Health Canada
www.hc-sc.gc.ca
Common menu bar links
Institutional links
Priority Substances List Assessment Report for Road Salts
3.0 Summary of Critical Information and Assessment of "Toxic" under CEPA 1999 (Continued)
3.1 CEPA 1999 64(a): Environment and CEPA 1999 64(b): Environment on which life depends
The approach for the environmental assessment of PSL substances involves characterizing the entry of, exposure to and effects of a substance, which ultimately permits the characterization of the risk posed by the substance (Environment Canada, 1997a). This may involve the use of both laboratory and field-derived data. Empirical field evidence linking environmental changes to entry of the substance can provide strong indications of actual risk.
It can be assumed that all road salts eventually enter the environment, whether from losses from storage or patrol yards, from roadway application or from disposal of snow. Although all chloride ions will ultimately be found in surface waters, ions from inorganic chloride road salts and ferrocyanides will be released to or move through soil, groundwater, surface water and air, and organisms can be exposed to road salts through all these media. Therefore, all environmental compartments are of concern for the assessment of road salts. Environmental effects can be either biotic or abiotic. Thus, endpoints of concern can include effects on individual organisms, populations or communities, as well as abiotic endpoints, such as deterioration of soil structure or lake stratification, which disrupts seasonal mixing of waters.
While there has not been systematic assessment of exposure to road salts for any given compartment across Canada that would permit detailed quantitative risk characterization at the national scale, many individual studies are available that permit the establishment of relationships between the use of road salts and their impacts on biotic and abiotic systems.
The following sections characterize the risks associated with releases of road salts into the environment, by reviewing the entry, exposure and effects related to five environmental compartments or biotic groups with regard to inorganic chloride salts. Data pertinent to ferrocyanides and their environmental risks are reviewed in a separate section (Section 3.7).
3.2 Groundwater
3.2.1 Introduction
This section summarizes the fate and impact of road salts on groundwater quality. Discussion in this section is based on a report prepared by Johnston et al. (2000). A summary of case studies (Morin, 2000) is also presented.
Data on concentrations of salts in groundwater are frequently discussed in published and unpublished literature in terms of exceedences of drinking water quality guidelines. These guidelines are set on the basis of taste thresholds at 200 mg sodium/L and 250 mg chloride/L (CCME, 1998). Although the use of groundwater as a drinking water resource is not considered in this assessment, a review of reported exceedences of a chloride concentration of 250 mg/L is pertinent to the assessment of potential impacts on aquatic biota. As reviewed in Section 3.3, the No-Observed-Effect Concentration (NOEC) for the 33-day survival of early life stage fathead minnow was 252 mg chloride/L. Furthermore, lethality data modelled for chronic exposure indicated that 10% of aquatic species would be affected at chloride concentrations of about 240 mg/L.
3.2.2 Groundwater principles
3.2.2.1 Hydrological cycle
The degree to which road salts may have an impact on the groundwater environment will vary significantly, depending on the road salt loading, climatic conditions, surface and subsurface soil conditions and location of the site within the overall hydrogeological environment.
Precipitation that falls across an area as rain or snow contributes to a number of hydrological processes. A portion of the precipitation will run off as overland flow from impervious surfaces, such as roadways and parking lots, and is eventually directed by open ditches or subsurface piped systems that discharge to streams, wetlands, lakes or other surface water features. Overland flow will result in the direct migration of surficial contaminants, like road salts, to surface water features, typically in the time frame of hours.
In certain settings, a portion of precipitation will infiltrate the shallow soil zone and move laterally through the unsaturated zone above the water table as interflow to the local discharge area. The transport of contaminants to surface water features as interflow typically occurs in the time frame of hours to days.
The balance of precipitation is subject to evapotranspiration or infiltrates as groundwater recharge to the water table. Contaminants introduced to the local, intermediate or regional groundwater flow systems may take several years to several hundred years before discharging as baseflow to surface water features.
3.2.2.2 Unsaturated flow and groundwater recharge
The amount of precipitation that infiltrates to the water table depends on a number of factors, including the duration and intensity of the rainfall event, the initial soil moisture content and the soil moisture characteristics. Based on the general temperature and precipitation pattern in Canada, the majority of groundwater recharge in a year is expected to occur in the late winter and early spring, as a result of winter snowmelt and spring rains (Gerber and Howard, 1997; Singer et al., 1997).
Vertical travel times through the unsaturated zone may be expected to range from less than 0.5 m/year to more than 20 m/year. In areas with thick unsaturated zones (i.e., 20-30 m), travel times for infiltrating precipitation to reach the water table may be in the range of 5-20 years in sandy soils. These travel times are important when considering the impacts to downgradient receptors and the time required before remedial measures implemented at the source would have noticeable effects.
3.2.2.3 Saturated flow and contaminant migration
Once below the water table, the movement of water is controlled by the fluid potential and the permeability of the geological material. When evaluating the migration of solutes, such as sodium, calcium and chloride ions, within groundwater environments, several other physical and chemical processes must be considered. Mechanical dispersion, molecular diffusion and density effects represent the primary physical transport mechanisms that affect the migration of sodium, calcium and chloride ions.
3.2.3 Impacts of road salts on groundwater quality
3.2.3.1 Factors controlling the migration of road salts
One of the primary concerns associated with inorganic sodium chloride road salts is their high solubility in water. The dissociated ions will migrate with infiltrating precipitation to the water table. The migration rate of chloride ion will be the same as that of the water. This is not the case for cations such as the sodium ion, which can be affected by cation exchange reactions in certain geological environments. The following equation presents the ion exchange reaction for sodium ion:
2Na+ + Ca(adsorbed) <> Ca2+ + 2Na(adsorbed)
In the equation, sodium ion is exchanged with calcium adsorbed to clay-rich materials. The net effect of this reaction is a decrease in sodium ion concentrations within the infiltrating groundwater and a Na:Cl ratio of less than 1. Reduced sodium ion concentrations in groundwater impacted by road salts have been documented by Howard and Beck (1993) and Pilon and Howard (1987).
3.2.3.2 Predicted road salt concentrations in groundwater
To provide an understanding of the impact that road salting may have on groundwater quality, an estimate of the amount of road salt available for dissolution in infiltrating precipitation is required. A simple spreadsheet mass balance model was developed to provide an indication of the equilibrium chloride ion concentration in shallow groundwater (Johnston et al., 2000). The model is based on the dissolution of the total chloride ion retained within the soil by recharging precipitation and does not consider any additional chloride ion from other sources. The calculated chloride ion concentrations can therefore be considered representative of regional-scale equilibrium chloride ion concentrations downgradient of a saltable road network. It should be noted that the model does not provide an indication of potential chloride ion concentrations immediately downgradient of a source area, such as an individual roadway, where higher chloride ion concentrations may be expected.
Input to the model includes groundwater recharge rates, road salt loadings and road densities typical of southern Ontario. Recharge rates ranging from 10 to 40 cm/year were selected based on a range of soil types. The chloride ion loadings were based on typical road salt application rates for both provincial and municipal road networks, as discussed in Section 2.2. The road salt available for infiltration to the water table was estimated based on the ranges reported by Wulkowicz and Saleem (1974), Scott (1980) and Howard and Haynes (1993). Road densities in the range of 1-15 two-lane-kilometres per square kilometre were selected based on the data presented in Morin and Perchanok (2000) and Stantec Consulting Ltd. (2000). Road densities of 0.01-0.1 two-lane-kilometres per square kilometre were assumed to be characteristic of low-density rural land use, road densities of 1-5 two-lane-kilometres per square kilometre were assumed to be characteristic of medium-density rural to semi-rural land use, and road densities of 10-15 two-lane-kilometres per square kilometre were assumed to be characteristic of urban land use.
Predicted chloride ion concentrations for a given recharge rate increase with increased loading, road network density and percentage chloride ion retained. Chloride concentrations decrease with increasing recharge rates due to the dilution effects of increased recharge. The highest chloride ion concentration, 3600 mg/L, was predicted for soil with a recharge rate of 10 cm/year, a loading of 40 tonnes chloride ion per two-lane-kilometre and 60% of the chloride ion applied available for dilution in recharging precipitation. For comparison, the chloride ion concentration for the same loading and percentage retained is 900 mg/L for soils with a recharge rate of 40 cm/year. It is expected that the percentage of chloride ion retained will vary significantly between a site with a recharge rate of 10 cm/year versus 40 cm/year. This is due to the fact that runoff would be higher for the site with a recharge rate of 10 cm/year, and therefore less chloride ion should be retained for dissolution in infiltrating precipitation.
Figures 12-14 present regional-scale groundwater chloride ion concentrations that were predicted for various road salt loadings, road densities and recharge rates. In each figure, an envelope corresponding to the maximum and minimum chloride ion concentrations is presented. Average chloride ion concentration envelopes for medium- and high-density road networks are also presented, assuming the average percentage chloride ion retained of 30% and 40%.
Figure 12 Estimated chloride concentrations in groundwater for various chloride application rates and a groundwater recharge rate of 10 cm/year (from Johnston et al., 2000)
Figure 13 Estimated chloride concentrations in groundwater for various chloride application rates and a groundwater recharge rate of 20 cm/year (from Johnston et al., 2000)
Figure 14 Estimated chloride concentrations in groundwater for various chloride application rates and a groundwater recharge rate of 40 cm/year (from Johnston et al., 2000)
For soils with recharge rates of 10 cm/year, typical of clay-rich soils, chloride ion concentrations may exceed 250 mg/L for chloride ion loadings of approximately 5 tonnes chloride per two-lane-kilometre under high-density road networks (Figure 12). Under medium road densities, chloride ion concentrations are predicted to remain below 250 mg/L for road salt loadings of approximately 10-15 tonnes chloride per two-lane-kilometre. For low-density road networks, typical of a rural setting, regional-scale chloride concentrations are predicted to remain well below 250 mg/L (Johnston et al., 2000).
Figure 13 presents the estimated groundwater concentrations for a soil with a recharge rate of 20 cm/year, typical of sand. Due to dilution from the increased recharge, slightly higher chloride ion loadings can be supported before a concentration of 250 mg/L is exceeded. Under high-and medium-density road networks, chloride ion loadings above 11 tonnes chloride per two-lane-kilometre and 25-34 tonnes chloride per two-lane-kilometre, respectively, may result in regional-scale chloride ion concentrations above 250 mg/L.
At recharge rates typical of very permeable soils, such as gravels or fractured rock, groundwater chloride ion concentrations exceed 250 mg/L only for high-density road networks. In this case, chloride ion loadings above about 16-22 tonnes chloride per two-lane-kilometre are predicted to result in concentrations above 250 mg/L (Figure 14).
The mass balance modelling completed for this section indicates that road salt loadings above approximately 15 tonnes chloride per two-lane-kilometre, or 24 tonnes sodium chloride per two-lane-kilometre, will result in regional-scale groundwater concentrations greater than 250 mg/L in any soil condition under high-density land uses. Considering that road salt loadings in southern Ontario, southern Quebec and the Maritime provinces are in the range of 20-50 tonnes sodium chloride per two-lane-kilometre, highly developed urban areas in these provinces are at the greatest risk of wide-scale groundwater impacts associated with road salting. In addition to groundwater impacts, surface water quality within these areas of Canada will be subject to the discharge of groundwater with elevated sodium and chloride ion concentrations as baseflow, particularly during the late summer and fall.
The mass balance modelling indicates that under low-density land use, chloride ion concentrations will remain well below 250 mg/L, reaching a maximum concentration of 24 mg/L for road salt loadings of 40 tonnes sodium chloride per two-lane-kilometre. It should be noted, however, that this concentration represents an average regional concentration, and higher concentrations may be expected immediately downgradient from individual roadways.
3.2.3.3 Plume migration and impact assessment
The above section provides an overview of the regional impact associated with non-point source loading from road salting in rural and urban areas. The impact on groundwater quality and plume development from point source loading associated with road salting along an individual roadway is discussed below.
To provide an indication of maximum chloride ion concentrations and the effects of dispersion on plume concentrations, a numerical solute transport model was developed for a simple unconfined sand aquifer system using the finite difference modelling programs MODFLOW (McDonald and Harbaugh, 1988) and MT3D (Zheng, 1990). The simplified aquifer was assigned a lateral hydraulic conductivity of 1 x 10-4 m/s, vertical hydraulic conductivity of 1 x 10-5 m/s, specific yield of 0.25, effective porosity of 0.30, longitudinal dispersivity of 10 m, and lateral and vertical transverse dispersivities of 1 m and 0.1 m, respectively. Linear constant-head boundaries were assigned at the upgradient (17.5 m) and downgradient (15 m) extents of the model to simulate a lateral hydraulic gradient of 0.005 m/m across the 500-m model length. Recharge was assigned at 30 cm/year to the entire model domain.
Within the simulated steady-state flow field, a square constant concentration source term of 1000 mg/L was assigned at the water table to represent the chloride ion loading from a section of roadway. This concentration is within the range (323-2930 mg/L, average 923 mg/L) for shallow groundwater beneath a four-lane arterial road in Kitchener, Ontario (Stantec Consulting Ltd., 2000). The formation and migration of an unretarded, dissolved-phase chloride ion plume was simulated for 5000 days, followed by the dissipation of the plume after removing the concentration source term.
At a simulated time of 100 days, typical of early summer following recharge of groundwater with elevated chloride ion concentrations from winter road salting, groundwater concentrations parallel to the plume showed the formation of a typical "cigar-shaped" plume as defined by the 10 mg/L contour. At 100 days, the 10 mg/L contour has migrated approximately 40 m downgradient from the source area, with the 250 mg/L contour restricted to within approximately 10 m of the source area. After a period of approximately 5 years, the plume has reached a steady-state condition as defined by the 10 mg/L contour in which the flux from the source area is balanced by the flux across the plume boundary. The 10 mg/L contour under steady-state conditions is located approximately 200 m downgradient from the source, while the 250 mg/L contour moved 20 m downgradient from the source area.
After a simulation period of approximately 5 years, the chloride ion source was removed to provide an indication of the time required for concentrations to dissipate. Chloride concentrations decreased rapidly after 50 days, with no simulated concentrations above 250 mg/L.
