A total of six monitoring wells were equipped with pressure and temperature transducers and were sampled monthly for major anions and field parameters. The provincial monitoring wells PGMN1 and PGMN2 were sampled monthly during the research while temperature and water table level were provided by the GRCA. The four pre-existing monitoring wells (LPS, LPD, MAS, and MAD), were reequipped in November 2021 with Van Essen TD-Divers (replacing Solinst transducers) for hourly monitoring of temperature and water level (well location on Figure 2, screened depth on Figure 5C).

A high resolution (HR) sampling location was selected for examining groundwater-surface water interaction and NO₃-N at a local scale on Landon’s Creek (Figure 2C). The instruments were installed on either side of the stream (Figure 2A). HR was instrumented with six 25 mm-diameter Solinst 615N Simple Well Point piezometers with a screen length of 20 cm (i.e., drive point piezometers installed using a post hammer). Four piezometers (DP1-4; screen top 1.6,1.8,3.0, and 1.4 m bgs) were installed in July 2022, two of which (DP3 and DP4) were equipped with Van Essen Micro-Divers to monitor groundwater levels on opposing sides of the stream. In May 2023, three similar piezometers were installed at the tributary (DP5-7; screen top 1.0, 1.8, and 0.4 m bgs). When targeting storm events in 2022 and 2023, these sites were used for high-frequency sampling.

Surface water instrumentation included an autosampler, pressure transducer, and in situ water quality stations. A Teledyne ISCO autosampler was use for targeting three storm events (Nov. 11–12, 2022; Mar. 22–23, 2023; Mar. 25, 2023). An in-stream Solinst pressure transducer, attached using aircraft cable to a steel fence post vertically hammered into the stream bed and protected with a perforated PVC pipe jacket, was used for temperature and stream stage measurements at 30 min intervals. Finally, three in situ water quality monitoring stations (labelled as SE, Figure 2) were used during the warm seasons at various sites from 2022 to 2024. These high-frequency monitoring stations were each equipped with a submersible Seabird Scientific Ultraviolet Nitrate Analyzer V2 (SUNA) and a YSI EXO Sonde (EXO). The SUNA is an in situ spectrophotometer which uses a wavelength range between 217–240 nm to measure NO₃-N concentrations in water (Johnson and Coletti, 2002; Seabird Scientific, 2024). Each EXO instrument was equipped with four probes for measuring electrical conductivity, temperature, pH, dissolved oxygen, and turbidity in stream water, but in the current project, only water temperature was used. The SUNA/EXO (SE) station SE1 (Figure 2B) was installed all 3 years (2022–2024) while SE2 (Figure 2D) at the outlet was in operation during 2023 and 2024. A third station, SE3 (Figure 2C), was used in 2024 for targeting storm events (Table 1).

Groundwater sampling from wells and drive point piezometers were conducted using either a submersible pump (Grundfos), peristaltic pump (GeoTech), or hand pump foot tubing (Waterra HDPE tubing), depending on well diameter, screen depth, and equipment availability. Wells and piezometers were purged by pumping three well volumes or the amount of water required for stabilization of field parameters measured with a YSI ProPlus multiparameter (sensors included: specific conductivity, temperature, pH, dissolved oxygen, and oxidation-reduction potential). Clean 1 L HDPE sample bottles were rinsed three times with fresh sample water, filled with no headspace for all groundwater and surface water samples. Surface water grab samples were taken at all steam gauging stations approximately 2.5 m from the bank of the creek using an extendable dip pole at a depth between 0.25 and 0.5 m at five locations along the stream (Figure 2A; Supplementary Figures S1–S5). Grab samples in the tributary (Figure 2C; Supplementary Figures S6–S8) were collected in the center of the creek at approximately 0.25 m from the surface by hand.

Three storm events (Nov. 11–12, 2022; Mar. 22–23, 2023; Mar. 25, 2023) were sampled with a Teledyne ISCO autosampler, which collected surface water samples every 2 hours and stored samples internally in 1 L HDPE bottles. Storm events assessed with the SUNA/EXO instruments were sampled at a frequency between 15 min and 1 hour depending on the sampling site and season (Table 1). Turbidity, sediment build-up, and sensor fouling impact the performance of the SUNA by impeding the UV light transmittance. The SUNA is designed to account for darker conditions by increasing the amount of time the sensor takes to scan its 256-channel spectrometer, but excessive fouling can still impact performance (Seabird Scientific, 2024). Throughout the deployment periods, sediments were manually removed from the sensors approximately once per month, with the 2024 season seeing cleaning occur every 2 weeks. Calibration with deionized water was conducted monthly for the SUNA and NO₃-N grab samples analyzed using ion chromatography were taken for comparison. The EXOs were also calibrated monthly following the procedures outlined by the YSI EXO user manual (Revision K; Xylem, 2020).

