Guidelines for Canadian Drinking Water Quality: Guideline Technical Document - Bromoxynil

Proposed guideline

A maximum acceptable concentration (MAC) of 0.03 mg/L (30 µg/L) is proposed for bromoxynil in drinking water.

Executive summary

This guideline technical document was prepared in collaboration with the Federal-Provincial-Territorial Committee on Drinking Water and is based on assessments of bromoxynil completed by Health Canada's Pest Management Regulatory Agency and supporting documents.

Table of Contents

• Proposed guideline
• Executive summary
  • Exposure
  • Health effects
  • Analytical and treatment considerations
  • Application of the guideline
  • International considerations
• 1.0 Exposure Considerations
  • 1.1 Sources and uses
  • 1.2 Substance identity
  • 1.3 Exposure
• 2.0 Health Considerations
• 2.1 Kinetics
• 2.2 Health effects
• 2.3 Effects in humans
• 2.4 Effects in animals
• 2.5 Genotoxicity and carcinogenicity
• 2.6 Mode of action
• 2.7 Selected key study
• 3.0 Derivation of the health-based value
• 4.0 Analytical and Treatment Considerations
  • 4.1 Analytical methods to detect bromoxynil
  • 4.2 Treatment considerations
  • 4.2.1 Municipal-scale
  • 4.2.1.1 Conventional treatment
  • 4.2.1.2 Activated carbon adsorption
  • 4.2.1.3 Membrane filtration
  • 4.2.1.4 Oxidation
  • 4.2.2 Residential-scale
• 5.0 Management strategies
  • 5.1 Monitoring
• 6.0 International Considerations
• 7.0 Rationale
• 8.0 References
• Appendix A: List of acronyms
• Appendix B: Canadian water quality data

1.0 Exposure Considerations

1.1 Sources and uses

Bromoxynil phenol (3,5-dibromo-4-hydroxybenzonitrile; parent compound) and its ester derivatives are commonly grouped together under the name “bromoxynil”. However, it is the phenolic form of bromoxynil that is known as “bromoxynil” (United States Environmental Protection Agency, U.S. EPA, 1998; CCME, 1999a). Henceforth, “bromoxynil” will be used to denote bromoxynil phenol.

Bromoxynil is a contact, post-emergent and selective benzonitrile herbicide used to control a wide spectrum of annual broadleaf weeds in food and feed crops (Health Canada, 2019a). It acts by inhibiting photosynthesis, which stops energy production and negatively affects plant respiration (U.S. EPA, 1998). Bromoxynil and its octanoate and heptanoate esters are currently registered in Canada for use in commercial class products. There are no registered domestic class products in Canada containing bromoxynil or its ester derivatives (Health Canada, 2019a). Bromoxynil was on Health Canada's Pest Management Regulatory Agency (PMRA) yearly list of “Top 10 Herbicide Active Ingredients Sold in Canada” in 2017 (the most recent year for which data are available) with more than 1,000,000 kg of active ingredient (bromoxynil) being sold for use on terrestrial feed and food crops (Health Canada, 2017).

Bromoxynil and its octanoate and heptanoate esters are released into the environment as spray drift that results from agricultural applications and as surface runoff from treated areas (Anderson et al., 1998; CCME, 1999a, Waite et al., 2005; Health Canada, 2019a). In addition, dust particles with adsorbed bromoxynil, bromoxynil octanoate and/or bromoxynil heptanoate can also contaminate surface water (CCME, 1999a).

Bromoxynil is moderately soluble in water (see Table 1). Based on the rapid aerobic and anaerobic conversion of bromoxynil to carbon dioxide, it has a low potential to contaminate groundwater (U.S. EPA, 1998; Health Canada, 2008). Its half-life in prairie wetland ponds is reported as 9-17 days, though residues (> 2 ng/L) may remain up to 120 days after application (Muir et al., 1991). In soil, bromoxynil's half-life is reported to be 4.12 days and its dissipation rate is reported to be 91.25% over 21 days (Chen, 2011).

Both bromoxynil octanoate and bromoxynil heptanoate are chemically and physically similar to each other so that information relating to the environmental fate of the octanoate ester is relevant to the heptanoate ester (U.S. EPA, 1998; Health Canada, 2008). Both esters are almost insoluble in water and readily hydrolyze to bromoxynil at alkaline pH, with half-lives of 1.7 to 34.1 days, depending on pH (CCME, 1998a; U.S. EPA, 1998; Health Canada, 2008). Bromoxynil octanoate is rapidly degraded in both water and soil with photolysis half-lives of 2-4.6 days and microbial degradation half-lives of 12 hours to 3.7 days. Bromoxynil octanoate is mobile in sand, sandy loam, and loam soils. Since bromoxynil octanoate readily degrades to bromoxynil via hydrolysis, the potential for ground water contamination (and exposure) from bromoxynil octanoate is low. Bromoxynil octanoate is also not expected to persist in surface waters (U.S. EPA, 1998).

The reported vapour pressure and Henry's law constant (see Table 1) indicate that bromoxynil and its octanoate and heptanoate esters should not readily volatilize from surface water environments and undergo long-range airborne transport (U.S. EPA, 1998; Health Canada, 2008).

1.2 Substance identity

Bromoxynil (CAS# 1689-84-5; C₇H₃Br₂NO) is a colourless, odourless solid (CCME, 1999b). Bromoxynil octanoate (CAS# 1689-99-2; C₁₅H₁₇Br₂NO₂) and bromoxynil heptanoate (CAS# 56634-95-8; C₁₄H₁₅Br₂NO₂) are brown crystalline solids (U.S. EPA, 1998). In their evaluations, the U.S. EPA (1998) and PMRA (2019a) focused on data for bromoxynil octanoate as it was found to be chemically and physically similar to bromoxynil heptanoate. Therefore, some of the physico-chemical properties of bromoxynil and bromoxynil octanoate are summarized in Table 1.

