Coupling the dual isotopes of water (δ 2H and δ 18O) and nitrate (δ 15N and δ 18O): a new framework for classifying current and legacy groundwater pollution

Nitrate contamination of groundwater is a concern globally, particularly in agricultural regions where decades of fertilizer nitrogen (N) use has led to a legacy of N accumulation in soils and groundwater. Linkages between current management practices and groundwater nitrate dynamics are often confounded by the legacy effect, and other processes unrelated to management. A coupled analysis of dual stable isotopes of water (δH2O = δ 2H and δ 18O) and nitrate (δNO3 −= δ 15N and δ 18O) can be a powerful approach to identify sources and processes responsible for groundwater pollution. To assess how management practices impact groundwater nitrate, we interpreted behavior of δH2O and δNO3 −, together with nitrate concentrations, in water samples collected from long-term monitoring wells in the Southern Willamette Valley (SWV), Oregon. The source(s) of nitrate and water varied among wells, suggesting that the nitrate concentration patterns were not uniform across the shallow aquifer of the valley. Analyzing the stability versus variability of a well’s corresponding δH2O and δNO3 − values over time revealed the mechanisms controlling nitrate concentrations. Wells with stable δH2O and δNO3 − values and nitrate concentrations were influenced by one water source with a long residence time and one nitrate source. Variable nitrate concentrations of other wells were attributed to dilution with an alternate water source, mixing of two nitrate sources, or variances in the release of legacy N from overlying soils. Denitrification was not an important process influencing well nitrate dynamics. Understanding the drivers of nitrate dynamics and interaction with legacy N is crucial for managing water quality improvement. This case study illustrates when and where such coupled stable isotope approaches might provide key insights to management on groundwater nitrate contamination issues.


