Priority Substances List Assessment Report for Ammonia in the Aquatic Environment

3.0 Assessment of "Toxic" Under CEPA 1999 (Continued)

3.1 CEPA 1999 64(a): Environment (Continued)

3.1.2 Environmental risk characterizatio (Continued)

3.1.2.3 Probabilistic risk assessment

Ecological risk assessment for ammonia using the conservative quotient method indicated that sewage effluents are a major source of toxicity to aquatic habitats. The results identified three case studies where a probabilistic risk assessment could be conducted due to the relative completeness of the data and the likelihood of negative impacts. These are:

1. Hamilton Harbour,
2. the North Saskatchewan River downstream of the Gold Bar municipal WWTP, and
3. the Red River downstream of Winnipeg's three municipal WWTPs.

These case studies are fairly typical of municipal wastewater discharges in Canada. Two are located on fairly large, yet slow rivers supporting large urban populations, while the other is a lake discharge situation with minimal water exchange and intensive urban development in the surrounding watershed. Considering that no two municipal wastewater discharge situations are the same, these case studies should provide a good, generic probabilistic risk assessment for this source of ammonia to fresh water.

3.1.2.3.1 Hamilton Harbour

Extensive sampling was done in 1998 (Charlton and Milne, 1999) to complement measurements taken routinely at a station in the centre of the harbour. The crosses on Figure 12 show the sewage outfall locations in Hamilton Harbour.

Figure 12 Hamilton Harbour

Figure 13 illustrates seasonal fluctuations in ammonia concentrations at the central station in Hamilton Harbour for the years 1986-1999. Weekly grab samples were collected for these years. For the last 6 years, the peak ammonia concentration found in early spring has been increasing. The rapid decrease of ammonia in late spring and summer is caused by nitrifying bacteria that produce nitrate from total ammonia. The nitrifying process is greatly reduced in the winter due to the sensitivity of the bacterial community to low temperatures. As a result, ammonia accumulates in the water during the winter.

As the nitrification process begins to consume ammonia in the spring due to warming temperatures, the proportion of un-ionized ammonia increases for the same reason. At the same time, increasing algal growth withdraws carbon dioxide from the water, and this causes the pH to rise, which also causes the proportion of un-ionized ammonia to increase. The net result is that the timing of the temperature and pH cycle produces increasing concentrations of un-ionized ammonia, even though the total concentration is decreasing in spring (Charlton and Milne, 1999).

The peak ammonia concentration therefore occurs in the spring and depends on the loading rate of ammonia (Rodgers et al., 1992). However, in 1997-1998, several unusual occurrences generated lower than usual ammonia concentrations in the harbour (the concentration was lower by about 0.4 mg/L). The winter of 1998 was unusually mild and the spring was relatively warm, resulting in slightly higher bacterial nitrification, which degraded some of the ammonia. The load from the Woodward sewage treatment plant, which is the main source of ammonia to the harbour, was lower in the fall and winter of 1997-1998. In addition, the ammonia load from the Burlington Skyway sewage plant was lower than normal the previous winter during interruption of operations at a local food processing plant. Thus, both higher temperature and lower ammonia load explain the lower than usual ammonia concentrations in 1998 (Charlton and Milne, 1999) (Figure 13).

Un-ionized ammonia concentrations were usually much higher in the Windermere Arm than elsewhere in the harbour (see Figure 12 for location). The Windermere Arm receives water discharging from the Windermere Basin at the southeast end of the harbour. The Woodward Avenue sewage treatment plant discharges into Redhill Creek near the upstream opening into Windermere Basin. The outflow from the basin comprises mostly treated sewage, along with creek water and combined sewer overflows. The sewage treatment plant is the major source of ammonia in the basin outflow (Charlton and Milne, 1999).

Figure 13 Seasonal fluctuations in ammonia at the central station in Hamilton Harbour

Figure 14 Total and un-ionized ammonia concentrations at the central station in Hamilton Harbour, 1-m depth

The data at the central station represented the overall condition of the harbour reasonably well. Even though the central station data were not always the same as the harbour mean, they lay partway between the extremes of the data. The main ammonia loads occur in shallow water near shore, so that those waters will generally have more severe ammonia conditions than will the centre of the harbour. In the centre, the harbour is about 25 m deep at the deepest point. During the spring and summer, the surface water warms, but the rate of warming is less in the lower water, causing a bilayer system to form, because the cool, lower layer is denser. Analyses of water samples collected at 1, 3, 5, 7 and 19 m at the central station show that, until mid-May, ammonia concentrations at all the depths were similar. With the warming in May, concentrations began to decline, but the lower, cooler depths declined the fastest. This seems paradoxical, because the nitrification rate is dependent on temperature. The cause of the rapidly declining ammonia concentrations in the bottom layer is a deep-water flow that brings lake water into the bottom of the harbour via the shipping canal. This dilutes and displaces some of the ammonia accumulated during winter. The bottom water at this time is a potential refuge for fish from high un-ionized ammonia concentrations. The relatively low temperature in the bottom water also favours the ionized form of ammonia. By the end of June, however, the dissolved oxygen in the bottom water is near zero, and this excludes fish and most other higher organisms. Thus, there is no real escape from high un-ionized ammonia in the surface water (Charlton and Milne, 1999).

