Health Canada
www.hc-sc.gc.ca
Common menu bar links
Institutional links
Priority Substances List Assessment Report for Releases from Primary and Secondary Copper Smelters and Copper Refineries - Releases from Primary and Secondary Zinc Smelters and Zinc Refineries
2.0 Summary of Information Critical to Assessment of "Toxic" under CEPA 1999 (Continued)
2.3.2 Releases to water
2.3.2.1 Fate and concentrations
Information on the fate and environmental concentrations resulting from direct release to water from CCR, CEZinc and CTO is summarized below. Further details are provided in Beak International (1999).
The concentrations of chemicals released into a river environment may be as high as the concentrations in effluent at the end-of-pipe, but then decrease in a downstream direction along the centreline of the plume. This is associated with the widening or lateral dispersion of the plume. At any point downstream, concentrations decline with lateral distance away from the plume centreline. Eventually, a point may be reached where the effluent is fully mixed with river water (hence, there is no further dilution unless the river flow is augmented) or the chemical concentrations arising from the effluent are small in relation to natural background concentrations and hence of little biological significance.
Chemical concentrations in the effluent have been summarized in Table 9 for each of the three facilities. In addition, the proportion of chemical dissolved or adsorbed is indicated for each effluent.
The spatial pattern of chemical concentrations arising from the discharge must be considered, in conjunction with the movements of receptor species, to determine the chemical exposure concentrations experienced by these organisms. Since recent monitoring data were not available, plume models were developed in these assessments for releases from each of the operations. Based on these models, the expected patterns of relevant chemical concentrations in the receiving water and sediment were defined.
Estimated Exposure Values (EEVs) for fish are calculated in these assessments using a conservative spatial averaging approach, consistent with Environment Canada (1997a) guidance for Tier II exposure assessments. The fish is assumed to reside in the plume, with a home range immediately downstream of the outfall. Thus, a realistic worst-case spatially averaged EEV is determined. It is likely that only larger fish (200-300 mm in length) would be able to maintain position for prolonged periods in a 1 m/s current, and thus be resident in the plume. Thus, based on Minns (1995), a home range on the order of 10 000 m2 is conservative for lake species. A smaller home range on the order of 1000 m2 may be more appropriate for stream and creek species; however, fish in large rivers may be closer to the lake model. The smaller home range area was utilized for spatial averaging to calculate an upper-limit EEV.
EEVs for pelagic invertebrates are calculated in these assessments by assuming that the organisms drift along the plume centreline at the same velocity as the receiving water. As they travel, they experience declining concentrations until they reach the point downstream where the plume is no longer discernible from background.
For the high-velocity receiving waters considered here, a 10-km trip takes only 3-4 hours. Most of the exposure above background occurs within the first few kilometres (i.e., within 1 hour). Thus, a reasonable maximum spatially averaged EEV is obtained by averaging exposure concentrations over a 1-km (20-minute) trip down the plume centreline.
EEVs for benthic invertebrates are calculated in these assessments as point concentrations at locations where habitat is generally suitable to support a benthic community. Spatial averaging is not appropriate, since movements of benthic invertebrates are extremely limited.
2.3.2.1.1 Canadian Copper Refinery
A screening-level model (Sayre, 1973; NCRP, 1996) was used to define the spatial pattern of metal concentrations in the St. Lawrence River arising from the MUC-WWTP effluent. The CCR facility contributes to this effluent as outlined in Table 7. The model gives the concentration of any chemical constituent arising from the discharge, at any point in the river downstream, based on advection and symmetric lateral dispersion processes.
This model assumes vertical mixing, which is typically complete within a downstream distance of 7*depth, or in this case about 50 m. It assumes that shoreline effects are negligible, as appropriate for the mid-river outfall of the MUC-WWTP. It also ignores any jetting or turbulent mixing that may occur in the immediate proximity of the outfall.
The lateral dispersion coefficient was 0.06*depth*velocity. A mid-river depth of 7 m and a velocity of 0.81 m/s were utilized in the model, based on cross-sectional area and flow information provided by Hudon and Sylvestre (1998). Using these input parameters and the chemical loadings for the MUC-WWTP for 1995 shown in Table 7, resulting chemical concentrations in water were estimated for the downstream St. Lawrence River. For each metal, the maximum concentration in the plume is the concentration in the effluent that can be seen in the first few metres downstream of the outfall. The concentration declines with further distance down the plume centreline, rapidly at first and then more slowly.
The concentration of metal dissolved in water (i.e., not adsorbed on suspended solids) was computed at each point in the plume using a sorption model, as follows:
Cdiss = Ctot/(1 + Kd * 10 * Css x 10-6)
where:
- Cdiss = dissolved metal concentration (mg/L),
- Ctot = total metal concentration (mg/L),
- Css = suspended solids concentration (mg/L), and
- Kd = distribution coefficient for soils (L/kg).
The Kd is increased by a factor of 10 to represent adsorption in aqueous systems, which is typically greater than that in consolidated soil (O'Conner and Connelly, 1980). Values of soil Kd were taken from Sheppard et al. (1992, 1999), except for Hg, where an aqueous value from Birge et al. (1987b) was used directly without adjustment. There is typically order of magnitude uncertainty in Kd values. Suspended solids were considered to show the same pattern of spatial dispersion described above for metals, declining from an initial concentration in MUC-WWTP effluent (36 mg/L) to a Lac St. Louis background value of 4 mg/L (Rondeau, 1993). The dissolved metal concentrations computed in this manner are only slightly below the total metal concentrations, with the greatest differences near the outfall where suspended solids concentrations are highest.
The metal concentrations in newly formed sediments that might arise at each point in the plume were computed from the difference between dissolved and total metal concentrations, as follows:
Csed = (Ctot - Cdiss) * 106/Css
where:
- Csed = metal concentration in sediment (mg/kg),
- Ctot = total metal concentration in water (mg/L),
- Cdiss = dissolved metal concentration in water (mg/L), and
- Css = suspended solids concentration in water (mg/L).
Whether sediments will actually settle out at a particular point depends upon the water velocity profile and particle size of solids. In this assessment, the location of areas likely to receive sedimentation is a matter of professional judgement, based on locations of embayments and backwaters in contact with the plume.
Similar calculations of concentrations of metals dissolved in water and deposited in newly formed sediments were performed using Lac St. Louis background concentrations in water (Rondeau, 1993) as a starting point. Regional background concentrations were then added to the incremental concentrations arising from the discharge (above) in order to estimate the overall concentrations experienced by aquatic biota at any point.
Based on these calculations and spatial averaging calculations for the mobile receptor species, as described above, the modelled annual average EEVs for different receptors exposed to available metals in the vicinity of the MUC-WWTP are listed in Table 26. The dissolved portion of metals in water (Cdiss) was considered to be available for fish, pelagic invertebrates and epibenthos, while the metal adsorbed to sediment (Csed) was considered potentially available to infaunal benthic organisms. The percentage of each exposure attributable to CCR, rather than regional background or various other municipal/ industrial users of the MUC-WWTP, is shown in parentheses.
- Spatial average for fish resident in plume with 30 x 40 m home range (average = 0.21*maximum plume concentration).
- Time average for zooplankton drift down centre of plume for 1000 m (average = 0.12*maximum plume concentration).
- Point concentration on plume centreline 1000 m from outfall.
- Point concentration in sediment on plume centreline 1000 m from outfall.
- Exceeds corresponding ENEV (Table 35).
2.3.2.1.2 Canadian Electrolytic Zinc
Empirical data from a 1990 plume study were utilized to develop a near-field descriptive model of the effluent dilution pattern at the CEZinc UNA effluent discharge to the Beauharnois Canal. The data included Zn and Se measurements on a series of transects across the UNA plume. Selenium was the best tracer of the plume, since concentrations were well above background in the receiving water.
