SEARCH

SEARCH BY CITATION

Keywords:

  • Whole effluent toxicity;
  • Industry;
  • Effluent;
  • Toxicity;
  • Priority substances

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

The purpose of this study was to examine broad-scale correlation between presence of priority substances and whole effluent toxicity (WET) across a range of industry types. Using regression analysis, we examined how chemical-based inferred toxicity predicted measured WET of the effluents. Whole effluent toxicity was determined using a suite of acute and chronic bioassays; chemical-based toxicity was inferred from concentrations of priority chemicals and from published chemical toxicity values. When inferred toxicity was corrected for bioavailable metal and ion concentrations, 43% of the variability in measured toxicity was explained. For many industries, priority contaminants accounted for WET, and their toxic action was generally additive. However, industry-specific analysis of the residuals highlighted effluent types for which there was over one order of magnitude variation in inferred and measured toxicity. In particular, chemical-based assessments tended to overestimate toxicity of effluents containing high concentrations of metals and to underestimate toxicity of pulp mill effluents.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Regulators rely on two approaches to quantify the toxicity of industrial wastes and effluents for the purpose of setting acceptable discharge levels. The first approach is chemical-based: chemical analyses, coupled with toxicity tests, are used to identify the nature and toxicity of putative substances in liquid waste (Table 1). These tools permit regulatory agencies to establish numeric water quality criteria on a chemical-by-chemical basis [1,2]. Chemical-based toxicity assessments enabled scientists to link individual compounds to damaging ecological effects [3,4]. As a result, priority substances lists were developed to restrict or eliminate discharge of toxic chemicals into the environment [5]. However, this approach relies on the knowledge of what chemicals are present in an effluent and their potential toxicity; industrial effluents are generally complex and poorly characterized mixtures of a large number of chemicals.

The second approach, whole effluent toxicity (WET) testing, is an integrative tool that measures the toxic effect of an effluent mixture as a whole and accounts for uncharacterized sources of toxicity and for toxic interactions (Table 1). For this reason, in 1984, the U.S. Environmental Protection Agency (U.S. EPA) recommended an “integrated strategy” of both biological testing and chemical analyses to achieve and maintain water quality standards [6]. To further this goal, all states implemented water quality standards consisting of both numeric chemical-specific criteria and narrative “free from toxics in toxic amounts” criteria (via WET testing) [7]. Regulations for environmental effects monitoring of pulp and paper effluents in Canada have also incorporated effluent guidelines based on WET testing [8].

In 1988, Environment Canada undertook a 5-year study to quantify and regulate toxicity of industrial effluents discharged into the St. Lawrence River [9]. A program was designed to meet the integrated needs of toxicity testing advanced by the U.S. EPA. Effluents from the 45 most-polluting industries in Quebec were assessed for toxicity using both chemical analyses and WET tests. Effluent samples were obtained from a broad spectrum of industries, including chemical, petrochemical, and metallurgical facilities.

Few studies have examined the relationship between chemical assessments based on priority substances and whole effluent toxicity. Existing studies have focused mainly on single effluents or on a single industry [10–13]. The data from this study provide a unique opportunity to examine broad-scale patterns in toxicity assessment across a large number of industries and a range of industrial activity.

The purpose of this study was to compare these chemistry-based toxicity assessments and WET results in search of industry-specific trends in toxicity. Furthermore, comparing inferred toxicity, which is estimated from individual chemicals, to measured WET permitted an evaluation of the extent to which priority substances accounted for WET in complex effluents.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Sample collection

Effluent samples were collected from 45 industries located along the shores of the St. Lawrence and Saguenay Rivers in Canada. These industries are recognized by Environment Canada and Environment Ministry of the province of Quebec as being the top industrial polluters [14] (Fig. 1 and Table 2). The samples comprise a range of industrial categories, including inorganic and organic chemical production, petroleum refining, metal extraction and refining (Al, Cu, Ti, Cr, Zn, Ag, Au, Pt, and Pd), metal surface treatment, and pulp and paper mill activity (Table 2).

Table Table 1.. Strengths and limitations of chemical-based and whole effluent toxicity approaches to toxicity assessmenta
Assessment approachStrengthsLimitations
  1. a Adapted from [7].

Chemical-specificMore complete toxicologyDoes not consider unknown/uncharac-terized substances
 Treatment systems designed for chemical-specific requirements 
  Bioavailability is not measured
  Cannot account for interactions in a mixture
 Fate can be modeled 
 Less expensive than biological measurements 
  Cannot directly assess biological impairment
 Can predict impacts 
Whole effluent toxicityAccounts for un-known/uncharacteri-zed chemicalsIncomplete toxicology (few species tested)
 Evaluates aggregate toxicity (interactions)Identification and treatment of toxicants difficult
 Toxicity is directly measuredBioaccumulation and persistence not measured
 Bioavailability is assessed 
  Effects of ambient conditions on toxic-ity not considered
 Can predict impacts 

Whole effluent samples were collected over three consecutive days using an automatic sampler (Manning Environmental, Santa Cruz, CA, USA) that collects 375 ml of wastewater every 15 min for 24 h, for a total of 40 L per day. For plants with more than one final effluent, composite samples were taken in proportion to the discharge of each effluent stream. During the sampling period, final effluent flow was measured on-site with a Parshall flume (Nortech Control Equipment, Laval, QC, Canada). All samples were kept at 4°C during sampling, transport, and storage, according to guidelines outlined by Vezeau [15].