Exposure concentrations for ammonia were developed from the collection and analysis of water samples from Hamilton Harbour. Two sampling locations were selected: (1) Windermere Arm, where ammonia concentrations are typically the highest in the harbour, and (2) a central station where ammonia concentrations are representative of the overall conditions in the harbour (Figure 12).

Sixty-eight samples were collected from Windermere Arm between March 31 and August 31, 1998 (Charlton and Milne, 1999). Concentrations of un-ionized ammonia ranged from 0.003 to 0.63 mg/L. At the central station, 21 samples were collected at a depth of 1 m between January and September 17, 1998. Concentrations of un-ionized ammonia are represented in Figure 14; concentrations ranged from ≤0.01 to 0.11 mg/L.

Due to the relatively high ammonia concentrations and length of exposures to ammonia in Hamilton Harbour, ecological risks were determined in three ways. The short-term acute CTV was used for a risk assessment of rainbow trout (O. mykiss) passing through Windermere Arm over a short period of time; the acute CTV was used to assess lethality risk to organisms in the harbour exposed to intermittent elevated concentrations. The chronic CTV was used to assess the risk in the harbour from exposure to long-term average concentrations. Craig (1999) analysed published trout toxicity data for short-term exposure to high ammonia concentrations. For un-ionized ammonia, the LC₅₀ for a 12-hour exposure was 0.74 mg/L, and the LC₁₀ was 0.074 mg/L.

Craig (1999) also analysed ammonia LC₅₀ data using the Water Environment Research Foundation (WERF, 1996) methodology for logistic regression analysis of community toxicity data. A concentration of 0.29 mg NH₃/L would, on average, theoretically produce 50% mortality in the most sensitive organisms representing the 5th percentile of the aquatic community. This value is approximately equal to the lowest acute effect level for freshwater species that was found in the published literature.

Ammonia sublethal toxicity data were analysed using the WERF regression analysis approach. At un-ionized ammonia concentrations above 0.041 mg/L, 5% of the species in an exposed community would exhibit a 20% reduction in growth or reproduction. This value is also just below the lowest reported chronic effect level for freshwater species (Table 6).

Figure 15 presents the risk analysis for ammonia in Windermere Arm, Hamilton Harbour. Of the weekly ammonia samples taken from Windermere Arm in 1998, there was a slight chance (<4%) that un-ionized ammonia concentrations could exceed the acute CTV of 0.29 mg/L (96-hour LC₅₀). The ammonia concentrations never reached that high a level in the rest of the harbour. Eighteen percent of samples contained un-ionized ammonia concentrations exceeding the short-term acute CTV for trout (12-hour LC₁₀) of 0.074 mg/L. Forty-five percent exceeded the chronic CTV (EC₂₀ growth/reproduction) of 0.041 mg/L (Figure 15).

This analysis assumes that weekly samples approximate 96-hour average ammonia concentrations and that the logarithmic trend line in Figure 15 approximates the actual percentile of time that fish are exposed up to a concentration (R² = 0.89). It indicates that in Windermere Arm, there is little probability that ammonia concentrations would be lethal to 50% of sensitive fish species that remained in the Arm for 96 hours. However, 30% of the time, ammonia concentrations would be expected to cause 10% mortality in a population of rainbow trout passing through the Arm over a 12-hour period. Forty-five percent of the time, the concentrations of unionized ammonia in the Arm would be expected to cause a 20% reduction in growth or reproduction in the most sensitive organisms present in the Arm for an extended period of time.

Because the conditions at the central sampling site in Hamilton Harbour were sufficiently similar to those of other stations around the harbour (Charlton and Milne, 1999), this site was used to estimate effects from ammonia in the rest of the harbour. Again, the analysis assumes that sampling times approximate the test exposures and that the logarithmic trend line in Figure 16 approximates the actual percentile of time that fish are exposed up to a concentration (R² = 0.95). The conditions in the rest of Hamilton Harbour were not as severe as in Windermere Arm. None of the samples reached the 96-hour acute CTV of 0.29 mg/L. However, Figure 16 illustrates that 8% of samples contained unionized ammonia concentrations exceeding the short-term acute CTV for trout (LC₁₀, 0.074 mg/L) and 36% exceeded the chronic CTV (EC₂₀ , 0.041 mg/L).

This means that in Hamilton Harbour, 8% of the time un-ionized ammonia concentrations would be expected to cause 10% mortality in a population of rainbow trout resident for at least 12 hours and 36% of the time un-ionized ammonia in the harbour would cause a 20% reduction in growth or reproduction of the most sensitive group of species in the harbour.

