Groundwater characterization of the eastern Minesing Wetlands in support of the endangered Hine’s emerald dragonfly (Somatochlora hineana)

The Minesing Wetlands contain Canada’s only known population of Hine’s emerald dragonfly (HED), a species listed as endangered since 2011. The HED relies on groundwater discharge areas to complete its life cycle and therefore the population is potentially sensitive to changes in groundwater. The goals of this study were to, (1) characterize groundwater discharge to the eastern portion of the Minesing Wetlands, which corresponds to the HED habitat, (2) document any significant changes in groundwater chemistry since a 1998 study, and (3) establish a comprehensive baseline water quality data set to facilitate future comparisons. Consistent with HED habitats in the USA, the chemistry of Minesing Wetlands groundwater was dominated by calcium and bicarbonate. Low groundwater chloride and nitrate concentrations indicated minimal impact from activities such as road salt application and fertilizer use. Nitrate was generally only found at the wetland margin where groundwater springs emerged from the base of the uplands. The dominant form of inorganic nitrogen in wetland groundwater was ammonium (max = 1.5 mg N/L), and soluble reactive phosphorus concentrations were also relatively high (8 to 122 µg/L). The most consistent change in wetland groundwater chemistry between the 1998 study and samples collected in 2015–2017 was that chloride concentrations in 10 of 23 wells increased over time, however the maximum mean well chloride (11.6 mg/L) remained well below the guideline for the protection of aquatic life (120 mg/L). To preserve HED habitat, potential impacts to groundwater need to be considered during land use planning for the recharge area.


Introduction
The globally rare Hine's emerald dragonfly (Somatochlora hineana; HED) is a medium-sized dragonfly (about 60 mm long) characterized by bright green eyes, a metallic green thorax with two lateral yellow stripes, and a blackish-brown abdomen. Fish and Wildlife Service 2001). The only known population in Canada is located in the Minesing Wetlands, Ontario. In 2011, the HED was listed as federally endangered in Canada (COSEWIC 2011) and provincially endangered in Ontario by the Committee on the Status of Species at Risk in Ontario. Although considerable effort to locate this species in other similar habitats within the province have been undertaken, it has yet to be found elsewhere (COSEWIC 2011). The Ministry of Natural Resources (MNR) released the HED Ontario Government Response Statement (Ministry of Natural Resources 2017) outlining the key focus areas for the protection and recovery of HED: (1) Protection of HED habitat and mitigation of threats through the adoption of best management practices to maintain current groundwater quantity and quality, and (2) Research and monitoring with the objective of increasing knowledge of HED biology in Ontario including the assessment of groundwater quantity and quality in recharge and discharge areas.

HED lifecycle
Hine's emerald dragonflies rely on groundwaterdependant ecosystems to carry out their life cycle. HED eggs are deposited in shallow waters in areas of herbaceous vegetation in marshes, meadow marshes, and fens (Lee et al. 2006). The larvae remain in cool, shallow, slowly-moving waters for 3 to 5 years before emerging as adults (Pulfer et al. 2013), then may live for 4 to 6 weeks. Kay et al. (2018) outlined some of the key functions that groundwater performs to enhance the survival of the early life stages of the HED. In addition to providing eggs and larvae with protection from desiccation, seasonally stable groundwater levels and temperatures also buffer the developing HED from thermal stress, including preventing their habitat from freezing solid during the winter. Furthermore, compared to typical stream environments, groundwater flows are usually slow and gentle, thereby minimizing egg losses due to habitat erosion.
Habitat requirements of the HED include groundwater-sourced wetlands that are dominated by graminoid (grass, rush, and sedge) vegetation and groundwater with a chemical composition dominated by calcium, magnesium, and bicarbonate (Cobb and Bradbury 2008). Larval nymphs use crayfish burrows for protection from desiccation during summer drought and from freezing during winter months (Pintor and Soluk 2006). While crayfish are predatory towards HED larvae, the availability of crayfish burrows may be a critical limiting factor to HED populations since the larvae would likely only survive for up to 2 weeks during drought conditions without the burrows (COSEWIC 2011;Pintor and Soluk 2006;Pulfer et al. 2013). Within the Minesing Wetlands, HED larval habitat has also been associated with the presence of digger crayfish (Fallicambarus fodiens; COSEWIC 2011). Digger crayfish burrows are most densely distributed in the fen communities, with scattered occurrences elsewhere throughout the Minesing Wetlands.
Alteration to the quantity and/or quality of groundwater discharging to the wetlands used by HED is the greatest threat to its survival. In the uplands adjacent to the eastern Minesing Wetlands, changes in land use could impact groundwater and thus HED habitat. While this upland area is largely agricultural, it is also on the edge of the city of Barrie, Ontario, which is undergoing rapid expansion. Furthermore, the anticipated effects of climate change could influence wetland characteristics by impacting the quantity and quality of groundwater recharge in the area. The objective of this study was to characterize the groundwater hydrology and geochemistry of Canada's only known HED habitat. This multi-year investigation provides information on the water quality conditions suitable for HED and also serves as a baseline study to monitor for changes in these parameters over time. The specific objectives of the geochemical analysis were to: 1) Characterize groundwater chemistry in the eastern portion of the wetlands in the main ecohydrological units (graminoid fen, discharge wetland system, flow-through wetland system), and recharge area, and 2) examine indicators of anthropogenic impact to the wetlands, including chloride and nitrate.