The data show concentrations that are highest near the outfall (about 10 m out from the bank) and declining in both downstream and lateral directions. The plume is not vertically mixed at the outfall (higher concentrations at depth) but is vertically mixed within 50 m. It was distinguishable above background at the time of the study to a maximum distance of a few hundred metres. The descriptive model reflects this pattern, with exponential decline from the outfall concentration in both lateral and downstream directions.
The Se loading at the time of this study was approximately 600 mg/s (52 kg/d). In order to generalize the model to accommodate different loadings for different chemicals and for the present-day combined effluent, the Se dilution model was multiplied by a loading ratio (new loading/600 mg/s). This has the effect of increasing or decreasing the plume concentrations and the extent of the plume above background as loadings are increased or decreased. However, the underlying plume shape (i.e., degree of change with distance) is assumed not to change with loading.
This purely descriptive near-field model does not conserve mass when extended in the downstream direction. Therefore, an alternate model is needed for the far-field region. A screening-level model for nearshore discharge (Sayre, 1973; NCRP, 1996) was used to define the far-field spatial pattern of chemical concentrations arising from the CEZinc combined effluent. A 10-km setback distance was used to achieve the observed lateral dispersion at the downstream edge of the near-field region. The model gives the concentration of any chemical arising from the discharge at any point in the Beauharnois Canal downstream, based on advection and one-sided lateral dispersion processes.
The lateral dispersion coefficient was 0.06*depth*velocity. An average depth of 5 m for the nearshore canal and a velocity of 0.5 m/s were utilized in the far-field model, based on cross-sectional areas from hydrographic charts and flow information provided by CEZinc (personal communication with facility operators). The depth and velocity are greater further out from shore (e.g., 7.5 m and 0.74 m/s). The flow data used were consistent with Hudon and Sylvestre (1998).
Using these models and input parameters and the maximum monthly or annual average loadings for 1995 from CEZinc (Tables 7 and 8), resulting chemical concentrations in water were estimated for the Beauharnois Canal downstream from CEZinc. For each chemical, the maximum concentration in the plume is the concentration in the effluent that can be seen in the first few metres downstream of the outfall. The concentration declines with further distance down the plume centreline (i.e., near the bank), rapidly at first and then more slowly.
At each point in the plume, the concentrations of metals dissolved in water (i.e., not adsorbed on suspended solids) and the concentrations in the solids that may contribute to newly formed sediments were computed as described in the preceding section. Ammonia was considered to be entirely dissolved (i.e., none adsorbed) and was not partitioned to sediments.
Similar calculations of concentrations of chemicals dissolved in water and deposited in newly formed sediments were performed using Lac St. Francis background concentrations in water (Rondeau, 1993) as a starting point. Regional background concentrations were then added to the incremental concentrations arising from the discharge (above) in order to estimate the overall concentrations experienced by aquatic biota at any point.
Based on these calculations and spatial averaging calculations for the mobile receptor species as described above, the modelled maximum short-term EEVs for different receptors exposed to metals in the vicinity of CEZinc are listed in Table 27. The percentage of each exposure attributable to CEZinc, rather than regional background, is shown in parentheses. Maximum short-term EEVs could not be estimated for Pb or ammonia, as data were not available to determine the factors for estimation of short-term loading rates from annual mean loading rates (see Tables 7-9). Exposure concentrations (mg/L) based on annual average loadings are, for Pb and ammonia respectively, 1.26 and 157 for fish, 0.46 and 16.8 for zooplankton, 0.41 and 7.99 for benthic-epifauna, and (in mg/kg) 1.70 and 0 for benthic-infauna.
2.3.2.1.3 Cominco-Trail Operations
Empirical data from a 1997 plume study (Frew, 1997) were utilized to develop a near-field descriptive model of the effluent dilution pattern at the C-III discharge from CTO to the Columbia River. The study was performed under low river flow conditions. The data included thallium (Tl) and other metal measurements across the plume at various points downstream, as well as photographs of dye dispersion. Thallium is a suitable tracer of the plume, since background river concentrations are low and Tl is mainly in dissolved form.
The Tl data show concentrations that are distinguishable above background for at least several kilometres downstream. The dye dispersion indicates initially rapid lateral mixing (in the channel behind an island where the discharge occurs) followed by slower lateral mixing. The descriptive model reflects this pattern, with exponential decline from the outfall concentration in both lateral and downstream directions.
- Spatial average for fish resident in plume with 40 x 40 m home range (average = 0.4*maximum plume concentration) -maximum monthly average loading.
- Time average for zooplankton drift down centre of plume for 1000 m (average = 0.031*maximum plume concentration) -maximum monthly average loading, except for Zn.
- Point concentration on plume centreline 1000 m from outfall - maximum monthly average loading.
- Point concentration in sediment on plume centreline 1000 m from outfall - maximum monthly average loading.
- Exceeds corresponding ENEV (Table 35).
- Based on maximum 4-day average loading.
- Zinc concentrations reflect the portion of loading dissolved or adsorbed at source.
The Tl loading at the time of this study was 172 mg/s (15 kg/d). To represent different loadings for different chemicals, the Tl dilution model was multiplied by a loading ratio (new loading/172 mg/s). This is the same approach that was used for the near-field portion of the CEZinc plume.
The same descriptive model was used to describe the near-field plume from the C-II outfall 0.8 km downstream. Concentrations rising from the two outfalls were added to estimate the combined concentrations at near-field locations further downstream.
For locations beyond several kilometres downstream from the C-III outfall, a far-field model was utilized, as previously described for CEZinc. A combined C-II + C-III loading was used, with a 90-km setback to achieve the observed lateral dispersion at the downstream edge of the near-field region. The lateral dispersion coefficient was 0.12*depth*velocity. A depth of 3 m and a velocity of 1 m/s were used, based on hydrological data from Aquametrix (1994) and MES (1997).
Using these models and input parameters and the maximum monthly or annual average chemical loadings from the zinc operations at Trail for 1998 (Tables 7 and 8), chemical concentrations in the downstream waters and newly formed sediments of the Columbia River were estimated, as previously described for CEZinc and CCR. The estimated maximum concentration in the plume (1 m from outfall) is approximately 1/25 of the effluent concentration. The concentration declines slowly with further distance down the plume centreline. The concentration increases again for some metals (particularly Zn) at the point where the C-II discharge joins the C-III plume, then continues to decline.
At each point in the plume, the concentrations of metals dissolved in water (i.e., not adsorbed on suspended solids) and the concentrations in solids that may contribute to newly formed sediments were computed as previously described for CEZinc and CCR. Ammonia was considered to be entirely dissolved (i.e., none adsorbed) and was not partitioned to sediments.
Similar calculations of concentrations of chemicals dissolved in water and deposited in newly formed sediments were performed using Birchbank background concentrations in water (MES, 1997), supplemented to account for the influence of the C-IV outfall and Stoney Creek just upstream of the C-III outfall, as a starting point. Birchbank is located about 3 km upstream of C-IV and Stoney Creek. These regional "background" concentrations were then added to the incremental concentrations arising from the CTO zinc operations (above) in order to estimate the overall concentrations experienced by aquatic biota at any point.
Based on these calculations and spatial averaging calculations for the mobile receptor species as described above, the modelled maximum short-term EEVs for different receptors exposed to metals, ammonia and fluoride in the vicinity of CTO are listed in Table 28. The proportion of each exposure attributable to CTO zinc operations, rather than regional background or lead operations or fertilizer operations, is shown in parentheses.
2.4 Effects characterization
2.4.1 Ecotoxicology
This section presents information on the effects of various release constituents on sensitive species of relevance to Canada. Effects information is summarized either as CTVs and ENEVs, expressed as concentrations in an exposure medium, or as critical loads. ENEVs and CTVs are, respectively, estimates of the upper limit of no effect and of low toxic effect concentrations. CTVs are derived from studies of toxic effects to relevant sensitive laboratory organisms (e.g., reduced reproduction in Daphnia). These laboratory effects data are the "measurement endpoints" for the assessment.