1.3 Exposure

The general Canadian population can potentially be exposed to bromoxynil through food and drinking water (Health Canada, 2019a).

Water monitoring data from the provinces and territories (municipal and non-municipal supplies), PMRA and Environment Canada, (2011) (Appendix B) were available for bromoxynil.

Data provided by the provinces and territories indicates that bromoxynil levels are below the method reporting limit (MRL) or method detection limit (MDL) in most samples collected. These samples were collected from a variety of water supplies in Canada including surface water and groundwater as well as treated and distributed water where monitoring occurred (British Columbia Ministry of Health, 2019; Indigenous Services Canada, 2019; Manitoba Sustainable Development 2019; Ministère de l'Environnement et de la Lutte contre les changements climatiques du Québec, 2019; Nova Scotia Environment, 2019; Saskatchewan Water Security Agency, 2019; Ontario Ministry of the Environment, Conservation and Parks, 2020). Table 2 summarizes the monitoring data for jurisdictions in which all reported samples were below the MDL. Table 3 summarizes the data for jurisdictions in which bromoxynil detections were reported. The maximum concentration of 6.69 μg/L was reported in the distribution system from surface water in Quebec.

There were no monitoring data available in New Brunswick, Newfoundland and Labrador, Prince Edward Island and Yukon (New Brunswick Department of Environment and Local Government, 2019; Newfoundland and Labrador Municipal Affairs and Environment, 2019; Prince Edward Island Department of Communities, Land and Environment, 2019; Yukon Environmental Health Services, 2019).

As part of its assessment, PMRA (2019a) noted that there were no quantifiable detections of bromoxynil in Canadian groundwater sources. Bromoxynil was detected in surface water in the provinces of Alberta, Manitoba, Saskatchewan, Ontario, and Quebec. The Canadian data show that the overall detection frequency is less than 50% in most studies, with maximum single concentrations in potential surface water sources of drinking water being less than or equal to 0.96 μg/L.

A study of 15 drinking water reservoirs (fed primarily from crop land snowmelt and occasional rainfall runoffs) in Alberta, Manitoba and Saskatchewan detected bromoxynil in 54% of 206 reservoir samples taken between 2003 and 2004, with mean and maximum concentrations of 2.4 ng/L and 384 ng/L, respectively (detection limit = 0.99 ng/L) (Donald et al., 2007). The authors stated that significantly higher mean concentrations were found in July after herbicide application (May to early July, 29.6±12.5 ng/L) as compared to April/May (1.5±0.3 ng/L) after the snowmelt runoff.

Rivers from areas in Quebec where corn and soy are intensely cultivated were also sampled for bromoxynil from 2015 to 2017. For these years, the average detection frequency was 6.4% and the maximum concentration was 0.51 μg/L in the Chibouet, des Hurons, Saint-Régis and Saint-Zéphirin Rivers. In 2015, the bromoxynil detection frequency was 18.2%, 18.2% and 27.3% in the Chaudière, Beaurivage and Le Bras Rivers, respectively, with detected bromoxynil concentrations < 5 μg/L. In 2016, the bromoxynil detection frequency was 10% in the Yamaska River, with detected bromoxynil concentrations < 5 μg/L (Giroux, 2019).

In 2001 and 2002, bromoxynil was detected at one municipal wastewater treatment plant out of seven plants surveyed in Quebec; the detection frequency was 3% and the maximum concentration was 0.26 μg/L (MDL = 0.02 μg/L) (Giroux and Therrien, 2005).

Health Canada (2020) has set maximum residue limits for a variety of food products (including fruits, vegetables, and animal tissue/organs) of 0.02-0.9 ppm of bromoxynil. Domestic and imported food products (i.e., fresh fruit, fresh vegetables, seeds, tree nuts, peanuts, and processed fruit and vegetable products) were sampled and tested by the Canadian Food Inspection Agency (CFIA) between April 1, 2015 and March 31, 2016. No samples (n = 798) tested positive for bromoxynil contamination (CFIA, 2019).

The potential bystander inhalation exposure via spray drift that results from agricultural applications of bromoxynil is expected to be negligible and is not considered to significantly contribute to the overall exposure to bromoxynil (Heath Canada, 2019a). Results from a 3-year (2003-2006) national air surveillance program by the Canadian Atmospheric Network for Currently Used Pesticides (CANCUP) showed atmospheric concentrations of bromoxynil varied within years and time periods with averages ranging from 4.7 to 1,840 pg/m³ (n = 8 locations; MDL = 1.2 pg/m³) (Yao et al., 2008).

Bromoxynil was detected in soil at one sampling site of the CANCUP surveillance study in 2005, the average concentration was 6.23 ng/g (MDL = 0.2 ng/g) (Yao et al. 2008).

2.0 Health Considerations

All pesticides, including bromoxynil, are regulated by PMRA. PMRA conducts extensive evaluations and cyclical reviews of pesticides, including unpublished and proprietary information, as well as foreign reviews by other regulatory agencies such as the U.S. EPA. As such, this health assessment is primarily based on PMRA's evaluations (Health Canada, 2008, 2019a) and supporting documentation. Additionally, any reviews and relevant literature available since PMRA's evaluations were completed were also considered.

2.1 Kinetics

Bromoxynil, bromoxynil octanoate and bromoxynil heptanoate pharmacokinetics are considered to be identical in animals (U.S. EPA, 1998; EFSA 2017). Therefore, pharmacokinetic studies using the octanoate or heptanoate ester are indicative of the pharmacokinetic behaviour of bromoxynil (U.S. EPA, 1998).

2.2 Health effects

The database for the toxicity of bromoxynil is adequate, covering several endpoints and various types of exposure (see U.S. EPA, 1998 for a more thorough review). As bromoxynil and bromoxynil octanoate have been determined to be toxicologically equivalent, studies on both forms were considered in this section (Health Canada, 2019a). In general, bromoxynil and its octanoate ester have a low to high acute oral toxicity in experimental animals depending on the species tested (i.e., guinea pigs being the most sensitive species). As well, bromoxynil and its octanoate ester have low dermal acute toxicity and were well-tolerated in repeated-dose dermal toxicity studies in rabbits. Repeated-dose oral toxicity studies in animals show that the liver is the most sensitive endpoint.