Site description
The provincially significant Minesing Wetlands, located approximately 80 km north of Toronto, Ontario, spans an area of over 6,000 hectares ( Fig. 1) and is bisected by the Nottawasaga River. Designated as a wetland of international importance by the Ramsar Convention on Wetlands, it is home to over 520 plant species and over 560 animal species, some of which are provincially, nationally, or globally rare (Bowles et al. 2007). The climate of the area is temperate, with mean daily temperatures ranging from 20.0 °C in the summer (July) to -5.6 °C in the winter (January) and average annual precipitation of 712 mm (Government of Canada 2021;Egbert, ON, station;2010-2017. Upgradient of the HED habitat, land use consists largely of agriculture within 5 km of the eastern edge of the Minesing Wetlands, beyond which lies the city of Barrie, with a population of approximately 146,000 (Statistics Canada 2016).
Hydrologically, the Minesing Wetlands can be subdivided into two regimes: the southeast fen and mixed boreal complex swamp, which are dominated by groundwater, and the remaining wetland areas to the west, which are dominated by surface water processes (Bowles et al. 2007;Fig. 1). Situated at the base of the upgradient succession of former shorelines of glacial lakes Payette, Nipissing and Algonquin, the glacial Lake Payette buried shoreline corresponds to the edge of the eastern Minesing Wetlands and represents the groundwater discharge front (Bradford 1999). Much of the groundwater recharge area for the eastern, groundwater-dominated aspects of the Minesing Wetlands corresponds to the Snow Valley Uplands (Beckers and Frind 2001). This area above the glacial Lake Algonquin shoreline has gently sloping topography extending between 260 and 310 masl with soils that are dominated by sand, gravel and till. Below the relic shorelines, the Minesing Wetlands are an expansive area that is relatively flat, ranging in elevation between 180 and 190 masl with organic soils underlain by glaciolacustrine clay and silt (Bradford 1999).
The southeast border of the Minesing Wetlands is bracketed by the largely buried ancestral shoreline of glacial Lake Payette (a lower stage of glacial Lake Algonquin) (Bradford 1999). The base of these shoreline deposits corresponds to the discharge wetland system, a mixed conifer swamp complex that is supported by relatively constant regional groundwater discharge (Bradford 1999;Beckers and Frind 2001). Moving west towards the fen, this groundwater discharge coalesces into spring-fed tributaries and the vegetation changes from the open mixed/conifer swamp to low, hummocky terrain supporting deciduous and mixed forests.
The flow-through wetland system transitions to a conifer swamp that develops tamarack-dominated components approaching the central shrub fen. It is characterized by very slow moving groundwater with the water table at or above the ground surface.
The graminoid fen ecohydrological unit is a minerotrophic, peat-accumulating ecosystem that corresponds to the HED Category 1 habitat and is bracketed by coniferous string islands. Fens have perennially saturated soils from groundwater inputs, with vegetation largely consisting of grasses, sedges, rushes, shrubs, and various broad-leaved plants.