Introduction
Chronic inputs of nitrogen (N) for agricultural production over time can lead to accumulation of surplus N in soils and groundwater. This legacy N contamination of nitrate (NO 3 − ) to groundwater systems has far-reaching consequences for human health and the environment, including impacts to drinking water sources or to groundwater-dependent ecosystems, like wetlands, rivers, and coastal areas (Hansen et al 2017). The U.S. Environmental Protection Agency (EPA) established a maximum contaminant level (MCL) for public drinking water of 10 mg NO 3 − -N l −1 primarily to reduce risk of methemoglobinemia in infants (USEPA 1995). Ingestion of water with NO 3 − concentrations at or even below the current MCL can increase risk of cancers, birth defects, and other adverse health effects (Hinsby et al 2012, Ward et al 2018. Furthermore, the leaching of legacy NO 3 − to the groundwater, and its subsequent discharge to surface waters, can cause eutrophication and seasonal hypoxia (Lewis et  Thus, understanding the current and legacy drivers of NO 3 − concentrations in groundwater is critical for water quality management.
Across the US, agricultural activities are the main source of N inputs to landscapes (Ruddy et al 2006, Galloway et al 2008, Sobota et al 2013, Sabo et al 2019. Nitrate concentrations in groundwater are driven by N inputs to the land, physical features impacting the flow rates of water through soils and aquifers, and redox conditions (DeSimone et al 2014). More than 20% of shallow domestic wells in agricultural areas of the US are reported to exceed the MCL , DeSimone et al 2014. In addition, drinking water NO 3 − violations in groundwater used for US public water supplies are largely influenced by cropland area, precipitation, and annual N surplus in the source area (Pennino et al 2020). Such elevated concentrations can persist for decades in groundwater aquifers, especially beneath agricultural lands with a legacy of N applications (Repert et al 2006, Puckett et al 2010, Katz et al 2014. Even if new N inputs cease, the release of diffuse sources of N, coupled with slow natural attenuation of groundwater NO 3 − in shallow aquifers (Mastrocicco et al 2010, Exner et al 2014, Dwivedi and Mohanty 2016, may lead to significant lags between management efforts and improvements to groundwater quality (Lindsey et al 2003, Howden et al 2010, Meals et al 2010, Van Meter et al 2016. In 2015, approximately 47% of the U.S. population was estimated to rely on groundwater for domestic purposes including drinking water (Dieter et al 2018). This percentage was much higher in Oregon, where ∼70% of the state population relies at least partially on groundwater for domestic use, with close to 95% of rural populations entirely dependent on groundwater from private domestic wells (ODEQ 2017a). Over the past three decades, water samples collected from both private and public wells across the state have shown widespread groundwater NO 3 − contamination (ODEQ 2017b). Specifically, an extensive groundwater survey of the southern Willamette Valley (SWV) in Oregon, where 90% of N inputs are attributed to agricultural practices (LCOG 2008), revealed that much of the shallow groundwater of the region was chronically contaminated with NO 3 − at concentrations exceeding natural levels, i.e. >3 mg NO 3 − -N l −1 , indicating anthropogenic causes (Madison and Brunett 1985). Designated as a Groundwater Management Area (GWMA) in 2004, the Oregon Department of Environmental Quality (ODEQ) has since sought to control NO 3 − contamination in the area by promoting best management practices (BMP's) that reduce N inputs. However, despite 15 years of mitigation efforts 57% of wells in the SWV-GWMA exhibit increasing NO 3 − concentrations (Piscitelli 2019).
Increasing trends emphasize the urgency to link management practices to variations in groundwater NO 3 − concentrations. However, the legacy of past management and N accumulation have complicated these simple linkages. Given the prevalence of legacy NO 3 − in agricultural areas (Van Meter et al 2016), simply tracking changes in NO 3 − concentrations over time has been inadequate to evaluate long-term effectiveness of management practices (Nestler et al 2011, Utom et al 2020. Rather, the addition of isotopic tools to identify sources and transformations of N in groundwater may be an effective means for classifying wells based on unique patterns (figure 1). This approach may be especially important when legacy effects confound the ability to directly link current NO 3 − levels with improved aboveground agricultural practices (Meals et al 2010, Hamilton 2012 − values and groundwater NO 3 − concentrations can also be used to ascertain N transformation processes (e.g. Mayer et al 2002, Minet et al 2017, Veale et al 2019, Utom et al 2020. However, identification of NO 3 − sources and/or processing based solely on the analysis of δNO 3 − can be complicated by overlapping source δNO 3 − values, potential mixing of NO 3 − sources, and isotopic changes from biogeochemical processes (Kendall et al 2007, Xue et al 2009, Zhang et al 2018, Zhu et al 2019. Legacy effects may also impact interpretation, as δNO 3 − values in groundwater could represent a mixture of different sources and times (Hu et al 2019). Thus, for more accurate interpretation, multiple investigative tools should be used simultaneously (Hu et  Combining δNO 3 − with δH 2 O to identify hydrologic parameters could provide a mechanistic approach for understanding groundwater NO 3 − dynamics and help to distinguish areas vulnerable to long-term N contamination due to legacy effects. The main objectives of this study were to assess whether coupling of dual stable isotopes of δH 2 O and δNO 3 − can resolve questions about sources and transformations of N in groundwater systems, and to develop an approach to identify some key mechanisms influencing NO 3 − dynamics (figure 1 and table 1). To meet these objectives, NO 3 − concentrations, as well as the dual stable isotopes of δH 2 O and δNO 3 − , were measured in groundwater and domestic wells of the SWV-GWMA. We hypothesized that Colored symbols across plots (a)-(c) distinguish between well behavior categories as follows: blue squares = stable; red triangles = dilution; yellow circles = mixing; green diamonds = leaching; and, gray hexagons = denitrification. coupled isotopic indicators of δH 2 O and δNO 3 − would act as a powerful tool for classifying wells based on N movement, potential N sources with distinct isotopic signals, and transformations of N in the groundwater, allowing for identifying wells where management practices might address contamination issues.