2.3 Effects in humans

In terms of acute exposures to bromoxynil or bromoxynil octanoate, no relevant human data were available.

2.4 Effects in animals

Toxicity studies were available for both bromoxynil and bromoxynil octanoate. Bromoxynil has been shown to be toxic to experimental animals with oral median lethal dose (LD₅₀) values reported for some species as follows: 63 mg/kg of bw in guinea pigs; 81-440 mg/kg bw in rats; 100-245 mg/kg bw in mice and 260-2,000 mg/kg bw in rabbits (U.S. EPA, 1998; CCME, 1999b). Dermal LD₅₀ values of > 2,000 mg/kg (abraded skin, male and female) in rabbits and inhalation median lethal concentration (LC₅₀) values of 0.150 mg/L (female) and 0.269 mg/L (male) in rats have been reported. For bromoxynil octanoate, the reported oral LD₅₀ values were 238 mg/kg bw (female) and 400 mg/kg bw (male) in rats; the dermal LD₅₀ values were 1,310 mg/kg (intact skin, female) and > 2,000 mg/kg (abraded skin, male) in rabbits, and the inhalation LC₅₀were 0.72 mg/L (female) and 0.81 mg/L (male) in rats (U.S. EPA, 1998).

2.5 Genotoxicity and carcinogenicity

Based on the collective data from several in vitro and in vivo tests, bromoxynil is not mutagenic or genotoxic (EFSA, 2017; Health Canada, 2019a).

Bromoxynil was negative in two in vivo studies (mouse micronucleus assays) (Holmstrom and McGregor, 1982; Holmstrom et al., 1991). Negative in vitro studies included unscheduled DNA synthesis (UDS) assay; transformation assay using cultures of mouse C3H/10T1/2 C18 cells; sister chromatid exchange assay using cultures of Chinese hamster ovary (CHO) cells (both with and without activation); mouse lymphoma forward mutation assay (without activation); chromosome aberration assay (CHO; without activation); CHO/ Hypoxanthine-guanine phosphoribosyltransferase (HGPRT) mutation assay (both with and without activation); and Ames study using Salmonella typhimurium (both with and without activation) (Cifone and Balinas, 1982; Galloway and Lebowitz, 1982; Myhr and McKeon, 1982; Rundell and Matthews, 1982; Cifone, 1991; Lawlor, 1991). However, positive results were obtained in three in vitro studies: mouse lymphoma forward mutation assay (with activation); bacterial DNA repair test using Escherichia coli indicator strain (with and without activation); and chromosome aberration assay (CHO; with activation) (Cifone and Balinas, 1982; Galloway and Lebowitz, 1982; Jagannath et al., 1982). Although bromoxynil was positive in three in vitro studies, negative results were obtained in the majority of in vitro studies as well as in in vivo tests.

Bromoxynil octanoate was negative in vitro (Ames study using Salmonella typhimurium [both with and without activation] and in UDS assay) and in vivo (mouse micronucleus assay) (Dillon, 1993; U.S. EPA, 1998).

Carcinogenicity studies using bromoxynil were conducted in rats and mice. Swiss albino mice (60/sex/dose) were administered 10-100 ppm (equivalent to 1.3-13 mg/kg bw per day) of technical grade bromoxynil in their diet. The combined incidence of hepatic adenomas and carcinomas in males increased in a dose-related manner; the increase being statistically significant at the highest dose. No treatment-related increased incidence of tumours was seen in female Swiss mice (Williams, 1994). CD-1 mice (60/sex/dose) were administered 20-300 ppm (equivalent to 3.1-46 mg/kg bw per day in males and 3.7-53 mg/kg bw per day in females) of technical grade bromoxynil in their diet. The combined incidence of hepatic adenomas and carcinomas in males was increased relative to control animals at every dose level tested but the dose response was not linear. An increased incidence of hepatic adenomas and carcinomas (combined) was noted in female CD-1 mice at the highest dose (U.S. EPA, 1998; Health Canada, 2019a).

No neoplastic lesions were associated with treatment in a two-year combined chronic feeding/carcinogenicity study where Sprague Dawley rats (70/sex/dose) were administered 60-600 ppm (equivalent to 2.6-28 mg/kg bw per day in males and 3.3-41 mg/kg bw per day in females) of technical grade bromoxynil in their diet (Hamada, 1988).

Bromoxynil was found to be carcinogenic in mice but not in rats. The U.S. EPA (1998) has classified bromoxynil as a Group C, possible human carcinogen based primarily on observed liver tumours in mice. The International Agency for Research on Cancer has not reviewed the carcinogenicity of bromoxynil. In its special review for bromoxynil and its associated end-use products, PMRA (2019a) concluded that the reference values selected for characterization of non-cancer risks resulting from repeated exposure to bromoxynil are protective of any residual concerns regarding the oncogenic potential of bromoxynil.

2.6 Mode of action

Bromoxynil is an uncoupling agent that disrupts oxidative phosphorylation, which may be responsible for the hepatotoxicity observed in toxicological studies (EFSA, 2017).

Bromoxynil has also been shown to increase the incidence of hepatic adenomas and carcinomas in mice. It was proposed that the generation of hepatocellular tumours could be the result of a mode of action involving activation of the peroxisome proliferator-activated receptor (PPARα) or the constitutive androstane receptor (CAR). As such, the PPARα activation leads to alterations in cell growth pathways, subsequent perturbation of cell growth and survival, then selective clonal expansion of pre-neoplastic cells, and ultimately, the production of hepatic tumours (Health Canada, 2019a). The CAR activation can cause an increase in hepatocyte replicative DNA synthesis, an induction of detoxification enzymes and transporters and an increase in altered hepatic foci and hepatocellular adenomas/carcinomas (Lake, 2018). The role of PPARα activity could not be clearly differentiated from that of CAR (Health Canada, 2019a).