Recharge area delineation and recharge rates
Particle tracking (both forward and reverse) was previously completed by Matrix Solutions (2014) to identify the contributing recharge areas to the discharge wetland ecohydrological zone, and by extension to the flow-through system and graminoid fen. The calibrated groundwater model for the Barrie Tier Three Risk Assessment (AquaResources et al. 2013) was used to delineate recharge areas. Four major hydrostratigraphic units composed of sand and gravel (A1 to A4 in order of depth) underlie the study area and are part of a regionally extensive and complex aquifer system (AquaResources et al. 2013). The regional aquifer A2 is the principal source of groundwater to the eastern portion of the wetland complex (Beckers and Frind 2001). At the request of the Nottawasaga Valley Conservation Authority (NVCA), particle tracks with travel times < 30 days were not used to delineate the fen recharge area (Matrix Solutions 2014).
The modelled contributing recharge area was approximately 325 km 2 . Particle tracks that ended within the flanks of the Minesing wetland primarily originated in the Snow Valley uplands region, northwest of the City of Barrie and southwest of the Town of Midhurst. Recharge originating from the Snow Valley Uplands was simulated to discharge from aquifer A2 to the eastern boarder of the wetland and corresponded to a < 25 year time of travel capture zone. Based on the particle tracking results, the delineated recharge area within the < 25 year zone covered approximately 18 km 2 , was situated predominately in the upland area east of the wetlands, and extended approximately 2-3 km southeast from the study site. The Pinegrove well was located in the modelled < 6 month time of travel zone whereas the Grenfel well was situated on the border of the 10-25 year time of travel zones (Fig. S1).
Recharge rates were relatively high with an average of 307 mm/year, varying from 50 to 450 mm/year depending on vegetation, slope, and soils. Recharge originating from within the < 25 year zone accounted for approximately 67% of the discharge volume to the fen, with the remainder having a greater travel time or occurring within 30 days. Land use in the recharge area is dominated by row cropland (i.e., wheat, oats, barley, corn, soybean, etc.) with deciduous or coniferous forests predominating on the flanks of the upland. Low density, rural residences in the area have private wells and septic systems. Several streams prograde off the flanks of the uplands into the wetland and provide cold and cool water fisheries habitats (Bowles et al. 2007).

Minesing Wetlands groundwater monitoring network
Groundwater monitoring in the Minesing Wetlands was conducted using a series of shallow monitoring wells and piezometers located along two transects, T3 and T6 ( Fig. 1, 2, Table S1). Originally established in 1996 by Bradford (1999), the transects commence at the glacial Lake Payette shoreline at the eastern margin of the wetlands and terminate in the fen. Seventeen of the original groundwater monitoring sites were located through field reconnaissance in 2009. However, other piezometers and monitoring wells that were originally located on these transects could not be found. Each groundwater monitoring site consisted of a standpipe screened just below the surface of the peat and a piezometer screened just above the mineral soil (screen mid-point range = 1.5 to 2.3 m). All monitoring wells were made of PVC and had screened intakes approximately 65 cm long with an inside diameter of 3.2 cm.
Nine additional monitoring sites were installed in October, 2013 to infill the existing Bradford (1999) T3 and T6 transects in areas of known co-occurrence of HED and digger crayfish (Fig. 2, Table S1). Two 6.4 cm diameter monitoring wells were installed at each additional site by hand-augering. These new installations consisted of, 1) a shallow standpipe that was typically screened approximately from 10 to 80 cm below the ground surface and, 2) a deeper piezometer, which was completed to the peat-marl clay sediment interface, generally located between 2.5 and 3 m below the ground surface, with a 70 cm, slot-10 screen.
Two upland groundwater monitoring wells were installed in 2014, one located on Simcoe County Forest property off Pinegrove Road and one on municipal road allowance off Grenfel Road (Fig. 2, Table S1). The purpose of these drilled monitoring wells was to characterize the groundwater quality and levels in the recharge area located upgradient of the HED habitat. The two monitoring wells were completed in the same stratigraphic/hydrostratigraphic unit as the wetland seepage front. Continuous water level logging commenced in the Pinegrove well in May 2014 and in March 2015 for the Grenfel well, although Loggers (Onset HOBO U20) installed in select wells, generally from May to November, 2014 to 2017, recorded water levels at 60 min intervals with an accuracy of ± 0.005 m. Raw water level data was adjusted for atmospheric pressure fluctuations and calibrated based on manual static readings three times per annum, during logger deployment, mid-summer, and logger retrieval.
The Minesing Wetlands groundwater quality dataset is comprised of 9 sampling events: October 2010, November 2013, September 2014, and May and November 2015, 2016, and 2017. Sampling events were generally conducted during periods without significant recent precipitation in order to minimize surface water runoff effects in the wetlands during groundwater sampling.
Water levels in each well were determined prior to purging in excess of three well volumes using either a manual foot valve pump (Waterra) or a portable peristaltic pump for the wetland wells and a Grundfos pump for the two recharge area wells. Following purging, each well was sampled and field-filtered to 0.45 µm for a suite of water quality parameters including, major anions and cations, nutrients, and select stable isotope analyses. Depending on the specific parameter, samples were stored either refrigerated or frozen until analysis. In May 2016 and 2017, grab samples were collected from groundwater seeps discharging at surface along the base of the eastern slope of the Minesing Wetlands (Fig. 2). Seep samples were filtered and analyzed in the same manner as groundwater samples collected from the wells and piezometers.

