Evaluating the influence of road salt on water quality of Ohio rivers over time

Applied Geochemistry 47 (2014) 25–35

Contents lists available at ScienceDirect

Applied Geochemistry journal homepage: www.elsevier.com/locate/apgeochem

Evaluating the influence of road salt on water quality of Ohio rivers over time Kelsey R. Dailey a,b,⇑, Kathleen A. Welch b, W. Berry Lyons a,b a b

School of Earth Sciences, The Ohio State University, 275 Mendenhall Lab, 125 S. Oval Mall, Columbus, OH 43210, USA Byrd Polar Research Center, The Ohio State University, 108 Scott Hall, 1090 Carmack Rd, Columbus, OH 43210, USA

a r t i c l e

i n f o

Article history: Available online 17 May 2014 Editorial handling by M. Kersten

a b s t r a c t Anthropogenic inputs have largely contributed to the increasing salinization of surface waters in central Ohio, USA. Major anthropogenic contributions to surface waters are chloride (Cl ) and sodium (Na+), derived primarily from inputs such as road salt. In 2012–2013, central Ohio rivers were sampled and waters analyzed for comparison with historical data. Higher Cl and Na+ concentrations and fluxes were observed in late winter as a result of increased road salt application during winter months. Increases in both chloride/bromide (Cl /Br ) ratios and nitrate (N-NO3 ) concentrations and fluxes were observed in March 2013 relative to June 2012, suggesting a mixture of road salt and fertilizer runoff influencing the rivers in late winter. For some rivers, increased Cl and Na+ concentrations and fluxes were observed at downstream sites near more urban areas of influence. Concentrations of Na+ were slightly lower than respective Cl concentrations (in equivalents). High Cl /Br mass ratios in the Ohio surface waters indicated the source of Cl was likely halite, or road salt. In addition, analysis of 36Cl/Cl ratios revealed low values suggestive of a substantial dissolved halite component, implying the addition of ‘‘old’’ Cl into the water system. Temporal trend analysis via the Mann–Kendall test identified increasing trends in Cl and Na+ concentration beginning in the 1960s at river locations with more complete historical datasets. An increasing trend in Cl flux through the 1960s was also identified in the Hocking River at Athens, Ohio. Our results were similar to other studies that examined road salt impacts in the northern US, but a lack of consistent long-term data hindered historical analysis for some rivers. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Many natural and anthropogenic factors affect the geochemistry of surface waters in both rural and urban areas. Much of the impact on fresh water quality in the US comes from non-point sources, like road and agricultural runoff, with population and land use playing an important role in determining water quality. The salinization of fresh water occurs by means of surface runoff related to agricultural practices and increasing suburban and urban coverage of roadways. Major components of anthropogenic inputs into rivers and streams are chloride (Cl ) and sodium (Na+), derived primarily from modern contributions of halite, or road salt, whereas anthropogenic input of nitrate largely comes from agricultural fertilizers and treated wastewater. The application of salt to highways and roads lowers the freezing point of water, thereby ⇑ Corresponding author at: School of Earth Sciences, The Ohio State University, 275 Mendenhall Lab, 125 S. Oval Mall, Columbus, OH 43210, USA. Tel.: +1 (440)667 4078. E-mail addresses: [email protected] (K.R. Dailey), [email protected] (K.A. Welch), [email protected] (W.B. Lyons). http://dx.doi.org/10.1016/j.apgeochem.2014.05.006 0883-2927/Ó 2014 Elsevier Ltd. All rights reserved.

melting snow and ice, but also creates saline water that runs off roadways into the local environment. This saline water is difficult to control and can travel a variety of pathways, flowing into the surrounding rivers and streams and infiltrating into the ground. Consequently, high-quality fresh water resources decrease, and roadside terrestrial habitats and aquatic ecosystems are degraded. Aesthetic effects arising from the increased salinization of water resources, such as salty taste, coupled with chloride’s ability to corrode steel and damage pipes can result in increased treatment costs for drinking water. Elevated Cl concentrations in natural waters can degrade bridges, road beds, and other infrastructure as a result of its corrosive nature (Transportation Research Board, 1991). The acidification of streams, alteration of the mortality and biodiversity of aquatic biota, and the mobilization of toxic trace metals due to organic matter mobilization and ion exchange can all result from the salinization of fresh waters (Heath et al., 1992; James et al., 2003; Amrhein and Strong, 1990). After World War II, large-scale implementation of road salt use began in the US by replacing abrasives as the main agent in clearing roadways. Coupled with the expanding US highway system at this time, salt usage increased quickly in the 1950s and 1960s,

26

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

leveling off in the 1970s when the conversion from abrasives to road salt was nearly complete and highway agencies began implementing better salt application practices as salt’s adverse effects were recognized (Transportation Research Board, 1991). Although elevated levels of Cl and Na+ have become more common in surface and ground waters in northern regions of the US, road salt is presently not regulated as a primary contaminant to fresh waters (USEPA, 1988). While no limit for Na+ concentration exists, the USEPA has only a secondary standard of 250 mg/L for Cl in drinking water to account for salty taste and a recommended chronic criterion of 230 mg/L Cl for at least 4 days in surface waters for aquatic life (USEPA, 2002; USEPA 1988). Surface waters that are exposed to abundant halite dissolution may present a different type of pollution not specifically covered by environmental policies and regulations. As a result, road salt use in the northeastern, midwestern, and Great Lakes region of the US and its impact on the quality of surface water and groundwater has been examined by a number of investigations in recent years (e.g. Chapra et al., 2009; Foos, 2003; Kaushal et al., 2005; Kelly et al., 2012; Koryak et al., 2001; Mason et al., 1999; Panno et al., 2006). The rate of salt usage for highway deicing in the US has increased yearly by about 2–3% since the late 1970s, while Ohio had an annual road salt application rate of 200 kg/ha in the 1990s (Chapra et al., 2009; Kelly et al., 2012). The Ohio Department of Transportation (ODOT) uses around 700,000 tons of road salt per year statewide, and the city of Columbus has used about 23,000 tons yearly over the past decade (ODOT, 2008; City of Columbus, OH, 2014). During the 2012–2013 winter season, nearly 22,000 tons of salt were used by the city, compared to less than 6000 tons used during the previous, milder winter of 2011–2012 (City of Columbus, OH, 2014). Previous work in Ohio has also demonstrated the detrimental environmental impact of point source pollution related to road salt. The Ohio EPA has found evidence in five Ohio communities of runoff from road salt storage piles contaminating public and private wells since 2009. Wells had to be abandoned in Preble County in southwest Ohio when the salty taste in drinking water could no longer be tolerated (Ohio EPA, 2011). In central Ohio, drinking water concerns pertinent to storm water runoff from salt piles has recently lead state organizations to create road salt storage guidelines for businesses and cities. Although these are not laws and have not been officially adopted, the guidelines request that road salt be stored outside of flood prone areas, a minimum of 90 m from streams and wells, and at least 30 m from ditches and storm drains (Ohio Water Resources Council, 2013). The metropolitan Columbus, Ohio, area has seen population growth in the past two decades, resulting in urbanization and suburban sprawl. Little is known about the correlation between urban and suburban development across the US and its impact on longterm changes in baseline salinity of fresh surface water. Historical data on fresh water quality exist for many rivers and streams throughout Ohio, however much of it has never been utilized to characterize trends through time. Due to chloride’s conservative nature, low natural background levels, and ability to be measured with reasonable precision, it serves as a reliable indicator of anthropogenic influence on water quality (Chapra et al., 2009). The overall goal of this research was to investigate the application of road salt, or halite, as the main source of Cl input to the Ohio surface waters. Additionally, possible anthropogenic induced trends in Cl and Na+ concentration through time were evaluated in a number of Ohio rivers that are located in areas of urban and suburban influence. Rivers studied include the Hocking River, Olentangy River, Little and Big Darby Creeks (Darby Creek), Alum Creek, and Little Miami River (Fig. 1).