Study location
The Willamette Valley, Oregon, USA, is a productive agricultural area with fine textured soils originating from the Missoula floods (O'Connor et al 2001). Characterized as having a modified maritime climate regime, the SWV-GWMA has cool, wet winters and warm, dry summers. Yearly precipitation ranges from 1020 to 1270 mm (with ∼80% occuring from October to March) and mean monthly air temperatures range from 3 • C-5 • C in January to 17 • C-20 • C in August (Uhrich and Wentz 1999). Though relatively flat-lying with very low relief (figure 2), a series of gently sloping and smoother terrace and floodplain surfaces have given the landscape an undulating or rolling topography moving out from the Willamette River (Roberts 1984). The region's mild climate and flat terrain is suited to produce orchard crops, nursery crops, blueberries, hay, and many types of grass grown for seed (Mueller-Warrant et al 2015).
Flowing mostly northward (figures 2 and S1), groundwater generally follows the contour of the land, similar to the flow of the Willamette River (Herrera et al 2014). Groundwater within the topmost shallow aquifer of the SWV-GWMA generally flows through the upper sedimentary unit, which is characterized by high permeability, high porosity, and high well yield (Conlon et al 2005). Horizontal hydraulic conductivities range from 1.06 × 10 −7 to 8.64 × 10 −2 m s −1 , vertical hydraulic conductivities from 7.06 × 10 −6 m s −1 , and storage coefficients from 3.00 × 10 −3 to 2.00 × 10 −1 . Flow tends to occur under unconfined conditions with typical water table fluctuations between 1.5 and 6 m of the surface (Conlon et al 2005). Data from USGS indicates that >80% of groundwater used throughout the Willamette Valley, which is principally recharged by direct infiltration of valley precipitation, is pumped from the uppermost alluvial aquifer layer (consisting of sand and gravel deposits) (Hinkle 1997) and used mostly for irrigation (Conlon et al 2005). Thus, regional water-quality monitoring has focused on the shallow groundwater (<25 m below land surface), which is likely most affected by anthropogenic activities (Hinkle 1997).
The southern part of the Willamette Valley was identified as a hot spot for N loading (Hoppe et al 2014) with NO 3 − contaminated groundwater (ODEQ 2004, Kite-Powell andHarding 2006). The SWV-GMWA (figure 2), which covers ∼600 km 2 of lowlands, was established in 2004 because of the high density of domestic and groundwater wells with elevated NO 3 − concentrations. The SWV-GWMA extends from Albany south to the city of Eugene. The boundaries approximate the limits of the underlying shallow alluvial aquifer, with the Willamette River flowing south-to-north through the center of the GWMA (figure 2). Agricultural land uses cover approximately 93% of the SWV-GWMA area (LCOG 2008).  When the SD of a parameter was <10% of its variability range, the parameter was initially identified as stable over time, and when it was >10%, it was initially identified as variable over time. We then assessed whether variable parameters were correlated within a well to further classify the behavior ( figure 1 and table 1).

Classification of wells
Theoretically, specific processes such as dilution with an alternate groundwater source, mixing of two groundwater sources with differing NO 3 − sources, leaching of legacy NO 3 − from overlying soils, and denitrification have unique isotopic signatures in this coupled dual isotope approach ( figure 1 and table 1 − became apparent in most of the wells of the SWV-GWMA (figure 3). However, well category was not related to well location across the SWV-GWMA (figure 4). Of the 39 total sampled wells, [NO 3 − ] in 28 wells varied over time. Nitrate trends in 85% of the wells (i.e. 33) could be classified based on concentration and isotopic patterns (figures 3(a)-(i)); overlapping processes in six wells, categorized as 'multiprocess' (figures 3(j)-(l)), make classification difficult using the coupled dual isotope approach alone.
Nitrate derived from fertilizers, soil organic matter, and animal manure/septic waste tend to have overlapping δ 18 O-NO 3 − values, in the range of −15‰ to +15‰ (Kendall et al 2007). Values of δ 18 O-NO 3 − in the 11 stable wells fell near the center of this range, extending from +0.2‰ to +8.5‰ ( figure 3(c)). However, δ 15 N-NO 3 − values tend to be more distinct, allowing for better discernment among these sources. Most synthetic fertilizers have δ 15 N-NO 3 − values in the range of −4‰ to +4‰, with some measured in the range of −8‰ to +7‰, while manure/septic waste tends to be more enriched in δ 15 N-NO 3 − , with typical values that range from +10‰ to +20‰ (Kendall et al 2007). Values of δ 15 N-NO 3 − in the stable wells ranged from 2.3‰ to 10.2 ‰ ( figure 3(c)). Together, the dual isotopes of δNO 3 − showed that synthetic fertilizer was the dominant agricultural NO 3 − source contributing to groundwater NO 3 − in the stable wells, with wells DW-5 and GW-8 potentially influenced by manure/septic waste sources ( figure 3(c)).   (table 1). While the groundwater composition of the wells was clearly impacted by a combination of NO 3 − sources, such as fertilizer sources, crop residues, and soil mineralization, our data precludes us from ascertaining the specific sources that mixed.