Analytical methods
Unless otherwise indicated, chemical analyses were conducted by Environment and Climate Change Canada (ECCC) at the Canada Centre for Inland Waters (CCIW), Burlington, Ontario. Sample concentrations were calibrated against multi-ion standards that were analyzed with the samples. When necessary, samples were diluted with Milli-Q water to bring concentrations within the working range of the standards.

Anions
Major anion concentrations were measured using a Dionex 2500 ion chromatograph on samples that had been refrigerated until analysis. Method detection limits (mdl) for chloride, nitrate-N, and sulfate were 0.01, 0.02, and 0.05 mg/L, respectively. Bicarbonate was determined by titration using a Hach Alkalinity Test Kit (Model AL-DT) with an assumed precision of ± 2%.

Cations
Samples for major cations (calcium, magnesium, sodium, potassium) were acidified with nitric acid prior to being refrigerated until analysis. Cation concentrations were determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using a Horiba Jobin Yvon Ultima 2 ICP. The method detection limits for calcium, magnesium, sodium and potassium were 0.03, 0.01, 0.03, and 0.20 mg/L, respectively.

Ammonium
Sub-samples for ammonium analysis were acidified with hydrochloric acid to a pH of approximately 5-6 and analyzed within 48 h or stored frozen until analysis. Ammonium concentrations (mdl = 0.01 mg N/L) were determined using a colorimetric method (Salicylate-Nitroprusside) by measuring absorbance at 640 nm on a Beckman-Coulter DU720 UV/visible spectrophotometer.

Soluble reactive phosphorus (SRP)
SRP samples were refrigerated until analysis within 48 h or less. SRP concentrations (mdl = 2 µg/L) were determined using an orthophosphate colorimetric technique, measuring the absorbance at 885 nm using a 10 cm path length cuvette on a Thermo Scientific Evolution 201 UV-visible spectrophotometer.

Water isotopes
Water isotope ratios ( 2 H/ 1 H, 18 O/ 16 O) were measured by laser absorption spectroscopy (Wassenaar et al. 2014) with an off-axis integrated cavity output spectroscopy (OA-ICOS) water isotope analyzer (Los Gatos model DLT-100) coupled to a CTC-PAL liquid autosampler at ECCC's Stable Isotope Laboratory, Saskatoon, Saskatchewan. Calibrated internal laboratory reference waters were used to normalize water isotope results to the Vienna Standard Mean Ocean Water (VSMOW) scale. Analytical precisions were ± 0.1‰ and ± 1.0‰ for δ 18 O and δ 2 H values, respectively.

Nitrate isotopes
Nitrate isotope ratios ( 15 N/ 14 N, 18 O/ 16 O) were determined using a modified version (Spoelstra et al. 2014) of the chemical denitrification method (McIlvin and Altabet 2005;Ryabenko et al. 2009) at the Environmental Isotope Lab, University of Waterloo, using a GV Trace Gas pre-concentrator system (GV instruments, Thermo Electron Corp., Manchester, UK) attached to a GV Isoprime mass spectrometer. Internal and internationally recognized reference standards (USGS-34, USGS-35, EGC-1, EGC-17) were used to calibrate the final δ 15 N and δ 18 O values, which were reported relative to atmospheric N 2 and VSMOW, respectively. Analytical precisions were ± 0.4‰ and ± 1.0‰ or better for δ 15 N and δ 18 O values, respectively.