- Spatial average for fish resident in plume with 40 x 40 m home range (average = 0.88*maximum plume concentration) -maximum monthly average loading.
- Time average for zooplankton drift down centre of plume for 1000 m (average = 0.45*maximum plume concentration) -maximum 4-day average loading.
- Point concentration on plume centreline 1000 m from outfall - maximum monthly average loading.
- Point concentration in sediment on plume centreline 1000 m from outfall - maximum monthly average loading.
- Exceeds corresponding ENEV (Table 35).
- Metal concentrations reflect the portion of loading dissolved or adsorbed at source.
ENEVS are derived from CTVs - by dividing by an appropriate application factor (e.g., 10) -and are intended to represent effect thresholds for receptors in the field. The critical effects of concern in the field (e.g., adverse reproductive effects on sensitive aquatic invertebrates) are the "assessment endpoints."
In these assessments, the critical effects of concern are harm to sensitive aquatic organisms (fish, invertebrates and plants) and soil-dwelling organisms (plants and decomposers). Although effects on wildlife were not examined, it was concluded in a recent review document (Welbourn, 1996) that there is very little evidence, based on the limited available data, that releases from copper smelters and refineries and zinc plants are causing adverse effects on wildlife under present conditions.
Critical loads may be defined as the amount of deposition required for contaminant levels to reach threshold effect values (e.g., ENEVs) in receiving media. These loads were calculated using transport and fate models as described below. For the assessment of the effects of metals released to the atmosphere, the conditions assumed in the modelled receiving media (i.e., sandy acidic soils and circumneutral to somewhat acidic lakes) were typical of those on the southern Canadian Shield. These assumptions were made because most of the metal production plants included in these assessments are located on the Shield. Furthermore, receiving media with properties similar to those assumed for fate modelling occur in many other regions of Canada.
To the extent possible, effect values derived in this section were estimated taking into account bioavailability. CTVs for metals were estimated based on either free-ion or total dissolved concentrations in waters (either surface or soil pore waters). Critical loads for metals were estimated as deposition rates of total soluble metals.
The magnitude of the application factor used to derive ENEVs typically increases with the uncertainties associated with effect estimates, arising, for example, from limitations in the amount or quality of toxicity data, and extrapolation of laboratory effects data to field conditions (Environment Canada, 1997a). Since toxicity data were relatively abundant in these assessments, test organisms were typically closely related to organisms likely to be encountered in the field, and because all of the release constituents examined occur naturally, application factors used were very small -typically 1, and never more than 2.
Because release constituents are present in the environment naturally, the upper bounds of natural bioavailable concentrations (typically 95th percentile values) were used to set lower bounds on ENEVs for these substances. This is justified based on the expectation that most natural organisms are unaffected by bioavailable concentrations typically found in nature. Of course, there may be exceptions. For example, near ore deposits, metal concentrations can be extremely high and harmful to some organisms. However, 95th percentiles of natural concentrations are usually not extreme, being within a factor of 2 or 3 of the geometric mean values (Bird et al., 1999). Harmful effects at such concentrations are therefore considered to be unlikely.
Several of the release constituents considered (e.g., Cu, Zn and Ni) are essential micronutrients for at least some organisms. Care must be taken to ensure that ENEVs for such substances are not within the deficiency range. This was accomplished by ensuring that the ENEVs selected are not below 95th percentile values of natural (bioavailable) concentrations, since concentrations equivalent to the 95th percentile natural values are expected to satisfy the nutritional needs of most organisms.
Since the releases being assessed include a mixture of chemical substances (e.g., several metals and SO2 are released together to air), their combined effects can be different (greater or less) from those of their individual constituents. As explained below, data available for metals suggest that combined effects may be assumed to be additive. An alternative, less conservative approach is to assume that release constituents act independently (i.e., that there are no combined effects). In this section, only effects of individual release constituents are described. Combined effects are considered as appropriate in Section 3.0.
2.4.1.1 Releases to air
2.4.1.1.1 Sulphur dioxide
Plants are among the most sensitive receptors affected by SO2 in air (FPACAQ, 1987). Information on effects of SO2 on vegetation has been reviewed by Linzon (1999). While in some cases the addition of low concentrations of SO2 can be beneficial (Linzon, 1999), in general, accumulations of sulphur in leaf tissue beyond certain threshold levels have harmful effects (Linzon et al., 1979). Sulphur dioxide enters leaves mainly through the stomata and is toxic to the metabolic processes taking place in the mesophyll cells (Linzon, 1972). Acute injury is caused by a rapid metabolic accumulation of bisulphite and sulphite. When the oxidation product, sulphate, accumulates beyond a threshold value that the plant cells can tolerate, chronic injury occurs.
Acute injury can occur as the result of plant exposure to high concentrations of SO2 for short periods of time (one to several hours). The injury develops within several hours to a few days after exposure and manifests itself usually as necrosis of foliage accompanied by certain metabolic effects. Chronic injury can occur as the result of plant exposure to constant or intermittent low concentrations of SO2 over long periods of time (over one day to one or more growing seasons). The injury can include metabolic effects (physiological and biochemical), chlorosis of foliage (becoming necrotic), reduction in plant growth and yield, and death of the plants. Results of acute and chronic studies with sensitive Canadian species are described briefly below.
Acute effects: Five studies reported acute injury to Canadian species at relatively low concentrations - between 524 and 1100 mg/m3. Four were experimental studies where receptor species and exposure concentrations were controlled, and one by Dreisinger (1965) involved examination of effects of SO2 in the field under natural (uncontrolled) conditions.
Metcalfe (1941) reported damaging begonia varieties in fumigations of 655 mg/m3 (0.25 ppm) SO2 for 1 hour under very humid conditions. Berry (1967) reported injuring foliage of potted eastern white pine at a concentration of 655 mg/m3 (0.25 ppm) SO2 for 1 hour. A high temperature (27ºC) and a high relative humidity (70%) were maintained during the exposure of eastern white pine to SO2 in a specially built greenhouse chamber. Murray et al. (1975) induced moderate to severe injury on several Kentucky bluegrass cultivars in artificial fumigations of 524 mg/m3 (0.20 ppm) SO2 for 2 hours. Karnosky (1976) produced acute injury on foliage of trembling aspen in artificial fumigations with 910 mg/m3 (0.35 ppm) SO2 for a period of 3 hours. Five aspen clones were tested, and three clones were injured slightly.
Dreisinger (1965) reported observations of acute injury to natural vegetation following SO2 fumigations originating from copper and nickel smelters in the Sudbury area during a 10-year period (1954-1963). The lowest SO2 concentrations for 1 hour found to be associated with a vegetative damaging fumigation were 1466 mg/m3 (0.56 ppm) for a crop species (buckwheat) and 1100 mg/m3 (0.42 ppm) for a forest tree species (trembling aspen).
Chronic effects: Six studies reported chronic injury to Canadian species at relatively low concentrations - between 21 and 74 mg/m3.
Two studies with sensitive species were conducted in controlled experiments. Shaw et al. (1993) conducted long-term open-air fumigation experiments on Scots pine seedlings, using a predetermined pattern of hourly mean values of SO2 based upon monitoring data from a site in central England. Although mean annual concentrations did not exceed 58 mg/m3 (0.022 ppm) SO2 over three years (1988, 1989 and 1990), up to 20% of the trees developed foliar necrosis during each growing season. Kropff et al. (1989) exposed broad bean plants under field conditions to a mean concentration of 74 mg/m3 (0.028 ppm) during the 1988 growing season. This resulted in a reduction of total dry-matter production of 9%, and a seed yield reduction of 10%, compared to controls.