Seasonal patterns in discharge at the Mt. Vernon station (S5) are readily evident, such as the significant increase during the spring snow melt, starting around February 15th in 2022, March 15th in 2023, and February 1st in 2024 (Figure 4A). Maximum spring season flow rates then reached 33.3, 39.5 and 38.5 m³/s in 2022, 2023, and 2024, respectively. The extreme peak in discharge in August 2024, was linked to the tail of Hurricane Debby. Summer daily averages in discharge range between 1.3 and 3.5 m³/s, reaching 0.4 m³/s in the relatively dry summer and fall of 2022 while minimum discharge in summer/fall 2023 and 2024 was 0.8 and 0.7 m³/s, respectively.

Groundwater levels in the sand aquifer (Figure 5C) responded to seasonal changes in inputs from precipitation and snowmelt (Figure 4). The maximum difference in groundwater levels was seen in PGMN1 (2.1 m; Figure 5C). The two in-stream drive points located at the HR site (DP5 and DP7) showed the least total change in water level (0.65 and 0.61 m respectively, Figure 5C). LPD, LPS, DP3, and DP4 were more sensitive to precipitation, as noted by the noisier data (Figures 4C,D), than MAD, MAS, and PGMN2 (Figure 4B).

For major anions across all groundwater and surface water sampling locations throughout the study period, only NO₃-N, Cl⁻, and SO4^2- had measurable concentrations (Table 2; Supplementary Material). Nitrite (NO₂ ⁻), bromide (Br⁻), and phosphate (PO^4-) were always below the minimum detection limit (MDL) at all sites (0.10 mg/L). Across all groundwater samples, only 25% had fluoride (Fl⁻) concentrations above the MDL (mean = 0.7 mg/L; maximum = 3.2 mg/L) while surface water above the MDL occurred in 46% of samples (mean = 0.6 mg/L; maximum = 3.6 mg/L). Other field parameters measured over the study period were specific conductivity, pH, dissolved oxygen, and oxidation-reduction potential. Groundwater and surface water mean, minimum, and maximum values for the field parameters are summarized in Table 2 (timeseries graphs for each sampling location and field parameter are in the Supplementary Figures S1–S5).

NO₃-N concentrations in the shallow monitoring wells MAS and LPS averaged 17.2 and 5.0 mg/L NO₃-N with maximum values 38.4 and 11.4, respectively (Figure 5A). The deeper wells, MAD and LPD, had averages of 2.8 and 0.4 mg NO₃-N/L, and maximum values 20.2 and 6.9, respectively. Samples from the PGMN1 well never exceeded the NO₃-N detection limit while the PGMN2 samples averaged 12.5, peaking at 17.6 mg/L. Mean NO₃-N concentrations in the shallow drive point piezometers (DP1 to DP7) ranged between 21.4 (DP1) and 1.0 (DP4) mg/L NO₃-N, with a maximum value of 40.7 mg/L occurring in DP2 (Figure 5A).

Average NO₃-N concentrations at the five sampling locations on the main creek (S1-S5) varied between 4.1 (Inlet) and 3.9 (S2) mg/L (Figure 5B). Results from the tributary sampling sites (S6-S8) had NO₃-N concentrations between 4.3 (S7) and 5.6 (S8) mg/L, but the point nearest the outlet (S8) having a higher average concentration than the inlet (S7) (Figure 5A).

The ISCO method captured storm event responses at a single location on Landon’s Creek tributary (S6). During the storm events (Figure 6), the NO₃-N concentrations varied between 5.5 and 6.5 mg/L (November 11–12, 2022), 1.1 and 1.9 mg/L (March 22–23, 2023), and 0.90 and 2.0 mg/L (25 March 2023). For the three events, NO₃-N concentrations decreased between 0.8 and 1.1 mg/L during the storm. Figures 6B,C were consecutive events, with the second occurring 41 h after the first. The sampling gap between the two events was 30 h, after which the NO₃-N returned to the pre-storm high of ∼2 mg/L before dropping again during the following event.