In summary, in 1998, many sites in Hamilton Harbour had concentrations of un-ionized ammonia that were up to 0.11 mg/L. Un-ionized ammonia concentrations in other areas were above concentrations that are generally safe for aquatic organisms. The ammonia concentrations in 1998 were unusually low. The central harbour site is reasonably similar to the rest of the harbour and, if anything, underestimates ammonia concentrations in shallow-water areas (Charlton and Milne, 1999).

It is concluded that un-ionized ammonia concentrations in the central station of Hamilton Harbour and in Windermere Arm are sufficiently high to cause significant adverse sublethal effects on sensitive organisms that could normally be expected to inhabit these areas. For short periods, the concentrations of un-ionized ammonia can be expected to be acutely lethal to a portion of the rainbow trout population in the harbour.

Figure 15 Un-ionized ammonia concentrations in Windermere Arm, Hamilton Harbour

Figure 16 Risk curve for un-ionized ammonia at the central station in Hamilton Harbour

3.1.2.3.2 North Saskatchewan River

Direct measurements of ammonia in the North Saskatchewan River showed that there is a zone of potential toxicity generated in the river downstream from the City of Edmonton's sewage effluents. The data for this study were collected at only one period of time (September 1993), although they were collected over a large reach of the river. While useful for a validation for the CORMIX plume dispersion modelling program, this study could not be used to adequately determine the risk to aquatic organisms.

Figure 17 Cumulative density function of NH₃ concentrations for August at 1 km along the centre of the plume in the North Saskatchewan River

An evaluation of the field monitoring data for the North Saskatchewan River revealed that the data were insufficient to enable development of an exposure cumulative density function (CDF). As an alternative, the CORMIX model was used to estimate levels of un-ionized ammonia at various distances downstream of the WWTP. Initial analyses with CORMIX indicated that ammonia levels in the North Saskatchewan River are typically highest in August. Therefore, the assessment was focused on this month to estimate exposure and risks. Because CORMIX is not a distributional model, exposure CDFs were developed. The steps are described in Appendix D. Figure 17 is an example of an exposure CDF for the North Saskatchewan River 1 km downstream of the Gold Bar WWTP. It represents the probability that a specific ammonia concentration will be exceeded. Reading the ammonia concentration from the bottom of the graph, for example, we can see that there is a 28% chance that ammonia concentrations will be greater than 0.02 mg/L 1 km from the outfall.

For each distance downstream of the North Saskatchewan River treatment plant, a risk curve was derived by combining the exposure CDF and the concentration-response relationship for percentage of biota adversely affected (as derived in the effects characterization). This was done by calculating the un-ionized ammonia concentration that caused effects ranging from 1 to 99% of species affected in 1% increments. Each effect concentration was then compared with the appropriate CDF for exposure to determine the proportion of the exposure values that exceeded the effects concentration. For 0% effect, 100% of the exposure values were greater; hence, each risk curve starts at 100% on the left-hand axis. Risk curves were generated for the centre-line plume 1, 2, 5, 10, 15 and 20 km downstream of the WWTP.

The risk curves are illustrated in Figures 18 to 23 and indicate that impact to aquatic organisms in the North Saskatchewan River decreases downstream of the outfall, as would be expected. For example, Figure 18 shows that at 1 km downstream, there is a 92.2% probability of at least 5% of the species exhibiting a 20% inhibition in growth or reproduction. However, at 20 km downstream, Figure 23 shows that there is a 9.8% probability of the same impact.

Figures 18-23 Risk curves for 1-20 km downstream of the wastewater treatment plant on the North Saskatchewan River

Enlarge image

Table 12 is a listing of probabilities of an effect (20% or greater reduction in growth/reproduction) to varying proportions of the aquatic community for each distance downstream of the WWTP for the month of August. This table shows that there is a decreasing probability of the defined toxic impact as more species are considered. For example, at 5 km, there is a 42% probability of an impact on 5% of the species, and there is a 0% probability of an impact on 25% of the species.

Table 12 Probability of species affected 1-20 km downstream of the wastewater treatment plant on the North Saskatchewan River in August

Distance from WWTP (km)	Prob. of 5% of species affected	Prob. of 10% of species affected	Prob. of 15% of species affected	Prob. of 20% of species affected	Prob. of 25% of species affected	Prob. of 50% of species affected
1	92	73	54	34	29	4
2	73	42	24	15	9	0
5	42	17	7	4	0	0
10	24	7	3	0	0	0
15	14	3	0	0	0	0
20	10	0	0	0	0	0

Figure 24 is a graphical representation of the impact gradient in the plume for 5% of species.

This analysis is, however, a conservative estimate of risk to aquatic biota in the North Saskatchewan River for the month of greatest impact. The uncertainty in quantifying the risk from ammonia is discussed in Section 3.1.2.4.