Sample processing

Bioassays. Four single-species toxicity tests were used to evaluate WET: the Microtox™ (AZUR Environmental, Carlsbad, CA, USA) bacterial luminescence assay, a microalgal growth inhibition assay, and a chronic survival and reproduction assay using a freshwater cladoceran (Table 3). The toxicity tests were chosen to fulfill criteria that included sensitivity, variation in biological complexity and feeding ecology, variation in assessment endpoints (acute, chronic, and sublethal), cost, rapidity, and replicability (Table 3). Results of the toxicity tests were combined to give a mean toxic score for each effluent sample according to the following equation, modified from Costan et al. [9]:

  • equation image(1)

where n = number of positive toxic responses (significantly different from control responses), N = number of bioassays used, and T = toxic units or 1/MATC (maximum allowable toxicant concentration). The MATC is calculated as the geometric mean of the lowest observable effect of concentration and the no observable effect of concentration endpoints [16].

thumbnail image

Figure Fig. 1.. Map of the St. Lawrence and Saguenay Rivers, Quebec, Canada. Letters denote locations of facilities whose effluents were sampled for this study.

Download figure to PowerPoint

Table Table 2.. Industrial effluent samplesa collected from the St. Lawrence River and their principal products and effluent discharges; facilities are defined by industrial activity
IndustrybPrincipal productsDischarge (103 m3d−1)WETcTEdMape
  1. a Samples were final effluents, based on 2- to 5-day sampling periods, and may be 24-h integrated final effluents or 24-h integrated composite effluents when no single final effluent existed.

  2. b Inorganic = inorganic chemical production facilities; surface = surface treatment facilities; metal = metallurgical industries; petrol = petrochemical refining industries; organic = organic chemical production facilities; paper = pulp and/or paper facilities.

  3. c WET = whole effluent toxicity or the toxicity of the mixture measured as a whole, expressed as log MATC−1 (maximum allowable toxicant concentration).

  4. d TE = toxic equivalents or the sum of the constituent chemical toxicities, expressed as log (concentration·EC50−1).

  5. e Indicates location of facilities on the map shown in Figure 1.

Inorganic 1Titanium oxide512.71.7J
Inorganic 2Explosives and propulsives220.6−0.3A
Inorganic 3Chlorine and alkali141.41.2D
Inorganic 4Titanium dioxide73.22.3L
Inorganic 5Niobium oxide61.60.6W
Inorganic 6Chlorine and alkali51.20.9M
Inorganic 7Elemental phosphorus11.81.7J
Surface 1High-performance engine parts51.30.4G
Surface 2Hydraulic systems assemblage0.11.30.3G
Surface 3Galvanized steel0.070.7−0.3E
Metal 1Zn and Cd ingots1821.80.8C
Metal 2Steel shot, cast iron, Ti slag1141.41.4L
Metal 3Stainless steel650.5−0.1L
Metal 4Al, anodes and cathodes for Al production650.70.7V
Metal 5Stainless steel461.51.5K
Metal 6Refined Cu, Ag, Au, Pt, and Pd250.4−0.5I
Metal 7Al ingots161.81.8Y
Metal 8Al ingots131.50.5X
Metal 9Al ingots61.41.1D
Metal 10Al ingots210.4M
Metal 11Al foil0.91.81.3N
Metal 12Al ingots0.31.80.9U
Organic 1Industrial adhesives and plastic resins5121.7H
Organic 2High-density polyethylene, specialty chemicals251.40.5I
Organic 3Ethanol91.21.1J
Organic 4Polyvinyl acetate20.2−0.2J
Petrol 1Refined petroleum products4621.3R
Petrol 2Refined petroleum products521.8I
Petrol 3Refined petroleum products20.2−0.3J
Petrol 4Refined petroleum products21.71.7I
Paper 1Newsprint1701.71.1Y
Paper 2Specialty paper products1441.81.4N
Paper 3Newsprint, kraft paper products1441.70.8P
Paper 4Newsprint, specialty paper products1032.81.8N
Paper 5Newsprint832.21.5X
Paper 6Newsprint, specialty paper products751.80.9O
Paper 7Newsprint622.72.6U
Paper 8Newsprint5021.6S
Paper 9Specialty paper products391.91.3Q
Paper 10Kraft paper products162.61.8V
Paper 11Fine paper products91.40.5D
Paper 12Newsprint92.42.1T
Paper 13Specialty paper products51.51V
Paper 14Hygienic paper products410.9E
Paper 15Kraft paper, newsprint121.5N
Table Table 3.. Suite of toxicity tests used to measure whole effluent toxicity
Test type (Organism)Endpoint and effectExposure timeReference
Microtox™ bacterial assay (Vibrio fischeri)Acute sublethal light inhibition15 min[38]
Microalgal assay (Selenastrum capricornutum)Chronic growth inhibition96 h[17]
Crustacean assay (Ceriodaphnia dubia)Chronic lethality7d[39]
Crustacean assay (C. dubia)Chronic reproduction inhibition7d[39]

Certain assays, such as the algal growth assay using Se-lanastrum capricornutum, require a sterile medium for algal growth [17], and the presence of suspended particulate matter may confound results. Therefore, effluents were filtered through a 0.45-μm Millipore™ filter prior to all toxicity analyses in order to remove suspended particulate matter.