Groundwater levels
Static water level elevations closely mirrored the wetland topography, with the average water table about 0.18 mbgs in the discharge system to 0.09 mbgs in the flow-through system. Averaged static water levels in the graminoid fen were essentially at ground surface for all wells within this unit, with shallow standing water often present. Flow direction followed the regional topography, sloping to the west towards the Nottawasaga River from the Snow Valley Uplands (Fig. 2). The discharge wetland system exhibited a topographic gradient of 4 m over roughly 400 m, whereas the flow-through wetland system and the graminoid fen were essentially flat. Along the T3 and T6 transects, minimal head differences between the paired shallow and deep piezometers indicated that groundwater gradients were largely horizontal with minimal downward gradient (Bradford 1999, this study). The prominently lateral flow was the result of thick, relatively laterally continuous glaciolacustrine silts and clays that underlie the wetland complex, acting as a regional impermeable unit and directing flow laterally instead of vertically.
Groundwater levels were more variable in 2016 compared to 2017, with 2016 having a more uneven distribution of precipitation, combined with less overall precipitation (Fig. 3). For instance, in 2016 there were six precipitation events over 20 mm, with one event over 45 mm. However, in 2017, there were seven events of over 20 mm but no single event exceeded 30 mm. The 45 mm precipitation event in 2016 was noted in all hydrographs with a one to four day lag period. This indicates that, even though the eastern aspect of the Minesing Wetlands was dominated by groundwater discharge, precipitation events had a significant influence on the overall water table elevation, especially in the summer months when evapotranspiration is significant.
The two recharge area wells (Pinegrove and Grenfel), installed with screens in a confined aquifer, displayed typical groundwater hydrographs for the water year, with lower water elevations in March-April and higher levels in the fall (October-November). The individual hydrographs exhibited a range of 2.5 and 1 m for the Pinegrove and Grenfel wells, respectively (Fig. S2).
The isotopic composition of groundwater in the Minesing Wetlands was confined to a narrow range of δ 18 O-H 2 O and δ 2 H-H 2 O values compared to those observed for meteoric water in the area (Fig. 4). δ 18 O-H 2 O and δ 2 H-H 2 O values determined from precipitation collected at Egbert, Ontario, located approximately 17 km from the study site, have previously been found to range from -19.4 to -3.9‰ and -142.5 to -26.0‰, respectively (IAEA/ WHO 2020; 1997-2002 monthly data). In contrast, groundwater collected for this study ranged from -13.1 to -9.4‰ (mean = -11.2‰) and -92.2 to -65.3‰ (mean = -77.1‰), respectively. Due to mixing of seasonal recharge along groundwater flow paths, the isotopic composition of groundwater tends to be similar to the mean values of annual precipitation. The relatively narrow range of the groundwater isotope values indicates that groundwater flow dominates the hydrology of the Minesing Wetlands, since the seasonal variability in isotopic composition (i.e., high values in the summer, low values for snowmelt) was not observed in the groundwater, despite the very shallow depth of collection. The δ18O and δ2H values of groundwater in the graminoid fen and flow-through wetland zones tended to be greater than those areas corresponding to groundwater source areas of the wetland (discharge system, springs, upland wells) (Fig. 5). These higher values were likely due to evaporation effects as the water moved laterally through the wetland. Evaporation would result in increasing δ 2 H:δ 18 O values along a slope of approximately five, which, given the narrow range of the groundwater isotope values overall, is not distinguishable from the slope (7.3) of the local meteoric water line (LMWL) (Fig. 5).

Major ion geochemistry
The chemical composition of groundwater in the Minesing Wetlands and surrounding recharge area was largely controlled by the mineral composition of the aquifer materials. Groundwater chemistry in the Minesing Wetlands was therefore dominated by calcium and bicarbonate (measured as alkalinity), with concentrations ranging from 23 to 93 mg/L and 96 to 408 mg/L, respectively (Fig. 6, S3-S10). Magnesium and sulfate were typically the second most abundant ions present. Over the time period sampled, the five ecohydrological units were not statistically different from one another with respect to major ion composition (Fig. 6), (Table 1). Mean sulfate values were higher in the recharge area and groundwater seeps compared to those of the discharge, graminoid and flow-through portions of the wetland, where sulfate reduction occurred in anaerobic, organic-rich wetland horizons at some sampling sites. The major ion geochemistry of the Minesing Wetlands was consistent with previous studies of HED sites in Wisconsin and Illinois that have shown HED being restricted throughout its range to groundwater-fed wetlands dominated by calcium/magnesium carbonate waters (Soluk et al. 1998  To determine if Minesing Wetlands groundwater chemistry has changed over time, we compared data from our three most recent sampling years (2015 to 2017) to those from July 1998 (Bradford 1999). For each well that was sampled during both studies, the single value for each parameter from 1998 was compared to the 95% confidence interval around the mean of the 2015-2017 values (Fig. 7, Table S2). Of the 23 wells used in the comparison, calcium concentrations were the same (n = 17) or higher (n = 5) in 1998. Similarly, bicarbonate concentrations were the same (n = 21) or higher (n = 2) Fig. 6 Mean major ion concentrations in groundwater wells grouped by ecohydrological classification for all sampling dates (2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017). The error bars correspond to two standard deviations of the mean in the Bradford (1999) study. In contrast, for the 11 wells that had a difference in chloride levels between the two periods, ten of those were lower in 1998. For these wells, the maximum mean concentration (5.6 mg/L, T3-5_D) remained well below the guideline for the protection of aquatic life (120 mg/L, CCME 2011). Only one well (T6-11-S) had a difference in Na concentrations between the two periods, increasing from 4.3 in 1998 to a mean of 11.3 in 2015-2017. Three wells had lower mean Mg concentrations in the current study, with the other 20 wells not significantly different compared to 1998 values.