Stream temperature varied slightly throughout the course of each storm. When compared to change in atmospheric temperature, surface water temperature changed by 1.9 °C (8.4–10.2) and air temperature by 6.8 °C (6.1–12.9) (Nov.11), 1.1 °C (3.1–4.2) and air temperature by 5.5 °C (2.2–7.7) (Mar. 22), and 1.1 °C (2.9–4.0) and air temperature by 10.8 °C (0.3–11.1) (Mar. 25). Shallow groundwater temperature (DP3, DP4) did not change during any of the three storms.

Change in water level for both surface water (S6) and shallow groundwater (DP3, DP4) varied over the course of each storm. S6 and DP4 showed similar changes the water level magnitude 2.8 vs. 2.9 cm (Nov. 11), 22.0 vs. 19.2 cm (Mar. 22), and 29.4 vs. 24.5 cm (Mar. 25) while DP3 saw changes of 1.4, 8.4, and 9.0 cm, respectively.

High resolution temporal data were captured at specific locations using the SUNA/EXO stations. All data collected with the SUNAs were filtered by removing measurements for which the difference between the light spectrum absorbance (up to 20,000 counts) and the dark spectrum absorbance (thermal noise; typically between 500–600 counts) readings was less than 1,500. It was found that values below this range indicated significant sediment build-up on the sensor, resulting in unreliable readings. Validation of the sampling method was done through comparing grab sample concentrations measured in lab with ion chromatography to SUNA readings (Figure 7). The SUNA consistently recorded higher concentrations of NO₃-N than the grab samples, averaging +0.75 mg/L [±0.77] (max.: 2.87 mg/L; min.: 0.28 mg/L).

Daily average NO₃-N concentration and water temperature were compared to daily precipitation. During the fall of 2022, a SUNA-EXO (SE) station was set up at the inlet (S4) and recorded a NO₃-N range of 1.24–2.28 mg/L. In 2023, the SE station at the outlet (S5) varied between 3.68 and 8.32 mg/L. During the 2024 sampling season, when all three SE stations were operational, NO₃-N concentrations varied between 2.18–5.65 (Inlet), 2.76–5.55 (Outlet), and 2.09–8.47 mg/L (HR). Daily averages of NO₃-N (considering only days with at least 8 h of measurements) had noisier readings from mid-April (DoY 100) to early August (DoY 220). When compared with the second part of the season, the noisier period corresponded to days with more daily precipitation (Figures 8A,C). Additionally, cooler temperatures in the tributary (S6) corresponded with higher concentrations of NO₃-N (Figures 8B,C).

During the 2024 sampling period, NO₃-N concentrations were compared to precipitation, water temperature, and change in stream level (i.e., stage height). The inlet and outlet SE stations were compared for two events, May 17 and 29 June 2024 (Figures 9A,B). NO₃-N peaked 5 days after the May event while it peaked just under 2 days after the June event at the inlet. The outlet peak for both events occurred after an additional 12 h. Peak values for the change in water level occurred approximately 12 h prior to all four peak values in NO₃-N.

For the first 10 days in August 2024 (DoY 213–222), three SE stations were compared (Inlet, Outlet, HR). NO₃-N notably dropped promptly over the course of a storm event at the HR site, also corresponding with an increase in water level (Figure 10). Such a relationship was not observed at the inlet and outlet SE stations during the same period (note: water level measurements at the inlet were lost during this period and NO₃-N concentrations are only available during daylight hours).

Sensitivity of the water table to precipitation (Figure 4) was assessed by comparing stream discharge at the outlet (Mt. Vernon) to the water levels recorded in the monitoring wells. The frequency of fluctuations for each well hydrograph was highest for MAD and MAS, followed by PGMN2, then LPD, LPS, and then PGMN1. The fluctuations in groundwater levels in LPD and LPS notably follow changes in discharge closely, perhaps due to the proximity of these wells to the main creek, which contrasts with PGMN1, a well significantly further from the creek, that followed a smooth seasonal pattern with lowest levels during summer. Given the fluctuations in LPD/S following stream discharge, it is possible that WC is either a losing or gaining stream at this location, depending on season and storm event, which could be affecting GW NO₃-N in LPS/D. For example, if WC is losing during high flows in the spring, the shallow groundwater would reflect NO₃-N concentrations in WC. During low flows, when WC is gaining, NO₃-N in LPS/D would be expected to reflect a different source, such as upgradient GW NO₃-N. While the pattern appears to reflect concentrations in LPS, it is not seen in LPD (see S1-5, Figure 5; Supplementary Figure S6). Therefore, LPD/S having lower relative NO₃-N as compared to MAD/S is likely caused by other factors, such as anthropogenic or biogenic factors discussed below.