Four studies were conducted under natural uncontrolled conditions. The results of forest studies conducted over a 10-year period in the area near Sudbury affected by sulphur fumes were reported by Linzon (1971). It was found that chronic effects on forest growth were prominent where SO2 air concentrations averaged 44 mg/m3 (0.017 ppm), the arithmetic mean for the growing season for the 10-year measurement period. Chronic effects were slight where SO2 annual concentrations averaged 21 mg/m3 (0.008 ppm). In Celna, Czechoslovakia, Materna et al. (1969) reported moderate chronic injury to foliage of Norway spruce trees under the influence of an average growing season concentration of 50 mg/m3 (0.019 ppm) SO2, which occurred during 1966 and 1967. In studies conducted on the occurrence of lichens at Sudbury (Leblanc et al., 1972), the number of epiphytes found growing on balsam poplar trees was drastically reduced in zones where the annual growing season mean levels of SO2 were over 52 m g/m3 (0.020 ppm), and slightly reduced in zones where the mean levels of SO2 were over 26 m g/m3 (0.010 ppm). Similarly, in Sweden (Skye, 1964), it was found that the survival of lichens was poorer in areas with an annual SO2 concentration of approximately 39 mg/m3 (0.015 ppm).
CTVs and ENEVs: CTVs and ENEVs estimated for chronic and acute conditions are summarized in Table 29. Generally, effects of SO2 observed in the field under natural conditions provide the best basis for estimation of effect thresholds (Linzon, 1999). Since the lowest chronic effect level identified (21 mg/m3) was reported for a field study conducted in Canada under natural conditions (Linzon, 1971), this mean growing season value was chosen as the chronic CTV. Information on acute effects of SO2 on sensitive species under natural field conditions is more limited. Only one study - in a natural Canadian forest - was identified, with effects reported at 1-hour concentrations as low as 1100 mg/m3 (Dreisinger, 1965). Since two other 1-hour controlled experimental studies reported effects at a lower concentration (655 mg/m3), a CTV of 900 mg/m3 was estimated, being the average of the experimental and the natural forest values.
Because of the uncertainties associated with these effects estimates (Linzon, 1999), a small application factor of approximately 2 was used to derive ENEVs - 450 mg/m3 for acute (1 hour) and 10 mg/m3 for chronic (annual or growing season) exposures. The chronic ENEV is several-fold above the estimated upper limit of natural annual mean SO2 concentrations (approximately 2.6 mg/m3) in Canada. Although an upper limit for 1 hour (acute) natural exposures was not identified, it is very unlikely given the low maximum natural annual average concentration that maximum 1-hour natural concentrations in Canada would exceed the acute ENEV.
2.4.1.1.2 Sulphate deposition
The following is a brief summary of information available on the likely impacts of sulphate deposition on aquatic organisms in the four acid-sensitive regions of eastern Canada examined in these assessments. Potential for effects is evaluated in relation to critical loads for wet sulphate deposition. Further details on effect thresholds for aquatic systems and methods of critical load estimation are provided in Environment Canada (1997c,d).
Effects on aquatic organisms - CTV and ENEV: In Canada, the pH of aquatic ecosystems is used as a surrogate parameter to represent the complex relationships between water chemistry and biological effects (Jeffries et al., 1999). It has been determined that sensitive aquatic ecosystems require the maintenance of a pH level of 6.0 or higher. For example, it has been observed that Canadian lakes with pH values of less than 6.0 have fewer species of fish than similar lakes with higher pH values (Environment Canada, 1997d). In the language of these assessments, sensitive fish species may be considered to be assessment endpoints, and pH 6.0 represents a CTV. Since this is the threshold used for evaluation of effects of sulphate deposition, the application factor used to derive the ENEV was 1.0. This is justified because the critical effects data are based on extensive field observations in acid-sensitive parts of Canada.
Effects on aquatic organisms - critical loads: Using a pH of 6.0 as a criterion, aquatic fate and transport models have been used to estimate critical load values for wet sulphate deposition in selected lake cluster receptor areas in eastern Canada (Environment Canada 1997c,d).
1 Derivation of values is detailed in Linzon (1999).
These critical loads may be described as CL5s, since these loadings are considered to be suitable for the protection of 95% of the lakes in the cluster. There is some probability that the other 5% would be adversely impacted. Wet sulphate critical loads for the four acid-sensitive receptor regions selected for examination in these assessments were estimated by visual examination of acid sensitivity maps. Values range from less than 6 kg/ha/a for the Kejimkujik region of Nova Scotia to about 13 kg/ha/a for the Sudbury/Muskoka area (Table 13).
2.4.1.1.3 Metals (Cu, Zn, Ni, Pb, Cd, As)
Bioavailability and the free-ion model: The response of organisms to a toxic substance requires both contact and susceptibility. Contact in this case means more than physical contact; it almost always requires absorption of the dissolved metal into the organism. Absorption requires the metal to be free to move in the environment around the organism and requires that the metal be able to pass through the membranes into the organism. Both of these processes are highly dependent on the chemical form of the metal, and both control the net bioavailability of the metal.
The mobility of the metal in the environment around the organism is essentially determined by the proportion of the element in the solution phase. In surface water systems, this is assumed to be the concentration in filtered water, and in sediment or soil it is the concentration in the (filtered) pore water. Metals attached to particles suspended in water are mobile, but are not as bioavailable as the metals in solution and, as a first approximation, can be ignored. The metals in solution exchange to some extent with those on the solid phase, and a model of this relationship is needed to predict bioavailability. There are many possible models. In this assessment, a simple linear partitioning model is used. It defines the ratio of solid-phase metal to solution-phase as a partition coefficient, Kd, which is assumed to depend upon many environmental factors.
The absorption of a metal into the organism requires the metal to pass through cell membranes. These can be as diverse as cells on the surface of fish gills or cells in plant roots. This typically requires diffusion of the metal to the membrane surface, movement through the membrane (this may be passive or, more rarely, metabolically facilitated), and, finally, movement of the metal away from the membrane inside the cell. The membrane is often envisioned as a cation exchanger, and metabolically facilitated passage through the membrane may involve carrier enzymes that are relatively specific to certain chemical forms of the metals. There are good evidence and support for the concept that the free-ion species of the metals are most able to pass through the membranes. This concept has been formalized in the free-ion activity model (FIAM) for metal absorption (Tessier et al., 1994). There is nothing exact or exclusive about this model, but it is the best at present for dealing with the issue of bioavailability. It should be noted that this approach does not account for the uptake of metals by ingestion and therefore underestimates exposure to metals for some organisms.
The application of FIAM requires a number of assumptions. For one, at environmental concentrations, it is assumed that activity and concentration are so similar that ion concentrations can be used rather than estimations of ion activities. It is also assumed that free-ion concentration can be estimated given the concentrations of the metal of concern and the important complexing ligands in the solution, as well as other geochemical characteristics such as pH and redox potential. Estimating chemical speciation is a rapidly evolving subject and has uncertainties associated with it, not only because of methodology but also because validation is very difficult. It is not easy to measure the free-ion concentration (or activity) of many metals. Despite the uncertainties, the estimation of free-ion concentrations is considered a more useful index of bioavailability than is reliance on total metal concentrations.