The above simulations indicate that impacts associated with road salting along individual highways may be limited in extent due to groundwater flow and transport processes, depending on the aquifer properties. Based on the modelling results, shallow wells and surface water features located within 20 m of salted roads are at the greatest risk of chloride ion impacts. The modelling does suggest that if road salting activities are discontinued, chloride ion concentrations within the shallow aquifer may improve significantly over time frames of months to years.
3.2.4 Case studies
This section presents modelling and field measurements for specific areas in Canada. Additional case studies involving impacts on groundwater quality associated with roadways and maintenance salt storage yards are presented in Morin (2000).
One method for assessing the regional-scale impacts associated with road salting was that applied by Howard and Haynes (1993) in the Highland Creek catchment located near Toronto, Ontario. Through detailed measurements of stream flow and chloride ion concentrations over a 3-year period, it was estimated that only 45% of the chloride ion applied as road salt within the basin was removed by surface water runoff. The remaining 55% was interpreted to be stored within the groundwater system. At current salt application rates, Howard and Haynes (1993) estimated an average chloride ion concentration of 426 ± 50 mg/L in groundwater discharging as baseflow to Highland Creek. While road salt application rates were not presented, the total loading over the basin was approximately 200 g sodium chloride/m2. Assuming a road density typical of an urban environment, the road salt application rate over the study area is estimated to be in the range of 20-30 tonnes sodium chloride per two-lane-kilometre, which is similar to that predicted based on the mass balance modelling completed as part of this report (Section 3.2.3.2).
Further evidence of the impact on groundwater quality and discharge to surface water features due to road salting in urban areas is presented in Howard and Beck (1993) and Eyles and Howard (1988). Chloride concentrations in springs discharging to Lake Ontario, along the Scarborough Bluffs near Toronto, were measured and found to be approximately 400 mg/L, similar to those estimated by Howard and Haynes (1993) for the Highland Creek watershed.
Long-term chloride ion concentrations from municipal supply wells provide further evidence of impacts on groundwater quality associated with winter road salting (Johnston et al., 2000). Figure 15 presents chloride ion concentrations from municipal water supply wells in an urban centre of southern Ontario. Data on current chloride ion application rates in the study area near the wells indicate that application rates have ranged from 18 to 36 tonnes sodium chloride per two-lane-kilometre, with an average application rate of 28 tonnes sodium chloride per two-lane-kilometre. Chloride concentrations beginning in the early 1960s were less than 50 mg/L and increased steadily to 250-300 mg/L in the late 1990s (Figure 15) (data after 1998 not plotted). One of the key issues with respect to the trends indicated in Figure 15 is that chloride ion concentrations are continuing to rise and have not reached steady state. Since 1998, chloride concentrations at these wells have increased at a rate similar to that seen since 1993 on Figure 15, with Well 3 now over 350 mg/L. However, it is noted that the groundwater being pumped by these wells represents precipitation that recharged the aquifer in the mid to late 1970s, with travel times from ground to surface on the order of 30 years. Hence, increased chloride concentrations observed at these production wells are the result of road salts that were spread 30 years ago. Given that these wells pump on average 3.3 million cubic metres per year and have 20-year capture zones that extend out over 12 km2, the concentrations predicted at these wells are considered to reflect regional-scale chloride concentrations. A comparison of the current concentrations of approximately 350 mg/L with the predicted concentrations of 300-450 mg/L indicates that the mass balance modelling approach presented in the previous sections provides reasonable estimates of chloride concentrations (Johnston et al., 2000).
Figure 15 Chloride concentrations in groundwater from municipal production wells in southern Ontario (from Johnston et al., 2000)
The time to steady state (the time when salt inputs are balanced by salt output) depends on local hydrogeological conditions and the size of the catchment area (Howard et al., 1993). Howard et al. (1993) used two numerical models to demonstrate temporal and spatial changes in water quality. The model FLOWPATH showed that chemically conservative contaminants such as chloride released within a few kilometres of rivers and Lake Ontario will discharge in about 5 years. Contaminants that are released in more central areas will take more than 100 years to be flushed from the groundwater system. When road salt is evenly distributed in a representative 460-km2 region of the Greater Toronto Area, Howard et al. (1993) demonstrated, using the FLOWPATH model, that the average chloride concentrations will reach steady-state values within 200 years of initial salt application. Using another model, AQUA, in a smaller area near Toronto, steady state was reached in just 30 years. At steady state, the levels of chloride at a distance of 200 m from the salted highways are 2-3 times the levels of chloride observed in the discharging baseflow.
Williams et al. (1997) investigated chloride concentrations in 20 springs in southeastern Ontario. Chloride concentrations ranged from 8.1 to 1149 mg/L. The higher chloride concentrations originated from groundwater that may be contaminated by road salt. Williams et al. (1999) continued this research, noting that the mean chloride concentrations were 2.1 mg/L at the Glen Majors Conservation Area and about 100 mg/L in rural areas. Chloride concentrations in springs near urban areas were higher (>200 mg/L), with the maximum concentration 1345 mg/L and mean 1092 mg/L. The spring with highest concentrations was adjacent to a highway and bridge. Chloride concentrations at these urban sites also increased between November 1996 and November 1997. Salinity contamination was related to road salt. Williams et al. (1999) noted that spatial patterns in road salt contamination were more readily detected by sampling springs than by sampling creeks, because chloride concentrations in spring waters (i.e., discharging groundwater) exhibited relatively little seasonal variability. In contrast, chloride concentrations in creeks were highly variable seasonally.
Howard and Taylor (1998) reported high concentrations of chloride in springs discharging from aquifers along the Scarborough Bluffs, a stretch of Lake Ontario shoreline east of Toronto, and along an urban-rural transect from the Oak Ridges Moraine to Metropolitan Toronto. Concentrations exceeding 2800 mg chloride/L were detected in springs issuing from shallow aquifers in the Oak Ridges Moraine-Toronto area. The primary source of the chloride contamination was road de-icing chemicals (Howard and Beck, 1993).
In 1970, an expressway was constructed at a distance of 50-200 m from wells used by the municipality of Trois-Rivières-Ouest to obtain groundwater. By 1976, an increase in salinity was noted (Gélinas and Locat, 1988). In the early 1980s, a plume of salty groundwater was identified. Chloride concentrations in samples from piezometers ranged between 3 and 1500 mg/L. Data collected in 1985 revealed that the shape and extent of the plume varied depending on the time of year and the quantity of water pumped from each of the six municipal pumping stations. Data for 1985-86 indicate that concentrations gradually increase throughout the spring (during snowmelt) to peak in August and September. Higher concentrations were observed for a municipal pumping station located downgradient from the highway. Concentrations at most wells upgradient from the highway were less than 5 mg/L. Mean chloride concentrations in one station downgradient (80 mg/L) were about 6 times higher than the average concentration in a nearby station upgradient (14 mg/L).
The municipality of Cap-de-la-Madeleine pumps water from 27 wells. While the quality of this water was considered to be extremely good, certain wells have experienced a gradual increase in the concentration of chloride since the 1970s that is attributed to the use of de-icing road salts. In areas where ditches were not connected to a drainage system, runoff is left to percolate through the soils (Delisle, 1999).
3.2.5 Conclusions
The potential for impacts on groundwater and surface water quality was evaluated using a mass balance technique that provides an indication of potential regional-scale chloride ion concentrations downgradient of saltable road networks. The mass balance modelling indicates that for road salt application rates above 20 tonnes sodium chloride per two-lane-kilometre, regional-scale groundwater chloride ion concentrations greater than 250 mg/L will likely result under high-density road networks, typical of urban areas. Considering road salt application rates throughout Canada, urban areas in southern Ontario, southern Quebec and the Maritime provinces are at the greatest risk of wide-scale groundwater impacts associated with road salting. The mass balance modelling completed as part of this report indicates that rural and semi-rural areas are not as likely to be impacted by chloride ion concentrations above 250 mg/L on a regional scale. However, local impacts along individual roadways have been documented.
Groundwater containing road salts will eventually upwell into the surface water or emerge as springs. Research has shown that only a portion of the road salts that are applied to roads are removed by surface water runoff, and a significant proportion of the salt may be stored within the groundwater system. Elevated concentrations of chlorides have been detected in both groundwater and groundwater that has emerged to the surface as springs. Predicted and measured concentrations in groundwaters and springs exceed those identified as lethality thresholds for many organisms, as identified in Section 3.3 (e.g., 10% of species can be expected to be affected by concentrations greater than 240 mg/L).
3.3 Aquatic ecosystems
3.3.1 Scoping and assessment approach
Aquatic ecosystems are vulnerable to various impacts from road salts. Evans and Frick (2001) assessed such impacts, focusing on a) a literature review of the effects of road salt applications on aquatic ecosystems and b) the lethal and sublethal effect levels of sodium chloride and other chloride salts as determined from laboratory studies. These studies were then used in the characterization of the risks of road salts. Highlights of this report are presented in this section of the Assessment Report.
The aquatic ecosystems that were considered in this assessment included a wide variety of habitats, including streams, rivers, wetlands, ponds and lakes. These ecosystems were located across Canada and in a wide variety of regions, some experiencing minimal and others intense anthropogenic stress. Many aquatic ecosystems are located in relatively pristine areas of Canada, where anthropogenic impacts are minimal. Nevertheless, such impacts are of concern, because pristine habitats are vulnerable to relatively minor anthropogenic stresses. Small shifts in species composition, particularly for phytoplankton, are highly indicative of such impacts. This is because the species assemblage of pristine environments is, in large measure, composed of taxa that are adapted to unperturbed environments; pollution-tolerant species generally are minor constituents of such assemblages. Therefore, there is the concern that small increases in chloride concentrations from highway runoff or loss from salt storage depots may have measurable impacts on such communities.
Other aquatic ecosystems are located in rural and agricultural areas. Taxa inhabiting these environments are experiencing a variety of stresses as a result of a variety of anthropogenic activities, including habitat loss and increased chemical inputs. With increased development in the Canadian countryside, including the loss of naturally occurring ponds and wetlands to agricultural usage, the narrow tens of metres of space on the sides of roadways may provide important habitat for displaced plant and wildlife communities. Moreover, this land can serve as migration corridors between larger, relatively undisturbed areas (Nadec, 2001).
Aquatic ecosystems are also found in urban areas, with many such aquatic ecosystems gradually being incorporated into the growing urban landscape. In such areas, provision is made to protect these ecosystems by limiting development around their immediate shoreline and by minimizing anthropogenic impacts in their immediate watershed. Well-known examples of such incorporated parkland areas include those established around the many creeks running through the Metropolitan Toronto area and parks such as Stanley Park in Vancouver. These aquatic systems, while no longer pristine, nevertheless merit protection against further environmental degradation. Increasingly, the design of subdivisions includes the incorporation of rural features such as lakes and creeks. Many of these engineered ecosystems are built within the natural landscape over existing aquatic habitats (e.g., ponds, wetlands, sloughs, creeks), which are rapidly recolonized by temporarily displaced aquatic organisms.
Stormwater detention ponds, which are often constructed in a natural aquatic ecosystem, e.g., a creek or wetland, are engineered habitats. Nevertheless, these ponds do provide habitat for a wide variety of organisms (Bishop et al., 2000). While clearly not ideal because of the variety of contaminants associated with them and altered hydrological regime, they provide aquatic habitat and resources for both the organism and the population (which requires a critical area for self-maintenance) in an increasingly fragmented urban and rural landscape.
Aquatic organisms considered in the assessment included all components of the aquatic ecosystem, i.e., bacteria, fungi, protozoans, zooplankton, benthic invertebrates, fish and amphibians. However, not all groups were equally well represented in the literature. For example, laboratory studies investigating chloride toxicity were heavily weighted towards fish, zooplankton and benthic invertebrates and for short-term exposures. Such studies, by their nature, focus on the capacity of species to endure short-term osmotic stress due to elevated salinities. Few studies investigated chronic toxicity; chronic toxicity was estimated using approaches followed by the U.S. EPA (1988) for setting water quality guidelines for chloride.
Relatively few field studies have been conducted specifically investigating the impacts of road salts on aquatic ecosystems. The majority of these studies were conducted in the 1970s and 1980s. Many of the studies that were found were conducted in the United States but have general application to the Canadian environment.
As provided for in CEPA 1999 and as noted in Environment Canada (1997a), a weight of evidence approach involving consideration of several lines of evidence is used to strengthen the confidence in assessment conclusions. These lines of evidence include consideration of short-term toxicity studies, estimates of acute toxicity, field studies, emerging information and the general knowledge of the use patterns of road salts in the Canadian environment. In addition, a tiered approach, using quotient-based assessments, is used to determine the likelihood of adverse effects occurring in aquatic ecosystems as a result of release of road salts into the environment.
3.3.2 Laboratory studies
The literature review of laboratory studies of chloride toxicity focused on obtaining information on sodium chloride toxicity, but also collected data for calcium, potassium and magnesium chloride toxicity, since these chlorides are also present in various road salt formulations. Toxicity data are essential for assessing the environmental concentrations and exposure durations that may be harmful. Both short-term (hours to days) toxicity and long-term (weeks to months) toxicity are of concern. In the following paragraphs, the highlights of these determinations are presented.
3.3.2.1 Short-term or acute toxicity studies
One exposure scenario that is of concern is short-term exposures (hours to days) to elevated chloride concentrations, particularly when the dilution capacity of the receiving water body is weak. Direct highway runoff, particularly into creeks, streams and other roadside waterways, is of particular concern. Organisms inhabiting larger water bodies such as rivers and lakes, with their greater dilution capacity, are less likely to be exposed to conditions in which acute toxicity is of concern. Organisms inhabiting wetlands, which are relatively small, stationary water bodies with slow water exchange, may experience significant acute and chronic toxicity.