On the south side of Whitemans Creek, only one well, PGMN2, was studied. It is screened at a depth comparable to that of MAD and PGMN1, yet it had a comparable average NO₃-N to MAS (15 and 17 mg/L). PGMN2 has a relatively deeper average GW level (∼8 m bgs) compared to PGMN1 (∼6 m bgs) and MAD/MAS (∼5.5 m bgs) and had smoother (less variable) well hydrographs which could imply less connection to surface. This could also imply a higher storativity, i.e., the aquifer at this location can store more water and effectively buffer impacts from surface inputs. Differences in storativity may impact response times and thus NO₃-N concentrations. Although both MAD/MAS and PGMN2 are screened into similar overburden material, coarse sand (Osman, 2017) and gravelly sand (Government of Ontario, 2017) respectively, the quaternary hydrostratigraphic units are different, with MAD/MAS in the Upper Erie phase aquifer (AFB1) and PGMN2 in outwash deposits (ATA2) (Bajc and Dodge, 2011). Further testing using pumping tests would help determine the extent to which storativity may be impacting responses.

Land use (Table 4) likely also impacted the results, since the MAD and MAS wells (2.8 and 17.2 mg NO₃-N/L) were surrounded by cash crops which are fertilized with nitrogen during the corn rotation. PGMN2 (12.68 mg NO₃-N/L) was similar in that nearby fields are corn or potatoes, whereas PGMN1 (n.d.) is in a tree nursery and the LPD and LPS wells (0.4 and 5.0 mg NO₃-N/L) are in a forested park (Figure 2).

Land use surrounding the PGMN2 well is notably different than other wells monitored. Moreover, PGMN2 is located less than 1 km from a housing subdivision in the Township of Burford, where residences use septic systems for wastewater treatment (County of Brant, 2023; Maxwell, 2020). The PGMN2 well is screened near a similar elevation (∼250 m asl) as the potentiometric surface located under the nearby houses (250–245 masl, Figure 3), from which elevated NO₃-N may be related, due to the septic systems which are known to contribute NO₃-N, along with many other contaminants, to groundwater systems globally (Gyimah et al., 2024; Robertson, 2021). Monthly sampling data for specific conductivity and chloride was notably higher in PGMN2 (1,348 μS/cm and 253 mg Cl⁻/L) than all other groundwater samples taken during the study (See Supplementary Table S1; Supplementary Figures S1, S2 for comparison). Additionally, the mixed agricultural land use (Figure 1) has recently consisted of potatoes and other vegetables along with corn and winter wheat (Table 4) on fields equipped with irrigation capabilities. These high concentrations measured at depths further below ground surface than the other wells (MAS, LPS) also screened near the water table may be indicative of legacy NO₃-N build up due to the relatively thicker vadose zone (Ascott et al., 2017a) or different antecedent GW geochemistry coming from the west, among other factors.

A comparable well to PGMN2 from Gardner et al. (2020) had high NO₃-N concentrations (average 13.5 mg NO₃-N/L) in Norfolk County, just south of LWC and in a sand plain aquifer, where proximity to residences on septic systems were identified as the likely source via timeseries patterns and isotopic signatures. However, not enough data was taken on the south side of the LWC catchment to be able to determine conclusively the sources of NO₃-N and its relationships to precipitation in PGMN2. Further study of NO₃-N isotopes could help identify sources of NO₃-N (Nikolenko et al., 2018). Some samplings of NO₃-N isotopes were conducted in 2023 at the site (for collection and processing methods see Arce-Rodriguez (2024); Supplementary Material), but PGMN2 did not show notable signatures for septic waste.

Heterotrophic denitrification, a redox transformation, in groundwater can occur under specific conditions, typically when NO₃-N concentrations are greater than 1 mg/L, dissolved oxygen is below 2 mg/L, pH is between 5.5 and 8.0, and chloride is below 500 mg/L (Rivett et al., 2008). Presence of electron acceptors (e.g., dissolved organic carbon) are also typically needed. To determine redox state of natural waters, methods such as cross-referencing dissolved oxygen (DO), manganese (Mn), iron (Fe), and NO₃-N to field measured oxidation-reduction potential (ORP), categorized into anoxic (−300 to 0 mV), suboxic (0–50 mV), and oxic (50–300 mV) (Gardner et al., 2020; Kehew, 2001; Stumm and Morgan, 1996). As the transformation of NO₃-N requires anoxic environments, when the redox potential is moderate-to-low, NO₃-N is not expected to be present (Grant Ferris et al., 2021; Kehew, 2001).