The accuracy of CORMIX predictions when compared with field measurements is discussed in the supporting document (Environment Canada, 2000). Validation results showed that the greatest difference between the model and measured values lies within the first kilometre of the plume. From the discharge to 1000 m, the accuracy of predictions varies from 95% to 40% of measured, with no particular trend. The prediction improves with distance downstream; it is 98% accurate at 5300 m. Although the near-field zone is of less interest than the far-field zone for the risk assessment on the North Saskatchewan River, it is important to know that CORMIX significantly underestimates ammonia concentration in the first 1000 m.

From this analysis, there is a significant likelihood of an ecological impact from the release of ammonia in sewage effluent from Edmonton's sewage system in the summer months. In this Assessment Report, the chronic impact is a 20% decrease in the rate of growth or inhibition of reproduction. Significance can be quantified in the matrix of probabilities of impacts with distance downstream of the outfall (Table12), such that we can predict that there is a 42% probability of a chronic impact on 5% or more of freshwater species in the North Saskatchewan River at 5 km below the outfall. This can be extended to predicting a 10% probability of this impact in a plume that is 20 km long by 80 m wide. The potential impacts are not likely to be as great in any other month, with very little impact expected from November to May.

The City of Edmonton is committed to reducing the ammonia concentration in its effluent to 5 mg/L in summer and 10 mg/L in winter by the year 2005 (Sawatzky, 1999).

3.1.2.3.3 Red River

The screening procedure for high ammonia concentrations in national rivers identified the Red River downstream of Winnipeg as a possibly impacted river (refer to the technical supporting document [Environment Canada, 2000]). Of the samples analysed by the province of Manitoba from 1988 to 1997, 27% of them exceeded the screening criteria of 0.02 mg un-ionized NH₃/L. This site was on the Red River at Selkirk, downstream of the Lockport Dam. The City of Winnipeg monitors water quality at the dam; 10% of samples were in excess of the chronic CTV of 0.04 mg un-ionized ammonia/L. These numbers warranted a detailed review of the data from the City of Winnipeg (data were provided by R. Ross, City of Winnipeg Water Pollution Control Centre, Chemistry Laboratory).

Figure 24 Probability of impacts in the North Saskatchewan River

Winnipeg operates three sewage treatment plants: two on the Red River, called the South End Water Pollution Control Centre (SEWPCC) and the North End Water Pollution Control Centre (NEWPCC), and one on the Assiniboine River, called the West End Water Pollution Control Centre (WEWPCC). The CORMIX screening exercise identified the North End and West End plants as being potentially problematic (Figure 25).

The West End plant is the smallest of the three plants and discharges secondary-treated effluent to the Assiniboine River near the western boundary of the city. The South End plant is located in the southern half of the city (St. Vital) on the Red River. It handles the second largest volume of sewage.

The North End facility is the City of Winnipeg's largest-capacity plant, discharging to the Red River about 24 km downstream of the confluence with the Assiniboine River, near the northern boundary. Lockport Dam 20 km downstream backs up the water through Winnipeg, so that it has an average width of 175 m and an average depth of 3.5 m at low flow. As the river is relatively deep with slow currents, this leads to slow vertical mixing but rapid horizontal mixing. The Red River provides relatively low rates of dilution at full mixing, ranging from about 11:1 to 69:1 at low and average river flows, respectively.

Figure 25 Winnipeg sewage treatment plants and sample sites

The City of Winnipeg monitors ammonia concentrations at four locations: (1) on the Assiniboine River at Winnipeg's Main Street Bridge, (2) on the Red River at Lockport Dam north of the city, (3) at Perimeter Bridge just north of the North End plant, and (4) at Winnipeg's Fort Garry Bridge downstream of the South End plant.

As shown in Figure 26, total ammonia concentrations have varied considerably at Lockport Dam, although the ammonia concentrations from the three WWTPs have not changed appreciably in the past 20 years. The City of Winnipeg did not change its sewage treatment processes during the period of data presented (1986-1997) in Figure 26 (Ross, 1998). A comparison of river flow data for the Red River with the total ammonia concentration at Lockport shows a weak negative correlation (−0.34).

There are some distinct patterns identifiable for flow rates and total ammonia concentrations. The predominant ones are the periodic fluctuation in both flow rates and total ammonia concentrations. Flows peak in the spring, while total ammonia peaks in late summer General trends can be observed between high ammonia concentrations and low flows for the years 1988-1991 and between low ammonia concentrations and high flows, exemplified by the years 1995-1997.