2. Methods Water quality data from two main sources were tabulated to identify long-term changes in ion concentrations in rivers at multiple locations throughout central Ohio. Historical water chemistry and flow data were obtained from the US Geological Survey (USGS) National Water Information System. Previously unpublished water chemistry data from the Lyons research group for the 2000s at The Ohio State University (OSU) School of Earth Sciences were also utilized in this study. A total of 14 sites including 12 sites examined in the past corresponding to USGS gauge stations and 2 new sites (OA, HA) were sampled in June 2012 and March 2013 (Fig. 1; Appendix A). Not every site sampled in 2012 was sampled again in 2013 either due to difficulties in reaching the water because of ice and snow or due to a lack of long-term consistent historical data for the sample site. For this study, sample locations have identifiers beginning with O, A, L, D, and H for the Olentangy River, Alum Creek, Little Miami River, Darby Creek, and Hocking River, respectively. Water samples were collected in two bottles at each sample location, one with a precleaned 125 mL wide mouth low-density polyethylene bottle and one with a 20 mL plastic scintillation vial with a poly cone cap. Before filling each bottle, water from the stream was used to rinse the bottles three times. The 125 mL sample at each location was for anion and cation analysis via ion chromatography using a Dionex Ion Chromatograph (DX-120). The 20 mL sample was analyzed for H2O isotopes using a Picarro L1102-i water isotope analyzer. The samples for ion analysis were filtered either on-site or shortly after collection in the Environmental Geochemistry lab on the OSU campus. Samples were filtered through Whatman 0.45 lm polypropylene filters using a plastic disposable filtering device. All samples were stored at 4 °C and in the dark until analyses were completed. Historical data utilized from OSU are from samples that were filtered similarly to the samples in this study. USGS data are from samples filtered according to USGS standard procedure using a 0.45 lm disposable capsule filter (Wilde et al., 2004). In order to resolve water contamination issues, it is essential to identify the source of contamination. Due to chemical similarities between the elements of chlorine and bromine, the concentration of Br relative to Cl in the Ohio rivers was utilized to aid in determining the origin of Cl . Bromide, like Cl , behaves conservatively in aquatic systems; however there are geochemical processes, such as precipitation of halite, that fractionate Br from Cl . As halite is precipitated, Br is preferentially excluded from the NaCl crystal lattice due to its much larger ion size relative to Cl (Oosting and Von Damm, 1996). Thus, waters affected by the dissolution of halite will have higher Cl /Br mass ratios due to the addition of Cl with little accompanying Br input. Previous studies have examined Cl /Br mass ratios of waters affected by the dissolution of halite (Davis et al. 1998, 2001; Panno et al. 2006). In addition, the analysis of the cosmogenic isotope 36Cl was utilized to interpret the origin of Cl in the Ohio rivers. Chlorine-36 is produced in the upper atmosphere by cosmic-ray spallation and has a long half-life of over 301,000 years. More recently, 36Cl was released into the atmosphere in large amounts from 1950s nuclear weapons testing, establishing it as a useful hydrologic tracer worldwide (Davis et al., 2001). It behaves the same as stable chlorine and mixes with marine derived stable chlorine in the atmosphere, which results in relatively low but measurable ratios of 36Cl/Cltotal in precipitation. The 36Cl/Cltotal ratio in the ocean has been estimated at 0.5 ± 0.3  10 15 (Argento et al., 2010). Consequently, halite taken from deposits formed by the evaporation of saline or seawater long ago has little 36Cl. As distance from the ocean increases, 36Cl/Cltotal ratios in groundwater generally increase near

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

27

Fig. 1. June 2012 and March 2013 Ohio study locations. Station names: LB, Oldtown; DF, Plain City; DC, West Jefferson; DD, Darbydale; DE, Darbyville; OE, Claridon; OD, Delaware; OC, Worthington; OA, Drake Union; AE, Kilbourne: HB, Lancaster; HC, Enterprise; HA, Logan; HD, Athens. All stations except OA and HA correspond to USGS gauge stations. Milford station was not sampled for this study but historical USGS data is included in discussion. For precise locations, see Appendix A.

the center of continents (Davis et al., 2003). The number of 36Cl atoms is small compared to total atoms of chlorine in water, so the ratio of 36Cl/Cltotal is conventionally multiplied by 1015 (Davis et al., 2003). In this paper, any reference to the 36Cl/Cl ratio refers to the value of the resulting ratio of (36Cl atoms/total Cl atoms)  1015. Samples collected in June 2012 for ion analysis from the Olentangy River and Darby Creek were composited and sent to Purdue University’s Rare Isotope Measurement Laboratory (PRIME Lab) in West Lafayette, Indiana, where they were analyzed for 36 Cl using accelerator mass spectrometry (Elmore and Phillips, 1987). The sample volumes remaining after ion analyses for the four 2012 Darby Creek samples and the four 2012 Olentangy River samples were combined to form two larger samples representing Darby Creek and Olentangy River, respectively. This was done to make sure there was sufficient Cl for analysis. As flow was not measured for this study upon sampling and USGS flow measurements for 2012 sample days are only available for some Olentangy River and Darby Creek locations, the compositing of samples for 36 Cl analysis was not flux-weighted. Therefore, the values obtained from the 36Cl analysis do not represent the average 36Cl composition or 36Cl/Cl ratio of the combined samples but instead represent a value that is between the 36Cl compositions of the end-member samples in each composite.

The presence of trends in Cl and Na+ concentrations, fluxes, and streamflow over time were determined for the sites with more complete historical datasets using the non-parametric Mann–Kendall test at the 95% confidence interval (Helsel and Hirsch, 2002). This test was utilized due to its low sensitivity to breaks in time series, a common occurrence in the available historical data. The Mann–Kendall test for trend analysis does not require that data conform to any particular distribution and has the null hypothesis, H0, that no trend exists. Kendall’s tau (s) measures the strength of the relationship between time and the y variable and has a value between 1 and +1, with a positive value indicating an increasing trend and a negative value indicating a decreasing trend. The Kendall S statistic also indicates how strong a trend is by measuring the dependence of y on x, demonstrated by the magnitude of difference from zero, where a large positive value indicates an increasing trend while a large negative value indicates a decreasing trend. Probability, i.e. the p-value, between S and sample size, n, was calculated using a two tailed test to statistically quantify the significance of a trend (Helsel and Hirsch, 2002). The null hypothesis is rejected when the p-value is less than the significance level, chosen to be 0.05 for this study, resulting in a statistically significant existing trend. Possible autocorrelation in the data was considered by calculating the adjusted variance of S, i.e. var(S), from Hamed

28

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

and Rao (1998). Addinsoft’s XLSTAT 2014 was used to perform the test. 3. Results 3.1. Water Isotopes Results of d18O and dD analysis for the June 2012 river samples are plotted with central Ohio precipitation samples from Leslie (2013) and the Global Meteoric Water Line (GMWL) in Fig. 2 (Craig, 1961). To ensure the ion data could be taken as a primary signal of water quality rather than from water that had undergone evapoconcentration, precipitation samples were included to compare the isotopic characteristics of central Ohio meteoric waters with the river samples in this study. The isotopic values measured for rivers in this study exhibit a narrow range in composition, from approximately 6‰ to 8‰ in d18O and 30‰ to 42‰ in dD. The June 2012 river samples fall on the same trend as the precipitation samples and exhibit a linear trend parallel to but slightly above the GMWL. Isotopic values reveal no strong characteristics of waters affected by excessive evaporation, indicating that the ion data from the rivers are reflective of the primary input. 3.2. Water chemistry Anion and cation data for the June 2012 and March 2013 (2012– 2013) river samples from central Ohio are reported in Table 1. Nitrate data are presented as N-NO3 , or nitrate as nitrogen. Available flow data for the 2012–2013 sample dates from the USGS for some locations are shown in Table 1 to provide a general idea of the magnitude of flow. From this flow data, Cl , Na+, and N-NO3 fluxes were calculated and are listed in Table 1. Precision of the ion chromatography data were calculated as the average percent difference between duplicate measurements, with Cl < 1%, Na+ < 2%, and N-NO3 < 2%. Accuracy was determined based on The USGS Office of Water Quality inter-laboratory comparison study that our lab participated in during 2011 and 2012. Our results via ion chromatography for Cl concentration were different from the most probable value by less than 3%. Likewise, Na+ differed by <1% and N-NO3 by <2%. Precision and accuracy for Br was not as good as the other ions due to the low concentrations present in the samples. The concentrations were close to the detection limit of 0.04 mg/L, where the percent difference between duplicates was as much as 25%. Accuracy for Br was calculated as a 2% difference compared to the most probable value for 2011, whereas for 2012 the difference was as much as 14%.