When comparing monthly groundwater samples over the course of the study the deeper wells averaged between 2.7 and 4.1 mg DO/L (MAD, LPD, PGMN1) while the shallow ranged from 6.7 to 11.2 mg DO/L (MAS, LPS, PGMN2, DP1-7). Only wells LPD and PGMN1 had, on average, reducing conditions with ORPs of −35.34 and −19.49 mV respectively. However, all groundwater samples recorded ORP values below zero at varying points in the study (See Supplementary Table S2; Supplementary Figures S3, S4 for further detail). Chloride concentrations never exceeded 350 mg/L in any groundwater samples and pH consistently ranged between 6.2 and 8.5 (Supplementary Tables S1, S2). Considering these factors, it seems unlikely that denitrification contributed to the noted lower concentrations of NO₃-N with depth between MAS and MAD. A stronger argument may be made for denitrification potentially occurring at LPS and LPD during times when DO fell below 2 mg/L. For PGMN2 and DP1-7 favorable conditions for denitrification were less likely, since all these monitoring points are screened just below the average water table depth and have relatively high DO and ORP values. However, sampling these wells required the use of foot valve hand pumping, which prevented a flowthrough cell from being used when sampling, thus may have artificially increased the measured DO values. PGMN1 had the strongest evidence for denitrification contributing to NO₃-N rarely being detected in the well, as the DO was consistently low (2.7 mg DO/L) and ORP averaged below 0 mV.

The 2023 NO₃-N isotopes samples from Arce-Rodriguez (2024) show indications of denitrification in only three of the 75 samples taken, all three originating from the MAD well. Further delineation of sources was not possible due to most samples falling within the Soil-N range, where sources of NO₃-N range from possible mineral fertilizers, naturally occurring in soil, manure, or septic sources (+3‰ to +8‰; Nikolenko et al., 2018).

During all storm events, water elevations in streams and groundwater level both rose within a short timeframe, within a day on Landon’s Creek (Figures 6, 10) or within two to 5 days in Whitemans Creek (Figure 9). Deeper monitoring wells further away from the tributary and creek (MAD/S, PGMN2) did not appear to respond in the same rapid fashion as seen at the high resolution site (DP3, DP4) or the shallow well near the main creek (LPD/S), which can be seen in the relative smoothness of water level measurements in Figure 4B as compared to Figures 4C,D. When linked with NO₃-N concentrations, the rapid response in the tributary to storm events initially lowered the NO₃-N concentration but was followed within a day by a return to pre-storm levels (e.g., 1.1 mg/L at the end of March 23 and 2.1 mg/L beginning 24 March 2023; Figure 6). In the LWC, no noticeable NO₃-N dilution was seen due to the storm events and, on the contrary, a noticeable increase occurred slightly delayed from the peak in discharge. Additionally, the outlet for both May and June/July lagged about half a day behind the peak in concentration occurring at the inlet (Figure 9). The August 1–10, 2024 rain event was the most thoroughly captured event using the methods attempted. Decreasing NO₃-N concentrations were observed at the HR site, followed by a rapid rise to pre-storm levels (e.g., ∼6.5 mg/L pre Aug. 6th storm, ∼5.5 mg/L mid storm, and return to ∼6.3 mg/L by Aug. 8th, 2024; Figure 10), while no notable drop in NO₃-N was recorded during the same period at the Inlet and Outlet stations (Figure 10).

This can be interpreted as the small tributary likely acting as a conduit for NO₃-N entering the main creek. From the research conducted at the HR site, Landon’s Creek appears to be a gaining stream, which could explain why it has particularly high NO₃-N concentrations. The creek supports coldwater species of brook, brown, and rainbow trout (personal communication with Trout Unlimited Canada [now Freshwater Conservation Canada) – Middle Grand Chapter, 2024; (OMNR, 2025)]. It maintains a higher temperature relative to LWC during the winter (min.: 2.3 °C–4 °C at S6 vs. 0 °C–0.2 °C at S1) and lower during the summer (max.: 16.4 °C–20 °C at S6 vs. 20.7 °C–22.5 °C at S1) (Supplementary Table S2), did not freeze over in the winter (both the inlet and S3 froze at times), and did not go dry in the summer. A similar study on the larger Bazile Creek in Nebraska (1958 km²) highlighted the higher impact of some groundwater-fed tributaries in conducting NO₃-N to the main creek, emphasizing the need for sampling lower-order streams (Richards et al., 2021).