Figure 26 Red River monthly mean flows and total ammonia concentrations

The highest ammonia levels in the Red River occur from August to November. Therefore, this time frame was chosen for estimating risks to aquatic biota. Because the concern is for chronic effects, the appropriate temporal scale for estimating exposures is monthly (many chronic toxicity tests in fish are close to 1 month in length). In a chronic exposure scenario, high exposures on some days tend to be balanced out by low exposures on other days, such that overall exposure tends towards some measure of centrality. The appropriate measure of centrality in the case of ammonia concentrations in water is a geometric mean, because the underlying distribution for concentrations of contaminants in the environment is typically lognormal (Ott, 1995). Developing CDFs (probability of being in an exposure range versus ammonia concentration) for ammonia exposure also requires a measure of dispersion about the geometric mean of the ammonia concentration. Because we are concerned with chronic exposures, the measure of dispersion should not be used to estimate day-to-day variation. Rather, the dispersion measure should be used to account for year-to-year variability of the monthly mean. As the wastewater treatment practices and receiving environment conditions had not been significantly altered in the last decade, the long-term monitoring data were used to make predictions about possible exposures in the future. That is, the variation in monthly geometric means in the past 11 years can be used to estimate expected variation in the future. Therefore, for each sampling station and each month, exposure CDFs were developed as described in Appendix E.

At Lockport Dam, approximately 20 km north of the city (Figure 25), the un-ionized ammonia concentrations exceeded the CTV for most of the months from July to January for the period 1986-1993 (Figure 27, grab samples taken weekly). This is not the pattern seen at the North Perimeter station upstream. At Lockport Dam, the periods when the CTV is exceeded occur in both low and high water flow periods. The months of high concentrations extend from July occasionally to January. This may have something to do with the effect of the dam on flows in this area of the Red River.

From these data, it appears that a major factor in high ammonia levels downstream of Winnipeg is the flow rate of the Red River. The Lockport Dam creates a still-water area for some 20 km back past Winnipeg. The combination of moderately high pH (usually above 8) and warm temperatures in the Red River drives the unionized ammonia concentrations above 0.04 mg/L. The City of Winnipeg provided sufficient information from its water quality monitoring program to conduct an analysis to determine the probability of impacts.

For each of the four sampling stations, and for each month under consideration, a risk curve was derived by combining the exposure CDF (as derived from exposure characterization) and the concentration-response curve (as derived from effects characterization). This was done by calculating the un-ionized ammonia concentration that caused effects ranging from 1% to 99% of species affected in 1% increments. Each effect concentration was then compared with the appropriate CDF for exposure in order to determine the proportion of the exposure values that exceeded the effects concentration. For 0% effect or more, 100% of the exposure values were greater; hence, each risk curve starts at 100% on the left-hand axis. The risk curves are presented in Figures 28-43. A good point of comparison between the risk curves is the probability that 5% or more of the species will be affected by the ammonia in the river. For example, in August at the Lockport Dam site, there is a 24.4% probability of at least 5% of the species exhibiting a 20% inhibition in growth or reproduction.

Table 13 lists the probabilities of an effect (20% reduction in growth/reproduction) on varying proportions of the aquatic community for each month at the four monitoring sites in the Red and Assiniboine rivers.

Figure 27 Un-ionized ammonia concentrations in the Red River at Lockport Dam, north of Winnipeg

Enlarge image

From this analysis, there is a significant likelihood of an ecological impact from the release of ammonia in sewage effluent from Winnipeg's sewage system. In this assessment, the "toxic impact" is defined as a 20% decrease in the rate of growth or inhibition of reproduction. Significance can be quantified in that probabilities of impacts to the most sensitive 5% or more of freshwater species in the Red River are from 10% to 31%, depending on the month of exposure and the location. Probabilities of the same degree of impact to the most sensitive 10% or more of species range from 5% to 21%, depending on the month and location. The stretch of river encompassed by this study is roughly 30 km in length.

Significant impacts on freshwater biota in the Assiniboine River, roughly 20 km downstream of the discharge, are not likely, although some degree of impact might be expected at this site in November Impacts upstream of the bridge could not be quantified.

The City of Winnipeg is currently conducting a site-specific ecotoxicological assessment of its municipal effluents.

3.1.2.4 Discussion of uncertainty

Uncertainty analyses seek to describe and interpret lack of knowledge that may be present in the implementation or interpretation of a risk analysis. The goal of uncertainty analyses is to provide the risk manager with the most complete information available on the expected outcomes of exposures. In risk analysis, scientific uncertainty derives from many sources, including inadequate scientific knowledge, natural variability, measurement error, sampling error and incorrect assumptions. Uncertainty can also arise from model mis-specification, including errors in statistics, parameters and initial conditions and failure to appropriately capture expert judgement (SETAC, 1997).

There are several major sources of uncertainty associated with the environmental risk assessment of ammonia. The principal source of uncertainty is the estimation of a chronic CTV at the low end of the toxicity scale. In this case, it was estimated to be 0.041 mg/L, with 95% prediction limits of 0.02-0.06 mg/L, which is just below the lowest measured EC₂₀ estimated from published toxicity studies. Other major sources of uncertainty are the period of actual exposure in fish, the application of a generic assessment to specific situations, the lack of recent ambient concentration data in most Canadian media, and the potential confounding toxicity from other components of sewage effluents.