Nutrients
Excess nutrients in aquatic ecosystems is a priority issue in the lower Great Lakes and intensive agricultural areas of southern Ontario (Robinson 2015), including the Nottawasaga River Watershed, in which the Minesing Wetlands are located. Given that the recharge area for the Minesing Wetlands included potential sources of elevated nutrients (nitrogen and phosphorus species), including agricultural lands and onsite wastewater treatment systems (i.e., septic systems), it was prudent to assess nutrient concentrations as part of the Minesing Wetlands water quality monitoring program. Our water quality analysis included three main nutrient parameters: soluble reactive phosphorus, ammonium, and nitrate (Fig. 8). Nitrate is especially relevant as it is a common groundwater contaminant in areas of intensive agriculture where high nitrogen fertilizer application rates and nitrate's high solubility make it very mobile in groundwater. Soluble reactive phosphorus has historically received less attention as a groundwater contaminant compared to nitrate although recent studies have shown that groundwater can be an important source of phosphorus to surface waters, thereby contributing to the eutrophication of some aquatic ecosystems (Holman et al. 2008;Meinikmann et al. 2015).
Nitrate concentrations in groundwater ranged from below detection (< 0.02 mg N/L) to 1.23 mg N/L for the most recent sampling years (2016-2017; Fig. 8).
In the recharge area, the Pinegrove and Grenfel wells had mean nitrate concentrations of 1.1 mg N/L and < mdl, respectively, with all samples being well below the Canadian Environmental Quality Guidelines for the Protection of Aquatic Life (2.95 mg N/L). The groundwater seeps emanating from the eastern slopes of the study area had the highest nitrate concentrations, albeit still well below guidelines. Presumably, groundwater discharging from the seeps had minimal contact with the near-surface layers where plant roots and microbial activity are more concentrated and could take up nitrate as a nutrient or energy source. Furthermore, compared to the down-gradient wetland areas, minimal contact with the organic-rich soil horizons would mean less opportunity for denitrification activity to reduce nitrate concentrations. A high number of non-detectable nitrate concentrations in the discharge, graminoid, and flow-through systems were most likely due to enhanced plant uptake and denitrification activity in the organic-rich horizons. Redox conditions favorable to nitrate reduction were supported by the fact that sulfate was also low or absent in some of these wells. It is known that wetlands are hot spots for denitrification and therefore the presence of nitrate in the shallow wetland groundwater may be highly episodic. Drought periods that decrease water levels promote oxidation of Fig. 7 Comparison of mean calcium, bicarbonate, and chloride concentrations of the 2015-2017 groundwater samples (circles with 95% confidence interval) and those collected in 1998 (Bradford 1999, diamonds) for individual wells. The red, green, and yellow symbols indicate wells in the discharge, graminoid fen, and flow-through ecohydrological units, respectively wetland organic matter, the mineralization of organic nitrogen to ammonium, and its subsequent nitrification to nitrate. However, once water tables rise again to saturate the organic horizons, anaerobic conditions rapidly develop and nitrate is consumed by denitrification. As such, the timing of sampling (e.g., summer vs. fall/spring) could have been a controlling factor of whether or not nitrate was detected in Minesing Wetlands groundwater. Greater plant uptake of nitrate in the spring and summer (i.e., prior to senescence) could also contribute to seasonal differences in groundwater nitrate concentrations. Dual stable isotope analysis of nitrate provides insights into the sources and cycling of nitrate in the environment. For example, nitrate derived from precipitation is isotopically distinct from nitrate in synthetic fertilizer sources (Kendall 1998;Kendall et al. 2007). Nitrate stable isotopes can also be used to detect the effects of denitrification activity on nitrate concentrations (Kendall 1998;Kendall et al. 2007). Isotopic analysis of groundwater seep samples and Pinegrove well samples collected in 2016-2017 indicate that the nitrate was produced from the nitrification of ammonium, rather than being directly from atmospheric deposition or leached nitrate fertilizer (Fig. 9). The isotope values indicate that the source of the ammonium was likely a combination of nitrogen fertilizer and soil organic nitrogen. The higher δ 15 N values for a few samples were likely the result of denitrification activity rather than a higher contribution of ammonium from a septic or manure source. Denitrification activity is known to increase δ 15 N:δ 18 O values of nitrate in a ratio of approximately 2:1, a trajectory that the two most 15 N-enriched groundwater samples seem to fit (Fig. 9). The red dashed lines correspond to the method detection limit (mdl) of 2 µg/L for SRP and 0.02 mg N/L for nitrate. The number of samples with values below the mdl for ammonium (0.01 mg N/L) and nitrate are indicated for each ecohydrological unit In contrast to nitrate, which is typically the dominant oxidized form of N in aerobic waters, ammonium is often dominant under anaerobic conditions. Groundwater ammonium concentrations ranged from < mdl to 1.5 mg N/L, with the highest values observed in the wetland areas and lower or undetectable levels for the recharge and seep sites. Decomposition of organic matter in the wetland and subsequent ammonification of organic nitrogen compounds was likely the dominant source of the groundwater ammonium. Furthermore, anaerobic conditions in wetland groundwater would inhibit the subsequent nitrification of ammonium to nitrate. With a volumeweighted average precipitation ammonium concentration of 0.25 mg N/L for 2016-17 (calculated from CAPMoN database (Egbert Station), ECCC 2022), atmospheric deposition might also be an important source of ammonium to the Minesing Wetlands.
Phosphorus is typically a limiting nutrient in freshwater ecosystems and elevated concentrations can result in excess algal and macrophyte growth and the subsequent detrimental impacts of eutrophication. The Ontario Interim Provincial Water Quality Objective for total phosphorus (TP) in rivers and lakes is 30 µg/L (MOEE 1994) and is meant to prevent excessive aquatic plant growth. Soluble reactive phosphorus (SRP) is a component of total phosphorus and approximates the amount of phosphorus that is readily available for use by plants and algae. Soluble reactive phosphorus was present in all groundwater samples and concentrations ranged from 8.2 to 122 µg/L (log transformed mean = 38.3 µg/L) for the 2016-2017 samples (Fig. 8), with no significant difference between the various ecohydrological units.
Atmospheric deposition is known to be an important source of phosphorus to nearby Lake Simcoe (Brown et al. 2011) and is likely a significant source of phosphorus to the eastern Minesing Wetlands. Decomposition of wetland organic matter and mineralization of organic forms of phosphorus are also likely to be important. Phosphorus retention by wetlands may include surface adsorption on soil minerals (Zurayk et al. 1997), precipitation in mineral form (Reddy et al. 1987), microbial immobilization (Newbold et al. 1983), and plant uptake (Reddy et al. 1995). Phosphorus is more mobile in the oxygen-depleted groundwater of wetland sediments since iron is likely to be reduced (i.e., Fe 2+ ) and therefore not available in an oxidized form that can bind phosphorus (Baken et al. 2015) and limit its concentration and mobility in groundwater. In a study of total phosphorus loading to the Nottawasaga River, Rutledge and Chow-Fraser (2019) found that total phosphorus concentrations increased at low dissolved oxygen concentrations in Willow Creek, a tributary that runs through the Minesing Wetlands just north of our study site. The findings of Rutledge and Chow-Fraser (2019) are consistent with the Minesing Wetlands being a source of phosphorus to Willow Creek and the Nottawasaga River.

