Wet–dry cycles influence methylmercury concentrations in water in seasonal prairie wetland ponds

Our three study sites were located within and near the St. Denis National Wildlife Area (SDNWA; 52°12.537′N, 106°5.240′W; Fig. 1) (Government of Canada 2022), part of the PPR of North America that extends from northwestern Alberta to northwestern Iowa. This region is characterized by hummocky glacial till, aspen parkland terrain and contains numerous small, depressional mineral soil wetlands. Annual precipitation at the site based on local observations from precipitation gauges and snow surveys (Bam et al. 2019) for the period 1994–2017 is approximately 360 mm·year⁻¹, with around 80 mm·year⁻¹ falling as snow. The locally measured monthly mean air temperature for the same period varies between −15.8 °C (in January) and 17.9 °C (in July).

We sampled 28 ponds classified using the Circular 39 classification system of Shaw and Fredine (1956) as temporary ponds with the duration of ponded water on the order of weeks (Type III), months or seasons (Type IV), and permanent sites with water year-round regardless of periods of drought (Type V wetlands; Table S1). Although one pond (Pond 66) was mesohaline (conductivity, which we used as a proxy for salinity, = 11 mS·cm⁻¹), the majority of ponds were either freshwater (<0.5 mS·cm⁻¹; n = 10) or oligohaline (0.5–5 mS·cm⁻¹; n = 17; Table S2). Sixteen ponds sampled were located in the eastern section of the SDNWA, which consists of a ring of semipermanent ponds at low elevations and an upland area that is 5–10 m higher in elevation. Low elevation ponds (Ponds 1, 2, 3, 65, 66, 67, 88, 97, 125 N, 130, and 139) have higher solute concentrations, whereas upland ponds (Ponds 100, 104, 110, 113, and 118) have lower solute concentrations (Kiss and Bedard-Haughn 2021), likely due to lower groundwater inputs. Three ponds were located in the western section of SDNWA (Ponds 26, 50, and 60). Six of our wetland ponds (Cultivated 1–3 and Grass 1–3) were located on private lands ∼2 km west of the SDNWA. Three additional sites (Organic 1, 2, and 4) were located on private land east of Vonda, SK ∼7.0 km north of the SDNWA (Fig. 1).

Prior to 1968, the entire SDNWA area was cultivated by local land owners. In 1977, land surrounding the wetlands in the low elevation area of the SDNWA was converted to brome grass (Bromus inermis Leyss.), alfalfa (Medicago sativa L.) and yellow sweet clover (Melilotus officinalis L.). Conversion to these grasses at the lower elevations was complete by 1983 (Conly and van der Kamp 2001). In May 2004, approximately half of the upland area (adjacent to Ponds 100, 104, and 118) was converted to a mix of perennials consisting mainly of intermediate wheatgrass (Elymus intermedius M. Bieb) and meadow brome (Bromus biebersteinii Roem. and Scult.). The other half (adjacent to Ponds 110, 113, and 118) was cultivated land and planted with flax (Linum usitatissimum; 2006), oats (Avena sp.; 2007), barley (Hordeum vulgare L.; 2008 and 2010), and canola (Brassica campestris L.; 2009, 2011). In years not listed, the land was in fallow.

For wetland ponds located on private lands immediately outside of and adjacent to the SDNWA, three ponds (Cultivated 1–3) were surrounded by cultivated fields planted with legumes in 2007 and a cereal crop in 2008 and 2009 and three ponds (Grass 1–3) were surrounded by grassland previously returned to fallow from cultivation practices in 1995 (Table S1, Fig. 1). Flora surrounding ponds in the grassland area consisted of aspen groves (Populus tremuloides Michx.), willows (Salix spp.), Medicago sativa, white sweet-clover (Melilotus alba Medic.), Canada thistle (Cirsium arvense (L.) Scop.), goldenrod (Solidago canadensis L.), and mixed tame and native grasses. The land adjacent to the three ponds north of the SDNWA (Organic 1, 2, and 4; Fig. 1) was managed as certified organic since 1985. Two Organic ponds (Organic 1 and 2) were adjacent to a crop of barley in 2007, plow down clover in 2008, and red fife wheat (Triticum aestivum L.) in 2009. Our third Organic pond (Organic 4) was surrounded by a plow down pea crop (Pisum sativum L.) in 2008 and red fife wheat in 2009. In dry years, cultivation occurred in the depressions left by Organic 2 and 4. With a few exceptions (Organic 2 and 4), all ponds were surrounded by some riparian vegetation, consisting of mostly willows and grasses.