An extensive review of the literature revealed several dozen studies that investigated the short-term or acute toxicity of sodium chloride to aquatic organisms; fewer studies were located for magnesium, calcium and potassium salts (Evans and Frick, 2001). These acute studies were subsequently grouped into four time intervals to estimate toxicity following various exposures to elevated chloride concentrations:
- <24 hours: For exposures of less than 1 day, four studies were located for fish (3) and a benthic organism (1) (Table 10). The LC50s ranged from 6063 to 30 330 mg chloride/L, with a geometric mean of 12 826 mg chloride/L. Natural salinities in this range are associated with estuaries, tidal marshes, oceans and inland saline lakes. Salinities in this range have also been associated with highway runoff (and spray) from multiple-lane highways, waste snow from urban areas, leachate and groundwater from patrol yards. Elevated chloride concentrations at the lower end of this range have also been observed in well water, a stream and a wetland near a road salt depot (see Sections 2.3, 2.4, 2.6 and 3.4).
- 24 hours: For exposures of 24 hours, seven studies were located testing fish (4) and cladocerans (3) (Table 11). The LC50s ranged from 1652 to 8553 mg chloride/L, with a geometric mean of 3746 mg chloride/L. Salinities in this range are associated with estuaries, tidal marshes and inland saline lakes. Salinities in this range have also been associated with highway runoff (and spray) from multiple-lane highways, waste snow from urban areas and leachate from patrol yards. Elevated chloride concentrations at the lower end of this range have also been observed in well water, a stream and a wetland near a road salt depot (see Sections 2.3, 2.4, 2.6 and 3.4).
- 3 or 4 days: Several studies were found investigating the 4-day toxicity of sodium chloride to a variety of organisms, and a smaller set of 3-day exposures was also found. The 3-day exposure data were converted into 4-day estimates using a conversion factor as described in Evans and Frick (2001). This resulted in 28 observations, including fish (13), cladocerans (7) and other invertebrates (8) (Table 12). Some species were the subject of several studies, e.g., Daphnia magna and fathead minnow (Pimephales promelas). There were differences in some LC50s, probably as a result of test conditions (e.g., use of reconstituted water in Birge et al., 1985) and/or differences in the physiological tolerances of the test organisms. In general, fish had greater tolerances than invertebrates. The LC50s ranged from 1400 to 13 085 mg chloride/L, with a geometric mean of 4033 mg chloride/L. The higher 4-day than 1-day tolerances is due, in large measure, to the inclusion of mosquito fish (Gambusia affinis, a hardy fish that is used worldwide for mosquito control) and American eel (Anguilla rostrata, a fish that spends its adult life in the ocean) in the data set. Salinities in this range are associated with estuaries, tidal marshes and inland saline lakes. Salinities in this range have also been associated with highway runoff (and spray) from multiple-lane highways, waste snow from urban areas and leachate from patrol yards. Elevated chloride concentrations at the lower end of this range have also been observed in well water, a stream and a wetland near a road salt depot (see Sections 2.3, 2.4, 2.6 and 3.4). Moreover, salinities in the 1000-4000 mg/L range have been reported for several creeks in the Metropolitan Toronto area.
- 7-10 days: Seventeen studies were found estimating chloride toxicity at exposure times of 7-10 days (Table 13). Exposures of this duration are at the upper end of acute exposure times but are not sufficiently long to be considered chronic for long-lived organisms such as amphibians and fish. These studies examined responses such as mortality, reduced growth, fecundity and failure to complete development, e.g., from egg to embryo. There were five fish tests, two amphibian tests, nine zooplankton tests and one algal test. The LC50s and EC50s ranged from 874 to 3660 mg chloride/L, with a geometric mean of 1840 mg chloride/L.
1 LC50/EC50 ratio calculated for studies where both parameters were estimated (i.e., Cowgill and Milazzo, 1990; Beak, 1999).
This salinity range is defined as subsaline (Hammer, 1986). Such salinities are found at the interface between the marine and freshwater realms and in inland subsaline lakes. Species diversity declines rapidly with increasing salinity in this range (Wetzel, 1983). Salinities in this range have been associated with highway runoff (and spray), waste snow, leachate from patrol yards, well water, a stream, a wetland near a road salt depot, and creeks and rivers in the Metropolitan Toronto area.
3.3.2.2 Chronic toxicity
Long-term or chronic toxicity to sodium chloride, which may be expected to occur over weeks to months, has seldom been investigated in the laboratory. Only two studies were located in the literature review.
Birge et al. (1985) estimated the 4-day LC50 toxicity of chloride to fathead minnow as 6570 mg/L. The NOEC for the 33-day early life stage test was 252 mg chloride/L and the Lowest-Observed-Effect Concentration (LOEC) was 352 mg/L, giving a geometric mean of 298 mg chloride/L as the estimated chronic value and an acute to chronic ratio (ACR) of 22.1. Birge et al. (1985) also tested Daphnia pulex. The 4-day LC50 was 1470 mg chloride/L. For the 21-day test, the chronic toxicity, calculated as the geometric mean of the NOEC (314 mg/L) and the LOEC (441 mg/L), was 372 mg chloride/L. This value is similar to that for fathead minnow. However, because of the lower acute toxicity value for Daphnia, the ACR was 3.95. Birge et al. (1985) conducted a second 4-day toxicity test with D. pulex using natural stream water and reported an EC50 of 3050 mg chloride/L, or twice the tolerance of that using reconstituted water. (If this value were used to estimate the ACR, the value would become 8.20.) Birge et al. (1985) calculated the geometric mean of the Daphnia and fathead minnow chronic toxicity values to estimate a chronic chloride toxicity concentration of 333 mg/L, although they also indicated some uncertainty in the D. pulex values.
In a later study, the U.S. EPA (1988) developed water quality criteria for chloride, relying largely on the results of the Birge et al. (1985) study for acute toxicity and ACR estimates. They used the same toxicity data and ACR (3.95) for Daphnia pulex. However, for fathead minnow, they estimated the chronic toxicity as 433.1 mg/L (based on the geometric mean of 9% impaired survival at 352 mg/L and 15% impaired survival at 533 mg/L), giving an ACR of 15.17. They also cited a rainbow trout (Salmo gairdneri, now Oncorhynchus mykiss) study conducted by Spehar (1987), who reported an acute toxicity of 6743 mg chloride/L, a chronic toxicity of 922.7 mg chloride/L and an ACR of 7.31. The geometric mean of these three ACRs was used to develop a final ACR of 7.59. A final acute genus toxicity value of 1720 mg/L, based on the most sensitive genus (Daphnia), was divided by this ACR to give a final chronic toxicity value of 226.5 mg/L.
A modification of the ACR approach was used to estimate the chronic toxicity of the organisms for which there were 4-day acute toxicity data. Two approaches could have been used. The first was to apply the Daphnia ACR (3.95) as developed by Birge et al. (1985) and later used by the U.S. Environmental Protection Agency (U.S. EPA, 1988) to the invertebrate data and the geometric mean (10.53) of the fish ACR data reported by the U.S. EPA. However, Birge et al. (1985) considered the acute toxicity data (and ACR) to be too low. Application of two different ratios also changes the general ordering of species sensitivities based on the 4-day acute toxicity data, so that fish now appear more sensitive and aquatic invertebrates more tolerant to long-term elevated chloride exposures. This seems highly unlikely, with aquatic organisms such as cladocerans and insect larvae being poorly represented in the marine and estuarine environment, while fish taxa such as the salmonids have strong marine affinities. A more suitable approach, then, is to follow the U.S. EPA approach and to use the mean ACR of 7.59 based on the geometric mean of the three ACR studies.
Using a mean ACR of 7.59, the acute toxicity of chloride was estimated as ranging from 184.5 mg/L for Ceriodaphnia dubia to 1724 mg/L for the black eel stage of the American eel, with a geometric mean value of 512.6 mg/L chloride (Evans and Frick, 2001). The lower value is approximately 24 times greater than the average chloride concentration (8 mg/L) of the world's river waters (Wetzel, 1983). The geometric mean of the chronic toxicities, 551.9 mg chloride/L, is at the lower end of the 500-1000 mg chloride/L range, where many researchers have reported losses in freshwater species with increasing salinity (Williams, 1987; Hart et al., 1990; Leland and Fend, 1998; Evans and Frick, 2001). A marked drop in the species diversity of freshwater ecosystems occurs as salinity increases from about 2000 mg/L to about 5000 mg/L (Figure 16; see also Wetzel, 1983). In general, for freshwater species, the number of species decreases as salinity increases, with the greatest and most rapid decrease in species numbers at chloride ion concentrations of 1000-3000 mg/L. Salinities in this 185-1724 mg/L range have been associated with highway runoff (and spray), waste snow, leachate from patrol yards and creeks in the Metropolitan Toronto area. The lower salinity values have been associated with chloride-impacted lakes.
3.3.3 Assessment of toxicity to aquatic ecosystems
Three approaches were used to assess the toxicity of road salts in the aquatic environment. These are a) toxicity thresholds and quotient-based risk characterization, as described in Environment Canada (1997a); b) field evidence of population or ecosystem effects; and c) consideration of abiotic effects. Briefy, these approaches can be described as follows:
- Toxicity thresholds and quotient-based risk characterization: This is a quotient-based approach, which is used to determine the likelihood of adverse effects occurring in aquatic ecosystems as a result of release of road salts into the environment. This approach is based on comparisons of concentrations of chloride in the aquatic environment resulting from the release of road salts (based on literature and analyses done for this assessment) and toxicity thresholds for chloride salts as identified in laboratory studies.
- Field evidence of population or ecosystem effects: The quotient-based approach generally focuses on the consideration of toxicity thresholds. However, community structure can be affected by environmental conditions that, while not necessarily exposing organisms to concentrations of contaminants that are themselves toxic, can result in selective advantage for certain populations or species and thereby result in shifts in populations and changes in community structure. Thus, field studies are particularly well suited for investigating the chronic effects of elevated chloride concentrations in the environment. However, because not all variables are controlled in such studies, it is possible that other, covarying variables affect some observed responses.
- Abiotic effects: Through their interactions with physical, chemical or biological components of the environment, substances may alter the abiotic environment and thereby affect ecosystems. Road salts may operate in two important ways. The first is by affecting the density of water and hence its mixing properties. This is especially important in relatively stationary water bodies such as lakes, ponds and wetlands. Second, chloride may increase the solubility of a variety of compounds, including metals.
3.3.3.1 Quotient-based risk characterization
The quotient-based environmental risk assessment is based on the procedures outlined in Environment Canada (1997a). Based on possible exposure scenarios, a hyperconservative or conservative Estimated Exposure Value (EEV) is selected. Next, an Estimated-No-Effects Value (ENEV) is determined by dividing an experimental Critical Toxicity Value (CTV) by an application factor to account for the uncertainty surrounding the extrapolation to chronic or pulsed exposure, the extrapolation from laboratory to field conditions, and interspecies and intraspecies variations in sensitivity. The CTV can be based on a wide range of toxicity responses, including mortality, growth, reproduction, fecundity, longevity, productivity, community structure and diversity.
Figure 16 Species diversity across a salinity and chloride gradient (from Wetzel, 1983)
After calculating ENEVs, a hyperconservative (Tier 1) quotient (EEV/ ENEV) is calculated for pertinent sensitive taxa in order to determine whether there is potential environmental risk from the release of that substance. If these quotients are less than 1, it can be concluded that the substance poses little risk to the environment, and the risk assessment is completed. If, however, the quotient is greater than 1 for a particular assessment endpoint, then the risk assessment for that endpoint proceeds to a Tier 2 analysis, where a more realistic but conservative quotient is calculated for each pertinent taxon. If these quotients are greater than 1, the assessment proceeds to a Tier 3 level. Here, the broader likelihood and magnitude of effects are considered. This latter approach involves a more thorough consideration of sources of variability and uncertainty in the risk analysis.
3.3.3.2 Tier 1 and Tier 2 assessments
3.3.3.2.1 Estimated Exposure Values (EEVs)
A Tier 1 assessment is based on a hyperconservative exposure scenario, generally assuming that the EEV is the maximum concentration measured or likely to be encountered in the Canadian environment. The highest reported sodium chloride concentrations in the aquatic environment have been associated with highway runoff, salt-contaminated snow and contaminated waters near salt storage depots.
In Ontario, chloride concentrations as high as 19 135 mg/L were reported in highway runoff from the Skyway Bridge (Mayer et al., 1998). Delisle et al. (1995) assessed the concentration of chloride in snow cleared from city streets in Montréal. The average concentration of chloride was 3851 mg/L, with a maximum reported concentration of 10 000 mg/L. Concentrations of chloride in waste snow in Montréal averaged 3115 mg/L for secondary streets and 5066 mg/L for primary streets (Delisle and Dériger, 2000; Section 2.3.1). Chloride concentrations in leachate from poorly maintained salt storage depots may reach 66 000 mg/L (Section 2.4.2.1). While no Canadian wetland studies have been identified, in Maine, chloride concentrations in surface waters of a bog located near a salt storage depot reached 13 500 mg/L; concentrations remained elevated throughout the March to October sampling period (Ohno, 1990). Therefore, for the purposes of Tier 1 assessments, an EEV of 10 000 mg/L was used to represent estimated maximum chloride concentrations. Such a concentration is representative of snowmelt draining from highways into small receiving water bodies such as roadside creeks and wetlands and approximates that observed in wetlands near poorly designed road salt storage depots.
Tier 2 assessments involve a further analysis of exposure and/or effects to calculate a quotient that is still conservative, but is more realistic than the hyperconservative quotient calculated in Tier 1 (Environment Canada, 1997a). In Canada, the best-documented impacts of road salt on chloride levels in groundwater and other aquatic systems have been for the Metropolitan Toronto area. Williams et al. (1997, 1999) demonstrated the impact of road salts on groundwater and the invertebrate communities inhabiting groundwater-fed springs. There is also a comprehensive water quality data set for the numerous creeks flowing through the Metropolitan Toronto area. The southern Ontario region is also an area of high road salt application rates. Hence, this data set was examined to estimate a Tier 2 EEV.