Leaching wells
In ∼25% of wells (i.e. ten wells), changes in [NO 3 − ] that ranged from 0.0 to 29.1 mg NO 3 − -N l −1 were classified as leaching of soil NO 3 − . The groundwater NO 3 − in these wells lacked any correlation with  3(i)). Seasonal precipitation and/or irrigation events are likely responsible for the release of fertilizer NO 3 − from overlying soils, leading to the leaching of excess NO 3 − into the groundwater.

Multi-process wells
For the six remaining wells, the [NO 3 − ] and isotopic patterns did not indicate one dominant process as being responsible for the NO 3 − trends, so they were given the categorization of multi-process (figures 3(j)-(l)). Concentrations of NO 3 − in these wells ranged from 0. in tandem with positive correlations between the dual δNO 3 − isotopes would seem to suggest denitrification processes are at play in wells DW-1524, GW-4S, GW-7, GW-18, and seasonally in GW-10 (table 1, figures 3(j) and (l)). However, the variability in δ 2 H-H 2 O and δ 15 N-NO 3 − values for the wells suggests that the influence of multiple sources cannot be ruled out. Thus, denitrification was not a dominant transformation pathway in any of the six wells (or in any of the wells throughout the SWV-GWMA). While we cannot distinguish the primary influences accounting for the variable [NO 3 − ] within the multiprocess wells, (i.e. whether multiple N transformation processes are occurring simultaneously, or mixing of water sources, and NO 3 − sources, or both), synthetic fertilizers and manure/septic sources appear to be the main contributors ( figure 3(l)).

Discussion
Given that NO 3 − is highly mobile and primarily originates from non-point sources, tracking its origins can be difficult. However, by analyzing δH 2 O and δNO 3 − in tandem we were able to identify multiple mechanisms and sources controlling groundwater [NO 3 − ]. We created a new framework for categorizing groundwater behavior (figure 1 and table 1), revealing insights into groundwatercontaminant interactions and helped identify where to target appropriate land management practices (Hansen et al 2017) to reduce groundwater [NO 3 − ]. While the overlap in isotopic values for multiple sources and the influence of isotopic fractionation pose limits, applying the coupled dual isotope approach at other locations could lead to more mechanistic understanding of the movement of water and contaminants within the groundwater. Experimenting with different management techniques in areas where groundwater [NO 3 − ] are known to be linked to contemporary land management practices could allow for unambiguous assessments of BMP's, eliminating the confounding effects of legacy lag-times (Meals et al 2010, Van Meter et al 2016).