δD (‰, VSMOW)

-15 -35 -55 -75 -95 -115 -16

-12

-8

-4

0

δ18O (‰, VSMOW) Fig. 2. Global Meteoric Water Line (Craig, 1961) with central Ohio precipitation samples (Leslie, 2013) and June 2012 central Ohio river samples. GMWL points chosen based on the formula: dD = 8d18O + 10â.

The range of Cl concentrations for June 2012 samples was 18.7–79.3 mg/L, with a mean of 46.6 mg/L and a median of 47.1 mg/L. The range of Na+ concentrations for June 2012 was 8.0–46.8 mg/L, with a mean of 27.3 mg/L and a median of 27.1 mg/L. The range of Cl concentrations for March 2013 samples was 33.1–71.9 mg/L, with a mean of 54.0 mg/L and a median of 58.0 mg/L. The range of Na+ concentrations for March 2013 was 15.6–39.0 mg/L, with a mean of 27.4 mg/L and a median of 28.9 mg/L. The highest 2012–2013 concentrations for Cl were observed in the Olentangy River at Worthington, Ohio (OC) just north of Columbus (Table 1; Fig. 1). The highest Na+ concentration for 2012 was also observed at OC, whereas the highest for 2013 was observed in the Hocking River at Lancaster, Ohio (HB) southeast of Columbus, with OC having the second highest. Concentrations of Na+ were slightly lower compared to Cl concentrations (in equivalents) at the majority of sites for both seasons (Fig. 3). Downstream profiles of 2012–2013 Cl and Na+ concentrations for sample locations on the Olentangy River and Darby Creek are shown with available flow and Cl and Na+ flux data in Table 2. Dashed spaces indicate either a lack of available data for the site or that no sample was taken. The range of Cl /Br mass ratios for June 2012 was 600–5400, with a mean of 2500 and a median of 1900 (Table 1). The range of Cl /Br mass ratios for March 2013 was 830–9600, with a mean of 5300 and a median of 5200. The 36Cl/Cl ratios for 2012 Olentangy River and Darby Creek samples were 17.4 ± 2.4 and 38.6 ± 2.8, respectively (Table 3). The blank had a value of 14 and the reported data have been corrected for the blank value. Time series of Cl and Na+ concentration and flux were created for river locations with consistent historical data (Figs. 4–6). Results from the historical analysis portion of this study are presented in Tables 4 and 5.

4. Discussion 4.1. Previous work on the impact of road salt in the northern US In the northern US, Cl additions via road salt have been abundant enough in the winter months to maintain extremely high concentrations in fresh waters across all seasons (Kaushal et al., 2005). Kaushal et al. (2005) observed rural stream Cl concentrations in the northeastern US that exceeded 100 mg/L throughout the year, suggesting changes in baseline salinity of surface waters and the prevalence of Cl contamination across geographic areas. Chloride concentrations for rivers in the Chicago Metropolitan area were also mostly above 100 mg/L from 2001 to 2005 and have been increasing with time (Kelly et al., 2012). Gardner and Carey (2004) demonstrated that urban storm sewer runoff continued to supply Cl and Na+ to urban portions of the Olentangy River in central Ohio more than 7 months after the last application of road salt. Novotny (2009) showed increases in salt concentrations related to road salt purchases over a 22-year period in 39 Minneapolis-St. Paul lakes and determined that even if Cl inputs from road salt ceased, it would take 10–30 years for Cl levels to reach predevelopment concentration. Chapra et al. (2009) reported that in the first six decades of the 20th century, Cl concentrations increased exponentially in almost all the Great Lakes of the US. These lakes showed peak Cl concentrations upwards of 30 mg/L around 1965–1975, with reduced levels in subsequent years, while current trends demonstrate that Cl is rising again. Pre-1960s elevated Cl concentrations were mostly due to industrial discharges from large production areas located near the Great Lakes, including Cleveland

29

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35 Table 1 Concentrations (mg/L), Cl /Br mass ratios, and flux (g/s) data for 2012–2013. Flow data (m3/s) represent daily means as reported by the USGS. Sample ID

Date

Cl

Na+

Br

N-NO3

Cl /Br

Flow

Cl flux

Na+ flux

N-NO3 flux

OA OC OD OE AE LB DC DD DE DF HB HC HD

6/19/12 6/19/12 6/19/12 6/19/12 6/19/12 6/21/12 6/19/12 6/19/12 6/21/12 6/19/12 6/21/12 6/21/12 6/21/12

67.4 79.3 27.0 34.7 46.0 18.7 26.3 49.2 47.1 33.9 55.0 53.1 68.1

36.9 46.8 12.2 20.8 25.1 8.0 13.0 28.1 27.1 19.0 43.6 29.7 45.2

0.037 0.049 0.014 0.030 0.052 0.006 0.008 0.013 0.012 0.006 0.015 0.054 0.113

0.50 1.29 2.51 1.03 7.90 4.25 1.17 1.11 0.81 0.53 0.67 1.56 0.50

1800 1600 1900 1100 880 3000 3500 3900 3800 5400 3600 980 600

– 0.9 1.3 – 0.2 2.5 1.0 – 2.4 – – 2.1 3.6

– 72 35 – 11 46 26 – 111 – – 110 243

– 42 16 – 6 20 13 – 64 – – 61 161

– 1.2 3.3 – 1.9 10 1.2 – 1.9 – – 3.2 1.8

OA OC DC DD DE HA HB HD

3/2/13 3/2/13 3/2/13 3/2/13 3/2/13 3/2/13 3/2/13 3/2/13

57.3 71.9 38.3 58.7 61.3 40.7 71.1 33.1

29.1 36.8 15.6 28.8 30.1 20.6 39.0 19.5

0.013 0.014 0.005 0.007 0.006 0.034 0.014 0.040

4.23 4.29 9.33 8.06 7.74 2.73 4.05 1.68

4400 5300 7000 8600 9600 1200 5200 830

– 32 8.5 – 24 – – 52

– 2279 326 – 1455 – – 1734

– 1168 133 – 714 – – 1023

– 136 79 – 184 – – 88

Precision of Cl : <1%, Na+: <2%, Br : 25%a, N-NO3: <2%. Accuracy of Cl : <3%, Na+: <1%, Br : 14%a, N-NO3: <2%. a See Section 3.2.

2.5 2.0

Na+ mM

4.2. Ohio data comparison

June 2012 March 2013 y=x all samples

1.5 1.0

y = 0.9404x - 0.1191 R² = 0.8702

0.5 0.0 0.0

0.5

1.0

1.5

2.0

2.5

Cl- mM Fig. 3. Na+ vs. Cl plot of 2012–2013 central Ohio river samples.

and Detroit. Starting in the late 1960s and continuing over the past 35 years, there have been major reductions in industrial discharges from this region, likely reflected in the observed decreases in lake Cl concentrations during the 1970s. However, recent increasing trends of Cl suggest input from other anthropogenic sources, such as increased road salt use (Chapra et al., 2009).