Samples were taken from ponds located in areas characterized as falling into four land use categories—land converted to a perennial grass cover in 1977 (Grassland) and 2004 (Recent Grassland), land cultivated using nonorganic methods (Cultivated), and land cultivated using certified organic methods (Organic). Throughout this manuscript, land use categories refer to the type of land use on the watershed supplying immediate runoff to the ponds. All Type IV and V ponds (1, 26, 50, 65, 66, 67, 97, and 125 N) were surrounded by land classified as both Grassland and Cultivated land, and as such we used only Type III wetlands in our analysis of the impact of land use (see below).

Sampling took place between 2006 and 2012 (Table S3). In 2006, we performed an initial survey examining whole water [MeHg] in 19 ponds throughout the SDNWA. Beginning in 2007, efforts were concentrated on the ponds with different land use on private lands. From 2008 onwards, we focused on Type III ponds in the SDNWA for which we had water level data (see below). Private wetland ponds were not sampled after 2009.

Surface water sampled for Hg analysis was collected using trace metal-free techniques from open water areas of the ponds beyond the fringe of riparian vegetation (St. Louis et al. 1994; Olson et al. 1997). Samples were taken by hand in shallow areas by wading into the open water area and avoiding the plume or from a fibreglass canoe or inflatable boat in deeper water. Samples were visually inspected for large particles and if present the sample was taken again. Total mercury (THg; all forms of Hg) samples were collected in pre-cleaned glass bottles and preserved using trace metal grade concentrated HCl to 1% of total sample volume. Until 2011, MeHg samples were taken in sterile fluorocarbon polymer and, as was standard in the early years of our sampling efforts, were immediately placed on ice and frozen within 48 h. In 2010, we switched to pre-cleaned glass bottles for collection and acidification similar to THg samples. On two occasions in 2007, MeHg and THg were obtained from dissolved and particulate phase of samples from seven sites. After collection as outlined above, samples were filtered into clean bottles using 0.45 µm muffled quartz fibre filters and Teflon filter holders. The filtrate was preserved in the same manner as whole water samples. Filters loaded with particulate samples were immediately frozen on dry ice.

Methylmercury concentrations were determined by cold vapour atomic florescence spectrometry (CVAFS) after distillation, ethylation, and separation (Bloom 1989; Horvat et al. 1993; Liang et al. 1994). Detection limits for MeHg in water were between 0.01 and 0.05 ng·L⁻¹. Matrix spike recoveries for MeHg were generally >80% and >90%. Total Hg was analyzed by CVAFS after BrCl oxidation and SnCl₂ reduction following EPA Method 1631 (U.S. EPA 2002). Recoveries of THg from a certified standard reference material (National Research Council DORM-3) run concurrent with samples were between 100% and 102%. Average THg recoveries from sample duplicates spiked with a known concentration of a Hg standard were 90%–113% with a mean of 99%. Detection limits were 0.01–0.05 ng·L⁻¹. Approximately 10%–15% of the samples were analyzed in duplicate and average relative differences among duplicates were 2.19% and 3.30% for THg and MeHg, respectively.

Ponds were sampled for temperature, dissolved oxygen concentrations, conductivity, and pH using a YSI probe. Samples for DOC and SO₄⁻² concentrations were taken from water filtered through 47 mm glass fibre filters and analyzed using standard methods (Stainton et al. 1977). Specific ultraviolet absorbance (SUVA) at λ = 254 nm (an indication of aromatic moieties in a sample and thus of the quality of the DOC) was determined as described in Hall et al. (2008).