Sodium and chloride
Elevated concentrations of sodium (Na) and especially chloride (Cl) have long been used as indicators  Kendall (1998) and Kendall et al. (2007). The slope of the isotopic shift expected for denitrification of groundwater nitrate is indicated (δ 15 N:δ 18 O = 2:1) of anthropogenic impact to surface and ground waters. In addition to natural, geological sources of sodium chloride, potential anthropogenic sources of elevated sodium and/or chloride in freshwaters include: the application of road salt for de-icing, synthetic fertilizers, sanitary (septic, municipal) and industrial wastewater. Due to its ease of analysis, relatively low cost, and mobility as an anion in groundwater, chloride is often analyzed as a water quality indicator (e.g., Zahn et al. 1993;Sérodes et al. 2021;ECCC and EPA 2022). In Canada, a maximum chloride concentration of 120 mg/L has been set as a water quality guideline for the protection of aquatic life (long-term exposure; CCME 2011).
Sodium and chloride concentrations for the 2016-2017 samples ranged from 1.8 to 27.1 mg/L (mean = 6.3 mg/L) and 1.0 to 18.9 mg/L (mean = 4.4 mg/L), respectively. Two wells (T6-9_D and T6-11_D) and several seeps, all of which were at the eastern edge of the Minesing Wetlands, had higher chloride and/or sodium concentrations compared to the rest of the sites (Fig. 10). Seep samples with chloride > 10 mg/L and sodium < 6 mg/L did not have the concomitant increase in sodium that would be expected from a sodium chloride source such as road salt (see Na:Cl slope in Fig. 10). However, cation exchange with calcium and magnesium in the soil and aquifer matrix could account for this observation. In contrast, T6-9_D and T6-11_D both had more sodium than would be expected from the simple addition of sodium chloride to Minesing groundwater (Fig. 10). It should be noted that sodium and chloride concentrations were still quite low compared to groundwaters contaminated by road salt (e.g., Williams et al. 2000;Howard and Maier 2007;Meriano et al. 2009). Chloride levels in the HED Category 1 habitat (graminoid fen) were lower than the typical background levels for the Lower Great Lakes Region of southern Ontario (10 to 30 mg/L, CCME 2011) and were well below the water quality guideline.