4.2.1. Cl and Na+ concentration Although the water samples in this study have lower solute concentrations than some of those described in referenced studies, evidence of anthropogenic input of Cl and Na+ to the Ohio rivers is present. Panno et al. (2006) considered Cl concentrations in the range of 50–1000s of mg/L to indicate water affected by road salt. Additionally, they indicated that background concentrations in shallow aquifers in northeastern Illinois ranged from <1 to 15 mg/L for both Cl and Na+. Chloride concentrations in central Ohio 2012–2013 river samples were mostly near or in the bottom part of the 50+ mg/L range, suggesting contributions from road salts (Table 1). Similarly, Stucker (2013) reported 2011 Cl concentrations of Olentangy River tributaries that represent the lowest order streams ranging from 48 to 255 mg/L. All of the Cl and Na+ concentrations of 2013 samples and all but four Na+ concentrations of 2012 samples listed for this study were above 15 mg/L. These concentrations suggest the anthropogenic addition of Cl and Na+ to these rivers if it is assumed central Ohio waters have background concentrations similar to the Illinois waters. The historical and 2012–2013 water chemistry data as well as the findings of other studies in the northern US show that Cl

Table 2 Downstream profiles of 2012–2013 Cl and Na+ concentrations and fluxes with flow for Olentangy River and Darby Creek sampling sites (see Fig. 1 for map). Flow data represent daily means as reported by the USGS. Na+ (mg/L)

Cl (mg/L)

Flow (m3/s)

Na+ flux (g/s)

Cl flux (g/s)

2012

2013

2012

2013

2012

2013

2012

2013

2012

2013

DF DC DD DE

33.9 26.3 49.2 47.1

– 38.3 58.7 61.3

19.0 13.0 28.1 27.1

– 15.6 28.8 30.1

– 1.0 – 2.4

– 8.5 – 24

– 26 – 111

– 326 – 1455

– 13 – 64

– 133 – 714

Olentangy River North OE OD OC South OA

34.7 27.0 79.3 67.4

– – 71.9 57.3

20.8 12.2 46.8 36.9

– – 36.8 29.1

– 1.3 0.9 –

– 24 32 –

– 35 72 –

– – 2279 –

– 16 42 –

– – 1168 –

Darby Creek North

South

30

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

Table 3 Chlorine-36 ratios of 2012 Olentangy River and Darby Creek samples. AMS measurement was done at Purdue University’s PRIME Lab. 36

Olentangy River Darby Creek

17.4 ± 2.4 38.6 ± 2.8

Error (%)

Cl (mg/L)b

36

8 5

53.3 38.4

2.52 1.57

Cl (atoms/L  107)

Flux (g/s)

40

300 250 200

Cl Na+

Cl- (mg/L)

Concentration (mg/L)

(Number of atoms of 36Cl/total number of atoms of Cl)  1015. Final concentration of the combined samples for each river.

150 100

30 20 10

June 2012

0

50

Cl- Flux (g/s)

a b

Cl/Cl ratioa

0 7000 6000 5000 4000 3000 2000 1000 0

600 400 200 0

Fig. 6. Time series of Cl concentration and flux at the Olentangy River, Claridon Ohio (OE). Historical data from USGS site 3223000.

Fig. 4. Time series of Cl and Na+ concentration and flux at the Hocking River, Athens, Ohio (HD). Historical data from USGS site 3159500.

and Na+ vary seasonally, with higher concentrations during winter months (Kaushal et al., 2005; Kelly et al., 2012). Higher Cl and Na+ concentrations in March 2013 were likely due to increased application and runoff of road salts during winter and the differences in salt application between the milder winter (less snowfall) of 2011–2012 compared to that of 2012–2013. Some sites (i.e. OA, OC, and HD) exhibited decreased concentrations of Cl and Na+ in March 2013 relative to June 2012 as a result of increased flow in March 2013 (Table 1). Despite this, Cl and Na+ fluxes were still

an order of magnitude higher in March relative to June for sites with available flow data, indicating increased inputs of both Cl and Na+ from road salt during winter months (Table 1).

4.2.2. Na+ vs. Cl Either increases in both Cl and Na+ concentrations or decreases in both ions were observed from June 2012 to March 2013 at all sites except the Hocking River at Lancaster, Ohio (HB). This pattern is expected in surface waters if halite is the main source of Cl and Na+ to the system (Kelly et al., 2012). Water affected by road salt usually has excess equivalent Cl compared to Na+, despite the fact that equimolar concentrations of both ions are present in

Concentration (mg/L)

150 125 100

Cl Na+

75 50 25 0 7000

Flux (g/s)

6000 5000 4000 3000 2000 1000 0

Fig. 5. Time series of Cl and Na+ concentration and flux at the Little Miami River, Milford, Ohio. Historical data from USGS site 3245500. Two Cl flux measurements (3/ 1980 = 11,600; 1/1994 = 9530) are not visible as they exceed the chosen y-axis scale.

31

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35 Table 4 Trend analysis results from the Mann–Kendall test for the Hocking River at Athens (HD) and the Little Miami River at Milford, Ohio.

s

Site

Variable

n

Date range

S

Hocking River, Athens (HD) (Fig. 4)

Cl Flow Cl flux Na+ Flow Na+ flux

303 303 303 199 199 199

10/55–10/65 10/55–10/65 10/55–10/65 10/55–12/62 10/55–12/62 10/55–12/62

19,762 7551 3928 4801 1797 1260

Little Miami River, Milford (Fig. 5)

Cl Flow Cl flux Na+ Flow Na+ flux

162 162 162 135 135 135

9/64–9/94 9/64–9/94 9/64–9/94 9/74–9/94 9/74–9/94 9/74–9/94

2820 6 1327 1303 435 472

var(S)

p-Value

Interpretation

inc/dec

0.434 0.165 0.086 0.245 0.091 0.064

3105328 3106047 3106113 881637 882146 882161

<0.0001 <0.0001 0.026 <0.0001 0.056 0.180

Reject H0, trend Reject H0, trend Reject H0, trend Reject H0, trend Accept H0, no trend Accept H0, no trend

inc dec inc inc – –

0.218 0.000 0.102 0.146 0.048 0.052

476429 476707 476721 276032 276362 276374

<0.0001 0.993 0.055 0.013 0.408 0.369

Reject H0, trend Accept H0, no trend Accept H0, no trend Reject H0, trend Accept H0, no trend Accept H0, no trend

inc – – inc – –

Table 5 Comparison of past Cl concentration and flux for locations with limited historical data. See Appendix A for USGS data sources. Olentangy River (OC)

Cl (mg/L)

Cl flux (g/s)

Data source

Darby Creek (DE)

Cl (mg/L)

8/23/1967 7/18/1969 6/29/1970 7/20/1971 6/27/1972 8/11/1972 6/20/1973 6/22/1974 8/8/1974 7/20/1976 6/23/1977

53 18 32 56 37 50 10 62 50 60 59

57 89 176 68 44 69 1263 93 68 54 45

USGS USGS USGS USGS USGS USGS USGS USGS USGS USGS USGS

8/24/1967 7/28/1969 6/25/1970 7/19/1971 8/20/1971 6/28/1972 8/15/1972 7/18/1973 8/14/1975 6/16/1976 7/8/1977