2.7 Selected key study

In its special review for bromoxynil and its associated end-use products (PSRD2019-01), Health Canada (2019a) identified the liver as the most sensitive target organ in the database. As such, an unpublished one-year oral toxicity study in the dog in which the endpoint includes liver effects has been considered the key study for the human health risk assessment of bromoxynil in drinking water (Health Canada, 2019a, 2019b).

In this study, technical grade bromoxynil was administered in gelatin capsules to groups of beagle dogs (6/sex/dose) at dose levels of 0 (control), 0.1, 0.3, 1.5 or 7.5 mg/kg bw per day. The administration of bromoxynil resulted in clinical signs such as panting for animals receiving 1.5 or 7.5 mg/kg bw per day, and salivation, liquid feces and pale gums principally for animals receiving 7.5 mg/kg bw per day. Reductions in body weight and body weight gain were recorded for male animals receiving 1.5 mg/kg bw per day or higher and in female animals receiving 7.5 mg/kg bw per day. Statistically significant reductions in red blood cell parameters (packed cell volume, hemoglobin and red blood cells) were recorded in high dose animals of both sexes. Reductions in these parameters were recorded in males at 1.5 mg/kg bw/day as well; however, they did not attain statistical significance. Changes in biochemical parameters included greater urea and ALT values and lower ALP and protein values for male and female animals receiving 7.5 mg/kg bw per day. Organ weight analysis indicated greater liver weights for animals that received 1.5 or 7.5 mg/kg per day, lower spleen weights for males that received 7.5 mg/kg bw per day and greater prostate weights for males that received 7.5 mg/kg bw per day (Health Canada, 2019b).

An oral no-observed-adverse-effect level (NOAEL) of 0.3 mg/kg bw per day was identified in this study based on increases in clinical signs and liver weight and decreases in body weight and body weight gain observed at the lowest-observed-adverse-effect level of 1.5 mg/kg bw per day (Health Canada, 2019a, 2019b).

3.0 Derivation of the health-based value

As noted above, the NOAEL of 0.3 mg/kg bw per day for increases in clinical signs and liver weight and decreases in body weight and body weight gain in dogs was selected as the basis for the current risk assessment. Using the NOAEL of 0.3 mg bromoxynil/kg bw per day, an acceptable daily intake (ADI) for bromoxynil (Health Canada, 2008, 2019a, 2019b) is calculated as follows:

Where:

• 0.3 mg/kg bw per day is the NOAEL, based on increases in clinical signs and liver weight and decreases in body weight and body weight gain in beagle dogs; and
• 100 is the uncertainty factor, selected to account for interspecies variation (×10) and intraspecies variation (×10).

Equation 1 - Text equivalent

Equation for ADI:

This equation calculates the acceptable daily intake (ADI) for Bromoxynil. The ADI intake is calculated by dividing the alterations in clinical chemistry for Bromoxynil (0.3 micrograms per kilogram of body weight per day) by the uncertainty factor of 100 which equals 0.003 micrograms per kilogram of body weight per day.

Based on the ADI of 0.003 mg/kg bw per day, a health-based value (HBV) for bromoxynil in drinking water was derived as follows:

Where:

• 0.003 mg/kg bw per day is the ADI calculated using a NOAEL of 0.3 mg/kg bw per day (Health Canada, 2019);
• 74 kg is the adult body weight (Health Canada, in preparation);
• 1.53 L per day is the daily volume of tap water consumed by an adult (Health Canada, in preparation).
• 0.20 is the default allocation factor for drinking water (Krishnan and Carrier, 2013).

Equation 2 - Text equivalent

Equation for HBV:

This equation calculates the health based value for Bromoxynil. The health based value is calculated by multiplying the tolerable daily intake of 0.003 micrograms per kilogram of body weight per day by a body weight of 74 kilograms and by an allocation factor of 0.20 and then dividing by a volume of water ingested per day of 1.53 litres. This equals a health based values of 0.03 micrograms per litre.

4.0 Analytical and Treatment Considerations

4.1 Analytical methods to detect bromoxynil

Standardized methods available for the analysis of bromoxynil in source and drinking water and their respective MDLs are summarized in Table 4. MDLs are dependent on the sample matrix, instrumentation, and selected operating conditions and will vary between individual laboratories. These methods are subject to a variety of interferences, which are outlined in the respective references.

A number of accredited laboratories in Canada were contacted to determine MDLs and MRLs for bromoxynil analysis. The MDLs were in the same order of magnitude as those reported in Table 4 and the MRLs ranged between 0.1 and 0.5 μg/L (AGAT Laboratories Ltd., 2019; ALS Environmental (Waterloo), 2019; Bureau Veritas Laboratories, 2019; CARO Analytical Services (Richmond Laboratory), 2019; Element Materials Technology Canada Inc., 2019; SGS Environmental Services, 2019).

Drinking water utilities should discuss sampling requirements with the accredited laboratory conducting the analysis to ensure that quality control procedures are met and that MRLs are low enough to ensure accurate monitoring at concentrations below the proposed MAC. Sample processing considerations and method interferences for the analysis of bromoxynil in drinking water (e.g., sample preservation, storage) can be found in the references listed in Table 4. It is important to note that quenching is critical if an oxidant is present in samples in order to prevent additional degradation of bromoxynil prior to analysis.

4.2 Treatment considerations

Treatment technologies are available to decrease bromoxynil concentrations with varying effectiveness in drinking water. These include activated carbon, membrane filtration, oxidation, advanced oxidation processes (AOPs) and biofiltration. Studies have concluded that granulated activated carbon (GAC) can remove bromoxynil from water (Baup et al., 2000, 2002; Yang et al., 2004). Published performance data for bromoxynil removal are only available for oxidation, AOPs technologies, and biofiltration and indicate wide range of removal efficiencies (39% to 99%) (Preuss et al. 1996; Bourgine et al., 1997; Chelme-Ayala et al., 2010a, 2010b, 2011). Pilot-scale testing is an important step for water utilities considering these processes for pesticide removal in drinking water. At the residential scale, certified treatment devices relying on reverse osmosis (RO) or activated carbon adsorption are expected to be effective for removal of bromoxynil.