Chloride concentrations were measured in several streams in the Toronto Remedial Action Plan watershed over 1990-1996 (see Evans and Frick, 2001; Table 9). Data were collected seasonally. Maximum chloride concentrations reported at individual stations included 2140-3780 mg/L for three stations on Etobicoke Creek, 3470 mg/L at Mimico Creek, 4310 mg/L at Black Creek, 96-4310 mg/L at five stations on the Humber River, 960-2610 mg/L at three stations on the Don River and 1390 mg/L at Highland Creek. Seasonal plots of chloride concentration variations in Highland Creek over 1990-1993 show several winter sampling periods in which chloride concentrations exceeded 1000 mg/L for what appear to be several periods of 1 week. Mean chloride concentrations at these sites ranged from 278 to 553 mg/L, with winter concentrations approximately twice those of summer. Williams et al. (1999) reported that an Ontario spring located near a highway and bridge had a mean chloride concentration of 1092 mg/L as a probable result of road salt contamination of groundwater. While the maximum chloride value observed in creeks and rivers in the Toronto area that have been contaminated by road salts is in the range of 2000-4000 mg/L, the frequency and duration of such occurrences are likely variable. Therefore, a less conservative estimate of 1000 mg/L was used as the EEV in the Tier 2 assessments. Chloride concentrations in this range have been commonly observed in Toronto-area creeks and rivers, in a contaminated spring in the Toronto area and in a bog contaminated by a salt storage depot (Wilcox, 1982). It is assumed that, for these scenarios, exposure times would be in the order of days.
3.3.3.2.2 Critical Toxicity Values (CTVs)
Data used for generating the CTVs used in Tier 1 and 2 assessments were obtained during the literature review as described above and presented in detail in Evans and Frick (2001). Suitable data were not found for all trophic groups. The most comprehensive data were from the 2- to 4-day studies. The endpoint most typically investigated was mortality (zooplankton, benthos), unsuccessful development of eggs to larval stages (EC50 data) and reduced growth (phytoplankton, macrophytes).
To calculate ENEVs, an application factor of 100 was used for toxicity data based on LC50 and EC50 data as recommended in the guidance manual (Environment Canada, 1997a) and 10 for toxicity data based on LC25 and EC25 data.
3.3.3.2.3 Conclusions for Tier 1 and Tier 2 assessments
The Tier 1 assessments indicated that risk quotients were greater than 1 for all taxa considered. Accordingly, a Tier 2 assessment was conducted for all taxa. Tier 2 assessments were conducted by reducing the EEV from 10 000 mg chloride/L to 1000 mg chloride/L. Results are presented in Table 14 (given the assumed differences in exposure scenarios between Tier 1 and Tier 2, all Tier 1 quotients are 10 times greater than the Tier 2 quotients). Even if the Tier 2 application factors were reduced by a factor of 10, most quotients would still exceed 1. Tier 3 assessments were therefore done for all groups of organisms considered in Tiers 1 and 2.
3.3.3.3 Tier 3 assessments with field validation
Tier 3 assessments include considerations of the distributions of exposures and/or effects in the environment (Environment Canada, 1997a). This is a challenging task, given the wide variety of road salt entries into the Canadian environment, e.g., leakage from salt storage depots, road and highway runoff (and spray) and transport via creeks into larger aquatic ecosystems, such as wetlands, rivers, ponds and lakes. It is also a complex task, as watercourses vary tremendously in their size, location in relation to a potential source of road salt source, flow regimes, etc.
While data on chloride concentrations have been collected as part of routine water quality monitoring programs, most data have not been subjected to rigorous analysis for spatial and temporal distributions and for causal factors. Thus, while Evans and Frick (2001) conducted an extensive review of the literature, a relatively small number of case history studies were located, especially for the Canadian environment. This situation may begin to change as researchers and agencies devote more effort to better understanding the impacts of road salts on chloride concentrations in the environment.
In order to conduct the Tier 3 assessment, a scenario approach is employed to describe the different settings in which road salts have or may have impacted the Canadian aquatic environment. This approach is based on information on road salt usage patterns across Canada, known information on chloride levels in the aquatic environment, toxicity data as described previously and reported environmental impacts or case studies. Many of the case studies are drawn from literature in the United States, where a greater number of studies have been conducted, but are representative of use in Canada.
3.3.3.3.1 Case 1: Road salt runoff and urban creeks, streams and small rivers in densely populated areas
As cities expand into the rural landscape, they enclose an increasingly large network of creeks, streams and rivers. Moreover, two-lane roads increase in density and, for larger cities, four-lane and larger highways are constructed, resulting in a highly altered watershed. Unlike the rural landscape, where a significant fraction of the runoff may be retained in relatively close proximity to the roadway, in the highly paved urban setting, much of the runoff is delivered in pulses into creeks, streams and rivers. Moreover, the absolute volume of road salt used is higher.
The Metropolitan Toronto area represents the clearest example of the impacts of a dense network of highways on the numerous creeks, streams and, eventually, rivers flowing through this densely populated urban area. As they flow south towards Lake Ontario, these creeks pick up more and more of the urban runoff. As previously noted, chloride concentrations can exceed several hundred milligrams per litre for much of the winter, with peak concentrations reaching 1000-4000 mg/L. Watercourses with high chloride concentrations include Etobicoke, Mimico, Black and Highland. The Don River can also have high chloride concentrations during much of the winter.
Schroeder and Solomon (1998) conducted a formal investigation of chloride concentrations and toxicity at three stations on the Don River. Chloride concentrations from November 1995 to April 1996 ranged up to about 950 mg/L at site 1, up to about 2600 mg/L at site 2 and up to about 1150 mg/L at site 3. These researchers also collected acute lethal toxicity data (LC50 and EC50) for 13 fish and 7 invertebrate taxa; values ranged from about 1000 to 30 000 mg/L. Next, the researchers superimposed the cumulative toxicity data on the cumulative chloride distribution data. They then estimated that 10% of the species would be expected to experience acute effects of chloride toxicity at least 90% of the time at two of the sites and at least 85% of the time at the third site.
Several field studies were located that provided corroborating evidence that chloride concentrations in the 500-1000 mg/L range can impact stream and creek ecosystems. The following paragraphs summarize some of these studies.
Williams et al. (1997) detected significant variations in macroinvertebrate community structure related to different concentrations of chloride in 20 groundwater-fed springs in southeastern Ontario in the Metropolitan Toronto area. The chloride content of these springs ranged from 8.1 to 1149 mg/L. Tipulidae and Ceratopogonidae were associated with springs containing higher chloride concentrations, whereas taxa such as Gammarus pseudolimnaeus (an amphipod) and Tubellaria (a flatworm) were found only in springs with low chloride concentrations. Crowther and Hynes (1977) reported that 20% of a test population of G. pseudolimnaeus dies with a 1-day exposure to 4121 mg sodium chloride/L (2500 mg chloride/L). The higher chloride concentrations in some of the springs originated from groundwater that apparently had been contaminated by road salt.
Crowther and Hynes (1977) experimentally investigated the effect of chloride concentrations on the drift of benthic invertebrates in Lutteral Creek, a trout stream located in southern Ontario. This creek is a small, spring-fed tributary of the Speed River. They also investigated chloride concentrations in Laurel Creek (annual discharge 19 x 106 m3), a tributary to the Grand River, which passes through urban Waterloo. At that time, Waterloo had a population of 32 000. Therefore, the general results of the study by Crowther and Hynes (1977) may apply to the creeks flowing through the numerous small towns in such regions as southern Ontario, Quebec and the Maritimes, where road salt is heavily used. Chloride levels in Laurel Creek varied seasonally, with peak concentrations as high as 1770 mg/L being observed for a 2-week period between January 5 and February 6, 1975.
Considering the elevated concentrations in Laurel Creek, a series of experiments was conducted in Lutteral Creek to assess the impacts of pulses of sodium chloride-laden water on benthic drift. Fifteen metres of Lutteral Creek were divided longitudinally by sheets of corrugated steel and benthic drift assessed on the control side and experiment side of the partition. Highlights are presented in Evans and Frick (2001).
Crowther and Hynes (1977) concluded their study by noting that, as chloride levels in streams approach 1000 mg/L, impacts on benthic drift may begin to appear in streams.
During their study, they noted that most of these effects would be expected to occur during winter and early spring. However, they warned that, as groundwater continued to become contaminated with chloride salts, stream flows might be contaminated by chloride-rich groundwater inflows during summer low-flow periods.
Dickman and Gochnauer (1978) studied the effects of exposure to sodium chloride on the density of bacteria and algae in Heyworth Stream, near Heyworth, Quebec, between July 24 and October 1, 1973. Sodium chloride was added at one location along the stream to maintain chloride concentrations of 1000 mg/L (1653 mg sodium chloride/L). This concentration was selected to simulate the impact of road salt from storm sewers on a receiving stream microflora ecosystem. It would also simulate the impacts of elevated chloride concentrations in groundwater of spring ecosystems. Sodium chloride concentrations at the control site were 2-3 mg/L. The stream was described as shallow, fast flowing and densely shaded. Artificial substrates (tiles) were placed in the creek at the experimental site and at an upstream control and recovered at weekly intervals. The treated site had lower algal diversity and lower standing crops of photosynthetic periphyton or algae, which was related to osmotic stress. Auxospore (a resting stage) formation of the diatom Cocconeis placentula was noted at the treated site on day 28 but not at the control site; auxospore formation is often triggered by environmental stress. Density of bacteria was greater at the treated site, which was attributed to a reduction in the grazing pressure on the bacterial population because of the reduced number of grazers, such as flagellates, ciliates and rhizopods. The incidence of diatom parasitism was lower at the treated site than at the control site, possibly because the sodium chloride inhibited fungal growth, as indicated by other researchers (Kszos et al., 1990; Rantamaki et al., 1992).
Other studies investigated the effects of salinity of benthic and algal community structure. Although the salinity was not due to road salt, these studies provide further support for the hypothesis that stream community composition is sensitive to salinity, with species composition changing as salinity increases towards 1000 mg chloride/L.
Leland and Fend (1998) investigated benthic distributions in the San Joaquin River Valley, California. Total dissolved solids concentrations ranged from 55 to 1700 mg/L. Distinct assemblages were observed at the high and low salinities, and the distribution of many taxa indicated salinity optima. Ephemeroptera (mayfly larvae) rarely occurred at salinities above 1000 mg/L (600 mg chloride/L). While other, covarying factors may have affected these relationships, the authors noted that patterns were not related to pesticide distributions or to water discharge and irrigation regime.
Short et al. (1991) investigated benthic invertebrates in a Kentucky stream subject to chloride seepage from nearby oil field operations. Ephemeropterans were the group least tolerant of elevated sodium chloride levels and were absent in regions where salinity exceeded 2000 mg chloride/L. Fish appeared to be more tolerant of these elevated salinities.
Blinn et al. (1981) investigated the seasonal dynamics of phytoplankton at three locations along the Chevelon Creek system in Arizona. The chloride concentration at site 1 was 88 mg/L at baseflow compared with 900-1100 mg/L at site 3, located 10.8 km downstream of site 1 and 3.5 km downstream of site 2 (chloride concentration at site 2 was not given). The elevated chloride concentration at site 3 was due to numerous seeps and springs from the canyon wall. There were marked differences in phytoplankton between the three sites, with species associated with differences in salinities.
Overall, these studies suggest that continuous exposure to chloride concentrations as low as 1000 mg/L (1653 mg sodium chloride/L) for time periods as short as 1 week can result in changes in stream periphyton and benthic communities. Furthermore, if these exposures persist, community composition remains different from lower-salinity upstream and downstream sites. Several severely chloride-contaminated streams and ponds adjacent to roadways have been identified in the Metropolitan Toronto area and surrounding municipalities. The number of affected streams and creeks is likely to increase with increased population growth in southern Ontario. Chloride-contaminated streams may also occur in other similar areas of Quebec and elsewhere where there is a dense network of highways and heavy road salt use. Areas in which continuous (or long-term) exposures are most likely to occur are seepages from road salt storage yards and contaminated groundwaters that later emerge as springs. Data suggest that pulses of sodium chloride-laden water during spring melt will have pronounced effects on stream communities, as will the continued release of high concentrations of sodium chloride from stream banks during summer and autumn months. Periphyton form the base of the food web in many creeks, being grazed upon by invertebrates, which in turn serve as forage for fish. Reduced fungal biomass resulting from increased sodium chloride concentrations may also impact invertebrates, forage fish and, ultimately, predatory fish communities.
3.3.3.3.2 Case 2: Road salt runoff and creeks, streams and rivers in less densely populated areas
Substantially lower chloride concentrations have been observed in creeks flowing through less densely populated areas with a smaller highway network. Moreover, at these low levels, which may be only 10-100 mg/L above background concentrations, many anthropogenic activities may be associated with this increase. For example, within the Waterford River Basin (Newfoundland), chloride concentrations averaged 70 mg/L at the Donovans Station (located in an industrial park), compared with 17 mg/L in an area that had no known industrial activities (Arsenault et al., 1985). High concentrations at the Donovans Station were related to industrial activities in addition to road salting and contamination from two road salt depots. Unfortunately, no study was specifically designed to investigate chloride concentrations immediately following road salt application and the following snowmelt.
Several studies conducted in the United States have shown similarly low chloride concentrations. One study shows how, in these low-impact streams, relatively high chloride pulses may occur. Boucher (1982) conducted a study investigating the impacts of road salt runoff on Penjajawoc Stream in Maine.
Chloride concentrations upstream of a commercial development averaged 5-15 mg/L over 1979-1982; within the development, concentrations averaged 10-50 mg/L. However, chloride levels rose during runoff events to reach a maximum concentration of 621 mg/L. Other studies have also associated the use of road salts with increased chloride concentrations. These concentrations have ranged between 10 and 150 mg/L in small urban and rural streams (Cherkauer and Ostenso, 1976; Gosz, 1977; Hoffman et al., 1981; Smith and Kaster, 1983; Prowse, 1987; August and Graupensperger, 1989; Demers and Sage, 1990; Mattson and Godfrey, 1994; Herlighy et al., 1998). Other factors may have affected some of the increases in average chloride concentrations in these streams.
However, where studies have been conducted to assess the effects of snowmelt on chloride levels, a clear pulse has been observed (Prowse, 1987). Prowse (1987) and Mattson and Godfrey (1994) also noted a relationship between the concentration of chloride in streams and highway density and length of the roadway that drains directly into the watercourse (e.g., where the road is adjacent to the watercourse).
The impacts on stream and creek communities of relatively small increases in chloride concentrations or larger increases over a short duration (hours to days) are uncertain.