31 20 20 30 42 30 39 20 32 37 30

20 216 97 55 31 87 52 95 26 78 87

7/20/2003

34.1

103

OSU

6/21/2012

47.1

111

6/9/2004 6/16/2005 6/25/2005 7/20/2005 8/22/2005 6/12/2007

25.1 43.9 50.1 39.9 47.8 44.5

98 81 50 132 70 66

OSU OSU OSU OSU OSU OSU

1/8/1970 2/18/1970 3/9/1970 2/3/1971 1/18/1972

36 22 20 41 27

122 249 398 68 291

6/19/2012

79.3

72

3/13/1972

28

271

USGS

16 24 9 33

739 597 2575 991

USGS USGS USGS USGS

61.3

1455

12/19/1969 2/18/1971 1/10/1972

34 40 67

477 459 1343

USGS USGS USGS

3/28/1973 2/8/1974 2/24/1975 3/23/1977

3/26/1973

29

961

USGS

3/2/2013

2/19/1974 2/27/1975 3/11/1976 3/3/1977 12/18/2004 1/12/2005 1/24/2005 2/27/2005 3/21/2005 12/27/2005 1/16/2006 1/24/2006 1/20/2007

120 17 59 33 40.7 10.6 10.3 46.7 80.0 50.6 37.8 31.7 10.0

999 1617 454 412 518 1766 342 324 301 3182 567 565 1266

USGS USGS USGS USGS OSU OSU OSU OSU OSU OSU OSU OSU OSU

3/2/2013

71.9

2279

unimpacted surface waters. Unlike conservative Cl ions, Na+ can take part in cation exchange reactions and thus is more likely to be retained in the subsurface (Kelly et al., 2012). For example, road salt derived sodium can displace dissolved Ca, Mg, and K from soil, increasing their concentrations in surrounding waters (Mason et al., 1999). Eventually the exchange process reaches a steady state where additional Na+ moves more conservatively through soil and dissolved Ca, Mg, and K return to previous levels (Mason et al., 1999). However, it should be noted that Na+ concentrations may

Cl flux (g/s)

Data source USGS USGS USGS USGS USGS USGS USGS USGS USGS USGS USGS

USGS USGS USGS USGS USGS

not reflect all the Na+ present in natural water systems due to these processes. For our study, the lower Na+ concentrations relative to Cl observed at most sites could reflect either the participation of Na+ in ion exchange processes or the use of liquid CaCl2 and other non Na+ salts by ODOT as road deicers in addition to halite (Fig. 3; ODOT, 2011). Sodium/chloride (Na+/Cl ) molar ratios were smaller in March relative to June, similar to the findings of Kelly et al. (2012). Additional comparison of Na+ vs. Cl in central Ohio rainwater precipitation from Leslie (2013) with our central Ohio river

32

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

samples showed precipitation to be more often enriched in Na+ relative to Cl , while rivers displayed the opposite. Although Na+/Cl ratios were higher in rainfall (mostly >1) compared to 2012–2013 river samples (mostly <1), concentrations of Na+ were 100 times higher on average in the rivers than in the rainfall. Neal and Kirchner (2000) demonstrated that as salt (NaCl) concentration in streams increases above the norm, more Na+ is incorporated into ion exchange reactions, thus lowering the overall Na+/Cl ratio. This supports our notion that Na+ is not behaving conservatively throughout the system, making it difficult to interpret anthropogenic induced changes in concentration. 4.2.3. Downstream comparison Just north of Columbus between sites OD and OC (Fig. 1), the Olentangy River flows through southern Delaware County, one of the fastest growing counties in the US over the past decade (Ohio Department of Development, 2013). Many of the residents commute by road into the Columbus metropolitan area. For our study, sites OC and OA showed elevated Cl and Na+ concentrations relative to upstream sites OE and OD for both seasons (Table 2). Similarly, Jacobs (2006) reported Olentangy River Cl and Na+ concentrations at two locations near OC and OA and found that concentrations were almost always higher at the downstream location across all seasons, owing the difference to increased input from road salt in the more urban area of the watershed (Jacobs, 2006). In addition to road runoff, treated wastewater in urban areas can be a major source of Cl input, especially during summer and autumn when flows are low and surface drainage is at a minimum (Kelly et al., 2010). Treated wastewater in central Ohio can include storm and road runoff as well as sanitary and industrial wastewater, making Cl source identification more difficult (Ohio EPA, 2014). For our study, there are six major wastewater treatment plants (WWTPs), defined as discharging more than 1 million gallons per day, discharging directly into the rivers of study upstream of sample locations (Fig. 1; Ohio EPA, 2014). The increases in Cl and Na+ concentration downstream at sites OC and OA were possibly influenced by the location of two major WWTPs discharging directly to the Olentangy River north of OC (Fig. 1; Table 2). This possibility is supported by the increases in Cl and Na+ flux observed in June 2012 at site OC, south of the WWTP, relative to site OD, north of the WWTP (Table 2). Additionally, the Hocking River and Olentangy River sites located downstream of WWTPs had the highest concentrations of Cl and Na+ for June 2012, also suggesting increased influence from treated wastewater. Yet, it has been demonstrated that Cl and Na+ from road salt runoff impacts urban portions of the Olentangy River throughout the entire year, so the distinction between input from treated wastewater and road runoff is difficult without additional analyses of these two sources (Gardner and Carey, 2004). Treated wastewater is known to be a considerable source of anthropogenic nitrate to surface waters, so N-NO3 fluxes would likely increase downstream of WWTPs if treated wastewater was a major contaminant source to these rivers (Kelly et al., 2010). This was not observed for the Olentangy River, as site OC downstream of a major WWTP had a lower N-NO3 flux for June 2012 than upstream site OD (Table 1). For Darby Creek, downstream profiles of 2012– 2013 samples showed higher Cl and Na+ concentrations and fluxes at the two locations south of the city of Columbus (Table 2). Sites DD and DE differed little in concentration for all ions and had elevated Cl and Na+ concentrations relative to upstream sites, but only DE is located downstream of a major WWTP (Fig. 1). This made input from treated wastewater at DE not clearly identifiable. Generally, higher Cl and Na+ concentrations and fluxes were observed downstream at the southernmost, more urban sites on the Olentangy River and Darby Creek for 2012–2013.

4.2.4. Cl /Br analysis Chlorine and bromine are present in natural water systems commonly as the anions Cl and Br . The geochemical behavior of both ions is generally conservative in solution, however many factors can affect Cl /Br mass ratios in the environment (Davis et al., 1998). In natural solids and fluids, the mass ratios of Cl / Br commonly range from 40 to 8000 (Davis et al., 1998). Due to its larger size, Br partitions into binary salts in a different manner than Cl . Bromide salts are known to be even more soluble than those of Cl , so when extreme evaporation permits the precipitation of halite, the resulting brine is left enriched in Br (Davis et al., 1998). Halite therefore has a low Br concentration compared to Cl , resulting in large mass ratios of Cl /Br in the material. Reported values for the lower and upper bromine content of fully investigated halite rocks range from 0.01 to 0.03 wt% Brtotal/ NaCl, with a distribution coefficient of 0.032 for Br between halite and brine (Braitsch, 1971; McCaffrey et al., 1987). Due to halite’s low Br content, the addition of Cl from road runoff to surface waters results in large Cl /Br mass ratios in these waters. Coastal areas exhibit Cl /Br natural mass ratios in precipitation near 250, whereas north central states display ratios around 50, with any ratio above 300 attributed to sources of Cl other than atmospheric deposition (Davis et al., 2001). Davis et al. (1998) reported that the Cl /Br mass ratios for waters affected by the dissolution of halite are generally in the range of 1000–10,000, compared to the seawater Cl /Br ratio of 290. Panno et al. (2006) determined that groundwater samples affected by road salt from northeastern Illinois had Cl /Br ratios of 1164–4225. For northeast Ohio, Foos (2003) reported average Cl /Br ratios near 1590 for road salt contaminated springs and seeps. The Cl /Br mass ratios for the majority of our 2012–2013 river samples were in the 1000– 10,000 Cl /Br mass ratio range (Table 1). Hence, by the criterion of Davis et al. (1998), they would be classified as being affected by halite dissolution. Only 4 out of 14 samples were outside the range, however they all were greater than 600 and thus well above the preanthropogenic mass ratios for Ohio groundwater near 70 (Davis et al., 2001). Our mean mass ratios of 2500 and 5300 and median mass ratios of 1900 and 5200 for 2012 and 2013 samples, respectively, suggest that all of the Ohio rivers have been affected by halite dissolution. To further evaluate the source of Cl in the Ohio rivers, Cl /Br mass ratios and N-NO3 concentrations were compared with those in Davis et al. (2001). The study reported a Cl concentration of 24.7 mg/L and a Cl /Br mass ratio of 948 in spring water from Big Spring, Iowa (Davis et al., 2001). Although Br data are not provided, the mass ratio suggests a concentration around 0.026 mg/L Br . The magnitude of the ratio alone reveals that a large amount of Cl was introduced into the system by means other than natural deposition. Suggested artificial sources of Cl other than road salt were agricultural sources like fertilizer and animal waste (Davis et al., 2001). Average Br concentrations for the 2012–2013 samples in our study were 0.032 mg/L and 0.017 mg/L respectively, while average Cl concentrations were 46.6 mg/L and 54.0 mg/L respectively (Table 1). Although Cl and Br concentrations are comparable in the Iowa sample and 2012–2013 central Ohio samples, the additional comparison of N-NO3 concentration suggests different primary sources of Cl between the two locations. The Iowa sample had a concentration of 12.6 mg/L N-NO3 , the largest in the study by far, thus favoring the agricultural source for the Cl . Although Cl concentration was higher in the Ohio samples, the NNO3 concentrations were lower than the Iowa water with a 2012 average of 1.83 mg/L and a 2013 average of 5.26 mg/L (Table 1). A 36 Cl/Cl ratio of 212 was also reported for the Iowa sample, much higher than what is characteristic of halite and what was observed for the central Ohio rivers (Table 3). As the Cl /Br mass ratio of the Iowa spring water is within the lower range of the Ohio