4.2.1 Municipal-scale

The selection of an appropriate treatment process will depend on many factors, including the raw water source and its characteristics, the operational conditions of the selected treatment method and the utility's treatment goals. Bench- or pilot-scale testing is recommended to ensure the source water can be successfully treated and optimal process design is established.

The surface water study by Donald et al. (2007) discussed in Section 1.3 also presented bromoxynil concentrations in treated water from 15 drinking water reservoirs. Each drinking water supply had various treatment processes, but all included chlorination and most included flocculation (alum) settling, activated carbon and/or sand filtration. The raw data were not provided thus only overall statistics could be determined, not the performance of individual treatment processes. The mean bromoxynil concentrations in reservoir and treated water were 2.4 ng/L and 1 ng/L, respectively (n = 163), with a maximum bromoxynil concentration in the treated water of 227 ng/L. From 12 paired samples, a mean bromoxynil reduction of 46% was determined (range between 0 and 98%).

When using oxidation or AOPs for pesticide removal in drinking water, it is important to be aware of the potential for formation of by-products due to degradation of the target compound (Ikehata and Gamal El-Din, 2006; Beduk et al., 2012; Li et al., 2019). The primary objective should be removal of the pesticide with the secondary objective being the minimization of by-product formation if they are of health concern. In addition, water utilities should consider the potential for the formation of disinfection by-products depending on the oxidant selected and the source water quality.

4.2.1.1 Conventional treatment

Conventional filtration (chemical coagulation, clarification, and rapid sand filtration) alone is not effective but the addition of chlorine during the disinfection step will reduce bromoxynil concentrations through oxidation. However, degradation processes like oxidation result in the formation of by-products (see Section 4.2.1.4). No studies on the removal of bromoxynil through conventional filtration were found in the literature.

4.2.1.2 Activated carbon adsorption

Activated carbon adsorption is a technology widely used to reduce the concentration of micropollutants, including a wide range of pesticides, in drinking water (Haist-Gulde and Happel, 2012; van der Aa et al., 2012). Activated carbon can be applied in two ways: slurry applications using powdered activated carbon (PAC) or fixed bed reactors with GAC (Chowdhury et al., 2013).

There is very limited published literature on the removal of bromoxynil using activated carbon and no data on performance. As such, prior to full-scale implementation it is essential to conduct appropriate pilot- or bench-scale testing. Bromoxynil removal from natural water using activated carbon can be negatively affected by competition from other contaminants or natural organic matter (NOM), biofilm development, temperature, influent concentration, activated carbon characteristics and hydraulic loading rate (Speth and Miltner, 1998; Haist-Gulde and Happel, 2012).

Data generated through bench-scale testing to determine adsorption coefficients for pesticides are useful in predicting whether activated carbon adsorbs a particular pesticide (U.S. EPA, 2011). In general, pesticides with an adsorption capacity constant (e.g., Freundlich coefficient (K)) greater than 200 µg/g∙(L/µg)1/n are considered to be amenable to removal by carbon adsorption (Speth and Adams, 1993; Speth and Miltner, 1998, U.S. EPA, 2011). The authors noted, however, that the capacity of activated carbon is affected by many factors, including the compound's ionic character and the solution pH.

Two bench-scale studies present Freundlich coefficients for the adsorption of bromoxynil, diuron and atrazine to different activated carbons (Baup et al., 2000, 2002). The authors reported in both batch studies that bromoxynil exhibited an adsorption in between diuron and atrazine, indicating that activated carbon is a good option for bromoxynil removal. Long-term studies, over a 90-day period, were conducted using both granular and powdered forms of activated carbon (Baup et al., 2002). The authors observed that short-term adsorption was better with PAC than GAC and stated that this may be due to better accessibility to pores and adsorption sites. A bench-scale study evaluating adsorption, showed that adsorption capacity declined with increasing pH (see Table 5) (Yang et al., 2004).

The use of PAC offers the advantage of providing virgin carbon when required (e.g., during the pesticide application season) (Miltner et al., 1989). The removal efficiency of PAC depends on the PAC type and dose, the contact time, the PAC characteristics (type, particle size), the adsorbability of the contaminant and the presence of NOM (Gustafson et al., 2003; Summers et al., 2010; Haist-Gulde and Happel, 2012; Chowdhury et al., 2013). In addition to the adsorption capacity of the GAC, the maximum operation time for GAC adsorbers to remove pesticides depends on the filter velocity, empty bed contact time, the GAC characteristics (type, particle size) the adsorbability of the contaminant and the organic background matrix (Haist-Gulde and Happel, 2012). Because GAC fixed bed adsorbers are typically operated on a continuous basis, the GAC can become fouled (or preloaded) with NOM and it may be completely or partially ineffective for pesticide removal (Knappe et al., 1999; Summers et al., 2010; Haist-Gulde and Happel, 2012; Chowdhury et al., 2013).

When bromoxynil is oxidized, degradation by-products can be formed as discussed further in Section 4.2.1.5. One bench-scale study evaluated the adsorption of bromoxynil to PAC as well as one of the ozone degradation by-products, 3-bromo-4,5-dihydroxybenzonitrile (Br-DHBN) (Schoutteten et al., 2016). Separate adsorption tests with initial bromoxynil or Br-DHBN concentration of 5 mg C/L were conducted by mixing 0-18 mg PAC to 80 mL solution on a shaker for three days at 25˚C. The authors stated that the results showed better adsorption of the degradation by-product, Br-DHBN, to PAC than bromoxynil.

