottom water during July–September throughout most of CB, with hypoxia and anoxia developing in the deep channel of the mesohaline (Figure 8; Officer et al., 1984; Li et al., 2015). This cycle is driven by the spring bloom and bacterial decomposition of phytoplankton biomass deposited during the course of the bloom (Malone, 1991; Smith and Kemp, 1995). Summer bottom water hypoxia has been documented in the mesohaline salinity zone throughout the twentieth century and, while interannual variability is substantial, the volume of hypoxia has increased significantly since WWII as anthropogenic N loading to the Bay increased (Brush, 2009; Murphy et al., 2011; Li et al., 2016; Testa et al., 2017). Under current conditions, it is estimated that the missing biomass of benthic animals due to hypoxia is equivalent to ∼ 10,000 metric tons of carbon annually which represents ∼5% of the Bay’s secondary production (Diaz and Rosenberg, 2008).

In addition, long-term increases in Chl, combined with increases in sediment loading to the Bay, have been linked to bay-wide declines in the spatial extent of seagrass beds during the 1960s and 70s due to reduced light penetration (Orth and Moore, 1983; Kemp et al., 2005). As a result, it is estimated that the spatial extent of seagrass decreased by 46% between the 1930s and the 1990s (Kemp et al., 2005). These decreases persisted through the 1990s (Orth et al., 2010; Harding et al., 2019) but show signs of increasing since ∼2006 during a sustained low-flow period (Testa et al., 2019).

Harmful algal blooms in CB are not a new phenomenon, but they may be increasing in frequency and diversity (Glibert et al., 2001). As in most coastal ecosystems, dinoflagellates account for most toxic or potentially toxic phytoplankton. Prorocentrum cordatum, Karlodinium veneficum, and Margalefidinium polykrikoides have occurred with increasing frequency in recent years (Tango et al., 2005; Li et al., 2015; Reece, 2015; Lin et al., 2018). Beginning in 2007, and blooms of Alexandrium monilatum also began occurring in the lower bay in most summers following blooms of M. polykrikoides (Reece, 2015). P. cordatum, M. aeruginosa and the diatom Pseudo-nitzschia spp. are common throughout the bay; K. veneficum is limited to the oligohaline and mesohaline zones; and Dinophysis acuminata and M. polykrikoides are limited to the mesohaline and polyhaline zones. All species were found in all seasons, but are generally most abundant during summer and fall²².

Coastal upwelling centers represent an extreme case of natural new nutrient inputs and are expected to exhibit equally extreme blooms of both benign and toxic phytoplankton (Kudela et al., 2010). Unlike most coastal ecosystems impacted by eutrophication, the physical, chemical, and ecological characteristics of eastern boundary currents, such as the California current system, are dominated by wind-driven coastal upwelling which enriches these waters with nutrients far in excess of riverine inputs of anthropogenic nutrients (Anderson et al., 2008).

The Santa Barbara Channel (SBC) in the Southern California Bight is an elongated ecosystem (4,000 km²) bounded to the north by the California coast, to the south by the Channel Islands 50 km offshore, and by narrow, relatively shallow sills to the west (475 m) and east (200 m) (Bograd et al., 2002). Between these sills lies the Santa Barbara Basin which is periodically flushed with cool, dense, oxygenated water that flows over the western sill and sinks into the basin. Waters below the deepest sill are dysaerobic (dissolved oxygen < 0.1 ml liter^–1) between flushing events (Bernhard and Reimers, 1991). Flushing event frequency of deep water is driven by interannual climate (ENSO) variability, which switches the basin between a stagnant water column and a ventilated one, a process that amplifies variability in the magnitude of oxygen depletion (Bograd et al., 2002; Wang et al., 2017).

Upwelling is by far the largest source of nutrients to the SBC ecosystem as a whole, especially during the late spring and early summer (Warrick et al., 2005; Howard et al., 2017). N input via upwelling is estimated to be ∼ 0.21 × 10⁹ kg yr^–1, a rate that is 2–3 orders-of-magnitude greater than the riverine inputs during wet, El Nino years (Warrick et al., 2005). The largest river flowing into the Channel is the Santa Clara River with an annual mean flow of 0.16 km³ yr^–1. Although total riverine N input is much less than that supplied by upwelling, the timing these inputs are seasonally out of phase with river runoff peaking during winter storms and upwelling during spring-summer (Warrick et al., 2005; Anderson et al., 2008). In addition, upwelling inputs occur at the western end of the basin while riverine inputs occur along the eastern boundary.