Demers (1992) investigated the effects of elevated chloride levels on the aquatic macroinvertebrates inhabiting four streams near the town of Newcomb in the Adirondack region of northern New York. All four streams were located along a 2-km stretch of a state highway. Chloride concentrations in the creeks were measured weekly, yielding an overall mean concentration of chloride in upstream locations of 0.61 mg/L compared with 5.23 mg/L in downstream areas. Artificial substrates were placed in riffle or fast-flowing sections of the stream upstream and downstream of the highway. Samplers were within 50-100 m of the road and were left in place between April 22 and June 3, 1988, at which time they were recovered to characterize the colonizing invertebrates.
Benthic diversity was lower in downstream than in upstream sites. There was an increase of Chironomidae in downstream sites, while Perlodidae (a stonefly family) and Ephemerellidae (a mayfly family) declined downstream of the highway. A similar increase in chironomid and decrease in mayfly and stonefly dominance were observed in the Humber River (Ontario) between November, when road salt had not yet been applied, and the following March, after road salt had been applied (Kersey and Mackay, 1981).
Other studies have reported differences in benthic communities upstream and downstream of highways, but these differences were related to the larger differences in flow regime, sediment suspended load, habitat and inorganic and organic contaminants (Molles and Gosz, 1980; Kersey and Mackay, 1981; Smith and Kaster, 1983; Maltby et al., 1995). In the absence of carefully designed field studies, the impacts of small increases in chloride concentrations on stream, creek and river ecosystems remain unknown. An increase in mean chloride concentration from about 3 mg/L (world average from rivers) to 30-300 mg/L may affect succession of species with different salinity optima. Recent paleolimnological studies are showing that algal species, in fact, have distinct salinity optima, and changes in lake salinity are being inferred based on long-term changes in algal communities (Dixit et al., 1999).
Some researchers are beginning to investigate the role that water chemistry plays in determining community structure in rivers.
For example, Rott et al. (1998) investigated periphyton assemblages in the Grand River, Ontario, where chloride concentrations ranged from 7.7 to 85.0 mg/L and conductivity from 180 to 540 µS/cm. Canonical correspondence analysis showed that the largest portion of variability in species composition in the river over the study period could be explained by a seasonal gradient related to temperature and latitudinal gradients of nitrite-nitrate, conductivity and chloride. Phosphorus, ammonia, pH, turbidity and oxygen were of lesser importance.
Chetelat et al. (1999) investigated periphyton biomass and community composition in 13 rivers in southern Ontario that differed in their nutrient concentrations and conductivity. Biomass was strongly correlated with total phosphorus (r 2 = 0.56) and even more strongly correlated with conductivity (r 2 = 0.71). They suggested that the predominance of Cladophora at high-conductivity sites was related to the higher calcium concentrations at this site. Conductivity ranged from 65 to 190 µS/cm. Major differences in species composition were observed between low- and high-conductivity sites.
It is not clear whether periphyton communities were responding directly to chloride or whether chloride is an indicator of other perturbations in the watershed. However, given newer studies involving algal community composition in lakes (e.g., Dixit et al., 1999), it is probable that at least some of the responses are directly due to small differences in salt concentrations.
3.3.3.3.3 Case 3: Road salt runoff and urban ponds and lakes
Ponds and lakes differ from creeks, streams and rivers in three important ways. First, ponds and lake systems do not have the pronounced and seasonally varying flow velocities characteristic of streams, creeks and rivers. Second, water residence time in ponds and lakes is longer, ranging from weeks to years. Third, ponds and lakes receive water from a variety of sources, and so their dilution capacity for contaminants can be higher than that for streams and creeks located near similar contaminant sources. Thus, chloride levels tend to be substantially lower in road salt-contaminated ponds, although the exposure time is generally much longer. The degree of contamination also depends on the degree of dilution that can be afforded. Generally, larger water bodies have a greater capacity for dilution.
There is a tremendous amount of variation in the degree to which road salt runoff affects the chloride concentration of urban ponds and lakes. However, in general, impacts on the Canadian environment have been modest, with chloride increases limited to the tens to hundreds of milligrams per litre rather than the thousands of milligrams per litre as observed for creeks in the Metropolitan Toronto area.
In Newfoundland, Arsenault et al. (1985) investigated chloride concentrations in five ponds in the Waterford River basin. Average concentrations ranged from 6 to 24 mg chloride/L, with the higher values associated with highway runoff and possibly runoff from a salt storage depot.
In Halifax, the First and Second Chain lakes have shown seasonal increases in chloride concentration, apparently related to road salting; chloride concentrations increased through the early 1980s to peak at 120-170 mg/L in the mid 1980s. Both lakes are shallow (mean depths of 4.1 and 3.2 m, respectively) and with a large urban watershed. More recently, Keizer et al. (1993) investigated the water quality of 51 lakes in the Halifax/Dartmouth Metro Area, comparing data collected in April 1991 with those collected in April 1980. The most pronounced change in these lakes was a near doubling in conductivity, with the majority of the increase due to increased concentrations of sodium and chloride. Whereas only 3 lakes had chloride concentrations greater than 100 mg/L in 1980, 12 had chloride concentrations exceeding 100 mg/L in 1991, with one lake reaching 197 mg/L.
In southeastern Ontario, Little Round Lake apparently became meromictic (or more strongly meromictic) in the 1950s as a result of road salt entering the lake from two highways and seepage from a salt storage shed (Smol et al., 1983). Chloride concentrations in the monimolimnion were 104 mg/L in about the early 1970s. This meromixis resulted in a reduction in seasonal nutrient exchange from deep to surface waters, a reduction in lake productivity and a change in algal community structure.
In Ontario, an ongoing study is providing evidence of elevated chloride concentrations in ponds next to highways. Chloride concentrations in ponds within 10 m of the highway average 88 mg/L (5-150 mg/L) for two-lane highways and 970 mg/L (84-2630 mg/L) for four-lane highways (Watson, 2000). Chloride concentrations remained elevated 50 m from the highway, with concentrations averaging 115 mg/L (35-368 mg/L) for two-lane and 1219 mg/L (50-3950 mg/L) for four-lane highways. Free and Mulamoottil (1983) reported chloride concentrations of up to 282 mg/L in Lake Wabekayne, a stormwater impoundment in Mississauga.
The Ontario Ministry of Transportation has unpublished data on the mean (May-October 1995) chloride concentrations in eight water bodies in the Humber River watershed (Scanton, 1999). Chloride concentrations averaged 10.6 mg/L at Lake St. George, 26.8 mg/L at Preston Lake, 47.0 mg/L at Wilcox Lake, 60.8 mg/L at Heart Lake, 107.9 mg/L at Claireville Reservoir, 110.3 mg/L at G. Ross Lord Dam, 174.0 mg/L at Lake Aquitanne and 408.9 mg/L at Grenedier Pond. The Humber River, like Highland Creek, is a major urban catchment basin. Road salt applications may be impacting the Humber River basin, as has been observed for the Highland Creek basin (Howard and Haynes, 1997).
More extreme examples of increased chloride concentrations in ponds and creeks in urban areas are to be found in U.S. studies. Cherkauer and Ostenso (1976) reported that Northridge Lakes in Milwaukee, Wisconsin, were affected by elevated chloride concentrations during winter ice cover. These lakes are shallow (9.2 m), artificial and interconnected. Chloride concentrations reached 2500 mg/L. Another extreme example is Irondequoit Bay, a small embayment located on the southern shore of Lake Ontario. This is a moderately deep bay (maximum depth 23.8 m, mean depth 6.8 m) surrounded by a dense network of highways. Chloride concentrations began increasing in this bay in the 1940s and continued increasing into the early 1970s. Chloride concentrations in deep waters reached more than 400 mg/L; vertical mixing of the water column was prevented (Bubeck et al., 1971). Remedial actions were developed; by the early 1980s, chloride concentrations in Ides Cove, within the bay, declined to a maximum of 140-150 mg/L, and vertical mixing was improved (Bubeck et al., 1995).
In Ann Arbor, Michigan, Fonda Lake was adversely affected by highway runoff that entered the lake via two drainage pipes. Chloride concentrations reached 177 mg/L in the mid 1960s, and vertical mixing was disrupted (Judd, 1969). After this was learned, the city reduced its salt usage in the subdivision and redirected runoff. However, chloride concentrations continued to increase because of increased inputs from the greater number of highways. Chloride concentrations reached a maximum of 720 mg/L in deep waters (Judd and Stegall, 1982). However, because the salt entered the lake in diffuse sources rather than through two drainage pipes, this salt mixed into the lake more effectively, and vertical mixing was not disrupted.
As already noted, few biological studies have accompanied these urban pond and lake studies. In addition to the Smol et al. (1983) study, Free and Mulamoottil (1983) noted a decrease in benthic invertebrate densities that occurred when bottom water became anoxic. Based on estimates of chronic toxicity, direct chloride effects on the individual species may be expected to occur at concentrations as low as 200 mg/L for individual species. However, recent studies involving phytoplankton suggest that for algae, at least, impacts may occur at even lower levels. This is discussed in the next section.
3.3.3.3.4 Case 4: Highway runoff and rural ponds and lakes
Chloride increases due to road salt have also been reported in rural ponds and lakes. As in the urban setting, the increase is related to pond/lake size and the degree of contact with the roadway.
In Terra Nova National Park in Newfoundland, chloride concentrations in Pine Hill Pond increased nearly 7-fold in spring from an average of 14 mg/L over April-December 1969 to 94 mg/L in April (Kerekes, 1974). This increase was related to road salt usage.
Underwood et al. (1986) surveyed 234 Nova Scotia lakes. The background chloride concentration was 8.1 mg/L, while the maximum concentration was 28.1 mg/L. The Nova Scotia Department of the Environment (Briggins et al., 1989) indicated that waters with chloride concentrations above 25 mg/L must have been receiving chloride inputs from anthropogenic sources. In watersheds with numerous highways, road salt would have been the probable source.
The Ministère des Transports du Québec (1980, 1999) investigated the impacts of road salt on Lac à la Truite, by Highway 15 and near Sainte-Agathe-des-Monts. The lake's drainage area, estimated at 728 ha, was affected by a 7-km stretch of highway, with a maximum slope of 5%. The lake had a surface area of 48.6 ha, a mean depth of 21.5 m and an estimated volume of 486 000 m3. In 1972, the average chloride concentration for the lake was 12 mg/L. This increased through the 1970s to reach a maximum concentration of 150 mg/L in 1979. This corresponds to an addition of approximately 67 068 kg of salt to the lake. The amount of salt applied to the roads was reduced, and chloride concentrations fell through the 1980s to reach 45 mg/L in 1990. While concentrations declined, they remained elevated at 42 mg/L at 3 m and 49 mg/L at 10 m.
In the United States, as in Canada, there is growing evidence of increasing chloride levels in lakes, with these increases being related, at least in part, to road salt application.
Several studies conducted in the United States provide further information on the probable impact of road salts on rural lakes and ponds that may be applicable to the Canadian environment. In Maine, Hanes et al. (1970) discussed the effects of road salt on a variety of aquatic ecosystems. They noted that farm ponds located near highways had sodium concentrations ranging from 1.4 to 115 mg/L and chloride concentrations ranging from <1 to 210 mg/L. Salt concentrations were higher in April 1966 than in July 1965. In 1967, chloride concentrations ranged from 1.4 to 221 mg/L, suggesting that chloride levels were increasing.
Sparkling Lake in the northern Wisconsin Lake District experienced an increase in chloride concentration due to contamination from road salt-laden groundwater (Bowser, 1992). The morphometry of Sparkling Lake was not reported. Road salt was initially applied to roads above the lake, and the salt apparently leached into the groundwater prior to entering the lake. Chloride concentrations in unaffected lakes and groundwater in the area were 0.3-0.5 mg/L compared with 2.61 mg/L in 1982 and 3.68 mg/L in 1991 in Sparkling Lake. The load of chloride required to produce such an increase in chloride concentration between 1982 and 1991 was calculated by Bowser (1992) to be 1200 kg/year.
Eilers and Selle (1991) compared conductivity, alkalinity, calcium and pH data collected at 149 northern Wisconsin lakes over 1925-1931 with data collected over 1973-1983. All parameters increased between the two time periods, with the greatest increases associated with increased land development on lake shorelines. Mean conductivity increased from 31.5 µS/cm in 1925 to 44.3 µS/cm in 1983 in developed watersheds and from 14.3 µS/cm in 1925 to 14.7 µS/cm in 1983 in undeveloped watersheds. Increased conductivity appeared to be associated with a combination of factors, including road salt, cultural eutrophication and changes in hydrology. The strongest increases were associated with lakes located near highways or paved roads.
The impacts of small increases in chloride concentration from road salt on lakes and ponds were not investigated in these studies. However, increased chloride (and other salts) concentrations traditionally have been viewed with concern (Beeton, 1969; Pringle et al., 1981) and actions taken to protect aquatic environments from these increased salt levels.
Small increases in chloride concentration are most likely to impact species composition and productivity. There is emerging evidence that microorganisms such as fungi, plankton and macroinvertebrates have salinity optima. Changes in salinity that alter the competitive balance between species should result in compositional changes at the base of the food web. Productivity may also increase. Evans and Frick (2001) addressed this issue by investigating what is known about the effects of salinity and conductivity on freshwater systems in general.
The addition of road salts to aquatic environments may enhance productivity through a variety of means. Road salt contains trace nutrients, including phosphorus (14-26 mg/kg) and nitrogen (7-4200 mg/kg). The addition of 100 mg chloride salt/L could result in an increase in phosphorus concentration of 1.4-2.6 µg/L and an increase in nitrogen concentration of 0.7-420 µg/L. Metals such as copper and zinc are also essential elements, and the addition of these metals to aquatic ecosystems may also enhance productivity. The systems that would be most vulnerable to a road salt-stimulated eutrophication would be low-productivity ecosystems. Paleolimnological studies in the Great Lakes have demonstrated enhanced productivity in the Bay of Quinte as early as the late 1600s, which was associated with land clearing and the increased inputs of nutrients (and trace metals) from the watershed (Schelske et al., 1983, 1985; Stoermer et al., 1985).