Table 13 Probabilities of impacts at four sites around Winnipeg

Month	Probability of 5% of species affected	Probability of 10% of species affected	Probability of 15% of species affected	Probability of 20% of species affected	Probability of 25% of species affected
Red River -- Fort Garry Bridge
August	19	11	7	5	4
September	21	13	9	7	5
October	10	5	3	2	0
November	18	11	8	7	5
Red River -- North Perimeter Bridge
August	31	21	16	13	11
September	25	15	11	8	6
October	16	10	7	5	4
November	19	12	8	6	5
Red River -- Lockport Dam
August	24	14	10	7	5
September	21	10	6	4	3
October	18	11	7	5	4
November	20	11	7	5	4
Assiniboine River -- Main Street Bridge
August	0	0	0	0	0
September	0	0	0	0	0
October	0	0	0	0	0
November	11	6	4	3	3

Regarding environmental exposure, there could be concentrations of ammonia in Canada that are higher than those identified and used in this assessment. Limited data were available for ammonia levels in air where the largest Canadian releases occur. For example, ammonia deposition rates at present across Canada are relatively low, although they can be very high in certain locales, typically associated with intensive livestock operations (i.e., the Lower Fraser Valley).

Of the three case studies used for WWTPs, two, Hamilton Harbour and Winnipeg, had well-documented water quality monitoring studies, and Edmonton had an intense, short-term study to determine ammonia dispersion in the plume that enabled modelling of the plume under variable conditions.

Analysis of ammonia released from pore water by the disposal of dredged sediment suggests that marine species might be adversely affected. However, this analysis should be viewed with caution due to the paucity of data with respect to marine species.

Regarding effects of ammonia on aquatic and terrestrial organisms, uncertainty inevitably surrounds the extrapolation from available toxicity data to potential ecosystem effects. The ammonia assessment is based on a few well-done freshwater field studies, modelling and extrapolation from laboratory toxicity work. The relatively small number of organisms that can be routinely cultured and tested in laboratory toxicity studies leads to this uncertainty when extrapolating these toxicity results to responses of natural populations. That said, the Europeans have documented ecological changes in their sensitive ecosystems likely as a result of the atmospheric deposition of ammonia. Canadian ecosystems will likely respond in similar ways.

Figures 28-31 Risk curves for Fort Garry, Red River

Enlarge image

To account for some of these uncertainties, conservative application factors were used as appropriate in the environmental risk analysis to derive ENEVs. An application factor is useful when few toxicity data are available and is, in general, environmentally protective, as it is a conservative approach. In addition, when there are many sources of uncertainty (e.g., sources of uncertainty in toxicity testing or exposure concentrations), application factors provide a relatively easy way to aggregate the multiple sources of uncertainty. In these cases, complicated statistical analysis may be impractical and costly.

Figures 32-35 Risk curves for Perimeter Bridge, Red River

Enlarge image

The toxicity of ammonia to warm-water species is limited; however, the database for toxicity of ammonia to cold-water species is good. The problem of determining actual periods of exposure is difficult due to the mobility of many fish species in rivers. Toxicity estimation in the North Saskatchewan River is problematic due to the unrestricted nature of the river and the mobility of fish. However, benthic surveys above and below the outfall support the conclusion of toxicity from the sewage effluent, although not necessarily from ammonia. In the Red River, there is limited travel for fish below the Lockport Dam, and Hamilton Harbour has a greatly restricted water flow, with Lake Ontario restricting fish travel. Thus, the likelihood of overestimating risk to species critical to the structure and function of the community or ecosystem was judged to be acceptable.

The use of a generic group of aquatic species that are native to much of Canada may still present a source of uncertainty, for this group of species is not resident in all sewage discharge situations across Canada. Some of the species used were not commonly found in each of the three case studies, although they were all potentially resident. The use of site-specific assemblages of species would eliminate this uncertainty, although toxicity information is sparse for many species commonly found across Canada. A site-specific assemblage of species would not necessarily be less sensitive than the generic one chosen; some local species, mountain whitefish (P. williamsoni), for example, are more sensitive than rainbow trout (O. mykiss) to ammonia.

Figures 36-39 Risk curves for Lockport Dam, Red River

Enlarge image

The major reviews of data conclude that pH has a larger effect than temperature on acute toxicity (WHO, 1986; U.S. EPA, 1998). There are no proven correlations between temperature and chronic toxicity due to the paucity of data. The effect on this assessment of assuming that toxicity increases, rather than decreases, with temperature is to make the late summer/early fall months the critical periods, whereas the risk from toxicity may be more spread out from spring to fall. In this case, this assessment would underestimate toxicity from ammonia in some months and overestimate it in others.

The probabilistic approach used for the case studies allows for a quantitative estimation of risk (analysed as a distribution of effects and exposure concentrations) and therefore incorporates many of the uncertainties associated with the effects and exposure characterization discussed above. The largest source of uncertainty lies in the estimate of the CTV.

