total phosphorus (TP), (d) dissolved inorganic phosphorus (DIP), and (e) suspended sediment (SS) to Chesapeake Bay and the northern Adriatic Sea from their largest tributaries (i.e., Susquehanna and Po, respectively) in 1985–2015. Mann‐Kendall (MK) trend slope and significance (p) values are shown in legend.
For the CB watershed, riverine export of TN was dominated by agriculture nonpoint sources (fertilizer and manure 54%), followed by contributions from atmospheric deposition (17%), point sources (16%), and urban sources (12%). Riverine export of TP was dominated by agriculture nonpoint sources (43%), followed by point sources (32%) and urban sources (11%) (Ator et al., 2011). Typically, nutrients accumulate in watersheds during dry periods of low flows and are transported to receiving waters via groundwater discharge and surface water runoff during wet periods and storm events (Shields et al., 2008; Tesi et al., 2013). Groundwater can represent a major fraction of riverine load (especially N). Bachman et al. (1998) estimated that base flow (a proxy of groundwater input) accounted for 17–80% (median 48%) of the TN load at 36 CB monitoring sites. In addition, riverine export can be strongly modulated by reservoirs. For example, Conowingo Reservoir and two others in the lower Susquehanna River historically trapped about 2%, 45%, and 70% of annual N, P, and SS load, respectively (Langland & Hainly, 1997).
Riverine inputs of TN to the NAS were also dominated by agriculture nonpoint sources (40%) followed by point and urban sources (27%), and groundwater (29%). The TP load mainly originated from point sources and urban sources (43%) and agriculture nonpoint sources (36%) (Volf et al., 2013). For the Po watershed, TN load comes from point sources (40%), nonpoint sources (20%), and groundwater and springs (40%), whereas TP loads come from point sources (80%) and nonpoint sources (20%) (Salvetti et al., 2006).
2.4.2. Controlling Factors
In the CB watershed, several controlling factors have been identified for nutrient export in the Susquehanna River (Zhang, Ball, et al., 2016). First, river flow dominates the interannual variability of constituent export. Second, land‐use patterns strongly affect the relative contribution of the subwatersheds. Specifically, long‐term median yields of N, P, and SS all correlate positively with the fraction of the watershed that is not forested. Third, riverine loads from different Susquehanna subwatersheds have generally declined due to reductions in source input over the same period. Finally, dams and reservoirs can strongly modulate the storage and release of particulate constituents and the extent of modulation has varied considerably over time as reservoirs fill. Two decades ago, the Conowingo Reservoir and two others in the lower Susquehanna River trapped about 2%, 45%, and 70% of annual N, P, and SS load, respectively (Langland & Hainly, 1997). Currently, the reservoir system is no longer an effective trap of these constituents (Hirsch, 2012; Langland, 2015; Zhang et al., 2013; Zhang, Hirsch, et al., 2016).
In the NAS watershed, the annual export of nutrient and sediment is directly linked to river flow and can be estimated using Po River loads as a reference (Cozzi & Giani, 2011). However, the short‐term dynamics of riverine loads are complicated during freshets by the effects of flow on the erosion, groundwater inputs, dilution, and biological processes in the river environment (Marchina et al., 2015; Tesi et al., 2013). The N and P loads in the marine environment have long‐term trends consistent with watershed inputs from industrial activities, agriculture, livestock, urban settlements, and atmospheric deposition; dynamics of their delivery are also modulated by the variable retention of these elements in soils and aquifers (Cozzi et al., 2019; Palmeri et al., 2005; Salvetti et al., 2006; Viaroli et al., 2018). In general, retention mechanisms differ considerably between N and P—sedimentation may be the major retention mechanism for particulate P during overland flow conditions (Cirmo & McDonnell, 1997; Hoffmann et al., 2009), whereas sorption/desorption reactions and denitrification are more important for P and N, respectively, during subsurface flow conditions (House, 2003; Withers & Jarvie, 2008).