More recent limnological studies are investigating community structure in a wide variety of lakes and as a function of a number of limnological variables, such as conductivity, nutrient concentrations and major ions. This literature was briefly examined to determine whether additional information could be found on the responses of aquatic organisms to small gradients in salinity in addition to the expected responses to phosphorus concentrations. Since these studies were conducted in freshwater lakes, conductivity was dominated primarily by calcium carbonates. The following are noted.
Dixit et al. (1999) assessed changes in water quality in the northeastern United States using diatoms that had become deposited in lake sediments as indicators of change. The composition of diatoms in the surface sediments was related to current water quality parameters, including pH, chloride concentration and total phosphorus concentration. Total phosphorus and chloride optima were developed for 235 of the common species (Table 15). Mathematical models were then developed to predict the pH, chloride concentration or total phosphorus concentration of the study lakes based on the diatom assemblage (i.e., not individual taxa) found on surface sediments. The diatom assemblage was then examined at the 30-cm sediment depth in core samples to determine whether or not the water quality had changed since before the 1850s. Marked deterioration in water quality was noted in hundreds of lakes. In these lakes, phosphorus and chloride concentrations clearly were higher in modern than in preindustrial times. Moreover, there has been a marked increase in the number of eutrophic lakes in the Coastal Lowlands/Plateau. This increase in the trophic status of these lakes was associated not only with increasing phosphorus concentrations but also with increasing chloride ion concentrations. Sodium and chloride concentrations were highly correlated in these lakes, suggesting that increased chloride concentrations were associated with road salting. However, it was also recognized that agriculture, silviculture and urbanization can contribute to increases in chloride concentration. Future research, based on more detailed paleolimnological studies, will resolve these issues. However, it is already apparent that diatoms have optimal chloride concentrations, and deviations in these concentrations are associated with changes in species composition. This also must occur for other organisms, such as other algal groups, plants, benthic invertebrates and zooplankton.
Studies, while few in number, strongly suggest that small increases in chloride concentration will result in shifts in phytoplankton community composition in ponds and lakes. Furthermore, it is highly likely that standing stocks of plants and animals will also be affected. Productivity is likely to be enhanced through increased phosphorus, nitrogen and trace element inputs.
3.3.3.3.5 Case 5: Salt storage depots and aquatic ecosystems
The actual number of road salt storage depots in Canada is unknown. Morin and Perchanok (2000) estimated that there were more than 1300 patrol yards at the provincial level and an unknown number of municipal and privately maintained yards. Snodgrass and Morin (2000) also noted that the design of these yards varies considerably across Canada. Some are well designed, with minimal salt leaching through the winter. Poorly designed yards may lose 22% or more of their salt through leaching. Salt concentrations in such leachate can exceed 80 000 mg/L (see Section 2.4). Ontario Ministry of Transportation (MTO, 1997) studies conducted in Ontario determined that 26% of patrol yards had well water concentrations that exceeded 500 mg/L. In another study, 69% of shallow groundwater samples had chloride concentrations that exceeded 250 mg/L. While leachate eventually will be diluted on entering a receiving water body, considerable damage may be inflicted on the aquatic environment on the dilution path. This is of particular concern for wetlands and small streams such as investigated by Dickman and Gochnauer (1978). Moreover, even after dilution, chloride concentrations may remain high enough to create meromixis in small lakes and/or to significantly raise their salinity.
Poorly designed or maintained salt storage depots clearly present a threat to the Canadian environment. Brief (hours to a few days) exposures to undiluted leachate are well within the toxic range for a variety of aquatic organisms. Where saline water collects for extended periods of time, as in a bog, sensitive marsh vegetation species may be killed and replaced by more salt-tolerant forms (Wilcox, 1982). Where saline water flows into lakes, the normal circulation patterns of the water column may be disrupted, and the lake becomes more strongly stratified. Relatively small increases in salinity to as little as 200-300 mg chloride/L fall within the chronic toxicity range for certain aquatic organisms.
Some studies have implicated road salt storage depots as having adversely affected chloride levels in the Canadian environment. Arsenault et al. (1985) related small increases in chloride concentrations at the Donovans Station on the Waterford River, Newfoundland, to salt storage depots in addition to potential impacts from the highway. In addition, Arp (2001) (Section 3.4.1.2) reported elevated sodium and chloride concentrations in creek and ditch water located downstream of a salt storage depot. Highest chloride levels occurred in summer. In southeastern Ontario, Smol et al. (1983) conducted a paleolimnological study on Little Round Lake, noting that the lake was oligotrophic prior to European settlement in the mid 1800s. With increased logging in the watershed, the lake became eutrophic, presumably because of increased nutrient and mineral inputs associated with land disturbance. A railroad and two highways were later built adjacent to the lake. The lake returned to oligotrophic conditions in the late 1960s as an apparent result of the lake becoming meromictic. The deep-water chloride concentration in the early 1970s was 104 mg/L, well above background levels. This meromixis prevented nutrients regenerated in the deep layer of the lake from being mixed back into surface waters. The elevated chloride concentrations (and meromixis) were related to highway runoff and seepage from a salt storage shed. Unfortunately, very few studies have been conducted around salt storage depots in Canada, despite the fact that there are more than a thousand such facilities in Canada and the fact that many clearly are contaminating groundwater and well water with chloride.
More limnological studies have been conducted around road salt storage depots in the United States. Such studies have shown a variety of impacts, which indicates that the proper management of such facilities is of environmental concern. Thus, the results of these studies have broad application to the Canadian environment, particularly for storage depots that are not properly maintained (i.e., covered and on an asphalt pad) and where best practices are not followed with respect to patrol yard washwater waste.
Wilcox (1982) investigated the effects of sodium chloride contamination by a road salt storage pile at Pinhook Bog in Indiana. In 1963, an uncovered sodium chloride storage depot was established overlooking the bog. Salt-laden runoff from the storage pile resulted in major alterations in the bog vegetation within a 2-ha area, as did runoff from the highway. The salt pile was covered in 1972; after winter 1980-81, road salt ceased to be stored at this site. Impacts were studied from 1979 to 1983. Sodium concentrations as high as 468 mg/L and chloride concentrations as high as 1215 mg/L were recorded in interstitial waters of the bog mat in areas of the strongest road salt impact (Wilcox, 1982). These readings were made in 1979, so higher concentrations may have occurred in earlier years. Chloride concentrations at control sites in 1980 and 1981 were 5-6 mg/L. The maximum single daily readings for salt-impacted locations were 1468 mg chloride/L in 1979, 982 mg chloride/L in 1980, 570 mg chloride/L in 1981 and 610 mg chloride/L in 1983. This indicates that decommissioning of the storage facility resulted in reductions in loadings and resulting concentrations in the bog, but that the chloride was subsequently largely retained in the wetland.
Native species, such as Sphagnum spp. and Larix laricina, were absent from the impacted areas of the bog, while salt-tolerant species, such as Typha angustifolia, invaded the bog. Salinity tolerances of various plant species were defined (Wilcox, 1982), and taxa were shown to be sensitive to sodium chloride concentrations in bog water as low as 280 mg/L (170 mg chloride/L) (Table 16). As salt concentrations decreased some 50% over 1980-1983, many endemic bog plants, including Sphagnum, recolonized the bog (Wilcox, 1982). Sphagnum began growing on low hummocks in areas where interstitial chloride concentrations had dropped to approximately 300 mg/L.
Tuchman et al. (1984) used core samples from lake sediments to investigate the impacts of historic salinization on diatoms of Fonda Lake, Michigan. A salt storage facility has been located adjacent to Fonda Lake since 1953; an asphalt pad was added at the salt storage facility in the early 1970s to reduce loss of salt. A reduction in diversity of algal species began in 1960. Diatom diversity reached a minimum in 1968, when a variety of salt-tolerant (or halophilic) taxa attained their highest relative abundance. In later years, diversity increased slightly and some halophilic taxa decreased in relative abundance, suggesting a decrease in salt loading to the lake. Unfortunately, lake salinity was not determined as part of this study. Zeeb and Smol (1991) extended the study of Fonda Lake to scaled chrysophytes. Mallamonas caudata, a chloride-indifferent taxon, dominated at all depths in the core. Mallamonas elongata, a widely distributed taxon found more commonly in eutrophic and alkaline waters, and Mallamonas pseudocoronata, a chloride-intolerant taxon, declined during the period when chloride concentrations apparently were highest.
Mallamonas tonsurrata, which occurs mainly in eutrophic lakes, became more abundant during this period. Major shifts in diatom and chrysophyte assemblages were associated with relatively small changes in salinity (i.e., from about 12 to 235 mg chloride/L, or about 20-387 mg sodium chloride/L).
Overall, these studies show that poorly designed or managed road salt storage depots can have a number of adverse impacts on the aquatic environment. This is of particular concern because biological monitoring programs have not been established around these yards to ensure that the local aquatic and terrestrial environments are not being damaged.
3.3.3.3.6 Case 6: Snow dumps
Snow collected from roadways to which road salts have been applied has been reported to contain chloride at concentrations reaching up to 2000-10 000 mg/L. Chloride concentrations in this range can be acutely toxic to a variety of organisms for exposures of 1 day or less. The majority of municipalities that remove roadside snow use surface dumps (Delisle and Dériger, 2000). The actual location of snow dumps is subject to significant variation and variable environmental regulation. The number of snow dumps is also unknown.
Snow deposited on low-lying lands can result in the increased chloride content of underlying soils and wetlands when the snow melts. Similarly, snow deposited on elevated areas eventually melts, transporting salt into the groundwater and into aquatic ecosystems, such as streams, wetlands and ponds.
Few studies have been conducted around snow dumps. However, elevated chloride concentrations (233-994 mg/L) have been reported in monitoring wells established to assess the impact of snow disposal on shallow groundwater (Morin, 2000). This contaminated groundwater could, in turn, contaminate groundwater-fed springs, as observed by Williams et al. (1999).
In British Columbia, Warrington and Phelan (1998) reported that the ionic composition of two lakes was shifted from carbonate dominated to chloride dominated as a result of salt-laden snow being pushed into the lakes from the highway. Salt-laden highway runoff may also have been a contributing factor. The biological effects were not investigated, but some could be inferred from recent paleolimnological studies of the effects of increased chloride concentrations on phytoplankton communities.
No other studies were located on the adverse impacts of surface snow dumps on aquatic ecosystems. However, given the probable thousands of snow dumps across Canada and the high chloride concentration in meltwaters, the potential harm to the Canadian aquatic environment of poorly located dumps is of concern. This is particularly so because there are not uniform regulations regarding snow dump practices and very few monitoring programs in place to ensure that the environment is not being harmed by these chloride releases. Unsuitable locations for snow dumps include wetlands, small ponds and lakes, and elevated areas that drain into small creeks. Groundwater contamination also needs to be considered.
3.3.3.4 Other effects
Road salts can have adverse effects on the aquatic environment in addition to their apparent direct effects on species composition and abundance. Chloride salts can affect aquatic systems through interactions with abiotic components of the environment. Notably, chloride salts tend to be more soluble than carbonate salts and can thus, through various reactions, enhance the mobility of trace metals in aquatic ecosystems. Increased salt loadings can affect water density, thereby affecting mixing processes in lakes and thus disrupting the ecological functioning of that ecosystem.
Increased concentrations of chloride salts in surface water systems can lead to the release of metals from sediments and suspended particulate matter. By competing for particulate binding sites, sodium chloride acts as an enhancer of dissolved and potentially bioavailable trace metals, such as cadmium, copper and zinc (Warren and Zimmerman, 1994). Cadmium, copper and zinc are acutely toxic to aquatic organisms at concentrations as low as 1 µg cadmium/L (rainbow trout), 6.5 µg copper/L (Daphnia magna) and 90 µg zinc/L (rainbow trout) (CCME, 1991). In addition, since road salt contains a variety of contaminants, including metals such as copper (0-14 µg/g), zinc (0.02-0.68 µg/g) and cadmium (MDOT, 1993; Maltby et al., 1995), snowmelt containing road salt is a potential source of metal contamination to receiving water bodies. Furthermore, because the highway surface contains a variety of metal contaminants from the automobile itself (lead, copper, zinc and cadmium), road salt can facilitate the mobilization and transport of these contaminants into the aquatic ecosystem (Maltby et al., 1995). These metals may accumulate in various reaches of streams on sedimentary particles (Maltby et al., 1995) or in stationary waters such as ponds. There, they have the potential to exert toxic effects, especially through processes involving sediments.
High concentrations of salts in sediment pore water were shown to augment the concentrations of dissolved heavy metals (e.g., cadmium) and resulting toxicity to a benthic invertebrate, Hyalella azteca (Mayer et al., 1999). Wang et al. (1991) found that 709 mg chloride/L (or 0.02 mol/L) substantially enhanced the release of mercury from freshwater sediments. Mercury can be acutely toxic to invertebrate species and fish in concentrations as low as 0.002 mg/L and 0.02 mg/L, respectively (CCME, 1991) and, when present in its methylated form, accumulates in the tissues of fish. Sodium chloride also enhances mercury mobilization from soils (MacLeod et al., 1996).
Road salt can also affect ponds and lakes by reducing their capacity for vertical mixing. Most lakes undergo vertical mixing, resulting in an exchange of deep and surface waters. Such exchanges are important for transferring oxygen-rich surface water to the deeper regions of the lake. In the absence of such exchanges, deep waters can become anoxic. Vertical exchanges are also important in transferring nutrients regenerated in the deeper regions of the lake to the surface, where they become available to the phytoplankton community. Vertical mixing is driven by the winds, which mix surface waters down to depths dependent on lake fetch, surrounding topography and water temperature. Vertical mixing is also driven by changes in water temperature that affect the density of water. Since water has a maximum density at 4°C, spring warming from freezing to 4°C is accompanied by an increase in water density, promoting the vertical exchange of warmer waters with colder and less dense deep waters. Lake cooling in autumn and lake warming to 4°C in spring are accompanied by a change in water density, which promotes the vertical exchange of waters.