Figures 40-43 Risk curves at Main Street Bridge, Assiniboine River

Enlarge image

In addition to uncertainty in the exposure and effects data, biological and ecological uncertainty needs to be considered. This includes consideration of the potential for organisms such as plants to recover from exposure and the effects of multiple stressors that are likely present. Ammonia is a constant component of sewage, but the concentrations will vary considerably based on social fluctuations from the city generating the effluent. It has been shown that organisms will tolerate higher concentrations of ammonia if the exposure is pulsed, rather than constant. Sewage effluent is also a complex mixture and is released to a variety of different ecosystems that will respond differently. Typical ecosystems include coastal marine/estuary systems, lake systems, and small and large river systems. Potential impacts are not strictly from ammonia, but could also be from excess chlorine, chlorinated compounds, chloramines, biochemical oxygen demand, chemical oxygen demand and metals.

In the North Saskatchewan River below Edmonton, confounding toxicity could arise from chloramines, as the City of Edmonton chloraminates its drinking water. A PSL toxicity assessment of chloramines is being conducted. Dissolved oxygen is typically high in the river, so lack of oxygen is not likely a component of the toxicity "package." Excessive organic matter could also generate an enriched environment that some pollutant-tolerant organisms will find attractive.

Hamilton Harbour is seriously impacted by the many industries and cities lining its shores. The sediments are generally contaminated with metals, polychlorinated biphenyls and polycyclic aromatic hydrocarbons. The bottom waters of the harbour become anoxic in summer, likely due to a combination of poor water exchange with Lake Ontario, biological degradation of organic matter from sewage, nitrification of ammonia and biological degradation of sediments.

Winnipeg does not disinfect its sewage, so there are no problems with excessive chlorine. It uses chlorine for drinking water, however, so there may be excessive chlorine from this exposure as well as some short-term exposure to chloramines as chlorine and ammonia react.

The Red River below Winnipeg is slow moving, but relatively deep. Some oxygen sag has been noted in previous surveys; however, excessive algal growth below the WWTPs augments this somewhat.

The information presented in this risk assessment shows that ammonia is a chemical of concern. However, it is important to remember that ammonia toxicity is being assessed independently of all other stressors, including other effects of ammonia, such as its effects as a nutrient on primary production and its effects on dissolved oxygen concentrations from nitrification and plant respiration.

3.1.2.5 Interpretation of ecological significance

The toxicity of ammonia downstream of a sewage outfall varies with many parameters, the most important of which are the concentration of ammonia in the effluent, the temperature of the water, the pH of the water, the flow rate of the water system and the flow rate of the effluent. Secondarily, the way in which the effluent enters the receiving environment is important; a multi-port diffuser dilutes the ammonia more rapidly in the water column so that it poses less of a risk than an effluent that enters a water system as a single plume. Discrete plumes from a point source tend to disperse slowly in water systems unless they are highly energetic, i.e., exposed to tides or strong currents. Temperature cannot be separated from the toxicity equation due to a lack of adequate information on the relative toxicity of ammonia at different temperatures.

The maximum risk of acute and chronic toxicity in an aquatic ecosystem downstream from a sewage outfall or other point source of ammonia occurs at a combination of low flows, high temperatures and high pH values, typically in late summer and early fall. The temperature and pH conditions will drive the proportion of un-ionized ammonia to chronically toxic levels, and the low flows ensure that there is not sufficient dilution capacity in the water system to accommodate the amount of ammonia present. The toxic risk is very low for waters all across Canada from December to April due to the low water temperatures and reduced pH levels. Most agricultural runoff of ammonia occurs in early spring before it can bind to soil and will not have a significant impact on aquatic ecosystems.

The ecological impact of ammonia in aquatic ecosystems is likely to occur through chronic toxicity to benthic invertebrates and fish populations as a result of reduced reproductive capacity and reduced growth of young. These are subtle impacts that will likely not be noticed for some time below an outfall. Typically what happens is a decline in the numbers of a sensitive species. Unless there is continual recruitment from unaffected populations, the affected population may die out over time. Toxic impacts on aquatic ecosystems can extend for many kilometres below a large sewage outfall. Impacts on fish populations are very difficult to determine due to the mobile nature of many fish species and to recruitment of fish from non-impacted areas. Benthic invertebrates are a much better indicator of impact, as they are not very mobile for much of their life cycle. Below the outfall at Edmonton, the diversity and benthic community structure were severely disrupted for over 20 km. Some of the pollution-sensitive insects did not make a comeback at 100 km downstream. It is difficult to determine if this impact is from ammonia, one of the other major components of sewage effluent, or a combination of factors.