2.4.3. Watershed Management
For CB, coordinated efforts have been implemented to reduce pollutant inputs since the first Chesapeake Bay Agreement was signed in 1983. Here we provide a brief overview of historical changes in management and practices associated with point, agricultural, and stormwater sources in the Susquehanna River (Zhang et al., 2013). For point sources, the more important historical controls were the P ban in detergents (Litke, 1999) and the adoption of increasingly effective nutrient removal technologies at wastewater treatment plants (Chesapeake Executive Council, 1988). For agriculture nonpoint sources, many strategies have focused on controlling fertilizer and manure applications, including regulations on storage and usage of animal manure and regulations on concentrated animal operations and feeding operations (New York State Department of Environmental Conservation, 2007; Pennsylvania Department of Environmental Protection, 2004; US Department of Agriculture and US Environmental Protection Agency, 1999). For stormwater sources, the USEPA initiated the National Pollutant Discharge Elimination System Phase I regulations in 1990 for Municipalities with Separate Storm Sewer Systems (MS4s) serving populations of 100,000 or more (US Environmental Protection Agency, 2000) and expanded the program in 1999 to include smaller MS4s (US Environmental Protection Agency, 2000). States also took actions to promulgate regulations on stormwater discharges in the late 1990s to early 2000s (New York State Department of Environmental Conservation, 2007; Pennsylvania Department of Environmental Protection, 2004).
For the NAS, efforts were made by Italian regulators to progressively reduce P content in detergents to 1% (6% for automatic laundry detergents) in 1989 (Marchetti et al., 1989; Rinaldi, 2014). In that year, industries also substituted sodium‐triphosphate‐based detergents with zeolite A (Glennie et al., 2002). The control of N loads has been improved recently with the adoption of the European Nitrates Directive (91/676/EEC). However, the reduction of N loads was less effective compared to that of P loads, because N is largely contributed by nonpoint sources (Viaroli et al., 2010, 2018). The creation of designated Authorities for specific hydrographic basins and of Regional Environmental Protection Agencies in 1994 further supported a coordinated monitoring of continental waters in Italy. The main measures adopted to reduce eutrophication from river loads were (a) establishment of spill basins and manure wastes for crop and livestock farming, (b) implementation of good farming practices in vulnerable areas, and (c) implementation of the Urban Waste Water Treatment Directive (91/271/EEC) (Bortone, 2014).
The N and P inputs in the NAS are not limited to riverine inputs. Direct inputs to the NAS occur via groundwater discharge and treatment plants, but are poorly quantified. Groundwater aquifers, particularly in the NE Adriatic region, are very sensitive to external pollution due to their large draining capacity of surface waters and low self‐cleaning potential (The EU.WATER Project, 2010). For treatment plants, several underwater pipelines have been built since the 1980s, especially in the NE Adriatic region. Despite a gradual upgrade of treatment plants from secondary to tertiary treatment, wastewater loads continue to be an important source of nutrients in the NAS (Cozzi et al., 2014; Scroccaro et al., 2010; Sekulić et al., 2004; Volf et al., 2018).
2.5. MAJOR CHALLENGES
2.5.1. Legacy Sources
Many restoration efforts around the world have not yet achieved significant progress in reducing riverine loads of nutrient and sediment due to challenges such as legacy inputs, which accumulate and are stored in groundwater aquifers and sediments. Such effects have been documented for watersheds in North America (e.g., Chesapeake Bay, Mississippi River, and Lake Erie) and Europe (Basu et al., 2010; Jarvie et al., 2013; Sharpley et al., 2013; Van Meter et al., 2017; Van Meter et al., 2017, 2018; Vero et al., 2017). For CB, there is strong evidence for the importance of legacy sources. For example, riverine loads in the Susquehanna remained relatively constant in the past 30 years despite strong reductions in anthropogenic inputs to watersheds (Zhang, Ball, et al., 2016). Such patterns may reflect the effects of legacy sources (Basu et al., 2010; Thompson et al., 2011). Van Meter et al. (2017) reported that N dynamics in the Susquehanna are dominated by groundwater legacies, with 18% of the current annual N input to the river being