Chemical analyses. Inferred toxicity was based on physical-chemical characterizations of the effluent [18]. A list of more than 120 priority toxic substances was established by the St. Lawrence Action Plan, guided by existing priority lists from Canadian [5] and American [6] federal environmental agencies. For each industry, pertinent chemical parameters for analysis were chosen from this priority list. Conventional parameters such as dissolved organic carbon; total organic carbon; biological oxygen demand; pH; suspended sediment; and concentrations of SO42−, PO43−, NO3, NH3, and CN were also measured. Detailed analysis methods for all chemicals and parameters are reported in Legault and Villeneuve [18].

Inferred toxicity of an effluent was expressed in toxic units. Each priority contaminant was assigned a toxic weighting factor, Ftox (Table 4); Ftox was calculated as the inverse of its mean EC50 (concentration causing an effect in 50% of the organisms tested) such that:

  • equation image(2)

EC50 values were obtained from the Aquatic Toxicity Information Retrieval Database [19]. Mean EC50 values were calculated from toxicity data encompassing a range of endpoints and test organisms and were chosen to reflect the suite of bioassays used to measure whole effluent toxicity.

A toxic unit value (TU) for a chemical i was then calculated as follows:

  • equation image(3)

where Ftoxi is the weighting factor assigned to chemical i. The inferred toxicity of an effluent is the sum of the TUs of its constituent chemicals. This numerical score can then be used to compare relative toxicity of different industries and industrial sectors, as well as to identify the predominant toxicants in an effluent.

Correction of chemical analyses for suspended particulates. No filtration was performed on samples prior to chemical analyses. To account for this discrepancy with bioassay methodology, chemical concentrations were corrected for particle-bound contaminants. Many chemical compounds, because of their molecular structure, are not highly water soluble and display an affinity with suspended particulate matter. The degree of association between a solute (contaminant) and a sor-bent (particulate matter) is indicated by the sorption partition coefficient (KD) of the solute. Sorption partition coefficient values for hydrophobic organic contaminants have been shown to be correlated to their octanol-water partition coefficients (KOW) and to the concentration and organic content of the suspended particulate matter [20]. The concentration of a contaminant (mg/L) that would be adsorbed to suspended particulate matter in the effluent was calculated using an equation modified from White et al. [21]:

  • equation image(4)

where KD is the sediment partition coefficient for the given contaminant, [SS] is the suspended sediment concentration, and 10−6 is a unit conversion factor. For example, pyrene has a KD of 10414. Organic effluent 2 contains elevated amounts of this organic compound and 42 mg/L of suspended particulate matter. For every 1 ppm (mg/L) of pyrene in solution, one can estimate {104.14 mg/kg·42 mg/L·106 kg/mg} = 34.6 ppm to be associated with suspended particulate matter in the effluent. Therefore, in each liter of organic effluent 2, 34.6 out of 35.6 mg, or 97%, of pyrene will be associated with suspended particulate matter and will subsequently be removed with particle filtration. Toxic units for all effluent samples were recalculated with particulate-corrected contaminant concentrations to reflect aqueous concentrations of contaminants in the sample.

Table Table 4.. List of inorganic and organic chemicals targeted as priorities for concern by the Quebec Ministry of Environmenta
Chemicals and groupsWeighting factorsb
  1. a Adapted from [40].

  2. b Weighting factors were calculated as the inverse of the mean toxicity, expressed as an EC50, for a given chemical. Toxicity data was collected from the EPA Aquatic Information Retrieval database [19]. Bioassays were chosen to reflect the assessment endpoints employed in whole effluent toxicity assessment.