Application of approach at SWV-GWMA
The variance in [NO 3 − ] and values of the coupled dual isotopic indicators of δH 2 O and δNO 3 − across space and time within the wells of the SWV-GWMA revealed the complex nature of groundwater NO 3 − transport throughout the relatively uniform shallow aquifer. We classified well behavior at this test site into five categories, with the percentage of wells in each category, from greatest to least, as follows: 28% stable, 26% leaching, 21% dilution, 15% multi-process, and 10% mixing. These results suggest that managing groundwater [NO 3 − ] in the region will require integration of different approaches, such as controlling NO 3 − sources and/or enhancing NO 3 − sinks across the landscape (Stigter et al 2011).
Synthetic fertilizers (69%), manure/septic sources (5%), or a mixture of the two (26%) were found to be the main sources of NO 3 − to the SWV-GWMA groundwater. These results align with a surface water modeling study based on conditions in the Willamette River Basin in 2002 that found agricultural fertilizer (27.2%) and animal manure (10.9%) were the largest contributors to incremental N stream loads (Wise and Johnson 2011). Similarly, Compton et al (2020) showed that agricultural activities accounted for 78% of the annual total N inputs to the entire Willamette River Basin for the years 2002-2006, with 69% of total inputs attributed to synthetic fertilizers and 7% to manure waste from permitted confined animal feeding operations (CAFOs) used as fertilizer. These numbers closely match those within the boundaries of the SWV-GWMA where agricultural crop activities contribute 90% of N inputs and CAFOs contribute 6% (LCOG 2008). Most of the nursery crops and grass seed of the region require significant inputs of synthetic N fertilizers (100-250 kg N ha −1 y −1 )  where a substantial amount can leach from the rooting zone into streams or the groundwater, especially when temporal asynchrony occurs between fertilizer application, crop N uptake, and hydrologic movement (Lin et al 2019).
Eight permitted CAFOs within the SWV-GWMA make up ∼2% of the land, and together contribute ∼6% of the total N inputs (LCOG 2008). The three largest operations account for ∼94% of the total CAFO N contributions and are closest to wells DW-10, GW-3, and GW-12. Average δ 15 N-NO 3 − values for these nearby wells are 8.8‰, 6.5‰, and 4.6‰, respectively. Typical values for manure waste tend to have δ 15 N-NO 3 − values ⩾10‰ (Kendall et al 2007), suggesting that a well's distance from a currently-permitted CAFO may not be the best parameter for revealing the true influence of animal agriculture on groundwater [NO 3 − ] in the region. The manure source signatures seen in two wells (DW-5 and GW-8) of the SWV-GWMA that are not close to any currently-permitted CAFOs could be due to the direct application of manure as a crop fertilizer to the surrounding agricultural fields, the legacy impact of past animal agriculture in the area, or the flow path and direction of groundwater.
Water isotopes were useful in elucidating the contributions of varying water sources and hydrological processes to the SWV-GWMA groundwater. Local valley precipitation was the main water source to the groundwater in 64% of the wells across the region, with evidence of Willamette River hyporheic water mixing with valley groundwater (Kendall and Caldwell 1998) in the remaining 34% of wells, which diluted [NO 3 − ] ( figure S2). This method worked well because the two sources were isotopically unique; however, the δ 2 H-H 2 O values of groundwater in each stable well were also isotopically distinct within the precipitation range ( figure 3(b)). These slight isotopic differences suggest that the shallow aquifer of the SWV-GWMA consists of highly compartmentalized groundwater pools that have limited lateral connectivity (Joshi et al 2018), likely due to the heterogeneity of the alluvial aquifer material. The slight but consistent isotopic differences also indicate that water isotopes could be a powerful tool even in locations without a broad range of isotopically distinct water sources.