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

samples, while Ohio waters have higher Cl and lower N-NO3 concentrations, Cl derived from road salt and not from agriculture was the likely main source of Cl in the Ohio rivers. All Ohio locations that were re-sampled in 2013 displayed an N-NO3 increase more than twice the concentration in 2012, while available NNO3 flux data showed increases of more than forty-fold (Table 1). Thus, the higher Cl /Br mass ratios observed in 2013 accompanied by notable increases in N-NO3 concentration and flux suggest influence from both road salt and the leaching of fertilizers from agricultural lands during the winter season. March 2013 increases in N-NO3 flux were accompanied by increased flow, indicating an N-NO3 source that was amplified by increased flow. Treated wastewater is known to be a nearly constant source of ions to urban waterways throughout the year and thus fluxes of contaminants from treated wastewater should be less seasonally controlled (Kelly et al., 2010). Therefore, the increase of N-NO3 observed in March 2013 was likely a result of the leaching of fertilizers and was not attributed to treated wastewater. Since precision and accuracy were not as good for Br as for the other ions, further evaluation regarding seasonal or source differences between Cl and Br was not possible. 4.2.5. Chlorine-36 analysis The expected natural background 36Cl/Cl ratio for regional Ohio groundwater is 350, while the 36Cl/Cl ratio for bedded halite is <5 (Davis et al., 2003; Davis et al., 2001). The Olentangy River and Darby Creek 36Cl/Cl ratios from 2012 composite samples of 17.4 and 38.6, respectively, suggest the addition of old stable Cl relative to natural background 36Cl deposition (Table 3). From these ratios, approximately 95% of the Olentangy River Cl and 89% of the Darby Creek Cl in 2012 composite samples is indicative of halite dissolution based on the mixing line between water containing old Cl with a 36Cl/Cl ratio near zero and preanthropogenic Ohio groundwater with a 36Cl/Cl ratio of 350. Assuming 90% of the Cl in the composite samples is from the dissolution of halite, corresponding Cl /Br ratios in the range of 900–9000 for the same samples would be expected based on the same mixing line between water indicative of halite dissolution with Cl /Br ratios of 1000–10,000 and preanthropogenic Ohio groundwater with a Cl /Br ratio of 70. The Olentangy River samples that were composited had Cl /Br ratios ranging 1100–1900 with an estimated composite ratio of 1600, whereas the Darby Creek samples that were composited had Cl /Br ratios ranging 3500–5400 with an estimated composite ratio of 4000. All of these values lie within the Cl /Br range of 900–9000 characteristic of waters with a 90% halite dissolution component for Cl . Previously reported 36Cl/Cl ratios near 165 were observed in 6 shallow Ohio wells, while some of these wells had accompanying Cl /Br ratios greater than 260 (Davis et al., 2003). The lower than expected 36Cl/Cl ratio was considered to be a result of road runoff and human influences, possibly road salt and agricultural chemicals. Additionally, Davis et al. (2001) found that the most prevalent artificial source of old Cl in shallow groundwater samples in the US seemed to be road salt used to control ice, with other sources related to agriculture and industry. The Olentangy River and Darby Creek surface waters had higher Cl /Br ratios and lower 36Cl/Cl ratios compared to Ohio groundwater samples in Davis et al. (2003), suggesting increased impact from road runoff and human influence on the surface relative to subsurface. Davis et al. (2003) attributed a low 36Cl/Cl ratio of 12.6 and a Cl /Br ratio of 2700 in Indiana groundwater, similar to our Ohio surface water findings, to the presence of old Cl derived from the mining and use of subsurface salt deposits used for deicing, fertilizers, and industrial effluents. The 36Cl analysis overall revealed the source of Cl in the samples was likely from old Cl deposited long ago, such as in halite deposits utilized for road salt today. While central Ohio

33

road salt is mined largely in northern Ohio, other areas of the midwestern US utilize salt mined from New York, Michigan, and Ontario, Canada (ODOT, 2008). Therefore, there is a need to characterize the 36Cl content of halite from varying geographic origins so the 36Cl variability observed in natural waters arising from halite dissolution can be diagnosed more accurately in future geochemical investigations. 4.3. Historical analysis Sample locations with adequate available historical data displayed increases in concentration through time. The site with the longest set of historical data was the Hocking River at Athens, Ohio (HD) dating back to 1955, although the data were not continuous. Chloride concentration at this location showed an increasing trend over the period from 1955 to 1965, as demonstrated by both simple linear regression and the Mann–Kendall results (Fig. 4; Table 4). Some Cl data were available for this location up until 1974; however it was collected much more sparingly and thus was not included in the analysis. Available Na+ concentration data spanned 1955–1962 and were also characteristic of a significant increasing trend (Fig. 4; Table 4). This 1960s timing reflects the beginning of large-scale road salt application, and changes seen in Cl were closely mirrored by those of Na+, supporting the argument that halite dissolution had impacted the river early on. Similarly, Kelly et al. (2012) observed notable increases in Cl concentration since the mid-1970s in surface waters near the Chicago, IL region, with the most accelerated increases observed in areas of rapidly changing rural to urban land use. In Fig. 4, the periodicity of the concentration time series reflects increased concentrations of both Cl and Na+ historically during the months October–December. Kelly et al. (2010) observed this same trend in concentration for some Illinois rivers, attributing the increases to the leaching of fertilizers from agricultural lands that are applied after the autumn harvest. Although we did not sample the Little Miami River at Milford in west central Ohio (Fig. 1), this location had one of the most complete datasets through the 1980s and 1990s and demonstrated increased Cl and Na+ concentrations over this time (Fig. 5). Mann–Kendall analysis revealed a statistically significant increasing trend in Cl concentration from 1964 to 1994, as did available Na+ data from 1974 to 1994 (Table 4). The only other available data for this location corresponded to a much smaller dataset for 1998– 2000 (n = 30) in which no statistically significant trends were identified for any variable. Increasing trends identified for Na+ concentration were of lesser magnitude than Cl for both rivers, demonstrated by the smaller S and s values (Table 4). This observation could reflect the more complex geochemistry of Na+ in natural waters relative to Cl . Overall, studying Na+ through time was difficult due to the limited amount of data available for this ion. To gain insight on the dominant functions that influenced the increasing trends observed in concentration, Mann–Kendall trend analysis was also performed on corresponding flow data and Cl and Na+ flux data for both locations (Table 4). A decreasing trend in flow, weaker in magnitude than the increasing trend for Cl concentration for the same time period, was observed for the Hocking River at Athens (HD) from 1955 to 1965. Furthermore, an increasing trend was identified for Cl flux. The flux time series shows periodic increases in Cl and Na+ flux characteristic of the months January–April (Fig. 4). This reflects the seasonably variable nature of road salt inputs, with the highest fluxes in winter and early spring (Kelly et al., 2010). The trend in Cl flux confirms that the increasing trend in Cl concentration at this location was primarily a result of changes in baseline salinity or input functions rather than hydrologic functions. For the same location, no significant trends were found in correlating flux and flow data for Na+. Similarly, no trends in flow or Cl and Na+ fluxes were observed at