On the field scale at the HR site, DP4 has the lowest average NO₃-N concentration (1.40 mg/L) and is located on an alfalfa field only receiving potash (potassium) as fertilizer. Highest concentrations are on the cash crop field (Corn-Soybean-Winter Wheat) (DP1:19.8 mg/L, DP2:17.8, DP3: 15.4) and in between are the bank/bed DPs (DP5:11.0 mg/L, DP6: 10.6, DP7: 10.6 (Figure 5A). Elevated levels of NO₃-N in shallow groundwater can be linked to fertilization under high water inputs, such as the case with irrigation (Su et al., 2022). While the HR site fields are not irrigated, as is the case with the majority of operations within the LWC catchment, the associated shallow GW levels also appear to correspond to surface inputs from precipitation (rain and snow), as highlighted by the increase in apparent noise of the timeseries where the water table is nearer to the surface (Figure 4). The delay time difference between stream stage (S6) and GW levels at the HR Site, where DP4 responds before S6 and DP3 responds after, may, in-part, be due to different land usage. Rapid (sub-daily delay) shallow groundwater verses stream response to precipitation events have been shown to vary in a small subcatchment, with GW responding before or after depending on land use (Gootman and Hubbart, 2021). The implication being that local land use can notably change response to precipitation and should be considered for assessing response to non-point source contaminants. However, more events than those captured in this study would be needed to draw a stronger connection to land use specifically, as differences in slope and relative location to S6 are present.

The shallow groundwater below a corn-soybean-winter wheat rotational field (Figure 1) averaged between 15–19 mg/L (DP1-3) and the groundwater in the nearby stream between 10–11 mg/L (DP5,7), while the tributary (S6) averaged 5 mg NO₃-N/L (Figure 5B). The difference between the average groundwater and surface water at the HR Site may be the result of water contributions from the shallow GW on the alfalfa field (1.40 mg/L at DP4) or deeper groundwater from further away (e.g., MAD averaging 2.2 mg/L) discharging to Landon’s Creek, causing a dilution. Alternatively, the difference may, in part, be due to the shallow GW and SW being transported through a wetland area (Figure 3C), as retention in the hyporheic zone via denitrification is a possibility (Boulton et al., 2010; Lefebvre et al., 2024).

Over the course of the study, it was noted that Landon’s Creek had a pattern of downstream increase of NO₃-N, implying the ability of the stream to retain nitrogen is less than the overall contribution, a common situation in groundwater fed streams near cropped fields (Hill, 2023). As the water travels downstream in Landon’s Creek the average concentration increases from 4.0 to 5.5 mg/L. In contrast, Whitemans Creek holds a relatively constant average of 4.1 mg/L between the inlet and the outlet (S1-5, Figure 5). Since the discharge also increases, either groundwater or surface water necessarily contributes to the stream flow. If the surface water from other small streams act similarly to Landon’s Creek, then Whitemans Creek must either be retaining NO₃-N, receiving discharge from the lower parts of the aquifer (i.e., PGMN1, MAD, LPD), or some combination to maintain a constant NO₃-N at the outlet when contributing SW is higher.

The scientific literature shows that the presence of wetlands can contribute to reduce non-point source pollutant loading to groundwater and surface water (Nsenga Kumwimba et al., 2023; Ranalli and Macalady, 2010; Walton et al., 2020). Natural attenuation of NO₃-N occurs mainly via denitrification, but plant assimilation, and microbial immobilization can also play a role, all of which can vary with location, season, hydrogeology. Some recent studies focused on small groundwater-fed wetlands used 222Rn as a natural tracer to delineate water source (Lefebvre et al., 2024), or in-stream high-frequency NO₃-N sampling (Crossley et al., 2025). They demonstrate both the potential for retention and the many simultaneous factors (e.g., temporal changes in geochemical conditions for denitrification or storm intensity) affecting this retention. Lefebvre et al. (2024) found that 80% of groundwater entering a small stream was via the wetland flow, where denitrification and plant assimilation can occur. In the current study, beavers were observed to create dams which flooded the HR site, something which potentially can decrease NO₃-N downstream transport and increase nutrient storage (Larsen et al., 2021). Here, the experimental methods were not designed to highlight these effects, but the presence of a riparian wetland inhabited by beavers in the HR site may decrease NO₃-N concentrations in the shallow groundwater of DP5-7 and in Landon’s Creek (S6-8). However, beaver activity on Landon’s Creek may also negatively impact nearby farms through flooding and the coldwater species through water temperature increase associated with stream bank deterioration.