Management pathways for the successful reduction of nonpoint source nutrients in coastal ecosystems

Abstract

Eutrophication remains a threat to coastal habitats and water quality worldwide. The U.S. Clean Water Act resulted in reductions of nutrient loading from point sources but management of nonpoint sources (NPS) of nutrients remains challenging despite efforts over at least three decades. The hydrological factors, best management practices (BMPs) and regulatory mechanisms that target nutrient NPS and improve coastal ecosystem function are poorly understood. We identified three case study sites in the U.S. with sufficient NPS management and monitoring history to quantify changes in estuarine habitat and water quality following BMP implementation and regulation targeting nutrient NPS. Utilizing publicly available data, we compared sites that are geographically distant and hydrologically distinct. We found that BMPs targeting NPS loads from surface waters into Roberts Bay (Florida) and Newport Bay (California) significantly reduced nutrient concentrations and harmful algal blooms within ~20 years. Improvements occurred despite concurrent human population growth within both watersheds. Conversely, we found that the majority of BMPs implemented within the Peconic Estuary (New York) watershed targeted surface waters despite a dominance of nitrogen inputs (97%) from groundwater and atmospheric sources. Declines in habitat and water quality in Peconic Estuary may be due to a failure to control the dominant nutrient sources and the long residence time of nitrogen in groundwater. Compared to surface water, reducing groundwater and atmospheric nutrients face greater technical and financial challenges. Improvements to Peconic Estuary may occur with further reductions in surface water inputs and as nutrients leach out of the groundwater. Although the effectiveness of specific NPS BMPs has been examined at small spatial scales, our study is the first to quantify improvements at a watershed scale. We showed that successful NPS management pathways are those which targeted the dominant sources of nutrients to coastal ecosystems and applied multiple BMPs within watersheds.

1. Introduction

Eutrophication has resulted in the global deterioration of aquatic ecosystems despite concerted efforts over the last 50 years to reduce nutrient pollution. Inadequate controls of nitrogen and phosphorous from watershed sources have negatively impacted ecosystem function (Deegan et al., 2012; Kemp et al., 2005), human health (Van Dolah, 2000), recreation (Pinto et al., 2010) and fisheries (Breitburg et al., 2009; Diaz and Rosenberg, 2008). Sources of excess nutrients include point sources, such as wastewater treatment discharge that have a clear point of origin, and nonpoint sources (NPS), such as stormwater runoff, groundwater and atmospheric deposition which lack clear origins (Craig and Roberts, 2015; Hutchings et al., 2014). Implementation of the Clean Water Act (1972) in the U.S. (Birkeland, 2001), stimulated efforts to reduce nutrient pollution, improve water quality and restore function to aquatic ecosystems (Bricker et al., 2008). Successful examples include Boston Harbor (Diaz et al., 2008), which exhibited significant improvements in water quality following wastewater treatment upgrades to remove excess nutrients from effluent. Though many studies have demonstrated improved ecosystem function when point source inputs were reduced (Borja et al., 2010; Gross and Hagy, 2017), NPS pollution often remains a barrier to achieving state and national water quality goals (Craig and Roberts, 2015). This challenge has also been observed in the Baltic Sea, where a 50–70% reduction in nutrient loads occurred following wastewater treatment improvements but failure to adequately address NPS, such as agricultural sources, has ultimately prevented nutrient load targets from being met (Reusch et al., 2018). Furthermore, changes in precipitation patterns due to climate change are predicted to increase nitrogen loading in the U.S., particularly in the Northeast (Sinha et al., 2017), and elsewhere globally, including the Baltic Sea (Meier et al., 2012). Thus, it is increasingly important to address both current and future sources of nutrient loads contributing to coastal eutrophication.

More than half of nutrient loads into coastal ecosystems originate from NPS (Greening and Janicki, 2006; Kennish et al., 2007; Paerl et al., 2004), including sources from agriculture (Kronvang et al., 2005), septic systems (Bicki et al., 1984; Lloyd, 2014), urban runoff (French et al., 2006), groundwater (Capone and Bautista, 1985) and atmospheric deposition (Carstensen et al., 2006). However, regulation of nutrient NPS is difficult in the U.S. because regulatory control authority was not provided in the Clean Water Act until 1987 when Section §1329 of the amendments specified that states needed greater leadership and support to address nutrient NPS (Birkeland, 2001). Though NPS programs addressing nutrient pollution have been developed across the U.S., only 19 states have enforceable regulations for agricultural NPS (Craig and Roberts, 2015) and most rely on voluntary implementation of best management practices (BMPs) (Tzankova, 2012). These BMPs are effective and practical approaches to preventing or reducing pollution generated by NPS. Additional challenges exist for NPS management because certain sources of nutrients may be easier to tackle through BMPs than others. For example, given the currently available technologies, long residence times of groundwater and regional scale management of atmospheric nitrogen, it is easier to address NPS nutrients emanating from surface waters rather than from groundwater or atmospheric deposition (EPA, 2015; King et al., 2012; RBF Consulting, 2013). However, there have been emission control programs that successfully decreased nutrient inputs from atmospheric deposition and resulted in water quality improvement (Boesch, 2019; Eshleman et al., 2013; Williamson et al., 2017). Further, the implementation of agricultural BMPs over 30 years in Denmark reduced nitrate concentrations in groundwater, but human health thresholds are still exceeded in some locations (Hansen et al., 2017).

Typically, the effectiveness of NPS management has only been evaluated for BMPs that targeted small scale (<2 km²) reductions in nutrient runoff (Dillaha et al., 1988; Fennessy and Cronk, 1997; Lee et al., 2010). Far less is known about the pathways that result in successful NPS management and the subsequent restoration of habitat and water quality for whole ecosystems, especially for estuaries (except via ecosystem modeling, e.g. Jansson et al., 2019; Khangaonkar et al., 2018). Therefore, we ask, are there examples of successful NPS management in coastal ecosystems? If so, what processes contributed to the successful reduction of NPS nutrient loads and improvements in habitat and water quality? To address these questions, we evaluated three U.S. estuaries that varied in management of NPS nutrients. We considered trends in indicators of estuary condition, including water column nutrient concentrations, phytoplankton biomass (chlorophyll a), density and cover of seagrass vegetation, including Zostera marina (eelgrass) and Thalassia testudinum (turtle grass), and density of ephemeral macroalgae (Ulva spp.). Ulvoid macroalgae were included in this analysis because they respond rapidly to nutrient enrichment (Nelson and Young, 2015; Teichberg et al., 2010) and blooms are common to shallow estuaries (Nixon et al., 2001). Since nutrient pollution rates are often correlated with human population density (Vitousek et al., 1997), we analyzed trends in human populations for the counties encompassing the watershed boundaries of each case study sites including Sarasota County (Roberts Bay, Florida), Orange County (Newport Bay, California), and Suffolk County (Peconic Estuary, New York). To identify management pathways targeting NPS nutrients that resulted in significant improvements to estuarine habitat and water quality, we quantified the number of BMPs targeting nutrient NPS and the length of time the BMPs were implemented. We expect that management pathways which targeted the dominant sources of watershed nutrients inputs would be most effective at reducing nutrient concentrations in estuaries. In addition, improvements in habitat and water quality would be greatest in estuaries where watershed agriculture and stormwater nutrient loading was greater than contributions from groundwater and atmospheric deposition due to the challenges of addressing the latter nutrient sources.