One of the major uncertainties in the estimation of free-ion concentrations is the variable effectiveness of dissolved organic matter as a ligand. Dissolved organic matter is typically the dominant ligand affecting the chemical speciation of metals, yet it is an amorphous substance that has proven very difficult to characterize quantitatively. In this assessment, two different geochemical speciation models were used. Each dealt with organic complexants differently. In one, the complexing characteristics of dissolved organic matter are modelled in terms of a diprotic acid, and parameters to reflect the interaction of a generic diprotic acid with the metals were used. In the other model, the behaviour of the natural dissolved organic materials was emulated with a specific mixture of six pure organic acids, for which the metal-ligand interaction parameters are well known. The results of the two models showed the same trends and were quantitatively very similar under some conditions of ligand concentration and pH and somewhat different under others. A geometric mean of the results was used. The two models were used to compute the fraction of dissolved metal present as free ion under what were considered low- and high-ligand concentrations, under low and high partial pressures of CO2 and over a range of pH. The pH range specified was relatively narrow, reflecting the acidic nature of the receiving aquatic and terrestrial environments on the Canadian Shield. The narrow, acidic pH range simplified the geochemical modelling and model interpretation, because at pH < 7, most of the dissolved metals are present as the free ions. The exceptions are Cu and to a lesser degree Pb, where the models indicated that complexes consume a substantial portion of the dissolved metal above about pH 6, depending as well on the other water chemistry attributes. Details of the modelling and interpretation are provided by Bird et al. (1999).
Arsenic, being a metalloid, differs in several ways from the metals being considered here. Arsenic may be present in the environment as a variety of species, normally as one of a number of oxyanions. FIAM is not applicable to such species. Further, the conversion between the chemical species is kinetically limited and at least partially mediated by microbes, and so is not suitably predicted with geochemical equilibrium models. It is also probable that more than one chemical species of As is responsible for toxic effects. It was therefore decided that the best conceptual model at this stage is to assume that all As in solution is bioavailable.
Effects on soil-dwelling organisms -data handling: There is a considerable amount of information in the literature on the uptake and effects of heavy metals on soil-dwelling organisms.
As a first-level analysis of information for the present assessment, only studies showing relatively sensitive effects were considered in detail. Documents supporting existing environmental guidelines and assessments (EC/HWC, 1993; EC/HC, 1994a,b; Environment Canada, 1996a-d, 1997e-i; CCME, 1997) were used to direct the literature search and to provide an expectation of sensitivity for each metal. There was further emphasis on recent literature, where possible, because it was more often complete and relevant with regard to free-ion effects and related geochemistry. Very few papers include free-ion measurements or estimates, and remarkably few provide enough ancillary information to allow users to estimate free-ion concentrations. By requiring information about or relevant to free-ion concentrations, some sensitive ecotoxicology studies that lacked such data were screened out. However, on the whole, this constraint did not seem to seriously affect the derivation of free-ion ENEVs.
Second-level analysis involved relevance, among other criteria. Relevant studies were those that used soils and organisms representative of those found on the Shield. Relevant ecotoxicology endpoints were also required. In the terrestrial environment, assessment endpoints were related to 1) the growth of native tree species and the efficacy of their root symbionts, and 2) the population of litter invertebrates and decomposers capable of maintaining steady-state levels of litter. A third assessment endpoint related to native plants in wetlands was proposed but in practice was no different from 1) above, because there were no data specific to wetland species.
All ecotoxicology studies reported effects on some biological endpoint, but statistical significance and effect level were not always reported. In this assessment, emphasis was placed on sublethal chronic effects at effect levels less than 50%. For example, an effect level of 25% (EC25) for reproductive performance in the long- term study would be a preferred endpoint. In many cases, it was possible to interpolate effect levels to EC25 even if only median effect levels such as EC50 were formally reported. Results were not considered unless there was evidence of a statistically significant or otherwise unambiguous toxic effect.
Estimation of the fraction of the solution-phase metal present as free ion was done for every study with sensitive species where pH and solution-phase concentration were provided. In the ecotoxicology studies that used whole soils, it was common that solution-phase concentrations were not provided and only the concentration on the total soil (solid and liquid phases, on a dry weight basis) was reported. Since these studies were by far the most common, it was necessary to develop a means to estimate the solution-phase concentration from data on total soil concentrations. Because sorption in soils is very complex and very soil-dependent, it was considered inaccurate to extrapolate the solid/liquid partition coefficient, Kd, from other soils. Extrapolation was done only if the study reported corresponding tissue concentrations in higher plants. Tissue concentration is perhaps one of the best indices of bioavailability, because it is based on the actions of the organism. There is a negative correlation between Kd and the plant:soil concentration ratio - high Kd indicates strong sorption and corresponds to low plant uptake. This relationship has been parameterized in a first-order, log-log regression model and this model, was used (Bird et al., 1999) to estimate Kd where solution-phase concentrations were not provided.
Once the ecotoxicology data were interpolated and adjusted as required, they were summarized and ranked from lowest to highest effect concentration. Numerous studies were listed. The 10 most sensitive were considered as possible sources of CTVs for soils. The methodology in CEPA/PSL assessments is to seek (as a starting point) relevant and reliable endpoints for the most sensitive species in the setting of concern. Thus, consideration was first given to the study showing relevant effects at the lowest effect concentration for each metal. This study was re-examined in detail to assure relevance and to compare it to the next few higher reported effects. Most confidence was placed in results from studies that were technically sound and unambiguously reported, and where the next higher effect concentrations were not markedly higher. Case-by-case arguments were developed and documented (Bird et al., 1999) to support the selection of the data that would become the CTV and ultimately the ENEV. All CTVs are reported as free-ion concentrations in soil pore water except for As, where CTVs are reported as concentrations of dissolved As in soil pore water.
The use of the 95th percentile background concentrations to set lower bounds for ENEVs required that these background data also be expressed as free-ion concentrations. In soils, statistical distributions of total soil background concentration were available, so the 95th percentile total concentration could be defined. The solid/liquid partition coefficient, Kd, values applied in the critical load modelling for this assessment (see below) were not considered directly applicable to naturally occurring metal compounds to convert total concentration to solution-phase concentration, because the Kds were often estimated for metals recently added to soil. Therefore, the 95th percentile values from distributions of empirically determined Kd values were used for the uncontaminated background soils to estimate the background solution-phase metal concentrations. An upper percentile Kd was considered appropriate because naturally occurring metal compounds should be less mobile than recently added metals. By assuming median pH and geochemical conditions, solution-phase background concentrations were converted to background free-ion concentrations.
Effects on soil-dwelling organisms - CTV and ENEV values: Chronic ENEVs for soil-dwelling organisms estimated for Cu, Zn, Ni, Pb, Cd and As are summarized in Table 30.
For Cu in soil, effect levels derived from the eight most sensitive studies (in order of decreasing sensitivity - Miles and Parker, 1979; van Gestel et al., 1991; Halsall, 1977; Chang and Broadbent, 1981; Schat and Ten Bookum, 1992; Korthals et al., 1996; Walsh et al., 1972, referenced and described in detail by Bird et al., 1999) spanned an order of magnitude in free-ion Cu concentrations in the solution phase, from 0.01 to 0.1 mg/L. This range encompassed both plants and decomposers. Upon detailed analysis, it was clear that the more sensitive studies were somewhat ambiguous. No one study was clearly the most appropriate to set the CTV, and a study in the midpoint of the range (Schat and Ten Bookum, 1992), with EC20 sublethal effects at 0.04 mg/L, was chosen. The effects were for root growth of a grass, and because there was no clear demarcation in effects between plants and decomposers among the sensitive studies, this concentration was used as the CTV for both plants and decomposers. It is an order of magnitude higher than the estimated 95th percentile background free-ion concentration, so the CTV was based solely on the effects data.