Activated carbon is expected to be effective for the removal of bromoxynil based on the limited published literature. As such, it is important to perform appropriate testing prior to full-scale implementation with source water under the proposed operating conditions to ensure that adequate bromoxynil removal is occurring.

4.2.1.3 Membrane filtration

In general, nanofiltration (NF) and RO are effective pressure-driven membrane processes for the removal of pesticides from drinking water (Van der Bruggen and Vandecasteele, 2003; U.S. EPA, 2011). The effectiveness of NF and RO for pesticide removal is dependent on the membrane characteristics, pesticide properties, feed water composition, operating conditions and membrane fouling (Van der Bruggen and Vandecasteele, 2003; Plakas and Karabelas, 2012).

Since the main mechanism for pesticide removal using NF and RO membranes is size exclusion, the molecular weight cut-off (MWCO) of the membrane is an important characteristic. In choosing a membrane, the molecular weight of bromoxynil (> 276 Da), should be considered. In addition to the sieving effect, retention of small pesticide molecules by larger pore size membranes can be influenced by the physicochemical interactions between the pesticide and the membrane surface (Plakas and Karabelas, 2012). Bellona et al. (2004) present a flow-chart using the characteristics of the pesticide in water (e.g. molecular weight, log Kow, molecular diameter) and those of the membrane (e.g., MWCO, pore size) to determine the potential for removal by membrane filtration. Bromoxynil is somewhat hydrophobic (log Kow > 2), and has a fairly low pKa, indicating potential for further removal through hydrophobic bonding to the membrane surface and electrostatic exclusion (Bellona et al., 2004; Plakas and Karabelas, 2012). Based on these physiochemical properties, it is expected that bromoxynil can be removed through membrane processes.

As there is no published literature on the removal of bromoxynil using membrane filtration, it is important to perform appropriate testing prior to full-scale implementation with membrane and source water under the proposed operating conditions to ensure that adequate bromoxynil removal is occurring.

4.2.1.4 Oxidation

Pilot-scale and bench-scale oxidation studies of bromoxynil using ozone (O₃) and ultraviolet (UV) photolysis can be effective treatment methods for removing bromoxynil from water depending on a variety of factors including contact time, oxidant dose and water matrix (pH, alkalinity and organic matter) (Bourgine et al., 1997; Chelme-Ayala et al., 2010a, 2010b; Chelme-Ayala et al., 2011). Bourgine et al. (1997) reported limited data from a pilot-scale test using UV photolysis to reduce the concentrations of several pesticides in groundwater including bromoxynil. The authors indicated that bromoxynil was easily degraded and that the UV photolysis achieved a greater than 85% removal. A bench-scale study conducted by Chelme-Ayala et al. (2010a) investigated degradation of bromoxynil using direct UV photolysis in natural water. The water samples were collected from the North Saskatchewan River, upstream of the City of Edmonton (W1) and from an irrigation return flow discharging into Redwater River (W2). The quality of natural waters is discussed in Table 6. The results indicated that 30 minutes and 56 minutes were required to achieve 50% degradation of bromoxynil in samples W1 and W2, respectively. In addition, higher UV doses were needed for degradation of bromoxynil in water samples with higher total organic carbon concentration (TOC) and higher alkalinity (see Table 6). The reported UV doses are much higher than the typically required UV doses for inactivation of microorganisms (Health Canada, 2019c, 2019d).

A bench-scale ozonation study reported an increase in degradation efficiency of bromoxynil with an increase in pH level (from 2.0 to 7.0) in buffered ultrapure water (Chelme-Ayala et al., 2010b). The different reactivity of bromoxynil was due to the higher reactivity of the deprotonated bromoxynil species formed when pH increased. The reaction rate constants ranged from 2.3x10² to 4.6x10² M^-1s^-1 at a pH level of 2.0 and from 6.0x10² to 6.4x10² M^-1s^-1 at a pH level of 7.0 for the reaction of ozone with bromoxynil using a different kinetic method (see Table 6). Overall, ozone oxidation reactions may proceed through two pathways. In reactions conducted under acidic conditions direct molecular ozonation is the dominant process, while an indirect mechanism by hydroxyl radicals (●OH) controls the reactions at pH greater than 10.0 (Hoigné and Bader, 1976). Gottschalk et al. (2000) indicated that both oxidation pathways, direct ozonation and hydroxyl radical oxidation, may proceed at pH levels greater than 7.0.

Bench-scale ozonation tests examined the effect of pH, alkalinity and humic acid levels on the degradation of bromoxynil in ultrapure and natural waters (sampling discussed above) (Chelme-Ayala et al., 2011) (see Table 7). An increase of the degradation rate of bromoxynil was observed with an increase in pH level from 7.0 to 9.0. Due to the scavenging effect on bicarbonate ions in the ozone radical chain reactions and the reaction of ozone with humic acids, higher degradation efficiency of bromoxynil was reported in water samples with lower levels of organic matter and alkalinity. The scavenging effect of carbonate and bicarbonate ions and sodium nitrite on photolytic degradation of bromoxynil was also reported by Kochany et al. (1990a, 1992).

When bromoxynil is oxidized, degradation by-products can be formed as discussed further in Section 4.2.1.5.

Advanced Oxidation Processes

Bench-scale tests reported a high reactivity of bromoxynil toward ●OH generated by ozone decomposition at basic conditions (pH > 10) (Chelme-Ayala et al., 2010b; Chelme-Ayala et al., 2011). Chelme-Ayala et al. (2010b) calculated a reaction rate constant of 8.45x10⁹ M^-1s^-1 of the reaction of ozone (0.015 mM) with bromoxynil concentration of 9.0 µM at pH 11.0.