34

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

the Little Miami River at Milford. Our Mann–Kendall trend analysis confirmed increasing trends in Cl and Na+ concentrations in these Ohio rivers during the latter part of the 20th century, while also highlighting differences between hydrologic functions and the material flow of both constituents through time. Despite the lack of significant trends for Na+ flux, Cl and Na+ flux in both rivers followed similar patterns of change, a characteristic of water affected by halite dissolution (Figs. 4 and 5). While changes in the material flow of Cl and Na+ were not as pronounced in our historical datasets as changes in concentration, the trends in concentration observed in these rivers have major implications for the increasing salinization of Ohio’s natural waters and water resources. Chloride concentration and flux for the Olentangy River at Claridon, Ohio (OE) from 1965 to 1977 are shown with the June 2012 Cl concentration plotted as a dashed line for comparison (Fig. 6). While no statistically significant trends were found for either variable during this time, the 2012 concentration is among the highest compared to past sampling dates. This observation supports the notion that Cl concentrations have remained elevated in recent years. Time series of Cl concentration and flux in other Ohio rivers were inconsistent due to lack of historical data collection, but comparison of past data with more recent data was still possible. Table 5 shows all available historical Cl concentration and flux data from the same seasons sampled in 2012–2013 with recent data for the Olentangy River at Worthington, Ohio (OC) and Darby Creek at Darbyville, Ohio (DE). Historic data from OSU at Highbanks Metro Park, located about 4.5 km north of the USGS gauge site where we sampled site OC, was included due to the close proximity of the sampling sites. For June 2012, Cl concentrations for both OC and DE were the highest observed amongst all other historical data for past summers. Similarly in 2013, DE had a Cl concentration higher than all those observed in the 1970s and OC had a Cl concentration higher than most late winter/early spring sampling dates in the past. Chloride fluxes at OC for 2013 and at DE for both 2012 and 2013 sampling dates were the second highest flux measurements observed in our datasets, indicating increased inputs of Cl to these rivers during 2012–2013 relative to earlier sampling years. The magnitude of these Cl fluxes suggests that the higher concentrations observed in 2012–2013 were influenced by changes in input function and were not strictly controlled by changes in flow. All historical Cl data from the USGS and OSU corresponding to the sites sampled in 2012–2013 as well as additional data shown in Figs. 4– 6 can be found in the Supplementary file for this study. 5. Conclusions This study presented several lines of evidence that indicate anthropogenic sources such as road salt have largely contributed to the salinization of central Ohio rivers. Chloride/bromide mass ratios of 2012–2013 samples in conjunction with chlorine-36 analysis support halite as the main source of Cl to the central Ohio rivers. Increases in Cl and Na+ fluxes in late winter observed both historically and in 2013 indicate road salt as the dominant halite source at this time of the year. Increases in both Cl /Br ratios and N-NO3 concentrations and fluxes in March 2013 suggest a mixture of road salt and fertilizer runoff influencing the rivers in late winter. In addition to road salt, anthropogenic Cl sources such as agricultural fertilizers and treated wastewater were assessed for their influence largely through comparison of spatial and seasonal differences in Cl and N-NO3 concentrations and fluxes. While inputs from these sources exist in the systems studied, there was little evidence indicating they are the main sources of Cl to the rivers. Comparison of river Na+ and Cl concentrations with those of central Ohio precipitation suggests Na+ is not behaving

conservatively in these systems, possibly due to the participation of Na+ in ion exchange processes. Chloride and sodium concentrations and fluxes followed similar patterns of change both in recent data and historically, supporting the argument that halite dissolution has impacted the rivers of study. Downstream data profiles of the Olentangy River and Darby Creek revealed increased Cl and Na+ concentrations and fluxes at downstream sites, possibly reflecting increased road salt application near the more urban areas of the watershed or increased influence from treated wastewater during the summer. Historical temporal trend analysis identified increasing trends in Cl and Na+ concentrations during the latter half of the 20th century at two different locations in Ohio. An increasing trend in Cl flux through the 1960s was also identified for one of our river locations, suggesting early impact on the river system from the onset of road salt as the primary agent for deicing. Available historical data for other rivers were inconsistent, but 2012–2013 samples still exhibited elevated Cl concentrations and fluxes relative to most historical observations from the same season. Changes in Na+ concentration over time were more difficult to identify because there is the least amount of recorded data for this ion and the geochemistry of Na+ is more complex than Cl . Observations of the select Ohio rivers in this study are similar to what have been observed elsewhere in the US. Both Cl and Na+ concentrations have been increasing in Ohio rivers since the 1960s as a result of human activities. These increasing concentrations may arise to have major implications if they persist for not only Ohio’s stream ecosystems, but drinking water supplies and infrastructure as well. Society’s need for both high quality fresh water resources and safe travel on roadways after snow and ice events demands the attention of public officials to road salt pollution. In addition to current anthropogenic inputs, impending influences from climate change and energy development in Ohio via hydraulic fracturing have the ability to put more stress on our fresh water resources in the future and must be anticipated. Acknowledgments We would like to thank Deborah Leslie and Susan Welch of The Ohio State University for providing their assistance in lab analyses and manuscript development. Funding for this Project was provided by the Shell Exploration and Production Company through the 2012 SURE Program at The Ohio State University School of Earth Sciences. We acknowledge helpful comments from the anonymous reviewers and editor. Chlorine-36 analysis was performed at Purdue University’s PRIME Lab in West Lafayette, IN. Appendix A Descriptions of 2012–2013 Ohio sample locations and the historical data utilized for analysis. All historical Cl-data from the USGS and OSU corresponding to these sites can be found in the supplementary file for this study. OA

OC

OD

Olentangy River, Drake Union, Ohio State Univ., 1849 Cannon Dr., Columbus, OH 43210; corresponds to data from OSU Olentangy River, Worthington, Olentangy Trail at OH315N and I-270E; merged with data from USGS site 3226800 & data from OSU at Highbanks Metro Park (4.5km north of USGS site) Olentangy River, Delaware, County Rd. 213/Main Rd. off US-23N; corresponds to USGS site 3225500

K.R. Dailey et al. / Applied Geochemistry 47 (2014) 25–35

OE

DC

DD

DE

DF HA HB HC HD LB AE

Olentangy River, Claridon, River Bend Campground, Whetstone River Rd.; merged with data from USGS site 3223000 Little Darby Creek, West Jefferson, east of Chestnut St. near E Main St.; corresponds to USGS site 3230310 & data from OSU Big Darby Creek, Darbydale, Harrisburg Georgesville Rd. near Osprey Lake; corresponds to USGS site 3230400 & data from OSU Big Darby Creek, Darbyville, OH-316E just east of downtown Darbyville; merged with data from USGS site 3230500 Big Darby Creek, Plain City, OH-161 at Butler Ave/County Highway 3; corresponds to USGS site 3230200 Hocking River, Logan, South of US-33 on State Route 93 at Kachelmacher Park Hocking River, Lancaster, Canal St. at S. Columbus St.; corresponds to USGS site 3156400 Hocking River, Enterprise (near Logan), Zeller Rd.; corresponds to USGS site 3157500 Hocking River, Athens, Stimson Ave near US-33; merged with data from USGS site 3159500 Little Miami River, Oldtown, US-68S near Clifton Rd.; corresponds to USGS site 3240000 Alum Creek, Kilbourne, County Rd. 34/N Galena Rd.; corresponds to USGS site 3228750

Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apgeochem.2014. 05.006. References Amrhein, C., Strong, J.E., 1990. The effect of deicing salts on trace metal mobility in roadside soils. J. Environ. Qual. 19, 765–772. Argento, D.C., Stone, J.O., Fifield, L.K., Tims, S.G., 2010. Chlorine-36 in seawater. Nucl. Instrum. Meth. B 268, 1226–1228. Braitsch, O., 1971. Salt Deposits; Their Origin and Composition. Springer-Verlag, New York. Chapra, S.C., Dove, A., Rockwell, D.C., 2009. Great lakes chloride trends: long-term mass balance and loading analysis. J. Great Lakes Res. 35, 272–284. City of Columbus, OH, 2014. Snow and Ice Plan. Department of Public Service. (last accessed 28.01.14. Craig, H., 1961. Isotopic variations in meteoric waters. Science 133, 1702–1703. Davis, S.N., Whittemore, D.O., Fabryka-Martin, J., 1998. Uses of chloride/bromide ratios in studies of potable water. Ground Water 36, 338–350. Davis, S.N., Cecil, L.D., Zreda, M., Moysey, S., 2001. Chlorine-36, bromide, and the origin of spring water. Chem. Geol. 179, 3–16. Davis, S.N., Moysey, S., Cecil, L.D., Zreda, M., 2003. Chlorine-36 in groundwater of the United States: empirical data. Hydrogeol. J. 11, 217–227. Elmore, D., Phillips, F.M., 1987. Accelerator mass spectrometry for measurement of long-lived radioisotopes. Science 236, 543–550. Foos, A., 2003. Spatial distribution of road salt contamination of natural springs and seeps, Cuyahoga Falls, Ohio, USA. Environ. Geol. 44, 14–19. Gardner, C.B., Carey, A.E., 2004. Trace metal and major ion inputs into the Olentangy River from an urban storm sewer. Environ. Sci. Technol. 38, 5319–5326. Hamed, K.H., Rao, A.R., 1998. A modified Mann–Kendall test for autocorrelated data. J. Hydrol. 204, 182–196. Heath, R.H., Kahl, J.S., Norton, S., 1992. Episodic stream acidification cause by atmospheric deposition of sea salts at Acadia National Park, Maine, United States. Water Resour. Res. 28 (4), 1081–1088.

35

Helsel, D.R., Hirsch, R.M., 2002. Techniques of water resources investigations of the United States Geological Survey. Book 4: Hydrologic Analysis and Interpretation. Statistical Methods in Water Resources. USGS, Reston, VA, 510p (Chapter A3). James, K.R., Cant, B., Ryan, T., 2003. Responses of freshwater biota to rising salinity levels and implications for saline water management: a review. Aust. J. Bot. 51, 703–713. Jacobs, A.N., 2006. Seasonal variations of major ions at two locations along the Olentangy River, Columbus, Ohio. B.A. Thesis, School of Earth Sciences, Ohio State Univ. Kaushal, S.S., Groffman, P.M., Likens, G.E., Belt, K.T., Stack, W.P., Kelly, V.R., Band, L.E., Fisher, G.T., 2005. Increased salinization of fresh water in the northeastern United States. Proc. Natl. Acad. Sci. USA 102, 13517–13520. Kelly, W.R., Panno, S.V., Hackley, K.C., Hwang, H., Martinsek, A.T., Markus, M., 2010. Using chloride and other ions to trace sewage and road salt in the Illinois Waterway. Appl. Geochem. 25, 661–673. Kelly, W.R., Panno, S.V., Hackley, K.C., 2012. Impacts of road salt runoff on water quality of the Chicago, Illinois, region. Environ. Eng. Geosci. 18, 65–81. Koryak, M., Stafford, L.J., Reilly, R.J., Magnuson, P.M., 2001. Highway deicing salt runoff events and major ion concentrations along a small urban stream. J. Freshw. Ecol. 16 (1), 125–134. Leslie, D., 2013. Applying stable isotopes d11B, d18O, and dD in geochemical and hydrological investigations. Ph.D. Thesis, School of Earth Sciences, Ohio State Univ. Mason, C.F., Fernandez, I.J., Norton, S.A., Katz, L.E., 1999. Deconstruction of the chemical effects of road salt on stream water chemistry. J. Environ. Qual. 28, 82– 91. McCaffrey, M.A., Lazar, B., Holland, H.D., 1987. The evaporation path of seawater and the coprecipitation of Br and K+ with halite. J. Sediment. Petrol. 57, 928– 937. Neal, C., Kirchner, J.W., 2000. Sodium and chloride levels in rainfall, mist, streamwater and groundwater at the Plynlimon catchments, mid-Wales: inferences on hydrological and chemical controls. Hydrol. Earth Syst. Sci. 4 (2), 295–310. Novotny, E.V., 2009. Road deicing salt impacts on urban water quality. Ph.D. Thesis, Univ. of Minnesota. Ohio Environmental Protection Agency (Ohio EPA), 2011. 2011 Annual Report, 34p. (last accessed 28.01.14). Ohio Environmental Protection Agency (Ohio EPA), 2014. Individual NDPES Permits, Division of Surface Water. (last accessed 28.01.14). Ohio Department of Development, 2013. Population Characteristics and Projections, Ohio’s Population. (last accessed 28.01.14). Ohio Department of Transportation (ODOT), 2008. Smart salt strategy: analysis of Ohio’s road salt market and 2008–2009 price increase. Bid Analysis and Review Team. (last accessed 28.01.14). Ohio Department of Transportation (ODOT), 2011. Snow and Ice Practices. Division of Operations, Office of Maintenance Administration. (last accessed 28.01.14). Ohio Water Resources Council, 2013. Recommendations for Salt Storage – Guidance for Protecting Ohio’s Water Resources. (last accessed 28.01.14). Oosting, S.E., Von Damm, K.L., 1996. Bromide/chloride fractionation in seafloor hydrothermal fluids from 9–10°N East Pacific Rise. Earth Planet. Sci. Lett. 144, 133–145. Panno, S.V., Hackley, K.C., Hwang, H.H., Greenberg, S.E., Krapac, I.G., Landsberger, S., O’Kelly, D.J., 2006. Characterization and identification of Na–Cl sources in ground water. Ground Water 44, 176–187. Stucker, J.D., 2013. The effects of urban land use type on low order stream geochemistry, Columbus, Ohio. M.S. Thesis, School of Earth Sciences, Ohio State Univ. Transportation Research Board, 1991. Highway Deicing: Comparing Salt and Calcium Magnesium Acetate. National Research Council, Special Report 235. U.S. Environmental Protection Agency (USEPA), 1988. Ambient Water Quality Criteria for Chloride—1988. Office of Water Regulations and Standards, EPA Pub. No. 440/5-88-001. U.S. Environmental Protection Agency (USEPA), 2002. Drinking water contaminants, national secondary drinking water regulations. Office of Ground Water and Drinking Water. (last accessed 28.01.14). Wilde, F.D., Radtke, D.B., Gibs, Jacob, Iwatsubo, R.T., 2004. Processing of Water Samples (Ver. 2.2): U.S. Geological Survey Techniques of Water-Resources Investigations, Book 9 (Chapter A5). (last accessed 28.01.14).