For Zn in soil, there was more demarcation of effect concentrations for plants and decomposers, so separate CTV values were set for each. The overall most sensitive study was for an Enchytraeid worm (Posthuma et al., 1997), and the study was superior among the others considered (referenced and described in detail by Bird et al., 1999) in providing the required data. The next most sensitive studies with invertebrates (Smit and van Gestel, 1996; Chang and Broadbent, 1981; Sheppard et al., 1993) had effect concentrations within about three-fold of the Posthuma et al. (1997) study, and so were supportive of the choice. The effect was an EC50 for decreased reproduction in a chronic exposure study, and the free-ion effect concentration used for the CTV was 0.28 mg/L. This is above the estimated 95th percentile background free-ion concentration, so the CTV for decomposers was based solely on the effects data. For plants, the most sensitive endpoint (Sheppard et al., 1993) was time to bloom initiation, and so was also a non-lethal effect in a chronic exposure study. The next most sensitive studies with plants (MacLean, 1974; Dixon and Buschena, 1988) had effect concentrations within about four-fold of the most sensitive, and so were supportive of the choice. The free-ion effect concentration used for the CTV was 0.46 mg/L, which is above the estimated 95th percentile background free-ion concentration, so that the CTV for plants was also based solely on the effects data.
- Concentrations for As are expressed on a total dissolved basis. All others are on a free-ion basis.
- Due to the use of application factors of 1, ENEVs are equal to CTVs.
For Ni in soil, the most sensitive study reported free-ion effect concentrations an order of magnitude below all the others, and so was not considered sufficiently supported to use as the CTV. The second-most sensitive study (Dixon, 1988) was more consistent. It reported a non-lethal endpoint with an EC72 effect level at a free-ion concentration of 0.2 mg/L. The effect was the mycorrhizal infection of oak roots, a very relevant endpoint for boreal forests where symbiotic mycorrhizal relationships are often essential to tree survival. Because no invertebrate studies showed similar sensitivity, this value was chosen as the CTV. The five next most sensitive studies (Wilke, 1988; Dixon and Buschena, 1988; Dixon, 1988; Taylor, 1989; Taylor et al., 1992, referenced and described in detail by Bird et al., 1999) had effect concentrations within four-fold of 0.2 mg/L, providing strong support for the choice of CTV. This value is almost 10-fold above the estimated 95th percentile background free-ion concentration, so the CTV was based solely on the effects data.
For Pb in soil, the fifth most sensitive study (Seiler and Paganelli, 1987) was chosen for the CTV, even though it was about 10-fold less sensitive than the most sensitive study. The effect concentration was 0.12 mg/L for an EC40 on shoot and root growth of spruce, a very relevant endpoint. The more sensitive studies (Balba et al., 1991; Miles and Parker, 1979) were for less relevant species (e.g., tomatoes) and had substantial uncertainties related to the Kd values and fractions of soluble Pb present as free ion. The four next most sensitive studies (Chang and Broadbent, 1981; Seiler and Paganelli, 1987; Wilke, 1988; Dixon and Buschena, 1988) were within three-fold of the CTV, and thus supported the choice. Apart from two studies with microbial endpoints, only plant-related endpoints were among the 10 most sensitive, and so the CTV for decomposers was set the same as for plants. The CTV is almost 20-fold above the estimated 95th percentile background free-ion concentration, and so was based solely on the effects data.
For Cd in soil, the most sensitive study (Ibekwe et al., 1996) was also the only soil-related study found that reported ecotoxicology data relative to free-ion concentrations. It used quite different techniques than all the other studies, in that free-ion concentrations were controlled in solution culture with specific chelating agents. Because the results were fivefold different from the next most sensitive and the technique was novel and not fully accepted in the literature, this study was not used for the CTV. The next most sensitive study (Wetzel and Werner, 1995) was used as the CTV, and it reported a non-lethal EC20 for a plant root symbiont with an effect concentration of 0.008 mg/L. No data for decomposers were among the 10 most sensitive studies. This CTV is 20-fold above the estimated 95th percentile background free-ion concentration, so the CTV was based solely on the effects data. This CTV is mid-way between, five-fold above and below, the studies ranked as less or more sensitive (Ibekwe et al., 1996; Wetzel and Werner, 1995; Bingham et al., 1975; Wilke, 1988; Taylor and Stadt, 1990, referenced and described in detail by Bird et al., 1999), and so is supported by the other studies.
For As in soil, the most sensitive study (Wetzel and Werner, 1995) was for plants and their symbionts grown on agar. It had an effect concentration five-fold lower than the next most sensitive studies. The seven next most sensitive studies (Steevens et al., 1972; Jacobs et al., 1970; Woolson, 1973; Jacobs and Keeney, 1970; Sheppard et al., 1982) had effect concentrations from 0.05 to 0.50 mg/L and were all for plants grown in soil. This is a more relevant growth medium, but the interpretation is complicated because in all seven cases, Kd was estimated from plant:soil concentration ratios. The choice for CTV (Woolson, 1973) was made because the underlying Kd was considered the most reliable of the seven studies. The effect level was an EC29 of 0.07 mg/L for a non-lethal effect on plant yield. This is more than 100-fold above the estimated 95th percentile background free-ion concentration, so the CTV for plants was based solely on the effects data. The only non-plant effect concentration among the 10 most sensitive studies was for phosphatase activity (Wilke, 1988), and this was used for a CTV for decomposer organisms. The effect level was an EC50 of 1.9 mg/L, well above background.
In all cases for CTVs in soil, the application factor was set to unity, in part because the measurement endpoints were quite applicable to the assessment endpoints related to tree and tree-symbiont growth and litter decomposition, but also because for some elements (particularly Zn) CTVs were not much above the 95th percentile natural background values. Thus, the CTVs become the ENEVs.11 When, for a given metal, ENEVs for plant-related and decomposer-related effects differed (i.e., for Zn and As), the lower value was selected as the "primary" ENEV, used in calculation of "primary" critical loads (below) and risk quotients (Section 3.0). The choices for ENEVs (Table 30), when compared on a similar total-metal bulk soil concentration basis, agreed well with existing summaries (CCME, 1991; Klepper and van de Meent, 1997).
Effects on soil-dwelling organisms - critical loads: The ENEVs represent the concentration in the soil that, if exceeded, could cause non-lethal (e.g., 40%) reductions in performance of key organisms. There are many processes that link the concentration in soil to the deposition flux of metal to the soil and vegetation surface. These processes were dealt with using a model to simulate transport of water and contaminant in soil. Several assumptions were required.
The first assumption for the model development was that only the soluble metal in the flux to the soil surface would be considered. The insoluble fraction was assumed to be very slowly released and of no consequence. Also, flux of metal through the soil was assumed to be by diffusion and convective mass flux with water. Downward flux of water resulted from excess precipitation and upward flux from evapotranspiration and capillary rise. Finally, the metals were assumed to be sorbed onto the immobile soil solids following a Kd relationship.
The model was formulated as an analytical solution to the convection/dispersion equation and was validated by Elrick et al. (1994, 1997).
The input parameters were selected to reflect boreal forest and Canadian Shield conditions (Sheppard et al., 1999). Parameters were assigned best-estimate values (medians of the distribution of possible values) for use in a deterministic manner and probability distributions of values for use in a probabilistic manner. For the probabilistic simulations (1000 repeated runs of the model), parameter values were chosen for each run from their statistical distributions, with care to consider appropriate correlations among parameters. The parameters set specifically for this assessment were net water flux (median 0.47, range 0.25-0.68 m/a), effective water velocity (median 3.6, range 3.5-10.7 m/a), moisture content (median 0.13, range 0.05-0.20 m3/m3), pH (median 5.1, range 3.5-7.0), dispersion coefficient (median 0.0067, range 0.005-0.01 m2/a) and Kd. The geometric means specified for Kd (L/kg) were Cu: 314, Zn: 63, Ni: 116, Pb: 534, Cd: 40 and As: 417. For each element, the geometric standard deviation (GSD) for Kd was set at 5, and truncations were set at two GSDs above and below the geometric mean.
Another key assumption was to use the model results at steady-state soil concentrations, which is desirable because concentrations are constant in time after steady state is reached. Steady state was defined at the 5-cm soil depth. The model results showed that steady state may not be reached until several centuries of constant flux have passed (Sheppard et al., 1999). Fluxes are unlikely to be constant for that long, although metalliferous areas in Europe have been successively exploited over such time periods. In the absence of some criterion to define time lapse, and recognizing that surface horizons will reach steady state much earlier than deeper horizons, only results at steady state were used. It should be recognized that because historic emissions from some facilities (and concomitant local deposition) were significantly greater than current emissions, metal concentrations in soils near some facilities must decrease to reach steady-state levels.