Bench-scale removal of bromoxynil using two AOPs, UV/H₂O₂ and O3/H2O2, was studied in natural water samples (sampling discussed in section 4.2.1.4) (Chelme-Ayala et al., 2010a; Chelme-Ayala et al., 2011). Both processes, UV/H₂O₂ and O₃/H₂O₂, reported an increase in the levels of degradation of bromoxynil compared to direct UV photolysis and molecular ozonation, respectively. The results in Table 8 showed that a higher degradation rate of bromoxynil was achieved in water samples having a lower TOC concentration and a lower alkalinity (samples W1). The presence of organic matter and bicarbonate ions in natural water reduced the concentration of ●OH available to react with the pesticides (Chelme-Ayala et al., 2010a; Chelme-Ayala et al., 2011).

Studies identified several degradation by-products of bromoxynil oxidation after UV photolysis and/or ozonation in aqueous solution: 3-bromo-4-hydroxybenzonitrile, 4-hydroxybenzonitrile, 3-bromo-4-hydroxy-5-nitrobenzonitrile, 4-hydroxy-3-nitrobenzonitrile, 3,5-dibromo-2,4-dihydroxybenzonitrile, 3-bromo-4,5-dihydroxybenzonitrile (Kochany and Chaudhry, 1990; Kochany et al., 1992; Marchado et al., 1995; Chelme-Ayala et al., 2010b). Chelme-Ayala et al. (2010b) suggested that the degradation of bromoxynil occurs via hydroxylation and debromination. Studies reported that some pesticide degradation by-products are more persistent than the parent molecule. As such, the application of higher oxidant doses may be required to degrade the target molecule and the degradation by-products generated during the treatment (Chelme-Ayala et al., 2011; Beduk et al., 2012; Li et al., 2019). A H₂O₂/O₃ ratio in the range 0.4-0.6 was suggested as an optimum ratio for pesticide degradation (Glaze et al., 1987; Acero and von Gunten, 2001; Can and Cakir, 2010). Although oxidation and AOPs appear to be effective treatment processes for removal of bromoxynil, water utilities need to consider the formation of the degradation by-products, depending on the oxidant selected and water quality.

4.2.1.5 Biofiltration

Biological treatment involves targeting the removal of the biodegradable organic material fraction. The effectiveness of biological treatment depends on the initial concentration, source water properties, the microbial community, the contact time, the soil properties and the temperature (Drewes et al., 2009; Diem et al., 2013). The main biological treatment processes for drinking water include riverbank filtration, rapid granular media filtration without the maintenance of a disinfectant residual across the bed and slow sand filtration.

A laboratory batch study was conducted to observe the biodegradation of bromoxynil using groundwater added to a salt medium (Preuss et al. 1996). Under anaerobic conditions, biodegradation of bromoxynil was > 99% after 32 days. No biodegradation was observed under aerobic conditions.

4.2.2 Residential-scale

In cases where bromoxynil removal is desired at the household level, for example, when a household obtains its drinking water from a private well, a residential drinking water treatment unit may be an option for decreasing bromoxynil concentrations in drinking water. Before a treatment unit is installed, the water should be tested to determine the general water chemistry and bromoxynil concentration in the source water. There is a lack of performance testing of treatment technologies; however, adsorption (activated carbon) and RO treatment units are expected to remove bromoxynil at the residential-scale. To verify that a treatment unit is effective, water entering and leaving the treatment unit should be sampled periodically and submitted to an accredited laboratory for analysis. Units can lose removal capacity through use and time and need to be maintained and/or replaced. Consumers should verify the expected longevity of the components in the treatment unit according to the manufacturer's recommendations and service it when required. Systems classified as residential scale may have a rated capacity to treat volumes greater than that needed for a single residence, and thus, may also be used in small systems.

Health Canada does not recommend specific brands of drinking water treatment units, but it strongly recommends that consumers use units that have been certified by an accredited certification body as meeting the appropriate NSF International Standard/American National Standard (NSF/ANSI) for drinking water treatment units. The purpose of these standards is to establish minimum requirements for the materials, design and construction of drinking water treatment units that can be tested by a third party. This ensures that materials in the unit do not leach contaminants into the drinking water (i.e., material safety). In addition, the standards include performance requirements that specify the removal that must be achieved for specific contaminants (e.g., reduction claim) that may be present in water supplies. Certification organizations (i.e., third party) provide assurance that a product conforms to applicable standards and must be accredited by the Standards Council of Canada (SCC). Accredited organizations in Canada include:

• CSA Group;
• NSF International;
• Water Quality Association;
• UL LLC;
• Bureau de Normalisation du Québec (available in French only);
• International Association of Plumbing and Mechanical Officials; and
• Truesdail Laboratories Inc.

An up-to-date list of accredited certification organizations can be obtained from the SCC.

The drinking water treatment technologies that are expected to be effective for bromoxynil removal at the residential-scale include adsorption and RO. Currently, bromoxynil is not included in the performance requirements of NSF/ANSI standards. However, consumers can use a treatment unit that is certified to the standards for adsorption or RO to ensure that the material safety has been tested. In addition, units that have been certified for the removal of atrazine are more likely to be effective for the removal of bromoxynil, as bromoxynil was shown to have better adsorption than atrazine (see Section 4.2.1.2).

Water that has been treated using RO may be corrosive to internal plumbing components. Therefore, these units should be installed only at the point-of-use. Also, as large quantities of influent water are needed to obtain the required volume of treated water, these units are generally not practical for point-of-entry installation.

5.0 Management strategies

All water utilities should implement a risk management approach, such as the source-to-tap or water safety plan approach, to ensure water safety (CCME, 2004; WHO, 2011, 2012). These approaches require a system assessment to characterize the source water, describe the treatment barriers that prevent or reduce contamination, to identify the conditions that can result in contamination, and to implement control measures. Operational monitoring is then established, and operational/management protocols are instituted (e.g., standard operating procedures, corrective actions and incident responses). Compliance monitoring is determined and other protocols to validate the water safety plan are implemented (e.g., record keeping, consumer satisfaction). Operator training is also required to ensure the effectiveness of the water safety plan at all times (Smeets et al., 2009).