Meromixis occurs in lakes when conditions develop that prevent the vertical exchange of surface and deep waters (Wetzel, 1983). This occurs when a sufficient gradient exists in salinity to override the effects of seasonal variations in water temperature on density and lake mixing. Road salts, by entering lakes through surface flow (overland runoff, ditches, streams) or groundwater discharge (seeps, springs), have the potential to impair the normal circulation of lakes. Small, moderately deep lakes will be the most vulnerable to such meromixis, especially in areas of heavy road salt application. Larger lakes are less vulnerable, because the intruding saltwater plumes experience greater dilution as the denser salt-laden water flows along the lake floor towards the deeper regions of the lake. In addition, larger lakes have greater fetches and hence more powerful wind-driven currents and other water exchanges.
The formation of meromixis can have a number of impacts on lakes. The low oxygen conditions that develop below the chemocline can result in the loss of all but the most resilient of deep-water benthic species. Zooplankton may be excluded from their deep-water daytime refuges, being forced to live in the brighter surface layers, where they may be more vulnerable to size-selective fish predation. Hypolimnetic fish such as lake trout may also be adversely affected. The onset of lake meromixis will affect sediment-water exchanges. Phosphorus and various metals are more readily released from poorly oxygenated than from well-oxygenated sediments (Wetzel, 1983). This increase in phosphorus release from the sediments may enhance the productivity and eutrophication of the lake, particularly if there is sufficient exchange at the chemocline. Alternatively, if there is limited exchange at the chemocline, production may be reduced, and a eutrophic lake may become oligotrophic (Smol et al., 1983).
3.3.3.5 Conclusions for aquatic ecosystems
Short-term acute toxic effects have been associated in the laboratory with high chloride concentrations. However, as exposure time increases, sensitivity to chloride increases.
For example, the 4-day LC50 is 1400 mg/L for Ceriodaphnia dubia (a cladoceran) and 1261 mg/L at 9 days (Cowgill and Milazzo, 1990). Exposure to such high lethal concentrations would most likely be associated with poorly managed salt storage depots, inappropriately placed snow dumps on wetland areas, roadside ditches in areas of high use, contaminated groundwater springs and small watercourses in heavily populated urban areas (e.g., Metropolitan Toronto) that have a dense network of highways and road salt application.
Longer-term toxicity occurs at substantially lower levels, estimated to begin at concentrations as low as 210 mg/L. Chloride concentrations at these levels have been observed in a variety of urban creeks, streams and lakes, although primarily in highly populated areas and small water bodies near highways with high use. Such levels are also expected to occur in the vicinity of poorly managed road salt storage depots and in inappropriately placed snow dumps, e.g., on wetlands, in small ponds, near headwater creeks and near contaminated groundwater-fed springs. Aquatic ecosystems experiencing such chloride levels are expected to be impaired.
A family of curves was prepared based on chloride toxicity data following 4-day, 7-day and predicted long-term (chronic) exposures (Figure 17). The toxicity data used to develop Figure 17 can be found in Evans and Frick (2001). One-day and less than 1-day data are excluded because studies were few in number. Data are plotted on a linear-log scale with 95% confidence intervals around each curve. A log-logistic model was used in the curve fitting (Environment Canada, 1997b). Based on the predicted data, it can be expected that 5% of species would be affected (median lethal concentration) at chloride concentrations of about 210 mg/L (Table 17), while 10% of species would be affected at chloride concentrations of about 240 mg/L.
Figures 18 and 19 show various chloride concentrations in the Canadian aquatic environment juxtaposed with concentrations causing adverse biological effects. The environmental chloride concentrations specified are assumed to arise from road salt contamination. Unless otherwise specified, all values for effects relate to lethality.
Figure 18 presents data on short-term exposure pertinent to a period of 5 days or less. The concentrations given are for rivers and creeks where chloride peaks associated with events such as freeze/thaw occurrences would not be expected to last for more than a few days at any given sampling location. The biological effects specified are for toxicity tests with an endpoint of 5 days or less (e.g., 4-day LC50).
Figure 19 presents data pertinent to exposure for a period of more than 5 days. The environmental concentrations given are mostly for lakes and ponds where chloride levels are generally expected to remain stable over a longer period. The biological effects specified are for toxicity tests with an endpoint of more than 5 days (e.g., 33-day LC50).
As previously noted, there is growing evidence of a widespread increase in chloride levels in rivers and lakes in the northeastern and midwestern United States, including lakes in a non-urban setting. For example, Dixit et al. (1999) investigated 257 lakes and reservoirs in the northeastern United States and noted that while 34% of the reservoirs and 6% of the lakes had background chloride concentrations of >8 mg/L prior to the 1850s, this had increased to 46% and 18%, respectively, by present times. Such increases are related, in part, to the number of highways in the area, land disturbance and road salt use. Furthermore, researchers are beginning to note changes in algal communities as these chloride levels increase from background levels of 1.8-3.6 mg/L to over 7 mg/L (Dixit et al., 1999). Studies conducted in Canada, although less detailed, are also pointing towards increased chloride levels, in the range at which changes in algal composition may be expected, in small ponds and lakes located near certain highways and in a variety of streams and lakes in an urban setting. However, it is important to note that the vast majority of these studies are limited to areas of heavy road salt usage, primarily southern Ontario, the Maritimes and Quebec.
1 Data are from Figure 17.
The U.S. EPA (1988) developed water quality guidelines for chloride and concluded that, except possibly where a locally important species is very sensitive, freshwater organisms and their uses should not be affected unacceptably if:
- the 4-day average concentration of chloride, when associated with sodium, does not exceed 230 mg/L more than once every 3 years on average;
- the 1-hour average chloride concentration does not exceed 860 mg/L more than once every 3 years on average.
They noted that these criteria will not be adequately protective when the chloride is associated with potassium, calcium or magnesium, and that because animals have a narrow range of acute sensitivities to chloride, excursions above this range might affect a substantial number of species. Evans and Frick (2001) also found evidence that potassium chloride and magnesium chloride salts were more toxic than sodium chloride. In addition, fish appeared to be less sensitive to calcium chloride than to sodium chloride, and the converse was true for invertebrates, although the data sets were small. Although the focus of the assessment was primarily on sodium chloride, they concluded that potassium chloride and magnesium chloride salts appear to be as toxic as or even more toxic than sodium chloride.
To conclude, it is considered that high concentrations of chloride associated with road salts may have immediate or long-term harmful effects on surface water systems, based on several exceedences of effect thresholds in the environment and on field evidence of ecosystem population effects. These effects are most likely to occur at improperly managed road salt storage depots, at inappropriately placed snow dumps and in small watercourses along a dense network of highways. At lower concentrations, increased chloride concentrations may affect community structure, diversity and productivity. Elevated chloride levels have been shown to increase metal bioavailability; by affecting density gradients in lakes, road salts may have a major impact on lake ecosystems, notably in terms of depth-dependent availability of oxygen and nutrients. Most instances in which this occurs are in areas of high road salt usage, primarily southern Ontario, Quebec and the Maritimes.
Figure 17 Experimental acute toxicity and predicted chronic toxicity for aquatic taxa (from Evans and Frick, 2001)
Figure 18 Representative short-term chloride concentrations in the Canadian aquatic environment associated with contamination by road salts and concentrations causing adverse biological effects following brief exposures
Figure 19 Representative long-term chloride concentrations in the Canadian aquatic environment associated with contamination by road salts and concentrations causing adverse biological effects following prolonged exposures
3.4 Soils
3.4.1 Soil salinization resulting from road salt application
This section summarizes the information presented in the report prepared by Morin et al. (2000) on the extent of soil salinization due to road salt application. The report presented soil maps for all provinces in Canada. This mapping effort is complemented with a few case studies designed to show how the salinity of soil solutions changes dynamically in salt-affected areas, including short distances from point and line salt sources, and with season. The salt ions examined are sodium, calcium, sulphate and chloride. Contributions due to other ions such as potassium and magnesium are neglected, because these generally contribute little to overall soil salinity in most locations.
Mapping of road salt inputs (annual inputs) and salt concentrations in soils was done as spatial averages compiled for Level 2 watersheds, as spatial averages at the municipality level and as spatial averages at the road level (mainly principal roads and expressways). These three levels of analysis were done because of the general connection between soil salinity and surface water salinity, i.e., most surface waters receive direct input from soils as a result of runoff, interflow, soil percolation and baseflow. Similarly, downslope soils often receive seepage water from higher surface water locations. Gradual dissolution of easily weathered soil and rock minerals further contributes to downslope soil salinization, as do agricultural and horticultural fertilizer applications and seepage from domestic and industrial sewage. Evaporation of water from dry soils in arid regions such as the Prairie provinces also contributes to soil salinity, which, in the extreme, leads to above- and below-ground salt accumulations, especially in landscape depressions and on hillslope seeps. Industrial activities, especially those associated with salt mining and with the sodium hydroxide-based extraction of fossil fuels from tar and oil sands, add further to environmental salt loading in areas specific to these activities. This section, however, focuses on the extent to which road salt applications in combination with atmospheric deposition contribute to the overall salinity in soils next to roads, and how these applications may affect the general background salinity of water as it reemerges from the soil in various downslope locations. In general, calculated average Level 2 soil salt concentrations from road salt loadings are expected to be lower than area-wide municipal averages, and area-wide municipal averages are expected to be lower than roadside soil averages. Hence, a three-way mapping effort was done to estimate:
- expected average salt concentrations in soil solutions based on total road and atmospheric salt inputs per Level 2 watershed;
- expected average salt concentrations in soil solutions based on total road and atmospheric salt inputs per municipal area; and
- expected average salt concentrations in soil solutions based on total road and atmospheric salt inputs in immediate roadside vicinities.
To quantify these averages, information contained in various national, provincial and municipal databases was compiled to determine:
- local average annual road salt applications, by municipality and county;
- local average annual atmospheric deposition regarding precipitation and wet sodium, calcium, chloride and sulphate ion deposition (a dry deposition database does not yet exist; including dry deposition further adds to overall salt loadings, especially in the Atlantic region on account of sea spray);
- average annual runoff; and
- local soil depth, clay content and organic matter content.
For the mapping effort, seasonal aspects are not considered, but soil salinities would vary depending on the volume of road salt loadings, weather and topography. Some of the seasonal aspects are addressed in this section by way of special case studies that deal with soil and surface water salinity in downslope locations from a salt depot.
3.4.1.1 Definitions and equations
Compiled data on atmospheric sodium, calcium, chloride and sulphate ion loadings refer to volume-weighted ion concentrations (wet deposition only), either in mg/L or in meq/L. Salt loadings, in contrast, are generally expressed in terms of g/m2 per year or kg/ha per year. In this report, atmospheric and road salt loadings are converted into total dissolved salt concentrations in soil solutions by dividing the calculated combined annual salt loadings by the expected annual soil percolation rate, where:
soil percolation rate (mm/year) = precipitation rate (mm/year)– actual evapotranspiration rate (mm/year) – surface runoff (mm/year)
For simplicity, annual soil percolation rates (mm/year) are estimated to be equivalent to stream discharge rates (mm/year), where stream discharge rates are obtained from watershed-specific hydrometric discharge observations. For the mapping effort, expected soil percolation rates are therefore set equal to stream discharge rates. Maps located in Morin et al. (2000) depict:
- millimetres of stream water discharge at a national scale, as obtained from Natural Resources Canada; and
- cubic metres of stream water discharge per Level 2 watershed (as obtained from the national stream discharge map).
Watersheds with highest runoff coefficients occur along the Pacific and the Atlantic coasts. Watersheds located in the southern part of the Prairie provinces have the lowest quantity of runoff to dissolve and dilute the salts. Areas with low rates of runoff are important to identify because these areas have high road salt concentrations in runoff.
For the mapping of these parameters, it was also essential to obtain national soil survey information on:
- soil clay content (as a % of mineral portion of soil) (particulate matter <2 mm);
- soil organic matter (in %); and
- cation exchange capacity (in meq/100 g oven-dry soil).
The soil solution estimates for total dissolved salt loadings and related ion concentrations were used to calculate:
- total cations and total anions (in meq/L);
- soil electrical conductivity (in mmho/cm or mS/cm);
- osmotic potential (in bar);
- sodium absorption ratio (dimensionless);
- exchangeable sodium ratio (dimensionless);
- exchangeable sodium percentage (in %);
- exchangeable sodium content (in meq/100 g oven-dry soil);
- soil clay and silt dispersion (as % of soil clay content); and
- relative changes in soil hydraulic conductivity (dimensionless).
Likely salt-induced effects on soils include substantial lowering of soil osmotic potentials, increased soil swelling, reduced soil stability (loss of soil structure), decreased soil permeability, increased potential for soil erosion, increased clay and silt dispersion, and increased turbidity in surface waters. All of these impacts get worse with increasing sodium content in the soil. This is because increased sodium ion contents decrease the affinity that soil particles have for each other on account of the strong affinity between sodium ions and water molecules. To some extent, salt-induced effects depend strongly on soil clay mineralogy (e.g., soil swelling and shrinking in soils containing large portions of montmorillonite). In some other respects, salt-induced effects are independent of soil mineralogy and are simply affected by electrolyte-mediated affinity among individual particles in the soil or in suspension (e.g., affinity is regulated by interparticulate electrolyte forces that determine the degree of particle coagulation and dispersion; see Ali et al., 1987). With increasing soil dispersion (e.g., when salinized soils receive low electrolyte irrigation water and are then subjected to water and wind erosion), not only does dispersion include fine soil particles, but these particles may also be carriers for environmental contaminants such as nutrients, heavy metals and microbiota by way of water-and wind-induced soil erosion.
Soils that have a relatively high clay content and continuously or periodically receive high sodium inputs will likely show the strongest impacts in the long run (i.e., should receive a high salt hazard rating). To map areas with a high salt hazard rating, a simple salt hazard index was obtained by setting:
salt hazard index = exchangeable sodium ratio x soil clay content (in %)
In this way, areas with highest hazard ratings were located for southern Ontario near the Greater Toronto Area and in southern Quebec around the Island of Montréal (Figure 20). Other urban areas in southern Ontario and southern Quebec also have high ratings.
Additional areas are found in the southern portions of the Prairie provinces, where soil salinization on account of natural processes is a major regional concern (see Eilers et al., 1995).
3.4.1.2 Case studies
Three case studies were conducted at Kejimkujik National Park, Nova Scotia, along select streets in Fredericton, New Brunswick, and in a forested area next to Fredericton's municipal salt depot (University of New Brunswick woodlot). These studies focused on:
- providing numbers for the general turnover rate of sodium ions in the soils of headwater basins at Kejimkujik National Park (an environment that is essentially free of road salts but is influenced by sea spray);
- assessing the status of soil salinity underneath lawns next to the curb of select roads in Fredericton during the fall season;
- examining how the salinity of surface water changes with distance from Fredericton's salt depot, by season; and
- examining the salinity of ditch water along the Fredericton-New Maryland Highway.
The results of these case studies indicated that the ability of soils and watersheds in the Kejimkujik National Park to retain incoming sodium and chloride ions is very low, i.e., sodium and chloride ions are readily flushed through the soils and through the basins of the park. Ion concentrations are highest in soils and streams during early fall, and lowest concentrations occur at the time of snowmelt.
Along the city streets of Fredericton, generally little sodium remains on the ion exchange sites in soils next to the road, as observed in early fall. The data indicate that ion exchangeable sodium is particularly low along the most frequently travelled roads at this time of year. Furthermore, exchangeable sodium concentrations are lower next to curbs than 2-3 m away from curbs. This difference indicates that repeated and substantial traffic splashing and road runoff during spring, summer and fall are flushing sodium out of the roadside soils.
Compared with Kejimkujik National Park, background sodium and chloride ion concentrations in soil solutions and in surface waters are somewhat smaller at the University of New Brunswick's woodlot (2 mg/L versus 3 mg/L, respectively). Chloride concentrations in ditch and stream water around the University of New Brunswick's woodlot, however, vary by location. Arp (2001) measured the electrical conductivity (later related to chloride concentrations) of ditch water along the Fredericton-New Maryland Highway and Bishop Drive for representative days in June, July, October (1999) and January (2000). Results are summarized in Figure 21 by way of box plots, according to the hydrological condition at each point of measurement. Ditches in this analysis were categorized according to the following three location characteristics:
- well-drained locations where the ditches tend to run dry (crest locations);
- poorly drained areas where water accumulates and becomes stagnant; and
- areas of continuous water flow, i.e., brook road crossings (culverts).
Figure 21 shows that electrical conductivities of ditch water peak in July, especially in areas where water is not stagnant.
Figure 20 Areas with low to medium to high road salt application hazards, based on product of soil clay content with exchangeable sodium ratio (from Morin et al., 2000)
At stagnant locations, electrical conductivity readings remain fairly high regardless of season and tend to be highest in June and July, but are lowest and generally least variable in the fall. Where streams, brooks and/or ditch water cross underneath the highways (by way of culverts), conductivities are lowest, at or near background level, due to continuous flushing - i.e., ditch water, as it enters streams and brooks, is quickly diluted, on a year-round basis. However, where ditch water drains into poorly drained, water-accumulating depressions and where the water turnover rate is low, dilution does not occur, and electrical conductivity levels (and, therefore, salinity levels) remain elevated year-round.
Chloride concentrations are highest at and near the salt depot, ranging from 300 to more than 1000 mg/L. Listed in Table 18 are chloride concentrations for stream and pond water with increasing stream catchment area of the Corbett Brook downstream from Fredericton's salt depot at various times of the year (May, June, July, October, January). At first, considerable dilution occurs in the nearby "Larch Swale Pond" (0 km in Table 18), which receives water from the 172-ha headwater system. Below this area, dilution continues either gradually or in jumps, with each jump being specific to the catchment area of each of the adjoining brooks. For example, at about 2 km away from the depot, Corbett Brook is joined by the O'Leary Brook. At that point, dilution increases sharply, but in proportion with the combined catchment area for O'Leary Brook and Corbett Brook.
Figure 21 Box plot showing 10th, 25th, 75th, and 90th percentile plus individual data above and below the 10th and 90th percentile for ditch water electrical conductivity and chloride concentrations along two highways near Fredericton's salt depot, outside the depot's catchment area for representative days in June, July, October 1999, and January 2000. "Well-drained", "cross-flow", and "stagnant" refer to three conditions: ridges from which water drains, stream/road crossings, and ditch depressions with no visible outflow, respectively (from Arp, 2001).
Season has a strong effect on electrical conductivity measurements and the related stream dilution factor. In principle, conductivity measurements are related to stream discharge rates: high discharge rates imply increased dilution, and therefore decreased conductivity values; low discharge rates imply less dilution, and therefore increased conductivity values. For intermittent stream flow conditions (as may occur in summer and early fall), the overall influence of the salt depot appears to be confined to the vicinity of the salt depot, while areas further removed from the salt depot receive less salt-enriched water. This "apparent" dilution with increasing catchment area at lower stream locations is greater during times of intermittent flow (such as July) than during times of continuous flow. In July, concentrations greater than 250 mg/L were observed up to 500 m from the depot and were still 100 mg/L at 1250 m from the depot.
3.4.1.3 Discussion
Road salt application effects on physical and chemical soil parameters such as osmotic potential, electrical conductivity, soil permeability, soil dispersion and exchangeable sodium ratio (see Morin et al., 2000, for detailed information on other physical and chemical parameters) are all calculated to be highest in areas next to salt depots and roads receiving salt inputs. At the municipal level, impacts of soil waters on downslope positions can be expected to vary greatly depending on the density of the road network, on local conditions, on the actual location of road salt applications and on the physical condition of salt storage facilities. For example, in highly paved urban areas, road salt applications are expected to produce higher overall salt concentrations in soils and surface waters than in less developed areas. Williams et al. (1999), for example, showed that salt concentrations in springs within the Metropolitan Toronto area increase as predominant land use changes from rural to urban. These authors also showed that:
- these salt concentrations peak during the time of salt application;
- some of these peaks have chloride ion concentrations in excess of 1000 mg/L; and
- secondary salt concentration peaks occur in midsummer (thereby supporting the results of one of the above case studies), with highest chloride ion concentrations in excess of 200 mg/L.
Table 18 Modelled electrical conductivity values and chloride concentrations for Corbett Brook, downstream from Fredericton's salt depot Catchment area (ha) Distance Electrical conductivity (mS) Chloride concentrations (mg/L)
Salt concentrations in surface water, groundwater and soil solutions, therefore, may be high at the time of road salt application, but elevated concentrations may recur later in summer during periods of drought when water evaporates and when salty water from, for example, subsoils seeps back into downslope locations. During times of high precipitation, low electrical conductivity water will help to flush briny water from these areas. Along roadsides, frequent splashing during rain events may further accelerate the displacement of briny water and the return to low soil exchangeable sodium percentage values, but this can occur effectively only in soils of high permeability in areas where water does not accumulate (i.e., does not stagnate for significant parts of the year).
Underneath paved surfaces (as in urban areas), soils may not benefit from seasonal salt flushing. Salt in these locations likely may remain concentrated or may accumulate even more with additional salt applications, because these locations are, at least in part, outside the paths of percolating soil water. These locations, therefore, become a source of soil and surface water salinity, by way of either slow seepage or high-salinity flushes during times of pavement repair or unusually wet weather.
The overall salt impacts on soils need to be evaluated in terms of landscape-level source/sink positioning. For example, the location of the roadside in relation to the road (above road, below road, same level as road, upwind from road, downwind from road) and terrain conditions along roads (e.g., open versus forested, well-drained versus poorly drained) have to be considered. Typically, salts are spread from road to roadside by vehicle spray, dust and runoff. Accordingly, soil salinity is expected to increase from ridge tops (recharge areas) to poorly drained depressions.
The process of salt dilution through subsequent precipitation and traffic splashing is important in terms of allowing much of the Canadian soil roadscape to recover partially or completely from seasonal road salt applications. Apparently, splashing from roadsides accelerates the leaching loss of winter-accumulated salt from the roadsides, as noted in one of the above case studies. However, during times of high exchangeable sodium percentage values and low electrical conductivity splash water, this effect could have deleterious effects on soil aggregation, e.g., soil clay and silt dispersion, which, in turn, can lead to reduced soil infiltration rates and thereby enhanced soil erosion by water and enhanced dispersion of soil particles by wind. This effect would be highest where roadsides are not covered by vegetation.
Figure 22 shows expected average values for total dissolved solids, osmotic potential and electrical conductivity of soil solutions and surface water at the Level 2 watershed level and at the municipal area level. An inspection of the total dissolved solids results indicates that these values are all higher than the "background" levels that would be expected due to local atmospheric deposition and subsequent evapotranspiration alone. These area-averaged concentrations, however, are far from being of concern, except for highly urbanized areas. As expected, calculated values for total dissolved solids, osmotic potential and electrical conductivity increase with decreasing area averaging.
Values for total dissolved solids, osmotic potential and electrical conductivity for soil solutions and surface water for roadway rights-of-way were calculated. These values are based on provincial road salt loading data, by maintenance district. Overall, these calculations indicate that areas immediately surrounding the right-of-way tend to have high annual averages for total dissolved solids, osmotic potential and electrical conductivity. Seasonally, these values would be higher through periods of salt application and substantially lower during periods of high flushing. High peaks, however, can be expected to return during midsummer in areas of water accumulation and may remain high in stagnant water pools or depressions with limited surface runoff, as discussed above.
The map in Figure 23 suggests that certain areas in Ontario and Quebec may have roads with elevated exchangeable sodium percentage values. These values are primarily associated with areas that have high road salt loadings. Figure 23 suggests that certain areas in the Prairie provinces may also have high exchangeable sodium percentage values. While roads in the Prairie provinces have low road salt loadings compared with most municipal areas in Ontario and Quebec, there is less precipitation to dissolve and dilute the salts. The situation is such that many soils in southern Alberta and southern Saskatchewan and some soils in southern Manitoba are already saline due to natural causes, especially in depressions and near wetlands and drainage courses. Here, road salts would add to the total electrolyte loads of soils, as these would subsequently accumulate in the depressed areas next to the roads. In general, the spatial patterns of the background levels of electrolyte ions in the soil solutions based on atmospheric deposition alone match the spatial patterns of soil salinity as displayed by Eilers et al. (1995) fairly well, thereby confirming the simple mass balance protocol of the model that produced the maps of this report.
Figure 22 Estimates for total dissolved solids (TDS) and electrical conductivity (EC) of average soil solution and surface waters, by level 2 watersheds and by municipal boundary, Ontario and Quebec (from Morin et al., 2000)
Increased soil salinization through road salt applications may affect soil biota (macro- and microflora and macro- and microfauna). For a simplified summary of expected plant sensitivities and related species-specific threshold tolerances to falling soil osmotic potentials, refer to Figure 24. Figure 24 associates the sensitivity of crops to soil salinity by demonstrating how crop yield decreases with increasing levels of electrical conductivity. The dashed boxes in this figure show the percentage of provincial roadways with certain levels of estimated soil salinity. Data in this graph suggest that salinity in soil solution alongside the majority of provincial roads may adversely affect many shrub and tree species. This graph also suggests that certain roadsides may have levels of soil salinity that can adversely affect salt-tolerant plants like wheatgrass. Impacts on terrestrial vegetation are discussed more comprehensively in Section 3.5.
Figure 23 Exchangeable sodium percentage (ESP) along roadsides, by provincial road maintenance district (from Morin et al., 2000)
Road salt effects on soils are expected to be highly dependent on the extent of vegetative cover. Soils that are bare likely suffer the worst impacts in terms of surface crusting, surface runoff, and clay and silt dispersal by water and wind. Soils that are originally covered by vegetation will also suffer impacts once the surface vegetation reacts negatively to high negative osmotic potentials in the soil.
Here, gradual loss of surface vegetation will lead to increased mineral soil exposure; this leads to gradually decreasing levels of soil organic matter and, therefore, increasing soil dispersion, and thus decreasing soil permeability and increasing soil erosion. Proper vegetation management, salt application and road maintenance practices would obviously go a long way to reducing any such effects. Additional details on soil salinity and many other aspects related to soil management and soil conservation can be found in, for example, Anon. (1992), Holms and Henry (1982), Steppuhn and Curtin (1992) and Acton and Gregorich (1995).
Other impacts of road salts not covered in this section deal with salt-induced release of heavy metals (e.g., zinc, copper, cobalt, mercury, cadmium) and base cations (calcium, magnesium, potassium). All of these may have positive or negative effects on soil flora and fauna, including:
- encouraging the growth of certain species in otherwise nutrient-poor ecosystems;
- enhancing the loss of plant nutrients such as calcium, magnesium and potassium due to ion exchange with sodium ions; and
- heavy metal mobilization in soil and aquatic habitats, depending on local circumstances.
Heavy metal mobilization occurs through formation of water-soluble chloride-metal cation complexes, thereby rendering many heavy metal cations more phyto-available. This would increase the entrance of heavy metals into the local food chain by way of plant uptake. The plants themselves could experience heavy metal toxicity symptoms in the worst-case scenarios. For an example of the processes involved, see Smolders and McLaughlin (1996) regarding the case of chloride-enhanced cadmium mobilization and subsequent phyto-uptake. For additional details regarding soil salinization effects, see, for example, Bresler et al. (1982), Eilers et al. (1995), Butler and Addison (2000) and Cain et al. (2001).
3.4.1.4 Conclusions
Application of road salts can result in deleterious effects on physical and chemical parameters of soils, especially in salt-affected areas that also suffer from general salt, soil and vegetation management neglect. Highest effects are calculated in areas that would be directly impacted by salt depots and along roadsides, especially in poorly drained depressions.
Electrolyte-related effects impact on soil structural stability, soil dispersion, soil permeability, soil swelling and crusting, soil electrical conductivity and soil osmotic potential. Surface waters downslope from salt-affected soils are also affected by briny seepage water or briny runoff with salt dispersed sediments. All of this has, in turn, abiotic and biotic impacts on the local environment. Abiotic impacts primarily deal with loss of soil stability during drying and wetting cycles and during periods of high surface runoff and wind; biological impacts primarily deal with osmotic stressors on soil macro-and microflora and macro- and microfauna and salt-induced mobilization of macro- and micronutrients, which, in turn, affect flora and fauna in situ and in downslope seepage areas.