Although the distribution of phytoplankton biomass in the SBC is complex, the long-term climatology of surface Chl concentration (derived from satellite ocean color images) reveals two consistent features: (1) maximum biomass north of the Channel Islands and over the Santa Barbara Basin and (2) a second maximum along the coast in the eastern SBC where the continental shelf widens (Warrick et al., 2005; Brzezinski and Washburn, 2011). It is estimated that new production²³ from river runoff accounts for 0.03–4% of total new production (river + upwelling) within the SBC. Since upwelling is generally lower during El Niño, river contributions are most significant during these years (∼4% of new production in 1998) (Warrick et al., 2005). However, while high phytoplankton biomass over the basin and north is fueled primarily by upwelling, the near shore maximum to the east is fueled by a combination of mechanisms including upwelling, diurnal internal waves, and river runoff (McPhee-Shaw et al., 2007). Upwelling dominates increases of inner−shelf DIN during March-May and accounts for more than half of annual advective inputs. Internal waves are an important source during summer, and riverine inputs are significant during winter.

The frequency of toxic phytoplankton blooms increased during the first decade of this century (Anderson et al., 2009). Episodic blooms of potentially toxic species (the diatom Pseudo-nitzschia spp. and the dinoflagellate Lingulodinium polyedrum) occur along the coastline of the Southern California Bight including the SBC (Bialonski et al., 2016). Blooms of Pseudo-nitzschia and Lingulodinium tend to occur during spring-summer when upwelling events are most frequent (Schnetzer et al., 2013). However, this does not preclude the possibility that the growth of these algae, their toxicity, and the frequency or duration of toxic events may be exacerbated by anthropogenic nutrient inputs (Anderson et al., 2008). Blooms can occur during fall or winter, presumably fueled by riverine inputs of nutrients (Kudela et al., 2010). Other toxin-producing phytoplankton that occur here include dinoflagellate species from the genera Akashiwo, Alexandrium, Cochlodinium, and Dinophysis.

Bottom water oxygen in the basin has been decreasing since about 1850 due to upper ocean warming (decreases in solubility and vertical mixing) (Wang et al., 2017), and the frequency and duration of hypoxia in California Current waters has been increasing as the Eastern Pacific oxygen minimum zone expands (Stramma et al., 2010; Wang et al., 2017). The spatial extent of oxygen depleted deep water is estimated to be on the order of 800 km². Low−oxygen water (mean = 0.5 ml liter^–1 during 2008–2011) from the oxygen minimum enters the basin over the western sill and is further depleted of oxygen by the decomposition of organic matter from productive surface water fueled primarily by upwelled nutrients and exacerbated by riverine inputs of anthropogenic nutrients (Ohkushi et al., 2013). An indication of the significance of the latter is the observation that dissolved oxygen in nearshore bottom water (within 10 km of the shoreline) has declined by up to four times faster than reported for more offshore waters over the last 15 years (Booth et al., 2014). The trend has been briefly interrupted when dense oxygenated water flows into the basin during transitions from El Niño to La Niña conditions (Wang et al., 2017). Thus, anthropogenic nutrients can have an impact on local spatial scales within the larger upwelling ecosystem (Howard et al., 2017).

The 1998 National Strategy for the Development of Regional Nutrient Criteria²⁸ presented an approach the U.S. Environmental Protection Agency (EPA) now uses to work with States and Tribes to adopt nutrient criteria as part of State water quality standards. Each State is required to submit a list of impaired and threatened waters within their jurisdiction and to establish priorities for the formulation of “Total Maximum Daily Loads” (TMDLs) of nutrients and other pollutants to these water bodies²⁹. Implementation by each State may occur via direct management actions and water quality trading programs³⁰. In addition to TMDLs, the EPA funds state-level diffuse source management programs that incorporate “Best Management Practices” (BMPs) (Liu et al., 2017). BMPs can be implemented to reduce nutrient loading from urban sources by using detention basins, constructed wetlands, vegetative swales³¹, and rain gardens. A critical BMP for agricultural fields is to improve nutrient use efficiency, i.e., select the right fertilizer and tune applications to match plant requirements and reduce nutrient losses. BMPs also include contour farming³², crop rotation, cover crops, no tillage, grassed waterways, constructed wetlands, grade stabilization structures, and vegetated buffer strips.