The model was run with an input of a unit flux density of soluble metal to the surface. Concentrations of free ion in solution at 5-cm depth were output. Because the model is linear with respect to concentration, the results could then be scaled with flux or concentration. The outcome was a ratio of steady-state soil concentration to input flux density. Because of linearity in the model, this ratio could be applied to any ENEV to predict the flux density that would result in that concentration.
It is assumed that natural background free-ion metal and contaminant free-ion metal have the same biological effect and the concentrations are additive. As a result, critical loads were defined as the flux density to the surface that will increment the expected median background free-ion concentration up to the concentration of the ENEV.
Deterministic runs with median values of all the parameters were used to define the "median" critical loads (CL50), and probabilistic runs were used to define the deviations from the median case. Deviations from median critical loads result from variations in background free-ion concentrations and in model parameters. The ENEVs were in all cases assumed to be invariant. Variations in background and model parameters were intended primarily to represent spatial variability across the Shield.
A median critical load (50th percentile) is the flux density that will result in steady-state concentrations that are higher than the ENEV in 50% of Shield soils. Similarly, when deposition equals the 10th percentile critical load (CL10), 10% of soils on the Shield are predicted to have steady-state concentrations higher than the ENEV. Effects equivalent to at least a 20% reduction in performance may be experienced by terrestrial organisms in this 10% of soils. Geometric mean, 25th and 10th percentile critical loads for soils are shown in Table 31.
Effects on aquatic organisms - data handling: In general, the handling of data and selection of CTVs for aquatic organisms were the same as for soil-dwelling organisms (Bird et al., 1999). In surface water, the assessment endpoints were related to 1) the survival of populations of pelagic and/or benthic invertebrates, 2) survival of populations of fish, and 3) the productivity of populations of aquatic plants.
The objective of this portion of the work was to establish levels causing low chronic effects on organisms. However, in the ecotoxicology data for aquatic organisms, there is greater emphasis on acute than on chronic studies. To deal with this, a database was developed with data from studies where both acute and chronic effects were reported. In general, chronic effect concentrations were lower than acute effect concentrations, as expected, but there were no consistent changes in the ratio with organism or element. As a result, a median chronic:acute effect concentration ratio of 0.35 was calculated and used to convert pertinent acute effect concentrations to chronic effect concentrations. In deriving CTVs, preference was given to results of chronic studies, but results from acute studies were used when they had effect concentrations more than 2.8-fold (2.8 is the inverse of 0.35) lower than the similar chronic studies.
Case-by-case arguments were developed and documented to support the selection of the data that would become the CTV (Bird et al., 1999). Often, pH and water hardness were used as criteria to select data. Data from studies using low pH (5.5-7.0) and water hardness (nominally below 10 mg/L) were preferred, since these match those typically encountered on the Canadian Shield. Because aquatic ecotoxicology has more routine bioassays with specific species, the decision was made to average the effect concentrations of studies using similar species and conditions, where the effect concentrations were comparable. In almost all cases, there were sufficient data to define CTVs for each of the three assessment endpoints. All CTVs and ENEVs are reported as free-ion concentrations except for As, where CTVs are reported as concentrations of dissolved As.
1 Different percentile critical loads are based on probabilistic evaluation of metal transport and fate and assume that the ENEV is invariant.
Effects on aquatic organisms - CTV and ENEV values: Chronic ENEVs for aquatic organisms estimated for Cu, Zn, Ni, Pb, Cd and As are summarized in Table 32.
For Cu and aquatic invertebrates, the study by Giesey et al. (1983) used appropriate low-hardness water and reported sensitive effects.
They reported both computed and measured free-ion concentrations, but for consistency the geochemical models applied throughout this report were used to generate the free-ion effect concentrations. The resulting CTV, when adjusted with the chronic:acute ratio, was 0.98 mg/L. This is below the 95th percentile background free-ion concentration of 1.0 mg/L, so the background free-ion value is used for the CTV.
For Cu and fish, several acute exposure studies were equally relevant (Cusimano et al., 1986; Anadu et al., 1989; Welsh et al., 1996), and the average acute effect concentration was 1.9 mg/L. Converted to chronic with the ratio 0.35, this becomes 0.67 mg/L, which is below the 95th percentile background free-ion concentration. Thus, as for the invertebrates, the CTV for Cu is set to the background free-ion value of 1.0 mg/L.
- Values for As are expressed on a total dissolved basis. All others are on a free-ion basis.
- Due to the use of application factors of 1, ENEVs are equal to CTVs.
- CTVs are below the 95th percentile background concentration. Therefore, ENEVs were set equal to the 95th percentile background value.
For Cu and aquatic plants, only data for algae were found (Gachter et al., 1973; Stokes, 1981; Vavilin et al., 1995). They were equally relevant, and the average effect concentration was 2.6 mg/L. This is above the 95th percentile background free-ion concentration, so the free-ion CTV was set at 2.6 mg/L.
For Zn and aquatic invertebrates, the only study among those considered that used low-hardness water was that of Belanger and Cherry (1990), and so it was chosen for the CTV. An effect level of EC20 for reproduction after seven days was interpolated from their data, and this was at a free-ion concentration of 48 mg/L. This is four-fold above the 95th percentile background free-ion concentration of 12 mg/L, and so the CTV was based solely on the effects data.
For Zn and fish, several acute exposure studies were relevant. The most sensitive chronic study had higher effect concentrations than the acute studies, and so was not used. The three most sensitive acute studies (Cusimano et al., 1986; Bradley and Sprague, 1985; Anadu et al., 1989) used the same species, Oncorhynchus mykiss (rainbow trout), and were comparable, so the effect concentrations were averaged. Because they were acute studies, the factor of 0.35 was used, resulting in a CTV of 39 mg/L. This is three-fold above the 95th percentile free-ion background concentration, so the CTV was based solely on effects data.
For Zn and aquatic plants, only data for algae were found, and the effect concentrations of the two most sensitive studies (Stokes, 1981; Bartlett et al., 1974) were averaged to give a CTV of 45 mg/L. This is well above background, so the CTV was based solely on effects data.
For Ni and aquatic invertebrates, the most sensitive study using appropriate water hardness was that of van Frankenhuyzen and Geen (1987). They reported growth and survival of caddisfly (Clistoronia magnifica), and from their data an EC25 could be interpolated. The free-ion effect concentration, chosen as the CTV, was 35 mg/L. This is 20-fold above the 95th percentile free-ion background concentration of 1.8 mg/L, and so was used as the CTV.
For Ni and fish, sensitive data by Nebeker et al. (1985) for Oncorhynchus mykiss were chosen in preference to another species because data for this species were available for most of the elements considered in this assessment, allowing consideration of additive toxicity. Effect levels of EC25 for growth at 53 mg/L were used for the CTV, and this concentration is almost 30-fold above the 95th percentile free-ion background concentration.
For Ni and aquatic plants, only data for algae were found, and those of Stokes (1981) were used for the CTV. The EC25 was 18 m g/L, and because this is well above the background, it was taken as the CTV.
For Pb and aquatic invertebrates, only the study of Mackie (1989) both was sensitive and used low-hardness water. Several species were studied; the most sensitive was Hyalella azteca, where an acute effect level (LC50) was reported at 20 mg/L. With adjustment using the chronic:acute ratio of 0.35, the CTV becomes 6 mg/L. This is 10-fold above the 95th percentile free-ion background concentration of 0.64 mg/L, and so was used as the CTV.