Water levels in ponds at the SDNWA have been monitored from 1968 using an approach presented in Conly et al. (2004). Briefly, a rod was inserted at the lowest point of the wetland pond to be used as a consistent point of reference for subsequent measurements. During the open water season (typically April–October), a measuring rod with a small circular metal base and graduated to 1 mm accuracy was used to measure the water level from the bottom of the pond. Measurements were typically taken monthly, although in some years measurements were taken only on a few occasions and in others as often as biweekly (Conly and van der Kamp 2001). To assess the impact of water levels, we narrowed our data set to only Type III wetlands at the SDNWA (Table 1) and examined water levels over the period 2000–2012. We chose this time period because it contained both a period of drought in the early part of the decade (2000–2004; Hanesiak et al. 2011) with subsequent wetter conditions (2005–2012). During the drought, the vast majority of Type III ponds held no water over the entire ice-free period (Table 1, Fig. S1). In 2005, a heavy spring melt resulting from larger than normal snow fall over the preceding winter (Pennock et al. 2010) increased water levels in most ponds and allowed us to sample these sites. Wet conditions due to greater than average spring precipitation persisted in the latter part of the decade and from 2006 to 2012, all ponds, with the exception of Pond 104 in 2008, contained water for some portion of each ice-free season. Based on the water level data, we categorized two types of ponds: (i) “Mainly Wet” ponds (Ponds 2, 3, 60, 88, 130, and 139; Fig. S1a) which either did not dry up at all over the course of each summer or stayed wet until October of each year from 2005 to 2012 and (ii) “Mainly Dry” ponds (Ponds 100, 104, 110, 113, and 118; Fig. S1b) which dried up in the earlier parts of each summer during 2005–2009 (Table 1). There were a few exceptions with Ponds 100, 110, and 118 tending to be wetter longer into the season in 2010–2012. Water level data were not available for the ponds located on private lands, and as such were not included in the statistical analysis on the impact of water levels. All of the ponds in the Mainly Wet category, with the exception of Ponds 60 and 88, were classified as Grassland sites (Table 1). In the Mainly Dry category, all of the sites were either Recent grass or Cultivated sites. We note that even during our “wet” periods, fill and spill conditions were still relatively isolated to freshet and/or large storm events. Finally, we do not have water level data for ponds outside the SDNWA; as such ponds in the Mainly Dry and Mainly Wet categories do not include Organic ponds.

Average concentrations of MeHg, THg, DOC, and SO₄⁻² and the proportion of THg that was MeHg (%MeHg) were calculated for each site by combining pond data from different sampling dates, thus, ponds were our experimental units. Forward multiple linear regressions were performed on data to determine which variables predicted [MeHg]. Variables included temperature, pH, conductivity, dissolved oxygen, SO₄⁻² and DOC concentrations, SUVA values, land use, wetland type, and hydrological status. Conductivity, MeHg, and SO₄⁻² concentrations were log transformed to meet requirements of normality, and to address the large range in conductivity and [SO₄⁻²] (correlations’ comparisons of both original and log variable are presented in Table S4). Smallest Akaike Information Criteria (AIC) and Bayesian Information Criteria were used for best-fit model selection. Data analysis was performed using R statistical software (R Development Core Team 2013), with the dplyr (Wickham et al. 2021) and Mu-MIn (Barton 2009) packages. t tests and one-way analyses of variance were used to test differences in DOC and log SO₄⁻² concentrations and SUVA with land use category and hydro-status.

Average MeHg and total Hg concentrations (all forms of Hg, [THg]) in unfiltered water from individual ponds ranged from 0.123 to 9.947 ng·L⁻¹ and 0.57 to 7.32 ng·L⁻¹, respectively (Figs. 2A and S2, Table S5). The %MeHg values ranged from 3.2% to 64.5% (Fig. 2B, Table S5). All of these measurements were similar to values observed in other Saskatchewan ponds in 2006–2007 (Hall et al. 2009) and in North Dakota in 2003–2004 (Sando et al. 2007), with the exception of Organic 2, where average [MeHg] in water was significantly higher than all other ponds in this study (9.947 ± 5.054 ng·L⁻¹). Total Hg concentrations in Organic 2 (7.31 ± 2.15 ng·L⁻¹) were also elevated compared to other sites; however, the difference was not as striking as with [MeHg].

Shallow ponds in this area are often turbid due to shallow depths, loose sediments, and constant mixing due to wind (de la Cruz 1979; cited in Murkin 1989). Since we did not filter the majority of our samples, and although we made every effort to exclude particulates when sampling, the amount of particulate matter in some samples was substantial, potentially leading to %MeHg values that were greater than 100%. To assess the general contributions of particles to overall whole water concentrations, MeHg and THg in both the dissolved and particulate phases were examined from seven sites on two occasions in 2007 (Table S6). In six out of seven sites, the percentage of MeHg and THg in the dissolved phase ranged from 73.0% to 98.4% and 85.7% to 99.7%, respectively. These results are similar to ponds in North Dakota where the majority of MeHg was found in the dissolved phase (Sando et al. 2007) or in the colloidal phase that may pass through 0.45 µm filters (Hsu Kim et al. 2013). In Organic 2, 53% of MeHg and 62% of THg was in the particulate phase (Table S6). This site was shallow with extremely fine and flocculate sediments and it was impossible to avoid large particulates in the sample and, as such, increased particulate matter in Organic 2 may explain high whole water concentrations at that site. Variability within whole water samples in Organic 2 was large and %MeHg values ranged from 13.5% to 177.3%. Concentrations of MeHg and THg in insects were also high at this site compared to others (Bates and Hall 2012), suggesting bioaccumulation in the invertebrate food chain was occurring in that pond. We reiterate a conclusion presented in Hoggarth et al. (2015) that, from a management perspective, measuring MeHg concentrations, as well as the %MeHg, in surface water may be sufficient to predict and assess MeHg risks to biota.