However, implementation has been slow and uneven because it is not a legal requirement, funding is insufficient, and outcomes are uncertain due to time lags between management actions and improvements in trophic status and hysteresis. In addition, although technical tools and management practices exist for cost-effective reductions in nutrient inputs, implementation remains the primary problem, which suggests that new policy approaches should be tried. A broader use of incentives, trading, and corporate stewardship is clearly needed.

Numerous policy instruments have been propagated in an effort to protect coastal ecosystems (Joint Research Centre [JRC], 2014). The Nitrates Directive³³ is intended to protect water resources against nitrates from agricultural sources, and the Urban Waste Water Treatment Directive (UWWT) aims at controlling emissions of nitrogen and phosphorus from point sources such as urban sewage effluents (Joint Research Centre [JRC], 2014). They both support the Water Framework Directive (WFD) that aims to achieve Good Ecological Status of water, including coastal waters, by 2020 (Pulido-Velazquez and Ward, 2017). The Marine Strategy Framework Directive has a specific descriptor focused on eutrophication to achieve Good Environmental Status in marine waters. These legal instruments are further complemented by international initiatives of Regional Seas that aim to control pollution from land-based sources and maritime transport, e.g., the Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR), the Helsinki Convention on the Protection of the Marine Environment of the Baltic Sea Area (HELCOM, 2018b), and the Convention for the Protection of the Mediterranean Sea (Barcelona Convention).

However, despite numerous policy directives and conventions, the most recent assessment of trophic status of European coastal ecosystems (European Environmental Agency [EEA], 2018) shows that 40% of coastal (continental shelf) waters and 66% of transitional waters (estuaries) failed to achieve the objective of good ecological status³⁴ (Poikane et al., 2019). Therefore, a new Fertilizing Products Regulation was adopted in 2019, and Carvalho et al. (2019), made the following recommendations to address the ongoing eutrophication problem:

It is too early to assess whether these new EU fertilizer regulations will be effective, particularly in view of sometimes lengthy recovery times (O’Higgins et al., 2014).

The Chinese Government regards agriculture as the primary driver of economic development in the twenty-first century, and Chinese policy calls for implementing Integrated Nutrient Management (INM). The major challenge is how to increase crop production to meet food-demand while also increasing nutrient use efficiency (NUE) to protect the environment (Zhang F. et al., 2012). Nutrient management strategies have been successful in terms of maintaining maximum crop production, but attempts to improve NUE have been met with limited success in industrial agriculture. As a result, agricultural ecosystems are maintained in a state of nutrient saturation and are inherently leaky because chronic surplus additions of N and P have been used to maximize production.

INM was launched in 2005 with funding for 5-years (6 billion Yuan or $923 million US) (Zhang F. et al., 2012). The overall goal was to maximize both crop production and NUE by informing “root zone-rhizosphere management³⁶ through soil testing and controlled fertilizer application. The objective is to manage soil nutrient supply in the root zone to match the quantity required for crop production. Rhizosphere management was proven to be an effective approach for increasing both crop productivity and NUE for sustainable agricultural production. However, there is still a long way to go to realize the objectives of NUE.

China’s agricultural advisory system is fragmented (Ma et al., 2013). Increasing agricultural production is the primary role of the Ministry of Agriculture while environmental protection is the primary role of the Ministry of Environmental Protection. However, the Ministry of Agriculture has the greatest influence on farmers through financial incentives. In addition, regional Governors focus on industrial development and employment. As a result, regional governments often pay little attention to sustainable food production and environmental protection (Ma et al., 2013).

Fortunately, the Chinese Government is aware of the importance of INM to sustainable production and supports its adoption on a national scale. The key steps of INM are to (1) optimize inputs of anthropogenic nutrients by considering all possible nutrient sources, (2) match nutrient applications to crop requirement in this context, and (3) reduce N losses from industrial agriculture (Zhang F. et al., 2012). This has been a challenge given the transitions China has been going through in terms of both evolving from an agrarian society to an industrial society and from a government-led system to a market-led system. Adapting to these trends will require Governmental support for agriculture to deemphasize fertilizer-subsidies and increase direct support for farmers who can demonstrate that they are not only contributing to increases in food production but are also increasing NUE and environmental protection.