Inorganic substances 
 Heavy metals 
  Antimony0
  Silver1,000
  Arsenic2
  Beryllium8
  Cadmium90
  Chromium0.4
  Copper30
  Mercury1.3 E4
  Nickel1
  Lead10
  Selenium0
  Thallium9
  Vanadium70
  Zinc5
 Other metals 
  Aluminum0.3
  Iron0.3
  Manganese0.1
  Molybdenum0.1
 Anions and others 
  Ammonia0.1
  Chlorine0
  Cyanide20
  Nitrites/nitrates0.5
  Phosphorus30
  Sulfurs500
  Tetrachloroacetaldehyde0
  1,1,2,2-Tetrachloroethane0
  Tetrachloroethylene0.1
  Carbon tetrachloride0
  trans 1,2-Dichloroethylene0.1
  trans 1,3-Dichlropropene0
  1,2,4-Trichlorobenzene0.1
  1,1,1-Trichlroethane0
  1,1,2-Trichloroethane0
  Trichloroethylene0
  Trichlorofluoromethane0
 Dioxins and furans 
  2,3,7,8-T4CDD equivalents1 E5
 PAHs 
  Acenaphthene0.1
  Acenaphthylene4
  Anthracene2
  Benzo[a]anthracene7
  Benzo[b]fluoranthene7
  Benzo[k]fluoranthene70
  Benzo[ghi]perylene5,000
  Benzo[a]pyrene200
  2-Chloronaphthalene0
  Chrysene0
  Dibenzo[ah]anthracene1.4 E4
  Fluoranthene0
Organic substances 
 Fatty acids 
  Linoleic2
  Linolenic2
  Oleic2
  Palmitic2
  Palmitoleic2
  Dichlorostearic2
  Stearic2
 Resin acids 
  Abeitic2
  Chlorodehydroabeitic2
  Dehydroabeitic2
  Dichlorodehydroabeitic2
  Isopimaric2
  Levopimaric2
  Neoabeitic2
  Palustric2
  Pimaric2
  Sandaracopimaric2
 PCBs 
  Total PCB1,000
  PCB-10161,000
  PCB-12211,000
  PCB-12321,000
  PCB-12421,000
  PCB-12481,000
  PCB-12541,000
  PCB-12601,000
  Fluorene0.5
  Indeno(1,2,3-cd)pyrene0
  2-Methylnaphthalene0.1
  Naphthalene0.1
  Phenanthrene2
  Pyrene0
 Oils and greases 
  Total oils/greases1
  Mineral oils/greases1
 Nonchlorinated phenols 
  Catechol0.1
  Cresols (o,m,p)0
  2,4-Dimethylphenol0.6
  4,6-Dinitro-o-cresol20
  2-4-Dinitrophenol0.1
  Eugenol20
  Guaiacol20
  Hydroxyphenol0
  Isoeugenol0
  3-Methyl-4-6-dinitrophenol0
  2-Nitrophenol0
  4-nitrophenol0.2
  Phenol0
 Chlorinated phenols 
  p-Chloro-m-cresol1,000
  2-Chlorophenol0
  4-Chloro-3-methylphenol6
  6-Chlorovanille0
  4,5-Dichlorocatechol0.5
 Nonhalogenated volatile organics 
  Acetone0
  Acroleine0.4
  Acrylonitrile0
  Benzene0
  Butylcyclooctane0
  2,4-Dinitrotoluene0.1
  Ethyl ether0
  Ethylbenzene0.1
  Ethylmethylcyclohexane0
  Isopropanol0
  Mesitylene0
  Methylcyclohexane0.1
  Nitrobenzene0
  Styrene0.1
  Toluene0.1
  Xylenes (o,m,p)0
 Halogenated volatile organics 
  bis-Chloromethyl ether0.8
  Bromodichloromethane0
  Bromoform0
  Bromomethane0
  Chlorobenzene0
  Chlorodibromomethane0
  Chloroethane0
  Chloroethylene0.1
  2-Chloroethyl vinyl ether1
  Chloroform0
  Chloromethane0
  4,5-Dichloroguaiacol0
  2,4-Dichlorophenol0.3
  5,6-Dichlorovanille0
  Monochlorophenols0
  Pentachlorophenol3
  Tetrachorocatechol0.4
  Tetrachloroguaiacol1
  2,3,4,6-Tetrachlorophenol1
  3,4,5-Trichlorocatechol1
  3,4,5-Trichloroguaiacol2
  4,5,6-Trichloroguaiacol0.1
  2,4,6-Trichlrophenol0.2
  Trichlorophenols0.1
  Trichlorosyringol0
 Phthalates 
  Butyl benzylphthalate0.1
  bis-(2-Ethylhexyl)phthalate170
  Di-n-butylphthalate0.2
  Diethylphthalate0
  Dimethylphthalate0.1
  Di-n-octylphthalate500
  Total phthalates2
 Soluble volatile organics 
  Aniline0
  Anthraquinone0
  Benzidine0.1
  bis-(2-Chloroethoxy)methane20
  bis-(2-Chloroethyl) ether1
  bis-(2-Chloroisopropyl) ether1
  1,2-cis-Dichloroethylene0
  1,3-cis-Dichloropropene0
  1,2-Dichlorobenzene0
  1,3-Dichlorobenzene0
  1,4-Dichlorobenzene0
  Dichlorodifluoromethane0
  1,1-Dichloroethane0
  1,2-Dichloroethane0
  1,1-Dichloroethylene0
  Dichloromethane0
  1,2-Dichloropropane0
  1,2-Dichloropropene0
  Hexachlorobenzene0
  Hexachloroethane0.2
  4-Bromophenyl phenyl ether0.4
  4-Chloropehnyl phenyl ether1
  3,3-Dichlorobenzidine20
  1,2-Diphenylhydrazine0.4
  2-Ethylhexanol0.1
  Hexachlorobutadiene10
  Hexachlorocyclopentadiene7
  Isophorone0
  3-Nitroaniline0
  Nitroso-n-dimethylamine0.2
  Nitroso-n-diphenylamine0.2
  Nitroso-n-di-n-propylamide0