Management implications for wells
Stable wells, i.e. those with relatively unchanging [NO 3 − ] and δ 2 H-H 2 O and δ 15 N-NO 3 − values (figures 3(a)-(c)), are unlikely to be immediately impacted by any new management modifications at the land surface. The stability of δ 2 H-H 2 O values suggests one slow-moving groundwater source to each stable well with long residence time (Broxton et al 2009, Thomas et al 2013. Given this, the stable δ 15 N-NO 3 − values, which indicate fertilizer-or manure/septic-derived NO 3 − sources, are likely signatures from past N inputs. While the [NO 3 − ] in stable wells appear to be disconnected from current surface inputs, the relatively low concentrations found in some wells (e.g. DW-9, GW-8,  suggest that land around them may be less susceptible to leaching of NO 3 − into the groundwater, or inputs of N in the past were more efficiently managed. The higher groundwater [NO 3 − ] of other stable wells (e.g. DW-10, GW-9, GW-27), however, could signify a long-term legacy of contaminated groundwater, which immediate land management changes could not resolve readily.
We found [NO 3 − ] variation was driven by dilution of an alternate groundwater source (Ogrinc et al 2019), the mixing of two NO 3 − sources (Kendall et al 2007), or the leaching of present-day (Minet et al 2017) or legacy N (Hu et al 2019) from overlying soils. The variable δ 2 H-H 2 O values in leaching wells suggest that groundwater within them has a short residence time (Broxton et al 2009, Thomas et al 2013, and thus the impact of surface management changes on groundwater [NO 3 − ] could potentially be assessed over relatively short timeframes. The residence time of groundwater in the dilution and mixing wells, however, is not as discernable. The source of high [NO 3 − ] could be from a stable groundwater pool with a long residence time, suggesting once again that legacy sources could be responsible for the contamination. Concentrations only decrease on the short-term when the contaminated water is influenced by another water supply (like the Willamette River) or another NO 3 − source ( figure S2). These wells could thus have long-term [NO 3 − ] contamination problems that are not addressed as quickly because evidence of other events (i.e. dilution by 'cleaner' river water or mixing with a lower concentration NO 3 − source; figure S2)  Denitrification was not found to be a dominant process in any of the wells of the SWV-GWMA. While many have found high denitrification in groundwater (Böttcher et al 1990, Tesoriero et al 2013, Minet et al 2017, others found it to be insignificant (Howard 1985, Wassenaar et al 2006, Jia et al 2020. In shallow, and even deep, aquifer systems, anaerobic conditions known to promote high levels of denitrification may be elusive (Hamilton andHelsel 1995, Lorite-Herrera andJiménez-Espinosa 2008). The absence of an adequate carbon source can also limit denitrification in soils and groundwater (Hiscock et al 1991, Rivett et al 2008, Weitzman et al 2014. Thus, the conditions necessary for denitrification were likely lacking across the SWV-GWMA. However, strategies that slow the movement of water through the soil profile or supplement low-organic soils with organic-rich carbon sources could increase denitrification.

Conclusions
Using the coupled dual isotope approach, we built a framework for classifying different processes responsible for groundwater [NO 3 − ] dynamics and confirmed the prevalence of legacy NO 3 − as a main contributor to groundwater contamination in an agricultural setting. Including δH 2 O and δNO 3 − analyses with standard [NO 3 − ] data could enable land managers to more effectively evaluate groundwater BMP's. The value of different improved N management strategies, such as the optimization of fertilizer use (rate, timing, location, and form), irrigation management, soil and tissue testing, cover crop adoption, and soil health promotion (Feaga et al 2004), may vary depending on the underlying behavior of the groundwater. Future work to elucidate fate and transport of groundwater N may benefit from the coupling of δH 2 O, δNO 3 − , and another discriminate isotope (e.g. boron, strontium, sulfate) or chemical tracers to further elucidate NO 3 − sources or processes.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/10.23719/1519089.

Acknowledgments
We thank Rich Myzak (Laboratory and Environmental Assessment Division of the Oregon Department of Environmental Quality) for his assistance with sample collection and analytical reporting. We thank Gary Bahr (Natural Resources Assessment Section of the Washington State Department of Agriculture) for his constructive comments and suggestions that helped improve and clarify this manuscript. This project was supported in part by an appointment to the Research Participation Program at the Office of Research and Development, U.S. Environmental Protection Agency, administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and EPA. This manuscript has undergone internal peerreview at the U.S. Environmental Protection Agency and has been approved for publication. The views expressed in this article are those of the author(s) and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency, U.S. Department of Energy, or the Oak Ridge Institute for Science and Education. Any mention of trade names, products, or services does not imply an endorsement by the U.S. Government or the U.S. Environmental Protection Agency.