Forward multiple linear regression on data from Type III ponds determined period of inundation (hydro-status), SO₄⁻² and DOC concentrations, SUVA values, conductivity, pH, temperature, and Organic and Grassland land use to be significant variables predicting MeHg concentrations (Table 2). Model selection was based on AIC values (Table S7) and Quantile–Quantile plots of residuals support appropriate fitting of the model. Coefficients were as follows: log [MeHg] in whole water (ng·L⁻¹) = −0.821 – (0.558*Mainly Wet) + (0.040* temperature (°C) − (0.647*log conductivity (mS·cm⁻¹)) − (0.435*pH) + (0.364*log SO₄⁻² (mg·L⁻¹)) + (0.007*DOC (mg·L⁻¹)) + (0.288*SUVA (L mg·C⁻¹·cm⁻¹)) + (0.825*Organic land use) + (1.016*Grassland use) + ε.

Considering that our ponds were all within 3 km of each other, the variation of water chemistry parameters (Table S2) may be surprising if unfamiliar with the high temporal and spatial variability of these prairie wetland systems (Euliss and Mushet 1996; Nachson et al. 2014). Concentrations of DOC and SO₄⁻² (range = 10.06–188.20 mg·L⁻¹ and 0.04–8168 mg·L⁻¹ for DOC and SO₄⁻², respectively), conductivity (range = 0.168–11.220 mS·cm⁻¹), and pH (range = 6.1–9.4) varied widely among ponds over the course of our sampling period. Similarly water temperature varied (range = 7.64 °C–32.11 °C) likely as a function of the difference in depth and climatic conditions over the open water season. Despite the variability, forward multiple linear regression on data from Type III ponds identified SO₄⁻² and DOC concentrations, SUVA values, conductivity, pH, and temperature to be significant variables predicting MeHg concentrations (Table 2). Predictive relationships between [MeHg] in whole water and both SO₄⁻² and DOC concentrations and SUVA is not surprising because previous studies have demonstrated that these parameters may control MeHg production by providing necessary substrates to methylating communities (Benoit et al. 1999; Aiken et al. 2003; Gorski et al. 2008; Hall et al. 2008; Janssen et al. 2022) (see below). A positive relationship between conductivity and MeHg reflects the fact that most of our systems are SO₄⁻² dominated. As methylation is by in large a microbially mediated process, we are also not surprised that temperature predicated [MeHg]. Finally, expectations on the processes in which pH may influence [MeHg] are less clear with early work showing inverse relationships between pH and Hg concentrations in fish (Winfrey and Rudd 1990) and that increasing concentrations of H⁺ facilitates the uptake of Hg by methylating organisms (Kelly et al. 2003). We note, however, that these findings were made in low pH lakes. Studies on the impact of pH in less acidic lakes are rare.

Average MeHg concentrations in water were higher in Type III wetland ponds surrounded by organic farming (p < 0.001) and grasslands (p < 0.001) compared to those surrounded by cultivated lands or lands that had been recently converted to grass (Fig. 3). The significant difference in [MeHg] and %MeHg in the Organic land use group could have been driven by data from Organic 2, which had high concentrations due to high particle loads (see above). Concentrations of THg in ponds surrounded by cultivated land were significantly lower than in ponds with other land use (p = 0.002; Fig. S3).

We present two explanations for why land use might indirectly influence [MeHg] in these environments; both are based on the finding that DOC concentrations were significantly lower in Cultivated ponds (Fig. S4A). First, because organic matter may influence rates of Hg methylation, impacts of land use on DOC mobilization to ponds may increase net MeHg concentrations in waters. For example, ponds with grassland riparian areas may have increased organic matter inputs due to heavily vegetated shorelines compared to ponds where the riparian vegetation is removed in an annual harvest. Ponds surrounded by organic farms may have increased organic matter inputs due to the use of green manures as fertilizers (Lundquist et al. 1999).