Data treatment and analysis

Modeling of free ion activity. Total metal concentrations were analyzed by acid digestion and atomic absorption spectroscopy as detailed in Standard Methods [22]: for As and Se, according to methods 3030, 3113, or 3114; for all remaining metals, according to methods 3030, 3111, or 3120. Free ion activity of trace metals and anions was determined using MI-NEQL [23], a computer program that can model free ion activity when given the total metal and anion concentration and chemical data of the effluent. Only those effluents in which metals and anions contributed greater than 15% of the inferred toxicity of the effluent were corrected for free ion activity. Previous trials on effluents in which metals and anions contributed less than 15% of the inferred toxicity showed that changes following correction were minute. Because the dissolved organic carbon in the effluent samples was uncharac-terized, it was assumed to consist entirely of fulvic acids. Fulvic acids are common components of humic substances, which remain soluble over the entire pH range and are considerably more stable than humic acids [24]. For each effluent, molarities of all ligands were calculated. Ionic strength was calculated as the product of the molarity of an ion and the square of its charge. Sample hardness values were not available; therefore, values were estimated from the receiving water hardness in the zone of the St. Lawrence system where the industry was located. Values fell into one of three categories: south shore hard water (≈30 mg/L CaCO4), north shore soft water ≈11 mg/L CaCO4), and upstream Saguenay water (≈2 mg/L CaCO4).

Least-squares linear regression analysis was used to determine how well inferred toxicity predicted WET. All data were log transformed and conformed to the assumptions of normality and homoscedasticity of residuals.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Regression analysis of WET plotted against inferred toxicity (Fig. 2A) revealed a significant relationship between the two toxicity measures (y = 0.87x + 0.34, r2 = 0.25, p < 0.01; Fig. 2A). Correction for particulate-bound contaminants decreased inferred toxicity values by less than a factor of 2.

The intercept of 0.34 was probably due to differences between the measurement endpoints used to calculate inferred toxicity and those used to calculate WET. Whole effluent toxicity measurement endpoints were expressed as MATC. Measurement endpoints obtained from the literature were reported as EC50 (concentration eliciting an observable effect on 50% of the test population), which, by definition, will always be a higher concentration than the MATC. This discrepancy drove the regression line through a higher intercept (i.e., 0.34 rather than 0) than if both measurement endpoints had been the same. Previous studies on the relationship between EC50 (acute) values and no observed effect concentration (chronic) values have shown that there is generally about a 10-fold difference between the two and that for 95% of all organic industrial chemicals studied, acute and chronic concentrations will differ by less than two orders of magnitude [25,26].

thumbnail image

Figure Fig. 2.. Regression analysis and plot of the relationship between whole effluent toxicity (WET) and inferred toxicity (toxic units), (A) prior to free ion activity correction and (B) following free ion correction with MINEQL. Inferred toxicity was calculated according to Equation 3 in the text. Whole effluent toxicity was calculated according to Equation 1. Plot symbols: • = petroleum facilities, ○ = inorganic chemical production facilities, ▾ = metallurgical facilities, ▿ = organic chemical production facilities, and ▪ = pulp and paper facilities. SEE = standard error of the estimate. Dotted arrow in Figure 2B indicates the change in inferred toxicity of inorganic facilities 1 and 4 following adjustment for toxicity of H+ ions (see text).

Download figure to PowerPoint

The relationship between WET and inferred toxicity was strengthened when bioavailable metal and ion concentrations were used (r2 = 0.43, p < 0.001; Fig. 2B). Inferred toxicity of effluents from inorganic chemical and metallurgical facilities, as well as inferred toxicity of some petrochemical and organic effluents, decreased after correction for free ion activity (Table 5). These effluents were characterized by high toxic contributions from metals (>35% of total inferred toxicity value) or from anions such as CN, PO43−, NO3, and SO42−, a significant fraction of which were likely complexed by trace metals or Ca, making them unavailable.

Most of the effluents with greater-than-predicted WET values were from pulp and paper facilities, whereas many of the samples with less-than-predicted WET values were from metallurgical industries and industries with high metal concentrations in their final effluents (Figure 2B). Industry-specific analysis of mean residual values reflected this observation (Fig. 3). A residual is the error (difference) between the actual y value and the estimated mean calculated in the regression; a positive residual therefore denoted that WET was greater than inferred toxicity for a given effluent, whereas a negative residual denoted the opposite result.

Table Table 5.. Decrease in inferred toxicity (IT) of metal- and ion-rich effluents after correction for metal bioavailability using MINEQL programa
IndustryTotal metals (mg/L)Bioavailable metals (mg/L)DOC (μg/L)IT beforebIT afterc% Change in IT
  1. a [23].

  2. b Inferred toxicity before correction for metal and ion bioavailability.

  3. c Inferred toxicity after correction for metal and ion bioavailability.

Petrochemical 330.111200.590
Metallurgical 31010.112170
Metallurgical 610.20.150.367
Inorganic 20.20550.464
Inorganic 7201054 E45061
Surface treatment 3500.2300.460
Surface treatment 210160253
Metallurgical 4200.10.220552
Metallurgical 230102.2903038
Metallurgical 810120334
Surface treatment 110.159230
Metallurgical 1201070728
Organic 2100.10.110325
Petrochemical 410.110605021
Metallurgical 550.20.1303017
Inorganic 125014017205016
Inorganic 5414949
Petrochemical 240.4070604
Inorganic 456051014201803

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Initial correspondence between WET and inferred toxicity

A significant relationship was detected between WET and inferred toxicity (Fig. 2A). Correction for contaminants adsorbed to particulate matter did not alter inferred toxicity, although other research has suggested that a considerable proportion of organic contaminants may be associated with suspended particulates [21]. For some effluents, this may have been due to the presence of highly soluble, polar organic contaminants, such as those typically discharged in pulp and paper facilities [27]. In such cases, most of the toxicity would be associated with the aqueous fraction of the mixture and not with the suspended particulates. Other factors, such as consistency in measurement endpoints and free ion activity and the ability of chemical-specific analyses to target putative contaminants, would likely be more important in structuring the relationship between inferred toxicity and WET for these samples.

Free ion correction of inferred toxicity

Correction of inferred toxicity for bioavailable metals and ions strengthened the relationship between WET and inferred toxicity, particularly in effluents from metallurgical and inorganic chemical production facilities (Fig. 2B). Toxicity predictions for trace metals are most meaningful when toxicity is evaluated using metal ion, not total metal, concentrations, because neither toxicity nor bioavailability is directly related to total metal concentrations [28]. The extent of the metal-particulate association depends on the concentration and chemistry of the particulate matter, water chemistry, temperature, and the presence of ligands and other metals [28]. It is possible to model the distribution of species of a metal and to predict the free ion activity from a computer program that employs principles of equilibrium distribution and mass conservation. However, predicting the free ion concentration of trace metals requires detailed and accurate chemical information. In this study, it was necessary to make several assumptions about the chemistry of the effluent samples and the receiving environment. Dissolved organic carbon was not chemically characterized and was assumed to be 100% fulvic acid. Calcium carbonate, the principal inorganic ligand found in freshwater, had to be estimated from the alkalinity of the receiving water (presumably used as process water). As well, colloidal matter was not characterized in this study. Results in Table 5 show that free ion activity correction did alter the concentration of bioavailable metals and, consequently, the inferred toxicity of the effluent. The percentage reduction in toxicity ranged from 3% to 90%. These results highlight the value of bioassays for trace metal toxicity studies, which measure the toxic effect of only the bioavailable fraction of metals and ions.

thumbnail image

Figure Fig. 3.. Plot of mean residual whole effluent toxicity (+ SE) by category of industrial activity. A positive residual value indicates that whole effluent toxicity (WET) is greater than inferred toxicity; a negative value indicates that inferred toxicity was greater than WET.

Download figure to PowerPoint

After methodological differences were accounted for, there was a strong correspondence between the toxicity of effluents and their chemical constituents (r2 = 0.43, SEE = 1.0; Fig. 2B). An average of 43% of WET could be accounted for by the presence of priority contaminants. These results are consistent with the assumption of additivity of toxicity, which implies that the sum toxicity of individual chemicals in a mixture is more or less equivalent to toxicity of the mixture as a whole.

Residual variation in correlation between inferred toxicity and WET

There were also groups of exceptional effluents whose WET did not correspond to toxicity estimates based on the presence and concentration of priority contaminants. The average toxic score (log 1/MATC) for metallurgical effluents was 0.8 units lower than predicted (residual value = —0.8), and the average for pulp and paper facilities was 0.7 higher than predicted (Fig. 3). Ranges in residuals for pulp and paper and for metallurgical effluents did not overlap with the other classes of effluents (Fig. 3).

Metallurgical facility effluents challenge the assumption of additivity of toxicity. Aluminum refineries were generally characterized by concentrations of trace metal ions that exceeded chronic and sometimes acute toxicity thresholds, particularly for Al2+, Al3+, Cu2+, Hg2+, Pb2+, Fe2+, Fe3+, Ni2+, and Zn2+. For the majority of metallurgical effluents, WET was at least one order of magnitude lower than predicted. This discrepancy may be potentially explained by negative interactions among metals and between metals and ligands. Negative interactions have been known to occur between Zn2+ and Cd2+ [29], Co2+ and Zn2+ [29], Cu2+ and Zn2+ [30], Cu2+ and Mn2+ [31], Cu2+ and Fe3+ [32], and Cd2+ and Fe3+ [33], all of which are found in these effluents.

Only one effluent with less-than-predicted toxicity, from organic facility 3, was not dominated by trace metals but contained >5 mg/L of phthalates, which contributed to 99% of the inferred toxicity of that sample. Variation in this sample was probably due to overestimation of phthalate toxicity, because phthalates in the effluent were uncharacterized and were assumed to be a highly toxic species. Phthalate toxicity can vary over several orders of magnitude, depending on the species [7].

There were several effluents whose WET was more than one order of magnitude higher than predicted (Fig. 2B). In the case of two inorganic chemical production facilities, 1 and 4, the discrepancy in toxicity values between the two estimates was due to the extremely low pH of the samples. With pH values of 0.5 and 1.5, respectively, any toxic effects of the effluent were most likely attributable to the toxicity of high H+ concentration. Laboratory studies on H+ toxicity have demonstrated chronic toxic effects on mature larvae of aquatic insects at pH ranges of 2.5 to 5.4 [34] and on algal species at a pH level of 3 to 4 [35]. However, H+ is not considered a priority contaminant, and therefore any toxic effect due to elevated concentrations of H+ would not have been considered in the calculation of inferred toxicity. When toxic contributions from hydrogen ion were added, inferred toxicity values increased by a factor of 6 for inorganic facility 1 and by a factor of 10 for inorganic facility 4. The direction and magnitude of change in inferred toxicity for these facilities is illustrated by the arrows in Fig. 2B. Adjustment of the inferred toxicity scores did not alter the predictive power of the regression equation. Although the lack of correspondence between inferred toxicity and WET for inorganic facilities 1 and 4 was due to a methodological detail and was easily corrected, the discrepancy nevertheless highlights the potential pitfalls in treating complex mixtures as a collection of priority chemicals.

The third effluent sample whose WET was greater than predicted came from paper facility 15, a kraft mill facility. This positive residual value is consistent with reports in the literature that toxicity of kraft mill effluents often cannot be accounted for by the presence of known organic chemicals [11,12] and that many chlorinated compounds in kraft mill effluents remain unidentified [36].

In the case of the remaining positive outliers, there is no obvious reason why the mixtures were more toxic than predicted. Most industrial organic chemicals act by simple chemical narcosis and are additive, by molecular volume, in their joint acute toxicity [37]. However, when mechanisms of toxic action are more specific, or when there is more than one mode of toxic action (as is the case with most trace metals), one can assume neither that joint toxic action of a mixture will be strictly additive nor that inferred toxicity will necessarily reflect WET.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

We report that, in general, concentrations of priority contaminants accounted for observed WET and that their toxic action was more or less additive. However, analysis of patterns in residuals revealed interesting exceptions. Whole effluent toxicity of effluents high in metal concentrations was lower than predicted, possibly because of the discrepancy between total and bioavailable metals or because of interactions among metals. When measured toxicity was much higher than predicted, it was difficult to determine whether this was due to positive interactions among chemicals or to the presence of unknown chemicals. Such positive outlier industries would likely be good candidates for a toxicity identification and evaluation to identify the putative substances.

From the results of this study, we must consider that un-characterized toxic components in industrial effluents and interactions between chemicals, both synergistic and antagonistic, may have a significant effect on the toxicity of complex effluents. Finally, it is only through broad-scale, cross-industry investigations that such trends can be elucidated.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

The authors would like to thank the personnel at the St. Lawrence Center, particularly Raymond Vezeau and Richard Legault. We also gratefully acknowledge Klaus L.E. Kaiser, Maria T. Maldonato, and Chris Payne for data and Jake Vander Zanden, Adrian de Bruyn, Anthony Ricciardi, Richard Carignan, and Landis Hare for comments. This study was funded by a National Sciences and Engineering Research Council of Canada grant to J.B. Rasmussen.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References
  • 1
    Canadian Council of Ministers of the Environment. 1991. Canadian Water Quality Guidelines, Appendix IX—A Protocol for the Derivation of Water Quality Guidelines for the Protection of Aquatic Life. Ottawa, ON.
  • 2
    U.S. Environmental Protection Agency. 1986. Quality criteria for water. EPA 440/5–86–001. Final Report. Washington, DC.
  • 3
    WatrasCJ, HuckabeeJW, eds., 1994. Mercury Pollution: Integration and Synthesis. Lewis, Boca Raton, FL, USA.
  • 4
    Elliott JE, Norstrom RJ, Keith JA. 1988. Organochlorines and eggshell thinning in northern gannets (Sula bassanus) from Eastern Canada, 1968–1984. Environ Pollut 52: 81102.
  • 5
    Government of Canada. 1995. The second priority substances list. In Canada Gazette, Part 1—December 16, 1995. Queen's Printer for Canada, Ottawa, ON.
  • 6
    U.S. Environmental Protection Agency. 1984. Development of water quality–based permit limitations for toxic pollutants: National policy. Fed Reg 49: 90169019.
  • 7
    U.S. Environmental Protection Agency. 1991. Technical support document for water quality–based toxics control. EPA 505/2–90–001. Technical Report. Washington, DC.
  • 8
    Government of Canada. 1992. Pulp and paper effluent regulations. In Canada Gazette, Part II—May 20, 1992. Queen's Printer for Canada, Ottawa, ON.
  • 9
    Costan G, Bermingham N, Blaise C, Ferard JF. 1993. Potential ecotoxic effects probe (PEEP): A novel index to assess and compare the toxic potential of industrial effluents. Environ Toxicol Water Qual 8: 115140.
  • 10
    Bruynsteyn A. 1977. Determination of the practical value of toxic loadings. CPAR Report 588–1. Environmental Protection Service, Ottawa, ON, Canada.
  • 11
    Holmbom B, Lehtinen K-J. 1980. Acute toxicity to fish of kraft mill wastewaters. Paperi ja Puu-Papper o Trä 62: 673684.
  • 12
    Leach JM, Chung TK. 1980. Development of a chemical toxicity assay for pulp mill effluents. EPA-600/2–80–206. Technical Report. U.S. Environmental Protection Agency, Cincinnati, OH.
  • 13
    Andrén C, Eklund B, Gravenfors E, Kukulska Z, Tarkpea M. 1998. A multivariate biological and chemical characterization of industrial effluents connected to municipal sewage treatment plants. Environ Toxicol Chem 17: 228233.
  • 14
    Environment Canada. 1992. St. Lawrence action plan: 1991–92 annual report. Montreal, QC.
  • 15
    Vezeau R. 1982. Protocoles d'echantillonage, de preservation et de preparation des echantillons pour les analyses des polluants prioritaires. Environment Canada, Montreal, QC.
  • 16
    Horning WB, Weber CI. 1985. Methods for estimating the chronic toxicity of effluents and receiving waters to freshwater organisms. EPA-660/3–84–080. U.S. Environmental Protection Agency, Cincinnati, OH.
  • 17
    Environment Canada. 1992. Biological test method. Growth inhibition test using the freshwater alga Selanastrum capricornu-tum. EPS 1/RM/25. Technical Report. Ottawa, ON.
  • 18
    Legault G, Villeneuve M. 1993. Le Chimiotox: Resultats d'evaluation chimio-toxique des etablissements industriels du Plan d'action Saint-Laurent, Vol 1—Equipe d'intervention St-Laurent, Direction des services techniques, Montreal, QC, Canada.
  • 19
    U.S. Environmental Protection Agency. 1991. Aquatic Information Retrieval Database (AQUIRE). Duluth, MN.
  • 20
    Allan R. 1986. The role of particulate matter in the fate of contaminants in aquatic ecosystems. NWRI Contribution 84–16. Inland Waters Directorate, Burlington, ON, Canada.
  • 21
    White PA, Rasmussen JB, Blaise C. 1996. Sorption of organic genotoxins to particulate matter in industrial effluents. Environ Mol Mutagen 27: 140151.
  • 22
    American Public Health Association, American Water Works Association, and Water Pollution Control Federation. 1992. Standard Methods for the Examination of Water and Wastewater, 18th ed. American Public Health Association, New York, NY.
  • 23
    Westall JC, Zachary JL, Morel FMM. 1976. MINEQL: A Computer Program for the Calculation of Chemical Equilibrium Composition Aqueous Systems. Technical Note 18. Massachusetts Institute of Technology, Cambridge, MA, USA.
  • 24
    Stumm W, Morgan J. 1981. Aquatic Chemistry. Wiley-Intersci-ence, Toronto, ON, Canada.
  • 25
    Kenaga EE. 1982. Predictability of chronic toxicity from acute toxicity of chemicals in fish and aquatic invertebrates. Environ Toxicol Chem 1: 347358.
  • 26
    Slooff W, van Oers JAM, de Zwart D. 1986. Margins of uncertainty in ecotoxicological hazard assessment. Environ Toxicol Chem 5: 841852.
  • 27
    Environment Canada. 1987. Aquatic toxicity of pulp and paper mill effluent: A review. EPS 4/PF/1. Ottawa, ON.
  • 28
    O'Donnel JR, Kaplan BM, Allen HE. 1985. Bioavailability of Trace Metals in Natural Waters. American Society for Testing and Materials, Philadelphia, PA.
  • 29
    Price NM, Morel FMM. 1990. Cadmium and cobalt substitution for zinc in a marine diatom. Nature 344: 658660.
  • 30
    Rueter J, Morel FMM. 1981. The interaction between zinc deficiency and copper toxicity as it affects the salicilic acid uptake mechanism in Thalassiosira pseudonana. Limnol Oceanogr 26: 6367.
  • 31
    Sunda W, Huntsman S. 1983. Effect of competitive interactions between manganese and copper on cellular manganese and growth in estuarine species of the marine diatom Thalassiosira. Limnol Oceanogr 28: 924934.
  • 32
    Foster P, Morel FMM. 1982. Reversal of cadmium toxicity in a diatom: An interaction between cadmium activity and iron. Lim-nol Oceanogr 27: 745752.
  • 33
    Harrison G, Morel FMM. 1983. Antagonism between cadmium and iron in the marine diatom Thalassiosira weisflogii. J Phycol 19: 495507.
  • 34
    Bell HL. 1971. Effect of low pH on the survival and emergence of aquatic insects. Water Res 5: 313319.
  • 35
    Haines TA. 1981. Acidic precipitation and its consequences for aquatic ecosystems. Trans Am Fish Soc 110: 669707.
  • 36
    Environment Canada. 1993. Fraser River action plan—Chemistry and toxicity of three wastewaters. DOE FRAP 1993–08. North Vancouver, BC.
  • 37
    Broderius SJ. 1991. Modeling the Joint Toxicity of Xenobiotics to Aquatic Organisms: Basic Concepts and Approaches. American Society for Testing and Materials, Philadelphia, PA.
  • 38
    Environment Canada. 1992. Biological test method. Toxicity test using luminescent bacteria (Photobacterium phosphoreum). EPS 1/RM/24. Technical Report. Ottawa, ON.
  • 39
    Environment Canada. 1992. Biological test method. Reproduction and survival assay using the cladoceran Ceriodaphnia dubia. EPS 1/RM/21. Technical Report. Ottawa, ON.
  • 40
    Gouvernement du Québec. 1990. Critères sur la qualité de l'eau, Vol 1. Ministère de l'environnement et la faune, Québec, QC, Canada.