5.1 Monitoring

Bromoxynil can be present in groundwater and surface water in areas where it is being used depending on the type and extent of its application, environmental factors (e.g., amount of precipitation, soil type, hydrogeological setting) and environmental fate (e.g., mobility, leaching potential, degradation) in the surrounding area. Water utilities should consider the potential for bromoxynil to enter source water (e.g., raw water supply to the drinking water system) based on site-specific considerations.

When it is determined that bromoxynil may be present and monitoring is necessary then surface and groundwater sources should be characterized to determine the concentration of bromoxynil. This should include monitoring of surface water sources during periods of peak use and rainfall events and/or monitoring of groundwater annually. Where baseline data indicate that bromoxynil is not present in source water, monitoring may be reduced.

Where treatment is required to remove bromoxynil, operational monitoring should be implemented to confirm whether the treatment process is functioning as required. The frequency of operational monitoring will depend on the water quality, fluctuations of the raw water concentrations and the treatment process. Responsible authorities should be aware of the impact of NOM on activated carbon systems, as it may impact water quality objectives for bromoxynil removal.

Where treatment is in place for bromoxynil removal, compliance monitoring (i.e., paired samples of source and treated water to confirm the efficacy of treatment) should be conducted at a minimum, on an annual basis. When routine operational monitoring indicates the potential for contaminant breakthrough, such as with GAC, monitoring should be conducted quarterly to plan for the regeneration or replacement of the media. When a degradation process is utilized, oxidation by-product formation should also be considered.

6.0 International Considerations

Other national and international organizations have drinking water guidelines, standards and/or guidance values for bromoxynil in drinking water. Variations in these values can be attributed to the age of the assessments or to differing policies and approaches, including the choice of key study and the use of different consumption rates, body weights and source allocation factors.

Australia has set a guideline value of 0.01 mg/L (10 µg/L) for bromoxynil in drinking water (NHMRC and NRMMC, 2011). This guideline value is based on a no-observed-effect level of 0.3 mg/kg bw per day from a 12-month dietary study in dogs which reported reduced bodyweight gain at 1.5 mg/kg bw per day and evidence of anemia at 7.5 mg/kg bw per day. The U.S. EPA does not have a regulatory value for bromoxynil. It concluded that there are no chronic or acute risks of concern for any subpopulations from drinking water (U.S. EPA, 1998). WHO has not set a specific guideline value for bromoxynil.

The EU does not have a specific chemical parametric value for individual pesticides. Instead, the EU has a value of 0.1 µg/L for any individual (single) pesticide, and a value of 0.5 µg/L for total pesticides found in drinking water. In establishing these values, the EU did not consider the science related to each pesticide, such as health effects. Instead, the values are based on a policy decision to keep pesticides out of drinking water.

7.0 Rationale

Bromoxynil is registered in Canada to control a wide spectrum of annual broadleaf weeds in food and feed crops. Despite its common use in Canada, data provided by provinces and territories that monitor for bromoxynil in source and drinking water indicate that when detected, levels of bromoxynil are below the proposed MAC. The liver is considered the target organ for bromoxynil toxicity. Although no human studies have investigated the effects of bromoxynil on the liver, animal studies conducted in several species (rats, mice and dogs) have consistently shown liver toxicity following bromoxynil exposure. Health Canada, in collaboration with the Federal-Provincial-Territorial Committee on Drinking Water, is proposing a MAC of 0.03 mg/L (30 µg/L) based on the following considerations:

• An HBV of 0.03 mg/L (30 µg/L) based on an increase in clinical signs and liver weight and decreases in body weight and body weight gain observed in dogs.
• Bromoxynil can be accurately measured at concentrations well below the proposed MAC.
• Bromoxynil is expected to be removed at the municipal and residential scale.

The proposed MAC is protective of potential health effects. As part of its ongoing guideline review process, Health Canada will continue to monitor new research in this area and recommend any change to this guideline technical document that it deems necessary.

8.0 References

Appendix A: List of acronyms

ADI

Acceptable daily intake

AHS 

Agricultural Health Study

ALP

Alkaline phosphatase

ALT 

Alanine aminotransferase

ANA

Antinuclear antibodies

ANSI

American National Standards Institute

AOP

Advanced oxidation process

bw

body weight

CANCUP 

Canadian Atmospheric Network for Currently Used Pesticides

CAR

Constitutive androstane receptor

CAS# 

Chemical Abstracts Service registry number

CCME

Canadian Council of Ministers of the Environment  

CDW

Federal-Provincial-Territorial Committee on Drinking Water

CFIA

Canadian Food Inspection Agency

CHO

Chinese hamster ovary

EFSA 

European Food Safety Authority

EU

European Union

FNIHB 

First Nations and Inuit Health Branch

GAC

Granulated activated carbon

GD

Gestation days

HBV

Health-based value

HGPRT

Hypoxanthine-guanine phosphoribosyltransferase

K

Freundlich coefficient

Kow 

n-Octanol:water partition coefficient

LC₅₀

Median lethal concentration

LD₅₀

Median lethal dose

LOAEL  

Lowest-observed-adverse-effect level

MAC 

Maximum acceptable concentration

MDL 

Method detection limit

MRL

Method reporting limit

MWCO 

Molecular weight cut-off

NF

Nanofiltration

NSF  

NSF International Standard

NHMRC

National Health and Medical Research Council (Australia)

NOAEL 

No-observed-adverse-effect level

NOM 

Natural organic matter

NRMCC

Natural Resources Management Ministerial Council (Australia)

NSF

NSF International

●OH 

Hydroxyl radicals

PAC

Powdered activated carbon

PMRA  

Pest Management Regulatory Agency

PPARa

Peroxisome proliferator-activated receptor   

RO 

Reverse osmosis

SCC 

Standards Council of Canada

TOC 

Total organic carbon concentration

UDS 

Unscheduled DNA synthesis

U.S. EPA

United States Environmental Protection Agency

UV  

Ultraviolet

WHO

World Health Organization

Appendix B: Canadian water quality data