Other contaminants
Using groundwater samples collected in 2010, Van Stempvoort et al. (2016 documented the occurrence of the common herbicide, glyphosate, and its degradate, aminomethylphosphonic acid (AMPA), in the Minesing Wetlands. Approximately 30% of groundwater samples had measureable concentrations of glyphosate, with the maximum concentration (21 ng/L) still well below the Canadian guidelines for drinking water (280,000 ng/L; Health Canada 2021) and for the protection of aquatic life (800,000 ng/L,  -term exposure; CCME 2012). No water quality guidelines exist for AMPA in Canada, however the maximum concentration found was also extremely low (39 ng/L). It was hypothesized that the presence of glyphosate and AMPA in the wetland was due to airborne transport from application to nearby roadsides or agricultural lands (Van Stempvoort et al. 2016). Glyphosate and AMPA detections in wetonly precipitation in the Nottawasaga River Watershed supported atmospheric deposition as a source of these chemicals to the wetland (Van Stempvoort et al. 2016).
The Minesing Wetlands groundwater samples collected in May and November 2017 were also analyzed for several wastewater tracer compounds, namely four artificial sweeteners (acesulfame, saccharin, cyclamate, sucralose) and an emerging environmental contaminant, sulfamate (Van Stempvoort et al. 2019). Artificial sweeteners, particularly acesulfame, are ubiquitous in sanitary wastewater due to their widespread usage and resistance to degradation, and therefore are increasingly being used as powerful tracers of wastewater in the environment (e.g., Buerge et al. 2009;Roy et al. 2014;Spoelstra et al. 2020;Van Stempvoort et al. 2020). Saccharin was the only artificial sweetener found in Minesing Wetlands groundwater. Given the low saccharin concentrations, and the absence of other artificial sweeteners, Van Stempvoort et al. (2019) concluded that the saccharin likely originated from an undetermined, non-wastewater source. Sulfamate is formed naturally in the atmosphere and is also widely produced and used by industry and therefore occurs in precipitation and wastewater (Castronova et al. 2017). Sulfamate concentrations in Minesing Wetlands groundwater (< 25 to 65,000 ng/L) were generally within the range found for atmospheric deposition (< 2,600 ng/L) (Van Stempvoort et al. 2019) and therefore did not indicate impact from wastewater (e.g., septic system effluent).

Conclusions
The major ion chemistry of groundwater sustaining Canada's only known HED population was dominated by calcium and bicarbonate, consistent with other HED habitats previously characterized in the USA. Although recent studies in the Minesing Wetlands have documented relatively low concentrations of some organic contaminants in the shallow groundwater, we did not find evidence of significant changes in the major ion chemistry of the groundwater between the original study (1998 sampling) and the latest sampling periods of this study (2015)(2016)(2017). Chloride and nitrate concentrations that were well below guidelines indicate that anthropogenic impacts from activities such as road salt application and nitrogen fertilizer use in the contributing recharge area, were minimal. Protection of the Hine's emerald dragonfly habitat in the Minesing Wetlands would benefit from an ongoing groundwater monitoring program designed to detect changes in both water quality and quantity that may occur due to changes in land use in the recharge area as well as the impact of local climate change.
Acknowledgements The authors acknowledge the work of several individuals that contributed to this study. Michael Saunders, Nottawasaga Valley Conservation Authority (NVCA) and Ross McKay (Environment and Climate Change Canada) assisted with groundwater field work. Geochemical analysis was completed by Sue Brown and Pam Collins, Environment and Climate Change Canada. Funding for this project was provided by the Ontario Ministry of Natural Resources and Forestry's (MNRF) Species at Risk Research Fund for Ontario (SARRFO) and Environment and Climate Change Canada. We also thank an anonymous reviewer that helped improve a previous version of this article.

Author contributions
Both authors contributed to the study design, field work, data interpretation, and manuscript preparation. Both authors read and approved the final manuscript.

Data availability
The geochemistry data reported in this manuscript is available in Table S3 of the Supporting Information and will be also available from Environment and Climate Change Canada's Data Catalogue (https:// catal ogue. ec. gc. ca/). Additional data used in this manuscript can be requested from the corresponding author.

Declarations
Competing interests The authors have no relevant financial or non-financial interests to disclose.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative