Priority Substances List Assessment Report for Inorganic Chloramines

3.0 Assessment of "Toxic" under CEPA 1999

3.1 CEPA 1999 64(a): Environment

The environmental risk assessment of PSL substances is based on the procedures outlined in Environment Canada (1997a). Analysis of exposure pathways and subsequent identification of sensitive receptors are used to select environmental assessment endpoints (e.g., adverse reproductive effects on sensitive fish species in a community). For each endpoint, a conservative Estimated Exposure Value (EEV) or Estimated Environmental Concentration (EEC) is selected and an Estimated No-Effects Value (ENEV) is determined by dividing a Critical Toxicity Value (CTV) by an application factor. A conservative (or hyperconservative) quotient (EEV/ENEV or EEC/ENEV) is calculated for each of the assessment endpoints in order to determine whether there is potential ecological risk in Canada. If these quotients are less than one, it can be concluded that the substance poses no significant risk to the environment, and the risk assessment is completed. If, however, the quotient is greater than one for a particular assessment endpoint, then the risk assessment for that endpoint proceeds to an analysis where more realistic assumptions are used and the probability and magnitude of effects are considered. This latter approach involves a more thorough consideration of the sources of variability and uncertainty in the risk analysis.

3.1.1 Assessment endpoints

In Canada, releases of inorganic chloramines are in aqueous solution destined for aquatic environments. Hence, this assessment is based on water-phase exposures. Trichloramine is not of environmental significance, since conditions rarely occur that support its formation. Because inorganic chloramines are in dynamic equilibrium with other residual chlorine species, and due to analytical limitations that usually preclude species-specific analysis, this assessment often uses data pertaining to TRC and TRO. In reality, observed effects are usually as a result of the simultaneous presence of different chlorine residuals whose individual effects are indistinguishable.

Releases of inorganic chloramines to surface waters are due to effluent, cooling water and potable water releases. However, some drinking water releases (e.g., due to outdoor water use, main breaks and leaks) are to soils.

Both monochloramine and dichloramine are water soluble and are known to volatilize into the air phase and partition to or react with sediments. However, there are no data regarding their presence or fate in air and sediment phases. Impacts due to residual chlorine have not been documented in any phase except for the water phase. Historically, their presence in air and sediment phases has not appeared to be as great a concern as concentrations in surface waters. Therefore, this assessment will focus on the water-phase exposures of inorganic chloramines.

Decay data indicate that inorganic chloramines are not persistent in the environment; thus, the assessment of releases focuses on exposures near point sources and acute and subacute effects on receptor organisms.

3.1.1.1 Releases to soils

The assessment endpoint for soils was mortality/recovery of microorganisms and soil processes. Soil microorganisms are important for nutrient cycling and decomposition of organic matter and are thus important for plant growth. Reductions in microbial populations can inhibit plant growth.

3.1.1.2 Releases to water

Assessment endpoints for surface waters (fresh water and seawater) include the survival of sensitive invertebrates and fish. Invertebrates are an essential component of aquatic ecosystems. Benthic invertebrates (e.g., amphipods and isopods) facilitate detrital decomposition. Both benthic and pelagic invertebrates consume bacteria and phytoplankton and themselves serve as sources of food for many fish species. For instance, cladocerans from the family Daphniidae, which includes Daphnia spp. and Ceriodaphnia spp., are ubiquitous in temperate lakes and ponds, as well as in quiescent sections of streams and rivers throughout North America. Daphnids are often more sensitive than other aquatic organisms to various chemicals and as such are good surrogates for the protection of other aquatic life. Cladocerans are ecologically important species, since they convert phytoplankton and bacteria into animal protein (Environment Canada, 1992). They are representative of other larger and smaller invertebrates that together act as food sources for many fish. They also form a significant portion of the diet of many fishes, including salmonids, which are themselves an important food, economic and cultural resource for Canadians. Some birds and terrestrial mammals also depend on the presence of fish as a food source.

Saltwater invertebrates are equally important to the functioning of marine and estuarine ecosystems. Amphipods are an abundant component of benthic communities in estuarine and marine environments and are a primary food source for certain species of whales and for many species of birds, fish and larger invertebrates.

Amphiporeia virginiana and E. washingtonianus are two important and commonly found amphipods in Canadian waters. Marine fish serve as an important food, economic and cultural resource for Canadians.

3.1.2 Environmental risk characterization

It was determined that the assessment in soils would be qualitative at the hyperconservative level rather than follow the quotient approach, since the available data did not facilitate a quantifiable approach.

The risk assessment in surface water followed a tiered risk assessment approach as identified in the guidance manual (Environment Canada, 1997a).

3.1.2.1 Soil organisms

No information is available that was directly relevant to the effects of inorganic chloramines in soils. However, the available evidence indicates that negative impacts on soil microorganisms from inorganic chloramines are unlikely. First, a proportion of the inorganic chloramine would be lost prior to entering the soil environment (e.g., from volatilization, photolysis, reaction with particulates) and hence would not come into contact with soil microbes. Upon infiltrating soils, the treated water would be exposed to a variety of organic materials that are extremely reactive with inorganic chloramines. These organic substances serve as effective reducing agents that change the form of inorganic chloramines and bind them to the soil matrix. Although there are limited data regarding these transformation products and their toxicity, their disinfection potential is usually considered limited. According to Zellmer et al. (1987), hypochlorous acid applied in the form of calcium hypochlorite will be immobilized and deactivated by a mineral soil (i.e., fine-silty clay loam).

The disinfection molecule in aqueous solution must come into contact with the microorganism in order for inactivation to occur. The presence of particulates can provide protection to microorganisms against disinfectants. The protection afforded bacteria associated with surface solids would most likely result from physical interference with the transport of the chloramine molecules towards the organism, because of a barrier of charges associated with the particulate (Gerba and Stagg, 1979). Microorganisms embedded in particulate matter may be afforded significant protection from a disinfectant (Berman et al., 1988).

In addition, it should be noted that there have been no historic accounts of environmental impacts resulting from inorganic chloramine release to soils or to any phase other than water.

Since populations of soil microorganisms and soil processes are not likely to be harmed from the application of inorganic chloramines to soils, the assessment of chloramine ri sk in soils did not proceed to Tier 2.

3.1.2.2 Aquatic organisms

3.1.2.2.1 Hyperconservative risk assessment

Receptors and assessment endpoints

For the hyperconservative assessment, the CTV was based on the most sensitive species receptor, and the assessment endpoint was mortality. The value of 0.01 mg/L was the lowest concentration resulting in the mortality of 50% of a test population found in the published literature respecting freshwater and marine biota based on the initial review of the literature. Kaniewska-Prus (1982) exposed D. magna to chloramines (80% monochloramine, 17% dichloramine and 3% trichloramine, measured by the orthotolidine method; see Kaniewska-Prus and Sztrantoicz, 1979) in a static freshwater bioassay and derived a 24-hour LC₅₀ of 0.0108 mg/L. Capuzzo (1979b) exposed the American oyster, C. virginica, to chloramines in continuous-flow seawater bioassays and derived a 30-minute LC₅₀ of 0.01 mg/L. The concentration 0.01 mg/L was then divided by an application factor to translate it to an ENEV. There are no rules for the selection of an application factor, but Environment Canada (1997a) suggests a maximum application factor of 100 for converting the lowest acute LC₅₀ or EC₅₀ from a data set to a hyperconservative ENEV. Therefore, the hyperconservative ENEV = 0.0001 mg/L (0.01 ÷ 100).

Hyperconservative assessment of chloramine exposure from effluents, cooling waters and drinking water

The hyperconservative evaluation was conducted early in the assessment process when there were very few data regarding inorganic chloramine concentrations in aquatic environments. As a result, maximum end-of-pipe concentrations were used as freshwater and saltwater EEVs. Based on the data available at the time, the hyperconservative EEVs were 3.0 mg/L for drinking water, 3.56 mg/L for chlorinated municipal effluent and 2.0 mg/L for chlorinated cooling water from electrical utilities. These represented maximum TRC concentrations described by various sources (Norecol Environmental Consultants Ltd. and Dayton and Knight Ltd., 1992; Government of Canada, 1993). TRC was used as a surrogate for inorganic chloramine since, in a worst-case scenario, all TRC present could be in the form of inorganic chloramine.

Using the above-indicated EEVs and ENEV, the following hyperconservative quotients were derived:

• treated potable water at its source:
  3.0 mg/L ÷ 0.0001 mg/L = 30 000
• chlorinated municipal effluent:
  3.56 mg/L ÷ 0.0001 mg/L = 35 600
• chlorinated cooling water:
  2.0 mg/L ÷ 0.0001 mg/L = 20 000

Quotients exceeded 1 by very large margins. Hence, the risk assessment for inorganic chloramine in surface waters proceeded to a conservative-level assessment.

3.1.2.2.2 Conservative risk assessment

Receptors and assessment endpoints

Table 5 Calculation of no-effects factors used in conservative assessments

Time (hours)	Time (minutes)	LC50 (mg/L)	LC0 (mg/L)	No-effects factor1
8	480	0.0809	0.0412	0.51
48	2880	0.0118	0.0074	0.63

A conservative approach was taken for the derivation of ENEVs using the most sensitive freshwater and saltwater species. This involved a rigorous evaluation of existing data and newly available data provided through additional acute toxicity testing, supported by the best available analytical chemistry (see Farrell and Wan, 2000: Appendices B-F).

The ENEVs derived for the conservative assessment were based on the recommendations identified in Appendix H in Farrell and Wan (2000). The ENEVs were derived for C. dubia because it was more sensitive to monochloramine than chinook salmon for exposure times greater than 1 hour and because, as a group, invertebrates were found to be many times more sensitive to inorganic chloramine and residual oxidant exposures than fish (Farrell and Wan, 2000). Although some freshwater and marine invertebrates from the open literature appeared to be more sensitive to monochloramine than C. dubia, the data set for C. dubia was very comprehensive. Using these data, Farrell and Wan (2000: Appendix H) derived a reference line (the lowest reference concentration for 50% lethality) against which to compare the sensitivity of other organisms. Using their analysis, an 8-hour LC₅₀ of 0.060 mg/L and an incipient LC₅₀ of 0.018 mg/L were derived for C. dubia. The incipient toxicity level is defined as the concentration of chemical that was lethal to 50% of the test organisms as a result of exposure for periods sufficiently long that the acute lethal action essentially ceases. The incipient level is also the asymptote of the toxicity curve or that part of the curve that is parallel to the time axis. The asymptote produced by the C. dubia LC₅₀ model occurred at 17.9 hours. Therefore, at times ≥17.9 hours, the LC₅₀ concentration (0.018 mg/L) remains essentially the same.

In order to account for data points that were below those of the C. dubia model, a species sensitivity application factor was used in conjunction with the incipient LC₅₀ (≥1073 ¹ minutes) of 0.018 mg/L. In keeping with the conservative approach of the Tier 2 assessment, it was still necessary to convert the LC₅₀ data to ENEVs. Farrell and Wan (2000: Appendix B) undertook monochloramine toxicity studies with C. dubia and derived power equations to describe the LT50 and LT0 data (valid for exposures of 10-3200 minutes).

Using these equations, the LC₅₀ and LC₀ are calculated for 8 hours and 48 hours, and the differences or ratios between the two values (i.e., 8-hour LC₀/8-hour LC₅₀ and 48-hour LC₀/48-hour LC₅₀) were used as application factors to calculate the ENEVs (see Table 5).

Therefore, the 8-hour ENEV for freshwater organisms can be calculated as follows:

The incipient ENEV (for durations ≥1073 minutes) for freshwater organisms can be calculated as:

The C. dubia mathematical model for acute toxicity was adopted as a point of reference for determining a suitable lower b ou nd ary line for marine invertebrates due to insufficient acute toxicity data with which to perform reliable modelling with marine and estuarine invertebrate species (Farrell and Wan, 2000). For the C. dubia mathematical model, a species sensitivity factor of 0.1 was recommended by Farrell and Wan (2000: Appendix G) for exposure times less than that of the "time to incipient LC₅₀" (i.e., 1073 minutes), and a species sensitivity factor of 0.25 was recommended for exposure times greater than 1073 minutes. These factors are suggested to account for CPO effects data identified in the open literature. In seawater, the addition of chloramines will result in the formation of bromamines. For marine environments, the species sensitivity factor of 0.25 was reasonable given that the EC₅₀ for sublethal effects on oyster larvae is <0.005 mg/L. Using the C. dubia mathematical model, an incipient LC₅₀ of 0.018 mg/L was predicted, and the species sensitivity factor of 0.25 lowers this to 0.0045 mg/L.

Using the C. dubia mathematical model for acute toxicity (Equation 1), the marine species sensitivity factors of 0.1 and 0.25 identified above and the no-effects factors identified in Table 5, ENEVs for saltwater organisms are calculated.

The 8-hour ENEV for saltwater organisms is calculated as follows:

The incipient ENEV (for durations ≥1073 minutes) for saltwater organisms is calculated as:

Conservative assessment of chloramine exposure from effluents and cooling water

The assessment of chloramines is national in scope, with a diversity of discharge and exposure scenarios. To facilitate and simplify the risk assessment, discharges with similar characteristics were grouped together into discharge categories. Assessments were conducted for discharges in each category (see Table 6).

There were insufficient measured concentrations to screen chloramine exposures in the environment. Hence, the conservative assessment focused largely on estimated data produced from mathematical models (see McCullum et al., 2000; Pasternak, 2000). For effluents, the purpose was to screen characteristics of chloramine discharges into Canadian surface waters and identify which discharge settings may have the greatest impact on receiving aquatic environments. Mixing models appropriate for describing the dispersive behaviour of residual chloramines in surface waters were used on a number of specific settings.

Characteristics of chloramine discharges from over 110 WWTPs were screened to identify the highest chloramine discharge loadings to surface waters. The highest discharge loading of chloramines (measured as TRC) in each of four regions of Canada was retained for modelling. These regions included western/coastal, Prairie, Great Lakes, and eastern/coastal. Eighteen discharge settings (12 wastewater, 5 drinking water treatment and 1 cooling water) in Canada were subjected to residual chloramine dispersion modelling in surface waters under defined conditions (see McCullum et al., 2000).

Table 6 Matrix of different discharge situations

Enlarge image

In keeping with the conservative intent, all effluent and cooling water discharges were treated equally as continuous discharges, even though some were intermittent. Also, inorganic chloramine decay was assumed to be slow (1 per day, half-life of 0.69 days). Hydrological data (flow rate, current velocity, depth and width) represented 20-year arithmetic means for the period 1976-1995, or shorter periods if there were insufficient data to determine the 20-year average. TRC was modelled, since appropriate chloramine concentrations were not available. It was not possible to estimate the proportion of TRC that was in the inorganic chloramine form. The use of TRC as a surrogate for inorganic chloramines meets the conservative objective of the assessment (McCullum et al., 2000).

Industrial effluents were considered for the Tier 2 assessment; however, these data were not of as high quality as the municipal data. Industrial effluents were considered to be part of the same discharge category as municipal effluents, and the results from municipal facilities were assumed to be representative of industrial si tuat ions . Th erefore, chloramine discharges from industrial facilities were considered to be equivalent in nature to chlorinated municipal wastewater discharges. One industrial discharge was modelled for the conservative assessment (the Clover Bar Generating Station cooling water discharge to the North Saskatchewan River at Edmonton, Alberta).

Results of the conservative assessment are presented in Tables 7 and 8, and these show that seven discharges (i.e., Rossdale WTP, E.L. Smith WTP, R.O. Pickard Environmental Centre, Saskatoon WWTP, Lethbridge WTP, Britannia WTP and Toronto Humber WWTP) produced quotients of 1-10, and four discharges (i.e., Ashbridges Bay WWTP, North Toronto WWTP, Okotoks WWTP and Clover Bar Generating Station) produced quotients of 10 or greater (McCullum et al., 2000). The discharges resulting in quotients greater than 10 were recommended for a probabilistic risk assessment. In addition, the Rossdale WTP was recommended for a probabilistic risk assessment since its discharge resulted in the highest quotients from a water treatment facility waste. This discharge is also intermittent, and thus it is different from the continuous WWTP discharges but similar to those of the Clover Bar Generating Station.

Table 7 Summary of conservative assessment EECs and quotients for effluents and cooling water discharges to rivers and marine environments

Enlarge image

Table 8 Summary of conservative assessment EECs and quotients for effluents to Lake Ontario ¹

Location	Approximate maximum EECs (mg/L) for distances (m) from source					Quotients (i.e., EEC ÷ ENEV) for distances (m) from source
	100	250	500	750	1000	100	250	500	750	1000
Ashbridges Bay WWTP
Perpendicular (to outlet) Parallel	0.076	0.049	0.034	0.026	0.022	14	9	6	5	4
	0.046	0.020	0.0077	0.0038	0.002	8	3	1	0.7	0.4
Toronto Humber WWTP
Perpendicular (to outlet) Parallel	0.044	0.028	0.019	0.015	0.012	8	5	3	3	2
	0.027	0.011	0.004	0.002	0.001	5	2	0.7	0.4	0.2

Although other discharges produced risk quotients greater than one, the above discharges were considered to be representative of discharges that may produce ecological risk to aquatic biota in Canada. The above list includes wastewater effluents, a WTP waste discharge and a cooling water discharge. Although no existing marine or estuarine discharge scenarios were recommended for Tier 3, this does not mean that new discharges to marine or estuarine environments could not produce negative ecological consequences. As shown in the toxicity assessment (Farrell and Wan, 2000), saltwater biota are very sensitive to residual oxidant exposures resulting from chloramine releases. In the event that a facility discharging chloramines were proposed, risk assessment methods such as those used in this PSL assessment should be used to ensure that discharges do not result in unacceptable ecological risk.

Conservative assessment of chloramine exposure from potable water

The purpose of the conservative assessment of chloramine exposure from potable water sources was to determine types of releases that may produce risk and to generate data that could be useful (if necessary) for risk management. A generic evaluation of drinking water was conducted using a simple dilution and decay model for streams having discharges ≤1.0 m³/s and mixing models for larger surface waters, including streams with discharges >1.0 m³/s, as well as lake and marine environments. This assessment was intended to serve as a management tool, with the focus being on identification of potential impacts resulting from discharges having different flow rates, chloramine concentrations and rates of decay. Models were used to estimate chloramine concentrations in water under various simplified discharge scenarios. Actual discharge events could not be modelled due to an absence of data pertaining to stream and drinking water flow rates, chloramine concentrations and in situ chloramine decay rates.

To simplify the assessment, chloraminated water releases were classified into the following flow rates: 0.0001, 0.001, 0.01, 0.1 and 1 m³/s. These represent a broad spectrum of discharges, from a small leak (0.0001 m³/s) to a very severe main break (1 m³/s). Also, data regarding chloramine concentrations, stream discharge and decay rates were selected to represent three geographically and climatically diverse case study locations: Abbotsford and Mission (coastal British Columbia), Edmonton (Prairie region) and Brantford (Great Lakes region). Rates of chloramine decay identified in the open literature and derived using water from each region were selected. This resulted in decay rate constants that represented high-demand (20 pe r day, half- life o f 0.03 days; Prairie region), medium-demand (4.97 per day, half-life of 0.14 days; Great Lakes region) and low-demand (0.734 per day, half-life of 0.94 days; coastal British Columbia) scenarios (Wisz et al., 1978; Milne, 1991; Norecol Environmental Consultants Ltd. and Dayton and Knight Ltd., 1993). For each region, high and low chloramine concentrations in their potable water supply reported by the three case study municipalities were used.

A small stream model was used to estimate EECs in streams with different discharge rates (0.001, 0.01, 0.1 and 1 m³/s). Mixing models derived by the University of Alberta were also used to determine chloramine fate and EECs in larger water bodies (Fraser River, North Saskatchewan River and Grand River) and in a generic lake and marine environment. EECs were compared with ENEVs for fresh water and salt water, and risk quotients were determined. No-effect values for short-term (8 hours, 0.015 mg/L) and long-term (17.9 hours, 0.0056 mg/L) chloramine exposures were compared with the EECs.

For small streams, results of the conservative assessment were summarized as dilution ratios relating to the approximate drinking water discharge rates that result in EECs that exceed the ENEV. Assuming direct discharge of potable water to a surface water, the results suggest that:

• In Mission and Abbotsford, short-duration (i.e., 8 hours) drinking water discharges containing 0.2 mg chloramine/L could impact on a small stream if diluted with surface water by a ratio of less than 1:10. A discharge with 1.020 mg chloramine/L could impact on a small stream if diluted with surface water by a ratio of less than 1:100 to 1:10. Long-duration discharges (≥17.9 hours) could impact on a small stream if diluted with surface water by a ratio of less than 1:100 to 1:10.
• In Edmonton and Brantford, Ontario, a short-duration discharge of 1.030-2.400 mg chloramine/L may produce impacts in a small stream when diluted with surface water by a ratio of less than 1:100. A long-term discharge with concentrations of 1.030-2.400 mg/L could produce impacts when diluted by a ratio of approximately less than 1:1000 to 1:100.

In reality, most discharges of chloraminated drinking water to surface waters would be indirect and would travel overland or through storm sewers prior to entering a surface water. This assessment suggests that pathways that have sufficiently high chemical decay and that are sufficiently long could decrease chloramine concentrations to levels that do not result in impacts to surface waters. A high-demand pathway may result from exposure to biological materials such as slimes and fungi and entrainment with high levels of suspended sediments containing various oxidizable organic substances. Soil infiltration and evaporation would influence losses en route to the surface waters. On the other hand, pathways not exposed to organic materials, without significant losses due to infiltration and evaporation, would not result in large chloramine losses. Drinking water releases occur from several sources, including distribution system leaks and breaks, lawn and garden watering, car and driveway washing, street cleaning, main flushing, fire fighting and relevant training, as well as industrial or commercial washdown and construction activities. These uses occur predominantly on land; hence, flows produced by these activities would usually incur some decay en route to a surface water.

The generic modelling indicated that a chloraminated flow of potable water from a typical garden hose could result in some marginal impacts if the discharge were direct to a very small stream and if decay was sufficiently slow. However, most flows of this nature are indirect. Environmental sampling conducted in the FVRD (i.e., Vanden Berg and Wade, 1997; Pasternak et al., 1998, 1999) could not detect measurable concentrations of chloramines in surface waters from indirect sources such as water used for lawn watering and equipment washdowns. These studies show that common indirect small flows with magnitudes of approximately ≤0.001 m³/s would not result in impacts to surface waters.

Larger flows with discharges of ≥0.01 m³/s, such as from large distribution system leaks, main breaks, fire hose discharges, main flushing, street washing and some industrial and commercial activities, may have a greater possibility of producing impacts. Direct discharges of approximately 0.01 m³/s and greater may potentially have an impact on small streams with discharges of approximately ≤0.1 m³/s in Abbotsford and Mission and streams with discharges of ≤1.0 m³/s in Edmonton and Brantford. Impacts may be greater near to the source of discharged water in Edmonton and Brantford, since chloramine concentrations in drinking water are reportedly higher in these municipalities than in Abbotsford and Mission. However, impacts may be localized in Edmonton and Brantford due to the presumed faster rate of chloramine decay. A summary of potable water flow rates that could impact on larger surface waters has been included in Table 9.

The Tier 2 assessment found that the greatest total number of accidental releases of chloraminated potable water occurred in the City of Edmonton (627 in 1996, 780 in 1995), followed by the City of Brantford (45 in 1996, 50 in 1995) and Mission/Abbotsford (7 in 1996 and 1997, 22 in 1994). In 1996, the regional water service area for Edmonton repaired one break per leak for approximately each 3.7 km of distribution main. In Brantford and Mission/Abbotsford, approximately one break per leak was repaired for each 8.5 and 45.7 km of distribution main, respectively, during 1996 (Pasternak, 2000). Although a larger number of drinking water releases occurred in Edmonton and Brantford, this does not necessarily mean that the greatest risk to aquatic biota occurs in these regions. The proximity and frequency of small streams with sensitive habitat and significant fish resources, destination of storm sewer discharge, as well as the number and magnitude of accidental chloramine releases are important factors affecting risk to aquatic biota.

Table 9 Summary of approximate drinking water discharges (m³/s) that may result in impacts to large surface waters ¹

Surface water	Drinking water discharges (m3/s)
	Short exposure2		Long exposure3
	Low chloramine concentration4	High chloramine concentration 5	Low chloramine concentration4	High chloramine concentration5
Fraser River	≥1	≥1	≥1	≥0.1
North Saskatchewan River	≥0.1	≥0.1	≥0.1	≥0.1
Grand River	≥0.01	≥0.01	≥0.01	≥0.001
Lake	≥0.1	≥0.1	≥0.1	≥0.01
Marine	≥1	≥1	≥1	≥1

In 1996, Edmonton and Brantford did not have a high topographical frequency of small streams, and a lower proportion of drinking water releases occurred in close proximity to their surface waters. Overall, the approximate average distance of a main break to a local surface water (named or unnamed) was approximately 180 m in Mission/Abbotsford and approximately 1200 m in both Edmonton and Brantford (Pasternak, 2000).

Data on the proportion of larger potable water discharges that flow to a surface water are not available; however, some indication of destination is suggested by the design of storm sewers. In Mission and Abbotsford, as well as in Brantford, storm sewers lead to local surface waters, ditches or retention areas. Sanitary sewers are not designed to capture surface water drainage in Brantford, Mission or Abbotsford (District of Mission, 1979; Eldridge, 1999). In Edmonton, approximately 75% of storm sewers flow to the North Saskatchewan River, while the remaining 25% are combined sanitary/storm sewers leading to the WWTP (Environment Canada, 1997b).

In Mission and Abbotsford, due to the abundance of streams in the area, overland and storm sewer pathways to these surface waters can be short. In other regions, the occurrence of small streams is less frequent; therefore, there are often greater distances to travel overland or via storm drains to surface waters. Particularly in the Prairie and Great Lakes regions, chloramine decay presented by long overland or storm sewer pathways and the high dilution resulting from the destination surface water may act to mitigate chloramine impacts.

Table 10 Characterization of main breaks and release of chloraminated potable water to Fergus Creek, British Columbia ¹

Date of spill	Description of main	Estimated volume released (m3)	Discharge rate (m3/s)	Duration(min)	Chloramine concentration (mg/L as TRC)2
17 October 1989	8" main	330	0.183	30	2.53
9 July 1990	aged concrete - asbestos	23 000	1.6-2.1	180-240	2.75

Other evidence of risk due to accidental releases of potable water

High ecological impact from accidental releases of potable water to small streams has been established. Two events, occurring on October 17, 1989, and July 9, 1990, in Surrey, British Columbia, resulted in devastating consequences to Fergus Creek (Table 10). Both events resulted in convictions under the Fisheries Act. Inorganic chloramine was identified as the culprit responsible for impacts (Nikl and Nikl, 1992). These main breaks occurred during a pilot study designed to determine the feasibility of chloramination for secondary treatment of drinking water in the Greater Vancouver Regional District.

Fergus Creek is a small stream for which discharge monitoring data do not exist; however, it has been estimated to have an average base flow rate of 0.065-0.130 m³/s. The stream is approximately 6 km long and flows from its headwaters in an agricultural area into the Little Campbell River. The lower 1.5 km of Fergus Creek support significant salmon habitat.

The first event released treated water that flowed approximately 1 km over lawns and through ditches and a storm drain to Fergus Creek and en route entrained large amounts of highly organic sediment. Eyewitnesses reported seeing fish attempting to leap out of the water, thereby suggesting a vigorous avoidance reaction by the fish. After the 30-minute spill, an estimated 1700-2000 juvenile coho salmon and lesser numbers of cutthroat trout (Oncorhynchus clarki) carcasses were observed along Fergus Creek downstream of the point of toxicant entry into the stream. Upstream of the point of entry, live juvenile coho salmon were abundant. No dead fish were observed in the Little Campbell River (Nikl and Nikl, 1992).

Flow from the second event travelled a short distance overland into the headwaters of Fergus Creek. The treated water then flowed for 4.5 km in Fergus Creek before entering the area of the stream containing fish habitat. The break resulted in an estimated 3000 fish carcasses of predominantly juvenile coho salmon. Throughout Fergus Creek, the stream bed was covered in dead stream insects and other invertebrates (Nikl and Nikl, 1992).

The results of the conservative assessment and evidence from actual events provide sufficient rationale to further the assessment of potable water to a probabilistic level. However, there is an absence of comprehensive data pertaining to potable water releases, particularly those that are accidental in nature. This absence seems to be due to their unpredictable nature.

All accounts of ecological impact resulting from chloramine-treated potable water releases are in the Lower Mainland area of British Columbia. There are no documented reports of impacts occurring from chloramine-treated potable water releases in any other region of Canada. The available information suggests that this is due to the high frequency of small streams with high water quality in the Lower Mainland of British Columbia.

3.1.2.2.3 Probabilistic assessment of chloramine exposure from effluents and cooling water

Receptors and assessment endpoints

The probabilistic risk assessment was focused on sensitive invertebrate and fish species commonly found in Canada. Sensitive receptors included the freshwater invertebrate, C. dubia, and a juvenile freshwater life stage of the anadromous fish, chinook salmo n. The chi nook salmo n w as chosen as a fish receptor, although this species was not found to be the most sensitive freshwater fish species, and in spite of the fact that it is not ubiquitous across Canada. However, it is related to other salmonid species, such as rainbow trout and coho salmon, which have a similar or greater sensitivity to chloramines, and together, salmonids are widely distributed throughout Canada (Scott and Crossman, 1973). As indicated in Section 2.4.2.3, coho salmon was found to be the most sensitive species to chloramine exposure (96-hour LC₅₀ = 0.07 mg/L).

To estimate risks of exposures of aquatic biota to chloramine, each exposure distribution was compared with three incipient lethality endpoints: 50% mortality to C. dubia (0.018 mg/L); and 50% (0.112 mg/L) and 20% (0.077 mg/L) mortality to chinook salmon. The probability that exposures exceed the endpoints was calculated. Each effect endpoint was modelled using time-series data so that the threshold concentration could be determined beyond which longer durations produced no additional mortality (incipient lethality) for C. dubia and little additional mortality (7-day LC₅₀) for chinook salmon. The details of this analysis are in Farrell and Wan (2000: Appendices B, C and H).

To bound risk estimates, months were selected that resulted in a high or a low risk to sensitive receptors. Selecting high- and low-risk months was difficult because aquatic biota are sensitive to chloramines at many life stages and acute lethality occurs at very low concentrations (Farrell and Wan, 2000: Appendix A). Each case study included specific months that were selected based on professional judgment, release patterns and organism life history. As will be discussed in the following paragraphs, limitations in the available hydrological and effluent data precluded an assessment for a portion of the selected months.

Probabilistic risk assessment of effluents and cooling water

McCullum et al. (2000) recommended that probabilistic risk assessments be conducted for chloramine releases from the Ashbridges Bay WWTP, North Toronto WWTP, Okotoks WWTP, Clover Bar Generating Station and Rossdale WTP. Although other discharges produced risk quotients greater than one, the above discharges were considered to be representative of discharges that may produce ecological risk to aquatic biota in Canada. The above list includes wastewater effluents, a WTP waste discharge and a cooling water discharge.

Inadequate hydrological data were available for a Tier 3 assessment of aquatic biota exposed to chloramines in the Sheep River near the Okotoks WWTP; hence, a probabilistic risk assessment could not be conducted for this discharge. However, a cursory analysis of the available data found that discharges from the Okotoks WWTP to the Sheep River and from the North Toronto WWTP to the Don River are similar. Both effluents are released to a small-sized and shallow river. However, the effluent from the North Toronto WWTP received less dilution than the discharge from the Okotoks WWTP; thus, one would expect a higher risk to biota from the Toronto facility. In 1996, discharge from the Okotoks WWTP contained an average TRC concentration of 3.0 mg/L and had an average flow rate of 3218 m³/d. The 20-year estimated mean (arithmetic, 1976-1995) flow rate for the Sheep River was 1 209 600 m³/d (Environment Canada, 1997b, 1999a). In 1998, Golder Associates Ltd. undertook sampling in the Sheep River and found measurable levels of CRC (mean 0.03 mg/L) close to the north bank up to 150 m downstream from the Okotoks WWTP (Golder Associates Ltd., 1998). These data are also described by Pasternak and Powell (2000). In 1996, discharge from the North Toronto WWTP contained an average TRC concentration of 1.6 mg/L and had an average flow rate of 33 528 m³/d. The estimated 20-year mean (arithmetic, 1976-1995) flow rate for the Don River was 226 400 m³/d (Environment Canada, 1997b, 1999a).

Discharges from the Rossdale WTP to the North Saskatchewan River are small (0.1-0.2 m³/s), intermittent and of short duration (typically 30 minutes). Such discharges require a "slug" model to estimate plume concentrations and plume duration. Although such models exist, their performance at predicting concentrations in the receiving environment was found to be poor. Furthermore, the risks from this source are likely to be much smaller than those from the Clover Bar Generating Station. The latter has larger discharges (21.2 m³/s) of longer duration (typically 4-24 hours) to the same river with approximately the same chlorine residual concentration in the effluent (0.3-1 mg/L). For these reasons, the Rossdale WTP case study was not pursued any further.

To characterize annual variation in exposure to inorganic chloramines, the Tier 3 assessment was conducted using effluent and hydrological data spanning 4 years. Longer time periods were not selected due to limitations in the available historic data and because many wastewater facilities have changed processes in recent years in an effort to improve effluent quality.

A river mixing model was used to predict downstream chloramine concentrations on a spatial grid in the North Saskatchewan River (and Don River case studies. A lake mixing model was used in the Lake Ontario case study. The equations for both models are described in McCullum et al. (2000).

The river mixing model assumes complete vertical mixing of the effluent plume and thus is a two-dimensional mass transport model. The assumption of instantaneous mixing may not be realistic under certain circumstances. For instance, layering may occur if the temperatures in effluent and receiving water are sufficiently different and if the discharge outlet does not result in effluent diffusion. In such circumstances, the model would underestimate risks due to chloramine exposure. The river channel is assumed to be rectangular, and distance coordinates are in dimensionless form for both the longitudinal and lateral directions. The model accounts for a channel of confined width, with concentration reflection occurring. Further, the model has been modified to take account of instantaneous chlorine demand and decay following release.

Parallel shore currents tend to dominate flow patterns in large lakes within a few kilometres of the shore. These currents have the capacity to transport and disperse effluents that have been discharged near the shoreline. For the Lake Ontario case study, we used a lake mixing model that assumed steady parallel shore currents and a continuous effluent source. This model also assumed a constant depth and near-instantaneous vertical mixing. As with the river mixing model, the lake mixing model was modified to account for instantaneous chlorine demand and chloramine decay following release.

In order to run the models, hydrometric, dispersion, chloramine decay and effluent data were gathered and summarized in a manner appropriate for use with the selected models for each of the case studies (see Moore et al., 2000).

Table 11 Input distributions and point estimates used in case studies for probabilistic risk assessment¹

Variable	Distribution	Month	Parameters
Clover Bar Generating Station - North Saskatchewan River
Discharge location	Point estimate	August	0.98
Initial chlorine demand	Point estimate	August	0.44
Velocity (m/s)	Point estimate	August	0.73
Effluent flow (m3/s)	Point estimate	August	21.2
Effluent concentration (mg/L)	Lognormal	August	Mean = 0.68, s = 0.10, reffluent:decay = -0.64
Stream flow (m3/s)	Lognormal	August	Mean = 208.1, s = 51.8
Transverse mixing coefficient (m2/s)	Lognormal	August	Mean = 0.09, 50%ile = 0.07, 90%ile = 0.34
Decay rate constant (/d)	Lognormal	August	Mean = 67.96, s = 62.88, reffluent:decay = -0.64
Depth (m)	Normal	August	Mean = 1.90, s = 0.30
Width (m)	Normal	August	Mean = 150, s = 10
Don River
Effluent concentration (mg/L)	Lognormal	January	Mean = 1.27, s = 0.36, reffluent:decay = -0.64
		March	Mean = 1.31, s = 0.40, reffluent:decay = -0.64
		August	Mean = 1.23, s = 0.40, reffluent:decay = -0.64
		October	Mean = 1.15, s = 0.46, reffluent:decay = -0.64
Effluent flow (m3/s)	Lognormal	January	Mean = 0.42, s = 0.01
		March	Mean = 0.42, s = 0.02
		August	Mean = 0.37, s = 0.08
		October	Mean = 0.40, s = 0.02
Stream flow (m3/s)	Lognormal	January	Mean = 5.51, s = 2.66
		March	Mean = 5.59, s = 1.47
		August	Mean = 3.55, s = 0.63
		October	Mean = 3.33, s = 1.67
Transverse mixing coefficient (m2/s)	Lognormal	January	Mean = 0.01, s = 0.01
		March	Mean = 0.01, s = 0.01
		August	Mean = 0.01, s = 0.01
		October	Mean = 0.01, s = 0.01
Decay rate constant (/d)	Lognormal	January	Mean = 2.64, s = 2.44, reffluent:decay = -0.64
		March	Mean = 2.64, s = 2.44, reffluent:decay = -0.64
		August	Mean = 67.96, s = 62.88, reffluent:decay = -0.64
		October	Mean = 35.30, s = 32.66, reffluent:decay = -0.64
Width (m)	Point estimate	January	20.6
	Point estimate	March	20.5
	Normal	August	Mean = 20.53, s = 0.64
	Normal	October	Mean = 20.33, s = 0.29
Velocity (m/s)	Point estimate	January	0.47
	Point estimate	March	0.33
	Lognormal	August	Mean = 0.22, s = 0.06
	Lognormal	October	Mean = 0.21, s = 0.02
Depth (m)	Point estimate	January	0.45
		March	0.45
		August	0.45
		October	0.45
Discharge location (0.5 centre, 0 & 1 bank)	Point estimate	January	0.8
		March	0.8
		August	0.8
		October	0.8
Initial chlorine demand	Point estimate	January	0.44
		March	0.44
		August	0.44
		October	0.44
Lake Ontario
Effluent concentration (mg/L)	Lognormal	January	Mean = 1.10, s = 0.21, reffluent:decay = -0.64
		April	Mean = 0.98, s = 0.24, reffluent:decay = -0.64
		July	Mean = 1.02, s = 0.19, reffluent:decay = -0.64
		October	Mean = 1.04, s = 0.22, reffluent:decay = -0.64
Effluent flow (m3/s)	Lognormal	January	Mean = 8.62, s = 0.84
		April	Mean = 8.14, s = 0.48
		July	Mean = 7.85, s = 0.39
		October	Mean = 7.72, s = 0.43
Decay rate constant (/d)	Lognormal	January	Mean = 35.30, s = 32.66, reffluent:decay = -0.64
		April	Mean = 35.30, s = 32.66, reffluent:decay = -0.64
		July	Mean = 35.30, s = 32.66, reffluent:decay = -0.64
		October	Mean = 35.30, s = 32.66, reffluent:decay = -0.64
Width of lake (m)	Point estimate	January	50 000
		April	50 000
		July	50 000
		October	50 000
Depth of mixing layer (m)	Point estimate	January	3
		April	3
		July	3
		October	3
Velocity along the shoreline (m/s)	Point estimate	January	0.12
		April	0.12
		July	0.12
		October	0.12
Longitudinal dispersion coefficient (m2/s)	Point estimate	January	50
		April	50
		July	50
		October	50
Lateral dispersion coefficient (m2/s)	Point estimate	January	0.2
		April	0.2
		July	0.2
		October	0.2
Initial chlorine demand	Point estimate	January	0.85
		April	0.85
		July	0.85
		October	0.85

Distributions of exposure were generated by Monte Carlo analysis using river and lake models (Table 11). The model equations are described by McCullum et al. (2000). The technique involved defining distributions to each variable used by the environmental fate models. The models used input values selected from each of the distributions. By running the models, output exposure values were generated. The Monte Carlo analysis involved repeating this process 10 000 times to generate a distribution of output values. For each case study site, separate analyses were conducted and exposure distributions were produced for locations on a grid near the effluent outfall. These analyses were repeated for different months for the Don River and Lake Ontario case studies. The Monte Carlo analysis was conducted using Crystal Ball, version 4.01. Using Monte Carlo simulation, Crystal Ball forecasts the range of possible results for a given situation, thereby allowing estimation of the likelihood of an event occurring. For complete details regarding the methodology used, consult Moore et al. (2000).

Don River: Probabilistic risk assessments were conducted for the Don River for the months of January, March, August and October. The results indicated that risks were greatest in January and lowest in August (Figure 8a-h). In January, risks to C. dubia are severe, with probabilities >80% for 50% or greater mortality for over half the width of the river at the greatest distance from the outfall modelled in this case study (1900 m) (Figure 8a,b). Farther downstream, the Don River discharges into Lake Ontario. By August, chloramine risks to C. dubia are much reduced, with probabilities <10% for 50% mortality or greater for the entire width of the river 1800 m from the outfall (Figure 8c,d). The risks in October are intermediate between those observed in January and August.

In January and March, chloramine risks to sensitive life stages of chinook salmon are moderate, with probabilities of up to 41% for 20% mortal ity 1900 m from the outfall (Figure 8e,f). By August, risks to salmon have decreased (Figure 8g,h), with probabilities of up to approximately 20% for 20% mortality 600 m from the outfall.

Sensitivity analyses from the January analyses indicated that the most important input variables influencing chloramine concentrations at the left bank (0 m, shore opposite from the discharge) were the transverse mixing coefficient initially (correlation coefficient or r > 0.5 up to 900 m from the outfall) and the effluent concentration (r > 0.5) and decay rate (r > -0.7) farther downstream.

The largest seasonal variation in input variables was observed with hydrological and decay parameters (see Table 12); thus, seasonal differences in risk may be attributed largely to these variables. The modelling exercise made use of discharge rates for the Don River, which varied from 3.3 m³/s for October to 5.6 m³/s for March, and decay rates ranging from 2.64 per day for January and March to 35.5 per day for October and 67.98 per day for August. If all model input variables were seasonally constant except for the river discharge rate, seasonal differences in EEC and risk would be proportional to variations in river discharge rates. The highest risk to aquatic biota would occur during the period of minimal dilution (i.e., during the summer and fall), and the lowest risk would occur during the months with the highest dilution (i.e., during the winter and spring). However, chloramine EEC decreases very rapidly during the months of August and October due to the estimated rapid rates of chloramine decay. The overall net effect of dilution and decay in this scenario is that risk is lowest during the summer months when decay is highest, in spite of the fact that discharge is lower than that experienced during other months of the year. The risk to aquatic biota is highest during winter and early spring when rates of decay are estimated to be very low, in spite of the fact that dilution of the effluent is highest at this time of the year.

Figure 8 Spatial distribution of risk for Ceriodaphnia dubia (Cd) and Oncorhynchus tshawytscha (Ot) exposed to inorganic chloramines discharged by the North Toronto WWTP to the Don River: (a) and (b) January - Cd; (c) and (d) August - Cd; (e) and (f) January - Ot; and (g) and (h) August - Ot (Moore et al., 2000)

Figure 8a Probability of 50% Mortality to Ceriodaphnia dubia

Figure 8b Probability of 50% Mortality to C. dubia

A limited number of samples were taken from the Don River in August-September 1998 to determine chloramine levels. The levels found 200 m (<0.005-0.013 mg/L) and 500 m (<0.005 mg/L) downstream along the shore closest to the outfall side were within factors of 5 and 2, respectively, of the 50th percentile concentrations predicted by the river model. Both values were within the output distributions predicted by the model. All other samples at these distances had levels below the analytical detection limit (0.005 mg/L), results that also correspond reasonably well with model predictions.

Lake Ontario (Ashbridges Bay): The risk analyses for Lake Ontario were done for the months of January, April, July and October. The results, however, indicated very little temporal variation in the estimated chloramine concentrations and hence risk. Risks to C. dubia and chinook salmon are highest by a marginal level in January. These January results are graphically depicted in Figure 9. This result is likely due to the lack of variation in the input variables for effluent concentration, effluent flow rate and decay rate. The results for all months indicate that the probabilities of 50% or greater mortality to C. dubia are fairly high (>40%) only in a narrow band (-250 m to +250 m) that runs parallel to the shoreline over a longitudinal distance of approximately 2000 m (Figure 9a,b). Probabilities of 20% or greater mortality to early life stages of chinook salmon are low even close to the outfall (Figure 9c,d).

Figure 8c Probability of 50% Mortality to Ceriodaphnia dubia

Figure 8d Probability of 50% Mortality to C. dubia

The lake model assumed complete and instantaneous mixing over the entire depth of the mixing layer (near the shore, this is assumed to equal the depth of the lake). Since the average depth of the lake in the vicinity of the discharge was determined to be approximately 3 m, this indicates the importance of rapid dilution as a factor affecting chloramine EECs.

Figure 8e Probability of 20% Mortality to O. tshawytscha

Figure 8f Probability of 20% Mortality to O. tshawytscha

Figure 8g Probability of 20% Mortality to O. tshawytscha

Figure 8h Probability of 20% Mortality to O. tshawytscha

North Saskatchewan River: The risk analyses of chloramines discharged from the Clover Bar Generating Station to the North Saskatchewan River were run only for the month of August, because of data limitations for other months of the year for stream flow, river width, river depth and cooling water discharge rates and concentrations. For August, we were able to obtain approximately 20 years of data for the stream flow, river width and river depth variables from the supporting document for the Priority Substances List assessment on ammonia (Environment Canada, 2000). Dilution is at its lowest in the North Saskatchewan River in August, thus making this time of year a potentially high-risk scenario.

The Clover Bar Generating Station releases chloramines intermittently for periods ranging from several hours to just over a day. Because incipient lethality for C. dubia is reached in approximately this time frame, we used a continuous-release model to simulate exposures, rather than a slug release model that we found performed very poorly.

Figure 9 Spatial distribution of risk for Ceriodaphnia dubia (Cd) and Oncorhynchus tshawytscha (Ot) exposed to inorganic chloramines discharged by Ashbridges Bay WWTP to Ashbridges Bay,
Lake Ontario: (a) and (b) January - Cd; (c) and (d) January - Ot (Moore et al., 2000)

Figure 9a Probability of 50% Mortality to Ceriodaphnia dubia

Figure 9b Probability of 50% Mortality to C. dubia

The analyses indicated that the plume from the Clover Bar Generating Station was narrow and remained close to the shoreline on the outfall side (Figure 10a-f). Except for a band approximately 45 m wide and 6500 m long, probabilities of 50% or greater mortality to C. dubia were <10% (Figure 10a,b). Risks were fairly high (>40% probability of 50% or greater mortality) in an area approximately 30 m wide and up to 3000 m downstream of the outfall. Risks to early life stages of chinook salmon were very low.

Figure 9c Probability of 20% Mortality to O. tshawytscha

Figure 9d Probability of 20% Mortality to O. tshawytscha

Conclusions

In the Don River, forecasted risks were most severe in January, with probabilities of >80% for 50% or greater mortality for C. dubia at 1900 m from the source. Lowest risk was forecasted for the month of August, with probabilities of up to 41% for 20% mortality at 1900 m from the outfall. For Lake Ontario in January, there was a probability of 2-68% for 50% mortality to C. dubia in a narrow, semi-elliptical band that was at least 500 m in width and extended approximately 6000 m. In July, the lowest risk was forecasted (range of 3-63% probability for 50% mortality in the zone 500 m in width and 4000 m in length). In the North Saskatchewan River, it appeared that elevated risk (i.e., >40% probability of 50% or greater mortality to C. dubia) was contained in a plume stretching to a maximum 30 m wide and approximately 3000 m long.

 

Table 12 Months identified for probabilistic assessment for each case study based on receptor life history and water body hydrology

Site	Invertebrates1		Chinooksalmon 2
	Low risk	High risk	Spring	Summer	Fall	Winter
Don River at Todmorden	August	October	March 3	June	October	January 4
Sheep River at Okotoks	June 3	October 4	March	June 3	October 4	no data
North Saskatchewan River at Edmonton	July 3	October	April	July 3	October	January 4
Lake Ontario at Ashbridges Bay	June	October	April	July	October	January

Since fish are less sensitive to chloramine than invertebrates, probabilities of risk to chinook salmon are lower than those for C. dubia. In the Don River, a 41% probability of 20% or greater mortality for chinook salmon at 1900 m from the source was forecasted for January. The forecasted risk dropped to its lowest in August (3% probability of 20% or greater mortality for chinook salmon at 1100 m from the source). For Lake Ontario, the highest risk was forecasted for January, at which time there was estimated to be a 3-40% probability of 20% or greater mortality to chinook salmon in a zone approximately 500 m wide and 3000 m long. In the North Saskatchewan River, a 52% probability of 20% or greater mortality to chinook salmon was forecasted in a narrow plume 1000 m from the source. This dropped to 4% probability of 20% or greater mortality to chinook salmon at approximately 4000 m from the source.

Although forecasted probabilities of risk for fish are lower than those for invertebrates, these may be important due to the longer period of time required for salmonids to regenerate. Therefore, lower probabilities of fish mortality may have population effects over the long term. Conversely, fish are mobile and have the ability to detect and avoid chloramine concentrations. Avoidance to chloramine has been reported at 0.05-0.11 mg/L for coho salmon and rainbow trout (Cherry et al., 1979). The avoidance effects may be offset by conditions in the effluent (e.g., elevated ammonia concentration and elevated water temperatures) that result in attraction. Data are not available to determine whether avoidance and/or attraction can affect the risk forecasts determined for this assessment.

Figure 10 Spatial distribution of risk for Ceriodaphnia dubia (Cd) and Oncorhynchus tshawytscha (Ot) exposed to inorganic chloramines discharged by Clover Bar Generating Station to the North
Saskatchewan River: (a) and (b) August - Cd; (c) to (f) August - Ot (Moore et al., 2000)

Figure 10a Probability of 50% Mortality to Ceriodaphnia dubia

Figure 10b Probability of 50% Mortality to C. dubia

The limited monitoring data for the Don River indicate that the river model may

have overpredicted chloramine concentrations somewhat. The same is likely true for Lake Ontario, although the comparison is made difficult by differences in orientation between the modelled scenario plume and the monitored plume. No monitoring data are available for the North Saskatchewan River.

3.1.2.3 Uncertainty

There a re several sources o f uncertainty in thi s assessment. Many of these uncertainties are founded in the reliance on existing data produced using traditional analytical methods. The existing methods have not permitted accurate differentiation between the chlorine species. Usually concentrations pertaining to inorganic chloramine are expressed as TRC or CRC, rather than monochloramine, dichloramine or trichloramine. In addition, the traditional methods have been prone to chemical interferences from various other chemical species, which result in false-positive measurements for various residual chlorine categories and congeners.

Figure 10c Probability of 20% Mortality to O. tshawytscha

Figure 10d Probability of 20% Mortality to O. tshawytscha

Figure 10e Probability of 50% Mortality to O. tshawytscha

Figure 10f Probability of 50% Mortality to O. tshawytscha

There is moderate uncertainty associated with production and loading estimates. Precise inorganic chloramine loading data are not available. Potable water, cooling water and wastewater treatment facilities typically measure chlorine residual as TRC and rarely conduct analyses for individual inorganic chloramine species. Therefore, in order to estimate exposure, it was necessary to assume that TRC concentrations were equal to inorganic chloramine concentrations. This may have resulted in overpredictions of inorganic chloramine concentrations in surface waters and conservative risk estimations.

The estimates for production and loading of chloramine from potable water include only those utilities intentionally producing chloramine for disinfection purposes. In fact, these values will be higher, since chloramines may be unintentionally formed due to the concurrent presence of ammonia in treated water. Also, inorganic chloramine may be formed in situ in surface waters when there is a release of FRC to a surface water containing sufficient concentrations of ammonia. It is not possible to quantify risks associated with in situ chloramine production. Although there are no data available characterizing risk from such sources, there is no reason to believe that the impacts resulting from such sources would be different from those presented in this report for inorganic chloramines.

There is moderate to high uncertainty associated with characterizations of inorganic chloramine decay. Several studies are available in the open literature that quantify rates of residual chlorine loss from the water column; however, few differentiate between chlorine species. Some studies report monochloramine or FRC decay as TRC only, which can be problematic, since there are differences in the decay rates of FRC, inorganic chloramines and organic chloramines. FRC species are generally more reactive than inorganic chloramines, and there are limited data regarding the decay of organochloramine species. Given its compositional uncertainty, decay rates expressed as TRC will not be precise indicators of inorganic chloramine decay without further speciation. Another limitation of the existing residual chlorine decay data is that most of the established analytical methods falsely measure several compounds as chlorine residual (Johnson, 1978; Milne, 1991; Harp, 1995) (see Section 2.1.2). This shortfall has particular implications for the probabilistic risk modelling that was conducted.

Sensitivity studies show that decay becomes more important in estimating environmental concentrations with distance from the source of input. Therefore, predictions in the far-field will have greater uncertainty than those in the near-field. The distributions and selection of decay data for risk modelling, however, were based on the literature and professional judgment rather than on site-specific data. Thus, there are some subjective uncertainties in the case studies that could not be accounted for in the model simulations. To address these uncertainties, site-specific studies would be required. In addition, future in situ chloramine decay studies should examine the loss of chloramine to suspended and bed sediments and chloramine's fate associated with these sediments (i.e., resultant transformation products, potential for release back to the water column). The toxicity of sediment-associated chloramines and reaction products requires evaluation.

There is moderate uncertainty in the dispersion coefficients used for the conservative and probabilistic modelling of effluents and cooling waters. Dispersion coefficients are highly site specific, and they were not available for all the surface waters subjected to modelling. The sensitivity analysis demonstrated that the transverse mixing coefficient has a strong influence on the width and downstream extent of the chloramine plume. The distribution for this variable was, however, based on literature and professional judgment rather than on site-specific data. To address these uncertainties, site-specific studies would be required.

Systematic monitoring surveys at the case study sites and other sites near chloramine sources would provide further data regarding chloramine exposure and would facilitate an analysis of risk using actual data. Together, monitoring programs and model development can complement each other, since monitoring data allow for model validation and calibration. Validated and calibrated models can then be used as an alternative or to support monitoring. Such extensive monitoring data were not available for this assessment.

The toxicity assessment indicated that coho salmon and rainbow trout may be more sensitive than chinook salmon. Therefore, the environmental risk for fish as presented in this report may have been higher if a more sensitive fish species had been selected as the receptor. On the other hand, the forecasted risk to fish may have been lower if a non-salmonid, such as bluegill (Lepomis macrochirus), had been chosen as the receptor.

In addition, the risk analysis did not involve the use of the entire concentration-response relationship; rather, the assessment was simplified by choosing 20% and 50% effect levels as endpoints. This was a practical consideration due to the number of exposure distributions (typically greater than 100) produced for each case study and the need to communicate risk in a direct manner. Toxicity data were also not adjusted for temperature. A discussion of temperature effects relevant to chloramine toxicity has been presented in a supporting document (see Farrell and Wan, 2000: Appendix A).

The approach taken in this assessment has been to examine risk to sensitive individual species and then to extrapolate effects to populations and ecosystems. This approach does not facilitate inferences regarding risk in the context of the community and does not describe indirect effects resulting via disruption of the food web. An evaluation of community-level risk would indicate the range of sensitivity that exists among individual species and would allow the ecological role of the more sensitive species to be better judged.

3.2 CEPA 1999 64(b): Environment upon which life depends

Although there are uncertainties regarding the fate of inorganic chloramines in the atmosphere, the available information shows that they would make a negligible contribution to tropospheric ozone formation and would not contribute at all to stratospheric ozone depletion.

3.3 Conclusions

CEPA 1999 64(a): Based on the available data, it has been concluded that inorganic chloramines in chlorinated effluents, cooling waters and treated potable water are entering the environment in a quantity or concentration or under conditions that have or may have an immediate or long-term harmful effect on the aquatic environment at various locations across Canada. Therefore, inorganic chloramines are considered to be "toxic" as defined in Paragraph 64(a) of CEPA 1999.

CEPA 1999 64(b): Based on the available data, it has been concluded that inorganic chloramines are not entering the environment in a quantit y or concentration or under conditions that constitute a da nger to the environment on which life depends. Therefore, inorganic chloramines are not considered to be "toxic" as defined in Paragraph 64(b) of CEPA 1999.

Overall conclusion:

Based on the critical assessment of relevant information, inorganic chloramines are considered to be "toxic" as defined in Section 64 of CEPA 1999.

3.4 Considerations for follow-up (further action)

The largest releasers of inorganic chloramines to the Canadian environment are municipal wastewater facilities, followed by potable and cooling water sources. Efforts to manage risk should involve limiting the exposure in surfacewaters from these sources. The chemistry of chloramines is extremely variable, and their persistence and fate will vary based on hydrological and climatic conditions, as well as water quality. Chloramine toxicity to freshwater and saltwater biota is also highly variable. As a result, reducing the exposure of aquatic biota to chloramines may involve an examination of regionally or locationally specific characteristics that affect chloramine risk. These would include decay, dilution and the presence of aquatic biota with sensitivity to inorganic chloramines.

The conservative-level assessment of chloramine-treated drinking water found that larger flows (e.g., from main breaks, punctures and large leaks, main flushing, and training of firefighters) destined for small streams with sensitive habitat can have devastating ecological consequences. However, low flow rate releases (e.g., from garden hoses or pinhole distribution system leaks) are unlikely to have any negative ecological consequences as long as the releases flow overland or via storm drains prior to entering a small stream.

Limiting exposure from unpredictable releases will prove most challenging. Reducing chloramine loading may be technologically feasible for point sources such as waste effluents or cooling waters, but not for geographically and temporally unpredictable releases from drinking water distribution systems. Regional-level control measures, potentially involving changes in treatment procedures, may have to be evaluated for regions with an abundance of aquatic environments that promote chloramine persistence, that provide low dilution and that contain sensitive aquatic ecosystems. Such measures must not compromise human health protection; selection of options must be based on optimization of treatment to ensure health protection, while minimizing or eliminating potential for harm to environmental organisms.

Although no existing marine or estuarine discharge scenarios were recommended for the probabilistic assessment, new discharges to marine and estuarine environments could produce negative ecological consequences. The marine environment contains aquatic organisms that are possibly even more sensitive to inorganic chloramines than freshwater species. Therefore, if a facility discharging chloramines to a marine environment is proposed, a precautionary risk assessment is recommended that evaluates site-specific characteristics that affect ecological risk.

Many of the input variables (e.g., effluent concentration, stream velocity) used in probabilistic modelling were based on limited data; hence, professional judgments were used to estimate distributions and to derive point estimates (e.g., transverse mixing coefficient and initial chlorine demand). Thus, there were several sources of uncertainty not accounted for in the analyses. Systematic monitoring studies at locations used as case studies for the probabilistic risk assessment and at other sites near chloramine sources would obviate the need for exposure modelling and allow for more confident predictions about risks of chloramines to aquatic biota in Canada. If it is determined in the risk management stage that comprehensive monitoring is required near chloramine sources, then consideration will need to be given to the standardization of a practical in-field sampling and analysis method that is able to distinguish between inorganic and organic chloramine species. If data (e.g., relevant to decay and transverse mixing) become available that may change risk forecasts, these should be considered, and the probabilistic assessment may need to be revisited.

Organic and inorganic chloramines are often found together; however, there are insufficient scientific data to allow an environmental risk assessment on the organic chloramine congeners to be conducted. In some instances, there may be a need to distinguish between inorganic and organic chloramine compounds for the purpose of risk management. As a unique group, existing analytical methods for organic chloramines would need improvement in order to allow measurement of individual congeners at the sub-microgram-per-litre level or to facilitate their measurement as a bulk parameter. Research may be required to determine the prevalence and fate of organic chloramines versus inorganic chloramines in waters receiving either chlorinated or chloraminated discharges. This could involve spatial and/or temporal considerations due to the disproportionation of the various types of chloramine species over time. Environmental fate models would require revision with this information. Also, the toxicological characteristics of organic chloramines may require scrutiny. This could be an extensive task given the wide range of organic nitrogen compounds known to be present in natural waters and the fact that only a fraction of them are present as easily measured small molecules. Simple toxicological testing protocols for estimating the toxicity of mixtures of organic chloramines (or organic and inorganic chloramines) may require development, since it will not likely be practical or possible to routinely measure the many possible organic chloramine species that are likely to be present in these waters.