For Pb and fish, the study of Davies et al. (1976) was most appropriate and used Oncorhynchus mykiss. A chronic EC25 for deformities in fry occurred at 18 mg/L, and because this is well above the 95th percentile free-ion background concentration, it was used as the CTV.
For Pb and aquatic plants, only data for algae were found, and those of Stokes (1981) were used. A chronic effect level (EC25) was observed at 39 mg/L, and this was used as the CTV.
For Cd and aquatic invertebrates, three studies (two species in Lawrence and Holoka, 1991; one species in Suedel et al., 1997) were comparable in methods and sensitivity, and so the effect concentrations were averaged. The exposures were chronic, and effect levels were no more than EC39. The average effect concentration, used as the CTV, was 0.18 mg/L. This is about two-fold above the 95th percentile free-ion background concentration of 0.084 mg/L, and so the CTV was based solely on effects data.
For Cd and fish, two studies using Oncorhynchus mykiss (Cusimano et al., 1986; Anadu et al., 1989) were comparable, so the average was used to derive the CTV. Both studies were of acute exposures, so the ratio 0.35 was used to estimate a chronic effect concentration of 0.25 mg/L. This is above the 95th percentile free-ion background concentration, and so was used as the CTV.
For Cd and aquatic plants, only data for algae were found, and two studies (Vocke et al., 1980; Stokes, 1981) used the same organisms and similar conditions. Thus, the effect concentrations were averaged to yield a free-ion CTV of 5.5 mg/L, which is well above background.
For As and aquatic invertebrates, the acute exposure study of Passino and Novak (1984), when adjusted using the chronic:acute ratio of 0.35, was more sensitive than the chronic studies found. The study used As(V) and Bosmina longirostris. After adjustment, the effect concentration was 300 mg/L (expressed as total dissolved As). This is well above the 95th percentile background concentration of 0.93 mg/L, and so the CTV was based solely on effects data.
For As and fish, the study of Birge et al. (1983a) is suitable. They used Oncorhynchus mykiss with chronic exposure to As(III), and an EC25 was interpolated. The CTV, based on total dissolved As, is 375 mg/L, well above background.
For As and aquatic plants, the effect concentrations in two studies (Vocke et al., 1980; Planas and Healey, 1978) were averaged, in part because of uncertainties in interpretation of the more sensitive study. The effect levels were EC42 and EC25, and the average total dissolved effect concentration, used as the CTV, was 21 mg/L. Both studies used As(V). This CTV is well above background.
In all cases for CTVs in aquatic systems, the application factor was set to unity because the measurement endpoints were considered to be quite applicable to the assessment endpoints and because, in some cases (e.g., for Cd and Zn), CTVs were not much above 95th percentile natural background values. Thus, in all cases, the CTVs become the ENEVs. The 95th percentile free-ion background concentration was used as the CTV (and ENEV) for Cu, for both aquatic invertebrates and fish. In all other cases, the CTVs (and ENEVs) were based on effects data. For each metal, the "primary" ENEV that was used in calculation of "primary" critical loads (below) and risk quotients (Section 3.0) was the lowest of the ENEVs estimated for invertebrates, fish and aquatic plants (Table 32). The choices for ENEVs, when compared on a similar dissolved-element basis, agreed very well with previous summaries (CCME, 1991; de Vries and Bakker, 1996). The agreement is not unexpected, since many of the same data were used in these summaries.
Effects on aquatic organisms - critical loads: As with the derivation of critical loads for soils, the first assumption in the aquatic environment is that only the soluble metal in the flux to the water surface need be considered. The water body is modelled as a mixing tank, with dilution water entering from the terrestrial catchment area. Loss of contaminant is by flushing downstream and burial in sediment. Transfer to sediment is modelled as a first-order process dependent on dissolved metal concentration and correlated to pH. The production of sediment by processes in the water column is independent of the rate of transfer of metal to sediment. It is assumed that once contaminants are buried by more than 10 cm of fresh sediment, they are effectively removed from the biotic environment. The transfer of contaminants to sediments is positively correlated to pH, so that in probabilistic analysis there is more transfer to the sediment if the water pH is higher.
The model parameters set for this assessment (Sheppard et al., 1999) were water body/lake area (median 1 x 105, range 1 x 104-7 x 107 m2), terrestrial catchment area (median 1 x 106, range 6 x 104-7 x 107 m2), water body/lake depth (median 4.7, range 0.7-27 m), sediment accumulation rate (median 0.17, range 0.012-2.6 kg/m2/a), thickness of new (biologically active) sediment (median 0.056, range 0.01-0.1 m), net precipitation (median 0.31, range 0.08-0.57 m/a), and water pH (median 6.2, range 5.5-7.0). The geometric mean, GSD and lower and upper bounds for the first-order rate constant for transfer to sediment, alpha (a-1), were specific for each element, but were all correlated to pH, with r=0.8. The geometric means (and GSDs) were Cu: 0.48 (4.0), Zn: 1.3 (3.7), Ni: 0.24 (7.5), Pb: 1.9 (3.7), Cd: 2.0 (3.7) and As: 1.5 (6.9).
The surface water model reaches steady state within a few years, so that considering results at steady state is not a difficult assumption. The model was run with an input of a unit flux density of soluble metal to the water surface, and concentrations of free ion in the water column were output. As with the soil model, the surface water model is linear with respect to concentration and can be scaled with flux or concentration. Critical loads were defined as the flux density to the water surface that will increase the existing, background free-ion concentration to the concentration of the ENEV.
The transfer of contaminants from the terrestrial catchment to the water body is difficult to model. In large part, this is related to the issue of time to steady state. With a constant flux of metal only to the surface of the water, steady state occurs in less than five years. But as discussed earlier, times to steady state for the top few centimetres of soil were up to several centuries. Thus, steady state for a watershed will be much longer, perhaps of the order of 104 years. Furthermore, high historic emissions have resulted in the accumulation of metals in the catchment areas of lakes located near some metal-processing facilities. Therefore, metal concentrations in these lakes would have to decrease to steady-state levels such as those estimated here.
It would not be reasonable to use the steady-state assumption for a watershed, and a decision to pick a specific time would affect how soil contamination is modelled. The delivery of contaminant from the terrestrial watershed to the water body is often parameterized as a delivery ratio. In theory, when atmospheric contamination is just beginning, the delivery ratio is near zero. Much of the contamination that deposited in the terrestrial environment is initially retained there. When the watershed is at true steady state, the delivery ratio is 1 by definition - the efflux from the terrestrial catchment area equals the influx. Thus, modelling the transfer of contaminant from the terrestrial catchment to the water body is very time-dependent. In addition, it is very site-specific and difficult to deal with generically. Here, it is assumed that there is no transfer from the terrestrial watershed, and in a separate calculation (Sheppard et al., 1999, Appendix) it is found that the potential underestimation of lake water concentration, assuming a delivery ratio of 0.25, is about fivefold.
As with the soil critical loads, deterministic runs using median values of all the parameters define "median" critical loads, and the probabilistic runs were used to define the deviations from the median case. Deviations from median critical loads result from variations in background free-ion concentrations and in the selected model parameter values. The ENEVs were assumed to be invariant. Variations in background and model parameters primarily represent spatial variability, because it was assumed that receiving environments could be anywhere on the Shield.
The 25th and 10th percentile critical loads were computed in addition to the median (50th percentile) critical loads. The interpretation of a 10th percentile critical load, for example, is that at that flux density, 10% of lakes on the Shield would have steady-state concentrations higher than the ENEV. Non-lethal effects (of the order of 20% reduction in performance) may be experienced by aquatic organisms in this 10% of lakes. The calculated critical loads are shown in Table 33.
11 As indicated, because application factors have been set to unity, ENEVs are equal to CTVs. It should be clearly recognized, however, that unlike typical ENEVs, these thresholds are levels that are known to cause low effects on sensitive organisms.