Second, ponds surrounded by grasslands tend to have increased hydraulic conductivity due to increased infiltration rates, macroporosity, and evapotranspiration in the riparian zone (Conly and van der Kamp 2001; Bodhinayake and Si 2004). As a result, infiltration of summer precipitation and snowmelt is likely to be higher and runoff smaller in undisturbed grasslands compared to cultivated lands. Increased upland infiltration may result in increased input of porewaters with high MeHg and DOC concentrations, or DOC that is more chemically complex. However, in very wet periods, water fluctuations tend to be smaller for grasslands wetlands compared to cultivated wetlands (Euliss and Mushet 1996). In fact, the conversion of cultivated lands at the SDNWA to grass in the 1980s resulted in a decrease in water levels in grassland ponds compared to cultivated ponds in subsequent years (van der Kamp et al. 1999, 2003).

Ideally, we would have liked to examine differences in [MeHg] among Type III, IV, and V ponds to determine if length of hydroperiod controlled concentrations. However, samples numbers in the more permanent ponds (Type IV and V) were small and thus we restricted our analysis to Type III ponds. Average [MeHg] and %MeHg in ponds classified into the Mainly Wet category were significantly higher than ponds classified to the Mainly Dry category (t test, [MeHg] p < 0.001 and %MeHg p < 0.001; Fig. 4). There was no significant difference in average [THg] in ponds with different hydro-status (p = 0.479; Fig. S5).

The high variability of hydrologic cycles that result in the alternation of wet–dry conditions of prairie wetlands may result in unusually high rates of MeHg production. We identify two strong possibilities to explain higher [MeHg] in ponds that had significant periods of flooding following years of drought conditions (Mainly Wet ponds). First, rewetting sediments increases the amount of labile DOC due to decomposition of flooded soils and vegetation growth in dried wetlands during periods of drought (Lundquist et al. 1999; Graham et al. 2013). We observed higher DOC in our Mainly Wet ponds (range = 15.0–181.8, mean = 63.9) than in the Mainly Dry ponds (range = 10.6–158.0 mg·L⁻¹, mean = 50.0 mg·L⁻¹; p = 0.030) and this correlated with [MeHg] (Figs. 4 and 5A). As well, infrequent, yet major, hydrological events (including snow melt and intense precipitation events) may result in temporary flooding of soils and terrestrial or riparian vegetation leading to increased DOC concentrations (cited in Brothers et al. 2014; Raymond et al. 2016; Pang et al. 2021). As such, because flooding of organic matter stimulates MeHg methylation (Hall et al. 2005; Windham-Myers et al. 2014; Eckley et al. 2015; Herrero Ortega et al. 2018), this high-quality, labile DOC can support increased microbial activity and Hg bioavailability to cells (Kelly et al. 1997; Hall and St. Louis 2004; St. Louis et al. 2004; Bravo and Cosio 2020). In a study examining the impact of the length of hydroperiod on [MeHg] on similar types of prairie ponds at Lostwood Reserve in North Dakota, Sando et al. (2007) found that whole water [MeHg] and %MeHg values were generally higher in seasonal and semipermanent wetlands compared to those from temporary and permanent wetlands. This was attributed to increased methylation due to the stimulation of the anaerobic microbial community following reflooding of dried ponds. Similar studies in the Everglades have also shown increases in [MeHg] in response elevated oxidized S due to wet–dry cycles (Krabbenhoft and Fink 2000; Janssen et al. 2022).