Ammonia is a fundamental building block of life; as such, it is a nutrient for primary producers. Some terrestrial ecosystems in Europe, especially coniferous forests, moors and fens, are being seriously affected by excess nitrogen, much of which is in the form of ammonia. The ammonia raining down on the Netherlands, Belgium, Germany and the United Kingdom is largely from intensive agricultural operations. In these cases, the ecological disturbance is through terrestrial eutrophication and a toxic reaction to beneficial mycorrhizae symbiotically associated with tree roots. The nutritional balance of the ecosystems is being upset so that the existing dominant plants are being destroyed or pushed out by plants more capable of using nitrogen. This same phenomenon is not happening in Canada, as we have more space with which to dilute the ammonia, and we do not have as many sensitive ecosystems in close proximity to ammonia sources. That said, there are potential instances of this occurring in Canada, in particular in the Lower Fraser Valley.

Aquatic eutrophication is generally limited not by ammonia, but by phosphorus. This was confirmed recently by a joint review by Environment Canada, Fisheries and Oceans Canada and Agriculture and Agri-Food Canada. Therefore, there is little ecological impact from aquatic eutrophication due to excessive ammonia concentrations. It follows that at least one example of a nitrogen-impacted water system exists in Canada. The Qu'Appelle Lakes in Saskatchewan downstream of Regina are likely impacted by ammonia loadings. The City of Regina does not remove nitrogen from its sewage effluent.

Benthic organisms and aquatic macrophytes below the Edmonton WWTP have been severely affected by the effluents, a major component of which is ammonia. The distribution and abundance of many aquatic insects have been altered, and the growth of aquatic macrophytes increases dramatically downstream of the outfall. Much of this can be attributed to excessive nutrients, including ammonia.

Due to the interaction between receiving water pH and temperature, those waters most at risk from sewage-related ammonia are those that are routinely basic in pH with a relatively warm summer temperature combined with low flows. In Canada, winter temperatures, regardless of pH, are low enough to keep the formation of unionized ammonia below the toxic threshold. Potentially toxic conditions typically start in May and can continue through to early October, depending on the water system and the yearly variation in pH, temperature and dissolved oxygen. In general, waters potentially sensitive to ammonia from WWTPs are found in southern areas of Alberta, Saskatchewan and Manitoba; southern Ontario; and the south shore of Quebec.

Most of the urban populations in the Maritime provinces and British Columbia discharge to a large river (St. John River, Fraser River), to lakes or directly to the ocean. There is little information on, or evidence of, potentially significant impacts of these discharges on their receiving environments, due largely to the high dilution capacity of the water bodies.

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

Ammonia does not deplete stratospheric ozone or contribute significantly to the formation of ground-level ozone, and its potential contribution to climate change is negligible.

3.3 Conclusions

CEPA 1999 64(a):

Based on available data on releases of ammonia from municipal wastewater treatment plants and the aquatic conditions routinely found downstream of many such outfalls in Canada, it has been concluded that ammonia is 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 environment or its biological diversity. Therefore, ammonia is considered to be "toxic" as defined under Paragraph 64(a) of CEPA 1999.

CEPA 1999 64(b):

Based on available data, it has been concluded that ammonia is not entering the environment in a quantity or concentration or under conditions that constitute or may constitute a danger to the environment on which life depends. Therefore, ammonia is not considered to be "toxic" as defined under Paragraph 64(b) of CEPA 1999.

Overall conclusion:

Based on critical assessment of information on ammonia relevant to the aquatic environment, ammonia is considered to be "toxic" as defined in Section 64 of CEPA 1999.

3.4 Considerations for follow-up (further action)

The conclusion of this assessment is based on analyses of risks posed by releases of ammonia from municipal WWTPs. Priority should therefore be given to consideration of options to reduce exposure to ammonia from municipal wastewater systems. Since the toxicity of ammonia is dependent on many site-specific variables, options to reduce exposure to ammonia from municipal wastewater systems should be examined on a site-specific basis. If a city or region has a water body with a large dilution capacity, then ammonia control may not be necessary, or perhaps an improved dilution system may be required. If, however, there is not a sufficient dilution capacity, then additional treatment may be required. This typically takes the form of converting ammonia to nitrate. A further step of converting nitrate to nitrogen reduces the possibility of nitrate toxicity and oversupply of nutrients, but is considerably more costly.

Results of conservative screening-level assessments suggest that releases of ammonia from several other sources may also be causing environmental harm, but available data were insufficient to establish the extent and magnitude of such harm. It is therefore recommended that additional data be obtained to determine whether options to reduce exposure to ammonia from such sources should be undertaken. The following data needs are listed in order of priority:

• concentrations in waters receiving inputs of ammonia from strong industrial point sources,
• concentrations in Canadian streams and rivers receiving runoff from agricultural sources, including manure-fertilized fields and intensive livestock operations (feedlots and dairies),
• deposition of ammonia from the atmosphere, both in background areas and near point and area sources in Canada, and
• concentrations of ammonia in Canadian marine waters at municipal wastewater outfalls.