2. Materials and Methods

2.1. Study Site Selection

We selected three U.S. estuaries with different NPS management strategies. Criteria for site selection included: 1) earliest available records showed that the system was dominated by nonpoint sources (our systems had <15% of nutrient loads from point sources), 2) availability of at least 20 years of water quality data, 3) evidence of habitat degradation due to nutrient pollution within the last 30 years and 4) watersheds that were moderate in size (<1000 km²). The watershed sizes of our sites were moderate compared to the largest estuarine watersheds in the U.S., Mississippi River Delta with 2.9 million km² and Chesapeake Bay with 165,759 km² (NOAA, 2020), and the smallest, Wells and Netarts Bays (22 and 31 km², respectively) (NEEA, 2020). We selected moderately sized watersheds so that the effects of management actions could be more readily identified compared to larger watersheds where detecting the effects nutrient NPS management is more difficult (Kennish et al., 2007). Based on these criteria, Roberts Bay (Florida), Newport Bay (California) and Peconic Estuary (New York) were selected as case study sites (Figure 1a).

Figure 1:

a. Map of the U.S. with states where case studies are located colored dark gray. 1. Roberts Bay (Florida). 2. Newport Bay (California). 3. Peconic Estuary (New York). Extent indicators of case study sites are approximate. b. Human population estimates for each county encompassing watershed boundaries for each case study sites.

2.2. Data Acquisition and Analyses

After investigating the history of nutrient pollution and NPS management in the three case study sites, habitat and water quality data were acquired from public sources (Supplement Table A-1). Utilizing existing datasets allowed us to conduct a broadscale temporal study, but it has distinct benefits and drawbacks. The benefits include reduced costs and time spent collecting data, as well as the ability to conduct temporal and cross-sectional analysis in a timely fashion (Pederson et al., 2020; Sprague et al., 2017). Problems arise with missing data and metadata as well as assumptions about quality control (Pederson et al., 2020; Sprague et al., 2017). We acknowledge that gaps, spatial bias and grouping of datasets may have affected our results. Additionally, we assumed that data were collected in accordance with standard methods and subject to appropriate quality assurance and quality control, however we performed additional inspections for data completeness and any obvious errors were excluded. The decision to exclude data was based on the availability of other sources of information and best professional judgement (EPA, 2016b). Nutrient concentrations less than methodological detection limits were replaced with half the detection limit (Helsel, 2006).

Due to departures from normality and homoscedasticity, data were analyzed using Trend Free Pre-whitening Mann-Kendall using the “modifiedmk” package in R v. 4.3 (McLeod, 2011; R Core Team, 2014; Sandeep, 2020; Wickham, 2009). Pre-whitening adjusts for serial autocorrelation in trend analyses (Danneberg, 2012). Annual scale trend analyses (Gilbert, 1987) were performed on raw data (Z. marina density and cover, T. testudinum cover, human population) or on data that were averaged by month then by year (water column nutrients, pheophytin-corrected chlorophyll a, Ulva spp. biomass).

To characterize recent land use for each system, we constructed intersections of estuarine drainage area polygons (NOAA, 1999) and watershed boundary datasets (USGS, 2013) using 12 digit hydrologic unit (HUC-12) polygons. The resulting polygons were then used to compute watershed scale information based on the National Land Cover Database (NLCD) 2011 (Homer et al., 2015; MRLC, 2006). We categorized Low, Medium and High Intensity development as impervious surfaces that cover 20–49%, 50–79% and ≥80% of total area, respectively based on the modified Anderson Land Cover Classification system (Anderson, 1976) used by MRLC (2006).

We identified the pathways taken to address nutrient NPS by classifying BMPs into types and quantified the number of BMP types that were utilized and the length of time over which BMPs were implemented for each nutrient source (Supplement Table B-1). Best management practices were found through published papers, reports, and personal communication with water quality management personnel for each case study site (Supplement Table B-1). The modeled nutrient NPS loads were identified by Total Maximum Daily Load (TMDL) documents (Peconic Estuary Program, 2007; Petrus and Lassiter, 2005; Santa Ana Regional Water Quality Control Board, 1998a). Total Maximum Daily Loads are maximum pollutant amounts allowed in a US water body and still maintain human health and ecosystem quality standards (U.S. Clean Water Act, 1972). These provided the earliest available quantitative estimates of nutrient loading for each system. We defined surface water as runoff over agricultural or natural lands and stormwater as runoff over urban areas. Groundwater was defined as water moving through sediment that is not visible at the surface and atmospheric deposition as the wet or dry particulates that settle onto land in the watershed or the surface water of the estuary.

Atmospheric deposition contributes significantly to nitrogen loading for some coastal estuaries (Paerl, 1997; Sigua and Tweedale, 2003), including Peconic Estuary (Peconic Estuary Program, 2007). However, sources of atmospheric nitrogen tend to occur at regional scales (Paerl et al., 2002) and are regulated by the state and federal governments (Swift, 2000; U.S. Clean Air Act, 1970). Therefore, we discuss the relative importance of atmospheric deposition for each system, but quantitative analysis of management efforts to reduce atmospheric deposition were beyond the scope of this study, which focused on site-specific, local management actions.

3. Case studies

3.1. Roberts Bay, Florida

3.1.1. Overview

Roberts Bay is a wide, shallow (average depth ~2 m), urbanized embayment located at the southern end of Sarasota Bay in Sarasota County (Figure 1a–1). It is the smallest of our three case study sites (3.3 km²) with a 146 km² watershed comprised of the Phillippi and Matheny Creek watersheds. It has porous, carbonate sediments (Barr, 1996) (Figure 1a–1). Roberts Bay has a 19–37 day flushing time (Sheng et al., 1991) due to limitation of tidal flushing by Siesta Key to the west (Figure 1a–1). The watershed is prone to frequent pulses of stormwater throughout the June to September rainy season (Tomasko et al., 2005).

Anthropogenic influence within the watershed dates back to the 1800’s when sawgrass (Cladium jamaicense) wetlands surrounding the bay were drained for agriculture and natural tidal creeks were straightened to allow for efficient drainage (Sarasota County, 2011). These modifications increased the rate of stormwater runoff into the bay (Sarasota County, 2011), reduced groundwater infiltration and natural removal of nutrients by sediments and vegetation (Reinelt and Horner, 1995). The estimated 1948 land cover was 61.7% forest, open area and park, 25% wetland, 9.5% agriculture, 3.2% open water with just 0.3% residential land use in the Roberts Bay North watershed (Sarasota County, 2011). Census data showed a 136% increase in the population of Sarasota County from 1976–2014 (Figure 1b, Table 1A). To accommodate increased urbanization, 72% of land uses were converted from forest and agricultural to residential and developed open space. By 2013, there were no remaining forests and just 1.3% of the land was used for agriculture. As urbanization increased, flood control became a key priority (Sarasota County, 2011).

Table 1.

Roberts Bay, Florida. Trend-Free Pre-whitened Mann-Kendall trend analyses on annual data (τ) for census data in Sarasota County (A), water column nutrient (B-E) and chlorophyll a (F) concentrations, and seagrass cover (G). Values in bold indicate significant trends (p < 0.05).

Factor	Years	n	τ	p value
A. Human population	1983–2014	32	0.991	<0.0001
B. NOX	1983–2014	30	−0.581	<0.0001
C. TKN	1983–2014	32	−0.454	0.0004
D. TN	1983–2014	30	−0.404	0.002
E. TP	1983–2014	32	−0.312	0.014
F. Chlorophyll α	1998–2014	17	0.066	0.753
G. Seagrass (primarily T. testudinum) cover	1988–2015	12	−0.333	0.251

By 1992, 61% percent of the nutrient loading into the bay arose from stormwater flow (Petrus and Lassiter, 2005), which raised concerns about the effects of elevated nutrients and pesticides in stormwater (McKnight et al., 1995; Sklar and Browder, 1998). The concurrent degradation of habitat and water quality including: chlorophyll a concentrations that reached 25 μg l⁻¹, loss of mangrove stands, declines in finfish and shellfish landings, reduction in wetland area, and a 28% loss of seagrass area between 1950–1988 were listed as motivations to include the greater Sarasota Bay in the National Estuary Program (NEP) in 1992 (SBNEP, 1992; Tomasko et al., 2005). When Sarasota Bay was included in the NEP, the water quality of Roberts Bay was ranked among the poorest of Sarasota Bay segments based on the Trophic State Index; a metric that scores water bodies based on nutrient concentrations, chlorophyll a and water clarity (SBNEP, 1992).

Stormwater is the greatest contributor of nutrients to Roberts Bay (Dillon and Chanton, 2008). However, enriched groundwater contributes significantly to the nitrogen load (Dillon and Chanton, 2008) and it was estimated that nutrients from septic systems constituted 33% of the 1992 loading (Petrus and Lassiter (2005), leading to groundwater contamination and deteriorated water quality (Sarasota County, 2011). The remaining sources of nutrients, treated wastewater and atmospheric deposition, respectively accounted for 3.6% and 1.5% of the nutrient load to Roberts Bay in 1992 (Petrus and Lassiter, 2005).

3.1.2. Data analyses

3.1.2.1. Water column nutrients

Nitrogen levels declined significantly from 1983 to 2014 and included decreases in average Nitrogen Oxide (NO_X), Total Nitrogen (TN) and Total Kjeldahl Nitrogen (TKN) concentrations by 89%, 41% and 34%, respectively (Figure 2a, Table 1B–D). Total Phosphorus (TP) also declined significantly by 2014 and was 57% lower than in 1983 (Figure 2a, Table 1E).

Figure 2:

Mean water column nutrients, chlorophyll a concentrations and seagrass cover for Roberts Bay (Florida) over time. a. Total Nitrogen (TN), Total Kjeldahl Nitrogen (TKN), Total Phosphorus (TP), Nitrogen Oxide (NO_X) concentrations from 1983 to 2014. b. Chlorophyll a concentration from 1998 to 2014 (1990 value included for reference but extraction methods prevented inclusion in trend analysis). c. Total seagrass cover (primarily Thalassia testudinum) in hectares from 1988–2012 with target value (139.5 hectares) indicated by horizontal line. Where present, error bars represent standard error.

3.1.2.2. Primary producers

Chlorophyll a did not change significantly from 1998–2014 (Figure 2b, Table 1F). Seagrass beds in Roberts Bay were dominated by T. testudinum (Tomasko et al., 1996) and, despite a decline in the year 2002 to 80% of historical values, no significant trend for seagrass cover was found (Table 1G;Figure 2c).

3.1.3. Best Management Practices and regulation of nutrient loads

Prior to the 1990s, stormwater management focused on volume control (Roy et al., 2008). However, a growing emphasis on ecosystem preservation resulted in a shift by municipal and regulatory agencies around the world to low impact development practices that reduce negative effects of stormwater on aquatic ecosystem health (Roy et al., 2008). For example, residents of Sarasota County were encouraged to construct vegetated stormwater ponds that reduce flood risk and improve water quality through biotic uptake of nutrients (Sarasota County, 2011; USDA, 1993) and the Marine Resources Council of Florida launched the Yards and Neighborhoods Program in Sarasota in 1994 (now called Florida Friendly Landscaping™) (Supplement Table B-1). The voluntary program encouraged residents and business owners to adopt BMPs restricting the application of fertilizers, weed and pest control products and to irrigate in a manner that reduced runoff of nutrients and toxics within the watershed (Supplement Table B-1).

Despite these efforts, the impaired status of Roberts Bay led the South Florida Water Management District to place the estuary on the Surface Water Improvement and Management (SWIM) priority list for restoration in 1995 and, in 1998, continued concerns about high chlorophyll a levels resulted in the placement of the bay on the Section 303(d) list of impaired water bodies under the U.S. Clean Water Act (Petrus and Lassiter, 2005). The waterbody remains on the impaired waters list until the state develops a TMDL. To address the water quality concerns, the Celery Fields Regional Stormwater Facility was constructed from 1998–2007 and mimics the nutrient uptake and stormwater buffering capacities of natural wetlands (Sarasota County, 2013). By 2005, TMDLs for nutrients and chlorophyll a were established for Roberts Bay (Petrus and Lassiter, 2005), but a coalition of environmental groups represented by EarthJustice sued the EPA in 2008 to expedite the creation of Numeric Nutrient Criteria (NNC) in Florida systems (Endres and Walker, 2015; Janicki Environmental, 2010). Numeric Nutrient Criteria are water quality criteria that are established by States, tribes, and other jurisdictions to protect designated uses from nutrient over enrichment (EPA, 2001). NNC often include causal (nitrogen and phosphorous) and response variables (e.g., chlorophyll a, macroalgae, seagrass, water clarity, and dissolved oxygen). There are various approaches for developing NNC ranging from reference condition approach to stressor response.

As a result, NNCs were adopted for TP, TN and chlorophyll a (EPA, 2016a; Florida Administrative Code, 2013). In addition, long-term concerns regarding groundwater contamination by septic systems (Magrin, 1984) led to a septic to sewer conversion project that began in 2009 and was completed October 2017 (Sarasota County, 2018).

3.1.4. Status of Roberts Bay

Efforts to reduce nutrient pollution and improve habitat quality in Roberts Bay have successfully lowered chlorophyll a concentrations below the 11 μg L⁻¹ water quality threshold under Florida’s impaired waters rule (Florida Administrative Code, 2013; Janicki Environmental, 2010). This reduction showed that Harmful Algal Blooms (HABs) – excess growth of micro and macroalgae that produce toxic or harmful effects to aquatic communities (Valiela et al., 1997) – no longer posed a threat to the system. Furthermore, seagrass cover in 2012 was just 12% below the 139.5 ha (348 acre) target (Janicki et al., 2008). These improvements resulted in the removal of Roberts Bay from Florida’s list of impaired waters in 2010 (303(d) list; EPA, 2012).

Upgrading wastewater facilities to tertiary treatment to remove inorganic nitrates and phosphates in the 1980s and 1990s may have resulted in early reductions of nutrients (Supplement Table B-2). However, improved treatment of wastewater is unlikely to have been the sole cause of water quality improvements since the greatest reductions in wastewater nutrients occurs when systems upgrade from primary to secondary treatment of wastewater (Van Drecht et al., 2009). For Roberts Bay, those upgrades were done in the 1970s (Supplement Table B-2). Therefore, the success is likely attributable to two decades of implementing at least nine types of BMPs targeting the dominant sources of nutrient loading (stormwater and surface water) (Figure 3a). Sarasota County should be able to reduce nutrient loads to ~6% of the remaining sources entering the system by applying additional BMPs to reduce nutrient loading in storm and surface water and due to the completed septic to sewer conversion project (Figure 3a).

Figure 3:

Bar plot showing the relationship between nitrogen NPS load source type and the percent of nutrient NPS load (black bars), number of BMPs (dark gray bars), and years since the first implementation of BMPs (light gray bars) for a. Roberts Bay (Florida), b. Newport Bay (Califonia), and c. Peconic Estuary (New York). Point sources were not included in this diagram.

3.2. Newport Bay, California

3.2.1. Overview

Newport Bay is a small (5.4 km²), shallow (average depth < 1 m), stratified (Nezlin et al., 2009) embayment in Orange County, southern California with a fairly small watershed (393 km²). The flushing time of the bay ranges from 0–30 days (Pednekar et al., 2005) and the primary freshwater source is San Diego Creek (Figure 1a–2), which deposits fine sediment from the watershed into the bay (Christensen et al., 1978).

From the 1860s-1980s, land use was dominated by ranching and agriculture (French et al., 2006; McKee, 1985 as cited in: Meixner et al., 2004). From 1968–2003, 95% of the land shifted from agriculture to commercial, industrial and residential uses (Pednekar et al., 2005). By 2013, more than 70% of the watershed consisted of high or medium intensity development and just 1.1% of the land was used for crops. Concurrently, human population of Orange County increased 77% from 1976–2014 (Figure 1b, Table 2A).

Table 2:

Newport Bay, California. Trend-Free Pre-whitened Mann-Kendall trend analyses on annual data (τ) for census data in Orange County (A), water column nutrient concentrations (B-E), macroalgal density (F), seagrass density (G) and cover (H). Values in bold indicate significant trends (p < 0.05).

Factor	Years	n	τ	p value
A. Human population	1976–2014	39	0.997	<0.0001
B. TKN	1976–2014	37	−0.628	<0.0001
C. NH4	1976–2014	37	−0.628	<0.0001
D. NOX	2000–2014	15	−0.297	0.15
E. TPO4	1976–2014	33	−0.415	<0.0001
F. Macroalgal density	1996–2014	13	−0.636	0.005
G. Seagrass (Z. marina) density	2004–2013	4	−0.333	1
H. Seagrass (Z. marina) cover	2003–2013	4	0.333	1

Early reports linked the poor water quality of Newport Bay to runoff from the watershed (Orange County, 1981), which has caused shellfish harvesting closures in Newport Bay since 1978 and restrictions of aquatic recreation due to high concentrations of fecal coliform (Ruffolo, 1999; Santa Ana Regional Water Quality Control Board, 1999). In 1997, the nonprofit group “Defend the Bay” sued EPA for failing to establish water quality regulations that would prevent additional nutrient discharges into the already eutrophic bay (Causin, B. pers. comm.). In 1998, TMDLs for nutrients and sediment were completed (EPA, 1998) as an effort to improve habitat and water quality.

Vogl (1966) documented ephemeral ulvoid macroalgae covering ~50–75% intertidal flats in Newport Bay as early as 1964. However, it was after a massive fish kill triggered by anoxia from decomposing Ulva spp. biomass in 1986 that citizens and regulatory agencies began to recognize the contribution of watershed-based nutrients to declining ecosystem condition (EPA, 1998; Lindgren, 1986). Despite this concern, Kamer et al. (2001) documented that macroalgae blooms continued in Newport Bay and some areas had 100% cover of U. expansa and U. intestinalis as late as 1997.

Nonpoint source nutrients have been the primary source of nutrients to Newport Bay for over 50 years. The Michelson Water Recycling Plant for the Irvine Ranch Water District, which had been using tertiary treatment and diverting discharge to the San Joaquin Marsh since 1967 (IRWD, 2015), accounted for 0% of nutrient loading into Newport Bay in 1997 (Santa Ana Regional Water Quality Control Board, 1998a). Three plant nurseries in the watershed (Hines, Bordiers, El Modeno) were identified as key NPS sources of nutrients (Tetra Tech, 1998) into Newport Bay and issued Waste Discharge Requirement permits. However, nurseries and irrigated agriculture are exempt from the National Pollution Discharge System (Craig and Roberts, 2015; Santa Ana Regional Water Quality Control Board, 1998a). In 1998, total agricultural surface water discharges accounted for 30% of the Newport Bay nutrient load whereas urban runoff, groundwater and atmospheric deposition accounted for 25% and 44% of nutrient loading, respectively (Santa Ana Regional Water Quality Control Board, 1998a). Later estimates by French et al. (2006) showed that by 2001, 63% of nitrate from San Diego Creek was derived from urban runoff (such as stormwater). Though the land use shifted away from agriculture, legacy nutrient inputs from decades of crop fertilization still enter the bay via groundwater (French et al., 2006). In addition to water column nutrient availability, nutrient flux out of sediment pore waters serve as an internal “slow release” fertilizer (Sutula et al., 2006) that may have contributed to the elevated nitrogen and phosphorus concentrations, which continued to support bloom forming macroalgae (Kamer et al., 2004) even when external loads decreased.

Estimates of atmospheric nitrogen deposition suggested that it could account for approximately one fifth of the nitrogen loading to the system. Meixner et al. (2004) quantified the dry deposition of nitrogen in San Joaquin Salt Marsh above Newport Bay to be 1.6–12 kg N ha⁻¹ yr⁻¹. Similarly, Fenn et al. (2010) estimated that total atmospheric nitrogen deposition (wet and dry) for the South Coast Air Basin (which includes Orange County) ranged from 11 to >25 kg N ha⁻¹ yr⁻¹. Based on the estimated size of Newport Bay, this rate suggests that at its maximum, atmospheric deposition could represent 21% of the 2012 TN target load (Santa Ana Regional Water Quality Control Board, 1998a).

3.2.2. Data analyses

3.2.2.1. Water column nutrients

Total Kjeldahl Nitrogen (TKN) concentrations were 61% lower in 2014 than in 1976 (p < 0.05, Table 2B, Figure 4a). Similarly, average annual ammonium (NH₄) and Total Phosphate (TPO₄) concentrations were 67% (p < 0.0001, Table 2C, Figure 4a) and 33% (p<0.05, Table 2E, Figure 4a) lower, respectively, in 2014 than in 1976 and resulted in a significant decline in nutrient concentrations over time. In contrast, trend analysis for NO_X concentration was not significant (Table 2D, Figure 4a).

Figure 4:

Mean water column nutrients, macroalgae biomass and seagrass density and cover for Newport Bay (California) over time. a. Total Kjeldahl Nitrogen (TKN), ammonium (NH₄), nitrogen oxide (NO_X) and total phosphate (TPO₄) concentrations from 1976 to 2014. b. Total biomass of bloom forming ulvoid macroalgae from 1996–2014. Horizontal line indicates minimum known threshold for negative impacts on benthic habitat quality indicators (Green et al. 2014). c. Seagrass (Zostera marina) short shoot density. d. Total seagrass (Z. marina) cover in hectares. Where present, error bars represent standard error. Chlorophyll a concentrations were not available for this estuary.

3.2.2.2. Primary producers

We found that Ulva spp. biomass was greatest in 1996 (1.83 kg ww m⁻² ± 0.43) and declined to 0 kg ww m⁻² by 2014 resulting in a significant negative trend (p=0.005, Table 2F, Figure 5b).

Figure 5:

Mean water column nutrients, chlorophyll a concentrations and seagrass density and cover for Peconic Estuary (New York) over time. a. Total Kjeldahl Nitrogen (TKN), ammonia (NH₃), Total Nitrogen (TN), Total Phosphate (TPO₄) and Nitrogen Oxide (NO_X) concentrations from 1976–2014. b. Chlorophyll a concentration from 1987 to 2014. c. Seagrass (Zostera marina) short shoot density. d. Total seagrass (Z. marina) cover in hectares. Where present, error bars represent standard error.

Eelgrass (Zostera marina) shoot density (number of short shoots m⁻²) declined 49% from 2004 to 2013 (Figure 5c) though no significant trend was detected (Table 2G). Although the total cover of Z. marina increased 12% from 2003–2013 (Figure 5d), this trend was also not significant (Table 2H).

3.2.3. Best Management Practices and regulation of nutrient loads

Concerns about the declining habitat and water quality of Newport Bay prompted the Santa Ana Regional Water Quality Control Board to take action by targeting agricultural fertilizer and sediment runoff (Boyle Engineering Corp., 1983; Orange County, 1981). Early efforts included encouraging farmers to adopt a number of erosion control practices such as: switching from overhead spraying to drip irrigation, planting native cover crop species and applying contour tilling to reduce runoff of sediment and nutrients to the bay (Supplement Table B-1).

Despite these early interventions, nutrients, fecal coliform and macroalgal biomass remained high and Newport Bay was added to the 303(d) list of impaired waters in 1996 (EPA, 1998). Moreover, in 1997, “Defend the Bay” sued the EPA for failing to establish TMDLs that are required when a water body is placed on the 303(d) list (Ruffolo, 1999). As such, in 1998 TMDLs were adopted to reduce sediment to 56,699 tonnes yr⁻¹ by 2009, nitrogen to 65,482 kg season⁻¹ and phosphorus to 28,159 kg season⁻¹ (Santa Ana Regional Water Quality Control Board, 1998a, b).

To attain these load reductions, additional BMPs were implemented to address nutrient pollution from agriculture, urban runoff and groundwater (Supplement Table B-1). A key strategy in reducing nutrient loads was the construction of 17 artificial wetlands throughout the watershed, which trap sediments and filter nutrients, beginning with the San Joaquin Marsh in 1997 (RBF Consulting, 2013). Many of the BMPs that targeted the reduction of fertilizer runoff, such as the use of drip irrigation and water recycling similarly reduced sediment loading into the bay (Supplement Table B-1). Recycling of irrigation water alone resulted in a 98% reduction of nutrient discharges from the three largest nurseries (Shibberu, 2010). Furthermore, the subsequent closure of all the aforementioned nurseries (Hines, Bordiers, El Modeno) likely contributed to the decline of watershed nutrient loads.

As the primary source of nutrient loading shifted from agriculture to urban runoff (French et al., 2006), Orange County implemented stormwater controls for urban areas (Supplement Table B-1). For example, the Peters Canyon Channel Water Capture and Reuse Pipeline project, completed in 2016, diverts runoff from Peters Canyon Wash to the Orange County Sanitation District thus reducing surface water nitrate loads into the bay (ESA, 2015). In addition to diverting stormwater runoff, the new pipeline also diverts groundwater-supported surface flows (Supplement Table B-1).

Though groundwater may contribute ~20% to the nutrient loading to the bay, due to the inherent difficulty of addressing nutrient pollution from groundwater, fewer groundwater BMPs have been implemented compared to agricultural and urban runoff BMPs (Supplement Table B-1). However, both surface and groundwater nitrogen loads were reduced by efforts such as the water recycling systems implemented by nurseries (Supplement Table B-1). Moreover, efforts to reduce nitrogen in groundwater benefitted from a TMDL requirement to reduce selenium in groundwater (RBF Consulting, 2013). For example, groundwater is pumped out and diverted for treatment (dewatering) during new construction within the watershed (RBF Consulting, 2013). This process removes both nitrogen and selenium from the groundwater (RBF Consulting, 2013).

3.2.4. Status of Newport Bay

Contrary to prior connections between nutrient pollution and increased human populations (Vitousek et al., 1997), habitat and water quality of Newport Bay have improved even as human population increased 77% over 39 years (Figure 1b). Total nitrogen loads to Newport Bay declined by approximately two orders of magnitude since 1978 (Schiff et al., 2014). In 2007, winter TN loads had been reduced to 30% lower than the 2012 target (Orange County, 2007). Furthermore, by 2008–2009, the suspended sediment load was 33% lower than the TMDL target (Orange County, 2010). Likely as a result of these nutrient reductions, the biomass of ephemeral macroalgae (Figure 5b) has been below the minimum threshold (~1 kg ww m⁻²) known to exert negative effects on benthic habitat quality in Newport Bay (Green et al., 2014) since 2002, with the exception of a bloom event in 2005 (Figure 5b). Furthermore, water quality and light availability have improved sufficiently to facilitate implementation of eelgrass (Z. marina) restoration pilot projects (Orange County Coastkeeper, 2015). Additionally, fecal coliform water quality objectives for water contact recreation have been attained for 84% of sites, though many sites still violated the 14 Mean Probable Number (MPN) per 100 ml criteria for human consumption of shellfish (California Water Resources Control Board, 2015; Santa Ana Regional Water Quality Control Board, 1999).

These improvements are not the result of point source reductions since point source discharges into the system were eliminated by at least 1961 (Supplement Table B-1). One reason for the improved water and habitat quality of Newport Bay could be the early implementation of BMPs that reduced nutrient loading from groundwater, surface water and stormwater (Figure 3b). Nutrient pollution of surface waters accounted for 30% of the nutrient loading in 1997 (Santa Ana Regional Water Quality Control Board, 1998a) and BMPs targeting surface water and sediment runoff began in the early 1980’s (Figure 3b, Supplement Table B-1). As urbanization replaced agriculture throughout the watershed, efforts to divert stormwater runoff (25% of the nutrient load) also began as early as 1983 (Supplement Table B-1). Moreover, despite the challenges associated with removing nutrients from groundwater, some diversion and treatment practices have been in place within the watershed since 1998 (Figure 3b, Supplement Table B-1).

Another factor contributing to the compliance with the nutrient TMDL may have been the chronic “exceptional drought” in California (Diffenbaugh et al., 2015) that reduced rainfall across the state to historic lows (Griffin and Anchukaitis, 2014). It is possible that the prolonged reduction in surface flow resulted in lower nutrient loadings than would occur during typical rainfall years. Around the world, reductions in nutrient loading have been linked to drought conditions (Møhlenberg, 1999). Conversely, some researchers documented that low flow during drought conditions resulted in increased rather than decreased nutrient concentrations (Alpine and Cloern, 1992; Beklioglu and Tan, 2008), which we did not find for Newport Bay.

Though the nitrogen TMDLs were achieved in 2001, the system has not been removed from the 303(d) impaired waters list (California Water Quality Control Board, 2016). Santa Ana region 303(d) listed sites were scheduled for review in 2016, however any update to the listing status of Newport Bay has not been announced as of August 2021 (CA Waterboards, 2020).

3.3. Peconic Estuary, New York

3.3.1. Overview

Peconic Estuary is a relatively large (498 km²) well-mixed coastal ecosystem located in Suffolk County on the eastern end of Long Island, New York (Figure 1a–3). With more than 100 embayments (Peconic Estuary Program, 2007), it has the largest watershed (709 km²) of our three study sites. The western basin has a flushing time of approximately two months (Suffolk County Department of Health Services, 1992), whereas the eastern basin flushes in just a few days. Additionally, the sediments within the watershed are composed of permeable sand and gravel (e.g. glacial till) resulting in shallow groundwater aquifers that are highly susceptible to nutrient loading from surface water infiltration (Eckhardt et al., 1989).

Most of the nutrient loading to the Peconic Estuary watershed can be attributed to a long history of agriculture and duck farming. Analysis of sediment cores from nearby Moriches Bay, Long Island suggested that agricultural influence on the system dates back to the early 1700’s (Clark, 1986). However, fertilizer inputs to the watershed have declined over time as the total acreage of farms declined by half from 1962–2007 and as farmers switched from potatoes to vineyards, which require less fertilizer (Cameron Engineering & Associates, 2012). By the 1980’s agricultural land comprised 10–12% of the watershed (Eckhardt et al., 1989; Figure 3c) though by 2012 farmland was approximately 6% of Suffolk County (Suffolk County, 2015).

In 1873, Pekin duck farming was introduced to the local economy (Suffolk County, 2009) and by 1962 there were 14 duck farms in the Peconic Estuary watershed (Cameron Engineering & Associates, 2012; Suffolk County, 2009). Though only one of duck farms originally present in the watershed remains open (Cameron Engineering & Associates, 2012; Hardy, 1976; Suffolk County, 2009), it is likely that there are legacy nutrient loads from duck waste and agricultural fertilizer (Eckhardt et al., 1989; Suffolk County Department of Health Services, 1992) given that groundwater reaching the estuary can have an estimated age of 15–120 years (Schubert, 1999).

The Peconic Estuary watershed had the highest population but lowest development of the three case study sites (~16% watershed area). Suffolk County population increased just 22% between 1976–2014 (Figure 1b, Table 3A). The long history of agriculture, duck farming and an increase in the number of septic systems has resulted in high nutrient loading (41%) to the groundwater of the Peconic Estuary watershed (Cameron Engineering & Associates, 2012; Lloyd, 2014; Peconic Estuary Program, 2007). However, the majority of nutrient loading (56%) is likely derived from atmospheric deposition (Peconic Estuary Program, 2007) resulting from fossil fuel combustion, volatile ammonia from fertilizers and manure in the surrounding area (Luo et al., 2002). However, Lloyd (2014) estimated the contribution of atmospheric deposition to nitrogen loading into Peconic Estuary to be closer to 24% and further noted that this value only included deposition to land surfaces and throughput to aquifers within the watershed. If direct deposition to surface waters were included, atmospheric deposition could account for nearly 84% of the nitrogen load to the system (Lloyd 2014). By one estimate, surface water, treated wastewater and stormwater comprised just 1% of the load each (Peconic Estuary Program, 2007). However, these loading estimates are somewhat uncertain as evidenced by Lloyd (2014) who calculated that agriculture and treated wastewater contributed 17% and 7%, respectively, to the nitrogen load to the system.

Table 3:

Peconic Estuary, New York. Trend-Free Pre-whitened Mann-Kendall trend analyses on annual data (τ) for census data in Suffolk County (A), water column nutrient (B-F) and chlorophyll a (G) concentrations, and seagrass density (H) and cover (I). Values in bold indicate significant trends (p < 0.05).

Factor	Years	n	τ	p value
A. Human population	1976–2014	39	0.986	<0.0001
B. TPO4	1976–2000	24	−0.526	0.0004
C. NH3	1976–2014	38	−0.216	0.006
D. NOX	1988–2014	27	0.372	0.0008
E. TKN	1976–2008	25	−0.246	0.096
F. TN	2000–2014	15	−0.037	0.753
G. Chlorophyll α	1976–2014	27	−0.516	0.0004
H. Seagrass (Z. marina) density	1997–2013	16	−0.989	<0.0001
I. Seagrass (Z. marina) cover	2000–2013	6	−0.154	0.495

A long history of nutrient pollution resulted in the decline of habitat and water quality of Peconic Estuary over time. Segments of the system were identified as eutrophic in the early 1970’s (Hardy, 1976). However, it was not until the 1985–1995 brown tide HABs (Aureococcus anophagefferens) (LaRoche et al., 1997), which decimated bay scallop (Argopecten irradians) populations (Gobler et al., 2005), that plans to reduce NPS nutrients were proposed (Peconic Estuary Program, 1995; Suffolk County Department of Health Services, 1992). Strong evidence from empirical research showed that A. anophagefferens increases in response to elevated dissolved organic nitrogen (DON) and low light availability (Gobler et al., 2005). The early brown tide may have been triggered by the collapse of phytoplankton blooms stimulated by a peak in groundwater nitrogen reaching the estuary (LaRoche et al., 1997). Though sustained brown tide has not occurred in the estuary since 1995, the frequency of highly toxic red tide HABs (Cochlodinium polykrikoides) increased over the last ten years contributing to shellfish and finfish mortality (Gobler et al., 2008). Additionally, a mahogany tide (Prorocentrum sp.) in 2015 was suspected of being responsible for reducing dissolved oxygen concentrations and triggering a massive die off of Atlantic menhaden (~300,000 Brevoortia tyrannus) (Civiletti and Blasl, 2015; Semple, 2015).

3.3.2. Data analyses

3.3.2.1. Water column nutrients

Nutrient concentrations primarily declined in Peconic Estuary from 1976–2014 (Figure 5a). Total Phosphate (TPO₄) concentrations declined significantly (p=0.0004, Table 3B) and were 75% lower on average in 2014 compared to 1976 (Figure 5a). Ammonia (NH₃) decreased during the same period and were 44% lower, respectively, in 2014 than 1976 (p=0.006, Table 3C, Figure 5a. NO_X concentrations declined 35% from 1988 to 2014 and this trend was significant (Figure 5a, Table 3D). However, changes in TKN and TN concentrations were not significant (Table 3E, F).

3.3.2.2. Primary producers

Chlorophyll a levels varied widely between 1976–2014, with average annual concentrations ranging from 2.3–26.6 μg l⁻¹ (Figure 5b). The 88% decline from 1976 to 2014 was significant (p=0.0004, Table 3G).

Average density of eelgrass (Z. marina) declined significantly from 1997–2013. Peak mean density (551 ± 59 short shoots m⁻²) occurred in 1999 but declined 90% by 2007 and remained at < 75 short shoots m⁻² (Figure 5c, Table 3H). There was no significant trend in Z. marina cover from 2000–2013 (Figure 5d, Table 5I).

3.3.3. Best Management Practices and regulation of nutrient loads

Early efforts to address eutrophication of Peconic Estuary began in 1968, when the New York State Department of Health required mandatory treatment of duck waste (Garvey, 1973; U.S. DOI, 1966). After 1968, few regulatory actions were taken to target nutrient pollution until after the 1985 brown tide event and subsequent collapse of the scallop fishery in 1986 (Supplement C Figure 1. Figure 1). By 1988, the Brown Tide Comprehensive Assessment and Management Program was developed and Peconic Estuary was included in the National Estuary Program (Kennish, 2000; Minei, 1989; Suffolk County Department of Health Services, 1992). Later, the Peconic Estuary Program Comprehensive Conservation and Management Plan detailed water quality targets including a 0.45 mg l⁻¹ for summer (July-Sept) TN and made recommendations for changes in infrastructure, tax credits, land use, fertilizer applications and BMPs that could be applied to achieve the targets (Peconic Estuary Program, 2001). As early as 1994, the cost of BMP implementation was offset through the Agricultural Nonpoint Source Abatement and Control Grant Program (New York State, 2016).

Despite these actions, continued poor water quality of the estuary resulted in the establishment of TMDLs for total nitrogen and dissolved oxygen (Peconic Estuary Program, 2007). To achieve the dissolved oxygen standards within the established 15 year timeframe, limits on TN loading were proposed for winter (180 kg day⁻¹ on average) and summer (140 kg day⁻¹ on average) (Peconic Estuary Program, 2007). If attained, these limits would represent 31% and 47% reductions in TN loading, respectively (Peconic Estuary Program, 2007). However, EPA (2013) reported that the difficulty of managing nutrient NPS meant that the pace of BMP implementation would hinder achievement of required load reductions by 2022.

Implementation of BMPs began in 2004 (Supplement Table B1) and largely focused on nutrient loading from fertilizer and stormwater runoff. For example, the Suffolk County Fertilizer Law restricts fertilizer application from November-April (EPA, 2013) and the New York State Dishwasher Detergent and Nutrient Runoff Law in 2010 (EPA, 2013) prohibits the application of fertilizers on impervious surfaces or within 6.1 meters of surface water. In addition, a suite of BMPs were implemented to reduce fertilizer runoff from golf courses and agricultural land (Supplement Table B-1). However, documentation of BMP implementation targeting fertilizer reduction has been sparse (Supplement Table B-1; A. Branco, pers. comm.). In a few locations throughout the watershed, efforts to reduce nutrient loading from stormwater can be attributed to the installation of Vortechs® systems (devices that trap sediment that bind nutrients), stormwater ponds and leeching basins that trap sediments and remove nutrients before the water enters the estuary (Supplement Table B-1). Lastly, there are greater efforts to promote low impact development in the watershed through efforts such as the utilization of permeable pavement (Supplement Table B-1).

3.3.4. Status of Peconic Estuary

Declines in TKN and NH₃ observed for Peconic Estuary between 1976–1986 (Figure 5a) likely occurred as a result of regulations requiring secondary treatment for all duck and municipal waste discharges by 1968 (Garvey, 1973) and from upgrades to secondary treatment by the mid 1970s (Supplement Table B-2). By 2009, all but two duck farms closed (Suffolk County, 2009), which further reduced nutrient input. We found that the 0.45 mg l⁻¹ summer TN target (Peconic Estuary Program, 2001) had been achieved by 2000 (Figure 5a).

Peconic Estuary may not attain the target load reductions (EPA, 2013) because although agricultural and duck waste loads have been reduced, the number of septic systems has increased over the same time period. It is likely that reductions in TKN and NH₃ have been insufficient since septic systems often contribute NO₃ to groundwater (Lu et al., 2008). Septic systems and cesspools may contribute much as 43% of the nitrogen loading to the system (Lloyd, 2014). Nitrogen pollution of groundwater in Suffolk County was evidenced by a 2006 survey, which found that 10% of private wells and 1% of community wells violated the 10 mg l⁻¹ Maximum Contaminant Level (MCL) for nitrate in drinking water and as many as 17% of wells in Suffolk County were considered “degraded” with nitrate concentrations exceeding 6 mg l⁻¹ (Suffolk County, 2007).

The difficulties associated with addressing nutrient loading from atmospheric deposition and groundwater are potentially linked to the continued habitat and water quality deterioration of Peconic Estuary. Though brown tide (A. anophagefferens) has not be prevalent since 1995 (Tettelbach et al., 2013), bay scallops (A. irradians) and other bivalves starved to death in 1985 and 1995 due to the poor nutritional quality of the blooms (Gobler et al., 2005; Gobler and Sunda, 2006) and have not recovered (Supplement A-3). Moreover, shading caused by A. anophagefferens blooms triggered a seagrass die-off within the system (Gobler and Sunda, 2006; Pickerell et al., 2005). The 87% loss of Z. marina density in just 16 years (1997–2013), despite efforts to reestablish seagrass populations (Pickerell et al., 2005; Stephenson, 2009), suggests that water quality has not improved sufficiently to promote seagrass restoration (Orth et al., 2006). More recently, harmful algae such as C. polykrikoides have bloomed annually since 2002 (Gobler et al., 2008) and chlorophyll a concentrations have increased significantly since 1986 (Figure 5b). The shift from brown to red tide may dampen efforts to restore scallop populations (Supplement A-3) within the estuary due to its toxic effects on scallop larvae (Tang and Gobler, 2009).

4. Discussion

We are the first to show how counties and water control boards have effectively managed nutrient NPS even as watershed human populations increased. One of the central paradigms in eutrophication research is that water quality inevitably declines in response to human population growth within the watershed (Howarth et al., 2002; Lotze et al., 2006; Vitousek et al., 1997). It was predicted that global NPS nitrogen and phosphorus pollution would continue to increase as the human population grows until 2050 (Van Drecht et al., 2009). Contrary to these predictions, we found that human populations in Sarasota and Orange Counties increased significantly over 38 years, yet both Roberts Bay and Newport Bay had significant improvements in habitat and water quality. Moreover, Peconic Estuary, which had the lowest population increase, is not on track to meet water quality targets. This shows that reductions in nutrient loading from stormwater, fertilizer and groundwater in Roberts Bay and Newport Bay exceeded the increased nutrient loading caused by population growth. Reductions in point source and NPS discharges to Tampa Bay similarly improved water and habitat quality despite increased human populations in the watershed (Greening and Janicki, 2006; Greening et al., 2014). Therefore, our work builds on evidence demonstrating that sufficient NPS nutrient management can compensate for human population growth and even reverse eutrophic conditions in coastal ecosystems.

We found that the successes in Roberts Bay and Newport Bay were due to the implementation of BMPs that reduced nutrient pollution in surface waters. Worldwide, nutrient NPS reduction efforts focus on surface waters (Table 4) because they can be managed at the watershed scale (Sharpley et al., 1994; Tilley and Brown, 1998). Though the majority of efforts in the Peconic Estuary watershed focused on surface water nutrient sources, surface waters comprised just 2% of the total load. The lack of groundwater management may explain why Peconic Estuary has not achieved the target reductions specified in the TMDL (EPA, 2013). Additionally, the residence times for nutrients flowing into estuaries via surface waters are on the order of hours to days (Greening and Janicki, 2006; Malone et al., 1988) compared to groundwater that may have decade-long travel times before reaching the receiving waterbody (Meals et al., 2010; Schubert, 1999). This may enhance lag times associated with the management and treatment of groundwater derived nutrients and delay restoration of habitat and water quality (Meals et al., 2010). For example, the effects of the Suffolk County fertilizer reduction law that became effective in 2009, may not be evident in Peconic Estuary until at least 2024 (assuming a 15 year groundwater residence time) (Schubert, 1999). Our work suggests that systems dominated by nutrient pollution from surface water and stormwater may be able to achieve more rapid improvements in eutrophic condition compared to systems with high groundwater loads and long groundwater residence times.

Table 4:

Review table of Nonpoint Source management practices for coastal estuaries and lagoons

Site	Evidence of eutrophication	Source of NPS nutrients	NPS BMPs	Targeted loading source
Chesapeake Bay, US	Phytoplankton blooms, macroalgal blooms, loss of seagrass	Agriculture	Fertilizer reductions, reduced spread of agricultural land, investigation into reducing airborne sources of nutrients	Surface water, groundwater, atmospheric deposition
Chesapeake Bay, US	Phytoplankton blooms, macroalgal blooms, loss of seagrass	Agriculture, livestock	Animal waste management, conservation tillage, cover crops, grassed buffers, manure transport, land retirements	Surface water
Peel-Harvey estuary, Australia	Macroalgae, phytoplankton	Cattle, sheep, agriculture	Banned clearing of native vegetation, changed timing of fertilization, changed fertilizer formulation to less water soluble	Surface water
Tuggerah Lakes, Australia	Phytoplankton blooms, macroalgal blooms, loss of seagrass	Dairy farms, citrus orchards	Reduced sediment and nutrient runoff, protected wetlands, trapped sediment	Surface water
Various sites in Denmark	Phytoplankton blooms	Agriculture, livestock	Eliminated direct discharge from farms, standardized nitrogen fertilizer values, reused of manure for fertilizer	Surface water
Danish Straits, Denmark	Nutrients, chlorophyll a, macrophytes, benthic macrofauna	Agriculture	Fertilizer reduction, buffer strips, restoration of wetlands	Surface water
Vistonis Lagoon, Greece	Nutrients, chlorophyll a, phytoplankton community composition	Sediment erosion, lifestock, agriculture	Reduced sediment runoff, protected wetlands, reforestation to trap sediment and reduce nutrients, banned grazing near lagoon, economic shift to different crops reduced fertilizer demand	Surface water
Zandvlei estuary, South Africa	Water weeds, Nutrients	Agriculture	Protected wetlands, diverted and treat water	Surface water

Best management practices such as changes in fertilizer application, water conservation measures (Yang et al., 2020) and NOx emission regulations (Degraeuwe et al., 2017; Haugen et al., 2018) should reduce nitrogen before it enters the groundwater. It will be worthwhile to determine whether the restricted land, air and sea traffic caused by COVID-19 quarantines result in reduced NOx emissions (Goldberg et al., 2020; Guevara et al., 2020; He et al., 2020) result in short term nitrogen reductions in surface waters and ultimately groundwater. If so, this would support NOx emissions reductions as a method to reduce groundwater nitrogen pollution. Thus, efforts to reduce nitrogen loading from atmospheric deposition will depend upon both international commitments to reduce NO_X emissions (Boersma et al., 2015) and tighter enforcement at the national and state levels (EPA, 2015). Despite growing concerns, few programs specifically examined the reduction of atmospheric nutrient sources (Williamson et al., 2017). However, since agriculture and livestock waste that contaminate surface and groundwater also contribute to atmospheric nitrogen loads (Hutchings et al., 2014; Luo et al., 2002), efforts to reduce nutrient loads from one source will likely reduce loading from other sources.

A discussion of water quality criteria at these sites would be incomplete without the referencing the role of litigation by environmental organizations. For example, the organization “Defend the Bay” brought suit against the EPA due to the failure to enact the TMDLs for Newport Bay (Ruffolo, 1999). In 1997, EPA settled the lawsuit and established a legally binding agreement and schedule by which the Santa Ana Regional Water Quality Control Board would establish TMDLs (EPA, 1997). Per the agreement, failure by the Board would result in EPA implementing its own water quality standards (EPA, 1997). Further involvement by EPA and Defend the Bay was not necessary. Similarly, in 2008 EarthJustice sued EPA on behalf of environmental groups including the Sierra Club (Pittman, 2012). In 2009, EarthJustice entered a consent decree with EPA to implement water quality criteria (Endres and Walker, 2015). The state of Florida also sued EPA, stating that the federal agency lacked the authority to implement NNCs (Flowers and Charles, 2012). In the end, EPA approved of Florida Department of Environmental Protection’s NNC (Flowers and Charles, 2012). To our knowledge, no such lawsuits have been filed over water quality for the Peconic Estuary. The merit and effectiveness of lawsuits in promoting water quality criteria are beyond the scope of this paper, but they have been one tool used by citizen groups.

Our work demonstrated the utility of existing, publicly available datasets to evaluate the effectiveness of NPS management in long term water quality trends of coastal ecosystems. However, such datasets may have inconsistencies or the user may be unaware of nuances in the original study design which can affect the interpretation of the results (Cheng and Phillips, 2014). One limitation of our use of the data was that it was correlative rather than causative. Designs such as Before Control Impact (BACI) enable researchers to identify specific causes related to habitat change. Roni et al. (2018) found that using Multiple Before After Control Impact (mBACI) and extensive post-treatment (EPT) (studies that evaluate changes in an environment before and after human disturbances or restoration treatments) were more effective at evaluating success of river restoration than case studies alone. Additionally, since restoration typically involves concurrent changes in several factors (e.g. BMPs, hydrology, land use) information theoretic modeling (Burnham and Anderson, 1998) may help identify the most important drivers of changes in water and habitat quality. Secondary data, subject to proper quality assurance and quality control measures, are useful tool to conduct studies to identify pathways of NPS management and restoration of coastal ecosystems.

5. Conclusions

Human-induced eutrophication threatens the ecological and economic integrity of coastal waters, highlighting the need to understand and address NPS nutrients. Our study showed that habitat and water quality improved in two estuaries in as little as 15 years when multiple BMPs were implemented that targeted the greatest sources of NPS nutrient loading. These improvements occurred despite increasing human population within the surrounding counties because nutrient management strategies were adapted to accommodate changes in dominant nonpoint sources entering the estuaries. Overall, our findings indicate that scientifically based adaptive and comprehensive nutrient management, which includes both regulatory actions and community-based initiatives (Gross and Hagy, 2017), should be included in successful frameworks for control of nonpoint source nutrients.

Highlights.

• Eutrophication can improve even with population increases in the watershed

• NPS are easier to control in smaller watersheds

• Demonstrate that estuarine health can improve quickly (within 15 years)

• Groundwater and atmospheric deposition remain key challenges for NPS management

• NPS management requires cooperation of government (local, state and federal)