Second, water-level declines expose anoxic wetland sediments or inundated soils to oxygen, thereby changing redox conditions (Coleman Wasik et al. 2015) resulting in the oxidation of sulfide to SO₄⁻², a required substrate for SRB methylators. After such, rewetting and re-establishment of anoxic conditions, may stimulate MeHg production rates (Eckley et al. 2015). At the SDNWA, Hoggarth et al. (2015) found that ponds with elevated SO₄⁻² concentrations had elevated MeHg concentrations compared to ponds with lower SO₄⁻² concentrations, but found no significant difference in potential rates of MeHg production with the use of stable Hg isotope sediment incubations. The elevated SO₄⁻² sites in Hoggarth et al. (2015) correspond to our Mainly Wet ponds which had SO₄⁻² concentrations ranges that were an order of magnitude higher than those in the Mainly Dry ponds (range = 0.19–2184 and 0.04–176 mg·L⁻¹ and mean = 703.5 and 25.9 mg·L⁻¹, respectively, Fig. 5B; p < 0.0001 for log [SO₄⁻²]). A number of studies (Benoit et al. 1999; Li and Cai 2012; Poulin et al. 2017; Jones et al. 2020; Huang et al. 2022; Janssen et al. 2022) have led to the development of a general paradigm that lakes or wetlands with lower concentrations of SO₄⁻² have lower SRB activity, and therefore less MeHg production, whereas those with higher SO₄⁻² concentrations result in inhibition of methylation due to formation of insoluble Hg–sulfide complexes not bioavailable to methylating organisms (Gilmour et al. 1998). Although we suspect high sulfide concentrations may have been present based on occasional H₂S odours and visual observation of black sediments, our data suggest that our sites may not fit this paradigm because Hoggarth et al. (2015) did not observe inhibition of potential methylation rates, and [MeHg] in sites with elevated SO₄⁻² were high compared to those in sites with lower SO₄⁻² concentrations. Prolonged flooding of the saline ring formed around the typical pond margin could also release SO₄⁻² into the pond while water levels are high (Mushet et al. 2018). Rates of SO₄⁻² reduction can be rapid in prairie wetland sediment with turnover of the porewater SO₄⁻² pool in hours to days (Dalcin Martins et al. 2017).

To further investigate the impact of S species on MeHg concentrations in our ponds, we used traditional geochemical speciation modelling. In the absence of details on DOC characterization, historic data from 18 ponds at SDNWA (Bam et al. 2019) were input into the geochemical modeling software Visual MINTEQ ver. 3.1 (Gustafsson 2014) to determine the dominant S species in anoxic bottom waters of our ponds. At pH = 8, models suggest that Hg species were dominated by small, ionic S containing species such as HgSH₂⁻ and HgS₂⁻² (Fig. S6). Under the assumption that bioavailable Hg enters methylating cells passively (which may or may not be the case; Hsu-Kim et al. 2013), ionic Hg species tend to be less bioavailable, and if our sites fit the accepted paradigm of inhibition of methylation in high S systems, we would have expected to see lower concentrations of MeHg in high SO₄⁻² sites. Higher MeHg concentrations in high SO₄⁻² sites suggest that the formation of HgS complexes are not rate limiting for MeHg production. It is possible that DOC is outcompeting sulfide species for Hg binding and facilitating MeHg production in high DOC, high SO₄⁻² sites.

The chemical composition of DOC also influences MeHg cycling in aquatic systems, in part because DOC and S species compete for Hg binding and thus controlling the ability of Hg to enter methylating cells (Miller et al. 2007). Highly complex aromatic moieties in DOC measured using SUVA mobilizes both inorganic and organic Hg facilitating increased inorganic Hg methylation and increased MeHg mobilization from sediments into other components of aquatic systems (Aiken et al. 2003, 2011). In our sites, SUVA ranged from 0.43 to 4.9 L mg·C⁻¹·cm⁻¹ and were not significantly different in Mainly Wet compared to Mainly Dry ponds (t test, p = 0.404 for SUVA; Fig. 5A). The range in SUVA values in our sites was similar to those in dark water systems that had significant relationships between MeHg concentrations and SUVA values (Louisiana wetlands range = 2.6–4.0 L mg·C⁻¹·cm⁻¹ (Hall et al. 2008), boreal lakes = 1.4–4.8 L mg·C⁻¹·cm⁻¹ (Bravo et al. 2016), and European lotic systems = 2.4–5.2 L mg·C⁻¹·cm⁻¹). The highest SUVA values in our sites were higher than those in North Dakota (1.2–2.1 L mg·C⁻¹·cm⁻¹ (Holloway et al. 2011)). Prairie wetland ponds typically experience intense irradiation due to long summer days, lower cloud cover, and reduced overhanging riparian or floating vegetation. As a result, due to intense photodegradation and despite high DOC concentrations, waters in many prairie ponds are not highly coloured and contain less aromatic and labile DOC (Waiser and Robarts 2004); however, some of our ponds do not follow this trend. Finally, more complex DOC (more complex DOC could be more aromatic, have more complicated active side groups, and be less bio- or photodegraded) in our Mainly Wet ponds may inhibit photolytic demethylation to a greater extent compared to less complex DOC in our Mainly Dry ponds, facilitating increased net MeHg concentrations in the water.

Investigating possible explanations on why prairie wetland ponds may not fit the accepted model is on-going in our lab, including the identification of putative methylating and demethylating microbes in ponds with differing MeHg properties and on-going studies investigating how DOC with different in optical characteristics may clarify the role of DOC composition in Hg cycling in SDNWA ponds.